You said earlier that pbo-readpixels-small worked with this patch on Fermi :( Oh well. <div class="gmail_extra"> <div class="gmail_quote">On Jul 22, 2016 5:39 AM, "Samuel Pitoiset" <<a href="mailto:samuel.pitoiset@gmail.com">samuel.pitoiset@gmail.com</a>> wrote: <blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex">This patch introduces the following regressions: bin/fcc-read-to-pbo-after-clear -auto bin/gl-2.1-pbo -auto -fbo bin/fbo-pbo-readpixels-small -auto bin/pbo-read-argb8888 -auto bin/pbo-readpixels-small -auto bin/mesa_pack_invert-readpixels -auto -fbo Tested on Fermi (GF119). On 07/16/2016 09:09 PM, Ilia Mirkin wrote: <blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex"> This is useful for pbo downloads, which are now accelerated with images. BGRA8 is a moderately common format to do that in. Signed-off-by: Ilia Mirkin <<a href="mailto:imirkin@alum.mit.edu" target="_blank">imirkin@alum.mit.edu</a>> --- This needs testing on SM20 and SM30. I've tested it on SM35 and bin/pbo-readpixels-small -auto worked fine. (Didn't until I properly fixed the various items.) src/gallium/drivers/nouveau/codegen/nv50_ir.cpp | 2 ++ src/gallium/drivers/nouveau/codegen/nv50_ir.h | 3 +++ src/gallium/drivers/nouveau/codegen/nv50_ir_from_tgsi.cpp | 2 ++ src/gallium/drivers/nouveau/codegen/nv50_ir_lowering_nvc0.cpp | 4 ++++ src/gallium/drivers/nouveau/nv50/g80_defs.xml.h | 1 + src/gallium/drivers/nouveau/nv50/nv50_formats.c | 3 ++- src/gallium/drivers/nouveau/nvc0/nvc0_tex.c | 2 ++ 7 files changed, 16 insertions(+), 1 deletion(-) diff --git a/src/gallium/drivers/nouveau/codegen/nv50_ir.cpp b/src/gallium/drivers/nouveau/codegen/nv50_ir.cpp index 2caebe8..179ad0b 100644 --- a/src/gallium/drivers/nouveau/codegen/nv50_ir.cpp +++ b/src/gallium/drivers/nouveau/codegen/nv50_ir.cpp @@ -1012,6 +1012,8 @@ const struct TexInstruction::ImgFormatDesc TexInstruction::formatTable[] = { "RG8_SNORM", 2, { 8, 8, 0, 0 }, SNORM }, { "R16_SNORM", 1, { 16, 0, 0, 0 }, SNORM }, { "R8_SNORM", 1, { 8, 0, 0, 0 }, SNORM }, + + { "BGRA8", 4, { 8, 8, 8, 8 }, UNORM, true }, }; void diff --git a/src/gallium/drivers/nouveau/codegen/nv50_ir.h b/src/gallium/drivers/nouveau/codegen/nv50_ir.h index 41804b6..6d2ee8b 100644 --- a/src/gallium/drivers/nouveau/codegen/nv50_ir.h +++ b/src/gallium/drivers/nouveau/codegen/nv50_ir.h @@ -412,6 +412,8 @@ enum ImgFormat FMT_R16_SNORM, FMT_R8_SNORM, + FMT_BGRA8, + IMG_FORMAT_COUNT, }; @@ -967,6 +969,7 @@ public: uint8_t components; uint8_t bits[4]; ImgType type; + bool bgra; }; static const struct ImgFormatDesc formatTable[IMG_FORMAT_COUNT]; diff --git a/src/gallium/drivers/nouveau/codegen/nv50_ir_from_tgsi.cpp b/src/gallium/drivers/nouveau/codegen/nv50_ir_from_tgsi.cpp index 61eb7f5..7dff08a 100644 --- a/src/gallium/drivers/nouveau/codegen/nv50_ir_from_tgsi.cpp +++ b/src/gallium/drivers/nouveau/codegen/nv50_ir_from_tgsi.cpp @@ -508,6 +508,8 @@ static nv50_ir::ImgFormat translateImgFormat(uint format) FMT_CASE(R8G8_SNORM, RG8_SNORM); FMT_CASE(R16_SNORM, R16_SNORM); FMT_CASE(R8_SNORM, R8_SNORM); + + FMT_CASE(B8G8R8A8_UNORM, BGRA8); } assert(!"Unexpected format"); diff --git a/src/gallium/drivers/nouveau/codegen/nv50_ir_lowering_nvc0.cpp b/src/gallium/drivers/nouveau/codegen/nv50_ir_lowering_nvc0.cpp index 18955eb..92bc0bb 100644 --- a/src/gallium/drivers/nouveau/codegen/nv50_ir_lowering_nvc0.cpp +++ b/src/gallium/drivers/nouveau/codegen/nv50_ir_lowering_nvc0.cpp @@ -1940,6 +1940,10 @@ NVC0LoweringPass::convertSurfaceFormat(TexInstruction *su) bld.mkCvt(OP_CVT, TYPE_F32, typedDst[i], TYPE_F16, typedDst[i]); } } + + if (format->bgra) { + std::swap(typedDst[0], typedDst[2]); + } } void diff --git a/src/gallium/drivers/nouveau/nv50/g80_defs.xml.h b/src/gallium/drivers/nouveau/nv50/g80_defs.xml.h index 5d40624..49bf860 100644 --- a/src/gallium/drivers/nouveau/nv50/g80_defs.xml.h +++ b/src/gallium/drivers/nouveau/nv50/g80_defs.xml.h @@ -177,6 +177,7 @@ WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. #define GK104_IMAGE_FORMAT_RG32_FLOAT 0x0000000d #define GK104_IMAGE_FORMAT_RG32_SINT 0x0000000e #define GK104_IMAGE_FORMAT_RG32_UINT 0x0000000f +#define GK104_IMAGE_FORMAT_BGRA8_UNORM 0x00000011 #define GK104_IMAGE_FORMAT_RGB10_A2_UNORM 0x00000013 #define GK104_IMAGE_FORMAT_RGB10_A2_UINT 0x00000015 #define GK104_IMAGE_FORMAT_RGBA8_UNORM 0x00000018 diff --git a/src/gallium/drivers/nouveau/nv50/nv50_formats.c b/src/gallium/drivers/nouveau/nv50/nv50_formats.c index 34d32d1..07c4419 100644 --- a/src/gallium/drivers/nouveau/nv50/nv50_formats.c +++ b/src/gallium/drivers/nouveau/nv50/nv50_formats.c @@ -50,6 +50,7 @@ #define U_IB PIPE_BIND_BLENDABLE | U_IR #define U_TD PIPE_BIND_SCANOUT | PIPE_BIND_DISPLAY_TARGET | U_TB #define U_TZ PIPE_BIND_DEPTH_STENCIL | U_T +#define U_ID U_TD | U_I #if NOUVEAU_DRIVER == 0xc0 # define U_TC U_TB # define U_IC U_IB @@ -122,7 +123,7 @@ const struct nvc0_format nvc0_format_table[PIPE_FORMAT_COUNT] = const struct nv50_format nv50_format_table[PIPE_FORMAT_COUNT] = #endif { - C4(A, B8G8R8A8_UNORM, BGRA8_UNORM, B, G, R, A, UNORM, A8B8G8R8, TD), + C4(A, B8G8R8A8_UNORM, BGRA8_UNORM, B, G, R, A, UNORM, A8B8G8R8, ID), F3(A, B8G8R8X8_UNORM, BGRX8_UNORM, B, G, R, xx, UNORM, A8B8G8R8, TD), C4(A, B8G8R8A8_SRGB, BGRA8_SRGB, B, G, R, A, UNORM, A8B8G8R8, TD), F3(A, B8G8R8X8_SRGB, BGRX8_SRGB, B, G, R, xx, UNORM, A8B8G8R8, TD), diff --git a/src/gallium/drivers/nouveau/nvc0/nvc0_tex.c b/src/gallium/drivers/nouveau/nvc0/nvc0_tex.c index 6e9e768..439f0f5 100644 --- a/src/gallium/drivers/nouveau/nvc0/nvc0_tex.c +++ b/src/gallium/drivers/nouveau/nvc0/nvc0_tex.c @@ -1162,6 +1162,7 @@ static const uint8_t nve4_su_format_map[PIPE_FORMAT_COUNT] = [PIPE_FORMAT_R16G16B16A16_SNORM] = GK104_IMAGE_FORMAT_RGBA16_SNORM, [PIPE_FORMAT_R16G16B16A16_SINT] = GK104_IMAGE_FORMAT_RGBA16_SINT, [PIPE_FORMAT_R16G16B16A16_UINT] = GK104_IMAGE_FORMAT_RGBA16_UINT, + [PIPE_FORMAT_B8G8R8A8_UNORM] = GK104_IMAGE_FORMAT_BGRA8_UNORM, [PIPE_FORMAT_R8G8B8A8_UNORM] = GK104_IMAGE_FORMAT_RGBA8_UNORM, [PIPE_FORMAT_R8G8B8A8_SNORM] = GK104_IMAGE_FORMAT_RGBA8_SNORM, [PIPE_FORMAT_R8G8B8A8_SINT] = GK104_IMAGE_FORMAT_RGBA8_SINT, @@ -1216,6 +1217,7 @@ static const uint16_t nve4_su_format_aux_map[PIPE_FORMAT_COUNT] = [PIPE_FORMAT_R10G10B10A2_UNORM] = 0x2a24, [PIPE_FORMAT_R10G10B10A2_UINT] = 0x2a24, + [PIPE_FORMAT_B8G8R8A8_UNORM] = 0x2a24, [PIPE_FORMAT_R8G8B8A8_UNORM] = 0x2a24, [PIPE_FORMAT_R8G8B8A8_SNORM] = 0x2a24, [PIPE_FORMAT_R8G8B8A8_SINT] = 0x2a24, </blockquote> </blockquote></div></div>