/* * Copyright 2010 Christoph Bumiller * * Permission is hereby granted, free of charge, to any person obtaining a * copy of this software and associated documentation files (the "Software"), * to deal in the Software without restriction, including without limitation * the rights to use, copy, modify, merge, publish, distribute, sublicense, * and/or sell copies of the Software, and to permit persons to whom the * Software is furnished to do so, subject to the following conditions: * * The above copyright notice and this permission notice shall be included in * all copies or substantial portions of the Software. * * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL * THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR * OTHER DEALINGS IN THE SOFTWARE. */ #include #include #include "util/u_format.h" #include "util/u_format_s3tc.h" #include "pipe/p_screen.h" #include "vl/vl_decoder.h" #include "vl/vl_video_buffer.h" #include "nouveau_vp3_video.h" #include "nvc0/nvc0_context.h" #include "nvc0/nvc0_screen.h" #include "nvc0/mme/com9097.mme.h" static boolean nvc0_screen_is_format_supported(struct pipe_screen *pscreen, enum pipe_format format, enum pipe_texture_target target, unsigned sample_count, unsigned bindings) { if (sample_count > 8) return FALSE; if (!(0x117 & (1 << sample_count))) /* 0, 1, 2, 4 or 8 */ return FALSE; if (!util_format_is_supported(format, bindings)) return FALSE; if ((bindings & PIPE_BIND_SAMPLER_VIEW) && (target != PIPE_BUFFER)) if (util_format_get_blocksizebits(format) == 3 * 32) return FALSE; /* transfers & shared are always supported */ bindings &= ~(PIPE_BIND_TRANSFER_READ | PIPE_BIND_TRANSFER_WRITE | PIPE_BIND_SHARED); return (nvc0_format_table[format].usage & bindings) == bindings; } static int nvc0_screen_get_param(struct pipe_screen *pscreen, enum pipe_cap param) { const uint16_t class_3d = nouveau_screen(pscreen)->class_3d; struct nouveau_device *dev = nouveau_screen(pscreen)->device; switch (param) { /* non-boolean caps */ case PIPE_CAP_MAX_TEXTURE_2D_LEVELS: case PIPE_CAP_MAX_TEXTURE_CUBE_LEVELS: return 15; case PIPE_CAP_MAX_TEXTURE_3D_LEVELS: return (class_3d >= NVE4_3D_CLASS) ? 13 : 12; case PIPE_CAP_MAX_TEXTURE_ARRAY_LAYERS: return 2048; case PIPE_CAP_MIN_TEXEL_OFFSET: return -8; case PIPE_CAP_MAX_TEXEL_OFFSET: return 7; case PIPE_CAP_MIN_TEXTURE_GATHER_OFFSET: return -32; case PIPE_CAP_MAX_TEXTURE_GATHER_OFFSET: return 31; case PIPE_CAP_MAX_TEXTURE_BUFFER_SIZE: return 65536; case PIPE_CAP_GLSL_FEATURE_LEVEL: return 400; case PIPE_CAP_MAX_RENDER_TARGETS: return 8; case PIPE_CAP_MAX_DUAL_SOURCE_RENDER_TARGETS: return 1; case PIPE_CAP_MAX_STREAM_OUTPUT_BUFFERS: return 4; case PIPE_CAP_MAX_STREAM_OUTPUT_SEPARATE_COMPONENTS: case PIPE_CAP_MAX_STREAM_OUTPUT_INTERLEAVED_COMPONENTS: return 128; case PIPE_CAP_MAX_GEOMETRY_OUTPUT_VERTICES: case PIPE_CAP_MAX_GEOMETRY_TOTAL_OUTPUT_COMPONENTS: return 1024; case PIPE_CAP_MAX_VERTEX_STREAMS: return 4; case PIPE_CAP_MAX_VERTEX_ATTRIB_STRIDE: return 2048; case PIPE_CAP_CONSTANT_BUFFER_OFFSET_ALIGNMENT: return 256; case PIPE_CAP_TEXTURE_BUFFER_OFFSET_ALIGNMENT: return 1; /* 256 for binding as RT, but that's not possible in GL */ case PIPE_CAP_MIN_MAP_BUFFER_ALIGNMENT: return NOUVEAU_MIN_BUFFER_MAP_ALIGN; case PIPE_CAP_MAX_VIEWPORTS: return NVC0_MAX_VIEWPORTS; case PIPE_CAP_MAX_TEXTURE_GATHER_COMPONENTS: return 4; case PIPE_CAP_TEXTURE_BORDER_COLOR_QUIRK: return PIPE_QUIRK_TEXTURE_BORDER_COLOR_SWIZZLE_NV50; case PIPE_CAP_ENDIANNESS: return PIPE_ENDIAN_LITTLE; /* supported caps */ case PIPE_CAP_TEXTURE_MIRROR_CLAMP: case PIPE_CAP_TEXTURE_SWIZZLE: case PIPE_CAP_TEXTURE_SHADOW_MAP: case PIPE_CAP_NPOT_TEXTURES: case PIPE_CAP_MIXED_FRAMEBUFFER_SIZES: case PIPE_CAP_ANISOTROPIC_FILTER: case PIPE_CAP_SEAMLESS_CUBE_MAP: case PIPE_CAP_CUBE_MAP_ARRAY: case PIPE_CAP_TEXTURE_BUFFER_OBJECTS: case PIPE_CAP_TEXTURE_MULTISAMPLE: case PIPE_CAP_TWO_SIDED_STENCIL: case PIPE_CAP_DEPTH_CLIP_DISABLE: case PIPE_CAP_POINT_SPRITE: case PIPE_CAP_TGSI_TEXCOORD: case PIPE_CAP_SM3: case PIPE_CAP_FRAGMENT_COLOR_CLAMPED: case PIPE_CAP_VERTEX_COLOR_UNCLAMPED: case PIPE_CAP_VERTEX_COLOR_CLAMPED: case PIPE_CAP_QUERY_TIMESTAMP: case PIPE_CAP_QUERY_TIME_ELAPSED: case PIPE_CAP_OCCLUSION_QUERY: case PIPE_CAP_STREAM_OUTPUT_PAUSE_RESUME: case PIPE_CAP_QUERY_PIPELINE_STATISTICS: case PIPE_CAP_BLEND_EQUATION_SEPARATE: case PIPE_CAP_INDEP_BLEND_ENABLE: case PIPE_CAP_INDEP_BLEND_FUNC: case PIPE_CAP_TGSI_FS_COORD_ORIGIN_UPPER_LEFT: case PIPE_CAP_TGSI_FS_COORD_PIXEL_CENTER_HALF_INTEGER: case PIPE_CAP_PRIMITIVE_RESTART: case PIPE_CAP_TGSI_INSTANCEID: case PIPE_CAP_VERTEX_ELEMENT_INSTANCE_DIVISOR: case PIPE_CAP_MIXED_COLORBUFFER_FORMATS: case PIPE_CAP_CONDITIONAL_RENDER: case PIPE_CAP_TEXTURE_BARRIER: case PIPE_CAP_QUADS_FOLLOW_PROVOKING_VERTEX_CONVENTION: case PIPE_CAP_START_INSTANCE: case PIPE_CAP_BUFFER_MAP_PERSISTENT_COHERENT: case PIPE_CAP_DRAW_INDIRECT: case PIPE_CAP_USER_CONSTANT_BUFFERS: case PIPE_CAP_USER_INDEX_BUFFERS: case PIPE_CAP_USER_VERTEX_BUFFERS: case PIPE_CAP_PREFER_BLIT_BASED_TEXTURE_TRANSFER: case PIPE_CAP_TEXTURE_QUERY_LOD: case PIPE_CAP_SAMPLE_SHADING: case PIPE_CAP_TEXTURE_GATHER_OFFSETS: case PIPE_CAP_TEXTURE_GATHER_SM5: case PIPE_CAP_TGSI_FS_FINE_DERIVATIVE: case PIPE_CAP_CONDITIONAL_RENDER_INVERTED: case PIPE_CAP_SAMPLER_VIEW_TARGET: case PIPE_CAP_CLIP_HALFZ: case PIPE_CAP_POLYGON_OFFSET_CLAMP: case PIPE_CAP_MULTISAMPLE_Z_RESOLVE: return 1; case PIPE_CAP_SEAMLESS_CUBE_MAP_PER_TEXTURE: return (class_3d >= NVE4_3D_CLASS) ? 1 : 0; case PIPE_CAP_COMPUTE: return (class_3d == NVE4_3D_CLASS) ? 1 : 0; /* unsupported caps */ case PIPE_CAP_TGSI_FS_COORD_ORIGIN_LOWER_LEFT: case PIPE_CAP_TGSI_FS_COORD_PIXEL_CENTER_INTEGER: case PIPE_CAP_SHADER_STENCIL_EXPORT: case PIPE_CAP_TGSI_CAN_COMPACT_CONSTANTS: case PIPE_CAP_VERTEX_BUFFER_OFFSET_4BYTE_ALIGNED_ONLY: case PIPE_CAP_VERTEX_BUFFER_STRIDE_4BYTE_ALIGNED_ONLY: case PIPE_CAP_VERTEX_ELEMENT_SRC_OFFSET_4BYTE_ALIGNED_ONLY: case PIPE_CAP_TGSI_VS_LAYER_VIEWPORT: case PIPE_CAP_FAKE_SW_MSAA: case PIPE_CAP_TGSI_VS_WINDOW_SPACE_POSITION: case PIPE_CAP_VERTEXID_NOBASE: return 0; case PIPE_CAP_VENDOR_ID: return 0x10de; case PIPE_CAP_DEVICE_ID: { uint64_t device_id; if (nouveau_getparam(dev, NOUVEAU_GETPARAM_PCI_DEVICE, &device_id)) { NOUVEAU_ERR("NOUVEAU_GETPARAM_PCI_DEVICE failed.\n"); return -1; } return device_id; } case PIPE_CAP_ACCELERATED: return 1; case PIPE_CAP_VIDEO_MEMORY: return dev->vram_size >> 20; case PIPE_CAP_UMA: return 0; } NOUVEAU_ERR("unknown PIPE_CAP %d\n", param); return 0; } static int nvc0_screen_get_shader_param(struct pipe_screen *pscreen, unsigned shader, enum pipe_shader_cap param) { const uint16_t class_3d = nouveau_screen(pscreen)->class_3d; switch (shader) { case PIPE_SHADER_VERTEX: /* case PIPE_SHADER_TESSELLATION_CONTROL: case PIPE_SHADER_TESSELLATION_EVALUATION: */ case PIPE_SHADER_GEOMETRY: case PIPE_SHADER_FRAGMENT: break; case PIPE_SHADER_COMPUTE: if (class_3d != NVE4_3D_CLASS) return 0; break; default: return 0; } switch (param) { case PIPE_SHADER_CAP_PREFERRED_IR: return PIPE_SHADER_IR_TGSI; case PIPE_SHADER_CAP_MAX_INSTRUCTIONS: case PIPE_SHADER_CAP_MAX_ALU_INSTRUCTIONS: case PIPE_SHADER_CAP_MAX_TEX_INSTRUCTIONS: case PIPE_SHADER_CAP_MAX_TEX_INDIRECTIONS: return 16384; case PIPE_SHADER_CAP_MAX_CONTROL_FLOW_DEPTH: return 16; case PIPE_SHADER_CAP_MAX_INPUTS: if (shader == PIPE_SHADER_VERTEX) return 32; /* NOTE: These only count our slots for GENERIC varyings. * The address space may be larger, but the actual hard limit seems to be * less than what the address space layout permits, so don't add TEXCOORD, * COLOR, etc. here. */ if (shader == PIPE_SHADER_FRAGMENT) return 0x1f0 / 16; /* Actually this counts CLIPVERTEX, which occupies the last generic slot, * and excludes 0x60 per-patch inputs. */ return 0x200 / 16; case PIPE_SHADER_CAP_MAX_OUTPUTS: return 32; case PIPE_SHADER_CAP_MAX_CONST_BUFFER_SIZE: return 65536; case PIPE_SHADER_CAP_MAX_CONST_BUFFERS: if (shader == PIPE_SHADER_COMPUTE && class_3d >= NVE4_3D_CLASS) return NVE4_MAX_PIPE_CONSTBUFS_COMPUTE; return NVC0_MAX_PIPE_CONSTBUFS; case PIPE_SHADER_CAP_INDIRECT_INPUT_ADDR: case PIPE_SHADER_CAP_INDIRECT_OUTPUT_ADDR: return shader != PIPE_SHADER_FRAGMENT; case PIPE_SHADER_CAP_INDIRECT_TEMP_ADDR: case PIPE_SHADER_CAP_INDIRECT_CONST_ADDR: return 1; case PIPE_SHADER_CAP_MAX_PREDS: return 0; case PIPE_SHADER_CAP_MAX_TEMPS: return NVC0_CAP_MAX_PROGRAM_TEMPS; case PIPE_SHADER_CAP_TGSI_CONT_SUPPORTED: return 1; case PIPE_SHADER_CAP_TGSI_SQRT_SUPPORTED: return 0; case PIPE_SHADER_CAP_SUBROUTINES: return 1; case PIPE_SHADER_CAP_INTEGERS: return 1; case PIPE_SHADER_CAP_MAX_TEXTURE_SAMPLERS: return 16; /* would be 32 in linked (OpenGL-style) mode */ case PIPE_SHADER_CAP_MAX_SAMPLER_VIEWS: return 16; /* XXX not sure if more are really safe */ default: NOUVEAU_ERR("unknown PIPE_SHADER_CAP %d\n", param); return 0; } } static float nvc0_screen_get_paramf(struct pipe_screen *pscreen, enum pipe_capf param) { switch (param) { case PIPE_CAPF_MAX_LINE_WIDTH: case PIPE_CAPF_MAX_LINE_WIDTH_AA: return 10.0f; case PIPE_CAPF_MAX_POINT_WIDTH: return 63.0f; case PIPE_CAPF_MAX_POINT_WIDTH_AA: return 63.375f; case PIPE_CAPF_MAX_TEXTURE_ANISOTROPY: return 16.0f; case PIPE_CAPF_MAX_TEXTURE_LOD_BIAS: return 15.0f; case PIPE_CAPF_GUARD_BAND_LEFT: case PIPE_CAPF_GUARD_BAND_TOP: return 0.0f; case PIPE_CAPF_GUARD_BAND_RIGHT: case PIPE_CAPF_GUARD_BAND_BOTTOM: return 0.0f; /* that or infinity */ } NOUVEAU_ERR("unknown PIPE_CAPF %d\n", param); return 0.0f; } static int nvc0_screen_get_compute_param(struct pipe_screen *pscreen, enum pipe_compute_cap param, void *data) { uint64_t *data64 = (uint64_t *)data; const uint16_t obj_class = nvc0_screen(pscreen)->compute->oclass; switch (param) { case PIPE_COMPUTE_CAP_GRID_DIMENSION: data64[0] = 3; return 8; case PIPE_COMPUTE_CAP_MAX_GRID_SIZE: data64[0] = (obj_class >= NVE4_COMPUTE_CLASS) ? 0x7fffffff : 65535; data64[1] = 65535; data64[2] = 65535; return 24; case PIPE_COMPUTE_CAP_MAX_BLOCK_SIZE: data64[0] = 1024; data64[1] = 1024; data64[2] = 64; return 24; case PIPE_COMPUTE_CAP_MAX_THREADS_PER_BLOCK: data64[0] = 1024; return 8; case PIPE_COMPUTE_CAP_MAX_GLOBAL_SIZE: /* g[] */ data64[0] = (uint64_t)1 << 40; return 8; case PIPE_COMPUTE_CAP_MAX_LOCAL_SIZE: /* s[] */ data64[0] = 48 << 10; return 8; case PIPE_COMPUTE_CAP_MAX_PRIVATE_SIZE: /* l[] */ data64[0] = 512 << 10; return 8; case PIPE_COMPUTE_CAP_MAX_INPUT_SIZE: /* c[], arbitrary limit */ data64[0] = 4096; return 8; default: return 0; } } static void nvc0_screen_destroy(struct pipe_screen *pscreen) { struct nvc0_screen *screen = nvc0_screen(pscreen); if (!nouveau_drm_screen_unref(&screen->base)) return; if (screen->base.fence.current) { struct nouveau_fence *current = NULL; /* nouveau_fence_wait will create a new current fence, so wait on the * _current_ one, and remove both. */ nouveau_fence_ref(screen->base.fence.current, ¤t); nouveau_fence_wait(current); nouveau_fence_ref(NULL, ¤t); nouveau_fence_ref(NULL, &screen->base.fence.current); } if (screen->base.pushbuf) screen->base.pushbuf->user_priv = NULL; if (screen->blitter) nvc0_blitter_destroy(screen); if (screen->pm.prog) { screen->pm.prog->code = NULL; /* hardcoded, don't FREE */ nvc0_program_destroy(NULL, screen->pm.prog); } nouveau_bo_ref(NULL, &screen->text); nouveau_bo_ref(NULL, &screen->uniform_bo); nouveau_bo_ref(NULL, &screen->tls); nouveau_bo_ref(NULL, &screen->txc); nouveau_bo_ref(NULL, &screen->fence.bo); nouveau_bo_ref(NULL, &screen->poly_cache); nouveau_bo_ref(NULL, &screen->parm); nouveau_heap_destroy(&screen->lib_code); nouveau_heap_destroy(&screen->text_heap); FREE(screen->tic.entries); nouveau_object_del(&screen->eng3d); nouveau_object_del(&screen->eng2d); nouveau_object_del(&screen->m2mf); nouveau_object_del(&screen->compute); nouveau_object_del(&screen->nvsw); nouveau_screen_fini(&screen->base); FREE(screen); } static int nvc0_graph_set_macro(struct nvc0_screen *screen, uint32_t m, unsigned pos, unsigned size, const uint32_t *data) { struct nouveau_pushbuf *push = screen->base.pushbuf; size /= 4; assert((pos + size) <= 0x800); BEGIN_NVC0(push, SUBC_3D(NVC0_GRAPH_MACRO_ID), 2); PUSH_DATA (push, (m - 0x3800) / 8); PUSH_DATA (push, pos); BEGIN_1IC0(push, SUBC_3D(NVC0_GRAPH_MACRO_UPLOAD_POS), size + 1); PUSH_DATA (push, pos); PUSH_DATAp(push, data, size); return pos + size; } static void nvc0_magic_3d_init(struct nouveau_pushbuf *push, uint16_t obj_class) { BEGIN_NVC0(push, SUBC_3D(0x10cc), 1); PUSH_DATA (push, 0xff); BEGIN_NVC0(push, SUBC_3D(0x10e0), 2); PUSH_DATA (push, 0xff); PUSH_DATA (push, 0xff); BEGIN_NVC0(push, SUBC_3D(0x10ec), 2); PUSH_DATA (push, 0xff); PUSH_DATA (push, 0xff); BEGIN_NVC0(push, SUBC_3D(0x074c), 1); PUSH_DATA (push, 0x3f); BEGIN_NVC0(push, SUBC_3D(0x16a8), 1); PUSH_DATA (push, (3 << 16) | 3); BEGIN_NVC0(push, SUBC_3D(0x1794), 1); PUSH_DATA (push, (2 << 16) | 2); if (obj_class < GM107_3D_CLASS) { BEGIN_NVC0(push, SUBC_3D(0x12ac), 1); PUSH_DATA (push, 0); } BEGIN_NVC0(push, SUBC_3D(0x0218), 1); PUSH_DATA (push, 0x10); BEGIN_NVC0(push, SUBC_3D(0x10fc), 1); PUSH_DATA (push, 0x10); BEGIN_NVC0(push, SUBC_3D(0x1290), 1); PUSH_DATA (push, 0x10); BEGIN_NVC0(push, SUBC_3D(0x12d8), 2); PUSH_DATA (push, 0x10); PUSH_DATA (push, 0x10); BEGIN_NVC0(push, SUBC_3D(0x1140), 1); PUSH_DATA (push, 0x10); BEGIN_NVC0(push, SUBC_3D(0x1610), 1); PUSH_DATA (push, 0xe); BEGIN_NVC0(push, NVC0_3D(VERTEX_ID_GEN_MODE), 1); PUSH_DATA (push, NVC0_3D_VERTEX_ID_GEN_MODE_DRAW_ARRAYS_ADD_START); BEGIN_NVC0(push, SUBC_3D(0x030c), 1); PUSH_DATA (push, 0); BEGIN_NVC0(push, SUBC_3D(0x0300), 1); PUSH_DATA (push, 3); BEGIN_NVC0(push, SUBC_3D(0x02d0), 1); PUSH_DATA (push, 0x3fffff); BEGIN_NVC0(push, SUBC_3D(0x0fdc), 1); PUSH_DATA (push, 1); BEGIN_NVC0(push, SUBC_3D(0x19c0), 1); PUSH_DATA (push, 1); if (obj_class < GM107_3D_CLASS) { BEGIN_NVC0(push, SUBC_3D(0x075c), 1); PUSH_DATA (push, 3); if (obj_class >= NVE4_3D_CLASS) { BEGIN_NVC0(push, SUBC_3D(0x07fc), 1); PUSH_DATA (push, 1); } } /* TODO: find out what software methods 0x1528, 0x1280 and (on nve4) 0x02dc * are supposed to do */ } static void nvc0_screen_fence_emit(struct pipe_screen *pscreen, u32 *sequence) { struct nvc0_screen *screen = nvc0_screen(pscreen); struct nouveau_pushbuf *push = screen->base.pushbuf; /* we need to do it after possible flush in MARK_RING */ *sequence = ++screen->base.fence.sequence; BEGIN_NVC0(push, NVC0_3D(QUERY_ADDRESS_HIGH), 4); PUSH_DATAh(push, screen->fence.bo->offset); PUSH_DATA (push, screen->fence.bo->offset); PUSH_DATA (push, *sequence); PUSH_DATA (push, NVC0_3D_QUERY_GET_FENCE | NVC0_3D_QUERY_GET_SHORT | (0xf << NVC0_3D_QUERY_GET_UNIT__SHIFT)); } static u32 nvc0_screen_fence_update(struct pipe_screen *pscreen) { struct nvc0_screen *screen = nvc0_screen(pscreen); return screen->fence.map[0]; } static int nvc0_screen_init_compute(struct nvc0_screen *screen) { screen->base.base.get_compute_param = nvc0_screen_get_compute_param; switch (screen->base.device->chipset & ~0xf) { case 0xc0: case 0xd0: /* Using COMPUTE has weird effects on 3D state, we need to * investigate this further before enabling it by default. */ if (debug_get_bool_option("NVC0_COMPUTE", FALSE)) return nvc0_screen_compute_setup(screen, screen->base.pushbuf); return 0; case 0xe0: return nve4_screen_compute_setup(screen, screen->base.pushbuf); case 0xf0: case 0x100: case 0x110: return 0; default: return -1; } } boolean nvc0_screen_resize_tls_area(struct nvc0_screen *screen, uint32_t lpos, uint32_t lneg, uint32_t cstack) { struct nouveau_bo *bo = NULL; int ret; uint64_t size = (lpos + lneg) * 32 + cstack; if (size >= (1 << 20)) { NOUVEAU_ERR("requested TLS size too large: 0x%"PRIx64"\n", size); return FALSE; } size *= (screen->base.device->chipset >= 0xe0) ? 64 : 48; /* max warps */ size = align(size, 0x8000); size *= screen->mp_count; size = align(size, 1 << 17); ret = nouveau_bo_new(screen->base.device, NOUVEAU_BO_VRAM, 1 << 17, size, NULL, &bo); if (ret) { NOUVEAU_ERR("failed to allocate TLS area, size: 0x%"PRIx64"\n", size); return FALSE; } nouveau_bo_ref(NULL, &screen->tls); screen->tls = bo; return TRUE; } #define FAIL_SCREEN_INIT(str, err) \ do { \ NOUVEAU_ERR(str, err); \ nvc0_screen_destroy(pscreen); \ return NULL; \ } while(0) struct pipe_screen * nvc0_screen_create(struct nouveau_device *dev) { struct nvc0_screen *screen; struct pipe_screen *pscreen; struct nouveau_object *chan; struct nouveau_pushbuf *push; uint64_t value; uint32_t obj_class; int ret; unsigned i; switch (dev->chipset & ~0xf) { case 0xc0: case 0xd0: case 0xe0: case 0xf0: case 0x100: case 0x110: break; default: return NULL; } screen = CALLOC_STRUCT(nvc0_screen); if (!screen) return NULL; pscreen = &screen->base.base; ret = nouveau_screen_init(&screen->base, dev); if (ret) { nvc0_screen_destroy(pscreen); return NULL; } chan = screen->base.channel; push = screen->base.pushbuf; push->user_priv = screen; push->rsvd_kick = 5; screen->base.vidmem_bindings |= PIPE_BIND_CONSTANT_BUFFER | PIPE_BIND_VERTEX_BUFFER | PIPE_BIND_INDEX_BUFFER | PIPE_BIND_COMMAND_ARGS_BUFFER; screen->base.sysmem_bindings |= PIPE_BIND_VERTEX_BUFFER | PIPE_BIND_INDEX_BUFFER; pscreen->destroy = nvc0_screen_destroy; pscreen->context_create = nvc0_create; pscreen->is_format_supported = nvc0_screen_is_format_supported; pscreen->get_param = nvc0_screen_get_param; pscreen->get_shader_param = nvc0_screen_get_shader_param; pscreen->get_paramf = nvc0_screen_get_paramf; pscreen->get_driver_query_info = nvc0_screen_get_driver_query_info; nvc0_screen_init_resource_functions(pscreen); screen->base.base.get_video_param = nouveau_vp3_screen_get_video_param; screen->base.base.is_video_format_supported = nouveau_vp3_screen_video_supported; ret = nouveau_bo_new(dev, NOUVEAU_BO_GART | NOUVEAU_BO_MAP, 0, 4096, NULL, &screen->fence.bo); if (ret) goto fail; nouveau_bo_map(screen->fence.bo, 0, NULL); screen->fence.map = screen->fence.bo->map; screen->base.fence.emit = nvc0_screen_fence_emit; screen->base.fence.update = nvc0_screen_fence_update; ret = nouveau_object_new(chan, (dev->chipset < 0xe0) ? 0x1f906e : 0x906e, 0x906e, NULL, 0, &screen->nvsw); if (ret) FAIL_SCREEN_INIT("Error creating SW object: %d\n", ret); switch (dev->chipset & ~0xf) { case 0x110: case 0x100: case 0xf0: obj_class = NVF0_P2MF_CLASS; break; case 0xe0: obj_class = NVE4_P2MF_CLASS; break; default: obj_class = NVC0_M2MF_CLASS; break; } ret = nouveau_object_new(chan, 0xbeef323f, obj_class, NULL, 0, &screen->m2mf); if (ret) FAIL_SCREEN_INIT("Error allocating PGRAPH context for M2MF: %d\n", ret); BEGIN_NVC0(push, SUBC_M2MF(NV01_SUBCHAN_OBJECT), 1); PUSH_DATA (push, screen->m2mf->oclass); if (screen->m2mf->oclass == NVE4_P2MF_CLASS) { BEGIN_NVC0(push, SUBC_COPY(NV01_SUBCHAN_OBJECT), 1); PUSH_DATA (push, 0xa0b5); } ret = nouveau_object_new(chan, 0xbeef902d, NVC0_2D_CLASS, NULL, 0, &screen->eng2d); if (ret) FAIL_SCREEN_INIT("Error allocating PGRAPH context for 2D: %d\n", ret); BEGIN_NVC0(push, SUBC_2D(NV01_SUBCHAN_OBJECT), 1); PUSH_DATA (push, screen->eng2d->oclass); BEGIN_NVC0(push, SUBC_2D(NVC0_2D_SINGLE_GPC), 1); PUSH_DATA (push, 0); BEGIN_NVC0(push, NVC0_2D(OPERATION), 1); PUSH_DATA (push, NV50_2D_OPERATION_SRCCOPY); BEGIN_NVC0(push, NVC0_2D(CLIP_ENABLE), 1); PUSH_DATA (push, 0); BEGIN_NVC0(push, NVC0_2D(COLOR_KEY_ENABLE), 1); PUSH_DATA (push, 0); BEGIN_NVC0(push, SUBC_2D(0x0884), 1); PUSH_DATA (push, 0x3f); BEGIN_NVC0(push, SUBC_2D(0x0888), 1); PUSH_DATA (push, 1); BEGIN_NVC0(push, NVC0_2D(COND_MODE), 1); PUSH_DATA (push, NV50_2D_COND_MODE_ALWAYS); BEGIN_NVC0(push, SUBC_2D(NVC0_GRAPH_NOTIFY_ADDRESS_HIGH), 2); PUSH_DATAh(push, screen->fence.bo->offset + 16); PUSH_DATA (push, screen->fence.bo->offset + 16); switch (dev->chipset & ~0xf) { case 0x110: obj_class = GM107_3D_CLASS; break; case 0x100: case 0xf0: obj_class = NVF0_3D_CLASS; break; case 0xe0: switch (dev->chipset) { case 0xea: obj_class = NVEA_3D_CLASS; break; default: obj_class = NVE4_3D_CLASS; break; } break; case 0xd0: obj_class = NVC8_3D_CLASS; break; case 0xc0: default: switch (dev->chipset) { case 0xc8: obj_class = NVC8_3D_CLASS; break; case 0xc1: obj_class = NVC1_3D_CLASS; break; default: obj_class = NVC0_3D_CLASS; break; } break; } ret = nouveau_object_new(chan, 0xbeef003d, obj_class, NULL, 0, &screen->eng3d); if (ret) FAIL_SCREEN_INIT("Error allocating PGRAPH context for 3D: %d\n", ret); screen->base.class_3d = obj_class; BEGIN_NVC0(push, SUBC_3D(NV01_SUBCHAN_OBJECT), 1); PUSH_DATA (push, screen->eng3d->oclass); BEGIN_NVC0(push, NVC0_3D(COND_MODE), 1); PUSH_DATA (push, NVC0_3D_COND_MODE_ALWAYS); if (debug_get_bool_option("NOUVEAU_SHADER_WATCHDOG", TRUE)) { /* kill shaders after about 1 second (at 100 MHz) */ BEGIN_NVC0(push, NVC0_3D(WATCHDOG_TIMER), 1); PUSH_DATA (push, 0x17); } IMMED_NVC0(push, NVC0_3D(ZETA_COMP_ENABLE), dev->drm_version >= 0x01000101); BEGIN_NVC0(push, NVC0_3D(RT_COMP_ENABLE(0)), 8); for (i = 0; i < 8; ++i) PUSH_DATA(push, dev->drm_version >= 0x01000101); BEGIN_NVC0(push, NVC0_3D(RT_CONTROL), 1); PUSH_DATA (push, 1); BEGIN_NVC0(push, NVC0_3D(CSAA_ENABLE), 1); PUSH_DATA (push, 0); BEGIN_NVC0(push, NVC0_3D(MULTISAMPLE_ENABLE), 1); PUSH_DATA (push, 0); BEGIN_NVC0(push, NVC0_3D(MULTISAMPLE_MODE), 1); PUSH_DATA (push, NVC0_3D_MULTISAMPLE_MODE_MS1); BEGIN_NVC0(push, NVC0_3D(MULTISAMPLE_CTRL), 1); PUSH_DATA (push, 0); BEGIN_NVC0(push, NVC0_3D(LINE_WIDTH_SEPARATE), 1); PUSH_DATA (push, 1); BEGIN_NVC0(push, NVC0_3D(PRIM_RESTART_WITH_DRAW_ARRAYS), 1); PUSH_DATA (push, 1); BEGIN_NVC0(push, NVC0_3D(BLEND_SEPARATE_ALPHA), 1); PUSH_DATA (push, 1); BEGIN_NVC0(push, NVC0_3D(BLEND_ENABLE_COMMON), 1); PUSH_DATA (push, 0); if (screen->eng3d->oclass < NVE4_3D_CLASS) { BEGIN_NVC0(push, NVC0_3D(TEX_MISC), 1); PUSH_DATA (push, NVC0_3D_TEX_MISC_SEAMLESS_CUBE_MAP); } else { BEGIN_NVC0(push, NVE4_3D(TEX_CB_INDEX), 1); PUSH_DATA (push, 15); } BEGIN_NVC0(push, NVC0_3D(CALL_LIMIT_LOG), 1); PUSH_DATA (push, 8); /* 128 */ BEGIN_NVC0(push, NVC0_3D(ZCULL_STATCTRS_ENABLE), 1); PUSH_DATA (push, 1); if (screen->eng3d->oclass >= NVC1_3D_CLASS) { BEGIN_NVC0(push, NVC0_3D(CACHE_SPLIT), 1); PUSH_DATA (push, NVC0_3D_CACHE_SPLIT_48K_SHARED_16K_L1); } nvc0_magic_3d_init(push, screen->eng3d->oclass); ret = nouveau_bo_new(dev, NOUVEAU_BO_VRAM, 1 << 17, 1 << 20, NULL, &screen->text); if (ret) goto fail; /* XXX: getting a page fault at the end of the code buffer every few * launches, don't use the last 256 bytes to work around them - prefetch ? */ nouveau_heap_init(&screen->text_heap, 0, (1 << 20) - 0x100); ret = nouveau_bo_new(dev, NOUVEAU_BO_VRAM, 1 << 12, 6 << 16, NULL, &screen->uniform_bo); if (ret) goto fail; PUSH_REFN (push, screen->uniform_bo, NOUVEAU_BO_VRAM | NOUVEAU_BO_WR); for (i = 0; i < 5; ++i) { /* TIC and TSC entries for each unit (nve4+ only) */ /* auxiliary constants (6 user clip planes, base instance id) */ BEGIN_NVC0(push, NVC0_3D(CB_SIZE), 3); PUSH_DATA (push, 512); PUSH_DATAh(push, screen->uniform_bo->offset + (5 << 16) + (i << 9)); PUSH_DATA (push, screen->uniform_bo->offset + (5 << 16) + (i << 9)); BEGIN_NVC0(push, NVC0_3D(CB_BIND(i)), 1); PUSH_DATA (push, (15 << 4) | 1); if (screen->eng3d->oclass >= NVE4_3D_CLASS) { unsigned j; BEGIN_1IC0(push, NVC0_3D(CB_POS), 9); PUSH_DATA (push, 0); for (j = 0; j < 8; ++j) PUSH_DATA(push, j); } else { BEGIN_NVC0(push, NVC0_3D(TEX_LIMITS(i)), 1); PUSH_DATA (push, 0x54); } } BEGIN_NVC0(push, NVC0_3D(LINKED_TSC), 1); PUSH_DATA (push, 0); /* return { 0.0, 0.0, 0.0, 0.0 } for out-of-bounds vtxbuf access */ BEGIN_NVC0(push, NVC0_3D(CB_SIZE), 3); PUSH_DATA (push, 256); PUSH_DATAh(push, screen->uniform_bo->offset + (5 << 16) + (6 << 9)); PUSH_DATA (push, screen->uniform_bo->offset + (5 << 16) + (6 << 9)); BEGIN_1IC0(push, NVC0_3D(CB_POS), 5); PUSH_DATA (push, 0); PUSH_DATAf(push, 0.0f); PUSH_DATAf(push, 0.0f); PUSH_DATAf(push, 0.0f); PUSH_DATAf(push, 0.0f); BEGIN_NVC0(push, NVC0_3D(VERTEX_RUNOUT_ADDRESS_HIGH), 2); PUSH_DATAh(push, screen->uniform_bo->offset + (5 << 16) + (6 << 9)); PUSH_DATA (push, screen->uniform_bo->offset + (5 << 16) + (6 << 9)); if (dev->drm_version >= 0x01000101) { ret = nouveau_getparam(dev, NOUVEAU_GETPARAM_GRAPH_UNITS, &value); if (ret) { NOUVEAU_ERR("NOUVEAU_GETPARAM_GRAPH_UNITS failed.\n"); goto fail; } } else { if (dev->chipset >= 0xe0 && dev->chipset < 0xf0) value = (8 << 8) | 4; else value = (16 << 8) | 4; } screen->mp_count = value >> 8; screen->mp_count_compute = screen->mp_count; nvc0_screen_resize_tls_area(screen, 128 * 16, 0, 0x200); BEGIN_NVC0(push, NVC0_3D(CODE_ADDRESS_HIGH), 2); PUSH_DATAh(push, screen->text->offset); PUSH_DATA (push, screen->text->offset); BEGIN_NVC0(push, NVC0_3D(TEMP_ADDRESS_HIGH), 4); PUSH_DATAh(push, screen->tls->offset); PUSH_DATA (push, screen->tls->offset); PUSH_DATA (push, screen->tls->size >> 32); PUSH_DATA (push, screen->tls->size); BEGIN_NVC0(push, NVC0_3D(WARP_TEMP_ALLOC), 1); PUSH_DATA (push, 0); BEGIN_NVC0(push, NVC0_3D(LOCAL_BASE), 1); PUSH_DATA (push, 0); if (screen->eng3d->oclass < GM107_3D_CLASS) { ret = nouveau_bo_new(dev, NOUVEAU_BO_VRAM, 1 << 17, 1 << 20, NULL, &screen->poly_cache); if (ret) goto fail; BEGIN_NVC0(push, NVC0_3D(VERTEX_QUARANTINE_ADDRESS_HIGH), 3); PUSH_DATAh(push, screen->poly_cache->offset); PUSH_DATA (push, screen->poly_cache->offset); PUSH_DATA (push, 3); } ret = nouveau_bo_new(dev, NOUVEAU_BO_VRAM, 1 << 17, 1 << 17, NULL, &screen->txc); if (ret) goto fail; BEGIN_NVC0(push, NVC0_3D(TIC_ADDRESS_HIGH), 3); PUSH_DATAh(push, screen->txc->offset); PUSH_DATA (push, screen->txc->offset); PUSH_DATA (push, NVC0_TIC_MAX_ENTRIES - 1); BEGIN_NVC0(push, NVC0_3D(TSC_ADDRESS_HIGH), 3); PUSH_DATAh(push, screen->txc->offset + 65536); PUSH_DATA (push, screen->txc->offset + 65536); PUSH_DATA (push, NVC0_TSC_MAX_ENTRIES - 1); BEGIN_NVC0(push, NVC0_3D(SCREEN_Y_CONTROL), 1); PUSH_DATA (push, 0); BEGIN_NVC0(push, NVC0_3D(WINDOW_OFFSET_X), 2); PUSH_DATA (push, 0); PUSH_DATA (push, 0); BEGIN_NVC0(push, NVC0_3D(ZCULL_REGION), 1); /* deactivate ZCULL */ PUSH_DATA (push, 0x3f); BEGIN_NVC0(push, NVC0_3D(CLIP_RECTS_MODE), 1); PUSH_DATA (push, NVC0_3D_CLIP_RECTS_MODE_INSIDE_ANY); BEGIN_NVC0(push, NVC0_3D(CLIP_RECT_HORIZ(0)), 8 * 2); for (i = 0; i < 8 * 2; ++i) PUSH_DATA(push, 0); BEGIN_NVC0(push, NVC0_3D(CLIP_RECTS_EN), 1); PUSH_DATA (push, 0); BEGIN_NVC0(push, NVC0_3D(CLIPID_ENABLE), 1); PUSH_DATA (push, 0); /* neither scissors, viewport nor stencil mask should affect clears */ BEGIN_NVC0(push, NVC0_3D(CLEAR_FLAGS), 1); PUSH_DATA (push, 0); BEGIN_NVC0(push, NVC0_3D(VIEWPORT_TRANSFORM_EN), 1); PUSH_DATA (push, 1); for (i = 0; i < NVC0_MAX_VIEWPORTS; i++) { BEGIN_NVC0(push, NVC0_3D(DEPTH_RANGE_NEAR(i)), 2); PUSH_DATAf(push, 0.0f); PUSH_DATAf(push, 1.0f); } BEGIN_NVC0(push, NVC0_3D(VIEW_VOLUME_CLIP_CTRL), 1); PUSH_DATA (push, NVC0_3D_VIEW_VOLUME_CLIP_CTRL_UNK1_UNK1); /* We use scissors instead of exact view volume clipping, * so they're always enabled. */ for (i = 0; i < NVC0_MAX_VIEWPORTS; i++) { BEGIN_NVC0(push, NVC0_3D(SCISSOR_ENABLE(i)), 3); PUSH_DATA (push, 1); PUSH_DATA (push, 8192 << 16); PUSH_DATA (push, 8192 << 16); } #define MK_MACRO(m, n) i = nvc0_graph_set_macro(screen, m, i, sizeof(n), n); i = 0; MK_MACRO(NVC0_3D_MACRO_VERTEX_ARRAY_PER_INSTANCE, mme9097_per_instance_bf); MK_MACRO(NVC0_3D_MACRO_BLEND_ENABLES, mme9097_blend_enables); MK_MACRO(NVC0_3D_MACRO_VERTEX_ARRAY_SELECT, mme9097_vertex_array_select); MK_MACRO(NVC0_3D_MACRO_TEP_SELECT, mme9097_tep_select); MK_MACRO(NVC0_3D_MACRO_GP_SELECT, mme9097_gp_select); MK_MACRO(NVC0_3D_MACRO_POLYGON_MODE_FRONT, mme9097_poly_mode_front); MK_MACRO(NVC0_3D_MACRO_POLYGON_MODE_BACK, mme9097_poly_mode_back); MK_MACRO(NVC0_3D_MACRO_DRAW_ARRAYS_INDIRECT, mme9097_draw_arrays_indirect); MK_MACRO(NVC0_3D_MACRO_DRAW_ELEMENTS_INDIRECT, mme9097_draw_elts_indirect); BEGIN_NVC0(push, NVC0_3D(RASTERIZE_ENABLE), 1); PUSH_DATA (push, 1); BEGIN_NVC0(push, NVC0_3D(RT_SEPARATE_FRAG_DATA), 1); PUSH_DATA (push, 1); BEGIN_NVC0(push, NVC0_3D(MACRO_GP_SELECT), 1); PUSH_DATA (push, 0x40); BEGIN_NVC0(push, NVC0_3D(LAYER), 1); PUSH_DATA (push, 0); BEGIN_NVC0(push, NVC0_3D(MACRO_TEP_SELECT), 1); PUSH_DATA (push, 0x30); BEGIN_NVC0(push, NVC0_3D(PATCH_VERTICES), 1); PUSH_DATA (push, 3); BEGIN_NVC0(push, NVC0_3D(SP_SELECT(2)), 1); PUSH_DATA (push, 0x20); BEGIN_NVC0(push, NVC0_3D(SP_SELECT(0)), 1); PUSH_DATA (push, 0x00); BEGIN_NVC0(push, NVC0_3D(POINT_COORD_REPLACE), 1); PUSH_DATA (push, 0); BEGIN_NVC0(push, NVC0_3D(POINT_RASTER_RULES), 1); PUSH_DATA (push, NVC0_3D_POINT_RASTER_RULES_OGL); IMMED_NVC0(push, NVC0_3D(EDGEFLAG), 1); if (nvc0_screen_init_compute(screen)) goto fail; PUSH_KICK (push); screen->tic.entries = CALLOC(4096, sizeof(void *)); screen->tsc.entries = screen->tic.entries + 2048; if (!nvc0_blitter_create(screen)) goto fail; nouveau_fence_new(&screen->base, &screen->base.fence.current, FALSE); return pscreen; fail: nvc0_screen_destroy(pscreen); return NULL; } int nvc0_screen_tic_alloc(struct nvc0_screen *screen, void *entry) { int i = screen->tic.next; while (screen->tic.lock[i / 32] & (1 << (i % 32))) i = (i + 1) & (NVC0_TIC_MAX_ENTRIES - 1); screen->tic.next = (i + 1) & (NVC0_TIC_MAX_ENTRIES - 1); if (screen->tic.entries[i]) nv50_tic_entry(screen->tic.entries[i])->id = -1; screen->tic.entries[i] = entry; return i; } int nvc0_screen_tsc_alloc(struct nvc0_screen *screen, void *entry) { int i = screen->tsc.next; while (screen->tsc.lock[i / 32] & (1 << (i % 32))) i = (i + 1) & (NVC0_TSC_MAX_ENTRIES - 1); screen->tsc.next = (i + 1) & (NVC0_TSC_MAX_ENTRIES - 1); if (screen->tsc.entries[i]) nv50_tsc_entry(screen->tsc.entries[i])->id = -1; screen->tsc.entries[i] = entry; return i; }