fixed subject line
On Thu, Nov 14, 2013 at 4:15 PM, Mark Mueller <markkmuel...@gmail.com>wrote:
> This patch consolidates the decision about formats that blorp_blt does and
> does not support to blorp instead of leaving that decision to callers. This
> opens the door to adding more functionality to blorp, including support for
> using GPU acceleration of processing and loading textures.
>
> This patch fixes piglit generated_tests/spec/glsl-1.50/compiler/-
> built-in-functions/op-div-mat4x3-float.geom.
>
>
> Signed-off-by: Mark Mueller <markkmuel...@gmail.com>
> ---
> src/mesa/drivers/dri/i965/brw_blorp.h | 8 +-
> src/mesa/drivers/dri/i965/brw_blorp_blit.cpp | 149
> ++++++++++++++--------
> src/mesa/drivers/dri/i965/brw_state.h | 5 +-
> src/mesa/drivers/dri/i965/brw_surface_formats.c | 26 ++--
> src/mesa/drivers/dri/i965/brw_wm_surface_state.c | 12 +-
> src/mesa/drivers/dri/i965/gen7_wm_surface_state.c | 4 +-
> 6 files changed, 131 insertions(+), 73 deletions(-)
>
> diff --git a/src/mesa/drivers/dri/i965/brw_blorp.h
> b/src/mesa/drivers/dri/i965/brw_blorp.h
> index 85bf099..5344851 100644
> --- a/src/mesa/drivers/dri/i965/brw_blorp.h
> +++ b/src/mesa/drivers/dri/i965/brw_blorp.h
> @@ -34,8 +34,7 @@ struct brw_context;
> extern "C" {
> #endif
>
> -void
> -brw_blorp_blit_miptrees(struct brw_context *brw,
> +bool brw_blorp_blit_miptrees(struct brw_context *brw,
> struct intel_mipmap_tree *src_mt,
> unsigned src_level, unsigned src_layer,
> struct intel_mipmap_tree *dst_mt,
> @@ -356,8 +355,13 @@ public:
> virtual uint32_t get_wm_prog(struct brw_context *brw,
> brw_blorp_prog_data **prog_data) const;
>
> + bool is_valid() {
> + return valid_parameters;
> + }
> +
> private:
> brw_blorp_blit_prog_key wm_prog_key;
> + bool valid_parameters;
> };
>
> /**
> diff --git a/src/mesa/drivers/dri/i965/brw_blorp_blit.cpp
> b/src/mesa/drivers/dri/i965/brw_blorp_blit.cpp
> index d54b926..00461a4 100644
> --- a/src/mesa/drivers/dri/i965/brw_blorp_blit.cpp
> +++ b/src/mesa/drivers/dri/i965/brw_blorp_blit.cpp
> @@ -123,7 +123,7 @@ find_miptree(GLbitfield buffer_bit, struct
> intel_renderbuffer *irb)
> return mt;
> }
>
> -void
> +bool
> brw_blorp_blit_miptrees(struct brw_context *brw,
> struct intel_mipmap_tree *src_mt,
> unsigned src_level, unsigned src_layer,
> @@ -135,16 +135,6 @@ brw_blorp_blit_miptrees(struct brw_context *brw,
> float dst_x1, float dst_y1,
> GLenum filter, bool mirror_x, bool mirror_y)
> {
> - /* Get ready to blit. This includes depth resolving the src and dst
> - * buffers if necessary. Note: it's not necessary to do a color
> resolve on
> - * the destination buffer because we use the standard render path to
> render
> - * to destination color buffers, and the standard render path is
> - * fast-color-aware.
> - */
> - intel_miptree_resolve_color(brw, src_mt);
> - intel_miptree_slice_resolve_depth(brw, src_mt, src_level, src_layer);
> - intel_miptree_slice_resolve_depth(brw, dst_mt, dst_level, dst_layer);
> -
> DBG("%s from %s mt %p %d %d (%f,%f) (%f,%f)"
> "to %s mt %p %d %d (%f,%f) (%f,%f) (flip %d,%d)\n",
> __FUNCTION__,
> @@ -162,12 +152,28 @@ brw_blorp_blit_miptrees(struct brw_context *brw,
> dst_x0, dst_y0,
> dst_x1, dst_y1,
> filter, mirror_x, mirror_y);
> +
> + if (!params.is_valid()) {
> + return false;
> + }
> +
> + /* Get ready to blit. This includes depth resolving the src and dst
> + * buffers if necessary. Note: it's not necessary to do a color
> resolve on
> + * the destination buffer because we use the standard render path to
> render
> + * to destination color buffers, and the standard render path is
> + * fast-color-aware.
> + */
> + intel_miptree_resolve_color(brw, src_mt);
> + intel_miptree_slice_resolve_depth(brw, src_mt, src_level, src_layer);
> + intel_miptree_slice_resolve_depth(brw, dst_mt, dst_level, dst_layer);
> +
> brw_blorp_exec(brw, ¶ms);
>
> intel_miptree_slice_set_needs_hiz_resolve(dst_mt, dst_level,
> dst_layer);
> + return true;
> }
>
> -static void
> +static bool
> do_blorp_blit(struct brw_context *brw, GLbitfield buffer_bit,
> struct intel_renderbuffer *src_irb,
> struct intel_renderbuffer *dst_irb,
> @@ -180,14 +186,17 @@ do_blorp_blit(struct brw_context *brw, GLbitfield
> buffer_bit,
> struct intel_mipmap_tree *dst_mt = find_miptree(buffer_bit, dst_irb);
>
> /* Do the blit */
> - brw_blorp_blit_miptrees(brw,
> - src_mt, src_irb->mt_level, src_irb->mt_layer,
> - dst_mt, dst_irb->mt_level, dst_irb->mt_layer,
> - srcX0, srcY0, srcX1, srcY1,
> - dstX0, dstY0, dstX1, dstY1,
> - filter, mirror_x, mirror_y);
> + if (!brw_blorp_blit_miptrees(brw,
> + src_mt, src_irb->mt_level,
> src_irb->mt_layer,
> + dst_mt, dst_irb->mt_level,
> dst_irb->mt_layer,
> + srcX0, srcY0, srcX1, srcY1,
> + dstX0, dstY0, dstX1, dstY1,
> + filter, mirror_x, mirror_y)) {
> + return false;
> + }
>
> intel_renderbuffer_set_needs_downsample(dst_irb);
> + return true;
> }
>
> static bool
> @@ -293,10 +302,13 @@ try_blorp_blit(struct brw_context *brw,
> }
> for (unsigned i = 0; i < ctx->DrawBuffer->_NumColorDrawBuffers;
> ++i) {
> dst_irb =
> intel_renderbuffer(ctx->DrawBuffer->_ColorDrawBuffers[i]);
> - if (dst_irb)
> - do_blorp_blit(brw, buffer_bit, src_irb, dst_irb, srcX0, srcY0,
> - srcX1, srcY1, dstX0, dstY0, dstX1, dstY1,
> - filter, mirror_x, mirror_y);
> + if (dst_irb) {
> + if (!do_blorp_blit(brw, buffer_bit, src_irb, dst_irb, srcX0,
> srcY0,
> + srcX1, srcY1, dstX0, dstY0, dstX1, dstY1,
> + filter, mirror_x, mirror_y)) {
> + return false;
> + }
> + }
> }
> break;
> case GL_DEPTH_BUFFER_BIT:
> @@ -306,9 +318,11 @@ try_blorp_blit(struct brw_context *brw,
>
> intel_renderbuffer(draw_fb->Attachment[BUFFER_DEPTH].Renderbuffer);
> if (!formats_match(buffer_bit, src_irb, dst_irb))
> return false;
> - do_blorp_blit(brw, buffer_bit, src_irb, dst_irb, srcX0, srcY0,
> - srcX1, srcY1, dstX0, dstY0, dstX1, dstY1,
> - filter, mirror_x, mirror_y);
> + if (!do_blorp_blit(brw, buffer_bit, src_irb, dst_irb, srcX0, srcY0,
> + srcX1, srcY1, dstX0, dstY0, dstX1, dstY1,
> + filter, mirror_x, mirror_y)) {
> + return false;
> + }
> break;
> case GL_STENCIL_BUFFER_BIT:
> src_irb =
> @@ -317,9 +331,11 @@ try_blorp_blit(struct brw_context *brw,
>
> intel_renderbuffer(draw_fb->Attachment[BUFFER_STENCIL].Renderbuffer);
> if (!formats_match(buffer_bit, src_irb, dst_irb))
> return false;
> - do_blorp_blit(brw, buffer_bit, src_irb, dst_irb, srcX0, srcY0,
> - srcX1, srcY1, dstX0, dstY0, dstX1, dstY1,
> - filter, mirror_x, mirror_y);
> + if (!do_blorp_blit(brw, buffer_bit, src_irb, dst_irb, srcX0, srcY0,
> + srcX1, srcY1, dstX0, dstY0, dstX1, dstY1,
> + filter, mirror_x, mirror_y)) {
> + return false;
> + }
> break;
> default:
> assert(false);
> @@ -349,27 +365,11 @@ brw_blorp_copytexsubimage(struct brw_context *brw,
> struct intel_mipmap_tree *src_mt = src_irb->mt;
> struct intel_mipmap_tree *dst_mt = intel_image->mt;
>
> - /* BLORP is not supported before Gen6. */
> - if (brw->gen < 6)
> - return false;
> -
> if (_mesa_get_format_base_format(src_mt->format) !=
> _mesa_get_format_base_format(dst_mt->format)) {
> return false;
> }
>
> - /* We can't handle format conversions between Z24 and other formats
> since
> - * we have to lie about the surface format. See the comments in
> - * brw_blorp_surface_info::set().
> - */
> - if ((src_mt->format == MESA_FORMAT_X8_Z24) !=
> - (dst_mt->format == MESA_FORMAT_X8_Z24)) {
> - return false;
> - }
> -
> - if (!brw->format_supported_as_render_target[dst_mt->format])
> - return false;
> -
> /* Source clipping shouldn't be necessary, since copytexsubimage (in
> * src/mesa/main/teximage.c) calls _mesa_clip_copytexsubimage() which
> * takes care of it.
> @@ -396,12 +396,14 @@ brw_blorp_copytexsubimage(struct brw_context *brw,
> mirror_y = true;
> }
>
> - brw_blorp_blit_miptrees(brw,
> - src_mt, src_irb->mt_level, src_irb->mt_layer,
> - dst_mt, dst_image->Level, dst_image->Face +
> slice,
> - srcX0, srcY0, srcX1, srcY1,
> - dstX0, dstY0, dstX1, dstY1,
> - GL_NEAREST, false, mirror_y);
> + if (!brw_blorp_blit_miptrees(brw,
> + src_mt, src_irb->mt_level,
> src_irb->mt_layer,
> + dst_mt, dst_image->Level, dst_image->Face
> + slice,
> + srcX0, srcY0, srcX1, srcY1,
> + dstX0, dstY0, dstX1, dstY1,
> + GL_NEAREST, false, mirror_y)) {
> + return false;
> + }
>
> /* If we're copying to a packed depth stencil texture and the source
> * framebuffer has separate stencil, we need to also copy the stencil
> data
> @@ -2069,10 +2071,52 @@
> brw_blorp_blit_params::brw_blorp_blit_params(struct brw_context *brw,
> GLfloat dst_x1, GLfloat
> dst_y1,
> GLenum filter,
> bool mirror_x, bool mirror_y)
> + : valid_parameters(false)
> {
> struct gl_context *ctx = &brw->ctx;
> - const struct gl_framebuffer *read_fb = ctx->ReadBuffer;
>
> + /* BLORP is not supported before Gen6. */
> + if (brw->gen < 6)
> + return;
> +
> + if (MESA_FORMAT_NONE == src_mt->format || MESA_FORMAT_COUNT <=
> src_mt->format) {
> + perf_debug("%s: Blorp doesn't yet support provided source
> format.\n", __FUNCTION__);
> + return;
> + }
> +
> + /* We can't handle format conversions between Z24 and other formats
> since
> + * we have to lie about the surface format. See the comments in
> + * brw_blorp_surface_info::set().
> + */
> + if ((src_mt->format == MESA_FORMAT_X8_Z24) !=
> + (dst_mt->format == MESA_FORMAT_X8_Z24)) {
> + return;
> + }
> +
> + /* blorp lies about some formats to placate the GPU. */
> + const GLint srcFormatTranslated = translate_tex_format(brw,
> src_mt->format, 0, true);
> + const GLint dstFormatTranslated = translate_tex_format(brw,
> dst_mt->format, 0, true);
> + if (0 == srcFormatTranslated || 0 == dstFormatTranslated) {
> + perf_debug("%s: compatible brw_surface format not found for source
> or target "
> + "texture mip tree. Source: %s Target: %s\n",
> + __FUNCTION__, _mesa_get_format_name(src_mt->format),
> + _mesa_get_format_name(dst_mt->format));
> + return;
> + }
> +
> + if (!brw->format_supported_as_render_target[dst_mt->format]) {
> + perf_debug("%s: The translated target miptree format is not
> supported as a render target: %s.\n",
> + __FUNCTION__, _mesa_get_format_name(dst_mt->format));
> + return;
> + }
> +
> + if (!brw_format_for_sampling(brw, srcFormatTranslated)) {
> + perf_debug("%s: The translated source miptree format is not
> supported for sampling: %s.\n",
> + __FUNCTION__, _mesa_get_format_name(src_mt->format));
> + return;
> + }
> +
> + const struct gl_framebuffer *read_fb = ctx->ReadBuffer;
> src.set(brw, src_mt, src_level, src_layer, false);
> dst.set(brw, dst_mt, dst_level, dst_layer, true);
>
> @@ -2123,7 +2167,7 @@ brw_blorp_blit_params::brw_blorp_blit_params(struct
> brw_context *brw,
> wm_prog_key.texture_data_type = BRW_REGISTER_TYPE_D;
> break;
> default:
> - assert(!"Unrecognized blorp format");
> + _mesa_debug(&brw->ctx, "Unrecognized blorp format");
> break;
> }
>
> @@ -2329,6 +2373,7 @@ brw_blorp_blit_params::brw_blorp_blit_params(struct
> brw_context *brw,
> src.x_offset *= 2;
> src.y_offset /= 2;
> }
> + valid_parameters = true;
> }
>
> uint32_t
> diff --git a/src/mesa/drivers/dri/i965/brw_state.h
> b/src/mesa/drivers/dri/i965/brw_state.h
> index 471f1da..49b02b6 100644
> --- a/src/mesa/drivers/dri/i965/brw_state.h
> +++ b/src/mesa/drivers/dri/i965/brw_state.h
> @@ -188,13 +188,14 @@ uint32_t brw_get_surface_tiling_bits(uint32_t
> tiling);
> uint32_t brw_get_surface_num_multisamples(unsigned num_samples);
>
> uint32_t brw_format_for_mesa_format(gl_format mesa_format);
> +bool brw_format_for_sampling(struct brw_context *brw, const unsigned
> brw_surface_sel);
>
> GLuint translate_tex_target(GLenum target);
>
> GLuint translate_tex_format(struct brw_context *brw,
> gl_format mesa_format,
> - GLenum depth_mode,
> - GLenum srgb_decode);
> + GLenum srgb_decode,
> + bool for_render);
>
> int brw_get_texture_swizzle(const struct gl_context *ctx,
> const struct gl_texture_object *t);
> diff --git a/src/mesa/drivers/dri/i965/brw_surface_formats.c
> b/src/mesa/drivers/dri/i965/brw_surface_formats.c
> index 7afca79..c9715a7 100644
> --- a/src/mesa/drivers/dri/i965/brw_surface_formats.c
> +++ b/src/mesa/drivers/dri/i965/brw_surface_formats.c
> @@ -306,6 +306,13 @@ const struct surface_format_info surface_formats[] = {
> SF( x, x, x, x, x, x, x, x, x, BRW_SURFACEFORMAT_R8G8B8_UINT)
> SF( x, x, x, x, x, x, x, x, x, BRW_SURFACEFORMAT_R8G8B8_SINT)
> };
> +
> +bool
> +brw_format_for_sampling(struct brw_context *brw, const unsigned
> brw_surface_sel)
> +{
> + return (brw_surface_sel && surface_formats[brw_surface_sel].sampling
> <= brw->gen * 10);
> +}
> +
> #undef x
> #undef Y
>
> @@ -686,24 +693,26 @@ brw_render_target_supported(struct brw_context *brw,
> GLuint
> translate_tex_format(struct brw_context *brw,
> gl_format mesa_format,
> - GLenum depth_mode,
> - GLenum srgb_decode)
> + GLenum srgb_decode,
> + bool for_render)
> {
> - struct gl_context *ctx = &brw->ctx;
> if (srgb_decode == GL_SKIP_DECODE_EXT)
> mesa_format = _mesa_get_srgb_format_linear(mesa_format);
>
> switch( mesa_format ) {
>
> + case MESA_FORMAT_S8:
> + return BRW_SURFACEFORMAT_R8_UNORM;
> +
> case MESA_FORMAT_Z16:
> - return BRW_SURFACEFORMAT_I16_UNORM;
> + return for_render ? BRW_SURFACEFORMAT_R16_UNORM :
> BRW_SURFACEFORMAT_I16_UNORM;
>
> case MESA_FORMAT_S8_Z24:
> case MESA_FORMAT_X8_Z24:
> - return BRW_SURFACEFORMAT_I24X8_UNORM;
> + return for_render ? BRW_SURFACEFORMAT_B8G8R8A8_UNORM :
> BRW_SURFACEFORMAT_I24X8_UNORM;
>
> case MESA_FORMAT_Z32_FLOAT:
> - return BRW_SURFACEFORMAT_I32_FLOAT;
> + return for_render ? BRW_SURFACEFORMAT_R32_FLOAT :
> BRW_SURFACEFORMAT_I32_FLOAT;
>
> case MESA_FORMAT_Z32_FLOAT_X24S8:
> return BRW_SURFACEFORMAT_R32G32_FLOAT;
> @@ -712,7 +721,7 @@ translate_tex_format(struct brw_context *brw,
> /* The value of this BRW_SURFACEFORMAT is 0, which tricks the
> * assertion below.
> */
> - return BRW_SURFACEFORMAT_R32G32B32A32_FLOAT;
> + return for_render ? BRW_SURFACEFORMAT_B8G8R8A8_UNORM :
> BRW_SURFACEFORMAT_R32G32B32A32_FLOAT;
>
> case MESA_FORMAT_SRGB_DXT1:
> if (brw->gen == 4 && !brw->is_g4x) {
> @@ -720,13 +729,12 @@ translate_tex_format(struct brw_context *brw,
> * skipping SRGB decode. It's not worth not supporting sRGB in
> * general to prevent this.
> */
> + struct gl_context *ctx = &brw->ctx;
> WARN_ONCE(true, "Demoting sRGB DXT1 texture to non-sRGB\n");
> mesa_format = MESA_FORMAT_RGB_DXT1;
> }
> - return brw_format_for_mesa_format(mesa_format);
>
> default:
> - assert(brw_format_for_mesa_format(mesa_format) != 0);
> return brw_format_for_mesa_format(mesa_format);
> }
> }
> diff --git a/src/mesa/drivers/dri/i965/brw_wm_surface_state.c
> b/src/mesa/drivers/dri/i965/brw_wm_surface_state.c
> index 662c975..4873353 100644
> --- a/src/mesa/drivers/dri/i965/brw_wm_surface_state.c
> +++ b/src/mesa/drivers/dri/i965/brw_wm_surface_state.c
> @@ -285,13 +285,13 @@ brw_update_texture_surface(struct gl_context *ctx,
> (void) for_gather; /* no w/a to apply for this gen */
>
> surf[0] = (translate_tex_target(tObj->Target) <<
> BRW_SURFACE_TYPE_SHIFT |
> - BRW_SURFACE_MIPMAPLAYOUT_BELOW <<
> BRW_SURFACE_MIPLAYOUT_SHIFT |
> - BRW_SURFACE_CUBEFACE_ENABLES |
> - (translate_tex_format(brw,
> + BRW_SURFACE_MIPMAPLAYOUT_BELOW <<
> BRW_SURFACE_MIPLAYOUT_SHIFT |
> + BRW_SURFACE_CUBEFACE_ENABLES |
> + (translate_tex_format(brw,
> mt->format,
> - tObj->DepthMode,
> - sampler->sRGBDecode) <<
> - BRW_SURFACE_FORMAT_SHIFT));
> + sampler->sRGBDecode,
> + false) <<
> + BRW_SURFACE_FORMAT_SHIFT));
>
> surf[1] = intelObj->mt->region->bo->offset + intelObj->mt->offset; /*
> reloc */
>
> diff --git a/src/mesa/drivers/dri/i965/gen7_wm_surface_state.c
> b/src/mesa/drivers/dri/i965/gen7_wm_surface_state.c
> index 3f4817d..13065ec 100644
> --- a/src/mesa/drivers/dri/i965/gen7_wm_surface_state.c
> +++ b/src/mesa/drivers/dri/i965/gen7_wm_surface_state.c
> @@ -291,8 +291,8 @@ gen7_update_texture_surface(struct gl_context *ctx,
>
> uint32_t tex_format = translate_tex_format(brw,
> mt->format,
> - tObj->DepthMode,
> - sampler->sRGBDecode);
> + sampler->sRGBDecode,
> + false);
>
> if (for_gather && tex_format == BRW_SURFACEFORMAT_R32G32_FLOAT)
> tex_format = BRW_SURFACEFORMAT_R32G32_FLOAT_LD;
> --
> 1.8.3.1
>
>
_______________________________________________
mesa-dev mailing list
mesa-dev@lists.freedesktop.org
http://lists.freedesktop.org/mailman/listinfo/mesa-dev
