diff options
374 files changed, 20498 insertions, 13052 deletions
diff --git a/configs/default b/configs/default index 9e4bb08ea9e..cb3ca1046f4 100644 --- a/configs/default +++ b/configs/default @@ -23,6 +23,7 @@ HOST_CC = $(CC) CFLAGS = -O CXXFLAGS = -O LDFLAGS = +HOST_CFLAGS = $(CFLAGS) GLU_CFLAGS = # Compiler for building demos/tests/etc diff --git a/configure.ac b/configure.ac index c5ef676e2e5..2881bb6bc25 100644 --- a/configure.ac +++ b/configure.ac @@ -1148,6 +1148,11 @@ yes) if test "$tracker" = egl && test "x$enable_egl" != xyes; then AC_MSG_ERROR([cannot build egl state tracker without EGL library]) fi + if test "$tracker" = xorg; then + PKG_CHECK_MODULES(XEXT, [xextproto >= 7.0.99.1], + HAVE_XEXTPROTO_71="yes"; DEFINES="$DEFINES -DHAVE_XEXTPROTO_71" + HAVE_XEXTPROTO_71="no") + fi done GALLIUM_STATE_TRACKERS_DIRS="$state_trackers" ;; @@ -1202,7 +1207,7 @@ if test "x$enable_gallium_radeon" = xyes; then fi dnl -dnl Gallium Radeon configuration +dnl Gallium Nouveau configuration dnl AC_ARG_ENABLE([gallium-nouveau], [AS_HELP_STRING([--enable-gallium-nouveau], diff --git a/docs/relnotes-7.5.2.html b/docs/relnotes-7.5.2.html index 32100142c0c..b638c0517dd 100644 --- a/docs/relnotes-7.5.2.html +++ b/docs/relnotes-7.5.2.html @@ -45,6 +45,10 @@ tbd <ul> <li>Assorted bug fixes for i965/i945 drivers <li>Fixed Gallium glDrawPixels(GL_STENCIL_INDEX) failure. +<li>Fixed GLSL linker/preprocessor version directive issue seen in Wine + (such as bug 23946) +<li>glUseProgram() is now compiled into display lists (bug 23746). +<li>glUniform functions are now compiled into display lists </ul> diff --git a/docs/relnotes-7.6.html b/docs/relnotes-7.6.html index 8a476378e8a..aaa36188d9d 100644 --- a/docs/relnotes-7.6.html +++ b/docs/relnotes-7.6.html @@ -50,6 +50,8 @@ This was written by Zack Rusin at Tungsten Graphics. <li>Rewritten radeon/r200/r300 driver using a buffer manager <li>radeon/r200/r300 GL_EXT_framebuffer_object support when used with kernel memory manager +<li>radeon/r200/r300 support for GL_ARB_occlusion_query</li> +<li>r300 driver supports OpenGL 1.5</li> <li>r300 driver support for GL_EXT_vertex_array_bgra, GL_EXT_texture_sRGB <li>i915/945 driver support for GL_ARB_point_sprite, GL_EXT_stencil_two_side and GL_ATI_separate_stencil extensions @@ -57,6 +59,10 @@ This was written by Zack Rusin at Tungsten Graphics. GL_ARB_fragment_program.</li> <li>Added configure --with-max-width=W, --with-max-height=H options to specify max framebuffer, viewport size. +<li>Initial version of Gallium llvmpipe driver. This is a new driver based + on LLVM which makes exensive use of run-time code generation. This is + an "alpha" stage driver. See the src/gallium/drivers/llvmpipe/README + file for more information. </ul> diff --git a/docs/relnotes-7.7.html b/docs/relnotes-7.7.html index 755e8ac5208..ca6ec55b601 100644 --- a/docs/relnotes-7.7.html +++ b/docs/relnotes-7.7.html @@ -37,6 +37,7 @@ tbd <li>GL_ARB_draw_elements_base_vertex (supported in Intel i965 and software drivers)</li> <li>GL_ARB_depth_clamp (supported in Intel i965 DRI and software drivers)</li> <li>GL_NV_depth_clamp (supported in Intel i965 DRI and software drivers)</li> +<li>GL_ARB_provoking_vertex (same as GL_EXT_provoking_vertex)</li> </ul> diff --git a/include/GL/gl.h b/include/GL/gl.h index 05362b97762..5f8bc2b708c 100644 --- a/include/GL/gl.h +++ b/include/GL/gl.h @@ -1740,6 +1740,9 @@ GLAPI void GLAPIENTRY glSeparableFilter2D( GLenum target, GLAPI void GLAPIENTRY glGetSeparableFilter( GLenum target, GLenum format, GLenum type, GLvoid *row, GLvoid *column, GLvoid *span ); +typedef void (APIENTRYP PFNGLBLENDCOLORPROC) (GLclampf red, GLclampf green, GLclampf blue, GLclampf alpha); +typedef void (APIENTRYP PFNGLBLENDEQUATIONPROC) (GLenum mode); + /* @@ -1945,6 +1948,18 @@ GLAPI void GLAPIENTRY glMultTransposeMatrixf( const GLfloat m[16] ); GLAPI void GLAPIENTRY glSampleCoverage( GLclampf value, GLboolean invert ); +typedef void (APIENTRYP PFNGLACTIVETEXTUREPROC) (GLenum texture); +typedef void (APIENTRYP PFNGLSAMPLECOVERAGEPROC) (GLclampf value, GLboolean invert); +typedef void (APIENTRYP PFNGLCOMPRESSEDTEXIMAGE3DPROC) (GLenum target, GLint level, GLenum internalformat, GLsizei width, GLsizei height, GLsizei depth, GLint border, GLsizei imageSize, const GLvoid *data); +typedef void (APIENTRYP PFNGLCOMPRESSEDTEXIMAGE2DPROC) (GLenum target, GLint level, GLenum internalformat, GLsizei width, GLsizei height, GLint border, GLsizei imageSize, const GLvoid *data); +typedef void (APIENTRYP PFNGLCOMPRESSEDTEXIMAGE1DPROC) (GLenum target, GLint level, GLenum internalformat, GLsizei width, GLint border, GLsizei imageSize, const GLvoid *data); +typedef void (APIENTRYP PFNGLCOMPRESSEDTEXSUBIMAGE3DPROC) (GLenum target, GLint level, GLint xoffset, GLint yoffset, GLint zoffset, GLsizei width, GLsizei height, GLsizei depth, GLenum format, GLsizei imageSize, const GLvoid *data); +typedef void (APIENTRYP PFNGLCOMPRESSEDTEXSUBIMAGE2DPROC) (GLenum target, GLint level, GLint xoffset, GLint yoffset, GLsizei width, GLsizei height, GLenum format, GLsizei imageSize, const GLvoid *data); +typedef void (APIENTRYP PFNGLCOMPRESSEDTEXSUBIMAGE1DPROC) (GLenum target, GLint level, GLint xoffset, GLsizei width, GLenum format, GLsizei imageSize, const GLvoid *data); +typedef void (APIENTRYP PFNGLGETCOMPRESSEDTEXIMAGEPROC) (GLenum target, GLint level, GLvoid *img); + + + /* * GL_ARB_multitexture (ARB extension 1 and OpenGL 1.2.1) */ diff --git a/progs/SConscript b/progs/SConscript index 620dd30e69c..66eaf9e5410 100644 --- a/progs/SConscript +++ b/progs/SConscript @@ -10,4 +10,5 @@ SConscript([ 'vpglsl/SConscript', 'fp/SConscript', 'wgl/SConscript', + 'perf/SConscript', ]) diff --git a/progs/demos/cubemap.c b/progs/demos/cubemap.c index 0df5ff09c33..20332b1d960 100644 --- a/progs/demos/cubemap.c +++ b/progs/demos/cubemap.c @@ -58,6 +58,9 @@ static GLint ClampIndex = 0; static GLboolean supportFBO = GL_FALSE; static GLboolean supportSeamless = GL_FALSE; static GLboolean seamless = GL_FALSE; +static GLuint TexObj = 0; +static GLint T0 = 0; +static GLint Frames = 0; static struct { @@ -282,6 +285,20 @@ static void draw( void ) glPopMatrix(); glutSwapBuffers(); + + Frames++; + + { + GLint t = glutGet(GLUT_ELAPSED_TIME); + if (t - T0 >= 5000) { + GLfloat seconds = (t - T0) / 1000.0; + GLfloat fps = Frames / seconds; + printf("%d frames in %6.3f seconds = %6.3f FPS\n", Frames, seconds, fps); + fflush(stdout); + T0 = t; + Frames = 0; + } + } } @@ -543,6 +560,10 @@ static void init( GLboolean useImageFiles ) printf("GL_RENDERER: %s\n", (char *) glGetString(GL_RENDERER)); + + glGenTextures(1, &TexObj); + glBindTexture(GL_TEXTURE_CUBE_MAP_ARB, TexObj); + if (useImageFiles) { load_envmaps(); } diff --git a/progs/demos/lodbias.c b/progs/demos/lodbias.c index 30b1ed13d5f..8d39bd605a7 100644 --- a/progs/demos/lodbias.c +++ b/progs/demos/lodbias.c @@ -43,6 +43,7 @@ static GLboolean Anim = GL_TRUE; static GLint Bias = 0, BiasStepSign = +1; /* ints avoid fp precision problem */ static GLint BiasMin = -400, BiasMax = 400; static int win = 0; +static GLuint TexObj = 0; static void @@ -214,6 +215,9 @@ static void Init( void ) glPixelStorei(GL_UNPACK_ALIGNMENT, 1); + glGenTextures(1, &TexObj); + glBindTexture(GL_TEXTURE_2D, TexObj); + if (glutExtensionSupported("GL_SGIS_generate_mipmap")) { /* test auto mipmap generation */ GLint width, height, i; diff --git a/progs/fp/add-sat.txt b/progs/fp/add-sat.txt new file mode 100644 index 00000000000..2253efb0855 --- /dev/null +++ b/progs/fp/add-sat.txt @@ -0,0 +1,6 @@ +!!ARBfp1.0 +TEMP R0; +MOV R0, fragment.color; +ADD_SAT R0, R0, R0; +MUL result.color, {0.5}.x, R0; +END diff --git a/progs/fp/mov-alias.txt b/progs/fp/mov-alias.txt new file mode 100644 index 00000000000..5f04e9c76e2 --- /dev/null +++ b/progs/fp/mov-alias.txt @@ -0,0 +1,6 @@ +!!ARBfp1.0 +TEMP R0; +MOV R0, fragment.color; +MOV R0, R0.zyxw; +MOV result.color, R0; +END diff --git a/progs/fp/mul-alias.txt b/progs/fp/mul-alias.txt new file mode 100644 index 00000000000..cf7d359e780 --- /dev/null +++ b/progs/fp/mul-alias.txt @@ -0,0 +1,6 @@ +!!ARBfp1.0 +TEMP R0; +MOV R0, fragment.color; +MUL R0, R0.zyxw, fragment.color; +MOV result.color, R0; +END diff --git a/progs/glsl/Makefile b/progs/glsl/Makefile index 8103a5cbca0..8928c833c0e 100644 --- a/progs/glsl/Makefile +++ b/progs/glsl/Makefile @@ -10,16 +10,15 @@ LIB_DEP = \ $(TOP)/$(LIB_DIR)/$(GLU_LIB_NAME) \ $(TOP)/$(LIB_DIR)/$(GLUT_LIB_NAME) -LIBS = -L$(TOP)/$(LIB_DIR) -l$(GLUT_LIB) -l$(GLEW_LIB) -l$(GLU_LIB) -l$(GL_LIB) $(APP_LIB_DEPS) - -INCLUDE_DIRS = -I$(TOP)/progs/util +LIBS = -L$(TOP)/$(LIB_DIR) -l$(GLUT_LIB) -l$(GLEW_LIB) -l$(GLU_LIB) \ + -l$(GL_LIB) $(APP_LIB_DEPS) # using : to avoid APP_CC pointing to CC loop -CC:=$(APP_CC) -CFLAGS += -I$(INCDIR) -LDLIBS=$(LIBS) +CC := $(APP_CC) +CFLAGS := -I$(INCDIR) $(CFLAGS) +LDLIBS = $(LIBS) -DEMO_SOURCES = \ +PROG_SOURCES = \ array.c \ bitmap.c \ brick.c \ @@ -59,8 +58,8 @@ UTIL_SOURCES = \ readtex.c UTIL_OBJS = $(UTIL_SOURCES:.c=.o) -PROG_OBJS = $(DEMO_SOURCES:.c=.o) -PROGS = $(DEMO_SOURCES:%.c=%) +PROG_OBJS = $(PROG_SOURCES:.c=.o) +PROGS = $(PROG_SOURCES:%.c=%) ##### TARGETS ##### diff --git a/progs/perf/Makefile b/progs/perf/Makefile new file mode 100644 index 00000000000..066eb8608ec --- /dev/null +++ b/progs/perf/Makefile @@ -0,0 +1,55 @@ +# progs/demos/Makefile + +TOP = ../.. +include $(TOP)/configs/current + +INCDIR = $(TOP)/include + +LIBS = -L$(TOP)/$(LIB_DIR) -l$(GLUT_LIB) -l$(GLEW_LIB) \ + -l$(GLU_LIB) -l$(GL_LIB) $(APP_LIB_DEPS) + +# using : to avoid APP_CC pointing to CC loop +CC := $(APP_CC) +CFLAGS += -I$(INCDIR) +LDLIBS = $(LIBS) + +PROG_SOURCES = \ + copytex.c \ + drawoverhead.c \ + fbobind.c \ + fill.c \ + genmipmap.c \ + readpixels.c \ + swapbuffers.c \ + teximage.c \ + vbo.c \ + vertexrate.c \ + +PROG_OBJS = $(PROG_SOURCES:.c=.o) + +PROGS = $(PROG_SOURCES:%.c=%) + + +UTIL_SOURCES = \ + common.c \ + glmain.c + +UTIL_HEADERS = \ + common.h \ + glmain.h + +UTIL_OBJS = $(UTIL_SOURCES:.c=.o) + + + +default: $(PROGS) + +$(PROG_OBJS): $(UTIL_HEADERS) + +$(PROGS): $(UTIL_OBJS) + + + +clean: + -rm -f $(PROGS) + -rm -f *.o *~ diff --git a/progs/perf/SConscript b/progs/perf/SConscript new file mode 100644 index 00000000000..a5ec9a7c2a0 --- /dev/null +++ b/progs/perf/SConscript @@ -0,0 +1,32 @@ +Import('env') + +if not env['GLUT']: + Return() + +env = env.Clone() + +env.Prepend(LIBS = ['$GLUT_LIB']) + +progs = [ + 'copytex', + 'drawoverhead', + 'fbobind', + 'fill', + 'genmipmap', + 'readpixels', + 'swapbuffers', + 'teximage', + 'vbo', + 'vertexrate', +] + +for prog in progs: + env.Program( + target = prog, + source = [ + prog + '.c', + 'common.c', + 'glmain.c', + ] + ) + diff --git a/progs/perf/common.c b/progs/perf/common.c new file mode 100644 index 00000000000..722f4b7b454 --- /dev/null +++ b/progs/perf/common.c @@ -0,0 +1,133 @@ +/* + * Copyright (C) 2009 VMware, Inc. 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, sublicense, + * 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 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 NONINFRINGEMENT. IN NO EVENT SHALL + * VMWARE 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. + */ + +/** + * Common perf code. This should be re-usable with other APIs. + */ + +#include "common.h" +#include "glmain.h" + +#include <stdio.h> +#include <stdlib.h> +#include <stdarg.h> + + +/* Need to add a fflush windows console with mingw, otherwise nothing + * shows up until program exit. May want to add logging here. + */ +void +perf_printf(const char *format, ...) +{ + va_list ap; + va_start(ap, format); + + fflush(stdout); + vfprintf(stdout, format, ap); + fflush(stdout); + + va_end(ap); +} + + + +/** + * Run function 'f' for enough iterations to reach a steady state. + * Return the rate (iterations/second). + */ +double +PerfMeasureRate(PerfRateFunc f) +{ + const double minDuration = 1.0; + double rate = 0.0, prevRate = 0.0; + unsigned subiters; + + /* Compute initial number of iterations to try. + * If the test function is pretty slow this helps to avoid + * extraordarily long run times. + */ + subiters = 2; + { + const double t0 = PerfGetTime(); + double t1; + do { + f(subiters); /* call the rendering function */ + t1 = PerfGetTime(); + subiters *= 2; + } while (t1 - t0 < 0.1 * minDuration); + } + /*perf_printf("initial subIters = %u\n", subiters);*/ + + while (1) { + const double t0 = PerfGetTime(); + unsigned iters = 0; + double t1; + + do { + f(subiters); /* call the rendering function */ + t1 = PerfGetTime(); + iters += subiters; + } while (t1 - t0 < minDuration); + + rate = iters / (t1 - t0); + + if (0) + perf_printf("prevRate %f rate %f ratio %f iters %u\n", + prevRate, rate, rate/prevRate, iters); + + /* Try and speed the search up by skipping a few steps: + */ + if (rate > prevRate * 1.6) + subiters *= 8; + else if (rate > prevRate * 1.2) + subiters *= 4; + else if (rate > prevRate * 1.05) + subiters *= 2; + else + break; + + prevRate = rate; + } + + if (0) + perf_printf("%s returning iters %u rate %f\n", __FUNCTION__, subiters, rate); + return rate; +} + + +/* Note static buffer, can only use once per printf. + */ +const char * +PerfHumanFloat( double d ) +{ + static char buf[80]; + + if (d > 1000000000.0) + snprintf(buf, sizeof(buf), "%.1f billion", d / 1000000000.0); + else if (d > 1000000.0) + snprintf(buf, sizeof(buf), "%.1f million", d / 1000000.0); + else if (d > 1000.0) + snprintf(buf, sizeof(buf), "%.1f thousand", d / 1000.0); + else + snprintf(buf, sizeof(buf), "%.1f", d); + + return buf; +} diff --git a/progs/perf/common.h b/progs/perf/common.h new file mode 100644 index 00000000000..6ea17402b52 --- /dev/null +++ b/progs/perf/common.h @@ -0,0 +1,44 @@ +/* + * Copyright (C) 2009 VMware, Inc. 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, sublicense, + * 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 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 NONINFRINGEMENT. IN NO EVENT SHALL + * VMWARE 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. + */ + + +#ifndef COMMON_H +#define COMMON_H + + +#include <stddef.h> /* for offsetof() */ + + +typedef void (*PerfRateFunc)(unsigned count); + + +extern double +PerfMeasureRate(PerfRateFunc f); + +const char * +PerfHumanFloat( double d ); + +extern void +perf_printf(const char *format, ...); + + +#endif /* COMMON_H */ + diff --git a/progs/perf/copytex.c b/progs/perf/copytex.c new file mode 100644 index 00000000000..f7a6b8aec39 --- /dev/null +++ b/progs/perf/copytex.c @@ -0,0 +1,214 @@ +/* + * Copyright (C) 2009 VMware, Inc. 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, sublicense, + * 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 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 NONINFRINGEMENT. IN NO EVENT SHALL + * VMWARE 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. + */ + +/** + * Measure glCopyTex[Sub]Image() rate. + * Create a large, off-screen framebuffer object for rendering and + * copying the texture data from it since we can't make really large + * on-screen windows. + * + * Brian Paul + * 22 Sep 2009 + */ + +#include <string.h> +#include "glmain.h" +#include "common.h" + +int WinWidth = 100, WinHeight = 100; + +static GLuint VBO, FBO, RBO, Tex; + +const GLsizei MinSize = 16, MaxSize = 4096; +static GLsizei TexSize; + +static const GLboolean DrawPoint = GL_TRUE; +static const GLboolean TexSubImage4 = GL_FALSE; + +struct vertex +{ + GLfloat x, y, s, t; +}; + +static const struct vertex vertices[1] = { + { 0.0, 0.0, 0.5, 0.5 }, +}; + +#define VOFFSET(F) ((void *) offsetof(struct vertex, F)) + + +/** Called from test harness/main */ +void +PerfInit(void) +{ + const GLenum filter = GL_LINEAR; + GLenum stat; + + if (!PerfExtensionSupported("GL_EXT_framebuffer_object")) { + perf_printf("copytex: GL_EXT_framebuffer_object not supported\n"); + exit(0); + } + + /* setup VBO */ + glGenBuffersARB(1, &VBO); + glBindBufferARB(GL_ARRAY_BUFFER_ARB, VBO); + glBufferDataARB(GL_ARRAY_BUFFER_ARB, sizeof(vertices), + vertices, GL_STATIC_DRAW_ARB); + + glVertexPointer(2, GL_FLOAT, sizeof(struct vertex), VOFFSET(x)); + glTexCoordPointer(2, GL_FLOAT, sizeof(struct vertex), VOFFSET(s)); + glEnableClientState(GL_VERTEX_ARRAY); + glEnableClientState(GL_TEXTURE_COORD_ARRAY); + + /* setup texture */ + glGenTextures(1, &Tex); + glBindTexture(GL_TEXTURE_2D, Tex); + glTexParameteri(GL_TEXTURE_2D, GL_TEXTURE_MIN_FILTER, filter); + glTexParameteri(GL_TEXTURE_2D, GL_TEXTURE_MAG_FILTER, filter); + glEnable(GL_TEXTURE_2D); + + /* setup rbo */ + glGenRenderbuffersEXT(1, &RBO); + glBindRenderbufferEXT(GL_RENDERBUFFER_EXT, RBO); + glRenderbufferStorageEXT(GL_RENDERBUFFER_EXT, GL_RGBA, MaxSize, MaxSize); + + /* setup fbo */ + glGenFramebuffersEXT(1, &FBO); + glBindFramebufferEXT(GL_FRAMEBUFFER, FBO); + glFramebufferRenderbufferEXT(GL_FRAMEBUFFER_EXT, + GL_COLOR_ATTACHMENT0_EXT, + GL_RENDERBUFFER_EXT, RBO); + + stat = glCheckFramebufferStatusEXT(GL_FRAMEBUFFER_EXT); + if (stat != GL_FRAMEBUFFER_COMPLETE_EXT) { + perf_printf("fboswitch: Error: incomplete FBO!\n"); + exit(1); + } + + /* clear the FBO */ + glDrawBuffer(GL_COLOR_ATTACHMENT0_EXT); + glViewport(0, 0, MaxSize, MaxSize); + glClear(GL_COLOR_BUFFER_BIT); +} + + +static void +CopyTexImage(unsigned count) +{ + unsigned i; + for (i = 1; i < count; i++) { + /* draw something */ + if (DrawPoint) + glDrawArrays(GL_POINTS, 0, 1); + + /* copy whole texture */ + glCopyTexImage2D(GL_TEXTURE_2D, 0, + GL_RGBA, 0, 0, TexSize, TexSize, 0); + } + glFinish(); +} + + +static void +CopyTexSubImage(unsigned count) +{ + unsigned i; + for (i = 1; i < count; i++) { + /* draw something */ + if (DrawPoint) + glDrawArrays(GL_POINTS, 0, 1); + + /* copy sub texture */ + if (TexSubImage4) { + /* four sub-copies */ + GLsizei half = TexSize / 2; + /* lower-left */ + glCopyTexSubImage2D(GL_TEXTURE_2D, 0, + 0, 0, 0, 0, half, half); + /* lower-right */ + glCopyTexSubImage2D(GL_TEXTURE_2D, 0, + half, 0, half, 0, half, half); + /* upper-left */ + glCopyTexSubImage2D(GL_TEXTURE_2D, 0, + 0, half, 0, half, half, half); + /* upper-right */ + glCopyTexSubImage2D(GL_TEXTURE_2D, 0, + half, half, half, half, half, half); + } + else { + /* one big copy */ + glCopyTexSubImage2D(GL_TEXTURE_2D, 0, + 0, 0, 0, 0, TexSize, TexSize); + } + } + glFinish(); +} + + +/** Called from test harness/main */ +void +PerfNextRound(void) +{ +} + + +/** Called from test harness/main */ +void +PerfDraw(void) +{ + double rate, mbPerSec; + GLint sub, maxTexSize; + + glGetIntegerv(GL_MAX_TEXTURE_SIZE, &maxTexSize); + + /* loop over whole/sub tex copy */ + for (sub = 0; sub < 2; sub++) { + + /* loop over texture sizes */ + for (TexSize = MinSize; TexSize <= MaxSize; TexSize *= 4) { + + if (TexSize <= maxTexSize) { + GLint bytesPerImage = 4 * TexSize * TexSize; + + if (sub == 0) + rate = PerfMeasureRate(CopyTexImage); + else { + /* setup empty dest texture */ + glTexImage2D(GL_TEXTURE_2D, 0, GL_RGBA, + TexSize, TexSize, 0, + GL_RGBA, GL_UNSIGNED_BYTE, NULL); + rate = PerfMeasureRate(CopyTexSubImage); + } + + mbPerSec = rate * bytesPerImage / (1024.0 * 1024.0); + } + else { + rate = 0.0; + mbPerSec = 0.0; + } + + perf_printf(" glCopyTex%sImage(%d x %d): %.1f copies/sec, %.1f Mpixels/sec\n", + (sub ? "Sub" : ""), TexSize, TexSize, rate, mbPerSec); + } + } + + exit(0); +} diff --git a/progs/perf/drawoverhead.c b/progs/perf/drawoverhead.c new file mode 100644 index 00000000000..f75c9bb74e8 --- /dev/null +++ b/progs/perf/drawoverhead.c @@ -0,0 +1,137 @@ +/* + * Copyright (C) 2009 VMware, Inc. 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, sublicense, + * 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 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 NONINFRINGEMENT. IN NO EVENT SHALL + * VMWARE 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. + */ + +/** + * Measure drawing overhead + * + * This is the first in a series of simple performance benchmarks. + * The code in this file should be as simple as possible to make it + * easily portable to other APIs. + * + * All the window-system stuff should be contained in glmain.c (or TBDmain.c). + * + * Brian Paul + * 15 Sep 2009 + */ + +#include "glmain.h" +#include "common.h" + + +int WinWidth = 100, WinHeight = 100; + +static GLuint VBO; + +struct vertex +{ + GLfloat x, y; +}; + +static const struct vertex vertices[4] = { + { -1.0, -1.0 }, + { 1.0, -1.0 }, + { 1.0, 1.0 }, + { -1.0, 1.0 } +}; + + +/** Called from test harness/main */ +void +PerfInit(void) +{ + /* setup VBO w/ vertex data */ + glGenBuffersARB(1, &VBO); + glBindBufferARB(GL_ARRAY_BUFFER_ARB, VBO); + glBufferDataARB(GL_ARRAY_BUFFER_ARB, + sizeof(vertices), vertices, GL_STATIC_DRAW_ARB); + glVertexPointer(2, GL_FLOAT, sizeof(struct vertex), (void *) 0); + glEnableClientState(GL_VERTEX_ARRAY); + + /* misc GL state */ + glAlphaFunc(GL_ALWAYS, 0.0); +} + + +static void +DrawNoStateChange(unsigned count) +{ + unsigned i; + for (i = 0; i < count; i++) { + glDrawArrays(GL_POINTS, 0, 4); + } + glFinish(); +} + + +static void +DrawNopStateChange(unsigned count) +{ + unsigned i; + for (i = 0; i < count; i++) { + glDisable(GL_ALPHA_TEST); + glDrawArrays(GL_POINTS, 0, 4); + } + glFinish(); +} + + +static void +DrawStateChange(unsigned count) +{ + unsigned i; + for (i = 0; i < count; i++) { + if (i & 1) + glEnable(GL_TEXTURE_GEN_S); + else + glDisable(GL_TEXTURE_GEN_S); + glDrawArrays(GL_POINTS, 0, 4); + } + glFinish(); +} + +void +PerfNextRound(void) +{ +} + +/** Called from test harness/main */ +void +PerfDraw(void) +{ + double rate0, rate1, rate2, overhead; + + rate0 = PerfMeasureRate(DrawNoStateChange); + perf_printf(" Draw only: %s draws/second\n", + PerfHumanFloat(rate0)); + + rate1 = PerfMeasureRate(DrawNopStateChange); + overhead = 1000.0 * (1.0 / rate1 - 1.0 / rate0); + perf_printf(" Draw w/ nop state change: %s draws/sec (overhead: %f ms/draw)\n", + PerfHumanFloat(rate1), overhead); + + rate2 = PerfMeasureRate(DrawStateChange); + overhead = 1000.0 * (1.0 / rate2 - 1.0 / rate0); + perf_printf(" Draw w/ state change: %s draws/sec (overhead: %f ms/draw)\n", + PerfHumanFloat(rate2), overhead); + + exit(0); +} + diff --git a/progs/perf/fbobind.c b/progs/perf/fbobind.c new file mode 100644 index 00000000000..fb52a93a2f6 --- /dev/null +++ b/progs/perf/fbobind.c @@ -0,0 +1,153 @@ +/* + * Copyright (C) 2009 VMware, Inc. 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, sublicense, + * 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 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 NONINFRINGEMENT. IN NO EVENT SHALL + * VMWARE 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. + */ + +/** + * Measure rate of binding/switching between FBO targets. + * Create two framebuffer objects for rendering to two textures. + * Ping pong between texturing from one and drawing into the other. + * + * Brian Paul + * 22 Sep 2009 + */ + +#include <string.h> +#include "glmain.h" +#include "common.h" + +int WinWidth = 100, WinHeight = 100; + +static GLuint VBO; + +static GLuint FBO[2], Tex[2]; + +static const GLsizei TexSize = 512; + +static const GLboolean DrawPoint = GL_TRUE; + +struct vertex +{ + GLfloat x, y, s, t; +}; + +static const struct vertex vertices[1] = { + { 0.0, 0.0, 0.5, 0.5 }, +}; + +#define VOFFSET(F) ((void *) offsetof(struct vertex, F)) + + +/** Called from test harness/main */ +void +PerfInit(void) +{ + const GLenum filter = GL_LINEAR; + GLenum stat; + int i; + + if (!PerfExtensionSupported("GL_EXT_framebuffer_object")) { + perf_printf("fboswitch: GL_EXT_framebuffer_object not supported\n"); + exit(0); + } + + /* setup VBO */ + glGenBuffersARB(1, &VBO); + glBindBufferARB(GL_ARRAY_BUFFER_ARB, VBO); + glBufferDataARB(GL_ARRAY_BUFFER_ARB, sizeof(vertices), + vertices, GL_STATIC_DRAW_ARB); + + glVertexPointer(2, GL_FLOAT, sizeof(struct vertex), VOFFSET(x)); + glTexCoordPointer(2, GL_FLOAT, sizeof(struct vertex), VOFFSET(s)); + glEnableClientState(GL_VERTEX_ARRAY); + glEnableClientState(GL_TEXTURE_COORD_ARRAY); + + glGenFramebuffersEXT(2, FBO); + glGenTextures(2, Tex); + + for (i = 0; i < 2; i++) { + /* setup texture */ + glBindTexture(GL_TEXTURE_2D, Tex[i]); + glTexImage2D(GL_TEXTURE_2D, 0, GL_RGBA, + TexSize, TexSize, 0, + GL_RGBA, GL_UNSIGNED_BYTE, NULL); + glTexParameteri(GL_TEXTURE_2D, GL_TEXTURE_MIN_FILTER, filter); + glTexParameteri(GL_TEXTURE_2D, GL_TEXTURE_MAG_FILTER, filter); + + + /* setup fbo */ + glBindFramebufferEXT(GL_FRAMEBUFFER, FBO[i]); + glFramebufferTexture2DEXT(GL_FRAMEBUFFER_EXT, + GL_COLOR_ATTACHMENT0_EXT, + GL_TEXTURE_2D, Tex[i], 0/*level*/); + stat = glCheckFramebufferStatusEXT(GL_FRAMEBUFFER_EXT); + if (stat != GL_FRAMEBUFFER_COMPLETE_EXT) { + perf_printf("fboswitch: Error: incomplete FBO!\n"); + exit(1); + } + + /* clear the FBO */ + glClear(GL_COLOR_BUFFER_BIT); + } + + glEnable(GL_TEXTURE_2D); +} + + +static void +FBOBind(unsigned count) +{ + unsigned i; + for (i = 1; i < count; i++) { + const GLuint dst = i & 1; + const GLuint src = 1 - dst; + + /* bind src texture */ + glBindTexture(GL_TEXTURE_2D, Tex[src]); + + /* bind dst fbo */ + glBindFramebufferEXT(GL_FRAMEBUFFER_EXT, FBO[dst]); + glDrawBuffer(GL_COLOR_ATTACHMENT0_EXT); + + /* draw something */ + if (DrawPoint) + glDrawArrays(GL_POINTS, 0, 1); + } + glFinish(); +} + + +/** Called from test harness/main */ +void +PerfNextRound(void) +{ +} + + +/** Called from test harness/main */ +void +PerfDraw(void) +{ + double rate; + + rate = PerfMeasureRate(FBOBind); + perf_printf(" FBO Binding: %1.f binds/sec\n", rate); + + exit(0); +} diff --git a/progs/perf/fill.c b/progs/perf/fill.c new file mode 100644 index 00000000000..279f2b5f189 --- /dev/null +++ b/progs/perf/fill.c @@ -0,0 +1,248 @@ +/* + * Copyright (C) 2009 VMware, Inc. 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, sublicense, + * 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 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 NONINFRINGEMENT. IN NO EVENT SHALL + * VMWARE 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. + */ + +/** + * Measure fill rates. + * + * Brian Paul + * 21 Sep 2009 + */ + +#include "glmain.h" +#include "common.h" + + +int WinWidth = 1000, WinHeight = 1000; + +static GLuint VBO, TexObj; + + +struct vertex +{ + GLfloat x, y, s, t, r, g, b, a; +}; + +#define VOFFSET(F) ((void *) offsetof(struct vertex, F)) + +static const struct vertex vertices[4] = { + /* x y s t r g b a */ + { -1.0, -1.0, 0.0, 0.0, 1.0, 0.0, 0.0, 0.5 }, + { 1.0, -1.0, 1.0, 0.0, 0.0, 1.0, 0.0, 0.5 }, + { 1.0, 1.0, 1.0, 1.0, 0.0, 0.0, 1.0, 0.5 }, + { -1.0, 1.0, 0.0, 1.0, 1.0, 1.0, 1.0, 0.5 } +}; + + +static const char *VertexShader = + "void main() \n" + "{ \n" + " gl_Position = ftransform(); \n" + " gl_TexCoord[0] = gl_MultiTexCoord0; \n" + " gl_FrontColor = gl_Color; \n" + "} \n"; + +/* simple fragment shader */ +static const char *FragmentShader1 = + "uniform sampler2D Tex; \n" + "void main() \n" + "{ \n" + " vec4 t = texture2D(Tex, gl_TexCoord[0].xy); \n" + " gl_FragColor = vec4(1.0) - t * gl_Color; \n" + "} \n"; + +/** + * A more complex fragment shader (but equivalent to first shader). + * A good optimizer should catch some of these no-op operations, but + * probably not all of them. + */ +static const char *FragmentShader2 = + "uniform sampler2D Tex; \n" + "void main() \n" + "{ \n" + " // as above \n" + " vec4 t = texture2D(Tex, gl_TexCoord[0].xy); \n" + " t = vec4(1.0) - t * gl_Color; \n" + + " vec4 u; \n" + + " // no-op negate/swizzle \n" + " u = -t.wzyx; \n" + " t = -u.wzyx; \n" + + " // no-op inverts \n" + " t = vec4(1.0) - t; \n" + " t = vec4(1.0) - t; \n" + + " // no-op min/max \n" + " t = min(t, t); \n" + " t = max(t, t); \n" + + " // no-op moves \n" + " u = t; \n" + " t = u; \n" + " u = t; \n" + " t = u; \n" + + " // no-op add/mul \n" + " t = (t + t + t + t) * 0.25; \n" + + " // no-op mul/sub \n" + " t = 3.0 * t - 2.0 * t; \n" + + " // no-op negate/min/max \n" + " t = -min(-t, -t); \n" + " t = -max(-t, -t); \n" + + " gl_FragColor = t; \n" + "} \n"; + +static GLuint ShaderProg1, ShaderProg2; + + + +/** Called from test harness/main */ +void +PerfInit(void) +{ + GLint u; + + /* setup VBO w/ vertex data */ + glGenBuffersARB(1, &VBO); + glBindBufferARB(GL_ARRAY_BUFFER_ARB, VBO); + glBufferDataARB(GL_ARRAY_BUFFER_ARB, + sizeof(vertices), vertices, GL_STATIC_DRAW_ARB); + glVertexPointer(2, GL_FLOAT, sizeof(struct vertex), VOFFSET(x)); + glTexCoordPointer(2, GL_FLOAT, sizeof(struct vertex), VOFFSET(s)); + glColorPointer(4, GL_FLOAT, sizeof(struct vertex), VOFFSET(r)); + glEnableClientState(GL_VERTEX_ARRAY); + glEnableClientState(GL_COLOR_ARRAY); + + /* setup texture */ + TexObj = PerfCheckerTexture(128, 128); + + /* setup shaders */ + ShaderProg1 = PerfShaderProgram(VertexShader, FragmentShader1); + glUseProgram(ShaderProg1); + u = glGetUniformLocation(ShaderProg1, "Tex"); + glUniform1i(u, 0); /* texture unit 0 */ + + ShaderProg2 = PerfShaderProgram(VertexShader, FragmentShader2); + glUseProgram(ShaderProg2); + u = glGetUniformLocation(ShaderProg2, "Tex"); + glUniform1i(u, 0); /* texture unit 0 */ + + glUseProgram(0); +} + + +static void +Ortho(void) +{ + glMatrixMode(GL_PROJECTION); + glLoadIdentity(); + glOrtho(-1.0, 1.0, -1.0, 1.0, -1.0, 1.0); + glMatrixMode(GL_MODELVIEW); + glLoadIdentity(); +} + + + +static void +DrawQuad(unsigned count) +{ + unsigned i; + glClear(GL_COLOR_BUFFER_BIT); + + for (i = 0; i < count; i++) { + glDrawArrays(GL_TRIANGLE_FAN, 0, 4); + + /* Avoid sending command buffers with huge numbers of fullscreen + * quads. Graphics schedulers don't always cope well with + * this... + */ + if (i % 128 == 0) { + PerfSwapBuffers(); + glClear(GL_COLOR_BUFFER_BIT); + } + } + + glFinish(); + + if (1) + PerfSwapBuffers(); +} + +void +PerfNextRound(void) +{ +} + +/** Called from test harness/main */ +void +PerfDraw(void) +{ + double rate; + double pixelsPerDraw = WinWidth * WinHeight; + + Ortho(); + + /* simple fill */ + rate = PerfMeasureRate(DrawQuad) * pixelsPerDraw; + perf_printf(" Simple fill: %s pixels/second\n", + PerfHumanFloat(rate)); + + /* blended fill */ + glEnable(GL_BLEND); + rate = PerfMeasureRate(DrawQuad) * pixelsPerDraw; + glDisable(GL_BLEND); + perf_printf(" Blended fill: %s pixels/second\n", + PerfHumanFloat(rate)); + + /* textured fill */ + glEnable(GL_TEXTURE_2D); + glEnableClientState(GL_TEXTURE_COORD_ARRAY); + rate = PerfMeasureRate(DrawQuad) * pixelsPerDraw; + glDisable(GL_TEXTURE_2D); + glDisableClientState(GL_TEXTURE_COORD_ARRAY); + perf_printf(" Textured fill: %s pixels/second\n", + PerfHumanFloat(rate)); + + /* shader1 fill */ + glUseProgram(ShaderProg1); + glEnableClientState(GL_TEXTURE_COORD_ARRAY); + rate = PerfMeasureRate(DrawQuad) * pixelsPerDraw; + glUseProgram(0); + glDisableClientState(GL_TEXTURE_COORD_ARRAY); + perf_printf(" Shader1 fill: %s pixels/second\n", + PerfHumanFloat(rate)); + + /* shader2 fill */ + glUseProgram(ShaderProg2); + glEnableClientState(GL_TEXTURE_COORD_ARRAY); + rate = PerfMeasureRate(DrawQuad) * pixelsPerDraw; + glUseProgram(0); + glDisableClientState(GL_TEXTURE_COORD_ARRAY); + perf_printf(" Shader2 fill: %s pixels/second\n", + PerfHumanFloat(rate)); + + exit(0); +} + diff --git a/progs/perf/genmipmap.c b/progs/perf/genmipmap.c new file mode 100644 index 00000000000..4b7d6ad155b --- /dev/null +++ b/progs/perf/genmipmap.c @@ -0,0 +1,136 @@ +/* + * Copyright (C) 2009 VMware, Inc. 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, sublicense, + * 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 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 NONINFRINGEMENT. IN NO EVENT SHALL + * VMWARE 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. + */ + +/** + * Measure glGenerateMipmap() speed. + * + * Brian Paul + * 24 Sep 2009 + */ + +#include <string.h> +#include "glmain.h" +#include "common.h" + + +int WinWidth = 100, WinHeight = 100; + +static GLboolean DrawPoint = GL_TRUE; +static GLuint VBO; +static GLuint TexObj = 0; +static GLint BaseLevel, MaxLevel; + +struct vertex +{ + GLfloat x, y, s, t; +}; + +static const struct vertex vertices[1] = { + { 0.0, 0.0, 0.5, 0.5 }, +}; + +#define VOFFSET(F) ((void *) offsetof(struct vertex, F)) + +/** Called from test harness/main */ +void +PerfInit(void) +{ + /* setup VBO w/ vertex data */ + glGenBuffersARB(1, &VBO); + glBindBufferARB(GL_ARRAY_BUFFER_ARB, VBO); + glBufferDataARB(GL_ARRAY_BUFFER_ARB, + sizeof(vertices), vertices, GL_STATIC_DRAW_ARB); + glVertexPointer(2, GL_FLOAT, sizeof(struct vertex), VOFFSET(x)); + glTexCoordPointer(2, GL_FLOAT, sizeof(struct vertex), VOFFSET(s)); + glEnableClientState(GL_VERTEX_ARRAY); + glEnableClientState(GL_TEXTURE_COORD_ARRAY); + + glGenTextures(1, &TexObj); + glBindTexture(GL_TEXTURE_2D, TexObj); + glTexParameteri(GL_TEXTURE_2D, GL_TEXTURE_MIN_FILTER, GL_NEAREST); + glTexParameteri(GL_TEXTURE_2D, GL_TEXTURE_MAG_FILTER, GL_NEAREST); + glEnable(GL_TEXTURE_2D); +} + + +static void +GenMipmap(unsigned count) +{ + unsigned i; + for (i = 0; i < count; i++) { + GLubyte texel[4]; + texel[0] = texel[1] = texel[2] = texel[3] = i & 0xff; + /* dirty the base image */ + glTexSubImage2D(GL_TEXTURE_2D, BaseLevel, + 0, 0, 1, 1, GL_RGBA, GL_UNSIGNED_BYTE, texel); + glGenerateMipmap(GL_TEXTURE_2D); + if (DrawPoint) + glDrawArrays(GL_POINTS, 0, 1); + } + glFinish(); +} + + +/** Called from test harness/main */ +void +PerfNextRound(void) +{ +} + + +/** Called from test harness/main */ +void +PerfDraw(void) +{ + const GLint NumLevels = 12; + const GLint TexWidth = 2048, TexHeight = 2048; + GLubyte *img; + double rate; + + /* Make 2K x 2K texture */ + img = (GLubyte *) malloc(TexWidth * TexHeight * 4); + memset(img, 128, TexWidth * TexHeight * 4); + glTexImage2D(GL_TEXTURE_2D, 0, + GL_RGBA, TexWidth, TexHeight, 0, + GL_RGBA, GL_UNSIGNED_BYTE, img); + free(img); + + perf_printf("Texture level[0] size: %d x %d, %d levels\n", + TexWidth, TexHeight, NumLevels); + + /* loop over base levels 0, 2, 4 */ + for (BaseLevel = 0; BaseLevel <= 4; BaseLevel += 2) { + + /* loop over max level */ + for (MaxLevel = NumLevels; MaxLevel > BaseLevel; MaxLevel--) { + + glTexParameteri(GL_TEXTURE_2D, GL_TEXTURE_BASE_LEVEL, BaseLevel); + glTexParameteri(GL_TEXTURE_2D, GL_TEXTURE_MAX_LEVEL, MaxLevel); + + rate = PerfMeasureRate(GenMipmap); + + perf_printf(" glGenerateMipmap(levels %d..%d): %.2f gens/sec\n", + BaseLevel + 1, MaxLevel, rate); + } + } + + exit(0); +} diff --git a/progs/perf/glmain.c b/progs/perf/glmain.c new file mode 100644 index 00000000000..69cdbce319e --- /dev/null +++ b/progs/perf/glmain.c @@ -0,0 +1,268 @@ +/* + * Copyright (C) 2009 VMware, Inc. 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, sublicense, + * 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 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 NONINFRINGEMENT. IN NO EVENT SHALL + * VMWARE 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. + */ + +/** + * OpenGL/GLUT common code for perf programs. + * Brian Paul + * 15 Sep 2009 + */ + + +#include <stdio.h> +#include "glmain.h" +#include <GL/glut.h> + + +static int Win; +static GLfloat Xrot = 0, Yrot = 0, Zrot = 0; + + +/** Return time in seconds */ +double +PerfGetTime(void) +{ + return glutGet(GLUT_ELAPSED_TIME) * 0.001; +} + + +void +PerfSwapBuffers(void) +{ + glutSwapBuffers(); +} + + +/** make simple checkerboard texture object */ +GLuint +PerfCheckerTexture(GLsizei width, GLsizei height) +{ + const GLenum filter = GL_NEAREST; + GLubyte *img = (GLubyte *) malloc(width * height * 4); + GLint i, j, k; + GLuint obj; + + k = 0; + for (i = 0; i < height; i++) { + for (j = 0; j < width; j++) { + GLubyte color; + if (((i / 8) ^ (j / 8)) & 1) { + color = 0xff; + } + else { + color = 0x0; + } + img[k++] = color; + img[k++] = color; + img[k++] = color; + img[k++] = color; + } + } + + glGenTextures(1, &obj); + glBindTexture(GL_TEXTURE_2D, obj); + glTexParameteri(GL_TEXTURE_2D, GL_TEXTURE_MIN_FILTER, filter); + glTexParameteri(GL_TEXTURE_2D, GL_TEXTURE_MAG_FILTER, filter); + glTexImage2D(GL_TEXTURE_2D, 0, GL_RGBA, width, height, 0, + GL_RGBA, GL_UNSIGNED_BYTE, img); + free(img); + + return obj; +} + + +static GLuint +CompileShader(GLenum type, const char *shader) +{ + GLuint sh; + GLint stat; + + sh = glCreateShader(type); + glShaderSource(sh, 1, (const GLchar **) &shader, NULL); + + glCompileShader(sh); + + glGetShaderiv(sh, GL_COMPILE_STATUS, &stat); + if (!stat) { + GLchar log[1000]; + GLsizei len; + glGetShaderInfoLog(sh, 1000, &len, log); + fprintf(stderr, "Error: problem compiling shader: %s\n", log); + exit(1); + } + + return sh; +} + + +/** Make shader program from given vert/frag shader text */ +GLuint +PerfShaderProgram(const char *vertShader, const char *fragShader) +{ + GLuint prog; + GLint stat; + + { + const char *version = (const char *) glGetString(GL_VERSION); + if ((version[0] != '2' && + version[0] != '3') || version[1] != '.') { + fprintf(stderr, "Error: GL version 2.x or better required\n"); + exit(1); + } + } + + prog = glCreateProgram(); + + if (vertShader) { + GLuint vs = CompileShader(GL_VERTEX_SHADER, vertShader); + glAttachShader(prog, vs); + } + if (fragShader) { + GLuint fs = CompileShader(GL_FRAGMENT_SHADER, fragShader); + glAttachShader(prog, fs); + } + + glLinkProgram(prog); + glGetProgramiv(prog, GL_LINK_STATUS, &stat); + if (!stat) { + GLchar log[1000]; + GLsizei len; + glGetProgramInfoLog(prog, 1000, &len, log); + fprintf(stderr, "Shader link error:\n%s\n", log); + exit(1); + } + + return prog; +} + + +int +PerfReshapeWindow( unsigned w, unsigned h ) +{ + if (glutGet(GLUT_SCREEN_WIDTH) < w || + glutGet(GLUT_SCREEN_HEIGHT) < h) + return 0; + + glutReshapeWindow( w, h ); + glutPostRedisplay(); + return 1; +} + + +GLboolean +PerfExtensionSupported(const char *ext) +{ + return glutExtensionSupported(ext); +} + + +static void +Idle(void) +{ + PerfNextRound(); +} + + +static void +Draw(void) +{ + PerfDraw(); + glutSwapBuffers(); +} + + +static void +Reshape(int width, int height) +{ + WinWidth = width; + WinHeight = height; + glViewport(0, 0, width, height); + glMatrixMode(GL_PROJECTION); + glLoadIdentity(); + glFrustum(-1.0, 1.0, -1.0, 1.0, 5.0, 25.0); + glMatrixMode(GL_MODELVIEW); + glLoadIdentity(); + glTranslatef(0.0, 0.0, -15.0); +} + + +static void +Key(unsigned char key, int x, int y) +{ + const GLfloat step = 3.0; + (void) x; + (void) y; + switch (key) { + case 'z': + Zrot -= step; + break; + case 'Z': + Zrot += step; + break; + case 27: + glutDestroyWindow(Win); + exit(0); + break; + } + glutPostRedisplay(); +} + + +static void +SpecialKey(int key, int x, int y) +{ + const GLfloat step = 3.0; + (void) x; + (void) y; + switch (key) { + case GLUT_KEY_UP: + Xrot -= step; + break; + case GLUT_KEY_DOWN: + Xrot += step; + break; + case GLUT_KEY_LEFT: + Yrot -= step; + break; + case GLUT_KEY_RIGHT: + Yrot += step; + break; + } + glutPostRedisplay(); +} + + +int +main(int argc, char *argv[]) +{ + glutInit(&argc, argv); + glutInitWindowSize(WinWidth, WinHeight); + glutInitDisplayMode(GLUT_RGB | GLUT_DOUBLE | GLUT_DEPTH | GLUT_STENCIL); + Win = glutCreateWindow(argv[0]); + glewInit(); + glutReshapeFunc(Reshape); + glutKeyboardFunc(Key); + glutSpecialFunc(SpecialKey); + glutDisplayFunc(Draw); + glutIdleFunc(Idle); + PerfInit(); + glutMainLoop(); + return 0; +} diff --git a/progs/perf/glmain.h b/progs/perf/glmain.h new file mode 100644 index 00000000000..d9bcd5f4e21 --- /dev/null +++ b/progs/perf/glmain.h @@ -0,0 +1,68 @@ +/* + * Copyright (C) 2009 VMware, Inc. 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, sublicense, + * 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 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 NONINFRINGEMENT. IN NO EVENT SHALL + * VMWARE 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. + */ + + +#ifndef GLMAIN_H +#define GLMAIN_H + + +#define GL_GLEXT_PROTOTYPES +#include <GL/glew.h> +#include <stdlib.h> +#include <math.h> + + +/** Test programs can use these vars/functions */ + +extern int WinWidth, WinHeight; + +extern double +PerfGetTime(void); + +extern void +PerfSwapBuffers(void); + +extern GLuint +PerfCheckerTexture(GLsizei width, GLsizei height); + +extern GLuint +PerfShaderProgram(const char *vertShader, const char *fragShader); + +extern int +PerfReshapeWindow( unsigned w, unsigned h ); + +extern GLboolean +PerfExtensionSupported(const char *ext); + + +/** Test programs must implement these functions **/ + +extern void +PerfInit(void); + +extern void +PerfNextRound(void); + +extern void +PerfDraw(void); + + +#endif /* GLMAIN_H */ diff --git a/progs/perf/readpixels.c b/progs/perf/readpixels.c new file mode 100644 index 00000000000..ac7dc426e93 --- /dev/null +++ b/progs/perf/readpixels.c @@ -0,0 +1,169 @@ +/* + * Copyright (C) 2009 VMware, Inc. 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, sublicense, + * 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 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 NONINFRINGEMENT. IN NO EVENT SHALL + * VMWARE 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. + */ + +/** + * Measure glReadPixels speed. + * XXX also read into a PBO? + * XXX also read from FBOs? + * + * Brian Paul + * 23 Sep 2009 + */ + +#include <string.h> +#include <assert.h> +#include "glmain.h" +#include "common.h" + +int WinWidth = 1000, WinHeight = 1000; + +static GLuint VBO; + +static const GLboolean DrawPoint = GL_TRUE; +static const GLboolean BufferSubDataInHalves = GL_TRUE; + +static const GLfloat Vertex0[2] = { 0.0, 0.0 }; + +static GLenum HaveDepthStencil; + +static GLenum ReadFormat, ReadType; +static GLint ReadWidth, ReadHeight; +static GLvoid *ReadBuffer; + + +/** Called from test harness/main */ +void +PerfInit(void) +{ + /* setup VBO */ + glGenBuffersARB(1, &VBO); + glBindBufferARB(GL_ARRAY_BUFFER_ARB, VBO); + glBufferDataARB(GL_ARRAY_BUFFER_ARB, sizeof(Vertex0), Vertex0, GL_STATIC_DRAW_ARB); + glVertexPointer(2, GL_FLOAT, sizeof(Vertex0), (void *) 0); + glEnableClientState(GL_VERTEX_ARRAY); + + glPixelStorei(GL_PACK_ALIGNMENT, 1); + + HaveDepthStencil = PerfExtensionSupported("GL_EXT_packed_depth_stencil"); + + glClear(GL_COLOR_BUFFER_BIT | GL_DEPTH_BUFFER_BIT | GL_STENCIL_BUFFER_BIT); + glEnable(GL_DEPTH_TEST); + glEnable(GL_STENCIL_TEST); +} + + +static void +ReadPixels(unsigned count) +{ + unsigned i; + for (i = 0; i < count; i++) { + /* read from random pos */ + GLint x, y; + + x = WinWidth - ReadWidth; + y = WinHeight - ReadHeight; + if (x > 0) + x = rand() % x; + if (y > 0) + y = rand() % y; + + if (DrawPoint) + glDrawArrays(GL_POINTS, 0, 1); + + glReadPixels(x, y, ReadWidth, ReadHeight, + ReadFormat, ReadType, ReadBuffer); + } + glFinish(); +} + + +static const GLsizei Sizes[] = { + 10, + 100, + 500, + 1000, + 0 +}; + + +static const struct { + GLenum format; + GLenum type; + const char *name; + GLuint pixel_size; +} DstFormats[] = { + { GL_RGBA, GL_UNSIGNED_BYTE, "RGBA/ubyte", 4 }, + { GL_BGRA, GL_UNSIGNED_BYTE, "BGRA/ubyte", 4 }, + { GL_RGB, GL_UNSIGNED_SHORT_5_6_5, "RGB/565", 2 }, + { GL_LUMINANCE, GL_UNSIGNED_BYTE, "L/ubyte", 1 }, + { GL_DEPTH_COMPONENT, GL_UNSIGNED_INT, "Z/uint", 4 }, + { GL_DEPTH_STENCIL_EXT, GL_UNSIGNED_INT_24_8_EXT, "Z+S/uint", 4 }, + { 0, 0, NULL, 0 } +}; + + + +/** Called from test harness/main */ +void +PerfNextRound(void) +{ +} + + +/** Called from test harness/main */ +void +PerfDraw(void) +{ + double rate, mbPerSec; + int fmt, sz; + + /* loop over formats */ + for (fmt = 0; DstFormats[fmt].format; fmt++) { + ReadFormat = DstFormats[fmt].format; + ReadType = DstFormats[fmt].type; + + /* loop over sizes */ + for (sz = 0; Sizes[sz]; sz++) { + int imgSize; + + ReadWidth = ReadHeight = Sizes[sz]; + imgSize = ReadWidth * ReadHeight * DstFormats[fmt].pixel_size; + ReadBuffer = malloc(imgSize); + + if (ReadFormat == GL_DEPTH_STENCIL_EXT && !HaveDepthStencil) { + rate = 0.0; + mbPerSec = 0.0; + } + else { + rate = PerfMeasureRate(ReadPixels); + mbPerSec = rate * imgSize / (1024.0 * 1024.0); + } + + perf_printf("glReadPixels(%d x %d, %s): %.1f images/sec, %.1f Mpixels/sec\n", + ReadWidth, ReadHeight, + DstFormats[fmt].name, rate, mbPerSec); + + free(ReadBuffer); + } + } + + exit(0); +} diff --git a/progs/perf/swapbuffers.c b/progs/perf/swapbuffers.c new file mode 100644 index 00000000000..63c7fc06f98 --- /dev/null +++ b/progs/perf/swapbuffers.c @@ -0,0 +1,161 @@ +/* + * Copyright (C) 2009 VMware, Inc. 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, sublicense, + * 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 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 NONINFRINGEMENT. IN NO EVENT SHALL + * VMWARE 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. + */ + +/** + * Measure SwapBuffers. + * + * Keith Whitwell + * 22 Sep 2009 + */ + +#include "glmain.h" +#include "common.h" + + +int WinWidth = 100, WinHeight = 100; +int real_WinWidth, real_WinHeight; /* don't know whats going on here */ + +static GLuint VBO; + +struct vertex +{ + GLfloat x, y; +}; + +static const struct vertex vertices[4] = { + { -1.0, -1.0 }, + { 1.0, -1.0 }, + { 1.0, 1.0 }, + { -1.0, 1.0 } +}; + + +/** Called from test harness/main */ +void +PerfInit(void) +{ + /* setup VBO w/ vertex data */ + glGenBuffersARB(1, &VBO); + glBindBufferARB(GL_ARRAY_BUFFER_ARB, VBO); + glBufferDataARB(GL_ARRAY_BUFFER_ARB, + sizeof(vertices), vertices, GL_STATIC_DRAW_ARB); + glVertexPointer(2, GL_FLOAT, sizeof(struct vertex), (void *) 0); + glEnableClientState(GL_VERTEX_ARRAY); + + /* misc GL state */ + glAlphaFunc(GL_ALWAYS, 0.0); +} + +static void +SwapNaked(unsigned count) +{ + unsigned i; + for (i = 0; i < count; i++) { + PerfSwapBuffers(); + } +} + + +static void +SwapClear(unsigned count) +{ + unsigned i; + for (i = 0; i < count; i++) { + glClear(GL_COLOR_BUFFER_BIT); + PerfSwapBuffers(); + } +} + +static void +SwapClearPoint(unsigned count) +{ + unsigned i; + for (i = 0; i < count; i++) { + glClear(GL_COLOR_BUFFER_BIT); + glDrawArrays(GL_POINTS, 0, 4); + PerfSwapBuffers(); + } +} + + +static const struct { + unsigned w; + unsigned h; +} sizes[] = { + { 320, 240 }, + { 640, 480 }, + { 1024, 768 }, + { 1200, 1024 }, + { 1600, 1200 } +}; + +void +PerfNextRound(void) +{ + static unsigned i; + + if (i < sizeof(sizes) / sizeof(sizes[0]) && + PerfReshapeWindow( sizes[i].w, sizes[i].h )) + { + perf_printf("Reshape %dx%d\n", sizes[i].w, sizes[i].h); + real_WinWidth = sizes[i].w; + real_WinHeight = sizes[i].h; + i++; + } + else { + exit(0); + } +} + + + + +/** Called from test harness/main */ +void +PerfDraw(void) +{ + double rate0; + + rate0 = PerfMeasureRate(SwapNaked); + perf_printf(" Swapbuffers %dx%d: %s swaps/second", + real_WinWidth, real_WinHeight, + PerfHumanFloat(rate0)); + perf_printf(" %s pixels/second\n", + PerfHumanFloat(rate0 * real_WinWidth * real_WinHeight)); + + + + rate0 = PerfMeasureRate(SwapClear); + perf_printf(" Swap/Clear %dx%d: %s swaps/second", + real_WinWidth, real_WinHeight, + PerfHumanFloat(rate0)); + perf_printf(" %s pixels/second\n", + PerfHumanFloat(rate0 * real_WinWidth * real_WinHeight)); + + + rate0 = PerfMeasureRate(SwapClearPoint); + perf_printf(" Swap/Clear/Draw %dx%d: %s swaps/second", + real_WinWidth, real_WinHeight, + PerfHumanFloat(rate0)); + perf_printf(" %s pixels/second\n", + PerfHumanFloat(rate0 * real_WinWidth * real_WinHeight)); +} + diff --git a/progs/perf/teximage.c b/progs/perf/teximage.c new file mode 100644 index 00000000000..a3005d0befd --- /dev/null +++ b/progs/perf/teximage.c @@ -0,0 +1,331 @@ +/* + * Copyright (C) 2009 VMware, Inc. 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, sublicense, + * 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 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 NONINFRINGEMENT. IN NO EVENT SHALL + * VMWARE 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. + */ + +/** + * Measure glTex[Sub]Image2D() and glGetTexImage() rate + * + * Brian Paul + * 16 Sep 2009 + */ + +#include "glmain.h" +#include "common.h" + + +int WinWidth = 100, WinHeight = 100; + +static GLuint VBO; +static GLuint TexObj = 0; +static GLubyte *TexImage = NULL; +static GLsizei TexSize; +static GLenum TexIntFormat, TexSrcFormat, TexSrcType; + +static const GLboolean DrawPoint = GL_TRUE; +static const GLboolean TexSubImage4 = GL_FALSE; + +enum { + MODE_CREATE_TEXIMAGE, + MODE_TEXIMAGE, + MODE_TEXSUBIMAGE, + MODE_GETTEXIMAGE, + MODE_COUNT +}; + +static const char *mode_name[MODE_COUNT] = +{ + "Create_TexImage", + "TexImage", + "TexSubImage", + "GetTexImage" +}; + + + +struct vertex +{ + GLfloat x, y, s, t; +}; + +static const struct vertex vertices[1] = { + { 0.0, 0.0, 0.5, 0.5 }, +}; + +#define VOFFSET(F) ((void *) offsetof(struct vertex, F)) + + +/** Called from test harness/main */ +void +PerfInit(void) +{ + /* setup VBO w/ vertex data */ + glGenBuffersARB(1, &VBO); + glBindBufferARB(GL_ARRAY_BUFFER_ARB, VBO); + glBufferDataARB(GL_ARRAY_BUFFER_ARB, + sizeof(vertices), vertices, GL_STATIC_DRAW_ARB); + glVertexPointer(2, GL_FLOAT, sizeof(struct vertex), VOFFSET(x)); + glTexCoordPointer(2, GL_FLOAT, sizeof(struct vertex), VOFFSET(s)); + glEnableClientState(GL_VERTEX_ARRAY); + glEnableClientState(GL_TEXTURE_COORD_ARRAY); + + /* texture */ + glGenTextures(1, &TexObj); + glBindTexture(GL_TEXTURE_2D, TexObj); + glTexParameteri(GL_TEXTURE_2D, GL_TEXTURE_MIN_FILTER, GL_NEAREST); + glTexParameteri(GL_TEXTURE_2D, GL_TEXTURE_MAG_FILTER, GL_NEAREST); + glEnable(GL_TEXTURE_2D); +} + + + + +static void +CreateUploadTexImage2D(unsigned count) +{ + unsigned i; + for (i = 0; i < count; i++) { + if (TexObj) + glDeleteTextures(1, &TexObj); + + glGenTextures(1, &TexObj); + glBindTexture(GL_TEXTURE_2D, TexObj); + glTexParameteri(GL_TEXTURE_2D, GL_TEXTURE_MIN_FILTER, GL_NEAREST); + glTexParameteri(GL_TEXTURE_2D, GL_TEXTURE_MAG_FILTER, GL_NEAREST); + + glTexImage2D(GL_TEXTURE_2D, 0, TexIntFormat, + TexSize, TexSize, 0, + TexSrcFormat, TexSrcType, TexImage); + + if (DrawPoint) + glDrawArrays(GL_POINTS, 0, 1); + } + glFinish(); +} + + +static void +UploadTexImage2D(unsigned count) +{ + unsigned i; + for (i = 0; i < count; i++) { + /* XXX is this equivalent to a glTexSubImage call since we're + * always specifying the same image size? That case isn't optimized + * in Mesa but may be optimized in other drivers. Note sure how + * much difference that might make. + */ + glTexImage2D(GL_TEXTURE_2D, 0, TexIntFormat, + TexSize, TexSize, 0, + TexSrcFormat, TexSrcType, TexImage); + if (DrawPoint) + glDrawArrays(GL_POINTS, 0, 1); + } + glFinish(); +} + + +static void +UploadTexSubImage2D(unsigned count) +{ + unsigned i; + for (i = 0; i < count; i++) { + if (TexSubImage4) { + GLsizei halfSize = (TexSize == 1) ? 1 : TexSize / 2; + GLsizei halfPos = TexSize - halfSize; + /* do glTexSubImage2D in four pieces */ + /* lower-left */ + glPixelStorei(GL_UNPACK_ROW_LENGTH, TexSize); + glTexSubImage2D(GL_TEXTURE_2D, 0, + 0, 0, halfSize, halfSize, + TexSrcFormat, TexSrcType, TexImage); + /* lower-right */ + glPixelStorei(GL_UNPACK_SKIP_PIXELS, halfPos); + glTexSubImage2D(GL_TEXTURE_2D, 0, + halfPos, 0, halfSize, halfSize, + TexSrcFormat, TexSrcType, TexImage); + /* upper-left */ + glPixelStorei(GL_UNPACK_SKIP_PIXELS, 0); + glPixelStorei(GL_UNPACK_SKIP_ROWS, halfPos); + glTexSubImage2D(GL_TEXTURE_2D, 0, + 0, halfPos, halfSize, halfSize, + TexSrcFormat, TexSrcType, TexImage); + /* upper-right */ + glPixelStorei(GL_UNPACK_SKIP_PIXELS, halfPos); + glPixelStorei(GL_UNPACK_SKIP_ROWS, halfPos); + glTexSubImage2D(GL_TEXTURE_2D, 0, + halfPos, halfPos, halfSize, halfSize, + TexSrcFormat, TexSrcType, TexImage); + /* reset the unpacking state */ + glPixelStorei(GL_UNPACK_SKIP_PIXELS, 0); + glPixelStorei(GL_UNPACK_SKIP_ROWS, 0); + glPixelStorei(GL_UNPACK_ROW_LENGTH, 0); + } + else { + /* replace whole texture image at once */ + glTexSubImage2D(GL_TEXTURE_2D, 0, + 0, 0, TexSize, TexSize, + TexSrcFormat, TexSrcType, TexImage); + } + if (DrawPoint) + glDrawArrays(GL_POINTS, 0, 1); + } + glFinish(); +} + + +static void +GetTexImage2D(unsigned count) +{ + unsigned i; + GLubyte *buf = (GLubyte *) malloc(TexSize * TexSize * 4); + for (i = 0; i < count; i++) { + glGetTexImage(GL_TEXTURE_2D, 0, + TexSrcFormat, TexSrcType, buf); + } + glFinish(); + free(buf); +} + + +/* XXX any other formats to measure? */ +static const struct { + GLenum format, type; + GLenum internal_format; + const char *name; + GLuint texel_size; + GLboolean full_test; +} SrcFormats[] = { + { GL_RGBA, GL_UNSIGNED_BYTE, GL_RGBA, "RGBA/ubyte", 4, GL_TRUE }, + { GL_RGB, GL_UNSIGNED_BYTE, GL_RGB, "RGB/ubyte", 3, GL_FALSE }, + { GL_RGB, GL_UNSIGNED_SHORT_5_6_5, GL_RGB, "RGB/565", 2, GL_FALSE }, + { GL_BGRA, GL_UNSIGNED_BYTE, GL_RGBA, "BGRA/ubyte", 4, GL_FALSE }, + { GL_LUMINANCE, GL_UNSIGNED_BYTE, GL_LUMINANCE, "L/ubyte", 1, GL_FALSE }, + { 0, 0, 0, NULL, 0, 0 } +}; + + +/** Called from test harness/main */ +void +PerfNextRound(void) +{ +} + + +/** Called from test harness/main */ +void +PerfDraw(void) +{ + GLint maxSize; + double rate; + GLint fmt, mode; + + glGetIntegerv(GL_MAX_TEXTURE_SIZE, &maxSize); + + /* loop over source data formats */ + for (fmt = 0; SrcFormats[fmt].format; fmt++) { + TexIntFormat = SrcFormats[fmt].internal_format; + TexSrcFormat = SrcFormats[fmt].format; + TexSrcType = SrcFormats[fmt].type; + + /* loop over glTexImage, glTexSubImage */ + for (mode = 0; mode < MODE_COUNT; mode++) { + GLuint minsz, maxsz; + + if (SrcFormats[fmt].full_test) { + minsz = 16; + maxsz = 4096; + } + else { + minsz = maxsz = 256; + if (mode == MODE_CREATE_TEXIMAGE) + continue; + } + + /* loop over a defined range of texture sizes, test only the + * ones which are legal for this driver. + */ + for (TexSize = minsz; TexSize <= maxsz; TexSize *= 4) { + double mbPerSec; + + if (TexSize <= maxSize) { + GLint bytesPerImage; + + bytesPerImage = TexSize * TexSize * SrcFormats[fmt].texel_size; + TexImage = malloc(bytesPerImage); + + switch (mode) { + case MODE_TEXIMAGE: + rate = PerfMeasureRate(UploadTexImage2D); + break; + + case MODE_CREATE_TEXIMAGE: + rate = PerfMeasureRate(CreateUploadTexImage2D); + break; + + case MODE_TEXSUBIMAGE: + /* create initial, empty texture */ + glTexImage2D(GL_TEXTURE_2D, 0, TexIntFormat, + TexSize, TexSize, 0, + TexSrcFormat, TexSrcType, NULL); + rate = PerfMeasureRate(UploadTexSubImage2D); + break; + + case MODE_GETTEXIMAGE: + glTexImage2D(GL_TEXTURE_2D, 0, TexIntFormat, + TexSize, TexSize, 0, + TexSrcFormat, TexSrcType, TexImage); + rate = PerfMeasureRate(GetTexImage2D); + break; + + default: + exit(1); + } + + mbPerSec = rate * bytesPerImage / (1024.0 * 1024.0); + free(TexImage); + + + { + unsigned err; + err = glGetError(); + if (err) { + perf_printf("non-zero glGetError() %d\n", err); + exit(1); + } + } + + } + else { + rate = 0; + mbPerSec = 0; + } + + perf_printf(" %s(%s %d x %d): " + "%.1f images/sec, %.1f MB/sec\n", + mode_name[mode], + SrcFormats[fmt].name, TexSize, TexSize, rate, mbPerSec); + } + + if (SrcFormats[fmt].full_test) + perf_printf("\n"); + } + } + + exit(0); +} diff --git a/progs/perf/vbo.c b/progs/perf/vbo.c new file mode 100644 index 00000000000..b326c056ec6 --- /dev/null +++ b/progs/perf/vbo.c @@ -0,0 +1,246 @@ +/* + * Copyright (C) 2009 VMware, Inc. 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, sublicense, + * 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 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 NONINFRINGEMENT. IN NO EVENT SHALL + * VMWARE 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. + */ + +/** + * Measure VBO upload speed. + * That is, measure glBufferDataARB() and glBufferSubDataARB(). + * + * Brian Paul + * 16 Sep 2009 + */ + +#include <string.h> +#include "glmain.h" +#include "common.h" + +/* Copy data out of a large array to avoid caching effects: + */ +#define DATA_SIZE (16*1024*1024) + +int WinWidth = 100, WinHeight = 100; + +static GLuint VBO; + +static GLsizei VBOSize = 0; +static GLsizei SubSize = 0; +static GLubyte *VBOData = NULL; /* array[DATA_SIZE] */ + +static const GLboolean DrawPoint = GL_TRUE; +static const GLboolean BufferSubDataInHalves = GL_TRUE; + +static const GLfloat Vertex0[2] = { 0.0, 0.0 }; + + +/** Called from test harness/main */ +void +PerfInit(void) +{ + /* setup VBO */ + glGenBuffersARB(1, &VBO); + glBindBufferARB(GL_ARRAY_BUFFER_ARB, VBO); + glVertexPointer(2, GL_FLOAT, sizeof(Vertex0), (void *) 0); + glEnableClientState(GL_VERTEX_ARRAY); +} + + +static void +UploadVBO(unsigned count) +{ + unsigned i; + unsigned total = 0; + unsigned src = 0; + + for (i = 0; i < count; i++) { + glBufferDataARB(GL_ARRAY_BUFFER, VBOSize, VBOData + src, GL_STREAM_DRAW_ARB); + glDrawArrays(GL_POINTS, 0, 1); + + /* Throw in an occasional flush to work around a driver crash: + */ + total += VBOSize; + if (total >= 16*1024*1024) { + glFlush(); + total = 0; + } + + src += VBOSize; + src %= DATA_SIZE; + } + glFinish(); +} + + +static void +UploadSubVBO(unsigned count) +{ + unsigned i; + unsigned src = 0; + + for (i = 0; i < count; i++) { + unsigned offset = (i * SubSize) % VBOSize; + glBufferSubDataARB(GL_ARRAY_BUFFER, offset, SubSize, VBOData + src); + + if (DrawPoint) { + glDrawArrays(GL_POINTS, offset / sizeof(Vertex0), 1); + } + + src += SubSize; + src %= DATA_SIZE; + } + glFinish(); +} + + +/* Do multiple small SubData uploads, then call DrawArrays. This may be a + * fairer comparison to back-to-back BufferData calls: + */ +static void +BatchUploadSubVBO(unsigned count) +{ + unsigned i = 0, j; + unsigned period = VBOSize / SubSize; + unsigned src = 0; + + while (i < count) { + for (j = 0; j < period && i < count; j++, i++) { + unsigned offset = j * SubSize; + glBufferSubDataARB(GL_ARRAY_BUFFER, offset, SubSize, VBOData + src); + } + + glDrawArrays(GL_POINTS, 0, 1); + + src += SubSize; + src %= DATA_SIZE; + } + glFinish(); +} + + +/** + * Test the sequence: + * create/load VBO + * draw + * destroy VBO + */ +static void +CreateDrawDestroyVBO(unsigned count) +{ + unsigned i; + for (i = 0; i < count; i++) { + GLuint vbo; + /* create/load */ + glGenBuffersARB(1, &vbo); + glBindBufferARB(GL_ARRAY_BUFFER_ARB, vbo); + glBufferDataARB(GL_ARRAY_BUFFER, VBOSize, VBOData, GL_STREAM_DRAW_ARB); + /* draw */ + glVertexPointer(2, GL_FLOAT, sizeof(Vertex0), (void *) 0); + glDrawArrays(GL_POINTS, 0, 1); + /* destroy */ + glDeleteBuffersARB(1, &vbo); + } + glFinish(); +} + + +static const GLsizei Sizes[] = { + 64, + 1024, + 16*1024, + 256*1024, + 1024*1024, + 16*1024*1024, + 0 /* end of list */ +}; + +void +PerfNextRound(void) +{ +} + +/** Called from test harness/main */ +void +PerfDraw(void) +{ + double rate, mbPerSec; + int i, sz; + + /* Load VBOData buffer with duplicated Vertex0. + */ + VBOData = calloc(DATA_SIZE, 1); + + for (i = 0; i < DATA_SIZE / sizeof(Vertex0); i++) { + memcpy(VBOData + i * sizeof(Vertex0), + Vertex0, + sizeof(Vertex0)); + } + + /* glBufferDataARB() + */ + for (sz = 0; Sizes[sz]; sz++) { + SubSize = VBOSize = Sizes[sz]; + rate = PerfMeasureRate(UploadVBO); + mbPerSec = rate * VBOSize / (1024.0 * 1024.0); + perf_printf(" glBufferDataARB(size = %d): %.1f MB/sec\n", + VBOSize, mbPerSec); + } + + /* glBufferSubDataARB() + */ + for (sz = 0; Sizes[sz]; sz++) { + SubSize = VBOSize = Sizes[sz]; + rate = PerfMeasureRate(UploadSubVBO); + mbPerSec = rate * VBOSize / (1024.0 * 1024.0); + perf_printf(" glBufferSubDataARB(size = %d): %.1f MB/sec\n", + VBOSize, mbPerSec); + } + + /* Batch upload + */ + VBOSize = 1024 * 1024; + glBufferDataARB(GL_ARRAY_BUFFER, VBOSize, VBOData, GL_STREAM_DRAW_ARB); + + for (sz = 0; Sizes[sz] < VBOSize; sz++) { + SubSize = Sizes[sz]; + rate = PerfMeasureRate(UploadSubVBO); + mbPerSec = rate * SubSize / (1024.0 * 1024.0); + perf_printf(" glBufferSubDataARB(size = %d, VBOSize = %d): %.1f MB/sec\n", + SubSize, VBOSize, mbPerSec); + } + + for (sz = 0; Sizes[sz] < VBOSize; sz++) { + SubSize = Sizes[sz]; + rate = PerfMeasureRate(BatchUploadSubVBO); + mbPerSec = rate * SubSize / (1024.0 * 1024.0); + perf_printf(" glBufferSubDataARB(size = %d, VBOSize = %d), batched: %.1f MB/sec\n", + SubSize, VBOSize, mbPerSec); + } + + /* Create/Draw/Destroy + */ + for (sz = 0; Sizes[sz]; sz++) { + SubSize = VBOSize = Sizes[sz]; + rate = PerfMeasureRate(CreateDrawDestroyVBO); + mbPerSec = rate * VBOSize / (1024.0 * 1024.0); + perf_printf(" VBO Create/Draw/Destroy(size = %d): %.1f MB/sec, %.1f draws/sec\n", + VBOSize, mbPerSec, rate); + } + + exit(0); +} diff --git a/progs/perf/vertexrate.c b/progs/perf/vertexrate.c new file mode 100644 index 00000000000..b5355525d06 --- /dev/null +++ b/progs/perf/vertexrate.c @@ -0,0 +1,276 @@ +/* + * Copyright (C) 2009 VMware, Inc. 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, sublicense, + * 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 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 NONINFRINGEMENT. IN NO EVENT SHALL + * VMWARE 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. + */ + +/** + * Measure simple vertex processing rate via: + * - immediate mode + * - vertex arrays + * - VBO vertex arrays + * - glDrawElements + * - VBO glDrawElements + * - glDrawRangeElements + * - VBO glDrawRangeElements + * + * Brian Paul + * 16 Sep 2009 + */ + +#include <assert.h> +#include <string.h> +#include "glmain.h" +#include "common.h" + + +#define MAX_VERTS (100 * 100) + +/** glVertex2/3/4 size */ +#define VERT_SIZE 4 + +int WinWidth = 500, WinHeight = 500; + +static GLuint VertexBO, ElementBO; + +static unsigned NumVerts = MAX_VERTS; +static unsigned VertBytes = VERT_SIZE * sizeof(float); +static float *VertexData = NULL; + +static unsigned NumElements = MAX_VERTS; +static GLuint *Elements = NULL; + + +/** + * Load VertexData buffer with a 2-D grid of points in the range [-1,1]^2. + */ +static void +InitializeVertexData(void) +{ + unsigned i; + float x = -1.0, y = -1.0; + float dx = 2.0 / 100; + float dy = 2.0 / 100; + + VertexData = (float *) malloc(NumVerts * VertBytes); + + for (i = 0; i < NumVerts; i++) { + VertexData[i * VERT_SIZE + 0] = x; + VertexData[i * VERT_SIZE + 1] = y; + VertexData[i * VERT_SIZE + 2] = 0.0; + VertexData[i * VERT_SIZE + 3] = 1.0; + x += dx; + if (x > 1.0) { + x = -1.0; + y += dy; + } + } + + Elements = (GLuint *) malloc(NumVerts * sizeof(GLuint)); + + for (i = 0; i < NumVerts; i++) { + Elements[i] = NumVerts - i - 1; + } +} + + +/** Called from test harness/main */ +void +PerfInit(void) +{ + InitializeVertexData(); + + /* setup VertexBO */ + glGenBuffersARB(1, &VertexBO); + glBindBufferARB(GL_ARRAY_BUFFER_ARB, VertexBO); + glBufferDataARB(GL_ARRAY_BUFFER_ARB, + NumVerts * VertBytes, VertexData, GL_STATIC_DRAW_ARB); + glEnableClientState(GL_VERTEX_ARRAY); + + /* setup ElementBO */ + glGenBuffersARB(1, &ElementBO); + glBindBufferARB(GL_ELEMENT_ARRAY_BUFFER_ARB, ElementBO); + glBufferDataARB(GL_ELEMENT_ARRAY_BUFFER_ARB, + NumElements * sizeof(GLuint), Elements, GL_STATIC_DRAW_ARB); +} + + +static void +DrawImmediate(unsigned count) +{ + unsigned i; + glBindBufferARB(GL_ELEMENT_ARRAY_BUFFER, 0); + glBindBufferARB(GL_ARRAY_BUFFER, 0); + for (i = 0; i < count; i++) { + unsigned j; + glBegin(GL_POINTS); + for (j = 0; j < NumVerts; j++) { +#if VERT_SIZE == 4 + glVertex4fv(VertexData + j * 4); +#elif VERT_SIZE == 3 + glVertex3fv(VertexData + j * 3); +#elif VERT_SIZE == 2 + glVertex2fv(VertexData + j * 2); +#else + abort(); +#endif + } + glEnd(); + } + glFinish(); + PerfSwapBuffers(); +} + + +static void +DrawArraysMem(unsigned count) +{ + unsigned i; + glBindBufferARB(GL_ELEMENT_ARRAY_BUFFER, 0); + glBindBufferARB(GL_ARRAY_BUFFER, 0); + glVertexPointer(VERT_SIZE, GL_FLOAT, VertBytes, VertexData); + for (i = 0; i < count; i++) { + glDrawArrays(GL_POINTS, 0, NumVerts); + } + glFinish(); + PerfSwapBuffers(); +} + + +static void +DrawArraysVBO(unsigned count) +{ + unsigned i; + glBindBufferARB(GL_ELEMENT_ARRAY_BUFFER, 0); + glBindBufferARB(GL_ARRAY_BUFFER, VertexBO); + glVertexPointer(VERT_SIZE, GL_FLOAT, VertBytes, (void *) 0); + for (i = 0; i < count; i++) { + glDrawArrays(GL_POINTS, 0, NumVerts); + } + glFinish(); + PerfSwapBuffers(); +} + + +static void +DrawElementsMem(unsigned count) +{ + unsigned i; + glBindBufferARB(GL_ELEMENT_ARRAY_BUFFER, 0); + glBindBufferARB(GL_ARRAY_BUFFER, 0); + glVertexPointer(VERT_SIZE, GL_FLOAT, VertBytes, VertexData); + for (i = 0; i < count; i++) { + glDrawElements(GL_POINTS, NumVerts, GL_UNSIGNED_INT, Elements); + } + glFinish(); + PerfSwapBuffers(); +} + + +static void +DrawElementsBO(unsigned count) +{ + unsigned i; + glBindBufferARB(GL_ELEMENT_ARRAY_BUFFER, ElementBO); + glBindBufferARB(GL_ARRAY_BUFFER, VertexBO); + glVertexPointer(VERT_SIZE, GL_FLOAT, VertBytes, (void *) 0); + for (i = 0; i < count; i++) { + glDrawElements(GL_POINTS, NumVerts, GL_UNSIGNED_INT, (void *) 0); + } + glFinish(); + PerfSwapBuffers(); +} + + +static void +DrawRangeElementsMem(unsigned count) +{ + unsigned i; + glBindBufferARB(GL_ELEMENT_ARRAY_BUFFER, 0); + glBindBufferARB(GL_ARRAY_BUFFER, 0); + glVertexPointer(VERT_SIZE, GL_FLOAT, VertBytes, VertexData); + for (i = 0; i < count; i++) { + glDrawRangeElements(GL_POINTS, 0, NumVerts - 1, + NumVerts, GL_UNSIGNED_INT, Elements); + } + glFinish(); + PerfSwapBuffers(); +} + + +static void +DrawRangeElementsBO(unsigned count) +{ + unsigned i; + glBindBufferARB(GL_ELEMENT_ARRAY_BUFFER, ElementBO); + glBindBufferARB(GL_ARRAY_BUFFER, VertexBO); + glVertexPointer(VERT_SIZE, GL_FLOAT, VertBytes, (void *) 0); + for (i = 0; i < count; i++) { + glDrawRangeElements(GL_POINTS, 0, NumVerts - 1, + NumVerts, GL_UNSIGNED_INT, (void *) 0); + } + glFinish(); + PerfSwapBuffers(); +} + +void +PerfNextRound(void) +{ +} + + +/** Called from test harness/main */ +void +PerfDraw(void) +{ + double rate; + + glClear(GL_COLOR_BUFFER_BIT | GL_DEPTH_BUFFER_BIT); + + perf_printf("Vertex rate (%d x Vertex%df)\n", NumVerts, VERT_SIZE); + + rate = PerfMeasureRate(DrawImmediate); + rate *= NumVerts; + perf_printf(" Immediate mode: %s verts/sec\n", PerfHumanFloat(rate)); + + rate = PerfMeasureRate(DrawArraysMem); + rate *= NumVerts; + perf_printf(" glDrawArrays: %s verts/sec\n", PerfHumanFloat(rate)); + + rate = PerfMeasureRate(DrawArraysVBO); + rate *= NumVerts; + perf_printf(" VBO glDrawArrays: %s verts/sec\n", PerfHumanFloat(rate)); + + rate = PerfMeasureRate(DrawElementsMem); + rate *= NumVerts; + perf_printf(" glDrawElements: %s verts/sec\n", PerfHumanFloat(rate)); + + rate = PerfMeasureRate(DrawElementsBO); + rate *= NumVerts; + perf_printf(" VBO glDrawElements: %s verts/sec\n", PerfHumanFloat(rate)); + + rate = PerfMeasureRate(DrawRangeElementsMem); + rate *= NumVerts; + perf_printf(" glDrawRangeElements: %s verts/sec\n", PerfHumanFloat(rate)); + + rate = PerfMeasureRate(DrawRangeElementsBO); + rate *= NumVerts; + perf_printf(" VBO glDrawRangeElements: %s verts/sec\n", PerfHumanFloat(rate)); + + exit(0); +} diff --git a/progs/tests/zreaddraw.c b/progs/tests/zreaddraw.c index 0821d5fb357..7dfed20cfb9 100644 --- a/progs/tests/zreaddraw.c +++ b/progs/tests/zreaddraw.c @@ -13,14 +13,16 @@ static GLint WinWidth = 500, WinHeight = 500; static GLboolean Invert = GL_FALSE; +static GLboolean TestPacking = GL_FALSE; +static GLboolean TestList = GL_FALSE; static void Display(void) { - GLfloat depth[100 * 100]; - GLfloat depth2[400 * 400]; - GLfloat min, max; - int i; + GLfloat depth[100 * 100 * 2]; + GLfloat depth2[400 * 400]; /* *2 to test pixelstore stuff */ + GLuint list; + GLenum depthType = GL_FLOAT; glClearColor(0.5, 0.5, 0.5, 1.0); glClear(GL_COLOR_BUFFER_BIT | GL_DEPTH_BUFFER_BIT); @@ -36,16 +38,32 @@ static void Display(void) glLoadIdentity(); glutSolidSphere(1.0, 20, 10); + if (TestPacking) { + glPixelStorei(GL_PACK_ROW_LENGTH, 120); + glPixelStorei(GL_PACK_SKIP_PIXELS, 5); + } + /* read the depth image */ - glReadPixels(0, 0, 100, 100, GL_DEPTH_COMPONENT, GL_FLOAT, depth); - min = max = depth[0]; - for (i = 1; i < 100 * 100; i++) { - if (depth[i] < min) - min = depth[i]; - if (depth[i] > max) - max = depth[i]; + glReadPixels(0, 0, 100, 100, GL_DEPTH_COMPONENT, depthType, depth); + if (depthType == GL_FLOAT) { + GLfloat min, max; + int i; + min = max = depth[0]; + for (i = 1; i < 100 * 100; i++) { + if (depth[i] < min) + min = depth[i]; + if (depth[i] > max) + max = depth[i]; + } + printf("Depth value range: [%f, %f]\n", min, max); + } + + if (TestPacking) { + glPixelStorei(GL_PACK_ROW_LENGTH, 0); + glPixelStorei(GL_PACK_SKIP_PIXELS, 0); + glPixelStorei(GL_UNPACK_ROW_LENGTH, 120); + glPixelStorei(GL_UNPACK_SKIP_PIXELS, 5); } - printf("Depth value range: [%f, %f]\n", min, max); /* draw depth image with scaling (into z buffer) */ glPixelZoom(4.0, 4.0); @@ -55,12 +73,27 @@ static void Display(void) glPixelTransferf(GL_DEPTH_SCALE, -1.0); glPixelTransferf(GL_DEPTH_BIAS, 1.0); } - glDrawPixels(100, 100, GL_DEPTH_COMPONENT, GL_FLOAT, depth); + if (TestList) { + list = glGenLists(1); + glNewList(list, GL_COMPILE); + glDrawPixels(100, 100, GL_DEPTH_COMPONENT, depthType, depth); + glEndList(); + glCallList(list); + glDeleteLists(list, 1); + } + else { + glDrawPixels(100, 100, GL_DEPTH_COMPONENT, depthType, depth); + } if (Invert) { glPixelTransferf(GL_DEPTH_SCALE, 1.0); glPixelTransferf(GL_DEPTH_BIAS, 0.0); } + if (TestPacking) { + glPixelStorei(GL_UNPACK_ROW_LENGTH, 0); + glPixelStorei(GL_UNPACK_SKIP_PIXELS, 0); + } + glDisable(GL_DEPTH_TEST); /* read back scaled depth image */ @@ -89,6 +122,14 @@ static void Key(unsigned char key, int x, int y) case 'i': Invert = !Invert; break; + case 'p': + TestPacking = !TestPacking; + printf("Test pixel pack/unpack: %d\n", TestPacking); + break; + case 'l': + TestList = !TestList; + printf("Test dlist: %d\n", TestList); + break; case 27: exit(0); break; diff --git a/progs/vp/vp-tris.c b/progs/vp/vp-tris.c index 97995accdd1..1356242d971 100644 --- a/progs/vp/vp-tris.c +++ b/progs/vp/vp-tris.c @@ -119,6 +119,12 @@ static void Init( void ) glBindProgramARB(GL_VERTEX_PROGRAM_ARB, prognum); glProgramStringARB(GL_VERTEX_PROGRAM_ARB, GL_PROGRAM_FORMAT_ASCII_ARB, sz, (const GLubyte *) buf); + if (glGetError()) { + printf("Program failed to compile:\n%s\n", buf); + printf("Error: %s\n", + (char *) glGetString(GL_PROGRAM_ERROR_STRING_ARB)); + exit(1); + } assert(glIsProgramARB(prognum)); } diff --git a/scons/crossmingw.py b/scons/crossmingw.py index 3aed4843502..9cb0b89e3c7 100644 --- a/scons/crossmingw.py +++ b/scons/crossmingw.py @@ -177,7 +177,7 @@ def generate(env): # default in recent gcc versions env.AppendUnique(CFLAGS = ['-gstabs']) - env.AppendUnique(LIBS = ['iberty']) + #env.AppendUnique(LIBS = ['iberty']) env.AppendUnique(SHLINKFLAGS = ['-Wl,--enable-stdcall-fixup']) #env.AppendUnique(SHLINKFLAGS = ['-Wl,--kill-at']) diff --git a/scons/dxsdk.py b/scons/dxsdk.py index de090e4f991..920cc2f689d 100644 --- a/scons/dxsdk.py +++ b/scons/dxsdk.py @@ -52,11 +52,20 @@ def generate(env): target_cpu = 'x64' else: raise SCons.Errors.InternalError, "Unsupported target machine" - include_dir = 'Include' + + include_dir = os.path.join(dxsdk_root, 'Include') + lib_dir = os.path.join(dxsdk_root, 'Lib', target_cpu) env.Append(CPPDEFINES = [('HAVE_DXSDK', '1')]) - env.Prepend(CPPPATH = [os.path.join(dxsdk_root, 'Include')]) - env.Prepend(LIBPATH = [os.path.join(dxsdk_root, 'Lib', target_cpu)]) + + gcc = 'gcc' in os.path.basename(env['CC']).split('-') + if gcc: + # Make GCC more forgiving towards Microsoft's headers + env.Prepend(CPPFLAGS = ['-isystem', include_dir]) + else: + env.Prepend(CPPPATH = [include_dir]) + + env.Prepend(LIBPATH = [lib_dir]) def exists(env): return get_dxsdk_root(env) is not None diff --git a/scons/gallium.py b/scons/gallium.py index 47b07744be9..38782ac7266 100644 --- a/scons/gallium.py +++ b/scons/gallium.py @@ -334,7 +334,11 @@ def generate(env): else: ccflags += ['-O3', '-g3'] if env['profile']: - ccflags += ['-pg'] + # See http://code.google.com/p/jrfonseca/wiki/Gprof2Dot#Which_options_should_I_pass_to_gcc_when_compiling_for_profiling? + ccflags += [ + '-fno-omit-frame-pointer', + '-fno-optimize-sibling-calls', + ] if env['machine'] == 'x86': ccflags += [ '-m32', diff --git a/scons/llvm.py b/scons/llvm.py index 46a8d829ca4..d3293bb404f 100644 --- a/scons/llvm.py +++ b/scons/llvm.py @@ -51,7 +51,9 @@ def generate(env): llvm_bin_dir = os.path.join(llvm_dir, llvm_subdir, 'bin') if not os.path.isdir(llvm_bin_dir): - raise SCons.Errors.InternalError, "LLVM build directory not found" + llvm_bin_dir = os.path.join(llvm_dir, 'bin') + if not os.path.isdir(llvm_bin_dir): + raise SCons.Errors.InternalError, "LLVM binary directory not found" env.PrependENVPath('PATH', llvm_bin_dir) @@ -65,6 +67,8 @@ def generate(env): except OSError: print 'llvm-config version %s failed' % version else: + if env['platform'] == 'windows': + env.Append(LIBS = ['imagehlp', 'psapi']) env['LINK'] = env['CXX'] env['LLVM_VERSION'] = version diff --git a/scons/winddk.py b/scons/winddk.py index afcea9909a6..4dac16ee66e 100644 --- a/scons/winddk.py +++ b/scons/winddk.py @@ -85,8 +85,6 @@ def get_winddk_paths(env, version, root): else: # TODO: take in consideration the host cpu bin_dir = os.path.join(root, 'bin', 'win64', 'x86', cpu_bin(target_cpu)) - - env.PrependENVPath('PATH', [bin_dir]) crt_inc_dir = os.path.join(root, 'inc', 'crt') if version_major >= 6000: @@ -98,17 +96,33 @@ def get_winddk_paths(env, version, root): sdk_inc_dir = os.path.join(root, 'inc', target_os) wdm_inc_dir = os.path.join(root, 'inc', 'ddk', 'wdm', target_os) - env.PrependENVPath('INCLUDE', [ - wdm_inc_dir, - ddk_inc_dir, - crt_inc_dir, - sdk_inc_dir, - ]) + if env['toolchain'] == 'winddk': + env.PrependENVPath('PATH', [bin_dir]) + env.PrependENVPath('INCLUDE', [ + wdm_inc_dir, + ddk_inc_dir, + crt_inc_dir, + sdk_inc_dir, + ]) + env.PrependENVPath('LIB', [ + os.path.join(root, 'lib', 'crt', target_cpu), + os.path.join(root, 'lib', target_os, target_cpu), + ]) + elif env['toolchain'] == 'crossmingw': + env.Prepend(CPPFLAGS = [ + '-isystem', ddk_inc_dir, + '-isystem', sdk_inc_dir, + ]) + else: + env.Prepend(CPPPATH = [ + wdm_inc_dir, + ddk_inc_dir, + sdk_inc_dir, + ]) + env.Prepend(LIBPATH = [ + os.path.join(root, 'lib', target_os, target_cpu), + ]) - env.PrependENVPath('LIB', [ - os.path.join(root, 'lib', 'crt', target_cpu), - os.path.join(root, 'lib', target_os, target_cpu), - ]) def generate(env): if not env.has_key('ENV'): @@ -120,9 +134,10 @@ def generate(env): get_winddk_paths(env, version, root) break - msvc_sa.generate(env) - mslib_sa.generate(env) - mslink_sa.generate(env) + if env['toolchain'] == 'winddk': + msvc_sa.generate(env) + mslib_sa.generate(env) + mslink_sa.generate(env) def exists(env): for version in versions: diff --git a/src/gallium/auxiliary/draw/draw_pt_post_vs.c b/src/gallium/auxiliary/draw/draw_pt_post_vs.c index 00d7197b132..e25f16c354a 100644 --- a/src/gallium/auxiliary/draw/draw_pt_post_vs.c +++ b/src/gallium/auxiliary/draw/draw_pt_post_vs.c @@ -104,7 +104,7 @@ static boolean post_vs_cliptest_viewport_gl( struct pt_post_vs *pvs, unsigned clipped = 0; unsigned j; - if (0) debug_printf("%s\n"); + if (0) debug_printf("%s\n", __FUNCTION__); for (j = 0; j < count; j++) { float *position = out->data[pos]; diff --git a/src/gallium/auxiliary/pipebuffer/pb_buffer_fenced.c b/src/gallium/auxiliary/pipebuffer/pb_buffer_fenced.c index 109ac7c9d63..d01f8666222 100644 --- a/src/gallium/auxiliary/pipebuffer/pb_buffer_fenced.c +++ b/src/gallium/auxiliary/pipebuffer/pb_buffer_fenced.c @@ -542,7 +542,7 @@ fenced_buffer_list_dump(struct fenced_buffer_list *fenced_list) debug_printf("%10p %7u %7u\n", fenced_buf, fenced_buf->base.base.size, - fenced_buf->base.base.reference.count); + p_atomic_read(&fenced_buf->base.base.reference.count)); curr = next; next = curr->next; } @@ -556,7 +556,7 @@ fenced_buffer_list_dump(struct fenced_buffer_list *fenced_list) debug_printf("%10p %7u %7u %10p %s\n", fenced_buf, fenced_buf->base.base.size, - fenced_buf->base.base.reference.count, + p_atomic_read(&fenced_buf->base.base.reference.count), fenced_buf->fence, signaled == 0 ? "y" : "n"); curr = next; diff --git a/src/gallium/auxiliary/pipebuffer/pb_bufmgr_debug.c b/src/gallium/auxiliary/pipebuffer/pb_bufmgr_debug.c index 1b4df28c707..6e3214ca9c9 100644 --- a/src/gallium/auxiliary/pipebuffer/pb_bufmgr_debug.c +++ b/src/gallium/auxiliary/pipebuffer/pb_bufmgr_debug.c @@ -350,7 +350,7 @@ pb_debug_manager_dump(struct pb_debug_manager *mgr) buf = LIST_ENTRY(struct pb_debug_buffer, curr, head); debug_printf("buffer = %p\n", buf); - debug_printf(" .size = %p\n", buf->base.base.size); + debug_printf(" .size = 0x%x\n", buf->base.base.size); debug_backtrace_dump(buf->create_backtrace, PB_DEBUG_CREATE_BACKTRACE); curr = next; diff --git a/src/gallium/auxiliary/tgsi/tgsi_sanity.c b/src/gallium/auxiliary/tgsi/tgsi_sanity.c index 8a13885da9b..53e13b30e63 100644 --- a/src/gallium/auxiliary/tgsi/tgsi_sanity.c +++ b/src/gallium/auxiliary/tgsi/tgsi_sanity.c @@ -358,7 +358,7 @@ epilog( boolean tgsi_sanity_check( - struct tgsi_token *tokens ) + const struct tgsi_token *tokens ) { struct sanity_check_ctx ctx; diff --git a/src/gallium/auxiliary/tgsi/tgsi_sanity.h b/src/gallium/auxiliary/tgsi/tgsi_sanity.h index ca45e94c7ad..52263ff8832 100644 --- a/src/gallium/auxiliary/tgsi/tgsi_sanity.h +++ b/src/gallium/auxiliary/tgsi/tgsi_sanity.h @@ -40,7 +40,7 @@ extern "C" { */ boolean tgsi_sanity_check( - struct tgsi_token *tokens ); + const struct tgsi_token *tokens ); #if defined __cplusplus } diff --git a/src/gallium/auxiliary/tgsi/tgsi_scan.c b/src/gallium/auxiliary/tgsi/tgsi_scan.c index c535788819f..0db4481a3d0 100644 --- a/src/gallium/auxiliary/tgsi/tgsi_scan.c +++ b/src/gallium/auxiliary/tgsi/tgsi_scan.c @@ -132,6 +132,7 @@ tgsi_scan_shader(const struct tgsi_token *tokens, if (file == TGSI_FILE_INPUT) { info->input_semantic_name[reg] = (ubyte)fulldecl->Semantic.SemanticName; info->input_semantic_index[reg] = (ubyte)fulldecl->Semantic.SemanticIndex; + info->input_interpolate[reg] = (ubyte)fulldecl->Declaration.Interpolate; info->num_inputs++; } else if (file == TGSI_FILE_OUTPUT) { diff --git a/src/gallium/auxiliary/tgsi/tgsi_scan.h b/src/gallium/auxiliary/tgsi/tgsi_scan.h index 2c1a75bc812..8a7ee0c7e4f 100644 --- a/src/gallium/auxiliary/tgsi/tgsi_scan.h +++ b/src/gallium/auxiliary/tgsi/tgsi_scan.h @@ -45,6 +45,7 @@ struct tgsi_shader_info ubyte num_outputs; ubyte input_semantic_name[PIPE_MAX_SHADER_INPUTS]; /**< TGSI_SEMANTIC_x */ ubyte input_semantic_index[PIPE_MAX_SHADER_INPUTS]; + ubyte input_interpolate[PIPE_MAX_SHADER_INPUTS]; ubyte output_semantic_name[PIPE_MAX_SHADER_OUTPUTS]; /**< TGSI_SEMANTIC_x */ ubyte output_semantic_index[PIPE_MAX_SHADER_OUTPUTS]; diff --git a/src/gallium/auxiliary/tgsi/tgsi_sse2.c b/src/gallium/auxiliary/tgsi/tgsi_sse2.c index 3cdf8b9f359..5f6b83b2362 100644 --- a/src/gallium/auxiliary/tgsi/tgsi_sse2.c +++ b/src/gallium/auxiliary/tgsi/tgsi_sse2.c @@ -39,8 +39,9 @@ #include "tgsi/tgsi_info.h" #include "tgsi/tgsi_parse.h" #include "tgsi/tgsi_util.h" -#include "tgsi_exec.h" -#include "tgsi_sse2.h" +#include "tgsi/tgsi_dump.h" +#include "tgsi/tgsi_exec.h" +#include "tgsi/tgsi_sse2.h" #include "rtasm/rtasm_x86sse.h" @@ -1360,6 +1361,32 @@ emit_store( const struct tgsi_full_instruction *inst, unsigned chan_index ) { + switch( inst->Instruction.Saturate ) { + case TGSI_SAT_NONE: + break; + + case TGSI_SAT_ZERO_ONE: + sse_maxps( + func, + make_xmm( xmm ), + get_temp( + TGSI_EXEC_TEMP_00000000_I, + TGSI_EXEC_TEMP_00000000_C ) ); + + sse_minps( + func, + make_xmm( xmm ), + get_temp( + TGSI_EXEC_TEMP_ONE_I, + TGSI_EXEC_TEMP_ONE_C ) ); + break; + + case TGSI_SAT_MINUS_PLUS_ONE: + assert( 0 ); + break; + } + + switch( reg->DstRegister.File ) { case TGSI_FILE_OUTPUT: emit_output( @@ -1388,19 +1415,6 @@ emit_store( default: assert( 0 ); } - - switch( inst->Instruction.Saturate ) { - case TGSI_SAT_NONE: - break; - - case TGSI_SAT_ZERO_ONE: - /* assert( 0 ); */ - break; - - case TGSI_SAT_MINUS_PLUS_ONE: - assert( 0 ); - break; - } } #define STORE( FUNC, INST, XMM, INDEX, CHAN )\ @@ -1431,11 +1445,11 @@ fetch_texel( struct tgsi_sampler **sampler, { float rgba[NUM_CHANNELS][QUAD_SIZE]; (*sampler)->get_samples(*sampler, - &store[0], - &store[4], - &store[8], - 0.0f, /*store[12], lodbias */ - rgba); + &store[0], /* s */ + &store[4], /* t */ + &store[8], /* r */ + store[12], /* lodbias */ + rgba); /* results */ memcpy( store, rgba, 16 * sizeof(float)); } @@ -1747,14 +1761,6 @@ emit_instruction( if (indirect_temp_reference(inst)) return FALSE; - /* we don't handle saturation/clamping yet */ - if (inst->Instruction.Saturate != TGSI_SAT_NONE) - return FALSE; - - /* need to use extra temps to fix SOA dependencies : */ - if (tgsi_check_soa_dependencies(inst)) - return FALSE; - switch (inst->Instruction.Opcode) { case TGSI_OPCODE_ARL: FOR_EACH_DST0_ENABLED_CHANNEL( *inst, chan_index ) { @@ -1768,8 +1774,10 @@ emit_instruction( case TGSI_OPCODE_MOV: case TGSI_OPCODE_SWZ: FOR_EACH_DST0_ENABLED_CHANNEL( *inst, chan_index ) { - FETCH( func, *inst, 0, 0, chan_index ); - STORE( func, *inst, 0, 0, chan_index ); + FETCH( func, *inst, 4 + chan_index, 0, chan_index ); + } + FOR_EACH_DST0_ENABLED_CHANNEL( *inst, chan_index ) { + STORE( func, *inst, 4 + chan_index, 0, chan_index ); } break; @@ -1847,7 +1855,6 @@ emit_instruction( break; case TGSI_OPCODE_RCP: - /* TGSI_OPCODE_RECIP */ FETCH( func, *inst, 0, 0, CHAN_X ); emit_rcp( func, 0, 0 ); FOR_EACH_DST0_ENABLED_CHANNEL( *inst, chan_index ) { @@ -1856,7 +1863,6 @@ emit_instruction( break; case TGSI_OPCODE_RSQ: - /* TGSI_OPCODE_RECIPSQRT */ FETCH( func, *inst, 0, 0, CHAN_X ); emit_abs( func, 0 ); emit_rsqrt( func, 1, 0 ); @@ -1954,7 +1960,6 @@ emit_instruction( break; case TGSI_OPCODE_DP3: - /* TGSI_OPCODE_DOT3 */ FETCH( func, *inst, 0, 0, CHAN_X ); FETCH( func, *inst, 1, 1, CHAN_X ); emit_mul( func, 0, 1 ); @@ -1972,7 +1977,6 @@ emit_instruction( break; case TGSI_OPCODE_DP4: - /* TGSI_OPCODE_DOT4 */ FETCH( func, *inst, 0, 0, CHAN_X ); FETCH( func, *inst, 1, 1, CHAN_X ); emit_mul( func, 0, 1 ); @@ -2043,17 +2047,14 @@ emit_instruction( break; case TGSI_OPCODE_SLT: - /* TGSI_OPCODE_SETLT */ emit_setcc( func, inst, cc_LessThan ); break; case TGSI_OPCODE_SGE: - /* TGSI_OPCODE_SETGE */ emit_setcc( func, inst, cc_NotLessThan ); break; case TGSI_OPCODE_MAD: - /* TGSI_OPCODE_MADD */ FOR_EACH_DST0_ENABLED_CHANNEL( *inst, chan_index ) { FETCH( func, *inst, 0, 0, chan_index ); FETCH( func, *inst, 1, 1, chan_index ); @@ -2283,7 +2284,7 @@ emit_instruction( break; case TGSI_OPCODE_SEQ: - return 0; + emit_setcc( func, inst, cc_Equal ); break; case TGSI_OPCODE_SFL: @@ -2291,7 +2292,7 @@ emit_instruction( break; case TGSI_OPCODE_SGT: - return 0; + emit_setcc( func, inst, cc_NotLessThanEqual ); break; case TGSI_OPCODE_SIN: @@ -2303,11 +2304,11 @@ emit_instruction( break; case TGSI_OPCODE_SLE: - return 0; + emit_setcc( func, inst, cc_LessThanEqual ); break; case TGSI_OPCODE_SNE: - return 0; + emit_setcc( func, inst, cc_NotEqual ); break; case TGSI_OPCODE_STR: @@ -2371,7 +2372,6 @@ emit_instruction( break; case TGSI_OPCODE_SSG: - /* TGSI_OPCODE_SGN */ FOR_EACH_DST0_ENABLED_CHANNEL( *inst, chan_index ) { FETCH( func, *inst, 0, 0, chan_index ); emit_sgn( func, 0, 0 ); @@ -2929,6 +2929,22 @@ tgsi_emit_sse2( parse.FullHeader.Processor.Processor == TGSI_PROCESSOR_VERTEX ? "vertex shader" : "fragment shader"); } + + if (tgsi_check_soa_dependencies(&parse.FullToken.FullInstruction)) { + uint opcode = parse.FullToken.FullInstruction.Instruction.Opcode; + + /* XXX: we only handle src/dst aliasing in a few opcodes + * currently. Need to use an additional temporay to hold + * the result in the cases where the code is too opaque to + * fix. + */ + if (opcode != TGSI_OPCODE_MOV && + opcode != TGSI_OPCODE_SWZ) { + debug_printf("Warning: src/dst aliasing in instruction" + " is not handled:\n"); + tgsi_dump_instruction(&parse.FullToken.FullInstruction, 1); + } + } break; case TGSI_TOKEN_TYPE_IMMEDIATE: diff --git a/src/gallium/auxiliary/tgsi/tgsi_ureg.c b/src/gallium/auxiliary/tgsi/tgsi_ureg.c index f7096bd8e2c..654426a903d 100644 --- a/src/gallium/auxiliary/tgsi/tgsi_ureg.c +++ b/src/gallium/auxiliary/tgsi/tgsi_ureg.c @@ -31,6 +31,7 @@ #include "tgsi/tgsi_ureg.h" #include "tgsi/tgsi_info.h" #include "tgsi/tgsi_dump.h" +#include "tgsi/tgsi_sanity.h" #include "util/u_memory.h" #include "util/u_math.h" @@ -70,6 +71,7 @@ struct ureg_tokens { #define UREG_MAX_INPUT PIPE_MAX_ATTRIBS #define UREG_MAX_OUTPUT PIPE_MAX_ATTRIBS +#define UREG_MAX_CONSTANT_RANGE 32 #define UREG_MAX_IMMEDIATE 32 #define UREG_MAX_TEMP 256 #define UREG_MAX_ADDR 2 @@ -86,8 +88,10 @@ struct ureg_program unsigned semantic_name; unsigned semantic_index; unsigned interp; - } input[UREG_MAX_INPUT]; - unsigned nr_inputs; + } fs_input[UREG_MAX_INPUT]; + unsigned nr_fs_inputs; + + unsigned vs_inputs[UREG_MAX_INPUT/32]; struct { unsigned semantic_name; @@ -107,9 +111,13 @@ struct ureg_program unsigned temps_active[UREG_MAX_TEMP / 32]; unsigned nr_temps; - unsigned nr_addrs; + struct { + unsigned first; + unsigned last; + } constant_range[UREG_MAX_CONSTANT_RANGE]; + unsigned nr_constant_ranges; - unsigned nr_constants; + unsigned nr_addrs; unsigned nr_instructions; struct ureg_tokens domain[2]; @@ -119,6 +127,9 @@ static union tgsi_any_token error_tokens[32]; static void tokens_error( struct ureg_tokens *tokens ) { + if (tokens->tokens && tokens->tokens != error_tokens) + FREE(tokens->tokens); + tokens->tokens = error_tokens; tokens->size = Elements(error_tokens); tokens->count = 0; @@ -228,25 +239,25 @@ ureg_src_register( unsigned file, -static struct ureg_src -ureg_DECL_input( struct ureg_program *ureg, - unsigned name, - unsigned index, - unsigned interp_mode ) +struct ureg_src +ureg_DECL_fs_input( struct ureg_program *ureg, + unsigned name, + unsigned index, + unsigned interp_mode ) { unsigned i; - for (i = 0; i < ureg->nr_inputs; i++) { - if (ureg->input[i].semantic_name == name && - ureg->input[i].semantic_index == index) + for (i = 0; i < ureg->nr_fs_inputs; i++) { + if (ureg->fs_input[i].semantic_name == name && + ureg->fs_input[i].semantic_index == index) goto out; } - if (ureg->nr_inputs < UREG_MAX_INPUT) { - ureg->input[i].semantic_name = name; - ureg->input[i].semantic_index = index; - ureg->input[i].interp = interp_mode; - ureg->nr_inputs++; + if (ureg->nr_fs_inputs < UREG_MAX_INPUT) { + ureg->fs_input[i].semantic_name = name; + ureg->fs_input[i].semantic_index = index; + ureg->fs_input[i].interp = interp_mode; + ureg->nr_fs_inputs++; } else { set_bad( ureg ); @@ -257,25 +268,14 @@ out: } - -struct ureg_src -ureg_DECL_fs_input( struct ureg_program *ureg, - unsigned name, - unsigned index, - unsigned interp ) -{ - assert(ureg->processor == TGSI_PROCESSOR_FRAGMENT); - return ureg_DECL_input( ureg, name, index, interp ); -} - - struct ureg_src ureg_DECL_vs_input( struct ureg_program *ureg, - unsigned name, unsigned index ) { assert(ureg->processor == TGSI_PROCESSOR_VERTEX); - return ureg_DECL_input( ureg, name, index, TGSI_INTERPOLATE_CONSTANT ); + + ureg->vs_inputs[index/32] |= 1 << (index % 32); + return ureg_src_register( TGSI_FILE_INPUT, index ); } @@ -313,9 +313,57 @@ out: * value or manage any constant_buffer contents -- that's the * resposibility of the calling code. */ -struct ureg_src ureg_DECL_constant(struct ureg_program *ureg ) +struct ureg_src ureg_DECL_constant(struct ureg_program *ureg, + unsigned index ) { - return ureg_src_register( TGSI_FILE_CONSTANT, ureg->nr_constants++ ); + unsigned minconst = index, maxconst = index; + unsigned i; + + /* Inside existing range? + */ + for (i = 0; i < ureg->nr_constant_ranges; i++) { + if (ureg->constant_range[i].first <= index && + ureg->constant_range[i].last >= index) + goto out; + } + + /* Extend existing range? + */ + for (i = 0; i < ureg->nr_constant_ranges; i++) { + if (ureg->constant_range[i].last == index - 1) { + ureg->constant_range[i].last = index; + goto out; + } + + if (ureg->constant_range[i].first == index + 1) { + ureg->constant_range[i].first = index; + goto out; + } + + minconst = MIN2(minconst, ureg->constant_range[i].first); + maxconst = MAX2(maxconst, ureg->constant_range[i].last); + } + + /* Create new range? + */ + if (ureg->nr_constant_ranges < UREG_MAX_CONSTANT_RANGE) { + i = ureg->nr_constant_ranges++; + ureg->constant_range[i].first = index; + ureg->constant_range[i].last = index; + } + + /* Collapse all ranges down to one: + */ + i = 0; + ureg->constant_range[0].first = minconst; + ureg->constant_range[0].last = maxconst; + ureg->nr_constant_ranges = 1; + +out: + assert(i < ureg->nr_constant_ranges); + assert(ureg->constant_range[i].first <= index); + assert(ureg->constant_range[i].last >= index); + return ureg_src_register( TGSI_FILE_CONSTANT, index ); } @@ -566,6 +614,19 @@ ureg_emit_dst( struct ureg_program *ureg, } +static void validate( unsigned opcode, + unsigned nr_dst, + unsigned nr_src ) +{ +#ifdef DEBUG + const struct tgsi_opcode_info *info = tgsi_get_opcode_info( opcode ); + assert(info); + if(info) { + assert(nr_dst == info->num_dst); + assert(nr_src == info->num_src); + } +#endif +} unsigned ureg_emit_insn(struct ureg_program *ureg, @@ -576,6 +637,8 @@ ureg_emit_insn(struct ureg_program *ureg, { union tgsi_any_token *out; + validate( opcode, num_dst, num_src ); + out = get_tokens( ureg, DOMAIN_INSN, 1 ); out[0].value = 0; out[0].insn.Type = TGSI_TOKEN_TYPE_INSTRUCTION; @@ -678,17 +741,6 @@ ureg_insn(struct ureg_program *ureg, unsigned insn, i; boolean saturate; -#ifdef DEBUG - { - const struct tgsi_opcode_info *info = tgsi_get_opcode_info( opcode ); - assert(info); - if(info) { - assert(nr_dst == info->num_dst); - assert(nr_src == info->num_src); - } - } -#endif - saturate = nr_dst ? dst[0].Saturate : FALSE; insn = ureg_emit_insn( ureg, opcode, saturate, nr_dst, nr_src ); @@ -702,6 +754,53 @@ ureg_insn(struct ureg_program *ureg, ureg_fixup_insn_size( ureg, insn ); } +void +ureg_tex_insn(struct ureg_program *ureg, + unsigned opcode, + const struct ureg_dst *dst, + unsigned nr_dst, + unsigned target, + const struct ureg_src *src, + unsigned nr_src ) +{ + unsigned insn, i; + boolean saturate; + + saturate = nr_dst ? dst[0].Saturate : FALSE; + + insn = ureg_emit_insn( ureg, opcode, saturate, nr_dst, nr_src ); + + ureg_emit_texture( ureg, insn, target ); \ + + for (i = 0; i < nr_dst; i++) + ureg_emit_dst( ureg, dst[i] ); + + for (i = 0; i < nr_src; i++) + ureg_emit_src( ureg, src[i] ); + + ureg_fixup_insn_size( ureg, insn ); +} + + +void +ureg_label_insn(struct ureg_program *ureg, + unsigned opcode, + const struct ureg_src *src, + unsigned nr_src, + unsigned *label_token ) +{ + unsigned insn, i; + + insn = ureg_emit_insn( ureg, opcode, FALSE, 0, nr_src ); + + ureg_emit_label( ureg, insn, label_token ); \ + + for (i = 0; i < nr_src; i++) + ureg_emit_src( ureg, src[i] ); + + ureg_fixup_insn_size( ureg, insn ); +} + static void emit_decl( struct ureg_program *ureg, @@ -777,13 +876,22 @@ static void emit_decls( struct ureg_program *ureg ) { unsigned i; - for (i = 0; i < ureg->nr_inputs; i++) { - emit_decl( ureg, - TGSI_FILE_INPUT, - i, - ureg->input[i].semantic_name, - ureg->input[i].semantic_index, - ureg->input[i].interp ); + if (ureg->processor == TGSI_PROCESSOR_VERTEX) { + for (i = 0; i < UREG_MAX_INPUT; i++) { + if (ureg->vs_inputs[i/32] & (1 << (i%32))) { + emit_decl_range( ureg, TGSI_FILE_INPUT, i, 1 ); + } + } + } + else { + for (i = 0; i < ureg->nr_fs_inputs; i++) { + emit_decl( ureg, + TGSI_FILE_INPUT, + i, + ureg->fs_input[i].semantic_name, + ureg->fs_input[i].semantic_index, + ureg->fs_input[i].interp ); + } } for (i = 0; i < ureg->nr_outputs; i++) { @@ -801,10 +909,13 @@ static void emit_decls( struct ureg_program *ureg ) ureg->sampler[i].Index, 1 ); } - if (ureg->nr_constants) { - emit_decl_range( ureg, - TGSI_FILE_CONSTANT, - 0, ureg->nr_constants ); + if (ureg->nr_constant_ranges) { + for (i = 0; i < ureg->nr_constant_ranges; i++) + emit_decl_range( ureg, + TGSI_FILE_CONSTANT, + ureg->constant_range[i].first, + (ureg->constant_range[i].last + 1 - + ureg->constant_range[i].first) ); } if (ureg->nr_temps) { @@ -890,6 +1001,15 @@ const struct tgsi_token *ureg_finalize( struct ureg_program *ureg ) ureg->domain[DOMAIN_DECL].count); tgsi_dump( tokens, 0 ); } + +#if DEBUG + if (tokens && !tgsi_sanity_check(tokens)) { + debug_printf("tgsi_ureg.c, sanity check failed on generated tokens:\n"); + tgsi_dump(tokens, 0); + assert(0); + } +#endif + return tokens; } @@ -911,6 +1031,25 @@ void *ureg_create_shader( struct ureg_program *ureg, } +const struct tgsi_token *ureg_get_tokens( struct ureg_program *ureg, + unsigned *nr_tokens ) +{ + const struct tgsi_token *tokens; + + ureg_finalize(ureg); + + tokens = &ureg->domain[DOMAIN_DECL].tokens[0].token; + + if (nr_tokens) + *nr_tokens = ureg->domain[DOMAIN_DECL].size; + + ureg->domain[DOMAIN_DECL].tokens = 0; + ureg->domain[DOMAIN_DECL].size = 0; + ureg->domain[DOMAIN_DECL].order = 0; + ureg->domain[DOMAIN_DECL].count = 0; + + return tokens; +} struct ureg_program *ureg_create( unsigned processor ) diff --git a/src/gallium/auxiliary/tgsi/tgsi_ureg.h b/src/gallium/auxiliary/tgsi/tgsi_ureg.h index acbca59040c..f04f443b9e7 100644 --- a/src/gallium/auxiliary/tgsi/tgsi_ureg.h +++ b/src/gallium/auxiliary/tgsi/tgsi_ureg.h @@ -82,10 +82,21 @@ ureg_create( unsigned processor ); const struct tgsi_token * ureg_finalize( struct ureg_program * ); +/* Create and return a shader: + */ void * ureg_create_shader( struct ureg_program *, struct pipe_context *pipe ); + +/* Alternately, return the built token stream and hand ownership of + * that memory to the caller: + */ +const struct tgsi_token * +ureg_get_tokens( struct ureg_program *ureg, + unsigned *nr_tokens ); + + void ureg_destroy( struct ureg_program * ); @@ -116,8 +127,7 @@ ureg_DECL_fs_input( struct ureg_program *, struct ureg_src ureg_DECL_vs_input( struct ureg_program *, - unsigned semantic_name, - unsigned semantic_index ); + unsigned index ); struct ureg_dst ureg_DECL_output( struct ureg_program *, @@ -130,7 +140,8 @@ ureg_DECL_immediate( struct ureg_program *, unsigned nr ); struct ureg_src -ureg_DECL_constant( struct ureg_program * ); +ureg_DECL_constant( struct ureg_program *, + unsigned index ); struct ureg_dst ureg_DECL_temporary( struct ureg_program * ); @@ -233,6 +244,24 @@ ureg_insn(struct ureg_program *ureg, unsigned nr_src ); +void +ureg_tex_insn(struct ureg_program *ureg, + unsigned opcode, + const struct ureg_dst *dst, + unsigned nr_dst, + unsigned target, + const struct ureg_src *src, + unsigned nr_src ); + + +void +ureg_label_insn(struct ureg_program *ureg, + unsigned opcode, + const struct ureg_src *src, + unsigned nr_src, + unsigned *label); + + /*********************************************************************** * Internal instruction helpers, don't call these directly: */ diff --git a/src/gallium/auxiliary/util/u_debug.h b/src/gallium/auxiliary/util/u_debug.h index 1380d98d7ee..b82e7cb4d40 100644 --- a/src/gallium/auxiliary/util/u_debug.h +++ b/src/gallium/auxiliary/util/u_debug.h @@ -65,6 +65,11 @@ extern "C" { #define __FUNCTION__ "???" #endif +#if defined(__GNUC__) +#define _util_printf_format(fmt, list) __attribute__ ((format (printf, fmt, list))) +#else +#define _util_printf_format(fmt, list) +#endif void _debug_vprintf(const char *format, va_list ap); @@ -82,14 +87,17 @@ _debug_printf(const char *format, ...) /** * Print debug messages. * - * The actual channel used to output debug message is platform specific. To - * avoid misformating or truncation, follow these rules of thumb: + * The actual channel used to output debug message is platform specific. To + * avoid misformating or truncation, follow these rules of thumb: * - output whole lines - * - avoid outputing large strings (512 bytes is the current maximum length + * - 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, ...) _util_printf_format(1,2); + +static INLINE void debug_printf(const char *format, ...) { #ifdef DEBUG diff --git a/src/gallium/auxiliary/util/u_format.csv b/src/gallium/auxiliary/util/u_format.csv index 6e82983e586..f1bf94f17dd 100644 --- a/src/gallium/auxiliary/util/u_format.csv +++ b/src/gallium/auxiliary/util/u_format.csv @@ -2,7 +2,7 @@ PIPE_FORMAT_A8R8G8B8_UNORM , arith , 1, 1, un8 , un8 , un8 , un8 , zyxw, PIPE_FORMAT_X8R8G8B8_UNORM , arith , 1, 1, un8 , un8 , un8 , un8 , zyx1, rgb PIPE_FORMAT_B8G8R8A8_UNORM , arith , 1, 1, un8 , un8 , un8 , un8 , yzwx, rgb PIPE_FORMAT_B8G8R8X8_UNORM , arith , 1, 1, un8 , un8 , un8 , un8 , yzw1, rgb -PIPE_FORMAT_A1R5G5B5_UNORM , arith , 1, 1, un1 , un5 , un5 , un5 , zyxw, rgb +PIPE_FORMAT_A1R5G5B5_UNORM , arith , 1, 1, un5 , un5 , un5 , un1 , zyxw, rgb PIPE_FORMAT_A4R4G4B4_UNORM , arith , 1, 1, un4 , un4 , un4 , un4 , zyxw, rgb PIPE_FORMAT_R5G6B5_UNORM , arith , 1, 1, un5 , un6 , un5 , , zyx1, rgb PIPE_FORMAT_A2B10G10R10_UNORM , arith , 1, 1, un10, un10, un10, un2 , xyzw, rgb diff --git a/src/gallium/auxiliary/util/u_math.h b/src/gallium/auxiliary/util/u_math.h index b12c97dfb4d..cd6a9fcc091 100644 --- a/src/gallium/auxiliary/util/u_math.h +++ b/src/gallium/auxiliary/util/u_math.h @@ -283,6 +283,14 @@ util_fast_pow(float x, float y) return util_fast_exp2(util_fast_log2(x) * y); } +/* Note that this counts zero as a power of two. + */ +static INLINE boolean +util_is_power_of_two( unsigned v ) +{ + return (v & (v-1)) == 0; +} + /** * Floor(x), returned as int. @@ -354,7 +362,9 @@ util_is_pot(unsigned x) * Find first bit set in word. Least significant bit is 1. * Return 0 if no bits set. */ -#if defined(_MSC_VER) && _MSC_VER >= 1300 +#if defined(_MSC_VER) && _MSC_VER >= 1300 && (_M_IX86 || _M_AMD64 || _M_IA64) +unsigned char _BitScanForward(unsigned long* Index, unsigned long Mask); +#pragma intrinsic(_BitScanForward) static INLINE unsigned long ffs( unsigned long u ) { diff --git a/src/gallium/auxiliary/util/u_simple_screen.c b/src/gallium/auxiliary/util/u_simple_screen.c index f01296b40fc..52382990155 100644 --- a/src/gallium/auxiliary/util/u_simple_screen.c +++ b/src/gallium/auxiliary/util/u_simple_screen.c @@ -52,8 +52,7 @@ pass_user_buffer_create(struct pipe_screen *screen, unsigned bytes) { struct pipe_buffer *buffer = - screen->winsys->user_buffer_create(screen->winsys, - ptr, bytes); + screen->winsys->user_buffer_create(screen->winsys, ptr, bytes); buffer->screen = screen; @@ -69,9 +68,8 @@ pass_surface_buffer_create(struct pipe_screen *screen, unsigned *stride) { struct pipe_buffer *buffer = - screen->winsys->surface_buffer_create(screen->winsys, - width, height, - format, usage, tex_usage, stride); + screen->winsys->surface_buffer_create(screen->winsys, width, height, + format, usage, tex_usage, stride); buffer->screen = screen; @@ -83,8 +81,7 @@ pass_buffer_map(struct pipe_screen *screen, struct pipe_buffer *buf, unsigned usage) { - return screen->winsys->buffer_map(screen->winsys, - buf, usage); + return screen->winsys->buffer_map(screen->winsys, buf, usage); } static void @@ -106,8 +103,7 @@ pass_flush_frontbuffer(struct pipe_screen *screen, struct pipe_surface *surf, void *context_private) { - screen->winsys->flush_frontbuffer(screen->winsys, - surf, context_private); + screen->winsys->flush_frontbuffer(screen->winsys, surf, context_private); } static void @@ -115,8 +111,7 @@ pass_fence_reference(struct pipe_screen *screen, struct pipe_fence_handle **ptr, struct pipe_fence_handle *fence) { - screen->winsys->fence_reference(screen->winsys, - ptr, fence); + screen->winsys->fence_reference(screen->winsys, ptr, fence); } static int @@ -124,8 +119,7 @@ pass_fence_signalled(struct pipe_screen *screen, struct pipe_fence_handle *fence, unsigned flag) { - return screen->winsys->fence_signalled(screen->winsys, - fence, flag); + return screen->winsys->fence_signalled(screen->winsys, fence, flag); } static int @@ -133,11 +127,11 @@ pass_fence_finish(struct pipe_screen *screen, struct pipe_fence_handle *fence, unsigned flag) { - return screen->winsys->fence_finish(screen->winsys, - fence, flag); + return screen->winsys->fence_finish(screen->winsys, fence, flag); } -void u_simple_screen_init(struct pipe_screen *screen) +void +u_simple_screen_init(struct pipe_screen *screen) { screen->buffer_create = pass_buffer_create; screen->user_buffer_create = pass_user_buffer_create; @@ -152,7 +146,8 @@ void u_simple_screen_init(struct pipe_screen *screen) screen->fence_finish = pass_fence_finish; } -const char* u_simple_screen_winsys_name(struct pipe_screen *screen) +const char * +u_simple_screen_winsys_name(struct pipe_screen *screen) { return screen->winsys->get_name(screen->winsys); } diff --git a/src/gallium/auxiliary/util/u_simple_shaders.c b/src/gallium/auxiliary/util/u_simple_shaders.c index ab754296fa8..0d706f9449d 100644 --- a/src/gallium/auxiliary/util/u_simple_shaders.c +++ b/src/gallium/auxiliary/util/u_simple_shaders.c @@ -34,14 +34,8 @@ #include "pipe/p_context.h" -#include "util/u_debug.h" -#include "pipe/p_defines.h" -#include "pipe/p_screen.h" #include "pipe/p_shader_tokens.h" - -#include "util/u_memory.h" #include "util/u_simple_shaders.h" - #include "tgsi/tgsi_ureg.h" @@ -67,9 +61,7 @@ util_make_vertex_passthrough_shader(struct pipe_context *pipe, struct ureg_src src; struct ureg_dst dst; - src = ureg_DECL_vs_input( ureg, - semantic_names[i], - semantic_indexes[i]); + src = ureg_DECL_vs_input( ureg, i ); dst = ureg_DECL_output( ureg, semantic_names[i], diff --git a/src/gallium/drivers/cell/ppu/cell_screen.c b/src/gallium/drivers/cell/ppu/cell_screen.c index bd48ce70050..d185c6b8497 100644 --- a/src/gallium/drivers/cell/ppu/cell_screen.c +++ b/src/gallium/drivers/cell/ppu/cell_screen.c @@ -41,7 +41,7 @@ static const char * cell_get_vendor(struct pipe_screen *screen) { - return "Tungsten Graphics, Inc."; + return "VMware, Inc."; } @@ -64,8 +64,6 @@ cell_get_param(struct pipe_screen *screen, int param) return 1; case PIPE_CAP_GLSL: return 1; - case PIPE_CAP_S3TC: - return 0; case PIPE_CAP_ANISOTROPIC_FILTER: return 0; case PIPE_CAP_POINT_SPRITE: diff --git a/src/gallium/drivers/i915simple/i915_context.c b/src/gallium/drivers/i915simple/i915_context.c index b43f7352456..e745f3342d1 100644 --- a/src/gallium/drivers/i915simple/i915_context.c +++ b/src/gallium/drivers/i915simple/i915_context.c @@ -175,12 +175,19 @@ i915_is_buffer_referenced(struct pipe_context *pipe, static void i915_destroy(struct pipe_context *pipe) { struct i915_context *i915 = i915_context(pipe); + int i; draw_destroy(i915->draw); if(i915->batch) i915->iws->batchbuffer_destroy(i915->batch); + /* unbind framebuffer */ + for (i = 0; i < PIPE_MAX_COLOR_BUFS; i++) { + pipe_surface_reference(&i915->framebuffer.cbufs[i], NULL); + } + pipe_surface_reference(&i915->framebuffer.zsbuf, NULL); + FREE(i915); } diff --git a/src/gallium/drivers/i915simple/i915_screen.c b/src/gallium/drivers/i915simple/i915_screen.c index 9f017a14cca..c66558c320e 100644 --- a/src/gallium/drivers/i915simple/i915_screen.c +++ b/src/gallium/drivers/i915simple/i915_screen.c @@ -46,7 +46,7 @@ static const char * i915_get_vendor(struct pipe_screen *screen) { - return "Tungsten Graphics, Inc."; + return "VMware, Inc."; } static const char * @@ -101,8 +101,6 @@ i915_get_param(struct pipe_screen *screen, int param) return 1; case PIPE_CAP_GLSL: return 0; - case PIPE_CAP_S3TC: - return 0; case PIPE_CAP_ANISOTROPIC_FILTER: return 0; case PIPE_CAP_POINT_SPRITE: diff --git a/src/gallium/drivers/i915simple/i915_state.c b/src/gallium/drivers/i915simple/i915_state.c index 0087dfa410f..7d48e6e84d5 100644 --- a/src/gallium/drivers/i915simple/i915_state.c +++ b/src/gallium/drivers/i915simple/i915_state.c @@ -588,9 +588,17 @@ static void i915_set_framebuffer_state(struct pipe_context *pipe, const struct pipe_framebuffer_state *fb) { struct i915_context *i915 = i915_context(pipe); + int i; + draw_flush(i915->draw); - i915->framebuffer = *fb; /* struct copy */ + i915->framebuffer.width = fb->width; + i915->framebuffer.height = fb->height; + i915->framebuffer.nr_cbufs = fb->nr_cbufs; + for (i = 0; i < PIPE_MAX_COLOR_BUFS; i++) { + pipe_surface_reference(&i915->framebuffer.cbufs[i], fb->cbufs[i]); + } + pipe_surface_reference(&i915->framebuffer.zsbuf, fb->zsbuf); i915->dirty |= I915_NEW_FRAMEBUFFER; } diff --git a/src/gallium/drivers/i965simple/brw_screen.c b/src/gallium/drivers/i965simple/brw_screen.c index b22e105f106..4a84c4db23f 100644 --- a/src/gallium/drivers/i965simple/brw_screen.c +++ b/src/gallium/drivers/i965simple/brw_screen.c @@ -39,7 +39,7 @@ static const char * brw_get_vendor( struct pipe_screen *screen ) { - return "Tungsten Graphics, Inc."; + return "VMware, Inc."; } @@ -85,8 +85,6 @@ brw_get_param(struct pipe_screen *screen, int param) return 1; case PIPE_CAP_GLSL: return 0; - case PIPE_CAP_S3TC: - return 0; case PIPE_CAP_ANISOTROPIC_FILTER: return 0; case PIPE_CAP_POINT_SPRITE: diff --git a/src/gallium/drivers/llvmpipe/Makefile b/src/gallium/drivers/llvmpipe/Makefile index 06c586e6bb9..cd7b6356d28 100644 --- a/src/gallium/drivers/llvmpipe/Makefile +++ b/src/gallium/drivers/llvmpipe/Makefile @@ -3,6 +3,8 @@ include $(TOP)/configs/current LIBNAME = llvmpipe +CFLAGS += -D__STDC_CONSTANT_MACROS -D__STDC_LIMIT_MACROS + C_SOURCES = \ lp_bld_alpha.c \ lp_bld_arit.c \ diff --git a/src/gallium/drivers/llvmpipe/README b/src/gallium/drivers/llvmpipe/README index 498d21dea6c..89d08834a3c 100644 --- a/src/gallium/drivers/llvmpipe/README +++ b/src/gallium/drivers/llvmpipe/README @@ -8,13 +8,16 @@ Done so far is: - the whole fragment pipeline is code generated in a single function + - input interpolation + - depth testing + - texture sampling (not all state/formats are supported) + - fragment shader TGSI translation - same level of support as the TGSI SSE2 exec machine, with the exception we don't fallback to TGSI interpretation when an unsupported opcode is found, but just ignore it - - texture sampling via an intrinsic call - done in SoA layout - input interpolation also code generated @@ -28,16 +31,17 @@ Done so far is: any width and length - not all operations are implemented for these types yet though -Most mesa/progs/demos/* work. Speed is on par with Keith's softpipe-opt branch, -which includes hand written fast implementations for common cases. +Most mesa/progs/demos/* work. To do (probably by this order): - code generate stipple and stencil testing - - code generate texture sampling + - translate the remaining bits of texture sampling state - translate TGSI control flow instructions, and all other remaining opcodes + + - integrate with the draw module for VS code generation - code generate the triangle setup and rasterization @@ -93,7 +97,7 @@ Alternatively, you can build it with GNU make, if you prefer, by invoking it as make linux-llvm -but the rest of these instructions assume scons is used. +but the rest of these instructions assume that scons is used. Using @@ -108,6 +112,9 @@ or export LD_LIBRARY_PATH=$PWD/build/linux-x86-debug/lib:$LD_LIBRARY_PATH +For performance evaluation pass debug=no to scons, and use the corresponding +lib directory without the "-debug" suffix. + Unit testing ============ @@ -119,7 +126,7 @@ build/linux-???-debug/gallium/drivers/llvmpipe: - lp_test_conv: SIMD vector conversion - lp_test_format: pixel unpacking/packing -Some of this tests can output results and benchmarks to a tab-seperated-file +Some of this tests can output results and benchmarks to a tab-separated-file for posterior analysis, e.g.: build/linux-x86_64-debug/gallium/drivers/llvmpipe/lp_test_blend -o blend.tsv @@ -133,10 +140,10 @@ Development Notes at the top of the lp_bld_*.c functions. - All lp_bld_*.[ch] are isolated from the rest of the driver, and could/may be - put in a standalone Gallium state -> LLVM IR translation module. + put in a stand-alone Gallium state -> LLVM IR translation module. - We use LLVM-C bindings for now. They are not documented, but follow the C++ interfaces very closely, and appear to be complete enough for code generation. See http://npcontemplation.blogspot.com/2008/06/secret-of-llvm-c-bindings.html - for a standalone example. + for a stand-alone example. diff --git a/src/gallium/drivers/llvmpipe/SConscript b/src/gallium/drivers/llvmpipe/SConscript index dea4b703c45..f4a9a3b22e2 100644 --- a/src/gallium/drivers/llvmpipe/SConscript +++ b/src/gallium/drivers/llvmpipe/SConscript @@ -3,7 +3,7 @@ Import('*') env = env.Clone() env.Tool('llvm') -if env.has_key('LLVM_VERSION') is False: +if not env.has_key('LLVM_VERSION'): print 'warning: LLVM not found: not building llvmpipe' Return() diff --git a/src/gallium/drivers/llvmpipe/lp_bld_alpha.c b/src/gallium/drivers/llvmpipe/lp_bld_alpha.c index 49c2f911af7..2b4bc5c819d 100644 --- a/src/gallium/drivers/llvmpipe/lp_bld_alpha.c +++ b/src/gallium/drivers/llvmpipe/lp_bld_alpha.c @@ -45,7 +45,7 @@ void lp_build_alpha_test(LLVMBuilderRef builder, const struct pipe_alpha_state *state, - union lp_type type, + struct lp_type type, struct lp_build_mask_context *mask, LLVMValueRef alpha, LLVMValueRef ref) diff --git a/src/gallium/drivers/llvmpipe/lp_bld_alpha.h b/src/gallium/drivers/llvmpipe/lp_bld_alpha.h index 9dbcdb4daab..634575670db 100644 --- a/src/gallium/drivers/llvmpipe/lp_bld_alpha.h +++ b/src/gallium/drivers/llvmpipe/lp_bld_alpha.h @@ -38,14 +38,14 @@ #include <llvm-c/Core.h> struct pipe_alpha_state; -union lp_type; +struct lp_type; struct lp_build_mask_context; void lp_build_alpha_test(LLVMBuilderRef builder, const struct pipe_alpha_state *state, - union lp_type type, + struct lp_type type, struct lp_build_mask_context *mask, LLVMValueRef alpha, LLVMValueRef ref); diff --git a/src/gallium/drivers/llvmpipe/lp_bld_arit.c b/src/gallium/drivers/llvmpipe/lp_bld_arit.c index be7442d00ae..31433318a7e 100644 --- a/src/gallium/drivers/llvmpipe/lp_bld_arit.c +++ b/src/gallium/drivers/llvmpipe/lp_bld_arit.c @@ -65,7 +65,7 @@ lp_build_min_simple(struct lp_build_context *bld, LLVMValueRef a, LLVMValueRef b) { - const union lp_type type = bld->type; + const struct lp_type type = bld->type; const char *intrinsic = NULL; LLVMValueRef cond; @@ -113,7 +113,7 @@ lp_build_max_simple(struct lp_build_context *bld, LLVMValueRef a, LLVMValueRef b) { - const union lp_type type = bld->type; + const struct lp_type type = bld->type; const char *intrinsic = NULL; LLVMValueRef cond; @@ -159,7 +159,7 @@ LLVMValueRef lp_build_comp(struct lp_build_context *bld, LLVMValueRef a) { - const union lp_type type = bld->type; + const struct lp_type type = bld->type; if(a == bld->one) return bld->zero; @@ -188,7 +188,7 @@ lp_build_add(struct lp_build_context *bld, LLVMValueRef a, LLVMValueRef b) { - const union lp_type type = bld->type; + const struct lp_type type = bld->type; LLVMValueRef res; if(a == bld->zero) @@ -241,7 +241,7 @@ lp_build_sub(struct lp_build_context *bld, LLVMValueRef a, LLVMValueRef b) { - const union lp_type type = bld->type; + const struct lp_type type = bld->type; LLVMValueRef res; if(b == bld->zero) @@ -405,7 +405,7 @@ lp_build_mul(struct lp_build_context *bld, LLVMValueRef a, LLVMValueRef b) { - const union lp_type type = bld->type; + const struct lp_type type = bld->type; if(a == bld->zero) return bld->zero; @@ -477,7 +477,7 @@ lp_build_div(struct lp_build_context *bld, LLVMValueRef a, LLVMValueRef b) { - const union lp_type type = bld->type; + const struct lp_type type = bld->type; if(a == bld->zero) return bld->zero; @@ -590,7 +590,7 @@ LLVMValueRef lp_build_abs(struct lp_build_context *bld, LLVMValueRef a) { - const union lp_type type = bld->type; + const struct lp_type type = bld->type; LLVMTypeRef vec_type = lp_build_vec_type(type); if(!type.sign) @@ -623,6 +623,47 @@ lp_build_abs(struct lp_build_context *bld, } +LLVMValueRef +lp_build_sgn(struct lp_build_context *bld, + LLVMValueRef a) +{ + const struct lp_type type = bld->type; + LLVMTypeRef vec_type = lp_build_vec_type(type); + LLVMValueRef cond; + LLVMValueRef res; + + /* Handle non-zero case */ + if(!type.sign) { + /* if not zero then sign must be positive */ + res = bld->one; + } + else if(type.floating) { + /* Take the sign bit and add it to 1 constant */ + LLVMTypeRef int_vec_type = lp_build_int_vec_type(type); + LLVMValueRef mask = lp_build_int_const_scalar(type, (unsigned long long)1 << (type.width - 1)); + LLVMValueRef sign; + LLVMValueRef one; + sign = LLVMBuildBitCast(bld->builder, a, int_vec_type, ""); + sign = LLVMBuildAnd(bld->builder, sign, mask, ""); + one = LLVMConstBitCast(bld->one, int_vec_type); + res = LLVMBuildOr(bld->builder, sign, one, ""); + res = LLVMBuildBitCast(bld->builder, res, vec_type, ""); + } + else + { + LLVMValueRef minus_one = lp_build_const_scalar(type, -1.0); + cond = lp_build_cmp(bld, PIPE_FUNC_GREATER, a, bld->zero); + res = lp_build_select(bld, cond, bld->one, minus_one); + } + + /* Handle zero */ + cond = lp_build_cmp(bld, PIPE_FUNC_EQUAL, a, bld->zero); + res = lp_build_select(bld, cond, bld->zero, bld->one); + + return res; +} + + enum lp_build_round_sse41_mode { LP_BUILD_ROUND_SSE41_NEAREST = 0, @@ -637,7 +678,7 @@ lp_build_round_sse41(struct lp_build_context *bld, LLVMValueRef a, enum lp_build_round_sse41_mode mode) { - const union lp_type type = bld->type; + const struct lp_type type = bld->type; LLVMTypeRef vec_type = lp_build_vec_type(type); const char *intrinsic; @@ -662,10 +703,28 @@ lp_build_round_sse41(struct lp_build_context *bld, LLVMValueRef +lp_build_round(struct lp_build_context *bld, + LLVMValueRef a) +{ + const struct lp_type type = bld->type; + + assert(type.floating); + +#if defined(PIPE_ARCH_X86) || defined(PIPE_ARCH_X86_64) + return lp_build_round_sse41(bld, a, LP_BUILD_ROUND_SSE41_NEAREST); +#endif + + /* FIXME */ + assert(0); + return bld->undef; +} + + +LLVMValueRef lp_build_floor(struct lp_build_context *bld, LLVMValueRef a) { - const union lp_type type = bld->type; + const struct lp_type type = bld->type; assert(type.floating); @@ -679,6 +738,42 @@ lp_build_floor(struct lp_build_context *bld, } +LLVMValueRef +lp_build_ceil(struct lp_build_context *bld, + LLVMValueRef a) +{ + const struct lp_type type = bld->type; + + assert(type.floating); + +#if defined(PIPE_ARCH_X86) || defined(PIPE_ARCH_X86_64) + return lp_build_round_sse41(bld, a, LP_BUILD_ROUND_SSE41_CEIL); +#endif + + /* FIXME */ + assert(0); + return bld->undef; +} + + +LLVMValueRef +lp_build_trunc(struct lp_build_context *bld, + LLVMValueRef a) +{ + const struct lp_type type = bld->type; + + assert(type.floating); + +#if defined(PIPE_ARCH_X86) || defined(PIPE_ARCH_X86_64) + return lp_build_round_sse41(bld, a, LP_BUILD_ROUND_SSE41_TRUNCATE); +#endif + + /* FIXME */ + assert(0); + return bld->undef; +} + + /** * Convert to integer, through whichever rounding method that's fastest, * typically truncating to zero. @@ -687,7 +782,7 @@ LLVMValueRef lp_build_int(struct lp_build_context *bld, LLVMValueRef a) { - const union lp_type type = bld->type; + const struct lp_type type = bld->type; LLVMTypeRef int_vec_type = lp_build_int_vec_type(type); assert(type.floating); @@ -710,7 +805,7 @@ LLVMValueRef lp_build_sqrt(struct lp_build_context *bld, LLVMValueRef a) { - const union lp_type type = bld->type; + const struct lp_type type = bld->type; LLVMTypeRef vec_type = lp_build_vec_type(type); char intrinsic[32]; @@ -728,7 +823,7 @@ LLVMValueRef lp_build_rcp(struct lp_build_context *bld, LLVMValueRef a) { - const union lp_type type = bld->type; + const struct lp_type type = bld->type; if(a == bld->zero) return bld->undef; @@ -759,7 +854,7 @@ LLVMValueRef lp_build_rsqrt(struct lp_build_context *bld, LLVMValueRef a) { - const union lp_type type = bld->type; + const struct lp_type type = bld->type; assert(type.floating); @@ -780,7 +875,7 @@ LLVMValueRef lp_build_cos(struct lp_build_context *bld, LLVMValueRef a) { - const union lp_type type = bld->type; + const struct lp_type type = bld->type; LLVMTypeRef vec_type = lp_build_vec_type(type); char intrinsic[32]; @@ -800,7 +895,7 @@ LLVMValueRef lp_build_sin(struct lp_build_context *bld, LLVMValueRef a) { - const union lp_type type = bld->type; + const struct lp_type type = bld->type; LLVMTypeRef vec_type = lp_build_vec_type(type); char intrinsic[32]; @@ -823,7 +918,8 @@ lp_build_pow(struct lp_build_context *bld, { /* TODO: optimize the constant case */ if(LLVMIsConstant(x) && LLVMIsConstant(y)) - debug_printf("%s: inefficient/imprecise constant arithmetic\n"); + debug_printf("%s: inefficient/imprecise constant arithmetic\n", + __FUNCTION__); return lp_build_exp2(bld, lp_build_mul(bld, lp_build_log2(bld, x), y)); } @@ -871,13 +967,14 @@ lp_build_polynomial(struct lp_build_context *bld, const double *coeffs, unsigned num_coeffs) { - const union lp_type type = bld->type; + const struct lp_type type = bld->type; LLVMValueRef res = NULL; unsigned i; /* TODO: optimize the constant case */ if(LLVMIsConstant(x)) - debug_printf("%s: inefficient/imprecise constant arithmetic\n"); + debug_printf("%s: inefficient/imprecise constant arithmetic\n", + __FUNCTION__); for (i = num_coeffs; i--; ) { LLVMValueRef coeff = lp_build_const_scalar(type, coeffs[i]); @@ -919,7 +1016,7 @@ lp_build_exp2_approx(struct lp_build_context *bld, LLVMValueRef *p_frac_part, LLVMValueRef *p_exp2) { - const union lp_type type = bld->type; + const struct lp_type type = bld->type; LLVMTypeRef vec_type = lp_build_vec_type(type); LLVMTypeRef int_vec_type = lp_build_int_vec_type(type); LLVMValueRef ipart = NULL; @@ -931,7 +1028,8 @@ lp_build_exp2_approx(struct lp_build_context *bld, if(p_exp2_int_part || p_frac_part || p_exp2) { /* TODO: optimize the constant case */ if(LLVMIsConstant(x)) - debug_printf("%s: inefficient/imprecise constant arithmetic\n"); + debug_printf("%s: inefficient/imprecise constant arithmetic\n", + __FUNCTION__); assert(type.floating && type.width == 32); @@ -1012,7 +1110,7 @@ lp_build_log2_approx(struct lp_build_context *bld, LLVMValueRef *p_floor_log2, LLVMValueRef *p_log2) { - const union lp_type type = bld->type; + const struct lp_type type = bld->type; LLVMTypeRef vec_type = lp_build_vec_type(type); LLVMTypeRef int_vec_type = lp_build_int_vec_type(type); @@ -1030,7 +1128,8 @@ lp_build_log2_approx(struct lp_build_context *bld, if(p_exp || p_floor_log2 || p_log2) { /* TODO: optimize the constant case */ if(LLVMIsConstant(x)) - debug_printf("%s: inefficient/imprecise constant arithmetic\n"); + debug_printf("%s: inefficient/imprecise constant arithmetic\n", + __FUNCTION__); assert(type.floating && type.width == 32); diff --git a/src/gallium/drivers/llvmpipe/lp_bld_arit.h b/src/gallium/drivers/llvmpipe/lp_bld_arit.h index 383c3c3313b..d68a97c4b87 100644 --- a/src/gallium/drivers/llvmpipe/lp_bld_arit.h +++ b/src/gallium/drivers/llvmpipe/lp_bld_arit.h @@ -40,7 +40,7 @@ #include <llvm-c/Core.h> -union lp_type type; +struct lp_type type; struct lp_build_context; @@ -106,10 +106,26 @@ lp_build_abs(struct lp_build_context *bld, LLVMValueRef a); LLVMValueRef +lp_build_sgn(struct lp_build_context *bld, + LLVMValueRef a); + +LLVMValueRef +lp_build_round(struct lp_build_context *bld, + LLVMValueRef a); + +LLVMValueRef lp_build_floor(struct lp_build_context *bld, LLVMValueRef a); LLVMValueRef +lp_build_ceil(struct lp_build_context *bld, + LLVMValueRef a); + +LLVMValueRef +lp_build_trunc(struct lp_build_context *bld, + LLVMValueRef a); + +LLVMValueRef lp_build_int(struct lp_build_context *bld, LLVMValueRef a); diff --git a/src/gallium/drivers/llvmpipe/lp_bld_blend.h b/src/gallium/drivers/llvmpipe/lp_bld_blend.h index d19e18846c2..da272e549f3 100644 --- a/src/gallium/drivers/llvmpipe/lp_bld_blend.h +++ b/src/gallium/drivers/llvmpipe/lp_bld_blend.h @@ -46,7 +46,7 @@ struct pipe_blend_state; -union lp_type; +struct lp_type; struct lp_build_context; @@ -74,7 +74,7 @@ lp_build_blend_func(struct lp_build_context *bld, LLVMValueRef lp_build_blend_aos(LLVMBuilderRef builder, const struct pipe_blend_state *blend, - union lp_type type, + struct lp_type type, LLVMValueRef src, LLVMValueRef dst, LLVMValueRef const_, @@ -84,7 +84,7 @@ lp_build_blend_aos(LLVMBuilderRef builder, void lp_build_blend_soa(LLVMBuilderRef builder, const struct pipe_blend_state *blend, - union lp_type type, + struct lp_type type, LLVMValueRef src[4], LLVMValueRef dst[4], LLVMValueRef const_[4], diff --git a/src/gallium/drivers/llvmpipe/lp_bld_blend_aos.c b/src/gallium/drivers/llvmpipe/lp_bld_blend_aos.c index c11a9398f87..d14f468ba93 100644 --- a/src/gallium/drivers/llvmpipe/lp_bld_blend_aos.c +++ b/src/gallium/drivers/llvmpipe/lp_bld_blend_aos.c @@ -303,7 +303,7 @@ lp_build_blend_func(struct lp_build_context *bld, LLVMValueRef lp_build_blend_aos(LLVMBuilderRef builder, const struct pipe_blend_state *blend, - union lp_type type, + struct lp_type type, LLVMValueRef src, LLVMValueRef dst, LLVMValueRef const_, diff --git a/src/gallium/drivers/llvmpipe/lp_bld_blend_soa.c b/src/gallium/drivers/llvmpipe/lp_bld_blend_soa.c index b92254a7d6f..9511299d558 100644 --- a/src/gallium/drivers/llvmpipe/lp_bld_blend_soa.c +++ b/src/gallium/drivers/llvmpipe/lp_bld_blend_soa.c @@ -199,7 +199,7 @@ lp_build_blend_soa_factor(struct lp_build_blend_soa_context *bld, void lp_build_blend_soa(LLVMBuilderRef builder, const struct pipe_blend_state *blend, - union lp_type type, + struct lp_type type, LLVMValueRef src[4], LLVMValueRef dst[4], LLVMValueRef con[4], diff --git a/src/gallium/drivers/llvmpipe/lp_bld_const.c b/src/gallium/drivers/llvmpipe/lp_bld_const.c index 21487365eae..c8eaa8c3940 100644 --- a/src/gallium/drivers/llvmpipe/lp_bld_const.c +++ b/src/gallium/drivers/llvmpipe/lp_bld_const.c @@ -42,7 +42,7 @@ unsigned -lp_mantissa(union lp_type type) +lp_mantissa(struct lp_type type) { assert(type.floating); @@ -72,7 +72,7 @@ lp_mantissa(union lp_type type) * Same as lp_const_scale(), but in terms of shifts. */ unsigned -lp_const_shift(union lp_type type) +lp_const_shift(struct lp_type type) { if(type.floating) return 0; @@ -86,7 +86,7 @@ lp_const_shift(union lp_type type) unsigned -lp_const_offset(union lp_type type) +lp_const_offset(struct lp_type type) { if(type.floating || type.fixed) return 0; @@ -104,7 +104,7 @@ lp_const_offset(union lp_type type) * else for the fixed points types and normalized integers. */ double -lp_const_scale(union lp_type type) +lp_const_scale(struct lp_type type) { unsigned long long llscale; double dscale; @@ -122,7 +122,7 @@ lp_const_scale(union lp_type type) * Minimum value representable by the type. */ double -lp_const_min(union lp_type type) +lp_const_min(struct lp_type type) { unsigned bits; @@ -158,7 +158,7 @@ lp_const_min(union lp_type type) * Maximum value representable by the type. */ double -lp_const_max(union lp_type type) +lp_const_max(struct lp_type type) { unsigned bits; @@ -190,7 +190,7 @@ lp_const_max(union lp_type type) double -lp_const_eps(union lp_type type) +lp_const_eps(struct lp_type type) { if (type.floating) { switch(type.width) { @@ -211,7 +211,7 @@ lp_const_eps(union lp_type type) LLVMValueRef -lp_build_undef(union lp_type type) +lp_build_undef(struct lp_type type) { LLVMTypeRef vec_type = lp_build_vec_type(type); return LLVMGetUndef(vec_type); @@ -219,7 +219,7 @@ lp_build_undef(union lp_type type) LLVMValueRef -lp_build_zero(union lp_type type) +lp_build_zero(struct lp_type type) { LLVMTypeRef vec_type = lp_build_vec_type(type); return LLVMConstNull(vec_type); @@ -227,7 +227,7 @@ lp_build_zero(union lp_type type) LLVMValueRef -lp_build_one(union lp_type type) +lp_build_one(struct lp_type type) { LLVMTypeRef elem_type; LLVMValueRef elems[LP_MAX_VECTOR_LENGTH]; @@ -269,7 +269,7 @@ lp_build_one(union lp_type type) LLVMValueRef -lp_build_const_scalar(union lp_type type, +lp_build_const_scalar(struct lp_type type, double val) { LLVMTypeRef elem_type = lp_build_elem_type(type); @@ -295,7 +295,7 @@ lp_build_const_scalar(union lp_type type, LLVMValueRef -lp_build_int_const_scalar(union lp_type type, +lp_build_int_const_scalar(struct lp_type type, long long val) { LLVMTypeRef elem_type = lp_build_int_elem_type(type); @@ -312,7 +312,7 @@ lp_build_int_const_scalar(union lp_type type, LLVMValueRef -lp_build_const_aos(union lp_type type, +lp_build_const_aos(struct lp_type type, double r, double g, double b, double a, const unsigned char *swizzle) { @@ -352,8 +352,8 @@ lp_build_const_aos(union lp_type type, LLVMValueRef -lp_build_const_mask_aos(union lp_type type, - boolean cond[4]) +lp_build_const_mask_aos(struct lp_type type, + const boolean cond[4]) { LLVMTypeRef elem_type = LLVMIntType(type.width); LLVMValueRef masks[LP_MAX_VECTOR_LENGTH]; diff --git a/src/gallium/drivers/llvmpipe/lp_bld_const.h b/src/gallium/drivers/llvmpipe/lp_bld_const.h index 1934530ea3c..ffb302f7366 100644 --- a/src/gallium/drivers/llvmpipe/lp_bld_const.h +++ b/src/gallium/drivers/llvmpipe/lp_bld_const.h @@ -42,67 +42,67 @@ #include <pipe/p_compiler.h> -union lp_type type; +struct lp_type type; unsigned -lp_mantissa(union lp_type type); +lp_mantissa(struct lp_type type); unsigned -lp_const_shift(union lp_type type); +lp_const_shift(struct lp_type type); unsigned -lp_const_offset(union lp_type type); +lp_const_offset(struct lp_type type); double -lp_const_scale(union lp_type type); +lp_const_scale(struct lp_type type); double -lp_const_min(union lp_type type); +lp_const_min(struct lp_type type); double -lp_const_max(union lp_type type); +lp_const_max(struct lp_type type); double -lp_const_eps(union lp_type type); +lp_const_eps(struct lp_type type); LLVMValueRef -lp_build_undef(union lp_type type); +lp_build_undef(struct lp_type type); LLVMValueRef -lp_build_zero(union lp_type type); +lp_build_zero(struct lp_type type); LLVMValueRef -lp_build_one(union lp_type type); +lp_build_one(struct lp_type type); LLVMValueRef -lp_build_const_scalar(union lp_type type, +lp_build_const_scalar(struct lp_type type, double val); LLVMValueRef -lp_build_int_const_scalar(union lp_type type, +lp_build_int_const_scalar(struct lp_type type, long long val); LLVMValueRef -lp_build_const_aos(union lp_type type, +lp_build_const_aos(struct lp_type type, double r, double g, double b, double a, const unsigned char *swizzle); LLVMValueRef -lp_build_const_mask_aos(union lp_type type, - boolean cond[4]); +lp_build_const_mask_aos(struct lp_type type, + const boolean cond[4]); #endif /* !LP_BLD_CONST_H */ diff --git a/src/gallium/drivers/llvmpipe/lp_bld_conv.c b/src/gallium/drivers/llvmpipe/lp_bld_conv.c index c8954c8a34f..186cac70f62 100644 --- a/src/gallium/drivers/llvmpipe/lp_bld_conv.c +++ b/src/gallium/drivers/llvmpipe/lp_bld_conv.c @@ -86,7 +86,7 @@ */ LLVMValueRef lp_build_clamped_float_to_unsigned_norm(LLVMBuilderRef builder, - union lp_type src_type, + struct lp_type src_type, unsigned dst_width, LLVMValueRef src) { @@ -122,7 +122,7 @@ lp_build_clamped_float_to_unsigned_norm(LLVMBuilderRef builder, int shift = dst_width - n; res = LLVMBuildShl(builder, res, lp_build_int_const_scalar(src_type, shift), ""); - /* Fill in the empty lower bits for added precision? */ + /* TODO: Fill in the empty lower bits for additional precision? */ #if 0 { LLVMValueRef msb; @@ -152,7 +152,7 @@ lp_build_clamped_float_to_unsigned_norm(LLVMBuilderRef builder, LLVMValueRef lp_build_unsigned_norm_to_float(LLVMBuilderRef builder, unsigned src_width, - union lp_type dst_type, + struct lp_type dst_type, LLVMValueRef src) { LLVMTypeRef vec_type = lp_build_vec_type(dst_type); @@ -244,12 +244,12 @@ lp_build_const_pack_shuffle(unsigned n) * Expand the bit width. * * This will only change the number of bits the values are represented, not the - * values themselved. + * values themselves. */ static void lp_build_expand(LLVMBuilderRef builder, - union lp_type src_type, - union lp_type dst_type, + struct lp_type src_type, + struct lp_type dst_type, LLVMValueRef src, LLVMValueRef *dst, unsigned num_dsts) { @@ -266,7 +266,7 @@ lp_build_expand(LLVMBuilderRef builder, dst[0] = src; while(src_type.width < dst_type.width) { - union lp_type new_type = src_type; + struct lp_type new_type = src_type; LLVMTypeRef new_vec_type; new_type.width *= 2; @@ -314,8 +314,8 @@ lp_build_expand(LLVMBuilderRef builder, */ static LLVMValueRef lp_build_pack2(LLVMBuilderRef builder, - union lp_type src_type, - union lp_type dst_type, + struct lp_type src_type, + struct lp_type dst_type, boolean clamped, LLVMValueRef lo, LLVMValueRef hi) @@ -391,11 +391,11 @@ lp_build_pack2(LLVMBuilderRef builder, * TODO: Handle saturation consistently. */ static LLVMValueRef -lp_build_trunc(LLVMBuilderRef builder, - union lp_type src_type, - union lp_type dst_type, - boolean clamped, - const LLVMValueRef *src, unsigned num_srcs) +lp_build_pack(LLVMBuilderRef builder, + struct lp_type src_type, + struct lp_type dst_type, + boolean clamped, + const LLVMValueRef *src, unsigned num_srcs) { LLVMValueRef tmp[LP_MAX_VECTOR_LENGTH]; unsigned i; @@ -410,7 +410,7 @@ lp_build_trunc(LLVMBuilderRef builder, tmp[i] = src[i]; while(src_type.width > dst_type.width) { - union lp_type new_type = src_type; + struct lp_type new_type = src_type; new_type.width /= 2; new_type.length *= 2; @@ -442,12 +442,12 @@ lp_build_trunc(LLVMBuilderRef builder, */ void lp_build_conv(LLVMBuilderRef builder, - union lp_type src_type, - union lp_type dst_type, + struct lp_type src_type, + struct lp_type dst_type, const LLVMValueRef *src, unsigned num_srcs, LLVMValueRef *dst, unsigned num_dsts) { - union lp_type tmp_type; + struct lp_type tmp_type; LLVMValueRef tmp[LP_MAX_VECTOR_LENGTH]; unsigned num_tmps; unsigned i; @@ -470,7 +470,7 @@ lp_build_conv(LLVMBuilderRef builder, * Clamp if necessary */ - if(src_type.value != dst_type.value) { + if(memcmp(&src_type, &dst_type, sizeof src_type) != 0) { struct lp_build_context bld; double src_min = lp_const_min(src_type); double dst_min = lp_const_min(dst_type); @@ -565,7 +565,7 @@ lp_build_conv(LLVMBuilderRef builder, if(tmp_type.width > dst_type.width) { assert(num_dsts == 1); - tmp[0] = lp_build_trunc(builder, tmp_type, dst_type, TRUE, tmp, num_tmps); + tmp[0] = lp_build_pack(builder, tmp_type, dst_type, TRUE, tmp, num_tmps); tmp_type.width = dst_type.width; tmp_type.length = dst_type.length; num_tmps = 1; @@ -656,8 +656,8 @@ lp_build_conv(LLVMBuilderRef builder, */ void lp_build_conv_mask(LLVMBuilderRef builder, - union lp_type src_type, - union lp_type dst_type, + struct lp_type src_type, + struct lp_type dst_type, const LLVMValueRef *src, unsigned num_srcs, LLVMValueRef *dst, unsigned num_dsts) { @@ -689,7 +689,7 @@ lp_build_conv_mask(LLVMBuilderRef builder, if(src_type.width > dst_type.width) { assert(num_dsts == 1); - dst[0] = lp_build_trunc(builder, src_type, dst_type, TRUE, src, num_srcs); + dst[0] = lp_build_pack(builder, src_type, dst_type, TRUE, src, num_srcs); } else if(src_type.width < dst_type.width) { assert(num_srcs == 1); diff --git a/src/gallium/drivers/llvmpipe/lp_bld_conv.h b/src/gallium/drivers/llvmpipe/lp_bld_conv.h index 05c1ef2a100..ca378804d2a 100644 --- a/src/gallium/drivers/llvmpipe/lp_bld_conv.h +++ b/src/gallium/drivers/llvmpipe/lp_bld_conv.h @@ -40,33 +40,33 @@ #include <llvm-c/Core.h> -union lp_type type; +struct lp_type type; LLVMValueRef lp_build_clamped_float_to_unsigned_norm(LLVMBuilderRef builder, - union lp_type src_type, + struct lp_type src_type, unsigned dst_width, LLVMValueRef src); LLVMValueRef lp_build_unsigned_norm_to_float(LLVMBuilderRef builder, unsigned src_width, - union lp_type dst_type, + struct lp_type dst_type, LLVMValueRef src); void lp_build_conv(LLVMBuilderRef builder, - union lp_type src_type, - union lp_type dst_type, + struct lp_type src_type, + struct lp_type dst_type, const LLVMValueRef *srcs, unsigned num_srcs, LLVMValueRef *dsts, unsigned num_dsts); void lp_build_conv_mask(LLVMBuilderRef builder, - union lp_type src_type, - union lp_type dst_type, + struct lp_type src_type, + struct lp_type dst_type, const LLVMValueRef *src, unsigned num_srcs, LLVMValueRef *dst, unsigned num_dsts); diff --git a/src/gallium/drivers/llvmpipe/lp_bld_depth.c b/src/gallium/drivers/llvmpipe/lp_bld_depth.c index 3f88a14b5d3..21c665c4d4c 100644 --- a/src/gallium/drivers/llvmpipe/lp_bld_depth.c +++ b/src/gallium/drivers/llvmpipe/lp_bld_depth.c @@ -71,11 +71,11 @@ /** * Return a type appropriate for depth/stencil testing. */ -union lp_type +struct lp_type lp_depth_type(const struct util_format_description *format_desc, unsigned length) { - union lp_type type; + struct lp_type type; unsigned swizzle; assert(format_desc->colorspace == UTIL_FORMAT_COLORSPACE_ZS); @@ -85,7 +85,7 @@ lp_depth_type(const struct util_format_description *format_desc, swizzle = format_desc->swizzle[0]; assert(swizzle < 4); - type.value = 0; + memset(&type, 0, sizeof type); type.width = format_desc->block.bits; if(format_desc->channel[swizzle].type == UTIL_FORMAT_TYPE_FLOAT) { @@ -114,7 +114,7 @@ lp_depth_type(const struct util_format_description *format_desc, void lp_build_depth_test(LLVMBuilderRef builder, const struct pipe_depth_state *state, - union lp_type type, + struct lp_type type, const struct util_format_description *format_desc, struct lp_build_mask_context *mask, LLVMValueRef src, @@ -211,5 +211,6 @@ lp_build_depth_test(LLVMBuilderRef builder, LLVMBuildStore(builder, dst, dst_ptr); } + /* FIXME */ assert(!state->occlusion_count); } diff --git a/src/gallium/drivers/llvmpipe/lp_bld_depth.h b/src/gallium/drivers/llvmpipe/lp_bld_depth.h index 5d2e042fcc5..79d6981bb51 100644 --- a/src/gallium/drivers/llvmpipe/lp_bld_depth.h +++ b/src/gallium/drivers/llvmpipe/lp_bld_depth.h @@ -41,11 +41,11 @@ struct pipe_depth_state; struct util_format_description; -union lp_type; +struct lp_type; struct lp_build_mask_context; -union lp_type +struct lp_type lp_depth_type(const struct util_format_description *format_desc, unsigned length); @@ -53,7 +53,7 @@ lp_depth_type(const struct util_format_description *format_desc, void lp_build_depth_test(LLVMBuilderRef builder, const struct pipe_depth_state *state, - union lp_type type, + struct lp_type type, const struct util_format_description *format_desc, struct lp_build_mask_context *mask, LLVMValueRef src, diff --git a/src/gallium/drivers/llvmpipe/lp_bld_flow.c b/src/gallium/drivers/llvmpipe/lp_bld_flow.c index 69ed014ff3d..dcc25fbff86 100644 --- a/src/gallium/drivers/llvmpipe/lp_bld_flow.c +++ b/src/gallium/drivers/llvmpipe/lp_bld_flow.c @@ -405,7 +405,7 @@ lp_build_mask_check(struct lp_build_mask_context *mask) void lp_build_mask_begin(struct lp_build_mask_context *mask, struct lp_build_flow_context *flow, - union lp_type type, + struct lp_type type, LLVMValueRef value) { memset(mask, 0, sizeof *mask); diff --git a/src/gallium/drivers/llvmpipe/lp_bld_flow.h b/src/gallium/drivers/llvmpipe/lp_bld_flow.h index 9d76e3064dd..e61999ff06b 100644 --- a/src/gallium/drivers/llvmpipe/lp_bld_flow.h +++ b/src/gallium/drivers/llvmpipe/lp_bld_flow.h @@ -38,7 +38,7 @@ #include <llvm-c/Core.h> -union lp_type; +struct lp_type; struct lp_build_flow_context; @@ -84,7 +84,7 @@ struct lp_build_mask_context void lp_build_mask_begin(struct lp_build_mask_context *mask, struct lp_build_flow_context *flow, - union lp_type type, + struct lp_type type, LLVMValueRef value); /** diff --git a/src/gallium/drivers/llvmpipe/lp_bld_format.h b/src/gallium/drivers/llvmpipe/lp_bld_format.h index 5ee06560932..6d3f6926193 100644 --- a/src/gallium/drivers/llvmpipe/lp_bld_format.h +++ b/src/gallium/drivers/llvmpipe/lp_bld_format.h @@ -39,7 +39,7 @@ #include "pipe/p_format.h" struct util_format_description; -union lp_type; +struct lp_type; /** @@ -103,7 +103,7 @@ lp_build_gather(LLVMBuilderRef builder, void lp_build_unpack_rgba_soa(LLVMBuilderRef builder, const struct util_format_description *format_desc, - union lp_type type, + struct lp_type type, LLVMValueRef packed, LLVMValueRef *rgba); @@ -111,7 +111,7 @@ lp_build_unpack_rgba_soa(LLVMBuilderRef builder, void lp_build_load_rgba_soa(LLVMBuilderRef builder, const struct util_format_description *format_desc, - union lp_type type, + struct lp_type type, LLVMValueRef base_ptr, LLVMValueRef offsets, LLVMValueRef *rgba); diff --git a/src/gallium/drivers/llvmpipe/lp_bld_format_soa.c b/src/gallium/drivers/llvmpipe/lp_bld_format_soa.c index 569e8d10a31..b5ff434e1ae 100644 --- a/src/gallium/drivers/llvmpipe/lp_bld_format_soa.c +++ b/src/gallium/drivers/llvmpipe/lp_bld_format_soa.c @@ -84,7 +84,7 @@ lp_build_gather(LLVMBuilderRef builder, static LLVMValueRef -lp_build_format_swizzle(union lp_type type, +lp_build_format_swizzle(struct lp_type type, const LLVMValueRef *inputs, enum util_format_swizzle swizzle) { @@ -110,7 +110,7 @@ lp_build_format_swizzle(union lp_type type, void lp_build_unpack_rgba_soa(LLVMBuilderRef builder, const struct util_format_description *format_desc, - union lp_type type, + struct lp_type type, LLVMValueRef packed, LLVMValueRef *rgba) { @@ -188,7 +188,7 @@ lp_build_unpack_rgba_soa(LLVMBuilderRef builder, void lp_build_load_rgba_soa(LLVMBuilderRef builder, const struct util_format_description *format_desc, - union lp_type type, + struct lp_type type, LLVMValueRef base_ptr, LLVMValueRef offsets, LLVMValueRef *rgba) diff --git a/src/gallium/drivers/llvmpipe/lp_bld_interp.c b/src/gallium/drivers/llvmpipe/lp_bld_interp.c index cfe20a0d75b..338dbca6d1e 100644 --- a/src/gallium/drivers/llvmpipe/lp_bld_interp.c +++ b/src/gallium/drivers/llvmpipe/lp_bld_interp.c @@ -292,7 +292,7 @@ void lp_build_interp_soa_init(struct lp_build_interp_soa_context *bld, const struct tgsi_token *tokens, LLVMBuilderRef builder, - union lp_type type, + struct lp_type type, LLVMValueRef a0_ptr, LLVMValueRef dadx_ptr, LLVMValueRef dady_ptr, diff --git a/src/gallium/drivers/llvmpipe/lp_bld_interp.h b/src/gallium/drivers/llvmpipe/lp_bld_interp.h index 9194f6233a7..9c57a10879b 100644 --- a/src/gallium/drivers/llvmpipe/lp_bld_interp.h +++ b/src/gallium/drivers/llvmpipe/lp_bld_interp.h @@ -83,7 +83,7 @@ void lp_build_interp_soa_init(struct lp_build_interp_soa_context *bld, const struct tgsi_token *tokens, LLVMBuilderRef builder, - union lp_type type, + struct lp_type type, LLVMValueRef a0_ptr, LLVMValueRef dadx_ptr, LLVMValueRef dady_ptr, diff --git a/src/gallium/drivers/llvmpipe/lp_bld_logic.c b/src/gallium/drivers/llvmpipe/lp_bld_logic.c index 8631efd6c3e..6b6f8207697 100644 --- a/src/gallium/drivers/llvmpipe/lp_bld_logic.c +++ b/src/gallium/drivers/llvmpipe/lp_bld_logic.c @@ -45,7 +45,7 @@ lp_build_cmp(struct lp_build_context *bld, LLVMValueRef a, LLVMValueRef b) { - const union lp_type type = bld->type; + const struct lp_type type = bld->type; LLVMTypeRef vec_type = lp_build_vec_type(type); LLVMTypeRef int_vec_type = lp_build_int_vec_type(type); LLVMValueRef zeros = LLVMConstNull(int_vec_type); @@ -301,7 +301,7 @@ lp_build_select(struct lp_build_context *bld, LLVMValueRef a, LLVMValueRef b) { - union lp_type type = bld->type; + struct lp_type type = bld->type; LLVMValueRef res; if(a == b) @@ -313,8 +313,6 @@ lp_build_select(struct lp_build_context *bld, b = LLVMBuildBitCast(bld->builder, b, int_vec_type, ""); } - /* TODO: On SSE4 we could do this with a single instruction -- PBLENDVB */ - a = LLVMBuildAnd(bld->builder, a, mask, ""); /* This often gets translated to PANDN, but sometimes the NOT is @@ -339,9 +337,9 @@ LLVMValueRef lp_build_select_aos(struct lp_build_context *bld, LLVMValueRef a, LLVMValueRef b, - boolean cond[4]) + const boolean cond[4]) { - const union lp_type type = bld->type; + const struct lp_type type = bld->type; const unsigned n = type.length; unsigned i, j; @@ -376,9 +374,9 @@ lp_build_select_aos(struct lp_build_context *bld, return LLVMBuildShuffleVector(bld->builder, a, b, LLVMConstVector(shuffles, n), ""); } + else { #if 0 - else if(0) { - /* FIXME: Unfortunately select of vectors do not work */ + /* XXX: Unfortunately select of vectors do not work */ /* Use a select */ LLVMTypeRef elem_type = LLVMInt1Type(); LLVMValueRef cond[LP_MAX_VECTOR_LENGTH]; @@ -388,10 +386,9 @@ lp_build_select_aos(struct lp_build_context *bld, cond[j + i] = LLVMConstInt(elem_type, cond[i] ? 1 : 0, 0); return LLVMBuildSelect(bld->builder, LLVMConstVector(cond, n), a, b, ""); - } -#endif - else { +#else LLVMValueRef mask = lp_build_const_mask_aos(type, cond); return lp_build_select(bld, mask, a, b); +#endif } } diff --git a/src/gallium/drivers/llvmpipe/lp_bld_logic.h b/src/gallium/drivers/llvmpipe/lp_bld_logic.h index 29b9e1c45b8..a4ee7723b5f 100644 --- a/src/gallium/drivers/llvmpipe/lp_bld_logic.h +++ b/src/gallium/drivers/llvmpipe/lp_bld_logic.h @@ -42,7 +42,7 @@ #include "pipe/p_defines.h" /* For PIPE_FUNC_xxx */ -union lp_type type; +struct lp_type type; struct lp_build_context; @@ -66,7 +66,7 @@ LLVMValueRef lp_build_select_aos(struct lp_build_context *bld, LLVMValueRef a, LLVMValueRef b, - boolean cond[4]); + const boolean cond[4]); #endif /* !LP_BLD_LOGIC_H */ diff --git a/src/gallium/drivers/llvmpipe/lp_bld_sample.h b/src/gallium/drivers/llvmpipe/lp_bld_sample.h index 6f565af76d1..403d0e48367 100644 --- a/src/gallium/drivers/llvmpipe/lp_bld_sample.h +++ b/src/gallium/drivers/llvmpipe/lp_bld_sample.h @@ -40,7 +40,7 @@ struct pipe_texture; struct pipe_sampler_state; -union lp_type; +struct lp_type; /** @@ -123,7 +123,7 @@ void lp_build_sample_soa(LLVMBuilderRef builder, const struct lp_sampler_static_state *static_state, struct lp_sampler_dynamic_state *dynamic_state, - union lp_type fp_type, + struct lp_type fp_type, unsigned unit, unsigned num_coords, const LLVMValueRef *coords, diff --git a/src/gallium/drivers/llvmpipe/lp_bld_sample_soa.c b/src/gallium/drivers/llvmpipe/lp_bld_sample_soa.c index 2913b17d3f8..8ca1be6f1be 100644 --- a/src/gallium/drivers/llvmpipe/lp_bld_sample_soa.c +++ b/src/gallium/drivers/llvmpipe/lp_bld_sample_soa.c @@ -97,15 +97,15 @@ struct lp_build_sample_context const struct util_format_description *format_desc; /** Incoming coordinates type and build context */ - union lp_type coord_type; + struct lp_type coord_type; struct lp_build_context coord_bld; /** Integer coordinates */ - union lp_type int_coord_type; + struct lp_type int_coord_type; struct lp_build_context int_coord_bld; /** Output texels type and build context */ - union lp_type texel_type; + struct lp_type texel_type; struct lp_build_context texel_bld; }; @@ -207,6 +207,12 @@ lp_build_sample_wrap(struct lp_build_sample_context *bld, case PIPE_TEX_WRAP_MIRROR_CLAMP: case PIPE_TEX_WRAP_MIRROR_CLAMP_TO_EDGE: case PIPE_TEX_WRAP_MIRROR_CLAMP_TO_BORDER: + /* FIXME */ + _debug_printf("warning: failed to translate texture wrap mode %u\n", wrap_mode); + coord = lp_build_max(int_coord_bld, coord, int_coord_bld->zero); + coord = lp_build_min(int_coord_bld, coord, length_minus_one); + break; + default: assert(0); } @@ -336,7 +342,7 @@ void lp_build_sample_soa(LLVMBuilderRef builder, const struct lp_sampler_static_state *static_state, struct lp_sampler_dynamic_state *dynamic_state, - union lp_type type, + struct lp_type type, unsigned unit, unsigned num_coords, const LLVMValueRef *coords, @@ -398,7 +404,13 @@ lp_build_sample_soa(LLVMBuilderRef builder, case PIPE_TEX_FILTER_ANISO: lp_build_sample_2d_linear_soa(&bld, s, t, width, height, stride, data_ptr, texel); break; + default: + assert(0); } + /* FIXME: respect static_state->min_mip_filter */; + /* FIXME: respect static_state->mag_img_filter */; + /* FIXME: respect static_state->prefilter */; + lp_build_sample_compare(&bld, p, texel); } diff --git a/src/gallium/drivers/llvmpipe/lp_bld_swizzle.c b/src/gallium/drivers/llvmpipe/lp_bld_swizzle.c index ac7eed9379a..64e81f7b1fe 100644 --- a/src/gallium/drivers/llvmpipe/lp_bld_swizzle.c +++ b/src/gallium/drivers/llvmpipe/lp_bld_swizzle.c @@ -64,7 +64,7 @@ LLVMValueRef lp_build_broadcast_scalar(struct lp_build_context *bld, LLVMValueRef scalar) { - const union lp_type type = bld->type; + const struct lp_type type = bld->type; LLVMValueRef res; unsigned i; @@ -83,7 +83,7 @@ lp_build_broadcast_aos(struct lp_build_context *bld, LLVMValueRef a, unsigned channel) { - const union lp_type type = bld->type; + const struct lp_type type = bld->type; const unsigned n = type.length; unsigned i, j; @@ -115,7 +115,7 @@ lp_build_broadcast_aos(struct lp_build_context *bld, * YY00 YY00 .... YY00 * YYYY YYYY .... YYYY <= output */ - union lp_type type4 = type; + struct lp_type type4 = type; const char shifts[4][2] = { { 1, 2}, {-1, 2}, @@ -161,7 +161,7 @@ lp_build_broadcast_aos(struct lp_build_context *bld, LLVMValueRef lp_build_swizzle1_aos(struct lp_build_context *bld, LLVMValueRef a, - unsigned char swizzle[4]) + const unsigned char swizzle[4]) { const unsigned n = bld->type.length; unsigned i, j; @@ -192,7 +192,7 @@ LLVMValueRef lp_build_swizzle2_aos(struct lp_build_context *bld, LLVMValueRef a, LLVMValueRef b, - unsigned char swizzle[4]) + const unsigned char swizzle[4]) { const unsigned n = bld->type.length; unsigned i, j; @@ -201,11 +201,12 @@ lp_build_swizzle2_aos(struct lp_build_context *bld, return lp_build_swizzle1_aos(bld, a, swizzle); if(a == b) { - swizzle[0] %= 4; - swizzle[1] %= 4; - swizzle[2] %= 4; - swizzle[3] %= 4; - return lp_build_swizzle1_aos(bld, a, swizzle); + unsigned char swizzle1[4]; + swizzle1[0] = swizzle[0] % 4; + swizzle1[1] = swizzle[1] % 4; + swizzle1[2] = swizzle[2] % 4; + swizzle1[3] = swizzle[3] % 4; + return lp_build_swizzle1_aos(bld, a, swizzle1); } if(swizzle[0] % 4 == 0 && diff --git a/src/gallium/drivers/llvmpipe/lp_bld_swizzle.h b/src/gallium/drivers/llvmpipe/lp_bld_swizzle.h index d7dd6a8a604..1f6da804488 100644 --- a/src/gallium/drivers/llvmpipe/lp_bld_swizzle.h +++ b/src/gallium/drivers/llvmpipe/lp_bld_swizzle.h @@ -40,7 +40,7 @@ #include <llvm-c/Core.h> -union lp_type type; +struct lp_type type; struct lp_build_context; @@ -73,7 +73,7 @@ lp_build_broadcast_aos(struct lp_build_context *bld, LLVMValueRef lp_build_swizzle1_aos(struct lp_build_context *bld, LLVMValueRef a, - unsigned char swizzle[4]); + const unsigned char swizzle[4]); /** @@ -85,7 +85,7 @@ LLVMValueRef lp_build_swizzle2_aos(struct lp_build_context *bld, LLVMValueRef a, LLVMValueRef b, - unsigned char swizzle[4]); + const unsigned char swizzle[4]); #endif /* !LP_BLD_SWIZZLE_H */ diff --git a/src/gallium/drivers/llvmpipe/lp_bld_tgsi.h b/src/gallium/drivers/llvmpipe/lp_bld_tgsi.h index 10c251c4162..eddb7a83fa2 100644 --- a/src/gallium/drivers/llvmpipe/lp_bld_tgsi.h +++ b/src/gallium/drivers/llvmpipe/lp_bld_tgsi.h @@ -39,7 +39,7 @@ struct tgsi_token; -union lp_type; +struct lp_type; struct lp_build_context; struct lp_build_mask_context; @@ -60,7 +60,7 @@ struct lp_build_sampler_soa void (*emit_fetch_texel)( struct lp_build_sampler_soa *sampler, LLVMBuilderRef builder, - union lp_type type, + struct lp_type type, unsigned unit, unsigned num_coords, const LLVMValueRef *coords, @@ -72,7 +72,7 @@ struct lp_build_sampler_soa void lp_build_tgsi_soa(LLVMBuilderRef builder, const struct tgsi_token *tokens, - union lp_type type, + struct lp_type type, struct lp_build_mask_context *mask, LLVMValueRef consts_ptr, const LLVMValueRef *pos, diff --git a/src/gallium/drivers/llvmpipe/lp_bld_tgsi_soa.c b/src/gallium/drivers/llvmpipe/lp_bld_tgsi_soa.c index 3ce379de12f..adc81569ed5 100644 --- a/src/gallium/drivers/llvmpipe/lp_bld_tgsi_soa.c +++ b/src/gallium/drivers/llvmpipe/lp_bld_tgsi_soa.c @@ -78,6 +78,11 @@ #define CHAN_Z 2 #define CHAN_W 3 +#define QUAD_TOP_LEFT 0 +#define QUAD_TOP_RIGHT 1 +#define QUAD_BOTTOM_LEFT 2 +#define QUAD_BOTTOM_RIGHT 3 + struct lp_build_tgsi_soa_context { @@ -97,6 +102,51 @@ struct lp_build_tgsi_soa_context }; +static const unsigned char +swizzle_left[4] = { + QUAD_TOP_LEFT, QUAD_TOP_LEFT, + QUAD_BOTTOM_LEFT, QUAD_BOTTOM_LEFT +}; + +static const unsigned char +swizzle_right[4] = { + QUAD_TOP_RIGHT, QUAD_TOP_RIGHT, + QUAD_BOTTOM_RIGHT, QUAD_BOTTOM_RIGHT +}; + +static const unsigned char +swizzle_top[4] = { + QUAD_TOP_LEFT, QUAD_TOP_RIGHT, + QUAD_TOP_LEFT, QUAD_TOP_RIGHT +}; + +static const unsigned char +swizzle_bottom[4] = { + QUAD_BOTTOM_LEFT, QUAD_BOTTOM_RIGHT, + QUAD_BOTTOM_LEFT, QUAD_BOTTOM_RIGHT +}; + + +static LLVMValueRef +emit_ddx(struct lp_build_tgsi_soa_context *bld, + LLVMValueRef src) +{ + LLVMValueRef src_left = lp_build_swizzle1_aos(&bld->base, src, swizzle_left); + LLVMValueRef src_right = lp_build_swizzle1_aos(&bld->base, src, swizzle_right); + return lp_build_sub(&bld->base, src_right, src_left); +} + + +static LLVMValueRef +emit_ddy(struct lp_build_tgsi_soa_context *bld, + LLVMValueRef src) +{ + LLVMValueRef src_top = lp_build_swizzle1_aos(&bld->base, src, swizzle_top); + LLVMValueRef src_bottom = lp_build_swizzle1_aos(&bld->base, src, swizzle_bottom); + return lp_build_sub(&bld->base, src_top, src_bottom); +} + + /** * Register fetch. */ @@ -167,6 +217,7 @@ emit_fetch( break; case TGSI_UTIL_SIGN_SET: + /* TODO: Use bitwese OR for floating point */ res = lp_build_abs( &bld->base, res ); res = LLVMBuildNeg( bld->base.builder, res, "" ); break; @@ -184,6 +235,36 @@ emit_fetch( /** + * Register fetch with derivatives. + */ +static void +emit_fetch_deriv( + struct lp_build_tgsi_soa_context *bld, + const struct tgsi_full_instruction *inst, + unsigned index, + const unsigned chan_index, + LLVMValueRef *res, + LLVMValueRef *ddx, + LLVMValueRef *ddy) +{ + LLVMValueRef src; + + src = emit_fetch(bld, inst, index, chan_index); + + if(res) + *res = src; + + /* TODO: use interpolation coeffs for inputs */ + + if(ddx) + *ddx = emit_ddx(bld, src); + + if(ddy) + *ddy = emit_ddy(bld, src); +} + + +/** * Register store. */ static void @@ -238,17 +319,18 @@ emit_store( * High-level instruction translators. */ + static void emit_tex( struct lp_build_tgsi_soa_context *bld, const struct tgsi_full_instruction *inst, boolean apply_lodbias, - boolean projected) + boolean projected, + LLVMValueRef *texel) { const uint unit = inst->FullSrcRegisters[1].SrcRegister.Index; LLVMValueRef lodbias; LLVMValueRef oow; LLVMValueRef coords[3]; - LLVMValueRef texel[4]; unsigned num_coords; unsigned i; @@ -293,10 +375,6 @@ emit_tex( struct lp_build_tgsi_soa_context *bld, bld->base.type, unit, num_coords, coords, lodbias, texel); - - FOR_EACH_DST0_ENABLED_CHANNEL( inst, i ) { - emit_store( bld, inst, 0, i, texel[i] ); - } } @@ -349,14 +427,6 @@ emit_kil( } -static void -emit_kilp( - struct lp_build_tgsi_soa_context *bld ) -{ - /* XXX todo / fix me */ -} - - /** * Check if inst src/dest regs use indirect addressing into temporary * register file. @@ -384,25 +454,35 @@ indirect_temp_reference(const struct tgsi_full_instruction *inst) static int emit_instruction( struct lp_build_tgsi_soa_context *bld, - struct tgsi_full_instruction *inst ) + const struct tgsi_full_instruction *inst, + const struct tgsi_opcode_info *info) { unsigned chan_index; LLVMValueRef src0, src1, src2; LLVMValueRef tmp0, tmp1, tmp2, tmp3, tmp4, tmp5, tmp6, tmp7; - LLVMValueRef dst0; + LLVMValueRef res; + LLVMValueRef dst0[NUM_CHANNELS]; /* we can't handle indirect addressing into temp register file yet */ if (indirect_temp_reference(inst)) return FALSE; + assert(info->num_dst <= 1); + if(info->num_dst) { + FOR_EACH_DST0_ENABLED_CHANNEL( inst, chan_index ) { + dst0[chan_index] = bld->base.undef; + } + } + switch (inst->Instruction.Opcode) { #if 0 case TGSI_OPCODE_ARL: + /* FIXME */ FOR_EACH_DST0_ENABLED_CHANNEL( inst, chan_index ) { tmp0 = emit_fetch( bld, inst, 0, chan_index ); emit_flr(bld, 0, 0); emit_f2it( bld, 0 ); - emit_store( bld, inst, 0, chan_index, tmp0); + dst0[chan_index] = tmp0; } break; #endif @@ -410,19 +490,17 @@ emit_instruction( case TGSI_OPCODE_MOV: case TGSI_OPCODE_SWZ: FOR_EACH_DST0_ENABLED_CHANNEL( inst, chan_index ) { - tmp0 = emit_fetch( bld, inst, 0, chan_index ); - emit_store( bld, inst, 0, chan_index, tmp0); + dst0[chan_index] = emit_fetch( bld, inst, 0, chan_index ); } break; case TGSI_OPCODE_LIT: if( IS_DST0_CHANNEL_ENABLED( inst, CHAN_X ) ) { - emit_store( bld, inst, 0, CHAN_X, bld->base.one); + dst0[CHAN_X] = bld->base.one; } if( IS_DST0_CHANNEL_ENABLED( inst, CHAN_Y ) ) { src0 = emit_fetch( bld, inst, 0, CHAN_X ); - dst0 = lp_build_max( &bld->base, src0, bld->base.zero); - emit_store( bld, inst, 0, CHAN_Y, dst0); + dst0[CHAN_Y] = lp_build_max( &bld->base, src0, bld->base.zero); } if( IS_DST0_CHANNEL_ENABLED( inst, CHAN_Z ) ) { /* XMM[1] = SrcReg[0].yyyy */ @@ -434,20 +512,19 @@ emit_instruction( tmp1 = lp_build_pow( &bld->base, tmp1, tmp2); tmp0 = emit_fetch( bld, inst, 0, CHAN_X ); tmp2 = lp_build_cmp(&bld->base, PIPE_FUNC_GREATER, tmp0, bld->base.zero); - dst0 = lp_build_select(&bld->base, tmp2, tmp1, bld->base.zero); - emit_store( bld, inst, 0, CHAN_Z, dst0); + dst0[CHAN_Z] = lp_build_select(&bld->base, tmp2, tmp1, bld->base.zero); } if( IS_DST0_CHANNEL_ENABLED( inst, CHAN_W ) ) { - emit_store( bld, inst, 0, CHAN_W, bld->base.one); + dst0[CHAN_W] = bld->base.one; } break; case TGSI_OPCODE_RCP: /* TGSI_OPCODE_RECIP */ src0 = emit_fetch( bld, inst, 0, CHAN_X ); - dst0 = lp_build_rcp(&bld->base, src0); + res = lp_build_rcp(&bld->base, src0); FOR_EACH_DST0_ENABLED_CHANNEL( inst, chan_index ) { - emit_store( bld, inst, 0, chan_index, dst0 ); + dst0[chan_index] = res; } break; @@ -455,9 +532,9 @@ emit_instruction( /* TGSI_OPCODE_RECIPSQRT */ src0 = emit_fetch( bld, inst, 0, CHAN_X ); src0 = lp_build_abs(&bld->base, src0); - dst0 = lp_build_rsqrt(&bld->base, src0); + res = lp_build_rsqrt(&bld->base, src0); FOR_EACH_DST0_ENABLED_CHANNEL( inst, chan_index ) { - emit_store( bld, inst, 0, chan_index, dst0 ); + dst0[chan_index] = res; } break; @@ -481,16 +558,15 @@ emit_instruction( lp_build_exp2_approx(&bld->base, src0, p_exp2_int_part, p_frac_part, p_exp2); if (IS_DST0_CHANNEL_ENABLED( inst, CHAN_X )) - emit_store( bld, inst, 0, CHAN_X, tmp0); + dst0[CHAN_X] = tmp0; if (IS_DST0_CHANNEL_ENABLED( inst, CHAN_Y )) - emit_store( bld, inst, 0, CHAN_Y, tmp1); + dst0[CHAN_Y] = tmp1; if (IS_DST0_CHANNEL_ENABLED( inst, CHAN_Z )) - emit_store( bld, inst, 0, CHAN_Z, tmp2); + dst0[CHAN_Z] = tmp2; } /* dst.w = 1.0 */ if (IS_DST0_CHANNEL_ENABLED( inst, CHAN_W )) { - tmp0 = bld->base.one; - emit_store( bld, inst, 0, CHAN_W, tmp0); + dst0[CHAN_W] = bld->base.one; } break; @@ -516,20 +592,18 @@ emit_instruction( /* dst.x = floor(lg2(abs(src.x))) */ if (IS_DST0_CHANNEL_ENABLED( inst, CHAN_X )) - emit_store( bld, inst, 0, CHAN_X, tmp0); + dst0[CHAN_X] = tmp0; /* dst.y = abs(src)/ex2(floor(lg2(abs(src.x)))) */ if (IS_DST0_CHANNEL_ENABLED( inst, CHAN_Y )) { - tmp1 = lp_build_div( &bld->base, src0, tmp1); - emit_store( bld, inst, 0, CHAN_Y, tmp1); + dst0[CHAN_Y] = lp_build_div( &bld->base, src0, tmp1); } /* dst.z = lg2(abs(src.x)) */ if (IS_DST0_CHANNEL_ENABLED( inst, CHAN_Z )) - emit_store( bld, inst, 0, CHAN_Z, tmp2); + dst0[CHAN_Z] = tmp2; } /* dst.w = 1.0 */ if (IS_DST0_CHANNEL_ENABLED( inst, CHAN_W )) { - tmp0 = bld->base.one; - emit_store( bld, inst, 0, CHAN_W, tmp0); + dst0[CHAN_W] = bld->base.one; } break; @@ -537,8 +611,7 @@ emit_instruction( FOR_EACH_DST0_ENABLED_CHANNEL( inst, chan_index ) { src0 = emit_fetch( bld, inst, 0, chan_index ); src1 = emit_fetch( bld, inst, 1, chan_index ); - dst0 = lp_build_mul(&bld->base, src0, src1); - emit_store( bld, inst, 0, chan_index, dst0); + dst0[chan_index] = lp_build_mul(&bld->base, src0, src1); } break; @@ -546,8 +619,7 @@ emit_instruction( FOR_EACH_DST0_ENABLED_CHANNEL( inst, chan_index ) { src0 = emit_fetch( bld, inst, 0, chan_index ); src1 = emit_fetch( bld, inst, 1, chan_index ); - dst0 = lp_build_add(&bld->base, src0, src1); - emit_store( bld, inst, 0, chan_index, dst0); + dst0[chan_index] = lp_build_add(&bld->base, src0, src1); } break; @@ -565,7 +637,7 @@ emit_instruction( tmp1 = lp_build_mul( &bld->base, tmp1, tmp2); tmp0 = lp_build_add( &bld->base, tmp0, tmp1); FOR_EACH_DST0_ENABLED_CHANNEL( inst, chan_index ) { - emit_store( bld, inst, 0, chan_index, tmp0); + dst0[chan_index] = tmp0; } break; @@ -587,28 +659,24 @@ emit_instruction( tmp1 = lp_build_mul( &bld->base, tmp1, tmp2); tmp0 = lp_build_add( &bld->base, tmp0, tmp1); FOR_EACH_DST0_ENABLED_CHANNEL( inst, chan_index ) { - emit_store( bld, inst, 0, chan_index, tmp0); + dst0[chan_index] = tmp0; } break; case TGSI_OPCODE_DST: IF_IS_DST0_CHANNEL_ENABLED( inst, CHAN_X ) { - tmp0 = bld->base.one; - emit_store( bld, inst, 0, CHAN_X, tmp0); + dst0[CHAN_X] = bld->base.one; } IF_IS_DST0_CHANNEL_ENABLED( inst, CHAN_Y ) { tmp0 = emit_fetch( bld, inst, 0, CHAN_Y ); tmp1 = emit_fetch( bld, inst, 1, CHAN_Y ); - tmp0 = lp_build_mul( &bld->base, tmp0, tmp1); - emit_store( bld, inst, 0, CHAN_Y, tmp0); + dst0[CHAN_Y] = lp_build_mul( &bld->base, tmp0, tmp1); } IF_IS_DST0_CHANNEL_ENABLED( inst, CHAN_Z ) { - tmp0 = emit_fetch( bld, inst, 0, CHAN_Z ); - emit_store( bld, inst, 0, CHAN_Z, tmp0); + dst0[CHAN_Z] = emit_fetch( bld, inst, 0, CHAN_Z ); } IF_IS_DST0_CHANNEL_ENABLED( inst, CHAN_W ) { - tmp0 = emit_fetch( bld, inst, 1, CHAN_W ); - emit_store( bld, inst, 0, CHAN_W, tmp0); + dst0[CHAN_W] = emit_fetch( bld, inst, 1, CHAN_W ); } break; @@ -616,8 +684,7 @@ emit_instruction( FOR_EACH_DST0_ENABLED_CHANNEL( inst, chan_index ) { src0 = emit_fetch( bld, inst, 0, chan_index ); src1 = emit_fetch( bld, inst, 1, chan_index ); - dst0 = lp_build_min( &bld->base, src0, src1 ); - emit_store( bld, inst, 0, chan_index, dst0); + dst0[chan_index] = lp_build_min( &bld->base, src0, src1 ); } break; @@ -625,8 +692,7 @@ emit_instruction( FOR_EACH_DST0_ENABLED_CHANNEL( inst, chan_index ) { src0 = emit_fetch( bld, inst, 0, chan_index ); src1 = emit_fetch( bld, inst, 1, chan_index ); - dst0 = lp_build_max( &bld->base, src0, src1 ); - emit_store( bld, inst, 0, chan_index, dst0); + dst0[chan_index] = lp_build_max( &bld->base, src0, src1 ); } break; @@ -636,8 +702,7 @@ emit_instruction( src0 = emit_fetch( bld, inst, 0, chan_index ); src1 = emit_fetch( bld, inst, 1, chan_index ); tmp0 = lp_build_cmp( &bld->base, PIPE_FUNC_LESS, src0, src1 ); - dst0 = lp_build_select( &bld->base, tmp0, bld->base.one, bld->base.zero ); - emit_store( bld, inst, 0, chan_index, dst0); + dst0[chan_index] = lp_build_select( &bld->base, tmp0, bld->base.one, bld->base.zero ); } break; @@ -647,8 +712,7 @@ emit_instruction( src0 = emit_fetch( bld, inst, 0, chan_index ); src1 = emit_fetch( bld, inst, 1, chan_index ); tmp0 = lp_build_cmp( &bld->base, PIPE_FUNC_GEQUAL, src0, src1 ); - dst0 = lp_build_select( &bld->base, tmp0, bld->base.one, bld->base.zero ); - emit_store( bld, inst, 0, chan_index, dst0); + dst0[chan_index] = lp_build_select( &bld->base, tmp0, bld->base.one, bld->base.zero ); } break; @@ -660,7 +724,7 @@ emit_instruction( tmp2 = emit_fetch( bld, inst, 2, chan_index ); tmp0 = lp_build_mul( &bld->base, tmp0, tmp1); tmp0 = lp_build_add( &bld->base, tmp0, tmp2); - emit_store( bld, inst, 0, chan_index, tmp0); + dst0[chan_index] = tmp0; } break; @@ -668,8 +732,7 @@ emit_instruction( FOR_EACH_DST0_ENABLED_CHANNEL( inst, chan_index ) { tmp0 = emit_fetch( bld, inst, 0, chan_index ); tmp1 = emit_fetch( bld, inst, 1, chan_index ); - tmp0 = lp_build_sub( &bld->base, tmp0, tmp1); - emit_store( bld, inst, 0, chan_index, tmp0); + dst0[chan_index] = lp_build_sub( &bld->base, tmp0, tmp1); } break; @@ -680,13 +743,19 @@ emit_instruction( src2 = emit_fetch( bld, inst, 2, chan_index ); tmp0 = lp_build_sub( &bld->base, src1, src2 ); tmp0 = lp_build_mul( &bld->base, src0, tmp0 ); - dst0 = lp_build_add( &bld->base, tmp0, src2 ); - emit_store( bld, inst, 0, chan_index, dst0 ); + dst0[chan_index] = lp_build_add( &bld->base, tmp0, src2 ); } break; case TGSI_OPCODE_CND: - return 0; + FOR_EACH_DST0_ENABLED_CHANNEL( inst, chan_index ) { + src0 = emit_fetch( bld, inst, 0, chan_index ); + src1 = emit_fetch( bld, inst, 1, chan_index ); + src2 = emit_fetch( bld, inst, 2, chan_index ); + tmp1 = lp_build_const_scalar(bld->base.type, 0.5); + tmp0 = lp_build_cmp( &bld->base, PIPE_FUNC_GREATER, src2, tmp1); + dst0[chan_index] = lp_build_select( &bld->base, tmp0, src0, src1 ); + } break; case TGSI_OPCODE_DP2A: @@ -700,45 +769,49 @@ emit_instruction( tmp1 = emit_fetch( bld, inst, 2, CHAN_X ); /* xmm1 = src[2].x */ tmp0 = lp_build_add( &bld->base, tmp0, tmp1); /* xmm0 = xmm0 + xmm1 */ FOR_EACH_DST0_ENABLED_CHANNEL( inst, chan_index ) { - emit_store( bld, inst, 0, chan_index, tmp0); /* dest[ch] = xmm0 */ + dst0[chan_index] = tmp0; /* dest[ch] = xmm0 */ } break; -#if 0 case TGSI_OPCODE_FRC: FOR_EACH_DST0_ENABLED_CHANNEL( inst, chan_index ) { - tmp0 = emit_fetch( bld, inst, 0, chan_index ); - emit_frc( bld, 0, 0 ); - emit_store( bld, inst, 0, chan_index, tmp0); + src0 = emit_fetch( bld, inst, 0, chan_index ); + tmp0 = lp_build_floor(&bld->base, src0); + tmp0 = lp_build_sub(&bld->base, tmp0, src0); + dst0[chan_index] = tmp0; } break; case TGSI_OPCODE_CLAMP: - return 0; + FOR_EACH_DST0_ENABLED_CHANNEL( inst, chan_index ) { + tmp0 = emit_fetch( bld, inst, 0, chan_index ); + src1 = emit_fetch( bld, inst, 1, chan_index ); + src2 = emit_fetch( bld, inst, 2, chan_index ); + tmp0 = lp_build_max(&bld->base, tmp0, src1); + tmp0 = lp_build_min(&bld->base, tmp0, src2); + dst0[chan_index] = tmp0; + } break; case TGSI_OPCODE_FLR: FOR_EACH_DST0_ENABLED_CHANNEL( inst, chan_index ) { tmp0 = emit_fetch( bld, inst, 0, chan_index ); - emit_flr( bld, 0, 0 ); - emit_store( bld, inst, 0, chan_index, tmp0); + dst0[chan_index] = lp_build_floor(&bld->base, tmp0); } break; case TGSI_OPCODE_ROUND: FOR_EACH_DST0_ENABLED_CHANNEL( inst, chan_index ) { tmp0 = emit_fetch( bld, inst, 0, chan_index ); - emit_rnd( bld, 0, 0 ); - emit_store( bld, inst, 0, chan_index, tmp0); + dst0[chan_index] = lp_build_round(&bld->base, tmp0); } break; -#endif case TGSI_OPCODE_EX2: { tmp0 = emit_fetch( bld, inst, 0, CHAN_X ); tmp0 = lp_build_exp2( &bld->base, tmp0); FOR_EACH_DST0_ENABLED_CHANNEL( inst, chan_index ) { - emit_store( bld, inst, 0, chan_index, tmp0); + dst0[chan_index] = tmp0; } break; } @@ -747,16 +820,16 @@ emit_instruction( tmp0 = emit_fetch( bld, inst, 0, CHAN_X ); tmp0 = lp_build_log2( &bld->base, tmp0); FOR_EACH_DST0_ENABLED_CHANNEL( inst, chan_index ) { - emit_store( bld, inst, 0, chan_index, tmp0); + dst0[chan_index] = tmp0; } break; case TGSI_OPCODE_POW: src0 = emit_fetch( bld, inst, 0, CHAN_X ); src1 = emit_fetch( bld, inst, 1, CHAN_X ); - dst0 = lp_build_pow( &bld->base, src0, src1 ); + res = lp_build_pow( &bld->base, src0, src1 ); FOR_EACH_DST0_ENABLED_CHANNEL( inst, chan_index ) { - emit_store( bld, inst, 0, chan_index, dst0 ); + dst0[chan_index] = res; } break; @@ -777,7 +850,7 @@ emit_instruction( tmp5 = tmp3; tmp5 = lp_build_mul( &bld->base, tmp5, tmp4); tmp2 = lp_build_sub( &bld->base, tmp2, tmp5); - emit_store( bld, inst, 0, CHAN_X, tmp2); + dst0[CHAN_X] = tmp2; } if( IS_DST0_CHANNEL_ENABLED( inst, CHAN_Y ) || IS_DST0_CHANNEL_ENABLED( inst, CHAN_Z ) ) { @@ -788,31 +861,30 @@ emit_instruction( tmp3 = lp_build_mul( &bld->base, tmp3, tmp2); tmp1 = lp_build_mul( &bld->base, tmp1, tmp5); tmp3 = lp_build_sub( &bld->base, tmp3, tmp1); - emit_store( bld, inst, 0, CHAN_Y, tmp3); + dst0[CHAN_Y] = tmp3; } IF_IS_DST0_CHANNEL_ENABLED( inst, CHAN_Z ) { tmp5 = lp_build_mul( &bld->base, tmp5, tmp4); tmp0 = lp_build_mul( &bld->base, tmp0, tmp2); tmp5 = lp_build_sub( &bld->base, tmp5, tmp0); - emit_store( bld, inst, 0, CHAN_Z, tmp5); + dst0[CHAN_Z] = tmp5; } IF_IS_DST0_CHANNEL_ENABLED( inst, CHAN_W ) { - tmp0 = bld->base.one; - emit_store( bld, inst, 0, CHAN_W, tmp0); + dst0[CHAN_W] = bld->base.one; } break; case TGSI_OPCODE_ABS: FOR_EACH_DST0_ENABLED_CHANNEL( inst, chan_index ) { tmp0 = emit_fetch( bld, inst, 0, chan_index ); - tmp0 = lp_build_abs( &bld->base, tmp0 ) ; - emit_store( bld, inst, 0, chan_index, tmp0); + dst0[chan_index] = lp_build_abs( &bld->base, tmp0 ); } break; case TGSI_OPCODE_RCC: + /* deprecated? */ + assert(0); return 0; - break; case TGSI_OPCODE_DPH: tmp0 = emit_fetch( bld, inst, 0, CHAN_X ); @@ -829,7 +901,7 @@ emit_instruction( tmp1 = emit_fetch( bld, inst, 1, CHAN_W ); tmp0 = lp_build_add( &bld->base, tmp0, tmp1); FOR_EACH_DST0_ENABLED_CHANNEL( inst, chan_index ) { - emit_store( bld, inst, 0, chan_index, tmp0); + dst0[chan_index] = tmp0; } break; @@ -837,25 +909,27 @@ emit_instruction( tmp0 = emit_fetch( bld, inst, 0, CHAN_X ); tmp0 = lp_build_cos( &bld->base, tmp0 ); FOR_EACH_DST0_ENABLED_CHANNEL( inst, chan_index ) { - emit_store( bld, inst, 0, chan_index, tmp0); + dst0[chan_index] = tmp0; } break; case TGSI_OPCODE_DDX: - return 0; + FOR_EACH_DST0_ENABLED_CHANNEL( inst, chan_index ) { + emit_fetch_deriv( bld, inst, 0, chan_index, NULL, &dst0[chan_index], NULL); + } break; case TGSI_OPCODE_DDY: - return 0; + FOR_EACH_DST0_ENABLED_CHANNEL( inst, chan_index ) { + emit_fetch_deriv( bld, inst, 0, chan_index, NULL, NULL, &dst0[chan_index]); + } break; -#if 0 case TGSI_OPCODE_KILP: /* predicated kill */ - emit_kilp( bld ); - return 0; /* XXX fix me */ + /* FIXME */ + return 0; break; -#endif case TGSI_OPCODE_KIL: /* conditional kill */ @@ -887,13 +961,14 @@ emit_instruction( src0 = emit_fetch( bld, inst, 0, chan_index ); src1 = emit_fetch( bld, inst, 1, chan_index ); tmp0 = lp_build_cmp( &bld->base, PIPE_FUNC_EQUAL, src0, src1 ); - dst0 = lp_build_select( &bld->base, tmp0, bld->base.one, bld->base.zero ); - emit_store( bld, inst, 0, chan_index, dst0); + dst0[chan_index] = lp_build_select( &bld->base, tmp0, bld->base.one, bld->base.zero ); } break; case TGSI_OPCODE_SFL: - return 0; + FOR_EACH_DST0_ENABLED_CHANNEL( inst, chan_index ) { + dst0[chan_index] = bld->base.zero; + } break; case TGSI_OPCODE_SGT: @@ -901,8 +976,7 @@ emit_instruction( src0 = emit_fetch( bld, inst, 0, chan_index ); src1 = emit_fetch( bld, inst, 1, chan_index ); tmp0 = lp_build_cmp( &bld->base, PIPE_FUNC_GREATER, src0, src1 ); - dst0 = lp_build_select( &bld->base, tmp0, bld->base.one, bld->base.zero ); - emit_store( bld, inst, 0, chan_index, dst0); + dst0[chan_index] = lp_build_select( &bld->base, tmp0, bld->base.one, bld->base.zero ); } break; @@ -910,7 +984,7 @@ emit_instruction( tmp0 = emit_fetch( bld, inst, 0, CHAN_X ); tmp0 = lp_build_sin( &bld->base, tmp0 ); FOR_EACH_DST0_ENABLED_CHANNEL( inst, chan_index ) { - emit_store( bld, inst, 0, chan_index, tmp0); + dst0[chan_index] = tmp0; } break; @@ -919,8 +993,7 @@ emit_instruction( src0 = emit_fetch( bld, inst, 0, chan_index ); src1 = emit_fetch( bld, inst, 1, chan_index ); tmp0 = lp_build_cmp( &bld->base, PIPE_FUNC_LEQUAL, src0, src1 ); - dst0 = lp_build_select( &bld->base, tmp0, bld->base.one, bld->base.zero ); - emit_store( bld, inst, 0, chan_index, dst0); + dst0[chan_index] = lp_build_select( &bld->base, tmp0, bld->base.one, bld->base.zero ); } break; @@ -929,85 +1002,99 @@ emit_instruction( src0 = emit_fetch( bld, inst, 0, chan_index ); src1 = emit_fetch( bld, inst, 1, chan_index ); tmp0 = lp_build_cmp( &bld->base, PIPE_FUNC_NOTEQUAL, src0, src1 ); - dst0 = lp_build_select( &bld->base, tmp0, bld->base.one, bld->base.zero ); - emit_store( bld, inst, 0, chan_index, dst0); + dst0[chan_index] = lp_build_select( &bld->base, tmp0, bld->base.one, bld->base.zero ); } break; case TGSI_OPCODE_STR: - return 0; + FOR_EACH_DST0_ENABLED_CHANNEL( inst, chan_index ) { + dst0[chan_index] = bld->base.one; + } break; case TGSI_OPCODE_TEX: - emit_tex( bld, inst, FALSE, FALSE ); + emit_tex( bld, inst, FALSE, FALSE, dst0 ); break; case TGSI_OPCODE_TXD: + /* FIXME */ return 0; break; case TGSI_OPCODE_UP2H: + /* deprecated */ + assert (0); return 0; break; case TGSI_OPCODE_UP2US: + /* deprecated */ + assert(0); return 0; break; case TGSI_OPCODE_UP4B: + /* deprecated */ + assert(0); return 0; break; case TGSI_OPCODE_UP4UB: + /* deprecated */ + assert(0); return 0; break; case TGSI_OPCODE_X2D: + /* deprecated? */ + assert(0); return 0; break; case TGSI_OPCODE_ARA: + /* deprecated */ + assert(0); return 0; break; #if 0 case TGSI_OPCODE_ARR: + /* FIXME */ FOR_EACH_DST0_ENABLED_CHANNEL( inst, chan_index ) { tmp0 = emit_fetch( bld, inst, 0, chan_index ); emit_rnd( bld, 0, 0 ); emit_f2it( bld, 0 ); - emit_store( bld, inst, 0, chan_index, tmp0); + dst0[chan_index] = tmp0; } break; #endif case TGSI_OPCODE_BRA: + /* deprecated */ + assert(0); return 0; break; case TGSI_OPCODE_CAL: + /* FIXME */ return 0; break; -#if 0 case TGSI_OPCODE_RET: - emit_ret( bld ); + /* FIXME */ + return 0; break; -#endif case TGSI_OPCODE_END: break; -#if 0 case TGSI_OPCODE_SSG: /* TGSI_OPCODE_SGN */ FOR_EACH_DST0_ENABLED_CHANNEL( inst, chan_index ) { tmp0 = emit_fetch( bld, inst, 0, chan_index ); - emit_sgn( bld, 0, 0 ); - emit_store( bld, inst, 0, chan_index, tmp0); + dst0[chan_index] = lp_build_sgn( &bld->base, tmp0 ); } break; -#endif case TGSI_OPCODE_CMP: FOR_EACH_DST0_ENABLED_CHANNEL( inst, chan_index ) { @@ -1015,34 +1102,29 @@ emit_instruction( src1 = emit_fetch( bld, inst, 1, chan_index ); src2 = emit_fetch( bld, inst, 2, chan_index ); tmp0 = lp_build_cmp( &bld->base, PIPE_FUNC_LESS, src0, bld->base.zero ); - dst0 = lp_build_select( &bld->base, tmp0, src1, src2); - emit_store( bld, inst, 0, chan_index, dst0); + dst0[chan_index] = lp_build_select( &bld->base, tmp0, src1, src2); } break; case TGSI_OPCODE_SCS: IF_IS_DST0_CHANNEL_ENABLED( inst, CHAN_X ) { tmp0 = emit_fetch( bld, inst, 0, CHAN_X ); - tmp0 = lp_build_cos( &bld->base, tmp0 ); - emit_store( bld, inst, 0, CHAN_X, tmp0); + dst0[CHAN_X] = lp_build_cos( &bld->base, tmp0 ); } IF_IS_DST0_CHANNEL_ENABLED( inst, CHAN_Y ) { tmp0 = emit_fetch( bld, inst, 0, CHAN_X ); - tmp0 = lp_build_sin( &bld->base, tmp0 ); - emit_store( bld, inst, 0, CHAN_Y, tmp0); + dst0[CHAN_Y] = lp_build_sin( &bld->base, tmp0 ); } IF_IS_DST0_CHANNEL_ENABLED( inst, CHAN_Z ) { - tmp0 = bld->base.zero; - emit_store( bld, inst, 0, CHAN_Z, tmp0); + dst0[CHAN_Z] = bld->base.zero; } IF_IS_DST0_CHANNEL_ENABLED( inst, CHAN_W ) { - tmp0 = bld->base.one; - emit_store( bld, inst, 0, CHAN_W, tmp0); + dst0[CHAN_W] = bld->base.one; } break; case TGSI_OPCODE_TXB: - emit_tex( bld, inst, TRUE, FALSE ); + emit_tex( bld, inst, TRUE, FALSE, dst0 ); break; case TGSI_OPCODE_NRM: @@ -1101,38 +1183,35 @@ emit_instruction( /* dst.x = xmm1 * src.x */ if (IS_DST0_CHANNEL_ENABLED(inst, CHAN_X)) { - tmp4 = lp_build_mul( &bld->base, tmp4, tmp1); - emit_store(bld, inst, 0, CHAN_X, tmp4); + dst0[CHAN_X] = lp_build_mul( &bld->base, tmp4, tmp1); } /* dst.y = xmm1 * src.y */ if (IS_DST0_CHANNEL_ENABLED(inst, CHAN_Y)) { - tmp5 = lp_build_mul( &bld->base, tmp5, tmp1); - emit_store(bld, inst, 0, CHAN_Y, tmp5); + dst0[CHAN_Y] = lp_build_mul( &bld->base, tmp5, tmp1); } /* dst.z = xmm1 * src.z */ if (IS_DST0_CHANNEL_ENABLED(inst, CHAN_Z)) { - tmp6 = lp_build_mul( &bld->base, tmp6, tmp1); - emit_store(bld, inst, 0, CHAN_Z, tmp6); + dst0[CHAN_Z] = lp_build_mul( &bld->base, tmp6, tmp1); } /* dst.w = xmm1 * src.w */ if (IS_DST0_CHANNEL_ENABLED(inst, CHAN_X) && dims == 4) { - tmp7 = lp_build_mul( &bld->base, tmp7, tmp1); - emit_store(bld, inst, 0, CHAN_W, tmp7); + dst0[CHAN_W] = lp_build_mul( &bld->base, tmp7, tmp1); } } - /* dst0.w = 1.0 */ + /* dst.w = 1.0 */ if (IS_DST0_CHANNEL_ENABLED(inst, CHAN_W) && dims == 3) { - tmp0 = bld->base.one; - emit_store(bld, inst, 0, CHAN_W, tmp0); + dst0[CHAN_W] = bld->base.one; } } break; case TGSI_OPCODE_DIV: + /* deprecated */ + assert( 0 ); return 0; break; @@ -1145,118 +1224,157 @@ emit_instruction( tmp1 = lp_build_mul( &bld->base, tmp1, tmp2); /* xmm1 = xmm1 * xmm2 */ tmp0 = lp_build_add( &bld->base, tmp0, tmp1); /* xmm0 = xmm0 + xmm1 */ FOR_EACH_DST0_ENABLED_CHANNEL( inst, chan_index ) { - emit_store( bld, inst, 0, chan_index, tmp0); /* dest[ch] = xmm0 */ + dst0[chan_index] = tmp0; /* dest[ch] = xmm0 */ } break; case TGSI_OPCODE_TXL: - emit_tex( bld, inst, TRUE, FALSE ); + emit_tex( bld, inst, TRUE, FALSE, dst0 ); break; case TGSI_OPCODE_TXP: - emit_tex( bld, inst, FALSE, TRUE ); + emit_tex( bld, inst, FALSE, TRUE, dst0 ); break; case TGSI_OPCODE_BRK: + /* FIXME */ return 0; break; case TGSI_OPCODE_IF: + /* FIXME */ return 0; break; case TGSI_OPCODE_BGNFOR: + /* deprecated */ + assert(0); return 0; break; case TGSI_OPCODE_REP: + /* deprecated */ + assert(0); return 0; break; case TGSI_OPCODE_ELSE: + /* FIXME */ return 0; break; case TGSI_OPCODE_ENDIF: + /* FIXME */ return 0; break; case TGSI_OPCODE_ENDFOR: + /* deprecated */ + assert(0); return 0; break; case TGSI_OPCODE_ENDREP: + /* deprecated */ + assert(0); return 0; break; case TGSI_OPCODE_PUSHA: + /* deprecated? */ + assert(0); return 0; break; case TGSI_OPCODE_POPA: + /* deprecated? */ + assert(0); return 0; break; case TGSI_OPCODE_CEIL: - return 0; + FOR_EACH_DST0_ENABLED_CHANNEL( inst, chan_index ) { + tmp0 = emit_fetch( bld, inst, 0, chan_index ); + dst0[chan_index] = lp_build_ceil(&bld->base, tmp0); + } break; case TGSI_OPCODE_I2F: + /* deprecated? */ + assert(0); return 0; break; case TGSI_OPCODE_NOT: + /* deprecated? */ + assert(0); return 0; break; -#if 0 case TGSI_OPCODE_TRUNC: FOR_EACH_DST0_ENABLED_CHANNEL( inst, chan_index ) { tmp0 = emit_fetch( bld, inst, 0, chan_index ); - emit_f2it( bld, 0 ); - emit_i2f( bld, 0 ); - emit_store( bld, inst, 0, chan_index, tmp0); + dst0[chan_index] = lp_build_trunc(&bld->base, tmp0); } break; -#endif case TGSI_OPCODE_SHL: + /* deprecated? */ + assert(0); return 0; break; case TGSI_OPCODE_SHR: + /* deprecated? */ + assert(0); return 0; break; case TGSI_OPCODE_AND: + /* deprecated? */ + assert(0); return 0; break; case TGSI_OPCODE_OR: + /* deprecated? */ + assert(0); return 0; break; case TGSI_OPCODE_MOD: + /* deprecated? */ + assert(0); return 0; break; case TGSI_OPCODE_XOR: + /* deprecated? */ + assert(0); return 0; break; case TGSI_OPCODE_SAD: + /* deprecated? */ + assert(0); return 0; break; case TGSI_OPCODE_TXF: + /* deprecated? */ + assert(0); return 0; break; case TGSI_OPCODE_TXQ: + /* deprecated? */ + assert(0); return 0; break; case TGSI_OPCODE_CONT: + /* deprecated? */ + assert(0); return 0; break; @@ -1268,10 +1386,28 @@ emit_instruction( return 0; break; + case TGSI_OPCODE_NOISE1: + case TGSI_OPCODE_NOISE2: + case TGSI_OPCODE_NOISE3: + case TGSI_OPCODE_NOISE4: + FOR_EACH_DST0_ENABLED_CHANNEL( inst, chan_index ) { + dst0[chan_index] = bld->base.zero; + } + break; + + case TGSI_OPCODE_NOP: + break; + default: return 0; } + if(info->num_dst) { + FOR_EACH_DST0_ENABLED_CHANNEL( inst, chan_index ) { + emit_store( bld, inst, 0, chan_index, dst0[chan_index]); + } + } + return 1; } @@ -1279,7 +1415,7 @@ emit_instruction( void lp_build_tgsi_soa(LLVMBuilderRef builder, const struct tgsi_token *tokens, - union lp_type type, + struct lp_type type, struct lp_build_mask_context *mask, LLVMValueRef consts_ptr, const LLVMValueRef *pos, @@ -1309,16 +1445,18 @@ lp_build_tgsi_soa(LLVMBuilderRef builder, switch( parse.FullToken.Token.Type ) { case TGSI_TOKEN_TYPE_DECLARATION: - /* Input already interpolated */ + /* Inputs already interpolated */ break; case TGSI_TOKEN_TYPE_INSTRUCTION: - if (!emit_instruction( &bld, &parse.FullToken.FullInstruction )) { + { unsigned opcode = parse.FullToken.FullInstruction.Instruction.Opcode; const struct tgsi_opcode_info *info = tgsi_get_opcode_info(opcode); - _debug_printf("warning: failed to translate tgsi opcode %s to LLVM\n", - info ? info->mnemonic : "<invalid>"); - } + if (!emit_instruction( &bld, &parse.FullToken.FullInstruction, info )) + _debug_printf("warning: failed to translate tgsi opcode %s to LLVM\n", + info ? info->mnemonic : "<invalid>"); + } + break; case TGSI_TOKEN_TYPE_IMMEDIATE: diff --git a/src/gallium/drivers/llvmpipe/lp_bld_type.c b/src/gallium/drivers/llvmpipe/lp_bld_type.c index 577644b7ab8..606243d6c5a 100644 --- a/src/gallium/drivers/llvmpipe/lp_bld_type.c +++ b/src/gallium/drivers/llvmpipe/lp_bld_type.c @@ -33,7 +33,7 @@ LLVMTypeRef -lp_build_elem_type(union lp_type type) +lp_build_elem_type(struct lp_type type) { if (type.floating) { switch(type.width) { @@ -55,7 +55,7 @@ lp_build_elem_type(union lp_type type) LLVMTypeRef -lp_build_vec_type(union lp_type type) +lp_build_vec_type(struct lp_type type) { LLVMTypeRef elem_type = lp_build_elem_type(type); return LLVMVectorType(elem_type, type.length); @@ -69,7 +69,7 @@ lp_build_vec_type(union lp_type type) * type and check for identity. */ boolean -lp_check_elem_type(union lp_type type, LLVMTypeRef elem_type) +lp_check_elem_type(struct lp_type type, LLVMTypeRef elem_type) { LLVMTypeKind elem_kind; @@ -107,7 +107,7 @@ lp_check_elem_type(union lp_type type, LLVMTypeRef elem_type) boolean -lp_check_vec_type(union lp_type type, LLVMTypeRef vec_type) +lp_check_vec_type(struct lp_type type, LLVMTypeRef vec_type) { LLVMTypeRef elem_type; @@ -128,7 +128,7 @@ lp_check_vec_type(union lp_type type, LLVMTypeRef vec_type) boolean -lp_check_value(union lp_type type, LLVMValueRef val) +lp_check_value(struct lp_type type, LLVMValueRef val) { LLVMTypeRef vec_type; @@ -143,25 +143,26 @@ lp_check_value(union lp_type type, LLVMValueRef val) LLVMTypeRef -lp_build_int_elem_type(union lp_type type) +lp_build_int_elem_type(struct lp_type type) { return LLVMIntType(type.width); } LLVMTypeRef -lp_build_int_vec_type(union lp_type type) +lp_build_int_vec_type(struct lp_type type) { LLVMTypeRef elem_type = lp_build_int_elem_type(type); return LLVMVectorType(elem_type, type.length); } -union lp_type -lp_int_type(union lp_type type) +struct lp_type +lp_int_type(struct lp_type type) { - union lp_type int_type; - int_type.value = 0; + struct lp_type int_type; + + memset(&int_type, 0, sizeof int_type); int_type.width = type.width; int_type.length = type.length; return int_type; @@ -171,7 +172,7 @@ lp_int_type(union lp_type type) void lp_build_context_init(struct lp_build_context *bld, LLVMBuilderRef builder, - union lp_type type) + struct lp_type type) { bld->builder = builder; bld->type = type; diff --git a/src/gallium/drivers/llvmpipe/lp_bld_type.h b/src/gallium/drivers/llvmpipe/lp_bld_type.h index 9933e0b45c3..ee5ca3483c1 100644 --- a/src/gallium/drivers/llvmpipe/lp_bld_type.h +++ b/src/gallium/drivers/llvmpipe/lp_bld_type.h @@ -56,58 +56,55 @@ * on the types used for intermediate computations, such as signed vs unsigned, * normalized values, or fixed point. */ -union lp_type { - struct { - /** - * Floating-point. Cannot be used with fixed. Integer numbers are - * represented by this zero. - */ - unsigned floating:1; - - /** - * Fixed-point. Cannot be used with floating. Integer numbers are - * represented by this zero. - */ - unsigned fixed:1; - - /** - * Whether it can represent negative values or not. - * - * If this is not set for floating point, it means that all values are - * assumed to be positive. - */ - unsigned sign:1; - - /** - * Whether values are normalized to fit [0, 1] interval, or [-1, 1] - * interval for signed types. - * - * For integer types it means the representable integer range should be - * interpreted as the interval above. - * - * For floating and fixed point formats it means the values should be - * clamped to the interval above. - */ - unsigned norm:1; - - /** - * Element width. - * - * For fixed point values, the fixed point is assumed to be at half the - * width. - */ - unsigned width:14; - - /** - * Vector length. - * - * width*length should be a power of two greater or equal to eight. - * - * @sa LP_MAX_VECTOR_LENGTH - */ - unsigned length:14; - }; - uint32_t value; +struct lp_type { + /** + * Floating-point. Cannot be used with fixed. Integer numbers are + * represented by this zero. + */ + unsigned floating:1; + + /** + * Fixed-point. Cannot be used with floating. Integer numbers are + * represented by this zero. + */ + unsigned fixed:1; + + /** + * Whether it can represent negative values or not. + * + * If this is not set for floating point, it means that all values are + * assumed to be positive. + */ + unsigned sign:1; + + /** + * Whether values are normalized to fit [0, 1] interval, or [-1, 1] + * interval for signed types. + * + * For integer types it means the representable integer range should be + * interpreted as the interval above. + * + * For floating and fixed point formats it means the values should be + * clamped to the interval above. + */ + unsigned norm:1; + + /** + * Element width. + * + * For fixed point values, the fixed point is assumed to be at half the + * width. + */ + unsigned width:14; + + /** + * Vector length. + * + * width*length should be a power of two greater or equal to eight. + * + * @sa LP_MAX_VECTOR_LENGTH + */ + unsigned length:14; }; @@ -124,7 +121,7 @@ struct lp_build_context * This not only describes the input/output LLVM types, but also whether * to normalize/clamp the results. */ - union lp_type type; + struct lp_type type; /** Same as lp_build_undef(type) */ LLVMValueRef undef; @@ -138,41 +135,41 @@ struct lp_build_context LLVMTypeRef -lp_build_elem_type(union lp_type type); +lp_build_elem_type(struct lp_type type); LLVMTypeRef -lp_build_vec_type(union lp_type type); +lp_build_vec_type(struct lp_type type); boolean -lp_check_elem_type(union lp_type type, LLVMTypeRef elem_type); +lp_check_elem_type(struct lp_type type, LLVMTypeRef elem_type); boolean -lp_check_vec_type(union lp_type type, LLVMTypeRef vec_type); +lp_check_vec_type(struct lp_type type, LLVMTypeRef vec_type); boolean -lp_check_value(union lp_type type, LLVMValueRef val); +lp_check_value(struct lp_type type, LLVMValueRef val); LLVMTypeRef -lp_build_int_elem_type(union lp_type type); +lp_build_int_elem_type(struct lp_type type); LLVMTypeRef -lp_build_int_vec_type(union lp_type type); +lp_build_int_vec_type(struct lp_type type); -union lp_type -lp_int_type(union lp_type type); +struct lp_type +lp_int_type(struct lp_type type); void lp_build_context_init(struct lp_build_context *bld, LLVMBuilderRef builder, - union lp_type type); + struct lp_type type); #endif /* !LP_BLD_TYPE_H */ diff --git a/src/gallium/drivers/llvmpipe/lp_clear.c b/src/gallium/drivers/llvmpipe/lp_clear.c index 580cca5b463..bdcff94b9bf 100644 --- a/src/gallium/drivers/llvmpipe/lp_clear.c +++ b/src/gallium/drivers/llvmpipe/lp_clear.c @@ -67,6 +67,7 @@ llvmpipe_clear(struct pipe_context *pipe, unsigned buffers, const float *rgba, util_pack_color(rgba, ps->format, &cv); lp_tile_cache_clear(llvmpipe->cbuf_cache[i], rgba, cv); } + llvmpipe->dirty_render_cache = TRUE; } if (buffers & PIPE_CLEAR_DEPTHSTENCIL) { diff --git a/src/gallium/drivers/llvmpipe/lp_context.c b/src/gallium/drivers/llvmpipe/lp_context.c index 233d1df0e10..a4b2bd8c2ad 100644 --- a/src/gallium/drivers/llvmpipe/lp_context.c +++ b/src/gallium/drivers/llvmpipe/lp_context.c @@ -141,8 +141,6 @@ llvmpipe_is_texture_referenced( struct pipe_context *pipe, return PIPE_REFERENCED_FOR_WRITE; } - /* FIXME: we also need to do the same for the texture cache */ - return PIPE_UNREFERENCED; } diff --git a/src/gallium/drivers/llvmpipe/lp_jit.c b/src/gallium/drivers/llvmpipe/lp_jit.c index 9465f763d50..b4a22ff4a97 100644 --- a/src/gallium/drivers/llvmpipe/lp_jit.c +++ b/src/gallium/drivers/llvmpipe/lp_jit.c @@ -152,7 +152,7 @@ lp_jit_screen_init(struct llvmpipe_screen *screen) screen->provider = LLVMCreateModuleProviderForExistingModule(screen->module); if (LLVMCreateJITCompiler(&screen->engine, screen->provider, 1, &error)) { - fprintf(stderr, "%s\n", error); + _debug_printf("%s\n", error); LLVMDisposeMessage(error); abort(); } diff --git a/src/gallium/drivers/llvmpipe/lp_jit.h b/src/gallium/drivers/llvmpipe/lp_jit.h index c3e3e1af672..58f716ede29 100644 --- a/src/gallium/drivers/llvmpipe/lp_jit.h +++ b/src/gallium/drivers/llvmpipe/lp_jit.h @@ -80,10 +80,9 @@ struct lp_jit_context struct tgsi_sampler **samplers; - /* TODO: alpha reference value */ float alpha_ref_value; - /* TODO: blend constant color */ + /* FIXME: store (also?) in floats */ uint8_t *blend_color; struct lp_jit_texture textures[PIPE_MAX_SAMPLERS]; diff --git a/src/gallium/drivers/llvmpipe/lp_screen.c b/src/gallium/drivers/llvmpipe/lp_screen.c index 125035771e5..05189274589 100644 --- a/src/gallium/drivers/llvmpipe/lp_screen.c +++ b/src/gallium/drivers/llvmpipe/lp_screen.c @@ -27,8 +27,6 @@ #include "util/u_memory.h" -#include "util/u_simple_screen.h" -#include "pipe/internal/p_winsys_screen.h" #include "pipe/p_defines.h" #include "pipe/p_screen.h" @@ -67,8 +65,6 @@ llvmpipe_get_param(struct pipe_screen *screen, int param) return 1; case PIPE_CAP_GLSL: return 1; - case PIPE_CAP_S3TC: - return 0; case PIPE_CAP_ANISOTROPIC_FILTER: return 0; case PIPE_CAP_POINT_SPRITE: @@ -86,7 +82,7 @@ llvmpipe_get_param(struct pipe_screen *screen, int param) case PIPE_CAP_MAX_TEXTURE_2D_LEVELS: return 13; /* max 4Kx4K */ case PIPE_CAP_MAX_TEXTURE_3D_LEVELS: - return 8; /* max 128x128x128 */ + return 9; /* max 256x256x256 */ case PIPE_CAP_MAX_TEXTURE_CUBE_LEVELS: return 13; /* max 4Kx4K */ case PIPE_CAP_TGSI_CONT_SUPPORTED: @@ -196,8 +192,7 @@ static void llvmpipe_destroy_screen( struct pipe_screen *_screen ) { struct llvmpipe_screen *screen = llvmpipe_screen(_screen); - - struct pipe_winsys *winsys = _screen->winsys; + struct llvmpipe_winsys *winsys = screen->winsys; lp_jit_screen_cleanup(screen); diff --git a/src/gallium/drivers/llvmpipe/lp_setup.c b/src/gallium/drivers/llvmpipe/lp_setup.c index f9e254efcae..2d2fc19a65f 100644 --- a/src/gallium/drivers/llvmpipe/lp_setup.c +++ b/src/gallium/drivers/llvmpipe/lp_setup.c @@ -163,8 +163,6 @@ shade_quads(struct llvmpipe_context *llvmpipe, else depth = NULL; - /* TODO: blend color */ - /* XXX: This will most likely fail on 32bit x86 without -mstackrealign */ assert(lp_check_alignment(mask, 16)); diff --git a/src/gallium/drivers/llvmpipe/lp_state_derived.c b/src/gallium/drivers/llvmpipe/lp_state_derived.c index e87976b9f36..30fb41ea65d 100644 --- a/src/gallium/drivers/llvmpipe/lp_state_derived.c +++ b/src/gallium/drivers/llvmpipe/lp_state_derived.c @@ -93,6 +93,23 @@ llvmpipe_get_vertex_info(struct llvmpipe_context *llvmpipe) vinfo->num_attribs = 0; for (i = 0; i < lpfs->info.num_inputs; i++) { int src; + enum interp_mode interp; + + switch (lpfs->info.input_interpolate[i]) { + case TGSI_INTERPOLATE_CONSTANT: + interp = INTERP_CONSTANT; + break; + case TGSI_INTERPOLATE_LINEAR: + interp = INTERP_LINEAR; + break; + case TGSI_INTERPOLATE_PERSPECTIVE: + interp = INTERP_PERSPECTIVE; + break; + default: + assert(0); + interp = INTERP_LINEAR; + } + switch (lpfs->info.input_semantic_name[i]) { case TGSI_SEMANTIC_POSITION: src = draw_find_vs_output(llvmpipe->draw, @@ -108,7 +125,7 @@ llvmpipe_get_vertex_info(struct llvmpipe_context *llvmpipe) case TGSI_SEMANTIC_FOG: src = draw_find_vs_output(llvmpipe->draw, TGSI_SEMANTIC_FOG, 0); - draw_emit_vertex_attr(vinfo, EMIT_4F, INTERP_PERSPECTIVE, src); + draw_emit_vertex_attr(vinfo, EMIT_4F, interp, src); break; case TGSI_SEMANTIC_GENERIC: @@ -116,7 +133,7 @@ llvmpipe_get_vertex_info(struct llvmpipe_context *llvmpipe) /* this includes texcoords and varying vars */ src = draw_find_vs_output(llvmpipe->draw, TGSI_SEMANTIC_GENERIC, lpfs->info.input_semantic_index[i]); - draw_emit_vertex_attr(vinfo, EMIT_4F, INTERP_PERSPECTIVE, src); + draw_emit_vertex_attr(vinfo, EMIT_4F, interp, src); break; default: diff --git a/src/gallium/drivers/llvmpipe/lp_state_fs.c b/src/gallium/drivers/llvmpipe/lp_state_fs.c index 618cf1ffb8d..9faed5a0b18 100644 --- a/src/gallium/drivers/llvmpipe/lp_state_fs.c +++ b/src/gallium/drivers/llvmpipe/lp_state_fs.c @@ -133,13 +133,13 @@ generate_pos0(LLVMBuilderRef builder, static void generate_depth(LLVMBuilderRef builder, const struct lp_fragment_shader_variant_key *key, - union lp_type src_type, + struct lp_type src_type, struct lp_build_mask_context *mask, LLVMValueRef src, LLVMValueRef dst_ptr) { const struct util_format_description *format_desc; - union lp_type dst_type; + struct lp_type dst_type; if(!key->depth.enabled) return; @@ -181,7 +181,7 @@ generate_fs(struct llvmpipe_context *lp, struct lp_fragment_shader *shader, const struct lp_fragment_shader_variant_key *key, LLVMBuilderRef builder, - union lp_type type, + struct lp_type type, LLVMValueRef context_ptr, unsigned i, const struct lp_build_interp_soa_context *interp, @@ -299,7 +299,7 @@ generate_fs(struct llvmpipe_context *lp, static void generate_blend(const struct pipe_blend_state *blend, LLVMBuilderRef builder, - union lp_type type, + struct lp_type type, LLVMValueRef context_ptr, LLVMValueRef mask, LLVMValueRef *src, @@ -364,8 +364,8 @@ generate_fragment(struct llvmpipe_context *lp, { struct llvmpipe_screen *screen = llvmpipe_screen(lp->pipe.screen); struct lp_fragment_shader_variant *variant; - union lp_type fs_type; - union lp_type blend_type; + struct lp_type fs_type; + struct lp_type blend_type; LLVMTypeRef fs_elem_type; LLVMTypeRef fs_vec_type; LLVMTypeRef fs_int_vec_type; @@ -431,7 +431,7 @@ generate_fragment(struct llvmpipe_context *lp, /* TODO: actually pick these based on the fs and color buffer * characteristics. */ - fs_type.value = 0; + memset(&fs_type, 0, sizeof fs_type); fs_type.floating = TRUE; /* floating point values */ fs_type.sign = TRUE; /* values are signed */ fs_type.norm = FALSE; /* values are not limited to [0,1] or [-1,1] */ @@ -439,7 +439,7 @@ generate_fragment(struct llvmpipe_context *lp, fs_type.length = 4; /* 4 element per vector */ num_fs = 4; - blend_type.value = 0; + memset(&blend_type, 0, sizeof blend_type); blend_type.floating = FALSE; /* values are integers */ blend_type.sign = FALSE; /* values are unsigned */ blend_type.norm = TRUE; /* values are in [0,1] or [-1,1] */ diff --git a/src/gallium/drivers/llvmpipe/lp_test.h b/src/gallium/drivers/llvmpipe/lp_test.h index 69aaae26e0a..a88e110c663 100644 --- a/src/gallium/drivers/llvmpipe/lp_test.h +++ b/src/gallium/drivers/llvmpipe/lp_test.h @@ -86,43 +86,43 @@ random_float(void); void -dump_type(FILE *fp, union lp_type type); +dump_type(FILE *fp, struct lp_type type); double -read_elem(union lp_type type, const void *src, unsigned index); +read_elem(struct lp_type type, const void *src, unsigned index); void -write_elem(union lp_type type, void *dst, unsigned index, double src); +write_elem(struct lp_type type, void *dst, unsigned index, double src); void -random_elem(union lp_type type, void *dst, unsigned index); +random_elem(struct lp_type type, void *dst, unsigned index); void -read_vec(union lp_type type, const void *src, double *dst); +read_vec(struct lp_type type, const void *src, double *dst); void -write_vec(union lp_type type, void *dst, const double *src); +write_vec(struct lp_type type, void *dst, const double *src); void -random_vec(union lp_type type, void *dst); +random_vec(struct lp_type type, void *dst); boolean -compare_vec_with_eps(union lp_type type, const void *res, const void *ref, double eps); +compare_vec_with_eps(struct lp_type type, const void *res, const void *ref, double eps); boolean -compare_vec(union lp_type type, const void *res, const void *ref); +compare_vec(struct lp_type type, const void *res, const void *ref); void -dump_vec(FILE *fp, union lp_type type, const void *src); +dump_vec(FILE *fp, struct lp_type type, const void *src); #endif /* !LP_TEST_H */ diff --git a/src/gallium/drivers/llvmpipe/lp_test_blend.c b/src/gallium/drivers/llvmpipe/lp_test_blend.c index 8dfad468e3c..94b661dcba4 100644 --- a/src/gallium/drivers/llvmpipe/lp_test_blend.c +++ b/src/gallium/drivers/llvmpipe/lp_test_blend.c @@ -80,7 +80,7 @@ static void write_tsv_row(FILE *fp, const struct pipe_blend_state *blend, enum vector_mode mode, - union lp_type type, + struct lp_type type, double cycles, boolean success) { @@ -125,7 +125,7 @@ static void dump_blend_type(FILE *fp, const struct pipe_blend_state *blend, enum vector_mode mode, - union lp_type type) + struct lp_type type) { fprintf(fp, "%s", mode ? "soa" : "aos"); @@ -153,7 +153,7 @@ static LLVMValueRef add_blend_test(LLVMModuleRef module, const struct pipe_blend_state *blend, enum vector_mode mode, - union lp_type type) + struct lp_type type) { LLVMTypeRef ret_type; LLVMTypeRef vec_type; @@ -467,7 +467,7 @@ test_one(unsigned verbose, FILE *fp, const struct pipe_blend_state *blend, enum vector_mode mode, - union lp_type type) + struct lp_type type) { LLVMModuleRef module = NULL; LLVMValueRef func = NULL; @@ -765,10 +765,10 @@ blend_funcs[] = { }; -const union lp_type blend_types[] = { +const struct lp_type blend_types[] = { /* float, fixed, sign, norm, width, len */ - {{ TRUE, FALSE, FALSE, TRUE, 32, 4 }}, /* f32 x 4 */ - {{ FALSE, FALSE, FALSE, TRUE, 8, 16 }}, /* u8n x 16 */ + { TRUE, FALSE, FALSE, TRUE, 32, 4 }, /* f32 x 4 */ + { FALSE, FALSE, FALSE, TRUE, 8, 16 }, /* u8n x 16 */ }; @@ -788,7 +788,7 @@ test_all(unsigned verbose, FILE *fp) const unsigned *alpha_dst_factor; struct pipe_blend_state blend; enum vector_mode mode; - const union lp_type *type; + const struct lp_type *type; bool success = TRUE; for(rgb_func = blend_funcs; rgb_func < &blend_funcs[num_funcs]; ++rgb_func) { @@ -841,27 +841,27 @@ test_some(unsigned verbose, FILE *fp, unsigned long n) const unsigned *alpha_dst_factor; struct pipe_blend_state blend; enum vector_mode mode; - const union lp_type *type; + const struct lp_type *type; unsigned long i; bool success = TRUE; for(i = 0; i < n; ++i) { - rgb_func = &blend_funcs[random() % num_funcs]; - alpha_func = &blend_funcs[random() % num_funcs]; - rgb_src_factor = &blend_factors[random() % num_factors]; - alpha_src_factor = &blend_factors[random() % num_factors]; + rgb_func = &blend_funcs[rand() % num_funcs]; + alpha_func = &blend_funcs[rand() % num_funcs]; + rgb_src_factor = &blend_factors[rand() % num_factors]; + alpha_src_factor = &blend_factors[rand() % num_factors]; do { - rgb_dst_factor = &blend_factors[random() % num_factors]; + rgb_dst_factor = &blend_factors[rand() % num_factors]; } while(*rgb_dst_factor == PIPE_BLENDFACTOR_SRC_ALPHA_SATURATE); do { - alpha_dst_factor = &blend_factors[random() % num_factors]; + alpha_dst_factor = &blend_factors[rand() % num_factors]; } while(*alpha_dst_factor == PIPE_BLENDFACTOR_SRC_ALPHA_SATURATE); - mode = random() & 1; + mode = rand() & 1; - type = &blend_types[random() % num_types]; + type = &blend_types[rand() % num_types]; memset(&blend, 0, sizeof blend); blend.blend_enable = 1; diff --git a/src/gallium/drivers/llvmpipe/lp_test_conv.c b/src/gallium/drivers/llvmpipe/lp_test_conv.c index e6489834af5..9dcf58e5dcd 100644 --- a/src/gallium/drivers/llvmpipe/lp_test_conv.c +++ b/src/gallium/drivers/llvmpipe/lp_test_conv.c @@ -59,8 +59,8 @@ write_tsv_header(FILE *fp) static void write_tsv_row(FILE *fp, - union lp_type src_type, - union lp_type dst_type, + struct lp_type src_type, + struct lp_type dst_type, double cycles, boolean success) { @@ -80,8 +80,8 @@ write_tsv_row(FILE *fp, static void dump_conv_types(FILE *fp, - union lp_type src_type, - union lp_type dst_type) + struct lp_type src_type, + struct lp_type dst_type) { fprintf(fp, "src_type="); dump_type(fp, src_type); @@ -96,8 +96,8 @@ dump_conv_types(FILE *fp, static LLVMValueRef add_conv_test(LLVMModuleRef module, - union lp_type src_type, unsigned num_srcs, - union lp_type dst_type, unsigned num_dsts) + struct lp_type src_type, unsigned num_srcs, + struct lp_type dst_type, unsigned num_dsts) { LLVMTypeRef args[2]; LLVMValueRef func; @@ -145,8 +145,8 @@ add_conv_test(LLVMModuleRef module, static boolean test_one(unsigned verbose, FILE *fp, - union lp_type src_type, - union lp_type dst_type) + struct lp_type src_type, + struct lp_type dst_type) { LLVMModuleRef module = NULL; LLVMValueRef func = NULL; @@ -343,35 +343,35 @@ test_one(unsigned verbose, } -const union lp_type conv_types[] = { +const struct lp_type conv_types[] = { /* float, fixed, sign, norm, width, len */ - {{ TRUE, FALSE, TRUE, TRUE, 32, 4 }}, - {{ TRUE, FALSE, TRUE, FALSE, 32, 4 }}, - {{ TRUE, FALSE, FALSE, TRUE, 32, 4 }}, - {{ TRUE, FALSE, FALSE, FALSE, 32, 4 }}, + { TRUE, FALSE, TRUE, TRUE, 32, 4 }, + { TRUE, FALSE, TRUE, FALSE, 32, 4 }, + { TRUE, FALSE, FALSE, TRUE, 32, 4 }, + { TRUE, FALSE, FALSE, FALSE, 32, 4 }, /* TODO: test fixed formats too */ - {{ FALSE, FALSE, TRUE, TRUE, 16, 8 }}, - {{ FALSE, FALSE, TRUE, FALSE, 16, 8 }}, - {{ FALSE, FALSE, FALSE, TRUE, 16, 8 }}, - {{ FALSE, FALSE, FALSE, FALSE, 16, 8 }}, - - {{ FALSE, FALSE, TRUE, TRUE, 32, 4 }}, - {{ FALSE, FALSE, TRUE, FALSE, 32, 4 }}, - {{ FALSE, FALSE, FALSE, TRUE, 32, 4 }}, - {{ FALSE, FALSE, FALSE, FALSE, 32, 4 }}, - - {{ FALSE, FALSE, TRUE, TRUE, 16, 8 }}, - {{ FALSE, FALSE, TRUE, FALSE, 16, 8 }}, - {{ FALSE, FALSE, FALSE, TRUE, 16, 8 }}, - {{ FALSE, FALSE, FALSE, FALSE, 16, 8 }}, - - {{ FALSE, FALSE, TRUE, TRUE, 8, 16 }}, - {{ FALSE, FALSE, TRUE, FALSE, 8, 16 }}, - {{ FALSE, FALSE, FALSE, TRUE, 8, 16 }}, - {{ FALSE, FALSE, FALSE, FALSE, 8, 16 }}, + { FALSE, FALSE, TRUE, TRUE, 16, 8 }, + { FALSE, FALSE, TRUE, FALSE, 16, 8 }, + { FALSE, FALSE, FALSE, TRUE, 16, 8 }, + { FALSE, FALSE, FALSE, FALSE, 16, 8 }, + + { FALSE, FALSE, TRUE, TRUE, 32, 4 }, + { FALSE, FALSE, TRUE, FALSE, 32, 4 }, + { FALSE, FALSE, FALSE, TRUE, 32, 4 }, + { FALSE, FALSE, FALSE, FALSE, 32, 4 }, + + { FALSE, FALSE, TRUE, TRUE, 16, 8 }, + { FALSE, FALSE, TRUE, FALSE, 16, 8 }, + { FALSE, FALSE, FALSE, TRUE, 16, 8 }, + { FALSE, FALSE, FALSE, FALSE, 16, 8 }, + + { FALSE, FALSE, TRUE, TRUE, 8, 16 }, + { FALSE, FALSE, TRUE, FALSE, 8, 16 }, + { FALSE, FALSE, FALSE, TRUE, 8, 16 }, + { FALSE, FALSE, FALSE, FALSE, 8, 16 }, }; @@ -381,8 +381,8 @@ const unsigned num_types = sizeof(conv_types)/sizeof(conv_types[0]); boolean test_all(unsigned verbose, FILE *fp) { - const union lp_type *src_type; - const union lp_type *dst_type; + const struct lp_type *src_type; + const struct lp_type *dst_type; bool success = TRUE; for(src_type = conv_types; src_type < &conv_types[num_types]; ++src_type) { @@ -407,16 +407,16 @@ test_all(unsigned verbose, FILE *fp) boolean test_some(unsigned verbose, FILE *fp, unsigned long n) { - const union lp_type *src_type; - const union lp_type *dst_type; + const struct lp_type *src_type; + const struct lp_type *dst_type; unsigned long i; bool success = TRUE; for(i = 0; i < n; ++i) { - src_type = &conv_types[random() % num_types]; + src_type = &conv_types[rand() % num_types]; do { - dst_type = &conv_types[random() % num_types]; + dst_type = &conv_types[rand() % num_types]; } while (src_type == dst_type || src_type->norm != dst_type->norm); if(!test_one(verbose, fp, *src_type, *dst_type)) diff --git a/src/gallium/drivers/llvmpipe/lp_test_main.c b/src/gallium/drivers/llvmpipe/lp_test_main.c index 49213fb4f0b..4592dc0b2d0 100644 --- a/src/gallium/drivers/llvmpipe/lp_test_main.c +++ b/src/gallium/drivers/llvmpipe/lp_test_main.c @@ -40,7 +40,7 @@ void dump_type(FILE *fp, - union lp_type type) + struct lp_type type) { fprintf(fp, "%s%s%u%sx%u", type.sign ? (type.floating || type.fixed ? "" : "s") : "u", @@ -52,7 +52,7 @@ dump_type(FILE *fp, double -read_elem(union lp_type type, const void *src, unsigned index) +read_elem(struct lp_type type, const void *src, unsigned index) { double scale = lp_const_scale(type); double value; @@ -115,7 +115,7 @@ read_elem(union lp_type type, const void *src, unsigned index) void -write_elem(union lp_type type, void *dst, unsigned index, double value) +write_elem(struct lp_type type, void *dst, unsigned index, double value) { assert(index < type.length); if(!type.sign && value < 0.0) @@ -184,11 +184,11 @@ write_elem(union lp_type type, void *dst, unsigned index, double value) void -random_elem(union lp_type type, void *dst, unsigned index) +random_elem(struct lp_type type, void *dst, unsigned index) { double value; assert(index < type.length); - value = (double)random()/(double)RAND_MAX; + value = (double)rand()/(double)RAND_MAX; if(!type.norm) { unsigned long long mask; if (type.floating) @@ -199,17 +199,17 @@ random_elem(union lp_type type, void *dst, unsigned index) mask = ((unsigned long long)1 << (type.width - 1)) - 1; else mask = ((unsigned long long)1 << type.width) - 1; - value += (double)(mask & random()); + value += (double)(mask & rand()); } if(!type.sign) - if(random() & 1) + if(rand() & 1) value = -value; write_elem(type, dst, index, value); } void -read_vec(union lp_type type, const void *src, double *dst) +read_vec(struct lp_type type, const void *src, double *dst) { unsigned i; for (i = 0; i < type.length; ++i) @@ -218,7 +218,7 @@ read_vec(union lp_type type, const void *src, double *dst) void -write_vec(union lp_type type, void *dst, const double *src) +write_vec(struct lp_type type, void *dst, const double *src) { unsigned i; for (i = 0; i < type.length; ++i) @@ -229,12 +229,12 @@ write_vec(union lp_type type, void *dst, const double *src) float random_float(void) { - return (float)((double)random()/(double)RAND_MAX); + return (float)((double)rand()/(double)RAND_MAX); } void -random_vec(union lp_type type, void *dst) +random_vec(struct lp_type type, void *dst) { unsigned i; for (i = 0; i < type.length; ++i) @@ -243,7 +243,7 @@ random_vec(union lp_type type, void *dst) boolean -compare_vec_with_eps(union lp_type type, const void *res, const void *ref, double eps) +compare_vec_with_eps(struct lp_type type, const void *res, const void *ref, double eps) { unsigned i; for (i = 0; i < type.length; ++i) { @@ -259,7 +259,7 @@ compare_vec_with_eps(union lp_type type, const void *res, const void *ref, doubl boolean -compare_vec(union lp_type type, const void *res, const void *ref) +compare_vec(struct lp_type type, const void *res, const void *ref) { double eps = lp_const_eps(type); return compare_vec_with_eps(type, res, ref, eps); @@ -267,7 +267,7 @@ compare_vec(union lp_type type, const void *res, const void *ref) void -dump_vec(FILE *fp, union lp_type type, const void *src) +dump_vec(FILE *fp, struct lp_type type, const void *src) { unsigned i; for (i = 0; i < type.length; ++i) { diff --git a/src/gallium/drivers/llvmpipe/lp_tex_sample_c.c b/src/gallium/drivers/llvmpipe/lp_tex_sample_c.c index 9a876f404df..a1365a045f1 100644 --- a/src/gallium/drivers/llvmpipe/lp_tex_sample_c.c +++ b/src/gallium/drivers/llvmpipe/lp_tex_sample_c.c @@ -1654,7 +1654,7 @@ lp_c_sampler_soa_destroy(struct lp_build_sampler_soa *sampler) static void lp_c_sampler_soa_emit_fetch_texel(struct lp_build_sampler_soa *_sampler, LLVMBuilderRef builder, - union lp_type type, + struct lp_type type, unsigned unit, unsigned num_coords, const LLVMValueRef *coords, diff --git a/src/gallium/drivers/llvmpipe/lp_tex_sample_llvm.c b/src/gallium/drivers/llvmpipe/lp_tex_sample_llvm.c index 7d31705d014..d2a6ae21f57 100644 --- a/src/gallium/drivers/llvmpipe/lp_tex_sample_llvm.c +++ b/src/gallium/drivers/llvmpipe/lp_tex_sample_llvm.c @@ -149,7 +149,7 @@ lp_llvm_sampler_soa_destroy(struct lp_build_sampler_soa *sampler) static void lp_llvm_sampler_soa_emit_fetch_texel(struct lp_build_sampler_soa *base, LLVMBuilderRef builder, - union lp_type type, + struct lp_type type, unsigned unit, unsigned num_coords, const LLVMValueRef *coords, diff --git a/src/gallium/drivers/llvmpipe/lp_tile_cache.c b/src/gallium/drivers/llvmpipe/lp_tile_cache.c index 143afec3d35..0c06b659a1f 100644 --- a/src/gallium/drivers/llvmpipe/lp_tile_cache.c +++ b/src/gallium/drivers/llvmpipe/lp_tile_cache.c @@ -44,10 +44,53 @@ #include "lp_tile_cache.h" +#define MAX_WIDTH 4096 +#define MAX_HEIGHT 4096 + + +enum llvmpipe_tile_status +{ + LP_TILE_STATUS_UNDEFINED = 0, + LP_TILE_STATUS_CLEAR = 1, + LP_TILE_STATUS_DEFINED = 2 +}; + + +struct llvmpipe_cached_tile +{ + enum llvmpipe_tile_status status; + + /** color in SOA format */ + uint8_t *color; +}; + + +struct llvmpipe_tile_cache +{ + struct pipe_screen *screen; + struct pipe_surface *surface; /**< the surface we're caching */ + struct pipe_transfer *transfer; + void *transfer_map; + + struct llvmpipe_cached_tile entries[MAX_WIDTH/TILE_SIZE][MAX_HEIGHT/TILE_SIZE]; + + uint8_t clear_color[4]; /**< for color bufs */ + uint clear_val; /**< for z+stencil, or packed color clear value */ + + struct llvmpipe_cached_tile *last_tile; /**< most recently retrieved tile */ +}; + + struct llvmpipe_tile_cache * lp_create_tile_cache( struct pipe_screen *screen ) { struct llvmpipe_tile_cache *tc; + int maxLevels, maxTexSize; + + /* sanity checking: max sure MAX_WIDTH/HEIGHT >= largest texture image */ + maxLevels = screen->get_param(screen, PIPE_CAP_MAX_TEXTURE_2D_LEVELS); + maxTexSize = 1 << (maxLevels - 1); + assert(MAX_WIDTH >= maxTexSize); tc = CALLOC_STRUCT( llvmpipe_tile_cache ); if(!tc) @@ -225,11 +268,14 @@ lp_flush_tile_cache(struct llvmpipe_tile_cache *tc) tc->clear_val); screen->transfer_unmap(screen, pt); + + tile->status = LP_TILE_STATUS_UNDEFINED; break; } case LP_TILE_STATUS_DEFINED: lp_put_tile_rgba_soa(pt, x, y, tile->color); + tile->status = LP_TILE_STATUS_UNDEFINED; break; } } @@ -257,7 +303,7 @@ lp_get_cached_tile(struct llvmpipe_tile_cache *tc, case LP_TILE_STATUS_UNDEFINED: /* get new tile data from transfer */ - lp_get_tile_rgba_soa(pt, x, y, tile->color); + lp_get_tile_rgba_soa(pt, x & ~(TILE_SIZE - 1), y & ~(TILE_SIZE - 1), tile->color); tile->status = LP_TILE_STATUS_DEFINED; break; diff --git a/src/gallium/drivers/llvmpipe/lp_tile_cache.h b/src/gallium/drivers/llvmpipe/lp_tile_cache.h index 6d8ba5ece7a..161bab37991 100644 --- a/src/gallium/drivers/llvmpipe/lp_tile_cache.h +++ b/src/gallium/drivers/llvmpipe/lp_tile_cache.h @@ -33,42 +33,7 @@ #include "lp_tile_soa.h" -enum llvmpipe_tile_status -{ - LP_TILE_STATUS_UNDEFINED = 0, - LP_TILE_STATUS_CLEAR = 1, - LP_TILE_STATUS_DEFINED = 2 -}; - - -struct llvmpipe_cached_tile -{ - enum llvmpipe_tile_status status; - - /** color in SOA format */ - uint8_t *color; -}; - - -/** XXX move these */ -#define MAX_WIDTH 2048 -#define MAX_HEIGHT 2048 - - -struct llvmpipe_tile_cache -{ - struct pipe_screen *screen; - struct pipe_surface *surface; /**< the surface we're caching */ - struct pipe_transfer *transfer; - void *transfer_map; - - struct llvmpipe_cached_tile entries[MAX_WIDTH/TILE_SIZE][MAX_HEIGHT/TILE_SIZE]; - - uint8_t clear_color[4]; /**< for color bufs */ - uint clear_val; /**< for z+stencil, or packed color clear value */ - - struct llvmpipe_cached_tile *last_tile; /**< most recently retrieved tile */ -}; +struct llvmpipe_tile_cache; /* opaque */ extern struct llvmpipe_tile_cache * diff --git a/src/gallium/drivers/nv04/nv04_screen.c b/src/gallium/drivers/nv04/nv04_screen.c index ff2febb668e..170ce3eb7e5 100644 --- a/src/gallium/drivers/nv04/nv04_screen.c +++ b/src/gallium/drivers/nv04/nv04_screen.c @@ -16,8 +16,6 @@ nv04_screen_get_param(struct pipe_screen *screen, int param) return 0; case PIPE_CAP_GLSL: return 0; - case PIPE_CAP_S3TC: - return 0; case PIPE_CAP_ANISOTROPIC_FILTER: return 0; case PIPE_CAP_POINT_SPRITE: diff --git a/src/gallium/drivers/nv10/nv10_screen.c b/src/gallium/drivers/nv10/nv10_screen.c index 4469b22d91a..ee5901e743e 100644 --- a/src/gallium/drivers/nv10/nv10_screen.c +++ b/src/gallium/drivers/nv10/nv10_screen.c @@ -15,8 +15,6 @@ nv10_screen_get_param(struct pipe_screen *screen, int param) return 0; case PIPE_CAP_GLSL: return 0; - case PIPE_CAP_S3TC: - return 0; case PIPE_CAP_ANISOTROPIC_FILTER: return 1; case PIPE_CAP_POINT_SPRITE: diff --git a/src/gallium/drivers/nv20/nv20_screen.c b/src/gallium/drivers/nv20/nv20_screen.c index e6924ad71eb..4eeacd1afd5 100644 --- a/src/gallium/drivers/nv20/nv20_screen.c +++ b/src/gallium/drivers/nv20/nv20_screen.c @@ -15,8 +15,6 @@ nv20_screen_get_param(struct pipe_screen *screen, int param) return 0; case PIPE_CAP_GLSL: return 0; - case PIPE_CAP_S3TC: - return 0; case PIPE_CAP_ANISOTROPIC_FILTER: return 1; case PIPE_CAP_POINT_SPRITE: diff --git a/src/gallium/drivers/nv30/nv30_screen.c b/src/gallium/drivers/nv30/nv30_screen.c index f8285e4455f..41af38450b5 100644 --- a/src/gallium/drivers/nv30/nv30_screen.c +++ b/src/gallium/drivers/nv30/nv30_screen.c @@ -22,8 +22,6 @@ nv30_screen_get_param(struct pipe_screen *pscreen, int param) return 1; case PIPE_CAP_GLSL: return 0; - case PIPE_CAP_S3TC: - return 0; case PIPE_CAP_ANISOTROPIC_FILTER: return 1; case PIPE_CAP_POINT_SPRITE: diff --git a/src/gallium/drivers/nv40/nv40_screen.c b/src/gallium/drivers/nv40/nv40_screen.c index 5d2a4216c5a..bd13dfddd1c 100644 --- a/src/gallium/drivers/nv40/nv40_screen.c +++ b/src/gallium/drivers/nv40/nv40_screen.c @@ -21,8 +21,6 @@ nv40_screen_get_param(struct pipe_screen *pscreen, int param) return 1; case PIPE_CAP_GLSL: return 0; - case PIPE_CAP_S3TC: - return 1; case PIPE_CAP_ANISOTROPIC_FILTER: return 1; case PIPE_CAP_POINT_SPRITE: diff --git a/src/gallium/drivers/nv50/nv50_context.c b/src/gallium/drivers/nv50/nv50_context.c index 6e8f4f9750d..fca078b174a 100644 --- a/src/gallium/drivers/nv50/nv50_context.c +++ b/src/gallium/drivers/nv50/nv50_context.c @@ -37,11 +37,12 @@ nv50_flush(struct pipe_context *pipe, unsigned flags, /* We need this in the ddx for reliable composite, not sure what we're * actually flushing. We generate all our own flushes with flags = 0. */ - WAIT_RING(chan, 3); + WAIT_RING(chan, 2); BEGIN_RING(chan, eng2d, 0x0110, 1); OUT_RING (chan, 0); - FIRE_RING(chan); + if (flags & PIPE_FLUSH_FRAME) + FIRE_RING(chan); } static void @@ -110,6 +111,9 @@ nv50_create(struct pipe_screen *pscreen, unsigned pctx_id) nv50->pipe.is_texture_referenced = nv50_is_texture_referenced; nv50->pipe.is_buffer_referenced = nv50_is_buffer_referenced; + screen->base.channel->user_private = nv50; + screen->base.channel->flush_notify = nv50_state_flush_notify; + nv50_init_surface_functions(nv50); nv50_init_state_functions(nv50); nv50_init_query_functions(nv50); diff --git a/src/gallium/drivers/nv50/nv50_context.h b/src/gallium/drivers/nv50/nv50_context.h index 1e9e8e49bfb..4608854d711 100644 --- a/src/gallium/drivers/nv50/nv50_context.h +++ b/src/gallium/drivers/nv50/nv50_context.h @@ -116,6 +116,7 @@ struct nv50_state { unsigned miptree_nr; struct nouveau_stateobj *vertprog; struct nouveau_stateobj *fragprog; + struct nouveau_stateobj *programs; struct nouveau_stateobj *vtxfmt; struct nouveau_stateobj *vtxbuf; struct nouveau_stateobj *vtxattr; @@ -190,10 +191,12 @@ extern void nv50_clear(struct pipe_context *pipe, unsigned buffers, /* nv50_program.c */ extern void nv50_vertprog_validate(struct nv50_context *nv50); extern void nv50_fragprog_validate(struct nv50_context *nv50); +extern void nv50_linkage_validate(struct nv50_context *nv50); extern void nv50_program_destroy(struct nv50_context *nv50, struct nv50_program *p); /* nv50_state_validate.c */ extern boolean nv50_state_validate(struct nv50_context *nv50); +extern void nv50_state_flush_notify(struct nouveau_channel *chan); /* nv50_tex.c */ extern void nv50_tex_validate(struct nv50_context *); diff --git a/src/gallium/drivers/nv50/nv50_program.c b/src/gallium/drivers/nv50/nv50_program.c index 4a838529de7..576d075318f 100644 --- a/src/gallium/drivers/nv50/nv50_program.c +++ b/src/gallium/drivers/nv50/nv50_program.c @@ -90,6 +90,10 @@ struct nv50_reg { int acc; /* instruction where this reg is last read (first insn == 1) */ }; +/* arbitrary limits */ +#define MAX_IF_DEPTH 4 +#define MAX_LOOP_DEPTH 4 + struct nv50_pc { struct nv50_program *p; @@ -112,11 +116,22 @@ struct nv50_pc { struct nv50_reg *temp_temp[16]; unsigned temp_temp_nr; + /* broadcast and destination replacement regs */ + struct nv50_reg *r_brdc; + struct nv50_reg *r_dst[4]; + unsigned interp_mode[32]; /* perspective interpolation registers */ struct nv50_reg *iv_p; struct nv50_reg *iv_c; + struct nv50_program_exec *if_cond; + struct nv50_program_exec *if_insn[MAX_IF_DEPTH]; + struct nv50_program_exec *br_join[MAX_IF_DEPTH]; + struct nv50_program_exec *br_loop[MAX_LOOP_DEPTH]; /* for BRK branch */ + int if_lvl, loop_lvl; + unsigned loop_pos[MAX_LOOP_DEPTH]; + /* current instruction and total number of insns */ unsigned insn_cur; unsigned insn_nr; @@ -124,6 +139,25 @@ struct nv50_pc { boolean allow32; }; +static INLINE void +ctor_reg(struct nv50_reg *reg, unsigned type, int index, int hw) +{ + reg->type = type; + reg->index = index; + reg->hw = hw; + reg->neg = 0; + reg->rhw = -1; + reg->acc = 0; +} + +static INLINE unsigned +popcnt4(uint32_t val) +{ + static const unsigned cnt[16] + = { 0, 1, 1, 2, 1, 2, 2, 3, 1, 2, 2, 3, 2, 3, 3, 4 }; + return cnt[val & 0xf]; +} + static void alloc_reg(struct nv50_pc *pc, struct nv50_reg *reg) { @@ -173,6 +207,10 @@ alloc_reg(struct nv50_pc *pc, struct nv50_reg *reg) assert(0); } +/* XXX: For shaders that aren't executed linearly (e.g. shaders that + * contain loops), we need to assign all hw regs to TGSI TEMPs early, + * lest we risk temp_temps overwriting regs alloc'd "later". + */ static struct nv50_reg * alloc_temp(struct nv50_pc *pc, struct nv50_reg *dst) { @@ -184,11 +222,8 @@ alloc_temp(struct nv50_pc *pc, struct nv50_reg *dst) for (i = 0; i < NV50_SU_MAX_TEMP; i++) { if (!pc->r_temp[i]) { - r = CALLOC_STRUCT(nv50_reg); - r->type = P_TEMP; - r->index = -1; - r->hw = i; - r->rhw = -1; + r = MALLOC_STRUCT(nv50_reg); + ctor_reg(r, P_TEMP, -1, i); pc->r_temp[i] = r; return r; } @@ -254,10 +289,8 @@ alloc_temp4(struct nv50_pc *pc, struct nv50_reg *dst[4], int idx) return alloc_temp4(pc, dst, idx + 4); for (i = 0; i < 4; i++) { - dst[i] = CALLOC_STRUCT(nv50_reg); - dst[i]->type = P_TEMP; - dst[i]->index = -1; - dst[i]->hw = idx + i; + dst[i] = MALLOC_STRUCT(nv50_reg); + ctor_reg(dst[i], P_TEMP, -1, idx + i); pc->r_temp[idx + i] = dst[i]; } @@ -309,7 +342,7 @@ ctor_immd(struct nv50_pc *pc, float x, float y, float z, float w) static struct nv50_reg * alloc_immd(struct nv50_pc *pc, float f) { - struct nv50_reg *r = CALLOC_STRUCT(nv50_reg); + struct nv50_reg *r = MALLOC_STRUCT(nv50_reg); unsigned hw; for (hw = 0; hw < pc->immd_nr * 4; hw++) @@ -319,9 +352,7 @@ alloc_immd(struct nv50_pc *pc, float f) if (hw == pc->immd_nr * 4) hw = ctor_immd(pc, f, -f, 0.5 * f, 0) * 4; - r->type = P_IMMD; - r->hw = hw; - r->index = -1; + ctor_reg(r, P_IMMD, -1, hw); return r; } @@ -543,6 +574,22 @@ check_swap_src_0_1(struct nv50_pc *pc, } static void +set_src_0_restricted(struct nv50_pc *pc, struct nv50_reg *src, + struct nv50_program_exec *e) +{ + struct nv50_reg *temp; + + if (src->type != P_TEMP) { + temp = temp_temp(pc); + emit_mov(pc, temp, src); + src = temp; + } + + alloc_reg(pc, src); + e->inst[0] |= (src->hw << 9); +} + +static void set_src_0(struct nv50_pc *pc, struct nv50_reg *src, struct nv50_program_exec *e) { if (src->type == P_ATTR) { @@ -744,7 +791,11 @@ emit_flop(struct nv50_pc *pc, unsigned sub, } set_dst(pc, dst, e); - set_src_0(pc, src, e); + + if (sub == 0 || sub == 2) + set_src_0_restricted(pc, src, e); + else + set_src_0(pc, src, e); emit(pc, e); } @@ -786,16 +837,20 @@ emit_precossin(struct nv50_pc *pc, struct nv50_reg *dst, struct nv50_reg *src) #define CVTOP_SAT 0x08 #define CVTOP_ABS 0x10 +/* 0x04 == 32 bit */ +/* 0x40 == dst is float */ +/* 0x80 == src is float */ #define CVT_F32_F32 0xc4 #define CVT_F32_S32 0x44 #define CVT_F32_U32 0x64 #define CVT_S32_F32 0x8c #define CVT_S32_S32 0x0c -#define CVT_F32_F32_ROP 0xcc +#define CVT_NEG 0x20 +#define CVT_RI 0x08 static void emit_cvt(struct nv50_pc *pc, struct nv50_reg *dst, struct nv50_reg *src, - int wp, unsigned cop, unsigned fmt) + int wp, unsigned cvn, unsigned fmt) { struct nv50_program_exec *e; @@ -804,7 +859,7 @@ emit_cvt(struct nv50_pc *pc, struct nv50_reg *dst, struct nv50_reg *src, e->inst[0] |= 0xa0000000; e->inst[1] |= 0x00004000; - e->inst[1] |= (cop << 16); + e->inst[1] |= (cvn << 16); e->inst[1] |= (fmt << 24); set_src_0(pc, src, e); @@ -821,53 +876,85 @@ emit_cvt(struct nv50_pc *pc, struct nv50_reg *dst, struct nv50_reg *src, emit(pc, e); } +/* nv50 Condition codes: + * 0x1 = LT + * 0x2 = EQ + * 0x3 = LE + * 0x4 = GT + * 0x5 = NE + * 0x6 = GE + * 0x7 = set condition code ? (used before bra.lt/le/gt/ge) + * 0x8 = unordered bit (allows NaN) + */ static void -emit_set(struct nv50_pc *pc, unsigned c_op, struct nv50_reg *dst, +emit_set(struct nv50_pc *pc, unsigned ccode, struct nv50_reg *dst, int wp, struct nv50_reg *src0, struct nv50_reg *src1) { + static const unsigned cc_swapped[8] = { 0, 4, 2, 6, 1, 5, 3, 7 }; + struct nv50_program_exec *e = exec(pc); - unsigned inv_cop[8] = { 0, 4, 2, 6, 1, 5, 3, 7 }; struct nv50_reg *rdst; - assert(c_op <= 7); + assert(ccode < 16); if (check_swap_src_0_1(pc, &src0, &src1)) - c_op = inv_cop[c_op]; + ccode = cc_swapped[ccode & 7] | (ccode & 8); rdst = dst; - if (dst->type != P_TEMP) + if (dst && dst->type != P_TEMP) dst = alloc_temp(pc, NULL); /* set.u32 */ set_long(pc, e); e->inst[0] |= 0xb0000000; - e->inst[1] |= (3 << 29); - e->inst[1] |= (c_op << 14); - /*XXX: breaks things, .u32 by default? - * decuda will disasm as .u16 and use .lo/.hi regs, but this - * doesn't seem to match what the hw actually does. - inst[1] |= 0x04000000; << breaks things.. .u32 by default? + e->inst[1] |= 0x60000000 | (ccode << 14); + + /* XXX: decuda will disasm as .u16 and use .lo/.hi regs, but + * that doesn't seem to match what the hw actually does + e->inst[1] |= 0x04000000; << breaks things, u32 by default ? */ - set_dst(pc, dst, e); + + if (wp >= 0) + set_pred_wr(pc, 1, wp, e); + if (dst) + set_dst(pc, dst, e); + else { + e->inst[0] |= 0x000001fc; + e->inst[1] |= 0x00000008; + } + set_src_0(pc, src0, e); set_src_1(pc, src1, e); - emit(pc, e); - /* cvt.f32.u32 */ - e = exec(pc); - e->inst[0] = 0xa0000001; - e->inst[1] = 0x64014780; - set_dst(pc, rdst, e); - set_src_0(pc, dst, e); emit(pc, e); + pc->if_cond = pc->p->exec_tail; /* record for OPCODE_IF */ - if (dst != rdst) + /* cvt.f32.u32/s32 (?) if we didn't only write the predicate */ + if (rdst) + emit_cvt(pc, rdst, dst, -1, CVTOP_ABS | CVTOP_RN, CVT_F32_S32); + if (rdst && rdst != dst) free_temp(pc, dst); } +static INLINE unsigned +map_tgsi_setop_cc(unsigned op) +{ + switch (op) { + case TGSI_OPCODE_SLT: return 0x1; + case TGSI_OPCODE_SGE: return 0x6; + case TGSI_OPCODE_SEQ: return 0x2; + case TGSI_OPCODE_SGT: return 0x4; + case TGSI_OPCODE_SLE: return 0x3; + case TGSI_OPCODE_SNE: return 0xd; + default: + assert(0); + return 0; + } +} + static INLINE void emit_flr(struct nv50_pc *pc, struct nv50_reg *dst, struct nv50_reg *src) { - emit_cvt(pc, dst, src, -1, CVTOP_FLOOR, CVT_F32_F32_ROP); + emit_cvt(pc, dst, src, -1, CVTOP_FLOOR, CVT_F32_F32 | CVT_RI); } static void @@ -890,6 +977,12 @@ emit_abs(struct nv50_pc *pc, struct nv50_reg *dst, struct nv50_reg *src) emit_cvt(pc, dst, src, -1, CVTOP_ABS, CVT_F32_F32); } +static INLINE void +emit_sat(struct nv50_pc *pc, struct nv50_reg *dst, struct nv50_reg *src) +{ + emit_cvt(pc, dst, src, -1, CVTOP_SAT, CVT_F32_F32); +} + static void emit_lit(struct nv50_pc *pc, struct nv50_reg **dst, unsigned mask, struct nv50_reg **src) @@ -1073,10 +1166,11 @@ emit_tex(struct nv50_pc *pc, struct nv50_reg **dst, unsigned mask, emit(pc, e); #if 1 - if (mask & 1) emit_mov(pc, dst[0], t[0]); - if (mask & 2) emit_mov(pc, dst[1], t[1]); - if (mask & 4) emit_mov(pc, dst[2], t[2]); - if (mask & 8) emit_mov(pc, dst[3], t[3]); + c = 0; + if (mask & 1) emit_mov(pc, dst[0], t[c++]); + if (mask & 2) emit_mov(pc, dst[1], t[c++]); + if (mask & 4) emit_mov(pc, dst[2], t[c++]); + if (mask & 8) emit_mov(pc, dst[3], t[c]); free_temp4(pc, t); #else @@ -1093,6 +1187,38 @@ emit_tex(struct nv50_pc *pc, struct nv50_reg **dst, unsigned mask, } static void +emit_branch(struct nv50_pc *pc, int pred, unsigned cc, + struct nv50_program_exec **join) +{ + struct nv50_program_exec *e = exec(pc); + + if (join) { + set_long(pc, e); + e->inst[0] |= 0xa0000002; + emit(pc, e); + *join = e; + e = exec(pc); + } + + set_long(pc, e); + e->inst[0] |= 0x10000002; + if (pred >= 0) + set_pred(pc, cc, pred, e); + emit(pc, e); +} + +static void +emit_nop(struct nv50_pc *pc) +{ + struct nv50_program_exec *e = exec(pc); + + e->inst[0] = 0xf0000000; + set_long(pc, e); + e->inst[1] = 0xe0000000; + emit(pc, e); +} + +static void convert_to_long(struct nv50_pc *pc, struct nv50_program_exec *e) { unsigned q = 0, m = ~0; @@ -1159,6 +1285,70 @@ negate_supported(const struct tgsi_full_instruction *insn, int i) } } +/* Return a read mask for source registers deduced from opcode & write mask. */ +static unsigned +nv50_tgsi_src_mask(const struct tgsi_full_instruction *insn, int c) +{ + unsigned x, mask = insn->FullDstRegisters[0].DstRegister.WriteMask; + + switch (insn->Instruction.Opcode) { + case TGSI_OPCODE_COS: + case TGSI_OPCODE_SIN: + return (mask & 0x8) | ((mask & 0x7) ? 0x1 : 0x0); + case TGSI_OPCODE_DP3: + return 0x7; + case TGSI_OPCODE_DP4: + case TGSI_OPCODE_DPH: + case TGSI_OPCODE_KIL: /* WriteMask ignored */ + return 0xf; + case TGSI_OPCODE_DST: + return mask & (c ? 0xa : 0x6); + case TGSI_OPCODE_EX2: + case TGSI_OPCODE_LG2: + case TGSI_OPCODE_POW: + case TGSI_OPCODE_RCP: + case TGSI_OPCODE_RSQ: + case TGSI_OPCODE_SCS: + return 0x1; + case TGSI_OPCODE_LIT: + return 0xb; + case TGSI_OPCODE_TEX: + case TGSI_OPCODE_TXP: + { + const struct tgsi_instruction_ext_texture *tex; + + assert(insn->Instruction.Extended); + tex = &insn->InstructionExtTexture; + + mask = 0x7; + if (insn->Instruction.Opcode == TGSI_OPCODE_TXP) + mask |= 0x8; + + switch (tex->Texture) { + case TGSI_TEXTURE_1D: + mask &= 0x9; + break; + case TGSI_TEXTURE_2D: + mask &= 0xb; + break; + default: + break; + } + } + return mask; + case TGSI_OPCODE_XPD: + x = 0; + if (mask & 1) x |= 0x6; + if (mask & 2) x |= 0x5; + if (mask & 4) x |= 0x3; + return x; + default: + break; + } + + return mask; +} + static struct nv50_reg * tgsi_dst(struct nv50_pc *pc, int c, const struct tgsi_full_dst_register *dst) { @@ -1258,93 +1448,175 @@ tgsi_src(struct nv50_pc *pc, int chan, const struct tgsi_full_src_register *src, return r; } -/* returns TRUE if instruction can overwrite sources before they're read */ +/* return TRUE for ops that produce only a single result */ static boolean -direct2dest_op(const struct tgsi_full_instruction *insn) +is_scalar_op(unsigned op) { - if (insn->Instruction.Saturate) - return FALSE; - - switch (insn->Instruction.Opcode) { + switch (op) { case TGSI_OPCODE_COS: + case TGSI_OPCODE_DP2: case TGSI_OPCODE_DP3: case TGSI_OPCODE_DP4: case TGSI_OPCODE_DPH: - case TGSI_OPCODE_KIL: - case TGSI_OPCODE_LIT: + case TGSI_OPCODE_EX2: + case TGSI_OPCODE_LG2: case TGSI_OPCODE_POW: case TGSI_OPCODE_RCP: case TGSI_OPCODE_RSQ: - case TGSI_OPCODE_SCS: case TGSI_OPCODE_SIN: + /* + case TGSI_OPCODE_KIL: + case TGSI_OPCODE_LIT: + case TGSI_OPCODE_SCS: + */ + return TRUE; + default: + return FALSE; + } +} + +/* Returns a bitmask indicating which dst components depend + * on source s, component c (reverse of nv50_tgsi_src_mask). + */ +static unsigned +nv50_tgsi_dst_revdep(unsigned op, int s, int c) +{ + if (is_scalar_op(op)) + return 0x1; + + switch (op) { + case TGSI_OPCODE_DST: + return (1 << c) & (s ? 0xa : 0x6); + case TGSI_OPCODE_XPD: + switch (c) { + case 0: return 0x6; + case 1: return 0x5; + case 2: return 0x3; + case 3: return 0x0; + default: + assert(0); + return 0x0; + } + case TGSI_OPCODE_LIT: + case TGSI_OPCODE_SCS: case TGSI_OPCODE_TEX: case TGSI_OPCODE_TXP: - return FALSE; + /* these take care of dangerous swizzles themselves */ + return 0x0; + case TGSI_OPCODE_IF: + case TGSI_OPCODE_KIL: + /* don't call this function for these ops */ + assert(0); + return 0; default: - return TRUE; + /* linear vector instruction */ + return (1 << c); } } +static INLINE boolean +has_pred(struct nv50_program_exec *e, unsigned cc) +{ + if (!is_long(e) || is_immd(e)) + return FALSE; + return ((e->inst[1] & 0x780) == (cc << 7)); +} + +/* on ENDIF see if we can do "@p0.neu single_op" instead of: + * join_at ENDIF + * @p0.eq bra ENDIF + * single_op + * ENDIF: nop.join + */ static boolean -nv50_program_tx_insn(struct nv50_pc *pc, const union tgsi_full_token *tok) +nv50_kill_branch(struct nv50_pc *pc) { - const struct tgsi_full_instruction *inst = &tok->FullInstruction; - struct nv50_reg *rdst[4], *dst[4], *src[3][4], *temp; + int lvl = pc->if_lvl; + + if (pc->if_insn[lvl]->next != pc->p->exec_tail) + return FALSE; + + /* if ccode == 'true', the BRA is from an ELSE and the predicate + * reg may no longer be valid, since we currently always use $p0 + */ + if (has_pred(pc->if_insn[lvl], 0xf)) + return FALSE; + assert(pc->if_insn[lvl] && pc->br_join[lvl]); + + /* We'll use the exec allocated for JOIN_AT (as we can't easily + * update prev's next); if exec_tail is BRK, update the pointer. + */ + if (pc->loop_lvl && pc->br_loop[pc->loop_lvl - 1] == pc->p->exec_tail) + pc->br_loop[pc->loop_lvl - 1] = pc->br_join[lvl]; + + pc->p->exec_size -= 4; /* remove JOIN_AT and BRA */ + + *pc->br_join[lvl] = *pc->p->exec_tail; + + FREE(pc->if_insn[lvl]); + FREE(pc->p->exec_tail); + + pc->p->exec_tail = pc->br_join[lvl]; + pc->p->exec_tail->next = NULL; + set_pred(pc, 0xd, 0, pc->p->exec_tail); + + return TRUE; +} + +static boolean +nv50_program_tx_insn(struct nv50_pc *pc, + const struct tgsi_full_instruction *inst) +{ + struct nv50_reg *rdst[4], *dst[4], *brdc, *src[3][4], *temp; unsigned mask, sat, unit; - boolean assimilate = FALSE; int i, c; mask = inst->FullDstRegisters[0].DstRegister.WriteMask; sat = inst->Instruction.Saturate == TGSI_SAT_ZERO_ONE; + memset(src, 0, sizeof(src)); + for (c = 0; c < 4; c++) { - if (mask & (1 << c)) + if ((mask & (1 << c)) && !pc->r_dst[c]) dst[c] = tgsi_dst(pc, c, &inst->FullDstRegisters[0]); else - dst[c] = NULL; - rdst[c] = NULL; - src[0][c] = NULL; - src[1][c] = NULL; - src[2][c] = NULL; + dst[c] = pc->r_dst[c]; + rdst[c] = dst[c]; } for (i = 0; i < inst->Instruction.NumSrcRegs; i++) { const struct tgsi_full_src_register *fs = &inst->FullSrcRegisters[i]; + unsigned src_mask; + boolean neg_supp; + + src_mask = nv50_tgsi_src_mask(inst, i); + neg_supp = negate_supported(inst, i); if (fs->SrcRegister.File == TGSI_FILE_SAMPLER) unit = fs->SrcRegister.Index; for (c = 0; c < 4; c++) - src[i][c] = tgsi_src(pc, c, fs, - negate_supported(inst, i)); + if (src_mask & (1 << c)) + src[i][c] = tgsi_src(pc, c, fs, neg_supp); } - if (sat) { - for (c = 0; c < 4; c++) { - rdst[c] = dst[c]; - dst[c] = temp_temp(pc); - } + brdc = temp = pc->r_brdc; + if (brdc && brdc->type != P_TEMP) { + temp = temp_temp(pc); + if (sat) + brdc = temp; } else - if (direct2dest_op(inst)) { + if (sat) { for (c = 0; c < 4; c++) { - if (!dst[c] || dst[c]->type != P_TEMP) - continue; - - for (i = c + 1; i < 4; i++) { - if (dst[c] == src[0][i] || - dst[c] == src[1][i] || - dst[c] == src[2][i]) - break; - } - if (i == 4) + if (!(mask & (1 << c)) || dst[c]->type == P_TEMP) continue; - - assimilate = TRUE; rdst[c] = dst[c]; - dst[c] = alloc_temp(pc, NULL); + dst[c] = temp_temp(pc); } } + assert(brdc || !is_scalar_op(inst->Instruction.Opcode)); + switch (inst->Instruction.Opcode) { case TGSI_OPCODE_ABS: for (c = 0; c < 4; c++) { @@ -1360,74 +1632,91 @@ nv50_program_tx_insn(struct nv50_pc *pc, const union tgsi_full_token *tok) emit_add(pc, dst[c], src[0][c], src[1][c]); } break; - case TGSI_OPCODE_COS: - temp = temp_temp(pc); - emit_precossin(pc, temp, src[0][0]); - emit_flop(pc, 5, temp, temp); + case TGSI_OPCODE_BGNLOOP: + pc->loop_pos[pc->loop_lvl++] = pc->p->exec_size; + break; + case TGSI_OPCODE_BRK: + emit_branch(pc, -1, 0, NULL); + assert(pc->loop_lvl > 0); + pc->br_loop[pc->loop_lvl - 1] = pc->p->exec_tail; + break; + case TGSI_OPCODE_CEIL: for (c = 0; c < 4; c++) { if (!(mask & (1 << c))) continue; - emit_mov(pc, dst[c], temp); + emit_cvt(pc, dst[c], src[0][c], -1, + CVTOP_CEIL, CVT_F32_F32 | CVT_RI); + } + break; + case TGSI_OPCODE_COS: + if (mask & 8) { + emit_precossin(pc, temp, src[0][3]); + emit_flop(pc, 5, dst[3], temp); + if (!(mask &= 7)) + break; + if (temp == dst[3]) + temp = brdc = temp_temp(pc); } + emit_precossin(pc, temp, src[0][0]); + emit_flop(pc, 5, brdc, temp); break; case TGSI_OPCODE_DP3: - temp = temp_temp(pc); emit_mul(pc, temp, src[0][0], src[1][0]); emit_mad(pc, temp, src[0][1], src[1][1], temp); - emit_mad(pc, temp, src[0][2], src[1][2], temp); - for (c = 0; c < 4; c++) { - if (!(mask & (1 << c))) - continue; - emit_mov(pc, dst[c], temp); - } + emit_mad(pc, brdc, src[0][2], src[1][2], temp); break; case TGSI_OPCODE_DP4: - temp = temp_temp(pc); emit_mul(pc, temp, src[0][0], src[1][0]); emit_mad(pc, temp, src[0][1], src[1][1], temp); emit_mad(pc, temp, src[0][2], src[1][2], temp); - emit_mad(pc, temp, src[0][3], src[1][3], temp); - for (c = 0; c < 4; c++) { - if (!(mask & (1 << c))) - continue; - emit_mov(pc, dst[c], temp); - } + emit_mad(pc, brdc, src[0][3], src[1][3], temp); break; case TGSI_OPCODE_DPH: - temp = temp_temp(pc); emit_mul(pc, temp, src[0][0], src[1][0]); emit_mad(pc, temp, src[0][1], src[1][1], temp); emit_mad(pc, temp, src[0][2], src[1][2], temp); - emit_add(pc, temp, src[1][3], temp); - for (c = 0; c < 4; c++) { - if (!(mask & (1 << c))) - continue; - emit_mov(pc, dst[c], temp); - } + emit_add(pc, brdc, src[1][3], temp); break; case TGSI_OPCODE_DST: - { - struct nv50_reg *one = alloc_immd(pc, 1.0); - if (mask & (1 << 0)) - emit_mov(pc, dst[0], one); if (mask & (1 << 1)) emit_mul(pc, dst[1], src[0][1], src[1][1]); if (mask & (1 << 2)) emit_mov(pc, dst[2], src[0][2]); if (mask & (1 << 3)) emit_mov(pc, dst[3], src[1][3]); - FREE(one); - } + if (mask & (1 << 0)) + emit_mov_immdval(pc, dst[0], 1.0f); + break; + case TGSI_OPCODE_ELSE: + emit_branch(pc, -1, 0, NULL); + pc->if_insn[--pc->if_lvl]->param.index = pc->p->exec_size; + pc->if_insn[pc->if_lvl++] = pc->p->exec_tail; + break; + case TGSI_OPCODE_ENDIF: + pc->if_insn[--pc->if_lvl]->param.index = pc->p->exec_size; + + /* try to replace branch over 1 insn with a predicated insn */ + if (nv50_kill_branch(pc) == TRUE) + break; + + if (pc->br_join[pc->if_lvl]) { + pc->br_join[pc->if_lvl]->param.index = pc->p->exec_size; + pc->br_join[pc->if_lvl] = NULL; + } + /* emit a NOP as join point, we could set it on the next + * one, but would have to make sure it is long and !immd + */ + emit_nop(pc); + pc->p->exec_tail->inst[1] |= 2; + break; + case TGSI_OPCODE_ENDLOOP: + emit_branch(pc, -1, 0, NULL); + pc->p->exec_tail->param.index = pc->loop_pos[--pc->loop_lvl]; + pc->br_loop[pc->loop_lvl]->param.index = pc->p->exec_size; break; case TGSI_OPCODE_EX2: - temp = temp_temp(pc); emit_preex2(pc, temp, src[0][0]); - emit_flop(pc, 6, temp, temp); - for (c = 0; c < 4; c++) { - if (!(mask & (1 << c))) - continue; - emit_mov(pc, dst[c], temp); - } + emit_flop(pc, 6, brdc, temp); break; case TGSI_OPCODE_FLR: for (c = 0; c < 4; c++) { @@ -1445,24 +1734,24 @@ nv50_program_tx_insn(struct nv50_pc *pc, const union tgsi_full_token *tok) emit_sub(pc, dst[c], src[0][c], temp); } break; + case TGSI_OPCODE_IF: + /* emitting a join_at may not be necessary */ + assert(pc->if_lvl < MAX_IF_DEPTH); + set_pred_wr(pc, 1, 0, pc->if_cond); + emit_branch(pc, 0, 2, &pc->br_join[pc->if_lvl]); + pc->if_insn[pc->if_lvl++] = pc->p->exec_tail; + break; case TGSI_OPCODE_KIL: emit_kil(pc, src[0][0]); emit_kil(pc, src[0][1]); emit_kil(pc, src[0][2]); emit_kil(pc, src[0][3]); - pc->p->cfg.fp.regs[2] |= 0x00100000; break; case TGSI_OPCODE_LIT: emit_lit(pc, &dst[0], mask, &src[0][0]); break; case TGSI_OPCODE_LG2: - temp = temp_temp(pc); - emit_flop(pc, 3, temp, src[0][0]); - for (c = 0; c < 4; c++) { - if (!(mask & (1 << c))) - continue; - emit_mov(pc, dst[c], temp); - } + emit_flop(pc, 3, brdc, src[0][0]); break; case TGSI_OPCODE_LRP: temp = temp_temp(pc); @@ -1510,31 +1799,18 @@ nv50_program_tx_insn(struct nv50_pc *pc, const union tgsi_full_token *tok) } break; case TGSI_OPCODE_POW: - temp = temp_temp(pc); - emit_pow(pc, temp, src[0][0], src[1][0]); - for (c = 0; c < 4; c++) { - if (!(mask & (1 << c))) - continue; - emit_mov(pc, dst[c], temp); - } + emit_pow(pc, brdc, src[0][0], src[1][0]); break; case TGSI_OPCODE_RCP: - for (c = 3; c >= 0; c--) { - if (!(mask & (1 << c))) - continue; - emit_flop(pc, 0, dst[c], src[0][0]); - } + emit_flop(pc, 0, brdc, src[0][0]); break; case TGSI_OPCODE_RSQ: - for (c = 3; c >= 0; c--) { - if (!(mask & (1 << c))) - continue; - emit_flop(pc, 2, dst[c], src[0][0]); - } + emit_flop(pc, 2, brdc, src[0][0]); break; case TGSI_OPCODE_SCS: temp = temp_temp(pc); - emit_precossin(pc, temp, src[0][0]); + if (mask & 3) + emit_precossin(pc, temp, src[0][0]); if (mask & (1 << 0)) emit_flop(pc, 5, dst[0], temp); if (mask & (1 << 1)) @@ -1544,28 +1820,29 @@ nv50_program_tx_insn(struct nv50_pc *pc, const union tgsi_full_token *tok) if (mask & (1 << 3)) emit_mov_immdval(pc, dst[3], 1.0); break; - case TGSI_OPCODE_SGE: - for (c = 0; c < 4; c++) { - if (!(mask & (1 << c))) - continue; - emit_set(pc, 6, dst[c], src[0][c], src[1][c]); - } - break; case TGSI_OPCODE_SIN: - temp = temp_temp(pc); - emit_precossin(pc, temp, src[0][0]); - emit_flop(pc, 4, temp, temp); - for (c = 0; c < 4; c++) { - if (!(mask & (1 << c))) - continue; - emit_mov(pc, dst[c], temp); + if (mask & 8) { + emit_precossin(pc, temp, src[0][3]); + emit_flop(pc, 4, dst[3], temp); + if (!(mask &= 7)) + break; + if (temp == dst[3]) + temp = brdc = temp_temp(pc); } + emit_precossin(pc, temp, src[0][0]); + emit_flop(pc, 4, brdc, temp); break; case TGSI_OPCODE_SLT: + case TGSI_OPCODE_SGE: + case TGSI_OPCODE_SEQ: + case TGSI_OPCODE_SGT: + case TGSI_OPCODE_SLE: + case TGSI_OPCODE_SNE: + i = map_tgsi_setop_cc(inst->Instruction.Opcode); for (c = 0; c < 4; c++) { if (!(mask & (1 << c))) continue; - emit_set(pc, 1, dst[c], src[0][c], src[1][c]); + emit_set(pc, i, dst[c], -1, src[0][c], src[1][c]); } break; case TGSI_OPCODE_SUB: @@ -1583,6 +1860,14 @@ nv50_program_tx_insn(struct nv50_pc *pc, const union tgsi_full_token *tok) emit_tex(pc, dst, mask, src[0], unit, inst->InstructionExtTexture.Texture, TRUE); break; + case TGSI_OPCODE_TRUNC: + for (c = 0; c < 4; c++) { + if (!(mask & (1 << c))) + continue; + emit_cvt(pc, dst[c], src[0][c], -1, + CVTOP_TRUNC, CVT_F32_F32 | CVT_RI); + } + break; case TGSI_OPCODE_XPD: temp = temp_temp(pc); if (mask & (1 << 0)) { @@ -1607,17 +1892,22 @@ nv50_program_tx_insn(struct nv50_pc *pc, const union tgsi_full_token *tok) return FALSE; } + if (brdc) { + if (sat) + emit_sat(pc, brdc, brdc); + for (c = 0; c < 4; c++) + if ((mask & (1 << c)) && dst[c] != brdc) + emit_mov(pc, dst[c], brdc); + } else if (sat) { for (c = 0; c < 4; c++) { if (!(mask & (1 << c))) continue; - emit_cvt(pc, rdst[c], dst[c], -1, CVTOP_SAT, - CVT_F32_F32); + /* in this case we saturate later */ + if (dst[c]->type == P_TEMP && dst[c]->index < 0) + continue; + emit_sat(pc, rdst[c], dst[c]); } - } else if (assimilate) { - for (c = 0; c < 4; c++) - if (rdst[c]) - assimilate_temp(pc, rdst[c], dst[c]); } for (i = 0; i < inst->Instruction.NumSrcRegs; i++) { @@ -1626,9 +1916,6 @@ nv50_program_tx_insn(struct nv50_pc *pc, const union tgsi_full_token *tok) continue; if (src[i][c]->index == -1 && src[i][c]->type == P_IMMD) FREE(src[i][c]); - else - if (src[i][c]->acc == pc->insn_cur) - release_hw(pc, src[i][c]); } } @@ -1636,180 +1923,271 @@ nv50_program_tx_insn(struct nv50_pc *pc, const union tgsi_full_token *tok) return TRUE; } -/* Adjust a bitmask that indicates what components of a source are used, - * we use this in tx_prep so we only load interpolants that are needed. - */ -static void -insn_adjust_mask(const struct tgsi_full_instruction *insn, unsigned *mask) -{ - const struct tgsi_instruction_ext_texture *tex; - - switch (insn->Instruction.Opcode) { - case TGSI_OPCODE_DP3: - *mask = 0x7; - break; - case TGSI_OPCODE_DP4: - case TGSI_OPCODE_DPH: - *mask = 0xF; - break; - case TGSI_OPCODE_LIT: - *mask = 0xB; - break; - case TGSI_OPCODE_RCP: - case TGSI_OPCODE_RSQ: - *mask = 0x1; - break; - case TGSI_OPCODE_TEX: - case TGSI_OPCODE_TXP: - assert(insn->Instruction.Extended); - tex = &insn->InstructionExtTexture; - - *mask = 0x7; - if (tex->Texture == TGSI_TEXTURE_1D) - *mask = 0x1; - else - if (tex->Texture == TGSI_TEXTURE_2D) - *mask = 0x3; - - if (insn->Instruction.Opcode == TGSI_OPCODE_TXP) - *mask |= 0x8; - break; - default: - break; - } -} - static void -prep_inspect_insn(struct nv50_pc *pc, const union tgsi_full_token *tok, - unsigned *r_usage[2]) +prep_inspect_insn(struct nv50_pc *pc, const struct tgsi_full_instruction *insn) { - const struct tgsi_full_instruction *insn; + struct nv50_reg *reg = NULL; const struct tgsi_full_src_register *src; const struct tgsi_dst_register *dst; + unsigned i, c, k, mask; - unsigned i, c, k, n, mask, *acc_p; - - insn = &tok->FullInstruction; dst = &insn->FullDstRegisters[0].DstRegister; mask = dst->WriteMask; - if (!r_usage[0]) - r_usage[0] = CALLOC(pc->temp_nr * 4, sizeof(unsigned)); - if (!r_usage[1]) - r_usage[1] = CALLOC(pc->attr_nr * 4, sizeof(unsigned)); + if (dst->File == TGSI_FILE_TEMPORARY) + reg = pc->temp; + else + if (dst->File == TGSI_FILE_OUTPUT) + reg = pc->result; - if (dst->File == TGSI_FILE_TEMPORARY) { + if (reg) { for (c = 0; c < 4; c++) { if (!(mask & (1 << c))) continue; - r_usage[0][dst->Index * 4 + c] = pc->insn_nr; + reg[dst->Index * 4 + c].acc = pc->insn_nr; } } for (i = 0; i < insn->Instruction.NumSrcRegs; i++) { src = &insn->FullSrcRegisters[i]; - switch (src->SrcRegister.File) { - case TGSI_FILE_TEMPORARY: - acc_p = r_usage[0]; - break; - case TGSI_FILE_INPUT: - acc_p = r_usage[1]; - break; - default: + if (src->SrcRegister.File == TGSI_FILE_TEMPORARY) + reg = pc->temp; + else + if (src->SrcRegister.File == TGSI_FILE_INPUT) + reg = pc->attr; + else continue; - } - insn_adjust_mask(insn, &mask); + mask = nv50_tgsi_src_mask(insn, i); for (c = 0; c < 4; c++) { if (!(mask & (1 << c))) continue; - k = tgsi_util_get_full_src_register_extswizzle(src, c); - switch (k) { - case TGSI_EXTSWIZZLE_X: - case TGSI_EXTSWIZZLE_Y: - case TGSI_EXTSWIZZLE_Z: - case TGSI_EXTSWIZZLE_W: - n = src->SrcRegister.Index * 4 + k; - acc_p[n] = pc->insn_nr; - break; - default: - break; - } + + if (k > TGSI_EXTSWIZZLE_W) + continue; + + reg[src->SrcRegister.Index * 4 + k].acc = pc->insn_nr; } } } +/* Returns a bitmask indicating which dst components need to be + * written to temporaries first to avoid 'corrupting' sources. + * + * m[i] (out) indicate component to write in the i-th position + * rdep[c] (in) bitmasks of dst[i] that require dst[c] as source + */ +static unsigned +nv50_revdep_reorder(unsigned m[4], unsigned rdep[4]) +{ + unsigned i, c, x, unsafe; + + for (c = 0; c < 4; c++) + m[c] = c; + + /* Swap as long as a dst component written earlier is depended on + * by one written later, but the next one isn't depended on by it. + */ + for (c = 0; c < 3; c++) { + if (rdep[m[c + 1]] & (1 << m[c])) + continue; /* if next one is depended on by us */ + for (i = c + 1; i < 4; i++) + /* if we are depended on by a later one */ + if (rdep[m[c]] & (1 << m[i])) + break; + if (i == 4) + continue; + /* now, swap */ + x = m[c]; + m[c] = m[c + 1]; + m[c + 1] = x; + + /* restart */ + c = 0; + } + + /* mark dependencies that could not be resolved by reordering */ + for (i = 0; i < 3; ++i) + for (c = i + 1; c < 4; ++c) + if (rdep[m[i]] & (1 << m[c])) + unsafe |= (1 << i); + + /* NOTE: $unsafe is with respect to order, not component */ + return unsafe; +} + +/* Select a suitable dst register for broadcasting scalar results, + * or return NULL if we have to allocate an extra TEMP. + * + * If e.g. only 1 component is written, we may also emit the final + * result to a write-only register. + */ +static struct nv50_reg * +tgsi_broadcast_dst(struct nv50_pc *pc, + const struct tgsi_full_dst_register *fd, unsigned mask) +{ + if (fd->DstRegister.File == TGSI_FILE_TEMPORARY) { + int c = ffs(~mask & fd->DstRegister.WriteMask); + if (c) + return tgsi_dst(pc, c - 1, fd); + } else { + int c = ffs(fd->DstRegister.WriteMask) - 1; + if ((1 << c) == fd->DstRegister.WriteMask) + return tgsi_dst(pc, c, fd); + } + + return NULL; +} + +/* Scan source swizzles and return a bitmask indicating dst regs that + * also occur among the src regs, and fill rdep for nv50_revdep_reoder. + */ static unsigned -load_fp_attrib(struct nv50_pc *pc, int i, unsigned *acc, int *mid, - int *aid, int *p_oid) +nv50_tgsi_scan_swizzle(const struct tgsi_full_instruction *insn, + unsigned rdep[4]) { - struct nv50_reg *iv; - int oid, c, n; - unsigned mask = 0; + const struct tgsi_full_dst_register *fd = &insn->FullDstRegisters[0]; + const struct tgsi_full_src_register *fs; + unsigned i, deqs = 0; - iv = (pc->interp_mode[i] & INTERP_CENTROID) ? pc->iv_c : pc->iv_p; + for (i = 0; i < 4; ++i) + rdep[i] = 0; - for (c = 0, n = i * 4; c < 4; c++, n++) { - oid = (*p_oid)++; - pc->attr[n].type = P_TEMP; - pc->attr[n].index = i; + for (i = 0; i < insn->Instruction.NumSrcRegs; i++) { + unsigned chn, mask = nv50_tgsi_src_mask(insn, i); + boolean neg_supp = negate_supported(insn, i); - if (pc->attr[n].acc == acc[n]) + fs = &insn->FullSrcRegisters[i]; + if (fs->SrcRegister.File != fd->DstRegister.File || + fs->SrcRegister.Index != fd->DstRegister.Index) continue; - mask |= (1 << c); - pc->attr[n].acc = acc[n]; - pc->attr[n].rhw = pc->attr[n].hw = -1; - alloc_reg(pc, &pc->attr[n]); + for (chn = 0; chn < 4; ++chn) { + unsigned s, c; + + if (!(mask & (1 << chn))) /* src is not read */ + continue; + c = tgsi_util_get_full_src_register_extswizzle(fs, chn); + s = tgsi_util_get_full_src_register_sign_mode(fs, chn); - pc->attr[n].rhw = (*aid)++; - emit_interp(pc, &pc->attr[n], iv, pc->interp_mode[i]); + if (c > TGSI_EXTSWIZZLE_W || + !(fd->DstRegister.WriteMask & (1 << c))) + continue; - pc->p->cfg.fp.map[(*mid) / 4] |= oid << (8 * ((*mid) % 4)); - (*mid)++; - pc->p->cfg.fp.regs[1] += 0x00010001; + /* no danger if src is copied to TEMP first */ + if ((s != TGSI_UTIL_SIGN_KEEP) && + (s != TGSI_UTIL_SIGN_TOGGLE || !neg_supp)) + continue; + + rdep[c] |= nv50_tgsi_dst_revdep( + insn->Instruction.Opcode, i, chn); + deqs |= (1 << c); + } } - return mask; + return deqs; } static boolean -nv50_program_tx_prep(struct nv50_pc *pc) +nv50_tgsi_insn(struct nv50_pc *pc, const union tgsi_full_token *tok) { - struct tgsi_parse_context p; - boolean ret = FALSE; - unsigned i, c; - unsigned fcol, bcol, fcrd, depr; + struct tgsi_full_instruction insn = tok->FullInstruction; + const struct tgsi_full_dst_register *fd; + unsigned i, deqs, rdep[4], m[4]; + + fd = &tok->FullInstruction.FullDstRegisters[0]; + deqs = nv50_tgsi_scan_swizzle(&insn, rdep); + + if (is_scalar_op(insn.Instruction.Opcode)) { + pc->r_brdc = tgsi_broadcast_dst(pc, fd, deqs); + if (!pc->r_brdc) + pc->r_brdc = temp_temp(pc); + return nv50_program_tx_insn(pc, &insn); + } + pc->r_brdc = NULL; + + if (!deqs) + return nv50_program_tx_insn(pc, &insn); + + deqs = nv50_revdep_reorder(m, rdep); + + for (i = 0; i < 4; ++i) { + assert(pc->r_dst[m[i]] == NULL); + + insn.FullDstRegisters[0].DstRegister.WriteMask = + fd->DstRegister.WriteMask & (1 << m[i]); + + if (!insn.FullDstRegisters[0].DstRegister.WriteMask) + continue; + + if (deqs & (1 << i)) + pc->r_dst[m[i]] = alloc_temp(pc, NULL); + + if (!nv50_program_tx_insn(pc, &insn)) + return FALSE; + } - /* count (centroid) perspective interpolations */ - unsigned centroid_loads = 0; - unsigned perspect_loads = 0; + for (i = 0; i < 4; i++) { + struct nv50_reg *reg = pc->r_dst[i]; + if (!reg) + continue; + pc->r_dst[i] = NULL; + + if (insn.Instruction.Saturate == TGSI_SAT_ZERO_ONE) + emit_sat(pc, tgsi_dst(pc, i, fd), reg); + else + emit_mov(pc, tgsi_dst(pc, i, fd), reg); + free_temp(pc, reg); + } - /* track register access for temps and attrs */ - unsigned *r_usage[2]; - r_usage[0] = NULL; - r_usage[1] = NULL; + return TRUE; +} - depr = fcol = bcol = fcrd = 0xffff; +static void +load_interpolant(struct nv50_pc *pc, struct nv50_reg *reg) +{ + struct nv50_reg *iv, **ppiv; + unsigned mode = pc->interp_mode[reg->index]; - if (pc->p->type == PIPE_SHADER_FRAGMENT) { - pc->p->cfg.fp.regs[0] = 0x01000404; - pc->p->cfg.fp.regs[1] = 0x00000400; + ppiv = (mode & INTERP_CENTROID) ? &pc->iv_c : &pc->iv_p; + iv = *ppiv; + + if ((mode & INTERP_PERSPECTIVE) && !iv) { + iv = *ppiv = alloc_temp(pc, NULL); + iv->rhw = popcnt4(pc->p->cfg.regs[1] >> 24) - 1; + + emit_interp(pc, iv, NULL, mode & INTERP_CENTROID); + emit_flop(pc, 0, iv, iv); + + /* XXX: when loading interpolants dynamically, move these + * to the program head, or make sure it can't be skipped. + */ } - tgsi_parse_init(&p, pc->p->pipe.tokens); - while (!tgsi_parse_end_of_tokens(&p)) { - const union tgsi_full_token *tok = &p.FullToken; + emit_interp(pc, reg, iv, mode); +} + +static boolean +nv50_program_tx_prep(struct nv50_pc *pc) +{ + struct tgsi_parse_context tp; + struct nv50_program *p = pc->p; + boolean ret = FALSE; + unsigned i, c, flat_nr = 0; + + tgsi_parse_init(&tp, pc->p->pipe.tokens); + while (!tgsi_parse_end_of_tokens(&tp)) { + const union tgsi_full_token *tok = &tp.FullToken; - tgsi_parse_token(&p); + tgsi_parse_token(&tp); switch (tok->Token.Type) { case TGSI_TOKEN_TYPE_IMMEDIATE: { const struct tgsi_full_immediate *imm = - &p.FullToken.FullImmediate; + &tp.FullToken.FullImmediate; ctor_immd(pc, imm->u[0].Float, imm->u[1].Float, @@ -1820,78 +2198,61 @@ nv50_program_tx_prep(struct nv50_pc *pc) case TGSI_TOKEN_TYPE_DECLARATION: { const struct tgsi_full_declaration *d; - unsigned last, first, mode; + unsigned si, last, first, mode; - d = &p.FullToken.FullDeclaration; + d = &tp.FullToken.FullDeclaration; first = d->DeclarationRange.First; last = d->DeclarationRange.Last; switch (d->Declaration.File) { case TGSI_FILE_TEMPORARY: - if (pc->temp_nr < (last + 1)) - pc->temp_nr = last + 1; break; case TGSI_FILE_OUTPUT: - if (pc->result_nr < (last + 1)) - pc->result_nr = last + 1; - - if (!d->Declaration.Semantic) + if (!d->Declaration.Semantic || + p->type == PIPE_SHADER_FRAGMENT) break; + si = d->Semantic.SemanticIndex; switch (d->Semantic.SemanticName) { - case TGSI_SEMANTIC_POSITION: - depr = first; - pc->p->cfg.fp.regs[2] |= 0x00000100; - pc->p->cfg.fp.regs[3] |= 0x00000011; + case TGSI_SEMANTIC_BCOLOR: + p->cfg.two_side[si].hw = first; + if (p->cfg.io_nr > first) + p->cfg.io_nr = first; + break; + case TGSI_SEMANTIC_PSIZE: + p->cfg.psiz = first; + if (p->cfg.io_nr > first) + p->cfg.io_nr = first; break; + /* + case TGSI_SEMANTIC_CLIP_DISTANCE: + p->cfg.clpd = MIN2(p->cfg.clpd, first); + break; + */ default: break; } - break; case TGSI_FILE_INPUT: { - if (pc->attr_nr < (last + 1)) - pc->attr_nr = last + 1; - - if (pc->p->type != PIPE_SHADER_FRAGMENT) + if (p->type != PIPE_SHADER_FRAGMENT) break; switch (d->Declaration.Interpolate) { case TGSI_INTERPOLATE_CONSTANT: mode = INTERP_FLAT; + flat_nr++; break; case TGSI_INTERPOLATE_PERSPECTIVE: mode = INTERP_PERSPECTIVE; + p->cfg.regs[1] |= 0x08 << 24; break; default: mode = INTERP_LINEAR; break; } - - if (d->Declaration.Semantic) { - switch (d->Semantic.SemanticName) { - case TGSI_SEMANTIC_POSITION: - fcrd = first; - break; - case TGSI_SEMANTIC_COLOR: - fcol = first; - mode = INTERP_PERSPECTIVE; - break; - case TGSI_SEMANTIC_BCOLOR: - bcol = first; - mode = INTERP_PERSPECTIVE; - break; - } - } - - if (d->Declaration.Centroid) { + if (d->Declaration.Centroid) mode |= INTERP_CENTROID; - if (mode & INTERP_PERSPECTIVE) - centroid_loads++; - } else - if (mode & INTERP_PERSPECTIVE) - perspect_loads++; assert(last < 32); for (i = first; i <= last; i++) @@ -1899,8 +2260,6 @@ nv50_program_tx_prep(struct nv50_pc *pc) } break; case TGSI_FILE_CONSTANT: - if (pc->param_nr < (last + 1)) - pc->param_nr = last + 1; break; case TGSI_FILE_SAMPLER: break; @@ -1913,182 +2272,157 @@ nv50_program_tx_prep(struct nv50_pc *pc) break; case TGSI_TOKEN_TYPE_INSTRUCTION: pc->insn_nr++; - prep_inspect_insn(pc, tok, r_usage); + prep_inspect_insn(pc, &tok->FullInstruction); break; default: break; } } - if (pc->temp_nr) { - pc->temp = CALLOC(pc->temp_nr * 4, sizeof(struct nv50_reg)); - if (!pc->temp) - goto out_err; + if (p->type == PIPE_SHADER_VERTEX) { + int rid = 0; - for (i = 0; i < pc->temp_nr; i++) { - for (c = 0; c < 4; c++) { - pc->temp[i*4+c].type = P_TEMP; - pc->temp[i*4+c].hw = -1; - pc->temp[i*4+c].rhw = -1; - pc->temp[i*4+c].index = i; - pc->temp[i*4+c].acc = r_usage[0][i*4+c]; + for (i = 0; i < pc->attr_nr * 4; ++i) { + if (pc->attr[i].acc) { + pc->attr[i].hw = rid++; + p->cfg.attr[i / 32] |= 1 << (i % 32); } } - } - - if (pc->attr_nr) { - int oid = 4, mid = 4, aid = 0; - /* oid = VP output id - * aid = FP attribute/interpolant id - * mid = VP output mapping field ID - */ - pc->attr = CALLOC(pc->attr_nr * 4, sizeof(struct nv50_reg)); - if (!pc->attr) - goto out_err; - - if (pc->p->type == PIPE_SHADER_FRAGMENT) { - /* position should be loaded first */ - if (fcrd != 0xffff) { - unsigned mask; - mid = 0; - mask = load_fp_attrib(pc, fcrd, r_usage[1], - &mid, &aid, &oid); - oid = 0; - pc->p->cfg.fp.regs[1] |= (mask << 24); - pc->p->cfg.fp.map[0] = 0x04040404 * fcrd; - } - pc->p->cfg.fp.map[0] += 0x03020100; - - /* should do MAD fcrd.xy, fcrd, SOME_CONST, fcrd */ - - if (perspect_loads) { - pc->iv_p = alloc_temp(pc, NULL); - - if (!(pc->p->cfg.fp.regs[1] & 0x08000000)) { - pc->p->cfg.fp.regs[1] |= 0x08000000; - pc->iv_p->rhw = aid++; - emit_interp(pc, pc->iv_p, NULL, - INTERP_LINEAR); - emit_flop(pc, 0, pc->iv_p, pc->iv_p); - } else { - pc->iv_p->rhw = aid - 1; - emit_flop(pc, 0, pc->iv_p, - &pc->attr[fcrd * 4 + 3]); - } - } + for (i = 0, rid = 0; i < pc->result_nr; ++i) { + p->cfg.io[i].hw = rid; + p->cfg.io[i].id_vp = i; - if (centroid_loads) { - pc->iv_c = alloc_temp(pc, NULL); - pc->iv_c->rhw = pc->iv_p ? aid - 1 : aid++; - emit_interp(pc, pc->iv_c, NULL, - INTERP_CENTROID); - emit_flop(pc, 0, pc->iv_c, pc->iv_c); - pc->p->cfg.fp.regs[1] |= 0x08000000; + for (c = 0; c < 4; ++c) { + int n = i * 4 + c; + if (!pc->result[n].acc) + continue; + pc->result[n].hw = rid++; + p->cfg.io[i].mask |= 1 << c; } + } - for (c = 0; c < 4; c++) { - /* I don't know what these values do, but - * let's set them like the blob does: - */ - if (fcol != 0xffff && r_usage[1][fcol * 4 + c]) - pc->p->cfg.fp.regs[0] += 0x00010000; - if (bcol != 0xffff && r_usage[1][bcol * 4 + c]) - pc->p->cfg.fp.regs[0] += 0x00010000; - } + for (c = 0; c < 2; ++c) + if (p->cfg.two_side[c].hw < 0x40) + p->cfg.two_side[c] = p->cfg.io[ + p->cfg.two_side[c].hw]; - for (i = 0; i < pc->attr_nr; i++) - load_fp_attrib(pc, i, r_usage[1], - &mid, &aid, &oid); + if (p->cfg.psiz < 0x40) + p->cfg.psiz = p->cfg.io[p->cfg.psiz].hw; + } else + if (p->type == PIPE_SHADER_FRAGMENT) { + int rid, aid; + unsigned n = 0, m = pc->attr_nr - flat_nr; - if (pc->iv_p) - free_temp(pc, pc->iv_p); - if (pc->iv_c) - free_temp(pc, pc->iv_c); + int base = (TGSI_SEMANTIC_POSITION == + p->info.input_semantic_name[0]) ? 0 : 1; - pc->p->cfg.fp.high_map = (mid / 4); - pc->p->cfg.fp.high_map += ((mid % 4) ? 1 : 0); - } else { - /* vertex program */ - for (i = 0; i < pc->attr_nr * 4; i++) { - pc->p->cfg.vp.attr[aid / 32] |= - (1 << (aid % 32)); - pc->attr[i].type = P_ATTR; - pc->attr[i].hw = aid++; - pc->attr[i].index = i / 4; + /* non-flat interpolants have to be mapped to + * the lower hardware IDs, so sort them: + */ + for (i = 0; i < pc->attr_nr; i++) { + if (pc->interp_mode[i] == INTERP_FLAT) { + p->cfg.io[m].id_vp = i + base; + p->cfg.io[m++].id_fp = i; + } else { + if (!(pc->interp_mode[i] & INTERP_PERSPECTIVE)) + p->cfg.io[n].linear = TRUE; + p->cfg.io[n].id_vp = i + base; + p->cfg.io[n++].id_fp = i; } } - } - if (pc->result_nr) { - int rid = 0; + if (!base) /* set w-coordinate mask from perspective interp */ + p->cfg.io[0].mask |= p->cfg.regs[1] >> 24; - pc->result = CALLOC(pc->result_nr * 4, sizeof(struct nv50_reg)); - if (!pc->result) - goto out_err; + aid = popcnt4( /* if fcrd isn't contained in cfg.io */ + base ? (p->cfg.regs[1] >> 24) : p->cfg.io[0].mask); - for (i = 0; i < pc->result_nr; i++) { - for (c = 0; c < 4; c++) { - if (pc->p->type == PIPE_SHADER_FRAGMENT) { - pc->result[i*4+c].type = P_TEMP; - pc->result[i*4+c].hw = -1; - pc->result[i*4+c].rhw = (i == depr) ? - -1 : rid++; - } else { - pc->result[i*4+c].type = P_RESULT; - pc->result[i*4+c].hw = rid++; - } - pc->result[i*4+c].index = i; - } + for (n = 0; n < pc->attr_nr; ++n) { + p->cfg.io[n].hw = rid = aid; + i = p->cfg.io[n].id_fp; - if (pc->p->type == PIPE_SHADER_FRAGMENT && - depr != 0xffff) { - pc->result[depr * 4 + 2].rhw = - (pc->result_nr - 1) * 4; + for (c = 0; c < 4; ++c) { + if (!pc->attr[i * 4 + c].acc) + continue; + pc->attr[i * 4 + c].rhw = rid++; + p->cfg.io[n].mask |= 1 << c; + + load_interpolant(pc, &pc->attr[i * 4 + c]); } + aid += popcnt4(p->cfg.io[n].mask); } - } - if (pc->param_nr) { - int rid = 0; + if (!base) + p->cfg.regs[1] |= p->cfg.io[0].mask << 24; - pc->param = CALLOC(pc->param_nr * 4, sizeof(struct nv50_reg)); - if (!pc->param) - goto out_err; + m = popcnt4(p->cfg.regs[1] >> 24); + + /* set count of non-position inputs and of non-flat + * non-position inputs for FP_INTERPOLANT_CTRL + */ + p->cfg.regs[1] |= aid - m; + + if (flat_nr) { + i = p->cfg.io[pc->attr_nr - flat_nr].hw; + p->cfg.regs[1] |= (i - m) << 16; + } else + p->cfg.regs[1] |= p->cfg.regs[1] << 16; + + /* mark color semantic for light-twoside */ + n = 0x40; + for (i = 0; i < pc->attr_nr; i++) { + ubyte si, sn; - for (i = 0; i < pc->param_nr; i++) { - for (c = 0; c < 4; c++) { - pc->param[i*4+c].type = P_CONST; - pc->param[i*4+c].hw = rid++; - pc->param[i*4+c].index = i; + sn = p->info.input_semantic_name[p->cfg.io[i].id_fp]; + si = p->info.input_semantic_index[p->cfg.io[i].id_fp]; + + if (sn == TGSI_SEMANTIC_COLOR) { + p->cfg.two_side[si] = p->cfg.io[i]; + + /* increase colour count */ + p->cfg.regs[0] += popcnt4( + p->cfg.two_side[si].mask) << 16; + + n = MIN2(n, p->cfg.io[i].hw - m); } } + if (n < 0x40) + p->cfg.regs[0] += n; + + /* Initialize FP results: + * FragDepth is always first TGSI and last hw output + */ + i = p->info.writes_z ? 4 : 0; + for (rid = 0; i < pc->result_nr * 4; i++) + pc->result[i].rhw = rid++; + if (p->info.writes_z) + pc->result[2].rhw = rid; + + p->cfg.high_result = rid; } if (pc->immd_nr) { int rid = 0; - pc->immd = CALLOC(pc->immd_nr * 4, sizeof(struct nv50_reg)); + pc->immd = MALLOC(pc->immd_nr * 4 * sizeof(struct nv50_reg)); if (!pc->immd) goto out_err; for (i = 0; i < pc->immd_nr; i++) { - for (c = 0; c < 4; c++) { - pc->immd[i*4+c].type = P_IMMD; - pc->immd[i*4+c].hw = rid++; - pc->immd[i*4+c].index = i; - } + for (c = 0; c < 4; c++, rid++) + ctor_reg(&pc->immd[rid], P_IMMD, i, rid); } } ret = TRUE; out_err: - if (r_usage[0]) - FREE(r_usage[0]); - if (r_usage[1]) - FREE(r_usage[1]); + if (pc->iv_p) + free_temp(pc, pc->iv_p); + if (pc->iv_c) + free_temp(pc, pc->iv_c); - tgsi_parse_free(&p); + tgsi_parse_free(&tp); return ret; } @@ -2110,18 +2444,165 @@ free_nv50_pc(struct nv50_pc *pc) } static boolean +ctor_nv50_pc(struct nv50_pc *pc, struct nv50_program *p) +{ + int i, c; + unsigned rtype[2] = { P_ATTR, P_RESULT }; + + pc->p = p; + pc->temp_nr = p->info.file_max[TGSI_FILE_TEMPORARY] + 1; + pc->attr_nr = p->info.file_max[TGSI_FILE_INPUT] + 1; + pc->result_nr = p->info.file_max[TGSI_FILE_OUTPUT] + 1; + pc->param_nr = p->info.file_max[TGSI_FILE_CONSTANT] + 1; + + p->cfg.high_temp = 4; + + p->cfg.two_side[0].hw = 0x40; + p->cfg.two_side[1].hw = 0x40; + + switch (p->type) { + case PIPE_SHADER_VERTEX: + p->cfg.psiz = 0x40; + p->cfg.clpd = 0x40; + p->cfg.io_nr = pc->result_nr; + break; + case PIPE_SHADER_FRAGMENT: + rtype[0] = rtype[1] = P_TEMP; + + p->cfg.regs[0] = 0x01000004; + p->cfg.io_nr = pc->attr_nr; + + if (p->info.writes_z) { + p->cfg.regs[2] |= 0x00000100; + p->cfg.regs[3] |= 0x00000011; + } + if (p->info.uses_kill) + p->cfg.regs[2] |= 0x00100000; + break; + } + + if (pc->temp_nr) { + pc->temp = MALLOC(pc->temp_nr * 4 * sizeof(struct nv50_reg)); + if (!pc->temp) + return FALSE; + + for (i = 0; i < pc->temp_nr * 4; ++i) + ctor_reg(&pc->temp[i], P_TEMP, i / 4, -1); + } + + if (pc->attr_nr) { + pc->attr = MALLOC(pc->attr_nr * 4 * sizeof(struct nv50_reg)); + if (!pc->attr) + return FALSE; + + for (i = 0; i < pc->attr_nr * 4; ++i) + ctor_reg(&pc->attr[i], rtype[0], i / 4, -1); + } + + if (pc->result_nr) { + unsigned nr = pc->result_nr * 4; + + pc->result = MALLOC(nr * sizeof(struct nv50_reg)); + if (!pc->result) + return FALSE; + + for (i = 0; i < nr; ++i) + ctor_reg(&pc->result[i], rtype[1], i / 4, -1); + } + + if (pc->param_nr) { + int rid = 0; + + pc->param = MALLOC(pc->param_nr * 4 * sizeof(struct nv50_reg)); + if (!pc->param) + return FALSE; + + for (i = 0; i < pc->param_nr; ++i) + for (c = 0; c < 4; ++c, ++rid) + ctor_reg(&pc->param[rid], P_CONST, i, rid); + } + + return TRUE; +} + +static void +nv50_fp_move_results(struct nv50_pc *pc) +{ + struct nv50_reg reg; + unsigned i; + + ctor_reg(®, P_TEMP, -1, -1); + + for (i = 0; i < pc->result_nr * 4; ++i) { + if (pc->result[i].rhw < 0 || pc->result[i].hw < 0) + continue; + if (pc->result[i].rhw != pc->result[i].hw) { + reg.hw = pc->result[i].rhw; + emit_mov(pc, ®, &pc->result[i]); + } + } +} + +static void +nv50_program_fixup_insns(struct nv50_pc *pc) +{ + struct nv50_program_exec *e, *prev = NULL, **bra_list; + unsigned i, n, pos; + + bra_list = CALLOC(pc->p->exec_size, sizeof(struct nv50_program_exec *)); + + /* Collect branch instructions, we need to adjust their offsets + * when converting 32 bit instructions to 64 bit ones + */ + for (n = 0, e = pc->p->exec_head; e; e = e->next) + if (e->param.index >= 0 && !e->param.mask) + bra_list[n++] = e; + + /* Make sure we don't have any single 32 bit instructions. */ + for (e = pc->p->exec_head, pos = 0; e; e = e->next) { + pos += is_long(e) ? 2 : 1; + + if ((pos & 1) && (!e->next || is_long(e->next))) { + for (i = 0; i < n; ++i) + if (bra_list[i]->param.index >= pos) + bra_list[i]->param.index += 1; + convert_to_long(pc, e); + ++pos; + } + if (e->next) + prev = e; + } + + assert(!is_immd(pc->p->exec_head)); + assert(!is_immd(pc->p->exec_tail)); + + /* last instruction must be long so it can have the end bit set */ + if (!is_long(pc->p->exec_tail)) { + convert_to_long(pc, pc->p->exec_tail); + if (prev) + convert_to_long(pc, prev); + } + assert(!(pc->p->exec_tail->inst[1] & 2)); + /* set the end-bit */ + pc->p->exec_tail->inst[1] |= 1; + + FREE(bra_list); +} + +static boolean nv50_program_tx(struct nv50_program *p) { struct tgsi_parse_context parse; struct nv50_pc *pc; - unsigned k; boolean ret; pc = CALLOC_STRUCT(nv50_pc); if (!pc) return FALSE; - pc->p = p; - pc->p->cfg.high_temp = 4; + + ret = ctor_nv50_pc(pc, p); + if (ret == FALSE) + goto out_cleanup; ret = nv50_program_tx_prep(pc); if (ret == FALSE) @@ -2141,7 +2622,7 @@ nv50_program_tx(struct nv50_program *p) switch (tok->Token.Type) { case TGSI_TOKEN_TYPE_INSTRUCTION: ++pc->insn_cur; - ret = nv50_program_tx_insn(pc, tok); + ret = nv50_tgsi_insn(pc, tok); if (ret == FALSE) goto out_err; break; @@ -2150,48 +2631,10 @@ nv50_program_tx(struct nv50_program *p) } } - if (p->type == PIPE_SHADER_FRAGMENT) { - struct nv50_reg out; - - out.type = P_TEMP; - for (k = 0; k < pc->result_nr * 4; k++) { - if (pc->result[k].rhw == -1) - continue; - if (pc->result[k].hw != pc->result[k].rhw) { - out.hw = pc->result[k].rhw; - emit_mov(pc, &out, &pc->result[k]); - } - if (pc->p->cfg.high_result < (pc->result[k].rhw + 1)) - pc->p->cfg.high_result = pc->result[k].rhw + 1; - } - } - - /* look for single half instructions and make them long */ - struct nv50_program_exec *e, *e_prev; + if (pc->p->type == PIPE_SHADER_FRAGMENT) + nv50_fp_move_results(pc); - for (k = 0, e = pc->p->exec_head, e_prev = NULL; e; e = e->next) { - if (!is_long(e)) - k++; - - if (!e->next || is_long(e->next)) { - if (k & 1) - convert_to_long(pc, e); - k = 0; - } - - if (e->next) - e_prev = e; - } - - if (!is_long(pc->p->exec_tail)) { - /* this may occur if moving FP results */ - assert(e_prev && !is_long(e_prev)); - convert_to_long(pc, e_prev); - convert_to_long(pc, pc->p->exec_tail); - } - - assert(is_long(pc->p->exec_tail) && !is_immd(pc->p->exec_head)); - pc->p->exec_tail->inst[1] |= 0x00000001; + nv50_program_fixup_insns(pc); p->param_nr = pc->param_nr * 4; p->immd_nr = pc->immd_nr * 4; @@ -2258,30 +2701,19 @@ nv50_program_validate_data(struct nv50_context *nv50, struct nv50_program *p) p->immd_nr, NV50_CB_PMISC); } - if (!p->data[1] && p->param_nr) { - struct nouveau_resource *heap = - nv50->screen->parm_heap[p->type]; - - if (nouveau_resource_alloc(heap, p->param_nr, p, &p->data[1])) { - while (heap->next && heap->size < p->param_nr) { - struct nv50_program *evict = heap->next->priv; - nouveau_resource_free(&evict->data[1]); - } - - if (nouveau_resource_alloc(heap, p->param_nr, p, - &p->data[1])) - assert(0); - } - } + assert(p->param_nr <= 128); if (p->param_nr) { - unsigned cbuf = NV50_CB_PVP; + unsigned cb; float *map = pipe_buffer_map(pscreen, nv50->constbuf[p->type], PIPE_BUFFER_USAGE_CPU_READ); - if (p->type == PIPE_SHADER_FRAGMENT) - cbuf = NV50_CB_PFP; - nv50_program_upload_data(nv50, map, p->data[1]->start, - p->param_nr, cbuf); + + if (p->type == PIPE_SHADER_VERTEX) + cb = NV50_CB_PVP; + else + cb = NV50_CB_PFP; + + nv50_program_upload_data(nv50, map, 0, p->param_nr, cb); pipe_buffer_unmap(pscreen, nv50->constbuf[p->type]); } } @@ -2303,32 +2735,41 @@ nv50_program_validate_code(struct nv50_context *nv50, struct nv50_program *p) upload = TRUE; } - if ((p->data[0] && p->data[0]->start != p->data_start[0]) || - (p->data[1] && p->data[1]->start != p->data_start[1])) { - for (e = p->exec_head; e; e = e->next) { - unsigned ei, ci, bs; + if (p->data[0] && p->data[0]->start != p->data_start[0]) + upload = TRUE; - if (e->param.index < 0) - continue; - bs = (e->inst[1] >> 22) & 0x07; - assert(bs < 2); - ei = e->param.shift >> 5; - ci = e->param.index + p->data[bs]->start; + if (!upload) + return; + + for (e = p->exec_head; e; e = e->next) { + unsigned ei, ci, bs; + + if (e->param.index < 0) + continue; + + if (e->param.mask == 0) { + assert(!(e->param.index & 1)); + /* seem to be 8 byte steps */ + ei = (e->param.index >> 1) + 0 /* START_ID */; - e->inst[ei] &= ~e->param.mask; - e->inst[ei] |= (ci << e->param.shift); + e->inst[0] &= 0xf0000fff; + e->inst[0] |= ei << 12; + continue; } - if (p->data[0]) - p->data_start[0] = p->data[0]->start; - if (p->data[1]) - p->data_start[1] = p->data[1]->start; + bs = (e->inst[1] >> 22) & 0x07; + assert(bs < 2); + ei = e->param.shift >> 5; + ci = e->param.index; + if (bs == 0) + ci += p->data[bs]->start; - upload = TRUE; + e->inst[ei] &= ~e->param.mask; + e->inst[ei] |= (ci << e->param.shift); } - if (!upload) - return; + if (p->data[0]) + p->data_start[0] = p->data[0]->start; #ifdef NV50_PROGRAM_DUMP NOUVEAU_ERR("-------\n"); @@ -2402,8 +2843,8 @@ nv50_vertprog_validate(struct nv50_context *nv50) so_reloc (so, p->bo, 0, NOUVEAU_BO_VRAM | NOUVEAU_BO_RD | NOUVEAU_BO_LOW, 0, 0); so_method(so, tesla, NV50TCL_VP_ATTR_EN_0, 2); - so_data (so, p->cfg.vp.attr[0]); - so_data (so, p->cfg.vp.attr[1]); + so_data (so, p->cfg.attr[0]); + so_data (so, p->cfg.attr[1]); so_method(so, tesla, NV50TCL_VP_REG_ALLOC_RESULT, 1); so_data (so, p->cfg.high_result); so_method(so, tesla, NV50TCL_VP_RESULT_MAP_SIZE, 2); @@ -2421,7 +2862,6 @@ nv50_fragprog_validate(struct nv50_context *nv50) struct nouveau_grobj *tesla = nv50->screen->tesla; struct nv50_program *p = nv50->fragprog; struct nouveau_stateobj *so; - unsigned i; if (!p->translated) { nv50_program_validate(nv50, p); @@ -2438,29 +2878,186 @@ nv50_fragprog_validate(struct nv50_context *nv50) NOUVEAU_BO_HIGH, 0, 0); so_reloc (so, p->bo, 0, NOUVEAU_BO_VRAM | NOUVEAU_BO_RD | NOUVEAU_BO_LOW, 0, 0); - so_method(so, tesla, NV50TCL_MAP_SEMANTIC_0, 4); - so_data (so, p->cfg.fp.regs[0]); /* 0x01000404 / 0x00040404 */ - so_data (so, 0x00000004); - so_data (so, 0x00000000); - so_data (so, 0x00000000); - so_method(so, tesla, NV50TCL_VP_RESULT_MAP(0), p->cfg.fp.high_map); - for (i = 0; i < p->cfg.fp.high_map; i++) - so_data(so, p->cfg.fp.map[i]); - so_method(so, tesla, NV50TCL_FP_INTERPOLANT_CTRL, 2); - so_data (so, p->cfg.fp.regs[1]); /* 0x08040404 / 0x0f000401 */ + so_method(so, tesla, NV50TCL_FP_REG_ALLOC_TEMP, 1); so_data (so, p->cfg.high_temp); so_method(so, tesla, NV50TCL_FP_RESULT_COUNT, 1); so_data (so, p->cfg.high_result); so_method(so, tesla, NV50TCL_FP_CTRL_UNK19A8, 1); - so_data (so, p->cfg.fp.regs[2]); + so_data (so, p->cfg.regs[2]); so_method(so, tesla, NV50TCL_FP_CTRL_UNK196C, 1); - so_data (so, p->cfg.fp.regs[3]); + so_data (so, p->cfg.regs[3]); so_method(so, tesla, NV50TCL_FP_START_ID, 1); so_data (so, 0); /* program start offset */ so_ref(so, &nv50->state.fragprog); so_ref(NULL, &so); } +static void +nv50_pntc_replace(struct nv50_context *nv50, uint32_t pntc[8], unsigned base) +{ + struct nv50_program *fp = nv50->fragprog; + struct nv50_program *vp = nv50->vertprog; + unsigned i, c, m = base; + + /* XXX: This can't work correctly in all cases yet, we either + * have to create TGSI_SEMANTIC_PNTC or sprite_coord_mode has + * to be per FP input instead of per VP output + */ + memset(pntc, 0, 8 * sizeof(uint32_t)); + + for (i = 0; i < fp->cfg.io_nr; i++) { + uint8_t sn, si; + uint8_t j = fp->cfg.io[i].id_vp, k = fp->cfg.io[i].id_fp; + unsigned n = popcnt4(fp->cfg.io[i].mask); + + if (fp->info.input_semantic_name[k] != TGSI_SEMANTIC_GENERIC) { + m += n; + continue; + } + + sn = vp->info.input_semantic_name[j]; + si = vp->info.input_semantic_index[j]; + + if (j < fp->cfg.io_nr && sn == TGSI_SEMANTIC_GENERIC) { + ubyte mode = + nv50->rasterizer->pipe.sprite_coord_mode[si]; + + if (mode == PIPE_SPRITE_COORD_NONE) { + m += n; + continue; + } + } + + /* this is either PointCoord or replaced by sprite coords */ + for (c = 0; c < 4; c++) { + if (!(fp->cfg.io[i].mask & (1 << c))) + continue; + pntc[m / 8] |= (c + 1) << ((m % 8) * 4); + ++m; + } + } +} + +static int +nv50_sreg4_map(uint32_t *p_map, int mid, uint32_t lin[4], + struct nv50_sreg4 *fpi, struct nv50_sreg4 *vpo) +{ + int c; + uint8_t mv = vpo->mask, mf = fpi->mask, oid = vpo->hw; + uint8_t *map = (uint8_t *)p_map; + + for (c = 0; c < 4; ++c) { + if (mf & 1) { + if (fpi->linear == TRUE) + lin[mid / 32] |= 1 << (mid % 32); + map[mid++] = (mv & 1) ? oid : ((c == 3) ? 0x41 : 0x40); + } + + oid += mv & 1; + mf >>= 1; + mv >>= 1; + } + + return mid; +} + +void +nv50_linkage_validate(struct nv50_context *nv50) +{ + struct nouveau_grobj *tesla = nv50->screen->tesla; + struct nv50_program *vp = nv50->vertprog; + struct nv50_program *fp = nv50->fragprog; + struct nouveau_stateobj *so; + struct nv50_sreg4 dummy, *vpo; + int i, n, c, m = 0; + uint32_t map[16], lin[4], reg[5], pcrd[8]; + + memset(map, 0, sizeof(map)); + memset(lin, 0, sizeof(lin)); + + reg[1] = 0x00000004; /* low and high clip distance map ids */ + reg[2] = 0x00000000; /* layer index map id (disabled, GP only) */ + reg[3] = 0x00000000; /* point size map id & enable */ + reg[0] = fp->cfg.regs[0]; /* colour semantic reg */ + reg[4] = fp->cfg.regs[1]; /* interpolant info */ + + dummy.linear = FALSE; + dummy.mask = 0xf; /* map all components of HPOS */ + m = nv50_sreg4_map(map, m, lin, &dummy, &vp->cfg.io[0]); + + dummy.mask = 0x0; + + if (vp->cfg.clpd < 0x40) { + for (c = 0; c < vp->cfg.clpd_nr; ++c) + map[m++] = vp->cfg.clpd + c; + reg[1] = (m << 8); + } + + reg[0] |= m << 8; /* adjust BFC0 id */ + + /* if light_twoside is active, it seems FFC0_ID == BFC0_ID is bad */ + if (nv50->rasterizer->pipe.light_twoside) { + vpo = &vp->cfg.two_side[0]; + + m = nv50_sreg4_map(map, m, lin, &fp->cfg.two_side[0], &vpo[0]); + m = nv50_sreg4_map(map, m, lin, &fp->cfg.two_side[1], &vpo[1]); + } + + reg[0] += m - 4; /* adjust FFC0 id */ + reg[4] |= m << 8; /* set mid where 'normal' FP inputs start */ + + i = 0; + if (fp->info.input_semantic_name[0] == TGSI_SEMANTIC_POSITION) + i = 1; + for (; i < fp->cfg.io_nr; i++) { + ubyte sn = fp->info.input_semantic_name[fp->cfg.io[i].id_fp]; + ubyte si = fp->info.input_semantic_index[fp->cfg.io[i].id_fp]; + + n = fp->cfg.io[i].id_vp; + if (n >= vp->cfg.io_nr || + vp->info.output_semantic_name[n] != sn || + vp->info.output_semantic_index[n] != si) + vpo = &dummy; + else + vpo = &vp->cfg.io[n]; + + m = nv50_sreg4_map(map, m, lin, &fp->cfg.io[i], vpo); + } + + if (nv50->rasterizer->pipe.point_size_per_vertex) { + map[m / 4] |= vp->cfg.psiz << ((m % 4) * 8); + reg[3] = (m++ << 4) | 1; + } + + /* now fill the stateobj */ + so = so_new(64, 0); + + n = (m + 3) / 4; + so_method(so, tesla, NV50TCL_VP_RESULT_MAP_SIZE, 1); + so_data (so, m); + so_method(so, tesla, NV50TCL_VP_RESULT_MAP(0), n); + so_datap (so, map, n); + + so_method(so, tesla, NV50TCL_MAP_SEMANTIC_0, 4); + so_datap (so, reg, 4); + + so_method(so, tesla, NV50TCL_FP_INTERPOLANT_CTRL, 1); + so_data (so, reg[4]); + + so_method(so, tesla, 0x1540, 4); + so_datap (so, lin, 4); + + if (nv50->rasterizer->pipe.point_sprite) { + nv50_pntc_replace(nv50, pcrd, (reg[4] >> 8) & 0xff); + + so_method(so, tesla, NV50TCL_POINT_COORD_REPLACE_MAP(0), 8); + so_datap (so, pcrd, 8); + } + + so_ref(so, &nv50->state.programs); + so_ref(NULL, &so); +} + void nv50_program_destroy(struct nv50_context *nv50, struct nv50_program *p) { @@ -2476,7 +3073,6 @@ nv50_program_destroy(struct nv50_context *nv50, struct nv50_program *p) nouveau_bo_ref(NULL, &p->bo); nouveau_resource_free(&p->data[0]); - nouveau_resource_free(&p->data[1]); p->translated = 0; } diff --git a/src/gallium/drivers/nv50/nv50_program.h b/src/gallium/drivers/nv50/nv50_program.h index 096e0476aab..d78dee083f1 100644 --- a/src/gallium/drivers/nv50/nv50_program.h +++ b/src/gallium/drivers/nv50/nv50_program.h @@ -15,6 +15,15 @@ struct nv50_program_exec { } param; }; +struct nv50_sreg4 { + uint8_t hw; + uint8_t id_vp; + uint8_t id_fp; + + uint8_t mask; + boolean linear; +}; + struct nv50_program { struct pipe_shader_state pipe; struct tgsi_shader_info info; @@ -24,8 +33,8 @@ struct nv50_program { struct nv50_program_exec *exec_head; struct nv50_program_exec *exec_tail; unsigned exec_size; - struct nouveau_resource *data[2]; - unsigned data_start[2]; + struct nouveau_resource *data[1]; + unsigned data_start[1]; struct nouveau_bo *bo; @@ -36,14 +45,20 @@ struct nv50_program { struct { unsigned high_temp; unsigned high_result; - struct { - unsigned attr[2]; - } vp; - struct { - unsigned regs[4]; - unsigned map[5]; - unsigned high_map; - } fp; + + uint32_t attr[2]; + uint32_t regs[4]; + + /* for VPs, io_nr doesn't count 'private' results (PSIZ etc.) */ + unsigned io_nr; + struct nv50_sreg4 io[PIPE_MAX_SHADER_OUTPUTS]; + + /* FP colour inputs, VP/GP back colour outputs */ + struct nv50_sreg4 two_side[2]; + + /* VP only */ + uint8_t clpd, clpd_nr; + uint8_t psiz; } cfg; }; diff --git a/src/gallium/drivers/nv50/nv50_screen.c b/src/gallium/drivers/nv50/nv50_screen.c index c7f80a22037..3b08e1b89fb 100644 --- a/src/gallium/drivers/nv50/nv50_screen.c +++ b/src/gallium/drivers/nv50/nv50_screen.c @@ -87,12 +87,10 @@ nv50_screen_get_param(struct pipe_screen *pscreen, int param) return 1; case PIPE_CAP_GLSL: return 0; - case PIPE_CAP_S3TC: - return 1; case PIPE_CAP_ANISOTROPIC_FILTER: return 1; case PIPE_CAP_POINT_SPRITE: - return 0; + return 1; case PIPE_CAP_MAX_RENDER_TARGETS: return 8; case PIPE_CAP_OCCLUSION_QUERY: diff --git a/src/gallium/drivers/nv50/nv50_state.c b/src/gallium/drivers/nv50/nv50_state.c index 4283808ed93..81fa3e34c59 100644 --- a/src/gallium/drivers/nv50/nv50_state.c +++ b/src/gallium/drivers/nv50/nv50_state.c @@ -276,6 +276,9 @@ nv50_rasterizer_state_create(struct pipe_context *pipe, so_method(so, tesla, 0x1684, 1); so_data (so, cso->flatshade_first ? 0 : 1); + so_method(so, tesla, NV50TCL_VERTEX_TWO_SIDE_ENABLE, 1); + so_data (so, cso->light_twoside); + so_method(so, tesla, NV50TCL_LINE_WIDTH, 1); so_data (so, fui(cso->line_width)); so_method(so, tesla, NV50TCL_LINE_SMOOTH_ENABLE, 1); @@ -294,6 +297,9 @@ nv50_rasterizer_state_create(struct pipe_context *pipe, so_method(so, tesla, NV50TCL_POINT_SIZE, 1); so_data (so, fui(cso->point_size)); + so_method(so, tesla, NV50TCL_POINT_SPRITE_ENABLE, 1); + so_data (so, cso->point_sprite); + so_method(so, tesla, NV50TCL_POLYGON_MODE_FRONT, 3); if (cso->front_winding == PIPE_WINDING_CCW) { so_data(so, nvgl_polygon_mode(cso->fill_ccw)); diff --git a/src/gallium/drivers/nv50/nv50_state_validate.c b/src/gallium/drivers/nv50/nv50_state_validate.c index d294356f75d..4ed76973c4b 100644 --- a/src/gallium/drivers/nv50/nv50_state_validate.c +++ b/src/gallium/drivers/nv50/nv50_state_validate.c @@ -189,6 +189,8 @@ nv50_state_emit(struct nv50_context *nv50) so_emit(chan, nv50->state.vertprog); if (nv50->state.dirty & NV50_NEW_FRAGPROG) so_emit(chan, nv50->state.fragprog); + if (nv50->state.dirty & (NV50_NEW_FRAGPROG | NV50_NEW_VERTPROG)) + so_emit(chan, nv50->state.programs); if (nv50->state.dirty & NV50_NEW_RASTERIZER) so_emit(chan, nv50->state.rast); if (nv50->state.dirty & NV50_NEW_BLEND_COLOUR) @@ -210,6 +212,12 @@ nv50_state_emit(struct nv50_context *nv50) so_emit(chan, nv50->state.vtxattr); } nv50->state.dirty = 0; +} + +void +nv50_state_flush_notify(struct nouveau_channel *chan) +{ + struct nv50_context *nv50 = chan->user_private; so_emit_reloc_markers(chan, nv50->state.fb); so_emit_reloc_markers(chan, nv50->state.vertprog); @@ -240,6 +248,9 @@ nv50_state_validate(struct nv50_context *nv50) if (nv50->dirty & (NV50_NEW_FRAGPROG | NV50_NEW_FRAGPROG_CB)) nv50_fragprog_validate(nv50); + if (nv50->dirty & (NV50_NEW_FRAGPROG | NV50_NEW_VERTPROG)) + nv50_linkage_validate(nv50); + if (nv50->dirty & NV50_NEW_RASTERIZER) so_ref(nv50->rasterizer->so, &nv50->state.rast); @@ -301,7 +312,7 @@ scissor_uptodate: goto viewport_uptodate; nv50->state.viewport_bypass = bypass; - so = so_new(12, 0); + so = so_new(14, 0); if (!bypass) { so_method(so, tesla, NV50TCL_VIEWPORT_TRANSLATE(0), 3); so_data (so, fui(nv50->viewport.translate[0])); @@ -314,12 +325,21 @@ scissor_uptodate: so_method(so, tesla, NV50TCL_VIEWPORT_TRANSFORM_EN, 1); so_data (so, 1); + /* 0x0000 = remove whole primitive only (xyz) + * 0x1018 = remove whole primitive only (xy), clamp z + * 0x1080 = clip primitive (xyz) + * 0x1098 = clip primitive (xy), clamp z + */ + so_method(so, tesla, NV50TCL_VIEW_VOLUME_CLIP_CTRL, 1); + so_data (so, 0x1080); /* no idea what 0f90 does */ so_method(so, tesla, 0x0f90, 1); so_data (so, 0); } else { so_method(so, tesla, NV50TCL_VIEWPORT_TRANSFORM_EN, 1); so_data (so, 0); + so_method(so, tesla, NV50TCL_VIEW_VOLUME_CLIP_CTRL, 1); + so_data (so, 0x0000); so_method(so, tesla, 0x0f90, 1); so_data (so, 1); } diff --git a/src/gallium/drivers/nv50/nv50_transfer.c b/src/gallium/drivers/nv50/nv50_transfer.c index e9c3562194b..bb7731855cd 100644 --- a/src/gallium/drivers/nv50/nv50_transfer.c +++ b/src/gallium/drivers/nv50/nv50_transfer.c @@ -89,14 +89,14 @@ nv50_transfer_rect_m2mf(struct pipe_screen *pscreen, if (src_bo->tile_flags) { BEGIN_RING(chan, m2mf, NV50_MEMORY_TO_MEMORY_FORMAT_TILING_POSITION_IN, 1); - OUT_RING (chan, (sy << 16) | sx); + OUT_RING (chan, (sy << 16) | (sx * cpp)); } else { src_offset += (line_count * src_pitch); } if (dst_bo->tile_flags) { BEGIN_RING(chan, m2mf, NV50_MEMORY_TO_MEMORY_FORMAT_TILING_POSITION_OUT, 1); - OUT_RING (chan, (dy << 16) | dx); + OUT_RING (chan, (dy << 16) | (dx * cpp)); } else { dst_offset += (line_count * dst_pitch); } diff --git a/src/gallium/drivers/r300/r300_context.h b/src/gallium/drivers/r300/r300_context.h index 6c5914baa35..52b1c9a6b26 100644 --- a/src/gallium/drivers/r300/r300_context.h +++ b/src/gallium/drivers/r300/r300_context.h @@ -184,8 +184,15 @@ struct r300_texture { /* Offsets into the buffer. */ unsigned offset[PIPE_MAX_TEXTURE_LEVELS]; - /* Stride (pitch?) of this texture in bytes */ - unsigned stride; + /** + * If non-zero, override the natural texture layout with + * a custom stride (in bytes). + * + * \note Mipmapping fails for textures with a non-natural layout! + * + * \sa r300_texture_get_stride + */ + unsigned stride_override; /* Total size of this texture, in bytes. */ unsigned size; diff --git a/src/gallium/drivers/r300/r300_emit.c b/src/gallium/drivers/r300/r300_emit.c index 1bc35c24867..a1b36ba2ed1 100644 --- a/src/gallium/drivers/r300/r300_emit.c +++ b/src/gallium/drivers/r300/r300_emit.c @@ -283,7 +283,7 @@ void r300_emit_fb_state(struct r300_context* r300, for (i = 0; i < fb->nr_cbufs; i++) { tex = (struct r300_texture*)fb->cbufs[i]->texture; assert(tex && tex->buffer && "cbuf is marked, but NULL!"); - pixpitch = tex->stride / tex->tex.block.size; + pixpitch = r300_texture_get_stride(tex, 0) / tex->tex.block.size; OUT_CS_REG_SEQ(R300_RB3D_COLOROFFSET0 + (4 * i), 1); OUT_CS_RELOC(tex->buffer, 0, 0, RADEON_GEM_DOMAIN_VRAM, 0); @@ -300,7 +300,7 @@ void r300_emit_fb_state(struct r300_context* r300, if (fb->zsbuf) { tex = (struct r300_texture*)fb->zsbuf->texture; assert(tex && tex->buffer && "zsbuf is marked, but NULL!"); - pixpitch = tex->stride / tex->tex.block.size; + pixpitch = r300_texture_get_stride(tex, 0) / tex->tex.block.size; OUT_CS_REG_SEQ(R300_ZB_DEPTHOFFSET, 1); OUT_CS_RELOC(tex->buffer, 0, 0, RADEON_GEM_DOMAIN_VRAM, 0); diff --git a/src/gallium/drivers/r300/r300_render.c b/src/gallium/drivers/r300/r300_render.c index d05a736dd96..737396d8d97 100644 --- a/src/gallium/drivers/r300/r300_render.c +++ b/src/gallium/drivers/r300/r300_render.c @@ -31,6 +31,7 @@ #include "r300_state_derived.h" /* r300_render: Vertex and index buffer primitive emission. */ +#define R300_MAX_VBO_SIZE (1024 * 1024) struct r300_render { /* Parent class */ @@ -46,7 +47,10 @@ struct r300_render { /* VBO */ struct pipe_buffer* vbo; - size_t vbo_alloc_size; + size_t vbo_size; + size_t vbo_offset; + size_t vbo_max_used; + void * vbo_ptr; }; static INLINE struct r300_render* @@ -75,19 +79,18 @@ static boolean r300_render_allocate_vertices(struct vbuf_render* render, struct pipe_screen* screen = r300->context.screen; size_t size = (size_t)vertex_size * (size_t)count; - if (r300render->vbo && (size > r300render->vbo_alloc_size)) { - pipe_buffer_reference(&r300render->vbo, NULL); - } - - if (!r300render->vbo) { + if (size + r300render->vbo_offset > r300render->vbo_size) + { r300render->vbo = pipe_buffer_create(screen, 64, PIPE_BUFFER_USAGE_VERTEX, - size); + R300_MAX_VBO_SIZE); + r300render->vbo_size = R300_MAX_VBO_SIZE; } - r300render->vbo_alloc_size = MAX2(size, r300render->vbo_alloc_size); r300render->vertex_size = vertex_size; + r300->vbo = r300render->vbo; + r300->vbo_offset = r300render->vbo_offset; return (r300render->vbo) ? TRUE : FALSE; } @@ -97,8 +100,10 @@ static void* r300_render_map_vertices(struct vbuf_render* render) struct r300_render* r300render = r300_render(render); struct pipe_screen* screen = r300render->r300->context.screen; - return (unsigned char*)pipe_buffer_map(screen, r300render->vbo, - PIPE_BUFFER_USAGE_CPU_WRITE); + r300render->vbo_ptr = pipe_buffer_map(screen, r300render->vbo, + PIPE_BUFFER_USAGE_CPU_WRITE); + + return (r300render->vbo_ptr + r300render->vbo_offset); } static void r300_render_unmap_vertices(struct vbuf_render* render, @@ -107,15 +112,24 @@ static void r300_render_unmap_vertices(struct vbuf_render* render, { struct r300_render* r300render = r300_render(render); struct pipe_screen* screen = r300render->r300->context.screen; + CS_LOCALS(r300render->r300); + BEGIN_CS(2); + OUT_CS_REG(R300_VAP_VF_MAX_VTX_INDX, max); + END_CS; + r300render->vbo_max_used = MAX2(r300render->vbo_max_used, + r300render->vertex_size * (max + 1)); pipe_buffer_unmap(screen, r300render->vbo); } static void r300_render_release_vertices(struct vbuf_render* render) { struct r300_render* r300render = r300_render(render); + struct r300_context* r300 = r300render->r300; - pipe_buffer_reference(&r300render->vbo, NULL); + r300render->vbo_offset += r300render->vbo_max_used; + r300render->vbo_max_used = 0; + r300->vbo = NULL; } static boolean r300_render_set_primitive(struct vbuf_render* render, @@ -163,14 +177,12 @@ static boolean r300_render_set_primitive(struct vbuf_render* render, return TRUE; } -static void prepare_render(struct r300_render* render, unsigned count) +static void r300_prepare_render(struct r300_render* render, unsigned count) { struct r300_context* r300 = render->r300; CS_LOCALS(r300); - r300->vbo = render->vbo; - r300_emit_dirty_state(r300); } @@ -183,7 +195,7 @@ static void r300_render_draw_arrays(struct vbuf_render* render, CS_LOCALS(r300); - prepare_render(r300render, count); + r300_prepare_render(r300render, count); DBG(r300, DBG_DRAW, "r300: Doing vbuf render, count %d\n", count); @@ -208,7 +220,7 @@ static void r300_render_draw(struct vbuf_render* render, CS_LOCALS(r300); - prepare_render(r300render, count); + r300_prepare_render(r300render, count); /* Send our indices into an index buffer. */ index_buffer = pipe_buffer_create(screen, 64, PIPE_BUFFER_USAGE_VERTEX, @@ -217,25 +229,7 @@ static void r300_render_draw(struct vbuf_render* render, return; } -/* - index_map = pipe_buffer_map(screen, index_buffer, - PIPE_BUFFER_USAGE_CPU_WRITE); - memcpy(index_map, indices, count); - pipe_buffer_unmap(screen, index_buffer); - - debug_printf("r300: Doing indexbuf render, count %d\n", count); - - BEGIN_CS(8); - OUT_CS_PKT3(R300_PACKET3_3D_DRAW_INDX_2, 0); - OUT_CS(R300_VAP_VF_CNTL__PRIM_WALK_INDICES | (count << 16) | - r300render->hwprim); - OUT_CS_PKT3(R300_PACKET3_INDX_BUFFER, 2); - OUT_CS(R300_INDX_BUFFER_ONE_REG_WR | (R300_VAP_PORT_IDX0 >> 2)); - OUT_CS_INDEX_RELOC(index_buffer, 0, count, RADEON_GEM_DOMAIN_GTT, 0, 0); - END_CS; */ - - BEGIN_CS(4 + (count+1)/2); - OUT_CS_REG(R300_VAP_VF_MAX_VTX_INDX, count); + BEGIN_CS(2 + (count+1)/2); OUT_CS_PKT3(R300_PACKET3_3D_DRAW_INDX_2, (count+1)/2); OUT_CS(R300_VAP_VF_CNTL__PRIM_WALK_INDICES | (count << 16) | r300render->hwprim); @@ -273,6 +267,10 @@ static struct vbuf_render* r300_render_create(struct r300_context* r300) r300render->base.release_vertices = r300_render_release_vertices; r300render->base.destroy = r300_render_destroy; + r300render->vbo = NULL; + r300render->vbo_size = 0; + r300render->vbo_offset = 0; + return &r300render->base; } diff --git a/src/gallium/drivers/r300/r300_screen.c b/src/gallium/drivers/r300/r300_screen.c index 15740f61252..3b5b1bbd37f 100644 --- a/src/gallium/drivers/r300/r300_screen.c +++ b/src/gallium/drivers/r300/r300_screen.c @@ -93,8 +93,6 @@ static int r300_get_param(struct pipe_screen* pscreen, int param) } else { return 0; } - case PIPE_CAP_S3TC: - return 1; case PIPE_CAP_ANISOTROPIC_FILTER: return 1; case PIPE_CAP_POINT_SPRITE: @@ -323,13 +321,14 @@ r300_get_tex_transfer(struct pipe_screen *screen, if (trans) { pipe_texture_reference(&trans->transfer.texture, texture); trans->transfer.format = texture->format; + trans->transfer.x = x; + trans->transfer.y = y; trans->transfer.width = w; trans->transfer.height = h; trans->transfer.block = texture->block; trans->transfer.nblocksx = texture->nblocksx[level]; trans->transfer.nblocksy = texture->nblocksy[level]; - trans->transfer.stride = align(pf_get_stride(&trans->transfer.block, - texture->width[level]), 32); + trans->transfer.stride = r300_texture_get_stride(tex, level); trans->transfer.usage = usage; trans->offset = offset; } @@ -356,7 +355,7 @@ static void* r300_transfer_map(struct pipe_screen* screen, if (transfer->usage != PIPE_TRANSFER_READ) { flags |= PIPE_BUFFER_USAGE_CPU_WRITE; } - + map = pipe_buffer_map(screen, tex->buffer, flags); if (!map) { diff --git a/src/gallium/drivers/r300/r300_surface.c b/src/gallium/drivers/r300/r300_surface.c index 96e6e4a77d4..cc6288cb519 100644 --- a/src/gallium/drivers/r300/r300_surface.c +++ b/src/gallium/drivers/r300/r300_surface.c @@ -29,7 +29,7 @@ static void r300_surface_setup(struct r300_context* r300, unsigned w, unsigned h) { struct r300_capabilities* caps = r300_screen(r300->context.screen)->caps; - unsigned pixpitch = dest->stride / dest->tex.block.size; + unsigned pixpitch = r300_texture_get_stride(dest, 0) / dest->tex.block.size; CS_LOCALS(r300); r300_emit_blend_state(r300, &blend_clear_state); @@ -100,7 +100,7 @@ static void r300_surface_fill(struct pipe_context* pipe, struct r300_context* r300 = r300_context(pipe); struct r300_capabilities* caps = r300_screen(pipe->screen)->caps; struct r300_texture* tex = (struct r300_texture*)dest->texture; - unsigned pixpitch = tex->stride / tex->tex.block.size; + unsigned pixpitch = r300_texture_get_stride(tex, 0) / tex->tex.block.size; boolean invalid = FALSE; CS_LOCALS(r300); @@ -233,7 +233,7 @@ static void r300_surface_copy(struct pipe_context* pipe, struct r300_capabilities* caps = r300_screen(pipe->screen)->caps; struct r300_texture* srctex = (struct r300_texture*)src->texture; struct r300_texture* desttex = (struct r300_texture*)dest->texture; - unsigned pixpitch = srctex->stride / srctex->tex.block.size; + unsigned pixpitch = r300_texture_get_stride(srctex, 0) / srctex->tex.block.size; boolean invalid = FALSE; float fsrcx = srcx, fsrcy = srcy, fdestx = destx, fdesty = desty; CS_LOCALS(r300); diff --git a/src/gallium/drivers/r300/r300_texture.c b/src/gallium/drivers/r300/r300_texture.c index 590052509cc..7c041d17f71 100644 --- a/src/gallium/drivers/r300/r300_texture.c +++ b/src/gallium/drivers/r300/r300_texture.c @@ -25,7 +25,6 @@ static void r300_setup_texture_state(struct r300_texture* tex, unsigned width, unsigned height, - unsigned pitch, unsigned levels) { struct r300_texture_state* state = &tex->state; @@ -38,7 +37,7 @@ static void r300_setup_texture_state(struct r300_texture* tex, /* XXX */ state->format1 = r300_translate_texformat(tex->tex.format); - state->format2 = pitch - 1; + state->format2 = r300_texture_get_stride(tex, 0); /* Assume (somewhat foolishly) that oversized textures will * not be permitted by the state tracker. */ @@ -49,14 +48,31 @@ static void r300_setup_texture_state(struct r300_texture* tex, state->format2 |= R500_TXHEIGHT_BIT11; } - debug_printf("r300: Set texture state (%dx%d, pitch %d, %d levels)\n", - width, height, pitch, levels); + debug_printf("r300: Set texture state (%dx%d, %d levels)\n", + width, height, levels); +} + +/** + * Return the stride, in bytes, of the texture images of the given texture + * at the given level. + */ +unsigned r300_texture_get_stride(struct r300_texture* tex, unsigned level) +{ + if (tex->stride_override) + return tex->stride_override; + + if (level > tex->tex.last_level) { + debug_printf("%s: level (%u) > last_level (%u)\n", __FUNCTION__, level, tex->tex.last_level); + return 0; + } + + return align(pf_get_stride(&tex->tex.block, tex->tex.width[level]), 32); } static void r300_setup_miptree(struct r300_texture* tex) { struct pipe_texture* base = &tex->tex; - int stride, size, offset; + int stride, size; int i; for (i = 0; i <= base->last_level; i++) { @@ -74,7 +90,7 @@ static void r300_setup_miptree(struct r300_texture* tex) * XXX * POT, uncompressed, unmippmapped textures can be aligned to 32, * instead of 64. */ - stride = align(pf_get_stride(&base->block, base->width[i]), 32); + stride = r300_texture_get_stride(tex, i); size = stride * base->nblocksy[i] * base->depth[i]; tex->offset[i] = align(tex->size, 32); @@ -84,10 +100,6 @@ static void r300_setup_miptree(struct r300_texture* tex) "(%dx%dx%d px, pitch %d bytes)\n", i, base->width[i], base->height[i], base->depth[i], stride); - /* Save stride of first level to the texture. */ - if (i == 0) { - tex->stride = stride; - } } } @@ -109,7 +121,7 @@ static struct pipe_texture* r300_setup_miptree(tex); r300_setup_texture_state(tex, template->width[0], template->height[0], - template->width[0], template->last_level); + template->last_level); tex->buffer = screen->buffer_create(screen, 1024, PIPE_BUFFER_USAGE_PIXEL, @@ -189,11 +201,10 @@ static struct pipe_texture* pipe_reference_init(&tex->tex.reference, 1); tex->tex.screen = screen; - tex->stride = *stride; + tex->stride_override = *stride; /* XXX */ - r300_setup_texture_state(tex, tex->tex.width[0], tex->tex.height[0], - tex->stride, 0); + r300_setup_texture_state(tex, tex->tex.width[0], tex->tex.height[0], 0); pipe_buffer_reference(&tex->buffer, buffer); @@ -221,7 +232,7 @@ boolean r300_get_texture_buffer(struct pipe_texture* texture, pipe_buffer_reference(buffer, tex->buffer); if (stride) { - *stride = tex->stride; + *stride = r300_texture_get_stride(tex, 0); } return TRUE; diff --git a/src/gallium/drivers/r300/r300_texture.h b/src/gallium/drivers/r300/r300_texture.h index 3b56f0307c0..697669147d8 100644 --- a/src/gallium/drivers/r300/r300_texture.h +++ b/src/gallium/drivers/r300/r300_texture.h @@ -30,8 +30,12 @@ #include "r300_context.h" #include "r300_reg.h" +struct r300_texture; + void r300_init_screen_texture_functions(struct pipe_screen* screen); +unsigned r300_texture_get_stride(struct r300_texture* tex, unsigned level); + /* Note the signature of R300_EASY_TX_FORMAT(A, R, G, B, FORMAT)... */ static INLINE uint32_t r300_translate_texformat(enum pipe_format format) { @@ -68,6 +72,9 @@ static INLINE uint32_t r300_translate_texformat(enum pipe_format format) /* W24_FP */ case PIPE_FORMAT_Z24S8_UNORM: return R300_EASY_TX_FORMAT(X, X, X, X, W24_FP); + /* Z5_Y6_X5 */ + case PIPE_FORMAT_R16_SNORM: + return R300_EASY_TX_FORMAT(X, X, X, X, Z5Y6X5); default: debug_printf("r300: Implementation error: " "Got unsupported texture format %s in %s\n", diff --git a/src/gallium/drivers/r300/r300_tgsi_to_rc.c b/src/gallium/drivers/r300/r300_tgsi_to_rc.c index d68a1041063..0913ca1bd5b 100644 --- a/src/gallium/drivers/r300/r300_tgsi_to_rc.c +++ b/src/gallium/drivers/r300/r300_tgsi_to_rc.c @@ -158,7 +158,6 @@ static unsigned translate_saturate(unsigned saturate) switch(saturate) { case TGSI_SAT_NONE: return SATURATE_OFF; case TGSI_SAT_ZERO_ONE: return SATURATE_ZERO_ONE; - case TGSI_SAT_MINUS_PLUS_ONE: return SATURATE_PLUS_MINUS_ONE; } fprintf(stderr, "Unknown saturate mode: %i\n", saturate); diff --git a/src/gallium/drivers/softpipe/Makefile b/src/gallium/drivers/softpipe/Makefile index 516e3992fdd..6ab37537628 100644 --- a/src/gallium/drivers/softpipe/Makefile +++ b/src/gallium/drivers/softpipe/Makefile @@ -6,26 +6,17 @@ LIBNAME = softpipe C_SOURCES = \ sp_fs_exec.c \ sp_fs_sse.c \ - sp_fs_llvm.c \ sp_clear.c \ sp_flush.c \ sp_query.c \ sp_context.c \ sp_draw_arrays.c \ - sp_prim_setup.c \ sp_prim_vbuf.c \ sp_quad_pipe.c \ - sp_quad_alpha_test.c \ - sp_quad_blend.c \ - sp_quad_colormask.c \ - sp_quad_coverage.c \ + sp_quad_stipple.c \ sp_quad_depth_test.c \ - sp_quad_earlyz.c \ sp_quad_fs.c \ - sp_quad_occlusion.c \ - sp_quad_output.c \ - sp_quad_stencil.c \ - sp_quad_stipple.c \ + sp_quad_blend.c \ sp_screen.c \ sp_setup.c \ sp_state_blend.c \ @@ -38,6 +29,7 @@ C_SOURCES = \ sp_state_vertex.c \ sp_texture.c \ sp_tex_sample.c \ + sp_tex_tile_cache.c \ sp_tile_cache.c \ sp_surface.c diff --git a/src/gallium/drivers/softpipe/SConscript b/src/gallium/drivers/softpipe/SConscript index f8720638a76..950c3d9955b 100644 --- a/src/gallium/drivers/softpipe/SConscript +++ b/src/gallium/drivers/softpipe/SConscript @@ -7,25 +7,16 @@ softpipe = env.ConvenienceLibrary( source = [ 'sp_fs_exec.c', 'sp_fs_sse.c', - 'sp_fs_llvm.c', 'sp_clear.c', 'sp_context.c', 'sp_draw_arrays.c', 'sp_flush.c', - 'sp_prim_setup.c', 'sp_prim_vbuf.c', 'sp_setup.c', - 'sp_quad_alpha_test.c', 'sp_quad_blend.c', 'sp_quad_pipe.c', - 'sp_quad_colormask.c', - 'sp_quad_coverage.c', 'sp_quad_depth_test.c', - 'sp_quad_earlyz.c', 'sp_quad_fs.c', - 'sp_quad_occlusion.c', - 'sp_quad_output.c', - 'sp_quad_stencil.c', 'sp_quad_stipple.c', 'sp_query.c', 'sp_screen.c', @@ -39,6 +30,7 @@ softpipe = env.ConvenienceLibrary( 'sp_state_vertex.c', 'sp_surface.c', 'sp_tex_sample.c', + 'sp_tex_tile_cache.c', 'sp_texture.c', 'sp_tile_cache.c', ]) diff --git a/src/gallium/drivers/softpipe/sp_clear.c b/src/gallium/drivers/softpipe/sp_clear.c index d3af18e162b..8fac8e6e05f 100644 --- a/src/gallium/drivers/softpipe/sp_clear.c +++ b/src/gallium/drivers/softpipe/sp_clear.c @@ -36,8 +36,6 @@ #include "util/u_pack_color.h" #include "sp_clear.h" #include "sp_context.h" -#include "sp_surface.h" -#include "sp_state.h" #include "sp_tile_cache.h" diff --git a/src/gallium/drivers/softpipe/sp_clear.h b/src/gallium/drivers/softpipe/sp_clear.h index 2e450672f58..9be3b86fe9f 100644 --- a/src/gallium/drivers/softpipe/sp_clear.h +++ b/src/gallium/drivers/softpipe/sp_clear.h @@ -32,7 +32,6 @@ #ifndef SP_CLEAR_H #define SP_CLEAR_H -#include "pipe/p_state.h" struct pipe_context; extern void diff --git a/src/gallium/drivers/softpipe/sp_context.c b/src/gallium/drivers/softpipe/sp_context.c index 86df320ea8a..e1e31ab047a 100644 --- a/src/gallium/drivers/softpipe/sp_context.c +++ b/src/gallium/drivers/softpipe/sp_context.c @@ -31,17 +31,18 @@ */ #include "draw/draw_context.h" +#include "draw/draw_vbuf.h" #include "pipe/p_defines.h" #include "util/u_math.h" #include "util/u_memory.h" #include "sp_clear.h" #include "sp_context.h" #include "sp_flush.h" -#include "sp_prim_setup.h" #include "sp_prim_vbuf.h" #include "sp_state.h" #include "sp_surface.h" #include "sp_tile_cache.h" +#include "sp_tex_tile_cache.h" #include "sp_texture.h" #include "sp_winsys.h" #include "sp_query.h" @@ -72,18 +73,16 @@ softpipe_unmap_transfers(struct softpipe_context *sp) { uint i; - for (i = 0; i < sp->framebuffer.nr_cbufs; i++) - sp_flush_tile_cache(sp, sp->cbuf_cache[i]); - sp_flush_tile_cache(sp, sp->zsbuf_cache); - for (i = 0; i < sp->framebuffer.nr_cbufs; i++) { sp_tile_cache_unmap_transfers(sp->cbuf_cache[i]); } + sp_tile_cache_unmap_transfers(sp->zsbuf_cache); } -static void softpipe_destroy( struct pipe_context *pipe ) +static void +softpipe_destroy( struct pipe_context *pipe ) { struct softpipe_context *softpipe = softpipe_context( pipe ); uint i; @@ -91,26 +90,16 @@ static void softpipe_destroy( struct pipe_context *pipe ) if (softpipe->draw) draw_destroy( softpipe->draw ); - for (i = 0; i < SP_NUM_QUAD_THREADS; i++) { - softpipe->quad[i].polygon_stipple->destroy( softpipe->quad[i].polygon_stipple ); - softpipe->quad[i].earlyz->destroy( softpipe->quad[i].earlyz ); - softpipe->quad[i].shade->destroy( softpipe->quad[i].shade ); - softpipe->quad[i].alpha_test->destroy( softpipe->quad[i].alpha_test ); - softpipe->quad[i].depth_test->destroy( softpipe->quad[i].depth_test ); - softpipe->quad[i].stencil_test->destroy( softpipe->quad[i].stencil_test ); - softpipe->quad[i].occlusion->destroy( softpipe->quad[i].occlusion ); - softpipe->quad[i].coverage->destroy( softpipe->quad[i].coverage ); - softpipe->quad[i].blend->destroy( softpipe->quad[i].blend ); - softpipe->quad[i].colormask->destroy( softpipe->quad[i].colormask ); - softpipe->quad[i].output->destroy( softpipe->quad[i].output ); - } + softpipe->quad.shade->destroy( softpipe->quad.shade ); + softpipe->quad.depth_test->destroy( softpipe->quad.depth_test ); + softpipe->quad.blend->destroy( softpipe->quad.blend ); for (i = 0; i < PIPE_MAX_COLOR_BUFS; i++) sp_destroy_tile_cache(softpipe->cbuf_cache[i]); sp_destroy_tile_cache(softpipe->zsbuf_cache); for (i = 0; i < PIPE_MAX_SAMPLERS; i++) - sp_destroy_tile_cache(softpipe->tex_cache[i]); + sp_destroy_tex_tile_cache(softpipe->tex_cache[i]); for (i = 0; i < Elements(softpipe->constants); i++) { if (softpipe->constants[i].buffer) { @@ -121,6 +110,15 @@ static void softpipe_destroy( struct pipe_context *pipe ) FREE( softpipe ); } + +/** + * if (the texture is being used as a framebuffer surface) + * return PIPE_REFERENCED_FOR_WRITE + * else if (the texture is a bound texture source) + * return PIPE_REFERENCED_FOR_READ XXX not done yet + * else + * return PIPE_UNREFERENCED + */ static unsigned int softpipe_is_texture_referenced( struct pipe_context *pipe, struct pipe_texture *texture, @@ -129,15 +127,17 @@ softpipe_is_texture_referenced( struct pipe_context *pipe, struct softpipe_context *softpipe = softpipe_context( pipe ); unsigned i; - if(softpipe->dirty_render_cache) { + if (softpipe->dirty_render_cache) { for (i = 0; i < softpipe->framebuffer.nr_cbufs; i++) { - if(softpipe->framebuffer.cbufs[i] && - softpipe->framebuffer.cbufs[i]->texture == texture) + if (softpipe->framebuffer.cbufs[i] && + softpipe->framebuffer.cbufs[i]->texture == texture) { return PIPE_REFERENCED_FOR_WRITE; + } } - if(softpipe->framebuffer.zsbuf && - softpipe->framebuffer.zsbuf->texture == texture) + if (softpipe->framebuffer.zsbuf && + softpipe->framebuffer.zsbuf->texture == texture) { return PIPE_REFERENCED_FOR_WRITE; + } } /* FIXME: we also need to do the same for the texture cache */ @@ -145,6 +145,7 @@ softpipe_is_texture_referenced( struct pipe_context *pipe, return PIPE_UNREFERENCED; } + static unsigned int softpipe_is_buffer_referenced( struct pipe_context *pipe, struct pipe_buffer *buf) @@ -152,6 +153,7 @@ softpipe_is_buffer_referenced( struct pipe_context *pipe, return PIPE_UNREFERENCED; } + struct pipe_context * softpipe_create( struct pipe_screen *screen ) { @@ -222,7 +224,6 @@ softpipe_create( struct pipe_screen *screen ) softpipe->pipe.is_buffer_referenced = softpipe_is_buffer_referenced; softpipe_init_query_funcs( softpipe ); - softpipe_init_texture_funcs( softpipe ); /* * Alloc caches for accessing drawing surfaces and textures. @@ -233,41 +234,14 @@ softpipe_create( struct pipe_screen *screen ) softpipe->zsbuf_cache = sp_create_tile_cache( screen ); for (i = 0; i < PIPE_MAX_SAMPLERS; i++) - softpipe->tex_cache[i] = sp_create_tile_cache( screen ); + softpipe->tex_cache[i] = sp_create_tex_tile_cache( screen ); /* setup quad rendering stages */ - for (i = 0; i < SP_NUM_QUAD_THREADS; i++) { - softpipe->quad[i].polygon_stipple = sp_quad_polygon_stipple_stage(softpipe); - softpipe->quad[i].earlyz = sp_quad_earlyz_stage(softpipe); - softpipe->quad[i].shade = sp_quad_shade_stage(softpipe); - softpipe->quad[i].alpha_test = sp_quad_alpha_test_stage(softpipe); - softpipe->quad[i].depth_test = sp_quad_depth_test_stage(softpipe); - softpipe->quad[i].stencil_test = sp_quad_stencil_test_stage(softpipe); - softpipe->quad[i].occlusion = sp_quad_occlusion_stage(softpipe); - softpipe->quad[i].coverage = sp_quad_coverage_stage(softpipe); - softpipe->quad[i].blend = sp_quad_blend_stage(softpipe); - softpipe->quad[i].colormask = sp_quad_colormask_stage(softpipe); - softpipe->quad[i].output = sp_quad_output_stage(softpipe); - } - - /* vertex shader samplers */ - for (i = 0; i < PIPE_MAX_SAMPLERS; i++) { - softpipe->tgsi.vert_samplers[i].base.get_samples = sp_get_samples_vertex; - softpipe->tgsi.vert_samplers[i].unit = i; - softpipe->tgsi.vert_samplers[i].sp = softpipe; - softpipe->tgsi.vert_samplers[i].cache = softpipe->tex_cache[i]; - softpipe->tgsi.vert_samplers_list[i] = &softpipe->tgsi.vert_samplers[i]; - } + softpipe->quad.shade = sp_quad_shade_stage(softpipe); + softpipe->quad.depth_test = sp_quad_depth_test_stage(softpipe); + softpipe->quad.blend = sp_quad_blend_stage(softpipe); - /* fragment shader samplers */ - for (i = 0; i < PIPE_MAX_SAMPLERS; i++) { - softpipe->tgsi.frag_samplers[i].base.get_samples = sp_get_samples_fragment; - softpipe->tgsi.frag_samplers[i].unit = i; - softpipe->tgsi.frag_samplers[i].sp = softpipe; - softpipe->tgsi.frag_samplers[i].cache = softpipe->tex_cache[i]; - softpipe->tgsi.frag_samplers_list[i] = &softpipe->tgsi.frag_samplers[i]; - } /* * Create drawing context and plug our rendering stage into it. @@ -281,30 +255,28 @@ softpipe_create( struct pipe_screen *screen ) (struct tgsi_sampler **) softpipe->tgsi.vert_samplers_list); - softpipe->setup = sp_draw_render_stage(softpipe); - if (!softpipe->setup) - goto fail; - if (debug_get_bool_option( "SP_NO_RAST", FALSE )) softpipe->no_rast = TRUE; - if (debug_get_bool_option( "SP_NO_VBUF", FALSE )) { - /* Deprecated path -- vbuf is the intended interface to the draw module: - */ - draw_set_rasterize_stage(softpipe->draw, softpipe->setup); - } - else { - sp_init_vbuf(softpipe); - } + softpipe->vbuf_backend = sp_create_vbuf_backend(softpipe); + if (!softpipe->vbuf_backend) + goto fail; + + softpipe->vbuf = draw_vbuf_stage(softpipe->draw, softpipe->vbuf_backend); + if (!softpipe->vbuf) + goto fail; + + draw_set_rasterize_stage(softpipe->draw, softpipe->vbuf); + draw_set_render(softpipe->draw, softpipe->vbuf_backend); + + /* plug in AA line/point stages */ draw_install_aaline_stage(softpipe->draw, &softpipe->pipe); draw_install_aapoint_stage(softpipe->draw, &softpipe->pipe); -#if USE_DRAW_STAGE_PSTIPPLE /* Do polygon stipple w/ texture map + frag prog? */ draw_install_pstipple_stage(softpipe->draw, &softpipe->pipe); -#endif sp_init_surface_functions(softpipe); diff --git a/src/gallium/drivers/softpipe/sp_context.h b/src/gallium/drivers/softpipe/sp_context.h index 7888c2f644b..43a195c8ef5 100644 --- a/src/gallium/drivers/softpipe/sp_context.h +++ b/src/gallium/drivers/softpipe/sp_context.h @@ -36,24 +36,13 @@ #include "draw/draw_vertex.h" #include "sp_quad_pipe.h" -#include "sp_tex_sample.h" -/** - * This is a temporary variable for testing draw-stage polygon stipple. - * If zero, do stipple in sp_quad_stipple.c - */ -#define USE_DRAW_STAGE_PSTIPPLE 1 - -/* Number of threads working on individual quads. - * Setting to 1 disables this feature. - */ -#define SP_NUM_QUAD_THREADS 1 - struct softpipe_vbuf_render; struct draw_context; struct draw_stage; struct softpipe_tile_cache; +struct softpipe_tex_tile_cache; struct sp_fragment_shader; struct sp_vertex_shader; @@ -62,12 +51,12 @@ struct softpipe_context { struct pipe_context pipe; /**< base class */ /** Constant state objects */ - const struct pipe_blend_state *blend; - const struct pipe_sampler_state *sampler[PIPE_MAX_SAMPLERS]; - const struct pipe_depth_stencil_alpha_state *depth_stencil; - const struct pipe_rasterizer_state *rasterizer; - const struct sp_fragment_shader *fs; - const struct sp_vertex_shader *vs; + struct pipe_blend_state *blend; + struct pipe_sampler_state *sampler[PIPE_MAX_SAMPLERS]; + struct pipe_depth_stencil_alpha_state *depth_stencil; + struct pipe_rasterizer_state *rasterizer; + struct sp_fragment_shader *fs; + struct sp_vertex_shader *vs; /** Other rendering state */ struct pipe_blend_color blend_color; @@ -107,7 +96,16 @@ struct softpipe_context { /** Which vertex shader output slot contains point size */ int psize_slot; - unsigned reduced_api_prim; /**< PIPE_PRIM_POINTS, _LINES or _TRIANGLES */ + /* The reduced version of the primitive supplied by the state + * tracker. + */ + unsigned reduced_api_prim; + + /* The reduced primitive after unfilled triangles, wide-line + * decomposition, etc, are taken into account. This is the + * primitive actually rasterized. + */ + unsigned reduced_prim; /** Derived from scissor and surface bounds: */ struct pipe_scissor_state cliprect; @@ -116,41 +114,33 @@ struct softpipe_context { /** Software quad rendering pipeline */ struct { - struct quad_stage *polygon_stipple; - struct quad_stage *earlyz; struct quad_stage *shade; - struct quad_stage *alpha_test; - struct quad_stage *stencil_test; struct quad_stage *depth_test; - struct quad_stage *occlusion; - struct quad_stage *coverage; struct quad_stage *blend; - struct quad_stage *colormask; - struct quad_stage *output; struct quad_stage *first; /**< points to one of the above stages */ - } quad[SP_NUM_QUAD_THREADS]; + } quad; /** TGSI exec things */ struct { - struct sp_shader_sampler vert_samplers[PIPE_MAX_SAMPLERS]; - struct sp_shader_sampler *vert_samplers_list[PIPE_MAX_SAMPLERS]; - struct sp_shader_sampler frag_samplers[PIPE_MAX_SAMPLERS]; - struct sp_shader_sampler *frag_samplers_list[PIPE_MAX_SAMPLERS]; + struct sp_sampler_varient *vert_samplers_list[PIPE_MAX_SAMPLERS]; + struct sp_sampler_varient *frag_samplers_list[PIPE_MAX_SAMPLERS]; } tgsi; /** The primitive drawing context */ struct draw_context *draw; - struct draw_stage *setup; + + /** Draw module backend */ + struct vbuf_render *vbuf_backend; struct draw_stage *vbuf; - struct softpipe_vbuf_render *vbuf_render; boolean dirty_render_cache; struct softpipe_tile_cache *cbuf_cache[PIPE_MAX_COLOR_BUFS]; struct softpipe_tile_cache *zsbuf_cache; - - struct softpipe_tile_cache *tex_cache[PIPE_MAX_SAMPLERS]; + + unsigned tex_timestamp; + struct softpipe_tex_tile_cache *tex_cache[PIPE_MAX_SAMPLERS]; unsigned use_sse : 1; unsigned dump_fs : 1; @@ -164,5 +154,9 @@ softpipe_context( struct pipe_context *pipe ) return (struct softpipe_context *)pipe; } +void +softpipe_reset_sampler_varients(struct softpipe_context *softpipe); + + #endif /* SP_CONTEXT_H */ diff --git a/src/gallium/drivers/softpipe/sp_flush.c b/src/gallium/drivers/softpipe/sp_flush.c index 4a14d49686e..e38b767cf2c 100644 --- a/src/gallium/drivers/softpipe/sp_flush.c +++ b/src/gallium/drivers/softpipe/sp_flush.c @@ -37,6 +37,7 @@ #include "sp_surface.h" #include "sp_state.h" #include "sp_tile_cache.h" +#include "sp_tex_tile_cache.h" #include "sp_winsys.h" @@ -52,17 +53,19 @@ softpipe_flush( struct pipe_context *pipe, if (flags & PIPE_FLUSH_TEXTURE_CACHE) { for (i = 0; i < softpipe->num_textures; i++) { - sp_flush_tile_cache(softpipe, softpipe->tex_cache[i]); + sp_flush_tex_tile_cache(softpipe->tex_cache[i]); } } - if (flags & PIPE_FLUSH_RENDER_CACHE) { + if (flags & PIPE_FLUSH_SWAPBUFFERS) { + /* If this is a swapbuffers, just flush color buffers. + * + * The zbuffer changes are not discarded, but held in the cache + * in the hope that a later clear will wipe them out. + */ for (i = 0; i < softpipe->framebuffer.nr_cbufs; i++) if (softpipe->cbuf_cache[i]) - sp_flush_tile_cache(softpipe, softpipe->cbuf_cache[i]); - - if (softpipe->zsbuf_cache) - sp_flush_tile_cache(softpipe, softpipe->zsbuf_cache); + sp_flush_tile_cache(softpipe->cbuf_cache[i]); /* Need this call for hardware buffers before swapbuffers. * @@ -71,7 +74,15 @@ softpipe_flush( struct pipe_context *pipe, * to unmap surfaces when flushing. */ softpipe_unmap_transfers(softpipe); - + } + else if (flags & PIPE_FLUSH_RENDER_CACHE) { + for (i = 0; i < softpipe->framebuffer.nr_cbufs; i++) + if (softpipe->cbuf_cache[i]) + sp_flush_tile_cache(softpipe->cbuf_cache[i]); + + if (softpipe->zsbuf_cache) + sp_flush_tile_cache(softpipe->zsbuf_cache); + softpipe->dirty_render_cache = FALSE; } diff --git a/src/gallium/drivers/softpipe/sp_fs_exec.c b/src/gallium/drivers/softpipe/sp_fs_exec.c index 9ee86fe7878..4076114d392 100644 --- a/src/gallium/drivers/softpipe/sp_fs_exec.c +++ b/src/gallium/drivers/softpipe/sp_fs_exec.c @@ -59,15 +59,34 @@ sp_exec_fragment_shader(const struct sp_fragment_shader *base) } +static void +exec_prepare( const struct sp_fragment_shader *base, + struct tgsi_exec_machine *machine, + struct tgsi_sampler **samplers ) +{ + /* + * Bind tokens/shader to the interpreter's machine state. + * Avoid redundant binding. + */ + if (machine->Tokens != base->shader.tokens) { + tgsi_exec_machine_bind_shader( machine, + base->shader.tokens, + PIPE_MAX_SAMPLERS, + samplers ); + } +} + + + /** * Compute quad X,Y,Z,W for the four fragments in a quad. * * This should really be part of the compiled shader. */ -void -sp_setup_pos_vector(const struct tgsi_interp_coef *coef, - float x, float y, - struct tgsi_exec_vector *quadpos) +static void +setup_pos_vector(const struct tgsi_interp_coef *coef, + float x, float y, + struct tgsi_exec_vector *quadpos) { uint chan; /* do X */ @@ -95,24 +114,6 @@ sp_setup_pos_vector(const struct tgsi_interp_coef *coef, } -static void -exec_prepare( const struct sp_fragment_shader *base, - struct tgsi_exec_machine *machine, - struct tgsi_sampler **samplers ) -{ - /* - * Bind tokens/shader to the interpreter's machine state. - * Avoid redundant binding. - */ - if (machine->Tokens != base->shader.tokens) { - tgsi_exec_machine_bind_shader( machine, - base->shader.tokens, - PIPE_MAX_SAMPLERS, - samplers ); - } -} - - /* TODO: hide the machine struct in here somewhere, remove from this * interface: */ @@ -122,11 +123,43 @@ exec_run( const struct sp_fragment_shader *base, struct quad_header *quad ) { /* Compute X, Y, Z, W vals for this quad */ - sp_setup_pos_vector(quad->posCoef, - (float)quad->input.x0, (float)quad->input.y0, - &machine->QuadPos); + setup_pos_vector(quad->posCoef, + (float)quad->input.x0, (float)quad->input.y0, + &machine->QuadPos); - return tgsi_exec_machine_run( machine ); + quad->inout.mask &= tgsi_exec_machine_run( machine ); + if (quad->inout.mask == 0) + return FALSE; + + /* store outputs */ + { + const ubyte *sem_name = base->info.output_semantic_name; + const ubyte *sem_index = base->info.output_semantic_index; + const uint n = base->info.num_outputs; + uint i; + for (i = 0; i < n; i++) { + switch (sem_name[i]) { + case TGSI_SEMANTIC_COLOR: + { + uint cbuf = sem_index[i]; + memcpy(quad->output.color[cbuf], + &machine->Outputs[i].xyzw[0].f[0], + sizeof(quad->output.color[0]) ); + } + break; + case TGSI_SEMANTIC_POSITION: + { + uint j; + for (j = 0; j < 4; j++) { + quad->output.depth[j] = machine->Outputs[i].xyzw[2].f[j]; + } + } + break; + } + } + } + + return TRUE; } diff --git a/src/gallium/drivers/softpipe/sp_fs_llvm.c b/src/gallium/drivers/softpipe/sp_fs_llvm.c deleted file mode 100644 index 95c0d982d12..00000000000 --- a/src/gallium/drivers/softpipe/sp_fs_llvm.c +++ /dev/null @@ -1,205 +0,0 @@ -/************************************************************************** - * - * Copyright 2007 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. - * - **************************************************************************/ - -/** - * Execute fragment shader using LLVM code generation. - * Authors: - * Zack Rusin - */ - -#include "sp_context.h" -#include "sp_state.h" -#include "sp_fs.h" - -#include "pipe/p_state.h" -#include "pipe/p_defines.h" -#include "util/u_memory.h" -#include "tgsi/tgsi_sse2.h" - -#if 0 - -/** - * Subclass of sp_fragment_shader - */ -struct sp_llvm_fragment_shader -{ - struct sp_fragment_shader base; - struct gallivm_prog *llvm_prog; -}; - - -static void -shade_quad_llvm(struct quad_stage *qs, - struct quad_header *quad) -{ - struct quad_shade_stage *qss = quad_shade_stage(qs); - struct softpipe_context *softpipe = qs->softpipe; - float dests[4][16][4] ALIGN16_ATTRIB; - float inputs[4][16][4] ALIGN16_ATTRIB; - const float fx = (float) quad->x0; - const float fy = (float) quad->y0; - struct gallivm_prog *llvm = qss->llvm_prog; - - inputs[0][0][0] = fx; - inputs[1][0][0] = fx + 1.0f; - inputs[2][0][0] = fx; - inputs[3][0][0] = fx + 1.0f; - - inputs[0][0][1] = fy; - inputs[1][0][1] = fy; - inputs[2][0][1] = fy + 1.0f; - inputs[3][0][1] = fy + 1.0f; - - - gallivm_prog_inputs_interpolate(llvm, inputs, quad->coef); - -#if DLLVM - debug_printf("MASK = %d\n", quad->mask); - for (int i = 0; i < 4; ++i) { - for (int j = 0; j < 2; ++j) { - debug_printf("IN(%d,%d) [%f %f %f %f]\n", i, j, - inputs[i][j][0], inputs[i][j][1], inputs[i][j][2], inputs[i][j][3]); - } - } -#endif - - quad->mask &= - gallivm_fragment_shader_exec(llvm, fx, fy, dests, inputs, - softpipe->mapped_constants[PIPE_SHADER_FRAGMENT], - qss->samplers); -#if DLLVM - debug_printf("OUT LLVM = 1[%f %f %f %f], 2[%f %f %f %f]\n", - dests[0][0][0], dests[0][0][1], dests[0][0][2], dests[0][0][3], - dests[0][1][0], dests[0][1][1], dests[0][1][2], dests[0][1][3]); -#endif - - /* store result color */ - if (qss->colorOutSlot >= 0) { - unsigned i; - /* XXX need to handle multiple color outputs someday */ - allvmrt(qss->stage.softpipe->fs->info.output_semantic_name[qss->colorOutSlot] - == TGSI_SEMANTIC_COLOR); - for (i = 0; i < QUAD_SIZE; ++i) { - quad->outputs.color[0][0][i] = dests[i][qss->colorOutSlot][0]; - quad->outputs.color[0][1][i] = dests[i][qss->colorOutSlot][1]; - quad->outputs.color[0][2][i] = dests[i][qss->colorOutSlot][2]; - quad->outputs.color[0][3][i] = dests[i][qss->colorOutSlot][3]; - } - } -#if DLLVM - for (int i = 0; i < QUAD_SIZE; ++i) { - debug_printf("QLLVM%d(%d) [%f, %f, %f, %f]\n", i, qss->colorOutSlot, - quad->outputs.color[0][0][i], - quad->outputs.color[0][1][i], - quad->outputs.color[0][2][i], - quad->outputs.color[0][3][i]); - } -#endif - - /* store result Z */ - if (qss->depthOutSlot >= 0) { - /* output[slot] is new Z */ - uint i; - for (i = 0; i < 4; i++) { - quad->outputs.depth[i] = dests[i][0][2]; - } - } - else { - /* copy input Z (which was interpolated by the executor) to output Z */ - uint i; - for (i = 0; i < 4; i++) { - quad->outputs.depth[i] = inputs[i][0][2]; - } - } -#if DLLVM - debug_printf("D [%f, %f, %f, %f] mask = %d\n", - quad->outputs.depth[0], - quad->outputs.depth[1], - quad->outputs.depth[2], - quad->outputs.depth[3], quad->mask); -#endif - - /* shader may cull fragments */ - if( quad->mask ) { - qs->next->run( qs->next, quad ); - } -} - - -unsigned -run_llvm_fs( const struct sp_fragment_shader *base, - struct foo *machine ) -{ -} - - -void -delete_llvm_fs( struct sp_fragment_shader *base ) -{ - FREE(base); -} - - -struct sp_fragment_shader * -softpipe_create_fs_llvm(struct softpipe_context *softpipe, - const struct pipe_shader_state *templ) -{ - struct sp_llvm_fragment_shader *shader = NULL; - - /* LLVM fragment shaders currently disabled: - */ - state = CALLOC_STRUCT(sp_llvm_shader_state); - if (!state) - return NULL; - - state->llvm_prog = 0; - - if (!gallivm_global_cpu_engine()) { - gallivm_cpu_engine_create(state->llvm_prog); - } - else - gallivm_cpu_jit_compile(gallivm_global_cpu_engine(), state->llvm_prog); - - if (shader) { - shader->base.run = run_llvm_fs; - shader->base.delete = delete_llvm_fs; - } - - return shader; -} - - -#else - -struct sp_fragment_shader * -softpipe_create_fs_llvm(struct softpipe_context *softpipe, - const struct pipe_shader_state *templ) -{ - return NULL; -} - -#endif diff --git a/src/gallium/drivers/softpipe/sp_fs_sse.c b/src/gallium/drivers/softpipe/sp_fs_sse.c index f4fa0905d74..9d3e4670eef 100644 --- a/src/gallium/drivers/softpipe/sp_fs_sse.c +++ b/src/gallium/drivers/softpipe/sp_fs_sse.c @@ -76,6 +76,43 @@ fs_sse_prepare( const struct sp_fragment_shader *base, } + +/** + * Compute quad X,Y,Z,W for the four fragments in a quad. + * + * This should really be part of the compiled shader. + */ +static void +setup_pos_vector(const struct tgsi_interp_coef *coef, + float x, float y, + struct tgsi_exec_vector *quadpos) +{ + uint chan; + /* do X */ + quadpos->xyzw[0].f[0] = x; + quadpos->xyzw[0].f[1] = x + 1; + quadpos->xyzw[0].f[2] = x; + quadpos->xyzw[0].f[3] = x + 1; + + /* do Y */ + quadpos->xyzw[1].f[0] = y; + quadpos->xyzw[1].f[1] = y; + quadpos->xyzw[1].f[2] = y + 1; + quadpos->xyzw[1].f[3] = y + 1; + + /* do Z and W for all fragments in the quad */ + for (chan = 2; chan < 4; chan++) { + const float dadx = coef->dadx[chan]; + const float dady = coef->dady[chan]; + const float a0 = coef->a0[chan] + dadx * x + dady * y; + quadpos->xyzw[chan].f[0] = a0; + quadpos->xyzw[chan].f[1] = a0 + dadx; + quadpos->xyzw[chan].f[2] = a0 + dady; + quadpos->xyzw[chan].f[3] = a0 + dadx + dady; + } +} + + /* TODO: codegenerate the whole run function, skip this wrapper. * TODO: break dependency on tgsi_exec_machine struct * TODO: push Position calculation into the generated shader @@ -89,9 +126,9 @@ fs_sse_run( const struct sp_fragment_shader *base, struct sp_sse_fragment_shader *shader = sp_sse_fragment_shader(base); /* Compute X, Y, Z, W vals for this quad -- place in temp[0] for now */ - sp_setup_pos_vector(quad->posCoef, - (float)quad->input.x0, (float)quad->input.y0, - machine->Temps); + setup_pos_vector(quad->posCoef, + (float)quad->input.x0, (float)quad->input.y0, + machine->Temps); /* init kill mask */ tgsi_set_kill_mask(machine, 0x0); @@ -104,7 +141,39 @@ fs_sse_run( const struct sp_fragment_shader *base, // , &machine->QuadPos ); - return ~(machine->Temps[TGSI_EXEC_TEMP_KILMASK_I].xyzw[TGSI_EXEC_TEMP_KILMASK_C].u[0]); + quad->inout.mask &= ~(machine->Temps[TGSI_EXEC_TEMP_KILMASK_I].xyzw[TGSI_EXEC_TEMP_KILMASK_C].u[0]); + if (quad->inout.mask == 0) + return FALSE; + + /* store outputs */ + { + const ubyte *sem_name = base->info.output_semantic_name; + const ubyte *sem_index = base->info.output_semantic_index; + const uint n = base->info.num_outputs; + uint i; + for (i = 0; i < n; i++) { + switch (sem_name[i]) { + case TGSI_SEMANTIC_COLOR: + { + uint cbuf = sem_index[i]; + memcpy(quad->output.color[cbuf], + &machine->Outputs[i].xyzw[0].f[0], + sizeof(quad->output.color[0]) ); + } + break; + case TGSI_SEMANTIC_POSITION: + { + uint j; + for (j = 0; j < 4; j++) { + quad->output.depth[j] = machine->Outputs[0].xyzw[2].f[j]; + } + } + break; + } + } + } + + return TRUE; } diff --git a/src/gallium/drivers/softpipe/sp_prim_setup.c b/src/gallium/drivers/softpipe/sp_prim_setup.c deleted file mode 100644 index 038ff04d4f1..00000000000 --- a/src/gallium/drivers/softpipe/sp_prim_setup.c +++ /dev/null @@ -1,190 +0,0 @@ -/************************************************************************** - * - * Copyright 2007 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. - * - **************************************************************************/ - -/** - * \brief A draw stage that drives our triangle setup routines from - * within the draw pipeline. One of two ways to drive setup, the - * other being in sp_prim_vbuf.c. - * - * \author Keith Whitwell <[email protected]> - * \author Brian Paul - */ - - -#include "sp_context.h" -#include "sp_setup.h" -#include "sp_state.h" -#include "sp_prim_setup.h" -#include "draw/draw_pipe.h" -#include "draw/draw_vertex.h" -#include "util/u_memory.h" - -/** - * Triangle setup info (derived from draw_stage). - * Also used for line drawing (taking some liberties). - */ -struct setup_stage { - struct draw_stage stage; /**< This must be first (base class) */ - - struct setup_context *setup; -}; - - - -/** - * Basically a cast wrapper. - */ -static INLINE struct setup_stage *setup_stage( struct draw_stage *stage ) -{ - return (struct setup_stage *)stage; -} - - -typedef const float (*cptrf4)[4]; - -static void -do_tri(struct draw_stage *stage, struct prim_header *prim) -{ - struct setup_stage *setup = setup_stage( stage ); - - setup_tri( setup->setup, - (cptrf4)prim->v[0]->data, - (cptrf4)prim->v[1]->data, - (cptrf4)prim->v[2]->data ); -} - -static void -do_line(struct draw_stage *stage, struct prim_header *prim) -{ - struct setup_stage *setup = setup_stage( stage ); - - setup_line( setup->setup, - (cptrf4)prim->v[0]->data, - (cptrf4)prim->v[1]->data ); -} - -static void -do_point(struct draw_stage *stage, struct prim_header *prim) -{ - struct setup_stage *setup = setup_stage( stage ); - - setup_point( setup->setup, - (cptrf4)prim->v[0]->data ); -} - - - - -static void setup_begin( struct draw_stage *stage ) -{ - struct setup_stage *setup = setup_stage(stage); - - setup_prepare( setup->setup ); - - stage->point = do_point; - stage->line = do_line; - stage->tri = do_tri; -} - - -static void setup_first_point( struct draw_stage *stage, - struct prim_header *header ) -{ - setup_begin(stage); - stage->point( stage, header ); -} - -static void setup_first_line( struct draw_stage *stage, - struct prim_header *header ) -{ - setup_begin(stage); - stage->line( stage, header ); -} - - -static void setup_first_tri( struct draw_stage *stage, - struct prim_header *header ) -{ - setup_begin(stage); - stage->tri( stage, header ); -} - - - -static void setup_flush( struct draw_stage *stage, - unsigned flags ) -{ - stage->point = setup_first_point; - stage->line = setup_first_line; - stage->tri = setup_first_tri; -} - - -static void reset_stipple_counter( struct draw_stage *stage ) -{ -} - - -static void render_destroy( struct draw_stage *stage ) -{ - struct setup_stage *ssetup = setup_stage(stage); - setup_destroy_context(ssetup->setup); - FREE( stage ); -} - - -/** - * Create a new primitive setup/render stage. - */ -struct draw_stage *sp_draw_render_stage( struct softpipe_context *softpipe ) -{ - struct setup_stage *sstage = CALLOC_STRUCT(setup_stage); - - sstage->setup = setup_create_context(softpipe); - sstage->stage.draw = softpipe->draw; - sstage->stage.point = setup_first_point; - sstage->stage.line = setup_first_line; - sstage->stage.tri = setup_first_tri; - sstage->stage.flush = setup_flush; - sstage->stage.reset_stipple_counter = reset_stipple_counter; - sstage->stage.destroy = render_destroy; - - return (struct draw_stage *)sstage; -} - -struct setup_context * -sp_draw_setup_context( struct draw_stage *stage ) -{ - struct setup_stage *ssetup = setup_stage(stage); - return ssetup->setup; -} - -void -sp_draw_flush( struct draw_stage *stage ) -{ - stage->flush( stage, 0 ); -} diff --git a/src/gallium/drivers/softpipe/sp_prim_setup.h b/src/gallium/drivers/softpipe/sp_prim_setup.h deleted file mode 100644 index 49bdd98ed87..00000000000 --- a/src/gallium/drivers/softpipe/sp_prim_setup.h +++ /dev/null @@ -1,85 +0,0 @@ -/************************************************************************** - * - * Copyright 2007 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. - * - **************************************************************************/ - - -#ifndef SP_PRIM_SETUP_H -#define SP_PRIM_SETUP_H - - -/** - * vbuf is a special stage to gather the stream of triangles, lines, points - * together and reconstruct vertex buffers for hardware upload. - * - * First attempt, work in progress. - * - * TODO: - * - separate out vertex buffer building and primitive emit, ie >1 draw per vb. - * - tell vbuf stage how to build hw vertices directly - * - pass vbuf stage a buffer pointer for direct emit to agp/vram. - * - * - * - * Vertices are just an array of floats, with all the attributes - * packed. We currently assume a layout like: - * - * attr[0][0..3] - window position - * attr[1..n][0..3] - remaining attributes. - * - * Attributes are assumed to be 4 floats wide but are packed so that - * all the enabled attributes run contiguously. - */ - - -struct draw_stage; -struct softpipe_context; - - -typedef void (*vbuf_draw_func)( struct pipe_context *pipe, - unsigned prim, - const ushort *elements, - unsigned nr_elements, - const void *vertex_buffer, - unsigned nr_vertices ); - - -extern struct draw_stage * -sp_draw_render_stage( struct softpipe_context *softpipe ); - -extern struct setup_context * -sp_draw_setup_context( struct draw_stage * ); - -extern void -sp_draw_flush( struct draw_stage * ); - - -extern struct draw_stage * -sp_draw_vbuf_stage( struct draw_context *draw_context, - struct pipe_context *pipe, - vbuf_draw_func draw ); - - -#endif /* SP_PRIM_SETUP_H */ diff --git a/src/gallium/drivers/softpipe/sp_prim_vbuf.c b/src/gallium/drivers/softpipe/sp_prim_vbuf.c index 42021789ea8..e603c20fc4c 100644 --- a/src/gallium/drivers/softpipe/sp_prim_vbuf.c +++ b/src/gallium/drivers/softpipe/sp_prim_vbuf.c @@ -37,13 +37,13 @@ #include "sp_context.h" +#include "sp_setup.h" #include "sp_state.h" #include "sp_prim_vbuf.h" -#include "sp_prim_setup.h" -#include "sp_setup.h" #include "draw/draw_context.h" #include "draw/draw_vbuf.h" #include "util/u_memory.h" +#include "util/u_prim.h" #define SP_MAX_VBUF_INDEXES 1024 @@ -58,6 +58,8 @@ struct softpipe_vbuf_render { struct vbuf_render base; struct softpipe_context *softpipe; + struct setup_context *setup; + uint prim; uint vertex_size; uint nr_vertices; @@ -74,6 +76,11 @@ softpipe_vbuf_render(struct vbuf_render *vbr) } + + + + + static const struct vertex_info * sp_vbuf_get_vertex_info(struct vbuf_render *vbr) { @@ -104,36 +111,6 @@ sp_vbuf_allocate_vertices(struct vbuf_render *vbr, static void sp_vbuf_release_vertices(struct vbuf_render *vbr) { -#if 0 - { - struct softpipe_vbuf_render *cvbr = softpipe_vbuf_render(vbr); - const struct vertex_info *info = - softpipe_get_vbuf_vertex_info(cvbr->softpipe); - const float *vtx = (const float *) cvbr->vertex_buffer; - uint i, j; - debug_printf("%s (vtx_size = %u, vtx_used = %u)\n", - __FUNCTION__, cvbr->vertex_size, cvbr->nr_vertices); - for (i = 0; i < cvbr->nr_vertices; i++) { - for (j = 0; j < info->num_attribs; j++) { - uint k; - switch (info->attrib[j].emit) { - case EMIT_4F: k = 4; break; - case EMIT_3F: k = 3; break; - case EMIT_2F: k = 2; break; - case EMIT_1F: k = 1; break; - default: assert(0); - } - debug_printf("Vert %u attr %u: ", i, j); - while (k-- > 0) { - debug_printf("%g ", vtx[0]); - vtx++; - } - debug_printf("\n"); - } - } - } -#endif - /* keep the old allocation for next time */ } @@ -159,14 +136,11 @@ static boolean sp_vbuf_set_primitive(struct vbuf_render *vbr, unsigned prim) { struct softpipe_vbuf_render *cvbr = softpipe_vbuf_render(vbr); - - /* XXX: break this dependency - make setup_context live under - * softpipe, rename the old "setup" draw stage to something else. - */ - struct setup_context *setup_ctx = sp_draw_setup_context(cvbr->softpipe->setup); + struct setup_context *setup_ctx = cvbr->setup; setup_prepare( setup_ctx ); + cvbr->softpipe->reduced_prim = u_reduced_prim(prim); cvbr->prim = prim; return TRUE; @@ -191,14 +165,9 @@ sp_vbuf_draw(struct vbuf_render *vbr, const ushort *indices, uint nr) struct softpipe_context *softpipe = cvbr->softpipe; const unsigned stride = softpipe->vertex_info_vbuf.size * sizeof(float); const void *vertex_buffer = cvbr->vertex_buffer; + struct setup_context *setup_ctx = cvbr->setup; unsigned i; - /* XXX: break this dependency - make setup_context live under - * softpipe, rename the old "setup" draw stage to something else. - */ - struct draw_stage *setup = softpipe->setup; - struct setup_context *setup_ctx = sp_draw_setup_context(setup); - switch (cvbr->prim) { case PIPE_PRIM_POINTS: for (i = 0; i < nr; i++) { @@ -237,14 +206,16 @@ sp_vbuf_draw(struct vbuf_render *vbr, const ushort *indices, uint nr) break; case PIPE_PRIM_TRIANGLES: - for (i = 2; i < nr; i += 3) { - if (softpipe->rasterizer->flatshade_first) { + if (softpipe->rasterizer->flatshade_first) { + for (i = 2; i < nr; i += 3) { setup_tri( setup_ctx, get_vert(vertex_buffer, indices[i-1], stride), get_vert(vertex_buffer, indices[i-0], stride), get_vert(vertex_buffer, indices[i-2], stride) ); } - else { + } + else { + for (i = 2; i < nr; i += 3) { setup_tri( setup_ctx, get_vert(vertex_buffer, indices[i-2], stride), get_vert(vertex_buffer, indices[i-1], stride), @@ -254,14 +225,16 @@ sp_vbuf_draw(struct vbuf_render *vbr, const ushort *indices, uint nr) break; case PIPE_PRIM_TRIANGLE_STRIP: - for (i = 2; i < nr; i += 1) { - if (softpipe->rasterizer->flatshade_first) { + if (softpipe->rasterizer->flatshade_first) { + for (i = 2; i < nr; i += 1) { setup_tri( setup_ctx, get_vert(vertex_buffer, indices[i+(i&1)-1], stride), get_vert(vertex_buffer, indices[i-(i&1)], stride), get_vert(vertex_buffer, indices[i-2], stride) ); } - else { + } + else { + for (i = 2; i < nr; i += 1) { setup_tri( setup_ctx, get_vert(vertex_buffer, indices[i+(i&1)-2], stride), get_vert(vertex_buffer, indices[i-(i&1)-1], stride), @@ -271,14 +244,16 @@ sp_vbuf_draw(struct vbuf_render *vbr, const ushort *indices, uint nr) break; case PIPE_PRIM_TRIANGLE_FAN: - for (i = 2; i < nr; i += 1) { - if (softpipe->rasterizer->flatshade_first) { + if (softpipe->rasterizer->flatshade_first) { + for (i = 2; i < nr; i += 1) { setup_tri( setup_ctx, get_vert(vertex_buffer, indices[i-0], stride), get_vert(vertex_buffer, indices[0], stride), get_vert(vertex_buffer, indices[i-1], stride) ); } - else { + } + else { + for (i = 2; i < nr; i += 1) { setup_tri( setup_ctx, get_vert(vertex_buffer, indices[0], stride), get_vert(vertex_buffer, indices[i-1], stride), @@ -288,8 +263,8 @@ sp_vbuf_draw(struct vbuf_render *vbr, const ushort *indices, uint nr) break; case PIPE_PRIM_QUADS: - for (i = 3; i < nr; i += 4) { - if (softpipe->rasterizer->flatshade_first) { + if (softpipe->rasterizer->flatshade_first) { + for (i = 3; i < nr; i += 4) { setup_tri( setup_ctx, get_vert(vertex_buffer, indices[i-2], stride), get_vert(vertex_buffer, indices[i-1], stride), @@ -299,7 +274,9 @@ sp_vbuf_draw(struct vbuf_render *vbr, const ushort *indices, uint nr) get_vert(vertex_buffer, indices[i-0], stride), get_vert(vertex_buffer, indices[i-3], stride) ); } - else { + } + else { + for (i = 3; i < nr; i += 4) { setup_tri( setup_ctx, get_vert(vertex_buffer, indices[i-3], stride), get_vert(vertex_buffer, indices[i-2], stride), @@ -314,8 +291,8 @@ sp_vbuf_draw(struct vbuf_render *vbr, const ushort *indices, uint nr) break; case PIPE_PRIM_QUAD_STRIP: - for (i = 3; i < nr; i += 2) { - if (softpipe->rasterizer->flatshade_first) { + if (softpipe->rasterizer->flatshade_first) { + for (i = 3; i < nr; i += 2) { setup_tri( setup_ctx, get_vert(vertex_buffer, indices[i-0], stride), get_vert(vertex_buffer, indices[i-1], stride), @@ -325,7 +302,9 @@ sp_vbuf_draw(struct vbuf_render *vbr, const ushort *indices, uint nr) get_vert(vertex_buffer, indices[i-0], stride), get_vert(vertex_buffer, indices[i-3], stride) ); } - else { + } + else { + for (i = 3; i < nr; i += 2) { setup_tri( setup_ctx, get_vert(vertex_buffer, indices[i-3], stride), get_vert(vertex_buffer, indices[i-2], stride), @@ -355,11 +334,6 @@ sp_vbuf_draw(struct vbuf_render *vbr, const ushort *indices, uint nr) default: assert(0); } - - /* XXX: why are we calling this??? If we had to call something, it - * would be a function in sp_setup.c: - */ - sp_draw_flush( setup ); } @@ -372,17 +346,12 @@ sp_vbuf_draw_arrays(struct vbuf_render *vbr, uint start, uint nr) { struct softpipe_vbuf_render *cvbr = softpipe_vbuf_render(vbr); struct softpipe_context *softpipe = cvbr->softpipe; + struct setup_context *setup_ctx = cvbr->setup; const unsigned stride = softpipe->vertex_info_vbuf.size * sizeof(float); const void *vertex_buffer = (void *) get_vert(cvbr->vertex_buffer, start, stride); unsigned i; - /* XXX: break this dependency - make setup_context live under - * softpipe, rename the old "setup" draw stage to something else. - */ - struct draw_stage *setup = softpipe->setup; - struct setup_context *setup_ctx = sp_draw_setup_context(setup); - switch (cvbr->prim) { case PIPE_PRIM_POINTS: for (i = 0; i < nr; i++) { @@ -421,14 +390,16 @@ sp_vbuf_draw_arrays(struct vbuf_render *vbr, uint start, uint nr) break; case PIPE_PRIM_TRIANGLES: - for (i = 2; i < nr; i += 3) { - if (softpipe->rasterizer->flatshade_first) { + if (softpipe->rasterizer->flatshade_first) { + for (i = 2; i < nr; i += 3) { setup_tri( setup_ctx, get_vert(vertex_buffer, i-1, stride), get_vert(vertex_buffer, i-0, stride), get_vert(vertex_buffer, i-2, stride) ); } - else { + } + else { + for (i = 2; i < nr; i += 3) { setup_tri( setup_ctx, get_vert(vertex_buffer, i-2, stride), get_vert(vertex_buffer, i-1, stride), @@ -438,14 +409,16 @@ sp_vbuf_draw_arrays(struct vbuf_render *vbr, uint start, uint nr) break; case PIPE_PRIM_TRIANGLE_STRIP: - for (i = 2; i < nr; i++) { - if (softpipe->rasterizer->flatshade_first) { + if (softpipe->rasterizer->flatshade_first) { + for (i = 2; i < nr; i++) { setup_tri( setup_ctx, get_vert(vertex_buffer, i+(i&1)-1, stride), get_vert(vertex_buffer, i-(i&1), stride), get_vert(vertex_buffer, i-2, stride) ); } - else { + } + else { + for (i = 2; i < nr; i++) { setup_tri( setup_ctx, get_vert(vertex_buffer, i+(i&1)-2, stride), get_vert(vertex_buffer, i-(i&1)-1, stride), @@ -455,14 +428,16 @@ sp_vbuf_draw_arrays(struct vbuf_render *vbr, uint start, uint nr) break; case PIPE_PRIM_TRIANGLE_FAN: - for (i = 2; i < nr; i += 1) { - if (softpipe->rasterizer->flatshade_first) { + if (softpipe->rasterizer->flatshade_first) { + for (i = 2; i < nr; i += 1) { setup_tri( setup_ctx, get_vert(vertex_buffer, i-0, stride), get_vert(vertex_buffer, 0, stride), get_vert(vertex_buffer, i-1, stride) ); } - else { + } + else { + for (i = 2; i < nr; i += 1) { setup_tri( setup_ctx, get_vert(vertex_buffer, 0, stride), get_vert(vertex_buffer, i-1, stride), @@ -472,8 +447,8 @@ sp_vbuf_draw_arrays(struct vbuf_render *vbr, uint start, uint nr) break; case PIPE_PRIM_QUADS: - for (i = 3; i < nr; i += 4) { - if (softpipe->rasterizer->flatshade_first) { + if (softpipe->rasterizer->flatshade_first) { + for (i = 3; i < nr; i += 4) { setup_tri( setup_ctx, get_vert(vertex_buffer, i-2, stride), get_vert(vertex_buffer, i-1, stride), @@ -483,7 +458,9 @@ sp_vbuf_draw_arrays(struct vbuf_render *vbr, uint start, uint nr) get_vert(vertex_buffer, i-0, stride), get_vert(vertex_buffer, i-3, stride) ); } - else { + } + else { + for (i = 3; i < nr; i += 4) { setup_tri( setup_ctx, get_vert(vertex_buffer, i-3, stride), get_vert(vertex_buffer, i-2, stride), @@ -497,8 +474,8 @@ sp_vbuf_draw_arrays(struct vbuf_render *vbr, uint start, uint nr) break; case PIPE_PRIM_QUAD_STRIP: - for (i = 3; i < nr; i += 2) { - if (softpipe->rasterizer->flatshade_first) { + if (softpipe->rasterizer->flatshade_first) { + for (i = 3; i < nr; i += 2) { setup_tri( setup_ctx, get_vert(vertex_buffer, i-0, stride), get_vert(vertex_buffer, i-1, stride), @@ -508,7 +485,9 @@ sp_vbuf_draw_arrays(struct vbuf_render *vbr, uint start, uint nr) get_vert(vertex_buffer, i-0, stride), get_vert(vertex_buffer, i-3, stride) ); } - else { + } + else { + for (i = 3; i < nr; i += 2) { setup_tri( setup_ctx, get_vert(vertex_buffer, i-3, stride), get_vert(vertex_buffer, i-2, stride), @@ -546,40 +525,38 @@ static void sp_vbuf_destroy(struct vbuf_render *vbr) { struct softpipe_vbuf_render *cvbr = softpipe_vbuf_render(vbr); - cvbr->softpipe->vbuf_render = NULL; + setup_destroy_context(cvbr->setup); FREE(cvbr); } /** - * Initialize the post-transform vertex buffer information for the given - * context. + * Create the post-transform vertex handler for the given context. */ -void -sp_init_vbuf(struct softpipe_context *sp) +struct vbuf_render * +sp_create_vbuf_backend(struct softpipe_context *sp) { - assert(sp->draw); + struct softpipe_vbuf_render *cvbr = CALLOC_STRUCT(softpipe_vbuf_render); - sp->vbuf_render = CALLOC_STRUCT(softpipe_vbuf_render); + assert(sp->draw); - sp->vbuf_render->base.max_indices = SP_MAX_VBUF_INDEXES; - sp->vbuf_render->base.max_vertex_buffer_bytes = SP_MAX_VBUF_SIZE; - sp->vbuf_render->base.get_vertex_info = sp_vbuf_get_vertex_info; - sp->vbuf_render->base.allocate_vertices = sp_vbuf_allocate_vertices; - sp->vbuf_render->base.map_vertices = sp_vbuf_map_vertices; - sp->vbuf_render->base.unmap_vertices = sp_vbuf_unmap_vertices; - sp->vbuf_render->base.set_primitive = sp_vbuf_set_primitive; - sp->vbuf_render->base.draw = sp_vbuf_draw; - sp->vbuf_render->base.draw_arrays = sp_vbuf_draw_arrays; - sp->vbuf_render->base.release_vertices = sp_vbuf_release_vertices; - sp->vbuf_render->base.destroy = sp_vbuf_destroy; + cvbr->base.max_indices = SP_MAX_VBUF_INDEXES; + cvbr->base.max_vertex_buffer_bytes = SP_MAX_VBUF_SIZE; - sp->vbuf_render->softpipe = sp; + cvbr->base.get_vertex_info = sp_vbuf_get_vertex_info; + cvbr->base.allocate_vertices = sp_vbuf_allocate_vertices; + cvbr->base.map_vertices = sp_vbuf_map_vertices; + cvbr->base.unmap_vertices = sp_vbuf_unmap_vertices; + cvbr->base.set_primitive = sp_vbuf_set_primitive; + cvbr->base.draw = sp_vbuf_draw; + cvbr->base.draw_arrays = sp_vbuf_draw_arrays; + cvbr->base.release_vertices = sp_vbuf_release_vertices; + cvbr->base.destroy = sp_vbuf_destroy; - sp->vbuf = draw_vbuf_stage(sp->draw, &sp->vbuf_render->base); + cvbr->softpipe = sp; - draw_set_rasterize_stage(sp->draw, sp->vbuf); + cvbr->setup = setup_create_context(cvbr->softpipe); - draw_set_render(sp->draw, &sp->vbuf_render->base); + return &cvbr->base; } diff --git a/src/gallium/drivers/softpipe/sp_prim_vbuf.h b/src/gallium/drivers/softpipe/sp_prim_vbuf.h index 1de9cc2a894..ad01cc2f289 100644 --- a/src/gallium/drivers/softpipe/sp_prim_vbuf.h +++ b/src/gallium/drivers/softpipe/sp_prim_vbuf.h @@ -31,8 +31,8 @@ struct softpipe_context; -extern void -sp_init_vbuf(struct softpipe_context *softpipe); +extern struct vbuf_render * +sp_create_vbuf_backend(struct softpipe_context *softpipe); #endif /* SP_VBUF_H */ diff --git a/src/gallium/drivers/softpipe/sp_quad.h b/src/gallium/drivers/softpipe/sp_quad.h index bd6c6cb9123..a3236bd1169 100644 --- a/src/gallium/drivers/softpipe/sp_quad.h +++ b/src/gallium/drivers/softpipe/sp_quad.h @@ -97,10 +97,10 @@ struct quad_header { struct quad_header_inout inout; struct quad_header_output output; - const struct tgsi_interp_coef *coef; + /* Redundant/duplicated: + */ const struct tgsi_interp_coef *posCoef; - - unsigned nr_attrs; + const struct tgsi_interp_coef *coef; }; #endif /* SP_QUAD_H */ diff --git a/src/gallium/drivers/softpipe/sp_quad_alpha_test.c b/src/gallium/drivers/softpipe/sp_quad_alpha_test.c deleted file mode 100644 index 0845bae0e68..00000000000 --- a/src/gallium/drivers/softpipe/sp_quad_alpha_test.c +++ /dev/null @@ -1,108 +0,0 @@ - -/** - * quad alpha test - */ - -#include "sp_context.h" -#include "sp_quad.h" -#include "sp_quad_pipe.h" -#include "pipe/p_defines.h" -#include "util/u_memory.h" - - -static void -alpha_test_quad(struct quad_stage *qs, struct quad_header *quad) -{ - struct softpipe_context *softpipe = qs->softpipe; - const float ref = softpipe->depth_stencil->alpha.ref_value; - unsigned passMask = 0x0, j; - const uint cbuf = 0; /* only output[0].alpha is tested */ - const float *aaaa = quad->output.color[cbuf][3]; - - switch (softpipe->depth_stencil->alpha.func) { - case PIPE_FUNC_NEVER: - break; - case PIPE_FUNC_LESS: - /* - * If mask were an array [4] we could do this SIMD-style: - * passMask = (quad->outputs.color[0][3] <= vec4(ref)); - */ - for (j = 0; j < QUAD_SIZE; j++) { - if (aaaa[j] < ref) { - passMask |= (1 << j); - } - } - break; - case PIPE_FUNC_EQUAL: - for (j = 0; j < QUAD_SIZE; j++) { - if (aaaa[j] == ref) { - passMask |= (1 << j); - } - } - break; - case PIPE_FUNC_LEQUAL: - for (j = 0; j < QUAD_SIZE; j++) { - if (aaaa[j] <= ref) { - passMask |= (1 << j); - } - } - break; - case PIPE_FUNC_GREATER: - for (j = 0; j < QUAD_SIZE; j++) { - if (aaaa[j] > ref) { - passMask |= (1 << j); - } - } - break; - case PIPE_FUNC_NOTEQUAL: - for (j = 0; j < QUAD_SIZE; j++) { - if (aaaa[j] != ref) { - passMask |= (1 << j); - } - } - break; - case PIPE_FUNC_GEQUAL: - for (j = 0; j < QUAD_SIZE; j++) { - if (aaaa[j] >= ref) { - passMask |= (1 << j); - } - } - break; - case PIPE_FUNC_ALWAYS: - passMask = MASK_ALL; - break; - default: - assert(0); - } - - quad->inout.mask &= passMask; - - if (quad->inout.mask) - qs->next->run(qs->next, quad); -} - - -static void alpha_test_begin(struct quad_stage *qs) -{ - qs->next->begin(qs->next); -} - - -static void alpha_test_destroy(struct quad_stage *qs) -{ - FREE( qs ); -} - - -struct quad_stage * -sp_quad_alpha_test_stage( struct softpipe_context *softpipe ) -{ - struct quad_stage *stage = CALLOC_STRUCT(quad_stage); - - stage->softpipe = softpipe; - stage->begin = alpha_test_begin; - stage->run = alpha_test_quad; - stage->destroy = alpha_test_destroy; - - return stage; -} diff --git a/src/gallium/drivers/softpipe/sp_quad_blend.c b/src/gallium/drivers/softpipe/sp_quad_blend.c index b1e18805c70..e243c63fa23 100644 --- a/src/gallium/drivers/softpipe/sp_quad_blend.c +++ b/src/gallium/drivers/softpipe/sp_quad_blend.c @@ -117,644 +117,865 @@ do { \ static void -logicop_quad(struct quad_stage *qs, struct quad_header *quad) +logicop_quad(struct quad_stage *qs, + float (*quadColor)[4], + float (*dest)[4]) { struct softpipe_context *softpipe = qs->softpipe; - uint cbuf; + ubyte src[4][4], dst[4][4], res[4][4]; + uint *src4 = (uint *) src; + uint *dst4 = (uint *) dst; + uint *res4 = (uint *) res; + uint j; + + + /* convert to ubyte */ + for (j = 0; j < 4; j++) { /* loop over R,G,B,A channels */ + dst[j][0] = float_to_ubyte(dest[j][0]); /* P0 */ + dst[j][1] = float_to_ubyte(dest[j][1]); /* P1 */ + dst[j][2] = float_to_ubyte(dest[j][2]); /* P2 */ + dst[j][3] = float_to_ubyte(dest[j][3]); /* P3 */ + + src[j][0] = float_to_ubyte(quadColor[j][0]); /* P0 */ + src[j][1] = float_to_ubyte(quadColor[j][1]); /* P1 */ + src[j][2] = float_to_ubyte(quadColor[j][2]); /* P2 */ + src[j][3] = float_to_ubyte(quadColor[j][3]); /* P3 */ + } - /* loop over colorbuffer outputs */ - for (cbuf = 0; cbuf < softpipe->framebuffer.nr_cbufs; cbuf++) { - float dest[4][QUAD_SIZE]; - ubyte src[4][4], dst[4][4], res[4][4]; - uint *src4 = (uint *) src; - uint *dst4 = (uint *) dst; - uint *res4 = (uint *) res; - struct softpipe_cached_tile * - tile = sp_get_cached_tile(softpipe, - softpipe->cbuf_cache[cbuf], - quad->input.x0, quad->input.y0); - float (*quadColor)[4] = quad->output.color[cbuf]; - uint i, j; + switch (softpipe->blend->logicop_func) { + case PIPE_LOGICOP_CLEAR: + for (j = 0; j < 4; j++) + res4[j] = 0; + break; + case PIPE_LOGICOP_NOR: + for (j = 0; j < 4; j++) + res4[j] = ~(src4[j] | dst4[j]); + break; + case PIPE_LOGICOP_AND_INVERTED: + for (j = 0; j < 4; j++) + res4[j] = ~src4[j] & dst4[j]; + break; + case PIPE_LOGICOP_COPY_INVERTED: + for (j = 0; j < 4; j++) + res4[j] = ~src4[j]; + break; + case PIPE_LOGICOP_AND_REVERSE: + for (j = 0; j < 4; j++) + res4[j] = src4[j] & ~dst4[j]; + break; + case PIPE_LOGICOP_INVERT: + for (j = 0; j < 4; j++) + res4[j] = ~dst4[j]; + break; + case PIPE_LOGICOP_XOR: + for (j = 0; j < 4; j++) + res4[j] = dst4[j] ^ src4[j]; + break; + case PIPE_LOGICOP_NAND: + for (j = 0; j < 4; j++) + res4[j] = ~(src4[j] & dst4[j]); + break; + case PIPE_LOGICOP_AND: + for (j = 0; j < 4; j++) + res4[j] = src4[j] & dst4[j]; + break; + case PIPE_LOGICOP_EQUIV: + for (j = 0; j < 4; j++) + res4[j] = ~(src4[j] ^ dst4[j]); + break; + case PIPE_LOGICOP_NOOP: + for (j = 0; j < 4; j++) + res4[j] = dst4[j]; + break; + case PIPE_LOGICOP_OR_INVERTED: + for (j = 0; j < 4; j++) + res4[j] = ~src4[j] | dst4[j]; + break; + case PIPE_LOGICOP_COPY: + for (j = 0; j < 4; j++) + res4[j] = src4[j]; + break; + case PIPE_LOGICOP_OR_REVERSE: + for (j = 0; j < 4; j++) + res4[j] = src4[j] | ~dst4[j]; + break; + case PIPE_LOGICOP_OR: + for (j = 0; j < 4; j++) + res4[j] = src4[j] | dst4[j]; + break; + case PIPE_LOGICOP_SET: + for (j = 0; j < 4; j++) + res4[j] = ~0; + break; + default: + assert(0); + } - /* get/swizzle dest colors */ - for (j = 0; j < QUAD_SIZE; j++) { - int x = (quad->input.x0 & (TILE_SIZE-1)) + (j & 1); - int y = (quad->input.y0 & (TILE_SIZE-1)) + (j >> 1); - for (i = 0; i < 4; i++) { - dest[i][j] = tile->data.color[y][x][i]; - } - } + for (j = 0; j < 4; j++) { + quadColor[j][0] = ubyte_to_float(res[j][0]); + quadColor[j][1] = ubyte_to_float(res[j][1]); + quadColor[j][2] = ubyte_to_float(res[j][2]); + quadColor[j][3] = ubyte_to_float(res[j][3]); + } +} - /* convert to ubyte */ - for (j = 0; j < 4; j++) { /* loop over R,G,B,A channels */ - dst[j][0] = float_to_ubyte(dest[j][0]); /* P0 */ - dst[j][1] = float_to_ubyte(dest[j][1]); /* P1 */ - dst[j][2] = float_to_ubyte(dest[j][2]); /* P2 */ - dst[j][3] = float_to_ubyte(dest[j][3]); /* P3 */ - - src[j][0] = float_to_ubyte(quadColor[j][0]); /* P0 */ - src[j][1] = float_to_ubyte(quadColor[j][1]); /* P1 */ - src[j][2] = float_to_ubyte(quadColor[j][2]); /* P2 */ - src[j][3] = float_to_ubyte(quadColor[j][3]); /* P3 */ - } - switch (softpipe->blend->logicop_func) { - case PIPE_LOGICOP_CLEAR: - for (j = 0; j < 4; j++) - res4[j] = 0; - break; - case PIPE_LOGICOP_NOR: - for (j = 0; j < 4; j++) - res4[j] = ~(src4[j] | dst4[j]); - break; - case PIPE_LOGICOP_AND_INVERTED: - for (j = 0; j < 4; j++) - res4[j] = ~src4[j] & dst4[j]; - break; - case PIPE_LOGICOP_COPY_INVERTED: - for (j = 0; j < 4; j++) - res4[j] = ~src4[j]; - break; - case PIPE_LOGICOP_AND_REVERSE: - for (j = 0; j < 4; j++) - res4[j] = src4[j] & ~dst4[j]; - break; - case PIPE_LOGICOP_INVERT: - for (j = 0; j < 4; j++) - res4[j] = ~dst4[j]; - break; - case PIPE_LOGICOP_XOR: - for (j = 0; j < 4; j++) - res4[j] = dst4[j] ^ src4[j]; - break; - case PIPE_LOGICOP_NAND: - for (j = 0; j < 4; j++) - res4[j] = ~(src4[j] & dst4[j]); - break; - case PIPE_LOGICOP_AND: - for (j = 0; j < 4; j++) - res4[j] = src4[j] & dst4[j]; - break; - case PIPE_LOGICOP_EQUIV: - for (j = 0; j < 4; j++) - res4[j] = ~(src4[j] ^ dst4[j]); - break; - case PIPE_LOGICOP_NOOP: - for (j = 0; j < 4; j++) - res4[j] = dst4[j]; - break; - case PIPE_LOGICOP_OR_INVERTED: - for (j = 0; j < 4; j++) - res4[j] = ~src4[j] | dst4[j]; - break; - case PIPE_LOGICOP_COPY: - for (j = 0; j < 4; j++) - res4[j] = src4[j]; - break; - case PIPE_LOGICOP_OR_REVERSE: - for (j = 0; j < 4; j++) - res4[j] = src4[j] | ~dst4[j]; - break; - case PIPE_LOGICOP_OR: - for (j = 0; j < 4; j++) - res4[j] = src4[j] | dst4[j]; - break; - case PIPE_LOGICOP_SET: - for (j = 0; j < 4; j++) - res4[j] = ~0; - break; - default: - assert(0); - } - for (j = 0; j < 4; j++) { - quadColor[j][0] = ubyte_to_float(res[j][0]); - quadColor[j][1] = ubyte_to_float(res[j][1]); - quadColor[j][2] = ubyte_to_float(res[j][2]); - quadColor[j][3] = ubyte_to_float(res[j][3]); - } +static void +blend_quad(struct quad_stage *qs, + float (*quadColor)[4], + float (*dest)[4]) +{ + static const float zero[4] = { 0, 0, 0, 0 }; + static const float one[4] = { 1, 1, 1, 1 }; + struct softpipe_context *softpipe = qs->softpipe; + float source[4][QUAD_SIZE]; + + /* + * Compute src/first term RGB + */ + switch (softpipe->blend->rgb_src_factor) { + case PIPE_BLENDFACTOR_ONE: + VEC4_COPY(source[0], quadColor[0]); /* R */ + VEC4_COPY(source[1], quadColor[1]); /* G */ + VEC4_COPY(source[2], quadColor[2]); /* B */ + break; + case PIPE_BLENDFACTOR_SRC_COLOR: + VEC4_MUL(source[0], quadColor[0], quadColor[0]); /* R */ + VEC4_MUL(source[1], quadColor[1], quadColor[1]); /* G */ + VEC4_MUL(source[2], quadColor[2], quadColor[2]); /* B */ + break; + case PIPE_BLENDFACTOR_SRC_ALPHA: + { + const float *alpha = quadColor[3]; + VEC4_MUL(source[0], quadColor[0], alpha); /* R */ + VEC4_MUL(source[1], quadColor[1], alpha); /* G */ + VEC4_MUL(source[2], quadColor[2], alpha); /* B */ + } + break; + case PIPE_BLENDFACTOR_DST_COLOR: + VEC4_MUL(source[0], quadColor[0], dest[0]); /* R */ + VEC4_MUL(source[1], quadColor[1], dest[1]); /* G */ + VEC4_MUL(source[2], quadColor[2], dest[2]); /* B */ + break; + case PIPE_BLENDFACTOR_DST_ALPHA: + { + const float *alpha = dest[3]; + VEC4_MUL(source[0], quadColor[0], alpha); /* R */ + VEC4_MUL(source[1], quadColor[1], alpha); /* G */ + VEC4_MUL(source[2], quadColor[2], alpha); /* B */ + } + break; + case PIPE_BLENDFACTOR_SRC_ALPHA_SATURATE: + { + const float *alpha = quadColor[3]; + float diff[4], temp[4]; + VEC4_SUB(diff, one, dest[3]); + VEC4_MIN(temp, alpha, diff); + VEC4_MUL(source[0], quadColor[0], temp); /* R */ + VEC4_MUL(source[1], quadColor[1], temp); /* G */ + VEC4_MUL(source[2], quadColor[2], temp); /* B */ + } + break; + case PIPE_BLENDFACTOR_CONST_COLOR: + { + float comp[4]; + VEC4_SCALAR(comp, softpipe->blend_color.color[0]); /* R */ + VEC4_MUL(source[0], quadColor[0], comp); /* R */ + VEC4_SCALAR(comp, softpipe->blend_color.color[1]); /* G */ + VEC4_MUL(source[1], quadColor[1], comp); /* G */ + VEC4_SCALAR(comp, softpipe->blend_color.color[2]); /* B */ + VEC4_MUL(source[2], quadColor[2], comp); /* B */ + } + break; + case PIPE_BLENDFACTOR_CONST_ALPHA: + { + float alpha[4]; + VEC4_SCALAR(alpha, softpipe->blend_color.color[3]); + VEC4_MUL(source[0], quadColor[0], alpha); /* R */ + VEC4_MUL(source[1], quadColor[1], alpha); /* G */ + VEC4_MUL(source[2], quadColor[2], alpha); /* B */ + } + break; + case PIPE_BLENDFACTOR_SRC1_COLOR: + assert(0); /* to do */ + break; + case PIPE_BLENDFACTOR_SRC1_ALPHA: + assert(0); /* to do */ + break; + case PIPE_BLENDFACTOR_ZERO: + VEC4_COPY(source[0], zero); /* R */ + VEC4_COPY(source[1], zero); /* G */ + VEC4_COPY(source[2], zero); /* B */ + break; + case PIPE_BLENDFACTOR_INV_SRC_COLOR: + { + float inv_comp[4]; + VEC4_SUB(inv_comp, one, quadColor[0]); /* R */ + VEC4_MUL(source[0], quadColor[0], inv_comp); /* R */ + VEC4_SUB(inv_comp, one, quadColor[1]); /* G */ + VEC4_MUL(source[1], quadColor[1], inv_comp); /* G */ + VEC4_SUB(inv_comp, one, quadColor[2]); /* B */ + VEC4_MUL(source[2], quadColor[2], inv_comp); /* B */ + } + break; + case PIPE_BLENDFACTOR_INV_SRC_ALPHA: + { + float inv_alpha[4]; + VEC4_SUB(inv_alpha, one, quadColor[3]); + VEC4_MUL(source[0], quadColor[0], inv_alpha); /* R */ + VEC4_MUL(source[1], quadColor[1], inv_alpha); /* G */ + VEC4_MUL(source[2], quadColor[2], inv_alpha); /* B */ + } + break; + case PIPE_BLENDFACTOR_INV_DST_ALPHA: + { + float inv_alpha[4]; + VEC4_SUB(inv_alpha, one, dest[3]); + VEC4_MUL(source[0], quadColor[0], inv_alpha); /* R */ + VEC4_MUL(source[1], quadColor[1], inv_alpha); /* G */ + VEC4_MUL(source[2], quadColor[2], inv_alpha); /* B */ + } + break; + case PIPE_BLENDFACTOR_INV_DST_COLOR: + { + float inv_comp[4]; + VEC4_SUB(inv_comp, one, dest[0]); /* R */ + VEC4_MUL(source[0], quadColor[0], inv_comp); /* R */ + VEC4_SUB(inv_comp, one, dest[1]); /* G */ + VEC4_MUL(source[1], quadColor[1], inv_comp); /* G */ + VEC4_SUB(inv_comp, one, dest[2]); /* B */ + VEC4_MUL(source[2], quadColor[2], inv_comp); /* B */ + } + break; + case PIPE_BLENDFACTOR_INV_CONST_COLOR: + { + float inv_comp[4]; + /* R */ + VEC4_SCALAR(inv_comp, 1.0f - softpipe->blend_color.color[0]); + VEC4_MUL(source[0], quadColor[0], inv_comp); + /* G */ + VEC4_SCALAR(inv_comp, 1.0f - softpipe->blend_color.color[1]); + VEC4_MUL(source[1], quadColor[1], inv_comp); + /* B */ + VEC4_SCALAR(inv_comp, 1.0f - softpipe->blend_color.color[2]); + VEC4_MUL(source[2], quadColor[2], inv_comp); + } + break; + case PIPE_BLENDFACTOR_INV_CONST_ALPHA: + { + float inv_alpha[4]; + VEC4_SCALAR(inv_alpha, 1.0f - softpipe->blend_color.color[3]); + VEC4_MUL(source[0], quadColor[0], inv_alpha); /* R */ + VEC4_MUL(source[1], quadColor[1], inv_alpha); /* G */ + VEC4_MUL(source[2], quadColor[2], inv_alpha); /* B */ + } + break; + case PIPE_BLENDFACTOR_INV_SRC1_COLOR: + assert(0); /* to do */ + break; + case PIPE_BLENDFACTOR_INV_SRC1_ALPHA: + assert(0); /* to do */ + break; + default: + assert(0); + } + + /* + * Compute src/first term A + */ + switch (softpipe->blend->alpha_src_factor) { + case PIPE_BLENDFACTOR_ONE: + VEC4_COPY(source[3], quadColor[3]); /* A */ + break; + case PIPE_BLENDFACTOR_SRC_COLOR: + /* fall-through */ + case PIPE_BLENDFACTOR_SRC_ALPHA: + { + const float *alpha = quadColor[3]; + VEC4_MUL(source[3], quadColor[3], alpha); /* A */ + } + break; + case PIPE_BLENDFACTOR_DST_COLOR: + /* fall-through */ + case PIPE_BLENDFACTOR_DST_ALPHA: + VEC4_MUL(source[3], quadColor[3], dest[3]); /* A */ + break; + case PIPE_BLENDFACTOR_SRC_ALPHA_SATURATE: + /* multiply alpha by 1.0 */ + VEC4_COPY(source[3], quadColor[3]); /* A */ + break; + case PIPE_BLENDFACTOR_CONST_COLOR: + /* fall-through */ + case PIPE_BLENDFACTOR_CONST_ALPHA: + { + float comp[4]; + VEC4_SCALAR(comp, softpipe->blend_color.color[3]); /* A */ + VEC4_MUL(source[3], quadColor[3], comp); /* A */ + } + break; + case PIPE_BLENDFACTOR_ZERO: + VEC4_COPY(source[3], zero); /* A */ + break; + case PIPE_BLENDFACTOR_INV_SRC_COLOR: + /* fall-through */ + case PIPE_BLENDFACTOR_INV_SRC_ALPHA: + { + float inv_alpha[4]; + VEC4_SUB(inv_alpha, one, quadColor[3]); + VEC4_MUL(source[3], quadColor[3], inv_alpha); /* A */ + } + break; + case PIPE_BLENDFACTOR_INV_DST_COLOR: + /* fall-through */ + case PIPE_BLENDFACTOR_INV_DST_ALPHA: + { + float inv_alpha[4]; + VEC4_SUB(inv_alpha, one, dest[3]); + VEC4_MUL(source[3], quadColor[3], inv_alpha); /* A */ + } + break; + case PIPE_BLENDFACTOR_INV_CONST_COLOR: + /* fall-through */ + case PIPE_BLENDFACTOR_INV_CONST_ALPHA: + { + float inv_comp[4]; + /* A */ + VEC4_SCALAR(inv_comp, 1.0f - softpipe->blend_color.color[3]); + VEC4_MUL(source[3], quadColor[3], inv_comp); + } + break; + default: + assert(0); } - /* pass quad to next stage */ - qs->next->run(qs->next, quad); -} + /* + * Compute dest/second term RGB + */ + switch (softpipe->blend->rgb_dst_factor) { + case PIPE_BLENDFACTOR_ONE: + /* dest = dest * 1 NO-OP, leave dest as-is */ + break; + case PIPE_BLENDFACTOR_SRC_COLOR: + VEC4_MUL(dest[0], dest[0], quadColor[0]); /* R */ + VEC4_MUL(dest[1], dest[1], quadColor[1]); /* G */ + VEC4_MUL(dest[2], dest[2], quadColor[2]); /* B */ + break; + case PIPE_BLENDFACTOR_SRC_ALPHA: + VEC4_MUL(dest[0], dest[0], quadColor[3]); /* R * A */ + VEC4_MUL(dest[1], dest[1], quadColor[3]); /* G * A */ + VEC4_MUL(dest[2], dest[2], quadColor[3]); /* B * A */ + break; + case PIPE_BLENDFACTOR_DST_ALPHA: + VEC4_MUL(dest[0], dest[0], dest[3]); /* R * A */ + VEC4_MUL(dest[1], dest[1], dest[3]); /* G * A */ + VEC4_MUL(dest[2], dest[2], dest[3]); /* B * A */ + break; + case PIPE_BLENDFACTOR_DST_COLOR: + VEC4_MUL(dest[0], dest[0], dest[0]); /* R */ + VEC4_MUL(dest[1], dest[1], dest[1]); /* G */ + VEC4_MUL(dest[2], dest[2], dest[2]); /* B */ + break; + case PIPE_BLENDFACTOR_SRC_ALPHA_SATURATE: + assert(0); /* illegal */ + break; + case PIPE_BLENDFACTOR_CONST_COLOR: + { + float comp[4]; + VEC4_SCALAR(comp, softpipe->blend_color.color[0]); /* R */ + VEC4_MUL(dest[0], dest[0], comp); /* R */ + VEC4_SCALAR(comp, softpipe->blend_color.color[1]); /* G */ + VEC4_MUL(dest[1], dest[1], comp); /* G */ + VEC4_SCALAR(comp, softpipe->blend_color.color[2]); /* B */ + VEC4_MUL(dest[2], dest[2], comp); /* B */ + } + break; + case PIPE_BLENDFACTOR_CONST_ALPHA: + { + float comp[4]; + VEC4_SCALAR(comp, softpipe->blend_color.color[3]); /* A */ + VEC4_MUL(dest[0], dest[0], comp); /* R */ + VEC4_MUL(dest[1], dest[1], comp); /* G */ + VEC4_MUL(dest[2], dest[2], comp); /* B */ + } + break; + case PIPE_BLENDFACTOR_ZERO: + VEC4_COPY(dest[0], zero); /* R */ + VEC4_COPY(dest[1], zero); /* G */ + VEC4_COPY(dest[2], zero); /* B */ + break; + case PIPE_BLENDFACTOR_SRC1_COLOR: + case PIPE_BLENDFACTOR_SRC1_ALPHA: + /* XXX what are these? */ + assert(0); + break; + case PIPE_BLENDFACTOR_INV_SRC_COLOR: + { + float inv_comp[4]; + VEC4_SUB(inv_comp, one, quadColor[0]); /* R */ + VEC4_MUL(dest[0], inv_comp, dest[0]); /* R */ + VEC4_SUB(inv_comp, one, quadColor[1]); /* G */ + VEC4_MUL(dest[1], inv_comp, dest[1]); /* G */ + VEC4_SUB(inv_comp, one, quadColor[2]); /* B */ + VEC4_MUL(dest[2], inv_comp, dest[2]); /* B */ + } + break; + case PIPE_BLENDFACTOR_INV_SRC_ALPHA: + { + float one_minus_alpha[QUAD_SIZE]; + VEC4_SUB(one_minus_alpha, one, quadColor[3]); + VEC4_MUL(dest[0], dest[0], one_minus_alpha); /* R */ + VEC4_MUL(dest[1], dest[1], one_minus_alpha); /* G */ + VEC4_MUL(dest[2], dest[2], one_minus_alpha); /* B */ + } + break; + case PIPE_BLENDFACTOR_INV_DST_ALPHA: + { + float inv_comp[4]; + VEC4_SUB(inv_comp, one, dest[3]); /* A */ + VEC4_MUL(dest[0], inv_comp, dest[0]); /* R */ + VEC4_MUL(dest[1], inv_comp, dest[1]); /* G */ + VEC4_MUL(dest[2], inv_comp, dest[2]); /* B */ + } + break; + case PIPE_BLENDFACTOR_INV_DST_COLOR: + { + float inv_comp[4]; + VEC4_SUB(inv_comp, one, dest[0]); /* R */ + VEC4_MUL(dest[0], dest[0], inv_comp); /* R */ + VEC4_SUB(inv_comp, one, dest[1]); /* G */ + VEC4_MUL(dest[1], dest[1], inv_comp); /* G */ + VEC4_SUB(inv_comp, one, dest[2]); /* B */ + VEC4_MUL(dest[2], dest[2], inv_comp); /* B */ + } + break; + case PIPE_BLENDFACTOR_INV_CONST_COLOR: + { + float inv_comp[4]; + /* R */ + VEC4_SCALAR(inv_comp, 1.0f - softpipe->blend_color.color[0]); + VEC4_MUL(dest[0], dest[0], inv_comp); + /* G */ + VEC4_SCALAR(inv_comp, 1.0f - softpipe->blend_color.color[1]); + VEC4_MUL(dest[1], dest[1], inv_comp); + /* B */ + VEC4_SCALAR(inv_comp, 1.0f - softpipe->blend_color.color[2]); + VEC4_MUL(dest[2], dest[2], inv_comp); + } + break; + case PIPE_BLENDFACTOR_INV_CONST_ALPHA: + { + float inv_comp[4]; + VEC4_SCALAR(inv_comp, 1.0f - softpipe->blend_color.color[3]); + VEC4_MUL(dest[0], dest[0], inv_comp); + VEC4_MUL(dest[1], dest[1], inv_comp); + VEC4_MUL(dest[2], dest[2], inv_comp); + } + break; + case PIPE_BLENDFACTOR_INV_SRC1_COLOR: + case PIPE_BLENDFACTOR_INV_SRC1_ALPHA: + /* XXX what are these? */ + assert(0); + break; + default: + assert(0); + } + /* + * Compute dest/second term A + */ + switch (softpipe->blend->alpha_dst_factor) { + case PIPE_BLENDFACTOR_ONE: + /* dest = dest * 1 NO-OP, leave dest as-is */ + break; + case PIPE_BLENDFACTOR_SRC_COLOR: + /* fall-through */ + case PIPE_BLENDFACTOR_SRC_ALPHA: + VEC4_MUL(dest[3], dest[3], quadColor[3]); /* A * A */ + break; + case PIPE_BLENDFACTOR_DST_COLOR: + /* fall-through */ + case PIPE_BLENDFACTOR_DST_ALPHA: + VEC4_MUL(dest[3], dest[3], dest[3]); /* A */ + break; + case PIPE_BLENDFACTOR_SRC_ALPHA_SATURATE: + assert(0); /* illegal */ + break; + case PIPE_BLENDFACTOR_CONST_COLOR: + /* fall-through */ + case PIPE_BLENDFACTOR_CONST_ALPHA: + { + float comp[4]; + VEC4_SCALAR(comp, softpipe->blend_color.color[3]); /* A */ + VEC4_MUL(dest[3], dest[3], comp); /* A */ + } + break; + case PIPE_BLENDFACTOR_ZERO: + VEC4_COPY(dest[3], zero); /* A */ + break; + case PIPE_BLENDFACTOR_INV_SRC_COLOR: + /* fall-through */ + case PIPE_BLENDFACTOR_INV_SRC_ALPHA: + { + float one_minus_alpha[QUAD_SIZE]; + VEC4_SUB(one_minus_alpha, one, quadColor[3]); + VEC4_MUL(dest[3], dest[3], one_minus_alpha); /* A */ + } + break; + case PIPE_BLENDFACTOR_INV_DST_COLOR: + /* fall-through */ + case PIPE_BLENDFACTOR_INV_DST_ALPHA: + { + float inv_comp[4]; + VEC4_SUB(inv_comp, one, dest[3]); /* A */ + VEC4_MUL(dest[3], inv_comp, dest[3]); /* A */ + } + break; + case PIPE_BLENDFACTOR_INV_CONST_COLOR: + /* fall-through */ + case PIPE_BLENDFACTOR_INV_CONST_ALPHA: + { + float inv_comp[4]; + VEC4_SCALAR(inv_comp, 1.0f - softpipe->blend_color.color[3]); + VEC4_MUL(dest[3], dest[3], inv_comp); + } + break; + default: + assert(0); + } + /* + * Combine RGB terms + */ + switch (softpipe->blend->rgb_func) { + case PIPE_BLEND_ADD: + VEC4_ADD_SAT(quadColor[0], source[0], dest[0]); /* R */ + VEC4_ADD_SAT(quadColor[1], source[1], dest[1]); /* G */ + VEC4_ADD_SAT(quadColor[2], source[2], dest[2]); /* B */ + break; + case PIPE_BLEND_SUBTRACT: + VEC4_SUB_SAT(quadColor[0], source[0], dest[0]); /* R */ + VEC4_SUB_SAT(quadColor[1], source[1], dest[1]); /* G */ + VEC4_SUB_SAT(quadColor[2], source[2], dest[2]); /* B */ + break; + case PIPE_BLEND_REVERSE_SUBTRACT: + VEC4_SUB_SAT(quadColor[0], dest[0], source[0]); /* R */ + VEC4_SUB_SAT(quadColor[1], dest[1], source[1]); /* G */ + VEC4_SUB_SAT(quadColor[2], dest[2], source[2]); /* B */ + break; + case PIPE_BLEND_MIN: + VEC4_MIN(quadColor[0], source[0], dest[0]); /* R */ + VEC4_MIN(quadColor[1], source[1], dest[1]); /* G */ + VEC4_MIN(quadColor[2], source[2], dest[2]); /* B */ + break; + case PIPE_BLEND_MAX: + VEC4_MAX(quadColor[0], source[0], dest[0]); /* R */ + VEC4_MAX(quadColor[1], source[1], dest[1]); /* G */ + VEC4_MAX(quadColor[2], source[2], dest[2]); /* B */ + break; + default: + assert(0); + } + + /* + * Combine A terms + */ + switch (softpipe->blend->alpha_func) { + case PIPE_BLEND_ADD: + VEC4_ADD_SAT(quadColor[3], source[3], dest[3]); /* A */ + break; + case PIPE_BLEND_SUBTRACT: + VEC4_SUB_SAT(quadColor[3], source[3], dest[3]); /* A */ + break; + case PIPE_BLEND_REVERSE_SUBTRACT: + VEC4_SUB_SAT(quadColor[3], dest[3], source[3]); /* A */ + break; + case PIPE_BLEND_MIN: + VEC4_MIN(quadColor[3], source[3], dest[3]); /* A */ + break; + case PIPE_BLEND_MAX: + VEC4_MAX(quadColor[3], source[3], dest[3]); /* A */ + break; + default: + assert(0); + } +} static void -blend_quad(struct quad_stage *qs, struct quad_header *quad) +colormask_quad(struct quad_stage *qs, + float (*quadColor)[4], + float (*dest)[4]) { - static const float zero[4] = { 0, 0, 0, 0 }; - static const float one[4] = { 1, 1, 1, 1 }; + struct softpipe_context *softpipe = qs->softpipe; + /* R */ + if (!(softpipe->blend->colormask & PIPE_MASK_R)) + COPY_4V(quadColor[0], dest[0]); + + /* G */ + if (!(softpipe->blend->colormask & PIPE_MASK_G)) + COPY_4V(quadColor[1], dest[1]); + + /* B */ + if (!(softpipe->blend->colormask & PIPE_MASK_B)) + COPY_4V(quadColor[2], dest[2]); + + /* A */ + if (!(softpipe->blend->colormask & PIPE_MASK_A)) + COPY_4V(quadColor[3], dest[3]); +} + + +static void +blend_fallback(struct quad_stage *qs, + struct quad_header *quads[], + unsigned nr) +{ struct softpipe_context *softpipe = qs->softpipe; - uint cbuf; + const struct pipe_blend_state *blend = softpipe->blend; + unsigned cbuf; + + for (cbuf = 0; cbuf < softpipe->framebuffer.nr_cbufs; cbuf++) + { + float dest[4][QUAD_SIZE]; + struct softpipe_cached_tile *tile + = sp_get_cached_tile(softpipe->cbuf_cache[cbuf], + quads[0]->input.x0, + quads[0]->input.y0); + uint q, i, j; + + for (q = 0; q < nr; q++) { + struct quad_header *quad = quads[q]; + float (*quadColor)[4] = quad->output.color[cbuf]; + const int itx = (quad->input.x0 & (TILE_SIZE-1)); + const int ity = (quad->input.y0 & (TILE_SIZE-1)); + + /* get/swizzle dest colors + */ + for (j = 0; j < QUAD_SIZE; j++) { + int x = itx + (j & 1); + int y = ity + (j >> 1); + for (i = 0; i < 4; i++) { + dest[i][j] = tile->data.color[y][x][i]; + } + } - if (softpipe->blend->logicop_enable) { - logicop_quad(qs, quad); - return; + + if (blend->logicop_enable) { + logicop_quad( qs, quadColor, dest ); + } + else if (blend->blend_enable) { + blend_quad( qs, quadColor, dest ); + } + + if (blend->colormask != 0xf) + colormask_quad( qs, quadColor, dest ); + + /* Output color values + */ + for (j = 0; j < QUAD_SIZE; j++) { + if (quad->inout.mask & (1 << j)) { + int x = itx + (j & 1); + int y = ity + (j >> 1); + for (i = 0; i < 4; i++) { /* loop over color chans */ + tile->data.color[y][x][i] = quadColor[i][j]; + } + } + } + } } +} - /* loop over colorbuffer outputs */ - for (cbuf = 0; cbuf < softpipe->framebuffer.nr_cbufs; cbuf++) { - float source[4][QUAD_SIZE], dest[4][QUAD_SIZE]; - struct softpipe_cached_tile *tile - = sp_get_cached_tile(softpipe, - softpipe->cbuf_cache[cbuf], - quad->input.x0, quad->input.y0); - float (*quadColor)[4] = quad->output.color[cbuf]; - uint i, j; +static void +blend_single_add_src_alpha_inv_src_alpha(struct quad_stage *qs, + struct quad_header *quads[], + unsigned nr) +{ + static const float one[4] = { 1, 1, 1, 1 }; + float one_minus_alpha[QUAD_SIZE]; + float dest[4][QUAD_SIZE]; + float source[4][QUAD_SIZE]; + uint i, j, q; + + struct softpipe_cached_tile *tile + = sp_get_cached_tile(qs->softpipe->cbuf_cache[0], + quads[0]->input.x0, + quads[0]->input.y0); + + for (q = 0; q < nr; q++) { + struct quad_header *quad = quads[q]; + float (*quadColor)[4] = quad->output.color[0]; + const float *alpha = quadColor[3]; + const int itx = (quad->input.x0 & (TILE_SIZE-1)); + const int ity = (quad->input.y0 & (TILE_SIZE-1)); + /* get/swizzle dest colors */ for (j = 0; j < QUAD_SIZE; j++) { - int x = (quad->input.x0 & (TILE_SIZE-1)) + (j & 1); - int y = (quad->input.y0 & (TILE_SIZE-1)) + (j >> 1); + int x = itx + (j & 1); + int y = ity + (j >> 1); for (i = 0; i < 4; i++) { dest[i][j] = tile->data.color[y][x][i]; } } - /* - * Compute src/first term RGB - */ - switch (softpipe->blend->rgb_src_factor) { - case PIPE_BLENDFACTOR_ONE: - VEC4_COPY(source[0], quadColor[0]); /* R */ - VEC4_COPY(source[1], quadColor[1]); /* G */ - VEC4_COPY(source[2], quadColor[2]); /* B */ - break; - case PIPE_BLENDFACTOR_SRC_COLOR: - VEC4_MUL(source[0], quadColor[0], quadColor[0]); /* R */ - VEC4_MUL(source[1], quadColor[1], quadColor[1]); /* G */ - VEC4_MUL(source[2], quadColor[2], quadColor[2]); /* B */ - break; - case PIPE_BLENDFACTOR_SRC_ALPHA: - { - const float *alpha = quadColor[3]; - VEC4_MUL(source[0], quadColor[0], alpha); /* R */ - VEC4_MUL(source[1], quadColor[1], alpha); /* G */ - VEC4_MUL(source[2], quadColor[2], alpha); /* B */ - } - break; - case PIPE_BLENDFACTOR_DST_COLOR: - VEC4_MUL(source[0], quadColor[0], dest[0]); /* R */ - VEC4_MUL(source[1], quadColor[1], dest[1]); /* G */ - VEC4_MUL(source[2], quadColor[2], dest[2]); /* B */ - break; - case PIPE_BLENDFACTOR_DST_ALPHA: - { - const float *alpha = dest[3]; - VEC4_MUL(source[0], quadColor[0], alpha); /* R */ - VEC4_MUL(source[1], quadColor[1], alpha); /* G */ - VEC4_MUL(source[2], quadColor[2], alpha); /* B */ - } - break; - case PIPE_BLENDFACTOR_SRC_ALPHA_SATURATE: - { - const float *alpha = quadColor[3]; - float diff[4], temp[4]; - VEC4_SUB(diff, one, dest[3]); - VEC4_MIN(temp, alpha, diff); - VEC4_MUL(source[0], quadColor[0], temp); /* R */ - VEC4_MUL(source[1], quadColor[1], temp); /* G */ - VEC4_MUL(source[2], quadColor[2], temp); /* B */ - } - break; - case PIPE_BLENDFACTOR_CONST_COLOR: - { - float comp[4]; - VEC4_SCALAR(comp, softpipe->blend_color.color[0]); /* R */ - VEC4_MUL(source[0], quadColor[0], comp); /* R */ - VEC4_SCALAR(comp, softpipe->blend_color.color[1]); /* G */ - VEC4_MUL(source[1], quadColor[1], comp); /* G */ - VEC4_SCALAR(comp, softpipe->blend_color.color[2]); /* B */ - VEC4_MUL(source[2], quadColor[2], comp); /* B */ - } - break; - case PIPE_BLENDFACTOR_CONST_ALPHA: - { - float alpha[4]; - VEC4_SCALAR(alpha, softpipe->blend_color.color[3]); - VEC4_MUL(source[0], quadColor[0], alpha); /* R */ - VEC4_MUL(source[1], quadColor[1], alpha); /* G */ - VEC4_MUL(source[2], quadColor[2], alpha); /* B */ - } - break; - case PIPE_BLENDFACTOR_SRC1_COLOR: - assert(0); /* to do */ - break; - case PIPE_BLENDFACTOR_SRC1_ALPHA: - assert(0); /* to do */ - break; - case PIPE_BLENDFACTOR_ZERO: - VEC4_COPY(source[0], zero); /* R */ - VEC4_COPY(source[1], zero); /* G */ - VEC4_COPY(source[2], zero); /* B */ - break; - case PIPE_BLENDFACTOR_INV_SRC_COLOR: - { - float inv_comp[4]; - VEC4_SUB(inv_comp, one, quadColor[0]); /* R */ - VEC4_MUL(source[0], quadColor[0], inv_comp); /* R */ - VEC4_SUB(inv_comp, one, quadColor[1]); /* G */ - VEC4_MUL(source[1], quadColor[1], inv_comp); /* G */ - VEC4_SUB(inv_comp, one, quadColor[2]); /* B */ - VEC4_MUL(source[2], quadColor[2], inv_comp); /* B */ - } - break; - case PIPE_BLENDFACTOR_INV_SRC_ALPHA: - { - float inv_alpha[4]; - VEC4_SUB(inv_alpha, one, quadColor[3]); - VEC4_MUL(source[0], quadColor[0], inv_alpha); /* R */ - VEC4_MUL(source[1], quadColor[1], inv_alpha); /* G */ - VEC4_MUL(source[2], quadColor[2], inv_alpha); /* B */ - } - break; - case PIPE_BLENDFACTOR_INV_DST_ALPHA: - { - float inv_alpha[4]; - VEC4_SUB(inv_alpha, one, dest[3]); - VEC4_MUL(source[0], quadColor[0], inv_alpha); /* R */ - VEC4_MUL(source[1], quadColor[1], inv_alpha); /* G */ - VEC4_MUL(source[2], quadColor[2], inv_alpha); /* B */ - } - break; - case PIPE_BLENDFACTOR_INV_DST_COLOR: - { - float inv_comp[4]; - VEC4_SUB(inv_comp, one, dest[0]); /* R */ - VEC4_MUL(source[0], quadColor[0], inv_comp); /* R */ - VEC4_SUB(inv_comp, one, dest[1]); /* G */ - VEC4_MUL(source[1], quadColor[1], inv_comp); /* G */ - VEC4_SUB(inv_comp, one, dest[2]); /* B */ - VEC4_MUL(source[2], quadColor[2], inv_comp); /* B */ - } - break; - case PIPE_BLENDFACTOR_INV_CONST_COLOR: - { - float inv_comp[4]; - /* R */ - VEC4_SCALAR(inv_comp, 1.0f - softpipe->blend_color.color[0]); - VEC4_MUL(source[0], quadColor[0], inv_comp); - /* G */ - VEC4_SCALAR(inv_comp, 1.0f - softpipe->blend_color.color[1]); - VEC4_MUL(source[1], quadColor[1], inv_comp); - /* B */ - VEC4_SCALAR(inv_comp, 1.0f - softpipe->blend_color.color[2]); - VEC4_MUL(source[2], quadColor[2], inv_comp); - } - break; - case PIPE_BLENDFACTOR_INV_CONST_ALPHA: - { - float inv_alpha[4]; - VEC4_SCALAR(inv_alpha, 1.0f - softpipe->blend_color.color[3]); - VEC4_MUL(source[0], quadColor[0], inv_alpha); /* R */ - VEC4_MUL(source[1], quadColor[1], inv_alpha); /* G */ - VEC4_MUL(source[2], quadColor[2], inv_alpha); /* B */ + VEC4_MUL(source[0], quadColor[0], alpha); /* R */ + VEC4_MUL(source[1], quadColor[1], alpha); /* G */ + VEC4_MUL(source[2], quadColor[2], alpha); /* B */ + VEC4_MUL(source[3], quadColor[3], alpha); /* A */ + + VEC4_SUB(one_minus_alpha, one, alpha); + VEC4_MUL(dest[0], dest[0], one_minus_alpha); /* R */ + VEC4_MUL(dest[1], dest[1], one_minus_alpha); /* G */ + VEC4_MUL(dest[2], dest[2], one_minus_alpha); /* B */ + VEC4_MUL(dest[3], dest[3], one_minus_alpha); /* B */ + + VEC4_ADD_SAT(quadColor[0], source[0], dest[0]); /* R */ + VEC4_ADD_SAT(quadColor[1], source[1], dest[1]); /* G */ + VEC4_ADD_SAT(quadColor[2], source[2], dest[2]); /* B */ + VEC4_ADD_SAT(quadColor[3], source[3], dest[3]); /* A */ + + for (j = 0; j < QUAD_SIZE; j++) { + if (quad->inout.mask & (1 << j)) { + int x = itx + (j & 1); + int y = ity + (j >> 1); + for (i = 0; i < 4; i++) { /* loop over color chans */ + tile->data.color[y][x][i] = quadColor[i][j]; + } } - break; - case PIPE_BLENDFACTOR_INV_SRC1_COLOR: - assert(0); /* to do */ - break; - case PIPE_BLENDFACTOR_INV_SRC1_ALPHA: - assert(0); /* to do */ - break; - default: - assert(0); } + } +} - /* - * Compute src/first term A - */ - switch (softpipe->blend->alpha_src_factor) { - case PIPE_BLENDFACTOR_ONE: - VEC4_COPY(source[3], quadColor[3]); /* A */ - break; - case PIPE_BLENDFACTOR_SRC_COLOR: - /* fall-through */ - case PIPE_BLENDFACTOR_SRC_ALPHA: - { - const float *alpha = quadColor[3]; - VEC4_MUL(source[3], quadColor[3], alpha); /* A */ - } - break; - case PIPE_BLENDFACTOR_DST_COLOR: - /* fall-through */ - case PIPE_BLENDFACTOR_DST_ALPHA: - VEC4_MUL(source[3], quadColor[3], dest[3]); /* A */ - break; - case PIPE_BLENDFACTOR_SRC_ALPHA_SATURATE: - /* multiply alpha by 1.0 */ - VEC4_COPY(source[3], quadColor[3]); /* A */ - break; - case PIPE_BLENDFACTOR_CONST_COLOR: - /* fall-through */ - case PIPE_BLENDFACTOR_CONST_ALPHA: - { - float comp[4]; - VEC4_SCALAR(comp, softpipe->blend_color.color[3]); /* A */ - VEC4_MUL(source[3], quadColor[3], comp); /* A */ - } - break; - case PIPE_BLENDFACTOR_ZERO: - VEC4_COPY(source[3], zero); /* A */ - break; - case PIPE_BLENDFACTOR_INV_SRC_COLOR: - /* fall-through */ - case PIPE_BLENDFACTOR_INV_SRC_ALPHA: - { - float inv_alpha[4]; - VEC4_SUB(inv_alpha, one, quadColor[3]); - VEC4_MUL(source[3], quadColor[3], inv_alpha); /* A */ - } - break; - case PIPE_BLENDFACTOR_INV_DST_COLOR: - /* fall-through */ - case PIPE_BLENDFACTOR_INV_DST_ALPHA: - { - float inv_alpha[4]; - VEC4_SUB(inv_alpha, one, dest[3]); - VEC4_MUL(source[3], quadColor[3], inv_alpha); /* A */ - } - break; - case PIPE_BLENDFACTOR_INV_CONST_COLOR: - /* fall-through */ - case PIPE_BLENDFACTOR_INV_CONST_ALPHA: - { - float inv_comp[4]; - /* A */ - VEC4_SCALAR(inv_comp, 1.0f - softpipe->blend_color.color[3]); - VEC4_MUL(source[3], quadColor[3], inv_comp); +static void +blend_single_add_one_one(struct quad_stage *qs, + struct quad_header *quads[], + unsigned nr) +{ + float dest[4][QUAD_SIZE]; + uint i, j, q; + + struct softpipe_cached_tile *tile + = sp_get_cached_tile(qs->softpipe->cbuf_cache[0], + quads[0]->input.x0, + quads[0]->input.y0); + + for (q = 0; q < nr; q++) { + struct quad_header *quad = quads[q]; + float (*quadColor)[4] = quad->output.color[0]; + const int itx = (quad->input.x0 & (TILE_SIZE-1)); + const int ity = (quad->input.y0 & (TILE_SIZE-1)); + + /* get/swizzle dest colors */ + for (j = 0; j < QUAD_SIZE; j++) { + int x = itx + (j & 1); + int y = ity + (j >> 1); + for (i = 0; i < 4; i++) { + dest[i][j] = tile->data.color[y][x][i]; } - break; - default: - assert(0); } + + VEC4_ADD_SAT(quadColor[0], quadColor[0], dest[0]); /* R */ + VEC4_ADD_SAT(quadColor[1], quadColor[1], dest[1]); /* G */ + VEC4_ADD_SAT(quadColor[2], quadColor[2], dest[2]); /* B */ + VEC4_ADD_SAT(quadColor[3], quadColor[3], dest[3]); /* A */ - - /* - * Compute dest/second term RGB - */ - switch (softpipe->blend->rgb_dst_factor) { - case PIPE_BLENDFACTOR_ONE: - /* dest = dest * 1 NO-OP, leave dest as-is */ - break; - case PIPE_BLENDFACTOR_SRC_COLOR: - VEC4_MUL(dest[0], dest[0], quadColor[0]); /* R */ - VEC4_MUL(dest[1], dest[1], quadColor[1]); /* G */ - VEC4_MUL(dest[2], dest[2], quadColor[2]); /* B */ - break; - case PIPE_BLENDFACTOR_SRC_ALPHA: - VEC4_MUL(dest[0], dest[0], quadColor[3]); /* R * A */ - VEC4_MUL(dest[1], dest[1], quadColor[3]); /* G * A */ - VEC4_MUL(dest[2], dest[2], quadColor[3]); /* B * A */ - break; - case PIPE_BLENDFACTOR_DST_ALPHA: - VEC4_MUL(dest[0], dest[0], dest[3]); /* R * A */ - VEC4_MUL(dest[1], dest[1], dest[3]); /* G * A */ - VEC4_MUL(dest[2], dest[2], dest[3]); /* B * A */ - break; - case PIPE_BLENDFACTOR_DST_COLOR: - VEC4_MUL(dest[0], dest[0], dest[0]); /* R */ - VEC4_MUL(dest[1], dest[1], dest[1]); /* G */ - VEC4_MUL(dest[2], dest[2], dest[2]); /* B */ - break; - case PIPE_BLENDFACTOR_SRC_ALPHA_SATURATE: - assert(0); /* illegal */ - break; - case PIPE_BLENDFACTOR_CONST_COLOR: - { - float comp[4]; - VEC4_SCALAR(comp, softpipe->blend_color.color[0]); /* R */ - VEC4_MUL(dest[0], dest[0], comp); /* R */ - VEC4_SCALAR(comp, softpipe->blend_color.color[1]); /* G */ - VEC4_MUL(dest[1], dest[1], comp); /* G */ - VEC4_SCALAR(comp, softpipe->blend_color.color[2]); /* B */ - VEC4_MUL(dest[2], dest[2], comp); /* B */ - } - break; - case PIPE_BLENDFACTOR_CONST_ALPHA: - { - float comp[4]; - VEC4_SCALAR(comp, softpipe->blend_color.color[3]); /* A */ - VEC4_MUL(dest[0], dest[0], comp); /* R */ - VEC4_MUL(dest[1], dest[1], comp); /* G */ - VEC4_MUL(dest[2], dest[2], comp); /* B */ - } - break; - case PIPE_BLENDFACTOR_ZERO: - VEC4_COPY(dest[0], zero); /* R */ - VEC4_COPY(dest[1], zero); /* G */ - VEC4_COPY(dest[2], zero); /* B */ - break; - case PIPE_BLENDFACTOR_SRC1_COLOR: - case PIPE_BLENDFACTOR_SRC1_ALPHA: - /* XXX what are these? */ - assert(0); - break; - case PIPE_BLENDFACTOR_INV_SRC_COLOR: - { - float inv_comp[4]; - VEC4_SUB(inv_comp, one, quadColor[0]); /* R */ - VEC4_MUL(dest[0], inv_comp, dest[0]); /* R */ - VEC4_SUB(inv_comp, one, quadColor[1]); /* G */ - VEC4_MUL(dest[1], inv_comp, dest[1]); /* G */ - VEC4_SUB(inv_comp, one, quadColor[2]); /* B */ - VEC4_MUL(dest[2], inv_comp, dest[2]); /* B */ - } - break; - case PIPE_BLENDFACTOR_INV_SRC_ALPHA: - { - float one_minus_alpha[QUAD_SIZE]; - VEC4_SUB(one_minus_alpha, one, quadColor[3]); - VEC4_MUL(dest[0], dest[0], one_minus_alpha); /* R */ - VEC4_MUL(dest[1], dest[1], one_minus_alpha); /* G */ - VEC4_MUL(dest[2], dest[2], one_minus_alpha); /* B */ - } - break; - case PIPE_BLENDFACTOR_INV_DST_ALPHA: - { - float inv_comp[4]; - VEC4_SUB(inv_comp, one, dest[3]); /* A */ - VEC4_MUL(dest[0], inv_comp, dest[0]); /* R */ - VEC4_MUL(dest[1], inv_comp, dest[1]); /* G */ - VEC4_MUL(dest[2], inv_comp, dest[2]); /* B */ - } - break; - case PIPE_BLENDFACTOR_INV_DST_COLOR: - { - float inv_comp[4]; - VEC4_SUB(inv_comp, one, dest[0]); /* R */ - VEC4_MUL(dest[0], dest[0], inv_comp); /* R */ - VEC4_SUB(inv_comp, one, dest[1]); /* G */ - VEC4_MUL(dest[1], dest[1], inv_comp); /* G */ - VEC4_SUB(inv_comp, one, dest[2]); /* B */ - VEC4_MUL(dest[2], dest[2], inv_comp); /* B */ - } - break; - case PIPE_BLENDFACTOR_INV_CONST_COLOR: - { - float inv_comp[4]; - /* R */ - VEC4_SCALAR(inv_comp, 1.0f - softpipe->blend_color.color[0]); - VEC4_MUL(dest[0], dest[0], inv_comp); - /* G */ - VEC4_SCALAR(inv_comp, 1.0f - softpipe->blend_color.color[1]); - VEC4_MUL(dest[1], dest[1], inv_comp); - /* B */ - VEC4_SCALAR(inv_comp, 1.0f - softpipe->blend_color.color[2]); - VEC4_MUL(dest[2], dest[2], inv_comp); - } - break; - case PIPE_BLENDFACTOR_INV_CONST_ALPHA: - { - float inv_comp[4]; - VEC4_SCALAR(inv_comp, 1.0f - softpipe->blend_color.color[3]); - VEC4_MUL(dest[0], dest[0], inv_comp); - VEC4_MUL(dest[1], dest[1], inv_comp); - VEC4_MUL(dest[2], dest[2], inv_comp); + for (j = 0; j < QUAD_SIZE; j++) { + if (quad->inout.mask & (1 << j)) { + int x = itx + (j & 1); + int y = ity + (j >> 1); + for (i = 0; i < 4; i++) { /* loop over color chans */ + tile->data.color[y][x][i] = quadColor[i][j]; + } } - break; - case PIPE_BLENDFACTOR_INV_SRC1_COLOR: - case PIPE_BLENDFACTOR_INV_SRC1_ALPHA: - /* XXX what are these? */ - assert(0); - break; - default: - assert(0); } + } +} - /* - * Compute dest/second term A - */ - switch (softpipe->blend->alpha_dst_factor) { - case PIPE_BLENDFACTOR_ONE: - /* dest = dest * 1 NO-OP, leave dest as-is */ - break; - case PIPE_BLENDFACTOR_SRC_COLOR: - /* fall-through */ - case PIPE_BLENDFACTOR_SRC_ALPHA: - VEC4_MUL(dest[3], dest[3], quadColor[3]); /* A * A */ - break; - case PIPE_BLENDFACTOR_DST_COLOR: - /* fall-through */ - case PIPE_BLENDFACTOR_DST_ALPHA: - VEC4_MUL(dest[3], dest[3], dest[3]); /* A */ - break; - case PIPE_BLENDFACTOR_SRC_ALPHA_SATURATE: - assert(0); /* illegal */ - break; - case PIPE_BLENDFACTOR_CONST_COLOR: - /* fall-through */ - case PIPE_BLENDFACTOR_CONST_ALPHA: - { - float comp[4]; - VEC4_SCALAR(comp, softpipe->blend_color.color[3]); /* A */ - VEC4_MUL(dest[3], dest[3], comp); /* A */ - } - break; - case PIPE_BLENDFACTOR_ZERO: - VEC4_COPY(dest[3], zero); /* A */ - break; - case PIPE_BLENDFACTOR_INV_SRC_COLOR: - /* fall-through */ - case PIPE_BLENDFACTOR_INV_SRC_ALPHA: - { - float one_minus_alpha[QUAD_SIZE]; - VEC4_SUB(one_minus_alpha, one, quadColor[3]); - VEC4_MUL(dest[3], dest[3], one_minus_alpha); /* A */ - } - break; - case PIPE_BLENDFACTOR_INV_DST_COLOR: - /* fall-through */ - case PIPE_BLENDFACTOR_INV_DST_ALPHA: - { - float inv_comp[4]; - VEC4_SUB(inv_comp, one, dest[3]); /* A */ - VEC4_MUL(dest[3], inv_comp, dest[3]); /* A */ - } - break; - case PIPE_BLENDFACTOR_INV_CONST_COLOR: - /* fall-through */ - case PIPE_BLENDFACTOR_INV_CONST_ALPHA: - { - float inv_comp[4]; - VEC4_SCALAR(inv_comp, 1.0f - softpipe->blend_color.color[3]); - VEC4_MUL(dest[3], dest[3], inv_comp); + +static void +single_output_color(struct quad_stage *qs, + struct quad_header *quads[], + unsigned nr) +{ + uint i, j, q; + + struct softpipe_cached_tile *tile + = sp_get_cached_tile(qs->softpipe->cbuf_cache[0], + quads[0]->input.x0, + quads[0]->input.y0); + + for (q = 0; q < nr; q++) { + struct quad_header *quad = quads[q]; + float (*quadColor)[4] = quad->output.color[0]; + const int itx = (quad->input.x0 & (TILE_SIZE-1)); + const int ity = (quad->input.y0 & (TILE_SIZE-1)); + + for (j = 0; j < QUAD_SIZE; j++) { + if (quad->inout.mask & (1 << j)) { + int x = itx + (j & 1); + int y = ity + (j >> 1); + for (i = 0; i < 4; i++) { /* loop over color chans */ + tile->data.color[y][x][i] = quadColor[i][j]; + } } - break; - default: - assert(0); } + } +} + +static void +blend_noop(struct quad_stage *qs, + struct quad_header *quads[], + unsigned nr) +{ +} - /* - * Combine RGB terms - */ - switch (softpipe->blend->rgb_func) { - case PIPE_BLEND_ADD: - VEC4_ADD_SAT(quadColor[0], source[0], dest[0]); /* R */ - VEC4_ADD_SAT(quadColor[1], source[1], dest[1]); /* G */ - VEC4_ADD_SAT(quadColor[2], source[2], dest[2]); /* B */ - break; - case PIPE_BLEND_SUBTRACT: - VEC4_SUB_SAT(quadColor[0], source[0], dest[0]); /* R */ - VEC4_SUB_SAT(quadColor[1], source[1], dest[1]); /* G */ - VEC4_SUB_SAT(quadColor[2], source[2], dest[2]); /* B */ - break; - case PIPE_BLEND_REVERSE_SUBTRACT: - VEC4_SUB_SAT(quadColor[0], dest[0], source[0]); /* R */ - VEC4_SUB_SAT(quadColor[1], dest[1], source[1]); /* G */ - VEC4_SUB_SAT(quadColor[2], dest[2], source[2]); /* B */ - break; - case PIPE_BLEND_MIN: - VEC4_MIN(quadColor[0], source[0], dest[0]); /* R */ - VEC4_MIN(quadColor[1], source[1], dest[1]); /* G */ - VEC4_MIN(quadColor[2], source[2], dest[2]); /* B */ - break; - case PIPE_BLEND_MAX: - VEC4_MAX(quadColor[0], source[0], dest[0]); /* R */ - VEC4_MAX(quadColor[1], source[1], dest[1]); /* G */ - VEC4_MAX(quadColor[2], source[2], dest[2]); /* B */ - break; - default: - assert(0); - } - /* - * Combine A terms - */ - switch (softpipe->blend->alpha_func) { - case PIPE_BLEND_ADD: - VEC4_ADD_SAT(quadColor[3], source[3], dest[3]); /* A */ - break; - case PIPE_BLEND_SUBTRACT: - VEC4_SUB_SAT(quadColor[3], source[3], dest[3]); /* A */ - break; - case PIPE_BLEND_REVERSE_SUBTRACT: - VEC4_SUB_SAT(quadColor[3], dest[3], source[3]); /* A */ - break; - case PIPE_BLEND_MIN: - VEC4_MIN(quadColor[3], source[3], dest[3]); /* A */ - break; - case PIPE_BLEND_MAX: - VEC4_MAX(quadColor[3], source[3], dest[3]); /* A */ - break; - default: - assert(0); +static void +choose_blend_quad(struct quad_stage *qs, + struct quad_header *quads[], + unsigned nr) +{ + struct softpipe_context *softpipe = qs->softpipe; + const struct pipe_blend_state *blend = softpipe->blend; + + qs->run = blend_fallback; + + if (softpipe->framebuffer.nr_cbufs == 0) { + qs->run = blend_noop; + } + else if (!softpipe->blend->logicop_enable && + softpipe->blend->colormask == 0xf) + { + if (!blend->blend_enable) { + qs->run = single_output_color; } + else if (blend->rgb_src_factor == blend->alpha_src_factor && + blend->rgb_dst_factor == blend->alpha_dst_factor && + blend->rgb_func == blend->alpha_func && + softpipe->framebuffer.nr_cbufs == 1) + { + if (blend->alpha_func == PIPE_BLEND_ADD) { + if (blend->rgb_src_factor == PIPE_BLENDFACTOR_ONE && + blend->rgb_dst_factor == PIPE_BLENDFACTOR_ONE) { + qs->run = blend_single_add_one_one; + } + else if (blend->rgb_src_factor == PIPE_BLENDFACTOR_SRC_ALPHA && + blend->rgb_dst_factor == PIPE_BLENDFACTOR_INV_SRC_ALPHA) + qs->run = blend_single_add_src_alpha_inv_src_alpha; - } /* cbuf loop */ + } + } + } - /* pass blended quad to next stage */ - qs->next->run(qs->next, quad); + qs->run(qs, quads, nr); } static void blend_begin(struct quad_stage *qs) { - qs->next->begin(qs->next); + qs->run = choose_blend_quad; } @@ -770,7 +991,7 @@ struct quad_stage *sp_quad_blend_stage( struct softpipe_context *softpipe ) stage->softpipe = softpipe; stage->begin = blend_begin; - stage->run = blend_quad; + stage->run = choose_blend_quad; stage->destroy = blend_destroy; return stage; diff --git a/src/gallium/drivers/softpipe/sp_quad_bufloop.c b/src/gallium/drivers/softpipe/sp_quad_bufloop.c deleted file mode 100644 index 953d8516b90..00000000000 --- a/src/gallium/drivers/softpipe/sp_quad_bufloop.c +++ /dev/null @@ -1,74 +0,0 @@ - -#include "util/u_memory.h" -#include "sp_context.h" -#include "sp_quad.h" -#include "sp_surface.h" -#include "sp_quad_pipe.h" - - -/** - * Loop over colorbuffers, passing quad to next stage each time. - */ -static void -cbuf_loop_quad(struct quad_stage *qs, struct quad_header *quad) -{ - struct softpipe_context *softpipe = qs->softpipe; - float tmp[PIPE_MAX_COLOR_BUFS][4][QUAD_SIZE]; - unsigned i; - - assert(sizeof(quad->outputs.color) == sizeof(tmp)); - assert(softpipe->framebuffer.nr_cbufs <= PIPE_MAX_COLOR_BUFS); - - /* make copy of original colors since they can get modified - * by blending and masking. - * XXX we won't have to do this if the fragment program actually emits - * N separate colors and we're drawing to N color buffers (MRT). - * But if we emitted one color and glDrawBuffer(GL_FRONT_AND_BACK) is - * in effect, we need to save/restore colors like this. - */ - memcpy(tmp, quad->outputs.color, sizeof(tmp)); - - for (i = 0; i < softpipe->framebuffer.nr_cbufs; i++) { - /* set current cbuffer */ -#if 0 /* obsolete & going away */ - softpipe->current_cbuf = i; -#endif - - /* pass blended quad to next stage */ - qs->next->run(qs->next, quad); - - /* restore quad's colors for next buffer */ - memcpy(quad->outputs.color, tmp, sizeof(tmp)); - } -} - - -static void cbuf_loop_begin(struct quad_stage *qs) -{ - qs->next->begin(qs->next); -} - - -static void cbuf_loop_destroy(struct quad_stage *qs) -{ - FREE( qs ); -} - - -/** - * Create the colorbuffer loop stage. - * This is used to implement multiple render targets and GL_FRONT_AND_BACK - * rendering. - */ -struct quad_stage *sp_quad_bufloop_stage( struct softpipe_context *softpipe ) -{ - struct quad_stage *stage = CALLOC_STRUCT(quad_stage); - - stage->softpipe = softpipe; - stage->begin = cbuf_loop_begin; - stage->run = cbuf_loop_quad; - stage->destroy = cbuf_loop_destroy; - - return stage; -} - diff --git a/src/gallium/drivers/softpipe/sp_quad_colormask.c b/src/gallium/drivers/softpipe/sp_quad_colormask.c deleted file mode 100644 index dc90e5d5e99..00000000000 --- a/src/gallium/drivers/softpipe/sp_quad_colormask.c +++ /dev/null @@ -1,116 +0,0 @@ -/************************************************************************** - * - * Copyright 2007 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. - * - **************************************************************************/ - -/** - * \brief quad colormask stage - * \author Brian Paul - */ - -#include "pipe/p_defines.h" -#include "util/u_math.h" -#include "util/u_memory.h" -#include "sp_context.h" -#include "sp_quad.h" -#include "sp_surface.h" -#include "sp_quad_pipe.h" -#include "sp_tile_cache.h" - - - -/** - * XXX colormask could be rolled into blending... - */ -static void -colormask_quad(struct quad_stage *qs, struct quad_header *quad) -{ - struct softpipe_context *softpipe = qs->softpipe; - uint cbuf; - - /* loop over colorbuffer outputs */ - for (cbuf = 0; cbuf < softpipe->framebuffer.nr_cbufs; cbuf++) { - float dest[4][QUAD_SIZE]; - struct softpipe_cached_tile *tile - = sp_get_cached_tile(softpipe, - softpipe->cbuf_cache[cbuf], - quad->input.x0, quad->input.y0); - float (*quadColor)[4] = quad->output.color[cbuf]; - uint i, j; - - /* get/swizzle dest colors */ - for (j = 0; j < QUAD_SIZE; j++) { - int x = (quad->input.x0 & (TILE_SIZE-1)) + (j & 1); - int y = (quad->input.y0 & (TILE_SIZE-1)) + (j >> 1); - for (i = 0; i < 4; i++) { - dest[i][j] = tile->data.color[y][x][i]; - } - } - - /* R */ - if (!(softpipe->blend->colormask & PIPE_MASK_R)) - COPY_4V(quadColor[0], dest[0]); - - /* G */ - if (!(softpipe->blend->colormask & PIPE_MASK_G)) - COPY_4V(quadColor[1], dest[1]); - - /* B */ - if (!(softpipe->blend->colormask & PIPE_MASK_B)) - COPY_4V(quadColor[2], dest[2]); - - /* A */ - if (!(softpipe->blend->colormask & PIPE_MASK_A)) - COPY_4V(quadColor[3], dest[3]); - } - - /* pass quad to next stage */ - qs->next->run(qs->next, quad); -} - - -static void colormask_begin(struct quad_stage *qs) -{ - qs->next->begin(qs->next); -} - - -static void colormask_destroy(struct quad_stage *qs) -{ - FREE( qs ); -} - - -struct quad_stage *sp_quad_colormask_stage( struct softpipe_context *softpipe ) -{ - struct quad_stage *stage = CALLOC_STRUCT(quad_stage); - - stage->softpipe = softpipe; - stage->begin = colormask_begin; - stage->run = colormask_quad; - stage->destroy = colormask_destroy; - - return stage; -} diff --git a/src/gallium/drivers/softpipe/sp_quad_coverage.c b/src/gallium/drivers/softpipe/sp_quad_coverage.c deleted file mode 100644 index 4aeee858705..00000000000 --- a/src/gallium/drivers/softpipe/sp_quad_coverage.c +++ /dev/null @@ -1,94 +0,0 @@ -/************************************************************************** - * - * Copyright 2007 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. - * - **************************************************************************/ - - -/** - * \brief Apply AA coverage to quad alpha valus - * \author Brian Paul - */ - - -#include "pipe/p_defines.h" -#include "util/u_memory.h" -#include "sp_context.h" -#include "sp_quad.h" -#include "sp_quad_pipe.h" - - -/** - * Multiply quad's alpha values by the fragment coverage. - */ -static void -coverage_quad(struct quad_stage *qs, struct quad_header *quad) -{ - struct softpipe_context *softpipe = qs->softpipe; - const uint prim = quad->input.prim; - - if ((softpipe->rasterizer->poly_smooth && prim == QUAD_PRIM_TRI) || - (softpipe->rasterizer->line_smooth && prim == QUAD_PRIM_LINE) || - (softpipe->rasterizer->point_smooth && prim == QUAD_PRIM_POINT)) { - uint cbuf; - - /* loop over colorbuffer outputs */ - for (cbuf = 0; cbuf < softpipe->framebuffer.nr_cbufs; cbuf++) { - float (*quadColor)[4] = quad->output.color[cbuf]; - unsigned j; - for (j = 0; j < QUAD_SIZE; j++) { - assert(quad->input.coverage[j] >= 0.0); - assert(quad->input.coverage[j] <= 1.0); - quadColor[3][j] *= quad->input.coverage[j]; - } - } - } - - qs->next->run(qs->next, quad); -} - - -static void coverage_begin(struct quad_stage *qs) -{ - qs->next->begin(qs->next); -} - - -static void coverage_destroy(struct quad_stage *qs) -{ - FREE( qs ); -} - - -struct quad_stage *sp_quad_coverage_stage( struct softpipe_context *softpipe ) -{ - struct quad_stage *stage = CALLOC_STRUCT(quad_stage); - - stage->softpipe = softpipe; - stage->begin = coverage_begin; - stage->run = coverage_quad; - stage->destroy = coverage_destroy; - - return stage; -} diff --git a/src/gallium/drivers/softpipe/sp_quad_depth_test.c b/src/gallium/drivers/softpipe/sp_quad_depth_test.c index d463930bae1..0ca86c4e1cb 100644 --- a/src/gallium/drivers/softpipe/sp_quad_depth_test.c +++ b/src/gallium/drivers/softpipe/sp_quad_depth_test.c @@ -31,61 +31,109 @@ #include "pipe/p_defines.h" #include "util/u_memory.h" +#include "tgsi/tgsi_scan.h" #include "sp_context.h" #include "sp_quad.h" #include "sp_surface.h" #include "sp_quad_pipe.h" #include "sp_tile_cache.h" +#include "sp_state.h" /* for sp_fragment_shader */ -/** - * Do depth testing for a quad. - * Not static since it's used by the stencil code. - */ +struct depth_data { + struct pipe_surface *ps; + enum pipe_format format; + unsigned bzzzz[QUAD_SIZE]; /**< Z values fetched from depth buffer */ + unsigned qzzzz[QUAD_SIZE]; /**< Z values from the quad */ + ubyte stencilVals[QUAD_SIZE]; + struct softpipe_cached_tile *tile; +}; -/* - * To increase efficiency, we should probably have multiple versions - * of this function that are specifically for Z16, Z32 and FP Z buffers. - * Try to effectively do that with codegen... - */ -void -sp_depth_test_quad(struct quad_stage *qs, struct quad_header *quad) + +static void +get_depth_stencil_values( struct depth_data *data, + const struct quad_header *quad ) { - struct softpipe_context *softpipe = qs->softpipe; - struct pipe_surface *ps = softpipe->framebuffer.zsbuf; - const enum pipe_format format = ps->format; - unsigned bzzzz[QUAD_SIZE]; /**< Z values fetched from depth buffer */ - unsigned qzzzz[QUAD_SIZE]; /**< Z values from the quad */ - unsigned zmask = 0; unsigned j; - struct softpipe_cached_tile *tile - = sp_get_cached_tile(softpipe, softpipe->zsbuf_cache, quad->input.x0, quad->input.y0); + const struct softpipe_cached_tile *tile = data->tile; + + switch (data->format) { + case PIPE_FORMAT_Z16_UNORM: + for (j = 0; j < QUAD_SIZE; j++) { + int x = quad->input.x0 % TILE_SIZE + (j & 1); + int y = quad->input.y0 % TILE_SIZE + (j >> 1); + data->bzzzz[j] = tile->data.depth16[y][x]; + } + break; + case PIPE_FORMAT_Z32_UNORM: + for (j = 0; j < QUAD_SIZE; j++) { + int x = quad->input.x0 % TILE_SIZE + (j & 1); + int y = quad->input.y0 % TILE_SIZE + (j >> 1); + data->bzzzz[j] = tile->data.depth32[y][x]; + } + break; + case PIPE_FORMAT_X8Z24_UNORM: + case PIPE_FORMAT_S8Z24_UNORM: + for (j = 0; j < QUAD_SIZE; j++) { + int x = quad->input.x0 % TILE_SIZE + (j & 1); + int y = quad->input.y0 % TILE_SIZE + (j >> 1); + data->bzzzz[j] = tile->data.depth32[y][x] & 0xffffff; + data->stencilVals[j] = tile->data.depth32[y][x] >> 24; + } + break; + case PIPE_FORMAT_Z24X8_UNORM: + case PIPE_FORMAT_Z24S8_UNORM: + for (j = 0; j < QUAD_SIZE; j++) { + int x = quad->input.x0 % TILE_SIZE + (j & 1); + int y = quad->input.y0 % TILE_SIZE + (j >> 1); + data->bzzzz[j] = tile->data.depth32[y][x] >> 8; + data->stencilVals[j] = tile->data.depth32[y][x] & 0xff; + } + break; + default: + assert(0); + } +} - assert(ps); /* shouldn't get here if there's no zbuffer */ +/* If the shader has not been run, interpolate the depth values + * ourselves. + */ +static void +interpolate_quad_depth( struct quad_header *quad ) +{ + const float fx = (float) quad->input.x0; + const float fy = (float) quad->input.y0; + const float dzdx = quad->posCoef->dadx[2]; + const float dzdy = quad->posCoef->dady[2]; + const float z0 = quad->posCoef->a0[2] + dzdx * fx + dzdy * fy; - /* - * Convert quad's float depth values to int depth values (qzzzz). + quad->output.depth[0] = z0; + quad->output.depth[1] = z0 + dzdx; + quad->output.depth[2] = z0 + dzdy; + quad->output.depth[3] = z0 + dzdx + dzdy; +} + + +static void +convert_quad_depth( struct depth_data *data, + const struct quad_header *quad ) +{ + unsigned j; + + /* Convert quad's float depth values to int depth values (qzzzz). * If the Z buffer stores integer values, we _have_ to do the depth * compares with integers (not floats). Otherwise, the float->int->float * conversion of Z values (which isn't an identity function) will cause * Z-fighting errors. - * - * Also, get the zbuffer values (bzzzz) from the cached tile. */ - switch (format) { + switch (data->format) { case PIPE_FORMAT_Z16_UNORM: { float scale = 65535.0; for (j = 0; j < QUAD_SIZE; j++) { - qzzzz[j] = (unsigned) (quad->output.depth[j] * scale); - } - - for (j = 0; j < QUAD_SIZE; j++) { - int x = quad->input.x0 % TILE_SIZE + (j & 1); - int y = quad->input.y0 % TILE_SIZE + (j >> 1); - bzzzz[j] = tile->data.depth16[y][x]; + data->qzzzz[j] = (unsigned) (quad->output.depth[j] * scale); } } break; @@ -94,47 +142,247 @@ sp_depth_test_quad(struct quad_stage *qs, struct quad_header *quad) double scale = (double) (uint) ~0UL; for (j = 0; j < QUAD_SIZE; j++) { - qzzzz[j] = (unsigned) (quad->output.depth[j] * scale); - } - - for (j = 0; j < QUAD_SIZE; j++) { - int x = quad->input.x0 % TILE_SIZE + (j & 1); - int y = quad->input.y0 % TILE_SIZE + (j >> 1); - bzzzz[j] = tile->data.depth32[y][x]; + data->qzzzz[j] = (unsigned) (quad->output.depth[j] * scale); } } break; case PIPE_FORMAT_X8Z24_UNORM: - /* fall-through */ case PIPE_FORMAT_S8Z24_UNORM: { float scale = (float) ((1 << 24) - 1); for (j = 0; j < QUAD_SIZE; j++) { - qzzzz[j] = (unsigned) (quad->output.depth[j] * scale); - } - - for (j = 0; j < QUAD_SIZE; j++) { - int x = quad->input.x0 % TILE_SIZE + (j & 1); - int y = quad->input.y0 % TILE_SIZE + (j >> 1); - bzzzz[j] = tile->data.depth32[y][x] & 0xffffff; + data->qzzzz[j] = (unsigned) (quad->output.depth[j] * scale); } } break; case PIPE_FORMAT_Z24X8_UNORM: - /* fall-through */ case PIPE_FORMAT_Z24S8_UNORM: { float scale = (float) ((1 << 24) - 1); for (j = 0; j < QUAD_SIZE; j++) { - qzzzz[j] = (unsigned) (quad->output.depth[j] * scale); + data->qzzzz[j] = (unsigned) (quad->output.depth[j] * scale); } + } + break; + default: + assert(0); + } +} - for (j = 0; j < QUAD_SIZE; j++) { - int x = quad->input.x0 % TILE_SIZE + (j & 1); - int y = quad->input.y0 % TILE_SIZE + (j >> 1); - bzzzz[j] = tile->data.depth32[y][x] >> 8; + + +static void +write_depth_stencil_values( struct depth_data *data, + struct quad_header *quad ) +{ + struct softpipe_cached_tile *tile = data->tile; + unsigned j; + + /* put updated Z values back into cached tile */ + switch (data->format) { + case PIPE_FORMAT_Z16_UNORM: + for (j = 0; j < QUAD_SIZE; j++) { + int x = quad->input.x0 % TILE_SIZE + (j & 1); + int y = quad->input.y0 % TILE_SIZE + (j >> 1); + tile->data.depth16[y][x] = (ushort) data->bzzzz[j]; + } + break; + case PIPE_FORMAT_X8Z24_UNORM: + case PIPE_FORMAT_Z32_UNORM: + for (j = 0; j < QUAD_SIZE; j++) { + int x = quad->input.x0 % TILE_SIZE + (j & 1); + int y = quad->input.y0 % TILE_SIZE + (j >> 1); + tile->data.depth32[y][x] = data->bzzzz[j]; + } + break; + case PIPE_FORMAT_S8Z24_UNORM: + for (j = 0; j < QUAD_SIZE; j++) { + int x = quad->input.x0 % TILE_SIZE + (j & 1); + int y = quad->input.y0 % TILE_SIZE + (j >> 1); + tile->data.depth32[y][x] = (data->stencilVals[j] << 24) | data->bzzzz[j]; + } + break; + case PIPE_FORMAT_Z24S8_UNORM: + for (j = 0; j < QUAD_SIZE; j++) { + int x = quad->input.x0 % TILE_SIZE + (j & 1); + int y = quad->input.y0 % TILE_SIZE + (j >> 1); + tile->data.depth32[y][x] = (data->bzzzz[j] << 8) | data->stencilVals[j]; + } + break; + case PIPE_FORMAT_Z24X8_UNORM: + for (j = 0; j < QUAD_SIZE; j++) { + int x = quad->input.x0 % TILE_SIZE + (j & 1); + int y = quad->input.y0 % TILE_SIZE + (j >> 1); + tile->data.depth32[y][x] = data->bzzzz[j] << 8; + } + break; + default: + assert(0); + } +} + + + + +/** Only 8-bit stencil supported */ +#define STENCIL_MAX 0xff + + +/** + * Do the basic stencil test (compare stencil buffer values against the + * reference value. + * + * \param data->stencilVals the stencil values from the stencil buffer + * \param func the stencil func (PIPE_FUNC_x) + * \param ref the stencil reference value + * \param valMask the stencil value mask indicating which bits of the stencil + * values and ref value are to be used. + * \return mask indicating which pixels passed the stencil test + */ +static unsigned +do_stencil_test(struct depth_data *data, + unsigned func, + unsigned ref, unsigned valMask) +{ + unsigned passMask = 0x0; + unsigned j; + + ref &= valMask; + + switch (func) { + case PIPE_FUNC_NEVER: + /* passMask = 0x0 */ + break; + case PIPE_FUNC_LESS: + for (j = 0; j < QUAD_SIZE; j++) { + if (ref < (data->stencilVals[j] & valMask)) { + passMask |= (1 << j); + } + } + break; + case PIPE_FUNC_EQUAL: + for (j = 0; j < QUAD_SIZE; j++) { + if (ref == (data->stencilVals[j] & valMask)) { + passMask |= (1 << j); + } + } + break; + case PIPE_FUNC_LEQUAL: + for (j = 0; j < QUAD_SIZE; j++) { + if (ref <= (data->stencilVals[j] & valMask)) { + passMask |= (1 << j); + } + } + break; + case PIPE_FUNC_GREATER: + for (j = 0; j < QUAD_SIZE; j++) { + if (ref > (data->stencilVals[j] & valMask)) { + passMask |= (1 << j); + } + } + break; + case PIPE_FUNC_NOTEQUAL: + for (j = 0; j < QUAD_SIZE; j++) { + if (ref != (data->stencilVals[j] & valMask)) { + passMask |= (1 << j); + } + } + break; + case PIPE_FUNC_GEQUAL: + for (j = 0; j < QUAD_SIZE; j++) { + if (ref >= (data->stencilVals[j] & valMask)) { + passMask |= (1 << j); + } + } + break; + case PIPE_FUNC_ALWAYS: + passMask = MASK_ALL; + break; + default: + assert(0); + } + + return passMask; +} + + +/** + * Apply the stencil operator to stencil values. + * + * \param data->stencilVals the stencil buffer values (read and written) + * \param mask indicates which pixels to update + * \param op the stencil operator (PIPE_STENCIL_OP_x) + * \param ref the stencil reference value + * \param wrtMask writemask controlling which bits are changed in the + * stencil values + */ +static void +apply_stencil_op(struct depth_data *data, + unsigned mask, unsigned op, ubyte ref, ubyte wrtMask) +{ + unsigned j; + ubyte newstencil[QUAD_SIZE]; + + for (j = 0; j < QUAD_SIZE; j++) { + newstencil[j] = data->stencilVals[j]; + } + + switch (op) { + case PIPE_STENCIL_OP_KEEP: + /* no-op */ + break; + case PIPE_STENCIL_OP_ZERO: + for (j = 0; j < QUAD_SIZE; j++) { + if (mask & (1 << j)) { + newstencil[j] = 0; + } + } + break; + case PIPE_STENCIL_OP_REPLACE: + for (j = 0; j < QUAD_SIZE; j++) { + if (mask & (1 << j)) { + newstencil[j] = ref; + } + } + break; + case PIPE_STENCIL_OP_INCR: + for (j = 0; j < QUAD_SIZE; j++) { + if (mask & (1 << j)) { + if (data->stencilVals[j] < STENCIL_MAX) { + newstencil[j] = data->stencilVals[j] + 1; + } + } + } + break; + case PIPE_STENCIL_OP_DECR: + for (j = 0; j < QUAD_SIZE; j++) { + if (mask & (1 << j)) { + if (data->stencilVals[j] > 0) { + newstencil[j] = data->stencilVals[j] - 1; + } + } + } + break; + case PIPE_STENCIL_OP_INCR_WRAP: + for (j = 0; j < QUAD_SIZE; j++) { + if (mask & (1 << j)) { + newstencil[j] = data->stencilVals[j] + 1; + } + } + break; + case PIPE_STENCIL_OP_DECR_WRAP: + for (j = 0; j < QUAD_SIZE; j++) { + if (mask & (1 << j)) { + newstencil[j] = data->stencilVals[j] - 1; + } + } + break; + case PIPE_STENCIL_OP_INVERT: + for (j = 0; j < QUAD_SIZE; j++) { + if (mask & (1 << j)) { + newstencil[j] = ~data->stencilVals[j]; } } break; @@ -142,6 +390,39 @@ sp_depth_test_quad(struct quad_stage *qs, struct quad_header *quad) assert(0); } + /* + * update the stencil values + */ + if (wrtMask != STENCIL_MAX) { + /* apply bit-wise stencil buffer writemask */ + for (j = 0; j < QUAD_SIZE; j++) { + data->stencilVals[j] = (wrtMask & newstencil[j]) | (~wrtMask & data->stencilVals[j]); + } + } + else { + for (j = 0; j < QUAD_SIZE; j++) { + data->stencilVals[j] = newstencil[j]; + } + } +} + + + +/* + * To increase efficiency, we should probably have multiple versions + * of this function that are specifically for Z16, Z32 and FP Z buffers. + * Try to effectively do that with codegen... + */ + +static boolean +depth_test_quad(struct quad_stage *qs, + struct depth_data *data, + struct quad_header *quad) +{ + struct softpipe_context *softpipe = qs->softpipe; + unsigned zmask = 0; + unsigned j; + switch (softpipe->depth_stencil->depth.func) { case PIPE_FUNC_NEVER: /* zmask = 0 */ @@ -151,37 +432,37 @@ sp_depth_test_quad(struct quad_stage *qs, struct quad_header *quad) * Like this: quad->mask &= (quad->outputs.depth < zzzz); */ for (j = 0; j < QUAD_SIZE; j++) { - if (qzzzz[j] < bzzzz[j]) + if (data->qzzzz[j] < data->bzzzz[j]) zmask |= 1 << j; } break; case PIPE_FUNC_EQUAL: for (j = 0; j < QUAD_SIZE; j++) { - if (qzzzz[j] == bzzzz[j]) + if (data->qzzzz[j] == data->bzzzz[j]) zmask |= 1 << j; } break; case PIPE_FUNC_LEQUAL: for (j = 0; j < QUAD_SIZE; j++) { - if (qzzzz[j] <= bzzzz[j]) + if (data->qzzzz[j] <= data->bzzzz[j]) zmask |= (1 << j); } break; case PIPE_FUNC_GREATER: for (j = 0; j < QUAD_SIZE; j++) { - if (qzzzz[j] > bzzzz[j]) + if (data->qzzzz[j] > data->bzzzz[j]) zmask |= (1 << j); } break; case PIPE_FUNC_NOTEQUAL: for (j = 0; j < QUAD_SIZE; j++) { - if (qzzzz[j] != bzzzz[j]) + if (data->qzzzz[j] != data->bzzzz[j]) zmask |= (1 << j); } break; case PIPE_FUNC_GEQUAL: for (j = 0; j < QUAD_SIZE; j++) { - if (qzzzz[j] >= bzzzz[j]) + if (data->qzzzz[j] >= data->bzzzz[j]) zmask |= (1 << j); } break; @@ -193,80 +474,480 @@ sp_depth_test_quad(struct quad_stage *qs, struct quad_header *quad) } quad->inout.mask &= zmask; + if (quad->inout.mask == 0) + return FALSE; + /* Update our internal copy only if writemask set. Even if + * depth.writemask is FALSE, may still need to write out buffer + * data due to stencil changes. + */ if (softpipe->depth_stencil->depth.writemask) { - - /* This is also efficient with sse / spe instructions: - */ for (j = 0; j < QUAD_SIZE; j++) { - if (quad->inout.mask & (1 << j)) { - bzzzz[j] = qzzzz[j]; - } + if (quad->inout.mask & (1 << j)) { + data->bzzzz[j] = data->qzzzz[j]; + } } + } - /* put updated Z values back into cached tile */ - switch (format) { - case PIPE_FORMAT_Z16_UNORM: - for (j = 0; j < QUAD_SIZE; j++) { - int x = quad->input.x0 % TILE_SIZE + (j & 1); - int y = quad->input.y0 % TILE_SIZE + (j >> 1); - tile->data.depth16[y][x] = (ushort) bzzzz[j]; + return TRUE; +} + + + +/** + * Do stencil (and depth) testing. Stenciling depends on the outcome of + * depth testing. + */ +static void +depth_stencil_test_quad(struct quad_stage *qs, + struct depth_data *data, + struct quad_header *quad) +{ + struct softpipe_context *softpipe = qs->softpipe; + unsigned func, zFailOp, zPassOp, failOp; + ubyte ref, wrtMask, valMask; + uint face = quad->input.facing; + + if (!softpipe->depth_stencil->stencil[1].enabled) { + /* single-sided stencil test, use front (face=0) state */ + face = 0; + } + + /* choose front or back face function, operator, etc */ + /* XXX we could do these initializations once per primitive */ + func = softpipe->depth_stencil->stencil[face].func; + failOp = softpipe->depth_stencil->stencil[face].fail_op; + zFailOp = softpipe->depth_stencil->stencil[face].zfail_op; + zPassOp = softpipe->depth_stencil->stencil[face].zpass_op; + ref = softpipe->depth_stencil->stencil[face].ref_value; + wrtMask = softpipe->depth_stencil->stencil[face].writemask; + valMask = softpipe->depth_stencil->stencil[face].valuemask; + + + /* do the stencil test first */ + { + unsigned passMask, failMask; + passMask = do_stencil_test(data, func, ref, valMask); + failMask = quad->inout.mask & ~passMask; + quad->inout.mask &= passMask; + + if (failOp != PIPE_STENCIL_OP_KEEP) { + apply_stencil_op(data, failMask, failOp, ref, wrtMask); + } + } + + if (quad->inout.mask) { + /* now the pixels that passed the stencil test are depth tested */ + if (softpipe->depth_stencil->depth.enabled) { + const unsigned origMask = quad->inout.mask; + + depth_test_quad(qs, data, quad); /* quad->mask is updated */ + + /* update stencil buffer values according to z pass/fail result */ + if (zFailOp != PIPE_STENCIL_OP_KEEP) { + const unsigned zFailMask = origMask & ~quad->inout.mask; + apply_stencil_op(data, zFailMask, zFailOp, ref, wrtMask); } - break; - case PIPE_FORMAT_X8Z24_UNORM: - /* fall-through */ - /* (yes, this falls through to a different case than above) */ - case PIPE_FORMAT_Z32_UNORM: - for (j = 0; j < QUAD_SIZE; j++) { - int x = quad->input.x0 % TILE_SIZE + (j & 1); - int y = quad->input.y0 % TILE_SIZE + (j >> 1); - tile->data.depth32[y][x] = bzzzz[j]; + + if (zPassOp != PIPE_STENCIL_OP_KEEP) { + const unsigned zPassMask = origMask & quad->inout.mask; + apply_stencil_op(data, zPassMask, zPassOp, ref, wrtMask); } - break; - case PIPE_FORMAT_S8Z24_UNORM: - for (j = 0; j < QUAD_SIZE; j++) { - int x = quad->input.x0 % TILE_SIZE + (j & 1); - int y = quad->input.y0 % TILE_SIZE + (j >> 1); - uint s8z24 = tile->data.depth32[y][x]; - s8z24 = (s8z24 & 0xff000000) | bzzzz[j]; - tile->data.depth32[y][x] = s8z24; + } + else { + /* no depth test, apply Zpass operator to stencil buffer values */ + apply_stencil_op(data, quad->inout.mask, zPassOp, ref, wrtMask); + } + } +} + + +#define ALPHATEST( FUNC, COMP ) \ + static int \ + alpha_test_quads_##FUNC( struct quad_stage *qs, \ + struct quad_header *quads[], \ + unsigned nr ) \ + { \ + const float ref = qs->softpipe->depth_stencil->alpha.ref_value; \ + const uint cbuf = 0; /* only output[0].alpha is tested */ \ + unsigned pass_nr = 0; \ + unsigned i; \ + \ + for (i = 0; i < nr; i++) { \ + const float *aaaa = quads[i]->output.color[cbuf][3]; \ + unsigned passMask = 0; \ + \ + if (aaaa[0] COMP ref) passMask |= (1 << 0); \ + if (aaaa[1] COMP ref) passMask |= (1 << 1); \ + if (aaaa[2] COMP ref) passMask |= (1 << 2); \ + if (aaaa[3] COMP ref) passMask |= (1 << 3); \ + \ + quads[i]->inout.mask &= passMask; \ + \ + if (quads[i]->inout.mask) \ + quads[pass_nr++] = quads[i]; \ + } \ + \ + return pass_nr; \ + } + + +ALPHATEST( LESS, < ) +ALPHATEST( EQUAL, == ) +ALPHATEST( LEQUAL, <= ) +ALPHATEST( GREATER, > ) +ALPHATEST( NOTEQUAL, != ) +ALPHATEST( GEQUAL, >= ) + + +/* XXX: Incorporate into shader using KILP. + */ +static int +alpha_test_quads(struct quad_stage *qs, + struct quad_header *quads[], + unsigned nr) +{ + switch (qs->softpipe->depth_stencil->alpha.func) { + case PIPE_FUNC_LESS: + return alpha_test_quads_LESS( qs, quads, nr ); + case PIPE_FUNC_EQUAL: + return alpha_test_quads_EQUAL( qs, quads, nr ); + break; + case PIPE_FUNC_LEQUAL: + return alpha_test_quads_LEQUAL( qs, quads, nr ); + case PIPE_FUNC_GREATER: + return alpha_test_quads_GREATER( qs, quads, nr ); + case PIPE_FUNC_NOTEQUAL: + return alpha_test_quads_NOTEQUAL( qs, quads, nr ); + case PIPE_FUNC_GEQUAL: + return alpha_test_quads_GEQUAL( qs, quads, nr ); + case PIPE_FUNC_ALWAYS: + return nr; + case PIPE_FUNC_NEVER: + default: + return 0; + } +} + +static unsigned mask_count[16] = +{ + 0, /* 0x0 */ + 1, /* 0x1 */ + 1, /* 0x2 */ + 2, /* 0x3 */ + 1, /* 0x4 */ + 2, /* 0x5 */ + 2, /* 0x6 */ + 3, /* 0x7 */ + 1, /* 0x8 */ + 2, /* 0x9 */ + 2, /* 0xa */ + 3, /* 0xb */ + 2, /* 0xc */ + 3, /* 0xd */ + 3, /* 0xe */ + 4, /* 0xf */ +}; + + + +static void +depth_test_quads_fallback(struct quad_stage *qs, + struct quad_header *quads[], + unsigned nr) +{ + unsigned i, pass = 0; + const struct sp_fragment_shader *fs = qs->softpipe->fs; + boolean interp_depth = !fs->info.writes_z; + struct depth_data data; + + + if (qs->softpipe->depth_stencil->alpha.enabled) { + nr = alpha_test_quads(qs, quads, nr); + } + + if (qs->softpipe->framebuffer.zsbuf && + (qs->softpipe->depth_stencil->depth.enabled || + qs->softpipe->depth_stencil->stencil[0].enabled)) { + + data.ps = qs->softpipe->framebuffer.zsbuf; + data.format = data.ps->format; + data.tile = sp_get_cached_tile(qs->softpipe->zsbuf_cache, + quads[0]->input.x0, + quads[0]->input.y0); + + for (i = 0; i < nr; i++) { + get_depth_stencil_values(&data, quads[i]); + + if (qs->softpipe->depth_stencil->depth.enabled) { + if (interp_depth) + interpolate_quad_depth(quads[i]); + + convert_quad_depth(&data, quads[i]); } - break; - case PIPE_FORMAT_Z24S8_UNORM: - for (j = 0; j < QUAD_SIZE; j++) { - int x = quad->input.x0 % TILE_SIZE + (j & 1); - int y = quad->input.y0 % TILE_SIZE + (j >> 1); - uint z24s8 = tile->data.depth32[y][x]; - z24s8 = (z24s8 & 0xff) | (bzzzz[j] << 8); - tile->data.depth32[y][x] = z24s8; + + if (qs->softpipe->depth_stencil->stencil[0].enabled) { + depth_stencil_test_quad(qs, &data, quads[i]); + write_depth_stencil_values(&data, quads[i]); } - break; - case PIPE_FORMAT_Z24X8_UNORM: - for (j = 0; j < QUAD_SIZE; j++) { - int x = quad->input.x0 % TILE_SIZE + (j & 1); - int y = quad->input.y0 % TILE_SIZE + (j >> 1); - tile->data.depth32[y][x] = bzzzz[j] << 8; + else { + if (!depth_test_quad(qs, &data, quads[i])) + continue; + + if (qs->softpipe->depth_stencil->depth.writemask) + write_depth_stencil_values(&data, quads[i]); } - break; - default: - assert(0); + + + quads[pass++] = quads[i]; + } + + nr = pass; + } + + if (qs->softpipe->active_query_count) { + for (i = 0; i < nr; i++) + qs->softpipe->occlusion_count += mask_count[quads[i]->inout.mask]; + } + + if (nr) + qs->next->run(qs->next, quads, nr); +} + +/* XXX: this function assumes setup function actually emits linear + * spans of quads. It seems a lot more natural to do (early) + * depth-testing on spans rather than quads. + */ +static void +depth_interp_z16_less_write(struct quad_stage *qs, + struct quad_header *quads[], + unsigned nr) +{ + unsigned i, pass = 0; + const unsigned ix = quads[0]->input.x0; + const unsigned iy = quads[0]->input.y0; + const float fx = (float) ix; + const float fy = (float) iy; + const float dzdx = quads[0]->posCoef->dadx[2]; + const float dzdy = quads[0]->posCoef->dady[2]; + const float z0 = quads[0]->posCoef->a0[2] + dzdx * fx + dzdy * fy; + struct softpipe_cached_tile *tile; + ushort (*depth16)[TILE_SIZE]; + ushort idepth[4], depth_step; + const float scale = 65535.0; + + idepth[0] = (ushort)((z0) * scale); + idepth[1] = (ushort)((z0 + dzdx) * scale); + idepth[2] = (ushort)((z0 + dzdy) * scale); + idepth[3] = (ushort)((z0 + dzdx + dzdy) * scale); + + depth_step = (ushort)(dzdx * 2 * scale); + + tile = sp_get_cached_tile(qs->softpipe->zsbuf_cache, ix, iy); + + depth16 = (ushort (*)[TILE_SIZE]) + &tile->data.depth16[iy % TILE_SIZE][ix % TILE_SIZE]; + + for (i = 0; i < nr; i++) { + unsigned outmask = quads[i]->inout.mask; + unsigned mask = 0; + + if ((outmask & 1) && idepth[0] < depth16[0][0]) { + depth16[0][0] = idepth[0]; + mask |= (1 << 0); + } + + if ((outmask & 2) && idepth[1] < depth16[0][1]) { + depth16[0][1] = idepth[1]; + mask |= (1 << 1); + } + + if ((outmask & 4) && idepth[2] < depth16[1][0]) { + depth16[1][0] = idepth[2]; + mask |= (1 << 2); + } + + if ((outmask & 8) && idepth[3] < depth16[1][1]) { + depth16[1][1] = idepth[3]; + mask |= (1 << 3); } + + idepth[0] += depth_step; + idepth[1] += depth_step; + idepth[2] += depth_step; + idepth[3] += depth_step; + + depth16 = (ushort (*)[TILE_SIZE]) &depth16[0][2]; + + quads[i]->inout.mask = mask; + if (quads[i]->inout.mask) + quads[pass++] = quads[i]; } + + if (pass) + qs->next->run(qs->next, quads, pass); + } static void -depth_test_quad(struct quad_stage *qs, struct quad_header *quad) +depth_interp_z16_lequal_write(struct quad_stage *qs, + struct quad_header *quads[], + unsigned nr) { - sp_depth_test_quad(qs, quad); + unsigned i, pass = 0; + const unsigned ix = quads[0]->input.x0; + const unsigned iy = quads[0]->input.y0; + const float fx = (float) ix; + const float fy = (float) iy; + const float dzdx = quads[0]->posCoef->dadx[2]; + const float dzdy = quads[0]->posCoef->dady[2]; + const float z0 = quads[0]->posCoef->a0[2] + dzdx * fx + dzdy * fy; + struct softpipe_cached_tile *tile; + ushort (*depth16)[TILE_SIZE]; + ushort idepth[4], depth_step; + const float scale = 65535.0; + + idepth[0] = (ushort)((z0) * scale); + idepth[1] = (ushort)((z0 + dzdx) * scale); + idepth[2] = (ushort)((z0 + dzdy) * scale); + idepth[3] = (ushort)((z0 + dzdx + dzdy) * scale); + + depth_step = (ushort)(dzdx * 2 * scale); + + tile = sp_get_cached_tile(qs->softpipe->zsbuf_cache, ix, iy); + + depth16 = (ushort (*)[TILE_SIZE]) + &tile->data.depth16[iy % TILE_SIZE][ix % TILE_SIZE]; + + for (i = 0; i < nr; i++) { + unsigned outmask = quads[i]->inout.mask; + unsigned mask = 0; + + if ((outmask & 1) && idepth[0] <= depth16[0][0]) { + depth16[0][0] = idepth[0]; + mask |= (1 << 0); + } + + if ((outmask & 2) && idepth[1] <= depth16[0][1]) { + depth16[0][1] = idepth[1]; + mask |= (1 << 1); + } + + if ((outmask & 4) && idepth[2] <= depth16[1][0]) { + depth16[1][0] = idepth[2]; + mask |= (1 << 2); + } + + if ((outmask & 8) && idepth[3] <= depth16[1][1]) { + depth16[1][1] = idepth[3]; + mask |= (1 << 3); + } + + idepth[0] += depth_step; + idepth[1] += depth_step; + idepth[2] += depth_step; + idepth[3] += depth_step; + + depth16 = (ushort (*)[TILE_SIZE]) &depth16[0][2]; + + quads[i]->inout.mask = mask; + if (quads[i]->inout.mask) + quads[pass++] = quads[i]; + } + + if (pass) + qs->next->run(qs->next, quads, pass); - if (quad->inout.mask) - qs->next->run(qs->next, quad); } + + + +static void +depth_noop(struct quad_stage *qs, + struct quad_header *quads[], + unsigned nr) +{ + qs->next->run(qs->next, quads, nr); +} + + + +static void +choose_depth_test(struct quad_stage *qs, + struct quad_header *quads[], + unsigned nr) +{ + boolean interp_depth = !qs->softpipe->fs->info.writes_z; + + boolean alpha = qs->softpipe->depth_stencil->alpha.enabled; + + boolean depth = (qs->softpipe->framebuffer.zsbuf && + qs->softpipe->depth_stencil->depth.enabled); + + unsigned depthfunc = qs->softpipe->depth_stencil->depth.func; + + boolean stencil = qs->softpipe->depth_stencil->stencil[0].enabled; + + boolean depthwrite = qs->softpipe->depth_stencil->depth.writemask; + + boolean occlusion = qs->softpipe->active_query_count; + + + if (!alpha && + !depth && + !stencil) { + qs->run = depth_noop; + } + else if (!alpha && + interp_depth && + depth && + depthwrite && + !occlusion && + !stencil) + { + switch (depthfunc) { + case PIPE_FUNC_LESS: + switch (qs->softpipe->framebuffer.zsbuf->format) { + case PIPE_FORMAT_Z16_UNORM: + qs->run = depth_interp_z16_less_write; + break; + default: + qs->run = depth_test_quads_fallback; + break; + } + break; + case PIPE_FUNC_LEQUAL: + switch (qs->softpipe->framebuffer.zsbuf->format) { + case PIPE_FORMAT_Z16_UNORM: + qs->run = depth_interp_z16_lequal_write; + break; + default: + qs->run = depth_test_quads_fallback; + break; + } + break; + default: + qs->run = depth_test_quads_fallback; + } + } + else { + qs->run = depth_test_quads_fallback; + } + + + qs->run( qs, quads, nr ); +} + + + + + static void depth_test_begin(struct quad_stage *qs) { + qs->run = choose_depth_test; qs->next->begin(qs->next); } @@ -283,7 +964,7 @@ struct quad_stage *sp_quad_depth_test_stage( struct softpipe_context *softpipe ) stage->softpipe = softpipe; stage->begin = depth_test_begin; - stage->run = depth_test_quad; + stage->run = choose_depth_test; stage->destroy = depth_test_destroy; return stage; diff --git a/src/gallium/drivers/softpipe/sp_quad_earlyz.c b/src/gallium/drivers/softpipe/sp_quad_earlyz.c deleted file mode 100644 index 496fd39ed1a..00000000000 --- a/src/gallium/drivers/softpipe/sp_quad_earlyz.c +++ /dev/null @@ -1,88 +0,0 @@ -/************************************************************************** - * - * Copyright 2007 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. - * - **************************************************************************/ - -/** - * \brief Quad early-z testing - */ - -#include "pipe/p_defines.h" -#include "util/u_memory.h" -#include "sp_quad.h" -#include "sp_quad_pipe.h" - - -/** - * All this stage does is compute the quad's Z values (which is normally - * done by the shading stage). - * The next stage will do the actual depth test. - */ -static void -earlyz_quad( - struct quad_stage *qs, - struct quad_header *quad ) -{ - const float fx = (float) quad->input.x0; - const float fy = (float) quad->input.y0; - const float dzdx = quad->posCoef->dadx[2]; - const float dzdy = quad->posCoef->dady[2]; - const float z0 = quad->posCoef->a0[2] + dzdx * fx + dzdy * fy; - - quad->output.depth[0] = z0; - quad->output.depth[1] = z0 + dzdx; - quad->output.depth[2] = z0 + dzdy; - quad->output.depth[3] = z0 + dzdx + dzdy; - - qs->next->run( qs->next, quad ); -} - -static void -earlyz_begin( - struct quad_stage *qs ) -{ - qs->next->begin( qs->next ); -} - -static void -earlyz_destroy( - struct quad_stage *qs ) -{ - FREE( qs ); -} - -struct quad_stage * -sp_quad_earlyz_stage( - struct softpipe_context *softpipe ) -{ - struct quad_stage *stage = CALLOC_STRUCT( quad_stage ); - - stage->softpipe = softpipe; - stage->begin = earlyz_begin; - stage->run = earlyz_quad; - stage->destroy = earlyz_destroy; - - return stage; -} diff --git a/src/gallium/drivers/softpipe/sp_quad_fs.c b/src/gallium/drivers/softpipe/sp_quad_fs.c index 28f8d1a60ea..1e7533d0f9e 100644 --- a/src/gallium/drivers/softpipe/sp_quad_fs.c +++ b/src/gallium/drivers/softpipe/sp_quad_fs.c @@ -68,72 +68,69 @@ quad_shade_stage(struct quad_stage *qs) /** * Execute fragment shader for the four fragments in the quad. */ -static void +static INLINE boolean shade_quad(struct quad_stage *qs, struct quad_header *quad) { struct quad_shade_stage *qss = quad_shade_stage( qs ); struct softpipe_context *softpipe = qs->softpipe; struct tgsi_exec_machine *machine = qss->machine; - boolean z_written; - - /* Consts do not require 16 byte alignment. */ - machine->Consts = softpipe->mapped_constants[PIPE_SHADER_FRAGMENT]; - - machine->InterpCoefs = quad->coef; /* run shader */ - quad->inout.mask &= softpipe->fs->run( softpipe->fs, machine, quad ); - - /* store outputs */ - z_written = FALSE; - { - const ubyte *sem_name = softpipe->fs->info.output_semantic_name; - const ubyte *sem_index = softpipe->fs->info.output_semantic_index; - const uint n = qss->stage.softpipe->fs->info.num_outputs; - uint i; - for (i = 0; i < n; i++) { - switch (sem_name[i]) { - case TGSI_SEMANTIC_COLOR: - { - uint cbuf = sem_index[i]; - memcpy(quad->output.color[cbuf], - &machine->Outputs[i].xyzw[0].f[0], - sizeof(quad->output.color[0]) ); - } - break; - case TGSI_SEMANTIC_POSITION: - { - uint j; - for (j = 0; j < 4; j++) { - quad->output.depth[j] = machine->Outputs[0].xyzw[2].f[j]; - } - z_written = TRUE; - } - break; - } + return softpipe->fs->run( softpipe->fs, machine, quad ); +} + + + +static void +coverage_quad(struct quad_stage *qs, struct quad_header *quad) +{ + struct softpipe_context *softpipe = qs->softpipe; + uint cbuf; + + /* loop over colorbuffer outputs */ + for (cbuf = 0; cbuf < softpipe->framebuffer.nr_cbufs; cbuf++) { + float (*quadColor)[4] = quad->output.color[cbuf]; + unsigned j; + for (j = 0; j < QUAD_SIZE; j++) { + assert(quad->input.coverage[j] >= 0.0); + assert(quad->input.coverage[j] <= 1.0); + quadColor[3][j] *= quad->input.coverage[j]; } } +} + - if (!z_written) { - /* compute Z values now, as in the quad earlyz stage */ - /* XXX we should really only do this if the earlyz stage is not used */ - const float fx = (float) quad->input.x0; - const float fy = (float) quad->input.y0; - const float dzdx = quad->posCoef->dadx[2]; - const float dzdy = quad->posCoef->dady[2]; - const float z0 = quad->posCoef->a0[2] + dzdx * fx + dzdy * fy; - - quad->output.depth[0] = z0; - quad->output.depth[1] = z0 + dzdx; - quad->output.depth[2] = z0 + dzdy; - quad->output.depth[3] = z0 + dzdx + dzdy; - } - /* shader may cull fragments */ - if (quad->inout.mask) { - qs->next->run( qs->next, quad ); +static void +shade_quads(struct quad_stage *qs, + struct quad_header *quads[], + unsigned nr) +{ + struct quad_shade_stage *qss = quad_shade_stage( qs ); + struct softpipe_context *softpipe = qs->softpipe; + struct tgsi_exec_machine *machine = qss->machine; + + unsigned i, pass = 0; + + machine->Consts = softpipe->mapped_constants[PIPE_SHADER_FRAGMENT]; + machine->InterpCoefs = quads[0]->coef; + + for (i = 0; i < nr; i++) { + if (!shade_quad(qs, quads[i])) + continue; + + if (/*do_coverage*/ 0) + coverage_quad( qs, quads[i] ); + + quads[pass++] = quads[i]; } + + if (pass) + qs->next->run(qs->next, quads, pass); } + + + /** @@ -174,7 +171,7 @@ sp_quad_shade_stage( struct softpipe_context *softpipe ) qss->stage.softpipe = softpipe; qss->stage.begin = shade_begin; - qss->stage.run = shade_quad; + qss->stage.run = shade_quads; qss->stage.destroy = shade_destroy; qss->machine = tgsi_exec_machine_create(); diff --git a/src/gallium/drivers/softpipe/sp_quad_occlusion.c b/src/gallium/drivers/softpipe/sp_quad_occlusion.c deleted file mode 100644 index dfa7ff3b1d1..00000000000 --- a/src/gallium/drivers/softpipe/sp_quad_occlusion.c +++ /dev/null @@ -1,85 +0,0 @@ -/************************************************************************** - * - * Copyright 2007 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. - * - **************************************************************************/ - - -/** - * \brief Quad occlusion counter stage - * \author Brian Paul - */ - - -#include "pipe/p_defines.h" -#include "util/u_memory.h" -#include "sp_context.h" -#include "sp_quad.h" -#include "sp_surface.h" -#include "sp_quad_pipe.h" - -static unsigned count_bits( unsigned val ) -{ - unsigned i; - - for (i = 0; val ; val >>= 1) - i += (val & 1); - - return i; -} - -static void -occlusion_count_quad(struct quad_stage *qs, struct quad_header *quad) -{ - struct softpipe_context *softpipe = qs->softpipe; - - softpipe->occlusion_count += count_bits(quad->inout.mask); - - qs->next->run(qs->next, quad); -} - - -static void occlusion_begin(struct quad_stage *qs) -{ - qs->next->begin(qs->next); -} - - -static void occlusion_destroy(struct quad_stage *qs) -{ - FREE( qs ); -} - - -struct quad_stage *sp_quad_occlusion_stage( struct softpipe_context *softpipe ) -{ - struct quad_stage *stage = CALLOC_STRUCT(quad_stage); - - stage->softpipe = softpipe; - stage->begin = occlusion_begin; - stage->run = occlusion_count_quad; - stage->destroy = occlusion_destroy; - - return stage; -} diff --git a/src/gallium/drivers/softpipe/sp_quad_output.c b/src/gallium/drivers/softpipe/sp_quad_output.c deleted file mode 100644 index 92d5f9f3c1a..00000000000 --- a/src/gallium/drivers/softpipe/sp_quad_output.c +++ /dev/null @@ -1,103 +0,0 @@ -/************************************************************************** - * - * Copyright 2007 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 "util/u_memory.h" -#include "sp_context.h" -#include "sp_quad.h" -#include "sp_surface.h" -#include "sp_quad_pipe.h" -#include "sp_tile_cache.h" - - -/** - * Last step of quad processing: write quad colors to the framebuffer, - * taking mask into account. - */ -static void -output_quad(struct quad_stage *qs, struct quad_header *quad) -{ - /* in-tile pos: */ - const int itx = quad->input.x0 % TILE_SIZE; - const int ity = quad->input.y0 % TILE_SIZE; - - struct softpipe_context *softpipe = qs->softpipe; - uint cbuf; - - /* loop over colorbuffer outputs */ - for (cbuf = 0; cbuf < softpipe->framebuffer.nr_cbufs; cbuf++) { - struct softpipe_cached_tile *tile - = sp_get_cached_tile(softpipe, - softpipe->cbuf_cache[cbuf], - quad->input.x0, quad->input.y0); - float (*quadColor)[4] = quad->output.color[cbuf]; - int i, j; - - /* get/swizzle dest colors */ - for (j = 0; j < QUAD_SIZE; j++) { - if (quad->inout.mask & (1 << j)) { - int x = itx + (j & 1); - int y = ity + (j >> 1); - for (i = 0; i < 4; i++) { /* loop over color chans */ - tile->data.color[y][x][i] = quadColor[i][j]; - } - if (0) { - debug_printf("sp write pixel %d,%d: %g, %g, %g\n", - quad->input.x0 + x, - quad->input.y0 + y, - quadColor[0][j], - quadColor[1][j], - quadColor[2][j]); - } - } - } - } -} - - -static void output_begin(struct quad_stage *qs) -{ - assert(qs->next == NULL); -} - - -static void output_destroy(struct quad_stage *qs) -{ - FREE( qs ); -} - - -struct quad_stage *sp_quad_output_stage( struct softpipe_context *softpipe ) -{ - struct quad_stage *stage = CALLOC_STRUCT(quad_stage); - - stage->softpipe = softpipe; - stage->begin = output_begin; - stage->run = output_quad; - stage->destroy = output_destroy; - - return stage; -} diff --git a/src/gallium/drivers/softpipe/sp_quad_pipe.c b/src/gallium/drivers/softpipe/sp_quad_pipe.c index b5f69b74264..1b5bab4eca6 100644 --- a/src/gallium/drivers/softpipe/sp_quad_pipe.c +++ b/src/gallium/drivers/softpipe/sp_quad_pipe.c @@ -31,88 +31,33 @@ #include "pipe/p_shader_tokens.h" static void -sp_push_quad_first( - struct softpipe_context *sp, - struct quad_stage *quad, - uint i ) +sp_push_quad_first( struct softpipe_context *sp, + struct quad_stage *quad ) { - quad->next = sp->quad[i].first; - sp->quad[i].first = quad; + quad->next = sp->quad.first; + sp->quad.first = quad; } -static void -sp_build_depth_stencil( - struct softpipe_context *sp, - uint i ) -{ - if (sp->depth_stencil->stencil[0].enabled || - sp->depth_stencil->stencil[1].enabled) { - sp_push_quad_first( sp, sp->quad[i].stencil_test, i ); - } - else if (sp->depth_stencil->depth.enabled && - sp->framebuffer.zsbuf) { - sp_push_quad_first( sp, sp->quad[i].depth_test, i ); - } -} void sp_build_quad_pipeline(struct softpipe_context *sp) { - uint i; - boolean early_depth_test = - sp->depth_stencil->depth.enabled && - sp->framebuffer.zsbuf && - !sp->depth_stencil->alpha.enabled && - !sp->fs->info.uses_kill && - !sp->fs->info.writes_z; - - /* build up the pipeline in reverse order... */ - for (i = 0; i < SP_NUM_QUAD_THREADS; i++) { - sp->quad[i].first = sp->quad[i].output; - - if (sp->blend->colormask != 0xf) { - sp_push_quad_first( sp, sp->quad[i].colormask, i ); - } + sp->depth_stencil->depth.enabled && + sp->framebuffer.zsbuf && + !sp->depth_stencil->alpha.enabled && + !sp->fs->info.uses_kill && + !sp->fs->info.writes_z; - if (sp->blend->blend_enable || - sp->blend->logicop_enable) { - sp_push_quad_first( sp, sp->quad[i].blend, i ); - } + sp->quad.first = sp->quad.blend; - if (sp->active_query_count) { - sp_push_quad_first( sp, sp->quad[i].occlusion, i ); - } - - if (sp->rasterizer->poly_smooth || - sp->rasterizer->line_smooth || - sp->rasterizer->point_smooth) { - sp_push_quad_first( sp, sp->quad[i].coverage, i ); - } - - if (!early_depth_test) { - sp_build_depth_stencil( sp, i ); - } - - if (sp->depth_stencil->alpha.enabled) { - sp_push_quad_first( sp, sp->quad[i].alpha_test, i ); - } - - /* XXX always enable shader? */ - if (1) { - sp_push_quad_first( sp, sp->quad[i].shade, i ); - } - - if (early_depth_test) { - sp_build_depth_stencil( sp, i ); - sp_push_quad_first( sp, sp->quad[i].earlyz, i ); - } - -#if !USE_DRAW_STAGE_PSTIPPLE - if (sp->rasterizer->poly_stipple_enable) { - sp_push_quad_first( sp, sp->quad[i].polygon_stipple, i ); - } -#endif + if (early_depth_test) { + sp_push_quad_first( sp, sp->quad.shade ); + sp_push_quad_first( sp, sp->quad.depth_test ); + } + else { + sp_push_quad_first( sp, sp->quad.depth_test ); + sp_push_quad_first( sp, sp->quad.shade ); } } diff --git a/src/gallium/drivers/softpipe/sp_quad_pipe.h b/src/gallium/drivers/softpipe/sp_quad_pipe.h index 0e40586ffc8..c0aa1348319 100644 --- a/src/gallium/drivers/softpipe/sp_quad_pipe.h +++ b/src/gallium/drivers/softpipe/sp_quad_pipe.h @@ -49,7 +49,7 @@ struct quad_stage { void (*begin)(struct quad_stage *qs); /** the stage action */ - void (*run)(struct quad_stage *qs, struct quad_header *quad); + void (*run)(struct quad_stage *qs, struct quad_header *quad[], unsigned nr); void (*destroy)(struct quad_stage *qs); }; @@ -69,6 +69,4 @@ struct quad_stage *sp_quad_output_stage( struct softpipe_context *softpipe ); void sp_build_quad_pipeline(struct softpipe_context *sp); -void sp_depth_test_quad(struct quad_stage *qs, struct quad_header *quad); - #endif /* SP_QUAD_PIPE_H */ diff --git a/src/gallium/drivers/softpipe/sp_quad_stencil.c b/src/gallium/drivers/softpipe/sp_quad_stencil.c deleted file mode 100644 index 5e9d447737d..00000000000 --- a/src/gallium/drivers/softpipe/sp_quad_stencil.c +++ /dev/null @@ -1,352 +0,0 @@ - -/** - * \brief Quad stencil testing - */ - - -#include "sp_context.h" -#include "sp_quad.h" -#include "sp_surface.h" -#include "sp_tile_cache.h" -#include "sp_quad_pipe.h" -#include "pipe/p_defines.h" -#include "util/u_memory.h" - - -/** Only 8-bit stencil supported */ -#define STENCIL_MAX 0xff - - -/** - * Do the basic stencil test (compare stencil buffer values against the - * reference value. - * - * \param stencilVals the stencil values from the stencil buffer - * \param func the stencil func (PIPE_FUNC_x) - * \param ref the stencil reference value - * \param valMask the stencil value mask indicating which bits of the stencil - * values and ref value are to be used. - * \return mask indicating which pixels passed the stencil test - */ -static unsigned -do_stencil_test(const ubyte stencilVals[QUAD_SIZE], unsigned func, - unsigned ref, unsigned valMask) -{ - unsigned passMask = 0x0; - unsigned j; - - ref &= valMask; - - switch (func) { - case PIPE_FUNC_NEVER: - /* passMask = 0x0 */ - break; - case PIPE_FUNC_LESS: - for (j = 0; j < QUAD_SIZE; j++) { - if (ref < (stencilVals[j] & valMask)) { - passMask |= (1 << j); - } - } - break; - case PIPE_FUNC_EQUAL: - for (j = 0; j < QUAD_SIZE; j++) { - if (ref == (stencilVals[j] & valMask)) { - passMask |= (1 << j); - } - } - break; - case PIPE_FUNC_LEQUAL: - for (j = 0; j < QUAD_SIZE; j++) { - if (ref <= (stencilVals[j] & valMask)) { - passMask |= (1 << j); - } - } - break; - case PIPE_FUNC_GREATER: - for (j = 0; j < QUAD_SIZE; j++) { - if (ref > (stencilVals[j] & valMask)) { - passMask |= (1 << j); - } - } - break; - case PIPE_FUNC_NOTEQUAL: - for (j = 0; j < QUAD_SIZE; j++) { - if (ref != (stencilVals[j] & valMask)) { - passMask |= (1 << j); - } - } - break; - case PIPE_FUNC_GEQUAL: - for (j = 0; j < QUAD_SIZE; j++) { - if (ref >= (stencilVals[j] & valMask)) { - passMask |= (1 << j); - } - } - break; - case PIPE_FUNC_ALWAYS: - passMask = MASK_ALL; - break; - default: - assert(0); - } - - return passMask; -} - - -/** - * Apply the stencil operator to stencil values. - * - * \param stencilVals the stencil buffer values (read and written) - * \param mask indicates which pixels to update - * \param op the stencil operator (PIPE_STENCIL_OP_x) - * \param ref the stencil reference value - * \param wrtMask writemask controlling which bits are changed in the - * stencil values - */ -static void -apply_stencil_op(ubyte stencilVals[QUAD_SIZE], - unsigned mask, unsigned op, ubyte ref, ubyte wrtMask) -{ - unsigned j; - ubyte newstencil[QUAD_SIZE]; - - for (j = 0; j < QUAD_SIZE; j++) { - newstencil[j] = stencilVals[j]; - } - - switch (op) { - case PIPE_STENCIL_OP_KEEP: - /* no-op */ - break; - case PIPE_STENCIL_OP_ZERO: - for (j = 0; j < QUAD_SIZE; j++) { - if (mask & (1 << j)) { - newstencil[j] = 0; - } - } - break; - case PIPE_STENCIL_OP_REPLACE: - for (j = 0; j < QUAD_SIZE; j++) { - if (mask & (1 << j)) { - newstencil[j] = ref; - } - } - break; - case PIPE_STENCIL_OP_INCR: - for (j = 0; j < QUAD_SIZE; j++) { - if (mask & (1 << j)) { - if (stencilVals[j] < STENCIL_MAX) { - newstencil[j] = stencilVals[j] + 1; - } - } - } - break; - case PIPE_STENCIL_OP_DECR: - for (j = 0; j < QUAD_SIZE; j++) { - if (mask & (1 << j)) { - if (stencilVals[j] > 0) { - newstencil[j] = stencilVals[j] - 1; - } - } - } - break; - case PIPE_STENCIL_OP_INCR_WRAP: - for (j = 0; j < QUAD_SIZE; j++) { - if (mask & (1 << j)) { - newstencil[j] = stencilVals[j] + 1; - } - } - break; - case PIPE_STENCIL_OP_DECR_WRAP: - for (j = 0; j < QUAD_SIZE; j++) { - if (mask & (1 << j)) { - newstencil[j] = stencilVals[j] - 1; - } - } - break; - case PIPE_STENCIL_OP_INVERT: - for (j = 0; j < QUAD_SIZE; j++) { - if (mask & (1 << j)) { - newstencil[j] = ~stencilVals[j]; - } - } - break; - default: - assert(0); - } - - /* - * update the stencil values - */ - if (wrtMask != STENCIL_MAX) { - /* apply bit-wise stencil buffer writemask */ - for (j = 0; j < QUAD_SIZE; j++) { - stencilVals[j] = (wrtMask & newstencil[j]) | (~wrtMask & stencilVals[j]); - } - } - else { - for (j = 0; j < QUAD_SIZE; j++) { - stencilVals[j] = newstencil[j]; - } - } -} - - -/** - * Do stencil (and depth) testing. Stenciling depends on the outcome of - * depth testing. - */ -static void -stencil_test_quad(struct quad_stage *qs, struct quad_header *quad) -{ - struct softpipe_context *softpipe = qs->softpipe; - struct pipe_surface *ps = softpipe->framebuffer.zsbuf; - unsigned func, zFailOp, zPassOp, failOp; - ubyte ref, wrtMask, valMask; - ubyte stencilVals[QUAD_SIZE]; - struct softpipe_cached_tile *tile - = sp_get_cached_tile(softpipe, softpipe->zsbuf_cache, quad->input.x0, quad->input.y0); - uint j; - uint face = quad->input.facing; - - if (!softpipe->depth_stencil->stencil[1].enabled) { - /* single-sided stencil test, use front (face=0) state */ - face = 0; - } - - /* choose front or back face function, operator, etc */ - /* XXX we could do these initializations once per primitive */ - func = softpipe->depth_stencil->stencil[face].func; - failOp = softpipe->depth_stencil->stencil[face].fail_op; - zFailOp = softpipe->depth_stencil->stencil[face].zfail_op; - zPassOp = softpipe->depth_stencil->stencil[face].zpass_op; - ref = softpipe->depth_stencil->stencil[face].ref_value; - wrtMask = softpipe->depth_stencil->stencil[face].writemask; - valMask = softpipe->depth_stencil->stencil[face].valuemask; - - assert(ps); /* shouldn't get here if there's no stencil buffer */ - - /* get stencil values from cached tile */ - switch (ps->format) { - case PIPE_FORMAT_S8Z24_UNORM: - for (j = 0; j < QUAD_SIZE; j++) { - int x = quad->input.x0 % TILE_SIZE + (j & 1); - int y = quad->input.y0 % TILE_SIZE + (j >> 1); - stencilVals[j] = tile->data.depth32[y][x] >> 24; - } - break; - case PIPE_FORMAT_Z24S8_UNORM: - for (j = 0; j < QUAD_SIZE; j++) { - int x = quad->input.x0 % TILE_SIZE + (j & 1); - int y = quad->input.y0 % TILE_SIZE + (j >> 1); - stencilVals[j] = tile->data.depth32[y][x] & 0xff; - } - break; - case PIPE_FORMAT_S8_UNORM: - for (j = 0; j < QUAD_SIZE; j++) { - int x = quad->input.x0 % TILE_SIZE + (j & 1); - int y = quad->input.y0 % TILE_SIZE + (j >> 1); - stencilVals[j] = tile->data.stencil8[y][x]; - } - break; - default: - assert(0); - } - - /* do the stencil test first */ - { - unsigned passMask, failMask; - passMask = do_stencil_test(stencilVals, func, ref, valMask); - failMask = quad->inout.mask & ~passMask; - quad->inout.mask &= passMask; - - if (failOp != PIPE_STENCIL_OP_KEEP) { - apply_stencil_op(stencilVals, failMask, failOp, ref, wrtMask); - } - } - - if (quad->inout.mask) { - /* now the pixels that passed the stencil test are depth tested */ - if (softpipe->depth_stencil->depth.enabled) { - const unsigned origMask = quad->inout.mask; - - sp_depth_test_quad(qs, quad); /* quad->mask is updated */ - - /* update stencil buffer values according to z pass/fail result */ - if (zFailOp != PIPE_STENCIL_OP_KEEP) { - const unsigned failMask = origMask & ~quad->inout.mask; - apply_stencil_op(stencilVals, failMask, zFailOp, ref, wrtMask); - } - - if (zPassOp != PIPE_STENCIL_OP_KEEP) { - const unsigned passMask = origMask & quad->inout.mask; - apply_stencil_op(stencilVals, passMask, zPassOp, ref, wrtMask); - } - } - else { - /* no depth test, apply Zpass operator to stencil buffer values */ - apply_stencil_op(stencilVals, quad->inout.mask, zPassOp, ref, wrtMask); - } - - } - - /* put new stencil values into cached tile */ - switch (ps->format) { - case PIPE_FORMAT_S8Z24_UNORM: - for (j = 0; j < QUAD_SIZE; j++) { - int x = quad->input.x0 % TILE_SIZE + (j & 1); - int y = quad->input.y0 % TILE_SIZE + (j >> 1); - uint s8z24 = tile->data.depth32[y][x]; - s8z24 = (stencilVals[j] << 24) | (s8z24 & 0xffffff); - tile->data.depth32[y][x] = s8z24; - } - break; - case PIPE_FORMAT_Z24S8_UNORM: - for (j = 0; j < QUAD_SIZE; j++) { - int x = quad->input.x0 % TILE_SIZE + (j & 1); - int y = quad->input.y0 % TILE_SIZE + (j >> 1); - uint z24s8 = tile->data.depth32[y][x]; - z24s8 = (z24s8 & 0xffffff00) | stencilVals[j]; - tile->data.depth32[y][x] = z24s8; - } - break; - case PIPE_FORMAT_S8_UNORM: - for (j = 0; j < QUAD_SIZE; j++) { - int x = quad->input.x0 % TILE_SIZE + (j & 1); - int y = quad->input.y0 % TILE_SIZE + (j >> 1); - tile->data.stencil8[y][x] = stencilVals[j]; - } - break; - default: - assert(0); - } - - if (quad->inout.mask) - qs->next->run(qs->next, quad); -} - - -static void stencil_begin(struct quad_stage *qs) -{ - qs->next->begin(qs->next); -} - - -static void stencil_destroy(struct quad_stage *qs) -{ - FREE( qs ); -} - - -struct quad_stage *sp_quad_stencil_test_stage( struct softpipe_context *softpipe ) -{ - struct quad_stage *stage = CALLOC_STRUCT(quad_stage); - - stage->softpipe = softpipe; - stage->begin = stencil_begin; - stage->run = stencil_test_quad; - stage->destroy = stencil_destroy; - - return stage; -} diff --git a/src/gallium/drivers/softpipe/sp_quad_stipple.c b/src/gallium/drivers/softpipe/sp_quad_stipple.c index 07162db7b6e..a0527a596a6 100644 --- a/src/gallium/drivers/softpipe/sp_quad_stipple.c +++ b/src/gallium/drivers/softpipe/sp_quad_stipple.c @@ -14,14 +14,20 @@ * Apply polygon stipple to quads produced by triangle rasterization */ static void -stipple_quad(struct quad_stage *qs, struct quad_header *quad) +stipple_quad(struct quad_stage *qs, struct quad_header *quads[], unsigned nr) { static const uint bit31 = 1 << 31; static const uint bit30 = 1 << 30; + unsigned pass = nr; + + struct softpipe_context *softpipe = qs->softpipe; + unsigned q; + + pass = 0; + + for (q = 0; q < nr; q++) { + struct quad_header *quad = quads[q]; - if (quad->input.prim == QUAD_PRIM_TRI) { - struct softpipe_context *softpipe = qs->softpipe; - /* need to invert Y to index into OpenGL's stipple pattern */ const int col0 = quad->input.x0 % 32; const int y0 = quad->input.y0; const int y1 = y0 + 1; @@ -41,13 +47,11 @@ stipple_quad(struct quad_stage *qs, struct quad_header *quad) if ((stipple1 & (bit30 >> col0)) == 0) quad->inout.mask &= ~MASK_BOTTOM_RIGHT; - if (!quad->inout.mask) { - /* all fragments failed stipple test, end of quad pipeline */ - return; - } + if (quad->inout.mask) + quads[pass++] = quad; } - qs->next->run(qs->next, quad); + qs->next->run(qs->next, quads, pass); } diff --git a/src/gallium/drivers/softpipe/sp_screen.c b/src/gallium/drivers/softpipe/sp_screen.c index 6178c4ac7e3..81fb7aa20c6 100644 --- a/src/gallium/drivers/softpipe/sp_screen.c +++ b/src/gallium/drivers/softpipe/sp_screen.c @@ -40,7 +40,7 @@ static const char * softpipe_get_vendor(struct pipe_screen *screen) { - return "Tungsten Graphics, Inc."; + return "VMware, Inc."; } @@ -65,8 +65,6 @@ softpipe_get_param(struct pipe_screen *screen, int param) return 1; case PIPE_CAP_GLSL: return 1; - case PIPE_CAP_S3TC: - return 0; case PIPE_CAP_ANISOTROPIC_FILTER: return 0; case PIPE_CAP_POINT_SPRITE: @@ -84,7 +82,7 @@ softpipe_get_param(struct pipe_screen *screen, int param) case PIPE_CAP_MAX_TEXTURE_2D_LEVELS: return 13; /* max 4Kx4K */ case PIPE_CAP_MAX_TEXTURE_3D_LEVELS: - return 8; /* max 128x128x128 */ + return 9; /* max 256x256x256 */ case PIPE_CAP_MAX_TEXTURE_CUBE_LEVELS: return 13; /* max 4Kx4K */ case PIPE_CAP_TGSI_CONT_SUPPORTED: @@ -137,10 +135,14 @@ softpipe_is_format_supported( struct pipe_screen *screen, target == PIPE_TEXTURE_CUBE); switch(format) { + case PIPE_FORMAT_L16_UNORM: + case PIPE_FORMAT_YCBCR_REV: + case PIPE_FORMAT_YCBCR: case PIPE_FORMAT_DXT1_RGB: case PIPE_FORMAT_DXT1_RGBA: case PIPE_FORMAT_DXT3_RGBA: case PIPE_FORMAT_DXT5_RGBA: + case PIPE_FORMAT_Z32_FLOAT: return FALSE; default: return TRUE; diff --git a/src/gallium/drivers/softpipe/sp_setup.c b/src/gallium/drivers/softpipe/sp_setup.c index de3ae3c3696..ade125662ad 100644 --- a/src/gallium/drivers/softpipe/sp_setup.c +++ b/src/gallium/drivers/softpipe/sp_setup.c @@ -33,7 +33,6 @@ */ #include "sp_context.h" -#include "sp_prim_setup.h" #include "sp_quad.h" #include "sp_quad_pipe.h" #include "sp_setup.h" @@ -61,87 +60,9 @@ struct edge { int lines; /**< number of lines on this edge */ }; -#if SP_NUM_QUAD_THREADS > 1 -/* Set to 1 if you want other threads to be instantly - * notified of pending jobs. - */ -#define INSTANT_NOTEMPTY_NOTIFY 0 - -struct thread_info -{ - struct setup_context *setup; - uint id; - pipe_thread handle; -}; - -struct quad_job; - -typedef void (* quad_job_routine)( struct setup_context *setup, uint thread, struct quad_job *job ); - -struct quad_job -{ - struct quad_header_input input; - struct quad_header_inout inout; - quad_job_routine routine; -}; - -#define NUM_QUAD_JOBS 64 - -struct quad_job_que -{ - struct quad_job jobs[NUM_QUAD_JOBS]; - uint first; - uint last; - pipe_mutex que_mutex; - pipe_condvar que_notfull_condvar; - pipe_condvar que_notempty_condvar; - uint jobs_added; - uint jobs_done; - pipe_condvar que_done_condvar; -}; - -static void -add_quad_job( struct quad_job_que *que, struct quad_header *quad, quad_job_routine routine ) -{ -#if INSTANT_NOTEMPTY_NOTIFY - boolean empty; -#endif - - /* Wait for empty slot, see if the que is empty. - */ - pipe_mutex_lock( que->que_mutex ); - while ((que->last + 1) % NUM_QUAD_JOBS == que->first) { -#if !INSTANT_NOTEMPTY_NOTIFY - pipe_condvar_broadcast( que->que_notempty_condvar ); -#endif - pipe_condvar_wait( que->que_notfull_condvar, que->que_mutex ); - } -#if INSTANT_NOTEMPTY_NOTIFY - empty = que->last == que->first; -#endif - que->jobs_added++; - pipe_mutex_unlock( que->que_mutex ); +#define MAX_QUADS 16 - /* Submit new job. - */ - que->jobs[que->last].input = quad->input; - que->jobs[que->last].inout = quad->inout; - que->jobs[que->last].routine = routine; - que->last = (que->last + 1) % NUM_QUAD_JOBS; - -#if INSTANT_NOTEMPTY_NOTIFY - /* If the que was empty, notify consumers there's a job to be done. - */ - if (empty) { - pipe_mutex_lock( que->que_mutex ); - pipe_condvar_broadcast( que->que_notempty_condvar ); - pipe_mutex_unlock( que->que_mutex ); - } -#endif -} - -#endif /** * Triangle setup info (derived from draw_stage). @@ -164,22 +85,19 @@ struct setup_context { struct edge emaj; float oneoverarea; + int facing; + + struct quad_header quad[MAX_QUADS]; + struct quad_header *quad_ptrs[MAX_QUADS]; + unsigned count; struct tgsi_interp_coef coef[PIPE_MAX_SHADER_INPUTS]; struct tgsi_interp_coef posCoef; /* For Z, W */ - struct quad_header quad; - -#if SP_NUM_QUAD_THREADS > 1 - struct quad_job_que que; - struct thread_info threads[SP_NUM_QUAD_THREADS]; -#endif struct { int left[2]; /**< [0] = row0, [1] = row1 */ int right[2]; int y; - unsigned y_flags; - unsigned mask; /**< mask of MASK_BOTTOM/TOP_LEFT/RIGHT bits */ } span; #if DEBUG_FRAGS @@ -190,67 +108,6 @@ struct setup_context { unsigned winding; /* which winding to cull */ }; -#if SP_NUM_QUAD_THREADS > 1 - -static PIPE_THREAD_ROUTINE( quad_thread, param ) -{ - struct thread_info *info = (struct thread_info *) param; - struct quad_job_que *que = &info->setup->que; - - for (;;) { - struct quad_job job; - boolean full; - - /* Wait for an available job. - */ - pipe_mutex_lock( que->que_mutex ); - while (que->last == que->first) - pipe_condvar_wait( que->que_notempty_condvar, que->que_mutex ); - - /* See if the que is full. - */ - full = (que->last + 1) % NUM_QUAD_JOBS == que->first; - - /* Take a job and remove it from que. - */ - job = que->jobs[que->first]; - que->first = (que->first + 1) % NUM_QUAD_JOBS; - - /* Notify the producer if the que is not full. - */ - if (full) - pipe_condvar_signal( que->que_notfull_condvar ); - pipe_mutex_unlock( que->que_mutex ); - - job.routine( info->setup, info->id, &job ); - - /* Notify the producer if that's the last finished job. - */ - pipe_mutex_lock( que->que_mutex ); - que->jobs_done++; - if (que->jobs_added == que->jobs_done) - pipe_condvar_signal( que->que_done_condvar ); - pipe_mutex_unlock( que->que_mutex ); - } - - return NULL; -} - -#define WAIT_FOR_COMPLETION(setup) \ - do {\ - pipe_mutex_lock( setup->que.que_mutex );\ - if (!INSTANT_NOTEMPTY_NOTIFY)\ - pipe_condvar_broadcast( setup->que.que_notempty_condvar );\ - while (setup->que.jobs_added != setup->que.jobs_done)\ - pipe_condvar_wait( setup->que.que_done_condvar, setup->que.que_mutex );\ - pipe_mutex_unlock( setup->que.que_mutex );\ - } while (0) - -#else - -#define WAIT_FOR_COMPLETION(setup) ((void) 0) - -#endif @@ -313,98 +170,18 @@ quad_clip( struct setup_context *setup, struct quad_header *quad ) * Emit a quad (pass to next stage) with clipping. */ static INLINE void -clip_emit_quad( struct setup_context *setup, struct quad_header *quad, uint thread ) +clip_emit_quad( struct setup_context *setup, struct quad_header *quad ) { quad_clip( setup, quad ); + if (quad->inout.mask) { struct softpipe_context *sp = setup->softpipe; - sp->quad[thread].first->run( sp->quad[thread].first, quad ); + sp->quad.first->run( sp->quad.first, &quad, 1 ); } } -#if SP_NUM_QUAD_THREADS > 1 - -static void -clip_emit_quad_job( struct setup_context *setup, uint thread, struct quad_job *job ) -{ - struct quad_header quad; - - quad.input = job->input; - quad.inout = job->inout; - quad.coef = setup->quad.coef; - quad.posCoef = setup->quad.posCoef; - quad.nr_attrs = setup->quad.nr_attrs; - clip_emit_quad( setup, &quad, thread ); -} - -#define CLIP_EMIT_QUAD(setup) add_quad_job( &setup->que, &setup->quad, clip_emit_quad_job ) - -#else - -#define CLIP_EMIT_QUAD(setup) clip_emit_quad( setup, &setup->quad, 0 ) - -#endif - -/** - * Emit a quad (pass to next stage). No clipping is done. - */ -static INLINE void -emit_quad( struct setup_context *setup, struct quad_header *quad, uint thread ) -{ - struct softpipe_context *sp = setup->softpipe; -#if DEBUG_FRAGS - uint mask = quad->inout.mask; -#endif - -#if DEBUG_FRAGS - if (mask & 1) setup->numFragsEmitted++; - if (mask & 2) setup->numFragsEmitted++; - if (mask & 4) setup->numFragsEmitted++; - if (mask & 8) setup->numFragsEmitted++; -#endif - sp->quad[thread].first->run( sp->quad[thread].first, quad ); -#if DEBUG_FRAGS - mask = quad->inout.mask; - if (mask & 1) setup->numFragsWritten++; - if (mask & 2) setup->numFragsWritten++; - if (mask & 4) setup->numFragsWritten++; - if (mask & 8) setup->numFragsWritten++; -#endif -} - -#if SP_NUM_QUAD_THREADS > 1 -static void -emit_quad_job( struct setup_context *setup, uint thread, struct quad_job *job ) -{ - struct quad_header quad; - - quad.input = job->input; - quad.inout = job->inout; - quad.coef = setup->quad.coef; - quad.posCoef = setup->quad.posCoef; - quad.nr_attrs = setup->quad.nr_attrs; - emit_quad( setup, &quad, thread ); -} - -#define EMIT_QUAD(setup,x,y,mask) do {\ - setup->quad.input.x0 = x;\ - setup->quad.input.y0 = y;\ - setup->quad.inout.mask = mask;\ - add_quad_job( &setup->que, &setup->quad, emit_quad_job );\ - } while (0) - -#else - -#define EMIT_QUAD(setup,x,y,mask) do {\ - setup->quad.input.x0 = x;\ - setup->quad.input.y0 = y;\ - setup->quad.inout.mask = mask;\ - emit_quad( setup, &setup->quad, 0 );\ - } while (0) - -#endif /** * Given an X or Y coordinate, return the block/quad coordinate that it @@ -412,7 +189,12 @@ emit_quad_job( struct setup_context *setup, uint thread, struct quad_job *job ) */ static INLINE int block( int x ) { - return x & ~1; + return x & ~(2-1); +} + +static INLINE int block_x( int x ) +{ + return x & ~(16-1); } @@ -421,72 +203,63 @@ static INLINE int block( int x ) */ static void flush_spans( struct setup_context *setup ) { + const int step = 16; const int xleft0 = setup->span.left[0]; const int xleft1 = setup->span.left[1]; const int xright0 = setup->span.right[0]; const int xright1 = setup->span.right[1]; - int minleft, maxright; + struct quad_stage *pipe = setup->softpipe->quad.first; + + + int minleft = block_x(MIN2(xleft0, xleft1)); + int maxright = MAX2(xright0, xright1); int x; - switch (setup->span.y_flags) { - case 0x3: - /* both odd and even lines written (both quad rows) */ - minleft = block(MIN2(xleft0, xleft1)); - maxright = block(MAX2(xright0, xright1)); - for (x = minleft; x <= maxright; x += 2) { - /* determine which of the four pixels is inside the span bounds */ - uint mask = 0x0; - if (x >= xleft0 && x < xright0) - mask |= MASK_TOP_LEFT; - if (x >= xleft1 && x < xright1) - mask |= MASK_BOTTOM_LEFT; - if (x+1 >= xleft0 && x+1 < xright0) - mask |= MASK_TOP_RIGHT; - if (x+1 >= xleft1 && x+1 < xright1) - mask |= MASK_BOTTOM_RIGHT; - if (mask) - EMIT_QUAD( setup, x, setup->span.y, mask ); - } - break; - - case 0x1: - /* only even line written (quad top row) */ - minleft = block(xleft0); - maxright = block(xright0); - for (x = minleft; x <= maxright; x += 2) { - uint mask = 0x0; - if (x >= xleft0 && x < xright0) - mask |= MASK_TOP_LEFT; - if (x+1 >= xleft0 && x+1 < xright0) - mask |= MASK_TOP_RIGHT; - if (mask) - EMIT_QUAD( setup, x, setup->span.y, mask ); - } - break; - - case 0x2: - /* only odd line written (quad bottom row) */ - minleft = block(xleft1); - maxright = block(xright1); - for (x = minleft; x <= maxright; x += 2) { - uint mask = 0x0; - if (x >= xleft1 && x < xright1) - mask |= MASK_BOTTOM_LEFT; - if (x+1 >= xleft1 && x+1 < xright1) - mask |= MASK_BOTTOM_RIGHT; - if (mask) - EMIT_QUAD( setup, x, setup->span.y, mask ); - } - break; + for (x = minleft; x < maxright; x += step) { + unsigned skip_left0 = CLAMP(xleft0 - x, 0, step); + unsigned skip_left1 = CLAMP(xleft1 - x, 0, step); + unsigned skip_right0 = CLAMP(x + step - xright0, 0, step); + unsigned skip_right1 = CLAMP(x + step - xright1, 0, step); + unsigned lx = x; + unsigned q = 0; - default: - return; + unsigned skipmask_left0 = (1U << skip_left0) - 1U; + unsigned skipmask_left1 = (1U << skip_left1) - 1U; + + /* These calculations fail when step == 32 and skip_right == 0. + */ + unsigned skipmask_right0 = ~0U << (unsigned)(step - skip_right0); + unsigned skipmask_right1 = ~0U << (unsigned)(step - skip_right1); + + unsigned mask0 = ~skipmask_left0 & ~skipmask_right0; + unsigned mask1 = ~skipmask_left1 & ~skipmask_right1; + + if (mask0 | mask1) { + do { + unsigned quadmask = (mask0 & 3) | ((mask1 & 3) << 2); + if (quadmask) { + setup->quad[q].input.x0 = lx; + setup->quad[q].input.y0 = setup->span.y; + setup->quad[q].input.facing = setup->facing; + setup->quad[q].inout.mask = quadmask; + setup->quad_ptrs[q] = &setup->quad[q]; + q++; + } + mask0 >>= 2; + mask1 >>= 2; + lx += 2; + } while (mask0 | mask1); + + pipe->run( pipe, setup->quad_ptrs, q ); + } } + setup->span.y = 0; - setup->span.y_flags = 0; setup->span.right[0] = 0; setup->span.right[1] = 0; + setup->span.left[0] = 1000000; /* greater than right[0] */ + setup->span.left[1] = 1000000; /* greater than right[1] */ } @@ -496,7 +269,7 @@ static void print_vertex(const struct setup_context *setup, { int i; debug_printf(" Vertex: (%p)\n", v); - for (i = 0; i < setup->quad.nr_attrs; i++) { + for (i = 0; i < setup->quad[0].nr_attrs; i++) { debug_printf(" %d: %f %f %f %f\n", i, v[i][0], v[i][1], v[i][2], v[i][3]); if (util_is_inf_or_nan(v[i][0])) { @@ -601,7 +374,9 @@ static boolean setup_sort_vertices( struct setup_context *setup, * - the GLSL gl_FrontFacing fragment attribute (bool) * - two-sided stencil test */ - setup->quad.input.facing = (det > 0.0) ^ (setup->softpipe->rasterizer->front_winding == PIPE_WINDING_CW); + setup->facing = + ((det > 0.0) ^ + (setup->softpipe->rasterizer->front_winding == PIPE_WINDING_CW)); return TRUE; } @@ -788,7 +563,7 @@ static void setup_tri_coefficients( struct setup_context *setup ) } if (spfs->info.input_semantic_name[fragSlot] == TGSI_SEMANTIC_FACE) { - setup->coef[fragSlot].a0[0] = 1.0f - setup->quad.input.facing; + setup->coef[fragSlot].a0[0] = 1.0f - setup->facing; setup->coef[fragSlot].dadx[0] = 0.0; setup->coef[fragSlot].dady[0] = 0.0; } @@ -844,11 +619,10 @@ static void subtriangle( struct setup_context *setup, /* clip top/bottom */ start_y = sy; - finish_y = sy + lines; - if (start_y < miny) start_y = miny; + finish_y = sy + lines; if (finish_y > maxy) finish_y = maxy; @@ -885,7 +659,6 @@ static void subtriangle( struct setup_context *setup, setup->span.left[_y&1] = left; setup->span.right[_y&1] = right; - setup->span.y_flags |= 1<<(_y&1); } } @@ -958,10 +731,9 @@ void setup_tri( struct setup_context *setup, setup_tri_coefficients( setup ); setup_tri_edges( setup ); - setup->quad.input.prim = QUAD_PRIM_TRI; + assert(setup->softpipe->reduced_prim == PIPE_PRIM_TRIANGLES); setup->span.y = 0; - setup->span.y_flags = 0; setup->span.right[0] = 0; setup->span.right[1] = 0; /* setup->span.z_mode = tri_z_mode( setup->ctx ); */ @@ -983,8 +755,6 @@ void setup_tri( struct setup_context *setup, flush_spans( setup ); - WAIT_FOR_COMPLETION(setup); - #if DEBUG_FRAGS printf("Tri: %u frags emitted, %u written\n", setup->numFragsEmitted, @@ -1101,7 +871,7 @@ setup_line_coefficients(struct setup_context *setup, } if (spfs->info.input_semantic_name[fragSlot] == TGSI_SEMANTIC_FACE) { - setup->coef[fragSlot].a0[0] = 1.0f - setup->quad.input.facing; + setup->coef[fragSlot].a0[0] = 1.0f - setup->facing; setup->coef[fragSlot].dadx[0] = 0.0; setup->coef[fragSlot].dady[0] = 0.0; } @@ -1122,20 +892,20 @@ plot(struct setup_context *setup, int x, int y) const int quadY = y - iy; const int mask = (1 << ix) << (2 * iy); - if (quadX != setup->quad.input.x0 || - quadY != setup->quad.input.y0) + if (quadX != setup->quad[0].input.x0 || + quadY != setup->quad[0].input.y0) { /* flush prev quad, start new quad */ - if (setup->quad.input.x0 != -1) - CLIP_EMIT_QUAD(setup); + if (setup->quad[0].input.x0 != -1) + clip_emit_quad( setup, &setup->quad[0] ); - setup->quad.input.x0 = quadX; - setup->quad.input.y0 = quadY; - setup->quad.inout.mask = 0x0; + setup->quad[0].input.x0 = quadX; + setup->quad[0].input.y0 = quadY; + setup->quad[0].inout.mask = 0x0; } - setup->quad.inout.mask |= mask; + setup->quad[0].inout.mask |= mask; } @@ -1195,17 +965,18 @@ setup_line(struct setup_context *setup, assert(dx >= 0); assert(dy >= 0); + assert(setup->softpipe->reduced_prim == PIPE_PRIM_LINES); + + setup->quad[0].input.x0 = setup->quad[0].input.y0 = -1; + setup->quad[0].inout.mask = 0x0; - setup->quad.input.x0 = setup->quad.input.y0 = -1; - setup->quad.inout.mask = 0x0; - setup->quad.input.prim = QUAD_PRIM_LINE; /* XXX temporary: set coverage to 1.0 so the line appears * if AA mode happens to be enabled. */ - setup->quad.input.coverage[0] = - setup->quad.input.coverage[1] = - setup->quad.input.coverage[2] = - setup->quad.input.coverage[3] = 1.0; + setup->quad[0].input.coverage[0] = + setup->quad[0].input.coverage[1] = + setup->quad[0].input.coverage[2] = + setup->quad[0].input.coverage[3] = 1.0; if (dx > dy) { /*** X-major line ***/ @@ -1249,11 +1020,9 @@ setup_line(struct setup_context *setup, } /* draw final quad */ - if (setup->quad.inout.mask) { - CLIP_EMIT_QUAD(setup); + if (setup->quad[0].inout.mask) { + clip_emit_quad( setup, &setup->quad[0] ); } - - WAIT_FOR_COMPLETION(setup); } @@ -1300,6 +1069,8 @@ setup_point( struct setup_context *setup, if (softpipe->no_rast) return; + assert(setup->softpipe->reduced_prim == PIPE_PRIM_POINTS); + /* For points, all interpolants are constant-valued. * However, for point sprites, we'll need to setup texcoords appropriately. * XXX: which coefficients are the texcoords??? @@ -1346,22 +1117,21 @@ setup_point( struct setup_context *setup, } if (spfs->info.input_semantic_name[fragSlot] == TGSI_SEMANTIC_FACE) { - setup->coef[fragSlot].a0[0] = 1.0f - setup->quad.input.facing; + setup->coef[fragSlot].a0[0] = 1.0f - setup->facing; setup->coef[fragSlot].dadx[0] = 0.0; setup->coef[fragSlot].dady[0] = 0.0; } } - setup->quad.input.prim = QUAD_PRIM_POINT; if (halfSize <= 0.5 && !round) { /* special case for 1-pixel points */ const int ix = ((int) x) & 1; const int iy = ((int) y) & 1; - setup->quad.input.x0 = (int) x - ix; - setup->quad.input.y0 = (int) y - iy; - setup->quad.inout.mask = (1 << ix) << (2 * iy); - CLIP_EMIT_QUAD(setup); + setup->quad[0].input.x0 = (int) x - ix; + setup->quad[0].input.y0 = (int) y - iy; + setup->quad[0].inout.mask = (1 << ix) << (2 * iy); + clip_emit_quad( setup, &setup->quad[0] ); } else { if (round) { @@ -1381,15 +1151,15 @@ setup_point( struct setup_context *setup, for (ix = ixmin; ix <= ixmax; ix += 2) { float dx, dy, dist2, cover; - setup->quad.inout.mask = 0x0; + setup->quad[0].inout.mask = 0x0; dx = (ix + 0.5f) - x; dy = (iy + 0.5f) - y; dist2 = dx * dx + dy * dy; if (dist2 <= rmax2) { cover = 1.0F - (dist2 - rmin2) * cscale; - setup->quad.input.coverage[QUAD_TOP_LEFT] = MIN2(cover, 1.0f); - setup->quad.inout.mask |= MASK_TOP_LEFT; + setup->quad[0].input.coverage[QUAD_TOP_LEFT] = MIN2(cover, 1.0f); + setup->quad[0].inout.mask |= MASK_TOP_LEFT; } dx = (ix + 1.5f) - x; @@ -1397,8 +1167,8 @@ setup_point( struct setup_context *setup, dist2 = dx * dx + dy * dy; if (dist2 <= rmax2) { cover = 1.0F - (dist2 - rmin2) * cscale; - setup->quad.input.coverage[QUAD_TOP_RIGHT] = MIN2(cover, 1.0f); - setup->quad.inout.mask |= MASK_TOP_RIGHT; + setup->quad[0].input.coverage[QUAD_TOP_RIGHT] = MIN2(cover, 1.0f); + setup->quad[0].inout.mask |= MASK_TOP_RIGHT; } dx = (ix + 0.5f) - x; @@ -1406,8 +1176,8 @@ setup_point( struct setup_context *setup, dist2 = dx * dx + dy * dy; if (dist2 <= rmax2) { cover = 1.0F - (dist2 - rmin2) * cscale; - setup->quad.input.coverage[QUAD_BOTTOM_LEFT] = MIN2(cover, 1.0f); - setup->quad.inout.mask |= MASK_BOTTOM_LEFT; + setup->quad[0].input.coverage[QUAD_BOTTOM_LEFT] = MIN2(cover, 1.0f); + setup->quad[0].inout.mask |= MASK_BOTTOM_LEFT; } dx = (ix + 1.5f) - x; @@ -1415,14 +1185,14 @@ setup_point( struct setup_context *setup, dist2 = dx * dx + dy * dy; if (dist2 <= rmax2) { cover = 1.0F - (dist2 - rmin2) * cscale; - setup->quad.input.coverage[QUAD_BOTTOM_RIGHT] = MIN2(cover, 1.0f); - setup->quad.inout.mask |= MASK_BOTTOM_RIGHT; + setup->quad[0].input.coverage[QUAD_BOTTOM_RIGHT] = MIN2(cover, 1.0f); + setup->quad[0].inout.mask |= MASK_BOTTOM_RIGHT; } - if (setup->quad.inout.mask) { - setup->quad.input.x0 = ix; - setup->quad.input.y0 = iy; - CLIP_EMIT_QUAD(setup); + if (setup->quad[0].inout.mask) { + setup->quad[0].input.x0 = ix; + setup->quad[0].input.y0 = iy; + clip_emit_quad( setup, &setup->quad[0] ); } } } @@ -1466,33 +1236,25 @@ setup_point( struct setup_context *setup, mask &= (MASK_BOTTOM_LEFT | MASK_TOP_LEFT); } - setup->quad.inout.mask = mask; - setup->quad.input.x0 = ix; - setup->quad.input.y0 = iy; - CLIP_EMIT_QUAD(setup); + setup->quad[0].inout.mask = mask; + setup->quad[0].input.x0 = ix; + setup->quad[0].input.y0 = iy; + clip_emit_quad( setup, &setup->quad[0] ); } } } } - - WAIT_FOR_COMPLETION(setup); } void setup_prepare( struct setup_context *setup ) { struct softpipe_context *sp = setup->softpipe; - unsigned i; if (sp->dirty) { softpipe_update_derived(sp); } - /* Note: nr_attrs is only used for debugging (vertex printing) */ - setup->quad.nr_attrs = draw_num_vs_outputs(sp->draw); - - for (i = 0; i < SP_NUM_QUAD_THREADS; i++) { - sp->quad[i].first->begin( sp->quad[i].first ); - } + sp->quad.first->begin( sp->quad.first ); if (sp->reduced_api_prim == PIPE_PRIM_TRIANGLES && sp->rasterizer->fill_cw == PIPE_POLYGON_MODE_FILL && @@ -1520,30 +1282,17 @@ void setup_destroy_context( struct setup_context *setup ) struct setup_context *setup_create_context( struct softpipe_context *softpipe ) { struct setup_context *setup = CALLOC_STRUCT(setup_context); -#if SP_NUM_QUAD_THREADS > 1 - uint i; -#endif + unsigned i; setup->softpipe = softpipe; - setup->quad.coef = setup->coef; - setup->quad.posCoef = &setup->posCoef; - -#if SP_NUM_QUAD_THREADS > 1 - setup->que.first = 0; - setup->que.last = 0; - pipe_mutex_init( setup->que.que_mutex ); - pipe_condvar_init( setup->que.que_notfull_condvar ); - pipe_condvar_init( setup->que.que_notempty_condvar ); - setup->que.jobs_added = 0; - setup->que.jobs_done = 0; - pipe_condvar_init( setup->que.que_done_condvar ); - for (i = 0; i < SP_NUM_QUAD_THREADS; i++) { - setup->threads[i].setup = setup; - setup->threads[i].id = i; - setup->threads[i].handle = pipe_thread_create( quad_thread, &setup->threads[i] ); + for (i = 0; i < MAX_QUADS; i++) { + setup->quad[i].coef = setup->coef; + setup->quad[i].posCoef = &setup->posCoef; } -#endif + + setup->span.left[0] = 1000000; /* greater than right[0] */ + setup->span.left[1] = 1000000; /* greater than right[1] */ return setup; } diff --git a/src/gallium/drivers/softpipe/sp_state.h b/src/gallium/drivers/softpipe/sp_state.h index 9776e978e3e..77ee3c1136b 100644 --- a/src/gallium/drivers/softpipe/sp_state.h +++ b/src/gallium/drivers/softpipe/sp_state.h @@ -87,6 +87,7 @@ struct sp_fragment_shader { struct sp_vertex_shader { struct pipe_shader_state shader; struct draw_vertex_shader *draw_data; + int max_sampler; /* -1 if no samplers */ }; diff --git a/src/gallium/drivers/softpipe/sp_state_blend.c b/src/gallium/drivers/softpipe/sp_state_blend.c index 384fe559afd..efed082f823 100644 --- a/src/gallium/drivers/softpipe/sp_state_blend.c +++ b/src/gallium/drivers/softpipe/sp_state_blend.c @@ -45,7 +45,7 @@ void softpipe_bind_blend_state( struct pipe_context *pipe, { struct softpipe_context *softpipe = softpipe_context(pipe); - softpipe->blend = (const struct pipe_blend_state *)blend; + softpipe->blend = (struct pipe_blend_state *)blend; softpipe->dirty |= SP_NEW_BLEND; } @@ -86,7 +86,7 @@ softpipe_bind_depth_stencil_state(struct pipe_context *pipe, { struct softpipe_context *softpipe = softpipe_context(pipe); - softpipe->depth_stencil = (const struct pipe_depth_stencil_alpha_state *)depth_stencil; + softpipe->depth_stencil = (struct pipe_depth_stencil_alpha_state *)depth_stencil; softpipe->dirty |= SP_NEW_DEPTH_STENCIL_ALPHA; } diff --git a/src/gallium/drivers/softpipe/sp_state_derived.c b/src/gallium/drivers/softpipe/sp_state_derived.c index 75551000c9b..1faeca1c2a3 100644 --- a/src/gallium/drivers/softpipe/sp_state_derived.c +++ b/src/gallium/drivers/softpipe/sp_state_derived.c @@ -32,7 +32,10 @@ #include "draw/draw_vertex.h" #include "draw/draw_private.h" #include "sp_context.h" +#include "sp_screen.h" #include "sp_state.h" +#include "sp_texture.h" +#include "sp_tex_tile_cache.h" /** @@ -65,24 +68,19 @@ softpipe_get_vertex_info(struct softpipe_context *softpipe) const struct sp_fragment_shader *spfs = softpipe->fs; const enum interp_mode colorInterp = softpipe->rasterizer->flatshade ? INTERP_CONSTANT : INTERP_LINEAR; + struct vertex_info *vinfo_vbuf = &softpipe->vertex_info_vbuf; + const uint num = draw_num_vs_outputs(softpipe->draw); uint i; - if (softpipe->vbuf) { - /* if using the post-transform vertex buffer, tell draw_vbuf to - * simply emit the whole post-xform vertex as-is: - */ - struct vertex_info *vinfo_vbuf = &softpipe->vertex_info_vbuf; - const uint num = draw_num_vs_outputs(softpipe->draw); - uint i; - - /* No longer any need to try and emit draw vertex_header info. - */ - vinfo_vbuf->num_attribs = 0; - for (i = 0; i < num; i++) { - draw_emit_vertex_attr(vinfo_vbuf, EMIT_4F, INTERP_PERSPECTIVE, i); - } - draw_compute_vertex_size(vinfo_vbuf); + /* Tell draw_vbuf to simply emit the whole post-xform vertex + * as-is. No longer any need to try and emit draw vertex_header + * info. + */ + vinfo_vbuf->num_attribs = 0; + for (i = 0; i < num; i++) { + draw_emit_vertex_attr(vinfo_vbuf, EMIT_4F, INTERP_PERSPECTIVE, i); } + draw_compute_vertex_size(vinfo_vbuf); /* * Loop over fragment shader inputs, searching for the matching output @@ -91,6 +89,23 @@ softpipe_get_vertex_info(struct softpipe_context *softpipe) vinfo->num_attribs = 0; for (i = 0; i < spfs->info.num_inputs; i++) { int src; + enum interp_mode interp; + + switch (spfs->info.input_interpolate[i]) { + case TGSI_INTERPOLATE_CONSTANT: + interp = INTERP_CONSTANT; + break; + case TGSI_INTERPOLATE_LINEAR: + interp = INTERP_LINEAR; + break; + case TGSI_INTERPOLATE_PERSPECTIVE: + interp = INTERP_PERSPECTIVE; + break; + default: + assert(0); + interp = INTERP_LINEAR; + } + switch (spfs->info.input_semantic_name[i]) { case TGSI_SEMANTIC_POSITION: src = draw_find_vs_output(softpipe->draw, @@ -106,7 +121,7 @@ softpipe_get_vertex_info(struct softpipe_context *softpipe) case TGSI_SEMANTIC_FOG: src = draw_find_vs_output(softpipe->draw, TGSI_SEMANTIC_FOG, 0); - draw_emit_vertex_attr(vinfo, EMIT_4F, INTERP_PERSPECTIVE, src); + draw_emit_vertex_attr(vinfo, EMIT_4F, interp, src); break; case TGSI_SEMANTIC_GENERIC: @@ -114,7 +129,7 @@ softpipe_get_vertex_info(struct softpipe_context *softpipe) /* this includes texcoords and varying vars */ src = draw_find_vs_output(softpipe->draw, TGSI_SEMANTIC_GENERIC, spfs->info.input_semantic_index[i]); - draw_emit_vertex_attr(vinfo, EMIT_4F, INTERP_PERSPECTIVE, src); + draw_emit_vertex_attr(vinfo, EMIT_4F, interp, src); break; default: @@ -164,11 +179,19 @@ softpipe_get_vbuf_vertex_info(struct softpipe_context *softpipe) static void compute_cliprect(struct softpipe_context *sp) { + /* SP_NEW_FRAMEBUFFER + */ uint surfWidth = sp->framebuffer.width; uint surfHeight = sp->framebuffer.height; + /* SP_NEW_RASTERIZER + */ if (sp->rasterizer->scissor) { - /* clip to scissor rect */ + + /* SP_NEW_SCISSOR + * + * clip to scissor rect: + */ sp->cliprect.minx = MAX2(sp->scissor.minx, 0); sp->cliprect.miny = MAX2(sp->scissor.miny, 0); sp->cliprect.maxx = MIN2(sp->scissor.maxx, surfWidth); @@ -184,27 +207,63 @@ compute_cliprect(struct softpipe_context *sp) } +static void +update_tgsi_samplers( struct softpipe_context *softpipe ) +{ + unsigned i; + + softpipe_reset_sampler_varients( softpipe ); + + for (i = 0; i < PIPE_MAX_SAMPLERS; i++) { + struct softpipe_tex_tile_cache *tc = softpipe->tex_cache[i]; + if (tc->texture) { + struct softpipe_texture *spt = softpipe_texture(tc->texture); + if (spt->timestamp != tc->timestamp) { + sp_tex_tile_cache_validate_texture( tc ); + /* + _debug_printf("INV %d %d\n", tc->timestamp, spt->timestamp); + */ + tc->timestamp = spt->timestamp; + } + } + } +} + + /* Hopefully this will remain quite simple, otherwise need to pull in * something like the state tracker mechanism. */ void softpipe_update_derived( struct softpipe_context *softpipe ) { + struct softpipe_screen *sp_screen = softpipe_screen(softpipe->pipe.screen); + + /* Check for updated textures. + */ + if (softpipe->tex_timestamp != sp_screen->timestamp) { + softpipe->tex_timestamp = sp_screen->timestamp; + softpipe->dirty |= SP_NEW_TEXTURE; + } + + if (softpipe->dirty & (SP_NEW_SAMPLER | + SP_NEW_TEXTURE | + SP_NEW_FS | + SP_NEW_VS)) + update_tgsi_samplers( softpipe ); + if (softpipe->dirty & (SP_NEW_RASTERIZER | SP_NEW_FS | SP_NEW_VS)) invalidate_vertex_layout( softpipe ); if (softpipe->dirty & (SP_NEW_SCISSOR | - SP_NEW_DEPTH_STENCIL_ALPHA | + SP_NEW_RASTERIZER | SP_NEW_FRAMEBUFFER)) compute_cliprect(softpipe); if (softpipe->dirty & (SP_NEW_BLEND | SP_NEW_DEPTH_STENCIL_ALPHA | SP_NEW_FRAMEBUFFER | - SP_NEW_RASTERIZER | - SP_NEW_FS | - SP_NEW_QUERY)) + SP_NEW_FS)) sp_build_quad_pipeline(softpipe); softpipe->dirty = 0; diff --git a/src/gallium/drivers/softpipe/sp_state_fs.c b/src/gallium/drivers/softpipe/sp_state_fs.c index 4330c203935..256faa94b84 100644 --- a/src/gallium/drivers/softpipe/sp_state_fs.c +++ b/src/gallium/drivers/softpipe/sp_state_fs.c @@ -31,9 +31,8 @@ #include "pipe/p_defines.h" #include "util/u_memory.h" -#include "pipe/internal/p_winsys_screen.h" -#include "pipe/p_shader_tokens.h" #include "draw/draw_context.h" +#include "draw/draw_vs.h" #include "tgsi/tgsi_dump.h" #include "tgsi/tgsi_scan.h" #include "tgsi/tgsi_parse.h" @@ -51,12 +50,9 @@ softpipe_create_fs_state(struct pipe_context *pipe, tgsi_dump(templ->tokens, 0); /* codegen */ - state = softpipe_create_fs_llvm( softpipe, templ ); + state = softpipe_create_fs_sse( softpipe, templ ); if (!state) { - state = softpipe_create_fs_sse( softpipe, templ ); - if (!state) { - state = softpipe_create_fs_exec( softpipe, templ ); - } + state = softpipe_create_fs_exec( softpipe, templ ); } assert(state); @@ -111,6 +107,8 @@ softpipe_create_vs_state(struct pipe_context *pipe, if (state->draw_data == NULL) goto fail; + state->max_sampler = state->draw_data->info.file_max[TGSI_FILE_SAMPLER]; + return state; fail: @@ -128,7 +126,7 @@ softpipe_bind_vs_state(struct pipe_context *pipe, void *vs) { struct softpipe_context *softpipe = softpipe_context(pipe); - softpipe->vs = (const struct sp_vertex_shader *)vs; + softpipe->vs = (struct sp_vertex_shader *) vs; draw_bind_vertex_shader(softpipe->draw, (softpipe->vs ? softpipe->vs->draw_data : NULL)); @@ -142,8 +140,7 @@ softpipe_delete_vs_state(struct pipe_context *pipe, void *vs) { struct softpipe_context *softpipe = softpipe_context(pipe); - struct sp_vertex_shader *state = - (struct sp_vertex_shader *)vs; + struct sp_vertex_shader *state = (struct sp_vertex_shader *) vs; draw_delete_vertex_shader(softpipe->draw, state->draw_data); FREE( state ); diff --git a/src/gallium/drivers/softpipe/sp_state_sampler.c b/src/gallium/drivers/softpipe/sp_state_sampler.c index cb517b02e44..db0b8ab76b1 100644 --- a/src/gallium/drivers/softpipe/sp_state_sampler.c +++ b/src/gallium/drivers/softpipe/sp_state_sampler.c @@ -32,21 +32,37 @@ #include "util/u_memory.h" #include "draw/draw_context.h" +#include "draw/draw_context.h" #include "sp_context.h" -#include "sp_context.h" #include "sp_state.h" #include "sp_texture.h" -#include "sp_tile_cache.h" -#include "draw/draw_context.h" +#include "sp_tex_sample.h" +#include "sp_tex_tile_cache.h" +struct sp_sampler { + struct pipe_sampler_state base; + struct sp_sampler_varient *varients; + struct sp_sampler_varient *current; +}; + +static struct sp_sampler *sp_sampler( struct pipe_sampler_state *sampler ) +{ + return (struct sp_sampler *)sampler; +} + void * softpipe_create_sampler_state(struct pipe_context *pipe, const struct pipe_sampler_state *sampler) { - return mem_dup(sampler, sizeof(*sampler)); + struct sp_sampler *sp_sampler = CALLOC_STRUCT(sp_sampler); + + sp_sampler->base = *sampler; + sp_sampler->varients = NULL; + + return (void *)sp_sampler; } @@ -97,7 +113,7 @@ softpipe_set_sampler_textures(struct pipe_context *pipe, struct pipe_texture *tex = i < num ? texture[i] : NULL; pipe_texture_reference(&softpipe->texture[i], tex); - sp_tile_cache_set_texture(pipe, softpipe->tex_cache[i], tex); + sp_tex_tile_cache_set_texture(softpipe->tex_cache[i], tex); } softpipe->num_textures = num; @@ -106,10 +122,111 @@ softpipe_set_sampler_textures(struct pipe_context *pipe, } +/** + * Find/create an sp_sampler_varient object for sampling the given texture, + * sampler and tex unit. + * + * Note that the tex unit is significant. We can't re-use a sampler + * varient for multiple texture units because the sampler varient contains + * the texture object pointer. If the texture object pointer were stored + * somewhere outside the sampler varient, we could re-use samplers for + * multiple texture units. + */ +static struct sp_sampler_varient * +get_sampler_varient( unsigned unit, + struct sp_sampler *sampler, + struct pipe_texture *texture, + unsigned processor ) +{ + struct softpipe_texture *sp_texture = softpipe_texture(texture); + struct sp_sampler_varient *v = NULL; + union sp_sampler_key key; + + /* if this fails, widen the key.unit field and update this assertion */ + assert(PIPE_MAX_SAMPLERS <= 16); + + key.bits.target = sp_texture->base.target; + key.bits.is_pot = sp_texture->pot; + key.bits.processor = processor; + key.bits.unit = unit; + key.bits.pad = 0; + + if (sampler->current && + key.value == sampler->current->key.value) { + v = sampler->current; + } + + if (v == NULL) { + for (v = sampler->varients; v; v = v->next) + if (v->key.value == key.value) + break; + + if (v == NULL) { + v = sp_create_sampler_varient( &sampler->base, key ); + v->next = sampler->varients; + sampler->varients = v; + } + } + + sampler->current = v; + return v; +} + + + + +void +softpipe_reset_sampler_varients(struct softpipe_context *softpipe) +{ + int i; + + /* It's a bit hard to build these samplers ahead of time -- don't + * really know which samplers are going to be used for vertex and + * fragment programs. + */ + for (i = 0; i <= softpipe->vs->max_sampler; i++) { + if (softpipe->sampler[i]) { + softpipe->tgsi.vert_samplers_list[i] = + get_sampler_varient( i, + sp_sampler(softpipe->sampler[i]), + softpipe->texture[i], + TGSI_PROCESSOR_VERTEX ); + + sp_sampler_varient_bind_texture( softpipe->tgsi.vert_samplers_list[i], + softpipe->tex_cache[i], + softpipe->texture[i] ); + } + } + + for (i = 0; i <= softpipe->fs->info.file_max[TGSI_FILE_SAMPLER]; i++) { + if (softpipe->sampler[i]) { + softpipe->tgsi.frag_samplers_list[i] = + get_sampler_varient( i, + sp_sampler(softpipe->sampler[i]), + softpipe->texture[i], + TGSI_PROCESSOR_FRAGMENT ); + + sp_sampler_varient_bind_texture( softpipe->tgsi.frag_samplers_list[i], + softpipe->tex_cache[i], + softpipe->texture[i] ); + } + } +} + + + void softpipe_delete_sampler_state(struct pipe_context *pipe, void *sampler) { + struct sp_sampler *sp_sampler = (struct sp_sampler *)sampler; + struct sp_sampler_varient *v, *tmp; + + for (v = sp_sampler->varients; v; v = tmp) { + tmp = v->next; + sp_sampler_varient_destroy(v); + } + FREE( sampler ); } diff --git a/src/gallium/drivers/softpipe/sp_state_surface.c b/src/gallium/drivers/softpipe/sp_state_surface.c index 7c06d864a75..c8f55c3cec4 100644 --- a/src/gallium/drivers/softpipe/sp_state_surface.c +++ b/src/gallium/drivers/softpipe/sp_state_surface.c @@ -53,7 +53,7 @@ softpipe_set_framebuffer_state(struct pipe_context *pipe, /* check if changing cbuf */ if (sp->framebuffer.cbufs[i] != fb->cbufs[i]) { /* flush old */ - sp_flush_tile_cache(sp, sp->cbuf_cache[i]); + sp_flush_tile_cache(sp->cbuf_cache[i]); /* assign new */ sp->framebuffer.cbufs[i] = fb->cbufs[i]; @@ -68,58 +68,28 @@ softpipe_set_framebuffer_state(struct pipe_context *pipe, /* zbuf changing? */ if (sp->framebuffer.zsbuf != fb->zsbuf) { /* flush old */ - sp_flush_tile_cache(sp, sp->zsbuf_cache); + sp_flush_tile_cache(sp->zsbuf_cache); /* assign new */ sp->framebuffer.zsbuf = fb->zsbuf; /* update cache */ sp_tile_cache_set_surface(sp->zsbuf_cache, fb->zsbuf); - } - -#if 0 - /* XXX combined depth/stencil here */ - - /* sbuf changing? */ - if (sp->framebuffer.sbuf != fb->sbuf) { - /* flush old */ - sp_flush_tile_cache(sp, sp->sbuf_cache_sep); - - /* assign new */ - sp->framebuffer.sbuf = fb->sbuf; - - /* update cache */ - if (fb->sbuf != fb->zbuf) { - /* separate stencil buf */ - sp->sbuf_cache = sp->sbuf_cache_sep; - sp_tile_cache_set_surface(sp->sbuf_cache, fb->sbuf); - } - else { - /* combined depth/stencil */ - sp->sbuf_cache = sp->zbuf_cache; - sp_tile_cache_set_surface(sp->sbuf_cache, fb->sbuf); - } - } -#endif - /* Tell draw module how deep the Z/depth buffer is */ - { - int depth_bits; - double mrd; + /* Tell draw module how deep the Z/depth buffer is */ if (sp->framebuffer.zsbuf) { + int depth_bits; + double mrd; depth_bits = pf_get_component_bits(sp->framebuffer.zsbuf->format, PIPE_FORMAT_COMP_Z); + if (depth_bits > 16) { + mrd = 0.0000001; + } + else { + mrd = 0.00002; + } + draw_set_mrd(sp->draw, mrd); } - else { - depth_bits = 0; - } - if (depth_bits > 16) { - mrd = 0.0000001; - } - else { - mrd = 0.00002; - } - draw_set_mrd(sp->draw, mrd); } sp->framebuffer.width = fb->width; diff --git a/src/gallium/drivers/softpipe/sp_tex_sample.c b/src/gallium/drivers/softpipe/sp_tex_sample.c index a1d3bade27a..c22ee86b66c 100644 --- a/src/gallium/drivers/softpipe/sp_tex_sample.c +++ b/src/gallium/drivers/softpipe/sp_tex_sample.c @@ -31,29 +31,33 @@ * * Authors: * Brian Paul + * Keith Whitwell */ -#include "sp_context.h" -#include "sp_quad.h" -#include "sp_surface.h" -#include "sp_texture.h" -#include "sp_tex_sample.h" -#include "sp_tile_cache.h" #include "pipe/p_context.h" #include "pipe/p_defines.h" +#include "pipe/p_shader_tokens.h" #include "util/u_math.h" #include "util/u_memory.h" +#include "sp_quad.h" /* only for #define QUAD_* tokens */ +#include "sp_tex_sample.h" +#include "sp_tex_tile_cache.h" /* - * Note, the FRAC macro has to work perfectly. Otherwise you'll sometimes - * see 1-pixel bands of improperly weighted linear-filtered textures. + * Return fractional part of 'f'. Used for computing interpolation weights. + * Need to be careful with negative values. + * Note, if this function isn't perfect you'll sometimes see 1-pixel bands + * of improperly weighted linear-filtered textures. * The tests/texwrap.c demo is a good test. - * Also note, FRAC(x) doesn't truly return the fractional part of x for x < 0. - * Instead, if x < 0 then FRAC(x) = 1 - true_frac(x). */ -#define FRAC(f) ((f) - util_ifloor(f)) +static INLINE float +frac(float f) +{ + return f - util_ifloor(f); +} + /** @@ -100,10 +104,16 @@ lerp_3d(float a, float b, float c, /** - * If A is a signed integer, A % B doesn't give the right value for A < 0 - * (in terms of texture repeat). Just casting to unsigned fixes that. + * Compute coord % size for repeat wrap modes. + * Note that if coord is a signed integer, coord % size doesn't give + * the right value for coord < 0 (in terms of texture repeat). Just + * casting to unsigned fixes that. */ -#define REMAINDER(A, B) ((unsigned) (A) % (unsigned) (B)) +static INLINE int +repeat(int coord, unsigned size) +{ + return (int) ((unsigned) coord % size); +} /** @@ -115,133 +125,153 @@ lerp_3d(float a, float b, float c, * \param icoord returns the integer texcoords * \return integer texture index */ -static INLINE void -nearest_texcoord_4(unsigned wrapMode, const float s[4], unsigned size, - int icoord[4]) +static void +wrap_nearest_repeat(const float s[4], unsigned size, int icoord[4]) { uint ch; - switch (wrapMode) { - case PIPE_TEX_WRAP_REPEAT: - /* s limited to [0,1) */ - /* i limited to [0,size-1] */ - for (ch = 0; ch < 4; ch++) { - int i = util_ifloor(s[ch] * size); - icoord[ch] = REMAINDER(i, size); - } - return; - case PIPE_TEX_WRAP_CLAMP: + /* s limited to [0,1) */ + /* i limited to [0,size-1] */ + for (ch = 0; ch < 4; ch++) { + int i = util_ifloor(s[ch] * size); + icoord[ch] = repeat(i, size); + } +} + + +static void +wrap_nearest_clamp(const float s[4], unsigned size, int icoord[4]) +{ + uint ch; + /* s limited to [0,1] */ + /* i limited to [0,size-1] */ + for (ch = 0; ch < 4; ch++) { + if (s[ch] <= 0.0F) + icoord[ch] = 0; + else if (s[ch] >= 1.0F) + icoord[ch] = size - 1; + else + icoord[ch] = util_ifloor(s[ch] * size); + } +} + + +static void +wrap_nearest_clamp_to_edge(const float s[4], unsigned size, int icoord[4]) +{ + uint ch; + /* s limited to [min,max] */ + /* i limited to [0, size-1] */ + const float min = 1.0F / (2.0F * size); + const float max = 1.0F - min; + for (ch = 0; ch < 4; ch++) { + if (s[ch] < min) + icoord[ch] = 0; + else if (s[ch] > max) + icoord[ch] = size - 1; + else + icoord[ch] = util_ifloor(s[ch] * size); + } +} + + +static void +wrap_nearest_clamp_to_border(const float s[4], unsigned size, int icoord[4]) +{ + uint ch; + /* s limited to [min,max] */ + /* i limited to [-1, size] */ + const float min = -1.0F / (2.0F * size); + const float max = 1.0F - min; + for (ch = 0; ch < 4; ch++) { + if (s[ch] <= min) + icoord[ch] = -1; + else if (s[ch] >= max) + icoord[ch] = size; + else + icoord[ch] = util_ifloor(s[ch] * size); + } +} + + +static void +wrap_nearest_mirror_repeat(const float s[4], unsigned size, int icoord[4]) +{ + uint ch; + const float min = 1.0F / (2.0F * size); + const float max = 1.0F - min; + for (ch = 0; ch < 4; ch++) { + const int flr = util_ifloor(s[ch]); + float u; + if (flr & 1) + u = 1.0F - (s[ch] - (float) flr); + else + u = s[ch] - (float) flr; + if (u < min) + icoord[ch] = 0; + else if (u > max) + icoord[ch] = size - 1; + else + icoord[ch] = util_ifloor(u * size); + } +} + + +static void +wrap_nearest_mirror_clamp(const float s[4], unsigned size, int icoord[4]) +{ + uint ch; + for (ch = 0; ch < 4; ch++) { /* s limited to [0,1] */ /* i limited to [0,size-1] */ - for (ch = 0; ch < 4; ch++) { - if (s[ch] <= 0.0F) - icoord[ch] = 0; - else if (s[ch] >= 1.0F) - icoord[ch] = size - 1; - else - icoord[ch] = util_ifloor(s[ch] * size); - } - return; - case PIPE_TEX_WRAP_CLAMP_TO_EDGE: - { - /* s limited to [min,max] */ - /* i limited to [0, size-1] */ - const float min = 1.0F / (2.0F * size); - const float max = 1.0F - min; - for (ch = 0; ch < 4; ch++) { - if (s[ch] < min) - icoord[ch] = 0; - else if (s[ch] > max) - icoord[ch] = size - 1; - else - icoord[ch] = util_ifloor(s[ch] * size); - } - } - return; - case PIPE_TEX_WRAP_CLAMP_TO_BORDER: - { - /* s limited to [min,max] */ - /* i limited to [-1, size] */ - const float min = -1.0F / (2.0F * size); - const float max = 1.0F - min; - for (ch = 0; ch < 4; ch++) { - if (s[ch] <= min) - icoord[ch] = -1; - else if (s[ch] >= max) - icoord[ch] = size; - else - icoord[ch] = util_ifloor(s[ch] * size); - } - } - return; - case PIPE_TEX_WRAP_MIRROR_REPEAT: - { - const float min = 1.0F / (2.0F * size); - const float max = 1.0F - min; - for (ch = 0; ch < 4; ch++) { - const int flr = util_ifloor(s[ch]); - float u; - if (flr & 1) - u = 1.0F - (s[ch] - (float) flr); - else - u = s[ch] - (float) flr; - if (u < min) - icoord[ch] = 0; - else if (u > max) - icoord[ch] = size - 1; - else - icoord[ch] = util_ifloor(u * size); - } - } - return; - case PIPE_TEX_WRAP_MIRROR_CLAMP: - for (ch = 0; ch < 4; ch++) { - /* s limited to [0,1] */ - /* i limited to [0,size-1] */ - const float u = fabsf(s[ch]); - if (u <= 0.0F) - icoord[ch] = 0; - else if (u >= 1.0F) - icoord[ch] = size - 1; - else - icoord[ch] = util_ifloor(u * size); - } - return; - case PIPE_TEX_WRAP_MIRROR_CLAMP_TO_EDGE: - { - /* s limited to [min,max] */ - /* i limited to [0, size-1] */ - const float min = 1.0F / (2.0F * size); - const float max = 1.0F - min; - for (ch = 0; ch < 4; ch++) { - const float u = fabsf(s[ch]); - if (u < min) - icoord[ch] = 0; - else if (u > max) - icoord[ch] = size - 1; - else - icoord[ch] = util_ifloor(u * size); - } - } - return; - case PIPE_TEX_WRAP_MIRROR_CLAMP_TO_BORDER: - { - /* s limited to [min,max] */ - /* i limited to [0, size-1] */ - const float min = -1.0F / (2.0F * size); - const float max = 1.0F - min; - for (ch = 0; ch < 4; ch++) { - const float u = fabsf(s[ch]); - if (u < min) - icoord[ch] = -1; - else if (u > max) - icoord[ch] = size; - else - icoord[ch] = util_ifloor(u * size); - } - } - return; - default: - assert(0); + const float u = fabsf(s[ch]); + if (u <= 0.0F) + icoord[ch] = 0; + else if (u >= 1.0F) + icoord[ch] = size - 1; + else + icoord[ch] = util_ifloor(u * size); + } +} + + +static void +wrap_nearest_mirror_clamp_to_edge(const float s[4], unsigned size, + int icoord[4]) +{ + uint ch; + /* s limited to [min,max] */ + /* i limited to [0, size-1] */ + const float min = 1.0F / (2.0F * size); + const float max = 1.0F - min; + for (ch = 0; ch < 4; ch++) { + const float u = fabsf(s[ch]); + if (u < min) + icoord[ch] = 0; + else if (u > max) + icoord[ch] = size - 1; + else + icoord[ch] = util_ifloor(u * size); + } +} + + +static void +wrap_nearest_mirror_clamp_to_border(const float s[4], unsigned size, + int icoord[4]) +{ + uint ch; + /* s limited to [min,max] */ + /* i limited to [0, size-1] */ + const float min = -1.0F / (2.0F * size); + const float max = 1.0F - min; + for (ch = 0; ch < 4; ch++) { + const float u = fabsf(s[ch]); + if (u < min) + icoord[ch] = -1; + else if (u > max) + icoord[ch] = size; + else + icoord[ch] = util_ifloor(u * size); } } @@ -256,125 +286,156 @@ nearest_texcoord_4(unsigned wrapMode, const float s[4], unsigned size, * \param w returns blend factor/weight between texture indexes * \param icoord returns the computed integer texture coords */ -static INLINE void -linear_texcoord_4(unsigned wrapMode, const float s[4], unsigned size, +static void +wrap_linear_repeat(const float s[4], unsigned size, + int icoord0[4], int icoord1[4], float w[4]) +{ + uint ch; + for (ch = 0; ch < 4; ch++) { + float u = s[ch] * size - 0.5F; + icoord0[ch] = repeat(util_ifloor(u), size); + icoord1[ch] = repeat(icoord0[ch] + 1, size); + w[ch] = frac(u); + } +} + + +static void +wrap_linear_clamp(const float s[4], unsigned size, int icoord0[4], int icoord1[4], float w[4]) { uint ch; + for (ch = 0; ch < 4; ch++) { + float u = CLAMP(s[ch], 0.0F, 1.0F); + u = u * size - 0.5f; + icoord0[ch] = util_ifloor(u); + icoord1[ch] = icoord0[ch] + 1; + w[ch] = frac(u); + } +} - switch (wrapMode) { - case PIPE_TEX_WRAP_REPEAT: - for (ch = 0; ch < 4; ch++) { - float u = s[ch] * size - 0.5F; - icoord0[ch] = REMAINDER(util_ifloor(u), size); - icoord1[ch] = REMAINDER(icoord0[ch] + 1, size); - w[ch] = FRAC(u); - } - break;; - case PIPE_TEX_WRAP_CLAMP: - for (ch = 0; ch < 4; ch++) { - float u = CLAMP(s[ch], 0.0F, 1.0F); - u = u * size - 0.5f; - icoord0[ch] = util_ifloor(u); - icoord1[ch] = icoord0[ch] + 1; - w[ch] = FRAC(u); - } - break;; - case PIPE_TEX_WRAP_CLAMP_TO_EDGE: - for (ch = 0; ch < 4; ch++) { - float u = CLAMP(s[ch], 0.0F, 1.0F); - u = u * size - 0.5f; - icoord0[ch] = util_ifloor(u); - icoord1[ch] = icoord0[ch] + 1; - if (icoord0[ch] < 0) - icoord0[ch] = 0; - if (icoord1[ch] >= (int) size) - icoord1[ch] = size - 1; - w[ch] = FRAC(u); - } - break;; - case PIPE_TEX_WRAP_CLAMP_TO_BORDER: - { - const float min = -1.0F / (2.0F * size); - const float max = 1.0F - min; - for (ch = 0; ch < 4; ch++) { - float u = CLAMP(s[ch], min, max); - u = u * size - 0.5f; - icoord0[ch] = util_ifloor(u); - icoord1[ch] = icoord0[ch] + 1; - w[ch] = FRAC(u); - } - } - break;; - case PIPE_TEX_WRAP_MIRROR_REPEAT: - for (ch = 0; ch < 4; ch++) { - const int flr = util_ifloor(s[ch]); - float u; - if (flr & 1) - u = 1.0F - (s[ch] - (float) flr); - else - u = s[ch] - (float) flr; - u = u * size - 0.5F; - icoord0[ch] = util_ifloor(u); - icoord1[ch] = icoord0[ch] + 1; - if (icoord0[ch] < 0) - icoord0[ch] = 0; - if (icoord1[ch] >= (int) size) - icoord1[ch] = size - 1; - w[ch] = FRAC(u); - } - break;; - case PIPE_TEX_WRAP_MIRROR_CLAMP: - for (ch = 0; ch < 4; ch++) { - float u = fabsf(s[ch]); - if (u >= 1.0F) - u = (float) size; - else - u *= size; - u -= 0.5F; - icoord0[ch] = util_ifloor(u); - icoord1[ch] = icoord0[ch] + 1; - w[ch] = FRAC(u); - } - break;; - case PIPE_TEX_WRAP_MIRROR_CLAMP_TO_EDGE: - for (ch = 0; ch < 4; ch++) { - float u = fabsf(s[ch]); - if (u >= 1.0F) - u = (float) size; - else - u *= size; - u -= 0.5F; - icoord0[ch] = util_ifloor(u); - icoord1[ch] = icoord0[ch] + 1; - if (icoord0[ch] < 0) - icoord0[ch] = 0; - if (icoord1[ch] >= (int) size) - icoord1[ch] = size - 1; - w[ch] = FRAC(u); - } - break;; - case PIPE_TEX_WRAP_MIRROR_CLAMP_TO_BORDER: - { - const float min = -1.0F / (2.0F * size); - const float max = 1.0F - min; - for (ch = 0; ch < 4; ch++) { - float u = fabsf(s[ch]); - if (u <= min) - u = min * size; - else if (u >= max) - u = max * size; - else - u *= size; - u -= 0.5F; - icoord0[ch] = util_ifloor(u); - icoord1[ch] = icoord0[ch] + 1; - w[ch] = FRAC(u); - } - } - break;; - default: - assert(0); + +static void +wrap_linear_clamp_to_edge(const float s[4], unsigned size, + int icoord0[4], int icoord1[4], float w[4]) +{ + uint ch; + for (ch = 0; ch < 4; ch++) { + float u = CLAMP(s[ch], 0.0F, 1.0F); + u = u * size - 0.5f; + icoord0[ch] = util_ifloor(u); + icoord1[ch] = icoord0[ch] + 1; + if (icoord0[ch] < 0) + icoord0[ch] = 0; + if (icoord1[ch] >= (int) size) + icoord1[ch] = size - 1; + w[ch] = frac(u); + } +} + + +static void +wrap_linear_clamp_to_border(const float s[4], unsigned size, + int icoord0[4], int icoord1[4], float w[4]) +{ + const float min = -1.0F / (2.0F * size); + const float max = 1.0F - min; + uint ch; + for (ch = 0; ch < 4; ch++) { + float u = CLAMP(s[ch], min, max); + u = u * size - 0.5f; + icoord0[ch] = util_ifloor(u); + icoord1[ch] = icoord0[ch] + 1; + w[ch] = frac(u); + } +} + + +static void +wrap_linear_mirror_repeat(const float s[4], unsigned size, + int icoord0[4], int icoord1[4], float w[4]) +{ + uint ch; + for (ch = 0; ch < 4; ch++) { + const int flr = util_ifloor(s[ch]); + float u; + if (flr & 1) + u = 1.0F - (s[ch] - (float) flr); + else + u = s[ch] - (float) flr; + u = u * size - 0.5F; + icoord0[ch] = util_ifloor(u); + icoord1[ch] = icoord0[ch] + 1; + if (icoord0[ch] < 0) + icoord0[ch] = 0; + if (icoord1[ch] >= (int) size) + icoord1[ch] = size - 1; + w[ch] = frac(u); + } +} + + +static void +wrap_linear_mirror_clamp(const float s[4], unsigned size, + int icoord0[4], int icoord1[4], float w[4]) +{ + uint ch; + for (ch = 0; ch < 4; ch++) { + float u = fabsf(s[ch]); + if (u >= 1.0F) + u = (float) size; + else + u *= size; + u -= 0.5F; + icoord0[ch] = util_ifloor(u); + icoord1[ch] = icoord0[ch] + 1; + w[ch] = frac(u); + } +} + + +static void +wrap_linear_mirror_clamp_to_edge(const float s[4], unsigned size, + int icoord0[4], int icoord1[4], float w[4]) +{ + uint ch; + for (ch = 0; ch < 4; ch++) { + float u = fabsf(s[ch]); + if (u >= 1.0F) + u = (float) size; + else + u *= size; + u -= 0.5F; + icoord0[ch] = util_ifloor(u); + icoord1[ch] = icoord0[ch] + 1; + if (icoord0[ch] < 0) + icoord0[ch] = 0; + if (icoord1[ch] >= (int) size) + icoord1[ch] = size - 1; + w[ch] = frac(u); + } +} + + +static void +wrap_linear_mirror_clamp_to_border(const float s[4], unsigned size, + int icoord0[4], int icoord1[4], float w[4]) +{ + const float min = -1.0F / (2.0F * size); + const float max = 1.0F - min; + uint ch; + for (ch = 0; ch < 4; ch++) { + float u = fabsf(s[ch]); + if (u <= min) + u = min * size; + else if (u >= max) + u = max * size; + else + u *= size; + u -= 0.5F; + icoord0[ch] = util_ifloor(u); + icoord1[ch] = icoord0[ch] + 1; + w[ch] = frac(u); } } @@ -383,27 +444,27 @@ linear_texcoord_4(unsigned wrapMode, const float s[4], unsigned size, * For RECT textures / unnormalized texcoords * Only a subset of wrap modes supported. */ -static INLINE void -nearest_texcoord_unnorm_4(unsigned wrapMode, const float s[4], unsigned size, - int icoord[4]) +static void +wrap_nearest_unorm_clamp(const float s[4], unsigned size, int icoord[4]) { uint ch; - switch (wrapMode) { - case PIPE_TEX_WRAP_CLAMP: - for (ch = 0; ch < 4; ch++) { - int i = util_ifloor(s[ch]); - icoord[ch]= CLAMP(i, 0, (int) size-1); - } - return; - case PIPE_TEX_WRAP_CLAMP_TO_EDGE: - /* fall-through */ - case PIPE_TEX_WRAP_CLAMP_TO_BORDER: - for (ch = 0; ch < 4; ch++) { - icoord[ch]= util_ifloor( CLAMP(s[ch], 0.5F, (float) size - 0.5F) ); - } - return; - default: - assert(0); + for (ch = 0; ch < 4; ch++) { + int i = util_ifloor(s[ch]); + icoord[ch]= CLAMP(i, 0, (int) size-1); + } +} + + +/** + * Handles clamp_to_edge and clamp_to_border: + */ +static void +wrap_nearest_unorm_clamp_to_border(const float s[4], unsigned size, + int icoord[4]) +{ + uint ch; + for (ch = 0; ch < 4; ch++) { + icoord[ch]= util_ifloor( CLAMP(s[ch], 0.5F, (float) size - 0.5F) ); } } @@ -412,358 +473,971 @@ nearest_texcoord_unnorm_4(unsigned wrapMode, const float s[4], unsigned size, * For RECT textures / unnormalized texcoords. * Only a subset of wrap modes supported. */ -static INLINE void -linear_texcoord_unnorm_4(unsigned wrapMode, const float s[4], unsigned size, - int icoord0[4], int icoord1[4], float w[4]) +static void +wrap_linear_unorm_clamp(const float s[4], unsigned size, + int icoord0[4], int icoord1[4], float w[4]) { uint ch; - switch (wrapMode) { - case PIPE_TEX_WRAP_CLAMP: - for (ch = 0; ch < 4; ch++) { - /* Not exactly what the spec says, but it matches NVIDIA output */ - float u = CLAMP(s[ch] - 0.5F, 0.0f, (float) size - 1.0f); - icoord0[ch] = util_ifloor(u); - icoord1[ch] = icoord0[ch] + 1; - w[ch] = FRAC(u); - } - return; - case PIPE_TEX_WRAP_CLAMP_TO_EDGE: - /* fall-through */ - case PIPE_TEX_WRAP_CLAMP_TO_BORDER: - for (ch = 0; ch < 4; ch++) { - float u = CLAMP(s[ch], 0.5F, (float) size - 0.5F); - u -= 0.5F; - icoord0[ch] = util_ifloor(u); - icoord1[ch] = icoord0[ch] + 1; - if (icoord1[ch] > (int) size - 1) - icoord1[ch] = size - 1; - w[ch] = FRAC(u); - } - break; - default: - assert(0); + for (ch = 0; ch < 4; ch++) { + /* Not exactly what the spec says, but it matches NVIDIA output */ + float u = CLAMP(s[ch] - 0.5F, 0.0f, (float) size - 1.0f); + icoord0[ch] = util_ifloor(u); + icoord1[ch] = icoord0[ch] + 1; + w[ch] = frac(u); } } -static unsigned -choose_cube_face(float rx, float ry, float rz, float *newS, float *newT) +static void +wrap_linear_unorm_clamp_to_border(const float s[4], unsigned size, + int icoord0[4], int icoord1[4], float w[4]) { - /* - major axis - direction target sc tc ma - ---------- ------------------------------- --- --- --- - +rx TEXTURE_CUBE_MAP_POSITIVE_X_EXT -rz -ry rx - -rx TEXTURE_CUBE_MAP_NEGATIVE_X_EXT +rz -ry rx - +ry TEXTURE_CUBE_MAP_POSITIVE_Y_EXT +rx +rz ry - -ry TEXTURE_CUBE_MAP_NEGATIVE_Y_EXT +rx -rz ry - +rz TEXTURE_CUBE_MAP_POSITIVE_Z_EXT +rx -ry rz - -rz TEXTURE_CUBE_MAP_NEGATIVE_Z_EXT -rx -ry rz - */ - const float arx = fabsf(rx), ary = fabsf(ry), arz = fabsf(rz); - unsigned face; - float sc, tc, ma; - - if (arx > ary && arx > arz) { - if (rx >= 0.0F) { - face = PIPE_TEX_FACE_POS_X; - sc = -rz; - tc = -ry; - ma = arx; + uint ch; + for (ch = 0; ch < 4; ch++) { + float u = CLAMP(s[ch], 0.5F, (float) size - 0.5F); + u -= 0.5F; + icoord0[ch] = util_ifloor(u); + icoord1[ch] = icoord0[ch] + 1; + if (icoord1[ch] > (int) size - 1) + icoord1[ch] = size - 1; + w[ch] = frac(u); + } +} + + + +/** + * Examine the quad's texture coordinates to compute the partial + * derivatives w.r.t X and Y, then compute lambda (level of detail). + */ +static float +compute_lambda_1d(const struct sp_sampler_varient *samp, + const float s[QUAD_SIZE], + const float t[QUAD_SIZE], + const float p[QUAD_SIZE], + float lodbias) +{ + const struct pipe_texture *texture = samp->texture; + const struct pipe_sampler_state *sampler = samp->sampler; + float dsdx = fabsf(s[QUAD_BOTTOM_RIGHT] - s[QUAD_BOTTOM_LEFT]); + float dsdy = fabsf(s[QUAD_TOP_LEFT] - s[QUAD_BOTTOM_LEFT]); + float rho = MAX2(dsdx, dsdy) * texture->width[0]; + float lambda; + + lambda = util_fast_log2(rho); + lambda += lodbias + sampler->lod_bias; + lambda = CLAMP(lambda, sampler->min_lod, sampler->max_lod); + + return lambda; +} + + +static float +compute_lambda_2d(const struct sp_sampler_varient *samp, + const float s[QUAD_SIZE], + const float t[QUAD_SIZE], + const float p[QUAD_SIZE], + float lodbias) +{ + const struct pipe_texture *texture = samp->texture; + const struct pipe_sampler_state *sampler = samp->sampler; + float dsdx = fabsf(s[QUAD_BOTTOM_RIGHT] - s[QUAD_BOTTOM_LEFT]); + float dsdy = fabsf(s[QUAD_TOP_LEFT] - s[QUAD_BOTTOM_LEFT]); + float dtdx = fabsf(t[QUAD_BOTTOM_RIGHT] - t[QUAD_BOTTOM_LEFT]); + float dtdy = fabsf(t[QUAD_TOP_LEFT] - t[QUAD_BOTTOM_LEFT]); + float maxx = MAX2(dsdx, dsdy) * texture->width[0]; + float maxy = MAX2(dtdx, dtdy) * texture->height[0]; + float rho = MAX2(maxx, maxy); + float lambda; + + lambda = util_fast_log2(rho); + lambda += lodbias + sampler->lod_bias; + lambda = CLAMP(lambda, sampler->min_lod, sampler->max_lod); + + return lambda; +} + + +static float +compute_lambda_3d(const struct sp_sampler_varient *samp, + const float s[QUAD_SIZE], + const float t[QUAD_SIZE], + const float p[QUAD_SIZE], + float lodbias) +{ + const struct pipe_texture *texture = samp->texture; + const struct pipe_sampler_state *sampler = samp->sampler; + float dsdx = fabsf(s[QUAD_BOTTOM_RIGHT] - s[QUAD_BOTTOM_LEFT]); + float dsdy = fabsf(s[QUAD_TOP_LEFT] - s[QUAD_BOTTOM_LEFT]); + float dtdx = fabsf(t[QUAD_BOTTOM_RIGHT] - t[QUAD_BOTTOM_LEFT]); + float dtdy = fabsf(t[QUAD_TOP_LEFT] - t[QUAD_BOTTOM_LEFT]); + float dpdx = fabsf(p[QUAD_BOTTOM_RIGHT] - p[QUAD_BOTTOM_LEFT]); + float dpdy = fabsf(p[QUAD_TOP_LEFT] - p[QUAD_BOTTOM_LEFT]); + float maxx = MAX2(dsdx, dsdy) * texture->width[0]; + float maxy = MAX2(dtdx, dtdy) * texture->height[0]; + float maxz = MAX2(dpdx, dpdy) * texture->depth[0]; + float rho, lambda; + + rho = MAX2(maxx, maxy); + rho = MAX2(rho, maxz); + + lambda = util_fast_log2(rho); + lambda += lodbias + sampler->lod_bias; + lambda = CLAMP(lambda, sampler->min_lod, sampler->max_lod); + + return lambda; +} + + +/** + * Compute lambda for a vertex texture sampler. + * Since there aren't derivatives to use, just return the LOD bias. + */ +static float +compute_lambda_vert(const struct sp_sampler_varient *samp, + const float s[QUAD_SIZE], + const float t[QUAD_SIZE], + const float p[QUAD_SIZE], + float lodbias) +{ + return lodbias; +} + + + +/** + * Get a texel from a texture, using the texture tile cache. + * + * \param addr the template tex address containing cube, z, face info. + * \param x the x coord of texel within 2D image + * \param y the y coord of texel within 2D image + * \param rgba the quad to put the texel/color into + * + * XXX maybe move this into sp_tex_tile_cache.c and merge with the + * sp_get_cached_tile_tex() function. Also, get 4 texels instead of 1... + */ + + + + +static INLINE const float * +get_texel_2d_no_border(const struct sp_sampler_varient *samp, + union tex_tile_address addr, int x, int y) +{ + const struct softpipe_tex_cached_tile *tile; + + addr.bits.x = x / TILE_SIZE; + addr.bits.y = y / TILE_SIZE; + y %= TILE_SIZE; + x %= TILE_SIZE; + + tile = sp_get_cached_tile_tex(samp->cache, addr); + + return &tile->data.color[y][x][0]; +} + + +static INLINE const float * +get_texel_2d(const struct sp_sampler_varient *samp, + union tex_tile_address addr, int x, int y) +{ + const struct pipe_texture *texture = samp->texture; + unsigned level = addr.bits.level; + + if (x < 0 || x >= (int) texture->width[level] || + y < 0 || y >= (int) texture->height[level]) { + return samp->sampler->border_color; + } + else { + return get_texel_2d_no_border( samp, addr, x, y ); + } +} + + +/* Gather a quad of adjacent texels within a tile: + */ +static INLINE void +get_texel_quad_2d_no_border_single_tile(const struct sp_sampler_varient *samp, + union tex_tile_address addr, + unsigned x, unsigned y, + const float *out[4]) +{ + const struct softpipe_tex_cached_tile *tile; + + addr.bits.x = x / TILE_SIZE; + addr.bits.y = y / TILE_SIZE; + y %= TILE_SIZE; + x %= TILE_SIZE; + + tile = sp_get_cached_tile_tex(samp->cache, addr); + + out[0] = &tile->data.color[y ][x ][0]; + out[1] = &tile->data.color[y ][x+1][0]; + out[2] = &tile->data.color[y+1][x ][0]; + out[3] = &tile->data.color[y+1][x+1][0]; +} + + +/* Gather a quad of potentially non-adjacent texels: + */ +static INLINE void +get_texel_quad_2d_no_border(const struct sp_sampler_varient *samp, + union tex_tile_address addr, + int x0, int y0, + int x1, int y1, + const float *out[4]) +{ + out[0] = get_texel_2d_no_border( samp, addr, x0, y0 ); + out[1] = get_texel_2d_no_border( samp, addr, x1, y0 ); + out[2] = get_texel_2d_no_border( samp, addr, x0, y1 ); + out[3] = get_texel_2d_no_border( samp, addr, x1, y1 ); +} + +/* Can involve a lot of unnecessary checks for border color: + */ +static INLINE void +get_texel_quad_2d(const struct sp_sampler_varient *samp, + union tex_tile_address addr, + int x0, int y0, + int x1, int y1, + const float *out[4]) +{ + out[0] = get_texel_2d( samp, addr, x0, y0 ); + out[1] = get_texel_2d( samp, addr, x1, y0 ); + out[3] = get_texel_2d( samp, addr, x1, y1 ); + out[2] = get_texel_2d( samp, addr, x0, y1 ); +} + + + +/* 3d varients: + */ +static INLINE const float * +get_texel_3d_no_border(const struct sp_sampler_varient *samp, + union tex_tile_address addr, int x, int y, int z) +{ + const struct softpipe_tex_cached_tile *tile; + + addr.bits.x = x / TILE_SIZE; + addr.bits.y = y / TILE_SIZE; + addr.bits.z = z; + y %= TILE_SIZE; + x %= TILE_SIZE; + + tile = sp_get_cached_tile_tex(samp->cache, addr); + + return &tile->data.color[y][x][0]; +} + + +static INLINE const float * +get_texel_3d(const struct sp_sampler_varient *samp, + union tex_tile_address addr, int x, int y, int z) +{ + const struct pipe_texture *texture = samp->texture; + unsigned level = addr.bits.level; + + if (x < 0 || x >= (int) texture->width[level] || + y < 0 || y >= (int) texture->height[level] || + z < 0 || z >= (int) texture->depth[level]) { + return samp->sampler->border_color; + } + else { + return get_texel_3d_no_border( samp, addr, x, y, z ); + } +} + + +/** + * Given the logbase2 of a mipmap's base level size and a mipmap level, + * return the size (in texels) of that mipmap level. + * For example, if level[0].width = 256 then base_pot will be 8. + * If level = 2, then we'll return 64 (the width at level=2). + * Return 1 if level > base_pot. + */ +static INLINE unsigned +pot_level_size(unsigned base_pot, unsigned level) +{ + return (base_pot >= level) ? (1 << (base_pot - level)) : 1; +} + + +/* Some image-filter fastpaths: + */ +static INLINE void +img_filter_2d_linear_repeat_POT(struct tgsi_sampler *tgsi_sampler, + const float s[QUAD_SIZE], + const float t[QUAD_SIZE], + const float p[QUAD_SIZE], + float lodbias, + float rgba[NUM_CHANNELS][QUAD_SIZE]) +{ + const struct sp_sampler_varient *samp = sp_sampler_varient(tgsi_sampler); + unsigned j; + unsigned level = samp->level; + unsigned xpot = pot_level_size(samp->xpot, level); + unsigned ypot = pot_level_size(samp->ypot, level); + unsigned xmax = (xpot - 1) & (TILE_SIZE - 1); /* MIN2(TILE_SIZE, xpot) - 1; */ + unsigned ymax = (ypot - 1) & (TILE_SIZE - 1); /* MIN2(TILE_SIZE, ypot) - 1; */ + union tex_tile_address addr; + + addr.value = 0; + addr.bits.level = samp->level; + + for (j = 0; j < QUAD_SIZE; j++) { + int c; + + float u = s[j] * xpot - 0.5F; + float v = t[j] * ypot - 0.5F; + + int uflr = util_ifloor(u); + int vflr = util_ifloor(v); + + float xw = u - (float)uflr; + float yw = v - (float)vflr; + + int x0 = uflr & (xpot - 1); + int y0 = vflr & (ypot - 1); + + const float *tx[4]; + + /* Can we fetch all four at once: + */ + if (x0 < xmax && y0 < ymax) { + get_texel_quad_2d_no_border_single_tile(samp, addr, x0, y0, tx); } else { - face = PIPE_TEX_FACE_NEG_X; - sc = rz; - tc = -ry; - ma = arx; + unsigned x1 = (x0 + 1) & (xpot - 1); + unsigned y1 = (y0 + 1) & (ypot - 1); + get_texel_quad_2d_no_border(samp, addr, x0, y0, x1, y1, tx); + } + + /* interpolate R, G, B, A */ + for (c = 0; c < 4; c++) { + rgba[c][j] = lerp_2d(xw, yw, + tx[0][c], tx[1][c], + tx[2][c], tx[3][c]); } } - else if (ary > arx && ary > arz) { - if (ry >= 0.0F) { - face = PIPE_TEX_FACE_POS_Y; - sc = rx; - tc = rz; - ma = ary; +} + + +static INLINE void +img_filter_2d_nearest_repeat_POT(struct tgsi_sampler *tgsi_sampler, + const float s[QUAD_SIZE], + const float t[QUAD_SIZE], + const float p[QUAD_SIZE], + float lodbias, + float rgba[NUM_CHANNELS][QUAD_SIZE]) +{ + const struct sp_sampler_varient *samp = sp_sampler_varient(tgsi_sampler); + unsigned j; + unsigned level = samp->level; + unsigned xpot = pot_level_size(samp->xpot, level); + unsigned ypot = pot_level_size(samp->ypot, level); + union tex_tile_address addr; + + addr.value = 0; + addr.bits.level = samp->level; + + for (j = 0; j < QUAD_SIZE; j++) { + int c; + + float u = s[j] * xpot; + float v = t[j] * ypot; + + int uflr = util_ifloor(u); + int vflr = util_ifloor(v); + + int x0 = uflr & (xpot - 1); + int y0 = vflr & (ypot - 1); + + const float *out = get_texel_2d_no_border(samp, addr, x0, y0); + + for (c = 0; c < 4; c++) { + rgba[c][j] = out[c]; } - else { - face = PIPE_TEX_FACE_NEG_Y; - sc = rx; - tc = -rz; - ma = ary; + } +} + + +static INLINE void +img_filter_2d_nearest_clamp_POT(struct tgsi_sampler *tgsi_sampler, + const float s[QUAD_SIZE], + const float t[QUAD_SIZE], + const float p[QUAD_SIZE], + float lodbias, + float rgba[NUM_CHANNELS][QUAD_SIZE]) +{ + const struct sp_sampler_varient *samp = sp_sampler_varient(tgsi_sampler); + unsigned j; + unsigned level = samp->level; + unsigned xpot = pot_level_size(samp->xpot, level); + unsigned ypot = pot_level_size(samp->ypot, level); + union tex_tile_address addr; + + addr.value = 0; + addr.bits.level = samp->level; + + for (j = 0; j < QUAD_SIZE; j++) { + int c; + + float u = s[j] * xpot; + float v = t[j] * ypot; + + int x0, y0; + const float *out; + + x0 = util_ifloor(u); + if (x0 < 0) + x0 = 0; + else if (x0 > xpot - 1) + x0 = xpot - 1; + + y0 = util_ifloor(v); + if (y0 < 0) + y0 = 0; + else if (y0 > ypot - 1) + y0 = ypot - 1; + + out = get_texel_2d_no_border(samp, addr, x0, y0); + + for (c = 0; c < 4; c++) { + rgba[c][j] = out[c]; } } - else { - if (rz > 0.0F) { - face = PIPE_TEX_FACE_POS_Z; - sc = rx; - tc = -ry; - ma = arz; +} + + +static void +img_filter_1d_nearest(struct tgsi_sampler *tgsi_sampler, + const float s[QUAD_SIZE], + const float t[QUAD_SIZE], + const float p[QUAD_SIZE], + float lodbias, + float rgba[NUM_CHANNELS][QUAD_SIZE]) +{ + const struct sp_sampler_varient *samp = sp_sampler_varient(tgsi_sampler); + const struct pipe_texture *texture = samp->texture; + unsigned level0, j; + int width; + int x[4]; + union tex_tile_address addr; + + level0 = samp->level; + width = texture->width[level0]; + + assert(width > 0); + + addr.value = 0; + addr.bits.level = samp->level; + + samp->nearest_texcoord_s(s, width, x); + + for (j = 0; j < QUAD_SIZE; j++) { + const float *out = get_texel_2d(samp, addr, x[j], 0); + int c; + for (c = 0; c < 4; c++) { + rgba[c][j] = out[c]; } - else { - face = PIPE_TEX_FACE_NEG_Z; - sc = -rx; - tc = -ry; - ma = arz; + } +} + + +static void +img_filter_2d_nearest(struct tgsi_sampler *tgsi_sampler, + const float s[QUAD_SIZE], + const float t[QUAD_SIZE], + const float p[QUAD_SIZE], + float lodbias, + float rgba[NUM_CHANNELS][QUAD_SIZE]) +{ + const struct sp_sampler_varient *samp = sp_sampler_varient(tgsi_sampler); + const struct pipe_texture *texture = samp->texture; + unsigned level0, j; + int width, height; + int x[4], y[4]; + union tex_tile_address addr; + + + level0 = samp->level; + width = texture->width[level0]; + height = texture->height[level0]; + + assert(width > 0); + assert(height > 0); + + addr.value = 0; + addr.bits.level = samp->level; + + samp->nearest_texcoord_s(s, width, x); + samp->nearest_texcoord_t(t, height, y); + + for (j = 0; j < QUAD_SIZE; j++) { + const float *out = get_texel_2d(samp, addr, x[j], y[j]); + int c; + for (c = 0; c < 4; c++) { + rgba[c][j] = out[c]; } } +} - *newS = ( sc / ma + 1.0F ) * 0.5F; - *newT = ( tc / ma + 1.0F ) * 0.5F; - return face; +static INLINE union tex_tile_address +face(union tex_tile_address addr, unsigned face ) +{ + addr.bits.face = face; + return addr; } -/** - * Examine the quad's texture coordinates to compute the partial - * derivatives w.r.t X and Y, then compute lambda (level of detail). - * - * This is only done for fragment shaders, not vertex shaders. - */ -static float -compute_lambda(const struct pipe_texture *tex, - const struct pipe_sampler_state *sampler, - const float s[QUAD_SIZE], - const float t[QUAD_SIZE], - const float p[QUAD_SIZE], - float lodbias) +static void +img_filter_cube_nearest(struct tgsi_sampler *tgsi_sampler, + const float s[QUAD_SIZE], + const float t[QUAD_SIZE], + const float p[QUAD_SIZE], + float lodbias, + float rgba[NUM_CHANNELS][QUAD_SIZE]) { - float rho, lambda; + const struct sp_sampler_varient *samp = sp_sampler_varient(tgsi_sampler); + const struct pipe_texture *texture = samp->texture; + const unsigned *faces = samp->faces; /* zero when not cube-mapping */ + unsigned level0, j; + int width, height; + int x[4], y[4]; + union tex_tile_address addr; - assert(sampler->normalized_coords); + level0 = samp->level; + width = texture->width[level0]; + height = texture->height[level0]; - assert(s); - { - float dsdx = s[QUAD_BOTTOM_RIGHT] - s[QUAD_BOTTOM_LEFT]; - float dsdy = s[QUAD_TOP_LEFT] - s[QUAD_BOTTOM_LEFT]; - dsdx = fabsf(dsdx); - dsdy = fabsf(dsdy); - rho = MAX2(dsdx, dsdy) * tex->width[0]; - } - if (t) { - float dtdx = t[QUAD_BOTTOM_RIGHT] - t[QUAD_BOTTOM_LEFT]; - float dtdy = t[QUAD_TOP_LEFT] - t[QUAD_BOTTOM_LEFT]; - float max; - dtdx = fabsf(dtdx); - dtdy = fabsf(dtdy); - max = MAX2(dtdx, dtdy) * tex->height[0]; - rho = MAX2(rho, max); + assert(width > 0); + assert(height > 0); + + addr.value = 0; + addr.bits.level = samp->level; + + samp->nearest_texcoord_s(s, width, x); + samp->nearest_texcoord_t(t, height, y); + + for (j = 0; j < QUAD_SIZE; j++) { + const float *out = get_texel_2d(samp, face(addr, faces[j]), x[j], y[j]); + int c; + for (c = 0; c < 4; c++) { + rgba[c][j] = out[c]; + } } - if (p) { - float dpdx = p[QUAD_BOTTOM_RIGHT] - p[QUAD_BOTTOM_LEFT]; - float dpdy = p[QUAD_TOP_LEFT] - p[QUAD_BOTTOM_LEFT]; - float max; - dpdx = fabsf(dpdx); - dpdy = fabsf(dpdy); - max = MAX2(dpdx, dpdy) * tex->depth[0]; - rho = MAX2(rho, max); +} + + +static void +img_filter_3d_nearest(struct tgsi_sampler *tgsi_sampler, + const float s[QUAD_SIZE], + const float t[QUAD_SIZE], + const float p[QUAD_SIZE], + float lodbias, + float rgba[NUM_CHANNELS][QUAD_SIZE]) +{ + const struct sp_sampler_varient *samp = sp_sampler_varient(tgsi_sampler); + const struct pipe_texture *texture = samp->texture; + unsigned level0, j; + int width, height, depth; + int x[4], y[4], z[4]; + union tex_tile_address addr; + + level0 = samp->level; + width = texture->width[level0]; + height = texture->height[level0]; + depth = texture->depth[level0]; + + assert(width > 0); + assert(height > 0); + assert(depth > 0); + + samp->nearest_texcoord_s(s, width, x); + samp->nearest_texcoord_t(t, height, y); + samp->nearest_texcoord_p(p, depth, z); + + addr.value = 0; + addr.bits.level = samp->level; + + for (j = 0; j < QUAD_SIZE; j++) { + const float *out = get_texel_3d(samp, addr, x[j], y[j], z[j]); + int c; + for (c = 0; c < 4; c++) { + rgba[c][j] = out[c]; + } } +} - lambda = util_fast_log2(rho); - lambda += lodbias + sampler->lod_bias; - lambda = CLAMP(lambda, sampler->min_lod, sampler->max_lod); - return lambda; +static void +img_filter_1d_linear(struct tgsi_sampler *tgsi_sampler, + const float s[QUAD_SIZE], + const float t[QUAD_SIZE], + const float p[QUAD_SIZE], + float lodbias, + float rgba[NUM_CHANNELS][QUAD_SIZE]) +{ + const struct sp_sampler_varient *samp = sp_sampler_varient(tgsi_sampler); + const struct pipe_texture *texture = samp->texture; + unsigned level0, j; + int width; + int x0[4], x1[4]; + float xw[4]; /* weights */ + union tex_tile_address addr; + + level0 = samp->level; + width = texture->width[level0]; + + assert(width > 0); + + addr.value = 0; + addr.bits.level = samp->level; + + samp->linear_texcoord_s(s, width, x0, x1, xw); + + for (j = 0; j < QUAD_SIZE; j++) { + const float *tx0 = get_texel_2d(samp, addr, x0[j], 0); + const float *tx1 = get_texel_2d(samp, addr, x1[j], 0); + int c; + + /* interpolate R, G, B, A */ + for (c = 0; c < 4; c++) { + rgba[c][j] = lerp(xw[j], tx0[c], tx1[c]); + } + } } -/** - * Do several things here: - * 1. Compute lambda from the texcoords, if needed - * 2. Determine if we're minifying or magnifying - * 3. If minifying, choose mipmap levels - * 4. Return image filter to use within mipmap images - * \param level0 Returns first mipmap level to sample from - * \param level1 Returns second mipmap level to sample from - * \param levelBlend Returns blend factor between levels, in [0,1] - * \param imgFilter Returns either the min or mag filter, depending on lambda - */ static void -choose_mipmap_levels(const struct pipe_texture *texture, - const struct pipe_sampler_state *sampler, +img_filter_2d_linear(struct tgsi_sampler *tgsi_sampler, const float s[QUAD_SIZE], const float t[QUAD_SIZE], const float p[QUAD_SIZE], - boolean computeLambda, float lodbias, - unsigned *level0, unsigned *level1, float *levelBlend, - unsigned *imgFilter) -{ - if (sampler->min_mip_filter == PIPE_TEX_MIPFILTER_NONE) { - /* no mipmap selection needed */ - *level0 = *level1 = CLAMP((int) sampler->min_lod, - 0, (int) texture->last_level); - - if (sampler->min_img_filter != sampler->mag_img_filter) { - /* non-mipmapped texture, but still need to determine if doing - * minification or magnification. - */ - float lambda = compute_lambda(texture, sampler, s, t, p, lodbias); - if (lambda <= 0.0) { - *imgFilter = sampler->mag_img_filter; - } - else { - *imgFilter = sampler->min_img_filter; - } + float rgba[NUM_CHANNELS][QUAD_SIZE]) +{ + const struct sp_sampler_varient *samp = sp_sampler_varient(tgsi_sampler); + const struct pipe_texture *texture = samp->texture; + unsigned level0, j; + int width, height; + int x0[4], y0[4], x1[4], y1[4]; + float xw[4], yw[4]; /* weights */ + union tex_tile_address addr; + + level0 = samp->level; + width = texture->width[level0]; + height = texture->height[level0]; + + assert(width > 0); + assert(height > 0); + + addr.value = 0; + addr.bits.level = samp->level; + + samp->linear_texcoord_s(s, width, x0, x1, xw); + samp->linear_texcoord_t(t, height, y0, y1, yw); + + for (j = 0; j < QUAD_SIZE; j++) { + const float *tx0 = get_texel_2d(samp, addr, x0[j], y0[j]); + const float *tx1 = get_texel_2d(samp, addr, x1[j], y0[j]); + const float *tx2 = get_texel_2d(samp, addr, x0[j], y1[j]); + const float *tx3 = get_texel_2d(samp, addr, x1[j], y1[j]); + int c; + + /* interpolate R, G, B, A */ + for (c = 0; c < 4; c++) { + rgba[c][j] = lerp_2d(xw[j], yw[j], + tx0[c], tx1[c], + tx2[c], tx3[c]); } - else { - *imgFilter = sampler->mag_img_filter; + } +} + + +static void +img_filter_cube_linear(struct tgsi_sampler *tgsi_sampler, + const float s[QUAD_SIZE], + const float t[QUAD_SIZE], + const float p[QUAD_SIZE], + float lodbias, + float rgba[NUM_CHANNELS][QUAD_SIZE]) +{ + const struct sp_sampler_varient *samp = sp_sampler_varient(tgsi_sampler); + const struct pipe_texture *texture = samp->texture; + const unsigned *faces = samp->faces; /* zero when not cube-mapping */ + unsigned level0, j; + int width, height; + int x0[4], y0[4], x1[4], y1[4]; + float xw[4], yw[4]; /* weights */ + union tex_tile_address addr; + + level0 = samp->level; + width = texture->width[level0]; + height = texture->height[level0]; + + assert(width > 0); + assert(height > 0); + + addr.value = 0; + addr.bits.level = samp->level; + + samp->linear_texcoord_s(s, width, x0, x1, xw); + samp->linear_texcoord_t(t, height, y0, y1, yw); + + for (j = 0; j < QUAD_SIZE; j++) { + union tex_tile_address addrj = face(addr, faces[j]); + const float *tx0 = get_texel_2d(samp, addrj, x0[j], y0[j]); + const float *tx1 = get_texel_2d(samp, addrj, x1[j], y0[j]); + const float *tx2 = get_texel_2d(samp, addrj, x0[j], y1[j]); + const float *tx3 = get_texel_2d(samp, addrj, x1[j], y1[j]); + int c; + + /* interpolate R, G, B, A */ + for (c = 0; c < 4; c++) { + rgba[c][j] = lerp_2d(xw[j], yw[j], + tx0[c], tx1[c], + tx2[c], tx3[c]); } } - else { - float lambda; +} - if (computeLambda) - /* fragment shader */ - lambda = compute_lambda(texture, sampler, s, t, p, lodbias); - else - /* vertex shader */ - lambda = lodbias; /* not really a bias, but absolute LOD */ - if (lambda <= 0.0) { /* XXX threshold depends on the filter */ - /* magnifying */ - *imgFilter = sampler->mag_img_filter; - *level0 = *level1 = 0; +static void +img_filter_3d_linear(struct tgsi_sampler *tgsi_sampler, + const float s[QUAD_SIZE], + const float t[QUAD_SIZE], + const float p[QUAD_SIZE], + float lodbias, + float rgba[NUM_CHANNELS][QUAD_SIZE]) +{ + const struct sp_sampler_varient *samp = sp_sampler_varient(tgsi_sampler); + const struct pipe_texture *texture = samp->texture; + unsigned level0, j; + int width, height, depth; + int x0[4], x1[4], y0[4], y1[4], z0[4], z1[4]; + float xw[4], yw[4], zw[4]; /* interpolation weights */ + union tex_tile_address addr; + + level0 = samp->level; + width = texture->width[level0]; + height = texture->height[level0]; + depth = texture->depth[level0]; + + addr.value = 0; + addr.bits.level = level0; + + assert(width > 0); + assert(height > 0); + assert(depth > 0); + + samp->linear_texcoord_s(s, width, x0, x1, xw); + samp->linear_texcoord_t(t, height, y0, y1, yw); + samp->linear_texcoord_p(p, depth, z0, z1, zw); + + for (j = 0; j < QUAD_SIZE; j++) { + int c; + + const float *tx00 = get_texel_3d(samp, addr, x0[j], y0[j], z0[j]); + const float *tx01 = get_texel_3d(samp, addr, x1[j], y0[j], z0[j]); + const float *tx02 = get_texel_3d(samp, addr, x0[j], y1[j], z0[j]); + const float *tx03 = get_texel_3d(samp, addr, x1[j], y1[j], z0[j]); + + const float *tx10 = get_texel_3d(samp, addr, x0[j], y0[j], z1[j]); + const float *tx11 = get_texel_3d(samp, addr, x1[j], y0[j], z1[j]); + const float *tx12 = get_texel_3d(samp, addr, x0[j], y1[j], z1[j]); + const float *tx13 = get_texel_3d(samp, addr, x1[j], y1[j], z1[j]); + + /* interpolate R, G, B, A */ + for (c = 0; c < 4; c++) { + rgba[c][j] = lerp_3d(xw[j], yw[j], zw[j], + tx00[c], tx01[c], + tx02[c], tx03[c], + tx10[c], tx11[c], + tx12[c], tx13[c]); } - else { - /* minifying */ - *imgFilter = sampler->min_img_filter; - - /* choose mipmap level(s) and compute the blend factor between them */ - if (sampler->min_mip_filter == PIPE_TEX_MIPFILTER_NEAREST) { - /* Nearest mipmap level */ - const int lvl = (int) (lambda + 0.5); - *level0 = - *level1 = CLAMP(lvl, 0, (int) texture->last_level); - } - else { - /* Linear interpolation between mipmap levels */ - const int lvl = (int) lambda; - *level0 = CLAMP(lvl, 0, (int) texture->last_level); - *level1 = CLAMP(lvl + 1, 0, (int) texture->last_level); - *levelBlend = FRAC(lambda); /* blending weight between levels */ + } +} + + +static void +mip_filter_linear(struct tgsi_sampler *tgsi_sampler, + const float s[QUAD_SIZE], + const float t[QUAD_SIZE], + const float p[QUAD_SIZE], + float lodbias, + float rgba[NUM_CHANNELS][QUAD_SIZE]) +{ + struct sp_sampler_varient *samp = sp_sampler_varient(tgsi_sampler); + const struct pipe_texture *texture = samp->texture; + int level0; + float lambda; + + lambda = samp->compute_lambda(samp, s, t, p, lodbias); + level0 = (int)lambda; + + if (lambda < 0.0) { + samp->level = 0; + samp->mag_img_filter( tgsi_sampler, s, t, p, 0, rgba ); + } + else if (level0 >= texture->last_level) { + samp->level = texture->last_level; + samp->min_img_filter( tgsi_sampler, s, t, p, 0, rgba ); + } + else { + float levelBlend = lambda - level0; + float rgba0[4][4]; + float rgba1[4][4]; + int c,j; + + samp->level = level0; + samp->min_img_filter( tgsi_sampler, s, t, p, 0, rgba0 ); + + samp->level = level0+1; + samp->min_img_filter( tgsi_sampler, s, t, p, 0, rgba1 ); + + for (j = 0; j < QUAD_SIZE; j++) { + for (c = 0; c < 4; c++) { + rgba[c][j] = lerp(levelBlend, rgba0[c][j], rgba1[c][j]); } } } } -/** - * Get a texel from a texture, using the texture tile cache. - * - * \param face the cube face in 0..5 - * \param level the mipmap level - * \param x the x coord of texel within 2D image - * \param y the y coord of texel within 2D image - * \param z which slice of a 3D texture - * \param rgba the quad to put the texel/color into - * \param j which element of the rgba quad to write to - * - * XXX maybe move this into sp_tile_cache.c and merge with the - * sp_get_cached_tile_tex() function. Also, get 4 texels instead of 1... - */ static void -get_texel(const struct tgsi_sampler *tgsi_sampler, - unsigned face, unsigned level, int x, int y, int z, - float rgba[NUM_CHANNELS][QUAD_SIZE], unsigned j) +mip_filter_nearest(struct tgsi_sampler *tgsi_sampler, + const float s[QUAD_SIZE], + const float t[QUAD_SIZE], + const float p[QUAD_SIZE], + float lodbias, + float rgba[NUM_CHANNELS][QUAD_SIZE]) { - const struct sp_shader_sampler *samp = sp_shader_sampler(tgsi_sampler); - struct softpipe_context *sp = samp->sp; - const uint unit = samp->unit; - const struct pipe_texture *texture = sp->texture[unit]; - const struct pipe_sampler_state *sampler = sp->sampler[unit]; + struct sp_sampler_varient *samp = sp_sampler_varient(tgsi_sampler); + const struct pipe_texture *texture = samp->texture; + float lambda; - if (x < 0 || x >= (int) texture->width[level] || - y < 0 || y >= (int) texture->height[level] || - z < 0 || z >= (int) texture->depth[level]) { - rgba[0][j] = sampler->border_color[0]; - rgba[1][j] = sampler->border_color[1]; - rgba[2][j] = sampler->border_color[2]; - rgba[3][j] = sampler->border_color[3]; + lambda = samp->compute_lambda(samp, s, t, p, lodbias); + + if (lambda < 0.0) { + samp->level = 0; + samp->mag_img_filter( tgsi_sampler, s, t, p, 0, rgba ); } else { - const int tx = x % TILE_SIZE; - const int ty = y % TILE_SIZE; - const struct softpipe_cached_tile *tile - = sp_get_cached_tile_tex(sp, samp->cache, - x, y, z, face, level); - rgba[0][j] = tile->data.color[ty][tx][0]; - rgba[1][j] = tile->data.color[ty][tx][1]; - rgba[2][j] = tile->data.color[ty][tx][2]; - rgba[3][j] = tile->data.color[ty][tx][3]; - if (0) - { - debug_printf("Get texel %f %f %f %f from %s\n", - rgba[0][j], rgba[1][j], rgba[2][j], rgba[3][j], - pf_name(texture->format)); - } + samp->level = (int)(lambda + 0.5) ; + samp->level = MIN2(samp->level, (int)texture->last_level); + samp->min_img_filter( tgsi_sampler, s, t, p, 0, rgba ); + } + +#if 0 + printf("RGBA %g %g %g %g, %g %g %g %g, %g %g %g %g, %g %g %g %g\n", + rgba[0][0], rgba[1][0], rgba[2][0], rgba[3][0], + rgba[0][1], rgba[1][1], rgba[2][1], rgba[3][1], + rgba[0][2], rgba[1][2], rgba[2][2], rgba[3][2], + rgba[0][3], rgba[1][3], rgba[2][3], rgba[3][3]); +#endif +} + + +static void +mip_filter_none(struct tgsi_sampler *tgsi_sampler, + const float s[QUAD_SIZE], + const float t[QUAD_SIZE], + const float p[QUAD_SIZE], + float lodbias, + float rgba[NUM_CHANNELS][QUAD_SIZE]) +{ + struct sp_sampler_varient *samp = sp_sampler_varient(tgsi_sampler); + float lambda = samp->compute_lambda(samp, s, t, p, lodbias); + + if (lambda < 0.0) { + samp->mag_img_filter( tgsi_sampler, s, t, p, 0, rgba ); + } + else { + samp->min_img_filter( tgsi_sampler, s, t, p, 0, rgba ); } } + /** - * Compare texcoord 'p' (aka R) against texture value 'rgba[0]' - * When we sampled the depth texture, the depth value was put into all - * RGBA channels. We look at the red channel here. - * \param rgba quad of (depth) texel values - * \param p texture 'P' components for four pixels in quad - * \param j which pixel in the quad to test [0..3] + * Specialized version of mip_filter_linear with hard-wired calls to + * 2d lambda calculation and 2d_linear_repeat_POT img filters. */ -static INLINE void -shadow_compare(const struct pipe_sampler_state *sampler, - float rgba[NUM_CHANNELS][QUAD_SIZE], - const float p[QUAD_SIZE], - uint j) +static void +mip_filter_linear_2d_linear_repeat_POT( + struct tgsi_sampler *tgsi_sampler, + const float s[QUAD_SIZE], + const float t[QUAD_SIZE], + const float p[QUAD_SIZE], + float lodbias, + float rgba[NUM_CHANNELS][QUAD_SIZE]) { - int k; - switch (sampler->compare_func) { - case PIPE_FUNC_LESS: - k = p[j] < rgba[0][j]; - break; - case PIPE_FUNC_LEQUAL: - k = p[j] <= rgba[0][j]; - break; - case PIPE_FUNC_GREATER: - k = p[j] > rgba[0][j]; - break; - case PIPE_FUNC_GEQUAL: - k = p[j] >= rgba[0][j]; - break; - case PIPE_FUNC_EQUAL: - k = p[j] == rgba[0][j]; - break; - case PIPE_FUNC_NOTEQUAL: - k = p[j] != rgba[0][j]; - break; - case PIPE_FUNC_ALWAYS: - k = 1; - break; - case PIPE_FUNC_NEVER: - k = 0; - break; - default: - k = 0; - assert(0); - break; + struct sp_sampler_varient *samp = sp_sampler_varient(tgsi_sampler); + const struct pipe_texture *texture = samp->texture; + int level0; + float lambda; + + lambda = compute_lambda_2d(samp, s, t, p, lodbias); + level0 = (int)lambda; + + /* Catches both negative and large values of level0: + */ + if ((unsigned)level0 >= texture->last_level) { + if (level0 < 0) + samp->level = 0; + else + samp->level = texture->last_level; + + img_filter_2d_linear_repeat_POT( tgsi_sampler, s, t, p, 0, rgba ); } + else { + float levelBlend = lambda - level0; + float rgba0[4][4]; + float rgba1[4][4]; + int c,j; - /* XXX returning result for default GL_DEPTH_TEXTURE_MODE = GL_LUMINANCE */ - rgba[0][j] = rgba[1][j] = rgba[2][j] = (float) k; - rgba[3][j] = 1.0F; + samp->level = level0; + img_filter_2d_linear_repeat_POT( tgsi_sampler, s, t, p, 0, rgba0 ); + + samp->level = level0+1; + img_filter_2d_linear_repeat_POT( tgsi_sampler, s, t, p, 0, rgba1 ); + + for (j = 0; j < QUAD_SIZE; j++) { + for (c = 0; c < 4; c++) { + rgba[c][j] = lerp(levelBlend, rgba0[c][j], rgba1[c][j]); + } + } + } } + /** - * As above, but do four z/texture comparisons. + * Do shadow/depth comparisons. */ -static INLINE void -shadow_compare4(const struct pipe_sampler_state *sampler, - float rgba[NUM_CHANNELS][QUAD_SIZE], - const float p[QUAD_SIZE]) +static void +sample_compare(struct tgsi_sampler *tgsi_sampler, + const float s[QUAD_SIZE], + const float t[QUAD_SIZE], + const float p[QUAD_SIZE], + float lodbias, + float rgba[NUM_CHANNELS][QUAD_SIZE]) { + struct sp_sampler_varient *samp = sp_sampler_varient(tgsi_sampler); + const struct pipe_sampler_state *sampler = samp->sampler; int j, k0, k1, k2, k3; float val; + samp->mip_filter( tgsi_sampler, s, t, p, lodbias, rgba ); + + /** + * Compare texcoord 'p' (aka R) against texture value 'rgba[0]' + * When we sampled the depth texture, the depth value was put into all + * RGBA channels. We look at the red channel here. + */ + /* compare four texcoords vs. four texture samples */ switch (sampler->compare_func) { case PIPE_FUNC_LESS: @@ -826,470 +1500,392 @@ shadow_compare4(const struct pipe_sampler_state *sampler, /** - * Common code for sampling 1D/2D/cube textures. - * Could probably extend for 3D... + * Compute which cube face is referenced by each texcoord and put that + * info into the sampler faces[] array. Then sample the cube faces */ static void -sp_get_samples_2d_common(const struct tgsi_sampler *tgsi_sampler, - const float s[QUAD_SIZE], - const float t[QUAD_SIZE], - const float p[QUAD_SIZE], - boolean computeLambda, - float lodbias, - float rgba[NUM_CHANNELS][QUAD_SIZE], - const unsigned faces[4]) -{ - const struct sp_shader_sampler *samp = sp_shader_sampler(tgsi_sampler); - const struct softpipe_context *sp = samp->sp; - const uint unit = samp->unit; - const struct pipe_texture *texture = sp->texture[unit]; - const struct pipe_sampler_state *sampler = sp->sampler[unit]; - unsigned level0, level1, j, imgFilter; - int width, height; - float levelBlend; - - choose_mipmap_levels(texture, sampler, s, t, p, computeLambda, lodbias, - &level0, &level1, &levelBlend, &imgFilter); - - assert(sampler->normalized_coords); - - width = texture->width[level0]; - height = texture->height[level0]; - - assert(width > 0); - - switch (imgFilter) { - case PIPE_TEX_FILTER_NEAREST: - { - int x[4], y[4]; - nearest_texcoord_4(sampler->wrap_s, s, width, x); - nearest_texcoord_4(sampler->wrap_t, t, height, y); - - for (j = 0; j < QUAD_SIZE; j++) { - get_texel(tgsi_sampler, faces[j], level0, x[j], y[j], 0, rgba, j); - if (sampler->compare_mode == PIPE_TEX_COMPARE_R_TO_TEXTURE) { - shadow_compare(sampler, rgba, p, j); - } +sample_cube(struct tgsi_sampler *tgsi_sampler, + const float s[QUAD_SIZE], + const float t[QUAD_SIZE], + const float p[QUAD_SIZE], + float lodbias, + float rgba[NUM_CHANNELS][QUAD_SIZE]) +{ + struct sp_sampler_varient *samp = sp_sampler_varient(tgsi_sampler); + unsigned j; + float ssss[4], tttt[4]; - if (level0 != level1) { - /* get texels from second mipmap level and blend */ - float rgba2[4][4]; - unsigned c; - x[j] /= 2; - y[j] /= 2; - get_texel(tgsi_sampler, faces[j], level1, x[j], y[j], 0, - rgba2, j); - if (sampler->compare_mode == PIPE_TEX_COMPARE_R_TO_TEXTURE){ - shadow_compare(sampler, rgba2, p, j); - } - - for (c = 0; c < NUM_CHANNELS; c++) { - rgba[c][j] = lerp(levelBlend, rgba[c][j], rgba2[c][j]); - } - } + /* + major axis + direction target sc tc ma + ---------- ------------------------------- --- --- --- + +rx TEXTURE_CUBE_MAP_POSITIVE_X_EXT -rz -ry rx + -rx TEXTURE_CUBE_MAP_NEGATIVE_X_EXT +rz -ry rx + +ry TEXTURE_CUBE_MAP_POSITIVE_Y_EXT +rx +rz ry + -ry TEXTURE_CUBE_MAP_NEGATIVE_Y_EXT +rx -rz ry + +rz TEXTURE_CUBE_MAP_POSITIVE_Z_EXT +rx -ry rz + -rz TEXTURE_CUBE_MAP_NEGATIVE_Z_EXT -rx -ry rz + */ + for (j = 0; j < QUAD_SIZE; j++) { + float rx = s[j]; + float ry = t[j]; + float rz = p[j]; + const float arx = fabsf(rx), ary = fabsf(ry), arz = fabsf(rz); + unsigned face; + float sc, tc, ma; + + if (arx >= ary && arx >= arz) { + if (rx >= 0.0F) { + face = PIPE_TEX_FACE_POS_X; + sc = -rz; + tc = -ry; + ma = arx; + } + else { + face = PIPE_TEX_FACE_NEG_X; + sc = rz; + tc = -ry; + ma = arx; } } - break; - case PIPE_TEX_FILTER_LINEAR: - case PIPE_TEX_FILTER_ANISO: + else if (ary >= arx && ary >= arz) { + if (ry >= 0.0F) { + face = PIPE_TEX_FACE_POS_Y; + sc = rx; + tc = rz; + ma = ary; + } + else { + face = PIPE_TEX_FACE_NEG_Y; + sc = rx; + tc = -rz; + ma = ary; + } + } + else { + if (rz > 0.0F) { + face = PIPE_TEX_FACE_POS_Z; + sc = rx; + tc = -ry; + ma = arz; + } + else { + face = PIPE_TEX_FACE_NEG_Z; + sc = -rx; + tc = -ry; + ma = arz; + } + } + { - int x0[4], y0[4], x1[4], y1[4]; - float xw[4], yw[4]; /* weights */ - - linear_texcoord_4(sampler->wrap_s, s, width, x0, x1, xw); - linear_texcoord_4(sampler->wrap_t, t, height, y0, y1, yw); - - for (j = 0; j < QUAD_SIZE; j++) { - float tx[4][4]; /* texels */ - int c; - get_texel(tgsi_sampler, faces[j], level0, x0[j], y0[j], 0, tx, 0); - get_texel(tgsi_sampler, faces[j], level0, x1[j], y0[j], 0, tx, 1); - get_texel(tgsi_sampler, faces[j], level0, x0[j], y1[j], 0, tx, 2); - get_texel(tgsi_sampler, faces[j], level0, x1[j], y1[j], 0, tx, 3); - if (sampler->compare_mode == PIPE_TEX_COMPARE_R_TO_TEXTURE) { - shadow_compare4(sampler, tx, p); - } + const float ima = 1.0 / ma; + ssss[j] = ( sc * ima + 1.0F ) * 0.5F; + tttt[j] = ( tc * ima + 1.0F ) * 0.5F; + samp->faces[j] = face; + } + } - /* interpolate R, G, B, A */ - for (c = 0; c < 4; c++) { - rgba[c][j] = lerp_2d(xw[j], yw[j], - tx[c][0], tx[c][1], - tx[c][2], tx[c][3]); - } + /* In our little pipeline, the compare stage is next. If compare + * is not active, this will point somewhere deeper into the + * pipeline, eg. to mip_filter or even img_filter. + */ + samp->compare(tgsi_sampler, ssss, tttt, NULL, lodbias, rgba); +} - if (level0 != level1) { - /* get texels from second mipmap level and blend */ - float rgba2[4][4]; - x0[j] /= 2; - y0[j] /= 2; - x1[j] /= 2; - y1[j] /= 2; - get_texel(tgsi_sampler, faces[j], level1, x0[j], y0[j], 0, tx, 0); - get_texel(tgsi_sampler, faces[j], level1, x1[j], y0[j], 0, tx, 1); - get_texel(tgsi_sampler, faces[j], level1, x0[j], y1[j], 0, tx, 2); - get_texel(tgsi_sampler, faces[j], level1, x1[j], y1[j], 0, tx, 3); - if (sampler->compare_mode == PIPE_TEX_COMPARE_R_TO_TEXTURE){ - shadow_compare4(sampler, tx, p); - } - - /* interpolate R, G, B, A */ - for (c = 0; c < 4; c++) { - rgba2[c][j] = lerp_2d(xw[j], yw[j], - tx[c][0], tx[c][1], tx[c][2], tx[c][3]); - } - - for (c = 0; c < NUM_CHANNELS; c++) { - rgba[c][j] = lerp(levelBlend, rgba[c][j], rgba2[c][j]); - } - } - } - } - break; + + +static wrap_nearest_func +get_nearest_unorm_wrap(unsigned mode) +{ + switch (mode) { + case PIPE_TEX_WRAP_CLAMP: + return wrap_nearest_unorm_clamp; + case PIPE_TEX_WRAP_CLAMP_TO_EDGE: + case PIPE_TEX_WRAP_CLAMP_TO_BORDER: + return wrap_nearest_unorm_clamp_to_border; default: assert(0); + return wrap_nearest_unorm_clamp; } } -static INLINE void -sp_get_samples_1d(const struct tgsi_sampler *sampler, - const float s[QUAD_SIZE], - const float t[QUAD_SIZE], - const float p[QUAD_SIZE], - boolean computeLambda, - float lodbias, - float rgba[NUM_CHANNELS][QUAD_SIZE]) +static wrap_nearest_func +get_nearest_wrap(unsigned mode) { - static const unsigned faces[4] = {0, 0, 0, 0}; - static const float tzero[4] = {0, 0, 0, 0}; - sp_get_samples_2d_common(sampler, s, tzero, NULL, - computeLambda, lodbias, rgba, faces); + switch (mode) { + case PIPE_TEX_WRAP_REPEAT: + return wrap_nearest_repeat; + case PIPE_TEX_WRAP_CLAMP: + return wrap_nearest_clamp; + case PIPE_TEX_WRAP_CLAMP_TO_EDGE: + return wrap_nearest_clamp_to_edge; + case PIPE_TEX_WRAP_CLAMP_TO_BORDER: + return wrap_nearest_clamp_to_border; + case PIPE_TEX_WRAP_MIRROR_REPEAT: + return wrap_nearest_mirror_repeat; + case PIPE_TEX_WRAP_MIRROR_CLAMP: + return wrap_nearest_mirror_clamp; + case PIPE_TEX_WRAP_MIRROR_CLAMP_TO_EDGE: + return wrap_nearest_mirror_clamp_to_edge; + case PIPE_TEX_WRAP_MIRROR_CLAMP_TO_BORDER: + return wrap_nearest_mirror_clamp_to_border; + default: + assert(0); + return wrap_nearest_repeat; + } } -static INLINE void -sp_get_samples_2d(const struct tgsi_sampler *sampler, - const float s[QUAD_SIZE], - const float t[QUAD_SIZE], - const float p[QUAD_SIZE], - boolean computeLambda, - float lodbias, - float rgba[NUM_CHANNELS][QUAD_SIZE]) +static wrap_linear_func +get_linear_unorm_wrap(unsigned mode) { - static const unsigned faces[4] = {0, 0, 0, 0}; - sp_get_samples_2d_common(sampler, s, t, p, - computeLambda, lodbias, rgba, faces); + switch (mode) { + case PIPE_TEX_WRAP_CLAMP: + return wrap_linear_unorm_clamp; + case PIPE_TEX_WRAP_CLAMP_TO_EDGE: + case PIPE_TEX_WRAP_CLAMP_TO_BORDER: + return wrap_linear_unorm_clamp_to_border; + default: + assert(0); + return wrap_linear_unorm_clamp; + } } -static INLINE void -sp_get_samples_3d(const struct tgsi_sampler *tgsi_sampler, - const float s[QUAD_SIZE], - const float t[QUAD_SIZE], - const float p[QUAD_SIZE], - boolean computeLambda, - float lodbias, - float rgba[NUM_CHANNELS][QUAD_SIZE]) +static wrap_linear_func +get_linear_wrap(unsigned mode) { - const struct sp_shader_sampler *samp = sp_shader_sampler(tgsi_sampler); - const struct softpipe_context *sp = samp->sp; - const uint unit = samp->unit; - const struct pipe_texture *texture = sp->texture[unit]; - const struct pipe_sampler_state *sampler = sp->sampler[unit]; - /* get/map pipe_surfaces corresponding to 3D tex slices */ - unsigned level0, level1, j, imgFilter; - int width, height, depth; - float levelBlend; - const uint face = 0; - - choose_mipmap_levels(texture, sampler, s, t, p, computeLambda, lodbias, - &level0, &level1, &levelBlend, &imgFilter); - - assert(sampler->normalized_coords); - - width = texture->width[level0]; - height = texture->height[level0]; - depth = texture->depth[level0]; - - assert(width > 0); - assert(height > 0); - assert(depth > 0); - - switch (imgFilter) { - case PIPE_TEX_FILTER_NEAREST: - { - int x[4], y[4], z[4]; - nearest_texcoord_4(sampler->wrap_s, s, width, x); - nearest_texcoord_4(sampler->wrap_t, t, height, y); - nearest_texcoord_4(sampler->wrap_r, p, depth, z); - for (j = 0; j < QUAD_SIZE; j++) { - get_texel(tgsi_sampler, face, level0, x[j], y[j], z[j], rgba, j); - if (level0 != level1) { - /* get texels from second mipmap level and blend */ - float rgba2[4][4]; - unsigned c; - x[j] /= 2; - y[j] /= 2; - z[j] /= 2; - get_texel(tgsi_sampler, face, level1, x[j], y[j], z[j], rgba2, j); - for (c = 0; c < NUM_CHANNELS; c++) { - rgba[c][j] = lerp(levelBlend, rgba2[c][j], rgba[c][j]); - } - } - } - } - break; - case PIPE_TEX_FILTER_LINEAR: - case PIPE_TEX_FILTER_ANISO: - { - int x0[4], x1[4], y0[4], y1[4], z0[4], z1[4]; - float xw[4], yw[4], zw[4]; /* interpolation weights */ - linear_texcoord_4(sampler->wrap_s, s, width, x0, x1, xw); - linear_texcoord_4(sampler->wrap_t, t, height, y0, y1, yw); - linear_texcoord_4(sampler->wrap_r, p, depth, z0, z1, zw); - - for (j = 0; j < QUAD_SIZE; j++) { - int c; - float tx0[4][4], tx1[4][4]; - get_texel(tgsi_sampler, face, level0, x0[j], y0[j], z0[j], tx0, 0); - get_texel(tgsi_sampler, face, level0, x1[j], y0[j], z0[j], tx0, 1); - get_texel(tgsi_sampler, face, level0, x0[j], y1[j], z0[j], tx0, 2); - get_texel(tgsi_sampler, face, level0, x1[j], y1[j], z0[j], tx0, 3); - get_texel(tgsi_sampler, face, level0, x0[j], y0[j], z1[j], tx1, 0); - get_texel(tgsi_sampler, face, level0, x1[j], y0[j], z1[j], tx1, 1); - get_texel(tgsi_sampler, face, level0, x0[j], y1[j], z1[j], tx1, 2); - get_texel(tgsi_sampler, face, level0, x1[j], y1[j], z1[j], tx1, 3); - - /* interpolate R, G, B, A */ - for (c = 0; c < 4; c++) { - rgba[c][j] = lerp_3d(xw[j], yw[j], zw[j], - tx0[c][0], tx0[c][1], - tx0[c][2], tx0[c][3], - tx1[c][0], tx1[c][1], - tx1[c][2], tx1[c][3]); - } - - if (level0 != level1) { - /* get texels from second mipmap level and blend */ - float rgba2[4][4]; - x0[j] /= 2; - y0[j] /= 2; - z0[j] /= 2; - x1[j] /= 2; - y1[j] /= 2; - z1[j] /= 2; - get_texel(tgsi_sampler, face, level1, x0[j], y0[j], z0[j], tx0, 0); - get_texel(tgsi_sampler, face, level1, x1[j], y0[j], z0[j], tx0, 1); - get_texel(tgsi_sampler, face, level1, x0[j], y1[j], z0[j], tx0, 2); - get_texel(tgsi_sampler, face, level1, x1[j], y1[j], z0[j], tx0, 3); - get_texel(tgsi_sampler, face, level1, x0[j], y0[j], z1[j], tx1, 0); - get_texel(tgsi_sampler, face, level1, x1[j], y0[j], z1[j], tx1, 1); - get_texel(tgsi_sampler, face, level1, x0[j], y1[j], z1[j], tx1, 2); - get_texel(tgsi_sampler, face, level1, x1[j], y1[j], z1[j], tx1, 3); - - /* interpolate R, G, B, A */ - for (c = 0; c < 4; c++) { - rgba2[c][j] = lerp_3d(xw[j], yw[j], zw[j], - tx0[c][0], tx0[c][1], - tx0[c][2], tx0[c][3], - tx1[c][0], tx1[c][1], - tx1[c][2], tx1[c][3]); - } - - /* blend mipmap levels */ - for (c = 0; c < NUM_CHANNELS; c++) { - rgba[c][j] = lerp(levelBlend, rgba[c][j], rgba2[c][j]); - } - } - } - } - break; + switch (mode) { + case PIPE_TEX_WRAP_REPEAT: + return wrap_linear_repeat; + case PIPE_TEX_WRAP_CLAMP: + return wrap_linear_clamp; + case PIPE_TEX_WRAP_CLAMP_TO_EDGE: + return wrap_linear_clamp_to_edge; + case PIPE_TEX_WRAP_CLAMP_TO_BORDER: + return wrap_linear_clamp_to_border; + case PIPE_TEX_WRAP_MIRROR_REPEAT: + return wrap_linear_mirror_repeat; + case PIPE_TEX_WRAP_MIRROR_CLAMP: + return wrap_linear_mirror_clamp; + case PIPE_TEX_WRAP_MIRROR_CLAMP_TO_EDGE: + return wrap_linear_mirror_clamp_to_edge; + case PIPE_TEX_WRAP_MIRROR_CLAMP_TO_BORDER: + return wrap_linear_mirror_clamp_to_border; default: assert(0); + return wrap_linear_repeat; } } -static void -sp_get_samples_cube(const struct tgsi_sampler *sampler, - const float s[QUAD_SIZE], - const float t[QUAD_SIZE], - const float p[QUAD_SIZE], - boolean computeLambda, - float lodbias, - float rgba[NUM_CHANNELS][QUAD_SIZE]) +static compute_lambda_func +get_lambda_func(const union sp_sampler_key key) { - unsigned faces[QUAD_SIZE], j; - float ssss[4], tttt[4]; - for (j = 0; j < QUAD_SIZE; j++) { - faces[j] = choose_cube_face(s[j], t[j], p[j], ssss + j, tttt + j); + if (key.bits.processor == TGSI_PROCESSOR_VERTEX) + return compute_lambda_vert; + + switch (key.bits.target) { + case PIPE_TEXTURE_1D: + return compute_lambda_1d; + case PIPE_TEXTURE_2D: + case PIPE_TEXTURE_CUBE: + return compute_lambda_2d; + case PIPE_TEXTURE_3D: + return compute_lambda_3d; + default: + assert(0); + return compute_lambda_1d; } - sp_get_samples_2d_common(sampler, ssss, tttt, NULL, - computeLambda, lodbias, rgba, faces); } -static void -sp_get_samples_rect(const struct tgsi_sampler *tgsi_sampler, - const float s[QUAD_SIZE], - const float t[QUAD_SIZE], - const float p[QUAD_SIZE], - boolean computeLambda, - float lodbias, - float rgba[NUM_CHANNELS][QUAD_SIZE]) -{ - const struct sp_shader_sampler *samp = sp_shader_sampler(tgsi_sampler); - const struct softpipe_context *sp = samp->sp; - const uint unit = samp->unit; - const struct pipe_texture *texture = sp->texture[unit]; - const struct pipe_sampler_state *sampler = sp->sampler[unit]; - const uint face = 0; - unsigned level0, level1, j, imgFilter; - int width, height; - float levelBlend; - - choose_mipmap_levels(texture, sampler, s, t, p, computeLambda, lodbias, - &level0, &level1, &levelBlend, &imgFilter); - - /* texture RECTS cannot be mipmapped */ - assert(level0 == level1); - - width = texture->width[level0]; - height = texture->height[level0]; - - assert(width > 0); - - switch (imgFilter) { - case PIPE_TEX_FILTER_NEAREST: - { - int x[4], y[4]; - nearest_texcoord_unnorm_4(sampler->wrap_s, s, width, x); - nearest_texcoord_unnorm_4(sampler->wrap_t, t, height, y); - for (j = 0; j < QUAD_SIZE; j++) { - get_texel(tgsi_sampler, face, level0, x[j], y[j], 0, rgba, j); - if (sampler->compare_mode == PIPE_TEX_COMPARE_R_TO_TEXTURE) { - shadow_compare(sampler, rgba, p, j); - } - } - } +static filter_func +get_img_filter(const union sp_sampler_key key, + unsigned filter, + const struct pipe_sampler_state *sampler) +{ + switch (key.bits.target) { + case PIPE_TEXTURE_1D: + if (filter == PIPE_TEX_FILTER_NEAREST) + return img_filter_1d_nearest; + else + return img_filter_1d_linear; break; - case PIPE_TEX_FILTER_LINEAR: - case PIPE_TEX_FILTER_ANISO: + case PIPE_TEXTURE_2D: + /* Try for fast path: + */ + if (key.bits.is_pot && + sampler->wrap_s == sampler->wrap_t && + sampler->normalized_coords) { - int x0[4], y0[4], x1[4], y1[4]; - float xw[4], yw[4]; /* weights */ - linear_texcoord_unnorm_4(sampler->wrap_s, s, width, x0, x1, xw); - linear_texcoord_unnorm_4(sampler->wrap_t, t, height, y0, y1, yw); - for (j = 0; j < QUAD_SIZE; j++) { - float tx[4][4]; /* texels */ - int c; - get_texel(tgsi_sampler, face, level0, x0[j], y0[j], 0, tx, 0); - get_texel(tgsi_sampler, face, level0, x1[j], y0[j], 0, tx, 1); - get_texel(tgsi_sampler, face, level0, x0[j], y1[j], 0, tx, 2); - get_texel(tgsi_sampler, face, level0, x1[j], y1[j], 0, tx, 3); - if (sampler->compare_mode == PIPE_TEX_COMPARE_R_TO_TEXTURE) { - shadow_compare4(sampler, tx, p); + switch (sampler->wrap_s) { + case PIPE_TEX_WRAP_REPEAT: + switch (filter) { + case PIPE_TEX_FILTER_NEAREST: + return img_filter_2d_nearest_repeat_POT; + case PIPE_TEX_FILTER_LINEAR: + return img_filter_2d_linear_repeat_POT; + default: + break; } - for (c = 0; c < 4; c++) { - rgba[c][j] = lerp_2d(xw[j], yw[j], - tx[c][0], tx[c][1], tx[c][2], tx[c][3]); + break; + case PIPE_TEX_WRAP_CLAMP: + switch (filter) { + case PIPE_TEX_FILTER_NEAREST: + return img_filter_2d_nearest_clamp_POT; + default: + break; } } } + /* Otherwise use default versions: + */ + if (filter == PIPE_TEX_FILTER_NEAREST) + return img_filter_2d_nearest; + else + return img_filter_2d_linear; + break; + case PIPE_TEXTURE_CUBE: + if (filter == PIPE_TEX_FILTER_NEAREST) + return img_filter_cube_nearest; + else + return img_filter_cube_linear; + break; + case PIPE_TEXTURE_3D: + if (filter == PIPE_TEX_FILTER_NEAREST) + return img_filter_3d_nearest; + else + return img_filter_3d_linear; break; default: assert(0); + return img_filter_1d_nearest; } } /** - * Common code for vertex/fragment program texture sampling. + * Bind the given texture object and texture cache to the sampler varient. */ -static INLINE void -sp_get_samples(struct tgsi_sampler *tgsi_sampler, - const float s[QUAD_SIZE], - const float t[QUAD_SIZE], - const float p[QUAD_SIZE], - boolean computeLambda, - float lodbias, - float rgba[NUM_CHANNELS][QUAD_SIZE]) +void +sp_sampler_varient_bind_texture( struct sp_sampler_varient *samp, + struct softpipe_tex_tile_cache *tex_cache, + const struct pipe_texture *texture ) { - const struct sp_shader_sampler *samp = sp_shader_sampler(tgsi_sampler); - const struct softpipe_context *sp = samp->sp; - const uint unit = samp->unit; - const struct pipe_texture *texture = sp->texture[unit]; - const struct pipe_sampler_state *sampler = sp->sampler[unit]; - - if (!texture) - return; + const struct pipe_sampler_state *sampler = samp->sampler; - switch (texture->target) { - case PIPE_TEXTURE_1D: - assert(sampler->normalized_coords); - sp_get_samples_1d(tgsi_sampler, s, t, p, computeLambda, lodbias, rgba); - break; - case PIPE_TEXTURE_2D: - if (sampler->normalized_coords) - sp_get_samples_2d(tgsi_sampler, s, t, p, computeLambda, lodbias, rgba); - else - sp_get_samples_rect(tgsi_sampler, s, t, p, computeLambda, lodbias, rgba); - break; - case PIPE_TEXTURE_3D: - assert(sampler->normalized_coords); - sp_get_samples_3d(tgsi_sampler, s, t, p, computeLambda, lodbias, rgba); - break; - case PIPE_TEXTURE_CUBE: - assert(sampler->normalized_coords); - sp_get_samples_cube(tgsi_sampler, s, t, p, computeLambda, lodbias, rgba); - break; - default: - assert(0); - } - -#if 0 /* DEBUG */ - { - int i; - printf("Sampled at %f, %f, %f:\n", s[0], t[0], p[0]); - for (i = 0; i < 4; i++) { - printf("Frag %d: %f %f %f %f\n", i, - rgba[0][i], - rgba[1][i], - rgba[2][i], - rgba[3][i]); - } - } -#endif + samp->texture = texture; + samp->cache = tex_cache; + samp->xpot = util_unsigned_logbase2( texture->width[0] ); + samp->ypot = util_unsigned_logbase2( texture->height[0] ); + samp->level = CLAMP((int) sampler->min_lod, 0, (int) texture->last_level); } -/** - * Called via tgsi_sampler::get_samples() when running a fragment shader. - * Get four filtered RGBA values from the sampler's texture. - */ void -sp_get_samples_fragment(struct tgsi_sampler *tgsi_sampler, - const float s[QUAD_SIZE], - const float t[QUAD_SIZE], - const float p[QUAD_SIZE], - float lodbias, - float rgba[NUM_CHANNELS][QUAD_SIZE]) +sp_sampler_varient_destroy( struct sp_sampler_varient *samp ) { - sp_get_samples(tgsi_sampler, s, t, p, TRUE, lodbias, rgba); + FREE(samp); } /** - * Called via tgsi_sampler::get_samples() when running a vertex shader. - * Get four filtered RGBA values from the sampler's texture. + * Create a sampler varient for a given set of non-orthogonal state. */ -void -sp_get_samples_vertex(struct tgsi_sampler *tgsi_sampler, - const float s[QUAD_SIZE], - const float t[QUAD_SIZE], - const float p[QUAD_SIZE], - float lodbias, - float rgba[NUM_CHANNELS][QUAD_SIZE]) +struct sp_sampler_varient * +sp_create_sampler_varient( const struct pipe_sampler_state *sampler, + const union sp_sampler_key key ) { - sp_get_samples(tgsi_sampler, s, t, p, FALSE, lodbias, rgba); + struct sp_sampler_varient *samp = CALLOC_STRUCT(sp_sampler_varient); + if (!samp) + return NULL; + + samp->sampler = sampler; + samp->key = key; + + /* Note that (for instance) linear_texcoord_s and + * nearest_texcoord_s may be active at the same time, if the + * sampler min_img_filter differs from its mag_img_filter. + */ + if (sampler->normalized_coords) { + samp->linear_texcoord_s = get_linear_wrap( sampler->wrap_s ); + samp->linear_texcoord_t = get_linear_wrap( sampler->wrap_t ); + samp->linear_texcoord_p = get_linear_wrap( sampler->wrap_r ); + + samp->nearest_texcoord_s = get_nearest_wrap( sampler->wrap_s ); + samp->nearest_texcoord_t = get_nearest_wrap( sampler->wrap_t ); + samp->nearest_texcoord_p = get_nearest_wrap( sampler->wrap_r ); + } + else { + samp->linear_texcoord_s = get_linear_unorm_wrap( sampler->wrap_s ); + samp->linear_texcoord_t = get_linear_unorm_wrap( sampler->wrap_t ); + samp->linear_texcoord_p = get_linear_unorm_wrap( sampler->wrap_r ); + + samp->nearest_texcoord_s = get_nearest_unorm_wrap( sampler->wrap_s ); + samp->nearest_texcoord_t = get_nearest_unorm_wrap( sampler->wrap_t ); + samp->nearest_texcoord_p = get_nearest_unorm_wrap( sampler->wrap_r ); + } + + samp->compute_lambda = get_lambda_func( key ); + + samp->min_img_filter = get_img_filter(key, sampler->min_img_filter, sampler); + samp->mag_img_filter = get_img_filter(key, sampler->mag_img_filter, sampler); + + switch (sampler->min_mip_filter) { + case PIPE_TEX_MIPFILTER_NONE: + if (sampler->min_img_filter == sampler->mag_img_filter) + samp->mip_filter = samp->min_img_filter; + else + samp->mip_filter = mip_filter_none; + break; + + case PIPE_TEX_MIPFILTER_NEAREST: + samp->mip_filter = mip_filter_nearest; + break; + + case PIPE_TEX_MIPFILTER_LINEAR: + if (key.bits.is_pot && + sampler->min_img_filter == sampler->mag_img_filter && + sampler->normalized_coords && + sampler->wrap_s == PIPE_TEX_WRAP_REPEAT && + sampler->wrap_t == PIPE_TEX_WRAP_REPEAT && + sampler->min_img_filter == PIPE_TEX_FILTER_LINEAR) + { + samp->mip_filter = mip_filter_linear_2d_linear_repeat_POT; + } + else + { + samp->mip_filter = mip_filter_linear; + } + break; + } + + if (sampler->compare_mode != FALSE) { + samp->compare = sample_compare; + } + else { + /* Skip compare operation by promoting the mip_filter function + * pointer: + */ + samp->compare = samp->mip_filter; + } + + if (key.bits.target == PIPE_TEXTURE_CUBE) { + samp->base.get_samples = sample_cube; + } + else { + samp->faces[0] = 0; + samp->faces[1] = 0; + samp->faces[2] = 0; + samp->faces[3] = 0; + + /* Skip cube face determination by promoting the compare + * function pointer: + */ + samp->base.get_samples = samp->compare; + } + + return samp; } diff --git a/src/gallium/drivers/softpipe/sp_tex_sample.h b/src/gallium/drivers/softpipe/sp_tex_sample.h index 40d8eb2c2a8..b0797711d37 100644 --- a/src/gallium/drivers/softpipe/sp_tex_sample.h +++ b/src/gallium/drivers/softpipe/sp_tex_sample.h @@ -31,43 +31,122 @@ #include "tgsi/tgsi_exec.h" +struct sp_sampler_varient; + +typedef void (*wrap_nearest_func)(const float s[4], + unsigned size, + int icoord[4]); + +typedef void (*wrap_linear_func)(const float s[4], + unsigned size, + int icoord0[4], + int icoord1[4], + float w[4]); + +typedef float (*compute_lambda_func)(const struct sp_sampler_varient *sampler, + const float s[QUAD_SIZE], + const float t[QUAD_SIZE], + const float p[QUAD_SIZE], + float lodbias); + +typedef void (*filter_func)(struct tgsi_sampler *tgsi_sampler, + const float s[QUAD_SIZE], + const float t[QUAD_SIZE], + const float p[QUAD_SIZE], + float lodbias, + float rgba[NUM_CHANNELS][QUAD_SIZE]); + + +union sp_sampler_key { + struct { + unsigned target:3; + unsigned is_pot:1; + unsigned processor:2; + unsigned unit:4; + unsigned pad:22; + } bits; + unsigned value; +}; /** * Subclass of tgsi_sampler */ -struct sp_shader_sampler +struct sp_sampler_varient { struct tgsi_sampler base; /**< base class */ - uint unit; - struct softpipe_context *sp; - struct softpipe_tile_cache *cache; + union sp_sampler_key key; + + /* The owner of this struct: + */ + const struct pipe_sampler_state *sampler; + + + /* Currently bound texture: + */ + const struct pipe_texture *texture; + struct softpipe_tex_tile_cache *cache; + + unsigned processor; + + /* For sp_get_samples_2d_linear_POT: + */ + unsigned xpot; + unsigned ypot; + unsigned level; + + unsigned faces[4]; + + wrap_nearest_func nearest_texcoord_s; + wrap_nearest_func nearest_texcoord_t; + wrap_nearest_func nearest_texcoord_p; + + wrap_linear_func linear_texcoord_s; + wrap_linear_func linear_texcoord_t; + wrap_linear_func linear_texcoord_p; + + filter_func min_img_filter; + filter_func mag_img_filter; + + compute_lambda_func compute_lambda; + + filter_func mip_filter; + filter_func compare; + + /* Linked list: + */ + struct sp_sampler_varient *next; }; +struct sp_sampler; +/* Create a sampler varient for a given set of non-orthogonal state. Currently the + */ +struct sp_sampler_varient * +sp_create_sampler_varient( const struct pipe_sampler_state *sampler, + const union sp_sampler_key key ); -static INLINE const struct sp_shader_sampler * -sp_shader_sampler(const struct tgsi_sampler *sampler) -{ - return (const struct sp_shader_sampler *) sampler; -} +void sp_sampler_varient_bind_texture( struct sp_sampler_varient *varient, + struct softpipe_tex_tile_cache *tex_cache, + const struct pipe_texture *tex ); +void sp_sampler_varient_destroy( struct sp_sampler_varient * ); -extern void -sp_get_samples_fragment(struct tgsi_sampler *tgsi_sampler, - const float s[QUAD_SIZE], - const float t[QUAD_SIZE], - const float p[QUAD_SIZE], - float lodbias, - float rgba[NUM_CHANNELS][QUAD_SIZE]); + + +static INLINE struct sp_sampler_varient * +sp_sampler_varient(const struct tgsi_sampler *sampler) +{ + return (struct sp_sampler_varient *) sampler; +} extern void -sp_get_samples_vertex(struct tgsi_sampler *tgsi_sampler, - const float s[QUAD_SIZE], - const float t[QUAD_SIZE], - const float p[QUAD_SIZE], - float lodbias, - float rgba[NUM_CHANNELS][QUAD_SIZE]); +sp_get_samples(struct tgsi_sampler *tgsi_sampler, + const float s[QUAD_SIZE], + const float t[QUAD_SIZE], + const float p[QUAD_SIZE], + float lodbias, + float rgba[NUM_CHANNELS][QUAD_SIZE]); #endif /* SP_TEX_SAMPLE_H */ diff --git a/src/gallium/drivers/softpipe/sp_tex_tile_cache.c b/src/gallium/drivers/softpipe/sp_tex_tile_cache.c new file mode 100644 index 00000000000..407a22a9f4b --- /dev/null +++ b/src/gallium/drivers/softpipe/sp_tex_tile_cache.c @@ -0,0 +1,273 @@ +/************************************************************************** + * + * Copyright 2007 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. + * + **************************************************************************/ + +/** + * Texture tile caching. + * + * Author: + * Brian Paul + */ + +#include "pipe/p_inlines.h" +#include "util/u_memory.h" +#include "util/u_tile.h" +#include "sp_context.h" +#include "sp_surface.h" +#include "sp_texture.h" +#include "sp_tex_tile_cache.h" + + + +struct softpipe_tex_tile_cache * +sp_create_tex_tile_cache( struct pipe_screen *screen ) +{ + struct softpipe_tex_tile_cache *tc; + uint pos; + + tc = CALLOC_STRUCT( softpipe_tex_tile_cache ); + if (tc) { + tc->screen = screen; + for (pos = 0; pos < NUM_ENTRIES; pos++) { + tc->entries[pos].addr.bits.invalid = 1; + } + tc->last_tile = &tc->entries[0]; /* any tile */ + } + return tc; +} + + +void +sp_destroy_tex_tile_cache(struct softpipe_tex_tile_cache *tc) +{ + struct pipe_screen *screen; + uint pos; + + for (pos = 0; pos < NUM_ENTRIES; pos++) { + /*assert(tc->entries[pos].x < 0);*/ + } + if (tc->transfer) { + screen = tc->transfer->texture->screen; + screen->tex_transfer_destroy(tc->transfer); + } + if (tc->tex_trans) { + screen = tc->tex_trans->texture->screen; + screen->tex_transfer_destroy(tc->tex_trans); + } + + FREE( tc ); +} + + + + +void +sp_tex_tile_cache_map_transfers(struct softpipe_tex_tile_cache *tc) +{ + if (tc->tex_trans && !tc->tex_trans_map) + tc->tex_trans_map = tc->screen->transfer_map(tc->screen, tc->tex_trans); +} + + +void +sp_tex_tile_cache_unmap_transfers(struct softpipe_tex_tile_cache *tc) +{ + if (tc->tex_trans_map) { + tc->screen->transfer_unmap(tc->screen, tc->tex_trans); + tc->tex_trans_map = NULL; + } +} + +/** + * Invalidate all cached tiles for the cached texture. + * Should be called when the texture is modified. + */ +void +sp_tex_tile_cache_validate_texture(struct softpipe_tex_tile_cache *tc) +{ + unsigned i; + + assert(tc); + assert(tc->texture); + + for (i = 0; i < NUM_ENTRIES; i++) { + tc->entries[i].addr.bits.invalid = 1; + } +} + +/** + * Specify the texture to cache. + */ +void +sp_tex_tile_cache_set_texture(struct softpipe_tex_tile_cache *tc, + struct pipe_texture *texture) +{ + uint i; + + assert(!tc->transfer); + + if (tc->texture != texture) { + pipe_texture_reference(&tc->texture, texture); + + if (tc->tex_trans) { + struct pipe_screen *screen = tc->tex_trans->texture->screen; + + if (tc->tex_trans_map) { + screen->transfer_unmap(screen, tc->tex_trans); + tc->tex_trans_map = NULL; + } + + screen->tex_transfer_destroy(tc->tex_trans); + tc->tex_trans = NULL; + } + + /* mark as entries as invalid/empty */ + /* XXX we should try to avoid this when the teximage hasn't changed */ + for (i = 0; i < NUM_ENTRIES; i++) { + tc->entries[i].addr.bits.invalid = 1; + } + + tc->tex_face = -1; /* any invalid value here */ + } +} + + + + +/** + * Flush the tile cache: write all dirty tiles back to the transfer. + * any tiles "flagged" as cleared will be "really" cleared. + */ +void +sp_flush_tex_tile_cache(struct softpipe_tex_tile_cache *tc) +{ + int pos; + + if (tc->texture) { + /* caching a texture, mark all entries as empty */ + for (pos = 0; pos < NUM_ENTRIES; pos++) { + tc->entries[pos].addr.bits.invalid = 1; + } + tc->tex_face = -1; + } + +} + + +/** + * Given the texture face, level, zslice, x and y values, compute + * the cache entry position/index where we'd hope to find the + * cached texture tile. + * This is basically a direct-map cache. + * XXX There's probably lots of ways in which we can improve this. + */ +static INLINE uint +tex_cache_pos( union tex_tile_address addr ) +{ + uint entry = (addr.bits.x + + addr.bits.y * 9 + + addr.bits.z * 3 + + addr.bits.face + + addr.bits.level * 7); + + return entry % NUM_ENTRIES; +} + +/** + * Similar to sp_get_cached_tile() but for textures. + * Tiles are read-only and indexed with more params. + */ +const struct softpipe_tex_cached_tile * +sp_find_cached_tile_tex(struct softpipe_tex_tile_cache *tc, + union tex_tile_address addr ) +{ + struct pipe_screen *screen = tc->screen; + struct softpipe_tex_cached_tile *tile; + + tile = tc->entries + tex_cache_pos( addr ); + + if (addr.value != tile->addr.value) { + + /* cache miss. Most misses are because we've invaldiated the + * texture cache previously -- most commonly on binding a new + * texture. Currently we effectively flush the cache on texture + * bind. + */ +#if 0 + _debug_printf("miss at %u: x=%d y=%d z=%d face=%d level=%d\n" + " tile %u: x=%d y=%d z=%d face=%d level=%d\n", + pos, x/TILE_SIZE, y/TILE_SIZE, z, face, level, + pos, tile->addr.bits.x, tile->addr.bits.y, tile->z, tile->face, tile->level); +#endif + + /* check if we need to get a new transfer */ + if (!tc->tex_trans || + tc->tex_face != addr.bits.face || + tc->tex_level != addr.bits.level || + tc->tex_z != addr.bits.z) { + /* get new transfer (view into texture) */ + + if (tc->tex_trans) { + if (tc->tex_trans_map) { + tc->screen->transfer_unmap(tc->screen, tc->tex_trans); + tc->tex_trans_map = NULL; + } + + screen->tex_transfer_destroy(tc->tex_trans); + tc->tex_trans = NULL; + } + + tc->tex_trans = + screen->get_tex_transfer(screen, tc->texture, + addr.bits.face, + addr.bits.level, + addr.bits.z, + PIPE_TRANSFER_READ, 0, 0, + tc->texture->width[addr.bits.level], + tc->texture->height[addr.bits.level]); + + tc->tex_trans_map = screen->transfer_map(screen, tc->tex_trans); + + tc->tex_face = addr.bits.face; + tc->tex_level = addr.bits.level; + tc->tex_z = addr.bits.z; + } + + /* get tile from the transfer (view into texture) */ + pipe_get_tile_rgba(tc->tex_trans, + addr.bits.x * TILE_SIZE, + addr.bits.y * TILE_SIZE, + TILE_SIZE, TILE_SIZE, + (float *) tile->data.color); + tile->addr = addr; + } + + tc->last_tile = tile; + return tile; +} + + + diff --git a/src/gallium/drivers/softpipe/sp_tex_tile_cache.h b/src/gallium/drivers/softpipe/sp_tex_tile_cache.h new file mode 100644 index 00000000000..ac6886a3df1 --- /dev/null +++ b/src/gallium/drivers/softpipe/sp_tex_tile_cache.h @@ -0,0 +1,155 @@ +/************************************************************************** + * + * Copyright 2007 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. + * + **************************************************************************/ + +#ifndef SP_TEX_TILE_CACHE_H +#define SP_TEX_TILE_CACHE_H + + +#include "pipe/p_compiler.h" + + +struct softpipe_context; +struct softpipe_tex_tile_cache; + + +/** + * Cache tile size (width and height). This needs to be a power of two. + */ +#define TILE_SIZE 64 + + +/* If we need to support > 4096, just expand this to be a 64 bit + * union, or consider tiling in Z as well. + */ +union tex_tile_address { + struct { + unsigned x:6; /* 4096 / TILE_SIZE */ + unsigned y:6; /* 4096 / TILE_SIZE */ + unsigned z:12; /* 4096 -- z not tiled */ + unsigned face:3; + unsigned level:4; + unsigned invalid:1; + } bits; + unsigned value; +}; + + +struct softpipe_tex_cached_tile +{ + union tex_tile_address addr; + union { + float color[TILE_SIZE][TILE_SIZE][4]; + } data; +}; + +#define NUM_ENTRIES 50 + +struct softpipe_tex_tile_cache +{ + struct pipe_screen *screen; + struct pipe_transfer *transfer; + void *transfer_map; + + struct pipe_texture *texture; /**< if caching a texture */ + unsigned timestamp; + + struct softpipe_tex_cached_tile entries[NUM_ENTRIES]; + + struct pipe_transfer *tex_trans; + void *tex_trans_map; + int tex_face, tex_level, tex_z; + + struct softpipe_tex_cached_tile *last_tile; /**< most recently retrieved tile */ +}; + + +extern struct softpipe_tex_tile_cache * +sp_create_tex_tile_cache( struct pipe_screen *screen ); + +extern void +sp_destroy_tex_tile_cache(struct softpipe_tex_tile_cache *tc); + + +extern void +sp_tex_tile_cache_map_transfers(struct softpipe_tex_tile_cache *tc); + +extern void +sp_tex_tile_cache_unmap_transfers(struct softpipe_tex_tile_cache *tc); + +extern void +sp_tex_tile_cache_set_texture(struct softpipe_tex_tile_cache *tc, + struct pipe_texture *texture); + +void +sp_tex_tile_cache_validate_texture(struct softpipe_tex_tile_cache *tc); + +extern void +sp_flush_tex_tile_cache(struct softpipe_tex_tile_cache *tc); + + + +extern const struct softpipe_tex_cached_tile * +sp_find_cached_tile_tex(struct softpipe_tex_tile_cache *tc, + union tex_tile_address addr ); + +static INLINE union tex_tile_address +tex_tile_address( unsigned x, + unsigned y, + unsigned z, + unsigned face, + unsigned level ) +{ + union tex_tile_address addr; + + addr.value = 0; + addr.bits.x = x / TILE_SIZE; + addr.bits.y = y / TILE_SIZE; + addr.bits.z = z; + addr.bits.face = face; + addr.bits.level = level; + + return addr; +} + +/* Quickly retrieve tile if it matches last lookup. + */ +static INLINE const struct softpipe_tex_cached_tile * +sp_get_cached_tile_tex(struct softpipe_tex_tile_cache *tc, + union tex_tile_address addr ) +{ + if (tc->last_tile->addr.value == addr.value) + return tc->last_tile; + + return sp_find_cached_tile_tex( tc, addr ); +} + + + + + +#endif /* SP_TEX_TILE_CACHE_H */ + diff --git a/src/gallium/drivers/softpipe/sp_texture.c b/src/gallium/drivers/softpipe/sp_texture.c index 70f09324311..49b51afda7e 100644 --- a/src/gallium/drivers/softpipe/sp_texture.c +++ b/src/gallium/drivers/softpipe/sp_texture.c @@ -30,26 +30,21 @@ * Michel Dänzer <[email protected]> */ -#include "pipe/p_context.h" #include "pipe/p_defines.h" #include "pipe/p_inlines.h" -#include "pipe/internal/p_winsys_screen.h" #include "util/u_math.h" #include "util/u_memory.h" #include "sp_context.h" #include "sp_state.h" #include "sp_texture.h" -#include "sp_tile_cache.h" #include "sp_screen.h" #include "sp_winsys.h" -/* Simple, maximally packed layout. - */ - - -/* Conventional allocation path for non-display textures: +/** + * Conventional allocation path for non-display textures: + * Use a simple, maximally packed layout. */ static boolean softpipe_texture_layout(struct pipe_screen *screen, @@ -89,6 +84,10 @@ softpipe_texture_layout(struct pipe_screen *screen, return spt->buffer != NULL; } + +/** + * Texture layout for simple color buffers. + */ static boolean softpipe_displaytarget_layout(struct pipe_screen *screen, struct softpipe_texture * spt) @@ -112,21 +111,22 @@ softpipe_displaytarget_layout(struct pipe_screen *screen, } - - - static struct pipe_texture * softpipe_texture_create(struct pipe_screen *screen, - const struct pipe_texture *templat) + const struct pipe_texture *template) { struct softpipe_texture *spt = CALLOC_STRUCT(softpipe_texture); if (!spt) return NULL; - spt->base = *templat; + spt->base = *template; pipe_reference_init(&spt->base.reference, 1); spt->base.screen = screen; + spt->pot = (util_is_power_of_two(template->width[0]) && + util_is_power_of_two(template->height[0]) && + util_is_power_of_two(template->depth[0])); + if (spt->base.tex_usage & (PIPE_TEXTURE_USAGE_DISPLAY_TARGET | PIPE_TEXTURE_USAGE_PRIMARY)) { if (!softpipe_displaytarget_layout(screen, spt)) @@ -222,6 +222,13 @@ softpipe_get_tex_surface(struct pipe_screen *screen, if (ps->usage & PIPE_BUFFER_USAGE_GPU_READ) ps->usage |= PIPE_BUFFER_USAGE_CPU_READ; + if (ps->usage & (PIPE_BUFFER_USAGE_CPU_WRITE | + PIPE_BUFFER_USAGE_GPU_WRITE)) { + /* Mark the surface as dirty. The tile cache will look for this. */ + spt->timestamp++; + softpipe_screen(screen)->timestamp++; + } + ps->face = face; ps->level = level; ps->zslice = zslice; @@ -342,14 +349,13 @@ softpipe_transfer_map( struct pipe_screen *screen, /* May want to different things here depending on read/write nature * of the map: */ - if (transfer->texture && transfer->usage != PIPE_TRANSFER_READ) - { + if (transfer->texture && transfer->usage != PIPE_TRANSFER_READ) { /* Do something to notify sharing contexts of a texture change. * In softpipe, that would mean flushing the texture cache. */ softpipe_screen(screen)->timestamp++; } - + xfer_map = map + softpipe_transfer(transfer)->offset + transfer->y / transfer->block.height * transfer->stride + transfer->x / transfer->block.width * transfer->block.size; @@ -360,7 +366,7 @@ softpipe_transfer_map( struct pipe_screen *screen, static void softpipe_transfer_unmap(struct pipe_screen *screen, - struct pipe_transfer *transfer) + struct pipe_transfer *transfer) { struct softpipe_texture *spt; @@ -371,18 +377,12 @@ softpipe_transfer_unmap(struct pipe_screen *screen, if (transfer->usage != PIPE_TRANSFER_READ) { /* Mark the texture as dirty to expire the tile caches. */ - spt->modified = TRUE; + spt->timestamp++; } } void -softpipe_init_texture_funcs(struct softpipe_context *sp) -{ -} - - -void softpipe_init_screen_texture_funcs(struct pipe_screen *screen) { screen->texture_create = softpipe_texture_create; @@ -404,7 +404,7 @@ softpipe_get_texture_buffer( struct pipe_texture *texture, struct pipe_buffer **buf, unsigned *stride ) { - struct softpipe_texture *tex = (struct softpipe_texture *)texture; + struct softpipe_texture *tex = (struct softpipe_texture *) texture; if (!tex) return FALSE; diff --git a/src/gallium/drivers/softpipe/sp_texture.h b/src/gallium/drivers/softpipe/sp_texture.h index 893aa7d11d8..2537ab6a40d 100644 --- a/src/gallium/drivers/softpipe/sp_texture.h +++ b/src/gallium/drivers/softpipe/sp_texture.h @@ -48,7 +48,11 @@ struct softpipe_texture */ struct pipe_buffer *buffer; - boolean modified; + /* True if texture images are power-of-two in all dimensions: + */ + boolean pot; + + unsigned timestamp; }; struct softpipe_transfer @@ -74,9 +78,6 @@ softpipe_transfer(struct pipe_transfer *pt) extern void -softpipe_init_texture_funcs( struct softpipe_context *softpipe ); - -extern void softpipe_init_screen_texture_funcs(struct pipe_screen *screen); diff --git a/src/gallium/drivers/softpipe/sp_tile_cache.c b/src/gallium/drivers/softpipe/sp_tile_cache.c index 1f9b8f1f4fb..de479709858 100644 --- a/src/gallium/drivers/softpipe/sp_tile_cache.c +++ b/src/gallium/drivers/softpipe/sp_tile_cache.c @@ -26,7 +26,7 @@ **************************************************************************/ /** - * Texture tile caching. + * Render target tile caching. * * Author: * Brian Paul @@ -35,38 +35,8 @@ #include "pipe/p_inlines.h" #include "util/u_memory.h" #include "util/u_tile.h" -#include "sp_context.h" -#include "sp_surface.h" -#include "sp_texture.h" #include "sp_tile_cache.h" -#define NUM_ENTRIES 50 - - -/** XXX move these */ -#define MAX_WIDTH 2048 -#define MAX_HEIGHT 2048 - - -struct softpipe_tile_cache -{ - struct pipe_screen *screen; - struct pipe_surface *surface; /**< the surface we're caching */ - struct pipe_transfer *transfer; - void *transfer_map; - struct pipe_texture *texture; /**< if caching a texture */ - struct softpipe_cached_tile entries[NUM_ENTRIES]; - uint clear_flags[(MAX_WIDTH / TILE_SIZE) * (MAX_HEIGHT / TILE_SIZE) / 32]; - float clear_color[4]; /**< for color bufs */ - uint clear_val; /**< for z+stencil, or packed color clear value */ - boolean depth_stencil; /**< Is the surface a depth/stencil format? */ - - struct pipe_transfer *tex_trans; - void *tex_trans_map; - int tex_face, tex_level, tex_z; - - struct softpipe_cached_tile tile; /**< scratch tile for clears */ -}; /** @@ -76,7 +46,7 @@ struct softpipe_tile_cache * a LRU replacement policy. */ #define CACHE_POS(x, y) \ - (((x) / TILE_SIZE + ((y) / TILE_SIZE) * 5) % NUM_ENTRIES) + (((x) + (y) * 5) % NUM_ENTRIES) @@ -84,12 +54,10 @@ struct softpipe_tile_cache * Is the tile at (x,y) in cleared state? */ static INLINE uint -is_clear_flag_set(const uint *bitvec, int x, int y) +is_clear_flag_set(const uint *bitvec, union tile_address addr) { int pos, bit; - x /= TILE_SIZE; - y /= TILE_SIZE; - pos = y * (MAX_WIDTH / TILE_SIZE) + x; + pos = addr.bits.y * (MAX_WIDTH / TILE_SIZE) + addr.bits.x; assert(pos / 32 < (MAX_WIDTH / TILE_SIZE) * (MAX_HEIGHT / TILE_SIZE) / 32); bit = bitvec[pos / 32] & (1 << (pos & 31)); return bit; @@ -100,12 +68,10 @@ is_clear_flag_set(const uint *bitvec, int x, int y) * Mark the tile at (x,y) as not cleared. */ static INLINE void -clear_clear_flag(uint *bitvec, int x, int y) +clear_clear_flag(uint *bitvec, union tile_address addr) { int pos; - x /= TILE_SIZE; - y /= TILE_SIZE; - pos = y * (MAX_WIDTH / TILE_SIZE) + x; + pos = addr.bits.y * (MAX_WIDTH / TILE_SIZE) + addr.bits.x; assert(pos / 32 < (MAX_WIDTH / TILE_SIZE) * (MAX_HEIGHT / TILE_SIZE) / 32); bitvec[pos / 32] &= ~(1 << (pos & 31)); } @@ -116,14 +82,20 @@ sp_create_tile_cache( struct pipe_screen *screen ) { struct softpipe_tile_cache *tc; uint pos; + int maxLevels, maxTexSize; + + /* sanity checking: max sure MAX_WIDTH/HEIGHT >= largest texture image */ + maxLevels = screen->get_param(screen, PIPE_CAP_MAX_TEXTURE_2D_LEVELS); + maxTexSize = 1 << (maxLevels - 1); + assert(MAX_WIDTH >= maxTexSize); tc = CALLOC_STRUCT( softpipe_tile_cache ); if (tc) { tc->screen = screen; for (pos = 0; pos < NUM_ENTRIES; pos++) { - tc->entries[pos].x = - tc->entries[pos].y = -1; + tc->entries[pos].addr.bits.invalid = 1; } + tc->last_tile = &tc->entries[0]; /* any tile */ } return tc; } @@ -142,10 +114,6 @@ sp_destroy_tile_cache(struct softpipe_tile_cache *tc) screen = tc->transfer->texture->screen; screen->tex_transfer_destroy(tc->transfer); } - if (tc->tex_trans) { - screen = tc->tex_trans->texture->screen; - screen->tex_transfer_destroy(tc->tex_trans); - } FREE( tc ); } @@ -158,8 +126,6 @@ void sp_tile_cache_set_surface(struct softpipe_tile_cache *tc, struct pipe_surface *ps) { - assert(!tc->texture); - if (tc->transfer) { struct pipe_screen *screen = tc->transfer->texture->screen; @@ -211,9 +177,6 @@ sp_tile_cache_map_transfers(struct softpipe_tile_cache *tc) { if (tc->transfer && !tc->transfer_map) tc->transfer_map = tc->screen->transfer_map(tc->screen, tc->transfer); - - if (tc->tex_trans && !tc->tex_trans_map) - tc->tex_trans_map = tc->screen->transfer_map(tc->screen, tc->tex_trans); } @@ -224,47 +187,6 @@ sp_tile_cache_unmap_transfers(struct softpipe_tile_cache *tc) tc->screen->transfer_unmap(tc->screen, tc->transfer); tc->transfer_map = NULL; } - - if (tc->tex_trans_map) { - tc->screen->transfer_unmap(tc->screen, tc->tex_trans); - tc->tex_trans_map = NULL; - } -} - - -/** - * Specify the texture to cache. - */ -void -sp_tile_cache_set_texture(struct pipe_context *pipe, - struct softpipe_tile_cache *tc, - struct pipe_texture *texture) -{ - uint i; - - assert(!tc->transfer); - - pipe_texture_reference(&tc->texture, texture); - - if (tc->tex_trans) { - struct pipe_screen *screen = tc->tex_trans->texture->screen; - - if (tc->tex_trans_map) { - screen->transfer_unmap(screen, tc->tex_trans); - tc->tex_trans_map = NULL; - } - - screen->tex_transfer_destroy(tc->tex_trans); - tc->tex_trans = NULL; - } - - /* mark as entries as invalid/empty */ - /* XXX we should try to avoid this when the teximage hasn't changed */ - for (i = 0; i < NUM_ENTRIES; i++) { - tc->entries[i].x = -1; - } - - tc->tex_face = -1; /* any invalid value here */ } @@ -308,7 +230,7 @@ clear_tile(struct softpipe_cached_tile *tile, switch (pf_get_size(format)) { case 1: - memset(tile->data.any, 0, TILE_SIZE * TILE_SIZE); + memset(tile->data.any, clear_value, TILE_SIZE * TILE_SIZE); break; case 2: if (clear_value == 0) { @@ -344,8 +266,7 @@ clear_tile(struct softpipe_cached_tile *tile, * Actually clear the tiles which were flagged as being in a clear state. */ static void -sp_tile_cache_flush_clear(struct pipe_context *pipe, - struct softpipe_tile_cache *tc) +sp_tile_cache_flush_clear(struct softpipe_tile_cache *tc) { struct pipe_transfer *pt = tc->transfer; const uint w = tc->transfer->width; @@ -359,13 +280,15 @@ sp_tile_cache_flush_clear(struct pipe_context *pipe, /* push the tile to all positions marked as clear */ for (y = 0; y < h; y += TILE_SIZE) { for (x = 0; x < w; x += TILE_SIZE) { - if (is_clear_flag_set(tc->clear_flags, x, y)) { + union tile_address addr = tile_address(x, y); + + if (is_clear_flag_set(tc->clear_flags, addr)) { pipe_put_tile_raw(pt, x, y, TILE_SIZE, TILE_SIZE, tc->tile.data.color32, 0/*STRIDE*/); /* do this? */ - clear_clear_flag(tc->clear_flags, x, y); + clear_clear_flag(tc->clear_flags, addr); numCleared++; } @@ -382,8 +305,7 @@ sp_tile_cache_flush_clear(struct pipe_context *pipe, * any tiles "flagged" as cleared will be "really" cleared. */ void -sp_flush_tile_cache(struct softpipe_context *softpipe, - struct softpipe_tile_cache *tc) +sp_flush_tile_cache(struct softpipe_tile_cache *tc) { struct pipe_transfer *pt = tc->transfer; int inuse = 0, pos; @@ -392,33 +314,30 @@ sp_flush_tile_cache(struct softpipe_context *softpipe, /* caching a drawing transfer */ for (pos = 0; pos < NUM_ENTRIES; pos++) { struct softpipe_cached_tile *tile = tc->entries + pos; - if (tile->x >= 0) { + if (!tile->addr.bits.invalid) { if (tc->depth_stencil) { pipe_put_tile_raw(pt, - tile->x, tile->y, TILE_SIZE, TILE_SIZE, + tile->addr.bits.x * TILE_SIZE, + tile->addr.bits.y * TILE_SIZE, + TILE_SIZE, TILE_SIZE, tile->data.depth32, 0/*STRIDE*/); } else { pipe_put_tile_rgba(pt, - tile->x, tile->y, TILE_SIZE, TILE_SIZE, + tile->addr.bits.x * TILE_SIZE, + tile->addr.bits.y * TILE_SIZE, + TILE_SIZE, TILE_SIZE, (float *) tile->data.color); } - tile->x = tile->y = -1; /* mark as empty */ + tile->addr.bits.invalid = 1; /* mark as empty */ inuse++; } } #if TILE_CLEAR_OPTIMIZATION - sp_tile_cache_flush_clear(&softpipe->pipe, tc); + sp_tile_cache_flush_clear(tc); #endif } - else if (tc->texture) { - /* caching a texture, mark all entries as empty */ - for (pos = 0; pos < NUM_ENTRIES; pos++) { - tc->entries[pos].x = -1; - } - tc->tex_face = -1; - } #if 0 debug_printf("flushed tiles in use: %d\n", inuse); @@ -431,40 +350,39 @@ sp_flush_tile_cache(struct softpipe_context *softpipe, * \param x, y position of tile, in pixels */ struct softpipe_cached_tile * -sp_get_cached_tile(struct softpipe_context *softpipe, - struct softpipe_tile_cache *tc, int x, int y) +sp_find_cached_tile(struct softpipe_tile_cache *tc, + union tile_address addr ) { struct pipe_transfer *pt = tc->transfer; - - /* tile pos in framebuffer: */ - const int tile_x = x & ~(TILE_SIZE - 1); - const int tile_y = y & ~(TILE_SIZE - 1); - + /* cache pos/entry: */ - const int pos = CACHE_POS(x, y); + const int pos = CACHE_POS(addr.bits.x, + addr.bits.y); struct softpipe_cached_tile *tile = tc->entries + pos; - if (tile_x != tile->x || - tile_y != tile->y) { + if (addr.value != tile->addr.value) { - if (tile->x != -1) { + if (tile->addr.bits.invalid == 0) { /* put dirty tile back in framebuffer */ if (tc->depth_stencil) { pipe_put_tile_raw(pt, - tile->x, tile->y, TILE_SIZE, TILE_SIZE, + tile->addr.bits.x * TILE_SIZE, + tile->addr.bits.y * TILE_SIZE, + TILE_SIZE, TILE_SIZE, tile->data.depth32, 0/*STRIDE*/); } else { pipe_put_tile_rgba(pt, - tile->x, tile->y, TILE_SIZE, TILE_SIZE, + tile->addr.bits.x * TILE_SIZE, + tile->addr.bits.y * TILE_SIZE, + TILE_SIZE, TILE_SIZE, (float *) tile->data.color); } } - tile->x = tile_x; - tile->y = tile_y; + tile->addr = addr; - if (is_clear_flag_set(tc->clear_flags, x, y)) { + if (is_clear_flag_set(tc->clear_flags, addr)) { /* don't get tile from framebuffer, just clear it */ if (tc->depth_stencil) { clear_tile(tile, pt->format, tc->clear_val); @@ -472,125 +390,33 @@ sp_get_cached_tile(struct softpipe_context *softpipe, else { clear_tile_rgba(tile, pt->format, tc->clear_color); } - clear_clear_flag(tc->clear_flags, x, y); + clear_clear_flag(tc->clear_flags, addr); } else { /* get new tile data from transfer */ if (tc->depth_stencil) { pipe_get_tile_raw(pt, - tile->x, tile->y, TILE_SIZE, TILE_SIZE, + tile->addr.bits.x * TILE_SIZE, + tile->addr.bits.y * TILE_SIZE, + TILE_SIZE, TILE_SIZE, tile->data.depth32, 0/*STRIDE*/); } else { pipe_get_tile_rgba(pt, - tile->x, tile->y, TILE_SIZE, TILE_SIZE, + tile->addr.bits.x * TILE_SIZE, + tile->addr.bits.y * TILE_SIZE, + TILE_SIZE, TILE_SIZE, (float *) tile->data.color); } } } + tc->last_tile = tile; return tile; } -/** - * Given the texture face, level, zslice, x and y values, compute - * the cache entry position/index where we'd hope to find the - * cached texture tile. - * This is basically a direct-map cache. - * XXX There's probably lots of ways in which we can improve this. - */ -static INLINE uint -tex_cache_pos(int x, int y, int z, int face, int level) -{ - uint entry = x + y * 9 + z * 3 + face + level * 7; - return entry % NUM_ENTRIES; -} - - -/** - * Similar to sp_get_cached_tile() but for textures. - * Tiles are read-only and indexed with more params. - */ -const struct softpipe_cached_tile * -sp_get_cached_tile_tex(struct softpipe_context *sp, - struct softpipe_tile_cache *tc, int x, int y, int z, - int face, int level) -{ - struct pipe_screen *screen = sp->pipe.screen; - /* tile pos in framebuffer: */ - const int tile_x = x & ~(TILE_SIZE - 1); - const int tile_y = y & ~(TILE_SIZE - 1); - /* cache pos/entry: */ - const uint pos = tex_cache_pos(x / TILE_SIZE, y / TILE_SIZE, z, - face, level); - struct softpipe_cached_tile *tile = tc->entries + pos; - - if (tc->texture) { - struct softpipe_texture *spt = softpipe_texture(tc->texture); - if (spt->modified) { - /* texture was modified, invalidate all cached tiles */ - uint p; - for (p = 0; p < NUM_ENTRIES; p++) { - tile = tc->entries + p; - tile->x = -1; - } - spt->modified = FALSE; - } - } - - if (tile_x != tile->x || - tile_y != tile->y || - z != tile->z || - face != tile->face || - level != tile->level) { - /* cache miss */ -#if 0 - printf("miss at %u x=%d y=%d z=%d face=%d level=%d\n", pos, - x/TILE_SIZE, y/TILE_SIZE, z, face, level); -#endif - /* check if we need to get a new transfer */ - if (!tc->tex_trans || - tc->tex_face != face || - tc->tex_level != level || - tc->tex_z != z) { - /* get new transfer (view into texture) */ - - if (tc->tex_trans) { - if (tc->tex_trans_map) { - tc->screen->transfer_unmap(tc->screen, tc->tex_trans); - tc->tex_trans_map = NULL; - } - - screen->tex_transfer_destroy(tc->tex_trans); - tc->tex_trans = NULL; - } - - tc->tex_trans = screen->get_tex_transfer(screen, tc->texture, face, level, z, - PIPE_TRANSFER_READ, 0, 0, - tc->texture->width[level], - tc->texture->height[level]); - tc->tex_trans_map = screen->transfer_map(screen, tc->tex_trans); - - tc->tex_face = face; - tc->tex_level = level; - tc->tex_z = z; - } - - /* get tile from the transfer (view into texture) */ - pipe_get_tile_rgba(tc->tex_trans, - tile_x, tile_y, TILE_SIZE, TILE_SIZE, - (float *) tile->data.color); - tile->x = tile_x; - tile->y = tile_y; - tile->z = z; - tile->face = face; - tile->level = level; - } - - return tile; -} /** @@ -621,6 +447,6 @@ sp_tile_cache_clear(struct softpipe_tile_cache *tc, const float *rgba, for (pos = 0; pos < NUM_ENTRIES; pos++) { struct softpipe_cached_tile *tile = tc->entries + pos; - tile->x = tile->y = -1; + tile->addr.bits.invalid = 1; } } diff --git a/src/gallium/drivers/softpipe/sp_tile_cache.h b/src/gallium/drivers/softpipe/sp_tile_cache.h index 8f247d0e580..a12092702a6 100644 --- a/src/gallium/drivers/softpipe/sp_tile_cache.h +++ b/src/gallium/drivers/softpipe/sp_tile_cache.h @@ -34,7 +34,6 @@ #include "pipe/p_compiler.h" -struct softpipe_context; struct softpipe_tile_cache; @@ -44,11 +43,23 @@ struct softpipe_tile_cache; #define TILE_SIZE 64 +/* If we need to support > 4096, just expand this to be a 64 bit + * union, or consider tiling in Z as well. + */ +union tile_address { + struct { + unsigned x:6; /* 4096 / TILE_SIZE */ + unsigned y:6; /* 4096 / TILE_SIZE */ + unsigned invalid:1; + unsigned pad:19; + } bits; + unsigned value; +}; + struct softpipe_cached_tile { - int x, y; /**< pos of tile in window coords */ - int z, face, level; /**< Extra texture indexes */ + union tile_address addr; union { float color[TILE_SIZE][TILE_SIZE][4]; uint color32[TILE_SIZE][TILE_SIZE]; @@ -59,6 +70,32 @@ struct softpipe_cached_tile } data; }; +#define NUM_ENTRIES 50 + + +/** XXX move these */ +#define MAX_WIDTH 4096 +#define MAX_HEIGHT 4096 + + +struct softpipe_tile_cache +{ + struct pipe_screen *screen; + struct pipe_surface *surface; /**< the surface we're caching */ + struct pipe_transfer *transfer; + void *transfer_map; + + struct softpipe_cached_tile entries[NUM_ENTRIES]; + uint clear_flags[(MAX_WIDTH / TILE_SIZE) * (MAX_HEIGHT / TILE_SIZE) / 32]; + float clear_color[4]; /**< for color bufs */ + uint clear_val; /**< for z+stencil, or packed color clear value */ + boolean depth_stencil; /**< Is the surface a depth/stencil format? */ + + struct softpipe_cached_tile tile; /**< scratch tile for clears */ + + struct softpipe_cached_tile *last_tile; /**< most recently retrieved tile */ +}; + extern struct softpipe_tile_cache * sp_create_tile_cache( struct pipe_screen *screen ); @@ -80,26 +117,45 @@ extern void sp_tile_cache_unmap_transfers(struct softpipe_tile_cache *tc); extern void -sp_tile_cache_set_texture(struct pipe_context *pipe, - struct softpipe_tile_cache *tc, - struct pipe_texture *texture); - -extern void -sp_flush_tile_cache(struct softpipe_context *softpipe, - struct softpipe_tile_cache *tc); +sp_flush_tile_cache(struct softpipe_tile_cache *tc); extern void sp_tile_cache_clear(struct softpipe_tile_cache *tc, const float *rgba, uint clearValue); extern struct softpipe_cached_tile * -sp_get_cached_tile(struct softpipe_context *softpipe, - struct softpipe_tile_cache *tc, int x, int y); +sp_find_cached_tile(struct softpipe_tile_cache *tc, + union tile_address addr ); + + +static INLINE union tile_address +tile_address( unsigned x, + unsigned y ) +{ + union tile_address addr; + + addr.value = 0; + addr.bits.x = x / TILE_SIZE; + addr.bits.y = y / TILE_SIZE; + + return addr; +} + +/* Quickly retrieve tile if it matches last lookup. + */ +static INLINE struct softpipe_cached_tile * +sp_get_cached_tile(struct softpipe_tile_cache *tc, + int x, int y ) +{ + union tile_address addr = tile_address( x, y ); + + if (tc->last_tile->addr.value == addr.value) + return tc->last_tile; + + return sp_find_cached_tile( tc, addr ); +} + -extern const struct softpipe_cached_tile * -sp_get_cached_tile_tex(struct softpipe_context *softpipe, - struct softpipe_tile_cache *tc, int x, int y, int z, - int face, int level); #endif /* SP_TILE_CACHE_H */ diff --git a/src/gallium/drivers/trace/tr_context.c b/src/gallium/drivers/trace/tr_context.c index ae0af4d0557..bf470b46ae1 100644 --- a/src/gallium/drivers/trace/tr_context.c +++ b/src/gallium/drivers/trace/tr_context.c @@ -125,11 +125,11 @@ trace_context_draw_block(struct trace_context *tr_ctx, int flag) } else if ((tr_ctx->draw_rule.blocker & flag) && (tr_ctx->draw_blocker & 4)) { boolean block = FALSE; - debug_printf("%s (%lu %lu) (%lu %lu) (%lu %u) (%lu %u)\n", __FUNCTION__, - tr_ctx->draw_rule.fs, tr_ctx->curr.fs, - tr_ctx->draw_rule.vs, tr_ctx->curr.vs, - tr_ctx->draw_rule.surf, 0, - tr_ctx->draw_rule.tex, 0); + debug_printf("%s (%p %p) (%p %p) (%p %u) (%p %u)\n", __FUNCTION__, + (void *) tr_ctx->draw_rule.fs, (void *) tr_ctx->curr.fs, + (void *) tr_ctx->draw_rule.vs, (void *) tr_ctx->curr.vs, + (void *) tr_ctx->draw_rule.surf, 0, + (void *) tr_ctx->draw_rule.tex, 0); if (tr_ctx->draw_rule.fs && tr_ctx->draw_rule.fs == tr_ctx->curr.fs) block = TRUE; diff --git a/src/gallium/include/pipe/p_defines.h b/src/gallium/include/pipe/p_defines.h index b01ab6d137c..f252d6df00d 100644 --- a/src/gallium/include/pipe/p_defines.h +++ b/src/gallium/include/pipe/p_defines.h @@ -281,7 +281,7 @@ enum pipe_transfer_usage { #define PIPE_CAP_NPOT_TEXTURES 2 #define PIPE_CAP_TWO_SIDED_STENCIL 3 #define PIPE_CAP_GLSL 4 /* XXX need something better */ -#define PIPE_CAP_S3TC 5 +#define PIPE_CAP_S3TC 5 /* XXX: deprecated; cap determined via supported sampler formats */ #define PIPE_CAP_ANISOTROPIC_FILTER 6 #define PIPE_CAP_POINT_SPRITE 7 #define PIPE_CAP_MAX_RENDER_TARGETS 8 diff --git a/src/gallium/include/pipe/p_inlines.h b/src/gallium/include/pipe/p_inlines.h index a5c1e8270a3..30a4aaf4093 100644 --- a/src/gallium/include/pipe/p_inlines.h +++ b/src/gallium/include/pipe/p_inlines.h @@ -118,7 +118,7 @@ pipe_buffer_write(struct pipe_screen *screen, unsigned offset, unsigned size, const void *data) { - uint8_t *map; + void *map; assert(offset < buf->size); assert(offset + size <= buf->size); @@ -129,7 +129,7 @@ pipe_buffer_write(struct pipe_screen *screen, PIPE_BUFFER_USAGE_FLUSH_EXPLICIT); assert(map); if(map) { - memcpy(map + offset, data, size); + memcpy((uint8_t *)map + offset, data, size); pipe_buffer_flush_mapped_range(screen, buf, offset, size); pipe_buffer_unmap(screen, buf); } @@ -141,7 +141,7 @@ pipe_buffer_read(struct pipe_screen *screen, unsigned offset, unsigned size, void *data) { - uint8_t *map; + void *map; assert(offset < buf->size); assert(offset + size <= buf->size); @@ -150,11 +150,31 @@ pipe_buffer_read(struct pipe_screen *screen, map = pipe_buffer_map_range(screen, buf, offset, size, PIPE_BUFFER_USAGE_CPU_READ); assert(map); if(map) { - memcpy(data, map + offset, size); + memcpy(data, (const uint8_t *)map + offset, size); pipe_buffer_unmap(screen, buf); } } +static INLINE void * +pipe_transfer_map( struct pipe_transfer *transf ) +{ + struct pipe_screen *screen = transf->texture->screen; + return screen->transfer_map(screen, transf); +} + +static INLINE void +pipe_transfer_unmap( struct pipe_transfer *transf ) +{ + struct pipe_screen *screen = transf->texture->screen; + screen->transfer_unmap(screen, transf); +} + +static INLINE void +pipe_transfer_destroy( struct pipe_transfer *transf ) +{ + struct pipe_screen *screen = transf->texture->screen; + screen->tex_transfer_destroy(transf); +} #ifdef __cplusplus } diff --git a/src/gallium/include/pipe/p_state.h b/src/gallium/include/pipe/p_state.h index 626bedb35a8..b59d6b7ae32 100644 --- a/src/gallium/include/pipe/p_state.h +++ b/src/gallium/include/pipe/p_state.h @@ -114,11 +114,29 @@ struct pipe_rasterizer_state * the vertex shader, clipping and viewport processing. Note that * a vertex shader is still needed though, to indicate the mapping * from vertex elements to fragment shader input semantics. + * + * XXX: considered for removal. */ unsigned bypass_vs_clip_and_viewport:1; - unsigned flatshade_first:1; /**< take color attribute from the first vertex of a primitive */ - unsigned gl_rasterization_rules:1; /**< enable tweaks for GL rasterization? */ + /** + * Use the first vertex of a primitive as the provoking vertex for + * flat shading. + */ + unsigned flatshade_first:1; + + /** + * When true, triangle rasterization uses (0.5, 0.5) pixel centers + * for determining pixel ownership. + * + * When false, triangle rasterization uses (0,0) pixel centers for + * determining pixel ownership. + * + * Triangle rasterization always uses a 'top,left' rule for pixel + * ownership, this just alters which point we consider the pixel + * center for that test. + */ + unsigned gl_rasterization_rules:1; float line_width; float point_size; /**< used when no per-vertex size */ @@ -307,7 +325,7 @@ struct pipe_transfer unsigned nblocksx; /**< allocated width in blocks */ unsigned nblocksy; /**< allocated height in blocks */ unsigned stride; /**< stride in bytes between rows of blocks */ - unsigned usage; /**< PIPE_TRANSFER_* */ + enum pipe_transfer_usage usage; /**< PIPE_TRANSFER_* */ struct pipe_texture *texture; /**< texture to transfer to/from */ unsigned face; diff --git a/src/gallium/include/pipe/p_thread.h b/src/gallium/include/pipe/p_thread.h index b1606dc6526..25e41482325 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) || defined(PIPE_OS_HAIKU) +#if defined(PIPE_OS_LINUX) || defined(PIPE_OS_BSD) || defined(PIPE_OS_SOLARIS) || defined(PIPE_OS_APPLE) || defined(PIPE_OS_HAIKU) #include <pthread.h> /* POSIX threads headers */ #include <stdio.h> /* 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) || defined(PIPE_OS_HAIKU) +#if defined(PIPE_OS_LINUX) || defined(PIPE_OS_BSD) || defined(PIPE_OS_SOLARIS) || defined(PIPE_OS_APPLE) || 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) || defined(PIPE_OS_HAIKU) +#if defined(PIPE_OS_LINUX) || defined(PIPE_OS_BSD) || defined(PIPE_OS_SOLARIS) || defined(PIPE_OS_APPLE) || 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) || defined(PIPE_OS_HAIKU) +#if defined(PIPE_OS_LINUX) || defined(PIPE_OS_BSD) || defined(PIPE_OS_SOLARIS) || defined(PIPE_OS_APPLE) || 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) || defined(PIPE_OS_HAIKU) +#if defined(PIPE_OS_LINUX) || defined(PIPE_OS_BSD) || defined(PIPE_OS_SOLARIS) || defined(PIPE_OS_APPLE) || defined(PIPE_OS_HAIKU) if (pthread_setspecific(tsd->key, value) != 0) { perror("pthread_set_specific() failed"); exit(-1); diff --git a/src/gallium/state_trackers/g3dvl/vl_basic_csc.c b/src/gallium/state_trackers/g3dvl/vl_basic_csc.c index 20d682de3fb..b1683b891be 100644 --- a/src/gallium/state_trackers/g3dvl/vl_basic_csc.c +++ b/src/gallium/state_trackers/g3dvl/vl_basic_csc.c @@ -51,6 +51,7 @@ static int vlResizeFrameBuffer struct vlBasicCSC *basic_csc; struct pipe_context *pipe; struct pipe_texture template; + float clear_color[4]; assert(csc); @@ -68,7 +69,12 @@ static int vlResizeFrameBuffer basic_csc->viewport.translate[1] = 0; basic_csc->viewport.translate[2] = 0; basic_csc->viewport.translate[3] = 0; - + + clear_color[0] = 0.0f; + clear_color[1] = 0.0f; + clear_color[2] = 0.0f; + clear_color[3] = 0.0f; + if (basic_csc->framebuffer_tex) { pipe_surface_reference(&basic_csc->framebuffer.cbufs[0], NULL); @@ -98,7 +104,7 @@ static int vlResizeFrameBuffer /* Clear to black, in case video doesn't fill the entire window */ pipe->set_framebuffer_state(pipe, &basic_csc->framebuffer); - pipe->clear(pipe, PIPE_CLEAR_COLOR, 0, 0.0f, 0); + pipe->clear(pipe, PIPE_CLEAR_COLOR, clear_color, 0.0f, 0); return 0; } @@ -425,7 +431,7 @@ static int vlCreateVertexShader */ for (i = 0; i < 2; ++i) { - inst = vl_inst4(TGSI_OPCODE_MADD, TGSI_FILE_OUTPUT, i, TGSI_FILE_INPUT, i, TGSI_FILE_CONSTANT, i * 2, TGSI_FILE_CONSTANT, i * 2 + 1); + inst = vl_inst4(TGSI_OPCODE_MAD, TGSI_FILE_OUTPUT, i, TGSI_FILE_INPUT, i, TGSI_FILE_CONSTANT, i * 2, TGSI_FILE_CONSTANT, i * 2 + 1); ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); } diff --git a/src/gallium/state_trackers/g3dvl/vl_r16snorm_mc_buf_shaders.inc b/src/gallium/state_trackers/g3dvl/vl_r16snorm_mc_buf_shaders.inc index ef4a4b2add9..34d93e1df0a 100644 --- a/src/gallium/state_trackers/g3dvl/vl_r16snorm_mc_buf_shaders.inc +++ b/src/gallium/state_trackers/g3dvl/vl_r16snorm_mc_buf_shaders.inc @@ -615,7 +615,7 @@ static int vlCreateFragmentShaderFieldPMB ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); /* floor t3, t3 ; Get rid of fractional part */ - inst = vl_inst2(TGSI_OPCODE_FLOOR, TGSI_FILE_TEMPORARY, 3, TGSI_FILE_TEMPORARY, 3); + inst = vl_inst2(TGSI_OPCODE_FLR, TGSI_FILE_TEMPORARY, 3, TGSI_FILE_TEMPORARY, 3); ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); /* mul t3, t3, c1.y ; Multiply by 2 */ @@ -632,7 +632,7 @@ static int vlCreateFragmentShaderFieldPMB /* TODO: Move to conditional tex fetch on t3 instead of lerp */ /* lerp t1, t3, t1, t2 ; Choose between top and bottom fields based on Y % 2 */ - inst = vl_inst4(TGSI_OPCODE_LERP, TGSI_FILE_TEMPORARY, 1, TGSI_FILE_TEMPORARY, 3, TGSI_FILE_TEMPORARY, 1, TGSI_FILE_TEMPORARY, 2); + inst = vl_inst4(TGSI_OPCODE_LRP, TGSI_FILE_TEMPORARY, 1, TGSI_FILE_TEMPORARY, 3, TGSI_FILE_TEMPORARY, 1, TGSI_FILE_TEMPORARY, 2); ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); /* add o0, t0, t1 ; Add ref and differential to form final output */ @@ -969,7 +969,7 @@ static int vlCreateFragmentShaderFrameBMB } /* lerp t1, c1.x, t1, t2 ; Blend past and future texels */ - inst = vl_inst4(TGSI_OPCODE_LERP, TGSI_FILE_TEMPORARY, 1, TGSI_FILE_CONSTANT, 1, TGSI_FILE_TEMPORARY, 1, TGSI_FILE_TEMPORARY, 2); + inst = vl_inst4(TGSI_OPCODE_LRP, TGSI_FILE_TEMPORARY, 1, TGSI_FILE_CONSTANT, 1, TGSI_FILE_TEMPORARY, 1, TGSI_FILE_TEMPORARY, 2); inst.FullSrcRegisters[0].SrcRegister.SwizzleX = TGSI_SWIZZLE_X; inst.FullSrcRegisters[0].SrcRegister.SwizzleY = TGSI_SWIZZLE_X; inst.FullSrcRegisters[0].SrcRegister.SwizzleZ = TGSI_SWIZZLE_X; @@ -1116,7 +1116,7 @@ static int vlCreateFragmentShaderFieldBMB ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); /* floor t3, t3 ; Get rid of fractional part */ - inst = vl_inst2(TGSI_OPCODE_FLOOR, TGSI_FILE_TEMPORARY, 3, TGSI_FILE_TEMPORARY, 3); + inst = vl_inst2(TGSI_OPCODE_FLR, TGSI_FILE_TEMPORARY, 3, TGSI_FILE_TEMPORARY, 3); ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); /* mul t3, t3, c1.y ; Multiply by 2 */ @@ -1143,7 +1143,7 @@ static int vlCreateFragmentShaderFieldBMB /* TODO: Move to conditional tex fetch on t3 instead of lerp */ /* lerp t1, t3, t1, t2 ; Choose between top and bottom fields based on Y % 2 */ - inst = vl_inst4(TGSI_OPCODE_LERP, TGSI_FILE_TEMPORARY, 1, TGSI_FILE_TEMPORARY, 3, TGSI_FILE_TEMPORARY, 1, TGSI_FILE_TEMPORARY, 2); + inst = vl_inst4(TGSI_OPCODE_LRP, TGSI_FILE_TEMPORARY, 1, TGSI_FILE_TEMPORARY, 3, TGSI_FILE_TEMPORARY, 1, TGSI_FILE_TEMPORARY, 2); ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); /* @@ -1158,11 +1158,11 @@ static int vlCreateFragmentShaderFieldBMB /* TODO: Move to conditional tex fetch on t3 instead of lerp */ /* lerp t2, t3, t4, t5 ; Choose between top and bottom fields based on Y % 2 */ - inst = vl_inst4(TGSI_OPCODE_LERP, TGSI_FILE_TEMPORARY, 2, TGSI_FILE_TEMPORARY, 3, TGSI_FILE_TEMPORARY, 4, TGSI_FILE_TEMPORARY, 5); + inst = vl_inst4(TGSI_OPCODE_LRP, TGSI_FILE_TEMPORARY, 2, TGSI_FILE_TEMPORARY, 3, TGSI_FILE_TEMPORARY, 4, TGSI_FILE_TEMPORARY, 5); ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); /* lerp t1, c1.x, t1, t2 ; Blend past and future texels */ - inst = vl_inst4(TGSI_OPCODE_LERP, TGSI_FILE_TEMPORARY, 1, TGSI_FILE_CONSTANT, 1, TGSI_FILE_TEMPORARY, 1, TGSI_FILE_TEMPORARY, 2); + inst = vl_inst4(TGSI_OPCODE_LRP, TGSI_FILE_TEMPORARY, 1, TGSI_FILE_CONSTANT, 1, TGSI_FILE_TEMPORARY, 1, TGSI_FILE_TEMPORARY, 2); inst.FullSrcRegisters[0].SrcRegister.SwizzleX = TGSI_SWIZZLE_X; inst.FullSrcRegisters[0].SrcRegister.SwizzleY = TGSI_SWIZZLE_X; inst.FullSrcRegisters[0].SrcRegister.SwizzleZ = TGSI_SWIZZLE_X; diff --git a/src/gallium/state_trackers/wgl/SConscript b/src/gallium/state_trackers/wgl/SConscript index 69b88618ecb..b05944a33b3 100644 --- a/src/gallium/state_trackers/wgl/SConscript +++ b/src/gallium/state_trackers/wgl/SConscript @@ -18,20 +18,17 @@ if env['platform'] in ['windows']: ]) sources = [ - 'icd/stw_icd.c', - - 'wgl/stw_wgl.c', - - 'shared/stw_context.c', - 'shared/stw_device.c', - 'shared/stw_framebuffer.c', - 'shared/stw_pixelformat.c', - 'shared/stw_extensionsstring.c', - 'shared/stw_extswapinterval.c', - 'shared/stw_getprocaddress.c', - 'shared/stw_extgallium.c', - 'shared/stw_arbpixelformat.c', - 'shared/stw_tls.c', + 'stw_context.c', + 'stw_device.c', + 'stw_ext_extensionsstring.c', + 'stw_ext_gallium.c', + 'stw_ext_pixelformat.c', + 'stw_ext_swapinterval.c', + 'stw_framebuffer.c', + 'stw_getprocaddress.c', + 'stw_pixelformat.c', + 'stw_tls.c', + 'stw_wgl.c', ] wgl = env.ConvenienceLibrary( diff --git a/src/gallium/state_trackers/wgl/opengl32.def b/src/gallium/state_trackers/wgl/opengl32.def index 596417ed844..5daa6ddd413 100644 --- a/src/gallium/state_trackers/wgl/opengl32.def +++ b/src/gallium/state_trackers/wgl/opengl32.def @@ -376,6 +376,7 @@ EXPORTS DrvDescribePixelFormat DrvGetLayerPaletteEntries DrvGetProcAddress + DrvPresentBuffers DrvRealizeLayerPalette DrvReleaseContext DrvSetCallbackProcs diff --git a/src/gallium/state_trackers/wgl/opengl32.mingw.def b/src/gallium/state_trackers/wgl/opengl32.mingw.def index 1f03ea3b375..6ebb31a6f1b 100644 --- a/src/gallium/state_trackers/wgl/opengl32.mingw.def +++ b/src/gallium/state_trackers/wgl/opengl32.mingw.def @@ -375,6 +375,7 @@ EXPORTS DrvDescribePixelFormat = DrvDescribePixelFormat@16 DrvGetLayerPaletteEntries = DrvGetLayerPaletteEntries@20 DrvGetProcAddress = DrvGetProcAddress@4 + DrvPresentBuffers = DrvPresentBuffers@8 DrvRealizeLayerPalette = DrvRealizeLayerPalette@12 DrvReleaseContext = DrvReleaseContext@4 DrvSetCallbackProcs = DrvSetCallbackProcs@8 diff --git a/src/gallium/state_trackers/wgl/shared/stw_context.c b/src/gallium/state_trackers/wgl/shared/stw_context.c deleted file mode 100644 index 4968ecc692d..00000000000 --- a/src/gallium/state_trackers/wgl/shared/stw_context.c +++ /dev/null @@ -1,382 +0,0 @@ -/************************************************************************** - * - * Copyright 2008 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 <windows.h> - -#include "main/mtypes.h" -#include "main/context.h" -#include "pipe/p_compiler.h" -#include "pipe/p_context.h" -#include "state_tracker/st_context.h" -#include "state_tracker/st_public.h" - -#ifdef DEBUG -#include "trace/tr_screen.h" -#include "trace/tr_context.h" -#endif - -#include "shared/stw_device.h" -#include "shared/stw_winsys.h" -#include "shared/stw_framebuffer.h" -#include "shared/stw_pixelformat.h" -#include "stw_public.h" -#include "stw_context.h" -#include "stw_tls.h" - - -static INLINE struct stw_context * -stw_context(GLcontext *glctx) -{ - if(!glctx) - return NULL; - assert(glctx->DriverCtx); - return (struct stw_context *)glctx->DriverCtx; -} - -static INLINE struct stw_context * -stw_current_context(void) -{ - /* We must check if multiple threads are being used or GET_CURRENT_CONTEXT - * might return the current context of the thread first seen. */ - _glapi_check_multithread(); - - { - GET_CURRENT_CONTEXT( glctx ); - return stw_context(glctx); - } -} - -BOOL -stw_copy_context( - UINT_PTR hglrcSrc, - UINT_PTR hglrcDst, - UINT mask ) -{ - struct stw_context *src; - struct stw_context *dst; - BOOL ret = FALSE; - - pipe_mutex_lock( stw_dev->ctx_mutex ); - - src = stw_lookup_context_locked( hglrcSrc ); - dst = stw_lookup_context_locked( hglrcDst ); - - if (src && dst) { - /* FIXME */ - assert(0); - (void) src; - (void) dst; - (void) mask; - } - - pipe_mutex_unlock( stw_dev->ctx_mutex ); - - return ret; -} - -BOOL -stw_share_lists( - UINT_PTR hglrc1, - UINT_PTR hglrc2 ) -{ - struct stw_context *ctx1; - struct stw_context *ctx2; - BOOL ret = FALSE; - - pipe_mutex_lock( stw_dev->ctx_mutex ); - - ctx1 = stw_lookup_context_locked( hglrc1 ); - ctx2 = stw_lookup_context_locked( hglrc2 ); - - if (ctx1 && ctx2 && - ctx1->iPixelFormat == ctx2->iPixelFormat) { - ret = _mesa_share_state(ctx2->st->ctx, ctx1->st->ctx); - } - - pipe_mutex_unlock( stw_dev->ctx_mutex ); - - return ret; -} - -static void -stw_viewport(GLcontext * glctx, GLint x, GLint y, - GLsizei width, GLsizei height) -{ - struct stw_context *ctx = (struct stw_context *)glctx->DriverCtx; - struct stw_framebuffer *fb; - - fb = stw_framebuffer_from_hdc( ctx->hdc ); - if(fb) { - stw_framebuffer_update(fb); - stw_framebuffer_release(fb); - } -} - -UINT_PTR -stw_create_layer_context( - HDC hdc, - int iLayerPlane ) -{ - int iPixelFormat; - const struct stw_pixelformat_info *pfi; - GLvisual visual; - struct stw_context *ctx = NULL; - struct pipe_screen *screen = NULL; - struct pipe_context *pipe = NULL; - - if(!stw_dev) - return 0; - - if (iLayerPlane != 0) - return 0; - - iPixelFormat = GetPixelFormat(hdc); - if(!iPixelFormat) - return 0; - - pfi = stw_pixelformat_get_info( iPixelFormat - 1 ); - stw_pixelformat_visual(&visual, pfi); - - ctx = CALLOC_STRUCT( stw_context ); - if (ctx == NULL) - goto no_ctx; - - ctx->hdc = hdc; - ctx->iPixelFormat = iPixelFormat; - - screen = stw_dev->screen; - -#ifdef DEBUG - /* Unwrap screen */ - if(stw_dev->trace_running) - screen = trace_screen(screen)->screen; -#endif - - pipe = stw_dev->stw_winsys->create_context( screen ); - if (pipe == NULL) - goto no_pipe; - -#ifdef DEBUG - /* Wrap context */ - if(stw_dev->trace_running) - pipe = trace_context_create(stw_dev->screen, pipe); -#endif - - /* pass to stw_flush_frontbuffer as context_private */ - assert(!pipe->priv); - pipe->priv = hdc; - - ctx->st = st_create_context( pipe, &visual, NULL ); - if (ctx->st == NULL) - goto no_st_ctx; - - ctx->st->ctx->DriverCtx = ctx; - ctx->st->ctx->Driver.Viewport = stw_viewport; - - pipe_mutex_lock( stw_dev->ctx_mutex ); - ctx->hglrc = handle_table_add(stw_dev->ctx_table, ctx); - pipe_mutex_unlock( stw_dev->ctx_mutex ); - if (!ctx->hglrc) - goto no_hglrc; - - return ctx->hglrc; - -no_hglrc: - st_destroy_context(ctx->st); - goto no_pipe; /* st_context_destroy already destroys pipe */ -no_st_ctx: - pipe->destroy( pipe ); -no_pipe: - FREE(ctx); -no_ctx: - return 0; -} - -BOOL -stw_delete_context( - UINT_PTR hglrc ) -{ - struct stw_context *ctx ; - BOOL ret = FALSE; - - if (!stw_dev) - return FALSE; - - pipe_mutex_lock( stw_dev->ctx_mutex ); - ctx = stw_lookup_context_locked(hglrc); - handle_table_remove(stw_dev->ctx_table, hglrc); - pipe_mutex_unlock( stw_dev->ctx_mutex ); - - if (ctx) { - struct stw_context *curctx = stw_current_context(); - - /* Unbind current if deleting current context. */ - if (curctx == ctx) - st_make_current( NULL, NULL, NULL ); - - st_destroy_context(ctx->st); - FREE(ctx); - - ret = TRUE; - } - - return ret; -} - -BOOL -stw_release_context( - UINT_PTR hglrc ) -{ - struct stw_context *ctx; - - if (!stw_dev) - return FALSE; - - pipe_mutex_lock( stw_dev->ctx_mutex ); - ctx = stw_lookup_context_locked( hglrc ); - pipe_mutex_unlock( stw_dev->ctx_mutex ); - - if (!ctx) - return FALSE; - - /* The expectation is that ctx is the same context which is - * current for this thread. We should check that and return False - * if not the case. - */ - if (ctx != stw_current_context()) - return FALSE; - - if (stw_make_current( NULL, 0 ) == FALSE) - return FALSE; - - return TRUE; -} - - -UINT_PTR -stw_get_current_context( void ) -{ - struct stw_context *ctx; - - ctx = stw_current_context(); - if(!ctx) - return 0; - - return ctx->hglrc; -} - -HDC -stw_get_current_dc( void ) -{ - struct stw_context *ctx; - - ctx = stw_current_context(); - if(!ctx) - return NULL; - - return ctx->hdc; -} - -BOOL -stw_make_current( - HDC hdc, - UINT_PTR hglrc ) -{ - struct stw_context *curctx = NULL; - struct stw_context *ctx = NULL; - struct stw_framebuffer *fb = NULL; - - if (!stw_dev) - goto fail; - - curctx = stw_current_context(); - if (curctx != NULL) { - if (curctx->hglrc != hglrc) - st_flush(curctx->st, PIPE_FLUSH_RENDER_CACHE, NULL); - - /* Return if already current. */ - if (curctx->hglrc == hglrc && curctx->hdc == hdc) { - ctx = curctx; - fb = stw_framebuffer_from_hdc( hdc ); - goto success; - } - } - - if (hdc == NULL || hglrc == 0) { - return st_make_current( NULL, NULL, NULL ); - } - - pipe_mutex_lock( stw_dev->ctx_mutex ); - ctx = stw_lookup_context_locked( hglrc ); - pipe_mutex_unlock( stw_dev->ctx_mutex ); - if(!ctx) - goto fail; - - fb = stw_framebuffer_from_hdc( hdc ); - if(!fb) { - /* Applications should call SetPixelFormat before creating a context, - * but not all do, and the opengl32 runtime seems to use a default pixel - * format in some cases, so we must create a framebuffer for those here - */ - int iPixelFormat = GetPixelFormat(hdc); - if(iPixelFormat) - fb = stw_framebuffer_create( hdc, iPixelFormat ); - if(!fb) - goto fail; - } - - if(fb->iPixelFormat != ctx->iPixelFormat) - goto fail; - - /* Lazy allocation of the frame buffer */ - if(!stw_framebuffer_allocate(fb)) - goto fail; - - /* Bind the new framebuffer */ - ctx->hdc = hdc; - - /* pass to stw_flush_frontbuffer as context_private */ - ctx->st->pipe->priv = hdc; - - if(!st_make_current( ctx->st, fb->stfb, fb->stfb )) - goto fail; - -success: - assert(fb); - if(fb) { - stw_framebuffer_update(fb); - stw_framebuffer_release(fb); - } - - return TRUE; - -fail: - if(fb) - stw_framebuffer_release(fb); - st_make_current( NULL, NULL, NULL ); - return FALSE; -} diff --git a/src/gallium/state_trackers/wgl/shared/stw_public.h b/src/gallium/state_trackers/wgl/shared/stw_public.h deleted file mode 100644 index 7fe9cfb3561..00000000000 --- a/src/gallium/state_trackers/wgl/shared/stw_public.h +++ /dev/null @@ -1,73 +0,0 @@ -/************************************************************************** - * - * Copyright 2009 VMware, Inc. - * 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. - * - **************************************************************************/ - -#ifndef STW_PUBLIC_H -#define STW_PUBLIC_H - -#include <windows.h> - -BOOL stw_copy_context( UINT_PTR hglrcSrc, - UINT_PTR hglrcDst, - UINT mask ); - -UINT_PTR stw_create_layer_context( HDC hdc, - int iLayerPlane ); - -BOOL stw_share_lists( UINT_PTR hglrc1, UINT_PTR hglrc2 ); - -BOOL stw_delete_context( UINT_PTR hglrc ); - -BOOL -stw_release_context( UINT_PTR dhglrc ); - -UINT_PTR stw_get_current_context( void ); - -HDC stw_get_current_dc( void ); - -BOOL stw_make_current( HDC hdc, UINT_PTR hglrc ); - -BOOL stw_swap_buffers( HDC hdc ); - -BOOL -stw_swap_layer_buffers( HDC hdc, UINT fuPlanes ); - -PROC stw_get_proc_address( LPCSTR lpszProc ); - -int stw_pixelformat_describe( HDC hdc, - int iPixelFormat, - UINT nBytes, - LPPIXELFORMATDESCRIPTOR ppfd ); - -int stw_pixelformat_get( HDC hdc ); - -BOOL stw_pixelformat_set( HDC hdc, - int iPixelFormat ); - -int stw_pixelformat_choose( HDC hdc, - CONST PIXELFORMATDESCRIPTOR *ppfd ); - -#endif diff --git a/src/gallium/state_trackers/wgl/icd/stw_icd.c b/src/gallium/state_trackers/wgl/stw_context.c index 347f40aa06b..f2f0264844a 100644 --- a/src/gallium/state_trackers/wgl/icd/stw_icd.c +++ b/src/gallium/state_trackers/wgl/stw_context.c @@ -26,18 +26,49 @@ **************************************************************************/ #include <windows.h> -#include <stdio.h> -#include "GL/gl.h" - -#include "util/u_debug.h" -#include "pipe/p_thread.h" - -#include "shared/stw_public.h" -#include "icd/stw_icd.h" +#include "main/mtypes.h" +#include "main/context.h" +#include "pipe/p_compiler.h" +#include "pipe/p_context.h" +#include "state_tracker/st_context.h" +#include "state_tracker/st_public.h" + +#ifdef DEBUG +#include "trace/tr_screen.h" +#include "trace/tr_context.h" +#endif + +#include "stw_icd.h" +#include "stw_device.h" +#include "stw_winsys.h" +#include "stw_framebuffer.h" +#include "stw_pixelformat.h" +#include "stw_context.h" +#include "stw_tls.h" + + +static INLINE struct stw_context * +stw_context(GLcontext *glctx) +{ + if(!glctx) + return NULL; + assert(glctx->DriverCtx); + return (struct stw_context *)glctx->DriverCtx; +} -#define DBG 0 +static INLINE struct stw_context * +stw_current_context(void) +{ + /* We must check if multiple threads are being used or GET_CURRENT_CONTEXT + * might return the current context of the thread first seen. */ + _glapi_check_multithread(); + { + GET_CURRENT_CONTEXT( glctx ); + return stw_context(glctx); + } +} BOOL APIENTRY DrvCopyContext( @@ -45,24 +76,64 @@ DrvCopyContext( DHGLRC dhrcDest, UINT fuMask ) { - return stw_copy_context(dhrcSource, dhrcDest, fuMask); -} + struct stw_context *src; + struct stw_context *dst; + BOOL ret = FALSE; + pipe_mutex_lock( stw_dev->ctx_mutex ); + + src = stw_lookup_context_locked( dhrcSource ); + dst = stw_lookup_context_locked( dhrcDest ); + + if (src && dst) { + /* FIXME */ + assert(0); + (void) src; + (void) dst; + (void) fuMask; + } -DHGLRC APIENTRY -DrvCreateLayerContext( - HDC hdc, - INT iLayerPlane ) + pipe_mutex_unlock( stw_dev->ctx_mutex ); + + return ret; +} + +BOOL APIENTRY +DrvShareLists( + DHGLRC dhglrc1, + DHGLRC dhglrc2 ) { - DHGLRC r; + struct stw_context *ctx1; + struct stw_context *ctx2; + BOOL ret = FALSE; + + pipe_mutex_lock( stw_dev->ctx_mutex ); - r = stw_create_layer_context( hdc, iLayerPlane ); + ctx1 = stw_lookup_context_locked( dhglrc1 ); + ctx2 = stw_lookup_context_locked( dhglrc2 ); + + if (ctx1 && ctx2 && + ctx1->iPixelFormat == ctx2->iPixelFormat) { + ret = _mesa_share_state(ctx2->st->ctx, ctx1->st->ctx); + } + + pipe_mutex_unlock( stw_dev->ctx_mutex ); - if (DBG) - debug_printf( "%s( %p, %i ) = %u\n", - __FUNCTION__, hdc, iLayerPlane, r ); + return ret; +} + +static void +stw_viewport(GLcontext * glctx, GLint x, GLint y, + GLsizei width, GLsizei height) +{ + struct stw_context *ctx = (struct stw_context *)glctx->DriverCtx; + struct stw_framebuffer *fb; - return r; + fb = stw_framebuffer_from_hdc( ctx->hdc ); + if(fb) { + stw_framebuffer_update(fb); + stw_framebuffer_release(fb); + } } DHGLRC APIENTRY @@ -72,114 +143,253 @@ DrvCreateContext( return DrvCreateLayerContext( hdc, 0 ); } -BOOL APIENTRY -DrvDeleteContext( - DHGLRC dhglrc ) +DHGLRC APIENTRY +DrvCreateLayerContext( + HDC hdc, + INT iLayerPlane ) { - BOOL r; + int iPixelFormat; + const struct stw_pixelformat_info *pfi; + GLvisual visual; + struct stw_context *ctx = NULL; + struct pipe_screen *screen = NULL; + struct pipe_context *pipe = NULL; - r = stw_delete_context( dhglrc ); + if(!stw_dev) + return 0; - if (DBG) - debug_printf( "%s( %u ) = %u\n", - __FUNCTION__, dhglrc, r ); + if (iLayerPlane != 0) + return 0; + + iPixelFormat = GetPixelFormat(hdc); + if(!iPixelFormat) + return 0; - return r; + pfi = stw_pixelformat_get_info( iPixelFormat - 1 ); + stw_pixelformat_visual(&visual, pfi); + + ctx = CALLOC_STRUCT( stw_context ); + if (ctx == NULL) + goto no_ctx; + + ctx->hdc = hdc; + ctx->iPixelFormat = iPixelFormat; + + screen = stw_dev->screen; + +#ifdef DEBUG + /* Unwrap screen */ + if(stw_dev->trace_running) + screen = trace_screen(screen)->screen; +#endif + + pipe = stw_dev->stw_winsys->create_context( screen ); + if (pipe == NULL) + goto no_pipe; + +#ifdef DEBUG + /* Wrap context */ + if(stw_dev->trace_running) + pipe = trace_context_create(stw_dev->screen, pipe); +#endif + + /* pass to stw_flush_frontbuffer as context_private */ + assert(!pipe->priv); + pipe->priv = hdc; + + ctx->st = st_create_context( pipe, &visual, NULL ); + if (ctx->st == NULL) + goto no_st_ctx; + + ctx->st->ctx->DriverCtx = ctx; + ctx->st->ctx->Driver.Viewport = stw_viewport; + + pipe_mutex_lock( stw_dev->ctx_mutex ); + ctx->dhglrc = handle_table_add(stw_dev->ctx_table, ctx); + pipe_mutex_unlock( stw_dev->ctx_mutex ); + if (!ctx->dhglrc) + goto no_hglrc; + + return ctx->dhglrc; + +no_hglrc: + st_destroy_context(ctx->st); + goto no_pipe; /* st_context_destroy already destroys pipe */ +no_st_ctx: + pipe->destroy( pipe ); +no_pipe: + FREE(ctx); +no_ctx: + return 0; } BOOL APIENTRY -DrvDescribeLayerPlane( - HDC hdc, - INT iPixelFormat, - INT iLayerPlane, - UINT nBytes, - LPLAYERPLANEDESCRIPTOR plpd ) +DrvDeleteContext( + DHGLRC dhglrc ) { - if (DBG) - debug_printf( "%s\n", __FUNCTION__ ); + struct stw_context *ctx ; + BOOL ret = FALSE; + + if (!stw_dev) + return FALSE; - return FALSE; -} + pipe_mutex_lock( stw_dev->ctx_mutex ); + ctx = stw_lookup_context_locked(dhglrc); + handle_table_remove(stw_dev->ctx_table, dhglrc); + pipe_mutex_unlock( stw_dev->ctx_mutex ); -LONG APIENTRY -DrvDescribePixelFormat( - HDC hdc, - INT iPixelFormat, - ULONG cjpfd, - PIXELFORMATDESCRIPTOR *ppfd ) -{ - LONG r; + if (ctx) { + struct stw_context *curctx = stw_current_context(); + + /* Unbind current if deleting current context. */ + if (curctx == ctx) + st_make_current( NULL, NULL, NULL ); - r = stw_pixelformat_describe( hdc, iPixelFormat, cjpfd, ppfd ); + st_destroy_context(ctx->st); + FREE(ctx); - if (DBG) - debug_printf( "%s( %p, %d, %u, %p ) = %d\n", - __FUNCTION__, hdc, iPixelFormat, cjpfd, ppfd, r ); + ret = TRUE; + } - return r; + return ret; } -int APIENTRY -DrvGetLayerPaletteEntries( - HDC hdc, - INT iLayerPlane, - INT iStart, - INT cEntries, - COLORREF *pcr ) +BOOL APIENTRY +DrvReleaseContext( + DHGLRC dhglrc ) { - if (DBG) - debug_printf( "%s\n", __FUNCTION__ ); + struct stw_context *ctx; - return 0; -} + if (!stw_dev) + return FALSE; -PROC APIENTRY -DrvGetProcAddress( - LPCSTR lpszProc ) -{ - PROC r; + pipe_mutex_lock( stw_dev->ctx_mutex ); + ctx = stw_lookup_context_locked( dhglrc ); + pipe_mutex_unlock( stw_dev->ctx_mutex ); - r = stw_get_proc_address( lpszProc ); + if (!ctx) + return FALSE; + + /* The expectation is that ctx is the same context which is + * current for this thread. We should check that and return False + * if not the case. + */ + if (ctx != stw_current_context()) + return FALSE; - if (DBG) - debug_printf( "%s( \"%s\" ) = %p\n", __FUNCTION__, lpszProc, r ); + if (stw_make_current( NULL, 0 ) == FALSE) + return FALSE; - return r; + return TRUE; } -BOOL APIENTRY -DrvRealizeLayerPalette( - HDC hdc, - INT iLayerPlane, - BOOL bRealize ) + +DHGLRC +stw_get_current_context( void ) { - if (DBG) - debug_printf( "%s\n", __FUNCTION__ ); + struct stw_context *ctx; - return FALSE; + ctx = stw_current_context(); + if(!ctx) + return 0; + + return ctx->dhglrc; } -BOOL APIENTRY -DrvReleaseContext( - DHGLRC dhglrc ) +HDC +stw_get_current_dc( void ) { - return stw_release_context(dhglrc); + struct stw_context *ctx; + + ctx = stw_current_context(); + if(!ctx) + return NULL; + + return ctx->hdc; } -void APIENTRY -DrvSetCallbackProcs( - INT nProcs, - PROC *pProcs ) +BOOL +stw_make_current( + HDC hdc, + DHGLRC dhglrc ) { - if (DBG) - debug_printf( "%s( %d, %p )\n", __FUNCTION__, nProcs, pProcs ); + struct stw_context *curctx = NULL; + struct stw_context *ctx = NULL; + struct stw_framebuffer *fb = NULL; - return; -} + if (!stw_dev) + goto fail; + + curctx = stw_current_context(); + if (curctx != NULL) { + if (curctx->dhglrc != dhglrc) + st_flush(curctx->st, PIPE_FLUSH_RENDER_CACHE, NULL); + + /* Return if already current. */ + if (curctx->dhglrc == dhglrc && curctx->hdc == hdc) { + ctx = curctx; + fb = stw_framebuffer_from_hdc( hdc ); + goto success; + } + } + + if (hdc == NULL || dhglrc == 0) { + return st_make_current( NULL, NULL, NULL ); + } + + pipe_mutex_lock( stw_dev->ctx_mutex ); + ctx = stw_lookup_context_locked( dhglrc ); + pipe_mutex_unlock( stw_dev->ctx_mutex ); + if(!ctx) + goto fail; + + fb = stw_framebuffer_from_hdc( hdc ); + if(!fb) { + /* Applications should call SetPixelFormat before creating a context, + * but not all do, and the opengl32 runtime seems to use a default pixel + * format in some cases, so we must create a framebuffer for those here + */ + int iPixelFormat = GetPixelFormat(hdc); + if(iPixelFormat) + fb = stw_framebuffer_create( hdc, iPixelFormat ); + if(!fb) + goto fail; + } + + if(fb->iPixelFormat != ctx->iPixelFormat) + goto fail; + + /* Lazy allocation of the frame buffer */ + if(!stw_framebuffer_allocate(fb)) + goto fail; + + /* Bind the new framebuffer */ + ctx->hdc = hdc; + + /* pass to stw_flush_frontbuffer as context_private */ + ctx->st->pipe->priv = hdc; + + if(!st_make_current( ctx->st, fb->stfb, fb->stfb )) + goto fail; + +success: + assert(fb); + if(fb) { + stw_framebuffer_update(fb); + stw_framebuffer_release(fb); + } + + return TRUE; +fail: + if(fb) + stw_framebuffer_release(fb); + st_make_current( NULL, NULL, NULL ); + return FALSE; +} /** - * Although WGL allows different dispatch entrypoints per context + * Although WGL allows different dispatch entrypoints per context */ static const GLCLTPROCTABLE cpt = { @@ -524,7 +734,6 @@ static const GLCLTPROCTABLE cpt = } }; - PGLCLTPROCTABLE APIENTRY DrvSetContext( HDC hdc, @@ -532,86 +741,9 @@ DrvSetContext( PFN_SETPROCTABLE pfnSetProcTable ) { PGLCLTPROCTABLE r = (PGLCLTPROCTABLE)&cpt; - + if (!stw_make_current( hdc, dhglrc )) r = NULL; - - if (DBG) - debug_printf( "%s( 0x%p, %u, 0x%p ) = %p\n", - __FUNCTION__, hdc, dhglrc, pfnSetProcTable, r ); - - return r; -} - -int APIENTRY -DrvSetLayerPaletteEntries( - HDC hdc, - INT iLayerPlane, - INT iStart, - INT cEntries, - CONST COLORREF *pcr ) -{ - if (DBG) - debug_printf( "%s\n", __FUNCTION__ ); - - return 0; -} - -BOOL APIENTRY -DrvSetPixelFormat( - HDC hdc, - LONG iPixelFormat ) -{ - BOOL r; - - r = stw_pixelformat_set( hdc, iPixelFormat ); - - if (DBG) - debug_printf( "%s( %p, %d ) = %s\n", __FUNCTION__, hdc, iPixelFormat, r ? "TRUE" : "FALSE" ); return r; } - -BOOL APIENTRY -DrvShareLists( - DHGLRC dhglrc1, - DHGLRC dhglrc2 ) -{ - if (DBG) - debug_printf( "%s\n", __FUNCTION__ ); - - return stw_share_lists(dhglrc1, dhglrc2); -} - -BOOL APIENTRY -DrvSwapBuffers( - HDC hdc ) -{ - if (DBG) - debug_printf( "%s( %p )\n", __FUNCTION__, hdc ); - - return stw_swap_buffers( hdc ); -} - -BOOL APIENTRY -DrvSwapLayerBuffers( - HDC hdc, - UINT fuPlanes ) -{ - if (DBG) - debug_printf( "%s\n", __FUNCTION__ ); - - return stw_swap_layer_buffers( hdc, fuPlanes ); -} - -BOOL APIENTRY -DrvValidateVersion( - ULONG ulVersion ) -{ - if (DBG) - debug_printf( "%s( %u )\n", __FUNCTION__, ulVersion ); - - /* TODO: get the expected version from the winsys */ - - return ulVersion == 1; -} diff --git a/src/gallium/state_trackers/wgl/shared/stw_context.h b/src/gallium/state_trackers/wgl/stw_context.h index 166471de5eb..256c27e21ef 100644 --- a/src/gallium/state_trackers/wgl/shared/stw_context.h +++ b/src/gallium/state_trackers/wgl/stw_context.h @@ -35,9 +35,15 @@ struct st_context; struct stw_context { struct st_context *st; - UINT_PTR hglrc; + DHGLRC dhglrc; int iPixelFormat; HDC hdc; }; +DHGLRC stw_get_current_context( void ); + +HDC stw_get_current_dc( void ); + +BOOL stw_make_current( HDC hdc, DHGLRC dhglrc ); + #endif /* STW_CONTEXT_H */ diff --git a/src/gallium/state_trackers/wgl/shared/stw_device.c b/src/gallium/state_trackers/wgl/stw_device.c index 0b6954915a6..985b8f0456a 100644 --- a/src/gallium/state_trackers/wgl/shared/stw_device.c +++ b/src/gallium/state_trackers/wgl/stw_device.c @@ -29,6 +29,7 @@ #include "glapi/glthread.h" #include "util/u_debug.h" +#include "util/u_math.h" #include "pipe/p_screen.h" #include "state_tracker/st_public.h" @@ -37,12 +38,12 @@ #include "trace/tr_texture.h" #endif -#include "shared/stw_device.h" -#include "shared/stw_winsys.h" -#include "shared/stw_pixelformat.h" -#include "shared/stw_public.h" -#include "shared/stw_tls.h" -#include "shared/stw_framebuffer.h" +#include "stw_device.h" +#include "stw_winsys.h" +#include "stw_pixelformat.h" +#include "stw_icd.h" +#include "stw_tls.h" +#include "stw_framebuffer.h" #ifdef WIN32_THREADS extern _glthread_Mutex OneTimeLock; @@ -62,38 +63,28 @@ stw_flush_frontbuffer(struct pipe_screen *screen, struct pipe_surface *surface, void *context_private ) { - const struct stw_winsys *stw_winsys = stw_dev->stw_winsys; HDC hdc = (HDC)context_private; struct stw_framebuffer *fb; fb = stw_framebuffer_from_hdc( hdc ); - /* fb can be NULL if window was destroyed already */ - if (fb) { + if (!fb) { + /* fb can be NULL if window was destroyed already */ + return; + } + #if DEBUG - { - struct pipe_surface *surface2; - - if(!st_get_framebuffer_surface( fb->stfb, ST_SURFACE_FRONT_LEFT, &surface2 )) - assert(0); - else - assert(surface2 == surface); - } -#endif + { + /* ensure that a random surface was not passed to us */ + struct pipe_surface *surface2; -#ifdef DEBUG - if(stw_dev->trace_running) { - screen = trace_screen(screen)->screen; - surface = trace_surface(surface)->surface; - } -#endif - } - - stw_winsys->flush_frontbuffer(screen, surface, hdc); - - if(fb) { - stw_framebuffer_update(fb); - stw_framebuffer_release(fb); + if(!st_get_framebuffer_surface( fb->stfb, ST_SURFACE_FRONT_LEFT, &surface2 )) + assert(0); + else + assert(surface2 == surface); } +#endif + + stw_framebuffer_present_locked(hdc, fb, ST_SURFACE_FRONT_LEFT); } @@ -126,6 +117,9 @@ stw_init(const struct stw_winsys *stw_winsys) if(!screen) goto error1; + if(stw_winsys->get_adapter_luid) + stw_winsys->get_adapter_luid(screen, &stw_dev->AdapterLuid); + #ifdef DEBUG stw_dev->screen = trace_screen_create(screen); stw_dev->trace_running = stw_dev->screen != screen ? TRUE : FALSE; @@ -182,7 +176,7 @@ stw_cleanup(void) /* Ensure all contexts are destroyed */ i = handle_table_get_first_handle(stw_dev->ctx_table); while (i) { - stw_delete_context(i); + DrvDeleteContext(i); i = handle_table_get_next_handle(stw_dev->ctx_table, i); } handle_table_destroy(stw_dev->ctx_table); @@ -212,7 +206,7 @@ stw_cleanup(void) struct stw_context * -stw_lookup_context_locked( UINT_PTR dhglrc ) +stw_lookup_context_locked( DHGLRC dhglrc ) { if (dhglrc == 0) return NULL; @@ -223,3 +217,28 @@ stw_lookup_context_locked( UINT_PTR dhglrc ) return (struct stw_context *) handle_table_get(stw_dev->ctx_table, dhglrc); } + +void APIENTRY +DrvSetCallbackProcs( + INT nProcs, + PROC *pProcs ) +{ + size_t size; + + if (stw_dev == NULL) + return; + + size = MIN2(nProcs * sizeof *pProcs, sizeof stw_dev->callbacks); + memcpy(&stw_dev->callbacks, pProcs, size); + + return; +} + + +BOOL APIENTRY +DrvValidateVersion( + ULONG ulVersion ) +{ + /* TODO: get the expected version from the winsys */ + return ulVersion == 1; +} diff --git a/src/gallium/state_trackers/wgl/shared/stw_device.h b/src/gallium/state_trackers/wgl/stw_device.h index e1bb9518dd1..0bf3b0da825 100644 --- a/src/gallium/state_trackers/wgl/shared/stw_device.h +++ b/src/gallium/state_trackers/wgl/stw_device.h @@ -29,11 +29,10 @@ #define STW_DEVICE_H_ -#include <windows.h> - #include "pipe/p_compiler.h" #include "pipe/p_thread.h" #include "util/u_handle_table.h" +#include "stw_icd.h" #include "stw_pixelformat.h" @@ -53,10 +52,14 @@ struct stw_device boolean trace_running; #endif + LUID AdapterLuid; + struct stw_pixelformat_info pixelformats[STW_MAX_PIXELFORMATS]; unsigned pixelformat_count; unsigned pixelformat_extended_count; + GLCALLBACKTABLE callbacks; + pipe_mutex ctx_mutex; struct handle_table *ctx_table; @@ -69,7 +72,7 @@ struct stw_device }; struct stw_context * -stw_lookup_context_locked( UINT_PTR hglrc ); +stw_lookup_context_locked( DHGLRC hglrc ); extern struct stw_device *stw_dev; diff --git a/src/gallium/state_trackers/wgl/shared/stw_extensionsstring.c b/src/gallium/state_trackers/wgl/stw_ext_extensionsstring.c index 62c859e1f92..62c859e1f92 100644 --- a/src/gallium/state_trackers/wgl/shared/stw_extensionsstring.c +++ b/src/gallium/state_trackers/wgl/stw_ext_extensionsstring.c diff --git a/src/gallium/state_trackers/wgl/shared/stw_extgallium.c b/src/gallium/state_trackers/wgl/stw_ext_gallium.c index fc22737d7e3..fb30ec5dba9 100644 --- a/src/gallium/state_trackers/wgl/shared/stw_extgallium.c +++ b/src/gallium/state_trackers/wgl/stw_ext_gallium.c @@ -27,9 +27,9 @@ #include "pipe/p_screen.h" -#include "stw_public.h" #include "stw_device.h" #include "stw_winsys.h" +#include "stw_ext_gallium.h" #ifdef DEBUG #include "trace/tr_screen.h" diff --git a/src/gallium/state_trackers/wgl/shared/stw_extgallium.h b/src/gallium/state_trackers/wgl/stw_ext_gallium.h index cc35f2bb7fe..cc35f2bb7fe 100644 --- a/src/gallium/state_trackers/wgl/shared/stw_extgallium.h +++ b/src/gallium/state_trackers/wgl/stw_ext_gallium.h diff --git a/src/gallium/state_trackers/wgl/shared/stw_arbpixelformat.c b/src/gallium/state_trackers/wgl/stw_ext_pixelformat.c index 0e2d4076993..8a9995aba8e 100644 --- a/src/gallium/state_trackers/wgl/shared/stw_arbpixelformat.c +++ b/src/gallium/state_trackers/wgl/stw_ext_pixelformat.c @@ -43,7 +43,6 @@ #include "pipe/p_compiler.h" #include "util/u_memory.h" -#include "stw_public.h" #include "stw_pixelformat.h" diff --git a/src/gallium/state_trackers/wgl/shared/stw_extswapinterval.c b/src/gallium/state_trackers/wgl/stw_ext_swapinterval.c index 9eac6a1d09d..9eac6a1d09d 100644 --- a/src/gallium/state_trackers/wgl/shared/stw_extswapinterval.c +++ b/src/gallium/state_trackers/wgl/stw_ext_swapinterval.c diff --git a/src/gallium/state_trackers/wgl/shared/stw_framebuffer.c b/src/gallium/state_trackers/wgl/stw_framebuffer.c index b8956bb5509..8a3e11b6b40 100644 --- a/src/gallium/state_trackers/wgl/shared/stw_framebuffer.c +++ b/src/gallium/state_trackers/wgl/stw_framebuffer.c @@ -1,8 +1,8 @@ /************************************************************************** - * - * Copyright 2008 Tungsten Graphics, Inc., Cedar Park, Texas. + * + * Copyright 2008-2009 Vmware, Inc. * 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 @@ -10,19 +10,19 @@ * 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 + * IN NO EVENT SHALL VMWARE 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 <windows.h> @@ -38,9 +38,9 @@ #include "trace/tr_texture.h" #endif +#include "stw_icd.h" #include "stw_framebuffer.h" #include "stw_device.h" -#include "stw_public.h" #include "stw_winsys.h" #include "stw_tls.h" @@ -83,6 +83,9 @@ stw_framebuffer_destroy_locked( *link = fb->next; fb->next = NULL; + if(fb->shared_surface) + stw_dev->stw_winsys->shared_surface_close(stw_dev->screen, fb->shared_surface); + st_unreference_framebuffer(fb->stfb); pipe_mutex_unlock( fb->mutex ); @@ -106,13 +109,18 @@ static INLINE void stw_framebuffer_get_size( struct stw_framebuffer *fb ) { unsigned width, height; - RECT rect; + RECT client_rect; + RECT window_rect; + POINT client_pos; assert(fb->hWnd); - GetClientRect( fb->hWnd, &rect ); - width = rect.right - rect.left; - height = rect.bottom - rect.top; + /* Get the client area size. */ + GetClientRect( fb->hWnd, &client_rect ); + assert(client_rect.left == 0); + assert(client_rect.top == 0); + width = client_rect.right - client_rect.left; + height = client_rect.bottom - client_rect.top; if(width < 1) width = 1; @@ -124,6 +132,31 @@ stw_framebuffer_get_size( struct stw_framebuffer *fb ) fb->width = width; fb->height = height; } + + client_pos.x = 0; + client_pos.y = 0; + ClientToScreen(fb->hWnd, &client_pos); + + GetWindowRect(fb->hWnd, &window_rect); + + fb->client_rect.left = client_pos.x - window_rect.left; + fb->client_rect.top = client_pos.y - window_rect.top; + fb->client_rect.right = fb->client_rect.left + fb->width; + fb->client_rect.bottom = fb->client_rect.top + fb->height; + +#if 0 + debug_printf("\n"); + debug_printf("%s: client_position = (%i, %i)\n", + __FUNCTION__, client_pos.x, client_pos.y); + debug_printf("%s: window_rect = (%i, %i) - (%i, %i)\n", + __FUNCTION__, + window_rect.left, window_rect.top, + window_rect.right, window_rect.bottom); + debug_printf("%s: client_rect = (%i, %i) - (%i, %i)\n", + __FUNCTION__, + fb->client_rect.left, fb->client_rect.top, + fb->client_rect.right, fb->client_rect.bottom); +#endif } @@ -155,6 +188,7 @@ stw_call_window_proc( * can be masked out by the application. */ LPWINDOWPOS lpWindowPos = (LPWINDOWPOS)pParams->lParam; if((lpWindowPos->flags & SWP_SHOWWINDOW) || + !(lpWindowPos->flags & SWP_NOMOVE) || !(lpWindowPos->flags & SWP_NOSIZE)) { fb = stw_framebuffer_from_hwnd( pParams->hwnd ); if(fb) { @@ -379,10 +413,10 @@ stw_framebuffer_from_hwnd( } -BOOL -stw_pixelformat_set( +BOOL APIENTRY +DrvSetPixelFormat( HDC hdc, - int iPixelFormat ) + LONG iPixelFormat ) { uint count; uint index; @@ -435,35 +469,24 @@ stw_pixelformat_get( } -BOOL -stw_swap_buffers( - HDC hdc ) +BOOL APIENTRY +DrvPresentBuffers(HDC hdc, PGLPRESENTBUFFERSDATA data) { struct stw_framebuffer *fb; struct pipe_screen *screen; struct pipe_surface *surface; + unsigned surface_index; + BOOL ret = FALSE; fb = stw_framebuffer_from_hdc( hdc ); if (fb == NULL) return FALSE; - if (!(fb->pfi->pfd.dwFlags & PFD_DOUBLEBUFFER)) { - stw_framebuffer_release(fb); - return TRUE; - } - - /* If we're swapping the buffer associated with the current context - * we have to flush any pending rendering commands first. - */ - st_notify_swapbuffers( fb->stfb ); - screen = stw_dev->screen; - - if(!st_get_framebuffer_surface( fb->stfb, ST_SURFACE_BACK_LEFT, &surface )) { - /* FIXME: this shouldn't happen, but does on glean */ - stw_framebuffer_release(fb); - return FALSE; - } + + surface_index = (unsigned)(uintptr_t)data->pPrivateData; + if(!st_get_framebuffer_surface( fb->stfb, surface_index, &surface )) + goto fail; #ifdef DEBUG if(stw_dev->trace_running) { @@ -472,22 +495,127 @@ stw_swap_buffers( } #endif - stw_dev->stw_winsys->flush_frontbuffer( screen, surface, hdc ); - + if(data->hSharedSurface != fb->hSharedSurface) { + if(fb->shared_surface) { + stw_dev->stw_winsys->shared_surface_close(screen, fb->shared_surface); + fb->shared_surface = NULL; + } + + fb->hSharedSurface = data->hSharedSurface; + + if(data->hSharedSurface && + stw_dev->stw_winsys->shared_surface_open) { + fb->shared_surface = stw_dev->stw_winsys->shared_surface_open(screen, fb->hSharedSurface); + } + } + + if(fb->shared_surface) { + stw_dev->stw_winsys->compose(screen, + surface, + fb->shared_surface, + &fb->client_rect, + data->PresentHistoryToken); + } + else { + stw_dev->stw_winsys->present( screen, surface, hdc ); + } + + ret = TRUE; + +fail: + stw_framebuffer_update(fb); + stw_framebuffer_release(fb); - - return TRUE; + + return ret; } +/** + * Queue a composition. + * + * It will drop the lock on success. + */ BOOL -stw_swap_layer_buffers( +stw_framebuffer_present_locked(HDC hdc, + struct stw_framebuffer *fb, + unsigned surface_index) +{ + if(stw_dev->callbacks.wglCbPresentBuffers && + stw_dev->stw_winsys->compose) { + GLCBPRESENTBUFFERSDATA data; + + memset(&data, 0, sizeof data); + data.magic1 = 2; + data.magic2 = 0; + data.AdapterLuid = stw_dev->AdapterLuid; + data.rect = fb->client_rect; + data.pPrivateData = (void *)(uintptr_t)surface_index; + + stw_framebuffer_release(fb); + + return stw_dev->callbacks.wglCbPresentBuffers(hdc, &data); + } + else { + struct pipe_screen *screen = stw_dev->screen; + struct pipe_surface *surface; + + if(!st_get_framebuffer_surface( fb->stfb, surface_index, &surface )) { + /* FIXME: this shouldn't happen, but does on glean */ + stw_framebuffer_release(fb); + return FALSE; + } + +#ifdef DEBUG + if(stw_dev->trace_running) { + screen = trace_screen(screen)->screen; + surface = trace_surface(surface)->surface; + } +#endif + + stw_dev->stw_winsys->present( screen, surface, hdc ); + + stw_framebuffer_update(fb); + + stw_framebuffer_release(fb); + + return TRUE; + } +} + + +BOOL APIENTRY +DrvSwapBuffers( + HDC hdc ) +{ + struct stw_framebuffer *fb; + + fb = stw_framebuffer_from_hdc( hdc ); + if (fb == NULL) + return FALSE; + + if (!(fb->pfi->pfd.dwFlags & PFD_DOUBLEBUFFER)) { + stw_framebuffer_release(fb); + return TRUE; + } + + /* If we're swapping the buffer associated with the current context + * we have to flush any pending rendering commands first. + */ + st_notify_swapbuffers( fb->stfb ); + + return stw_framebuffer_present_locked(hdc, fb, ST_SURFACE_BACK_LEFT); +} + + +BOOL APIENTRY +DrvSwapLayerBuffers( HDC hdc, UINT fuPlanes ) { if(fuPlanes & WGL_SWAP_MAIN_PLANE) - return stw_swap_buffers(hdc); + return DrvSwapBuffers(hdc); return FALSE; } diff --git a/src/gallium/state_trackers/wgl/shared/stw_framebuffer.h b/src/gallium/state_trackers/wgl/stw_framebuffer.h index 13d29f37e48..5afbe749086 100644 --- a/src/gallium/state_trackers/wgl/shared/stw_framebuffer.h +++ b/src/gallium/state_trackers/wgl/stw_framebuffer.h @@ -73,9 +73,20 @@ struct stw_framebuffer /* FIXME: Make this work for multiple contexts bound to the same framebuffer */ boolean must_resize; + unsigned width; unsigned height; + /** + * Client area rectangle, relative to the window upper-left corner. + * + * @sa GLCBPRESENTBUFFERSDATA::rect. + */ + RECT client_rect; + + HANDLE hSharedSurface; + struct stw_shared_surface *shared_surface; + /** * This is protected by stw_device::fb_mutex, not the mutex above. * @@ -126,6 +137,11 @@ BOOL stw_framebuffer_allocate( struct stw_framebuffer *fb ); +BOOL +stw_framebuffer_present_locked(HDC hdc, + struct stw_framebuffer *fb, + unsigned surface_index); + void stw_framebuffer_update( struct stw_framebuffer *fb); diff --git a/src/gallium/state_trackers/wgl/shared/stw_getprocaddress.c b/src/gallium/state_trackers/wgl/stw_getprocaddress.c index 879ced925a5..8875dc22f3d 100644 --- a/src/gallium/state_trackers/wgl/shared/stw_getprocaddress.c +++ b/src/gallium/state_trackers/wgl/stw_getprocaddress.c @@ -33,8 +33,7 @@ #include <GL/wglext.h> #include "glapi/glapi.h" -#include "stw_public.h" -#include "stw_extgallium.h" +#include "stw_ext_gallium.h" struct stw_extension_entry { @@ -68,8 +67,8 @@ static const struct stw_extension_entry stw_extension_entries[] = { { NULL, NULL } }; -PROC -stw_get_proc_address( +PROC APIENTRY +DrvGetProcAddress( LPCSTR lpszProc ) { const struct stw_extension_entry *entry; diff --git a/src/gallium/state_trackers/wgl/icd/stw_icd.h b/src/gallium/state_trackers/wgl/stw_icd.h index cbc1a665481..02eb543fef0 100644 --- a/src/gallium/state_trackers/wgl/icd/stw_icd.h +++ b/src/gallium/state_trackers/wgl/stw_icd.h @@ -1,6 +1,6 @@ /************************************************************************** * - * Copyright 2008 Tungsten Graphics, Inc., Cedar Park, Texas. + * Copyright 2008-2009 Vmware, Inc. * All Rights Reserved. * * Permission is hereby granted, free of charge, to any person obtaining a @@ -18,7 +18,7 @@ * 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 + * IN NO EVENT SHALL VMWARE 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. @@ -388,6 +388,113 @@ typedef struct _GLCLTPROCTABLE typedef VOID (APIENTRY * PFN_SETPROCTABLE)(PGLCLTPROCTABLE); +/** + * Presentation data passed to opengl32!wglCbPresentBuffers. + * + * Pure software drivers don't need to worry about this -- if they stick to the + * GDI API then will integrate with the Desktop Window Manager (DWM) without + * problems. Hardware drivers, however, cannot present directly to the primary + * surface while the DWM is active, as DWM gets exclusive access to the primary + * surface. + * + * Proper DWM integration requires: + * - advertise the PFD_SUPPORT_COMPOSITION flag + * - redirect glFlush/glfinish/wglSwapBuffers into a surface shared with the + * DWM process. + * + * @sa http://www.opengl.org/pipeline/article/vol003_7/ + * @sa http://blogs.msdn.com/greg_schechter/archive/2006/05/02/588934.aspx + */ +typedef struct _GLCBPRESENTBUFFERSDATA +{ + /** + * wglCbPresentBuffers enforces this to be 2. + */ + DWORD magic1; + + /** + * wglCbPresentBuffers enforces to be 0 or 1, but it is most commonly + * set to 0. + */ + DWORD magic2; + + /** + * Locally unique identifier (LUID) of the graphics adapter. + * + * This should contain the value returned by D3DKMTOpenAdapterFromHdc. It + * is passed to dwmapi!DwmpDxGetWindowSharedSurface in order to obtain + * the shared surface handle for the bound drawable (window). + * + * @sa http://msdn.microsoft.com/en-us/library/ms799177.aspx + */ + LUID AdapterLuid; + + /** + * This is passed unmodified to DrvPresentBuffers + */ + LPVOID pPrivateData; + + /** + * Client area rectangle to update, relative to the window upper-left corner. + */ + RECT rect; +} GLCBPRESENTBUFFERSDATA, *PGLCBPRESENTBUFFERSDATA; + +/** + * Callbacks supplied to DrvSetCallbackProcs by the OpenGL runtime. + * + * Pointers to several callback functions in opengl32.dll. + */ +typedef struct _GLCALLBACKTABLE +{ + /** Unused */ + PROC wglCbSetCurrentValue; + + /** Unused */ + PROC wglCbGetCurrentValue; + + /** Unused */ + PROC wglCbGetDhglrc; + + /** Unused */ + PROC wglCbGetDdHandle; + + /** + * Queue a present composition. + * + * Makes the runtime call DrvPresentBuffers with the composition information. + */ + BOOL (APIENTRY *wglCbPresentBuffers)(HDC hdc, PGLCBPRESENTBUFFERSDATA data); + +} GLCALLBACKTABLE; + +typedef struct _GLPRESENTBUFFERSDATA +{ + /** + * The shared surface handle. + * + * Return by dwmapi!DwmpDxGetWindowSharedSurface. + * + * @sa http://channel9.msdn.com/forums/TechOff/251261-Help-Getting-the-shared-window-texture-out-of-DWM-/ + */ + HANDLE hSharedSurface; + + LUID AdapterLuid; + + /** + * Present history token. + * + * This is returned by dwmapi!DwmpDxGetWindowSharedSurface and + * should be passed to D3DKMTRender in D3DKMT_RENDER::PresentHistoryToken. + * + * @sa http://msdn.microsoft.com/en-us/library/ms799176.aspx + */ + ULONGLONG PresentHistoryToken; + + /** Same as GLCBPRESENTBUFFERSDATA::pPrivateData */ + LPVOID pPrivateData; +} GLPRESENTBUFFERSDATA, *PGLPRESENTBUFFERSDATA; + BOOL APIENTRY DrvCopyContext( DHGLRC dhrcSource, @@ -435,6 +542,9 @@ DrvGetProcAddress( LPCSTR lpszProc ); BOOL APIENTRY +DrvPresentBuffers(HDC hdc, PGLPRESENTBUFFERSDATA data); + +BOOL APIENTRY DrvRealizeLayerPalette( HDC hdc, INT iLayerPlane, diff --git a/src/gallium/state_trackers/wgl/shared/stw_pixelformat.c b/src/gallium/state_trackers/wgl/stw_pixelformat.c index c296744838b..7abe5d9f7fa 100644 --- a/src/gallium/state_trackers/wgl/shared/stw_pixelformat.c +++ b/src/gallium/state_trackers/wgl/stw_pixelformat.c @@ -34,9 +34,9 @@ #include "util/u_debug.h" +#include "stw_icd.h" #include "stw_device.h" #include "stw_pixelformat.h" -#include "stw_public.h" #include "stw_tls.h" @@ -154,8 +154,11 @@ stw_pixelformat_add( pfi->pfd.dwFlags = PFD_SUPPORT_OPENGL; /* TODO: also support non-native pixel formats */ - pfi->pfd.dwFlags |= PFD_DRAW_TO_WINDOW ; - + pfi->pfd.dwFlags |= PFD_DRAW_TO_WINDOW; + + /* See http://www.opengl.org/pipeline/article/vol003_7/ */ + pfi->pfd.dwFlags |= PFD_SUPPORT_COMPOSITION; + if (doublebuffer) pfi->pfd.dwFlags |= PFD_DOUBLEBUFFER | PFD_SWAP_COPY; @@ -288,12 +291,12 @@ stw_pixelformat_visual(GLvisual *visual, } -int -stw_pixelformat_describe( +LONG APIENTRY +DrvDescribePixelFormat( HDC hdc, - int iPixelFormat, - UINT nBytes, - LPPIXELFORMATDESCRIPTOR ppfd ) + INT iPixelFormat, + ULONG cjpfd, + PIXELFORMATDESCRIPTOR *ppfd ) { uint count; uint index; @@ -306,7 +309,7 @@ stw_pixelformat_describe( if (ppfd == NULL) return count; - if (index >= count || nBytes != sizeof( PIXELFORMATDESCRIPTOR )) + if (index >= count || cjpfd != sizeof( PIXELFORMATDESCRIPTOR )) return 0; pfi = stw_pixelformat_get_info( index ); @@ -316,6 +319,52 @@ stw_pixelformat_describe( return count; } +BOOL APIENTRY +DrvDescribeLayerPlane( + HDC hdc, + INT iPixelFormat, + INT iLayerPlane, + UINT nBytes, + LPLAYERPLANEDESCRIPTOR plpd ) +{ + assert(0); + return FALSE; +} + +int APIENTRY +DrvGetLayerPaletteEntries( + HDC hdc, + INT iLayerPlane, + INT iStart, + INT cEntries, + COLORREF *pcr ) +{ + assert(0); + return 0; +} + +int APIENTRY +DrvSetLayerPaletteEntries( + HDC hdc, + INT iLayerPlane, + INT iStart, + INT cEntries, + CONST COLORREF *pcr ) +{ + assert(0); + return 0; +} + +BOOL APIENTRY +DrvRealizeLayerPalette( + HDC hdc, + INT iLayerPlane, + BOOL bRealize ) +{ + assert(0); + return FALSE; +} + /* Only used by the wgl code, but have it here to avoid exporting the * pixelformat.h functionality. */ diff --git a/src/gallium/state_trackers/wgl/shared/stw_pixelformat.h b/src/gallium/state_trackers/wgl/stw_pixelformat.h index bec429231b2..3a690b35bad 100644 --- a/src/gallium/state_trackers/wgl/shared/stw_pixelformat.h +++ b/src/gallium/state_trackers/wgl/stw_pixelformat.h @@ -30,6 +30,10 @@ #include <windows.h> +#ifndef PFD_SUPPORT_COMPOSITION +#define PFD_SUPPORT_COMPOSITION 0x00008000 +#endif + #include "main/mtypes.h" #include "pipe/p_compiler.h" @@ -62,4 +66,11 @@ void stw_pixelformat_visual(GLvisual *visual, const struct stw_pixelformat_info *pfi ); +int +stw_pixelformat_choose( HDC hdc, + CONST PIXELFORMATDESCRIPTOR *ppfd ); + +int +stw_pixelformat_get(HDC hdc); + #endif /* STW_PIXELFORMAT_H */ diff --git a/src/gallium/state_trackers/wgl/shared/stw_tls.c b/src/gallium/state_trackers/wgl/stw_tls.c index 4bd6a9289c9..4bd6a9289c9 100644 --- a/src/gallium/state_trackers/wgl/shared/stw_tls.c +++ b/src/gallium/state_trackers/wgl/stw_tls.c diff --git a/src/gallium/state_trackers/wgl/shared/stw_tls.h b/src/gallium/state_trackers/wgl/stw_tls.h index fbf8b1cbee4..fbf8b1cbee4 100644 --- a/src/gallium/state_trackers/wgl/shared/stw_tls.h +++ b/src/gallium/state_trackers/wgl/stw_tls.h diff --git a/src/gallium/state_trackers/wgl/wgl/stw_wgl.c b/src/gallium/state_trackers/wgl/stw_wgl.c index a131292f7ae..bb199fdd252 100644 --- a/src/gallium/state_trackers/wgl/wgl/stw_wgl.c +++ b/src/gallium/state_trackers/wgl/stw_wgl.c @@ -28,7 +28,9 @@ #include <windows.h> #include "util/u_debug.h" -#include "shared/stw_public.h" +#include "stw_icd.h" +#include "stw_context.h" +#include "stw_pixelformat.h" #include "stw_wgl.h" @@ -38,16 +40,16 @@ wglCopyContext( HGLRC hglrcDst, UINT mask ) { - return stw_copy_context( (UINT_PTR)hglrcSrc, - (UINT_PTR)hglrcDst, - mask ); + return DrvCopyContext( (DHGLRC)(UINT_PTR)hglrcSrc, + (DHGLRC)(UINT_PTR)hglrcDst, + mask ); } WINGDIAPI HGLRC APIENTRY wglCreateContext( HDC hdc ) { - return wglCreateLayerContext(hdc, 0); + return (HGLRC) DrvCreateContext(hdc); } WINGDIAPI HGLRC APIENTRY @@ -55,21 +57,21 @@ wglCreateLayerContext( HDC hdc, int iLayerPlane ) { - return (HGLRC) stw_create_layer_context( hdc, iLayerPlane ); + return (HGLRC) DrvCreateLayerContext( hdc, iLayerPlane ); } WINGDIAPI BOOL APIENTRY wglDeleteContext( HGLRC hglrc ) { - return stw_delete_context( (UINT_PTR)hglrc ); + return DrvDeleteContext((DHGLRC)(UINT_PTR)hglrc ); } WINGDIAPI HGLRC APIENTRY wglGetCurrentContext( VOID ) { - return (HGLRC)stw_get_current_context(); + return (HGLRC)(UINT_PTR)stw_get_current_context(); } WINGDIAPI HDC APIENTRY @@ -83,7 +85,7 @@ wglMakeCurrent( HDC hdc, HGLRC hglrc ) { - return stw_make_current( hdc, (UINT_PTR)hglrc ); + return DrvSetContext( hdc, (DHGLRC)(UINT_PTR)hglrc, NULL ) ? TRUE : FALSE; } @@ -91,7 +93,7 @@ WINGDIAPI BOOL APIENTRY wglSwapBuffers( HDC hdc ) { - return stw_swap_buffers( hdc ); + return DrvSwapBuffers( hdc ); } @@ -100,14 +102,14 @@ wglSwapLayerBuffers( HDC hdc, UINT fuPlanes ) { - return stw_swap_layer_buffers( hdc, fuPlanes ); + return DrvSwapLayerBuffers( hdc, fuPlanes ); } WINGDIAPI PROC APIENTRY wglGetProcAddress( LPCSTR lpszProc ) { - return stw_get_proc_address( lpszProc ); + return DrvGetProcAddress( lpszProc ); } @@ -141,7 +143,7 @@ wglDescribePixelFormat( UINT nBytes, LPPIXELFORMATDESCRIPTOR ppfd ) { - return stw_pixelformat_describe( hdc, iPixelFormat, nBytes, ppfd ); + return DrvDescribePixelFormat( hdc, iPixelFormat, nBytes, ppfd ); } WINGDIAPI int APIENTRY @@ -160,7 +162,7 @@ wglSetPixelFormat( if (ppfd->nSize != sizeof( PIXELFORMATDESCRIPTOR )) return FALSE; - return stw_pixelformat_set( hdc, iPixelFormat ); + return DrvSetPixelFormat( hdc, iPixelFormat ); } @@ -186,7 +188,8 @@ wglShareLists( HGLRC hglrc1, HGLRC hglrc2 ) { - return stw_share_lists( (UINT_PTR)hglrc1, (UINT_PTR)hglrc2);; + return DrvShareLists((DHGLRC)(UINT_PTR)hglrc1, + (DHGLRC)(UINT_PTR)hglrc2); } WINGDIAPI BOOL APIENTRY @@ -264,15 +267,7 @@ wglDescribeLayerPlane( UINT nBytes, LPLAYERPLANEDESCRIPTOR plpd ) { - (void) hdc; - (void) iPixelFormat; - (void) iLayerPlane; - (void) nBytes; - (void) plpd; - - assert( 0 ); - - return FALSE; + return DrvDescribeLayerPlane(hdc, iPixelFormat, iLayerPlane, nBytes, plpd); } WINGDIAPI int APIENTRY @@ -283,15 +278,7 @@ wglSetLayerPaletteEntries( int cEntries, CONST COLORREF *pcr ) { - (void) hdc; - (void) iLayerPlane; - (void) iStart; - (void) cEntries; - (void) pcr; - - assert( 0 ); - - return 0; + return DrvSetLayerPaletteEntries(hdc, iLayerPlane, iStart, cEntries, pcr); } WINGDIAPI int APIENTRY @@ -302,15 +289,7 @@ wglGetLayerPaletteEntries( int cEntries, COLORREF *pcr ) { - (void) hdc; - (void) iLayerPlane; - (void) iStart; - (void) cEntries; - (void) pcr; - - assert( 0 ); - - return 0; + return DrvGetLayerPaletteEntries(hdc, iLayerPlane, iStart, cEntries, pcr); } WINGDIAPI BOOL APIENTRY diff --git a/src/gallium/state_trackers/wgl/wgl/stw_wgl.h b/src/gallium/state_trackers/wgl/stw_wgl.h index a98179944aa..a98179944aa 100644 --- a/src/gallium/state_trackers/wgl/wgl/stw_wgl.h +++ b/src/gallium/state_trackers/wgl/stw_wgl.h diff --git a/src/gallium/state_trackers/wgl/shared/stw_winsys.h b/src/gallium/state_trackers/wgl/stw_winsys.h index c0bf82c9ed7..1ead47d6e63 100644 --- a/src/gallium/state_trackers/wgl/shared/stw_winsys.h +++ b/src/gallium/state_trackers/wgl/stw_winsys.h @@ -1,6 +1,6 @@ /************************************************************************** * - * Copyright 2008 Tungsten Graphics, Inc., Cedar Park, Texas. + * Copyright 2008-2009 Vmware, Inc. * All Rights Reserved. * * Permission is hereby granted, free of charge, to any person obtaining a @@ -18,7 +18,7 @@ * 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 + * IN NO EVENT SHALL VMWARE 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. @@ -36,6 +36,8 @@ struct pipe_screen; struct pipe_context; struct pipe_surface; +struct stw_shared_surface; + struct stw_winsys { struct pipe_screen * @@ -44,10 +46,52 @@ struct stw_winsys struct pipe_context * (*create_context)( struct pipe_screen *screen ); + /** + * Present the color buffer to the window associated with the device context. + */ + void + (*present)( struct pipe_screen *screen, + struct pipe_surface *surf, + HDC hDC ); + + /** + * Locally unique identifier (LUID) of the graphics adapter. + * + * @sa GLCBPRESENTBUFFERSDATA::AdapterLuid; + */ + boolean + (*get_adapter_luid)( struct pipe_screen *screen, + LUID *pAdapterLuid ); + + /** + * Open a shared surface (optional). + * + * @sa GLCBPRESENTBUFFERSDATA::hSharedSurface; + */ + struct stw_shared_surface * + (*shared_surface_open)(struct pipe_screen *screen, + HANDLE hSharedSurface); + + /** + * Open a shared surface (optional). + */ + void + (*shared_surface_close)(struct pipe_screen *screen, + struct stw_shared_surface *surface); + + /** + * Compose into a shared (optional). + * + * Blit the color buffer into a shared surface. + * + * @sa GLPRESENTBUFFERSDATA::PresentHistoryToken. + */ void - (*flush_frontbuffer)( struct pipe_screen *screen, - struct pipe_surface *surf, - HDC hDC ); + (*compose)( struct pipe_screen *screen, + struct pipe_surface *src, + struct stw_shared_surface *dest, + LPCRECT pRect, + ULONGLONG PresentHistoryToken ); }; boolean diff --git a/src/gallium/state_trackers/xorg/xorg_composite.c b/src/gallium/state_trackers/xorg/xorg_composite.c index 15c955450d0..9d15a615f15 100644 --- a/src/gallium/state_trackers/xorg/xorg_composite.c +++ b/src/gallium/state_trackers/xorg/xorg_composite.c @@ -12,9 +12,9 @@ struct xorg_composite_blend { int op:8; unsigned rgb_src_factor:5; /**< PIPE_BLENDFACTOR_x */ - unsigned rgb_dst_factor:5; /**< PIPE_BLENDFACTOR_x */ - unsigned alpha_src_factor:5; /**< PIPE_BLENDFACTOR_x */ + + unsigned rgb_dst_factor:5; /**< PIPE_BLENDFACTOR_x */ unsigned alpha_dst_factor:5; /**< PIPE_BLENDFACTOR_x */ }; @@ -57,24 +57,6 @@ pixel_to_float4(Pixel pixel, float *color) color[3] = ((float)a) / 255.; } -static INLINE void -render_pixel_to_float4(PictFormatPtr format, - CARD32 pixel, float *color) -{ - CARD32 r, g, b, a; - - debug_assert(format->type == PictTypeDirect); - - r = (pixel >> format->direct.red) & format->direct.redMask; - g = (pixel >> format->direct.green) & format->direct.greenMask; - b = (pixel >> format->direct.blue) & format->direct.blueMask; - a = (pixel >> format->direct.alpha) & format->direct.alphaMask; - color[0] = ((float)r) / ((float)format->direct.redMask); - color[1] = ((float)g) / ((float)format->direct.greenMask); - color[2] = ((float)b) / ((float)format->direct.blueMask); - color[3] = ((float)a) / ((float)format->direct.alphaMask); -} - struct acceleration_info { int op : 16; int with_mask : 1; @@ -150,24 +132,22 @@ setup_vertex_data0(struct exa_context *ctx, int srcX, int srcY, int maskX, int maskY, int dstX, int dstY, int width, int height) { - float vertices[4][2][4]; - /* 1st vertex */ - setup_vertex0(vertices[0], dstX, dstY, + setup_vertex0(ctx->vertices2[0], dstX, dstY, ctx->solid_color); /* 2nd vertex */ - setup_vertex0(vertices[1], dstX + width, dstY, + setup_vertex0(ctx->vertices2[1], dstX + width, dstY, ctx->solid_color); /* 3rd vertex */ - setup_vertex0(vertices[2], dstX + width, dstY + height, + setup_vertex0(ctx->vertices2[2], dstX + width, dstY + height, ctx->solid_color); /* 4th vertex */ - setup_vertex0(vertices[3], dstX, dstY + height, + setup_vertex0(ctx->vertices2[3], dstX, dstY + height, ctx->solid_color); return pipe_user_buffer_create(ctx->pipe->screen, - vertices, - sizeof(vertices)); + ctx->vertices2, + sizeof(ctx->vertices2)); } static INLINE void @@ -189,7 +169,6 @@ setup_vertex_data1(struct exa_context *ctx, int srcX, int srcY, int maskX, int maskY, int dstX, int dstY, int width, int height) { - float vertices[4][2][4]; float s0, t0, s1, t1; struct pipe_texture *src = ctx->bound_textures[0]; @@ -199,21 +178,21 @@ setup_vertex_data1(struct exa_context *ctx, t1 = srcY + height / src->height[0]; /* 1st vertex */ - setup_vertex1(vertices[0], dstX, dstY, + setup_vertex1(ctx->vertices2[0], dstX, dstY, s0, t0); /* 2nd vertex */ - setup_vertex1(vertices[1], dstX + width, dstY, + setup_vertex1(ctx->vertices2[1], dstX + width, dstY, s1, t0); /* 3rd vertex */ - setup_vertex1(vertices[2], dstX + width, dstY + height, + setup_vertex1(ctx->vertices2[2], dstX + width, dstY + height, s1, t1); /* 4th vertex */ - setup_vertex1(vertices[3], dstX, dstY + height, + setup_vertex1(ctx->vertices2[3], dstX, dstY + height, s0, t1); return pipe_user_buffer_create(ctx->pipe->screen, - vertices, - sizeof(vertices)); + ctx->vertices2, + sizeof(ctx->vertices2)); } static struct pipe_buffer * @@ -222,24 +201,22 @@ setup_vertex_data_tex(struct exa_context *ctx, float s0, float t0, float s1, float t1, float z) { - float vertices[4][2][4]; - /* 1st vertex */ - setup_vertex1(vertices[0], x0, y0, + setup_vertex1(ctx->vertices2[0], x0, y0, s0, t0); /* 2nd vertex */ - setup_vertex1(vertices[1], x1, y0, + setup_vertex1(ctx->vertices2[1], x1, y0, s1, t0); /* 3rd vertex */ - setup_vertex1(vertices[2], x1, y1, + setup_vertex1(ctx->vertices2[2], x1, y1, s1, t1); /* 4th vertex */ - setup_vertex1(vertices[3], x0, y1, + setup_vertex1(ctx->vertices2[3], x0, y1, s0, t1); return pipe_user_buffer_create(ctx->pipe->screen, - vertices, - sizeof(vertices)); + ctx->vertices2, + sizeof(ctx->vertices2)); } @@ -269,7 +246,6 @@ setup_vertex_data2(struct exa_context *ctx, int srcX, int srcY, int maskX, int maskY, int dstX, int dstY, int width, int height) { - float vertices[4][3][4]; float st0[4], st1[4]; struct pipe_texture *src = ctx->bound_textures[0]; struct pipe_texture *mask = ctx->bound_textures[0]; @@ -285,21 +261,21 @@ setup_vertex_data2(struct exa_context *ctx, st1[3] = maskY + height / mask->height[0]; /* 1st vertex */ - setup_vertex2(vertices[0], dstX, dstY, + setup_vertex2(ctx->vertices3[0], dstX, dstY, st0[0], st0[1], st1[0], st1[1]); /* 2nd vertex */ - setup_vertex2(vertices[1], dstX + width, dstY, + setup_vertex2(ctx->vertices3[1], dstX + width, dstY, st0[2], st0[1], st1[2], st1[1]); /* 3rd vertex */ - setup_vertex2(vertices[2], dstX + width, dstY + height, + setup_vertex2(ctx->vertices3[2], dstX + width, dstY + height, st0[2], st0[3], st1[2], st1[3]); /* 4th vertex */ - setup_vertex2(vertices[3], dstX, dstY + height, + setup_vertex2(ctx->vertices3[3], dstX, dstY + height, st0[0], st0[3], st1[0], st1[3]); return pipe_user_buffer_create(ctx->pipe->screen, - vertices, - sizeof(vertices)); + ctx->vertices3, + sizeof(ctx->vertices3)); } boolean xorg_composite_accelerated(int op, @@ -311,22 +287,25 @@ boolean xorg_composite_accelerated(int op, unsigned accel_ops_count = sizeof(accelerated_ops)/sizeof(struct acceleration_info); - - /*FIXME: currently accel is disabled */ - return FALSE; - - if (pSrcPicture) { - /* component alpha not supported */ - if (pSrcPicture->componentAlpha) + if (pSrcPicture->pSourcePict) { + /* Gradients not yet supported */ + if (pSrcPicture->pSourcePict->type != SourcePictTypeSolidFill) return FALSE; - /* fills not supported */ - if (pSrcPicture->pSourcePict) + + /* Solid source with mask not yet handled properly */ + if (pMaskPicture) return FALSE; } for (i = 0; i < accel_ops_count; ++i) { if (op == accelerated_ops[i].op) { - if (pMaskPicture && !accelerated_ops[i].with_mask) + /* Check for unsupported component alpha */ + if ((pSrcPicture->componentAlpha && + !accelerated_ops[i].component_alpha) || + (pMaskPicture && + (!accelerated_ops[i].with_mask || + (pMaskPicture->componentAlpha && + !accelerated_ops[i].component_alpha)))) return FALSE; return TRUE; } @@ -359,6 +338,9 @@ bind_framebuffer_state(struct exa_context *exa, struct exa_pixmap_priv *pDst) state.zsbuf = 0; cso_set_framebuffer(exa->cso, &state); + + /* we do fire and forget for the framebuffer, this is the forget part */ + pipe_surface_reference(&surface, NULL); } enum AxisOrientation { @@ -391,7 +373,7 @@ bind_viewport_state(struct exa_context *exa, struct exa_pixmap_priv *pDst) int width = pDst->tex->width[0]; int height = pDst->tex->height[0]; - debug_printf("Bind viewport (%d, %d)\n", width, height); + /*debug_printf("Bind viewport (%d, %d)\n", width, height);*/ set_viewport(exa, width, height, Y0_TOP); } @@ -400,14 +382,9 @@ static void bind_blend_state(struct exa_context *exa, int op, PicturePtr pSrcPicture, PicturePtr pMaskPicture) { - boolean component_alpha = (pSrcPicture) ? - pSrcPicture->componentAlpha : FALSE; struct xorg_composite_blend blend_opt; struct pipe_blend_state blend; - if (component_alpha) { - op = PictOpOver; - } blend_opt = blend_for_op(op); memset(&blend, 0, sizeof(struct pipe_blend_state)); @@ -448,9 +425,9 @@ bind_shaders(struct exa_context *exa, int op, if (pSrcPicture->pSourcePict->type == SourcePictTypeSolidFill) { fs_traits |= FS_SOLID_FILL; vs_traits |= VS_SOLID_FILL; - render_pixel_to_float4(pSrcPicture->pFormat, - pSrcPicture->pSourcePict->solidFill.color, - exa->solid_color); + debug_assert(pSrcPicture->format == PICT_a8r8g8b8); + pixel_to_float4(pSrcPicture->pSourcePict->solidFill.color, + exa->solid_color); exa->has_solid_color = TRUE; } else { debug_assert("!gradients not supported"); @@ -488,6 +465,12 @@ bind_samplers(struct exa_context *exa, int op, memset(&src_sampler, 0, sizeof(struct pipe_sampler_state)); memset(&mask_sampler, 0, sizeof(struct pipe_sampler_state)); + if ((pSrc && exa->pipe->is_texture_referenced(exa->pipe, pSrc->tex, 0, 0) & + PIPE_REFERENCED_FOR_WRITE) || + (pMask && exa->pipe->is_texture_referenced(exa->pipe, pMask->tex, 0, 0) & + PIPE_REFERENCED_FOR_WRITE)) + exa->pipe->flush(exa->pipe, PIPE_FLUSH_RENDER_CACHE, NULL); + if (pSrcPicture && pSrc) { unsigned src_wrap = render_repeat_to_gallium( pSrcPicture->repeatType); @@ -653,17 +636,12 @@ boolean xorg_solid_bind_state(struct exa_context *exa, exa->solid_color[3] = 1.f; +#if 0 debug_printf("Color Pixel=(%d, %d, %d, %d), RGBA=(%f, %f, %f, %f)\n", (fg >> 24) & 0xff, (fg >> 16) & 0xff, (fg >> 8) & 0xff, (fg >> 0) & 0xff, exa->solid_color[0], exa->solid_color[1], exa->solid_color[2], exa->solid_color[3]); - -#if 0 - exa->solid_color[0] = 1.f; - exa->solid_color[1] = 0.f; - exa->solid_color[2] = 0.f; - exa->solid_color[3] = 1.f; #endif vs_traits = VS_SOLID_FILL; @@ -680,7 +658,7 @@ boolean xorg_solid_bind_state(struct exa_context *exa, cso_set_vertex_shader_handle(exa->cso, shader.vs); cso_set_fragment_shader_handle(exa->cso, shader.fs); - return FALSE; + return TRUE; } void xorg_solid(struct exa_context *exa, @@ -689,31 +667,26 @@ void xorg_solid(struct exa_context *exa, { struct pipe_context *pipe = exa->pipe; struct pipe_buffer *buf = 0; - float vertices[4][2][4]; - - x0 = 10; y0 = 10; - x1 = 300; y1 = 300; /* 1st vertex */ - setup_vertex0(vertices[0], x0, y0, + setup_vertex0(exa->vertices2[0], x0, y0, exa->solid_color); /* 2nd vertex */ - setup_vertex0(vertices[1], x1, y0, + setup_vertex0(exa->vertices2[1], x1, y0, exa->solid_color); /* 3rd vertex */ - setup_vertex0(vertices[2], x1, y1, + setup_vertex0(exa->vertices2[2], x1, y1, exa->solid_color); /* 4th vertex */ - setup_vertex0(vertices[3], x0, y1, + setup_vertex0(exa->vertices2[3], x0, y1, exa->solid_color); buf = pipe_user_buffer_create(exa->pipe->screen, - vertices, - sizeof(vertices)); + exa->vertices2, + sizeof(exa->vertices2)); if (buf) { - debug_printf("Drawing buf is %p\n", buf); util_draw_vertex_buffer(pipe, buf, 0, PIPE_PRIM_TRIANGLE_FAN, 4, /* verts */ @@ -844,11 +817,6 @@ static void renderer_copy_texture(struct exa_context *exa, assert(dst->width[0] != 0); assert(dst->height[0] != 0); -#if 0 - debug_printf("copy texture [%f, %f, %f, %f], [%f, %f, %f, %f]\n", - sx1, sy1, sx2, sy2, dx1, dy1, dx2, dy2); -#endif - #if 1 s0 = sx1 / src->width[0]; s1 = sx2 / src->width[0]; @@ -861,8 +829,16 @@ static void renderer_copy_texture(struct exa_context *exa, t1 = 1; #endif - assert(screen->is_format_supported(screen, dst_surf->format, PIPE_TEXTURE_2D, - PIPE_TEXTURE_USAGE_RENDER_TARGET, 0)); +#if 1 + debug_printf("copy texture src=[%f, %f, %f, %f], dst=[%f, %f, %f, %f], tex=[%f, %f, %f, %f]\n", + sx1, sy1, sx2, sy2, dx1, dy1, dx2, dy2, + s0, t0, s1, t1); +#endif + + assert(screen->is_format_supported(screen, dst_surf->format, + PIPE_TEXTURE_2D, + PIPE_TEXTURE_USAGE_RENDER_TARGET, + 0)); /* save state (restored below) */ cso_save_blend(exa->cso); @@ -907,6 +883,8 @@ static void renderer_copy_texture(struct exa_context *exa, /* texture */ cso_set_sampler_textures(exa->cso, 1, &src); + bind_rasterizer_state(exa); + /* shaders */ shader = xorg_shaders_get(exa->shaders, VS_COMPOSITE, @@ -926,6 +904,8 @@ static void renderer_copy_texture(struct exa_context *exa, fb.cbufs[i] = 0; } cso_set_framebuffer(exa->cso, &fb); + setup_vs_constant_buffer(exa, fb.width, fb.height); + setup_fs_constant_buffer(exa); /* draw quad */ buf = setup_vertex_data_tex(exa, @@ -955,6 +935,61 @@ static void renderer_copy_texture(struct exa_context *exa, pipe_surface_reference(&dst_surf, NULL); } + +static struct pipe_texture * +create_sampler_texture(struct exa_context *ctx, + struct pipe_texture *src) +{ + enum pipe_format format; + struct pipe_context *pipe = ctx->pipe; + struct pipe_screen *screen = pipe->screen; + struct pipe_texture *pt; + struct pipe_texture templ; + + pipe->flush(pipe, PIPE_FLUSH_RENDER_CACHE, NULL); + + /* the coming in texture should already have that invariance */ + debug_assert(screen->is_format_supported(screen, src->format, + PIPE_TEXTURE_2D, + PIPE_TEXTURE_USAGE_SAMPLER, 0)); + + format = src->format; + + memset(&templ, 0, sizeof(templ)); + templ.target = PIPE_TEXTURE_2D; + templ.format = format; + templ.last_level = 0; + templ.width[0] = src->width[0]; + templ.height[0] = src->height[0]; + templ.depth[0] = 1; + pf_get_block(format, &templ.block); + templ.tex_usage = PIPE_TEXTURE_USAGE_SAMPLER; + + pt = screen->texture_create(screen, &templ); + + debug_assert(!pt || pipe_is_referenced(&pt->reference)); + + if (!pt) + return NULL; + + { + /* copy source framebuffer surface into texture */ + struct pipe_surface *ps_read = screen->get_tex_surface( + screen, src, 0, 0, 0, PIPE_BUFFER_USAGE_GPU_READ); + struct pipe_surface *ps_tex = screen->get_tex_surface( + screen, pt, 0, 0, 0, PIPE_BUFFER_USAGE_GPU_WRITE ); + pipe->surface_copy(pipe, + ps_tex, /* dest */ + 0, 0, /* destx/y */ + ps_read, + 0, 0, src->width[0], src->height[0]); + pipe_surface_reference(&ps_read, NULL); + pipe_surface_reference(&ps_tex, NULL); + } + + return pt; +} + void xorg_copy_pixmap(struct exa_context *ctx, struct exa_pixmap_priv *dst_priv, int dx, int dy, struct exa_pixmap_priv *src_priv, int sx, int sy, @@ -966,6 +1001,10 @@ void xorg_copy_pixmap(struct exa_context *ctx, struct pipe_texture *dst = dst_priv->tex; struct pipe_texture *src = src_priv->tex; + if (ctx->pipe->is_texture_referenced(ctx->pipe, src, 0, 0) & + PIPE_REFERENCED_FOR_WRITE) + ctx->pipe->flush(ctx->pipe, PIPE_FLUSH_RENDER_CACHE, NULL); + dst_loc[0] = dx; dst_loc[1] = dy; dst_loc[2] = width; @@ -1003,17 +1042,25 @@ void xorg_copy_pixmap(struct exa_context *ctx, if (src_loc[2] >= 0 && src_loc[3] >= 0 && dst_loc[2] >= 0 && dst_loc[3] >= 0) { + struct pipe_texture *temp_src = src; + + if (src == dst) + temp_src = create_sampler_texture(ctx, src); + renderer_copy_texture(ctx, - src, + temp_src, src_loc[0], - src_loc[1] + src_loc[3], - src_loc[0] + src_loc[2], src_loc[1], + src_loc[0] + src_loc[2], + src_loc[1] + src_loc[3], dst, dst_loc[0], - dst_loc[1] + dst_loc[3], + dst_loc[1], dst_loc[0] + dst_loc[2], - dst_loc[1]); + dst_loc[1] + dst_loc[3]); + + if (src == dst) + pipe_texture_reference(&temp_src, NULL); } } diff --git a/src/gallium/state_trackers/xorg/xorg_crtc.c b/src/gallium/state_trackers/xorg/xorg_crtc.c index fe08bde9efc..67fe29a69da 100644 --- a/src/gallium/state_trackers/xorg/xorg_crtc.c +++ b/src/gallium/state_trackers/xorg/xorg_crtc.c @@ -42,8 +42,12 @@ #include "xorg_tracker.h" #include "xf86Modes.h" +#ifdef HAVE_XEXTPROTO_71 +#include <X11/extensions/dpmsconst.h> +#else #define DPMS_SERVER #include <X11/extensions/dpms.h> +#endif #include "pipe/p_inlines.h" #include "util/u_rect.h" diff --git a/src/gallium/state_trackers/xorg/xorg_exa.c b/src/gallium/state_trackers/xorg/xorg_exa.c index 312dab1544c..3f48ab98ac5 100644 --- a/src/gallium/state_trackers/xorg/xorg_exa.c +++ b/src/gallium/state_trackers/xorg/xorg_exa.c @@ -47,6 +47,9 @@ #include "util/u_rect.h" +#define DEBUG_SOLID 0 +#define DISABLE_ACCEL 0 + /* * Helper functions */ @@ -72,6 +75,9 @@ exa_get_pipe_format(int depth, enum pipe_format *format, int *bbp) assert(*bbp == 16); break; case 8: + *format = PIPE_FORMAT_I8_UNORM; + assert(*bbp == 8); + break; case 4: case 1: *format = PIPE_FORMAT_A8R8G8B8_UNORM; /* bad bad bad */ @@ -253,7 +259,7 @@ ExaDone(PixmapPtr pPixmap) #if 1 xorg_exa_flush(exa, PIPE_FLUSH_RENDER_CACHE, NULL); #else - xorg_finish(exa); + xorg_exa_finish(exa); #endif xorg_exa_common_done(exa); } @@ -276,24 +282,36 @@ ExaPrepareSolid(PixmapPtr pPixmap, int alu, Pixel planeMask, Pixel fg) struct exa_pixmap_priv *priv = exaGetPixmapDriverPrivate(pPixmap); struct exa_context *exa = ms->exa; - debug_printf("ExaPrepareSolid - test\n"); - if (pPixmap->drawable.depth < 15) - return FALSE; - +#if 1 + debug_printf("ExaPrepareSolid(0x%x)\n", fg); +#endif if (!EXA_PM_IS_SOLID(&pPixmap->drawable, planeMask)) return FALSE; if (!priv || !priv->tex) return FALSE; + if (!exa->scrn->is_format_supported(exa->scrn, priv->tex->format, + priv->tex->target, + PIPE_TEXTURE_USAGE_RENDER_TARGET, 0)) + return FALSE; + if (alu != GXcopy) return FALSE; if (!exa->pipe) return FALSE; - debug_printf(" ExaPrepareSolid(0x%x)\n", fg); + +#if DEBUG_SOLID + fg = 0xffff0000; +#endif + +#if DISABLE_ACCEL + return FALSE; +#else return xorg_solid_bind_state(exa, priv, fg); +#endif } static void @@ -310,11 +328,42 @@ ExaSolid(PixmapPtr pPixmap, int x0, int y0, int x1, int y1) if (x0 == 0 && y0 == 0 && x1 == priv->tex->width[0] && y1 == priv->tex->height[0]) { - exa->ctx->clear(exa->ctx, PIPE_CLEAR_COLOR | PIPE_CLEAR_DEPTHSTENCIL, + exa->ctx->clear(exa->pipe, PIPE_CLEAR_COLOR, exa->solid_color, 1., 0); } else #endif - xorg_solid(exa, priv, x0, y0, x1, y1) ; + +#if DEBUG_SOLID + exa->solid_color[0] = 0.f; + exa->solid_color[1] = 1.f; + exa->solid_color[2] = 0.f; + exa->solid_color[3] = 1.f; + xorg_solid(exa, priv, 0, 0, 1024, 768); + exa->solid_color[0] = 1.f; + exa->solid_color[1] = 0.f; + exa->solid_color[2] = 0.f; + exa->solid_color[3] = 1.f; + xorg_solid(exa, priv, 0, 0, 300, 300); + xorg_solid(exa, priv, 300, 300, 350, 350); + xorg_solid(exa, priv, 350, 350, 500, 500); + + xorg_solid(exa, priv, + priv->tex->width[0] - 10, + priv->tex->height[0] - 10, + priv->tex->width[0], + priv->tex->height[0]); + + exa->solid_color[0] = 0.f; + exa->solid_color[1] = 0.f; + exa->solid_color[2] = 1.f; + exa->solid_color[3] = 1.f; + + exa->has_solid_color = FALSE; + ExaPrepareCopy(pPixmap, pPixmap, 0, 0, GXcopy, 0xffffffff); + ExaCopy(pPixmap, 0, 0, 50, 50, 500, 500); +#else + xorg_solid(exa, priv, x0, y0, x1, y1) ; +#endif } static Bool @@ -332,15 +381,20 @@ ExaPrepareCopy(PixmapPtr pSrcPixmap, PixmapPtr pDstPixmap, int xdir, if (alu != GXcopy) return FALSE; - if (pSrcPixmap->drawable.depth < 15 || pDstPixmap->drawable.depth < 15) - return FALSE; - if (!EXA_PM_IS_SOLID(&pSrcPixmap->drawable, planeMask)) return FALSE; if (!priv || !src_priv) return FALSE; + if (!exa->scrn->is_format_supported(exa->scrn, priv->tex->format, + priv->tex->target, + PIPE_TEXTURE_USAGE_RENDER_TARGET, 0) || + !exa->scrn->is_format_supported(exa->scrn, src_priv->tex->format, + src_priv->tex->target, + PIPE_TEXTURE_USAGE_SAMPLER, 0)) + return FALSE; + if (!priv->tex || !src_priv->tex) return FALSE; @@ -350,8 +404,11 @@ ExaPrepareCopy(PixmapPtr pSrcPixmap, PixmapPtr pDstPixmap, int xdir, exa->copy.src = src_priv; exa->copy.dst = priv; - /*XXX disabled until some issues with syncing are fixed */ +#if DISABLE_ACCEL return FALSE; +#else + return TRUE; +#endif } static void @@ -381,14 +438,45 @@ ExaPrepareComposite(int op, PicturePtr pSrcPicture, ScrnInfoPtr pScrn = xf86Screens[pDst->drawable.pScreen->myNum]; modesettingPtr ms = modesettingPTR(pScrn); struct exa_context *exa = ms->exa; + struct exa_pixmap_priv *priv; debug_printf("ExaPrepareComposite\n"); + priv = exaGetPixmapDriverPrivate(pDst); + if (!priv || !priv->tex || + !exa->scrn->is_format_supported(exa->scrn, priv->tex->format, + priv->tex->target, + PIPE_TEXTURE_USAGE_RENDER_TARGET, 0)) + return FALSE; + + if (pSrc) { + priv = exaGetPixmapDriverPrivate(pSrc); + if (!priv || !priv->tex || + !exa->scrn->is_format_supported(exa->scrn, priv->tex->format, + priv->tex->target, + PIPE_TEXTURE_USAGE_SAMPLER, 0)) + return FALSE; + } + + if (pMask) { + priv = exaGetPixmapDriverPrivate(pMask); + if (!priv || !priv->tex || + !exa->scrn->is_format_supported(exa->scrn, priv->tex->format, + priv->tex->target, + PIPE_TEXTURE_USAGE_SAMPLER, 0)) + return FALSE; + } + +#if DISABLE_ACCEL + (void) exa; + return FALSE; +#else return xorg_composite_bind_state(exa, op, pSrcPicture, pMaskPicture, pDstPicture, - exaGetPixmapDriverPrivate(pSrc), - exaGetPixmapDriverPrivate(pMask), + pSrc ? exaGetPixmapDriverPrivate(pSrc) : NULL, + pMask ? exaGetPixmapDriverPrivate(pMask) : NULL, exaGetPixmapDriverPrivate(pDst)); +#endif } static void @@ -411,10 +499,13 @@ ExaCheckComposite(int op, PicturePtr pSrcPicture, PicturePtr pMaskPicture, PicturePtr pDstPicture) { - return xorg_composite_accelerated(op, - pSrcPicture, - pMaskPicture, - pDstPicture); + boolean accelerated = xorg_composite_accelerated(op, + pSrcPicture, + pMaskPicture, + pDstPicture); + debug_printf("ExaCheckComposite(%d, %p, %p, %p) = %d\n", + op, pSrcPicture, pMaskPicture, pDstPicture, accelerated); + return accelerated; } static void * @@ -433,14 +524,11 @@ static void ExaDestroyPixmap(ScreenPtr pScreen, void *dPriv) { struct exa_pixmap_priv *priv = (struct exa_pixmap_priv *)dPriv; - ScrnInfoPtr pScrn = xf86Screens[pScreen->myNum]; - modesettingPtr ms = modesettingPTR(pScrn); if (!priv) return; - if (priv->tex) - ms->screen->texture_destroy(priv->tex); + pipe_texture_reference(&priv->tex, NULL); xfree(priv); } @@ -583,8 +671,8 @@ ExaModifyPixmapHeader(PixmapPtr pPixmap, int width, int height, struct pipe_surface *dst_surf; struct pipe_surface *src_surf; - dst_surf = exa->scrn->get_tex_surface(exa->scrn, texture, 0, 0, 0, - PIPE_BUFFER_USAGE_GPU_WRITE); + dst_surf = exa->scrn->get_tex_surface( + exa->scrn, texture, 0, 0, 0, PIPE_BUFFER_USAGE_GPU_WRITE); src_surf = exa_gpu_surface(exa, priv); exa->pipe->surface_copy(exa->pipe, dst_surf, 0, 0, src_surf, 0, 0, min(width, texture->width[0]), @@ -621,6 +709,8 @@ ExaModifyPixmapHeader(PixmapPtr pPixmap, int width, int height, #endif pipe_texture_reference(&priv->tex, texture); + /* the texture we create has one reference */ + pipe_texture_reference(&texture, NULL); } return TRUE; diff --git a/src/gallium/state_trackers/xorg/xorg_exa.h b/src/gallium/state_trackers/xorg/xorg_exa.h index 43949b04a44..fe1f1cd103c 100644 --- a/src/gallium/state_trackers/xorg/xorg_exa.h +++ b/src/gallium/state_trackers/xorg/xorg_exa.h @@ -32,8 +32,11 @@ struct exa_context struct exa_pixmap_priv *src; struct exa_pixmap_priv *dst; } copy; -}; + /* we should combine these two */ + float vertices2[4][2][4]; + float vertices3[4][3][4]; +}; struct exa_pixmap_priv { diff --git a/src/gallium/state_trackers/xorg/xorg_exa_tgsi.c b/src/gallium/state_trackers/xorg/xorg_exa_tgsi.c index 694eded09a2..bb5a42af37e 100644 --- a/src/gallium/state_trackers/xorg/xorg_exa_tgsi.c +++ b/src/gallium/state_trackers/xorg/xorg_exa_tgsi.c @@ -52,8 +52,7 @@ struct xorg_shaders { static const char over_op[] = "SUB TEMP[3], CONST[0].wwww, TEMP[1].wwww\n" - "MUL TEMP[3], TEMP[0], TEMP[3]\n" - "ADD TEMP[0], TEMP[3], TEMP[0]\n"; + "MAD TEMP[3], TEMP[0], TEMP[3], TEMP[0]\n"; static INLINE void @@ -79,8 +78,7 @@ vs_normalize_coords(struct ureg_program *ureg, struct ureg_src coords, { struct ureg_dst tmp = ureg_DECL_temporary(ureg); struct ureg_src ret; - ureg_MUL(ureg, tmp, coords, const0); - ureg_ADD(ureg, tmp, ureg_src(tmp), const1); + ureg_MAD(ureg, tmp, coords, const0, const1); ret = ureg_src(tmp); ureg_release_temporary(ureg, tmp); return ret; @@ -241,42 +239,39 @@ create_vs(struct pipe_context *pipe, boolean is_fill = vs_traits & VS_FILL; boolean is_composite = vs_traits & VS_COMPOSITE; boolean has_mask = vs_traits & VS_MASK; + unsigned input_slot = 0; ureg = ureg_create(TGSI_PROCESSOR_VERTEX); if (ureg == NULL) return 0; - const0 = ureg_DECL_constant(ureg); - const1 = ureg_DECL_constant(ureg); + const0 = ureg_DECL_constant(ureg, 0); + const1 = ureg_DECL_constant(ureg, 1); /* it has to be either a fill or a composite op */ debug_assert(is_fill ^ is_composite); - src = ureg_DECL_vs_input(ureg, - TGSI_SEMANTIC_POSITION, 0); + src = ureg_DECL_vs_input(ureg, input_slot++); dst = ureg_DECL_output(ureg, TGSI_SEMANTIC_POSITION, 0); src = vs_normalize_coords(ureg, src, const0, const1); ureg_MOV(ureg, dst, src); - if (is_composite) { - src = ureg_DECL_vs_input(ureg, - TGSI_SEMANTIC_GENERIC, 1); - dst = ureg_DECL_output(ureg, TGSI_SEMANTIC_GENERIC, 1); + src = ureg_DECL_vs_input(ureg, input_slot++); + dst = ureg_DECL_output(ureg, TGSI_SEMANTIC_GENERIC, 0); ureg_MOV(ureg, dst, src); } + if (is_fill) { - src = ureg_DECL_vs_input(ureg, - TGSI_SEMANTIC_COLOR, 0); + src = ureg_DECL_vs_input(ureg, input_slot++); dst = ureg_DECL_output(ureg, TGSI_SEMANTIC_COLOR, 0); ureg_MOV(ureg, dst, src); } if (has_mask) { - src = ureg_DECL_vs_input(ureg, - TGSI_SEMANTIC_GENERIC, 2); - dst = ureg_DECL_output(ureg, TGSI_SEMANTIC_POSITION, 2); + src = ureg_DECL_vs_input(ureg, input_slot++); + dst = ureg_DECL_output(ureg, TGSI_SEMANTIC_GENERIC, 2); ureg_MOV(ureg, dst, src); } @@ -315,7 +310,7 @@ create_fs(struct pipe_context *pipe, if (is_composite) { src_sampler = ureg_DECL_sampler(ureg, 0); src_input = ureg_DECL_fs_input(ureg, - TGSI_SEMANTIC_POSITION, + TGSI_SEMANTIC_GENERIC, 0, TGSI_INTERPOLATE_PERSPECTIVE); } else { @@ -335,7 +330,7 @@ create_fs(struct pipe_context *pipe, if (has_mask) { mask_sampler = ureg_DECL_sampler(ureg, 1); mask_pos = ureg_DECL_fs_input(ureg, - TGSI_SEMANTIC_POSITION, + TGSI_SEMANTIC_GENERIC, 1, TGSI_INTERPOLATE_PERSPECTIVE); } @@ -370,11 +365,11 @@ create_fs(struct pipe_context *pipe, else src = out; - coords = ureg_DECL_constant(ureg); - const0124 = ureg_DECL_constant(ureg); - matrow0 = ureg_DECL_constant(ureg); - matrow1 = ureg_DECL_constant(ureg); - matrow2 = ureg_DECL_constant(ureg); + coords = ureg_DECL_constant(ureg, 0); + const0124 = ureg_DECL_constant(ureg, 1); + matrow0 = ureg_DECL_constant(ureg, 2); + matrow1 = ureg_DECL_constant(ureg, 3); + matrow2 = ureg_DECL_constant(ureg, 4); if (is_lingrad) { linear_gradient(ureg, src, @@ -470,7 +465,7 @@ struct xorg_shader xorg_shaders_get(struct xorg_shaders *sc, unsigned vs_traits, unsigned fs_traits) { - struct xorg_shader shader = {0}; + struct xorg_shader shader = { NULL, NULL }; void *vs, *fs; vs = shader_from_cache(sc->exa->pipe, PIPE_SHADER_VERTEX, diff --git a/src/gallium/state_trackers/xorg/xorg_output.c b/src/gallium/state_trackers/xorg/xorg_output.c index 950af942f5b..26f45f8d645 100644 --- a/src/gallium/state_trackers/xorg/xorg_output.c +++ b/src/gallium/state_trackers/xorg/xorg_output.c @@ -42,8 +42,12 @@ #include <sys/stat.h> #include <sys/types.h> +#ifdef HAVE_XEXTPROTO_71 +#include <X11/extensions/dpmsconst.h> +#else #define DPMS_SERVER #include <X11/extensions/dpms.h> +#endif #include "X11/Xatom.h" diff --git a/src/gallium/state_trackers/xorg/xorg_tracker.h b/src/gallium/state_trackers/xorg/xorg_tracker.h index b1ab783a15a..2f7050bcb74 100644 --- a/src/gallium/state_trackers/xorg/xorg_tracker.h +++ b/src/gallium/state_trackers/xorg/xorg_tracker.h @@ -151,5 +151,11 @@ crtc_cursor_destroy(xf86CrtcPtr crtc); void output_init(ScrnInfoPtr pScrn); +/*********************************************************************** + * xorg_xv.c + */ +void +xorg_init_video(ScreenPtr pScreen); + #endif /* _XORG_TRACKER_H_ */ diff --git a/src/gallium/state_trackers/xorg/xorg_xv.c b/src/gallium/state_trackers/xorg/xorg_xv.c new file mode 100644 index 00000000000..88955d47fd3 --- /dev/null +++ b/src/gallium/state_trackers/xorg/xorg_xv.c @@ -0,0 +1,212 @@ +#include "xorg_tracker.h" + +#include <xf86xv.h> +#include <X11/extensions/Xv.h> +#include <fourcc.h> + +/*XXX get these from pipe's texture limits */ +#define IMAGE_MAX_WIDTH 2048 +#define IMAGE_MAX_HEIGHT 2048 + +#define MAKE_ATOM(a) MakeAtom(a, sizeof(a) - 1, TRUE) + +static Atom xvBrightness, xvContrast; + +#define NUM_TEXTURED_ATTRIBUTES 2 +static XF86AttributeRec TexturedAttributes[NUM_TEXTURED_ATTRIBUTES] = { + {XvSettable | XvGettable, -128, 127, "XV_BRIGHTNESS"}, + {XvSettable | XvGettable, 0, 255, "XV_CONTRAST"} +}; + +#define NUM_FORMATS 3 +static XF86VideoFormatRec Formats[NUM_FORMATS] = { + {15, TrueColor}, {16, TrueColor}, {24, TrueColor} +}; + +static XF86VideoEncodingRec DummyEncoding[1] = { + { + 0, + "XV_IMAGE", + IMAGE_MAX_WIDTH, IMAGE_MAX_HEIGHT, + {1, 1} + } +}; + +#define NUM_IMAGES 2 +static XF86ImageRec Images[NUM_IMAGES] = { + XVIMAGE_UYVY, + XVIMAGE_YUY2, +}; + +struct xorg_xv_port_priv { + RegionRec clip; +}; + + +static void +stop_video(ScrnInfoPtr pScrn, pointer data, Bool shutdown) +{ +} + +static int +set_port_attribute(ScrnInfoPtr pScrn, + Atom attribute, INT32 value, pointer data) +{ + return 0; +} + +static int +get_port_attribute(ScrnInfoPtr pScrn, + Atom attribute, INT32 * value, pointer data) +{ + return 0; +} + +static void +query_best_size(ScrnInfoPtr pScrn, + Bool motion, + short vid_w, short vid_h, + short drw_w, short drw_h, + unsigned int *p_w, unsigned int *p_h, pointer data) +{ +} + +static int +put_image(ScrnInfoPtr pScrn, + short src_x, short src_y, + short drw_x, short drw_y, + short src_w, short src_h, + short drw_w, short drw_h, + int id, unsigned char *buf, + short width, short height, + Bool sync, RegionPtr clipBoxes, pointer data, + DrawablePtr pDraw) +{ + return 0; +} + +static int +query_image_attributes(ScrnInfoPtr pScrn, + int id, + unsigned short *w, unsigned short *h, + int *pitches, int *offsets) +{ + return 0; +} + +static struct xorg_xv_port_priv * +port_priv_create(ScreenPtr pScreen) +{ + /*ScrnInfoPtr pScrn = xf86Screens[pScreen->myNum];*/ + /*modesettingPtr ms = modesettingPTR(pScrn);*/ + struct xorg_xv_port_priv *priv = NULL; + + priv = calloc(1, sizeof(struct xorg_xv_port_priv)); + + if (!priv) + return NULL; + + REGION_NULL(pScreen, &priv->clip); + + return priv; +} + +static XF86VideoAdaptorPtr +xorg_setup_textured_adapter(ScreenPtr pScreen) +{ + /*ScrnInfoPtr pScrn = xf86Screens[pScreen->myNum];*/ + /*modesettingPtr ms = modesettingPTR(pScrn);*/ + XF86VideoAdaptorPtr adapt; + XF86AttributePtr attrs; + DevUnion *dev_unions; + int nports = 16, i; + int nattributes; + + nattributes = NUM_TEXTURED_ATTRIBUTES; + + adapt = calloc(1, sizeof(XF86VideoAdaptorRec)); + dev_unions = calloc(nports, sizeof(DevUnion)); + attrs = calloc(nattributes, sizeof(XF86AttributeRec)); + if (adapt == NULL || dev_unions == NULL || attrs == NULL) { + free(adapt); + free(dev_unions); + free(attrs); + return NULL; + } + + adapt->type = XvWindowMask | XvInputMask | XvImageMask; + adapt->flags = 0; + adapt->name = "Gallium3D Textured Video"; + adapt->nEncodings = 1; + adapt->pEncodings = DummyEncoding; + adapt->nFormats = NUM_FORMATS; + adapt->pFormats = Formats; + adapt->nPorts = 0; + adapt->pPortPrivates = dev_unions; + adapt->nAttributes = nattributes; + adapt->pAttributes = attrs; + memcpy(attrs, TexturedAttributes, nattributes * sizeof(XF86AttributeRec)); + adapt->nImages = NUM_IMAGES; + adapt->pImages = Images; + adapt->PutVideo = NULL; + adapt->PutStill = NULL; + adapt->GetVideo = NULL; + adapt->GetStill = NULL; + adapt->StopVideo = stop_video; + adapt->SetPortAttribute = set_port_attribute; + adapt->GetPortAttribute = get_port_attribute; + adapt->QueryBestSize = query_best_size; + adapt->PutImage = put_image; + adapt->QueryImageAttributes = query_image_attributes; + + for (i = 0; i < nports; i++) { + struct xorg_xv_port_priv *priv = + port_priv_create(pScreen); + + adapt->pPortPrivates[i].ptr = (pointer) (priv); + adapt->nPorts++; + } + + return adapt; +} + +void +xorg_init_video(ScreenPtr pScreen) +{ + ScrnInfoPtr pScrn = xf86Screens[pScreen->myNum]; + /*modesettingPtr ms = modesettingPTR(pScrn);*/ + XF86VideoAdaptorPtr *adaptors, *new_adaptors = NULL; + XF86VideoAdaptorPtr textured_adapter; + int num_adaptors; + + num_adaptors = xf86XVListGenericAdaptors(pScrn, &adaptors); + new_adaptors = malloc((num_adaptors + 1) * sizeof(XF86VideoAdaptorPtr *)); + if (new_adaptors == NULL) + return; + + memcpy(new_adaptors, adaptors, num_adaptors * sizeof(XF86VideoAdaptorPtr)); + adaptors = new_adaptors; + + /* Add the adaptors supported by our hardware. First, set up the atoms + * that will be used by both output adaptors. + */ + xvBrightness = MAKE_ATOM("XV_BRIGHTNESS"); + xvContrast = MAKE_ATOM("XV_CONTRAST"); + + textured_adapter = xorg_setup_textured_adapter(pScreen); + + debug_assert(textured_adapter); + + if (textured_adapter) { + adaptors[num_adaptors++] = textured_adapter; + } + + if (num_adaptors) { + xf86XVScreenInit(pScreen, adaptors, num_adaptors); + } else { + xf86DrvMsg(pScrn->scrnIndex, X_WARNING, + "Disabling Xv because no adaptors could be initialized.\n"); + } + + free(adaptors); +} diff --git a/src/gallium/winsys/drm/intel/gem/intel_drm_api.c b/src/gallium/winsys/drm/intel/gem/intel_drm_api.c index 4c5a1d2ea89..0fd5cdd9697 100644 --- a/src/gallium/winsys/drm/intel/gem/intel_drm_api.c +++ b/src/gallium/winsys/drm/intel/gem/intel_drm_api.c @@ -7,6 +7,7 @@ #include "i915simple/i915_context.h" #include "i915simple/i915_screen.h" +#include "trace/tr_drm.h" /* * Helper functions @@ -198,5 +199,5 @@ struct drm_api intel_drm_api = struct drm_api * drm_api_create() { - return &intel_drm_api; + return trace_drm_create(&intel_drm_api); } diff --git a/src/gallium/winsys/drm/nouveau/drm/nouveau_drm_api.c b/src/gallium/winsys/drm/nouveau/drm/nouveau_drm_api.c index 091cbbcfed1..117ca6059bb 100644 --- a/src/gallium/winsys/drm/nouveau/drm/nouveau_drm_api.c +++ b/src/gallium/winsys/drm/nouveau/drm/nouveau_drm_api.c @@ -112,7 +112,7 @@ nouveau_drm_create_screen(struct drm_api *api, int fd, return NULL; } - if (arg->mode == DRM_CREATE_DRI1) { + if (arg && arg->mode == DRM_CREATE_DRI1) { struct nouveau_dri *nvdri = dri1->ddx_info; enum pipe_format format; diff --git a/src/gallium/winsys/drm/nouveau/xorg/Makefile b/src/gallium/winsys/drm/nouveau/xorg/Makefile new file mode 100644 index 00000000000..f0d3b337e83 --- /dev/null +++ b/src/gallium/winsys/drm/nouveau/xorg/Makefile @@ -0,0 +1,61 @@ +TARGET = modesetting_drv.so +CFILES = $(wildcard ./*.c) +OBJECTS = $(patsubst ./%.c,./%.o,$(CFILES)) +TOP = ../../../../../.. + +include $(TOP)/configs/current + +INCLUDES = \ + $(shell pkg-config --cflags-only-I pixman-1 xorg-server libdrm xproto) \ + -I../gem \ + -I$(TOP)/src/gallium/include \ + -I$(TOP)/src/gallium/drivers \ + -I$(TOP)/src/gallium/auxiliary \ + -I$(TOP)/src/mesa \ + -I$(TOP)/include \ + -I$(TOP)/src/egl/main + +LIBS = \ + $(TOP)/src/gallium/state_trackers/xorg/libxorgtracker.a \ + $(TOP)/src/gallium/winsys/drm/nouveau/drm/libnouveaudrm.a \ + $(TOP)/src/gallium/drivers/nv04/libnv04.a \ + $(TOP)/src/gallium/drivers/nv10/libnv10.a \ + $(TOP)/src/gallium/drivers/nv20/libnv20.a \ + $(TOP)/src/gallium/drivers/nv30/libnv30.a \ + $(TOP)/src/gallium/drivers/nv40/libnv40.a \ + $(TOP)/src/gallium/drivers/nv50/libnv50.a \ + $(TOP)/src/gallium/drivers/nouveau/libnouveau.a \ + $(GALLIUM_AUXILIARIES) + +DRIVER_DEFINES = \ + -DHAVE_CONFIG_H + + +############################################# + + + +all default: $(TARGET) + +$(TARGET): $(OBJECTS) Makefile $(TOP)/src/gallium/state_trackers/xorg/libxorgtracker.a $(LIBS) + $(TOP)/bin/mklib -noprefix -o $@ \ + $(OBJECTS) $(LIBS) $(shell pkg-config --libs libdrm) -ldrm_nouveau + +clean: + rm -rf $(OBJECTS) $(TARGET) + +install: + $(INSTALL) -d $(DESTDIR)/$(XORG_DRIVER_INSTALL_DIR) + $(MINSTALL) -m 755 $(TARGET) $(DESTDIR)/$(XORG_DRIVER_INSTALL_DIR) + + +############################################## + + +.c.o: + $(CC) -c $(CFLAGS) $(INCLUDES) $(DRIVER_DEFINES) $< -o $@ + + +############################################## + +.PHONY = all clean install diff --git a/src/gallium/winsys/drm/nouveau/xorg/nouveau_xorg.c b/src/gallium/winsys/drm/nouveau/xorg/nouveau_xorg.c new file mode 100644 index 00000000000..a669b3080aa --- /dev/null +++ b/src/gallium/winsys/drm/nouveau/xorg/nouveau_xorg.c @@ -0,0 +1,149 @@ +/* + * Copyright 2008 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. + * + * + * Author: Alan Hourihane <[email protected]> + * Author: Jakob Bornecrantz <[email protected]> + * + */ + +#include "../../../../state_trackers/xorg/xorg_winsys.h" + +static void nouveau_xorg_identify(int flags); +static Bool nouveau_xorg_pci_probe(DriverPtr driver, int entity_num, + struct pci_device *device, + intptr_t match_data); + +static const struct pci_id_match nouveau_xorg_device_match[] = { + { 0x10de, PCI_MATCH_ANY, PCI_MATCH_ANY, PCI_MATCH_ANY, + 0x00030000, 0x00ffffff, 0 }, + { 0x12d2, PCI_MATCH_ANY, PCI_MATCH_ANY, PCI_MATCH_ANY, + 0x00030000, 0x00ffffff, 0 }, + {0, 0, 0}, +}; + +static SymTabRec nouveau_xorg_chipsets[] = { + {PCI_MATCH_ANY, "NVIDIA Graphics Device"}, + {-1, NULL} +}; + +static PciChipsets nouveau_xorg_pci_devices[] = { + {PCI_MATCH_ANY, PCI_MATCH_ANY, NULL}, + {-1, -1, NULL} +}; + +static XF86ModuleVersionInfo nouveau_xorg_version = { + "modesetting", + MODULEVENDORSTRING, + MODINFOSTRING1, + MODINFOSTRING2, + XORG_VERSION_CURRENT, + 0, 1, 0, /* major, minor, patch */ + ABI_CLASS_VIDEODRV, + ABI_VIDEODRV_VERSION, + MOD_CLASS_VIDEODRV, + {0, 0, 0, 0} +}; + +/* + * Xorg driver exported structures + */ + +_X_EXPORT DriverRec modesetting = { + 1, + "modesetting", + nouveau_xorg_identify, + NULL, + xorg_tracker_available_options, + NULL, + 0, + NULL, + nouveau_xorg_device_match, + nouveau_xorg_pci_probe +}; + +static MODULESETUPPROTO(nouveau_xorg_setup); + +_X_EXPORT XF86ModuleData modesettingModuleData = { + &nouveau_xorg_version, + nouveau_xorg_setup, + NULL +}; + +/* + * Xorg driver functions + */ + +static pointer +nouveau_xorg_setup(pointer module, pointer opts, int *errmaj, int *errmin) +{ + static Bool setupDone = 0; + + /* This module should be loaded only once, but check to be sure. + */ + if (!setupDone) { + setupDone = 1; + xf86AddDriver(&modesetting, module, HaveDriverFuncs); + + /* + * The return value must be non-NULL on success even though there + * is no TearDownProc. + */ + return (pointer) 1; + } else { + if (errmaj) + *errmaj = LDR_ONCEONLY; + return NULL; + } +} + +static void +nouveau_xorg_identify(int flags) +{ + xf86PrintChipsets("modesetting", "Driver for Modesetting Kernel Drivers", + nouveau_xorg_chipsets); +} + +static Bool +nouveau_xorg_pci_probe(DriverPtr driver, + int entity_num, struct pci_device *device, intptr_t match_data) +{ + ScrnInfoPtr scrn = NULL; + EntityInfoPtr entity; + + scrn = xf86ConfigPciEntity(scrn, 0, entity_num, nouveau_xorg_pci_devices, + NULL, NULL, NULL, NULL, NULL); + if (scrn != NULL) { + scrn->driverVersion = 1; + scrn->driverName = "i915"; + scrn->name = "modesetting"; + scrn->Probe = NULL; + + entity = xf86GetEntityInfo(entity_num); + + /* Use all the functions from the xorg tracker */ + xorg_tracker_set_functions(scrn); + } + return scrn != NULL; +} diff --git a/src/gallium/winsys/g3dvl/xsp_winsys.c b/src/gallium/winsys/g3dvl/xsp_winsys.c index 698c2856a4f..37d60ce5406 100644 --- a/src/gallium/winsys/g3dvl/xsp_winsys.c +++ b/src/gallium/winsys/g3dvl/xsp_winsys.c @@ -105,6 +105,7 @@ static struct pipe_buffer* xsp_surface_buffer_create unsigned height, enum pipe_format format, unsigned usage, + unsigned tex_usage, unsigned *stride ) { diff --git a/src/gallium/winsys/gdi/SConscript b/src/gallium/winsys/gdi/SConscript index 86eb9ef55ed..f5e6d36d89c 100644 --- a/src/gallium/winsys/gdi/SConscript +++ b/src/gallium/winsys/gdi/SConscript @@ -5,35 +5,39 @@ Import('*') if env['platform'] == 'windows': - env = env.Clone() + env = env.Clone() - env.Append(CPPPATH = [ - '#src/gallium/state_trackers/wgl', - ]) + env.Append(CPPPATH = [ + '#src/gallium/state_trackers/wgl', + ]) - env.Append(LIBS = [ - 'gdi32', - 'user32', - 'kernel32', - 'ws2_32', - ]) + env.Append(LIBS = [ + 'gdi32', + 'user32', + 'kernel32', + 'ws2_32', + ]) - sources = [ - 'gdi_softpipe_winsys.c', - ] - - if env['gcc']: - sources += ['#src/gallium/state_trackers/wgl/opengl32.mingw.def'] - else: - sources += ['#src/gallium/state_trackers/wgl/opengl32.def'] - - drivers = [ - trace, - softpipe, - ] + if 'llvmpipe' in env['drivers']: + sources = ['gdi_llvmpipe_winsys.c'] + drivers = [llvmpipe] + env.Tool('llvm') + elif 'softpipe' in env['drivers']: + sources = ['gdi_softpipe_winsys.c'] + drivers = [softpipe] + else: + print 'warning: softpipe or llvmpipe not selected, gdi winsys disabled' + Return() + + if env['gcc']: + sources += ['#src/gallium/state_trackers/wgl/opengl32.mingw.def'] + else: + sources += ['#src/gallium/state_trackers/wgl/opengl32.def'] + + drivers += [trace] - env.SharedLibrary( - target ='opengl32', - source = sources, - LIBS = wgl + glapi + mesa + drivers + auxiliaries + env['LIBS'], - ) + env.SharedLibrary( + target ='opengl32', + source = sources, + LIBS = wgl + glapi + mesa + drivers + auxiliaries + env['LIBS'], + ) diff --git a/src/gallium/winsys/gdi/gdi_llvmpipe_winsys.c b/src/gallium/winsys/gdi/gdi_llvmpipe_winsys.c new file mode 100644 index 00000000000..e8bc0f55ac4 --- /dev/null +++ b/src/gallium/winsys/gdi/gdi_llvmpipe_winsys.c @@ -0,0 +1,288 @@ +/************************************************************************** + * + * Copyright 2009 VMware, Inc. + * 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 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 + * THE COPYRIGHT HOLDERS, AUTHORS 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. + * + * The above copyright notice and this permission notice (including the + * next paragraph) shall be included in all copies or substantial portions + * of the Software. + * + * + **************************************************************************/ + +/** + * @file + * LLVMpipe support. + * + * @author Jose Fonseca <[email protected]> + */ + + +#include <windows.h> + +#include "pipe/p_format.h" +#include "pipe/p_context.h" +#include "pipe/p_inlines.h" +#include "util/u_math.h" +#include "util/u_memory.h" +#include "llvmpipe/lp_winsys.h" +#include "llvmpipe/lp_texture.h" +#include "stw_winsys.h" + + +struct gdi_llvmpipe_displaytarget +{ + enum pipe_format format; + struct pipe_format_block block; + unsigned width; + unsigned height; + unsigned stride; + + unsigned size; + + void *data; + + BITMAPINFO bmi; +}; + + +/** Cast wrapper */ +static INLINE struct gdi_llvmpipe_displaytarget * +gdi_llvmpipe_displaytarget( struct llvmpipe_displaytarget *buf ) +{ + return (struct gdi_llvmpipe_displaytarget *)buf; +} + + +static boolean +gdi_llvmpipe_is_displaytarget_format_supported( struct llvmpipe_winsys *ws, + enum pipe_format format ) +{ + switch(format) { + case PIPE_FORMAT_X8R8G8B8_UNORM: + case PIPE_FORMAT_A8R8G8B8_UNORM: + return TRUE; + + /* TODO: Support other formats possible with BMPs, as described in + * http://msdn.microsoft.com/en-us/library/dd183376(VS.85).aspx */ + + default: + return FALSE; + } +} + + +static void * +gdi_llvmpipe_displaytarget_map(struct llvmpipe_winsys *ws, + struct llvmpipe_displaytarget *dt, + unsigned flags ) +{ + struct gdi_llvmpipe_displaytarget *gdt = gdi_llvmpipe_displaytarget(dt); + + return gdt->data; +} + + +static void +gdi_llvmpipe_displaytarget_unmap(struct llvmpipe_winsys *ws, + struct llvmpipe_displaytarget *dt ) +{ + +} + + +static void +gdi_llvmpipe_displaytarget_destroy(struct llvmpipe_winsys *winsys, + struct llvmpipe_displaytarget *dt) +{ + struct gdi_llvmpipe_displaytarget *gdt = gdi_llvmpipe_displaytarget(dt); + + align_free(gdt->data); + FREE(gdt); +} + + +/** + * Round n up to next multiple. + */ +static INLINE unsigned +round_up(unsigned n, unsigned multiple) +{ + return (n + multiple - 1) & ~(multiple - 1); +} + + +static struct llvmpipe_displaytarget * +gdi_llvmpipe_displaytarget_create(struct llvmpipe_winsys *winsys, + enum pipe_format format, + unsigned width, unsigned height, + unsigned alignment, + unsigned *stride) +{ + struct gdi_llvmpipe_displaytarget *gdt; + unsigned cpp; + unsigned bpp; + + gdt = CALLOC_STRUCT(gdi_llvmpipe_displaytarget); + if(!gdt) + goto no_gdt; + + gdt->format = format; + gdt->width = width; + gdt->height = height; + + bpp = pf_get_bits(format); + cpp = pf_get_size(format); + + gdt->stride = round_up(width * cpp, alignment); + gdt->size = gdt->stride * height; + + gdt->data = align_malloc(gdt->size, alignment); + if(!gdt->data) + goto no_data; + + gdt->bmi.bmiHeader.biSize = sizeof(BITMAPINFOHEADER); + gdt->bmi.bmiHeader.biWidth = gdt->stride / cpp; + gdt->bmi.bmiHeader.biHeight= -(long)height; + gdt->bmi.bmiHeader.biPlanes = 1; + gdt->bmi.bmiHeader.biBitCount = bpp; + gdt->bmi.bmiHeader.biCompression = BI_RGB; + gdt->bmi.bmiHeader.biSizeImage = 0; + gdt->bmi.bmiHeader.biXPelsPerMeter = 0; + gdt->bmi.bmiHeader.biYPelsPerMeter = 0; + gdt->bmi.bmiHeader.biClrUsed = 0; + gdt->bmi.bmiHeader.biClrImportant = 0; + + *stride = gdt->stride; + return (struct llvmpipe_displaytarget *)gdt; + +no_data: + FREE(gdt); +no_gdt: + return NULL; +} + + +static void +gdi_llvmpipe_displaytarget_display(struct llvmpipe_winsys *winsys, + struct llvmpipe_displaytarget *dt, + void *context_private) +{ + assert(0); +} + + +static void +gdi_llvmpipe_destroy(struct llvmpipe_winsys *winsys) +{ + FREE(winsys); +} + + +static struct pipe_screen * +gdi_llvmpipe_screen_create(void) +{ + static struct llvmpipe_winsys *winsys; + struct pipe_screen *screen; + + winsys = CALLOC_STRUCT(llvmpipe_winsys); + if(!winsys) + goto no_winsys; + + winsys->destroy = gdi_llvmpipe_destroy; + winsys->is_displaytarget_format_supported = gdi_llvmpipe_is_displaytarget_format_supported; + winsys->displaytarget_create = gdi_llvmpipe_displaytarget_create; + winsys->displaytarget_map = gdi_llvmpipe_displaytarget_map; + winsys->displaytarget_unmap = gdi_llvmpipe_displaytarget_unmap; + winsys->displaytarget_display = gdi_llvmpipe_displaytarget_display; + winsys->displaytarget_destroy = gdi_llvmpipe_displaytarget_destroy; + + screen = llvmpipe_create_screen(winsys); + if(!screen) + goto no_screen; + + return screen; + +no_screen: + FREE(winsys); +no_winsys: + return NULL; +} + + +static struct pipe_context * +gdi_llvmpipe_context_create(struct pipe_screen *screen) +{ + return llvmpipe_create(screen); +} + + +static void +gdi_llvmpipe_present(struct pipe_screen *screen, + struct pipe_surface *surface, + HDC hDC) +{ + struct llvmpipe_texture *texture; + struct gdi_llvmpipe_displaytarget *gdt; + + texture = llvmpipe_texture(surface->texture); + gdt = gdi_llvmpipe_displaytarget(texture->dt); + + StretchDIBits(hDC, + 0, 0, gdt->width, gdt->height, + 0, 0, gdt->width, gdt->height, + gdt->data, &gdt->bmi, 0, SRCCOPY); +} + + +static const struct stw_winsys stw_winsys = { + &gdi_llvmpipe_screen_create, + &gdi_llvmpipe_context_create, + &gdi_llvmpipe_present, + NULL, /* get_adapter_luid */ + NULL, /* shared_surface_open */ + NULL, /* shared_surface_close */ + NULL /* compose */ +}; + + +BOOL WINAPI +DllMain(HINSTANCE hinstDLL, DWORD fdwReason, LPVOID lpReserved) +{ + switch (fdwReason) { + case DLL_PROCESS_ATTACH: + if (!stw_init(&stw_winsys)) { + return FALSE; + } + return stw_init_thread(); + + case DLL_THREAD_ATTACH: + return stw_init_thread(); + + case DLL_THREAD_DETACH: + stw_cleanup_thread(); + break; + + case DLL_PROCESS_DETACH: + stw_cleanup_thread(); + stw_cleanup(); + break; + } + return TRUE; +} diff --git a/src/gallium/winsys/gdi/gdi_softpipe_winsys.c b/src/gallium/winsys/gdi/gdi_softpipe_winsys.c index 66120a6a983..5e0ccf32f48 100644 --- a/src/gallium/winsys/gdi/gdi_softpipe_winsys.c +++ b/src/gallium/winsys/gdi/gdi_softpipe_winsys.c @@ -46,7 +46,7 @@ #include "util/u_memory.h" #include "softpipe/sp_winsys.h" #include "softpipe/sp_texture.h" -#include "shared/stw_winsys.h" +#include "stw_winsys.h" struct gdi_softpipe_buffer @@ -269,9 +269,9 @@ gdi_softpipe_context_create(struct pipe_screen *screen) static void -gdi_softpipe_flush_frontbuffer(struct pipe_screen *screen, - struct pipe_surface *surface, - HDC hDC) +gdi_softpipe_present(struct pipe_screen *screen, + struct pipe_surface *surface, + HDC hDC) { struct softpipe_texture *texture; struct gdi_softpipe_buffer *buffer; @@ -304,7 +304,11 @@ gdi_softpipe_flush_frontbuffer(struct pipe_screen *screen, static const struct stw_winsys stw_winsys = { &gdi_softpipe_screen_create, &gdi_softpipe_context_create, - &gdi_softpipe_flush_frontbuffer + &gdi_softpipe_present, + NULL, /* get_adapter_luid */ + NULL, /* shared_surface_open */ + NULL, /* shared_surface_close */ + NULL /* compose */ }; diff --git a/src/gallium/winsys/xlib/Makefile b/src/gallium/winsys/xlib/Makefile index 522f6dc5aee..3a1945d92c5 100644 --- a/src/gallium/winsys/xlib/Makefile +++ b/src/gallium/winsys/xlib/Makefile @@ -34,6 +34,7 @@ XLIB_WINSYS_SOURCES = \ xlib_brw_aub.c \ xlib_brw_context.c \ xlib_brw_screen.c \ + xlib_llvmpipe.c \ xlib_softpipe.c \ xlib_trace.c diff --git a/src/gallium/winsys/xlib/xlib_llvmpipe.c b/src/gallium/winsys/xlib/xlib_llvmpipe.c index bc876591c0d..3dd15e099b0 100644 --- a/src/gallium/winsys/xlib/xlib_llvmpipe.c +++ b/src/gallium/winsys/xlib/xlib_llvmpipe.c @@ -33,6 +33,8 @@ */ +#if defined(GALLIUM_LLVMPIPE) + #include "xm_api.h" #undef ASSERT @@ -459,3 +461,4 @@ struct xm_driver xlib_llvmpipe_driver = +#endif /* GALLIUM_LLVMPIPE */ diff --git a/src/gallium/winsys/xlib/xlib_softpipe.c b/src/gallium/winsys/xlib/xlib_softpipe.c index 67fea023a3b..260b39e2a0f 100644 --- a/src/gallium/winsys/xlib/xlib_softpipe.c +++ b/src/gallium/winsys/xlib/xlib_softpipe.c @@ -75,9 +75,6 @@ struct xmesa_pipe_winsys { struct pipe_winsys base; /* struct xmesa_visual *xm_visual; */ -#ifdef USE_XSHM - int shm; -#endif }; @@ -93,11 +90,6 @@ xm_buffer( struct pipe_buffer *buf ) /** * X Shared Memory Image extension code */ -#ifdef USE_XSHM -#define XSHM_ENABLED(b) ((b)->shm) -#else -#define XSHM_ENABLED(b) 0 -#endif #ifdef USE_XSHM @@ -116,23 +108,23 @@ mesaHandleXError(Display *dpy, XErrorEvent *event) } -static GLboolean alloc_shm(struct xm_buffer *buf, unsigned size) +static char *alloc_shm(struct xm_buffer *buf, unsigned size) { XShmSegmentInfo *const shminfo = & buf->shminfo; shminfo->shmid = shmget(IPC_PRIVATE, size, IPC_CREAT|0777); if (shminfo->shmid < 0) { - return GL_FALSE; + return NULL; } shminfo->shmaddr = (char *) shmat(shminfo->shmid, 0, 0); if (shminfo->shmaddr == (char *) -1) { shmctl(shminfo->shmid, IPC_RMID, 0); - return GL_FALSE; + return NULL; } shminfo->readOnly = False; - return GL_TRUE; + return shminfo->shmaddr; } @@ -258,25 +250,30 @@ xlib_softpipe_display_surface(struct xmesa_buffer *b, return; #ifdef USE_XSHM - if (XSHM_ENABLED(xm_buf) && (xm_buf->tempImage == NULL)) { - assert(surf->texture->block.width == 1); - assert(surf->texture->block.height == 1); - alloc_shm_ximage(xm_buf, b, spt->stride[surf->level] / - surf->texture->block.size, surf->height); - } -#endif + if (xm_buf->shm) + { + if (xm_buf->tempImage == NULL) + { + assert(surf->texture->block.width == 1); + assert(surf->texture->block.height == 1); + alloc_shm_ximage(xm_buf, b, spt->stride[surf->level] / + surf->texture->block.size, surf->height); + } - ximage = (XSHM_ENABLED(xm_buf)) ? xm_buf->tempImage : b->tempImage; - ximage->data = xm_buf->data; + ximage = xm_buf->tempImage; + ximage->data = xm_buf->data; - /* display image in Window */ -#ifdef USE_XSHM - if (XSHM_ENABLED(xm_buf)) { + /* _debug_printf("XSHM\n"); */ XShmPutImage(b->xm_visual->display, b->drawable, b->gc, ximage, 0, 0, 0, 0, surf->width, surf->height, False); - } else + } + else #endif { + /* display image in Window */ + ximage = b->tempImage; + ximage->data = xm_buf->data; + /* check that the XImage has been previously initialized */ assert(ximage->format); assert(ximage->bitmap_unit); @@ -286,6 +283,7 @@ xlib_softpipe_display_surface(struct xmesa_buffer *b, ximage->height = surf->height; ximage->bytes_per_line = spt->stride[surf->level]; + /* _debug_printf("XPUT\n"); */ XPutImage(b->xm_visual->display, b->drawable, b->gc, ximage, 0, 0, 0, 0, surf->width, surf->height); } @@ -322,21 +320,6 @@ xm_buffer_create(struct pipe_winsys *pws, unsigned size) { struct xm_buffer *buffer = CALLOC_STRUCT(xm_buffer); -#ifdef USE_XSHM - struct xmesa_pipe_winsys *xpws = (struct xmesa_pipe_winsys *) pws; - - buffer->shminfo.shmid = -1; - buffer->shminfo.shmaddr = (char *) -1; - - if (xpws->shm && (usage & PIPE_BUFFER_USAGE_PIXEL) != 0) { - buffer->shm = xpws->shm; - - if (alloc_shm(buffer, size)) { - buffer->data = buffer->shminfo.shmaddr; - buffer->shm = 1; - } - } -#endif pipe_reference_init(&buffer->base.reference, 1); buffer->base.alignment = alignment; @@ -363,9 +346,6 @@ xm_user_buffer_create(struct pipe_winsys *pws, void *ptr, unsigned bytes) buffer->base.size = bytes; buffer->userBuffer = TRUE; buffer->data = ptr; -#ifdef USE_XSHM - buffer->shm = 0; -#endif return &buffer->base; } @@ -381,16 +361,44 @@ xm_surface_buffer_create(struct pipe_winsys *winsys, { const unsigned alignment = 64; struct pipe_format_block block; - unsigned nblocksx, nblocksy; + unsigned nblocksx, nblocksy, size; pf_get_block(format, &block); nblocksx = pf_get_nblocksx(&block, width); nblocksy = pf_get_nblocksy(&block, height); *stride = align(nblocksx * block.size, alignment); + size = *stride * nblocksy; + +#ifdef USE_XSHM + if (!debug_get_bool_option("XLIB_NO_SHM", FALSE)) + { + struct xm_buffer *buffer = CALLOC_STRUCT(xm_buffer); + + pipe_reference_init(&buffer->base.reference, 1); + buffer->base.alignment = alignment; + buffer->base.usage = usage; + buffer->base.size = size; + buffer->userBuffer = FALSE; + buffer->shminfo.shmid = -1; + buffer->shminfo.shmaddr = (char *) -1; + buffer->shm = TRUE; + + buffer->data = alloc_shm(buffer, size); + if (!buffer->data) + goto out; + + return &buffer->base; + + out: + if (buffer) + FREE(buffer); + } +#endif + return winsys->buffer_create(winsys, alignment, usage, - *stride * nblocksy); + size); } diff --git a/src/glx/x11/glx_pbuffer.c b/src/glx/x11/glx_pbuffer.c index 88a17df6f8d..501500afc3d 100644 --- a/src/glx/x11/glx_pbuffer.c +++ b/src/glx/x11/glx_pbuffer.c @@ -39,6 +39,30 @@ #include "glxextensions.h" #include "glcontextmodes.h" +#define WARN_ONCE_GLX_1_3(a, b) { \ + static int warned=1; \ + if(warned) { \ + warn_GLX_1_3((a), b ); \ + warned=0; \ + } \ + } + +/** + * Emit a warning when clients use GLX 1.3 functions on pre-1.3 systems. + */ +static void +warn_GLX_1_3(Display *dpy, const char *function_name) +{ + __GLXdisplayPrivate *priv = __glXInitialize(dpy); + + if (priv->minorVersion < 3) { + fprintf(stderr, + "WARNING: Application calling GLX 1.3 function \"%s\" " + "when GLX 1.3 is not supported! This is an application bug!\n", + function_name); + } +} + /** * Change a drawable's attribute. @@ -559,6 +583,8 @@ glXCreatePbuffer(Display * dpy, GLXFBConfig config, const int *attrib_list) width = 0; height = 0; + WARN_ONCE_GLX_1_3(dpy, __func__); + for (i = 0; attrib_list[i * 2]; i++) { switch (attrib_list[i * 2]) { case GLX_PBUFFER_WIDTH: @@ -592,6 +618,7 @@ PUBLIC void glXQueryDrawable(Display * dpy, GLXDrawable drawable, int attribute, unsigned int *value) { + WARN_ONCE_GLX_1_3(dpy, __func__); GetDrawableAttribute(dpy, drawable, attribute, value); } @@ -645,6 +672,8 @@ PUBLIC GLXPixmap glXCreatePixmap(Display * dpy, GLXFBConfig config, Pixmap pixmap, const int *attrib_list) { + WARN_ONCE_GLX_1_3(dpy, __func__); + return CreateDrawable(dpy, (__GLcontextModes *) config, (Drawable) pixmap, attrib_list, X_GLXCreatePixmap); } @@ -654,6 +683,8 @@ PUBLIC GLXWindow glXCreateWindow(Display * dpy, GLXFBConfig config, Window win, const int *attrib_list) { + WARN_ONCE_GLX_1_3(dpy, __func__); + return CreateDrawable(dpy, (__GLcontextModes *) config, (Drawable) win, attrib_list, X_GLXCreateWindow); } @@ -662,6 +693,8 @@ glXCreateWindow(Display * dpy, GLXFBConfig config, Window win, PUBLIC void glXDestroyPixmap(Display * dpy, GLXPixmap pixmap) { + WARN_ONCE_GLX_1_3(dpy, __func__); + DestroyDrawable(dpy, (GLXDrawable) pixmap, X_GLXDestroyPixmap); } @@ -669,6 +702,8 @@ glXDestroyPixmap(Display * dpy, GLXPixmap pixmap) PUBLIC void glXDestroyWindow(Display * dpy, GLXWindow win) { + WARN_ONCE_GLX_1_3(dpy, __func__); + DestroyDrawable(dpy, (GLXDrawable) win, X_GLXDestroyWindow); } @@ -688,3 +723,4 @@ GLX_ALIAS_VOID(glXGetSelectedEventSGIX, (Display * dpy, GLXDrawable drawable, unsigned long *mask), (dpy, drawable, mask), glXGetSelectedEvent) + diff --git a/src/glx/x11/glxhash.c b/src/glx/x11/glxhash.c index 74cd4f344df..6f2c51d39dd 100644 --- a/src/glx/x11/glxhash.c +++ b/src/glx/x11/glxhash.c @@ -77,6 +77,7 @@ #include <stdio.h> #include <stdlib.h> +#include <string.h> #define HASH_MAGIC 0xdeadbeef #define HASH_DEBUG 0 @@ -87,9 +88,13 @@ #define HASH_ALLOC malloc #define HASH_FREE free -#define HASH_RANDOM_DECL -#define HASH_RANDOM_INIT(seed) srandom(seed) -#define HASH_RANDOM random() +#define HASH_RANDOM_DECL struct random_data rd; int32_t rv; char rs[256] +#define HASH_RANDOM_INIT(seed) \ + do { \ + (void) memset(&rd, 0, sizeof(rd)); \ + (void) initstate_r(seed, rs, sizeof(rs), &rd); \ + } while(0) +#define HASH_RANDOM ((void) random_r(&rd, &rv), rv) #define HASH_RANDOM_DESTROY typedef struct __glxHashBucket diff --git a/src/mesa/drivers/common/driverfuncs.c b/src/mesa/drivers/common/driverfuncs.c index a9f3c8e7271..f09106b77c3 100644 --- a/src/mesa/drivers/common/driverfuncs.c +++ b/src/mesa/drivers/common/driverfuncs.c @@ -56,6 +56,7 @@ #include "swrast/swrast.h" #include "driverfuncs.h" +#include "meta.h" @@ -100,11 +101,11 @@ _mesa_init_driver_functions(struct dd_function_table *driver) driver->TexSubImage2D = _mesa_store_texsubimage2d; driver->TexSubImage3D = _mesa_store_texsubimage3d; driver->GetTexImage = _mesa_get_teximage; - driver->CopyTexImage1D = _swrast_copy_teximage1d; - driver->CopyTexImage2D = _swrast_copy_teximage2d; - driver->CopyTexSubImage1D = _swrast_copy_texsubimage1d; - driver->CopyTexSubImage2D = _swrast_copy_texsubimage2d; - driver->CopyTexSubImage3D = _swrast_copy_texsubimage3d; + driver->CopyTexImage1D = _mesa_meta_CopyTexImage1D; + driver->CopyTexImage2D = _mesa_meta_CopyTexImage2D; + driver->CopyTexSubImage1D = _mesa_meta_CopyTexSubImage1D; + driver->CopyTexSubImage2D = _mesa_meta_CopyTexSubImage2D; + driver->CopyTexSubImage3D = _mesa_meta_CopyTexSubImage3D; driver->GenerateMipmap = _mesa_generate_mipmap; driver->TestProxyTexImage = _mesa_test_proxy_teximage; driver->CompressedTexImage1D = _mesa_store_compressed_teximage1d; @@ -129,10 +130,10 @@ _mesa_init_driver_functions(struct dd_function_table *driver) driver->UpdateTexturePalette = NULL; /* imaging */ - driver->CopyColorTable = _swrast_CopyColorTable; - driver->CopyColorSubTable = _swrast_CopyColorSubTable; - driver->CopyConvolutionFilter1D = _swrast_CopyConvolutionFilter1D; - driver->CopyConvolutionFilter2D = _swrast_CopyConvolutionFilter2D; + driver->CopyColorTable = _mesa_meta_CopyColorTable; + driver->CopyColorSubTable = _mesa_meta_CopyColorSubTable; + driver->CopyConvolutionFilter1D = _mesa_meta_CopyConvolutionFilter1D; + driver->CopyConvolutionFilter2D = _mesa_meta_CopyConvolutionFilter2D; /* Vertex/fragment programs */ driver->BindProgram = NULL; diff --git a/src/mesa/drivers/common/meta.c b/src/mesa/drivers/common/meta.c index 0b9781027e7..b445323aabe 100644 --- a/src/mesa/drivers/common/meta.c +++ b/src/mesa/drivers/common/meta.c @@ -38,6 +38,8 @@ #include "main/blend.h" #include "main/bufferobj.h" #include "main/buffers.h" +#include "main/colortab.h" +#include "main/convolve.h" #include "main/depth.h" #include "main/enable.h" #include "main/fbobject.h" @@ -49,6 +51,7 @@ #include "main/readpix.h" #include "main/scissor.h" #include "main/shaders.h" +#include "main/state.h" #include "main/stencil.h" #include "main/texobj.h" #include "main/texenv.h" @@ -63,6 +66,33 @@ #include "drivers/common/meta.h" +/** Return offset in bytes of the field within a vertex struct */ +#define OFFSET(FIELD) ((void *) offsetof(struct vertex, FIELD)) + + +/** + * Flags passed to _mesa_meta_begin(). + */ +/*@{*/ +#define META_ALL ~0x0 +#define META_ALPHA_TEST 0x1 +#define META_BLEND 0x2 /**< includes logicop */ +#define META_COLOR_MASK 0x4 +#define META_DEPTH_TEST 0x8 +#define META_FOG 0x10 +#define META_PIXEL_STORE 0x20 +#define META_PIXEL_TRANSFER 0x40 +#define META_RASTERIZATION 0x80 +#define META_SCISSOR 0x100 +#define META_SHADER 0x200 +#define META_STENCIL_TEST 0x400 +#define META_TRANSFORM 0x800 /**< modelview, projection, clip planes */ +#define META_TEXTURE 0x1000 +#define META_VERTEX 0x2000 +#define META_VIEWPORT 0x4000 +/*@}*/ + + /** * State which we may save/restore across meta ops. * XXX this may be incomplete... @@ -90,6 +120,17 @@ struct save_state /** META_PIXEL_STORE */ struct gl_pixelstore_attrib Pack, Unpack; + /** META_PIXEL_TRANSFER */ + GLfloat RedBias, RedScale; + GLfloat GreenBias, GreenScale; + GLfloat BlueBias, BlueScale; + GLfloat AlphaBias, AlphaScale; + GLfloat DepthBias, DepthScale; + GLboolean MapColorFlag; + GLboolean Convolution1DEnabled; + GLboolean Convolution2DEnabled; + GLboolean Separable2DEnabled; + /** META_RASTERIZATION */ GLenum FrontPolygonMode, BackPolygonMode; GLboolean PolygonOffset; @@ -195,7 +236,6 @@ struct copypix_state struct drawpix_state { GLuint ArrayObj; - GLuint VBO; GLuint StencilFP; /**< Fragment program for drawing stencil images */ GLuint DepthFP; /**< Fragment program for drawing depth images */ @@ -233,12 +273,12 @@ struct gl_meta_state struct temp_texture TempTex; - struct blit_state Blit; /**< For _mesa_meta_blit_framebuffer() */ - struct clear_state Clear; /**< For _mesa_meta_clear() */ - struct copypix_state CopyPix; /**< For _mesa_meta_copy_pixels() */ - struct drawpix_state DrawPix; /**< For _mesa_meta_draw_pixels() */ - struct bitmap_state Bitmap; /**< For _mesa_meta_bitmap() */ - struct gen_mipmap_state Mipmap; /**< For _mesa_meta_generate_mipmap() */ + struct blit_state Blit; /**< For _mesa_meta_BlitFramebuffer() */ + struct clear_state Clear; /**< For _mesa_meta_Clear() */ + struct copypix_state CopyPix; /**< For _mesa_meta_CopyPixels() */ + struct drawpix_state DrawPix; /**< For _mesa_meta_DrawPixels() */ + struct bitmap_state Bitmap; /**< For _mesa_meta_Bitmap() */ + struct gen_mipmap_state Mipmap; /**< For _mesa_meta_GenerateMipmap() */ }; @@ -286,7 +326,6 @@ _mesa_meta_free(GLcontext *ctx) _mesa_DeleteVertexArraysAPPLE(1, &meta->CopyPix.ArrayObj); /* glDrawPixels */ - _mesa_DeleteBuffersARB(1, & meta->DrawPix.VBO); _mesa_DeleteVertexArraysAPPLE(1, &meta->DrawPix.ArrayObj); _mesa_DeletePrograms(1, &meta->DrawPix.DepthFP); _mesa_DeletePrograms(1, &meta->DrawPix.StencilFP); @@ -359,6 +398,35 @@ _mesa_meta_begin(GLcontext *ctx, GLbitfield state) ctx->Unpack = ctx->DefaultPacking; } + if (state & META_PIXEL_TRANSFER) { + save->RedScale = ctx->Pixel.RedScale; + save->RedBias = ctx->Pixel.RedBias; + save->GreenScale = ctx->Pixel.GreenScale; + save->GreenBias = ctx->Pixel.GreenBias; + save->BlueScale = ctx->Pixel.BlueScale; + save->BlueBias = ctx->Pixel.BlueBias; + save->AlphaScale = ctx->Pixel.AlphaScale; + save->AlphaBias = ctx->Pixel.AlphaBias; + save->MapColorFlag = ctx->Pixel.MapColorFlag; + save->Convolution1DEnabled = ctx->Pixel.Convolution1DEnabled; + save->Convolution2DEnabled = ctx->Pixel.Convolution2DEnabled; + save->Separable2DEnabled = ctx->Pixel.Separable2DEnabled; + ctx->Pixel.RedScale = 1.0F; + ctx->Pixel.RedBias = 0.0F; + ctx->Pixel.GreenScale = 1.0F; + ctx->Pixel.GreenBias = 0.0F; + ctx->Pixel.BlueScale = 1.0F; + ctx->Pixel.BlueBias = 0.0F; + ctx->Pixel.AlphaScale = 1.0F; + ctx->Pixel.AlphaBias = 0.0F; + ctx->Pixel.MapColorFlag = GL_FALSE; + ctx->Pixel.Convolution1DEnabled = GL_FALSE; + ctx->Pixel.Convolution2DEnabled = GL_FALSE; + ctx->Pixel.Separable2DEnabled = GL_FALSE; + /* XXX more state */ + ctx->NewState |=_NEW_PIXEL; + } + if (state & META_RASTERIZATION) { save->FrontPolygonMode = ctx->Polygon.FrontMode; save->BackPolygonMode = ctx->Polygon.BackMode; @@ -411,11 +479,6 @@ _mesa_meta_begin(GLcontext *ctx, GLbitfield state) save->ClientActiveUnit = ctx->Array.ActiveTexture; save->EnvMode = ctx->Texture.Unit[0].EnvMode; - if (ctx->Texture._EnabledUnits | - ctx->Texture._EnabledCoordUnits | - ctx->Texture._TexGenEnabled | - ctx->Texture._TexMatEnabled) { - /* Disable all texture units */ for (u = 0; u < ctx->Const.MaxTextureUnits; u++) { save->TexEnabled[u] = ctx->Texture.Unit[u].Enabled; @@ -434,7 +497,6 @@ _mesa_meta_begin(GLcontext *ctx, GLbitfield state) _mesa_set_enable(ctx, GL_TEXTURE_GEN_Q, GL_FALSE); } } - } /* save current texture objects for unit[0] only */ for (tgt = 0; tgt < NUM_TEXTURE_TARGETS; tgt++) { @@ -560,6 +622,23 @@ _mesa_meta_end(GLcontext *ctx) ctx->Unpack = save->Unpack; } + if (state & META_PIXEL_TRANSFER) { + ctx->Pixel.RedScale = save->RedScale; + ctx->Pixel.RedBias = save->RedBias; + ctx->Pixel.GreenScale = save->GreenScale; + ctx->Pixel.GreenBias = save->GreenBias; + ctx->Pixel.BlueScale = save->BlueScale; + ctx->Pixel.BlueBias = save->BlueBias; + ctx->Pixel.AlphaScale = save->AlphaScale; + ctx->Pixel.AlphaBias = save->AlphaBias; + ctx->Pixel.MapColorFlag = save->MapColorFlag; + ctx->Pixel.Convolution1DEnabled = save->Convolution1DEnabled; + ctx->Pixel.Convolution2DEnabled = save->Convolution2DEnabled; + ctx->Pixel.Separable2DEnabled = save->Separable2DEnabled; + /* XXX more state */ + ctx->NewState |=_NEW_PIXEL; + } + if (state & META_RASTERIZATION) { _mesa_PolygonMode(GL_FRONT, save->FrontPolygonMode); _mesa_PolygonMode(GL_BACK, save->BackPolygonMode); @@ -691,7 +770,6 @@ _mesa_meta_end(GLcontext *ctx) _mesa_MatrixMode(save->MatrixMode); - save->ClipPlanesEnabled = ctx->Transform.ClipPlanesEnabled; if (save->ClipPlanesEnabled) { GLuint i; for (i = 0; i < ctx->Const.MaxClipPlanes; i++) { @@ -727,9 +805,6 @@ _mesa_meta_end(GLcontext *ctx) if (save->Lighting) { _mesa_set_enable(ctx, GL_LIGHTING, GL_TRUE); } - if (save->Fog) { - _mesa_set_enable(ctx, GL_FOG, GL_TRUE); - } } @@ -974,10 +1049,10 @@ init_blit_depth_pixels(GLcontext *ctx) * of texture mapping and polygon rendering. */ void -_mesa_meta_blit_framebuffer(GLcontext *ctx, - GLint srcX0, GLint srcY0, GLint srcX1, GLint srcY1, - GLint dstX0, GLint dstY0, GLint dstX1, GLint dstY1, - GLbitfield mask, GLenum filter) +_mesa_meta_BlitFramebuffer(GLcontext *ctx, + GLint srcX0, GLint srcY0, GLint srcX1, GLint srcY1, + GLint dstX0, GLint dstY0, GLint dstX1, GLint dstY1, + GLbitfield mask, GLenum filter) { struct blit_state *blit = &ctx->Meta->Blit; struct temp_texture *tex = get_temp_texture(ctx); @@ -988,7 +1063,10 @@ _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 +1105,8 @@ _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 +1119,23 @@ _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); @@ -1122,10 +1198,13 @@ _mesa_meta_blit_framebuffer(GLcontext *ctx, * Meta implementation of ctx->Driver.Clear() in terms of polygon rendering. */ void -_mesa_meta_clear(GLcontext *ctx, GLbitfield buffers) +_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 +1229,8 @@ _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 +1280,25 @@ _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 */ @@ -1237,13 +1317,16 @@ _mesa_meta_clear(GLcontext *ctx, GLbitfield buffers) * of texture mapping and polygon rendering. */ void -_mesa_meta_copy_pixels(GLcontext *ctx, GLint srcX, GLint srcY, - GLsizei width, GLsizei height, - GLint dstX, GLint dstY, GLenum type) +_mesa_meta_CopyPixels(GLcontext *ctx, GLint srcX, GLint srcY, + GLsizei width, GLsizei height, + GLint dstX, GLint dstY, GLenum type) { 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 +1364,8 @@ _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 +1384,26 @@ _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); @@ -1375,9 +1456,8 @@ tiled_draw_pixels(GLcontext *ctx, tileUnpack.SkipRows = unpack->SkipRows + j; - _mesa_meta_draw_pixels(ctx, tileX, tileY, - tileWidth, tileHeight, - format, type, &tileUnpack, pixels); + _mesa_meta_DrawPixels(ctx, tileX, tileY, tileWidth, tileHeight, + format, type, &tileUnpack, pixels); } } } @@ -1486,20 +1566,24 @@ init_draw_depth_pixels(GLcontext *ctx) * of texture mapping and polygon rendering. */ void -_mesa_meta_draw_pixels(GLcontext *ctx, - GLint x, GLint y, GLsizei width, GLsizei height, - GLenum format, GLenum type, - const struct gl_pixelstore_attrib *unpack, - const GLvoid *pixels) +_mesa_meta_DrawPixels(GLcontext *ctx, + GLint x, GLint y, GLsizei width, GLsizei height, + GLenum format, GLenum type, + const struct gl_pixelstore_attrib *unpack, + const GLvoid *pixels) { struct drawpix_state *drawpix = &ctx->Meta->DrawPix; 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; + GLuint vbo; /* * Determine if we can do the glDrawPixels with texture mapping. @@ -1579,32 +1663,6 @@ _mesa_meta_draw_pixels(GLcontext *ctx, META_VIEWPORT | metaExtraSave)); - if (drawpix->ArrayObj == 0) { - /* one-time setup */ - - /* create vertex array object */ - _mesa_GenVertexArrays(1, &drawpix->ArrayObj); - _mesa_BindVertexArray(drawpix->ArrayObj); - - /* create vertex array buffer */ - _mesa_GenBuffersARB(1, &drawpix->VBO); - _mesa_BindBufferARB(GL_ARRAY_BUFFER_ARB, drawpix->VBO); - _mesa_BufferDataARB(GL_ARRAY_BUFFER_ARB, sizeof(verts), - 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_EnableClientState(GL_VERTEX_ARRAY); - _mesa_EnableClientState(GL_TEXTURE_COORD_ARRAY); - } - else { - _mesa_BindVertexArray(drawpix->ArrayObj); - _mesa_BindBufferARB(GL_ARRAY_BUFFER_ARB, drawpix->VBO); - } - newTex = alloc_texture(tex, width, height, texIntFormat); /* vertex positions, texcoords (after texture allocation!) */ @@ -1615,30 +1673,45 @@ _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); + if (drawpix->ArrayObj == 0) { + /* one-time setup: create vertex array object */ + _mesa_GenVertexArrays(1, &drawpix->ArrayObj); } + _mesa_BindVertexArray(drawpix->ArrayObj); + + /* create vertex array buffer */ + _mesa_GenBuffersARB(1, &vbo); + _mesa_BindBufferARB(GL_ARRAY_BUFFER_ARB, vbo); + _mesa_BufferDataARB(GL_ARRAY_BUFFER_ARB, sizeof(verts), + verts, GL_DYNAMIC_DRAW_ARB); + + /* setup vertex arrays */ + _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); /* set given unpack params */ ctx->Unpack = *unpack; @@ -1709,6 +1782,8 @@ _mesa_meta_draw_pixels(GLcontext *ctx, _mesa_set_enable(ctx, tex->Target, GL_FALSE); + _mesa_DeleteBuffersARB(1, &vbo); + /* restore unpack params */ ctx->Unpack = unpackSave; @@ -1724,7 +1799,7 @@ _mesa_meta_draw_pixels(GLcontext *ctx, * improve performance a lot. */ void -_mesa_meta_bitmap(GLcontext *ctx, +_mesa_meta_Bitmap(GLcontext *ctx, GLint x, GLint y, GLsizei width, GLsizei height, const struct gl_pixelstore_attrib *unpack, const GLubyte *bitmap1) @@ -1733,7 +1808,10 @@ _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; @@ -1776,13 +1854,9 @@ _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); @@ -1803,32 +1877,32 @@ _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 */ @@ -1865,23 +1939,37 @@ _mesa_meta_bitmap(GLcontext *ctx, } +/** + * Called via ctx->Driver.GenerateMipmap() + * Note: texture borders and 3D texture support not yet complete. + */ void -_mesa_meta_generate_mipmap(GLcontext *ctx, GLenum target, - struct gl_texture_object *texObj) +_mesa_meta_GenerateMipmap(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; const GLenum magFilterSave = texObj->MagFilter; + const GLint baseLevelSave = texObj->BaseLevel; + const GLint maxLevelSave = texObj->MaxLevel; + const GLboolean genMipmapSave = texObj->GenerateMipmap; + const GLenum wrapSSave = texObj->WrapS; + const GLenum wrapTSave = texObj->WrapT; + const GLenum wrapRSave = texObj->WrapR; const GLuint fboSave = ctx->DrawBuffer->Name; GLenum faceTarget; - GLuint level; + GLuint dstLevel; GLuint border = 0; /* check for fallbacks */ - if (!ctx->Extensions.EXT_framebuffer_object) { + if (!ctx->Extensions.EXT_framebuffer_object || + target == GL_TEXTURE_3D) { _mesa_generate_mipmap(ctx, target, texObj); return; } @@ -1911,11 +1999,8 @@ _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); } @@ -1933,12 +2018,16 @@ _mesa_meta_generate_mipmap(GLcontext *ctx, GLenum target, _mesa_TexParameteri(target, GL_TEXTURE_MIN_FILTER, GL_LINEAR); _mesa_TexParameteri(target, GL_TEXTURE_MAG_FILTER, GL_LINEAR); + _mesa_TexParameteri(target, GL_GENERATE_MIPMAP, GL_FALSE); + _mesa_TexParameteri(target, GL_TEXTURE_WRAP_S, GL_CLAMP_TO_EDGE); + _mesa_TexParameteri(target, GL_TEXTURE_WRAP_T, GL_CLAMP_TO_EDGE); + _mesa_TexParameteri(target, GL_TEXTURE_WRAP_R, GL_CLAMP_TO_EDGE); + _mesa_set_enable(ctx, target, GL_TRUE); /* setup texcoords once (XXX what about border?) */ switch (faceTarget) { - case GL_TEXTURE_CUBE_MAP_POSITIVE_X: - break; + case GL_TEXTURE_1D: case GL_TEXTURE_2D: verts[0].s = 0.0F; verts[0].t = 0.0F; @@ -1953,63 +2042,180 @@ _mesa_meta_generate_mipmap(GLcontext *ctx, GLenum target, verts[3].t = 1.0F; verts[3].r = 0.0F; break; + case GL_TEXTURE_3D: + abort(); + break; + default: + /* cube face */ + { + static const GLfloat st[4][2] = { + {0.0f, 0.0f}, {1.0f, 0.0f}, {1.0f, 1.0f}, {0.0f, 1.0f} + }; + GLuint i; + + /* loop over quad verts */ + for (i = 0; i < 4; i++) { + /* Compute sc = +/-scale and tc = +/-scale. + * Not +/-1 to avoid cube face selection ambiguity near the edges, + * though that can still sometimes happen with this scale factor... + */ + const GLfloat scale = 0.9999f; + const GLfloat sc = (2.0f * st[i][0] - 1.0f) * scale; + const GLfloat tc = (2.0f * st[i][1] - 1.0f) * scale; + + switch (faceTarget) { + case GL_TEXTURE_CUBE_MAP_POSITIVE_X: + verts[i].s = 1.0f; + verts[i].t = -tc; + verts[i].r = -sc; + break; + case GL_TEXTURE_CUBE_MAP_NEGATIVE_X: + verts[i].s = -1.0f; + verts[i].t = -tc; + verts[i].r = sc; + break; + case GL_TEXTURE_CUBE_MAP_POSITIVE_Y: + verts[i].s = sc; + verts[i].t = 1.0f; + verts[i].r = tc; + break; + case GL_TEXTURE_CUBE_MAP_NEGATIVE_Y: + verts[i].s = sc; + verts[i].t = -1.0f; + verts[i].r = -tc; + break; + case GL_TEXTURE_CUBE_MAP_POSITIVE_Z: + verts[i].s = sc; + verts[i].t = -tc; + verts[i].r = 1.0f; + break; + case GL_TEXTURE_CUBE_MAP_NEGATIVE_Z: + verts[i].s = -sc; + verts[i].t = -tc; + verts[i].r = -1.0f; + break; + default: + assert(0); + } + } + } } + _mesa_set_enable(ctx, target, GL_TRUE); + + /* texture is already locked, unlock now */ + _mesa_unlock_texture(ctx, texObj); - for (level = baseLevel + 1; level <= maxLevel; level++) { + for (dstLevel = baseLevel + 1; dstLevel <= maxLevel; dstLevel++) { const struct gl_texture_image *srcImage; - const GLuint srcLevel = level - 1; - GLsizei srcWidth, srcHeight; - GLsizei newWidth, newHeight; + const GLuint srcLevel = dstLevel - 1; + GLsizei srcWidth, srcHeight, srcDepth; + GLsizei dstWidth, dstHeight, dstDepth; GLenum status; - srcImage = _mesa_select_tex_image(ctx, texObj, target, srcLevel); + srcImage = _mesa_select_tex_image(ctx, texObj, faceTarget, srcLevel); assert(srcImage->Border == 0); /* XXX we can fix this */ + /* src size w/out border */ srcWidth = srcImage->Width - 2 * border; srcHeight = srcImage->Height - 2 * border; + srcDepth = srcImage->Depth - 2 * border; - newWidth = MAX2(1, srcWidth / 2) + 2 * border; - newHeight = MAX2(1, srcHeight / 2) + 2 * border; + /* new dst size w/ border */ + dstWidth = MAX2(1, srcWidth / 2) + 2 * border; + dstHeight = MAX2(1, srcHeight / 2) + 2 * border; + dstDepth = MAX2(1, srcDepth / 2) + 2 * border; - if (newWidth == srcImage->Width && newHeight == srcImage->Height) { - break; + if (dstWidth == srcImage->Width && + dstHeight == srcImage->Height && + dstDepth == srcImage->Depth) { + /* all done */ + break; } - /* Create empty image */ - _mesa_TexImage2D(GL_TEXTURE_2D, level, srcImage->InternalFormat, - newWidth, newHeight, border, - GL_RGBA, GL_UNSIGNED_BYTE, NULL); + /* Create empty dest image */ + if (target == GL_TEXTURE_1D) { + _mesa_TexImage1D(target, dstLevel, srcImage->InternalFormat, + dstWidth, border, + GL_RGBA, GL_UNSIGNED_BYTE, NULL); + } + else if (target == GL_TEXTURE_3D) { + _mesa_TexImage3D(target, dstLevel, srcImage->InternalFormat, + dstWidth, dstHeight, dstDepth, border, + GL_RGBA, GL_UNSIGNED_BYTE, NULL); + } + else { + /* 2D or cube */ + _mesa_TexImage2D(faceTarget, dstLevel, srcImage->InternalFormat, + dstWidth, dstHeight, border, + GL_RGBA, GL_UNSIGNED_BYTE, NULL); + + if (target == GL_TEXTURE_CUBE_MAP) { + /* If texturing from a cube, we need to make sure all src faces + * have been defined (even if we're not sampling from them.) + * Otherwise the texture object will be 'incomplete' and + * texturing from it will not be allowed. + */ + GLuint face; + for (face = 0; face < 6; face++) { + if (!texObj->Image[face][srcLevel] || + texObj->Image[face][srcLevel]->Width != srcWidth) { + _mesa_TexImage2D(GL_TEXTURE_CUBE_MAP_POSITIVE_X + face, + srcLevel, srcImage->InternalFormat, + srcWidth, srcHeight, border, + GL_RGBA, GL_UNSIGNED_BYTE, NULL); + } + } + } + } - /* vertex positions */ + /* setup vertex positions */ { verts[0].x = 0.0F; verts[0].y = 0.0F; - verts[1].x = (GLfloat) newWidth; + verts[1].x = (GLfloat) dstWidth; verts[1].y = 0.0F; - verts[2].x = (GLfloat) newWidth; - verts[2].y = (GLfloat) newHeight; + verts[2].x = (GLfloat) dstWidth; + verts[2].y = (GLfloat) dstHeight; verts[3].x = 0.0F; - verts[3].y = (GLfloat) newHeight; + verts[3].y = (GLfloat) dstHeight; /* upload new vertex data */ _mesa_BufferSubDataARB(GL_ARRAY_BUFFER_ARB, 0, sizeof(verts), verts); } /* limit sampling to src level */ - _mesa_TexParameteri(GL_TEXTURE_2D, GL_TEXTURE_BASE_LEVEL, srcLevel); - _mesa_TexParameteri(GL_TEXTURE_2D, GL_TEXTURE_MAX_LEVEL, srcLevel); - - /* Set to draw into the current level */ - _mesa_FramebufferTexture2DEXT(GL_FRAMEBUFFER_EXT, - GL_COLOR_ATTACHMENT0_EXT, - target, - texObj->Name, - level); + _mesa_TexParameteri(target, GL_TEXTURE_BASE_LEVEL, srcLevel); + _mesa_TexParameteri(target, GL_TEXTURE_MAX_LEVEL, srcLevel); + + /* Set to draw into the current dstLevel */ + if (target == GL_TEXTURE_1D) { + _mesa_FramebufferTexture1DEXT(GL_FRAMEBUFFER_EXT, + GL_COLOR_ATTACHMENT0_EXT, + target, + texObj->Name, + dstLevel); + } + else if (target == GL_TEXTURE_3D) { + GLint zoffset = 0; /* XXX unfinished */ + _mesa_FramebufferTexture3DEXT(GL_FRAMEBUFFER_EXT, + GL_COLOR_ATTACHMENT0_EXT, + target, + texObj->Name, + dstLevel, zoffset); + } + else { + /* 2D / cube */ + _mesa_FramebufferTexture2DEXT(GL_FRAMEBUFFER_EXT, + GL_COLOR_ATTACHMENT0_EXT, + faceTarget, + texObj->Name, + dstLevel); + } - /* Choose to render to the color attachment. */ _mesa_DrawBuffer(GL_COLOR_ATTACHMENT0_EXT); + /* sanity check */ status = _mesa_CheckFramebufferStatusEXT (GL_FRAMEBUFFER_EXT); if (status != GL_FRAMEBUFFER_COMPLETE_EXT) { abort(); @@ -2019,11 +2225,386 @@ _mesa_meta_generate_mipmap(GLcontext *ctx, GLenum target, _mesa_DrawArrays(GL_TRIANGLE_FAN, 0, 4); } + _mesa_lock_texture(ctx, texObj); /* relock */ + _mesa_meta_end(ctx); _mesa_TexParameteri(target, GL_TEXTURE_MIN_FILTER, minFilterSave); _mesa_TexParameteri(target, GL_TEXTURE_MAG_FILTER, magFilterSave); + _mesa_TexParameteri(target, GL_TEXTURE_BASE_LEVEL, baseLevelSave); + _mesa_TexParameteri(target, GL_TEXTURE_MAX_LEVEL, maxLevelSave); + _mesa_TexParameteri(target, GL_GENERATE_MIPMAP, genMipmapSave); + _mesa_TexParameteri(target, GL_TEXTURE_WRAP_S, wrapSSave); + _mesa_TexParameteri(target, GL_TEXTURE_WRAP_T, wrapTSave); + _mesa_TexParameteri(target, GL_TEXTURE_WRAP_R, wrapRSave); - /* restore (XXX add to meta_begin/end()? */ _mesa_BindFramebufferEXT(GL_FRAMEBUFFER_EXT, fboSave); } + + +/** + * Determine the GL data type to use for the temporary image read with + * ReadPixels() and passed to Tex[Sub]Image(). + */ +static GLenum +get_temp_image_type(GLcontext *ctx, GLenum baseFormat) +{ + switch (baseFormat) { + case GL_RGBA: + case GL_RGB: + case GL_ALPHA: + case GL_LUMINANCE: + case GL_LUMINANCE_ALPHA: + case GL_INTENSITY: + if (ctx->DrawBuffer->Visual.redBits <= 8) + return GL_UNSIGNED_BYTE; + else if (ctx->DrawBuffer->Visual.redBits <= 8) + return GL_UNSIGNED_SHORT; + else + return GL_FLOAT; + case GL_DEPTH_COMPONENT: + return GL_UNSIGNED_INT; + case GL_DEPTH_STENCIL: + return GL_UNSIGNED_INT_24_8; + default: + _mesa_problem(ctx, "Unexpected format in get_temp_image_type()"); + return 0; + } +} + + +/** + * Helper for _mesa_meta_CopyTexImage1/2D() functions. + * Have to be careful with locking and meta state for pixel transfer. + */ +static void +copy_tex_image(GLcontext *ctx, GLuint dims, GLenum target, GLint level, + GLenum internalFormat, GLint x, GLint y, + GLsizei width, GLsizei height, GLint border) +{ + struct gl_texture_unit *texUnit; + struct gl_texture_object *texObj; + struct gl_texture_image *texImage; + GLsizei postConvWidth = width, postConvHeight = height; + GLenum format, type; + GLint bpp; + void *buf; + + texUnit = _mesa_get_current_tex_unit(ctx); + texObj = _mesa_select_tex_object(ctx, texUnit, target); + texImage = _mesa_get_tex_image(ctx, texObj, target, level); + + format = _mesa_base_tex_format(ctx, internalFormat); + type = get_temp_image_type(ctx, format); + bpp = _mesa_bytes_per_pixel(format, type); + if (bpp <= 0) { + _mesa_problem(ctx, "Bad bpp in meta copy_tex_image()"); + return; + } + + /* + * Alloc image buffer (XXX could use a PBO) + */ + buf = _mesa_malloc(width * height * bpp); + if (!buf) { + _mesa_error(ctx, GL_OUT_OF_MEMORY, "glCopyTexImage%uD", dims); + return; + } + + _mesa_unlock_texture(ctx, texObj); /* need to unlock first */ + + /* + * Read image from framebuffer (disable pixel transfer ops) + */ + _mesa_meta_begin(ctx, META_PIXEL_STORE | META_PIXEL_TRANSFER); + ctx->Driver.ReadPixels(ctx, x, y, width, height, + format, type, &ctx->Pack, buf); + _mesa_meta_end(ctx); + + /* + * Prepare for new texture image size/data + */ +#if FEATURE_convolve + if (_mesa_is_color_format(internalFormat)) { + _mesa_adjust_image_for_convolution(ctx, 2, + &postConvWidth, &postConvHeight); + } +#endif + + if (texImage->Data) { + ctx->Driver.FreeTexImageData(ctx, texImage); + } + + _mesa_init_teximage_fields(ctx, target, texImage, + postConvWidth, postConvHeight, 1, + border, internalFormat); + + /* + * Store texture data (with pixel transfer ops) + */ + _mesa_meta_begin(ctx, META_PIXEL_STORE); + + _mesa_update_state(ctx); /* to update pixel transfer state */ + + if (target == GL_TEXTURE_1D) { + ctx->Driver.TexImage1D(ctx, target, level, internalFormat, + width, border, format, type, + buf, &ctx->Unpack, texObj, texImage); + } + else { + ctx->Driver.TexImage2D(ctx, target, level, internalFormat, + width, height, border, format, type, + buf, &ctx->Unpack, texObj, texImage); + } + _mesa_meta_end(ctx); + + _mesa_lock_texture(ctx, texObj); /* re-lock */ + + _mesa_free(buf); +} + + +void +_mesa_meta_CopyTexImage1D(GLcontext *ctx, GLenum target, GLint level, + GLenum internalFormat, GLint x, GLint y, + GLsizei width, GLint border) +{ + copy_tex_image(ctx, 1, target, level, internalFormat, x, y, + width, 1, border); +} + + +void +_mesa_meta_CopyTexImage2D(GLcontext *ctx, GLenum target, GLint level, + GLenum internalFormat, GLint x, GLint y, + GLsizei width, GLsizei height, GLint border) +{ + copy_tex_image(ctx, 2, target, level, internalFormat, x, y, + width, height, border); +} + + + +/** + * Helper for _mesa_meta_CopyTexSubImage1/2/3D() functions. + * Have to be careful with locking and meta state for pixel transfer. + */ +static void +copy_tex_sub_image(GLcontext *ctx, GLuint dims, GLenum target, GLint level, + GLint xoffset, GLint yoffset, GLint zoffset, + GLint x, GLint y, + GLsizei width, GLsizei height) +{ + struct gl_texture_unit *texUnit; + struct gl_texture_object *texObj; + struct gl_texture_image *texImage; + GLenum format, type; + GLint bpp; + void *buf; + + texUnit = _mesa_get_current_tex_unit(ctx); + texObj = _mesa_select_tex_object(ctx, texUnit, target); + texImage = _mesa_select_tex_image(ctx, texObj, target, level); + + format = texImage->TexFormat->BaseFormat; + type = get_temp_image_type(ctx, format); + bpp = _mesa_bytes_per_pixel(format, type); + if (bpp <= 0) { + _mesa_problem(ctx, "Bad bpp in meta copy_tex_sub_image()"); + return; + } + + /* + * Alloc image buffer (XXX could use a PBO) + */ + buf = _mesa_malloc(width * height * bpp); + if (!buf) { + _mesa_error(ctx, GL_OUT_OF_MEMORY, "glCopyTexSubImage%uD", dims); + return; + } + + _mesa_unlock_texture(ctx, texObj); /* need to unlock first */ + + /* + * Read image from framebuffer (disable pixel transfer ops) + */ + _mesa_meta_begin(ctx, META_PIXEL_STORE | META_PIXEL_TRANSFER); + ctx->Driver.ReadPixels(ctx, x, y, width, height, + format, type, &ctx->Pack, buf); + _mesa_meta_end(ctx); + + _mesa_update_state(ctx); /* to update pixel transfer state */ + + /* + * Store texture data (with pixel transfer ops) + */ + _mesa_meta_begin(ctx, META_PIXEL_STORE); + if (target == GL_TEXTURE_1D) { + ctx->Driver.TexSubImage1D(ctx, target, level, xoffset, + width, format, type, buf, + &ctx->Unpack, texObj, texImage); + } + else if (target == GL_TEXTURE_3D) { + ctx->Driver.TexSubImage3D(ctx, target, level, xoffset, yoffset, zoffset, + width, height, 1, format, type, buf, + &ctx->Unpack, texObj, texImage); + } + else { + ctx->Driver.TexSubImage2D(ctx, target, level, xoffset, yoffset, + width, height, format, type, buf, + &ctx->Unpack, texObj, texImage); + } + _mesa_meta_end(ctx); + + _mesa_lock_texture(ctx, texObj); /* re-lock */ + + _mesa_free(buf); +} + + +void +_mesa_meta_CopyTexSubImage1D(GLcontext *ctx, GLenum target, GLint level, + GLint xoffset, + GLint x, GLint y, GLsizei width) +{ + copy_tex_sub_image(ctx, 1, target, level, xoffset, 0, 0, + x, y, width, 1); +} + + +void +_mesa_meta_CopyTexSubImage2D(GLcontext *ctx, GLenum target, GLint level, + GLint xoffset, GLint yoffset, + GLint x, GLint y, + GLsizei width, GLsizei height) +{ + copy_tex_sub_image(ctx, 2, target, level, xoffset, yoffset, 0, + x, y, width, height); +} + + +void +_mesa_meta_CopyTexSubImage3D(GLcontext *ctx, GLenum target, GLint level, + GLint xoffset, GLint yoffset, GLint zoffset, + GLint x, GLint y, + GLsizei width, GLsizei height) +{ + copy_tex_sub_image(ctx, 3, target, level, xoffset, yoffset, zoffset, + x, y, width, height); +} + + +void +_mesa_meta_CopyColorTable(GLcontext *ctx, + GLenum target, GLenum internalformat, + GLint x, GLint y, GLsizei width) +{ + GLfloat *buf; + + buf = (GLfloat *) _mesa_malloc(width * 4 * sizeof(GLfloat)); + if (!buf) { + _mesa_error(ctx, GL_OUT_OF_MEMORY, "glCopyColorTable"); + return; + } + + /* + * Read image from framebuffer (disable pixel transfer ops) + */ + _mesa_meta_begin(ctx, META_PIXEL_STORE | META_PIXEL_TRANSFER); + ctx->Driver.ReadPixels(ctx, x, y, width, 1, + GL_RGBA, GL_FLOAT, &ctx->Pack, buf); + + _mesa_ColorTable(target, internalformat, width, GL_RGBA, GL_FLOAT, buf); + + _mesa_meta_end(ctx); + + _mesa_free(buf); +} + + +void +_mesa_meta_CopyColorSubTable(GLcontext *ctx,GLenum target, GLsizei start, + GLint x, GLint y, GLsizei width) +{ + GLfloat *buf; + + buf = (GLfloat *) _mesa_malloc(width * 4 * sizeof(GLfloat)); + if (!buf) { + _mesa_error(ctx, GL_OUT_OF_MEMORY, "glCopyColorSubTable"); + return; + } + + /* + * Read image from framebuffer (disable pixel transfer ops) + */ + _mesa_meta_begin(ctx, META_PIXEL_STORE | META_PIXEL_TRANSFER); + ctx->Driver.ReadPixels(ctx, x, y, width, 1, + GL_RGBA, GL_FLOAT, &ctx->Pack, buf); + + _mesa_ColorSubTable(target, start, width, GL_RGBA, GL_FLOAT, buf); + + _mesa_meta_end(ctx); + + _mesa_free(buf); +} + + +void +_mesa_meta_CopyConvolutionFilter1D(GLcontext *ctx, GLenum target, + GLenum internalFormat, + GLint x, GLint y, GLsizei width) +{ + GLfloat *buf; + + buf = (GLfloat *) _mesa_malloc(width * 4 * sizeof(GLfloat)); + if (!buf) { + _mesa_error(ctx, GL_OUT_OF_MEMORY, "glCopyConvolutionFilter2D"); + return; + } + + /* + * Read image from framebuffer (disable pixel transfer ops) + */ + _mesa_meta_begin(ctx, META_PIXEL_STORE | META_PIXEL_TRANSFER); + _mesa_update_state(ctx); + ctx->Driver.ReadPixels(ctx, x, y, width, 1, + GL_RGBA, GL_FLOAT, &ctx->Pack, buf); + + _mesa_ConvolutionFilter1D(target, internalFormat, width, + GL_RGBA, GL_FLOAT, buf); + + _mesa_meta_end(ctx); + + _mesa_free(buf); +} + + +void +_mesa_meta_CopyConvolutionFilter2D(GLcontext *ctx, GLenum target, + GLenum internalFormat, GLint x, GLint y, + GLsizei width, GLsizei height) +{ + GLfloat *buf; + + buf = (GLfloat *) _mesa_malloc(width * height * 4 * sizeof(GLfloat)); + if (!buf) { + _mesa_error(ctx, GL_OUT_OF_MEMORY, "glCopyConvolutionFilter2D"); + return; + } + + /* + * Read image from framebuffer (disable pixel transfer ops) + */ + _mesa_meta_begin(ctx, META_PIXEL_STORE | META_PIXEL_TRANSFER); + _mesa_update_state(ctx); + + ctx->Driver.ReadPixels(ctx, x, y, width, height, + GL_RGBA, GL_FLOAT, &ctx->Pack, buf); + + _mesa_ConvolutionFilter2D(target, internalFormat, width, height, + GL_RGBA, GL_FLOAT, buf); + + _mesa_meta_end(ctx); + + _mesa_free(buf); +} diff --git a/src/mesa/drivers/common/meta.h b/src/mesa/drivers/common/meta.h index 171ad27f265..7f659528dc0 100644 --- a/src/mesa/drivers/common/meta.h +++ b/src/mesa/drivers/common/meta.h @@ -27,29 +27,6 @@ #define META_H -/** - * Flags passed to _mesa_meta_begin(). - * XXX these flags may evolve... - */ -/*@{*/ -#define META_ALPHA_TEST 0x1 -#define META_BLEND 0x2 /**< includes logicop */ -#define META_COLOR_MASK 0x4 -#define META_DEPTH_TEST 0x8 -#define META_FOG 0x10 -#define META_RASTERIZATION 0x20 -#define META_SCISSOR 0x40 -#define META_SHADER 0x80 -#define META_STENCIL_TEST 0x100 -#define META_TRANSFORM 0x200 /**< modelview, projection */ -#define META_TEXTURE 0x400 -#define META_VERTEX 0x800 -#define META_VIEWPORT 0x1000 -#define META_PIXEL_STORE 0x2000 -#define META_ALL ~0x0 -/*@}*/ - - extern void _mesa_meta_init(GLcontext *ctx); @@ -57,35 +34,81 @@ extern void _mesa_meta_free(GLcontext *ctx); extern void -_mesa_meta_blit_framebuffer(GLcontext *ctx, - GLint srcX0, GLint srcY0, GLint srcX1, GLint srcY1, - GLint dstX0, GLint dstY0, GLint dstX1, GLint dstY1, - GLbitfield mask, GLenum filter); +_mesa_meta_BlitFramebuffer(GLcontext *ctx, + GLint srcX0, GLint srcY0, GLint srcX1, GLint srcY1, + GLint dstX0, GLint dstY0, GLint dstX1, GLint dstY1, + GLbitfield mask, GLenum filter); extern void -_mesa_meta_clear(GLcontext *ctx, GLbitfield buffers); +_mesa_meta_Clear(GLcontext *ctx, GLbitfield buffers); extern void -_mesa_meta_copy_pixels(GLcontext *ctx, GLint srcx, GLint srcy, - GLsizei width, GLsizei height, - GLint dstx, GLint dsty, GLenum type); +_mesa_meta_CopyPixels(GLcontext *ctx, GLint srcx, GLint srcy, + GLsizei width, GLsizei height, + GLint dstx, GLint dsty, GLenum type); extern void -_mesa_meta_draw_pixels(GLcontext *ctx, - GLint x, GLint y, GLsizei width, GLsizei height, - GLenum format, GLenum type, - const struct gl_pixelstore_attrib *unpack, - const GLvoid *pixels); +_mesa_meta_DrawPixels(GLcontext *ctx, + GLint x, GLint y, GLsizei width, GLsizei height, + GLenum format, GLenum type, + const struct gl_pixelstore_attrib *unpack, + const GLvoid *pixels); extern void -_mesa_meta_bitmap(GLcontext *ctx, +_mesa_meta_Bitmap(GLcontext *ctx, GLint x, GLint y, GLsizei width, GLsizei height, const struct gl_pixelstore_attrib *unpack, const GLubyte *bitmap); extern void -_mesa_meta_generate_mipmap(GLcontext *ctx, GLenum target, - struct gl_texture_object *texObj); +_mesa_meta_GenerateMipmap(GLcontext *ctx, GLenum target, + struct gl_texture_object *texObj); + +extern void +_mesa_meta_CopyTexImage1D(GLcontext *ctx, GLenum target, GLint level, + GLenum internalFormat, GLint x, GLint y, + GLsizei width, GLint border); + +extern void +_mesa_meta_CopyTexImage2D(GLcontext *ctx, GLenum target, GLint level, + GLenum internalFormat, GLint x, GLint y, + GLsizei width, GLsizei height, GLint border); + +extern void +_mesa_meta_CopyTexSubImage1D(GLcontext *ctx, GLenum target, GLint level, + GLint xoffset, + GLint x, GLint y, GLsizei width); + +extern void +_mesa_meta_CopyTexSubImage2D(GLcontext *ctx, GLenum target, GLint level, + GLint xoffset, GLint yoffset, + GLint x, GLint y, + GLsizei width, GLsizei height); + +extern void +_mesa_meta_CopyTexSubImage3D(GLcontext *ctx, GLenum target, GLint level, + GLint xoffset, GLint yoffset, GLint zoffset, + GLint x, GLint y, + GLsizei width, GLsizei height); + +extern void +_mesa_meta_CopyColorTable(GLcontext *ctx, + GLenum target, GLenum internalformat, + GLint x, GLint y, GLsizei width); + +extern void +_mesa_meta_CopyColorSubTable(GLcontext *ctx,GLenum target, GLsizei start, + GLint x, GLint y, GLsizei width); + +extern void +_mesa_meta_CopyConvolutionFilter1D(GLcontext *ctx, GLenum target, + GLenum internalFormat, + GLint x, GLint y, GLsizei width); + +extern void +_mesa_meta_CopyConvolutionFilter2D(GLcontext *ctx, GLenum target, + GLenum internalFormat, GLint x, GLint y, + GLsizei width, GLsizei height); #endif /* META_H */ diff --git a/src/mesa/drivers/dri/common/extension_helper.h b/src/mesa/drivers/dri/common/extension_helper.h index 2c89a9f58a9..5e86324eece 100644 --- a/src/mesa/drivers/dri/common/extension_helper.h +++ b/src/mesa/drivers/dri/common/extension_helper.h @@ -4417,10 +4417,11 @@ static const char SpriteParameterivSGIX_names[] = ""; #endif -#if defined(need_GL_EXT_provoking_vertex) +#if defined(need_GL_EXT_provoking_vertex) || defined(need_GL_ARB_provoking_vertex) static const char ProvokingVertexEXT_names[] = "i\0" /* Parameter signature */ "glProvokingVertexEXT\0" + "glProvokingVertex\0" ""; #endif @@ -5194,6 +5195,13 @@ static const struct dri_extension_function GL_ARB_point_parameters_functions[] = }; #endif +#if defined(need_GL_ARB_provoking_vertex) +static const struct dri_extension_function GL_ARB_provoking_vertex_functions[] = { + { ProvokingVertexEXT_names, ProvokingVertexEXT_remap_index, -1 }, + { NULL, 0, 0 } +}; +#endif + #if defined(need_GL_ARB_shader_objects) static const struct dri_extension_function GL_ARB_shader_objects_functions[] = { { UniformMatrix3fvARB_names, UniformMatrix3fvARB_remap_index, -1 }, diff --git a/src/mesa/drivers/dri/ffb/ffb_tex.c b/src/mesa/drivers/dri/ffb/ffb_tex.c index 69d30aedbaf..95058e9069f 100644 --- a/src/mesa/drivers/dri/ffb/ffb_tex.c +++ b/src/mesa/drivers/dri/ffb/ffb_tex.c @@ -30,24 +30,6 @@ #include "ffb_tex.h" /* No texture unit, all software. */ -/* XXX this function isn't needed since _mesa_init_driver_functions() - * will make all these assignments. - */ void ffbDDInitTexFuncs(GLcontext *ctx) { - /* - ctx->Driver.ChooseTextureFormat = _mesa_choose_tex_format; - ctx->Driver.TexImage1D = _mesa_store_teximage1d; - ctx->Driver.TexImage2D = _mesa_store_teximage2d; - ctx->Driver.TexImage3D = _mesa_store_teximage3d; - ctx->Driver.TexSubImage1D = _mesa_store_texsubimage1d; - ctx->Driver.TexSubImage2D = _mesa_store_texsubimage2d; - ctx->Driver.TexSubImage3D = _mesa_store_texsubimage3d; - ctx->Driver.CopyTexImage1D = _swrast_copy_teximage1d; - ctx->Driver.CopyTexImage2D = _swrast_copy_teximage2d; - ctx->Driver.CopyTexSubImage1D = _swrast_copy_texsubimage1d; - ctx->Driver.CopyTexSubImage2D = _swrast_copy_texsubimage2d; - ctx->Driver.CopyTexSubImage3D = _swrast_copy_texsubimage3d; - ctx->Driver.TestProxyTexImage = _mesa_test_proxy_teximage; - */ } diff --git a/src/mesa/drivers/dri/glcore/Makefile b/src/mesa/drivers/dri/glcore/Makefile deleted file mode 100644 index ac7e1de9285..00000000000 --- a/src/mesa/drivers/dri/glcore/Makefile +++ /dev/null @@ -1,84 +0,0 @@ -# src/mesa/drivers/dri/glcore/Makefile - -TOP = ../../../../.. -include $(TOP)/configs/current - -LIBNAME = glcore_dri.so - -DRIVER_SOURCES = glcore_driver.c \ - $(TOP)/src/mesa/drivers/common/driverfuncs.c \ - ../common/dri_util.c - -C_SOURCES = \ - $(DRIVER_SOURCES) \ - $(DRI_SOURCES) - - -# Include directories -INCLUDE_DIRS = \ - -I. \ - -I../common \ - -I../dri_client \ - -I../dri_client/imports \ - -Iserver \ - -I$(TOP)/include \ - -I$(DRM_SOURCE_PATH)/shared-core \ - -I$(TOP)/src/mesa \ - -I$(TOP)/src/mesa/main \ - -I$(TOP)/src/mesa/glapi \ - -I$(TOP)/src/mesa/math \ - -I$(TOP)/src/mesa/transform \ - -I$(TOP)/src/mesa/shader \ - -I$(TOP)/src/mesa/swrast \ - -I$(TOP)/src/mesa/swrast_setup - -# Core Mesa objects -MESA_MODULES = $(TOP)/src/mesa/libmesa.a - -# Libraries that the driver shared lib depends on -LIB_DEPS = -lm -lpthread -lc -# LIB_DEPS = -lGL -lm -lpthread -lc - - -ASM_SOURCES = - -OBJECTS = $(C_SOURCES:.c=.o) \ - $(ASM_SOURCES:.S=.o) - - -##### RULES ##### - -.c.o: - $(CC) -c $(INCLUDE_DIRS) $(CFLAGS) $(DEFINES) $< -o $@ - -.S.o: - $(CC) -c $(INCLUDE_DIRS) $(CFLAGS) $(DEFINES) $< -o $@ - - -##### TARGETS ##### - -default: depend $(TOP)/$(LIB_DIR)/$(LIBNAME) - - -$(TOP)/$(LIB_DIR)/$(LIBNAME): $(OBJECTS) $(MESA_MODULES) $(WINOBJ) Makefile - CC="$(CC)" CXX="$(CXX)" $(TOP)/bin/mklib -o $(LIBNAME) -noprefix -install $(TOP)/$(LIB_DIR) \ - $(OBJECTS) $(WINLIB) $(LIB_DEPS) $(WINOBJ) $(MESA_MODULES) - - -depend: $(C_SOURCES) $(ASM_SOURCES) - rm -f depend - touch depend - $(MKDEP) $(MKDEP_OPTIONS) $(INCLUDE_DIRS) $(C_SOURCES) $(ASM_SOURCES) \ - > /dev/null - - -# Emacs tags -tags: - etags `find . -name \*.[ch]` `find ../include` - - -clean: - -rm -f *.o server/*.o - - -include depend diff --git a/src/mesa/drivers/dri/i965/brw_context.c b/src/mesa/drivers/dri/i965/brw_context.c index 3c5b8483197..c300c33adce 100644 --- a/src/mesa/drivers/dri/i965/brw_context.c +++ b/src/mesa/drivers/dri/i965/brw_context.c @@ -125,6 +125,7 @@ GLboolean brwCreateContext( const __GLcontextModes *mesaVis, /* We want the GLSL compiler to emit code that uses condition codes */ ctx->Shader.EmitCondCodes = GL_TRUE; + ctx->Shader.EmitNVTempInitialization = GL_TRUE; ctx->Const.VertexProgram.MaxNativeInstructions = (16 * 1024); ctx->Const.VertexProgram.MaxAluInstructions = 0; diff --git a/src/mesa/drivers/dri/i965/brw_curbe.c b/src/mesa/drivers/dri/i965/brw_curbe.c index 0b0e6931a06..4be6c77aa1e 100644 --- a/src/mesa/drivers/dri/i965/brw_curbe.c +++ b/src/mesa/drivers/dri/i965/brw_curbe.c @@ -248,6 +248,9 @@ static void prepare_constant_buffer(struct brw_context *brw) GLuint offset = brw->curbe.vs_start * 16; GLuint nr = brw->vs.prog_data->nr_params / 4; + if (brw->vertex_program->IsNVProgram) + _mesa_load_tracked_matrices(ctx); + /* Updates the ParamaterValues[i] pointers for all parameters of the * basic type of PROGRAM_STATE_VAR. */ diff --git a/src/mesa/drivers/dri/i965/brw_state.h b/src/mesa/drivers/dri/i965/brw_state.h index 78572356a3d..5335eac8951 100644 --- a/src/mesa/drivers/dri/i965/brw_state.h +++ b/src/mesa/drivers/dri/i965/brw_state.h @@ -86,9 +86,6 @@ const struct brw_tracked_state brw_psp_urb_cbs; const struct brw_tracked_state brw_pipe_control; -const struct brw_tracked_state brw_clear_surface_cache; -const struct brw_tracked_state brw_clear_batch_cache; - const struct brw_tracked_state brw_drawing_rect; const struct brw_tracked_state brw_indices; const struct brw_tracked_state brw_vertices; @@ -165,7 +162,7 @@ GLboolean brw_cached_batch_struct( struct brw_context *brw, const void *data, GLuint sz ); void brw_destroy_batch_cache( struct brw_context *brw ); -void brw_clear_batch_cache_flush( struct brw_context *brw ); +void brw_clear_batch_cache( struct brw_context *brw ); /* brw_wm_surface_state.c */ dri_bo * diff --git a/src/mesa/drivers/dri/i965/brw_state_batch.c b/src/mesa/drivers/dri/i965/brw_state_batch.c index 811940edc05..7821898cf9b 100644 --- a/src/mesa/drivers/dri/i965/brw_state_batch.c +++ b/src/mesa/drivers/dri/i965/brw_state_batch.c @@ -79,7 +79,7 @@ GLboolean brw_cached_batch_struct( struct brw_context *brw, return GL_TRUE; } -static void clear_batch_cache( struct brw_context *brw ) +void brw_clear_batch_cache( struct brw_context *brw ) { struct brw_cached_batch_item *item = brw->cached_batch_items; @@ -93,18 +93,7 @@ static void clear_batch_cache( struct brw_context *brw ) brw->cached_batch_items = NULL; } -void brw_clear_batch_cache_flush( struct brw_context *brw ) -{ - clear_batch_cache(brw); - - brw->state.dirty.mesa |= ~0; - brw->state.dirty.brw |= ~0; - brw->state.dirty.cache |= ~0; -} - - - void brw_destroy_batch_cache( struct brw_context *brw ) { - clear_batch_cache(brw); + brw_clear_batch_cache(brw); } diff --git a/src/mesa/drivers/dri/i965/brw_state_upload.c b/src/mesa/drivers/dri/i965/brw_state_upload.c index 414620d0b39..b817b741e77 100644 --- a/src/mesa/drivers/dri/i965/brw_state_upload.c +++ b/src/mesa/drivers/dri/i965/brw_state_upload.c @@ -287,6 +287,7 @@ void brw_validate_state( struct brw_context *brw ) if (brw->emit_state_always) { state->mesa |= ~0; state->brw |= ~0; + state->cache |= ~0; } if (brw->fragment_program != ctx->FragmentProgram._Current) { @@ -305,7 +306,7 @@ void brw_validate_state( struct brw_context *brw ) return; if (brw->state.dirty.brw & BRW_NEW_CONTEXT) - brw_clear_batch_cache_flush(brw); + brw_clear_batch_cache(brw); brw->intel.Fallback = 0; diff --git a/src/mesa/drivers/dri/i965/brw_vs.c b/src/mesa/drivers/dri/i965/brw_vs.c index e3111c66800..f0c79efbd96 100644 --- a/src/mesa/drivers/dri/i965/brw_vs.c +++ b/src/mesa/drivers/dri/i965/brw_vs.c @@ -90,8 +90,6 @@ static void brw_upload_vs_prog(struct brw_context *brw) struct brw_vertex_program *vp = (struct brw_vertex_program *)brw->vertex_program; - assert (vp && !vp->program.IsNVProgram); - memset(&key, 0, sizeof(key)); /* Just upload the program verbatim for now. Always send it all 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/intel/intel_buffer_objects.c b/src/mesa/drivers/dri/intel/intel_buffer_objects.c index c55c5c426e0..7f6fb66d52f 100644 --- a/src/mesa/drivers/dri/intel/intel_buffer_objects.c +++ b/src/mesa/drivers/dri/intel/intel_buffer_objects.c @@ -207,8 +207,12 @@ intel_bufferobj_subdata(GLcontext * ctx, if (intel_obj->sys_buffer) memcpy((char *)intel_obj->sys_buffer + offset, data, size); - else + else { + /* Flush any existing batchbuffer that might reference this data. */ + intelFlush(ctx); + dri_bo_subdata(intel_obj->buffer, offset, size, data); + } } diff --git a/src/mesa/drivers/dri/intel/intel_clear.c b/src/mesa/drivers/dri/intel/intel_clear.c index bce23724b38..fb62f0f430a 100644 --- a/src/mesa/drivers/dri/intel/intel_clear.c +++ b/src/mesa/drivers/dri/intel/intel_clear.c @@ -38,6 +38,7 @@ #include "intel_fbo.h" #include "intel_pixel.h" #include "intel_regions.h" +#include "intel_batchbuffer.h" #define FILE_DEBUG_FLAG DEBUG_BLIT @@ -170,7 +171,8 @@ intelClear(GLcontext *ctx, GLbitfield mask) } DBG("\n"); } - _mesa_meta_clear(&intel->ctx, tri_mask); + + _mesa_meta_Clear(&intel->ctx, tri_mask); } if (swrast_mask) { diff --git a/src/mesa/drivers/dri/intel/intel_context.c b/src/mesa/drivers/dri/intel/intel_context.c index 7e21b94acc4..d49d95768db 100644 --- a/src/mesa/drivers/dri/intel/intel_context.c +++ b/src/mesa/drivers/dri/intel/intel_context.c @@ -507,7 +507,8 @@ intel_flush(GLcontext *ctx, GLboolean needs_mi_flush) if (screen->dri2.loader && (screen->dri2.loader->base.version >= 2) - && (screen->dri2.loader->flushFrontBuffer != NULL)) { + && (screen->dri2.loader->flushFrontBuffer != NULL) && + intel->driDrawable && intel->driDrawable->loaderPrivate) { (*screen->dri2.loader->flushFrontBuffer)(intel->driDrawable, intel->driDrawable->loaderPrivate); @@ -587,11 +588,6 @@ intelInitDriverFunctions(struct dd_function_table *functions) functions->GetString = intelGetString; functions->UpdateState = intelInvalidateState; - functions->CopyColorTable = _swrast_CopyColorTable; - functions->CopyColorSubTable = _swrast_CopyColorSubTable; - functions->CopyConvolutionFilter1D = _swrast_CopyConvolutionFilter1D; - functions->CopyConvolutionFilter2D = _swrast_CopyConvolutionFilter2D; - intelInitTextureFuncs(functions); intelInitTextureImageFuncs(functions); intelInitTextureSubImageFuncs(functions); @@ -922,6 +918,14 @@ intelDestroyContext(__DRIcontextPrivate * driContextPriv) GLboolean intelUnbindContext(__DRIcontextPrivate * driContextPriv) { + struct intel_context *intel = + (struct intel_context *) driContextPriv->driverPrivate; + + /* Deassociate the context with the drawables. + */ + intel->driDrawable = NULL; + intel->driReadDrawable = NULL; + return GL_TRUE; } diff --git a/src/mesa/drivers/dri/intel/intel_fbo.c b/src/mesa/drivers/dri/intel/intel_fbo.c index 804c0348401..8dfb24290d5 100644 --- a/src/mesa/drivers/dri/intel/intel_fbo.c +++ b/src/mesa/drivers/dri/intel/intel_fbo.c @@ -715,5 +715,5 @@ intel_fbo_init(struct intel_context *intel) intel->ctx.Driver.FinishRenderTexture = intel_finish_render_texture; intel->ctx.Driver.ResizeBuffers = intel_resize_buffers; intel->ctx.Driver.ValidateFramebuffer = intel_validate_framebuffer; - intel->ctx.Driver.BlitFramebuffer = _mesa_meta_blit_framebuffer; + intel->ctx.Driver.BlitFramebuffer = _mesa_meta_BlitFramebuffer; } diff --git a/src/mesa/drivers/dri/intel/intel_generatemipmap.c b/src/mesa/drivers/dri/intel/intel_generatemipmap.c index fe986092db6..5958b36c971 100644 --- a/src/mesa/drivers/dri/intel/intel_generatemipmap.c +++ b/src/mesa/drivers/dri/intel/intel_generatemipmap.c @@ -125,6 +125,8 @@ intel_generate_mipmap_2d(GLcontext *ctx, GLuint fb_name; GLboolean success = GL_FALSE; struct gl_framebuffer *saved_fbo = NULL; + struct gl_buffer_object *saved_array_buffer = NULL; + struct gl_buffer_object *saved_element_buffer = NULL; _mesa_PushAttrib(GL_ENABLE_BIT | GL_TEXTURE_BIT | GL_CURRENT_BIT | GL_COLOR_BUFFER_BIT | @@ -133,6 +135,16 @@ intel_generate_mipmap_2d(GLcontext *ctx, old_active_texture = ctx->Texture.CurrentUnit; _mesa_reference_framebuffer(&saved_fbo, ctx->DrawBuffer); + /* use default array/index buffers */ + _mesa_reference_buffer_object(ctx, &saved_array_buffer, + ctx->Array.ArrayBufferObj); + _mesa_reference_buffer_object(ctx, &ctx->Array.ArrayBufferObj, + ctx->Shared->NullBufferObj); + _mesa_reference_buffer_object(ctx, &saved_element_buffer, + ctx->Array.ElementArrayBufferObj); + _mesa_reference_buffer_object(ctx, &ctx->Array.ElementArrayBufferObj, + ctx->Shared->NullBufferObj); + _mesa_Disable(GL_POLYGON_STIPPLE); _mesa_Disable(GL_DEPTH_TEST); _mesa_Disable(GL_STENCIL_TEST); @@ -205,6 +217,15 @@ fail: meta_restore_fragment_program(&intel->meta); meta_restore_vertex_program(&intel->meta); + /* restore array/index buffers */ + _mesa_reference_buffer_object(ctx, &ctx->Array.ArrayBufferObj, + saved_array_buffer); + _mesa_reference_buffer_object(ctx, &saved_array_buffer, NULL); + _mesa_reference_buffer_object(ctx, &ctx->Array.ElementArrayBufferObj, + saved_element_buffer); + _mesa_reference_buffer_object(ctx, &saved_element_buffer, NULL); + + _mesa_DeleteFramebuffersEXT(1, &fb_name); _mesa_ActiveTextureARB(GL_TEXTURE0_ARB + old_active_texture); if (saved_fbo) @@ -223,10 +244,6 @@ fail: * * The texture object's miptree must be mapped. * - * It would be really nice if this was just called by Mesa whenever mipmaps - * needed to be regenerated, rather than us having to remember to do so in - * each texture image modification path. - * * This function should also include an accelerated path. */ void diff --git a/src/mesa/drivers/dri/intel/intel_pixel_copy.c b/src/mesa/drivers/dri/intel/intel_pixel_copy.c index 07ca8f7ddb5..f058b3c8e4d 100644 --- a/src/mesa/drivers/dri/intel/intel_pixel_copy.c +++ b/src/mesa/drivers/dri/intel/intel_pixel_copy.c @@ -240,5 +240,5 @@ intelCopyPixels(GLcontext * ctx, return; /* this will use swrast if needed */ - _mesa_meta_copy_pixels(ctx, srcx, srcy, width, height, destx, desty, type); + _mesa_meta_CopyPixels(ctx, srcx, srcy, width, height, destx, desty, type); } diff --git a/src/mesa/drivers/dri/intel/intel_pixel_draw.c b/src/mesa/drivers/dri/intel/intel_pixel_draw.c index 7fbb89fd6aa..5ffa847fd4a 100644 --- a/src/mesa/drivers/dri/intel/intel_pixel_draw.c +++ b/src/mesa/drivers/dri/intel/intel_pixel_draw.c @@ -54,7 +54,7 @@ #include "intel_fbo.h" -/** XXX compare perf of this vs. _mesa_meta_draw_pixels(STENCIL) */ +/** XXX compare perf of this vs. _mesa_meta_DrawPixels(STENCIL) */ static GLboolean intel_stencil_drawpixels(GLcontext * ctx, GLint x, GLint y, @@ -265,7 +265,7 @@ intelDrawPixels(GLcontext * ctx, /* 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 + * Fall back to swrast because even the _mesa_meta_DrawPixels() approach * isn't working because of an apparent stencil bug. */ if (intel_stencil_drawpixels(ctx, x, y, width, height, format, type, @@ -280,6 +280,6 @@ intelDrawPixels(GLcontext * ctx, } #endif - _mesa_meta_draw_pixels(ctx, x, y, width, height, format, type, - unpack, pixels); + _mesa_meta_DrawPixels(ctx, x, y, width, height, format, type, + unpack, pixels); } diff --git a/src/mesa/drivers/dri/intel/intel_span.c b/src/mesa/drivers/dri/intel/intel_span.c index 1e739434577..28eabbc0054 100644 --- a/src/mesa/drivers/dri/intel/intel_span.c +++ b/src/mesa/drivers/dri/intel/intel_span.c @@ -448,24 +448,22 @@ intel_map_unmap_framebuffer(struct intel_context *intel, struct gl_framebuffer *fb, GLboolean map) { - GLcontext *ctx = &intel->ctx; - GLuint i, 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 { + GLuint i; + + /* color draw buffers */ + for (i = 0; i < fb->_NumColorDrawBuffers; i++) { if (map) - intel_renderbuffer_map(intel, fb->_ColorReadBuffer); + intel_renderbuffer_map(intel, fb->_ColorDrawBuffers[i]); else - intel_renderbuffer_unmap(intel, fb->_ColorReadBuffer); + intel_renderbuffer_unmap(intel, fb->_ColorDrawBuffers[i]); } + /* color read buffer */ + if (map) + intel_renderbuffer_map(intel, fb->_ColorReadBuffer); + else + intel_renderbuffer_unmap(intel, fb->_ColorReadBuffer); + /* check for render to textures */ for (i = 0; i < BUFFER_COUNT; i++) { struct gl_renderbuffer_attachment *att = @@ -484,20 +482,17 @@ intel_map_unmap_framebuffer(struct intel_context *intel, /* depth buffer (Note wrapper!) */ if (fb->_DepthBuffer) { if (map) - intel_renderbuffer_map(intel, fb->_DepthBuffer->Wrapped); + intel_renderbuffer_map(intel, fb->_DepthBuffer->Wrapped); else - intel_renderbuffer_unmap(intel, - fb->_DepthBuffer->Wrapped); + intel_renderbuffer_unmap(intel, fb->_DepthBuffer->Wrapped); } /* stencil buffer (Note wrapper!) */ if (fb->_StencilBuffer) { if (map) - intel_renderbuffer_map(intel, - fb->_StencilBuffer->Wrapped); + intel_renderbuffer_map(intel, fb->_StencilBuffer->Wrapped); else - intel_renderbuffer_unmap(intel, - fb->_StencilBuffer->Wrapped); + intel_renderbuffer_unmap(intel, fb->_StencilBuffer->Wrapped); } } @@ -524,7 +519,8 @@ intelSpanRenderStart(GLcontext * ctx) } intel_map_unmap_framebuffer(intel, ctx->DrawBuffer, GL_TRUE); - intel_map_unmap_framebuffer(intel, ctx->ReadBuffer, GL_TRUE); + if (ctx->ReadBuffer != ctx->DrawBuffer) + intel_map_unmap_framebuffer(intel, ctx->ReadBuffer, GL_TRUE); } /** @@ -547,7 +543,8 @@ intelSpanRenderFinish(GLcontext * ctx) } intel_map_unmap_framebuffer(intel, ctx->DrawBuffer, GL_FALSE); - intel_map_unmap_framebuffer(intel, ctx->ReadBuffer, GL_FALSE); + if (ctx->ReadBuffer != ctx->DrawBuffer) + intel_map_unmap_framebuffer(intel, ctx->ReadBuffer, GL_FALSE); UNLOCK_HARDWARE(intel); } diff --git a/src/mesa/drivers/dri/intel/intel_tex_copy.c b/src/mesa/drivers/dri/intel/intel_tex_copy.c index 028b49c14d3..ac557a92005 100644 --- a/src/mesa/drivers/dri/intel/intel_tex_copy.c +++ b/src/mesa/drivers/dri/intel/intel_tex_copy.c @@ -30,7 +30,8 @@ #include "main/image.h" #include "main/teximage.h" #include "main/mipmap.h" -#include "swrast/swrast.h" + +#include "drivers/common/meta.h" #include "intel_screen.h" #include "intel_context.h" @@ -90,7 +91,6 @@ do_copy_texsubimage(struct intel_context *intel, GLint x, GLint y, GLsizei width, GLsizei height) { GLcontext *ctx = &intel->ctx; - struct gl_texture_object *texObj = intelImage->base.TexObject; const struct intel_region *src = get_teximage_source(intel, internalFormat); @@ -170,11 +170,6 @@ do_copy_texsubimage(struct intel_context *intel, UNLOCK_HARDWARE(intel); - /* GL_SGIS_generate_mipmap */ - if (intelImage->level == texObj->BaseLevel && texObj->GenerateMipmap) { - intel_generate_mipmap(ctx, target, texObj); - } - return GL_TRUE; } @@ -221,8 +216,8 @@ intelCopyTexImage1D(GLcontext * ctx, GLenum target, GLint level, return; fail: - _swrast_copy_teximage1d(ctx, target, level, internalFormat, x, y, - width, border); + _mesa_meta_CopyTexImage1D(ctx, target, level, internalFormat, x, y, + width, border); } @@ -269,8 +264,8 @@ intelCopyTexImage2D(GLcontext * ctx, GLenum target, GLint level, return; fail: - _swrast_copy_teximage2d(ctx, target, level, internalFormat, x, y, - width, height, border); + _mesa_meta_CopyTexImage2D(ctx, target, level, internalFormat, x, y, + width, height, border); } @@ -294,7 +289,7 @@ intelCopyTexSubImage1D(GLcontext * ctx, GLenum target, GLint level, if (!do_copy_texsubimage(intel_context(ctx), target, intel_texture_image(texImage), internalFormat, xoffset, 0, x, y, width, 1)) { - _swrast_copy_texsubimage1d(ctx, target, level, xoffset, x, y, width); + _mesa_meta_CopyTexSubImage1D(ctx, target, level, xoffset, x, y, width); } } @@ -320,10 +315,10 @@ intelCopyTexSubImage2D(GLcontext * ctx, GLenum target, GLint level, internalFormat, xoffset, yoffset, x, y, width, height)) { - DBG("%s - fallback to swrast\n", __FUNCTION__); + DBG("%s - fallback to _mesa_meta_CopyTexSubImage2D\n", __FUNCTION__); - _swrast_copy_texsubimage2d(ctx, target, level, - xoffset, yoffset, x, y, width, height); + _mesa_meta_CopyTexSubImage2D(ctx, target, level, + xoffset, yoffset, x, y, width, height); } } diff --git a/src/mesa/drivers/dri/intel/intel_tex_image.c b/src/mesa/drivers/dri/intel/intel_tex_image.c index a206fe6805b..66201b1f465 100644 --- a/src/mesa/drivers/dri/intel/intel_tex_image.c +++ b/src/mesa/drivers/dri/intel/intel_tex_image.c @@ -546,11 +546,6 @@ intelTexImage(GLcontext * ctx, } UNLOCK_HARDWARE(intel); - - /* GL_SGIS_generate_mipmap */ - if (level == texObj->BaseLevel && texObj->GenerateMipmap) { - intel_generate_mipmap(ctx, target, texObj); - } } diff --git a/src/mesa/drivers/dri/intel/intel_tex_subimage.c b/src/mesa/drivers/dri/intel/intel_tex_subimage.c index 89037073f84..751ec2c98c2 100644 --- a/src/mesa/drivers/dri/intel/intel_tex_subimage.c +++ b/src/mesa/drivers/dri/intel/intel_tex_subimage.c @@ -129,11 +129,6 @@ intelTexSubimage(GLcontext * ctx, } UNLOCK_HARDWARE(intel); - - /* GL_SGIS_generate_mipmap */ - if (level == texObj->BaseLevel && texObj->GenerateMipmap) { - intel_generate_mipmap(ctx, target, texObj); - } } diff --git a/src/mesa/drivers/dri/r200/Makefile b/src/mesa/drivers/dri/r200/Makefile index 42635bf9d90..fbce70c37bc 100644 --- a/src/mesa/drivers/dri/r200/Makefile +++ b/src/mesa/drivers/dri/r200/Makefile @@ -55,8 +55,7 @@ C_SOURCES = $(COMMON_SOURCES) $(DRIVER_SOURCES) X86_SOURCES = -DRIVER_DEFINES = -DRADEON_COMMON=1 -DRADEON_COMMON_FOR_R200 \ - -Wall +DRIVER_DEFINES = -DRADEON_R200 -Wall DRI_LIB_DEPS += $(RADEON_LDFLAGS) diff --git a/src/mesa/drivers/dri/r300/Makefile b/src/mesa/drivers/dri/r300/Makefile index 5d8d6f66589..fe775eac99a 100644 --- a/src/mesa/drivers/dri/r300/Makefile +++ b/src/mesa/drivers/dri/r300/Makefile @@ -62,8 +62,7 @@ DRIVER_SOURCES = \ C_SOURCES = $(COMMON_SOURCES) $(DRIVER_SOURCES) -DRIVER_DEFINES = -DCOMPILE_R300 -DR200_MERGED=0 \ - -DRADEON_COMMON=1 -DRADEON_COMMON_FOR_R300 \ +DRIVER_DEFINES = -DRADEON_R300 # -DRADEON_BO_TRACK \ -Wall diff --git a/src/mesa/drivers/dri/r300/compiler/r300_fragprog_emit.c b/src/mesa/drivers/dri/r300/compiler/r300_fragprog_emit.c index 305dc074ee8..c7227bbd15b 100644 --- a/src/mesa/drivers/dri/r300/compiler/r300_fragprog_emit.c +++ b/src/mesa/drivers/dri/r300/compiler/r300_fragprog_emit.c @@ -352,7 +352,7 @@ void r300BuildFragmentProgramHwCode(struct r300_fragment_program_compiler *compi if (emit.current_node < 3) { int shift = 3 - emit.current_node; int i; - for(i = 0; i <= emit.current_node; ++i) + for(i = emit.current_node; i >= 0; --i) code->code_addr[shift + i] = code->code_addr[i]; for(i = 0; i < shift; ++i) code->code_addr[i] = 0; diff --git a/src/mesa/drivers/dri/r300/compiler/r3xx_vertprog.c b/src/mesa/drivers/dri/r300/compiler/r3xx_vertprog.c index 93a516105e6..dad27fc98e6 100644 --- a/src/mesa/drivers/dri/r300/compiler/r3xx_vertprog.c +++ b/src/mesa/drivers/dri/r300/compiler/r3xx_vertprog.c @@ -89,6 +89,7 @@ static unsigned long t_dst_index(struct r300_vertex_program_code *vp, static unsigned long t_src_class(gl_register_file file) { switch (file) { + case PROGRAM_BUILTIN: case PROGRAM_TEMPORARY: return PVS_SRC_REG_TEMPORARY; case PROGRAM_INPUT: diff --git a/src/mesa/drivers/dri/r300/compiler/radeon_program.c b/src/mesa/drivers/dri/r300/compiler/radeon_program.c index bbbf0dd7768..b636f90a968 100644 --- a/src/mesa/drivers/dri/r300/compiler/radeon_program.c +++ b/src/mesa/drivers/dri/r300/compiler/radeon_program.c @@ -150,14 +150,37 @@ void rc_mesa_to_rc_program(struct radeon_compiler * c, struct gl_program * progr c->Program.InputsRead = program->InputsRead; c->Program.OutputsWritten = program->OutputsWritten; - for(i = 0; i < program->Parameters->NumParameters; ++i) { - struct rc_constant constant; + int isNVProgram = 0; - constant.Type = RC_CONSTANT_EXTERNAL; - constant.Size = 4; - constant.u.External = i; + if (program->Target == GL_VERTEX_PROGRAM_ARB) { + struct gl_vertex_program * vp = (struct gl_vertex_program *) program; + isNVProgram = vp->IsNVProgram; + } + + if (isNVProgram) { + /* NV_vertex_program has a fixed-sized constant environment. + * This could be handled more efficiently for programs that + * do not use relative addressing. + */ + for(i = 0; i < 96; ++i) { + struct rc_constant constant; - rc_constants_add(&c->Program.Constants, &constant); + constant.Type = RC_CONSTANT_EXTERNAL; + constant.Size = 4; + constant.u.External = i; + + rc_constants_add(&c->Program.Constants, &constant); + } + } else { + for(i = 0; i < program->Parameters->NumParameters; ++i) { + struct rc_constant constant; + + constant.Type = RC_CONSTANT_EXTERNAL; + constant.Size = 4; + constant.u.External = i; + + rc_constants_add(&c->Program.Constants, &constant); + } } } diff --git a/src/mesa/drivers/dri/r300/compiler/radeon_program_alu.c b/src/mesa/drivers/dri/r300/compiler/radeon_program_alu.c index 8071899eaa4..f23ce301cab 100644 --- a/src/mesa/drivers/dri/r300/compiler/radeon_program_alu.c +++ b/src/mesa/drivers/dri/r300/compiler/radeon_program_alu.c @@ -511,23 +511,23 @@ static void sincos_constants(struct radeon_compiler* c, GLuint *constants) * MAD dest, tmp.y, weight, tmp.x */ static void sin_approx( - struct radeon_compiler* c, struct rc_instruction * after, + struct radeon_compiler* c, struct rc_instruction * before, struct prog_dst_register dst, struct prog_src_register src, const GLuint* constants) { GLuint tempreg = rc_find_free_temporary(c); - emit2(c, after->Prev, OPCODE_MUL, 0, dstregtmpmask(tempreg, WRITEMASK_XY), + emit2(c, before->Prev, OPCODE_MUL, 0, dstregtmpmask(tempreg, WRITEMASK_XY), swizzle(src, SWIZZLE_X, SWIZZLE_X, SWIZZLE_X, SWIZZLE_X), srcreg(PROGRAM_CONSTANT, constants[0])); - emit3(c, after->Prev, OPCODE_MAD, 0, dstregtmpmask(tempreg, WRITEMASK_X), + emit3(c, before->Prev, OPCODE_MAD, 0, dstregtmpmask(tempreg, WRITEMASK_X), swizzle(srcreg(PROGRAM_TEMPORARY, tempreg), SWIZZLE_Y, SWIZZLE_Y, SWIZZLE_Y, SWIZZLE_Y), absolute(swizzle(src, SWIZZLE_X, SWIZZLE_X, SWIZZLE_X, SWIZZLE_X)), swizzle(srcreg(PROGRAM_TEMPORARY, tempreg), SWIZZLE_X, SWIZZLE_X, SWIZZLE_X, SWIZZLE_X)); - emit3(c, after->Prev, OPCODE_MAD, 0, dstregtmpmask(tempreg, WRITEMASK_Y), + emit3(c, before->Prev, OPCODE_MAD, 0, dstregtmpmask(tempreg, WRITEMASK_Y), swizzle(srcreg(PROGRAM_TEMPORARY, tempreg), SWIZZLE_X, SWIZZLE_X, SWIZZLE_X, SWIZZLE_X), absolute(swizzle(srcreg(PROGRAM_TEMPORARY, tempreg), SWIZZLE_X, SWIZZLE_X, SWIZZLE_X, SWIZZLE_X)), negate(swizzle(srcreg(PROGRAM_TEMPORARY, tempreg), SWIZZLE_X, SWIZZLE_X, SWIZZLE_X, SWIZZLE_X))); - emit3(c, after->Prev, OPCODE_MAD, 0, dst, + emit3(c, before->Prev, OPCODE_MAD, 0, dst, swizzle(srcreg(PROGRAM_TEMPORARY, tempreg), SWIZZLE_Y, SWIZZLE_Y, SWIZZLE_Y, SWIZZLE_Y), swizzle(srcreg(PROGRAM_CONSTANT, constants[0]), SWIZZLE_W, SWIZZLE_W, SWIZZLE_W, SWIZZLE_W), swizzle(srcreg(PROGRAM_TEMPORARY, tempreg), SWIZZLE_X, SWIZZLE_X, SWIZZLE_X, SWIZZLE_X)); @@ -567,7 +567,7 @@ GLboolean radeonTransformTrigSimple(struct radeon_compiler* c, swizzle(srcreg(PROGRAM_CONSTANT, constants[1]), SWIZZLE_W, SWIZZLE_W, SWIZZLE_W, SWIZZLE_W), negate(swizzle(srcreg(PROGRAM_CONSTANT, constants[0]), SWIZZLE_Z, SWIZZLE_Z, SWIZZLE_Z, SWIZZLE_Z))); - sin_approx(c, inst->Prev, inst->I.DstReg, + sin_approx(c, inst, inst->I.DstReg, swizzle(srcreg(PROGRAM_TEMPORARY, tempreg), SWIZZLE_W, SWIZZLE_W, SWIZZLE_W, SWIZZLE_W), constants); } else if (inst->I.Opcode == OPCODE_SIN) { @@ -582,7 +582,7 @@ GLboolean radeonTransformTrigSimple(struct radeon_compiler* c, swizzle(srcreg(PROGRAM_CONSTANT, constants[1]), SWIZZLE_W, SWIZZLE_W, SWIZZLE_W, SWIZZLE_W), negate(swizzle(srcreg(PROGRAM_CONSTANT, constants[0]), SWIZZLE_Z, SWIZZLE_Z, SWIZZLE_Z, SWIZZLE_Z))); - sin_approx(c, inst->Prev, inst->I.DstReg, + sin_approx(c, inst, inst->I.DstReg, swizzle(srcreg(PROGRAM_TEMPORARY, tempreg), SWIZZLE_W, SWIZZLE_W, SWIZZLE_W, SWIZZLE_W), constants); } else { @@ -600,12 +600,12 @@ GLboolean radeonTransformTrigSimple(struct radeon_compiler* c, struct prog_dst_register dst = inst->I.DstReg; dst.WriteMask = inst->I.DstReg.WriteMask & WRITEMASK_X; - sin_approx(c, inst->Prev, dst, + sin_approx(c, inst, dst, swizzle(srcreg(PROGRAM_TEMPORARY, tempreg), SWIZZLE_X, SWIZZLE_X, SWIZZLE_X, SWIZZLE_X), constants); dst.WriteMask = inst->I.DstReg.WriteMask & WRITEMASK_Y; - sin_approx(c, inst->Prev, dst, + sin_approx(c, inst, dst, swizzle(srcreg(PROGRAM_TEMPORARY, tempreg), SWIZZLE_Y, SWIZZLE_Y, SWIZZLE_Y, SWIZZLE_Y), constants); } diff --git a/src/mesa/drivers/dri/r300/r300_fragprog_common.c b/src/mesa/drivers/dri/r300/r300_fragprog_common.c index 469c278b510..0bdc90b4a84 100644 --- a/src/mesa/drivers/dri/r300/r300_fragprog_common.c +++ b/src/mesa/drivers/dri/r300/r300_fragprog_common.c @@ -99,8 +99,8 @@ static void insert_WPOS_trailer(struct r300_fragment_program_compiler *compiler, { int i; + fp->wpos_attr = FRAG_ATTRIB_MAX; if (!(compiler->Base.Program.InputsRead & FRAG_BIT_WPOS)) { - fp->wpos_attr = FRAG_ATTRIB_MAX; return; } @@ -112,6 +112,13 @@ static void insert_WPOS_trailer(struct r300_fragment_program_compiler *compiler, } } + /* No free texcoord found, fall-back to software rendering */ + if (fp->wpos_attr == FRAG_ATTRIB_MAX) + { + compiler->Base.Error = 1; + return; + } + rc_transform_fragment_wpos(&compiler->Base, FRAG_ATTRIB_WPOS, fp->wpos_attr); } @@ -127,8 +134,8 @@ static void rewriteFog(struct r300_fragment_program_compiler *compiler, struct r struct prog_src_register src; int i; + fp->fog_attr = FRAG_ATTRIB_MAX; if (!(compiler->Base.Program.InputsRead & FRAG_BIT_FOGC)) { - fp->fog_attr = FRAG_ATTRIB_MAX; return; } @@ -140,6 +147,13 @@ static void rewriteFog(struct r300_fragment_program_compiler *compiler, struct r } } + /* No free texcoord found, fall-back to software rendering */ + if (fp->fog_attr == FRAG_ATTRIB_MAX) + { + compiler->Base.Error = 1; + return; + } + memset(&src, 0, sizeof(src)); src.File = PROGRAM_INPUT; src.Index = fp->fog_attr; diff --git a/src/mesa/drivers/dri/r300/r300_render.c b/src/mesa/drivers/dri/r300/r300_render.c index b5ddfdc9f82..3cd38753b8a 100644 --- a/src/mesa/drivers/dri/r300/r300_render.c +++ b/src/mesa/drivers/dri/r300/r300_render.c @@ -475,7 +475,7 @@ void r300SwitchFallback(GLcontext *ctx, uint32_t bit, GLboolean mode) /* update only if we have disabled all tcl fallbacks */ if (rmesa->options.hw_tcl_enabled) { - if ((old_fallback & R300_RASTER_FALLBACK_MASK) == bit) { + if ((old_fallback & R300_TCL_FALLBACK_MASK) == bit) { R300_STATECHANGE(rmesa, vap_cntl_status); rmesa->hw.vap_cntl_status.cmd[1] &= ~R300_VAP_TCL_BYPASS; } diff --git a/src/mesa/drivers/dri/r300/r300_vertprog.c b/src/mesa/drivers/dri/r300/r300_vertprog.c index dd0f27f9cba..2f7b67c1431 100644 --- a/src/mesa/drivers/dri/r300/r300_vertprog.c +++ b/src/mesa/drivers/dri/r300/r300_vertprog.c @@ -43,6 +43,7 @@ USE OR OTHER DEALINGS IN THE SOFTWARE. #include "compiler/radeon_compiler.h" #include "compiler/radeon_nqssadce.h" #include "r300_context.h" +#include "r300_fragprog_common.h" #include "r300_state.h" /** @@ -203,6 +204,34 @@ static void t_inputs_outputs(struct r300_vertex_program_compiler * c) } } +/** + * The NV_vertex_program spec mandates that all registers be + * initialized to zero. We do this here unconditionally. + * + * \note We rely on dead-code elimination in the compiler. + */ +static void initialize_NV_registers(struct radeon_compiler * compiler) +{ + unsigned int reg; + struct rc_instruction * inst; + + for(reg = 0; reg < 12; ++reg) { + inst = rc_insert_new_instruction(compiler, &compiler->Program.Instructions); + inst->I.Opcode = OPCODE_MOV; + inst->I.DstReg.File = PROGRAM_TEMPORARY; + inst->I.DstReg.Index = reg; + inst->I.SrcReg[0].File = PROGRAM_BUILTIN; + inst->I.SrcReg[0].Swizzle = SWIZZLE_0000; + } + + inst = rc_insert_new_instruction(compiler, &compiler->Program.Instructions); + inst->I.Opcode = OPCODE_ARL; + inst->I.DstReg.File = PROGRAM_ADDRESS; + inst->I.DstReg.Index = 0; + inst->I.DstReg.WriteMask = WRITEMASK_X; + inst->I.SrcReg[0].File = PROGRAM_BUILTIN; + inst->I.SrcReg[0].Swizzle = SWIZZLE_0000; +} static struct r300_vertex_program *build_program(GLcontext *ctx, struct r300_vertex_program_key *wanted_key, @@ -234,6 +263,9 @@ static struct r300_vertex_program *build_program(GLcontext *ctx, rc_mesa_to_rc_program(&compiler.Base, &vp->Base->Base); + if (mesa_vp->IsNVProgram) + initialize_NV_registers(&compiler.Base); + rc_move_output(&compiler.Base, VERT_RESULT_PSIZ, VERT_RESULT_PSIZ, WRITEMASK_X); if (vp->key.WPosAttr != FRAG_ATTRIB_MAX) { @@ -267,6 +299,20 @@ struct r300_vertex_program * r300SelectAndTranslateVertexShader(GLcontext *ctx) struct r300_vertex_program *vp; vpc = (struct r300_vertex_program_cont *)ctx->VertexProgram._Current; + + if (!r300->selected_fp) { + /* This can happen when GetProgramiv is called to check + * whether the program runs natively. + * + * To be honest, this is not a very good solution, + * but solving the problem of reporting good values + * for those queries is tough anyway considering that + * we recompile vertex programs based on the precise + * fragment program that is in use. + */ + r300SelectAndTranslateFragmentShader(ctx); + } + wanted_key.FpReads = r300->selected_fp->InputsRead; wanted_key.FogAttr = r300->selected_fp->fog_attr; wanted_key.WPosAttr = r300->selected_fp->wpos_attr; diff --git a/src/mesa/drivers/dri/r300/radeon_context.h b/src/mesa/drivers/dri/r300/radeon_context.h index 250570f6b89..da4812d3234 100644 --- a/src/mesa/drivers/dri/r300/radeon_context.h +++ b/src/mesa/drivers/dri/r300/radeon_context.h @@ -51,26 +51,12 @@ WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. #include "radeon_screen.h" -#if R200_MERGED -extern void radeonFallback(GLcontext * ctx, GLuint bit, GLboolean mode); - -#define FALLBACK( radeon, bit, mode ) do { \ - if ( 0 ) fprintf( stderr, "FALLBACK in %s: #%d=%d\n", \ - __FUNCTION__, bit, mode ); \ - radeonFallback( (radeon)->glCtx, bit, mode ); \ -} while (0) -#else #define FALLBACK( radeon, bit, mode ) fprintf(stderr, "%s:%s\n", __LINE__, __FILE__); -#endif /* TCL fallbacks */ extern void radeonTclFallback(GLcontext * ctx, GLuint bit, GLboolean mode); -#if R200_MERGED -#define TCL_FALLBACK( ctx, bit, mode ) radeonTclFallback( ctx, bit, mode ) -#else #define TCL_FALLBACK( ctx, bit, mode ) ; -#endif #endif /* __RADEON_CONTEXT_H__ */ diff --git a/src/mesa/drivers/dri/r600/Makefile b/src/mesa/drivers/dri/r600/Makefile index d925a2dfe39..7d5a7b1ab6f 100644 --- a/src/mesa/drivers/dri/r600/Makefile +++ b/src/mesa/drivers/dri/r600/Makefile @@ -29,6 +29,7 @@ COMMON_SOURCES = \ RADEON_COMMON_SOURCES = \ radeon_bo_legacy.c \ radeon_common_context.c \ + radeon_buffer_objects.c \ radeon_common.c \ radeon_cs_legacy.c \ radeon_dma.c \ @@ -65,8 +66,7 @@ DRIVER_SOURCES = \ C_SOURCES = $(COMMON_SOURCES) $(DRIVER_SOURCES) -DRIVER_DEFINES = -DCOMPILE_R600 -DR200_MERGED=0 \ - -DRADEON_COMMON=1 -DRADEON_COMMON_FOR_R600 \ +DRIVER_DEFINES = -DRADEON_R600 \ # -DRADEON_BO_TRACK \ -Wall diff --git a/src/mesa/drivers/dri/r600/r600_context.c b/src/mesa/drivers/dri/r600/r600_context.c index 07a7bcf11f1..969144ba123 100644 --- a/src/mesa/drivers/dri/r600/r600_context.c +++ b/src/mesa/drivers/dri/r600/r600_context.c @@ -59,6 +59,7 @@ WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. #include "radeon_debug.h" #include "r600_context.h" #include "radeon_common_context.h" +#include "radeon_buffer_objects.h" #include "radeon_span.h" #include "r600_cmdbuf.h" #include "r600_emit.h" @@ -257,6 +258,7 @@ GLboolean r600CreateContext(const __GLcontextModes * glVisual, r600InitTextureFuncs(&functions); r700InitShaderFuncs(&functions); r700InitIoctlFuncs(&functions); + radeonInitBufferObjectFuncs(&functions); if (!radeonInitContext(&r600->radeon, &functions, glVisual, driContextPriv, @@ -284,8 +286,8 @@ GLboolean r600CreateContext(const __GLcontextModes * glVisual, ctx->Const.MaxTextureMaxAnisotropy = 16.0; ctx->Const.MaxTextureLodBias = 16.0; - ctx->Const.MaxTextureLevels = 13; - ctx->Const.MaxTextureRectSize = 4096; + ctx->Const.MaxTextureLevels = 13; /* hw support 14 */ + ctx->Const.MaxTextureRectSize = 4096; /* hw support 8192 */ ctx->Const.MinPointSize = 0x0001 / 8.0; ctx->Const.MinPointSizeAA = 0x0001 / 8.0; @@ -330,26 +332,27 @@ GLboolean r600CreateContext(const __GLcontextModes * glVisual, _tnl_allow_pixel_fog(ctx, GL_FALSE); _tnl_allow_vertex_fog(ctx, GL_TRUE); - /* currently bogus data */ - ctx->Const.VertexProgram.MaxInstructions = VSF_MAX_FRAGMENT_LENGTH / 4; - ctx->Const.VertexProgram.MaxNativeInstructions = - VSF_MAX_FRAGMENT_LENGTH / 4; - ctx->Const.VertexProgram.MaxNativeAttribs = 16; /* r420 */ - ctx->Const.VertexProgram.MaxTemps = 32; - ctx->Const.VertexProgram.MaxNativeTemps = - /*VSF_MAX_FRAGMENT_TEMPS */ 32; - ctx->Const.VertexProgram.MaxNativeParameters = 256; /* r420 */ - ctx->Const.VertexProgram.MaxNativeAddressRegs = 1; - - ctx->Const.FragmentProgram.MaxNativeTemps = PFS_NUM_TEMP_REGS; - ctx->Const.FragmentProgram.MaxNativeAttribs = 11; /* copy i915... */ - ctx->Const.FragmentProgram.MaxNativeParameters = PFS_NUM_CONST_REGS; - ctx->Const.FragmentProgram.MaxNativeAluInstructions = PFS_MAX_ALU_INST; - ctx->Const.FragmentProgram.MaxNativeTexInstructions = PFS_MAX_TEX_INST; - ctx->Const.FragmentProgram.MaxNativeInstructions = - PFS_MAX_ALU_INST + PFS_MAX_TEX_INST; - ctx->Const.FragmentProgram.MaxNativeTexIndirections = - PFS_MAX_TEX_INDIRECT; + /* 256 for reg-based consts, inline consts also supported */ + ctx->Const.VertexProgram.MaxInstructions = 8192; /* in theory no limit */ + ctx->Const.VertexProgram.MaxNativeInstructions = 8192; + ctx->Const.VertexProgram.MaxNativeAttribs = 160; + ctx->Const.VertexProgram.MaxTemps = 128; + ctx->Const.VertexProgram.MaxNativeTemps = 128; + ctx->Const.VertexProgram.MaxNativeParameters = 256; + ctx->Const.VertexProgram.MaxNativeAddressRegs = 1; /* ??? */ + + ctx->Const.FragmentProgram.MaxNativeTemps = 128; + ctx->Const.FragmentProgram.MaxNativeAttribs = 32; + ctx->Const.FragmentProgram.MaxNativeParameters = 256; + ctx->Const.FragmentProgram.MaxNativeAluInstructions = 8192; + /* 8 per clause on r6xx, 16 on rv670/r7xx */ + if ((screen->chip_family == CHIP_FAMILY_RV670) || + (screen->chip_family >= CHIP_FAMILY_RV770)) + ctx->Const.FragmentProgram.MaxNativeTexInstructions = 16; + else + ctx->Const.FragmentProgram.MaxNativeTexInstructions = 8; + ctx->Const.FragmentProgram.MaxNativeInstructions = 8192; + ctx->Const.FragmentProgram.MaxNativeTexIndirections = 8; /* ??? */ ctx->Const.FragmentProgram.MaxNativeAddressRegs = 0; /* and these are?? */ ctx->VertexProgram._MaintainTnlProgram = GL_TRUE; ctx->FragmentProgram._MaintainTexEnvProgram = GL_TRUE; @@ -374,6 +377,8 @@ GLboolean r600CreateContext(const __GLcontextModes * glVisual, _mesa_enable_extension(ctx, "GL_EXT_texture_compression_s3tc"); } + r700InitDraw(ctx); + radeon_fbo_init(&r600->radeon); radeonInitSpanFuncs( ctx ); @@ -385,9 +390,6 @@ GLboolean r600CreateContext(const __GLcontextModes * glVisual, if (driQueryOptionb(&r600->radeon.optionCache, "no_rast")) { radeon_warning("disabling 3D acceleration\n"); -#if R200_MERGED - FALLBACK(&r600->radeon, RADEON_FALLBACK_DISABLE, 1); -#endif } return GL_TRUE; diff --git a/src/mesa/drivers/dri/r600/r600_context.h b/src/mesa/drivers/dri/r600/r600_context.h index 8ae05a301c7..a296ea23fa4 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; @@ -85,29 +86,10 @@ extern int hw_tcl_on; #include "tnl_dd/t_dd_vertex.h" #undef TAG -#define PFS_MAX_ALU_INST 64 -#define PFS_MAX_TEX_INST 64 -#define PFS_MAX_TEX_INDIRECT 4 -#define PFS_NUM_TEMP_REGS 32 -#define PFS_NUM_CONST_REGS 16 - -#define R600_MAX_AOS_ARRAYS 16 - -#define REG_COORDS 0 -#define REG_COLOR0 1 -#define REG_TEX0 2 - #define R600_FALLBACK_NONE 0 #define R600_FALLBACK_TCL 1 #define R600_FALLBACK_RAST 2 -enum -{ - NO_SHIFT = 0, - LEFT_SHIFT = 1, - RIGHT_SHIFT = 2, -}; - struct r600_hw_state { struct radeon_state_atom sq; struct radeon_state_atom db; @@ -144,6 +126,34 @@ struct r600_hw_state { struct radeon_state_atom tx_brdr_clr; }; +typedef struct StreamDesc +{ + GLint size; //number of data element + GLenum type; //data element type + GLsizei stride; + + struct radeon_bo *bo; + GLint bo_offset; + + GLuint dwords; + GLuint dst_loc; + GLuint _signed; + GLboolean normalize; + GLboolean is_named_bo; + GLubyte element; +} StreamDesc; + +typedef struct r700_index_buffer +{ + struct radeon_bo *bo; + int bo_offset; + + GLboolean is_32bit; + GLuint count; + + GLboolean bHostIb; +} r700_index_buffer; + /** * \brief R600 context structure. */ @@ -155,11 +165,16 @@ struct r600_context { struct r600_hw_state atoms; + struct r700_vertex_program *selected_vp; + /* Vertex buffers */ GLvector4f dummy_attrib[_TNL_ATTRIB_MAX]; GLvector4f *temp_attrib[_TNL_ATTRIB_MAX]; + GLint nNumActiveAos; + StreamDesc stream_desc[VERT_ATTRIB_MAX]; + struct r700_index_buffer ind_buf; }; #define R700_CONTEXT(ctx) ((context_t *)(ctx->DriverCtx)) @@ -193,6 +208,7 @@ extern GLboolean r700SyncSurf(context_t *context, extern void r700SetupStreams(GLcontext * ctx); extern void r700Start3D(context_t *context); extern void r600InitAtoms(context_t *context); +extern void r700InitDraw(GLcontext *ctx); #define RADEON_D_CAPTURE 0 #define RADEON_D_PLAYBACK 1 diff --git a/src/mesa/drivers/dri/r600/r600_reg_auto_r6xx.h b/src/mesa/drivers/dri/r600/r600_reg_auto_r6xx.h index 9d5aa3c7e49..edd85b0facc 100644 --- a/src/mesa/drivers/dri/r600/r600_reg_auto_r6xx.h +++ b/src/mesa/drivers/dri/r600/r600_reg_auto_r6xx.h @@ -1366,6 +1366,7 @@ enum { DB_DEPTH_INFO__READ_SIZE_bit = 1 << 3, DB_DEPTH_INFO__ARRAY_MODE_mask = 0x0f << 15, DB_DEPTH_INFO__ARRAY_MODE_shift = 15, + ARRAY_1D_TILED_THIN1 = 0x02, ARRAY_2D_TILED_THIN1 = 0x04, TILE_SURFACE_ENABLE_bit = 1 << 25, TILE_COMPACT_bit = 1 << 26, @@ -1449,6 +1450,7 @@ enum { CB_COLOR0_INFO__ARRAY_MODE_shift = 8, ARRAY_LINEAR_GENERAL = 0x00, ARRAY_LINEAR_ALIGNED = 0x01, +/* ARRAY_1D_TILED_THIN1 = 0x02, */ /* ARRAY_2D_TILED_THIN1 = 0x04, */ NUMBER_TYPE_mask = 0x07 << 12, NUMBER_TYPE_shift = 12, diff --git a/src/mesa/drivers/dri/r600/r700_assembler.c b/src/mesa/drivers/dri/r600/r700_assembler.c index 2d8480daaf7..903b6968be1 100644 --- a/src/mesa/drivers/dri/r600/r700_assembler.c +++ b/src/mesa/drivers/dri/r600/r700_assembler.c @@ -213,7 +213,7 @@ GLboolean is_reduction_opcode(PVSDWORD* dest) { if (dest->dst.op3 == 0) { - if ( (dest->dst.opcode == SQ_OP2_INST_DOT4 || dest->dst.opcode == SQ_OP2_INST_DOT4_IEEE) ) + if ( (dest->dst.opcode == SQ_OP2_INST_DOT4 || dest->dst.opcode == SQ_OP2_INST_DOT4_IEEE || dest->dst.opcode == SQ_OP2_INST_CUBE) ) { return GL_TRUE; } @@ -350,6 +350,7 @@ unsigned int r700GetNumOperands(r700_AssemblerBase* pAsm) case SQ_OP2_INST_PRED_SETNE: case SQ_OP2_INST_DOT4: case SQ_OP2_INST_DOT4_IEEE: + case SQ_OP2_INST_CUBE: return 2; case SQ_OP2_INST_MOV: @@ -469,6 +470,9 @@ int Init_r700_AssemblerBase(SHADER_PIPE_TYPE spt, r700_AssemblerBase* pAsm, R700 pAsm->number_of_inputs = 0; + pAsm->is_tex = GL_FALSE; + pAsm->need_tex_barrier = GL_FALSE; + return 0; } @@ -682,7 +686,7 @@ GLboolean add_tex_instruction(r700_AssemblerBase* pAsm, // If this clause constains any TEX instruction that is dependent on a previous instruction, // set the barrier bit - if( pAsm->pInstDeps[pAsm->uiCurInst].nDstDep > (-1) ) + if( pAsm->pInstDeps[pAsm->uiCurInst].nDstDep > (-1) || pAsm->need_tex_barrier == GL_TRUE ) { pAsm->cf_current_tex_clause_ptr->m_Word1.f.barrier = 0x1; } @@ -786,6 +790,133 @@ GLboolean assemble_vfetch_instruction(r700_AssemblerBase* pAsm, return GL_TRUE; } +GLboolean assemble_vfetch_instruction2(r700_AssemblerBase* pAsm, + GLuint destination_register, + GLenum type, + GLint size, + GLubyte element, + GLuint _signed, + GLboolean normalize, + VTX_FETCH_METHOD * pFetchMethod) +{ + GLuint client_size_inbyte; + GLuint data_format; + GLuint mega_fetch_count; + GLuint is_mega_fetch_flag; + + R700VertexGenericFetch* vfetch_instruction_ptr; + R700VertexGenericFetch* assembled_vfetch_instruction_ptr + = pAsm->vfetch_instruction_ptr_array[element]; + + if (assembled_vfetch_instruction_ptr == NULL) + { + vfetch_instruction_ptr = (R700VertexGenericFetch*) CALLOC_STRUCT(R700VertexGenericFetch); + if (vfetch_instruction_ptr == NULL) + { + return GL_FALSE; + } + Init_R700VertexGenericFetch(vfetch_instruction_ptr); + } + else + { + vfetch_instruction_ptr = assembled_vfetch_instruction_ptr; + } + + data_format = GetSurfaceFormat(type, size, &client_size_inbyte); + + if(GL_TRUE == pFetchMethod->bEnableMini) //More conditions here + { + //TODO : mini fetch + } + else + { + mega_fetch_count = MEGA_FETCH_BYTES - 1; + is_mega_fetch_flag = 0x1; + pFetchMethod->mega_fetch_remainder = MEGA_FETCH_BYTES - client_size_inbyte; + } + + vfetch_instruction_ptr->m_Word0.f.vtx_inst = SQ_VTX_INST_FETCH; + vfetch_instruction_ptr->m_Word0.f.fetch_type = SQ_VTX_FETCH_VERTEX_DATA; + vfetch_instruction_ptr->m_Word0.f.fetch_whole_quad = 0x0; + + vfetch_instruction_ptr->m_Word0.f.buffer_id = element; + vfetch_instruction_ptr->m_Word0.f.src_gpr = 0x0; + vfetch_instruction_ptr->m_Word0.f.src_rel = SQ_ABSOLUTE; + vfetch_instruction_ptr->m_Word0.f.src_sel_x = SQ_SEL_X; + vfetch_instruction_ptr->m_Word0.f.mega_fetch_count = mega_fetch_count; + + vfetch_instruction_ptr->m_Word1.f.dst_sel_x = (size < 1) ? SQ_SEL_0 : SQ_SEL_X; + vfetch_instruction_ptr->m_Word1.f.dst_sel_y = (size < 2) ? SQ_SEL_0 : SQ_SEL_Y; + vfetch_instruction_ptr->m_Word1.f.dst_sel_z = (size < 3) ? SQ_SEL_0 : SQ_SEL_Z; + vfetch_instruction_ptr->m_Word1.f.dst_sel_w = (size < 4) ? SQ_SEL_1 : SQ_SEL_W; + + vfetch_instruction_ptr->m_Word1.f.use_const_fields = 1; + vfetch_instruction_ptr->m_Word1.f.data_format = data_format; + vfetch_instruction_ptr->m_Word2.f.endian_swap = SQ_ENDIAN_NONE; + + if(1 == _signed) + { + vfetch_instruction_ptr->m_Word1.f.format_comp_all = SQ_FORMAT_COMP_SIGNED; + } + else + { + vfetch_instruction_ptr->m_Word1.f.format_comp_all = SQ_FORMAT_COMP_UNSIGNED; + } + + if(GL_TRUE == normalize) + { + vfetch_instruction_ptr->m_Word1.f.num_format_all = SQ_NUM_FORMAT_NORM; + } + else + { + vfetch_instruction_ptr->m_Word1.f.num_format_all = SQ_NUM_FORMAT_INT; + } + + // Destination register + vfetch_instruction_ptr->m_Word1_GPR.f.dst_gpr = destination_register; + vfetch_instruction_ptr->m_Word1_GPR.f.dst_rel = SQ_ABSOLUTE; + + vfetch_instruction_ptr->m_Word2.f.offset = 0; + vfetch_instruction_ptr->m_Word2.f.const_buf_no_stride = 0x0; + + vfetch_instruction_ptr->m_Word2.f.mega_fetch = is_mega_fetch_flag; + + if (assembled_vfetch_instruction_ptr == NULL) + { + if ( GL_FALSE == add_vfetch_instruction(pAsm, (R700VertexInstruction *)vfetch_instruction_ptr) ) + { + return GL_FALSE; + } + + if (pAsm->vfetch_instruction_ptr_array[element] != NULL) + { + return GL_FALSE; + } + else + { + pAsm->vfetch_instruction_ptr_array[element] = vfetch_instruction_ptr; + } + } + + return GL_TRUE; +} + +GLboolean cleanup_vfetch_instructions(r700_AssemblerBase* pAsm) +{ + GLint i; + pAsm->cf_current_clause_type = CF_EMPTY_CLAUSE; + pAsm->cf_current_vtx_clause_ptr = NULL; + + for (i=0; i<VERT_ATTRIB_MAX; i++) + { + pAsm->vfetch_instruction_ptr_array[ i ] = NULL; + } + + cleanup_vfetch_shaderinst(pAsm->pR700Shader); + + return GL_TRUE; +} + GLuint gethelpr(r700_AssemblerBase* pAsm) { GLuint r = pAsm->uHelpReg; @@ -1149,41 +1280,55 @@ 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) + if(pAsm->aArgSubst[1] >= 0) { - case PROGRAM_TEMPORARY: bValidTexCoord = GL_TRUE; - - pAsm->S[0].src.reg = pILInst->SrcReg[0].Index + pAsm->starting_temp_register_number; + setaddrmode_PVSSRC(&(pAsm->S[0].src), ADDR_ABSOLUTE); 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: + pAsm->S[0].src.reg = pAsm->aArgSubst[1]; + } + else + { + switch (pILInst->SrcReg[0].File) { + case PROGRAM_CONSTANT: + case PROGRAM_LOCAL_PARAM: + case PROGRAM_ENV_PARAM: + case PROGRAM_STATE_VAR: + break; + case PROGRAM_TEMPORARY: bValidTexCoord = GL_TRUE; - - pAsm->S[0].src.reg = pAsm->uiFP_AttributeMap[pILInst->SrcReg[0].Index]; - pAsm->S[0].src.rtype = SRC_REG_INPUT; - } + 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; + } break; + } } if(GL_TRUE == bValidTexCoord) - { + { setaddrmode_PVSSRC(&(pAsm->S[0].src), ADDR_ABSOLUTE); } else @@ -1201,7 +1346,7 @@ 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; } @@ -1933,9 +2078,9 @@ GLboolean assemble_alu_instruction(r700_AssemblerBase *pAsm) GLuint contiguous_slots_needed; GLuint uNumSrc = r700GetNumOperands(pAsm); - GLuint channel_swizzle, j; - GLuint chan_counter[4] = {0, 0, 0, 0}; - PVSSRC * pSource[3]; + //GLuint channel_swizzle, j; + //GLuint chan_counter[4] = {0, 0, 0, 0}; + //PVSSRC * pSource[3]; GLboolean bSplitInst = GL_FALSE; if (1 == pAsm->D.dst.math) @@ -1947,7 +2092,9 @@ GLboolean assemble_alu_instruction(r700_AssemblerBase *pAsm) { is_single_scalar_operation = GL_FALSE; number_of_scalar_operations = 4; - + +/* current assembler doesn't do more than 1 register per source */ +#if 0 /* check read port, only very preliminary algorithm, not count in src0/1 same comp case and prev slot repeat case; also not count relative addressing. TODO: improve performance. */ @@ -1982,6 +2129,7 @@ GLboolean assemble_alu_instruction(r700_AssemblerBase *pAsm) { bSplitInst = GL_TRUE; } +#endif } contiguous_slots_needed = 0; @@ -2016,7 +2164,7 @@ GLboolean assemble_alu_instruction(r700_AssemblerBase *pAsm) return GL_FALSE; } - if (pAsm->D.dst.math == 0) + if (uNumSrc > 1) { // Process source 1 current_source_index = 1; @@ -2202,7 +2350,7 @@ GLboolean next_ins(r700_AssemblerBase *pAsm) { struct prog_instruction *pILInst = &(pAsm->pILInst[pAsm->uiCurInst]); - if( GL_TRUE == IsTex(pILInst->Opcode) ) + if( GL_TRUE == pAsm->is_tex ) { if (pILInst->TexSrcTarget == TEXTURE_RECT_INDEX) { if( GL_FALSE == assemble_tex_instruction(pAsm, GL_FALSE) ) @@ -2246,7 +2394,8 @@ GLboolean next_ins(r700_AssemblerBase *pAsm) pAsm->S[0].bits = 0; pAsm->S[1].bits = 0; pAsm->S[2].bits = 0; - + pAsm->is_tex = GL_FALSE; + pAsm->need_tex_barrier = GL_FALSE; return GL_TRUE; } @@ -2870,6 +3019,11 @@ GLboolean assemble_LIT(r700_AssemblerBase *pAsm) return GL_FALSE; } + if( GL_FALSE == assemble_src(pAsm, 0, -1) ) + { + return GL_FALSE; + } + /* dst.y = max(src.x, 0.0) */ pAsm->D.dst.opcode = SQ_OP2_INST_MAX; pAsm->D.dst.rtype = dstType; @@ -2881,11 +3035,6 @@ GLboolean assemble_LIT(r700_AssemblerBase *pAsm) pAsm->S[0].src.rtype = srcType; pAsm->S[0].src.reg = srcReg; setaddrmode_PVSSRC(&(pAsm->S[0].src), ADDR_ABSOLUTE); - noneg_PVSSRC(&(pAsm->S[0].src)); - pAsm->S[0].src.swizzlex = SQ_SEL_X; - pAsm->S[0].src.swizzley = SQ_SEL_X; - pAsm->S[0].src.swizzlez = SQ_SEL_X; - pAsm->S[0].src.swizzlew = SQ_SEL_X; pAsm->S[1].src.rtype = SRC_REG_TEMPORARY; pAsm->S[1].src.reg = tmp; setaddrmode_PVSSRC(&(pAsm->S[1].src), ADDR_ABSOLUTE); @@ -2899,34 +3048,47 @@ GLboolean assemble_LIT(r700_AssemblerBase *pAsm) return GL_FALSE; } - /* before: dst.w = log(src.y) - * after : dst.x = log(src.y) - * why change dest register is that dst.w has been initialized as 1 before - */ + if( GL_FALSE == assemble_src(pAsm, 0, -1) ) + { + return GL_FALSE; + } + + swizzleagain_PVSSRC(&(pAsm->S[0].src), SQ_SEL_Y, SQ_SEL_Y, SQ_SEL_Y, SQ_SEL_Y); + + /* dst.z = log(src.y) */ pAsm->D.dst.opcode = SQ_OP2_INST_LOG_CLAMPED; pAsm->D.dst.math = 1; pAsm->D.dst.rtype = dstType; pAsm->D.dst.reg = dstReg; - pAsm->D.dst.writex = 1; + pAsm->D.dst.writex = 0; pAsm->D.dst.writey = 0; - pAsm->D.dst.writez = 0; + pAsm->D.dst.writez = 1; pAsm->D.dst.writew = 0; pAsm->S[0].src.rtype = srcType; pAsm->S[0].src.reg = srcReg; setaddrmode_PVSSRC(&(pAsm->S[0].src), ADDR_ABSOLUTE); - noneg_PVSSRC(&(pAsm->S[0].src)); - pAsm->S[0].src.swizzlex = SQ_SEL_Y; - pAsm->S[0].src.swizzley = SQ_SEL_Y; - pAsm->S[0].src.swizzlez = SQ_SEL_Y; - pAsm->S[0].src.swizzlew = SQ_SEL_Y; if( GL_FALSE == next_ins(pAsm) ) { return GL_FALSE; } - /* before: tmp.x = amd MUL_LIT(src.w, dst.w, src.x ) */ - /* after : tmp.x = amd MUL_LIT(src.w, dst.x, src.x ) */ + if( GL_FALSE == assemble_src(pAsm, 0, -1) ) + { + return GL_FALSE; + } + + if( GL_FALSE == assemble_src(pAsm, 0, 2) ) + { + return GL_FALSE; + } + + swizzleagain_PVSSRC(&(pAsm->S[0].src), SQ_SEL_W, SQ_SEL_W, SQ_SEL_W, SQ_SEL_W); + + swizzleagain_PVSSRC(&(pAsm->S[2].src), SQ_SEL_X, SQ_SEL_X, SQ_SEL_X, SQ_SEL_X); + + /* tmp.x = amd MUL_LIT(src.w, dst.z, src.x ) */ pAsm->D.dst.opcode = SQ_OP3_INST_MUL_LIT; + pAsm->D.dst.math = 1; pAsm->D.dst.op3 = 1; pAsm->D.dst.rtype = DST_REG_TEMPORARY; pAsm->D.dst.reg = tmp; @@ -2938,29 +3100,19 @@ GLboolean assemble_LIT(r700_AssemblerBase *pAsm) pAsm->S[0].src.rtype = srcType; pAsm->S[0].src.reg = srcReg; setaddrmode_PVSSRC(&(pAsm->S[0].src), ADDR_ABSOLUTE); - noneg_PVSSRC(&(pAsm->S[0].src)); - pAsm->S[0].src.swizzlex = SQ_SEL_W; - pAsm->S[0].src.swizzley = SQ_SEL_W; - pAsm->S[0].src.swizzlez = SQ_SEL_W; - pAsm->S[0].src.swizzlew = SQ_SEL_W; pAsm->S[1].src.rtype = SRC_REG_TEMPORARY; pAsm->S[1].src.reg = dstReg; setaddrmode_PVSSRC(&(pAsm->S[1].src), ADDR_ABSOLUTE); noneg_PVSSRC(&(pAsm->S[1].src)); - pAsm->S[1].src.swizzlex = SQ_SEL_X; - pAsm->S[1].src.swizzley = SQ_SEL_X; - pAsm->S[1].src.swizzlez = SQ_SEL_X; - pAsm->S[1].src.swizzlew = SQ_SEL_X; + pAsm->S[1].src.swizzlex = SQ_SEL_Z; + pAsm->S[1].src.swizzley = SQ_SEL_Z; + pAsm->S[1].src.swizzlez = SQ_SEL_Z; + pAsm->S[1].src.swizzlew = SQ_SEL_Z; pAsm->S[2].src.rtype = srcType; pAsm->S[2].src.reg = srcReg; setaddrmode_PVSSRC(&(pAsm->S[2].src), ADDR_ABSOLUTE); - noneg_PVSSRC(&(pAsm->S[2].src)); - pAsm->S[2].src.swizzlex = SQ_SEL_X; - pAsm->S[2].src.swizzley = SQ_SEL_X; - pAsm->S[2].src.swizzlez = SQ_SEL_X; - pAsm->S[2].src.swizzlew = SQ_SEL_X; if( GL_FALSE == next_ins(pAsm) ) { @@ -3366,7 +3518,10 @@ GLboolean assemble_STP(r700_AssemblerBase *pAsm) GLboolean assemble_TEX(r700_AssemblerBase *pAsm) { GLboolean src_const; + GLboolean need_barrier = GL_FALSE; + checkop1(pAsm); + switch (pAsm->pILInst[pAsm->uiCurInst].SrcReg[0].File) { case PROGRAM_CONSTANT: @@ -3374,29 +3529,30 @@ 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: + default: 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; + need_barrier = GL_TRUE; } - switch (pAsm->pILInst[pAsm->uiCurInst].Opcode) + switch (pAsm->pILInst[pAsm->uiCurInst].Opcode) { case OPCODE_TEX: - 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 */ - pAsm->D.dst.opcode = SQ_TEX_INST_SAMPLE; + case OPCODE_TXP: break; default: radeon_error("Internal error: bad texture op (not TEX)\n"); @@ -3404,6 +3560,190 @@ GLboolean assemble_TEX(r700_AssemblerBase *pAsm) break; } + if (pAsm->pILInst[pAsm->uiCurInst].Opcode == OPCODE_TXP) + { + GLuint tmp = gethelpr(pAsm); + pAsm->D.dst.opcode = SQ_OP2_INST_RECIP_IEEE; + pAsm->D.dst.math = 1; + setaddrmode_PVSDST(&(pAsm->D.dst), ADDR_ABSOLUTE); + pAsm->D.dst.rtype = DST_REG_TEMPORARY; + pAsm->D.dst.reg = tmp; + pAsm->D.dst.writew = 1; + + if( GL_FALSE == assemble_src(pAsm, 0, -1) ) + { + return GL_FALSE; + } + swizzleagain_PVSSRC(&(pAsm->S[0].src), SQ_SEL_W, SQ_SEL_W, SQ_SEL_W, SQ_SEL_W); + if( GL_FALSE == next_ins(pAsm) ) + { + return GL_FALSE; + } + + pAsm->D.dst.opcode = SQ_OP2_INST_MUL; + setaddrmode_PVSDST(&(pAsm->D.dst), ADDR_ABSOLUTE); + pAsm->D.dst.rtype = DST_REG_TEMPORARY; + pAsm->D.dst.reg = tmp; + pAsm->D.dst.writex = 1; + pAsm->D.dst.writey = 1; + pAsm->D.dst.writez = 1; + pAsm->D.dst.writew = 0; + + if( GL_FALSE == assemble_src(pAsm, 0, -1) ) + { + return GL_FALSE; + } + setaddrmode_PVSSRC(&(pAsm->S[1].src), ADDR_ABSOLUTE); + pAsm->S[1].src.rtype = SRC_REG_TEMPORARY; + pAsm->S[1].src.reg = tmp; + setswizzle_PVSSRC(&(pAsm->S[1].src), SQ_SEL_W); + + if( GL_FALSE == next_ins(pAsm) ) + { + return GL_FALSE; + } + + pAsm->aArgSubst[1] = tmp; + need_barrier = GL_TRUE; + } + + if (pAsm->pILInst[pAsm->uiCurInst].TexSrcTarget == TEXTURE_CUBE_INDEX ) + { + GLuint tmp1 = gethelpr(pAsm); + GLuint tmp2 = gethelpr(pAsm); + + /* tmp1.xyzw = CUBE(R0.zzxy, R0.yxzz) */ + pAsm->D.dst.opcode = SQ_OP2_INST_CUBE; + setaddrmode_PVSDST(&(pAsm->D.dst), ADDR_ABSOLUTE); + pAsm->D.dst.rtype = DST_REG_TEMPORARY; + pAsm->D.dst.reg = tmp1; + nomask_PVSDST(&(pAsm->D.dst)); + + if( GL_FALSE == assemble_src(pAsm, 0, -1) ) + { + return GL_FALSE; + } + + if( GL_FALSE == assemble_src(pAsm, 0, 1) ) + { + return GL_FALSE; + } + + swizzleagain_PVSSRC(&(pAsm->S[0].src), SQ_SEL_Z, SQ_SEL_Z, SQ_SEL_X, SQ_SEL_Y); + swizzleagain_PVSSRC(&(pAsm->S[1].src), SQ_SEL_Y, SQ_SEL_X, SQ_SEL_Z, SQ_SEL_Z); + + if( GL_FALSE == next_ins(pAsm) ) + { + return GL_FALSE; + } + + /* tmp1.z = ABS(tmp1.z) dont have abs support in assembler currently + * have to do explicit instruction + */ + pAsm->D.dst.opcode = SQ_OP2_INST_MAX; + setaddrmode_PVSDST(&(pAsm->D.dst), ADDR_ABSOLUTE); + pAsm->D.dst.rtype = DST_REG_TEMPORARY; + pAsm->D.dst.reg = tmp1; + pAsm->D.dst.writez = 1; + + setaddrmode_PVSSRC(&(pAsm->S[0].src), ADDR_ABSOLUTE); + pAsm->S[0].src.rtype = SRC_REG_TEMPORARY; + pAsm->S[0].src.reg = tmp1; + noswizzle_PVSSRC(&(pAsm->S[0].src)); + pAsm->S[1].bits = pAsm->S[0].bits; + flipneg_PVSSRC(&(pAsm->S[1].src)); + + next_ins(pAsm); + + /* tmp1.z = RCP_e(|tmp1.z|) */ + pAsm->D.dst.opcode = SQ_OP2_INST_RECIP_IEEE; + pAsm->D.dst.math = 1; + setaddrmode_PVSDST(&(pAsm->D.dst), ADDR_ABSOLUTE); + pAsm->D.dst.rtype = DST_REG_TEMPORARY; + pAsm->D.dst.reg = tmp1; + pAsm->D.dst.writez = 1; + + setaddrmode_PVSSRC(&(pAsm->S[0].src), ADDR_ABSOLUTE); + pAsm->S[0].src.rtype = SRC_REG_TEMPORARY; + pAsm->S[0].src.reg = tmp1; + pAsm->S[0].src.swizzlex = SQ_SEL_Z; + + next_ins(pAsm); + + /* MULADD R0.x, R0.x, PS1, (0x3FC00000, 1.5f).x + * MULADD R0.y, R0.y, PS1, (0x3FC00000, 1.5f).x + * muladd has no writemask, have to use another temp + * also no support for imm constants, so add 1 here + */ + pAsm->D.dst.opcode = SQ_OP3_INST_MULADD; + pAsm->D.dst.op3 = 1; + setaddrmode_PVSDST(&(pAsm->D.dst), ADDR_ABSOLUTE); + pAsm->D.dst.rtype = DST_REG_TEMPORARY; + pAsm->D.dst.reg = tmp2; + + setaddrmode_PVSSRC(&(pAsm->S[0].src), ADDR_ABSOLUTE); + pAsm->S[0].src.rtype = SRC_REG_TEMPORARY; + pAsm->S[0].src.reg = tmp1; + noswizzle_PVSSRC(&(pAsm->S[0].src)); + setaddrmode_PVSSRC(&(pAsm->S[1].src), ADDR_ABSOLUTE); + pAsm->S[1].src.rtype = SRC_REG_TEMPORARY; + pAsm->S[1].src.reg = tmp1; + setswizzle_PVSSRC(&(pAsm->S[1].src), SQ_SEL_Z); + setaddrmode_PVSSRC(&(pAsm->S[2].src), ADDR_ABSOLUTE); + pAsm->S[2].src.rtype = SRC_REG_TEMPORARY; + pAsm->S[2].src.reg = tmp1; + setswizzle_PVSSRC(&(pAsm->S[2].src), SQ_SEL_1); + + next_ins(pAsm); + + /* ADD the remaining .5 */ + pAsm->D.dst.opcode = SQ_OP2_INST_ADD; + setaddrmode_PVSDST(&(pAsm->D.dst), ADDR_ABSOLUTE); + pAsm->D.dst.rtype = DST_REG_TEMPORARY; + pAsm->D.dst.reg = tmp2; + pAsm->D.dst.writex = 1; + pAsm->D.dst.writey = 1; + pAsm->D.dst.writez = 0; + pAsm->D.dst.writew = 0; + + setaddrmode_PVSSRC(&(pAsm->S[0].src), ADDR_ABSOLUTE); + pAsm->S[0].src.rtype = SRC_REG_TEMPORARY; + pAsm->S[0].src.reg = tmp2; + noswizzle_PVSSRC(&(pAsm->S[0].src)); + setaddrmode_PVSSRC(&(pAsm->S[1].src), ADDR_ABSOLUTE); + pAsm->S[1].src.rtype = SRC_REG_TEMPORARY; + pAsm->S[1].src.reg = 252; // SQ_ALU_SRC_0_5 + noswizzle_PVSSRC(&(pAsm->S[1].src)); + + next_ins(pAsm); + + /* tmp1.xy = temp2.xy */ + pAsm->D.dst.opcode = SQ_OP2_INST_MOV; + setaddrmode_PVSDST(&(pAsm->D.dst), ADDR_ABSOLUTE); + pAsm->D.dst.rtype = DST_REG_TEMPORARY; + pAsm->D.dst.reg = tmp1; + pAsm->D.dst.writex = 1; + pAsm->D.dst.writey = 1; + pAsm->D.dst.writez = 0; + pAsm->D.dst.writew = 0; + + setaddrmode_PVSSRC(&(pAsm->S[0].src), ADDR_ABSOLUTE); + pAsm->S[0].src.rtype = SRC_REG_TEMPORARY; + pAsm->S[0].src.reg = tmp2; + noswizzle_PVSSRC(&(pAsm->S[0].src)); + + next_ins(pAsm); + pAsm->aArgSubst[1] = tmp1; + need_barrier = GL_TRUE; + + } + + pAsm->D.dst.opcode = SQ_TEX_INST_SAMPLE; + pAsm->is_tex = GL_TRUE; + if ( GL_TRUE == need_barrier ) + { + pAsm->need_tex_barrier = GL_TRUE; + } // Set src1 to tex unit id pAsm->S[1].src.reg = pAsm->pILInst[pAsm->uiCurInst].TexSrcUnit; pAsm->S[1].src.rtype = SRC_REG_TEMPORARY; @@ -3418,16 +3758,31 @@ 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(pAsm->pILInst[pAsm->uiCurInst].Opcode == OPCODE_TXP) { - return GL_FALSE; + /* hopefully did swizzles before */ + noswizzle_PVSSRC(&(pAsm->S[0].src)); } + + if(pAsm->pILInst[pAsm->uiCurInst].TexSrcTarget == TEXTURE_CUBE_INDEX) + { + /* SAMPLE dst, tmp.yxwy, CUBE */ + pAsm->S[0].src.swizzlex = SQ_SEL_Y; + pAsm->S[0].src.swizzley = SQ_SEL_X; + pAsm->S[0].src.swizzlez = SQ_SEL_W; + pAsm->S[0].src.swizzlew = SQ_SEL_Y; + } + + if ( GL_FALSE == next_ins(pAsm) ) + { + return GL_FALSE; + } return GL_TRUE; } diff --git a/src/mesa/drivers/dri/r600/r700_assembler.h b/src/mesa/drivers/dri/r600/r700_assembler.h index f9c4d849c65..0d4283e4bad 100644 --- a/src/mesa/drivers/dri/r600/r700_assembler.h +++ b/src/mesa/drivers/dri/r600/r700_assembler.h @@ -374,6 +374,10 @@ typedef struct r700_AssemblerBase struct prog_instruction * pILInst; GLuint uiCurInst; GLboolean bR6xx; + /* helper to decide which type of instruction to assemble */ + GLboolean is_tex; + /* we inserted helper intructions and need barrier on next TEX ins */ + GLboolean need_tex_barrier; } r700_AssemblerBase; //Internal use @@ -411,6 +415,15 @@ GLboolean assemble_vfetch_instruction(r700_AssemblerBase* pAsm, GLuint number_of_elements, GLenum dataElementType, VTX_FETCH_METHOD* pFetchMethod); +GLboolean assemble_vfetch_instruction2(r700_AssemblerBase* pAsm, + GLuint destination_register, + GLenum type, + GLint size, + GLubyte element, + GLuint _signed, + GLboolean normalize, + VTX_FETCH_METHOD * pFetchMethod); +GLboolean cleanup_vfetch_instructions(r700_AssemblerBase* pAsm); GLuint gethelpr(r700_AssemblerBase* pAsm); void resethelpr(r700_AssemblerBase* pAsm); void checkop_init(r700_AssemblerBase* pAsm); diff --git a/src/mesa/drivers/dri/r600/r700_chip.c b/src/mesa/drivers/dri/r600/r700_chip.c index 312cacffda3..3b7f6fffe03 100644 --- a/src/mesa/drivers/dri/r600/r700_chip.c +++ b/src/mesa/drivers/dri/r600/r700_chip.c @@ -208,11 +208,97 @@ static void r700SetupVTXConstants(GLcontext * ctx, } +extern int getTypeSize(GLenum type); +static void r700SetupVTXConstants2(GLcontext * ctx, + void * pAos, + StreamDesc * pStreamDesc) +{ + context_t *context = R700_CONTEXT(ctx); + struct radeon_aos * paos = (struct radeon_aos *)pAos; + unsigned int nVBsize; + BATCH_LOCALS(&context->radeon); + + unsigned int uSQ_VTX_CONSTANT_WORD0_0; + unsigned int uSQ_VTX_CONSTANT_WORD1_0; + unsigned int uSQ_VTX_CONSTANT_WORD2_0 = 0; + unsigned int uSQ_VTX_CONSTANT_WORD3_0 = 0; + unsigned int uSQ_VTX_CONSTANT_WORD6_0 = 0; + + if (!paos->bo) + return; + + if ((context->radeon.radeonScreen->chip_family == CHIP_FAMILY_RV610) || + (context->radeon.radeonScreen->chip_family == CHIP_FAMILY_RV620) || + (context->radeon.radeonScreen->chip_family == CHIP_FAMILY_RS780) || + (context->radeon.radeonScreen->chip_family == CHIP_FAMILY_RS880) || + (context->radeon.radeonScreen->chip_family == CHIP_FAMILY_RV710)) + r700SyncSurf(context, paos->bo, RADEON_GEM_DOMAIN_GTT, 0, TC_ACTION_ENA_bit); + else + r700SyncSurf(context, paos->bo, RADEON_GEM_DOMAIN_GTT, 0, VC_ACTION_ENA_bit); + + if(0 == pStreamDesc->stride) + { + nVBsize = paos->count * pStreamDesc->size * getTypeSize(pStreamDesc->type); + } + else + { + nVBsize = paos->count * pStreamDesc->stride; + } + + uSQ_VTX_CONSTANT_WORD0_0 = paos->offset; + uSQ_VTX_CONSTANT_WORD1_0 = nVBsize - 1; + + SETfield(uSQ_VTX_CONSTANT_WORD2_0, 0, BASE_ADDRESS_HI_shift, BASE_ADDRESS_HI_mask); /* TODO */ + SETfield(uSQ_VTX_CONSTANT_WORD2_0, pStreamDesc->stride, SQ_VTX_CONSTANT_WORD2_0__STRIDE_shift, + SQ_VTX_CONSTANT_WORD2_0__STRIDE_mask); + SETfield(uSQ_VTX_CONSTANT_WORD2_0, GetSurfaceFormat(pStreamDesc->type, pStreamDesc->size, NULL), + SQ_VTX_CONSTANT_WORD2_0__DATA_FORMAT_shift, + SQ_VTX_CONSTANT_WORD2_0__DATA_FORMAT_mask); /* TODO : trace back api for initial data type, not only GL_FLOAT */ + + if(GL_TRUE == pStreamDesc->normalize) + { + SETfield(uSQ_VTX_CONSTANT_WORD2_0, SQ_NUM_FORMAT_NORM, + SQ_VTX_CONSTANT_WORD2_0__NUM_FORMAT_ALL_shift, SQ_VTX_CONSTANT_WORD2_0__NUM_FORMAT_ALL_mask); + } + //else + //{ + // SETfield(uSQ_VTX_CONSTANT_WORD2_0, SQ_NUM_FORMAT_INT, + // SQ_VTX_CONSTANT_WORD2_0__NUM_FORMAT_ALL_shift, SQ_VTX_CONSTANT_WORD2_0__NUM_FORMAT_ALL_mask); + //} + + if(1 == pStreamDesc->_signed) + { + SETbit(uSQ_VTX_CONSTANT_WORD2_0, SQ_VTX_CONSTANT_WORD2_0__FORMAT_COMP_ALL_bit); + } + + SETfield(uSQ_VTX_CONSTANT_WORD3_0, 1, MEM_REQUEST_SIZE_shift, MEM_REQUEST_SIZE_mask); + SETfield(uSQ_VTX_CONSTANT_WORD6_0, SQ_TEX_VTX_VALID_BUFFER, + SQ_TEX_RESOURCE_WORD6_0__TYPE_shift, SQ_TEX_RESOURCE_WORD6_0__TYPE_mask); + + BEGIN_BATCH_NO_AUTOSTATE(9 + 2); + + R600_OUT_BATCH(CP_PACKET3(R600_IT_SET_RESOURCE, 7)); + R600_OUT_BATCH((pStreamDesc->element + SQ_FETCH_RESOURCE_VS_OFFSET) * FETCH_RESOURCE_STRIDE); + R600_OUT_BATCH(uSQ_VTX_CONSTANT_WORD0_0); + R600_OUT_BATCH(uSQ_VTX_CONSTANT_WORD1_0); + R600_OUT_BATCH(uSQ_VTX_CONSTANT_WORD2_0); + R600_OUT_BATCH(uSQ_VTX_CONSTANT_WORD3_0); + R600_OUT_BATCH(0); + R600_OUT_BATCH(0); + R600_OUT_BATCH(uSQ_VTX_CONSTANT_WORD6_0); + R600_OUT_BATCH_RELOC(uSQ_VTX_CONSTANT_WORD0_0, + paos->bo, + uSQ_VTX_CONSTANT_WORD0_0, + RADEON_GEM_DOMAIN_GTT, 0, 0); + END_BATCH(); + COMMIT_BATCH(); + +} + 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; @@ -221,7 +307,7 @@ 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, @@ -237,8 +323,7 @@ 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__); @@ -258,14 +343,24 @@ 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)) { - /* currently aos are packed */ - r700SetupVTXConstants(ctx, - i, - (void*)(&context->radeon.tcl.aos[j]), - (unsigned int)context->radeon.tcl.aos[j].components, - (unsigned int)context->radeon.tcl.aos[j].stride * 4, - (unsigned int)context->radeon.tcl.aos[j].count); + if(vp->mesa_program->Base.InputsRead & (1 << i)) + { + if(1 == context->selected_vp->uiVersion) + { + /* currently aos are packed */ + r700SetupVTXConstants(ctx, + i, + (void*)(&context->radeon.tcl.aos[j]), + (unsigned int)context->radeon.tcl.aos[j].components, + (unsigned int)context->radeon.tcl.aos[j].stride * 4, + (unsigned int)context->radeon.tcl.aos[j].count); + } + else + { /* context->selected_vp->uiVersion == 2 : aos not always packed */ + r700SetupVTXConstants2(ctx, + (void*)(&context->radeon.tcl.aos[j]), + &(context->stream_desc[j])); + } j++; } } @@ -353,7 +448,7 @@ static void r700SetDepthTarget(context_t *context) SETfield(r700->DB_DEPTH_INFO.u32All, DEPTH_16, DB_DEPTH_INFO__FORMAT_shift, DB_DEPTH_INFO__FORMAT_mask); } - SETfield(r700->DB_DEPTH_INFO.u32All, ARRAY_2D_TILED_THIN1, + SETfield(r700->DB_DEPTH_INFO.u32All, ARRAY_1D_TILED_THIN1, DB_DEPTH_INFO__ARRAY_MODE_shift, DB_DEPTH_INFO__ARRAY_MODE_mask); /* r700->DB_PREFETCH_LIMIT.bits.DEPTH_HEIGHT_TILE_MAX = (context->currentDraw->h >> 3) - 1; */ /* z buffer sie may much bigger than what need, so use actual used h. */ } 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..5627984cf96 100644 --- a/src/mesa/drivers/dri/r600/r700_render.c +++ b/src/mesa/drivers/dri/r600/r700_render.c @@ -43,6 +43,7 @@ #include "tnl/t_context.h" #include "tnl/t_vertex.h" #include "tnl/t_pipeline.h" +#include "vbo/vbo_context.h" #include "r600_context.h" #include "r600_cmdbuf.h" @@ -53,6 +54,7 @@ #include "r700_fragprog.h" #include "r700_state.h" +#include "radeon_buffer_objects.h" #include "radeon_common_context.h" void r700WaitForIdle(context_t *context); @@ -260,6 +262,16 @@ static void r700RunRenderPrimitive(GLcontext * ctx, int start, int end, int prim TNLcontext *tnl = TNL_CONTEXT(ctx); struct vertex_buffer *vb = &tnl->vb; + GLboolean bUseDrawIndex; + if( (NULL != context->ind_buf.bo) && (GL_TRUE != context->ind_buf.bHostIb) ) + { + bUseDrawIndex = GL_TRUE; + } + else + { + bUseDrawIndex = GL_FALSE; + } + type = r700PrimitiveType(prim); num_indices = r700NumVerts(end - start, prim); @@ -270,67 +282,146 @@ static void r700RunRenderPrimitive(GLcontext * ctx, int start, int end, int prim if (type < 0 || num_indices <= 0) return; + if(GL_TRUE == bUseDrawIndex) + { + total_emit = 3 /* VGT_PRIMITIVE_TYPE */ + + 2 /* VGT_INDEX_TYPE */ + + 2 /* NUM_INSTANCES */ + + 5+2; /* DRAW_INDEX */ + } + else + { total_emit = 3 /* VGT_PRIMITIVE_TYPE */ - + 2 /* VGT_INDEX_TYPE */ - + 2 /* NUM_INSTANCES */ - + num_indices + 3; /* DRAW_INDEX_IMMD */ + + 2 /* VGT_INDEX_TYPE */ + + 2 /* NUM_INSTANCES */ + + num_indices + 3; /* DRAW_INDEX_IMMD */ + } - BEGIN_BATCH_NO_AUTOSTATE(total_emit); + BEGIN_BATCH_NO_AUTOSTATE(total_emit); // prim - SETfield(vgt_primitive_type, type, - VGT_PRIMITIVE_TYPE__PRIM_TYPE_shift, VGT_PRIMITIVE_TYPE__PRIM_TYPE_mask); - R600_OUT_BATCH(CP_PACKET3(R600_IT_SET_CONFIG_REG, 1)); - R600_OUT_BATCH(mmVGT_PRIMITIVE_TYPE - ASIC_CONFIG_BASE_INDEX); - R600_OUT_BATCH(vgt_primitive_type); + SETfield(vgt_primitive_type, type, + VGT_PRIMITIVE_TYPE__PRIM_TYPE_shift, VGT_PRIMITIVE_TYPE__PRIM_TYPE_mask); + R600_OUT_BATCH(CP_PACKET3(R600_IT_SET_CONFIG_REG, 1)); + R600_OUT_BATCH(mmVGT_PRIMITIVE_TYPE - ASIC_CONFIG_BASE_INDEX); + R600_OUT_BATCH(vgt_primitive_type); // index type - SETfield(vgt_index_type, DI_INDEX_SIZE_32_BIT, INDEX_TYPE_shift, INDEX_TYPE_mask); - R600_OUT_BATCH(CP_PACKET3(R600_IT_INDEX_TYPE, 0)); - R600_OUT_BATCH(vgt_index_type); + SETfield(vgt_index_type, DI_INDEX_SIZE_32_BIT, INDEX_TYPE_shift, INDEX_TYPE_mask); + + if(GL_TRUE == bUseDrawIndex) + { + if(GL_TRUE != context->ind_buf.is_32bit) + { + SETfield(vgt_index_type, DI_INDEX_SIZE_16_BIT, INDEX_TYPE_shift, INDEX_TYPE_mask); + } + } + + R600_OUT_BATCH(CP_PACKET3(R600_IT_INDEX_TYPE, 0)); + R600_OUT_BATCH(vgt_index_type); // num instances R600_OUT_BATCH(CP_PACKET3(R600_IT_NUM_INSTANCES, 0)); R600_OUT_BATCH(1); // draw packet - vgt_num_indices = num_indices; + vgt_num_indices = num_indices; + + if(GL_TRUE == bUseDrawIndex) + { + SETfield(vgt_draw_initiator, DI_SRC_SEL_DMA, SOURCE_SELECT_shift, SOURCE_SELECT_mask); + } + else + { SETfield(vgt_draw_initiator, DI_SRC_SEL_IMMEDIATE, SOURCE_SELECT_shift, SOURCE_SELECT_mask); + } + SETfield(vgt_draw_initiator, DI_MAJOR_MODE_0, MAJOR_MODE_shift, MAJOR_MODE_mask); + if(GL_TRUE == bUseDrawIndex) + { + R600_OUT_BATCH(CP_PACKET3(R600_IT_DRAW_INDEX, 3)); + R600_OUT_BATCH(context->ind_buf.bo_offset); + R600_OUT_BATCH(0); + R600_OUT_BATCH(vgt_num_indices); + R600_OUT_BATCH(vgt_draw_initiator); + R600_OUT_BATCH_RELOC(context->ind_buf.bo_offset, + context->ind_buf.bo, + context->ind_buf.bo_offset, + RADEON_GEM_DOMAIN_GTT, 0, 0); + } + else + { R600_OUT_BATCH(CP_PACKET3(R600_IT_DRAW_INDEX_IMMD, (num_indices + 1))); R600_OUT_BATCH(vgt_num_indices); R600_OUT_BATCH(vgt_draw_initiator); + } + if(NULL == context->ind_buf.bo) + { for (i = start; i < (start + num_indices); i++) { - if(vb->Elts) - R600_OUT_BATCH(vb->Elts[i]); - else - R600_OUT_BATCH(i); + if(vb->Elts) + { + R600_OUT_BATCH(vb->Elts[i]); + } + else + R600_OUT_BATCH(i); } - END_BATCH(); - COMMIT_BATCH(); + } + else + { + if(GL_TRUE == context->ind_buf.bHostIb) + { + if(GL_TRUE != context->ind_buf.is_32bit) + { + GLushort * pIndex = (GLushort*)ADD_POINTERS(context->ind_buf.bo->ptr, context->ind_buf.bo_offset); + pIndex += start; + for (i = 0; i < num_indices; i++) + { + R600_OUT_BATCH(*pIndex); + pIndex++; + } + } + else + { + GLuint * pIndex = (GLuint*)ADD_POINTERS(context->ind_buf.bo->ptr, context->ind_buf.bo_offset); + pIndex += start; + + for (i = 0; i < num_indices; i++) + { + R600_OUT_BATCH(*pIndex); + pIndex++; + } + } + } + } + END_BATCH(); + COMMIT_BATCH(); } /* start 3d, idle, cb/db flush */ #define PRE_EMIT_STATE_BUFSZ 10 + 5 + 14 -static GLuint r700PredictRenderSize(GLcontext* ctx) +static GLuint r700PredictRenderSize(GLcontext* ctx, GLuint nr_prims) { context_t *context = R700_CONTEXT(ctx); - TNLcontext *tnl = TNL_CONTEXT(ctx); - struct r700_vertex_program *vpc - = (struct r700_vertex_program *)ctx->VertexProgram._Current; - struct vertex_buffer *vb = &tnl->vb; + struct r700_vertex_program *vp = context->selected_vp; 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++) - dwords += vb->Primitive[i].count + 10; + if (nr_prims) + dwords += nr_prims * 14; + else { + TNLcontext *tnl = TNL_CONTEXT(ctx); + struct vertex_buffer *vb = &tnl->vb; + + for (i = 0; i < vb->PrimitiveCount; i++) + dwords += vb->Primitive[i].count + 10; + } state_size = radeonCountStateEmitSize(&context->radeon); flushed = rcommonEnsureCmdBufSpace(&context->radeon, dwords + state_size, __FUNCTION__); @@ -365,13 +456,12 @@ 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); r600UpdateTextureState(ctx); - GLuint emit_end = r700PredictRenderSize(ctx) + GLuint emit_end = r700PredictRenderSize(ctx, 0) + context->radeon.cmdbuf.cs->cdw; r700SetupStreams(ctx); @@ -427,7 +517,10 @@ 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)) @@ -476,4 +569,587 @@ const struct tnl_pipeline_stage *r700_pipeline[] = 0, }; +#define CONVERT( TYPE, MACRO ) do { \ + GLuint i, j, sz; \ + sz = input->Size; \ + if (input->Normalized) { \ + for (i = 0; i < count; i++) { \ + const TYPE *in = (TYPE *)src_ptr; \ + for (j = 0; j < sz; j++) { \ + *dst_ptr++ = MACRO(*in); \ + in++; \ + } \ + src_ptr += stride; \ + } \ + } else { \ + for (i = 0; i < count; i++) { \ + const TYPE *in = (TYPE *)src_ptr; \ + for (j = 0; j < sz; j++) { \ + *dst_ptr++ = (GLfloat)(*in); \ + in++; \ + } \ + src_ptr += stride; \ + } \ + } \ +} while (0) + +/** + * Convert attribute data type to float + * If the attribute uses named buffer object replace the bo with newly allocated bo + */ +static void r700ConvertAttrib(GLcontext *ctx, int count, + const struct gl_client_array *input, + struct StreamDesc *attr) +{ + context_t *context = R700_CONTEXT(ctx); + const GLvoid *src_ptr; + GLboolean mapped_named_bo = GL_FALSE; + GLfloat *dst_ptr; + GLuint stride; + + stride = (input->StrideB == 0) ? getTypeSize(input->Type) * input->Size : input->StrideB; + + /* Convert value for first element only */ + if (input->StrideB == 0) + { + count = 1; + } + + if (input->BufferObj->Name) + { + if (!input->BufferObj->Pointer) + { + ctx->Driver.MapBuffer(ctx, GL_ARRAY_BUFFER, GL_READ_ONLY_ARB, input->BufferObj); + mapped_named_bo = GL_TRUE; + } + + src_ptr = ADD_POINTERS(input->BufferObj->Pointer, input->Ptr); + } + else + { + src_ptr = input->Ptr; + } + + radeonAllocDmaRegion(&context->radeon, &attr->bo, &attr->bo_offset, + sizeof(GLfloat) * input->Size * count, 32); + dst_ptr = (GLfloat *)ADD_POINTERS(attr->bo->ptr, attr->bo_offset); + + assert(src_ptr != NULL); + + switch (input->Type) + { + case GL_DOUBLE: + CONVERT(GLdouble, (GLfloat)); + break; + case GL_UNSIGNED_INT: + CONVERT(GLuint, UINT_TO_FLOAT); + break; + case GL_INT: + CONVERT(GLint, INT_TO_FLOAT); + break; + case GL_UNSIGNED_SHORT: + CONVERT(GLushort, USHORT_TO_FLOAT); + break; + case GL_SHORT: + CONVERT(GLshort, SHORT_TO_FLOAT); + break; + case GL_UNSIGNED_BYTE: + assert(input->Format != GL_BGRA); + CONVERT(GLubyte, UBYTE_TO_FLOAT); + break; + case GL_BYTE: + CONVERT(GLbyte, BYTE_TO_FLOAT); + break; + default: + assert(0); + break; + } + + if (mapped_named_bo) + { + ctx->Driver.UnmapBuffer(ctx, GL_ARRAY_BUFFER, input->BufferObj); + } +} + +static void r700AlignDataToDword(GLcontext *ctx, + const struct gl_client_array *input, + int count, + struct StreamDesc *attr) +{ + context_t *context = R700_CONTEXT(ctx); + const int dst_stride = (input->StrideB + 3) & ~3; + const int size = getTypeSize(input->Type) * input->Size * count; + GLboolean mapped_named_bo = GL_FALSE; + + radeonAllocDmaRegion(&context->radeon, &attr->bo, &attr->bo_offset, size, 32); + + if (!input->BufferObj->Pointer) + { + ctx->Driver.MapBuffer(ctx, GL_ARRAY_BUFFER, GL_READ_ONLY_ARB, input->BufferObj); + mapped_named_bo = GL_TRUE; + } + + { + GLvoid *src_ptr = ADD_POINTERS(input->BufferObj->Pointer, input->Ptr); + GLvoid *dst_ptr = ADD_POINTERS(attr->bo->ptr, attr->bo_offset); + int i; + + for (i = 0; i < count; ++i) + { + _mesa_memcpy(dst_ptr, src_ptr, input->StrideB); + src_ptr += input->StrideB; + dst_ptr += dst_stride; + } + } + + if (mapped_named_bo) + { + ctx->Driver.UnmapBuffer(ctx, GL_ARRAY_BUFFER, input->BufferObj); + } + + attr->stride = dst_stride; +} + +static void r700SetupStreams2(GLcontext *ctx, const struct gl_client_array *input[], int count) +{ + context_t *context = R700_CONTEXT(ctx); + GLuint stride; + int ret; + int i, index; + + R600_STATECHANGE(context, vtx); + + for(index = 0; index < context->nNumActiveAos; index++) + { + struct radeon_aos *aos = &context->radeon.tcl.aos[index]; + i = context->stream_desc[index].element; + + stride = (input[i]->StrideB == 0) ? getTypeSize(input[i]->Type) * input[i]->Size : input[i]->StrideB; + + if (input[i]->Type == GL_DOUBLE || input[i]->Type == GL_UNSIGNED_INT || input[i]->Type == GL_INT || +#if MESA_BIG_ENDIAN + getTypeSize(input[i]->Type) != 4 || +#endif + stride < 4) + { + r700ConvertAttrib(ctx, count, input[i], &context->stream_desc[index]); + } + else + { + if (input[i]->BufferObj->Name) + { + if (stride % 4 != 0) + { + assert(((intptr_t) input[i]->Ptr) % input[i]->StrideB == 0); + r700AlignDataToDword(ctx, input[i], count, &context->stream_desc[index]); + context->stream_desc[index].is_named_bo = GL_FALSE; + } + else + { + context->stream_desc[index].stride = input[i]->StrideB; + context->stream_desc[index].bo_offset = (intptr_t) input[i]->Ptr; + context->stream_desc[index].bo = get_radeon_buffer_object(input[i]->BufferObj)->bo; + context->stream_desc[index].is_named_bo = GL_TRUE; + } + } + else + { + int size; + int local_count = count; + uint32_t *dst; + + if (input[i]->StrideB == 0) + { + size = getTypeSize(input[i]->Type) * input[i]->Size; + local_count = 1; + } + else + { + size = getTypeSize(input[i]->Type) * input[i]->Size * local_count; + } + + radeonAllocDmaRegion(&context->radeon, &context->stream_desc[index].bo, + &context->stream_desc[index].bo_offset, size, 32); + assert(context->stream_desc[index].bo->ptr != NULL); + dst = (uint32_t *)ADD_POINTERS(context->stream_desc[index].bo->ptr, + context->stream_desc[index].bo_offset); + + switch (context->stream_desc[index].dwords) + { + case 1: + radeonEmitVec4(dst, input[i]->Ptr, input[i]->StrideB, local_count); + context->stream_desc[index].stride = 4; + break; + case 2: + radeonEmitVec8(dst, input[i]->Ptr, input[i]->StrideB, local_count); + context->stream_desc[index].stride = 8; + break; + case 3: + radeonEmitVec12(dst, input[i]->Ptr, input[i]->StrideB, local_count); + context->stream_desc[index].stride = 12; + break; + case 4: + radeonEmitVec16(dst, input[i]->Ptr, input[i]->StrideB, local_count); + context->stream_desc[index].stride = 16; + break; + default: + assert(0); + break; + } + } + } + + aos->count = context->stream_desc[index].stride == 0 ? 1 : count; + aos->stride = context->stream_desc[index].stride / sizeof(float); + aos->components = context->stream_desc[index].dwords; + aos->bo = context->stream_desc[index].bo; + aos->offset = context->stream_desc[index].bo_offset; + + if(context->stream_desc[index].is_named_bo) + { + radeon_cs_space_add_persistent_bo(context->radeon.cmdbuf.cs, + context->stream_desc[index].bo, + RADEON_GEM_DOMAIN_GTT, 0); + } + } + + context->radeon.tcl.aos_count = context->nNumActiveAos; + ret = radeon_cs_space_check_with_bo(context->radeon.cmdbuf.cs, + first_elem(&context->radeon.dma.reserved)->bo, + RADEON_GEM_DOMAIN_GTT, 0); +} + +static void r700FreeData(GLcontext *ctx) +{ + /* Need to zero tcl.aos[n].bo and tcl.elt_dma_bo + * to prevent double unref in radeonReleaseArrays + * called during context destroy + */ + context_t *context = R700_CONTEXT(ctx); + + int i; + + for (i = 0; i < context->nNumActiveAos; i++) + { + if (!context->stream_desc[i].is_named_bo) + { + radeon_bo_unref(context->stream_desc[i].bo); + } + context->radeon.tcl.aos[i].bo = NULL; + } + + if (context->ind_buf.bo != NULL) + { + if(context->ind_buf.bHostIb != GL_TRUE) + { + radeon_bo_unref(context->ind_buf.bo); + } + else + { + FREE(context->ind_buf.bo->ptr); + FREE(context->ind_buf.bo); + context->ind_buf.bo = NULL; + } + } +} + +static void r700FixupIndexBuffer(GLcontext *ctx, const struct _mesa_index_buffer *mesa_ind_buf) +{ + context_t *context = R700_CONTEXT(ctx); + GLvoid *src_ptr; + GLuint *out; + int i; + GLboolean mapped_named_bo = GL_FALSE; + + if (mesa_ind_buf->obj->Name && !mesa_ind_buf->obj->Pointer) + { + ctx->Driver.MapBuffer(ctx, GL_ELEMENT_ARRAY_BUFFER, GL_READ_ONLY_ARB, mesa_ind_buf->obj); + mapped_named_bo = GL_TRUE; + assert(mesa_ind_buf->obj->Pointer != NULL); + } + src_ptr = ADD_POINTERS(mesa_ind_buf->obj->Pointer, mesa_ind_buf->ptr); + + if (mesa_ind_buf->type == GL_UNSIGNED_BYTE) + { + GLuint size = sizeof(GLushort) * ((mesa_ind_buf->count + 1) & ~1); + GLubyte *in = (GLubyte *)src_ptr; + + if(context->ind_buf.bHostIb != GL_TRUE) + { + radeonAllocDmaRegion(&context->radeon, &context->ind_buf.bo, + &context->ind_buf.bo_offset, size, 4); + + assert(context->ind_buf.bo->ptr != NULL); + out = (GLuint *)ADD_POINTERS(context->ind_buf.bo->ptr, context->ind_buf.bo_offset); + } + else + { + context->ind_buf.bo = MALLOC_STRUCT(radeon_bo); + context->ind_buf.bo->ptr = ALIGN_MALLOC(size, 4); + context->ind_buf.bo_offset = 0; + out = (GLuint *)context->ind_buf.bo->ptr; + } + + for (i = 0; i + 1 < mesa_ind_buf->count; i += 2) + { + *out++ = in[i] | in[i + 1] << 16; + } + + if (i < mesa_ind_buf->count) + { + *out++ = in[i]; + } + +#if MESA_BIG_ENDIAN + } + else + { /* if (mesa_ind_buf->type == GL_UNSIGNED_SHORT) */ + GLushort *in = (GLushort *)src_ptr; + GLuint size = sizeof(GLushort) * ((mesa_ind_buf->count + 1) & ~1); + + if(context->ind_buf.bHostIb != GL_TRUE) + { + radeonAllocDmaRegion(&context->radeon, &context->ind_buf.bo, + &context->ind_buf.bo_offset, size, 4); + + assert(context->ind_buf.bo->ptr != NULL); + out = (GLuint *)ADD_POINTERS(context->ind_buf.bo->ptr, context->ind_buf.bo_offset); + } + else + { + context->ind_buf.bo = MALLOC_STRUCT(radeon_bo); + context->ind_buf.bo->ptr = ALIGN_MALLOC(size, 4); + context->ind_buf.bo_offset = 0; + out = (GLuint *)context->ind_buf.bo->ptr; + } + + for (i = 0; i + 1 < mesa_ind_buf->count; i += 2) + { + *out++ = in[i] | in[i + 1] << 16; + } + + if (i < mesa_ind_buf->count) + { + *out++ = in[i]; + } +#endif + } + + context->ind_buf.is_32bit = GL_FALSE; + context->ind_buf.count = mesa_ind_buf->count; + + if (mapped_named_bo) + { + ctx->Driver.UnmapBuffer(ctx, GL_ELEMENT_ARRAY_BUFFER, mesa_ind_buf->obj); + } +} + +static void r700SetupIndexBuffer(GLcontext *ctx, const struct _mesa_index_buffer *mesa_ind_buf) +{ + context_t *context = R700_CONTEXT(ctx); + + if (!mesa_ind_buf) { + context->ind_buf.bo = NULL; + return; + } + + context->ind_buf.bHostIb = GL_FALSE; + +#if MESA_BIG_ENDIAN + if (mesa_ind_buf->type == GL_UNSIGNED_INT) + { +#else + if (mesa_ind_buf->type != GL_UNSIGNED_BYTE) + { +#endif + const GLvoid *src_ptr; + GLvoid *dst_ptr; + GLboolean mapped_named_bo = GL_FALSE; + + if (mesa_ind_buf->obj->Name && !mesa_ind_buf->obj->Pointer) + { + ctx->Driver.MapBuffer(ctx, GL_ELEMENT_ARRAY_BUFFER, GL_READ_ONLY_ARB, mesa_ind_buf->obj); + assert(mesa_ind_buf->obj->Pointer != NULL); + mapped_named_bo = GL_TRUE; + } + + src_ptr = ADD_POINTERS(mesa_ind_buf->obj->Pointer, mesa_ind_buf->ptr); + + const GLuint size = mesa_ind_buf->count * getTypeSize(mesa_ind_buf->type); + + if(context->ind_buf.bHostIb != GL_TRUE) + { + radeonAllocDmaRegion(&context->radeon, &context->ind_buf.bo, + &context->ind_buf.bo_offset, size, 4); + assert(context->ind_buf.bo->ptr != NULL); + dst_ptr = ADD_POINTERS(context->ind_buf.bo->ptr, context->ind_buf.bo_offset); + } + else + { + context->ind_buf.bo = MALLOC_STRUCT(radeon_bo); + context->ind_buf.bo->ptr = ALIGN_MALLOC(size, 4); + context->ind_buf.bo_offset = 0; + dst_ptr = context->ind_buf.bo->ptr; + } + + _mesa_memcpy(dst_ptr, src_ptr, size); + + context->ind_buf.is_32bit = (mesa_ind_buf->type == GL_UNSIGNED_INT); + context->ind_buf.count = mesa_ind_buf->count; + + if (mapped_named_bo) + { + ctx->Driver.UnmapBuffer(ctx, GL_ELEMENT_ARRAY_BUFFER, mesa_ind_buf->obj); + } + } + else + { + r700FixupIndexBuffer(ctx, mesa_ind_buf); + } +} + +static GLboolean r700TryDrawPrims(GLcontext *ctx, + const struct gl_client_array *arrays[], + const struct _mesa_prim *prim, + GLuint nr_prims, + const struct _mesa_index_buffer *ib, + GLuint min_index, + GLuint max_index ) +{ + context_t *context = R700_CONTEXT(ctx); + radeonContextPtr radeon = &context->radeon; + GLuint i, id = 0; + struct radeon_renderbuffer *rrb; + + if (ctx->NewState) + { + _mesa_update_state( ctx ); + } + + _tnl_UpdateFixedFunctionProgram(ctx); + r700SetVertexFormat(ctx, arrays, max_index + 1); + r700SetupIndexBuffer(ctx, ib); + /* shaders need to be updated before buffers are validated */ + r700UpdateShaders2(ctx); + if (!r600ValidateBuffers(ctx)) + return GL_FALSE; + + /* always emit CB base to prevent + * lock ups on some chips. + */ + R600_STATECHANGE(context, cb_target); + /* mark vtx as dirty since it changes per-draw */ + R600_STATECHANGE(context, vtx); + + r700SetScissor(context); + r700SetupVertexProgram(ctx); + r700SetupFragmentProgram(ctx); + r600UpdateTextureState(ctx); + + GLuint emit_end = r700PredictRenderSize(ctx, nr_prims) + + context->radeon.cmdbuf.cs->cdw; + + r700SetupStreams2(ctx, arrays, max_index + 1); + + radeonEmitState(radeon); + + radeon_debug_add_indent(); + for (i = 0; i < nr_prims; ++i) + { + r700RunRenderPrimitive(ctx, + prim[i].start, + prim[i].start + prim[i].count, + prim[i].mode); + } + radeon_debug_remove_indent(); + + /* Flush render op cached for last several quads. */ + r700WaitForIdleClean(context); + + rrb = radeon_get_colorbuffer(&context->radeon); + if (rrb && rrb->bo) + r700SyncSurf(context, rrb->bo, 0, RADEON_GEM_DOMAIN_VRAM, + CB_ACTION_ENA_bit | (1 << (id + 6))); + + rrb = radeon_get_depthbuffer(&context->radeon); + if (rrb && rrb->bo) + r700SyncSurf(context, rrb->bo, 0, RADEON_GEM_DOMAIN_VRAM, + DB_ACTION_ENA_bit | DB_DEST_BASE_ENA_bit); + + r700FreeData(ctx); + + if (emit_end < context->radeon.cmdbuf.cs->cdw) + { + WARN_ONCE("Rendering was %d commands larger than predicted size." + " We might overflow command buffer.\n", context->radeon.cmdbuf.cs->cdw - emit_end); + } + + return GL_TRUE; +} + +static void r700DrawPrimsRe(GLcontext *ctx, + const struct gl_client_array *arrays[], + const struct _mesa_prim *prim, + GLuint nr_prims, + const struct _mesa_index_buffer *ib, + GLboolean index_bounds_valid, + GLuint min_index, + GLuint max_index) +{ + GLboolean retval = GL_FALSE; + + /* This check should get folded into just the places that + * min/max index are really needed. + */ + if (!index_bounds_valid) { + vbo_get_minmax_index(ctx, prim, ib, &min_index, &max_index); + } + + if (min_index) { + vbo_rebase_prims( ctx, arrays, prim, nr_prims, ib, min_index, max_index, r700DrawPrimsRe ); + return; + } + + /* Make an attempt at drawing */ + retval = r700TryDrawPrims(ctx, arrays, prim, nr_prims, ib, min_index, max_index); + + /* If failed run tnl pipeline - it should take care of fallbacks */ + if (!retval) + _tnl_draw_prims(ctx, arrays, prim, nr_prims, ib, min_index, max_index); +} + +static void r700DrawPrims(GLcontext *ctx, + const struct gl_client_array *arrays[], + const struct _mesa_prim *prim, + GLuint nr_prims, + const struct _mesa_index_buffer *ib, + GLboolean index_bounds_valid, + GLuint min_index, + GLuint max_index) +{ + context_t *context = R700_CONTEXT(ctx); + + /* For non indexed drawing, using tnl pipe. */ + if(!ib) + { + context->ind_buf.bo = NULL; + + _tnl_vbo_draw_prims(ctx, arrays, prim, nr_prims, ib, + index_bounds_valid, min_index, max_index); + return; + } + + r700DrawPrimsRe(ctx, arrays, prim, nr_prims, ib, index_bounds_valid, min_index, max_index); +} + +void r700InitDraw(GLcontext *ctx) +{ + struct vbo_context *vbo = vbo_context(ctx); + + /* to be enabled */ + /* + vbo->draw_prims = r700DrawPrims; + */ +} + diff --git a/src/mesa/drivers/dri/r600/r700_shader.c b/src/mesa/drivers/dri/r600/r700_shader.c index b4fd51c1370..955ea4e4e1d 100644 --- a/src/mesa/drivers/dri/r600/r700_shader.c +++ b/src/mesa/drivers/dri/r600/r700_shader.c @@ -60,6 +60,55 @@ void AddInstToList(TypedShaderList * plstCFInstructions, R700ShaderInstruction * plstCFInstructions->uNumOfNode++; } +void TakeInstOutFromList(TypedShaderList * plstCFInstructions, R700ShaderInstruction * pInst) +{ + GLuint ulIndex = 0; + GLboolean bFound = GL_FALSE; + R700ShaderInstruction * pPrevInst = NULL; + R700ShaderInstruction * pCurInst = plstCFInstructions->pHead; + + /* Need go thro list to make sure pInst is there. */ + while(NULL != pCurInst) + { + if(pCurInst == pInst) + { + bFound = GL_TRUE; + break; + } + + pPrevInst = pCurInst; + pCurInst = pCurInst->pNextInst; + } + if(GL_TRUE == bFound) + { + plstCFInstructions->uNumOfNode--; + + pCurInst = pInst->pNextInst; + ulIndex = pInst->m_uIndex; + while(NULL != pCurInst) + { + pCurInst->m_uIndex = ulIndex; + ulIndex++; + pCurInst = pCurInst->pNextInst; + } + + if(plstCFInstructions->pHead == pInst) + { + plstCFInstructions->pHead = pInst->pNextInst; + } + if(plstCFInstructions->pTail == pInst) + { + plstCFInstructions->pTail = pPrevInst; + } + if(NULL != pPrevInst) + { + pPrevInst->pNextInst = pInst->pNextInst; + } + + FREE(pInst); + } +} + void Init_R700_Shader(R700_Shader * pShader) { pShader->Type = R700_SHADER_INVALID; @@ -488,6 +537,47 @@ void DebugPrint(void) { } +void cleanup_vfetch_shaderinst(R700_Shader *pShader) +{ + R700ShaderInstruction *pInst; + R700ShaderInstruction *pInstToFree; + R700VertexInstruction *pVTXInst; + R700ControlFlowInstruction *pCFInst; + + pInst = pShader->lstVTXInstructions.pHead; + while(NULL != pInst) + { + pVTXInst = (R700VertexInstruction *)pInst; + pShader->uShaderBinaryDWORDSize -= GetInstructionSize(pVTXInst->m_ShaderInstType); + + if(NULL != pVTXInst->m_pLinkedGenericClause) + { + pCFInst = (R700ControlFlowInstruction*)(pVTXInst->m_pLinkedGenericClause); + + TakeInstOutFromList(&(pShader->lstCFInstructions), + (R700ShaderInstruction*)pCFInst); + + pShader->uShaderBinaryDWORDSize -= GetInstructionSize(pCFInst->m_ShaderInstType); + } + + pInst = pInst->pNextInst; + }; + + //destroy each item in pShader->lstVTXInstructions; + pInst = pShader->lstVTXInstructions.pHead; + while(NULL != pInst) + { + pInstToFree = pInst; + pInst = pInst->pNextInst; + FREE(pInstToFree); + }; + + //set NULL pShader->lstVTXInstructions + pShader->lstVTXInstructions.pHead=NULL; + pShader->lstVTXInstructions.pTail=NULL; + pShader->lstVTXInstructions.uNumOfNode=0; +} + void Clean_Up_Shader(R700_Shader *pShader) { FREE(pShader->pProgram); diff --git a/src/mesa/drivers/dri/r600/r700_shader.h b/src/mesa/drivers/dri/r600/r700_shader.h index bfd01e1a93a..997cb05aaf8 100644 --- a/src/mesa/drivers/dri/r600/r700_shader.h +++ b/src/mesa/drivers/dri/r600/r700_shader.h @@ -143,6 +143,7 @@ void LoadProgram(R700_Shader *pShader); void UpdateShaderRegisters(R700_Shader *pShader); void DeleteInstructions(R700_Shader *pShader); void DebugPrint(void); +void cleanup_vfetch_shaderinst(R700_Shader *pShader); void Clean_Up_Shader(R700_Shader *pShader); diff --git a/src/mesa/drivers/dri/r600/r700_state.c b/src/mesa/drivers/dri/r600/r700_state.c index e91aa43118f..fbff109455d 100644 --- a/src/mesa/drivers/dri/r600/r700_state.c +++ b/src/mesa/drivers/dri/r600/r700_state.c @@ -92,7 +92,25 @@ void r700UpdateShaders (GLcontext * ctx) //---------------------------------- } } - r700SelectVertexShader(ctx); + r700SelectVertexShader(ctx, 1); + r700UpdateStateParameters(ctx, _NEW_PROGRAM | _NEW_PROGRAM_CONSTANTS); + context->radeon.NewGLState = 0; +} + +void r700UpdateShaders2(GLcontext * ctx) +{ + context_t *context = R700_CONTEXT(ctx); + + /* should only happenen once, just after context is created */ + /* TODO: shouldn't we fallback to sw here? */ + if (!ctx->FragmentProgram._Current) { + _mesa_fprintf(stderr, "No ctx->FragmentProgram._Current!!\n"); + return; + } + + r700SelectFragmentShader(ctx); + + r700SelectVertexShader(ctx, 2); r700UpdateStateParameters(ctx, _NEW_PROGRAM | _NEW_PROGRAM_CONSTANTS); context->radeon.NewGLState = 0; } @@ -845,9 +863,9 @@ static void r700PointSize(GLcontext * ctx, GLfloat size) size = CLAMP(size, ctx->Const.MinPointSize, ctx->Const.MaxPointSize); /* format is 12.4 fixed point */ - SETfield(r700->PA_SU_POINT_SIZE.u32All, (int)(size * 16), + SETfield(r700->PA_SU_POINT_SIZE.u32All, (int)(size * 8.0), PA_SU_POINT_SIZE__HEIGHT_shift, PA_SU_POINT_SIZE__HEIGHT_mask); - SETfield(r700->PA_SU_POINT_SIZE.u32All, (int)(size * 16), + SETfield(r700->PA_SU_POINT_SIZE.u32All, (int)(size * 8.0), PA_SU_POINT_SIZE__WIDTH_shift, PA_SU_POINT_SIZE__WIDTH_mask); } @@ -862,11 +880,11 @@ static void r700PointParameter(GLcontext * ctx, GLenum pname, const GLfloat * pa /* format is 12.4 fixed point */ switch (pname) { case GL_POINT_SIZE_MIN: - SETfield(r700->PA_SU_POINT_MINMAX.u32All, (int)(ctx->Point.MinSize * 16.0), + SETfield(r700->PA_SU_POINT_MINMAX.u32All, (int)(ctx->Point.MinSize * 8.0), MIN_SIZE_shift, MIN_SIZE_mask); break; case GL_POINT_SIZE_MAX: - SETfield(r700->PA_SU_POINT_MINMAX.u32All, (int)(ctx->Point.MaxSize * 16.0), + SETfield(r700->PA_SU_POINT_MINMAX.u32All, (int)(ctx->Point.MaxSize * 8.0), MAX_SIZE_shift, MAX_SIZE_mask); break; case GL_POINT_DISTANCE_ATTENUATION: @@ -1130,20 +1148,25 @@ static void r700PolygonOffset(GLcontext * ctx, GLfloat factor, GLfloat units) // context_t *context = R700_CONTEXT(ctx); R700_CHIP_CONTEXT *r700 = (R700_CHIP_CONTEXT*)(&context->hw); GLfloat constant = units; + GLchar depth = 0; + + R600_STATECHANGE(context, poly); switch (ctx->Visual.depthBits) { case 16: constant *= 4.0; + depth = -16; break; case 24: constant *= 2.0; + depth = -24; break; } factor *= 12.0; - - R600_STATECHANGE(context, poly); - + SETfield(r700->PA_SU_POLY_OFFSET_DB_FMT_CNTL.u32All, depth, + POLY_OFFSET_NEG_NUM_DB_BITS_shift, POLY_OFFSET_NEG_NUM_DB_BITS_mask); + //r700->PA_SU_POLY_OFFSET_CLAMP.f32All = constant; //??? r700->PA_SU_POLY_OFFSET_FRONT_SCALE.f32All = factor; r700->PA_SU_POLY_OFFSET_FRONT_OFFSET.f32All = constant; r700->PA_SU_POLY_OFFSET_BACK_SCALE.f32All = factor; @@ -1272,10 +1295,11 @@ void r700SetScissor(context_t *context) //--------------- return; } if (context->radeon.state.scissor.enabled) { + /* r600 has exclusive scissors */ x1 = context->radeon.state.scissor.rect.x1; y1 = context->radeon.state.scissor.rect.y1; - x2 = context->radeon.state.scissor.rect.x2; - y2 = context->radeon.state.scissor.rect.y2; + x2 = context->radeon.state.scissor.rect.x2 + 1; + y2 = context->radeon.state.scissor.rect.y2 + 1; } else { if (context->radeon.radeonScreen->driScreen->dri2.enabled) { x1 = 0; diff --git a/src/mesa/drivers/dri/r600/r700_state.h b/src/mesa/drivers/dri/r600/r700_state.h index 0f53d5b4c59..209189d8d72 100644 --- a/src/mesa/drivers/dri/r600/r700_state.h +++ b/src/mesa/drivers/dri/r600/r700_state.h @@ -35,6 +35,7 @@ extern void r700UpdateStateParameters(GLcontext * ctx, GLuint new_state); extern void r700UpdateShaders (GLcontext * ctx); +extern void r700UpdateShaders2(GLcontext * ctx); extern void r700UpdateViewportOffset(GLcontext * ctx); diff --git a/src/mesa/drivers/dri/r600/r700_vertprog.c b/src/mesa/drivers/dri/r600/r700_vertprog.c index d107f99e7ba..e7a209be9d8 100644 --- a/src/mesa/drivers/dri/r600/r700_vertprog.c +++ b/src/mesa/drivers/dri/r600/r700_vertprog.c @@ -35,12 +35,14 @@ #include "main/mtypes.h" #include "tnl/t_context.h" +#include "shader/program.h" #include "shader/prog_parameter.h" #include "shader/prog_statevars.h" #include "radeon_debug.h" #include "r600_context.h" #include "r600_cmdbuf.h" +#include "shader/programopt.c" #include "r700_debug.h" #include "r700_vertprog.h" @@ -157,7 +159,35 @@ GLboolean Process_Vertex_Program_Vfetch_Instructions( return GL_TRUE; } -void Map_Vertex_Program(struct r700_vertex_program *vp, +GLboolean Process_Vertex_Program_Vfetch_Instructions2( + GLcontext *ctx, + struct r700_vertex_program *vp, + struct gl_vertex_program *mesa_vp) +{ + int i; + context_t *context = R700_CONTEXT(ctx); + + VTX_FETCH_METHOD vtxFetchMethod; + vtxFetchMethod.bEnableMini = GL_FALSE; + vtxFetchMethod.mega_fetch_remainder = 0; + + for(i=0; i<context->nNumActiveAos; i++) + { + assemble_vfetch_instruction2(&vp->r700AsmCode, + vp->r700AsmCode.ucVP_AttributeMap[context->stream_desc[i].element], + context->stream_desc[i].type, + context->stream_desc[i].size, + context->stream_desc[i].element, + context->stream_desc[i]._signed, + context->stream_desc[i].normalize, + &vtxFetchMethod); + } + + return GL_TRUE; +} + +void Map_Vertex_Program(GLcontext *ctx, + struct r700_vertex_program *vp, struct gl_vertex_program *mesa_vp) { GLuint ui; @@ -173,11 +203,22 @@ void Map_Vertex_Program(struct r700_vertex_program *vp, pAsm->number_used_registers += num_inputs; // Create VFETCH instructions for inputs - if (GL_TRUE != Process_Vertex_Program_Vfetch_Instructions(vp, mesa_vp) ) - { - radeon_error("Calling Process_Vertex_Program_Vfetch_Instructions return error. \n"); - return; //error - } + if(1 == vp->uiVersion) + { + if (GL_TRUE != Process_Vertex_Program_Vfetch_Instructions(vp, mesa_vp) ) + { + radeon_error("Calling Process_Vertex_Program_Vfetch_Instructions return error. \n"); + return; + } + } + else + { + if (GL_TRUE != Process_Vertex_Program_Vfetch_Instructions2(ctx, vp, mesa_vp) ) + { + radeon_error("Calling Process_Vertex_Program_Vfetch_Instructions2 return error. \n"); + return; + } + } // Map Outputs pAsm->number_of_exports = Map_Vertex_Output(pAsm, mesa_vp, pAsm->number_used_registers); @@ -258,28 +299,61 @@ 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, + GLint nVer) { + 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->uiVersion = nVer; + vp->mesa_program = (struct gl_vertex_program *)_mesa_clone_program(ctx, &mesa_vp->Base); + + if (mesa_vp->IsPositionInvariant) + { + _mesa_insert_mvp_code(ctx, vp->mesa_program); + } + + 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 ); + Map_Vertex_Program(ctx, vp, vp->mesa_program ); - if(GL_FALSE == Find_Instruction_Dependencies_vp(vp, mesa_vp)) + if(GL_FALSE == Find_Instruction_Dependencies_vp(vp, vp->mesa_program)) { - return GL_FALSE; + return NULL; } - if(GL_FALSE == AssembleInstr(mesa_vp->Base.NumInstructions, - &(mesa_vp->Base.Instructions[0]), + if(GL_FALSE == AssembleInstr(vp->mesa_program->Base.NumInstructions, + &(vp->mesa_program->Base.Instructions[0]), &(vp->r700AsmCode)) ) { - return GL_FALSE; + return NULL; } - if(GL_FALSE == Process_Vertex_Exports(&(vp->r700AsmCode), mesa_vp->Base.OutputsWritten) ) + if(GL_FALSE == Process_Vertex_Exports(&(vp->r700AsmCode), vp->mesa_program->Base.OutputsWritten) ) { - return GL_FALSE; + return NULL; } vp->r700Shader.nRegs = (vp->r700AsmCode.number_used_registers == 0) ? 0 @@ -289,72 +363,216 @@ GLboolean r700TranslateVertexShader(struct r700_vertex_program *vp, vp->translated = GL_TRUE; - return GL_TRUE; + return vp; } -void r700SelectVertexShader(GLcontext *ctx) +void r700SelectVertexShader(GLcontext *ctx, GLint nVersion) { 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; + GLbitfield InputsRead; - if (context->radeon.NewGLState & (_NEW_PROGRAM_CONSTANTS|_NEW_PROGRAM)) - { - vpc->needUpdateVF = 1; - } + vpc = (struct r700_vertex_program_cont *)ctx->VertexProgram._Current; - if (context->radeon.radeonScreen->chip_family < CHIP_FAMILY_RV770) + InputsRead = vpc->mesa_program.Base.InputsRead; + if (vpc->mesa_program.IsPositionInvariant) { - vpc->r700AsmCode.bR6xx = 1; - } - + InputsRead |= VERT_BIT_POS; + } + + for (vp = vpc->progs; vp; vp = vp->next) + { + 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(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 (match) + { + context->selected_vp = vp; + return; + } + } + + vp = r700TranslateVertexShader(ctx, &(vpc->mesa_program), nVersion); + if(!vp) + { + radeon_error("Failed to translate vertex shader. \n"); + return; + } + vp->next = vpc->progs; + vpc->progs = vp; + context->selected_vp = vp; + return; +} + +int getTypeSize(GLenum type) +{ + switch (type) + { + case GL_DOUBLE: + return sizeof(GLdouble); + case GL_FLOAT: + return sizeof(GLfloat); + case GL_INT: + return sizeof(GLint); + case GL_UNSIGNED_INT: + return sizeof(GLuint); + case GL_SHORT: + return sizeof(GLshort); + case GL_UNSIGNED_SHORT: + return sizeof(GLushort); + case GL_BYTE: + return sizeof(GLbyte); + case GL_UNSIGNED_BYTE: + return sizeof(GLubyte); + default: + assert(0); + return 0; + } +} + +static void r700TranslateAttrib(GLcontext *ctx, GLuint unLoc, int count, const struct gl_client_array *input) +{ + context_t *context = R700_CONTEXT(ctx); + + StreamDesc * pStreamDesc = &(context->stream_desc[context->nNumActiveAos]); + + GLuint stride; + + stride = (input->StrideB == 0) ? getTypeSize(input->Type) * input->Size + : input->StrideB; + + if (input->Type == GL_DOUBLE || input->Type == GL_UNSIGNED_INT || input->Type == GL_INT || +#if MESA_BIG_ENDIAN + getTypeSize(input->Type) != 4 || +#endif + stride < 4) + { + pStreamDesc->type = GL_FLOAT; + + if (input->StrideB == 0) + { + pStreamDesc->stride = 0; + } + else + { + pStreamDesc->stride = sizeof(GLfloat) * input->Size; + } + pStreamDesc->dwords = input->Size; + pStreamDesc->is_named_bo = GL_FALSE; + } + else + { + pStreamDesc->type = input->Type; + pStreamDesc->dwords = (getTypeSize(input->Type) * input->Size + 3)/ 4; + if (!input->BufferObj->Name) + { + if (input->StrideB == 0) + { + pStreamDesc->stride = 0; + } + else + { + pStreamDesc->stride = (getTypeSize(pStreamDesc->type) * input->Size + 3) & ~3; + } + + pStreamDesc->is_named_bo = GL_FALSE; + } + } - if(GL_FALSE == vpc->translated) { - r700TranslateVertexShader(vpc, &(vpc->mesa_program) ); + pStreamDesc->size = input->Size; + pStreamDesc->dst_loc = context->nNumActiveAos; + pStreamDesc->element = unLoc; + + switch (pStreamDesc->type) + { //GetSurfaceFormat + case GL_FLOAT: + pStreamDesc->_signed = 0; + pStreamDesc->normalize = GL_FALSE; + break; + case GL_SHORT: + pStreamDesc->_signed = 1; + pStreamDesc->normalize = input->Normalized; + break; + case GL_BYTE: + pStreamDesc->_signed = 1; + pStreamDesc->normalize = input->Normalized; + break; + case GL_UNSIGNED_SHORT: + pStreamDesc->_signed = 0; + pStreamDesc->normalize = input->Normalized; + break; + case GL_UNSIGNED_BYTE: + pStreamDesc->_signed = 0; + pStreamDesc->normalize = input->Normalized; + break; + default: + case GL_INT: + case GL_UNSIGNED_INT: + case GL_DOUBLE: + assert(0); + break; } + context->nNumActiveAos++; +} + +void r700SetVertexFormat(GLcontext *ctx, const struct gl_client_array *arrays[], int count) +{ + context_t *context = R700_CONTEXT(ctx); + struct r700_vertex_program *vpc + = (struct r700_vertex_program *)ctx->VertexProgram._Current; + + struct gl_vertex_program * mesa_vp = (struct gl_vertex_program *)&(vpc->mesa_program); + unsigned int unLoc = 0; + unsigned int unBit = mesa_vp->Base.InputsRead; + context->nNumActiveAos = 0; + + while(unBit) + { + if(unBit & 1) + { + r700TranslateAttrib(ctx, unLoc, count, arrays[unLoc]); + } + + unBit >>= 1; + ++unLoc; + } } 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 +628,7 @@ 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..f9a3e395ee9 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; @@ -52,13 +52,20 @@ struct r700_vertex_program GLboolean translated; GLboolean loaded; - GLboolean needUpdateVF; + GLint uiVersion; void * shaderbo; 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, @@ -69,19 +76,28 @@ unsigned int Map_Vertex_Input(r700_AssemblerBase *pAsm, GLboolean Process_Vertex_Program_Vfetch_Instructions( struct r700_vertex_program *vp, struct gl_vertex_program *mesa_vp); -void Map_Vertex_Program(struct r700_vertex_program *vp, +GLboolean Process_Vertex_Program_Vfetch_Instructions2( + GLcontext *ctx, + struct r700_vertex_program *vp, + struct gl_vertex_program *mesa_vp); +void Map_Vertex_Program(GLcontext *ctx, + struct r700_vertex_program *vp, struct gl_vertex_program *mesa_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 gl_vertex_program *mesa_vp); +struct r700_vertex_program* r700TranslateVertexShader(GLcontext *ctx, + struct gl_vertex_program *mesa_vp, + GLint nVer); /* Interface */ -extern void r700SelectVertexShader(GLcontext *ctx); +extern void r700SelectVertexShader(GLcontext *ctx, GLint nVersion); +extern void r700SetVertexFormat(GLcontext *ctx, const struct gl_client_array *arrays[], int count); extern GLboolean r700SetupVertexProgram(GLcontext * ctx); extern void * r700GetActiveVpShaderBo(GLcontext * ctx); +extern int getTypeSize(GLenum type); + #endif /* _R700_VERTPROG_H_ */ diff --git a/src/mesa/drivers/dri/r600/radeon_buffer_objects.c b/src/mesa/drivers/dri/r600/radeon_buffer_objects.c new file mode 120000 index 00000000000..f6a5f664701 --- /dev/null +++ b/src/mesa/drivers/dri/r600/radeon_buffer_objects.c @@ -0,0 +1 @@ +../radeon/radeon_buffer_objects.c
\ No newline at end of file diff --git a/src/mesa/drivers/dri/r600/radeon_buffer_objects.h b/src/mesa/drivers/dri/r600/radeon_buffer_objects.h new file mode 120000 index 00000000000..2f134fd17b8 --- /dev/null +++ b/src/mesa/drivers/dri/r600/radeon_buffer_objects.h @@ -0,0 +1 @@ +../radeon/radeon_buffer_objects.h
\ No newline at end of file diff --git a/src/mesa/drivers/dri/radeon/Makefile b/src/mesa/drivers/dri/radeon/Makefile index 1f286776b5f..b1efc72872f 100644 --- a/src/mesa/drivers/dri/radeon/Makefile +++ b/src/mesa/drivers/dri/radeon/Makefile @@ -47,8 +47,7 @@ C_SOURCES = \ $(DRIVER_SOURCES) \ $(CS_SOURCES) -DRIVER_DEFINES = -DRADEON_COMMON=0 \ - -Wall +DRIVER_DEFINES = -DRADEON_R100 -Wall DRI_LIB_DEPS += $(RADEON_LDFLAGS) diff --git a/src/mesa/drivers/dri/radeon/radeon_buffer_objects.c b/src/mesa/drivers/dri/radeon/radeon_buffer_objects.c index a24b6dac265..8fac5c6c512 100644 --- a/src/mesa/drivers/dri/radeon/radeon_buffer_objects.c +++ b/src/mesa/drivers/dri/radeon/radeon_buffer_objects.c @@ -187,7 +187,11 @@ radeonMapBuffer(GLcontext * ctx, radeon_bo_map(radeon_obj->bo, access == GL_WRITE_ONLY_ARB); - return obj->Pointer = radeon_obj->bo->ptr; + obj->Pointer = radeon_obj->bo->ptr; + obj->Length = obj->Size; + obj->Offset = 0; + + return obj->Pointer; } @@ -203,9 +207,12 @@ radeonUnmapBuffer(GLcontext * ctx, if (radeon_obj->bo != NULL) { radeon_bo_unmap(radeon_obj->bo); - obj->Pointer = NULL; } + obj->Pointer = NULL; + obj->Offset = 0; + obj->Length = 0; + return GL_TRUE; } diff --git a/src/mesa/drivers/dri/radeon/radeon_common.c b/src/mesa/drivers/dri/radeon/radeon_common.c index a4c7b40798a..f8a4cdb4954 100644 --- a/src/mesa/drivers/dri/radeon/radeon_common.c +++ b/src/mesa/drivers/dri/radeon/radeon_common.c @@ -232,13 +232,13 @@ void radeonUpdateScissor( GLcontext *ctx ) __DRIdrawablePrivate *dPriv = radeon_get_drawable(rmesa); x1 += dPriv->x; - x2 += dPriv->x + 1; + x2 += dPriv->x; min_x += dPriv->x; - max_x += dPriv->x + 1; + max_x += dPriv->x; y1 += dPriv->y; - y2 += dPriv->y + 1; + y2 += dPriv->y; min_y += dPriv->y; - max_y += dPriv->y + 1; + max_y += dPriv->y; } rmesa->state.scissor.rect.x1 = CLAMP(x1, min_x, max_x); @@ -1345,5 +1345,5 @@ void rcommonBeginBatch(radeonContextPtr rmesa, int n, void radeonUserClear(GLcontext *ctx, GLuint mask) { - _mesa_meta_clear(ctx, mask); + _mesa_meta_Clear(ctx, mask); } diff --git a/src/mesa/drivers/dri/radeon/radeon_common_context.c b/src/mesa/drivers/dri/radeon/radeon_common_context.c index 71ee06d9a79..6b9b1e3c5e4 100644 --- a/src/mesa/drivers/dri/radeon/radeon_common_context.c +++ b/src/mesa/drivers/dri/radeon/radeon_common_context.c @@ -47,7 +47,7 @@ WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. #include "swrast_setup/swrast_setup.h" #include "tnl/tnl.h" -#if RADEON_COMMON && defined(RADEON_COMMON_FOR_R600) /* +r6/r7 */ +#if defined(RADEON_R600) #include "r600_context.h" #endif @@ -227,11 +227,8 @@ GLboolean radeonInitContext(radeonContextPtr radeon, fthrottle_mode = driQueryOptioni(&radeon->optionCache, "fthrottle_mode"); radeon->iw.irq_seq = -1; radeon->irqsEmitted = 0; - if (IS_R600_CLASS(radeon->radeonScreen)) - radeon->do_irqs = 0; - else - radeon->do_irqs = (fthrottle_mode == DRI_CONF_FTHROTTLE_IRQS && - radeon->radeonScreen->irq); + radeon->do_irqs = (fthrottle_mode == DRI_CONF_FTHROTTLE_IRQS && + radeon->radeonScreen->irq); radeon->do_usleeps = (fthrottle_mode == DRI_CONF_FTHROTTLE_USLEEPS); diff --git a/src/mesa/drivers/dri/radeon/radeon_debug.c b/src/mesa/drivers/dri/radeon/radeon_debug.c index 3b6f0038037..413000b6c06 100644 --- a/src/mesa/drivers/dri/radeon/radeon_debug.c +++ b/src/mesa/drivers/dri/radeon/radeon_debug.c @@ -39,7 +39,7 @@ static const struct dri_debug_control debug_control[] = { {"fall", RADEON_FALLBACKS}, {"tex", RADEON_TEXTURE}, {"ioctl", RADEON_IOCTL}, - {"verts", RADEON_RENDER}, + {"verts", RADEON_VERTS}, {"render", RADEON_RENDER}, {"swrender", RADEON_SWRENDER}, {"state", RADEON_STATE}, diff --git a/src/mesa/drivers/dri/radeon/radeon_debug.h b/src/mesa/drivers/dri/radeon/radeon_debug.h index 2a8302293b2..26da31c1c48 100644 --- a/src/mesa/drivers/dri/radeon/radeon_debug.h +++ b/src/mesa/drivers/dri/radeon/radeon_debug.h @@ -149,21 +149,22 @@ static inline void radeon_debug_remove_indent(void) } } + /* From http://gcc. gnu.org/onlinedocs/gcc-3.2.3/gcc/Variadic-Macros.html . I suppose we could inline this and use macro to fetch out __LINE__ and stuff in case we run into trouble with other compilers ... GLUE! */ -#define WARN_ONCE(a, ...) { \ - static int warn##__LINE__=1; \ - if(warn##__LINE__){ \ +#define WARN_ONCE(a, ...) do { \ + static int __warn_once=1; \ + if(__warn_once){ \ radeon_warning("*********************************WARN_ONCE*********************************\n"); \ radeon_warning("File %s function %s line %d\n", \ __FILE__, __FUNCTION__, __LINE__); \ radeon_warning( (a), ## __VA_ARGS__);\ radeon_warning("***************************************************************************\n"); \ - warn##__LINE__=0;\ + __warn_once=0;\ } \ - } + } while(0) #endif diff --git a/src/mesa/drivers/dri/radeon/radeon_dma.c b/src/mesa/drivers/dri/radeon/radeon_dma.c index 2eefa3f2b1b..c6edbae9a1b 100644 --- a/src/mesa/drivers/dri/radeon/radeon_dma.c +++ b/src/mesa/drivers/dri/radeon/radeon_dma.c @@ -207,7 +207,6 @@ again_alloc: counter on unused buffers for later freeing them from begin of list */ dma_bo = last_elem(&rmesa->dma.free); - assert(dma_bo->bo->cref == 1); remove_from_list(dma_bo); insert_at_head(&rmesa->dma.reserved, dma_bo); } @@ -263,7 +262,7 @@ void radeonAllocDmaRegion(radeonContextPtr rmesa, void radeonFreeDmaRegions(radeonContextPtr rmesa) { - struct radeon_dma_bo *dma_bo = CALLOC_STRUCT(radeon_dma_bo); + struct radeon_dma_bo *dma_bo; struct radeon_dma_bo *temp; if (RADEON_DEBUG & RADEON_DMA) fprintf(stderr, "%s\n", __FUNCTION__); @@ -307,6 +306,10 @@ static int radeon_bo_is_idle(struct radeon_bo* bo) WARN_ONCE("Your libdrm or kernel doesn't have support for busy query.\n" "This may cause small performance drop for you.\n"); } + /* Protect against bug in legacy bo handling that causes bos stay + * referenced even after they should be freed */ + if (bo->cref != 1) + return 0; return ret != -EBUSY; } @@ -343,7 +346,9 @@ void radeonReleaseDmaRegions(radeonContextPtr rmesa) foreach_s(dma_bo, temp, &rmesa->dma.wait) { if (dma_bo->expire_counter == time) { WARN_ONCE("Leaking dma buffer object!\n"); - radeon_bo_unref(dma_bo->bo); + /* force free of buffer so we don't realy start + * leaking stuff now*/ + while ((dma_bo->bo = radeon_bo_unref(dma_bo->bo))) {} remove_from_list(dma_bo); FREE(dma_bo); continue; diff --git a/src/mesa/drivers/dri/radeon/radeon_fbo.c b/src/mesa/drivers/dri/radeon/radeon_fbo.c index d83b166742c..7ac53ec0ca3 100644 --- a/src/mesa/drivers/dri/radeon/radeon_fbo.c +++ b/src/mesa/drivers/dri/radeon/radeon_fbo.c @@ -583,7 +583,7 @@ void radeon_fbo_init(struct radeon_context *radeon) radeon->glCtx->Driver.FinishRenderTexture = radeon_finish_render_texture; radeon->glCtx->Driver.ResizeBuffers = radeon_resize_buffers; radeon->glCtx->Driver.ValidateFramebuffer = radeon_validate_framebuffer; - radeon->glCtx->Driver.BlitFramebuffer = _mesa_meta_blit_framebuffer; + radeon->glCtx->Driver.BlitFramebuffer = _mesa_meta_BlitFramebuffer; } diff --git a/src/mesa/drivers/dri/radeon/radeon_screen.c b/src/mesa/drivers/dri/radeon/radeon_screen.c index 5ffb55db5ef..573eb6c9c18 100644 --- a/src/mesa/drivers/dri/radeon/radeon_screen.c +++ b/src/mesa/drivers/dri/radeon/radeon_screen.c @@ -48,17 +48,17 @@ WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. #include "radeon_screen.h" #include "radeon_common.h" #include "radeon_span.h" -#if !RADEON_COMMON +#if defined(RADEON_R100) #include "radeon_context.h" #include "radeon_tex.h" -#elif RADEON_COMMON && defined(RADEON_COMMON_FOR_R200) +#elif defined(RADEON_R200) #include "r200_context.h" #include "r200_ioctl.h" #include "r200_tex.h" -#elif RADEON_COMMON && defined(RADEON_COMMON_FOR_R300) +#elif defined(RADEON_R300) #include "r300_context.h" #include "r300_tex.h" -#elif RADEON_COMMON && defined(RADEON_COMMON_FOR_R600) +#elif defined(RADEON_R600) #include "r600_context.h" #include "r700_driconf.h" /* +r6/r7 */ #include "r600_tex.h" /* +r6/r7 */ @@ -82,7 +82,7 @@ DRI_CONF_OPT_BEGIN_V(command_buffer_size,int,def, # min ":" # max ) \ DRI_CONF_DESC(de,"Grösse des Befehlspuffers (in KB)") \ DRI_CONF_OPT_END -#if !RADEON_COMMON /* R100 */ +#if defined(RADEON_R100) /* R100 */ PUBLIC const char __driConfigOptions[] = DRI_CONF_BEGIN DRI_CONF_SECTION_PERFORMANCE @@ -109,7 +109,7 @@ DRI_CONF_BEGIN DRI_CONF_END; static const GLuint __driNConfigOptions = 15; -#elif RADEON_COMMON && defined(RADEON_COMMON_FOR_R200) +#elif defined(RADEON_R200) PUBLIC const char __driConfigOptions[] = DRI_CONF_BEGIN @@ -147,7 +147,7 @@ extern const struct dri_extension NV_vp_extension[]; extern const struct dri_extension ATI_fs_extension[]; extern const struct dri_extension point_extensions[]; -#elif RADEON_COMMON && (defined(RADEON_COMMON_FOR_R300) || defined(RADEON_COMMON_FOR_R600)) +#elif defined(RADEON_R300) || defined(RADEON_R600) #define DRI_CONF_FP_OPTIMIZATION_SPEED 0 #define DRI_CONF_FP_OPTIMIZATION_QUALITY 1 @@ -220,7 +220,7 @@ static const GLuint __driNConfigOptions = 17; extern const struct dri_extension gl_20_extension[]; -#endif /* RADEON_COMMON && defined(RADEON_COMMON_FOR_R300) */ +#endif extern const struct dri_extension card_extensions[]; extern const struct dri_extension mm_extensions[]; @@ -337,7 +337,7 @@ radeonFillInModes( __DRIscreenPrivate *psp, return (const __DRIconfig **) configs; } -#if !RADEON_COMMON +#if defined(RADEON_R100) static const __DRItexOffsetExtension radeonTexOffsetExtension = { { __DRI_TEX_OFFSET, __DRI_TEX_OFFSET_VERSION }, radeonSetTexOffset, @@ -350,7 +350,7 @@ static const __DRItexBufferExtension radeonTexBufferExtension = { }; #endif -#if RADEON_COMMON && defined(RADEON_COMMON_FOR_R200) +#if defined(RADEON_R200) static const __DRIallocateExtension r200AllocateExtension = { { __DRI_ALLOCATE, __DRI_ALLOCATE_VERSION }, r200AllocateMemoryMESA, @@ -370,7 +370,7 @@ static const __DRItexBufferExtension r200TexBufferExtension = { }; #endif -#if RADEON_COMMON && defined(RADEON_COMMON_FOR_R300) +#if defined(RADEON_R300) static const __DRItexOffsetExtension r300texOffsetExtension = { { __DRI_TEX_OFFSET, __DRI_TEX_OFFSET_VERSION }, r300SetTexOffset, @@ -383,7 +383,7 @@ static const __DRItexBufferExtension r300TexBufferExtension = { }; #endif -#if RADEON_COMMON && defined(RADEON_COMMON_FOR_R600) +#if defined(RADEON_R600) static const __DRItexOffsetExtension r600texOffsetExtension = { { __DRI_TEX_OFFSET, __DRI_TEX_OFFSET_VERSION }, r600SetTexOffset, /* +r6/r7 */ @@ -1222,22 +1222,22 @@ radeonCreateScreen( __DRIscreenPrivate *sPriv ) screen->extensions[i++] = &driMediaStreamCounterExtension.base; } -#if !RADEON_COMMON +#if defined(RADEON_R100) screen->extensions[i++] = &radeonTexOffsetExtension.base; #endif -#if RADEON_COMMON && defined(RADEON_COMMON_FOR_R200) +#if defined(RADEON_R200) if (IS_R200_CLASS(screen)) screen->extensions[i++] = &r200AllocateExtension.base; screen->extensions[i++] = &r200texOffsetExtension.base; #endif -#if RADEON_COMMON && defined(RADEON_COMMON_FOR_R300) +#if defined(RADEON_R300) screen->extensions[i++] = &r300texOffsetExtension.base; #endif -#if RADEON_COMMON && defined(RADEON_COMMON_FOR_R600) +#if defined(RADEON_R600) screen->extensions[i++] = &r600texOffsetExtension.base; #endif @@ -1376,22 +1376,22 @@ radeonCreateScreen2(__DRIscreenPrivate *sPriv) screen->extensions[i++] = &driMediaStreamCounterExtension.base; } -#if !RADEON_COMMON +#if defined(RADEON_R100) screen->extensions[i++] = &radeonTexBufferExtension.base; #endif -#if RADEON_COMMON && defined(RADEON_COMMON_FOR_R200) +#if defined(RADEON_R200) if (IS_R200_CLASS(screen)) screen->extensions[i++] = &r200AllocateExtension.base; screen->extensions[i++] = &r200TexBufferExtension.base; #endif -#if RADEON_COMMON && defined(RADEON_COMMON_FOR_R300) +#if defined(RADEON_R300) screen->extensions[i++] = &r300TexBufferExtension.base; #endif -#if RADEON_COMMON && defined(RADEON_COMMON_FOR_R600) +#if defined(RADEON_R600) screen->extensions[i++] = &r600TexBufferExtension.base; #endif @@ -1589,22 +1589,22 @@ radeonDestroyBuffer(__DRIdrawablePrivate *driDrawPriv) static const __DRIconfig ** radeonInitScreen(__DRIscreenPrivate *psp) { -#if !RADEON_COMMON +#if defined(RADEON_R100) static const char *driver_name = "Radeon"; static const __DRIutilversion2 ddx_expected = { 4, 5, 0, 0 }; static const __DRIversion dri_expected = { 4, 0, 0 }; static const __DRIversion drm_expected = { 1, 6, 0 }; -#elif RADEON_COMMON && defined(RADEON_COMMON_FOR_R200) +#elif defined(RADEON_R200) static const char *driver_name = "R200"; static const __DRIutilversion2 ddx_expected = { 4, 5, 0, 0 }; static const __DRIversion dri_expected = { 4, 0, 0 }; static const __DRIversion drm_expected = { 1, 6, 0 }; -#elif RADEON_COMMON && defined(RADEON_COMMON_FOR_R300) +#elif defined(RADEON_R300) static const char *driver_name = "R300"; static const __DRIutilversion2 ddx_expected = { 4, 5, 0, 0 }; static const __DRIversion dri_expected = { 4, 0, 0 }; static const __DRIversion drm_expected = { 1, 24, 0 }; -#elif RADEON_COMMON && defined(RADEON_COMMON_FOR_R600) +#elif defined(RADEON_R600) static const char *driver_name = "R600"; static const __DRIutilversion2 ddx_expected = { 4, 5, 0, 0 }; static const __DRIversion dri_expected = { 4, 0, 0 }; @@ -1630,13 +1630,13 @@ radeonInitScreen(__DRIscreenPrivate *psp) * Hello chicken. Hello egg. How are you two today? */ driInitExtensions( NULL, card_extensions, GL_FALSE ); -#if RADEON_COMMON && defined(RADEON_COMMON_FOR_R200) +#if defined(RADEON_R200) driInitExtensions( NULL, blend_extensions, GL_FALSE ); driInitSingleExtension( NULL, ARB_vp_extension ); driInitSingleExtension( NULL, NV_vp_extension ); driInitSingleExtension( NULL, ATI_fs_extension ); driInitExtensions( NULL, point_extensions, GL_FALSE ); -#elif (defined(RADEON_COMMON_FOR_R300) || defined(RADEON_COMMON_FOR_R600)) +#elif (defined(RADEON_R300) || defined(RADEON_R600)) driInitSingleExtension( NULL, gl_20_extension ); #endif @@ -1684,13 +1684,13 @@ __DRIconfig **radeonInitScreen2(__DRIscreenPrivate *psp) */ driInitExtensions( NULL, card_extensions, GL_FALSE ); driInitExtensions( NULL, mm_extensions, GL_FALSE ); -#if RADEON_COMMON && defined(RADEON_COMMON_FOR_R200) +#if defined(RADEON_R200) driInitExtensions( NULL, blend_extensions, GL_FALSE ); driInitSingleExtension( NULL, ARB_vp_extension ); driInitSingleExtension( NULL, NV_vp_extension ); driInitSingleExtension( NULL, ATI_fs_extension ); driInitExtensions( NULL, point_extensions, GL_FALSE ); -#elif (defined(RADEON_COMMON_FOR_R300) || defined(RADEON_COMMON_FOR_R600)) +#elif (defined(RADEON_R300) || defined(RADEON_R600)) driInitSingleExtension( NULL, gl_20_extension ); #endif @@ -1772,13 +1772,13 @@ getSwapInfo( __DRIdrawablePrivate *dPriv, __DRIswapInfo * sInfo ) const struct __DriverAPIRec driDriverAPI = { .InitScreen = radeonInitScreen, .DestroyScreen = radeonDestroyScreen, -#if RADEON_COMMON && defined(RADEON_COMMON_FOR_R200) +#if defined(RADEON_R200) .CreateContext = r200CreateContext, .DestroyContext = r200DestroyContext, -#elif RADEON_COMMON && defined(RADEON_COMMON_FOR_R600) +#elif defined(RADEON_R600) .CreateContext = r600CreateContext, .DestroyContext = radeonDestroyContext, -#elif RADEON_COMMON && defined(RADEON_COMMON_FOR_R300) +#elif defined(RADEON_R300) .CreateContext = r300CreateContext, .DestroyContext = radeonDestroyContext, #else diff --git a/src/mesa/drivers/dri/radeon/radeon_span.c b/src/mesa/drivers/dri/radeon/radeon_span.c index 4e100d854ed..0c49c3713af 100644 --- a/src/mesa/drivers/dri/radeon/radeon_span.c +++ b/src/mesa/drivers/dri/radeon/radeon_span.c @@ -55,7 +55,7 @@ static void radeonSetSpanFunctions(struct radeon_renderbuffer *rrb); /* r200 depth buffer is always tiled - this is the formula according to the docs unless I typo'ed in it */ -#if defined(RADEON_COMMON_FOR_R200) +#if defined(RADEON_R200) static GLubyte *r200_depth_2byte(const struct radeon_renderbuffer * rrb, GLint x, GLint y) { @@ -106,6 +106,141 @@ static GLubyte *r200_depth_4byte(const struct radeon_renderbuffer * rrb, } #endif +/* r600 tiling + * two main types: + * - 1D (akin to macro-linear/micro-tiled on older asics) + * - 2D (akin to macro-tiled/micro-tiled on older asics) + * only 1D tiling is implemented below + */ +#if defined(RADEON_R600) +static inline GLint r600_1d_tile_helper(const struct radeon_renderbuffer * rrb, + GLint x, GLint y, GLint is_depth, GLint is_stencil) +{ + GLint element_bytes = rrb->cpp; + GLint num_samples = 1; + GLint tile_width = 8; + GLint tile_height = 8; + GLint tile_thickness = 1; + GLint pitch_elements = rrb->pitch / element_bytes; + GLint height = rrb->base.Height; + GLint z = 0; + GLint sample_number = 0; + /* */ + GLint tile_bytes; + GLint tiles_per_row; + GLint tiles_per_slice; + GLint slice_offset; + GLint tile_row_index; + GLint tile_column_index; + GLint tile_offset; + GLint pixel_number = 0; + GLint element_offset; + GLint offset = 0; + + tile_bytes = tile_width * tile_height * tile_thickness * element_bytes * num_samples; + tiles_per_row = pitch_elements / tile_width; + tiles_per_slice = tiles_per_row * (height / tile_height); + slice_offset = (z / tile_thickness) * tiles_per_slice * tile_bytes; + tile_row_index = y / tile_height; + tile_column_index = x / tile_width; + tile_offset = ((tile_row_index * tiles_per_row) + tile_column_index) * tile_bytes; + + if (is_depth) { + GLint pixel_offset = 0; + + pixel_number |= ((x >> 0) & 1) << 0; // pn[0] = x[0] + pixel_number |= ((y >> 0) & 1) << 1; // pn[1] = y[0] + pixel_number |= ((x >> 1) & 1) << 2; // pn[2] = x[1] + pixel_number |= ((y >> 1) & 1) << 3; // pn[3] = y[1] + pixel_number |= ((x >> 2) & 1) << 4; // pn[4] = x[2] + pixel_number |= ((y >> 2) & 1) << 5; // pn[5] = y[2] + switch (element_bytes) { + case 2: + pixel_offset = pixel_number * element_bytes * num_samples; + break; + case 4: + /* stencil and depth data are stored separately within a tile. + * stencil is stored in a contiguous tile before the depth tile. + * stencil element is 1 byte, depth element is 3 bytes. + * stencil tile is 64 bytes. + */ + if (is_stencil) + pixel_offset = pixel_number * 1 * num_samples; + else + pixel_offset = (pixel_number * 3 * num_samples) + 64; + break; + } + element_offset = pixel_offset + (sample_number * element_bytes); + } else { + GLint sample_offset; + + switch (element_bytes) { + case 1: + pixel_number |= ((x >> 0) & 1) << 0; // pn[0] = x[0] + pixel_number |= ((x >> 1) & 1) << 1; // pn[1] = x[1] + pixel_number |= ((x >> 2) & 1) << 2; // pn[2] = x[2] + pixel_number |= ((y >> 1) & 1) << 3; // pn[3] = y[1] + pixel_number |= ((y >> 0) & 1) << 4; // pn[4] = y[0] + pixel_number |= ((y >> 2) & 1) << 5; // pn[5] = y[2] + break; + case 2: + pixel_number |= ((x >> 0) & 1) << 0; // pn[0] = x[0] + pixel_number |= ((x >> 1) & 1) << 1; // pn[1] = x[1] + pixel_number |= ((x >> 2) & 1) << 2; // pn[2] = x[2] + pixel_number |= ((y >> 0) & 1) << 3; // pn[3] = y[0] + pixel_number |= ((y >> 1) & 1) << 4; // pn[4] = y[1] + pixel_number |= ((y >> 2) & 1) << 5; // pn[5] = y[2] + break; + case 4: + pixel_number |= ((x >> 0) & 1) << 0; // pn[0] = x[0] + pixel_number |= ((x >> 1) & 1) << 1; // pn[1] = x[1] + pixel_number |= ((y >> 0) & 1) << 2; // pn[2] = y[0] + pixel_number |= ((x >> 2) & 1) << 3; // pn[3] = x[2] + pixel_number |= ((y >> 1) & 1) << 4; // pn[4] = y[1] + pixel_number |= ((y >> 2) & 1) << 5; // pn[5] = y[2] + break; + } + sample_offset = sample_number * (tile_bytes / num_samples); + element_offset = sample_offset + (pixel_number * element_bytes); + } + offset = slice_offset + tile_offset + element_offset; + return offset; +} + +/* depth buffers */ +static GLubyte *r600_ptr_depth(const struct radeon_renderbuffer * rrb, + GLint x, GLint y) +{ + GLubyte *ptr = rrb->bo->ptr; + GLint offset = r600_1d_tile_helper(rrb, x, y, 1, 0); + return &ptr[offset]; +} + +static GLubyte *r600_ptr_stencil(const struct radeon_renderbuffer * rrb, + GLint x, GLint y) +{ + GLubyte *ptr = rrb->bo->ptr; + GLint offset = r600_1d_tile_helper(rrb, x, y, 1, 1); + return &ptr[offset]; +} + +static GLubyte *r600_ptr_color(const struct radeon_renderbuffer * rrb, + GLint x, GLint y) +{ + GLubyte *ptr = rrb->bo->ptr; + uint32_t mask = RADEON_BO_FLAGS_MACRO_TILE | RADEON_BO_FLAGS_MICRO_TILE; + GLint offset; + + if (rrb->has_surface || !(rrb->bo->flags & mask)) { + offset = x * rrb->cpp + y * rrb->pitch; + } else { + offset = r600_1d_tile_helper(rrb, x, y, 0, 0); + } + return &ptr[offset]; +} + +#else + /* radeon tiling on r300-r500 has 4 states, macro-linear/micro-linear macro-linear/micro-tiled @@ -197,7 +332,10 @@ static GLubyte *radeon_ptr_2byte_8x2(const struct radeon_renderbuffer * rrb, return &ptr[offset]; } -#ifndef COMPILE_R300 +#endif + +#ifndef RADEON_R300 +#ifndef RADEON_R600 static uint32_t z24s8_to_s8z24(uint32_t val) { @@ -210,6 +348,7 @@ s8z24_to_z24s8(uint32_t val) return (val >> 24) | (val << 8); } #endif +#endif /* * Note that all information needed to access pixels in a renderbuffer @@ -270,7 +409,11 @@ s8z24_to_z24s8(uint32_t val) #define TAG(x) radeon##x##_RGB565 #define TAG2(x,y) radeon##x##_RGB565##y +#if defined(RADEON_R600) +#define GET_PTR(X,Y) r600_ptr_color(rrb, (X) + x_off, (Y) + y_off) +#else #define GET_PTR(X,Y) radeon_ptr_2byte_8x2(rrb, (X) + x_off, (Y) + y_off) +#endif #include "spantmp2.h" /* 16 bit, ARGB1555 color spanline and pixel functions @@ -280,7 +423,11 @@ s8z24_to_z24s8(uint32_t val) #define TAG(x) radeon##x##_ARGB1555 #define TAG2(x,y) radeon##x##_ARGB1555##y +#if defined(RADEON_R600) +#define GET_PTR(X,Y) r600_ptr_color(rrb, (X) + x_off, (Y) + y_off) +#else #define GET_PTR(X,Y) radeon_ptr_2byte_8x2(rrb, (X) + x_off, (Y) + y_off) +#endif #include "spantmp2.h" /* 16 bit, RGBA4 color spanline and pixel functions @@ -290,7 +437,11 @@ s8z24_to_z24s8(uint32_t val) #define TAG(x) radeon##x##_ARGB4444 #define TAG2(x,y) radeon##x##_ARGB4444##y +#if defined(RADEON_R600) +#define GET_PTR(X,Y) r600_ptr_color(rrb, (X) + x_off, (Y) + y_off) +#else #define GET_PTR(X,Y) radeon_ptr_2byte_8x2(rrb, (X) + x_off, (Y) + y_off) +#endif #include "spantmp2.h" /* 32 bit, xRGB8888 color spanline and pixel functions @@ -300,11 +451,19 @@ s8z24_to_z24s8(uint32_t val) #define TAG(x) radeon##x##_xRGB8888 #define TAG2(x,y) radeon##x##_xRGB8888##y +#if defined(RADEON_R600) +#define GET_VALUE(_x, _y) ((*(GLuint*)(r600_ptr_color(rrb, _x + x_off, _y + y_off)) | 0xff000000)) +#define PUT_VALUE(_x, _y, d) { \ + GLuint *_ptr = (GLuint*)r600_ptr_color( rrb, _x + x_off, _y + y_off ); \ + *_ptr = d; \ +} while (0) +#else #define GET_VALUE(_x, _y) ((*(GLuint*)(radeon_ptr_4byte(rrb, _x + x_off, _y + y_off)) | 0xff000000)) #define PUT_VALUE(_x, _y, d) { \ GLuint *_ptr = (GLuint*)radeon_ptr_4byte( rrb, _x + x_off, _y + y_off ); \ *_ptr = d; \ } while (0) +#endif #include "spantmp2.h" /* 32 bit, ARGB8888 color spanline and pixel functions @@ -314,11 +473,19 @@ s8z24_to_z24s8(uint32_t val) #define TAG(x) radeon##x##_ARGB8888 #define TAG2(x,y) radeon##x##_ARGB8888##y +#if defined(RADEON_R600) +#define GET_VALUE(_x, _y) (*(GLuint*)(r600_ptr_color(rrb, _x + x_off, _y + y_off))) +#define PUT_VALUE(_x, _y, d) { \ + GLuint *_ptr = (GLuint*)r600_ptr_color( rrb, _x + x_off, _y + y_off ); \ + *_ptr = d; \ +} while (0) +#else #define GET_VALUE(_x, _y) (*(GLuint*)(radeon_ptr_4byte(rrb, _x + x_off, _y + y_off))) #define PUT_VALUE(_x, _y, d) { \ GLuint *_ptr = (GLuint*)radeon_ptr_4byte( rrb, _x + x_off, _y + y_off ); \ *_ptr = d; \ } while (0) +#endif #include "spantmp2.h" /* ================================================================ @@ -339,17 +506,23 @@ s8z24_to_z24s8(uint32_t val) */ #define VALUE_TYPE GLushort -#if defined(RADEON_COMMON_FOR_R200) +#if defined(RADEON_R200) #define WRITE_DEPTH( _x, _y, d ) \ *(GLushort *)r200_depth_2byte(rrb, _x + x_off, _y + y_off) = d +#elif defined(RADEON_R600) +#define WRITE_DEPTH( _x, _y, d ) \ + *(GLushort *)r600_ptr_depth(rrb, _x + x_off, _y + y_off) = d #else #define WRITE_DEPTH( _x, _y, d ) \ *(GLushort *)radeon_ptr_2byte_8x2(rrb, _x + x_off, _y + y_off) = d #endif -#if defined(RADEON_COMMON_FOR_R200) +#if defined(RADEON_R200) #define READ_DEPTH( d, _x, _y ) \ d = *(GLushort *)r200_depth_2byte(rrb, _x + x_off, _y + y_off) +#elif defined(RADEON_R600) +#define READ_DEPTH( d, _x, _y ) \ + d = *(GLushort *)r600_ptr_depth(rrb, _x + x_off, _y + y_off) #else #define READ_DEPTH( d, _x, _y ) \ d = *(GLushort *)radeon_ptr_2byte_8x2(rrb, _x + x_off, _y + y_off) @@ -365,7 +538,7 @@ s8z24_to_z24s8(uint32_t val) */ #define VALUE_TYPE GLuint -#if defined(COMPILE_R300) +#if defined(RADEON_R300) #define WRITE_DEPTH( _x, _y, d ) \ do { \ GLuint *_ptr = (GLuint*)radeon_ptr_4byte( rrb, _x + x_off, _y + y_off ); \ @@ -374,7 +547,16 @@ do { \ tmp |= ((d << 8) & 0xffffff00); \ *_ptr = tmp; \ } while (0) -#elif defined(RADEON_COMMON_FOR_R200) +#elif defined(RADEON_R600) +#define WRITE_DEPTH( _x, _y, d ) \ +do { \ + GLuint *_ptr = (GLuint*)r600_ptr_depth( rrb, _x + x_off, _y + y_off ); \ + GLuint tmp = *_ptr; \ + tmp &= 0xff000000; \ + tmp |= ((d) & 0x00ffffff); \ + *_ptr = tmp; \ +} while (0) +#elif defined(RADEON_R200) #define WRITE_DEPTH( _x, _y, d ) \ do { \ GLuint *_ptr = (GLuint*)r200_depth_4byte( rrb, _x + x_off, _y + y_off ); \ @@ -394,12 +576,17 @@ do { \ } while (0) #endif -#if defined(COMPILE_R300) +#if defined(RADEON_R300) #define READ_DEPTH( d, _x, _y ) \ do { \ d = (*(GLuint*)(radeon_ptr_4byte(rrb, _x + x_off, _y + y_off)) & 0xffffff00) >> 8; \ }while(0) -#elif defined(RADEON_COMMON_FOR_R200) +#elif defined(RADEON_R600) +#define READ_DEPTH( d, _x, _y ) \ + do { \ + d = (*(GLuint*)(r600_ptr_depth(rrb, _x + x_off, _y + y_off)) & 0x00ffffff); \ + }while(0) +#elif defined(RADEON_R200) #define READ_DEPTH( d, _x, _y ) \ do { \ d = *(GLuint*)(r200_depth_4byte(rrb, _x + x_off, _y + y_off)) & 0x00ffffff; \ @@ -420,13 +607,27 @@ do { \ */ #define VALUE_TYPE GLuint -#if defined(COMPILE_R300) +#if defined(RADEON_R300) #define WRITE_DEPTH( _x, _y, d ) \ do { \ GLuint *_ptr = (GLuint*)radeon_ptr_4byte( rrb, _x + x_off, _y + y_off ); \ *_ptr = d; \ } while (0) -#elif defined(RADEON_COMMON_FOR_R200) +#elif defined(RADEON_R600) +#define WRITE_DEPTH( _x, _y, d ) \ +do { \ + GLuint *_ptr = (GLuint*)r600_ptr_depth( rrb, _x + x_off, _y + y_off ); \ + GLuint tmp = *_ptr; \ + tmp &= 0xff000000; \ + tmp |= (((d) >> 8) & 0x00ffffff); \ + *_ptr = tmp; \ + _ptr = (GLuint*)r600_ptr_stencil(rrb, _x + x_off, _y + y_off); \ + tmp = *_ptr; \ + tmp &= 0xffffff00; \ + tmp |= (d) & 0xff; \ + *_ptr = tmp; \ +} while (0) +#elif defined(RADEON_R200) #define WRITE_DEPTH( _x, _y, d ) \ do { \ GLuint *_ptr = (GLuint*)r200_depth_4byte( rrb, _x + x_off, _y + y_off ); \ @@ -442,12 +643,18 @@ do { \ } while (0) #endif -#if defined(COMPILE_R300) +#if defined(RADEON_R300) #define READ_DEPTH( d, _x, _y ) \ do { \ d = (*(GLuint*)(radeon_ptr_4byte(rrb, _x + x_off, _y + y_off))); \ }while(0) -#elif defined(RADEON_COMMON_FOR_R200) +#elif defined(RADEON_R600) +#define READ_DEPTH( d, _x, _y ) \ + do { \ + d = ((*(GLuint*)(r600_ptr_depth(rrb, _x + x_off, _y + y_off))) << 8) & 0xffffff00; \ + d |= (*(GLuint*)(r600_ptr_stencil(rrb, _x + x_off, _y + y_off))) & 0x000000ff; \ + }while(0) +#elif defined(RADEON_R200) #define READ_DEPTH( d, _x, _y ) \ do { \ d = s8z24_to_z24s8(*(GLuint*)(r200_depth_4byte(rrb, _x + x_off, _y + y_off))); \ @@ -467,7 +674,7 @@ do { \ /* 24 bit depth, 8 bit stencil depthbuffer functions */ -#ifdef COMPILE_R300 +#ifdef RADEON_R300 #define WRITE_STENCIL( _x, _y, d ) \ do { \ GLuint *_ptr = (GLuint*)radeon_ptr_4byte(rrb, _x + x_off, _y + y_off); \ @@ -476,7 +683,16 @@ do { \ tmp |= (d) & 0xff; \ *_ptr = tmp; \ } while (0) -#elif defined(RADEON_COMMON_FOR_R200) +#elif defined(RADEON_R600) +#define WRITE_STENCIL( _x, _y, d ) \ +do { \ + GLuint *_ptr = (GLuint*)r600_ptr_stencil(rrb, _x + x_off, _y + y_off); \ + GLuint tmp = *_ptr; \ + tmp &= 0xffffff00; \ + tmp |= (d) & 0xff; \ + *_ptr = tmp; \ +} while (0) +#elif defined(RADEON_R200) #define WRITE_STENCIL( _x, _y, d ) \ do { \ GLuint *_ptr = (GLuint*)r200_depth_4byte(rrb, _x + x_off, _y + y_off); \ @@ -496,14 +712,21 @@ do { \ } while (0) #endif -#ifdef COMPILE_R300 +#ifdef RADEON_R300 #define READ_STENCIL( d, _x, _y ) \ do { \ GLuint *_ptr = (GLuint*)radeon_ptr_4byte( rrb, _x + x_off, _y + y_off ); \ GLuint tmp = *_ptr; \ d = tmp & 0x000000ff; \ } while (0) -#elif defined(RADEON_COMMON_FOR_R200) +#elif defined(RADEON_R600) +#define READ_STENCIL( d, _x, _y ) \ +do { \ + GLuint *_ptr = (GLuint*)r600_ptr_stencil( rrb, _x + x_off, _y + y_off ); \ + GLuint tmp = *_ptr; \ + d = tmp & 0x000000ff; \ +} while (0) +#elif defined(RADEON_R200) #define READ_STENCIL( d, _x, _y ) \ do { \ GLuint *_ptr = (GLuint*)r200_depth_4byte( rrb, _x + x_off, _y + y_off ); \ diff --git a/src/mesa/drivers/dri/radeon/radeon_texstate.c b/src/mesa/drivers/dri/radeon/radeon_texstate.c index 9d252aa74c5..ae41b90efe5 100644 --- a/src/mesa/drivers/dri/radeon/radeon_texstate.c +++ b/src/mesa/drivers/dri/radeon/radeon_texstate.c @@ -833,11 +833,14 @@ static void import_tex_obj_state( r100ContextPtr rmesa, cmd[TEX_PP_TXFORMAT] |= texobj->pp_txformat & TEXOBJ_TXFORMAT_MASK; cmd[TEX_PP_BORDER_COLOR] = texobj->pp_border_color; - if (texobj->base.Target == GL_TEXTURE_RECTANGLE_NV) { - GLuint *txr_cmd = RADEON_DB_STATE( txr[unit] ); + if (texobj->pp_txformat & RADEON_TXFORMAT_NON_POWER2) { + uint32_t *txr_cmd = &rmesa->hw.txr[unit].cmd[TXR_CMD_0]; txr_cmd[TXR_PP_TEX_SIZE] = texobj->pp_txsize; /* NPOT only! */ txr_cmd[TXR_PP_TEX_PITCH] = texobj->pp_txpitch; /* NPOT only! */ - RADEON_DB_STATECHANGE( rmesa, &rmesa->hw.txr[unit] ); + RADEON_STATECHANGE( rmesa, txr[unit] ); + } + + if (texobj->base.Target == GL_TEXTURE_RECTANGLE_NV) { se_coord_fmt |= RADEON_VTX_ST0_NONPARAMETRIC << unit; } else { @@ -1114,7 +1117,6 @@ static GLboolean radeon_validate_texture(GLcontext *ctx, struct gl_texture_objec RADEON_STATECHANGE( rmesa, ctx ); rmesa->hw.ctx.cmd[CTX_PP_CNTL] |= (RADEON_TEX_0_ENABLE | RADEON_TEX_BLEND_0_ENABLE) << unit; - RADEON_STATECHANGE( rmesa, tcl ); rmesa->hw.tcl.cmd[TCL_OUTPUT_VTXFMT] |= RADEON_ST_BIT(unit); diff --git a/src/mesa/drivers/dri/radeon/radeon_texture.c b/src/mesa/drivers/dri/radeon/radeon_texture.c index fad3d1cedaf..049284ef8c5 100644 --- a/src/mesa/drivers/dri/radeon/radeon_texture.c +++ b/src/mesa/drivers/dri/radeon/radeon_texture.c @@ -659,11 +659,6 @@ static void radeon_teximage( if (dims == 3) _mesa_free(dstImageOffsets); } - - /* SGIS_generate_mipmap */ - if (level == texObj->BaseLevel && texObj->GenerateMipmap) { - radeon_generate_mipmap(ctx, target, texObj); - } } _mesa_unmap_teximage_pbo(ctx, packing); @@ -792,11 +787,6 @@ static void radeon_texsubimage(GLcontext* ctx, int dims, GLenum target, int leve format, type, pixels, packing)) _mesa_error(ctx, GL_OUT_OF_MEMORY, "glTexSubImage"); } - - /* GL_SGIS_generate_mipmap */ - if (level == texObj->BaseLevel && texObj->GenerateMipmap) { - radeon_generate_mipmap(ctx, target, texObj); - } } radeon_teximage_unmap(image); diff --git a/src/mesa/drivers/dri/s3v/s3v_tex.c b/src/mesa/drivers/dri/s3v/s3v_tex.c index 9b92519862a..ec1182f34f7 100644 --- a/src/mesa/drivers/dri/s3v/s3v_tex.c +++ b/src/mesa/drivers/dri/s3v/s3v_tex.c @@ -536,24 +536,13 @@ void s3vInitTextureFuncs( GLcontext *ctx ) #endif ctx->Driver.TexEnv = s3vTexEnv; - ctx->Driver.ChooseTextureFormat = _mesa_choose_tex_format; - ctx->Driver.TexImage1D = _mesa_store_teximage1d; ctx->Driver.TexImage2D = s3vTexImage2D; - ctx->Driver.TexImage3D = _mesa_store_teximage3d; - ctx->Driver.TexSubImage1D = _mesa_store_texsubimage1d; ctx->Driver.TexSubImage2D = s3vTexSubImage2D; - ctx->Driver.TexSubImage3D = _mesa_store_texsubimage3d; - ctx->Driver.CopyTexImage1D = _swrast_copy_teximage1d; - ctx->Driver.CopyTexImage2D = _swrast_copy_teximage2d; - ctx->Driver.CopyTexSubImage1D = _swrast_copy_texsubimage1d; - ctx->Driver.CopyTexSubImage2D = _swrast_copy_texsubimage2d; - ctx->Driver.CopyTexSubImage3D = _swrast_copy_texsubimage3d; ctx->Driver.BindTexture = s3vBindTexture; ctx->Driver.DeleteTexture = s3vDeleteTexture; ctx->Driver.TexParameter = s3vTexParameter; ctx->Driver.UpdateTexturePalette = 0; ctx->Driver.IsTextureResident = s3vIsTextureResident; - ctx->Driver.TestProxyTexImage = _mesa_test_proxy_teximage; s3vInitTextureObjects( ctx ); } diff --git a/src/mesa/drivers/dri/swrast/swrast.c b/src/mesa/drivers/dri/swrast/swrast.c index b36cd0274a8..f4947daa063 100644 --- a/src/mesa/drivers/dri/swrast/swrast.c +++ b/src/mesa/drivers/dri/swrast/swrast.c @@ -43,6 +43,7 @@ #include "tnl/t_pipeline.h" #include "vbo/vbo.h" #include "drivers/common/driverfuncs.h" +#include "drivers/common/meta.h" #include "utils.h" #include "swrast_priv.h" @@ -650,6 +651,8 @@ driCreateNewContext(__DRIscreen *screen, const __DRIconfig *config, _mesa_enable_2_0_extensions(mesaCtx); _mesa_enable_2_1_extensions(mesaCtx); + _mesa_meta_init(mesaCtx); + return ctx; } @@ -661,6 +664,7 @@ driDestroyContext(__DRIcontext *ctx) if (ctx) { mesaCtx = &ctx->Base; + _mesa_meta_free(mesaCtx); _swsetup_DestroyContext( mesaCtx ); _swrast_DestroyContext( mesaCtx ); _tnl_DestroyContext( mesaCtx ); diff --git a/src/mesa/drivers/dri/tdfx/tdfx_tex.c b/src/mesa/drivers/dri/tdfx/tdfx_tex.c index 1f7257eaead..f6a48b3ae12 100644 --- a/src/mesa/drivers/dri/tdfx/tdfx_tex.c +++ b/src/mesa/drivers/dri/tdfx/tdfx_tex.c @@ -176,6 +176,54 @@ logbase2(int n) } +static void +tdfxGenerateMipmap(GLcontext *ctx, GLenum target, + struct gl_texture_object *texObj) +{ + GLint mipWidth, mipHeight; + tdfxMipMapLevel *mip; + struct gl_texture_image *mipImage; /* the new/next image */ + struct gl_texture_image *texImage; + const GLint maxLevels = _mesa_max_texture_levels(ctx, texObj->Target); + GLint level = texObj->BaseLevel; + GLsizei width, height, texelBytes; + const tdfxMipMapLevel *mml; + + texImage = _mesa_get_tex_image(ctx, texObj, target, level); + assert(!texImage->IsCompressed); + + mml = TDFX_TEXIMAGE_DATA(texImage); + + width = texImage->Width; + height = texImage->Height; + while (level < texObj->MaxLevel && level < maxLevels - 1) { + mipWidth = width / 2; + if (!mipWidth) { + mipWidth = 1; + } + mipHeight = height / 2; + if (!mipHeight) { + mipHeight = 1; + } + if ((mipWidth == width) && (mipHeight == height)) { + break; + } + ++level; + mipImage = _mesa_select_tex_image(ctx, texObj, target, level); + mip = TDFX_TEXIMAGE_DATA(mipImage); + _mesa_halve2x2_teximage2d(ctx, + texImage, + texelBytes, + mml->width, mml->height, + texImage->Data, mipImage->Data); + texImage = mipImage; + mml = mip; + width = mipWidth; + height = mipHeight; + } +} + + /* * Compute various texture image parameters. * Input: w, h - source texture width and height @@ -1397,45 +1445,6 @@ tdfxTexImage2D(GLcontext *ctx, GLenum target, GLint level, width, height, 1, format, type, pixels, packing); } - - /* GL_SGIS_generate_mipmap */ - if (level == texObj->BaseLevel && texObj->GenerateMipmap) { - GLint mipWidth, mipHeight; - tdfxMipMapLevel *mip; - struct gl_texture_image *mipImage; - const GLint maxLevels = _mesa_max_texture_levels(ctx, texObj->Target); - - assert(!texImage->IsCompressed); - - while (level < texObj->MaxLevel && level < maxLevels - 1) { - mipWidth = width / 2; - if (!mipWidth) { - mipWidth = 1; - } - mipHeight = height / 2; - if (!mipHeight) { - mipHeight = 1; - } - if ((mipWidth == width) && (mipHeight == height)) { - break; - } - _mesa_TexImage2D(target, ++level, internalFormat, - mipWidth, mipHeight, border, - format, type, - NULL); - mipImage = _mesa_select_tex_image(ctx, texObj, target, level); - mip = TDFX_TEXIMAGE_DATA(mipImage); - _mesa_halve2x2_teximage2d(ctx, - texImage, - texelBytes, - mml->width, mml->height, - texImage->Data, mipImage->Data); - texImage = mipImage; - mml = mip; - width = mipWidth; - height = mipHeight; - } - } } RevalidateTexture(ctx, texObj); @@ -1507,44 +1516,6 @@ tdfxTexSubImage2D(GLcontext *ctx, GLenum target, GLint level, format, type, pixels, packing); } - /* GL_SGIS_generate_mipmap */ - if (level == texObj->BaseLevel && texObj->GenerateMipmap) { - GLint mipWidth, mipHeight; - tdfxMipMapLevel *mip; - struct gl_texture_image *mipImage; - const GLint maxLevels = _mesa_max_texture_levels(ctx, texObj->Target); - - assert(!texImage->IsCompressed); - - width = texImage->Width; - height = texImage->Height; - while (level < texObj->MaxLevel && level < maxLevels - 1) { - mipWidth = width / 2; - if (!mipWidth) { - mipWidth = 1; - } - mipHeight = height / 2; - if (!mipHeight) { - mipHeight = 1; - } - if ((mipWidth == width) && (mipHeight == height)) { - break; - } - ++level; - mipImage = _mesa_select_tex_image(ctx, texObj, target, level); - mip = TDFX_TEXIMAGE_DATA(mipImage); - _mesa_halve2x2_teximage2d(ctx, - texImage, - texelBytes, - mml->width, mml->height, - texImage->Data, mipImage->Data); - texImage = mipImage; - mml = mip; - width = mipWidth; - height = mipHeight; - } - } - ti->reloadImages = GL_TRUE; /* signal the image needs to be reloaded */ fxMesa->new_state |= TDFX_NEW_TEXTURE; /* XXX this might be a bit much */ } @@ -1703,11 +1674,6 @@ tdfxCompressedTexImage2D (GLcontext *ctx, GLenum target, MEMCPY(texImage->Data, data, texImage->CompressedSize); } - /* GL_SGIS_generate_mipmap */ - if (level == texObj->BaseLevel && texObj->GenerateMipmap) { - assert(!texImage->IsCompressed); - } - RevalidateTexture(ctx, texObj); ti->reloadImages = GL_TRUE; @@ -1770,11 +1736,6 @@ tdfxCompressedTexSubImage2D( GLcontext *ctx, GLenum target, texImage->Data); } - /* GL_SGIS_generate_mipmap */ - if (level == texObj->BaseLevel && texObj->GenerateMipmap) { - assert(!texImage->IsCompressed); - } - RevalidateTexture(ctx, texObj); ti->reloadImages = GL_TRUE; @@ -1914,4 +1875,5 @@ void tdfxInitTextureFuncs( struct dd_function_table *functions ) functions->CompressedTexImage2D = tdfxCompressedTexImage2D; functions->CompressedTexSubImage2D = tdfxCompressedTexSubImage2D; functions->UpdateTexturePalette = tdfxUpdateTexturePalette; + functions->GenerateMipmap = tdfxGenerateMipmap; } diff --git a/src/mesa/drivers/dri/unichrome/via_tex.c b/src/mesa/drivers/dri/unichrome/via_tex.c index d2010f09074..54073e76918 100644 --- a/src/mesa/drivers/dri/unichrome/via_tex.c +++ b/src/mesa/drivers/dri/unichrome/via_tex.c @@ -818,11 +818,6 @@ static void viaTexImage(GLcontext *ctx, } } - /* GL_SGIS_generate_mipmap */ - if (level == texObj->BaseLevel && texObj->GenerateMipmap) { - _mesa_generate_mipmap(ctx, target, texObj); - } - _mesa_unmap_teximage_pbo(ctx, packing); } diff --git a/src/mesa/drivers/osmesa/osmesa.c b/src/mesa/drivers/osmesa/osmesa.c index 904659e3450..692657a5dfd 100644 --- a/src/mesa/drivers/osmesa/osmesa.c +++ b/src/mesa/drivers/osmesa/osmesa.c @@ -50,6 +50,7 @@ #include "tnl/t_context.h" #include "tnl/t_pipeline.h" #include "drivers/common/driverfuncs.h" +#include "drivers/common/meta.h" #include "vbo/vbo.h" @@ -1258,6 +1259,8 @@ OSMesaCreateContextExt( GLenum format, GLint depthBits, GLint stencilBits, osmesa->bInd = bind; osmesa->aInd = aind; + _mesa_meta_init(&osmesa->mesa); + /* Initialize the software rasterizer and helper modules. */ { GLcontext *ctx = &osmesa->mesa; @@ -1304,6 +1307,8 @@ OSMesaDestroyContext( OSMesaContext osmesa ) if (osmesa->rb) _mesa_reference_renderbuffer(&osmesa->rb, NULL); + _mesa_meta_free( &osmesa->mesa ); + _swsetup_DestroyContext( &osmesa->mesa ); _tnl_DestroyContext( &osmesa->mesa ); _vbo_DestroyContext( &osmesa->mesa ); diff --git a/src/mesa/drivers/windows/gdi/mesa.def b/src/mesa/drivers/windows/gdi/mesa.def index bd3e5b21373..5abcd1d927e 100644 --- a/src/mesa/drivers/windows/gdi/mesa.def +++ b/src/mesa/drivers/windows/gdi/mesa.def @@ -943,6 +943,15 @@ EXPORTS _mesa_update_framebuffer_visual _mesa_use_program _mesa_Viewport + _mesa_meta_CopyColorSubTable + _mesa_meta_CopyColorTable + _mesa_meta_CopyConvolutionFilter1D + _mesa_meta_CopyConvolutionFilter2D + _mesa_meta_CopyTexImage1D + _mesa_meta_CopyTexImage2D + _mesa_meta_CopyTexSubImage1D + _mesa_meta_CopyTexSubImage2D + _mesa_meta_CopyTexSubImage3D _mesa_wait_query _swrast_Accum _swrast_Bitmap @@ -953,15 +962,6 @@ EXPORTS _swrast_Clear _swrast_choose_line _swrast_choose_triangle - _swrast_CopyColorSubTable - _swrast_CopyColorTable - _swrast_CopyConvolutionFilter1D - _swrast_CopyConvolutionFilter2D - _swrast_copy_teximage1d - _swrast_copy_teximage2d - _swrast_copy_texsubimage1d - _swrast_copy_texsubimage2d - _swrast_copy_texsubimage3d _swrast_CreateContext _swrast_DestroyContext _swrast_exec_fragment_program diff --git a/src/mesa/drivers/windows/gdi/wmesa.c b/src/mesa/drivers/windows/gdi/wmesa.c index e1971db6933..8929b22af16 100644 --- a/src/mesa/drivers/windows/gdi/wmesa.c +++ b/src/mesa/drivers/windows/gdi/wmesa.c @@ -12,6 +12,7 @@ #include "framebuffer.h" #include "renderbuffer.h" #include "drivers/common/driverfuncs.h" +#include "drivers/common/meta.h" #include "vbo/vbo.h" #include "swrast/swrast.h" #include "swrast_setup/swrast_setup.h" @@ -1515,6 +1516,8 @@ WMesaContext WMesaCreateContext(HDC hDC, _mesa_enable_2_0_extensions(ctx); _mesa_enable_2_1_extensions(ctx); + _mesa_meta_init(ctx); + /* Initialize the software rasterizer and helper modules. */ if (!_swrast_CreateContext(ctx) || !_vbo_CreateContext(ctx) || @@ -1558,6 +1561,8 @@ void WMesaDestroyContext( WMesaContext pwc ) DeleteObject(pwc->clearPen); DeleteObject(pwc->clearBrush); + _mesa_meta_free(ctx); + _swsetup_DestroyContext(ctx); _tnl_DestroyContext(ctx); _vbo_DestroyContext(ctx); diff --git a/src/mesa/drivers/windows/gldirect/mesasw/gld_wgl_mesasw.c b/src/mesa/drivers/windows/gldirect/mesasw/gld_wgl_mesasw.c index 342a742867a..7ac425a1092 100644 --- a/src/mesa/drivers/windows/gldirect/mesasw/gld_wgl_mesasw.c +++ b/src/mesa/drivers/windows/gldirect/mesasw/gld_wgl_mesasw.c @@ -1346,6 +1346,8 @@ static void wmesa_update_state_first_time( struct swrast_device_driver *swdd = _swrast_GetDeviceDriverReference( ctx ); TNLcontext *tnl = TNL_CONTEXT(ctx); + _mesa_init_driver_functions(&ctx->Driver); + /* * XXX these function pointers could be initialized just once during * context creation since they don't depend on any state changes. @@ -1362,8 +1364,6 @@ static void wmesa_update_state_first_time( ctx->Driver.Viewport = wmesa_viewport; - ctx->Driver.Accum = _swrast_Accum; - ctx->Driver.Bitmap = _swrast_Bitmap; ctx->Driver.Clear = clear; ctx->Driver.Flush = flush; @@ -1371,28 +1371,6 @@ static void wmesa_update_state_first_time( ctx->Driver.ClearColor = clear_color; ctx->Driver.Enable = enable; - ctx->Driver.CopyPixels = _swrast_CopyPixels; - ctx->Driver.DrawPixels = _swrast_DrawPixels; - ctx->Driver.ReadPixels = _swrast_ReadPixels; - - ctx->Driver.ChooseTextureFormat = _mesa_choose_tex_format; - ctx->Driver.TexImage1D = _mesa_store_teximage1d; - ctx->Driver.TexImage2D = _mesa_store_teximage2d; - ctx->Driver.TexImage3D = _mesa_store_teximage3d; - ctx->Driver.TexSubImage1D = _mesa_store_texsubimage1d; - ctx->Driver.TexSubImage2D = _mesa_store_texsubimage2d; - ctx->Driver.TexSubImage3D = _mesa_store_texsubimage3d; - ctx->Driver.TestProxyTexImage = _mesa_test_proxy_teximage; - - ctx->Driver.CopyTexImage1D = _swrast_copy_teximage1d; - ctx->Driver.CopyTexImage2D = _swrast_copy_teximage2d; - ctx->Driver.CopyTexSubImage1D = _swrast_copy_texsubimage1d; - ctx->Driver.CopyTexSubImage2D = _swrast_copy_texsubimage2d; - ctx->Driver.CopyTexSubImage3D = _swrast_copy_texsubimage3d; - ctx->Driver.CopyColorTable = _swrast_CopyColorTable; - ctx->Driver.CopyColorSubTable = _swrast_CopyColorSubTable; - ctx->Driver.CopyConvolutionFilter1D = _swrast_CopyConvolutionFilter1D; - ctx->Driver.CopyConvolutionFilter2D = _swrast_CopyConvolutionFilter2D; // Does not apply for Mesa 5.x //ctx->Driver.BaseCompressedTexFormat = _mesa_base_compressed_texformat; diff --git a/src/mesa/drivers/windows/icd/mesa.def b/src/mesa/drivers/windows/icd/mesa.def index 465b380a0cb..25ac08a2f0f 100644 --- a/src/mesa/drivers/windows/icd/mesa.def +++ b/src/mesa/drivers/windows/icd/mesa.def @@ -75,6 +75,15 @@ EXPORTS _mesa_strcmp _mesa_test_proxy_teximage _mesa_Viewport + _mesa_meta_CopyColorSubTable + _mesa_meta_CopyColorTable + _mesa_meta_CopyConvolutionFilter1D + _mesa_meta_CopyConvolutionFilter2D + _mesa_meta_CopyTexImage1D + _mesa_meta_CopyTexImage2D + _mesa_meta_CopyTexSubImage1D + _mesa_meta_CopyTexSubImage2D + _mesa_meta_CopyTexSubImage3D _swrast_Accum _swrast_Bitmap _swrast_CopyPixels @@ -84,15 +93,6 @@ EXPORTS _swrast_Clear _swrast_choose_line _swrast_choose_triangle - _swrast_CopyColorSubTable - _swrast_CopyColorTable - _swrast_CopyConvolutionFilter1D - _swrast_CopyConvolutionFilter2D - _swrast_copy_teximage1d - _swrast_copy_teximage2d - _swrast_copy_texsubimage1d - _swrast_copy_texsubimage2d - _swrast_copy_texsubimage3d _swrast_CreateContext _swrast_DestroyContext _swrast_InvalidateState diff --git a/src/mesa/drivers/x11/xm_api.c b/src/mesa/drivers/x11/xm_api.c index 662c61ae7e7..79b058634cf 100644 --- a/src/mesa/drivers/x11/xm_api.c +++ b/src/mesa/drivers/x11/xm_api.c @@ -1648,8 +1648,7 @@ XMesaContext XMesaCreateContext( XMesaVisual v, XMesaContext share_list ) xmesa_register_swrast_functions( mesaCtx ); _swsetup_Wakeup(mesaCtx); - if (TEST_META_FUNCS) - _mesa_meta_init(mesaCtx); + _mesa_meta_init(mesaCtx); return c; } @@ -1665,8 +1664,7 @@ void XMesaDestroyContext( XMesaContext c ) FXdestroyContext( XMESA_BUFFER(mesaCtx->DrawBuffer) ); #endif - if (TEST_META_FUNCS) - _mesa_meta_free( mesaCtx ); + _mesa_meta_free( mesaCtx ); _swsetup_DestroyContext( mesaCtx ); _swrast_DestroyContext( mesaCtx ); diff --git a/src/mesa/drivers/x11/xm_dd.c b/src/mesa/drivers/x11/xm_dd.c index 4e9c001cc76..e2d4aa9b2d1 100644 --- a/src/mesa/drivers/x11/xm_dd.c +++ b/src/mesa/drivers/x11/xm_dd.c @@ -1150,11 +1150,11 @@ xmesa_init_driver_functions( XMesaVisual xmvisual, driver->Enable = enable; driver->Viewport = xmesa_viewport; if (TEST_META_FUNCS) { - driver->Clear = _mesa_meta_clear; - driver->CopyPixels = _mesa_meta_copy_pixels; - driver->BlitFramebuffer = _mesa_meta_blit_framebuffer; - driver->DrawPixels = _mesa_meta_draw_pixels; - driver->Bitmap = _mesa_meta_bitmap; + driver->Clear = _mesa_meta_Clear; + driver->CopyPixels = _mesa_meta_CopyPixels; + driver->BlitFramebuffer = _mesa_meta_BlitFramebuffer; + driver->DrawPixels = _mesa_meta_DrawPixels; + driver->Bitmap = _mesa_meta_Bitmap; } else { driver->Clear = clear_buffers; diff --git a/src/mesa/glapi/EXT_provoking_vertex.xml b/src/mesa/glapi/EXT_provoking_vertex.xml index f528a2c7d36..71d2c729091 100644 --- a/src/mesa/glapi/EXT_provoking_vertex.xml +++ b/src/mesa/glapi/EXT_provoking_vertex.xml @@ -19,4 +19,17 @@ </category> +<category name="GL_ARB_provoking_vertex" number="64"> + + <enum name="FIRST_VERTEX_CONVENTION" value="0x8E4D"/> + <enum name="LAST_VERTEX_CONVENTION" value="0x8E4E"/> + <enum name="PROVOKING_VERTEX" value="0x8E4F"/> + <enum name="QUADS_FOLLOW_PROVOKING_VERTEX_CONVENTION" value="0x8E4C"/> + + <function name="ProvokingVertex" alias="ProvokingVertexEXT"> + <param name="mode" type="GLenum"/> + </function> + +</category> + </OpenGLAPI> diff --git a/src/mesa/glapi/glapitemp.h b/src/mesa/glapi/glapitemp.h index 3da0f631435..d9a3690f2ad 100644 --- a/src/mesa/glapi/glapitemp.h +++ b/src/mesa/glapi/glapitemp.h @@ -5677,6 +5677,11 @@ KEYWORD1 void KEYWORD2 NAME(ProvokingVertexEXT)(GLenum mode) DISPATCH(ProvokingVertexEXT, (mode), (F, "glProvokingVertexEXT(0x%x);\n", mode)); } +KEYWORD1 void KEYWORD2 NAME(ProvokingVertex)(GLenum mode) +{ + DISPATCH(ProvokingVertexEXT, (mode), (F, "glProvokingVertex(0x%x);\n", mode)); +} + KEYWORD1_ALT void KEYWORD2 NAME(_dispatch_stub_788)(GLenum target, GLenum pname, GLvoid ** params); KEYWORD1_ALT void KEYWORD2 NAME(_dispatch_stub_788)(GLenum target, GLenum pname, GLvoid ** params) @@ -6923,6 +6928,7 @@ static _glapi_proc UNUSED_TABLE_NAME[] = { TABLE_ENTRY(RenderbufferStorage), TABLE_ENTRY(BlitFramebuffer), TABLE_ENTRY(FramebufferTextureLayer), + TABLE_ENTRY(ProvokingVertex), }; #endif /*UNUSED_TABLE_NAME*/ diff --git a/src/mesa/glapi/glprocs.h b/src/mesa/glapi/glprocs.h index fc0fcd331a3..c29f8b57be9 100644 --- a/src/mesa/glapi/glprocs.h +++ b/src/mesa/glapi/glprocs.h @@ -1147,6 +1147,7 @@ static const char gl_string_table[] = "glRenderbufferStorage\0" "glBlitFramebuffer\0" "glFramebufferTextureLayer\0" + "glProvokingVertex\0" ; @@ -2353,6 +2354,7 @@ static const glprocs_table_t static_functions[] = { NAME_FUNC_OFFSET(19722, glRenderbufferStorageEXT, glRenderbufferStorageEXT, NULL, _gloffset_RenderbufferStorageEXT), NAME_FUNC_OFFSET(19744, gl_dispatch_stub_783, gl_dispatch_stub_783, NULL, _gloffset_BlitFramebufferEXT), NAME_FUNC_OFFSET(19762, glFramebufferTextureLayerEXT, glFramebufferTextureLayerEXT, NULL, _gloffset_FramebufferTextureLayerEXT), + NAME_FUNC_OFFSET(19788, glProvokingVertexEXT, glProvokingVertexEXT, NULL, _gloffset_ProvokingVertexEXT), NAME_FUNC_OFFSET(-1, NULL, NULL, NULL, 0) }; diff --git a/src/mesa/glapi/mesadef.py b/src/mesa/glapi/mesadef.py index 0f410fc4825..342c9cde460 100644 --- a/src/mesa/glapi/mesadef.py +++ b/src/mesa/glapi/mesadef.py @@ -155,6 +155,15 @@ def PrintTail(): print '\t_mesa_strcmp' print '\t_mesa_test_proxy_teximage' print '\t_mesa_Viewport' + print '\t_mesa_meta_CopyColorSubTable' + print '\t_mesa_meta_CopyColorTable' + print '\t_mesa_meta_CopyConvolutionFilter1D' + print '\t_mesa_meta_CopyConvolutionFilter2D' + print '\t_mesa_meta_CopyTexImage1D' + print '\t_mesa_meta_CopyTexImage2D' + print '\t_mesa_meta_CopyTexSubImage1D' + print '\t_mesa_meta_CopyTexSubImage2D' + print '\t_mesa_meta_CopyTexSubImage3D' print '\t_swrast_Accum' print '\t_swrast_alloc_buffers' print '\t_swrast_Bitmap' @@ -164,15 +173,6 @@ def PrintTail(): print '\t_swrast_Clear' print '\t_swrast_choose_line' print '\t_swrast_choose_triangle' - print '\t_swrast_CopyColorSubTable' - print '\t_swrast_CopyColorTable' - print '\t_swrast_CopyConvolutionFilter1D' - print '\t_swrast_CopyConvolutionFilter2D' - print '\t_swrast_copy_teximage1d' - print '\t_swrast_copy_teximage2d' - print '\t_swrast_copy_texsubimage1d' - print '\t_swrast_copy_texsubimage2d' - print '\t_swrast_copy_texsubimage3d' print '\t_swrast_CreateContext' print '\t_swrast_DestroyContext' print '\t_swrast_InvalidateState' diff --git a/src/mesa/main/api_exec.c b/src/mesa/main/api_exec.c index 02550ae1088..63176390751 100644 --- a/src/mesa/main/api_exec.c +++ b/src/mesa/main/api_exec.c @@ -54,13 +54,9 @@ #endif #include "clear.h" #include "clip.h" -#if FEATURE_colortable #include "colortab.h" -#endif #include "context.h" -#if FEATURE_convolve #include "convolve.h" -#endif #include "depth.h" #if FEATURE_dlist #include "dlist.h" @@ -93,9 +89,7 @@ #include "macros.h" #include "matrix.h" #include "multisample.h" -#if FEATURE_pixel_transfer #include "pixel.h" -#endif #include "pixelstore.h" #include "points.h" #include "polygon.h" @@ -284,17 +278,9 @@ _mesa_init_exec_table(struct _glapi_table *exec) SET_MapGrid2f(exec, _mesa_MapGrid2f); #endif SET_MultMatrixd(exec, _mesa_MultMatrixd); -#if FEATURE_pixel_transfer - SET_GetPixelMapfv(exec, _mesa_GetPixelMapfv); - SET_GetPixelMapuiv(exec, _mesa_GetPixelMapuiv); - SET_GetPixelMapusv(exec, _mesa_GetPixelMapusv); - SET_PixelMapfv(exec, _mesa_PixelMapfv); - SET_PixelMapuiv(exec, _mesa_PixelMapuiv); - SET_PixelMapusv(exec, _mesa_PixelMapusv); - SET_PixelTransferf(exec, _mesa_PixelTransferf); - SET_PixelTransferi(exec, _mesa_PixelTransferi); - SET_PixelZoom(exec, _mesa_PixelZoom); -#endif + + _mesa_init_pixel_dispatch(exec); + SET_PixelStoref(exec, _mesa_PixelStoref); SET_PointSize(exec, _mesa_PointSize); SET_PolygonMode(exec, _mesa_PolygonMode); @@ -395,32 +381,8 @@ _mesa_init_exec_table(struct _glapi_table *exec) SET_BlendEquation(exec, _mesa_BlendEquation); SET_BlendEquationSeparateEXT(exec, _mesa_BlendEquationSeparateEXT); -#if FEATURE_colortable - SET_ColorSubTable(exec, _mesa_ColorSubTable); - SET_ColorTable(exec, _mesa_ColorTable); - SET_ColorTableParameterfv(exec, _mesa_ColorTableParameterfv); - SET_ColorTableParameteriv(exec, _mesa_ColorTableParameteriv); - SET_CopyColorSubTable(exec, _mesa_CopyColorSubTable); - SET_CopyColorTable(exec, _mesa_CopyColorTable); - SET_GetColorTable(exec, _mesa_GetColorTable); - SET_GetColorTableParameterfv(exec, _mesa_GetColorTableParameterfv); - SET_GetColorTableParameteriv(exec, _mesa_GetColorTableParameteriv); -#endif - -#if FEATURE_convolve - SET_ConvolutionFilter1D(exec, _mesa_ConvolutionFilter1D); - SET_ConvolutionFilter2D(exec, _mesa_ConvolutionFilter2D); - SET_ConvolutionParameterf(exec, _mesa_ConvolutionParameterf); - SET_ConvolutionParameterfv(exec, _mesa_ConvolutionParameterfv); - SET_ConvolutionParameteri(exec, _mesa_ConvolutionParameteri); - SET_ConvolutionParameteriv(exec, _mesa_ConvolutionParameteriv); - SET_CopyConvolutionFilter1D(exec, _mesa_CopyConvolutionFilter1D); - SET_CopyConvolutionFilter2D(exec, _mesa_CopyConvolutionFilter2D); - SET_GetConvolutionFilter(exec, _mesa_GetConvolutionFilter); - SET_GetConvolutionParameterfv(exec, _mesa_GetConvolutionParameterfv); - SET_GetConvolutionParameteriv(exec, _mesa_GetConvolutionParameteriv); - SET_SeparableFilter2D(exec, _mesa_SeparableFilter2D); -#endif + _mesa_init_colortable_dispatch(exec); + _mesa_init_convolve_dispatch(exec); #if FEATURE_histogram SET_GetHistogram(exec, _mesa_GetHistogram); SET_GetHistogramParameterfv(exec, _mesa_GetHistogramParameterfv); @@ -428,7 +390,6 @@ _mesa_init_exec_table(struct _glapi_table *exec) SET_GetMinmax(exec, _mesa_GetMinmax); SET_GetMinmaxParameterfv(exec, _mesa_GetMinmaxParameterfv); SET_GetMinmaxParameteriv(exec, _mesa_GetMinmaxParameteriv); - SET_GetSeparableFilter(exec, _mesa_GetSeparableFilter); SET_Histogram(exec, _mesa_Histogram); SET_Minmax(exec, _mesa_Minmax); SET_ResetHistogram(exec, _mesa_ResetHistogram); diff --git a/src/mesa/main/api_validate.c b/src/mesa/main/api_validate.c index 4a1448deee6..e71e5a6ce86 100644 --- a/src/mesa/main/api_validate.c +++ b/src/mesa/main/api_validate.c @@ -52,6 +52,51 @@ index_bytes(GLenum type, GLsizei count) /** + * Find the max index in the given element/index buffer + */ +GLuint +_mesa_max_buffer_index(GLcontext *ctx, GLuint count, GLenum type, + const void *indices, + struct gl_buffer_object *elementBuf) +{ + const GLubyte *map = NULL; + GLuint max = 0; + GLuint i; + + if (_mesa_is_bufferobj(elementBuf)) { + /* elements are in a user-defined buffer object. need to map it */ + map = ctx->Driver.MapBuffer(ctx, GL_ELEMENT_ARRAY_BUFFER, + GL_READ_ONLY, elementBuf); + /* Actual address is the sum of pointers */ + indices = (const GLvoid *) ADD_POINTERS(map, (const GLubyte *) indices); + } + + if (type == GL_UNSIGNED_INT) { + for (i = 0; i < count; i++) + if (((GLuint *) indices)[i] > max) + max = ((GLuint *) indices)[i]; + } + else if (type == GL_UNSIGNED_SHORT) { + for (i = 0; i < count; i++) + if (((GLushort *) indices)[i] > max) + max = ((GLushort *) indices)[i]; + } + else { + ASSERT(type == GL_UNSIGNED_BYTE); + for (i = 0; i < count; i++) + if (((GLubyte *) indices)[i] > max) + max = ((GLubyte *) indices)[i]; + } + + if (map) { + ctx->Driver.UnmapBuffer(ctx, GL_ELEMENT_ARRAY_BUFFER_ARB, elementBuf); + } + + return max; +} + + +/** * Check if OK to draw arrays/elements. */ static GLboolean diff --git a/src/mesa/main/api_validate.h b/src/mesa/main/api_validate.h index 1d3ae157d73..6064d15fe6c 100644 --- a/src/mesa/main/api_validate.h +++ b/src/mesa/main/api_validate.h @@ -30,6 +30,12 @@ #include "mtypes.h" + +extern GLuint +_mesa_max_buffer_index(GLcontext *ctx, GLuint count, GLenum type, + const void *indices, + struct gl_buffer_object *elementBuf); + extern GLboolean _mesa_validate_DrawArrays(GLcontext *ctx, GLenum mode, GLint start, GLsizei count); diff --git a/src/mesa/main/bufferobj.c b/src/mesa/main/bufferobj.c index b95e00af5bb..189b5e16558 100644 --- a/src/mesa/main/bufferobj.c +++ b/src/mesa/main/bufferobj.c @@ -1173,7 +1173,7 @@ _mesa_BufferDataARB(GLenum target, GLsizeiptrARB size, ASSERT(ctx->Driver.BufferData); if (!ctx->Driver.BufferData( ctx, target, size, data, usage, bufObj )) { - _mesa_error(ctx, GL_OUT_OF_MEMORY, "glBufferDataARB(access)"); + _mesa_error(ctx, GL_OUT_OF_MEMORY, "glBufferDataARB()"); } } @@ -1262,7 +1262,7 @@ _mesa_MapBufferARB(GLenum target, GLenum access) ASSERT(ctx->Driver.MapBuffer); map = ctx->Driver.MapBuffer( ctx, target, access, bufObj ); if (!map) { - _mesa_error(ctx, GL_OUT_OF_MEMORY, "glMapBufferARB(access)"); + _mesa_error(ctx, GL_OUT_OF_MEMORY, "glMapBufferARB(map failed)"); return NULL; } else { @@ -1593,7 +1593,7 @@ _mesa_MapBufferRange(GLenum target, GLintptr offset, GLsizeiptr length, map = ctx->Driver.MapBufferRange(ctx, target, offset, length, access, bufObj); if (!map) { - _mesa_error(ctx, GL_OUT_OF_MEMORY, "glMapBufferARB(access)"); + _mesa_error(ctx, GL_OUT_OF_MEMORY, "glMapBufferARB(map failed)"); } else { /* The driver callback should have set all these fields. diff --git a/src/mesa/main/colortab.c b/src/mesa/main/colortab.c index 5a7de5f2098..5ede76c1fb1 100644 --- a/src/mesa/main/colortab.c +++ b/src/mesa/main/colortab.c @@ -31,6 +31,11 @@ #include "macros.h" #include "state.h" #include "teximage.h" +#include "texstate.h" +#include "glapi/dispatch.h" + + +#if FEATURE_colortable /** @@ -278,7 +283,7 @@ _mesa_ColorTable( GLenum target, GLenum internalFormat, static const GLfloat one[4] = { 1.0, 1.0, 1.0, 1.0 }; static const GLfloat zero[4] = { 0.0, 0.0, 0.0, 0.0 }; GET_CURRENT_CONTEXT(ctx); - struct gl_texture_unit *texUnit = &ctx->Texture.Unit[ctx->Texture.CurrentUnit]; + struct gl_texture_unit *texUnit = _mesa_get_current_tex_unit(ctx); struct gl_texture_object *texObj = NULL; struct gl_color_table *table = NULL; GLboolean proxy = GL_FALSE; @@ -443,7 +448,7 @@ _mesa_ColorSubTable( GLenum target, GLsizei start, static const GLfloat one[4] = { 1.0, 1.0, 1.0, 1.0 }; static const GLfloat zero[4] = { 0.0, 0.0, 0.0, 0.0 }; GET_CURRENT_CONTEXT(ctx); - struct gl_texture_unit *texUnit = &ctx->Texture.Unit[ctx->Texture.CurrentUnit]; + struct gl_texture_unit *texUnit = _mesa_get_current_tex_unit(ctx); struct gl_texture_object *texObj = NULL; struct gl_color_table *table = NULL; const GLfloat *scale = one, *bias = zero; @@ -535,37 +540,44 @@ _mesa_ColorSubTable( GLenum target, GLsizei start, -void GLAPIENTRY +static void GLAPIENTRY _mesa_CopyColorTable(GLenum target, GLenum internalformat, GLint x, GLint y, GLsizei width) { GET_CURRENT_CONTEXT(ctx); ASSERT_OUTSIDE_BEGIN_END_AND_FLUSH(ctx); - /* Select buffer to read from */ + if (!ctx->ReadBuffer->_ColorReadBuffer) { + return; /* no readbuffer - OK */ + } + ctx->Driver.CopyColorTable( ctx, target, internalformat, x, y, width ); } -void GLAPIENTRY +static void GLAPIENTRY _mesa_CopyColorSubTable(GLenum target, GLsizei start, GLint x, GLint y, GLsizei width) { GET_CURRENT_CONTEXT(ctx); ASSERT_OUTSIDE_BEGIN_END_AND_FLUSH(ctx); + if (!ctx->ReadBuffer->_ColorReadBuffer) { + return; /* no readbuffer - OK */ + } + ctx->Driver.CopyColorSubTable( ctx, target, start, x, y, width ); } -void GLAPIENTRY +static void GLAPIENTRY _mesa_GetColorTable( GLenum target, GLenum format, GLenum type, GLvoid *data ) { GET_CURRENT_CONTEXT(ctx); - struct gl_texture_unit *texUnit = &ctx->Texture.Unit[ctx->Texture.CurrentUnit]; + struct gl_texture_unit *texUnit = _mesa_get_current_tex_unit(ctx); struct gl_color_table *table = NULL; GLfloat rgba[MAX_COLOR_TABLE_SIZE][4]; ASSERT_OUTSIDE_BEGIN_END_AND_FLUSH(ctx); @@ -694,7 +706,7 @@ _mesa_GetColorTable( GLenum target, GLenum format, -void GLAPIENTRY +static void GLAPIENTRY _mesa_ColorTableParameterfv(GLenum target, GLenum pname, const GLfloat *params) { GLfloat *scale, *bias; @@ -739,7 +751,7 @@ _mesa_ColorTableParameterfv(GLenum target, GLenum pname, const GLfloat *params) -void GLAPIENTRY +static void GLAPIENTRY _mesa_ColorTableParameteriv(GLenum target, GLenum pname, const GLint *params) { GLfloat fparams[4]; @@ -762,11 +774,11 @@ _mesa_ColorTableParameteriv(GLenum target, GLenum pname, const GLint *params) -void GLAPIENTRY +static void GLAPIENTRY _mesa_GetColorTableParameterfv( GLenum target, GLenum pname, GLfloat *params ) { GET_CURRENT_CONTEXT(ctx); - struct gl_texture_unit *texUnit = &ctx->Texture.Unit[ctx->Texture.CurrentUnit]; + struct gl_texture_unit *texUnit = _mesa_get_current_tex_unit(ctx); struct gl_color_table *table = NULL; ASSERT_OUTSIDE_BEGIN_END(ctx); @@ -889,11 +901,11 @@ _mesa_GetColorTableParameterfv( GLenum target, GLenum pname, GLfloat *params ) -void GLAPIENTRY +static void GLAPIENTRY _mesa_GetColorTableParameteriv( GLenum target, GLenum pname, GLint *params ) { GET_CURRENT_CONTEXT(ctx); - struct gl_texture_unit *texUnit = &ctx->Texture.Unit[ctx->Texture.CurrentUnit]; + struct gl_texture_unit *texUnit = _mesa_get_current_tex_unit(ctx); struct gl_color_table *table = NULL; ASSERT_OUTSIDE_BEGIN_END(ctx); @@ -1044,6 +1056,25 @@ _mesa_GetColorTableParameteriv( GLenum target, GLenum pname, GLint *params ) } } + +void +_mesa_init_colortable_dispatch(struct _glapi_table *disp) +{ + SET_ColorSubTable(disp, _mesa_ColorSubTable); + SET_ColorTable(disp, _mesa_ColorTable); + SET_ColorTableParameterfv(disp, _mesa_ColorTableParameterfv); + SET_ColorTableParameteriv(disp, _mesa_ColorTableParameteriv); + SET_CopyColorSubTable(disp, _mesa_CopyColorSubTable); + SET_CopyColorTable(disp, _mesa_CopyColorTable); + SET_GetColorTable(disp, _mesa_GetColorTable); + SET_GetColorTableParameterfv(disp, _mesa_GetColorTableParameterfv); + SET_GetColorTableParameteriv(disp, _mesa_GetColorTableParameteriv); +} + + +#endif /* FEATURE_colortable */ + + /**********************************************************************/ /***** Initialization *****/ /**********************************************************************/ diff --git a/src/mesa/main/colortab.h b/src/mesa/main/colortab.h index b6ff737a65f..652fb582463 100644 --- a/src/mesa/main/colortab.h +++ b/src/mesa/main/colortab.h @@ -27,9 +27,16 @@ #define COLORTAB_H -#include "mtypes.h" +#include "main/mtypes.h" -#if _HAVE_FULL_GL +#if FEATURE_colortable + +#define _MESA_INIT_COLORTABLE_FUNCTIONS(driver, impl) \ + do { \ + (driver)->CopyColorTable = impl ## CopyColorTable; \ + (driver)->CopyColorSubTable = impl ## CopyColorSubTable; \ + (driver)->UpdateTexturePalette = impl ## UpdateTexturePalette; \ + } while (0) extern void GLAPIENTRY _mesa_ColorTable( GLenum target, GLenum internalformat, @@ -41,32 +48,35 @@ _mesa_ColorSubTable( GLenum target, GLsizei start, GLsizei count, GLenum format, GLenum type, const GLvoid *table ); -extern void GLAPIENTRY -_mesa_CopyColorSubTable(GLenum target, GLsizei start, - GLint x, GLint y, GLsizei width); - -extern void GLAPIENTRY -_mesa_CopyColorTable(GLenum target, GLenum internalformat, - GLint x, GLint y, GLsizei width); +extern void +_mesa_init_colortable_dispatch(struct _glapi_table *disp); -extern void GLAPIENTRY -_mesa_GetColorTable( GLenum target, GLenum format, - GLenum type, GLvoid *table ); +#else /* FEATURE_colortable */ -extern void GLAPIENTRY -_mesa_ColorTableParameterfv(GLenum target, GLenum pname, - const GLfloat *params); +#define _MESA_INIT_COLORTABLE_FUNCTIONS(driver, impl) do { } while (0) -extern void GLAPIENTRY -_mesa_ColorTableParameteriv(GLenum target, GLenum pname, - const GLint *params); +static INLINE void GLAPIENTRY +_mesa_ColorTable( GLenum target, GLenum internalformat, + GLsizei width, GLenum format, GLenum type, + const GLvoid *table ) +{ + ASSERT_NO_FEATURE(); +} -extern void GLAPIENTRY -_mesa_GetColorTableParameterfv( GLenum target, GLenum pname, GLfloat *params ); +static INLINE void GLAPIENTRY +_mesa_ColorSubTable( GLenum target, GLsizei start, + GLsizei count, GLenum format, GLenum type, + const GLvoid *table ) +{ + ASSERT_NO_FEATURE(); +} -extern void GLAPIENTRY -_mesa_GetColorTableParameteriv( GLenum target, GLenum pname, GLint *params ); +static INLINE void +_mesa_init_colortable_dispatch(struct _glapi_table *disp) +{ +} +#endif /* FEATURE_colortable */ extern void @@ -81,20 +91,5 @@ _mesa_init_colortables( GLcontext *ctx ); extern void _mesa_free_colortables_data( GLcontext *ctx ); -#else - -/** No-op */ -#define _mesa_init_colortable( p ) ((void) 0) - -/** No-op */ -#define _mesa_free_colortable_data( p ) ((void) 0) - -/** No-op */ -#define _mesa_init_colortables( p ) ((void)0) - -/** No-op */ -#define _mesa_free_colortables_data( p ) ((void)0) - -#endif -#endif +#endif /* COLORTAB_H */ diff --git a/src/mesa/main/compiler.h b/src/mesa/main/compiler.h index 9319505a75d..380663ec971 100644 --- a/src/mesa/main/compiler.h +++ b/src/mesa/main/compiler.h @@ -107,8 +107,7 @@ extern "C" { /** * finite macro. */ -#if defined(_WIN32) && !defined(__WIN32__) && !defined(__CYGWIN__) && !defined(BUILD_FOR_SNAP) -# define __WIN32__ +#if defined(_MSC_VER) # define finite _finite #elif defined(__WATCOMC__) # define finite _finite diff --git a/src/mesa/main/context.c b/src/mesa/main/context.c index f6d4ac45957..1907c799df9 100644 --- a/src/mesa/main/context.c +++ b/src/mesa/main/context.c @@ -90,9 +90,7 @@ #include "blend.h" #include "buffers.h" #include "bufferobj.h" -#if FEATURE_colortable #include "colortab.h" -#endif #include "context.h" #include "cpuinfo.h" #include "debug.h" @@ -686,9 +684,7 @@ init_attrib_groups(GLcontext *ctx) #endif _mesa_init_buffer_objects( ctx ); _mesa_init_color( ctx ); -#if FEATURE_colortable _mesa_init_colortables( ctx ); -#endif _mesa_init_current( ctx ); _mesa_init_depth( ctx ); _mesa_init_debug( ctx ); @@ -1015,9 +1011,7 @@ _mesa_free_context_data( GLcontext *ctx ) _mesa_free_texture_data( ctx ); _mesa_free_matrix_data( ctx ); _mesa_free_viewport_data( ctx ); -#if FEATURE_colortable _mesa_free_colortables_data( ctx ); -#endif _mesa_free_program_data(ctx); _mesa_free_shader_state(ctx); #if FEATURE_ARB_occlusion_query diff --git a/src/mesa/main/convolve.c b/src/mesa/main/convolve.c index 70951112a18..8db3e79d384 100644 --- a/src/mesa/main/convolve.c +++ b/src/mesa/main/convolve.c @@ -40,6 +40,10 @@ #include "mtypes.h" #include "pixel.h" #include "state.h" +#include "glapi/dispatch.h" + + +#if FEATURE_convolve /* @@ -256,7 +260,7 @@ _mesa_ConvolutionFilter2D(GLenum target, GLenum internalFormat, GLsizei width, G } -void GLAPIENTRY +static void GLAPIENTRY _mesa_ConvolutionParameterf(GLenum target, GLenum pname, GLfloat param) { GET_CURRENT_CONTEXT(ctx); @@ -299,7 +303,7 @@ _mesa_ConvolutionParameterf(GLenum target, GLenum pname, GLfloat param) } -void GLAPIENTRY +static void GLAPIENTRY _mesa_ConvolutionParameterfv(GLenum target, GLenum pname, const GLfloat *params) { GET_CURRENT_CONTEXT(ctx); @@ -351,7 +355,7 @@ _mesa_ConvolutionParameterfv(GLenum target, GLenum pname, const GLfloat *params) } -void GLAPIENTRY +static void GLAPIENTRY _mesa_ConvolutionParameteri(GLenum target, GLenum pname, GLint param) { GET_CURRENT_CONTEXT(ctx); @@ -394,7 +398,7 @@ _mesa_ConvolutionParameteri(GLenum target, GLenum pname, GLint param) } -void GLAPIENTRY +static void GLAPIENTRY _mesa_ConvolutionParameteriv(GLenum target, GLenum pname, const GLint *params) { GET_CURRENT_CONTEXT(ctx); @@ -459,7 +463,7 @@ _mesa_ConvolutionParameteriv(GLenum target, GLenum pname, const GLint *params) } -void GLAPIENTRY +static void GLAPIENTRY _mesa_CopyConvolutionFilter1D(GLenum target, GLenum internalFormat, GLint x, GLint y, GLsizei width) { GLint baseFormat; @@ -482,12 +486,16 @@ _mesa_CopyConvolutionFilter1D(GLenum target, GLenum internalFormat, GLint x, GLi return; } + if (!ctx->ReadBuffer->_ColorReadBuffer) { + return; /* no readbuffer - OK */ + } + ctx->Driver.CopyConvolutionFilter1D( ctx, target, internalFormat, x, y, width); } -void GLAPIENTRY +static void GLAPIENTRY _mesa_CopyConvolutionFilter2D(GLenum target, GLenum internalFormat, GLint x, GLint y, GLsizei width, GLsizei height) { GLint baseFormat; @@ -514,12 +522,16 @@ _mesa_CopyConvolutionFilter2D(GLenum target, GLenum internalFormat, GLint x, GLi return; } + if (!ctx->ReadBuffer->_ColorReadBuffer) { + return; /* no readbuffer - OK */ + } + ctx->Driver.CopyConvolutionFilter2D( ctx, target, internalFormat, x, y, width, height ); } -void GLAPIENTRY +static void GLAPIENTRY _mesa_GetConvolutionFilter(GLenum target, GLenum format, GLenum type, GLvoid *image) { @@ -578,7 +590,7 @@ _mesa_GetConvolutionFilter(GLenum target, GLenum format, GLenum type, } -void GLAPIENTRY +static void GLAPIENTRY _mesa_GetConvolutionParameterfv(GLenum target, GLenum pname, GLfloat *params) { GET_CURRENT_CONTEXT(ctx); @@ -639,7 +651,7 @@ _mesa_GetConvolutionParameterfv(GLenum target, GLenum pname, GLfloat *params) } -void GLAPIENTRY +static void GLAPIENTRY _mesa_GetConvolutionParameteriv(GLenum target, GLenum pname, GLint *params) { GET_CURRENT_CONTEXT(ctx); @@ -709,7 +721,7 @@ _mesa_GetConvolutionParameteriv(GLenum target, GLenum pname, GLint *params) } -void GLAPIENTRY +static void GLAPIENTRY _mesa_GetSeparableFilter(GLenum target, GLenum format, GLenum type, GLvoid *row, GLvoid *column, GLvoid *span) { @@ -776,7 +788,7 @@ _mesa_GetSeparableFilter(GLenum target, GLenum format, GLenum type, } -void GLAPIENTRY +static void GLAPIENTRY _mesa_SeparableFilter2D(GLenum target, GLenum internalFormat, GLsizei width, GLsizei height, GLenum format, GLenum type, const GLvoid *row, const GLvoid *column) { const GLint colStart = MAX_CONVOLUTION_WIDTH * 4; @@ -1425,3 +1437,25 @@ _mesa_adjust_image_for_convolution(const GLcontext *ctx, GLuint dimensions, *height = *height - (MAX2(ctx->Separable2D.Height, 1) - 1); } } + + +void +_mesa_init_convolve_dispatch(struct _glapi_table *disp) +{ + SET_ConvolutionFilter1D(disp, _mesa_ConvolutionFilter1D); + SET_ConvolutionFilter2D(disp, _mesa_ConvolutionFilter2D); + SET_ConvolutionParameterf(disp, _mesa_ConvolutionParameterf); + SET_ConvolutionParameterfv(disp, _mesa_ConvolutionParameterfv); + SET_ConvolutionParameteri(disp, _mesa_ConvolutionParameteri); + SET_ConvolutionParameteriv(disp, _mesa_ConvolutionParameteriv); + SET_CopyConvolutionFilter1D(disp, _mesa_CopyConvolutionFilter1D); + SET_CopyConvolutionFilter2D(disp, _mesa_CopyConvolutionFilter2D); + SET_GetConvolutionFilter(disp, _mesa_GetConvolutionFilter); + SET_GetConvolutionParameterfv(disp, _mesa_GetConvolutionParameterfv); + SET_GetConvolutionParameteriv(disp, _mesa_GetConvolutionParameteriv); + SET_SeparableFilter2D(disp, _mesa_SeparableFilter2D); + SET_GetSeparableFilter(disp, _mesa_GetSeparableFilter); +} + + +#endif /* FEATURE_convolve */ diff --git a/src/mesa/main/convolve.h b/src/mesa/main/convolve.h index 4505cdae01e..59492bc7c54 100644 --- a/src/mesa/main/convolve.h +++ b/src/mesa/main/convolve.h @@ -28,10 +28,17 @@ #define CONVOLVE_H -#include "mtypes.h" +#include "main/mtypes.h" -#if _HAVE_FULL_GL +#if FEATURE_convolve + +#define _MESA_INIT_CONVOLVE_FUNCTIONS(driver, impl) \ + do { \ + (driver)->CopyConvolutionFilter1D = impl ## CopyConvolutionFilter1D; \ + (driver)->CopyConvolutionFilter2D = impl ## CopyConvolutionFilter2D; \ + } while (0) + extern void GLAPIENTRY _mesa_ConvolutionFilter1D(GLenum target, GLenum internalformat, GLsizei width, GLenum format, GLenum type, const GLvoid *image); @@ -41,74 +48,79 @@ _mesa_ConvolutionFilter2D(GLenum target, GLenum internalformat, GLsizei width, GLsizei height, GLenum format, GLenum type, const GLvoid *image); -extern void GLAPIENTRY -_mesa_ConvolutionParameterf(GLenum target, GLenum pname, GLfloat params); - -extern void GLAPIENTRY -_mesa_ConvolutionParameterfv(GLenum target, GLenum pname, - const GLfloat *params); - -extern void GLAPIENTRY -_mesa_ConvolutionParameteri(GLenum target, GLenum pname, GLint params); - -extern void GLAPIENTRY -_mesa_ConvolutionParameteriv(GLenum target, GLenum pname, const GLint *params); - -extern void GLAPIENTRY -_mesa_CopyConvolutionFilter1D(GLenum target, GLenum internalformat, - GLint x, GLint y, GLsizei width); +extern void +_mesa_convolve_1d_image(const GLcontext *ctx, GLsizei *width, + const GLfloat *srcImage, GLfloat *dstImage); -extern void GLAPIENTRY -_mesa_CopyConvolutionFilter2D(GLenum target, GLenum internalformat, - GLint x, GLint y, GLsizei width, GLsizei height); +extern void +_mesa_convolve_2d_image(const GLcontext *ctx, GLsizei *width, GLsizei *height, + const GLfloat *srcImage, GLfloat *dstImage); -extern void GLAPIENTRY -_mesa_GetConvolutionFilter(GLenum target, GLenum format, GLenum type, - GLvoid *image); +extern void +_mesa_convolve_sep_image(const GLcontext *ctx, + GLsizei *width, GLsizei *height, + const GLfloat *srcImage, GLfloat *dstImage); -extern void GLAPIENTRY -_mesa_GetConvolutionParameterfv(GLenum target, GLenum pname, GLfloat *params); +extern void +_mesa_adjust_image_for_convolution(const GLcontext *ctx, GLuint dimensions, + GLsizei *width, GLsizei *height); -extern void GLAPIENTRY -_mesa_GetConvolutionParameteriv(GLenum target, GLenum pname, GLint *params); +extern void +_mesa_init_convolve_dispatch(struct _glapi_table *disp); -extern void GLAPIENTRY -_mesa_GetSeparableFilter(GLenum target, GLenum format, GLenum type, - GLvoid *row, GLvoid *column, GLvoid *span); +#else /* FEATURE_convolve */ -extern void GLAPIENTRY -_mesa_SeparableFilter2D(GLenum target, GLenum internalformat, - GLsizei width, GLsizei height, - GLenum format, GLenum type, - const GLvoid *row, const GLvoid *column); +#define _MESA_INIT_CONVOLVE_FUNCTIONS(driver, impl) do { } while (0) +static INLINE void GLAPIENTRY +_mesa_ConvolutionFilter1D(GLenum target, GLenum internalformat, GLsizei width, + GLenum format, GLenum type, const GLvoid *image) +{ + ASSERT_NO_FEATURE(); +} +static INLINE void GLAPIENTRY +_mesa_ConvolutionFilter2D(GLenum target, GLenum internalformat, GLsizei width, + GLsizei height, GLenum format, GLenum type, + const GLvoid *image) +{ + ASSERT_NO_FEATURE(); +} -extern void +static INLINE void _mesa_convolve_1d_image(const GLcontext *ctx, GLsizei *width, - const GLfloat *srcImage, GLfloat *dstImage); - + const GLfloat *srcImage, GLfloat *dstImage) +{ + ASSERT_NO_FEATURE(); +} -extern void +static INLINE void _mesa_convolve_2d_image(const GLcontext *ctx, GLsizei *width, GLsizei *height, - const GLfloat *srcImage, GLfloat *dstImage); + const GLfloat *srcImage, GLfloat *dstImage) +{ + ASSERT_NO_FEATURE(); +} -extern void +static INLINE void _mesa_convolve_sep_image(const GLcontext *ctx, GLsizei *width, GLsizei *height, - const GLfloat *srcImage, GLfloat *dstImage); - + const GLfloat *srcImage, GLfloat *dstImage) +{ + ASSERT_NO_FEATURE(); +} -extern void +static INLINE void _mesa_adjust_image_for_convolution(const GLcontext *ctx, GLuint dimensions, - GLsizei *width, GLsizei *height); + GLsizei *width, GLsizei *height) +{ +} + +static INLINE void +_mesa_init_convolve_dispatch(struct _glapi_table *disp) +{ +} -#else -#define _mesa_adjust_image_for_convolution(c, d, w, h) ((void)0) -#define _mesa_convolve_1d_image(c,w,s,d) ((void)0) -#define _mesa_convolve_2d_image(c,w,h,s,d) ((void)0) -#define _mesa_convolve_sep_image(c,w,h,s,d) ((void)0) -#endif +#endif /* FEATURE_convolve */ -#endif +#endif /* CONVOLVE_H */ diff --git a/src/mesa/main/dlist.c b/src/mesa/main/dlist.c index d721f699fd1..9c25de41871 100644 --- a/src/mesa/main/dlist.c +++ b/src/mesa/main/dlist.c @@ -329,6 +329,34 @@ typedef enum OPCODE_STENCIL_OP_SEPARATE, OPCODE_STENCIL_MASK_SEPARATE, + /* GL_ARB_shader_objects */ + OPCODE_USE_PROGRAM, + OPCODE_UNIFORM_1F, + OPCODE_UNIFORM_2F, + OPCODE_UNIFORM_3F, + OPCODE_UNIFORM_4F, + OPCODE_UNIFORM_1FV, + OPCODE_UNIFORM_2FV, + OPCODE_UNIFORM_3FV, + OPCODE_UNIFORM_4FV, + OPCODE_UNIFORM_1I, + OPCODE_UNIFORM_2I, + OPCODE_UNIFORM_3I, + OPCODE_UNIFORM_4I, + OPCODE_UNIFORM_1IV, + OPCODE_UNIFORM_2IV, + OPCODE_UNIFORM_3IV, + OPCODE_UNIFORM_4IV, + OPCODE_UNIFORM_MATRIX22, + OPCODE_UNIFORM_MATRIX33, + OPCODE_UNIFORM_MATRIX44, + OPCODE_UNIFORM_MATRIX23, + OPCODE_UNIFORM_MATRIX32, + OPCODE_UNIFORM_MATRIX24, + OPCODE_UNIFORM_MATRIX42, + OPCODE_UNIFORM_MATRIX34, + OPCODE_UNIFORM_MATRIX43, + /* GL_EXT_framebuffer_blit */ OPCODE_BLIT_FRAMEBUFFER, @@ -576,6 +604,30 @@ _mesa_delete_list(GLcontext *ctx, struct gl_display_list *dlist) n += InstSize[n[0].opcode]; break; #endif + case OPCODE_UNIFORM_1FV: + case OPCODE_UNIFORM_2FV: + case OPCODE_UNIFORM_3FV: + case OPCODE_UNIFORM_4FV: + case OPCODE_UNIFORM_1IV: + case OPCODE_UNIFORM_2IV: + case OPCODE_UNIFORM_3IV: + case OPCODE_UNIFORM_4IV: + _mesa_free(n[3].data); + n += InstSize[n[0].opcode]; + break; + case OPCODE_UNIFORM_MATRIX22: + case OPCODE_UNIFORM_MATRIX33: + case OPCODE_UNIFORM_MATRIX44: + case OPCODE_UNIFORM_MATRIX24: + case OPCODE_UNIFORM_MATRIX42: + case OPCODE_UNIFORM_MATRIX23: + case OPCODE_UNIFORM_MATRIX32: + case OPCODE_UNIFORM_MATRIX34: + case OPCODE_UNIFORM_MATRIX43: + _mesa_free(n[4].data); + n += InstSize[n[0].opcode]; + break; + case OPCODE_CONTINUE: n = (Node *) n[1].next; _mesa_free(block); @@ -5804,6 +5856,493 @@ save_ProvokingVertexEXT(GLenum mode) } +/* aka UseProgram() */ +static void GLAPIENTRY +save_UseProgramObjectARB(GLhandleARB program) +{ + GET_CURRENT_CONTEXT(ctx); + Node *n; + ASSERT_OUTSIDE_SAVE_BEGIN_END_AND_FLUSH(ctx); + n = ALLOC_INSTRUCTION(ctx, OPCODE_USE_PROGRAM, 1); + if (n) { + n[1].ui = program; + } + if (ctx->ExecuteFlag) { + CALL_UseProgramObjectARB(ctx->Exec, (program)); + } +} + + +static void GLAPIENTRY +save_Uniform1fARB(GLint location, GLfloat x) +{ + GET_CURRENT_CONTEXT(ctx); + Node *n; + ASSERT_OUTSIDE_SAVE_BEGIN_END_AND_FLUSH(ctx); + n = ALLOC_INSTRUCTION(ctx, OPCODE_UNIFORM_1F, 2); + if (n) { + n[1].i = location; + n[2].f = x; + } + if (ctx->ExecuteFlag) { + CALL_Uniform1fARB(ctx->Exec, (location, x)); + } +} + + +static void GLAPIENTRY +save_Uniform2fARB(GLint location, GLfloat x, GLfloat y) +{ + GET_CURRENT_CONTEXT(ctx); + Node *n; + ASSERT_OUTSIDE_SAVE_BEGIN_END_AND_FLUSH(ctx); + n = ALLOC_INSTRUCTION(ctx, OPCODE_UNIFORM_2F, 3); + if (n) { + n[1].i = location; + n[2].f = x; + n[3].f = y; + } + if (ctx->ExecuteFlag) { + CALL_Uniform2fARB(ctx->Exec, (location, x, y)); + } +} + + +static void GLAPIENTRY +save_Uniform3fARB(GLint location, GLfloat x, GLfloat y, GLfloat z) +{ + GET_CURRENT_CONTEXT(ctx); + Node *n; + ASSERT_OUTSIDE_SAVE_BEGIN_END_AND_FLUSH(ctx); + n = ALLOC_INSTRUCTION(ctx, OPCODE_UNIFORM_3F, 4); + if (n) { + n[1].i = location; + n[2].f = x; + n[3].f = y; + n[4].f = z; + } + if (ctx->ExecuteFlag) { + CALL_Uniform3fARB(ctx->Exec, (location, x, y, z)); + } +} + + +static void GLAPIENTRY +save_Uniform4fARB(GLint location, GLfloat x, GLfloat y, GLfloat z, GLfloat w) +{ + GET_CURRENT_CONTEXT(ctx); + Node *n; + ASSERT_OUTSIDE_SAVE_BEGIN_END_AND_FLUSH(ctx); + n = ALLOC_INSTRUCTION(ctx, OPCODE_UNIFORM_4F, 5); + if (n) { + n[1].i = location; + n[2].f = x; + n[3].f = y; + n[4].f = z; + n[5].f = w; + } + if (ctx->ExecuteFlag) { + CALL_Uniform4fARB(ctx->Exec, (location, x, y, z, w)); + } +} + + +/** Return copy of memory */ +static void * +memdup(const void *src, GLsizei bytes) +{ + void *b = bytes >= 0 ? _mesa_malloc(bytes) : NULL; + if (b) + _mesa_memcpy(b, src, bytes); + return b; +} + + +static void GLAPIENTRY +save_Uniform1fvARB(GLint location, GLsizei count, const GLfloat *v) +{ + GET_CURRENT_CONTEXT(ctx); + Node *n; + ASSERT_OUTSIDE_SAVE_BEGIN_END_AND_FLUSH(ctx); + n = ALLOC_INSTRUCTION(ctx, OPCODE_UNIFORM_1FV, 3); + if (n) { + n[1].i = location; + n[2].i = count; + n[3].data = memdup(v, count * 1 * sizeof(GLfloat)); + } + if (ctx->ExecuteFlag) { + CALL_Uniform1fvARB(ctx->Exec, (location, count, v)); + } +} + +static void GLAPIENTRY +save_Uniform2fvARB(GLint location, GLsizei count, const GLfloat *v) +{ + GET_CURRENT_CONTEXT(ctx); + Node *n; + ASSERT_OUTSIDE_SAVE_BEGIN_END_AND_FLUSH(ctx); + n = ALLOC_INSTRUCTION(ctx, OPCODE_UNIFORM_2FV, 3); + if (n) { + n[1].i = location; + n[2].i = count; + n[3].data = memdup(v, count * 2 * sizeof(GLfloat)); + } + if (ctx->ExecuteFlag) { + CALL_Uniform2fvARB(ctx->Exec, (location, count, v)); + } +} + +static void GLAPIENTRY +save_Uniform3fvARB(GLint location, GLsizei count, const GLfloat *v) +{ + GET_CURRENT_CONTEXT(ctx); + Node *n; + ASSERT_OUTSIDE_SAVE_BEGIN_END_AND_FLUSH(ctx); + n = ALLOC_INSTRUCTION(ctx, OPCODE_UNIFORM_3FV, 3); + if (n) { + n[1].i = location; + n[2].i = count; + n[3].data = memdup(v, count * 3 * sizeof(GLfloat)); + } + if (ctx->ExecuteFlag) { + CALL_Uniform3fvARB(ctx->Exec, (location, count, v)); + } +} + +static void GLAPIENTRY +save_Uniform4fvARB(GLint location, GLsizei count, const GLfloat *v) +{ + GET_CURRENT_CONTEXT(ctx); + Node *n; + ASSERT_OUTSIDE_SAVE_BEGIN_END_AND_FLUSH(ctx); + n = ALLOC_INSTRUCTION(ctx, OPCODE_UNIFORM_4FV, 3); + if (n) { + n[1].i = location; + n[2].i = count; + n[3].data = memdup(v, count * 4 * sizeof(GLfloat)); + } + if (ctx->ExecuteFlag) { + CALL_Uniform4fvARB(ctx->Exec, (location, count, v)); + } +} + + +static void GLAPIENTRY +save_Uniform1iARB(GLint location, GLint x) +{ + GET_CURRENT_CONTEXT(ctx); + Node *n; + ASSERT_OUTSIDE_SAVE_BEGIN_END_AND_FLUSH(ctx); + n = ALLOC_INSTRUCTION(ctx, OPCODE_UNIFORM_1I, 2); + if (n) { + n[1].i = location; + n[2].i = x; + } + if (ctx->ExecuteFlag) { + CALL_Uniform1iARB(ctx->Exec, (location, x)); + } +} + +static void GLAPIENTRY +save_Uniform2iARB(GLint location, GLint x, GLint y) +{ + GET_CURRENT_CONTEXT(ctx); + Node *n; + ASSERT_OUTSIDE_SAVE_BEGIN_END_AND_FLUSH(ctx); + n = ALLOC_INSTRUCTION(ctx, OPCODE_UNIFORM_2I, 3); + if (n) { + n[1].i = location; + n[2].i = x; + n[3].i = y; + } + if (ctx->ExecuteFlag) { + CALL_Uniform2iARB(ctx->Exec, (location, x, y)); + } +} + +static void GLAPIENTRY +save_Uniform3iARB(GLint location, GLint x, GLint y, GLint z) +{ + GET_CURRENT_CONTEXT(ctx); + Node *n; + ASSERT_OUTSIDE_SAVE_BEGIN_END_AND_FLUSH(ctx); + n = ALLOC_INSTRUCTION(ctx, OPCODE_UNIFORM_3I, 4); + if (n) { + n[1].i = location; + n[2].i = x; + n[3].i = y; + n[4].i = z; + } + if (ctx->ExecuteFlag) { + CALL_Uniform3iARB(ctx->Exec, (location, x, y, z)); + } +} + +static void GLAPIENTRY +save_Uniform4iARB(GLint location, GLint x, GLint y, GLint z, GLint w) +{ + GET_CURRENT_CONTEXT(ctx); + Node *n; + ASSERT_OUTSIDE_SAVE_BEGIN_END_AND_FLUSH(ctx); + n = ALLOC_INSTRUCTION(ctx, OPCODE_UNIFORM_4I, 5); + if (n) { + n[1].i = location; + n[2].i = x; + n[3].i = y; + n[4].i = z; + n[5].i = w; + } + if (ctx->ExecuteFlag) { + CALL_Uniform4iARB(ctx->Exec, (location, x, y, z, w)); + } +} + + + +static void GLAPIENTRY +save_Uniform1ivARB(GLint location, GLsizei count, const GLint *v) +{ + GET_CURRENT_CONTEXT(ctx); + Node *n; + ASSERT_OUTSIDE_SAVE_BEGIN_END_AND_FLUSH(ctx); + n = ALLOC_INSTRUCTION(ctx, OPCODE_UNIFORM_1IV, 3); + if (n) { + n[1].i = location; + n[2].i = count; + n[3].data = memdup(v, count * 1 * sizeof(GLint)); + } + if (ctx->ExecuteFlag) { + CALL_Uniform1ivARB(ctx->Exec, (location, count, v)); + } +} + +static void GLAPIENTRY +save_Uniform2ivARB(GLint location, GLsizei count, const GLint *v) +{ + GET_CURRENT_CONTEXT(ctx); + Node *n; + ASSERT_OUTSIDE_SAVE_BEGIN_END_AND_FLUSH(ctx); + n = ALLOC_INSTRUCTION(ctx, OPCODE_UNIFORM_2IV, 3); + if (n) { + n[1].i = location; + n[2].i = count; + n[3].data = memdup(v, count * 2 * sizeof(GLint)); + } + if (ctx->ExecuteFlag) { + CALL_Uniform2ivARB(ctx->Exec, (location, count, v)); + } +} + +static void GLAPIENTRY +save_Uniform3ivARB(GLint location, GLsizei count, const GLint *v) +{ + GET_CURRENT_CONTEXT(ctx); + Node *n; + ASSERT_OUTSIDE_SAVE_BEGIN_END_AND_FLUSH(ctx); + n = ALLOC_INSTRUCTION(ctx, OPCODE_UNIFORM_3IV, 3); + if (n) { + n[1].i = location; + n[2].i = count; + n[3].data = memdup(v, count * 3 * sizeof(GLint)); + } + if (ctx->ExecuteFlag) { + CALL_Uniform3ivARB(ctx->Exec, (location, count, v)); + } +} + +static void GLAPIENTRY +save_Uniform4ivARB(GLint location, GLsizei count, const GLint *v) +{ + GET_CURRENT_CONTEXT(ctx); + Node *n; + ASSERT_OUTSIDE_SAVE_BEGIN_END_AND_FLUSH(ctx); + n = ALLOC_INSTRUCTION(ctx, OPCODE_UNIFORM_4IV, 3); + if (n) { + n[1].i = location; + n[2].i = count; + n[3].data = memdup(v, count * 4 * sizeof(GLfloat)); + } + if (ctx->ExecuteFlag) { + CALL_Uniform4ivARB(ctx->Exec, (location, count, v)); + } +} + + +static void GLAPIENTRY +save_UniformMatrix2fvARB(GLint location, GLsizei count, GLboolean transpose, + const GLfloat *m) +{ + GET_CURRENT_CONTEXT(ctx); + Node *n; + ASSERT_OUTSIDE_SAVE_BEGIN_END_AND_FLUSH(ctx); + n = ALLOC_INSTRUCTION(ctx, OPCODE_UNIFORM_MATRIX22, 4); + if (n) { + n[1].i = location; + n[2].i = count; + n[3].b = transpose; + n[4].data = memdup(m, count * 2 * 2 * sizeof(GLfloat)); + } + if (ctx->ExecuteFlag) { + CALL_UniformMatrix2fvARB(ctx->Exec, (location, count, transpose, m)); + } +} + +static void GLAPIENTRY +save_UniformMatrix3fvARB(GLint location, GLsizei count, GLboolean transpose, + const GLfloat *m) +{ + GET_CURRENT_CONTEXT(ctx); + Node *n; + ASSERT_OUTSIDE_SAVE_BEGIN_END_AND_FLUSH(ctx); + n = ALLOC_INSTRUCTION(ctx, OPCODE_UNIFORM_MATRIX33, 4); + if (n) { + n[1].i = location; + n[2].i = count; + n[3].b = transpose; + n[4].data = memdup(m, count * 3 * 3 * sizeof(GLfloat)); + } + if (ctx->ExecuteFlag) { + CALL_UniformMatrix3fvARB(ctx->Exec, (location, count, transpose, m)); + } +} + +static void GLAPIENTRY +save_UniformMatrix4fvARB(GLint location, GLsizei count, GLboolean transpose, + const GLfloat *m) +{ + GET_CURRENT_CONTEXT(ctx); + Node *n; + ASSERT_OUTSIDE_SAVE_BEGIN_END_AND_FLUSH(ctx); + n = ALLOC_INSTRUCTION(ctx, OPCODE_UNIFORM_MATRIX44, 4); + if (n) { + n[1].i = location; + n[2].i = count; + n[3].b = transpose; + n[4].data = memdup(m, count * 4 * 4 * sizeof(GLfloat)); + } + if (ctx->ExecuteFlag) { + CALL_UniformMatrix4fvARB(ctx->Exec, (location, count, transpose, m)); + } +} + + +static void GLAPIENTRY +save_UniformMatrix2x3fv(GLint location, GLsizei count, GLboolean transpose, + const GLfloat *m) +{ + GET_CURRENT_CONTEXT(ctx); + Node *n; + ASSERT_OUTSIDE_SAVE_BEGIN_END_AND_FLUSH(ctx); + n = ALLOC_INSTRUCTION(ctx, OPCODE_UNIFORM_MATRIX23, 4); + if (n) { + n[1].i = location; + n[2].i = count; + n[3].b = transpose; + n[4].data = memdup(m, count * 2 * 3 * sizeof(GLfloat)); + } + if (ctx->ExecuteFlag) { + CALL_UniformMatrix2x3fv(ctx->Exec, (location, count, transpose, m)); + } +} + +static void GLAPIENTRY +save_UniformMatrix3x2fv(GLint location, GLsizei count, GLboolean transpose, + const GLfloat *m) +{ + GET_CURRENT_CONTEXT(ctx); + Node *n; + ASSERT_OUTSIDE_SAVE_BEGIN_END_AND_FLUSH(ctx); + n = ALLOC_INSTRUCTION(ctx, OPCODE_UNIFORM_MATRIX32, 4); + if (n) { + n[1].i = location; + n[2].i = count; + n[3].b = transpose; + n[4].data = memdup(m, count * 3 * 2 * sizeof(GLfloat)); + } + if (ctx->ExecuteFlag) { + CALL_UniformMatrix3x2fv(ctx->Exec, (location, count, transpose, m)); + } +} + + +static void GLAPIENTRY +save_UniformMatrix2x4fv(GLint location, GLsizei count, GLboolean transpose, + const GLfloat *m) +{ + GET_CURRENT_CONTEXT(ctx); + Node *n; + ASSERT_OUTSIDE_SAVE_BEGIN_END_AND_FLUSH(ctx); + n = ALLOC_INSTRUCTION(ctx, OPCODE_UNIFORM_MATRIX24, 4); + if (n) { + n[1].i = location; + n[2].i = count; + n[3].b = transpose; + n[4].data = memdup(m, count * 2 * 4 * sizeof(GLfloat)); + } + if (ctx->ExecuteFlag) { + CALL_UniformMatrix2x4fv(ctx->Exec, (location, count, transpose, m)); + } +} + +static void GLAPIENTRY +save_UniformMatrix4x2fv(GLint location, GLsizei count, GLboolean transpose, + const GLfloat *m) +{ + GET_CURRENT_CONTEXT(ctx); + Node *n; + ASSERT_OUTSIDE_SAVE_BEGIN_END_AND_FLUSH(ctx); + n = ALLOC_INSTRUCTION(ctx, OPCODE_UNIFORM_MATRIX42, 4); + if (n) { + n[1].i = location; + n[2].i = count; + n[3].b = transpose; + n[4].data = memdup(m, count * 4 * 2 * sizeof(GLfloat)); + } + if (ctx->ExecuteFlag) { + CALL_UniformMatrix4x2fv(ctx->Exec, (location, count, transpose, m)); + } +} + + +static void GLAPIENTRY +save_UniformMatrix3x4fv(GLint location, GLsizei count, GLboolean transpose, + const GLfloat *m) +{ + GET_CURRENT_CONTEXT(ctx); + Node *n; + ASSERT_OUTSIDE_SAVE_BEGIN_END_AND_FLUSH(ctx); + n = ALLOC_INSTRUCTION(ctx, OPCODE_UNIFORM_MATRIX34, 4); + if (n) { + n[1].i = location; + n[2].i = count; + n[3].b = transpose; + n[4].data = memdup(m, count * 3 * 4 * sizeof(GLfloat)); + } + if (ctx->ExecuteFlag) { + CALL_UniformMatrix3x4fv(ctx->Exec, (location, count, transpose, m)); + } +} + +static void GLAPIENTRY +save_UniformMatrix4x3fv(GLint location, GLsizei count, GLboolean transpose, + const GLfloat *m) +{ + GET_CURRENT_CONTEXT(ctx); + Node *n; + ASSERT_OUTSIDE_SAVE_BEGIN_END_AND_FLUSH(ctx); + n = ALLOC_INSTRUCTION(ctx, OPCODE_UNIFORM_MATRIX43, 4); + if (n) { + n[1].i = location; + n[2].i = count; + n[3].b = transpose; + n[4].data = memdup(m, count * 4 * 3 * sizeof(GLfloat)); + } + if (ctx->ExecuteFlag) { + CALL_UniformMatrix4x3fv(ctx->Exec, (location, count, transpose, m)); + } +} + + /** * Save an error-generating command into display list. @@ -6640,6 +7179,98 @@ execute_list(GLcontext *ctx, GLuint list) n[9].i, n[10].e)); break; #endif + + case OPCODE_USE_PROGRAM: + CALL_UseProgramObjectARB(ctx->Exec, (n[1].ui)); + break; + case OPCODE_UNIFORM_1F: + CALL_Uniform1fARB(ctx->Exec, (n[1].i, n[2].f)); + break; + case OPCODE_UNIFORM_2F: + CALL_Uniform2fARB(ctx->Exec, (n[1].i, n[2].f, n[3].f)); + break; + case OPCODE_UNIFORM_3F: + CALL_Uniform3fARB(ctx->Exec, (n[1].i, n[2].f, n[3].f, n[4].f)); + break; + case OPCODE_UNIFORM_4F: + CALL_Uniform4fARB(ctx->Exec, + (n[1].i, n[2].f, n[3].f, n[4].f, n[5].f)); + break; + case OPCODE_UNIFORM_1FV: + CALL_Uniform1fvARB(ctx->Exec, (n[1].i, n[2].i, n[3].data)); + break; + case OPCODE_UNIFORM_2FV: + CALL_Uniform2fvARB(ctx->Exec, (n[1].i, n[2].i, n[3].data)); + break; + case OPCODE_UNIFORM_3FV: + CALL_Uniform3fvARB(ctx->Exec, (n[1].i, n[2].i, n[3].data)); + break; + case OPCODE_UNIFORM_4FV: + CALL_Uniform4fvARB(ctx->Exec, (n[1].i, n[2].i, n[3].data)); + break; + case OPCODE_UNIFORM_1I: + CALL_Uniform1iARB(ctx->Exec, (n[1].i, n[2].i)); + break; + case OPCODE_UNIFORM_2I: + CALL_Uniform2iARB(ctx->Exec, (n[1].i, n[2].i, n[3].i)); + break; + case OPCODE_UNIFORM_3I: + CALL_Uniform3iARB(ctx->Exec, (n[1].i, n[2].i, n[3].i, n[4].i)); + break; + case OPCODE_UNIFORM_4I: + CALL_Uniform4iARB(ctx->Exec, + (n[1].i, n[2].i, n[3].i, n[4].i, n[5].i)); + break; + case OPCODE_UNIFORM_1IV: + CALL_Uniform1ivARB(ctx->Exec, (n[1].i, n[2].i, n[3].data)); + break; + case OPCODE_UNIFORM_2IV: + CALL_Uniform2ivARB(ctx->Exec, (n[1].i, n[2].i, n[3].data)); + break; + case OPCODE_UNIFORM_3IV: + CALL_Uniform3ivARB(ctx->Exec, (n[1].i, n[2].i, n[3].data)); + break; + case OPCODE_UNIFORM_4IV: + CALL_Uniform4ivARB(ctx->Exec, (n[1].i, n[2].i, n[3].data)); + break; + + case OPCODE_UNIFORM_MATRIX22: + CALL_UniformMatrix2fvARB(ctx->Exec, + (n[1].i, n[2].i, n[3].b, n[4].data)); + break; + case OPCODE_UNIFORM_MATRIX33: + CALL_UniformMatrix3fvARB(ctx->Exec, + (n[1].i, n[2].i, n[3].b, n[4].data)); + break; + case OPCODE_UNIFORM_MATRIX44: + CALL_UniformMatrix4fvARB(ctx->Exec, + (n[1].i, n[2].i, n[3].b, n[4].data)); + break; + case OPCODE_UNIFORM_MATRIX23: + CALL_UniformMatrix2x3fv(ctx->Exec, + (n[1].i, n[2].i, n[3].b, n[4].data)); + break; + case OPCODE_UNIFORM_MATRIX32: + CALL_UniformMatrix3x2fv(ctx->Exec, + (n[1].i, n[2].i, n[3].b, n[4].data)); + break; + case OPCODE_UNIFORM_MATRIX24: + CALL_UniformMatrix2x4fv(ctx->Exec, + (n[1].i, n[2].i, n[3].b, n[4].data)); + break; + case OPCODE_UNIFORM_MATRIX42: + CALL_UniformMatrix4x2fv(ctx->Exec, + (n[1].i, n[2].i, n[3].b, n[4].data)); + break; + case OPCODE_UNIFORM_MATRIX34: + CALL_UniformMatrix3x4fv(ctx->Exec, + (n[1].i, n[2].i, n[3].b, n[4].data)); + break; + case OPCODE_UNIFORM_MATRIX43: + CALL_UniformMatrix4x3fv(ctx->Exec, + (n[1].i, n[2].i, n[3].b, n[4].data)); + break; + case OPCODE_TEX_BUMP_PARAMETER_ATI: { GLfloat values[4]; @@ -8309,6 +8940,34 @@ _mesa_init_dlist_table(struct _glapi_table *table) SET_BlitFramebufferEXT(table, save_BlitFramebufferEXT); #endif + /* GL_ARB_shader_objects */ + SET_UseProgramObjectARB(table, save_UseProgramObjectARB); + SET_Uniform1fARB(table, save_Uniform1fARB); + SET_Uniform2fARB(table, save_Uniform2fARB); + SET_Uniform3fARB(table, save_Uniform3fARB); + SET_Uniform4fARB(table, save_Uniform4fARB); + SET_Uniform1fvARB(table, save_Uniform1fvARB); + SET_Uniform2fvARB(table, save_Uniform2fvARB); + SET_Uniform3fvARB(table, save_Uniform3fvARB); + SET_Uniform4fvARB(table, save_Uniform4fvARB); + SET_Uniform1iARB(table, save_Uniform1iARB); + SET_Uniform2iARB(table, save_Uniform2iARB); + SET_Uniform3iARB(table, save_Uniform3iARB); + SET_Uniform4iARB(table, save_Uniform4iARB); + SET_Uniform1ivARB(table, save_Uniform1ivARB); + SET_Uniform2ivARB(table, save_Uniform2ivARB); + SET_Uniform3ivARB(table, save_Uniform3ivARB); + SET_Uniform4ivARB(table, save_Uniform4ivARB); + SET_UniformMatrix2fvARB(table, save_UniformMatrix2fvARB); + SET_UniformMatrix3fvARB(table, save_UniformMatrix3fvARB); + SET_UniformMatrix4fvARB(table, save_UniformMatrix4fvARB); + SET_UniformMatrix2x3fv(table, save_UniformMatrix2x3fv); + SET_UniformMatrix3x2fv(table, save_UniformMatrix3x2fv); + SET_UniformMatrix2x4fv(table, save_UniformMatrix2x4fv); + SET_UniformMatrix4x2fv(table, save_UniformMatrix4x2fv); + SET_UniformMatrix3x4fv(table, save_UniformMatrix3x4fv); + SET_UniformMatrix4x3fv(table, save_UniformMatrix4x3fv); + /* ARB 30/31/32. GL_ARB_shader_objects, GL_ARB_vertex/fragment_shader */ SET_BindAttribLocationARB(table, exec_BindAttribLocationARB); SET_GetAttribLocationARB(table, exec_GetAttribLocationARB); diff --git a/src/mesa/main/enable.c b/src/mesa/main/enable.c index 4c1f46102d1..8e6b01f73b9 100644 --- a/src/mesa/main/enable.c +++ b/src/mesa/main/enable.c @@ -233,7 +233,7 @@ enable_texture(GLcontext *ctx, GLboolean state, GLbitfield texBit) const GLbitfield newenabled = state ? (texUnit->Enabled | texBit) : (texUnit->Enabled & ~texBit); - if (!ctx->DrawBuffer->Visual.rgbMode || texUnit->Enabled == newenabled) + if (texUnit->Enabled == newenabled) return GL_FALSE; FLUSH_VERTICES(ctx, _NEW_TEXTURE); @@ -935,11 +935,6 @@ _mesa_set_enable(GLcontext *ctx, GLenum cap, GLboolean state) /* GL_EXT_depth_bounds_test */ case GL_DEPTH_BOUNDS_TEST_EXT: CHECK_EXTENSION(EXT_depth_bounds_test, cap); - if (state && ctx->DrawBuffer->Visual.depthBits == 0) { - _mesa_warning(ctx, - "glEnable(GL_DEPTH_BOUNDS_TEST_EXT) but no depth buffer"); - return; - } if (ctx->Depth.BoundsTest == state) return; FLUSH_VERTICES(ctx, _NEW_DEPTH); diff --git a/src/mesa/main/enums.c b/src/mesa/main/enums.c index 2d1594eb7a2..606d50c59ad 100644 --- a/src/mesa/main/enums.c +++ b/src/mesa/main/enums.c @@ -519,6 +519,7 @@ LONGSTRING static const char enum_string_table[] = "GL_FEEDBACK_BUFFER_SIZE\0" "GL_FEEDBACK_BUFFER_TYPE\0" "GL_FILL\0" + "GL_FIRST_VERTEX_CONVENTION\0" "GL_FIRST_VERTEX_CONVENTION_EXT\0" "GL_FLAT\0" "GL_FLOAT\0" @@ -702,6 +703,7 @@ LONGSTRING static const char enum_string_table[] = "GL_INVERSE_TRANSPOSE_NV\0" "GL_INVERT\0" "GL_KEEP\0" + "GL_LAST_VERTEX_CONVENTION\0" "GL_LAST_VERTEX_CONVENTION_EXT\0" "GL_LEFT\0" "GL_LEQUAL\0" @@ -1288,6 +1290,7 @@ LONGSTRING static const char enum_string_table[] = "GL_PROJECTION\0" "GL_PROJECTION_MATRIX\0" "GL_PROJECTION_STACK_DEPTH\0" + "GL_PROVOKING_VERTEX\0" "GL_PROVOKING_VERTEX_EXT\0" "GL_PROXY_COLOR_TABLE\0" "GL_PROXY_HISTOGRAM\0" @@ -1309,6 +1312,7 @@ LONGSTRING static const char enum_string_table[] = "GL_Q\0" "GL_QUADRATIC_ATTENUATION\0" "GL_QUADS\0" + "GL_QUADS_FOLLOW_PROVOKING_VERTEX_CONVENTION\0" "GL_QUADS_FOLLOW_PROVOKING_VERTEX_CONVENTION_EXT\0" "GL_QUAD_MESH_SUN\0" "GL_QUAD_STRIP\0" @@ -1896,7 +1900,7 @@ LONGSTRING static const char enum_string_table[] = "GL_ZOOM_Y\0" ; -static const enum_elt all_enums[1858] = +static const enum_elt all_enums[1862] = { { 0, 0x00000600 }, /* GL_2D */ { 6, 0x00001407 }, /* GL_2_BYTES */ @@ -2381,1460 +2385,1464 @@ static const enum_elt all_enums[1858] = { 9670, 0x00000DF1 }, /* GL_FEEDBACK_BUFFER_SIZE */ { 9694, 0x00000DF2 }, /* GL_FEEDBACK_BUFFER_TYPE */ { 9718, 0x00001B02 }, /* GL_FILL */ - { 9726, 0x00008E4D }, /* GL_FIRST_VERTEX_CONVENTION_EXT */ - { 9757, 0x00001D00 }, /* GL_FLAT */ - { 9765, 0x00001406 }, /* GL_FLOAT */ - { 9774, 0x00008B5A }, /* GL_FLOAT_MAT2 */ - { 9788, 0x00008B5A }, /* GL_FLOAT_MAT2_ARB */ - { 9806, 0x00008B65 }, /* GL_FLOAT_MAT2x3 */ - { 9822, 0x00008B66 }, /* GL_FLOAT_MAT2x4 */ - { 9838, 0x00008B5B }, /* GL_FLOAT_MAT3 */ - { 9852, 0x00008B5B }, /* GL_FLOAT_MAT3_ARB */ - { 9870, 0x00008B67 }, /* GL_FLOAT_MAT3x2 */ - { 9886, 0x00008B68 }, /* GL_FLOAT_MAT3x4 */ - { 9902, 0x00008B5C }, /* GL_FLOAT_MAT4 */ - { 9916, 0x00008B5C }, /* GL_FLOAT_MAT4_ARB */ - { 9934, 0x00008B69 }, /* GL_FLOAT_MAT4x2 */ - { 9950, 0x00008B6A }, /* GL_FLOAT_MAT4x3 */ - { 9966, 0x00008B50 }, /* GL_FLOAT_VEC2 */ - { 9980, 0x00008B50 }, /* GL_FLOAT_VEC2_ARB */ - { 9998, 0x00008B51 }, /* GL_FLOAT_VEC3 */ - { 10012, 0x00008B51 }, /* GL_FLOAT_VEC3_ARB */ - { 10030, 0x00008B52 }, /* GL_FLOAT_VEC4 */ - { 10044, 0x00008B52 }, /* GL_FLOAT_VEC4_ARB */ - { 10062, 0x00000B60 }, /* GL_FOG */ - { 10069, 0x00000080 }, /* GL_FOG_BIT */ - { 10080, 0x00000B66 }, /* GL_FOG_COLOR */ - { 10093, 0x00008451 }, /* GL_FOG_COORD */ - { 10106, 0x00008451 }, /* GL_FOG_COORDINATE */ - { 10124, 0x00008457 }, /* GL_FOG_COORDINATE_ARRAY */ - { 10148, 0x0000889D }, /* GL_FOG_COORDINATE_ARRAY_BUFFER_BINDING */ - { 10187, 0x0000889D }, /* GL_FOG_COORDINATE_ARRAY_BUFFER_BINDING_ARB */ - { 10230, 0x00008456 }, /* GL_FOG_COORDINATE_ARRAY_POINTER */ - { 10262, 0x00008455 }, /* GL_FOG_COORDINATE_ARRAY_STRIDE */ - { 10293, 0x00008454 }, /* GL_FOG_COORDINATE_ARRAY_TYPE */ - { 10322, 0x00008450 }, /* GL_FOG_COORDINATE_SOURCE */ - { 10347, 0x00008457 }, /* GL_FOG_COORD_ARRAY */ - { 10366, 0x0000889D }, /* GL_FOG_COORD_ARRAY_BUFFER_BINDING */ - { 10400, 0x00008456 }, /* GL_FOG_COORD_ARRAY_POINTER */ - { 10427, 0x00008455 }, /* GL_FOG_COORD_ARRAY_STRIDE */ - { 10453, 0x00008454 }, /* GL_FOG_COORD_ARRAY_TYPE */ - { 10477, 0x00008450 }, /* GL_FOG_COORD_SRC */ - { 10494, 0x00000B62 }, /* GL_FOG_DENSITY */ - { 10509, 0x0000855A }, /* GL_FOG_DISTANCE_MODE_NV */ - { 10533, 0x00000B64 }, /* GL_FOG_END */ - { 10544, 0x00000C54 }, /* GL_FOG_HINT */ - { 10556, 0x00000B61 }, /* GL_FOG_INDEX */ - { 10569, 0x00000B65 }, /* GL_FOG_MODE */ - { 10581, 0x00008198 }, /* GL_FOG_OFFSET_SGIX */ - { 10600, 0x00008199 }, /* GL_FOG_OFFSET_VALUE_SGIX */ - { 10625, 0x00000B63 }, /* GL_FOG_START */ - { 10638, 0x00008452 }, /* GL_FRAGMENT_DEPTH */ - { 10656, 0x00008804 }, /* GL_FRAGMENT_PROGRAM_ARB */ - { 10680, 0x00008B30 }, /* GL_FRAGMENT_SHADER */ - { 10699, 0x00008B30 }, /* GL_FRAGMENT_SHADER_ARB */ - { 10722, 0x00008B8B }, /* GL_FRAGMENT_SHADER_DERIVATIVE_HINT */ - { 10757, 0x00008D40 }, /* GL_FRAMEBUFFER */ - { 10772, 0x00008215 }, /* GL_FRAMEBUFFER_ATTACHMENT_ALPHA_SIZE */ - { 10809, 0x00008214 }, /* GL_FRAMEBUFFER_ATTACHMENT_BLUE_SIZE */ - { 10845, 0x00008210 }, /* GL_FRAMEBUFFER_ATTACHMENT_COLOR_ENCODING */ - { 10886, 0x00008211 }, /* GL_FRAMEBUFFER_ATTACHMENT_COMPONENT_TYPE */ - { 10927, 0x00008216 }, /* GL_FRAMEBUFFER_ATTACHMENT_DEPTH_SIZE */ - { 10964, 0x00008213 }, /* GL_FRAMEBUFFER_ATTACHMENT_GREEN_SIZE */ - { 11001, 0x00008CD1 }, /* GL_FRAMEBUFFER_ATTACHMENT_OBJECT_NAME */ - { 11039, 0x00008CD1 }, /* GL_FRAMEBUFFER_ATTACHMENT_OBJECT_NAME_EXT */ - { 11081, 0x00008CD0 }, /* GL_FRAMEBUFFER_ATTACHMENT_OBJECT_TYPE */ - { 11119, 0x00008CD0 }, /* GL_FRAMEBUFFER_ATTACHMENT_OBJECT_TYPE_EXT */ - { 11161, 0x00008212 }, /* GL_FRAMEBUFFER_ATTACHMENT_RED_SIZE */ - { 11196, 0x00008217 }, /* GL_FRAMEBUFFER_ATTACHMENT_STENCIL_SIZE */ - { 11235, 0x00008CD4 }, /* GL_FRAMEBUFFER_ATTACHMENT_TEXTURE_3D_ZOFFSET_EXT */ - { 11284, 0x00008CD3 }, /* GL_FRAMEBUFFER_ATTACHMENT_TEXTURE_CUBE_MAP_FACE */ - { 11332, 0x00008CD3 }, /* GL_FRAMEBUFFER_ATTACHMENT_TEXTURE_CUBE_MAP_FACE_EXT */ - { 11384, 0x00008CD4 }, /* GL_FRAMEBUFFER_ATTACHMENT_TEXTURE_LAYER */ - { 11424, 0x00008CD4 }, /* GL_FRAMEBUFFER_ATTACHMENT_TEXTURE_LAYER_EXT */ - { 11468, 0x00008CD2 }, /* GL_FRAMEBUFFER_ATTACHMENT_TEXTURE_LEVEL */ - { 11508, 0x00008CD2 }, /* GL_FRAMEBUFFER_ATTACHMENT_TEXTURE_LEVEL_EXT */ - { 11552, 0x00008CA6 }, /* GL_FRAMEBUFFER_BINDING_EXT */ - { 11579, 0x00008CD5 }, /* GL_FRAMEBUFFER_COMPLETE */ - { 11603, 0x00008CD5 }, /* GL_FRAMEBUFFER_COMPLETE_EXT */ - { 11631, 0x00008218 }, /* GL_FRAMEBUFFER_DEFAULT */ - { 11654, 0x00008D40 }, /* GL_FRAMEBUFFER_EXT */ - { 11673, 0x00008CD6 }, /* GL_FRAMEBUFFER_INCOMPLETE_ATTACHMENT */ - { 11710, 0x00008CD6 }, /* GL_FRAMEBUFFER_INCOMPLETE_ATTACHMENT_EXT */ - { 11751, 0x00008CD9 }, /* GL_FRAMEBUFFER_INCOMPLETE_DIMENSIONS_EXT */ - { 11792, 0x00008CDB }, /* GL_FRAMEBUFFER_INCOMPLETE_DRAW_BUFFER_EXT */ - { 11834, 0x00008CD8 }, /* GL_FRAMEBUFFER_INCOMPLETE_DUPLICATE_ATTACHMENT_EXT */ - { 11885, 0x00008CDA }, /* GL_FRAMEBUFFER_INCOMPLETE_FORMATS_EXT */ - { 11923, 0x00008CD7 }, /* GL_FRAMEBUFFER_INCOMPLETE_MISSING_ATTACHMENT */ - { 11968, 0x00008CD7 }, /* GL_FRAMEBUFFER_INCOMPLETE_MISSING_ATTACHMENT_EXT */ - { 12017, 0x00008D56 }, /* GL_FRAMEBUFFER_INCOMPLETE_MULTISAMPLE */ - { 12055, 0x00008CDC }, /* GL_FRAMEBUFFER_INCOMPLETE_READ_BUFFER_EXT */ - { 12097, 0x00008CDE }, /* GL_FRAMEBUFFER_STATUS_ERROR_EXT */ - { 12129, 0x00008219 }, /* GL_FRAMEBUFFER_UNDEFINED */ - { 12154, 0x00008CDD }, /* GL_FRAMEBUFFER_UNSUPPORTED */ - { 12181, 0x00008CDD }, /* GL_FRAMEBUFFER_UNSUPPORTED_EXT */ - { 12212, 0x00000404 }, /* GL_FRONT */ - { 12221, 0x00000408 }, /* GL_FRONT_AND_BACK */ - { 12239, 0x00000B46 }, /* GL_FRONT_FACE */ - { 12253, 0x00000400 }, /* GL_FRONT_LEFT */ - { 12267, 0x00000401 }, /* GL_FRONT_RIGHT */ - { 12282, 0x00008006 }, /* GL_FUNC_ADD */ - { 12294, 0x00008006 }, /* GL_FUNC_ADD_EXT */ - { 12310, 0x0000800B }, /* GL_FUNC_REVERSE_SUBTRACT */ - { 12335, 0x0000800B }, /* GL_FUNC_REVERSE_SUBTRACT_EXT */ - { 12364, 0x0000800A }, /* GL_FUNC_SUBTRACT */ - { 12381, 0x0000800A }, /* GL_FUNC_SUBTRACT_EXT */ - { 12402, 0x00008191 }, /* GL_GENERATE_MIPMAP */ - { 12421, 0x00008192 }, /* GL_GENERATE_MIPMAP_HINT */ - { 12445, 0x00008192 }, /* GL_GENERATE_MIPMAP_HINT_SGIS */ - { 12474, 0x00008191 }, /* GL_GENERATE_MIPMAP_SGIS */ - { 12498, 0x00000206 }, /* GL_GEQUAL */ - { 12508, 0x00000204 }, /* GL_GREATER */ - { 12519, 0x00001904 }, /* GL_GREEN */ - { 12528, 0x00000D19 }, /* GL_GREEN_BIAS */ - { 12542, 0x00000D53 }, /* GL_GREEN_BITS */ - { 12556, 0x00000D18 }, /* GL_GREEN_SCALE */ - { 12571, 0x00008000 }, /* GL_HINT_BIT */ - { 12583, 0x00008024 }, /* GL_HISTOGRAM */ - { 12596, 0x0000802B }, /* GL_HISTOGRAM_ALPHA_SIZE */ - { 12620, 0x0000802B }, /* GL_HISTOGRAM_ALPHA_SIZE_EXT */ - { 12648, 0x0000802A }, /* GL_HISTOGRAM_BLUE_SIZE */ - { 12671, 0x0000802A }, /* GL_HISTOGRAM_BLUE_SIZE_EXT */ - { 12698, 0x00008024 }, /* GL_HISTOGRAM_EXT */ - { 12715, 0x00008027 }, /* GL_HISTOGRAM_FORMAT */ - { 12735, 0x00008027 }, /* GL_HISTOGRAM_FORMAT_EXT */ - { 12759, 0x00008029 }, /* GL_HISTOGRAM_GREEN_SIZE */ - { 12783, 0x00008029 }, /* GL_HISTOGRAM_GREEN_SIZE_EXT */ - { 12811, 0x0000802C }, /* GL_HISTOGRAM_LUMINANCE_SIZE */ - { 12839, 0x0000802C }, /* GL_HISTOGRAM_LUMINANCE_SIZE_EXT */ - { 12871, 0x00008028 }, /* GL_HISTOGRAM_RED_SIZE */ - { 12893, 0x00008028 }, /* GL_HISTOGRAM_RED_SIZE_EXT */ - { 12919, 0x0000802D }, /* GL_HISTOGRAM_SINK */ - { 12937, 0x0000802D }, /* GL_HISTOGRAM_SINK_EXT */ - { 12959, 0x00008026 }, /* GL_HISTOGRAM_WIDTH */ - { 12978, 0x00008026 }, /* GL_HISTOGRAM_WIDTH_EXT */ - { 13001, 0x0000862A }, /* GL_IDENTITY_NV */ - { 13016, 0x00008150 }, /* GL_IGNORE_BORDER_HP */ - { 13036, 0x00008B9B }, /* GL_IMPLEMENTATION_COLOR_READ_FORMAT_OES */ - { 13076, 0x00008B9A }, /* GL_IMPLEMENTATION_COLOR_READ_TYPE_OES */ - { 13114, 0x00001E02 }, /* GL_INCR */ - { 13122, 0x00008507 }, /* GL_INCR_WRAP */ - { 13135, 0x00008507 }, /* GL_INCR_WRAP_EXT */ - { 13152, 0x00008222 }, /* GL_INDEX */ - { 13161, 0x00008077 }, /* GL_INDEX_ARRAY */ - { 13176, 0x00008899 }, /* GL_INDEX_ARRAY_BUFFER_BINDING */ - { 13206, 0x00008899 }, /* GL_INDEX_ARRAY_BUFFER_BINDING_ARB */ - { 13240, 0x00008091 }, /* GL_INDEX_ARRAY_POINTER */ - { 13263, 0x00008086 }, /* GL_INDEX_ARRAY_STRIDE */ - { 13285, 0x00008085 }, /* GL_INDEX_ARRAY_TYPE */ - { 13305, 0x00000D51 }, /* GL_INDEX_BITS */ - { 13319, 0x00000C20 }, /* GL_INDEX_CLEAR_VALUE */ - { 13340, 0x00000BF1 }, /* GL_INDEX_LOGIC_OP */ - { 13358, 0x00000C30 }, /* GL_INDEX_MODE */ - { 13372, 0x00000D13 }, /* GL_INDEX_OFFSET */ - { 13388, 0x00000D12 }, /* GL_INDEX_SHIFT */ - { 13403, 0x00000C21 }, /* GL_INDEX_WRITEMASK */ - { 13422, 0x00008B84 }, /* GL_INFO_LOG_LENGTH */ - { 13441, 0x00001404 }, /* GL_INT */ - { 13448, 0x00008049 }, /* GL_INTENSITY */ - { 13461, 0x0000804C }, /* GL_INTENSITY12 */ - { 13476, 0x0000804C }, /* GL_INTENSITY12_EXT */ - { 13495, 0x0000804D }, /* GL_INTENSITY16 */ - { 13510, 0x0000804D }, /* GL_INTENSITY16_EXT */ - { 13529, 0x0000804A }, /* GL_INTENSITY4 */ - { 13543, 0x0000804A }, /* GL_INTENSITY4_EXT */ - { 13561, 0x0000804B }, /* GL_INTENSITY8 */ - { 13575, 0x0000804B }, /* GL_INTENSITY8_EXT */ - { 13593, 0x00008049 }, /* GL_INTENSITY_EXT */ - { 13610, 0x00008575 }, /* GL_INTERPOLATE */ - { 13625, 0x00008575 }, /* GL_INTERPOLATE_ARB */ - { 13644, 0x00008575 }, /* GL_INTERPOLATE_EXT */ - { 13663, 0x00008B53 }, /* GL_INT_VEC2 */ - { 13675, 0x00008B53 }, /* GL_INT_VEC2_ARB */ - { 13691, 0x00008B54 }, /* GL_INT_VEC3 */ - { 13703, 0x00008B54 }, /* GL_INT_VEC3_ARB */ - { 13719, 0x00008B55 }, /* GL_INT_VEC4 */ - { 13731, 0x00008B55 }, /* GL_INT_VEC4_ARB */ - { 13747, 0x00000500 }, /* GL_INVALID_ENUM */ - { 13763, 0x00000506 }, /* GL_INVALID_FRAMEBUFFER_OPERATION */ - { 13796, 0x00000506 }, /* GL_INVALID_FRAMEBUFFER_OPERATION_EXT */ - { 13833, 0x00000502 }, /* GL_INVALID_OPERATION */ - { 13854, 0x00000501 }, /* GL_INVALID_VALUE */ - { 13871, 0x0000862B }, /* GL_INVERSE_NV */ - { 13885, 0x0000862D }, /* GL_INVERSE_TRANSPOSE_NV */ - { 13909, 0x0000150A }, /* GL_INVERT */ - { 13919, 0x00001E00 }, /* GL_KEEP */ - { 13927, 0x00008E4E }, /* GL_LAST_VERTEX_CONVENTION_EXT */ - { 13957, 0x00000406 }, /* GL_LEFT */ - { 13965, 0x00000203 }, /* GL_LEQUAL */ - { 13975, 0x00000201 }, /* GL_LESS */ - { 13983, 0x00004000 }, /* GL_LIGHT0 */ - { 13993, 0x00004001 }, /* GL_LIGHT1 */ - { 14003, 0x00004002 }, /* GL_LIGHT2 */ - { 14013, 0x00004003 }, /* GL_LIGHT3 */ - { 14023, 0x00004004 }, /* GL_LIGHT4 */ - { 14033, 0x00004005 }, /* GL_LIGHT5 */ - { 14043, 0x00004006 }, /* GL_LIGHT6 */ - { 14053, 0x00004007 }, /* GL_LIGHT7 */ - { 14063, 0x00000B50 }, /* GL_LIGHTING */ - { 14075, 0x00000040 }, /* GL_LIGHTING_BIT */ - { 14091, 0x00000B53 }, /* GL_LIGHT_MODEL_AMBIENT */ - { 14114, 0x000081F8 }, /* GL_LIGHT_MODEL_COLOR_CONTROL */ - { 14143, 0x000081F8 }, /* GL_LIGHT_MODEL_COLOR_CONTROL_EXT */ - { 14176, 0x00000B51 }, /* GL_LIGHT_MODEL_LOCAL_VIEWER */ - { 14204, 0x00000B52 }, /* GL_LIGHT_MODEL_TWO_SIDE */ - { 14228, 0x00001B01 }, /* GL_LINE */ - { 14236, 0x00002601 }, /* GL_LINEAR */ - { 14246, 0x00001208 }, /* GL_LINEAR_ATTENUATION */ - { 14268, 0x00008170 }, /* GL_LINEAR_CLIPMAP_LINEAR_SGIX */ - { 14298, 0x0000844F }, /* GL_LINEAR_CLIPMAP_NEAREST_SGIX */ - { 14329, 0x00002703 }, /* GL_LINEAR_MIPMAP_LINEAR */ - { 14353, 0x00002701 }, /* GL_LINEAR_MIPMAP_NEAREST */ - { 14378, 0x00000001 }, /* GL_LINES */ - { 14387, 0x00000004 }, /* GL_LINE_BIT */ - { 14399, 0x00000002 }, /* GL_LINE_LOOP */ - { 14412, 0x00000707 }, /* GL_LINE_RESET_TOKEN */ - { 14432, 0x00000B20 }, /* GL_LINE_SMOOTH */ - { 14447, 0x00000C52 }, /* GL_LINE_SMOOTH_HINT */ - { 14467, 0x00000B24 }, /* GL_LINE_STIPPLE */ - { 14483, 0x00000B25 }, /* GL_LINE_STIPPLE_PATTERN */ - { 14507, 0x00000B26 }, /* GL_LINE_STIPPLE_REPEAT */ - { 14530, 0x00000003 }, /* GL_LINE_STRIP */ - { 14544, 0x00000702 }, /* GL_LINE_TOKEN */ - { 14558, 0x00000B21 }, /* GL_LINE_WIDTH */ - { 14572, 0x00000B23 }, /* GL_LINE_WIDTH_GRANULARITY */ - { 14598, 0x00000B22 }, /* GL_LINE_WIDTH_RANGE */ - { 14618, 0x00008B82 }, /* GL_LINK_STATUS */ - { 14633, 0x00000B32 }, /* GL_LIST_BASE */ - { 14646, 0x00020000 }, /* GL_LIST_BIT */ - { 14658, 0x00000B33 }, /* GL_LIST_INDEX */ - { 14672, 0x00000B30 }, /* GL_LIST_MODE */ - { 14685, 0x00000101 }, /* GL_LOAD */ - { 14693, 0x00000BF1 }, /* GL_LOGIC_OP */ - { 14705, 0x00000BF0 }, /* GL_LOGIC_OP_MODE */ - { 14722, 0x00008CA1 }, /* GL_LOWER_LEFT */ - { 14736, 0x00001909 }, /* GL_LUMINANCE */ - { 14749, 0x00008041 }, /* GL_LUMINANCE12 */ - { 14764, 0x00008047 }, /* GL_LUMINANCE12_ALPHA12 */ - { 14787, 0x00008047 }, /* GL_LUMINANCE12_ALPHA12_EXT */ - { 14814, 0x00008046 }, /* GL_LUMINANCE12_ALPHA4 */ - { 14836, 0x00008046 }, /* GL_LUMINANCE12_ALPHA4_EXT */ - { 14862, 0x00008041 }, /* GL_LUMINANCE12_EXT */ - { 14881, 0x00008042 }, /* GL_LUMINANCE16 */ - { 14896, 0x00008048 }, /* GL_LUMINANCE16_ALPHA16 */ - { 14919, 0x00008048 }, /* GL_LUMINANCE16_ALPHA16_EXT */ - { 14946, 0x00008042 }, /* GL_LUMINANCE16_EXT */ - { 14965, 0x0000803F }, /* GL_LUMINANCE4 */ - { 14979, 0x00008043 }, /* GL_LUMINANCE4_ALPHA4 */ - { 15000, 0x00008043 }, /* GL_LUMINANCE4_ALPHA4_EXT */ - { 15025, 0x0000803F }, /* GL_LUMINANCE4_EXT */ - { 15043, 0x00008044 }, /* GL_LUMINANCE6_ALPHA2 */ - { 15064, 0x00008044 }, /* GL_LUMINANCE6_ALPHA2_EXT */ - { 15089, 0x00008040 }, /* GL_LUMINANCE8 */ - { 15103, 0x00008045 }, /* GL_LUMINANCE8_ALPHA8 */ - { 15124, 0x00008045 }, /* GL_LUMINANCE8_ALPHA8_EXT */ - { 15149, 0x00008040 }, /* GL_LUMINANCE8_EXT */ - { 15167, 0x0000190A }, /* GL_LUMINANCE_ALPHA */ - { 15186, 0x00000D90 }, /* GL_MAP1_COLOR_4 */ - { 15202, 0x00000DD0 }, /* GL_MAP1_GRID_DOMAIN */ - { 15222, 0x00000DD1 }, /* GL_MAP1_GRID_SEGMENTS */ - { 15244, 0x00000D91 }, /* GL_MAP1_INDEX */ - { 15258, 0x00000D92 }, /* GL_MAP1_NORMAL */ - { 15273, 0x00000D93 }, /* GL_MAP1_TEXTURE_COORD_1 */ - { 15297, 0x00000D94 }, /* GL_MAP1_TEXTURE_COORD_2 */ - { 15321, 0x00000D95 }, /* GL_MAP1_TEXTURE_COORD_3 */ - { 15345, 0x00000D96 }, /* GL_MAP1_TEXTURE_COORD_4 */ - { 15369, 0x00000D97 }, /* GL_MAP1_VERTEX_3 */ - { 15386, 0x00000D98 }, /* GL_MAP1_VERTEX_4 */ - { 15403, 0x00008660 }, /* GL_MAP1_VERTEX_ATTRIB0_4_NV */ - { 15431, 0x0000866A }, /* GL_MAP1_VERTEX_ATTRIB10_4_NV */ - { 15460, 0x0000866B }, /* GL_MAP1_VERTEX_ATTRIB11_4_NV */ - { 15489, 0x0000866C }, /* GL_MAP1_VERTEX_ATTRIB12_4_NV */ - { 15518, 0x0000866D }, /* GL_MAP1_VERTEX_ATTRIB13_4_NV */ - { 15547, 0x0000866E }, /* GL_MAP1_VERTEX_ATTRIB14_4_NV */ - { 15576, 0x0000866F }, /* GL_MAP1_VERTEX_ATTRIB15_4_NV */ - { 15605, 0x00008661 }, /* GL_MAP1_VERTEX_ATTRIB1_4_NV */ - { 15633, 0x00008662 }, /* GL_MAP1_VERTEX_ATTRIB2_4_NV */ - { 15661, 0x00008663 }, /* GL_MAP1_VERTEX_ATTRIB3_4_NV */ - { 15689, 0x00008664 }, /* GL_MAP1_VERTEX_ATTRIB4_4_NV */ - { 15717, 0x00008665 }, /* GL_MAP1_VERTEX_ATTRIB5_4_NV */ - { 15745, 0x00008666 }, /* GL_MAP1_VERTEX_ATTRIB6_4_NV */ - { 15773, 0x00008667 }, /* GL_MAP1_VERTEX_ATTRIB7_4_NV */ - { 15801, 0x00008668 }, /* GL_MAP1_VERTEX_ATTRIB8_4_NV */ - { 15829, 0x00008669 }, /* GL_MAP1_VERTEX_ATTRIB9_4_NV */ - { 15857, 0x00000DB0 }, /* GL_MAP2_COLOR_4 */ - { 15873, 0x00000DD2 }, /* GL_MAP2_GRID_DOMAIN */ - { 15893, 0x00000DD3 }, /* GL_MAP2_GRID_SEGMENTS */ - { 15915, 0x00000DB1 }, /* GL_MAP2_INDEX */ - { 15929, 0x00000DB2 }, /* GL_MAP2_NORMAL */ - { 15944, 0x00000DB3 }, /* GL_MAP2_TEXTURE_COORD_1 */ - { 15968, 0x00000DB4 }, /* GL_MAP2_TEXTURE_COORD_2 */ - { 15992, 0x00000DB5 }, /* GL_MAP2_TEXTURE_COORD_3 */ - { 16016, 0x00000DB6 }, /* GL_MAP2_TEXTURE_COORD_4 */ - { 16040, 0x00000DB7 }, /* GL_MAP2_VERTEX_3 */ - { 16057, 0x00000DB8 }, /* GL_MAP2_VERTEX_4 */ - { 16074, 0x00008670 }, /* GL_MAP2_VERTEX_ATTRIB0_4_NV */ - { 16102, 0x0000867A }, /* GL_MAP2_VERTEX_ATTRIB10_4_NV */ - { 16131, 0x0000867B }, /* GL_MAP2_VERTEX_ATTRIB11_4_NV */ - { 16160, 0x0000867C }, /* GL_MAP2_VERTEX_ATTRIB12_4_NV */ - { 16189, 0x0000867D }, /* GL_MAP2_VERTEX_ATTRIB13_4_NV */ - { 16218, 0x0000867E }, /* GL_MAP2_VERTEX_ATTRIB14_4_NV */ - { 16247, 0x0000867F }, /* GL_MAP2_VERTEX_ATTRIB15_4_NV */ - { 16276, 0x00008671 }, /* GL_MAP2_VERTEX_ATTRIB1_4_NV */ - { 16304, 0x00008672 }, /* GL_MAP2_VERTEX_ATTRIB2_4_NV */ - { 16332, 0x00008673 }, /* GL_MAP2_VERTEX_ATTRIB3_4_NV */ - { 16360, 0x00008674 }, /* GL_MAP2_VERTEX_ATTRIB4_4_NV */ - { 16388, 0x00008675 }, /* GL_MAP2_VERTEX_ATTRIB5_4_NV */ - { 16416, 0x00008676 }, /* GL_MAP2_VERTEX_ATTRIB6_4_NV */ - { 16444, 0x00008677 }, /* GL_MAP2_VERTEX_ATTRIB7_4_NV */ - { 16472, 0x00008678 }, /* GL_MAP2_VERTEX_ATTRIB8_4_NV */ - { 16500, 0x00008679 }, /* GL_MAP2_VERTEX_ATTRIB9_4_NV */ - { 16528, 0x00000D10 }, /* GL_MAP_COLOR */ - { 16541, 0x00000010 }, /* GL_MAP_FLUSH_EXPLICIT_BIT */ - { 16567, 0x00000008 }, /* GL_MAP_INVALIDATE_BUFFER_BIT */ - { 16596, 0x00000004 }, /* GL_MAP_INVALIDATE_RANGE_BIT */ - { 16624, 0x00000001 }, /* GL_MAP_READ_BIT */ - { 16640, 0x00000D11 }, /* GL_MAP_STENCIL */ - { 16655, 0x00000020 }, /* GL_MAP_UNSYNCHRONIZED_BIT */ - { 16681, 0x00000002 }, /* GL_MAP_WRITE_BIT */ - { 16698, 0x000088C0 }, /* GL_MATRIX0_ARB */ - { 16713, 0x00008630 }, /* GL_MATRIX0_NV */ - { 16727, 0x000088CA }, /* GL_MATRIX10_ARB */ - { 16743, 0x000088CB }, /* GL_MATRIX11_ARB */ - { 16759, 0x000088CC }, /* GL_MATRIX12_ARB */ - { 16775, 0x000088CD }, /* GL_MATRIX13_ARB */ - { 16791, 0x000088CE }, /* GL_MATRIX14_ARB */ - { 16807, 0x000088CF }, /* GL_MATRIX15_ARB */ - { 16823, 0x000088D0 }, /* GL_MATRIX16_ARB */ - { 16839, 0x000088D1 }, /* GL_MATRIX17_ARB */ - { 16855, 0x000088D2 }, /* GL_MATRIX18_ARB */ - { 16871, 0x000088D3 }, /* GL_MATRIX19_ARB */ - { 16887, 0x000088C1 }, /* GL_MATRIX1_ARB */ - { 16902, 0x00008631 }, /* GL_MATRIX1_NV */ - { 16916, 0x000088D4 }, /* GL_MATRIX20_ARB */ - { 16932, 0x000088D5 }, /* GL_MATRIX21_ARB */ - { 16948, 0x000088D6 }, /* GL_MATRIX22_ARB */ - { 16964, 0x000088D7 }, /* GL_MATRIX23_ARB */ - { 16980, 0x000088D8 }, /* GL_MATRIX24_ARB */ - { 16996, 0x000088D9 }, /* GL_MATRIX25_ARB */ - { 17012, 0x000088DA }, /* GL_MATRIX26_ARB */ - { 17028, 0x000088DB }, /* GL_MATRIX27_ARB */ - { 17044, 0x000088DC }, /* GL_MATRIX28_ARB */ - { 17060, 0x000088DD }, /* GL_MATRIX29_ARB */ - { 17076, 0x000088C2 }, /* GL_MATRIX2_ARB */ - { 17091, 0x00008632 }, /* GL_MATRIX2_NV */ - { 17105, 0x000088DE }, /* GL_MATRIX30_ARB */ - { 17121, 0x000088DF }, /* GL_MATRIX31_ARB */ - { 17137, 0x000088C3 }, /* GL_MATRIX3_ARB */ - { 17152, 0x00008633 }, /* GL_MATRIX3_NV */ - { 17166, 0x000088C4 }, /* GL_MATRIX4_ARB */ - { 17181, 0x00008634 }, /* GL_MATRIX4_NV */ - { 17195, 0x000088C5 }, /* GL_MATRIX5_ARB */ - { 17210, 0x00008635 }, /* GL_MATRIX5_NV */ - { 17224, 0x000088C6 }, /* GL_MATRIX6_ARB */ - { 17239, 0x00008636 }, /* GL_MATRIX6_NV */ - { 17253, 0x000088C7 }, /* GL_MATRIX7_ARB */ - { 17268, 0x00008637 }, /* GL_MATRIX7_NV */ - { 17282, 0x000088C8 }, /* GL_MATRIX8_ARB */ - { 17297, 0x000088C9 }, /* GL_MATRIX9_ARB */ - { 17312, 0x00008844 }, /* GL_MATRIX_INDEX_ARRAY_ARB */ - { 17338, 0x00008849 }, /* GL_MATRIX_INDEX_ARRAY_POINTER_ARB */ - { 17372, 0x00008846 }, /* GL_MATRIX_INDEX_ARRAY_SIZE_ARB */ - { 17403, 0x00008848 }, /* GL_MATRIX_INDEX_ARRAY_STRIDE_ARB */ - { 17436, 0x00008847 }, /* GL_MATRIX_INDEX_ARRAY_TYPE_ARB */ - { 17467, 0x00000BA0 }, /* GL_MATRIX_MODE */ - { 17482, 0x00008840 }, /* GL_MATRIX_PALETTE_ARB */ - { 17504, 0x00008008 }, /* GL_MAX */ - { 17511, 0x00008073 }, /* GL_MAX_3D_TEXTURE_SIZE */ - { 17534, 0x000088FF }, /* GL_MAX_ARRAY_TEXTURE_LAYERS_EXT */ - { 17566, 0x00000D35 }, /* GL_MAX_ATTRIB_STACK_DEPTH */ - { 17592, 0x00000D3B }, /* GL_MAX_CLIENT_ATTRIB_STACK_DEPTH */ - { 17625, 0x00008177 }, /* GL_MAX_CLIPMAP_DEPTH_SGIX */ - { 17651, 0x00008178 }, /* GL_MAX_CLIPMAP_VIRTUAL_DEPTH_SGIX */ - { 17685, 0x00000D32 }, /* GL_MAX_CLIP_PLANES */ - { 17704, 0x00008CDF }, /* GL_MAX_COLOR_ATTACHMENTS_EXT */ - { 17733, 0x000080B3 }, /* GL_MAX_COLOR_MATRIX_STACK_DEPTH */ - { 17765, 0x000080B3 }, /* GL_MAX_COLOR_MATRIX_STACK_DEPTH_SGI */ - { 17801, 0x00008B4D }, /* GL_MAX_COMBINED_TEXTURE_IMAGE_UNITS */ - { 17837, 0x00008B4D }, /* GL_MAX_COMBINED_TEXTURE_IMAGE_UNITS_ARB */ - { 17877, 0x0000801B }, /* GL_MAX_CONVOLUTION_HEIGHT */ - { 17903, 0x0000801B }, /* GL_MAX_CONVOLUTION_HEIGHT_EXT */ - { 17933, 0x0000801A }, /* GL_MAX_CONVOLUTION_WIDTH */ - { 17958, 0x0000801A }, /* GL_MAX_CONVOLUTION_WIDTH_EXT */ - { 17987, 0x0000851C }, /* GL_MAX_CUBE_MAP_TEXTURE_SIZE */ - { 18016, 0x0000851C }, /* GL_MAX_CUBE_MAP_TEXTURE_SIZE_ARB */ - { 18049, 0x00008824 }, /* GL_MAX_DRAW_BUFFERS */ - { 18069, 0x00008824 }, /* GL_MAX_DRAW_BUFFERS_ARB */ - { 18093, 0x00008824 }, /* GL_MAX_DRAW_BUFFERS_ATI */ - { 18117, 0x000080E9 }, /* GL_MAX_ELEMENTS_INDICES */ - { 18141, 0x000080E8 }, /* GL_MAX_ELEMENTS_VERTICES */ - { 18166, 0x00000D30 }, /* GL_MAX_EVAL_ORDER */ - { 18184, 0x00008008 }, /* GL_MAX_EXT */ - { 18195, 0x00008B49 }, /* GL_MAX_FRAGMENT_UNIFORM_COMPONENTS */ - { 18230, 0x00008B49 }, /* GL_MAX_FRAGMENT_UNIFORM_COMPONENTS_ARB */ - { 18269, 0x00000D31 }, /* GL_MAX_LIGHTS */ - { 18283, 0x00000B31 }, /* GL_MAX_LIST_NESTING */ - { 18303, 0x00008841 }, /* GL_MAX_MATRIX_PALETTE_STACK_DEPTH_ARB */ - { 18341, 0x00000D36 }, /* GL_MAX_MODELVIEW_STACK_DEPTH */ - { 18370, 0x00000D37 }, /* GL_MAX_NAME_STACK_DEPTH */ - { 18394, 0x00008842 }, /* GL_MAX_PALETTE_MATRICES_ARB */ - { 18422, 0x00000D34 }, /* GL_MAX_PIXEL_MAP_TABLE */ - { 18445, 0x000088B1 }, /* GL_MAX_PROGRAM_ADDRESS_REGISTERS_ARB */ - { 18482, 0x0000880B }, /* GL_MAX_PROGRAM_ALU_INSTRUCTIONS_ARB */ - { 18518, 0x000088AD }, /* GL_MAX_PROGRAM_ATTRIBS_ARB */ - { 18545, 0x000088F5 }, /* GL_MAX_PROGRAM_CALL_DEPTH_NV */ - { 18574, 0x000088B5 }, /* GL_MAX_PROGRAM_ENV_PARAMETERS_ARB */ - { 18608, 0x000088F4 }, /* GL_MAX_PROGRAM_EXEC_INSTRUCTIONS_NV */ - { 18644, 0x000088F6 }, /* GL_MAX_PROGRAM_IF_DEPTH_NV */ - { 18671, 0x000088A1 }, /* GL_MAX_PROGRAM_INSTRUCTIONS_ARB */ - { 18703, 0x000088B4 }, /* GL_MAX_PROGRAM_LOCAL_PARAMETERS_ARB */ - { 18739, 0x000088F8 }, /* GL_MAX_PROGRAM_LOOP_COUNT_NV */ - { 18768, 0x000088F7 }, /* GL_MAX_PROGRAM_LOOP_DEPTH_NV */ - { 18797, 0x0000862F }, /* GL_MAX_PROGRAM_MATRICES_ARB */ - { 18825, 0x0000862E }, /* GL_MAX_PROGRAM_MATRIX_STACK_DEPTH_ARB */ - { 18863, 0x000088B3 }, /* GL_MAX_PROGRAM_NATIVE_ADDRESS_REGISTERS_ARB */ - { 18907, 0x0000880E }, /* GL_MAX_PROGRAM_NATIVE_ALU_INSTRUCTIONS_ARB */ - { 18950, 0x000088AF }, /* GL_MAX_PROGRAM_NATIVE_ATTRIBS_ARB */ - { 18984, 0x000088A3 }, /* GL_MAX_PROGRAM_NATIVE_INSTRUCTIONS_ARB */ - { 19023, 0x000088AB }, /* GL_MAX_PROGRAM_NATIVE_PARAMETERS_ARB */ - { 19060, 0x000088A7 }, /* GL_MAX_PROGRAM_NATIVE_TEMPORARIES_ARB */ - { 19098, 0x00008810 }, /* GL_MAX_PROGRAM_NATIVE_TEX_INDIRECTIONS_ARB */ - { 19141, 0x0000880F }, /* GL_MAX_PROGRAM_NATIVE_TEX_INSTRUCTIONS_ARB */ - { 19184, 0x000088A9 }, /* GL_MAX_PROGRAM_PARAMETERS_ARB */ - { 19214, 0x000088A5 }, /* GL_MAX_PROGRAM_TEMPORARIES_ARB */ - { 19245, 0x0000880D }, /* GL_MAX_PROGRAM_TEX_INDIRECTIONS_ARB */ - { 19281, 0x0000880C }, /* GL_MAX_PROGRAM_TEX_INSTRUCTIONS_ARB */ - { 19317, 0x00000D38 }, /* GL_MAX_PROJECTION_STACK_DEPTH */ - { 19347, 0x000084F8 }, /* GL_MAX_RECTANGLE_TEXTURE_SIZE_ARB */ - { 19381, 0x000084F8 }, /* GL_MAX_RECTANGLE_TEXTURE_SIZE_NV */ - { 19414, 0x000084E8 }, /* GL_MAX_RENDERBUFFER_SIZE_EXT */ - { 19443, 0x00008D57 }, /* GL_MAX_SAMPLES */ - { 19458, 0x00009111 }, /* GL_MAX_SERVER_WAIT_TIMEOUT */ - { 19485, 0x00008504 }, /* GL_MAX_SHININESS_NV */ - { 19505, 0x00008505 }, /* GL_MAX_SPOT_EXPONENT_NV */ - { 19529, 0x00008871 }, /* GL_MAX_TEXTURE_COORDS */ - { 19551, 0x00008871 }, /* GL_MAX_TEXTURE_COORDS_ARB */ - { 19577, 0x00008872 }, /* GL_MAX_TEXTURE_IMAGE_UNITS */ - { 19604, 0x00008872 }, /* GL_MAX_TEXTURE_IMAGE_UNITS_ARB */ - { 19635, 0x000084FD }, /* GL_MAX_TEXTURE_LOD_BIAS */ - { 19659, 0x000084FF }, /* GL_MAX_TEXTURE_MAX_ANISOTROPY_EXT */ - { 19693, 0x00000D33 }, /* GL_MAX_TEXTURE_SIZE */ - { 19713, 0x00000D39 }, /* GL_MAX_TEXTURE_STACK_DEPTH */ - { 19740, 0x000084E2 }, /* GL_MAX_TEXTURE_UNITS */ - { 19761, 0x000084E2 }, /* GL_MAX_TEXTURE_UNITS_ARB */ - { 19786, 0x0000862F }, /* GL_MAX_TRACK_MATRICES_NV */ - { 19811, 0x0000862E }, /* GL_MAX_TRACK_MATRIX_STACK_DEPTH_NV */ - { 19846, 0x00008B4B }, /* GL_MAX_VARYING_FLOATS */ - { 19868, 0x00008B4B }, /* GL_MAX_VARYING_FLOATS_ARB */ - { 19894, 0x00008869 }, /* GL_MAX_VERTEX_ATTRIBS */ - { 19916, 0x00008869 }, /* GL_MAX_VERTEX_ATTRIBS_ARB */ - { 19942, 0x00008B4C }, /* GL_MAX_VERTEX_TEXTURE_IMAGE_UNITS */ - { 19976, 0x00008B4C }, /* GL_MAX_VERTEX_TEXTURE_IMAGE_UNITS_ARB */ - { 20014, 0x00008B4A }, /* GL_MAX_VERTEX_UNIFORM_COMPONENTS */ - { 20047, 0x00008B4A }, /* GL_MAX_VERTEX_UNIFORM_COMPONENTS_ARB */ - { 20084, 0x000086A4 }, /* GL_MAX_VERTEX_UNITS_ARB */ - { 20108, 0x00000D3A }, /* GL_MAX_VIEWPORT_DIMS */ - { 20129, 0x00008007 }, /* GL_MIN */ - { 20136, 0x0000802E }, /* GL_MINMAX */ - { 20146, 0x0000802E }, /* GL_MINMAX_EXT */ - { 20160, 0x0000802F }, /* GL_MINMAX_FORMAT */ - { 20177, 0x0000802F }, /* GL_MINMAX_FORMAT_EXT */ - { 20198, 0x00008030 }, /* GL_MINMAX_SINK */ - { 20213, 0x00008030 }, /* GL_MINMAX_SINK_EXT */ - { 20232, 0x00008007 }, /* GL_MIN_EXT */ - { 20243, 0x00008370 }, /* GL_MIRRORED_REPEAT */ - { 20262, 0x00008370 }, /* GL_MIRRORED_REPEAT_ARB */ - { 20285, 0x00008370 }, /* GL_MIRRORED_REPEAT_IBM */ - { 20308, 0x00008742 }, /* GL_MIRROR_CLAMP_ATI */ - { 20328, 0x00008742 }, /* GL_MIRROR_CLAMP_EXT */ - { 20348, 0x00008912 }, /* GL_MIRROR_CLAMP_TO_BORDER_EXT */ - { 20378, 0x00008743 }, /* GL_MIRROR_CLAMP_TO_EDGE_ATI */ - { 20406, 0x00008743 }, /* GL_MIRROR_CLAMP_TO_EDGE_EXT */ - { 20434, 0x00001700 }, /* GL_MODELVIEW */ - { 20447, 0x00001700 }, /* GL_MODELVIEW0_ARB */ - { 20465, 0x0000872A }, /* GL_MODELVIEW10_ARB */ - { 20484, 0x0000872B }, /* GL_MODELVIEW11_ARB */ - { 20503, 0x0000872C }, /* GL_MODELVIEW12_ARB */ - { 20522, 0x0000872D }, /* GL_MODELVIEW13_ARB */ - { 20541, 0x0000872E }, /* GL_MODELVIEW14_ARB */ - { 20560, 0x0000872F }, /* GL_MODELVIEW15_ARB */ - { 20579, 0x00008730 }, /* GL_MODELVIEW16_ARB */ - { 20598, 0x00008731 }, /* GL_MODELVIEW17_ARB */ - { 20617, 0x00008732 }, /* GL_MODELVIEW18_ARB */ - { 20636, 0x00008733 }, /* GL_MODELVIEW19_ARB */ - { 20655, 0x0000850A }, /* GL_MODELVIEW1_ARB */ - { 20673, 0x00008734 }, /* GL_MODELVIEW20_ARB */ - { 20692, 0x00008735 }, /* GL_MODELVIEW21_ARB */ - { 20711, 0x00008736 }, /* GL_MODELVIEW22_ARB */ - { 20730, 0x00008737 }, /* GL_MODELVIEW23_ARB */ - { 20749, 0x00008738 }, /* GL_MODELVIEW24_ARB */ - { 20768, 0x00008739 }, /* GL_MODELVIEW25_ARB */ - { 20787, 0x0000873A }, /* GL_MODELVIEW26_ARB */ - { 20806, 0x0000873B }, /* GL_MODELVIEW27_ARB */ - { 20825, 0x0000873C }, /* GL_MODELVIEW28_ARB */ - { 20844, 0x0000873D }, /* GL_MODELVIEW29_ARB */ - { 20863, 0x00008722 }, /* GL_MODELVIEW2_ARB */ - { 20881, 0x0000873E }, /* GL_MODELVIEW30_ARB */ - { 20900, 0x0000873F }, /* GL_MODELVIEW31_ARB */ - { 20919, 0x00008723 }, /* GL_MODELVIEW3_ARB */ - { 20937, 0x00008724 }, /* GL_MODELVIEW4_ARB */ - { 20955, 0x00008725 }, /* GL_MODELVIEW5_ARB */ - { 20973, 0x00008726 }, /* GL_MODELVIEW6_ARB */ - { 20991, 0x00008727 }, /* GL_MODELVIEW7_ARB */ - { 21009, 0x00008728 }, /* GL_MODELVIEW8_ARB */ - { 21027, 0x00008729 }, /* GL_MODELVIEW9_ARB */ - { 21045, 0x00000BA6 }, /* GL_MODELVIEW_MATRIX */ - { 21065, 0x00008629 }, /* GL_MODELVIEW_PROJECTION_NV */ - { 21092, 0x00000BA3 }, /* GL_MODELVIEW_STACK_DEPTH */ - { 21117, 0x00002100 }, /* GL_MODULATE */ - { 21129, 0x00008744 }, /* GL_MODULATE_ADD_ATI */ - { 21149, 0x00008745 }, /* GL_MODULATE_SIGNED_ADD_ATI */ - { 21176, 0x00008746 }, /* GL_MODULATE_SUBTRACT_ATI */ - { 21201, 0x00000103 }, /* GL_MULT */ - { 21209, 0x0000809D }, /* GL_MULTISAMPLE */ - { 21224, 0x000086B2 }, /* GL_MULTISAMPLE_3DFX */ - { 21244, 0x0000809D }, /* GL_MULTISAMPLE_ARB */ - { 21263, 0x20000000 }, /* GL_MULTISAMPLE_BIT */ - { 21282, 0x20000000 }, /* GL_MULTISAMPLE_BIT_3DFX */ - { 21306, 0x20000000 }, /* GL_MULTISAMPLE_BIT_ARB */ - { 21329, 0x00008534 }, /* GL_MULTISAMPLE_FILTER_HINT_NV */ - { 21359, 0x00002A25 }, /* GL_N3F_V3F */ - { 21370, 0x00000D70 }, /* GL_NAME_STACK_DEPTH */ - { 21390, 0x0000150E }, /* GL_NAND */ - { 21398, 0x00002600 }, /* GL_NEAREST */ - { 21409, 0x0000844E }, /* GL_NEAREST_CLIPMAP_LINEAR_SGIX */ - { 21440, 0x0000844D }, /* GL_NEAREST_CLIPMAP_NEAREST_SGIX */ - { 21472, 0x00002702 }, /* GL_NEAREST_MIPMAP_LINEAR */ - { 21497, 0x00002700 }, /* GL_NEAREST_MIPMAP_NEAREST */ - { 21523, 0x00000200 }, /* GL_NEVER */ - { 21532, 0x00001102 }, /* GL_NICEST */ - { 21542, 0x00000000 }, /* GL_NONE */ - { 21550, 0x00001505 }, /* GL_NOOP */ - { 21558, 0x00001508 }, /* GL_NOR */ - { 21565, 0x00000BA1 }, /* GL_NORMALIZE */ - { 21578, 0x00008075 }, /* GL_NORMAL_ARRAY */ - { 21594, 0x00008897 }, /* GL_NORMAL_ARRAY_BUFFER_BINDING */ - { 21625, 0x00008897 }, /* GL_NORMAL_ARRAY_BUFFER_BINDING_ARB */ - { 21660, 0x0000808F }, /* GL_NORMAL_ARRAY_POINTER */ - { 21684, 0x0000807F }, /* GL_NORMAL_ARRAY_STRIDE */ - { 21707, 0x0000807E }, /* GL_NORMAL_ARRAY_TYPE */ - { 21728, 0x00008511 }, /* GL_NORMAL_MAP */ - { 21742, 0x00008511 }, /* GL_NORMAL_MAP_ARB */ - { 21760, 0x00008511 }, /* GL_NORMAL_MAP_NV */ - { 21777, 0x00000205 }, /* GL_NOTEQUAL */ - { 21789, 0x00000000 }, /* GL_NO_ERROR */ - { 21801, 0x000086A2 }, /* GL_NUM_COMPRESSED_TEXTURE_FORMATS */ - { 21835, 0x000086A2 }, /* GL_NUM_COMPRESSED_TEXTURE_FORMATS_ARB */ - { 21873, 0x00008B89 }, /* GL_OBJECT_ACTIVE_ATTRIBUTES_ARB */ - { 21905, 0x00008B8A }, /* GL_OBJECT_ACTIVE_ATTRIBUTE_MAX_LENGTH_ARB */ - { 21947, 0x00008B86 }, /* GL_OBJECT_ACTIVE_UNIFORMS_ARB */ - { 21977, 0x00008B87 }, /* GL_OBJECT_ACTIVE_UNIFORM_MAX_LENGTH_ARB */ - { 22017, 0x00008B85 }, /* GL_OBJECT_ATTACHED_OBJECTS_ARB */ - { 22048, 0x00008B81 }, /* GL_OBJECT_COMPILE_STATUS_ARB */ - { 22077, 0x00008B80 }, /* GL_OBJECT_DELETE_STATUS_ARB */ - { 22105, 0x00008B84 }, /* GL_OBJECT_INFO_LOG_LENGTH_ARB */ - { 22135, 0x00002401 }, /* GL_OBJECT_LINEAR */ - { 22152, 0x00008B82 }, /* GL_OBJECT_LINK_STATUS_ARB */ - { 22178, 0x00002501 }, /* GL_OBJECT_PLANE */ - { 22194, 0x00008B88 }, /* GL_OBJECT_SHADER_SOURCE_LENGTH_ARB */ - { 22229, 0x00008B4F }, /* GL_OBJECT_SUBTYPE_ARB */ - { 22251, 0x00009112 }, /* GL_OBJECT_TYPE */ - { 22266, 0x00008B4E }, /* GL_OBJECT_TYPE_ARB */ - { 22285, 0x00008B83 }, /* GL_OBJECT_VALIDATE_STATUS_ARB */ - { 22315, 0x00008165 }, /* GL_OCCLUSION_TEST_HP */ - { 22336, 0x00008166 }, /* GL_OCCLUSION_TEST_RESULT_HP */ - { 22364, 0x00000001 }, /* GL_ONE */ - { 22371, 0x00008004 }, /* GL_ONE_MINUS_CONSTANT_ALPHA */ - { 22399, 0x00008004 }, /* GL_ONE_MINUS_CONSTANT_ALPHA_EXT */ - { 22431, 0x00008002 }, /* GL_ONE_MINUS_CONSTANT_COLOR */ - { 22459, 0x00008002 }, /* GL_ONE_MINUS_CONSTANT_COLOR_EXT */ - { 22491, 0x00000305 }, /* GL_ONE_MINUS_DST_ALPHA */ - { 22514, 0x00000307 }, /* GL_ONE_MINUS_DST_COLOR */ - { 22537, 0x00000303 }, /* GL_ONE_MINUS_SRC_ALPHA */ - { 22560, 0x00000301 }, /* GL_ONE_MINUS_SRC_COLOR */ - { 22583, 0x00008598 }, /* GL_OPERAND0_ALPHA */ - { 22601, 0x00008598 }, /* GL_OPERAND0_ALPHA_ARB */ - { 22623, 0x00008598 }, /* GL_OPERAND0_ALPHA_EXT */ - { 22645, 0x00008590 }, /* GL_OPERAND0_RGB */ - { 22661, 0x00008590 }, /* GL_OPERAND0_RGB_ARB */ - { 22681, 0x00008590 }, /* GL_OPERAND0_RGB_EXT */ - { 22701, 0x00008599 }, /* GL_OPERAND1_ALPHA */ - { 22719, 0x00008599 }, /* GL_OPERAND1_ALPHA_ARB */ - { 22741, 0x00008599 }, /* GL_OPERAND1_ALPHA_EXT */ - { 22763, 0x00008591 }, /* GL_OPERAND1_RGB */ - { 22779, 0x00008591 }, /* GL_OPERAND1_RGB_ARB */ - { 22799, 0x00008591 }, /* GL_OPERAND1_RGB_EXT */ - { 22819, 0x0000859A }, /* GL_OPERAND2_ALPHA */ - { 22837, 0x0000859A }, /* GL_OPERAND2_ALPHA_ARB */ - { 22859, 0x0000859A }, /* GL_OPERAND2_ALPHA_EXT */ - { 22881, 0x00008592 }, /* GL_OPERAND2_RGB */ - { 22897, 0x00008592 }, /* GL_OPERAND2_RGB_ARB */ - { 22917, 0x00008592 }, /* GL_OPERAND2_RGB_EXT */ - { 22937, 0x0000859B }, /* GL_OPERAND3_ALPHA_NV */ - { 22958, 0x00008593 }, /* GL_OPERAND3_RGB_NV */ - { 22977, 0x00001507 }, /* GL_OR */ - { 22983, 0x00000A01 }, /* GL_ORDER */ - { 22992, 0x0000150D }, /* GL_OR_INVERTED */ - { 23007, 0x0000150B }, /* GL_OR_REVERSE */ - { 23021, 0x00000505 }, /* GL_OUT_OF_MEMORY */ - { 23038, 0x00000D05 }, /* GL_PACK_ALIGNMENT */ - { 23056, 0x0000806C }, /* GL_PACK_IMAGE_HEIGHT */ - { 23077, 0x00008758 }, /* GL_PACK_INVERT_MESA */ - { 23097, 0x00000D01 }, /* GL_PACK_LSB_FIRST */ - { 23115, 0x00000D02 }, /* GL_PACK_ROW_LENGTH */ - { 23134, 0x0000806B }, /* GL_PACK_SKIP_IMAGES */ - { 23154, 0x00000D04 }, /* GL_PACK_SKIP_PIXELS */ - { 23174, 0x00000D03 }, /* GL_PACK_SKIP_ROWS */ - { 23192, 0x00000D00 }, /* GL_PACK_SWAP_BYTES */ - { 23211, 0x00008B92 }, /* GL_PALETTE4_R5_G6_B5_OES */ - { 23236, 0x00008B94 }, /* GL_PALETTE4_RGB5_A1_OES */ - { 23260, 0x00008B90 }, /* GL_PALETTE4_RGB8_OES */ - { 23281, 0x00008B93 }, /* GL_PALETTE4_RGBA4_OES */ - { 23303, 0x00008B91 }, /* GL_PALETTE4_RGBA8_OES */ - { 23325, 0x00008B97 }, /* GL_PALETTE8_R5_G6_B5_OES */ - { 23350, 0x00008B99 }, /* GL_PALETTE8_RGB5_A1_OES */ - { 23374, 0x00008B95 }, /* GL_PALETTE8_RGB8_OES */ - { 23395, 0x00008B98 }, /* GL_PALETTE8_RGBA4_OES */ - { 23417, 0x00008B96 }, /* GL_PALETTE8_RGBA8_OES */ - { 23439, 0x00000700 }, /* GL_PASS_THROUGH_TOKEN */ - { 23461, 0x00000C50 }, /* GL_PERSPECTIVE_CORRECTION_HINT */ - { 23492, 0x00000C79 }, /* GL_PIXEL_MAP_A_TO_A */ - { 23512, 0x00000CB9 }, /* GL_PIXEL_MAP_A_TO_A_SIZE */ - { 23537, 0x00000C78 }, /* GL_PIXEL_MAP_B_TO_B */ - { 23557, 0x00000CB8 }, /* GL_PIXEL_MAP_B_TO_B_SIZE */ - { 23582, 0x00000C77 }, /* GL_PIXEL_MAP_G_TO_G */ - { 23602, 0x00000CB7 }, /* GL_PIXEL_MAP_G_TO_G_SIZE */ - { 23627, 0x00000C75 }, /* GL_PIXEL_MAP_I_TO_A */ - { 23647, 0x00000CB5 }, /* GL_PIXEL_MAP_I_TO_A_SIZE */ - { 23672, 0x00000C74 }, /* GL_PIXEL_MAP_I_TO_B */ - { 23692, 0x00000CB4 }, /* GL_PIXEL_MAP_I_TO_B_SIZE */ - { 23717, 0x00000C73 }, /* GL_PIXEL_MAP_I_TO_G */ - { 23737, 0x00000CB3 }, /* GL_PIXEL_MAP_I_TO_G_SIZE */ - { 23762, 0x00000C70 }, /* GL_PIXEL_MAP_I_TO_I */ - { 23782, 0x00000CB0 }, /* GL_PIXEL_MAP_I_TO_I_SIZE */ - { 23807, 0x00000C72 }, /* GL_PIXEL_MAP_I_TO_R */ - { 23827, 0x00000CB2 }, /* GL_PIXEL_MAP_I_TO_R_SIZE */ - { 23852, 0x00000C76 }, /* GL_PIXEL_MAP_R_TO_R */ - { 23872, 0x00000CB6 }, /* GL_PIXEL_MAP_R_TO_R_SIZE */ - { 23897, 0x00000C71 }, /* GL_PIXEL_MAP_S_TO_S */ - { 23917, 0x00000CB1 }, /* GL_PIXEL_MAP_S_TO_S_SIZE */ - { 23942, 0x00000020 }, /* GL_PIXEL_MODE_BIT */ - { 23960, 0x000088EB }, /* GL_PIXEL_PACK_BUFFER */ - { 23981, 0x000088ED }, /* GL_PIXEL_PACK_BUFFER_BINDING */ - { 24010, 0x000088ED }, /* GL_PIXEL_PACK_BUFFER_BINDING_EXT */ - { 24043, 0x000088EB }, /* GL_PIXEL_PACK_BUFFER_EXT */ - { 24068, 0x000088EC }, /* GL_PIXEL_UNPACK_BUFFER */ - { 24091, 0x000088EF }, /* GL_PIXEL_UNPACK_BUFFER_BINDING */ - { 24122, 0x000088EF }, /* GL_PIXEL_UNPACK_BUFFER_BINDING_EXT */ - { 24157, 0x000088EC }, /* GL_PIXEL_UNPACK_BUFFER_EXT */ - { 24184, 0x00001B00 }, /* GL_POINT */ - { 24193, 0x00000000 }, /* GL_POINTS */ - { 24203, 0x00000002 }, /* GL_POINT_BIT */ - { 24216, 0x00008129 }, /* GL_POINT_DISTANCE_ATTENUATION */ - { 24246, 0x00008129 }, /* GL_POINT_DISTANCE_ATTENUATION_ARB */ - { 24280, 0x00008129 }, /* GL_POINT_DISTANCE_ATTENUATION_EXT */ - { 24314, 0x00008129 }, /* GL_POINT_DISTANCE_ATTENUATION_SGIS */ - { 24349, 0x00008128 }, /* GL_POINT_FADE_THRESHOLD_SIZE */ - { 24378, 0x00008128 }, /* GL_POINT_FADE_THRESHOLD_SIZE_ARB */ - { 24411, 0x00008128 }, /* GL_POINT_FADE_THRESHOLD_SIZE_EXT */ - { 24444, 0x00008128 }, /* GL_POINT_FADE_THRESHOLD_SIZE_SGIS */ - { 24478, 0x00000B11 }, /* GL_POINT_SIZE */ - { 24492, 0x00000B13 }, /* GL_POINT_SIZE_GRANULARITY */ - { 24518, 0x00008127 }, /* GL_POINT_SIZE_MAX */ - { 24536, 0x00008127 }, /* GL_POINT_SIZE_MAX_ARB */ - { 24558, 0x00008127 }, /* GL_POINT_SIZE_MAX_EXT */ - { 24580, 0x00008127 }, /* GL_POINT_SIZE_MAX_SGIS */ - { 24603, 0x00008126 }, /* GL_POINT_SIZE_MIN */ - { 24621, 0x00008126 }, /* GL_POINT_SIZE_MIN_ARB */ - { 24643, 0x00008126 }, /* GL_POINT_SIZE_MIN_EXT */ - { 24665, 0x00008126 }, /* GL_POINT_SIZE_MIN_SGIS */ - { 24688, 0x00000B12 }, /* GL_POINT_SIZE_RANGE */ - { 24708, 0x00000B10 }, /* GL_POINT_SMOOTH */ - { 24724, 0x00000C51 }, /* GL_POINT_SMOOTH_HINT */ - { 24745, 0x00008861 }, /* GL_POINT_SPRITE */ - { 24761, 0x00008861 }, /* GL_POINT_SPRITE_ARB */ - { 24781, 0x00008CA0 }, /* GL_POINT_SPRITE_COORD_ORIGIN */ - { 24810, 0x00008861 }, /* GL_POINT_SPRITE_NV */ - { 24829, 0x00008863 }, /* GL_POINT_SPRITE_R_MODE_NV */ - { 24855, 0x00000701 }, /* GL_POINT_TOKEN */ - { 24870, 0x00000009 }, /* GL_POLYGON */ - { 24881, 0x00000008 }, /* GL_POLYGON_BIT */ - { 24896, 0x00000B40 }, /* GL_POLYGON_MODE */ - { 24912, 0x00008039 }, /* GL_POLYGON_OFFSET_BIAS */ - { 24935, 0x00008038 }, /* GL_POLYGON_OFFSET_FACTOR */ - { 24960, 0x00008037 }, /* GL_POLYGON_OFFSET_FILL */ - { 24983, 0x00002A02 }, /* GL_POLYGON_OFFSET_LINE */ - { 25006, 0x00002A01 }, /* GL_POLYGON_OFFSET_POINT */ - { 25030, 0x00002A00 }, /* GL_POLYGON_OFFSET_UNITS */ - { 25054, 0x00000B41 }, /* GL_POLYGON_SMOOTH */ - { 25072, 0x00000C53 }, /* GL_POLYGON_SMOOTH_HINT */ - { 25095, 0x00000B42 }, /* GL_POLYGON_STIPPLE */ - { 25114, 0x00000010 }, /* GL_POLYGON_STIPPLE_BIT */ - { 25137, 0x00000703 }, /* GL_POLYGON_TOKEN */ - { 25154, 0x00001203 }, /* GL_POSITION */ - { 25166, 0x000080BB }, /* GL_POST_COLOR_MATRIX_ALPHA_BIAS */ - { 25198, 0x000080BB }, /* GL_POST_COLOR_MATRIX_ALPHA_BIAS_SGI */ - { 25234, 0x000080B7 }, /* GL_POST_COLOR_MATRIX_ALPHA_SCALE */ - { 25267, 0x000080B7 }, /* GL_POST_COLOR_MATRIX_ALPHA_SCALE_SGI */ - { 25304, 0x000080BA }, /* GL_POST_COLOR_MATRIX_BLUE_BIAS */ - { 25335, 0x000080BA }, /* GL_POST_COLOR_MATRIX_BLUE_BIAS_SGI */ - { 25370, 0x000080B6 }, /* GL_POST_COLOR_MATRIX_BLUE_SCALE */ - { 25402, 0x000080B6 }, /* GL_POST_COLOR_MATRIX_BLUE_SCALE_SGI */ - { 25438, 0x000080D2 }, /* GL_POST_COLOR_MATRIX_COLOR_TABLE */ - { 25471, 0x000080B9 }, /* GL_POST_COLOR_MATRIX_GREEN_BIAS */ - { 25503, 0x000080B9 }, /* GL_POST_COLOR_MATRIX_GREEN_BIAS_SGI */ - { 25539, 0x000080B5 }, /* GL_POST_COLOR_MATRIX_GREEN_SCALE */ - { 25572, 0x000080B5 }, /* GL_POST_COLOR_MATRIX_GREEN_SCALE_SGI */ - { 25609, 0x000080B8 }, /* GL_POST_COLOR_MATRIX_RED_BIAS */ - { 25639, 0x000080B8 }, /* GL_POST_COLOR_MATRIX_RED_BIAS_SGI */ - { 25673, 0x000080B4 }, /* GL_POST_COLOR_MATRIX_RED_SCALE */ - { 25704, 0x000080B4 }, /* GL_POST_COLOR_MATRIX_RED_SCALE_SGI */ - { 25739, 0x00008023 }, /* GL_POST_CONVOLUTION_ALPHA_BIAS */ - { 25770, 0x00008023 }, /* GL_POST_CONVOLUTION_ALPHA_BIAS_EXT */ - { 25805, 0x0000801F }, /* GL_POST_CONVOLUTION_ALPHA_SCALE */ - { 25837, 0x0000801F }, /* GL_POST_CONVOLUTION_ALPHA_SCALE_EXT */ - { 25873, 0x00008022 }, /* GL_POST_CONVOLUTION_BLUE_BIAS */ - { 25903, 0x00008022 }, /* GL_POST_CONVOLUTION_BLUE_BIAS_EXT */ - { 25937, 0x0000801E }, /* GL_POST_CONVOLUTION_BLUE_SCALE */ - { 25968, 0x0000801E }, /* GL_POST_CONVOLUTION_BLUE_SCALE_EXT */ - { 26003, 0x000080D1 }, /* GL_POST_CONVOLUTION_COLOR_TABLE */ - { 26035, 0x00008021 }, /* GL_POST_CONVOLUTION_GREEN_BIAS */ - { 26066, 0x00008021 }, /* GL_POST_CONVOLUTION_GREEN_BIAS_EXT */ - { 26101, 0x0000801D }, /* GL_POST_CONVOLUTION_GREEN_SCALE */ - { 26133, 0x0000801D }, /* GL_POST_CONVOLUTION_GREEN_SCALE_EXT */ - { 26169, 0x00008020 }, /* GL_POST_CONVOLUTION_RED_BIAS */ - { 26198, 0x00008020 }, /* GL_POST_CONVOLUTION_RED_BIAS_EXT */ - { 26231, 0x0000801C }, /* GL_POST_CONVOLUTION_RED_SCALE */ - { 26261, 0x0000801C }, /* GL_POST_CONVOLUTION_RED_SCALE_EXT */ - { 26295, 0x0000817B }, /* GL_POST_TEXTURE_FILTER_BIAS_RANGE_SGIX */ - { 26334, 0x00008179 }, /* GL_POST_TEXTURE_FILTER_BIAS_SGIX */ - { 26367, 0x0000817C }, /* GL_POST_TEXTURE_FILTER_SCALE_RANGE_SGIX */ - { 26407, 0x0000817A }, /* GL_POST_TEXTURE_FILTER_SCALE_SGIX */ - { 26441, 0x00008578 }, /* GL_PREVIOUS */ - { 26453, 0x00008578 }, /* GL_PREVIOUS_ARB */ - { 26469, 0x00008578 }, /* GL_PREVIOUS_EXT */ - { 26485, 0x00008577 }, /* GL_PRIMARY_COLOR */ - { 26502, 0x00008577 }, /* GL_PRIMARY_COLOR_ARB */ - { 26523, 0x00008577 }, /* GL_PRIMARY_COLOR_EXT */ - { 26544, 0x000088B0 }, /* GL_PROGRAM_ADDRESS_REGISTERS_ARB */ - { 26577, 0x00008805 }, /* GL_PROGRAM_ALU_INSTRUCTIONS_ARB */ - { 26609, 0x000088AC }, /* GL_PROGRAM_ATTRIBS_ARB */ - { 26632, 0x00008677 }, /* GL_PROGRAM_BINDING_ARB */ - { 26655, 0x0000864B }, /* GL_PROGRAM_ERROR_POSITION_ARB */ - { 26685, 0x0000864B }, /* GL_PROGRAM_ERROR_POSITION_NV */ - { 26714, 0x00008874 }, /* GL_PROGRAM_ERROR_STRING_ARB */ - { 26742, 0x00008876 }, /* GL_PROGRAM_FORMAT_ARB */ - { 26764, 0x00008875 }, /* GL_PROGRAM_FORMAT_ASCII_ARB */ - { 26792, 0x000088A0 }, /* GL_PROGRAM_INSTRUCTIONS_ARB */ - { 26820, 0x00008627 }, /* GL_PROGRAM_LENGTH_ARB */ - { 26842, 0x00008627 }, /* GL_PROGRAM_LENGTH_NV */ - { 26863, 0x000088B2 }, /* GL_PROGRAM_NATIVE_ADDRESS_REGISTERS_ARB */ - { 26903, 0x00008808 }, /* GL_PROGRAM_NATIVE_ALU_INSTRUCTIONS_ARB */ - { 26942, 0x000088AE }, /* GL_PROGRAM_NATIVE_ATTRIBS_ARB */ - { 26972, 0x000088A2 }, /* GL_PROGRAM_NATIVE_INSTRUCTIONS_ARB */ - { 27007, 0x000088AA }, /* GL_PROGRAM_NATIVE_PARAMETERS_ARB */ - { 27040, 0x000088A6 }, /* GL_PROGRAM_NATIVE_TEMPORARIES_ARB */ - { 27074, 0x0000880A }, /* GL_PROGRAM_NATIVE_TEX_INDIRECTIONS_ARB */ - { 27113, 0x00008809 }, /* GL_PROGRAM_NATIVE_TEX_INSTRUCTIONS_ARB */ - { 27152, 0x00008B40 }, /* GL_PROGRAM_OBJECT_ARB */ - { 27174, 0x000088A8 }, /* GL_PROGRAM_PARAMETERS_ARB */ - { 27200, 0x00008644 }, /* GL_PROGRAM_PARAMETER_NV */ - { 27224, 0x00008647 }, /* GL_PROGRAM_RESIDENT_NV */ - { 27247, 0x00008628 }, /* GL_PROGRAM_STRING_ARB */ - { 27269, 0x00008628 }, /* GL_PROGRAM_STRING_NV */ - { 27290, 0x00008646 }, /* GL_PROGRAM_TARGET_NV */ - { 27311, 0x000088A4 }, /* GL_PROGRAM_TEMPORARIES_ARB */ - { 27338, 0x00008807 }, /* GL_PROGRAM_TEX_INDIRECTIONS_ARB */ - { 27370, 0x00008806 }, /* GL_PROGRAM_TEX_INSTRUCTIONS_ARB */ - { 27402, 0x000088B6 }, /* GL_PROGRAM_UNDER_NATIVE_LIMITS_ARB */ - { 27437, 0x00001701 }, /* GL_PROJECTION */ - { 27451, 0x00000BA7 }, /* GL_PROJECTION_MATRIX */ - { 27472, 0x00000BA4 }, /* GL_PROJECTION_STACK_DEPTH */ - { 27498, 0x00008E4F }, /* GL_PROVOKING_VERTEX_EXT */ - { 27522, 0x000080D3 }, /* GL_PROXY_COLOR_TABLE */ - { 27543, 0x00008025 }, /* GL_PROXY_HISTOGRAM */ - { 27562, 0x00008025 }, /* GL_PROXY_HISTOGRAM_EXT */ - { 27585, 0x000080D5 }, /* GL_PROXY_POST_COLOR_MATRIX_COLOR_TABLE */ - { 27624, 0x000080D4 }, /* GL_PROXY_POST_CONVOLUTION_COLOR_TABLE */ - { 27662, 0x00008063 }, /* GL_PROXY_TEXTURE_1D */ - { 27682, 0x00008C19 }, /* GL_PROXY_TEXTURE_1D_ARRAY_EXT */ - { 27712, 0x00008063 }, /* GL_PROXY_TEXTURE_1D_EXT */ - { 27736, 0x00008064 }, /* GL_PROXY_TEXTURE_2D */ - { 27756, 0x00008C1B }, /* GL_PROXY_TEXTURE_2D_ARRAY_EXT */ - { 27786, 0x00008064 }, /* GL_PROXY_TEXTURE_2D_EXT */ - { 27810, 0x00008070 }, /* GL_PROXY_TEXTURE_3D */ - { 27830, 0x000080BD }, /* GL_PROXY_TEXTURE_COLOR_TABLE_SGI */ - { 27863, 0x0000851B }, /* GL_PROXY_TEXTURE_CUBE_MAP */ - { 27889, 0x0000851B }, /* GL_PROXY_TEXTURE_CUBE_MAP_ARB */ - { 27919, 0x000084F7 }, /* GL_PROXY_TEXTURE_RECTANGLE_ARB */ - { 27950, 0x000084F7 }, /* GL_PROXY_TEXTURE_RECTANGLE_NV */ - { 27980, 0x00002003 }, /* GL_Q */ - { 27985, 0x00001209 }, /* GL_QUADRATIC_ATTENUATION */ - { 28010, 0x00000007 }, /* GL_QUADS */ - { 28019, 0x00008E4C }, /* GL_QUADS_FOLLOW_PROVOKING_VERTEX_CONVENTION_EXT */ - { 28067, 0x00008614 }, /* GL_QUAD_MESH_SUN */ - { 28084, 0x00000008 }, /* GL_QUAD_STRIP */ - { 28098, 0x00008864 }, /* GL_QUERY_COUNTER_BITS */ - { 28120, 0x00008864 }, /* GL_QUERY_COUNTER_BITS_ARB */ - { 28146, 0x00008866 }, /* GL_QUERY_RESULT */ - { 28162, 0x00008866 }, /* GL_QUERY_RESULT_ARB */ - { 28182, 0x00008867 }, /* GL_QUERY_RESULT_AVAILABLE */ - { 28208, 0x00008867 }, /* GL_QUERY_RESULT_AVAILABLE_ARB */ - { 28238, 0x00002002 }, /* GL_R */ - { 28243, 0x00002A10 }, /* GL_R3_G3_B2 */ - { 28255, 0x00019262 }, /* GL_RASTER_POSITION_UNCLIPPED_IBM */ - { 28288, 0x00000C02 }, /* GL_READ_BUFFER */ - { 28303, 0x00008CA8 }, /* GL_READ_FRAMEBUFFER */ - { 28323, 0x00008CAA }, /* GL_READ_FRAMEBUFFER_BINDING_EXT */ - { 28355, 0x00008CA8 }, /* GL_READ_FRAMEBUFFER_EXT */ - { 28379, 0x000088B8 }, /* GL_READ_ONLY */ - { 28392, 0x000088B8 }, /* GL_READ_ONLY_ARB */ - { 28409, 0x000088BA }, /* GL_READ_WRITE */ - { 28423, 0x000088BA }, /* GL_READ_WRITE_ARB */ - { 28441, 0x00001903 }, /* GL_RED */ - { 28448, 0x00008016 }, /* GL_REDUCE */ - { 28458, 0x00008016 }, /* GL_REDUCE_EXT */ - { 28472, 0x00000D15 }, /* GL_RED_BIAS */ - { 28484, 0x00000D52 }, /* GL_RED_BITS */ - { 28496, 0x00000D14 }, /* GL_RED_SCALE */ - { 28509, 0x00008512 }, /* GL_REFLECTION_MAP */ - { 28527, 0x00008512 }, /* GL_REFLECTION_MAP_ARB */ - { 28549, 0x00008512 }, /* GL_REFLECTION_MAP_NV */ - { 28570, 0x00001C00 }, /* GL_RENDER */ - { 28580, 0x00008D41 }, /* GL_RENDERBUFFER */ - { 28596, 0x00008D53 }, /* GL_RENDERBUFFER_ALPHA_SIZE */ - { 28623, 0x00008CA7 }, /* GL_RENDERBUFFER_BINDING_EXT */ - { 28651, 0x00008D52 }, /* GL_RENDERBUFFER_BLUE_SIZE */ - { 28677, 0x00008D54 }, /* GL_RENDERBUFFER_DEPTH_SIZE */ - { 28704, 0x00008D41 }, /* GL_RENDERBUFFER_EXT */ - { 28724, 0x00008D51 }, /* GL_RENDERBUFFER_GREEN_SIZE */ - { 28751, 0x00008D43 }, /* GL_RENDERBUFFER_HEIGHT */ - { 28774, 0x00008D43 }, /* GL_RENDERBUFFER_HEIGHT_EXT */ - { 28801, 0x00008D44 }, /* GL_RENDERBUFFER_INTERNAL_FORMAT */ - { 28833, 0x00008D44 }, /* GL_RENDERBUFFER_INTERNAL_FORMAT_EXT */ - { 28869, 0x00008D50 }, /* GL_RENDERBUFFER_RED_SIZE */ - { 28894, 0x00008CAB }, /* GL_RENDERBUFFER_SAMPLES */ - { 28918, 0x00008D55 }, /* GL_RENDERBUFFER_STENCIL_SIZE */ - { 28947, 0x00008D42 }, /* GL_RENDERBUFFER_WIDTH */ - { 28969, 0x00008D42 }, /* GL_RENDERBUFFER_WIDTH_EXT */ - { 28995, 0x00001F01 }, /* GL_RENDERER */ - { 29007, 0x00000C40 }, /* GL_RENDER_MODE */ - { 29022, 0x00002901 }, /* GL_REPEAT */ - { 29032, 0x00001E01 }, /* GL_REPLACE */ - { 29043, 0x00008062 }, /* GL_REPLACE_EXT */ - { 29058, 0x00008153 }, /* GL_REPLICATE_BORDER_HP */ - { 29081, 0x0000803A }, /* GL_RESCALE_NORMAL */ - { 29099, 0x0000803A }, /* GL_RESCALE_NORMAL_EXT */ - { 29121, 0x00000102 }, /* GL_RETURN */ - { 29131, 0x00001907 }, /* GL_RGB */ - { 29138, 0x00008052 }, /* GL_RGB10 */ - { 29147, 0x00008059 }, /* GL_RGB10_A2 */ - { 29159, 0x00008059 }, /* GL_RGB10_A2_EXT */ - { 29175, 0x00008052 }, /* GL_RGB10_EXT */ - { 29188, 0x00008053 }, /* GL_RGB12 */ - { 29197, 0x00008053 }, /* GL_RGB12_EXT */ - { 29210, 0x00008054 }, /* GL_RGB16 */ - { 29219, 0x00008054 }, /* GL_RGB16_EXT */ - { 29232, 0x0000804E }, /* GL_RGB2_EXT */ - { 29244, 0x0000804F }, /* GL_RGB4 */ - { 29252, 0x0000804F }, /* GL_RGB4_EXT */ - { 29264, 0x000083A1 }, /* GL_RGB4_S3TC */ - { 29277, 0x00008050 }, /* GL_RGB5 */ - { 29285, 0x00008057 }, /* GL_RGB5_A1 */ - { 29296, 0x00008057 }, /* GL_RGB5_A1_EXT */ - { 29311, 0x00008050 }, /* GL_RGB5_EXT */ - { 29323, 0x00008051 }, /* GL_RGB8 */ - { 29331, 0x00008051 }, /* GL_RGB8_EXT */ - { 29343, 0x00001908 }, /* GL_RGBA */ - { 29351, 0x0000805A }, /* GL_RGBA12 */ - { 29361, 0x0000805A }, /* GL_RGBA12_EXT */ - { 29375, 0x0000805B }, /* GL_RGBA16 */ - { 29385, 0x0000805B }, /* GL_RGBA16_EXT */ - { 29399, 0x00008055 }, /* GL_RGBA2 */ - { 29408, 0x00008055 }, /* GL_RGBA2_EXT */ - { 29421, 0x00008056 }, /* GL_RGBA4 */ - { 29430, 0x000083A5 }, /* GL_RGBA4_DXT5_S3TC */ - { 29449, 0x00008056 }, /* GL_RGBA4_EXT */ - { 29462, 0x000083A3 }, /* GL_RGBA4_S3TC */ - { 29476, 0x00008058 }, /* GL_RGBA8 */ - { 29485, 0x00008058 }, /* GL_RGBA8_EXT */ - { 29498, 0x00008F97 }, /* GL_RGBA8_SNORM */ - { 29513, 0x000083A4 }, /* GL_RGBA_DXT5_S3TC */ - { 29531, 0x00000C31 }, /* GL_RGBA_MODE */ - { 29544, 0x000083A2 }, /* GL_RGBA_S3TC */ - { 29557, 0x00008F93 }, /* GL_RGBA_SNORM */ - { 29571, 0x000083A0 }, /* GL_RGB_S3TC */ - { 29583, 0x00008573 }, /* GL_RGB_SCALE */ - { 29596, 0x00008573 }, /* GL_RGB_SCALE_ARB */ - { 29613, 0x00008573 }, /* GL_RGB_SCALE_EXT */ - { 29630, 0x00000407 }, /* GL_RIGHT */ - { 29639, 0x00002000 }, /* GL_S */ - { 29644, 0x00008B5D }, /* GL_SAMPLER_1D */ - { 29658, 0x00008B61 }, /* GL_SAMPLER_1D_SHADOW */ - { 29679, 0x00008B5E }, /* GL_SAMPLER_2D */ - { 29693, 0x00008B62 }, /* GL_SAMPLER_2D_SHADOW */ - { 29714, 0x00008B5F }, /* GL_SAMPLER_3D */ - { 29728, 0x00008B60 }, /* GL_SAMPLER_CUBE */ - { 29744, 0x000080A9 }, /* GL_SAMPLES */ - { 29755, 0x000086B4 }, /* GL_SAMPLES_3DFX */ - { 29771, 0x000080A9 }, /* GL_SAMPLES_ARB */ - { 29786, 0x00008914 }, /* GL_SAMPLES_PASSED */ - { 29804, 0x00008914 }, /* GL_SAMPLES_PASSED_ARB */ - { 29826, 0x0000809E }, /* GL_SAMPLE_ALPHA_TO_COVERAGE */ - { 29854, 0x0000809E }, /* GL_SAMPLE_ALPHA_TO_COVERAGE_ARB */ - { 29886, 0x0000809F }, /* GL_SAMPLE_ALPHA_TO_ONE */ - { 29909, 0x0000809F }, /* GL_SAMPLE_ALPHA_TO_ONE_ARB */ - { 29936, 0x000080A8 }, /* GL_SAMPLE_BUFFERS */ - { 29954, 0x000086B3 }, /* GL_SAMPLE_BUFFERS_3DFX */ - { 29977, 0x000080A8 }, /* GL_SAMPLE_BUFFERS_ARB */ - { 29999, 0x000080A0 }, /* GL_SAMPLE_COVERAGE */ - { 30018, 0x000080A0 }, /* GL_SAMPLE_COVERAGE_ARB */ - { 30041, 0x000080AB }, /* GL_SAMPLE_COVERAGE_INVERT */ - { 30067, 0x000080AB }, /* GL_SAMPLE_COVERAGE_INVERT_ARB */ - { 30097, 0x000080AA }, /* GL_SAMPLE_COVERAGE_VALUE */ - { 30122, 0x000080AA }, /* GL_SAMPLE_COVERAGE_VALUE_ARB */ - { 30151, 0x00080000 }, /* GL_SCISSOR_BIT */ - { 30166, 0x00000C10 }, /* GL_SCISSOR_BOX */ - { 30181, 0x00000C11 }, /* GL_SCISSOR_TEST */ - { 30197, 0x0000845E }, /* GL_SECONDARY_COLOR_ARRAY */ - { 30222, 0x0000889C }, /* GL_SECONDARY_COLOR_ARRAY_BUFFER_BINDING */ - { 30262, 0x0000889C }, /* GL_SECONDARY_COLOR_ARRAY_BUFFER_BINDING_ARB */ - { 30306, 0x0000845D }, /* GL_SECONDARY_COLOR_ARRAY_POINTER */ - { 30339, 0x0000845A }, /* GL_SECONDARY_COLOR_ARRAY_SIZE */ - { 30369, 0x0000845C }, /* GL_SECONDARY_COLOR_ARRAY_STRIDE */ - { 30401, 0x0000845B }, /* GL_SECONDARY_COLOR_ARRAY_TYPE */ - { 30431, 0x00001C02 }, /* GL_SELECT */ - { 30441, 0x00000DF3 }, /* GL_SELECTION_BUFFER_POINTER */ - { 30469, 0x00000DF4 }, /* GL_SELECTION_BUFFER_SIZE */ - { 30494, 0x00008012 }, /* GL_SEPARABLE_2D */ - { 30510, 0x000081FA }, /* GL_SEPARATE_SPECULAR_COLOR */ - { 30537, 0x000081FA }, /* GL_SEPARATE_SPECULAR_COLOR_EXT */ - { 30568, 0x0000150F }, /* GL_SET */ - { 30575, 0x00008B48 }, /* GL_SHADER_OBJECT_ARB */ - { 30596, 0x00008B88 }, /* GL_SHADER_SOURCE_LENGTH */ - { 30620, 0x00008B4F }, /* GL_SHADER_TYPE */ - { 30635, 0x00000B54 }, /* GL_SHADE_MODEL */ - { 30650, 0x00008B8C }, /* GL_SHADING_LANGUAGE_VERSION */ - { 30678, 0x000080BF }, /* GL_SHADOW_AMBIENT_SGIX */ - { 30701, 0x000081FB }, /* GL_SHARED_TEXTURE_PALETTE_EXT */ - { 30731, 0x00001601 }, /* GL_SHININESS */ - { 30744, 0x00001402 }, /* GL_SHORT */ - { 30753, 0x00009119 }, /* GL_SIGNALED */ - { 30765, 0x00008F9C }, /* GL_SIGNED_NORMALIZED */ - { 30786, 0x000081F9 }, /* GL_SINGLE_COLOR */ - { 30802, 0x000081F9 }, /* GL_SINGLE_COLOR_EXT */ - { 30822, 0x000085CC }, /* GL_SLICE_ACCUM_SUN */ - { 30841, 0x00008C46 }, /* GL_SLUMINANCE */ - { 30855, 0x00008C47 }, /* GL_SLUMINANCE8 */ - { 30870, 0x00008C45 }, /* GL_SLUMINANCE8_ALPHA8 */ - { 30892, 0x00008C44 }, /* GL_SLUMINANCE_ALPHA */ - { 30912, 0x00001D01 }, /* GL_SMOOTH */ - { 30922, 0x00000B23 }, /* GL_SMOOTH_LINE_WIDTH_GRANULARITY */ - { 30955, 0x00000B22 }, /* GL_SMOOTH_LINE_WIDTH_RANGE */ - { 30982, 0x00000B13 }, /* GL_SMOOTH_POINT_SIZE_GRANULARITY */ - { 31015, 0x00000B12 }, /* GL_SMOOTH_POINT_SIZE_RANGE */ - { 31042, 0x00008588 }, /* GL_SOURCE0_ALPHA */ - { 31059, 0x00008588 }, /* GL_SOURCE0_ALPHA_ARB */ - { 31080, 0x00008588 }, /* GL_SOURCE0_ALPHA_EXT */ - { 31101, 0x00008580 }, /* GL_SOURCE0_RGB */ - { 31116, 0x00008580 }, /* GL_SOURCE0_RGB_ARB */ - { 31135, 0x00008580 }, /* GL_SOURCE0_RGB_EXT */ - { 31154, 0x00008589 }, /* GL_SOURCE1_ALPHA */ - { 31171, 0x00008589 }, /* GL_SOURCE1_ALPHA_ARB */ - { 31192, 0x00008589 }, /* GL_SOURCE1_ALPHA_EXT */ - { 31213, 0x00008581 }, /* GL_SOURCE1_RGB */ - { 31228, 0x00008581 }, /* GL_SOURCE1_RGB_ARB */ - { 31247, 0x00008581 }, /* GL_SOURCE1_RGB_EXT */ - { 31266, 0x0000858A }, /* GL_SOURCE2_ALPHA */ - { 31283, 0x0000858A }, /* GL_SOURCE2_ALPHA_ARB */ - { 31304, 0x0000858A }, /* GL_SOURCE2_ALPHA_EXT */ - { 31325, 0x00008582 }, /* GL_SOURCE2_RGB */ - { 31340, 0x00008582 }, /* GL_SOURCE2_RGB_ARB */ - { 31359, 0x00008582 }, /* GL_SOURCE2_RGB_EXT */ - { 31378, 0x0000858B }, /* GL_SOURCE3_ALPHA_NV */ - { 31398, 0x00008583 }, /* GL_SOURCE3_RGB_NV */ - { 31416, 0x00001202 }, /* GL_SPECULAR */ - { 31428, 0x00002402 }, /* GL_SPHERE_MAP */ - { 31442, 0x00001206 }, /* GL_SPOT_CUTOFF */ - { 31457, 0x00001204 }, /* GL_SPOT_DIRECTION */ - { 31475, 0x00001205 }, /* GL_SPOT_EXPONENT */ - { 31492, 0x00008588 }, /* GL_SRC0_ALPHA */ - { 31506, 0x00008580 }, /* GL_SRC0_RGB */ - { 31518, 0x00008589 }, /* GL_SRC1_ALPHA */ - { 31532, 0x00008581 }, /* GL_SRC1_RGB */ - { 31544, 0x0000858A }, /* GL_SRC2_ALPHA */ - { 31558, 0x00008582 }, /* GL_SRC2_RGB */ - { 31570, 0x00000302 }, /* GL_SRC_ALPHA */ - { 31583, 0x00000308 }, /* GL_SRC_ALPHA_SATURATE */ - { 31605, 0x00000300 }, /* GL_SRC_COLOR */ - { 31618, 0x00008C40 }, /* GL_SRGB */ - { 31626, 0x00008C41 }, /* GL_SRGB8 */ - { 31635, 0x00008C43 }, /* GL_SRGB8_ALPHA8 */ - { 31651, 0x00008C42 }, /* GL_SRGB_ALPHA */ - { 31665, 0x00000503 }, /* GL_STACK_OVERFLOW */ - { 31683, 0x00000504 }, /* GL_STACK_UNDERFLOW */ - { 31702, 0x000088E6 }, /* GL_STATIC_COPY */ - { 31717, 0x000088E6 }, /* GL_STATIC_COPY_ARB */ - { 31736, 0x000088E4 }, /* GL_STATIC_DRAW */ - { 31751, 0x000088E4 }, /* GL_STATIC_DRAW_ARB */ - { 31770, 0x000088E5 }, /* GL_STATIC_READ */ - { 31785, 0x000088E5 }, /* GL_STATIC_READ_ARB */ - { 31804, 0x00001802 }, /* GL_STENCIL */ - { 31815, 0x00008D20 }, /* GL_STENCIL_ATTACHMENT */ - { 31837, 0x00008D20 }, /* GL_STENCIL_ATTACHMENT_EXT */ - { 31863, 0x00008801 }, /* GL_STENCIL_BACK_FAIL */ - { 31884, 0x00008801 }, /* GL_STENCIL_BACK_FAIL_ATI */ - { 31909, 0x00008800 }, /* GL_STENCIL_BACK_FUNC */ - { 31930, 0x00008800 }, /* GL_STENCIL_BACK_FUNC_ATI */ - { 31955, 0x00008802 }, /* GL_STENCIL_BACK_PASS_DEPTH_FAIL */ - { 31987, 0x00008802 }, /* GL_STENCIL_BACK_PASS_DEPTH_FAIL_ATI */ - { 32023, 0x00008803 }, /* GL_STENCIL_BACK_PASS_DEPTH_PASS */ - { 32055, 0x00008803 }, /* GL_STENCIL_BACK_PASS_DEPTH_PASS_ATI */ - { 32091, 0x00008CA3 }, /* GL_STENCIL_BACK_REF */ - { 32111, 0x00008CA4 }, /* GL_STENCIL_BACK_VALUE_MASK */ - { 32138, 0x00008CA5 }, /* GL_STENCIL_BACK_WRITEMASK */ - { 32164, 0x00000D57 }, /* GL_STENCIL_BITS */ - { 32180, 0x00000400 }, /* GL_STENCIL_BUFFER_BIT */ - { 32202, 0x00000B91 }, /* GL_STENCIL_CLEAR_VALUE */ - { 32225, 0x00000B94 }, /* GL_STENCIL_FAIL */ - { 32241, 0x00000B92 }, /* GL_STENCIL_FUNC */ - { 32257, 0x00001901 }, /* GL_STENCIL_INDEX */ - { 32274, 0x00008D49 }, /* GL_STENCIL_INDEX16_EXT */ - { 32297, 0x00008D46 }, /* GL_STENCIL_INDEX1_EXT */ - { 32319, 0x00008D47 }, /* GL_STENCIL_INDEX4_EXT */ - { 32341, 0x00008D48 }, /* GL_STENCIL_INDEX8_EXT */ - { 32363, 0x00008D45 }, /* GL_STENCIL_INDEX_EXT */ - { 32384, 0x00000B95 }, /* GL_STENCIL_PASS_DEPTH_FAIL */ - { 32411, 0x00000B96 }, /* GL_STENCIL_PASS_DEPTH_PASS */ - { 32438, 0x00000B97 }, /* GL_STENCIL_REF */ - { 32453, 0x00000B90 }, /* GL_STENCIL_TEST */ - { 32469, 0x00008910 }, /* GL_STENCIL_TEST_TWO_SIDE_EXT */ - { 32498, 0x00000B93 }, /* GL_STENCIL_VALUE_MASK */ - { 32520, 0x00000B98 }, /* GL_STENCIL_WRITEMASK */ - { 32541, 0x00000C33 }, /* GL_STEREO */ - { 32551, 0x000085BE }, /* GL_STORAGE_CACHED_APPLE */ - { 32575, 0x000085BD }, /* GL_STORAGE_PRIVATE_APPLE */ - { 32600, 0x000085BF }, /* GL_STORAGE_SHARED_APPLE */ - { 32624, 0x000088E2 }, /* GL_STREAM_COPY */ - { 32639, 0x000088E2 }, /* GL_STREAM_COPY_ARB */ - { 32658, 0x000088E0 }, /* GL_STREAM_DRAW */ - { 32673, 0x000088E0 }, /* GL_STREAM_DRAW_ARB */ - { 32692, 0x000088E1 }, /* GL_STREAM_READ */ - { 32707, 0x000088E1 }, /* GL_STREAM_READ_ARB */ - { 32726, 0x00000D50 }, /* GL_SUBPIXEL_BITS */ - { 32743, 0x000084E7 }, /* GL_SUBTRACT */ - { 32755, 0x000084E7 }, /* GL_SUBTRACT_ARB */ - { 32771, 0x00009113 }, /* GL_SYNC_CONDITION */ - { 32789, 0x00009116 }, /* GL_SYNC_FENCE */ - { 32803, 0x00009115 }, /* GL_SYNC_FLAGS */ - { 32817, 0x00000001 }, /* GL_SYNC_FLUSH_COMMANDS_BIT */ - { 32844, 0x00009117 }, /* GL_SYNC_GPU_COMMANDS_COMPLETE */ - { 32874, 0x00009114 }, /* GL_SYNC_STATUS */ - { 32889, 0x00002001 }, /* GL_T */ - { 32894, 0x00002A2A }, /* GL_T2F_C3F_V3F */ - { 32909, 0x00002A2C }, /* GL_T2F_C4F_N3F_V3F */ - { 32928, 0x00002A29 }, /* GL_T2F_C4UB_V3F */ - { 32944, 0x00002A2B }, /* GL_T2F_N3F_V3F */ - { 32959, 0x00002A27 }, /* GL_T2F_V3F */ - { 32970, 0x00002A2D }, /* GL_T4F_C4F_N3F_V4F */ - { 32989, 0x00002A28 }, /* GL_T4F_V4F */ - { 33000, 0x00008031 }, /* GL_TABLE_TOO_LARGE_EXT */ - { 33023, 0x00001702 }, /* GL_TEXTURE */ - { 33034, 0x000084C0 }, /* GL_TEXTURE0 */ - { 33046, 0x000084C0 }, /* GL_TEXTURE0_ARB */ - { 33062, 0x000084C1 }, /* GL_TEXTURE1 */ - { 33074, 0x000084CA }, /* GL_TEXTURE10 */ - { 33087, 0x000084CA }, /* GL_TEXTURE10_ARB */ - { 33104, 0x000084CB }, /* GL_TEXTURE11 */ - { 33117, 0x000084CB }, /* GL_TEXTURE11_ARB */ - { 33134, 0x000084CC }, /* GL_TEXTURE12 */ - { 33147, 0x000084CC }, /* GL_TEXTURE12_ARB */ - { 33164, 0x000084CD }, /* GL_TEXTURE13 */ - { 33177, 0x000084CD }, /* GL_TEXTURE13_ARB */ - { 33194, 0x000084CE }, /* GL_TEXTURE14 */ - { 33207, 0x000084CE }, /* GL_TEXTURE14_ARB */ - { 33224, 0x000084CF }, /* GL_TEXTURE15 */ - { 33237, 0x000084CF }, /* GL_TEXTURE15_ARB */ - { 33254, 0x000084D0 }, /* GL_TEXTURE16 */ - { 33267, 0x000084D0 }, /* GL_TEXTURE16_ARB */ - { 33284, 0x000084D1 }, /* GL_TEXTURE17 */ - { 33297, 0x000084D1 }, /* GL_TEXTURE17_ARB */ - { 33314, 0x000084D2 }, /* GL_TEXTURE18 */ - { 33327, 0x000084D2 }, /* GL_TEXTURE18_ARB */ - { 33344, 0x000084D3 }, /* GL_TEXTURE19 */ - { 33357, 0x000084D3 }, /* GL_TEXTURE19_ARB */ - { 33374, 0x000084C1 }, /* GL_TEXTURE1_ARB */ - { 33390, 0x000084C2 }, /* GL_TEXTURE2 */ - { 33402, 0x000084D4 }, /* GL_TEXTURE20 */ - { 33415, 0x000084D4 }, /* GL_TEXTURE20_ARB */ - { 33432, 0x000084D5 }, /* GL_TEXTURE21 */ - { 33445, 0x000084D5 }, /* GL_TEXTURE21_ARB */ - { 33462, 0x000084D6 }, /* GL_TEXTURE22 */ - { 33475, 0x000084D6 }, /* GL_TEXTURE22_ARB */ - { 33492, 0x000084D7 }, /* GL_TEXTURE23 */ - { 33505, 0x000084D7 }, /* GL_TEXTURE23_ARB */ - { 33522, 0x000084D8 }, /* GL_TEXTURE24 */ - { 33535, 0x000084D8 }, /* GL_TEXTURE24_ARB */ - { 33552, 0x000084D9 }, /* GL_TEXTURE25 */ - { 33565, 0x000084D9 }, /* GL_TEXTURE25_ARB */ - { 33582, 0x000084DA }, /* GL_TEXTURE26 */ - { 33595, 0x000084DA }, /* GL_TEXTURE26_ARB */ - { 33612, 0x000084DB }, /* GL_TEXTURE27 */ - { 33625, 0x000084DB }, /* GL_TEXTURE27_ARB */ - { 33642, 0x000084DC }, /* GL_TEXTURE28 */ - { 33655, 0x000084DC }, /* GL_TEXTURE28_ARB */ - { 33672, 0x000084DD }, /* GL_TEXTURE29 */ - { 33685, 0x000084DD }, /* GL_TEXTURE29_ARB */ - { 33702, 0x000084C2 }, /* GL_TEXTURE2_ARB */ - { 33718, 0x000084C3 }, /* GL_TEXTURE3 */ - { 33730, 0x000084DE }, /* GL_TEXTURE30 */ - { 33743, 0x000084DE }, /* GL_TEXTURE30_ARB */ - { 33760, 0x000084DF }, /* GL_TEXTURE31 */ - { 33773, 0x000084DF }, /* GL_TEXTURE31_ARB */ - { 33790, 0x000084C3 }, /* GL_TEXTURE3_ARB */ - { 33806, 0x000084C4 }, /* GL_TEXTURE4 */ - { 33818, 0x000084C4 }, /* GL_TEXTURE4_ARB */ - { 33834, 0x000084C5 }, /* GL_TEXTURE5 */ - { 33846, 0x000084C5 }, /* GL_TEXTURE5_ARB */ - { 33862, 0x000084C6 }, /* GL_TEXTURE6 */ - { 33874, 0x000084C6 }, /* GL_TEXTURE6_ARB */ - { 33890, 0x000084C7 }, /* GL_TEXTURE7 */ - { 33902, 0x000084C7 }, /* GL_TEXTURE7_ARB */ - { 33918, 0x000084C8 }, /* GL_TEXTURE8 */ - { 33930, 0x000084C8 }, /* GL_TEXTURE8_ARB */ - { 33946, 0x000084C9 }, /* GL_TEXTURE9 */ - { 33958, 0x000084C9 }, /* GL_TEXTURE9_ARB */ - { 33974, 0x00000DE0 }, /* GL_TEXTURE_1D */ - { 33988, 0x00008C18 }, /* GL_TEXTURE_1D_ARRAY_EXT */ - { 34012, 0x00000DE1 }, /* GL_TEXTURE_2D */ - { 34026, 0x00008C1A }, /* GL_TEXTURE_2D_ARRAY_EXT */ - { 34050, 0x0000806F }, /* GL_TEXTURE_3D */ - { 34064, 0x0000805F }, /* GL_TEXTURE_ALPHA_SIZE */ - { 34086, 0x0000805F }, /* GL_TEXTURE_ALPHA_SIZE_EXT */ - { 34112, 0x0000813C }, /* GL_TEXTURE_BASE_LEVEL */ - { 34134, 0x00008068 }, /* GL_TEXTURE_BINDING_1D */ - { 34156, 0x00008C1C }, /* GL_TEXTURE_BINDING_1D_ARRAY_EXT */ - { 34188, 0x00008069 }, /* GL_TEXTURE_BINDING_2D */ - { 34210, 0x00008C1D }, /* GL_TEXTURE_BINDING_2D_ARRAY_EXT */ - { 34242, 0x0000806A }, /* GL_TEXTURE_BINDING_3D */ - { 34264, 0x00008514 }, /* GL_TEXTURE_BINDING_CUBE_MAP */ - { 34292, 0x00008514 }, /* GL_TEXTURE_BINDING_CUBE_MAP_ARB */ - { 34324, 0x000084F6 }, /* GL_TEXTURE_BINDING_RECTANGLE_ARB */ - { 34357, 0x000084F6 }, /* GL_TEXTURE_BINDING_RECTANGLE_NV */ - { 34389, 0x00040000 }, /* GL_TEXTURE_BIT */ - { 34404, 0x0000805E }, /* GL_TEXTURE_BLUE_SIZE */ - { 34425, 0x0000805E }, /* GL_TEXTURE_BLUE_SIZE_EXT */ - { 34450, 0x00001005 }, /* GL_TEXTURE_BORDER */ - { 34468, 0x00001004 }, /* GL_TEXTURE_BORDER_COLOR */ - { 34492, 0x00008171 }, /* GL_TEXTURE_CLIPMAP_CENTER_SGIX */ - { 34523, 0x00008176 }, /* GL_TEXTURE_CLIPMAP_DEPTH_SGIX */ - { 34553, 0x00008172 }, /* GL_TEXTURE_CLIPMAP_FRAME_SGIX */ - { 34583, 0x00008175 }, /* GL_TEXTURE_CLIPMAP_LOD_OFFSET_SGIX */ - { 34618, 0x00008173 }, /* GL_TEXTURE_CLIPMAP_OFFSET_SGIX */ - { 34649, 0x00008174 }, /* GL_TEXTURE_CLIPMAP_VIRTUAL_DEPTH_SGIX */ - { 34687, 0x000080BC }, /* GL_TEXTURE_COLOR_TABLE_SGI */ - { 34714, 0x000081EF }, /* GL_TEXTURE_COLOR_WRITEMASK_SGIS */ - { 34746, 0x000080BF }, /* GL_TEXTURE_COMPARE_FAIL_VALUE_ARB */ - { 34780, 0x0000884D }, /* GL_TEXTURE_COMPARE_FUNC */ - { 34804, 0x0000884D }, /* GL_TEXTURE_COMPARE_FUNC_ARB */ - { 34832, 0x0000884C }, /* GL_TEXTURE_COMPARE_MODE */ - { 34856, 0x0000884C }, /* GL_TEXTURE_COMPARE_MODE_ARB */ - { 34884, 0x0000819B }, /* GL_TEXTURE_COMPARE_OPERATOR_SGIX */ - { 34917, 0x0000819A }, /* GL_TEXTURE_COMPARE_SGIX */ - { 34941, 0x00001003 }, /* GL_TEXTURE_COMPONENTS */ - { 34963, 0x000086A1 }, /* GL_TEXTURE_COMPRESSED */ - { 34985, 0x000086A1 }, /* GL_TEXTURE_COMPRESSED_ARB */ - { 35011, 0x000086A3 }, /* GL_TEXTURE_COMPRESSED_FORMATS_ARB */ - { 35045, 0x000086A0 }, /* GL_TEXTURE_COMPRESSED_IMAGE_SIZE */ - { 35078, 0x000086A0 }, /* GL_TEXTURE_COMPRESSED_IMAGE_SIZE_ARB */ - { 35115, 0x000084EF }, /* GL_TEXTURE_COMPRESSION_HINT */ - { 35143, 0x000084EF }, /* GL_TEXTURE_COMPRESSION_HINT_ARB */ - { 35175, 0x00008078 }, /* GL_TEXTURE_COORD_ARRAY */ - { 35198, 0x0000889A }, /* GL_TEXTURE_COORD_ARRAY_BUFFER_BINDING */ - { 35236, 0x0000889A }, /* GL_TEXTURE_COORD_ARRAY_BUFFER_BINDING_ARB */ - { 35278, 0x00008092 }, /* GL_TEXTURE_COORD_ARRAY_POINTER */ - { 35309, 0x00008088 }, /* GL_TEXTURE_COORD_ARRAY_SIZE */ - { 35337, 0x0000808A }, /* GL_TEXTURE_COORD_ARRAY_STRIDE */ - { 35367, 0x00008089 }, /* GL_TEXTURE_COORD_ARRAY_TYPE */ - { 35395, 0x00008513 }, /* GL_TEXTURE_CUBE_MAP */ - { 35415, 0x00008513 }, /* GL_TEXTURE_CUBE_MAP_ARB */ - { 35439, 0x00008516 }, /* GL_TEXTURE_CUBE_MAP_NEGATIVE_X */ - { 35470, 0x00008516 }, /* GL_TEXTURE_CUBE_MAP_NEGATIVE_X_ARB */ - { 35505, 0x00008518 }, /* GL_TEXTURE_CUBE_MAP_NEGATIVE_Y */ - { 35536, 0x00008518 }, /* GL_TEXTURE_CUBE_MAP_NEGATIVE_Y_ARB */ - { 35571, 0x0000851A }, /* GL_TEXTURE_CUBE_MAP_NEGATIVE_Z */ - { 35602, 0x0000851A }, /* GL_TEXTURE_CUBE_MAP_NEGATIVE_Z_ARB */ - { 35637, 0x00008515 }, /* GL_TEXTURE_CUBE_MAP_POSITIVE_X */ - { 35668, 0x00008515 }, /* GL_TEXTURE_CUBE_MAP_POSITIVE_X_ARB */ - { 35703, 0x00008517 }, /* GL_TEXTURE_CUBE_MAP_POSITIVE_Y */ - { 35734, 0x00008517 }, /* GL_TEXTURE_CUBE_MAP_POSITIVE_Y_ARB */ - { 35769, 0x00008519 }, /* GL_TEXTURE_CUBE_MAP_POSITIVE_Z */ - { 35800, 0x00008519 }, /* GL_TEXTURE_CUBE_MAP_POSITIVE_Z_ARB */ - { 35835, 0x000088F4 }, /* GL_TEXTURE_CUBE_MAP_SEAMLESS */ - { 35864, 0x00008071 }, /* GL_TEXTURE_DEPTH */ - { 35881, 0x0000884A }, /* GL_TEXTURE_DEPTH_SIZE */ - { 35903, 0x0000884A }, /* GL_TEXTURE_DEPTH_SIZE_ARB */ - { 35929, 0x00002300 }, /* GL_TEXTURE_ENV */ - { 35944, 0x00002201 }, /* GL_TEXTURE_ENV_COLOR */ - { 35965, 0x00002200 }, /* GL_TEXTURE_ENV_MODE */ - { 35985, 0x00008500 }, /* GL_TEXTURE_FILTER_CONTROL */ - { 36011, 0x00002500 }, /* GL_TEXTURE_GEN_MODE */ - { 36031, 0x00000C63 }, /* GL_TEXTURE_GEN_Q */ - { 36048, 0x00000C62 }, /* GL_TEXTURE_GEN_R */ - { 36065, 0x00000C60 }, /* GL_TEXTURE_GEN_S */ - { 36082, 0x00000C61 }, /* GL_TEXTURE_GEN_T */ - { 36099, 0x0000819D }, /* GL_TEXTURE_GEQUAL_R_SGIX */ - { 36124, 0x0000805D }, /* GL_TEXTURE_GREEN_SIZE */ - { 36146, 0x0000805D }, /* GL_TEXTURE_GREEN_SIZE_EXT */ - { 36172, 0x00001001 }, /* GL_TEXTURE_HEIGHT */ - { 36190, 0x000080ED }, /* GL_TEXTURE_INDEX_SIZE_EXT */ - { 36216, 0x00008061 }, /* GL_TEXTURE_INTENSITY_SIZE */ - { 36242, 0x00008061 }, /* GL_TEXTURE_INTENSITY_SIZE_EXT */ - { 36272, 0x00001003 }, /* GL_TEXTURE_INTERNAL_FORMAT */ - { 36299, 0x0000819C }, /* GL_TEXTURE_LEQUAL_R_SGIX */ - { 36324, 0x00008501 }, /* GL_TEXTURE_LOD_BIAS */ - { 36344, 0x00008501 }, /* GL_TEXTURE_LOD_BIAS_EXT */ - { 36368, 0x00008190 }, /* GL_TEXTURE_LOD_BIAS_R_SGIX */ - { 36395, 0x0000818E }, /* GL_TEXTURE_LOD_BIAS_S_SGIX */ - { 36422, 0x0000818F }, /* GL_TEXTURE_LOD_BIAS_T_SGIX */ - { 36449, 0x00008060 }, /* GL_TEXTURE_LUMINANCE_SIZE */ - { 36475, 0x00008060 }, /* GL_TEXTURE_LUMINANCE_SIZE_EXT */ - { 36505, 0x00002800 }, /* GL_TEXTURE_MAG_FILTER */ - { 36527, 0x00000BA8 }, /* GL_TEXTURE_MATRIX */ - { 36545, 0x000084FE }, /* GL_TEXTURE_MAX_ANISOTROPY_EXT */ - { 36575, 0x0000836B }, /* GL_TEXTURE_MAX_CLAMP_R_SGIX */ - { 36603, 0x00008369 }, /* GL_TEXTURE_MAX_CLAMP_S_SGIX */ - { 36631, 0x0000836A }, /* GL_TEXTURE_MAX_CLAMP_T_SGIX */ - { 36659, 0x0000813D }, /* GL_TEXTURE_MAX_LEVEL */ - { 36680, 0x0000813B }, /* GL_TEXTURE_MAX_LOD */ - { 36699, 0x00002801 }, /* GL_TEXTURE_MIN_FILTER */ - { 36721, 0x0000813A }, /* GL_TEXTURE_MIN_LOD */ - { 36740, 0x00008066 }, /* GL_TEXTURE_PRIORITY */ - { 36760, 0x000085B7 }, /* GL_TEXTURE_RANGE_LENGTH_APPLE */ - { 36790, 0x000085B8 }, /* GL_TEXTURE_RANGE_POINTER_APPLE */ - { 36821, 0x000084F5 }, /* GL_TEXTURE_RECTANGLE_ARB */ - { 36846, 0x000084F5 }, /* GL_TEXTURE_RECTANGLE_NV */ - { 36870, 0x0000805C }, /* GL_TEXTURE_RED_SIZE */ - { 36890, 0x0000805C }, /* GL_TEXTURE_RED_SIZE_EXT */ - { 36914, 0x00008067 }, /* GL_TEXTURE_RESIDENT */ - { 36934, 0x00000BA5 }, /* GL_TEXTURE_STACK_DEPTH */ - { 36957, 0x000088F1 }, /* GL_TEXTURE_STENCIL_SIZE */ - { 36981, 0x000085BC }, /* GL_TEXTURE_STORAGE_HINT_APPLE */ - { 37011, 0x00008065 }, /* GL_TEXTURE_TOO_LARGE_EXT */ - { 37036, 0x0000888F }, /* GL_TEXTURE_UNSIGNED_REMAP_MODE_NV */ - { 37070, 0x00001000 }, /* GL_TEXTURE_WIDTH */ - { 37087, 0x00008072 }, /* GL_TEXTURE_WRAP_R */ - { 37105, 0x00002802 }, /* GL_TEXTURE_WRAP_S */ - { 37123, 0x00002803 }, /* GL_TEXTURE_WRAP_T */ - { 37141, 0x0000911B }, /* GL_TIMEOUT_EXPIRED */ - { 37160, 0xFFFFFFFFFFFFFFFF }, /* GL_TIMEOUT_IGNORED */ - { 37179, 0x000088BF }, /* GL_TIME_ELAPSED_EXT */ - { 37199, 0x00008648 }, /* GL_TRACK_MATRIX_NV */ - { 37218, 0x00008649 }, /* GL_TRACK_MATRIX_TRANSFORM_NV */ - { 37247, 0x00001000 }, /* GL_TRANSFORM_BIT */ - { 37264, 0x000084E6 }, /* GL_TRANSPOSE_COLOR_MATRIX */ - { 37290, 0x000084E6 }, /* GL_TRANSPOSE_COLOR_MATRIX_ARB */ - { 37320, 0x000088B7 }, /* GL_TRANSPOSE_CURRENT_MATRIX_ARB */ - { 37352, 0x000084E3 }, /* GL_TRANSPOSE_MODELVIEW_MATRIX */ - { 37382, 0x000084E3 }, /* GL_TRANSPOSE_MODELVIEW_MATRIX_ARB */ - { 37416, 0x0000862C }, /* GL_TRANSPOSE_NV */ - { 37432, 0x000084E4 }, /* GL_TRANSPOSE_PROJECTION_MATRIX */ - { 37463, 0x000084E4 }, /* GL_TRANSPOSE_PROJECTION_MATRIX_ARB */ - { 37498, 0x000084E5 }, /* GL_TRANSPOSE_TEXTURE_MATRIX */ - { 37526, 0x000084E5 }, /* GL_TRANSPOSE_TEXTURE_MATRIX_ARB */ - { 37558, 0x00000004 }, /* GL_TRIANGLES */ - { 37571, 0x00000006 }, /* GL_TRIANGLE_FAN */ - { 37587, 0x00008615 }, /* GL_TRIANGLE_MESH_SUN */ - { 37608, 0x00000005 }, /* GL_TRIANGLE_STRIP */ - { 37626, 0x00000001 }, /* GL_TRUE */ - { 37634, 0x00000CF5 }, /* GL_UNPACK_ALIGNMENT */ - { 37654, 0x0000806E }, /* GL_UNPACK_IMAGE_HEIGHT */ - { 37677, 0x00000CF1 }, /* GL_UNPACK_LSB_FIRST */ - { 37697, 0x00000CF2 }, /* GL_UNPACK_ROW_LENGTH */ - { 37718, 0x0000806D }, /* GL_UNPACK_SKIP_IMAGES */ - { 37740, 0x00000CF4 }, /* GL_UNPACK_SKIP_PIXELS */ - { 37762, 0x00000CF3 }, /* GL_UNPACK_SKIP_ROWS */ - { 37782, 0x00000CF0 }, /* GL_UNPACK_SWAP_BYTES */ - { 37803, 0x00009118 }, /* GL_UNSIGNALED */ - { 37817, 0x00001401 }, /* GL_UNSIGNED_BYTE */ - { 37834, 0x00008362 }, /* GL_UNSIGNED_BYTE_2_3_3_REV */ - { 37861, 0x00008032 }, /* GL_UNSIGNED_BYTE_3_3_2 */ - { 37884, 0x00001405 }, /* GL_UNSIGNED_INT */ - { 37900, 0x00008036 }, /* GL_UNSIGNED_INT_10_10_10_2 */ - { 37927, 0x000084FA }, /* GL_UNSIGNED_INT_24_8 */ - { 37948, 0x000084FA }, /* GL_UNSIGNED_INT_24_8_NV */ - { 37972, 0x00008368 }, /* GL_UNSIGNED_INT_2_10_10_10_REV */ - { 38003, 0x00008035 }, /* GL_UNSIGNED_INT_8_8_8_8 */ - { 38027, 0x00008367 }, /* GL_UNSIGNED_INT_8_8_8_8_REV */ - { 38055, 0x00008C17 }, /* GL_UNSIGNED_NORMALIZED */ - { 38078, 0x00001403 }, /* GL_UNSIGNED_SHORT */ - { 38096, 0x00008366 }, /* GL_UNSIGNED_SHORT_1_5_5_5_REV */ - { 38126, 0x00008033 }, /* GL_UNSIGNED_SHORT_4_4_4_4 */ - { 38152, 0x00008365 }, /* GL_UNSIGNED_SHORT_4_4_4_4_REV */ - { 38182, 0x00008034 }, /* GL_UNSIGNED_SHORT_5_5_5_1 */ - { 38208, 0x00008363 }, /* GL_UNSIGNED_SHORT_5_6_5 */ - { 38232, 0x00008364 }, /* GL_UNSIGNED_SHORT_5_6_5_REV */ - { 38260, 0x000085BA }, /* GL_UNSIGNED_SHORT_8_8_APPLE */ - { 38288, 0x000085BA }, /* GL_UNSIGNED_SHORT_8_8_MESA */ - { 38315, 0x000085BB }, /* GL_UNSIGNED_SHORT_8_8_REV_APPLE */ - { 38347, 0x000085BB }, /* GL_UNSIGNED_SHORT_8_8_REV_MESA */ - { 38378, 0x00008CA2 }, /* GL_UPPER_LEFT */ - { 38392, 0x00002A20 }, /* GL_V2F */ - { 38399, 0x00002A21 }, /* GL_V3F */ - { 38406, 0x00008B83 }, /* GL_VALIDATE_STATUS */ - { 38425, 0x00001F00 }, /* GL_VENDOR */ - { 38435, 0x00001F02 }, /* GL_VERSION */ - { 38446, 0x00008074 }, /* GL_VERTEX_ARRAY */ - { 38462, 0x000085B5 }, /* GL_VERTEX_ARRAY_BINDING */ - { 38486, 0x000085B5 }, /* GL_VERTEX_ARRAY_BINDING_APPLE */ - { 38516, 0x00008896 }, /* GL_VERTEX_ARRAY_BUFFER_BINDING */ - { 38547, 0x00008896 }, /* GL_VERTEX_ARRAY_BUFFER_BINDING_ARB */ - { 38582, 0x0000808E }, /* GL_VERTEX_ARRAY_POINTER */ - { 38606, 0x0000807A }, /* GL_VERTEX_ARRAY_SIZE */ - { 38627, 0x0000807C }, /* GL_VERTEX_ARRAY_STRIDE */ - { 38650, 0x0000807B }, /* GL_VERTEX_ARRAY_TYPE */ - { 38671, 0x00008650 }, /* GL_VERTEX_ATTRIB_ARRAY0_NV */ - { 38698, 0x0000865A }, /* GL_VERTEX_ATTRIB_ARRAY10_NV */ - { 38726, 0x0000865B }, /* GL_VERTEX_ATTRIB_ARRAY11_NV */ - { 38754, 0x0000865C }, /* GL_VERTEX_ATTRIB_ARRAY12_NV */ - { 38782, 0x0000865D }, /* GL_VERTEX_ATTRIB_ARRAY13_NV */ - { 38810, 0x0000865E }, /* GL_VERTEX_ATTRIB_ARRAY14_NV */ - { 38838, 0x0000865F }, /* GL_VERTEX_ATTRIB_ARRAY15_NV */ - { 38866, 0x00008651 }, /* GL_VERTEX_ATTRIB_ARRAY1_NV */ - { 38893, 0x00008652 }, /* GL_VERTEX_ATTRIB_ARRAY2_NV */ - { 38920, 0x00008653 }, /* GL_VERTEX_ATTRIB_ARRAY3_NV */ - { 38947, 0x00008654 }, /* GL_VERTEX_ATTRIB_ARRAY4_NV */ - { 38974, 0x00008655 }, /* GL_VERTEX_ATTRIB_ARRAY5_NV */ - { 39001, 0x00008656 }, /* GL_VERTEX_ATTRIB_ARRAY6_NV */ - { 39028, 0x00008657 }, /* GL_VERTEX_ATTRIB_ARRAY7_NV */ - { 39055, 0x00008658 }, /* GL_VERTEX_ATTRIB_ARRAY8_NV */ - { 39082, 0x00008659 }, /* GL_VERTEX_ATTRIB_ARRAY9_NV */ - { 39109, 0x0000889F }, /* GL_VERTEX_ATTRIB_ARRAY_BUFFER_BINDING */ - { 39147, 0x0000889F }, /* GL_VERTEX_ATTRIB_ARRAY_BUFFER_BINDING_ARB */ - { 39189, 0x00008622 }, /* GL_VERTEX_ATTRIB_ARRAY_ENABLED */ - { 39220, 0x00008622 }, /* GL_VERTEX_ATTRIB_ARRAY_ENABLED_ARB */ - { 39255, 0x0000886A }, /* GL_VERTEX_ATTRIB_ARRAY_NORMALIZED */ - { 39289, 0x0000886A }, /* GL_VERTEX_ATTRIB_ARRAY_NORMALIZED_ARB */ - { 39327, 0x00008645 }, /* GL_VERTEX_ATTRIB_ARRAY_POINTER */ - { 39358, 0x00008645 }, /* GL_VERTEX_ATTRIB_ARRAY_POINTER_ARB */ - { 39393, 0x00008623 }, /* GL_VERTEX_ATTRIB_ARRAY_SIZE */ - { 39421, 0x00008623 }, /* GL_VERTEX_ATTRIB_ARRAY_SIZE_ARB */ - { 39453, 0x00008624 }, /* GL_VERTEX_ATTRIB_ARRAY_STRIDE */ - { 39483, 0x00008624 }, /* GL_VERTEX_ATTRIB_ARRAY_STRIDE_ARB */ - { 39517, 0x00008625 }, /* GL_VERTEX_ATTRIB_ARRAY_TYPE */ - { 39545, 0x00008625 }, /* GL_VERTEX_ATTRIB_ARRAY_TYPE_ARB */ - { 39577, 0x000086A7 }, /* GL_VERTEX_BLEND_ARB */ - { 39597, 0x00008620 }, /* GL_VERTEX_PROGRAM_ARB */ - { 39619, 0x0000864A }, /* GL_VERTEX_PROGRAM_BINDING_NV */ - { 39648, 0x00008620 }, /* GL_VERTEX_PROGRAM_NV */ - { 39669, 0x00008642 }, /* GL_VERTEX_PROGRAM_POINT_SIZE */ - { 39698, 0x00008642 }, /* GL_VERTEX_PROGRAM_POINT_SIZE_ARB */ - { 39731, 0x00008642 }, /* GL_VERTEX_PROGRAM_POINT_SIZE_NV */ - { 39763, 0x00008643 }, /* GL_VERTEX_PROGRAM_TWO_SIDE */ - { 39790, 0x00008643 }, /* GL_VERTEX_PROGRAM_TWO_SIDE_ARB */ - { 39821, 0x00008643 }, /* GL_VERTEX_PROGRAM_TWO_SIDE_NV */ - { 39851, 0x00008B31 }, /* GL_VERTEX_SHADER */ - { 39868, 0x00008B31 }, /* GL_VERTEX_SHADER_ARB */ - { 39889, 0x00008621 }, /* GL_VERTEX_STATE_PROGRAM_NV */ - { 39916, 0x00000BA2 }, /* GL_VIEWPORT */ - { 39928, 0x00000800 }, /* GL_VIEWPORT_BIT */ - { 39944, 0x0000911D }, /* GL_WAIT_FAILED */ - { 39959, 0x000086AD }, /* GL_WEIGHT_ARRAY_ARB */ - { 39979, 0x0000889E }, /* GL_WEIGHT_ARRAY_BUFFER_BINDING */ - { 40010, 0x0000889E }, /* GL_WEIGHT_ARRAY_BUFFER_BINDING_ARB */ - { 40045, 0x000086AC }, /* GL_WEIGHT_ARRAY_POINTER_ARB */ - { 40073, 0x000086AB }, /* GL_WEIGHT_ARRAY_SIZE_ARB */ - { 40098, 0x000086AA }, /* GL_WEIGHT_ARRAY_STRIDE_ARB */ - { 40125, 0x000086A9 }, /* GL_WEIGHT_ARRAY_TYPE_ARB */ - { 40150, 0x000086A6 }, /* GL_WEIGHT_SUM_UNITY_ARB */ - { 40174, 0x000081D4 }, /* GL_WRAP_BORDER_SUN */ - { 40193, 0x000088B9 }, /* GL_WRITE_ONLY */ - { 40207, 0x000088B9 }, /* GL_WRITE_ONLY_ARB */ - { 40225, 0x00001506 }, /* GL_XOR */ - { 40232, 0x000085B9 }, /* GL_YCBCR_422_APPLE */ - { 40251, 0x00008757 }, /* GL_YCBCR_MESA */ - { 40265, 0x00000000 }, /* GL_ZERO */ - { 40273, 0x00000D16 }, /* GL_ZOOM_X */ - { 40283, 0x00000D17 }, /* GL_ZOOM_Y */ + { 9726, 0x00008E4D }, /* GL_FIRST_VERTEX_CONVENTION */ + { 9753, 0x00008E4D }, /* GL_FIRST_VERTEX_CONVENTION_EXT */ + { 9784, 0x00001D00 }, /* GL_FLAT */ + { 9792, 0x00001406 }, /* GL_FLOAT */ + { 9801, 0x00008B5A }, /* GL_FLOAT_MAT2 */ + { 9815, 0x00008B5A }, /* GL_FLOAT_MAT2_ARB */ + { 9833, 0x00008B65 }, /* GL_FLOAT_MAT2x3 */ + { 9849, 0x00008B66 }, /* GL_FLOAT_MAT2x4 */ + { 9865, 0x00008B5B }, /* GL_FLOAT_MAT3 */ + { 9879, 0x00008B5B }, /* GL_FLOAT_MAT3_ARB */ + { 9897, 0x00008B67 }, /* GL_FLOAT_MAT3x2 */ + { 9913, 0x00008B68 }, /* GL_FLOAT_MAT3x4 */ + { 9929, 0x00008B5C }, /* GL_FLOAT_MAT4 */ + { 9943, 0x00008B5C }, /* GL_FLOAT_MAT4_ARB */ + { 9961, 0x00008B69 }, /* GL_FLOAT_MAT4x2 */ + { 9977, 0x00008B6A }, /* GL_FLOAT_MAT4x3 */ + { 9993, 0x00008B50 }, /* GL_FLOAT_VEC2 */ + { 10007, 0x00008B50 }, /* GL_FLOAT_VEC2_ARB */ + { 10025, 0x00008B51 }, /* GL_FLOAT_VEC3 */ + { 10039, 0x00008B51 }, /* GL_FLOAT_VEC3_ARB */ + { 10057, 0x00008B52 }, /* GL_FLOAT_VEC4 */ + { 10071, 0x00008B52 }, /* GL_FLOAT_VEC4_ARB */ + { 10089, 0x00000B60 }, /* GL_FOG */ + { 10096, 0x00000080 }, /* GL_FOG_BIT */ + { 10107, 0x00000B66 }, /* GL_FOG_COLOR */ + { 10120, 0x00008451 }, /* GL_FOG_COORD */ + { 10133, 0x00008451 }, /* GL_FOG_COORDINATE */ + { 10151, 0x00008457 }, /* GL_FOG_COORDINATE_ARRAY */ + { 10175, 0x0000889D }, /* GL_FOG_COORDINATE_ARRAY_BUFFER_BINDING */ + { 10214, 0x0000889D }, /* GL_FOG_COORDINATE_ARRAY_BUFFER_BINDING_ARB */ + { 10257, 0x00008456 }, /* GL_FOG_COORDINATE_ARRAY_POINTER */ + { 10289, 0x00008455 }, /* GL_FOG_COORDINATE_ARRAY_STRIDE */ + { 10320, 0x00008454 }, /* GL_FOG_COORDINATE_ARRAY_TYPE */ + { 10349, 0x00008450 }, /* GL_FOG_COORDINATE_SOURCE */ + { 10374, 0x00008457 }, /* GL_FOG_COORD_ARRAY */ + { 10393, 0x0000889D }, /* GL_FOG_COORD_ARRAY_BUFFER_BINDING */ + { 10427, 0x00008456 }, /* GL_FOG_COORD_ARRAY_POINTER */ + { 10454, 0x00008455 }, /* GL_FOG_COORD_ARRAY_STRIDE */ + { 10480, 0x00008454 }, /* GL_FOG_COORD_ARRAY_TYPE */ + { 10504, 0x00008450 }, /* GL_FOG_COORD_SRC */ + { 10521, 0x00000B62 }, /* GL_FOG_DENSITY */ + { 10536, 0x0000855A }, /* GL_FOG_DISTANCE_MODE_NV */ + { 10560, 0x00000B64 }, /* GL_FOG_END */ + { 10571, 0x00000C54 }, /* GL_FOG_HINT */ + { 10583, 0x00000B61 }, /* GL_FOG_INDEX */ + { 10596, 0x00000B65 }, /* GL_FOG_MODE */ + { 10608, 0x00008198 }, /* GL_FOG_OFFSET_SGIX */ + { 10627, 0x00008199 }, /* GL_FOG_OFFSET_VALUE_SGIX */ + { 10652, 0x00000B63 }, /* GL_FOG_START */ + { 10665, 0x00008452 }, /* GL_FRAGMENT_DEPTH */ + { 10683, 0x00008804 }, /* GL_FRAGMENT_PROGRAM_ARB */ + { 10707, 0x00008B30 }, /* GL_FRAGMENT_SHADER */ + { 10726, 0x00008B30 }, /* GL_FRAGMENT_SHADER_ARB */ + { 10749, 0x00008B8B }, /* GL_FRAGMENT_SHADER_DERIVATIVE_HINT */ + { 10784, 0x00008D40 }, /* GL_FRAMEBUFFER */ + { 10799, 0x00008215 }, /* GL_FRAMEBUFFER_ATTACHMENT_ALPHA_SIZE */ + { 10836, 0x00008214 }, /* GL_FRAMEBUFFER_ATTACHMENT_BLUE_SIZE */ + { 10872, 0x00008210 }, /* GL_FRAMEBUFFER_ATTACHMENT_COLOR_ENCODING */ + { 10913, 0x00008211 }, /* GL_FRAMEBUFFER_ATTACHMENT_COMPONENT_TYPE */ + { 10954, 0x00008216 }, /* GL_FRAMEBUFFER_ATTACHMENT_DEPTH_SIZE */ + { 10991, 0x00008213 }, /* GL_FRAMEBUFFER_ATTACHMENT_GREEN_SIZE */ + { 11028, 0x00008CD1 }, /* GL_FRAMEBUFFER_ATTACHMENT_OBJECT_NAME */ + { 11066, 0x00008CD1 }, /* GL_FRAMEBUFFER_ATTACHMENT_OBJECT_NAME_EXT */ + { 11108, 0x00008CD0 }, /* GL_FRAMEBUFFER_ATTACHMENT_OBJECT_TYPE */ + { 11146, 0x00008CD0 }, /* GL_FRAMEBUFFER_ATTACHMENT_OBJECT_TYPE_EXT */ + { 11188, 0x00008212 }, /* GL_FRAMEBUFFER_ATTACHMENT_RED_SIZE */ + { 11223, 0x00008217 }, /* GL_FRAMEBUFFER_ATTACHMENT_STENCIL_SIZE */ + { 11262, 0x00008CD4 }, /* GL_FRAMEBUFFER_ATTACHMENT_TEXTURE_3D_ZOFFSET_EXT */ + { 11311, 0x00008CD3 }, /* GL_FRAMEBUFFER_ATTACHMENT_TEXTURE_CUBE_MAP_FACE */ + { 11359, 0x00008CD3 }, /* GL_FRAMEBUFFER_ATTACHMENT_TEXTURE_CUBE_MAP_FACE_EXT */ + { 11411, 0x00008CD4 }, /* GL_FRAMEBUFFER_ATTACHMENT_TEXTURE_LAYER */ + { 11451, 0x00008CD4 }, /* GL_FRAMEBUFFER_ATTACHMENT_TEXTURE_LAYER_EXT */ + { 11495, 0x00008CD2 }, /* GL_FRAMEBUFFER_ATTACHMENT_TEXTURE_LEVEL */ + { 11535, 0x00008CD2 }, /* GL_FRAMEBUFFER_ATTACHMENT_TEXTURE_LEVEL_EXT */ + { 11579, 0x00008CA6 }, /* GL_FRAMEBUFFER_BINDING_EXT */ + { 11606, 0x00008CD5 }, /* GL_FRAMEBUFFER_COMPLETE */ + { 11630, 0x00008CD5 }, /* GL_FRAMEBUFFER_COMPLETE_EXT */ + { 11658, 0x00008218 }, /* GL_FRAMEBUFFER_DEFAULT */ + { 11681, 0x00008D40 }, /* GL_FRAMEBUFFER_EXT */ + { 11700, 0x00008CD6 }, /* GL_FRAMEBUFFER_INCOMPLETE_ATTACHMENT */ + { 11737, 0x00008CD6 }, /* GL_FRAMEBUFFER_INCOMPLETE_ATTACHMENT_EXT */ + { 11778, 0x00008CD9 }, /* GL_FRAMEBUFFER_INCOMPLETE_DIMENSIONS_EXT */ + { 11819, 0x00008CDB }, /* GL_FRAMEBUFFER_INCOMPLETE_DRAW_BUFFER_EXT */ + { 11861, 0x00008CD8 }, /* GL_FRAMEBUFFER_INCOMPLETE_DUPLICATE_ATTACHMENT_EXT */ + { 11912, 0x00008CDA }, /* GL_FRAMEBUFFER_INCOMPLETE_FORMATS_EXT */ + { 11950, 0x00008CD7 }, /* GL_FRAMEBUFFER_INCOMPLETE_MISSING_ATTACHMENT */ + { 11995, 0x00008CD7 }, /* GL_FRAMEBUFFER_INCOMPLETE_MISSING_ATTACHMENT_EXT */ + { 12044, 0x00008D56 }, /* GL_FRAMEBUFFER_INCOMPLETE_MULTISAMPLE */ + { 12082, 0x00008CDC }, /* GL_FRAMEBUFFER_INCOMPLETE_READ_BUFFER_EXT */ + { 12124, 0x00008CDE }, /* GL_FRAMEBUFFER_STATUS_ERROR_EXT */ + { 12156, 0x00008219 }, /* GL_FRAMEBUFFER_UNDEFINED */ + { 12181, 0x00008CDD }, /* GL_FRAMEBUFFER_UNSUPPORTED */ + { 12208, 0x00008CDD }, /* GL_FRAMEBUFFER_UNSUPPORTED_EXT */ + { 12239, 0x00000404 }, /* GL_FRONT */ + { 12248, 0x00000408 }, /* GL_FRONT_AND_BACK */ + { 12266, 0x00000B46 }, /* GL_FRONT_FACE */ + { 12280, 0x00000400 }, /* GL_FRONT_LEFT */ + { 12294, 0x00000401 }, /* GL_FRONT_RIGHT */ + { 12309, 0x00008006 }, /* GL_FUNC_ADD */ + { 12321, 0x00008006 }, /* GL_FUNC_ADD_EXT */ + { 12337, 0x0000800B }, /* GL_FUNC_REVERSE_SUBTRACT */ + { 12362, 0x0000800B }, /* GL_FUNC_REVERSE_SUBTRACT_EXT */ + { 12391, 0x0000800A }, /* GL_FUNC_SUBTRACT */ + { 12408, 0x0000800A }, /* GL_FUNC_SUBTRACT_EXT */ + { 12429, 0x00008191 }, /* GL_GENERATE_MIPMAP */ + { 12448, 0x00008192 }, /* GL_GENERATE_MIPMAP_HINT */ + { 12472, 0x00008192 }, /* GL_GENERATE_MIPMAP_HINT_SGIS */ + { 12501, 0x00008191 }, /* GL_GENERATE_MIPMAP_SGIS */ + { 12525, 0x00000206 }, /* GL_GEQUAL */ + { 12535, 0x00000204 }, /* GL_GREATER */ + { 12546, 0x00001904 }, /* GL_GREEN */ + { 12555, 0x00000D19 }, /* GL_GREEN_BIAS */ + { 12569, 0x00000D53 }, /* GL_GREEN_BITS */ + { 12583, 0x00000D18 }, /* GL_GREEN_SCALE */ + { 12598, 0x00008000 }, /* GL_HINT_BIT */ + { 12610, 0x00008024 }, /* GL_HISTOGRAM */ + { 12623, 0x0000802B }, /* GL_HISTOGRAM_ALPHA_SIZE */ + { 12647, 0x0000802B }, /* GL_HISTOGRAM_ALPHA_SIZE_EXT */ + { 12675, 0x0000802A }, /* GL_HISTOGRAM_BLUE_SIZE */ + { 12698, 0x0000802A }, /* GL_HISTOGRAM_BLUE_SIZE_EXT */ + { 12725, 0x00008024 }, /* GL_HISTOGRAM_EXT */ + { 12742, 0x00008027 }, /* GL_HISTOGRAM_FORMAT */ + { 12762, 0x00008027 }, /* GL_HISTOGRAM_FORMAT_EXT */ + { 12786, 0x00008029 }, /* GL_HISTOGRAM_GREEN_SIZE */ + { 12810, 0x00008029 }, /* GL_HISTOGRAM_GREEN_SIZE_EXT */ + { 12838, 0x0000802C }, /* GL_HISTOGRAM_LUMINANCE_SIZE */ + { 12866, 0x0000802C }, /* GL_HISTOGRAM_LUMINANCE_SIZE_EXT */ + { 12898, 0x00008028 }, /* GL_HISTOGRAM_RED_SIZE */ + { 12920, 0x00008028 }, /* GL_HISTOGRAM_RED_SIZE_EXT */ + { 12946, 0x0000802D }, /* GL_HISTOGRAM_SINK */ + { 12964, 0x0000802D }, /* GL_HISTOGRAM_SINK_EXT */ + { 12986, 0x00008026 }, /* GL_HISTOGRAM_WIDTH */ + { 13005, 0x00008026 }, /* GL_HISTOGRAM_WIDTH_EXT */ + { 13028, 0x0000862A }, /* GL_IDENTITY_NV */ + { 13043, 0x00008150 }, /* GL_IGNORE_BORDER_HP */ + { 13063, 0x00008B9B }, /* GL_IMPLEMENTATION_COLOR_READ_FORMAT_OES */ + { 13103, 0x00008B9A }, /* GL_IMPLEMENTATION_COLOR_READ_TYPE_OES */ + { 13141, 0x00001E02 }, /* GL_INCR */ + { 13149, 0x00008507 }, /* GL_INCR_WRAP */ + { 13162, 0x00008507 }, /* GL_INCR_WRAP_EXT */ + { 13179, 0x00008222 }, /* GL_INDEX */ + { 13188, 0x00008077 }, /* GL_INDEX_ARRAY */ + { 13203, 0x00008899 }, /* GL_INDEX_ARRAY_BUFFER_BINDING */ + { 13233, 0x00008899 }, /* GL_INDEX_ARRAY_BUFFER_BINDING_ARB */ + { 13267, 0x00008091 }, /* GL_INDEX_ARRAY_POINTER */ + { 13290, 0x00008086 }, /* GL_INDEX_ARRAY_STRIDE */ + { 13312, 0x00008085 }, /* GL_INDEX_ARRAY_TYPE */ + { 13332, 0x00000D51 }, /* GL_INDEX_BITS */ + { 13346, 0x00000C20 }, /* GL_INDEX_CLEAR_VALUE */ + { 13367, 0x00000BF1 }, /* GL_INDEX_LOGIC_OP */ + { 13385, 0x00000C30 }, /* GL_INDEX_MODE */ + { 13399, 0x00000D13 }, /* GL_INDEX_OFFSET */ + { 13415, 0x00000D12 }, /* GL_INDEX_SHIFT */ + { 13430, 0x00000C21 }, /* GL_INDEX_WRITEMASK */ + { 13449, 0x00008B84 }, /* GL_INFO_LOG_LENGTH */ + { 13468, 0x00001404 }, /* GL_INT */ + { 13475, 0x00008049 }, /* GL_INTENSITY */ + { 13488, 0x0000804C }, /* GL_INTENSITY12 */ + { 13503, 0x0000804C }, /* GL_INTENSITY12_EXT */ + { 13522, 0x0000804D }, /* GL_INTENSITY16 */ + { 13537, 0x0000804D }, /* GL_INTENSITY16_EXT */ + { 13556, 0x0000804A }, /* GL_INTENSITY4 */ + { 13570, 0x0000804A }, /* GL_INTENSITY4_EXT */ + { 13588, 0x0000804B }, /* GL_INTENSITY8 */ + { 13602, 0x0000804B }, /* GL_INTENSITY8_EXT */ + { 13620, 0x00008049 }, /* GL_INTENSITY_EXT */ + { 13637, 0x00008575 }, /* GL_INTERPOLATE */ + { 13652, 0x00008575 }, /* GL_INTERPOLATE_ARB */ + { 13671, 0x00008575 }, /* GL_INTERPOLATE_EXT */ + { 13690, 0x00008B53 }, /* GL_INT_VEC2 */ + { 13702, 0x00008B53 }, /* GL_INT_VEC2_ARB */ + { 13718, 0x00008B54 }, /* GL_INT_VEC3 */ + { 13730, 0x00008B54 }, /* GL_INT_VEC3_ARB */ + { 13746, 0x00008B55 }, /* GL_INT_VEC4 */ + { 13758, 0x00008B55 }, /* GL_INT_VEC4_ARB */ + { 13774, 0x00000500 }, /* GL_INVALID_ENUM */ + { 13790, 0x00000506 }, /* GL_INVALID_FRAMEBUFFER_OPERATION */ + { 13823, 0x00000506 }, /* GL_INVALID_FRAMEBUFFER_OPERATION_EXT */ + { 13860, 0x00000502 }, /* GL_INVALID_OPERATION */ + { 13881, 0x00000501 }, /* GL_INVALID_VALUE */ + { 13898, 0x0000862B }, /* GL_INVERSE_NV */ + { 13912, 0x0000862D }, /* GL_INVERSE_TRANSPOSE_NV */ + { 13936, 0x0000150A }, /* GL_INVERT */ + { 13946, 0x00001E00 }, /* GL_KEEP */ + { 13954, 0x00008E4E }, /* GL_LAST_VERTEX_CONVENTION */ + { 13980, 0x00008E4E }, /* GL_LAST_VERTEX_CONVENTION_EXT */ + { 14010, 0x00000406 }, /* GL_LEFT */ + { 14018, 0x00000203 }, /* GL_LEQUAL */ + { 14028, 0x00000201 }, /* GL_LESS */ + { 14036, 0x00004000 }, /* GL_LIGHT0 */ + { 14046, 0x00004001 }, /* GL_LIGHT1 */ + { 14056, 0x00004002 }, /* GL_LIGHT2 */ + { 14066, 0x00004003 }, /* GL_LIGHT3 */ + { 14076, 0x00004004 }, /* GL_LIGHT4 */ + { 14086, 0x00004005 }, /* GL_LIGHT5 */ + { 14096, 0x00004006 }, /* GL_LIGHT6 */ + { 14106, 0x00004007 }, /* GL_LIGHT7 */ + { 14116, 0x00000B50 }, /* GL_LIGHTING */ + { 14128, 0x00000040 }, /* GL_LIGHTING_BIT */ + { 14144, 0x00000B53 }, /* GL_LIGHT_MODEL_AMBIENT */ + { 14167, 0x000081F8 }, /* GL_LIGHT_MODEL_COLOR_CONTROL */ + { 14196, 0x000081F8 }, /* GL_LIGHT_MODEL_COLOR_CONTROL_EXT */ + { 14229, 0x00000B51 }, /* GL_LIGHT_MODEL_LOCAL_VIEWER */ + { 14257, 0x00000B52 }, /* GL_LIGHT_MODEL_TWO_SIDE */ + { 14281, 0x00001B01 }, /* GL_LINE */ + { 14289, 0x00002601 }, /* GL_LINEAR */ + { 14299, 0x00001208 }, /* GL_LINEAR_ATTENUATION */ + { 14321, 0x00008170 }, /* GL_LINEAR_CLIPMAP_LINEAR_SGIX */ + { 14351, 0x0000844F }, /* GL_LINEAR_CLIPMAP_NEAREST_SGIX */ + { 14382, 0x00002703 }, /* GL_LINEAR_MIPMAP_LINEAR */ + { 14406, 0x00002701 }, /* GL_LINEAR_MIPMAP_NEAREST */ + { 14431, 0x00000001 }, /* GL_LINES */ + { 14440, 0x00000004 }, /* GL_LINE_BIT */ + { 14452, 0x00000002 }, /* GL_LINE_LOOP */ + { 14465, 0x00000707 }, /* GL_LINE_RESET_TOKEN */ + { 14485, 0x00000B20 }, /* GL_LINE_SMOOTH */ + { 14500, 0x00000C52 }, /* GL_LINE_SMOOTH_HINT */ + { 14520, 0x00000B24 }, /* GL_LINE_STIPPLE */ + { 14536, 0x00000B25 }, /* GL_LINE_STIPPLE_PATTERN */ + { 14560, 0x00000B26 }, /* GL_LINE_STIPPLE_REPEAT */ + { 14583, 0x00000003 }, /* GL_LINE_STRIP */ + { 14597, 0x00000702 }, /* GL_LINE_TOKEN */ + { 14611, 0x00000B21 }, /* GL_LINE_WIDTH */ + { 14625, 0x00000B23 }, /* GL_LINE_WIDTH_GRANULARITY */ + { 14651, 0x00000B22 }, /* GL_LINE_WIDTH_RANGE */ + { 14671, 0x00008B82 }, /* GL_LINK_STATUS */ + { 14686, 0x00000B32 }, /* GL_LIST_BASE */ + { 14699, 0x00020000 }, /* GL_LIST_BIT */ + { 14711, 0x00000B33 }, /* GL_LIST_INDEX */ + { 14725, 0x00000B30 }, /* GL_LIST_MODE */ + { 14738, 0x00000101 }, /* GL_LOAD */ + { 14746, 0x00000BF1 }, /* GL_LOGIC_OP */ + { 14758, 0x00000BF0 }, /* GL_LOGIC_OP_MODE */ + { 14775, 0x00008CA1 }, /* GL_LOWER_LEFT */ + { 14789, 0x00001909 }, /* GL_LUMINANCE */ + { 14802, 0x00008041 }, /* GL_LUMINANCE12 */ + { 14817, 0x00008047 }, /* GL_LUMINANCE12_ALPHA12 */ + { 14840, 0x00008047 }, /* GL_LUMINANCE12_ALPHA12_EXT */ + { 14867, 0x00008046 }, /* GL_LUMINANCE12_ALPHA4 */ + { 14889, 0x00008046 }, /* GL_LUMINANCE12_ALPHA4_EXT */ + { 14915, 0x00008041 }, /* GL_LUMINANCE12_EXT */ + { 14934, 0x00008042 }, /* GL_LUMINANCE16 */ + { 14949, 0x00008048 }, /* GL_LUMINANCE16_ALPHA16 */ + { 14972, 0x00008048 }, /* GL_LUMINANCE16_ALPHA16_EXT */ + { 14999, 0x00008042 }, /* GL_LUMINANCE16_EXT */ + { 15018, 0x0000803F }, /* GL_LUMINANCE4 */ + { 15032, 0x00008043 }, /* GL_LUMINANCE4_ALPHA4 */ + { 15053, 0x00008043 }, /* GL_LUMINANCE4_ALPHA4_EXT */ + { 15078, 0x0000803F }, /* GL_LUMINANCE4_EXT */ + { 15096, 0x00008044 }, /* GL_LUMINANCE6_ALPHA2 */ + { 15117, 0x00008044 }, /* GL_LUMINANCE6_ALPHA2_EXT */ + { 15142, 0x00008040 }, /* GL_LUMINANCE8 */ + { 15156, 0x00008045 }, /* GL_LUMINANCE8_ALPHA8 */ + { 15177, 0x00008045 }, /* GL_LUMINANCE8_ALPHA8_EXT */ + { 15202, 0x00008040 }, /* GL_LUMINANCE8_EXT */ + { 15220, 0x0000190A }, /* GL_LUMINANCE_ALPHA */ + { 15239, 0x00000D90 }, /* GL_MAP1_COLOR_4 */ + { 15255, 0x00000DD0 }, /* GL_MAP1_GRID_DOMAIN */ + { 15275, 0x00000DD1 }, /* GL_MAP1_GRID_SEGMENTS */ + { 15297, 0x00000D91 }, /* GL_MAP1_INDEX */ + { 15311, 0x00000D92 }, /* GL_MAP1_NORMAL */ + { 15326, 0x00000D93 }, /* GL_MAP1_TEXTURE_COORD_1 */ + { 15350, 0x00000D94 }, /* GL_MAP1_TEXTURE_COORD_2 */ + { 15374, 0x00000D95 }, /* GL_MAP1_TEXTURE_COORD_3 */ + { 15398, 0x00000D96 }, /* GL_MAP1_TEXTURE_COORD_4 */ + { 15422, 0x00000D97 }, /* GL_MAP1_VERTEX_3 */ + { 15439, 0x00000D98 }, /* GL_MAP1_VERTEX_4 */ + { 15456, 0x00008660 }, /* GL_MAP1_VERTEX_ATTRIB0_4_NV */ + { 15484, 0x0000866A }, /* GL_MAP1_VERTEX_ATTRIB10_4_NV */ + { 15513, 0x0000866B }, /* GL_MAP1_VERTEX_ATTRIB11_4_NV */ + { 15542, 0x0000866C }, /* GL_MAP1_VERTEX_ATTRIB12_4_NV */ + { 15571, 0x0000866D }, /* GL_MAP1_VERTEX_ATTRIB13_4_NV */ + { 15600, 0x0000866E }, /* GL_MAP1_VERTEX_ATTRIB14_4_NV */ + { 15629, 0x0000866F }, /* GL_MAP1_VERTEX_ATTRIB15_4_NV */ + { 15658, 0x00008661 }, /* GL_MAP1_VERTEX_ATTRIB1_4_NV */ + { 15686, 0x00008662 }, /* GL_MAP1_VERTEX_ATTRIB2_4_NV */ + { 15714, 0x00008663 }, /* GL_MAP1_VERTEX_ATTRIB3_4_NV */ + { 15742, 0x00008664 }, /* GL_MAP1_VERTEX_ATTRIB4_4_NV */ + { 15770, 0x00008665 }, /* GL_MAP1_VERTEX_ATTRIB5_4_NV */ + { 15798, 0x00008666 }, /* GL_MAP1_VERTEX_ATTRIB6_4_NV */ + { 15826, 0x00008667 }, /* GL_MAP1_VERTEX_ATTRIB7_4_NV */ + { 15854, 0x00008668 }, /* GL_MAP1_VERTEX_ATTRIB8_4_NV */ + { 15882, 0x00008669 }, /* GL_MAP1_VERTEX_ATTRIB9_4_NV */ + { 15910, 0x00000DB0 }, /* GL_MAP2_COLOR_4 */ + { 15926, 0x00000DD2 }, /* GL_MAP2_GRID_DOMAIN */ + { 15946, 0x00000DD3 }, /* GL_MAP2_GRID_SEGMENTS */ + { 15968, 0x00000DB1 }, /* GL_MAP2_INDEX */ + { 15982, 0x00000DB2 }, /* GL_MAP2_NORMAL */ + { 15997, 0x00000DB3 }, /* GL_MAP2_TEXTURE_COORD_1 */ + { 16021, 0x00000DB4 }, /* GL_MAP2_TEXTURE_COORD_2 */ + { 16045, 0x00000DB5 }, /* GL_MAP2_TEXTURE_COORD_3 */ + { 16069, 0x00000DB6 }, /* GL_MAP2_TEXTURE_COORD_4 */ + { 16093, 0x00000DB7 }, /* GL_MAP2_VERTEX_3 */ + { 16110, 0x00000DB8 }, /* GL_MAP2_VERTEX_4 */ + { 16127, 0x00008670 }, /* GL_MAP2_VERTEX_ATTRIB0_4_NV */ + { 16155, 0x0000867A }, /* GL_MAP2_VERTEX_ATTRIB10_4_NV */ + { 16184, 0x0000867B }, /* GL_MAP2_VERTEX_ATTRIB11_4_NV */ + { 16213, 0x0000867C }, /* GL_MAP2_VERTEX_ATTRIB12_4_NV */ + { 16242, 0x0000867D }, /* GL_MAP2_VERTEX_ATTRIB13_4_NV */ + { 16271, 0x0000867E }, /* GL_MAP2_VERTEX_ATTRIB14_4_NV */ + { 16300, 0x0000867F }, /* GL_MAP2_VERTEX_ATTRIB15_4_NV */ + { 16329, 0x00008671 }, /* GL_MAP2_VERTEX_ATTRIB1_4_NV */ + { 16357, 0x00008672 }, /* GL_MAP2_VERTEX_ATTRIB2_4_NV */ + { 16385, 0x00008673 }, /* GL_MAP2_VERTEX_ATTRIB3_4_NV */ + { 16413, 0x00008674 }, /* GL_MAP2_VERTEX_ATTRIB4_4_NV */ + { 16441, 0x00008675 }, /* GL_MAP2_VERTEX_ATTRIB5_4_NV */ + { 16469, 0x00008676 }, /* GL_MAP2_VERTEX_ATTRIB6_4_NV */ + { 16497, 0x00008677 }, /* GL_MAP2_VERTEX_ATTRIB7_4_NV */ + { 16525, 0x00008678 }, /* GL_MAP2_VERTEX_ATTRIB8_4_NV */ + { 16553, 0x00008679 }, /* GL_MAP2_VERTEX_ATTRIB9_4_NV */ + { 16581, 0x00000D10 }, /* GL_MAP_COLOR */ + { 16594, 0x00000010 }, /* GL_MAP_FLUSH_EXPLICIT_BIT */ + { 16620, 0x00000008 }, /* GL_MAP_INVALIDATE_BUFFER_BIT */ + { 16649, 0x00000004 }, /* GL_MAP_INVALIDATE_RANGE_BIT */ + { 16677, 0x00000001 }, /* GL_MAP_READ_BIT */ + { 16693, 0x00000D11 }, /* GL_MAP_STENCIL */ + { 16708, 0x00000020 }, /* GL_MAP_UNSYNCHRONIZED_BIT */ + { 16734, 0x00000002 }, /* GL_MAP_WRITE_BIT */ + { 16751, 0x000088C0 }, /* GL_MATRIX0_ARB */ + { 16766, 0x00008630 }, /* GL_MATRIX0_NV */ + { 16780, 0x000088CA }, /* GL_MATRIX10_ARB */ + { 16796, 0x000088CB }, /* GL_MATRIX11_ARB */ + { 16812, 0x000088CC }, /* GL_MATRIX12_ARB */ + { 16828, 0x000088CD }, /* GL_MATRIX13_ARB */ + { 16844, 0x000088CE }, /* GL_MATRIX14_ARB */ + { 16860, 0x000088CF }, /* GL_MATRIX15_ARB */ + { 16876, 0x000088D0 }, /* GL_MATRIX16_ARB */ + { 16892, 0x000088D1 }, /* GL_MATRIX17_ARB */ + { 16908, 0x000088D2 }, /* GL_MATRIX18_ARB */ + { 16924, 0x000088D3 }, /* GL_MATRIX19_ARB */ + { 16940, 0x000088C1 }, /* GL_MATRIX1_ARB */ + { 16955, 0x00008631 }, /* GL_MATRIX1_NV */ + { 16969, 0x000088D4 }, /* GL_MATRIX20_ARB */ + { 16985, 0x000088D5 }, /* GL_MATRIX21_ARB */ + { 17001, 0x000088D6 }, /* GL_MATRIX22_ARB */ + { 17017, 0x000088D7 }, /* GL_MATRIX23_ARB */ + { 17033, 0x000088D8 }, /* GL_MATRIX24_ARB */ + { 17049, 0x000088D9 }, /* GL_MATRIX25_ARB */ + { 17065, 0x000088DA }, /* GL_MATRIX26_ARB */ + { 17081, 0x000088DB }, /* GL_MATRIX27_ARB */ + { 17097, 0x000088DC }, /* GL_MATRIX28_ARB */ + { 17113, 0x000088DD }, /* GL_MATRIX29_ARB */ + { 17129, 0x000088C2 }, /* GL_MATRIX2_ARB */ + { 17144, 0x00008632 }, /* GL_MATRIX2_NV */ + { 17158, 0x000088DE }, /* GL_MATRIX30_ARB */ + { 17174, 0x000088DF }, /* GL_MATRIX31_ARB */ + { 17190, 0x000088C3 }, /* GL_MATRIX3_ARB */ + { 17205, 0x00008633 }, /* GL_MATRIX3_NV */ + { 17219, 0x000088C4 }, /* GL_MATRIX4_ARB */ + { 17234, 0x00008634 }, /* GL_MATRIX4_NV */ + { 17248, 0x000088C5 }, /* GL_MATRIX5_ARB */ + { 17263, 0x00008635 }, /* GL_MATRIX5_NV */ + { 17277, 0x000088C6 }, /* GL_MATRIX6_ARB */ + { 17292, 0x00008636 }, /* GL_MATRIX6_NV */ + { 17306, 0x000088C7 }, /* GL_MATRIX7_ARB */ + { 17321, 0x00008637 }, /* GL_MATRIX7_NV */ + { 17335, 0x000088C8 }, /* GL_MATRIX8_ARB */ + { 17350, 0x000088C9 }, /* GL_MATRIX9_ARB */ + { 17365, 0x00008844 }, /* GL_MATRIX_INDEX_ARRAY_ARB */ + { 17391, 0x00008849 }, /* GL_MATRIX_INDEX_ARRAY_POINTER_ARB */ + { 17425, 0x00008846 }, /* GL_MATRIX_INDEX_ARRAY_SIZE_ARB */ + { 17456, 0x00008848 }, /* GL_MATRIX_INDEX_ARRAY_STRIDE_ARB */ + { 17489, 0x00008847 }, /* GL_MATRIX_INDEX_ARRAY_TYPE_ARB */ + { 17520, 0x00000BA0 }, /* GL_MATRIX_MODE */ + { 17535, 0x00008840 }, /* GL_MATRIX_PALETTE_ARB */ + { 17557, 0x00008008 }, /* GL_MAX */ + { 17564, 0x00008073 }, /* GL_MAX_3D_TEXTURE_SIZE */ + { 17587, 0x000088FF }, /* GL_MAX_ARRAY_TEXTURE_LAYERS_EXT */ + { 17619, 0x00000D35 }, /* GL_MAX_ATTRIB_STACK_DEPTH */ + { 17645, 0x00000D3B }, /* GL_MAX_CLIENT_ATTRIB_STACK_DEPTH */ + { 17678, 0x00008177 }, /* GL_MAX_CLIPMAP_DEPTH_SGIX */ + { 17704, 0x00008178 }, /* GL_MAX_CLIPMAP_VIRTUAL_DEPTH_SGIX */ + { 17738, 0x00000D32 }, /* GL_MAX_CLIP_PLANES */ + { 17757, 0x00008CDF }, /* GL_MAX_COLOR_ATTACHMENTS_EXT */ + { 17786, 0x000080B3 }, /* GL_MAX_COLOR_MATRIX_STACK_DEPTH */ + { 17818, 0x000080B3 }, /* GL_MAX_COLOR_MATRIX_STACK_DEPTH_SGI */ + { 17854, 0x00008B4D }, /* GL_MAX_COMBINED_TEXTURE_IMAGE_UNITS */ + { 17890, 0x00008B4D }, /* GL_MAX_COMBINED_TEXTURE_IMAGE_UNITS_ARB */ + { 17930, 0x0000801B }, /* GL_MAX_CONVOLUTION_HEIGHT */ + { 17956, 0x0000801B }, /* GL_MAX_CONVOLUTION_HEIGHT_EXT */ + { 17986, 0x0000801A }, /* GL_MAX_CONVOLUTION_WIDTH */ + { 18011, 0x0000801A }, /* GL_MAX_CONVOLUTION_WIDTH_EXT */ + { 18040, 0x0000851C }, /* GL_MAX_CUBE_MAP_TEXTURE_SIZE */ + { 18069, 0x0000851C }, /* GL_MAX_CUBE_MAP_TEXTURE_SIZE_ARB */ + { 18102, 0x00008824 }, /* GL_MAX_DRAW_BUFFERS */ + { 18122, 0x00008824 }, /* GL_MAX_DRAW_BUFFERS_ARB */ + { 18146, 0x00008824 }, /* GL_MAX_DRAW_BUFFERS_ATI */ + { 18170, 0x000080E9 }, /* GL_MAX_ELEMENTS_INDICES */ + { 18194, 0x000080E8 }, /* GL_MAX_ELEMENTS_VERTICES */ + { 18219, 0x00000D30 }, /* GL_MAX_EVAL_ORDER */ + { 18237, 0x00008008 }, /* GL_MAX_EXT */ + { 18248, 0x00008B49 }, /* GL_MAX_FRAGMENT_UNIFORM_COMPONENTS */ + { 18283, 0x00008B49 }, /* GL_MAX_FRAGMENT_UNIFORM_COMPONENTS_ARB */ + { 18322, 0x00000D31 }, /* GL_MAX_LIGHTS */ + { 18336, 0x00000B31 }, /* GL_MAX_LIST_NESTING */ + { 18356, 0x00008841 }, /* GL_MAX_MATRIX_PALETTE_STACK_DEPTH_ARB */ + { 18394, 0x00000D36 }, /* GL_MAX_MODELVIEW_STACK_DEPTH */ + { 18423, 0x00000D37 }, /* GL_MAX_NAME_STACK_DEPTH */ + { 18447, 0x00008842 }, /* GL_MAX_PALETTE_MATRICES_ARB */ + { 18475, 0x00000D34 }, /* GL_MAX_PIXEL_MAP_TABLE */ + { 18498, 0x000088B1 }, /* GL_MAX_PROGRAM_ADDRESS_REGISTERS_ARB */ + { 18535, 0x0000880B }, /* GL_MAX_PROGRAM_ALU_INSTRUCTIONS_ARB */ + { 18571, 0x000088AD }, /* GL_MAX_PROGRAM_ATTRIBS_ARB */ + { 18598, 0x000088F5 }, /* GL_MAX_PROGRAM_CALL_DEPTH_NV */ + { 18627, 0x000088B5 }, /* GL_MAX_PROGRAM_ENV_PARAMETERS_ARB */ + { 18661, 0x000088F4 }, /* GL_MAX_PROGRAM_EXEC_INSTRUCTIONS_NV */ + { 18697, 0x000088F6 }, /* GL_MAX_PROGRAM_IF_DEPTH_NV */ + { 18724, 0x000088A1 }, /* GL_MAX_PROGRAM_INSTRUCTIONS_ARB */ + { 18756, 0x000088B4 }, /* GL_MAX_PROGRAM_LOCAL_PARAMETERS_ARB */ + { 18792, 0x000088F8 }, /* GL_MAX_PROGRAM_LOOP_COUNT_NV */ + { 18821, 0x000088F7 }, /* GL_MAX_PROGRAM_LOOP_DEPTH_NV */ + { 18850, 0x0000862F }, /* GL_MAX_PROGRAM_MATRICES_ARB */ + { 18878, 0x0000862E }, /* GL_MAX_PROGRAM_MATRIX_STACK_DEPTH_ARB */ + { 18916, 0x000088B3 }, /* GL_MAX_PROGRAM_NATIVE_ADDRESS_REGISTERS_ARB */ + { 18960, 0x0000880E }, /* GL_MAX_PROGRAM_NATIVE_ALU_INSTRUCTIONS_ARB */ + { 19003, 0x000088AF }, /* GL_MAX_PROGRAM_NATIVE_ATTRIBS_ARB */ + { 19037, 0x000088A3 }, /* GL_MAX_PROGRAM_NATIVE_INSTRUCTIONS_ARB */ + { 19076, 0x000088AB }, /* GL_MAX_PROGRAM_NATIVE_PARAMETERS_ARB */ + { 19113, 0x000088A7 }, /* GL_MAX_PROGRAM_NATIVE_TEMPORARIES_ARB */ + { 19151, 0x00008810 }, /* GL_MAX_PROGRAM_NATIVE_TEX_INDIRECTIONS_ARB */ + { 19194, 0x0000880F }, /* GL_MAX_PROGRAM_NATIVE_TEX_INSTRUCTIONS_ARB */ + { 19237, 0x000088A9 }, /* GL_MAX_PROGRAM_PARAMETERS_ARB */ + { 19267, 0x000088A5 }, /* GL_MAX_PROGRAM_TEMPORARIES_ARB */ + { 19298, 0x0000880D }, /* GL_MAX_PROGRAM_TEX_INDIRECTIONS_ARB */ + { 19334, 0x0000880C }, /* GL_MAX_PROGRAM_TEX_INSTRUCTIONS_ARB */ + { 19370, 0x00000D38 }, /* GL_MAX_PROJECTION_STACK_DEPTH */ + { 19400, 0x000084F8 }, /* GL_MAX_RECTANGLE_TEXTURE_SIZE_ARB */ + { 19434, 0x000084F8 }, /* GL_MAX_RECTANGLE_TEXTURE_SIZE_NV */ + { 19467, 0x000084E8 }, /* GL_MAX_RENDERBUFFER_SIZE_EXT */ + { 19496, 0x00008D57 }, /* GL_MAX_SAMPLES */ + { 19511, 0x00009111 }, /* GL_MAX_SERVER_WAIT_TIMEOUT */ + { 19538, 0x00008504 }, /* GL_MAX_SHININESS_NV */ + { 19558, 0x00008505 }, /* GL_MAX_SPOT_EXPONENT_NV */ + { 19582, 0x00008871 }, /* GL_MAX_TEXTURE_COORDS */ + { 19604, 0x00008871 }, /* GL_MAX_TEXTURE_COORDS_ARB */ + { 19630, 0x00008872 }, /* GL_MAX_TEXTURE_IMAGE_UNITS */ + { 19657, 0x00008872 }, /* GL_MAX_TEXTURE_IMAGE_UNITS_ARB */ + { 19688, 0x000084FD }, /* GL_MAX_TEXTURE_LOD_BIAS */ + { 19712, 0x000084FF }, /* GL_MAX_TEXTURE_MAX_ANISOTROPY_EXT */ + { 19746, 0x00000D33 }, /* GL_MAX_TEXTURE_SIZE */ + { 19766, 0x00000D39 }, /* GL_MAX_TEXTURE_STACK_DEPTH */ + { 19793, 0x000084E2 }, /* GL_MAX_TEXTURE_UNITS */ + { 19814, 0x000084E2 }, /* GL_MAX_TEXTURE_UNITS_ARB */ + { 19839, 0x0000862F }, /* GL_MAX_TRACK_MATRICES_NV */ + { 19864, 0x0000862E }, /* GL_MAX_TRACK_MATRIX_STACK_DEPTH_NV */ + { 19899, 0x00008B4B }, /* GL_MAX_VARYING_FLOATS */ + { 19921, 0x00008B4B }, /* GL_MAX_VARYING_FLOATS_ARB */ + { 19947, 0x00008869 }, /* GL_MAX_VERTEX_ATTRIBS */ + { 19969, 0x00008869 }, /* GL_MAX_VERTEX_ATTRIBS_ARB */ + { 19995, 0x00008B4C }, /* GL_MAX_VERTEX_TEXTURE_IMAGE_UNITS */ + { 20029, 0x00008B4C }, /* GL_MAX_VERTEX_TEXTURE_IMAGE_UNITS_ARB */ + { 20067, 0x00008B4A }, /* GL_MAX_VERTEX_UNIFORM_COMPONENTS */ + { 20100, 0x00008B4A }, /* GL_MAX_VERTEX_UNIFORM_COMPONENTS_ARB */ + { 20137, 0x000086A4 }, /* GL_MAX_VERTEX_UNITS_ARB */ + { 20161, 0x00000D3A }, /* GL_MAX_VIEWPORT_DIMS */ + { 20182, 0x00008007 }, /* GL_MIN */ + { 20189, 0x0000802E }, /* GL_MINMAX */ + { 20199, 0x0000802E }, /* GL_MINMAX_EXT */ + { 20213, 0x0000802F }, /* GL_MINMAX_FORMAT */ + { 20230, 0x0000802F }, /* GL_MINMAX_FORMAT_EXT */ + { 20251, 0x00008030 }, /* GL_MINMAX_SINK */ + { 20266, 0x00008030 }, /* GL_MINMAX_SINK_EXT */ + { 20285, 0x00008007 }, /* GL_MIN_EXT */ + { 20296, 0x00008370 }, /* GL_MIRRORED_REPEAT */ + { 20315, 0x00008370 }, /* GL_MIRRORED_REPEAT_ARB */ + { 20338, 0x00008370 }, /* GL_MIRRORED_REPEAT_IBM */ + { 20361, 0x00008742 }, /* GL_MIRROR_CLAMP_ATI */ + { 20381, 0x00008742 }, /* GL_MIRROR_CLAMP_EXT */ + { 20401, 0x00008912 }, /* GL_MIRROR_CLAMP_TO_BORDER_EXT */ + { 20431, 0x00008743 }, /* GL_MIRROR_CLAMP_TO_EDGE_ATI */ + { 20459, 0x00008743 }, /* GL_MIRROR_CLAMP_TO_EDGE_EXT */ + { 20487, 0x00001700 }, /* GL_MODELVIEW */ + { 20500, 0x00001700 }, /* GL_MODELVIEW0_ARB */ + { 20518, 0x0000872A }, /* GL_MODELVIEW10_ARB */ + { 20537, 0x0000872B }, /* GL_MODELVIEW11_ARB */ + { 20556, 0x0000872C }, /* GL_MODELVIEW12_ARB */ + { 20575, 0x0000872D }, /* GL_MODELVIEW13_ARB */ + { 20594, 0x0000872E }, /* GL_MODELVIEW14_ARB */ + { 20613, 0x0000872F }, /* GL_MODELVIEW15_ARB */ + { 20632, 0x00008730 }, /* GL_MODELVIEW16_ARB */ + { 20651, 0x00008731 }, /* GL_MODELVIEW17_ARB */ + { 20670, 0x00008732 }, /* GL_MODELVIEW18_ARB */ + { 20689, 0x00008733 }, /* GL_MODELVIEW19_ARB */ + { 20708, 0x0000850A }, /* GL_MODELVIEW1_ARB */ + { 20726, 0x00008734 }, /* GL_MODELVIEW20_ARB */ + { 20745, 0x00008735 }, /* GL_MODELVIEW21_ARB */ + { 20764, 0x00008736 }, /* GL_MODELVIEW22_ARB */ + { 20783, 0x00008737 }, /* GL_MODELVIEW23_ARB */ + { 20802, 0x00008738 }, /* GL_MODELVIEW24_ARB */ + { 20821, 0x00008739 }, /* GL_MODELVIEW25_ARB */ + { 20840, 0x0000873A }, /* GL_MODELVIEW26_ARB */ + { 20859, 0x0000873B }, /* GL_MODELVIEW27_ARB */ + { 20878, 0x0000873C }, /* GL_MODELVIEW28_ARB */ + { 20897, 0x0000873D }, /* GL_MODELVIEW29_ARB */ + { 20916, 0x00008722 }, /* GL_MODELVIEW2_ARB */ + { 20934, 0x0000873E }, /* GL_MODELVIEW30_ARB */ + { 20953, 0x0000873F }, /* GL_MODELVIEW31_ARB */ + { 20972, 0x00008723 }, /* GL_MODELVIEW3_ARB */ + { 20990, 0x00008724 }, /* GL_MODELVIEW4_ARB */ + { 21008, 0x00008725 }, /* GL_MODELVIEW5_ARB */ + { 21026, 0x00008726 }, /* GL_MODELVIEW6_ARB */ + { 21044, 0x00008727 }, /* GL_MODELVIEW7_ARB */ + { 21062, 0x00008728 }, /* GL_MODELVIEW8_ARB */ + { 21080, 0x00008729 }, /* GL_MODELVIEW9_ARB */ + { 21098, 0x00000BA6 }, /* GL_MODELVIEW_MATRIX */ + { 21118, 0x00008629 }, /* GL_MODELVIEW_PROJECTION_NV */ + { 21145, 0x00000BA3 }, /* GL_MODELVIEW_STACK_DEPTH */ + { 21170, 0x00002100 }, /* GL_MODULATE */ + { 21182, 0x00008744 }, /* GL_MODULATE_ADD_ATI */ + { 21202, 0x00008745 }, /* GL_MODULATE_SIGNED_ADD_ATI */ + { 21229, 0x00008746 }, /* GL_MODULATE_SUBTRACT_ATI */ + { 21254, 0x00000103 }, /* GL_MULT */ + { 21262, 0x0000809D }, /* GL_MULTISAMPLE */ + { 21277, 0x000086B2 }, /* GL_MULTISAMPLE_3DFX */ + { 21297, 0x0000809D }, /* GL_MULTISAMPLE_ARB */ + { 21316, 0x20000000 }, /* GL_MULTISAMPLE_BIT */ + { 21335, 0x20000000 }, /* GL_MULTISAMPLE_BIT_3DFX */ + { 21359, 0x20000000 }, /* GL_MULTISAMPLE_BIT_ARB */ + { 21382, 0x00008534 }, /* GL_MULTISAMPLE_FILTER_HINT_NV */ + { 21412, 0x00002A25 }, /* GL_N3F_V3F */ + { 21423, 0x00000D70 }, /* GL_NAME_STACK_DEPTH */ + { 21443, 0x0000150E }, /* GL_NAND */ + { 21451, 0x00002600 }, /* GL_NEAREST */ + { 21462, 0x0000844E }, /* GL_NEAREST_CLIPMAP_LINEAR_SGIX */ + { 21493, 0x0000844D }, /* GL_NEAREST_CLIPMAP_NEAREST_SGIX */ + { 21525, 0x00002702 }, /* GL_NEAREST_MIPMAP_LINEAR */ + { 21550, 0x00002700 }, /* GL_NEAREST_MIPMAP_NEAREST */ + { 21576, 0x00000200 }, /* GL_NEVER */ + { 21585, 0x00001102 }, /* GL_NICEST */ + { 21595, 0x00000000 }, /* GL_NONE */ + { 21603, 0x00001505 }, /* GL_NOOP */ + { 21611, 0x00001508 }, /* GL_NOR */ + { 21618, 0x00000BA1 }, /* GL_NORMALIZE */ + { 21631, 0x00008075 }, /* GL_NORMAL_ARRAY */ + { 21647, 0x00008897 }, /* GL_NORMAL_ARRAY_BUFFER_BINDING */ + { 21678, 0x00008897 }, /* GL_NORMAL_ARRAY_BUFFER_BINDING_ARB */ + { 21713, 0x0000808F }, /* GL_NORMAL_ARRAY_POINTER */ + { 21737, 0x0000807F }, /* GL_NORMAL_ARRAY_STRIDE */ + { 21760, 0x0000807E }, /* GL_NORMAL_ARRAY_TYPE */ + { 21781, 0x00008511 }, /* GL_NORMAL_MAP */ + { 21795, 0x00008511 }, /* GL_NORMAL_MAP_ARB */ + { 21813, 0x00008511 }, /* GL_NORMAL_MAP_NV */ + { 21830, 0x00000205 }, /* GL_NOTEQUAL */ + { 21842, 0x00000000 }, /* GL_NO_ERROR */ + { 21854, 0x000086A2 }, /* GL_NUM_COMPRESSED_TEXTURE_FORMATS */ + { 21888, 0x000086A2 }, /* GL_NUM_COMPRESSED_TEXTURE_FORMATS_ARB */ + { 21926, 0x00008B89 }, /* GL_OBJECT_ACTIVE_ATTRIBUTES_ARB */ + { 21958, 0x00008B8A }, /* GL_OBJECT_ACTIVE_ATTRIBUTE_MAX_LENGTH_ARB */ + { 22000, 0x00008B86 }, /* GL_OBJECT_ACTIVE_UNIFORMS_ARB */ + { 22030, 0x00008B87 }, /* GL_OBJECT_ACTIVE_UNIFORM_MAX_LENGTH_ARB */ + { 22070, 0x00008B85 }, /* GL_OBJECT_ATTACHED_OBJECTS_ARB */ + { 22101, 0x00008B81 }, /* GL_OBJECT_COMPILE_STATUS_ARB */ + { 22130, 0x00008B80 }, /* GL_OBJECT_DELETE_STATUS_ARB */ + { 22158, 0x00008B84 }, /* GL_OBJECT_INFO_LOG_LENGTH_ARB */ + { 22188, 0x00002401 }, /* GL_OBJECT_LINEAR */ + { 22205, 0x00008B82 }, /* GL_OBJECT_LINK_STATUS_ARB */ + { 22231, 0x00002501 }, /* GL_OBJECT_PLANE */ + { 22247, 0x00008B88 }, /* GL_OBJECT_SHADER_SOURCE_LENGTH_ARB */ + { 22282, 0x00008B4F }, /* GL_OBJECT_SUBTYPE_ARB */ + { 22304, 0x00009112 }, /* GL_OBJECT_TYPE */ + { 22319, 0x00008B4E }, /* GL_OBJECT_TYPE_ARB */ + { 22338, 0x00008B83 }, /* GL_OBJECT_VALIDATE_STATUS_ARB */ + { 22368, 0x00008165 }, /* GL_OCCLUSION_TEST_HP */ + { 22389, 0x00008166 }, /* GL_OCCLUSION_TEST_RESULT_HP */ + { 22417, 0x00000001 }, /* GL_ONE */ + { 22424, 0x00008004 }, /* GL_ONE_MINUS_CONSTANT_ALPHA */ + { 22452, 0x00008004 }, /* GL_ONE_MINUS_CONSTANT_ALPHA_EXT */ + { 22484, 0x00008002 }, /* GL_ONE_MINUS_CONSTANT_COLOR */ + { 22512, 0x00008002 }, /* GL_ONE_MINUS_CONSTANT_COLOR_EXT */ + { 22544, 0x00000305 }, /* GL_ONE_MINUS_DST_ALPHA */ + { 22567, 0x00000307 }, /* GL_ONE_MINUS_DST_COLOR */ + { 22590, 0x00000303 }, /* GL_ONE_MINUS_SRC_ALPHA */ + { 22613, 0x00000301 }, /* GL_ONE_MINUS_SRC_COLOR */ + { 22636, 0x00008598 }, /* GL_OPERAND0_ALPHA */ + { 22654, 0x00008598 }, /* GL_OPERAND0_ALPHA_ARB */ + { 22676, 0x00008598 }, /* GL_OPERAND0_ALPHA_EXT */ + { 22698, 0x00008590 }, /* GL_OPERAND0_RGB */ + { 22714, 0x00008590 }, /* GL_OPERAND0_RGB_ARB */ + { 22734, 0x00008590 }, /* GL_OPERAND0_RGB_EXT */ + { 22754, 0x00008599 }, /* GL_OPERAND1_ALPHA */ + { 22772, 0x00008599 }, /* GL_OPERAND1_ALPHA_ARB */ + { 22794, 0x00008599 }, /* GL_OPERAND1_ALPHA_EXT */ + { 22816, 0x00008591 }, /* GL_OPERAND1_RGB */ + { 22832, 0x00008591 }, /* GL_OPERAND1_RGB_ARB */ + { 22852, 0x00008591 }, /* GL_OPERAND1_RGB_EXT */ + { 22872, 0x0000859A }, /* GL_OPERAND2_ALPHA */ + { 22890, 0x0000859A }, /* GL_OPERAND2_ALPHA_ARB */ + { 22912, 0x0000859A }, /* GL_OPERAND2_ALPHA_EXT */ + { 22934, 0x00008592 }, /* GL_OPERAND2_RGB */ + { 22950, 0x00008592 }, /* GL_OPERAND2_RGB_ARB */ + { 22970, 0x00008592 }, /* GL_OPERAND2_RGB_EXT */ + { 22990, 0x0000859B }, /* GL_OPERAND3_ALPHA_NV */ + { 23011, 0x00008593 }, /* GL_OPERAND3_RGB_NV */ + { 23030, 0x00001507 }, /* GL_OR */ + { 23036, 0x00000A01 }, /* GL_ORDER */ + { 23045, 0x0000150D }, /* GL_OR_INVERTED */ + { 23060, 0x0000150B }, /* GL_OR_REVERSE */ + { 23074, 0x00000505 }, /* GL_OUT_OF_MEMORY */ + { 23091, 0x00000D05 }, /* GL_PACK_ALIGNMENT */ + { 23109, 0x0000806C }, /* GL_PACK_IMAGE_HEIGHT */ + { 23130, 0x00008758 }, /* GL_PACK_INVERT_MESA */ + { 23150, 0x00000D01 }, /* GL_PACK_LSB_FIRST */ + { 23168, 0x00000D02 }, /* GL_PACK_ROW_LENGTH */ + { 23187, 0x0000806B }, /* GL_PACK_SKIP_IMAGES */ + { 23207, 0x00000D04 }, /* GL_PACK_SKIP_PIXELS */ + { 23227, 0x00000D03 }, /* GL_PACK_SKIP_ROWS */ + { 23245, 0x00000D00 }, /* GL_PACK_SWAP_BYTES */ + { 23264, 0x00008B92 }, /* GL_PALETTE4_R5_G6_B5_OES */ + { 23289, 0x00008B94 }, /* GL_PALETTE4_RGB5_A1_OES */ + { 23313, 0x00008B90 }, /* GL_PALETTE4_RGB8_OES */ + { 23334, 0x00008B93 }, /* GL_PALETTE4_RGBA4_OES */ + { 23356, 0x00008B91 }, /* GL_PALETTE4_RGBA8_OES */ + { 23378, 0x00008B97 }, /* GL_PALETTE8_R5_G6_B5_OES */ + { 23403, 0x00008B99 }, /* GL_PALETTE8_RGB5_A1_OES */ + { 23427, 0x00008B95 }, /* GL_PALETTE8_RGB8_OES */ + { 23448, 0x00008B98 }, /* GL_PALETTE8_RGBA4_OES */ + { 23470, 0x00008B96 }, /* GL_PALETTE8_RGBA8_OES */ + { 23492, 0x00000700 }, /* GL_PASS_THROUGH_TOKEN */ + { 23514, 0x00000C50 }, /* GL_PERSPECTIVE_CORRECTION_HINT */ + { 23545, 0x00000C79 }, /* GL_PIXEL_MAP_A_TO_A */ + { 23565, 0x00000CB9 }, /* GL_PIXEL_MAP_A_TO_A_SIZE */ + { 23590, 0x00000C78 }, /* GL_PIXEL_MAP_B_TO_B */ + { 23610, 0x00000CB8 }, /* GL_PIXEL_MAP_B_TO_B_SIZE */ + { 23635, 0x00000C77 }, /* GL_PIXEL_MAP_G_TO_G */ + { 23655, 0x00000CB7 }, /* GL_PIXEL_MAP_G_TO_G_SIZE */ + { 23680, 0x00000C75 }, /* GL_PIXEL_MAP_I_TO_A */ + { 23700, 0x00000CB5 }, /* GL_PIXEL_MAP_I_TO_A_SIZE */ + { 23725, 0x00000C74 }, /* GL_PIXEL_MAP_I_TO_B */ + { 23745, 0x00000CB4 }, /* GL_PIXEL_MAP_I_TO_B_SIZE */ + { 23770, 0x00000C73 }, /* GL_PIXEL_MAP_I_TO_G */ + { 23790, 0x00000CB3 }, /* GL_PIXEL_MAP_I_TO_G_SIZE */ + { 23815, 0x00000C70 }, /* GL_PIXEL_MAP_I_TO_I */ + { 23835, 0x00000CB0 }, /* GL_PIXEL_MAP_I_TO_I_SIZE */ + { 23860, 0x00000C72 }, /* GL_PIXEL_MAP_I_TO_R */ + { 23880, 0x00000CB2 }, /* GL_PIXEL_MAP_I_TO_R_SIZE */ + { 23905, 0x00000C76 }, /* GL_PIXEL_MAP_R_TO_R */ + { 23925, 0x00000CB6 }, /* GL_PIXEL_MAP_R_TO_R_SIZE */ + { 23950, 0x00000C71 }, /* GL_PIXEL_MAP_S_TO_S */ + { 23970, 0x00000CB1 }, /* GL_PIXEL_MAP_S_TO_S_SIZE */ + { 23995, 0x00000020 }, /* GL_PIXEL_MODE_BIT */ + { 24013, 0x000088EB }, /* GL_PIXEL_PACK_BUFFER */ + { 24034, 0x000088ED }, /* GL_PIXEL_PACK_BUFFER_BINDING */ + { 24063, 0x000088ED }, /* GL_PIXEL_PACK_BUFFER_BINDING_EXT */ + { 24096, 0x000088EB }, /* GL_PIXEL_PACK_BUFFER_EXT */ + { 24121, 0x000088EC }, /* GL_PIXEL_UNPACK_BUFFER */ + { 24144, 0x000088EF }, /* GL_PIXEL_UNPACK_BUFFER_BINDING */ + { 24175, 0x000088EF }, /* GL_PIXEL_UNPACK_BUFFER_BINDING_EXT */ + { 24210, 0x000088EC }, /* GL_PIXEL_UNPACK_BUFFER_EXT */ + { 24237, 0x00001B00 }, /* GL_POINT */ + { 24246, 0x00000000 }, /* GL_POINTS */ + { 24256, 0x00000002 }, /* GL_POINT_BIT */ + { 24269, 0x00008129 }, /* GL_POINT_DISTANCE_ATTENUATION */ + { 24299, 0x00008129 }, /* GL_POINT_DISTANCE_ATTENUATION_ARB */ + { 24333, 0x00008129 }, /* GL_POINT_DISTANCE_ATTENUATION_EXT */ + { 24367, 0x00008129 }, /* GL_POINT_DISTANCE_ATTENUATION_SGIS */ + { 24402, 0x00008128 }, /* GL_POINT_FADE_THRESHOLD_SIZE */ + { 24431, 0x00008128 }, /* GL_POINT_FADE_THRESHOLD_SIZE_ARB */ + { 24464, 0x00008128 }, /* GL_POINT_FADE_THRESHOLD_SIZE_EXT */ + { 24497, 0x00008128 }, /* GL_POINT_FADE_THRESHOLD_SIZE_SGIS */ + { 24531, 0x00000B11 }, /* GL_POINT_SIZE */ + { 24545, 0x00000B13 }, /* GL_POINT_SIZE_GRANULARITY */ + { 24571, 0x00008127 }, /* GL_POINT_SIZE_MAX */ + { 24589, 0x00008127 }, /* GL_POINT_SIZE_MAX_ARB */ + { 24611, 0x00008127 }, /* GL_POINT_SIZE_MAX_EXT */ + { 24633, 0x00008127 }, /* GL_POINT_SIZE_MAX_SGIS */ + { 24656, 0x00008126 }, /* GL_POINT_SIZE_MIN */ + { 24674, 0x00008126 }, /* GL_POINT_SIZE_MIN_ARB */ + { 24696, 0x00008126 }, /* GL_POINT_SIZE_MIN_EXT */ + { 24718, 0x00008126 }, /* GL_POINT_SIZE_MIN_SGIS */ + { 24741, 0x00000B12 }, /* GL_POINT_SIZE_RANGE */ + { 24761, 0x00000B10 }, /* GL_POINT_SMOOTH */ + { 24777, 0x00000C51 }, /* GL_POINT_SMOOTH_HINT */ + { 24798, 0x00008861 }, /* GL_POINT_SPRITE */ + { 24814, 0x00008861 }, /* GL_POINT_SPRITE_ARB */ + { 24834, 0x00008CA0 }, /* GL_POINT_SPRITE_COORD_ORIGIN */ + { 24863, 0x00008861 }, /* GL_POINT_SPRITE_NV */ + { 24882, 0x00008863 }, /* GL_POINT_SPRITE_R_MODE_NV */ + { 24908, 0x00000701 }, /* GL_POINT_TOKEN */ + { 24923, 0x00000009 }, /* GL_POLYGON */ + { 24934, 0x00000008 }, /* GL_POLYGON_BIT */ + { 24949, 0x00000B40 }, /* GL_POLYGON_MODE */ + { 24965, 0x00008039 }, /* GL_POLYGON_OFFSET_BIAS */ + { 24988, 0x00008038 }, /* GL_POLYGON_OFFSET_FACTOR */ + { 25013, 0x00008037 }, /* GL_POLYGON_OFFSET_FILL */ + { 25036, 0x00002A02 }, /* GL_POLYGON_OFFSET_LINE */ + { 25059, 0x00002A01 }, /* GL_POLYGON_OFFSET_POINT */ + { 25083, 0x00002A00 }, /* GL_POLYGON_OFFSET_UNITS */ + { 25107, 0x00000B41 }, /* GL_POLYGON_SMOOTH */ + { 25125, 0x00000C53 }, /* GL_POLYGON_SMOOTH_HINT */ + { 25148, 0x00000B42 }, /* GL_POLYGON_STIPPLE */ + { 25167, 0x00000010 }, /* GL_POLYGON_STIPPLE_BIT */ + { 25190, 0x00000703 }, /* GL_POLYGON_TOKEN */ + { 25207, 0x00001203 }, /* GL_POSITION */ + { 25219, 0x000080BB }, /* GL_POST_COLOR_MATRIX_ALPHA_BIAS */ + { 25251, 0x000080BB }, /* GL_POST_COLOR_MATRIX_ALPHA_BIAS_SGI */ + { 25287, 0x000080B7 }, /* GL_POST_COLOR_MATRIX_ALPHA_SCALE */ + { 25320, 0x000080B7 }, /* GL_POST_COLOR_MATRIX_ALPHA_SCALE_SGI */ + { 25357, 0x000080BA }, /* GL_POST_COLOR_MATRIX_BLUE_BIAS */ + { 25388, 0x000080BA }, /* GL_POST_COLOR_MATRIX_BLUE_BIAS_SGI */ + { 25423, 0x000080B6 }, /* GL_POST_COLOR_MATRIX_BLUE_SCALE */ + { 25455, 0x000080B6 }, /* GL_POST_COLOR_MATRIX_BLUE_SCALE_SGI */ + { 25491, 0x000080D2 }, /* GL_POST_COLOR_MATRIX_COLOR_TABLE */ + { 25524, 0x000080B9 }, /* GL_POST_COLOR_MATRIX_GREEN_BIAS */ + { 25556, 0x000080B9 }, /* GL_POST_COLOR_MATRIX_GREEN_BIAS_SGI */ + { 25592, 0x000080B5 }, /* GL_POST_COLOR_MATRIX_GREEN_SCALE */ + { 25625, 0x000080B5 }, /* GL_POST_COLOR_MATRIX_GREEN_SCALE_SGI */ + { 25662, 0x000080B8 }, /* GL_POST_COLOR_MATRIX_RED_BIAS */ + { 25692, 0x000080B8 }, /* GL_POST_COLOR_MATRIX_RED_BIAS_SGI */ + { 25726, 0x000080B4 }, /* GL_POST_COLOR_MATRIX_RED_SCALE */ + { 25757, 0x000080B4 }, /* GL_POST_COLOR_MATRIX_RED_SCALE_SGI */ + { 25792, 0x00008023 }, /* GL_POST_CONVOLUTION_ALPHA_BIAS */ + { 25823, 0x00008023 }, /* GL_POST_CONVOLUTION_ALPHA_BIAS_EXT */ + { 25858, 0x0000801F }, /* GL_POST_CONVOLUTION_ALPHA_SCALE */ + { 25890, 0x0000801F }, /* GL_POST_CONVOLUTION_ALPHA_SCALE_EXT */ + { 25926, 0x00008022 }, /* GL_POST_CONVOLUTION_BLUE_BIAS */ + { 25956, 0x00008022 }, /* GL_POST_CONVOLUTION_BLUE_BIAS_EXT */ + { 25990, 0x0000801E }, /* GL_POST_CONVOLUTION_BLUE_SCALE */ + { 26021, 0x0000801E }, /* GL_POST_CONVOLUTION_BLUE_SCALE_EXT */ + { 26056, 0x000080D1 }, /* GL_POST_CONVOLUTION_COLOR_TABLE */ + { 26088, 0x00008021 }, /* GL_POST_CONVOLUTION_GREEN_BIAS */ + { 26119, 0x00008021 }, /* GL_POST_CONVOLUTION_GREEN_BIAS_EXT */ + { 26154, 0x0000801D }, /* GL_POST_CONVOLUTION_GREEN_SCALE */ + { 26186, 0x0000801D }, /* GL_POST_CONVOLUTION_GREEN_SCALE_EXT */ + { 26222, 0x00008020 }, /* GL_POST_CONVOLUTION_RED_BIAS */ + { 26251, 0x00008020 }, /* GL_POST_CONVOLUTION_RED_BIAS_EXT */ + { 26284, 0x0000801C }, /* GL_POST_CONVOLUTION_RED_SCALE */ + { 26314, 0x0000801C }, /* GL_POST_CONVOLUTION_RED_SCALE_EXT */ + { 26348, 0x0000817B }, /* GL_POST_TEXTURE_FILTER_BIAS_RANGE_SGIX */ + { 26387, 0x00008179 }, /* GL_POST_TEXTURE_FILTER_BIAS_SGIX */ + { 26420, 0x0000817C }, /* GL_POST_TEXTURE_FILTER_SCALE_RANGE_SGIX */ + { 26460, 0x0000817A }, /* GL_POST_TEXTURE_FILTER_SCALE_SGIX */ + { 26494, 0x00008578 }, /* GL_PREVIOUS */ + { 26506, 0x00008578 }, /* GL_PREVIOUS_ARB */ + { 26522, 0x00008578 }, /* GL_PREVIOUS_EXT */ + { 26538, 0x00008577 }, /* GL_PRIMARY_COLOR */ + { 26555, 0x00008577 }, /* GL_PRIMARY_COLOR_ARB */ + { 26576, 0x00008577 }, /* GL_PRIMARY_COLOR_EXT */ + { 26597, 0x000088B0 }, /* GL_PROGRAM_ADDRESS_REGISTERS_ARB */ + { 26630, 0x00008805 }, /* GL_PROGRAM_ALU_INSTRUCTIONS_ARB */ + { 26662, 0x000088AC }, /* GL_PROGRAM_ATTRIBS_ARB */ + { 26685, 0x00008677 }, /* GL_PROGRAM_BINDING_ARB */ + { 26708, 0x0000864B }, /* GL_PROGRAM_ERROR_POSITION_ARB */ + { 26738, 0x0000864B }, /* GL_PROGRAM_ERROR_POSITION_NV */ + { 26767, 0x00008874 }, /* GL_PROGRAM_ERROR_STRING_ARB */ + { 26795, 0x00008876 }, /* GL_PROGRAM_FORMAT_ARB */ + { 26817, 0x00008875 }, /* GL_PROGRAM_FORMAT_ASCII_ARB */ + { 26845, 0x000088A0 }, /* GL_PROGRAM_INSTRUCTIONS_ARB */ + { 26873, 0x00008627 }, /* GL_PROGRAM_LENGTH_ARB */ + { 26895, 0x00008627 }, /* GL_PROGRAM_LENGTH_NV */ + { 26916, 0x000088B2 }, /* GL_PROGRAM_NATIVE_ADDRESS_REGISTERS_ARB */ + { 26956, 0x00008808 }, /* GL_PROGRAM_NATIVE_ALU_INSTRUCTIONS_ARB */ + { 26995, 0x000088AE }, /* GL_PROGRAM_NATIVE_ATTRIBS_ARB */ + { 27025, 0x000088A2 }, /* GL_PROGRAM_NATIVE_INSTRUCTIONS_ARB */ + { 27060, 0x000088AA }, /* GL_PROGRAM_NATIVE_PARAMETERS_ARB */ + { 27093, 0x000088A6 }, /* GL_PROGRAM_NATIVE_TEMPORARIES_ARB */ + { 27127, 0x0000880A }, /* GL_PROGRAM_NATIVE_TEX_INDIRECTIONS_ARB */ + { 27166, 0x00008809 }, /* GL_PROGRAM_NATIVE_TEX_INSTRUCTIONS_ARB */ + { 27205, 0x00008B40 }, /* GL_PROGRAM_OBJECT_ARB */ + { 27227, 0x000088A8 }, /* GL_PROGRAM_PARAMETERS_ARB */ + { 27253, 0x00008644 }, /* GL_PROGRAM_PARAMETER_NV */ + { 27277, 0x00008647 }, /* GL_PROGRAM_RESIDENT_NV */ + { 27300, 0x00008628 }, /* GL_PROGRAM_STRING_ARB */ + { 27322, 0x00008628 }, /* GL_PROGRAM_STRING_NV */ + { 27343, 0x00008646 }, /* GL_PROGRAM_TARGET_NV */ + { 27364, 0x000088A4 }, /* GL_PROGRAM_TEMPORARIES_ARB */ + { 27391, 0x00008807 }, /* GL_PROGRAM_TEX_INDIRECTIONS_ARB */ + { 27423, 0x00008806 }, /* GL_PROGRAM_TEX_INSTRUCTIONS_ARB */ + { 27455, 0x000088B6 }, /* GL_PROGRAM_UNDER_NATIVE_LIMITS_ARB */ + { 27490, 0x00001701 }, /* GL_PROJECTION */ + { 27504, 0x00000BA7 }, /* GL_PROJECTION_MATRIX */ + { 27525, 0x00000BA4 }, /* GL_PROJECTION_STACK_DEPTH */ + { 27551, 0x00008E4F }, /* GL_PROVOKING_VERTEX */ + { 27571, 0x00008E4F }, /* GL_PROVOKING_VERTEX_EXT */ + { 27595, 0x000080D3 }, /* GL_PROXY_COLOR_TABLE */ + { 27616, 0x00008025 }, /* GL_PROXY_HISTOGRAM */ + { 27635, 0x00008025 }, /* GL_PROXY_HISTOGRAM_EXT */ + { 27658, 0x000080D5 }, /* GL_PROXY_POST_COLOR_MATRIX_COLOR_TABLE */ + { 27697, 0x000080D4 }, /* GL_PROXY_POST_CONVOLUTION_COLOR_TABLE */ + { 27735, 0x00008063 }, /* GL_PROXY_TEXTURE_1D */ + { 27755, 0x00008C19 }, /* GL_PROXY_TEXTURE_1D_ARRAY_EXT */ + { 27785, 0x00008063 }, /* GL_PROXY_TEXTURE_1D_EXT */ + { 27809, 0x00008064 }, /* GL_PROXY_TEXTURE_2D */ + { 27829, 0x00008C1B }, /* GL_PROXY_TEXTURE_2D_ARRAY_EXT */ + { 27859, 0x00008064 }, /* GL_PROXY_TEXTURE_2D_EXT */ + { 27883, 0x00008070 }, /* GL_PROXY_TEXTURE_3D */ + { 27903, 0x000080BD }, /* GL_PROXY_TEXTURE_COLOR_TABLE_SGI */ + { 27936, 0x0000851B }, /* GL_PROXY_TEXTURE_CUBE_MAP */ + { 27962, 0x0000851B }, /* GL_PROXY_TEXTURE_CUBE_MAP_ARB */ + { 27992, 0x000084F7 }, /* GL_PROXY_TEXTURE_RECTANGLE_ARB */ + { 28023, 0x000084F7 }, /* GL_PROXY_TEXTURE_RECTANGLE_NV */ + { 28053, 0x00002003 }, /* GL_Q */ + { 28058, 0x00001209 }, /* GL_QUADRATIC_ATTENUATION */ + { 28083, 0x00000007 }, /* GL_QUADS */ + { 28092, 0x00008E4C }, /* GL_QUADS_FOLLOW_PROVOKING_VERTEX_CONVENTION */ + { 28136, 0x00008E4C }, /* GL_QUADS_FOLLOW_PROVOKING_VERTEX_CONVENTION_EXT */ + { 28184, 0x00008614 }, /* GL_QUAD_MESH_SUN */ + { 28201, 0x00000008 }, /* GL_QUAD_STRIP */ + { 28215, 0x00008864 }, /* GL_QUERY_COUNTER_BITS */ + { 28237, 0x00008864 }, /* GL_QUERY_COUNTER_BITS_ARB */ + { 28263, 0x00008866 }, /* GL_QUERY_RESULT */ + { 28279, 0x00008866 }, /* GL_QUERY_RESULT_ARB */ + { 28299, 0x00008867 }, /* GL_QUERY_RESULT_AVAILABLE */ + { 28325, 0x00008867 }, /* GL_QUERY_RESULT_AVAILABLE_ARB */ + { 28355, 0x00002002 }, /* GL_R */ + { 28360, 0x00002A10 }, /* GL_R3_G3_B2 */ + { 28372, 0x00019262 }, /* GL_RASTER_POSITION_UNCLIPPED_IBM */ + { 28405, 0x00000C02 }, /* GL_READ_BUFFER */ + { 28420, 0x00008CA8 }, /* GL_READ_FRAMEBUFFER */ + { 28440, 0x00008CAA }, /* GL_READ_FRAMEBUFFER_BINDING_EXT */ + { 28472, 0x00008CA8 }, /* GL_READ_FRAMEBUFFER_EXT */ + { 28496, 0x000088B8 }, /* GL_READ_ONLY */ + { 28509, 0x000088B8 }, /* GL_READ_ONLY_ARB */ + { 28526, 0x000088BA }, /* GL_READ_WRITE */ + { 28540, 0x000088BA }, /* GL_READ_WRITE_ARB */ + { 28558, 0x00001903 }, /* GL_RED */ + { 28565, 0x00008016 }, /* GL_REDUCE */ + { 28575, 0x00008016 }, /* GL_REDUCE_EXT */ + { 28589, 0x00000D15 }, /* GL_RED_BIAS */ + { 28601, 0x00000D52 }, /* GL_RED_BITS */ + { 28613, 0x00000D14 }, /* GL_RED_SCALE */ + { 28626, 0x00008512 }, /* GL_REFLECTION_MAP */ + { 28644, 0x00008512 }, /* GL_REFLECTION_MAP_ARB */ + { 28666, 0x00008512 }, /* GL_REFLECTION_MAP_NV */ + { 28687, 0x00001C00 }, /* GL_RENDER */ + { 28697, 0x00008D41 }, /* GL_RENDERBUFFER */ + { 28713, 0x00008D53 }, /* GL_RENDERBUFFER_ALPHA_SIZE */ + { 28740, 0x00008CA7 }, /* GL_RENDERBUFFER_BINDING_EXT */ + { 28768, 0x00008D52 }, /* GL_RENDERBUFFER_BLUE_SIZE */ + { 28794, 0x00008D54 }, /* GL_RENDERBUFFER_DEPTH_SIZE */ + { 28821, 0x00008D41 }, /* GL_RENDERBUFFER_EXT */ + { 28841, 0x00008D51 }, /* GL_RENDERBUFFER_GREEN_SIZE */ + { 28868, 0x00008D43 }, /* GL_RENDERBUFFER_HEIGHT */ + { 28891, 0x00008D43 }, /* GL_RENDERBUFFER_HEIGHT_EXT */ + { 28918, 0x00008D44 }, /* GL_RENDERBUFFER_INTERNAL_FORMAT */ + { 28950, 0x00008D44 }, /* GL_RENDERBUFFER_INTERNAL_FORMAT_EXT */ + { 28986, 0x00008D50 }, /* GL_RENDERBUFFER_RED_SIZE */ + { 29011, 0x00008CAB }, /* GL_RENDERBUFFER_SAMPLES */ + { 29035, 0x00008D55 }, /* GL_RENDERBUFFER_STENCIL_SIZE */ + { 29064, 0x00008D42 }, /* GL_RENDERBUFFER_WIDTH */ + { 29086, 0x00008D42 }, /* GL_RENDERBUFFER_WIDTH_EXT */ + { 29112, 0x00001F01 }, /* GL_RENDERER */ + { 29124, 0x00000C40 }, /* GL_RENDER_MODE */ + { 29139, 0x00002901 }, /* GL_REPEAT */ + { 29149, 0x00001E01 }, /* GL_REPLACE */ + { 29160, 0x00008062 }, /* GL_REPLACE_EXT */ + { 29175, 0x00008153 }, /* GL_REPLICATE_BORDER_HP */ + { 29198, 0x0000803A }, /* GL_RESCALE_NORMAL */ + { 29216, 0x0000803A }, /* GL_RESCALE_NORMAL_EXT */ + { 29238, 0x00000102 }, /* GL_RETURN */ + { 29248, 0x00001907 }, /* GL_RGB */ + { 29255, 0x00008052 }, /* GL_RGB10 */ + { 29264, 0x00008059 }, /* GL_RGB10_A2 */ + { 29276, 0x00008059 }, /* GL_RGB10_A2_EXT */ + { 29292, 0x00008052 }, /* GL_RGB10_EXT */ + { 29305, 0x00008053 }, /* GL_RGB12 */ + { 29314, 0x00008053 }, /* GL_RGB12_EXT */ + { 29327, 0x00008054 }, /* GL_RGB16 */ + { 29336, 0x00008054 }, /* GL_RGB16_EXT */ + { 29349, 0x0000804E }, /* GL_RGB2_EXT */ + { 29361, 0x0000804F }, /* GL_RGB4 */ + { 29369, 0x0000804F }, /* GL_RGB4_EXT */ + { 29381, 0x000083A1 }, /* GL_RGB4_S3TC */ + { 29394, 0x00008050 }, /* GL_RGB5 */ + { 29402, 0x00008057 }, /* GL_RGB5_A1 */ + { 29413, 0x00008057 }, /* GL_RGB5_A1_EXT */ + { 29428, 0x00008050 }, /* GL_RGB5_EXT */ + { 29440, 0x00008051 }, /* GL_RGB8 */ + { 29448, 0x00008051 }, /* GL_RGB8_EXT */ + { 29460, 0x00001908 }, /* GL_RGBA */ + { 29468, 0x0000805A }, /* GL_RGBA12 */ + { 29478, 0x0000805A }, /* GL_RGBA12_EXT */ + { 29492, 0x0000805B }, /* GL_RGBA16 */ + { 29502, 0x0000805B }, /* GL_RGBA16_EXT */ + { 29516, 0x00008055 }, /* GL_RGBA2 */ + { 29525, 0x00008055 }, /* GL_RGBA2_EXT */ + { 29538, 0x00008056 }, /* GL_RGBA4 */ + { 29547, 0x000083A5 }, /* GL_RGBA4_DXT5_S3TC */ + { 29566, 0x00008056 }, /* GL_RGBA4_EXT */ + { 29579, 0x000083A3 }, /* GL_RGBA4_S3TC */ + { 29593, 0x00008058 }, /* GL_RGBA8 */ + { 29602, 0x00008058 }, /* GL_RGBA8_EXT */ + { 29615, 0x00008F97 }, /* GL_RGBA8_SNORM */ + { 29630, 0x000083A4 }, /* GL_RGBA_DXT5_S3TC */ + { 29648, 0x00000C31 }, /* GL_RGBA_MODE */ + { 29661, 0x000083A2 }, /* GL_RGBA_S3TC */ + { 29674, 0x00008F93 }, /* GL_RGBA_SNORM */ + { 29688, 0x000083A0 }, /* GL_RGB_S3TC */ + { 29700, 0x00008573 }, /* GL_RGB_SCALE */ + { 29713, 0x00008573 }, /* GL_RGB_SCALE_ARB */ + { 29730, 0x00008573 }, /* GL_RGB_SCALE_EXT */ + { 29747, 0x00000407 }, /* GL_RIGHT */ + { 29756, 0x00002000 }, /* GL_S */ + { 29761, 0x00008B5D }, /* GL_SAMPLER_1D */ + { 29775, 0x00008B61 }, /* GL_SAMPLER_1D_SHADOW */ + { 29796, 0x00008B5E }, /* GL_SAMPLER_2D */ + { 29810, 0x00008B62 }, /* GL_SAMPLER_2D_SHADOW */ + { 29831, 0x00008B5F }, /* GL_SAMPLER_3D */ + { 29845, 0x00008B60 }, /* GL_SAMPLER_CUBE */ + { 29861, 0x000080A9 }, /* GL_SAMPLES */ + { 29872, 0x000086B4 }, /* GL_SAMPLES_3DFX */ + { 29888, 0x000080A9 }, /* GL_SAMPLES_ARB */ + { 29903, 0x00008914 }, /* GL_SAMPLES_PASSED */ + { 29921, 0x00008914 }, /* GL_SAMPLES_PASSED_ARB */ + { 29943, 0x0000809E }, /* GL_SAMPLE_ALPHA_TO_COVERAGE */ + { 29971, 0x0000809E }, /* GL_SAMPLE_ALPHA_TO_COVERAGE_ARB */ + { 30003, 0x0000809F }, /* GL_SAMPLE_ALPHA_TO_ONE */ + { 30026, 0x0000809F }, /* GL_SAMPLE_ALPHA_TO_ONE_ARB */ + { 30053, 0x000080A8 }, /* GL_SAMPLE_BUFFERS */ + { 30071, 0x000086B3 }, /* GL_SAMPLE_BUFFERS_3DFX */ + { 30094, 0x000080A8 }, /* GL_SAMPLE_BUFFERS_ARB */ + { 30116, 0x000080A0 }, /* GL_SAMPLE_COVERAGE */ + { 30135, 0x000080A0 }, /* GL_SAMPLE_COVERAGE_ARB */ + { 30158, 0x000080AB }, /* GL_SAMPLE_COVERAGE_INVERT */ + { 30184, 0x000080AB }, /* GL_SAMPLE_COVERAGE_INVERT_ARB */ + { 30214, 0x000080AA }, /* GL_SAMPLE_COVERAGE_VALUE */ + { 30239, 0x000080AA }, /* GL_SAMPLE_COVERAGE_VALUE_ARB */ + { 30268, 0x00080000 }, /* GL_SCISSOR_BIT */ + { 30283, 0x00000C10 }, /* GL_SCISSOR_BOX */ + { 30298, 0x00000C11 }, /* GL_SCISSOR_TEST */ + { 30314, 0x0000845E }, /* GL_SECONDARY_COLOR_ARRAY */ + { 30339, 0x0000889C }, /* GL_SECONDARY_COLOR_ARRAY_BUFFER_BINDING */ + { 30379, 0x0000889C }, /* GL_SECONDARY_COLOR_ARRAY_BUFFER_BINDING_ARB */ + { 30423, 0x0000845D }, /* GL_SECONDARY_COLOR_ARRAY_POINTER */ + { 30456, 0x0000845A }, /* GL_SECONDARY_COLOR_ARRAY_SIZE */ + { 30486, 0x0000845C }, /* GL_SECONDARY_COLOR_ARRAY_STRIDE */ + { 30518, 0x0000845B }, /* GL_SECONDARY_COLOR_ARRAY_TYPE */ + { 30548, 0x00001C02 }, /* GL_SELECT */ + { 30558, 0x00000DF3 }, /* GL_SELECTION_BUFFER_POINTER */ + { 30586, 0x00000DF4 }, /* GL_SELECTION_BUFFER_SIZE */ + { 30611, 0x00008012 }, /* GL_SEPARABLE_2D */ + { 30627, 0x000081FA }, /* GL_SEPARATE_SPECULAR_COLOR */ + { 30654, 0x000081FA }, /* GL_SEPARATE_SPECULAR_COLOR_EXT */ + { 30685, 0x0000150F }, /* GL_SET */ + { 30692, 0x00008B48 }, /* GL_SHADER_OBJECT_ARB */ + { 30713, 0x00008B88 }, /* GL_SHADER_SOURCE_LENGTH */ + { 30737, 0x00008B4F }, /* GL_SHADER_TYPE */ + { 30752, 0x00000B54 }, /* GL_SHADE_MODEL */ + { 30767, 0x00008B8C }, /* GL_SHADING_LANGUAGE_VERSION */ + { 30795, 0x000080BF }, /* GL_SHADOW_AMBIENT_SGIX */ + { 30818, 0x000081FB }, /* GL_SHARED_TEXTURE_PALETTE_EXT */ + { 30848, 0x00001601 }, /* GL_SHININESS */ + { 30861, 0x00001402 }, /* GL_SHORT */ + { 30870, 0x00009119 }, /* GL_SIGNALED */ + { 30882, 0x00008F9C }, /* GL_SIGNED_NORMALIZED */ + { 30903, 0x000081F9 }, /* GL_SINGLE_COLOR */ + { 30919, 0x000081F9 }, /* GL_SINGLE_COLOR_EXT */ + { 30939, 0x000085CC }, /* GL_SLICE_ACCUM_SUN */ + { 30958, 0x00008C46 }, /* GL_SLUMINANCE */ + { 30972, 0x00008C47 }, /* GL_SLUMINANCE8 */ + { 30987, 0x00008C45 }, /* GL_SLUMINANCE8_ALPHA8 */ + { 31009, 0x00008C44 }, /* GL_SLUMINANCE_ALPHA */ + { 31029, 0x00001D01 }, /* GL_SMOOTH */ + { 31039, 0x00000B23 }, /* GL_SMOOTH_LINE_WIDTH_GRANULARITY */ + { 31072, 0x00000B22 }, /* GL_SMOOTH_LINE_WIDTH_RANGE */ + { 31099, 0x00000B13 }, /* GL_SMOOTH_POINT_SIZE_GRANULARITY */ + { 31132, 0x00000B12 }, /* GL_SMOOTH_POINT_SIZE_RANGE */ + { 31159, 0x00008588 }, /* GL_SOURCE0_ALPHA */ + { 31176, 0x00008588 }, /* GL_SOURCE0_ALPHA_ARB */ + { 31197, 0x00008588 }, /* GL_SOURCE0_ALPHA_EXT */ + { 31218, 0x00008580 }, /* GL_SOURCE0_RGB */ + { 31233, 0x00008580 }, /* GL_SOURCE0_RGB_ARB */ + { 31252, 0x00008580 }, /* GL_SOURCE0_RGB_EXT */ + { 31271, 0x00008589 }, /* GL_SOURCE1_ALPHA */ + { 31288, 0x00008589 }, /* GL_SOURCE1_ALPHA_ARB */ + { 31309, 0x00008589 }, /* GL_SOURCE1_ALPHA_EXT */ + { 31330, 0x00008581 }, /* GL_SOURCE1_RGB */ + { 31345, 0x00008581 }, /* GL_SOURCE1_RGB_ARB */ + { 31364, 0x00008581 }, /* GL_SOURCE1_RGB_EXT */ + { 31383, 0x0000858A }, /* GL_SOURCE2_ALPHA */ + { 31400, 0x0000858A }, /* GL_SOURCE2_ALPHA_ARB */ + { 31421, 0x0000858A }, /* GL_SOURCE2_ALPHA_EXT */ + { 31442, 0x00008582 }, /* GL_SOURCE2_RGB */ + { 31457, 0x00008582 }, /* GL_SOURCE2_RGB_ARB */ + { 31476, 0x00008582 }, /* GL_SOURCE2_RGB_EXT */ + { 31495, 0x0000858B }, /* GL_SOURCE3_ALPHA_NV */ + { 31515, 0x00008583 }, /* GL_SOURCE3_RGB_NV */ + { 31533, 0x00001202 }, /* GL_SPECULAR */ + { 31545, 0x00002402 }, /* GL_SPHERE_MAP */ + { 31559, 0x00001206 }, /* GL_SPOT_CUTOFF */ + { 31574, 0x00001204 }, /* GL_SPOT_DIRECTION */ + { 31592, 0x00001205 }, /* GL_SPOT_EXPONENT */ + { 31609, 0x00008588 }, /* GL_SRC0_ALPHA */ + { 31623, 0x00008580 }, /* GL_SRC0_RGB */ + { 31635, 0x00008589 }, /* GL_SRC1_ALPHA */ + { 31649, 0x00008581 }, /* GL_SRC1_RGB */ + { 31661, 0x0000858A }, /* GL_SRC2_ALPHA */ + { 31675, 0x00008582 }, /* GL_SRC2_RGB */ + { 31687, 0x00000302 }, /* GL_SRC_ALPHA */ + { 31700, 0x00000308 }, /* GL_SRC_ALPHA_SATURATE */ + { 31722, 0x00000300 }, /* GL_SRC_COLOR */ + { 31735, 0x00008C40 }, /* GL_SRGB */ + { 31743, 0x00008C41 }, /* GL_SRGB8 */ + { 31752, 0x00008C43 }, /* GL_SRGB8_ALPHA8 */ + { 31768, 0x00008C42 }, /* GL_SRGB_ALPHA */ + { 31782, 0x00000503 }, /* GL_STACK_OVERFLOW */ + { 31800, 0x00000504 }, /* GL_STACK_UNDERFLOW */ + { 31819, 0x000088E6 }, /* GL_STATIC_COPY */ + { 31834, 0x000088E6 }, /* GL_STATIC_COPY_ARB */ + { 31853, 0x000088E4 }, /* GL_STATIC_DRAW */ + { 31868, 0x000088E4 }, /* GL_STATIC_DRAW_ARB */ + { 31887, 0x000088E5 }, /* GL_STATIC_READ */ + { 31902, 0x000088E5 }, /* GL_STATIC_READ_ARB */ + { 31921, 0x00001802 }, /* GL_STENCIL */ + { 31932, 0x00008D20 }, /* GL_STENCIL_ATTACHMENT */ + { 31954, 0x00008D20 }, /* GL_STENCIL_ATTACHMENT_EXT */ + { 31980, 0x00008801 }, /* GL_STENCIL_BACK_FAIL */ + { 32001, 0x00008801 }, /* GL_STENCIL_BACK_FAIL_ATI */ + { 32026, 0x00008800 }, /* GL_STENCIL_BACK_FUNC */ + { 32047, 0x00008800 }, /* GL_STENCIL_BACK_FUNC_ATI */ + { 32072, 0x00008802 }, /* GL_STENCIL_BACK_PASS_DEPTH_FAIL */ + { 32104, 0x00008802 }, /* GL_STENCIL_BACK_PASS_DEPTH_FAIL_ATI */ + { 32140, 0x00008803 }, /* GL_STENCIL_BACK_PASS_DEPTH_PASS */ + { 32172, 0x00008803 }, /* GL_STENCIL_BACK_PASS_DEPTH_PASS_ATI */ + { 32208, 0x00008CA3 }, /* GL_STENCIL_BACK_REF */ + { 32228, 0x00008CA4 }, /* GL_STENCIL_BACK_VALUE_MASK */ + { 32255, 0x00008CA5 }, /* GL_STENCIL_BACK_WRITEMASK */ + { 32281, 0x00000D57 }, /* GL_STENCIL_BITS */ + { 32297, 0x00000400 }, /* GL_STENCIL_BUFFER_BIT */ + { 32319, 0x00000B91 }, /* GL_STENCIL_CLEAR_VALUE */ + { 32342, 0x00000B94 }, /* GL_STENCIL_FAIL */ + { 32358, 0x00000B92 }, /* GL_STENCIL_FUNC */ + { 32374, 0x00001901 }, /* GL_STENCIL_INDEX */ + { 32391, 0x00008D49 }, /* GL_STENCIL_INDEX16_EXT */ + { 32414, 0x00008D46 }, /* GL_STENCIL_INDEX1_EXT */ + { 32436, 0x00008D47 }, /* GL_STENCIL_INDEX4_EXT */ + { 32458, 0x00008D48 }, /* GL_STENCIL_INDEX8_EXT */ + { 32480, 0x00008D45 }, /* GL_STENCIL_INDEX_EXT */ + { 32501, 0x00000B95 }, /* GL_STENCIL_PASS_DEPTH_FAIL */ + { 32528, 0x00000B96 }, /* GL_STENCIL_PASS_DEPTH_PASS */ + { 32555, 0x00000B97 }, /* GL_STENCIL_REF */ + { 32570, 0x00000B90 }, /* GL_STENCIL_TEST */ + { 32586, 0x00008910 }, /* GL_STENCIL_TEST_TWO_SIDE_EXT */ + { 32615, 0x00000B93 }, /* GL_STENCIL_VALUE_MASK */ + { 32637, 0x00000B98 }, /* GL_STENCIL_WRITEMASK */ + { 32658, 0x00000C33 }, /* GL_STEREO */ + { 32668, 0x000085BE }, /* GL_STORAGE_CACHED_APPLE */ + { 32692, 0x000085BD }, /* GL_STORAGE_PRIVATE_APPLE */ + { 32717, 0x000085BF }, /* GL_STORAGE_SHARED_APPLE */ + { 32741, 0x000088E2 }, /* GL_STREAM_COPY */ + { 32756, 0x000088E2 }, /* GL_STREAM_COPY_ARB */ + { 32775, 0x000088E0 }, /* GL_STREAM_DRAW */ + { 32790, 0x000088E0 }, /* GL_STREAM_DRAW_ARB */ + { 32809, 0x000088E1 }, /* GL_STREAM_READ */ + { 32824, 0x000088E1 }, /* GL_STREAM_READ_ARB */ + { 32843, 0x00000D50 }, /* GL_SUBPIXEL_BITS */ + { 32860, 0x000084E7 }, /* GL_SUBTRACT */ + { 32872, 0x000084E7 }, /* GL_SUBTRACT_ARB */ + { 32888, 0x00009113 }, /* GL_SYNC_CONDITION */ + { 32906, 0x00009116 }, /* GL_SYNC_FENCE */ + { 32920, 0x00009115 }, /* GL_SYNC_FLAGS */ + { 32934, 0x00000001 }, /* GL_SYNC_FLUSH_COMMANDS_BIT */ + { 32961, 0x00009117 }, /* GL_SYNC_GPU_COMMANDS_COMPLETE */ + { 32991, 0x00009114 }, /* GL_SYNC_STATUS */ + { 33006, 0x00002001 }, /* GL_T */ + { 33011, 0x00002A2A }, /* GL_T2F_C3F_V3F */ + { 33026, 0x00002A2C }, /* GL_T2F_C4F_N3F_V3F */ + { 33045, 0x00002A29 }, /* GL_T2F_C4UB_V3F */ + { 33061, 0x00002A2B }, /* GL_T2F_N3F_V3F */ + { 33076, 0x00002A27 }, /* GL_T2F_V3F */ + { 33087, 0x00002A2D }, /* GL_T4F_C4F_N3F_V4F */ + { 33106, 0x00002A28 }, /* GL_T4F_V4F */ + { 33117, 0x00008031 }, /* GL_TABLE_TOO_LARGE_EXT */ + { 33140, 0x00001702 }, /* GL_TEXTURE */ + { 33151, 0x000084C0 }, /* GL_TEXTURE0 */ + { 33163, 0x000084C0 }, /* GL_TEXTURE0_ARB */ + { 33179, 0x000084C1 }, /* GL_TEXTURE1 */ + { 33191, 0x000084CA }, /* GL_TEXTURE10 */ + { 33204, 0x000084CA }, /* GL_TEXTURE10_ARB */ + { 33221, 0x000084CB }, /* GL_TEXTURE11 */ + { 33234, 0x000084CB }, /* GL_TEXTURE11_ARB */ + { 33251, 0x000084CC }, /* GL_TEXTURE12 */ + { 33264, 0x000084CC }, /* GL_TEXTURE12_ARB */ + { 33281, 0x000084CD }, /* GL_TEXTURE13 */ + { 33294, 0x000084CD }, /* GL_TEXTURE13_ARB */ + { 33311, 0x000084CE }, /* GL_TEXTURE14 */ + { 33324, 0x000084CE }, /* GL_TEXTURE14_ARB */ + { 33341, 0x000084CF }, /* GL_TEXTURE15 */ + { 33354, 0x000084CF }, /* GL_TEXTURE15_ARB */ + { 33371, 0x000084D0 }, /* GL_TEXTURE16 */ + { 33384, 0x000084D0 }, /* GL_TEXTURE16_ARB */ + { 33401, 0x000084D1 }, /* GL_TEXTURE17 */ + { 33414, 0x000084D1 }, /* GL_TEXTURE17_ARB */ + { 33431, 0x000084D2 }, /* GL_TEXTURE18 */ + { 33444, 0x000084D2 }, /* GL_TEXTURE18_ARB */ + { 33461, 0x000084D3 }, /* GL_TEXTURE19 */ + { 33474, 0x000084D3 }, /* GL_TEXTURE19_ARB */ + { 33491, 0x000084C1 }, /* GL_TEXTURE1_ARB */ + { 33507, 0x000084C2 }, /* GL_TEXTURE2 */ + { 33519, 0x000084D4 }, /* GL_TEXTURE20 */ + { 33532, 0x000084D4 }, /* GL_TEXTURE20_ARB */ + { 33549, 0x000084D5 }, /* GL_TEXTURE21 */ + { 33562, 0x000084D5 }, /* GL_TEXTURE21_ARB */ + { 33579, 0x000084D6 }, /* GL_TEXTURE22 */ + { 33592, 0x000084D6 }, /* GL_TEXTURE22_ARB */ + { 33609, 0x000084D7 }, /* GL_TEXTURE23 */ + { 33622, 0x000084D7 }, /* GL_TEXTURE23_ARB */ + { 33639, 0x000084D8 }, /* GL_TEXTURE24 */ + { 33652, 0x000084D8 }, /* GL_TEXTURE24_ARB */ + { 33669, 0x000084D9 }, /* GL_TEXTURE25 */ + { 33682, 0x000084D9 }, /* GL_TEXTURE25_ARB */ + { 33699, 0x000084DA }, /* GL_TEXTURE26 */ + { 33712, 0x000084DA }, /* GL_TEXTURE26_ARB */ + { 33729, 0x000084DB }, /* GL_TEXTURE27 */ + { 33742, 0x000084DB }, /* GL_TEXTURE27_ARB */ + { 33759, 0x000084DC }, /* GL_TEXTURE28 */ + { 33772, 0x000084DC }, /* GL_TEXTURE28_ARB */ + { 33789, 0x000084DD }, /* GL_TEXTURE29 */ + { 33802, 0x000084DD }, /* GL_TEXTURE29_ARB */ + { 33819, 0x000084C2 }, /* GL_TEXTURE2_ARB */ + { 33835, 0x000084C3 }, /* GL_TEXTURE3 */ + { 33847, 0x000084DE }, /* GL_TEXTURE30 */ + { 33860, 0x000084DE }, /* GL_TEXTURE30_ARB */ + { 33877, 0x000084DF }, /* GL_TEXTURE31 */ + { 33890, 0x000084DF }, /* GL_TEXTURE31_ARB */ + { 33907, 0x000084C3 }, /* GL_TEXTURE3_ARB */ + { 33923, 0x000084C4 }, /* GL_TEXTURE4 */ + { 33935, 0x000084C4 }, /* GL_TEXTURE4_ARB */ + { 33951, 0x000084C5 }, /* GL_TEXTURE5 */ + { 33963, 0x000084C5 }, /* GL_TEXTURE5_ARB */ + { 33979, 0x000084C6 }, /* GL_TEXTURE6 */ + { 33991, 0x000084C6 }, /* GL_TEXTURE6_ARB */ + { 34007, 0x000084C7 }, /* GL_TEXTURE7 */ + { 34019, 0x000084C7 }, /* GL_TEXTURE7_ARB */ + { 34035, 0x000084C8 }, /* GL_TEXTURE8 */ + { 34047, 0x000084C8 }, /* GL_TEXTURE8_ARB */ + { 34063, 0x000084C9 }, /* GL_TEXTURE9 */ + { 34075, 0x000084C9 }, /* GL_TEXTURE9_ARB */ + { 34091, 0x00000DE0 }, /* GL_TEXTURE_1D */ + { 34105, 0x00008C18 }, /* GL_TEXTURE_1D_ARRAY_EXT */ + { 34129, 0x00000DE1 }, /* GL_TEXTURE_2D */ + { 34143, 0x00008C1A }, /* GL_TEXTURE_2D_ARRAY_EXT */ + { 34167, 0x0000806F }, /* GL_TEXTURE_3D */ + { 34181, 0x0000805F }, /* GL_TEXTURE_ALPHA_SIZE */ + { 34203, 0x0000805F }, /* GL_TEXTURE_ALPHA_SIZE_EXT */ + { 34229, 0x0000813C }, /* GL_TEXTURE_BASE_LEVEL */ + { 34251, 0x00008068 }, /* GL_TEXTURE_BINDING_1D */ + { 34273, 0x00008C1C }, /* GL_TEXTURE_BINDING_1D_ARRAY_EXT */ + { 34305, 0x00008069 }, /* GL_TEXTURE_BINDING_2D */ + { 34327, 0x00008C1D }, /* GL_TEXTURE_BINDING_2D_ARRAY_EXT */ + { 34359, 0x0000806A }, /* GL_TEXTURE_BINDING_3D */ + { 34381, 0x00008514 }, /* GL_TEXTURE_BINDING_CUBE_MAP */ + { 34409, 0x00008514 }, /* GL_TEXTURE_BINDING_CUBE_MAP_ARB */ + { 34441, 0x000084F6 }, /* GL_TEXTURE_BINDING_RECTANGLE_ARB */ + { 34474, 0x000084F6 }, /* GL_TEXTURE_BINDING_RECTANGLE_NV */ + { 34506, 0x00040000 }, /* GL_TEXTURE_BIT */ + { 34521, 0x0000805E }, /* GL_TEXTURE_BLUE_SIZE */ + { 34542, 0x0000805E }, /* GL_TEXTURE_BLUE_SIZE_EXT */ + { 34567, 0x00001005 }, /* GL_TEXTURE_BORDER */ + { 34585, 0x00001004 }, /* GL_TEXTURE_BORDER_COLOR */ + { 34609, 0x00008171 }, /* GL_TEXTURE_CLIPMAP_CENTER_SGIX */ + { 34640, 0x00008176 }, /* GL_TEXTURE_CLIPMAP_DEPTH_SGIX */ + { 34670, 0x00008172 }, /* GL_TEXTURE_CLIPMAP_FRAME_SGIX */ + { 34700, 0x00008175 }, /* GL_TEXTURE_CLIPMAP_LOD_OFFSET_SGIX */ + { 34735, 0x00008173 }, /* GL_TEXTURE_CLIPMAP_OFFSET_SGIX */ + { 34766, 0x00008174 }, /* GL_TEXTURE_CLIPMAP_VIRTUAL_DEPTH_SGIX */ + { 34804, 0x000080BC }, /* GL_TEXTURE_COLOR_TABLE_SGI */ + { 34831, 0x000081EF }, /* GL_TEXTURE_COLOR_WRITEMASK_SGIS */ + { 34863, 0x000080BF }, /* GL_TEXTURE_COMPARE_FAIL_VALUE_ARB */ + { 34897, 0x0000884D }, /* GL_TEXTURE_COMPARE_FUNC */ + { 34921, 0x0000884D }, /* GL_TEXTURE_COMPARE_FUNC_ARB */ + { 34949, 0x0000884C }, /* GL_TEXTURE_COMPARE_MODE */ + { 34973, 0x0000884C }, /* GL_TEXTURE_COMPARE_MODE_ARB */ + { 35001, 0x0000819B }, /* GL_TEXTURE_COMPARE_OPERATOR_SGIX */ + { 35034, 0x0000819A }, /* GL_TEXTURE_COMPARE_SGIX */ + { 35058, 0x00001003 }, /* GL_TEXTURE_COMPONENTS */ + { 35080, 0x000086A1 }, /* GL_TEXTURE_COMPRESSED */ + { 35102, 0x000086A1 }, /* GL_TEXTURE_COMPRESSED_ARB */ + { 35128, 0x000086A3 }, /* GL_TEXTURE_COMPRESSED_FORMATS_ARB */ + { 35162, 0x000086A0 }, /* GL_TEXTURE_COMPRESSED_IMAGE_SIZE */ + { 35195, 0x000086A0 }, /* GL_TEXTURE_COMPRESSED_IMAGE_SIZE_ARB */ + { 35232, 0x000084EF }, /* GL_TEXTURE_COMPRESSION_HINT */ + { 35260, 0x000084EF }, /* GL_TEXTURE_COMPRESSION_HINT_ARB */ + { 35292, 0x00008078 }, /* GL_TEXTURE_COORD_ARRAY */ + { 35315, 0x0000889A }, /* GL_TEXTURE_COORD_ARRAY_BUFFER_BINDING */ + { 35353, 0x0000889A }, /* GL_TEXTURE_COORD_ARRAY_BUFFER_BINDING_ARB */ + { 35395, 0x00008092 }, /* GL_TEXTURE_COORD_ARRAY_POINTER */ + { 35426, 0x00008088 }, /* GL_TEXTURE_COORD_ARRAY_SIZE */ + { 35454, 0x0000808A }, /* GL_TEXTURE_COORD_ARRAY_STRIDE */ + { 35484, 0x00008089 }, /* GL_TEXTURE_COORD_ARRAY_TYPE */ + { 35512, 0x00008513 }, /* GL_TEXTURE_CUBE_MAP */ + { 35532, 0x00008513 }, /* GL_TEXTURE_CUBE_MAP_ARB */ + { 35556, 0x00008516 }, /* GL_TEXTURE_CUBE_MAP_NEGATIVE_X */ + { 35587, 0x00008516 }, /* GL_TEXTURE_CUBE_MAP_NEGATIVE_X_ARB */ + { 35622, 0x00008518 }, /* GL_TEXTURE_CUBE_MAP_NEGATIVE_Y */ + { 35653, 0x00008518 }, /* GL_TEXTURE_CUBE_MAP_NEGATIVE_Y_ARB */ + { 35688, 0x0000851A }, /* GL_TEXTURE_CUBE_MAP_NEGATIVE_Z */ + { 35719, 0x0000851A }, /* GL_TEXTURE_CUBE_MAP_NEGATIVE_Z_ARB */ + { 35754, 0x00008515 }, /* GL_TEXTURE_CUBE_MAP_POSITIVE_X */ + { 35785, 0x00008515 }, /* GL_TEXTURE_CUBE_MAP_POSITIVE_X_ARB */ + { 35820, 0x00008517 }, /* GL_TEXTURE_CUBE_MAP_POSITIVE_Y */ + { 35851, 0x00008517 }, /* GL_TEXTURE_CUBE_MAP_POSITIVE_Y_ARB */ + { 35886, 0x00008519 }, /* GL_TEXTURE_CUBE_MAP_POSITIVE_Z */ + { 35917, 0x00008519 }, /* GL_TEXTURE_CUBE_MAP_POSITIVE_Z_ARB */ + { 35952, 0x000088F4 }, /* GL_TEXTURE_CUBE_MAP_SEAMLESS */ + { 35981, 0x00008071 }, /* GL_TEXTURE_DEPTH */ + { 35998, 0x0000884A }, /* GL_TEXTURE_DEPTH_SIZE */ + { 36020, 0x0000884A }, /* GL_TEXTURE_DEPTH_SIZE_ARB */ + { 36046, 0x00002300 }, /* GL_TEXTURE_ENV */ + { 36061, 0x00002201 }, /* GL_TEXTURE_ENV_COLOR */ + { 36082, 0x00002200 }, /* GL_TEXTURE_ENV_MODE */ + { 36102, 0x00008500 }, /* GL_TEXTURE_FILTER_CONTROL */ + { 36128, 0x00002500 }, /* GL_TEXTURE_GEN_MODE */ + { 36148, 0x00000C63 }, /* GL_TEXTURE_GEN_Q */ + { 36165, 0x00000C62 }, /* GL_TEXTURE_GEN_R */ + { 36182, 0x00000C60 }, /* GL_TEXTURE_GEN_S */ + { 36199, 0x00000C61 }, /* GL_TEXTURE_GEN_T */ + { 36216, 0x0000819D }, /* GL_TEXTURE_GEQUAL_R_SGIX */ + { 36241, 0x0000805D }, /* GL_TEXTURE_GREEN_SIZE */ + { 36263, 0x0000805D }, /* GL_TEXTURE_GREEN_SIZE_EXT */ + { 36289, 0x00001001 }, /* GL_TEXTURE_HEIGHT */ + { 36307, 0x000080ED }, /* GL_TEXTURE_INDEX_SIZE_EXT */ + { 36333, 0x00008061 }, /* GL_TEXTURE_INTENSITY_SIZE */ + { 36359, 0x00008061 }, /* GL_TEXTURE_INTENSITY_SIZE_EXT */ + { 36389, 0x00001003 }, /* GL_TEXTURE_INTERNAL_FORMAT */ + { 36416, 0x0000819C }, /* GL_TEXTURE_LEQUAL_R_SGIX */ + { 36441, 0x00008501 }, /* GL_TEXTURE_LOD_BIAS */ + { 36461, 0x00008501 }, /* GL_TEXTURE_LOD_BIAS_EXT */ + { 36485, 0x00008190 }, /* GL_TEXTURE_LOD_BIAS_R_SGIX */ + { 36512, 0x0000818E }, /* GL_TEXTURE_LOD_BIAS_S_SGIX */ + { 36539, 0x0000818F }, /* GL_TEXTURE_LOD_BIAS_T_SGIX */ + { 36566, 0x00008060 }, /* GL_TEXTURE_LUMINANCE_SIZE */ + { 36592, 0x00008060 }, /* GL_TEXTURE_LUMINANCE_SIZE_EXT */ + { 36622, 0x00002800 }, /* GL_TEXTURE_MAG_FILTER */ + { 36644, 0x00000BA8 }, /* GL_TEXTURE_MATRIX */ + { 36662, 0x000084FE }, /* GL_TEXTURE_MAX_ANISOTROPY_EXT */ + { 36692, 0x0000836B }, /* GL_TEXTURE_MAX_CLAMP_R_SGIX */ + { 36720, 0x00008369 }, /* GL_TEXTURE_MAX_CLAMP_S_SGIX */ + { 36748, 0x0000836A }, /* GL_TEXTURE_MAX_CLAMP_T_SGIX */ + { 36776, 0x0000813D }, /* GL_TEXTURE_MAX_LEVEL */ + { 36797, 0x0000813B }, /* GL_TEXTURE_MAX_LOD */ + { 36816, 0x00002801 }, /* GL_TEXTURE_MIN_FILTER */ + { 36838, 0x0000813A }, /* GL_TEXTURE_MIN_LOD */ + { 36857, 0x00008066 }, /* GL_TEXTURE_PRIORITY */ + { 36877, 0x000085B7 }, /* GL_TEXTURE_RANGE_LENGTH_APPLE */ + { 36907, 0x000085B8 }, /* GL_TEXTURE_RANGE_POINTER_APPLE */ + { 36938, 0x000084F5 }, /* GL_TEXTURE_RECTANGLE_ARB */ + { 36963, 0x000084F5 }, /* GL_TEXTURE_RECTANGLE_NV */ + { 36987, 0x0000805C }, /* GL_TEXTURE_RED_SIZE */ + { 37007, 0x0000805C }, /* GL_TEXTURE_RED_SIZE_EXT */ + { 37031, 0x00008067 }, /* GL_TEXTURE_RESIDENT */ + { 37051, 0x00000BA5 }, /* GL_TEXTURE_STACK_DEPTH */ + { 37074, 0x000088F1 }, /* GL_TEXTURE_STENCIL_SIZE */ + { 37098, 0x000085BC }, /* GL_TEXTURE_STORAGE_HINT_APPLE */ + { 37128, 0x00008065 }, /* GL_TEXTURE_TOO_LARGE_EXT */ + { 37153, 0x0000888F }, /* GL_TEXTURE_UNSIGNED_REMAP_MODE_NV */ + { 37187, 0x00001000 }, /* GL_TEXTURE_WIDTH */ + { 37204, 0x00008072 }, /* GL_TEXTURE_WRAP_R */ + { 37222, 0x00002802 }, /* GL_TEXTURE_WRAP_S */ + { 37240, 0x00002803 }, /* GL_TEXTURE_WRAP_T */ + { 37258, 0x0000911B }, /* GL_TIMEOUT_EXPIRED */ + { 37277, 0xFFFFFFFFFFFFFFFF }, /* GL_TIMEOUT_IGNORED */ + { 37296, 0x000088BF }, /* GL_TIME_ELAPSED_EXT */ + { 37316, 0x00008648 }, /* GL_TRACK_MATRIX_NV */ + { 37335, 0x00008649 }, /* GL_TRACK_MATRIX_TRANSFORM_NV */ + { 37364, 0x00001000 }, /* GL_TRANSFORM_BIT */ + { 37381, 0x000084E6 }, /* GL_TRANSPOSE_COLOR_MATRIX */ + { 37407, 0x000084E6 }, /* GL_TRANSPOSE_COLOR_MATRIX_ARB */ + { 37437, 0x000088B7 }, /* GL_TRANSPOSE_CURRENT_MATRIX_ARB */ + { 37469, 0x000084E3 }, /* GL_TRANSPOSE_MODELVIEW_MATRIX */ + { 37499, 0x000084E3 }, /* GL_TRANSPOSE_MODELVIEW_MATRIX_ARB */ + { 37533, 0x0000862C }, /* GL_TRANSPOSE_NV */ + { 37549, 0x000084E4 }, /* GL_TRANSPOSE_PROJECTION_MATRIX */ + { 37580, 0x000084E4 }, /* GL_TRANSPOSE_PROJECTION_MATRIX_ARB */ + { 37615, 0x000084E5 }, /* GL_TRANSPOSE_TEXTURE_MATRIX */ + { 37643, 0x000084E5 }, /* GL_TRANSPOSE_TEXTURE_MATRIX_ARB */ + { 37675, 0x00000004 }, /* GL_TRIANGLES */ + { 37688, 0x00000006 }, /* GL_TRIANGLE_FAN */ + { 37704, 0x00008615 }, /* GL_TRIANGLE_MESH_SUN */ + { 37725, 0x00000005 }, /* GL_TRIANGLE_STRIP */ + { 37743, 0x00000001 }, /* GL_TRUE */ + { 37751, 0x00000CF5 }, /* GL_UNPACK_ALIGNMENT */ + { 37771, 0x0000806E }, /* GL_UNPACK_IMAGE_HEIGHT */ + { 37794, 0x00000CF1 }, /* GL_UNPACK_LSB_FIRST */ + { 37814, 0x00000CF2 }, /* GL_UNPACK_ROW_LENGTH */ + { 37835, 0x0000806D }, /* GL_UNPACK_SKIP_IMAGES */ + { 37857, 0x00000CF4 }, /* GL_UNPACK_SKIP_PIXELS */ + { 37879, 0x00000CF3 }, /* GL_UNPACK_SKIP_ROWS */ + { 37899, 0x00000CF0 }, /* GL_UNPACK_SWAP_BYTES */ + { 37920, 0x00009118 }, /* GL_UNSIGNALED */ + { 37934, 0x00001401 }, /* GL_UNSIGNED_BYTE */ + { 37951, 0x00008362 }, /* GL_UNSIGNED_BYTE_2_3_3_REV */ + { 37978, 0x00008032 }, /* GL_UNSIGNED_BYTE_3_3_2 */ + { 38001, 0x00001405 }, /* GL_UNSIGNED_INT */ + { 38017, 0x00008036 }, /* GL_UNSIGNED_INT_10_10_10_2 */ + { 38044, 0x000084FA }, /* GL_UNSIGNED_INT_24_8 */ + { 38065, 0x000084FA }, /* GL_UNSIGNED_INT_24_8_NV */ + { 38089, 0x00008368 }, /* GL_UNSIGNED_INT_2_10_10_10_REV */ + { 38120, 0x00008035 }, /* GL_UNSIGNED_INT_8_8_8_8 */ + { 38144, 0x00008367 }, /* GL_UNSIGNED_INT_8_8_8_8_REV */ + { 38172, 0x00008C17 }, /* GL_UNSIGNED_NORMALIZED */ + { 38195, 0x00001403 }, /* GL_UNSIGNED_SHORT */ + { 38213, 0x00008366 }, /* GL_UNSIGNED_SHORT_1_5_5_5_REV */ + { 38243, 0x00008033 }, /* GL_UNSIGNED_SHORT_4_4_4_4 */ + { 38269, 0x00008365 }, /* GL_UNSIGNED_SHORT_4_4_4_4_REV */ + { 38299, 0x00008034 }, /* GL_UNSIGNED_SHORT_5_5_5_1 */ + { 38325, 0x00008363 }, /* GL_UNSIGNED_SHORT_5_6_5 */ + { 38349, 0x00008364 }, /* GL_UNSIGNED_SHORT_5_6_5_REV */ + { 38377, 0x000085BA }, /* GL_UNSIGNED_SHORT_8_8_APPLE */ + { 38405, 0x000085BA }, /* GL_UNSIGNED_SHORT_8_8_MESA */ + { 38432, 0x000085BB }, /* GL_UNSIGNED_SHORT_8_8_REV_APPLE */ + { 38464, 0x000085BB }, /* GL_UNSIGNED_SHORT_8_8_REV_MESA */ + { 38495, 0x00008CA2 }, /* GL_UPPER_LEFT */ + { 38509, 0x00002A20 }, /* GL_V2F */ + { 38516, 0x00002A21 }, /* GL_V3F */ + { 38523, 0x00008B83 }, /* GL_VALIDATE_STATUS */ + { 38542, 0x00001F00 }, /* GL_VENDOR */ + { 38552, 0x00001F02 }, /* GL_VERSION */ + { 38563, 0x00008074 }, /* GL_VERTEX_ARRAY */ + { 38579, 0x000085B5 }, /* GL_VERTEX_ARRAY_BINDING */ + { 38603, 0x000085B5 }, /* GL_VERTEX_ARRAY_BINDING_APPLE */ + { 38633, 0x00008896 }, /* GL_VERTEX_ARRAY_BUFFER_BINDING */ + { 38664, 0x00008896 }, /* GL_VERTEX_ARRAY_BUFFER_BINDING_ARB */ + { 38699, 0x0000808E }, /* GL_VERTEX_ARRAY_POINTER */ + { 38723, 0x0000807A }, /* GL_VERTEX_ARRAY_SIZE */ + { 38744, 0x0000807C }, /* GL_VERTEX_ARRAY_STRIDE */ + { 38767, 0x0000807B }, /* GL_VERTEX_ARRAY_TYPE */ + { 38788, 0x00008650 }, /* GL_VERTEX_ATTRIB_ARRAY0_NV */ + { 38815, 0x0000865A }, /* GL_VERTEX_ATTRIB_ARRAY10_NV */ + { 38843, 0x0000865B }, /* GL_VERTEX_ATTRIB_ARRAY11_NV */ + { 38871, 0x0000865C }, /* GL_VERTEX_ATTRIB_ARRAY12_NV */ + { 38899, 0x0000865D }, /* GL_VERTEX_ATTRIB_ARRAY13_NV */ + { 38927, 0x0000865E }, /* GL_VERTEX_ATTRIB_ARRAY14_NV */ + { 38955, 0x0000865F }, /* GL_VERTEX_ATTRIB_ARRAY15_NV */ + { 38983, 0x00008651 }, /* GL_VERTEX_ATTRIB_ARRAY1_NV */ + { 39010, 0x00008652 }, /* GL_VERTEX_ATTRIB_ARRAY2_NV */ + { 39037, 0x00008653 }, /* GL_VERTEX_ATTRIB_ARRAY3_NV */ + { 39064, 0x00008654 }, /* GL_VERTEX_ATTRIB_ARRAY4_NV */ + { 39091, 0x00008655 }, /* GL_VERTEX_ATTRIB_ARRAY5_NV */ + { 39118, 0x00008656 }, /* GL_VERTEX_ATTRIB_ARRAY6_NV */ + { 39145, 0x00008657 }, /* GL_VERTEX_ATTRIB_ARRAY7_NV */ + { 39172, 0x00008658 }, /* GL_VERTEX_ATTRIB_ARRAY8_NV */ + { 39199, 0x00008659 }, /* GL_VERTEX_ATTRIB_ARRAY9_NV */ + { 39226, 0x0000889F }, /* GL_VERTEX_ATTRIB_ARRAY_BUFFER_BINDING */ + { 39264, 0x0000889F }, /* GL_VERTEX_ATTRIB_ARRAY_BUFFER_BINDING_ARB */ + { 39306, 0x00008622 }, /* GL_VERTEX_ATTRIB_ARRAY_ENABLED */ + { 39337, 0x00008622 }, /* GL_VERTEX_ATTRIB_ARRAY_ENABLED_ARB */ + { 39372, 0x0000886A }, /* GL_VERTEX_ATTRIB_ARRAY_NORMALIZED */ + { 39406, 0x0000886A }, /* GL_VERTEX_ATTRIB_ARRAY_NORMALIZED_ARB */ + { 39444, 0x00008645 }, /* GL_VERTEX_ATTRIB_ARRAY_POINTER */ + { 39475, 0x00008645 }, /* GL_VERTEX_ATTRIB_ARRAY_POINTER_ARB */ + { 39510, 0x00008623 }, /* GL_VERTEX_ATTRIB_ARRAY_SIZE */ + { 39538, 0x00008623 }, /* GL_VERTEX_ATTRIB_ARRAY_SIZE_ARB */ + { 39570, 0x00008624 }, /* GL_VERTEX_ATTRIB_ARRAY_STRIDE */ + { 39600, 0x00008624 }, /* GL_VERTEX_ATTRIB_ARRAY_STRIDE_ARB */ + { 39634, 0x00008625 }, /* GL_VERTEX_ATTRIB_ARRAY_TYPE */ + { 39662, 0x00008625 }, /* GL_VERTEX_ATTRIB_ARRAY_TYPE_ARB */ + { 39694, 0x000086A7 }, /* GL_VERTEX_BLEND_ARB */ + { 39714, 0x00008620 }, /* GL_VERTEX_PROGRAM_ARB */ + { 39736, 0x0000864A }, /* GL_VERTEX_PROGRAM_BINDING_NV */ + { 39765, 0x00008620 }, /* GL_VERTEX_PROGRAM_NV */ + { 39786, 0x00008642 }, /* GL_VERTEX_PROGRAM_POINT_SIZE */ + { 39815, 0x00008642 }, /* GL_VERTEX_PROGRAM_POINT_SIZE_ARB */ + { 39848, 0x00008642 }, /* GL_VERTEX_PROGRAM_POINT_SIZE_NV */ + { 39880, 0x00008643 }, /* GL_VERTEX_PROGRAM_TWO_SIDE */ + { 39907, 0x00008643 }, /* GL_VERTEX_PROGRAM_TWO_SIDE_ARB */ + { 39938, 0x00008643 }, /* GL_VERTEX_PROGRAM_TWO_SIDE_NV */ + { 39968, 0x00008B31 }, /* GL_VERTEX_SHADER */ + { 39985, 0x00008B31 }, /* GL_VERTEX_SHADER_ARB */ + { 40006, 0x00008621 }, /* GL_VERTEX_STATE_PROGRAM_NV */ + { 40033, 0x00000BA2 }, /* GL_VIEWPORT */ + { 40045, 0x00000800 }, /* GL_VIEWPORT_BIT */ + { 40061, 0x0000911D }, /* GL_WAIT_FAILED */ + { 40076, 0x000086AD }, /* GL_WEIGHT_ARRAY_ARB */ + { 40096, 0x0000889E }, /* GL_WEIGHT_ARRAY_BUFFER_BINDING */ + { 40127, 0x0000889E }, /* GL_WEIGHT_ARRAY_BUFFER_BINDING_ARB */ + { 40162, 0x000086AC }, /* GL_WEIGHT_ARRAY_POINTER_ARB */ + { 40190, 0x000086AB }, /* GL_WEIGHT_ARRAY_SIZE_ARB */ + { 40215, 0x000086AA }, /* GL_WEIGHT_ARRAY_STRIDE_ARB */ + { 40242, 0x000086A9 }, /* GL_WEIGHT_ARRAY_TYPE_ARB */ + { 40267, 0x000086A6 }, /* GL_WEIGHT_SUM_UNITY_ARB */ + { 40291, 0x000081D4 }, /* GL_WRAP_BORDER_SUN */ + { 40310, 0x000088B9 }, /* GL_WRITE_ONLY */ + { 40324, 0x000088B9 }, /* GL_WRITE_ONLY_ARB */ + { 40342, 0x00001506 }, /* GL_XOR */ + { 40349, 0x000085B9 }, /* GL_YCBCR_422_APPLE */ + { 40368, 0x00008757 }, /* GL_YCBCR_MESA */ + { 40382, 0x00000000 }, /* GL_ZERO */ + { 40390, 0x00000D16 }, /* GL_ZOOM_X */ + { 40400, 0x00000D17 }, /* GL_ZOOM_Y */ }; static const unsigned reduced_enums[1347] = { 476, /* GL_FALSE */ - 692, /* GL_LINES */ - 694, /* GL_LINE_LOOP */ - 701, /* GL_LINE_STRIP */ - 1744, /* GL_TRIANGLES */ - 1747, /* GL_TRIANGLE_STRIP */ - 1745, /* GL_TRIANGLE_FAN */ - 1272, /* GL_QUADS */ - 1275, /* GL_QUAD_STRIP */ - 1159, /* GL_POLYGON */ - 1171, /* GL_POLYGON_STIPPLE_BIT */ - 1120, /* GL_PIXEL_MODE_BIT */ - 679, /* GL_LIGHTING_BIT */ - 505, /* GL_FOG_BIT */ + 694, /* GL_LINES */ + 696, /* GL_LINE_LOOP */ + 703, /* GL_LINE_STRIP */ + 1748, /* GL_TRIANGLES */ + 1751, /* GL_TRIANGLE_STRIP */ + 1749, /* GL_TRIANGLE_FAN */ + 1275, /* GL_QUADS */ + 1279, /* GL_QUAD_STRIP */ + 1161, /* GL_POLYGON */ + 1173, /* GL_POLYGON_STIPPLE_BIT */ + 1122, /* GL_PIXEL_MODE_BIT */ + 681, /* GL_LIGHTING_BIT */ + 506, /* GL_FOG_BIT */ 8, /* GL_ACCUM */ - 711, /* GL_LOAD */ - 1327, /* GL_RETURN */ - 992, /* GL_MULT */ + 713, /* GL_LOAD */ + 1331, /* GL_RETURN */ + 994, /* GL_MULT */ 23, /* GL_ADD */ - 1008, /* GL_NEVER */ - 669, /* GL_LESS */ + 1010, /* GL_NEVER */ + 671, /* GL_LESS */ 466, /* GL_EQUAL */ - 668, /* GL_LEQUAL */ - 591, /* GL_GREATER */ - 1023, /* GL_NOTEQUAL */ - 590, /* GL_GEQUAL */ + 670, /* GL_LEQUAL */ + 592, /* GL_GREATER */ + 1025, /* GL_NOTEQUAL */ + 591, /* GL_GEQUAL */ 47, /* GL_ALWAYS */ - 1468, /* GL_SRC_COLOR */ - 1053, /* GL_ONE_MINUS_SRC_COLOR */ - 1466, /* GL_SRC_ALPHA */ - 1052, /* GL_ONE_MINUS_SRC_ALPHA */ + 1472, /* GL_SRC_COLOR */ + 1055, /* GL_ONE_MINUS_SRC_COLOR */ + 1470, /* GL_SRC_ALPHA */ + 1054, /* GL_ONE_MINUS_SRC_ALPHA */ 445, /* GL_DST_ALPHA */ - 1050, /* GL_ONE_MINUS_DST_ALPHA */ + 1052, /* GL_ONE_MINUS_DST_ALPHA */ 446, /* GL_DST_COLOR */ - 1051, /* GL_ONE_MINUS_DST_COLOR */ - 1467, /* GL_SRC_ALPHA_SATURATE */ - 578, /* GL_FRONT_LEFT */ - 579, /* GL_FRONT_RIGHT */ + 1053, /* GL_ONE_MINUS_DST_COLOR */ + 1471, /* GL_SRC_ALPHA_SATURATE */ + 579, /* GL_FRONT_LEFT */ + 580, /* GL_FRONT_RIGHT */ 69, /* GL_BACK_LEFT */ 70, /* GL_BACK_RIGHT */ - 575, /* GL_FRONT */ + 576, /* GL_FRONT */ 68, /* GL_BACK */ - 667, /* GL_LEFT */ - 1369, /* GL_RIGHT */ - 576, /* GL_FRONT_AND_BACK */ + 669, /* GL_LEFT */ + 1373, /* GL_RIGHT */ + 577, /* GL_FRONT_AND_BACK */ 63, /* GL_AUX0 */ 64, /* GL_AUX1 */ 65, /* GL_AUX2 */ 66, /* GL_AUX3 */ - 657, /* GL_INVALID_ENUM */ - 661, /* GL_INVALID_VALUE */ - 660, /* GL_INVALID_OPERATION */ - 1473, /* GL_STACK_OVERFLOW */ - 1474, /* GL_STACK_UNDERFLOW */ - 1078, /* GL_OUT_OF_MEMORY */ - 658, /* GL_INVALID_FRAMEBUFFER_OPERATION */ + 658, /* GL_INVALID_ENUM */ + 662, /* GL_INVALID_VALUE */ + 661, /* GL_INVALID_OPERATION */ + 1477, /* GL_STACK_OVERFLOW */ + 1478, /* GL_STACK_UNDERFLOW */ + 1080, /* GL_OUT_OF_MEMORY */ + 659, /* GL_INVALID_FRAMEBUFFER_OPERATION */ 0, /* GL_2D */ 2, /* GL_3D */ 3, /* GL_3D_COLOR */ 4, /* GL_3D_COLOR_TEXTURE */ 6, /* GL_4D_COLOR_TEXTURE */ - 1098, /* GL_PASS_THROUGH_TOKEN */ - 1158, /* GL_POINT_TOKEN */ - 702, /* GL_LINE_TOKEN */ - 1172, /* GL_POLYGON_TOKEN */ + 1100, /* GL_PASS_THROUGH_TOKEN */ + 1160, /* GL_POINT_TOKEN */ + 704, /* GL_LINE_TOKEN */ + 1174, /* GL_POLYGON_TOKEN */ 74, /* GL_BITMAP_TOKEN */ 444, /* GL_DRAW_PIXEL_TOKEN */ 301, /* GL_COPY_PIXEL_TOKEN */ - 695, /* GL_LINE_RESET_TOKEN */ + 697, /* GL_LINE_RESET_TOKEN */ 469, /* GL_EXP */ 470, /* GL_EXP2 */ 337, /* GL_CW */ 125, /* GL_CCW */ 146, /* GL_COEFF */ - 1075, /* GL_ORDER */ + 1077, /* GL_ORDER */ 382, /* GL_DOMAIN */ 311, /* GL_CURRENT_COLOR */ 314, /* GL_CURRENT_INDEX */ @@ -3846,67 +3854,67 @@ static const unsigned reduced_enums[1347] = 328, /* GL_CURRENT_RASTER_POSITION */ 329, /* GL_CURRENT_RASTER_POSITION_VALID */ 326, /* GL_CURRENT_RASTER_DISTANCE */ - 1151, /* GL_POINT_SMOOTH */ - 1140, /* GL_POINT_SIZE */ - 1150, /* GL_POINT_SIZE_RANGE */ - 1141, /* GL_POINT_SIZE_GRANULARITY */ - 696, /* GL_LINE_SMOOTH */ - 703, /* GL_LINE_WIDTH */ - 705, /* GL_LINE_WIDTH_RANGE */ - 704, /* GL_LINE_WIDTH_GRANULARITY */ - 698, /* GL_LINE_STIPPLE */ - 699, /* GL_LINE_STIPPLE_PATTERN */ - 700, /* GL_LINE_STIPPLE_REPEAT */ - 710, /* GL_LIST_MODE */ - 875, /* GL_MAX_LIST_NESTING */ - 707, /* GL_LIST_BASE */ - 709, /* GL_LIST_INDEX */ - 1161, /* GL_POLYGON_MODE */ - 1168, /* GL_POLYGON_SMOOTH */ - 1170, /* GL_POLYGON_STIPPLE */ + 1153, /* GL_POINT_SMOOTH */ + 1142, /* GL_POINT_SIZE */ + 1152, /* GL_POINT_SIZE_RANGE */ + 1143, /* GL_POINT_SIZE_GRANULARITY */ + 698, /* GL_LINE_SMOOTH */ + 705, /* GL_LINE_WIDTH */ + 707, /* GL_LINE_WIDTH_RANGE */ + 706, /* GL_LINE_WIDTH_GRANULARITY */ + 700, /* GL_LINE_STIPPLE */ + 701, /* GL_LINE_STIPPLE_PATTERN */ + 702, /* GL_LINE_STIPPLE_REPEAT */ + 712, /* GL_LIST_MODE */ + 877, /* GL_MAX_LIST_NESTING */ + 709, /* GL_LIST_BASE */ + 711, /* GL_LIST_INDEX */ + 1163, /* GL_POLYGON_MODE */ + 1170, /* GL_POLYGON_SMOOTH */ + 1172, /* GL_POLYGON_STIPPLE */ 455, /* GL_EDGE_FLAG */ 304, /* GL_CULL_FACE */ 305, /* GL_CULL_FACE_MODE */ - 577, /* GL_FRONT_FACE */ - 678, /* GL_LIGHTING */ - 683, /* GL_LIGHT_MODEL_LOCAL_VIEWER */ - 684, /* GL_LIGHT_MODEL_TWO_SIDE */ - 680, /* GL_LIGHT_MODEL_AMBIENT */ - 1415, /* GL_SHADE_MODEL */ + 578, /* GL_FRONT_FACE */ + 680, /* GL_LIGHTING */ + 685, /* GL_LIGHT_MODEL_LOCAL_VIEWER */ + 686, /* GL_LIGHT_MODEL_TWO_SIDE */ + 682, /* GL_LIGHT_MODEL_AMBIENT */ + 1419, /* GL_SHADE_MODEL */ 193, /* GL_COLOR_MATERIAL_FACE */ 194, /* GL_COLOR_MATERIAL_PARAMETER */ 192, /* GL_COLOR_MATERIAL */ - 504, /* GL_FOG */ - 526, /* GL_FOG_INDEX */ - 522, /* GL_FOG_DENSITY */ - 530, /* GL_FOG_START */ - 524, /* GL_FOG_END */ - 527, /* GL_FOG_MODE */ - 506, /* GL_FOG_COLOR */ + 505, /* GL_FOG */ + 527, /* GL_FOG_INDEX */ + 523, /* GL_FOG_DENSITY */ + 531, /* GL_FOG_START */ + 525, /* GL_FOG_END */ + 528, /* GL_FOG_MODE */ + 507, /* GL_FOG_COLOR */ 369, /* GL_DEPTH_RANGE */ 376, /* GL_DEPTH_TEST */ 379, /* GL_DEPTH_WRITEMASK */ 357, /* GL_DEPTH_CLEAR_VALUE */ 368, /* GL_DEPTH_FUNC */ 12, /* GL_ACCUM_CLEAR_VALUE */ - 1509, /* GL_STENCIL_TEST */ - 1497, /* GL_STENCIL_CLEAR_VALUE */ - 1499, /* GL_STENCIL_FUNC */ - 1511, /* GL_STENCIL_VALUE_MASK */ - 1498, /* GL_STENCIL_FAIL */ - 1506, /* GL_STENCIL_PASS_DEPTH_FAIL */ - 1507, /* GL_STENCIL_PASS_DEPTH_PASS */ - 1508, /* GL_STENCIL_REF */ - 1512, /* GL_STENCIL_WRITEMASK */ - 844, /* GL_MATRIX_MODE */ - 1013, /* GL_NORMALIZE */ - 1838, /* GL_VIEWPORT */ - 987, /* GL_MODELVIEW_STACK_DEPTH */ - 1251, /* GL_PROJECTION_STACK_DEPTH */ - 1719, /* GL_TEXTURE_STACK_DEPTH */ - 985, /* GL_MODELVIEW_MATRIX */ - 1250, /* GL_PROJECTION_MATRIX */ - 1702, /* GL_TEXTURE_MATRIX */ + 1513, /* GL_STENCIL_TEST */ + 1501, /* GL_STENCIL_CLEAR_VALUE */ + 1503, /* GL_STENCIL_FUNC */ + 1515, /* GL_STENCIL_VALUE_MASK */ + 1502, /* GL_STENCIL_FAIL */ + 1510, /* GL_STENCIL_PASS_DEPTH_FAIL */ + 1511, /* GL_STENCIL_PASS_DEPTH_PASS */ + 1512, /* GL_STENCIL_REF */ + 1516, /* GL_STENCIL_WRITEMASK */ + 846, /* GL_MATRIX_MODE */ + 1015, /* GL_NORMALIZE */ + 1842, /* GL_VIEWPORT */ + 989, /* GL_MODELVIEW_STACK_DEPTH */ + 1253, /* GL_PROJECTION_STACK_DEPTH */ + 1723, /* GL_TEXTURE_STACK_DEPTH */ + 987, /* GL_MODELVIEW_MATRIX */ + 1252, /* GL_PROJECTION_MATRIX */ + 1706, /* GL_TEXTURE_MATRIX */ 61, /* GL_ATTRIB_STACK_DEPTH */ 136, /* GL_CLIENT_ATTRIB_STACK_DEPTH */ 43, /* GL_ALPHA_TEST */ @@ -3916,162 +3924,162 @@ static const unsigned reduced_enums[1347] = 78, /* GL_BLEND_DST */ 87, /* GL_BLEND_SRC */ 75, /* GL_BLEND */ - 713, /* GL_LOGIC_OP_MODE */ - 631, /* GL_INDEX_LOGIC_OP */ + 715, /* GL_LOGIC_OP_MODE */ + 632, /* GL_INDEX_LOGIC_OP */ 191, /* GL_COLOR_LOGIC_OP */ 67, /* GL_AUX_BUFFERS */ 392, /* GL_DRAW_BUFFER */ - 1285, /* GL_READ_BUFFER */ - 1396, /* GL_SCISSOR_BOX */ - 1397, /* GL_SCISSOR_TEST */ - 630, /* GL_INDEX_CLEAR_VALUE */ - 635, /* GL_INDEX_WRITEMASK */ + 1289, /* GL_READ_BUFFER */ + 1400, /* GL_SCISSOR_BOX */ + 1401, /* GL_SCISSOR_TEST */ + 631, /* GL_INDEX_CLEAR_VALUE */ + 636, /* GL_INDEX_WRITEMASK */ 188, /* GL_COLOR_CLEAR_VALUE */ 230, /* GL_COLOR_WRITEMASK */ - 632, /* GL_INDEX_MODE */ - 1362, /* GL_RGBA_MODE */ + 633, /* GL_INDEX_MODE */ + 1366, /* GL_RGBA_MODE */ 391, /* GL_DOUBLEBUFFER */ - 1513, /* GL_STEREO */ - 1320, /* GL_RENDER_MODE */ - 1099, /* GL_PERSPECTIVE_CORRECTION_HINT */ - 1152, /* GL_POINT_SMOOTH_HINT */ - 697, /* GL_LINE_SMOOTH_HINT */ - 1169, /* GL_POLYGON_SMOOTH_HINT */ - 525, /* GL_FOG_HINT */ - 1683, /* GL_TEXTURE_GEN_S */ - 1684, /* GL_TEXTURE_GEN_T */ - 1682, /* GL_TEXTURE_GEN_R */ - 1681, /* GL_TEXTURE_GEN_Q */ - 1112, /* GL_PIXEL_MAP_I_TO_I */ - 1118, /* GL_PIXEL_MAP_S_TO_S */ - 1114, /* GL_PIXEL_MAP_I_TO_R */ - 1110, /* GL_PIXEL_MAP_I_TO_G */ - 1108, /* GL_PIXEL_MAP_I_TO_B */ - 1106, /* GL_PIXEL_MAP_I_TO_A */ - 1116, /* GL_PIXEL_MAP_R_TO_R */ - 1104, /* GL_PIXEL_MAP_G_TO_G */ - 1102, /* GL_PIXEL_MAP_B_TO_B */ - 1100, /* GL_PIXEL_MAP_A_TO_A */ - 1113, /* GL_PIXEL_MAP_I_TO_I_SIZE */ - 1119, /* GL_PIXEL_MAP_S_TO_S_SIZE */ - 1115, /* GL_PIXEL_MAP_I_TO_R_SIZE */ - 1111, /* GL_PIXEL_MAP_I_TO_G_SIZE */ - 1109, /* GL_PIXEL_MAP_I_TO_B_SIZE */ - 1107, /* GL_PIXEL_MAP_I_TO_A_SIZE */ - 1117, /* GL_PIXEL_MAP_R_TO_R_SIZE */ - 1105, /* GL_PIXEL_MAP_G_TO_G_SIZE */ - 1103, /* GL_PIXEL_MAP_B_TO_B_SIZE */ - 1101, /* GL_PIXEL_MAP_A_TO_A_SIZE */ - 1756, /* GL_UNPACK_SWAP_BYTES */ - 1751, /* GL_UNPACK_LSB_FIRST */ - 1752, /* GL_UNPACK_ROW_LENGTH */ - 1755, /* GL_UNPACK_SKIP_ROWS */ - 1754, /* GL_UNPACK_SKIP_PIXELS */ - 1749, /* GL_UNPACK_ALIGNMENT */ - 1087, /* GL_PACK_SWAP_BYTES */ - 1082, /* GL_PACK_LSB_FIRST */ - 1083, /* GL_PACK_ROW_LENGTH */ - 1086, /* GL_PACK_SKIP_ROWS */ - 1085, /* GL_PACK_SKIP_PIXELS */ - 1079, /* GL_PACK_ALIGNMENT */ - 791, /* GL_MAP_COLOR */ - 796, /* GL_MAP_STENCIL */ - 634, /* GL_INDEX_SHIFT */ - 633, /* GL_INDEX_OFFSET */ - 1298, /* GL_RED_SCALE */ - 1296, /* GL_RED_BIAS */ - 1856, /* GL_ZOOM_X */ - 1857, /* GL_ZOOM_Y */ - 595, /* GL_GREEN_SCALE */ - 593, /* GL_GREEN_BIAS */ + 1517, /* GL_STEREO */ + 1324, /* GL_RENDER_MODE */ + 1101, /* GL_PERSPECTIVE_CORRECTION_HINT */ + 1154, /* GL_POINT_SMOOTH_HINT */ + 699, /* GL_LINE_SMOOTH_HINT */ + 1171, /* GL_POLYGON_SMOOTH_HINT */ + 526, /* GL_FOG_HINT */ + 1687, /* GL_TEXTURE_GEN_S */ + 1688, /* GL_TEXTURE_GEN_T */ + 1686, /* GL_TEXTURE_GEN_R */ + 1685, /* GL_TEXTURE_GEN_Q */ + 1114, /* GL_PIXEL_MAP_I_TO_I */ + 1120, /* GL_PIXEL_MAP_S_TO_S */ + 1116, /* GL_PIXEL_MAP_I_TO_R */ + 1112, /* GL_PIXEL_MAP_I_TO_G */ + 1110, /* GL_PIXEL_MAP_I_TO_B */ + 1108, /* GL_PIXEL_MAP_I_TO_A */ + 1118, /* GL_PIXEL_MAP_R_TO_R */ + 1106, /* GL_PIXEL_MAP_G_TO_G */ + 1104, /* GL_PIXEL_MAP_B_TO_B */ + 1102, /* GL_PIXEL_MAP_A_TO_A */ + 1115, /* GL_PIXEL_MAP_I_TO_I_SIZE */ + 1121, /* GL_PIXEL_MAP_S_TO_S_SIZE */ + 1117, /* GL_PIXEL_MAP_I_TO_R_SIZE */ + 1113, /* GL_PIXEL_MAP_I_TO_G_SIZE */ + 1111, /* GL_PIXEL_MAP_I_TO_B_SIZE */ + 1109, /* GL_PIXEL_MAP_I_TO_A_SIZE */ + 1119, /* GL_PIXEL_MAP_R_TO_R_SIZE */ + 1107, /* GL_PIXEL_MAP_G_TO_G_SIZE */ + 1105, /* GL_PIXEL_MAP_B_TO_B_SIZE */ + 1103, /* GL_PIXEL_MAP_A_TO_A_SIZE */ + 1760, /* GL_UNPACK_SWAP_BYTES */ + 1755, /* GL_UNPACK_LSB_FIRST */ + 1756, /* GL_UNPACK_ROW_LENGTH */ + 1759, /* GL_UNPACK_SKIP_ROWS */ + 1758, /* GL_UNPACK_SKIP_PIXELS */ + 1753, /* GL_UNPACK_ALIGNMENT */ + 1089, /* GL_PACK_SWAP_BYTES */ + 1084, /* GL_PACK_LSB_FIRST */ + 1085, /* GL_PACK_ROW_LENGTH */ + 1088, /* GL_PACK_SKIP_ROWS */ + 1087, /* GL_PACK_SKIP_PIXELS */ + 1081, /* GL_PACK_ALIGNMENT */ + 793, /* GL_MAP_COLOR */ + 798, /* GL_MAP_STENCIL */ + 635, /* GL_INDEX_SHIFT */ + 634, /* GL_INDEX_OFFSET */ + 1302, /* GL_RED_SCALE */ + 1300, /* GL_RED_BIAS */ + 1860, /* GL_ZOOM_X */ + 1861, /* GL_ZOOM_Y */ + 596, /* GL_GREEN_SCALE */ + 594, /* GL_GREEN_BIAS */ 93, /* GL_BLUE_SCALE */ 91, /* GL_BLUE_BIAS */ 42, /* GL_ALPHA_SCALE */ 40, /* GL_ALPHA_BIAS */ 370, /* GL_DEPTH_SCALE */ 350, /* GL_DEPTH_BIAS */ - 870, /* GL_MAX_EVAL_ORDER */ - 874, /* GL_MAX_LIGHTS */ - 853, /* GL_MAX_CLIP_PLANES */ - 920, /* GL_MAX_TEXTURE_SIZE */ - 880, /* GL_MAX_PIXEL_MAP_TABLE */ - 849, /* GL_MAX_ATTRIB_STACK_DEPTH */ - 877, /* GL_MAX_MODELVIEW_STACK_DEPTH */ - 878, /* GL_MAX_NAME_STACK_DEPTH */ - 906, /* GL_MAX_PROJECTION_STACK_DEPTH */ - 921, /* GL_MAX_TEXTURE_STACK_DEPTH */ - 935, /* GL_MAX_VIEWPORT_DIMS */ - 850, /* GL_MAX_CLIENT_ATTRIB_STACK_DEPTH */ - 1523, /* GL_SUBPIXEL_BITS */ - 629, /* GL_INDEX_BITS */ - 1297, /* GL_RED_BITS */ - 594, /* GL_GREEN_BITS */ + 872, /* GL_MAX_EVAL_ORDER */ + 876, /* GL_MAX_LIGHTS */ + 855, /* GL_MAX_CLIP_PLANES */ + 922, /* GL_MAX_TEXTURE_SIZE */ + 882, /* GL_MAX_PIXEL_MAP_TABLE */ + 851, /* GL_MAX_ATTRIB_STACK_DEPTH */ + 879, /* GL_MAX_MODELVIEW_STACK_DEPTH */ + 880, /* GL_MAX_NAME_STACK_DEPTH */ + 908, /* GL_MAX_PROJECTION_STACK_DEPTH */ + 923, /* GL_MAX_TEXTURE_STACK_DEPTH */ + 937, /* GL_MAX_VIEWPORT_DIMS */ + 852, /* GL_MAX_CLIENT_ATTRIB_STACK_DEPTH */ + 1527, /* GL_SUBPIXEL_BITS */ + 630, /* GL_INDEX_BITS */ + 1301, /* GL_RED_BITS */ + 595, /* GL_GREEN_BITS */ 92, /* GL_BLUE_BITS */ 41, /* GL_ALPHA_BITS */ 351, /* GL_DEPTH_BITS */ - 1495, /* GL_STENCIL_BITS */ + 1499, /* GL_STENCIL_BITS */ 14, /* GL_ACCUM_RED_BITS */ 13, /* GL_ACCUM_GREEN_BITS */ 10, /* GL_ACCUM_BLUE_BITS */ 9, /* GL_ACCUM_ALPHA_BITS */ - 1001, /* GL_NAME_STACK_DEPTH */ + 1003, /* GL_NAME_STACK_DEPTH */ 62, /* GL_AUTO_NORMAL */ - 737, /* GL_MAP1_COLOR_4 */ - 740, /* GL_MAP1_INDEX */ - 741, /* GL_MAP1_NORMAL */ - 742, /* GL_MAP1_TEXTURE_COORD_1 */ - 743, /* GL_MAP1_TEXTURE_COORD_2 */ - 744, /* GL_MAP1_TEXTURE_COORD_3 */ - 745, /* GL_MAP1_TEXTURE_COORD_4 */ - 746, /* GL_MAP1_VERTEX_3 */ - 747, /* GL_MAP1_VERTEX_4 */ - 764, /* GL_MAP2_COLOR_4 */ - 767, /* GL_MAP2_INDEX */ - 768, /* GL_MAP2_NORMAL */ - 769, /* GL_MAP2_TEXTURE_COORD_1 */ - 770, /* GL_MAP2_TEXTURE_COORD_2 */ - 771, /* GL_MAP2_TEXTURE_COORD_3 */ - 772, /* GL_MAP2_TEXTURE_COORD_4 */ - 773, /* GL_MAP2_VERTEX_3 */ - 774, /* GL_MAP2_VERTEX_4 */ - 738, /* GL_MAP1_GRID_DOMAIN */ - 739, /* GL_MAP1_GRID_SEGMENTS */ - 765, /* GL_MAP2_GRID_DOMAIN */ - 766, /* GL_MAP2_GRID_SEGMENTS */ - 1606, /* GL_TEXTURE_1D */ - 1608, /* GL_TEXTURE_2D */ + 739, /* GL_MAP1_COLOR_4 */ + 742, /* GL_MAP1_INDEX */ + 743, /* GL_MAP1_NORMAL */ + 744, /* GL_MAP1_TEXTURE_COORD_1 */ + 745, /* GL_MAP1_TEXTURE_COORD_2 */ + 746, /* GL_MAP1_TEXTURE_COORD_3 */ + 747, /* GL_MAP1_TEXTURE_COORD_4 */ + 748, /* GL_MAP1_VERTEX_3 */ + 749, /* GL_MAP1_VERTEX_4 */ + 766, /* GL_MAP2_COLOR_4 */ + 769, /* GL_MAP2_INDEX */ + 770, /* GL_MAP2_NORMAL */ + 771, /* GL_MAP2_TEXTURE_COORD_1 */ + 772, /* GL_MAP2_TEXTURE_COORD_2 */ + 773, /* GL_MAP2_TEXTURE_COORD_3 */ + 774, /* GL_MAP2_TEXTURE_COORD_4 */ + 775, /* GL_MAP2_VERTEX_3 */ + 776, /* GL_MAP2_VERTEX_4 */ + 740, /* GL_MAP1_GRID_DOMAIN */ + 741, /* GL_MAP1_GRID_SEGMENTS */ + 767, /* GL_MAP2_GRID_DOMAIN */ + 768, /* GL_MAP2_GRID_SEGMENTS */ + 1610, /* GL_TEXTURE_1D */ + 1612, /* GL_TEXTURE_2D */ 479, /* GL_FEEDBACK_BUFFER_POINTER */ 480, /* GL_FEEDBACK_BUFFER_SIZE */ 481, /* GL_FEEDBACK_BUFFER_TYPE */ - 1406, /* GL_SELECTION_BUFFER_POINTER */ - 1407, /* GL_SELECTION_BUFFER_SIZE */ - 1724, /* GL_TEXTURE_WIDTH */ - 1688, /* GL_TEXTURE_HEIGHT */ - 1643, /* GL_TEXTURE_COMPONENTS */ - 1627, /* GL_TEXTURE_BORDER_COLOR */ - 1626, /* GL_TEXTURE_BORDER */ + 1410, /* GL_SELECTION_BUFFER_POINTER */ + 1411, /* GL_SELECTION_BUFFER_SIZE */ + 1728, /* GL_TEXTURE_WIDTH */ + 1692, /* GL_TEXTURE_HEIGHT */ + 1647, /* GL_TEXTURE_COMPONENTS */ + 1631, /* GL_TEXTURE_BORDER_COLOR */ + 1630, /* GL_TEXTURE_BORDER */ 383, /* GL_DONT_CARE */ 477, /* GL_FASTEST */ - 1009, /* GL_NICEST */ + 1011, /* GL_NICEST */ 48, /* GL_AMBIENT */ 380, /* GL_DIFFUSE */ - 1455, /* GL_SPECULAR */ - 1173, /* GL_POSITION */ - 1458, /* GL_SPOT_DIRECTION */ - 1459, /* GL_SPOT_EXPONENT */ - 1457, /* GL_SPOT_CUTOFF */ + 1459, /* GL_SPECULAR */ + 1175, /* GL_POSITION */ + 1462, /* GL_SPOT_DIRECTION */ + 1463, /* GL_SPOT_EXPONENT */ + 1461, /* GL_SPOT_CUTOFF */ 275, /* GL_CONSTANT_ATTENUATION */ - 687, /* GL_LINEAR_ATTENUATION */ - 1271, /* GL_QUADRATIC_ATTENUATION */ + 689, /* GL_LINEAR_ATTENUATION */ + 1274, /* GL_QUADRATIC_ATTENUATION */ 244, /* GL_COMPILE */ 245, /* GL_COMPILE_AND_EXECUTE */ 120, /* GL_BYTE */ - 1758, /* GL_UNSIGNED_BYTE */ - 1420, /* GL_SHORT */ - 1769, /* GL_UNSIGNED_SHORT */ - 637, /* GL_INT */ - 1761, /* GL_UNSIGNED_INT */ - 485, /* GL_FLOAT */ + 1762, /* GL_UNSIGNED_BYTE */ + 1424, /* GL_SHORT */ + 1773, /* GL_UNSIGNED_SHORT */ + 638, /* GL_INT */ + 1765, /* GL_UNSIGNED_INT */ + 486, /* GL_FLOAT */ 1, /* GL_2_BYTES */ 5, /* GL_3_BYTES */ 7, /* GL_4_BYTES */ @@ -4081,284 +4089,284 @@ static const unsigned reduced_enums[1347] = 52, /* GL_AND_REVERSE */ 299, /* GL_COPY */ 51, /* GL_AND_INVERTED */ - 1011, /* GL_NOOP */ - 1852, /* GL_XOR */ - 1074, /* GL_OR */ - 1012, /* GL_NOR */ + 1013, /* GL_NOOP */ + 1856, /* GL_XOR */ + 1076, /* GL_OR */ + 1014, /* GL_NOR */ 467, /* GL_EQUIV */ - 664, /* GL_INVERT */ - 1077, /* GL_OR_REVERSE */ + 665, /* GL_INVERT */ + 1079, /* GL_OR_REVERSE */ 300, /* GL_COPY_INVERTED */ - 1076, /* GL_OR_INVERTED */ - 1002, /* GL_NAND */ - 1411, /* GL_SET */ + 1078, /* GL_OR_INVERTED */ + 1004, /* GL_NAND */ + 1415, /* GL_SET */ 464, /* GL_EMISSION */ - 1419, /* GL_SHININESS */ + 1423, /* GL_SHININESS */ 49, /* GL_AMBIENT_AND_DIFFUSE */ 190, /* GL_COLOR_INDEXES */ - 952, /* GL_MODELVIEW */ - 1249, /* GL_PROJECTION */ - 1541, /* GL_TEXTURE */ + 954, /* GL_MODELVIEW */ + 1251, /* GL_PROJECTION */ + 1545, /* GL_TEXTURE */ 147, /* GL_COLOR */ 346, /* GL_DEPTH */ - 1481, /* GL_STENCIL */ + 1485, /* GL_STENCIL */ 189, /* GL_COLOR_INDEX */ - 1500, /* GL_STENCIL_INDEX */ + 1504, /* GL_STENCIL_INDEX */ 358, /* GL_DEPTH_COMPONENT */ - 1293, /* GL_RED */ - 592, /* GL_GREEN */ + 1297, /* GL_RED */ + 593, /* GL_GREEN */ 90, /* GL_BLUE */ 31, /* GL_ALPHA */ - 1328, /* GL_RGB */ - 1347, /* GL_RGBA */ - 715, /* GL_LUMINANCE */ - 736, /* GL_LUMINANCE_ALPHA */ + 1332, /* GL_RGB */ + 1351, /* GL_RGBA */ + 717, /* GL_LUMINANCE */ + 738, /* GL_LUMINANCE_ALPHA */ 73, /* GL_BITMAP */ - 1129, /* GL_POINT */ - 685, /* GL_LINE */ + 1131, /* GL_POINT */ + 687, /* GL_LINE */ 482, /* GL_FILL */ - 1302, /* GL_RENDER */ + 1306, /* GL_RENDER */ 478, /* GL_FEEDBACK */ - 1405, /* GL_SELECT */ - 484, /* GL_FLAT */ - 1430, /* GL_SMOOTH */ - 665, /* GL_KEEP */ - 1322, /* GL_REPLACE */ - 619, /* GL_INCR */ + 1409, /* GL_SELECT */ + 485, /* GL_FLAT */ + 1434, /* GL_SMOOTH */ + 666, /* GL_KEEP */ + 1326, /* GL_REPLACE */ + 620, /* GL_INCR */ 342, /* GL_DECR */ - 1784, /* GL_VENDOR */ - 1319, /* GL_RENDERER */ - 1785, /* GL_VERSION */ + 1788, /* GL_VENDOR */ + 1323, /* GL_RENDERER */ + 1789, /* GL_VERSION */ 471, /* GL_EXTENSIONS */ - 1370, /* GL_S */ - 1532, /* GL_T */ - 1282, /* GL_R */ - 1270, /* GL_Q */ - 988, /* GL_MODULATE */ + 1374, /* GL_S */ + 1536, /* GL_T */ + 1286, /* GL_R */ + 1273, /* GL_Q */ + 990, /* GL_MODULATE */ 341, /* GL_DECAL */ - 1678, /* GL_TEXTURE_ENV_MODE */ - 1677, /* GL_TEXTURE_ENV_COLOR */ - 1676, /* GL_TEXTURE_ENV */ + 1682, /* GL_TEXTURE_ENV_MODE */ + 1681, /* GL_TEXTURE_ENV_COLOR */ + 1680, /* GL_TEXTURE_ENV */ 472, /* GL_EYE_LINEAR */ - 1035, /* GL_OBJECT_LINEAR */ - 1456, /* GL_SPHERE_MAP */ - 1680, /* GL_TEXTURE_GEN_MODE */ - 1037, /* GL_OBJECT_PLANE */ + 1037, /* GL_OBJECT_LINEAR */ + 1460, /* GL_SPHERE_MAP */ + 1684, /* GL_TEXTURE_GEN_MODE */ + 1039, /* GL_OBJECT_PLANE */ 473, /* GL_EYE_PLANE */ - 1003, /* GL_NEAREST */ - 686, /* GL_LINEAR */ - 1007, /* GL_NEAREST_MIPMAP_NEAREST */ - 691, /* GL_LINEAR_MIPMAP_NEAREST */ - 1006, /* GL_NEAREST_MIPMAP_LINEAR */ - 690, /* GL_LINEAR_MIPMAP_LINEAR */ - 1701, /* GL_TEXTURE_MAG_FILTER */ - 1709, /* GL_TEXTURE_MIN_FILTER */ - 1726, /* GL_TEXTURE_WRAP_S */ - 1727, /* GL_TEXTURE_WRAP_T */ + 1005, /* GL_NEAREST */ + 688, /* GL_LINEAR */ + 1009, /* GL_NEAREST_MIPMAP_NEAREST */ + 693, /* GL_LINEAR_MIPMAP_NEAREST */ + 1008, /* GL_NEAREST_MIPMAP_LINEAR */ + 692, /* GL_LINEAR_MIPMAP_LINEAR */ + 1705, /* GL_TEXTURE_MAG_FILTER */ + 1713, /* GL_TEXTURE_MIN_FILTER */ + 1730, /* GL_TEXTURE_WRAP_S */ + 1731, /* GL_TEXTURE_WRAP_T */ 126, /* GL_CLAMP */ - 1321, /* GL_REPEAT */ - 1167, /* GL_POLYGON_OFFSET_UNITS */ - 1166, /* GL_POLYGON_OFFSET_POINT */ - 1165, /* GL_POLYGON_OFFSET_LINE */ - 1283, /* GL_R3_G3_B2 */ - 1781, /* GL_V2F */ - 1782, /* GL_V3F */ + 1325, /* GL_REPEAT */ + 1169, /* GL_POLYGON_OFFSET_UNITS */ + 1168, /* GL_POLYGON_OFFSET_POINT */ + 1167, /* GL_POLYGON_OFFSET_LINE */ + 1287, /* GL_R3_G3_B2 */ + 1785, /* GL_V2F */ + 1786, /* GL_V3F */ 123, /* GL_C4UB_V2F */ 124, /* GL_C4UB_V3F */ 121, /* GL_C3F_V3F */ - 1000, /* GL_N3F_V3F */ + 1002, /* GL_N3F_V3F */ 122, /* GL_C4F_N3F_V3F */ - 1537, /* GL_T2F_V3F */ - 1539, /* GL_T4F_V4F */ - 1535, /* GL_T2F_C4UB_V3F */ - 1533, /* GL_T2F_C3F_V3F */ - 1536, /* GL_T2F_N3F_V3F */ - 1534, /* GL_T2F_C4F_N3F_V3F */ - 1538, /* GL_T4F_C4F_N3F_V4F */ + 1541, /* GL_T2F_V3F */ + 1543, /* GL_T4F_V4F */ + 1539, /* GL_T2F_C4UB_V3F */ + 1537, /* GL_T2F_C3F_V3F */ + 1540, /* GL_T2F_N3F_V3F */ + 1538, /* GL_T2F_C4F_N3F_V3F */ + 1542, /* GL_T4F_C4F_N3F_V4F */ 139, /* GL_CLIP_PLANE0 */ 140, /* GL_CLIP_PLANE1 */ 141, /* GL_CLIP_PLANE2 */ 142, /* GL_CLIP_PLANE3 */ 143, /* GL_CLIP_PLANE4 */ 144, /* GL_CLIP_PLANE5 */ - 670, /* GL_LIGHT0 */ - 671, /* GL_LIGHT1 */ - 672, /* GL_LIGHT2 */ - 673, /* GL_LIGHT3 */ - 674, /* GL_LIGHT4 */ - 675, /* GL_LIGHT5 */ - 676, /* GL_LIGHT6 */ - 677, /* GL_LIGHT7 */ - 596, /* GL_HINT_BIT */ + 672, /* GL_LIGHT0 */ + 673, /* GL_LIGHT1 */ + 674, /* GL_LIGHT2 */ + 675, /* GL_LIGHT3 */ + 676, /* GL_LIGHT4 */ + 677, /* GL_LIGHT5 */ + 678, /* GL_LIGHT6 */ + 679, /* GL_LIGHT7 */ + 597, /* GL_HINT_BIT */ 277, /* GL_CONSTANT_COLOR */ - 1048, /* GL_ONE_MINUS_CONSTANT_COLOR */ + 1050, /* GL_ONE_MINUS_CONSTANT_COLOR */ 272, /* GL_CONSTANT_ALPHA */ - 1046, /* GL_ONE_MINUS_CONSTANT_ALPHA */ + 1048, /* GL_ONE_MINUS_CONSTANT_ALPHA */ 76, /* GL_BLEND_COLOR */ - 580, /* GL_FUNC_ADD */ - 936, /* GL_MIN */ - 846, /* GL_MAX */ + 581, /* GL_FUNC_ADD */ + 938, /* GL_MIN */ + 848, /* GL_MAX */ 81, /* GL_BLEND_EQUATION */ - 584, /* GL_FUNC_SUBTRACT */ - 582, /* GL_FUNC_REVERSE_SUBTRACT */ + 585, /* GL_FUNC_SUBTRACT */ + 583, /* GL_FUNC_REVERSE_SUBTRACT */ 280, /* GL_CONVOLUTION_1D */ 281, /* GL_CONVOLUTION_2D */ - 1408, /* GL_SEPARABLE_2D */ + 1412, /* GL_SEPARABLE_2D */ 284, /* GL_CONVOLUTION_BORDER_MODE */ 288, /* GL_CONVOLUTION_FILTER_SCALE */ 286, /* GL_CONVOLUTION_FILTER_BIAS */ - 1294, /* GL_REDUCE */ + 1298, /* GL_REDUCE */ 290, /* GL_CONVOLUTION_FORMAT */ 294, /* GL_CONVOLUTION_WIDTH */ 292, /* GL_CONVOLUTION_HEIGHT */ - 861, /* GL_MAX_CONVOLUTION_WIDTH */ - 859, /* GL_MAX_CONVOLUTION_HEIGHT */ - 1206, /* GL_POST_CONVOLUTION_RED_SCALE */ - 1202, /* GL_POST_CONVOLUTION_GREEN_SCALE */ - 1197, /* GL_POST_CONVOLUTION_BLUE_SCALE */ - 1193, /* GL_POST_CONVOLUTION_ALPHA_SCALE */ - 1204, /* GL_POST_CONVOLUTION_RED_BIAS */ - 1200, /* GL_POST_CONVOLUTION_GREEN_BIAS */ - 1195, /* GL_POST_CONVOLUTION_BLUE_BIAS */ - 1191, /* GL_POST_CONVOLUTION_ALPHA_BIAS */ - 597, /* GL_HISTOGRAM */ - 1254, /* GL_PROXY_HISTOGRAM */ - 613, /* GL_HISTOGRAM_WIDTH */ - 603, /* GL_HISTOGRAM_FORMAT */ - 609, /* GL_HISTOGRAM_RED_SIZE */ - 605, /* GL_HISTOGRAM_GREEN_SIZE */ - 600, /* GL_HISTOGRAM_BLUE_SIZE */ - 598, /* GL_HISTOGRAM_ALPHA_SIZE */ - 607, /* GL_HISTOGRAM_LUMINANCE_SIZE */ - 611, /* GL_HISTOGRAM_SINK */ - 937, /* GL_MINMAX */ - 939, /* GL_MINMAX_FORMAT */ - 941, /* GL_MINMAX_SINK */ - 1540, /* GL_TABLE_TOO_LARGE_EXT */ - 1760, /* GL_UNSIGNED_BYTE_3_3_2 */ - 1771, /* GL_UNSIGNED_SHORT_4_4_4_4 */ - 1773, /* GL_UNSIGNED_SHORT_5_5_5_1 */ - 1766, /* GL_UNSIGNED_INT_8_8_8_8 */ - 1762, /* GL_UNSIGNED_INT_10_10_10_2 */ - 1164, /* GL_POLYGON_OFFSET_FILL */ - 1163, /* GL_POLYGON_OFFSET_FACTOR */ - 1162, /* GL_POLYGON_OFFSET_BIAS */ - 1325, /* GL_RESCALE_NORMAL */ + 863, /* GL_MAX_CONVOLUTION_WIDTH */ + 861, /* GL_MAX_CONVOLUTION_HEIGHT */ + 1208, /* GL_POST_CONVOLUTION_RED_SCALE */ + 1204, /* GL_POST_CONVOLUTION_GREEN_SCALE */ + 1199, /* GL_POST_CONVOLUTION_BLUE_SCALE */ + 1195, /* GL_POST_CONVOLUTION_ALPHA_SCALE */ + 1206, /* GL_POST_CONVOLUTION_RED_BIAS */ + 1202, /* GL_POST_CONVOLUTION_GREEN_BIAS */ + 1197, /* GL_POST_CONVOLUTION_BLUE_BIAS */ + 1193, /* GL_POST_CONVOLUTION_ALPHA_BIAS */ + 598, /* GL_HISTOGRAM */ + 1257, /* GL_PROXY_HISTOGRAM */ + 614, /* GL_HISTOGRAM_WIDTH */ + 604, /* GL_HISTOGRAM_FORMAT */ + 610, /* GL_HISTOGRAM_RED_SIZE */ + 606, /* GL_HISTOGRAM_GREEN_SIZE */ + 601, /* GL_HISTOGRAM_BLUE_SIZE */ + 599, /* GL_HISTOGRAM_ALPHA_SIZE */ + 608, /* GL_HISTOGRAM_LUMINANCE_SIZE */ + 612, /* GL_HISTOGRAM_SINK */ + 939, /* GL_MINMAX */ + 941, /* GL_MINMAX_FORMAT */ + 943, /* GL_MINMAX_SINK */ + 1544, /* GL_TABLE_TOO_LARGE_EXT */ + 1764, /* GL_UNSIGNED_BYTE_3_3_2 */ + 1775, /* GL_UNSIGNED_SHORT_4_4_4_4 */ + 1777, /* GL_UNSIGNED_SHORT_5_5_5_1 */ + 1770, /* GL_UNSIGNED_INT_8_8_8_8 */ + 1766, /* GL_UNSIGNED_INT_10_10_10_2 */ + 1166, /* GL_POLYGON_OFFSET_FILL */ + 1165, /* GL_POLYGON_OFFSET_FACTOR */ + 1164, /* GL_POLYGON_OFFSET_BIAS */ + 1329, /* GL_RESCALE_NORMAL */ 36, /* GL_ALPHA4 */ 38, /* GL_ALPHA8 */ 32, /* GL_ALPHA12 */ 34, /* GL_ALPHA16 */ - 726, /* GL_LUMINANCE4 */ - 732, /* GL_LUMINANCE8 */ - 716, /* GL_LUMINANCE12 */ - 722, /* GL_LUMINANCE16 */ - 727, /* GL_LUMINANCE4_ALPHA4 */ - 730, /* GL_LUMINANCE6_ALPHA2 */ - 733, /* GL_LUMINANCE8_ALPHA8 */ - 719, /* GL_LUMINANCE12_ALPHA4 */ - 717, /* GL_LUMINANCE12_ALPHA12 */ - 723, /* GL_LUMINANCE16_ALPHA16 */ - 638, /* GL_INTENSITY */ - 643, /* GL_INTENSITY4 */ - 645, /* GL_INTENSITY8 */ - 639, /* GL_INTENSITY12 */ - 641, /* GL_INTENSITY16 */ - 1337, /* GL_RGB2_EXT */ - 1338, /* GL_RGB4 */ - 1341, /* GL_RGB5 */ - 1345, /* GL_RGB8 */ - 1329, /* GL_RGB10 */ - 1333, /* GL_RGB12 */ - 1335, /* GL_RGB16 */ - 1352, /* GL_RGBA2 */ - 1354, /* GL_RGBA4 */ - 1342, /* GL_RGB5_A1 */ - 1358, /* GL_RGBA8 */ - 1330, /* GL_RGB10_A2 */ - 1348, /* GL_RGBA12 */ - 1350, /* GL_RGBA16 */ - 1716, /* GL_TEXTURE_RED_SIZE */ - 1686, /* GL_TEXTURE_GREEN_SIZE */ - 1624, /* GL_TEXTURE_BLUE_SIZE */ - 1611, /* GL_TEXTURE_ALPHA_SIZE */ - 1699, /* GL_TEXTURE_LUMINANCE_SIZE */ - 1690, /* GL_TEXTURE_INTENSITY_SIZE */ - 1323, /* GL_REPLACE_EXT */ - 1258, /* GL_PROXY_TEXTURE_1D */ - 1261, /* GL_PROXY_TEXTURE_2D */ - 1722, /* GL_TEXTURE_TOO_LARGE_EXT */ - 1711, /* GL_TEXTURE_PRIORITY */ - 1718, /* GL_TEXTURE_RESIDENT */ - 1614, /* GL_TEXTURE_BINDING_1D */ - 1616, /* GL_TEXTURE_BINDING_2D */ - 1618, /* GL_TEXTURE_BINDING_3D */ - 1084, /* GL_PACK_SKIP_IMAGES */ - 1080, /* GL_PACK_IMAGE_HEIGHT */ - 1753, /* GL_UNPACK_SKIP_IMAGES */ - 1750, /* GL_UNPACK_IMAGE_HEIGHT */ - 1610, /* GL_TEXTURE_3D */ - 1264, /* GL_PROXY_TEXTURE_3D */ - 1673, /* GL_TEXTURE_DEPTH */ - 1725, /* GL_TEXTURE_WRAP_R */ - 847, /* GL_MAX_3D_TEXTURE_SIZE */ - 1786, /* GL_VERTEX_ARRAY */ - 1014, /* GL_NORMAL_ARRAY */ + 728, /* GL_LUMINANCE4 */ + 734, /* GL_LUMINANCE8 */ + 718, /* GL_LUMINANCE12 */ + 724, /* GL_LUMINANCE16 */ + 729, /* GL_LUMINANCE4_ALPHA4 */ + 732, /* GL_LUMINANCE6_ALPHA2 */ + 735, /* GL_LUMINANCE8_ALPHA8 */ + 721, /* GL_LUMINANCE12_ALPHA4 */ + 719, /* GL_LUMINANCE12_ALPHA12 */ + 725, /* GL_LUMINANCE16_ALPHA16 */ + 639, /* GL_INTENSITY */ + 644, /* GL_INTENSITY4 */ + 646, /* GL_INTENSITY8 */ + 640, /* GL_INTENSITY12 */ + 642, /* GL_INTENSITY16 */ + 1341, /* GL_RGB2_EXT */ + 1342, /* GL_RGB4 */ + 1345, /* GL_RGB5 */ + 1349, /* GL_RGB8 */ + 1333, /* GL_RGB10 */ + 1337, /* GL_RGB12 */ + 1339, /* GL_RGB16 */ + 1356, /* GL_RGBA2 */ + 1358, /* GL_RGBA4 */ + 1346, /* GL_RGB5_A1 */ + 1362, /* GL_RGBA8 */ + 1334, /* GL_RGB10_A2 */ + 1352, /* GL_RGBA12 */ + 1354, /* GL_RGBA16 */ + 1720, /* GL_TEXTURE_RED_SIZE */ + 1690, /* GL_TEXTURE_GREEN_SIZE */ + 1628, /* GL_TEXTURE_BLUE_SIZE */ + 1615, /* GL_TEXTURE_ALPHA_SIZE */ + 1703, /* GL_TEXTURE_LUMINANCE_SIZE */ + 1694, /* GL_TEXTURE_INTENSITY_SIZE */ + 1327, /* GL_REPLACE_EXT */ + 1261, /* GL_PROXY_TEXTURE_1D */ + 1264, /* GL_PROXY_TEXTURE_2D */ + 1726, /* GL_TEXTURE_TOO_LARGE_EXT */ + 1715, /* GL_TEXTURE_PRIORITY */ + 1722, /* GL_TEXTURE_RESIDENT */ + 1618, /* GL_TEXTURE_BINDING_1D */ + 1620, /* GL_TEXTURE_BINDING_2D */ + 1622, /* GL_TEXTURE_BINDING_3D */ + 1086, /* GL_PACK_SKIP_IMAGES */ + 1082, /* GL_PACK_IMAGE_HEIGHT */ + 1757, /* GL_UNPACK_SKIP_IMAGES */ + 1754, /* GL_UNPACK_IMAGE_HEIGHT */ + 1614, /* GL_TEXTURE_3D */ + 1267, /* GL_PROXY_TEXTURE_3D */ + 1677, /* GL_TEXTURE_DEPTH */ + 1729, /* GL_TEXTURE_WRAP_R */ + 849, /* GL_MAX_3D_TEXTURE_SIZE */ + 1790, /* GL_VERTEX_ARRAY */ + 1016, /* GL_NORMAL_ARRAY */ 148, /* GL_COLOR_ARRAY */ - 623, /* GL_INDEX_ARRAY */ - 1651, /* GL_TEXTURE_COORD_ARRAY */ + 624, /* GL_INDEX_ARRAY */ + 1655, /* GL_TEXTURE_COORD_ARRAY */ 456, /* GL_EDGE_FLAG_ARRAY */ - 1792, /* GL_VERTEX_ARRAY_SIZE */ - 1794, /* GL_VERTEX_ARRAY_TYPE */ - 1793, /* GL_VERTEX_ARRAY_STRIDE */ - 1019, /* GL_NORMAL_ARRAY_TYPE */ - 1018, /* GL_NORMAL_ARRAY_STRIDE */ + 1796, /* GL_VERTEX_ARRAY_SIZE */ + 1798, /* GL_VERTEX_ARRAY_TYPE */ + 1797, /* GL_VERTEX_ARRAY_STRIDE */ + 1021, /* GL_NORMAL_ARRAY_TYPE */ + 1020, /* GL_NORMAL_ARRAY_STRIDE */ 152, /* GL_COLOR_ARRAY_SIZE */ 154, /* GL_COLOR_ARRAY_TYPE */ 153, /* GL_COLOR_ARRAY_STRIDE */ - 628, /* GL_INDEX_ARRAY_TYPE */ - 627, /* GL_INDEX_ARRAY_STRIDE */ - 1655, /* GL_TEXTURE_COORD_ARRAY_SIZE */ - 1657, /* GL_TEXTURE_COORD_ARRAY_TYPE */ - 1656, /* GL_TEXTURE_COORD_ARRAY_STRIDE */ + 629, /* GL_INDEX_ARRAY_TYPE */ + 628, /* GL_INDEX_ARRAY_STRIDE */ + 1659, /* GL_TEXTURE_COORD_ARRAY_SIZE */ + 1661, /* GL_TEXTURE_COORD_ARRAY_TYPE */ + 1660, /* GL_TEXTURE_COORD_ARRAY_STRIDE */ 460, /* GL_EDGE_FLAG_ARRAY_STRIDE */ - 1791, /* GL_VERTEX_ARRAY_POINTER */ - 1017, /* GL_NORMAL_ARRAY_POINTER */ + 1795, /* GL_VERTEX_ARRAY_POINTER */ + 1019, /* GL_NORMAL_ARRAY_POINTER */ 151, /* GL_COLOR_ARRAY_POINTER */ - 626, /* GL_INDEX_ARRAY_POINTER */ - 1654, /* GL_TEXTURE_COORD_ARRAY_POINTER */ + 627, /* GL_INDEX_ARRAY_POINTER */ + 1658, /* GL_TEXTURE_COORD_ARRAY_POINTER */ 459, /* GL_EDGE_FLAG_ARRAY_POINTER */ - 993, /* GL_MULTISAMPLE */ - 1382, /* GL_SAMPLE_ALPHA_TO_COVERAGE */ - 1384, /* GL_SAMPLE_ALPHA_TO_ONE */ - 1389, /* GL_SAMPLE_COVERAGE */ - 1386, /* GL_SAMPLE_BUFFERS */ - 1377, /* GL_SAMPLES */ - 1393, /* GL_SAMPLE_COVERAGE_VALUE */ - 1391, /* GL_SAMPLE_COVERAGE_INVERT */ + 995, /* GL_MULTISAMPLE */ + 1386, /* GL_SAMPLE_ALPHA_TO_COVERAGE */ + 1388, /* GL_SAMPLE_ALPHA_TO_ONE */ + 1393, /* GL_SAMPLE_COVERAGE */ + 1390, /* GL_SAMPLE_BUFFERS */ + 1381, /* GL_SAMPLES */ + 1397, /* GL_SAMPLE_COVERAGE_VALUE */ + 1395, /* GL_SAMPLE_COVERAGE_INVERT */ 195, /* GL_COLOR_MATRIX */ 197, /* GL_COLOR_MATRIX_STACK_DEPTH */ - 855, /* GL_MAX_COLOR_MATRIX_STACK_DEPTH */ - 1189, /* GL_POST_COLOR_MATRIX_RED_SCALE */ - 1185, /* GL_POST_COLOR_MATRIX_GREEN_SCALE */ - 1180, /* GL_POST_COLOR_MATRIX_BLUE_SCALE */ - 1176, /* GL_POST_COLOR_MATRIX_ALPHA_SCALE */ - 1187, /* GL_POST_COLOR_MATRIX_RED_BIAS */ - 1183, /* GL_POST_COLOR_MATRIX_GREEN_BIAS */ - 1178, /* GL_POST_COLOR_MATRIX_BLUE_BIAS */ - 1174, /* GL_POST_COLOR_MATRIX_ALPHA_BIAS */ - 1634, /* GL_TEXTURE_COLOR_TABLE_SGI */ - 1265, /* GL_PROXY_TEXTURE_COLOR_TABLE_SGI */ - 1636, /* GL_TEXTURE_COMPARE_FAIL_VALUE_ARB */ + 857, /* GL_MAX_COLOR_MATRIX_STACK_DEPTH */ + 1191, /* GL_POST_COLOR_MATRIX_RED_SCALE */ + 1187, /* GL_POST_COLOR_MATRIX_GREEN_SCALE */ + 1182, /* GL_POST_COLOR_MATRIX_BLUE_SCALE */ + 1178, /* GL_POST_COLOR_MATRIX_ALPHA_SCALE */ + 1189, /* GL_POST_COLOR_MATRIX_RED_BIAS */ + 1185, /* GL_POST_COLOR_MATRIX_GREEN_BIAS */ + 1180, /* GL_POST_COLOR_MATRIX_BLUE_BIAS */ + 1176, /* GL_POST_COLOR_MATRIX_ALPHA_BIAS */ + 1638, /* GL_TEXTURE_COLOR_TABLE_SGI */ + 1268, /* GL_PROXY_TEXTURE_COLOR_TABLE_SGI */ + 1640, /* GL_TEXTURE_COMPARE_FAIL_VALUE_ARB */ 80, /* GL_BLEND_DST_RGB */ 89, /* GL_BLEND_SRC_RGB */ 79, /* GL_BLEND_DST_ALPHA */ 88, /* GL_BLEND_SRC_ALPHA */ 201, /* GL_COLOR_TABLE */ - 1199, /* GL_POST_CONVOLUTION_COLOR_TABLE */ - 1182, /* GL_POST_COLOR_MATRIX_COLOR_TABLE */ - 1253, /* GL_PROXY_COLOR_TABLE */ - 1257, /* GL_PROXY_POST_CONVOLUTION_COLOR_TABLE */ - 1256, /* GL_PROXY_POST_COLOR_MATRIX_COLOR_TABLE */ + 1201, /* GL_POST_CONVOLUTION_COLOR_TABLE */ + 1184, /* GL_POST_COLOR_MATRIX_COLOR_TABLE */ + 1256, /* GL_PROXY_COLOR_TABLE */ + 1260, /* GL_PROXY_POST_CONVOLUTION_COLOR_TABLE */ + 1259, /* GL_PROXY_POST_COLOR_MATRIX_COLOR_TABLE */ 225, /* GL_COLOR_TABLE_SCALE */ 205, /* GL_COLOR_TABLE_BIAS */ 210, /* GL_COLOR_TABLE_FORMAT */ @@ -4371,380 +4379,380 @@ static const unsigned reduced_enums[1347] = 216, /* GL_COLOR_TABLE_INTENSITY_SIZE */ 71, /* GL_BGR */ 72, /* GL_BGRA */ - 869, /* GL_MAX_ELEMENTS_VERTICES */ - 868, /* GL_MAX_ELEMENTS_INDICES */ - 1689, /* GL_TEXTURE_INDEX_SIZE_EXT */ + 871, /* GL_MAX_ELEMENTS_VERTICES */ + 870, /* GL_MAX_ELEMENTS_INDICES */ + 1693, /* GL_TEXTURE_INDEX_SIZE_EXT */ 145, /* GL_CLIP_VOLUME_CLIPPING_HINT_EXT */ - 1146, /* GL_POINT_SIZE_MIN */ - 1142, /* GL_POINT_SIZE_MAX */ - 1136, /* GL_POINT_FADE_THRESHOLD_SIZE */ - 1132, /* GL_POINT_DISTANCE_ATTENUATION */ + 1148, /* GL_POINT_SIZE_MIN */ + 1144, /* GL_POINT_SIZE_MAX */ + 1138, /* GL_POINT_FADE_THRESHOLD_SIZE */ + 1134, /* GL_POINT_DISTANCE_ATTENUATION */ 127, /* GL_CLAMP_TO_BORDER */ 130, /* GL_CLAMP_TO_EDGE */ - 1710, /* GL_TEXTURE_MIN_LOD */ - 1708, /* GL_TEXTURE_MAX_LOD */ - 1613, /* GL_TEXTURE_BASE_LEVEL */ - 1707, /* GL_TEXTURE_MAX_LEVEL */ - 616, /* GL_IGNORE_BORDER_HP */ + 1714, /* GL_TEXTURE_MIN_LOD */ + 1712, /* GL_TEXTURE_MAX_LOD */ + 1617, /* GL_TEXTURE_BASE_LEVEL */ + 1711, /* GL_TEXTURE_MAX_LEVEL */ + 617, /* GL_IGNORE_BORDER_HP */ 276, /* GL_CONSTANT_BORDER_HP */ - 1324, /* GL_REPLICATE_BORDER_HP */ + 1328, /* GL_REPLICATE_BORDER_HP */ 282, /* GL_CONVOLUTION_BORDER_COLOR */ - 1043, /* GL_OCCLUSION_TEST_HP */ - 1044, /* GL_OCCLUSION_TEST_RESULT_HP */ - 688, /* GL_LINEAR_CLIPMAP_LINEAR_SGIX */ - 1628, /* GL_TEXTURE_CLIPMAP_CENTER_SGIX */ - 1630, /* GL_TEXTURE_CLIPMAP_FRAME_SGIX */ - 1632, /* GL_TEXTURE_CLIPMAP_OFFSET_SGIX */ - 1633, /* GL_TEXTURE_CLIPMAP_VIRTUAL_DEPTH_SGIX */ - 1631, /* GL_TEXTURE_CLIPMAP_LOD_OFFSET_SGIX */ - 1629, /* GL_TEXTURE_CLIPMAP_DEPTH_SGIX */ - 851, /* GL_MAX_CLIPMAP_DEPTH_SGIX */ - 852, /* GL_MAX_CLIPMAP_VIRTUAL_DEPTH_SGIX */ - 1209, /* GL_POST_TEXTURE_FILTER_BIAS_SGIX */ - 1211, /* GL_POST_TEXTURE_FILTER_SCALE_SGIX */ - 1208, /* GL_POST_TEXTURE_FILTER_BIAS_RANGE_SGIX */ - 1210, /* GL_POST_TEXTURE_FILTER_SCALE_RANGE_SGIX */ - 1697, /* GL_TEXTURE_LOD_BIAS_S_SGIX */ - 1698, /* GL_TEXTURE_LOD_BIAS_T_SGIX */ - 1696, /* GL_TEXTURE_LOD_BIAS_R_SGIX */ - 586, /* GL_GENERATE_MIPMAP */ - 587, /* GL_GENERATE_MIPMAP_HINT */ - 528, /* GL_FOG_OFFSET_SGIX */ - 529, /* GL_FOG_OFFSET_VALUE_SGIX */ - 1642, /* GL_TEXTURE_COMPARE_SGIX */ - 1641, /* GL_TEXTURE_COMPARE_OPERATOR_SGIX */ - 1693, /* GL_TEXTURE_LEQUAL_R_SGIX */ - 1685, /* GL_TEXTURE_GEQUAL_R_SGIX */ + 1045, /* GL_OCCLUSION_TEST_HP */ + 1046, /* GL_OCCLUSION_TEST_RESULT_HP */ + 690, /* GL_LINEAR_CLIPMAP_LINEAR_SGIX */ + 1632, /* GL_TEXTURE_CLIPMAP_CENTER_SGIX */ + 1634, /* GL_TEXTURE_CLIPMAP_FRAME_SGIX */ + 1636, /* GL_TEXTURE_CLIPMAP_OFFSET_SGIX */ + 1637, /* GL_TEXTURE_CLIPMAP_VIRTUAL_DEPTH_SGIX */ + 1635, /* GL_TEXTURE_CLIPMAP_LOD_OFFSET_SGIX */ + 1633, /* GL_TEXTURE_CLIPMAP_DEPTH_SGIX */ + 853, /* GL_MAX_CLIPMAP_DEPTH_SGIX */ + 854, /* GL_MAX_CLIPMAP_VIRTUAL_DEPTH_SGIX */ + 1211, /* GL_POST_TEXTURE_FILTER_BIAS_SGIX */ + 1213, /* GL_POST_TEXTURE_FILTER_SCALE_SGIX */ + 1210, /* GL_POST_TEXTURE_FILTER_BIAS_RANGE_SGIX */ + 1212, /* GL_POST_TEXTURE_FILTER_SCALE_RANGE_SGIX */ + 1701, /* GL_TEXTURE_LOD_BIAS_S_SGIX */ + 1702, /* GL_TEXTURE_LOD_BIAS_T_SGIX */ + 1700, /* GL_TEXTURE_LOD_BIAS_R_SGIX */ + 587, /* GL_GENERATE_MIPMAP */ + 588, /* GL_GENERATE_MIPMAP_HINT */ + 529, /* GL_FOG_OFFSET_SGIX */ + 530, /* GL_FOG_OFFSET_VALUE_SGIX */ + 1646, /* GL_TEXTURE_COMPARE_SGIX */ + 1645, /* GL_TEXTURE_COMPARE_OPERATOR_SGIX */ + 1697, /* GL_TEXTURE_LEQUAL_R_SGIX */ + 1689, /* GL_TEXTURE_GEQUAL_R_SGIX */ 359, /* GL_DEPTH_COMPONENT16 */ 362, /* GL_DEPTH_COMPONENT24 */ 365, /* GL_DEPTH_COMPONENT32 */ 306, /* GL_CULL_VERTEX_EXT */ 308, /* GL_CULL_VERTEX_OBJECT_POSITION_EXT */ 307, /* GL_CULL_VERTEX_EYE_POSITION_EXT */ - 1849, /* GL_WRAP_BORDER_SUN */ - 1635, /* GL_TEXTURE_COLOR_WRITEMASK_SGIS */ - 681, /* GL_LIGHT_MODEL_COLOR_CONTROL */ - 1423, /* GL_SINGLE_COLOR */ - 1409, /* GL_SEPARATE_SPECULAR_COLOR */ - 1418, /* GL_SHARED_TEXTURE_PALETTE_EXT */ - 539, /* GL_FRAMEBUFFER_ATTACHMENT_COLOR_ENCODING */ - 540, /* GL_FRAMEBUFFER_ATTACHMENT_COMPONENT_TYPE */ - 547, /* GL_FRAMEBUFFER_ATTACHMENT_RED_SIZE */ - 542, /* GL_FRAMEBUFFER_ATTACHMENT_GREEN_SIZE */ - 538, /* GL_FRAMEBUFFER_ATTACHMENT_BLUE_SIZE */ - 537, /* GL_FRAMEBUFFER_ATTACHMENT_ALPHA_SIZE */ - 541, /* GL_FRAMEBUFFER_ATTACHMENT_DEPTH_SIZE */ - 548, /* GL_FRAMEBUFFER_ATTACHMENT_STENCIL_SIZE */ - 559, /* GL_FRAMEBUFFER_DEFAULT */ - 572, /* GL_FRAMEBUFFER_UNDEFINED */ + 1853, /* GL_WRAP_BORDER_SUN */ + 1639, /* GL_TEXTURE_COLOR_WRITEMASK_SGIS */ + 683, /* GL_LIGHT_MODEL_COLOR_CONTROL */ + 1427, /* GL_SINGLE_COLOR */ + 1413, /* GL_SEPARATE_SPECULAR_COLOR */ + 1422, /* GL_SHARED_TEXTURE_PALETTE_EXT */ + 540, /* GL_FRAMEBUFFER_ATTACHMENT_COLOR_ENCODING */ + 541, /* GL_FRAMEBUFFER_ATTACHMENT_COMPONENT_TYPE */ + 548, /* GL_FRAMEBUFFER_ATTACHMENT_RED_SIZE */ + 543, /* GL_FRAMEBUFFER_ATTACHMENT_GREEN_SIZE */ + 539, /* GL_FRAMEBUFFER_ATTACHMENT_BLUE_SIZE */ + 538, /* GL_FRAMEBUFFER_ATTACHMENT_ALPHA_SIZE */ + 542, /* GL_FRAMEBUFFER_ATTACHMENT_DEPTH_SIZE */ + 549, /* GL_FRAMEBUFFER_ATTACHMENT_STENCIL_SIZE */ + 560, /* GL_FRAMEBUFFER_DEFAULT */ + 573, /* GL_FRAMEBUFFER_UNDEFINED */ 372, /* GL_DEPTH_STENCIL_ATTACHMENT */ - 622, /* GL_INDEX */ - 1759, /* GL_UNSIGNED_BYTE_2_3_3_REV */ - 1774, /* GL_UNSIGNED_SHORT_5_6_5 */ - 1775, /* GL_UNSIGNED_SHORT_5_6_5_REV */ - 1772, /* GL_UNSIGNED_SHORT_4_4_4_4_REV */ - 1770, /* GL_UNSIGNED_SHORT_1_5_5_5_REV */ - 1767, /* GL_UNSIGNED_INT_8_8_8_8_REV */ - 1765, /* GL_UNSIGNED_INT_2_10_10_10_REV */ - 1705, /* GL_TEXTURE_MAX_CLAMP_S_SGIX */ - 1706, /* GL_TEXTURE_MAX_CLAMP_T_SGIX */ - 1704, /* GL_TEXTURE_MAX_CLAMP_R_SGIX */ - 944, /* GL_MIRRORED_REPEAT */ - 1365, /* GL_RGB_S3TC */ - 1340, /* GL_RGB4_S3TC */ - 1363, /* GL_RGBA_S3TC */ - 1357, /* GL_RGBA4_S3TC */ - 1361, /* GL_RGBA_DXT5_S3TC */ - 1355, /* GL_RGBA4_DXT5_S3TC */ + 623, /* GL_INDEX */ + 1763, /* GL_UNSIGNED_BYTE_2_3_3_REV */ + 1778, /* GL_UNSIGNED_SHORT_5_6_5 */ + 1779, /* GL_UNSIGNED_SHORT_5_6_5_REV */ + 1776, /* GL_UNSIGNED_SHORT_4_4_4_4_REV */ + 1774, /* GL_UNSIGNED_SHORT_1_5_5_5_REV */ + 1771, /* GL_UNSIGNED_INT_8_8_8_8_REV */ + 1769, /* GL_UNSIGNED_INT_2_10_10_10_REV */ + 1709, /* GL_TEXTURE_MAX_CLAMP_S_SGIX */ + 1710, /* GL_TEXTURE_MAX_CLAMP_T_SGIX */ + 1708, /* GL_TEXTURE_MAX_CLAMP_R_SGIX */ + 946, /* GL_MIRRORED_REPEAT */ + 1369, /* GL_RGB_S3TC */ + 1344, /* GL_RGB4_S3TC */ + 1367, /* GL_RGBA_S3TC */ + 1361, /* GL_RGBA4_S3TC */ + 1365, /* GL_RGBA_DXT5_S3TC */ + 1359, /* GL_RGBA4_DXT5_S3TC */ 264, /* GL_COMPRESSED_RGB_S3TC_DXT1_EXT */ 259, /* GL_COMPRESSED_RGBA_S3TC_DXT1_EXT */ 260, /* GL_COMPRESSED_RGBA_S3TC_DXT3_EXT */ 261, /* GL_COMPRESSED_RGBA_S3TC_DXT5_EXT */ - 1005, /* GL_NEAREST_CLIPMAP_NEAREST_SGIX */ - 1004, /* GL_NEAREST_CLIPMAP_LINEAR_SGIX */ - 689, /* GL_LINEAR_CLIPMAP_NEAREST_SGIX */ - 515, /* GL_FOG_COORDINATE_SOURCE */ - 507, /* GL_FOG_COORD */ - 531, /* GL_FRAGMENT_DEPTH */ + 1007, /* GL_NEAREST_CLIPMAP_NEAREST_SGIX */ + 1006, /* GL_NEAREST_CLIPMAP_LINEAR_SGIX */ + 691, /* GL_LINEAR_CLIPMAP_NEAREST_SGIX */ + 516, /* GL_FOG_COORDINATE_SOURCE */ + 508, /* GL_FOG_COORD */ + 532, /* GL_FRAGMENT_DEPTH */ 312, /* GL_CURRENT_FOG_COORD */ - 514, /* GL_FOG_COORDINATE_ARRAY_TYPE */ - 513, /* GL_FOG_COORDINATE_ARRAY_STRIDE */ - 512, /* GL_FOG_COORDINATE_ARRAY_POINTER */ - 509, /* GL_FOG_COORDINATE_ARRAY */ + 515, /* GL_FOG_COORDINATE_ARRAY_TYPE */ + 514, /* GL_FOG_COORDINATE_ARRAY_STRIDE */ + 513, /* GL_FOG_COORDINATE_ARRAY_POINTER */ + 510, /* GL_FOG_COORDINATE_ARRAY */ 199, /* GL_COLOR_SUM */ 332, /* GL_CURRENT_SECONDARY_COLOR */ - 1402, /* GL_SECONDARY_COLOR_ARRAY_SIZE */ - 1404, /* GL_SECONDARY_COLOR_ARRAY_TYPE */ - 1403, /* GL_SECONDARY_COLOR_ARRAY_STRIDE */ - 1401, /* GL_SECONDARY_COLOR_ARRAY_POINTER */ - 1398, /* GL_SECONDARY_COLOR_ARRAY */ + 1406, /* GL_SECONDARY_COLOR_ARRAY_SIZE */ + 1408, /* GL_SECONDARY_COLOR_ARRAY_TYPE */ + 1407, /* GL_SECONDARY_COLOR_ARRAY_STRIDE */ + 1405, /* GL_SECONDARY_COLOR_ARRAY_POINTER */ + 1402, /* GL_SECONDARY_COLOR_ARRAY */ 330, /* GL_CURRENT_RASTER_SECONDARY_COLOR */ 28, /* GL_ALIASED_POINT_SIZE_RANGE */ 27, /* GL_ALIASED_LINE_WIDTH_RANGE */ - 1542, /* GL_TEXTURE0 */ - 1544, /* GL_TEXTURE1 */ - 1566, /* GL_TEXTURE2 */ - 1588, /* GL_TEXTURE3 */ - 1594, /* GL_TEXTURE4 */ - 1596, /* GL_TEXTURE5 */ - 1598, /* GL_TEXTURE6 */ - 1600, /* GL_TEXTURE7 */ - 1602, /* GL_TEXTURE8 */ - 1604, /* GL_TEXTURE9 */ - 1545, /* GL_TEXTURE10 */ - 1547, /* GL_TEXTURE11 */ - 1549, /* GL_TEXTURE12 */ - 1551, /* GL_TEXTURE13 */ - 1553, /* GL_TEXTURE14 */ - 1555, /* GL_TEXTURE15 */ - 1557, /* GL_TEXTURE16 */ - 1559, /* GL_TEXTURE17 */ - 1561, /* GL_TEXTURE18 */ - 1563, /* GL_TEXTURE19 */ - 1567, /* GL_TEXTURE20 */ - 1569, /* GL_TEXTURE21 */ - 1571, /* GL_TEXTURE22 */ - 1573, /* GL_TEXTURE23 */ - 1575, /* GL_TEXTURE24 */ - 1577, /* GL_TEXTURE25 */ - 1579, /* GL_TEXTURE26 */ - 1581, /* GL_TEXTURE27 */ - 1583, /* GL_TEXTURE28 */ - 1585, /* GL_TEXTURE29 */ - 1589, /* GL_TEXTURE30 */ - 1591, /* GL_TEXTURE31 */ + 1546, /* GL_TEXTURE0 */ + 1548, /* GL_TEXTURE1 */ + 1570, /* GL_TEXTURE2 */ + 1592, /* GL_TEXTURE3 */ + 1598, /* GL_TEXTURE4 */ + 1600, /* GL_TEXTURE5 */ + 1602, /* GL_TEXTURE6 */ + 1604, /* GL_TEXTURE7 */ + 1606, /* GL_TEXTURE8 */ + 1608, /* GL_TEXTURE9 */ + 1549, /* GL_TEXTURE10 */ + 1551, /* GL_TEXTURE11 */ + 1553, /* GL_TEXTURE12 */ + 1555, /* GL_TEXTURE13 */ + 1557, /* GL_TEXTURE14 */ + 1559, /* GL_TEXTURE15 */ + 1561, /* GL_TEXTURE16 */ + 1563, /* GL_TEXTURE17 */ + 1565, /* GL_TEXTURE18 */ + 1567, /* GL_TEXTURE19 */ + 1571, /* GL_TEXTURE20 */ + 1573, /* GL_TEXTURE21 */ + 1575, /* GL_TEXTURE22 */ + 1577, /* GL_TEXTURE23 */ + 1579, /* GL_TEXTURE24 */ + 1581, /* GL_TEXTURE25 */ + 1583, /* GL_TEXTURE26 */ + 1585, /* GL_TEXTURE27 */ + 1587, /* GL_TEXTURE28 */ + 1589, /* GL_TEXTURE29 */ + 1593, /* GL_TEXTURE30 */ + 1595, /* GL_TEXTURE31 */ 18, /* GL_ACTIVE_TEXTURE */ 133, /* GL_CLIENT_ACTIVE_TEXTURE */ - 922, /* GL_MAX_TEXTURE_UNITS */ - 1737, /* GL_TRANSPOSE_MODELVIEW_MATRIX */ - 1740, /* GL_TRANSPOSE_PROJECTION_MATRIX */ - 1742, /* GL_TRANSPOSE_TEXTURE_MATRIX */ - 1734, /* GL_TRANSPOSE_COLOR_MATRIX */ - 1524, /* GL_SUBTRACT */ - 909, /* GL_MAX_RENDERBUFFER_SIZE_EXT */ + 924, /* GL_MAX_TEXTURE_UNITS */ + 1741, /* GL_TRANSPOSE_MODELVIEW_MATRIX */ + 1744, /* GL_TRANSPOSE_PROJECTION_MATRIX */ + 1746, /* GL_TRANSPOSE_TEXTURE_MATRIX */ + 1738, /* GL_TRANSPOSE_COLOR_MATRIX */ + 1528, /* GL_SUBTRACT */ + 911, /* GL_MAX_RENDERBUFFER_SIZE_EXT */ 247, /* GL_COMPRESSED_ALPHA */ 251, /* GL_COMPRESSED_LUMINANCE */ 252, /* GL_COMPRESSED_LUMINANCE_ALPHA */ 249, /* GL_COMPRESSED_INTENSITY */ 255, /* GL_COMPRESSED_RGB */ 256, /* GL_COMPRESSED_RGBA */ - 1649, /* GL_TEXTURE_COMPRESSION_HINT */ - 1714, /* GL_TEXTURE_RECTANGLE_ARB */ - 1621, /* GL_TEXTURE_BINDING_RECTANGLE_ARB */ - 1268, /* GL_PROXY_TEXTURE_RECTANGLE_ARB */ - 907, /* GL_MAX_RECTANGLE_TEXTURE_SIZE_ARB */ + 1653, /* GL_TEXTURE_COMPRESSION_HINT */ + 1718, /* GL_TEXTURE_RECTANGLE_ARB */ + 1625, /* GL_TEXTURE_BINDING_RECTANGLE_ARB */ + 1271, /* GL_PROXY_TEXTURE_RECTANGLE_ARB */ + 909, /* GL_MAX_RECTANGLE_TEXTURE_SIZE_ARB */ 371, /* GL_DEPTH_STENCIL */ - 1763, /* GL_UNSIGNED_INT_24_8 */ - 918, /* GL_MAX_TEXTURE_LOD_BIAS */ - 1703, /* GL_TEXTURE_MAX_ANISOTROPY_EXT */ - 919, /* GL_MAX_TEXTURE_MAX_ANISOTROPY_EXT */ - 1679, /* GL_TEXTURE_FILTER_CONTROL */ - 1694, /* GL_TEXTURE_LOD_BIAS */ + 1767, /* GL_UNSIGNED_INT_24_8 */ + 920, /* GL_MAX_TEXTURE_LOD_BIAS */ + 1707, /* GL_TEXTURE_MAX_ANISOTROPY_EXT */ + 921, /* GL_MAX_TEXTURE_MAX_ANISOTROPY_EXT */ + 1683, /* GL_TEXTURE_FILTER_CONTROL */ + 1698, /* GL_TEXTURE_LOD_BIAS */ 232, /* GL_COMBINE4 */ - 912, /* GL_MAX_SHININESS_NV */ - 913, /* GL_MAX_SPOT_EXPONENT_NV */ - 620, /* GL_INCR_WRAP */ + 914, /* GL_MAX_SHININESS_NV */ + 915, /* GL_MAX_SPOT_EXPONENT_NV */ + 621, /* GL_INCR_WRAP */ 343, /* GL_DECR_WRAP */ - 964, /* GL_MODELVIEW1_ARB */ - 1020, /* GL_NORMAL_MAP */ - 1299, /* GL_REFLECTION_MAP */ - 1658, /* GL_TEXTURE_CUBE_MAP */ - 1619, /* GL_TEXTURE_BINDING_CUBE_MAP */ - 1666, /* GL_TEXTURE_CUBE_MAP_POSITIVE_X */ - 1660, /* GL_TEXTURE_CUBE_MAP_NEGATIVE_X */ - 1668, /* GL_TEXTURE_CUBE_MAP_POSITIVE_Y */ - 1662, /* GL_TEXTURE_CUBE_MAP_NEGATIVE_Y */ - 1670, /* GL_TEXTURE_CUBE_MAP_POSITIVE_Z */ - 1664, /* GL_TEXTURE_CUBE_MAP_NEGATIVE_Z */ - 1266, /* GL_PROXY_TEXTURE_CUBE_MAP */ - 863, /* GL_MAX_CUBE_MAP_TEXTURE_SIZE */ - 999, /* GL_MULTISAMPLE_FILTER_HINT_NV */ - 523, /* GL_FOG_DISTANCE_MODE_NV */ + 966, /* GL_MODELVIEW1_ARB */ + 1022, /* GL_NORMAL_MAP */ + 1303, /* GL_REFLECTION_MAP */ + 1662, /* GL_TEXTURE_CUBE_MAP */ + 1623, /* GL_TEXTURE_BINDING_CUBE_MAP */ + 1670, /* GL_TEXTURE_CUBE_MAP_POSITIVE_X */ + 1664, /* GL_TEXTURE_CUBE_MAP_NEGATIVE_X */ + 1672, /* GL_TEXTURE_CUBE_MAP_POSITIVE_Y */ + 1666, /* GL_TEXTURE_CUBE_MAP_NEGATIVE_Y */ + 1674, /* GL_TEXTURE_CUBE_MAP_POSITIVE_Z */ + 1668, /* GL_TEXTURE_CUBE_MAP_NEGATIVE_Z */ + 1269, /* GL_PROXY_TEXTURE_CUBE_MAP */ + 865, /* GL_MAX_CUBE_MAP_TEXTURE_SIZE */ + 1001, /* GL_MULTISAMPLE_FILTER_HINT_NV */ + 524, /* GL_FOG_DISTANCE_MODE_NV */ 475, /* GL_EYE_RADIAL_NV */ 474, /* GL_EYE_PLANE_ABSOLUTE_NV */ 231, /* GL_COMBINE */ 238, /* GL_COMBINE_RGB */ 233, /* GL_COMBINE_ALPHA */ - 1366, /* GL_RGB_SCALE */ + 1370, /* GL_RGB_SCALE */ 24, /* GL_ADD_SIGNED */ - 648, /* GL_INTERPOLATE */ + 649, /* GL_INTERPOLATE */ 271, /* GL_CONSTANT */ - 1215, /* GL_PRIMARY_COLOR */ - 1212, /* GL_PREVIOUS */ - 1438, /* GL_SOURCE0_RGB */ - 1444, /* GL_SOURCE1_RGB */ - 1450, /* GL_SOURCE2_RGB */ - 1454, /* GL_SOURCE3_RGB_NV */ - 1435, /* GL_SOURCE0_ALPHA */ - 1441, /* GL_SOURCE1_ALPHA */ - 1447, /* GL_SOURCE2_ALPHA */ - 1453, /* GL_SOURCE3_ALPHA_NV */ - 1057, /* GL_OPERAND0_RGB */ - 1063, /* GL_OPERAND1_RGB */ - 1069, /* GL_OPERAND2_RGB */ - 1073, /* GL_OPERAND3_RGB_NV */ - 1054, /* GL_OPERAND0_ALPHA */ - 1060, /* GL_OPERAND1_ALPHA */ - 1066, /* GL_OPERAND2_ALPHA */ - 1072, /* GL_OPERAND3_ALPHA_NV */ - 1787, /* GL_VERTEX_ARRAY_BINDING */ - 1712, /* GL_TEXTURE_RANGE_LENGTH_APPLE */ - 1713, /* GL_TEXTURE_RANGE_POINTER_APPLE */ - 1853, /* GL_YCBCR_422_APPLE */ - 1776, /* GL_UNSIGNED_SHORT_8_8_APPLE */ - 1778, /* GL_UNSIGNED_SHORT_8_8_REV_APPLE */ - 1721, /* GL_TEXTURE_STORAGE_HINT_APPLE */ - 1515, /* GL_STORAGE_PRIVATE_APPLE */ - 1514, /* GL_STORAGE_CACHED_APPLE */ - 1516, /* GL_STORAGE_SHARED_APPLE */ - 1425, /* GL_SLICE_ACCUM_SUN */ - 1274, /* GL_QUAD_MESH_SUN */ - 1746, /* GL_TRIANGLE_MESH_SUN */ - 1826, /* GL_VERTEX_PROGRAM_ARB */ - 1837, /* GL_VERTEX_STATE_PROGRAM_NV */ - 1813, /* GL_VERTEX_ATTRIB_ARRAY_ENABLED */ - 1819, /* GL_VERTEX_ATTRIB_ARRAY_SIZE */ - 1821, /* GL_VERTEX_ATTRIB_ARRAY_STRIDE */ - 1823, /* GL_VERTEX_ATTRIB_ARRAY_TYPE */ + 1217, /* GL_PRIMARY_COLOR */ + 1214, /* GL_PREVIOUS */ + 1442, /* GL_SOURCE0_RGB */ + 1448, /* GL_SOURCE1_RGB */ + 1454, /* GL_SOURCE2_RGB */ + 1458, /* GL_SOURCE3_RGB_NV */ + 1439, /* GL_SOURCE0_ALPHA */ + 1445, /* GL_SOURCE1_ALPHA */ + 1451, /* GL_SOURCE2_ALPHA */ + 1457, /* GL_SOURCE3_ALPHA_NV */ + 1059, /* GL_OPERAND0_RGB */ + 1065, /* GL_OPERAND1_RGB */ + 1071, /* GL_OPERAND2_RGB */ + 1075, /* GL_OPERAND3_RGB_NV */ + 1056, /* GL_OPERAND0_ALPHA */ + 1062, /* GL_OPERAND1_ALPHA */ + 1068, /* GL_OPERAND2_ALPHA */ + 1074, /* GL_OPERAND3_ALPHA_NV */ + 1791, /* GL_VERTEX_ARRAY_BINDING */ + 1716, /* GL_TEXTURE_RANGE_LENGTH_APPLE */ + 1717, /* GL_TEXTURE_RANGE_POINTER_APPLE */ + 1857, /* GL_YCBCR_422_APPLE */ + 1780, /* GL_UNSIGNED_SHORT_8_8_APPLE */ + 1782, /* GL_UNSIGNED_SHORT_8_8_REV_APPLE */ + 1725, /* GL_TEXTURE_STORAGE_HINT_APPLE */ + 1519, /* GL_STORAGE_PRIVATE_APPLE */ + 1518, /* GL_STORAGE_CACHED_APPLE */ + 1520, /* GL_STORAGE_SHARED_APPLE */ + 1429, /* GL_SLICE_ACCUM_SUN */ + 1278, /* GL_QUAD_MESH_SUN */ + 1750, /* GL_TRIANGLE_MESH_SUN */ + 1830, /* GL_VERTEX_PROGRAM_ARB */ + 1841, /* GL_VERTEX_STATE_PROGRAM_NV */ + 1817, /* GL_VERTEX_ATTRIB_ARRAY_ENABLED */ + 1823, /* GL_VERTEX_ATTRIB_ARRAY_SIZE */ + 1825, /* GL_VERTEX_ATTRIB_ARRAY_STRIDE */ + 1827, /* GL_VERTEX_ATTRIB_ARRAY_TYPE */ 334, /* GL_CURRENT_VERTEX_ATTRIB */ - 1228, /* GL_PROGRAM_LENGTH_ARB */ - 1242, /* GL_PROGRAM_STRING_ARB */ - 986, /* GL_MODELVIEW_PROJECTION_NV */ - 615, /* GL_IDENTITY_NV */ - 662, /* GL_INVERSE_NV */ - 1739, /* GL_TRANSPOSE_NV */ - 663, /* GL_INVERSE_TRANSPOSE_NV */ - 893, /* GL_MAX_PROGRAM_MATRIX_STACK_DEPTH_ARB */ - 892, /* GL_MAX_PROGRAM_MATRICES_ARB */ - 800, /* GL_MATRIX0_NV */ - 812, /* GL_MATRIX1_NV */ - 824, /* GL_MATRIX2_NV */ - 828, /* GL_MATRIX3_NV */ - 830, /* GL_MATRIX4_NV */ - 832, /* GL_MATRIX5_NV */ - 834, /* GL_MATRIX6_NV */ - 836, /* GL_MATRIX7_NV */ + 1230, /* GL_PROGRAM_LENGTH_ARB */ + 1244, /* GL_PROGRAM_STRING_ARB */ + 988, /* GL_MODELVIEW_PROJECTION_NV */ + 616, /* GL_IDENTITY_NV */ + 663, /* GL_INVERSE_NV */ + 1743, /* GL_TRANSPOSE_NV */ + 664, /* GL_INVERSE_TRANSPOSE_NV */ + 895, /* GL_MAX_PROGRAM_MATRIX_STACK_DEPTH_ARB */ + 894, /* GL_MAX_PROGRAM_MATRICES_ARB */ + 802, /* GL_MATRIX0_NV */ + 814, /* GL_MATRIX1_NV */ + 826, /* GL_MATRIX2_NV */ + 830, /* GL_MATRIX3_NV */ + 832, /* GL_MATRIX4_NV */ + 834, /* GL_MATRIX5_NV */ + 836, /* GL_MATRIX6_NV */ + 838, /* GL_MATRIX7_NV */ 318, /* GL_CURRENT_MATRIX_STACK_DEPTH_ARB */ 315, /* GL_CURRENT_MATRIX_ARB */ - 1829, /* GL_VERTEX_PROGRAM_POINT_SIZE */ - 1832, /* GL_VERTEX_PROGRAM_TWO_SIDE */ - 1240, /* GL_PROGRAM_PARAMETER_NV */ - 1817, /* GL_VERTEX_ATTRIB_ARRAY_POINTER */ - 1244, /* GL_PROGRAM_TARGET_NV */ - 1241, /* GL_PROGRAM_RESIDENT_NV */ - 1731, /* GL_TRACK_MATRIX_NV */ - 1732, /* GL_TRACK_MATRIX_TRANSFORM_NV */ - 1827, /* GL_VERTEX_PROGRAM_BINDING_NV */ - 1222, /* GL_PROGRAM_ERROR_POSITION_ARB */ + 1833, /* GL_VERTEX_PROGRAM_POINT_SIZE */ + 1836, /* GL_VERTEX_PROGRAM_TWO_SIDE */ + 1242, /* GL_PROGRAM_PARAMETER_NV */ + 1821, /* GL_VERTEX_ATTRIB_ARRAY_POINTER */ + 1246, /* GL_PROGRAM_TARGET_NV */ + 1243, /* GL_PROGRAM_RESIDENT_NV */ + 1735, /* GL_TRACK_MATRIX_NV */ + 1736, /* GL_TRACK_MATRIX_TRANSFORM_NV */ + 1831, /* GL_VERTEX_PROGRAM_BINDING_NV */ + 1224, /* GL_PROGRAM_ERROR_POSITION_ARB */ 355, /* GL_DEPTH_CLAMP */ - 1795, /* GL_VERTEX_ATTRIB_ARRAY0_NV */ - 1802, /* GL_VERTEX_ATTRIB_ARRAY1_NV */ - 1803, /* GL_VERTEX_ATTRIB_ARRAY2_NV */ - 1804, /* GL_VERTEX_ATTRIB_ARRAY3_NV */ - 1805, /* GL_VERTEX_ATTRIB_ARRAY4_NV */ - 1806, /* GL_VERTEX_ATTRIB_ARRAY5_NV */ - 1807, /* GL_VERTEX_ATTRIB_ARRAY6_NV */ - 1808, /* GL_VERTEX_ATTRIB_ARRAY7_NV */ - 1809, /* GL_VERTEX_ATTRIB_ARRAY8_NV */ - 1810, /* GL_VERTEX_ATTRIB_ARRAY9_NV */ - 1796, /* GL_VERTEX_ATTRIB_ARRAY10_NV */ - 1797, /* GL_VERTEX_ATTRIB_ARRAY11_NV */ - 1798, /* GL_VERTEX_ATTRIB_ARRAY12_NV */ - 1799, /* GL_VERTEX_ATTRIB_ARRAY13_NV */ - 1800, /* GL_VERTEX_ATTRIB_ARRAY14_NV */ - 1801, /* GL_VERTEX_ATTRIB_ARRAY15_NV */ - 748, /* GL_MAP1_VERTEX_ATTRIB0_4_NV */ - 755, /* GL_MAP1_VERTEX_ATTRIB1_4_NV */ - 756, /* GL_MAP1_VERTEX_ATTRIB2_4_NV */ - 757, /* GL_MAP1_VERTEX_ATTRIB3_4_NV */ - 758, /* GL_MAP1_VERTEX_ATTRIB4_4_NV */ - 759, /* GL_MAP1_VERTEX_ATTRIB5_4_NV */ - 760, /* GL_MAP1_VERTEX_ATTRIB6_4_NV */ - 761, /* GL_MAP1_VERTEX_ATTRIB7_4_NV */ - 762, /* GL_MAP1_VERTEX_ATTRIB8_4_NV */ - 763, /* GL_MAP1_VERTEX_ATTRIB9_4_NV */ - 749, /* GL_MAP1_VERTEX_ATTRIB10_4_NV */ - 750, /* GL_MAP1_VERTEX_ATTRIB11_4_NV */ - 751, /* GL_MAP1_VERTEX_ATTRIB12_4_NV */ - 752, /* GL_MAP1_VERTEX_ATTRIB13_4_NV */ - 753, /* GL_MAP1_VERTEX_ATTRIB14_4_NV */ - 754, /* GL_MAP1_VERTEX_ATTRIB15_4_NV */ - 775, /* GL_MAP2_VERTEX_ATTRIB0_4_NV */ - 782, /* GL_MAP2_VERTEX_ATTRIB1_4_NV */ - 783, /* GL_MAP2_VERTEX_ATTRIB2_4_NV */ - 784, /* GL_MAP2_VERTEX_ATTRIB3_4_NV */ - 785, /* GL_MAP2_VERTEX_ATTRIB4_4_NV */ - 786, /* GL_MAP2_VERTEX_ATTRIB5_4_NV */ - 787, /* GL_MAP2_VERTEX_ATTRIB6_4_NV */ - 1221, /* GL_PROGRAM_BINDING_ARB */ - 789, /* GL_MAP2_VERTEX_ATTRIB8_4_NV */ - 790, /* GL_MAP2_VERTEX_ATTRIB9_4_NV */ - 776, /* GL_MAP2_VERTEX_ATTRIB10_4_NV */ - 777, /* GL_MAP2_VERTEX_ATTRIB11_4_NV */ - 778, /* GL_MAP2_VERTEX_ATTRIB12_4_NV */ - 779, /* GL_MAP2_VERTEX_ATTRIB13_4_NV */ - 780, /* GL_MAP2_VERTEX_ATTRIB14_4_NV */ - 781, /* GL_MAP2_VERTEX_ATTRIB15_4_NV */ - 1647, /* GL_TEXTURE_COMPRESSED_IMAGE_SIZE */ - 1644, /* GL_TEXTURE_COMPRESSED */ - 1025, /* GL_NUM_COMPRESSED_TEXTURE_FORMATS */ + 1799, /* GL_VERTEX_ATTRIB_ARRAY0_NV */ + 1806, /* GL_VERTEX_ATTRIB_ARRAY1_NV */ + 1807, /* GL_VERTEX_ATTRIB_ARRAY2_NV */ + 1808, /* GL_VERTEX_ATTRIB_ARRAY3_NV */ + 1809, /* GL_VERTEX_ATTRIB_ARRAY4_NV */ + 1810, /* GL_VERTEX_ATTRIB_ARRAY5_NV */ + 1811, /* GL_VERTEX_ATTRIB_ARRAY6_NV */ + 1812, /* GL_VERTEX_ATTRIB_ARRAY7_NV */ + 1813, /* GL_VERTEX_ATTRIB_ARRAY8_NV */ + 1814, /* GL_VERTEX_ATTRIB_ARRAY9_NV */ + 1800, /* GL_VERTEX_ATTRIB_ARRAY10_NV */ + 1801, /* GL_VERTEX_ATTRIB_ARRAY11_NV */ + 1802, /* GL_VERTEX_ATTRIB_ARRAY12_NV */ + 1803, /* GL_VERTEX_ATTRIB_ARRAY13_NV */ + 1804, /* GL_VERTEX_ATTRIB_ARRAY14_NV */ + 1805, /* GL_VERTEX_ATTRIB_ARRAY15_NV */ + 750, /* GL_MAP1_VERTEX_ATTRIB0_4_NV */ + 757, /* GL_MAP1_VERTEX_ATTRIB1_4_NV */ + 758, /* GL_MAP1_VERTEX_ATTRIB2_4_NV */ + 759, /* GL_MAP1_VERTEX_ATTRIB3_4_NV */ + 760, /* GL_MAP1_VERTEX_ATTRIB4_4_NV */ + 761, /* GL_MAP1_VERTEX_ATTRIB5_4_NV */ + 762, /* GL_MAP1_VERTEX_ATTRIB6_4_NV */ + 763, /* GL_MAP1_VERTEX_ATTRIB7_4_NV */ + 764, /* GL_MAP1_VERTEX_ATTRIB8_4_NV */ + 765, /* GL_MAP1_VERTEX_ATTRIB9_4_NV */ + 751, /* GL_MAP1_VERTEX_ATTRIB10_4_NV */ + 752, /* GL_MAP1_VERTEX_ATTRIB11_4_NV */ + 753, /* GL_MAP1_VERTEX_ATTRIB12_4_NV */ + 754, /* GL_MAP1_VERTEX_ATTRIB13_4_NV */ + 755, /* GL_MAP1_VERTEX_ATTRIB14_4_NV */ + 756, /* GL_MAP1_VERTEX_ATTRIB15_4_NV */ + 777, /* GL_MAP2_VERTEX_ATTRIB0_4_NV */ + 784, /* GL_MAP2_VERTEX_ATTRIB1_4_NV */ + 785, /* GL_MAP2_VERTEX_ATTRIB2_4_NV */ + 786, /* GL_MAP2_VERTEX_ATTRIB3_4_NV */ + 787, /* GL_MAP2_VERTEX_ATTRIB4_4_NV */ + 788, /* GL_MAP2_VERTEX_ATTRIB5_4_NV */ + 789, /* GL_MAP2_VERTEX_ATTRIB6_4_NV */ + 1223, /* GL_PROGRAM_BINDING_ARB */ + 791, /* GL_MAP2_VERTEX_ATTRIB8_4_NV */ + 792, /* GL_MAP2_VERTEX_ATTRIB9_4_NV */ + 778, /* GL_MAP2_VERTEX_ATTRIB10_4_NV */ + 779, /* GL_MAP2_VERTEX_ATTRIB11_4_NV */ + 780, /* GL_MAP2_VERTEX_ATTRIB12_4_NV */ + 781, /* GL_MAP2_VERTEX_ATTRIB13_4_NV */ + 782, /* GL_MAP2_VERTEX_ATTRIB14_4_NV */ + 783, /* GL_MAP2_VERTEX_ATTRIB15_4_NV */ + 1651, /* GL_TEXTURE_COMPRESSED_IMAGE_SIZE */ + 1648, /* GL_TEXTURE_COMPRESSED */ + 1027, /* GL_NUM_COMPRESSED_TEXTURE_FORMATS */ 269, /* GL_COMPRESSED_TEXTURE_FORMATS */ - 934, /* GL_MAX_VERTEX_UNITS_ARB */ + 936, /* GL_MAX_VERTEX_UNITS_ARB */ 22, /* GL_ACTIVE_VERTEX_UNITS_ARB */ - 1848, /* GL_WEIGHT_SUM_UNITY_ARB */ - 1825, /* GL_VERTEX_BLEND_ARB */ + 1852, /* GL_WEIGHT_SUM_UNITY_ARB */ + 1829, /* GL_VERTEX_BLEND_ARB */ 336, /* GL_CURRENT_WEIGHT_ARB */ - 1847, /* GL_WEIGHT_ARRAY_TYPE_ARB */ - 1846, /* GL_WEIGHT_ARRAY_STRIDE_ARB */ - 1845, /* GL_WEIGHT_ARRAY_SIZE_ARB */ - 1844, /* GL_WEIGHT_ARRAY_POINTER_ARB */ - 1841, /* GL_WEIGHT_ARRAY_ARB */ + 1851, /* GL_WEIGHT_ARRAY_TYPE_ARB */ + 1850, /* GL_WEIGHT_ARRAY_STRIDE_ARB */ + 1849, /* GL_WEIGHT_ARRAY_SIZE_ARB */ + 1848, /* GL_WEIGHT_ARRAY_POINTER_ARB */ + 1845, /* GL_WEIGHT_ARRAY_ARB */ 384, /* GL_DOT3_RGB */ 385, /* GL_DOT3_RGBA */ 263, /* GL_COMPRESSED_RGB_FXT1_3DFX */ 258, /* GL_COMPRESSED_RGBA_FXT1_3DFX */ - 994, /* GL_MULTISAMPLE_3DFX */ - 1387, /* GL_SAMPLE_BUFFERS_3DFX */ - 1378, /* GL_SAMPLES_3DFX */ - 975, /* GL_MODELVIEW2_ARB */ - 978, /* GL_MODELVIEW3_ARB */ - 979, /* GL_MODELVIEW4_ARB */ - 980, /* GL_MODELVIEW5_ARB */ - 981, /* GL_MODELVIEW6_ARB */ - 982, /* GL_MODELVIEW7_ARB */ - 983, /* GL_MODELVIEW8_ARB */ - 984, /* GL_MODELVIEW9_ARB */ - 954, /* GL_MODELVIEW10_ARB */ - 955, /* GL_MODELVIEW11_ARB */ - 956, /* GL_MODELVIEW12_ARB */ - 957, /* GL_MODELVIEW13_ARB */ - 958, /* GL_MODELVIEW14_ARB */ - 959, /* GL_MODELVIEW15_ARB */ - 960, /* GL_MODELVIEW16_ARB */ - 961, /* GL_MODELVIEW17_ARB */ - 962, /* GL_MODELVIEW18_ARB */ - 963, /* GL_MODELVIEW19_ARB */ - 965, /* GL_MODELVIEW20_ARB */ - 966, /* GL_MODELVIEW21_ARB */ - 967, /* GL_MODELVIEW22_ARB */ - 968, /* GL_MODELVIEW23_ARB */ - 969, /* GL_MODELVIEW24_ARB */ - 970, /* GL_MODELVIEW25_ARB */ - 971, /* GL_MODELVIEW26_ARB */ - 972, /* GL_MODELVIEW27_ARB */ - 973, /* GL_MODELVIEW28_ARB */ - 974, /* GL_MODELVIEW29_ARB */ - 976, /* GL_MODELVIEW30_ARB */ - 977, /* GL_MODELVIEW31_ARB */ + 996, /* GL_MULTISAMPLE_3DFX */ + 1391, /* GL_SAMPLE_BUFFERS_3DFX */ + 1382, /* GL_SAMPLES_3DFX */ + 977, /* GL_MODELVIEW2_ARB */ + 980, /* GL_MODELVIEW3_ARB */ + 981, /* GL_MODELVIEW4_ARB */ + 982, /* GL_MODELVIEW5_ARB */ + 983, /* GL_MODELVIEW6_ARB */ + 984, /* GL_MODELVIEW7_ARB */ + 985, /* GL_MODELVIEW8_ARB */ + 986, /* GL_MODELVIEW9_ARB */ + 956, /* GL_MODELVIEW10_ARB */ + 957, /* GL_MODELVIEW11_ARB */ + 958, /* GL_MODELVIEW12_ARB */ + 959, /* GL_MODELVIEW13_ARB */ + 960, /* GL_MODELVIEW14_ARB */ + 961, /* GL_MODELVIEW15_ARB */ + 962, /* GL_MODELVIEW16_ARB */ + 963, /* GL_MODELVIEW17_ARB */ + 964, /* GL_MODELVIEW18_ARB */ + 965, /* GL_MODELVIEW19_ARB */ + 967, /* GL_MODELVIEW20_ARB */ + 968, /* GL_MODELVIEW21_ARB */ + 969, /* GL_MODELVIEW22_ARB */ + 970, /* GL_MODELVIEW23_ARB */ + 971, /* GL_MODELVIEW24_ARB */ + 972, /* GL_MODELVIEW25_ARB */ + 973, /* GL_MODELVIEW26_ARB */ + 974, /* GL_MODELVIEW27_ARB */ + 975, /* GL_MODELVIEW28_ARB */ + 976, /* GL_MODELVIEW29_ARB */ + 978, /* GL_MODELVIEW30_ARB */ + 979, /* GL_MODELVIEW31_ARB */ 389, /* GL_DOT3_RGB_EXT */ 387, /* GL_DOT3_RGBA_EXT */ - 948, /* GL_MIRROR_CLAMP_EXT */ - 951, /* GL_MIRROR_CLAMP_TO_EDGE_EXT */ - 989, /* GL_MODULATE_ADD_ATI */ - 990, /* GL_MODULATE_SIGNED_ADD_ATI */ - 991, /* GL_MODULATE_SUBTRACT_ATI */ - 1854, /* GL_YCBCR_MESA */ - 1081, /* GL_PACK_INVERT_MESA */ + 950, /* GL_MIRROR_CLAMP_EXT */ + 953, /* GL_MIRROR_CLAMP_TO_EDGE_EXT */ + 991, /* GL_MODULATE_ADD_ATI */ + 992, /* GL_MODULATE_SIGNED_ADD_ATI */ + 993, /* GL_MODULATE_SUBTRACT_ATI */ + 1858, /* GL_YCBCR_MESA */ + 1083, /* GL_PACK_INVERT_MESA */ 339, /* GL_DEBUG_OBJECT_MESA */ 340, /* GL_DEBUG_PRINT_MESA */ 338, /* GL_DEBUG_ASSERT_MESA */ @@ -4758,24 +4766,24 @@ static const unsigned reduced_enums[1347] = 447, /* GL_DU8DV8_ATI */ 114, /* GL_BUMP_ENVMAP_ATI */ 118, /* GL_BUMP_TARGET_ATI */ - 1486, /* GL_STENCIL_BACK_FUNC */ - 1484, /* GL_STENCIL_BACK_FAIL */ - 1488, /* GL_STENCIL_BACK_PASS_DEPTH_FAIL */ - 1490, /* GL_STENCIL_BACK_PASS_DEPTH_PASS */ - 532, /* GL_FRAGMENT_PROGRAM_ARB */ - 1219, /* GL_PROGRAM_ALU_INSTRUCTIONS_ARB */ - 1247, /* GL_PROGRAM_TEX_INSTRUCTIONS_ARB */ - 1246, /* GL_PROGRAM_TEX_INDIRECTIONS_ARB */ - 1231, /* GL_PROGRAM_NATIVE_ALU_INSTRUCTIONS_ARB */ - 1237, /* GL_PROGRAM_NATIVE_TEX_INSTRUCTIONS_ARB */ - 1236, /* GL_PROGRAM_NATIVE_TEX_INDIRECTIONS_ARB */ - 882, /* GL_MAX_PROGRAM_ALU_INSTRUCTIONS_ARB */ - 905, /* GL_MAX_PROGRAM_TEX_INSTRUCTIONS_ARB */ - 904, /* GL_MAX_PROGRAM_TEX_INDIRECTIONS_ARB */ - 895, /* GL_MAX_PROGRAM_NATIVE_ALU_INSTRUCTIONS_ARB */ - 901, /* GL_MAX_PROGRAM_NATIVE_TEX_INSTRUCTIONS_ARB */ - 900, /* GL_MAX_PROGRAM_NATIVE_TEX_INDIRECTIONS_ARB */ - 865, /* GL_MAX_DRAW_BUFFERS */ + 1490, /* GL_STENCIL_BACK_FUNC */ + 1488, /* GL_STENCIL_BACK_FAIL */ + 1492, /* GL_STENCIL_BACK_PASS_DEPTH_FAIL */ + 1494, /* GL_STENCIL_BACK_PASS_DEPTH_PASS */ + 533, /* GL_FRAGMENT_PROGRAM_ARB */ + 1221, /* GL_PROGRAM_ALU_INSTRUCTIONS_ARB */ + 1249, /* GL_PROGRAM_TEX_INSTRUCTIONS_ARB */ + 1248, /* GL_PROGRAM_TEX_INDIRECTIONS_ARB */ + 1233, /* GL_PROGRAM_NATIVE_ALU_INSTRUCTIONS_ARB */ + 1239, /* GL_PROGRAM_NATIVE_TEX_INSTRUCTIONS_ARB */ + 1238, /* GL_PROGRAM_NATIVE_TEX_INDIRECTIONS_ARB */ + 884, /* GL_MAX_PROGRAM_ALU_INSTRUCTIONS_ARB */ + 907, /* GL_MAX_PROGRAM_TEX_INSTRUCTIONS_ARB */ + 906, /* GL_MAX_PROGRAM_TEX_INDIRECTIONS_ARB */ + 897, /* GL_MAX_PROGRAM_NATIVE_ALU_INSTRUCTIONS_ARB */ + 903, /* GL_MAX_PROGRAM_NATIVE_TEX_INSTRUCTIONS_ARB */ + 902, /* GL_MAX_PROGRAM_NATIVE_TEX_INDIRECTIONS_ARB */ + 867, /* GL_MAX_DRAW_BUFFERS */ 393, /* GL_DRAW_BUFFER0 */ 396, /* GL_DRAW_BUFFER1 */ 417, /* GL_DRAW_BUFFER2 */ @@ -4793,253 +4801,253 @@ static const unsigned reduced_enums[1347] = 409, /* GL_DRAW_BUFFER14 */ 412, /* GL_DRAW_BUFFER15 */ 82, /* GL_BLEND_EQUATION_ALPHA */ - 845, /* GL_MATRIX_PALETTE_ARB */ - 876, /* GL_MAX_MATRIX_PALETTE_STACK_DEPTH_ARB */ - 879, /* GL_MAX_PALETTE_MATRICES_ARB */ + 847, /* GL_MATRIX_PALETTE_ARB */ + 878, /* GL_MAX_MATRIX_PALETTE_STACK_DEPTH_ARB */ + 881, /* GL_MAX_PALETTE_MATRICES_ARB */ 321, /* GL_CURRENT_PALETTE_MATRIX_ARB */ - 839, /* GL_MATRIX_INDEX_ARRAY_ARB */ + 841, /* GL_MATRIX_INDEX_ARRAY_ARB */ 316, /* GL_CURRENT_MATRIX_INDEX_ARB */ - 841, /* GL_MATRIX_INDEX_ARRAY_SIZE_ARB */ - 843, /* GL_MATRIX_INDEX_ARRAY_TYPE_ARB */ - 842, /* GL_MATRIX_INDEX_ARRAY_STRIDE_ARB */ - 840, /* GL_MATRIX_INDEX_ARRAY_POINTER_ARB */ - 1674, /* GL_TEXTURE_DEPTH_SIZE */ + 843, /* GL_MATRIX_INDEX_ARRAY_SIZE_ARB */ + 845, /* GL_MATRIX_INDEX_ARRAY_TYPE_ARB */ + 844, /* GL_MATRIX_INDEX_ARRAY_STRIDE_ARB */ + 842, /* GL_MATRIX_INDEX_ARRAY_POINTER_ARB */ + 1678, /* GL_TEXTURE_DEPTH_SIZE */ 377, /* GL_DEPTH_TEXTURE_MODE */ - 1639, /* GL_TEXTURE_COMPARE_MODE */ - 1637, /* GL_TEXTURE_COMPARE_FUNC */ + 1643, /* GL_TEXTURE_COMPARE_MODE */ + 1641, /* GL_TEXTURE_COMPARE_FUNC */ 242, /* GL_COMPARE_R_TO_TEXTURE */ - 1153, /* GL_POINT_SPRITE */ + 1155, /* GL_POINT_SPRITE */ 296, /* GL_COORD_REPLACE */ - 1157, /* GL_POINT_SPRITE_R_MODE_NV */ - 1276, /* GL_QUERY_COUNTER_BITS */ + 1159, /* GL_POINT_SPRITE_R_MODE_NV */ + 1280, /* GL_QUERY_COUNTER_BITS */ 323, /* GL_CURRENT_QUERY */ - 1278, /* GL_QUERY_RESULT */ - 1280, /* GL_QUERY_RESULT_AVAILABLE */ - 928, /* GL_MAX_VERTEX_ATTRIBS */ - 1815, /* GL_VERTEX_ATTRIB_ARRAY_NORMALIZED */ + 1282, /* GL_QUERY_RESULT */ + 1284, /* GL_QUERY_RESULT_AVAILABLE */ + 930, /* GL_MAX_VERTEX_ATTRIBS */ + 1819, /* GL_VERTEX_ATTRIB_ARRAY_NORMALIZED */ 375, /* GL_DEPTH_STENCIL_TO_RGBA_NV */ 374, /* GL_DEPTH_STENCIL_TO_BGRA_NV */ - 914, /* GL_MAX_TEXTURE_COORDS */ - 916, /* GL_MAX_TEXTURE_IMAGE_UNITS */ - 1224, /* GL_PROGRAM_ERROR_STRING_ARB */ - 1226, /* GL_PROGRAM_FORMAT_ASCII_ARB */ - 1225, /* GL_PROGRAM_FORMAT_ARB */ - 1723, /* GL_TEXTURE_UNSIGNED_REMAP_MODE_NV */ + 916, /* GL_MAX_TEXTURE_COORDS */ + 918, /* GL_MAX_TEXTURE_IMAGE_UNITS */ + 1226, /* GL_PROGRAM_ERROR_STRING_ARB */ + 1228, /* GL_PROGRAM_FORMAT_ASCII_ARB */ + 1227, /* GL_PROGRAM_FORMAT_ARB */ + 1727, /* GL_TEXTURE_UNSIGNED_REMAP_MODE_NV */ 353, /* GL_DEPTH_BOUNDS_TEST_EXT */ 352, /* GL_DEPTH_BOUNDS_EXT */ 53, /* GL_ARRAY_BUFFER */ 461, /* GL_ELEMENT_ARRAY_BUFFER */ 54, /* GL_ARRAY_BUFFER_BINDING */ 462, /* GL_ELEMENT_ARRAY_BUFFER_BINDING */ - 1789, /* GL_VERTEX_ARRAY_BUFFER_BINDING */ - 1015, /* GL_NORMAL_ARRAY_BUFFER_BINDING */ + 1793, /* GL_VERTEX_ARRAY_BUFFER_BINDING */ + 1017, /* GL_NORMAL_ARRAY_BUFFER_BINDING */ 149, /* GL_COLOR_ARRAY_BUFFER_BINDING */ - 624, /* GL_INDEX_ARRAY_BUFFER_BINDING */ - 1652, /* GL_TEXTURE_COORD_ARRAY_BUFFER_BINDING */ + 625, /* GL_INDEX_ARRAY_BUFFER_BINDING */ + 1656, /* GL_TEXTURE_COORD_ARRAY_BUFFER_BINDING */ 457, /* GL_EDGE_FLAG_ARRAY_BUFFER_BINDING */ - 1399, /* GL_SECONDARY_COLOR_ARRAY_BUFFER_BINDING */ - 510, /* GL_FOG_COORDINATE_ARRAY_BUFFER_BINDING */ - 1842, /* GL_WEIGHT_ARRAY_BUFFER_BINDING */ - 1811, /* GL_VERTEX_ATTRIB_ARRAY_BUFFER_BINDING */ - 1227, /* GL_PROGRAM_INSTRUCTIONS_ARB */ - 888, /* GL_MAX_PROGRAM_INSTRUCTIONS_ARB */ - 1233, /* GL_PROGRAM_NATIVE_INSTRUCTIONS_ARB */ - 897, /* GL_MAX_PROGRAM_NATIVE_INSTRUCTIONS_ARB */ - 1245, /* GL_PROGRAM_TEMPORARIES_ARB */ - 903, /* GL_MAX_PROGRAM_TEMPORARIES_ARB */ - 1235, /* GL_PROGRAM_NATIVE_TEMPORARIES_ARB */ - 899, /* GL_MAX_PROGRAM_NATIVE_TEMPORARIES_ARB */ - 1239, /* GL_PROGRAM_PARAMETERS_ARB */ - 902, /* GL_MAX_PROGRAM_PARAMETERS_ARB */ - 1234, /* GL_PROGRAM_NATIVE_PARAMETERS_ARB */ - 898, /* GL_MAX_PROGRAM_NATIVE_PARAMETERS_ARB */ - 1220, /* GL_PROGRAM_ATTRIBS_ARB */ - 883, /* GL_MAX_PROGRAM_ATTRIBS_ARB */ - 1232, /* GL_PROGRAM_NATIVE_ATTRIBS_ARB */ - 896, /* GL_MAX_PROGRAM_NATIVE_ATTRIBS_ARB */ - 1218, /* GL_PROGRAM_ADDRESS_REGISTERS_ARB */ - 881, /* GL_MAX_PROGRAM_ADDRESS_REGISTERS_ARB */ - 1230, /* GL_PROGRAM_NATIVE_ADDRESS_REGISTERS_ARB */ - 894, /* GL_MAX_PROGRAM_NATIVE_ADDRESS_REGISTERS_ARB */ - 889, /* GL_MAX_PROGRAM_LOCAL_PARAMETERS_ARB */ - 885, /* GL_MAX_PROGRAM_ENV_PARAMETERS_ARB */ - 1248, /* GL_PROGRAM_UNDER_NATIVE_LIMITS_ARB */ - 1736, /* GL_TRANSPOSE_CURRENT_MATRIX_ARB */ - 1289, /* GL_READ_ONLY */ - 1850, /* GL_WRITE_ONLY */ - 1291, /* GL_READ_WRITE */ + 1403, /* GL_SECONDARY_COLOR_ARRAY_BUFFER_BINDING */ + 511, /* GL_FOG_COORDINATE_ARRAY_BUFFER_BINDING */ + 1846, /* GL_WEIGHT_ARRAY_BUFFER_BINDING */ + 1815, /* GL_VERTEX_ATTRIB_ARRAY_BUFFER_BINDING */ + 1229, /* GL_PROGRAM_INSTRUCTIONS_ARB */ + 890, /* GL_MAX_PROGRAM_INSTRUCTIONS_ARB */ + 1235, /* GL_PROGRAM_NATIVE_INSTRUCTIONS_ARB */ + 899, /* GL_MAX_PROGRAM_NATIVE_INSTRUCTIONS_ARB */ + 1247, /* GL_PROGRAM_TEMPORARIES_ARB */ + 905, /* GL_MAX_PROGRAM_TEMPORARIES_ARB */ + 1237, /* GL_PROGRAM_NATIVE_TEMPORARIES_ARB */ + 901, /* GL_MAX_PROGRAM_NATIVE_TEMPORARIES_ARB */ + 1241, /* GL_PROGRAM_PARAMETERS_ARB */ + 904, /* GL_MAX_PROGRAM_PARAMETERS_ARB */ + 1236, /* GL_PROGRAM_NATIVE_PARAMETERS_ARB */ + 900, /* GL_MAX_PROGRAM_NATIVE_PARAMETERS_ARB */ + 1222, /* GL_PROGRAM_ATTRIBS_ARB */ + 885, /* GL_MAX_PROGRAM_ATTRIBS_ARB */ + 1234, /* GL_PROGRAM_NATIVE_ATTRIBS_ARB */ + 898, /* GL_MAX_PROGRAM_NATIVE_ATTRIBS_ARB */ + 1220, /* GL_PROGRAM_ADDRESS_REGISTERS_ARB */ + 883, /* GL_MAX_PROGRAM_ADDRESS_REGISTERS_ARB */ + 1232, /* GL_PROGRAM_NATIVE_ADDRESS_REGISTERS_ARB */ + 896, /* GL_MAX_PROGRAM_NATIVE_ADDRESS_REGISTERS_ARB */ + 891, /* GL_MAX_PROGRAM_LOCAL_PARAMETERS_ARB */ + 887, /* GL_MAX_PROGRAM_ENV_PARAMETERS_ARB */ + 1250, /* GL_PROGRAM_UNDER_NATIVE_LIMITS_ARB */ + 1740, /* GL_TRANSPOSE_CURRENT_MATRIX_ARB */ + 1293, /* GL_READ_ONLY */ + 1854, /* GL_WRITE_ONLY */ + 1295, /* GL_READ_WRITE */ 102, /* GL_BUFFER_ACCESS */ 105, /* GL_BUFFER_MAPPED */ 107, /* GL_BUFFER_MAP_POINTER */ - 1730, /* GL_TIME_ELAPSED_EXT */ - 799, /* GL_MATRIX0_ARB */ - 811, /* GL_MATRIX1_ARB */ - 823, /* GL_MATRIX2_ARB */ - 827, /* GL_MATRIX3_ARB */ - 829, /* GL_MATRIX4_ARB */ - 831, /* GL_MATRIX5_ARB */ - 833, /* GL_MATRIX6_ARB */ - 835, /* GL_MATRIX7_ARB */ - 837, /* GL_MATRIX8_ARB */ - 838, /* GL_MATRIX9_ARB */ - 801, /* GL_MATRIX10_ARB */ - 802, /* GL_MATRIX11_ARB */ - 803, /* GL_MATRIX12_ARB */ - 804, /* GL_MATRIX13_ARB */ - 805, /* GL_MATRIX14_ARB */ - 806, /* GL_MATRIX15_ARB */ - 807, /* GL_MATRIX16_ARB */ - 808, /* GL_MATRIX17_ARB */ - 809, /* GL_MATRIX18_ARB */ - 810, /* GL_MATRIX19_ARB */ - 813, /* GL_MATRIX20_ARB */ - 814, /* GL_MATRIX21_ARB */ - 815, /* GL_MATRIX22_ARB */ - 816, /* GL_MATRIX23_ARB */ - 817, /* GL_MATRIX24_ARB */ - 818, /* GL_MATRIX25_ARB */ - 819, /* GL_MATRIX26_ARB */ - 820, /* GL_MATRIX27_ARB */ - 821, /* GL_MATRIX28_ARB */ - 822, /* GL_MATRIX29_ARB */ - 825, /* GL_MATRIX30_ARB */ - 826, /* GL_MATRIX31_ARB */ - 1519, /* GL_STREAM_DRAW */ - 1521, /* GL_STREAM_READ */ - 1517, /* GL_STREAM_COPY */ - 1477, /* GL_STATIC_DRAW */ - 1479, /* GL_STATIC_READ */ - 1475, /* GL_STATIC_COPY */ + 1734, /* GL_TIME_ELAPSED_EXT */ + 801, /* GL_MATRIX0_ARB */ + 813, /* GL_MATRIX1_ARB */ + 825, /* GL_MATRIX2_ARB */ + 829, /* GL_MATRIX3_ARB */ + 831, /* GL_MATRIX4_ARB */ + 833, /* GL_MATRIX5_ARB */ + 835, /* GL_MATRIX6_ARB */ + 837, /* GL_MATRIX7_ARB */ + 839, /* GL_MATRIX8_ARB */ + 840, /* GL_MATRIX9_ARB */ + 803, /* GL_MATRIX10_ARB */ + 804, /* GL_MATRIX11_ARB */ + 805, /* GL_MATRIX12_ARB */ + 806, /* GL_MATRIX13_ARB */ + 807, /* GL_MATRIX14_ARB */ + 808, /* GL_MATRIX15_ARB */ + 809, /* GL_MATRIX16_ARB */ + 810, /* GL_MATRIX17_ARB */ + 811, /* GL_MATRIX18_ARB */ + 812, /* GL_MATRIX19_ARB */ + 815, /* GL_MATRIX20_ARB */ + 816, /* GL_MATRIX21_ARB */ + 817, /* GL_MATRIX22_ARB */ + 818, /* GL_MATRIX23_ARB */ + 819, /* GL_MATRIX24_ARB */ + 820, /* GL_MATRIX25_ARB */ + 821, /* GL_MATRIX26_ARB */ + 822, /* GL_MATRIX27_ARB */ + 823, /* GL_MATRIX28_ARB */ + 824, /* GL_MATRIX29_ARB */ + 827, /* GL_MATRIX30_ARB */ + 828, /* GL_MATRIX31_ARB */ + 1523, /* GL_STREAM_DRAW */ + 1525, /* GL_STREAM_READ */ + 1521, /* GL_STREAM_COPY */ + 1481, /* GL_STATIC_DRAW */ + 1483, /* GL_STATIC_READ */ + 1479, /* GL_STATIC_COPY */ 451, /* GL_DYNAMIC_DRAW */ 453, /* GL_DYNAMIC_READ */ 449, /* GL_DYNAMIC_COPY */ - 1121, /* GL_PIXEL_PACK_BUFFER */ - 1125, /* GL_PIXEL_UNPACK_BUFFER */ - 1122, /* GL_PIXEL_PACK_BUFFER_BINDING */ - 1126, /* GL_PIXEL_UNPACK_BUFFER_BINDING */ + 1123, /* GL_PIXEL_PACK_BUFFER */ + 1127, /* GL_PIXEL_UNPACK_BUFFER */ + 1124, /* GL_PIXEL_PACK_BUFFER_BINDING */ + 1128, /* GL_PIXEL_UNPACK_BUFFER_BINDING */ 347, /* GL_DEPTH24_STENCIL8 */ - 1720, /* GL_TEXTURE_STENCIL_SIZE */ - 1672, /* GL_TEXTURE_CUBE_MAP_SEAMLESS */ - 884, /* GL_MAX_PROGRAM_CALL_DEPTH_NV */ - 887, /* GL_MAX_PROGRAM_IF_DEPTH_NV */ - 891, /* GL_MAX_PROGRAM_LOOP_DEPTH_NV */ - 890, /* GL_MAX_PROGRAM_LOOP_COUNT_NV */ - 848, /* GL_MAX_ARRAY_TEXTURE_LAYERS_EXT */ - 1510, /* GL_STENCIL_TEST_TWO_SIDE_EXT */ + 1724, /* GL_TEXTURE_STENCIL_SIZE */ + 1676, /* GL_TEXTURE_CUBE_MAP_SEAMLESS */ + 886, /* GL_MAX_PROGRAM_CALL_DEPTH_NV */ + 889, /* GL_MAX_PROGRAM_IF_DEPTH_NV */ + 893, /* GL_MAX_PROGRAM_LOOP_DEPTH_NV */ + 892, /* GL_MAX_PROGRAM_LOOP_COUNT_NV */ + 850, /* GL_MAX_ARRAY_TEXTURE_LAYERS_EXT */ + 1514, /* GL_STENCIL_TEST_TWO_SIDE_EXT */ 17, /* GL_ACTIVE_STENCIL_FACE_EXT */ - 949, /* GL_MIRROR_CLAMP_TO_BORDER_EXT */ - 1380, /* GL_SAMPLES_PASSED */ + 951, /* GL_MIRROR_CLAMP_TO_BORDER_EXT */ + 1384, /* GL_SAMPLES_PASSED */ 109, /* GL_BUFFER_SERIALIZED_MODIFY_APPLE */ 104, /* GL_BUFFER_FLUSHING_UNMAP_APPLE */ - 533, /* GL_FRAGMENT_SHADER */ - 1835, /* GL_VERTEX_SHADER */ - 1238, /* GL_PROGRAM_OBJECT_ARB */ - 1412, /* GL_SHADER_OBJECT_ARB */ - 872, /* GL_MAX_FRAGMENT_UNIFORM_COMPONENTS */ - 932, /* GL_MAX_VERTEX_UNIFORM_COMPONENTS */ - 926, /* GL_MAX_VARYING_FLOATS */ - 930, /* GL_MAX_VERTEX_TEXTURE_IMAGE_UNITS */ - 857, /* GL_MAX_COMBINED_TEXTURE_IMAGE_UNITS */ - 1041, /* GL_OBJECT_TYPE_ARB */ - 1414, /* GL_SHADER_TYPE */ - 498, /* GL_FLOAT_VEC2 */ - 500, /* GL_FLOAT_VEC3 */ - 502, /* GL_FLOAT_VEC4 */ - 651, /* GL_INT_VEC2 */ - 653, /* GL_INT_VEC3 */ - 655, /* GL_INT_VEC4 */ + 534, /* GL_FRAGMENT_SHADER */ + 1839, /* GL_VERTEX_SHADER */ + 1240, /* GL_PROGRAM_OBJECT_ARB */ + 1416, /* GL_SHADER_OBJECT_ARB */ + 874, /* GL_MAX_FRAGMENT_UNIFORM_COMPONENTS */ + 934, /* GL_MAX_VERTEX_UNIFORM_COMPONENTS */ + 928, /* GL_MAX_VARYING_FLOATS */ + 932, /* GL_MAX_VERTEX_TEXTURE_IMAGE_UNITS */ + 859, /* GL_MAX_COMBINED_TEXTURE_IMAGE_UNITS */ + 1043, /* GL_OBJECT_TYPE_ARB */ + 1418, /* GL_SHADER_TYPE */ + 499, /* GL_FLOAT_VEC2 */ + 501, /* GL_FLOAT_VEC3 */ + 503, /* GL_FLOAT_VEC4 */ + 652, /* GL_INT_VEC2 */ + 654, /* GL_INT_VEC3 */ + 656, /* GL_INT_VEC4 */ 94, /* GL_BOOL */ 96, /* GL_BOOL_VEC2 */ 98, /* GL_BOOL_VEC3 */ 100, /* GL_BOOL_VEC4 */ - 486, /* GL_FLOAT_MAT2 */ - 490, /* GL_FLOAT_MAT3 */ - 494, /* GL_FLOAT_MAT4 */ - 1371, /* GL_SAMPLER_1D */ - 1373, /* GL_SAMPLER_2D */ - 1375, /* GL_SAMPLER_3D */ - 1376, /* GL_SAMPLER_CUBE */ - 1372, /* GL_SAMPLER_1D_SHADOW */ - 1374, /* GL_SAMPLER_2D_SHADOW */ - 488, /* GL_FLOAT_MAT2x3 */ - 489, /* GL_FLOAT_MAT2x4 */ - 492, /* GL_FLOAT_MAT3x2 */ - 493, /* GL_FLOAT_MAT3x4 */ - 496, /* GL_FLOAT_MAT4x2 */ - 497, /* GL_FLOAT_MAT4x3 */ + 487, /* GL_FLOAT_MAT2 */ + 491, /* GL_FLOAT_MAT3 */ + 495, /* GL_FLOAT_MAT4 */ + 1375, /* GL_SAMPLER_1D */ + 1377, /* GL_SAMPLER_2D */ + 1379, /* GL_SAMPLER_3D */ + 1380, /* GL_SAMPLER_CUBE */ + 1376, /* GL_SAMPLER_1D_SHADOW */ + 1378, /* GL_SAMPLER_2D_SHADOW */ + 489, /* GL_FLOAT_MAT2x3 */ + 490, /* GL_FLOAT_MAT2x4 */ + 493, /* GL_FLOAT_MAT3x2 */ + 494, /* GL_FLOAT_MAT3x4 */ + 497, /* GL_FLOAT_MAT4x2 */ + 498, /* GL_FLOAT_MAT4x3 */ 345, /* GL_DELETE_STATUS */ 246, /* GL_COMPILE_STATUS */ - 706, /* GL_LINK_STATUS */ - 1783, /* GL_VALIDATE_STATUS */ - 636, /* GL_INFO_LOG_LENGTH */ + 708, /* GL_LINK_STATUS */ + 1787, /* GL_VALIDATE_STATUS */ + 637, /* GL_INFO_LOG_LENGTH */ 56, /* GL_ATTACHED_SHADERS */ 20, /* GL_ACTIVE_UNIFORMS */ 21, /* GL_ACTIVE_UNIFORM_MAX_LENGTH */ - 1413, /* GL_SHADER_SOURCE_LENGTH */ + 1417, /* GL_SHADER_SOURCE_LENGTH */ 15, /* GL_ACTIVE_ATTRIBUTES */ 16, /* GL_ACTIVE_ATTRIBUTE_MAX_LENGTH */ - 535, /* GL_FRAGMENT_SHADER_DERIVATIVE_HINT */ - 1416, /* GL_SHADING_LANGUAGE_VERSION */ + 536, /* GL_FRAGMENT_SHADER_DERIVATIVE_HINT */ + 1420, /* GL_SHADING_LANGUAGE_VERSION */ 322, /* GL_CURRENT_PROGRAM */ - 1090, /* GL_PALETTE4_RGB8_OES */ - 1092, /* GL_PALETTE4_RGBA8_OES */ - 1088, /* GL_PALETTE4_R5_G6_B5_OES */ - 1091, /* GL_PALETTE4_RGBA4_OES */ - 1089, /* GL_PALETTE4_RGB5_A1_OES */ - 1095, /* GL_PALETTE8_RGB8_OES */ - 1097, /* GL_PALETTE8_RGBA8_OES */ - 1093, /* GL_PALETTE8_R5_G6_B5_OES */ - 1096, /* GL_PALETTE8_RGBA4_OES */ - 1094, /* GL_PALETTE8_RGB5_A1_OES */ - 618, /* GL_IMPLEMENTATION_COLOR_READ_TYPE_OES */ - 617, /* GL_IMPLEMENTATION_COLOR_READ_FORMAT_OES */ - 1768, /* GL_UNSIGNED_NORMALIZED */ - 1607, /* GL_TEXTURE_1D_ARRAY_EXT */ - 1259, /* GL_PROXY_TEXTURE_1D_ARRAY_EXT */ - 1609, /* GL_TEXTURE_2D_ARRAY_EXT */ - 1262, /* GL_PROXY_TEXTURE_2D_ARRAY_EXT */ - 1615, /* GL_TEXTURE_BINDING_1D_ARRAY_EXT */ - 1617, /* GL_TEXTURE_BINDING_2D_ARRAY_EXT */ - 1469, /* GL_SRGB */ - 1470, /* GL_SRGB8 */ - 1472, /* GL_SRGB_ALPHA */ - 1471, /* GL_SRGB8_ALPHA8 */ - 1429, /* GL_SLUMINANCE_ALPHA */ - 1428, /* GL_SLUMINANCE8_ALPHA8 */ - 1426, /* GL_SLUMINANCE */ - 1427, /* GL_SLUMINANCE8 */ + 1092, /* GL_PALETTE4_RGB8_OES */ + 1094, /* GL_PALETTE4_RGBA8_OES */ + 1090, /* GL_PALETTE4_R5_G6_B5_OES */ + 1093, /* GL_PALETTE4_RGBA4_OES */ + 1091, /* GL_PALETTE4_RGB5_A1_OES */ + 1097, /* GL_PALETTE8_RGB8_OES */ + 1099, /* GL_PALETTE8_RGBA8_OES */ + 1095, /* GL_PALETTE8_R5_G6_B5_OES */ + 1098, /* GL_PALETTE8_RGBA4_OES */ + 1096, /* GL_PALETTE8_RGB5_A1_OES */ + 619, /* GL_IMPLEMENTATION_COLOR_READ_TYPE_OES */ + 618, /* GL_IMPLEMENTATION_COLOR_READ_FORMAT_OES */ + 1772, /* GL_UNSIGNED_NORMALIZED */ + 1611, /* GL_TEXTURE_1D_ARRAY_EXT */ + 1262, /* GL_PROXY_TEXTURE_1D_ARRAY_EXT */ + 1613, /* GL_TEXTURE_2D_ARRAY_EXT */ + 1265, /* GL_PROXY_TEXTURE_2D_ARRAY_EXT */ + 1619, /* GL_TEXTURE_BINDING_1D_ARRAY_EXT */ + 1621, /* GL_TEXTURE_BINDING_2D_ARRAY_EXT */ + 1473, /* GL_SRGB */ + 1474, /* GL_SRGB8 */ + 1476, /* GL_SRGB_ALPHA */ + 1475, /* GL_SRGB8_ALPHA8 */ + 1433, /* GL_SLUMINANCE_ALPHA */ + 1432, /* GL_SLUMINANCE8_ALPHA8 */ + 1430, /* GL_SLUMINANCE */ + 1431, /* GL_SLUMINANCE8 */ 267, /* GL_COMPRESSED_SRGB */ 268, /* GL_COMPRESSED_SRGB_ALPHA */ 265, /* GL_COMPRESSED_SLUMINANCE */ 266, /* GL_COMPRESSED_SLUMINANCE_ALPHA */ - 1155, /* GL_POINT_SPRITE_COORD_ORIGIN */ - 714, /* GL_LOWER_LEFT */ - 1780, /* GL_UPPER_LEFT */ - 1492, /* GL_STENCIL_BACK_REF */ - 1493, /* GL_STENCIL_BACK_VALUE_MASK */ - 1494, /* GL_STENCIL_BACK_WRITEMASK */ + 1157, /* GL_POINT_SPRITE_COORD_ORIGIN */ + 716, /* GL_LOWER_LEFT */ + 1784, /* GL_UPPER_LEFT */ + 1496, /* GL_STENCIL_BACK_REF */ + 1497, /* GL_STENCIL_BACK_VALUE_MASK */ + 1498, /* GL_STENCIL_BACK_WRITEMASK */ 442, /* GL_DRAW_FRAMEBUFFER_BINDING_EXT */ - 1305, /* GL_RENDERBUFFER_BINDING_EXT */ - 1286, /* GL_READ_FRAMEBUFFER */ + 1309, /* GL_RENDERBUFFER_BINDING_EXT */ + 1290, /* GL_READ_FRAMEBUFFER */ 441, /* GL_DRAW_FRAMEBUFFER */ - 1287, /* GL_READ_FRAMEBUFFER_BINDING_EXT */ - 1315, /* GL_RENDERBUFFER_SAMPLES */ - 545, /* GL_FRAMEBUFFER_ATTACHMENT_OBJECT_TYPE */ - 543, /* GL_FRAMEBUFFER_ATTACHMENT_OBJECT_NAME */ - 554, /* GL_FRAMEBUFFER_ATTACHMENT_TEXTURE_LEVEL */ - 550, /* GL_FRAMEBUFFER_ATTACHMENT_TEXTURE_CUBE_MAP_FACE */ - 552, /* GL_FRAMEBUFFER_ATTACHMENT_TEXTURE_LAYER */ - 557, /* GL_FRAMEBUFFER_COMPLETE */ - 561, /* GL_FRAMEBUFFER_INCOMPLETE_ATTACHMENT */ - 567, /* GL_FRAMEBUFFER_INCOMPLETE_MISSING_ATTACHMENT */ - 565, /* GL_FRAMEBUFFER_INCOMPLETE_DUPLICATE_ATTACHMENT_EXT */ - 563, /* GL_FRAMEBUFFER_INCOMPLETE_DIMENSIONS_EXT */ - 566, /* GL_FRAMEBUFFER_INCOMPLETE_FORMATS_EXT */ - 564, /* GL_FRAMEBUFFER_INCOMPLETE_DRAW_BUFFER_EXT */ - 570, /* GL_FRAMEBUFFER_INCOMPLETE_READ_BUFFER_EXT */ - 573, /* GL_FRAMEBUFFER_UNSUPPORTED */ - 571, /* GL_FRAMEBUFFER_STATUS_ERROR_EXT */ - 854, /* GL_MAX_COLOR_ATTACHMENTS_EXT */ + 1291, /* GL_READ_FRAMEBUFFER_BINDING_EXT */ + 1319, /* GL_RENDERBUFFER_SAMPLES */ + 546, /* GL_FRAMEBUFFER_ATTACHMENT_OBJECT_TYPE */ + 544, /* GL_FRAMEBUFFER_ATTACHMENT_OBJECT_NAME */ + 555, /* GL_FRAMEBUFFER_ATTACHMENT_TEXTURE_LEVEL */ + 551, /* GL_FRAMEBUFFER_ATTACHMENT_TEXTURE_CUBE_MAP_FACE */ + 553, /* GL_FRAMEBUFFER_ATTACHMENT_TEXTURE_LAYER */ + 558, /* GL_FRAMEBUFFER_COMPLETE */ + 562, /* GL_FRAMEBUFFER_INCOMPLETE_ATTACHMENT */ + 568, /* GL_FRAMEBUFFER_INCOMPLETE_MISSING_ATTACHMENT */ + 566, /* GL_FRAMEBUFFER_INCOMPLETE_DUPLICATE_ATTACHMENT_EXT */ + 564, /* GL_FRAMEBUFFER_INCOMPLETE_DIMENSIONS_EXT */ + 567, /* GL_FRAMEBUFFER_INCOMPLETE_FORMATS_EXT */ + 565, /* GL_FRAMEBUFFER_INCOMPLETE_DRAW_BUFFER_EXT */ + 571, /* GL_FRAMEBUFFER_INCOMPLETE_READ_BUFFER_EXT */ + 574, /* GL_FRAMEBUFFER_UNSUPPORTED */ + 572, /* GL_FRAMEBUFFER_STATUS_ERROR_EXT */ + 856, /* GL_MAX_COLOR_ATTACHMENTS_EXT */ 155, /* GL_COLOR_ATTACHMENT0 */ 157, /* GL_COLOR_ATTACHMENT1 */ 171, /* GL_COLOR_ATTACHMENT2 */ @@ -5057,56 +5065,56 @@ static const unsigned reduced_enums[1347] = 166, /* GL_COLOR_ATTACHMENT14 */ 168, /* GL_COLOR_ATTACHMENT15 */ 348, /* GL_DEPTH_ATTACHMENT */ - 1482, /* GL_STENCIL_ATTACHMENT */ - 536, /* GL_FRAMEBUFFER */ - 1303, /* GL_RENDERBUFFER */ - 1317, /* GL_RENDERBUFFER_WIDTH */ - 1310, /* GL_RENDERBUFFER_HEIGHT */ - 1312, /* GL_RENDERBUFFER_INTERNAL_FORMAT */ - 1505, /* GL_STENCIL_INDEX_EXT */ - 1502, /* GL_STENCIL_INDEX1_EXT */ - 1503, /* GL_STENCIL_INDEX4_EXT */ - 1504, /* GL_STENCIL_INDEX8_EXT */ - 1501, /* GL_STENCIL_INDEX16_EXT */ - 1314, /* GL_RENDERBUFFER_RED_SIZE */ - 1309, /* GL_RENDERBUFFER_GREEN_SIZE */ - 1306, /* GL_RENDERBUFFER_BLUE_SIZE */ - 1304, /* GL_RENDERBUFFER_ALPHA_SIZE */ - 1307, /* GL_RENDERBUFFER_DEPTH_SIZE */ - 1316, /* GL_RENDERBUFFER_STENCIL_SIZE */ - 569, /* GL_FRAMEBUFFER_INCOMPLETE_MULTISAMPLE */ - 910, /* GL_MAX_SAMPLES */ - 1273, /* GL_QUADS_FOLLOW_PROVOKING_VERTEX_CONVENTION_EXT */ - 483, /* GL_FIRST_VERTEX_CONVENTION_EXT */ - 666, /* GL_LAST_VERTEX_CONVENTION_EXT */ - 1252, /* GL_PROVOKING_VERTEX_EXT */ + 1486, /* GL_STENCIL_ATTACHMENT */ + 537, /* GL_FRAMEBUFFER */ + 1307, /* GL_RENDERBUFFER */ + 1321, /* GL_RENDERBUFFER_WIDTH */ + 1314, /* GL_RENDERBUFFER_HEIGHT */ + 1316, /* GL_RENDERBUFFER_INTERNAL_FORMAT */ + 1509, /* GL_STENCIL_INDEX_EXT */ + 1506, /* GL_STENCIL_INDEX1_EXT */ + 1507, /* GL_STENCIL_INDEX4_EXT */ + 1508, /* GL_STENCIL_INDEX8_EXT */ + 1505, /* GL_STENCIL_INDEX16_EXT */ + 1318, /* GL_RENDERBUFFER_RED_SIZE */ + 1313, /* GL_RENDERBUFFER_GREEN_SIZE */ + 1310, /* GL_RENDERBUFFER_BLUE_SIZE */ + 1308, /* GL_RENDERBUFFER_ALPHA_SIZE */ + 1311, /* GL_RENDERBUFFER_DEPTH_SIZE */ + 1320, /* GL_RENDERBUFFER_STENCIL_SIZE */ + 570, /* GL_FRAMEBUFFER_INCOMPLETE_MULTISAMPLE */ + 912, /* GL_MAX_SAMPLES */ + 1276, /* GL_QUADS_FOLLOW_PROVOKING_VERTEX_CONVENTION */ + 483, /* GL_FIRST_VERTEX_CONVENTION */ + 667, /* GL_LAST_VERTEX_CONVENTION */ + 1254, /* GL_PROVOKING_VERTEX */ 302, /* GL_COPY_READ_BUFFER */ 303, /* GL_COPY_WRITE_BUFFER */ - 1364, /* GL_RGBA_SNORM */ - 1360, /* GL_RGBA8_SNORM */ - 1422, /* GL_SIGNED_NORMALIZED */ - 911, /* GL_MAX_SERVER_WAIT_TIMEOUT */ - 1040, /* GL_OBJECT_TYPE */ - 1526, /* GL_SYNC_CONDITION */ - 1531, /* GL_SYNC_STATUS */ - 1528, /* GL_SYNC_FLAGS */ - 1527, /* GL_SYNC_FENCE */ - 1530, /* GL_SYNC_GPU_COMMANDS_COMPLETE */ - 1757, /* GL_UNSIGNALED */ - 1421, /* GL_SIGNALED */ + 1368, /* GL_RGBA_SNORM */ + 1364, /* GL_RGBA8_SNORM */ + 1426, /* GL_SIGNED_NORMALIZED */ + 913, /* GL_MAX_SERVER_WAIT_TIMEOUT */ + 1042, /* GL_OBJECT_TYPE */ + 1530, /* GL_SYNC_CONDITION */ + 1535, /* GL_SYNC_STATUS */ + 1532, /* GL_SYNC_FLAGS */ + 1531, /* GL_SYNC_FENCE */ + 1534, /* GL_SYNC_GPU_COMMANDS_COMPLETE */ + 1761, /* GL_UNSIGNALED */ + 1425, /* GL_SIGNALED */ 46, /* GL_ALREADY_SIGNALED */ - 1728, /* GL_TIMEOUT_EXPIRED */ + 1732, /* GL_TIMEOUT_EXPIRED */ 270, /* GL_CONDITION_SATISFIED */ - 1840, /* GL_WAIT_FAILED */ + 1844, /* GL_WAIT_FAILED */ 468, /* GL_EVAL_BIT */ - 1284, /* GL_RASTER_POSITION_UNCLIPPED_IBM */ - 708, /* GL_LIST_BIT */ - 1623, /* GL_TEXTURE_BIT */ - 1395, /* GL_SCISSOR_BIT */ + 1288, /* GL_RASTER_POSITION_UNCLIPPED_IBM */ + 710, /* GL_LIST_BIT */ + 1627, /* GL_TEXTURE_BIT */ + 1399, /* GL_SCISSOR_BIT */ 29, /* GL_ALL_ATTRIB_BITS */ - 996, /* GL_MULTISAMPLE_BIT */ + 998, /* GL_MULTISAMPLE_BIT */ 30, /* GL_ALL_CLIENT_ATTRIB_BITS */ - 1729, /* GL_TIMEOUT_IGNORED */ + 1733, /* GL_TIMEOUT_IGNORED */ }; typedef int (*cfunc)(const void *, const void *); diff --git a/src/mesa/main/execmem.c b/src/mesa/main/execmem.c index 57c1e117c89..4c6139985fe 100644 --- a/src/mesa/main/execmem.c +++ b/src/mesa/main/execmem.c @@ -80,11 +80,10 @@ init_heap(void) exec_heap = mmInit( 0, EXEC_HEAP_SIZE ); if (!exec_mem) - exec_mem = (unsigned char *) mmap(0, EXEC_HEAP_SIZE, - PROT_EXEC | PROT_READ | PROT_WRITE, - MAP_PRIVATE | MAP_ANONYMOUS, -1, 0); + exec_mem = mmap(NULL, EXEC_HEAP_SIZE, PROT_EXEC | PROT_READ | PROT_WRITE, + MAP_PRIVATE | MAP_ANONYMOUS, -1, 0); - return (exec_mem != NULL); + return (exec_mem != MAP_FAILED); } diff --git a/src/mesa/main/extensions.c b/src/mesa/main/extensions.c index 73adb0e2176..2992abd0751 100644 --- a/src/mesa/main/extensions.c +++ b/src/mesa/main/extensions.c @@ -63,6 +63,7 @@ static const struct { { OFF, "GL_ARB_pixel_buffer_object", F(EXT_pixel_buffer_object) }, { OFF, "GL_ARB_point_parameters", F(EXT_point_parameters) }, { OFF, "GL_ARB_point_sprite", F(ARB_point_sprite) }, + { OFF, "GL_ARB_provoking_vertex", F(EXT_provoking_vertex) }, { OFF, "GL_ARB_seamless_cube_map", F(ARB_seamless_cube_map) }, { OFF, "GL_ARB_shader_objects", F(ARB_shader_objects) }, { OFF, "GL_ARB_shading_language_100", F(ARB_shading_language_100) }, diff --git a/src/mesa/main/fbobject.c b/src/mesa/main/fbobject.c index 825a23090b5..13f49da5a78 100644 --- a/src/mesa/main/fbobject.c +++ b/src/mesa/main/fbobject.c @@ -46,6 +46,10 @@ #include "texstore.h" +/** Set this to 1 to help debug FBO incompleteness problems */ +#define DEBUG_FBO 0 + + /** * Notes: * @@ -308,8 +312,8 @@ _mesa_framebuffer_renderbuffer(GLcontext *ctx, struct gl_framebuffer *fb, static void att_incomplete(const char *msg) { -#if 0 - _mesa_printf("attachment incomplete: %s\n", msg); +#if DEBUG_FBO + _mesa_debug(NULL, "attachment incomplete: %s\n", msg); #else (void) msg; #endif @@ -317,6 +321,23 @@ att_incomplete(const char *msg) /** + * For debug only. + */ +static void +fbo_incomplete(const char *msg, int index) +{ +#if DEBUG_FBO + _mesa_debug(NULL, "FBO Incomplete: %s [%d]\n", msg, index); +#else + (void) msg; + (void) index; +#endif +} + + + + +/** * Test if an attachment point is complete and update its Complete field. * \param format if GL_COLOR, this is a color attachment point, * if GL_DEPTH, this is a depth component attachment point, @@ -468,20 +489,6 @@ test_attachment_completeness(const GLcontext *ctx, GLenum format, /** - * Helpful for debugging - */ -static void -fbo_incomplete(const char *msg, int index) -{ - (void) msg; - (void) index; - /* - _mesa_debug(NULL, "FBO Incomplete: %s [%d]\n", msg, index); - */ -} - - -/** * Test if the given framebuffer object is complete and update its * Status field with the results. * Calls the ctx->Driver.ValidateFramebuffer() function to allow the @@ -1953,13 +1960,13 @@ _mesa_GenerateMipmapEXT(GLenum target) _mesa_lock_texture(ctx, texObj); if (target == GL_TEXTURE_CUBE_MAP) { - int face; - + GLuint face; for (face = 0; face < 6; face++) ctx->Driver.GenerateMipmap(ctx, GL_TEXTURE_CUBE_MAP_POSITIVE_X_ARB + face, texObj); - } else { + } + else { ctx->Driver.GenerateMipmap(ctx, target, texObj); } _mesa_unlock_texture(ctx, texObj); diff --git a/src/mesa/main/mfeatures.h b/src/mesa/main/mfeatures.h index e23cdb1f426..6318934c6bf 100644 --- a/src/mesa/main/mfeatures.h +++ b/src/mesa/main/mfeatures.h @@ -36,6 +36,38 @@ #define _HAVE_FULL_GL 1 #endif +/* assert that a feature is disabled and should never be used */ +#define ASSERT_NO_FEATURE() ASSERT(0) + +/** + * A feature can be anything. But most of them share certain characteristics. + * + * When a feature defines driver entries, they can be initialized by + * _MESA_INIT_<FEATURE>_FUNCTIONS + * + * When a feature defines vtxfmt entries, they can be initialized and + * installed by + * _MESA_INIT_<FEATURE>_VTXFMT + * _mesa_install_<feature>_vtxfmt + * + * When a feature defines dispatch entries, they are initialized by + * _mesa_init_<feature>_dispatch + * + * When a feature has states, they are initialized and freed by + * _mesa_init_<feature> + * _mesa_free_<feature>_data + * + * Except for states, the others compile to no-op when a feature is disabled. + * + * The GLAPIENTRYs and helper functions defined by a feature should also + * compile to no-op when it is disabled. But to save typings and to catch + * bugs, some of them may be unavailable, or compile to ASSERT_NO_FEATURE() + * when the feature is disabled. + * + * A feature following the conventions may be used without knowing if it is + * enabled or not. + */ + #define FEATURE_accum _HAVE_FULL_GL #define FEATURE_attrib_stack _HAVE_FULL_GL #define FEATURE_colortable _HAVE_FULL_GL diff --git a/src/mesa/main/mtypes.h b/src/mesa/main/mtypes.h index 823c6dc29c6..d7bf7689f3c 100644 --- a/src/mesa/main/mtypes.h +++ b/src/mesa/main/mtypes.h @@ -1834,7 +1834,6 @@ struct gl_vertex_program struct gl_program Base; /**< base class */ GLboolean IsNVProgram; /**< is this a GL_NV_vertex_program program? */ GLboolean IsPositionInvariant; - void *TnlData; /**< should probably use Base.DriverData */ }; @@ -2080,6 +2079,7 @@ struct gl_shader_state GLboolean EmitContReturn; /**< Emit CONT/RET opcodes? */ GLboolean EmitCondCodes; /**< Use condition codes? */ GLboolean EmitComments; /**< Annotated instructions */ + GLboolean EmitNVTempInitialization; /**< 0-fill NV temp registers */ void *MemPool; GLbitfield Flags; /**< Mask of GLSL_x flags */ struct gl_sl_pragmas DefaultPragmas; /**< Default #pragma settings */ diff --git a/src/mesa/main/pixel.c b/src/mesa/main/pixel.c index fcef6dfd422..3820ebd8894 100644 --- a/src/mesa/main/pixel.c +++ b/src/mesa/main/pixel.c @@ -36,13 +36,17 @@ #include "macros.h" #include "pixel.h" #include "mtypes.h" +#include "glapi/dispatch.h" + + +#if FEATURE_pixel_transfer /**********************************************************************/ /***** glPixelZoom *****/ /**********************************************************************/ -void GLAPIENTRY +static void GLAPIENTRY _mesa_PixelZoom( GLfloat xfactor, GLfloat yfactor ) { GET_CURRENT_CONTEXT(ctx); @@ -163,7 +167,7 @@ validate_pbo_access(GLcontext *ctx, struct gl_pixelstore_attrib *pack, } -void GLAPIENTRY +static void GLAPIENTRY _mesa_PixelMapfv( GLenum map, GLsizei mapsize, const GLfloat *values ) { GET_CURRENT_CONTEXT(ctx); @@ -205,7 +209,7 @@ _mesa_PixelMapfv( GLenum map, GLsizei mapsize, const GLfloat *values ) } -void GLAPIENTRY +static void GLAPIENTRY _mesa_PixelMapuiv(GLenum map, GLsizei mapsize, const GLuint *values ) { GLfloat fvalues[MAX_PIXEL_MAP_TABLE]; @@ -261,7 +265,7 @@ _mesa_PixelMapuiv(GLenum map, GLsizei mapsize, const GLuint *values ) } -void GLAPIENTRY +static void GLAPIENTRY _mesa_PixelMapusv(GLenum map, GLsizei mapsize, const GLushort *values ) { GLfloat fvalues[MAX_PIXEL_MAP_TABLE]; @@ -317,7 +321,7 @@ _mesa_PixelMapusv(GLenum map, GLsizei mapsize, const GLushort *values ) } -void GLAPIENTRY +static void GLAPIENTRY _mesa_GetPixelMapfv( GLenum map, GLfloat *values ) { GET_CURRENT_CONTEXT(ctx); @@ -362,7 +366,7 @@ _mesa_GetPixelMapfv( GLenum map, GLfloat *values ) } -void GLAPIENTRY +static void GLAPIENTRY _mesa_GetPixelMapuiv( GLenum map, GLuint *values ) { GET_CURRENT_CONTEXT(ctx); @@ -406,7 +410,7 @@ _mesa_GetPixelMapuiv( GLenum map, GLuint *values ) } -void GLAPIENTRY +static void GLAPIENTRY _mesa_GetPixelMapusv( GLenum map, GLushort *values ) { GET_CURRENT_CONTEXT(ctx); @@ -468,7 +472,7 @@ _mesa_GetPixelMapusv( GLenum map, GLushort *values ) * Implements glPixelTransfer[fi] whether called immediately or from a * display list. */ -void GLAPIENTRY +static void GLAPIENTRY _mesa_PixelTransferf( GLenum pname, GLfloat param ) { GET_CURRENT_CONTEXT(ctx); @@ -662,7 +666,7 @@ _mesa_PixelTransferf( GLenum pname, GLfloat param ) } -void GLAPIENTRY +static void GLAPIENTRY _mesa_PixelTransferi( GLenum pname, GLint param ) { _mesa_PixelTransferf( pname, (GLfloat) param ); @@ -756,6 +760,24 @@ void _mesa_update_pixel( GLcontext *ctx, GLuint new_state ) } +void +_mesa_init_pixel_dispatch(struct _glapi_table *disp) +{ + SET_GetPixelMapfv(disp, _mesa_GetPixelMapfv); + SET_GetPixelMapuiv(disp, _mesa_GetPixelMapuiv); + SET_GetPixelMapusv(disp, _mesa_GetPixelMapusv); + SET_PixelMapfv(disp, _mesa_PixelMapfv); + SET_PixelMapuiv(disp, _mesa_PixelMapuiv); + SET_PixelMapusv(disp, _mesa_PixelMapusv); + SET_PixelTransferf(disp, _mesa_PixelTransferf); + SET_PixelTransferi(disp, _mesa_PixelTransferi); + SET_PixelZoom(disp, _mesa_PixelZoom); +} + + +#endif /* FEATURE_pixel_transfer */ + + /**********************************************************************/ /***** Initialization *****/ /**********************************************************************/ diff --git a/src/mesa/main/pixel.h b/src/mesa/main/pixel.h index cb6c5262a39..f4d3f1efdb0 100644 --- a/src/mesa/main/pixel.h +++ b/src/mesa/main/pixel.h @@ -33,48 +33,35 @@ #define PIXEL_H -#include "mtypes.h" +#include "main/mtypes.h" -/** \name API functions */ -/*@{*/ +#if FEATURE_pixel_transfer -extern void GLAPIENTRY -_mesa_GetPixelMapfv( GLenum map, GLfloat *values ); - -extern void GLAPIENTRY -_mesa_GetPixelMapuiv( GLenum map, GLuint *values ); - -extern void GLAPIENTRY -_mesa_GetPixelMapusv( GLenum map, GLushort *values ); - -extern void GLAPIENTRY -_mesa_PixelMapfv( GLenum map, GLsizei mapsize, const GLfloat *values ); - -extern void GLAPIENTRY -_mesa_PixelMapuiv(GLenum map, GLsizei mapsize, const GLuint *values ); - -extern void GLAPIENTRY -_mesa_PixelMapusv(GLenum map, GLsizei mapsize, const GLushort *values ); +extern void +_mesa_update_pixel( GLcontext *ctx, GLuint newstate ); -extern void GLAPIENTRY -_mesa_PixelTransferf( GLenum pname, GLfloat param ); +extern void +_mesa_init_pixel_dispatch( struct _glapi_table * disp ); -extern void GLAPIENTRY -_mesa_PixelTransferi( GLenum pname, GLint param ); +#else /* FEATURE_pixel_transfer */ -extern void GLAPIENTRY -_mesa_PixelZoom( GLfloat xfactor, GLfloat yfactor ); +static INLINE void +_mesa_update_pixel(GLcontext *ctx, GLuint newstate) +{ +} -/*@}*/ +static INLINE void +_mesa_init_pixel_dispatch(struct _glapi_table *disp) +{ +} +#endif /* FEATURE_pixel_transfer */ -extern void -_mesa_update_pixel( GLcontext *ctx, GLuint newstate ); extern void _mesa_init_pixel( GLcontext * ctx ); /*@}*/ -#endif +#endif /* PIXEL_H */ diff --git a/src/mesa/main/state.c b/src/mesa/main/state.c index 140a998df2e..f10e6b04b7c 100644 --- a/src/mesa/main/state.c +++ b/src/mesa/main/state.c @@ -40,9 +40,7 @@ #include "framebuffer.h" #include "light.h" #include "matrix.h" -#if FEATURE_pixel_transfer #include "pixel.h" -#endif #include "shader/program.h" #include "shader/prog_parameter.h" #include "state.h" @@ -585,10 +583,8 @@ _mesa_update_state_locked( GLcontext *ctx ) if (new_state & (_NEW_STENCIL | _NEW_BUFFERS)) _mesa_update_stencil( ctx ); -#if FEATURE_pixel_transfer if (new_state & _MESA_NEW_TRANSFER_STATE) _mesa_update_pixel( ctx, new_state ); -#endif if (new_state & _DD_NEW_SEPARATE_SPECULAR) update_separate_specular( ctx ); diff --git a/src/mesa/main/texenvprogram.c b/src/mesa/main/texenvprogram.c index 2f3e47e69ec..d7e77e759e0 100644 --- a/src/mesa/main/texenvprogram.c +++ b/src/mesa/main/texenvprogram.c @@ -276,6 +276,7 @@ need_saturate( GLuint mode ) return GL_TRUE; default: assert(0); + return GL_FALSE; } } diff --git a/src/mesa/main/texgen.c b/src/mesa/main/texgen.c index e3feb024c31..b3ecfc784e3 100644 --- a/src/mesa/main/texgen.c +++ b/src/mesa/main/texgen.c @@ -35,6 +35,7 @@ #include "main/enums.h" #include "main/macros.h" #include "main/texgen.h" +#include "main/texstate.h" #include "math/m_matrix.h" @@ -79,7 +80,7 @@ _mesa_TexGenfv( GLenum coord, GLenum pname, const GLfloat *params ) return; } - texUnit = &ctx->Texture.Unit[ctx->Texture.CurrentUnit]; + texUnit = _mesa_get_current_tex_unit(ctx); texgen = get_texgen(texUnit, coord); if (!texgen) { @@ -231,7 +232,7 @@ _mesa_GetTexGendv( GLenum coord, GLenum pname, GLdouble *params ) return; } - texUnit = &ctx->Texture.Unit[ctx->Texture.CurrentUnit]; + texUnit = _mesa_get_current_tex_unit(ctx); texgen = get_texgen(texUnit, coord); if (!texgen) { @@ -269,7 +270,7 @@ _mesa_GetTexGenfv( GLenum coord, GLenum pname, GLfloat *params ) return; } - texUnit = &ctx->Texture.Unit[ctx->Texture.CurrentUnit]; + texUnit = _mesa_get_current_tex_unit(ctx); texgen = get_texgen(texUnit, coord); if (!texgen) { @@ -307,7 +308,7 @@ _mesa_GetTexGeniv( GLenum coord, GLenum pname, GLint *params ) return; } - texUnit = &ctx->Texture.Unit[ctx->Texture.CurrentUnit]; + texUnit = _mesa_get_current_tex_unit(ctx); texgen = get_texgen(texUnit, coord); if (!texgen) { diff --git a/src/mesa/main/teximage.c b/src/mesa/main/teximage.c index 82283030409..465da6b0469 100644 --- a/src/mesa/main/teximage.c +++ b/src/mesa/main/teximage.c @@ -33,9 +33,8 @@ #include "glheader.h" #include "bufferobj.h" #include "context.h" -#if FEATURE_convolve #include "convolve.h" -#endif +#include "enums.h" #include "fbobject.h" #include "framebuffer.h" #include "hash.h" @@ -2075,6 +2074,23 @@ update_fbo_texture(GLcontext *ctx, struct gl_texture_object *texObj, } +/** + * If the texture object's GenerateMipmap flag is set and we've + * changed the texture base level image, regenerate the rest of the + * mipmap levels now. + */ +static INLINE void +check_gen_mipmap(GLcontext *ctx, GLenum target, + struct gl_texture_object *texObj, GLint level) +{ + ASSERT(target != GL_TEXTURE_CUBE_MAP); + if (texObj->GenerateMipmap && level == texObj->BaseLevel) { + ASSERT(ctx->Driver.GenerateMipmap); + ctx->Driver.GenerateMipmap(ctx, target, texObj); + } +} + + /** Debug helper: override the user-requested internal format */ static GLenum override_internal_format(GLenum internalFormat, GLint width, GLint height) @@ -2131,6 +2147,13 @@ _mesa_TexImage1D( GLenum target, GLint level, GLint internalFormat, GET_CURRENT_CONTEXT(ctx); ASSERT_OUTSIDE_BEGIN_END_AND_FLUSH(ctx); + if (MESA_VERBOSE & (VERBOSE_API|VERBOSE_TEXTURE)) + _mesa_debug(ctx, "glTexImage1D %s %d %s %d %d %s %s %p\n", + _mesa_lookup_enum_by_nr(target), level, + _mesa_lookup_enum_by_nr(internalFormat), width, border, + _mesa_lookup_enum_by_nr(format), + _mesa_lookup_enum_by_nr(type), pixels); + internalFormat = override_internal_format(internalFormat, width, 1); #if FEATURE_convolve @@ -2161,36 +2184,36 @@ _mesa_TexImage1D( GLenum target, GLint level, GLint internalFormat, texImage = _mesa_get_tex_image(ctx, texObj, target, level); if (!texImage) { _mesa_error(ctx, GL_OUT_OF_MEMORY, "glTexImage1D"); - goto out; - } - - if (texImage->Data) { - ctx->Driver.FreeTexImageData( ctx, texImage ); } + else { + if (texImage->Data) { + ctx->Driver.FreeTexImageData( ctx, texImage ); + } + + ASSERT(texImage->Data == NULL); + + clear_teximage_fields(texImage); /* not really needed, but helpful */ + _mesa_init_teximage_fields(ctx, target, texImage, + postConvWidth, 1, 1, + border, internalFormat); + + /* Give the texture to the driver. <pixels> may be null. */ + ASSERT(ctx->Driver.TexImage1D); + ctx->Driver.TexImage1D(ctx, target, level, internalFormat, + width, border, format, type, pixels, + &ctx->Unpack, texObj, texImage); + + ASSERT(texImage->TexFormat); - ASSERT(texImage->Data == NULL); - - clear_teximage_fields(texImage); /* not really needed, but helpful */ - _mesa_init_teximage_fields(ctx, target, texImage, - postConvWidth, 1, 1, - border, internalFormat); - - ASSERT(ctx->Driver.TexImage1D); - - /* Give the texture to the driver! <pixels> may be null! */ - (*ctx->Driver.TexImage1D)(ctx, target, level, internalFormat, - width, border, format, type, pixels, - &ctx->Unpack, texObj, texImage); - - ASSERT(texImage->TexFormat); - - update_fbo_texture(ctx, texObj, face, level); - - /* state update */ - texObj->_Complete = GL_FALSE; - ctx->NewState |= _NEW_TEXTURE; - } - out: + check_gen_mipmap(ctx, target, texObj, level); + + update_fbo_texture(ctx, texObj, face, level); + + /* state update */ + texObj->_Complete = GL_FALSE; + ctx->NewState |= _NEW_TEXTURE; + } + } _mesa_unlock_texture(ctx, texObj); } else if (target == GL_PROXY_TEXTURE_1D) { @@ -2230,6 +2253,13 @@ _mesa_TexImage2D( GLenum target, GLint level, GLint internalFormat, GET_CURRENT_CONTEXT(ctx); ASSERT_OUTSIDE_BEGIN_END_AND_FLUSH(ctx); + if (MESA_VERBOSE & (VERBOSE_API|VERBOSE_TEXTURE)) + _mesa_debug(ctx, "glTexImage2D %s %d %s %d %d %d %s %s %p\n", + _mesa_lookup_enum_by_nr(target), level, + _mesa_lookup_enum_by_nr(internalFormat), width, height, + border, _mesa_lookup_enum_by_nr(format), + _mesa_lookup_enum_by_nr(type), pixels); + internalFormat = override_internal_format(internalFormat, width, height); #if FEATURE_convolve @@ -2269,35 +2299,35 @@ _mesa_TexImage2D( GLenum target, GLint level, GLint internalFormat, texImage = _mesa_get_tex_image(ctx, texObj, target, level); if (!texImage) { _mesa_error(ctx, GL_OUT_OF_MEMORY, "glTexImage2D"); - goto out; - } - - if (texImage->Data) { - ctx->Driver.FreeTexImageData( ctx, texImage ); } - - ASSERT(texImage->Data == NULL); - clear_teximage_fields(texImage); /* not really needed, but helpful */ - _mesa_init_teximage_fields(ctx, target, texImage, - postConvWidth, postConvHeight, 1, - border, internalFormat); - - ASSERT(ctx->Driver.TexImage2D); - - /* Give the texture to the driver! <pixels> may be null! */ - (*ctx->Driver.TexImage2D)(ctx, target, level, internalFormat, - width, height, border, format, type, pixels, - &ctx->Unpack, texObj, texImage); - - ASSERT(texImage->TexFormat); - - update_fbo_texture(ctx, texObj, face, level); - - /* state update */ - texObj->_Complete = GL_FALSE; - ctx->NewState |= _NEW_TEXTURE; - } - out: + else { + if (texImage->Data) { + ctx->Driver.FreeTexImageData( ctx, texImage ); + } + + ASSERT(texImage->Data == NULL); + clear_teximage_fields(texImage); /* not really needed, but helpful */ + _mesa_init_teximage_fields(ctx, target, texImage, + postConvWidth, postConvHeight, 1, + border, internalFormat); + + /* Give the texture to the driver. <pixels> may be null. */ + ASSERT(ctx->Driver.TexImage2D); + ctx->Driver.TexImage2D(ctx, target, level, internalFormat, + width, height, border, format, type, + pixels, &ctx->Unpack, texObj, texImage); + + ASSERT(texImage->TexFormat); + + check_gen_mipmap(ctx, target, texObj, level); + + update_fbo_texture(ctx, texObj, face, level); + + /* state update */ + texObj->_Complete = GL_FALSE; + ctx->NewState |= _NEW_TEXTURE; + } + } _mesa_unlock_texture(ctx, texObj); } else if (target == GL_PROXY_TEXTURE_2D || @@ -2346,6 +2376,13 @@ _mesa_TexImage3D( GLenum target, GLint level, GLint internalFormat, GET_CURRENT_CONTEXT(ctx); ASSERT_OUTSIDE_BEGIN_END_AND_FLUSH(ctx); + if (MESA_VERBOSE & (VERBOSE_API|VERBOSE_TEXTURE)) + _mesa_debug(ctx, "glTexImage3D %s %d %s %d %d %d %d %s %s %p\n", + _mesa_lookup_enum_by_nr(target), level, + _mesa_lookup_enum_by_nr(internalFormat), width, height, + depth, border, _mesa_lookup_enum_by_nr(format), + _mesa_lookup_enum_by_nr(type), pixels); + internalFormat = override_internal_format(internalFormat, width, height); if (target == GL_TEXTURE_3D || @@ -2372,35 +2409,35 @@ _mesa_TexImage3D( GLenum target, GLint level, GLint internalFormat, texImage = _mesa_get_tex_image(ctx, texObj, target, level); if (!texImage) { _mesa_error(ctx, GL_OUT_OF_MEMORY, "glTexImage3D"); - goto out; - } - - if (texImage->Data) { - ctx->Driver.FreeTexImageData( ctx, texImage ); } - - ASSERT(texImage->Data == NULL); - clear_teximage_fields(texImage); /* not really needed, but helpful */ - _mesa_init_teximage_fields(ctx, target, texImage, - width, height, depth, - border, internalFormat); + else { + if (texImage->Data) { + ctx->Driver.FreeTexImageData( ctx, texImage ); + } + + ASSERT(texImage->Data == NULL); + clear_teximage_fields(texImage); /* not really needed, but helpful */ + _mesa_init_teximage_fields(ctx, target, texImage, + width, height, depth, + border, internalFormat); - ASSERT(ctx->Driver.TexImage3D); + /* Give the texture to the driver. <pixels> may be null. */ + ASSERT(ctx->Driver.TexImage3D); + ctx->Driver.TexImage3D(ctx, target, level, internalFormat, + width, height, depth, border, format, type, + pixels, &ctx->Unpack, texObj, texImage); - /* Give the texture to the driver! <pixels> may be null! */ - (*ctx->Driver.TexImage3D)(ctx, target, level, internalFormat, - width, height, depth, border, format, type, - pixels, &ctx->Unpack, texObj, texImage); + ASSERT(texImage->TexFormat); - ASSERT(texImage->TexFormat); + check_gen_mipmap(ctx, target, texObj, level); - update_fbo_texture(ctx, texObj, face, level); + update_fbo_texture(ctx, texObj, face, level); - /* state update */ - texObj->_Complete = GL_FALSE; - ctx->NewState |= _NEW_TEXTURE; + /* state update */ + texObj->_Complete = GL_FALSE; + ctx->NewState |= _NEW_TEXTURE; + } } - out: _mesa_unlock_texture(ctx, texObj); } else if (target == GL_PROXY_TEXTURE_3D || @@ -2455,6 +2492,12 @@ _mesa_TexSubImage1D( GLenum target, GLint level, GET_CURRENT_CONTEXT(ctx); ASSERT_OUTSIDE_BEGIN_END_AND_FLUSH(ctx); + if (MESA_VERBOSE & (VERBOSE_API|VERBOSE_TEXTURE)) + _mesa_debug(ctx, "glTexSubImage1D %s %d %d %d %s %s %p\n", + _mesa_lookup_enum_by_nr(target), level, + xoffset, width, _mesa_lookup_enum_by_nr(format), + _mesa_lookup_enum_by_nr(type), pixels); + if (ctx->NewState & _MESA_NEW_TRANSFER_STATE) _mesa_update_state(ctx); @@ -2480,23 +2523,24 @@ _mesa_TexSubImage1D( GLenum target, GLint level, texImage = _mesa_select_tex_image(ctx, texObj, target, level); if (subtexture_error_check2(ctx, 1, target, level, xoffset, 0, 0, - postConvWidth, 1, 1, format, type, texImage)) { - goto out; /* error was detected */ + postConvWidth, 1, 1, + format, type, texImage)) { + /* error was recorded */ } + else if (width > 0) { + /* If we have a border, xoffset=-1 is legal. Bias by border width */ + xoffset += texImage->Border; - if (width == 0) - goto out; /* no-op, not an error */ + ASSERT(ctx->Driver.TexSubImage1D); + ctx->Driver.TexSubImage1D(ctx, target, level, xoffset, width, + format, type, pixels, &ctx->Unpack, + texObj, texImage); - /* If we have a border, xoffset=-1 is legal. Bias by border width */ - xoffset += texImage->Border; + check_gen_mipmap(ctx, target, texObj, level); - ASSERT(ctx->Driver.TexSubImage1D); - (*ctx->Driver.TexSubImage1D)(ctx, target, level, xoffset, width, - format, type, pixels, &ctx->Unpack, - texObj, texImage); - ctx->NewState |= _NEW_TEXTURE; + ctx->NewState |= _NEW_TEXTURE; + } } - out: _mesa_unlock_texture(ctx, texObj); } @@ -2515,6 +2559,13 @@ _mesa_TexSubImage2D( GLenum target, GLint level, GET_CURRENT_CONTEXT(ctx); ASSERT_OUTSIDE_BEGIN_END_AND_FLUSH(ctx); + if (MESA_VERBOSE & (VERBOSE_API|VERBOSE_TEXTURE)) + _mesa_debug(ctx, "glTexSubImage2D %s %d %d %d %d %d %s %s %p\n", + _mesa_lookup_enum_by_nr(target), level, + xoffset, yoffset, width, height, + _mesa_lookup_enum_by_nr(format), + _mesa_lookup_enum_by_nr(type), pixels); + if (ctx->NewState & _MESA_NEW_TRANSFER_STATE) _mesa_update_state(ctx); @@ -2533,30 +2584,31 @@ _mesa_TexSubImage2D( GLenum target, GLint level, texUnit = _mesa_get_current_tex_unit(ctx); texObj = _mesa_select_tex_object(ctx, texUnit, target); + _mesa_lock_texture(ctx, texObj); { texImage = _mesa_select_tex_image(ctx, texObj, target, level); if (subtexture_error_check2(ctx, 2, target, level, xoffset, yoffset, 0, - postConvWidth, postConvHeight, 1, format, type, - texImage)) { - goto out; /* error was detected */ + postConvWidth, postConvHeight, 1, + format, type, texImage)) { + /* error was recorded */ } + else if (width > 0 && height >= 0) { + /* If we have a border, xoffset=-1 is legal. Bias by border width */ + xoffset += texImage->Border; + yoffset += texImage->Border; + + ASSERT(ctx->Driver.TexSubImage2D); + ctx->Driver.TexSubImage2D(ctx, target, level, xoffset, yoffset, + width, height, format, type, pixels, + &ctx->Unpack, texObj, texImage); - if (width == 0 || height == 0) - goto out; /* no-op, not an error */ + check_gen_mipmap(ctx, target, texObj, level); - /* If we have a border, xoffset=-1 is legal. Bias by border width */ - xoffset += texImage->Border; - yoffset += texImage->Border; - - ASSERT(ctx->Driver.TexSubImage2D); - (*ctx->Driver.TexSubImage2D)(ctx, target, level, xoffset, yoffset, - width, height, format, type, pixels, - &ctx->Unpack, texObj, texImage); - ctx->NewState |= _NEW_TEXTURE; + ctx->NewState |= _NEW_TEXTURE; + } } - out: _mesa_unlock_texture(ctx, texObj); } @@ -2575,6 +2627,13 @@ _mesa_TexSubImage3D( GLenum target, GLint level, GET_CURRENT_CONTEXT(ctx); ASSERT_OUTSIDE_BEGIN_END_AND_FLUSH(ctx); + if (MESA_VERBOSE & (VERBOSE_API|VERBOSE_TEXTURE)) + _mesa_debug(ctx, "glTexSubImage3D %s %d %d %d %d %d %d %d %s %s %p\n", + _mesa_lookup_enum_by_nr(target), level, + xoffset, yoffset, zoffset, width, height, depth, + _mesa_lookup_enum_by_nr(format), + _mesa_lookup_enum_by_nr(type), pixels); + if (ctx->NewState & _MESA_NEW_TRANSFER_STATE) _mesa_update_state(ctx); @@ -2590,28 +2649,30 @@ _mesa_TexSubImage3D( GLenum target, GLint level, { texImage = _mesa_select_tex_image(ctx, texObj, target, level); - if (subtexture_error_check2(ctx, 3, target, level, xoffset, yoffset, zoffset, - width, height, depth, format, type, texImage)) { - goto out; /* error was detected */ + if (subtexture_error_check2(ctx, 3, target, level, + xoffset, yoffset, zoffset, + width, height, depth, + format, type, texImage)) { + /* error was recorded */ } + else if (width > 0 && height > 0 && height > 0) { + /* If we have a border, xoffset=-1 is legal. Bias by border width */ + xoffset += texImage->Border; + yoffset += texImage->Border; + zoffset += texImage->Border; - if (width == 0 || height == 0 || height == 0) - goto out; /* no-op, not an error */ + ASSERT(ctx->Driver.TexSubImage3D); + ctx->Driver.TexSubImage3D(ctx, target, level, + xoffset, yoffset, zoffset, + width, height, depth, + format, type, pixels, + &ctx->Unpack, texObj, texImage ); - /* If we have a border, xoffset=-1 is legal. Bias by border width */ - xoffset += texImage->Border; - yoffset += texImage->Border; - zoffset += texImage->Border; + check_gen_mipmap(ctx, target, texObj, level); - ASSERT(ctx->Driver.TexSubImage3D); - (*ctx->Driver.TexSubImage3D)(ctx, target, level, - xoffset, yoffset, zoffset, - width, height, depth, - format, type, pixels, - &ctx->Unpack, texObj, texImage ); - ctx->NewState |= _NEW_TEXTURE; + ctx->NewState |= _NEW_TEXTURE; + } } - out: _mesa_unlock_texture(ctx, texObj); } @@ -2631,6 +2692,12 @@ _mesa_CopyTexImage1D( GLenum target, GLint level, GET_CURRENT_CONTEXT(ctx); ASSERT_OUTSIDE_BEGIN_END_AND_FLUSH(ctx); + if (MESA_VERBOSE & (VERBOSE_API|VERBOSE_TEXTURE)) + _mesa_debug(ctx, "glCopyTexImage1D %s %d %s %d %d %d %d\n", + _mesa_lookup_enum_by_nr(target), level, + _mesa_lookup_enum_by_nr(internalFormat), + x, y, width, border); + if (ctx->NewState & NEW_COPY_TEX_STATE) _mesa_update_state(ctx); @@ -2646,38 +2713,39 @@ _mesa_CopyTexImage1D( GLenum target, GLint level, texUnit = _mesa_get_current_tex_unit(ctx); texObj = _mesa_select_tex_object(ctx, texUnit, target); + _mesa_lock_texture(ctx, texObj); { texImage = _mesa_get_tex_image(ctx, texObj, target, level); if (!texImage) { _mesa_error(ctx, GL_OUT_OF_MEMORY, "glCopyTexImage1D"); - goto out; } + else { + if (texImage->Data) { + ctx->Driver.FreeTexImageData( ctx, texImage ); + } - if (texImage->Data) { - ctx->Driver.FreeTexImageData( ctx, texImage ); - } - - ASSERT(texImage->Data == NULL); + ASSERT(texImage->Data == NULL); - clear_teximage_fields(texImage); /* not really needed, but helpful */ - _mesa_init_teximage_fields(ctx, target, texImage, postConvWidth, 1, 1, - border, internalFormat); + clear_teximage_fields(texImage); /* not really needed, but helpful */ + _mesa_init_teximage_fields(ctx, target, texImage, postConvWidth, 1, 1, + border, internalFormat); + ASSERT(ctx->Driver.CopyTexImage1D); + ctx->Driver.CopyTexImage1D(ctx, target, level, internalFormat, + x, y, width, border); - ASSERT(ctx->Driver.CopyTexImage1D); - (*ctx->Driver.CopyTexImage1D)(ctx, target, level, internalFormat, - x, y, width, border); + ASSERT(texImage->TexFormat); - ASSERT(texImage->TexFormat); + check_gen_mipmap(ctx, target, texObj, level); - update_fbo_texture(ctx, texObj, face, level); + update_fbo_texture(ctx, texObj, face, level); - /* state update */ - texObj->_Complete = GL_FALSE; - ctx->NewState |= _NEW_TEXTURE; + /* state update */ + texObj->_Complete = GL_FALSE; + ctx->NewState |= _NEW_TEXTURE; + } } - out: _mesa_unlock_texture(ctx, texObj); } @@ -2696,6 +2764,12 @@ _mesa_CopyTexImage2D( GLenum target, GLint level, GLenum internalFormat, GET_CURRENT_CONTEXT(ctx); ASSERT_OUTSIDE_BEGIN_END_AND_FLUSH(ctx); + if (MESA_VERBOSE & (VERBOSE_API|VERBOSE_TEXTURE)) + _mesa_debug(ctx, "glCopyTexImage2D %s %d %s %d %d %d %d %d\n", + _mesa_lookup_enum_by_nr(target), level, + _mesa_lookup_enum_by_nr(internalFormat), + x, y, width, height, border); + if (ctx->NewState & NEW_COPY_TEX_STATE) _mesa_update_state(ctx); @@ -2719,33 +2793,34 @@ _mesa_CopyTexImage2D( GLenum target, GLint level, GLenum internalFormat, if (!texImage) { _mesa_error(ctx, GL_OUT_OF_MEMORY, "glCopyTexImage2D"); - goto out; } - - if (texImage->Data) { - ctx->Driver.FreeTexImageData( ctx, texImage ); - } - - ASSERT(texImage->Data == NULL); + else { + if (texImage->Data) { + ctx->Driver.FreeTexImageData( ctx, texImage ); + } + + ASSERT(texImage->Data == NULL); + + clear_teximage_fields(texImage); /* not really needed, but helpful */ + _mesa_init_teximage_fields(ctx, target, texImage, + postConvWidth, postConvHeight, 1, + border, internalFormat); + + ASSERT(ctx->Driver.CopyTexImage2D); + ctx->Driver.CopyTexImage2D(ctx, target, level, internalFormat, + x, y, width, height, border); - clear_teximage_fields(texImage); /* not really needed, but helpful */ - _mesa_init_teximage_fields(ctx, target, texImage, - postConvWidth, postConvHeight, 1, - border, internalFormat); - - ASSERT(ctx->Driver.CopyTexImage2D); - (*ctx->Driver.CopyTexImage2D)(ctx, target, level, internalFormat, - x, y, width, height, border); - - ASSERT(texImage->TexFormat); + ASSERT(texImage->TexFormat); - update_fbo_texture(ctx, texObj, face, level); + check_gen_mipmap(ctx, target, texObj, level); - /* state update */ - texObj->_Complete = GL_FALSE; - ctx->NewState |= _NEW_TEXTURE; + update_fbo_texture(ctx, texObj, face, level); + + /* state update */ + texObj->_Complete = GL_FALSE; + ctx->NewState |= _NEW_TEXTURE; + } } - out: _mesa_unlock_texture(ctx, texObj); } @@ -2764,6 +2839,11 @@ _mesa_CopyTexSubImage1D( GLenum target, GLint level, GET_CURRENT_CONTEXT(ctx); ASSERT_OUTSIDE_BEGIN_END_AND_FLUSH(ctx); + if (MESA_VERBOSE & (VERBOSE_API|VERBOSE_TEXTURE)) + _mesa_debug(ctx, "glCopyTexSubImage1D %s %d %d %d %d %d\n", + _mesa_lookup_enum_by_nr(target), + level, xoffset, x, y, width); + if (ctx->NewState & NEW_COPY_TEX_STATE) _mesa_update_state(ctx); @@ -2785,23 +2865,25 @@ _mesa_CopyTexSubImage1D( GLenum target, GLint level, if (copytexsubimage_error_check2(ctx, 1, target, level, xoffset, 0, 0, postConvWidth, 1, - texImage)) - goto out; - + texImage)) { + /* error was recorded */ + } + else { + /* If we have a border, xoffset=-1 is legal. Bias by border width */ + xoffset += texImage->Border; - /* If we have a border, xoffset=-1 is legal. Bias by border width */ - xoffset += texImage->Border; + if (_mesa_clip_copytexsubimage(ctx, &xoffset, &yoffset, &x, &y, + &width, &height)) { + ASSERT(ctx->Driver.CopyTexSubImage1D); + ctx->Driver.CopyTexSubImage1D(ctx, target, level, + xoffset, x, y, width); - if (_mesa_clip_copytexsubimage(ctx, &xoffset, &yoffset, &x, &y, - &width, &height)) { - ASSERT(ctx->Driver.CopyTexSubImage1D); - ctx->Driver.CopyTexSubImage1D(ctx, target, level, - xoffset, x, y, width); - } + check_gen_mipmap(ctx, target, texObj, level); - ctx->NewState |= _NEW_TEXTURE; + ctx->NewState |= _NEW_TEXTURE; + } + } } - out: _mesa_unlock_texture(ctx, texObj); } @@ -2819,6 +2901,11 @@ _mesa_CopyTexSubImage2D( GLenum target, GLint level, GET_CURRENT_CONTEXT(ctx); ASSERT_OUTSIDE_BEGIN_END_AND_FLUSH(ctx); + if (MESA_VERBOSE & (VERBOSE_API|VERBOSE_TEXTURE)) + _mesa_debug(ctx, "glCopyTexSubImage2D %s %d %d %d %d %d %d %d\n", + _mesa_lookup_enum_by_nr(target), + level, xoffset, yoffset, x, y, width, height); + if (ctx->NewState & NEW_COPY_TEX_STATE) _mesa_update_state(ctx); @@ -2839,24 +2926,29 @@ _mesa_CopyTexSubImage2D( GLenum target, GLint level, } #endif - if (copytexsubimage_error_check2(ctx, 2, target, level, xoffset, yoffset, 0, - postConvWidth, postConvHeight, texImage)) - goto out; + if (copytexsubimage_error_check2(ctx, 2, target, level, + xoffset, yoffset, 0, + postConvWidth, postConvHeight, + texImage)) { + /* error was recorded */ + } + else { + /* If we have a border, xoffset=-1 is legal. Bias by border width */ + xoffset += texImage->Border; + yoffset += texImage->Border; - /* If we have a border, xoffset=-1 is legal. Bias by border width */ - xoffset += texImage->Border; - yoffset += texImage->Border; + if (_mesa_clip_copytexsubimage(ctx, &xoffset, &yoffset, &x, &y, + &width, &height)) { + ASSERT(ctx->Driver.CopyTexSubImage2D); + ctx->Driver.CopyTexSubImage2D(ctx, target, level, xoffset, yoffset, + x, y, width, height); - if (_mesa_clip_copytexsubimage(ctx, &xoffset, &yoffset, &x, &y, - &width, &height)) { - ASSERT(ctx->Driver.CopyTexSubImage2D); - ctx->Driver.CopyTexSubImage2D(ctx, target, level, - xoffset, yoffset, x, y, width, height); - } + check_gen_mipmap(ctx, target, texObj, level); - ctx->NewState |= _NEW_TEXTURE; + ctx->NewState |= _NEW_TEXTURE; + } + } } - out: _mesa_unlock_texture(ctx, texObj); } @@ -2874,6 +2966,11 @@ _mesa_CopyTexSubImage3D( GLenum target, GLint level, GET_CURRENT_CONTEXT(ctx); ASSERT_OUTSIDE_BEGIN_END_AND_FLUSH(ctx); + if (MESA_VERBOSE & (VERBOSE_API|VERBOSE_TEXTURE)) + _mesa_debug(ctx, "glCopyTexSubImage3D %s %d %d %d %d %d %d %d %d\n", + _mesa_lookup_enum_by_nr(target), + level, xoffset, yoffset, zoffset, x, y, width, height); + if (ctx->NewState & NEW_COPY_TEX_STATE) _mesa_update_state(ctx); @@ -2896,25 +2993,28 @@ _mesa_CopyTexSubImage3D( GLenum target, GLint level, if (copytexsubimage_error_check2(ctx, 3, target, level, xoffset, yoffset, zoffset, postConvWidth, postConvHeight, - texImage)) - goto out; - - /* If we have a border, xoffset=-1 is legal. Bias by border width */ - xoffset += texImage->Border; - yoffset += texImage->Border; - zoffset += texImage->Border; - - if (_mesa_clip_copytexsubimage(ctx, &xoffset, &yoffset, &x, &y, - &width, &height)) { - ASSERT(ctx->Driver.CopyTexSubImage3D); - ctx->Driver.CopyTexSubImage3D(ctx, target, level, - xoffset, yoffset, zoffset, - x, y, width, height); + texImage)) { + /* error was recored */ } + else { + /* If we have a border, xoffset=-1 is legal. Bias by border width */ + xoffset += texImage->Border; + yoffset += texImage->Border; + zoffset += texImage->Border; + + if (_mesa_clip_copytexsubimage(ctx, &xoffset, &yoffset, &x, &y, + &width, &height)) { + ASSERT(ctx->Driver.CopyTexSubImage3D); + ctx->Driver.CopyTexSubImage3D(ctx, target, level, + xoffset, yoffset, zoffset, + x, y, width, height); - ctx->NewState |= _NEW_TEXTURE; + check_gen_mipmap(ctx, target, texObj, level); + + ctx->NewState |= _NEW_TEXTURE; + } + } } - out: _mesa_unlock_texture(ctx, texObj); } @@ -3122,6 +3222,12 @@ _mesa_CompressedTexImage1DARB(GLenum target, GLint level, GET_CURRENT_CONTEXT(ctx); ASSERT_OUTSIDE_BEGIN_END_AND_FLUSH(ctx); + if (MESA_VERBOSE & (VERBOSE_API|VERBOSE_TEXTURE)) + _mesa_debug(ctx, "glCompressedTexImage1DARB %s %d %s %d %d %d %p\n", + _mesa_lookup_enum_by_nr(target), level, + _mesa_lookup_enum_by_nr(internalFormat), + width, border, imageSize, data); + if (target == GL_TEXTURE_1D) { /* non-proxy target */ struct gl_texture_unit *texUnit; @@ -3142,28 +3248,29 @@ _mesa_CompressedTexImage1DARB(GLenum target, GLint level, texImage = _mesa_get_tex_image(ctx, texObj, target, level); if (!texImage) { _mesa_error(ctx, GL_OUT_OF_MEMORY, "glCompressedTexImage1D"); - goto out; } - - if (texImage->Data) { - ctx->Driver.FreeTexImageData( ctx, texImage ); - } - ASSERT(texImage->Data == NULL); + else { + if (texImage->Data) { + ctx->Driver.FreeTexImageData( ctx, texImage ); + } + ASSERT(texImage->Data == NULL); - _mesa_init_teximage_fields(ctx, target, texImage, width, 1, 1, - border, internalFormat); + _mesa_init_teximage_fields(ctx, target, texImage, width, 1, 1, + border, internalFormat); - ASSERT(ctx->Driver.CompressedTexImage1D); - (*ctx->Driver.CompressedTexImage1D)(ctx, target, level, - internalFormat, width, border, - imageSize, data, - texObj, texImage); + ASSERT(ctx->Driver.CompressedTexImage1D); + ctx->Driver.CompressedTexImage1D(ctx, target, level, + internalFormat, width, border, + imageSize, data, + texObj, texImage); - /* state update */ - texObj->_Complete = GL_FALSE; - ctx->NewState |= _NEW_TEXTURE; + check_gen_mipmap(ctx, target, texObj, level); + + /* state update */ + texObj->_Complete = GL_FALSE; + ctx->NewState |= _NEW_TEXTURE; + } } - out: _mesa_unlock_texture(ctx, texObj); } else if (target == GL_PROXY_TEXTURE_1D) { @@ -3216,6 +3323,12 @@ _mesa_CompressedTexImage2DARB(GLenum target, GLint level, GET_CURRENT_CONTEXT(ctx); ASSERT_OUTSIDE_BEGIN_END_AND_FLUSH(ctx); + if (MESA_VERBOSE & (VERBOSE_API|VERBOSE_TEXTURE)) + _mesa_debug(ctx, "glCompressedTexImage2DARB %s %d %s %d %d %d %d %p\n", + _mesa_lookup_enum_by_nr(target), level, + _mesa_lookup_enum_by_nr(internalFormat), + width, height, border, imageSize, data); + if (target == GL_TEXTURE_2D || (ctx->Extensions.ARB_texture_cube_map && target >= GL_TEXTURE_CUBE_MAP_POSITIVE_X_ARB && @@ -3239,28 +3352,29 @@ _mesa_CompressedTexImage2DARB(GLenum target, GLint level, texImage = _mesa_get_tex_image(ctx, texObj, target, level); if (!texImage) { _mesa_error(ctx, GL_OUT_OF_MEMORY, "glCompressedTexImage2D"); - goto out; - } - - if (texImage->Data) { - ctx->Driver.FreeTexImageData( ctx, texImage ); } - ASSERT(texImage->Data == NULL); - - _mesa_init_teximage_fields(ctx, target, texImage, width, height, 1, - border, internalFormat); - - ASSERT(ctx->Driver.CompressedTexImage2D); - (*ctx->Driver.CompressedTexImage2D)(ctx, target, level, - internalFormat, width, height, - border, imageSize, data, - texObj, texImage); - - /* state update */ - texObj->_Complete = GL_FALSE; - ctx->NewState |= _NEW_TEXTURE; - } - out: + else { + if (texImage->Data) { + ctx->Driver.FreeTexImageData( ctx, texImage ); + } + ASSERT(texImage->Data == NULL); + + _mesa_init_teximage_fields(ctx, target, texImage, width, height, 1, + border, internalFormat); + + ASSERT(ctx->Driver.CompressedTexImage2D); + ctx->Driver.CompressedTexImage2D(ctx, target, level, + internalFormat, width, height, + border, imageSize, data, + texObj, texImage); + + check_gen_mipmap(ctx, target, texObj, level); + + /* state update */ + texObj->_Complete = GL_FALSE; + ctx->NewState |= _NEW_TEXTURE; + } + } _mesa_unlock_texture(ctx, texObj); } else if (target == GL_PROXY_TEXTURE_2D || @@ -3315,6 +3429,12 @@ _mesa_CompressedTexImage3DARB(GLenum target, GLint level, GET_CURRENT_CONTEXT(ctx); ASSERT_OUTSIDE_BEGIN_END_AND_FLUSH(ctx); + if (MESA_VERBOSE & (VERBOSE_API|VERBOSE_TEXTURE)) + _mesa_debug(ctx, "glCompressedTexImage3DARB %s %d %s %d %d %d %d %d %p\n", + _mesa_lookup_enum_by_nr(target), level, + _mesa_lookup_enum_by_nr(internalFormat), + width, height, depth, border, imageSize, data); + if (target == GL_TEXTURE_3D) { /* non-proxy target */ struct gl_texture_unit *texUnit; @@ -3334,29 +3454,31 @@ _mesa_CompressedTexImage3DARB(GLenum target, GLint level, texImage = _mesa_get_tex_image(ctx, texObj, target, level); if (!texImage) { _mesa_error(ctx, GL_OUT_OF_MEMORY, "glCompressedTexImage3D"); - goto out; - } - - if (texImage->Data) { - ctx->Driver.FreeTexImageData( ctx, texImage ); } - ASSERT(texImage->Data == NULL); - - _mesa_init_teximage_fields(ctx, target, texImage, width, height, depth, - border, internalFormat); - - ASSERT(ctx->Driver.CompressedTexImage3D); - (*ctx->Driver.CompressedTexImage3D)(ctx, target, level, - internalFormat, - width, height, depth, - border, imageSize, data, - texObj, texImage); - - /* state update */ - texObj->_Complete = GL_FALSE; - ctx->NewState |= _NEW_TEXTURE; - } - out: + else { + if (texImage->Data) { + ctx->Driver.FreeTexImageData( ctx, texImage ); + } + ASSERT(texImage->Data == NULL); + + _mesa_init_teximage_fields(ctx, target, texImage, + width, height, depth, + border, internalFormat); + + ASSERT(ctx->Driver.CompressedTexImage3D); + ctx->Driver.CompressedTexImage3D(ctx, target, level, + internalFormat, + width, height, depth, + border, imageSize, data, + texObj, texImage); + + check_gen_mipmap(ctx, target, texObj, level); + + /* state update */ + texObj->_Complete = GL_FALSE; + ctx->NewState |= _NEW_TEXTURE; + } + } _mesa_unlock_texture(ctx, texObj); } else if (target == GL_PROXY_TEXTURE_3D) { @@ -3422,6 +3544,7 @@ _mesa_CompressedTexSubImage1DARB(GLenum target, GLint level, GLint xoffset, texUnit = _mesa_get_current_tex_unit(ctx); texObj = _mesa_select_tex_object(ctx, texUnit, target); + _mesa_lock_texture(ctx, texObj); { texImage = _mesa_select_tex_image(ctx, texObj, target, level); @@ -3430,26 +3553,25 @@ _mesa_CompressedTexSubImage1DARB(GLenum target, GLint level, GLint xoffset, if ((GLint) format != texImage->InternalFormat) { _mesa_error(ctx, GL_INVALID_OPERATION, "glCompressedTexSubImage1D(format)"); - goto out; } + else if ((width == 1 || width == 2) && + (GLuint) width != texImage->Width) { + _mesa_error(ctx, GL_INVALID_VALUE, + "glCompressedTexSubImage1D(width)"); + } + else if (width > 0) { + if (ctx->Driver.CompressedTexSubImage1D) { + ctx->Driver.CompressedTexSubImage1D(ctx, target, level, + xoffset, width, + format, imageSize, data, + texObj, texImage); + } - if ((width == 1 || width == 2) && (GLuint) width != texImage->Width) { - _mesa_error(ctx, GL_INVALID_VALUE, "glCompressedTexSubImage1D(width)"); - goto out; - } - - if (width == 0) - goto out; /* no-op, not an error */ + check_gen_mipmap(ctx, target, texObj, level); - if (ctx->Driver.CompressedTexSubImage1D) { - (*ctx->Driver.CompressedTexSubImage1D)(ctx, target, level, - xoffset, width, - format, imageSize, data, - texObj, texImage); + ctx->NewState |= _NEW_TEXTURE; } - ctx->NewState |= _NEW_TEXTURE; } - out: _mesa_unlock_texture(ctx, texObj); } @@ -3479,6 +3601,7 @@ _mesa_CompressedTexSubImage2DARB(GLenum target, GLint level, GLint xoffset, texUnit = _mesa_get_current_tex_unit(ctx); texObj = _mesa_select_tex_object(ctx, texUnit, target); + _mesa_lock_texture(ctx, texObj); { texImage = _mesa_select_tex_image(ctx, texObj, target, level); @@ -3487,27 +3610,26 @@ _mesa_CompressedTexSubImage2DARB(GLenum target, GLint level, GLint xoffset, if ((GLint) format != texImage->InternalFormat) { _mesa_error(ctx, GL_INVALID_OPERATION, "glCompressedTexSubImage2D(format)"); - goto out; } - - if (((width == 1 || width == 2) && (GLuint) width != texImage->Width) || - ((height == 1 || height == 2) && (GLuint) height != texImage->Height)) { + else if (((width == 1 || width == 2) + && (GLuint) width != texImage->Width) || + ((height == 1 || height == 2) + && (GLuint) height != texImage->Height)) { _mesa_error(ctx, GL_INVALID_VALUE, "glCompressedTexSubImage2D(size)"); - goto out; } - - if (width == 0 || height == 0) - goto out; /* no-op, not an error */ - - if (ctx->Driver.CompressedTexSubImage2D) { - (*ctx->Driver.CompressedTexSubImage2D)(ctx, target, level, + else if (width > 0 && height > 0) { + if (ctx->Driver.CompressedTexSubImage2D) { + ctx->Driver.CompressedTexSubImage2D(ctx, target, level, xoffset, yoffset, width, height, format, imageSize, data, texObj, texImage); + } + + check_gen_mipmap(ctx, target, texObj, level); + + ctx->NewState |= _NEW_TEXTURE; } - ctx->NewState |= _NEW_TEXTURE; } - out: _mesa_unlock_texture(ctx, texObj); } @@ -3536,6 +3658,7 @@ _mesa_CompressedTexSubImage3DARB(GLenum target, GLint level, GLint xoffset, texUnit = _mesa_get_current_tex_unit(ctx); texObj = _mesa_select_tex_object(ctx, texUnit, target); + _mesa_lock_texture(ctx, texObj); { texImage = _mesa_select_tex_image(ctx, texObj, target, level); @@ -3544,29 +3667,29 @@ _mesa_CompressedTexSubImage3DARB(GLenum target, GLint level, GLint xoffset, if ((GLint) format != texImage->InternalFormat) { _mesa_error(ctx, GL_INVALID_OPERATION, "glCompressedTexSubImage3D(format)"); - goto out; } - - if (((width == 1 || width == 2) && (GLuint) width != texImage->Width) || - ((height == 1 || height == 2) && (GLuint) height != texImage->Height) || - ((depth == 1 || depth == 2) && (GLuint) depth != texImage->Depth)) { + else if (((width == 1 || width == 2) + && (GLuint) width != texImage->Width) || + ((height == 1 || height == 2) + && (GLuint) height != texImage->Height) || + ((depth == 1 || depth == 2) + && (GLuint) depth != texImage->Depth)) { _mesa_error(ctx, GL_INVALID_VALUE, "glCompressedTexSubImage3D(size)"); - goto out; } - - if (width == 0 || height == 0 || depth == 0) - goto out; /* no-op, not an error */ - - if (ctx->Driver.CompressedTexSubImage3D) { - (*ctx->Driver.CompressedTexSubImage3D)(ctx, target, level, + else if (width > 0 && height > 0 && depth > 0) { + if (ctx->Driver.CompressedTexSubImage3D) { + ctx->Driver.CompressedTexSubImage3D(ctx, target, level, xoffset, yoffset, zoffset, width, height, depth, format, imageSize, data, texObj, texImage); + } + + check_gen_mipmap(ctx, target, texObj, level); + + ctx->NewState |= _NEW_TEXTURE; } - ctx->NewState |= _NEW_TEXTURE; } - out: _mesa_unlock_texture(ctx, texObj); } diff --git a/src/mesa/main/texobj.c b/src/mesa/main/texobj.c index d09c4392506..678845ece06 100644 --- a/src/mesa/main/texobj.c +++ b/src/mesa/main/texobj.c @@ -29,9 +29,7 @@ #include "mfeatures.h" -#if FEATURE_colortable #include "colortab.h" -#endif #include "context.h" #include "enums.h" #include "fbobject.h" @@ -194,9 +192,7 @@ _mesa_delete_texture_object( GLcontext *ctx, struct gl_texture_object *texObj ) */ texObj->Target = 0x99; -#if FEATURE_colortable _mesa_free_colortable_data(&texObj->Palette); -#endif /* free the texture images */ for (face = 0; face < 6; face++) { diff --git a/src/mesa/main/texparam.c b/src/mesa/main/texparam.c index 05d144270ec..b2fbe2205ba 100644 --- a/src/mesa/main/texparam.c +++ b/src/mesa/main/texparam.c @@ -70,7 +70,7 @@ validate_texture_wrap_mode(GLcontext * ctx, GLenum target, GLenum wrap) return GL_TRUE; } - _mesa_error( ctx, GL_INVALID_VALUE, "glTexParameter(param=0x%x)", wrap ); + _mesa_error( ctx, GL_INVALID_ENUM, "glTexParameter(param=0x%x)", wrap ); return GL_FALSE; } @@ -210,7 +210,7 @@ set_tex_parameteri(GLcontext *ctx, } /* fall-through */ default: - _mesa_error( ctx, GL_INVALID_VALUE, "glTexParameter(param=0x%x)", + _mesa_error( ctx, GL_INVALID_ENUM, "glTexParameter(param=0x%x)", params[0] ); } return GL_FALSE; @@ -225,7 +225,7 @@ set_tex_parameteri(GLcontext *ctx, texObj->MagFilter = params[0]; return GL_TRUE; default: - _mesa_error( ctx, GL_INVALID_VALUE, "glTexParameter(param=0x%x)", + _mesa_error( ctx, GL_INVALID_ENUM, "glTexParameter(param=0x%x)", params[0]); } return GL_FALSE; diff --git a/src/mesa/main/texrender.c b/src/mesa/main/texrender.c index cc74d58fbd1..53be83b05ce 100644 --- a/src/mesa/main/texrender.c +++ b/src/mesa/main/texrender.c @@ -398,6 +398,14 @@ texture_put_mono_values(GLcontext *ctx, struct gl_renderbuffer *rb, static void +store_nop(struct gl_texture_image *texImage, + GLint col, GLint row, GLint img, + const void *texel) +{ +} + + +static void delete_texture_wrapper(struct gl_renderbuffer *rb) { ASSERT(rb->RefCount == 0); @@ -462,7 +470,10 @@ update_wrapper(GLcontext *ctx, const struct gl_renderbuffer_attachment *att) ASSERT(trb->TexImage); trb->Store = trb->TexImage->TexFormat->StoreTexel; - ASSERT(trb->Store); + if (!trb->Store) { + /* we'll never draw into some textures (compressed formats) */ + trb->Store = store_nop; + } if (att->Texture->Target == GL_TEXTURE_1D_ARRAY_EXT) { trb->Yoffset = att->Zoffset; diff --git a/src/mesa/main/texstate.c b/src/mesa/main/texstate.c index 861c5f37c4e..b9311d0ffc9 100644 --- a/src/mesa/main/texstate.c +++ b/src/mesa/main/texstate.c @@ -31,9 +31,7 @@ #include "glheader.h" #include "mfeatures.h" #include "colormac.h" -#if FEATURE_colortable #include "colortab.h" -#endif #include "context.h" #include "enums.h" #include "macros.h" @@ -753,9 +751,7 @@ _mesa_init_texture(GLcontext *ctx) ctx->Texture.CurrentUnit = 0; /* multitexture */ ctx->Texture._EnabledUnits = 0x0; ctx->Texture.SharedPalette = GL_FALSE; -#if FEATURE_colortable _mesa_init_colortable(&ctx->Texture.Palette); -#endif for (u = 0; u < MAX_TEXTURE_UNITS; u++) init_texture_unit(ctx, u); @@ -796,10 +792,8 @@ _mesa_free_texture_data(GLcontext *ctx) for (tgt = 0; tgt < NUM_TEXTURE_TARGETS; tgt++) ctx->Driver.DeleteTexture(ctx, ctx->Texture.ProxyTex[tgt]); -#if FEATURE_colortable for (u = 0; u < MAX_TEXTURE_IMAGE_UNITS; u++) _mesa_free_colortable_data(&ctx->Texture.Unit[u].ColorTable); -#endif } diff --git a/src/mesa/main/texstore.c b/src/mesa/main/texstore.c index a22db628d3e..f553a898f9e 100644 --- a/src/mesa/main/texstore.c +++ b/src/mesa/main/texstore.c @@ -56,9 +56,7 @@ #include "bufferobj.h" #include "colormac.h" #include "context.h" -#if FEATURE_convolve #include "convolve.h" -#endif #include "image.h" #include "macros.h" #include "mipmap.h" @@ -3302,32 +3300,9 @@ _mesa_set_fetch_functions(struct gl_texture_image *texImage, GLuint dims) } -/** - * Choose the actual storage format for a new texture image. - * Mainly, this is a wrapper for the driver's ChooseTextureFormat() function. - * Also set some other texImage fields related to texture compression, etc. - * \param ctx rendering context - * \param texImage the gl_texture_image - * \param dims texture dimensions (1, 2 or 3) - * \param format the user-specified format parameter - * \param type the user-specified type parameter - * \param internalFormat the user-specified internal format hint - */ static void -choose_texture_format(GLcontext *ctx, struct gl_texture_image *texImage, - GLuint dims, - GLenum format, GLenum type, GLint internalFormat) +compute_texture_size(GLcontext *ctx, struct gl_texture_image *texImage) { - ASSERT(dims == 1 || dims == 2 || dims == 3); - ASSERT(ctx->Driver.ChooseTextureFormat); - - texImage->TexFormat - = ctx->Driver.ChooseTextureFormat(ctx, internalFormat, format, type); - - ASSERT(texImage->TexFormat); - - _mesa_set_fetch_functions(texImage, dims); - if (texImage->TexFormat->TexelBytes == 0) { /* must be a compressed format */ texImage->IsCompressed = GL_TRUE; @@ -3365,7 +3340,12 @@ _mesa_store_teximage1d(GLcontext *ctx, GLenum target, GLint level, GLint sizeInBytes; (void) border; - choose_texture_format(ctx, texImage, 1, format, type, internalFormat); + texImage->TexFormat + = ctx->Driver.ChooseTextureFormat(ctx, internalFormat, format, type); + ASSERT(texImage->TexFormat); + + _mesa_set_fetch_functions(texImage, 1); + compute_texture_size(ctx, texImage); /* allocate memory */ if (texImage->IsCompressed) @@ -3403,11 +3383,6 @@ _mesa_store_teximage1d(GLcontext *ctx, GLenum target, GLint level, } } - /* GL_SGIS_generate_mipmap */ - if (level == texObj->BaseLevel && texObj->GenerateMipmap) { - ctx->Driver.GenerateMipmap(ctx, target, texObj); - } - _mesa_unmap_teximage_pbo(ctx, packing); } @@ -3434,7 +3409,12 @@ _mesa_store_teximage2d(GLcontext *ctx, GLenum target, GLint level, GLint texelBytes, sizeInBytes; (void) border; - choose_texture_format(ctx, texImage, 2, format, type, internalFormat); + texImage->TexFormat + = ctx->Driver.ChooseTextureFormat(ctx, internalFormat, format, type); + ASSERT(texImage->TexFormat); + + _mesa_set_fetch_functions(texImage, 2); + compute_texture_size(ctx, texImage); texelBytes = texImage->TexFormat->TexelBytes; @@ -3481,11 +3461,6 @@ _mesa_store_teximage2d(GLcontext *ctx, GLenum target, GLint level, } } - /* GL_SGIS_generate_mipmap */ - if (level == texObj->BaseLevel && texObj->GenerateMipmap) { - ctx->Driver.GenerateMipmap(ctx, target, texObj); - } - _mesa_unmap_teximage_pbo(ctx, packing); } @@ -3508,7 +3483,12 @@ _mesa_store_teximage3d(GLcontext *ctx, GLenum target, GLint level, GLint texelBytes, sizeInBytes; (void) border; - choose_texture_format(ctx, texImage, 3, format, type, internalFormat); + texImage->TexFormat + = ctx->Driver.ChooseTextureFormat(ctx, internalFormat, format, type); + ASSERT(texImage->TexFormat); + + _mesa_set_fetch_functions(texImage, 3); + compute_texture_size(ctx, texImage); texelBytes = texImage->TexFormat->TexelBytes; @@ -3555,11 +3535,6 @@ _mesa_store_teximage3d(GLcontext *ctx, GLenum target, GLint level, } } - /* GL_SGIS_generate_mipmap */ - if (level == texObj->BaseLevel && texObj->GenerateMipmap) { - ctx->Driver.GenerateMipmap(ctx, target, texObj); - } - _mesa_unmap_teximage_pbo(ctx, packing); } @@ -3601,11 +3576,6 @@ _mesa_store_texsubimage1d(GLcontext *ctx, GLenum target, GLint level, } } - /* GL_SGIS_generate_mipmap */ - if (level == texObj->BaseLevel && texObj->GenerateMipmap) { - ctx->Driver.GenerateMipmap(ctx, target, texObj); - } - _mesa_unmap_teximage_pbo(ctx, packing); } @@ -3654,11 +3624,6 @@ _mesa_store_texsubimage2d(GLcontext *ctx, GLenum target, GLint level, } } - /* GL_SGIS_generate_mipmap */ - if (level == texObj->BaseLevel && texObj->GenerateMipmap) { - ctx->Driver.GenerateMipmap(ctx, target, texObj); - } - _mesa_unmap_teximage_pbo(ctx, packing); } @@ -3707,11 +3672,6 @@ _mesa_store_texsubimage3d(GLcontext *ctx, GLenum target, GLint level, } } - /* GL_SGIS_generate_mipmap */ - if (level == texObj->BaseLevel && texObj->GenerateMipmap) { - ctx->Driver.GenerateMipmap(ctx, target, texObj); - } - _mesa_unmap_teximage_pbo(ctx, packing); } @@ -3762,7 +3722,12 @@ _mesa_store_compressed_teximage2d(GLcontext *ctx, GLenum target, GLint level, ASSERT(texImage->Depth == 1); ASSERT(texImage->Data == NULL); /* was freed in glCompressedTexImage2DARB */ - choose_texture_format(ctx, texImage, 2, 0, 0, internalFormat); + texImage->TexFormat + = ctx->Driver.ChooseTextureFormat(ctx, internalFormat, 0, 0); + ASSERT(texImage->TexFormat); + + _mesa_set_fetch_functions(texImage, 2); + compute_texture_size(ctx, texImage); /* allocate storage */ texImage->Data = _mesa_alloc_texmemory(imageSize); @@ -3781,11 +3746,6 @@ _mesa_store_compressed_teximage2d(GLcontext *ctx, GLenum target, GLint level, ASSERT(texImage->CompressedSize == (GLuint) imageSize); MEMCPY(texImage->Data, data, imageSize); - /* GL_SGIS_generate_mipmap */ - if (level == texObj->BaseLevel && texObj->GenerateMipmap) { - ctx->Driver.GenerateMipmap(ctx, target, texObj); - } - _mesa_unmap_teximage_pbo(ctx, &ctx->Unpack); } @@ -3891,11 +3851,6 @@ _mesa_store_compressed_texsubimage2d(GLcontext *ctx, GLenum target, src += srcRowStride; } - /* GL_SGIS_generate_mipmap */ - if (level == texObj->BaseLevel && texObj->GenerateMipmap) { - ctx->Driver.GenerateMipmap(ctx, target, texObj); - } - _mesa_unmap_teximage_pbo(ctx, &ctx->Unpack); } diff --git a/src/mesa/shader/nvprogram.c b/src/mesa/shader/nvprogram.c index d6469b17bea..471a7358a2f 100644 --- a/src/mesa/shader/nvprogram.c +++ b/src/mesa/shader/nvprogram.c @@ -509,7 +509,79 @@ _mesa_GetVertexAttribPointervNV(GLuint index, GLenum pname, GLvoid **pointer) *pointer = (GLvoid *) ctx->Array.ArrayObj->VertexAttrib[index].Ptr; } +void +_mesa_emit_nv_temp_initialization(GLcontext *ctx, + struct gl_program *program) +{ + struct prog_instruction *inst; + int i; + + if (!ctx->Shader.EmitNVTempInitialization) + return; + + /* We'll swizzle up a zero temporary so we can use it for the + * ARL. + */ + if (program->NumTemporaries == 0) + program->NumTemporaries = 1; + + _mesa_insert_instructions(program, 0, program->NumTemporaries + 1); + + for (i = 0; i < program->NumTemporaries; i++) { + struct prog_instruction *inst = &program->Instructions[i]; + + inst->Opcode = OPCODE_SWZ; + inst->DstReg.File = PROGRAM_TEMPORARY; + inst->DstReg.Index = i; + inst->DstReg.WriteMask = WRITEMASK_XYZW; + inst->SrcReg[0].File = PROGRAM_TEMPORARY; + inst->SrcReg[0].Index = 0; + inst->SrcReg[0].Swizzle = MAKE_SWIZZLE4(SWIZZLE_ZERO, + SWIZZLE_ZERO, + SWIZZLE_ZERO, + SWIZZLE_ZERO); + } + + inst = &program->Instructions[i]; + inst->Opcode = OPCODE_ARL; + inst->DstReg.File = PROGRAM_ADDRESS; + inst->DstReg.Index = 0; + inst->DstReg.WriteMask = WRITEMASK_XYZW; + inst->SrcReg[0].File = PROGRAM_TEMPORARY; + inst->SrcReg[0].Index = 0; + inst->SrcReg[0].Swizzle = SWIZZLE_XXXX; + + if (program->NumAddressRegs == 0) + program->NumAddressRegs = 1; +} + +void +_mesa_setup_nv_temporary_count(GLcontext *ctx, struct gl_program *program) +{ + int i; + program->NumTemporaries = 0; + for (i = 0; i < program->NumInstructions; i++) { + struct prog_instruction *inst = &program->Instructions[i]; + + if (inst->DstReg.File == PROGRAM_TEMPORARY) { + program->NumTemporaries = MAX2(program->NumTemporaries, + inst->DstReg.Index + 1); + } + if (inst->SrcReg[0].File == PROGRAM_TEMPORARY) { + program->NumTemporaries = MAX2(program->NumTemporaries, + inst->SrcReg[0].Index + 1); + } + if (inst->SrcReg[1].File == PROGRAM_TEMPORARY) { + program->NumTemporaries = MAX2(program->NumTemporaries, + inst->SrcReg[1].Index + 1); + } + if (inst->SrcReg[2].File == PROGRAM_TEMPORARY) { + program->NumTemporaries = MAX2(program->NumTemporaries, + inst->SrcReg[2].Index + 1); + } + } +} /** * Load/parse/compile a program. diff --git a/src/mesa/shader/nvprogram.h b/src/mesa/shader/nvprogram.h index bfac165b5ed..8ee59661bd0 100644 --- a/src/mesa/shader/nvprogram.h +++ b/src/mesa/shader/nvprogram.h @@ -103,5 +103,11 @@ extern void GLAPIENTRY _mesa_GetProgramNamedParameterdvNV(GLuint id, GLsizei len, const GLubyte *name, GLdouble *params); +extern void +_mesa_setup_nv_temporary_count(GLcontext *ctx, struct gl_program *program); + +extern void +_mesa_emit_nv_temp_initialization(GLcontext *ctx, + struct gl_program *program); #endif diff --git a/src/mesa/shader/nvvertparse.c b/src/mesa/shader/nvvertparse.c index f5e2df26708..8574016050c 100644 --- a/src/mesa/shader/nvvertparse.c +++ b/src/mesa/shader/nvvertparse.c @@ -44,6 +44,7 @@ #include "nvprogram.h" #include "nvvertparse.h" #include "prog_instruction.h" +#include "prog_parameter.h" #include "prog_print.h" #include "program.h" @@ -1345,6 +1346,9 @@ _mesa_parse_nv_vertex_program(GLcontext *ctx, GLenum dstTarget, if (Parse_Program(&parseState, instBuffer)) { + gl_state_index state_tokens[STATE_LENGTH] = {0, 0, 0, 0, 0}; + int i; + /* successful parse! */ if (parseState.isStateProgram) { @@ -1398,6 +1402,29 @@ _mesa_parse_nv_vertex_program(GLcontext *ctx, GLenum dstTarget, _mesa_fprint_program_opt(stdout, &program->Base, PROG_PRINT_NV, 0); _mesa_printf("------------------------------\n"); #endif + + if (program->Base.Parameters) + _mesa_free_parameter_list(program->Base.Parameters); + + program->Base.Parameters = _mesa_new_parameter_list (); + program->Base.NumParameters = 0; + + state_tokens[0] = STATE_VERTEX_PROGRAM; + state_tokens[1] = STATE_ENV; + /* Add refs to all of the potential params, in order. If we want to not + * upload everything, _mesa_layout_parameters is the answer. + */ + for (i = 0; i < MAX_NV_VERTEX_PROGRAM_PARAMS; i++) { + GLint index; + state_tokens[2] = i; + index = _mesa_add_state_reference(program->Base.Parameters, + state_tokens); + assert(index == i); + } + program->Base.NumParameters = program->Base.Parameters->NumParameters; + + _mesa_setup_nv_temporary_count(ctx, &program->Base); + _mesa_emit_nv_temp_initialization(ctx, &program->Base); } else { /* Error! */ diff --git a/src/mesa/shader/prog_execute.c b/src/mesa/shader/prog_execute.c index c8a762f8ff7..69b81e724a2 100644 --- a/src/mesa/shader/prog_execute.c +++ b/src/mesa/shader/prog_execute.c @@ -1555,17 +1555,12 @@ _mesa_execute_program(GLcontext * ctx, case OPCODE_TXB: /* GL_ARB_fragment_program only */ /* Texel lookup with LOD bias */ { - const GLuint unit = machine->Samplers[inst->TexSrcUnit]; - const struct gl_texture_unit *texUnit = &ctx->Texture.Unit[unit]; GLfloat texcoord[4], color[4], lodBias; fetch_vector4(&inst->SrcReg[0], machine, texcoord); /* texcoord[3] is the bias to add to lambda */ - lodBias = texUnit->LodBias + texcoord[3]; - if (texUnit->_Current) { - lodBias += texUnit->_Current->LodBias; - } + lodBias = texcoord[3]; fetch_texel(ctx, machine, inst, texcoord, lodBias, color); diff --git a/src/mesa/shader/prog_instruction.h b/src/mesa/shader/prog_instruction.h index 39a221eeaba..1c687bc16cf 100644 --- a/src/mesa/shader/prog_instruction.h +++ b/src/mesa/shader/prog_instruction.h @@ -121,7 +121,6 @@ /*@{*/ #define SATURATE_OFF 0 #define SATURATE_ZERO_ONE 1 -#define SATURATE_PLUS_MINUS_ONE 2 /*@}*/ diff --git a/src/mesa/shader/prog_parameter.c b/src/mesa/shader/prog_parameter.c index 6b9e73b2cbd..2f029b02e50 100644 --- a/src/mesa/shader/prog_parameter.c +++ b/src/mesa/shader/prog_parameter.c @@ -100,6 +100,7 @@ _mesa_free_parameter_list(struct gl_program_parameter_list *paramList) * \param type type of parameter, such as * \param name the parameter name, will be duplicated/copied! * \param size number of elements in 'values' vector (1..4, or more) + * \param datatype GL_FLOAT, GL_FLOAT_VECx, GL_INT, GL_INT_VECx or GL_NONE. * \param values initial parameter value, up to 4 GLfloats, or NULL * \param state state indexes, or NULL * \return index of new parameter in the list, or -1 if error (out of mem) diff --git a/src/mesa/shader/program.c b/src/mesa/shader/program.c index 963478fccdb..532adf4d360 100644 --- a/src/mesa/shader/program.c +++ b/src/mesa/shader/program.c @@ -351,13 +351,6 @@ _mesa_delete_program(GLcontext *ctx, struct gl_program *prog) _mesa_free_parameter_list(prog->Attributes); } - /* XXX this is a little ugly */ - if (prog->Target == GL_VERTEX_PROGRAM_ARB) { - struct gl_vertex_program *vprog = (struct gl_vertex_program *) prog; - if (vprog->TnlData) - _mesa_free(vprog->TnlData); - } - _mesa_free(prog); } @@ -502,6 +495,7 @@ _mesa_clone_program(GLcontext *ctx, const struct gl_program *prog) = (const struct gl_vertex_program *) prog; struct gl_vertex_program *vpc = (struct gl_vertex_program *) clone; vpc->IsPositionInvariant = vp->IsPositionInvariant; + vpc->IsNVProgram = vp->IsNVProgram; } break; case GL_FRAGMENT_PROGRAM_ARB: diff --git a/src/mesa/shader/program_parse.tab.c b/src/mesa/shader/program_parse.tab.c index c51b4d2f282..c0f50147bd9 100644 --- a/src/mesa/shader/program_parse.tab.c +++ b/src/mesa/shader/program_parse.tab.c @@ -789,28 +789,28 @@ static const yytype_uint16 yyrline[] = 517, 518, 519, 520, 521, 522, 523, 526, 538, 546, 563, 570, 589, 600, 620, 645, 654, 687, 694, 709, 759, 801, 812, 833, 843, 849, 880, 897, 897, 899, - 906, 918, 919, 920, 923, 935, 947, 965, 976, 988, - 990, 991, 992, 993, 996, 996, 996, 996, 997, 1000, - 1004, 1009, 1016, 1023, 1030, 1053, 1076, 1077, 1078, 1079, - 1080, 1081, 1084, 1102, 1106, 1112, 1116, 1120, 1124, 1133, - 1142, 1146, 1151, 1157, 1168, 1168, 1169, 1171, 1175, 1179, - 1183, 1189, 1189, 1191, 1207, 1230, 1233, 1244, 1250, 1256, - 1257, 1264, 1270, 1276, 1284, 1290, 1296, 1304, 1310, 1316, - 1324, 1325, 1328, 1329, 1330, 1331, 1332, 1333, 1334, 1335, - 1336, 1337, 1338, 1341, 1350, 1354, 1358, 1364, 1373, 1377, - 1381, 1390, 1394, 1400, 1406, 1413, 1418, 1426, 1436, 1438, - 1446, 1452, 1456, 1460, 1466, 1477, 1486, 1490, 1495, 1499, - 1503, 1507, 1513, 1520, 1524, 1530, 1538, 1549, 1556, 1560, - 1566, 1576, 1587, 1591, 1609, 1618, 1621, 1627, 1631, 1635, - 1641, 1652, 1657, 1662, 1667, 1672, 1677, 1685, 1688, 1693, - 1706, 1714, 1725, 1733, 1733, 1735, 1735, 1737, 1747, 1752, - 1759, 1769, 1778, 1783, 1790, 1800, 1810, 1822, 1822, 1823, - 1823, 1825, 1835, 1843, 1853, 1861, 1869, 1878, 1889, 1893, - 1899, 1900, 1901, 1904, 1904, 1907, 1942, 1946, 1946, 1949, - 1955, 1963, 1976, 1985, 1994, 1998, 2007, 2016, 2027, 2034, - 2039, 2048, 2060, 2063, 2072, 2083, 2084, 2085, 2088, 2089, - 2090, 2093, 2094, 2097, 2098, 2101, 2102, 2105, 2116, 2127, - 2138, 2159, 2160 + 906, 918, 919, 920, 923, 937, 951, 969, 980, 992, + 994, 995, 996, 997, 1000, 1000, 1000, 1000, 1001, 1004, + 1008, 1013, 1020, 1027, 1034, 1057, 1080, 1081, 1082, 1083, + 1084, 1085, 1088, 1106, 1110, 1116, 1120, 1124, 1128, 1137, + 1146, 1150, 1155, 1161, 1172, 1172, 1173, 1175, 1179, 1183, + 1187, 1193, 1193, 1195, 1211, 1234, 1237, 1248, 1254, 1260, + 1261, 1268, 1274, 1280, 1288, 1294, 1300, 1308, 1314, 1320, + 1328, 1329, 1332, 1333, 1334, 1335, 1336, 1337, 1338, 1339, + 1340, 1341, 1342, 1345, 1354, 1358, 1362, 1368, 1377, 1381, + 1385, 1394, 1398, 1404, 1410, 1417, 1422, 1430, 1440, 1442, + 1450, 1456, 1460, 1464, 1470, 1481, 1490, 1494, 1499, 1503, + 1507, 1511, 1517, 1524, 1528, 1534, 1542, 1553, 1560, 1564, + 1570, 1580, 1591, 1595, 1613, 1622, 1625, 1631, 1635, 1639, + 1645, 1656, 1661, 1666, 1671, 1676, 1681, 1689, 1692, 1697, + 1710, 1718, 1729, 1737, 1737, 1739, 1739, 1741, 1751, 1756, + 1763, 1773, 1782, 1787, 1794, 1804, 1814, 1826, 1826, 1827, + 1827, 1829, 1839, 1847, 1857, 1865, 1873, 1882, 1893, 1897, + 1903, 1904, 1905, 1908, 1908, 1911, 1946, 1950, 1950, 1953, + 1959, 1967, 1980, 1989, 1998, 2002, 2011, 2020, 2031, 2038, + 2043, 2052, 2064, 2067, 2076, 2087, 2088, 2089, 2092, 2093, + 2094, 2097, 2098, 2101, 2102, 2105, 2106, 2109, 2120, 2131, + 2142, 2163, 2164 }; #endif @@ -2971,8 +2971,10 @@ yyreduce: #line 924 "program_parse.y" { if (((yyvsp[(1) - (1)].integer) < 0) || ((yyvsp[(1) - (1)].integer) > 63)) { - yyerror(& (yylsp[(1) - (1)]), state, - "relative address offset too large (positive)"); + char s[100]; + _mesa_snprintf(s, sizeof(s), + "relative address offset too large (%d)", (yyvsp[(1) - (1)].integer)); + yyerror(& (yylsp[(1) - (1)]), state, s); YYERROR; } else { (yyval.integer) = (yyvsp[(1) - (1)].integer); @@ -2983,11 +2985,13 @@ yyreduce: case 75: /* Line 1455 of yacc.c */ -#line 936 "program_parse.y" +#line 938 "program_parse.y" { if (((yyvsp[(1) - (1)].integer) < 0) || ((yyvsp[(1) - (1)].integer) > 64)) { - yyerror(& (yylsp[(1) - (1)]), state, - "relative address offset too large (negative)"); + char s[100]; + _mesa_snprintf(s, sizeof(s), + "relative address offset too large (%d)", (yyvsp[(1) - (1)].integer)); + yyerror(& (yylsp[(1) - (1)]), state, s); YYERROR; } else { (yyval.integer) = (yyvsp[(1) - (1)].integer); @@ -2998,7 +3002,7 @@ yyreduce: case 76: /* Line 1455 of yacc.c */ -#line 948 "program_parse.y" +#line 952 "program_parse.y" { struct asm_symbol *const s = (struct asm_symbol *) _mesa_symbol_table_find_symbol(state->st, 0, (yyvsp[(1) - (1)].string)); @@ -3019,7 +3023,7 @@ yyreduce: case 77: /* Line 1455 of yacc.c */ -#line 966 "program_parse.y" +#line 970 "program_parse.y" { if ((yyvsp[(1) - (1)].swiz_mask).mask != WRITEMASK_X) { yyerror(& (yylsp[(1) - (1)]), state, "invalid address component selector"); @@ -3033,7 +3037,7 @@ yyreduce: case 78: /* Line 1455 of yacc.c */ -#line 977 "program_parse.y" +#line 981 "program_parse.y" { if ((yyvsp[(1) - (1)].swiz_mask).mask != WRITEMASK_X) { yyerror(& (yylsp[(1) - (1)]), state, @@ -3048,21 +3052,21 @@ yyreduce: case 83: /* Line 1455 of yacc.c */ -#line 993 "program_parse.y" +#line 997 "program_parse.y" { (yyval.swiz_mask).swizzle = SWIZZLE_NOOP; (yyval.swiz_mask).mask = WRITEMASK_XYZW; ;} break; case 88: /* Line 1455 of yacc.c */ -#line 997 "program_parse.y" +#line 1001 "program_parse.y" { (yyval.swiz_mask).swizzle = SWIZZLE_NOOP; (yyval.swiz_mask).mask = WRITEMASK_XYZW; ;} break; case 89: /* Line 1455 of yacc.c */ -#line 1001 "program_parse.y" +#line 1005 "program_parse.y" { (yyval.dst_reg) = (yyvsp[(2) - (3)].dst_reg); ;} @@ -3071,7 +3075,7 @@ yyreduce: case 90: /* Line 1455 of yacc.c */ -#line 1005 "program_parse.y" +#line 1009 "program_parse.y" { (yyval.dst_reg) = (yyvsp[(2) - (3)].dst_reg); ;} @@ -3080,7 +3084,7 @@ yyreduce: case 91: /* Line 1455 of yacc.c */ -#line 1009 "program_parse.y" +#line 1013 "program_parse.y" { (yyval.dst_reg).CondMask = COND_TR; (yyval.dst_reg).CondSwizzle = SWIZZLE_NOOP; @@ -3091,7 +3095,7 @@ yyreduce: case 92: /* Line 1455 of yacc.c */ -#line 1017 "program_parse.y" +#line 1021 "program_parse.y" { (yyval.dst_reg) = (yyvsp[(1) - (2)].dst_reg); (yyval.dst_reg).CondSwizzle = (yyvsp[(2) - (2)].swiz_mask).swizzle; @@ -3101,7 +3105,7 @@ yyreduce: case 93: /* Line 1455 of yacc.c */ -#line 1024 "program_parse.y" +#line 1028 "program_parse.y" { (yyval.dst_reg) = (yyvsp[(1) - (2)].dst_reg); (yyval.dst_reg).CondSwizzle = (yyvsp[(2) - (2)].swiz_mask).swizzle; @@ -3111,7 +3115,7 @@ yyreduce: case 94: /* Line 1455 of yacc.c */ -#line 1031 "program_parse.y" +#line 1035 "program_parse.y" { const int cond = _mesa_parse_cc((yyvsp[(1) - (1)].string)); if ((cond == 0) || ((yyvsp[(1) - (1)].string)[2] != '\0')) { @@ -3137,7 +3141,7 @@ yyreduce: case 95: /* Line 1455 of yacc.c */ -#line 1054 "program_parse.y" +#line 1058 "program_parse.y" { const int cond = _mesa_parse_cc((yyvsp[(1) - (1)].string)); if ((cond == 0) || ((yyvsp[(1) - (1)].string)[2] != '\0')) { @@ -3163,7 +3167,7 @@ yyreduce: case 102: /* Line 1455 of yacc.c */ -#line 1085 "program_parse.y" +#line 1089 "program_parse.y" { struct asm_symbol *const s = declare_variable(state, (yyvsp[(2) - (4)].string), at_attrib, & (yylsp[(2) - (4)])); @@ -3184,7 +3188,7 @@ yyreduce: case 103: /* Line 1455 of yacc.c */ -#line 1103 "program_parse.y" +#line 1107 "program_parse.y" { (yyval.attrib) = (yyvsp[(2) - (2)].attrib); ;} @@ -3193,7 +3197,7 @@ yyreduce: case 104: /* Line 1455 of yacc.c */ -#line 1107 "program_parse.y" +#line 1111 "program_parse.y" { (yyval.attrib) = (yyvsp[(2) - (2)].attrib); ;} @@ -3202,7 +3206,7 @@ yyreduce: case 105: /* Line 1455 of yacc.c */ -#line 1113 "program_parse.y" +#line 1117 "program_parse.y" { (yyval.attrib) = VERT_ATTRIB_POS; ;} @@ -3211,7 +3215,7 @@ yyreduce: case 106: /* Line 1455 of yacc.c */ -#line 1117 "program_parse.y" +#line 1121 "program_parse.y" { (yyval.attrib) = VERT_ATTRIB_WEIGHT; ;} @@ -3220,7 +3224,7 @@ yyreduce: case 107: /* Line 1455 of yacc.c */ -#line 1121 "program_parse.y" +#line 1125 "program_parse.y" { (yyval.attrib) = VERT_ATTRIB_NORMAL; ;} @@ -3229,7 +3233,7 @@ yyreduce: case 108: /* Line 1455 of yacc.c */ -#line 1125 "program_parse.y" +#line 1129 "program_parse.y" { if (!state->ctx->Extensions.EXT_secondary_color) { yyerror(& (yylsp[(2) - (2)]), state, "GL_EXT_secondary_color not supported"); @@ -3243,7 +3247,7 @@ yyreduce: case 109: /* Line 1455 of yacc.c */ -#line 1134 "program_parse.y" +#line 1138 "program_parse.y" { if (!state->ctx->Extensions.EXT_fog_coord) { yyerror(& (yylsp[(1) - (1)]), state, "GL_EXT_fog_coord not supported"); @@ -3257,7 +3261,7 @@ yyreduce: case 110: /* Line 1455 of yacc.c */ -#line 1143 "program_parse.y" +#line 1147 "program_parse.y" { (yyval.attrib) = VERT_ATTRIB_TEX0 + (yyvsp[(2) - (2)].integer); ;} @@ -3266,7 +3270,7 @@ yyreduce: case 111: /* Line 1455 of yacc.c */ -#line 1147 "program_parse.y" +#line 1151 "program_parse.y" { yyerror(& (yylsp[(1) - (4)]), state, "GL_ARB_matrix_palette not supported"); YYERROR; @@ -3276,7 +3280,7 @@ yyreduce: case 112: /* Line 1455 of yacc.c */ -#line 1152 "program_parse.y" +#line 1156 "program_parse.y" { (yyval.attrib) = VERT_ATTRIB_GENERIC0 + (yyvsp[(3) - (4)].integer); ;} @@ -3285,7 +3289,7 @@ yyreduce: case 113: /* Line 1455 of yacc.c */ -#line 1158 "program_parse.y" +#line 1162 "program_parse.y" { if ((unsigned) (yyvsp[(1) - (1)].integer) >= state->limits->MaxAttribs) { yyerror(& (yylsp[(1) - (1)]), state, "invalid vertex attribute reference"); @@ -3299,7 +3303,7 @@ yyreduce: case 117: /* Line 1455 of yacc.c */ -#line 1172 "program_parse.y" +#line 1176 "program_parse.y" { (yyval.attrib) = FRAG_ATTRIB_WPOS; ;} @@ -3308,7 +3312,7 @@ yyreduce: case 118: /* Line 1455 of yacc.c */ -#line 1176 "program_parse.y" +#line 1180 "program_parse.y" { (yyval.attrib) = FRAG_ATTRIB_COL0 + (yyvsp[(2) - (2)].integer); ;} @@ -3317,7 +3321,7 @@ yyreduce: case 119: /* Line 1455 of yacc.c */ -#line 1180 "program_parse.y" +#line 1184 "program_parse.y" { (yyval.attrib) = FRAG_ATTRIB_FOGC; ;} @@ -3326,7 +3330,7 @@ yyreduce: case 120: /* Line 1455 of yacc.c */ -#line 1184 "program_parse.y" +#line 1188 "program_parse.y" { (yyval.attrib) = FRAG_ATTRIB_TEX0 + (yyvsp[(2) - (2)].integer); ;} @@ -3335,7 +3339,7 @@ yyreduce: case 123: /* Line 1455 of yacc.c */ -#line 1192 "program_parse.y" +#line 1196 "program_parse.y" { struct asm_symbol *const s = declare_variable(state, (yyvsp[(2) - (3)].string), at_param, & (yylsp[(2) - (3)])); @@ -3354,7 +3358,7 @@ yyreduce: case 124: /* Line 1455 of yacc.c */ -#line 1208 "program_parse.y" +#line 1212 "program_parse.y" { if (((yyvsp[(4) - (6)].integer) != 0) && ((unsigned) (yyvsp[(4) - (6)].integer) != (yyvsp[(6) - (6)].temp_sym).param_binding_length)) { yyerror(& (yylsp[(4) - (6)]), state, @@ -3379,7 +3383,7 @@ yyreduce: case 125: /* Line 1455 of yacc.c */ -#line 1230 "program_parse.y" +#line 1234 "program_parse.y" { (yyval.integer) = 0; ;} @@ -3388,7 +3392,7 @@ yyreduce: case 126: /* Line 1455 of yacc.c */ -#line 1234 "program_parse.y" +#line 1238 "program_parse.y" { if (((yyvsp[(1) - (1)].integer) < 1) || ((unsigned) (yyvsp[(1) - (1)].integer) >= state->limits->MaxParameters)) { yyerror(& (yylsp[(1) - (1)]), state, "invalid parameter array size"); @@ -3402,7 +3406,7 @@ yyreduce: case 127: /* Line 1455 of yacc.c */ -#line 1245 "program_parse.y" +#line 1249 "program_parse.y" { (yyval.temp_sym) = (yyvsp[(2) - (2)].temp_sym); ;} @@ -3411,7 +3415,7 @@ yyreduce: case 128: /* Line 1455 of yacc.c */ -#line 1251 "program_parse.y" +#line 1255 "program_parse.y" { (yyval.temp_sym) = (yyvsp[(3) - (4)].temp_sym); ;} @@ -3420,7 +3424,7 @@ yyreduce: case 130: /* Line 1455 of yacc.c */ -#line 1258 "program_parse.y" +#line 1262 "program_parse.y" { (yyvsp[(1) - (3)].temp_sym).param_binding_length += (yyvsp[(3) - (3)].temp_sym).param_binding_length; (yyval.temp_sym) = (yyvsp[(1) - (3)].temp_sym); @@ -3430,7 +3434,7 @@ yyreduce: case 131: /* Line 1455 of yacc.c */ -#line 1265 "program_parse.y" +#line 1269 "program_parse.y" { memset(& (yyval.temp_sym), 0, sizeof((yyval.temp_sym))); (yyval.temp_sym).param_binding_begin = ~0; @@ -3441,7 +3445,7 @@ yyreduce: case 132: /* Line 1455 of yacc.c */ -#line 1271 "program_parse.y" +#line 1275 "program_parse.y" { memset(& (yyval.temp_sym), 0, sizeof((yyval.temp_sym))); (yyval.temp_sym).param_binding_begin = ~0; @@ -3452,7 +3456,7 @@ yyreduce: case 133: /* Line 1455 of yacc.c */ -#line 1277 "program_parse.y" +#line 1281 "program_parse.y" { memset(& (yyval.temp_sym), 0, sizeof((yyval.temp_sym))); (yyval.temp_sym).param_binding_begin = ~0; @@ -3463,7 +3467,7 @@ yyreduce: case 134: /* Line 1455 of yacc.c */ -#line 1285 "program_parse.y" +#line 1289 "program_parse.y" { memset(& (yyval.temp_sym), 0, sizeof((yyval.temp_sym))); (yyval.temp_sym).param_binding_begin = ~0; @@ -3474,7 +3478,7 @@ yyreduce: case 135: /* Line 1455 of yacc.c */ -#line 1291 "program_parse.y" +#line 1295 "program_parse.y" { memset(& (yyval.temp_sym), 0, sizeof((yyval.temp_sym))); (yyval.temp_sym).param_binding_begin = ~0; @@ -3485,7 +3489,7 @@ yyreduce: case 136: /* Line 1455 of yacc.c */ -#line 1297 "program_parse.y" +#line 1301 "program_parse.y" { memset(& (yyval.temp_sym), 0, sizeof((yyval.temp_sym))); (yyval.temp_sym).param_binding_begin = ~0; @@ -3496,7 +3500,7 @@ yyreduce: case 137: /* Line 1455 of yacc.c */ -#line 1305 "program_parse.y" +#line 1309 "program_parse.y" { memset(& (yyval.temp_sym), 0, sizeof((yyval.temp_sym))); (yyval.temp_sym).param_binding_begin = ~0; @@ -3507,7 +3511,7 @@ yyreduce: case 138: /* Line 1455 of yacc.c */ -#line 1311 "program_parse.y" +#line 1315 "program_parse.y" { memset(& (yyval.temp_sym), 0, sizeof((yyval.temp_sym))); (yyval.temp_sym).param_binding_begin = ~0; @@ -3518,7 +3522,7 @@ yyreduce: case 139: /* Line 1455 of yacc.c */ -#line 1317 "program_parse.y" +#line 1321 "program_parse.y" { memset(& (yyval.temp_sym), 0, sizeof((yyval.temp_sym))); (yyval.temp_sym).param_binding_begin = ~0; @@ -3529,98 +3533,98 @@ yyreduce: case 140: /* Line 1455 of yacc.c */ -#line 1324 "program_parse.y" +#line 1328 "program_parse.y" { memcpy((yyval.state), (yyvsp[(1) - (1)].state), sizeof((yyval.state))); ;} break; case 141: /* Line 1455 of yacc.c */ -#line 1325 "program_parse.y" +#line 1329 "program_parse.y" { memcpy((yyval.state), (yyvsp[(2) - (2)].state), sizeof((yyval.state))); ;} break; case 142: /* Line 1455 of yacc.c */ -#line 1328 "program_parse.y" +#line 1332 "program_parse.y" { memcpy((yyval.state), (yyvsp[(2) - (2)].state), sizeof((yyval.state))); ;} break; case 143: /* Line 1455 of yacc.c */ -#line 1329 "program_parse.y" +#line 1333 "program_parse.y" { memcpy((yyval.state), (yyvsp[(2) - (2)].state), sizeof((yyval.state))); ;} break; case 144: /* Line 1455 of yacc.c */ -#line 1330 "program_parse.y" +#line 1334 "program_parse.y" { memcpy((yyval.state), (yyvsp[(2) - (2)].state), sizeof((yyval.state))); ;} break; case 145: /* Line 1455 of yacc.c */ -#line 1331 "program_parse.y" +#line 1335 "program_parse.y" { memcpy((yyval.state), (yyvsp[(2) - (2)].state), sizeof((yyval.state))); ;} break; case 146: /* Line 1455 of yacc.c */ -#line 1332 "program_parse.y" +#line 1336 "program_parse.y" { memcpy((yyval.state), (yyvsp[(2) - (2)].state), sizeof((yyval.state))); ;} break; case 147: /* Line 1455 of yacc.c */ -#line 1333 "program_parse.y" +#line 1337 "program_parse.y" { memcpy((yyval.state), (yyvsp[(2) - (2)].state), sizeof((yyval.state))); ;} break; case 148: /* Line 1455 of yacc.c */ -#line 1334 "program_parse.y" +#line 1338 "program_parse.y" { memcpy((yyval.state), (yyvsp[(2) - (2)].state), sizeof((yyval.state))); ;} break; case 149: /* Line 1455 of yacc.c */ -#line 1335 "program_parse.y" +#line 1339 "program_parse.y" { memcpy((yyval.state), (yyvsp[(2) - (2)].state), sizeof((yyval.state))); ;} break; case 150: /* Line 1455 of yacc.c */ -#line 1336 "program_parse.y" +#line 1340 "program_parse.y" { memcpy((yyval.state), (yyvsp[(2) - (2)].state), sizeof((yyval.state))); ;} break; case 151: /* Line 1455 of yacc.c */ -#line 1337 "program_parse.y" +#line 1341 "program_parse.y" { memcpy((yyval.state), (yyvsp[(2) - (2)].state), sizeof((yyval.state))); ;} break; case 152: /* Line 1455 of yacc.c */ -#line 1338 "program_parse.y" +#line 1342 "program_parse.y" { memcpy((yyval.state), (yyvsp[(2) - (2)].state), sizeof((yyval.state))); ;} break; case 153: /* Line 1455 of yacc.c */ -#line 1342 "program_parse.y" +#line 1346 "program_parse.y" { memset((yyval.state), 0, sizeof((yyval.state))); (yyval.state)[0] = STATE_MATERIAL; @@ -3632,7 +3636,7 @@ yyreduce: case 154: /* Line 1455 of yacc.c */ -#line 1351 "program_parse.y" +#line 1355 "program_parse.y" { (yyval.integer) = (yyvsp[(1) - (1)].integer); ;} @@ -3641,7 +3645,7 @@ yyreduce: case 155: /* Line 1455 of yacc.c */ -#line 1355 "program_parse.y" +#line 1359 "program_parse.y" { (yyval.integer) = STATE_EMISSION; ;} @@ -3650,7 +3654,7 @@ yyreduce: case 156: /* Line 1455 of yacc.c */ -#line 1359 "program_parse.y" +#line 1363 "program_parse.y" { (yyval.integer) = STATE_SHININESS; ;} @@ -3659,7 +3663,7 @@ yyreduce: case 157: /* Line 1455 of yacc.c */ -#line 1365 "program_parse.y" +#line 1369 "program_parse.y" { memset((yyval.state), 0, sizeof((yyval.state))); (yyval.state)[0] = STATE_LIGHT; @@ -3671,7 +3675,7 @@ yyreduce: case 158: /* Line 1455 of yacc.c */ -#line 1374 "program_parse.y" +#line 1378 "program_parse.y" { (yyval.integer) = (yyvsp[(1) - (1)].integer); ;} @@ -3680,7 +3684,7 @@ yyreduce: case 159: /* Line 1455 of yacc.c */ -#line 1378 "program_parse.y" +#line 1382 "program_parse.y" { (yyval.integer) = STATE_POSITION; ;} @@ -3689,7 +3693,7 @@ yyreduce: case 160: /* Line 1455 of yacc.c */ -#line 1382 "program_parse.y" +#line 1386 "program_parse.y" { if (!state->ctx->Extensions.EXT_point_parameters) { yyerror(& (yylsp[(1) - (1)]), state, "GL_ARB_point_parameters not supported"); @@ -3703,7 +3707,7 @@ yyreduce: case 161: /* Line 1455 of yacc.c */ -#line 1391 "program_parse.y" +#line 1395 "program_parse.y" { (yyval.integer) = (yyvsp[(2) - (2)].integer); ;} @@ -3712,7 +3716,7 @@ yyreduce: case 162: /* Line 1455 of yacc.c */ -#line 1395 "program_parse.y" +#line 1399 "program_parse.y" { (yyval.integer) = STATE_HALF_VECTOR; ;} @@ -3721,7 +3725,7 @@ yyreduce: case 163: /* Line 1455 of yacc.c */ -#line 1401 "program_parse.y" +#line 1405 "program_parse.y" { (yyval.integer) = STATE_SPOT_DIRECTION; ;} @@ -3730,7 +3734,7 @@ yyreduce: case 164: /* Line 1455 of yacc.c */ -#line 1407 "program_parse.y" +#line 1411 "program_parse.y" { (yyval.state)[0] = (yyvsp[(2) - (2)].state)[0]; (yyval.state)[1] = (yyvsp[(2) - (2)].state)[1]; @@ -3740,7 +3744,7 @@ yyreduce: case 165: /* Line 1455 of yacc.c */ -#line 1414 "program_parse.y" +#line 1418 "program_parse.y" { memset((yyval.state), 0, sizeof((yyval.state))); (yyval.state)[0] = STATE_LIGHTMODEL_AMBIENT; @@ -3750,7 +3754,7 @@ yyreduce: case 166: /* Line 1455 of yacc.c */ -#line 1419 "program_parse.y" +#line 1423 "program_parse.y" { memset((yyval.state), 0, sizeof((yyval.state))); (yyval.state)[0] = STATE_LIGHTMODEL_SCENECOLOR; @@ -3761,7 +3765,7 @@ yyreduce: case 167: /* Line 1455 of yacc.c */ -#line 1427 "program_parse.y" +#line 1431 "program_parse.y" { memset((yyval.state), 0, sizeof((yyval.state))); (yyval.state)[0] = STATE_LIGHTPROD; @@ -3774,7 +3778,7 @@ yyreduce: case 169: /* Line 1455 of yacc.c */ -#line 1439 "program_parse.y" +#line 1443 "program_parse.y" { memset((yyval.state), 0, sizeof((yyval.state))); (yyval.state)[0] = (yyvsp[(3) - (3)].integer); @@ -3785,7 +3789,7 @@ yyreduce: case 170: /* Line 1455 of yacc.c */ -#line 1447 "program_parse.y" +#line 1451 "program_parse.y" { (yyval.integer) = STATE_TEXENV_COLOR; ;} @@ -3794,7 +3798,7 @@ yyreduce: case 171: /* Line 1455 of yacc.c */ -#line 1453 "program_parse.y" +#line 1457 "program_parse.y" { (yyval.integer) = STATE_AMBIENT; ;} @@ -3803,7 +3807,7 @@ yyreduce: case 172: /* Line 1455 of yacc.c */ -#line 1457 "program_parse.y" +#line 1461 "program_parse.y" { (yyval.integer) = STATE_DIFFUSE; ;} @@ -3812,7 +3816,7 @@ yyreduce: case 173: /* Line 1455 of yacc.c */ -#line 1461 "program_parse.y" +#line 1465 "program_parse.y" { (yyval.integer) = STATE_SPECULAR; ;} @@ -3821,7 +3825,7 @@ yyreduce: case 174: /* Line 1455 of yacc.c */ -#line 1467 "program_parse.y" +#line 1471 "program_parse.y" { if ((unsigned) (yyvsp[(1) - (1)].integer) >= state->MaxLights) { yyerror(& (yylsp[(1) - (1)]), state, "invalid light selector"); @@ -3835,7 +3839,7 @@ yyreduce: case 175: /* Line 1455 of yacc.c */ -#line 1478 "program_parse.y" +#line 1482 "program_parse.y" { memset((yyval.state), 0, sizeof((yyval.state))); (yyval.state)[0] = STATE_TEXGEN; @@ -3847,7 +3851,7 @@ yyreduce: case 176: /* Line 1455 of yacc.c */ -#line 1487 "program_parse.y" +#line 1491 "program_parse.y" { (yyval.integer) = STATE_TEXGEN_EYE_S; ;} @@ -3856,7 +3860,7 @@ yyreduce: case 177: /* Line 1455 of yacc.c */ -#line 1491 "program_parse.y" +#line 1495 "program_parse.y" { (yyval.integer) = STATE_TEXGEN_OBJECT_S; ;} @@ -3865,7 +3869,7 @@ yyreduce: case 178: /* Line 1455 of yacc.c */ -#line 1496 "program_parse.y" +#line 1500 "program_parse.y" { (yyval.integer) = STATE_TEXGEN_EYE_S - STATE_TEXGEN_EYE_S; ;} @@ -3874,7 +3878,7 @@ yyreduce: case 179: /* Line 1455 of yacc.c */ -#line 1500 "program_parse.y" +#line 1504 "program_parse.y" { (yyval.integer) = STATE_TEXGEN_EYE_T - STATE_TEXGEN_EYE_S; ;} @@ -3883,7 +3887,7 @@ yyreduce: case 180: /* Line 1455 of yacc.c */ -#line 1504 "program_parse.y" +#line 1508 "program_parse.y" { (yyval.integer) = STATE_TEXGEN_EYE_R - STATE_TEXGEN_EYE_S; ;} @@ -3892,7 +3896,7 @@ yyreduce: case 181: /* Line 1455 of yacc.c */ -#line 1508 "program_parse.y" +#line 1512 "program_parse.y" { (yyval.integer) = STATE_TEXGEN_EYE_Q - STATE_TEXGEN_EYE_S; ;} @@ -3901,7 +3905,7 @@ yyreduce: case 182: /* Line 1455 of yacc.c */ -#line 1514 "program_parse.y" +#line 1518 "program_parse.y" { memset((yyval.state), 0, sizeof((yyval.state))); (yyval.state)[0] = (yyvsp[(2) - (2)].integer); @@ -3911,7 +3915,7 @@ yyreduce: case 183: /* Line 1455 of yacc.c */ -#line 1521 "program_parse.y" +#line 1525 "program_parse.y" { (yyval.integer) = STATE_FOG_COLOR; ;} @@ -3920,7 +3924,7 @@ yyreduce: case 184: /* Line 1455 of yacc.c */ -#line 1525 "program_parse.y" +#line 1529 "program_parse.y" { (yyval.integer) = STATE_FOG_PARAMS; ;} @@ -3929,7 +3933,7 @@ yyreduce: case 185: /* Line 1455 of yacc.c */ -#line 1531 "program_parse.y" +#line 1535 "program_parse.y" { memset((yyval.state), 0, sizeof((yyval.state))); (yyval.state)[0] = STATE_CLIPPLANE; @@ -3940,7 +3944,7 @@ yyreduce: case 186: /* Line 1455 of yacc.c */ -#line 1539 "program_parse.y" +#line 1543 "program_parse.y" { if ((unsigned) (yyvsp[(1) - (1)].integer) >= state->MaxClipPlanes) { yyerror(& (yylsp[(1) - (1)]), state, "invalid clip plane selector"); @@ -3954,7 +3958,7 @@ yyreduce: case 187: /* Line 1455 of yacc.c */ -#line 1550 "program_parse.y" +#line 1554 "program_parse.y" { memset((yyval.state), 0, sizeof((yyval.state))); (yyval.state)[0] = (yyvsp[(2) - (2)].integer); @@ -3964,7 +3968,7 @@ yyreduce: case 188: /* Line 1455 of yacc.c */ -#line 1557 "program_parse.y" +#line 1561 "program_parse.y" { (yyval.integer) = STATE_POINT_SIZE; ;} @@ -3973,7 +3977,7 @@ yyreduce: case 189: /* Line 1455 of yacc.c */ -#line 1561 "program_parse.y" +#line 1565 "program_parse.y" { (yyval.integer) = STATE_POINT_ATTENUATION; ;} @@ -3982,7 +3986,7 @@ yyreduce: case 190: /* Line 1455 of yacc.c */ -#line 1567 "program_parse.y" +#line 1571 "program_parse.y" { (yyval.state)[0] = (yyvsp[(1) - (5)].state)[0]; (yyval.state)[1] = (yyvsp[(1) - (5)].state)[1]; @@ -3995,7 +3999,7 @@ yyreduce: case 191: /* Line 1455 of yacc.c */ -#line 1577 "program_parse.y" +#line 1581 "program_parse.y" { (yyval.state)[0] = (yyvsp[(1) - (2)].state)[0]; (yyval.state)[1] = (yyvsp[(1) - (2)].state)[1]; @@ -4008,7 +4012,7 @@ yyreduce: case 192: /* Line 1455 of yacc.c */ -#line 1587 "program_parse.y" +#line 1591 "program_parse.y" { (yyval.state)[2] = 0; (yyval.state)[3] = 3; @@ -4018,7 +4022,7 @@ yyreduce: case 193: /* Line 1455 of yacc.c */ -#line 1592 "program_parse.y" +#line 1596 "program_parse.y" { /* It seems logical that the matrix row range specifier would have * to specify a range or more than one row (i.e., $5 > $3). @@ -4039,7 +4043,7 @@ yyreduce: case 194: /* Line 1455 of yacc.c */ -#line 1610 "program_parse.y" +#line 1614 "program_parse.y" { (yyval.state)[0] = (yyvsp[(2) - (3)].state)[0]; (yyval.state)[1] = (yyvsp[(2) - (3)].state)[1]; @@ -4050,7 +4054,7 @@ yyreduce: case 195: /* Line 1455 of yacc.c */ -#line 1618 "program_parse.y" +#line 1622 "program_parse.y" { (yyval.integer) = 0; ;} @@ -4059,7 +4063,7 @@ yyreduce: case 196: /* Line 1455 of yacc.c */ -#line 1622 "program_parse.y" +#line 1626 "program_parse.y" { (yyval.integer) = (yyvsp[(1) - (1)].integer); ;} @@ -4068,7 +4072,7 @@ yyreduce: case 197: /* Line 1455 of yacc.c */ -#line 1628 "program_parse.y" +#line 1632 "program_parse.y" { (yyval.integer) = STATE_MATRIX_INVERSE; ;} @@ -4077,7 +4081,7 @@ yyreduce: case 198: /* Line 1455 of yacc.c */ -#line 1632 "program_parse.y" +#line 1636 "program_parse.y" { (yyval.integer) = STATE_MATRIX_TRANSPOSE; ;} @@ -4086,7 +4090,7 @@ yyreduce: case 199: /* Line 1455 of yacc.c */ -#line 1636 "program_parse.y" +#line 1640 "program_parse.y" { (yyval.integer) = STATE_MATRIX_INVTRANS; ;} @@ -4095,7 +4099,7 @@ yyreduce: case 200: /* Line 1455 of yacc.c */ -#line 1642 "program_parse.y" +#line 1646 "program_parse.y" { if ((yyvsp[(1) - (1)].integer) > 3) { yyerror(& (yylsp[(1) - (1)]), state, "invalid matrix row reference"); @@ -4109,7 +4113,7 @@ yyreduce: case 201: /* Line 1455 of yacc.c */ -#line 1653 "program_parse.y" +#line 1657 "program_parse.y" { (yyval.state)[0] = STATE_MODELVIEW_MATRIX; (yyval.state)[1] = (yyvsp[(2) - (2)].integer); @@ -4119,7 +4123,7 @@ yyreduce: case 202: /* Line 1455 of yacc.c */ -#line 1658 "program_parse.y" +#line 1662 "program_parse.y" { (yyval.state)[0] = STATE_PROJECTION_MATRIX; (yyval.state)[1] = 0; @@ -4129,7 +4133,7 @@ yyreduce: case 203: /* Line 1455 of yacc.c */ -#line 1663 "program_parse.y" +#line 1667 "program_parse.y" { (yyval.state)[0] = STATE_MVP_MATRIX; (yyval.state)[1] = 0; @@ -4139,7 +4143,7 @@ yyreduce: case 204: /* Line 1455 of yacc.c */ -#line 1668 "program_parse.y" +#line 1672 "program_parse.y" { (yyval.state)[0] = STATE_TEXTURE_MATRIX; (yyval.state)[1] = (yyvsp[(2) - (2)].integer); @@ -4149,7 +4153,7 @@ yyreduce: case 205: /* Line 1455 of yacc.c */ -#line 1673 "program_parse.y" +#line 1677 "program_parse.y" { yyerror(& (yylsp[(1) - (4)]), state, "GL_ARB_matrix_palette not supported"); YYERROR; @@ -4159,7 +4163,7 @@ yyreduce: case 206: /* Line 1455 of yacc.c */ -#line 1678 "program_parse.y" +#line 1682 "program_parse.y" { (yyval.state)[0] = STATE_PROGRAM_MATRIX; (yyval.state)[1] = (yyvsp[(3) - (4)].integer); @@ -4169,7 +4173,7 @@ yyreduce: case 207: /* Line 1455 of yacc.c */ -#line 1685 "program_parse.y" +#line 1689 "program_parse.y" { (yyval.integer) = 0; ;} @@ -4178,7 +4182,7 @@ yyreduce: case 208: /* Line 1455 of yacc.c */ -#line 1689 "program_parse.y" +#line 1693 "program_parse.y" { (yyval.integer) = (yyvsp[(2) - (3)].integer); ;} @@ -4187,7 +4191,7 @@ yyreduce: case 209: /* Line 1455 of yacc.c */ -#line 1694 "program_parse.y" +#line 1698 "program_parse.y" { /* Since GL_ARB_vertex_blend isn't supported, only modelview matrix * zero is valid. @@ -4204,7 +4208,7 @@ yyreduce: case 210: /* Line 1455 of yacc.c */ -#line 1707 "program_parse.y" +#line 1711 "program_parse.y" { /* Since GL_ARB_matrix_palette isn't supported, just let any value * through here. The error will be generated later. @@ -4216,7 +4220,7 @@ yyreduce: case 211: /* Line 1455 of yacc.c */ -#line 1715 "program_parse.y" +#line 1719 "program_parse.y" { if ((unsigned) (yyvsp[(1) - (1)].integer) >= state->MaxProgramMatrices) { yyerror(& (yylsp[(1) - (1)]), state, "invalid program matrix selector"); @@ -4230,7 +4234,7 @@ yyreduce: case 212: /* Line 1455 of yacc.c */ -#line 1726 "program_parse.y" +#line 1730 "program_parse.y" { memset((yyval.state), 0, sizeof((yyval.state))); (yyval.state)[0] = STATE_DEPTH_RANGE; @@ -4240,7 +4244,7 @@ yyreduce: case 217: /* Line 1455 of yacc.c */ -#line 1738 "program_parse.y" +#line 1742 "program_parse.y" { memset((yyval.state), 0, sizeof((yyval.state))); (yyval.state)[0] = state->state_param_enum; @@ -4253,7 +4257,7 @@ yyreduce: case 218: /* Line 1455 of yacc.c */ -#line 1748 "program_parse.y" +#line 1752 "program_parse.y" { (yyval.state)[0] = (yyvsp[(1) - (1)].integer); (yyval.state)[1] = (yyvsp[(1) - (1)].integer); @@ -4263,7 +4267,7 @@ yyreduce: case 219: /* Line 1455 of yacc.c */ -#line 1753 "program_parse.y" +#line 1757 "program_parse.y" { (yyval.state)[0] = (yyvsp[(1) - (3)].integer); (yyval.state)[1] = (yyvsp[(3) - (3)].integer); @@ -4273,7 +4277,7 @@ yyreduce: case 220: /* Line 1455 of yacc.c */ -#line 1760 "program_parse.y" +#line 1764 "program_parse.y" { memset((yyval.state), 0, sizeof((yyval.state))); (yyval.state)[0] = state->state_param_enum; @@ -4286,7 +4290,7 @@ yyreduce: case 221: /* Line 1455 of yacc.c */ -#line 1770 "program_parse.y" +#line 1774 "program_parse.y" { memset((yyval.state), 0, sizeof((yyval.state))); (yyval.state)[0] = state->state_param_enum; @@ -4299,7 +4303,7 @@ yyreduce: case 222: /* Line 1455 of yacc.c */ -#line 1779 "program_parse.y" +#line 1783 "program_parse.y" { (yyval.state)[0] = (yyvsp[(1) - (1)].integer); (yyval.state)[1] = (yyvsp[(1) - (1)].integer); @@ -4309,7 +4313,7 @@ yyreduce: case 223: /* Line 1455 of yacc.c */ -#line 1784 "program_parse.y" +#line 1788 "program_parse.y" { (yyval.state)[0] = (yyvsp[(1) - (3)].integer); (yyval.state)[1] = (yyvsp[(3) - (3)].integer); @@ -4319,7 +4323,7 @@ yyreduce: case 224: /* Line 1455 of yacc.c */ -#line 1791 "program_parse.y" +#line 1795 "program_parse.y" { memset((yyval.state), 0, sizeof((yyval.state))); (yyval.state)[0] = state->state_param_enum; @@ -4332,7 +4336,7 @@ yyreduce: case 225: /* Line 1455 of yacc.c */ -#line 1801 "program_parse.y" +#line 1805 "program_parse.y" { if ((unsigned) (yyvsp[(1) - (1)].integer) >= state->limits->MaxEnvParams) { yyerror(& (yylsp[(1) - (1)]), state, "invalid environment parameter reference"); @@ -4345,7 +4349,7 @@ yyreduce: case 226: /* Line 1455 of yacc.c */ -#line 1811 "program_parse.y" +#line 1815 "program_parse.y" { if ((unsigned) (yyvsp[(1) - (1)].integer) >= state->limits->MaxLocalParams) { yyerror(& (yylsp[(1) - (1)]), state, "invalid local parameter reference"); @@ -4358,7 +4362,7 @@ yyreduce: case 231: /* Line 1455 of yacc.c */ -#line 1826 "program_parse.y" +#line 1830 "program_parse.y" { (yyval.vector).count = 4; (yyval.vector).data[0] = (yyvsp[(1) - (1)].real); @@ -4371,7 +4375,7 @@ yyreduce: case 232: /* Line 1455 of yacc.c */ -#line 1836 "program_parse.y" +#line 1840 "program_parse.y" { (yyval.vector).count = 1; (yyval.vector).data[0] = (yyvsp[(1) - (1)].real); @@ -4384,7 +4388,7 @@ yyreduce: case 233: /* Line 1455 of yacc.c */ -#line 1844 "program_parse.y" +#line 1848 "program_parse.y" { (yyval.vector).count = 1; (yyval.vector).data[0] = (float) (yyvsp[(1) - (1)].integer); @@ -4397,7 +4401,7 @@ yyreduce: case 234: /* Line 1455 of yacc.c */ -#line 1854 "program_parse.y" +#line 1858 "program_parse.y" { (yyval.vector).count = 4; (yyval.vector).data[0] = (yyvsp[(2) - (3)].real); @@ -4410,7 +4414,7 @@ yyreduce: case 235: /* Line 1455 of yacc.c */ -#line 1862 "program_parse.y" +#line 1866 "program_parse.y" { (yyval.vector).count = 4; (yyval.vector).data[0] = (yyvsp[(2) - (5)].real); @@ -4423,7 +4427,7 @@ yyreduce: case 236: /* Line 1455 of yacc.c */ -#line 1871 "program_parse.y" +#line 1875 "program_parse.y" { (yyval.vector).count = 4; (yyval.vector).data[0] = (yyvsp[(2) - (7)].real); @@ -4436,7 +4440,7 @@ yyreduce: case 237: /* Line 1455 of yacc.c */ -#line 1880 "program_parse.y" +#line 1884 "program_parse.y" { (yyval.vector).count = 4; (yyval.vector).data[0] = (yyvsp[(2) - (9)].real); @@ -4449,7 +4453,7 @@ yyreduce: case 238: /* Line 1455 of yacc.c */ -#line 1890 "program_parse.y" +#line 1894 "program_parse.y" { (yyval.real) = ((yyvsp[(1) - (2)].negate)) ? -(yyvsp[(2) - (2)].real) : (yyvsp[(2) - (2)].real); ;} @@ -4458,7 +4462,7 @@ yyreduce: case 239: /* Line 1455 of yacc.c */ -#line 1894 "program_parse.y" +#line 1898 "program_parse.y" { (yyval.real) = (float)(((yyvsp[(1) - (2)].negate)) ? -(yyvsp[(2) - (2)].integer) : (yyvsp[(2) - (2)].integer)); ;} @@ -4467,35 +4471,35 @@ yyreduce: case 240: /* Line 1455 of yacc.c */ -#line 1899 "program_parse.y" +#line 1903 "program_parse.y" { (yyval.negate) = FALSE; ;} break; case 241: /* Line 1455 of yacc.c */ -#line 1900 "program_parse.y" +#line 1904 "program_parse.y" { (yyval.negate) = TRUE; ;} break; case 242: /* Line 1455 of yacc.c */ -#line 1901 "program_parse.y" +#line 1905 "program_parse.y" { (yyval.negate) = FALSE; ;} break; case 243: /* Line 1455 of yacc.c */ -#line 1904 "program_parse.y" +#line 1908 "program_parse.y" { (yyval.integer) = (yyvsp[(2) - (2)].integer); ;} break; case 245: /* Line 1455 of yacc.c */ -#line 1908 "program_parse.y" +#line 1912 "program_parse.y" { /* NV_fragment_program_option defines the size qualifiers in a * fairly broken way. "SHORT" or "LONG" can optionally be used @@ -4534,7 +4538,7 @@ yyreduce: case 246: /* Line 1455 of yacc.c */ -#line 1942 "program_parse.y" +#line 1946 "program_parse.y" { ;} break; @@ -4542,14 +4546,14 @@ yyreduce: case 247: /* Line 1455 of yacc.c */ -#line 1946 "program_parse.y" +#line 1950 "program_parse.y" { (yyval.integer) = (yyvsp[(1) - (1)].integer); ;} break; case 249: /* Line 1455 of yacc.c */ -#line 1950 "program_parse.y" +#line 1954 "program_parse.y" { if (!declare_variable(state, (yyvsp[(3) - (3)].string), (yyvsp[(0) - (3)].integer), & (yylsp[(3) - (3)]))) { YYERROR; @@ -4560,7 +4564,7 @@ yyreduce: case 250: /* Line 1455 of yacc.c */ -#line 1956 "program_parse.y" +#line 1960 "program_parse.y" { if (!declare_variable(state, (yyvsp[(1) - (1)].string), (yyvsp[(0) - (1)].integer), & (yylsp[(1) - (1)]))) { YYERROR; @@ -4571,7 +4575,7 @@ yyreduce: case 251: /* Line 1455 of yacc.c */ -#line 1964 "program_parse.y" +#line 1968 "program_parse.y" { struct asm_symbol *const s = declare_variable(state, (yyvsp[(3) - (5)].string), at_output, & (yylsp[(3) - (5)])); @@ -4587,7 +4591,7 @@ yyreduce: case 252: /* Line 1455 of yacc.c */ -#line 1977 "program_parse.y" +#line 1981 "program_parse.y" { if (state->mode == ARB_vertex) { (yyval.result) = VERT_RESULT_HPOS; @@ -4601,7 +4605,7 @@ yyreduce: case 253: /* Line 1455 of yacc.c */ -#line 1986 "program_parse.y" +#line 1990 "program_parse.y" { if (state->mode == ARB_vertex) { (yyval.result) = VERT_RESULT_FOGC; @@ -4615,7 +4619,7 @@ yyreduce: case 254: /* Line 1455 of yacc.c */ -#line 1995 "program_parse.y" +#line 1999 "program_parse.y" { (yyval.result) = (yyvsp[(2) - (2)].result); ;} @@ -4624,7 +4628,7 @@ yyreduce: case 255: /* Line 1455 of yacc.c */ -#line 1999 "program_parse.y" +#line 2003 "program_parse.y" { if (state->mode == ARB_vertex) { (yyval.result) = VERT_RESULT_PSIZ; @@ -4638,7 +4642,7 @@ yyreduce: case 256: /* Line 1455 of yacc.c */ -#line 2008 "program_parse.y" +#line 2012 "program_parse.y" { if (state->mode == ARB_vertex) { (yyval.result) = VERT_RESULT_TEX0 + (yyvsp[(3) - (3)].integer); @@ -4652,7 +4656,7 @@ yyreduce: case 257: /* Line 1455 of yacc.c */ -#line 2017 "program_parse.y" +#line 2021 "program_parse.y" { if (state->mode == ARB_fragment) { (yyval.result) = FRAG_RESULT_DEPTH; @@ -4666,7 +4670,7 @@ yyreduce: case 258: /* Line 1455 of yacc.c */ -#line 2028 "program_parse.y" +#line 2032 "program_parse.y" { (yyval.result) = (yyvsp[(2) - (3)].integer) + (yyvsp[(3) - (3)].integer); ;} @@ -4675,7 +4679,7 @@ yyreduce: case 259: /* Line 1455 of yacc.c */ -#line 2034 "program_parse.y" +#line 2038 "program_parse.y" { (yyval.integer) = (state->mode == ARB_vertex) ? VERT_RESULT_COL0 @@ -4686,7 +4690,7 @@ yyreduce: case 260: /* Line 1455 of yacc.c */ -#line 2040 "program_parse.y" +#line 2044 "program_parse.y" { if (state->mode == ARB_vertex) { (yyval.integer) = VERT_RESULT_COL0; @@ -4700,7 +4704,7 @@ yyreduce: case 261: /* Line 1455 of yacc.c */ -#line 2049 "program_parse.y" +#line 2053 "program_parse.y" { if (state->mode == ARB_vertex) { (yyval.integer) = VERT_RESULT_BFC0; @@ -4714,7 +4718,7 @@ yyreduce: case 262: /* Line 1455 of yacc.c */ -#line 2060 "program_parse.y" +#line 2064 "program_parse.y" { (yyval.integer) = 0; ;} @@ -4723,7 +4727,7 @@ yyreduce: case 263: /* Line 1455 of yacc.c */ -#line 2064 "program_parse.y" +#line 2068 "program_parse.y" { if (state->mode == ARB_vertex) { (yyval.integer) = 0; @@ -4737,7 +4741,7 @@ yyreduce: case 264: /* Line 1455 of yacc.c */ -#line 2073 "program_parse.y" +#line 2077 "program_parse.y" { if (state->mode == ARB_vertex) { (yyval.integer) = 1; @@ -4751,91 +4755,91 @@ yyreduce: case 265: /* Line 1455 of yacc.c */ -#line 2083 "program_parse.y" +#line 2087 "program_parse.y" { (yyval.integer) = 0; ;} break; case 266: /* Line 1455 of yacc.c */ -#line 2084 "program_parse.y" +#line 2088 "program_parse.y" { (yyval.integer) = 0; ;} break; case 267: /* Line 1455 of yacc.c */ -#line 2085 "program_parse.y" +#line 2089 "program_parse.y" { (yyval.integer) = 1; ;} break; case 268: /* Line 1455 of yacc.c */ -#line 2088 "program_parse.y" +#line 2092 "program_parse.y" { (yyval.integer) = 0; ;} break; case 269: /* Line 1455 of yacc.c */ -#line 2089 "program_parse.y" +#line 2093 "program_parse.y" { (yyval.integer) = 0; ;} break; case 270: /* Line 1455 of yacc.c */ -#line 2090 "program_parse.y" +#line 2094 "program_parse.y" { (yyval.integer) = 1; ;} break; case 271: /* Line 1455 of yacc.c */ -#line 2093 "program_parse.y" +#line 2097 "program_parse.y" { (yyval.integer) = 0; ;} break; case 272: /* Line 1455 of yacc.c */ -#line 2094 "program_parse.y" +#line 2098 "program_parse.y" { (yyval.integer) = (yyvsp[(2) - (3)].integer); ;} break; case 273: /* Line 1455 of yacc.c */ -#line 2097 "program_parse.y" +#line 2101 "program_parse.y" { (yyval.integer) = 0; ;} break; case 274: /* Line 1455 of yacc.c */ -#line 2098 "program_parse.y" +#line 2102 "program_parse.y" { (yyval.integer) = (yyvsp[(2) - (3)].integer); ;} break; case 275: /* Line 1455 of yacc.c */ -#line 2101 "program_parse.y" +#line 2105 "program_parse.y" { (yyval.integer) = 0; ;} break; case 276: /* Line 1455 of yacc.c */ -#line 2102 "program_parse.y" +#line 2106 "program_parse.y" { (yyval.integer) = (yyvsp[(2) - (3)].integer); ;} break; case 277: /* Line 1455 of yacc.c */ -#line 2106 "program_parse.y" +#line 2110 "program_parse.y" { if ((unsigned) (yyvsp[(1) - (1)].integer) >= state->MaxTextureCoordUnits) { yyerror(& (yylsp[(1) - (1)]), state, "invalid texture coordinate unit selector"); @@ -4849,7 +4853,7 @@ yyreduce: case 278: /* Line 1455 of yacc.c */ -#line 2117 "program_parse.y" +#line 2121 "program_parse.y" { if ((unsigned) (yyvsp[(1) - (1)].integer) >= state->MaxTextureImageUnits) { yyerror(& (yylsp[(1) - (1)]), state, "invalid texture image unit selector"); @@ -4863,7 +4867,7 @@ yyreduce: case 279: /* Line 1455 of yacc.c */ -#line 2128 "program_parse.y" +#line 2132 "program_parse.y" { if ((unsigned) (yyvsp[(1) - (1)].integer) >= state->MaxTextureUnits) { yyerror(& (yylsp[(1) - (1)]), state, "invalid texture unit selector"); @@ -4877,7 +4881,7 @@ yyreduce: case 280: /* Line 1455 of yacc.c */ -#line 2139 "program_parse.y" +#line 2143 "program_parse.y" { struct asm_symbol *exist = (struct asm_symbol *) _mesa_symbol_table_find_symbol(state->st, 0, (yyvsp[(2) - (4)].string)); @@ -4901,7 +4905,7 @@ yyreduce: /* Line 1455 of yacc.c */ -#line 4905 "program_parse.tab.c" +#line 4909 "program_parse.tab.c" default: break; } YY_SYMBOL_PRINT ("-> $$ =", yyr1[yyn], &yyval, &yyloc); @@ -5120,7 +5124,7 @@ yyreturn: /* Line 1675 of yacc.c */ -#line 2163 "program_parse.y" +#line 2167 "program_parse.y" void diff --git a/src/mesa/shader/program_parse.y b/src/mesa/shader/program_parse.y index f70ea92cb05..9703e8e670f 100644 --- a/src/mesa/shader/program_parse.y +++ b/src/mesa/shader/program_parse.y @@ -923,8 +923,10 @@ addrRegRelOffset: { $$ = 0; } addrRegPosOffset: INTEGER { if (($1 < 0) || ($1 > 63)) { - yyerror(& @1, state, - "relative address offset too large (positive)"); + char s[100]; + _mesa_snprintf(s, sizeof(s), + "relative address offset too large (%d)", $1); + yyerror(& @1, state, s); YYERROR; } else { $$ = $1; @@ -935,8 +937,10 @@ addrRegPosOffset: INTEGER addrRegNegOffset: INTEGER { if (($1 < 0) || ($1 > 64)) { - yyerror(& @1, state, - "relative address offset too large (negative)"); + char s[100]; + _mesa_snprintf(s, sizeof(s), + "relative address offset too large (%d)", $1); + yyerror(& @1, state, s); YYERROR; } else { $$ = $1; diff --git a/src/mesa/shader/shader_api.c b/src/mesa/shader/shader_api.c index 54a25dfaf07..178b7d0dbaf 100644 --- a/src/mesa/shader/shader_api.c +++ b/src/mesa/shader/shader_api.c @@ -1474,6 +1474,21 @@ _mesa_link_program(GLcontext *ctx, GLuint program) FLUSH_VERTICES(ctx, _NEW_PROGRAM); _slang_link(ctx, program, shProg); + + /* debug code */ + if (0) { + GLuint i; + + _mesa_printf("Link %u shaders in program %u: %s\n", + shProg->NumShaders, shProg->Name, + shProg->LinkStatus ? "Success" : "Failed"); + + for (i = 0; i < shProg->NumShaders; i++) { + _mesa_printf(" shader %u, type 0x%x\n", + shProg->Shaders[i]->Name, + shProg->Shaders[i]->Type); + } + } } diff --git a/src/mesa/shader/slang/slang_builtin.c b/src/mesa/shader/slang/slang_builtin.c index bef0f856534..e5809509c93 100644 --- a/src/mesa/shader/slang/slang_builtin.c +++ b/src/mesa/shader/slang/slang_builtin.c @@ -741,7 +741,7 @@ static const struct input_info vertInputs[] = { { "gl_MultiTexCoord5", VERT_ATTRIB_TEX5, GL_FLOAT_VEC4, SWIZZLE_NOOP }, { "gl_MultiTexCoord6", VERT_ATTRIB_TEX6, GL_FLOAT_VEC4, SWIZZLE_NOOP }, { "gl_MultiTexCoord7", VERT_ATTRIB_TEX7, GL_FLOAT_VEC4, SWIZZLE_NOOP }, - { NULL, 0, SWIZZLE_NOOP } + { NULL, 0, GL_NONE, SWIZZLE_NOOP } }; /** Predefined fragment shader inputs */ @@ -754,7 +754,7 @@ static const struct input_info fragInputs[] = { { "gl_FogFragCoord", FRAG_ATTRIB_FOGC, GL_FLOAT, SWIZZLE_XXXX }, { "gl_FrontFacing", FRAG_ATTRIB_FACE, GL_FLOAT, SWIZZLE_XXXX }, { "gl_PointCoord", FRAG_ATTRIB_PNTC, GL_FLOAT_VEC2, SWIZZLE_XYZW }, - { NULL, 0, SWIZZLE_NOOP } + { NULL, 0, GL_NONE, SWIZZLE_NOOP } }; diff --git a/src/mesa/shader/slang/slang_codegen.c b/src/mesa/shader/slang/slang_codegen.c index 349f432deca..703af9f8740 100644 --- a/src/mesa/shader/slang/slang_codegen.c +++ b/src/mesa/shader/slang/slang_codegen.c @@ -2775,7 +2775,7 @@ _slang_gen_while(slang_assemble_ctx * A, slang_operation *oper) * body code (child[1]) */ slang_ir_node *loop, *breakIf, *body; - GLboolean isConst, constTrue; + GLboolean isConst, constTrue = GL_FALSE; if (!A->EmitContReturn) { /* We don't want to emit CONT instructions. If this while-loop has diff --git a/src/mesa/shader/slang/slang_link.c b/src/mesa/shader/slang/slang_link.c index 169c07f8cea..8f2b40d5dfb 100644 --- a/src/mesa/shader/slang/slang_link.c +++ b/src/mesa/shader/slang/slang_link.c @@ -546,6 +546,32 @@ _slang_update_inputs_outputs(struct gl_program *prog) +/** + * Remove extra #version directives from the concatenated source string. + * Disable the extra ones by converting first two chars to //, a comment. + * This is a bit of hack to work around a preprocessor bug that only + * allows one #version directive per source. + */ +static void +remove_extra_version_directives(GLchar *source) +{ + GLuint verCount = 0; + while (1) { + char *ver = _mesa_strstr(source, "#version"); + if (ver) { + verCount++; + if (verCount > 1) { + ver[0] = '/'; + ver[1] = '/'; + } + source += 8; + } + else { + break; + } + } +} + /** @@ -593,6 +619,8 @@ concat_shaders(struct gl_shader_program *shProg, GLenum shaderType) _mesa_printf("---NEW CONCATENATED SHADER---:\n%s\n------------\n", source); */ + remove_extra_version_directives(source); + newShader = CALLOC_STRUCT(gl_shader); newShader->Type = shaderType; newShader->Source = source; diff --git a/src/mesa/sources.mak b/src/mesa/sources.mak index fa2a6307a4b..7107538cee7 100644 --- a/src/mesa/sources.mak +++ b/src/mesa/sources.mak @@ -115,7 +115,6 @@ SWRAST_SOURCES = \ swrast/s_feedback.c \ swrast/s_fog.c \ swrast/s_fragprog.c \ - swrast/s_imaging.c \ swrast/s_lines.c \ swrast/s_logic.c \ swrast/s_masking.c \ @@ -125,7 +124,6 @@ SWRAST_SOURCES = \ swrast/s_stencil.c \ swrast/s_texcombine.c \ swrast/s_texfilter.c \ - swrast/s_texstore.c \ swrast/s_triangle.c \ swrast/s_zoom.c diff --git a/src/mesa/sparc/glapi_sparc.S b/src/mesa/sparc/glapi_sparc.S index 50a20370349..aaa17e6a3b9 100644 --- a/src/mesa/sparc/glapi_sparc.S +++ b/src/mesa/sparc/glapi_sparc.S @@ -1362,6 +1362,7 @@ gl_dispatch_functions_start: GL_STUB_ALIAS(glIsRenderbuffer, glIsRenderbufferEXT) GL_STUB_ALIAS(glRenderbufferStorage, glRenderbufferStorageEXT) GL_STUB_ALIAS(glFramebufferTextureLayer, glFramebufferTextureLayerEXT) + GL_STUB_ALIAS(glProvokingVertex, glProvokingVertexEXT) .globl gl_dispatch_functions_end HIDDEN(gl_dispatch_functions_end) diff --git a/src/mesa/state_tracker/st_cb_blit.c b/src/mesa/state_tracker/st_cb_blit.c index c741940bcf4..5626e25323e 100644 --- a/src/mesa/state_tracker/st_cb_blit.c +++ b/src/mesa/state_tracker/st_cb_blit.c @@ -39,6 +39,7 @@ #include "shader/prog_print.h" #include "st_context.h" +#include "st_texture.h" #include "st_program.h" #include "st_cb_blit.h" #include "st_cb_fbo.h" @@ -110,17 +111,50 @@ st_BlitFramebuffer(GLcontext *ctx, } if (mask & GL_COLOR_BUFFER_BIT) { - struct st_renderbuffer *srcRb = - st_renderbuffer(readFB->_ColorReadBuffer); - struct st_renderbuffer *dstRb = - st_renderbuffer(drawFB->_ColorDrawBuffers[0]); - struct pipe_surface *srcSurf = srcRb->surface; - struct pipe_surface *dstSurf = dstRb->surface; - - util_blit_pixels(st->blit, - srcSurf, srcX0, srcY0, srcX1, srcY1, - dstSurf, dstX0, dstY0, dstX1, dstY1, - 0.0, pFilter); + struct gl_renderbuffer_attachment *srcAtt = + &readFB->Attachment[readFB->_ColorReadBufferIndex]; + + if(srcAtt->Type == GL_TEXTURE) { + struct pipe_screen *screen = ctx->st->pipe->screen; + const struct st_texture_object *srcObj = + st_texture_object(srcAtt->Texture); + struct st_renderbuffer *dstRb = + st_renderbuffer(drawFB->_ColorDrawBuffers[0]); + struct pipe_surface *srcSurf; + struct pipe_surface *dstSurf = dstRb->surface; + + if (!srcObj->pt) + return; + + srcSurf = screen->get_tex_surface(screen, + srcObj->pt, + srcAtt->CubeMapFace, + srcAtt->TextureLevel, + srcAtt->Zoffset, + PIPE_BUFFER_USAGE_GPU_READ); + if(!srcSurf) + return; + + util_blit_pixels(st->blit, + srcSurf, srcX0, srcY0, srcX1, srcY1, + dstSurf, dstX0, dstY0, dstX1, dstY1, + 0.0, pFilter); + + pipe_surface_reference(&srcSurf, NULL); + } + else { + struct st_renderbuffer *srcRb = + st_renderbuffer(readFB->_ColorReadBuffer); + struct st_renderbuffer *dstRb = + st_renderbuffer(drawFB->_ColorDrawBuffers[0]); + struct pipe_surface *srcSurf = srcRb->surface; + struct pipe_surface *dstSurf = dstRb->surface; + + util_blit_pixels(st->blit, + srcSurf, srcX0, srcY0, srcX1, srcY1, + dstSurf, dstX0, dstY0, dstX1, dstY1, + 0.0, pFilter); + } } if (mask & depthStencil) { diff --git a/src/mesa/state_tracker/st_cb_drawpixels.c b/src/mesa/state_tracker/st_cb_drawpixels.c index a9cafbf8cdc..99f3ba678bb 100644 --- a/src/mesa/state_tracker/st_cb_drawpixels.c +++ b/src/mesa/state_tracker/st_cb_drawpixels.c @@ -98,7 +98,7 @@ is_passthrough_program(const struct gl_fragment_program *prog) static struct st_fragment_program * combined_drawpix_fragment_program(GLcontext *ctx) { - struct st_context *st = ctx->st; + struct st_context *st = st_context(ctx); struct st_fragment_program *stfp; if (st->pixel_xfer.program->serialNo == st->pixel_xfer.xfer_prog_sn @@ -445,8 +445,8 @@ draw_quad(GLcontext *ctx, GLfloat x0, GLfloat y0, GLfloat z, GLfloat x1, GLfloat y1, const GLfloat *color, GLboolean invertTex, GLfloat maxXcoord, GLfloat maxYcoord) { - struct st_context *st = ctx->st; - struct pipe_context *pipe = ctx->st->pipe; + struct st_context *st = st_context(ctx); + struct pipe_context *pipe = st->pipe; GLfloat verts[4][3][4]; /* four verts, three attribs, XYZW */ /* setup vertex data */ @@ -540,9 +540,9 @@ draw_textured_quad(GLcontext *ctx, GLint x, GLint y, GLfloat z, const GLfloat *color, GLboolean invertTex) { - struct st_context *st = ctx->st; - struct pipe_context *pipe = ctx->st->pipe; - struct cso_context *cso = ctx->st->cso_context; + struct st_context *st = st_context(ctx); + struct pipe_context *pipe = st->pipe; + struct cso_context *cso = st->cso_context; GLfloat x0, y0, x1, y1; GLsizei maxSize; @@ -652,7 +652,7 @@ draw_stencil_pixels(GLcontext *ctx, GLint x, GLint y, const struct gl_pixelstore_attrib *unpack, const GLvoid *pixels) { - struct st_context *st = ctx->st; + struct st_context *st = st_context(ctx); struct pipe_context *pipe = st->pipe; struct pipe_screen *screen = pipe->screen; struct st_renderbuffer *strb; @@ -793,7 +793,7 @@ st_DrawPixels(GLcontext *ctx, GLint x, GLint y, GLsizei width, GLsizei height, { struct st_fragment_program *stfp; struct st_vertex_program *stvp; - struct st_context *st = ctx->st; + struct st_context *st = st_context(ctx); struct pipe_surface *ps; const GLfloat *color; @@ -811,21 +811,21 @@ st_DrawPixels(GLcontext *ctx, GLint x, GLint y, GLsizei width, GLsizei height, if (format == GL_DEPTH_COMPONENT) { ps = st->state.framebuffer.zsbuf; - stfp = make_fragment_shader_z(ctx->st); - stvp = st_make_passthrough_vertex_shader(ctx->st, GL_TRUE); + stfp = make_fragment_shader_z(st); + stvp = st_make_passthrough_vertex_shader(st, GL_TRUE); color = ctx->Current.RasterColor; } else { ps = st->state.framebuffer.cbufs[0]; stfp = combined_drawpix_fragment_program(ctx); - stvp = st_make_passthrough_vertex_shader(ctx->st, GL_FALSE); + stvp = st_make_passthrough_vertex_shader(st, GL_FALSE); color = NULL; } /* draw with textured quad */ { struct pipe_texture *pt - = make_texture(ctx->st, width, height, format, type, unpack, pixels); + = make_texture(st, width, height, format, type, unpack, pixels); if (pt) { draw_textured_quad(ctx, x, y, ctx->Current.RasterPos[2], width, height, ctx->Pixel.ZoomX, ctx->Pixel.ZoomY, @@ -942,7 +942,7 @@ st_CopyPixels(GLcontext *ctx, GLint srcx, GLint srcy, GLsizei width, GLsizei height, GLint dstx, GLint dsty, GLenum type) { - struct st_context *st = ctx->st; + struct st_context *st = st_context(ctx); struct pipe_context *pipe = st->pipe; struct pipe_screen *screen = pipe->screen; struct st_renderbuffer *rbRead; @@ -995,14 +995,14 @@ st_CopyPixels(GLcontext *ctx, GLint srcx, GLint srcy, rbRead = st_get_color_read_renderbuffer(ctx); color = NULL; stfp = combined_drawpix_fragment_program(ctx); - stvp = st_make_passthrough_vertex_shader(ctx->st, GL_FALSE); + stvp = st_make_passthrough_vertex_shader(st, GL_FALSE); } else { assert(type == GL_DEPTH); rbRead = st_renderbuffer(ctx->ReadBuffer->_DepthBuffer); color = ctx->Current.Attrib[VERT_ATTRIB_COLOR0]; - stfp = make_fragment_shader_z(ctx->st); - stvp = st_make_passthrough_vertex_shader(ctx->st, GL_TRUE); + stfp = make_fragment_shader_z(st); + stvp = st_make_passthrough_vertex_shader(st, GL_TRUE); } srcFormat = rbRead->texture->format; @@ -1059,7 +1059,7 @@ st_CopyPixels(GLcontext *ctx, GLint srcx, GLint srcy, assert(pth <= maxSize); } - pt = st_texture_create(ctx->st, PIPE_TEXTURE_2D, texFormat, 0, + pt = st_texture_create(st, PIPE_TEXTURE_2D, texFormat, 0, ptw, pth, 1, PIPE_TEXTURE_USAGE_SAMPLER); if (!pt) diff --git a/src/mesa/state_tracker/st_cb_fbo.c b/src/mesa/state_tracker/st_cb_fbo.c index a96602878e4..fe0a1214933 100644 --- a/src/mesa/state_tracker/st_cb_fbo.c +++ b/src/mesa/state_tracker/st_cb_fbo.c @@ -258,6 +258,7 @@ st_new_renderbuffer_fb(enum pipe_format format, int samples, boolean sw) strb->Base.ClassID = 0x4242; /* just a unique value */ strb->Base.NumSamples = samples; strb->format = format; + init_renderbuffer_bits(strb, format); strb->software = sw; switch (format) { diff --git a/src/mesa/state_tracker/st_cb_program.c b/src/mesa/state_tracker/st_cb_program.c index 4398ab28392..b2d5c39a3a0 100644 --- a/src/mesa/state_tracker/st_cb_program.c +++ b/src/mesa/state_tracker/st_cb_program.c @@ -45,6 +45,7 @@ #include "st_context.h" #include "st_program.h" #include "st_atom_shader.h" +#include "st_mesa_to_tgsi.h" #include "st_cb_program.h" @@ -152,7 +153,7 @@ st_delete_program(GLcontext *ctx, struct gl_program *prog) } if (stvp->state.tokens) { - _mesa_free((void *) stvp->state.tokens); + st_free_tokens(stvp->state.tokens); stvp->state.tokens = NULL; } } @@ -167,7 +168,7 @@ st_delete_program(GLcontext *ctx, struct gl_program *prog) } if (stfp->state.tokens) { - _mesa_free((void *) stfp->state.tokens); + st_free_tokens(stfp->state.tokens); stfp->state.tokens = NULL; } @@ -214,7 +215,7 @@ static void st_program_string_notify( GLcontext *ctx, } if (stfp->state.tokens) { - _mesa_free((void *) stfp->state.tokens); + st_free_tokens(stfp->state.tokens); stfp->state.tokens = NULL; } @@ -242,7 +243,7 @@ static void st_program_string_notify( GLcontext *ctx, } if (stvp->state.tokens) { - _mesa_free((void *) stvp->state.tokens); + st_free_tokens(stvp->state.tokens); stvp->state.tokens = NULL; } diff --git a/src/mesa/state_tracker/st_cb_texture.c b/src/mesa/state_tracker/st_cb_texture.c index abaf9d2c353..3520c986b47 100644 --- a/src/mesa/state_tracker/st_cb_texture.c +++ b/src/mesa/state_tracker/st_cb_texture.c @@ -658,8 +658,10 @@ st_TexImage(GLcontext * ctx, format, type, pixels, unpack, "glTexImage"); } - if (!pixels) - return; + + /* Note: we can't check for pixels==NULL until after we've allocated + * memory for the texture. + */ /* See if we can do texture compression with a blit/render. */ @@ -670,6 +672,9 @@ st_TexImage(GLcontext * ctx, stImage->pt->format, stImage->pt->target, PIPE_TEXTURE_USAGE_RENDER_TARGET, 0)) { + if (!pixels) + goto done; + if (compress_with_blit(ctx, target, level, 0, 0, 0, width, height, depth, format, type, pixels, unpack, texImage)) { goto done; @@ -711,6 +716,9 @@ st_TexImage(GLcontext * ctx, return; } + if (!pixels) + goto done; + DBG("Upload image %dx%dx%d row_len %x pitch %x\n", width, height, depth, width * texelBytes, dstRowStride); @@ -753,17 +761,13 @@ st_TexImage(GLcontext * ctx, } } +done: _mesa_unmap_teximage_pbo(ctx, unpack); -done: if (stImage->pt && texImage->Data) { st_texture_image_unmap(ctx->st, stImage); texImage->Data = NULL; } - - if (level == texObj->BaseLevel && texObj->GenerateMipmap) { - ctx->Driver.GenerateMipmap(ctx, target, texObj); - } } @@ -1096,7 +1100,7 @@ st_TexSubimage(GLcontext *ctx, GLint dims, GLenum target, GLint level, if (!texImage->Data) { _mesa_error(ctx, GL_OUT_OF_MEMORY, "glTexSubImage"); - return; + goto done; } src = (const GLubyte *) pixels; @@ -1127,17 +1131,13 @@ st_TexSubimage(GLcontext *ctx, GLint dims, GLenum target, GLint level, } } +done: _mesa_unmap_teximage_pbo(ctx, packing); -done: if (stImage->pt) { st_texture_image_unmap(ctx->st, stImage); texImage->Data = NULL; } - - if (level == texObj->BaseLevel && texObj->GenerateMipmap) { - ctx->Driver.GenerateMipmap(ctx, target, texObj); - } } @@ -1601,10 +1601,6 @@ st_copy_texsubimage(GLcontext *ctx, destX, destY, destZ, srcX, srcY, width, height); } - - if (level == texObj->BaseLevel && texObj->GenerateMipmap) { - ctx->Driver.GenerateMipmap(ctx, target, texObj); - } } diff --git a/src/mesa/state_tracker/st_draw.c b/src/mesa/state_tracker/st_draw.c index 503a5f34a3f..c76bff91819 100644 --- a/src/mesa/state_tracker/st_draw.c +++ b/src/mesa/state_tracker/st_draw.c @@ -328,23 +328,29 @@ get_arrays_bounds(const struct st_vertex_program *vp, const GLubyte **low, const GLubyte **high) { const GLubyte *low_addr = NULL; + const GLubyte *high_addr = NULL; GLuint attr; - GLint stride; for (attr = 0; attr < vp->num_inputs; attr++) { const GLuint mesaAttr = vp->index_to_input[attr]; + const GLint stride = arrays[mesaAttr]->StrideB; const GLubyte *start = arrays[mesaAttr]->Ptr; - stride = arrays[mesaAttr]->StrideB; + const unsigned sz = (arrays[mesaAttr]->Size * + _mesa_sizeof_type(arrays[mesaAttr]->Type)); + const GLubyte *end = start + (max_index * stride) + sz; + if (attr == 0) { low_addr = start; + high_addr = end; } else { low_addr = MIN2(low_addr, start); + high_addr = MAX2(high_addr, end); } } *low = low_addr; - *high = low_addr + (max_index + 1) * stride; + *high = high_addr; } diff --git a/src/mesa/state_tracker/st_framebuffer.c b/src/mesa/state_tracker/st_framebuffer.c index ca32b2e573c..5c0d335d62d 100644 --- a/src/mesa/state_tracker/st_framebuffer.c +++ b/src/mesa/state_tracker/st_framebuffer.c @@ -66,7 +66,7 @@ st_create_framebuffer( const __GLcontextModes *visual, else { /* Only allocate front buffer right now if we're single buffered. * If double-buffered, allocate front buffer on demand later. - * See check_create_front_buffers(). + * See check_create_front_buffers() and st_set_framebuffer_surface(). */ struct gl_renderbuffer *rb = st_new_renderbuffer_fb(colorFormat, samples, FALSE); @@ -170,8 +170,20 @@ st_set_framebuffer_surface(struct st_framebuffer *stfb, strb = st_renderbuffer(stfb->Base.Attachment[surfIndex].Renderbuffer); - /* fail */ - if (!strb) return; + if (!strb) { + if (surfIndex == ST_SURFACE_FRONT_LEFT) { + /* Delayed creation when the window system supplies a fake front buffer */ + struct st_renderbuffer *strb_back + = st_renderbuffer(stfb->Base.Attachment[ST_SURFACE_BACK_LEFT].Renderbuffer); + struct gl_renderbuffer *rb + = st_new_renderbuffer_fb(surf->format, strb_back->Base.NumSamples, FALSE); + _mesa_add_renderbuffer(&stfb->Base, BUFFER_FRONT_LEFT, rb); + strb = st_renderbuffer(rb); + } else { + /* fail */ + return; + } + } /* replace the renderbuffer's surface/texture pointers */ pipe_surface_reference( &strb->surface, surf ); diff --git a/src/mesa/state_tracker/st_mesa_to_tgsi.c b/src/mesa/state_tracker/st_mesa_to_tgsi.c index 2ab12d3cf3f..b0a1b529f1e 100644 --- a/src/mesa/state_tracker/st_mesa_to_tgsi.c +++ b/src/mesa/state_tracker/st_mesa_to_tgsi.c @@ -27,836 +27,648 @@ /* * \author - * Michal Krol + * Michal Krol, + * Keith Whitwell */ #include "pipe/p_compiler.h" #include "pipe/p_shader_tokens.h" -#include "tgsi/tgsi_parse.h" -#include "tgsi/tgsi_build.h" -#include "tgsi/tgsi_util.h" -#include "tgsi/tgsi_dump.h" -#include "tgsi/tgsi_sanity.h" +#include "pipe/p_state.h" +#include "tgsi/tgsi_ureg.h" #include "st_mesa_to_tgsi.h" #include "shader/prog_instruction.h" #include "shader/prog_parameter.h" #include "shader/prog_print.h" #include "util/u_debug.h" +#include "util/u_math.h" +#include "util/u_memory.h" + +struct label { + unsigned branch_target; + unsigned token; +}; + +struct st_translate { + struct ureg_program *ureg; + + struct ureg_dst temps[MAX_PROGRAM_TEMPS]; + struct ureg_src *constants; + struct ureg_dst outputs[PIPE_MAX_SHADER_OUTPUTS]; + struct ureg_src inputs[PIPE_MAX_SHADER_INPUTS]; + struct ureg_dst address[1]; + struct ureg_src samplers[PIPE_MAX_SAMPLERS]; + + const GLuint *inputMapping; + const GLuint *outputMapping; + + /* For every instruction that contains a label (eg CALL), keep + * details so that we can go back afterwards and emit the correct + * tgsi instruction number for each label. + */ + struct label *labels; + unsigned labels_size; + unsigned labels_count; + + /* Keep a record of the tgsi instruction number that each mesa + * instruction starts at, will be used to fix up labels after + * translation. + */ + unsigned *insn; + unsigned insn_size; + unsigned insn_count; + + unsigned procType; /**< TGSI_PROCESSOR_VERTEX/FRAGMENT */ + + boolean error; +}; + + +static unsigned *get_label( struct st_translate *t, + unsigned branch_target ) +{ + unsigned i; + + if (t->labels_count + 1 >= t->labels_size) { + unsigned old_size = t->labels_size; + t->labels_size = 1 << (util_logbase2(t->labels_size) + 1); + t->labels = REALLOC( t->labels, + old_size * sizeof t->labels[0], + t->labels_size * sizeof t->labels[0] ); + if (t->labels == NULL) { + static unsigned dummy; + t->error = TRUE; + return &dummy; + } + } + + i = t->labels_count++; + t->labels[i].branch_target = branch_target; + return &t->labels[i].token; +} + + +static void set_insn_start( struct st_translate *t, + unsigned start ) +{ + if (t->insn_count + 1 >= t->insn_size) { + unsigned old_size = t->insn_size; + t->insn_size = 1 << (util_logbase2(t->insn_size) + 1); + t->insn = REALLOC( t->insn, + old_size * sizeof t->insn[0], + t->insn_size * sizeof t->insn[0] ); + if (t->insn == NULL) { + t->error = TRUE; + return; + } + } + + t->insn[t->insn_count++] = start; +} + /* * Map mesa register file to TGSI register file. */ -static GLuint -map_register_file( - gl_register_file file, - GLuint index, - const GLuint immediateMapping[], - GLboolean indirectAccess ) +static struct ureg_dst +dst_register( struct st_translate *t, + gl_register_file file, + GLuint index ) { switch( file ) { case PROGRAM_UNDEFINED: - return TGSI_FILE_NULL; + return ureg_dst_undef(); + case PROGRAM_TEMPORARY: - return TGSI_FILE_TEMPORARY; - /*case PROGRAM_LOCAL_PARAM:*/ - /*case PROGRAM_ENV_PARAM:*/ - - /* Because of the longstanding problem with mesa arb shaders - * where constants, immediates and state variables are all - * bundled together as PROGRAM_STATE_VAR, we can't tell from the - * mesa register file whether this is a CONSTANT or an - * IMMEDIATE, hence we need all the other information. - */ - case PROGRAM_STATE_VAR: - case PROGRAM_NAMED_PARAM: - case PROGRAM_UNIFORM: - if (!indirectAccess && immediateMapping && immediateMapping[index] != ~0) - return TGSI_FILE_IMMEDIATE; - else - return TGSI_FILE_CONSTANT; - case PROGRAM_CONSTANT: - if (indirectAccess) - return TGSI_FILE_CONSTANT; - assert(immediateMapping[index] != ~0); - return TGSI_FILE_IMMEDIATE; - case PROGRAM_INPUT: - return TGSI_FILE_INPUT; + if (ureg_dst_is_undef(t->temps[index])) + t->temps[index] = ureg_DECL_temporary( t->ureg ); + + return t->temps[index]; + case PROGRAM_OUTPUT: - return TGSI_FILE_OUTPUT; + return t->outputs[t->outputMapping[index]]; + case PROGRAM_ADDRESS: - return TGSI_FILE_ADDRESS; + return t->address[index]; + default: assert( 0 ); - return TGSI_FILE_NULL; + return ureg_dst_undef(); } } -/** - * Map mesa register file index to TGSI index. - * Take special care when processing input and output indices. - * \param file one of TGSI_FILE_x - * \param index the mesa register file index - * \param inputMapping maps Mesa input indexes to TGSI input indexes - * \param outputMapping maps Mesa output indexes to TGSI output indexes - */ -static GLuint -map_register_file_index( - GLuint procType, - GLuint file, - GLuint index, - GLuint *swizzle, - const GLuint inputMapping[], - const GLuint outputMapping[], - const GLuint immediateMapping[], - GLboolean indirectAccess ) + +static struct ureg_src +src_register( struct st_translate *t, + gl_register_file file, + GLuint index ) { switch( file ) { - case TGSI_FILE_INPUT: - /* inputs are mapped according to the user-defined map */ - return inputMapping[index]; + case PROGRAM_UNDEFINED: + return ureg_src_undef(); - case TGSI_FILE_OUTPUT: - return outputMapping[index]; + case PROGRAM_TEMPORARY: + if (ureg_dst_is_undef(t->temps[index])) + t->temps[index] = ureg_DECL_temporary( t->ureg ); + return ureg_src(t->temps[index]); - case TGSI_FILE_IMMEDIATE: - if (indirectAccess) - return index; - assert(immediateMapping[index] != ~0); - return immediateMapping[index]; + case PROGRAM_STATE_VAR: + case PROGRAM_NAMED_PARAM: + case PROGRAM_UNIFORM: + case PROGRAM_CONSTANT: + return t->constants[index]; + + case PROGRAM_INPUT: + return t->inputs[t->inputMapping[index]]; + + case PROGRAM_OUTPUT: + return ureg_src(t->outputs[t->outputMapping[index]]); /* not needed? */ + + case PROGRAM_ADDRESS: + return ureg_src(t->address[index]); default: - return index; + assert( 0 ); + return ureg_src_undef(); } } -/* + +/** * Map mesa texture target to TGSI texture target. */ -static GLuint -map_texture_target( - GLuint textarget, - GLboolean shadow ) +static unsigned +translate_texture_target( GLuint textarget, + GLboolean shadow ) { - switch( textarget ) { - case TEXTURE_1D_INDEX: - if (shadow) - return TGSI_TEXTURE_SHADOW1D; - else - return TGSI_TEXTURE_1D; - case TEXTURE_2D_INDEX: - if (shadow) - return TGSI_TEXTURE_SHADOW2D; - else - return TGSI_TEXTURE_2D; - case TEXTURE_3D_INDEX: - return TGSI_TEXTURE_3D; - case TEXTURE_CUBE_INDEX: - return TGSI_TEXTURE_CUBE; - case TEXTURE_RECT_INDEX: - if (shadow) - return TGSI_TEXTURE_SHADOWRECT; - else - return TGSI_TEXTURE_RECT; - default: - assert( 0 ); + if (shadow) { + switch( textarget ) { + case TEXTURE_1D_INDEX: return TGSI_TEXTURE_SHADOW1D; + case TEXTURE_2D_INDEX: return TGSI_TEXTURE_SHADOW2D; + case TEXTURE_RECT_INDEX: return TGSI_TEXTURE_SHADOWRECT; + default: break; + } } - return TGSI_TEXTURE_1D; -} - -static GLuint -convert_sat( - GLuint sat ) -{ - switch( sat ) { - case SATURATE_OFF: - return TGSI_SAT_NONE; - case SATURATE_ZERO_ONE: - return TGSI_SAT_ZERO_ONE; - case SATURATE_PLUS_MINUS_ONE: - return TGSI_SAT_MINUS_PLUS_ONE; + switch( textarget ) { + case TEXTURE_1D_INDEX: return TGSI_TEXTURE_1D; + case TEXTURE_2D_INDEX: return TGSI_TEXTURE_2D; + case TEXTURE_3D_INDEX: return TGSI_TEXTURE_3D; + case TEXTURE_CUBE_INDEX: return TGSI_TEXTURE_CUBE; + case TEXTURE_RECT_INDEX: return TGSI_TEXTURE_RECT; default: assert( 0 ); - return TGSI_SAT_NONE; + return TGSI_TEXTURE_1D; } } -static GLuint -convert_writemask( - GLuint writemask ) + +static struct ureg_dst +translate_dst( struct st_translate *t, + const struct prog_dst_register *DstReg, + boolean saturate ) { - assert( WRITEMASK_X == TGSI_WRITEMASK_X ); - assert( WRITEMASK_Y == TGSI_WRITEMASK_Y ); - assert( WRITEMASK_Z == TGSI_WRITEMASK_Z ); - assert( WRITEMASK_W == TGSI_WRITEMASK_W ); - assert( (writemask & ~TGSI_WRITEMASK_XYZW) == 0 ); + struct ureg_dst dst = dst_register( t, + DstReg->File, + DstReg->Index ); - return writemask; + dst = ureg_writemask( dst, + DstReg->WriteMask ); + + if (saturate) + dst = ureg_saturate( dst ); + + if (DstReg->RelAddr) + dst = ureg_dst_indirect( dst, ureg_src(t->address[0]) ); + + return dst; } -static struct tgsi_full_immediate -make_immediate(const float *value, uint size) + +static struct ureg_src +translate_src( struct st_translate *t, + const struct prog_src_register *SrcReg ) { - struct tgsi_full_immediate imm; - unsigned i; + struct ureg_src src = src_register( t, SrcReg->File, SrcReg->Index ); + + src = ureg_swizzle( src, + GET_SWZ( SrcReg->Swizzle, 0 ) & 0x3, + GET_SWZ( SrcReg->Swizzle, 1 ) & 0x3, + GET_SWZ( SrcReg->Swizzle, 2 ) & 0x3, + GET_SWZ( SrcReg->Swizzle, 3 ) & 0x3); - imm = tgsi_default_full_immediate(); - imm.Immediate.NrTokens += size; - imm.Immediate.DataType = TGSI_IMM_FLOAT32; + if (SrcReg->Negate == NEGATE_XYZW) + src = ureg_negate(src); - for (i = 0; i < size; i++) - imm.u[i].Float = value[i]; + if (SrcReg->Abs) + src = ureg_abs(src); - return imm; + if (SrcReg->RelAddr) + src = ureg_src_indirect( src, ureg_src(t->address[0])); + + return src; } -static void -compile_instruction( - const struct prog_instruction *inst, - struct tgsi_full_instruction *fullinst, - const GLuint inputMapping[], - const GLuint outputMapping[], - const GLuint immediateMapping[], - GLboolean indirectAccess, - GLuint preamble_size, - GLuint procType, - GLboolean *insideSubroutine, - GLint wposTemp) + +static struct ureg_src swizzle_4v( struct ureg_src src, + const unsigned *swz ) { - GLuint i; - struct tgsi_full_dst_register *fulldst; - struct tgsi_full_src_register *fullsrc; - - *fullinst = tgsi_default_full_instruction(); - - fullinst->Instruction.Saturate = convert_sat( inst->SaturateMode ); - fullinst->Instruction.NumDstRegs = _mesa_num_inst_dst_regs( inst->Opcode ); - fullinst->Instruction.NumSrcRegs = _mesa_num_inst_src_regs( inst->Opcode ); - - fulldst = &fullinst->FullDstRegisters[0]; - fulldst->DstRegister.File = map_register_file( inst->DstReg.File, 0, NULL, GL_FALSE ); - fulldst->DstRegister.Index = map_register_file_index( - procType, - fulldst->DstRegister.File, - inst->DstReg.Index, - NULL, - inputMapping, - outputMapping, - NULL, - GL_FALSE ); - fulldst->DstRegister.WriteMask = convert_writemask( inst->DstReg.WriteMask ); - if (inst->DstReg.RelAddr) { - fulldst->DstRegister.Indirect = 1; - fulldst->DstRegisterInd.File = TGSI_FILE_ADDRESS; - fulldst->DstRegisterInd.Index = 0; + return ureg_swizzle( src, swz[0], swz[1], swz[2], swz[3] ); +} + + +/** + * Translate SWZ instructions into a single MAD. EG: + * + * SWZ dst, src.x-y10 + * + * becomes: + * + * MAD dst {1,-1,0,0}, src.xyxx, {0,0,1,0} + */ +static void emit_swz( struct st_translate *t, + struct ureg_dst dst, + const struct prog_src_register *SrcReg ) +{ + struct ureg_program *ureg = t->ureg; + struct ureg_src src = src_register( t, SrcReg->File, SrcReg->Index ); + + unsigned negate_mask = SrcReg->Negate; + + unsigned one_mask = ((GET_SWZ(SrcReg->Swizzle, 0) == SWIZZLE_ONE) << 0 | + (GET_SWZ(SrcReg->Swizzle, 1) == SWIZZLE_ONE) << 1 | + (GET_SWZ(SrcReg->Swizzle, 2) == SWIZZLE_ONE) << 2 | + (GET_SWZ(SrcReg->Swizzle, 3) == SWIZZLE_ONE) << 3); + + unsigned zero_mask = ((GET_SWZ(SrcReg->Swizzle, 0) == SWIZZLE_ZERO) << 0 | + (GET_SWZ(SrcReg->Swizzle, 1) == SWIZZLE_ZERO) << 1 | + (GET_SWZ(SrcReg->Swizzle, 2) == SWIZZLE_ZERO) << 2 | + (GET_SWZ(SrcReg->Swizzle, 3) == SWIZZLE_ZERO) << 3); + + unsigned negative_one_mask = one_mask & negate_mask; + unsigned positive_one_mask = one_mask & ~negate_mask; + + struct ureg_src imm; + unsigned i; + unsigned mul_swizzle[4] = {0,0,0,0}; + unsigned add_swizzle[4] = {0,0,0,0}; + unsigned src_swizzle[4] = {0,0,0,0}; + boolean need_add = FALSE; + boolean need_mul = FALSE; + + if (dst.WriteMask == 0) + return; + + /* Is this just a MOV? + */ + if (zero_mask == 0 && + one_mask == 0 && + (negate_mask == 0 || negate_mask == TGSI_WRITEMASK_XYZW)) + { + ureg_MOV( ureg, dst, translate_src( t, SrcReg )); + return; } - for (i = 0; i < fullinst->Instruction.NumSrcRegs; i++) { - GLuint j; - GLuint swizzle = inst->SrcReg[i].Swizzle; +#define IMM_ZERO 0 +#define IMM_ONE 1 +#define IMM_NEG_ONE 2 - fullsrc = &fullinst->FullSrcRegisters[i]; + imm = ureg_imm3f( ureg, 0, 1, -1 ); - if (procType == TGSI_PROCESSOR_FRAGMENT && - inst->SrcReg[i].File == PROGRAM_INPUT && - inst->SrcReg[i].Index == FRAG_ATTRIB_WPOS) { - /* special case of INPUT[WPOS] */ - fullsrc->SrcRegister.File = TGSI_FILE_TEMPORARY; - fullsrc->SrcRegister.Index = wposTemp; - } - else { - /* any other src register */ - fullsrc->SrcRegister.File = map_register_file( - inst->SrcReg[i].File, - inst->SrcReg[i].Index, - immediateMapping, - indirectAccess ); - fullsrc->SrcRegister.Index = map_register_file_index( - procType, - fullsrc->SrcRegister.File, - inst->SrcReg[i].Index, - &swizzle, - inputMapping, - outputMapping, - immediateMapping, - indirectAccess ); - } + for (i = 0; i < 4; i++) { + unsigned bit = 1 << i; - /* swizzle (ext swizzle also depends on negation) */ - { - GLuint swz[4]; - GLboolean extended = (inst->SrcReg[i].Negate != NEGATE_NONE && - inst->SrcReg[i].Negate != NEGATE_XYZW); - for( j = 0; j < 4; j++ ) { - swz[j] = GET_SWZ( swizzle, j ); - if (swz[j] > SWIZZLE_W) - extended = GL_TRUE; + if (dst.WriteMask & bit) { + if (positive_one_mask & bit) { + mul_swizzle[i] = IMM_ZERO; + add_swizzle[i] = IMM_ONE; + need_add = TRUE; } - if (extended) { - for (j = 0; j < 4; j++) { - tgsi_util_set_src_register_extswizzle(&fullsrc->SrcRegisterExtSwz, - swz[j], j); - } + else if (negative_one_mask & bit) { + mul_swizzle[i] = IMM_ZERO; + add_swizzle[i] = IMM_NEG_ONE; + need_add = TRUE; + } + else if (zero_mask & bit) { + mul_swizzle[i] = IMM_ZERO; + add_swizzle[i] = IMM_ZERO; + need_add = TRUE; } else { - for (j = 0; j < 4; j++) { - tgsi_util_set_src_register_swizzle(&fullsrc->SrcRegister, - swz[j], j); + add_swizzle[i] = IMM_ZERO; + src_swizzle[i] = GET_SWZ(SrcReg->Swizzle, i); + need_mul = TRUE; + if (negate_mask & bit) { + mul_swizzle[i] = IMM_NEG_ONE; + } + else { + mul_swizzle[i] = IMM_ONE; } } } + } - if( inst->SrcReg[i].Negate == NEGATE_XYZW ) { - fullsrc->SrcRegister.Negate = 1; - } - else if( inst->SrcReg[i].Negate != NEGATE_NONE ) { - if( inst->SrcReg[i].Negate & NEGATE_X ) { - fullsrc->SrcRegisterExtSwz.NegateX = 1; - } - if( inst->SrcReg[i].Negate & NEGATE_Y ) { - fullsrc->SrcRegisterExtSwz.NegateY = 1; - } - if( inst->SrcReg[i].Negate & NEGATE_Z ) { - fullsrc->SrcRegisterExtSwz.NegateZ = 1; - } - if( inst->SrcReg[i].Negate & NEGATE_W ) { - fullsrc->SrcRegisterExtSwz.NegateW = 1; - } - } + if (need_mul && need_add) { + ureg_MAD( ureg, + dst, + swizzle_4v( src, src_swizzle ), + swizzle_4v( imm, mul_swizzle ), + swizzle_4v( imm, add_swizzle ) ); + } + else if (need_mul) { + ureg_MUL( ureg, + dst, + swizzle_4v( src, src_swizzle ), + swizzle_4v( imm, mul_swizzle ) ); + } + else if (need_add) { + ureg_MOV( ureg, + dst, + swizzle_4v( imm, add_swizzle ) ); + } + else { + assert(0); + } - if( inst->SrcReg[i].Abs ) { - fullsrc->SrcRegisterExtMod.Absolute = 1; - } +#undef IMM_ZERO +#undef IMM_ONE +#undef IMM_NEG_ONE +} - if( inst->SrcReg[i].RelAddr ) { - fullsrc->SrcRegister.Indirect = 1; - fullsrc->SrcRegisterInd.File = TGSI_FILE_ADDRESS; - fullsrc->SrcRegisterInd.Index = 0; - } - } - switch( inst->Opcode ) { +static unsigned +translate_opcode( unsigned op ) +{ + switch( op ) { case OPCODE_ARL: - fullinst->Instruction.Opcode = TGSI_OPCODE_ARL; - break; + return TGSI_OPCODE_ARL; case OPCODE_ABS: - fullinst->Instruction.Opcode = TGSI_OPCODE_ABS; - break; + return TGSI_OPCODE_ABS; case OPCODE_ADD: - fullinst->Instruction.Opcode = TGSI_OPCODE_ADD; - break; + return TGSI_OPCODE_ADD; case OPCODE_BGNLOOP: - fullinst->Instruction.Opcode = TGSI_OPCODE_BGNLOOP; - fullinst->InstructionExtLabel.Label = inst->BranchTarget + preamble_size; - break; + return TGSI_OPCODE_BGNLOOP; case OPCODE_BGNSUB: - fullinst->Instruction.Opcode = TGSI_OPCODE_BGNSUB; - *insideSubroutine = GL_TRUE; - break; + return TGSI_OPCODE_BGNSUB; case OPCODE_BRA: - fullinst->Instruction.Opcode = TGSI_OPCODE_BRA; - break; + return TGSI_OPCODE_BRA; case OPCODE_BRK: - fullinst->Instruction.Opcode = TGSI_OPCODE_BRK; - break; + return TGSI_OPCODE_BRK; case OPCODE_CAL: - fullinst->Instruction.Opcode = TGSI_OPCODE_CAL; - fullinst->InstructionExtLabel.Label = inst->BranchTarget + preamble_size; - break; + return TGSI_OPCODE_CAL; case OPCODE_CMP: - fullinst->Instruction.Opcode = TGSI_OPCODE_CMP; - break; + return TGSI_OPCODE_CMP; case OPCODE_CONT: - fullinst->Instruction.Opcode = TGSI_OPCODE_CONT; - break; + return TGSI_OPCODE_CONT; case OPCODE_COS: - fullinst->Instruction.Opcode = TGSI_OPCODE_COS; - break; + return TGSI_OPCODE_COS; case OPCODE_DDX: - fullinst->Instruction.Opcode = TGSI_OPCODE_DDX; - break; + return TGSI_OPCODE_DDX; case OPCODE_DDY: - fullinst->Instruction.Opcode = TGSI_OPCODE_DDY; - break; + return TGSI_OPCODE_DDY; case OPCODE_DP2: - fullinst->Instruction.Opcode = TGSI_OPCODE_DP2; - break; + return TGSI_OPCODE_DP2; case OPCODE_DP2A: - fullinst->Instruction.Opcode = TGSI_OPCODE_DP2A; - break; + return TGSI_OPCODE_DP2A; case OPCODE_DP3: - fullinst->Instruction.Opcode = TGSI_OPCODE_DP3; - break; + return TGSI_OPCODE_DP3; case OPCODE_DP4: - fullinst->Instruction.Opcode = TGSI_OPCODE_DP4; - break; + return TGSI_OPCODE_DP4; case OPCODE_DPH: - fullinst->Instruction.Opcode = TGSI_OPCODE_DPH; - break; + return TGSI_OPCODE_DPH; case OPCODE_DST: - fullinst->Instruction.Opcode = TGSI_OPCODE_DST; - break; + return TGSI_OPCODE_DST; case OPCODE_ELSE: - fullinst->Instruction.Opcode = TGSI_OPCODE_ELSE; - fullinst->InstructionExtLabel.Label = inst->BranchTarget + preamble_size; - break; + return TGSI_OPCODE_ELSE; case OPCODE_ENDIF: - fullinst->Instruction.Opcode = TGSI_OPCODE_ENDIF; - break; + return TGSI_OPCODE_ENDIF; case OPCODE_ENDLOOP: - fullinst->Instruction.Opcode = TGSI_OPCODE_ENDLOOP; - fullinst->InstructionExtLabel.Label = inst->BranchTarget + preamble_size; - break; + return TGSI_OPCODE_ENDLOOP; case OPCODE_ENDSUB: - fullinst->Instruction.Opcode = TGSI_OPCODE_ENDSUB; - *insideSubroutine = GL_FALSE; - break; + return TGSI_OPCODE_ENDSUB; case OPCODE_EX2: - fullinst->Instruction.Opcode = TGSI_OPCODE_EX2; - break; + return TGSI_OPCODE_EX2; case OPCODE_EXP: - fullinst->Instruction.Opcode = TGSI_OPCODE_EXP; - break; + return TGSI_OPCODE_EXP; case OPCODE_FLR: - fullinst->Instruction.Opcode = TGSI_OPCODE_FLR; - break; + return TGSI_OPCODE_FLR; case OPCODE_FRC: - fullinst->Instruction.Opcode = TGSI_OPCODE_FRC; - break; + return TGSI_OPCODE_FRC; case OPCODE_IF: - fullinst->Instruction.Opcode = TGSI_OPCODE_IF; - fullinst->InstructionExtLabel.Label = inst->BranchTarget + preamble_size; - break; + return TGSI_OPCODE_IF; case OPCODE_TRUNC: - fullinst->Instruction.Opcode = TGSI_OPCODE_TRUNC; - break; + return TGSI_OPCODE_TRUNC; case OPCODE_KIL: - /* conditional */ - fullinst->Instruction.Opcode = TGSI_OPCODE_KIL; - break; + return TGSI_OPCODE_KIL; case OPCODE_KIL_NV: - /* predicated */ - assert(inst->DstReg.CondMask == COND_TR); - fullinst->Instruction.Opcode = TGSI_OPCODE_KILP; - break; + return TGSI_OPCODE_KILP; case OPCODE_LG2: - fullinst->Instruction.Opcode = TGSI_OPCODE_LG2; - break; + return TGSI_OPCODE_LG2; case OPCODE_LOG: - fullinst->Instruction.Opcode = TGSI_OPCODE_LOG; - break; + return TGSI_OPCODE_LOG; case OPCODE_LIT: - fullinst->Instruction.Opcode = TGSI_OPCODE_LIT; - break; + return TGSI_OPCODE_LIT; case OPCODE_LRP: - fullinst->Instruction.Opcode = TGSI_OPCODE_LRP; - break; + return TGSI_OPCODE_LRP; case OPCODE_MAD: - fullinst->Instruction.Opcode = TGSI_OPCODE_MAD; - break; + return TGSI_OPCODE_MAD; case OPCODE_MAX: - fullinst->Instruction.Opcode = TGSI_OPCODE_MAX; - break; + return TGSI_OPCODE_MAX; case OPCODE_MIN: - fullinst->Instruction.Opcode = TGSI_OPCODE_MIN; - break; + return TGSI_OPCODE_MIN; case OPCODE_MOV: - fullinst->Instruction.Opcode = TGSI_OPCODE_MOV; - break; + return TGSI_OPCODE_MOV; case OPCODE_MUL: - fullinst->Instruction.Opcode = TGSI_OPCODE_MUL; - break; + return TGSI_OPCODE_MUL; case OPCODE_NOISE1: - fullinst->Instruction.Opcode = TGSI_OPCODE_NOISE1; - break; + return TGSI_OPCODE_NOISE1; case OPCODE_NOISE2: - fullinst->Instruction.Opcode = TGSI_OPCODE_NOISE2; - break; + return TGSI_OPCODE_NOISE2; case OPCODE_NOISE3: - fullinst->Instruction.Opcode = TGSI_OPCODE_NOISE3; - break; + return TGSI_OPCODE_NOISE3; case OPCODE_NOISE4: - fullinst->Instruction.Opcode = TGSI_OPCODE_NOISE4; - break; + return TGSI_OPCODE_NOISE4; case OPCODE_NOP: - fullinst->Instruction.Opcode = TGSI_OPCODE_NOP; - break; + return TGSI_OPCODE_NOP; case OPCODE_NRM3: - fullinst->Instruction.Opcode = TGSI_OPCODE_NRM; - break; + return TGSI_OPCODE_NRM; case OPCODE_NRM4: - fullinst->Instruction.Opcode = TGSI_OPCODE_NRM4; - break; + return TGSI_OPCODE_NRM4; case OPCODE_POW: - fullinst->Instruction.Opcode = TGSI_OPCODE_POW; - break; + return TGSI_OPCODE_POW; case OPCODE_RCP: - fullinst->Instruction.Opcode = TGSI_OPCODE_RCP; - break; + return TGSI_OPCODE_RCP; case OPCODE_RET: - /* If RET is used inside main (not a real subroutine) we may want - * to execute END instead of RET. TBD... - */ - if (1 /* *insideSubroutine */) { - fullinst->Instruction.Opcode = TGSI_OPCODE_RET; - } - else { - /* inside main() pseudo-function */ - fullinst->Instruction.Opcode = TGSI_OPCODE_END; - } - break; + return TGSI_OPCODE_RET; case OPCODE_RSQ: - fullinst->Instruction.Opcode = TGSI_OPCODE_RSQ; - break; + return TGSI_OPCODE_RSQ; case OPCODE_SCS: - fullinst->Instruction.Opcode = TGSI_OPCODE_SCS; - fulldst->DstRegister.WriteMask &= TGSI_WRITEMASK_XY; - break; + return TGSI_OPCODE_SCS; case OPCODE_SEQ: - fullinst->Instruction.Opcode = TGSI_OPCODE_SEQ; - break; + return TGSI_OPCODE_SEQ; case OPCODE_SGE: - fullinst->Instruction.Opcode = TGSI_OPCODE_SGE; - break; + return TGSI_OPCODE_SGE; case OPCODE_SGT: - fullinst->Instruction.Opcode = TGSI_OPCODE_SGT; - break; + return TGSI_OPCODE_SGT; case OPCODE_SIN: - fullinst->Instruction.Opcode = TGSI_OPCODE_SIN; - break; + return TGSI_OPCODE_SIN; case OPCODE_SLE: - fullinst->Instruction.Opcode = TGSI_OPCODE_SLE; - break; + return TGSI_OPCODE_SLE; case OPCODE_SLT: - fullinst->Instruction.Opcode = TGSI_OPCODE_SLT; - break; + return TGSI_OPCODE_SLT; case OPCODE_SNE: - fullinst->Instruction.Opcode = TGSI_OPCODE_SNE; - break; + return TGSI_OPCODE_SNE; case OPCODE_SSG: - fullinst->Instruction.Opcode = TGSI_OPCODE_SSG; - break; + return TGSI_OPCODE_SSG; case OPCODE_SUB: - fullinst->Instruction.Opcode = TGSI_OPCODE_SUB; - break; + return TGSI_OPCODE_SUB; case OPCODE_SWZ: - fullinst->Instruction.Opcode = TGSI_OPCODE_SWZ; - break; + return TGSI_OPCODE_SWZ; case OPCODE_TEX: - /* ordinary texture lookup */ - fullinst->Instruction.Opcode = TGSI_OPCODE_TEX; - fullinst->Instruction.NumSrcRegs = 2; - fullinst->InstructionExtTexture.Texture = - map_texture_target( inst->TexSrcTarget, inst->TexShadow ); - fullinst->FullSrcRegisters[1].SrcRegister.File = TGSI_FILE_SAMPLER; - fullinst->FullSrcRegisters[1].SrcRegister.Index = inst->TexSrcUnit; - break; + return TGSI_OPCODE_TEX; case OPCODE_TXB: - /* texture lookup with LOD bias */ - fullinst->Instruction.Opcode = TGSI_OPCODE_TXB; - fullinst->Instruction.NumSrcRegs = 2; - fullinst->InstructionExtTexture.Texture = - map_texture_target( inst->TexSrcTarget, inst->TexShadow ); - fullinst->FullSrcRegisters[1].SrcRegister.File = TGSI_FILE_SAMPLER; - fullinst->FullSrcRegisters[1].SrcRegister.Index = inst->TexSrcUnit; - break; + return TGSI_OPCODE_TXB; case OPCODE_TXD: - /* texture lookup with explicit partial derivatives */ - fullinst->Instruction.Opcode = TGSI_OPCODE_TXD; - fullinst->Instruction.NumSrcRegs = 4; - fullinst->InstructionExtTexture.Texture = - map_texture_target( inst->TexSrcTarget, inst->TexShadow ); - /* src[0] = coord, src[1] = d[strq]/dx, src[2] = d[strq]/dy */ - fullinst->FullSrcRegisters[3].SrcRegister.File = TGSI_FILE_SAMPLER; - fullinst->FullSrcRegisters[3].SrcRegister.Index = inst->TexSrcUnit; - break; + return TGSI_OPCODE_TXD; case OPCODE_TXL: - /* texture lookup with explicit LOD */ - fullinst->Instruction.Opcode = TGSI_OPCODE_TXL; - fullinst->Instruction.NumSrcRegs = 2; - fullinst->InstructionExtTexture.Texture = - map_texture_target( inst->TexSrcTarget, inst->TexShadow ); - fullinst->FullSrcRegisters[1].SrcRegister.File = TGSI_FILE_SAMPLER; - fullinst->FullSrcRegisters[1].SrcRegister.Index = inst->TexSrcUnit; - break; + return TGSI_OPCODE_TXL; case OPCODE_TXP: - /* texture lookup with divide by Q component */ - /* convert to TEX w/ special flag for division */ - fullinst->Instruction.Opcode = TGSI_OPCODE_TXP; - fullinst->Instruction.NumSrcRegs = 2; - fullinst->InstructionExtTexture.Texture = - map_texture_target( inst->TexSrcTarget, inst->TexShadow ); - fullinst->FullSrcRegisters[1].SrcRegister.File = TGSI_FILE_SAMPLER; - fullinst->FullSrcRegisters[1].SrcRegister.Index = inst->TexSrcUnit; - break; + return TGSI_OPCODE_TXP; case OPCODE_XPD: - fullinst->Instruction.Opcode = TGSI_OPCODE_XPD; - fulldst->DstRegister.WriteMask &= TGSI_WRITEMASK_XYZ; - break; + return TGSI_OPCODE_XPD; case OPCODE_END: - fullinst->Instruction.Opcode = TGSI_OPCODE_END; - break; + return TGSI_OPCODE_END; default: assert( 0 ); + return TGSI_OPCODE_NOP; } } -/** - * \param usage_mask bitfield of TGSI_WRITEMASK_{XYZW} tokens - */ -static struct tgsi_full_declaration -make_input_decl( - GLuint index, - GLboolean interpolate_info, - GLuint interpolate, - GLuint usage_mask, - GLboolean semantic_info, - GLuint semantic_name, - GLbitfield semantic_index, - GLbitfield input_flags) -{ - struct tgsi_full_declaration decl; - - assert(semantic_name < TGSI_SEMANTIC_COUNT); - - decl = tgsi_default_full_declaration(); - decl.Declaration.File = TGSI_FILE_INPUT; - decl.Declaration.UsageMask = usage_mask; - decl.Declaration.Semantic = semantic_info; - decl.DeclarationRange.First = index; - decl.DeclarationRange.Last = index; - if (semantic_info) { - decl.Semantic.SemanticName = semantic_name; - decl.Semantic.SemanticIndex = semantic_index; - } - if (interpolate_info) { - decl.Declaration.Interpolate = interpolate; - } - if (input_flags & PROG_PARAM_BIT_CENTROID) - decl.Declaration.Centroid = 1; - if (input_flags & PROG_PARAM_BIT_INVARIANT) - decl.Declaration.Invariant = 1; - - return decl; -} -/** - * \param usage_mask bitfield of TGSI_WRITEMASK_{XYZW} tokens - */ -static struct tgsi_full_declaration -make_output_decl( - GLuint index, - GLuint semantic_name, - GLuint semantic_index, - GLuint usage_mask, - GLbitfield output_flags) +static void +compile_instruction( + struct st_translate *t, + const struct prog_instruction *inst ) { - struct tgsi_full_declaration decl; - - assert(semantic_name < TGSI_SEMANTIC_COUNT); - - decl = tgsi_default_full_declaration(); - decl.Declaration.File = TGSI_FILE_OUTPUT; - decl.Declaration.UsageMask = usage_mask; - decl.Declaration.Semantic = 1; - decl.DeclarationRange.First = index; - decl.DeclarationRange.Last = index; - decl.Semantic.SemanticName = semantic_name; - decl.Semantic.SemanticIndex = semantic_index; - if (output_flags & PROG_PARAM_BIT_CENTROID) - decl.Declaration.Centroid = 1; - if (output_flags & PROG_PARAM_BIT_INVARIANT) - decl.Declaration.Invariant = 1; - - return decl; -} - + struct ureg_program *ureg = t->ureg; + GLuint i; + struct ureg_dst dst[1]; + struct ureg_src src[4]; + unsigned num_dst; + unsigned num_src; -static struct tgsi_full_declaration -make_temp_decl( - GLuint start_index, - GLuint end_index ) -{ - struct tgsi_full_declaration decl; - decl = tgsi_default_full_declaration(); - decl.Declaration.File = TGSI_FILE_TEMPORARY; - decl.DeclarationRange.First = start_index; - decl.DeclarationRange.Last = end_index; - return decl; -} + num_dst = _mesa_num_inst_dst_regs( inst->Opcode ); + num_src = _mesa_num_inst_src_regs( inst->Opcode ); -static struct tgsi_full_declaration -make_addr_decl( - GLuint start_index, - GLuint end_index ) -{ - struct tgsi_full_declaration decl; + if (num_dst) + dst[0] = translate_dst( t, + &inst->DstReg, + inst->SaturateMode ); - decl = tgsi_default_full_declaration(); - decl.Declaration.File = TGSI_FILE_ADDRESS; - decl.DeclarationRange.First = start_index; - decl.DeclarationRange.Last = end_index; - return decl; -} + for (i = 0; i < num_src; i++) + src[i] = translate_src( t, &inst->SrcReg[i] ); -static struct tgsi_full_declaration -make_sampler_decl(GLuint index) -{ - struct tgsi_full_declaration decl; - decl = tgsi_default_full_declaration(); - decl.Declaration.File = TGSI_FILE_SAMPLER; - decl.DeclarationRange.First = index; - decl.DeclarationRange.Last = index; - return decl; -} - -/** Reference into a constant buffer */ -static struct tgsi_full_declaration -make_constant_decl(GLuint first, GLuint last) -{ - struct tgsi_full_declaration decl; - decl = tgsi_default_full_declaration(); - decl.Declaration.File = TGSI_FILE_CONSTANT; - decl.DeclarationRange.First = first; - decl.DeclarationRange.Last = last; - return decl; -} + switch( inst->Opcode ) { + case OPCODE_SWZ: + emit_swz( t, dst[0], &inst->SrcReg[0] ); + return; + case OPCODE_BGNLOOP: + case OPCODE_CAL: + case OPCODE_ELSE: + case OPCODE_ENDLOOP: + case OPCODE_IF: + assert(num_dst == 0); + ureg_label_insn( ureg, + translate_opcode( inst->Opcode ), + src, num_src, + get_label( t, inst->BranchTarget )); + return; + case OPCODE_TEX: + case OPCODE_TXB: + case OPCODE_TXD: + case OPCODE_TXL: + case OPCODE_TXP: + src[num_src++] = t->samplers[inst->TexSrcUnit]; + ureg_tex_insn( ureg, + translate_opcode( inst->Opcode ), + dst, num_dst, + translate_texture_target( inst->TexSrcTarget, + inst->TexShadow ), + src, num_src ); + return; -/** - * Find the temporaries which are used in the given program. - */ -static void -find_temporaries(const struct gl_program *program, - GLboolean tempsUsed[MAX_PROGRAM_TEMPS]) -{ - GLuint i, j; + case OPCODE_SCS: + dst[0] = ureg_writemask(dst[0], TGSI_WRITEMASK_XY ); + ureg_insn( ureg, + translate_opcode( inst->Opcode ), + dst, num_dst, + src, num_src ); + break; - for (i = 0; i < MAX_PROGRAM_TEMPS; i++) - tempsUsed[i] = GL_FALSE; + case OPCODE_XPD: + dst[0] = ureg_writemask(dst[0], TGSI_WRITEMASK_XYZ ); + ureg_insn( ureg, + translate_opcode( inst->Opcode ), + dst, num_dst, + src, num_src ); + break; - for (i = 0; i < program->NumInstructions; i++) { - const struct prog_instruction *inst = program->Instructions + i; - const GLuint n = _mesa_num_inst_src_regs( inst->Opcode ); - for (j = 0; j < n; j++) { - if (inst->SrcReg[j].File == PROGRAM_TEMPORARY) - tempsUsed[inst->SrcReg[j].Index] = GL_TRUE; - if (inst->DstReg.File == PROGRAM_TEMPORARY) - tempsUsed[inst->DstReg.Index] = GL_TRUE; - } + default: + ureg_insn( ureg, + translate_opcode( inst->Opcode ), + dst, num_dst, + src, num_src ); + break; } } /** - * Find an unused temporary in the tempsUsed array. + * Emit the TGSI instructions for inverting the WPOS y coordinate. */ -static int -find_free_temporary(GLboolean tempsUsed[MAX_PROGRAM_TEMPS]) -{ - int i; - for (i = 0; i < MAX_PROGRAM_TEMPS; i++) { - if (!tempsUsed[i]) { - tempsUsed[i] = GL_TRUE; - return i; - } - } - return -1; -} - - -/** helper for building simple TGSI instruction, one src register */ -static void -build_tgsi_instruction1(struct tgsi_full_instruction *inst, - int opcode, - int dstFile, int dstIndex, int writemask, - int srcFile1, int srcIndex1) -{ - *inst = tgsi_default_full_instruction(); - - inst->Instruction.Opcode = opcode; - - inst->Instruction.NumDstRegs = 1; - inst->FullDstRegisters[0].DstRegister.File = dstFile; - inst->FullDstRegisters[0].DstRegister.Index = dstIndex; - inst->FullDstRegisters[0].DstRegister.WriteMask = writemask; - - inst->Instruction.NumSrcRegs = 1; - inst->FullSrcRegisters[0].SrcRegister.File = srcFile1; - inst->FullSrcRegisters[0].SrcRegister.Index = srcIndex1; -} - - -/** helper for building simple TGSI instruction, two src registers */ static void -build_tgsi_instruction2(struct tgsi_full_instruction *inst, - int opcode, - int dstFile, int dstIndex, int writemask, - int srcFile1, int srcIndex1, - int srcFile2, int srcIndex2) +emit_inverted_wpos( struct st_translate *t, + const struct gl_program *program ) { - *inst = tgsi_default_full_instruction(); - - inst->Instruction.Opcode = opcode; + struct ureg_program *ureg = t->ureg; - inst->Instruction.NumDstRegs = 1; - inst->FullDstRegisters[0].DstRegister.File = dstFile; - inst->FullDstRegisters[0].DstRegister.Index = dstIndex; - inst->FullDstRegisters[0].DstRegister.WriteMask = writemask; + /* Fragment program uses fragment position input. + * Need to replace instances of INPUT[WPOS] with temp T + * where T = INPUT[WPOS] by y is inverted. + */ + static const gl_state_index winSizeState[STATE_LENGTH] + = { STATE_INTERNAL, STATE_FB_SIZE, 0, 0, 0 }; + + /* XXX: note we are modifying the incoming shader here! Need to + * do this before emitting the constant decls below, or this + * will be missed: + */ + unsigned winHeightConst = _mesa_add_state_reference(program->Parameters, + winSizeState); - inst->Instruction.NumSrcRegs = 2; - inst->FullSrcRegisters[0].SrcRegister.File = srcFile1; - inst->FullSrcRegisters[0].SrcRegister.Index = srcIndex1; - inst->FullSrcRegisters[1].SrcRegister.File = srcFile2; - inst->FullSrcRegisters[1].SrcRegister.Index = srcIndex2; -} + struct ureg_src winsize = ureg_DECL_constant( ureg, winHeightConst ); + struct ureg_dst wpos_temp = ureg_DECL_temporary( ureg ); + struct ureg_src wpos_input = t->inputs[t->inputMapping[FRAG_ATTRIB_WPOS]]; + /* MOV wpos_temp, input[wpos] + */ + ureg_MOV( ureg, wpos_temp, wpos_input ); + /* SUB wpos_temp.y, winsize_const, wpos_input + */ + ureg_SUB( ureg, + ureg_writemask(wpos_temp, TGSI_WRITEMASK_Y ), + winsize, + wpos_input); -/** - * Emit the TGSI instructions for inverting the WPOS y coordinate. - */ -static int -emit_inverted_wpos(struct tgsi_token *tokens, - int wpos_temp, - int winsize_const, - int wpos_input, - struct tgsi_header *header, int maxTokens) -{ - struct tgsi_full_instruction fullinst; - int ti = 0; - - /* MOV wpos_temp.xzw, input[wpos]; */ - build_tgsi_instruction1(&fullinst, - TGSI_OPCODE_MOV, - TGSI_FILE_TEMPORARY, wpos_temp, WRITEMASK_XZW, - TGSI_FILE_INPUT, 0); - - ti += tgsi_build_full_instruction(&fullinst, - &tokens[ti], - header, - maxTokens - ti); - - /* SUB wpos_temp.y, const[winsize_const] - input[wpos_input]; */ - build_tgsi_instruction2(&fullinst, - TGSI_OPCODE_SUB, - TGSI_FILE_TEMPORARY, wpos_temp, WRITEMASK_Y, - TGSI_FILE_CONSTANT, winsize_const, - TGSI_FILE_INPUT, wpos_input); - - ti += tgsi_build_full_instruction(&fullinst, - &tokens[ti], - header, - maxTokens - ti); - - return ti; + /* Use wpos_temp as position input from here on: + */ + t->inputs[t->inputMapping[FRAG_ATTRIB_WPOS]] = ureg_src(wpos_temp); } - - /** * Translate Mesa program to TGSI format. * \param program the program to translate @@ -864,20 +676,19 @@ emit_inverted_wpos(struct tgsi_token *tokens, * \param inputMapping maps Mesa fragment program inputs to TGSI generic * input indexes * \param inputSemanticName the TGSI_SEMANTIC flag for each input - * \param inputSemanticIndex the semantic index (ex: which texcoord) for each input + * \param inputSemanticIndex the semantic index (ex: which texcoord) for + * each input * \param interpMode the TGSI_INTERPOLATE_LINEAR/PERSP mode for each input - * \param numOutputs number of output registers used * \param outputMapping maps Mesa fragment program outputs to TGSI * generic outputs * \param outputSemanticName the TGSI_SEMANTIC flag for each output - * \param outputSemanticIndex the semantic index (ex: which texcoord) for each output - * \param tokens array to store translated tokens in - * \param maxTokens size of the tokens array + * \param outputSemanticIndex the semantic index (ex: which texcoord) for + * each output * - * \return number of tokens placed in 'tokens' buffer, or zero if error + * \return array of translated tokens, caller's responsibility to free */ -GLuint +const struct tgsi_token * st_translate_mesa_program( GLcontext *ctx, uint procType, @@ -892,252 +703,124 @@ st_translate_mesa_program( const GLuint outputMapping[], const ubyte outputSemanticName[], const ubyte outputSemanticIndex[], - const GLbitfield outputFlags[], - struct tgsi_token *tokens, - GLuint maxTokens ) + const GLbitfield outputFlags[] ) { - GLuint i; - GLuint ti; /* token index */ - struct tgsi_header *header; - struct tgsi_processor *processor; - GLuint preamble_size = 0; - GLuint immediates[1000]; - GLuint numImmediates = 0; - GLboolean insideSubroutine = GL_FALSE; - GLboolean indirectAccess = GL_FALSE; - GLboolean tempsUsed[MAX_PROGRAM_TEMPS + 1]; - GLint wposTemp = -1, winHeightConst = -1; - - assert(procType == TGSI_PROCESSOR_FRAGMENT || - procType == TGSI_PROCESSOR_VERTEX); - - find_temporaries(program, tempsUsed); - - if (procType == TGSI_PROCESSOR_FRAGMENT) { - if (program->InputsRead & FRAG_BIT_WPOS) { - /* Fragment program uses fragment position input. - * Need to replace instances of INPUT[WPOS] with temp T - * where T = INPUT[WPOS] by y is inverted. - */ - static const gl_state_index winSizeState[STATE_LENGTH] - = { STATE_INTERNAL, STATE_FB_SIZE, 0, 0, 0 }; - winHeightConst = _mesa_add_state_reference(program->Parameters, - winSizeState); - wposTemp = find_free_temporary(tempsUsed); - } - } - + struct st_translate translate, *t; + struct ureg_program *ureg; + const struct tgsi_token *tokens = NULL; + unsigned i; - *(struct tgsi_version *) &tokens[0] = tgsi_build_version(); + t = &translate; + memset(t, 0, sizeof *t); - header = (struct tgsi_header *) &tokens[1]; - *header = tgsi_build_header(); + t->procType = procType; + t->inputMapping = inputMapping; + t->outputMapping = outputMapping; + t->ureg = ureg_create( procType ); + if (t->ureg == NULL) + return NULL; - processor = (struct tgsi_processor *) &tokens[2]; - *processor = tgsi_build_processor( procType, header ); + ureg = t->ureg; - ti = 3; + /*_mesa_print_program(program);*/ /* * Declare input attributes. */ if (procType == TGSI_PROCESSOR_FRAGMENT) { for (i = 0; i < numInputs; i++) { - struct tgsi_full_declaration fulldecl; - fulldecl = make_input_decl(i, - GL_TRUE, interpMode[i], - TGSI_WRITEMASK_XYZW, - GL_TRUE, inputSemanticName[i], - inputSemanticIndex[i], - inputFlags[i]); - ti += tgsi_build_full_declaration(&fulldecl, - &tokens[ti], - header, - maxTokens - ti ); + t->inputs[i] = ureg_DECL_fs_input(ureg, + inputSemanticName[i], + inputSemanticIndex[i], + interpMode[i]); } - } - else { - /* vertex prog */ - /* XXX: this could probaby be merged with the clause above. - * the only difference is the semantic tags. - */ - for (i = 0; i < numInputs; i++) { - struct tgsi_full_declaration fulldecl; - fulldecl = make_input_decl(i, - GL_FALSE, 0, - TGSI_WRITEMASK_XYZW, - GL_FALSE, 0, 0, - inputFlags[i]); - ti += tgsi_build_full_declaration(&fulldecl, - &tokens[ti], - header, - maxTokens - ti ); + + if (program->InputsRead & FRAG_BIT_WPOS) { + /* Must do this after setting up t->inputs, and before + * emitting constant references, below: + */ + emit_inverted_wpos( t, program ); } - } - /* - * Declare output attributes. - */ - if (procType == TGSI_PROCESSOR_FRAGMENT) { + /* + * Declare output attributes. + */ for (i = 0; i < numOutputs; i++) { - struct tgsi_full_declaration fulldecl; switch (outputSemanticName[i]) { case TGSI_SEMANTIC_POSITION: - fulldecl = make_output_decl(i, - TGSI_SEMANTIC_POSITION, /* Z / Depth */ - outputSemanticIndex[i], - TGSI_WRITEMASK_Z, - outputFlags[i]); + t->outputs[i] = ureg_DECL_output( ureg, + TGSI_SEMANTIC_POSITION, /* Z / Depth */ + outputSemanticIndex[i] ); + + t->outputs[i] = ureg_writemask( t->outputs[i], + TGSI_WRITEMASK_Z ); break; case TGSI_SEMANTIC_COLOR: - fulldecl = make_output_decl(i, - TGSI_SEMANTIC_COLOR, - outputSemanticIndex[i], - TGSI_WRITEMASK_XYZW, - outputFlags[i]); + t->outputs[i] = ureg_DECL_output( ureg, + TGSI_SEMANTIC_COLOR, + outputSemanticIndex[i] ); break; default: assert(0); return 0; } - ti += tgsi_build_full_declaration(&fulldecl, - &tokens[ti], - header, - maxTokens - ti ); } } else { - /* vertex prog */ - for (i = 0; i < numOutputs; i++) { - struct tgsi_full_declaration fulldecl; - fulldecl = make_output_decl(i, - outputSemanticName[i], - outputSemanticIndex[i], - TGSI_WRITEMASK_XYZW, - outputFlags[i]); - ti += tgsi_build_full_declaration(&fulldecl, - &tokens[ti], - header, - maxTokens - ti ); + for (i = 0; i < numInputs; i++) { + t->inputs[i] = ureg_DECL_vs_input(ureg, i); } - } - /* temporary decls */ - { - GLboolean inside_range = GL_FALSE; - GLuint start_range = 0; - - tempsUsed[MAX_PROGRAM_TEMPS] = GL_FALSE; - for (i = 0; i < MAX_PROGRAM_TEMPS + 1; i++) { - if (tempsUsed[i] && !inside_range) { - inside_range = GL_TRUE; - start_range = i; - } - else if (!tempsUsed[i] && inside_range) { - struct tgsi_full_declaration fulldecl; - - inside_range = GL_FALSE; - fulldecl = make_temp_decl( start_range, i - 1 ); - ti += tgsi_build_full_declaration( - &fulldecl, - &tokens[ti], - header, - maxTokens - ti ); - } + for (i = 0; i < numOutputs; i++) { + t->outputs[i] = ureg_DECL_output( ureg, + outputSemanticName[i], + outputSemanticIndex[i] ); } } /* Declare address register. - */ + */ if (program->NumAddressRegs > 0) { - struct tgsi_full_declaration fulldecl; - assert( program->NumAddressRegs == 1 ); - - fulldecl = make_addr_decl( 0, 0 ); - ti += tgsi_build_full_declaration( - &fulldecl, - &tokens[ti], - header, - maxTokens - ti ); - - indirectAccess = GL_TRUE; + t->address[0] = ureg_DECL_address( ureg ); } - /* immediates/literals */ - memset(immediates, ~0, sizeof(immediates)); - /* Emit immediates only when there is no address register in use. - * FIXME: Be smarter and recognize param arrays -- indirect addressing is - * only valid within the referenced array. + /* Emit constants and immediates. Mesa uses a single index space + * for these, so we put all the translated regs in t->constants. */ - if (program->Parameters && !indirectAccess) { - for (i = 0; i < program->Parameters->NumParameters; i++) { - if (program->Parameters->Parameters[i].Type == PROGRAM_CONSTANT) { - struct tgsi_full_immediate fullimm; - - fullimm = make_immediate( program->Parameters->ParameterValues[i], 4 ); - ti += tgsi_build_full_immediate( - &fullimm, - &tokens[ti], - header, - maxTokens - ti ); - immediates[i] = numImmediates; - numImmediates++; - } - } - } - - /* constant buffer refs */ if (program->Parameters) { - GLint start = -1, end = -1; - + + t->constants = CALLOC( program->Parameters->NumParameters, + sizeof t->constants[0] ); + if (t->constants == NULL) + goto out; + for (i = 0; i < program->Parameters->NumParameters; i++) { - GLboolean emit = (i == program->Parameters->NumParameters - 1); - GLboolean matches; - switch (program->Parameters->Parameters[i].Type) { case PROGRAM_ENV_PARAM: case PROGRAM_STATE_VAR: case PROGRAM_NAMED_PARAM: case PROGRAM_UNIFORM: - matches = GL_TRUE; + t->constants[i] = ureg_DECL_constant( ureg, i ); break; + + /* Emit immediates only when there is no address register + * in use. FIXME: Be smarter and recognize param arrays: + * indirect addressing is only valid within the referenced + * array. + */ case PROGRAM_CONSTANT: - matches = indirectAccess; + if (program->NumAddressRegs > 0) + t->constants[i] = ureg_DECL_constant( ureg, i ); + else + t->constants[i] = + ureg_DECL_immediate( ureg, + program->Parameters->ParameterValues[i], + 4 ); break; default: - matches = GL_FALSE; - } - - if (matches) { - if (start == -1) { - /* begin a sequence */ - start = i; - end = i; - } - else { - /* continue sequence */ - end = i; - } - } - else { - if (start != -1) { - /* end of sequence */ - emit = GL_TRUE; - } - } - - if (emit && start >= 0) { - struct tgsi_full_declaration fulldecl; - - fulldecl = make_constant_decl( start, end ); - ti += tgsi_build_full_declaration( - &fulldecl, - &tokens[ti], - header, - maxTokens - ti ); - start = end = -1; + break; } } } @@ -1145,58 +828,55 @@ st_translate_mesa_program( /* texture samplers */ for (i = 0; i < ctx->Const.MaxTextureImageUnits; i++) { if (program->SamplersUsed & (1 << i)) { - struct tgsi_full_declaration fulldecl; - - fulldecl = make_sampler_decl( i ); - ti += tgsi_build_full_declaration( - &fulldecl, - &tokens[ti], - header, - maxTokens - ti ); + t->samplers[i] = ureg_DECL_sampler( ureg, i ); } } - /* invert WPOS fragment input */ - if (wposTemp >= 0) { - ti += emit_inverted_wpos(&tokens[ti], wposTemp, winHeightConst, - inputMapping[FRAG_ATTRIB_WPOS], - header, maxTokens - ti); - preamble_size = 2; /* two instructions added */ + /* Emit each instruction in turn: + */ + for (i = 0; i < program->NumInstructions; i++) { + set_insn_start( t, ureg_get_instruction_number( ureg )); + compile_instruction( t, &program->Instructions[i] ); } - for (i = 0; i < program->NumInstructions; i++) { - struct tgsi_full_instruction fullinst; - - compile_instruction( - &program->Instructions[i], - &fullinst, - inputMapping, - outputMapping, - immediates, - indirectAccess, - preamble_size, - procType, - &insideSubroutine, - wposTemp); - - ti += tgsi_build_full_instruction( - &fullinst, - &tokens[ti], - header, - maxTokens - ti ); + /* Fix up all emitted labels: + */ + for (i = 0; i < t->labels_count; i++) { + ureg_fixup_label( ureg, + t->labels[i].token, + t->insn[t->labels[i].branch_target] ); } -#if DEBUG - if(!tgsi_sanity_check(tokens)) { - debug_printf("Due to sanity check failure(s) above the following shader program is invalid:\n"); - debug_printf("\nOriginal program:\n%s", program->String); - debug_printf("\nMesa program:\n"); + tokens = ureg_get_tokens( ureg, NULL ); + ureg_destroy( ureg ); + +out: + FREE(t->insn); + FREE(t->labels); + FREE(t->constants); + + if (t->error) { + debug_printf("%s: translate error flag set\n", __FUNCTION__); + FREE((void *)tokens); + tokens = NULL; + } + + if (!tokens) { + debug_printf("%s: failed to translate Mesa program:\n", __FUNCTION__); _mesa_print_program(program); - debug_printf("\nTGSI program:\n"); - tgsi_dump(tokens, 0); assert(0); } -#endif - return ti; + return tokens; +} + + +/** + * Tokens cannot be free with _mesa_free otherwise the builtin gallium + * malloc debugging will get confused. + */ +void +st_free_tokens(const struct tgsi_token *tokens) +{ + FREE((void *)tokens); } diff --git a/src/mesa/state_tracker/st_mesa_to_tgsi.h b/src/mesa/state_tracker/st_mesa_to_tgsi.h index b465b3bddcf..c0d1ff59e1f 100644 --- a/src/mesa/state_tracker/st_mesa_to_tgsi.h +++ b/src/mesa/state_tracker/st_mesa_to_tgsi.h @@ -39,7 +39,7 @@ extern "C" { struct tgsi_token; struct gl_program; -GLuint +const struct tgsi_token * st_translate_mesa_program( GLcontext *ctx, uint procType, @@ -54,9 +54,10 @@ st_translate_mesa_program( const GLuint outputMapping[], const ubyte outputSemanticName[], const ubyte outputSemanticIndex[], - const GLbitfield outputFlags[], - struct tgsi_token *tokens, - GLuint maxTokens ); + const GLbitfield outputFlags[] ); + +void +st_free_tokens(const struct tgsi_token *tokens); #if defined __cplusplus diff --git a/src/mesa/state_tracker/st_program.c b/src/mesa/state_tracker/st_program.c index d2da20ae424..927f60cc7e9 100644 --- a/src/mesa/state_tracker/st_program.c +++ b/src/mesa/state_tracker/st_program.c @@ -49,9 +49,6 @@ #include "cso_cache/cso_context.h" -#define ST_MAX_SHADER_TOKENS (8 * 1024) - - #define TGSI_DEBUG 0 @@ -70,12 +67,9 @@ st_translate_vertex_program(struct st_context *st, const ubyte *outputSemanticIndex) { struct pipe_context *pipe = st->pipe; - struct tgsi_token *tokens; GLuint defaultOutputMapping[VERT_RESULT_MAX]; - struct pipe_shader_state vs; GLuint attr, i; GLuint num_generic = 0; - GLuint num_tokens; ubyte vs_input_semantic_name[PIPE_MAX_SHADER_INPUTS]; ubyte vs_input_semantic_index[PIPE_MAX_SHADER_INPUTS]; @@ -88,14 +82,7 @@ st_translate_vertex_program(struct st_context *st, GLbitfield input_flags[MAX_PROGRAM_INPUTS]; GLbitfield output_flags[MAX_PROGRAM_OUTPUTS]; - tokens = (struct tgsi_token *)MALLOC(ST_MAX_SHADER_TOKENS * sizeof *tokens); - if(!tokens) { - /* FIXME: propagate error to the caller */ - assert(0); - return; - } - - memset(&vs, 0, sizeof(vs)); +// memset(&vs, 0, sizeof(vs)); memset(input_flags, 0, sizeof(input_flags)); memset(output_flags, 0, sizeof(output_flags)); @@ -330,7 +317,7 @@ st_translate_vertex_program(struct st_context *st, /* free old shader state, if any */ if (stvp->state.tokens) { - _mesa_free((void *) stvp->state.tokens); + st_free_tokens(stvp->state.tokens); stvp->state.tokens = NULL; } if (stvp->driver_shader) { @@ -338,43 +325,32 @@ st_translate_vertex_program(struct st_context *st, stvp->driver_shader = NULL; } - /* XXX: fix static allocation of tokens: - */ - num_tokens = st_translate_mesa_program(st->ctx, - TGSI_PROCESSOR_VERTEX, - &stvp->Base.Base, - /* inputs */ - vs_num_inputs, - stvp->input_to_index, - vs_input_semantic_name, - vs_input_semantic_index, - NULL, - input_flags, - /* outputs */ - vs_num_outputs, - outputMapping, - vs_output_semantic_name, - vs_output_semantic_index, - output_flags, - /* tokenized result */ - tokens, ST_MAX_SHADER_TOKENS); - - assert(num_tokens < ST_MAX_SHADER_TOKENS); - - vs.tokens = (struct tgsi_token *) - _mesa_realloc(tokens, - ST_MAX_SHADER_TOKENS * sizeof *tokens, - num_tokens * sizeof *tokens); + stvp->state.tokens = + st_translate_mesa_program(st->ctx, + TGSI_PROCESSOR_VERTEX, + &stvp->Base.Base, + /* inputs */ + vs_num_inputs, + stvp->input_to_index, + vs_input_semantic_name, + vs_input_semantic_index, + NULL, + input_flags, + /* outputs */ + vs_num_outputs, + outputMapping, + vs_output_semantic_name, + vs_output_semantic_index, + output_flags ); stvp->num_inputs = vs_num_inputs; - stvp->state = vs; /* struct copy */ - stvp->driver_shader = pipe->create_vs_state(pipe, &vs); + stvp->driver_shader = pipe->create_vs_state(pipe, &stvp->state); if (0) _mesa_print_program(&stvp->Base.Base); if (TGSI_DEBUG) - tgsi_dump( vs.tokens, 0 ); + tgsi_dump( stvp->state.tokens, 0 ); } @@ -383,7 +359,6 @@ st_translate_vertex_program(struct st_context *st, * Translate a Mesa fragment shader into a TGSI shader. * \param inputMapping to map fragment program input registers to TGSI * input slots - * \param tokensOut destination for TGSI tokens * \return pointer to cached pipe_shader object. */ void @@ -392,16 +367,13 @@ st_translate_fragment_program(struct st_context *st, const GLuint inputMapping[]) { struct pipe_context *pipe = st->pipe; - struct tgsi_token *tokens; GLuint outputMapping[FRAG_RESULT_MAX]; GLuint defaultInputMapping[FRAG_ATTRIB_MAX]; - struct pipe_shader_state fs; GLuint interpMode[16]; /* XXX size? */ GLuint attr; const GLbitfield inputsRead = stfp->Base.Base.InputsRead; GLuint vslot = 0; GLuint num_generic = 0; - GLuint num_tokens; uint fs_num_inputs = 0; @@ -412,14 +384,7 @@ st_translate_fragment_program(struct st_context *st, GLbitfield input_flags[MAX_PROGRAM_INPUTS]; GLbitfield output_flags[MAX_PROGRAM_OUTPUTS]; - tokens = (struct tgsi_token *)MALLOC(ST_MAX_SHADER_TOKENS * sizeof *tokens); - if(!tokens) { - /* FIXME: propagate error to the caller */ - assert(0); - return; - } - - memset(&fs, 0, sizeof(fs)); +// memset(&fs, 0, sizeof(fs)); memset(input_flags, 0, sizeof(input_flags)); memset(output_flags, 0, sizeof(output_flags)); @@ -541,42 +506,31 @@ st_translate_fragment_program(struct st_context *st, if (!inputMapping) inputMapping = defaultInputMapping; - /* XXX: fix static allocation of tokens: - */ - num_tokens = st_translate_mesa_program(st->ctx, - TGSI_PROCESSOR_FRAGMENT, - &stfp->Base.Base, - /* inputs */ - fs_num_inputs, - inputMapping, - stfp->input_semantic_name, - stfp->input_semantic_index, - interpMode, - input_flags, - /* outputs */ - fs_num_outputs, - outputMapping, - fs_output_semantic_name, - fs_output_semantic_index, - output_flags, - /* tokenized result */ - tokens, ST_MAX_SHADER_TOKENS); - - assert(num_tokens < ST_MAX_SHADER_TOKENS); - - fs.tokens = (struct tgsi_token *) - _mesa_realloc(tokens, - ST_MAX_SHADER_TOKENS * sizeof *tokens, - num_tokens * sizeof *tokens); - - stfp->state = fs; /* struct copy */ - stfp->driver_shader = pipe->create_fs_state(pipe, &fs); + stfp->state.tokens = + st_translate_mesa_program(st->ctx, + TGSI_PROCESSOR_FRAGMENT, + &stfp->Base.Base, + /* inputs */ + fs_num_inputs, + inputMapping, + stfp->input_semantic_name, + stfp->input_semantic_index, + interpMode, + input_flags, + /* outputs */ + fs_num_outputs, + outputMapping, + fs_output_semantic_name, + fs_output_semantic_index, + output_flags ); + + stfp->driver_shader = pipe->create_fs_state(pipe, &stfp->state); if (0) _mesa_print_program(&stfp->Base.Base); if (TGSI_DEBUG) - tgsi_dump( fs.tokens, 0/*TGSI_DUMP_VERBOSE*/ ); + tgsi_dump( stfp->state.tokens, 0/*TGSI_DUMP_VERBOSE*/ ); } diff --git a/src/mesa/swrast/s_fragprog.c b/src/mesa/swrast/s_fragprog.c index 613a91b0ecd..77a77f0bcbb 100644 --- a/src/mesa/swrast/s_fragprog.c +++ b/src/mesa/swrast/s_fragprog.c @@ -89,6 +89,8 @@ fetch_texel_lod( GLcontext *ctx, const GLfloat texcoord[4], GLfloat lambda, * Fetch a texel with the given partial derivatives to compute a level * of detail in the mipmap. * Called via machine->FetchTexelDeriv() + * \param lodBias the lod bias which may be specified by a TXB instruction, + * otherwise zero. */ static void fetch_texel_deriv( GLcontext *ctx, const GLfloat texcoord[4], @@ -96,7 +98,8 @@ fetch_texel_deriv( GLcontext *ctx, const GLfloat texcoord[4], GLfloat lodBias, GLuint unit, GLfloat color[4] ) { SWcontext *swrast = SWRAST_CONTEXT(ctx); - const struct gl_texture_object *texObj = ctx->Texture.Unit[unit]._Current; + const struct gl_texture_unit *texUnit = &ctx->Texture.Unit[unit]; + const struct gl_texture_object *texObj = texUnit->_Current; if (texObj) { const struct gl_texture_image *texImg = @@ -108,10 +111,12 @@ fetch_texel_deriv( GLcontext *ctx, const GLfloat texcoord[4], lambda = _swrast_compute_lambda(texdx[0], texdy[0], /* ds/dx, ds/dy */ texdx[1], texdy[1], /* dt/dx, dt/dy */ - texdx[3], texdy[2], /* dq/dx, dq/dy */ + texdx[3], texdy[3], /* dq/dx, dq/dy */ texW, texH, texcoord[0], texcoord[1], texcoord[3], - 1.0F / texcoord[3]) + lodBias; + 1.0F / texcoord[3]); + + lambda += lodBias + texUnit->LodBias + texObj->LodBias; lambda = CLAMP(lambda, texObj->MinLod, texObj->MaxLod); diff --git a/src/mesa/swrast/s_imaging.c b/src/mesa/swrast/s_imaging.c deleted file mode 100644 index 3578b713f61..00000000000 --- a/src/mesa/swrast/s_imaging.c +++ /dev/null @@ -1,196 +0,0 @@ -/* - * Mesa 3-D graphics library - * Version: 6.5 - * - * Copyright (C) 1999-2005 Brian Paul 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, sublicense, - * 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 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 NONINFRINGEMENT. IN NO EVENT SHALL - * BRIAN PAUL 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. - */ - -/* KW: Moved these here to remove knowledge of swrast from core mesa. - * Should probably pull the entire software implementation of these - * extensions into either swrast or a sister module. - */ - -#include "main/glheader.h" -#include "main/colortab.h" -#include "main/convolve.h" -#include "s_context.h" -#include "s_span.h" - - -void -_swrast_CopyColorTable( GLcontext *ctx, - GLenum target, GLenum internalformat, - GLint x, GLint y, GLsizei width) -{ - GLchan data[MAX_WIDTH][4]; - struct gl_buffer_object *bufferSave; - - if (!ctx->ReadBuffer->_ColorReadBuffer) { - /* no readbuffer - OK */ - return; - } - - if (width > MAX_WIDTH) - width = MAX_WIDTH; - - swrast_render_start(ctx); - - /* read the data from framebuffer */ - _swrast_read_rgba_span( ctx, ctx->ReadBuffer->_ColorReadBuffer, - width, x, y, CHAN_TYPE, data ); - - swrast_render_finish(ctx); - - /* save PBO binding */ - bufferSave = ctx->Unpack.BufferObj; - ctx->Unpack.BufferObj = ctx->Shared->NullBufferObj; - - _mesa_ColorTable(target, internalformat, width, GL_RGBA, CHAN_TYPE, data); - - /* restore PBO binding */ - ctx->Unpack.BufferObj = bufferSave; -} - - -void -_swrast_CopyColorSubTable( GLcontext *ctx,GLenum target, GLsizei start, - GLint x, GLint y, GLsizei width) -{ - GLchan data[MAX_WIDTH][4]; - struct gl_buffer_object *bufferSave; - - if (!ctx->ReadBuffer->_ColorReadBuffer) { - /* no readbuffer - OK */ - return; - } - - if (width > MAX_WIDTH) - width = MAX_WIDTH; - - swrast_render_start(ctx); - - /* read the data from framebuffer */ - _swrast_read_rgba_span( ctx, ctx->ReadBuffer->_ColorReadBuffer, - width, x, y, CHAN_TYPE, data ); - - swrast_render_finish(ctx); - - /* save PBO binding */ - bufferSave = ctx->Unpack.BufferObj; - ctx->Unpack.BufferObj = ctx->Shared->NullBufferObj; - - _mesa_ColorSubTable(target, start, width, GL_RGBA, CHAN_TYPE, data); - - /* restore PBO binding */ - ctx->Unpack.BufferObj = bufferSave; -} - - -void -_swrast_CopyConvolutionFilter1D(GLcontext *ctx, GLenum target, - GLenum internalFormat, - GLint x, GLint y, GLsizei width) -{ - GLchan rgba[MAX_CONVOLUTION_WIDTH][4]; - struct gl_buffer_object *bufferSave; - - if (!ctx->ReadBuffer->_ColorReadBuffer) { - /* no readbuffer - OK */ - return; - } - - swrast_render_start(ctx); - - /* read the data from framebuffer */ - _swrast_read_rgba_span( ctx, ctx->ReadBuffer->_ColorReadBuffer, - width, x, y, CHAN_TYPE, rgba ); - - swrast_render_finish(ctx); - - /* save PBO binding */ - bufferSave = ctx->Unpack.BufferObj; - ctx->Unpack.BufferObj = ctx->Shared->NullBufferObj; - - /* store as convolution filter */ - _mesa_ConvolutionFilter1D(target, internalFormat, width, - GL_RGBA, CHAN_TYPE, rgba); - - /* restore PBO binding */ - ctx->Unpack.BufferObj = bufferSave; -} - - -void -_swrast_CopyConvolutionFilter2D(GLcontext *ctx, GLenum target, - GLenum internalFormat, - GLint x, GLint y, GLsizei width, GLsizei height) -{ - struct gl_pixelstore_attrib packSave; - GLchan rgba[MAX_CONVOLUTION_HEIGHT][MAX_CONVOLUTION_WIDTH][4]; - GLint i; - struct gl_buffer_object *bufferSave; - - if (!ctx->ReadBuffer->_ColorReadBuffer) { - /* no readbuffer - OK */ - return; - } - - swrast_render_start(ctx); - - /* read pixels from framebuffer */ - for (i = 0; i < height; i++) { - _swrast_read_rgba_span( ctx, ctx->ReadBuffer->_ColorReadBuffer, - width, x, y + i, CHAN_TYPE, rgba[i] ); - } - - swrast_render_finish(ctx); - - /* - * HACK: save & restore context state so we can store this as a - * convolution filter via the GL api. Doesn't call any callbacks - * hanging off ctx->Unpack statechanges. - */ - - packSave = ctx->Unpack; /* save pixel packing params */ - - ctx->Unpack.Alignment = 1; - ctx->Unpack.RowLength = MAX_CONVOLUTION_WIDTH; - ctx->Unpack.SkipPixels = 0; - ctx->Unpack.SkipRows = 0; - ctx->Unpack.ImageHeight = 0; - ctx->Unpack.SkipImages = 0; - ctx->Unpack.SwapBytes = GL_FALSE; - ctx->Unpack.LsbFirst = GL_FALSE; - ctx->Unpack.BufferObj = ctx->Shared->NullBufferObj; - ctx->NewState |= _NEW_PACKUNPACK; - - /* save PBO binding */ - bufferSave = ctx->Unpack.BufferObj; - ctx->Unpack.BufferObj = ctx->Shared->NullBufferObj; - - _mesa_ConvolutionFilter2D(target, internalFormat, width, height, - GL_RGBA, CHAN_TYPE, rgba); - - /* restore PBO binding */ - ctx->Unpack.BufferObj = bufferSave; - - ctx->Unpack = packSave; /* restore pixel packing params */ - ctx->NewState |= _NEW_PACKUNPACK; -} diff --git a/src/mesa/swrast/s_texfilter.c b/src/mesa/swrast/s_texfilter.c index 6b1f9346475..004d4e05ae5 100644 --- a/src/mesa/swrast/s_texfilter.c +++ b/src/mesa/swrast/s_texfilter.c @@ -1862,7 +1862,7 @@ choose_cube_face(const struct gl_texture_object *texObj, GLuint face; GLfloat sc, tc, ma; - if (arx > ary && arx > arz) { + if (arx >= ary && arx >= arz) { if (rx >= 0.0F) { face = FACE_POS_X; sc = -rz; @@ -1876,7 +1876,7 @@ choose_cube_face(const struct gl_texture_object *texObj, ma = arx; } } - else if (ary > arx && ary > arz) { + else if (ary >= arx && ary >= arz) { if (ry >= 0.0F) { face = FACE_POS_Y; sc = rx; @@ -1905,8 +1905,12 @@ choose_cube_face(const struct gl_texture_object *texObj, } } - newCoord[0] = ( sc / ma + 1.0F ) * 0.5F; - newCoord[1] = ( tc / ma + 1.0F ) * 0.5F; + { + const float ima = 1.0F / ma; + newCoord[0] = ( sc * ima + 1.0F ) * 0.5F; + newCoord[1] = ( tc * ima + 1.0F ) * 0.5F; + } + return (const struct gl_texture_image **) texObj->Image[face]; } diff --git a/src/mesa/swrast/s_texstore.c b/src/mesa/swrast/s_texstore.c deleted file mode 100644 index f9ff9ad6a42..00000000000 --- a/src/mesa/swrast/s_texstore.c +++ /dev/null @@ -1,601 +0,0 @@ -/* - * Mesa 3-D graphics library - * Version: 6.5.2 - * - * Copyright (C) 1999-2006 Brian Paul 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, sublicense, - * 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 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 NONINFRINGEMENT. IN NO EVENT SHALL - * BRIAN PAUL 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. - */ - -/* - * Authors: - * Brian Paul - */ - - -/* - * The functions in this file are mostly related to software texture fallbacks. - * This includes texture image transfer/packing and texel fetching. - * Hardware drivers will likely override most of this. - */ - - - -#include "main/glheader.h" -#include "main/imports.h" -#include "main/colormac.h" -#include "main/context.h" -#include "main/convolve.h" -#include "main/image.h" -#include "main/macros.h" -#include "main/mipmap.h" -#include "main/texformat.h" -#include "main/teximage.h" -#include "main/texstore.h" - -#include "s_context.h" -#include "s_depth.h" -#include "s_span.h" - - -/** - * Read an RGBA image from the frame buffer. - * This is used by glCopyTex[Sub]Image[12]D(). - * \param x window source x - * \param y window source y - * \param width image width - * \param height image height - * \param type datatype for returned GL_RGBA image - * \return pointer to image - */ -static GLvoid * -read_color_image( GLcontext *ctx, GLint x, GLint y, GLenum type, - GLsizei width, GLsizei height ) -{ - struct gl_renderbuffer *rb = ctx->ReadBuffer->_ColorReadBuffer; - const GLint pixelSize = _mesa_bytes_per_pixel(GL_RGBA, type); - const GLint stride = width * pixelSize; - GLint row; - GLubyte *image, *dst; - - image = (GLubyte *) _mesa_malloc(width * height * pixelSize); - if (!image) - return NULL; - - swrast_render_start(ctx); - - dst = image; - for (row = 0; row < height; row++) { - _swrast_read_rgba_span(ctx, rb, width, x, y + row, type, dst); - dst += stride; - } - - swrast_render_finish(ctx); - - return image; -} - - -/** - * As above, but read data from depth buffer. Returned as GLuints. - * \sa read_color_image - */ -static GLuint * -read_depth_image( GLcontext *ctx, GLint x, GLint y, - GLsizei width, GLsizei height ) -{ - struct gl_renderbuffer *rb = ctx->ReadBuffer->_DepthBuffer; - GLuint *image, *dst; - GLint i; - - image = (GLuint *) _mesa_malloc(width * height * sizeof(GLuint)); - if (!image) - return NULL; - - swrast_render_start(ctx); - - dst = image; - for (i = 0; i < height; i++) { - _swrast_read_depth_span_uint(ctx, rb, width, x, y + i, dst); - dst += width; - } - - swrast_render_finish(ctx); - - return image; -} - - -/** - * As above, but read data from depth+stencil buffers. - */ -static GLuint * -read_depth_stencil_image(GLcontext *ctx, GLint x, GLint y, - GLsizei width, GLsizei height) -{ - struct gl_renderbuffer *depthRb = ctx->ReadBuffer->_DepthBuffer; - struct gl_renderbuffer *stencilRb = ctx->ReadBuffer->_StencilBuffer; - GLuint *image, *dst; - GLint i; - - ASSERT(depthRb); - ASSERT(stencilRb); - - image = (GLuint *) _mesa_malloc(width * height * sizeof(GLuint)); - if (!image) - return NULL; - - swrast_render_start(ctx); - - /* read from depth buffer */ - dst = image; - if (depthRb->DataType == GL_UNSIGNED_INT) { - for (i = 0; i < height; i++) { - _swrast_get_row(ctx, depthRb, width, x, y + i, dst, sizeof(GLuint)); - dst += width; - } - } - else { - GLushort z16[MAX_WIDTH]; - ASSERT(depthRb->DataType == GL_UNSIGNED_SHORT); - for (i = 0; i < height; i++) { - GLint j; - _swrast_get_row(ctx, depthRb, width, x, y + i, z16, sizeof(GLushort)); - /* convert GLushorts to GLuints */ - for (j = 0; j < width; j++) { - dst[j] = z16[j]; - } - dst += width; - } - } - - /* put depth values into bits 0xffffff00 */ - if (ctx->ReadBuffer->Visual.depthBits == 24) { - GLint j; - for (j = 0; j < width * height; j++) { - image[j] <<= 8; - } - } - else if (ctx->ReadBuffer->Visual.depthBits == 16) { - GLint j; - for (j = 0; j < width * height; j++) { - image[j] = (image[j] << 16) | (image[j] & 0xff00); - } - } - else { - /* this handles arbitrary depthBits >= 12 */ - const GLint rShift = ctx->ReadBuffer->Visual.depthBits; - const GLint lShift = 32 - rShift; - GLint j; - for (j = 0; j < width * height; j++) { - GLuint z = (image[j] << lShift); - image[j] = z | (z >> rShift); - } - } - - /* read stencil values and interleave into image array */ - dst = image; - for (i = 0; i < height; i++) { - GLstencil stencil[MAX_WIDTH]; - GLint j; - ASSERT(8 * sizeof(GLstencil) == stencilRb->StencilBits); - _swrast_get_row(ctx, stencilRb, width, x, y + i, - stencil, sizeof(GLstencil)); - for (j = 0; j < width; j++) { - dst[j] = (dst[j] & 0xffffff00) | (stencil[j] & 0xff); - } - dst += width; - } - - swrast_render_finish(ctx); - - return image; -} - - -static GLboolean -is_depth_format(GLenum format) -{ - switch (format) { - case GL_DEPTH_COMPONENT: - case GL_DEPTH_COMPONENT16: - case GL_DEPTH_COMPONENT24: - case GL_DEPTH_COMPONENT32: - return GL_TRUE; - default: - return GL_FALSE; - } -} - - -static GLboolean -is_depth_stencil_format(GLenum format) -{ - switch (format) { - case GL_DEPTH_STENCIL_EXT: - case GL_DEPTH24_STENCIL8_EXT: - return GL_TRUE; - default: - return GL_FALSE; - } -} - - -/* - * Fallback for Driver.CopyTexImage1D(). - */ -void -_swrast_copy_teximage1d( GLcontext *ctx, GLenum target, GLint level, - GLenum internalFormat, - GLint x, GLint y, GLsizei width, GLint border ) -{ - struct gl_texture_unit *texUnit; - struct gl_texture_object *texObj; - struct gl_texture_image *texImage; - - texUnit = &ctx->Texture.Unit[ctx->Texture.CurrentUnit]; - texObj = _mesa_select_tex_object(ctx, texUnit, target); - ASSERT(texObj); - texImage = _mesa_select_tex_image(ctx, texObj, target, level); - ASSERT(texImage); - - ASSERT(ctx->Driver.TexImage1D); - - if (is_depth_format(internalFormat)) { - /* read depth image from framebuffer */ - GLuint *image = read_depth_image(ctx, x, y, width, 1); - if (!image) { - _mesa_error(ctx, GL_OUT_OF_MEMORY, "glCopyTexImage1D"); - return; - } - /* call glTexImage1D to redefine the texture */ - ctx->Driver.TexImage1D(ctx, target, level, internalFormat, - width, border, - GL_DEPTH_COMPONENT, GL_UNSIGNED_INT, image, - &ctx->DefaultPacking, texObj, texImage); - _mesa_free(image); - } - else if (is_depth_stencil_format(internalFormat)) { - /* read depth/stencil image from framebuffer */ - GLuint *image = read_depth_stencil_image(ctx, x, y, width, 1); - if (!image) { - _mesa_error(ctx, GL_OUT_OF_MEMORY, "glCopyTexImage1D"); - return; - } - /* call glTexImage1D to redefine the texture */ - ctx->Driver.TexImage1D(ctx, target, level, internalFormat, - width, border, - GL_DEPTH_STENCIL_EXT, GL_UNSIGNED_INT_24_8_EXT, - image, &ctx->DefaultPacking, texObj, texImage); - _mesa_free(image); - } - else { - /* read RGBA image from framebuffer */ - const GLenum format = GL_RGBA; - const GLenum type = ctx->ReadBuffer->_ColorReadBuffer->DataType; - GLvoid *image = read_color_image(ctx, x, y, type, width, 1); - if (!image) { - _mesa_error(ctx, GL_OUT_OF_MEMORY, "glCopyTexImage1D"); - return; - } - /* call glTexImage1D to redefine the texture */ - ctx->Driver.TexImage1D(ctx, target, level, internalFormat, - width, border, format, type, image, - &ctx->DefaultPacking, texObj, texImage); - _mesa_free(image); - } - - /* GL_SGIS_generate_mipmap */ - if (level == texObj->BaseLevel && texObj->GenerateMipmap) { - ctx->Driver.GenerateMipmap(ctx, target, texObj); - } -} - - -/** - * Fallback for Driver.CopyTexImage2D(). - * - * We implement CopyTexImage by reading the image from the framebuffer - * then passing it to the ctx->Driver.TexImage2D() function. - * - * Device drivers should try to implement direct framebuffer->texture copies. - */ -void -_swrast_copy_teximage2d( GLcontext *ctx, GLenum target, GLint level, - GLenum internalFormat, - GLint x, GLint y, GLsizei width, GLsizei height, - GLint border ) -{ - struct gl_texture_unit *texUnit; - struct gl_texture_object *texObj; - struct gl_texture_image *texImage; - - texUnit = &ctx->Texture.Unit[ctx->Texture.CurrentUnit]; - texObj = _mesa_select_tex_object(ctx, texUnit, target); - ASSERT(texObj); - texImage = _mesa_select_tex_image(ctx, texObj, target, level); - ASSERT(texImage); - - ASSERT(ctx->Driver.TexImage2D); - - if (is_depth_format(internalFormat)) { - /* read depth image from framebuffer */ - GLuint *image = read_depth_image(ctx, x, y, width, height); - if (!image) { - _mesa_error(ctx, GL_OUT_OF_MEMORY, "glCopyTexImage2D"); - return; - } - /* call glTexImage2D to redefine the texture */ - ctx->Driver.TexImage2D(ctx, target, level, internalFormat, - width, height, border, - GL_DEPTH_COMPONENT, GL_UNSIGNED_INT, image, - &ctx->DefaultPacking, texObj, texImage); - _mesa_free(image); - } - else if (is_depth_stencil_format(internalFormat)) { - GLuint *image = read_depth_stencil_image(ctx, x, y, width, height); - if (!image) { - _mesa_error(ctx, GL_OUT_OF_MEMORY, "glCopyTexImage2D"); - return; - } - /* call glTexImage2D to redefine the texture */ - ctx->Driver.TexImage2D(ctx, target, level, internalFormat, - width, height, border, - GL_DEPTH_STENCIL_EXT, GL_UNSIGNED_INT_24_8_EXT, - image, &ctx->DefaultPacking, texObj, texImage); - _mesa_free(image); - } - else { - /* read RGBA image from framebuffer */ - const GLenum format = GL_RGBA; - const GLenum type = ctx->ReadBuffer->_ColorReadBuffer->DataType; - GLvoid *image = read_color_image(ctx, x, y, type, width, height); - if (!image) { - _mesa_error(ctx, GL_OUT_OF_MEMORY, "glCopyTexImage2D"); - return; - } - /* call glTexImage2D to redefine the texture */ - ctx->Driver.TexImage2D(ctx, target, level, internalFormat, - width, height, border, format, type, image, - &ctx->DefaultPacking, texObj, texImage); - _mesa_free(image); - } - - /* GL_SGIS_generate_mipmap */ - if (level == texObj->BaseLevel && texObj->GenerateMipmap) { - ctx->Driver.GenerateMipmap(ctx, target, texObj); - } -} - - -/* - * Fallback for Driver.CopyTexSubImage1D(). - */ -void -_swrast_copy_texsubimage1d( GLcontext *ctx, GLenum target, GLint level, - GLint xoffset, GLint x, GLint y, GLsizei width ) -{ - struct gl_texture_unit *texUnit; - struct gl_texture_object *texObj; - struct gl_texture_image *texImage; - - texUnit = &ctx->Texture.Unit[ctx->Texture.CurrentUnit]; - texObj = _mesa_select_tex_object(ctx, texUnit, target); - ASSERT(texObj); - texImage = _mesa_select_tex_image(ctx, texObj, target, level); - ASSERT(texImage); - - ASSERT(ctx->Driver.TexImage1D); - - if (texImage->_BaseFormat == GL_DEPTH_COMPONENT) { - /* read depth image from framebuffer */ - GLuint *image = read_depth_image(ctx, x, y, width, 1); - if (!image) { - _mesa_error(ctx, GL_OUT_OF_MEMORY, "glCopyTexSubImage1D"); - return; - } - - /* call glTexSubImage1D to redefine the texture */ - ctx->Driver.TexSubImage1D(ctx, target, level, xoffset, width, - GL_DEPTH_COMPONENT, GL_UNSIGNED_INT, image, - &ctx->DefaultPacking, texObj, texImage); - _mesa_free(image); - } - else if (texImage->_BaseFormat == GL_DEPTH_STENCIL_EXT) { - /* read depth/stencil image from framebuffer */ - GLuint *image = read_depth_stencil_image(ctx, x, y, width, 1); - if (!image) { - _mesa_error(ctx, GL_OUT_OF_MEMORY, "glCopyTexSubImage1D"); - return; - } - /* call glTexImage1D to redefine the texture */ - ctx->Driver.TexSubImage1D(ctx, target, level, xoffset, width, - GL_DEPTH_STENCIL_EXT, GL_UNSIGNED_INT_24_8_EXT, - image, &ctx->DefaultPacking, texObj, texImage); - _mesa_free(image); - } - else { - /* read RGBA image from framebuffer */ - const GLenum format = GL_RGBA; - const GLenum type = ctx->ReadBuffer->_ColorReadBuffer->DataType; - GLvoid *image = read_color_image(ctx, x, y, type, width, 1); - if (!image) { - _mesa_error( ctx, GL_OUT_OF_MEMORY, "glCopyTexSubImage1D" ); - return; - } - /* now call glTexSubImage1D to do the real work */ - ctx->Driver.TexSubImage1D(ctx, target, level, xoffset, width, - format, type, image, - &ctx->DefaultPacking, texObj, texImage); - _mesa_free(image); - } - - /* GL_SGIS_generate_mipmap */ - if (level == texObj->BaseLevel && texObj->GenerateMipmap) { - ctx->Driver.GenerateMipmap(ctx, target, texObj); - } -} - - -/** - * Fallback for Driver.CopyTexSubImage2D(). - * - * Read the image from the framebuffer then hand it - * off to ctx->Driver.TexSubImage2D(). - */ -void -_swrast_copy_texsubimage2d( GLcontext *ctx, - GLenum target, GLint level, - GLint xoffset, GLint yoffset, - GLint x, GLint y, GLsizei width, GLsizei height ) -{ - struct gl_texture_unit *texUnit; - struct gl_texture_object *texObj; - struct gl_texture_image *texImage; - - texUnit = &ctx->Texture.Unit[ctx->Texture.CurrentUnit]; - texObj = _mesa_select_tex_object(ctx, texUnit, target); - ASSERT(texObj); - texImage = _mesa_select_tex_image(ctx, texObj, target, level); - ASSERT(texImage); - - ASSERT(ctx->Driver.TexImage2D); - - if (texImage->_BaseFormat == GL_DEPTH_COMPONENT) { - /* read depth image from framebuffer */ - GLuint *image = read_depth_image(ctx, x, y, width, height); - if (!image) { - _mesa_error(ctx, GL_OUT_OF_MEMORY, "glCopyTexSubImage2D"); - return; - } - /* call glTexImage2D to redefine the texture */ - ctx->Driver.TexSubImage2D(ctx, target, level, - xoffset, yoffset, width, height, - GL_DEPTH_COMPONENT, GL_UNSIGNED_INT, image, - &ctx->DefaultPacking, texObj, texImage); - _mesa_free(image); - } - else if (texImage->_BaseFormat == GL_DEPTH_STENCIL_EXT) { - /* read depth/stencil image from framebuffer */ - GLuint *image = read_depth_stencil_image(ctx, x, y, width, height); - if (!image) { - _mesa_error(ctx, GL_OUT_OF_MEMORY, "glCopyTexSubImage2D"); - return; - } - /* call glTexImage2D to redefine the texture */ - ctx->Driver.TexSubImage2D(ctx, target, level, - xoffset, yoffset, width, height, - GL_DEPTH_STENCIL_EXT, GL_UNSIGNED_INT_24_8_EXT, - image, &ctx->DefaultPacking, texObj, texImage); - _mesa_free(image); - } - else { - /* read RGBA image from framebuffer */ - const GLenum format = GL_RGBA; - const GLenum type = ctx->ReadBuffer->_ColorReadBuffer->DataType; - GLvoid *image = read_color_image(ctx, x, y, type, width, height); - if (!image) { - _mesa_error( ctx, GL_OUT_OF_MEMORY, "glCopyTexSubImage2D" ); - return; - } - /* now call glTexSubImage2D to do the real work */ - ctx->Driver.TexSubImage2D(ctx, target, level, - xoffset, yoffset, width, height, - format, type, image, - &ctx->DefaultPacking, texObj, texImage); - _mesa_free(image); - } - - /* GL_SGIS_generate_mipmap */ - if (level == texObj->BaseLevel && texObj->GenerateMipmap) { - ctx->Driver.GenerateMipmap(ctx, target, texObj); - } -} - - -/* - * Fallback for Driver.CopyTexSubImage3D(). - */ -void -_swrast_copy_texsubimage3d( GLcontext *ctx, - GLenum target, GLint level, - GLint xoffset, GLint yoffset, GLint zoffset, - GLint x, GLint y, GLsizei width, GLsizei height ) -{ - struct gl_texture_unit *texUnit; - struct gl_texture_object *texObj; - struct gl_texture_image *texImage; - - texUnit = &ctx->Texture.Unit[ctx->Texture.CurrentUnit]; - texObj = _mesa_select_tex_object(ctx, texUnit, target); - ASSERT(texObj); - texImage = _mesa_select_tex_image(ctx, texObj, target, level); - ASSERT(texImage); - - ASSERT(ctx->Driver.TexImage3D); - - if (texImage->_BaseFormat == GL_DEPTH_COMPONENT) { - /* read depth image from framebuffer */ - GLuint *image = read_depth_image(ctx, x, y, width, height); - if (!image) { - _mesa_error(ctx, GL_OUT_OF_MEMORY, "glCopyTexSubImage3D"); - return; - } - /* call glTexImage3D to redefine the texture */ - ctx->Driver.TexSubImage3D(ctx, target, level, - xoffset, yoffset, zoffset, width, height, 1, - GL_DEPTH_COMPONENT, GL_UNSIGNED_INT, image, - &ctx->DefaultPacking, texObj, texImage); - _mesa_free(image); - } - else if (texImage->_BaseFormat == GL_DEPTH_STENCIL_EXT) { - /* read depth/stencil image from framebuffer */ - GLuint *image = read_depth_stencil_image(ctx, x, y, width, height); - if (!image) { - _mesa_error(ctx, GL_OUT_OF_MEMORY, "glCopyTexSubImage3D"); - return; - } - /* call glTexImage3D to redefine the texture */ - ctx->Driver.TexSubImage3D(ctx, target, level, - xoffset, yoffset, zoffset, width, height, 1, - GL_DEPTH_STENCIL_EXT, GL_UNSIGNED_INT_24_8_EXT, - image, &ctx->DefaultPacking, texObj, texImage); - _mesa_free(image); - } - else { - /* read RGBA image from framebuffer */ - const GLenum format = GL_RGBA; - const GLenum type = ctx->ReadBuffer->_ColorReadBuffer->DataType; - GLvoid *image = read_color_image(ctx, x, y, type, width, height); - if (!image) { - _mesa_error( ctx, GL_OUT_OF_MEMORY, "glCopyTexSubImage3D" ); - return; - } - /* now call glTexSubImage3D to do the real work */ - ctx->Driver.TexSubImage3D(ctx, target, level, - xoffset, yoffset, zoffset, width, height, 1, - format, type, image, - &ctx->DefaultPacking, texObj, texImage); - _mesa_free(image); - } - - /* GL_SGIS_generate_mipmap */ - if (level == texObj->BaseLevel && texObj->GenerateMipmap) { - ctx->Driver.GenerateMipmap(ctx, target, texObj); - } -} diff --git a/src/mesa/swrast/swrast.h b/src/mesa/swrast/swrast.h index c319ca62f9c..c183b315b67 100644 --- a/src/mesa/swrast/swrast.h +++ b/src/mesa/swrast/swrast.h @@ -207,60 +207,6 @@ extern void _swrast_print_vertex( GLcontext *ctx, const SWvertex *v ); -/* - * Imaging fallbacks (a better solution should be found, perhaps - * moving all the imaging fallback code to a new module) - */ -extern void -_swrast_CopyConvolutionFilter2D(GLcontext *ctx, GLenum target, - GLenum internalFormat, - GLint x, GLint y, GLsizei width, - GLsizei height); -extern void -_swrast_CopyConvolutionFilter1D(GLcontext *ctx, GLenum target, - GLenum internalFormat, - GLint x, GLint y, GLsizei width); -extern void -_swrast_CopyColorSubTable( GLcontext *ctx,GLenum target, GLsizei start, - GLint x, GLint y, GLsizei width); -extern void -_swrast_CopyColorTable( GLcontext *ctx, - GLenum target, GLenum internalformat, - GLint x, GLint y, GLsizei width); - - -/* - * Texture fallbacks. Could also live in a new module - * with the rest of the texture store fallbacks? - */ -extern void -_swrast_copy_teximage1d(GLcontext *ctx, GLenum target, GLint level, - GLenum internalFormat, - GLint x, GLint y, GLsizei width, GLint border); - -extern void -_swrast_copy_teximage2d(GLcontext *ctx, GLenum target, GLint level, - GLenum internalFormat, - GLint x, GLint y, GLsizei width, GLsizei height, - GLint border); - - -extern void -_swrast_copy_texsubimage1d(GLcontext *ctx, GLenum target, GLint level, - GLint xoffset, GLint x, GLint y, GLsizei width); - -extern void -_swrast_copy_texsubimage2d(GLcontext *ctx, - GLenum target, GLint level, - GLint xoffset, GLint yoffset, - GLint x, GLint y, GLsizei width, GLsizei height); - -extern void -_swrast_copy_texsubimage3d(GLcontext *ctx, - GLenum target, GLint level, - GLint xoffset, GLint yoffset, GLint zoffset, - GLint x, GLint y, GLsizei width, GLsizei height); - extern void _swrast_eject_texture_images(GLcontext *ctx); diff --git a/src/mesa/vbo/vbo_exec_array.c b/src/mesa/vbo/vbo_exec_array.c index b9550d6106c..774cffc451b 100644 --- a/src/mesa/vbo/vbo_exec_array.c +++ b/src/mesa/vbo/vbo_exec_array.c @@ -672,29 +672,69 @@ vbo_exec_DrawRangeElementsBaseVertex(GLenum mode, const GLvoid *indices, GLint basevertex) { + static GLuint warnCount = 0; GET_CURRENT_CONTEXT(ctx); if (!_mesa_validate_DrawRangeElements( ctx, mode, start, end, count, type, indices, basevertex )) return; + /* NOTE: It's important that 'end' is a reasonable value. + * in _tnl_draw_prims(), we use end to determine how many vertices + * to transform. If it's too large, we can unnecessarily split prims + * or we can read/write out of memory in several different places! + */ + if (end >= ctx->Array.ArrayObj->_MaxElement) { /* the max element is out of bounds of one or more enabled arrays */ - _mesa_warning(ctx, "glDraw[Range]Elements{,BaseVertex}(start %u, end %u, " - "count %d, type 0x%x, indices=%p, base=%d)\n" - "\tindex=%u is out of bounds (max=%u) " - "Element Buffer %u (size %d)", - start, end, count, type, indices, end, basevertex, - ctx->Array.ArrayObj->_MaxElement - 1, - ctx->Array.ElementArrayBufferObj->Name, - ctx->Array.ElementArrayBufferObj->Size); + warnCount++; + + if (warnCount < 10) { + _mesa_warning(ctx, "glDraw[Range]Elements(start %u, end %u, count %d, " + "type 0x%x, indices=%p)\n" + "\tend is out of bounds (max=%u) " + "Element Buffer %u (size %d)\n" + "\tThis should probably be fixed in the application.", + start, end, count, type, indices, + ctx->Array.ArrayObj->_MaxElement - 1, + ctx->Array.ElementArrayBufferObj->Name, + ctx->Array.ElementArrayBufferObj->Size); + } if (0) dump_element_buffer(ctx, type); if (0) _mesa_print_arrays(ctx); - return; + +#ifdef DEBUG + /* 'end' was out of bounds, but now let's check the actual array + * indexes to see if any of them are out of bounds. If so, warn + * and skip the draw to avoid potential segfault, etc. + */ + { + GLuint max = _mesa_max_buffer_index(ctx, count, type, indices, + ctx->Array.ElementArrayBufferObj); + if (max >= ctx->Array.ArrayObj->_MaxElement) { + if (warnCount < 10) { + _mesa_warning(ctx, "glDraw[Range]Elements(start %u, end %u, " + "count %d, type 0x%x, indices=%p)\n" + "\tindex=%u is out of bounds (max=%u) " + "Element Buffer %u (size %d)\n" + "\tSkipping the glDrawRangeElements() call", + start, end, count, type, indices, max, + ctx->Array.ArrayObj->_MaxElement - 1, + ctx->Array.ElementArrayBufferObj->Name, + ctx->Array.ElementArrayBufferObj->Size); + } + return; + } + /* XXX we could also find the min index and compare to 'start' + * to see if start is correct. But it's more likely to get the + * upper bound wrong. + */ + } +#endif } else if (0) { _mesa_printf("glDraw[Range]Elements{,BaseVertex}" diff --git a/src/mesa/vbo/vbo_rebase.c b/src/mesa/vbo/vbo_rebase.c index 799a25fc1cb..55a82ee369e 100644 --- a/src/mesa/vbo/vbo_rebase.c +++ b/src/mesa/vbo/vbo_rebase.c @@ -127,7 +127,10 @@ void vbo_rebase_prims( GLcontext *ctx, _mesa_printf("%s %d..%d\n", __FUNCTION__, min_index, max_index); - if (ib && ctx->Extensions.ARB_draw_elements_base_vertex) { + /* XXX this path is disabled for now. + * There's rendering corruption in some apps when it's enabled. + */ + if (0 && ib && ctx->Extensions.ARB_draw_elements_base_vertex) { /* If we can just tell the hardware or the TNL to interpret our * indices with a different base, do so. */ diff --git a/src/mesa/x86-64/glapi_x86-64.S b/src/mesa/x86-64/glapi_x86-64.S index 907deb4d2fc..72d0532906a 100644 --- a/src/mesa/x86-64/glapi_x86-64.S +++ b/src/mesa/x86-64/glapi_x86-64.S @@ -30393,6 +30393,7 @@ GL_PREFIX(_dispatch_stub_794): .globl GL_PREFIX(RenderbufferStorage) ; .set GL_PREFIX(RenderbufferStorage), GL_PREFIX(RenderbufferStorageEXT) .globl GL_PREFIX(BlitFramebuffer) ; .set GL_PREFIX(BlitFramebuffer), GL_PREFIX(_dispatch_stub_783) .globl GL_PREFIX(FramebufferTextureLayer) ; .set GL_PREFIX(FramebufferTextureLayer), GL_PREFIX(FramebufferTextureLayerEXT) + .globl GL_PREFIX(ProvokingVertex) ; .set GL_PREFIX(ProvokingVertex), GL_PREFIX(ProvokingVertexEXT) #if defined(GLX_USE_TLS) && defined(__linux__) .section ".note.ABI-tag", "a" diff --git a/src/mesa/x86/glapi_x86.S b/src/mesa/x86/glapi_x86.S index 5f12b4fb6a0..12c77f434ec 100644 --- a/src/mesa/x86/glapi_x86.S +++ b/src/mesa/x86/glapi_x86.S @@ -1316,6 +1316,7 @@ GLNAME(gl_dispatch_functions_start): GL_STUB_ALIAS(IsRenderbuffer, _gloffset_IsRenderbufferEXT, IsRenderbuffer@4, IsRenderbufferEXT, IsRenderbufferEXT@4) GL_STUB_ALIAS(RenderbufferStorage, _gloffset_RenderbufferStorageEXT, RenderbufferStorage@16, RenderbufferStorageEXT, RenderbufferStorageEXT@16) GL_STUB_ALIAS(FramebufferTextureLayer, _gloffset_FramebufferTextureLayerEXT, FramebufferTextureLayer@20, FramebufferTextureLayerEXT, FramebufferTextureLayerEXT@20) + GL_STUB_ALIAS(ProvokingVertex, _gloffset_ProvokingVertexEXT, ProvokingVertex@4, ProvokingVertexEXT, ProvokingVertexEXT@4) GLOBL GLNAME(gl_dispatch_functions_end) HIDDEN(GLNAME(gl_dispatch_functions_end)) diff --git a/src/xvmc/context.c b/src/xvmc/context.c index 273f6580292..9c2b6648bb5 100644 --- a/src/xvmc/context.c +++ b/src/xvmc/context.c @@ -1,6 +1,7 @@ #include <assert.h> #include <X11/Xlib.h> #include <X11/extensions/XvMClib.h> +#include <X11/Xlibint.h> #include <pipe/p_context.h> #include <vl_display.h> #include <vl_screen.h> @@ -137,6 +138,7 @@ Status XvMCCreateContext(Display *display, XvPortID port, int surface_type_id, i struct vlScreen *vl_scrn; struct vlContext *vl_ctx; struct pipe_context *pipe; + Display *dpy = display; assert(display); @@ -176,6 +178,7 @@ Status XvMCCreateContext(Display *display, XvPortID port, int surface_type_id, i context->port = port; context->privData = vl_ctx; + SyncHandle(); return Success; } diff --git a/src/xvmc/subpicture.c b/src/xvmc/subpicture.c index c8f70c90d0c..09e9c98e6af 100644 --- a/src/xvmc/subpicture.c +++ b/src/xvmc/subpicture.c @@ -2,6 +2,7 @@ #include <X11/Xlib.h> #include <X11/extensions/Xvlib.h> #include <X11/extensions/XvMC.h> +#include <X11/Xlibint.h> Status XvMCCreateSubpicture ( @@ -13,6 +14,7 @@ Status XvMCCreateSubpicture int xvimage_id ) { + Display *dpy = display; assert(display); if (!context) @@ -38,7 +40,8 @@ Status XvMCCreateSubpicture subpicture->component_order[2] = 0; subpicture->component_order[3] = 0; /* TODO: subpicture->privData = ;*/ - + + SyncHandle(); return Success; } diff --git a/src/xvmc/surface.c b/src/xvmc/surface.c index 7c5f45bd346..fea351b84f0 100644 --- a/src/xvmc/surface.c +++ b/src/xvmc/surface.c @@ -1,6 +1,7 @@ #include <assert.h> #include <X11/Xlib.h> #include <X11/extensions/XvMC.h> +#include <X11/Xlibint.h> #include <vl_display.h> #include <vl_screen.h> #include <vl_context.h> @@ -61,6 +62,7 @@ Status XvMCCreateSurface(Display *display, XvMCContext *context, XvMCSurface *su { struct vlContext *vl_ctx; struct vlSurface *vl_sfc; + Display *dpy = display; assert(display); @@ -90,6 +92,7 @@ Status XvMCCreateSurface(Display *display, XvMCContext *context, XvMCSurface *su surface->height = context->height; surface->privData = vl_sfc; + SyncHandle(); return Success; } diff --git a/windows/VC7/mesa/mesa/mesa.vcproj b/windows/VC7/mesa/mesa/mesa.vcproj index 4d5d4c744b6..4d7df741702 100644 --- a/windows/VC7/mesa/mesa/mesa.vcproj +++ b/windows/VC7/mesa/mesa/mesa.vcproj @@ -416,9 +416,6 @@ RelativePath="..\..\..\..\src\mesa\swrast\s_fragprog.c">
</File>
<File
- RelativePath="..\..\..\..\src\mesa\swrast\s_imaging.c">
- </File>
- <File
RelativePath="..\..\..\..\src\mesa\swrast\s_lines.c">
</File>
<File
@@ -446,9 +443,6 @@ RelativePath="..\..\..\..\src\mesa\swrast\s_texfilter.c">
</File>
<File
- RelativePath="..\..\..\..\src\mesa\swrast\s_texstore.c">
- </File>
- <File
RelativePath="..\..\..\..\src\mesa\swrast\s_triangle.c">
</File>
<File
diff --git a/windows/VC8/mesa/mesa/mesa.vcproj b/windows/VC8/mesa/mesa/mesa.vcproj index 068da1612d4..f15c4435f61 100644 --- a/windows/VC8/mesa/mesa/mesa.vcproj +++ b/windows/VC8/mesa/mesa/mesa.vcproj @@ -783,10 +783,6 @@ >
</File>
<File
- RelativePath="..\..\..\..\src\mesa\swrast\s_imaging.c"
- >
- </File>
- <File
RelativePath="..\..\..\..\src\mesa\swrast\s_lines.c"
>
</File>
@@ -823,10 +819,6 @@ >
</File>
<File
- RelativePath="..\..\..\..\src\mesa\swrast\s_texstore.c"
- >
- </File>
- <File
RelativePath="..\..\..\..\src\mesa\swrast\s_triangle.c"
>
</File>
|