nvfx: new 2D: use new 2D engine in Gallium

This patch implements nv04_surface_copy/fill using the new 2D engine module.
It supports falling back to the 3D engine using the u_blitter module, which will be
added in a later patch.

Also adds support for using the 3D engine, reusing the u_blitter module
created for r300.
This is used for unswizzling and copies between swizzled surfaces.

1 files changed, 319 insertions, 27 deletions

diff --git a/src/gallium/drivers/nvfx/nvfx_surface.c b/src/gallium/drivers/nvfx/nvfx_surface.c
index a605d2b7545..a97f342c646 100644
--- a/src/gallium/drivers/nvfx/nvfx_surface.c
+++ b/src/gallium/drivers/nvfx/nvfx_surface.c
@@ -26,33 +26,319 @@
  *
  **************************************************************************/
 
+#include "pipe/p_context.h"
+#include "pipe/p_format.h"
+#include "util/u_format.h"
+#include "util/u_math.h"
+#include "util/u_memory.h"
+#include "util/u_pack_color.h"
+#include "util/u_rect.h"
+#include "util/u_blitter.h"
+
+#include "nouveau/nouveau_winsys.h"
+#include "nouveau/nouveau_util.h"
+#include "nouveau/nouveau_screen.h"
 #include "nvfx_context.h"
+#include "nvfx_screen.h"
 #include "nvfx_resource.h"
-#include "pipe/p_defines.h"
-#include "util/u_inlines.h"
-#include "util/u_pack_color.h"
+#include "nv04_2d.h"
+
+#include <nouveau/nouveau_bo.h>
+
+static INLINE void
+nvfx_region_set_format(struct nv04_region* rgn, enum pipe_format format)
+{
+	unsigned bits = util_format_get_blocksizebits(format);
+	switch(bits)
+	{
+	case 8:
+		rgn->bpps = 0;
+		break;
+	case 16:
+		rgn->bpps = 1;
+		break;
+	case 32:
+		rgn->bpps = 2;
+		break;
+	default:
+		assert(util_is_pot(bits));
+		int shift = log2i(bits) - 3;
+		assert(shift >= 2);
+		rgn->bpps = 2;
+		shift -= 2;
+
+		rgn->x = util_format_get_nblocksx(format, rgn->x) << shift;
+		rgn->y = util_format_get_nblocksy(format, rgn->y);
+	}
+}
+
+static INLINE void
+nvfx_region_fixup_swizzled(struct nv04_region* rgn, unsigned zslice, unsigned width, unsigned height, unsigned depth)
+{
+	// TODO: move this code to surface creation?
+	if((depth <= 1) && (height <= 1 || width <= 2))
+		rgn->pitch = width << rgn->bpps;
+	else if(depth > 1 && height <= 2 && width <= 2)
+	{
+		rgn->pitch = width << rgn->bpps;
+		rgn->offset += (zslice * width * height) << rgn->bpps;
+	}
+	else
+	{
+		rgn->pitch = 0;
+		rgn->z = zslice;
+		rgn->w = width;
+		rgn->h = height;
+		rgn->d = depth;
+	}
+}
+
+static INLINE void
+nvfx_region_init_for_surface(struct nv04_region* rgn, struct nvfx_surface* surf, unsigned x, unsigned y)
+{
+	rgn->bo = ((struct nvfx_resource*)surf->base.texture)->bo;
+	rgn->offset = surf->base.offset;
+	rgn->pitch = surf->pitch;
+	rgn->x = x;
+	rgn->y = y;
+	rgn->z = 0;
+
+	nvfx_region_set_format(rgn, surf->base.format);
+	if(!(surf->base.texture->flags & NVFX_RESOURCE_FLAG_LINEAR))
+		nvfx_region_fixup_swizzled(rgn, surf->base.zslice, surf->base.width, surf->base.height, u_minify(surf->base.texture->depth0, surf->base.level));
+}
+
+static INLINE void
+nvfx_region_init_for_subresource(struct nv04_region* rgn, struct pipe_resource* pt, struct pipe_subresource sub, unsigned x, unsigned y, unsigned z)
+{
+	rgn->bo = ((struct nvfx_resource*)pt)->bo;
+	rgn->offset = nvfx_subresource_offset(pt, sub.face, sub.level, z);
+	rgn->pitch = nvfx_subresource_pitch(pt, sub.level);
+	rgn->x = x;
+	rgn->y = y;
+	rgn->z = 0;
+
+	nvfx_region_set_format(rgn, pt->format);
+	if(!(pt->flags & NVFX_RESOURCE_FLAG_LINEAR))
+		nvfx_region_fixup_swizzled(rgn, z, u_minify(pt->width0, sub.level), u_minify(pt->height0, sub.level), u_minify(pt->depth0, sub.level));
+}
+
+// TODO: actually test this for all formats, it's probably wrong for some...
+
+static INLINE int
+nvfx_surface_format(enum pipe_format format)
+{
+	switch(util_format_get_blocksize(format)) {
+	case 1:
+		return NV04_CONTEXT_SURFACES_2D_FORMAT_Y8;
+	case 2:
+		//return NV04_CONTEXT_SURFACES_2D_FORMAT_Y16;
+		return NV04_CONTEXT_SURFACES_2D_FORMAT_R5G6B5;
+	case 4:
+		//if(format == PIPE_FORMAT_B8G8R8X8_UNORM || format == PIPE_FORMAT_B8G8R8A8_UNORM)
+			return NV04_CONTEXT_SURFACES_2D_FORMAT_A8R8G8B8;
+		//else
+		//	return NV04_CONTEXT_SURFACES_2D_FORMAT_Y32;
+	default:
+		return -1;
+	}
+}
+
+static INLINE int
+nv04_scaled_image_format(enum pipe_format format)
+{
+	switch(util_format_get_blocksize(format)) {
+	case 1:
+		return NV03_SCALED_IMAGE_FROM_MEMORY_COLOR_FORMAT_Y8;
+	case 2:
+		//if(format == PIPE_FORMAT_B5G5R5A1_UNORM)
+		//	return NV03_SCALED_IMAGE_FROM_MEMORY_COLOR_FORMAT_A1R5G5B5;
+		//else
+			return NV03_SCALED_IMAGE_FROM_MEMORY_COLOR_FORMAT_R5G6B5;
+	case 4:
+		if(format == PIPE_FORMAT_B8G8R8X8_UNORM)
+			return NV03_SCALED_IMAGE_FROM_MEMORY_COLOR_FORMAT_X8R8G8B8;
+		else
+			return NV03_SCALED_IMAGE_FROM_MEMORY_COLOR_FORMAT_A8R8G8B8;
+	default:
+		return -1;
+	}
+}
+
+static struct blitter_context*
+nvfx_get_blitter(struct pipe_context* pipe, int copy)
+{
+	struct nvfx_context* nvfx = nvfx_context(pipe);
+
+	struct blitter_context* blitter = nvfx->blitter;
+	if(!blitter)
+		nvfx->blitter = blitter = util_blitter_create(pipe);
+
+	util_blitter_save_blend(blitter, nvfx->blend);
+	util_blitter_save_depth_stencil_alpha(blitter, nvfx->zsa);
+	util_blitter_save_stencil_ref(blitter, &nvfx->stencil_ref);
+	util_blitter_save_rasterizer(blitter, nvfx->rasterizer);
+	util_blitter_save_fragment_shader(blitter, nvfx->fragprog);
+	util_blitter_save_vertex_shader(blitter, nvfx->vertprog);
+	util_blitter_save_viewport(blitter, &nvfx->viewport);
+	util_blitter_save_framebuffer(blitter, &nvfx->framebuffer);
+	util_blitter_save_clip(blitter, &nvfx->clip);
+	util_blitter_save_vertex_elements(blitter, nvfx->vtxelt);
+	util_blitter_save_vertex_buffers(blitter, nvfx->vtxbuf_nr, nvfx->vtxbuf);
+
+	if(copy)
+	{
+		util_blitter_save_fragment_sampler_states(blitter, nvfx->nr_samplers, (void**)nvfx->tex_sampler);
+		util_blitter_save_fragment_sampler_views(blitter, nvfx->nr_textures, nvfx->fragment_sampler_views);
+	}
+
+	return blitter;
+}
+
+static unsigned
+nvfx_region_clone(struct nv04_2d_context* ctx, struct nv04_region* rgn, unsigned w, unsigned h, boolean for_read)
+{
+	unsigned begin = nv04_region_begin(rgn, w, h);
+	unsigned end = nv04_region_end(rgn, w, h);
+	unsigned size = end - begin;
+	struct nouveau_bo* bo = 0;
+	nouveau_bo_new(rgn->bo->device, NOUVEAU_BO_MAP | NOUVEAU_BO_GART, 256, size, &bo);
+
+	if(for_read || (size > ((w * h) << rgn->bpps)))
+		nv04_memcpy(ctx, bo, 0, rgn->bo, rgn->offset + begin, size);
+
+	rgn->bo = bo;
+	rgn->offset = -begin;
+	return begin;
+}
 
 static void
-nvfx_surface_copy(struct pipe_context *pipe,
-		  struct pipe_resource *dest, struct pipe_subresource subdst,
-		  unsigned destx, unsigned desty, unsigned destz,
-		  struct pipe_resource *src, struct pipe_subresource subsrc,
+nvfx_resource_copy_region(struct pipe_context *pipe,
+		  struct pipe_resource *dstr, struct pipe_subresource subdst,
+		  unsigned dstx, unsigned dsty, unsigned dstz,
+		  struct pipe_resource *srcr, struct pipe_subresource subsrc,
 		  unsigned srcx, unsigned srcy, unsigned srcz,
-		  unsigned width, unsigned height)
+		  unsigned w, unsigned h)
 {
-	struct nvfx_context *nvfx = nvfx_context(pipe);
-	struct nv04_surface_2d *eng2d = nvfx->screen->eng2d;
-	struct pipe_surface *ps_dst, *ps_src;
+	struct nv04_2d_context *ctx = nvfx_screen(pipe->screen)->eng2d;
+	struct nv04_region dst, src;
+
+	if(!w || !h)
+		return;
+
+	static int copy_threshold = -1;
+	if(copy_threshold < 0)
+	{
+		copy_threshold = debug_get_num_option("NOUVEAU_COPY_THRESHOLD", 0);
+		if(copy_threshold < 0)
+			copy_threshold = 0;
+	}
 
-	ps_src = nvfx_miptree_surface_new(pipe->screen, src, subsrc.face,
-					  subsrc.level, srcz, 0 /* bind flags */);
-	ps_dst = nvfx_miptree_surface_new(pipe->screen, dest, subdst.face,
-					  subdst.level, destz, 0 /* bindflags */);
+	int dst_to_gpu = dstr->usage != PIPE_USAGE_DYNAMIC && dstr->usage != PIPE_USAGE_STAGING;
+	int src_on_gpu = nvfx_resource_on_gpu(srcr);
 
-	eng2d->copy(eng2d, ps_dst, destx, desty, ps_src, srcx, srcy, width, height);
+	nvfx_region_init_for_subresource(&dst, dstr, subdst, dstx, dsty, dstz);
+	nvfx_region_init_for_subresource(&src, srcr, subsrc, srcx, srcy, srcz);
+	w = util_format_get_stride(dstr->format, w) >> dst.bpps;
+	h = util_format_get_nblocksy(dstr->format, h);
 
-	nvfx_miptree_surface_del(ps_src);
-	nvfx_miptree_surface_del(ps_dst);
+	int ret;
+	boolean small = (w * h <= copy_threshold);
+	if((!dst_to_gpu || !src_on_gpu) && small)
+		ret = -1; /* use the CPU */
+	else
+		ret = nv04_region_copy_2d(ctx, &dst, &src, w, h,
+			dstr->target == PIPE_BUFFER ? -1 : nvfx_surface_format(dstr->format),
+			dstr->target == PIPE_BUFFER ? -1 : nv04_scaled_image_format(dstr->format),
+			dst_to_gpu, src_on_gpu);
+	if(!ret)
+	{}
+	else if(ret > 0 && dstr->bind & PIPE_BIND_RENDER_TARGET && srcr->bind & PIPE_BIND_SAMPLER_VIEW)
+	{
+		struct blitter_context* blitter = nvfx_get_blitter(pipe, 1);
+		util_blitter_copy_region(blitter, dstr, subdst, dstx, dsty, dstz, srcr, subsrc, srcx, srcy, srcz, w, h, TRUE);
+	}
+	else
+	{
+		struct nv04_region dstt = dst;
+		struct nv04_region srct = src;
+		unsigned dstbegin = 0;
+
+		if(!small)
+		{
+			if(src_on_gpu)
+				nvfx_region_clone(ctx, &srct, w, h, TRUE);
+
+			if(dst_to_gpu)
+				dstbegin = nvfx_region_clone(ctx, &dstt, w, h, FALSE);
+		}
+
+		nv04_region_copy_cpu(&dstt, &srct, w, h);
+
+		if(srct.bo != src.bo)
+			nouveau_screen_bo_release(pipe->screen, srct.bo);
+
+		if(dstt.bo != dst.bo)
+		{
+			nv04_memcpy(ctx, dst.bo, dst.offset + dstbegin, dstt.bo, 0, dstt.bo->size);
+			nouveau_screen_bo_release(pipe->screen, dstt.bo);
+		}
+	}
+}
+
+static int
+nvfx_surface_fill(struct pipe_context* pipe, struct pipe_surface *dsts,
+		  unsigned dx, unsigned dy, unsigned w, unsigned h, unsigned value)
+{
+	struct nv04_2d_context *ctx = nvfx_screen(pipe->screen)->eng2d;
+	struct nv04_region dst;
+	/* Always try to use the GPU right now, if possible
+	 * If the user wanted the surface data on the CPU, he would have cleared with memset */
+
+	// we don't care about interior pixel order since we set all them to the same value
+	nvfx_region_init_for_surface(&dst, (struct nvfx_surface*)dsts, dx, dy);
+	w = util_format_get_stride(dsts->format, w) >> dst.bpps;
+	h = util_format_get_nblocksy(dsts->format, h);
+
+	int ret = nv04_region_fill_2d(ctx, &dst, w, h, value);
+	if(ret > 0 && dsts->texture->bind & PIPE_BIND_RENDER_TARGET)
+		return 1;
+	else if(ret)
+	{
+		struct nv04_region dstt = dst;
+		unsigned dstbegin = 0;
+
+		if(nvfx_resource_on_gpu(dsts->texture))
+			dstbegin = nvfx_region_clone(ctx, &dstt, w, h, FALSE);
+
+		nv04_region_fill_cpu(&dstt, w, h, value);
+
+		if(dstt.bo != dst.bo)
+		{
+			nv04_memcpy(ctx, dst.bo, dst.offset + dstbegin, dstt.bo, 0, dstt.bo->size);
+			nouveau_screen_bo_release(pipe->screen, dstt.bo);
+		}
+	}
+
+	return 0;
+}
+
+
+void
+nvfx_screen_surface_takedown(struct pipe_screen *pscreen)
+{
+	nv04_2d_context_takedown(nvfx_screen(pscreen)->eng2d);
+	nvfx_screen(pscreen)->eng2d = 0;
+}
+
+int
+nvfx_screen_surface_init(struct pipe_screen *pscreen)
+{
+	struct nv04_2d_context* ctx = nv04_2d_context_init(nouveau_screen(pscreen)->channel);
+	if(!ctx)
+		return -1;
+	nvfx_screen(pscreen)->eng2d = ctx;
+	return 0;
 }
 
 static void
@@ -62,12 +348,16 @@ nvfx_clear_render_target(struct pipe_context *pipe,
 			 unsigned dstx, unsigned dsty,
 			 unsigned width, unsigned height)
 {
-	struct nvfx_context *nvfx = nvfx_context(pipe);
-	struct nv04_surface_2d *eng2d = nvfx->screen->eng2d;
 	union util_color uc;
 	util_pack_color(rgba, dst->format, &uc);
 
-	eng2d->fill(eng2d, dst, dstx, dsty, width, height, uc.ui);
+	if(util_format_get_blocksizebits(dst->format) > 32
+		|| nvfx_surface_fill(pipe, dst, dstx, dsty, width, height, uc.ui))
+	{
+		// TODO: probably should use hardware clear here instead if possible
+		struct blitter_context* blitter = nvfx_get_blitter(pipe, 0);
+		util_blitter_clear_render_target(blitter, dst, rgba, dstx, dsty, width, height);
+	}
 }
 
 static void
@@ -79,18 +369,20 @@ nvfx_clear_depth_stencil(struct pipe_context *pipe,
 			 unsigned dstx, unsigned dsty,
 			 unsigned width, unsigned height)
 {
-	struct nvfx_context *nvfx = nvfx_context(pipe);
-	struct nv04_surface_2d *eng2d = nvfx->screen->eng2d;
-
-	eng2d->fill(eng2d, dst, dstx, dsty, width, height,
-		    util_pack_z_stencil(dst->format, depth, stencil));
+	if(util_format_get_blocksizebits(dst->format) > 32
+		|| nvfx_surface_fill(pipe, dst, dstx, dsty, width, height, util_pack_z_stencil(dst->format, depth, stencil)))
+	{
+		// TODO: probably should use hardware clear here instead if possible
+		struct blitter_context* blitter = nvfx_get_blitter(pipe, 0);
+		util_blitter_clear_depth_stencil(blitter, dst, clear_flags, depth, stencil, dstx, dsty, width, height);
+	}
 }
 
 
 void
 nvfx_init_surface_functions(struct nvfx_context *nvfx)
 {
-	nvfx->pipe.resource_copy_region = nvfx_surface_copy;
+	nvfx->pipe.resource_copy_region = nvfx_resource_copy_region;
 	nvfx->pipe.clear_render_target = nvfx_clear_render_target;
 	nvfx->pipe.clear_depth_stencil = nvfx_clear_depth_stencil;
 }