From: Luca Barbieri Date: Tue, 3 Aug 2010 04:24:22 +0000 (+0200) Subject: nvfx: new 2D: use new 2D engine in Gallium X-Git-Url: https://git.libre-soc.org/?a=commitdiff_plain;h=9ed0686e8e6157ce38c9fa284cd7399289a2e698;p=mesa.git 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. --- diff --git a/src/gallium/drivers/nvfx/Makefile b/src/gallium/drivers/nvfx/Makefile index 6536343e440..2834f8984c7 100644 --- a/src/gallium/drivers/nvfx/Makefile +++ b/src/gallium/drivers/nvfx/Makefile @@ -4,7 +4,6 @@ include $(TOP)/configs/current LIBNAME = nvfx C_SOURCES = \ - nv04_surface_2d.c \ nv04_2d.c \ nvfx_buffer.c \ nvfx_context.c \ diff --git a/src/gallium/drivers/nvfx/nv04_surface_2d.c b/src/gallium/drivers/nvfx/nv04_surface_2d.c deleted file mode 100644 index cd0f4ce4c93..00000000000 --- a/src/gallium/drivers/nvfx/nv04_surface_2d.c +++ /dev/null @@ -1,533 +0,0 @@ -#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 "nouveau/nouveau_winsys.h" -#include "nouveau/nouveau_util.h" -#include "nouveau/nouveau_screen.h" -#include "nv04_surface_2d.h" -#include "nvfx_resource.h" - -static INLINE int -nv04_surface_format(enum pipe_format format) -{ - switch (format) { - case PIPE_FORMAT_A8_UNORM: - case PIPE_FORMAT_L8_UNORM: - case PIPE_FORMAT_I8_UNORM: - return NV04_CONTEXT_SURFACES_2D_FORMAT_Y8; - case PIPE_FORMAT_R16_SNORM: - case PIPE_FORMAT_B5G6R5_UNORM: - case PIPE_FORMAT_Z16_UNORM: - case PIPE_FORMAT_L8A8_UNORM: - return NV04_CONTEXT_SURFACES_2D_FORMAT_R5G6B5; - case PIPE_FORMAT_B8G8R8X8_UNORM: - case PIPE_FORMAT_B8G8R8A8_UNORM: - return NV04_CONTEXT_SURFACES_2D_FORMAT_A8R8G8B8; - case PIPE_FORMAT_S8_USCALED_Z24_UNORM: - case PIPE_FORMAT_X8Z24_UNORM: - return NV04_CONTEXT_SURFACES_2D_FORMAT_Y32; - default: - return -1; - } -} - -static INLINE int -nv04_rect_format(enum pipe_format format) -{ - switch (format) { - case PIPE_FORMAT_A8_UNORM: - return NV04_GDI_RECTANGLE_TEXT_COLOR_FORMAT_A8R8G8B8; - case PIPE_FORMAT_B5G6R5_UNORM: - case PIPE_FORMAT_L8A8_UNORM: - case PIPE_FORMAT_Z16_UNORM: - return NV04_GDI_RECTANGLE_TEXT_COLOR_FORMAT_A16R5G6B5; - case PIPE_FORMAT_B8G8R8X8_UNORM: - case PIPE_FORMAT_B8G8R8A8_UNORM: - case PIPE_FORMAT_S8_USCALED_Z24_UNORM: - case PIPE_FORMAT_X8Z24_UNORM: - return NV04_GDI_RECTANGLE_TEXT_COLOR_FORMAT_A8R8G8B8; - default: - return -1; - } -} - -static INLINE int -nv04_scaled_image_format(enum pipe_format format) -{ - switch (format) { - case PIPE_FORMAT_A8_UNORM: - case PIPE_FORMAT_L8_UNORM: - case PIPE_FORMAT_I8_UNORM: - return NV03_SCALED_IMAGE_FROM_MEMORY_COLOR_FORMAT_Y8; - case PIPE_FORMAT_B5G5R5A1_UNORM: - return NV03_SCALED_IMAGE_FROM_MEMORY_COLOR_FORMAT_A1R5G5B5; - case PIPE_FORMAT_B8G8R8A8_UNORM: - return NV03_SCALED_IMAGE_FROM_MEMORY_COLOR_FORMAT_A8R8G8B8; - case PIPE_FORMAT_B8G8R8X8_UNORM: - return NV03_SCALED_IMAGE_FROM_MEMORY_COLOR_FORMAT_X8R8G8B8; - case PIPE_FORMAT_B5G6R5_UNORM: - case PIPE_FORMAT_R16_SNORM: - case PIPE_FORMAT_L8A8_UNORM: - return NV03_SCALED_IMAGE_FROM_MEMORY_COLOR_FORMAT_R5G6B5; - default: - return -1; - } -} - -static INLINE unsigned -nv04_swizzle_bits_square(unsigned x, unsigned y) -{ - unsigned u = (x & 0x001) << 0 | - (x & 0x002) << 1 | - (x & 0x004) << 2 | - (x & 0x008) << 3 | - (x & 0x010) << 4 | - (x & 0x020) << 5 | - (x & 0x040) << 6 | - (x & 0x080) << 7 | - (x & 0x100) << 8 | - (x & 0x200) << 9 | - (x & 0x400) << 10 | - (x & 0x800) << 11; - - unsigned v = (y & 0x001) << 1 | - (y & 0x002) << 2 | - (y & 0x004) << 3 | - (y & 0x008) << 4 | - (y & 0x010) << 5 | - (y & 0x020) << 6 | - (y & 0x040) << 7 | - (y & 0x080) << 8 | - (y & 0x100) << 9 | - (y & 0x200) << 10 | - (y & 0x400) << 11 | - (y & 0x800) << 12; - return v | u; -} - -/* rectangular swizzled textures are linear concatenations of swizzled square tiles */ -static INLINE unsigned -nv04_swizzle_bits(unsigned x, unsigned y, unsigned w, unsigned h) -{ - unsigned s = MIN2(w, h); - unsigned m = s - 1; - return (((x | y) & ~m) * s) | nv04_swizzle_bits_square(x & m, y & m); -} - -static int -nv04_surface_copy_swizzle(struct nv04_surface_2d *ctx, - struct pipe_surface *dst, int dx, int dy, - struct pipe_surface *src, int sx, int sy, - int w, int h) -{ - struct nouveau_channel *chan = ctx->swzsurf->channel; - struct nouveau_grobj *swzsurf = ctx->swzsurf; - struct nouveau_grobj *sifm = ctx->sifm; - struct nouveau_bo *src_bo = ctx->buf(src); - struct nouveau_bo *dst_bo = ctx->buf(dst); - const unsigned src_pitch = ((struct nv04_surface *)src)->pitch; - /* Max width & height may not be the same on all HW, but must be POT */ - const unsigned max_w = 1024; - const unsigned max_h = 1024; - unsigned sub_w = w > max_w ? max_w : w; - unsigned sub_h = h > max_h ? max_h : h; - unsigned x; - unsigned y; - - /* Swizzled surfaces must be POT */ - assert(util_is_pot(dst->width) && util_is_pot(dst->height)); - - /* If area is too large to copy in one shot we must copy it in POT chunks to meet alignment requirements */ - assert(sub_w == w || util_is_pot(sub_w)); - assert(sub_h == h || util_is_pot(sub_h)); - - MARK_RING (chan, 8 + ((w+sub_w)/sub_w)*((h+sub_h)/sub_h)*17, 2 + - ((w+sub_w)/sub_w)*((h+sub_h)/sub_h)*2); - - BEGIN_RING(chan, swzsurf, NV04_SWIZZLED_SURFACE_DMA_IMAGE, 1); - OUT_RELOCo(chan, dst_bo, - NOUVEAU_BO_VRAM | NOUVEAU_BO_WR); - - BEGIN_RING(chan, swzsurf, NV04_SWIZZLED_SURFACE_FORMAT, 1); - OUT_RING (chan, nv04_surface_format(dst->format) | - log2i(dst->width) << NV04_SWIZZLED_SURFACE_FORMAT_BASE_SIZE_U_SHIFT | - log2i(dst->height) << NV04_SWIZZLED_SURFACE_FORMAT_BASE_SIZE_V_SHIFT); - - BEGIN_RING(chan, sifm, NV03_SCALED_IMAGE_FROM_MEMORY_DMA_IMAGE, 1); - OUT_RELOCo(chan, src_bo, - NOUVEAU_BO_GART | NOUVEAU_BO_VRAM | NOUVEAU_BO_RD); - BEGIN_RING(chan, sifm, NV04_SCALED_IMAGE_FROM_MEMORY_SURFACE, 1); - OUT_RING (chan, swzsurf->handle); - - for (y = 0; y < h; y += sub_h) { - sub_h = MIN2(sub_h, h - y); - - for (x = 0; x < w; x += sub_w) { - sub_w = MIN2(sub_w, w - x); - - assert(!(dst->offset & 63)); - - BEGIN_RING(chan, swzsurf, NV04_SWIZZLED_SURFACE_OFFSET, 1); - OUT_RELOCl(chan, dst_bo, dst->offset, - NOUVEAU_BO_VRAM | NOUVEAU_BO_WR); - - BEGIN_RING(chan, sifm, NV05_SCALED_IMAGE_FROM_MEMORY_COLOR_CONVERSION, 9); - OUT_RING (chan, NV05_SCALED_IMAGE_FROM_MEMORY_COLOR_CONVERSION_TRUNCATE); - OUT_RING (chan, nv04_scaled_image_format(src->format)); - OUT_RING (chan, NV03_SCALED_IMAGE_FROM_MEMORY_OPERATION_SRCCOPY); - OUT_RING (chan, (x + dx) | ((y + dy) << NV03_SCALED_IMAGE_FROM_MEMORY_CLIP_POINT_Y_SHIFT)); - OUT_RING (chan, sub_h << NV03_SCALED_IMAGE_FROM_MEMORY_CLIP_SIZE_H_SHIFT | sub_w); - OUT_RING (chan, (x + dx) | ((y + dy) << NV03_SCALED_IMAGE_FROM_MEMORY_OUT_POINT_Y_SHIFT)); - OUT_RING (chan, sub_h << NV03_SCALED_IMAGE_FROM_MEMORY_OUT_SIZE_H_SHIFT | sub_w); - OUT_RING (chan, 1 << 20); - OUT_RING (chan, 1 << 20); - - BEGIN_RING(chan, sifm, NV03_SCALED_IMAGE_FROM_MEMORY_SIZE, 4); - OUT_RING (chan, sub_h << NV03_SCALED_IMAGE_FROM_MEMORY_SIZE_H_SHIFT | sub_w); - OUT_RING (chan, src_pitch | - NV03_SCALED_IMAGE_FROM_MEMORY_FORMAT_ORIGIN_CENTER | - NV03_SCALED_IMAGE_FROM_MEMORY_FORMAT_FILTER_POINT_SAMPLE); - OUT_RELOCl(chan, src_bo, src->offset + (sy+y) * src_pitch + (sx+x) * util_format_get_blocksize(src->texture->format), - NOUVEAU_BO_GART | NOUVEAU_BO_VRAM | NOUVEAU_BO_RD); - OUT_RING (chan, 0); - } - } - - return 0; -} - -static int -nv04_surface_copy_m2mf(struct nv04_surface_2d *ctx, - struct pipe_surface *dst, int dx, int dy, - struct pipe_surface *src, int sx, int sy, int w, int h) -{ - struct nouveau_channel *chan = ctx->m2mf->channel; - struct nouveau_grobj *m2mf = ctx->m2mf; - struct nouveau_bo *src_bo = ctx->buf(src); - struct nouveau_bo *dst_bo = ctx->buf(dst); - unsigned src_pitch = ((struct nv04_surface *)src)->pitch; - unsigned dst_pitch = ((struct nv04_surface *)dst)->pitch; - unsigned dst_offset = dst->offset + dy * dst_pitch + - dx * util_format_get_blocksize(dst->texture->format); - unsigned src_offset = src->offset + sy * src_pitch + - sx * util_format_get_blocksize(src->texture->format); - - MARK_RING (chan, 3 + ((h / 2047) + 1) * 9, 2 + ((h / 2047) + 1) * 2); - BEGIN_RING(chan, m2mf, NV04_MEMORY_TO_MEMORY_FORMAT_DMA_BUFFER_IN, 2); - OUT_RELOCo(chan, src_bo, - NOUVEAU_BO_GART | NOUVEAU_BO_VRAM | NOUVEAU_BO_RD); - OUT_RELOCo(chan, dst_bo, - NOUVEAU_BO_GART | NOUVEAU_BO_VRAM | NOUVEAU_BO_WR); - - while (h) { - int count = (h > 2047) ? 2047 : h; - - BEGIN_RING(chan, m2mf, NV04_MEMORY_TO_MEMORY_FORMAT_OFFSET_IN, 8); - OUT_RELOCl(chan, src_bo, src_offset, - NOUVEAU_BO_VRAM | NOUVEAU_BO_GART | NOUVEAU_BO_RD); - OUT_RELOCl(chan, dst_bo, dst_offset, - NOUVEAU_BO_VRAM | NOUVEAU_BO_GART | NOUVEAU_BO_WR); - OUT_RING (chan, src_pitch); - OUT_RING (chan, dst_pitch); - OUT_RING (chan, w * util_format_get_blocksize(src->texture->format)); - OUT_RING (chan, count); - OUT_RING (chan, 0x0101); - OUT_RING (chan, 0); - - h -= count; - src_offset += src_pitch * count; - dst_offset += dst_pitch * count; - } - - return 0; -} - -static int -nv04_surface_copy_blit(struct nv04_surface_2d *ctx, struct pipe_surface *dst, - int dx, int dy, struct pipe_surface *src, int sx, int sy, - int w, int h) -{ - struct nouveau_channel *chan = ctx->surf2d->channel; - struct nouveau_grobj *surf2d = ctx->surf2d; - struct nouveau_grobj *blit = ctx->blit; - struct nouveau_bo *src_bo = ctx->buf(src); - struct nouveau_bo *dst_bo = ctx->buf(dst); - unsigned src_pitch = ((struct nv04_surface *)src)->pitch; - unsigned dst_pitch = ((struct nv04_surface *)dst)->pitch; - int format; - - format = nv04_surface_format(dst->format); - if (format < 0) - return 1; - - MARK_RING (chan, 12, 4); - BEGIN_RING(chan, surf2d, NV04_CONTEXT_SURFACES_2D_DMA_IMAGE_SOURCE, 2); - OUT_RELOCo(chan, src_bo, NOUVEAU_BO_VRAM | NOUVEAU_BO_RD); - OUT_RELOCo(chan, dst_bo, NOUVEAU_BO_VRAM | NOUVEAU_BO_WR); - BEGIN_RING(chan, surf2d, NV04_CONTEXT_SURFACES_2D_FORMAT, 4); - OUT_RING (chan, format); - OUT_RING (chan, (dst_pitch << 16) | src_pitch); - OUT_RELOCl(chan, src_bo, src->offset, NOUVEAU_BO_VRAM | NOUVEAU_BO_RD); - OUT_RELOCl(chan, dst_bo, dst->offset, NOUVEAU_BO_VRAM | NOUVEAU_BO_WR); - - BEGIN_RING(chan, blit, 0x0300, 3); - OUT_RING (chan, (sy << 16) | sx); - OUT_RING (chan, (dy << 16) | dx); - OUT_RING (chan, ( h << 16) | w); - - return 0; -} - -static void -nv04_surface_copy(struct nv04_surface_2d *ctx, struct pipe_surface *dst, - int dx, int dy, struct pipe_surface *src, int sx, int sy, - int w, int h) -{ - int src_linear = src->texture->flags & NVFX_RESOURCE_FLAG_LINEAR; - int dst_linear = dst->texture->flags & NVFX_RESOURCE_FLAG_LINEAR; - - assert(src->format == dst->format); - - /* Setup transfer to swizzle the texture to vram if needed */ - if (src_linear && !dst_linear && w > 1 && h > 1) { - nv04_surface_copy_swizzle(ctx, dst, dx, dy, src, sx, sy, w, h); - return; - } - - /* Use M2MF instead of the blitter since it always works - * Any possible performance drop is likely to be not very significant - * and dwarfed anyway by the current buffer management problems - */ - nv04_surface_copy_m2mf(ctx, dst, dx, dy, src, sx, sy, w, h); -} - -static void -nv04_surface_fill(struct nv04_surface_2d *ctx, struct pipe_surface *dst, - int dx, int dy, int w, int h, unsigned value) -{ - struct nouveau_channel *chan = ctx->surf2d->channel; - struct nouveau_grobj *surf2d = ctx->surf2d; - struct nouveau_grobj *rect = ctx->rect; - struct nouveau_bo *dst_bo = ctx->buf(dst); - unsigned dst_pitch = ((struct nv04_surface *)dst)->pitch; - int cs2d_format, gdirect_format; - - cs2d_format = nv04_surface_format(dst->format); - assert(cs2d_format >= 0); - - gdirect_format = nv04_rect_format(dst->format); - assert(gdirect_format >= 0); - - MARK_RING (chan, 16, 4); - BEGIN_RING(chan, surf2d, NV04_CONTEXT_SURFACES_2D_DMA_IMAGE_SOURCE, 2); - OUT_RELOCo(chan, dst_bo, NOUVEAU_BO_VRAM | NOUVEAU_BO_WR); - OUT_RELOCo(chan, dst_bo, NOUVEAU_BO_VRAM | NOUVEAU_BO_WR); - BEGIN_RING(chan, surf2d, NV04_CONTEXT_SURFACES_2D_FORMAT, 4); - OUT_RING (chan, cs2d_format); - OUT_RING (chan, (dst_pitch << 16) | dst_pitch); - OUT_RELOCl(chan, dst_bo, dst->offset, NOUVEAU_BO_VRAM | NOUVEAU_BO_WR); - OUT_RELOCl(chan, dst_bo, dst->offset, NOUVEAU_BO_VRAM | NOUVEAU_BO_WR); - - BEGIN_RING(chan, rect, NV04_GDI_RECTANGLE_TEXT_COLOR_FORMAT, 1); - OUT_RING (chan, gdirect_format); - BEGIN_RING(chan, rect, NV04_GDI_RECTANGLE_TEXT_COLOR1_A, 1); - OUT_RING (chan, value); - BEGIN_RING(chan, rect, - NV04_GDI_RECTANGLE_TEXT_UNCLIPPED_RECTANGLE_POINT(0), 2); - OUT_RING (chan, (dx << 16) | dy); - OUT_RING (chan, ( w << 16) | h); -} - -void -nv04_surface_2d_takedown(struct nv04_surface_2d **pctx) -{ - struct nv04_surface_2d *ctx; - - if (!pctx || !*pctx) - return; - ctx = *pctx; - *pctx = NULL; - - nouveau_notifier_free(&ctx->ntfy); - nouveau_grobj_free(&ctx->m2mf); - nouveau_grobj_free(&ctx->surf2d); - nouveau_grobj_free(&ctx->swzsurf); - nouveau_grobj_free(&ctx->rect); - nouveau_grobj_free(&ctx->blit); - nouveau_grobj_free(&ctx->sifm); - - FREE(ctx); -} - -struct nv04_surface_2d * -nv04_surface_2d_init(struct nouveau_screen *screen) -{ - struct nv04_surface_2d *ctx = CALLOC_STRUCT(nv04_surface_2d); - struct nouveau_channel *chan = screen->channel; - unsigned handle = 0x88000000, class; - int ret; - - if (!ctx) - return NULL; - - ret = nouveau_notifier_alloc(chan, handle++, 1, &ctx->ntfy); - if (ret) { - nv04_surface_2d_takedown(&ctx); - return NULL; - } - - ret = nouveau_grobj_alloc(chan, handle++, 0x0039, &ctx->m2mf); - if (ret) { - nv04_surface_2d_takedown(&ctx); - return NULL; - } - - BEGIN_RING(chan, ctx->m2mf, NV04_MEMORY_TO_MEMORY_FORMAT_DMA_NOTIFY, 1); - OUT_RING (chan, ctx->ntfy->handle); - - if (chan->device->chipset < 0x10) - class = NV04_CONTEXT_SURFACES_2D; - else - class = NV10_CONTEXT_SURFACES_2D; - - ret = nouveau_grobj_alloc(chan, handle++, class, &ctx->surf2d); - if (ret) { - nv04_surface_2d_takedown(&ctx); - return NULL; - } - - BEGIN_RING(chan, ctx->surf2d, - NV04_CONTEXT_SURFACES_2D_DMA_IMAGE_SOURCE, 2); - OUT_RING (chan, chan->vram->handle); - OUT_RING (chan, chan->vram->handle); - - if (chan->device->chipset < 0x10) - class = NV04_IMAGE_BLIT; - else - class = NV12_IMAGE_BLIT; - - ret = nouveau_grobj_alloc(chan, handle++, class, &ctx->blit); - if (ret) { - nv04_surface_2d_takedown(&ctx); - return NULL; - } - - BEGIN_RING(chan, ctx->blit, NV01_IMAGE_BLIT_DMA_NOTIFY, 1); - OUT_RING (chan, ctx->ntfy->handle); - BEGIN_RING(chan, ctx->blit, NV04_IMAGE_BLIT_SURFACE, 1); - OUT_RING (chan, ctx->surf2d->handle); - BEGIN_RING(chan, ctx->blit, NV01_IMAGE_BLIT_OPERATION, 1); - OUT_RING (chan, NV01_IMAGE_BLIT_OPERATION_SRCCOPY); - - ret = nouveau_grobj_alloc(chan, handle++, NV04_GDI_RECTANGLE_TEXT, - &ctx->rect); - if (ret) { - nv04_surface_2d_takedown(&ctx); - return NULL; - } - - BEGIN_RING(chan, ctx->rect, NV04_GDI_RECTANGLE_TEXT_DMA_NOTIFY, 1); - OUT_RING (chan, ctx->ntfy->handle); - BEGIN_RING(chan, ctx->rect, NV04_GDI_RECTANGLE_TEXT_SURFACE, 1); - OUT_RING (chan, ctx->surf2d->handle); - BEGIN_RING(chan, ctx->rect, NV04_GDI_RECTANGLE_TEXT_OPERATION, 1); - OUT_RING (chan, NV04_GDI_RECTANGLE_TEXT_OPERATION_SRCCOPY); - BEGIN_RING(chan, ctx->rect, - NV04_GDI_RECTANGLE_TEXT_MONOCHROME_FORMAT, 1); - OUT_RING (chan, NV04_GDI_RECTANGLE_TEXT_MONOCHROME_FORMAT_LE); - - switch (chan->device->chipset & 0xf0) { - case 0x00: - case 0x10: - class = NV04_SWIZZLED_SURFACE; - break; - case 0x20: - class = NV20_SWIZZLED_SURFACE; - break; - case 0x30: - class = NV30_SWIZZLED_SURFACE; - break; - case 0x40: - case 0x60: - class = NV40_SWIZZLED_SURFACE; - break; - default: - /* Famous last words: this really can't happen.. */ - assert(0); - break; - } - - ret = nouveau_grobj_alloc(chan, handle++, class, &ctx->swzsurf); - if (ret) { - nv04_surface_2d_takedown(&ctx); - return NULL; - } - - switch (chan->device->chipset & 0xf0) { - case 0x10: - case 0x20: - class = NV10_SCALED_IMAGE_FROM_MEMORY; - break; - case 0x30: - class = NV30_SCALED_IMAGE_FROM_MEMORY; - break; - case 0x40: - case 0x60: - class = NV40_SCALED_IMAGE_FROM_MEMORY; - break; - default: - class = NV04_SCALED_IMAGE_FROM_MEMORY; - break; - } - - ret = nouveau_grobj_alloc(chan, handle++, class, &ctx->sifm); - if (ret) { - nv04_surface_2d_takedown(&ctx); - return NULL; - } - - ctx->copy = nv04_surface_copy; - ctx->fill = nv04_surface_fill; - return ctx; -} - -struct nv04_surface* -nv04_surface_wrap_for_render(struct pipe_screen *pscreen, - struct nv04_surface_2d* eng2d, struct nv04_surface* ns) -{ - struct pipe_resource templ; - struct pipe_resource* temp_tex; - struct nv04_surface* temp_ns; - int temp_flags; - - temp_flags = ns->base.usage; - - ns->base.usage = 0; - - memset(&templ, 0, sizeof(templ)); - templ.format = ns->base.texture->format; - templ.target = PIPE_TEXTURE_2D; - templ.width0 = ns->base.width; - templ.height0 = ns->base.height; - templ.depth0 = 1; - templ.last_level = 0; - - // TODO: this is probably wrong and we should specifically handle multisampling somehow once it is implemented - templ.nr_samples = ns->base.texture->nr_samples; - - templ.bind = ns->base.texture->bind | PIPE_BIND_RENDER_TARGET; - - temp_tex = pscreen->resource_create(pscreen, &templ); - temp_ns = (struct nv04_surface*)pscreen->get_tex_surface(pscreen, temp_tex, 0, 0, 0, temp_flags); - temp_ns->backing = ns; - - if(1) /* hmm */ - eng2d->copy(eng2d, &temp_ns->backing->base, - 0, 0, &ns->base, - 0, 0, ns->base.width, ns->base.height); - - return temp_ns; -} diff --git a/src/gallium/drivers/nvfx/nv04_surface_2d.h b/src/gallium/drivers/nvfx/nv04_surface_2d.h deleted file mode 100644 index b9020dbe961..00000000000 --- a/src/gallium/drivers/nvfx/nv04_surface_2d.h +++ /dev/null @@ -1,41 +0,0 @@ -#ifndef __NV04_SURFACE_2D_H__ -#define __NV04_SURFACE_2D_H__ - -#include "pipe/p_state.h" - -struct nouveau_screen; - -struct nv04_surface { - struct pipe_surface base; - unsigned pitch; - struct nv04_surface* backing; -}; - -struct nv04_surface_2d { - struct nouveau_notifier *ntfy; - struct nouveau_grobj *surf2d; - struct nouveau_grobj *swzsurf; - struct nouveau_grobj *m2mf; - struct nouveau_grobj *rect; - struct nouveau_grobj *blit; - struct nouveau_grobj *sifm; - - struct nouveau_bo *(*buf)(struct pipe_surface *); - - void (*copy)(struct nv04_surface_2d *, struct pipe_surface *dst, - int dx, int dy, struct pipe_surface *src, int sx, int sy, - int w, int h); - void (*fill)(struct nv04_surface_2d *, struct pipe_surface *dst, - int dx, int dy, int w, int h, unsigned value); -}; - -struct nv04_surface_2d * -nv04_surface_2d_init(struct nouveau_screen *screen); - -void -nv04_surface_2d_takedown(struct nv04_surface_2d **); - -struct nv04_surface* -nv04_surface_wrap_for_render(struct pipe_screen *pscreen, struct nv04_surface_2d* eng2d, struct nv04_surface* ns); - -#endif diff --git a/src/gallium/drivers/nvfx/nvfx_context.h b/src/gallium/drivers/nvfx/nvfx_context.h index 89f94c10bd1..278be94d525 100644 --- a/src/gallium/drivers/nvfx/nvfx_context.h +++ b/src/gallium/drivers/nvfx/nvfx_context.h @@ -13,6 +13,7 @@ #include "util/u_inlines.h" #include "draw/draw_vertex.h" +#include "util/u_blitter.h" #include "nouveau/nouveau_winsys.h" #include "nouveau/nouveau_gldefs.h" @@ -88,6 +89,7 @@ struct nvfx_context { unsigned is_nv4x; /* either 0 or ~0 */ struct draw_context *draw; + struct blitter_context* blitter; /* HW state derived from pipe states */ struct nvfx_state state; diff --git a/src/gallium/drivers/nvfx/nvfx_miptree.c b/src/gallium/drivers/nvfx/nvfx_miptree.c index 27bfa24b282..b8ec726624d 100644 --- a/src/gallium/drivers/nvfx/nvfx_miptree.c +++ b/src/gallium/drivers/nvfx/nvfx_miptree.c @@ -8,8 +8,7 @@ #include "state_tracker/drm_driver.h" #include "nouveau/nouveau_winsys.h" #include "nouveau/nouveau_screen.h" -#include "nv04_surface_2d.h" -#include "nvfx_context.h" +#include "nvfx_screen.h" #include "nvfx_resource.h" #include "nvfx_transfer.h" @@ -231,9 +230,9 @@ nvfx_miptree_surface_new(struct pipe_screen *pscreen, struct pipe_resource *pt, unsigned face, unsigned level, unsigned zslice, unsigned flags) { - struct nv04_surface *ns; + struct nvfx_surface *ns; - ns = CALLOC_STRUCT(nv04_surface); + ns = CALLOC_STRUCT(nvfx_surface); if (!ns) return NULL; pipe_resource_reference(&ns->base.texture, pt); @@ -254,15 +253,6 @@ nvfx_miptree_surface_new(struct pipe_screen *pscreen, struct pipe_resource *pt, void nvfx_miptree_surface_del(struct pipe_surface *ps) { - struct nv04_surface* ns = (struct nv04_surface*)ps; - if(ns->backing) - { - struct nvfx_screen* screen = (struct nvfx_screen*)ps->texture->screen; - if(1 /*ns->backing->base.usage & PIPE_BIND_BLIT_DESTINATION*/) - screen->eng2d->copy(screen->eng2d, &ns->backing->base, 0, 0, ps, 0, 0, ns->base.width, ns->base.height); - nvfx_miptree_surface_del(&ns->backing->base); - } - pipe_resource_reference(&ps->texture, NULL); FREE(ps); } diff --git a/src/gallium/drivers/nvfx/nvfx_resource.h b/src/gallium/drivers/nvfx/nvfx_resource.h index 0e24ec2f1f9..42d04ebb370 100644 --- a/src/gallium/drivers/nvfx/nvfx_resource.h +++ b/src/gallium/drivers/nvfx/nvfx_resource.h @@ -46,6 +46,11 @@ struct nvfx_miptree { unsigned level_offset[NVFX_MAX_TEXTURE_LEVELS]; }; +struct nvfx_surface { + struct pipe_surface base; + unsigned pitch; +}; + static INLINE struct nvfx_resource *nvfx_resource(struct pipe_resource *resource) { diff --git a/src/gallium/drivers/nvfx/nvfx_screen.c b/src/gallium/drivers/nvfx/nvfx_screen.c index d354bd166a7..a1b7c218f11 100644 --- a/src/gallium/drivers/nvfx/nvfx_screen.c +++ b/src/gallium/drivers/nvfx/nvfx_screen.c @@ -233,7 +233,6 @@ nvfx_screen_surface_format_supported(struct pipe_screen *pscreen, return FALSE; } - static void nvfx_screen_destroy(struct pipe_screen *pscreen) { @@ -245,7 +244,7 @@ nvfx_screen_destroy(struct pipe_screen *pscreen) nouveau_notifier_free(&screen->query); nouveau_notifier_free(&screen->sync); nouveau_grobj_free(&screen->eng3d); - nv04_surface_2d_takedown(&screen->eng2d); + nvfx_screen_surface_takedown(pscreen); nouveau_screen_fini(&screen->base); @@ -451,8 +450,7 @@ nvfx_screen_create(struct pipe_winsys *ws, struct nouveau_device *dev) } /* 2D engine setup */ - screen->eng2d = nv04_surface_2d_init(&screen->base); - screen->eng2d->buf = nvfx_surface_buffer; + nvfx_screen_surface_init(pscreen); /* Notifier for sync purposes */ ret = nouveau_notifier_alloc(chan, 0xbeef0301, 1, &screen->sync); diff --git a/src/gallium/drivers/nvfx/nvfx_screen.h b/src/gallium/drivers/nvfx/nvfx_screen.h index 5e1c3945aef..4dedbe9cb40 100644 --- a/src/gallium/drivers/nvfx/nvfx_screen.h +++ b/src/gallium/drivers/nvfx/nvfx_screen.h @@ -3,9 +3,9 @@ #include "util/u_double_list.h" #include "nouveau/nouveau_screen.h" -#include "nv04_surface_2d.h" +#include "nvfx_context.h" -struct nvfx_context; +struct nv04_2d_context; struct nvfx_screen { struct nouveau_screen base; @@ -20,7 +20,6 @@ struct nvfx_screen { unsigned index_buffer_reloc_flags; /* HW graphics objects */ - struct nv04_surface_2d *eng2d; struct nouveau_grobj *eng3d; struct nouveau_notifier *sync; @@ -32,6 +31,8 @@ struct nvfx_screen { /* Vtxprog resources */ struct nouveau_resource *vp_exec_heap; struct nouveau_resource *vp_data_heap; + + struct nv04_2d_context* eng2d; }; static INLINE struct nvfx_screen * @@ -40,4 +41,7 @@ nvfx_screen(struct pipe_screen *screen) return (struct nvfx_screen *)screen; } +int nvfx_screen_surface_init(struct pipe_screen *pscreen); +void nvfx_screen_surface_takedown(struct pipe_screen *pscreen); + #endif diff --git a/src/gallium/drivers/nvfx/nvfx_state_fb.c b/src/gallium/drivers/nvfx/nvfx_state_fb.c index 657e315f067..e111d11627f 100644 --- a/src/gallium/drivers/nvfx/nvfx_state_fb.c +++ b/src/gallium/drivers/nvfx/nvfx_state_fb.c @@ -2,8 +2,6 @@ #include "nvfx_resource.h" #include "nouveau/nouveau_util.h" - - void nvfx_state_framebuffer_validate(struct nvfx_context *nvfx) { @@ -31,7 +29,7 @@ nvfx_state_framebuffer_validate(struct nvfx_context *nvfx) rt_enable |= (NV34TCL_RT_ENABLE_COLOR0 << i); nvfx->hw_rt[i].bo = ((struct nvfx_miptree*)fb->cbufs[i]->texture)->base.bo; nvfx->hw_rt[i].offset = fb->cbufs[i]->offset; - nvfx->hw_rt[i].pitch = ((struct nv04_surface *)fb->cbufs[i])->pitch; + nvfx->hw_rt[i].pitch = ((struct nvfx_surface *)fb->cbufs[i])->pitch; } for(; i < 4; ++i) nvfx->hw_rt[i].bo = 0; @@ -44,7 +42,7 @@ nvfx_state_framebuffer_validate(struct nvfx_context *nvfx) zeta_format = fb->zsbuf->format; nvfx->hw_zeta.bo = ((struct nvfx_miptree*)fb->zsbuf->texture)->base.bo; nvfx->hw_zeta.offset = fb->zsbuf->offset; - nvfx->hw_zeta.pitch = ((struct nv04_surface *)fb->zsbuf)->pitch; + nvfx->hw_zeta.pitch = ((struct nvfx_surface *)fb->zsbuf)->pitch; } else nvfx->hw_zeta.bo = 0; 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 + +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; }