diff options
Diffstat (limited to 'src/mesa/drivers')
43 files changed, 792 insertions, 550 deletions
diff --git a/src/mesa/drivers/common/meta.c b/src/mesa/drivers/common/meta.c index 6e0905aa493..05909cfa307 100644 --- a/src/mesa/drivers/common/meta.c +++ b/src/mesa/drivers/common/meta.c @@ -63,6 +63,10 @@ #include "drivers/common/meta.h" +/** Return offset in bytes of the field within a vertex struct */ +#define OFFSET(FIELD) ((void *) offsetof(struct vertex, FIELD)) + + /** * State which we may save/restore across meta ops. * XXX this may be incomplete... @@ -988,7 +992,10 @@ _mesa_meta_blit_framebuffer(GLcontext *ctx, const GLint srcH = abs(srcY1 - srcY0); const GLboolean srcFlipX = srcX1 < srcX0; const GLboolean srcFlipY = srcY1 < srcY0; - GLfloat verts[4][4]; /* four verts of X,Y,S,T */ + struct vertex { + GLfloat x, y, s, t; + }; + struct vertex verts[4]; GLboolean newTex; if (srcW > maxTexSize || srcH > maxTexSize) { @@ -1027,10 +1034,8 @@ _mesa_meta_blit_framebuffer(GLcontext *ctx, NULL, GL_DYNAMIC_DRAW_ARB); /* setup vertex arrays */ - _mesa_VertexPointer(2, GL_FLOAT, sizeof(verts[0]), - (void *) (0 * sizeof(GLfloat))); - _mesa_TexCoordPointer(2, GL_FLOAT, sizeof(verts[0]), - (void *) (2 * sizeof(GLfloat))); + _mesa_VertexPointer(2, GL_FLOAT, sizeof(struct vertex), OFFSET(x)); + _mesa_TexCoordPointer(2, GL_FLOAT, sizeof(struct vertex), OFFSET(s)); _mesa_EnableClientState(GL_VERTEX_ARRAY); _mesa_EnableClientState(GL_TEXTURE_COORD_ARRAY); } @@ -1043,23 +1048,23 @@ _mesa_meta_blit_framebuffer(GLcontext *ctx, /* vertex positions/texcoords (after texture allocation!) */ { - verts[0][0] = (GLfloat) dstX0; - verts[0][1] = (GLfloat) dstY0; - verts[1][0] = (GLfloat) dstX1; - verts[1][1] = (GLfloat) dstY0; - verts[2][0] = (GLfloat) dstX1; - verts[2][1] = (GLfloat) dstY1; - verts[3][0] = (GLfloat) dstX0; - verts[3][1] = (GLfloat) dstY1; - - verts[0][2] = 0.0F; - verts[0][3] = 0.0F; - verts[1][2] = tex->Sright; - verts[1][3] = 0.0F; - verts[2][2] = tex->Sright; - verts[2][3] = tex->Ttop; - verts[3][2] = 0.0F; - verts[3][3] = tex->Ttop; + verts[0].x = (GLfloat) dstX0; + verts[0].y = (GLfloat) dstY0; + verts[1].x = (GLfloat) dstX1; + verts[1].y = (GLfloat) dstY0; + verts[2].x = (GLfloat) dstX1; + verts[2].y = (GLfloat) dstY1; + verts[3].x = (GLfloat) dstX0; + verts[3].y = (GLfloat) dstY1; + + verts[0].s = 0.0F; + verts[0].t = 0.0F; + verts[1].s = tex->Sright; + verts[1].t = 0.0F; + verts[2].s = tex->Sright; + verts[2].t = tex->Ttop; + verts[3].s = 0.0F; + verts[3].t = tex->Ttop; /* upload new vertex data */ _mesa_BufferSubDataARB(GL_ARRAY_BUFFER_ARB, 0, sizeof(verts), verts); @@ -1125,7 +1130,10 @@ void _mesa_meta_clear(GLcontext *ctx, GLbitfield buffers) { struct clear_state *clear = &ctx->Meta->Clear; - GLfloat verts[4][7]; /* four verts of X,Y,Z,R,G,B,A */ + struct vertex { + GLfloat x, y, z, r, g, b, a; + }; + struct vertex verts[4]; /* save all state but scissor, pixel pack/unpack */ GLbitfield metaSave = META_ALL - META_SCISSOR - META_PIXEL_STORE; @@ -1150,10 +1158,8 @@ _mesa_meta_clear(GLcontext *ctx, GLbitfield buffers) NULL, GL_DYNAMIC_DRAW_ARB); /* setup vertex arrays */ - _mesa_VertexPointer(3, GL_FLOAT, sizeof(verts[0]), - (void *) (0 * sizeof(GLfloat))); - _mesa_ColorPointer(4, GL_FLOAT, sizeof(verts[0]), - (void *) (3 * sizeof(GLfloat))); + _mesa_VertexPointer(3, GL_FLOAT, sizeof(struct vertex), OFFSET(x)); + _mesa_ColorPointer(4, GL_FLOAT, sizeof(struct vertex), OFFSET(r)); _mesa_EnableClientState(GL_VERTEX_ARRAY); _mesa_EnableClientState(GL_COLOR_ARRAY); } @@ -1203,22 +1209,25 @@ _mesa_meta_clear(GLcontext *ctx, GLbitfield buffers) const GLfloat z = 1.0 - 2.0 * ctx->Depth.Clear; GLuint i; - verts[0][0] = x0; - verts[0][1] = y0; - verts[0][2] = z; - verts[1][0] = x1; - verts[1][1] = y0; - verts[1][2] = z; - verts[2][0] = x1; - verts[2][1] = y1; - verts[2][2] = z; - verts[3][0] = x0; - verts[3][1] = y1; - verts[3][2] = z; + verts[0].x = x0; + verts[0].y = y0; + verts[0].z = z; + verts[1].x = x1; + verts[1].y = y0; + verts[1].z = z; + verts[2].x = x1; + verts[2].y = y1; + verts[2].z = z; + verts[3].x = x0; + verts[3].y = y1; + verts[3].z = z; /* vertex colors */ for (i = 0; i < 4; i++) { - COPY_4FV(&verts[i][3], ctx->Color.ClearColor); + verts[i].r = ctx->Color.ClearColor[0]; + verts[i].g = ctx->Color.ClearColor[1]; + verts[i].b = ctx->Color.ClearColor[2]; + verts[i].a = ctx->Color.ClearColor[3]; } /* upload new vertex data */ @@ -1243,7 +1252,10 @@ _mesa_meta_copy_pixels(GLcontext *ctx, GLint srcX, GLint srcY, { struct copypix_state *copypix = &ctx->Meta->CopyPix; struct temp_texture *tex = get_temp_texture(ctx); - GLfloat verts[4][5]; /* four verts of X,Y,Z,S,T */ + struct vertex { + GLfloat x, y, z, s, t; + }; + struct vertex verts[4]; GLboolean newTex; GLenum intFormat = GL_RGBA; @@ -1281,10 +1293,8 @@ _mesa_meta_copy_pixels(GLcontext *ctx, GLint srcX, GLint srcY, NULL, GL_DYNAMIC_DRAW_ARB); /* setup vertex arrays */ - _mesa_VertexPointer(3, GL_FLOAT, sizeof(verts[0]), - (void *) (0 * sizeof(GLfloat))); - _mesa_TexCoordPointer(2, GL_FLOAT, sizeof(verts[0]), - (void *) (3 * sizeof(GLfloat))); + _mesa_VertexPointer(3, GL_FLOAT, sizeof(struct vertex), OFFSET(x)); + _mesa_TexCoordPointer(2, GL_FLOAT, sizeof(struct vertex), OFFSET(s)); _mesa_EnableClientState(GL_VERTEX_ARRAY); _mesa_EnableClientState(GL_TEXTURE_COORD_ARRAY); } @@ -1303,26 +1313,26 @@ _mesa_meta_copy_pixels(GLcontext *ctx, GLint srcX, GLint srcY, const GLfloat dstY1 = dstY + height * ctx->Pixel.ZoomY; const GLfloat z = ctx->Current.RasterPos[2]; - verts[0][0] = dstX0; - verts[0][1] = dstY0; - verts[0][2] = z; - verts[0][3] = 0.0F; - verts[0][4] = 0.0F; - verts[1][0] = dstX1; - verts[1][1] = dstY0; - verts[1][2] = z; - verts[1][3] = tex->Sright; - verts[1][4] = 0.0F; - verts[2][0] = dstX1; - verts[2][1] = dstY1; - verts[2][2] = z; - verts[2][3] = tex->Sright; - verts[2][4] = tex->Ttop; - verts[3][0] = dstX0; - verts[3][1] = dstY1; - verts[3][2] = z; - verts[3][3] = 0.0F; - verts[3][4] = tex->Ttop; + verts[0].x = dstX0; + verts[0].y = dstY0; + verts[0].z = z; + verts[0].s = 0.0F; + verts[0].t = 0.0F; + verts[1].x = dstX1; + verts[1].y = dstY0; + verts[1].z = z; + verts[1].s = tex->Sright; + verts[1].t = 0.0F; + verts[2].x = dstX1; + verts[2].y = dstY1; + verts[2].z = z; + verts[2].s = tex->Sright; + verts[2].t = tex->Ttop; + verts[3].x = dstX0; + verts[3].y = dstY1; + verts[3].z = z; + verts[3].s = 0.0F; + verts[3].t = tex->Ttop; /* upload new vertex data */ _mesa_BufferSubDataARB(GL_ARRAY_BUFFER_ARB, 0, sizeof(verts), verts); @@ -1496,7 +1506,10 @@ _mesa_meta_draw_pixels(GLcontext *ctx, struct temp_texture *tex = get_temp_texture(ctx); const struct gl_pixelstore_attrib unpackSave = ctx->Unpack; const GLuint origStencilMask = ctx->Stencil.WriteMask[0]; - GLfloat verts[4][5]; /* four verts of X,Y,Z,S,T */ + struct vertex { + GLfloat x, y, z, s, t; + }; + struct vertex verts[4]; GLenum texIntFormat; GLboolean fallback, newTex; GLbitfield metaExtraSave = 0x0; @@ -1512,13 +1525,18 @@ _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; } else if (_mesa_is_stencil_format(format)) { if (ctx->Extensions.ARB_fragment_program && + ctx->Pixel.IndexShift == 0 && + ctx->Pixel.IndexOffset == 0 && type == GL_UNSIGNED_BYTE) { /* We'll store stencil as alpha. This only works for GLubyte * image data because of how incoming values are mapped to alpha @@ -1588,10 +1606,8 @@ _mesa_meta_draw_pixels(GLcontext *ctx, NULL, GL_DYNAMIC_DRAW_ARB); /* setup vertex arrays */ - _mesa_VertexPointer(3, GL_FLOAT, sizeof(verts[0]), - (void *) (0 * sizeof(GLfloat))); - _mesa_TexCoordPointer(2, GL_FLOAT, sizeof(verts[0]), - (void *) (3 * sizeof(GLfloat))); + _mesa_VertexPointer(3, GL_FLOAT, sizeof(struct vertex), OFFSET(x)); + _mesa_TexCoordPointer(2, GL_FLOAT, sizeof(struct vertex), OFFSET(s)); _mesa_EnableClientState(GL_VERTEX_ARRAY); _mesa_EnableClientState(GL_TEXTURE_COORD_ARRAY); } @@ -1610,26 +1626,26 @@ _mesa_meta_draw_pixels(GLcontext *ctx, const GLfloat y1 = y + height * ctx->Pixel.ZoomY; const GLfloat z = ctx->Current.RasterPos[2]; - verts[0][0] = x0; - verts[0][1] = y0; - verts[0][2] = z; - verts[0][3] = 0.0F; - verts[0][4] = 0.0F; - verts[1][0] = x1; - verts[1][1] = y0; - verts[1][2] = z; - verts[1][3] = tex->Sright; - verts[1][4] = 0.0F; - verts[2][0] = x1; - verts[2][1] = y1; - verts[2][2] = z; - verts[2][3] = tex->Sright; - verts[2][4] = tex->Ttop; - verts[3][0] = x0; - verts[3][1] = y1; - verts[3][2] = z; - verts[3][3] = 0.0F; - verts[3][4] = tex->Ttop; + verts[0].x = x0; + verts[0].y = y0; + verts[0].z = z; + verts[0].s = 0.0F; + verts[0].t = 0.0F; + verts[1].x = x1; + verts[1].y = y0; + verts[1].z = z; + verts[1].s = tex->Sright; + verts[1].t = 0.0F; + verts[2].x = x1; + verts[2].y = y1; + verts[2].z = z; + verts[2].s = tex->Sright; + verts[2].t = tex->Ttop; + verts[3].x = x0; + verts[3].y = y1; + verts[3].z = z; + verts[3].s = 0.0F; + verts[3].t = tex->Ttop; /* upload new vertex data */ _mesa_BufferSubDataARB(GL_ARRAY_BUFFER_ARB, 0, sizeof(verts), verts); @@ -1653,6 +1669,13 @@ _mesa_meta_draw_pixels(GLcontext *ctx, _mesa_ColorMask(GL_FALSE, GL_FALSE, GL_FALSE, GL_FALSE); _mesa_set_enable(ctx, GL_STENCIL_TEST, GL_TRUE); + + /* set all stencil bits to 0 */ + _mesa_StencilOp(GL_REPLACE, GL_REPLACE, GL_REPLACE); + _mesa_StencilFunc(GL_ALWAYS, 0, 255); + _mesa_DrawArrays(GL_TRIANGLE_FAN, 0, 4); + + /* set stencil bits to 1 where needed */ _mesa_StencilOp(GL_KEEP, GL_KEEP, GL_REPLACE); _mesa_BindProgram(GL_FRAGMENT_PROGRAM_ARB, drawpix->StencilFP); @@ -1721,7 +1744,10 @@ _mesa_meta_bitmap(GLcontext *ctx, struct temp_texture *tex = get_bitmap_temp_texture(ctx); const GLenum texIntFormat = GL_ALPHA; const struct gl_pixelstore_attrib unpackSave = *unpack; - GLfloat verts[4][9]; /* four verts of X,Y,Z,S,T,R,G,B,A */ + struct vertex { + GLfloat x, y, z, s, t, r, g, b, a; + }; + struct vertex verts[4]; GLboolean newTex; GLubyte *bitmap8; @@ -1764,13 +1790,9 @@ _mesa_meta_bitmap(GLcontext *ctx, NULL, GL_DYNAMIC_DRAW_ARB); /* setup vertex arrays */ - _mesa_VertexPointer(3, GL_FLOAT, sizeof(verts[0]), - (void *) (0 * sizeof(GLfloat))); - _mesa_TexCoordPointer(2, GL_FLOAT, sizeof(verts[0]), - (void *) (3 * sizeof(GLfloat))); - _mesa_ColorPointer(4, GL_FLOAT, sizeof(verts[0]), - (void *) (5 * sizeof(GLfloat))); - + _mesa_VertexPointer(3, GL_FLOAT, sizeof(struct vertex), OFFSET(x)); + _mesa_TexCoordPointer(2, GL_FLOAT, sizeof(struct vertex), OFFSET(s)); + _mesa_ColorPointer(4, GL_FLOAT, sizeof(struct vertex), OFFSET(r)); _mesa_EnableClientState(GL_VERTEX_ARRAY); _mesa_EnableClientState(GL_TEXTURE_COORD_ARRAY); _mesa_EnableClientState(GL_COLOR_ARRAY); @@ -1791,32 +1813,32 @@ _mesa_meta_bitmap(GLcontext *ctx, const GLfloat z = ctx->Current.RasterPos[2]; GLuint i; - verts[0][0] = x0; - verts[0][1] = y0; - verts[0][2] = z; - verts[0][3] = 0.0F; - verts[0][4] = 0.0F; - verts[1][0] = x1; - verts[1][1] = y0; - verts[1][2] = z; - verts[1][3] = tex->Sright; - verts[1][4] = 0.0F; - verts[2][0] = x1; - verts[2][1] = y1; - verts[2][2] = z; - verts[2][3] = tex->Sright; - verts[2][4] = tex->Ttop; - verts[3][0] = x0; - verts[3][1] = y1; - verts[3][2] = z; - verts[3][3] = 0.0F; - verts[3][4] = tex->Ttop; + verts[0].x = x0; + verts[0].y = y0; + verts[0].z = z; + verts[0].s = 0.0F; + verts[0].t = 0.0F; + verts[1].x = x1; + verts[1].y = y0; + verts[1].z = z; + verts[1].s = tex->Sright; + verts[1].t = 0.0F; + verts[2].x = x1; + verts[2].y = y1; + verts[2].z = z; + verts[2].s = tex->Sright; + verts[2].t = tex->Ttop; + verts[3].x = x0; + verts[3].y = y1; + verts[3].z = z; + verts[3].s = 0.0F; + verts[3].t = tex->Ttop; for (i = 0; i < 4; i++) { - verts[i][5] = ctx->Current.RasterColor[0]; - verts[i][6] = ctx->Current.RasterColor[1]; - verts[i][7] = ctx->Current.RasterColor[2]; - verts[i][8] = ctx->Current.RasterColor[3]; + verts[i].r = ctx->Current.RasterColor[0]; + verts[i].g = ctx->Current.RasterColor[1]; + verts[i].b = ctx->Current.RasterColor[2]; + verts[i].a = ctx->Current.RasterColor[3]; } /* upload new vertex data */ @@ -1858,7 +1880,10 @@ _mesa_meta_generate_mipmap(GLcontext *ctx, GLenum target, struct gl_texture_object *texObj) { struct gen_mipmap_state *mipmap = &ctx->Meta->Mipmap; - struct { GLfloat x, y, s, t, r; } verts[4]; + struct vertex { + GLfloat x, y, s, t, r; + }; + struct vertex verts[4]; const GLuint baseLevel = texObj->BaseLevel; const GLuint maxLevel = texObj->MaxLevel; const GLenum minFilterSave = texObj->MinFilter; @@ -1899,11 +1924,8 @@ _mesa_meta_generate_mipmap(GLcontext *ctx, GLenum target, NULL, GL_DYNAMIC_DRAW_ARB); /* setup vertex arrays */ - _mesa_VertexPointer(2, GL_FLOAT, sizeof(verts[0]), - (void *) (0 * sizeof(GLfloat))); - _mesa_TexCoordPointer(3, GL_FLOAT, sizeof(verts[0]), - (void *) (2 * sizeof(GLfloat))); - + _mesa_VertexPointer(2, GL_FLOAT, sizeof(struct vertex), OFFSET(x)); + _mesa_TexCoordPointer(3, GL_FLOAT, sizeof(struct vertex), OFFSET(s)); _mesa_EnableClientState(GL_VERTEX_ARRAY); _mesa_EnableClientState(GL_TEXTURE_COORD_ARRAY); } @@ -1933,13 +1955,13 @@ _mesa_meta_generate_mipmap(GLcontext *ctx, GLenum target, verts[0].r = 0.0F; verts[1].s = 1.0F; verts[1].t = 0.0F; + verts[1].r = 0.0F; + verts[2].s = 1.0F; + verts[2].t = 1.0F; verts[2].r = 0.0F; - verts[3].s = 1.0F; + verts[3].s = 0.0F; verts[3].t = 1.0F; verts[3].r = 0.0F; - verts[4].s = 0.0F; - verts[4].t = 1.0F; - verts[4].r = 0.0F; break; } diff --git a/src/mesa/drivers/dri/common/extension_helper.h b/src/mesa/drivers/dri/common/extension_helper.h index 40a030ce0da..2c89a9f58a9 100644 --- a/src/mesa/drivers/dri/common/extension_helper.h +++ b/src/mesa/drivers/dri/common/extension_helper.h @@ -316,6 +316,13 @@ static const char CombinerOutputNV_names[] = ""; #endif +#if defined(need_GL_NV_vertex_program) +static const char VertexAttribs3fvNV_names[] = + "iip\0" /* Parameter signature */ + "glVertexAttribs3fvNV\0" + ""; +#endif + #if defined(need_GL_VERSION_2_0) || defined(need_GL_ARB_shader_objects) static const char Uniform2fARB_names[] = "iff\0" /* Parameter signature */ @@ -575,6 +582,13 @@ static const char MatrixIndexusvARB_names[] = ""; #endif +#if defined(need_GL_ARB_draw_elements_base_vertex) +static const char DrawElementsBaseVertex_names[] = + "iiipi\0" /* Parameter signature */ + "glDrawElementsBaseVertex\0" + ""; +#endif + #if defined(need_GL_VERSION_2_0) || defined(need_GL_ARB_vertex_program) static const char DisableVertexAttribArrayARB_names[] = "i\0" /* Parameter signature */ @@ -970,10 +984,10 @@ static const char GlobalAlphaFactordSUN_names[] = ""; #endif -#if defined(need_GL_NV_vertex_program) -static const char VertexAttribs3fvNV_names[] = +#if defined(need_GL_NV_register_combiners) +static const char GetFinalCombinerInputParameterfvNV_names[] = "iip\0" /* Parameter signature */ - "glVertexAttribs3fvNV\0" + "glGetFinalCombinerInputParameterfvNV\0" ""; #endif @@ -3918,6 +3932,13 @@ static const char MapBufferARB_names[] = ""; #endif +#if defined(need_GL_ARB_draw_elements_base_vertex) +static const char MultiDrawElementsBaseVertex_names[] = + "ipipip\0" /* Parameter signature */ + "glMultiDrawElementsBaseVertex\0" + ""; +#endif + #if defined(need_GL_EXT_coordinate_frame) static const char Binormal3svEXT_names[] = "p\0" /* Parameter signature */ @@ -4554,10 +4575,10 @@ static const char UniformMatrix3x4fv_names[] = ""; #endif -#if defined(need_GL_EXT_coordinate_frame) -static const char Binormal3fvEXT_names[] = - "p\0" /* Parameter signature */ - "glBinormal3fvEXT\0" +#if defined(need_GL_ARB_draw_elements_base_vertex) +static const char DrawRangeElementsBaseVertex_names[] = + "iiiiipi\0" /* Parameter signature */ + "glDrawRangeElementsBaseVertex\0" ""; #endif @@ -4710,10 +4731,10 @@ static const char GetFragmentLightfvSGIX_names[] = ""; #endif -#if defined(need_GL_NV_register_combiners) -static const char GetFinalCombinerInputParameterfvNV_names[] = - "iip\0" /* Parameter signature */ - "glGetFinalCombinerInputParameterfvNV\0" +#if defined(need_GL_EXT_coordinate_frame) +static const char Binormal3fvEXT_names[] = + "p\0" /* Parameter signature */ + "glBinormal3fvEXT\0" ""; #endif @@ -5090,6 +5111,15 @@ static const struct dri_extension_function GL_ARB_draw_buffers_functions[] = { }; #endif +#if defined(need_GL_ARB_draw_elements_base_vertex) +static const struct dri_extension_function GL_ARB_draw_elements_base_vertex_functions[] = { + { DrawElementsBaseVertex_names, DrawElementsBaseVertex_remap_index, -1 }, + { MultiDrawElementsBaseVertex_names, MultiDrawElementsBaseVertex_remap_index, -1 }, + { DrawRangeElementsBaseVertex_names, DrawRangeElementsBaseVertex_remap_index, -1 }, + { NULL, 0, 0 } +}; +#endif + #if defined(need_GL_ARB_framebuffer_object) static const struct dri_extension_function GL_ARB_framebuffer_object_functions[] = { { BlitFramebufferEXT_names, BlitFramebufferEXT_remap_index, -1 }, @@ -5520,8 +5550,8 @@ static const struct dri_extension_function GL_EXT_coordinate_frame_functions[] = { Binormal3dvEXT_names, Binormal3dvEXT_remap_index, -1 }, { Tangent3iEXT_names, Tangent3iEXT_remap_index, -1 }, { Tangent3bvEXT_names, Tangent3bvEXT_remap_index, -1 }, - { Binormal3fvEXT_names, Binormal3fvEXT_remap_index, -1 }, { Tangent3bEXT_names, Tangent3bEXT_remap_index, -1 }, + { Binormal3fvEXT_names, Binormal3fvEXT_remap_index, -1 }, { BinormalPointerEXT_names, BinormalPointerEXT_remap_index, -1 }, { Tangent3svEXT_names, Tangent3svEXT_remap_index, -1 }, { Binormal3bEXT_names, Binormal3bEXT_remap_index, -1 }, @@ -5963,6 +5993,7 @@ static const struct dri_extension_function GL_NV_point_sprite_functions[] = { static const struct dri_extension_function GL_NV_register_combiners_functions[] = { { CombinerOutputNV_names, CombinerOutputNV_remap_index, -1 }, { CombinerParameterfvNV_names, CombinerParameterfvNV_remap_index, -1 }, + { GetFinalCombinerInputParameterfvNV_names, GetFinalCombinerInputParameterfvNV_remap_index, -1 }, { GetCombinerOutputParameterfvNV_names, GetCombinerOutputParameterfvNV_remap_index, -1 }, { FinalCombinerInputNV_names, FinalCombinerInputNV_remap_index, -1 }, { GetCombinerInputParameterfvNV_names, GetCombinerInputParameterfvNV_remap_index, -1 }, @@ -5971,7 +6002,6 @@ static const struct dri_extension_function GL_NV_register_combiners_functions[] { GetFinalCombinerInputParameterivNV_names, GetFinalCombinerInputParameterivNV_remap_index, -1 }, { CombinerInputNV_names, CombinerInputNV_remap_index, -1 }, { CombinerParameterfNV_names, CombinerParameterfNV_remap_index, -1 }, - { GetFinalCombinerInputParameterfvNV_names, GetFinalCombinerInputParameterfvNV_remap_index, -1 }, { GetCombinerInputParameterivNV_names, GetCombinerInputParameterivNV_remap_index, -1 }, { CombinerParameterivNV_names, CombinerParameterivNV_remap_index, -1 }, { NULL, 0, 0 } @@ -5998,6 +6028,7 @@ static const struct dri_extension_function GL_NV_vertex_array_range_functions[] static const struct dri_extension_function GL_NV_vertex_program_functions[] = { { VertexAttrib4ubvNV_names, VertexAttrib4ubvNV_remap_index, -1 }, { VertexAttrib4svNV_names, VertexAttrib4svNV_remap_index, -1 }, + { VertexAttribs3fvNV_names, VertexAttribs3fvNV_remap_index, -1 }, { VertexAttribs1dvNV_names, VertexAttribs1dvNV_remap_index, -1 }, { VertexAttrib1fvNV_names, VertexAttrib1fvNV_remap_index, -1 }, { VertexAttrib4fNV_names, VertexAttrib4fNV_remap_index, -1 }, @@ -6006,7 +6037,6 @@ static const struct dri_extension_function GL_NV_vertex_program_functions[] = { { VertexAttribs3dvNV_names, VertexAttribs3dvNV_remap_index, -1 }, { VertexAttribs4fvNV_names, VertexAttribs4fvNV_remap_index, -1 }, { VertexAttrib2sNV_names, VertexAttrib2sNV_remap_index, -1 }, - { VertexAttribs3fvNV_names, VertexAttribs3fvNV_remap_index, -1 }, { ProgramEnvParameter4fvARB_names, ProgramEnvParameter4fvARB_remap_index, -1 }, { LoadProgramNV_names, LoadProgramNV_remap_index, -1 }, { VertexAttrib4fvNV_names, VertexAttrib4fvNV_remap_index, -1 }, diff --git a/src/mesa/drivers/dri/i965/brw_cc.c b/src/mesa/drivers/dri/i965/brw_cc.c index c724218cf56..1088a7a6070 100644 --- a/src/mesa/drivers/dri/i965/brw_cc.c +++ b/src/mesa/drivers/dri/i965/brw_cc.c @@ -39,12 +39,14 @@ static void prepare_cc_vp( struct brw_context *brw ) { + GLcontext *ctx = &brw->intel.ctx; struct brw_cc_viewport ccv; memset(&ccv, 0, sizeof(ccv)); - ccv.min_depth = 0.0; - ccv.max_depth = 1.0; + /* _NEW_VIEWPORT */ + ccv.min_depth = ctx->Viewport.Near; + ccv.max_depth = ctx->Viewport.Far; dri_bo_unreference(brw->cc.vp_bo); brw->cc.vp_bo = brw_cache_data( &brw->cache, BRW_CC_VP, &ccv, NULL, 0 ); @@ -52,7 +54,7 @@ static void prepare_cc_vp( struct brw_context *brw ) const struct brw_tracked_state brw_cc_vp = { .dirty = { - .mesa = 0, + .mesa = _NEW_VIEWPORT, .brw = BRW_NEW_CONTEXT, .cache = 0 }, diff --git a/src/mesa/drivers/dri/i965/brw_clip_state.c b/src/mesa/drivers/dri/i965/brw_clip_state.c index 5762c9577c6..234b3744bfc 100644 --- a/src/mesa/drivers/dri/i965/brw_clip_state.c +++ b/src/mesa/drivers/dri/i965/brw_clip_state.c @@ -43,11 +43,14 @@ struct brw_clip_unit_key { unsigned int curbe_offset; unsigned int nr_urb_entries, urb_size; + + GLboolean depth_clamp; }; static void clip_unit_populate_key(struct brw_context *brw, struct brw_clip_unit_key *key) { + GLcontext *ctx = &brw->intel.ctx; memset(key, 0, sizeof(*key)); /* CACHE_NEW_CLIP_PROG */ @@ -62,6 +65,9 @@ clip_unit_populate_key(struct brw_context *brw, struct brw_clip_unit_key *key) /* BRW_NEW_URB_FENCE */ key->nr_urb_entries = brw->urb.nr_clip_entries; key->urb_size = brw->urb.vsize; + + /* _NEW_TRANSOFORM */ + key->depth_clamp = ctx->Transform.DepthClamp; } static dri_bo * @@ -117,7 +123,8 @@ clip_unit_create_from_key(struct brw_context *brw, clip.clip5.userclip_enable_flags = 0x7f; clip.clip5.userclip_must_clip = 1; clip.clip5.guard_band_enable = 0; - clip.clip5.viewport_z_clip_enable = 1; + if (!key->depth_clamp) + clip.clip5.viewport_z_clip_enable = 1; clip.clip5.viewport_xy_clip_enable = 1; clip.clip5.vertex_position_space = BRW_CLIP_NDCSPACE; clip.clip5.api_mode = BRW_CLIP_API_OGL; @@ -168,7 +175,7 @@ static void upload_clip_unit( struct brw_context *brw ) const struct brw_tracked_state brw_clip_unit = { .dirty = { - .mesa = 0, + .mesa = _NEW_TRANSFORM, .brw = (BRW_NEW_CURBE_OFFSETS | BRW_NEW_URB_FENCE), .cache = CACHE_NEW_CLIP_PROG diff --git a/src/mesa/drivers/dri/i965/brw_context.h b/src/mesa/drivers/dri/i965/brw_context.h index 26a64d65280..a5209ac41be 100644 --- a/src/mesa/drivers/dri/i965/brw_context.h +++ b/src/mesa/drivers/dri/i965/brw_context.h @@ -390,6 +390,8 @@ 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 682094ff139..44bb7bd5882 100644 --- a/src/mesa/drivers/dri/i965/brw_draw.c +++ b/src/mesa/drivers/dri/i965/brw_draw.c @@ -25,13 +25,15 @@ * **************************************************************************/ -#include <stdlib.h> #include "main/glheader.h" #include "main/context.h" #include "main/state.h" -#include "main/api_validate.h" #include "main/enums.h" +#include "tnl/tnl.h" +#include "vbo/vbo_context.h" +#include "swrast/swrast.h" +#include "swrast_setup/swrast_setup.h" #include "brw_draw.h" #include "brw_defines.h" @@ -42,11 +44,6 @@ #include "intel_batchbuffer.h" #include "intel_buffer_objects.h" -#include "tnl/tnl.h" -#include "vbo/vbo_context.h" -#include "swrast/swrast.h" -#include "swrast_setup/swrast_setup.h" - #define FILE_DEBUG_FLAG DEBUG_BATCH static GLuint prim_to_hw_prim[GL_POLYGON+1] = { @@ -145,7 +142,7 @@ static void brw_emit_prim(struct brw_context *brw, prim_packet.start_vert_location += brw->ib.start_vertex_offset; prim_packet.instance_count = 1; prim_packet.start_instance_location = 0; - prim_packet.base_vert_location = 0; + prim_packet.base_vert_location = prim->basevertex; /* Can't wrap here, since we rely on the validated state. */ brw->no_batch_wrap = GL_TRUE; @@ -187,6 +184,7 @@ 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.sizes[i/16] |= (brw->vb.inputs[i].glarray->Size - 1) << diff --git a/src/mesa/drivers/dri/i965/brw_draw_upload.c b/src/mesa/drivers/dri/i965/brw_draw_upload.c index d49fb0fd951..765ae5a2fe5 100644 --- a/src/mesa/drivers/dri/i965/brw_draw_upload.c +++ b/src/mesa/drivers/dri/i965/brw_draw_upload.c @@ -25,9 +25,9 @@ * **************************************************************************/ -#include <stdlib.h> #include "main/glheader.h" +#include "main/bufferobj.h" #include "main/context.h" #include "main/state.h" #include "main/api_validate.h" @@ -384,7 +384,7 @@ static void brw_prepare_vertices(struct brw_context *brw) input->element_size = get_size(input->glarray->Type) * input->glarray->Size; - if (input->glarray->BufferObj->Name != 0) { + if (_mesa_is_bufferobj(input->glarray->BufferObj)) { struct intel_buffer_object *intel_buffer = intel_buffer_object(input->glarray->BufferObj); @@ -423,7 +423,7 @@ 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) { @@ -623,7 +623,7 @@ static void brw_prepare_indices(struct brw_context *brw) /* Turn into a proper VBO: */ - if (!bufferobj->Name) { + if (!_mesa_is_bufferobj(bufferobj)) { brw->ib.start_vertex_offset = 0; /* Get new bufferobj, offset: diff --git a/src/mesa/drivers/dri/i965/brw_vs_emit.c b/src/mesa/drivers/dri/i965/brw_vs_emit.c index 584fdbdfc37..1638ef81115 100644 --- a/src/mesa/drivers/dri/i965/brw_vs_emit.c +++ b/src/mesa/drivers/dri/i965/brw_vs_emit.c @@ -1270,9 +1270,27 @@ post_vs_emit( struct brw_vs_compile *c, } static uint32_t -get_predicate(uint32_t swizzle) +get_predicate(const struct prog_instruction *inst) { - switch (swizzle) { + if (inst->DstReg.CondMask == COND_TR) + return BRW_PREDICATE_NONE; + + /* All of GLSL only produces predicates for COND_NE and one channel per + * vector. Fail badly if someone starts doing something else, as it might + * mean infinite looping or something. + * + * We'd like to support all the condition codes, but our hardware doesn't + * quite match the Mesa IR, which is modeled after the NV extensions. For + * those, the instruction may update the condition codes or not, then any + * later instruction may use one of those condition codes. For gen4, the + * instruction may update the flags register based on one of the condition + * codes output by the instruction, and then further instructions may + * predicate on that. We can probably support this, but it won't + * necessarily be easy. + */ + assert(inst->DstReg.CondMask == COND_NE); + + switch (inst->DstReg.CondSwizzle) { case SWIZZLE_XXXX: return BRW_PREDICATE_ALIGN16_REPLICATE_X; case SWIZZLE_YYYY: @@ -1282,7 +1300,8 @@ get_predicate(uint32_t swizzle) case SWIZZLE_WWWW: return BRW_PREDICATE_ALIGN16_REPLICATE_W; default: - _mesa_problem(NULL, "Unexpected predicate: 0x%08x\n", swizzle); + _mesa_problem(NULL, "Unexpected predicate: 0x%08x\n", + inst->DstReg.CondMask); return BRW_PREDICATE_NORMAL; } } @@ -1294,6 +1313,7 @@ void brw_vs_emit(struct brw_vs_compile *c ) #define MAX_IF_DEPTH 32 #define MAX_LOOP_DEPTH 32 struct brw_compile *p = &c->func; + struct brw_context *brw = p->brw; const GLuint nr_insns = c->vp->program.Base.NumInstructions; GLuint insn, if_depth = 0, loop_depth = 0; GLuint end_offset = 0; @@ -1492,8 +1512,8 @@ void brw_vs_emit(struct brw_vs_compile *c ) case OPCODE_IF: assert(if_depth < MAX_IF_DEPTH); if_inst[if_depth] = brw_IF(p, BRW_EXECUTE_8); - if_inst[if_depth]->header.predicate_control = - get_predicate(inst->DstReg.CondSwizzle); + /* Note that brw_IF smashes the predicate_control field. */ + if_inst[if_depth]->header.predicate_control = get_predicate(inst); if_depth++; break; case OPCODE_ELSE: @@ -1503,45 +1523,48 @@ void brw_vs_emit(struct brw_vs_compile *c ) assert(if_depth > 0); brw_ENDIF(p, if_inst[--if_depth]); break; -#if 0 case OPCODE_BGNLOOP: loop_inst[loop_depth++] = brw_DO(p, BRW_EXECUTE_8); break; case OPCODE_BRK: + brw_set_predicate_control(p, get_predicate(inst)); brw_BREAK(p); - brw_set_predicate_control(p, BRW_PREDICATE_NONE); + brw_set_predicate_control(p, BRW_PREDICATE_NONE); break; case OPCODE_CONT: + brw_set_predicate_control(p, get_predicate(inst)); brw_CONT(p); brw_set_predicate_control(p, BRW_PREDICATE_NONE); break; case OPCODE_ENDLOOP: { struct brw_instruction *inst0, *inst1; + GLuint br = 1; + loop_depth--; + + if (BRW_IS_IGDNG(brw)) + br = 2; + inst0 = inst1 = brw_WHILE(p, loop_inst[loop_depth]); /* patch all the BREAK/CONT instructions from last BEGINLOOP */ while (inst0 > loop_inst[loop_depth]) { inst0--; if (inst0->header.opcode == BRW_OPCODE_BREAK) { - inst0->bits3.if_else.jump_count = inst1 - inst0 + 1; + inst0->bits3.if_else.jump_count = br * (inst1 - inst0 + 1); inst0->bits3.if_else.pop_count = 0; } else if (inst0->header.opcode == BRW_OPCODE_CONTINUE) { - inst0->bits3.if_else.jump_count = inst1 - inst0; + inst0->bits3.if_else.jump_count = br * (inst1 - inst0); inst0->bits3.if_else.pop_count = 0; } } } break; -#else - (void) loop_inst; - (void) loop_depth; -#endif case OPCODE_BRA: - brw_set_predicate_control(p, BRW_PREDICATE_NORMAL); + brw_set_predicate_control(p, get_predicate(inst)); brw_ADD(p, brw_ip_reg(), brw_ip_reg(), brw_imm_d(1*16)); - brw_set_predicate_control_flag_value(p, 0xff); + brw_set_predicate_control(p, BRW_PREDICATE_NONE); break; case OPCODE_CAL: brw_set_access_mode(p, BRW_ALIGN_1); diff --git a/src/mesa/drivers/dri/i965/brw_wm.h b/src/mesa/drivers/dri/i965/brw_wm.h index ae98b5492db..872b1f3ecf4 100644 --- a/src/mesa/drivers/dri/i965/brw_wm.h +++ b/src/mesa/drivers/dri/i965/brw_wm.h @@ -38,6 +38,8 @@ #include "brw_context.h" #include "brw_eu.h" +#define SATURATE (1<<5) + /* A big lookup table is used to figure out which and how many * additional regs will inserted before the main payload in the WM * program execution. These mainly relate to depth and stencil @@ -203,7 +205,6 @@ struct brw_wm_compile { GLuint fp_temp; GLuint fp_interp_emitted; GLuint fp_fragcolor_emitted; - GLuint fp_deriv_emitted; struct prog_src_register pixel_xy; struct prog_src_register delta_xy; @@ -299,5 +300,10 @@ void brw_wm_lookup_iz( GLuint line_aa, GLboolean brw_wm_is_glsl(const struct gl_fragment_program *fp); void brw_wm_glsl_emit(struct brw_context *brw, struct brw_wm_compile *c); +void emit_ddxy(struct brw_compile *p, + const struct brw_reg *dst, + GLuint mask, + GLboolean is_ddx, + const struct brw_reg *arg0); #endif diff --git a/src/mesa/drivers/dri/i965/brw_wm_emit.c b/src/mesa/drivers/dri/i965/brw_wm_emit.c index 981864323ec..bf80a2942a4 100644 --- a/src/mesa/drivers/dri/i965/brw_wm_emit.c +++ b/src/mesa/drivers/dri/i965/brw_wm_emit.c @@ -34,8 +34,6 @@ #include "brw_context.h" #include "brw_wm.h" -#define SATURATE (1<<5) - /* Not quite sure how correct this is - need to understand horiz * vs. vertical strides a little better. */ @@ -281,6 +279,79 @@ static void emit_frontfacing( struct brw_compile *p, brw_set_predicate_control_flag_value(p, 0xff); } +/* For OPCODE_DDX and OPCODE_DDY, per channel of output we've got input + * looking like: + * + * arg0: ss0.tl ss0.tr ss0.bl ss0.br ss1.tl ss1.tr ss1.bl ss1.br + * + * and we're trying to produce: + * + * DDX DDY + * dst: (ss0.tr - ss0.tl) (ss0.tl - ss0.bl) + * (ss0.tr - ss0.tl) (ss0.tr - ss0.br) + * (ss0.br - ss0.bl) (ss0.tl - ss0.bl) + * (ss0.br - ss0.bl) (ss0.tr - ss0.br) + * (ss1.tr - ss1.tl) (ss1.tl - ss1.bl) + * (ss1.tr - ss1.tl) (ss1.tr - ss1.br) + * (ss1.br - ss1.bl) (ss1.tl - ss1.bl) + * (ss1.br - ss1.bl) (ss1.tr - ss1.br) + * + * and add another set of two more subspans if in 16-pixel dispatch mode. + * + * For DDX, it ends up being easy: width = 2, horiz=0 gets us the same result + * for each pair, and vertstride = 2 jumps us 2 elements after processing a + * pair. But for DDY, it's harder, as we want to produce the pairs swizzled + * between each other. We could probably do it like ddx and swizzle the right + * order later, but bail for now and just produce + * ((ss0.tl - ss0.bl)x4 (ss1.tl - ss1.bl)x4) + */ +void emit_ddxy(struct brw_compile *p, + const struct brw_reg *dst, + GLuint mask, + GLboolean is_ddx, + const struct brw_reg *arg0) +{ + int i; + struct brw_reg src0, src1; + + if (mask & SATURATE) + brw_set_saturate(p, 1); + for (i = 0; i < 4; i++ ) { + if (mask & (1<<i)) { + if (is_ddx) { + src0 = brw_reg(arg0[i].file, arg0[i].nr, 1, + BRW_REGISTER_TYPE_F, + BRW_VERTICAL_STRIDE_2, + BRW_WIDTH_2, + BRW_HORIZONTAL_STRIDE_0, + BRW_SWIZZLE_XYZW, WRITEMASK_XYZW); + src1 = brw_reg(arg0[i].file, arg0[i].nr, 0, + BRW_REGISTER_TYPE_F, + BRW_VERTICAL_STRIDE_2, + BRW_WIDTH_2, + BRW_HORIZONTAL_STRIDE_0, + BRW_SWIZZLE_XYZW, WRITEMASK_XYZW); + } else { + src0 = brw_reg(arg0[i].file, arg0[i].nr, 0, + BRW_REGISTER_TYPE_F, + BRW_VERTICAL_STRIDE_4, + BRW_WIDTH_4, + BRW_HORIZONTAL_STRIDE_0, + BRW_SWIZZLE_XYZW, WRITEMASK_XYZW); + src1 = brw_reg(arg0[i].file, arg0[i].nr, 2, + BRW_REGISTER_TYPE_F, + BRW_VERTICAL_STRIDE_4, + BRW_WIDTH_4, + BRW_HORIZONTAL_STRIDE_0, + BRW_SWIZZLE_XYZW, WRITEMASK_XYZW); + } + brw_ADD(p, dst[i], src0, negate(src1)); + } + } + if (mask & SATURATE) + brw_set_saturate(p, 0); +} + static void emit_alu1( struct brw_compile *p, struct brw_instruction *(*func)(struct brw_compile *, struct brw_reg, @@ -1272,6 +1343,14 @@ void brw_wm_emit( struct brw_wm_compile *c ) emit_alu1(p, brw_RNDD, dst, dst_flags, args[0]); break; + case OPCODE_DDX: + emit_ddxy(p, dst, dst_flags, GL_TRUE, args[0]); + break; + + case OPCODE_DDY: + emit_ddxy(p, dst, dst_flags, GL_FALSE, args[0]); + break; + case OPCODE_DP3: emit_dp3(p, dst, dst_flags, args[0], args[1]); break; diff --git a/src/mesa/drivers/dri/i965/brw_wm_fp.c b/src/mesa/drivers/dri/i965/brw_wm_fp.c index 123fe841c3f..4e3edfbbffa 100644 --- a/src/mesa/drivers/dri/i965/brw_wm_fp.c +++ b/src/mesa/drivers/dri/i965/brw_wm_fp.c @@ -494,38 +494,6 @@ static void emit_interp( struct brw_wm_compile *c, c->fp_interp_emitted |= 1<<idx; } -static void emit_ddx( struct brw_wm_compile *c, - const struct prog_instruction *inst ) -{ - GLuint idx = inst->SrcReg[0].Index; - struct prog_src_register interp = src_reg(PROGRAM_PAYLOAD, idx); - - c->fp_deriv_emitted |= 1<<idx; - emit_op(c, - OPCODE_DDX, - inst->DstReg, - 0, - interp, - get_pixel_w(c), - src_undef()); -} - -static void emit_ddy( struct brw_wm_compile *c, - const struct prog_instruction *inst ) -{ - GLuint idx = inst->SrcReg[0].Index; - struct prog_src_register interp = src_reg(PROGRAM_PAYLOAD, idx); - - c->fp_deriv_emitted |= 1<<idx; - emit_op(c, - OPCODE_DDY, - inst->DstReg, - 0, - interp, - get_pixel_w(c), - src_undef()); -} - /*********************************************************************** * Hacks to extend the program parameter and constant lists. */ @@ -1186,12 +1154,6 @@ void brw_wm_pass_fp( struct brw_wm_compile *c ) */ out->DstReg.WriteMask = 0; break; - case OPCODE_DDX: - emit_ddx(c, inst); - break; - case OPCODE_DDY: - emit_ddy(c, inst); - break; case OPCODE_END: emit_fb_write(c); break; diff --git a/src/mesa/drivers/dri/i965/brw_wm_glsl.c b/src/mesa/drivers/dri/i965/brw_wm_glsl.c index 7c210abbce3..c9fe1dd8ad2 100644 --- a/src/mesa/drivers/dri/i965/brw_wm_glsl.c +++ b/src/mesa/drivers/dri/i965/brw_wm_glsl.c @@ -22,6 +22,7 @@ static struct brw_reg get_dst_reg(struct brw_wm_compile *c, GLboolean brw_wm_is_glsl(const struct gl_fragment_program *fp) { int i; + for (i = 0; i < fp->Base.NumInstructions; i++) { const struct prog_instruction *inst = &fp->Base.Instructions[i]; switch (inst->Opcode) { @@ -31,8 +32,6 @@ GLboolean brw_wm_is_glsl(const struct gl_fragment_program *fp) case OPCODE_CAL: case OPCODE_BRK: case OPCODE_RET: - case OPCODE_DDX: - case OPCODE_DDY: case OPCODE_NOISE1: case OPCODE_NOISE2: case OPCODE_NOISE3: @@ -293,7 +292,7 @@ static void prealloc_reg(struct brw_wm_compile *c) int i, j; struct brw_reg reg; int urb_read_length = 0; - GLuint inputs = FRAG_BIT_WPOS | c->fp_interp_emitted | c->fp_deriv_emitted; + GLuint inputs = FRAG_BIT_WPOS | c->fp_interp_emitted; GLuint reg_index = 0; memset(c->used_grf, GL_FALSE, sizeof(c->used_grf)); @@ -1474,61 +1473,6 @@ static void emit_sne(struct brw_wm_compile *c, emit_sop(c, inst, BRW_CONDITIONAL_NEQ); } -static void emit_ddx(struct brw_wm_compile *c, - const struct prog_instruction *inst) -{ - struct brw_compile *p = &c->func; - GLuint mask = inst->DstReg.WriteMask; - struct brw_reg interp[4]; - struct brw_reg dst; - struct brw_reg src0, w; - GLuint nr, i; - src0 = get_src_reg(c, inst, 0, 0); - w = get_src_reg(c, inst, 1, 3); - nr = src0.nr; - interp[0] = brw_vec1_grf(nr, 0); - interp[1] = brw_vec1_grf(nr, 4); - interp[2] = brw_vec1_grf(nr+1, 0); - interp[3] = brw_vec1_grf(nr+1, 4); - brw_set_saturate(p, inst->SaturateMode != SATURATE_OFF); - for(i = 0; i < 4; i++ ) { - if (mask & (1<<i)) { - dst = get_dst_reg(c, inst, i); - brw_MOV(p, dst, interp[i]); - brw_MUL(p, dst, dst, w); - } - } - brw_set_saturate(p, 0); -} - -static void emit_ddy(struct brw_wm_compile *c, - const struct prog_instruction *inst) -{ - struct brw_compile *p = &c->func; - GLuint mask = inst->DstReg.WriteMask; - struct brw_reg interp[4]; - struct brw_reg dst; - struct brw_reg src0, w; - GLuint nr, i; - - src0 = get_src_reg(c, inst, 0, 0); - nr = src0.nr; - w = get_src_reg(c, inst, 1, 3); - interp[0] = brw_vec1_grf(nr, 0); - interp[1] = brw_vec1_grf(nr, 4); - interp[2] = brw_vec1_grf(nr+1, 0); - interp[3] = brw_vec1_grf(nr+1, 4); - brw_set_saturate(p, inst->SaturateMode != SATURATE_OFF); - for(i = 0; i < 4; i++ ) { - if (mask & (1<<i)) { - dst = get_dst_reg(c, inst, i); - brw_MOV(p, dst, suboffset(interp[i], 1)); - brw_MUL(p, dst, dst, w); - } - } - brw_set_saturate(p, 0); -} - static INLINE struct brw_reg high_words( struct brw_reg reg ) { return stride( suboffset( retype( reg, BRW_REGISTER_TYPE_W ), 1 ), @@ -2780,6 +2724,21 @@ static void post_wm_emit( struct brw_wm_compile *c ) brw_resolve_cals(&c->func); } +static void +get_argument_regs(struct brw_wm_compile *c, + const struct prog_instruction *inst, + int index, + struct brw_reg *regs, + int mask) +{ + int i; + + for (i = 0; i < 4; i++) { + if (mask & (1 << i)) + regs[i] = get_src_reg(c, inst, index, i); + } +} + static void brw_wm_emit_glsl(struct brw_context *brw, struct brw_wm_compile *c) { #define MAX_IF_DEPTH 32 @@ -2797,6 +2756,9 @@ static void brw_wm_emit_glsl(struct brw_context *brw, struct brw_wm_compile *c) for (i = 0; i < c->nr_fp_insns; i++) { const struct prog_instruction *inst = &c->prog_instructions[i]; + int dst_flags; + struct brw_reg args[3][4], dst[4]; + int j; c->cur_inst = i; @@ -2814,6 +2776,10 @@ static void brw_wm_emit_glsl(struct brw_context *brw, struct brw_wm_compile *c) else brw_set_conditionalmod(p, BRW_CONDITIONAL_NONE); + dst_flags = inst->DstReg.WriteMask; + if (inst->SaturateMode == SATURATE_ZERO_ONE) + dst_flags |= SATURATE; + switch (inst->Opcode) { case WM_PIXELXY: emit_pixel_xy(c, inst); @@ -2899,10 +2865,16 @@ static void brw_wm_emit_glsl(struct brw_context *brw, struct brw_wm_compile *c) emit_min_max(c, inst); break; case OPCODE_DDX: - emit_ddx(c, inst); - break; case OPCODE_DDY: - emit_ddy(c, inst); + for (j = 0; j < 4; j++) { + if (inst->DstReg.WriteMask & (1 << j)) + dst[j] = get_dst_reg(c, inst, j); + else + dst[j] = brw_null_reg(); + } + get_argument_regs(c, inst, 0, args[0], WRITEMASK_XYZW); + emit_ddxy(p, dst, dst_flags, (inst->Opcode == OPCODE_DDX), + args[0]); break; case OPCODE_SLT: emit_slt(c, inst); diff --git a/src/mesa/drivers/dri/i965/brw_wm_pass1.c b/src/mesa/drivers/dri/i965/brw_wm_pass1.c index 9c68bfd78b2..b4493940292 100644 --- a/src/mesa/drivers/dri/i965/brw_wm_pass1.c +++ b/src/mesa/drivers/dri/i965/brw_wm_pass1.c @@ -178,6 +178,11 @@ void brw_wm_pass1( struct brw_wm_compile *c ) read1 = writemask; break; + case OPCODE_DDX: + case OPCODE_DDY: + read0 = writemask; + break; + case OPCODE_MAD: case OPCODE_CMP: case OPCODE_LRP: diff --git a/src/mesa/drivers/dri/i965/brw_wm_state.c b/src/mesa/drivers/dri/i965/brw_wm_state.c index 39f8c6d522b..361f91292be 100644 --- a/src/mesa/drivers/dri/i965/brw_wm_state.c +++ b/src/mesa/drivers/dri/i965/brw_wm_state.c @@ -107,6 +107,12 @@ wm_unit_populate_key(struct brw_context *brw, struct brw_wm_unit_key *key) /* as far as we can tell */ key->computes_depth = (fp->Base.OutputsWritten & (1 << FRAG_RESULT_DEPTH)) != 0; + /* BRW_NEW_DEPTH_BUFFER + * Override for NULL depthbuffer case, required by the Pixel Shader Computed + * Depth field. + */ + if (brw->state.depth_region == NULL) + key->computes_depth = 0; /* _NEW_COLOR */ key->uses_kill = fp->UsesKill || ctx->Color.AlphaEnabled; @@ -300,6 +306,7 @@ const struct brw_tracked_state brw_wm_unit = { .brw = (BRW_NEW_FRAGMENT_PROGRAM | BRW_NEW_CURBE_OFFSETS | + BRW_NEW_DEPTH_BUFFER | BRW_NEW_NR_WM_SURFACES), .cache = (CACHE_NEW_WM_PROG | diff --git a/src/mesa/drivers/dri/i965/brw_wm_surface_state.c b/src/mesa/drivers/dri/i965/brw_wm_surface_state.c index 096f74394eb..51539ac1e73 100644 --- a/src/mesa/drivers/dri/i965/brw_wm_surface_state.c +++ b/src/mesa/drivers/dri/i965/brw_wm_surface_state.c @@ -545,15 +545,20 @@ brw_update_renderbuffer_surface(struct brw_context *brw, irb->texformat->MesaFormat); } key.tiling = region->tiling; - key.width = region->width; - key.height = region->height; + if (brw->intel.intelScreen->driScrnPriv->dri2.enabled) { + key.width = rb->Width; + key.height = rb->Height; + } else { + key.width = region->width; + key.height = region->height; + } key.pitch = region->pitch; key.cpp = region->cpp; key.draw_offset = region->draw_offset; /* cur 3d or cube face offset */ } else { key.surface_type = BRW_SURFACE_NULL; key.surface_format = BRW_SURFACEFORMAT_B8G8R8A8_UNORM; - key.tiling = 0; + key.tiling = I915_TILING_X; key.width = 1; key.height = 1; key.cpp = 4; @@ -629,7 +634,7 @@ brw_update_renderbuffer_surface(struct brw_context *brw, drm_intel_bo_emit_reloc(brw->wm.surf_bo[unit], offsetof(struct brw_surface_state, ss1), region_bo, - surf.ss1.base_addr, + surf.ss1.base_addr - region_bo->offset, I915_GEM_DOMAIN_RENDER, I915_GEM_DOMAIN_RENDER); } diff --git a/src/mesa/drivers/dri/intel/intel_blit.c b/src/mesa/drivers/dri/intel/intel_blit.c index 0c5be4c7989..43141c509c7 100644 --- a/src/mesa/drivers/dri/intel/intel_blit.c +++ b/src/mesa/drivers/dri/intel/intel_blit.c @@ -26,9 +26,6 @@ **************************************************************************/ -#include <stdio.h> -#include <errno.h> - #include "main/mtypes.h" #include "main/context.h" #include "main/enums.h" diff --git a/src/mesa/drivers/dri/intel/intel_chipset.h b/src/mesa/drivers/dri/intel/intel_chipset.h index 3c38f1676c9..3dc8653a735 100644 --- a/src/mesa/drivers/dri/intel/intel_chipset.h +++ b/src/mesa/drivers/dri/intel/intel_chipset.h @@ -66,6 +66,7 @@ #define PCI_CHIP_Q45_G 0x2E12 #define PCI_CHIP_G45_G 0x2E22 #define PCI_CHIP_G41_G 0x2E32 +#define PCI_CHIP_B43_G 0x2E42 #define PCI_CHIP_ILD_G 0x0042 #define PCI_CHIP_ILM_G 0x0046 @@ -83,7 +84,8 @@ #define IS_G45(devid) (devid == PCI_CHIP_IGD_E_G || \ devid == PCI_CHIP_Q45_G || \ devid == PCI_CHIP_G45_G || \ - devid == PCI_CHIP_G41_G) + devid == PCI_CHIP_G41_G || \ + devid == PCI_CHIP_B43_G) #define IS_GM45(devid) (devid == PCI_CHIP_GM45_GM) #define IS_G4X(devid) (IS_G45(devid) || IS_GM45(devid)) diff --git a/src/mesa/drivers/dri/intel/intel_context.c b/src/mesa/drivers/dri/intel/intel_context.c index 89f99f7ffdb..7e21b94acc4 100644 --- a/src/mesa/drivers/dri/intel/intel_context.c +++ b/src/mesa/drivers/dri/intel/intel_context.c @@ -162,6 +162,9 @@ intelGetString(GLcontext * ctx, GLenum name) case PCI_CHIP_G41_G: chipset = "Intel(R) G41"; break; + case PCI_CHIP_B43_G: + chipset = "Intel(R) B43"; + break; case PCI_CHIP_ILD_G: chipset = "Intel(R) IGDNG_D"; break; @@ -641,10 +644,6 @@ intelInitContext(struct intel_context *intel, intel->maxBatchSize = BATCH_SZ; intel->bufmgr = intelScreen->bufmgr; - - if (0) /* for debug */ - drm_intel_bufmgr_set_debug(intel->bufmgr, 1); - intel->ttm = intelScreen->ttm; if (intel->ttm) { int bo_reuse_mode; diff --git a/src/mesa/drivers/dri/intel/intel_extensions.c b/src/mesa/drivers/dri/intel/intel_extensions.c index 2e61c556d85..5431cf90a1b 100644 --- a/src/mesa/drivers/dri/intel/intel_extensions.c +++ b/src/mesa/drivers/dri/intel/intel_extensions.c @@ -31,6 +31,7 @@ #define need_GL_ARB_copy_buffer +#define need_GL_ARB_draw_elements_base_vertex #define need_GL_ARB_framebuffer_object #define need_GL_ARB_map_buffer_range #define need_GL_ARB_occlusion_query @@ -73,6 +74,7 @@ */ static const struct dri_extension card_extensions[] = { { "GL_ARB_copy_buffer", GL_ARB_copy_buffer_functions }, + { "GL_ARB_draw_elements_base_vertex", GL_ARB_draw_elements_base_vertex_functions }, { "GL_ARB_half_float_pixel", NULL }, { "GL_ARB_map_buffer_range", GL_ARB_map_buffer_range_functions }, { "GL_ARB_multitexture", NULL }, @@ -139,6 +141,7 @@ static const struct dri_extension i915_extensions[] = { /** i965-only extensions */ static const struct dri_extension brw_extensions[] = { + { "GL_ARB_depth_clamp", NULL }, { "GL_ARB_depth_texture", NULL }, { "GL_ARB_fragment_program", NULL }, { "GL_ARB_fragment_program_shadow", NULL }, diff --git a/src/mesa/drivers/dri/intel/intel_pixel_bitmap.c b/src/mesa/drivers/dri/intel/intel_pixel_bitmap.c index b543a0bbc3e..9a0bcc07a5b 100644 --- a/src/mesa/drivers/dri/intel/intel_pixel_bitmap.c +++ b/src/mesa/drivers/dri/intel/intel_pixel_bitmap.c @@ -209,7 +209,7 @@ do_blit_bitmap( GLcontext *ctx, if (!dst) return GL_FALSE; - if (unpack->BufferObj->Name) { + if (_mesa_is_bufferobj(unpack->BufferObj)) { bitmap = map_pbo(ctx, width, height, unpack, bitmap); if (bitmap == NULL) return GL_TRUE; /* even though this is an error, we're done */ @@ -329,7 +329,7 @@ out: if (INTEL_DEBUG & DEBUG_SYNC) intel_batchbuffer_flush(intel->batch); - if (unpack->BufferObj->Name) { + if (_mesa_is_bufferobj(unpack->BufferObj)) { /* done with PBO so unmap it now */ ctx->Driver.UnmapBuffer(ctx, GL_PIXEL_UNPACK_BUFFER_EXT, unpack->BufferObj); @@ -418,7 +418,7 @@ intel_texture_bitmap(GLcontext * ctx, return GL_FALSE; } - if (unpack->BufferObj->Name) { + if (_mesa_is_bufferobj(unpack->BufferObj)) { bitmap = map_pbo(ctx, width, height, unpack, bitmap); if (bitmap == NULL) return GL_TRUE; /* even though this is an error, we're done */ @@ -428,7 +428,7 @@ intel_texture_bitmap(GLcontext * ctx, a8_bitmap = _mesa_calloc(width * height); _mesa_expand_bitmap(width, height, unpack, bitmap, a8_bitmap, width, 0xff); - if (unpack->BufferObj->Name) { + if (_mesa_is_bufferobj(unpack->BufferObj)) { /* done with PBO so unmap it now */ ctx->Driver.UnmapBuffer(ctx, GL_PIXEL_UNPACK_BUFFER_EXT, unpack->BufferObj); diff --git a/src/mesa/drivers/dri/intel/intel_pixel_draw.c b/src/mesa/drivers/dri/intel/intel_pixel_draw.c index 8c113881d6b..7fbb89fd6aa 100644 --- a/src/mesa/drivers/dri/intel/intel_pixel_draw.c +++ b/src/mesa/drivers/dri/intel/intel_pixel_draw.c @@ -42,6 +42,7 @@ #include "main/depth.h" #include "main/hash.h" #include "main/blend.h" +#include "swrast/swrast.h" #include "drivers/common/meta.h" #include "intel_context.h" @@ -260,9 +261,24 @@ intelDrawPixels(GLcontext * ctx, const struct gl_pixelstore_attrib *unpack, const GLvoid * pixels) { +#if 0 + /* XXX this function doesn't seem to work reliably even when all + * the pre-requisite conditions are met. + * Note that this function is never hit with conform. + * Fall back to swrast because even the _mesa_meta_draw_pixels() approach + * isn't working because of an apparent stencil bug. + */ if (intel_stencil_drawpixels(ctx, x, y, width, height, format, type, unpack, pixels)) return; +#else + (void) intel_stencil_drawpixels; /* silence warning */ + if (format == GL_STENCIL_INDEX) { + _swrast_DrawPixels(ctx, x, y, width, height, format, type, + unpack, pixels); + return; + } +#endif _mesa_meta_draw_pixels(ctx, x, y, width, height, format, type, unpack, pixels); diff --git a/src/mesa/drivers/dri/intel/intel_pixel_read.c b/src/mesa/drivers/dri/intel/intel_pixel_read.c index 8713463ace2..bc67f6242a1 100644 --- a/src/mesa/drivers/dri/intel/intel_pixel_read.c +++ b/src/mesa/drivers/dri/intel/intel_pixel_read.c @@ -180,16 +180,7 @@ do_blit_readpixels(GLcontext * ctx, if (!src) return GL_FALSE; - if (pack->BufferObj->Name) { - /* 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 { + if (!_mesa_is_bufferobj(pack->BufferObj)) { /* PBO only for now: */ if (INTEL_DEBUG & DEBUG_PIXEL) diff --git a/src/mesa/drivers/dri/intel/intel_regions.c b/src/mesa/drivers/dri/intel/intel_regions.c index 068a3f33797..a86c66a8443 100644 --- a/src/mesa/drivers/dri/intel/intel_regions.c +++ b/src/mesa/drivers/dri/intel/intel_regions.c @@ -196,6 +196,13 @@ intel_region_alloc(struct intel_context *intel, else height = ALIGN(height, 2); + /* 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); diff --git a/src/mesa/drivers/dri/intel/intel_span.c b/src/mesa/drivers/dri/intel/intel_span.c index 8df49908806..1e739434577 100644 --- a/src/mesa/drivers/dri/intel/intel_span.c +++ b/src/mesa/drivers/dri/intel/intel_span.c @@ -444,23 +444,32 @@ intel_renderbuffer_unmap(struct intel_context *intel, * _ColorReadBuffer, _DepthBuffer or _StencilBuffer fields. */ static void -intel_map_unmap_buffers(struct intel_context *intel, GLboolean map) +intel_map_unmap_framebuffer(struct intel_context *intel, + struct gl_framebuffer *fb, + GLboolean map) { GLcontext *ctx = &intel->ctx; GLuint i, j; - /* color draw buffers */ - for (j = 0; j < ctx->DrawBuffer->_NumColorDrawBuffers; j++) { + /* color buffers */ + if (fb == ctx->DrawBuffer) { + for (j = 0; j < fb->_NumColorDrawBuffers; j++) { + if (map) + intel_renderbuffer_map(intel, fb->_ColorDrawBuffers[j]); + else + intel_renderbuffer_unmap(intel, fb->_ColorDrawBuffers[j]); + } + } else { if (map) - intel_renderbuffer_map(intel, ctx->DrawBuffer->_ColorDrawBuffers[j]); + intel_renderbuffer_map(intel, fb->_ColorReadBuffer); else - intel_renderbuffer_unmap(intel, ctx->DrawBuffer->_ColorDrawBuffers[j]); + intel_renderbuffer_unmap(intel, fb->_ColorReadBuffer); } /* check for render to textures */ for (i = 0; i < BUFFER_COUNT; i++) { struct gl_renderbuffer_attachment *att = - ctx->DrawBuffer->Attachment + i; + fb->Attachment + i; struct gl_texture_object *tex = att->Texture; if (tex) { /* render to texture */ @@ -472,34 +481,26 @@ intel_map_unmap_buffers(struct intel_context *intel, GLboolean map) } } - /* color read buffers */ - if (map) - intel_renderbuffer_map(intel, ctx->ReadBuffer->_ColorReadBuffer); - else - intel_renderbuffer_unmap(intel, ctx->ReadBuffer->_ColorReadBuffer); - /* depth buffer (Note wrapper!) */ - if (ctx->DrawBuffer->_DepthBuffer) { + if (fb->_DepthBuffer) { if (map) - intel_renderbuffer_map(intel, ctx->DrawBuffer->_DepthBuffer->Wrapped); + intel_renderbuffer_map(intel, fb->_DepthBuffer->Wrapped); else intel_renderbuffer_unmap(intel, - ctx->DrawBuffer->_DepthBuffer->Wrapped); + fb->_DepthBuffer->Wrapped); } /* stencil buffer (Note wrapper!) */ - if (ctx->DrawBuffer->_StencilBuffer) { + if (fb->_StencilBuffer) { if (map) intel_renderbuffer_map(intel, - ctx->DrawBuffer->_StencilBuffer->Wrapped); + fb->_StencilBuffer->Wrapped); else intel_renderbuffer_unmap(intel, - ctx->DrawBuffer->_StencilBuffer->Wrapped); + fb->_StencilBuffer->Wrapped); } } - - /** * Prepare for software rendering. Map current read/draw framebuffers' * renderbuffes and all currently bound texture objects. @@ -522,7 +523,8 @@ intelSpanRenderStart(GLcontext * ctx) } } - intel_map_unmap_buffers(intel, GL_TRUE); + intel_map_unmap_framebuffer(intel, ctx->DrawBuffer, GL_TRUE); + intel_map_unmap_framebuffer(intel, ctx->ReadBuffer, GL_TRUE); } /** @@ -544,7 +546,8 @@ intelSpanRenderFinish(GLcontext * ctx) } } - intel_map_unmap_buffers(intel, GL_FALSE); + intel_map_unmap_framebuffer(intel, ctx->DrawBuffer, GL_FALSE); + intel_map_unmap_framebuffer(intel, ctx->ReadBuffer, GL_FALSE); UNLOCK_HARDWARE(intel); } @@ -711,6 +714,9 @@ intel_set_span_functions(struct intel_context *intel, intel_YTile_InitStencilPointers_z24_s8(rb); break; } + } else { + _mesa_problem(NULL, + "Unexpected ActualFormat in intelSetSpanFunctions"); } break; default: diff --git a/src/mesa/drivers/dri/intel/intel_tex_image.c b/src/mesa/drivers/dri/intel/intel_tex_image.c index c5f52208376..a206fe6805b 100644 --- a/src/mesa/drivers/dri/intel/intel_tex_image.c +++ b/src/mesa/drivers/dri/intel/intel_tex_image.c @@ -1,15 +1,11 @@ -#include <stdlib.h> -#include <stdio.h> - #include "main/glheader.h" #include "main/macros.h" #include "main/mtypes.h" #include "main/enums.h" -#include "main/colortab.h" +#include "main/bufferobj.h" #include "main/convolve.h" #include "main/context.h" -#include "main/simple_list.h" #include "main/texcompress.h" #include "main/texformat.h" #include "main/texgetimage.h" @@ -206,7 +202,7 @@ try_pbo_upload(struct intel_context *intel, GLuint src_offset, src_stride; GLuint dst_offset, dst_stride; - if (unpack->BufferObj->Name == 0 || + if (!_mesa_is_bufferobj(unpack->BufferObj) || intel->ctx._ImageTransferState || unpack->SkipPixels || unpack->SkipRows) { DBG("%s: failure 1\n", __FUNCTION__); @@ -264,7 +260,7 @@ try_pbo_zcopy(struct intel_context *intel, GLuint src_offset, src_stride; GLuint dst_offset, dst_stride; - if (unpack->BufferObj->Name == 0 || + if (!_mesa_is_bufferobj(unpack->BufferObj) || intel->ctx._ImageTransferState || unpack->SkipPixels || unpack->SkipRows) { DBG("%s: failure 1\n", __FUNCTION__); @@ -427,7 +423,7 @@ intelTexImage(GLcontext * ctx, */ if (dims <= 2 && intelImage->mt && - unpack->BufferObj->Name != 0 && + _mesa_is_bufferobj(unpack->BufferObj) && check_pbo_format(internalFormat, format, type, intelImage->base.TexFormat)) { diff --git a/src/mesa/drivers/dri/r300/r300_cmdbuf.c b/src/mesa/drivers/dri/r300/r300_cmdbuf.c index 0fe32a5443a..da5b7ba642f 100644 --- a/src/mesa/drivers/dri/r300/r300_cmdbuf.c +++ b/src/mesa/drivers/dri/r300/r300_cmdbuf.c @@ -684,11 +684,7 @@ void r300InitCmdBuf(r300ContextPtr r300) r300->hw.rb3d_dither_ctl.cmd[0] = cmdpacket0(r300->radeon.radeonScreen, R300_RB3D_DITHER_CTL, 9); ALLOC_STATE(rb3d_aaresolve_ctl, always, 2, 0); r300->hw.rb3d_aaresolve_ctl.cmd[0] = cmdpacket0(r300->radeon.radeonScreen, R300_RB3D_AARESOLVE_CTL, 1); - if ((r300->radeon.radeonScreen->chip_family >= CHIP_FAMILY_RV515) || - ( !r300->radeon.radeonScreen->kernel_mm && ( - (r300->radeon.radeonScreen->chip_family == CHIP_FAMILY_RS400) || - (r300->radeon.radeonScreen->chip_family == CHIP_FAMILY_RV410) || - (r300->radeon.radeonScreen->chip_family == CHIP_FAMILY_R420) ) ) ) { + if (r300->radeon.radeonScreen->chip_family >= CHIP_FAMILY_RV350) { ALLOC_STATE(rb3d_discard_src_pixel_lte_threshold, always, 3, 0); } else { ALLOC_STATE(rb3d_discard_src_pixel_lte_threshold, never, 3, 0); @@ -697,6 +693,14 @@ void r300InitCmdBuf(r300ContextPtr r300) ALLOC_STATE(zs, always, R300_ZS_CMDSIZE, 0); r300->hw.zs.cmd[R300_ZS_CMD_0] = cmdpacket0(r300->radeon.radeonScreen, R300_ZB_CNTL, 3); + if (is_r500) { + if (r300->radeon.radeonScreen->kernel_mm) + ALLOC_STATE(zsb, always, R300_ZSB_CMDSIZE, 0); + else + ALLOC_STATE(zsb, never, R300_ZSB_CMDSIZE, 0); + r300->hw.zsb.cmd[R300_ZSB_CMD_0] = + cmdpacket0(r300->radeon.radeonScreen, R500_ZB_STENCILREFMASK_BF, 1); + } ALLOC_STATE(zstencil_format, always, 5, 0); r300->hw.zstencil_format.cmd[0] = diff --git a/src/mesa/drivers/dri/r300/r300_context.h b/src/mesa/drivers/dri/r300/r300_context.h index 1dadcc0a697..518d5cdbf4f 100644 --- a/src/mesa/drivers/dri/r300/r300_context.h +++ b/src/mesa/drivers/dri/r300/r300_context.h @@ -234,6 +234,10 @@ typedef struct r300_context *r300ContextPtr; #define R300_ZS_CNTL_2 3 #define R300_ZS_CMDSIZE 4 +#define R300_ZSB_CMD_0 0 +#define R300_ZSB_CNTL_0 1 +#define R300_ZSB_CMDSIZE 2 + #define R300_ZB_CMD_0 0 #define R300_ZB_OFFSET 1 #define R300_ZB_PITCH 2 @@ -343,6 +347,7 @@ struct r300_hw_state { struct radeon_state_atom rb3d_aaresolve_ctl; /* (4E88) */ struct radeon_state_atom rb3d_discard_src_pixel_lte_threshold; /* (4E88) I saw it only written on RV350 hardware.. */ struct radeon_state_atom zs; /* zstencil control (4F00) */ + struct radeon_state_atom zsb; /* zstencil bf */ struct radeon_state_atom zstencil_format; struct radeon_state_atom zb; /* z buffer (4F20) */ struct radeon_state_atom zb_depthclearvalue; /* (4F28) */ diff --git a/src/mesa/drivers/dri/r300/r300_reg.h b/src/mesa/drivers/dri/r300/r300_reg.h index 98512d778e3..b9ccd098dc8 100644 --- a/src/mesa/drivers/dri/r300/r300_reg.h +++ b/src/mesa/drivers/dri/r300/r300_reg.h @@ -2313,6 +2313,8 @@ USE OR OTHER DEALINGS IN THE SOFTWARE. # define R300_Z_WRITE_ENABLE (1 << 2) # define R300_Z_SIGNED_COMPARE (1 << 3) # define R300_STENCIL_FRONT_BACK (1 << 4) +# define R400_ZSIGNED_MAGNITUDE (1 << 5) +# define R500_STENCIL_REFMASK_FRONT_BACK (1 << 6) #define R300_ZB_ZSTENCILCNTL 0x4f04 /* functions */ diff --git a/src/mesa/drivers/dri/r300/r300_state.c b/src/mesa/drivers/dri/r300/r300_state.c index d4c3ecee669..3060f49aaf8 100644 --- a/src/mesa/drivers/dri/r300/r300_state.c +++ b/src/mesa/drivers/dri/r300/r300_state.c @@ -590,7 +590,9 @@ static void r300SetDepthState(GLcontext * ctx) r300ContextPtr r300 = R300_CONTEXT(ctx); R300_STATECHANGE(r300, zs); - r300->hw.zs.cmd[R300_ZS_CNTL_0] &= R300_STENCIL_ENABLE|R300_STENCIL_FRONT_BACK; + r300->hw.zs.cmd[R300_ZS_CNTL_0] &= (R300_STENCIL_ENABLE | + R300_STENCIL_FRONT_BACK | + R500_STENCIL_REFMASK_FRONT_BACK); r300->hw.zs.cmd[R300_ZS_CNTL_1] &= ~(R300_ZS_MASK << R300_Z_FUNC_SHIFT); if (ctx->Depth.Test) { @@ -604,11 +606,16 @@ static void r300SetDepthState(GLcontext * ctx) static void r300CatchStencilFallback(GLcontext *ctx) { + r300ContextPtr rmesa = R300_CONTEXT(ctx); const unsigned back = ctx->Stencil._BackFace; - if (ctx->Stencil._Enabled && (ctx->Stencil.Ref[0] != ctx->Stencil.Ref[back] - || ctx->Stencil.ValueMask[0] != ctx->Stencil.ValueMask[back] - || ctx->Stencil.WriteMask[0] != ctx->Stencil.WriteMask[back])) { + if (rmesa->radeon.radeonScreen->kernel_mm && + (rmesa->radeon.radeonScreen->chip_family >= CHIP_FAMILY_RV515)) { + r300SwitchFallback(ctx, R300_FALLBACK_STENCIL_TWOSIDE, GL_FALSE); + } else if (ctx->Stencil._Enabled && + (ctx->Stencil.Ref[0] != ctx->Stencil.Ref[back] + || ctx->Stencil.ValueMask[0] != ctx->Stencil.ValueMask[back] + || ctx->Stencil.WriteMask[0] != ctx->Stencil.WriteMask[back])) { r300SwitchFallback(ctx, R300_FALLBACK_STENCIL_TWOSIDE, GL_TRUE); } else { r300SwitchFallback(ctx, R300_FALLBACK_STENCIL_TWOSIDE, GL_FALSE); @@ -915,11 +922,24 @@ static void r300StencilFuncSeparate(GLcontext * ctx, GLenum face, rmesa->hw.zs.cmd[R300_ZS_CNTL_1] |= (flag << R300_S_BACK_FUNC_SHIFT); rmesa->hw.zs.cmd[R300_ZS_CNTL_2] |= refmask; + + if (rmesa->radeon.radeonScreen->chip_family >= CHIP_FAMILY_RV515) { + rmesa->hw.zs.cmd[R300_ZS_CNTL_0] |= R500_STENCIL_REFMASK_FRONT_BACK; + R300_STATECHANGE(rmesa, zsb); + refmask = ((ctx->Stencil.Ref[back] & 0xff) << R300_STENCILREF_SHIFT) + | ((ctx->Stencil.ValueMask[back] & 0xff) << R300_STENCILMASK_SHIFT); + + rmesa->hw.zsb.cmd[R300_ZSB_CNTL_0] &= + ~((R300_STENCILREF_MASK << R300_STENCILREF_SHIFT) | + (R300_STENCILREF_MASK << R300_STENCILMASK_SHIFT)); + rmesa->hw.zsb.cmd[R300_ZSB_CNTL_0] |= refmask; + } } static void r300StencilMaskSeparate(GLcontext * ctx, GLenum face, GLuint mask) { r300ContextPtr rmesa = R300_CONTEXT(ctx); + const unsigned back = ctx->Stencil._BackFace; r300CatchStencilFallback(ctx); @@ -931,6 +951,13 @@ static void r300StencilMaskSeparate(GLcontext * ctx, GLenum face, GLuint mask) (ctx->Stencil. WriteMask[0] & R300_STENCILREF_MASK) << R300_STENCILWRITEMASK_SHIFT; + if (rmesa->radeon.radeonScreen->chip_family >= CHIP_FAMILY_RV515) { + R300_STATECHANGE(rmesa, zsb); + rmesa->hw.zsb.cmd[R300_ZSB_CNTL_0] |= + (ctx->Stencil. + WriteMask[back] & R300_STENCILREF_MASK) << + R300_STENCILWRITEMASK_SHIFT; + } } static void r300StencilOpSeparate(GLcontext * ctx, GLenum face, diff --git a/src/mesa/drivers/dri/r600/r600_context.h b/src/mesa/drivers/dri/r600/r600_context.h index 8ae05a301c7..c59df7505af 100644 --- a/src/mesa/drivers/dri/r600/r600_context.h +++ b/src/mesa/drivers/dri/r600/r600_context.h @@ -51,6 +51,7 @@ WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. #include "r700_chip.h" #include "r600_tex.h" #include "r700_oglprog.h" +#include "r700_vertprog.h" struct r600_context; typedef struct r600_context context_t; @@ -155,6 +156,8 @@ struct r600_context { struct r600_hw_state atoms; + struct r700_vertex_program *selected_vp; + /* Vertex buffers */ GLvector4f dummy_attrib[_TNL_ATTRIB_MAX]; diff --git a/src/mesa/drivers/dri/r600/r600_texstate.c b/src/mesa/drivers/dri/r600/r600_texstate.c index fff6e743860..7d7e77d355b 100644 --- a/src/mesa/drivers/dri/r600/r600_texstate.c +++ b/src/mesa/drivers/dri/r600/r600_texstate.c @@ -69,7 +69,7 @@ void r600UpdateTextureState(GLcontext * ctx) for (unit = 0; unit < R700_MAX_TEXTURE_UNITS; unit++) { texUnit = &ctx->Texture.Unit[unit]; t = radeon_tex_obj(ctx->Texture.Unit[unit]._Current); - + r700->textures[unit] = NULL; if (texUnit->_ReallyEnabled) { if (!t) continue; @@ -608,6 +608,11 @@ static void setup_hardware_state(context_t *rmesa, struct gl_texture_object *tex int firstlevel = t->mt ? t->mt->firstLevel : 0; GLuint uTexelPitch, row_align; + if (rmesa->radeon.radeonScreen->driScreen->dri2.enabled && + t->image_override && + t->bo) + return; + firstImage = t->base.Image[0][firstlevel]; if (!t->image_override) { diff --git a/src/mesa/drivers/dri/r600/r700_assembler.c b/src/mesa/drivers/dri/r600/r700_assembler.c index 2d8480daaf7..fda66927253 100644 --- a/src/mesa/drivers/dri/r600/r700_assembler.c +++ b/src/mesa/drivers/dri/r600/r700_assembler.c @@ -1149,41 +1149,49 @@ GLboolean tex_dst(r700_AssemblerBase *pAsm) GLboolean tex_src(r700_AssemblerBase *pAsm) { struct prog_instruction *pILInst = &(pAsm->pILInst[pAsm->uiCurInst]); - + GLboolean bValidTexCoord = GL_FALSE; - switch (pILInst->SrcReg[0].File) - { + switch (pILInst->SrcReg[0].File) { + case PROGRAM_CONSTANT: + case PROGRAM_LOCAL_PARAM: + case PROGRAM_ENV_PARAM: + case PROGRAM_STATE_VAR: + bValidTexCoord = GL_TRUE; + setaddrmode_PVSSRC(&(pAsm->S[0].src), ADDR_ABSOLUTE); + pAsm->S[0].src.rtype = SRC_REG_TEMPORARY; + pAsm->S[0].src.reg = pAsm->aArgSubst[1]; + break; case PROGRAM_TEMPORARY: - bValidTexCoord = GL_TRUE; - - pAsm->S[0].src.reg = pILInst->SrcReg[0].Index + pAsm->starting_temp_register_number; - pAsm->S[0].src.rtype = SRC_REG_TEMPORARY; - - break; + bValidTexCoord = GL_TRUE; + pAsm->S[0].src.reg = pILInst->SrcReg[0].Index + + pAsm->starting_temp_register_number; + pAsm->S[0].src.rtype = SRC_REG_TEMPORARY; + break; case PROGRAM_INPUT: - switch (pILInst->SrcReg[0].Index) - { - case FRAG_ATTRIB_COL0: - case FRAG_ATTRIB_COL1: - case FRAG_ATTRIB_TEX0: - case FRAG_ATTRIB_TEX1: - case FRAG_ATTRIB_TEX2: - case FRAG_ATTRIB_TEX3: - case FRAG_ATTRIB_TEX4: - case FRAG_ATTRIB_TEX5: - case FRAG_ATTRIB_TEX6: - case FRAG_ATTRIB_TEX7: - bValidTexCoord = GL_TRUE; - - pAsm->S[0].src.reg = pAsm->uiFP_AttributeMap[pILInst->SrcReg[0].Index]; - pAsm->S[0].src.rtype = SRC_REG_INPUT; - } - break; + switch (pILInst->SrcReg[0].Index) + { + case FRAG_ATTRIB_COL0: + case FRAG_ATTRIB_COL1: + case FRAG_ATTRIB_TEX0: + case FRAG_ATTRIB_TEX1: + case FRAG_ATTRIB_TEX2: + case FRAG_ATTRIB_TEX3: + case FRAG_ATTRIB_TEX4: + case FRAG_ATTRIB_TEX5: + case FRAG_ATTRIB_TEX6: + case FRAG_ATTRIB_TEX7: + bValidTexCoord = GL_TRUE; + pAsm->S[0].src.reg = + pAsm->uiFP_AttributeMap[pILInst->SrcReg[0].Index]; + pAsm->S[0].src.rtype = SRC_REG_INPUT; + break; + } + break; } if(GL_TRUE == bValidTexCoord) - { + { setaddrmode_PVSSRC(&(pAsm->S[0].src), ADDR_ABSOLUTE); } else @@ -1201,7 +1209,7 @@ GLboolean tex_src(r700_AssemblerBase *pAsm) pAsm->S[0].src.negy = (pILInst->SrcReg[0].Negate >> 1) & 0x1; pAsm->S[0].src.negz = (pILInst->SrcReg[0].Negate >> 2) & 0x1; pAsm->S[0].src.negw = (pILInst->SrcReg[0].Negate >> 3) & 0x1; - + return GL_TRUE; } @@ -2202,7 +2210,9 @@ GLboolean next_ins(r700_AssemblerBase *pAsm) { struct prog_instruction *pILInst = &(pAsm->pILInst[pAsm->uiCurInst]); - if( GL_TRUE == IsTex(pILInst->Opcode) ) + if( GL_TRUE == IsTex(pILInst->Opcode) && + /* handle const moves to temp register */ + !(pAsm->D.dst.opcode == SQ_OP2_INST_MOV) ) { if (pILInst->TexSrcTarget == TEXTURE_RECT_INDEX) { if( GL_FALSE == assemble_tex_instruction(pAsm, GL_FALSE) ) @@ -3374,28 +3384,30 @@ GLboolean assemble_TEX(r700_AssemblerBase *pAsm) case PROGRAM_ENV_PARAM: case PROGRAM_STATE_VAR: src_const = GL_TRUE; + break; case PROGRAM_TEMPORARY: case PROGRAM_INPUT: src_const = GL_FALSE; + break; } - if (GL_TRUE == src_const) + if (GL_TRUE == src_const) { - radeon_error("TODO: Texture coordinates from a constant register not supported.\n"); - return GL_FALSE; + if ( GL_FALSE == mov_temp(pAsm, 0) ) + return GL_FALSE; } - switch (pAsm->pILInst[pAsm->uiCurInst].Opcode) + switch (pAsm->pILInst[pAsm->uiCurInst].Opcode) { case OPCODE_TEX: - pAsm->D.dst.opcode = SQ_TEX_INST_SAMPLE; + pAsm->D.dst.opcode = SQ_TEX_INST_SAMPLE; break; - case OPCODE_TXB: + case OPCODE_TXB: radeon_error("do not support TXB yet\n"); return GL_FALSE; break; - case OPCODE_TXP: - /* TODO : tex proj version : divid first 3 components by 4th */ + case OPCODE_TXP: + /* TODO : tex proj version : divid first 3 components by 4th */ pAsm->D.dst.opcode = SQ_TEX_INST_SAMPLE; break; default: @@ -3418,13 +3430,13 @@ GLboolean assemble_TEX(r700_AssemblerBase *pAsm) { return GL_FALSE; } - + if( GL_FALSE == tex_src(pAsm) ) { return GL_FALSE; } - if ( GL_FALSE == next_ins(pAsm) ) + if ( GL_FALSE == next_ins(pAsm) ) { return GL_FALSE; } diff --git a/src/mesa/drivers/dri/r600/r700_chip.c b/src/mesa/drivers/dri/r600/r700_chip.c index 37bff56f5a9..1b560591974 100644 --- a/src/mesa/drivers/dri/r600/r700_chip.c +++ b/src/mesa/drivers/dri/r600/r700_chip.c @@ -52,38 +52,40 @@ static void r700SendTexState(GLcontext *ctx, struct radeon_state_atom *atom) radeon_print(RADEON_STATE, RADEON_VERBOSE, "%s\n", __func__); for (i = 0; i < R700_TEXTURE_NUMBERUNITS; i++) { - radeonTexObj *t = r700->textures[i]; - if (t) { - if (!t->image_override) - bo = t->mt->bo; - else - bo = t->bo; - if (bo) { - - r700SyncSurf(context, bo, - RADEON_GEM_DOMAIN_GTT|RADEON_GEM_DOMAIN_VRAM, - 0, TC_ACTION_ENA_bit); - - BEGIN_BATCH_NO_AUTOSTATE(9 + 4); - R600_OUT_BATCH(CP_PACKET3(R600_IT_SET_RESOURCE, 7)); - R600_OUT_BATCH(i * 7); - R600_OUT_BATCH(r700->textures[i]->SQ_TEX_RESOURCE0); - R600_OUT_BATCH(r700->textures[i]->SQ_TEX_RESOURCE1); - R600_OUT_BATCH(r700->textures[i]->SQ_TEX_RESOURCE2); - R600_OUT_BATCH(r700->textures[i]->SQ_TEX_RESOURCE3); - R600_OUT_BATCH(r700->textures[i]->SQ_TEX_RESOURCE4); - R600_OUT_BATCH(r700->textures[i]->SQ_TEX_RESOURCE5); - R600_OUT_BATCH(r700->textures[i]->SQ_TEX_RESOURCE6); - R600_OUT_BATCH_RELOC(r700->textures[i]->SQ_TEX_RESOURCE2, - bo, - 0, - RADEON_GEM_DOMAIN_GTT|RADEON_GEM_DOMAIN_VRAM, 0, 0); - R600_OUT_BATCH_RELOC(r700->textures[i]->SQ_TEX_RESOURCE3, - bo, - r700->textures[i]->SQ_TEX_RESOURCE3, - RADEON_GEM_DOMAIN_GTT|RADEON_GEM_DOMAIN_VRAM, 0, 0); - END_BATCH(); - COMMIT_BATCH(); + if (ctx->Texture.Unit[i]._ReallyEnabled) { + radeonTexObj *t = r700->textures[i]; + if (t) { + if (!t->image_override) + bo = t->mt->bo; + else + bo = t->bo; + if (bo) { + + r700SyncSurf(context, bo, + RADEON_GEM_DOMAIN_GTT|RADEON_GEM_DOMAIN_VRAM, + 0, TC_ACTION_ENA_bit); + + BEGIN_BATCH_NO_AUTOSTATE(9 + 4); + R600_OUT_BATCH(CP_PACKET3(R600_IT_SET_RESOURCE, 7)); + R600_OUT_BATCH(i * 7); + R600_OUT_BATCH(r700->textures[i]->SQ_TEX_RESOURCE0); + R600_OUT_BATCH(r700->textures[i]->SQ_TEX_RESOURCE1); + R600_OUT_BATCH(r700->textures[i]->SQ_TEX_RESOURCE2); + R600_OUT_BATCH(r700->textures[i]->SQ_TEX_RESOURCE3); + R600_OUT_BATCH(r700->textures[i]->SQ_TEX_RESOURCE4); + R600_OUT_BATCH(r700->textures[i]->SQ_TEX_RESOURCE5); + R600_OUT_BATCH(r700->textures[i]->SQ_TEX_RESOURCE6); + R600_OUT_BATCH_RELOC(r700->textures[i]->SQ_TEX_RESOURCE2, + bo, + 0, + RADEON_GEM_DOMAIN_GTT|RADEON_GEM_DOMAIN_VRAM, 0, 0); + R600_OUT_BATCH_RELOC(r700->textures[i]->SQ_TEX_RESOURCE3, + bo, + r700->textures[i]->SQ_TEX_RESOURCE3, + RADEON_GEM_DOMAIN_GTT|RADEON_GEM_DOMAIN_VRAM, 0, 0); + END_BATCH(); + COMMIT_BATCH(); + } } } } @@ -98,16 +100,18 @@ static void r700SendTexSamplerState(GLcontext *ctx, struct radeon_state_atom *at radeon_print(RADEON_STATE, RADEON_VERBOSE, "%s\n", __func__); for (i = 0; i < R700_TEXTURE_NUMBERUNITS; i++) { - radeonTexObj *t = r700->textures[i]; - if (t) { - BEGIN_BATCH_NO_AUTOSTATE(5); - R600_OUT_BATCH(CP_PACKET3(R600_IT_SET_SAMPLER, 3)); - R600_OUT_BATCH(i * 3); - R600_OUT_BATCH(r700->textures[i]->SQ_TEX_SAMPLER0); - R600_OUT_BATCH(r700->textures[i]->SQ_TEX_SAMPLER1); - R600_OUT_BATCH(r700->textures[i]->SQ_TEX_SAMPLER2); - END_BATCH(); - COMMIT_BATCH(); + if (ctx->Texture.Unit[i]._ReallyEnabled) { + radeonTexObj *t = r700->textures[i]; + if (t) { + BEGIN_BATCH_NO_AUTOSTATE(5); + R600_OUT_BATCH(CP_PACKET3(R600_IT_SET_SAMPLER, 3)); + R600_OUT_BATCH(i * 3); + R600_OUT_BATCH(r700->textures[i]->SQ_TEX_SAMPLER0); + R600_OUT_BATCH(r700->textures[i]->SQ_TEX_SAMPLER1); + R600_OUT_BATCH(r700->textures[i]->SQ_TEX_SAMPLER2); + END_BATCH(); + COMMIT_BATCH(); + } } } } @@ -121,16 +125,18 @@ static void r700SendTexBorderColorState(GLcontext *ctx, struct radeon_state_atom radeon_print(RADEON_STATE, RADEON_VERBOSE, "%s\n", __func__); for (i = 0; i < R700_TEXTURE_NUMBERUNITS; i++) { - radeonTexObj *t = r700->textures[i]; - if (t) { - BEGIN_BATCH_NO_AUTOSTATE(2 + 4); - R600_OUT_BATCH_REGSEQ((TD_PS_SAMPLER0_BORDER_RED + (i * 16)), 4); - R600_OUT_BATCH(r700->textures[i]->TD_PS_SAMPLER0_BORDER_RED); - R600_OUT_BATCH(r700->textures[i]->TD_PS_SAMPLER0_BORDER_GREEN); - R600_OUT_BATCH(r700->textures[i]->TD_PS_SAMPLER0_BORDER_BLUE); - R600_OUT_BATCH(r700->textures[i]->TD_PS_SAMPLER0_BORDER_ALPHA); - END_BATCH(); - COMMIT_BATCH(); + if (ctx->Texture.Unit[i]._ReallyEnabled) { + radeonTexObj *t = r700->textures[i]; + if (t) { + BEGIN_BATCH_NO_AUTOSTATE(2 + 4); + R600_OUT_BATCH_REGSEQ((TD_PS_SAMPLER0_BORDER_RED + (i * 16)), 4); + R600_OUT_BATCH(r700->textures[i]->TD_PS_SAMPLER0_BORDER_RED); + R600_OUT_BATCH(r700->textures[i]->TD_PS_SAMPLER0_BORDER_GREEN); + R600_OUT_BATCH(r700->textures[i]->TD_PS_SAMPLER0_BORDER_BLUE); + R600_OUT_BATCH(r700->textures[i]->TD_PS_SAMPLER0_BORDER_ALPHA); + END_BATCH(); + COMMIT_BATCH(); + } } } } @@ -205,8 +211,7 @@ static void r700SetupVTXConstants(GLcontext * ctx, void r700SetupStreams(GLcontext *ctx) { context_t *context = R700_CONTEXT(ctx); - struct r700_vertex_program *vpc - = (struct r700_vertex_program *)ctx->VertexProgram._Current; + struct r700_vertex_program *vp = context->selected_vp; TNLcontext *tnl = TNL_CONTEXT(ctx); struct vertex_buffer *vb = &tnl->vb; unsigned int i, j = 0; @@ -215,7 +220,7 @@ void r700SetupStreams(GLcontext *ctx) R600_STATECHANGE(context, vtx); for(i=0; i<VERT_ATTRIB_MAX; i++) { - if(vpc->mesa_program.Base.InputsRead & (1 << i)) { + if(vp->mesa_program->Base.InputsRead & (1 << i)) { rcommon_emit_vector(ctx, &context->radeon.tcl.aos[j], vb->AttribPtr[i]->data, @@ -231,8 +236,7 @@ void r700SetupStreams(GLcontext *ctx) static void r700SendVTXState(GLcontext *ctx, struct radeon_state_atom *atom) { context_t *context = R700_CONTEXT(ctx); - struct r700_vertex_program *vpc - = (struct r700_vertex_program *)ctx->VertexProgram._Current; + struct r700_vertex_program *vp = context->selected_vp; unsigned int i, j = 0; BATCH_LOCALS(&context->radeon); radeon_print(RADEON_STATE, RADEON_VERBOSE, "%s\n", __func__); @@ -252,7 +256,7 @@ static void r700SendVTXState(GLcontext *ctx, struct radeon_state_atom *atom) COMMIT_BATCH(); for(i=0; i<VERT_ATTRIB_MAX; i++) { - if(vpc->mesa_program.Base.InputsRead & (1 << i)) { + if(vp->mesa_program->Base.InputsRead & (1 << i)) { /* currently aos are packed */ r700SetupVTXConstants(ctx, i, @@ -1176,9 +1180,11 @@ static int check_tx(GLcontext *ctx, struct radeon_state_atom *atom) R700_CHIP_CONTEXT *r700 = (R700_CHIP_CONTEXT*)(&context->hw); for (i = 0; i < R700_TEXTURE_NUMBERUNITS; i++) { - radeonTexObj *t = r700->textures[i]; - if (t) - count++; + if (ctx->Texture.Unit[i]._ReallyEnabled) { + radeonTexObj *t = r700->textures[i]; + if (t) + count++; + } } radeon_print(RADEON_STATE, RADEON_TRACE, "%s %d\n", __func__, count); return count * 31; diff --git a/src/mesa/drivers/dri/r600/r700_oglprog.c b/src/mesa/drivers/dri/r600/r700_oglprog.c index 3c8c1fd7a34..5290ef31be3 100644 --- a/src/mesa/drivers/dri/r600/r700_oglprog.c +++ b/src/mesa/drivers/dri/r600/r700_oglprog.c @@ -46,7 +46,7 @@ static struct gl_program *r700NewProgram(GLcontext * ctx, { struct gl_program *pProgram = NULL; - struct r700_vertex_program *vp; + struct r700_vertex_program_cont *vpc; struct r700_fragment_program *fp; radeon_print(RADEON_SHADER, RADEON_VERBOSE, @@ -56,16 +56,11 @@ static struct gl_program *r700NewProgram(GLcontext * ctx, { case GL_VERTEX_STATE_PROGRAM_NV: case GL_VERTEX_PROGRAM_ARB: - vp = CALLOC_STRUCT(r700_vertex_program); + vpc = CALLOC_STRUCT(r700_vertex_program_cont); pProgram = _mesa_init_vertex_program(ctx, - &vp->mesa_program, + &vpc->mesa_program, target, id); - vp->translated = GL_FALSE; - vp->loaded = GL_FALSE; - - vp->shaderbo = NULL; - break; case GL_FRAGMENT_PROGRAM_NV: case GL_FRAGMENT_PROGRAM_ARB: @@ -89,7 +84,8 @@ static struct gl_program *r700NewProgram(GLcontext * ctx, static void r700DeleteProgram(GLcontext * ctx, struct gl_program *prog) { - struct r700_vertex_program * vp; + struct r700_vertex_program_cont * vpc; + struct r700_vertex_program *vp, *tmp; struct r700_fragment_program * fp; radeon_print(RADEON_SHADER, RADEON_VERBOSE, @@ -99,14 +95,20 @@ static void r700DeleteProgram(GLcontext * ctx, struct gl_program *prog) { case GL_VERTEX_STATE_PROGRAM_NV: case GL_VERTEX_PROGRAM_ARB: - vp = (struct r700_vertex_program*)prog; - /* Release DMA region */ - - r600DeleteShader(ctx, vp->shaderbo); - - /* Clean up */ - Clean_Up_Assembler(&(vp->r700AsmCode)); - Clean_Up_Shader(&(vp->r700Shader)); + vpc = (struct r700_vertex_program_cont*)prog; + vp = vpc->progs; + while (vp) { + tmp = vp->next; + /* Release DMA region */ + + r600DeleteShader(ctx, vp->shaderbo); + + /* Clean up */ + Clean_Up_Assembler(&(vp->r700AsmCode)); + Clean_Up_Shader(&(vp->r700Shader)); + _mesa_free(vp); + vp = tmp; + } break; case GL_FRAGMENT_PROGRAM_NV: case GL_FRAGMENT_PROGRAM_ARB: diff --git a/src/mesa/drivers/dri/r600/r700_render.c b/src/mesa/drivers/dri/r600/r700_render.c index 3566bf3ca78..b1c3648ca56 100644 --- a/src/mesa/drivers/dri/r600/r700_render.c +++ b/src/mesa/drivers/dri/r600/r700_render.c @@ -319,14 +319,13 @@ static GLuint r700PredictRenderSize(GLcontext* ctx) { context_t *context = R700_CONTEXT(ctx); TNLcontext *tnl = TNL_CONTEXT(ctx); - struct r700_vertex_program *vpc - = (struct r700_vertex_program *)ctx->VertexProgram._Current; + struct r700_vertex_program *vp = context->selected_vp; struct vertex_buffer *vb = &tnl->vb; GLboolean flushed; GLuint dwords, i; GLuint state_size; /* pre calculate aos count so state prediction works */ - context->radeon.tcl.aos_count = _mesa_bitcount(vpc->mesa_program.Base.InputsRead); + context->radeon.tcl.aos_count = _mesa_bitcount(vp->mesa_program->Base.InputsRead); dwords = PRE_EMIT_STATE_BUFSZ; for (i = 0; i < vb->PrimitiveCount; i++) @@ -365,7 +364,6 @@ static GLboolean r700RunRender(GLcontext * ctx, /* mark vtx as dirty since it changes per-draw */ R600_STATECHANGE(context, vtx); - r700UpdateShaders(ctx); r700SetScissor(context); r700SetupVertexProgram(ctx); r700SetupFragmentProgram(ctx); @@ -427,7 +425,10 @@ static GLboolean r700RunTCLRender(GLcontext * ctx, /*----------------------*/ /* TODO : sw fallback */ + /* Need shader bo's setup before bo check */ + r700UpdateShaders(ctx); /** + * Ensure all enabled and complete textures are uploaded along with any buffers being used. */ if(!r600ValidateBuffers(ctx)) diff --git a/src/mesa/drivers/dri/r600/r700_state.c b/src/mesa/drivers/dri/r600/r700_state.c index 93b4ebfdb96..e91aa43118f 100644 --- a/src/mesa/drivers/dri/r600/r700_state.c +++ b/src/mesa/drivers/dri/r600/r700_state.c @@ -1280,8 +1280,8 @@ void r700SetScissor(context_t *context) //--------------- if (context->radeon.radeonScreen->driScreen->dri2.enabled) { x1 = 0; y1 = 0; - x2 = rrb->base.Width - 1; - y2 = rrb->base.Height - 1; + x2 = rrb->base.Width; + y2 = rrb->base.Height; } else { x1 = rrb->dPriv->x; y1 = rrb->dPriv->y; diff --git a/src/mesa/drivers/dri/r600/r700_vertprog.c b/src/mesa/drivers/dri/r600/r700_vertprog.c index d107f99e7ba..8c2b0071df9 100644 --- a/src/mesa/drivers/dri/r600/r700_vertprog.c +++ b/src/mesa/drivers/dri/r600/r700_vertprog.c @@ -35,6 +35,7 @@ #include "main/mtypes.h" #include "tnl/t_context.h" +#include "shader/program.h" #include "shader/prog_parameter.h" #include "shader/prog_statevars.h" @@ -258,28 +259,54 @@ GLboolean Find_Instruction_Dependencies_vp(struct r700_vertex_program *vp, return GL_TRUE; } -GLboolean r700TranslateVertexShader(struct r700_vertex_program *vp, - struct gl_vertex_program *mesa_vp) +struct r700_vertex_program* r700TranslateVertexShader(GLcontext *ctx, + struct gl_vertex_program *mesa_vp) { + context_t *context = R700_CONTEXT(ctx); + struct r700_vertex_program *vp; + TNLcontext *tnl = TNL_CONTEXT(ctx); + struct vertex_buffer *vb = &tnl->vb; + unsigned int unBit; + unsigned int i; + + vp = _mesa_calloc(sizeof(*vp)); + vp->mesa_program = (struct gl_vertex_program *)_mesa_clone_program(ctx, &mesa_vp->Base); + + for(i=0; i<VERT_ATTRIB_MAX; i++) + { + unBit = 1 << i; + if(vp->mesa_program->Base.InputsRead & unBit) /* ctx->Array.ArrayObj->xxxxxxx */ + { + vp->aos_desc[i].size = vb->AttribPtr[i]->size; + vp->aos_desc[i].stride = vb->AttribPtr[i]->size * sizeof(GL_FLOAT);/* when emit array, data is packed. vb->AttribPtr[i]->stride;*/ + vp->aos_desc[i].type = GL_FLOAT; + } + } + + if (context->radeon.radeonScreen->chip_family < CHIP_FAMILY_RV770) + { + vp->r700AsmCode.bR6xx = 1; + } + //Init_Program Init_r700_AssemblerBase(SPT_VP, &(vp->r700AsmCode), &(vp->r700Shader) ); Map_Vertex_Program( vp, mesa_vp ); if(GL_FALSE == Find_Instruction_Dependencies_vp(vp, mesa_vp)) { - return GL_FALSE; + return NULL; } if(GL_FALSE == AssembleInstr(mesa_vp->Base.NumInstructions, &(mesa_vp->Base.Instructions[0]), &(vp->r700AsmCode)) ) { - return GL_FALSE; + return NULL; } if(GL_FALSE == Process_Vertex_Exports(&(vp->r700AsmCode), mesa_vp->Base.OutputsWritten) ) { - return GL_FALSE; + return NULL; } vp->r700Shader.nRegs = (vp->r700AsmCode.number_used_registers == 0) ? 0 @@ -289,72 +316,82 @@ GLboolean r700TranslateVertexShader(struct r700_vertex_program *vp, vp->translated = GL_TRUE; - return GL_TRUE; + return vp; } void r700SelectVertexShader(GLcontext *ctx) { context_t *context = R700_CONTEXT(ctx); - struct r700_vertex_program *vpc - = (struct r700_vertex_program *)ctx->VertexProgram._Current; + struct r700_vertex_program_cont *vpc; + struct r700_vertex_program *vp; TNLcontext *tnl = TNL_CONTEXT(ctx); struct vertex_buffer *vb = &tnl->vb; unsigned int unBit; unsigned int i; + GLboolean match; + vpc = (struct r700_vertex_program_cont *)ctx->VertexProgram._Current; + +#if 0 if (context->radeon.NewGLState & (_NEW_PROGRAM_CONSTANTS|_NEW_PROGRAM)) { vpc->needUpdateVF = 1; } +#endif - if (context->radeon.radeonScreen->chip_family < CHIP_FAMILY_RV770) + for (vp = vpc->progs; vp; vp = vp->next) { - vpc->r700AsmCode.bR6xx = 1; - } - + match = GL_TRUE; for(i=0; i<VERT_ATTRIB_MAX; i++) { unBit = 1 << i; - if(vpc->mesa_program.Base.InputsRead & unBit) /* ctx->Array.ArrayObj->xxxxxxx */ + if(vpc->mesa_program.Base.InputsRead & unBit) { - vpc->aos_desc[i].size = vb->AttribPtr[i]->size; - vpc->aos_desc[i].stride = vb->AttribPtr[i]->size * sizeof(GL_FLOAT);/* when emit array, data is packed. vb->AttribPtr[i]->stride;*/ - vpc->aos_desc[i].type = GL_FLOAT; + if (vp->aos_desc[i].size != vb->AttribPtr[i]->size) + match = GL_FALSE; + break; } } - - if(GL_FALSE == vpc->translated) { - r700TranslateVertexShader(vpc, &(vpc->mesa_program) ); + if (match) + { + context->selected_vp = vp; + return; } + } + + vp = r700TranslateVertexShader(ctx, &(vpc->mesa_program) ); + if(!vp) + { + radeon_error("Failed to translate vertex shader. \n"); + return; + } + vp->next = vpc->progs; + vpc->progs = vp; + context->selected_vp = vp; + return; } void * r700GetActiveVpShaderBo(GLcontext * ctx) { - struct r700_vertex_program *vp - = (struct r700_vertex_program *)ctx->VertexProgram._Current; + context_t *context = R700_CONTEXT(ctx); + struct r700_vertex_program *vp = context->selected_vp;; - return vp->shaderbo; + if (vp) + return vp->shaderbo; + else + return NULL; } GLboolean r700SetupVertexProgram(GLcontext * ctx) { context_t *context = R700_CONTEXT(ctx); R700_CHIP_CONTEXT *r700 = (R700_CHIP_CONTEXT*)(&context->hw); - struct r700_vertex_program *vp - = (struct r700_vertex_program *)ctx->VertexProgram._Current; + struct r700_vertex_program *vp = context->selected_vp; struct gl_program_parameter_list *paramList; unsigned int unNumParamData; unsigned int ui; - if (vp->needUpdateVF) - { - vp->loaded = GL_FALSE; - vp->r700Shader.bNeedsAssembly = GL_TRUE; - Process_Vertex_Program_Vfetch_Instructions(vp, &(vp->mesa_program)); - r600DeleteShader(ctx, vp->shaderbo); - } - if(GL_FALSE == vp->loaded) { if(vp->r700Shader.bNeedsAssembly == GL_TRUE) @@ -410,7 +447,7 @@ GLboolean r700SetupVertexProgram(GLcontext * ctx) */ /* sent out shader constants. */ - paramList = vp->mesa_program.Base.Parameters; + paramList = vp->mesa_program->Base.Parameters; if(NULL != paramList) { _mesa_load_state_parameters(ctx, paramList); diff --git a/src/mesa/drivers/dri/r600/r700_vertprog.h b/src/mesa/drivers/dri/r600/r700_vertprog.h index e2e65021fd3..c48764c43ba 100644 --- a/src/mesa/drivers/dri/r600/r700_vertprog.h +++ b/src/mesa/drivers/dri/r600/r700_vertprog.h @@ -43,7 +43,7 @@ typedef struct ArrayDesc //TEMP struct r700_vertex_program { - struct gl_vertex_program mesa_program; /* Must be first */ + struct gl_vertex_program *mesa_program; /* Must be first */ struct r700_vertex_program *next; @@ -59,6 +59,13 @@ struct r700_vertex_program ArrayDesc aos_desc[VERT_ATTRIB_MAX]; }; +struct r700_vertex_program_cont +{ + struct gl_vertex_program mesa_program; + + struct r700_vertex_program *progs; +}; + //Internal unsigned int Map_Vertex_Output(r700_AssemblerBase *pAsm, struct gl_vertex_program *mesa_vp, @@ -74,7 +81,7 @@ void Map_Vertex_Program(struct r700_vertex_program *vp, GLboolean Find_Instruction_Dependencies_vp(struct r700_vertex_program *vp, struct gl_vertex_program *mesa_vp); -GLboolean r700TranslateVertexShader(struct r700_vertex_program *vp, +struct r700_vertex_program* r700TranslateVertexShader(GLcontext *ctx, struct gl_vertex_program *mesa_vp); /* Interface */ diff --git a/src/mesa/drivers/dri/radeon/radeon_common.c b/src/mesa/drivers/dri/radeon/radeon_common.c index e53eb0904d2..a4c7b40798a 100644 --- a/src/mesa/drivers/dri/radeon/radeon_common.c +++ b/src/mesa/drivers/dri/radeon/radeon_common.c @@ -1234,7 +1234,9 @@ int rcommonFlushCmdBuf(radeonContextPtr rmesa, const char *caller) UNLOCK_HARDWARE(rmesa); if (ret) { - fprintf(stderr, "drmRadeonCmdBuffer: %d\n", ret); + fprintf(stderr, "drmRadeonCmdBuffer: %d. Kernel failed to " + "parse or rejected command stream. See dmesg " + "for more info.\n", ret); _mesa_exit(ret); } diff --git a/src/mesa/drivers/dri/radeon/radeon_debug.c b/src/mesa/drivers/dri/radeon/radeon_debug.c index a1ed39683f1..3b6f0038037 100644 --- a/src/mesa/drivers/dri/radeon/radeon_debug.c +++ b/src/mesa/drivers/dri/radeon/radeon_debug.c @@ -32,6 +32,9 @@ #include "radeon_debug.h" #include "radeon_common_context.h" +#include <stdarg.h> +#include <stdio.h> + static const struct dri_debug_control debug_control[] = { {"fall", RADEON_FALLBACKS}, {"tex", RADEON_TEXTURE}, @@ -85,10 +88,10 @@ void _radeon_debug_remove_indent(void) } } -extern void _radeon_print(const radeon_debug_type_t type, +void _radeon_print(const radeon_debug_type_t type, const radeon_debug_level_t level, const char* message, - va_list values) + ...) { GET_CURRENT_CONTEXT(ctx); if (ctx) { @@ -97,5 +100,8 @@ extern void _radeon_print(const radeon_debug_type_t type, if (radeon->debug.indent_depth) fprintf(stderr, "%s", radeon->debug.indent); } + va_list values; + va_start( values, message ); vfprintf(stderr, message, values); + va_end( values ); } diff --git a/src/mesa/drivers/dri/radeon/radeon_debug.h b/src/mesa/drivers/dri/radeon/radeon_debug.h index 132e27351da..2a8302293b2 100644 --- a/src/mesa/drivers/dri/radeon/radeon_debug.h +++ b/src/mesa/drivers/dri/radeon/radeon_debug.h @@ -30,8 +30,7 @@ #ifndef RADEON_DEBUG_H_INCLUDED #define RADEON_DEBUG_H_INCLUDED -#include <stdarg.h> -#include <stdio.h> +#include <stdlib.h> typedef enum radeon_debug_levels { RADEON_CRITICAL = 0, /* Only errors */ @@ -102,57 +101,36 @@ static inline int radeon_is_debug_enabled(const radeon_debug_type_t type, extern void _radeon_print(const radeon_debug_type_t type, const radeon_debug_level_t level, const char* message, - va_list values); -/** - * Format attribute requires declaration for setting it. Don't ask me why! - */ -static inline void radeon_print(const radeon_debug_type_t type, - const radeon_debug_level_t level, - const char* message, - ...) __attribute__((format(printf,3,4))); - + ...) __attribute__((format(printf,3,4))); /** * Print out debug message if channel specified by type is enabled * and compile time debugging level is at least as high as level parameter */ -static inline void radeon_print(const radeon_debug_type_t type, - const radeon_debug_level_t level, - const char* message, - ...) -{ - /* Compile out if level of message is too high */ - if (radeon_is_debug_enabled(type, level)) { - - va_list values; - va_start( values, message ); - _radeon_print(type, level, message, values); - va_end( values ); - } -} +#define radeon_print(type, level, message, ...) do { \ + const radeon_debug_level_t _debug_level = (level); \ + const radeon_debug_type_t _debug_type = (type); \ + /* Compile out if level of message is too high */ \ + if (radeon_is_debug_enabled(type, level)) { \ + _radeon_print(_debug_type, _debug_level, \ + (message), ## __VA_ARGS__); \ + } \ +} while(0) -static inline void radeon_error(const char* message, ...) __attribute__((format(printf,1,2))); /** * printf style function for writing error messages. */ -static inline void radeon_error(const char* message, ...) -{ - va_list values; - va_start( values, message ); - radeon_print(RADEON_GENERAL, RADEON_CRITICAL, message, values); - va_end( values ); -} +#define radeon_error(message, ...) do { \ + radeon_print(RADEON_GENERAL, RADEON_CRITICAL, \ + (message), ## __VA_ARGS__); \ +} while(0) -static inline void radeon_warning(const char* message, ...) __attribute__((format(printf,1,2))); /** * printf style function for writing warnings. */ -static inline void radeon_warning(const char* message, ...) -{ - va_list values; - va_start( values, message ); - radeon_print(RADEON_GENERAL, RADEON_IMPORTANT, message, values); - va_end( values ); -} +#define radeon_warning(message, ...) do { \ + radeon_print(RADEON_GENERAL, RADEON_IMPORTANT, \ + (message), ## __VA_ARGS__); \ +} while(0) extern void radeon_init_debug(void); extern void _radeon_debug_add_indent(void); diff --git a/src/mesa/drivers/dri/swrast/swrast.c b/src/mesa/drivers/dri/swrast/swrast.c index a858af30c11..3016987d561 100644 --- a/src/mesa/drivers/dri/swrast/swrast.c +++ b/src/mesa/drivers/dri/swrast/swrast.c @@ -62,6 +62,7 @@ #define need_GL_SGI_color_table /* sw extensions not associated with some GL version */ +#define need_GL_ARB_draw_elements_base_vertex #define need_GL_ARB_shader_objects #define need_GL_ARB_vertex_array_object #define need_GL_ARB_vertex_program @@ -95,6 +96,8 @@ const struct dri_extension card_extensions[] = { "GL_EXT_histogram", GL_EXT_histogram_functions }, { "GL_SGI_color_table", GL_SGI_color_table_functions }, + { "GL_ARB_depth_clamp", NULL }, + { "GL_ARB_draw_elements_base_vertex", GL_ARB_draw_elements_base_vertex_functions }, { "GL_ARB_shader_objects", GL_ARB_shader_objects_functions }, { "GL_ARB_vertex_array_object", GL_ARB_vertex_array_object_functions }, { "GL_ARB_vertex_program", GL_ARB_vertex_program_functions }, @@ -109,6 +112,7 @@ const struct dri_extension card_extensions[] = { "GL_EXT_paletted_texture", GL_EXT_paletted_texture_functions }, { "GL_EXT_stencil_two_side", GL_EXT_stencil_two_side_functions }, { "GL_MESA_resize_buffers", GL_MESA_resize_buffers_functions }, + { "GL_NV_depth_clamp", NULL }, { "GL_NV_vertex_program", GL_NV_vertex_program_functions }, { "GL_NV_fragment_program", GL_NV_fragment_program_functions }, { NULL, NULL } diff --git a/src/mesa/drivers/x11/xm_api.c b/src/mesa/drivers/x11/xm_api.c index 2c7be9f182d..662c61ae7e7 100644 --- a/src/mesa/drivers/x11/xm_api.c +++ b/src/mesa/drivers/x11/xm_api.c @@ -1317,6 +1317,7 @@ xmesa_convert_from_x_visual_type( int visualType ) #define need_GL_SGI_color_table /* sw extensions not associated with some GL version */ +#define need_GL_ARB_draw_elements_base_vertex #define need_GL_ARB_shader_objects #define need_GL_ARB_sync #define need_GL_ARB_vertex_program @@ -1347,6 +1348,8 @@ const struct dri_extension card_extensions[] = { "GL_EXT_histogram", GL_EXT_histogram_functions }, { "GL_SGI_color_table", GL_SGI_color_table_functions }, + { "GL_ARB_depth_clamp", NULL }, + { "GL_ARB_draw_elements_base_vertex", GL_ARB_draw_elements_base_vertex_functions }, { "GL_ARB_shader_objects", GL_ARB_shader_objects_functions }, { "GL_ARB_sync", GL_ARB_sync_functions }, { "GL_ARB_vertex_program", GL_ARB_vertex_program_functions }, @@ -1358,6 +1361,7 @@ const struct dri_extension card_extensions[] = { "GL_EXT_gpu_program_parameters", GL_EXT_gpu_program_parameters_functions }, { "GL_EXT_paletted_texture", GL_EXT_paletted_texture_functions }, { "GL_MESA_resize_buffers", GL_MESA_resize_buffers_functions }, + { "GL_NV_depth_clamp", NULL }, { "GL_NV_vertex_program", GL_NV_vertex_program_functions }, { "GL_NV_fragment_program", GL_NV_fragment_program_functions }, { NULL, NULL } |