From a48ee529a256611676884262ced218ad6c05f961 Mon Sep 17 00:00:00 2001 From: Marc Dietrich Date: Mon, 31 Aug 2009 08:56:33 -0700 Subject: mesa: fix 32bit cross compilation on a 64bit machine When cross compiling on a 64bit machine, gen_matypes.c is build for the host machine (64bit) but must generates code for the target machine (32bit). This causes wrong offsets all over the place and crashes googleearth on my machine. Solution is to add -m32 when cross compiling. Attached patch is compatible with linux-x86-32 and autoconf based builds. --- src/mesa/x86/Makefile | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'src') diff --git a/src/mesa/x86/Makefile b/src/mesa/x86/Makefile index aa49a9134a..69122c114b 100644 --- a/src/mesa/x86/Makefile +++ b/src/mesa/x86/Makefile @@ -21,7 +21,7 @@ clean: gen_matypes: gen_matypes.c - $(HOST_CC) $(INCLUDE_DIRS) $(HOST_CFLAGS) gen_matypes.c -o gen_matypes + $(HOST_CC) $(ARCH_FLAGS) $(INCLUDE_DIRS) $(HOST_CFLAGS) gen_matypes.c -o gen_matypes # need some special rules here, unfortunately matypes.h: ../main/mtypes.h ../tnl/t_context.h gen_matypes -- cgit v1.2.3 From 2241665dc6d77a992edfc49a9d9d9ed8d1b52e60 Mon Sep 17 00:00:00 2001 From: Brian Paul Date: Mon, 31 Aug 2009 11:14:16 -0600 Subject: mesa: fix saturation logic in emit_texenv() We need to clamp/saturate after each texenv stage, not just the last one. Fixes glean texEnv failure for softpipe (and probably other fragment program- based drivers). --- src/mesa/main/texenvprogram.c | 8 +++++++- 1 file changed, 7 insertions(+), 1 deletion(-) (limited to 'src') diff --git a/src/mesa/main/texenvprogram.c b/src/mesa/main/texenvprogram.c index 3ff30058ec..050a3a99fe 100644 --- a/src/mesa/main/texenvprogram.c +++ b/src/mesa/main/texenvprogram.c @@ -1097,7 +1097,7 @@ static struct ureg emit_texenv(struct texenv_fragment_program *p, GLuint unit) { struct state_key *key = p->state; - GLboolean saturate = (unit < p->last_tex_stage); + GLboolean saturate; GLuint rgb_shift, alpha_shift; struct ureg out, shift; struct ureg dest; @@ -1125,6 +1125,11 @@ emit_texenv(struct texenv_fragment_program *p, GLuint unit) break; } + /* If we'll do rgb/alpha shifting don't saturate in emit_combine(). + * We don't want to clamp twice. + */ + saturate = !(rgb_shift || alpha_shift); + /* If this is the very last calculation, emit direct to output reg: */ if (key->separate_specular || @@ -1173,6 +1178,7 @@ emit_texenv(struct texenv_fragment_program *p, GLuint unit) /* Deal with the final shift: */ if (alpha_shift || rgb_shift) { + saturate = GL_TRUE; /* always saturate at this point */ if (rgb_shift == alpha_shift) { shift = register_scalar_const(p, (GLfloat)(1< Date: Mon, 31 Aug 2009 11:17:59 -0600 Subject: mesa: added const qualifiers, move local var --- src/mesa/main/texenvprogram.c | 15 ++++++++------- 1 file changed, 8 insertions(+), 7 deletions(-) (limited to 'src') diff --git a/src/mesa/main/texenvprogram.c b/src/mesa/main/texenvprogram.c index 050a3a99fe..454d97506e 100644 --- a/src/mesa/main/texenvprogram.c +++ b/src/mesa/main/texenvprogram.c @@ -936,7 +936,7 @@ static struct ureg emit_combine_source( struct texenv_fragment_program *p, } } -static GLboolean args_match( struct state_key *key, GLuint unit ) +static GLboolean args_match( const struct state_key *key, GLuint unit ) { GLuint i, nr = key->unit[unit].NumArgsRGB; @@ -1096,11 +1096,10 @@ static struct ureg emit_combine( struct texenv_fragment_program *p, static struct ureg emit_texenv(struct texenv_fragment_program *p, GLuint unit) { - struct state_key *key = p->state; + const struct state_key *key = p->state; GLboolean saturate; GLuint rgb_shift, alpha_shift; - struct ureg out, shift; - struct ureg dest; + struct ureg out, dest; if (!key->unit[unit].enabled) { return get_source(p, SRC_PREVIOUS, 0); @@ -1152,7 +1151,6 @@ emit_texenv(struct texenv_fragment_program *p, GLuint unit) } else if (key->unit[unit].ModeRGB == MODE_DOT3_RGBA_EXT || key->unit[unit].ModeRGB == MODE_DOT3_RGBA) { - out = emit_combine( p, dest, WRITEMASK_XYZW, saturate, unit, key->unit[unit].NumArgsRGB, @@ -1178,7 +1176,10 @@ emit_texenv(struct texenv_fragment_program *p, GLuint unit) /* Deal with the final shift: */ if (alpha_shift || rgb_shift) { + struct ureg shift; + saturate = GL_TRUE; /* always saturate at this point */ + if (rgb_shift == alpha_shift) { shift = register_scalar_const(p, (GLfloat)(1<state; + const struct state_key *key = p->state; GLuint i; for (i = 0; i < key->unit[unit].NumArgsRGB; i++) { @@ -1298,7 +1299,7 @@ load_texunit_sources( struct texenv_fragment_program *p, int unit ) static GLboolean load_texunit_bumpmap( struct texenv_fragment_program *p, int unit ) { - struct state_key *key = p->state; + const struct state_key *key = p->state; GLuint bumpedUnitNr = key->unit[unit].OptRGB[1].Source - SRC_TEXTURE0; struct ureg texcDst, bumpMapRes; struct ureg constdudvcolor = register_const4f(p, 0.0, 0.0, 0.0, 1.0); -- cgit v1.2.3 From fcf0804c05faefd196ed5525c068ee4cd30c5312 Mon Sep 17 00:00:00 2001 From: Brian Paul Date: Mon, 31 Aug 2009 13:28:31 -0600 Subject: swrast: can't use deferred texture/shading if using KIL instruction If the fragment program uses KIL, we have to execute it before z/stencil testing. Otherwise, deferred texture/shading lets us skip shading for pixels that fail z/stencil testing. --- src/mesa/swrast/s_context.c | 3 +++ 1 file changed, 3 insertions(+) (limited to 'src') diff --git a/src/mesa/swrast/s_context.c b/src/mesa/swrast/s_context.c index e7c2ace32c..abf0008565 100644 --- a/src/mesa/swrast/s_context.c +++ b/src/mesa/swrast/s_context.c @@ -220,6 +220,9 @@ _swrast_update_deferred_texture(GLcontext *ctx) /* Z comes from fragment program/shader */ swrast->_DeferredTexture = GL_FALSE; } + else if (fprog && fprog->UsesKill) { + swrast->_DeferredTexture = GL_FALSE; + } else if (ctx->Query.CurrentOcclusionObject) { /* occlusion query depends on shader discard/kill results */ swrast->_DeferredTexture = GL_FALSE; -- cgit v1.2.3 From 3f785080db33d437893564dded325452770699be Mon Sep 17 00:00:00 2001 From: Brian Paul Date: Mon, 31 Aug 2009 17:54:46 -0600 Subject: swrast: fix selection/feedback regression This fixes a conform selection/feedback regression introduced by commit 8f4d66c5f893b49eb3973aa3b31a856314c045c7 --- src/mesa/swrast/s_triangle.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'src') diff --git a/src/mesa/swrast/s_triangle.c b/src/mesa/swrast/s_triangle.c index 1d2fed7169..1ab0e19f92 100644 --- a/src/mesa/swrast/s_triangle.c +++ b/src/mesa/swrast/s_triangle.c @@ -61,7 +61,7 @@ _swrast_culltriangle( GLcontext *ctx, GLfloat fy = v2->attrib[FRAG_ATTRIB_WPOS][1] - v0->attrib[FRAG_ATTRIB_WPOS][1]; GLfloat c = ex*fy-ey*fx; - if (c * swrast->_BackfaceSign * swrast->_BackfaceCullSign < 0.0F) + if (c * swrast->_BackfaceSign * swrast->_BackfaceCullSign <= 0.0F) return GL_FALSE; return GL_TRUE; -- cgit v1.2.3 From 1d7a989b104f02042fadfeec5bd20654bbb6fea6 Mon Sep 17 00:00:00 2001 From: Keith Whitwell Date: Tue, 1 Sep 2009 11:11:28 +0100 Subject: draw: remove unused variable --- src/gallium/auxiliary/draw/draw_pipe.c | 1 - 1 file changed, 1 deletion(-) (limited to 'src') diff --git a/src/gallium/auxiliary/draw/draw_pipe.c b/src/gallium/auxiliary/draw/draw_pipe.c index be2f0f27f2..3cfae2de83 100644 --- a/src/gallium/auxiliary/draw/draw_pipe.c +++ b/src/gallium/auxiliary/draw/draw_pipe.c @@ -232,7 +232,6 @@ void draw_pipeline_run( struct draw_context *draw, unsigned count ) { char *verts = (char *)vertices; - unsigned i; draw->pipeline.verts = verts; draw->pipeline.vertex_stride = stride; -- cgit v1.2.3 From de343680a3e6b2a588f8b3c844828b8d9e6cb6a5 Mon Sep 17 00:00:00 2001 From: Keith Whitwell Date: Wed, 2 Sep 2009 17:58:52 +0100 Subject: util: add version of u_blit_pixels which takes a writemask Values outside the writemask are set in the destination to {0,0,0,1} --- src/gallium/auxiliary/util/u_blit.c | 55 ++++++++++++++++++++------- src/gallium/auxiliary/util/u_blit.h | 11 ++++++ src/gallium/auxiliary/util/u_simple_shaders.c | 45 +++++++++++++++++++++- src/gallium/auxiliary/util/u_simple_shaders.h | 4 ++ 4 files changed, 100 insertions(+), 15 deletions(-) (limited to 'src') diff --git a/src/gallium/auxiliary/util/u_blit.c b/src/gallium/auxiliary/util/u_blit.c index cda6dbd46d..c516317d70 100644 --- a/src/gallium/auxiliary/util/u_blit.c +++ b/src/gallium/auxiliary/util/u_blit.c @@ -62,7 +62,7 @@ struct blit_state struct pipe_viewport_state viewport; void *vs; - void *fs; + void *fs[TGSI_WRITEMASK_XYZW + 1]; struct pipe_buffer *vbuf; /**< quad vertices */ unsigned vbuf_slot; @@ -125,7 +125,7 @@ util_create_blit(struct pipe_context *pipe, struct cso_context *cso) } /* fragment shader */ - ctx->fs = util_make_fragment_tex_shader(pipe); + ctx->fs[TGSI_WRITEMASK_XYZW] = util_make_fragment_tex_shader(pipe); ctx->vbuf = NULL; /* init vertex data that doesn't change */ @@ -146,9 +146,13 @@ void util_destroy_blit(struct blit_state *ctx) { struct pipe_context *pipe = ctx->pipe; + unsigned i; pipe->delete_vs_state(pipe, ctx->vs); - pipe->delete_fs_state(pipe, ctx->fs); + + for (i = 0; i < Elements(ctx->fs); i++) + if (ctx->fs[i]) + pipe->delete_fs_state(pipe, ctx->fs[i]); pipe_buffer_reference(&ctx->vbuf, NULL); @@ -299,14 +303,15 @@ regions_overlap(int srcX0, int srcY0, * XXX need some control over blitting Z and/or stencil. */ void -util_blit_pixels(struct blit_state *ctx, - struct pipe_surface *src, - int srcX0, int srcY0, - int srcX1, int srcY1, - struct pipe_surface *dst, - int dstX0, int dstY0, - int dstX1, int dstY1, - float z, uint filter) +util_blit_pixels_writemask(struct blit_state *ctx, + struct pipe_surface *src, + int srcX0, int srcY0, + int srcX1, int srcY1, + struct pipe_surface *dst, + int dstX0, int dstY0, + int dstX1, int dstY1, + float z, uint filter, + uint writemask) { struct pipe_context *pipe = ctx->pipe; struct pipe_screen *screen = pipe->screen; @@ -426,8 +431,11 @@ util_blit_pixels(struct blit_state *ctx, /* texture */ cso_set_sampler_textures(ctx->cso, 1, &tex); + if (ctx->fs[writemask] == NULL) + ctx->fs[writemask] = util_make_fragment_tex_shader_writemask(pipe, writemask); + /* shaders */ - cso_set_fragment_shader_handle(ctx->cso, ctx->fs); + cso_set_fragment_shader_handle(ctx->cso, ctx->fs[writemask]); cso_set_vertex_shader_handle(ctx->cso, ctx->vs); /* drawing dest */ @@ -462,6 +470,27 @@ util_blit_pixels(struct blit_state *ctx, } +void +util_blit_pixels(struct blit_state *ctx, + struct pipe_surface *src, + int srcX0, int srcY0, + int srcX1, int srcY1, + struct pipe_surface *dst, + int dstX0, int dstY0, + int dstX1, int dstY1, + float z, uint filter ) +{ + util_blit_pixels_writemask( ctx, src, + srcX0, srcY0, + srcX1, srcY1, + dst, + dstX0, dstY0, + dstX1, dstY1, + z, filter, + TGSI_WRITEMASK_XYZW ); +} + + /* Release vertex buffer at end of frame to avoid synchronous * rendering. */ @@ -535,7 +564,7 @@ util_blit_pixels_tex(struct blit_state *ctx, cso_set_sampler_textures(ctx->cso, 1, &tex); /* shaders */ - cso_set_fragment_shader_handle(ctx->cso, ctx->fs); + cso_set_fragment_shader_handle(ctx->cso, ctx->fs[TGSI_WRITEMASK_XYZW]); cso_set_vertex_shader_handle(ctx->cso, ctx->vs); /* drawing dest */ diff --git a/src/gallium/auxiliary/util/u_blit.h b/src/gallium/auxiliary/util/u_blit.h index c35beceda8..a102021529 100644 --- a/src/gallium/auxiliary/util/u_blit.h +++ b/src/gallium/auxiliary/util/u_blit.h @@ -60,6 +60,17 @@ util_blit_pixels(struct blit_state *ctx, int dstX1, int dstY1, float z, uint filter); +void +util_blit_pixels_writemask(struct blit_state *ctx, + struct pipe_surface *src, + int srcX0, int srcY0, + int srcX1, int srcY1, + struct pipe_surface *dst, + int dstX0, int dstY0, + int dstX1, int dstY1, + float z, uint filter, + uint writemask); + extern void util_blit_pixels_tex(struct blit_state *ctx, struct pipe_texture *tex, diff --git a/src/gallium/auxiliary/util/u_simple_shaders.c b/src/gallium/auxiliary/util/u_simple_shaders.c index e519c354d2..acc5b83c62 100644 --- a/src/gallium/auxiliary/util/u_simple_shaders.c +++ b/src/gallium/auxiliary/util/u_simple_shaders.c @@ -152,11 +152,14 @@ util_make_vertex_passthrough_shader(struct pipe_context *pipe, /** * Make simple fragment texture shader: - * TEX OUT[0], IN[0], SAMP[0], 2D; + * IMM {0,0,0,1} // (if writemask != 0xf) + * MOV OUT[0], IMM[0] // (if writemask != 0xf) + * TEX OUT[0].writemask, IN[0], SAMP[0], 2D; * END; */ void * -util_make_fragment_tex_shader(struct pipe_context *pipe) +util_make_fragment_tex_shader_writemask(struct pipe_context *pipe, + unsigned writemask ) { struct pipe_shader_state shader; struct tgsi_token tokens[100]; @@ -217,12 +220,43 @@ util_make_fragment_tex_shader(struct pipe_context *pipe) header, Elements(tokens) - ti); + + if (writemask != TGSI_WRITEMASK_XYZW) { + struct tgsi_full_immediate imm; + static const float value[4] = { 0, 0, 0, 1 }; + + imm = tgsi_default_full_immediate(); + imm.Immediate.NrTokens += 4; + imm.Immediate.DataType = TGSI_IMM_FLOAT32; + imm.u.Pointer = value; + + ti += tgsi_build_full_immediate(&imm, + &tokens[ti], + header, + Elements(tokens) - ti ); + + /* MOV instruction */ + inst = tgsi_default_full_instruction(); + inst.Instruction.Opcode = TGSI_OPCODE_MOV; + inst.Instruction.NumDstRegs = 1; + inst.FullDstRegisters[0].DstRegister.File = TGSI_FILE_OUTPUT; + inst.FullDstRegisters[0].DstRegister.Index = 0; + inst.Instruction.NumSrcRegs = 1; + inst.FullSrcRegisters[0].SrcRegister.File = TGSI_FILE_IMMEDIATE; + inst.FullSrcRegisters[0].SrcRegister.Index = 0; + ti += tgsi_build_full_instruction(&inst, + &tokens[ti], + header, + Elements(tokens) - ti ); + } + /* TEX instruction */ inst = tgsi_default_full_instruction(); inst.Instruction.Opcode = TGSI_OPCODE_TEX; inst.Instruction.NumDstRegs = 1; inst.FullDstRegisters[0].DstRegister.File = TGSI_FILE_OUTPUT; inst.FullDstRegisters[0].DstRegister.Index = 0; + inst.FullDstRegisters[0].DstRegister.WriteMask = writemask; inst.Instruction.NumSrcRegs = 2; inst.InstructionExtTexture.Texture = TGSI_TEXTURE_2D; inst.FullSrcRegisters[0].SrcRegister.File = TGSI_FILE_INPUT; @@ -253,6 +287,13 @@ util_make_fragment_tex_shader(struct pipe_context *pipe) return pipe->create_fs_state(pipe, &shader); } +void * +util_make_fragment_tex_shader(struct pipe_context *pipe ) +{ + return util_make_fragment_tex_shader_writemask( pipe, + TGSI_WRITEMASK_XYZW ); +} + diff --git a/src/gallium/auxiliary/util/u_simple_shaders.h b/src/gallium/auxiliary/util/u_simple_shaders.h index 6f8d96af9b..d2e80d6eb4 100644 --- a/src/gallium/auxiliary/util/u_simple_shaders.h +++ b/src/gallium/auxiliary/util/u_simple_shaders.h @@ -49,6 +49,10 @@ util_make_vertex_passthrough_shader(struct pipe_context *pipe, const uint *semantic_indexes); +extern void * +util_make_fragment_tex_shader_writemask(struct pipe_context *pipe, + unsigned writemask ); + extern void * util_make_fragment_tex_shader(struct pipe_context *pipe); -- cgit v1.2.3 From e79054cc4090a2be346236236c9e18ae85cad43d Mon Sep 17 00:00:00 2001 From: Keith Whitwell Date: Wed, 2 Sep 2009 18:30:11 +0100 Subject: st/mesa: Do GL_RGBA->GL_RGB texsubimage on hardware State tracker currently backs GL_RGB textures with RGBA almost always. This means we need to maintain A==1 in these textures to give correct GL_RGB sampling results. This change offloads the RGBA->RGB copy to hardware using the new writemask version of u_blit_pixels. More src/dstLogical/dstActual triples could be shifted to hardware by this technique in future patches. --- src/mesa/state_tracker/st_cb_texture.c | 65 ++++++++++++++++++++++++++-------- 1 file changed, 51 insertions(+), 14 deletions(-) (limited to 'src') diff --git a/src/mesa/state_tracker/st_cb_texture.c b/src/mesa/state_tracker/st_cb_texture.c index ee71c012c6..2d37d24ea8 100644 --- a/src/mesa/state_tracker/st_cb_texture.c +++ b/src/mesa/state_tracker/st_cb_texture.c @@ -56,6 +56,7 @@ #include "pipe/p_context.h" #include "pipe/p_defines.h" #include "pipe/p_inlines.h" +#include "pipe/p_shader_tokens.h" #include "util/u_tile.h" #include "util/u_blit.h" #include "util/u_surface.h" @@ -1396,6 +1397,36 @@ fallback_copy_texsubimage(GLcontext *ctx, GLenum target, GLint level, } +static unsigned +compatible_src_dst_formats(const struct gl_renderbuffer *src, + const struct gl_texture_image *dst) +{ + const GLenum srcFormat = src->_BaseFormat; + const GLenum dstLogicalFormat = dst->_BaseFormat; + + if (srcFormat == dstLogicalFormat) { + /* This is the same as matching_base_formats, which should + * always pass, as it did previously. + */ + return TGSI_WRITEMASK_XYZW; + } + else if (srcFormat == GL_RGBA && + dstLogicalFormat == GL_RGB) { + /* Add a single special case to cope with RGBA->RGB transfers, + * setting A to 1.0 to cope with situations where the RGB + * destination is actually stored as RGBA. + */ + return TGSI_WRITEMASK_XYZ; /* A ==> 1.0 */ + } + else { + /* Otherwise fail. + */ + return 0; + } +} + + + /** * Do a CopyTex[Sub]Image1/2/3D() using a hardware (blit) path if possible. * Note that the region to copy has already been clipped so we know we @@ -1425,6 +1456,9 @@ st_copy_texsubimage(GLcontext *ctx, enum pipe_format dest_format, src_format; GLboolean use_fallback = GL_TRUE; GLboolean matching_base_formats; + GLuint format_writemask; + struct pipe_surface *dest_surface = NULL; + GLboolean do_flip = (st_fb_orientation(ctx->ReadBuffer) == Y_0_TOP); /* any rendering in progress must flushed before we grab the fb image */ st_flush(ctx->st, PIPE_FLUSH_RENDER_CACHE, NULL); @@ -1495,13 +1529,14 @@ st_copy_texsubimage(GLcontext *ctx, * textured-quad paths. */ matching_base_formats = (strb->Base._BaseFormat == texImage->_BaseFormat); + format_writemask = compatible_src_dst_formats(&strb->Base, texImage); - if (matching_base_formats && ctx->_ImageTransferState == 0x0) { - /* try potential hardware path */ - struct pipe_surface *dest_surface = NULL; - boolean do_flip = (st_fb_orientation(ctx->ReadBuffer) == Y_0_TOP); + if (ctx->_ImageTransferState == 0x0) { - if (src_format == dest_format && !do_flip) { + if (matching_base_formats && + src_format == dest_format && + !do_flip) + { /* use surface_copy() / blit */ dest_surface = screen->get_tex_surface(screen, stImage->pt, @@ -1521,7 +1556,8 @@ st_copy_texsubimage(GLcontext *ctx, width, height); use_fallback = GL_FALSE; } - else if (screen->is_format_supported(screen, src_format, + else if (format_writemask && + screen->is_format_supported(screen, src_format, PIPE_TEXTURE_2D, PIPE_TEXTURE_USAGE_SAMPLER, 0) && @@ -1545,14 +1581,15 @@ st_copy_texsubimage(GLcontext *ctx, srcY0 = srcY; srcY1 = srcY0 + height; } - util_blit_pixels(ctx->st->blit, - strb->surface, - srcX, srcY0, - srcX + width, srcY1, - dest_surface, - destX, destY, - destX + width, destY + height, - 0.0, PIPE_TEX_MIPFILTER_NEAREST); + util_blit_pixels_writemask(ctx->st->blit, + strb->surface, + srcX, srcY0, + srcX + width, srcY1, + dest_surface, + destX, destY, + destX + width, destY + height, + 0.0, PIPE_TEX_MIPFILTER_NEAREST, + format_writemask); use_fallback = GL_FALSE; } -- cgit v1.2.3 From c3a0624ef01c3b347ed8a9fd3bab578b52bd9130 Mon Sep 17 00:00:00 2001 From: Brian Paul Date: Thu, 3 Sep 2009 14:39:53 -0600 Subject: st/mesa: fix glCopyPixels(GL_STENCIL_INDEX) inverted position If the renderbuffer orientation is Y=0=TOP we need to invert the dstY position. --- src/mesa/state_tracker/st_cb_drawpixels.c | 4 ++++ 1 file changed, 4 insertions(+) (limited to 'src') diff --git a/src/mesa/state_tracker/st_cb_drawpixels.c b/src/mesa/state_tracker/st_cb_drawpixels.c index 8b5094a04f..adb349c1f1 100644 --- a/src/mesa/state_tracker/st_cb_drawpixels.c +++ b/src/mesa/state_tracker/st_cb_drawpixels.c @@ -842,6 +842,10 @@ copy_stencil_pixels(GLcontext *ctx, GLint srcx, GLint srcy, else usage = PIPE_TRANSFER_WRITE; + if (st_fb_orientation(ctx->DrawBuffer) == Y_0_TOP) { + dsty = rbDraw->Base.Height - dsty - height; + } + ptDraw = st_cond_flush_get_tex_transfer(st_context(ctx), rbDraw->texture, 0, 0, 0, usage, dstx, dsty, -- cgit v1.2.3 From 67c286d20e6f1256950d81aab3f6b54cd4926602 Mon Sep 17 00:00:00 2001 From: Brian Paul Date: Thu, 3 Sep 2009 15:04:49 -0600 Subject: st/mesa: silence uninitialized var warnings --- src/mesa/state_tracker/st_cb_texture.c | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) (limited to 'src') diff --git a/src/mesa/state_tracker/st_cb_texture.c b/src/mesa/state_tracker/st_cb_texture.c index 2d37d24ea8..2db28ef0a4 100644 --- a/src/mesa/state_tracker/st_cb_texture.c +++ b/src/mesa/state_tracker/st_cb_texture.c @@ -526,9 +526,9 @@ st_TexImage(GLcontext * ctx, struct st_texture_image *stImage = st_texture_image(texImage); GLint postConvWidth, postConvHeight; GLint texelBytes, sizeInBytes; - GLuint dstRowStride; + GLuint dstRowStride = 0; struct gl_pixelstore_attrib unpackNB; - enum pipe_transfer_usage transfer_usage; + enum pipe_transfer_usage transfer_usage = 0; DBG("%s target %s level %d %dx%dx%d border %d\n", __FUNCTION__, _mesa_lookup_enum_by_nr(target), level, width, height, depth, border); -- cgit v1.2.3 From 1c32caf075ce4015ba50d0aa9f0a2ff924c21970 Mon Sep 17 00:00:00 2001 From: Brian Paul Date: Thu, 3 Sep 2009 15:23:15 -0600 Subject: gallium/xlib: silence uninitialized var warning --- src/gallium/winsys/xlib/xlib_brw_screen.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'src') diff --git a/src/gallium/winsys/xlib/xlib_brw_screen.c b/src/gallium/winsys/xlib/xlib_brw_screen.c index fe8dfff767..a78efb10fa 100644 --- a/src/gallium/winsys/xlib/xlib_brw_screen.c +++ b/src/gallium/winsys/xlib/xlib_brw_screen.c @@ -364,7 +364,7 @@ void xlib_brw_buffer_subdata_typed( struct pipe_winsys *pws, unsigned data_type ) { unsigned aub_type = DW_GENERAL_STATE; - unsigned aub_sub_type; + unsigned aub_sub_type = 0; switch (data_type) { case BRW_CC_VP: -- cgit v1.2.3 From 32156f3a1165624cd4e2d0d337809fb046bd9285 Mon Sep 17 00:00:00 2001 From: Brian Paul Date: Thu, 3 Sep 2009 15:44:22 -0600 Subject: gallium/xlib: silence unitialized var warning --- src/gallium/state_trackers/glx/xlib/xm_api.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'src') diff --git a/src/gallium/state_trackers/glx/xlib/xm_api.c b/src/gallium/state_trackers/glx/xlib/xm_api.c index a3d1651653..45ad694e23 100644 --- a/src/gallium/state_trackers/glx/xlib/xm_api.c +++ b/src/gallium/state_trackers/glx/xlib/xm_api.c @@ -743,7 +743,7 @@ XMesaContext XMesaCreateContext( XMesaVisual v, XMesaContext share_list ) { static GLboolean firstTime = GL_TRUE; struct pipe_screen *screen; - struct pipe_context *pipe; + struct pipe_context *pipe = NULL; XMesaContext c; GLcontext *mesaCtx; uint pf; -- cgit v1.2.3 From cab307ce6b32d2ffdb0eb3bb5bae93c6fb9305fb Mon Sep 17 00:00:00 2001 From: Brian Paul Date: Fri, 15 May 2009 09:14:24 -0600 Subject: i965: rename var: s/tmp/vs_inputs/ (cherry picked from commit 840c09fc71542fdfc71edd2a2802925d467567bb) --- src/mesa/drivers/dri/i965/brw_draw_upload.c | 16 ++++++++-------- 1 file changed, 8 insertions(+), 8 deletions(-) (limited to 'src') diff --git a/src/mesa/drivers/dri/i965/brw_draw_upload.c b/src/mesa/drivers/dri/i965/brw_draw_upload.c index b91b20bec6..1b8bcc14ec 100644 --- a/src/mesa/drivers/dri/i965/brw_draw_upload.c +++ b/src/mesa/drivers/dri/i965/brw_draw_upload.c @@ -343,7 +343,7 @@ static void brw_prepare_vertices(struct brw_context *brw) { GLcontext *ctx = &brw->intel.ctx; struct intel_context *intel = intel_context(ctx); - GLuint tmp = brw->vs.prog_data->inputs_read; + GLbitfield vs_inputs = brw->vs.prog_data->inputs_read; GLuint i; const unsigned char *ptr = NULL; GLuint interleave = 0; @@ -362,11 +362,11 @@ static void brw_prepare_vertices(struct brw_context *brw) _mesa_printf("%s %d..%d\n", __FUNCTION__, min_index, max_index); /* Accumulate the list of enabled arrays. */ - while (tmp) { - GLuint i = _mesa_ffsll(tmp)-1; + while (vs_inputs) { + GLuint i = _mesa_ffsll(vs_inputs) - 1; struct brw_vertex_element *input = &brw->vb.inputs[i]; - tmp &= ~(1<intel.ctx; struct intel_context *intel = intel_context(ctx); - GLuint tmp = brw->vs.prog_data->inputs_read; + GLbitfield vs_inputs = brw->vs.prog_data->inputs_read; struct brw_vertex_element *enabled[VERT_ATTRIB_MAX]; GLuint i; GLuint nr_enabled = 0; /* Accumulate the list of enabled arrays. */ - while (tmp) { - i = _mesa_ffsll(tmp)-1; + while (vs_inputs) { + i = _mesa_ffsll(vs_inputs) - 1; struct brw_vertex_element *input = &brw->vb.inputs[i]; - tmp &= ~(1< Date: Tue, 23 Jun 2009 19:30:25 -0700 Subject: i965: Set the max index buffer address correctly according to the docs. It's the last addressable byte, not the byte after the end of the buffer. (cherry picked from commit b72dea5441e8e9226dabf1826fa3bc129c7bc281) --- src/mesa/drivers/dri/i965/brw_draw_upload.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'src') diff --git a/src/mesa/drivers/dri/i965/brw_draw_upload.c b/src/mesa/drivers/dri/i965/brw_draw_upload.c index 1b8bcc14ec..3ef56a0068 100644 --- a/src/mesa/drivers/dri/i965/brw_draw_upload.c +++ b/src/mesa/drivers/dri/i965/brw_draw_upload.c @@ -635,7 +635,7 @@ static void brw_emit_indices(struct brw_context *brw) if (index_buffer == NULL) return; - ib_size = get_size(index_buffer->type) * index_buffer->count; + ib_size = get_size(index_buffer->type) * index_buffer->count - 1; /* Emit the indexbuffer packet: */ -- cgit v1.2.3 From 7e26bdb849b75f4aeb69cf8b1fdffbc461265490 Mon Sep 17 00:00:00 2001 From: Eric Anholt Date: Mon, 3 Aug 2009 14:46:18 -0700 Subject: i965: Calculate enabled[] and nr_enabled once and re-use the values. The code duplication bothered me. (cherry picked from commit 9b9cb30d128fc5f1ba77287696ecd508e640efde) --- src/mesa/drivers/dri/i965/brw_context.h | 3 ++ src/mesa/drivers/dri/i965/brw_draw_upload.c | 44 ++++++++++------------------- 2 files changed, 18 insertions(+), 29 deletions(-) (limited to 'src') diff --git a/src/mesa/drivers/dri/i965/brw_context.h b/src/mesa/drivers/dri/i965/brw_context.h index 577497bf6b..e52fc3f374 100644 --- a/src/mesa/drivers/dri/i965/brw_context.h +++ b/src/mesa/drivers/dri/i965/brw_context.h @@ -477,6 +477,9 @@ struct brw_context struct { struct brw_vertex_element inputs[VERT_ATTRIB_MAX]; + struct brw_vertex_element *enabled[VERT_ATTRIB_MAX]; + GLuint nr_enabled; + #define BRW_NR_UPLOAD_BUFS 17 #define BRW_UPLOAD_INIT_SIZE (128*1024) diff --git a/src/mesa/drivers/dri/i965/brw_draw_upload.c b/src/mesa/drivers/dri/i965/brw_draw_upload.c index 3ef56a0068..c1fe85908b 100644 --- a/src/mesa/drivers/dri/i965/brw_draw_upload.c +++ b/src/mesa/drivers/dri/i965/brw_draw_upload.c @@ -350,9 +350,6 @@ static void brw_prepare_vertices(struct brw_context *brw) unsigned int min_index = brw->vb.min_index; unsigned int max_index = brw->vb.max_index; - struct brw_vertex_element *enabled[VERT_ATTRIB_MAX]; - GLuint nr_enabled = 0; - struct brw_vertex_element *upload[VERT_ATTRIB_MAX]; GLuint nr_uploads = 0; @@ -362,12 +359,13 @@ static void brw_prepare_vertices(struct brw_context *brw) _mesa_printf("%s %d..%d\n", __FUNCTION__, min_index, max_index); /* Accumulate the list of enabled arrays. */ + brw->vb.nr_enabled = 0; while (vs_inputs) { GLuint i = _mesa_ffsll(vs_inputs) - 1; struct brw_vertex_element *input = &brw->vb.inputs[i]; vs_inputs &= ~(1 << i); - enabled[nr_enabled++] = input; + brw->vb.enabled[brw->vb.nr_enabled++] = input; } /* XXX: In the rare cases where this happens we fallback all @@ -376,13 +374,13 @@ static void brw_prepare_vertices(struct brw_context *brw) * cases with > 17 vertex attributes enabled, so it probably * isn't an issue at this point. */ - if (nr_enabled >= BRW_VEP_MAX) { + if (brw->vb.nr_enabled >= BRW_VEP_MAX) { intel->Fallback = 1; return; } - for (i = 0; i < nr_enabled; i++) { - struct brw_vertex_element *input = enabled[i]; + for (i = 0; i < brw->vb.nr_enabled; i++) { + struct brw_vertex_element *input = brw->vb.enabled[i]; input->element_size = get_size(input->glarray->Type) * input->glarray->Size; input->count = input->glarray->StrideB ? max_index + 1 - min_index : 1; @@ -466,8 +464,8 @@ static void brw_prepare_vertices(struct brw_context *brw) brw_prepare_query_begin(brw); - for (i = 0; i < nr_enabled; i++) { - struct brw_vertex_element *input = enabled[i]; + for (i = 0; i < brw->vb.nr_enabled; i++) { + struct brw_vertex_element *input = brw->vb.enabled[i]; brw_add_validated_bo(brw, input->bo); } @@ -477,19 +475,7 @@ static void brw_emit_vertices(struct brw_context *brw) { GLcontext *ctx = &brw->intel.ctx; struct intel_context *intel = intel_context(ctx); - GLbitfield vs_inputs = brw->vs.prog_data->inputs_read; - struct brw_vertex_element *enabled[VERT_ATTRIB_MAX]; GLuint i; - GLuint nr_enabled = 0; - - /* Accumulate the list of enabled arrays. */ - while (vs_inputs) { - i = _mesa_ffsll(vs_inputs) - 1; - struct brw_vertex_element *input = &brw->vb.inputs[i]; - - vs_inputs &= ~(1 << i); - enabled[nr_enabled++] = input; - } brw_emit_query_begin(brw); @@ -499,12 +485,12 @@ static void brw_emit_vertices(struct brw_context *brw) * are interleaved or from the same VBO. TBD if this makes a * performance difference. */ - BEGIN_BATCH(1 + nr_enabled * 4, IGNORE_CLIPRECTS); + BEGIN_BATCH(1 + brw->vb.nr_enabled * 4, IGNORE_CLIPRECTS); OUT_BATCH((CMD_VERTEX_BUFFER << 16) | - ((1 + nr_enabled * 4) - 2)); + ((1 + brw->vb.nr_enabled * 4) - 2)); - for (i = 0; i < nr_enabled; i++) { - struct brw_vertex_element *input = enabled[i]; + for (i = 0; i < brw->vb.nr_enabled; i++) { + struct brw_vertex_element *input = brw->vb.enabled[i]; OUT_BATCH((i << BRW_VB0_INDEX_SHIFT) | BRW_VB0_ACCESS_VERTEXDATA | @@ -517,10 +503,10 @@ static void brw_emit_vertices(struct brw_context *brw) } ADVANCE_BATCH(); - BEGIN_BATCH(1 + nr_enabled * 2, IGNORE_CLIPRECTS); - OUT_BATCH((CMD_VERTEX_ELEMENT << 16) | ((1 + nr_enabled * 2) - 2)); - for (i = 0; i < nr_enabled; i++) { - struct brw_vertex_element *input = enabled[i]; + BEGIN_BATCH(1 + brw->vb.nr_enabled * 2, IGNORE_CLIPRECTS); + OUT_BATCH((CMD_VERTEX_ELEMENT << 16) | ((1 + brw->vb.nr_enabled * 2) - 2)); + for (i = 0; i < brw->vb.nr_enabled; i++) { + struct brw_vertex_element *input = brw->vb.enabled[i]; uint32_t format = get_surface_type(input->glarray->Type, input->glarray->Size, input->glarray->Format, -- cgit v1.2.3 From 9eca0e5350377148976e0d1200f98bd20ac28197 Mon Sep 17 00:00:00 2001 From: Eric Anholt Date: Thu, 30 Jul 2009 13:40:29 -0700 Subject: i965: Don't emit bad packets when no VBs are referenced. It appears that sometimes Mesa (and I suppose a VS could as well) emits a program which references no vertex data, and thus we end up with nr_enabled == 0 even though some VBs are enabled. We'd end up emitting VB/VE packet headers of 0xffffffff in that case, leading to GPU hangs. Bug #22945 (wine with an uncompiled VS) (cherry picked from commit d1fbfd0f962347e4153db3852292d44de5aea863) --- src/mesa/drivers/dri/i965/brw_draw_upload.c | 22 ++++++++++++++++++++++ 1 file changed, 22 insertions(+) (limited to 'src') diff --git a/src/mesa/drivers/dri/i965/brw_draw_upload.c b/src/mesa/drivers/dri/i965/brw_draw_upload.c index c1fe85908b..e7a87b6e09 100644 --- a/src/mesa/drivers/dri/i965/brw_draw_upload.c +++ b/src/mesa/drivers/dri/i965/brw_draw_upload.c @@ -479,6 +479,28 @@ static void brw_emit_vertices(struct brw_context *brw) brw_emit_query_begin(brw); + /* If the VS doesn't read any inputs (calculating vertex position from + * a state variable for some reason, for example), emit a single pad + * VERTEX_ELEMENT struct and bail. + * + * The stale VB state stays in place, but they don't do anything unless + * a VE loads from them. + */ + if (brw->vb.nr_enabled == 0) { + BEGIN_BATCH(3, IGNORE_CLIPRECTS); + OUT_BATCH((CMD_VERTEX_ELEMENT << 16) | 1); + OUT_BATCH((0 << BRW_VE0_INDEX_SHIFT) | + BRW_VE0_VALID | + (BRW_SURFACEFORMAT_R32G32B32A32_FLOAT << BRW_VE0_FORMAT_SHIFT) | + (0 << BRW_VE0_SRC_OFFSET_SHIFT)); + OUT_BATCH((BRW_VE1_COMPONENT_STORE_0 << BRW_VE1_COMPONENT_0_SHIFT) | + (BRW_VE1_COMPONENT_STORE_0 << BRW_VE1_COMPONENT_1_SHIFT) | + (BRW_VE1_COMPONENT_STORE_0 << BRW_VE1_COMPONENT_2_SHIFT) | + (BRW_VE1_COMPONENT_STORE_1_FLT << BRW_VE1_COMPONENT_3_SHIFT)); + ADVANCE_BATCH(); + return; + } + /* Now emit VB and VEP state packets. * * This still defines a hardware VB for each input, even if they -- cgit v1.2.3 From 456a16491bc757af6ba9a98e5706a78f748e9a79 Mon Sep 17 00:00:00 2001 From: Eric Anholt Date: Mon, 3 Aug 2009 15:24:02 -0700 Subject: i965: Make sure the VS URB size is big enough to fit a VF VUE. This fix is just from code and docs inspection, but it may fix hangs on some applications. (cherry picked from commit e93848e595176ae0bad3bfe64e0ca63fd089bb72) --- src/mesa/drivers/dri/i965/brw_vs_emit.c | 9 ++++++++- 1 file changed, 8 insertions(+), 1 deletion(-) (limited to 'src') diff --git a/src/mesa/drivers/dri/i965/brw_vs_emit.c b/src/mesa/drivers/dri/i965/brw_vs_emit.c index b69616d6e5..6792c3a34e 100644 --- a/src/mesa/drivers/dri/i965/brw_vs_emit.c +++ b/src/mesa/drivers/dri/i965/brw_vs_emit.c @@ -68,6 +68,7 @@ static void release_tmps( struct brw_vs_compile *c ) static void brw_vs_alloc_regs( struct brw_vs_compile *c ) { GLuint i, reg = 0, mrf; + int attributes_in_vue; #if 0 if (c->vp->program.Base.Parameters->NumParameters >= 6) @@ -201,7 +202,13 @@ static void brw_vs_alloc_regs( struct brw_vs_compile *c ) */ c->prog_data.urb_read_length = (c->nr_inputs + 1) / 2; - c->prog_data.urb_entry_size = (c->nr_outputs + 2 + 3) / 4; + /* The VS VUEs are shared by VF (outputting our inputs) and VS, so size + * them to fit the biggest thing they need to. + */ + attributes_in_vue = MAX2(c->nr_outputs, c->nr_inputs); + + c->prog_data.urb_entry_size = (attributes_in_vue + 2 + 3) / 4; + c->prog_data.total_grf = reg; if (INTEL_DEBUG & DEBUG_VS) { -- cgit v1.2.3 From 217af32c2d6afab5e1907cc16fb4b6feb982abe7 Mon Sep 17 00:00:00 2001 From: Eric Anholt Date: Mon, 3 Aug 2009 17:12:43 -0700 Subject: i965: Even if no VS inputs are set, still load some amount of URB as required. See comment on Vertex URB Entry Read Length for VS_STATE. This, combined with the previous three commits, fixes #22945. (cherry picked from commit e340d4f9866db4bae391288e83a630a310b0dd2b) --- src/mesa/drivers/dri/i965/brw_vs_emit.c | 11 +++++++++++ 1 file changed, 11 insertions(+) (limited to 'src') diff --git a/src/mesa/drivers/dri/i965/brw_vs_emit.c b/src/mesa/drivers/dri/i965/brw_vs_emit.c index 6792c3a34e..b95079a5d1 100644 --- a/src/mesa/drivers/dri/i965/brw_vs_emit.c +++ b/src/mesa/drivers/dri/i965/brw_vs_emit.c @@ -124,6 +124,11 @@ static void brw_vs_alloc_regs( struct brw_vs_compile *c ) reg++; } } + /* If there are no inputs, we'll still be reading one attribute's worth + * because it's required -- see urb_read_length setting. + */ + if (c->nr_inputs == 0) + reg++; /* Allocate outputs: TODO: could organize the non-position outputs * to go straight into message regs. @@ -201,6 +206,12 @@ static void brw_vs_alloc_regs( struct brw_vs_compile *c ) * vertex urb, so is half the amount: */ c->prog_data.urb_read_length = (c->nr_inputs + 1) / 2; + /* Setting this field to 0 leads to undefined behavior according to the + * the VS_STATE docs. Our VUEs will always have at least one attribute + * sitting in them, even if it's padding. + */ + if (c->prog_data.urb_read_length == 0) + c->prog_data.urb_read_length = 1; /* The VS VUEs are shared by VF (outputting our inputs) and VS, so size * them to fit the biggest thing they need to. -- cgit v1.2.3 From 63b3fa2bcecc75a116ce651da435d205ccd43584 Mon Sep 17 00:00:00 2001 From: Eric Anholt Date: Mon, 3 Aug 2009 17:55:14 -0700 Subject: i965: Assert that the offset in the VBO is below the VBO size. This avoids sending a bad buffer address to the GPU due to programmer error, and is permitted by the ARB_vbo spec. Note that we still have the opportunity to dereference past the end of the GPU, because we aren't clipping to a correct _MaxElement, but that appears to be harder than it should be. This gets us the 90% solution. Bug #19911. (cherry picked from commit d7430d942f6c7950a92367aeb13b80cf76ccad78) --- src/mesa/drivers/dri/i965/brw_draw_upload.c | 14 ++++++++++++++ 1 file changed, 14 insertions(+) (limited to 'src') diff --git a/src/mesa/drivers/dri/i965/brw_draw_upload.c b/src/mesa/drivers/dri/i965/brw_draw_upload.c index e7a87b6e09..05079c043a 100644 --- a/src/mesa/drivers/dri/i965/brw_draw_upload.c +++ b/src/mesa/drivers/dri/i965/brw_draw_upload.c @@ -396,6 +396,20 @@ static void brw_prepare_vertices(struct brw_context *brw) dri_bo_reference(input->bo); input->offset = (unsigned long)input->glarray->Ptr; input->stride = input->glarray->StrideB; + + /* This is a common place to reach if the user mistakenly supplies + * a pointer in place of a VBO offset. If we just let it go through, + * we may end up dereferencing a pointer beyond the bounds of the + * GTT. We would hope that the VBO's max_index would save us, but + * Mesa appears to hand us min/max values not clipped to the + * array object's _MaxElement, and _MaxElement frequently appears + * to be wrong anyway. + * + * The VBO spec allows application termination in this case, and it's + * probably a service to the poor programmer to do so rather than + * trying to just not render. + */ + assert(input->offset < input->bo->size); } else { if (input->bo != NULL) { /* Already-uploaded vertex data is present from a previous -- cgit v1.2.3 From f396263651b867be39800bfd274b1be3d78ef60f Mon Sep 17 00:00:00 2001 From: Eric Anholt Date: Tue, 4 Aug 2009 12:39:22 -0700 Subject: i965: Fix RECT shadow sampling by not losing the other texcoords. Bug #20821 (cherry picked from commit 191e028de20b2f954621b652aa77b06d0e93652a) --- src/mesa/drivers/dri/i965/brw_wm_fp.c | 6 +++++- 1 file changed, 5 insertions(+), 1 deletion(-) (limited to 'src') diff --git a/src/mesa/drivers/dri/i965/brw_wm_fp.c b/src/mesa/drivers/dri/i965/brw_wm_fp.c index 49aad281d7..8cf3a1fd13 100644 --- a/src/mesa/drivers/dri/i965/brw_wm_fp.c +++ b/src/mesa/drivers/dri/i965/brw_wm_fp.c @@ -705,7 +705,11 @@ static void precalc_tex( struct brw_wm_compile *c, tmpcoord, 0, inst->SrcReg[0], - scale, + src_swizzle(scale, + SWIZZLE_X, + SWIZZLE_Y, + SWIZZLE_ONE, + SWIZZLE_ONE), src_undef()); coord = src_reg_from_dst(tmpcoord); -- cgit v1.2.3 From 83e6c67363d77d89ab4b6e2e202c8634d582c907 Mon Sep 17 00:00:00 2001 From: Eric Anholt Date: Tue, 4 Aug 2009 13:42:30 -0700 Subject: i965: Spell "conditional" correctly. --- src/mesa/drivers/dri/i965/brw_eu.c | 2 +- src/mesa/drivers/dri/i965/brw_eu_emit.c | 26 +++++++++++++------------- src/mesa/drivers/dri/i965/brw_structs.h | 2 +- 3 files changed, 15 insertions(+), 15 deletions(-) (limited to 'src') diff --git a/src/mesa/drivers/dri/i965/brw_eu.c b/src/mesa/drivers/dri/i965/brw_eu.c index c53efba599..1df561386e 100644 --- a/src/mesa/drivers/dri/i965/brw_eu.c +++ b/src/mesa/drivers/dri/i965/brw_eu.c @@ -62,7 +62,7 @@ void brw_set_predicate_control( struct brw_compile *p, GLuint pc ) void brw_set_conditionalmod( struct brw_compile *p, GLuint conditional ) { - p->current->header.destreg__conditonalmod = conditional; + p->current->header.destreg__conditionalmod = conditional; } void brw_set_access_mode( struct brw_compile *p, GLuint access_mode ) diff --git a/src/mesa/drivers/dri/i965/brw_eu_emit.c b/src/mesa/drivers/dri/i965/brw_eu_emit.c index 2a147fb8c3..48243e1574 100644 --- a/src/mesa/drivers/dri/i965/brw_eu_emit.c +++ b/src/mesa/drivers/dri/i965/brw_eu_emit.c @@ -377,8 +377,8 @@ static struct brw_instruction *next_insn( struct brw_compile *p, /* Reset this one-shot flag: */ - if (p->current->header.destreg__conditonalmod) { - p->current->header.destreg__conditonalmod = 0; + if (p->current->header.destreg__conditionalmod) { + p->current->header.destreg__conditionalmod = 0; p->current->header.predicate_control = BRW_PREDICATE_NORMAL; } @@ -746,7 +746,7 @@ void brw_CMP(struct brw_compile *p, { struct brw_instruction *insn = next_insn(p, BRW_OPCODE_CMP); - insn->header.destreg__conditonalmod = conditional; + insn->header.destreg__conditionalmod = conditional; brw_set_dest(insn, dest); brw_set_src0(insn, src0); brw_set_src1(insn, src1); @@ -790,7 +790,7 @@ void brw_math( struct brw_compile *p, * instructions. */ insn->header.predicate_control = 0; - insn->header.destreg__conditonalmod = msg_reg_nr; + insn->header.destreg__conditionalmod = msg_reg_nr; brw_set_dest(insn, dest); brw_set_src0(insn, src); @@ -826,7 +826,7 @@ void brw_math_16( struct brw_compile *p, brw_set_compression_control(p, BRW_COMPRESSION_NONE); insn = next_insn(p, BRW_OPCODE_SEND); - insn->header.destreg__conditonalmod = msg_reg_nr; + insn->header.destreg__conditionalmod = msg_reg_nr; brw_set_dest(insn, dest); brw_set_src0(insn, src); @@ -842,7 +842,7 @@ void brw_math_16( struct brw_compile *p, */ insn = next_insn(p, BRW_OPCODE_SEND); insn->header.compression_control = BRW_COMPRESSION_2NDHALF; - insn->header.destreg__conditonalmod = msg_reg_nr+1; + insn->header.destreg__conditionalmod = msg_reg_nr+1; brw_set_dest(insn, offset(dest,1)); brw_set_src0(insn, src); @@ -888,7 +888,7 @@ void brw_dp_WRITE_16( struct brw_compile *p, insn->header.predicate_control = 0; /* XXX */ insn->header.compression_control = BRW_COMPRESSION_NONE; - insn->header.destreg__conditonalmod = msg_reg_nr; + insn->header.destreg__conditionalmod = msg_reg_nr; brw_set_dest(insn, dest); brw_set_src0(insn, src); @@ -933,7 +933,7 @@ void brw_dp_READ_16( struct brw_compile *p, insn->header.predicate_control = 0; /* XXX */ insn->header.compression_control = BRW_COMPRESSION_NONE; - insn->header.destreg__conditonalmod = msg_reg_nr; + insn->header.destreg__conditionalmod = msg_reg_nr; brw_set_dest(insn, dest); /* UW? */ brw_set_src0(insn, retype(brw_vec8_grf(0, 0), BRW_REGISTER_TYPE_UW)); @@ -986,7 +986,7 @@ void brw_dp_READ_4( struct brw_compile *p, insn->header.predicate_control = BRW_PREDICATE_NONE; insn->header.compression_control = BRW_COMPRESSION_NONE; - insn->header.destreg__conditonalmod = msg_reg_nr; + insn->header.destreg__conditionalmod = msg_reg_nr; insn->header.mask_control = BRW_MASK_DISABLE; /* cast dest to a uword[8] vector */ @@ -1059,7 +1059,7 @@ void brw_dp_READ_4_vs(struct brw_compile *p, insn->header.predicate_control = BRW_PREDICATE_NONE; insn->header.compression_control = BRW_COMPRESSION_NONE; - insn->header.destreg__conditonalmod = msg_reg_nr; + insn->header.destreg__conditionalmod = msg_reg_nr; insn->header.mask_control = BRW_MASK_DISABLE; /*insn->header.access_mode = BRW_ALIGN_16;*/ @@ -1092,7 +1092,7 @@ void brw_fb_WRITE(struct brw_compile *p, insn->header.predicate_control = 0; /* XXX */ insn->header.compression_control = BRW_COMPRESSION_NONE; - insn->header.destreg__conditonalmod = msg_reg_nr; + insn->header.destreg__conditionalmod = msg_reg_nr; brw_set_dest(insn, dest); brw_set_src0(insn, src0); @@ -1187,7 +1187,7 @@ void brw_SAMPLE(struct brw_compile *p, insn->header.predicate_control = 0; /* XXX */ insn->header.compression_control = BRW_COMPRESSION_NONE; - insn->header.destreg__conditonalmod = msg_reg_nr; + insn->header.destreg__conditionalmod = msg_reg_nr; brw_set_dest(insn, dest); brw_set_src0(insn, src0); @@ -1238,7 +1238,7 @@ void brw_urb_WRITE(struct brw_compile *p, brw_set_src0(insn, src0); brw_set_src1(insn, brw_imm_d(0)); - insn->header.destreg__conditonalmod = msg_reg_nr; + insn->header.destreg__conditionalmod = msg_reg_nr; brw_set_urb_message(insn, allocate, diff --git a/src/mesa/drivers/dri/i965/brw_structs.h b/src/mesa/drivers/dri/i965/brw_structs.h index 89e2981203..6011a4aa0a 100644 --- a/src/mesa/drivers/dri/i965/brw_structs.h +++ b/src/mesa/drivers/dri/i965/brw_structs.h @@ -1168,7 +1168,7 @@ struct brw_instruction GLuint predicate_control:4; GLuint predicate_inverse:1; GLuint execution_size:3; - GLuint destreg__conditonalmod:4; /* destreg - send, conditionalmod - others */ + GLuint destreg__conditionalmod:4; /* destreg - send, conditionalmod - others */ GLuint pad0:2; GLuint debug_control:1; GLuint saturate:1; -- cgit v1.2.3 From 3d6c73513c5918a3f833ebf402803946b78828b8 Mon Sep 17 00:00:00 2001 From: Eric Anholt Date: Tue, 4 Aug 2009 14:13:27 -0700 Subject: i965: Emit conditional code updates as required for GLSL VS if statements. Previously, we'd be branching based on whatever condition code happened to be laying around. (cherry picked from commit 7007f8b352763af89805f287153cb7972bff0523) --- src/mesa/drivers/dri/i965/brw_vs_emit.c | 13 +++++++++++++ 1 file changed, 13 insertions(+) (limited to 'src') diff --git a/src/mesa/drivers/dri/i965/brw_vs_emit.c b/src/mesa/drivers/dri/i965/brw_vs_emit.c index b95079a5d1..4887cdea5b 100644 --- a/src/mesa/drivers/dri/i965/brw_vs_emit.c +++ b/src/mesa/drivers/dri/i965/brw_vs_emit.c @@ -1448,6 +1448,19 @@ void brw_vs_emit(struct brw_vs_compile *c ) "unknown"); } + /* Set the predication update on the last instruction of the native + * instruction sequence. + * + * This would be problematic if it was set on a math instruction, + * but that shouldn't be the case with the current GLSL compiler. + */ + if (inst->CondUpdate) { + struct brw_instruction *hw_insn = &p->store[p->nr_insn - 1]; + + assert(hw_insn->header.destreg__conditionalmod == 0); + hw_insn->header.destreg__conditionalmod = BRW_CONDITIONAL_NZ; + } + if ((inst->DstReg.File == PROGRAM_OUTPUT) && (inst->DstReg.Index != VERT_RESULT_HPOS) && c->output_regs[inst->DstReg.Index].used_in_src) { -- cgit v1.2.3 From a0b7850f1da6e367ed3380fbf0a59fa5af992546 Mon Sep 17 00:00:00 2001 From: Brian Paul Date: Thu, 18 Jun 2009 09:23:58 -0600 Subject: i965: asst clean-ups, etc in brw_vs_emit() (cherry picked from commit fd7d764514c540987549c3ea88a2d669b0f0ea58) --- src/mesa/drivers/dri/i965/brw_vs_emit.c | 21 ++++++++++----------- 1 file changed, 10 insertions(+), 11 deletions(-) (limited to 'src') diff --git a/src/mesa/drivers/dri/i965/brw_vs_emit.c b/src/mesa/drivers/dri/i965/brw_vs_emit.c index 4887cdea5b..7c09222281 100644 --- a/src/mesa/drivers/dri/i965/brw_vs_emit.c +++ b/src/mesa/drivers/dri/i965/brw_vs_emit.c @@ -1195,15 +1195,14 @@ post_vs_emit( struct brw_vs_compile *c, */ void brw_vs_emit(struct brw_vs_compile *c ) { -#define MAX_IFSN 32 +#define MAX_IF_DEPTH 32 struct brw_compile *p = &c->func; - GLuint nr_insns = c->vp->program.Base.NumInstructions; - GLuint insn, if_insn = 0; + const GLuint nr_insns = c->vp->program.Base.NumInstructions; + GLuint insn, if_depth = 0; GLuint end_offset = 0; struct brw_instruction *end_inst, *last_inst; - struct brw_instruction *if_inst[MAX_IFSN]; - struct brw_indirect stack_index = brw_indirect(0, 0); - + struct brw_instruction *if_inst[MAX_IF_DEPTH]; + const struct brw_indirect stack_index = brw_indirect(0, 0); GLuint index; GLuint file; @@ -1394,15 +1393,15 @@ void brw_vs_emit(struct brw_vs_compile *c ) emit_xpd(p, dst, args[0], args[1]); break; case OPCODE_IF: - assert(if_insn < MAX_IFSN); - if_inst[if_insn++] = brw_IF(p, BRW_EXECUTE_8); + assert(if_depth < MAX_IF_DEPTH); + if_inst[if_depth++] = brw_IF(p, BRW_EXECUTE_8); break; case OPCODE_ELSE: - if_inst[if_insn-1] = brw_ELSE(p, if_inst[if_insn-1]); + if_inst[if_depth-1] = brw_ELSE(p, if_inst[if_depth-1]); break; case OPCODE_ENDIF: - assert(if_insn > 0); - brw_ENDIF(p, if_inst[--if_insn]); + assert(if_depth > 0); + brw_ENDIF(p, if_inst[--if_depth]); break; case OPCODE_BRA: brw_set_predicate_control(p, BRW_PREDICATE_NORMAL); -- cgit v1.2.3 From 8c764d5c34548d56e912cdc5990adffb0e8fd9f7 Mon Sep 17 00:00:00 2001 From: Eric Anholt Date: Tue, 4 Aug 2009 18:02:31 -0700 Subject: i965: Respect CondSwizzle in OPCODE_IF. Fixes piglit glsl-vs-if-bool and progs/glsl/twoside, and will likely be useful for the looping code. Bug #18992 (cherry picked from commit 78c022acd0b37bf8b32f04313d76255255e769c1) (cherry picked from commit 63d7a2f53fb38e170f4e55f2b599e918edf2c512) --- src/mesa/drivers/dri/i965/brw_vs_emit.c | 22 +++++++++++++++++++++- 1 file changed, 21 insertions(+), 1 deletion(-) (limited to 'src') diff --git a/src/mesa/drivers/dri/i965/brw_vs_emit.c b/src/mesa/drivers/dri/i965/brw_vs_emit.c index 7c09222281..72cd36a603 100644 --- a/src/mesa/drivers/dri/i965/brw_vs_emit.c +++ b/src/mesa/drivers/dri/i965/brw_vs_emit.c @@ -1190,6 +1190,23 @@ post_vs_emit( struct brw_vs_compile *c, brw_set_src1(end_inst, brw_imm_d(offset * 16)); } +static uint32_t +get_predicate(uint32_t swizzle) +{ + switch (swizzle) { + case SWIZZLE_XXXX: + return BRW_PREDICATE_ALIGN16_REPLICATE_X; + case SWIZZLE_YYYY: + return BRW_PREDICATE_ALIGN16_REPLICATE_Y; + case SWIZZLE_ZZZZ: + return BRW_PREDICATE_ALIGN16_REPLICATE_Z; + case SWIZZLE_WWWW: + return BRW_PREDICATE_ALIGN16_REPLICATE_W; + default: + _mesa_problem(NULL, "Unexpected predicate: 0x%08x\n", swizzle); + return BRW_PREDICATE_NORMAL; + } +} /* Emit the vertex program instructions here. */ @@ -1394,7 +1411,10 @@ void brw_vs_emit(struct brw_vs_compile *c ) break; case OPCODE_IF: assert(if_depth < MAX_IF_DEPTH); - if_inst[if_depth++] = brw_IF(p, BRW_EXECUTE_8); + if_inst[if_depth] = brw_IF(p, BRW_EXECUTE_8); + if_inst[if_depth]->header.predicate_control = + get_predicate(inst->DstReg.CondSwizzle); + if_depth++; break; case OPCODE_ELSE: if_inst[if_depth-1] = brw_ELSE(p, if_inst[if_depth-1]); -- cgit v1.2.3 From 94d3b832cc5c73544c478804531e16644483d8be Mon Sep 17 00:00:00 2001 From: Eric Anholt Date: Wed, 5 Aug 2009 20:12:15 -0700 Subject: i965: Fix source depth reg setting for FSes reading and writing to depth. For some IZ setups, we'd forget to account for the source depth register being present, so we'd both read the wrong reg, and write output depth to the wrong reg. Bug #22603. (cherry picked from commit f44916414ecd2b888c8a680d56b7467ccdff6886) --- src/mesa/drivers/dri/i965/brw_wm.c | 2 ++ src/mesa/drivers/dri/i965/brw_wm.h | 1 + src/mesa/drivers/dri/i965/brw_wm_iz.c | 3 ++- 3 files changed, 5 insertions(+), 1 deletion(-) (limited to 'src') diff --git a/src/mesa/drivers/dri/i965/brw_wm.c b/src/mesa/drivers/dri/i965/brw_wm.c index 8a3b7df9c7..d4e22d5939 100644 --- a/src/mesa/drivers/dri/i965/brw_wm.c +++ b/src/mesa/drivers/dri/i965/brw_wm.c @@ -202,6 +202,7 @@ static void brw_wm_populate_key( struct brw_context *brw, /* BRW_NEW_FRAGMENT_PROGRAM */ const struct brw_fragment_program *fp = (struct brw_fragment_program *)brw->fragment_program; + GLboolean uses_depth = (fp->program.Base.InputsRead & (1 << FRAG_ATTRIB_WPOS)) != 0; GLuint lookup = 0; GLuint line_aa; GLuint i; @@ -263,6 +264,7 @@ static void brw_wm_populate_key( struct brw_context *brw, brw_wm_lookup_iz(line_aa, lookup, + uses_depth, key); diff --git a/src/mesa/drivers/dri/i965/brw_wm.h b/src/mesa/drivers/dri/i965/brw_wm.h index 295fed851b..307b5e59dd 100644 --- a/src/mesa/drivers/dri/i965/brw_wm.h +++ b/src/mesa/drivers/dri/i965/brw_wm.h @@ -286,6 +286,7 @@ void brw_wm_print_program( struct brw_wm_compile *c, void brw_wm_lookup_iz( GLuint line_aa, GLuint lookup, + GLboolean ps_uses_depth, struct brw_wm_prog_key *key ); GLboolean brw_wm_is_glsl(const struct gl_fragment_program *fp); diff --git a/src/mesa/drivers/dri/i965/brw_wm_iz.c b/src/mesa/drivers/dri/i965/brw_wm_iz.c index bd60ac9b31..7e2b1c79de 100644 --- a/src/mesa/drivers/dri/i965/brw_wm_iz.c +++ b/src/mesa/drivers/dri/i965/brw_wm_iz.c @@ -118,6 +118,7 @@ const struct { void brw_wm_lookup_iz( GLuint line_aa, GLuint lookup, + GLboolean ps_uses_depth, struct brw_wm_prog_key *key ) { GLuint reg = 2; @@ -127,7 +128,7 @@ void brw_wm_lookup_iz( GLuint line_aa, if (lookup & IZ_PS_COMPUTES_DEPTH_BIT) key->computes_depth = 1; - if (wm_iz_table[lookup].sd_present) { + if (wm_iz_table[lookup].sd_present || ps_uses_depth) { key->source_depth_reg = reg; reg += 2; } -- cgit v1.2.3 From 945d34e88ab7413299227fea56acc746010bb2e9 Mon Sep 17 00:00:00 2001 From: Eric Anholt Date: Fri, 7 Aug 2009 18:09:31 -0700 Subject: intel: Align untiled region height to 2 according to 965 docs. This may or may not be required pre-965, but it doesn't seem unlikely, and I'd rather be safe. (cherry picked from commit b053474378633249be0e9f24010650ffb816229a) --- src/mesa/drivers/dri/intel/intel_regions.c | 7 +++++++ 1 file changed, 7 insertions(+) (limited to 'src') diff --git a/src/mesa/drivers/dri/intel/intel_regions.c b/src/mesa/drivers/dri/intel/intel_regions.c index 0aa5b8c02c..b8d2dec4cb 100644 --- a/src/mesa/drivers/dri/intel/intel_regions.c +++ b/src/mesa/drivers/dri/intel/intel_regions.c @@ -114,6 +114,13 @@ intel_region_alloc(struct intel_context *intel, { dri_bo *buffer; + /* If we're untiled, we have to align to 2 rows high because the + * data port accesses 2x2 blocks even if the bottom row isn't to be + * rendered, so failure to align means we could walk off the end of the + * GTT and fault. + */ + height = ALIGN(height, 2); + if (expect_accelerated_upload) { buffer = drm_intel_bo_alloc_for_render(intel->bufmgr, "region", pitch * cpp * height, 64); -- cgit v1.2.3 From a9504edee344936ef496341174d4c2d76ee0c44d Mon Sep 17 00:00:00 2001 From: Eric Anholt Date: Tue, 18 Aug 2009 18:48:54 -0700 Subject: intel: Align cubemap texture height to its padding requirements. (cherry picked from commit a70e1315846cd5e8d6f2b622821ff8262fe7179d) (cherry picked from commit 29e51c3872531366570d032147abad50f8a3c1af) --- src/mesa/drivers/dri/i965/brw_tex_layout.c | 10 ++++++++++ 1 file changed, 10 insertions(+) (limited to 'src') diff --git a/src/mesa/drivers/dri/i965/brw_tex_layout.c b/src/mesa/drivers/dri/i965/brw_tex_layout.c index 51a617fcb4..3ab27c26ef 100644 --- a/src/mesa/drivers/dri/i965/brw_tex_layout.c +++ b/src/mesa/drivers/dri/i965/brw_tex_layout.c @@ -119,6 +119,16 @@ GLboolean brw_miptree_layout( struct intel_context *intel, struct intel_mipmap_t } } + /* The 965's sampler lays cachelines out according to how accesses + * in the texture surfaces run, so they may be "vertical" through + * memory. As a result, the docs say in Surface Padding Requirements: + * Sampling Engine Surfaces that two extra rows of padding are required. + * We don't know of similar requirements for pre-965, but given that + * those docs are silent on padding requirements in general, let's play + * it safe. + */ + if (mt->target == GL_TEXTURE_CUBE_MAP) + mt->total_height += 2; break; } -- cgit v1.2.3 From 1d1dfc96c48681c455e1174a60bd325b33a2784f Mon Sep 17 00:00:00 2001 From: Eric Anholt Date: Thu, 16 Jul 2009 15:57:22 -0700 Subject: i965: Add missing state dependency of sf_unit on _NEW_BUFFERS. (cherry picked from commit 99174e7630676307f618c252755a20ba61ad9158) --- src/mesa/drivers/dri/i965/brw_sf_state.c | 6 ++++-- 1 file changed, 4 insertions(+), 2 deletions(-) (limited to 'src') diff --git a/src/mesa/drivers/dri/i965/brw_sf_state.c b/src/mesa/drivers/dri/i965/brw_sf_state.c index c99918724b..38733c8c73 100644 --- a/src/mesa/drivers/dri/i965/brw_sf_state.c +++ b/src/mesa/drivers/dri/i965/brw_sf_state.c @@ -233,7 +233,7 @@ sf_unit_create_from_key(struct brw_context *brw, struct brw_sf_unit_key *key, else if (sf.sf6.line_width <= 0x2) sf.sf6.line_width = 0; - /* _NEW_POINT */ + /* _NEW_BUFFERS */ key->render_to_fbo = brw->intel.ctx.DrawBuffer->Name != 0; if (!key->render_to_fbo) { /* Rendering to an OpenGL window */ @@ -263,6 +263,7 @@ sf_unit_create_from_key(struct brw_context *brw, struct brw_sf_unit_key *key, } /* XXX clamp max depends on AA vs. non-AA */ + /* _NEW_POINT */ sf.sf7.sprite_point = key->point_sprite; sf.sf7.point_size = CLAMP(rint(key->point_size), 1, 255) * (1<<3); sf.sf7.use_point_size_state = !key->point_attenuated; @@ -328,7 +329,8 @@ const struct brw_tracked_state brw_sf_unit = { .mesa = (_NEW_POLYGON | _NEW_LINE | _NEW_POINT | - _NEW_SCISSOR), + _NEW_SCISSOR | + _NEW_BUFFERS), .brw = BRW_NEW_URB_FENCE, .cache = (CACHE_NEW_SF_VP | CACHE_NEW_SF_PROG) -- cgit v1.2.3 From 04081a164ca6160404d87dccbfc641bfd46428e0 Mon Sep 17 00:00:00 2001 From: Eric Anholt Date: Fri, 19 Jun 2009 21:43:22 -0700 Subject: intel: Move intel_pixel_read.c to shared for use with i965. (cherry picked from commit dcfe0d66bfff9a55741aee298b7ffb051a48f0d3) --- src/mesa/drivers/dri/i915/intel_pixel_read.c | 307 +------------------------- src/mesa/drivers/dri/intel/intel_pixel_read.c | 306 +++++++++++++++++++++++++ 2 files changed, 307 insertions(+), 306 deletions(-) mode change 100644 => 120000 src/mesa/drivers/dri/i915/intel_pixel_read.c create mode 100644 src/mesa/drivers/dri/intel/intel_pixel_read.c (limited to 'src') diff --git a/src/mesa/drivers/dri/i915/intel_pixel_read.c b/src/mesa/drivers/dri/i915/intel_pixel_read.c deleted file mode 100644 index 56087aacd4..0000000000 --- a/src/mesa/drivers/dri/i915/intel_pixel_read.c +++ /dev/null @@ -1,306 +0,0 @@ -/************************************************************************** - * - * Copyright 2003 Tungsten Graphics, Inc., Cedar Park, Texas. - * All Rights Reserved. - * - * 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, sub license, 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 (including the - * next paragraph) 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 NON-INFRINGEMENT. - * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS 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 "main/glheader.h" -#include "main/enums.h" -#include "main/mtypes.h" -#include "main/macros.h" -#include "main/image.h" -#include "main/bufferobj.h" -#include "swrast/swrast.h" - -#include "intel_screen.h" -#include "intel_context.h" -#include "intel_batchbuffer.h" -#include "intel_blit.h" -#include "intel_buffers.h" -#include "intel_regions.h" -#include "intel_pixel.h" -#include "intel_buffer_objects.h" - -/* For many applications, the new ability to pull the source buffers - * back out of the GTT and then do the packing/conversion operations - * in software will be as much of an improvement as trying to get the - * blitter and/or texture engine to do the work. - * - * This step is gated on private backbuffers. - * - * Obviously the frontbuffer can't be pulled back, so that is either - * an argument for blit/texture readpixels, or for blitting to a - * temporary and then pulling that back. - * - * When the destination is a pbo, however, it's not clear if it is - * ever going to be pulled to main memory (though the access param - * will be a good hint). So it sounds like we do want to be able to - * choose between blit/texture implementation on the gpu and pullback - * and cpu-based copying. - * - * Unless you can magically turn client memory into a PBO for the - * duration of this call, there will be a cpu-based copying step in - * any case. - */ - - -static GLboolean -do_texture_readpixels(GLcontext * ctx, - GLint x, GLint y, GLsizei width, GLsizei height, - GLenum format, GLenum type, - const struct gl_pixelstore_attrib *pack, - struct intel_region *dest_region) -{ -#if 0 - struct intel_context *intel = intel_context(ctx); - intelScreenPrivate *screen = intel->intelScreen; - GLint pitch = pack->RowLength ? pack->RowLength : width; - __DRIdrawablePrivate *dPriv = intel->driDrawable; - int textureFormat; - GLenum glTextureFormat; - int destFormat, depthFormat, destPitch; - drm_clip_rect_t tmp; - - if (INTEL_DEBUG & DEBUG_PIXEL) - fprintf(stderr, "%s\n", __FUNCTION__); - - - if (ctx->_ImageTransferState || - pack->SwapBytes || pack->LsbFirst || !pack->Invert) { - if (INTEL_DEBUG & DEBUG_PIXEL) - fprintf(stderr, "%s: check_color failed\n", __FUNCTION__); - return GL_FALSE; - } - - intel->vtbl.meta_texrect_source(intel, intel_readbuf_region(intel)); - - if (!intel->vtbl.meta_render_dest(intel, dest_region, type, format)) { - if (INTEL_DEBUG & DEBUG_PIXEL) - fprintf(stderr, "%s: couldn't set dest %s/%s\n", - __FUNCTION__, - _mesa_lookup_enum_by_nr(type), - _mesa_lookup_enum_by_nr(format)); - return GL_FALSE; - } - - LOCK_HARDWARE(intel); - - if (intel->driDrawable->numClipRects) { - intel->vtbl.install_meta_state(intel); - intel->vtbl.meta_no_depth_write(intel); - intel->vtbl.meta_no_stencil_write(intel); - - if (!driClipRectToFramebuffer(ctx->ReadBuffer, &x, &y, &width, &height)) { - UNLOCK_HARDWARE(intel); - SET_STATE(i830, state); - if (INTEL_DEBUG & DEBUG_PIXEL) - fprintf(stderr, "%s: cliprect failed\n", __FUNCTION__); - return GL_TRUE; - } - - y = dPriv->h - y - height; - x += dPriv->x; - y += dPriv->y; - - - /* Set the frontbuffer up as a large rectangular texture. - */ - intel->vtbl.meta_tex_rect_source(intel, src_region, textureFormat); - - - intel->vtbl.meta_texture_blend_replace(i830, glTextureFormat); - - - /* Set the 3d engine to draw into the destination region: - */ - - intel->vtbl.meta_draw_region(intel, dest_region); - intel->vtbl.meta_draw_format(intel, destFormat, depthFormat); /* ?? */ - - - /* Draw a single quad, no cliprects: - */ - intel->vtbl.meta_disable_cliprects(intel); - - intel->vtbl.draw_quad(intel, - 0, width, 0, height, - 0x00ff00ff, x, x + width, y, y + height); - - intel->vtbl.leave_meta_state(intel); - } - UNLOCK_HARDWARE(intel); - - intel_region_wait_fence(ctx, dest_region); /* required by GL */ - return GL_TRUE; -#endif - - return GL_FALSE; -} - - - - -static GLboolean -do_blit_readpixels(GLcontext * ctx, - GLint x, GLint y, GLsizei width, GLsizei height, - GLenum format, GLenum type, - const struct gl_pixelstore_attrib *pack, GLvoid * pixels) -{ - struct intel_context *intel = intel_context(ctx); - struct intel_region *src = intel_readbuf_region(intel); - struct intel_buffer_object *dst = intel_buffer_object(pack->BufferObj); - GLuint dst_offset; - GLuint rowLength; - - if (INTEL_DEBUG & DEBUG_PIXEL) - _mesa_printf("%s\n", __FUNCTION__); - - if (!src) - return GL_FALSE; - - if (dst) { - /* XXX This validation should be done by core mesa: - */ - if (!_mesa_validate_pbo_access(2, pack, width, height, 1, - format, type, pixels)) { - _mesa_error(ctx, GL_INVALID_OPERATION, "glDrawPixels"); - return GL_TRUE; - } - } - else { - /* PBO only for now: - */ - if (INTEL_DEBUG & DEBUG_PIXEL) - _mesa_printf("%s - not PBO\n", __FUNCTION__); - return GL_FALSE; - } - - - if (ctx->_ImageTransferState || - !intel_check_blit_format(src, format, type)) { - if (INTEL_DEBUG & DEBUG_PIXEL) - _mesa_printf("%s - bad format for blit\n", __FUNCTION__); - return GL_FALSE; - } - - if (pack->Alignment != 1 || pack->SwapBytes || pack->LsbFirst) { - if (INTEL_DEBUG & DEBUG_PIXEL) - _mesa_printf("%s: bad packing params\n", __FUNCTION__); - return GL_FALSE; - } - - if (pack->RowLength > 0) - rowLength = pack->RowLength; - else - rowLength = width; - - if (pack->Invert) { - if (INTEL_DEBUG & DEBUG_PIXEL) - _mesa_printf("%s: MESA_PACK_INVERT not done yet\n", __FUNCTION__); - return GL_FALSE; - } - else { - rowLength = -rowLength; - } - - /* XXX 64-bit cast? */ - dst_offset = (GLuint) _mesa_image_address(2, pack, pixels, width, height, - format, type, 0, 0, 0); - - - /* Although the blits go on the command buffer, need to do this and - * fire with lock held to guarentee cliprects are correct. - */ - intelFlush(&intel->ctx); - LOCK_HARDWARE(intel); - - if (intel->driDrawable->numClipRects) { - GLboolean all = (width * height * src->cpp == dst->Base.Size && - x == 0 && dst_offset == 0); - - dri_bo *dst_buffer = intel_bufferobj_buffer(intel, dst, - all ? INTEL_WRITE_FULL : - INTEL_WRITE_PART); - __DRIdrawablePrivate *dPriv = intel->driDrawable; - int nbox = dPriv->numClipRects; - drm_clip_rect_t *box = dPriv->pClipRects; - drm_clip_rect_t rect; - drm_clip_rect_t src_rect; - int i; - - src_rect.x1 = dPriv->x + x; - src_rect.y1 = dPriv->y + dPriv->h - (y + height); - src_rect.x2 = src_rect.x1 + width; - src_rect.y2 = src_rect.y1 + height; - - - - for (i = 0; i < nbox; i++) { - if (!intel_intersect_cliprects(&rect, &src_rect, &box[i])) - continue; - - intelEmitCopyBlit(intel, - src->cpp, - src->pitch, src->buffer, 0, src->tiling, - rowLength, dst_buffer, dst_offset, GL_FALSE, - rect.x1, - rect.y1, - rect.x1 - src_rect.x1, - rect.y2 - src_rect.y2, - rect.x2 - rect.x1, rect.y2 - rect.y1, - GL_COPY); - } - } - UNLOCK_HARDWARE(intel); - - if (INTEL_DEBUG & DEBUG_PIXEL) - _mesa_printf("%s - DONE\n", __FUNCTION__); - - return GL_TRUE; -} - -void -intelReadPixels(GLcontext * ctx, - GLint x, GLint y, GLsizei width, GLsizei height, - GLenum format, GLenum type, - const struct gl_pixelstore_attrib *pack, GLvoid * pixels) -{ - if (INTEL_DEBUG & DEBUG_PIXEL) - fprintf(stderr, "%s\n", __FUNCTION__); - - intelFlush(ctx); - - if (do_blit_readpixels - (ctx, x, y, width, height, format, type, pack, pixels)) - return; - - if (do_texture_readpixels - (ctx, x, y, width, height, format, type, pack, pixels)) - return; - - if (INTEL_DEBUG & DEBUG_PIXEL) - _mesa_printf("%s: fallback to swrast\n", __FUNCTION__); - - _swrast_ReadPixels(ctx, x, y, width, height, format, type, pack, pixels); -} diff --git a/src/mesa/drivers/dri/i915/intel_pixel_read.c b/src/mesa/drivers/dri/i915/intel_pixel_read.c new file mode 120000 index 0000000000..cc4589f4d4 --- /dev/null +++ b/src/mesa/drivers/dri/i915/intel_pixel_read.c @@ -0,0 +1 @@ +../intel/intel_pixel_read.c \ No newline at end of file diff --git a/src/mesa/drivers/dri/intel/intel_pixel_read.c b/src/mesa/drivers/dri/intel/intel_pixel_read.c new file mode 100644 index 0000000000..2c57b470f5 --- /dev/null +++ b/src/mesa/drivers/dri/intel/intel_pixel_read.c @@ -0,0 +1,306 @@ +/************************************************************************** + * + * Copyright 2003 Tungsten Graphics, Inc., Cedar Park, Texas. + * All Rights Reserved. + * + * 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, sub license, 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 (including the + * next paragraph) 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 NON-INFRINGEMENT. + * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS 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 "main/glheader.h" +#include "main/enums.h" +#include "main/mtypes.h" +#include "main/macros.h" +#include "main/image.h" +#include "main/bufferobj.h" +#include "swrast/swrast.h" + +#include "intel_screen.h" +#include "intel_context.h" +#include "intel_batchbuffer.h" +#include "intel_blit.h" +#include "intel_buffers.h" +#include "intel_regions.h" +#include "intel_pixel.h" +#include "intel_buffer_objects.h" + +/* For many applications, the new ability to pull the source buffers + * back out of the GTT and then do the packing/conversion operations + * in software will be as much of an improvement as trying to get the + * blitter and/or texture engine to do the work. + * + * This step is gated on private backbuffers. + * + * Obviously the frontbuffer can't be pulled back, so that is either + * an argument for blit/texture readpixels, or for blitting to a + * temporary and then pulling that back. + * + * When the destination is a pbo, however, it's not clear if it is + * ever going to be pulled to main memory (though the access param + * will be a good hint). So it sounds like we do want to be able to + * choose between blit/texture implementation on the gpu and pullback + * and cpu-based copying. + * + * Unless you can magically turn client memory into a PBO for the + * duration of this call, there will be a cpu-based copying step in + * any case. + */ + + +static GLboolean +do_texture_readpixels(GLcontext * ctx, + GLint x, GLint y, GLsizei width, GLsizei height, + GLenum format, GLenum type, + const struct gl_pixelstore_attrib *pack, + struct intel_region *dest_region) +{ +#if 0 + struct intel_context *intel = intel_context(ctx); + intelScreenPrivate *screen = intel->intelScreen; + GLint pitch = pack->RowLength ? pack->RowLength : width; + __DRIdrawablePrivate *dPriv = intel->driDrawable; + int textureFormat; + GLenum glTextureFormat; + int destFormat, depthFormat, destPitch; + drm_clip_rect_t tmp; + + if (INTEL_DEBUG & DEBUG_PIXEL) + fprintf(stderr, "%s\n", __FUNCTION__); + + + if (ctx->_ImageTransferState || + pack->SwapBytes || pack->LsbFirst || !pack->Invert) { + if (INTEL_DEBUG & DEBUG_PIXEL) + fprintf(stderr, "%s: check_color failed\n", __FUNCTION__); + return GL_FALSE; + } + + intel->vtbl.meta_texrect_source(intel, intel_readbuf_region(intel)); + + if (!intel->vtbl.meta_render_dest(intel, dest_region, type, format)) { + if (INTEL_DEBUG & DEBUG_PIXEL) + fprintf(stderr, "%s: couldn't set dest %s/%s\n", + __FUNCTION__, + _mesa_lookup_enum_by_nr(type), + _mesa_lookup_enum_by_nr(format)); + return GL_FALSE; + } + + LOCK_HARDWARE(intel); + + if (intel->driDrawable->numClipRects) { + intel->vtbl.install_meta_state(intel); + intel->vtbl.meta_no_depth_write(intel); + intel->vtbl.meta_no_stencil_write(intel); + + if (!driClipRectToFramebuffer(ctx->ReadBuffer, &x, &y, &width, &height)) { + UNLOCK_HARDWARE(intel); + SET_STATE(i830, state); + if (INTEL_DEBUG & DEBUG_PIXEL) + fprintf(stderr, "%s: cliprect failed\n", __FUNCTION__); + return GL_TRUE; + } + + y = dPriv->h - y - height; + x += dPriv->x; + y += dPriv->y; + + + /* Set the frontbuffer up as a large rectangular texture. + */ + intel->vtbl.meta_tex_rect_source(intel, src_region, textureFormat); + + + intel->vtbl.meta_texture_blend_replace(i830, glTextureFormat); + + + /* Set the 3d engine to draw into the destination region: + */ + + intel->vtbl.meta_draw_region(intel, dest_region); + intel->vtbl.meta_draw_format(intel, destFormat, depthFormat); /* ?? */ + + + /* Draw a single quad, no cliprects: + */ + intel->vtbl.meta_disable_cliprects(intel); + + intel->vtbl.draw_quad(intel, + 0, width, 0, height, + 0x00ff00ff, x, x + width, y, y + height); + + intel->vtbl.leave_meta_state(intel); + } + UNLOCK_HARDWARE(intel); + + intel_region_wait_fence(ctx, dest_region); /* required by GL */ + return GL_TRUE; +#endif + + return GL_FALSE; +} + + + + +static GLboolean +do_blit_readpixels(GLcontext * ctx, + GLint x, GLint y, GLsizei width, GLsizei height, + GLenum format, GLenum type, + const struct gl_pixelstore_attrib *pack, GLvoid * pixels) +{ + struct intel_context *intel = intel_context(ctx); + struct intel_region *src = intel_readbuf_region(intel); + struct intel_buffer_object *dst = intel_buffer_object(pack->BufferObj); + GLuint dst_offset; + GLuint rowLength; + + if (INTEL_DEBUG & DEBUG_PIXEL) + _mesa_printf("%s\n", __FUNCTION__); + + if (!src) + return GL_FALSE; + + if (dst) { + /* XXX This validation should be done by core mesa: + */ + if (!_mesa_validate_pbo_access(2, pack, width, height, 1, + format, type, pixels)) { + _mesa_error(ctx, GL_INVALID_OPERATION, "glDrawPixels"); + return GL_TRUE; + } + } + else { + /* PBO only for now: + */ + if (INTEL_DEBUG & DEBUG_PIXEL) + _mesa_printf("%s - not PBO\n", __FUNCTION__); + return GL_FALSE; + } + + + if (ctx->_ImageTransferState || + !intel_check_blit_format(src, format, type)) { + if (INTEL_DEBUG & DEBUG_PIXEL) + _mesa_printf("%s - bad format for blit\n", __FUNCTION__); + return GL_FALSE; + } + + if (pack->Alignment != 1 || pack->SwapBytes || pack->LsbFirst) { + if (INTEL_DEBUG & DEBUG_PIXEL) + _mesa_printf("%s: bad packing params\n", __FUNCTION__); + return GL_FALSE; + } + + if (pack->RowLength > 0) + rowLength = pack->RowLength; + else + rowLength = width; + + if (pack->Invert) { + if (INTEL_DEBUG & DEBUG_PIXEL) + _mesa_printf("%s: MESA_PACK_INVERT not done yet\n", __FUNCTION__); + return GL_FALSE; + } + else { + rowLength = -rowLength; + } + + /* XXX 64-bit cast? */ + dst_offset = (GLuint) _mesa_image_address(2, pack, pixels, width, height, + format, type, 0, 0, 0); + + + /* Although the blits go on the command buffer, need to do this and + * fire with lock held to guarentee cliprects are correct. + */ + intelFlush(&intel->ctx); + LOCK_HARDWARE(intel); + + if (intel->driDrawable->numClipRects) { + GLboolean all = (width * height * src->cpp == dst->Base.Size && + x == 0 && dst_offset == 0); + + dri_bo *dst_buffer = intel_bufferobj_buffer(intel, dst, + all ? INTEL_WRITE_FULL : + INTEL_WRITE_PART); + __DRIdrawablePrivate *dPriv = intel->driDrawable; + int nbox = dPriv->numClipRects; + drm_clip_rect_t *box = dPriv->pClipRects; + drm_clip_rect_t rect; + drm_clip_rect_t src_rect; + int i; + + src_rect.x1 = dPriv->x + x; + src_rect.y1 = dPriv->y + dPriv->h - (y + height); + src_rect.x2 = src_rect.x1 + width; + src_rect.y2 = src_rect.y1 + height; + + + + for (i = 0; i < nbox; i++) { + if (!intel_intersect_cliprects(&rect, &src_rect, &box[i])) + continue; + + intelEmitCopyBlit(intel, + src->cpp, + src->pitch, src->buffer, 0, src->tiling, + rowLength, dst_buffer, dst_offset, GL_FALSE, + rect.x1, + rect.y1, + rect.x1 - src_rect.x1, + rect.y2 - src_rect.y2, + rect.x2 - rect.x1, rect.y2 - rect.y1, + GL_COPY); + } + } + UNLOCK_HARDWARE(intel); + + if (INTEL_DEBUG & DEBUG_PIXEL) + _mesa_printf("%s - DONE\n", __FUNCTION__); + + return GL_TRUE; +} + +void +intelReadPixels(GLcontext * ctx, + GLint x, GLint y, GLsizei width, GLsizei height, + GLenum format, GLenum type, + const struct gl_pixelstore_attrib *pack, GLvoid * pixels) +{ + if (INTEL_DEBUG & DEBUG_PIXEL) + fprintf(stderr, "%s\n", __FUNCTION__); + + intelFlush(ctx); + + if (do_blit_readpixels + (ctx, x, y, width, height, format, type, pack, pixels)) + return; + + if (do_texture_readpixels + (ctx, x, y, width, height, format, type, pack, pixels)) + return; + + if (INTEL_DEBUG & DEBUG_PIXEL) + _mesa_printf("%s: fallback to swrast\n", __FUNCTION__); + + _swrast_ReadPixels(ctx, x, y, width, height, format, type, pack, pixels); +} -- cgit v1.2.3 From 2855ee82c6d74066e8d9e44b17b2ce3b5782110e Mon Sep 17 00:00:00 2001 From: Eric Anholt Date: Fri, 19 Jun 2009 22:03:37 -0700 Subject: intel: Update Mesa state before span setup in glReadPixels. We could have mapped the wrong set of draw buffers. Noticed while looking into a DRI2 glean ReadPixels issue. (cherry picked from commit afc981ee46791838f3cb83e11eb33938aa3efc83) --- src/mesa/drivers/dri/i965/intel_pixel_read.c | 1 + src/mesa/drivers/dri/intel/intel_pixel.c | 4 +--- src/mesa/drivers/dri/intel/intel_pixel_read.c | 11 +++++++++++ 3 files changed, 13 insertions(+), 3 deletions(-) create mode 120000 src/mesa/drivers/dri/i965/intel_pixel_read.c (limited to 'src') diff --git a/src/mesa/drivers/dri/i965/intel_pixel_read.c b/src/mesa/drivers/dri/i965/intel_pixel_read.c new file mode 120000 index 0000000000..cc4589f4d4 --- /dev/null +++ b/src/mesa/drivers/dri/i965/intel_pixel_read.c @@ -0,0 +1 @@ +../intel/intel_pixel_read.c \ No newline at end of file diff --git a/src/mesa/drivers/dri/intel/intel_pixel.c b/src/mesa/drivers/dri/intel/intel_pixel.c index fc0ac0b79c..defb80f85d 100644 --- a/src/mesa/drivers/dri/intel/intel_pixel.c +++ b/src/mesa/drivers/dri/intel/intel_pixel.c @@ -342,10 +342,8 @@ intelInitPixelFuncs(struct dd_function_table *functions) functions->Bitmap = intelBitmap; functions->CopyPixels = intelCopyPixels; functions->DrawPixels = intelDrawPixels; -#ifdef I915 - functions->ReadPixels = intelReadPixels; -#endif } + functions->ReadPixels = intelReadPixels; } void diff --git a/src/mesa/drivers/dri/intel/intel_pixel_read.c b/src/mesa/drivers/dri/intel/intel_pixel_read.c index 2c57b470f5..538da28c50 100644 --- a/src/mesa/drivers/dri/intel/intel_pixel_read.c +++ b/src/mesa/drivers/dri/intel/intel_pixel_read.c @@ -291,6 +291,7 @@ intelReadPixels(GLcontext * ctx, intelFlush(ctx); +#ifdef I915 if (do_blit_readpixels (ctx, x, y, width, height, format, type, pack, pixels)) return; @@ -298,9 +299,19 @@ intelReadPixels(GLcontext * ctx, if (do_texture_readpixels (ctx, x, y, width, height, format, type, pack, pixels)) return; +#endif if (INTEL_DEBUG & DEBUG_PIXEL) _mesa_printf("%s: fallback to swrast\n", __FUNCTION__); + /* Update Mesa state before calling down into _swrast_ReadPixels, as + * the spans code requires the computed buffer states to be up to date, + * but _swrast_ReadPixels only updates Mesa state after setting up + * the spans code. + */ + + if (ctx->NewState) + _mesa_update_state(ctx); + _swrast_ReadPixels(ctx, x, y, width, height, format, type, pack, pixels); } -- cgit v1.2.3 From cf820a045f0626718ec147ebb26e31f82ec0b4fb Mon Sep 17 00:00:00 2001 From: Eric Anholt Date: Fri, 19 Jun 2009 22:12:52 -0700 Subject: intel: Also get the DRI2 front buffer when doing front buffer reading. (cherry picked from commit df70d3049a396af3601d2a1747770635a74120bb) --- src/mesa/drivers/dri/i965/Makefile | 1 + src/mesa/drivers/dri/intel/intel_buffers.c | 17 +++++++++++++++++ src/mesa/drivers/dri/intel/intel_context.c | 4 +++- src/mesa/drivers/dri/intel/intel_context.h | 8 ++++++++ 4 files changed, 29 insertions(+), 1 deletion(-) (limited to 'src') diff --git a/src/mesa/drivers/dri/i965/Makefile b/src/mesa/drivers/dri/i965/Makefile index 2934414d99..81c2fdb0dc 100644 --- a/src/mesa/drivers/dri/i965/Makefile +++ b/src/mesa/drivers/dri/i965/Makefile @@ -22,6 +22,7 @@ DRIVER_SOURCES = \ intel_pixel_bitmap.c \ intel_pixel_copy.c \ intel_pixel_draw.c \ + intel_pixel_read.c \ intel_state.c \ intel_swapbuffers.c \ intel_tex.c \ diff --git a/src/mesa/drivers/dri/intel/intel_buffers.c b/src/mesa/drivers/dri/intel/intel_buffers.c index d2fad9e4ea..44e34330a6 100644 --- a/src/mesa/drivers/dri/intel/intel_buffers.c +++ b/src/mesa/drivers/dri/intel/intel_buffers.c @@ -345,6 +345,23 @@ intelDrawBuffer(GLcontext * ctx, GLenum mode) static void intelReadBuffer(GLcontext * ctx, GLenum mode) { + if ((ctx->DrawBuffer != NULL) && (ctx->DrawBuffer->Name == 0)) { + struct intel_context *const intel = intel_context(ctx); + const GLboolean was_front_buffer_reading = + intel->is_front_buffer_reading; + + intel->is_front_buffer_reading = (mode == GL_FRONT_LEFT) + || (mode == GL_FRONT); + + /* If we weren't front-buffer reading before but we are now, make sure + * that the front-buffer has actually been allocated. + */ + if (!was_front_buffer_reading && intel->is_front_buffer_reading) { + intel_update_renderbuffers(intel->driContext, + intel->driContext->driDrawablePriv); + } + } + if (ctx->ReadBuffer == ctx->DrawBuffer) { /* This will update FBO completeness status. * A framebuffer will be incomplete if the GL_READ_BUFFER setting diff --git a/src/mesa/drivers/dri/intel/intel_context.c b/src/mesa/drivers/dri/intel/intel_context.c index 50ae677d20..977fad9313 100644 --- a/src/mesa/drivers/dri/intel/intel_context.c +++ b/src/mesa/drivers/dri/intel/intel_context.c @@ -220,7 +220,9 @@ intel_update_renderbuffers(__DRIcontext *context, __DRIdrawable *drawable) struct intel_renderbuffer *stencil_rb; i = 0; - if ((intel->is_front_buffer_rendering || !intel_fb->color_rb[1]) + if ((intel->is_front_buffer_rendering || + intel->is_front_buffer_reading || + !intel_fb->color_rb[1]) && intel_fb->color_rb[0]) { attachments[i++] = __DRI_BUFFER_FRONT_LEFT; attachments[i++] = intel_bits_per_pixel(intel_fb->color_rb[0]); diff --git a/src/mesa/drivers/dri/intel/intel_context.h b/src/mesa/drivers/dri/intel/intel_context.h index b3db561fd5..e2b3943fb5 100644 --- a/src/mesa/drivers/dri/intel/intel_context.h +++ b/src/mesa/drivers/dri/intel/intel_context.h @@ -295,6 +295,14 @@ struct intel_context * easily. */ GLboolean is_front_buffer_rendering; + /** + * Track whether front-buffer is the current read target. + * + * This is closely associated with is_front_buffer_rendering, but may + * be set separately. The DRI2 fake front buffer must be referenced + * either way. + */ + GLboolean is_front_buffer_reading; drm_clip_rect_t fboRect; /**< cliprect for FBO rendering */ -- cgit v1.2.3 From b2cba25f9eecf2063c3b98d66ade59cd9e50990e Mon Sep 17 00:00:00 2001 From: Eric Anholt Date: Mon, 22 Jun 2009 08:52:52 -0700 Subject: i965: Fix warnings in intel_pixel_read.c. (cherry picked from commit c80ce5ac90b1e0ac7a72cd41c314aa2000bfecf5) --- src/mesa/drivers/dri/intel/intel_pixel_read.c | 4 ++++ 1 file changed, 4 insertions(+) (limited to 'src') diff --git a/src/mesa/drivers/dri/intel/intel_pixel_read.c b/src/mesa/drivers/dri/intel/intel_pixel_read.c index 538da28c50..0370255614 100644 --- a/src/mesa/drivers/dri/intel/intel_pixel_read.c +++ b/src/mesa/drivers/dri/intel/intel_pixel_read.c @@ -31,6 +31,7 @@ #include "main/macros.h" #include "main/image.h" #include "main/bufferobj.h" +#include "main/state.h" #include "swrast/swrast.h" #include "intel_screen.h" @@ -299,6 +300,9 @@ intelReadPixels(GLcontext * ctx, if (do_texture_readpixels (ctx, x, y, width, height, format, type, pack, pixels)) return; +#else + (void)do_blit_readpixels; + (void)do_texture_readpixels; #endif if (INTEL_DEBUG & DEBUG_PIXEL) -- cgit v1.2.3 From 8de625c7cf639c583e8bf43acb1214010989bb64 Mon Sep 17 00:00:00 2001 From: Brian Paul Date: Tue, 8 Sep 2009 12:21:42 -0600 Subject: i965: fix incorrect test for vertex position attribute --- src/mesa/drivers/dri/i965/brw_context.h | 2 ++ src/mesa/drivers/dri/i965/brw_draw.c | 1 + src/mesa/drivers/dri/i965/brw_draw_upload.c | 2 +- 3 files changed, 4 insertions(+), 1 deletion(-) (limited to 'src') diff --git a/src/mesa/drivers/dri/i965/brw_context.h b/src/mesa/drivers/dri/i965/brw_context.h index e52fc3f374..5cf12fb353 100644 --- a/src/mesa/drivers/dri/i965/brw_context.h +++ b/src/mesa/drivers/dri/i965/brw_context.h @@ -386,6 +386,8 @@ struct brw_cached_batch_item { struct brw_vertex_element { const struct gl_client_array *glarray; + /** The corresponding Mesa vertex attribute */ + gl_vert_attrib attrib; /** Size of a complete element */ GLuint element_size; /** Number of uploaded elements for this input. */ diff --git a/src/mesa/drivers/dri/i965/brw_draw.c b/src/mesa/drivers/dri/i965/brw_draw.c index 5342622a73..54b0661db8 100644 --- a/src/mesa/drivers/dri/i965/brw_draw.c +++ b/src/mesa/drivers/dri/i965/brw_draw.c @@ -185,6 +185,7 @@ static void brw_merge_inputs( struct brw_context *brw, for (i = 0; i < VERT_ATTRIB_MAX; i++) { brw->vb.inputs[i].glarray = arrays[i]; + brw->vb.inputs[i].attrib = (gl_vert_attrib) i; if (arrays[i]->StrideB != 0) brw->vb.info.varying |= 1 << i; diff --git a/src/mesa/drivers/dri/i965/brw_draw_upload.c b/src/mesa/drivers/dri/i965/brw_draw_upload.c index 05079c043a..fd9c3915c4 100644 --- a/src/mesa/drivers/dri/i965/brw_draw_upload.c +++ b/src/mesa/drivers/dri/i965/brw_draw_upload.c @@ -422,7 +422,7 @@ static void brw_prepare_vertices(struct brw_context *brw) /* Queue the buffer object up to be uploaded in the next pass, * when we've decided if we're doing interleaved or not. */ - if (i == 0) { + if (input->attrib == VERT_ATTRIB_POS) { /* Position array not properly enabled: */ if (input->glarray->StrideB == 0) { -- cgit v1.2.3 From 42943a4cf90d66ade212feea29a29a3702eec6c1 Mon Sep 17 00:00:00 2001 From: Brian Paul Date: Tue, 8 Sep 2009 14:45:24 -0600 Subject: mesa: bump version to 7.5.2 I'm not 100% sure there'll be a 7.5.2 release, but just in case. --- Makefile | 2 +- configs/default | 2 +- src/mesa/main/version.h | 6 +++--- 3 files changed, 5 insertions(+), 5 deletions(-) (limited to 'src') diff --git a/Makefile b/Makefile index 6ad07e6137..636be5c28c 100644 --- a/Makefile +++ b/Makefile @@ -182,7 +182,7 @@ ultrix-gcc: # Rules for making release tarballs -VERSION=7.5.1 +VERSION=7.5.2 DIRECTORY = Mesa-$(VERSION) LIB_NAME = MesaLib-$(VERSION) DEMO_NAME = MesaDemos-$(VERSION) diff --git a/configs/default b/configs/default index 2981bb9a8f..773a6204d3 100644 --- a/configs/default +++ b/configs/default @@ -10,7 +10,7 @@ CONFIG_NAME = default # Version info MESA_MAJOR=7 MESA_MINOR=5 -MESA_TINY=1 +MESA_TINY=2 MESA_VERSION = $(MESA_MAJOR).$(MESA_MINOR).$(MESA_TINY) # external projects. This should be useless now that we use libdrm. diff --git a/src/mesa/main/version.h b/src/mesa/main/version.h index 50c1143ccc..995e46d318 100644 --- a/src/mesa/main/version.h +++ b/src/mesa/main/version.h @@ -1,6 +1,6 @@ /* * Mesa 3-D graphics library - * Version: 7.5.1 + * Version: 7.5.2 * * Copyright (C) 1999-2009 Brian Paul All Rights Reserved. * @@ -30,8 +30,8 @@ /* Mesa version */ #define MESA_MAJOR 7 #define MESA_MINOR 5 -#define MESA_PATCH 1 -#define MESA_VERSION_STRING "7.5.1" +#define MESA_PATCH 2 +#define MESA_VERSION_STRING "7.5.2-devel" /* To make version comparison easy */ #define MESA_VERSION(a,b,c) (((a) << 16) + ((b) << 8) + (c)) -- cgit v1.2.3 From 89a765e92b4847f80848c8be89efbce2d021434b Mon Sep 17 00:00:00 2001 From: Brian Paul Date: Wed, 9 Sep 2009 08:23:11 -0600 Subject: mesa: disable GL_LUMINANCE case in _mesa_meta_draw_pixels() Works around a bug found on i965. See bug 23670. --- src/mesa/drivers/common/meta.c | 5 ++++- 1 file changed, 4 insertions(+), 1 deletion(-) (limited to 'src') diff --git a/src/mesa/drivers/common/meta.c b/src/mesa/drivers/common/meta.c index 965fb8697e..28e49b6898 100644 --- a/src/mesa/drivers/common/meta.c +++ b/src/mesa/drivers/common/meta.c @@ -1446,7 +1446,10 @@ _mesa_meta_draw_pixels(GLcontext *ctx, if (_mesa_is_color_format(format)) { /* use more compact format when possible */ - if (format == GL_LUMINANCE || format == GL_LUMINANCE_ALPHA) + /* XXX disable special case for GL_LUMINANCE for now to work around + * apparent i965 driver bug (see bug #23670). + */ + if (/*format == GL_LUMINANCE ||*/ format == GL_LUMINANCE_ALPHA) texIntFormat = format; else texIntFormat = GL_RGBA; -- cgit v1.2.3 From c6c44bf48124dd5b4661014a8d58482c5a54557f Mon Sep 17 00:00:00 2001 From: aljen Date: Sat, 5 Sep 2009 23:06:53 +0200 Subject: gallium: Added HaikuOS platform --- src/gallium/auxiliary/util/u_debug.h | 2 ++ src/gallium/auxiliary/util/u_network.c | 6 +++--- src/gallium/auxiliary/util/u_network.h | 2 +- src/gallium/auxiliary/util/u_stream_stdc.c | 2 +- src/gallium/auxiliary/util/u_time.c | 12 ++++++------ src/gallium/auxiliary/util/u_time.h | 9 +++++++-- src/gallium/include/pipe/p_compiler.h | 4 +++- src/gallium/include/pipe/p_config.h | 3 +++ src/gallium/include/pipe/p_thread.h | 10 +++++----- 9 files changed, 31 insertions(+), 19 deletions(-) (limited to 'src') diff --git a/src/gallium/auxiliary/util/u_debug.h b/src/gallium/auxiliary/util/u_debug.h index d42b65ce28..1380d98d7e 100644 --- a/src/gallium/auxiliary/util/u_debug.h +++ b/src/gallium/auxiliary/util/u_debug.h @@ -88,6 +88,7 @@ _debug_printf(const char *format, ...) * - avoid outputing large strings (512 bytes is the current maximum length * that is guaranteed to be printed in all platforms) */ +#if !defined(PIPE_OS_HAIKU) static INLINE void debug_printf(const char *format, ...) { @@ -101,6 +102,7 @@ debug_printf(const char *format, ...) #endif } +#endif /* !PIPE_OS_HAIKU */ /* * ... isn't portable so we need to pass arguments in parentheses. diff --git a/src/gallium/auxiliary/util/u_network.c b/src/gallium/auxiliary/util/u_network.c index 07d804ecdb..bc4b758406 100644 --- a/src/gallium/auxiliary/util/u_network.c +++ b/src/gallium/auxiliary/util/u_network.c @@ -6,7 +6,7 @@ #if defined(PIPE_SUBSYSTEM_WINDOWS_USER) # include # include -#elif defined(PIPE_OS_LINUX) +#elif defined(PIPE_OS_LINUX) || defined(PIPE_OS_HAIKU) # include # include # include @@ -54,7 +54,7 @@ u_socket_close(int s) if (s < 0) return; -#if defined(PIPE_OS_LINUX) +#if defined(PIPE_OS_LINUX) || defined(PIPE_OS_HAIKU) shutdown(s, SHUT_RDWR); close(s); #elif defined(PIPE_SUBSYSTEM_WINDOWS_USER) @@ -169,7 +169,7 @@ u_socket_listen_on_port(uint16_t portnum) void u_socket_block(int s, boolean block) { -#if defined(PIPE_OS_LINUX) +#if defined(PIPE_OS_LINUX) || defined(PIPE_OS_HAIKU) int old = fcntl(s, F_GETFL, 0); if (old == -1) return; diff --git a/src/gallium/auxiliary/util/u_network.h b/src/gallium/auxiliary/util/u_network.h index 14d3884427..8c778f492c 100644 --- a/src/gallium/auxiliary/util/u_network.h +++ b/src/gallium/auxiliary/util/u_network.h @@ -6,7 +6,7 @@ #if defined(PIPE_SUBSYSTEM_WINDOWS_USER) # define PIPE_HAVE_SOCKETS -#elif defined(PIPE_OS_LINUX) +#elif defined(PIPE_OS_LINUX) || defined(PIPE_OS_HAIKU) # define PIPE_HAVE_SOCKETS #endif diff --git a/src/gallium/auxiliary/util/u_stream_stdc.c b/src/gallium/auxiliary/util/u_stream_stdc.c index d8f648e5dd..5cd05b2904 100644 --- a/src/gallium/auxiliary/util/u_stream_stdc.c +++ b/src/gallium/auxiliary/util/u_stream_stdc.c @@ -32,7 +32,7 @@ #include "pipe/p_config.h" -#if defined(PIPE_OS_LINUX) || defined(PIPE_OS_BSD) || defined(PIPE_SUBSYSTEM_WINDOWS_USER) || defined(PIPE_OS_SOLARIS) +#if defined(PIPE_OS_LINUX) || defined(PIPE_OS_BSD) || defined(PIPE_SUBSYSTEM_WINDOWS_USER) || defined(PIPE_OS_SOLARIS) || defined(PIPE_OS_HAIKU) #include diff --git a/src/gallium/auxiliary/util/u_time.c b/src/gallium/auxiliary/util/u_time.c index c16cdd0b22..b958a98635 100644 --- a/src/gallium/auxiliary/util/u_time.c +++ b/src/gallium/auxiliary/util/u_time.c @@ -35,7 +35,7 @@ #include "pipe/p_config.h" -#if defined(PIPE_OS_LINUX) || defined(PIPE_OS_BSD) || defined(PIPE_OS_SOLARIS) || defined(PIPE_OS_APPLE) +#if defined(PIPE_OS_LINUX) || defined(PIPE_OS_BSD) || defined(PIPE_OS_SOLARIS) || defined(PIPE_OS_APPLE) || defined(PIPE_OS_HAIKU) #include #elif defined(PIPE_SUBSYSTEM_WINDOWS_DISPLAY) #include @@ -77,7 +77,7 @@ util_time_get_frequency(void) void util_time_get(struct util_time *t) { -#if defined(PIPE_OS_LINUX) || defined(PIPE_OS_BSD) || defined(PIPE_OS_SOLARIS) || defined(PIPE_OS_APPLE) +#if defined(PIPE_OS_LINUX) || defined(PIPE_OS_BSD) || defined(PIPE_OS_SOLARIS) || defined(PIPE_OS_APPLE) || defined(PIPE_OS_HAIKU) gettimeofday(&t->tv, NULL); #elif defined(PIPE_SUBSYSTEM_WINDOWS_DISPLAY) LONGLONG temp; @@ -102,7 +102,7 @@ util_time_add(const struct util_time *t1, int64_t usecs, struct util_time *t2) { -#if defined(PIPE_OS_LINUX) || defined(PIPE_OS_BSD) || defined(PIPE_OS_SOLARIS) || defined(PIPE_OS_APPLE) +#if defined(PIPE_OS_LINUX) || defined(PIPE_OS_BSD) || defined(PIPE_OS_SOLARIS) || defined(PIPE_OS_APPLE) || defined(PIPE_OS_HAIKU) t2->tv.tv_sec = t1->tv.tv_sec + usecs / 1000000; t2->tv.tv_usec = t1->tv.tv_usec + usecs % 1000000; #elif defined(PIPE_SUBSYSTEM_WINDOWS_DISPLAY) || defined(PIPE_SUBSYSTEM_WINDOWS_USER) || defined(PIPE_SUBSYSTEM_WINDOWS_CE) @@ -124,7 +124,7 @@ int64_t util_time_diff(const struct util_time *t1, const struct util_time *t2) { -#if defined(PIPE_OS_LINUX) || defined(PIPE_OS_BSD) || defined(PIPE_OS_SOLARIS) || defined(PIPE_OS_APPLE) +#if defined(PIPE_OS_LINUX) || defined(PIPE_OS_BSD) || defined(PIPE_OS_SOLARIS) || defined(PIPE_OS_APPLE) || defined(PIPE_OS_HAIKU) return (t2->tv.tv_usec - t1->tv.tv_usec) + (t2->tv.tv_sec - t1->tv.tv_sec)*1000000; #elif defined(PIPE_SUBSYSTEM_WINDOWS_DISPLAY) || defined(PIPE_SUBSYSTEM_WINDOWS_USER) || defined(PIPE_SUBSYSTEM_WINDOWS_CE) @@ -144,7 +144,7 @@ util_time_micros( void ) util_time_get(&t1); -#if defined(PIPE_OS_LINUX) || defined(PIPE_OS_BSD) || defined(PIPE_OS_SOLARIS) || defined(PIPE_OS_APPLE) +#if defined(PIPE_OS_LINUX) || defined(PIPE_OS_BSD) || defined(PIPE_OS_SOLARIS) || defined(PIPE_OS_APPLE) || defined(PIPE_OS_HAIKU) return t1.tv.tv_usec + t1.tv.tv_sec*1000000LL; #elif defined(PIPE_SUBSYSTEM_WINDOWS_DISPLAY) || defined(PIPE_SUBSYSTEM_WINDOWS_USER) || defined(PIPE_SUBSYSTEM_WINDOWS_CE) util_time_get_frequency(); @@ -166,7 +166,7 @@ static INLINE int util_time_compare(const struct util_time *t1, const struct util_time *t2) { -#if defined(PIPE_OS_LINUX) || defined(PIPE_OS_BSD) || defined(PIPE_OS_SOLARIS) || defined(PIPE_OS_APPLE) +#if defined(PIPE_OS_LINUX) || defined(PIPE_OS_BSD) || defined(PIPE_OS_SOLARIS) || defined(PIPE_OS_APPLE) || defined(PIPE_OS_HAIKU) if (t1->tv.tv_sec < t2->tv.tv_sec) return -1; else if(t1->tv.tv_sec > t2->tv.tv_sec) diff --git a/src/gallium/auxiliary/util/u_time.h b/src/gallium/auxiliary/util/u_time.h index 7a5c54d9b2..0ce1cc4a34 100644 --- a/src/gallium/auxiliary/util/u_time.h +++ b/src/gallium/auxiliary/util/u_time.h @@ -43,6 +43,11 @@ #include /* usleep */ #endif +#if defined(PIPE_OS_HAIKU) +#include /* timeval */ +#include +#endif + #include "pipe/p_compiler.h" @@ -58,7 +63,7 @@ extern "C" { */ struct util_time { -#if defined(PIPE_OS_LINUX) || defined(PIPE_OS_BSD) || defined(PIPE_OS_SOLARIS) || defined(PIPE_OS_APPLE) +#if defined(PIPE_OS_LINUX) || defined(PIPE_OS_BSD) || defined(PIPE_OS_SOLARIS) || defined(PIPE_OS_APPLE) || defined(PIPE_OS_HAIKU) struct timeval tv; #else int64_t counter; @@ -89,7 +94,7 @@ util_time_timeout(const struct util_time *start, const struct util_time *end, const struct util_time *curr); -#if defined(PIPE_OS_LINUX) || defined(PIPE_OS_BSD) || defined(PIPE_OS_SOLARIS) || defined(PIPE_OS_APPLE) +#if defined(PIPE_OS_LINUX) || defined(PIPE_OS_BSD) || defined(PIPE_OS_SOLARIS) || defined(PIPE_OS_HAIKU) #define util_time_sleep usleep #else void diff --git a/src/gallium/include/pipe/p_compiler.h b/src/gallium/include/pipe/p_compiler.h index e6a67f8c2f..c13cffceb0 100644 --- a/src/gallium/include/pipe/p_compiler.h +++ b/src/gallium/include/pipe/p_compiler.h @@ -93,9 +93,11 @@ typedef int _Bool; #endif +#ifndef __HAIKU__ typedef unsigned int uint; -typedef unsigned char ubyte; typedef unsigned short ushort; +#endif +typedef unsigned char ubyte; #if 0 #define boolean bool diff --git a/src/gallium/include/pipe/p_config.h b/src/gallium/include/pipe/p_config.h index 4152d6ac36..de99957d9d 100644 --- a/src/gallium/include/pipe/p_config.h +++ b/src/gallium/include/pipe/p_config.h @@ -140,6 +140,9 @@ #define PIPE_OS_WINDOWS #endif +#if defined(__HAIKU__) +#define PIPE_OS_HAIKU +#endif /* * Subsystem. diff --git a/src/gallium/include/pipe/p_thread.h b/src/gallium/include/pipe/p_thread.h index 96e8e08744..b1606dc652 100644 --- a/src/gallium/include/pipe/p_thread.h +++ b/src/gallium/include/pipe/p_thread.h @@ -39,7 +39,7 @@ #include "util/u_debug.h" /* for assert */ -#if defined(PIPE_OS_LINUX) || defined(PIPE_OS_BSD) || defined(PIPE_OS_SOLARIS) +#if defined(PIPE_OS_LINUX) || defined(PIPE_OS_BSD) || defined(PIPE_OS_SOLARIS) || defined(PIPE_OS_HAIKU) #include /* POSIX threads headers */ #include /* for perror() */ @@ -213,7 +213,7 @@ typedef unsigned pipe_condvar; */ typedef struct { -#if defined(PIPE_OS_LINUX) || defined(PIPE_OS_BSD) || defined(PIPE_OS_SOLARIS) +#if defined(PIPE_OS_LINUX) || defined(PIPE_OS_BSD) || defined(PIPE_OS_SOLARIS) || defined(PIPE_OS_HAIKU) pthread_key_t key; #elif defined(PIPE_SUBSYSTEM_WINDOWS_USER) DWORD key; @@ -228,7 +228,7 @@ typedef struct { static INLINE void pipe_tsd_init(pipe_tsd *tsd) { -#if defined(PIPE_OS_LINUX) || defined(PIPE_OS_BSD) || defined(PIPE_OS_SOLARIS) +#if defined(PIPE_OS_LINUX) || defined(PIPE_OS_BSD) || defined(PIPE_OS_SOLARIS) || defined(PIPE_OS_HAIKU) if (pthread_key_create(&tsd->key, NULL/*free*/) != 0) { perror("pthread_key_create(): failed to allocate key for thread specific data"); exit(-1); @@ -245,7 +245,7 @@ pipe_tsd_get(pipe_tsd *tsd) if (tsd->initMagic != (int) PIPE_TSD_INIT_MAGIC) { pipe_tsd_init(tsd); } -#if defined(PIPE_OS_LINUX) || defined(PIPE_OS_BSD) || defined(PIPE_OS_SOLARIS) +#if defined(PIPE_OS_LINUX) || defined(PIPE_OS_BSD) || defined(PIPE_OS_SOLARIS) || defined(PIPE_OS_HAIKU) return pthread_getspecific(tsd->key); #elif defined(PIPE_SUBSYSTEM_WINDOWS_USER) assert(0); @@ -262,7 +262,7 @@ pipe_tsd_set(pipe_tsd *tsd, void *value) if (tsd->initMagic != (int) PIPE_TSD_INIT_MAGIC) { pipe_tsd_init(tsd); } -#if defined(PIPE_OS_LINUX) || defined(PIPE_OS_BSD) || defined(PIPE_OS_SOLARIS) +#if defined(PIPE_OS_LINUX) || defined(PIPE_OS_BSD) || defined(PIPE_OS_SOLARIS) || defined(PIPE_OS_HAIKU) if (pthread_setspecific(tsd->key, value) != 0) { perror("pthread_set_specific() failed"); exit(-1); -- cgit v1.2.3