From 026b85e62b3d8812afb5f04df29aeac28c52b331 Mon Sep 17 00:00:00 2001 From: marha Date: Sat, 12 Feb 2011 15:47:35 +0000 Subject: xkeyboard-config pixman xkbcompt mesa git update 12 Feb 2011 --- mesalib/src/mesa/Makefile | 412 +- mesalib/src/mesa/drivers/dri/Makefile.template | 12 +- .../windows/gldirect/dx9/gld_primitive_dx9.c | 1 - .../windows/gldirect/mesasw/gld_wgl_mesasw.c | 1 - mesalib/src/mesa/main/mtypes.h | 6740 ++++++++++---------- mesalib/src/mesa/program/ir_to_mesa.cpp | 6537 +++++++++---------- mesalib/src/mesa/state_tracker/st_cb_fbo.c | 1259 ++-- mesalib/src/mesa/state_tracker/st_manager.c | 15 +- mesalib/src/mesa/swrast/s_trispan.h | 31 - mesalib/src/mesa/x86/read_rgba_span_x86.S | 1364 ++-- 10 files changed, 8198 insertions(+), 8174 deletions(-) delete mode 100644 mesalib/src/mesa/swrast/s_trispan.h (limited to 'mesalib/src/mesa') diff --git a/mesalib/src/mesa/Makefile b/mesalib/src/mesa/Makefile index a6025e990..36cbdd994 100644 --- a/mesalib/src/mesa/Makefile +++ b/mesalib/src/mesa/Makefile @@ -1,188 +1,224 @@ -# src/mesa/Makefile - -TOP = ../.. -include $(TOP)/configs/current - -MESA_LIBS := libmesa.a libmesagallium.a -DEPENDS := depend - -MESA_OBJ_DIR := . - - -include sources.mak - -# adjust object dirs -MESA_OBJECTS := $(addprefix $(MESA_OBJ_DIR)/, $(MESA_OBJECTS)) -MESA_GALLIUM_OBJECTS := $(addprefix $(MESA_OBJ_DIR)/, $(MESA_GALLIUM_OBJECTS)) - -# define preprocessor flags -MESA_CPPFLAGS := $(API_DEFINES) $(DEFINES) - -# append include dirs -MESA_CPPFLAGS += $(INCLUDE_DIRS) - -# tidy compiler flags -CFLAGS := $(filter-out $(DEFINES), $(CFLAGS)) -CXXFLAGS := $(filter-out $(DEFINES), $(CXXFLAGS)) - -# LLVM is needed for the state tracker -MESA_CFLAGS := $(LLVM_CFLAGS) - -define mesa-cc-c - @mkdir -p $(dir $@) - $(CC) -c -o $@ $< $($(1)_CPPFLAGS) $($(1)_CFLAGS) $(CFLAGS) -endef - -define mesa-cxx-c - @mkdir -p $(dir $@) - $(CXX) -c -o $@ $< $($(1)_CPPFLAGS) $($(1)_CFLAGS) $(CXXFLAGS) -endef - -$(MESA_OBJ_DIR)/%.o: %.c - $(call mesa-cc-c,MESA) - -$(MESA_OBJ_DIR)/%.o: %.cpp - $(call mesa-cxx-c,MESA) - -$(MESA_OBJ_DIR)/%.o: %.S - $(call mesa-cc-c,MESA) - -# Default: build dependencies, then asm_subdirs, GLSL built-in lib, -# then convenience libs (.a) and finally the device drivers: -default: $(DEPENDS) asm_subdirs $(MESA_LIBS) driver_subdirs - -main/api_exec_es1.c: main/APIspec.xml main/es_generator.py main/APIspecutil.py main/APIspec.py - $(PYTHON2) $(PYTHON_FLAGS) main/es_generator.py -S main/APIspec.xml -V GLES1.1 > $@ - -main/api_exec_es2.c: main/APIspec.xml main/es_generator.py main/APIspecutil.py main/APIspec.py - $(PYTHON2) $(PYTHON_FLAGS) main/es_generator.py -S main/APIspec.xml -V GLES2.0 > $@ - -###################################################################### -# Helper libraries used by many drivers: - -# Make archive of core mesa object files -libmesa.a: $(MESA_OBJECTS) $(GLSL_LIBS) - @ $(MKLIB) -o mesa -static $(MESA_OBJECTS) $(GLSL_LIBS) - -# Make archive of subset of core mesa object files for gallium -libmesagallium.a: $(MESA_GALLIUM_OBJECTS) $(GLSL_LIBS) - @ $(MKLIB) -o mesagallium -static $(MESA_GALLIUM_OBJECTS) $(GLSL_LIBS) - -###################################################################### -# Device drivers -driver_subdirs: $(MESA_LIBS) - @ (cd drivers && $(MAKE)) - - -###################################################################### -# Assembly subdirs -asm_subdirs: - @ if echo "$(ASM_FLAGS)" | grep -q USE_X86_ASM ; then \ - (cd x86 && $(MAKE)) || exit 1 ; \ - fi - @ if echo "$(ASM_FLAGS)" | grep -q USE_X86_64_ASM ; then \ - (cd x86 && $(MAKE)) || exit 1 ; \ - (cd x86-64 && $(MAKE)) || exit 1 ; \ - fi - - -###################################################################### -# Dependency generation - -depend: $(ALL_SOURCES) - @ echo "running $(MKDEP)" - @ touch depend - @$(MKDEP) $(MKDEP_OPTIONS) -p$(MESA_OBJ_DIR)/ $(MESA_CPPFLAGS) \ - $(ALL_SOURCES) > /dev/null 2>/dev/null - -###################################################################### -# Installation rules - -# this isn't fleshed out yet but is probably the way to go in the future -new_install: - (cd drivers && $(MAKE) install) - - -# XXX replace this with new_install above someday -install: default - @for driver in $(DRIVER_DIRS) ; do \ - case "$$driver" in \ - osmesa) if [ "$(DRIVER_DIRS)" = osmesa ]; then \ - $(MAKE) install-headers install-osmesa || exit 1 ; \ - else \ - $(MAKE) install-osmesa || exit 1 ; \ - fi ;; \ - dri) $(MAKE) install-libgl install-dri || exit 1 ;; \ - *) $(MAKE) install-libgl || exit 1 ;; \ - esac ; \ - done - -pcedit = \ - -e 's,@INSTALL_DIR@,$(INSTALL_DIR),' \ - -e 's,@INSTALL_LIB_DIR@,$(INSTALL_LIB_DIR),' \ - -e 's,@INSTALL_INC_DIR@,$(INSTALL_INC_DIR),' \ - -e 's,@VERSION@,$(MESA_MAJOR).$(MESA_MINOR).$(MESA_TINY),' \ - - -gl_pcedit = sed \ - $(pcedit) \ - -e 's,@GL_PC_REQ_PRIV@,$(GL_PC_REQ_PRIV),' \ - -e 's,@GL_PC_LIB_PRIV@,$(GL_PC_LIB_PRIV),' \ - -e 's,@GL_PC_CFLAGS@,$(GL_PC_CFLAGS),' \ - -e 's,@GLX_TLS@,$(GLX_TLS),' \ - -e 's,@GL_LIB@,$(GL_LIB),' - -gl.pc: gl.pc.in - $(gl_pcedit) $< > $@ - -osmesa_pcedit = sed \ - $(pcedit) \ - -e 's,@OSMESA_LIB@,$(OSMESA_LIB),' \ - -e 's,@OSMESA_PC_REQ@,$(OSMESA_PC_REQ),' \ - -e 's,@OSMESA_PC_LIB_PRIV@,$(OSMESA_PC_LIB_PRIV),' - -osmesa.pc: osmesa.pc.in - $(osmesa_pcedit) $< > $@ - -install-headers: - $(INSTALL) -d $(DESTDIR)$(INSTALL_INC_DIR)/GL - $(INSTALL) -m 644 $(TOP)/include/GL/*.h \ - $(DESTDIR)$(INSTALL_INC_DIR)/GL - -install-libgl: default gl.pc install-headers - $(INSTALL) -d $(DESTDIR)$(INSTALL_LIB_DIR) - $(INSTALL) -d $(DESTDIR)$(INSTALL_LIB_DIR)/pkgconfig - $(MINSTALL) $(TOP)/$(LIB_DIR)/$(GL_LIB_GLOB) \ - $(DESTDIR)$(INSTALL_LIB_DIR) - $(INSTALL) -m 644 gl.pc $(DESTDIR)$(INSTALL_LIB_DIR)/pkgconfig - -install-osmesa: default osmesa.pc - $(INSTALL) -d $(DESTDIR)$(INSTALL_LIB_DIR) - $(INSTALL) -d $(DESTDIR)$(INSTALL_LIB_DIR)/pkgconfig - $(MINSTALL) $(TOP)/$(LIB_DIR)/$(OSMESA_LIB_GLOB) \ - $(DESTDIR)$(INSTALL_LIB_DIR) - $(INSTALL) -m 644 osmesa.pc $(DESTDIR)$(INSTALL_LIB_DIR)/pkgconfig - -install-dri: default - cd drivers/dri && $(MAKE) install - - - -# Emacs tags -tags: - etags `find . -name \*.[ch]` $(TOP)/include/GL/*.h - -clean: - -rm -f */*.o - -rm -f */*/*.o - -rm -f depend depend.bak libmesa.a libmesagallium.a - -rm -f drivers/*/*.o - -rm -f *.pc - -@cd drivers/dri && $(MAKE) clean - -@cd drivers/x11 && $(MAKE) clean - -@cd drivers/osmesa && $(MAKE) clean - -@cd x86 && $(MAKE) clean - -@cd x86-64 && $(MAKE) clean - - --include $(DEPENDS) +# src/mesa/Makefile + +TOP = ../.. +include $(TOP)/configs/current + +MESA_LIBS := libmesa.a libmesagallium.a +DEPENDS := depend + +MESA_OBJ_DIR := . +DRICORE_OBJ_DIR := objs-dricore + +include sources.mak + +# adjust object dirs +MESA_OBJECTS := $(addprefix $(MESA_OBJ_DIR)/, $(MESA_OBJECTS)) +MESA_GALLIUM_OBJECTS := $(addprefix $(MESA_OBJ_DIR)/, $(MESA_GALLIUM_OBJECTS)) + +DRICORE_OBJECTS := $(addprefix $(DRICORE_OBJ_DIR)/, $(MESA_OBJECTS)) + +# define preprocessor flags +MESA_CPPFLAGS := $(API_DEFINES) $(DEFINES) + +# append include dirs +MESA_CPPFLAGS += $(INCLUDE_DIRS) + +DRICORE_CPPFLAGS = $(MESA_CPPFLAGS) + +# tidy compiler flags +CFLAGS := $(filter-out $(DEFINES), $(CFLAGS)) +CXXFLAGS := $(filter-out $(DEFINES), $(CXXFLAGS)) + +# LLVM is needed for the state tracker +MESA_CFLAGS := $(LLVM_CFLAGS) $(CFLAGS) +DRICORE_CFLAGS := $(LLVM_CFLAGS) $(DRI_CFLAGS) + +MESA_CXXFLAGS := $(LLVM_CFLAGS) $(CXXFLAGS) +DRICORE_CXXFLAGS := $(LLVM_CFLAGS) $(DRI_CXXFLAGS) + +define mesa-cc-c + @mkdir -p $(dir $@) + $(CC) -c -o $@ $< $($(1)_CPPFLAGS) $($(1)_CFLAGS) +endef + +define mesa-cxx-c + @mkdir -p $(dir $@) + $(CXX) -c -o $@ $< $($(1)_CPPFLAGS) $($(1)_CFLAGS) $($(1)_CXXFLAGS) +endef + +$(MESA_OBJ_DIR)/%.o: %.c + $(call mesa-cc-c,MESA) + +$(MESA_OBJ_DIR)/%.o: %.cpp + $(call mesa-cxx-c,MESA) + +$(MESA_OBJ_DIR)/%.o: %.S + $(call mesa-cc-c,MESA) + +$(DRICORE_OBJ_DIR)/%.o: %.c + $(call mesa-cc-c,DRICORE) + +$(DRICORE_OBJ_DIR)/%.o: %.cpp + $(call mesa-cxx-c,DRICORE) + +$(DRICORE_OBJ_DIR)/%.o: %.S + $(call mesa-cc-c,DRICORE) + +# Default: build dependencies, then asm_subdirs, GLSL built-in lib, +# then convenience libs (.a) and finally the device drivers: +default: $(DEPENDS) asm_subdirs $(MESA_LIBS) $(DRICORE_LIBS) driver_subdirs + +main/api_exec_es1.c: main/APIspec.xml main/es_generator.py main/APIspecutil.py main/APIspec.py + $(PYTHON2) $(PYTHON_FLAGS) main/es_generator.py -S main/APIspec.xml -V GLES1.1 > $@ + +main/api_exec_es2.c: main/APIspec.xml main/es_generator.py main/APIspecutil.py main/APIspec.py + $(PYTHON2) $(PYTHON_FLAGS) main/es_generator.py -S main/APIspec.xml -V GLES2.0 > $@ + +###################################################################### +# Helper libraries used by many drivers: + +# Make archive of core mesa object files +libmesa.a: $(MESA_OBJECTS) $(GLSL_LIBS) + @ $(MKLIB) -o mesa -static $(MESA_OBJECTS) $(GLSL_LIBS) + +# Shared dricore library for classic DRI drivers +$(TOP)/$(LIB_DIR)/libdricore.so: $(DRICORE_OBJECTS) $(DRICORE_GLSL_LIBS) + @$(MKLIB) -o $@ -linker '$(CXX)' -ldflags '$(LDFLAGS)' \ + -cplusplus -noprefix \ + -install $(TOP)/$(LIB_DIR) -id $(DRI_DRIVER_INSTALL_DIR)/$@.dylib \ + $(DRICORE_LIB_DEPS) $(DRICORE_OBJECTS) + +# Make archive of subset of core mesa object files for gallium +libmesagallium.a: $(MESA_GALLIUM_OBJECTS) $(GLSL_LIBS) + @ $(MKLIB) -o mesagallium -static $(MESA_GALLIUM_OBJECTS) $(GLSL_LIBS) + +###################################################################### +# Device drivers +driver_subdirs: $(MESA_LIBS) $(DRICORE_LIBS) + @ (cd drivers && $(MAKE)) + + +###################################################################### +# Assembly subdirs +asm_subdirs: + @ if echo "$(ASM_FLAGS)" | grep -q USE_X86_ASM ; then \ + (cd x86 && $(MAKE)) || exit 1 ; \ + fi + @ if echo "$(ASM_FLAGS)" | grep -q USE_X86_64_ASM ; then \ + (cd x86 && $(MAKE)) || exit 1 ; \ + (cd x86-64 && $(MAKE)) || exit 1 ; \ + fi + + +###################################################################### +# Dependency generation + +depend: $(ALL_SOURCES) + @ echo "running $(MKDEP)" + @ touch depend + @$(MKDEP) $(MKDEP_OPTIONS) -p$(MESA_OBJ_DIR)/ $(MESA_CPPFLAGS) \ + $(ALL_SOURCES) > /dev/null 2>/dev/null + +###################################################################### +# Installation rules + +# this isn't fleshed out yet but is probably the way to go in the future +new_install: + (cd drivers && $(MAKE) install) + +ifneq (,$(DRICORE_LIBS)) +DRICORE_INSTALL_TARGET = install-dricore +endif + +# XXX replace this with new_install above someday +install: default $(DRICORE_INSTALL_TARGET) + @for driver in $(DRIVER_DIRS) ; do \ + case "$$driver" in \ + osmesa) if [ "$(DRIVER_DIRS)" = osmesa ]; then \ + $(MAKE) install-headers install-osmesa || exit 1 ; \ + else \ + $(MAKE) install-osmesa || exit 1 ; \ + fi ;; \ + dri) $(MAKE) install-libgl install-dri || exit 1 ;; \ + *) $(MAKE) install-libgl || exit 1 ;; \ + esac ; \ + done + +pcedit = \ + -e 's,@INSTALL_DIR@,$(INSTALL_DIR),' \ + -e 's,@INSTALL_LIB_DIR@,$(INSTALL_LIB_DIR),' \ + -e 's,@INSTALL_INC_DIR@,$(INSTALL_INC_DIR),' \ + -e 's,@VERSION@,$(MESA_MAJOR).$(MESA_MINOR).$(MESA_TINY),' \ + + +gl_pcedit = sed \ + $(pcedit) \ + -e 's,@GL_PC_REQ_PRIV@,$(GL_PC_REQ_PRIV),' \ + -e 's,@GL_PC_LIB_PRIV@,$(GL_PC_LIB_PRIV),' \ + -e 's,@GL_PC_CFLAGS@,$(GL_PC_CFLAGS),' \ + -e 's,@GLX_TLS@,$(GLX_TLS),' \ + -e 's,@GL_LIB@,$(GL_LIB),' + +gl.pc: gl.pc.in + $(gl_pcedit) $< > $@ + +osmesa_pcedit = sed \ + $(pcedit) \ + -e 's,@OSMESA_LIB@,$(OSMESA_LIB),' \ + -e 's,@OSMESA_PC_REQ@,$(OSMESA_PC_REQ),' \ + -e 's,@OSMESA_PC_LIB_PRIV@,$(OSMESA_PC_LIB_PRIV),' + +osmesa.pc: osmesa.pc.in + $(osmesa_pcedit) $< > $@ + +install-headers: + $(INSTALL) -d $(DESTDIR)$(INSTALL_INC_DIR)/GL + $(INSTALL) -m 644 $(TOP)/include/GL/*.h \ + $(DESTDIR)$(INSTALL_INC_DIR)/GL + +install-libgl: default gl.pc install-headers + $(INSTALL) -d $(DESTDIR)$(INSTALL_LIB_DIR) + $(INSTALL) -d $(DESTDIR)$(INSTALL_LIB_DIR)/pkgconfig + $(MINSTALL) $(TOP)/$(LIB_DIR)/$(GL_LIB_GLOB) \ + $(DESTDIR)$(INSTALL_LIB_DIR) + $(INSTALL) -m 644 gl.pc $(DESTDIR)$(INSTALL_LIB_DIR)/pkgconfig + +install-osmesa: default osmesa.pc + $(INSTALL) -d $(DESTDIR)$(INSTALL_LIB_DIR) + $(INSTALL) -d $(DESTDIR)$(INSTALL_LIB_DIR)/pkgconfig + $(MINSTALL) $(TOP)/$(LIB_DIR)/$(OSMESA_LIB_GLOB) \ + $(DESTDIR)$(INSTALL_LIB_DIR) + $(INSTALL) -m 644 osmesa.pc $(DESTDIR)$(INSTALL_LIB_DIR)/pkgconfig + +install-dri: default + cd drivers/dri && $(MAKE) install + +# We don't need MINSTALL here because we're not installing symbolic links +install-dricore: default + $(INSTALL) -d $(DESTDIR)$(DRI_DRIVER_INSTALL_DIR) + $(INSTALL) -m 755 $(DRICORE_LIBS) $(DESTDIR)$(DRI_DRIVER_INSTALL_DIR) + + +# Emacs tags +tags: + etags `find . -name \*.[ch]` $(TOP)/include/GL/*.h + +clean-dricore: + -rm -f libdricore.so + -rm -f $(DRICORE_LIBS) + -rm -rf $(DRICORE_OBJ_DIR) + +clean: clean-dricore + -rm -f */*.o + -rm -f */*/*.o + -rm -f depend depend.bak libmesa.a libmesagallium.a + -rm -f drivers/*/*.o + -rm -f *.pc + -@cd drivers/dri && $(MAKE) clean + -@cd drivers/x11 && $(MAKE) clean + -@cd drivers/osmesa && $(MAKE) clean + -@cd x86 && $(MAKE) clean + -@cd x86-64 && $(MAKE) clean + + +-include $(DEPENDS) diff --git a/mesalib/src/mesa/drivers/dri/Makefile.template b/mesalib/src/mesa/drivers/dri/Makefile.template index 6be554af7..588210f8a 100644 --- a/mesalib/src/mesa/drivers/dri/Makefile.template +++ b/mesalib/src/mesa/drivers/dri/Makefile.template @@ -1,7 +1,5 @@ # -*-makefile-*- -MESA_MODULES = $(TOP)/src/mesa/libmesa.a - COMMON_GALLIUM_SOURCES = \ ../common/utils.c \ ../common/vblank.c \ @@ -39,13 +37,13 @@ CXXFLAGS += $(API_DEFINES) ##### RULES ##### .c.o: - $(CC) -c $(INCLUDES) $(CFLAGS) $(DRIVER_DEFINES) $< -o $@ + $(CC) -c $(INCLUDES) $(DRI_CFLAGS) $(DRIVER_DEFINES) $< -o $@ .cpp.o: - $(CC) -c $(INCLUDES) $(CXXFLAGS) $(DRIVER_DEFINES) $< -o $@ + $(CC) -c $(INCLUDES) $(DRI_CXXFLAGS) $(DRIVER_DEFINES) $< -o $@ .S.o: - $(CC) -c $(INCLUDES) $(CFLAGS) $(DRIVER_DEFINES) $< -o $@ + $(CC) -c $(INCLUDES) $(DRI_CFLAGS) $(DRIVER_DEFINES) $< -o $@ ##### TARGETS ##### @@ -57,10 +55,10 @@ default: subdirs lib lib: symlinks subdirs depend @$(MAKE) $(LIBNAME) $(TOP)/$(LIB_DIR)/$(LIBNAME) -$(LIBNAME): $(OBJECTS) $(MESA_MODULES) $(EXTRA_MODULES) Makefile \ +$(LIBNAME): $(OBJECTS) $(EXTRA_MODULES) $(MESA_MODULES) Makefile \ $(TOP)/src/mesa/drivers/dri/Makefile.template $(TOP)/src/mesa/drivers/dri/common/dri_test.o $(MKLIB) -o $@.tmp -noprefix -linker '$(CXX)' -ldflags '$(LDFLAGS)' \ - $(OBJECTS) $(MESA_MODULES) $(EXTRA_MODULES) $(DRI_LIB_DEPS) + $(OBJECTS) $(EXTRA_MODULES) $(DRI_LIB_DEPS) $(CXX) $(CFLAGS) -o $@.test $(TOP)/src/mesa/drivers/dri/common/dri_test.o $@.tmp $(DRI_LIB_DEPS) @rm -f $@.test mv -f $@.tmp $@ diff --git a/mesalib/src/mesa/drivers/windows/gldirect/dx9/gld_primitive_dx9.c b/mesalib/src/mesa/drivers/windows/gldirect/dx9/gld_primitive_dx9.c index aaed87028..256ffcc38 100644 --- a/mesalib/src/mesa/drivers/windows/gldirect/dx9/gld_primitive_dx9.c +++ b/mesalib/src/mesa/drivers/windows/gldirect/dx9/gld_primitive_dx9.c @@ -62,7 +62,6 @@ #include "swrast/s_depth.h" #include "swrast/s_lines.h" #include "swrast/s_triangle.h" -#include "swrast/s_trispan.h" #include "tnl/tnl.h" #include "tnl/t_context.h" #include "tnl/t_pipeline.h" diff --git a/mesalib/src/mesa/drivers/windows/gldirect/mesasw/gld_wgl_mesasw.c b/mesalib/src/mesa/drivers/windows/gldirect/mesasw/gld_wgl_mesasw.c index 36ebe9f31..02f0dfacd 100644 --- a/mesalib/src/mesa/drivers/windows/gldirect/mesasw/gld_wgl_mesasw.c +++ b/mesalib/src/mesa/drivers/windows/gldirect/mesasw/gld_wgl_mesasw.c @@ -61,7 +61,6 @@ #include "swrast/s_depth.h" #include "swrast/s_lines.h" #include "swrast/s_triangle.h" -#include "swrast/s_trispan.h" #include "tnl/tnl.h" #include "tnl/t_context.h" #include "tnl/t_pipeline.h" diff --git a/mesalib/src/mesa/main/mtypes.h b/mesalib/src/mesa/main/mtypes.h index 6f0aac6ee..020595bd0 100644 --- a/mesalib/src/mesa/main/mtypes.h +++ b/mesalib/src/mesa/main/mtypes.h @@ -1,3371 +1,3369 @@ -/* - * Mesa 3-D graphics library - * Version: 7.7 - * - * Copyright (C) 1999-2008 Brian Paul All Rights Reserved. - * 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 - * 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. - */ - -/** - * \file mtypes.h - * Main Mesa data structures. - * - * Please try to mark derived values with a leading underscore ('_'). - */ - -#ifndef MTYPES_H -#define MTYPES_H - - -#include "main/glheader.h" -#include "main/config.h" -#include "main/mfeatures.h" -#include "glapi/glapi.h" -#include "math/m_matrix.h" /* GLmatrix */ -#include "main/simple_list.h" /* struct simple_node */ -#include "main/formats.h" /* MESA_FORMAT_COUNT */ - - -/** - * Color channel data type. - */ -#if CHAN_BITS == 8 - typedef GLubyte GLchan; -#define CHAN_MAX 255 -#define CHAN_MAXF 255.0F -#define CHAN_TYPE GL_UNSIGNED_BYTE -#elif CHAN_BITS == 16 - typedef GLushort GLchan; -#define CHAN_MAX 65535 -#define CHAN_MAXF 65535.0F -#define CHAN_TYPE GL_UNSIGNED_SHORT -#elif CHAN_BITS == 32 - typedef GLfloat GLchan; -#define CHAN_MAX 1.0 -#define CHAN_MAXF 1.0F -#define CHAN_TYPE GL_FLOAT -#else -#error "illegal number of color channel bits" -#endif - - -/** - * Stencil buffer data type. - */ -#if STENCIL_BITS==8 - typedef GLubyte GLstencil; -#elif STENCIL_BITS==16 - typedef GLushort GLstencil; -#else -# error "illegal number of stencil bits" -#endif - - -/** - * \name 64-bit extension of GLbitfield. - */ -/*@{*/ -typedef GLuint64 GLbitfield64; - -/** Set a single bit */ -#define BITFIELD64_BIT(b) (1ULL << (b)) - - -/** - * \name Some forward type declarations - */ -/*@{*/ -struct _mesa_HashTable; -struct gl_attrib_node; -struct gl_list_extensions; -struct gl_meta_state; -struct gl_pixelstore_attrib; -struct gl_program_cache; -struct gl_texture_format; -struct gl_texture_image; -struct gl_texture_object; -struct gl_context; -struct st_context; -/*@}*/ - - -/** Extra draw modes beyond GL_POINTS, GL_TRIANGLE_FAN, etc */ -#define PRIM_OUTSIDE_BEGIN_END (GL_POLYGON+1) -#define PRIM_INSIDE_UNKNOWN_PRIM (GL_POLYGON+2) -#define PRIM_UNKNOWN (GL_POLYGON+3) - - -/** - * Shader stages. Note that these will become 5 with tessellation. - * These MUST have the same values as gallium's PIPE_SHADER_* - */ -typedef enum -{ - MESA_SHADER_VERTEX = 0, - MESA_SHADER_FRAGMENT = 1, - MESA_SHADER_GEOMETRY = 2, - MESA_SHADER_TYPES = 3 -} gl_shader_type; - - - -/** - * Indexes for vertex program attributes. - * GL_NV_vertex_program aliases generic attributes over the conventional - * attributes. In GL_ARB_vertex_program shader the aliasing is optional. - * In GL_ARB_vertex_shader / OpenGL 2.0 the aliasing is disallowed (the - * generic attributes are distinct/separate). - */ -typedef enum -{ - VERT_ATTRIB_POS = 0, - VERT_ATTRIB_WEIGHT = 1, - VERT_ATTRIB_NORMAL = 2, - VERT_ATTRIB_COLOR0 = 3, - VERT_ATTRIB_COLOR1 = 4, - VERT_ATTRIB_FOG = 5, - VERT_ATTRIB_COLOR_INDEX = 6, - VERT_ATTRIB_POINT_SIZE = 6, /*alias*/ - VERT_ATTRIB_EDGEFLAG = 7, - VERT_ATTRIB_TEX0 = 8, - VERT_ATTRIB_TEX1 = 9, - VERT_ATTRIB_TEX2 = 10, - VERT_ATTRIB_TEX3 = 11, - VERT_ATTRIB_TEX4 = 12, - VERT_ATTRIB_TEX5 = 13, - VERT_ATTRIB_TEX6 = 14, - VERT_ATTRIB_TEX7 = 15, - VERT_ATTRIB_GENERIC0 = 16, - VERT_ATTRIB_GENERIC1 = 17, - VERT_ATTRIB_GENERIC2 = 18, - VERT_ATTRIB_GENERIC3 = 19, - VERT_ATTRIB_GENERIC4 = 20, - VERT_ATTRIB_GENERIC5 = 21, - VERT_ATTRIB_GENERIC6 = 22, - VERT_ATTRIB_GENERIC7 = 23, - VERT_ATTRIB_GENERIC8 = 24, - VERT_ATTRIB_GENERIC9 = 25, - VERT_ATTRIB_GENERIC10 = 26, - VERT_ATTRIB_GENERIC11 = 27, - VERT_ATTRIB_GENERIC12 = 28, - VERT_ATTRIB_GENERIC13 = 29, - VERT_ATTRIB_GENERIC14 = 30, - VERT_ATTRIB_GENERIC15 = 31, - VERT_ATTRIB_MAX = 32 -} gl_vert_attrib; - -/** - * Bitflags for vertex attributes. - * These are used in bitfields in many places. - */ -/*@{*/ -#define VERT_BIT_POS (1 << VERT_ATTRIB_POS) -#define VERT_BIT_WEIGHT (1 << VERT_ATTRIB_WEIGHT) -#define VERT_BIT_NORMAL (1 << VERT_ATTRIB_NORMAL) -#define VERT_BIT_COLOR0 (1 << VERT_ATTRIB_COLOR0) -#define VERT_BIT_COLOR1 (1 << VERT_ATTRIB_COLOR1) -#define VERT_BIT_FOG (1 << VERT_ATTRIB_FOG) -#define VERT_BIT_COLOR_INDEX (1 << VERT_ATTRIB_COLOR_INDEX) -#define VERT_BIT_EDGEFLAG (1 << VERT_ATTRIB_EDGEFLAG) -#define VERT_BIT_TEX0 (1 << VERT_ATTRIB_TEX0) -#define VERT_BIT_TEX1 (1 << VERT_ATTRIB_TEX1) -#define VERT_BIT_TEX2 (1 << VERT_ATTRIB_TEX2) -#define VERT_BIT_TEX3 (1 << VERT_ATTRIB_TEX3) -#define VERT_BIT_TEX4 (1 << VERT_ATTRIB_TEX4) -#define VERT_BIT_TEX5 (1 << VERT_ATTRIB_TEX5) -#define VERT_BIT_TEX6 (1 << VERT_ATTRIB_TEX6) -#define VERT_BIT_TEX7 (1 << VERT_ATTRIB_TEX7) -#define VERT_BIT_GENERIC0 (1 << VERT_ATTRIB_GENERIC0) -#define VERT_BIT_GENERIC1 (1 << VERT_ATTRIB_GENERIC1) -#define VERT_BIT_GENERIC2 (1 << VERT_ATTRIB_GENERIC2) -#define VERT_BIT_GENERIC3 (1 << VERT_ATTRIB_GENERIC3) -#define VERT_BIT_GENERIC4 (1 << VERT_ATTRIB_GENERIC4) -#define VERT_BIT_GENERIC5 (1 << VERT_ATTRIB_GENERIC5) -#define VERT_BIT_GENERIC6 (1 << VERT_ATTRIB_GENERIC6) -#define VERT_BIT_GENERIC7 (1 << VERT_ATTRIB_GENERIC7) -#define VERT_BIT_GENERIC8 (1 << VERT_ATTRIB_GENERIC8) -#define VERT_BIT_GENERIC9 (1 << VERT_ATTRIB_GENERIC9) -#define VERT_BIT_GENERIC10 (1 << VERT_ATTRIB_GENERIC10) -#define VERT_BIT_GENERIC11 (1 << VERT_ATTRIB_GENERIC11) -#define VERT_BIT_GENERIC12 (1 << VERT_ATTRIB_GENERIC12) -#define VERT_BIT_GENERIC13 (1 << VERT_ATTRIB_GENERIC13) -#define VERT_BIT_GENERIC14 (1 << VERT_ATTRIB_GENERIC14) -#define VERT_BIT_GENERIC15 (1 << VERT_ATTRIB_GENERIC15) - -#define VERT_BIT_TEX(u) (1 << (VERT_ATTRIB_TEX0 + (u))) -#define VERT_BIT_GENERIC(g) (1 << (VERT_ATTRIB_GENERIC0 + (g))) -/*@}*/ - - -/** - * Indexes for vertex program result attributes - */ -typedef enum -{ - VERT_RESULT_HPOS = 0, - VERT_RESULT_COL0 = 1, - VERT_RESULT_COL1 = 2, - VERT_RESULT_FOGC = 3, - VERT_RESULT_TEX0 = 4, - VERT_RESULT_TEX1 = 5, - VERT_RESULT_TEX2 = 6, - VERT_RESULT_TEX3 = 7, - VERT_RESULT_TEX4 = 8, - VERT_RESULT_TEX5 = 9, - VERT_RESULT_TEX6 = 10, - VERT_RESULT_TEX7 = 11, - VERT_RESULT_PSIZ = 12, - VERT_RESULT_BFC0 = 13, - VERT_RESULT_BFC1 = 14, - VERT_RESULT_EDGE = 15, - VERT_RESULT_VAR0 = 16, /**< shader varying */ - VERT_RESULT_MAX = (VERT_RESULT_VAR0 + MAX_VARYING) -} gl_vert_result; - - -/*********************************************/ - -/** - * Indexes for geometry program attributes. - */ -typedef enum -{ - GEOM_ATTRIB_POSITION = 0, - GEOM_ATTRIB_COLOR0 = 1, - GEOM_ATTRIB_COLOR1 = 2, - GEOM_ATTRIB_SECONDARY_COLOR0 = 3, - GEOM_ATTRIB_SECONDARY_COLOR1 = 4, - GEOM_ATTRIB_FOG_FRAG_COORD = 5, - GEOM_ATTRIB_POINT_SIZE = 6, - GEOM_ATTRIB_CLIP_VERTEX = 7, - GEOM_ATTRIB_PRIMITIVE_ID = 8, - GEOM_ATTRIB_TEX_COORD = 9, - - GEOM_ATTRIB_VAR0 = 16, - GEOM_ATTRIB_MAX = (GEOM_ATTRIB_VAR0 + MAX_VARYING) -} gl_geom_attrib; - -/** - * Bitflags for geometry attributes. - * These are used in bitfields in many places. - */ -/*@{*/ -#define GEOM_BIT_COLOR0 (1 << GEOM_ATTRIB_COLOR0) -#define GEOM_BIT_COLOR1 (1 << GEOM_ATTRIB_COLOR1) -#define GEOM_BIT_SCOLOR0 (1 << GEOM_ATTRIB_SECONDARY_COLOR0) -#define GEOM_BIT_SCOLOR1 (1 << GEOM_ATTRIB_SECONDARY_COLOR1) -#define GEOM_BIT_TEX_COORD (1 << GEOM_ATTRIB_TEX_COORD) -#define GEOM_BIT_FOG_COORD (1 << GEOM_ATTRIB_FOG_FRAG_COORD) -#define GEOM_BIT_POSITION (1 << GEOM_ATTRIB_POSITION) -#define GEOM_BIT_POINT_SIDE (1 << GEOM_ATTRIB_POINT_SIZE) -#define GEOM_BIT_CLIP_VERTEX (1 << GEOM_ATTRIB_CLIP_VERTEX) -#define GEOM_BIT_PRIM_ID (1 << GEOM_ATTRIB_PRIMITIVE_ID) -#define GEOM_BIT_VAR0 (1 << GEOM_ATTRIB_VAR0) - -#define GEOM_BIT_VAR(g) (1 << (GEOM_BIT_VAR0 + (g))) -/*@}*/ - - -/** - * Indexes for geometry program result attributes - */ -typedef enum -{ - GEOM_RESULT_POS = 0, - GEOM_RESULT_COL0 = 1, - GEOM_RESULT_COL1 = 2, - GEOM_RESULT_SCOL0 = 3, - GEOM_RESULT_SCOL1 = 4, - GEOM_RESULT_FOGC = 5, - GEOM_RESULT_TEX0 = 6, - GEOM_RESULT_TEX1 = 7, - GEOM_RESULT_TEX2 = 8, - GEOM_RESULT_TEX3 = 9, - GEOM_RESULT_TEX4 = 10, - GEOM_RESULT_TEX5 = 11, - GEOM_RESULT_TEX6 = 12, - GEOM_RESULT_TEX7 = 13, - GEOM_RESULT_PSIZ = 14, - GEOM_RESULT_CLPV = 15, - GEOM_RESULT_PRID = 16, - GEOM_RESULT_LAYR = 17, - GEOM_RESULT_VAR0 = 18, /**< shader varying, should really be 16 */ - /* ### we need to -2 because var0 is 18 instead 16 like in the others */ - GEOM_RESULT_MAX = (GEOM_RESULT_VAR0 + MAX_VARYING - 2) -} gl_geom_result; - - -/** - * Indexes for fragment program input attributes. - */ -typedef enum -{ - FRAG_ATTRIB_WPOS = 0, - FRAG_ATTRIB_COL0 = 1, - FRAG_ATTRIB_COL1 = 2, - FRAG_ATTRIB_FOGC = 3, - FRAG_ATTRIB_TEX0 = 4, - FRAG_ATTRIB_TEX1 = 5, - FRAG_ATTRIB_TEX2 = 6, - FRAG_ATTRIB_TEX3 = 7, - FRAG_ATTRIB_TEX4 = 8, - FRAG_ATTRIB_TEX5 = 9, - FRAG_ATTRIB_TEX6 = 10, - FRAG_ATTRIB_TEX7 = 11, - FRAG_ATTRIB_FACE = 12, /**< front/back face */ - FRAG_ATTRIB_PNTC = 13, /**< sprite/point coord */ - FRAG_ATTRIB_VAR0 = 14, /**< shader varying */ - FRAG_ATTRIB_MAX = (FRAG_ATTRIB_VAR0 + MAX_VARYING) -} gl_frag_attrib; - -/** - * Bitflags for fragment program input attributes. - */ -/*@{*/ -#define FRAG_BIT_WPOS (1 << FRAG_ATTRIB_WPOS) -#define FRAG_BIT_COL0 (1 << FRAG_ATTRIB_COL0) -#define FRAG_BIT_COL1 (1 << FRAG_ATTRIB_COL1) -#define FRAG_BIT_FOGC (1 << FRAG_ATTRIB_FOGC) -#define FRAG_BIT_FACE (1 << FRAG_ATTRIB_FACE) -#define FRAG_BIT_PNTC (1 << FRAG_ATTRIB_PNTC) -#define FRAG_BIT_TEX0 (1 << FRAG_ATTRIB_TEX0) -#define FRAG_BIT_TEX1 (1 << FRAG_ATTRIB_TEX1) -#define FRAG_BIT_TEX2 (1 << FRAG_ATTRIB_TEX2) -#define FRAG_BIT_TEX3 (1 << FRAG_ATTRIB_TEX3) -#define FRAG_BIT_TEX4 (1 << FRAG_ATTRIB_TEX4) -#define FRAG_BIT_TEX5 (1 << FRAG_ATTRIB_TEX5) -#define FRAG_BIT_TEX6 (1 << FRAG_ATTRIB_TEX6) -#define FRAG_BIT_TEX7 (1 << FRAG_ATTRIB_TEX7) -#define FRAG_BIT_VAR0 (1 << FRAG_ATTRIB_VAR0) - -#define FRAG_BIT_TEX(U) (FRAG_BIT_TEX0 << (U)) -#define FRAG_BIT_VAR(V) (FRAG_BIT_VAR0 << (V)) - -#define FRAG_BITS_TEX_ANY (FRAG_BIT_TEX0| \ - FRAG_BIT_TEX1| \ - FRAG_BIT_TEX2| \ - FRAG_BIT_TEX3| \ - FRAG_BIT_TEX4| \ - FRAG_BIT_TEX5| \ - FRAG_BIT_TEX6| \ - FRAG_BIT_TEX7) -/*@}*/ - - -/** - * Fragment program results - */ -typedef enum -{ - FRAG_RESULT_DEPTH = 0, - FRAG_RESULT_STENCIL = 1, - FRAG_RESULT_COLOR = 2, - FRAG_RESULT_DATA0 = 3, - FRAG_RESULT_MAX = (FRAG_RESULT_DATA0 + MAX_DRAW_BUFFERS) -} gl_frag_result; - - -/** - * Indexes for all renderbuffers - */ -typedef enum -{ - /* the four standard color buffers */ - BUFFER_FRONT_LEFT, - BUFFER_BACK_LEFT, - BUFFER_FRONT_RIGHT, - BUFFER_BACK_RIGHT, - BUFFER_DEPTH, - BUFFER_STENCIL, - BUFFER_ACCUM, - /* optional aux buffer */ - BUFFER_AUX0, - /* generic renderbuffers */ - BUFFER_COLOR0, - BUFFER_COLOR1, - BUFFER_COLOR2, - BUFFER_COLOR3, - BUFFER_COLOR4, - BUFFER_COLOR5, - BUFFER_COLOR6, - BUFFER_COLOR7, - BUFFER_COUNT -} gl_buffer_index; - -/** - * Bit flags for all renderbuffers - */ -#define BUFFER_BIT_FRONT_LEFT (1 << BUFFER_FRONT_LEFT) -#define BUFFER_BIT_BACK_LEFT (1 << BUFFER_BACK_LEFT) -#define BUFFER_BIT_FRONT_RIGHT (1 << BUFFER_FRONT_RIGHT) -#define BUFFER_BIT_BACK_RIGHT (1 << BUFFER_BACK_RIGHT) -#define BUFFER_BIT_AUX0 (1 << BUFFER_AUX0) -#define BUFFER_BIT_AUX1 (1 << BUFFER_AUX1) -#define BUFFER_BIT_AUX2 (1 << BUFFER_AUX2) -#define BUFFER_BIT_AUX3 (1 << BUFFER_AUX3) -#define BUFFER_BIT_DEPTH (1 << BUFFER_DEPTH) -#define BUFFER_BIT_STENCIL (1 << BUFFER_STENCIL) -#define BUFFER_BIT_ACCUM (1 << BUFFER_ACCUM) -#define BUFFER_BIT_COLOR0 (1 << BUFFER_COLOR0) -#define BUFFER_BIT_COLOR1 (1 << BUFFER_COLOR1) -#define BUFFER_BIT_COLOR2 (1 << BUFFER_COLOR2) -#define BUFFER_BIT_COLOR3 (1 << BUFFER_COLOR3) -#define BUFFER_BIT_COLOR4 (1 << BUFFER_COLOR4) -#define BUFFER_BIT_COLOR5 (1 << BUFFER_COLOR5) -#define BUFFER_BIT_COLOR6 (1 << BUFFER_COLOR6) -#define BUFFER_BIT_COLOR7 (1 << BUFFER_COLOR7) - -/** - * Mask of all the color buffer bits (but not accum). - */ -#define BUFFER_BITS_COLOR (BUFFER_BIT_FRONT_LEFT | \ - BUFFER_BIT_BACK_LEFT | \ - BUFFER_BIT_FRONT_RIGHT | \ - BUFFER_BIT_BACK_RIGHT | \ - BUFFER_BIT_AUX0 | \ - BUFFER_BIT_COLOR0 | \ - BUFFER_BIT_COLOR1 | \ - BUFFER_BIT_COLOR2 | \ - BUFFER_BIT_COLOR3 | \ - BUFFER_BIT_COLOR4 | \ - BUFFER_BIT_COLOR5 | \ - BUFFER_BIT_COLOR6 | \ - BUFFER_BIT_COLOR7) - - -/** - * Framebuffer configuration (aka visual / pixelformat) - * Note: some of these fields should be boolean, but it appears that - * code in drivers/dri/common/util.c requires int-sized fields. - */ -struct gl_config -{ - GLboolean rgbMode; - GLboolean floatMode; - GLboolean colorIndexMode; /* XXX is this used anywhere? */ - GLuint doubleBufferMode; - GLuint stereoMode; - - GLboolean haveAccumBuffer; - GLboolean haveDepthBuffer; - GLboolean haveStencilBuffer; - - GLint redBits, greenBits, blueBits, alphaBits; /* bits per comp */ - GLuint redMask, greenMask, blueMask, alphaMask; - GLint rgbBits; /* total bits for rgb */ - GLint indexBits; /* total bits for colorindex */ - - GLint accumRedBits, accumGreenBits, accumBlueBits, accumAlphaBits; - GLint depthBits; - GLint stencilBits; - - GLint numAuxBuffers; - - GLint level; - - /* EXT_visual_rating / GLX 1.2 */ - GLint visualRating; - - /* EXT_visual_info / GLX 1.2 */ - GLint transparentPixel; - /* colors are floats scaled to ints */ - GLint transparentRed, transparentGreen, transparentBlue, transparentAlpha; - GLint transparentIndex; - - /* ARB_multisample / SGIS_multisample */ - GLint sampleBuffers; - GLint samples; - - /* SGIX_pbuffer / GLX 1.3 */ - GLint maxPbufferWidth; - GLint maxPbufferHeight; - GLint maxPbufferPixels; - GLint optimalPbufferWidth; /* Only for SGIX_pbuffer. */ - GLint optimalPbufferHeight; /* Only for SGIX_pbuffer. */ - - /* OML_swap_method */ - GLint swapMethod; - - /* EXT_texture_from_pixmap */ - GLint bindToTextureRgb; - GLint bindToTextureRgba; - GLint bindToMipmapTexture; - GLint bindToTextureTargets; - GLint yInverted; - - /* EXT_framebuffer_sRGB */ - GLint sRGBCapable; -}; - - -/** - * Data structure for color tables - */ -struct gl_color_table -{ - GLenum InternalFormat; /**< The user-specified format */ - GLenum _BaseFormat; /**< GL_ALPHA, GL_RGBA, GL_RGB, etc */ - GLuint Size; /**< number of entries in table */ - GLfloat *TableF; /**< Color table, floating point values */ - GLubyte *TableUB; /**< Color table, ubyte values */ - GLubyte RedSize; - GLubyte GreenSize; - GLubyte BlueSize; - GLubyte AlphaSize; - GLubyte LuminanceSize; - GLubyte IntensitySize; -}; - - -/** - * \name Bit flags used for updating material values. - */ -/*@{*/ -#define MAT_ATTRIB_FRONT_AMBIENT 0 -#define MAT_ATTRIB_BACK_AMBIENT 1 -#define MAT_ATTRIB_FRONT_DIFFUSE 2 -#define MAT_ATTRIB_BACK_DIFFUSE 3 -#define MAT_ATTRIB_FRONT_SPECULAR 4 -#define MAT_ATTRIB_BACK_SPECULAR 5 -#define MAT_ATTRIB_FRONT_EMISSION 6 -#define MAT_ATTRIB_BACK_EMISSION 7 -#define MAT_ATTRIB_FRONT_SHININESS 8 -#define MAT_ATTRIB_BACK_SHININESS 9 -#define MAT_ATTRIB_FRONT_INDEXES 10 -#define MAT_ATTRIB_BACK_INDEXES 11 -#define MAT_ATTRIB_MAX 12 - -#define MAT_ATTRIB_AMBIENT(f) (MAT_ATTRIB_FRONT_AMBIENT+(f)) -#define MAT_ATTRIB_DIFFUSE(f) (MAT_ATTRIB_FRONT_DIFFUSE+(f)) -#define MAT_ATTRIB_SPECULAR(f) (MAT_ATTRIB_FRONT_SPECULAR+(f)) -#define MAT_ATTRIB_EMISSION(f) (MAT_ATTRIB_FRONT_EMISSION+(f)) -#define MAT_ATTRIB_SHININESS(f)(MAT_ATTRIB_FRONT_SHININESS+(f)) -#define MAT_ATTRIB_INDEXES(f) (MAT_ATTRIB_FRONT_INDEXES+(f)) - -#define MAT_INDEX_AMBIENT 0 -#define MAT_INDEX_DIFFUSE 1 -#define MAT_INDEX_SPECULAR 2 - -#define MAT_BIT_FRONT_AMBIENT (1< ) */ - GLfloat _NormSpotDirection[4]; /**< normalized spotlight direction */ - GLfloat _VP_inf_spot_attenuation; - - GLfloat _SpotExpTable[EXP_TABLE_SIZE][2]; /**< to replace a pow() call */ - GLfloat _MatAmbient[2][3]; /**< material ambient * light ambient */ - GLfloat _MatDiffuse[2][3]; /**< material diffuse * light diffuse */ - GLfloat _MatSpecular[2][3]; /**< material spec * light specular */ - GLfloat _dli; /**< CI diffuse light intensity */ - GLfloat _sli; /**< CI specular light intensity */ - /*@}*/ -}; - - -/** - * Light model state. - */ -struct gl_lightmodel -{ - GLfloat Ambient[4]; /**< ambient color */ - GLboolean LocalViewer; /**< Local (or infinite) view point? */ - GLboolean TwoSide; /**< Two (or one) sided lighting? */ - GLenum ColorControl; /**< either GL_SINGLE_COLOR - * or GL_SEPARATE_SPECULAR_COLOR */ -}; - - -/** - * Material state. - */ -struct gl_material -{ - GLfloat Attrib[MAT_ATTRIB_MAX][4]; -}; - - -/** - * Accumulation buffer attribute group (GL_ACCUM_BUFFER_BIT) - */ -struct gl_accum_attrib -{ - GLfloat ClearColor[4]; /**< Accumulation buffer clear color */ -}; - - -/** - * Color buffer attribute group (GL_COLOR_BUFFER_BIT). - */ -struct gl_colorbuffer_attrib -{ - GLuint ClearIndex; /**< Index to use for glClear */ - GLclampf ClearColor[4]; /**< Color to use for glClear */ - - GLuint IndexMask; /**< Color index write mask */ - GLubyte ColorMask[MAX_DRAW_BUFFERS][4];/**< Each flag is 0xff or 0x0 */ - - GLenum DrawBuffer[MAX_DRAW_BUFFERS]; /**< Which buffer to draw into */ - - /** - * \name alpha testing - */ - /*@{*/ - GLboolean AlphaEnabled; /**< Alpha test enabled flag */ - GLenum AlphaFunc; /**< Alpha test function */ - GLclampf AlphaRef; /**< Alpha reference value */ - /*@}*/ - - /** - * \name Blending - */ - /*@{*/ - GLbitfield BlendEnabled; /**< Per-buffer blend enable flags */ - GLfloat BlendColor[4]; /**< Blending color */ - struct - { - GLenum SrcRGB; /**< RGB blend source term */ - GLenum DstRGB; /**< RGB blend dest term */ - GLenum SrcA; /**< Alpha blend source term */ - GLenum DstA; /**< Alpha blend dest term */ - GLenum EquationRGB; /**< GL_ADD, GL_SUBTRACT, etc. */ - GLenum EquationA; /**< GL_ADD, GL_SUBTRACT, etc. */ - } Blend[MAX_DRAW_BUFFERS]; - /** Are the blend func terms currently different for each buffer/target? */ - GLboolean _BlendFuncPerBuffer; - /** Are the blend equations currently different for each buffer/target? */ - GLboolean _BlendEquationPerBuffer; - /*@}*/ - - /** - * \name Logic op - */ - /*@{*/ - GLenum LogicOp; /**< Logic operator */ - GLboolean IndexLogicOpEnabled; /**< Color index logic op enabled flag */ - GLboolean ColorLogicOpEnabled; /**< RGBA logic op enabled flag */ - GLboolean _LogicOpEnabled; /**< RGBA logic op + EXT_blend_logic_op enabled flag */ - /*@}*/ - - GLboolean DitherFlag; /**< Dither enable flag */ - - GLenum ClampFragmentColor; /**< GL_TRUE, GL_FALSE or GL_FIXED_ONLY_ARB */ - GLenum ClampReadColor; /**< GL_TRUE, GL_FALSE or GL_FIXED_ONLY_ARB */ - - GLboolean sRGBEnabled; /**< Framebuffer sRGB blending/updating requested */ -}; - - -/** - * Current attribute group (GL_CURRENT_BIT). - */ -struct gl_current_attrib -{ - /** - * \name Current vertex attributes. - * \note Values are valid only after FLUSH_VERTICES has been called. - * \note Index and Edgeflag current values are stored as floats in the - * SIX and SEVEN attribute slots. - */ - GLfloat Attrib[VERT_ATTRIB_MAX][4]; /**< Position, color, texcoords, etc */ - - /** - * \name Current raster position attributes (always valid). - * \note This set of attributes is very similar to the SWvertex struct. - */ - /*@{*/ - GLfloat RasterPos[4]; - GLfloat RasterDistance; - GLfloat RasterColor[4]; - GLfloat RasterSecondaryColor[4]; - GLfloat RasterTexCoords[MAX_TEXTURE_COORD_UNITS][4]; - GLboolean RasterPosValid; - /*@}*/ -}; - - -/** - * Depth buffer attribute group (GL_DEPTH_BUFFER_BIT). - */ -struct gl_depthbuffer_attrib -{ - GLenum Func; /**< Function for depth buffer compare */ - GLclampd Clear; /**< Value to clear depth buffer to */ - GLboolean Test; /**< Depth buffering enabled flag */ - GLboolean Mask; /**< Depth buffer writable? */ - GLboolean BoundsTest; /**< GL_EXT_depth_bounds_test */ - GLfloat BoundsMin, BoundsMax;/**< GL_EXT_depth_bounds_test */ -}; - - -/** - * Evaluator attribute group (GL_EVAL_BIT). - */ -struct gl_eval_attrib -{ - /** - * \name Enable bits - */ - /*@{*/ - GLboolean Map1Color4; - GLboolean Map1Index; - GLboolean Map1Normal; - GLboolean Map1TextureCoord1; - GLboolean Map1TextureCoord2; - GLboolean Map1TextureCoord3; - GLboolean Map1TextureCoord4; - GLboolean Map1Vertex3; - GLboolean Map1Vertex4; - GLboolean Map1Attrib[16]; /* GL_NV_vertex_program */ - GLboolean Map2Color4; - GLboolean Map2Index; - GLboolean Map2Normal; - GLboolean Map2TextureCoord1; - GLboolean Map2TextureCoord2; - GLboolean Map2TextureCoord3; - GLboolean Map2TextureCoord4; - GLboolean Map2Vertex3; - GLboolean Map2Vertex4; - GLboolean Map2Attrib[16]; /* GL_NV_vertex_program */ - GLboolean AutoNormal; - /*@}*/ - - /** - * \name Map Grid endpoints and divisions and calculated du values - */ - /*@{*/ - GLint MapGrid1un; - GLfloat MapGrid1u1, MapGrid1u2, MapGrid1du; - GLint MapGrid2un, MapGrid2vn; - GLfloat MapGrid2u1, MapGrid2u2, MapGrid2du; - GLfloat MapGrid2v1, MapGrid2v2, MapGrid2dv; - /*@}*/ -}; - - -/** - * Fog attribute group (GL_FOG_BIT). - */ -struct gl_fog_attrib -{ - GLboolean Enabled; /**< Fog enabled flag */ - GLfloat Color[4]; /**< Fog color */ - GLfloat Density; /**< Density >= 0.0 */ - GLfloat Start; /**< Start distance in eye coords */ - GLfloat End; /**< End distance in eye coords */ - GLfloat Index; /**< Fog index */ - GLenum Mode; /**< Fog mode */ - GLboolean ColorSumEnabled; - GLenum FogCoordinateSource; /**< GL_EXT_fog_coord */ - GLfloat _Scale; /**< (End == Start) ? 1.0 : 1.0 / (End - Start) */ -}; - - -/** - * \brief Layout qualifiers for gl_FragDepth. - * - * Extension AMD_conservative_depth allows gl_FragDepth to be redeclared with - * a layout qualifier. - * - * \see enum ir_depth_layout - */ -enum gl_frag_depth_layout { - FRAG_DEPTH_LAYOUT_NONE, /**< No layout is specified. */ - FRAG_DEPTH_LAYOUT_ANY, - FRAG_DEPTH_LAYOUT_GREATER, - FRAG_DEPTH_LAYOUT_LESS, - FRAG_DEPTH_LAYOUT_UNCHANGED -}; - - -/** - * Hint attribute group (GL_HINT_BIT). - * - * Values are always one of GL_FASTEST, GL_NICEST, or GL_DONT_CARE. - */ -struct gl_hint_attrib -{ - GLenum PerspectiveCorrection; - GLenum PointSmooth; - GLenum LineSmooth; - GLenum PolygonSmooth; - GLenum Fog; - GLenum ClipVolumeClipping; /**< GL_EXT_clip_volume_hint */ - GLenum TextureCompression; /**< GL_ARB_texture_compression */ - GLenum GenerateMipmap; /**< GL_SGIS_generate_mipmap */ - GLenum FragmentShaderDerivative; /**< GL_ARB_fragment_shader */ -}; - -/** - * Light state flags. - */ -/*@{*/ -#define LIGHT_SPOT 0x1 -#define LIGHT_LOCAL_VIEWER 0x2 -#define LIGHT_POSITIONAL 0x4 -#define LIGHT_NEED_VERTICES (LIGHT_POSITIONAL|LIGHT_LOCAL_VIEWER) -/*@}*/ - - -/** - * Lighting attribute group (GL_LIGHT_BIT). - */ -struct gl_light_attrib -{ - struct gl_light Light[MAX_LIGHTS]; /**< Array of light sources */ - struct gl_lightmodel Model; /**< Lighting model */ - - /** - * Must flush FLUSH_VERTICES before referencing: - */ - /*@{*/ - struct gl_material Material; /**< Includes front & back values */ - /*@}*/ - - GLboolean Enabled; /**< Lighting enabled flag */ - GLenum ShadeModel; /**< GL_FLAT or GL_SMOOTH */ - GLenum ProvokingVertex; /**< GL_EXT_provoking_vertex */ - GLenum ColorMaterialFace; /**< GL_FRONT, BACK or FRONT_AND_BACK */ - GLenum ColorMaterialMode; /**< GL_AMBIENT, GL_DIFFUSE, etc */ - GLbitfield ColorMaterialBitmask; /**< bitmask formed from Face and Mode */ - GLboolean ColorMaterialEnabled; - GLenum ClampVertexColor; - - struct gl_light EnabledList; /**< List sentinel */ - - /** - * Derived state for optimizations: - */ - /*@{*/ - GLboolean _NeedEyeCoords; - GLboolean _NeedVertices; /**< Use fast shader? */ - GLbitfield _Flags; /**< LIGHT_* flags, see above */ - GLfloat _BaseColor[2][3]; - /*@}*/ -}; - - -/** - * Line attribute group (GL_LINE_BIT). - */ -struct gl_line_attrib -{ - GLboolean SmoothFlag; /**< GL_LINE_SMOOTH enabled? */ - GLboolean StippleFlag; /**< GL_LINE_STIPPLE enabled? */ - GLushort StipplePattern; /**< Stipple pattern */ - GLint StippleFactor; /**< Stipple repeat factor */ - GLfloat Width; /**< Line width */ -}; - - -/** - * Display list attribute group (GL_LIST_BIT). - */ -struct gl_list_attrib -{ - GLuint ListBase; -}; - - -/** - * Multisample attribute group (GL_MULTISAMPLE_BIT). - */ -struct gl_multisample_attrib -{ - GLboolean Enabled; - GLboolean _Enabled; /**< true if Enabled and multisample buffer */ - GLboolean SampleAlphaToCoverage; - GLboolean SampleAlphaToOne; - GLboolean SampleCoverage; - GLfloat SampleCoverageValue; - GLboolean SampleCoverageInvert; -}; - - -/** - * A pixelmap (see glPixelMap) - */ -struct gl_pixelmap -{ - GLint Size; - GLfloat Map[MAX_PIXEL_MAP_TABLE]; - GLubyte Map8[MAX_PIXEL_MAP_TABLE]; /**< converted to 8-bit color */ -}; - - -/** - * Collection of all pixelmaps - */ -struct gl_pixelmaps -{ - struct gl_pixelmap RtoR; /**< i.e. GL_PIXEL_MAP_R_TO_R */ - struct gl_pixelmap GtoG; - struct gl_pixelmap BtoB; - struct gl_pixelmap AtoA; - struct gl_pixelmap ItoR; - struct gl_pixelmap ItoG; - struct gl_pixelmap ItoB; - struct gl_pixelmap ItoA; - struct gl_pixelmap ItoI; - struct gl_pixelmap StoS; -}; - - -/** - * Pixel attribute group (GL_PIXEL_MODE_BIT). - */ -struct gl_pixel_attrib -{ - GLenum ReadBuffer; /**< source buffer for glRead/CopyPixels() */ - - /*--- Begin Pixel Transfer State ---*/ - /* Fields are in the order in which they're applied... */ - - /** Scale & Bias (index shift, offset) */ - /*@{*/ - GLfloat RedBias, RedScale; - GLfloat GreenBias, GreenScale; - GLfloat BlueBias, BlueScale; - GLfloat AlphaBias, AlphaScale; - GLfloat DepthBias, DepthScale; - GLint IndexShift, IndexOffset; - /*@}*/ - - /* Pixel Maps */ - /* Note: actual pixel maps are not part of this attrib group */ - GLboolean MapColorFlag; - GLboolean MapStencilFlag; - - /*--- End Pixel Transfer State ---*/ - - /** glPixelZoom */ - GLfloat ZoomX, ZoomY; - - /** GL_SGI_texture_color_table */ - GLfloat TextureColorTableScale[4]; /**< RGBA */ - GLfloat TextureColorTableBias[4]; /**< RGBA */ -}; - - -/** - * Point attribute group (GL_POINT_BIT). - */ -struct gl_point_attrib -{ - GLboolean SmoothFlag; /**< True if GL_POINT_SMOOTH is enabled */ - GLfloat Size; /**< User-specified point size */ - GLfloat Params[3]; /**< GL_EXT_point_parameters */ - GLfloat MinSize, MaxSize; /**< GL_EXT_point_parameters */ - GLfloat Threshold; /**< GL_EXT_point_parameters */ - GLboolean _Attenuated; /**< True if Params != [1, 0, 0] */ - GLboolean PointSprite; /**< GL_NV/ARB_point_sprite */ - GLboolean CoordReplace[MAX_TEXTURE_COORD_UNITS]; /**< GL_ARB_point_sprite*/ - GLenum SpriteRMode; /**< GL_NV_point_sprite (only!) */ - GLenum SpriteOrigin; /**< GL_ARB_point_sprite */ -}; - - -/** - * Polygon attribute group (GL_POLYGON_BIT). - */ -struct gl_polygon_attrib -{ - GLenum FrontFace; /**< Either GL_CW or GL_CCW */ - GLenum FrontMode; /**< Either GL_POINT, GL_LINE or GL_FILL */ - GLenum BackMode; /**< Either GL_POINT, GL_LINE or GL_FILL */ - GLboolean _FrontBit; /**< 0=GL_CCW, 1=GL_CW */ - GLboolean CullFlag; /**< Culling on/off flag */ - GLboolean SmoothFlag; /**< True if GL_POLYGON_SMOOTH is enabled */ - GLboolean StippleFlag; /**< True if GL_POLYGON_STIPPLE is enabled */ - GLenum CullFaceMode; /**< Culling mode GL_FRONT or GL_BACK */ - GLfloat OffsetFactor; /**< Polygon offset factor, from user */ - GLfloat OffsetUnits; /**< Polygon offset units, from user */ - GLboolean OffsetPoint; /**< Offset in GL_POINT mode */ - GLboolean OffsetLine; /**< Offset in GL_LINE mode */ - GLboolean OffsetFill; /**< Offset in GL_FILL mode */ -}; - - -/** - * Scissor attributes (GL_SCISSOR_BIT). - */ -struct gl_scissor_attrib -{ - GLboolean Enabled; /**< Scissor test enabled? */ - GLint X, Y; /**< Lower left corner of box */ - GLsizei Width, Height; /**< Size of box */ -}; - - -/** - * Stencil attribute group (GL_STENCIL_BUFFER_BIT). - * - * Three sets of stencil data are tracked so that OpenGL 2.0, - * GL_EXT_stencil_two_side, and GL_ATI_separate_stencil can all be supported - * simultaneously. In each of the stencil state arrays, element 0 corresponds - * to GL_FRONT. Element 1 corresponds to the OpenGL 2.0 / - * GL_ATI_separate_stencil GL_BACK state. Element 2 corresponds to the - * GL_EXT_stencil_two_side GL_BACK state. - * - * The derived value \c _BackFace is either 1 or 2 depending on whether or - * not GL_STENCIL_TEST_TWO_SIDE_EXT is enabled. - * - * The derived value \c _TestTwoSide is set when the front-face and back-face - * stencil state are different. - */ -struct gl_stencil_attrib -{ - GLboolean Enabled; /**< Enabled flag */ - GLboolean TestTwoSide; /**< GL_EXT_stencil_two_side */ - GLubyte ActiveFace; /**< GL_EXT_stencil_two_side (0 or 2) */ - GLboolean _Enabled; /**< Enabled and stencil buffer present */ - GLboolean _TestTwoSide; - GLubyte _BackFace; /**< Current back stencil state (1 or 2) */ - GLenum Function[3]; /**< Stencil function */ - GLenum FailFunc[3]; /**< Fail function */ - GLenum ZPassFunc[3]; /**< Depth buffer pass function */ - GLenum ZFailFunc[3]; /**< Depth buffer fail function */ - GLint Ref[3]; /**< Reference value */ - GLuint ValueMask[3]; /**< Value mask */ - GLuint WriteMask[3]; /**< Write mask */ - GLuint Clear; /**< Clear value */ -}; - - -/** - * An index for each type of texture object. These correspond to the GL - * texture target enums, such as GL_TEXTURE_2D, GL_TEXTURE_CUBE_MAP, etc. - * Note: the order is from highest priority to lowest priority. - */ -typedef enum -{ - TEXTURE_2D_ARRAY_INDEX, - TEXTURE_1D_ARRAY_INDEX, - TEXTURE_CUBE_INDEX, - TEXTURE_3D_INDEX, - TEXTURE_RECT_INDEX, - TEXTURE_2D_INDEX, - TEXTURE_1D_INDEX, - NUM_TEXTURE_TARGETS -} gl_texture_index; - - -/** - * Bit flags for each type of texture object - * Used for Texture.Unit[]._ReallyEnabled flags. - */ -/*@{*/ -#define TEXTURE_2D_ARRAY_BIT (1 << TEXTURE_2D_ARRAY_INDEX) -#define TEXTURE_1D_ARRAY_BIT (1 << TEXTURE_1D_ARRAY_INDEX) -#define TEXTURE_CUBE_BIT (1 << TEXTURE_CUBE_INDEX) -#define TEXTURE_3D_BIT (1 << TEXTURE_3D_INDEX) -#define TEXTURE_RECT_BIT (1 << TEXTURE_RECT_INDEX) -#define TEXTURE_2D_BIT (1 << TEXTURE_2D_INDEX) -#define TEXTURE_1D_BIT (1 << TEXTURE_1D_INDEX) -/*@}*/ - - -/** - * TexGenEnabled flags. - */ -/*@{*/ -#define S_BIT 1 -#define T_BIT 2 -#define R_BIT 4 -#define Q_BIT 8 -#define STR_BITS (S_BIT | T_BIT | R_BIT) -/*@}*/ - - -/** - * Bit flag versions of the corresponding GL_ constants. - */ -/*@{*/ -#define TEXGEN_SPHERE_MAP 0x1 -#define TEXGEN_OBJ_LINEAR 0x2 -#define TEXGEN_EYE_LINEAR 0x4 -#define TEXGEN_REFLECTION_MAP_NV 0x8 -#define TEXGEN_NORMAL_MAP_NV 0x10 - -#define TEXGEN_NEED_NORMALS (TEXGEN_SPHERE_MAP | \ - TEXGEN_REFLECTION_MAP_NV | \ - TEXGEN_NORMAL_MAP_NV) -#define TEXGEN_NEED_EYE_COORD (TEXGEN_SPHERE_MAP | \ - TEXGEN_REFLECTION_MAP_NV | \ - TEXGEN_NORMAL_MAP_NV | \ - TEXGEN_EYE_LINEAR) -/*@}*/ - - - -/** Tex-gen enabled for texture unit? */ -#define ENABLE_TEXGEN(unit) (1 << (unit)) - -/** Non-identity texture matrix for texture unit? */ -#define ENABLE_TEXMAT(unit) (1 << (unit)) - - -/** - * Texel fetch function prototype. We use texel fetch functions to - * extract RGBA, color indexes and depth components out of 1D, 2D and 3D - * texture images. These functions help to isolate us from the gritty - * details of all the various texture image encodings. - * - * \param texImage texture image. - * \param col texel column. - * \param row texel row. - * \param img texel image level/layer. - * \param texelOut output texel (up to 4 GLchans) - */ -typedef void (*FetchTexelFuncC)( const struct gl_texture_image *texImage, - GLint col, GLint row, GLint img, - GLchan *texelOut ); - -/** - * As above, but returns floats. - * Used for depth component images and for upcoming signed/float - * texture images. - */ -typedef void (*FetchTexelFuncF)( const struct gl_texture_image *texImage, - GLint col, GLint row, GLint img, - GLfloat *texelOut ); - - -typedef void (*StoreTexelFunc)(struct gl_texture_image *texImage, - GLint col, GLint row, GLint img, - const void *texel); - - -/** - * Texture image state. Describes the dimensions of a texture image, - * the texel format and pointers to Texel Fetch functions. - */ -struct gl_texture_image -{ - GLint InternalFormat; /**< Internal format as given by the user */ - GLenum _BaseFormat; /**< Either GL_RGB, GL_RGBA, GL_ALPHA, - * GL_LUMINANCE, GL_LUMINANCE_ALPHA, - * GL_INTENSITY, GL_COLOR_INDEX, - * GL_DEPTH_COMPONENT or GL_DEPTH_STENCIL_EXT - * only. Used for choosing TexEnv arithmetic. - */ - GLuint TexFormat; /**< The actual format: MESA_FORMAT_x */ - - GLuint Border; /**< 0 or 1 */ - GLuint Width; /**< = 2^WidthLog2 + 2*Border */ - GLuint Height; /**< = 2^HeightLog2 + 2*Border */ - GLuint Depth; /**< = 2^DepthLog2 + 2*Border */ - GLuint Width2; /**< = Width - 2*Border */ - GLuint Height2; /**< = Height - 2*Border */ - GLuint Depth2; /**< = Depth - 2*Border */ - GLuint WidthLog2; /**< = log2(Width2) */ - GLuint HeightLog2; /**< = log2(Height2) */ - GLuint DepthLog2; /**< = log2(Depth2) */ - GLuint MaxLog2; /**< = MAX(WidthLog2, HeightLog2) */ - GLfloat WidthScale; /**< used for mipmap LOD computation */ - GLfloat HeightScale; /**< used for mipmap LOD computation */ - GLfloat DepthScale; /**< used for mipmap LOD computation */ - GLboolean IsClientData; /**< Data owned by client? */ - GLboolean _IsPowerOfTwo; /**< Are all dimensions powers of two? */ - - struct gl_texture_object *TexObject; /**< Pointer back to parent object */ - - FetchTexelFuncC FetchTexelc; /**< GLchan texel fetch function pointer */ - FetchTexelFuncF FetchTexelf; /**< Float texel fetch function pointer */ - - GLuint RowStride; /**< Padded width in units of texels */ - GLuint *ImageOffsets; /**< if 3D texture: array [Depth] of offsets to - each 2D slice in 'Data', in texels */ - GLvoid *Data; /**< Image data, accessed via FetchTexel() */ - - /** - * \name For device driver: - */ - /*@{*/ - void *DriverData; /**< Arbitrary device driver data */ - /*@}*/ -}; - - -/** - * Indexes for cube map faces. - */ -typedef enum -{ - FACE_POS_X = 0, - FACE_NEG_X = 1, - FACE_POS_Y = 2, - FACE_NEG_Y = 3, - FACE_POS_Z = 4, - FACE_NEG_Z = 5, - MAX_FACES = 6 -} gl_face_index; - - -/** - * Texture object state. Contains the array of mipmap images, border color, - * wrap modes, filter modes, shadow/texcompare state, and the per-texture - * color palette. - */ -struct gl_texture_object -{ - _glthread_Mutex Mutex; /**< for thread safety */ - GLint RefCount; /**< reference count */ - GLuint Name; /**< the user-visible texture object ID */ - GLenum Target; /**< GL_TEXTURE_1D, GL_TEXTURE_2D, etc. */ - GLfloat Priority; /**< in [0,1] */ - union { - GLfloat f[4]; - GLuint ui[4]; - GLint i[4]; - } BorderColor; /**< Interpreted according to texture format */ - GLenum WrapS; /**< S-axis texture image wrap mode */ - GLenum WrapT; /**< T-axis texture image wrap mode */ - GLenum WrapR; /**< R-axis texture image wrap mode */ - GLenum MinFilter; /**< minification filter */ - GLenum MagFilter; /**< magnification filter */ - GLfloat MinLod; /**< min lambda, OpenGL 1.2 */ - GLfloat MaxLod; /**< max lambda, OpenGL 1.2 */ - GLfloat LodBias; /**< OpenGL 1.4 */ - GLint BaseLevel; /**< min mipmap level, OpenGL 1.2 */ - GLint MaxLevel; /**< max mipmap level, OpenGL 1.2 */ - GLfloat MaxAnisotropy; /**< GL_EXT_texture_filter_anisotropic */ - GLenum CompareMode; /**< GL_ARB_shadow */ - GLenum CompareFunc; /**< GL_ARB_shadow */ - GLfloat CompareFailValue; /**< GL_ARB_shadow_ambient */ - GLenum DepthMode; /**< GL_ARB_depth_texture */ - GLint _MaxLevel; /**< actual max mipmap level (q in the spec) */ - GLfloat _MaxLambda; /**< = _MaxLevel - BaseLevel (q - b in spec) */ - GLint CropRect[4]; /**< GL_OES_draw_texture */ - GLenum Swizzle[4]; /**< GL_EXT_texture_swizzle */ - GLuint _Swizzle; /**< same as Swizzle, but SWIZZLE_* format */ - GLboolean GenerateMipmap; /**< GL_SGIS_generate_mipmap */ - GLboolean _Complete; /**< Is texture object complete? */ - GLboolean _RenderToTexture; /**< Any rendering to this texture? */ - GLboolean Purgeable; /**< Is the buffer purgeable under memory pressure? */ - GLenum sRGBDecode; /**< GL_DECODE_EXT or GL_SKIP_DECODE_EXT */ - - /** Actual texture images, indexed by [cube face] and [mipmap level] */ - struct gl_texture_image *Image[MAX_FACES][MAX_TEXTURE_LEVELS]; - - /** GL_EXT_paletted_texture */ - struct gl_color_table Palette; - - /** - * \name For device driver. - * Note: instead of attaching driver data to this pointer, it's preferable - * to instead use this struct as a base class for your own texture object - * class. Driver->NewTextureObject() can be used to implement the - * allocation. - */ - void *DriverData; /**< Arbitrary device driver data */ -}; - - -/** Up to four combiner sources are possible with GL_NV_texture_env_combine4 */ -#define MAX_COMBINER_TERMS 4 - - -/** - * Texture combine environment state. - */ -struct gl_tex_env_combine_state -{ - GLenum ModeRGB; /**< GL_REPLACE, GL_DECAL, GL_ADD, etc. */ - GLenum ModeA; /**< GL_REPLACE, GL_DECAL, GL_ADD, etc. */ - /** Source terms: GL_PRIMARY_COLOR, GL_TEXTURE, etc */ - GLenum SourceRGB[MAX_COMBINER_TERMS]; - GLenum SourceA[MAX_COMBINER_TERMS]; - /** Source operands: GL_SRC_COLOR, GL_ONE_MINUS_SRC_COLOR, etc */ - GLenum OperandRGB[MAX_COMBINER_TERMS]; - GLenum OperandA[MAX_COMBINER_TERMS]; - GLuint ScaleShiftRGB; /**< 0, 1 or 2 */ - GLuint ScaleShiftA; /**< 0, 1 or 2 */ - GLuint _NumArgsRGB; /**< Number of inputs used for the RGB combiner */ - GLuint _NumArgsA; /**< Number of inputs used for the A combiner */ -}; - - -/** - * Texture coord generation state. - */ -struct gl_texgen -{ - GLenum Mode; /**< GL_EYE_LINEAR, GL_SPHERE_MAP, etc */ - GLbitfield _ModeBit; /**< TEXGEN_x bit corresponding to Mode */ - GLfloat ObjectPlane[4]; - GLfloat EyePlane[4]; -}; - - -/** - * Texture unit state. Contains enable flags, texture environment/function/ - * combiners, texgen state, pointers to current texture objects and - * post-filter color tables. - */ -struct gl_texture_unit -{ - GLbitfield Enabled; /**< bitmask of TEXTURE_*_BIT flags */ - GLbitfield _ReallyEnabled; /**< 0 or exactly one of TEXTURE_*_BIT flags */ - - GLenum EnvMode; /**< GL_MODULATE, GL_DECAL, GL_BLEND, etc. */ - GLfloat EnvColor[4]; - - struct gl_texgen GenS; - struct gl_texgen GenT; - struct gl_texgen GenR; - struct gl_texgen GenQ; - GLbitfield TexGenEnabled; /**< Bitwise-OR of [STRQ]_BIT values */ - GLbitfield _GenFlags; /**< Bitwise-OR of Gen[STRQ]._ModeBit */ - - GLfloat LodBias; /**< for biasing mipmap levels */ - GLenum BumpTarget; - GLfloat RotMatrix[4]; /* 2x2 matrix */ - - /** - * \name GL_EXT_texture_env_combine - */ - struct gl_tex_env_combine_state Combine; - - /** - * Derived state based on \c EnvMode and the \c BaseFormat of the - * currently enabled texture. - */ - struct gl_tex_env_combine_state _EnvMode; - - /** - * Currently enabled combiner state. This will point to either - * \c Combine or \c _EnvMode. - */ - struct gl_tex_env_combine_state *_CurrentCombine; - - /** Current texture object pointers */ - struct gl_texture_object *CurrentTex[NUM_TEXTURE_TARGETS]; - - /** Points to highest priority, complete and enabled texture object */ - struct gl_texture_object *_Current; - - /** GL_SGI_texture_color_table */ - /*@{*/ - struct gl_color_table ColorTable; - struct gl_color_table ProxyColorTable; - GLboolean ColorTableEnabled; - /*@}*/ -}; - - -/** - * Texture attribute group (GL_TEXTURE_BIT). - */ -struct gl_texture_attrib -{ - GLuint CurrentUnit; /**< GL_ACTIVE_TEXTURE */ - struct gl_texture_unit Unit[MAX_COMBINED_TEXTURE_IMAGE_UNITS]; - - struct gl_texture_object *ProxyTex[NUM_TEXTURE_TARGETS]; - - /** GL_ARB_seamless_cubemap */ - GLboolean CubeMapSeamless; - - /** GL_EXT_shared_texture_palette */ - GLboolean SharedPalette; - struct gl_color_table Palette; - - /** Texture units/samplers used by vertex or fragment texturing */ - GLbitfield _EnabledUnits; - - /** Texture coord units/sets used for fragment texturing */ - GLbitfield _EnabledCoordUnits; - - /** Texture coord units that have texgen enabled */ - GLbitfield _TexGenEnabled; - - /** Texture coord units that have non-identity matrices */ - GLbitfield _TexMatEnabled; - - /** Bitwise-OR of all Texture.Unit[i]._GenFlags */ - GLbitfield _GenFlags; -}; - - -/** - * Transformation attribute group (GL_TRANSFORM_BIT). - */ -struct gl_transform_attrib -{ - GLenum MatrixMode; /**< Matrix mode */ - GLfloat EyeUserPlane[MAX_CLIP_PLANES][4]; /**< User clip planes */ - GLfloat _ClipUserPlane[MAX_CLIP_PLANES][4]; /**< derived */ - GLbitfield ClipPlanesEnabled; /**< on/off bitmask */ - GLboolean Normalize; /**< Normalize all normals? */ - GLboolean RescaleNormals; /**< GL_EXT_rescale_normal */ - GLboolean RasterPositionUnclipped; /**< GL_IBM_rasterpos_clip */ - GLboolean DepthClamp; /**< GL_ARB_depth_clamp */ - - GLfloat CullEyePos[4]; - GLfloat CullObjPos[4]; -}; - - -/** - * Viewport attribute group (GL_VIEWPORT_BIT). - */ -struct gl_viewport_attrib -{ - GLint X, Y; /**< position */ - GLsizei Width, Height; /**< size */ - GLfloat Near, Far; /**< Depth buffer range */ - GLmatrix _WindowMap; /**< Mapping transformation as a matrix. */ -}; - - -/** - * GL_ARB_vertex/pixel_buffer_object buffer object - */ -struct gl_buffer_object -{ - _glthread_Mutex Mutex; - GLint RefCount; - GLuint Name; - GLenum Usage; /**< GL_STREAM_DRAW_ARB, GL_STREAM_READ_ARB, etc. */ - GLsizeiptrARB Size; /**< Size of buffer storage in bytes */ - GLubyte *Data; /**< Location of storage either in RAM or VRAM. */ - /** Fields describing a mapped buffer */ - /*@{*/ - GLbitfield AccessFlags; /**< Mask of GL_MAP_x_BIT flags */ - GLvoid *Pointer; /**< User-space address of mapping */ - GLintptr Offset; /**< Mapped offset */ - GLsizeiptr Length; /**< Mapped length */ - /*@}*/ - GLboolean Written; /**< Ever written to? (for debugging) */ - GLboolean Purgeable; /**< Is the buffer purgeable under memory pressure? */ -}; - - -/** - * Client pixel packing/unpacking attributes - */ -struct gl_pixelstore_attrib -{ - GLint Alignment; - GLint RowLength; - GLint SkipPixels; - GLint SkipRows; - GLint ImageHeight; - GLint SkipImages; - GLboolean SwapBytes; - GLboolean LsbFirst; - GLboolean ClientStorage; /**< GL_APPLE_client_storage */ - GLboolean Invert; /**< GL_MESA_pack_invert */ - struct gl_buffer_object *BufferObj; /**< GL_ARB_pixel_buffer_object */ -}; - - -/** - * Client vertex array attributes - */ -struct gl_client_array -{ - GLint Size; /**< components per element (1,2,3,4) */ - GLenum Type; /**< datatype: GL_FLOAT, GL_INT, etc */ - GLenum Format; /**< default: GL_RGBA, but may be GL_BGRA */ - GLsizei Stride; /**< user-specified stride */ - GLsizei StrideB; /**< actual stride in bytes */ - const GLubyte *Ptr; /**< Points to array data */ - GLboolean Enabled; /**< Enabled flag is a boolean */ - GLboolean Normalized; /**< GL_ARB_vertex_program */ - GLboolean Integer; /**< Integer-valued? */ - GLuint InstanceDivisor; /**< GL_ARB_instanced_arrays */ - GLuint _ElementSize; /**< size of each element in bytes */ - - struct gl_buffer_object *BufferObj;/**< GL_ARB_vertex_buffer_object */ - GLuint _MaxElement; /**< max element index into array buffer + 1 */ -}; - - -/** - * Collection of vertex arrays. Defined by the GL_APPLE_vertex_array_object - * extension, but a nice encapsulation in any case. - */ -struct gl_array_object -{ - /** Name of the array object as received from glGenVertexArrayAPPLE. */ - GLuint Name; - - GLint RefCount; - _glthread_Mutex Mutex; - GLboolean VBOonly; /**< require all arrays to live in VBOs? */ - - /** Conventional vertex arrays */ - /*@{*/ - struct gl_client_array Vertex; - struct gl_client_array Weight; - struct gl_client_array Normal; - struct gl_client_array Color; - struct gl_client_array SecondaryColor; - struct gl_client_array FogCoord; - struct gl_client_array Index; - struct gl_client_array EdgeFlag; - struct gl_client_array TexCoord[MAX_TEXTURE_COORD_UNITS]; - struct gl_client_array PointSize; - /*@}*/ - - /** - * Generic arrays for vertex programs/shaders. - * For NV vertex programs, these attributes alias and take priority - * over the conventional attribs above. For ARB vertex programs and - * GLSL vertex shaders, these attributes are separate. - */ - struct gl_client_array VertexAttrib[MAX_VERTEX_GENERIC_ATTRIBS]; - - /** Mask of _NEW_ARRAY_* values indicating which arrays are enabled */ - GLbitfield _Enabled; - - /** - * Min of all enabled arrays' _MaxElement. When arrays reside inside VBOs - * we can determine the max legal (in bounds) glDrawElements array index. - */ - GLuint _MaxElement; -}; - - -/** - * Vertex array state - */ -struct gl_array_attrib -{ - /** Currently bound array object. See _mesa_BindVertexArrayAPPLE() */ - struct gl_array_object *ArrayObj; - - /** The default vertex array object */ - struct gl_array_object *DefaultArrayObj; - - /** Array objects (GL_ARB/APPLE_vertex_array_object) */ - struct _mesa_HashTable *Objects; - - GLint ActiveTexture; /**< Client Active Texture */ - GLuint LockFirst; /**< GL_EXT_compiled_vertex_array */ - GLuint LockCount; /**< GL_EXT_compiled_vertex_array */ - - /** GL 3.1 (slightly different from GL_NV_primitive_restart) */ - GLboolean PrimitiveRestart; - GLuint RestartIndex; - - GLbitfield NewState; /**< mask of _NEW_ARRAY_* values */ - - /* GL_ARB_vertex_buffer_object */ - struct gl_buffer_object *ArrayBufferObj; - struct gl_buffer_object *ElementArrayBufferObj; -}; - - -/** - * Feedback buffer state - */ -struct gl_feedback -{ - GLenum Type; - GLbitfield _Mask; /**< FB_* bits */ - GLfloat *Buffer; - GLuint BufferSize; - GLuint Count; -}; - - -/** - * Selection buffer state - */ -struct gl_selection -{ - GLuint *Buffer; /**< selection buffer */ - GLuint BufferSize; /**< size of the selection buffer */ - GLuint BufferCount; /**< number of values in the selection buffer */ - GLuint Hits; /**< number of records in the selection buffer */ - GLuint NameStackDepth; /**< name stack depth */ - GLuint NameStack[MAX_NAME_STACK_DEPTH]; /**< name stack */ - GLboolean HitFlag; /**< hit flag */ - GLfloat HitMinZ; /**< minimum hit depth */ - GLfloat HitMaxZ; /**< maximum hit depth */ -}; - - -/** - * 1-D Evaluator control points - */ -struct gl_1d_map -{ - GLuint Order; /**< Number of control points */ - GLfloat u1, u2, du; /**< u1, u2, 1.0/(u2-u1) */ - GLfloat *Points; /**< Points to contiguous control points */ -}; - - -/** - * 2-D Evaluator control points - */ -struct gl_2d_map -{ - GLuint Uorder; /**< Number of control points in U dimension */ - GLuint Vorder; /**< Number of control points in V dimension */ - GLfloat u1, u2, du; - GLfloat v1, v2, dv; - GLfloat *Points; /**< Points to contiguous control points */ -}; - - -/** - * All evaluator control point state - */ -struct gl_evaluators -{ - /** - * \name 1-D maps - */ - /*@{*/ - struct gl_1d_map Map1Vertex3; - struct gl_1d_map Map1Vertex4; - struct gl_1d_map Map1Index; - struct gl_1d_map Map1Color4; - struct gl_1d_map Map1Normal; - struct gl_1d_map Map1Texture1; - struct gl_1d_map Map1Texture2; - struct gl_1d_map Map1Texture3; - struct gl_1d_map Map1Texture4; - struct gl_1d_map Map1Attrib[16]; /**< GL_NV_vertex_program */ - /*@}*/ - - /** - * \name 2-D maps - */ - /*@{*/ - struct gl_2d_map Map2Vertex3; - struct gl_2d_map Map2Vertex4; - struct gl_2d_map Map2Index; - struct gl_2d_map Map2Color4; - struct gl_2d_map Map2Normal; - struct gl_2d_map Map2Texture1; - struct gl_2d_map Map2Texture2; - struct gl_2d_map Map2Texture3; - struct gl_2d_map Map2Texture4; - struct gl_2d_map Map2Attrib[16]; /**< GL_NV_vertex_program */ - /*@}*/ -}; - - -/** - * Names of the various vertex/fragment program register files, etc. - * - * NOTE: first four tokens must fit into 2 bits (see t_vb_arbprogram.c) - * All values should fit in a 4-bit field. - * - * NOTE: PROGRAM_ENV_PARAM, PROGRAM_STATE_VAR, PROGRAM_NAMED_PARAM, - * PROGRAM_CONSTANT, and PROGRAM_UNIFORM can all be considered to - * be "uniform" variables since they can only be set outside glBegin/End. - * They're also all stored in the same Parameters array. - */ -typedef enum -{ - PROGRAM_TEMPORARY, /**< machine->Temporary[] */ - PROGRAM_INPUT, /**< machine->Inputs[] */ - PROGRAM_OUTPUT, /**< machine->Outputs[] */ - PROGRAM_VARYING, /**< machine->Inputs[]/Outputs[] */ - PROGRAM_LOCAL_PARAM, /**< gl_program->LocalParams[] */ - PROGRAM_ENV_PARAM, /**< gl_program->Parameters[] */ - PROGRAM_STATE_VAR, /**< gl_program->Parameters[] */ - PROGRAM_NAMED_PARAM, /**< gl_program->Parameters[] */ - PROGRAM_CONSTANT, /**< gl_program->Parameters[] */ - PROGRAM_UNIFORM, /**< gl_program->Parameters[] */ - PROGRAM_WRITE_ONLY, /**< A dummy, write-only register */ - PROGRAM_ADDRESS, /**< machine->AddressReg */ - PROGRAM_SAMPLER, /**< for shader samplers, compile-time only */ - PROGRAM_SYSTEM_VALUE,/**< InstanceId, PrimitiveID, etc. */ - PROGRAM_UNDEFINED, /**< Invalid/TBD value */ - PROGRAM_FILE_MAX -} gl_register_file; - - -/** - * If the register file is PROGRAM_SYSTEM_VALUE, the register index will be - * one of these values. - */ -typedef enum -{ - SYSTEM_VALUE_FRONT_FACE, /**< Fragment shader only (not done yet) */ - SYSTEM_VALUE_INSTANCE_ID, /**< Vertex shader only */ - SYSTEM_VALUE_MAX /**< Number of values */ -} gl_system_value; - - -/** Vertex and fragment instructions */ -struct prog_instruction; -struct gl_program_parameter_list; -struct gl_uniform_list; - - -/** - * Base class for any kind of program object - */ -struct gl_program -{ - GLuint Id; - GLubyte *String; /**< Null-terminated program text */ - GLint RefCount; - GLenum Target; /**< GL_VERTEX/FRAGMENT_PROGRAM_ARB, GL_FRAGMENT_PROGRAM_NV */ - GLenum Format; /**< String encoding format */ - GLboolean Resident; - - struct prog_instruction *Instructions; - - GLbitfield InputsRead; /**< Bitmask of which input regs are read */ - GLbitfield64 OutputsWritten; /**< Bitmask of which output regs are written */ - GLbitfield SystemValuesRead; /**< Bitmask of SYSTEM_VALUE_x inputs used */ - GLbitfield InputFlags[MAX_PROGRAM_INPUTS]; /**< PROG_PARAM_BIT_x flags */ - GLbitfield OutputFlags[MAX_PROGRAM_OUTPUTS]; /**< PROG_PARAM_BIT_x flags */ - GLbitfield TexturesUsed[MAX_TEXTURE_UNITS]; /**< TEXTURE_x_BIT bitmask */ - GLbitfield SamplersUsed; /**< Bitfield of which samplers are used */ - GLbitfield ShadowSamplers; /**< Texture units used for shadow sampling. */ - - - /** Named parameters, constants, etc. from program text */ - struct gl_program_parameter_list *Parameters; - /** Numbered local parameters */ - GLfloat LocalParams[MAX_PROGRAM_LOCAL_PARAMS][4]; - - /** Vertex/fragment shader varying vars */ - struct gl_program_parameter_list *Varying; - /** Vertex program user-defined attributes */ - struct gl_program_parameter_list *Attributes; - - /** Map from sampler unit to texture unit (set by glUniform1i()) */ - GLubyte SamplerUnits[MAX_SAMPLERS]; - /** Which texture target is being sampled (TEXTURE_1D/2D/3D/etc_INDEX) */ - gl_texture_index SamplerTargets[MAX_SAMPLERS]; - - /** Bitmask of which register files are read/written with indirect - * addressing. Mask of (1 << PROGRAM_x) bits. - */ - GLbitfield IndirectRegisterFiles; - - /** Logical counts */ - /*@{*/ - GLuint NumInstructions; - GLuint NumTemporaries; - GLuint NumParameters; - GLuint NumAttributes; - GLuint NumAddressRegs; - GLuint NumAluInstructions; - GLuint NumTexInstructions; - GLuint NumTexIndirections; - /*@}*/ - /** Native, actual h/w counts */ - /*@{*/ - GLuint NumNativeInstructions; - GLuint NumNativeTemporaries; - GLuint NumNativeParameters; - GLuint NumNativeAttributes; - GLuint NumNativeAddressRegs; - GLuint NumNativeAluInstructions; - GLuint NumNativeTexInstructions; - GLuint NumNativeTexIndirections; - /*@}*/ -}; - - -/** Vertex program object */ -struct gl_vertex_program -{ - struct gl_program Base; /**< base class */ - GLboolean IsNVProgram; /**< is this a GL_NV_vertex_program program? */ - GLboolean IsPositionInvariant; -}; - - -/** Geometry program object */ -struct gl_geometry_program -{ - struct gl_program Base; /**< base class */ - - GLint VerticesOut; - GLenum InputType; /**< GL_POINTS, GL_LINES, GL_LINES_ADJACENCY_ARB, - GL_TRIANGLES, or GL_TRIANGLES_ADJACENCY_ARB */ - GLenum OutputType; /**< GL_POINTS, GL_LINE_STRIP or GL_TRIANGLE_STRIP */ -}; - - -/** Fragment program object */ -struct gl_fragment_program -{ - struct gl_program Base; /**< base class */ - GLenum FogOption; - GLboolean UsesKill; /**< shader uses KIL instruction */ - GLboolean OriginUpperLeft; - GLboolean PixelCenterInteger; - enum gl_frag_depth_layout FragDepthLayout; -}; - - -/** - * State common to vertex and fragment programs. - */ -struct gl_program_state -{ - GLint ErrorPos; /* GL_PROGRAM_ERROR_POSITION_ARB/NV */ - const char *ErrorString; /* GL_PROGRAM_ERROR_STRING_ARB/NV */ -}; - - -/** - * Context state for vertex programs. - */ -struct gl_vertex_program_state -{ - GLboolean Enabled; /**< User-set GL_VERTEX_PROGRAM_ARB/NV flag */ - GLboolean _Enabled; /**< Enabled and _valid_ user program? */ - GLboolean PointSizeEnabled; /**< GL_VERTEX_PROGRAM_POINT_SIZE_ARB/NV */ - GLboolean TwoSideEnabled; /**< GL_VERTEX_PROGRAM_TWO_SIDE_ARB/NV */ - struct gl_vertex_program *Current; /**< User-bound vertex program */ - - /** Currently enabled and valid vertex program (including internal - * programs, user-defined vertex programs and GLSL vertex shaders). - * This is the program we must use when rendering. - */ - struct gl_vertex_program *_Current; - - GLfloat Parameters[MAX_PROGRAM_ENV_PARAMS][4]; /**< Env params */ - - /* For GL_NV_vertex_program only: */ - GLenum TrackMatrix[MAX_PROGRAM_ENV_PARAMS / 4]; - GLenum TrackMatrixTransform[MAX_PROGRAM_ENV_PARAMS / 4]; - - /** Should fixed-function T&L be implemented with a vertex prog? */ - GLboolean _MaintainTnlProgram; - - /** Program to emulate fixed-function T&L (see above) */ - struct gl_vertex_program *_TnlProgram; - - /** Cache of fixed-function programs */ - struct gl_program_cache *Cache; - - GLboolean _Overriden; -}; - - -/** - * Context state for geometry programs. - */ -struct gl_geometry_program_state -{ - GLboolean Enabled; /**< GL_ARB_GEOMETRY_SHADER4 */ - GLboolean _Enabled; /**< Enabled and valid program? */ - struct gl_geometry_program *Current; /**< user-bound geometry program */ - - /** Currently enabled and valid program (including internal programs - * and compiled shader programs). - */ - struct gl_geometry_program *_Current; - - GLfloat Parameters[MAX_PROGRAM_ENV_PARAMS][4]; /**< Env params */ - - /** Cache of fixed-function programs */ - struct gl_program_cache *Cache; -}; - -/** - * Context state for fragment programs. - */ -struct gl_fragment_program_state -{ - GLboolean Enabled; /**< User-set fragment program enable flag */ - GLboolean _Enabled; /**< Enabled and _valid_ user program? */ - struct gl_fragment_program *Current; /**< User-bound fragment program */ - - /** Currently enabled and valid fragment program (including internal - * programs, user-defined fragment programs and GLSL fragment shaders). - * This is the program we must use when rendering. - */ - struct gl_fragment_program *_Current; - - GLfloat Parameters[MAX_PROGRAM_ENV_PARAMS][4]; /**< Env params */ - - /** Should fixed-function texturing be implemented with a fragment prog? */ - GLboolean _MaintainTexEnvProgram; - - /** Program to emulate fixed-function texture env/combine (see above) */ - struct gl_fragment_program *_TexEnvProgram; - - /** Cache of fixed-function programs */ - struct gl_program_cache *Cache; -}; - - -/** - * ATI_fragment_shader runtime state - */ -#define ATI_FS_INPUT_PRIMARY 0 -#define ATI_FS_INPUT_SECONDARY 1 - -struct atifs_instruction; -struct atifs_setupinst; - -/** - * ATI fragment shader - */ -struct ati_fragment_shader -{ - GLuint Id; - GLint RefCount; - struct atifs_instruction *Instructions[2]; - struct atifs_setupinst *SetupInst[2]; - GLfloat Constants[8][4]; - GLbitfield LocalConstDef; /**< Indicates which constants have been set */ - GLubyte numArithInstr[2]; - GLubyte regsAssigned[2]; - GLubyte NumPasses; /**< 1 or 2 */ - GLubyte cur_pass; - GLubyte last_optype; - GLboolean interpinp1; - GLboolean isValid; - GLuint swizzlerq; -}; - -/** - * Context state for GL_ATI_fragment_shader - */ -struct gl_ati_fragment_shader_state -{ - GLboolean Enabled; - GLboolean _Enabled; /**< enabled and valid shader? */ - GLboolean Compiling; - GLfloat GlobalConstants[8][4]; - struct ati_fragment_shader *Current; -}; - - -/** - * Occlusion/timer query object. - */ -struct gl_query_object -{ - GLenum Target; /**< The query target, when active */ - GLuint Id; /**< hash table ID/name */ - GLuint64EXT Result; /**< the counter */ - GLboolean Active; /**< inside Begin/EndQuery */ - GLboolean Ready; /**< result is ready? */ -}; - - -/** - * Context state for query objects. - */ -struct gl_query_state -{ - struct _mesa_HashTable *QueryObjects; - struct gl_query_object *CurrentOcclusionObject; /* GL_ARB_occlusion_query */ - struct gl_query_object *CurrentTimerObject; /* GL_EXT_timer_query */ - - /** GL_NV_conditional_render */ - struct gl_query_object *CondRenderQuery; - - /** GL_EXT_transform_feedback */ - struct gl_query_object *PrimitivesGenerated; - struct gl_query_object *PrimitivesWritten; - - /** GL_ARB_timer_query */ - struct gl_query_object *TimeElapsed; - - GLenum CondRenderMode; -}; - - -/** Sync object state */ -struct gl_sync_object { - struct simple_node link; - GLenum Type; /**< GL_SYNC_FENCE */ - GLuint Name; /**< Fence name */ - GLint RefCount; /**< Reference count */ - GLboolean DeletePending; /**< Object was deleted while there were still - * live references (e.g., sync not yet finished) - */ - GLenum SyncCondition; - GLbitfield Flags; /**< Flags passed to glFenceSync */ - GLuint StatusFlag:1; /**< Has the sync object been signaled? */ -}; - - -/** Set by #pragma directives */ -struct gl_sl_pragmas -{ - GLboolean IgnoreOptimize; /**< ignore #pragma optimize(on/off) ? */ - GLboolean IgnoreDebug; /**< ignore #pragma debug(on/off) ? */ - GLboolean Optimize; /**< defaults on */ - GLboolean Debug; /**< defaults off */ -}; - - -/** - * A GLSL vertex or fragment shader object. - */ -struct gl_shader -{ - GLenum Type; /**< GL_FRAGMENT_SHADER || GL_VERTEX_SHADER || GL_GEOMETRY_SHADER_ARB (first field!) */ - GLuint Name; /**< AKA the handle */ - GLint RefCount; /**< Reference count */ - GLboolean DeletePending; - GLboolean CompileStatus; - GLboolean Main; /**< shader defines main() */ - GLboolean UnresolvedRefs; - const GLchar *Source; /**< Source code string */ - GLuint SourceChecksum; /**< for debug/logging purposes */ - struct gl_program *Program; /**< Post-compile assembly code */ - GLchar *InfoLog; - struct gl_sl_pragmas Pragmas; - - unsigned Version; /**< GLSL version used for linking */ - - struct exec_list *ir; - struct glsl_symbol_table *symbols; - - /** Shaders containing built-in functions that are used for linking. */ - struct gl_shader *builtins_to_link[16]; - unsigned num_builtins_to_link; -}; - - -/** - * A GLSL program object. - * Basically a linked collection of vertex and fragment shaders. - */ -struct gl_shader_program -{ - GLenum Type; /**< Always GL_SHADER_PROGRAM (internal token) */ - GLuint Name; /**< aka handle or ID */ - GLint RefCount; /**< Reference count */ - GLboolean DeletePending; - - GLuint NumShaders; /**< number of attached shaders */ - struct gl_shader **Shaders; /**< List of attached the shaders */ - - /** User-defined attribute bindings (glBindAttribLocation) */ - struct gl_program_parameter_list *Attributes; - - /** Transform feedback varyings */ - struct { - GLenum BufferMode; - GLuint NumVarying; - GLchar **VaryingNames; /**< Array [NumVarying] of char * */ - } TransformFeedback; - - /** Geometry shader state - copied into gl_geometry_program at link time */ - struct { - GLint VerticesOut; - GLenum InputType; /**< GL_POINTS, GL_LINES, GL_LINES_ADJACENCY_ARB, - GL_TRIANGLES, or GL_TRIANGLES_ADJACENCY_ARB */ - GLenum OutputType; /**< GL_POINTS, GL_LINE_STRIP or GL_TRIANGLE_STRIP */ - } Geom; - - /* post-link info: */ - struct gl_vertex_program *VertexProgram; /**< Linked vertex program */ - struct gl_fragment_program *FragmentProgram; /**< Linked fragment prog */ - struct gl_geometry_program *GeometryProgram; /**< Linked geometry prog */ - struct gl_uniform_list *Uniforms; - struct gl_program_parameter_list *Varying; - GLboolean LinkStatus; /**< GL_LINK_STATUS */ - GLboolean Validated; - GLboolean _Used; /**< Ever used for drawing? */ - GLchar *InfoLog; - - unsigned Version; /**< GLSL version used for linking */ - - /** - * Per-stage shaders resulting from the first stage of linking. - * - * Set of linked shaders for this program. The array is accessed using the - * \c MESA_SHADER_* defines. Entries for non-existent stages will be - * \c NULL. - */ - struct gl_shader *_LinkedShaders[MESA_SHADER_TYPES]; -}; - - -#define GLSL_DUMP 0x1 /**< Dump shaders to stdout */ -#define GLSL_LOG 0x2 /**< Write shaders to files */ -#define GLSL_OPT 0x4 /**< Force optimizations (override pragmas) */ -#define GLSL_NO_OPT 0x8 /**< Force no optimizations (override pragmas) */ -#define GLSL_UNIFORMS 0x10 /**< Print glUniform calls */ -#define GLSL_NOP_VERT 0x20 /**< Force no-op vertex shaders */ -#define GLSL_NOP_FRAG 0x40 /**< Force no-op fragment shaders */ -#define GLSL_USE_PROG 0x80 /**< Log glUseProgram calls */ - - -/** - * Context state for GLSL vertex/fragment shaders. - */ -struct gl_shader_state -{ - /** - * Programs used for rendering - * - * There is a separate program set for each shader stage. If - * GL_EXT_separate_shader_objects is not supported, each of these must point - * to \c NULL or to the same program. - */ - struct gl_shader_program *CurrentVertexProgram; - struct gl_shader_program *CurrentGeometryProgram; - struct gl_shader_program *CurrentFragmentProgram; - - /** - * Program used by glUniform calls. - * - * Explicitly set by \c glUseProgram and \c glActiveProgramEXT. - */ - struct gl_shader_program *ActiveProgram; - - void *MemPool; - - GLbitfield Flags; /**< Mask of GLSL_x flags */ -}; - -/** - * Compiler options for a single GLSL shaders type - */ -struct gl_shader_compiler_options -{ - /** Driver-selectable options: */ - GLboolean EmitCondCodes; /**< Use condition codes? */ - GLboolean EmitNVTempInitialization; /**< 0-fill NV temp registers */ - /** - * Attempts to flatten all ir_if (OPCODE_IF) for GPUs that can't - * support control flow. - */ - GLboolean EmitNoIfs; - GLboolean EmitNoLoops; - GLboolean EmitNoFunctions; - GLboolean EmitNoCont; /**< Emit CONT opcode? */ - GLboolean EmitNoMainReturn; /**< Emit CONT/RET opcodes? */ - GLboolean EmitNoNoise; /**< Emit NOISE opcodes? */ - GLboolean EmitNoPow; /**< Emit POW opcodes? */ - - /** - * \name Forms of indirect addressing the driver cannot do. - */ - /*@{*/ - GLboolean EmitNoIndirectInput; /**< No indirect addressing of inputs */ - GLboolean EmitNoIndirectOutput; /**< No indirect addressing of outputs */ - GLboolean EmitNoIndirectTemp; /**< No indirect addressing of temps */ - GLboolean EmitNoIndirectUniform; /**< No indirect addressing of constants */ - /*@}*/ - - GLuint MaxUnrollIterations; - - struct gl_sl_pragmas DefaultPragmas; /**< Default #pragma settings */ -}; - -/** - * Transform feedback object state - */ -struct gl_transform_feedback_object -{ - GLuint Name; /**< AKA the object ID */ - GLint RefCount; - GLboolean Active; /**< Is transform feedback enabled? */ - GLboolean Paused; /**< Is transform feedback paused? */ - - /** The feedback buffers */ - GLuint BufferNames[MAX_FEEDBACK_ATTRIBS]; - struct gl_buffer_object *Buffers[MAX_FEEDBACK_ATTRIBS]; - - /** Start of feedback data in dest buffer */ - GLintptr Offset[MAX_FEEDBACK_ATTRIBS]; - /** Max data to put into dest buffer (in bytes) */ - GLsizeiptr Size[MAX_FEEDBACK_ATTRIBS]; -}; - - -/** - * Context state for transform feedback. - */ -struct gl_transform_feedback -{ - GLenum Mode; /**< GL_POINTS, GL_LINES or GL_TRIANGLES */ - - GLboolean RasterDiscard; /**< GL_RASTERIZER_DISCARD */ - - /** The general binding point (GL_TRANSFORM_FEEDBACK_BUFFER) */ - struct gl_buffer_object *CurrentBuffer; - - /** The table of all transform feedback objects */ - struct _mesa_HashTable *Objects; - - /** The current xform-fb object (GL_TRANSFORM_FEEDBACK_BINDING) */ - struct gl_transform_feedback_object *CurrentObject; - - /** The default xform-fb object (Name==0) */ - struct gl_transform_feedback_object *DefaultObject; -}; - - - -/** - * State which can be shared by multiple contexts: - */ -struct gl_shared_state -{ - _glthread_Mutex Mutex; /**< for thread safety */ - GLint RefCount; /**< Reference count */ - struct _mesa_HashTable *DisplayList; /**< Display lists hash table */ - struct _mesa_HashTable *TexObjects; /**< Texture objects hash table */ - - /** Default texture objects (shared by all texture units) */ - struct gl_texture_object *DefaultTex[NUM_TEXTURE_TARGETS]; - - /** Fallback texture used when a bound texture is incomplete */ - struct gl_texture_object *FallbackTex; - - /** - * \name Thread safety and statechange notification for texture - * objects. - * - * \todo Improve the granularity of locking. - */ - /*@{*/ - _glthread_Mutex TexMutex; /**< texobj thread safety */ - GLuint TextureStateStamp; /**< state notification for shared tex */ - /*@}*/ - - /** Default buffer object for vertex arrays that aren't in VBOs */ - struct gl_buffer_object *NullBufferObj; - - /** - * \name Vertex/geometry/fragment programs - */ - /*@{*/ - struct _mesa_HashTable *Programs; /**< All vertex/fragment programs */ - struct gl_vertex_program *DefaultVertexProgram; - struct gl_fragment_program *DefaultFragmentProgram; - struct gl_geometry_program *DefaultGeometryProgram; - /*@}*/ - - /* GL_ATI_fragment_shader */ - struct _mesa_HashTable *ATIShaders; - struct ati_fragment_shader *DefaultFragmentShader; - - struct _mesa_HashTable *BufferObjects; - - /** Table of both gl_shader and gl_shader_program objects */ - struct _mesa_HashTable *ShaderObjects; - - /* GL_EXT_framebuffer_object */ - struct _mesa_HashTable *RenderBuffers; - struct _mesa_HashTable *FrameBuffers; - - /* GL_ARB_sync */ - struct simple_node SyncObjects; - - void *DriverData; /**< Device driver shared state */ -}; - - - - -/** - * A renderbuffer stores colors or depth values or stencil values. - * A framebuffer object will have a collection of these. - * Data are read/written to the buffer with a handful of Get/Put functions. - * - * Instances of this object are allocated with the Driver's NewRenderbuffer - * hook. Drivers will likely wrap this class inside a driver-specific - * class to simulate inheritance. - */ -struct gl_renderbuffer -{ -#define RB_MAGIC 0xaabbccdd - int Magic; /** XXX TEMPORARY DEBUG INFO */ - _glthread_Mutex Mutex; /**< for thread safety */ - GLuint ClassID; /**< Useful for drivers */ - GLuint Name; - GLint RefCount; - GLuint Width, Height; - GLboolean Purgeable; /**< Is the buffer purgeable under memory pressure? */ - - GLenum InternalFormat; /**< The user-specified format */ - GLenum _BaseFormat; /**< Either GL_RGB, GL_RGBA, GL_DEPTH_COMPONENT or - GL_STENCIL_INDEX. */ - GLuint Format; /**< The actual format: MESA_FORMAT_x */ - - GLubyte NumSamples; - - GLenum DataType; /**< Type of values passed to the Get/Put functions */ - GLvoid *Data; /**< This may not be used by some kinds of RBs */ - - /* Used to wrap one renderbuffer around another: */ - struct gl_renderbuffer *Wrapped; - - /* Delete this renderbuffer */ - void (*Delete)(struct gl_renderbuffer *rb); - - /* Allocate new storage for this renderbuffer */ - GLboolean (*AllocStorage)(struct gl_context *ctx, struct gl_renderbuffer *rb, - GLenum internalFormat, - GLuint width, GLuint height); - - /* Lock/Unlock are called before/after calling the Get/Put functions. - * Not sure this is the right place for these yet. - void (*Lock)(struct gl_context *ctx, struct gl_renderbuffer *rb); - void (*Unlock)(struct gl_context *ctx, struct gl_renderbuffer *rb); - */ - - /* Return a pointer to the element/pixel at (x,y). - * Should return NULL if the buffer memory can't be directly addressed. - */ - void *(*GetPointer)(struct gl_context *ctx, struct gl_renderbuffer *rb, - GLint x, GLint y); - - /* Get/Read a row of values. - * The values will be of format _BaseFormat and type DataType. - */ - void (*GetRow)(struct gl_context *ctx, struct gl_renderbuffer *rb, GLuint count, - GLint x, GLint y, void *values); - - /* Get/Read values at arbitrary locations. - * The values will be of format _BaseFormat and type DataType. - */ - void (*GetValues)(struct gl_context *ctx, struct gl_renderbuffer *rb, GLuint count, - const GLint x[], const GLint y[], void *values); - - /* Put/Write a row of values. - * The values will be of format _BaseFormat and type DataType. - */ - void (*PutRow)(struct gl_context *ctx, struct gl_renderbuffer *rb, GLuint count, - GLint x, GLint y, const void *values, const GLubyte *mask); - - /* Put/Write a row of RGB values. This is a special-case routine that's - * only used for RGBA renderbuffers when the source data is GL_RGB. That's - * a common case for glDrawPixels and some triangle routines. - * The values will be of format GL_RGB and type DataType. - */ - void (*PutRowRGB)(struct gl_context *ctx, struct gl_renderbuffer *rb, GLuint count, - GLint x, GLint y, const void *values, const GLubyte *mask); - - - /* Put/Write a row of identical values. - * The values will be of format _BaseFormat and type DataType. - */ - void (*PutMonoRow)(struct gl_context *ctx, struct gl_renderbuffer *rb, GLuint count, - GLint x, GLint y, const void *value, const GLubyte *mask); - - /* Put/Write values at arbitrary locations. - * The values will be of format _BaseFormat and type DataType. - */ - void (*PutValues)(struct gl_context *ctx, struct gl_renderbuffer *rb, GLuint count, - const GLint x[], const GLint y[], const void *values, - const GLubyte *mask); - /* Put/Write identical values at arbitrary locations. - * The values will be of format _BaseFormat and type DataType. - */ - void (*PutMonoValues)(struct gl_context *ctx, struct gl_renderbuffer *rb, - GLuint count, const GLint x[], const GLint y[], - const void *value, const GLubyte *mask); -}; - - -/** - * A renderbuffer attachment points to either a texture object (and specifies - * a mipmap level, cube face or 3D texture slice) or points to a renderbuffer. - */ -struct gl_renderbuffer_attachment -{ - GLenum Type; /**< \c GL_NONE or \c GL_TEXTURE or \c GL_RENDERBUFFER_EXT */ - GLboolean Complete; - - /** - * If \c Type is \c GL_RENDERBUFFER_EXT, this stores a pointer to the - * application supplied renderbuffer object. - */ - struct gl_renderbuffer *Renderbuffer; - - /** - * If \c Type is \c GL_TEXTURE, this stores a pointer to the application - * supplied texture object. - */ - struct gl_texture_object *Texture; - GLuint TextureLevel; /**< Attached mipmap level. */ - GLuint CubeMapFace; /**< 0 .. 5, for cube map textures. */ - GLuint Zoffset; /**< Slice for 3D textures, or layer for both 1D - * and 2D array textures */ -}; - - -/** - * A framebuffer is a collection of renderbuffers (color, depth, stencil, etc). - * In C++ terms, think of this as a base class from which device drivers - * will make derived classes. - */ -struct gl_framebuffer -{ - _glthread_Mutex Mutex; /**< for thread safety */ - /** - * If zero, this is a window system framebuffer. If non-zero, this - * is a FBO framebuffer; note that for some devices (i.e. those with - * a natural pixel coordinate system for FBOs that differs from the - * OpenGL/Mesa coordinate system), this means that the viewport, - * polygon face orientation, and polygon stipple will have to be inverted. - */ - GLuint Name; - - GLint RefCount; - GLboolean DeletePending; - - /** - * The framebuffer's visual. Immutable if this is a window system buffer. - * Computed from attachments if user-made FBO. - */ - struct gl_config Visual; - - GLboolean Initialized; - - GLuint Width, Height; /**< size of frame buffer in pixels */ - - /** \name Drawing bounds (Intersection of buffer size and scissor box) */ - /*@{*/ - GLint _Xmin, _Xmax; /**< inclusive */ - GLint _Ymin, _Ymax; /**< exclusive */ - /*@}*/ - - /** \name Derived Z buffer stuff */ - /*@{*/ - GLuint _DepthMax; /**< Max depth buffer value */ - GLfloat _DepthMaxF; /**< Float max depth buffer value */ - GLfloat _MRD; /**< minimum resolvable difference in Z values */ - /*@}*/ - - /** One of the GL_FRAMEBUFFER_(IN)COMPLETE_* tokens */ - GLenum _Status; - - /** Integer color values */ - GLboolean _IntegerColor; - - /** Array of all renderbuffer attachments, indexed by BUFFER_* tokens. */ - struct gl_renderbuffer_attachment Attachment[BUFFER_COUNT]; - - /* In unextended OpenGL these vars are part of the GL_COLOR_BUFFER - * attribute group and GL_PIXEL attribute group, respectively. - */ - GLenum ColorDrawBuffer[MAX_DRAW_BUFFERS]; - GLenum ColorReadBuffer; - - /** Computed from ColorDraw/ReadBuffer above */ - GLuint _NumColorDrawBuffers; - GLint _ColorDrawBufferIndexes[MAX_DRAW_BUFFERS]; /**< BUFFER_x or -1 */ - GLint _ColorReadBufferIndex; /* -1 = None */ - struct gl_renderbuffer *_ColorDrawBuffers[MAX_DRAW_BUFFERS]; - struct gl_renderbuffer *_ColorReadBuffer; - - /** The Actual depth/stencil buffers to use. May be wrappers around the - * depth/stencil buffers attached above. */ - struct gl_renderbuffer *_DepthBuffer; - struct gl_renderbuffer *_StencilBuffer; - - /** Delete this framebuffer */ - void (*Delete)(struct gl_framebuffer *fb); -}; - - -/** - * Precision info for shader datatypes. See glGetShaderPrecisionFormat(). - */ -struct gl_precision -{ - GLushort RangeMin; /**< min value exponent */ - GLushort RangeMax; /**< max value exponent */ - GLushort Precision; /**< number of mantissa bits */ -}; - - -/** - * Limits for vertex and fragment programs/shaders. - */ -struct gl_program_constants -{ - /* logical limits */ - GLuint MaxInstructions; - GLuint MaxAluInstructions; - GLuint MaxTexInstructions; - GLuint MaxTexIndirections; - GLuint MaxAttribs; - GLuint MaxTemps; - GLuint MaxAddressRegs; - GLuint MaxParameters; - GLuint MaxLocalParams; - GLuint MaxEnvParams; - /* native/hardware limits */ - GLuint MaxNativeInstructions; - GLuint MaxNativeAluInstructions; - GLuint MaxNativeTexInstructions; - GLuint MaxNativeTexIndirections; - GLuint MaxNativeAttribs; - GLuint MaxNativeTemps; - GLuint MaxNativeAddressRegs; - GLuint MaxNativeParameters; - /* For shaders */ - GLuint MaxUniformComponents; - /* GL_ARB_geometry_shader4 */ - GLuint MaxGeometryTextureImageUnits; - GLuint MaxGeometryVaryingComponents; - GLuint MaxVertexVaryingComponents; - GLuint MaxGeometryUniformComponents; - GLuint MaxGeometryOutputVertices; - GLuint MaxGeometryTotalOutputComponents; - /* ES 2.0 and GL_ARB_ES2_compatibility */ - struct gl_precision LowFloat, MediumFloat, HighFloat; - struct gl_precision LowInt, MediumInt, HighInt; -}; - - -/** - * Constants which may be overridden by device driver during context creation - * but are never changed after that. - */ -struct gl_constants -{ - GLint MaxTextureMbytes; /**< Max memory per image, in MB */ - GLint MaxTextureLevels; /**< Max mipmap levels. */ - GLint Max3DTextureLevels; /**< Max mipmap levels for 3D textures */ - GLint MaxCubeTextureLevels; /**< Max mipmap levels for cube textures */ - GLint MaxArrayTextureLayers; /**< Max layers in array textures */ - GLint MaxTextureRectSize; /**< Max rectangle texture size, in pixes */ - GLuint MaxTextureCoordUnits; - GLuint MaxTextureImageUnits; - GLuint MaxVertexTextureImageUnits; - GLuint MaxCombinedTextureImageUnits; - GLuint MaxTextureUnits; /**< = MIN(CoordUnits, ImageUnits) */ - GLfloat MaxTextureMaxAnisotropy; /**< GL_EXT_texture_filter_anisotropic */ - GLfloat MaxTextureLodBias; /**< GL_EXT_texture_lod_bias */ - - GLuint MaxArrayLockSize; - - GLint SubPixelBits; - - GLfloat MinPointSize, MaxPointSize; /**< aliased */ - GLfloat MinPointSizeAA, MaxPointSizeAA; /**< antialiased */ - GLfloat PointSizeGranularity; - GLfloat MinLineWidth, MaxLineWidth; /**< aliased */ - GLfloat MinLineWidthAA, MaxLineWidthAA; /**< antialiased */ - GLfloat LineWidthGranularity; - - GLuint MaxColorTableSize; - - GLuint MaxClipPlanes; - GLuint MaxLights; - GLfloat MaxShininess; /**< GL_NV_light_max_exponent */ - GLfloat MaxSpotExponent; /**< GL_NV_light_max_exponent */ - - GLuint MaxViewportWidth, MaxViewportHeight; - - struct gl_program_constants VertexProgram; /**< GL_ARB_vertex_program */ - struct gl_program_constants FragmentProgram; /**< GL_ARB_fragment_program */ - struct gl_program_constants GeometryProgram; /**< GL_ARB_geometry_shader4 */ - GLuint MaxProgramMatrices; - GLuint MaxProgramMatrixStackDepth; - - /** vertex array / buffer object bounds checking */ - GLboolean CheckArrayBounds; - - GLuint MaxDrawBuffers; /**< GL_ARB_draw_buffers */ - - GLuint MaxColorAttachments; /**< GL_EXT_framebuffer_object */ - GLuint MaxRenderbufferSize; /**< GL_EXT_framebuffer_object */ - GLuint MaxSamples; /**< GL_ARB_framebuffer_object */ - - GLuint MaxVarying; /**< Number of float[4] varying parameters */ - - GLuint GLSLVersion; /**< GLSL version supported (ex: 120 = 1.20) */ - - /** Which texture units support GL_ATI_envmap_bumpmap as targets */ - GLbitfield SupportedBumpUnits; - - /** - * Maximum amount of time, measured in nanseconds, that the server can wait. - */ - GLuint64 MaxServerWaitTimeout; - - /** GL_EXT_provoking_vertex */ - GLboolean QuadsFollowProvokingVertexConvention; - - /** OpenGL version 3.0 */ - GLbitfield ContextFlags; /**< Ex: GL_CONTEXT_FLAG_FORWARD_COMPATIBLE_BIT */ - - /** OpenGL version 3.2 */ - GLbitfield ProfileMask; /**< Mask of CONTEXT_x_PROFILE_BIT */ - - /** GL_EXT_transform_feedback */ - GLuint MaxTransformFeedbackSeparateAttribs; - GLuint MaxTransformFeedbackSeparateComponents; - GLuint MaxTransformFeedbackInterleavedComponents; - - /** GL_EXT_gpu_shader4 */ - GLint MinProgramTexelOffset, MaxProgramTexelOffset; - - /* GL_EXT_framebuffer_sRGB */ - GLboolean sRGBCapable; /* can enable sRGB blend/update on FBOs */ -}; - - -/** - * Enable flag for each OpenGL extension. Different device drivers will - * enable different extensions at runtime. - */ -struct gl_extensions -{ - GLboolean dummy; /* don't remove this! */ - GLboolean dummy_true; /* Set true by _mesa_init_extensions(). */ - GLboolean dummy_false; /* Set false by _mesa_init_extensions(). */ - GLboolean ARB_ES2_compatibility; - GLboolean ARB_blend_func_extended; - GLboolean ARB_copy_buffer; - GLboolean ARB_depth_buffer_float; - GLboolean ARB_depth_clamp; - GLboolean ARB_depth_texture; - GLboolean ARB_draw_buffers; - GLboolean ARB_draw_buffers_blend; - GLboolean ARB_draw_elements_base_vertex; - GLboolean ARB_draw_instanced; - GLboolean ARB_fragment_coord_conventions; - GLboolean ARB_fragment_program; - GLboolean ARB_fragment_program_shadow; - GLboolean ARB_fragment_shader; - GLboolean ARB_framebuffer_object; - GLboolean ARB_explicit_attrib_location; - GLboolean ARB_geometry_shader4; - GLboolean ARB_half_float_pixel; - GLboolean ARB_half_float_vertex; - GLboolean ARB_instanced_arrays; - GLboolean ARB_map_buffer_range; - GLboolean ARB_multisample; - GLboolean ARB_multitexture; - GLboolean ARB_occlusion_query; - GLboolean ARB_occlusion_query2; - GLboolean ARB_point_sprite; - GLboolean ARB_sampler_objects; - GLboolean ARB_seamless_cube_map; - GLboolean ARB_shader_objects; - GLboolean ARB_shader_stencil_export; - GLboolean ARB_shading_language_100; - GLboolean ARB_shadow; - GLboolean ARB_shadow_ambient; - GLboolean ARB_sync; - GLboolean ARB_texture_border_clamp; - GLboolean ARB_texture_buffer_object; - GLboolean ARB_texture_compression; - GLboolean ARB_texture_compression_rgtc; - GLboolean ARB_texture_cube_map; - GLboolean ARB_texture_env_combine; - GLboolean ARB_texture_env_crossbar; - GLboolean ARB_texture_env_dot3; - GLboolean ARB_texture_float; - GLboolean ARB_texture_mirrored_repeat; - GLboolean ARB_texture_multisample; - GLboolean ARB_texture_non_power_of_two; - GLboolean ARB_texture_rg; - GLboolean ARB_texture_rgb10_a2ui; - GLboolean ARB_timer_query; - GLboolean ARB_transform_feedback2; - GLboolean ARB_transpose_matrix; - GLboolean ARB_uniform_buffer_object; - GLboolean ARB_vertex_array_object; - GLboolean ARB_vertex_buffer_object; - GLboolean ARB_vertex_program; - GLboolean ARB_vertex_shader; - GLboolean ARB_vertex_type_2_10_10_10_rev; - GLboolean ARB_window_pos; - GLboolean EXT_abgr; - GLboolean EXT_bgra; - GLboolean EXT_blend_color; - GLboolean EXT_blend_equation_separate; - GLboolean EXT_blend_func_separate; - GLboolean EXT_blend_logic_op; - GLboolean EXT_blend_minmax; - GLboolean EXT_blend_subtract; - GLboolean EXT_clip_volume_hint; - GLboolean EXT_compiled_vertex_array; - GLboolean EXT_copy_texture; - GLboolean EXT_depth_bounds_test; - GLboolean EXT_draw_buffers2; - GLboolean EXT_draw_range_elements; - GLboolean EXT_fog_coord; - GLboolean EXT_framebuffer_blit; - GLboolean EXT_framebuffer_multisample; - GLboolean EXT_framebuffer_object; - GLboolean EXT_framebuffer_sRGB; - GLboolean EXT_gpu_program_parameters; - GLboolean EXT_gpu_shader4; - GLboolean EXT_multi_draw_arrays; - GLboolean EXT_paletted_texture; - GLboolean EXT_packed_depth_stencil; - GLboolean EXT_packed_float; - GLboolean EXT_packed_pixels; - GLboolean EXT_pixel_buffer_object; - GLboolean EXT_point_parameters; - GLboolean EXT_polygon_offset; - GLboolean EXT_provoking_vertex; - GLboolean EXT_rescale_normal; - GLboolean EXT_shadow_funcs; - GLboolean EXT_secondary_color; - GLboolean EXT_separate_shader_objects; - GLboolean EXT_separate_specular_color; - GLboolean EXT_shared_texture_palette; - GLboolean EXT_stencil_wrap; - GLboolean EXT_stencil_two_side; - GLboolean EXT_subtexture; - GLboolean EXT_texture; - GLboolean EXT_texture_object; - GLboolean EXT_texture3D; - GLboolean EXT_texture_array; - GLboolean EXT_texture_compression_s3tc; - GLboolean EXT_texture_env_add; - GLboolean EXT_texture_env_combine; - GLboolean EXT_texture_env_dot3; - GLboolean EXT_texture_filter_anisotropic; - GLboolean EXT_texture_integer; - GLboolean EXT_texture_lod_bias; - GLboolean EXT_texture_mirror_clamp; - GLboolean EXT_texture_shared_exponent; - GLboolean EXT_texture_sRGB; - GLboolean EXT_texture_sRGB_decode; - GLboolean EXT_texture_swizzle; - GLboolean EXT_transform_feedback; - GLboolean EXT_timer_query; - GLboolean EXT_vertex_array; - GLboolean EXT_vertex_array_bgra; - GLboolean EXT_vertex_array_set; - GLboolean OES_standard_derivatives; - /* vendor extensions */ - GLboolean AMD_conservative_depth; - GLboolean APPLE_client_storage; - GLboolean APPLE_packed_pixels; - GLboolean APPLE_vertex_array_object; - GLboolean APPLE_object_purgeable; - GLboolean ATI_envmap_bumpmap; - GLboolean ATI_texture_mirror_once; - GLboolean ATI_texture_env_combine3; - GLboolean ATI_fragment_shader; - GLboolean ATI_separate_stencil; - GLboolean IBM_rasterpos_clip; - GLboolean IBM_multimode_draw_arrays; - GLboolean MESA_pack_invert; - GLboolean MESA_resize_buffers; - GLboolean MESA_ycbcr_texture; - GLboolean MESA_texture_array; - GLboolean MESA_texture_signed_rgba; - GLboolean NV_blend_square; - GLboolean NV_conditional_render; - GLboolean NV_fragment_program; - GLboolean NV_fragment_program_option; - GLboolean NV_light_max_exponent; - GLboolean NV_point_sprite; - GLboolean NV_primitive_restart; - GLboolean NV_texgen_reflection; - GLboolean NV_texture_env_combine4; - GLboolean NV_texture_rectangle; - GLboolean NV_vertex_program; - GLboolean NV_vertex_program1_1; - GLboolean OES_read_format; - GLboolean SGI_texture_color_table; - GLboolean SGIS_generate_mipmap; - GLboolean SGIS_texture_edge_clamp; - GLboolean SGIS_texture_lod; - GLboolean TDFX_texture_compression_FXT1; - GLboolean S3_s3tc; - GLboolean OES_EGL_image; - GLboolean OES_draw_texture; - GLboolean EXT_texture_format_BGRA8888; - GLboolean extension_sentinel; - /** The extension string */ - const GLubyte *String; - /** Number of supported extensions */ - GLuint Count; -}; - - -/** - * A stack of matrices (projection, modelview, color, texture, etc). - */ -struct gl_matrix_stack -{ - GLmatrix *Top; /**< points into Stack */ - GLmatrix *Stack; /**< array [MaxDepth] of GLmatrix */ - GLuint Depth; /**< 0 <= Depth < MaxDepth */ - GLuint MaxDepth; /**< size of Stack[] array */ - GLuint DirtyFlag; /**< _NEW_MODELVIEW or _NEW_PROJECTION, for example */ -}; - - -/** - * \name Bits for image transfer operations - * \sa __struct gl_contextRec::ImageTransferState. - */ -/*@{*/ -#define IMAGE_SCALE_BIAS_BIT 0x1 -#define IMAGE_SHIFT_OFFSET_BIT 0x2 -#define IMAGE_MAP_COLOR_BIT 0x4 -#define IMAGE_CLAMP_BIT 0x800 - - -/** Pixel Transfer ops */ -#define IMAGE_BITS (IMAGE_SCALE_BIAS_BIT | \ - IMAGE_SHIFT_OFFSET_BIT | \ - IMAGE_MAP_COLOR_BIT) - -/** - * \name Bits to indicate what state has changed. - */ -/*@{*/ -#define _NEW_MODELVIEW (1 << 0) /**< gl_context::ModelView */ -#define _NEW_PROJECTION (1 << 1) /**< gl_context::Projection */ -#define _NEW_TEXTURE_MATRIX (1 << 2) /**< gl_context::TextureMatrix */ -#define _NEW_COLOR (1 << 3) /**< gl_context::Color */ -#define _NEW_DEPTH (1 << 4) /**< gl_context::Depth */ -#define _NEW_EVAL (1 << 5) /**< gl_context::Eval, EvalMap */ -#define _NEW_FOG (1 << 6) /**< gl_context::Fog */ -#define _NEW_HINT (1 << 7) /**< gl_context::Hint */ -#define _NEW_LIGHT (1 << 8) /**< gl_context::Light */ -#define _NEW_LINE (1 << 9) /**< gl_context::Line */ -#define _NEW_PIXEL (1 << 10) /**< gl_context::Pixel */ -#define _NEW_POINT (1 << 11) /**< gl_context::Point */ -#define _NEW_POLYGON (1 << 12) /**< gl_context::Polygon */ -#define _NEW_POLYGONSTIPPLE (1 << 13) /**< gl_context::PolygonStipple */ -#define _NEW_SCISSOR (1 << 14) /**< gl_context::Scissor */ -#define _NEW_STENCIL (1 << 15) /**< gl_context::Stencil */ -#define _NEW_TEXTURE (1 << 16) /**< gl_context::Texture */ -#define _NEW_TRANSFORM (1 << 17) /**< gl_context::Transform */ -#define _NEW_VIEWPORT (1 << 18) /**< gl_context::Viewport */ -#define _NEW_PACKUNPACK (1 << 19) /**< gl_context::Pack, Unpack */ -#define _NEW_ARRAY (1 << 20) /**< gl_context::Array */ -#define _NEW_RENDERMODE (1 << 21) /**< gl_context::RenderMode, etc */ -#define _NEW_BUFFERS (1 << 22) /**< gl_context::Visual, DrawBuffer, */ -#define _NEW_CURRENT_ATTRIB (1 << 23) /**< gl_context::Current */ -#define _NEW_MULTISAMPLE (1 << 24) /**< gl_context::Multisample */ -#define _NEW_TRACK_MATRIX (1 << 25) /**< gl_context::VertexProgram */ -#define _NEW_PROGRAM (1 << 26) /**< New program/shader state */ -#define _NEW_PROGRAM_CONSTANTS (1 << 27) -#define _NEW_BUFFER_OBJECT (1 << 28) -#define _NEW_ALL ~0 -/*@}*/ - - -/** - * \name Bits to track array state changes - * - * Also used to summarize array enabled. - */ -/*@{*/ -#define _NEW_ARRAY_VERTEX VERT_BIT_POS -#define _NEW_ARRAY_WEIGHT VERT_BIT_WEIGHT -#define _NEW_ARRAY_NORMAL VERT_BIT_NORMAL -#define _NEW_ARRAY_COLOR0 VERT_BIT_COLOR0 -#define _NEW_ARRAY_COLOR1 VERT_BIT_COLOR1 -#define _NEW_ARRAY_FOGCOORD VERT_BIT_FOG -#define _NEW_ARRAY_INDEX VERT_BIT_COLOR_INDEX -#define _NEW_ARRAY_EDGEFLAG VERT_BIT_EDGEFLAG -#define _NEW_ARRAY_POINT_SIZE VERT_BIT_COLOR_INDEX /* aliased */ -#define _NEW_ARRAY_TEXCOORD_0 VERT_BIT_TEX0 -#define _NEW_ARRAY_TEXCOORD_1 VERT_BIT_TEX1 -#define _NEW_ARRAY_TEXCOORD_2 VERT_BIT_TEX2 -#define _NEW_ARRAY_TEXCOORD_3 VERT_BIT_TEX3 -#define _NEW_ARRAY_TEXCOORD_4 VERT_BIT_TEX4 -#define _NEW_ARRAY_TEXCOORD_5 VERT_BIT_TEX5 -#define _NEW_ARRAY_TEXCOORD_6 VERT_BIT_TEX6 -#define _NEW_ARRAY_TEXCOORD_7 VERT_BIT_TEX7 -#define _NEW_ARRAY_ATTRIB_0 VERT_BIT_GENERIC0 /* start at bit 16 */ -#define _NEW_ARRAY_ALL 0xffffffff - - -#define _NEW_ARRAY_TEXCOORD(i) (_NEW_ARRAY_TEXCOORD_0 << (i)) -#define _NEW_ARRAY_ATTRIB(i) (_NEW_ARRAY_ATTRIB_0 << (i)) -/*@}*/ - - - -/** - * \name A bunch of flags that we think might be useful to drivers. - * - * Set in the __struct gl_contextRec::_TriangleCaps bitfield. - */ -/*@{*/ -#define DD_FLATSHADE 0x1 -#define DD_SEPARATE_SPECULAR 0x2 -#define DD_TRI_CULL_FRONT_BACK 0x4 /* special case on some hw */ -#define DD_TRI_LIGHT_TWOSIDE 0x8 -#define DD_TRI_UNFILLED 0x10 -#define DD_TRI_SMOOTH 0x20 -#define DD_TRI_STIPPLE 0x40 -#define DD_TRI_OFFSET 0x80 -#define DD_LINE_SMOOTH 0x100 -#define DD_LINE_STIPPLE 0x200 -#define DD_POINT_SMOOTH 0x400 -#define DD_POINT_ATTEN 0x800 -#define DD_TRI_TWOSTENCIL 0x1000 -/*@}*/ - - -/** - * \name Define the state changes under which each of these bits might change - */ -/*@{*/ -#define _DD_NEW_FLATSHADE _NEW_LIGHT -#define _DD_NEW_SEPARATE_SPECULAR (_NEW_LIGHT | _NEW_FOG | _NEW_PROGRAM) -#define _DD_NEW_TRI_CULL_FRONT_BACK _NEW_POLYGON -#define _DD_NEW_TRI_LIGHT_TWOSIDE _NEW_LIGHT -#define _DD_NEW_TRI_UNFILLED _NEW_POLYGON -#define _DD_NEW_TRI_SMOOTH _NEW_POLYGON -#define _DD_NEW_TRI_STIPPLE _NEW_POLYGON -#define _DD_NEW_TRI_OFFSET _NEW_POLYGON -#define _DD_NEW_LINE_SMOOTH _NEW_LINE -#define _DD_NEW_LINE_STIPPLE _NEW_LINE -#define _DD_NEW_LINE_WIDTH _NEW_LINE -#define _DD_NEW_POINT_SMOOTH _NEW_POINT -#define _DD_NEW_POINT_SIZE _NEW_POINT -#define _DD_NEW_POINT_ATTEN _NEW_POINT -/*@}*/ - - -/** - * Composite state flags - */ -/*@{*/ -#define _MESA_NEW_NEED_EYE_COORDS (_NEW_LIGHT | \ - _NEW_TEXTURE | \ - _NEW_POINT | \ - _NEW_PROGRAM | \ - _NEW_MODELVIEW) - -#define _MESA_NEW_NEED_NORMALS (_NEW_LIGHT | \ - _NEW_TEXTURE) - -#define _MESA_NEW_TRANSFER_STATE (_NEW_PIXEL) -/*@}*/ - - - - -/* This has to be included here. */ -#include "dd.h" - - -/** - * Display list flags. - * Strictly this is a tnl-private concept, but it doesn't seem - * worthwhile adding a tnl private structure just to hold this one bit - * of information: - */ -#define DLIST_DANGLING_REFS 0x1 - - -/** Opaque declaration of display list payload data type */ -union gl_dlist_node; - - -/** - * Provide a location where information about a display list can be - * collected. Could be extended with driverPrivate structures, - * etc. in the future. - */ -struct gl_display_list -{ - GLuint Name; - GLbitfield Flags; /**< DLIST_x flags */ - /** The dlist commands are in a linked list of nodes */ - union gl_dlist_node *Head; -}; - - -/** - * State used during display list compilation and execution. - */ -struct gl_dlist_state -{ - GLuint CallDepth; /**< Current recursion calling depth */ - - struct gl_display_list *CurrentList; /**< List currently being compiled */ - union gl_dlist_node *CurrentBlock; /**< Pointer to current block of nodes */ - GLuint CurrentPos; /**< Index into current block of nodes */ - - GLvertexformat ListVtxfmt; - - GLubyte ActiveAttribSize[VERT_ATTRIB_MAX]; - GLfloat CurrentAttrib[VERT_ATTRIB_MAX][4]; - - GLubyte ActiveMaterialSize[MAT_ATTRIB_MAX]; - GLfloat CurrentMaterial[MAT_ATTRIB_MAX][4]; - - GLubyte ActiveIndex; - GLfloat CurrentIndex; - - GLubyte ActiveEdgeFlag; - GLboolean CurrentEdgeFlag; - - struct { - /* State known to have been set by the currently-compiling display - * list. Used to eliminate some redundant state changes. - */ - GLenum ShadeModel; - } Current; -}; - - -/** - * Enum for the OpenGL APIs we know about and may support. - */ -typedef enum -{ - API_OPENGL, - API_OPENGLES, - API_OPENGLES2 -} gl_api; - - -/** - * Mesa rendering context. - * - * This is the central context data structure for Mesa. Almost all - * OpenGL state is contained in this structure. - * Think of this as a base class from which device drivers will derive - * sub classes. - * - * The struct gl_context typedef names this structure. - */ -struct gl_context -{ - /** State possibly shared with other contexts in the address space */ - struct gl_shared_state *Shared; - - /** \name API function pointer tables */ - /*@{*/ - gl_api API; - struct _glapi_table *Save; /**< Display list save functions */ - struct _glapi_table *Exec; /**< Execute functions */ - struct _glapi_table *CurrentDispatch; /**< == Save or Exec !! */ - /*@}*/ - - struct gl_config Visual; - struct gl_framebuffer *DrawBuffer; /**< buffer for writing */ - struct gl_framebuffer *ReadBuffer; /**< buffer for reading */ - struct gl_framebuffer *WinSysDrawBuffer; /**< set with MakeCurrent */ - struct gl_framebuffer *WinSysReadBuffer; /**< set with MakeCurrent */ - - /** - * Device driver function pointer table - */ - struct dd_function_table Driver; - - void *DriverCtx; /**< Points to device driver context/state */ - - /** Core/Driver constants */ - struct gl_constants Const; - - /** \name The various 4x4 matrix stacks */ - /*@{*/ - struct gl_matrix_stack ModelviewMatrixStack; - struct gl_matrix_stack ProjectionMatrixStack; - struct gl_matrix_stack TextureMatrixStack[MAX_TEXTURE_UNITS]; - struct gl_matrix_stack ProgramMatrixStack[MAX_PROGRAM_MATRICES]; - struct gl_matrix_stack *CurrentStack; /**< Points to one of the above stacks */ - /*@}*/ - - /** Combined modelview and projection matrix */ - GLmatrix _ModelProjectMatrix; - - /** \name Display lists */ - struct gl_dlist_state ListState; - - GLboolean ExecuteFlag; /**< Execute GL commands? */ - GLboolean CompileFlag; /**< Compile GL commands into display list? */ - - /** Extension information */ - struct gl_extensions Extensions; - - /** Version info */ - GLuint VersionMajor, VersionMinor; - char *VersionString; - - /** \name State attribute stack (for glPush/PopAttrib) */ - /*@{*/ - GLuint AttribStackDepth; - struct gl_attrib_node *AttribStack[MAX_ATTRIB_STACK_DEPTH]; - /*@}*/ - - /** \name Renderer attribute groups - * - * We define a struct for each attribute group to make pushing and popping - * attributes easy. Also it's a good organization. - */ - /*@{*/ - struct gl_accum_attrib Accum; /**< Accum buffer attributes */ - struct gl_colorbuffer_attrib Color; /**< Color buffer attributes */ - struct gl_current_attrib Current; /**< Current attributes */ - struct gl_depthbuffer_attrib Depth; /**< Depth buffer attributes */ - struct gl_eval_attrib Eval; /**< Eval attributes */ - struct gl_fog_attrib Fog; /**< Fog attributes */ - struct gl_hint_attrib Hint; /**< Hint attributes */ - struct gl_light_attrib Light; /**< Light attributes */ - struct gl_line_attrib Line; /**< Line attributes */ - struct gl_list_attrib List; /**< List attributes */ - struct gl_multisample_attrib Multisample; - struct gl_pixel_attrib Pixel; /**< Pixel attributes */ - struct gl_point_attrib Point; /**< Point attributes */ - struct gl_polygon_attrib Polygon; /**< Polygon attributes */ - GLuint PolygonStipple[32]; /**< Polygon stipple */ - struct gl_scissor_attrib Scissor; /**< Scissor attributes */ - struct gl_stencil_attrib Stencil; /**< Stencil buffer attributes */ - struct gl_texture_attrib Texture; /**< Texture attributes */ - struct gl_transform_attrib Transform; /**< Transformation attributes */ - struct gl_viewport_attrib Viewport; /**< Viewport attributes */ - /*@}*/ - - /** \name Client attribute stack */ - /*@{*/ - GLuint ClientAttribStackDepth; - struct gl_attrib_node *ClientAttribStack[MAX_CLIENT_ATTRIB_STACK_DEPTH]; - /*@}*/ - - /** \name Client attribute groups */ - /*@{*/ - struct gl_array_attrib Array; /**< Vertex arrays */ - struct gl_pixelstore_attrib Pack; /**< Pixel packing */ - struct gl_pixelstore_attrib Unpack; /**< Pixel unpacking */ - struct gl_pixelstore_attrib DefaultPacking; /**< Default params */ - /*@}*/ - - /** \name Other assorted state (not pushed/popped on attribute stack) */ - /*@{*/ - struct gl_pixelmaps PixelMaps; - - struct gl_evaluators EvalMap; /**< All evaluators */ - struct gl_feedback Feedback; /**< Feedback */ - struct gl_selection Select; /**< Selection */ - - struct gl_program_state Program; /**< general program state */ - struct gl_vertex_program_state VertexProgram; - struct gl_fragment_program_state FragmentProgram; - struct gl_geometry_program_state GeometryProgram; - struct gl_ati_fragment_shader_state ATIFragmentShader; - - struct gl_shader_state Shader; /**< GLSL shader object state */ - struct gl_shader_compiler_options ShaderCompilerOptions[MESA_SHADER_TYPES]; - - struct gl_query_state Query; /**< occlusion, timer queries */ - - struct gl_transform_feedback TransformFeedback; - - struct gl_buffer_object *CopyReadBuffer; /**< GL_ARB_copy_buffer */ - struct gl_buffer_object *CopyWriteBuffer; /**< GL_ARB_copy_buffer */ - /*@}*/ - - struct gl_meta_state *Meta; /**< for "meta" operations */ - - /* GL_EXT_framebuffer_object */ - struct gl_renderbuffer *CurrentRenderbuffer; - - GLenum ErrorValue; /**< Last error code */ - - /** - * Recognize and silence repeated error debug messages in buggy apps. - */ - const char *ErrorDebugFmtString; - GLuint ErrorDebugCount; - - GLenum RenderMode; /**< either GL_RENDER, GL_SELECT, GL_FEEDBACK */ - GLbitfield NewState; /**< bitwise-or of _NEW_* flags */ - - GLboolean ViewportInitialized; /**< has viewport size been initialized? */ - - GLbitfield varying_vp_inputs; /**< mask of VERT_BIT_* flags */ - - /** \name Derived state */ - /*@{*/ - /** Bitwise-or of DD_* flags. Note that this bitfield may be used before - * state validation so they need to always be current. - */ - GLbitfield _TriangleCaps; - GLbitfield _ImageTransferState;/**< bitwise-or of IMAGE_*_BIT flags */ - GLfloat _EyeZDir[3]; - GLfloat _ModelViewInvScale; - GLboolean _NeedEyeCoords; - GLboolean _ForceEyeCoords; - - GLuint TextureStateTimestamp; /**< detect changes to shared state */ - - struct gl_shine_tab *_ShineTable[2]; /**< Active shine tables */ - struct gl_shine_tab *_ShineTabList; /**< MRU list of inactive shine tables */ - /**@}*/ - - struct gl_list_extensions *ListExt; /**< driver dlist extensions */ - - /** \name For debugging/development only */ - /*@{*/ - GLboolean FirstTimeCurrent; - /*@}*/ - - /** Dither disable via MESA_NO_DITHER env var */ - GLboolean NoDither; - - /** software compression/decompression supported or not */ - GLboolean Mesa_DXTn; - - GLboolean TextureFormatSupported[MESA_FORMAT_COUNT]; - - /** - * Use dp4 (rather than mul/mad) instructions for position - * transformation? - */ - GLboolean mvp_with_dp4; - - /** - * \name Hooks for module contexts. - * - * These will eventually live in the driver or elsewhere. - */ - /*@{*/ - void *swrast_context; - void *swsetup_context; - void *swtnl_context; - void *swtnl_im; - struct st_context *st; - void *aelt_context; - /*@}*/ -}; - - -#ifdef DEBUG -extern int MESA_VERBOSE; -extern int MESA_DEBUG_FLAGS; -# define MESA_FUNCTION __FUNCTION__ -#else -# define MESA_VERBOSE 0 -# define MESA_DEBUG_FLAGS 0 -# define MESA_FUNCTION "a function" -# ifndef NDEBUG -# define NDEBUG -# endif -#endif - - -/** The MESA_VERBOSE var is a bitmask of these flags */ -enum _verbose -{ - VERBOSE_VARRAY = 0x0001, - VERBOSE_TEXTURE = 0x0002, - VERBOSE_MATERIAL = 0x0004, - VERBOSE_PIPELINE = 0x0008, - VERBOSE_DRIVER = 0x0010, - VERBOSE_STATE = 0x0020, - VERBOSE_API = 0x0040, - VERBOSE_DISPLAY_LIST = 0x0100, - VERBOSE_LIGHTING = 0x0200, - VERBOSE_PRIMS = 0x0400, - VERBOSE_VERTS = 0x0800, - VERBOSE_DISASSEM = 0x1000, - VERBOSE_DRAW = 0x2000, - VERBOSE_SWAPBUFFERS = 0x4000 -}; - - -/** The MESA_DEBUG_FLAGS var is a bitmask of these flags */ -enum _debug -{ - DEBUG_ALWAYS_FLUSH = 0x1 -}; - - - -#endif /* MTYPES_H */ +/* + * Mesa 3-D graphics library + * Version: 7.7 + * + * Copyright (C) 1999-2008 Brian Paul All Rights Reserved. + * 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 + * 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. + */ + +/** + * \file mtypes.h + * Main Mesa data structures. + * + * Please try to mark derived values with a leading underscore ('_'). + */ + +#ifndef MTYPES_H +#define MTYPES_H + + +#include "main/glheader.h" +#include "main/config.h" +#include "main/mfeatures.h" +#include "glapi/glapi.h" +#include "math/m_matrix.h" /* GLmatrix */ +#include "main/simple_list.h" /* struct simple_node */ +#include "main/formats.h" /* MESA_FORMAT_COUNT */ + + +/** + * Color channel data type. + */ +#if CHAN_BITS == 8 + typedef GLubyte GLchan; +#define CHAN_MAX 255 +#define CHAN_MAXF 255.0F +#define CHAN_TYPE GL_UNSIGNED_BYTE +#elif CHAN_BITS == 16 + typedef GLushort GLchan; +#define CHAN_MAX 65535 +#define CHAN_MAXF 65535.0F +#define CHAN_TYPE GL_UNSIGNED_SHORT +#elif CHAN_BITS == 32 + typedef GLfloat GLchan; +#define CHAN_MAX 1.0 +#define CHAN_MAXF 1.0F +#define CHAN_TYPE GL_FLOAT +#else +#error "illegal number of color channel bits" +#endif + + +/** + * Stencil buffer data type. + */ +#if STENCIL_BITS==8 + typedef GLubyte GLstencil; +#elif STENCIL_BITS==16 + typedef GLushort GLstencil; +#else +# error "illegal number of stencil bits" +#endif + + +/** + * \name 64-bit extension of GLbitfield. + */ +/*@{*/ +typedef GLuint64 GLbitfield64; + +/** Set a single bit */ +#define BITFIELD64_BIT(b) (1ULL << (b)) + + +/** + * \name Some forward type declarations + */ +/*@{*/ +struct _mesa_HashTable; +struct gl_attrib_node; +struct gl_list_extensions; +struct gl_meta_state; +struct gl_pixelstore_attrib; +struct gl_program_cache; +struct gl_texture_format; +struct gl_texture_image; +struct gl_texture_object; +struct gl_context; +struct st_context; +/*@}*/ + + +/** Extra draw modes beyond GL_POINTS, GL_TRIANGLE_FAN, etc */ +#define PRIM_OUTSIDE_BEGIN_END (GL_POLYGON+1) +#define PRIM_INSIDE_UNKNOWN_PRIM (GL_POLYGON+2) +#define PRIM_UNKNOWN (GL_POLYGON+3) + + +/** + * Shader stages. Note that these will become 5 with tessellation. + * These MUST have the same values as gallium's PIPE_SHADER_* + */ +typedef enum +{ + MESA_SHADER_VERTEX = 0, + MESA_SHADER_FRAGMENT = 1, + MESA_SHADER_GEOMETRY = 2, + MESA_SHADER_TYPES = 3 +} gl_shader_type; + + + +/** + * Indexes for vertex program attributes. + * GL_NV_vertex_program aliases generic attributes over the conventional + * attributes. In GL_ARB_vertex_program shader the aliasing is optional. + * In GL_ARB_vertex_shader / OpenGL 2.0 the aliasing is disallowed (the + * generic attributes are distinct/separate). + */ +typedef enum +{ + VERT_ATTRIB_POS = 0, + VERT_ATTRIB_WEIGHT = 1, + VERT_ATTRIB_NORMAL = 2, + VERT_ATTRIB_COLOR0 = 3, + VERT_ATTRIB_COLOR1 = 4, + VERT_ATTRIB_FOG = 5, + VERT_ATTRIB_COLOR_INDEX = 6, + VERT_ATTRIB_POINT_SIZE = 6, /*alias*/ + VERT_ATTRIB_EDGEFLAG = 7, + VERT_ATTRIB_TEX0 = 8, + VERT_ATTRIB_TEX1 = 9, + VERT_ATTRIB_TEX2 = 10, + VERT_ATTRIB_TEX3 = 11, + VERT_ATTRIB_TEX4 = 12, + VERT_ATTRIB_TEX5 = 13, + VERT_ATTRIB_TEX6 = 14, + VERT_ATTRIB_TEX7 = 15, + VERT_ATTRIB_GENERIC0 = 16, + VERT_ATTRIB_GENERIC1 = 17, + VERT_ATTRIB_GENERIC2 = 18, + VERT_ATTRIB_GENERIC3 = 19, + VERT_ATTRIB_GENERIC4 = 20, + VERT_ATTRIB_GENERIC5 = 21, + VERT_ATTRIB_GENERIC6 = 22, + VERT_ATTRIB_GENERIC7 = 23, + VERT_ATTRIB_GENERIC8 = 24, + VERT_ATTRIB_GENERIC9 = 25, + VERT_ATTRIB_GENERIC10 = 26, + VERT_ATTRIB_GENERIC11 = 27, + VERT_ATTRIB_GENERIC12 = 28, + VERT_ATTRIB_GENERIC13 = 29, + VERT_ATTRIB_GENERIC14 = 30, + VERT_ATTRIB_GENERIC15 = 31, + VERT_ATTRIB_MAX = 32 +} gl_vert_attrib; + +/** + * Bitflags for vertex attributes. + * These are used in bitfields in many places. + */ +/*@{*/ +#define VERT_BIT_POS (1 << VERT_ATTRIB_POS) +#define VERT_BIT_WEIGHT (1 << VERT_ATTRIB_WEIGHT) +#define VERT_BIT_NORMAL (1 << VERT_ATTRIB_NORMAL) +#define VERT_BIT_COLOR0 (1 << VERT_ATTRIB_COLOR0) +#define VERT_BIT_COLOR1 (1 << VERT_ATTRIB_COLOR1) +#define VERT_BIT_FOG (1 << VERT_ATTRIB_FOG) +#define VERT_BIT_COLOR_INDEX (1 << VERT_ATTRIB_COLOR_INDEX) +#define VERT_BIT_EDGEFLAG (1 << VERT_ATTRIB_EDGEFLAG) +#define VERT_BIT_TEX0 (1 << VERT_ATTRIB_TEX0) +#define VERT_BIT_TEX1 (1 << VERT_ATTRIB_TEX1) +#define VERT_BIT_TEX2 (1 << VERT_ATTRIB_TEX2) +#define VERT_BIT_TEX3 (1 << VERT_ATTRIB_TEX3) +#define VERT_BIT_TEX4 (1 << VERT_ATTRIB_TEX4) +#define VERT_BIT_TEX5 (1 << VERT_ATTRIB_TEX5) +#define VERT_BIT_TEX6 (1 << VERT_ATTRIB_TEX6) +#define VERT_BIT_TEX7 (1 << VERT_ATTRIB_TEX7) +#define VERT_BIT_GENERIC0 (1 << VERT_ATTRIB_GENERIC0) +#define VERT_BIT_GENERIC1 (1 << VERT_ATTRIB_GENERIC1) +#define VERT_BIT_GENERIC2 (1 << VERT_ATTRIB_GENERIC2) +#define VERT_BIT_GENERIC3 (1 << VERT_ATTRIB_GENERIC3) +#define VERT_BIT_GENERIC4 (1 << VERT_ATTRIB_GENERIC4) +#define VERT_BIT_GENERIC5 (1 << VERT_ATTRIB_GENERIC5) +#define VERT_BIT_GENERIC6 (1 << VERT_ATTRIB_GENERIC6) +#define VERT_BIT_GENERIC7 (1 << VERT_ATTRIB_GENERIC7) +#define VERT_BIT_GENERIC8 (1 << VERT_ATTRIB_GENERIC8) +#define VERT_BIT_GENERIC9 (1 << VERT_ATTRIB_GENERIC9) +#define VERT_BIT_GENERIC10 (1 << VERT_ATTRIB_GENERIC10) +#define VERT_BIT_GENERIC11 (1 << VERT_ATTRIB_GENERIC11) +#define VERT_BIT_GENERIC12 (1 << VERT_ATTRIB_GENERIC12) +#define VERT_BIT_GENERIC13 (1 << VERT_ATTRIB_GENERIC13) +#define VERT_BIT_GENERIC14 (1 << VERT_ATTRIB_GENERIC14) +#define VERT_BIT_GENERIC15 (1 << VERT_ATTRIB_GENERIC15) + +#define VERT_BIT_TEX(u) (1 << (VERT_ATTRIB_TEX0 + (u))) +#define VERT_BIT_GENERIC(g) (1 << (VERT_ATTRIB_GENERIC0 + (g))) +/*@}*/ + + +/** + * Indexes for vertex program result attributes + */ +typedef enum +{ + VERT_RESULT_HPOS = 0, + VERT_RESULT_COL0 = 1, + VERT_RESULT_COL1 = 2, + VERT_RESULT_FOGC = 3, + VERT_RESULT_TEX0 = 4, + VERT_RESULT_TEX1 = 5, + VERT_RESULT_TEX2 = 6, + VERT_RESULT_TEX3 = 7, + VERT_RESULT_TEX4 = 8, + VERT_RESULT_TEX5 = 9, + VERT_RESULT_TEX6 = 10, + VERT_RESULT_TEX7 = 11, + VERT_RESULT_PSIZ = 12, + VERT_RESULT_BFC0 = 13, + VERT_RESULT_BFC1 = 14, + VERT_RESULT_EDGE = 15, + VERT_RESULT_VAR0 = 16, /**< shader varying */ + VERT_RESULT_MAX = (VERT_RESULT_VAR0 + MAX_VARYING) +} gl_vert_result; + + +/*********************************************/ + +/** + * Indexes for geometry program attributes. + */ +typedef enum +{ + GEOM_ATTRIB_POSITION = 0, + GEOM_ATTRIB_COLOR0 = 1, + GEOM_ATTRIB_COLOR1 = 2, + GEOM_ATTRIB_SECONDARY_COLOR0 = 3, + GEOM_ATTRIB_SECONDARY_COLOR1 = 4, + GEOM_ATTRIB_FOG_FRAG_COORD = 5, + GEOM_ATTRIB_POINT_SIZE = 6, + GEOM_ATTRIB_CLIP_VERTEX = 7, + GEOM_ATTRIB_PRIMITIVE_ID = 8, + GEOM_ATTRIB_TEX_COORD = 9, + + GEOM_ATTRIB_VAR0 = 16, + GEOM_ATTRIB_MAX = (GEOM_ATTRIB_VAR0 + MAX_VARYING) +} gl_geom_attrib; + +/** + * Bitflags for geometry attributes. + * These are used in bitfields in many places. + */ +/*@{*/ +#define GEOM_BIT_COLOR0 (1 << GEOM_ATTRIB_COLOR0) +#define GEOM_BIT_COLOR1 (1 << GEOM_ATTRIB_COLOR1) +#define GEOM_BIT_SCOLOR0 (1 << GEOM_ATTRIB_SECONDARY_COLOR0) +#define GEOM_BIT_SCOLOR1 (1 << GEOM_ATTRIB_SECONDARY_COLOR1) +#define GEOM_BIT_TEX_COORD (1 << GEOM_ATTRIB_TEX_COORD) +#define GEOM_BIT_FOG_COORD (1 << GEOM_ATTRIB_FOG_FRAG_COORD) +#define GEOM_BIT_POSITION (1 << GEOM_ATTRIB_POSITION) +#define GEOM_BIT_POINT_SIDE (1 << GEOM_ATTRIB_POINT_SIZE) +#define GEOM_BIT_CLIP_VERTEX (1 << GEOM_ATTRIB_CLIP_VERTEX) +#define GEOM_BIT_PRIM_ID (1 << GEOM_ATTRIB_PRIMITIVE_ID) +#define GEOM_BIT_VAR0 (1 << GEOM_ATTRIB_VAR0) + +#define GEOM_BIT_VAR(g) (1 << (GEOM_BIT_VAR0 + (g))) +/*@}*/ + + +/** + * Indexes for geometry program result attributes + */ +typedef enum +{ + GEOM_RESULT_POS = 0, + GEOM_RESULT_COL0 = 1, + GEOM_RESULT_COL1 = 2, + GEOM_RESULT_SCOL0 = 3, + GEOM_RESULT_SCOL1 = 4, + GEOM_RESULT_FOGC = 5, + GEOM_RESULT_TEX0 = 6, + GEOM_RESULT_TEX1 = 7, + GEOM_RESULT_TEX2 = 8, + GEOM_RESULT_TEX3 = 9, + GEOM_RESULT_TEX4 = 10, + GEOM_RESULT_TEX5 = 11, + GEOM_RESULT_TEX6 = 12, + GEOM_RESULT_TEX7 = 13, + GEOM_RESULT_PSIZ = 14, + GEOM_RESULT_CLPV = 15, + GEOM_RESULT_PRID = 16, + GEOM_RESULT_LAYR = 17, + GEOM_RESULT_VAR0 = 18, /**< shader varying, should really be 16 */ + /* ### we need to -2 because var0 is 18 instead 16 like in the others */ + GEOM_RESULT_MAX = (GEOM_RESULT_VAR0 + MAX_VARYING - 2) +} gl_geom_result; + + +/** + * Indexes for fragment program input attributes. + */ +typedef enum +{ + FRAG_ATTRIB_WPOS = 0, + FRAG_ATTRIB_COL0 = 1, + FRAG_ATTRIB_COL1 = 2, + FRAG_ATTRIB_FOGC = 3, + FRAG_ATTRIB_TEX0 = 4, + FRAG_ATTRIB_TEX1 = 5, + FRAG_ATTRIB_TEX2 = 6, + FRAG_ATTRIB_TEX3 = 7, + FRAG_ATTRIB_TEX4 = 8, + FRAG_ATTRIB_TEX5 = 9, + FRAG_ATTRIB_TEX6 = 10, + FRAG_ATTRIB_TEX7 = 11, + FRAG_ATTRIB_FACE = 12, /**< front/back face */ + FRAG_ATTRIB_PNTC = 13, /**< sprite/point coord */ + FRAG_ATTRIB_VAR0 = 14, /**< shader varying */ + FRAG_ATTRIB_MAX = (FRAG_ATTRIB_VAR0 + MAX_VARYING) +} gl_frag_attrib; + +/** + * Bitflags for fragment program input attributes. + */ +/*@{*/ +#define FRAG_BIT_WPOS (1 << FRAG_ATTRIB_WPOS) +#define FRAG_BIT_COL0 (1 << FRAG_ATTRIB_COL0) +#define FRAG_BIT_COL1 (1 << FRAG_ATTRIB_COL1) +#define FRAG_BIT_FOGC (1 << FRAG_ATTRIB_FOGC) +#define FRAG_BIT_FACE (1 << FRAG_ATTRIB_FACE) +#define FRAG_BIT_PNTC (1 << FRAG_ATTRIB_PNTC) +#define FRAG_BIT_TEX0 (1 << FRAG_ATTRIB_TEX0) +#define FRAG_BIT_TEX1 (1 << FRAG_ATTRIB_TEX1) +#define FRAG_BIT_TEX2 (1 << FRAG_ATTRIB_TEX2) +#define FRAG_BIT_TEX3 (1 << FRAG_ATTRIB_TEX3) +#define FRAG_BIT_TEX4 (1 << FRAG_ATTRIB_TEX4) +#define FRAG_BIT_TEX5 (1 << FRAG_ATTRIB_TEX5) +#define FRAG_BIT_TEX6 (1 << FRAG_ATTRIB_TEX6) +#define FRAG_BIT_TEX7 (1 << FRAG_ATTRIB_TEX7) +#define FRAG_BIT_VAR0 (1 << FRAG_ATTRIB_VAR0) + +#define FRAG_BIT_TEX(U) (FRAG_BIT_TEX0 << (U)) +#define FRAG_BIT_VAR(V) (FRAG_BIT_VAR0 << (V)) + +#define FRAG_BITS_TEX_ANY (FRAG_BIT_TEX0| \ + FRAG_BIT_TEX1| \ + FRAG_BIT_TEX2| \ + FRAG_BIT_TEX3| \ + FRAG_BIT_TEX4| \ + FRAG_BIT_TEX5| \ + FRAG_BIT_TEX6| \ + FRAG_BIT_TEX7) +/*@}*/ + + +/** + * Fragment program results + */ +typedef enum +{ + FRAG_RESULT_DEPTH = 0, + FRAG_RESULT_STENCIL = 1, + FRAG_RESULT_COLOR = 2, + FRAG_RESULT_DATA0 = 3, + FRAG_RESULT_MAX = (FRAG_RESULT_DATA0 + MAX_DRAW_BUFFERS) +} gl_frag_result; + + +/** + * Indexes for all renderbuffers + */ +typedef enum +{ + /* the four standard color buffers */ + BUFFER_FRONT_LEFT, + BUFFER_BACK_LEFT, + BUFFER_FRONT_RIGHT, + BUFFER_BACK_RIGHT, + BUFFER_DEPTH, + BUFFER_STENCIL, + BUFFER_ACCUM, + /* optional aux buffer */ + BUFFER_AUX0, + /* generic renderbuffers */ + BUFFER_COLOR0, + BUFFER_COLOR1, + BUFFER_COLOR2, + BUFFER_COLOR3, + BUFFER_COLOR4, + BUFFER_COLOR5, + BUFFER_COLOR6, + BUFFER_COLOR7, + BUFFER_COUNT +} gl_buffer_index; + +/** + * Bit flags for all renderbuffers + */ +#define BUFFER_BIT_FRONT_LEFT (1 << BUFFER_FRONT_LEFT) +#define BUFFER_BIT_BACK_LEFT (1 << BUFFER_BACK_LEFT) +#define BUFFER_BIT_FRONT_RIGHT (1 << BUFFER_FRONT_RIGHT) +#define BUFFER_BIT_BACK_RIGHT (1 << BUFFER_BACK_RIGHT) +#define BUFFER_BIT_AUX0 (1 << BUFFER_AUX0) +#define BUFFER_BIT_AUX1 (1 << BUFFER_AUX1) +#define BUFFER_BIT_AUX2 (1 << BUFFER_AUX2) +#define BUFFER_BIT_AUX3 (1 << BUFFER_AUX3) +#define BUFFER_BIT_DEPTH (1 << BUFFER_DEPTH) +#define BUFFER_BIT_STENCIL (1 << BUFFER_STENCIL) +#define BUFFER_BIT_ACCUM (1 << BUFFER_ACCUM) +#define BUFFER_BIT_COLOR0 (1 << BUFFER_COLOR0) +#define BUFFER_BIT_COLOR1 (1 << BUFFER_COLOR1) +#define BUFFER_BIT_COLOR2 (1 << BUFFER_COLOR2) +#define BUFFER_BIT_COLOR3 (1 << BUFFER_COLOR3) +#define BUFFER_BIT_COLOR4 (1 << BUFFER_COLOR4) +#define BUFFER_BIT_COLOR5 (1 << BUFFER_COLOR5) +#define BUFFER_BIT_COLOR6 (1 << BUFFER_COLOR6) +#define BUFFER_BIT_COLOR7 (1 << BUFFER_COLOR7) + +/** + * Mask of all the color buffer bits (but not accum). + */ +#define BUFFER_BITS_COLOR (BUFFER_BIT_FRONT_LEFT | \ + BUFFER_BIT_BACK_LEFT | \ + BUFFER_BIT_FRONT_RIGHT | \ + BUFFER_BIT_BACK_RIGHT | \ + BUFFER_BIT_AUX0 | \ + BUFFER_BIT_COLOR0 | \ + BUFFER_BIT_COLOR1 | \ + BUFFER_BIT_COLOR2 | \ + BUFFER_BIT_COLOR3 | \ + BUFFER_BIT_COLOR4 | \ + BUFFER_BIT_COLOR5 | \ + BUFFER_BIT_COLOR6 | \ + BUFFER_BIT_COLOR7) + + +/** + * Framebuffer configuration (aka visual / pixelformat) + * Note: some of these fields should be boolean, but it appears that + * code in drivers/dri/common/util.c requires int-sized fields. + */ +struct gl_config +{ + GLboolean rgbMode; + GLboolean floatMode; + GLboolean colorIndexMode; /* XXX is this used anywhere? */ + GLuint doubleBufferMode; + GLuint stereoMode; + + GLboolean haveAccumBuffer; + GLboolean haveDepthBuffer; + GLboolean haveStencilBuffer; + + GLint redBits, greenBits, blueBits, alphaBits; /* bits per comp */ + GLuint redMask, greenMask, blueMask, alphaMask; + GLint rgbBits; /* total bits for rgb */ + GLint indexBits; /* total bits for colorindex */ + + GLint accumRedBits, accumGreenBits, accumBlueBits, accumAlphaBits; + GLint depthBits; + GLint stencilBits; + + GLint numAuxBuffers; + + GLint level; + + /* EXT_visual_rating / GLX 1.2 */ + GLint visualRating; + + /* EXT_visual_info / GLX 1.2 */ + GLint transparentPixel; + /* colors are floats scaled to ints */ + GLint transparentRed, transparentGreen, transparentBlue, transparentAlpha; + GLint transparentIndex; + + /* ARB_multisample / SGIS_multisample */ + GLint sampleBuffers; + GLint samples; + + /* SGIX_pbuffer / GLX 1.3 */ + GLint maxPbufferWidth; + GLint maxPbufferHeight; + GLint maxPbufferPixels; + GLint optimalPbufferWidth; /* Only for SGIX_pbuffer. */ + GLint optimalPbufferHeight; /* Only for SGIX_pbuffer. */ + + /* OML_swap_method */ + GLint swapMethod; + + /* EXT_texture_from_pixmap */ + GLint bindToTextureRgb; + GLint bindToTextureRgba; + GLint bindToMipmapTexture; + GLint bindToTextureTargets; + GLint yInverted; + + /* EXT_framebuffer_sRGB */ + GLint sRGBCapable; +}; + + +/** + * Data structure for color tables + */ +struct gl_color_table +{ + GLenum InternalFormat; /**< The user-specified format */ + GLenum _BaseFormat; /**< GL_ALPHA, GL_RGBA, GL_RGB, etc */ + GLuint Size; /**< number of entries in table */ + GLfloat *TableF; /**< Color table, floating point values */ + GLubyte *TableUB; /**< Color table, ubyte values */ + GLubyte RedSize; + GLubyte GreenSize; + GLubyte BlueSize; + GLubyte AlphaSize; + GLubyte LuminanceSize; + GLubyte IntensitySize; +}; + + +/** + * \name Bit flags used for updating material values. + */ +/*@{*/ +#define MAT_ATTRIB_FRONT_AMBIENT 0 +#define MAT_ATTRIB_BACK_AMBIENT 1 +#define MAT_ATTRIB_FRONT_DIFFUSE 2 +#define MAT_ATTRIB_BACK_DIFFUSE 3 +#define MAT_ATTRIB_FRONT_SPECULAR 4 +#define MAT_ATTRIB_BACK_SPECULAR 5 +#define MAT_ATTRIB_FRONT_EMISSION 6 +#define MAT_ATTRIB_BACK_EMISSION 7 +#define MAT_ATTRIB_FRONT_SHININESS 8 +#define MAT_ATTRIB_BACK_SHININESS 9 +#define MAT_ATTRIB_FRONT_INDEXES 10 +#define MAT_ATTRIB_BACK_INDEXES 11 +#define MAT_ATTRIB_MAX 12 + +#define MAT_ATTRIB_AMBIENT(f) (MAT_ATTRIB_FRONT_AMBIENT+(f)) +#define MAT_ATTRIB_DIFFUSE(f) (MAT_ATTRIB_FRONT_DIFFUSE+(f)) +#define MAT_ATTRIB_SPECULAR(f) (MAT_ATTRIB_FRONT_SPECULAR+(f)) +#define MAT_ATTRIB_EMISSION(f) (MAT_ATTRIB_FRONT_EMISSION+(f)) +#define MAT_ATTRIB_SHININESS(f)(MAT_ATTRIB_FRONT_SHININESS+(f)) +#define MAT_ATTRIB_INDEXES(f) (MAT_ATTRIB_FRONT_INDEXES+(f)) + +#define MAT_INDEX_AMBIENT 0 +#define MAT_INDEX_DIFFUSE 1 +#define MAT_INDEX_SPECULAR 2 + +#define MAT_BIT_FRONT_AMBIENT (1< ) */ + GLfloat _NormSpotDirection[4]; /**< normalized spotlight direction */ + GLfloat _VP_inf_spot_attenuation; + + GLfloat _SpotExpTable[EXP_TABLE_SIZE][2]; /**< to replace a pow() call */ + GLfloat _MatAmbient[2][3]; /**< material ambient * light ambient */ + GLfloat _MatDiffuse[2][3]; /**< material diffuse * light diffuse */ + GLfloat _MatSpecular[2][3]; /**< material spec * light specular */ + GLfloat _dli; /**< CI diffuse light intensity */ + GLfloat _sli; /**< CI specular light intensity */ + /*@}*/ +}; + + +/** + * Light model state. + */ +struct gl_lightmodel +{ + GLfloat Ambient[4]; /**< ambient color */ + GLboolean LocalViewer; /**< Local (or infinite) view point? */ + GLboolean TwoSide; /**< Two (or one) sided lighting? */ + GLenum ColorControl; /**< either GL_SINGLE_COLOR + * or GL_SEPARATE_SPECULAR_COLOR */ +}; + + +/** + * Material state. + */ +struct gl_material +{ + GLfloat Attrib[MAT_ATTRIB_MAX][4]; +}; + + +/** + * Accumulation buffer attribute group (GL_ACCUM_BUFFER_BIT) + */ +struct gl_accum_attrib +{ + GLfloat ClearColor[4]; /**< Accumulation buffer clear color */ +}; + + +/** + * Color buffer attribute group (GL_COLOR_BUFFER_BIT). + */ +struct gl_colorbuffer_attrib +{ + GLuint ClearIndex; /**< Index to use for glClear */ + GLclampf ClearColor[4]; /**< Color to use for glClear */ + + GLuint IndexMask; /**< Color index write mask */ + GLubyte ColorMask[MAX_DRAW_BUFFERS][4];/**< Each flag is 0xff or 0x0 */ + + GLenum DrawBuffer[MAX_DRAW_BUFFERS]; /**< Which buffer to draw into */ + + /** + * \name alpha testing + */ + /*@{*/ + GLboolean AlphaEnabled; /**< Alpha test enabled flag */ + GLenum AlphaFunc; /**< Alpha test function */ + GLclampf AlphaRef; /**< Alpha reference value */ + /*@}*/ + + /** + * \name Blending + */ + /*@{*/ + GLbitfield BlendEnabled; /**< Per-buffer blend enable flags */ + GLfloat BlendColor[4]; /**< Blending color */ + struct + { + GLenum SrcRGB; /**< RGB blend source term */ + GLenum DstRGB; /**< RGB blend dest term */ + GLenum SrcA; /**< Alpha blend source term */ + GLenum DstA; /**< Alpha blend dest term */ + GLenum EquationRGB; /**< GL_ADD, GL_SUBTRACT, etc. */ + GLenum EquationA; /**< GL_ADD, GL_SUBTRACT, etc. */ + } Blend[MAX_DRAW_BUFFERS]; + /** Are the blend func terms currently different for each buffer/target? */ + GLboolean _BlendFuncPerBuffer; + /** Are the blend equations currently different for each buffer/target? */ + GLboolean _BlendEquationPerBuffer; + /*@}*/ + + /** + * \name Logic op + */ + /*@{*/ + GLenum LogicOp; /**< Logic operator */ + GLboolean IndexLogicOpEnabled; /**< Color index logic op enabled flag */ + GLboolean ColorLogicOpEnabled; /**< RGBA logic op enabled flag */ + GLboolean _LogicOpEnabled; /**< RGBA logic op + EXT_blend_logic_op enabled flag */ + /*@}*/ + + GLboolean DitherFlag; /**< Dither enable flag */ + + GLenum ClampFragmentColor; /**< GL_TRUE, GL_FALSE or GL_FIXED_ONLY_ARB */ + GLenum ClampReadColor; /**< GL_TRUE, GL_FALSE or GL_FIXED_ONLY_ARB */ + + GLboolean sRGBEnabled; /**< Framebuffer sRGB blending/updating requested */ +}; + + +/** + * Current attribute group (GL_CURRENT_BIT). + */ +struct gl_current_attrib +{ + /** + * \name Current vertex attributes. + * \note Values are valid only after FLUSH_VERTICES has been called. + * \note Index and Edgeflag current values are stored as floats in the + * SIX and SEVEN attribute slots. + */ + GLfloat Attrib[VERT_ATTRIB_MAX][4]; /**< Position, color, texcoords, etc */ + + /** + * \name Current raster position attributes (always valid). + * \note This set of attributes is very similar to the SWvertex struct. + */ + /*@{*/ + GLfloat RasterPos[4]; + GLfloat RasterDistance; + GLfloat RasterColor[4]; + GLfloat RasterSecondaryColor[4]; + GLfloat RasterTexCoords[MAX_TEXTURE_COORD_UNITS][4]; + GLboolean RasterPosValid; + /*@}*/ +}; + + +/** + * Depth buffer attribute group (GL_DEPTH_BUFFER_BIT). + */ +struct gl_depthbuffer_attrib +{ + GLenum Func; /**< Function for depth buffer compare */ + GLclampd Clear; /**< Value to clear depth buffer to */ + GLboolean Test; /**< Depth buffering enabled flag */ + GLboolean Mask; /**< Depth buffer writable? */ + GLboolean BoundsTest; /**< GL_EXT_depth_bounds_test */ + GLfloat BoundsMin, BoundsMax;/**< GL_EXT_depth_bounds_test */ +}; + + +/** + * Evaluator attribute group (GL_EVAL_BIT). + */ +struct gl_eval_attrib +{ + /** + * \name Enable bits + */ + /*@{*/ + GLboolean Map1Color4; + GLboolean Map1Index; + GLboolean Map1Normal; + GLboolean Map1TextureCoord1; + GLboolean Map1TextureCoord2; + GLboolean Map1TextureCoord3; + GLboolean Map1TextureCoord4; + GLboolean Map1Vertex3; + GLboolean Map1Vertex4; + GLboolean Map1Attrib[16]; /* GL_NV_vertex_program */ + GLboolean Map2Color4; + GLboolean Map2Index; + GLboolean Map2Normal; + GLboolean Map2TextureCoord1; + GLboolean Map2TextureCoord2; + GLboolean Map2TextureCoord3; + GLboolean Map2TextureCoord4; + GLboolean Map2Vertex3; + GLboolean Map2Vertex4; + GLboolean Map2Attrib[16]; /* GL_NV_vertex_program */ + GLboolean AutoNormal; + /*@}*/ + + /** + * \name Map Grid endpoints and divisions and calculated du values + */ + /*@{*/ + GLint MapGrid1un; + GLfloat MapGrid1u1, MapGrid1u2, MapGrid1du; + GLint MapGrid2un, MapGrid2vn; + GLfloat MapGrid2u1, MapGrid2u2, MapGrid2du; + GLfloat MapGrid2v1, MapGrid2v2, MapGrid2dv; + /*@}*/ +}; + + +/** + * Fog attribute group (GL_FOG_BIT). + */ +struct gl_fog_attrib +{ + GLboolean Enabled; /**< Fog enabled flag */ + GLfloat Color[4]; /**< Fog color */ + GLfloat Density; /**< Density >= 0.0 */ + GLfloat Start; /**< Start distance in eye coords */ + GLfloat End; /**< End distance in eye coords */ + GLfloat Index; /**< Fog index */ + GLenum Mode; /**< Fog mode */ + GLboolean ColorSumEnabled; + GLenum FogCoordinateSource; /**< GL_EXT_fog_coord */ + GLfloat _Scale; /**< (End == Start) ? 1.0 : 1.0 / (End - Start) */ +}; + + +/** + * \brief Layout qualifiers for gl_FragDepth. + * + * Extension AMD_conservative_depth allows gl_FragDepth to be redeclared with + * a layout qualifier. + * + * \see enum ir_depth_layout + */ +enum gl_frag_depth_layout { + FRAG_DEPTH_LAYOUT_NONE, /**< No layout is specified. */ + FRAG_DEPTH_LAYOUT_ANY, + FRAG_DEPTH_LAYOUT_GREATER, + FRAG_DEPTH_LAYOUT_LESS, + FRAG_DEPTH_LAYOUT_UNCHANGED +}; + + +/** + * Hint attribute group (GL_HINT_BIT). + * + * Values are always one of GL_FASTEST, GL_NICEST, or GL_DONT_CARE. + */ +struct gl_hint_attrib +{ + GLenum PerspectiveCorrection; + GLenum PointSmooth; + GLenum LineSmooth; + GLenum PolygonSmooth; + GLenum Fog; + GLenum ClipVolumeClipping; /**< GL_EXT_clip_volume_hint */ + GLenum TextureCompression; /**< GL_ARB_texture_compression */ + GLenum GenerateMipmap; /**< GL_SGIS_generate_mipmap */ + GLenum FragmentShaderDerivative; /**< GL_ARB_fragment_shader */ +}; + +/** + * Light state flags. + */ +/*@{*/ +#define LIGHT_SPOT 0x1 +#define LIGHT_LOCAL_VIEWER 0x2 +#define LIGHT_POSITIONAL 0x4 +#define LIGHT_NEED_VERTICES (LIGHT_POSITIONAL|LIGHT_LOCAL_VIEWER) +/*@}*/ + + +/** + * Lighting attribute group (GL_LIGHT_BIT). + */ +struct gl_light_attrib +{ + struct gl_light Light[MAX_LIGHTS]; /**< Array of light sources */ + struct gl_lightmodel Model; /**< Lighting model */ + + /** + * Must flush FLUSH_VERTICES before referencing: + */ + /*@{*/ + struct gl_material Material; /**< Includes front & back values */ + /*@}*/ + + GLboolean Enabled; /**< Lighting enabled flag */ + GLenum ShadeModel; /**< GL_FLAT or GL_SMOOTH */ + GLenum ProvokingVertex; /**< GL_EXT_provoking_vertex */ + GLenum ColorMaterialFace; /**< GL_FRONT, BACK or FRONT_AND_BACK */ + GLenum ColorMaterialMode; /**< GL_AMBIENT, GL_DIFFUSE, etc */ + GLbitfield ColorMaterialBitmask; /**< bitmask formed from Face and Mode */ + GLboolean ColorMaterialEnabled; + GLenum ClampVertexColor; + + struct gl_light EnabledList; /**< List sentinel */ + + /** + * Derived state for optimizations: + */ + /*@{*/ + GLboolean _NeedEyeCoords; + GLboolean _NeedVertices; /**< Use fast shader? */ + GLbitfield _Flags; /**< LIGHT_* flags, see above */ + GLfloat _BaseColor[2][3]; + /*@}*/ +}; + + +/** + * Line attribute group (GL_LINE_BIT). + */ +struct gl_line_attrib +{ + GLboolean SmoothFlag; /**< GL_LINE_SMOOTH enabled? */ + GLboolean StippleFlag; /**< GL_LINE_STIPPLE enabled? */ + GLushort StipplePattern; /**< Stipple pattern */ + GLint StippleFactor; /**< Stipple repeat factor */ + GLfloat Width; /**< Line width */ +}; + + +/** + * Display list attribute group (GL_LIST_BIT). + */ +struct gl_list_attrib +{ + GLuint ListBase; +}; + + +/** + * Multisample attribute group (GL_MULTISAMPLE_BIT). + */ +struct gl_multisample_attrib +{ + GLboolean Enabled; + GLboolean _Enabled; /**< true if Enabled and multisample buffer */ + GLboolean SampleAlphaToCoverage; + GLboolean SampleAlphaToOne; + GLboolean SampleCoverage; + GLfloat SampleCoverageValue; + GLboolean SampleCoverageInvert; +}; + + +/** + * A pixelmap (see glPixelMap) + */ +struct gl_pixelmap +{ + GLint Size; + GLfloat Map[MAX_PIXEL_MAP_TABLE]; + GLubyte Map8[MAX_PIXEL_MAP_TABLE]; /**< converted to 8-bit color */ +}; + + +/** + * Collection of all pixelmaps + */ +struct gl_pixelmaps +{ + struct gl_pixelmap RtoR; /**< i.e. GL_PIXEL_MAP_R_TO_R */ + struct gl_pixelmap GtoG; + struct gl_pixelmap BtoB; + struct gl_pixelmap AtoA; + struct gl_pixelmap ItoR; + struct gl_pixelmap ItoG; + struct gl_pixelmap ItoB; + struct gl_pixelmap ItoA; + struct gl_pixelmap ItoI; + struct gl_pixelmap StoS; +}; + + +/** + * Pixel attribute group (GL_PIXEL_MODE_BIT). + */ +struct gl_pixel_attrib +{ + GLenum ReadBuffer; /**< source buffer for glRead/CopyPixels() */ + + /*--- Begin Pixel Transfer State ---*/ + /* Fields are in the order in which they're applied... */ + + /** Scale & Bias (index shift, offset) */ + /*@{*/ + GLfloat RedBias, RedScale; + GLfloat GreenBias, GreenScale; + GLfloat BlueBias, BlueScale; + GLfloat AlphaBias, AlphaScale; + GLfloat DepthBias, DepthScale; + GLint IndexShift, IndexOffset; + /*@}*/ + + /* Pixel Maps */ + /* Note: actual pixel maps are not part of this attrib group */ + GLboolean MapColorFlag; + GLboolean MapStencilFlag; + + /*--- End Pixel Transfer State ---*/ + + /** glPixelZoom */ + GLfloat ZoomX, ZoomY; + + /** GL_SGI_texture_color_table */ + GLfloat TextureColorTableScale[4]; /**< RGBA */ + GLfloat TextureColorTableBias[4]; /**< RGBA */ +}; + + +/** + * Point attribute group (GL_POINT_BIT). + */ +struct gl_point_attrib +{ + GLboolean SmoothFlag; /**< True if GL_POINT_SMOOTH is enabled */ + GLfloat Size; /**< User-specified point size */ + GLfloat Params[3]; /**< GL_EXT_point_parameters */ + GLfloat MinSize, MaxSize; /**< GL_EXT_point_parameters */ + GLfloat Threshold; /**< GL_EXT_point_parameters */ + GLboolean _Attenuated; /**< True if Params != [1, 0, 0] */ + GLboolean PointSprite; /**< GL_NV/ARB_point_sprite */ + GLboolean CoordReplace[MAX_TEXTURE_COORD_UNITS]; /**< GL_ARB_point_sprite*/ + GLenum SpriteRMode; /**< GL_NV_point_sprite (only!) */ + GLenum SpriteOrigin; /**< GL_ARB_point_sprite */ +}; + + +/** + * Polygon attribute group (GL_POLYGON_BIT). + */ +struct gl_polygon_attrib +{ + GLenum FrontFace; /**< Either GL_CW or GL_CCW */ + GLenum FrontMode; /**< Either GL_POINT, GL_LINE or GL_FILL */ + GLenum BackMode; /**< Either GL_POINT, GL_LINE or GL_FILL */ + GLboolean _FrontBit; /**< 0=GL_CCW, 1=GL_CW */ + GLboolean CullFlag; /**< Culling on/off flag */ + GLboolean SmoothFlag; /**< True if GL_POLYGON_SMOOTH is enabled */ + GLboolean StippleFlag; /**< True if GL_POLYGON_STIPPLE is enabled */ + GLenum CullFaceMode; /**< Culling mode GL_FRONT or GL_BACK */ + GLfloat OffsetFactor; /**< Polygon offset factor, from user */ + GLfloat OffsetUnits; /**< Polygon offset units, from user */ + GLboolean OffsetPoint; /**< Offset in GL_POINT mode */ + GLboolean OffsetLine; /**< Offset in GL_LINE mode */ + GLboolean OffsetFill; /**< Offset in GL_FILL mode */ +}; + + +/** + * Scissor attributes (GL_SCISSOR_BIT). + */ +struct gl_scissor_attrib +{ + GLboolean Enabled; /**< Scissor test enabled? */ + GLint X, Y; /**< Lower left corner of box */ + GLsizei Width, Height; /**< Size of box */ +}; + + +/** + * Stencil attribute group (GL_STENCIL_BUFFER_BIT). + * + * Three sets of stencil data are tracked so that OpenGL 2.0, + * GL_EXT_stencil_two_side, and GL_ATI_separate_stencil can all be supported + * simultaneously. In each of the stencil state arrays, element 0 corresponds + * to GL_FRONT. Element 1 corresponds to the OpenGL 2.0 / + * GL_ATI_separate_stencil GL_BACK state. Element 2 corresponds to the + * GL_EXT_stencil_two_side GL_BACK state. + * + * The derived value \c _BackFace is either 1 or 2 depending on whether or + * not GL_STENCIL_TEST_TWO_SIDE_EXT is enabled. + * + * The derived value \c _TestTwoSide is set when the front-face and back-face + * stencil state are different. + */ +struct gl_stencil_attrib +{ + GLboolean Enabled; /**< Enabled flag */ + GLboolean TestTwoSide; /**< GL_EXT_stencil_two_side */ + GLubyte ActiveFace; /**< GL_EXT_stencil_two_side (0 or 2) */ + GLboolean _Enabled; /**< Enabled and stencil buffer present */ + GLboolean _TestTwoSide; + GLubyte _BackFace; /**< Current back stencil state (1 or 2) */ + GLenum Function[3]; /**< Stencil function */ + GLenum FailFunc[3]; /**< Fail function */ + GLenum ZPassFunc[3]; /**< Depth buffer pass function */ + GLenum ZFailFunc[3]; /**< Depth buffer fail function */ + GLint Ref[3]; /**< Reference value */ + GLuint ValueMask[3]; /**< Value mask */ + GLuint WriteMask[3]; /**< Write mask */ + GLuint Clear; /**< Clear value */ +}; + + +/** + * An index for each type of texture object. These correspond to the GL + * texture target enums, such as GL_TEXTURE_2D, GL_TEXTURE_CUBE_MAP, etc. + * Note: the order is from highest priority to lowest priority. + */ +typedef enum +{ + TEXTURE_2D_ARRAY_INDEX, + TEXTURE_1D_ARRAY_INDEX, + TEXTURE_CUBE_INDEX, + TEXTURE_3D_INDEX, + TEXTURE_RECT_INDEX, + TEXTURE_2D_INDEX, + TEXTURE_1D_INDEX, + NUM_TEXTURE_TARGETS +} gl_texture_index; + + +/** + * Bit flags for each type of texture object + * Used for Texture.Unit[]._ReallyEnabled flags. + */ +/*@{*/ +#define TEXTURE_2D_ARRAY_BIT (1 << TEXTURE_2D_ARRAY_INDEX) +#define TEXTURE_1D_ARRAY_BIT (1 << TEXTURE_1D_ARRAY_INDEX) +#define TEXTURE_CUBE_BIT (1 << TEXTURE_CUBE_INDEX) +#define TEXTURE_3D_BIT (1 << TEXTURE_3D_INDEX) +#define TEXTURE_RECT_BIT (1 << TEXTURE_RECT_INDEX) +#define TEXTURE_2D_BIT (1 << TEXTURE_2D_INDEX) +#define TEXTURE_1D_BIT (1 << TEXTURE_1D_INDEX) +/*@}*/ + + +/** + * TexGenEnabled flags. + */ +/*@{*/ +#define S_BIT 1 +#define T_BIT 2 +#define R_BIT 4 +#define Q_BIT 8 +#define STR_BITS (S_BIT | T_BIT | R_BIT) +/*@}*/ + + +/** + * Bit flag versions of the corresponding GL_ constants. + */ +/*@{*/ +#define TEXGEN_SPHERE_MAP 0x1 +#define TEXGEN_OBJ_LINEAR 0x2 +#define TEXGEN_EYE_LINEAR 0x4 +#define TEXGEN_REFLECTION_MAP_NV 0x8 +#define TEXGEN_NORMAL_MAP_NV 0x10 + +#define TEXGEN_NEED_NORMALS (TEXGEN_SPHERE_MAP | \ + TEXGEN_REFLECTION_MAP_NV | \ + TEXGEN_NORMAL_MAP_NV) +#define TEXGEN_NEED_EYE_COORD (TEXGEN_SPHERE_MAP | \ + TEXGEN_REFLECTION_MAP_NV | \ + TEXGEN_NORMAL_MAP_NV | \ + TEXGEN_EYE_LINEAR) +/*@}*/ + + + +/** Tex-gen enabled for texture unit? */ +#define ENABLE_TEXGEN(unit) (1 << (unit)) + +/** Non-identity texture matrix for texture unit? */ +#define ENABLE_TEXMAT(unit) (1 << (unit)) + + +/** + * Texel fetch function prototype. We use texel fetch functions to + * extract RGBA, color indexes and depth components out of 1D, 2D and 3D + * texture images. These functions help to isolate us from the gritty + * details of all the various texture image encodings. + * + * \param texImage texture image. + * \param col texel column. + * \param row texel row. + * \param img texel image level/layer. + * \param texelOut output texel (up to 4 GLchans) + */ +typedef void (*FetchTexelFuncC)( const struct gl_texture_image *texImage, + GLint col, GLint row, GLint img, + GLchan *texelOut ); + +/** + * As above, but returns floats. + * Used for depth component images and for upcoming signed/float + * texture images. + */ +typedef void (*FetchTexelFuncF)( const struct gl_texture_image *texImage, + GLint col, GLint row, GLint img, + GLfloat *texelOut ); + + +typedef void (*StoreTexelFunc)(struct gl_texture_image *texImage, + GLint col, GLint row, GLint img, + const void *texel); + + +/** + * Texture image state. Describes the dimensions of a texture image, + * the texel format and pointers to Texel Fetch functions. + */ +struct gl_texture_image +{ + GLint InternalFormat; /**< Internal format as given by the user */ + GLenum _BaseFormat; /**< Either GL_RGB, GL_RGBA, GL_ALPHA, + * GL_LUMINANCE, GL_LUMINANCE_ALPHA, + * GL_INTENSITY, GL_COLOR_INDEX, + * GL_DEPTH_COMPONENT or GL_DEPTH_STENCIL_EXT + * only. Used for choosing TexEnv arithmetic. + */ + GLuint TexFormat; /**< The actual format: MESA_FORMAT_x */ + + GLuint Border; /**< 0 or 1 */ + GLuint Width; /**< = 2^WidthLog2 + 2*Border */ + GLuint Height; /**< = 2^HeightLog2 + 2*Border */ + GLuint Depth; /**< = 2^DepthLog2 + 2*Border */ + GLuint Width2; /**< = Width - 2*Border */ + GLuint Height2; /**< = Height - 2*Border */ + GLuint Depth2; /**< = Depth - 2*Border */ + GLuint WidthLog2; /**< = log2(Width2) */ + GLuint HeightLog2; /**< = log2(Height2) */ + GLuint DepthLog2; /**< = log2(Depth2) */ + GLuint MaxLog2; /**< = MAX(WidthLog2, HeightLog2) */ + GLfloat WidthScale; /**< used for mipmap LOD computation */ + GLfloat HeightScale; /**< used for mipmap LOD computation */ + GLfloat DepthScale; /**< used for mipmap LOD computation */ + GLboolean IsClientData; /**< Data owned by client? */ + GLboolean _IsPowerOfTwo; /**< Are all dimensions powers of two? */ + + struct gl_texture_object *TexObject; /**< Pointer back to parent object */ + + FetchTexelFuncC FetchTexelc; /**< GLchan texel fetch function pointer */ + FetchTexelFuncF FetchTexelf; /**< Float texel fetch function pointer */ + + GLuint RowStride; /**< Padded width in units of texels */ + GLuint *ImageOffsets; /**< if 3D texture: array [Depth] of offsets to + each 2D slice in 'Data', in texels */ + GLvoid *Data; /**< Image data, accessed via FetchTexel() */ + + /** + * \name For device driver: + */ + /*@{*/ + void *DriverData; /**< Arbitrary device driver data */ + /*@}*/ +}; + + +/** + * Indexes for cube map faces. + */ +typedef enum +{ + FACE_POS_X = 0, + FACE_NEG_X = 1, + FACE_POS_Y = 2, + FACE_NEG_Y = 3, + FACE_POS_Z = 4, + FACE_NEG_Z = 5, + MAX_FACES = 6 +} gl_face_index; + + +/** + * Texture object state. Contains the array of mipmap images, border color, + * wrap modes, filter modes, shadow/texcompare state, and the per-texture + * color palette. + */ +struct gl_texture_object +{ + _glthread_Mutex Mutex; /**< for thread safety */ + GLint RefCount; /**< reference count */ + GLuint Name; /**< the user-visible texture object ID */ + GLenum Target; /**< GL_TEXTURE_1D, GL_TEXTURE_2D, etc. */ + GLfloat Priority; /**< in [0,1] */ + union { + GLfloat f[4]; + GLuint ui[4]; + GLint i[4]; + } BorderColor; /**< Interpreted according to texture format */ + GLenum WrapS; /**< S-axis texture image wrap mode */ + GLenum WrapT; /**< T-axis texture image wrap mode */ + GLenum WrapR; /**< R-axis texture image wrap mode */ + GLenum MinFilter; /**< minification filter */ + GLenum MagFilter; /**< magnification filter */ + GLfloat MinLod; /**< min lambda, OpenGL 1.2 */ + GLfloat MaxLod; /**< max lambda, OpenGL 1.2 */ + GLfloat LodBias; /**< OpenGL 1.4 */ + GLint BaseLevel; /**< min mipmap level, OpenGL 1.2 */ + GLint MaxLevel; /**< max mipmap level, OpenGL 1.2 */ + GLfloat MaxAnisotropy; /**< GL_EXT_texture_filter_anisotropic */ + GLenum CompareMode; /**< GL_ARB_shadow */ + GLenum CompareFunc; /**< GL_ARB_shadow */ + GLfloat CompareFailValue; /**< GL_ARB_shadow_ambient */ + GLenum DepthMode; /**< GL_ARB_depth_texture */ + GLint _MaxLevel; /**< actual max mipmap level (q in the spec) */ + GLfloat _MaxLambda; /**< = _MaxLevel - BaseLevel (q - b in spec) */ + GLint CropRect[4]; /**< GL_OES_draw_texture */ + GLenum Swizzle[4]; /**< GL_EXT_texture_swizzle */ + GLuint _Swizzle; /**< same as Swizzle, but SWIZZLE_* format */ + GLboolean GenerateMipmap; /**< GL_SGIS_generate_mipmap */ + GLboolean _Complete; /**< Is texture object complete? */ + GLboolean _RenderToTexture; /**< Any rendering to this texture? */ + GLboolean Purgeable; /**< Is the buffer purgeable under memory pressure? */ + GLenum sRGBDecode; /**< GL_DECODE_EXT or GL_SKIP_DECODE_EXT */ + + /** Actual texture images, indexed by [cube face] and [mipmap level] */ + struct gl_texture_image *Image[MAX_FACES][MAX_TEXTURE_LEVELS]; + + /** GL_EXT_paletted_texture */ + struct gl_color_table Palette; + + /** + * \name For device driver. + * Note: instead of attaching driver data to this pointer, it's preferable + * to instead use this struct as a base class for your own texture object + * class. Driver->NewTextureObject() can be used to implement the + * allocation. + */ + void *DriverData; /**< Arbitrary device driver data */ +}; + + +/** Up to four combiner sources are possible with GL_NV_texture_env_combine4 */ +#define MAX_COMBINER_TERMS 4 + + +/** + * Texture combine environment state. + */ +struct gl_tex_env_combine_state +{ + GLenum ModeRGB; /**< GL_REPLACE, GL_DECAL, GL_ADD, etc. */ + GLenum ModeA; /**< GL_REPLACE, GL_DECAL, GL_ADD, etc. */ + /** Source terms: GL_PRIMARY_COLOR, GL_TEXTURE, etc */ + GLenum SourceRGB[MAX_COMBINER_TERMS]; + GLenum SourceA[MAX_COMBINER_TERMS]; + /** Source operands: GL_SRC_COLOR, GL_ONE_MINUS_SRC_COLOR, etc */ + GLenum OperandRGB[MAX_COMBINER_TERMS]; + GLenum OperandA[MAX_COMBINER_TERMS]; + GLuint ScaleShiftRGB; /**< 0, 1 or 2 */ + GLuint ScaleShiftA; /**< 0, 1 or 2 */ + GLuint _NumArgsRGB; /**< Number of inputs used for the RGB combiner */ + GLuint _NumArgsA; /**< Number of inputs used for the A combiner */ +}; + + +/** + * Texture coord generation state. + */ +struct gl_texgen +{ + GLenum Mode; /**< GL_EYE_LINEAR, GL_SPHERE_MAP, etc */ + GLbitfield _ModeBit; /**< TEXGEN_x bit corresponding to Mode */ + GLfloat ObjectPlane[4]; + GLfloat EyePlane[4]; +}; + + +/** + * Texture unit state. Contains enable flags, texture environment/function/ + * combiners, texgen state, pointers to current texture objects and + * post-filter color tables. + */ +struct gl_texture_unit +{ + GLbitfield Enabled; /**< bitmask of TEXTURE_*_BIT flags */ + GLbitfield _ReallyEnabled; /**< 0 or exactly one of TEXTURE_*_BIT flags */ + + GLenum EnvMode; /**< GL_MODULATE, GL_DECAL, GL_BLEND, etc. */ + GLfloat EnvColor[4]; + + struct gl_texgen GenS; + struct gl_texgen GenT; + struct gl_texgen GenR; + struct gl_texgen GenQ; + GLbitfield TexGenEnabled; /**< Bitwise-OR of [STRQ]_BIT values */ + GLbitfield _GenFlags; /**< Bitwise-OR of Gen[STRQ]._ModeBit */ + + GLfloat LodBias; /**< for biasing mipmap levels */ + GLenum BumpTarget; + GLfloat RotMatrix[4]; /* 2x2 matrix */ + + /** + * \name GL_EXT_texture_env_combine + */ + struct gl_tex_env_combine_state Combine; + + /** + * Derived state based on \c EnvMode and the \c BaseFormat of the + * currently enabled texture. + */ + struct gl_tex_env_combine_state _EnvMode; + + /** + * Currently enabled combiner state. This will point to either + * \c Combine or \c _EnvMode. + */ + struct gl_tex_env_combine_state *_CurrentCombine; + + /** Current texture object pointers */ + struct gl_texture_object *CurrentTex[NUM_TEXTURE_TARGETS]; + + /** Points to highest priority, complete and enabled texture object */ + struct gl_texture_object *_Current; + + /** GL_SGI_texture_color_table */ + /*@{*/ + struct gl_color_table ColorTable; + struct gl_color_table ProxyColorTable; + GLboolean ColorTableEnabled; + /*@}*/ +}; + + +/** + * Texture attribute group (GL_TEXTURE_BIT). + */ +struct gl_texture_attrib +{ + GLuint CurrentUnit; /**< GL_ACTIVE_TEXTURE */ + struct gl_texture_unit Unit[MAX_COMBINED_TEXTURE_IMAGE_UNITS]; + + struct gl_texture_object *ProxyTex[NUM_TEXTURE_TARGETS]; + + /** GL_ARB_seamless_cubemap */ + GLboolean CubeMapSeamless; + + /** GL_EXT_shared_texture_palette */ + GLboolean SharedPalette; + struct gl_color_table Palette; + + /** Texture units/samplers used by vertex or fragment texturing */ + GLbitfield _EnabledUnits; + + /** Texture coord units/sets used for fragment texturing */ + GLbitfield _EnabledCoordUnits; + + /** Texture coord units that have texgen enabled */ + GLbitfield _TexGenEnabled; + + /** Texture coord units that have non-identity matrices */ + GLbitfield _TexMatEnabled; + + /** Bitwise-OR of all Texture.Unit[i]._GenFlags */ + GLbitfield _GenFlags; +}; + + +/** + * Transformation attribute group (GL_TRANSFORM_BIT). + */ +struct gl_transform_attrib +{ + GLenum MatrixMode; /**< Matrix mode */ + GLfloat EyeUserPlane[MAX_CLIP_PLANES][4]; /**< User clip planes */ + GLfloat _ClipUserPlane[MAX_CLIP_PLANES][4]; /**< derived */ + GLbitfield ClipPlanesEnabled; /**< on/off bitmask */ + GLboolean Normalize; /**< Normalize all normals? */ + GLboolean RescaleNormals; /**< GL_EXT_rescale_normal */ + GLboolean RasterPositionUnclipped; /**< GL_IBM_rasterpos_clip */ + GLboolean DepthClamp; /**< GL_ARB_depth_clamp */ + + GLfloat CullEyePos[4]; + GLfloat CullObjPos[4]; +}; + + +/** + * Viewport attribute group (GL_VIEWPORT_BIT). + */ +struct gl_viewport_attrib +{ + GLint X, Y; /**< position */ + GLsizei Width, Height; /**< size */ + GLfloat Near, Far; /**< Depth buffer range */ + GLmatrix _WindowMap; /**< Mapping transformation as a matrix. */ +}; + + +/** + * GL_ARB_vertex/pixel_buffer_object buffer object + */ +struct gl_buffer_object +{ + _glthread_Mutex Mutex; + GLint RefCount; + GLuint Name; + GLenum Usage; /**< GL_STREAM_DRAW_ARB, GL_STREAM_READ_ARB, etc. */ + GLsizeiptrARB Size; /**< Size of buffer storage in bytes */ + GLubyte *Data; /**< Location of storage either in RAM or VRAM. */ + /** Fields describing a mapped buffer */ + /*@{*/ + GLbitfield AccessFlags; /**< Mask of GL_MAP_x_BIT flags */ + GLvoid *Pointer; /**< User-space address of mapping */ + GLintptr Offset; /**< Mapped offset */ + GLsizeiptr Length; /**< Mapped length */ + /*@}*/ + GLboolean Written; /**< Ever written to? (for debugging) */ + GLboolean Purgeable; /**< Is the buffer purgeable under memory pressure? */ +}; + + +/** + * Client pixel packing/unpacking attributes + */ +struct gl_pixelstore_attrib +{ + GLint Alignment; + GLint RowLength; + GLint SkipPixels; + GLint SkipRows; + GLint ImageHeight; + GLint SkipImages; + GLboolean SwapBytes; + GLboolean LsbFirst; + GLboolean ClientStorage; /**< GL_APPLE_client_storage */ + GLboolean Invert; /**< GL_MESA_pack_invert */ + struct gl_buffer_object *BufferObj; /**< GL_ARB_pixel_buffer_object */ +}; + + +/** + * Client vertex array attributes + */ +struct gl_client_array +{ + GLint Size; /**< components per element (1,2,3,4) */ + GLenum Type; /**< datatype: GL_FLOAT, GL_INT, etc */ + GLenum Format; /**< default: GL_RGBA, but may be GL_BGRA */ + GLsizei Stride; /**< user-specified stride */ + GLsizei StrideB; /**< actual stride in bytes */ + const GLubyte *Ptr; /**< Points to array data */ + GLboolean Enabled; /**< Enabled flag is a boolean */ + GLboolean Normalized; /**< GL_ARB_vertex_program */ + GLboolean Integer; /**< Integer-valued? */ + GLuint InstanceDivisor; /**< GL_ARB_instanced_arrays */ + GLuint _ElementSize; /**< size of each element in bytes */ + + struct gl_buffer_object *BufferObj;/**< GL_ARB_vertex_buffer_object */ + GLuint _MaxElement; /**< max element index into array buffer + 1 */ +}; + + +/** + * Collection of vertex arrays. Defined by the GL_APPLE_vertex_array_object + * extension, but a nice encapsulation in any case. + */ +struct gl_array_object +{ + /** Name of the array object as received from glGenVertexArrayAPPLE. */ + GLuint Name; + + GLint RefCount; + _glthread_Mutex Mutex; + GLboolean VBOonly; /**< require all arrays to live in VBOs? */ + + /** Conventional vertex arrays */ + /*@{*/ + struct gl_client_array Vertex; + struct gl_client_array Weight; + struct gl_client_array Normal; + struct gl_client_array Color; + struct gl_client_array SecondaryColor; + struct gl_client_array FogCoord; + struct gl_client_array Index; + struct gl_client_array EdgeFlag; + struct gl_client_array TexCoord[MAX_TEXTURE_COORD_UNITS]; + struct gl_client_array PointSize; + /*@}*/ + + /** + * Generic arrays for vertex programs/shaders. + * For NV vertex programs, these attributes alias and take priority + * over the conventional attribs above. For ARB vertex programs and + * GLSL vertex shaders, these attributes are separate. + */ + struct gl_client_array VertexAttrib[MAX_VERTEX_GENERIC_ATTRIBS]; + + /** Mask of _NEW_ARRAY_* values indicating which arrays are enabled */ + GLbitfield _Enabled; + + /** + * Min of all enabled arrays' _MaxElement. When arrays reside inside VBOs + * we can determine the max legal (in bounds) glDrawElements array index. + */ + GLuint _MaxElement; +}; + + +/** + * Vertex array state + */ +struct gl_array_attrib +{ + /** Currently bound array object. See _mesa_BindVertexArrayAPPLE() */ + struct gl_array_object *ArrayObj; + + /** The default vertex array object */ + struct gl_array_object *DefaultArrayObj; + + /** Array objects (GL_ARB/APPLE_vertex_array_object) */ + struct _mesa_HashTable *Objects; + + GLint ActiveTexture; /**< Client Active Texture */ + GLuint LockFirst; /**< GL_EXT_compiled_vertex_array */ + GLuint LockCount; /**< GL_EXT_compiled_vertex_array */ + + /** GL 3.1 (slightly different from GL_NV_primitive_restart) */ + GLboolean PrimitiveRestart; + GLuint RestartIndex; + + GLbitfield NewState; /**< mask of _NEW_ARRAY_* values */ + + /* GL_ARB_vertex_buffer_object */ + struct gl_buffer_object *ArrayBufferObj; + struct gl_buffer_object *ElementArrayBufferObj; +}; + + +/** + * Feedback buffer state + */ +struct gl_feedback +{ + GLenum Type; + GLbitfield _Mask; /**< FB_* bits */ + GLfloat *Buffer; + GLuint BufferSize; + GLuint Count; +}; + + +/** + * Selection buffer state + */ +struct gl_selection +{ + GLuint *Buffer; /**< selection buffer */ + GLuint BufferSize; /**< size of the selection buffer */ + GLuint BufferCount; /**< number of values in the selection buffer */ + GLuint Hits; /**< number of records in the selection buffer */ + GLuint NameStackDepth; /**< name stack depth */ + GLuint NameStack[MAX_NAME_STACK_DEPTH]; /**< name stack */ + GLboolean HitFlag; /**< hit flag */ + GLfloat HitMinZ; /**< minimum hit depth */ + GLfloat HitMaxZ; /**< maximum hit depth */ +}; + + +/** + * 1-D Evaluator control points + */ +struct gl_1d_map +{ + GLuint Order; /**< Number of control points */ + GLfloat u1, u2, du; /**< u1, u2, 1.0/(u2-u1) */ + GLfloat *Points; /**< Points to contiguous control points */ +}; + + +/** + * 2-D Evaluator control points + */ +struct gl_2d_map +{ + GLuint Uorder; /**< Number of control points in U dimension */ + GLuint Vorder; /**< Number of control points in V dimension */ + GLfloat u1, u2, du; + GLfloat v1, v2, dv; + GLfloat *Points; /**< Points to contiguous control points */ +}; + + +/** + * All evaluator control point state + */ +struct gl_evaluators +{ + /** + * \name 1-D maps + */ + /*@{*/ + struct gl_1d_map Map1Vertex3; + struct gl_1d_map Map1Vertex4; + struct gl_1d_map Map1Index; + struct gl_1d_map Map1Color4; + struct gl_1d_map Map1Normal; + struct gl_1d_map Map1Texture1; + struct gl_1d_map Map1Texture2; + struct gl_1d_map Map1Texture3; + struct gl_1d_map Map1Texture4; + struct gl_1d_map Map1Attrib[16]; /**< GL_NV_vertex_program */ + /*@}*/ + + /** + * \name 2-D maps + */ + /*@{*/ + struct gl_2d_map Map2Vertex3; + struct gl_2d_map Map2Vertex4; + struct gl_2d_map Map2Index; + struct gl_2d_map Map2Color4; + struct gl_2d_map Map2Normal; + struct gl_2d_map Map2Texture1; + struct gl_2d_map Map2Texture2; + struct gl_2d_map Map2Texture3; + struct gl_2d_map Map2Texture4; + struct gl_2d_map Map2Attrib[16]; /**< GL_NV_vertex_program */ + /*@}*/ +}; + + +/** + * Names of the various vertex/fragment program register files, etc. + * + * NOTE: first four tokens must fit into 2 bits (see t_vb_arbprogram.c) + * All values should fit in a 4-bit field. + * + * NOTE: PROGRAM_ENV_PARAM, PROGRAM_STATE_VAR, PROGRAM_NAMED_PARAM, + * PROGRAM_CONSTANT, and PROGRAM_UNIFORM can all be considered to + * be "uniform" variables since they can only be set outside glBegin/End. + * They're also all stored in the same Parameters array. + */ +typedef enum +{ + PROGRAM_TEMPORARY, /**< machine->Temporary[] */ + PROGRAM_INPUT, /**< machine->Inputs[] */ + PROGRAM_OUTPUT, /**< machine->Outputs[] */ + PROGRAM_VARYING, /**< machine->Inputs[]/Outputs[] */ + PROGRAM_LOCAL_PARAM, /**< gl_program->LocalParams[] */ + PROGRAM_ENV_PARAM, /**< gl_program->Parameters[] */ + PROGRAM_STATE_VAR, /**< gl_program->Parameters[] */ + PROGRAM_NAMED_PARAM, /**< gl_program->Parameters[] */ + PROGRAM_CONSTANT, /**< gl_program->Parameters[] */ + PROGRAM_UNIFORM, /**< gl_program->Parameters[] */ + PROGRAM_WRITE_ONLY, /**< A dummy, write-only register */ + PROGRAM_ADDRESS, /**< machine->AddressReg */ + PROGRAM_SAMPLER, /**< for shader samplers, compile-time only */ + PROGRAM_SYSTEM_VALUE,/**< InstanceId, PrimitiveID, etc. */ + PROGRAM_UNDEFINED, /**< Invalid/TBD value */ + PROGRAM_FILE_MAX +} gl_register_file; + + +/** + * If the register file is PROGRAM_SYSTEM_VALUE, the register index will be + * one of these values. + */ +typedef enum +{ + SYSTEM_VALUE_FRONT_FACE, /**< Fragment shader only (not done yet) */ + SYSTEM_VALUE_INSTANCE_ID, /**< Vertex shader only */ + SYSTEM_VALUE_MAX /**< Number of values */ +} gl_system_value; + + +/** Vertex and fragment instructions */ +struct prog_instruction; +struct gl_program_parameter_list; +struct gl_uniform_list; + + +/** + * Base class for any kind of program object + */ +struct gl_program +{ + GLuint Id; + GLubyte *String; /**< Null-terminated program text */ + GLint RefCount; + GLenum Target; /**< GL_VERTEX/FRAGMENT_PROGRAM_ARB, GL_FRAGMENT_PROGRAM_NV */ + GLenum Format; /**< String encoding format */ + GLboolean Resident; + + struct prog_instruction *Instructions; + + GLbitfield InputsRead; /**< Bitmask of which input regs are read */ + GLbitfield64 OutputsWritten; /**< Bitmask of which output regs are written */ + GLbitfield SystemValuesRead; /**< Bitmask of SYSTEM_VALUE_x inputs used */ + GLbitfield InputFlags[MAX_PROGRAM_INPUTS]; /**< PROG_PARAM_BIT_x flags */ + GLbitfield OutputFlags[MAX_PROGRAM_OUTPUTS]; /**< PROG_PARAM_BIT_x flags */ + GLbitfield TexturesUsed[MAX_TEXTURE_UNITS]; /**< TEXTURE_x_BIT bitmask */ + GLbitfield SamplersUsed; /**< Bitfield of which samplers are used */ + GLbitfield ShadowSamplers; /**< Texture units used for shadow sampling. */ + + + /** Named parameters, constants, etc. from program text */ + struct gl_program_parameter_list *Parameters; + /** Numbered local parameters */ + GLfloat LocalParams[MAX_PROGRAM_LOCAL_PARAMS][4]; + + /** Vertex/fragment shader varying vars */ + struct gl_program_parameter_list *Varying; + /** Vertex program user-defined attributes */ + struct gl_program_parameter_list *Attributes; + + /** Map from sampler unit to texture unit (set by glUniform1i()) */ + GLubyte SamplerUnits[MAX_SAMPLERS]; + /** Which texture target is being sampled (TEXTURE_1D/2D/3D/etc_INDEX) */ + gl_texture_index SamplerTargets[MAX_SAMPLERS]; + + /** Bitmask of which register files are read/written with indirect + * addressing. Mask of (1 << PROGRAM_x) bits. + */ + GLbitfield IndirectRegisterFiles; + + /** Logical counts */ + /*@{*/ + GLuint NumInstructions; + GLuint NumTemporaries; + GLuint NumParameters; + GLuint NumAttributes; + GLuint NumAddressRegs; + GLuint NumAluInstructions; + GLuint NumTexInstructions; + GLuint NumTexIndirections; + /*@}*/ + /** Native, actual h/w counts */ + /*@{*/ + GLuint NumNativeInstructions; + GLuint NumNativeTemporaries; + GLuint NumNativeParameters; + GLuint NumNativeAttributes; + GLuint NumNativeAddressRegs; + GLuint NumNativeAluInstructions; + GLuint NumNativeTexInstructions; + GLuint NumNativeTexIndirections; + /*@}*/ +}; + + +/** Vertex program object */ +struct gl_vertex_program +{ + struct gl_program Base; /**< base class */ + GLboolean IsNVProgram; /**< is this a GL_NV_vertex_program program? */ + GLboolean IsPositionInvariant; +}; + + +/** Geometry program object */ +struct gl_geometry_program +{ + struct gl_program Base; /**< base class */ + + GLint VerticesOut; + GLenum InputType; /**< GL_POINTS, GL_LINES, GL_LINES_ADJACENCY_ARB, + GL_TRIANGLES, or GL_TRIANGLES_ADJACENCY_ARB */ + GLenum OutputType; /**< GL_POINTS, GL_LINE_STRIP or GL_TRIANGLE_STRIP */ +}; + + +/** Fragment program object */ +struct gl_fragment_program +{ + struct gl_program Base; /**< base class */ + GLenum FogOption; + GLboolean UsesKill; /**< shader uses KIL instruction */ + GLboolean OriginUpperLeft; + GLboolean PixelCenterInteger; + enum gl_frag_depth_layout FragDepthLayout; +}; + + +/** + * State common to vertex and fragment programs. + */ +struct gl_program_state +{ + GLint ErrorPos; /* GL_PROGRAM_ERROR_POSITION_ARB/NV */ + const char *ErrorString; /* GL_PROGRAM_ERROR_STRING_ARB/NV */ +}; + + +/** + * Context state for vertex programs. + */ +struct gl_vertex_program_state +{ + GLboolean Enabled; /**< User-set GL_VERTEX_PROGRAM_ARB/NV flag */ + GLboolean _Enabled; /**< Enabled and _valid_ user program? */ + GLboolean PointSizeEnabled; /**< GL_VERTEX_PROGRAM_POINT_SIZE_ARB/NV */ + GLboolean TwoSideEnabled; /**< GL_VERTEX_PROGRAM_TWO_SIDE_ARB/NV */ + struct gl_vertex_program *Current; /**< User-bound vertex program */ + + /** Currently enabled and valid vertex program (including internal + * programs, user-defined vertex programs and GLSL vertex shaders). + * This is the program we must use when rendering. + */ + struct gl_vertex_program *_Current; + + GLfloat Parameters[MAX_PROGRAM_ENV_PARAMS][4]; /**< Env params */ + + /* For GL_NV_vertex_program only: */ + GLenum TrackMatrix[MAX_PROGRAM_ENV_PARAMS / 4]; + GLenum TrackMatrixTransform[MAX_PROGRAM_ENV_PARAMS / 4]; + + /** Should fixed-function T&L be implemented with a vertex prog? */ + GLboolean _MaintainTnlProgram; + + /** Program to emulate fixed-function T&L (see above) */ + struct gl_vertex_program *_TnlProgram; + + /** Cache of fixed-function programs */ + struct gl_program_cache *Cache; + + GLboolean _Overriden; +}; + + +/** + * Context state for geometry programs. + */ +struct gl_geometry_program_state +{ + GLboolean Enabled; /**< GL_ARB_GEOMETRY_SHADER4 */ + GLboolean _Enabled; /**< Enabled and valid program? */ + struct gl_geometry_program *Current; /**< user-bound geometry program */ + + /** Currently enabled and valid program (including internal programs + * and compiled shader programs). + */ + struct gl_geometry_program *_Current; + + GLfloat Parameters[MAX_PROGRAM_ENV_PARAMS][4]; /**< Env params */ + + /** Cache of fixed-function programs */ + struct gl_program_cache *Cache; +}; + +/** + * Context state for fragment programs. + */ +struct gl_fragment_program_state +{ + GLboolean Enabled; /**< User-set fragment program enable flag */ + GLboolean _Enabled; /**< Enabled and _valid_ user program? */ + struct gl_fragment_program *Current; /**< User-bound fragment program */ + + /** Currently enabled and valid fragment program (including internal + * programs, user-defined fragment programs and GLSL fragment shaders). + * This is the program we must use when rendering. + */ + struct gl_fragment_program *_Current; + + GLfloat Parameters[MAX_PROGRAM_ENV_PARAMS][4]; /**< Env params */ + + /** Should fixed-function texturing be implemented with a fragment prog? */ + GLboolean _MaintainTexEnvProgram; + + /** Program to emulate fixed-function texture env/combine (see above) */ + struct gl_fragment_program *_TexEnvProgram; + + /** Cache of fixed-function programs */ + struct gl_program_cache *Cache; +}; + + +/** + * ATI_fragment_shader runtime state + */ +#define ATI_FS_INPUT_PRIMARY 0 +#define ATI_FS_INPUT_SECONDARY 1 + +struct atifs_instruction; +struct atifs_setupinst; + +/** + * ATI fragment shader + */ +struct ati_fragment_shader +{ + GLuint Id; + GLint RefCount; + struct atifs_instruction *Instructions[2]; + struct atifs_setupinst *SetupInst[2]; + GLfloat Constants[8][4]; + GLbitfield LocalConstDef; /**< Indicates which constants have been set */ + GLubyte numArithInstr[2]; + GLubyte regsAssigned[2]; + GLubyte NumPasses; /**< 1 or 2 */ + GLubyte cur_pass; + GLubyte last_optype; + GLboolean interpinp1; + GLboolean isValid; + GLuint swizzlerq; +}; + +/** + * Context state for GL_ATI_fragment_shader + */ +struct gl_ati_fragment_shader_state +{ + GLboolean Enabled; + GLboolean _Enabled; /**< enabled and valid shader? */ + GLboolean Compiling; + GLfloat GlobalConstants[8][4]; + struct ati_fragment_shader *Current; +}; + + +/** + * Occlusion/timer query object. + */ +struct gl_query_object +{ + GLenum Target; /**< The query target, when active */ + GLuint Id; /**< hash table ID/name */ + GLuint64EXT Result; /**< the counter */ + GLboolean Active; /**< inside Begin/EndQuery */ + GLboolean Ready; /**< result is ready? */ +}; + + +/** + * Context state for query objects. + */ +struct gl_query_state +{ + struct _mesa_HashTable *QueryObjects; + struct gl_query_object *CurrentOcclusionObject; /* GL_ARB_occlusion_query */ + struct gl_query_object *CurrentTimerObject; /* GL_EXT_timer_query */ + + /** GL_NV_conditional_render */ + struct gl_query_object *CondRenderQuery; + + /** GL_EXT_transform_feedback */ + struct gl_query_object *PrimitivesGenerated; + struct gl_query_object *PrimitivesWritten; + + /** GL_ARB_timer_query */ + struct gl_query_object *TimeElapsed; + + GLenum CondRenderMode; +}; + + +/** Sync object state */ +struct gl_sync_object { + struct simple_node link; + GLenum Type; /**< GL_SYNC_FENCE */ + GLuint Name; /**< Fence name */ + GLint RefCount; /**< Reference count */ + GLboolean DeletePending; /**< Object was deleted while there were still + * live references (e.g., sync not yet finished) + */ + GLenum SyncCondition; + GLbitfield Flags; /**< Flags passed to glFenceSync */ + GLuint StatusFlag:1; /**< Has the sync object been signaled? */ +}; + + +/** Set by #pragma directives */ +struct gl_sl_pragmas +{ + GLboolean IgnoreOptimize; /**< ignore #pragma optimize(on/off) ? */ + GLboolean IgnoreDebug; /**< ignore #pragma debug(on/off) ? */ + GLboolean Optimize; /**< defaults on */ + GLboolean Debug; /**< defaults off */ +}; + + +/** + * A GLSL vertex or fragment shader object. + */ +struct gl_shader +{ + GLenum Type; /**< GL_FRAGMENT_SHADER || GL_VERTEX_SHADER || GL_GEOMETRY_SHADER_ARB (first field!) */ + GLuint Name; /**< AKA the handle */ + GLint RefCount; /**< Reference count */ + GLboolean DeletePending; + GLboolean CompileStatus; + const GLchar *Source; /**< Source code string */ + GLuint SourceChecksum; /**< for debug/logging purposes */ + struct gl_program *Program; /**< Post-compile assembly code */ + GLchar *InfoLog; + struct gl_sl_pragmas Pragmas; + + unsigned Version; /**< GLSL version used for linking */ + + struct exec_list *ir; + struct glsl_symbol_table *symbols; + + /** Shaders containing built-in functions that are used for linking. */ + struct gl_shader *builtins_to_link[16]; + unsigned num_builtins_to_link; +}; + + +/** + * A GLSL program object. + * Basically a linked collection of vertex and fragment shaders. + */ +struct gl_shader_program +{ + GLenum Type; /**< Always GL_SHADER_PROGRAM (internal token) */ + GLuint Name; /**< aka handle or ID */ + GLint RefCount; /**< Reference count */ + GLboolean DeletePending; + + GLuint NumShaders; /**< number of attached shaders */ + struct gl_shader **Shaders; /**< List of attached the shaders */ + + /** User-defined attribute bindings (glBindAttribLocation) */ + struct gl_program_parameter_list *Attributes; + + /** Transform feedback varyings */ + struct { + GLenum BufferMode; + GLuint NumVarying; + GLchar **VaryingNames; /**< Array [NumVarying] of char * */ + } TransformFeedback; + + /** Geometry shader state - copied into gl_geometry_program at link time */ + struct { + GLint VerticesOut; + GLenum InputType; /**< GL_POINTS, GL_LINES, GL_LINES_ADJACENCY_ARB, + GL_TRIANGLES, or GL_TRIANGLES_ADJACENCY_ARB */ + GLenum OutputType; /**< GL_POINTS, GL_LINE_STRIP or GL_TRIANGLE_STRIP */ + } Geom; + + /* post-link info: */ + struct gl_vertex_program *VertexProgram; /**< Linked vertex program */ + struct gl_fragment_program *FragmentProgram; /**< Linked fragment prog */ + struct gl_geometry_program *GeometryProgram; /**< Linked geometry prog */ + struct gl_uniform_list *Uniforms; + struct gl_program_parameter_list *Varying; + GLboolean LinkStatus; /**< GL_LINK_STATUS */ + GLboolean Validated; + GLboolean _Used; /**< Ever used for drawing? */ + GLchar *InfoLog; + + unsigned Version; /**< GLSL version used for linking */ + + /** + * Per-stage shaders resulting from the first stage of linking. + * + * Set of linked shaders for this program. The array is accessed using the + * \c MESA_SHADER_* defines. Entries for non-existent stages will be + * \c NULL. + */ + struct gl_shader *_LinkedShaders[MESA_SHADER_TYPES]; +}; + + +#define GLSL_DUMP 0x1 /**< Dump shaders to stdout */ +#define GLSL_LOG 0x2 /**< Write shaders to files */ +#define GLSL_OPT 0x4 /**< Force optimizations (override pragmas) */ +#define GLSL_NO_OPT 0x8 /**< Force no optimizations (override pragmas) */ +#define GLSL_UNIFORMS 0x10 /**< Print glUniform calls */ +#define GLSL_NOP_VERT 0x20 /**< Force no-op vertex shaders */ +#define GLSL_NOP_FRAG 0x40 /**< Force no-op fragment shaders */ +#define GLSL_USE_PROG 0x80 /**< Log glUseProgram calls */ + + +/** + * Context state for GLSL vertex/fragment shaders. + */ +struct gl_shader_state +{ + /** + * Programs used for rendering + * + * There is a separate program set for each shader stage. If + * GL_EXT_separate_shader_objects is not supported, each of these must point + * to \c NULL or to the same program. + */ + struct gl_shader_program *CurrentVertexProgram; + struct gl_shader_program *CurrentGeometryProgram; + struct gl_shader_program *CurrentFragmentProgram; + + /** + * Program used by glUniform calls. + * + * Explicitly set by \c glUseProgram and \c glActiveProgramEXT. + */ + struct gl_shader_program *ActiveProgram; + + void *MemPool; + + GLbitfield Flags; /**< Mask of GLSL_x flags */ +}; + +/** + * Compiler options for a single GLSL shaders type + */ +struct gl_shader_compiler_options +{ + /** Driver-selectable options: */ + GLboolean EmitCondCodes; /**< Use condition codes? */ + GLboolean EmitNVTempInitialization; /**< 0-fill NV temp registers */ + /** + * Attempts to flatten all ir_if (OPCODE_IF) for GPUs that can't + * support control flow. + */ + GLboolean EmitNoIfs; + GLboolean EmitNoLoops; + GLboolean EmitNoFunctions; + GLboolean EmitNoCont; /**< Emit CONT opcode? */ + GLboolean EmitNoMainReturn; /**< Emit CONT/RET opcodes? */ + GLboolean EmitNoNoise; /**< Emit NOISE opcodes? */ + GLboolean EmitNoPow; /**< Emit POW opcodes? */ + + /** + * \name Forms of indirect addressing the driver cannot do. + */ + /*@{*/ + GLboolean EmitNoIndirectInput; /**< No indirect addressing of inputs */ + GLboolean EmitNoIndirectOutput; /**< No indirect addressing of outputs */ + GLboolean EmitNoIndirectTemp; /**< No indirect addressing of temps */ + GLboolean EmitNoIndirectUniform; /**< No indirect addressing of constants */ + /*@}*/ + + GLuint MaxUnrollIterations; + + struct gl_sl_pragmas DefaultPragmas; /**< Default #pragma settings */ +}; + +/** + * Transform feedback object state + */ +struct gl_transform_feedback_object +{ + GLuint Name; /**< AKA the object ID */ + GLint RefCount; + GLboolean Active; /**< Is transform feedback enabled? */ + GLboolean Paused; /**< Is transform feedback paused? */ + + /** The feedback buffers */ + GLuint BufferNames[MAX_FEEDBACK_ATTRIBS]; + struct gl_buffer_object *Buffers[MAX_FEEDBACK_ATTRIBS]; + + /** Start of feedback data in dest buffer */ + GLintptr Offset[MAX_FEEDBACK_ATTRIBS]; + /** Max data to put into dest buffer (in bytes) */ + GLsizeiptr Size[MAX_FEEDBACK_ATTRIBS]; +}; + + +/** + * Context state for transform feedback. + */ +struct gl_transform_feedback +{ + GLenum Mode; /**< GL_POINTS, GL_LINES or GL_TRIANGLES */ + + GLboolean RasterDiscard; /**< GL_RASTERIZER_DISCARD */ + + /** The general binding point (GL_TRANSFORM_FEEDBACK_BUFFER) */ + struct gl_buffer_object *CurrentBuffer; + + /** The table of all transform feedback objects */ + struct _mesa_HashTable *Objects; + + /** The current xform-fb object (GL_TRANSFORM_FEEDBACK_BINDING) */ + struct gl_transform_feedback_object *CurrentObject; + + /** The default xform-fb object (Name==0) */ + struct gl_transform_feedback_object *DefaultObject; +}; + + + +/** + * State which can be shared by multiple contexts: + */ +struct gl_shared_state +{ + _glthread_Mutex Mutex; /**< for thread safety */ + GLint RefCount; /**< Reference count */ + struct _mesa_HashTable *DisplayList; /**< Display lists hash table */ + struct _mesa_HashTable *TexObjects; /**< Texture objects hash table */ + + /** Default texture objects (shared by all texture units) */ + struct gl_texture_object *DefaultTex[NUM_TEXTURE_TARGETS]; + + /** Fallback texture used when a bound texture is incomplete */ + struct gl_texture_object *FallbackTex; + + /** + * \name Thread safety and statechange notification for texture + * objects. + * + * \todo Improve the granularity of locking. + */ + /*@{*/ + _glthread_Mutex TexMutex; /**< texobj thread safety */ + GLuint TextureStateStamp; /**< state notification for shared tex */ + /*@}*/ + + /** Default buffer object for vertex arrays that aren't in VBOs */ + struct gl_buffer_object *NullBufferObj; + + /** + * \name Vertex/geometry/fragment programs + */ + /*@{*/ + struct _mesa_HashTable *Programs; /**< All vertex/fragment programs */ + struct gl_vertex_program *DefaultVertexProgram; + struct gl_fragment_program *DefaultFragmentProgram; + struct gl_geometry_program *DefaultGeometryProgram; + /*@}*/ + + /* GL_ATI_fragment_shader */ + struct _mesa_HashTable *ATIShaders; + struct ati_fragment_shader *DefaultFragmentShader; + + struct _mesa_HashTable *BufferObjects; + + /** Table of both gl_shader and gl_shader_program objects */ + struct _mesa_HashTable *ShaderObjects; + + /* GL_EXT_framebuffer_object */ + struct _mesa_HashTable *RenderBuffers; + struct _mesa_HashTable *FrameBuffers; + + /* GL_ARB_sync */ + struct simple_node SyncObjects; + + void *DriverData; /**< Device driver shared state */ +}; + + + + +/** + * A renderbuffer stores colors or depth values or stencil values. + * A framebuffer object will have a collection of these. + * Data are read/written to the buffer with a handful of Get/Put functions. + * + * Instances of this object are allocated with the Driver's NewRenderbuffer + * hook. Drivers will likely wrap this class inside a driver-specific + * class to simulate inheritance. + */ +struct gl_renderbuffer +{ +#define RB_MAGIC 0xaabbccdd + int Magic; /** XXX TEMPORARY DEBUG INFO */ + _glthread_Mutex Mutex; /**< for thread safety */ + GLuint ClassID; /**< Useful for drivers */ + GLuint Name; + GLint RefCount; + GLuint Width, Height; + GLboolean Purgeable; /**< Is the buffer purgeable under memory pressure? */ + + GLenum InternalFormat; /**< The user-specified format */ + GLenum _BaseFormat; /**< Either GL_RGB, GL_RGBA, GL_DEPTH_COMPONENT or + GL_STENCIL_INDEX. */ + GLuint Format; /**< The actual format: MESA_FORMAT_x */ + + GLubyte NumSamples; + + GLenum DataType; /**< Type of values passed to the Get/Put functions */ + GLvoid *Data; /**< This may not be used by some kinds of RBs */ + + /* Used to wrap one renderbuffer around another: */ + struct gl_renderbuffer *Wrapped; + + /* Delete this renderbuffer */ + void (*Delete)(struct gl_renderbuffer *rb); + + /* Allocate new storage for this renderbuffer */ + GLboolean (*AllocStorage)(struct gl_context *ctx, struct gl_renderbuffer *rb, + GLenum internalFormat, + GLuint width, GLuint height); + + /* Lock/Unlock are called before/after calling the Get/Put functions. + * Not sure this is the right place for these yet. + void (*Lock)(struct gl_context *ctx, struct gl_renderbuffer *rb); + void (*Unlock)(struct gl_context *ctx, struct gl_renderbuffer *rb); + */ + + /* Return a pointer to the element/pixel at (x,y). + * Should return NULL if the buffer memory can't be directly addressed. + */ + void *(*GetPointer)(struct gl_context *ctx, struct gl_renderbuffer *rb, + GLint x, GLint y); + + /* Get/Read a row of values. + * The values will be of format _BaseFormat and type DataType. + */ + void (*GetRow)(struct gl_context *ctx, struct gl_renderbuffer *rb, GLuint count, + GLint x, GLint y, void *values); + + /* Get/Read values at arbitrary locations. + * The values will be of format _BaseFormat and type DataType. + */ + void (*GetValues)(struct gl_context *ctx, struct gl_renderbuffer *rb, GLuint count, + const GLint x[], const GLint y[], void *values); + + /* Put/Write a row of values. + * The values will be of format _BaseFormat and type DataType. + */ + void (*PutRow)(struct gl_context *ctx, struct gl_renderbuffer *rb, GLuint count, + GLint x, GLint y, const void *values, const GLubyte *mask); + + /* Put/Write a row of RGB values. This is a special-case routine that's + * only used for RGBA renderbuffers when the source data is GL_RGB. That's + * a common case for glDrawPixels and some triangle routines. + * The values will be of format GL_RGB and type DataType. + */ + void (*PutRowRGB)(struct gl_context *ctx, struct gl_renderbuffer *rb, GLuint count, + GLint x, GLint y, const void *values, const GLubyte *mask); + + + /* Put/Write a row of identical values. + * The values will be of format _BaseFormat and type DataType. + */ + void (*PutMonoRow)(struct gl_context *ctx, struct gl_renderbuffer *rb, GLuint count, + GLint x, GLint y, const void *value, const GLubyte *mask); + + /* Put/Write values at arbitrary locations. + * The values will be of format _BaseFormat and type DataType. + */ + void (*PutValues)(struct gl_context *ctx, struct gl_renderbuffer *rb, GLuint count, + const GLint x[], const GLint y[], const void *values, + const GLubyte *mask); + /* Put/Write identical values at arbitrary locations. + * The values will be of format _BaseFormat and type DataType. + */ + void (*PutMonoValues)(struct gl_context *ctx, struct gl_renderbuffer *rb, + GLuint count, const GLint x[], const GLint y[], + const void *value, const GLubyte *mask); +}; + + +/** + * A renderbuffer attachment points to either a texture object (and specifies + * a mipmap level, cube face or 3D texture slice) or points to a renderbuffer. + */ +struct gl_renderbuffer_attachment +{ + GLenum Type; /**< \c GL_NONE or \c GL_TEXTURE or \c GL_RENDERBUFFER_EXT */ + GLboolean Complete; + + /** + * If \c Type is \c GL_RENDERBUFFER_EXT, this stores a pointer to the + * application supplied renderbuffer object. + */ + struct gl_renderbuffer *Renderbuffer; + + /** + * If \c Type is \c GL_TEXTURE, this stores a pointer to the application + * supplied texture object. + */ + struct gl_texture_object *Texture; + GLuint TextureLevel; /**< Attached mipmap level. */ + GLuint CubeMapFace; /**< 0 .. 5, for cube map textures. */ + GLuint Zoffset; /**< Slice for 3D textures, or layer for both 1D + * and 2D array textures */ +}; + + +/** + * A framebuffer is a collection of renderbuffers (color, depth, stencil, etc). + * In C++ terms, think of this as a base class from which device drivers + * will make derived classes. + */ +struct gl_framebuffer +{ + _glthread_Mutex Mutex; /**< for thread safety */ + /** + * If zero, this is a window system framebuffer. If non-zero, this + * is a FBO framebuffer; note that for some devices (i.e. those with + * a natural pixel coordinate system for FBOs that differs from the + * OpenGL/Mesa coordinate system), this means that the viewport, + * polygon face orientation, and polygon stipple will have to be inverted. + */ + GLuint Name; + + GLint RefCount; + GLboolean DeletePending; + + /** + * The framebuffer's visual. Immutable if this is a window system buffer. + * Computed from attachments if user-made FBO. + */ + struct gl_config Visual; + + GLboolean Initialized; + + GLuint Width, Height; /**< size of frame buffer in pixels */ + + /** \name Drawing bounds (Intersection of buffer size and scissor box) */ + /*@{*/ + GLint _Xmin, _Xmax; /**< inclusive */ + GLint _Ymin, _Ymax; /**< exclusive */ + /*@}*/ + + /** \name Derived Z buffer stuff */ + /*@{*/ + GLuint _DepthMax; /**< Max depth buffer value */ + GLfloat _DepthMaxF; /**< Float max depth buffer value */ + GLfloat _MRD; /**< minimum resolvable difference in Z values */ + /*@}*/ + + /** One of the GL_FRAMEBUFFER_(IN)COMPLETE_* tokens */ + GLenum _Status; + + /** Integer color values */ + GLboolean _IntegerColor; + + /** Array of all renderbuffer attachments, indexed by BUFFER_* tokens. */ + struct gl_renderbuffer_attachment Attachment[BUFFER_COUNT]; + + /* In unextended OpenGL these vars are part of the GL_COLOR_BUFFER + * attribute group and GL_PIXEL attribute group, respectively. + */ + GLenum ColorDrawBuffer[MAX_DRAW_BUFFERS]; + GLenum ColorReadBuffer; + + /** Computed from ColorDraw/ReadBuffer above */ + GLuint _NumColorDrawBuffers; + GLint _ColorDrawBufferIndexes[MAX_DRAW_BUFFERS]; /**< BUFFER_x or -1 */ + GLint _ColorReadBufferIndex; /* -1 = None */ + struct gl_renderbuffer *_ColorDrawBuffers[MAX_DRAW_BUFFERS]; + struct gl_renderbuffer *_ColorReadBuffer; + + /** The Actual depth/stencil buffers to use. May be wrappers around the + * depth/stencil buffers attached above. */ + struct gl_renderbuffer *_DepthBuffer; + struct gl_renderbuffer *_StencilBuffer; + + /** Delete this framebuffer */ + void (*Delete)(struct gl_framebuffer *fb); +}; + + +/** + * Precision info for shader datatypes. See glGetShaderPrecisionFormat(). + */ +struct gl_precision +{ + GLushort RangeMin; /**< min value exponent */ + GLushort RangeMax; /**< max value exponent */ + GLushort Precision; /**< number of mantissa bits */ +}; + + +/** + * Limits for vertex and fragment programs/shaders. + */ +struct gl_program_constants +{ + /* logical limits */ + GLuint MaxInstructions; + GLuint MaxAluInstructions; + GLuint MaxTexInstructions; + GLuint MaxTexIndirections; + GLuint MaxAttribs; + GLuint MaxTemps; + GLuint MaxAddressRegs; + GLuint MaxParameters; + GLuint MaxLocalParams; + GLuint MaxEnvParams; + /* native/hardware limits */ + GLuint MaxNativeInstructions; + GLuint MaxNativeAluInstructions; + GLuint MaxNativeTexInstructions; + GLuint MaxNativeTexIndirections; + GLuint MaxNativeAttribs; + GLuint MaxNativeTemps; + GLuint MaxNativeAddressRegs; + GLuint MaxNativeParameters; + /* For shaders */ + GLuint MaxUniformComponents; + /* GL_ARB_geometry_shader4 */ + GLuint MaxGeometryTextureImageUnits; + GLuint MaxGeometryVaryingComponents; + GLuint MaxVertexVaryingComponents; + GLuint MaxGeometryUniformComponents; + GLuint MaxGeometryOutputVertices; + GLuint MaxGeometryTotalOutputComponents; + /* ES 2.0 and GL_ARB_ES2_compatibility */ + struct gl_precision LowFloat, MediumFloat, HighFloat; + struct gl_precision LowInt, MediumInt, HighInt; +}; + + +/** + * Constants which may be overridden by device driver during context creation + * but are never changed after that. + */ +struct gl_constants +{ + GLint MaxTextureMbytes; /**< Max memory per image, in MB */ + GLint MaxTextureLevels; /**< Max mipmap levels. */ + GLint Max3DTextureLevels; /**< Max mipmap levels for 3D textures */ + GLint MaxCubeTextureLevels; /**< Max mipmap levels for cube textures */ + GLint MaxArrayTextureLayers; /**< Max layers in array textures */ + GLint MaxTextureRectSize; /**< Max rectangle texture size, in pixes */ + GLuint MaxTextureCoordUnits; + GLuint MaxTextureImageUnits; + GLuint MaxVertexTextureImageUnits; + GLuint MaxCombinedTextureImageUnits; + GLuint MaxTextureUnits; /**< = MIN(CoordUnits, ImageUnits) */ + GLfloat MaxTextureMaxAnisotropy; /**< GL_EXT_texture_filter_anisotropic */ + GLfloat MaxTextureLodBias; /**< GL_EXT_texture_lod_bias */ + + GLuint MaxArrayLockSize; + + GLint SubPixelBits; + + GLfloat MinPointSize, MaxPointSize; /**< aliased */ + GLfloat MinPointSizeAA, MaxPointSizeAA; /**< antialiased */ + GLfloat PointSizeGranularity; + GLfloat MinLineWidth, MaxLineWidth; /**< aliased */ + GLfloat MinLineWidthAA, MaxLineWidthAA; /**< antialiased */ + GLfloat LineWidthGranularity; + + GLuint MaxColorTableSize; + + GLuint MaxClipPlanes; + GLuint MaxLights; + GLfloat MaxShininess; /**< GL_NV_light_max_exponent */ + GLfloat MaxSpotExponent; /**< GL_NV_light_max_exponent */ + + GLuint MaxViewportWidth, MaxViewportHeight; + + struct gl_program_constants VertexProgram; /**< GL_ARB_vertex_program */ + struct gl_program_constants FragmentProgram; /**< GL_ARB_fragment_program */ + struct gl_program_constants GeometryProgram; /**< GL_ARB_geometry_shader4 */ + GLuint MaxProgramMatrices; + GLuint MaxProgramMatrixStackDepth; + + /** vertex array / buffer object bounds checking */ + GLboolean CheckArrayBounds; + + GLuint MaxDrawBuffers; /**< GL_ARB_draw_buffers */ + + GLuint MaxColorAttachments; /**< GL_EXT_framebuffer_object */ + GLuint MaxRenderbufferSize; /**< GL_EXT_framebuffer_object */ + GLuint MaxSamples; /**< GL_ARB_framebuffer_object */ + + GLuint MaxVarying; /**< Number of float[4] varying parameters */ + + GLuint GLSLVersion; /**< GLSL version supported (ex: 120 = 1.20) */ + + /** Which texture units support GL_ATI_envmap_bumpmap as targets */ + GLbitfield SupportedBumpUnits; + + /** + * Maximum amount of time, measured in nanseconds, that the server can wait. + */ + GLuint64 MaxServerWaitTimeout; + + /** GL_EXT_provoking_vertex */ + GLboolean QuadsFollowProvokingVertexConvention; + + /** OpenGL version 3.0 */ + GLbitfield ContextFlags; /**< Ex: GL_CONTEXT_FLAG_FORWARD_COMPATIBLE_BIT */ + + /** OpenGL version 3.2 */ + GLbitfield ProfileMask; /**< Mask of CONTEXT_x_PROFILE_BIT */ + + /** GL_EXT_transform_feedback */ + GLuint MaxTransformFeedbackSeparateAttribs; + GLuint MaxTransformFeedbackSeparateComponents; + GLuint MaxTransformFeedbackInterleavedComponents; + + /** GL_EXT_gpu_shader4 */ + GLint MinProgramTexelOffset, MaxProgramTexelOffset; + + /* GL_EXT_framebuffer_sRGB */ + GLboolean sRGBCapable; /* can enable sRGB blend/update on FBOs */ +}; + + +/** + * Enable flag for each OpenGL extension. Different device drivers will + * enable different extensions at runtime. + */ +struct gl_extensions +{ + GLboolean dummy; /* don't remove this! */ + GLboolean dummy_true; /* Set true by _mesa_init_extensions(). */ + GLboolean dummy_false; /* Set false by _mesa_init_extensions(). */ + GLboolean ARB_ES2_compatibility; + GLboolean ARB_blend_func_extended; + GLboolean ARB_copy_buffer; + GLboolean ARB_depth_buffer_float; + GLboolean ARB_depth_clamp; + GLboolean ARB_depth_texture; + GLboolean ARB_draw_buffers; + GLboolean ARB_draw_buffers_blend; + GLboolean ARB_draw_elements_base_vertex; + GLboolean ARB_draw_instanced; + GLboolean ARB_fragment_coord_conventions; + GLboolean ARB_fragment_program; + GLboolean ARB_fragment_program_shadow; + GLboolean ARB_fragment_shader; + GLboolean ARB_framebuffer_object; + GLboolean ARB_explicit_attrib_location; + GLboolean ARB_geometry_shader4; + GLboolean ARB_half_float_pixel; + GLboolean ARB_half_float_vertex; + GLboolean ARB_instanced_arrays; + GLboolean ARB_map_buffer_range; + GLboolean ARB_multisample; + GLboolean ARB_multitexture; + GLboolean ARB_occlusion_query; + GLboolean ARB_occlusion_query2; + GLboolean ARB_point_sprite; + GLboolean ARB_sampler_objects; + GLboolean ARB_seamless_cube_map; + GLboolean ARB_shader_objects; + GLboolean ARB_shader_stencil_export; + GLboolean ARB_shading_language_100; + GLboolean ARB_shadow; + GLboolean ARB_shadow_ambient; + GLboolean ARB_sync; + GLboolean ARB_texture_border_clamp; + GLboolean ARB_texture_buffer_object; + GLboolean ARB_texture_compression; + GLboolean ARB_texture_compression_rgtc; + GLboolean ARB_texture_cube_map; + GLboolean ARB_texture_env_combine; + GLboolean ARB_texture_env_crossbar; + GLboolean ARB_texture_env_dot3; + GLboolean ARB_texture_float; + GLboolean ARB_texture_mirrored_repeat; + GLboolean ARB_texture_multisample; + GLboolean ARB_texture_non_power_of_two; + GLboolean ARB_texture_rg; + GLboolean ARB_texture_rgb10_a2ui; + GLboolean ARB_timer_query; + GLboolean ARB_transform_feedback2; + GLboolean ARB_transpose_matrix; + GLboolean ARB_uniform_buffer_object; + GLboolean ARB_vertex_array_object; + GLboolean ARB_vertex_buffer_object; + GLboolean ARB_vertex_program; + GLboolean ARB_vertex_shader; + GLboolean ARB_vertex_type_2_10_10_10_rev; + GLboolean ARB_window_pos; + GLboolean EXT_abgr; + GLboolean EXT_bgra; + GLboolean EXT_blend_color; + GLboolean EXT_blend_equation_separate; + GLboolean EXT_blend_func_separate; + GLboolean EXT_blend_logic_op; + GLboolean EXT_blend_minmax; + GLboolean EXT_blend_subtract; + GLboolean EXT_clip_volume_hint; + GLboolean EXT_compiled_vertex_array; + GLboolean EXT_copy_texture; + GLboolean EXT_depth_bounds_test; + GLboolean EXT_draw_buffers2; + GLboolean EXT_draw_range_elements; + GLboolean EXT_fog_coord; + GLboolean EXT_framebuffer_blit; + GLboolean EXT_framebuffer_multisample; + GLboolean EXT_framebuffer_object; + GLboolean EXT_framebuffer_sRGB; + GLboolean EXT_gpu_program_parameters; + GLboolean EXT_gpu_shader4; + GLboolean EXT_multi_draw_arrays; + GLboolean EXT_paletted_texture; + GLboolean EXT_packed_depth_stencil; + GLboolean EXT_packed_float; + GLboolean EXT_packed_pixels; + GLboolean EXT_pixel_buffer_object; + GLboolean EXT_point_parameters; + GLboolean EXT_polygon_offset; + GLboolean EXT_provoking_vertex; + GLboolean EXT_rescale_normal; + GLboolean EXT_shadow_funcs; + GLboolean EXT_secondary_color; + GLboolean EXT_separate_shader_objects; + GLboolean EXT_separate_specular_color; + GLboolean EXT_shared_texture_palette; + GLboolean EXT_stencil_wrap; + GLboolean EXT_stencil_two_side; + GLboolean EXT_subtexture; + GLboolean EXT_texture; + GLboolean EXT_texture_object; + GLboolean EXT_texture3D; + GLboolean EXT_texture_array; + GLboolean EXT_texture_compression_s3tc; + GLboolean EXT_texture_env_add; + GLboolean EXT_texture_env_combine; + GLboolean EXT_texture_env_dot3; + GLboolean EXT_texture_filter_anisotropic; + GLboolean EXT_texture_integer; + GLboolean EXT_texture_lod_bias; + GLboolean EXT_texture_mirror_clamp; + GLboolean EXT_texture_shared_exponent; + GLboolean EXT_texture_sRGB; + GLboolean EXT_texture_sRGB_decode; + GLboolean EXT_texture_swizzle; + GLboolean EXT_transform_feedback; + GLboolean EXT_timer_query; + GLboolean EXT_vertex_array; + GLboolean EXT_vertex_array_bgra; + GLboolean EXT_vertex_array_set; + GLboolean OES_standard_derivatives; + /* vendor extensions */ + GLboolean AMD_conservative_depth; + GLboolean APPLE_client_storage; + GLboolean APPLE_packed_pixels; + GLboolean APPLE_vertex_array_object; + GLboolean APPLE_object_purgeable; + GLboolean ATI_envmap_bumpmap; + GLboolean ATI_texture_mirror_once; + GLboolean ATI_texture_env_combine3; + GLboolean ATI_fragment_shader; + GLboolean ATI_separate_stencil; + GLboolean IBM_rasterpos_clip; + GLboolean IBM_multimode_draw_arrays; + GLboolean MESA_pack_invert; + GLboolean MESA_resize_buffers; + GLboolean MESA_ycbcr_texture; + GLboolean MESA_texture_array; + GLboolean MESA_texture_signed_rgba; + GLboolean NV_blend_square; + GLboolean NV_conditional_render; + GLboolean NV_fragment_program; + GLboolean NV_fragment_program_option; + GLboolean NV_light_max_exponent; + GLboolean NV_point_sprite; + GLboolean NV_primitive_restart; + GLboolean NV_texgen_reflection; + GLboolean NV_texture_env_combine4; + GLboolean NV_texture_rectangle; + GLboolean NV_vertex_program; + GLboolean NV_vertex_program1_1; + GLboolean OES_read_format; + GLboolean SGI_texture_color_table; + GLboolean SGIS_generate_mipmap; + GLboolean SGIS_texture_edge_clamp; + GLboolean SGIS_texture_lod; + GLboolean TDFX_texture_compression_FXT1; + GLboolean S3_s3tc; + GLboolean OES_EGL_image; + GLboolean OES_draw_texture; + GLboolean EXT_texture_format_BGRA8888; + GLboolean extension_sentinel; + /** The extension string */ + const GLubyte *String; + /** Number of supported extensions */ + GLuint Count; +}; + + +/** + * A stack of matrices (projection, modelview, color, texture, etc). + */ +struct gl_matrix_stack +{ + GLmatrix *Top; /**< points into Stack */ + GLmatrix *Stack; /**< array [MaxDepth] of GLmatrix */ + GLuint Depth; /**< 0 <= Depth < MaxDepth */ + GLuint MaxDepth; /**< size of Stack[] array */ + GLuint DirtyFlag; /**< _NEW_MODELVIEW or _NEW_PROJECTION, for example */ +}; + + +/** + * \name Bits for image transfer operations + * \sa __struct gl_contextRec::ImageTransferState. + */ +/*@{*/ +#define IMAGE_SCALE_BIAS_BIT 0x1 +#define IMAGE_SHIFT_OFFSET_BIT 0x2 +#define IMAGE_MAP_COLOR_BIT 0x4 +#define IMAGE_CLAMP_BIT 0x800 + + +/** Pixel Transfer ops */ +#define IMAGE_BITS (IMAGE_SCALE_BIAS_BIT | \ + IMAGE_SHIFT_OFFSET_BIT | \ + IMAGE_MAP_COLOR_BIT) + +/** + * \name Bits to indicate what state has changed. + */ +/*@{*/ +#define _NEW_MODELVIEW (1 << 0) /**< gl_context::ModelView */ +#define _NEW_PROJECTION (1 << 1) /**< gl_context::Projection */ +#define _NEW_TEXTURE_MATRIX (1 << 2) /**< gl_context::TextureMatrix */ +#define _NEW_COLOR (1 << 3) /**< gl_context::Color */ +#define _NEW_DEPTH (1 << 4) /**< gl_context::Depth */ +#define _NEW_EVAL (1 << 5) /**< gl_context::Eval, EvalMap */ +#define _NEW_FOG (1 << 6) /**< gl_context::Fog */ +#define _NEW_HINT (1 << 7) /**< gl_context::Hint */ +#define _NEW_LIGHT (1 << 8) /**< gl_context::Light */ +#define _NEW_LINE (1 << 9) /**< gl_context::Line */ +#define _NEW_PIXEL (1 << 10) /**< gl_context::Pixel */ +#define _NEW_POINT (1 << 11) /**< gl_context::Point */ +#define _NEW_POLYGON (1 << 12) /**< gl_context::Polygon */ +#define _NEW_POLYGONSTIPPLE (1 << 13) /**< gl_context::PolygonStipple */ +#define _NEW_SCISSOR (1 << 14) /**< gl_context::Scissor */ +#define _NEW_STENCIL (1 << 15) /**< gl_context::Stencil */ +#define _NEW_TEXTURE (1 << 16) /**< gl_context::Texture */ +#define _NEW_TRANSFORM (1 << 17) /**< gl_context::Transform */ +#define _NEW_VIEWPORT (1 << 18) /**< gl_context::Viewport */ +#define _NEW_PACKUNPACK (1 << 19) /**< gl_context::Pack, Unpack */ +#define _NEW_ARRAY (1 << 20) /**< gl_context::Array */ +#define _NEW_RENDERMODE (1 << 21) /**< gl_context::RenderMode, etc */ +#define _NEW_BUFFERS (1 << 22) /**< gl_context::Visual, DrawBuffer, */ +#define _NEW_CURRENT_ATTRIB (1 << 23) /**< gl_context::Current */ +#define _NEW_MULTISAMPLE (1 << 24) /**< gl_context::Multisample */ +#define _NEW_TRACK_MATRIX (1 << 25) /**< gl_context::VertexProgram */ +#define _NEW_PROGRAM (1 << 26) /**< New program/shader state */ +#define _NEW_PROGRAM_CONSTANTS (1 << 27) +#define _NEW_BUFFER_OBJECT (1 << 28) +#define _NEW_ALL ~0 +/*@}*/ + + +/** + * \name Bits to track array state changes + * + * Also used to summarize array enabled. + */ +/*@{*/ +#define _NEW_ARRAY_VERTEX VERT_BIT_POS +#define _NEW_ARRAY_WEIGHT VERT_BIT_WEIGHT +#define _NEW_ARRAY_NORMAL VERT_BIT_NORMAL +#define _NEW_ARRAY_COLOR0 VERT_BIT_COLOR0 +#define _NEW_ARRAY_COLOR1 VERT_BIT_COLOR1 +#define _NEW_ARRAY_FOGCOORD VERT_BIT_FOG +#define _NEW_ARRAY_INDEX VERT_BIT_COLOR_INDEX +#define _NEW_ARRAY_EDGEFLAG VERT_BIT_EDGEFLAG +#define _NEW_ARRAY_POINT_SIZE VERT_BIT_COLOR_INDEX /* aliased */ +#define _NEW_ARRAY_TEXCOORD_0 VERT_BIT_TEX0 +#define _NEW_ARRAY_TEXCOORD_1 VERT_BIT_TEX1 +#define _NEW_ARRAY_TEXCOORD_2 VERT_BIT_TEX2 +#define _NEW_ARRAY_TEXCOORD_3 VERT_BIT_TEX3 +#define _NEW_ARRAY_TEXCOORD_4 VERT_BIT_TEX4 +#define _NEW_ARRAY_TEXCOORD_5 VERT_BIT_TEX5 +#define _NEW_ARRAY_TEXCOORD_6 VERT_BIT_TEX6 +#define _NEW_ARRAY_TEXCOORD_7 VERT_BIT_TEX7 +#define _NEW_ARRAY_ATTRIB_0 VERT_BIT_GENERIC0 /* start at bit 16 */ +#define _NEW_ARRAY_ALL 0xffffffff + + +#define _NEW_ARRAY_TEXCOORD(i) (_NEW_ARRAY_TEXCOORD_0 << (i)) +#define _NEW_ARRAY_ATTRIB(i) (_NEW_ARRAY_ATTRIB_0 << (i)) +/*@}*/ + + + +/** + * \name A bunch of flags that we think might be useful to drivers. + * + * Set in the __struct gl_contextRec::_TriangleCaps bitfield. + */ +/*@{*/ +#define DD_FLATSHADE 0x1 +#define DD_SEPARATE_SPECULAR 0x2 +#define DD_TRI_CULL_FRONT_BACK 0x4 /* special case on some hw */ +#define DD_TRI_LIGHT_TWOSIDE 0x8 +#define DD_TRI_UNFILLED 0x10 +#define DD_TRI_SMOOTH 0x20 +#define DD_TRI_STIPPLE 0x40 +#define DD_TRI_OFFSET 0x80 +#define DD_LINE_SMOOTH 0x100 +#define DD_LINE_STIPPLE 0x200 +#define DD_POINT_SMOOTH 0x400 +#define DD_POINT_ATTEN 0x800 +#define DD_TRI_TWOSTENCIL 0x1000 +/*@}*/ + + +/** + * \name Define the state changes under which each of these bits might change + */ +/*@{*/ +#define _DD_NEW_FLATSHADE _NEW_LIGHT +#define _DD_NEW_SEPARATE_SPECULAR (_NEW_LIGHT | _NEW_FOG | _NEW_PROGRAM) +#define _DD_NEW_TRI_CULL_FRONT_BACK _NEW_POLYGON +#define _DD_NEW_TRI_LIGHT_TWOSIDE _NEW_LIGHT +#define _DD_NEW_TRI_UNFILLED _NEW_POLYGON +#define _DD_NEW_TRI_SMOOTH _NEW_POLYGON +#define _DD_NEW_TRI_STIPPLE _NEW_POLYGON +#define _DD_NEW_TRI_OFFSET _NEW_POLYGON +#define _DD_NEW_LINE_SMOOTH _NEW_LINE +#define _DD_NEW_LINE_STIPPLE _NEW_LINE +#define _DD_NEW_LINE_WIDTH _NEW_LINE +#define _DD_NEW_POINT_SMOOTH _NEW_POINT +#define _DD_NEW_POINT_SIZE _NEW_POINT +#define _DD_NEW_POINT_ATTEN _NEW_POINT +/*@}*/ + + +/** + * Composite state flags + */ +/*@{*/ +#define _MESA_NEW_NEED_EYE_COORDS (_NEW_LIGHT | \ + _NEW_TEXTURE | \ + _NEW_POINT | \ + _NEW_PROGRAM | \ + _NEW_MODELVIEW) + +#define _MESA_NEW_NEED_NORMALS (_NEW_LIGHT | \ + _NEW_TEXTURE) + +#define _MESA_NEW_TRANSFER_STATE (_NEW_PIXEL) +/*@}*/ + + + + +/* This has to be included here. */ +#include "dd.h" + + +/** + * Display list flags. + * Strictly this is a tnl-private concept, but it doesn't seem + * worthwhile adding a tnl private structure just to hold this one bit + * of information: + */ +#define DLIST_DANGLING_REFS 0x1 + + +/** Opaque declaration of display list payload data type */ +union gl_dlist_node; + + +/** + * Provide a location where information about a display list can be + * collected. Could be extended with driverPrivate structures, + * etc. in the future. + */ +struct gl_display_list +{ + GLuint Name; + GLbitfield Flags; /**< DLIST_x flags */ + /** The dlist commands are in a linked list of nodes */ + union gl_dlist_node *Head; +}; + + +/** + * State used during display list compilation and execution. + */ +struct gl_dlist_state +{ + GLuint CallDepth; /**< Current recursion calling depth */ + + struct gl_display_list *CurrentList; /**< List currently being compiled */ + union gl_dlist_node *CurrentBlock; /**< Pointer to current block of nodes */ + GLuint CurrentPos; /**< Index into current block of nodes */ + + GLvertexformat ListVtxfmt; + + GLubyte ActiveAttribSize[VERT_ATTRIB_MAX]; + GLfloat CurrentAttrib[VERT_ATTRIB_MAX][4]; + + GLubyte ActiveMaterialSize[MAT_ATTRIB_MAX]; + GLfloat CurrentMaterial[MAT_ATTRIB_MAX][4]; + + GLubyte ActiveIndex; + GLfloat CurrentIndex; + + GLubyte ActiveEdgeFlag; + GLboolean CurrentEdgeFlag; + + struct { + /* State known to have been set by the currently-compiling display + * list. Used to eliminate some redundant state changes. + */ + GLenum ShadeModel; + } Current; +}; + + +/** + * Enum for the OpenGL APIs we know about and may support. + */ +typedef enum +{ + API_OPENGL, + API_OPENGLES, + API_OPENGLES2 +} gl_api; + + +/** + * Mesa rendering context. + * + * This is the central context data structure for Mesa. Almost all + * OpenGL state is contained in this structure. + * Think of this as a base class from which device drivers will derive + * sub classes. + * + * The struct gl_context typedef names this structure. + */ +struct gl_context +{ + /** State possibly shared with other contexts in the address space */ + struct gl_shared_state *Shared; + + /** \name API function pointer tables */ + /*@{*/ + gl_api API; + struct _glapi_table *Save; /**< Display list save functions */ + struct _glapi_table *Exec; /**< Execute functions */ + struct _glapi_table *CurrentDispatch; /**< == Save or Exec !! */ + /*@}*/ + + struct gl_config Visual; + struct gl_framebuffer *DrawBuffer; /**< buffer for writing */ + struct gl_framebuffer *ReadBuffer; /**< buffer for reading */ + struct gl_framebuffer *WinSysDrawBuffer; /**< set with MakeCurrent */ + struct gl_framebuffer *WinSysReadBuffer; /**< set with MakeCurrent */ + + /** + * Device driver function pointer table + */ + struct dd_function_table Driver; + + void *DriverCtx; /**< Points to device driver context/state */ + + /** Core/Driver constants */ + struct gl_constants Const; + + /** \name The various 4x4 matrix stacks */ + /*@{*/ + struct gl_matrix_stack ModelviewMatrixStack; + struct gl_matrix_stack ProjectionMatrixStack; + struct gl_matrix_stack TextureMatrixStack[MAX_TEXTURE_UNITS]; + struct gl_matrix_stack ProgramMatrixStack[MAX_PROGRAM_MATRICES]; + struct gl_matrix_stack *CurrentStack; /**< Points to one of the above stacks */ + /*@}*/ + + /** Combined modelview and projection matrix */ + GLmatrix _ModelProjectMatrix; + + /** \name Display lists */ + struct gl_dlist_state ListState; + + GLboolean ExecuteFlag; /**< Execute GL commands? */ + GLboolean CompileFlag; /**< Compile GL commands into display list? */ + + /** Extension information */ + struct gl_extensions Extensions; + + /** Version info */ + GLuint VersionMajor, VersionMinor; + char *VersionString; + + /** \name State attribute stack (for glPush/PopAttrib) */ + /*@{*/ + GLuint AttribStackDepth; + struct gl_attrib_node *AttribStack[MAX_ATTRIB_STACK_DEPTH]; + /*@}*/ + + /** \name Renderer attribute groups + * + * We define a struct for each attribute group to make pushing and popping + * attributes easy. Also it's a good organization. + */ + /*@{*/ + struct gl_accum_attrib Accum; /**< Accum buffer attributes */ + struct gl_colorbuffer_attrib Color; /**< Color buffer attributes */ + struct gl_current_attrib Current; /**< Current attributes */ + struct gl_depthbuffer_attrib Depth; /**< Depth buffer attributes */ + struct gl_eval_attrib Eval; /**< Eval attributes */ + struct gl_fog_attrib Fog; /**< Fog attributes */ + struct gl_hint_attrib Hint; /**< Hint attributes */ + struct gl_light_attrib Light; /**< Light attributes */ + struct gl_line_attrib Line; /**< Line attributes */ + struct gl_list_attrib List; /**< List attributes */ + struct gl_multisample_attrib Multisample; + struct gl_pixel_attrib Pixel; /**< Pixel attributes */ + struct gl_point_attrib Point; /**< Point attributes */ + struct gl_polygon_attrib Polygon; /**< Polygon attributes */ + GLuint PolygonStipple[32]; /**< Polygon stipple */ + struct gl_scissor_attrib Scissor; /**< Scissor attributes */ + struct gl_stencil_attrib Stencil; /**< Stencil buffer attributes */ + struct gl_texture_attrib Texture; /**< Texture attributes */ + struct gl_transform_attrib Transform; /**< Transformation attributes */ + struct gl_viewport_attrib Viewport; /**< Viewport attributes */ + /*@}*/ + + /** \name Client attribute stack */ + /*@{*/ + GLuint ClientAttribStackDepth; + struct gl_attrib_node *ClientAttribStack[MAX_CLIENT_ATTRIB_STACK_DEPTH]; + /*@}*/ + + /** \name Client attribute groups */ + /*@{*/ + struct gl_array_attrib Array; /**< Vertex arrays */ + struct gl_pixelstore_attrib Pack; /**< Pixel packing */ + struct gl_pixelstore_attrib Unpack; /**< Pixel unpacking */ + struct gl_pixelstore_attrib DefaultPacking; /**< Default params */ + /*@}*/ + + /** \name Other assorted state (not pushed/popped on attribute stack) */ + /*@{*/ + struct gl_pixelmaps PixelMaps; + + struct gl_evaluators EvalMap; /**< All evaluators */ + struct gl_feedback Feedback; /**< Feedback */ + struct gl_selection Select; /**< Selection */ + + struct gl_program_state Program; /**< general program state */ + struct gl_vertex_program_state VertexProgram; + struct gl_fragment_program_state FragmentProgram; + struct gl_geometry_program_state GeometryProgram; + struct gl_ati_fragment_shader_state ATIFragmentShader; + + struct gl_shader_state Shader; /**< GLSL shader object state */ + struct gl_shader_compiler_options ShaderCompilerOptions[MESA_SHADER_TYPES]; + + struct gl_query_state Query; /**< occlusion, timer queries */ + + struct gl_transform_feedback TransformFeedback; + + struct gl_buffer_object *CopyReadBuffer; /**< GL_ARB_copy_buffer */ + struct gl_buffer_object *CopyWriteBuffer; /**< GL_ARB_copy_buffer */ + /*@}*/ + + struct gl_meta_state *Meta; /**< for "meta" operations */ + + /* GL_EXT_framebuffer_object */ + struct gl_renderbuffer *CurrentRenderbuffer; + + GLenum ErrorValue; /**< Last error code */ + + /** + * Recognize and silence repeated error debug messages in buggy apps. + */ + const char *ErrorDebugFmtString; + GLuint ErrorDebugCount; + + GLenum RenderMode; /**< either GL_RENDER, GL_SELECT, GL_FEEDBACK */ + GLbitfield NewState; /**< bitwise-or of _NEW_* flags */ + + GLboolean ViewportInitialized; /**< has viewport size been initialized? */ + + GLbitfield varying_vp_inputs; /**< mask of VERT_BIT_* flags */ + + /** \name Derived state */ + /*@{*/ + /** Bitwise-or of DD_* flags. Note that this bitfield may be used before + * state validation so they need to always be current. + */ + GLbitfield _TriangleCaps; + GLbitfield _ImageTransferState;/**< bitwise-or of IMAGE_*_BIT flags */ + GLfloat _EyeZDir[3]; + GLfloat _ModelViewInvScale; + GLboolean _NeedEyeCoords; + GLboolean _ForceEyeCoords; + + GLuint TextureStateTimestamp; /**< detect changes to shared state */ + + struct gl_shine_tab *_ShineTable[2]; /**< Active shine tables */ + struct gl_shine_tab *_ShineTabList; /**< MRU list of inactive shine tables */ + /**@}*/ + + struct gl_list_extensions *ListExt; /**< driver dlist extensions */ + + /** \name For debugging/development only */ + /*@{*/ + GLboolean FirstTimeCurrent; + /*@}*/ + + /** Dither disable via MESA_NO_DITHER env var */ + GLboolean NoDither; + + /** software compression/decompression supported or not */ + GLboolean Mesa_DXTn; + + GLboolean TextureFormatSupported[MESA_FORMAT_COUNT]; + + /** + * Use dp4 (rather than mul/mad) instructions for position + * transformation? + */ + GLboolean mvp_with_dp4; + + /** + * \name Hooks for module contexts. + * + * These will eventually live in the driver or elsewhere. + */ + /*@{*/ + void *swrast_context; + void *swsetup_context; + void *swtnl_context; + void *swtnl_im; + struct st_context *st; + void *aelt_context; + /*@}*/ +}; + + +#ifdef DEBUG +extern int MESA_VERBOSE; +extern int MESA_DEBUG_FLAGS; +# define MESA_FUNCTION __FUNCTION__ +#else +# define MESA_VERBOSE 0 +# define MESA_DEBUG_FLAGS 0 +# define MESA_FUNCTION "a function" +# ifndef NDEBUG +# define NDEBUG +# endif +#endif + + +/** The MESA_VERBOSE var is a bitmask of these flags */ +enum _verbose +{ + VERBOSE_VARRAY = 0x0001, + VERBOSE_TEXTURE = 0x0002, + VERBOSE_MATERIAL = 0x0004, + VERBOSE_PIPELINE = 0x0008, + VERBOSE_DRIVER = 0x0010, + VERBOSE_STATE = 0x0020, + VERBOSE_API = 0x0040, + VERBOSE_DISPLAY_LIST = 0x0100, + VERBOSE_LIGHTING = 0x0200, + VERBOSE_PRIMS = 0x0400, + VERBOSE_VERTS = 0x0800, + VERBOSE_DISASSEM = 0x1000, + VERBOSE_DRAW = 0x2000, + VERBOSE_SWAPBUFFERS = 0x4000 +}; + + +/** The MESA_DEBUG_FLAGS var is a bitmask of these flags */ +enum _debug +{ + DEBUG_ALWAYS_FLUSH = 0x1 +}; + + + +#endif /* MTYPES_H */ diff --git a/mesalib/src/mesa/program/ir_to_mesa.cpp b/mesalib/src/mesa/program/ir_to_mesa.cpp index d0ec23fc8..aa5f007e9 100644 --- a/mesalib/src/mesa/program/ir_to_mesa.cpp +++ b/mesalib/src/mesa/program/ir_to_mesa.cpp @@ -1,3267 +1,3270 @@ -/* - * Copyright (C) 2005-2007 Brian Paul All Rights Reserved. - * Copyright (C) 2008 VMware, Inc. All Rights Reserved. - * Copyright © 2010 Intel Corporation - * - * 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 (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 NONINFRINGEMENT. IN NO EVENT SHALL - * THE AUTHORS OR COPYRIGHT HOLDERS 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. - */ - -/** - * \file ir_to_mesa.cpp - * - * Translate GLSL IR to Mesa's gl_program representation. - */ - -#include -#include "main/compiler.h" -#include "ir.h" -#include "ir_visitor.h" -#include "ir_print_visitor.h" -#include "ir_expression_flattening.h" -#include "glsl_types.h" -#include "glsl_parser_extras.h" -#include "../glsl/program.h" -#include "ir_optimization.h" -#include "ast.h" - -extern "C" { -#include "main/mtypes.h" -#include "main/shaderapi.h" -#include "main/shaderobj.h" -#include "main/uniforms.h" -#include "program/hash_table.h" -#include "program/prog_instruction.h" -#include "program/prog_optimize.h" -#include "program/prog_print.h" -#include "program/program.h" -#include "program/prog_uniform.h" -#include "program/prog_parameter.h" -#include "program/sampler.h" -} - -static int swizzle_for_size(int size); - -/** - * This struct is a corresponding struct to Mesa prog_src_register, with - * wider fields. - */ -typedef struct ir_to_mesa_src_reg { - ir_to_mesa_src_reg(int file, int index, const glsl_type *type) - { - this->file = (gl_register_file) file; - this->index = index; - if (type && (type->is_scalar() || type->is_vector() || type->is_matrix())) - this->swizzle = swizzle_for_size(type->vector_elements); - else - this->swizzle = SWIZZLE_XYZW; - this->negate = 0; - this->reladdr = NULL; - } - - ir_to_mesa_src_reg() - { - this->file = PROGRAM_UNDEFINED; - this->index = 0; - this->swizzle = 0; - this->negate = 0; - this->reladdr = NULL; - } - - gl_register_file file; /**< PROGRAM_* from Mesa */ - int index; /**< temporary index, VERT_ATTRIB_*, FRAG_ATTRIB_*, etc. */ - GLuint swizzle; /**< SWIZZLE_XYZWONEZERO swizzles from Mesa. */ - int negate; /**< NEGATE_XYZW mask from mesa */ - /** Register index should be offset by the integer in this reg. */ - ir_to_mesa_src_reg *reladdr; -} ir_to_mesa_src_reg; - -typedef struct ir_to_mesa_dst_reg { - int file; /**< PROGRAM_* from Mesa */ - int index; /**< temporary index, VERT_ATTRIB_*, FRAG_ATTRIB_*, etc. */ - int writemask; /**< Bitfield of WRITEMASK_[XYZW] */ - GLuint cond_mask:4; - /** Register index should be offset by the integer in this reg. */ - ir_to_mesa_src_reg *reladdr; -} ir_to_mesa_dst_reg; - -extern ir_to_mesa_src_reg ir_to_mesa_undef; - -class ir_to_mesa_instruction : public exec_node { -public: - /* Callers of this ralloc-based new need not call delete. It's - * easier to just ralloc_free 'ctx' (or any of its ancestors). */ - static void* operator new(size_t size, void *ctx) - { - void *node; - - node = rzalloc_size(ctx, size); - assert(node != NULL); - - return node; - } - - enum prog_opcode op; - ir_to_mesa_dst_reg dst_reg; - ir_to_mesa_src_reg src_reg[3]; - /** Pointer to the ir source this tree came from for debugging */ - ir_instruction *ir; - GLboolean cond_update; - bool saturate; - int sampler; /**< sampler index */ - int tex_target; /**< One of TEXTURE_*_INDEX */ - GLboolean tex_shadow; - - class function_entry *function; /* Set on OPCODE_CAL or OPCODE_BGNSUB */ -}; - -class variable_storage : public exec_node { -public: - variable_storage(ir_variable *var, gl_register_file file, int index) - : file(file), index(index), var(var) - { - /* empty */ - } - - gl_register_file file; - int index; - ir_variable *var; /* variable that maps to this, if any */ -}; - -class function_entry : public exec_node { -public: - ir_function_signature *sig; - - /** - * identifier of this function signature used by the program. - * - * At the point that Mesa instructions for function calls are - * generated, we don't know the address of the first instruction of - * the function body. So we make the BranchTarget that is called a - * small integer and rewrite them during set_branchtargets(). - */ - int sig_id; - - /** - * Pointer to first instruction of the function body. - * - * Set during function body emits after main() is processed. - */ - ir_to_mesa_instruction *bgn_inst; - - /** - * Index of the first instruction of the function body in actual - * Mesa IR. - * - * Set after convertion from ir_to_mesa_instruction to prog_instruction. - */ - int inst; - - /** Storage for the return value. */ - ir_to_mesa_src_reg return_reg; -}; - -class ir_to_mesa_visitor : public ir_visitor { -public: - ir_to_mesa_visitor(); - ~ir_to_mesa_visitor(); - - function_entry *current_function; - - struct gl_context *ctx; - struct gl_program *prog; - struct gl_shader_program *shader_program; - struct gl_shader_compiler_options *options; - - int next_temp; - - variable_storage *find_variable_storage(ir_variable *var); - - function_entry *get_function_signature(ir_function_signature *sig); - - ir_to_mesa_src_reg get_temp(const glsl_type *type); - void reladdr_to_temp(ir_instruction *ir, - ir_to_mesa_src_reg *reg, int *num_reladdr); - - struct ir_to_mesa_src_reg src_reg_for_float(float val); - - /** - * \name Visit methods - * - * As typical for the visitor pattern, there must be one \c visit method for - * each concrete subclass of \c ir_instruction. Virtual base classes within - * the hierarchy should not have \c visit methods. - */ - /*@{*/ - virtual void visit(ir_variable *); - virtual void visit(ir_loop *); - virtual void visit(ir_loop_jump *); - virtual void visit(ir_function_signature *); - virtual void visit(ir_function *); - virtual void visit(ir_expression *); - virtual void visit(ir_swizzle *); - virtual void visit(ir_dereference_variable *); - virtual void visit(ir_dereference_array *); - virtual void visit(ir_dereference_record *); - virtual void visit(ir_assignment *); - virtual void visit(ir_constant *); - virtual void visit(ir_call *); - virtual void visit(ir_return *); - virtual void visit(ir_discard *); - virtual void visit(ir_texture *); - virtual void visit(ir_if *); - /*@}*/ - - struct ir_to_mesa_src_reg result; - - /** List of variable_storage */ - exec_list variables; - - /** List of function_entry */ - exec_list function_signatures; - int next_signature_id; - - /** List of ir_to_mesa_instruction */ - exec_list instructions; - - ir_to_mesa_instruction *ir_to_mesa_emit_op0(ir_instruction *ir, - enum prog_opcode op); - - ir_to_mesa_instruction *ir_to_mesa_emit_op1(ir_instruction *ir, - enum prog_opcode op, - ir_to_mesa_dst_reg dst, - ir_to_mesa_src_reg src0); - - ir_to_mesa_instruction *ir_to_mesa_emit_op2(ir_instruction *ir, - enum prog_opcode op, - ir_to_mesa_dst_reg dst, - ir_to_mesa_src_reg src0, - ir_to_mesa_src_reg src1); - - ir_to_mesa_instruction *ir_to_mesa_emit_op3(ir_instruction *ir, - enum prog_opcode op, - ir_to_mesa_dst_reg dst, - ir_to_mesa_src_reg src0, - ir_to_mesa_src_reg src1, - ir_to_mesa_src_reg src2); - - /** - * Emit the correct dot-product instruction for the type of arguments - * - * \sa ir_to_mesa_emit_op2 - */ - void ir_to_mesa_emit_dp(ir_instruction *ir, - ir_to_mesa_dst_reg dst, - ir_to_mesa_src_reg src0, - ir_to_mesa_src_reg src1, - unsigned elements); - - void ir_to_mesa_emit_scalar_op1(ir_instruction *ir, - enum prog_opcode op, - ir_to_mesa_dst_reg dst, - ir_to_mesa_src_reg src0); - - void ir_to_mesa_emit_scalar_op2(ir_instruction *ir, - enum prog_opcode op, - ir_to_mesa_dst_reg dst, - ir_to_mesa_src_reg src0, - ir_to_mesa_src_reg src1); - - void emit_scs(ir_instruction *ir, enum prog_opcode op, - ir_to_mesa_dst_reg dst, - const ir_to_mesa_src_reg &src); - - GLboolean try_emit_mad(ir_expression *ir, - int mul_operand); - GLboolean try_emit_sat(ir_expression *ir); - - void emit_swz(ir_expression *ir); - - bool process_move_condition(ir_rvalue *ir); - - void copy_propagate(void); - - void *mem_ctx; -}; - -ir_to_mesa_src_reg ir_to_mesa_undef = ir_to_mesa_src_reg(PROGRAM_UNDEFINED, 0, NULL); - -ir_to_mesa_dst_reg ir_to_mesa_undef_dst = { - PROGRAM_UNDEFINED, 0, SWIZZLE_NOOP, COND_TR, NULL, -}; - -ir_to_mesa_dst_reg ir_to_mesa_address_reg = { - PROGRAM_ADDRESS, 0, WRITEMASK_X, COND_TR, NULL -}; - -static void -fail_link(struct gl_shader_program *prog, const char *fmt, ...) PRINTFLIKE(2, 3); - -static void -fail_link(struct gl_shader_program *prog, const char *fmt, ...) -{ - va_list args; - va_start(args, fmt); - ralloc_vasprintf_append(&prog->InfoLog, fmt, args); - va_end(args); - - prog->LinkStatus = GL_FALSE; -} - -static int -swizzle_for_size(int size) -{ - int size_swizzles[4] = { - MAKE_SWIZZLE4(SWIZZLE_X, SWIZZLE_X, SWIZZLE_X, SWIZZLE_X), - MAKE_SWIZZLE4(SWIZZLE_X, SWIZZLE_Y, SWIZZLE_Y, SWIZZLE_Y), - MAKE_SWIZZLE4(SWIZZLE_X, SWIZZLE_Y, SWIZZLE_Z, SWIZZLE_Z), - MAKE_SWIZZLE4(SWIZZLE_X, SWIZZLE_Y, SWIZZLE_Z, SWIZZLE_W), - }; - - assert((size >= 1) && (size <= 4)); - return size_swizzles[size - 1]; -} - -ir_to_mesa_instruction * -ir_to_mesa_visitor::ir_to_mesa_emit_op3(ir_instruction *ir, - enum prog_opcode op, - ir_to_mesa_dst_reg dst, - ir_to_mesa_src_reg src0, - ir_to_mesa_src_reg src1, - ir_to_mesa_src_reg src2) -{ - ir_to_mesa_instruction *inst = new(mem_ctx) ir_to_mesa_instruction(); - int num_reladdr = 0; - - /* If we have to do relative addressing, we want to load the ARL - * reg directly for one of the regs, and preload the other reladdr - * sources into temps. - */ - num_reladdr += dst.reladdr != NULL; - num_reladdr += src0.reladdr != NULL; - num_reladdr += src1.reladdr != NULL; - num_reladdr += src2.reladdr != NULL; - - reladdr_to_temp(ir, &src2, &num_reladdr); - reladdr_to_temp(ir, &src1, &num_reladdr); - reladdr_to_temp(ir, &src0, &num_reladdr); - - if (dst.reladdr) { - ir_to_mesa_emit_op1(ir, OPCODE_ARL, ir_to_mesa_address_reg, - *dst.reladdr); - - num_reladdr--; - } - assert(num_reladdr == 0); - - inst->op = op; - inst->dst_reg = dst; - inst->src_reg[0] = src0; - inst->src_reg[1] = src1; - inst->src_reg[2] = src2; - inst->ir = ir; - - inst->function = NULL; - - this->instructions.push_tail(inst); - - return inst; -} - - -ir_to_mesa_instruction * -ir_to_mesa_visitor::ir_to_mesa_emit_op2(ir_instruction *ir, - enum prog_opcode op, - ir_to_mesa_dst_reg dst, - ir_to_mesa_src_reg src0, - ir_to_mesa_src_reg src1) -{ - return ir_to_mesa_emit_op3(ir, op, dst, src0, src1, ir_to_mesa_undef); -} - -ir_to_mesa_instruction * -ir_to_mesa_visitor::ir_to_mesa_emit_op1(ir_instruction *ir, - enum prog_opcode op, - ir_to_mesa_dst_reg dst, - ir_to_mesa_src_reg src0) -{ - assert(dst.writemask != 0); - return ir_to_mesa_emit_op3(ir, op, dst, - src0, ir_to_mesa_undef, ir_to_mesa_undef); -} - -ir_to_mesa_instruction * -ir_to_mesa_visitor::ir_to_mesa_emit_op0(ir_instruction *ir, - enum prog_opcode op) -{ - return ir_to_mesa_emit_op3(ir, op, ir_to_mesa_undef_dst, - ir_to_mesa_undef, - ir_to_mesa_undef, - ir_to_mesa_undef); -} - -void -ir_to_mesa_visitor::ir_to_mesa_emit_dp(ir_instruction *ir, - ir_to_mesa_dst_reg dst, - ir_to_mesa_src_reg src0, - ir_to_mesa_src_reg src1, - unsigned elements) -{ - static const gl_inst_opcode dot_opcodes[] = { - OPCODE_DP2, OPCODE_DP3, OPCODE_DP4 - }; - - ir_to_mesa_emit_op3(ir, dot_opcodes[elements - 2], - dst, src0, src1, ir_to_mesa_undef); -} - -inline ir_to_mesa_dst_reg -ir_to_mesa_dst_reg_from_src(ir_to_mesa_src_reg reg) -{ - ir_to_mesa_dst_reg dst_reg; - - dst_reg.file = reg.file; - dst_reg.index = reg.index; - dst_reg.writemask = WRITEMASK_XYZW; - dst_reg.cond_mask = COND_TR; - dst_reg.reladdr = reg.reladdr; - - return dst_reg; -} - -inline ir_to_mesa_src_reg -ir_to_mesa_src_reg_from_dst(ir_to_mesa_dst_reg reg) -{ - return ir_to_mesa_src_reg(reg.file, reg.index, NULL); -} - -/** - * Emits Mesa scalar opcodes to produce unique answers across channels. - * - * Some Mesa opcodes are scalar-only, like ARB_fp/vp. The src X - * channel determines the result across all channels. So to do a vec4 - * of this operation, we want to emit a scalar per source channel used - * to produce dest channels. - */ -void -ir_to_mesa_visitor::ir_to_mesa_emit_scalar_op2(ir_instruction *ir, - enum prog_opcode op, - ir_to_mesa_dst_reg dst, - ir_to_mesa_src_reg orig_src0, - ir_to_mesa_src_reg orig_src1) -{ - int i, j; - int done_mask = ~dst.writemask; - - /* Mesa RCP is a scalar operation splatting results to all channels, - * like ARB_fp/vp. So emit as many RCPs as necessary to cover our - * dst channels. - */ - for (i = 0; i < 4; i++) { - GLuint this_mask = (1 << i); - ir_to_mesa_instruction *inst; - ir_to_mesa_src_reg src0 = orig_src0; - ir_to_mesa_src_reg src1 = orig_src1; - - if (done_mask & this_mask) - continue; - - GLuint src0_swiz = GET_SWZ(src0.swizzle, i); - GLuint src1_swiz = GET_SWZ(src1.swizzle, i); - for (j = i + 1; j < 4; j++) { - /* If there is another enabled component in the destination that is - * derived from the same inputs, generate its value on this pass as - * well. - */ - if (!(done_mask & (1 << j)) && - GET_SWZ(src0.swizzle, j) == src0_swiz && - GET_SWZ(src1.swizzle, j) == src1_swiz) { - this_mask |= (1 << j); - } - } - src0.swizzle = MAKE_SWIZZLE4(src0_swiz, src0_swiz, - src0_swiz, src0_swiz); - src1.swizzle = MAKE_SWIZZLE4(src1_swiz, src1_swiz, - src1_swiz, src1_swiz); - - inst = ir_to_mesa_emit_op2(ir, op, - dst, - src0, - src1); - inst->dst_reg.writemask = this_mask; - done_mask |= this_mask; - } -} - -void -ir_to_mesa_visitor::ir_to_mesa_emit_scalar_op1(ir_instruction *ir, - enum prog_opcode op, - ir_to_mesa_dst_reg dst, - ir_to_mesa_src_reg src0) -{ - ir_to_mesa_src_reg undef = ir_to_mesa_undef; - - undef.swizzle = SWIZZLE_XXXX; - - ir_to_mesa_emit_scalar_op2(ir, op, dst, src0, undef); -} - -/** - * Emit an OPCODE_SCS instruction - * - * The \c SCS opcode functions a bit differently than the other Mesa (or - * ARB_fragment_program) opcodes. Instead of splatting its result across all - * four components of the destination, it writes one value to the \c x - * component and another value to the \c y component. - * - * \param ir IR instruction being processed - * \param op Either \c OPCODE_SIN or \c OPCODE_COS depending on which - * value is desired. - * \param dst Destination register - * \param src Source register - */ -void -ir_to_mesa_visitor::emit_scs(ir_instruction *ir, enum prog_opcode op, - ir_to_mesa_dst_reg dst, - const ir_to_mesa_src_reg &src) -{ - /* Vertex programs cannot use the SCS opcode. - */ - if (this->prog->Target == GL_VERTEX_PROGRAM_ARB) { - ir_to_mesa_emit_scalar_op1(ir, op, dst, src); - return; - } - - const unsigned component = (op == OPCODE_SIN) ? 0 : 1; - const unsigned scs_mask = (1U << component); - int done_mask = ~dst.writemask; - ir_to_mesa_src_reg tmp; - - assert(op == OPCODE_SIN || op == OPCODE_COS); - - /* If there are compnents in the destination that differ from the component - * that will be written by the SCS instrution, we'll need a temporary. - */ - if (scs_mask != unsigned(dst.writemask)) { - tmp = get_temp(glsl_type::vec4_type); - } - - for (unsigned i = 0; i < 4; i++) { - unsigned this_mask = (1U << i); - ir_to_mesa_src_reg src0 = src; - - if ((done_mask & this_mask) != 0) - continue; - - /* The source swizzle specified which component of the source generates - * sine / cosine for the current component in the destination. The SCS - * instruction requires that this value be swizzle to the X component. - * Replace the current swizzle with a swizzle that puts the source in - * the X component. - */ - unsigned src0_swiz = GET_SWZ(src.swizzle, i); - - src0.swizzle = MAKE_SWIZZLE4(src0_swiz, src0_swiz, - src0_swiz, src0_swiz); - for (unsigned j = i + 1; j < 4; j++) { - /* If there is another enabled component in the destination that is - * derived from the same inputs, generate its value on this pass as - * well. - */ - if (!(done_mask & (1 << j)) && - GET_SWZ(src0.swizzle, j) == src0_swiz) { - this_mask |= (1 << j); - } - } - - if (this_mask != scs_mask) { - ir_to_mesa_instruction *inst; - ir_to_mesa_dst_reg tmp_dst = ir_to_mesa_dst_reg_from_src(tmp); - - /* Emit the SCS instruction. - */ - inst = ir_to_mesa_emit_op1(ir, OPCODE_SCS, tmp_dst, src0); - inst->dst_reg.writemask = scs_mask; - - /* Move the result of the SCS instruction to the desired location in - * the destination. - */ - tmp.swizzle = MAKE_SWIZZLE4(component, component, - component, component); - inst = ir_to_mesa_emit_op1(ir, OPCODE_SCS, dst, tmp); - inst->dst_reg.writemask = this_mask; - } else { - /* Emit the SCS instruction to write directly to the destination. - */ - ir_to_mesa_instruction *inst = - ir_to_mesa_emit_op1(ir, OPCODE_SCS, dst, src0); - inst->dst_reg.writemask = scs_mask; - } - - done_mask |= this_mask; - } -} - -struct ir_to_mesa_src_reg -ir_to_mesa_visitor::src_reg_for_float(float val) -{ - ir_to_mesa_src_reg src_reg(PROGRAM_CONSTANT, -1, NULL); - - src_reg.index = _mesa_add_unnamed_constant(this->prog->Parameters, - &val, 1, &src_reg.swizzle); - - return src_reg; -} - -static int -type_size(const struct glsl_type *type) -{ - unsigned int i; - int size; - - switch (type->base_type) { - case GLSL_TYPE_UINT: - case GLSL_TYPE_INT: - case GLSL_TYPE_FLOAT: - case GLSL_TYPE_BOOL: - if (type->is_matrix()) { - return type->matrix_columns; - } else { - /* Regardless of size of vector, it gets a vec4. This is bad - * packing for things like floats, but otherwise arrays become a - * mess. Hopefully a later pass over the code can pack scalars - * down if appropriate. - */ - return 1; - } - case GLSL_TYPE_ARRAY: - assert(type->length > 0); - return type_size(type->fields.array) * type->length; - case GLSL_TYPE_STRUCT: - size = 0; - for (i = 0; i < type->length; i++) { - size += type_size(type->fields.structure[i].type); - } - return size; - case GLSL_TYPE_SAMPLER: - /* Samplers take up one slot in UNIFORMS[], but they're baked in - * at link time. - */ - return 1; - default: - assert(0); - return 0; - } -} - -/** - * In the initial pass of codegen, we assign temporary numbers to - * intermediate results. (not SSA -- variable assignments will reuse - * storage). Actual register allocation for the Mesa VM occurs in a - * pass over the Mesa IR later. - */ -ir_to_mesa_src_reg -ir_to_mesa_visitor::get_temp(const glsl_type *type) -{ - ir_to_mesa_src_reg src_reg; - int swizzle[4]; - int i; - - src_reg.file = PROGRAM_TEMPORARY; - src_reg.index = next_temp; - src_reg.reladdr = NULL; - next_temp += type_size(type); - - if (type->is_array() || type->is_record()) { - src_reg.swizzle = SWIZZLE_NOOP; - } else { - for (i = 0; i < type->vector_elements; i++) - swizzle[i] = i; - for (; i < 4; i++) - swizzle[i] = type->vector_elements - 1; - src_reg.swizzle = MAKE_SWIZZLE4(swizzle[0], swizzle[1], - swizzle[2], swizzle[3]); - } - src_reg.negate = 0; - - return src_reg; -} - -variable_storage * -ir_to_mesa_visitor::find_variable_storage(ir_variable *var) -{ - - variable_storage *entry; - - foreach_iter(exec_list_iterator, iter, this->variables) { - entry = (variable_storage *)iter.get(); - - if (entry->var == var) - return entry; - } - - return NULL; -} - -void -ir_to_mesa_visitor::visit(ir_variable *ir) -{ - if (strcmp(ir->name, "gl_FragCoord") == 0) { - struct gl_fragment_program *fp = (struct gl_fragment_program *)this->prog; - - fp->OriginUpperLeft = ir->origin_upper_left; - fp->PixelCenterInteger = ir->pixel_center_integer; - - } else if (strcmp(ir->name, "gl_FragDepth") == 0) { - struct gl_fragment_program *fp = (struct gl_fragment_program *)this->prog; - switch (ir->depth_layout) { - case ir_depth_layout_none: - fp->FragDepthLayout = FRAG_DEPTH_LAYOUT_NONE; - break; - case ir_depth_layout_any: - fp->FragDepthLayout = FRAG_DEPTH_LAYOUT_ANY; - break; - case ir_depth_layout_greater: - fp->FragDepthLayout = FRAG_DEPTH_LAYOUT_GREATER; - break; - case ir_depth_layout_less: - fp->FragDepthLayout = FRAG_DEPTH_LAYOUT_LESS; - break; - case ir_depth_layout_unchanged: - fp->FragDepthLayout = FRAG_DEPTH_LAYOUT_UNCHANGED; - break; - default: - assert(0); - break; - } - } - - if (ir->mode == ir_var_uniform && strncmp(ir->name, "gl_", 3) == 0) { - unsigned int i; - const struct gl_builtin_uniform_desc *statevar; - - for (i = 0; _mesa_builtin_uniform_desc[i].name; i++) { - if (strcmp(ir->name, _mesa_builtin_uniform_desc[i].name) == 0) - break; - } - - if (!_mesa_builtin_uniform_desc[i].name) { - fail_link(this->shader_program, - "Failed to find builtin uniform `%s'\n", ir->name); - return; - } - - statevar = &_mesa_builtin_uniform_desc[i]; - - int array_count; - if (ir->type->is_array()) { - array_count = ir->type->length; - } else { - array_count = 1; - } - - /* Check if this statevar's setup in the STATE file exactly - * matches how we'll want to reference it as a - * struct/array/whatever. If not, then we need to move it into - * temporary storage and hope that it'll get copy-propagated - * out. - */ - for (i = 0; i < statevar->num_elements; i++) { - if (statevar->elements[i].swizzle != SWIZZLE_XYZW) { - break; - } - } - - struct variable_storage *storage; - ir_to_mesa_dst_reg dst; - if (i == statevar->num_elements) { - /* We'll set the index later. */ - storage = new(mem_ctx) variable_storage(ir, PROGRAM_STATE_VAR, -1); - this->variables.push_tail(storage); - - dst = ir_to_mesa_undef_dst; - } else { - storage = new(mem_ctx) variable_storage(ir, PROGRAM_TEMPORARY, - this->next_temp); - this->variables.push_tail(storage); - this->next_temp += type_size(ir->type); - - dst = ir_to_mesa_dst_reg_from_src(ir_to_mesa_src_reg(PROGRAM_TEMPORARY, - storage->index, - NULL)); - } - - - for (int a = 0; a < array_count; a++) { - for (unsigned int i = 0; i < statevar->num_elements; i++) { - struct gl_builtin_uniform_element *element = &statevar->elements[i]; - int tokens[STATE_LENGTH]; - - memcpy(tokens, element->tokens, sizeof(element->tokens)); - if (ir->type->is_array()) { - tokens[1] = a; - } - - int index = _mesa_add_state_reference(this->prog->Parameters, - (gl_state_index *)tokens); - - if (storage->file == PROGRAM_STATE_VAR) { - if (storage->index == -1) { - storage->index = index; - } else { - assert(index == - (int)(storage->index + a * statevar->num_elements + i)); - } - } else { - ir_to_mesa_src_reg src(PROGRAM_STATE_VAR, index, NULL); - src.swizzle = element->swizzle; - ir_to_mesa_emit_op1(ir, OPCODE_MOV, dst, src); - /* even a float takes up a whole vec4 reg in a struct/array. */ - dst.index++; - } - } - } - if (storage->file == PROGRAM_TEMPORARY && - dst.index != storage->index + type_size(ir->type)) { - fail_link(this->shader_program, - "failed to load builtin uniform `%s' (%d/%d regs loaded)\n", - ir->name, dst.index - storage->index, - type_size(ir->type)); - } - } -} - -void -ir_to_mesa_visitor::visit(ir_loop *ir) -{ - ir_dereference_variable *counter = NULL; - - if (ir->counter != NULL) - counter = new(ir) ir_dereference_variable(ir->counter); - - if (ir->from != NULL) { - assert(ir->counter != NULL); - - ir_assignment *a = new(ir) ir_assignment(counter, ir->from, NULL); - - a->accept(this); - delete a; - } - - ir_to_mesa_emit_op0(NULL, OPCODE_BGNLOOP); - - if (ir->to) { - ir_expression *e = - new(ir) ir_expression(ir->cmp, glsl_type::bool_type, - counter, ir->to); - ir_if *if_stmt = new(ir) ir_if(e); - - ir_loop_jump *brk = new(ir) ir_loop_jump(ir_loop_jump::jump_break); - - if_stmt->then_instructions.push_tail(brk); - - if_stmt->accept(this); - - delete if_stmt; - delete e; - delete brk; - } - - visit_exec_list(&ir->body_instructions, this); - - if (ir->increment) { - ir_expression *e = - new(ir) ir_expression(ir_binop_add, counter->type, - counter, ir->increment); - - ir_assignment *a = new(ir) ir_assignment(counter, e, NULL); - - a->accept(this); - delete a; - delete e; - } - - ir_to_mesa_emit_op0(NULL, OPCODE_ENDLOOP); -} - -void -ir_to_mesa_visitor::visit(ir_loop_jump *ir) -{ - switch (ir->mode) { - case ir_loop_jump::jump_break: - ir_to_mesa_emit_op0(NULL, OPCODE_BRK); - break; - case ir_loop_jump::jump_continue: - ir_to_mesa_emit_op0(NULL, OPCODE_CONT); - break; - } -} - - -void -ir_to_mesa_visitor::visit(ir_function_signature *ir) -{ - assert(0); - (void)ir; -} - -void -ir_to_mesa_visitor::visit(ir_function *ir) -{ - /* Ignore function bodies other than main() -- we shouldn't see calls to - * them since they should all be inlined before we get to ir_to_mesa. - */ - if (strcmp(ir->name, "main") == 0) { - const ir_function_signature *sig; - exec_list empty; - - sig = ir->matching_signature(&empty); - - assert(sig); - - foreach_iter(exec_list_iterator, iter, sig->body) { - ir_instruction *ir = (ir_instruction *)iter.get(); - - ir->accept(this); - } - } -} - -GLboolean -ir_to_mesa_visitor::try_emit_mad(ir_expression *ir, int mul_operand) -{ - int nonmul_operand = 1 - mul_operand; - ir_to_mesa_src_reg a, b, c; - - ir_expression *expr = ir->operands[mul_operand]->as_expression(); - if (!expr || expr->operation != ir_binop_mul) - return false; - - expr->operands[0]->accept(this); - a = this->result; - expr->operands[1]->accept(this); - b = this->result; - ir->operands[nonmul_operand]->accept(this); - c = this->result; - - this->result = get_temp(ir->type); - ir_to_mesa_emit_op3(ir, OPCODE_MAD, - ir_to_mesa_dst_reg_from_src(this->result), a, b, c); - - return true; -} - -GLboolean -ir_to_mesa_visitor::try_emit_sat(ir_expression *ir) -{ - /* Saturates were only introduced to vertex programs in - * NV_vertex_program3, so don't give them to drivers in the VP. - */ - if (this->prog->Target == GL_VERTEX_PROGRAM_ARB) - return false; - - ir_rvalue *sat_src = ir->as_rvalue_to_saturate(); - if (!sat_src) - return false; - - sat_src->accept(this); - ir_to_mesa_src_reg src = this->result; - - this->result = get_temp(ir->type); - ir_to_mesa_instruction *inst; - inst = ir_to_mesa_emit_op1(ir, OPCODE_MOV, - ir_to_mesa_dst_reg_from_src(this->result), - src); - inst->saturate = true; - - return true; -} - -void -ir_to_mesa_visitor::reladdr_to_temp(ir_instruction *ir, - ir_to_mesa_src_reg *reg, int *num_reladdr) -{ - if (!reg->reladdr) - return; - - ir_to_mesa_emit_op1(ir, OPCODE_ARL, ir_to_mesa_address_reg, *reg->reladdr); - - if (*num_reladdr != 1) { - ir_to_mesa_src_reg temp = get_temp(glsl_type::vec4_type); - - ir_to_mesa_emit_op1(ir, OPCODE_MOV, - ir_to_mesa_dst_reg_from_src(temp), *reg); - *reg = temp; - } - - (*num_reladdr)--; -} - -void -ir_to_mesa_visitor::emit_swz(ir_expression *ir) -{ - /* Assume that the vector operator is in a form compatible with OPCODE_SWZ. - * This means that each of the operands is either an immediate value of -1, - * 0, or 1, or is a component from one source register (possibly with - * negation). - */ - uint8_t components[4] = { 0 }; - bool negate[4] = { false }; - ir_variable *var = NULL; - - for (unsigned i = 0; i < ir->type->vector_elements; i++) { - ir_rvalue *op = ir->operands[i]; - - assert(op->type->is_scalar()); - - while (op != NULL) { - switch (op->ir_type) { - case ir_type_constant: { - - assert(op->type->is_scalar()); - - const ir_constant *const c = op->as_constant(); - if (c->is_one()) { - components[i] = SWIZZLE_ONE; - } else if (c->is_zero()) { - components[i] = SWIZZLE_ZERO; - } else if (c->is_negative_one()) { - components[i] = SWIZZLE_ONE; - negate[i] = true; - } else { - assert(!"SWZ constant must be 0.0 or 1.0."); - } - - op = NULL; - break; - } - - case ir_type_dereference_variable: { - ir_dereference_variable *const deref = - (ir_dereference_variable *) op; - - assert((var == NULL) || (deref->var == var)); - components[i] = SWIZZLE_X; - var = deref->var; - op = NULL; - break; - } - - case ir_type_expression: { - ir_expression *const expr = (ir_expression *) op; - - assert(expr->operation == ir_unop_neg); - negate[i] = true; - - op = expr->operands[0]; - break; - } - - case ir_type_swizzle: { - ir_swizzle *const swiz = (ir_swizzle *) op; - - components[i] = swiz->mask.x; - op = swiz->val; - break; - } - - default: - assert(!"Should not get here."); - return; - } - } - } - - assert(var != NULL); - - ir_dereference_variable *const deref = - new(mem_ctx) ir_dereference_variable(var); - - this->result.file = PROGRAM_UNDEFINED; - deref->accept(this); - if (this->result.file == PROGRAM_UNDEFINED) { - ir_print_visitor v; - printf("Failed to get tree for expression operand:\n"); - deref->accept(&v); - exit(1); - } - - ir_to_mesa_src_reg src; - - src = this->result; - src.swizzle = MAKE_SWIZZLE4(components[0], - components[1], - components[2], - components[3]); - src.negate = ((unsigned(negate[0]) << 0) - | (unsigned(negate[1]) << 1) - | (unsigned(negate[2]) << 2) - | (unsigned(negate[3]) << 3)); - - /* Storage for our result. Ideally for an assignment we'd be using the - * actual storage for the result here, instead. - */ - const ir_to_mesa_src_reg result_src = get_temp(ir->type); - ir_to_mesa_dst_reg result_dst = ir_to_mesa_dst_reg_from_src(result_src); - - /* Limit writes to the channels that will be used by result_src later. - * This does limit this temp's use as a temporary for multi-instruction - * sequences. - */ - result_dst.writemask = (1 << ir->type->vector_elements) - 1; - - ir_to_mesa_emit_op1(ir, OPCODE_SWZ, result_dst, src); - this->result = result_src; -} - -void -ir_to_mesa_visitor::visit(ir_expression *ir) -{ - unsigned int operand; - struct ir_to_mesa_src_reg op[Elements(ir->operands)]; - struct ir_to_mesa_src_reg result_src; - struct ir_to_mesa_dst_reg result_dst; - - /* Quick peephole: Emit OPCODE_MAD(a, b, c) instead of ADD(MUL(a, b), c) - */ - if (ir->operation == ir_binop_add) { - if (try_emit_mad(ir, 1)) - return; - if (try_emit_mad(ir, 0)) - return; - } - if (try_emit_sat(ir)) - return; - - if (ir->operation == ir_quadop_vector) { - this->emit_swz(ir); - return; - } - - for (operand = 0; operand < ir->get_num_operands(); operand++) { - this->result.file = PROGRAM_UNDEFINED; - ir->operands[operand]->accept(this); - if (this->result.file == PROGRAM_UNDEFINED) { - ir_print_visitor v; - printf("Failed to get tree for expression operand:\n"); - ir->operands[operand]->accept(&v); - exit(1); - } - op[operand] = this->result; - - /* Matrix expression operands should have been broken down to vector - * operations already. - */ - assert(!ir->operands[operand]->type->is_matrix()); - } - - int vector_elements = ir->operands[0]->type->vector_elements; - if (ir->operands[1]) { - vector_elements = MAX2(vector_elements, - ir->operands[1]->type->vector_elements); - } - - this->result.file = PROGRAM_UNDEFINED; - - /* Storage for our result. Ideally for an assignment we'd be using - * the actual storage for the result here, instead. - */ - result_src = get_temp(ir->type); - /* convenience for the emit functions below. */ - result_dst = ir_to_mesa_dst_reg_from_src(result_src); - /* Limit writes to the channels that will be used by result_src later. - * This does limit this temp's use as a temporary for multi-instruction - * sequences. - */ - result_dst.writemask = (1 << ir->type->vector_elements) - 1; - - switch (ir->operation) { - case ir_unop_logic_not: - ir_to_mesa_emit_op2(ir, OPCODE_SEQ, result_dst, - op[0], src_reg_for_float(0.0)); - break; - case ir_unop_neg: - op[0].negate = ~op[0].negate; - result_src = op[0]; - break; - case ir_unop_abs: - ir_to_mesa_emit_op1(ir, OPCODE_ABS, result_dst, op[0]); - break; - case ir_unop_sign: - ir_to_mesa_emit_op1(ir, OPCODE_SSG, result_dst, op[0]); - break; - case ir_unop_rcp: - ir_to_mesa_emit_scalar_op1(ir, OPCODE_RCP, result_dst, op[0]); - break; - - case ir_unop_exp2: - ir_to_mesa_emit_scalar_op1(ir, OPCODE_EX2, result_dst, op[0]); - break; - case ir_unop_exp: - case ir_unop_log: - assert(!"not reached: should be handled by ir_explog_to_explog2"); - break; - case ir_unop_log2: - ir_to_mesa_emit_scalar_op1(ir, OPCODE_LG2, result_dst, op[0]); - break; - case ir_unop_sin: - ir_to_mesa_emit_scalar_op1(ir, OPCODE_SIN, result_dst, op[0]); - break; - case ir_unop_cos: - ir_to_mesa_emit_scalar_op1(ir, OPCODE_COS, result_dst, op[0]); - break; - case ir_unop_sin_reduced: - emit_scs(ir, OPCODE_SIN, result_dst, op[0]); - break; - case ir_unop_cos_reduced: - emit_scs(ir, OPCODE_COS, result_dst, op[0]); - break; - - case ir_unop_dFdx: - ir_to_mesa_emit_op1(ir, OPCODE_DDX, result_dst, op[0]); - break; - case ir_unop_dFdy: - ir_to_mesa_emit_op1(ir, OPCODE_DDY, result_dst, op[0]); - break; - - case ir_unop_noise: { - const enum prog_opcode opcode = - prog_opcode(OPCODE_NOISE1 - + (ir->operands[0]->type->vector_elements) - 1); - assert((opcode >= OPCODE_NOISE1) && (opcode <= OPCODE_NOISE4)); - - ir_to_mesa_emit_op1(ir, opcode, result_dst, op[0]); - break; - } - - case ir_binop_add: - ir_to_mesa_emit_op2(ir, OPCODE_ADD, result_dst, op[0], op[1]); - break; - case ir_binop_sub: - ir_to_mesa_emit_op2(ir, OPCODE_SUB, result_dst, op[0], op[1]); - break; - - case ir_binop_mul: - ir_to_mesa_emit_op2(ir, OPCODE_MUL, result_dst, op[0], op[1]); - break; - case ir_binop_div: - assert(!"not reached: should be handled by ir_div_to_mul_rcp"); - case ir_binop_mod: - assert(!"ir_binop_mod should have been converted to b * fract(a/b)"); - break; - - case ir_binop_less: - ir_to_mesa_emit_op2(ir, OPCODE_SLT, result_dst, op[0], op[1]); - break; - case ir_binop_greater: - ir_to_mesa_emit_op2(ir, OPCODE_SGT, result_dst, op[0], op[1]); - break; - case ir_binop_lequal: - ir_to_mesa_emit_op2(ir, OPCODE_SLE, result_dst, op[0], op[1]); - break; - case ir_binop_gequal: - ir_to_mesa_emit_op2(ir, OPCODE_SGE, result_dst, op[0], op[1]); - break; - case ir_binop_equal: - ir_to_mesa_emit_op2(ir, OPCODE_SEQ, result_dst, op[0], op[1]); - break; - case ir_binop_nequal: - ir_to_mesa_emit_op2(ir, OPCODE_SNE, result_dst, op[0], op[1]); - break; - case ir_binop_all_equal: - /* "==" operator producing a scalar boolean. */ - if (ir->operands[0]->type->is_vector() || - ir->operands[1]->type->is_vector()) { - ir_to_mesa_src_reg temp = get_temp(glsl_type::vec4_type); - ir_to_mesa_emit_op2(ir, OPCODE_SNE, - ir_to_mesa_dst_reg_from_src(temp), op[0], op[1]); - ir_to_mesa_emit_dp(ir, result_dst, temp, temp, vector_elements); - ir_to_mesa_emit_op2(ir, OPCODE_SEQ, - result_dst, result_src, src_reg_for_float(0.0)); - } else { - ir_to_mesa_emit_op2(ir, OPCODE_SEQ, result_dst, op[0], op[1]); - } - break; - case ir_binop_any_nequal: - /* "!=" operator producing a scalar boolean. */ - if (ir->operands[0]->type->is_vector() || - ir->operands[1]->type->is_vector()) { - ir_to_mesa_src_reg temp = get_temp(glsl_type::vec4_type); - ir_to_mesa_emit_op2(ir, OPCODE_SNE, - ir_to_mesa_dst_reg_from_src(temp), op[0], op[1]); - ir_to_mesa_emit_dp(ir, result_dst, temp, temp, vector_elements); - ir_to_mesa_emit_op2(ir, OPCODE_SNE, - result_dst, result_src, src_reg_for_float(0.0)); - } else { - ir_to_mesa_emit_op2(ir, OPCODE_SNE, result_dst, op[0], op[1]); - } - break; - - case ir_unop_any: - assert(ir->operands[0]->type->is_vector()); - ir_to_mesa_emit_dp(ir, result_dst, op[0], op[0], - ir->operands[0]->type->vector_elements); - ir_to_mesa_emit_op2(ir, OPCODE_SNE, - result_dst, result_src, src_reg_for_float(0.0)); - break; - - case ir_binop_logic_xor: - ir_to_mesa_emit_op2(ir, OPCODE_SNE, result_dst, op[0], op[1]); - break; - - case ir_binop_logic_or: - /* This could be a saturated add and skip the SNE. */ - ir_to_mesa_emit_op2(ir, OPCODE_ADD, - result_dst, - op[0], op[1]); - - ir_to_mesa_emit_op2(ir, OPCODE_SNE, - result_dst, - result_src, src_reg_for_float(0.0)); - break; - - case ir_binop_logic_and: - /* the bool args are stored as float 0.0 or 1.0, so "mul" gives us "and". */ - ir_to_mesa_emit_op2(ir, OPCODE_MUL, - result_dst, - op[0], op[1]); - break; - - case ir_binop_dot: - assert(ir->operands[0]->type->is_vector()); - assert(ir->operands[0]->type == ir->operands[1]->type); - ir_to_mesa_emit_dp(ir, result_dst, op[0], op[1], - ir->operands[0]->type->vector_elements); - break; - - case ir_unop_sqrt: - /* sqrt(x) = x * rsq(x). */ - ir_to_mesa_emit_scalar_op1(ir, OPCODE_RSQ, result_dst, op[0]); - ir_to_mesa_emit_op2(ir, OPCODE_MUL, result_dst, result_src, op[0]); - /* For incoming channels <= 0, set the result to 0. */ - op[0].negate = ~op[0].negate; - ir_to_mesa_emit_op3(ir, OPCODE_CMP, result_dst, - op[0], result_src, src_reg_for_float(0.0)); - break; - case ir_unop_rsq: - ir_to_mesa_emit_scalar_op1(ir, OPCODE_RSQ, result_dst, op[0]); - break; - case ir_unop_i2f: - case ir_unop_b2f: - case ir_unop_b2i: - /* Mesa IR lacks types, ints are stored as truncated floats. */ - result_src = op[0]; - break; - case ir_unop_f2i: - ir_to_mesa_emit_op1(ir, OPCODE_TRUNC, result_dst, op[0]); - break; - case ir_unop_f2b: - case ir_unop_i2b: - ir_to_mesa_emit_op2(ir, OPCODE_SNE, result_dst, - op[0], src_reg_for_float(0.0)); - break; - case ir_unop_trunc: - ir_to_mesa_emit_op1(ir, OPCODE_TRUNC, result_dst, op[0]); - break; - case ir_unop_ceil: - op[0].negate = ~op[0].negate; - ir_to_mesa_emit_op1(ir, OPCODE_FLR, result_dst, op[0]); - result_src.negate = ~result_src.negate; - break; - case ir_unop_floor: - ir_to_mesa_emit_op1(ir, OPCODE_FLR, result_dst, op[0]); - break; - case ir_unop_fract: - ir_to_mesa_emit_op1(ir, OPCODE_FRC, result_dst, op[0]); - break; - - case ir_binop_min: - ir_to_mesa_emit_op2(ir, OPCODE_MIN, result_dst, op[0], op[1]); - break; - case ir_binop_max: - ir_to_mesa_emit_op2(ir, OPCODE_MAX, result_dst, op[0], op[1]); - break; - case ir_binop_pow: - ir_to_mesa_emit_scalar_op2(ir, OPCODE_POW, result_dst, op[0], op[1]); - break; - - case ir_unop_bit_not: - case ir_unop_u2f: - case ir_binop_lshift: - case ir_binop_rshift: - case ir_binop_bit_and: - case ir_binop_bit_xor: - case ir_binop_bit_or: - case ir_unop_round_even: - assert(!"GLSL 1.30 features unsupported"); - break; - - case ir_quadop_vector: - /* This operation should have already been handled. - */ - assert(!"Should not get here."); - break; - } - - this->result = result_src; -} - - -void -ir_to_mesa_visitor::visit(ir_swizzle *ir) -{ - ir_to_mesa_src_reg src_reg; - int i; - int swizzle[4]; - - /* Note that this is only swizzles in expressions, not those on the left - * hand side of an assignment, which do write masking. See ir_assignment - * for that. - */ - - ir->val->accept(this); - src_reg = this->result; - assert(src_reg.file != PROGRAM_UNDEFINED); - - for (i = 0; i < 4; i++) { - if (i < ir->type->vector_elements) { - switch (i) { - case 0: - swizzle[i] = GET_SWZ(src_reg.swizzle, ir->mask.x); - break; - case 1: - swizzle[i] = GET_SWZ(src_reg.swizzle, ir->mask.y); - break; - case 2: - swizzle[i] = GET_SWZ(src_reg.swizzle, ir->mask.z); - break; - case 3: - swizzle[i] = GET_SWZ(src_reg.swizzle, ir->mask.w); - break; - } - } else { - /* If the type is smaller than a vec4, replicate the last - * channel out. - */ - swizzle[i] = swizzle[ir->type->vector_elements - 1]; - } - } - - src_reg.swizzle = MAKE_SWIZZLE4(swizzle[0], - swizzle[1], - swizzle[2], - swizzle[3]); - - this->result = src_reg; -} - -void -ir_to_mesa_visitor::visit(ir_dereference_variable *ir) -{ - variable_storage *entry = find_variable_storage(ir->var); - ir_variable *var = ir->var; - - if (!entry) { - switch (var->mode) { - case ir_var_uniform: - entry = new(mem_ctx) variable_storage(var, PROGRAM_UNIFORM, - var->location); - this->variables.push_tail(entry); - break; - case ir_var_in: - case ir_var_inout: - /* The linker assigns locations for varyings and attributes, - * including deprecated builtins (like gl_Color), user-assign - * generic attributes (glBindVertexLocation), and - * user-defined varyings. - * - * FINISHME: We would hit this path for function arguments. Fix! - */ - assert(var->location != -1); - entry = new(mem_ctx) variable_storage(var, - PROGRAM_INPUT, - var->location); - if (this->prog->Target == GL_VERTEX_PROGRAM_ARB && - var->location >= VERT_ATTRIB_GENERIC0) { - _mesa_add_attribute(this->prog->Attributes, - var->name, - _mesa_sizeof_glsl_type(var->type->gl_type), - var->type->gl_type, - var->location - VERT_ATTRIB_GENERIC0); - } - break; - case ir_var_out: - assert(var->location != -1); - entry = new(mem_ctx) variable_storage(var, - PROGRAM_OUTPUT, - var->location); - break; - case ir_var_system_value: - entry = new(mem_ctx) variable_storage(var, - PROGRAM_SYSTEM_VALUE, - var->location); - break; - case ir_var_auto: - case ir_var_temporary: - entry = new(mem_ctx) variable_storage(var, PROGRAM_TEMPORARY, - this->next_temp); - this->variables.push_tail(entry); - - next_temp += type_size(var->type); - break; - } - - if (!entry) { - printf("Failed to make storage for %s\n", var->name); - exit(1); - } - } - - this->result = ir_to_mesa_src_reg(entry->file, entry->index, var->type); -} - -void -ir_to_mesa_visitor::visit(ir_dereference_array *ir) -{ - ir_constant *index; - ir_to_mesa_src_reg src_reg; - int element_size = type_size(ir->type); - - index = ir->array_index->constant_expression_value(); - - ir->array->accept(this); - src_reg = this->result; - - if (index) { - src_reg.index += index->value.i[0] * element_size; - } else { - ir_to_mesa_src_reg array_base = this->result; - /* Variable index array dereference. It eats the "vec4" of the - * base of the array and an index that offsets the Mesa register - * index. - */ - ir->array_index->accept(this); - - ir_to_mesa_src_reg index_reg; - - if (element_size == 1) { - index_reg = this->result; - } else { - index_reg = get_temp(glsl_type::float_type); - - ir_to_mesa_emit_op2(ir, OPCODE_MUL, - ir_to_mesa_dst_reg_from_src(index_reg), - this->result, src_reg_for_float(element_size)); - } - - src_reg.reladdr = ralloc(mem_ctx, ir_to_mesa_src_reg); - memcpy(src_reg.reladdr, &index_reg, sizeof(index_reg)); - } - - /* If the type is smaller than a vec4, replicate the last channel out. */ - if (ir->type->is_scalar() || ir->type->is_vector()) - src_reg.swizzle = swizzle_for_size(ir->type->vector_elements); - else - src_reg.swizzle = SWIZZLE_NOOP; - - this->result = src_reg; -} - -void -ir_to_mesa_visitor::visit(ir_dereference_record *ir) -{ - unsigned int i; - const glsl_type *struct_type = ir->record->type; - int offset = 0; - - ir->record->accept(this); - - for (i = 0; i < struct_type->length; i++) { - if (strcmp(struct_type->fields.structure[i].name, ir->field) == 0) - break; - offset += type_size(struct_type->fields.structure[i].type); - } - - /* If the type is smaller than a vec4, replicate the last channel out. */ - if (ir->type->is_scalar() || ir->type->is_vector()) - this->result.swizzle = swizzle_for_size(ir->type->vector_elements); - else - this->result.swizzle = SWIZZLE_NOOP; - - this->result.index += offset; -} - -/** - * We want to be careful in assignment setup to hit the actual storage - * instead of potentially using a temporary like we might with the - * ir_dereference handler. - */ -static struct ir_to_mesa_dst_reg -get_assignment_lhs(ir_dereference *ir, ir_to_mesa_visitor *v) -{ - /* The LHS must be a dereference. If the LHS is a variable indexed array - * access of a vector, it must be separated into a series conditional moves - * before reaching this point (see ir_vec_index_to_cond_assign). - */ - assert(ir->as_dereference()); - ir_dereference_array *deref_array = ir->as_dereference_array(); - if (deref_array) { - assert(!deref_array->array->type->is_vector()); - } - - /* Use the rvalue deref handler for the most part. We'll ignore - * swizzles in it and write swizzles using writemask, though. - */ - ir->accept(v); - return ir_to_mesa_dst_reg_from_src(v->result); -} - -/** - * Process the condition of a conditional assignment - * - * Examines the condition of a conditional assignment to generate the optimal - * first operand of a \c CMP instruction. If the condition is a relational - * operator with 0 (e.g., \c ir_binop_less), the value being compared will be - * used as the source for the \c CMP instruction. Otherwise the comparison - * is processed to a boolean result, and the boolean result is used as the - * operand to the CMP instruction. - */ -bool -ir_to_mesa_visitor::process_move_condition(ir_rvalue *ir) -{ - ir_rvalue *src_ir = ir; - bool negate = true; - bool switch_order = false; - - ir_expression *const expr = ir->as_expression(); - if ((expr != NULL) && (expr->get_num_operands() == 2)) { - bool zero_on_left = false; - - if (expr->operands[0]->is_zero()) { - src_ir = expr->operands[1]; - zero_on_left = true; - } else if (expr->operands[1]->is_zero()) { - src_ir = expr->operands[0]; - zero_on_left = false; - } - - /* a is - 0 + - 0 + - * (a < 0) T F F ( a < 0) T F F - * (0 < a) F F T (-a < 0) F F T - * (a <= 0) T T F (-a < 0) F F T (swap order of other operands) - * (0 <= a) F T T ( a < 0) T F F (swap order of other operands) - * (a > 0) F F T (-a < 0) F F T - * (0 > a) T F F ( a < 0) T F F - * (a >= 0) F T T ( a < 0) T F F (swap order of other operands) - * (0 >= a) T T F (-a < 0) F F T (swap order of other operands) - * - * Note that exchanging the order of 0 and 'a' in the comparison simply - * means that the value of 'a' should be negated. - */ - if (src_ir != ir) { - switch (expr->operation) { - case ir_binop_less: - switch_order = false; - negate = zero_on_left; - break; - - case ir_binop_greater: - switch_order = false; - negate = !zero_on_left; - break; - - case ir_binop_lequal: - switch_order = true; - negate = !zero_on_left; - break; - - case ir_binop_gequal: - switch_order = true; - negate = zero_on_left; - break; - - default: - /* This isn't the right kind of comparison afterall, so make sure - * the whole condition is visited. - */ - src_ir = ir; - break; - } - } - } - - src_ir->accept(this); - - /* We use the OPCODE_CMP (a < 0 ? b : c) for conditional moves, and the - * condition we produced is 0.0 or 1.0. By flipping the sign, we can - * choose which value OPCODE_CMP produces without an extra instruction - * computing the condition. - */ - if (negate) - this->result.negate = ~this->result.negate; - - return switch_order; -} - -void -ir_to_mesa_visitor::visit(ir_assignment *ir) -{ - struct ir_to_mesa_dst_reg l; - struct ir_to_mesa_src_reg r; - int i; - - ir->rhs->accept(this); - r = this->result; - - l = get_assignment_lhs(ir->lhs, this); - - /* FINISHME: This should really set to the correct maximal writemask for each - * FINISHME: component written (in the loops below). This case can only - * FINISHME: occur for matrices, arrays, and structures. - */ - if (ir->write_mask == 0) { - assert(!ir->lhs->type->is_scalar() && !ir->lhs->type->is_vector()); - l.writemask = WRITEMASK_XYZW; - } else if (ir->lhs->type->is_scalar()) { - /* FINISHME: This hack makes writing to gl_FragDepth, which lives in the - * FINISHME: W component of fragment shader output zero, work correctly. - */ - l.writemask = WRITEMASK_XYZW; - } else { - int swizzles[4]; - int first_enabled_chan = 0; - int rhs_chan = 0; - - assert(ir->lhs->type->is_vector()); - l.writemask = ir->write_mask; - - for (int i = 0; i < 4; i++) { - if (l.writemask & (1 << i)) { - first_enabled_chan = GET_SWZ(r.swizzle, i); - break; - } - } - - /* Swizzle a small RHS vector into the channels being written. - * - * glsl ir treats write_mask as dictating how many channels are - * present on the RHS while Mesa IR treats write_mask as just - * showing which channels of the vec4 RHS get written. - */ - for (int i = 0; i < 4; i++) { - if (l.writemask & (1 << i)) - swizzles[i] = GET_SWZ(r.swizzle, rhs_chan++); - else - swizzles[i] = first_enabled_chan; - } - r.swizzle = MAKE_SWIZZLE4(swizzles[0], swizzles[1], - swizzles[2], swizzles[3]); - } - - assert(l.file != PROGRAM_UNDEFINED); - assert(r.file != PROGRAM_UNDEFINED); - - if (ir->condition) { - const bool switch_order = this->process_move_condition(ir->condition); - ir_to_mesa_src_reg condition = this->result; - - for (i = 0; i < type_size(ir->lhs->type); i++) { - if (switch_order) { - ir_to_mesa_emit_op3(ir, OPCODE_CMP, l, - condition, ir_to_mesa_src_reg_from_dst(l), r); - } else { - ir_to_mesa_emit_op3(ir, OPCODE_CMP, l, - condition, r, ir_to_mesa_src_reg_from_dst(l)); - } - - l.index++; - r.index++; - } - } else { - for (i = 0; i < type_size(ir->lhs->type); i++) { - ir_to_mesa_emit_op1(ir, OPCODE_MOV, l, r); - l.index++; - r.index++; - } - } -} - - -void -ir_to_mesa_visitor::visit(ir_constant *ir) -{ - ir_to_mesa_src_reg src_reg; - GLfloat stack_vals[4] = { 0 }; - GLfloat *values = stack_vals; - unsigned int i; - - /* Unfortunately, 4 floats is all we can get into - * _mesa_add_unnamed_constant. So, make a temp to store an - * aggregate constant and move each constant value into it. If we - * get lucky, copy propagation will eliminate the extra moves. - */ - - if (ir->type->base_type == GLSL_TYPE_STRUCT) { - ir_to_mesa_src_reg temp_base = get_temp(ir->type); - ir_to_mesa_dst_reg temp = ir_to_mesa_dst_reg_from_src(temp_base); - - foreach_iter(exec_list_iterator, iter, ir->components) { - ir_constant *field_value = (ir_constant *)iter.get(); - int size = type_size(field_value->type); - - assert(size > 0); - - field_value->accept(this); - src_reg = this->result; - - for (i = 0; i < (unsigned int)size; i++) { - ir_to_mesa_emit_op1(ir, OPCODE_MOV, temp, src_reg); - - src_reg.index++; - temp.index++; - } - } - this->result = temp_base; - return; - } - - if (ir->type->is_array()) { - ir_to_mesa_src_reg temp_base = get_temp(ir->type); - ir_to_mesa_dst_reg temp = ir_to_mesa_dst_reg_from_src(temp_base); - int size = type_size(ir->type->fields.array); - - assert(size > 0); - - for (i = 0; i < ir->type->length; i++) { - ir->array_elements[i]->accept(this); - src_reg = this->result; - for (int j = 0; j < size; j++) { - ir_to_mesa_emit_op1(ir, OPCODE_MOV, temp, src_reg); - - src_reg.index++; - temp.index++; - } - } - this->result = temp_base; - return; - } - - if (ir->type->is_matrix()) { - ir_to_mesa_src_reg mat = get_temp(ir->type); - ir_to_mesa_dst_reg mat_column = ir_to_mesa_dst_reg_from_src(mat); - - for (i = 0; i < ir->type->matrix_columns; i++) { - assert(ir->type->base_type == GLSL_TYPE_FLOAT); - values = &ir->value.f[i * ir->type->vector_elements]; - - src_reg = ir_to_mesa_src_reg(PROGRAM_CONSTANT, -1, NULL); - src_reg.index = _mesa_add_unnamed_constant(this->prog->Parameters, - values, - ir->type->vector_elements, - &src_reg.swizzle); - ir_to_mesa_emit_op1(ir, OPCODE_MOV, mat_column, src_reg); - - mat_column.index++; - } - - this->result = mat; - return; - } - - src_reg.file = PROGRAM_CONSTANT; - switch (ir->type->base_type) { - case GLSL_TYPE_FLOAT: - values = &ir->value.f[0]; - break; - case GLSL_TYPE_UINT: - for (i = 0; i < ir->type->vector_elements; i++) { - values[i] = ir->value.u[i]; - } - break; - case GLSL_TYPE_INT: - for (i = 0; i < ir->type->vector_elements; i++) { - values[i] = ir->value.i[i]; - } - break; - case GLSL_TYPE_BOOL: - for (i = 0; i < ir->type->vector_elements; i++) { - values[i] = ir->value.b[i]; - } - break; - default: - assert(!"Non-float/uint/int/bool constant"); - } - - this->result = ir_to_mesa_src_reg(PROGRAM_CONSTANT, -1, ir->type); - this->result.index = _mesa_add_unnamed_constant(this->prog->Parameters, - values, - ir->type->vector_elements, - &this->result.swizzle); -} - -function_entry * -ir_to_mesa_visitor::get_function_signature(ir_function_signature *sig) -{ - function_entry *entry; - - foreach_iter(exec_list_iterator, iter, this->function_signatures) { - entry = (function_entry *)iter.get(); - - if (entry->sig == sig) - return entry; - } - - entry = ralloc(mem_ctx, function_entry); - entry->sig = sig; - entry->sig_id = this->next_signature_id++; - entry->bgn_inst = NULL; - - /* Allocate storage for all the parameters. */ - foreach_iter(exec_list_iterator, iter, sig->parameters) { - ir_variable *param = (ir_variable *)iter.get(); - variable_storage *storage; - - storage = find_variable_storage(param); - assert(!storage); - - storage = new(mem_ctx) variable_storage(param, PROGRAM_TEMPORARY, - this->next_temp); - this->variables.push_tail(storage); - - this->next_temp += type_size(param->type); - } - - if (!sig->return_type->is_void()) { - entry->return_reg = get_temp(sig->return_type); - } else { - entry->return_reg = ir_to_mesa_undef; - } - - this->function_signatures.push_tail(entry); - return entry; -} - -void -ir_to_mesa_visitor::visit(ir_call *ir) -{ - ir_to_mesa_instruction *call_inst; - ir_function_signature *sig = ir->get_callee(); - function_entry *entry = get_function_signature(sig); - int i; - - /* Process in parameters. */ - exec_list_iterator sig_iter = sig->parameters.iterator(); - foreach_iter(exec_list_iterator, iter, *ir) { - ir_rvalue *param_rval = (ir_rvalue *)iter.get(); - ir_variable *param = (ir_variable *)sig_iter.get(); - - if (param->mode == ir_var_in || - param->mode == ir_var_inout) { - variable_storage *storage = find_variable_storage(param); - assert(storage); - - param_rval->accept(this); - ir_to_mesa_src_reg r = this->result; - - ir_to_mesa_dst_reg l; - l.file = storage->file; - l.index = storage->index; - l.reladdr = NULL; - l.writemask = WRITEMASK_XYZW; - l.cond_mask = COND_TR; - - for (i = 0; i < type_size(param->type); i++) { - ir_to_mesa_emit_op1(ir, OPCODE_MOV, l, r); - l.index++; - r.index++; - } - } - - sig_iter.next(); - } - assert(!sig_iter.has_next()); - - /* Emit call instruction */ - call_inst = ir_to_mesa_emit_op1(ir, OPCODE_CAL, - ir_to_mesa_undef_dst, ir_to_mesa_undef); - call_inst->function = entry; - - /* Process out parameters. */ - sig_iter = sig->parameters.iterator(); - foreach_iter(exec_list_iterator, iter, *ir) { - ir_rvalue *param_rval = (ir_rvalue *)iter.get(); - ir_variable *param = (ir_variable *)sig_iter.get(); - - if (param->mode == ir_var_out || - param->mode == ir_var_inout) { - variable_storage *storage = find_variable_storage(param); - assert(storage); - - ir_to_mesa_src_reg r; - r.file = storage->file; - r.index = storage->index; - r.reladdr = NULL; - r.swizzle = SWIZZLE_NOOP; - r.negate = 0; - - param_rval->accept(this); - ir_to_mesa_dst_reg l = ir_to_mesa_dst_reg_from_src(this->result); - - for (i = 0; i < type_size(param->type); i++) { - ir_to_mesa_emit_op1(ir, OPCODE_MOV, l, r); - l.index++; - r.index++; - } - } - - sig_iter.next(); - } - assert(!sig_iter.has_next()); - - /* Process return value. */ - this->result = entry->return_reg; -} - -void -ir_to_mesa_visitor::visit(ir_texture *ir) -{ - ir_to_mesa_src_reg result_src, coord, lod_info, projector; - ir_to_mesa_dst_reg result_dst, coord_dst; - ir_to_mesa_instruction *inst = NULL; - prog_opcode opcode = OPCODE_NOP; - - ir->coordinate->accept(this); - - /* Put our coords in a temp. We'll need to modify them for shadow, - * projection, or LOD, so the only case we'd use it as is is if - * we're doing plain old texturing. Mesa IR optimization should - * handle cleaning up our mess in that case. - */ - coord = get_temp(glsl_type::vec4_type); - coord_dst = ir_to_mesa_dst_reg_from_src(coord); - ir_to_mesa_emit_op1(ir, OPCODE_MOV, coord_dst, - this->result); - - if (ir->projector) { - ir->projector->accept(this); - projector = this->result; - } - - /* Storage for our result. Ideally for an assignment we'd be using - * the actual storage for the result here, instead. - */ - result_src = get_temp(glsl_type::vec4_type); - result_dst = ir_to_mesa_dst_reg_from_src(result_src); - - switch (ir->op) { - case ir_tex: - opcode = OPCODE_TEX; - break; - case ir_txb: - opcode = OPCODE_TXB; - ir->lod_info.bias->accept(this); - lod_info = this->result; - break; - case ir_txl: - opcode = OPCODE_TXL; - ir->lod_info.lod->accept(this); - lod_info = this->result; - break; - case ir_txd: - case ir_txf: - assert(!"GLSL 1.30 features unsupported"); - break; - } - - if (ir->projector) { - if (opcode == OPCODE_TEX) { - /* Slot the projector in as the last component of the coord. */ - coord_dst.writemask = WRITEMASK_W; - ir_to_mesa_emit_op1(ir, OPCODE_MOV, coord_dst, projector); - coord_dst.writemask = WRITEMASK_XYZW; - opcode = OPCODE_TXP; - } else { - ir_to_mesa_src_reg coord_w = coord; - coord_w.swizzle = SWIZZLE_WWWW; - - /* For the other TEX opcodes there's no projective version - * since the last slot is taken up by lod info. Do the - * projective divide now. - */ - coord_dst.writemask = WRITEMASK_W; - ir_to_mesa_emit_op1(ir, OPCODE_RCP, coord_dst, projector); - - coord_dst.writemask = WRITEMASK_XYZ; - ir_to_mesa_emit_op2(ir, OPCODE_MUL, coord_dst, coord, coord_w); - - coord_dst.writemask = WRITEMASK_XYZW; - coord.swizzle = SWIZZLE_XYZW; - } - } - - if (ir->shadow_comparitor) { - /* Slot the shadow value in as the second to last component of the - * coord. - */ - ir->shadow_comparitor->accept(this); - coord_dst.writemask = WRITEMASK_Z; - ir_to_mesa_emit_op1(ir, OPCODE_MOV, coord_dst, this->result); - coord_dst.writemask = WRITEMASK_XYZW; - } - - if (opcode == OPCODE_TXL || opcode == OPCODE_TXB) { - /* Mesa IR stores lod or lod bias in the last channel of the coords. */ - coord_dst.writemask = WRITEMASK_W; - ir_to_mesa_emit_op1(ir, OPCODE_MOV, coord_dst, lod_info); - coord_dst.writemask = WRITEMASK_XYZW; - } - - inst = ir_to_mesa_emit_op1(ir, opcode, result_dst, coord); - - if (ir->shadow_comparitor) - inst->tex_shadow = GL_TRUE; - - inst->sampler = _mesa_get_sampler_uniform_value(ir->sampler, - this->shader_program, - this->prog); - - const glsl_type *sampler_type = ir->sampler->type; - - switch (sampler_type->sampler_dimensionality) { - case GLSL_SAMPLER_DIM_1D: - inst->tex_target = (sampler_type->sampler_array) - ? TEXTURE_1D_ARRAY_INDEX : TEXTURE_1D_INDEX; - break; - case GLSL_SAMPLER_DIM_2D: - inst->tex_target = (sampler_type->sampler_array) - ? TEXTURE_2D_ARRAY_INDEX : TEXTURE_2D_INDEX; - break; - case GLSL_SAMPLER_DIM_3D: - inst->tex_target = TEXTURE_3D_INDEX; - break; - case GLSL_SAMPLER_DIM_CUBE: - inst->tex_target = TEXTURE_CUBE_INDEX; - break; - case GLSL_SAMPLER_DIM_RECT: - inst->tex_target = TEXTURE_RECT_INDEX; - break; - case GLSL_SAMPLER_DIM_BUF: - assert(!"FINISHME: Implement ARB_texture_buffer_object"); - break; - default: - assert(!"Should not get here."); - } - - this->result = result_src; -} - -void -ir_to_mesa_visitor::visit(ir_return *ir) -{ - if (ir->get_value()) { - ir_to_mesa_dst_reg l; - int i; - - assert(current_function); - - ir->get_value()->accept(this); - ir_to_mesa_src_reg r = this->result; - - l = ir_to_mesa_dst_reg_from_src(current_function->return_reg); - - for (i = 0; i < type_size(current_function->sig->return_type); i++) { - ir_to_mesa_emit_op1(ir, OPCODE_MOV, l, r); - l.index++; - r.index++; - } - } - - ir_to_mesa_emit_op0(ir, OPCODE_RET); -} - -void -ir_to_mesa_visitor::visit(ir_discard *ir) -{ - struct gl_fragment_program *fp = (struct gl_fragment_program *)this->prog; - - if (ir->condition) { - ir->condition->accept(this); - this->result.negate = ~this->result.negate; - ir_to_mesa_emit_op1(ir, OPCODE_KIL, ir_to_mesa_undef_dst, this->result); - } else { - ir_to_mesa_emit_op0(ir, OPCODE_KIL_NV); - } - - fp->UsesKill = GL_TRUE; -} - -void -ir_to_mesa_visitor::visit(ir_if *ir) -{ - ir_to_mesa_instruction *cond_inst, *if_inst, *else_inst = NULL; - ir_to_mesa_instruction *prev_inst; - - prev_inst = (ir_to_mesa_instruction *)this->instructions.get_tail(); - - ir->condition->accept(this); - assert(this->result.file != PROGRAM_UNDEFINED); - - if (this->options->EmitCondCodes) { - cond_inst = (ir_to_mesa_instruction *)this->instructions.get_tail(); - - /* See if we actually generated any instruction for generating - * the condition. If not, then cook up a move to a temp so we - * have something to set cond_update on. - */ - if (cond_inst == prev_inst) { - ir_to_mesa_src_reg temp = get_temp(glsl_type::bool_type); - cond_inst = ir_to_mesa_emit_op1(ir->condition, OPCODE_MOV, - ir_to_mesa_dst_reg_from_src(temp), - result); - } - cond_inst->cond_update = GL_TRUE; - - if_inst = ir_to_mesa_emit_op0(ir->condition, OPCODE_IF); - if_inst->dst_reg.cond_mask = COND_NE; - } else { - if_inst = ir_to_mesa_emit_op1(ir->condition, - OPCODE_IF, ir_to_mesa_undef_dst, - this->result); - } - - this->instructions.push_tail(if_inst); - - visit_exec_list(&ir->then_instructions, this); - - if (!ir->else_instructions.is_empty()) { - else_inst = ir_to_mesa_emit_op0(ir->condition, OPCODE_ELSE); - visit_exec_list(&ir->else_instructions, this); - } - - if_inst = ir_to_mesa_emit_op1(ir->condition, OPCODE_ENDIF, - ir_to_mesa_undef_dst, ir_to_mesa_undef); -} - -ir_to_mesa_visitor::ir_to_mesa_visitor() -{ - result.file = PROGRAM_UNDEFINED; - next_temp = 1; - next_signature_id = 1; - current_function = NULL; - mem_ctx = ralloc_context(NULL); -} - -ir_to_mesa_visitor::~ir_to_mesa_visitor() -{ - ralloc_free(mem_ctx); -} - -static struct prog_src_register -mesa_src_reg_from_ir_src_reg(ir_to_mesa_src_reg reg) -{ - struct prog_src_register mesa_reg; - - mesa_reg.File = reg.file; - assert(reg.index < (1 << INST_INDEX_BITS)); - mesa_reg.Index = reg.index; - mesa_reg.Swizzle = reg.swizzle; - mesa_reg.RelAddr = reg.reladdr != NULL; - mesa_reg.Negate = reg.negate; - mesa_reg.Abs = 0; - mesa_reg.HasIndex2 = GL_FALSE; - mesa_reg.RelAddr2 = 0; - mesa_reg.Index2 = 0; - - return mesa_reg; -} - -static void -set_branchtargets(ir_to_mesa_visitor *v, - struct prog_instruction *mesa_instructions, - int num_instructions) -{ - int if_count = 0, loop_count = 0; - int *if_stack, *loop_stack; - int if_stack_pos = 0, loop_stack_pos = 0; - int i, j; - - for (i = 0; i < num_instructions; i++) { - switch (mesa_instructions[i].Opcode) { - case OPCODE_IF: - if_count++; - break; - case OPCODE_BGNLOOP: - loop_count++; - break; - case OPCODE_BRK: - case OPCODE_CONT: - mesa_instructions[i].BranchTarget = -1; - break; - default: - break; - } - } - - if_stack = rzalloc_array(v->mem_ctx, int, if_count); - loop_stack = rzalloc_array(v->mem_ctx, int, loop_count); - - for (i = 0; i < num_instructions; i++) { - switch (mesa_instructions[i].Opcode) { - case OPCODE_IF: - if_stack[if_stack_pos] = i; - if_stack_pos++; - break; - case OPCODE_ELSE: - mesa_instructions[if_stack[if_stack_pos - 1]].BranchTarget = i; - if_stack[if_stack_pos - 1] = i; - break; - case OPCODE_ENDIF: - mesa_instructions[if_stack[if_stack_pos - 1]].BranchTarget = i; - if_stack_pos--; - break; - case OPCODE_BGNLOOP: - loop_stack[loop_stack_pos] = i; - loop_stack_pos++; - break; - case OPCODE_ENDLOOP: - loop_stack_pos--; - /* Rewrite any breaks/conts at this nesting level (haven't - * already had a BranchTarget assigned) to point to the end - * of the loop. - */ - for (j = loop_stack[loop_stack_pos]; j < i; j++) { - if (mesa_instructions[j].Opcode == OPCODE_BRK || - mesa_instructions[j].Opcode == OPCODE_CONT) { - if (mesa_instructions[j].BranchTarget == -1) { - mesa_instructions[j].BranchTarget = i; - } - } - } - /* The loop ends point at each other. */ - mesa_instructions[i].BranchTarget = loop_stack[loop_stack_pos]; - mesa_instructions[loop_stack[loop_stack_pos]].BranchTarget = i; - break; - case OPCODE_CAL: - foreach_iter(exec_list_iterator, iter, v->function_signatures) { - function_entry *entry = (function_entry *)iter.get(); - - if (entry->sig_id == mesa_instructions[i].BranchTarget) { - mesa_instructions[i].BranchTarget = entry->inst; - break; - } - } - break; - default: - break; - } - } -} - -static void -print_program(struct prog_instruction *mesa_instructions, - ir_instruction **mesa_instruction_annotation, - int num_instructions) -{ - ir_instruction *last_ir = NULL; - int i; - int indent = 0; - - for (i = 0; i < num_instructions; i++) { - struct prog_instruction *mesa_inst = mesa_instructions + i; - ir_instruction *ir = mesa_instruction_annotation[i]; - - fprintf(stdout, "%3d: ", i); - - if (last_ir != ir && ir) { - int j; - - for (j = 0; j < indent; j++) { - fprintf(stdout, " "); - } - ir->print(); - printf("\n"); - last_ir = ir; - - fprintf(stdout, " "); /* line number spacing. */ - } - - indent = _mesa_fprint_instruction_opt(stdout, mesa_inst, indent, - PROG_PRINT_DEBUG, NULL); - } -} - -static void -count_resources(struct gl_program *prog) -{ - unsigned int i; - - prog->SamplersUsed = 0; - - for (i = 0; i < prog->NumInstructions; i++) { - struct prog_instruction *inst = &prog->Instructions[i]; - - if (_mesa_is_tex_instruction(inst->Opcode)) { - prog->SamplerTargets[inst->TexSrcUnit] = - (gl_texture_index)inst->TexSrcTarget; - prog->SamplersUsed |= 1 << inst->TexSrcUnit; - if (inst->TexShadow) { - prog->ShadowSamplers |= 1 << inst->TexSrcUnit; - } - } - } - - _mesa_update_shader_textures_used(prog); -} - -struct uniform_sort { - struct gl_uniform *u; - int pos; -}; - -/* The shader_program->Uniforms list is almost sorted in increasing - * uniform->{Frag,Vert}Pos locations, but not quite when there are - * uniforms shared between targets. We need to add parameters in - * increasing order for the targets. - */ -static int -sort_uniforms(const void *a, const void *b) -{ - struct uniform_sort *u1 = (struct uniform_sort *)a; - struct uniform_sort *u2 = (struct uniform_sort *)b; - - return u1->pos - u2->pos; -} - -/* Add the uniforms to the parameters. The linker chose locations - * in our parameters lists (which weren't created yet), which the - * uniforms code will use to poke values into our parameters list - * when uniforms are updated. - */ -static void -add_uniforms_to_parameters_list(struct gl_shader_program *shader_program, - struct gl_shader *shader, - struct gl_program *prog) -{ - unsigned int i; - unsigned int next_sampler = 0, num_uniforms = 0; - struct uniform_sort *sorted_uniforms; - - sorted_uniforms = ralloc_array(NULL, struct uniform_sort, - shader_program->Uniforms->NumUniforms); - - for (i = 0; i < shader_program->Uniforms->NumUniforms; i++) { - struct gl_uniform *uniform = shader_program->Uniforms->Uniforms + i; - int parameter_index = -1; - - switch (shader->Type) { - case GL_VERTEX_SHADER: - parameter_index = uniform->VertPos; - break; - case GL_FRAGMENT_SHADER: - parameter_index = uniform->FragPos; - break; - case GL_GEOMETRY_SHADER: - parameter_index = uniform->GeomPos; - break; - } - - /* Only add uniforms used in our target. */ - if (parameter_index != -1) { - sorted_uniforms[num_uniforms].pos = parameter_index; - sorted_uniforms[num_uniforms].u = uniform; - num_uniforms++; - } - } - - qsort(sorted_uniforms, num_uniforms, sizeof(struct uniform_sort), - sort_uniforms); - - for (i = 0; i < num_uniforms; i++) { - struct gl_uniform *uniform = sorted_uniforms[i].u; - int parameter_index = sorted_uniforms[i].pos; - const glsl_type *type = uniform->Type; - unsigned int size; - - if (type->is_vector() || - type->is_scalar()) { - size = type->vector_elements; - } else { - size = type_size(type) * 4; - } - - gl_register_file file; - if (type->is_sampler() || - (type->is_array() && type->fields.array->is_sampler())) { - file = PROGRAM_SAMPLER; - } else { - file = PROGRAM_UNIFORM; - } - - GLint index = _mesa_lookup_parameter_index(prog->Parameters, -1, - uniform->Name); - - if (index < 0) { - index = _mesa_add_parameter(prog->Parameters, file, - uniform->Name, size, type->gl_type, - NULL, NULL, 0x0); - - /* Sampler uniform values are stored in prog->SamplerUnits, - * and the entry in that array is selected by this index we - * store in ParameterValues[]. - */ - if (file == PROGRAM_SAMPLER) { - for (unsigned int j = 0; j < size / 4; j++) - prog->Parameters->ParameterValues[index + j][0] = next_sampler++; - } - - /* The location chosen in the Parameters list here (returned - * from _mesa_add_uniform) has to match what the linker chose. - */ - if (index != parameter_index) { - fail_link(shader_program, "Allocation of uniform `%s' to target " - "failed (%d vs %d)\n", - uniform->Name, index, parameter_index); - } - } - } - - ralloc_free(sorted_uniforms); -} - -static void -set_uniform_initializer(struct gl_context *ctx, void *mem_ctx, - struct gl_shader_program *shader_program, - const char *name, const glsl_type *type, - ir_constant *val) -{ - if (type->is_record()) { - ir_constant *field_constant; - - field_constant = (ir_constant *)val->components.get_head(); - - for (unsigned int i = 0; i < type->length; i++) { - const glsl_type *field_type = type->fields.structure[i].type; - const char *field_name = ralloc_asprintf(mem_ctx, "%s.%s", name, - type->fields.structure[i].name); - set_uniform_initializer(ctx, mem_ctx, shader_program, field_name, - field_type, field_constant); - field_constant = (ir_constant *)field_constant->next; - } - return; - } - - int loc = _mesa_get_uniform_location(ctx, shader_program, name); - - if (loc == -1) { - fail_link(shader_program, - "Couldn't find uniform for initializer %s\n", name); - return; - } - - for (unsigned int i = 0; i < (type->is_array() ? type->length : 1); i++) { - ir_constant *element; - const glsl_type *element_type; - if (type->is_array()) { - element = val->array_elements[i]; - element_type = type->fields.array; - } else { - element = val; - element_type = type; - } - - void *values; - - if (element_type->base_type == GLSL_TYPE_BOOL) { - int *conv = ralloc_array(mem_ctx, int, element_type->components()); - for (unsigned int j = 0; j < element_type->components(); j++) { - conv[j] = element->value.b[j]; - } - values = (void *)conv; - element_type = glsl_type::get_instance(GLSL_TYPE_INT, - element_type->vector_elements, - 1); - } else { - values = &element->value; - } - - if (element_type->is_matrix()) { - _mesa_uniform_matrix(ctx, shader_program, - element_type->matrix_columns, - element_type->vector_elements, - loc, 1, GL_FALSE, (GLfloat *)values); - loc += element_type->matrix_columns; - } else { - _mesa_uniform(ctx, shader_program, loc, element_type->matrix_columns, - values, element_type->gl_type); - loc += type_size(element_type); - } - } -} - -static void -set_uniform_initializers(struct gl_context *ctx, - struct gl_shader_program *shader_program) -{ - void *mem_ctx = NULL; - - for (unsigned int i = 0; i < MESA_SHADER_TYPES; i++) { - struct gl_shader *shader = shader_program->_LinkedShaders[i]; - - if (shader == NULL) - continue; - - foreach_iter(exec_list_iterator, iter, *shader->ir) { - ir_instruction *ir = (ir_instruction *)iter.get(); - ir_variable *var = ir->as_variable(); - - if (!var || var->mode != ir_var_uniform || !var->constant_value) - continue; - - if (!mem_ctx) - mem_ctx = ralloc_context(NULL); - - set_uniform_initializer(ctx, mem_ctx, shader_program, var->name, - var->type, var->constant_value); - } - } - - ralloc_free(mem_ctx); -} - -/* - * On a basic block basis, tracks available PROGRAM_TEMPORARY register - * channels for copy propagation and updates following instructions to - * use the original versions. - * - * The ir_to_mesa_visitor lazily produces code assuming that this pass - * will occur. As an example, a TXP production before this pass: - * - * 0: MOV TEMP[1], INPUT[4].xyyy; - * 1: MOV TEMP[1].w, INPUT[4].wwww; - * 2: TXP TEMP[2], TEMP[1], texture[0], 2D; - * - * and after: - * - * 0: MOV TEMP[1], INPUT[4].xyyy; - * 1: MOV TEMP[1].w, INPUT[4].wwww; - * 2: TXP TEMP[2], INPUT[4].xyyw, texture[0], 2D; - * - * which allows for dead code elimination on TEMP[1]'s writes. - */ -void -ir_to_mesa_visitor::copy_propagate(void) -{ - ir_to_mesa_instruction **acp = rzalloc_array(mem_ctx, - ir_to_mesa_instruction *, - this->next_temp * 4); - - foreach_iter(exec_list_iterator, iter, this->instructions) { - ir_to_mesa_instruction *inst = (ir_to_mesa_instruction *)iter.get(); - - assert(inst->dst_reg.file != PROGRAM_TEMPORARY - || inst->dst_reg.index < this->next_temp); - - /* First, do any copy propagation possible into the src regs. */ - for (int r = 0; r < 3; r++) { - ir_to_mesa_instruction *first = NULL; - bool good = true; - int acp_base = inst->src_reg[r].index * 4; - - if (inst->src_reg[r].file != PROGRAM_TEMPORARY || - inst->src_reg[r].reladdr) - continue; - - /* See if we can find entries in the ACP consisting of MOVs - * from the same src register for all the swizzled channels - * of this src register reference. - */ - for (int i = 0; i < 4; i++) { - int src_chan = GET_SWZ(inst->src_reg[r].swizzle, i); - ir_to_mesa_instruction *copy_chan = acp[acp_base + src_chan]; - - if (!copy_chan) { - good = false; - break; - } - - if (!first) { - first = copy_chan; - } else { - if (first->src_reg[0].file != copy_chan->src_reg[0].file || - first->src_reg[0].index != copy_chan->src_reg[0].index) { - good = false; - break; - } - } - } - - if (good) { - /* We've now validated that we can copy-propagate to - * replace this src register reference. Do it. - */ - inst->src_reg[r].file = first->src_reg[0].file; - inst->src_reg[r].index = first->src_reg[0].index; - - int swizzle = 0; - for (int i = 0; i < 4; i++) { - int src_chan = GET_SWZ(inst->src_reg[r].swizzle, i); - ir_to_mesa_instruction *copy_inst = acp[acp_base + src_chan]; - swizzle |= (GET_SWZ(copy_inst->src_reg[0].swizzle, src_chan) << - (3 * i)); - } - inst->src_reg[r].swizzle = swizzle; - } - } - - switch (inst->op) { - case OPCODE_BGNLOOP: - case OPCODE_ENDLOOP: - case OPCODE_ELSE: - case OPCODE_ENDIF: - /* End of a basic block, clear the ACP entirely. */ - memset(acp, 0, sizeof(*acp) * this->next_temp * 4); - break; - - default: - /* Continuing the block, clear any written channels from - * the ACP. - */ - if (inst->dst_reg.file == PROGRAM_TEMPORARY && inst->dst_reg.reladdr) { - /* Any temporary might be written, so no copy propagation - * across this instruction. - */ - memset(acp, 0, sizeof(*acp) * this->next_temp * 4); - } else if (inst->dst_reg.file == PROGRAM_OUTPUT && - inst->dst_reg.reladdr) { - /* Any output might be written, so no copy propagation - * from outputs across this instruction. - */ - for (int r = 0; r < this->next_temp; r++) { - for (int c = 0; c < 4; c++) { - if (acp[4 * r + c]->src_reg[0].file == PROGRAM_OUTPUT) - acp[4 * r + c] = NULL; - } - } - } else if (inst->dst_reg.file == PROGRAM_TEMPORARY || - inst->dst_reg.file == PROGRAM_OUTPUT) { - /* Clear where it's used as dst. */ - if (inst->dst_reg.file == PROGRAM_TEMPORARY) { - for (int c = 0; c < 4; c++) { - if (inst->dst_reg.writemask & (1 << c)) { - acp[4 * inst->dst_reg.index + c] = NULL; - } - } - } - - /* Clear where it's used as src. */ - for (int r = 0; r < this->next_temp; r++) { - for (int c = 0; c < 4; c++) { - if (!acp[4 * r + c]) - continue; - - int src_chan = GET_SWZ(acp[4 * r + c]->src_reg[0].swizzle, c); - - if (acp[4 * r + c]->src_reg[0].file == inst->dst_reg.file && - acp[4 * r + c]->src_reg[0].index == inst->dst_reg.index && - inst->dst_reg.writemask & (1 << src_chan)) - { - acp[4 * r + c] = NULL; - } - } - } - } - break; - } - - /* If this is a copy, add it to the ACP. */ - if (inst->op == OPCODE_MOV && - inst->dst_reg.file == PROGRAM_TEMPORARY && - !inst->dst_reg.reladdr && - !inst->saturate && - !inst->src_reg[0].reladdr && - !inst->src_reg[0].negate) { - for (int i = 0; i < 4; i++) { - if (inst->dst_reg.writemask & (1 << i)) { - acp[4 * inst->dst_reg.index + i] = inst; - } - } - } - } - - ralloc_free(acp); -} - - -/** - * Convert a shader's GLSL IR into a Mesa gl_program. - */ -static struct gl_program * -get_mesa_program(struct gl_context *ctx, - struct gl_shader_program *shader_program, - struct gl_shader *shader) -{ - ir_to_mesa_visitor v; - struct prog_instruction *mesa_instructions, *mesa_inst; - ir_instruction **mesa_instruction_annotation; - int i; - struct gl_program *prog; - GLenum target; - const char *target_string; - GLboolean progress; - struct gl_shader_compiler_options *options = - &ctx->ShaderCompilerOptions[_mesa_shader_type_to_index(shader->Type)]; - - switch (shader->Type) { - case GL_VERTEX_SHADER: - target = GL_VERTEX_PROGRAM_ARB; - target_string = "vertex"; - break; - case GL_FRAGMENT_SHADER: - target = GL_FRAGMENT_PROGRAM_ARB; - target_string = "fragment"; - break; - case GL_GEOMETRY_SHADER: - target = GL_GEOMETRY_PROGRAM_NV; - target_string = "geometry"; - break; - default: - assert(!"should not be reached"); - return NULL; - } - - validate_ir_tree(shader->ir); - - prog = ctx->Driver.NewProgram(ctx, target, shader_program->Name); - if (!prog) - return NULL; - prog->Parameters = _mesa_new_parameter_list(); - prog->Varying = _mesa_new_parameter_list(); - prog->Attributes = _mesa_new_parameter_list(); - v.ctx = ctx; - v.prog = prog; - v.shader_program = shader_program; - v.options = options; - - add_uniforms_to_parameters_list(shader_program, shader, prog); - - /* Emit Mesa IR for main(). */ - visit_exec_list(shader->ir, &v); - v.ir_to_mesa_emit_op0(NULL, OPCODE_END); - - /* Now emit bodies for any functions that were used. */ - do { - progress = GL_FALSE; - - foreach_iter(exec_list_iterator, iter, v.function_signatures) { - function_entry *entry = (function_entry *)iter.get(); - - if (!entry->bgn_inst) { - v.current_function = entry; - - entry->bgn_inst = v.ir_to_mesa_emit_op0(NULL, OPCODE_BGNSUB); - entry->bgn_inst->function = entry; - - visit_exec_list(&entry->sig->body, &v); - - ir_to_mesa_instruction *last; - last = (ir_to_mesa_instruction *)v.instructions.get_tail(); - if (last->op != OPCODE_RET) - v.ir_to_mesa_emit_op0(NULL, OPCODE_RET); - - ir_to_mesa_instruction *end; - end = v.ir_to_mesa_emit_op0(NULL, OPCODE_ENDSUB); - end->function = entry; - - progress = GL_TRUE; - } - } - } while (progress); - - prog->NumTemporaries = v.next_temp; - - int num_instructions = 0; - foreach_iter(exec_list_iterator, iter, v.instructions) { - num_instructions++; - } - - mesa_instructions = - (struct prog_instruction *)calloc(num_instructions, - sizeof(*mesa_instructions)); - mesa_instruction_annotation = ralloc_array(v.mem_ctx, ir_instruction *, - num_instructions); - - v.copy_propagate(); - - /* Convert ir_mesa_instructions into prog_instructions. - */ - mesa_inst = mesa_instructions; - i = 0; - foreach_iter(exec_list_iterator, iter, v.instructions) { - const ir_to_mesa_instruction *inst = (ir_to_mesa_instruction *)iter.get(); - - mesa_inst->Opcode = inst->op; - mesa_inst->CondUpdate = inst->cond_update; - if (inst->saturate) - mesa_inst->SaturateMode = SATURATE_ZERO_ONE; - mesa_inst->DstReg.File = inst->dst_reg.file; - mesa_inst->DstReg.Index = inst->dst_reg.index; - mesa_inst->DstReg.CondMask = inst->dst_reg.cond_mask; - mesa_inst->DstReg.WriteMask = inst->dst_reg.writemask; - mesa_inst->DstReg.RelAddr = inst->dst_reg.reladdr != NULL; - mesa_inst->SrcReg[0] = mesa_src_reg_from_ir_src_reg(inst->src_reg[0]); - mesa_inst->SrcReg[1] = mesa_src_reg_from_ir_src_reg(inst->src_reg[1]); - mesa_inst->SrcReg[2] = mesa_src_reg_from_ir_src_reg(inst->src_reg[2]); - mesa_inst->TexSrcUnit = inst->sampler; - mesa_inst->TexSrcTarget = inst->tex_target; - mesa_inst->TexShadow = inst->tex_shadow; - mesa_instruction_annotation[i] = inst->ir; - - /* Set IndirectRegisterFiles. */ - if (mesa_inst->DstReg.RelAddr) - prog->IndirectRegisterFiles |= 1 << mesa_inst->DstReg.File; - - /* Update program's bitmask of indirectly accessed register files */ - for (unsigned src = 0; src < 3; src++) - if (mesa_inst->SrcReg[src].RelAddr) - prog->IndirectRegisterFiles |= 1 << mesa_inst->SrcReg[src].File; - - if (options->EmitNoIfs && mesa_inst->Opcode == OPCODE_IF) { - fail_link(shader_program, "Couldn't flatten if statement\n"); - } - - switch (mesa_inst->Opcode) { - case OPCODE_BGNSUB: - inst->function->inst = i; - mesa_inst->Comment = strdup(inst->function->sig->function_name()); - break; - case OPCODE_ENDSUB: - mesa_inst->Comment = strdup(inst->function->sig->function_name()); - break; - case OPCODE_CAL: - mesa_inst->BranchTarget = inst->function->sig_id; /* rewritten later */ - break; - case OPCODE_ARL: - prog->NumAddressRegs = 1; - break; - default: - break; - } - - mesa_inst++; - i++; - - if (!shader_program->LinkStatus) - break; - } - - if (!shader_program->LinkStatus) { - free(mesa_instructions); - _mesa_reference_program(ctx, &shader->Program, NULL); - return NULL; - } - - set_branchtargets(&v, mesa_instructions, num_instructions); - - if (ctx->Shader.Flags & GLSL_DUMP) { - printf("\n"); - printf("GLSL IR for linked %s program %d:\n", target_string, - shader_program->Name); - _mesa_print_ir(shader->ir, NULL); - printf("\n"); - printf("\n"); - printf("Mesa IR for linked %s program %d:\n", target_string, - shader_program->Name); - print_program(mesa_instructions, mesa_instruction_annotation, - num_instructions); - } - - prog->Instructions = mesa_instructions; - prog->NumInstructions = num_instructions; - - do_set_program_inouts(shader->ir, prog); - count_resources(prog); - - _mesa_reference_program(ctx, &shader->Program, prog); - - if ((ctx->Shader.Flags & GLSL_NO_OPT) == 0) { - _mesa_optimize_program(ctx, prog); - } - - return prog; -} - -extern "C" { - -/** - * Called via ctx->Driver.CompilerShader(). - * This is a no-op. - * XXX can we remove the ctx->Driver.CompileShader() hook? - */ -GLboolean -_mesa_ir_compile_shader(struct gl_context *ctx, struct gl_shader *shader) -{ - assert(shader->CompileStatus); - (void) ctx; - - return GL_TRUE; -} - - -/** - * Link a shader. - * Called via ctx->Driver.LinkShader() - * This actually involves converting GLSL IR into Mesa gl_programs with - * code lowering and other optimizations. - */ -GLboolean -_mesa_ir_link_shader(struct gl_context *ctx, struct gl_shader_program *prog) -{ - assert(prog->LinkStatus); - - for (unsigned i = 0; i < MESA_SHADER_TYPES; i++) { - if (prog->_LinkedShaders[i] == NULL) - continue; - - bool progress; - exec_list *ir = prog->_LinkedShaders[i]->ir; - const struct gl_shader_compiler_options *options = - &ctx->ShaderCompilerOptions[_mesa_shader_type_to_index(prog->_LinkedShaders[i]->Type)]; - - do { - progress = false; - - /* Lowering */ - do_mat_op_to_vec(ir); - lower_instructions(ir, (MOD_TO_FRACT | DIV_TO_MUL_RCP | EXP_TO_EXP2 - | LOG_TO_LOG2 - | ((options->EmitNoPow) ? POW_TO_EXP2 : 0))); - - progress = do_lower_jumps(ir, true, true, options->EmitNoMainReturn, options->EmitNoCont, options->EmitNoLoops) || progress; - - progress = do_common_optimization(ir, true, options->MaxUnrollIterations) || progress; - - progress = lower_quadop_vector(ir, true) || progress; - - if (options->EmitNoIfs) { - progress = lower_discard(ir) || progress; - progress = lower_if_to_cond_assign(ir) || progress; - } - - if (options->EmitNoNoise) - progress = lower_noise(ir) || progress; - - /* If there are forms of indirect addressing that the driver - * cannot handle, perform the lowering pass. - */ - if (options->EmitNoIndirectInput || options->EmitNoIndirectOutput - || options->EmitNoIndirectTemp || options->EmitNoIndirectUniform) - progress = - lower_variable_index_to_cond_assign(ir, - options->EmitNoIndirectInput, - options->EmitNoIndirectOutput, - options->EmitNoIndirectTemp, - options->EmitNoIndirectUniform) - || progress; - - progress = do_vec_index_to_cond_assign(ir) || progress; - } while (progress); - - validate_ir_tree(ir); - } - - for (unsigned i = 0; i < MESA_SHADER_TYPES; i++) { - struct gl_program *linked_prog; - - if (prog->_LinkedShaders[i] == NULL) - continue; - - linked_prog = get_mesa_program(ctx, prog, prog->_LinkedShaders[i]); - - if (linked_prog) { - bool ok = true; - - switch (prog->_LinkedShaders[i]->Type) { - case GL_VERTEX_SHADER: - _mesa_reference_vertprog(ctx, &prog->VertexProgram, - (struct gl_vertex_program *)linked_prog); - ok = ctx->Driver.ProgramStringNotify(ctx, GL_VERTEX_PROGRAM_ARB, - linked_prog); - break; - case GL_FRAGMENT_SHADER: - _mesa_reference_fragprog(ctx, &prog->FragmentProgram, - (struct gl_fragment_program *)linked_prog); - ok = ctx->Driver.ProgramStringNotify(ctx, GL_FRAGMENT_PROGRAM_ARB, - linked_prog); - break; - case GL_GEOMETRY_SHADER: - _mesa_reference_geomprog(ctx, &prog->GeometryProgram, - (struct gl_geometry_program *)linked_prog); - ok = ctx->Driver.ProgramStringNotify(ctx, GL_GEOMETRY_PROGRAM_NV, - linked_prog); - break; - } - if (!ok) { - return GL_FALSE; - } - } - - _mesa_reference_program(ctx, &linked_prog, NULL); - } - - return GL_TRUE; -} - - -/** - * Compile a GLSL shader. Called via glCompileShader(). - */ -void -_mesa_glsl_compile_shader(struct gl_context *ctx, struct gl_shader *shader) -{ - struct _mesa_glsl_parse_state *state = - new(shader) _mesa_glsl_parse_state(ctx, shader->Type, shader); - - const char *source = shader->Source; - /* Check if the user called glCompileShader without first calling - * glShaderSource. This should fail to compile, but not raise a GL_ERROR. - */ - if (source == NULL) { - shader->CompileStatus = GL_FALSE; - return; - } - - state->error = preprocess(state, &source, &state->info_log, - &ctx->Extensions, ctx->API); - - if (ctx->Shader.Flags & GLSL_DUMP) { - printf("GLSL source for shader %d:\n", shader->Name); - printf("%s\n", shader->Source); - } - - if (!state->error) { - _mesa_glsl_lexer_ctor(state, source); - _mesa_glsl_parse(state); - _mesa_glsl_lexer_dtor(state); - } - - ralloc_free(shader->ir); - shader->ir = new(shader) exec_list; - if (!state->error && !state->translation_unit.is_empty()) - _mesa_ast_to_hir(shader->ir, state); - - if (!state->error && !shader->ir->is_empty()) { - validate_ir_tree(shader->ir); - - /* Do some optimization at compile time to reduce shader IR size - * and reduce later work if the same shader is linked multiple times - */ - while (do_common_optimization(shader->ir, false, 32)) - ; - - validate_ir_tree(shader->ir); - } - - shader->symbols = state->symbols; - - shader->CompileStatus = !state->error; - shader->InfoLog = state->info_log; - shader->Version = state->language_version; - memcpy(shader->builtins_to_link, state->builtins_to_link, - sizeof(shader->builtins_to_link[0]) * state->num_builtins_to_link); - shader->num_builtins_to_link = state->num_builtins_to_link; - - if (ctx->Shader.Flags & GLSL_LOG) { - _mesa_write_shader_to_file(shader); - } - - if (ctx->Shader.Flags & GLSL_DUMP) { - if (shader->CompileStatus) { - printf("GLSL IR for shader %d:\n", shader->Name); - _mesa_print_ir(shader->ir, NULL); - printf("\n\n"); - } else { - printf("GLSL shader %d failed to compile.\n", shader->Name); - } - if (shader->InfoLog && shader->InfoLog[0] != 0) { - printf("GLSL shader %d info log:\n", shader->Name); - printf("%s\n", shader->InfoLog); - } - } - - /* Retain any live IR, but trash the rest. */ - reparent_ir(shader->ir, shader->ir); - - ralloc_free(state); - - if (shader->CompileStatus) { - if (!ctx->Driver.CompileShader(ctx, shader)) - shader->CompileStatus = GL_FALSE; - } -} - - -/** - * Link a GLSL shader program. Called via glLinkProgram(). - */ -void -_mesa_glsl_link_shader(struct gl_context *ctx, struct gl_shader_program *prog) -{ - unsigned int i; - - _mesa_clear_shader_program_data(ctx, prog); - - prog->LinkStatus = GL_TRUE; - - for (i = 0; i < prog->NumShaders; i++) { - if (!prog->Shaders[i]->CompileStatus) { - fail_link(prog, "linking with uncompiled shader"); - prog->LinkStatus = GL_FALSE; - } - } - - prog->Varying = _mesa_new_parameter_list(); - _mesa_reference_vertprog(ctx, &prog->VertexProgram, NULL); - _mesa_reference_fragprog(ctx, &prog->FragmentProgram, NULL); - _mesa_reference_geomprog(ctx, &prog->GeometryProgram, NULL); - - if (prog->LinkStatus) { - link_shaders(ctx, prog); - } - - if (prog->LinkStatus) { - if (!ctx->Driver.LinkShader(ctx, prog)) { - prog->LinkStatus = GL_FALSE; - } - } - - set_uniform_initializers(ctx, prog); - - if (ctx->Shader.Flags & GLSL_DUMP) { - if (!prog->LinkStatus) { - printf("GLSL shader program %d failed to link\n", prog->Name); - } - - if (prog->InfoLog && prog->InfoLog[0] != 0) { - printf("GLSL shader program %d info log:\n", prog->Name); - printf("%s\n", prog->InfoLog); - } - } -} - -} /* extern "C" */ +/* + * Copyright (C) 2005-2007 Brian Paul All Rights Reserved. + * Copyright (C) 2008 VMware, Inc. All Rights Reserved. + * Copyright © 2010 Intel Corporation + * + * 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 (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 NONINFRINGEMENT. IN NO EVENT SHALL + * THE AUTHORS OR COPYRIGHT HOLDERS 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. + */ + +/** + * \file ir_to_mesa.cpp + * + * Translate GLSL IR to Mesa's gl_program representation. + */ + +#include +#include "main/compiler.h" +#include "ir.h" +#include "ir_visitor.h" +#include "ir_print_visitor.h" +#include "ir_expression_flattening.h" +#include "glsl_types.h" +#include "glsl_parser_extras.h" +#include "../glsl/program.h" +#include "ir_optimization.h" +#include "ast.h" + +extern "C" { +#include "main/mtypes.h" +#include "main/shaderapi.h" +#include "main/shaderobj.h" +#include "main/uniforms.h" +#include "program/hash_table.h" +#include "program/prog_instruction.h" +#include "program/prog_optimize.h" +#include "program/prog_print.h" +#include "program/program.h" +#include "program/prog_uniform.h" +#include "program/prog_parameter.h" +#include "program/sampler.h" +} + +static int swizzle_for_size(int size); + +/** + * This struct is a corresponding struct to Mesa prog_src_register, with + * wider fields. + */ +typedef struct ir_to_mesa_src_reg { + ir_to_mesa_src_reg(int file, int index, const glsl_type *type) + { + this->file = (gl_register_file) file; + this->index = index; + if (type && (type->is_scalar() || type->is_vector() || type->is_matrix())) + this->swizzle = swizzle_for_size(type->vector_elements); + else + this->swizzle = SWIZZLE_XYZW; + this->negate = 0; + this->reladdr = NULL; + } + + ir_to_mesa_src_reg() + { + this->file = PROGRAM_UNDEFINED; + this->index = 0; + this->swizzle = 0; + this->negate = 0; + this->reladdr = NULL; + } + + gl_register_file file; /**< PROGRAM_* from Mesa */ + int index; /**< temporary index, VERT_ATTRIB_*, FRAG_ATTRIB_*, etc. */ + GLuint swizzle; /**< SWIZZLE_XYZWONEZERO swizzles from Mesa. */ + int negate; /**< NEGATE_XYZW mask from mesa */ + /** Register index should be offset by the integer in this reg. */ + ir_to_mesa_src_reg *reladdr; +} ir_to_mesa_src_reg; + +typedef struct ir_to_mesa_dst_reg { + int file; /**< PROGRAM_* from Mesa */ + int index; /**< temporary index, VERT_ATTRIB_*, FRAG_ATTRIB_*, etc. */ + int writemask; /**< Bitfield of WRITEMASK_[XYZW] */ + GLuint cond_mask:4; + /** Register index should be offset by the integer in this reg. */ + ir_to_mesa_src_reg *reladdr; +} ir_to_mesa_dst_reg; + +extern ir_to_mesa_src_reg ir_to_mesa_undef; + +class ir_to_mesa_instruction : public exec_node { +public: + /* Callers of this ralloc-based new need not call delete. It's + * easier to just ralloc_free 'ctx' (or any of its ancestors). */ + static void* operator new(size_t size, void *ctx) + { + void *node; + + node = rzalloc_size(ctx, size); + assert(node != NULL); + + return node; + } + + enum prog_opcode op; + ir_to_mesa_dst_reg dst_reg; + ir_to_mesa_src_reg src_reg[3]; + /** Pointer to the ir source this tree came from for debugging */ + ir_instruction *ir; + GLboolean cond_update; + bool saturate; + int sampler; /**< sampler index */ + int tex_target; /**< One of TEXTURE_*_INDEX */ + GLboolean tex_shadow; + + class function_entry *function; /* Set on OPCODE_CAL or OPCODE_BGNSUB */ +}; + +class variable_storage : public exec_node { +public: + variable_storage(ir_variable *var, gl_register_file file, int index) + : file(file), index(index), var(var) + { + /* empty */ + } + + gl_register_file file; + int index; + ir_variable *var; /* variable that maps to this, if any */ +}; + +class function_entry : public exec_node { +public: + ir_function_signature *sig; + + /** + * identifier of this function signature used by the program. + * + * At the point that Mesa instructions for function calls are + * generated, we don't know the address of the first instruction of + * the function body. So we make the BranchTarget that is called a + * small integer and rewrite them during set_branchtargets(). + */ + int sig_id; + + /** + * Pointer to first instruction of the function body. + * + * Set during function body emits after main() is processed. + */ + ir_to_mesa_instruction *bgn_inst; + + /** + * Index of the first instruction of the function body in actual + * Mesa IR. + * + * Set after convertion from ir_to_mesa_instruction to prog_instruction. + */ + int inst; + + /** Storage for the return value. */ + ir_to_mesa_src_reg return_reg; +}; + +class ir_to_mesa_visitor : public ir_visitor { +public: + ir_to_mesa_visitor(); + ~ir_to_mesa_visitor(); + + function_entry *current_function; + + struct gl_context *ctx; + struct gl_program *prog; + struct gl_shader_program *shader_program; + struct gl_shader_compiler_options *options; + + int next_temp; + + variable_storage *find_variable_storage(ir_variable *var); + + function_entry *get_function_signature(ir_function_signature *sig); + + ir_to_mesa_src_reg get_temp(const glsl_type *type); + void reladdr_to_temp(ir_instruction *ir, + ir_to_mesa_src_reg *reg, int *num_reladdr); + + struct ir_to_mesa_src_reg src_reg_for_float(float val); + + /** + * \name Visit methods + * + * As typical for the visitor pattern, there must be one \c visit method for + * each concrete subclass of \c ir_instruction. Virtual base classes within + * the hierarchy should not have \c visit methods. + */ + /*@{*/ + virtual void visit(ir_variable *); + virtual void visit(ir_loop *); + virtual void visit(ir_loop_jump *); + virtual void visit(ir_function_signature *); + virtual void visit(ir_function *); + virtual void visit(ir_expression *); + virtual void visit(ir_swizzle *); + virtual void visit(ir_dereference_variable *); + virtual void visit(ir_dereference_array *); + virtual void visit(ir_dereference_record *); + virtual void visit(ir_assignment *); + virtual void visit(ir_constant *); + virtual void visit(ir_call *); + virtual void visit(ir_return *); + virtual void visit(ir_discard *); + virtual void visit(ir_texture *); + virtual void visit(ir_if *); + /*@}*/ + + struct ir_to_mesa_src_reg result; + + /** List of variable_storage */ + exec_list variables; + + /** List of function_entry */ + exec_list function_signatures; + int next_signature_id; + + /** List of ir_to_mesa_instruction */ + exec_list instructions; + + ir_to_mesa_instruction *ir_to_mesa_emit_op0(ir_instruction *ir, + enum prog_opcode op); + + ir_to_mesa_instruction *ir_to_mesa_emit_op1(ir_instruction *ir, + enum prog_opcode op, + ir_to_mesa_dst_reg dst, + ir_to_mesa_src_reg src0); + + ir_to_mesa_instruction *ir_to_mesa_emit_op2(ir_instruction *ir, + enum prog_opcode op, + ir_to_mesa_dst_reg dst, + ir_to_mesa_src_reg src0, + ir_to_mesa_src_reg src1); + + ir_to_mesa_instruction *ir_to_mesa_emit_op3(ir_instruction *ir, + enum prog_opcode op, + ir_to_mesa_dst_reg dst, + ir_to_mesa_src_reg src0, + ir_to_mesa_src_reg src1, + ir_to_mesa_src_reg src2); + + /** + * Emit the correct dot-product instruction for the type of arguments + * + * \sa ir_to_mesa_emit_op2 + */ + void ir_to_mesa_emit_dp(ir_instruction *ir, + ir_to_mesa_dst_reg dst, + ir_to_mesa_src_reg src0, + ir_to_mesa_src_reg src1, + unsigned elements); + + void ir_to_mesa_emit_scalar_op1(ir_instruction *ir, + enum prog_opcode op, + ir_to_mesa_dst_reg dst, + ir_to_mesa_src_reg src0); + + void ir_to_mesa_emit_scalar_op2(ir_instruction *ir, + enum prog_opcode op, + ir_to_mesa_dst_reg dst, + ir_to_mesa_src_reg src0, + ir_to_mesa_src_reg src1); + + void emit_scs(ir_instruction *ir, enum prog_opcode op, + ir_to_mesa_dst_reg dst, + const ir_to_mesa_src_reg &src); + + GLboolean try_emit_mad(ir_expression *ir, + int mul_operand); + GLboolean try_emit_sat(ir_expression *ir); + + void emit_swz(ir_expression *ir); + + bool process_move_condition(ir_rvalue *ir); + + void copy_propagate(void); + + void *mem_ctx; +}; + +ir_to_mesa_src_reg ir_to_mesa_undef = ir_to_mesa_src_reg(PROGRAM_UNDEFINED, 0, NULL); + +ir_to_mesa_dst_reg ir_to_mesa_undef_dst = { + PROGRAM_UNDEFINED, 0, SWIZZLE_NOOP, COND_TR, NULL, +}; + +ir_to_mesa_dst_reg ir_to_mesa_address_reg = { + PROGRAM_ADDRESS, 0, WRITEMASK_X, COND_TR, NULL +}; + +static void +fail_link(struct gl_shader_program *prog, const char *fmt, ...) PRINTFLIKE(2, 3); + +static void +fail_link(struct gl_shader_program *prog, const char *fmt, ...) +{ + va_list args; + va_start(args, fmt); + ralloc_vasprintf_append(&prog->InfoLog, fmt, args); + va_end(args); + + prog->LinkStatus = GL_FALSE; +} + +static int +swizzle_for_size(int size) +{ + int size_swizzles[4] = { + MAKE_SWIZZLE4(SWIZZLE_X, SWIZZLE_X, SWIZZLE_X, SWIZZLE_X), + MAKE_SWIZZLE4(SWIZZLE_X, SWIZZLE_Y, SWIZZLE_Y, SWIZZLE_Y), + MAKE_SWIZZLE4(SWIZZLE_X, SWIZZLE_Y, SWIZZLE_Z, SWIZZLE_Z), + MAKE_SWIZZLE4(SWIZZLE_X, SWIZZLE_Y, SWIZZLE_Z, SWIZZLE_W), + }; + + assert((size >= 1) && (size <= 4)); + return size_swizzles[size - 1]; +} + +ir_to_mesa_instruction * +ir_to_mesa_visitor::ir_to_mesa_emit_op3(ir_instruction *ir, + enum prog_opcode op, + ir_to_mesa_dst_reg dst, + ir_to_mesa_src_reg src0, + ir_to_mesa_src_reg src1, + ir_to_mesa_src_reg src2) +{ + ir_to_mesa_instruction *inst = new(mem_ctx) ir_to_mesa_instruction(); + int num_reladdr = 0; + + /* If we have to do relative addressing, we want to load the ARL + * reg directly for one of the regs, and preload the other reladdr + * sources into temps. + */ + num_reladdr += dst.reladdr != NULL; + num_reladdr += src0.reladdr != NULL; + num_reladdr += src1.reladdr != NULL; + num_reladdr += src2.reladdr != NULL; + + reladdr_to_temp(ir, &src2, &num_reladdr); + reladdr_to_temp(ir, &src1, &num_reladdr); + reladdr_to_temp(ir, &src0, &num_reladdr); + + if (dst.reladdr) { + ir_to_mesa_emit_op1(ir, OPCODE_ARL, ir_to_mesa_address_reg, + *dst.reladdr); + + num_reladdr--; + } + assert(num_reladdr == 0); + + inst->op = op; + inst->dst_reg = dst; + inst->src_reg[0] = src0; + inst->src_reg[1] = src1; + inst->src_reg[2] = src2; + inst->ir = ir; + + inst->function = NULL; + + this->instructions.push_tail(inst); + + return inst; +} + + +ir_to_mesa_instruction * +ir_to_mesa_visitor::ir_to_mesa_emit_op2(ir_instruction *ir, + enum prog_opcode op, + ir_to_mesa_dst_reg dst, + ir_to_mesa_src_reg src0, + ir_to_mesa_src_reg src1) +{ + return ir_to_mesa_emit_op3(ir, op, dst, src0, src1, ir_to_mesa_undef); +} + +ir_to_mesa_instruction * +ir_to_mesa_visitor::ir_to_mesa_emit_op1(ir_instruction *ir, + enum prog_opcode op, + ir_to_mesa_dst_reg dst, + ir_to_mesa_src_reg src0) +{ + assert(dst.writemask != 0); + return ir_to_mesa_emit_op3(ir, op, dst, + src0, ir_to_mesa_undef, ir_to_mesa_undef); +} + +ir_to_mesa_instruction * +ir_to_mesa_visitor::ir_to_mesa_emit_op0(ir_instruction *ir, + enum prog_opcode op) +{ + return ir_to_mesa_emit_op3(ir, op, ir_to_mesa_undef_dst, + ir_to_mesa_undef, + ir_to_mesa_undef, + ir_to_mesa_undef); +} + +void +ir_to_mesa_visitor::ir_to_mesa_emit_dp(ir_instruction *ir, + ir_to_mesa_dst_reg dst, + ir_to_mesa_src_reg src0, + ir_to_mesa_src_reg src1, + unsigned elements) +{ + static const gl_inst_opcode dot_opcodes[] = { + OPCODE_DP2, OPCODE_DP3, OPCODE_DP4 + }; + + ir_to_mesa_emit_op3(ir, dot_opcodes[elements - 2], + dst, src0, src1, ir_to_mesa_undef); +} + +inline ir_to_mesa_dst_reg +ir_to_mesa_dst_reg_from_src(ir_to_mesa_src_reg reg) +{ + ir_to_mesa_dst_reg dst_reg; + + dst_reg.file = reg.file; + dst_reg.index = reg.index; + dst_reg.writemask = WRITEMASK_XYZW; + dst_reg.cond_mask = COND_TR; + dst_reg.reladdr = reg.reladdr; + + return dst_reg; +} + +inline ir_to_mesa_src_reg +ir_to_mesa_src_reg_from_dst(ir_to_mesa_dst_reg reg) +{ + return ir_to_mesa_src_reg(reg.file, reg.index, NULL); +} + +/** + * Emits Mesa scalar opcodes to produce unique answers across channels. + * + * Some Mesa opcodes are scalar-only, like ARB_fp/vp. The src X + * channel determines the result across all channels. So to do a vec4 + * of this operation, we want to emit a scalar per source channel used + * to produce dest channels. + */ +void +ir_to_mesa_visitor::ir_to_mesa_emit_scalar_op2(ir_instruction *ir, + enum prog_opcode op, + ir_to_mesa_dst_reg dst, + ir_to_mesa_src_reg orig_src0, + ir_to_mesa_src_reg orig_src1) +{ + int i, j; + int done_mask = ~dst.writemask; + + /* Mesa RCP is a scalar operation splatting results to all channels, + * like ARB_fp/vp. So emit as many RCPs as necessary to cover our + * dst channels. + */ + for (i = 0; i < 4; i++) { + GLuint this_mask = (1 << i); + ir_to_mesa_instruction *inst; + ir_to_mesa_src_reg src0 = orig_src0; + ir_to_mesa_src_reg src1 = orig_src1; + + if (done_mask & this_mask) + continue; + + GLuint src0_swiz = GET_SWZ(src0.swizzle, i); + GLuint src1_swiz = GET_SWZ(src1.swizzle, i); + for (j = i + 1; j < 4; j++) { + /* If there is another enabled component in the destination that is + * derived from the same inputs, generate its value on this pass as + * well. + */ + if (!(done_mask & (1 << j)) && + GET_SWZ(src0.swizzle, j) == src0_swiz && + GET_SWZ(src1.swizzle, j) == src1_swiz) { + this_mask |= (1 << j); + } + } + src0.swizzle = MAKE_SWIZZLE4(src0_swiz, src0_swiz, + src0_swiz, src0_swiz); + src1.swizzle = MAKE_SWIZZLE4(src1_swiz, src1_swiz, + src1_swiz, src1_swiz); + + inst = ir_to_mesa_emit_op2(ir, op, + dst, + src0, + src1); + inst->dst_reg.writemask = this_mask; + done_mask |= this_mask; + } +} + +void +ir_to_mesa_visitor::ir_to_mesa_emit_scalar_op1(ir_instruction *ir, + enum prog_opcode op, + ir_to_mesa_dst_reg dst, + ir_to_mesa_src_reg src0) +{ + ir_to_mesa_src_reg undef = ir_to_mesa_undef; + + undef.swizzle = SWIZZLE_XXXX; + + ir_to_mesa_emit_scalar_op2(ir, op, dst, src0, undef); +} + +/** + * Emit an OPCODE_SCS instruction + * + * The \c SCS opcode functions a bit differently than the other Mesa (or + * ARB_fragment_program) opcodes. Instead of splatting its result across all + * four components of the destination, it writes one value to the \c x + * component and another value to the \c y component. + * + * \param ir IR instruction being processed + * \param op Either \c OPCODE_SIN or \c OPCODE_COS depending on which + * value is desired. + * \param dst Destination register + * \param src Source register + */ +void +ir_to_mesa_visitor::emit_scs(ir_instruction *ir, enum prog_opcode op, + ir_to_mesa_dst_reg dst, + const ir_to_mesa_src_reg &src) +{ + /* Vertex programs cannot use the SCS opcode. + */ + if (this->prog->Target == GL_VERTEX_PROGRAM_ARB) { + ir_to_mesa_emit_scalar_op1(ir, op, dst, src); + return; + } + + const unsigned component = (op == OPCODE_SIN) ? 0 : 1; + const unsigned scs_mask = (1U << component); + int done_mask = ~dst.writemask; + ir_to_mesa_src_reg tmp; + + assert(op == OPCODE_SIN || op == OPCODE_COS); + + /* If there are compnents in the destination that differ from the component + * that will be written by the SCS instrution, we'll need a temporary. + */ + if (scs_mask != unsigned(dst.writemask)) { + tmp = get_temp(glsl_type::vec4_type); + } + + for (unsigned i = 0; i < 4; i++) { + unsigned this_mask = (1U << i); + ir_to_mesa_src_reg src0 = src; + + if ((done_mask & this_mask) != 0) + continue; + + /* The source swizzle specified which component of the source generates + * sine / cosine for the current component in the destination. The SCS + * instruction requires that this value be swizzle to the X component. + * Replace the current swizzle with a swizzle that puts the source in + * the X component. + */ + unsigned src0_swiz = GET_SWZ(src.swizzle, i); + + src0.swizzle = MAKE_SWIZZLE4(src0_swiz, src0_swiz, + src0_swiz, src0_swiz); + for (unsigned j = i + 1; j < 4; j++) { + /* If there is another enabled component in the destination that is + * derived from the same inputs, generate its value on this pass as + * well. + */ + if (!(done_mask & (1 << j)) && + GET_SWZ(src0.swizzle, j) == src0_swiz) { + this_mask |= (1 << j); + } + } + + if (this_mask != scs_mask) { + ir_to_mesa_instruction *inst; + ir_to_mesa_dst_reg tmp_dst = ir_to_mesa_dst_reg_from_src(tmp); + + /* Emit the SCS instruction. + */ + inst = ir_to_mesa_emit_op1(ir, OPCODE_SCS, tmp_dst, src0); + inst->dst_reg.writemask = scs_mask; + + /* Move the result of the SCS instruction to the desired location in + * the destination. + */ + tmp.swizzle = MAKE_SWIZZLE4(component, component, + component, component); + inst = ir_to_mesa_emit_op1(ir, OPCODE_SCS, dst, tmp); + inst->dst_reg.writemask = this_mask; + } else { + /* Emit the SCS instruction to write directly to the destination. + */ + ir_to_mesa_instruction *inst = + ir_to_mesa_emit_op1(ir, OPCODE_SCS, dst, src0); + inst->dst_reg.writemask = scs_mask; + } + + done_mask |= this_mask; + } +} + +struct ir_to_mesa_src_reg +ir_to_mesa_visitor::src_reg_for_float(float val) +{ + ir_to_mesa_src_reg src_reg(PROGRAM_CONSTANT, -1, NULL); + + src_reg.index = _mesa_add_unnamed_constant(this->prog->Parameters, + &val, 1, &src_reg.swizzle); + + return src_reg; +} + +static int +type_size(const struct glsl_type *type) +{ + unsigned int i; + int size; + + switch (type->base_type) { + case GLSL_TYPE_UINT: + case GLSL_TYPE_INT: + case GLSL_TYPE_FLOAT: + case GLSL_TYPE_BOOL: + if (type->is_matrix()) { + return type->matrix_columns; + } else { + /* Regardless of size of vector, it gets a vec4. This is bad + * packing for things like floats, but otherwise arrays become a + * mess. Hopefully a later pass over the code can pack scalars + * down if appropriate. + */ + return 1; + } + case GLSL_TYPE_ARRAY: + assert(type->length > 0); + return type_size(type->fields.array) * type->length; + case GLSL_TYPE_STRUCT: + size = 0; + for (i = 0; i < type->length; i++) { + size += type_size(type->fields.structure[i].type); + } + return size; + case GLSL_TYPE_SAMPLER: + /* Samplers take up one slot in UNIFORMS[], but they're baked in + * at link time. + */ + return 1; + default: + assert(0); + return 0; + } +} + +/** + * In the initial pass of codegen, we assign temporary numbers to + * intermediate results. (not SSA -- variable assignments will reuse + * storage). Actual register allocation for the Mesa VM occurs in a + * pass over the Mesa IR later. + */ +ir_to_mesa_src_reg +ir_to_mesa_visitor::get_temp(const glsl_type *type) +{ + ir_to_mesa_src_reg src_reg; + int swizzle[4]; + int i; + + src_reg.file = PROGRAM_TEMPORARY; + src_reg.index = next_temp; + src_reg.reladdr = NULL; + next_temp += type_size(type); + + if (type->is_array() || type->is_record()) { + src_reg.swizzle = SWIZZLE_NOOP; + } else { + for (i = 0; i < type->vector_elements; i++) + swizzle[i] = i; + for (; i < 4; i++) + swizzle[i] = type->vector_elements - 1; + src_reg.swizzle = MAKE_SWIZZLE4(swizzle[0], swizzle[1], + swizzle[2], swizzle[3]); + } + src_reg.negate = 0; + + return src_reg; +} + +variable_storage * +ir_to_mesa_visitor::find_variable_storage(ir_variable *var) +{ + + variable_storage *entry; + + foreach_iter(exec_list_iterator, iter, this->variables) { + entry = (variable_storage *)iter.get(); + + if (entry->var == var) + return entry; + } + + return NULL; +} + +void +ir_to_mesa_visitor::visit(ir_variable *ir) +{ + if (strcmp(ir->name, "gl_FragCoord") == 0) { + struct gl_fragment_program *fp = (struct gl_fragment_program *)this->prog; + + fp->OriginUpperLeft = ir->origin_upper_left; + fp->PixelCenterInteger = ir->pixel_center_integer; + + } else if (strcmp(ir->name, "gl_FragDepth") == 0) { + struct gl_fragment_program *fp = (struct gl_fragment_program *)this->prog; + switch (ir->depth_layout) { + case ir_depth_layout_none: + fp->FragDepthLayout = FRAG_DEPTH_LAYOUT_NONE; + break; + case ir_depth_layout_any: + fp->FragDepthLayout = FRAG_DEPTH_LAYOUT_ANY; + break; + case ir_depth_layout_greater: + fp->FragDepthLayout = FRAG_DEPTH_LAYOUT_GREATER; + break; + case ir_depth_layout_less: + fp->FragDepthLayout = FRAG_DEPTH_LAYOUT_LESS; + break; + case ir_depth_layout_unchanged: + fp->FragDepthLayout = FRAG_DEPTH_LAYOUT_UNCHANGED; + break; + default: + assert(0); + break; + } + } + + if (ir->mode == ir_var_uniform && strncmp(ir->name, "gl_", 3) == 0) { + unsigned int i; + const struct gl_builtin_uniform_desc *statevar; + + for (i = 0; _mesa_builtin_uniform_desc[i].name; i++) { + if (strcmp(ir->name, _mesa_builtin_uniform_desc[i].name) == 0) + break; + } + + if (!_mesa_builtin_uniform_desc[i].name) { + fail_link(this->shader_program, + "Failed to find builtin uniform `%s'\n", ir->name); + return; + } + + statevar = &_mesa_builtin_uniform_desc[i]; + + int array_count; + if (ir->type->is_array()) { + array_count = ir->type->length; + } else { + array_count = 1; + } + + /* Check if this statevar's setup in the STATE file exactly + * matches how we'll want to reference it as a + * struct/array/whatever. If not, then we need to move it into + * temporary storage and hope that it'll get copy-propagated + * out. + */ + for (i = 0; i < statevar->num_elements; i++) { + if (statevar->elements[i].swizzle != SWIZZLE_XYZW) { + break; + } + } + + struct variable_storage *storage; + ir_to_mesa_dst_reg dst; + if (i == statevar->num_elements) { + /* We'll set the index later. */ + storage = new(mem_ctx) variable_storage(ir, PROGRAM_STATE_VAR, -1); + this->variables.push_tail(storage); + + dst = ir_to_mesa_undef_dst; + } else { + storage = new(mem_ctx) variable_storage(ir, PROGRAM_TEMPORARY, + this->next_temp); + this->variables.push_tail(storage); + this->next_temp += type_size(ir->type); + + dst = ir_to_mesa_dst_reg_from_src(ir_to_mesa_src_reg(PROGRAM_TEMPORARY, + storage->index, + NULL)); + } + + + for (int a = 0; a < array_count; a++) { + for (unsigned int i = 0; i < statevar->num_elements; i++) { + struct gl_builtin_uniform_element *element = &statevar->elements[i]; + int tokens[STATE_LENGTH]; + + memcpy(tokens, element->tokens, sizeof(element->tokens)); + if (ir->type->is_array()) { + tokens[1] = a; + } + + int index = _mesa_add_state_reference(this->prog->Parameters, + (gl_state_index *)tokens); + + if (storage->file == PROGRAM_STATE_VAR) { + if (storage->index == -1) { + storage->index = index; + } else { + assert(index == + (int)(storage->index + a * statevar->num_elements + i)); + } + } else { + ir_to_mesa_src_reg src(PROGRAM_STATE_VAR, index, NULL); + src.swizzle = element->swizzle; + ir_to_mesa_emit_op1(ir, OPCODE_MOV, dst, src); + /* even a float takes up a whole vec4 reg in a struct/array. */ + dst.index++; + } + } + } + if (storage->file == PROGRAM_TEMPORARY && + dst.index != storage->index + type_size(ir->type)) { + fail_link(this->shader_program, + "failed to load builtin uniform `%s' (%d/%d regs loaded)\n", + ir->name, dst.index - storage->index, + type_size(ir->type)); + } + } +} + +void +ir_to_mesa_visitor::visit(ir_loop *ir) +{ + ir_dereference_variable *counter = NULL; + + if (ir->counter != NULL) + counter = new(ir) ir_dereference_variable(ir->counter); + + if (ir->from != NULL) { + assert(ir->counter != NULL); + + ir_assignment *a = new(ir) ir_assignment(counter, ir->from, NULL); + + a->accept(this); + delete a; + } + + ir_to_mesa_emit_op0(NULL, OPCODE_BGNLOOP); + + if (ir->to) { + ir_expression *e = + new(ir) ir_expression(ir->cmp, glsl_type::bool_type, + counter, ir->to); + ir_if *if_stmt = new(ir) ir_if(e); + + ir_loop_jump *brk = new(ir) ir_loop_jump(ir_loop_jump::jump_break); + + if_stmt->then_instructions.push_tail(brk); + + if_stmt->accept(this); + + delete if_stmt; + delete e; + delete brk; + } + + visit_exec_list(&ir->body_instructions, this); + + if (ir->increment) { + ir_expression *e = + new(ir) ir_expression(ir_binop_add, counter->type, + counter, ir->increment); + + ir_assignment *a = new(ir) ir_assignment(counter, e, NULL); + + a->accept(this); + delete a; + delete e; + } + + ir_to_mesa_emit_op0(NULL, OPCODE_ENDLOOP); +} + +void +ir_to_mesa_visitor::visit(ir_loop_jump *ir) +{ + switch (ir->mode) { + case ir_loop_jump::jump_break: + ir_to_mesa_emit_op0(NULL, OPCODE_BRK); + break; + case ir_loop_jump::jump_continue: + ir_to_mesa_emit_op0(NULL, OPCODE_CONT); + break; + } +} + + +void +ir_to_mesa_visitor::visit(ir_function_signature *ir) +{ + assert(0); + (void)ir; +} + +void +ir_to_mesa_visitor::visit(ir_function *ir) +{ + /* Ignore function bodies other than main() -- we shouldn't see calls to + * them since they should all be inlined before we get to ir_to_mesa. + */ + if (strcmp(ir->name, "main") == 0) { + const ir_function_signature *sig; + exec_list empty; + + sig = ir->matching_signature(&empty); + + assert(sig); + + foreach_iter(exec_list_iterator, iter, sig->body) { + ir_instruction *ir = (ir_instruction *)iter.get(); + + ir->accept(this); + } + } +} + +GLboolean +ir_to_mesa_visitor::try_emit_mad(ir_expression *ir, int mul_operand) +{ + int nonmul_operand = 1 - mul_operand; + ir_to_mesa_src_reg a, b, c; + + ir_expression *expr = ir->operands[mul_operand]->as_expression(); + if (!expr || expr->operation != ir_binop_mul) + return false; + + expr->operands[0]->accept(this); + a = this->result; + expr->operands[1]->accept(this); + b = this->result; + ir->operands[nonmul_operand]->accept(this); + c = this->result; + + this->result = get_temp(ir->type); + ir_to_mesa_emit_op3(ir, OPCODE_MAD, + ir_to_mesa_dst_reg_from_src(this->result), a, b, c); + + return true; +} + +GLboolean +ir_to_mesa_visitor::try_emit_sat(ir_expression *ir) +{ + /* Saturates were only introduced to vertex programs in + * NV_vertex_program3, so don't give them to drivers in the VP. + */ + if (this->prog->Target == GL_VERTEX_PROGRAM_ARB) + return false; + + ir_rvalue *sat_src = ir->as_rvalue_to_saturate(); + if (!sat_src) + return false; + + sat_src->accept(this); + ir_to_mesa_src_reg src = this->result; + + this->result = get_temp(ir->type); + ir_to_mesa_instruction *inst; + inst = ir_to_mesa_emit_op1(ir, OPCODE_MOV, + ir_to_mesa_dst_reg_from_src(this->result), + src); + inst->saturate = true; + + return true; +} + +void +ir_to_mesa_visitor::reladdr_to_temp(ir_instruction *ir, + ir_to_mesa_src_reg *reg, int *num_reladdr) +{ + if (!reg->reladdr) + return; + + ir_to_mesa_emit_op1(ir, OPCODE_ARL, ir_to_mesa_address_reg, *reg->reladdr); + + if (*num_reladdr != 1) { + ir_to_mesa_src_reg temp = get_temp(glsl_type::vec4_type); + + ir_to_mesa_emit_op1(ir, OPCODE_MOV, + ir_to_mesa_dst_reg_from_src(temp), *reg); + *reg = temp; + } + + (*num_reladdr)--; +} + +void +ir_to_mesa_visitor::emit_swz(ir_expression *ir) +{ + /* Assume that the vector operator is in a form compatible with OPCODE_SWZ. + * This means that each of the operands is either an immediate value of -1, + * 0, or 1, or is a component from one source register (possibly with + * negation). + */ + uint8_t components[4] = { 0 }; + bool negate[4] = { false }; + ir_variable *var = NULL; + + for (unsigned i = 0; i < ir->type->vector_elements; i++) { + ir_rvalue *op = ir->operands[i]; + + assert(op->type->is_scalar()); + + while (op != NULL) { + switch (op->ir_type) { + case ir_type_constant: { + + assert(op->type->is_scalar()); + + const ir_constant *const c = op->as_constant(); + if (c->is_one()) { + components[i] = SWIZZLE_ONE; + } else if (c->is_zero()) { + components[i] = SWIZZLE_ZERO; + } else if (c->is_negative_one()) { + components[i] = SWIZZLE_ONE; + negate[i] = true; + } else { + assert(!"SWZ constant must be 0.0 or 1.0."); + } + + op = NULL; + break; + } + + case ir_type_dereference_variable: { + ir_dereference_variable *const deref = + (ir_dereference_variable *) op; + + assert((var == NULL) || (deref->var == var)); + components[i] = SWIZZLE_X; + var = deref->var; + op = NULL; + break; + } + + case ir_type_expression: { + ir_expression *const expr = (ir_expression *) op; + + assert(expr->operation == ir_unop_neg); + negate[i] = true; + + op = expr->operands[0]; + break; + } + + case ir_type_swizzle: { + ir_swizzle *const swiz = (ir_swizzle *) op; + + components[i] = swiz->mask.x; + op = swiz->val; + break; + } + + default: + assert(!"Should not get here."); + return; + } + } + } + + assert(var != NULL); + + ir_dereference_variable *const deref = + new(mem_ctx) ir_dereference_variable(var); + + this->result.file = PROGRAM_UNDEFINED; + deref->accept(this); + if (this->result.file == PROGRAM_UNDEFINED) { + ir_print_visitor v; + printf("Failed to get tree for expression operand:\n"); + deref->accept(&v); + exit(1); + } + + ir_to_mesa_src_reg src; + + src = this->result; + src.swizzle = MAKE_SWIZZLE4(components[0], + components[1], + components[2], + components[3]); + src.negate = ((unsigned(negate[0]) << 0) + | (unsigned(negate[1]) << 1) + | (unsigned(negate[2]) << 2) + | (unsigned(negate[3]) << 3)); + + /* Storage for our result. Ideally for an assignment we'd be using the + * actual storage for the result here, instead. + */ + const ir_to_mesa_src_reg result_src = get_temp(ir->type); + ir_to_mesa_dst_reg result_dst = ir_to_mesa_dst_reg_from_src(result_src); + + /* Limit writes to the channels that will be used by result_src later. + * This does limit this temp's use as a temporary for multi-instruction + * sequences. + */ + result_dst.writemask = (1 << ir->type->vector_elements) - 1; + + ir_to_mesa_emit_op1(ir, OPCODE_SWZ, result_dst, src); + this->result = result_src; +} + +void +ir_to_mesa_visitor::visit(ir_expression *ir) +{ + unsigned int operand; + struct ir_to_mesa_src_reg op[Elements(ir->operands)]; + struct ir_to_mesa_src_reg result_src; + struct ir_to_mesa_dst_reg result_dst; + + /* Quick peephole: Emit OPCODE_MAD(a, b, c) instead of ADD(MUL(a, b), c) + */ + if (ir->operation == ir_binop_add) { + if (try_emit_mad(ir, 1)) + return; + if (try_emit_mad(ir, 0)) + return; + } + if (try_emit_sat(ir)) + return; + + if (ir->operation == ir_quadop_vector) { + this->emit_swz(ir); + return; + } + + for (operand = 0; operand < ir->get_num_operands(); operand++) { + this->result.file = PROGRAM_UNDEFINED; + ir->operands[operand]->accept(this); + if (this->result.file == PROGRAM_UNDEFINED) { + ir_print_visitor v; + printf("Failed to get tree for expression operand:\n"); + ir->operands[operand]->accept(&v); + exit(1); + } + op[operand] = this->result; + + /* Matrix expression operands should have been broken down to vector + * operations already. + */ + assert(!ir->operands[operand]->type->is_matrix()); + } + + int vector_elements = ir->operands[0]->type->vector_elements; + if (ir->operands[1]) { + vector_elements = MAX2(vector_elements, + ir->operands[1]->type->vector_elements); + } + + this->result.file = PROGRAM_UNDEFINED; + + /* Storage for our result. Ideally for an assignment we'd be using + * the actual storage for the result here, instead. + */ + result_src = get_temp(ir->type); + /* convenience for the emit functions below. */ + result_dst = ir_to_mesa_dst_reg_from_src(result_src); + /* Limit writes to the channels that will be used by result_src later. + * This does limit this temp's use as a temporary for multi-instruction + * sequences. + */ + result_dst.writemask = (1 << ir->type->vector_elements) - 1; + + switch (ir->operation) { + case ir_unop_logic_not: + ir_to_mesa_emit_op2(ir, OPCODE_SEQ, result_dst, + op[0], src_reg_for_float(0.0)); + break; + case ir_unop_neg: + op[0].negate = ~op[0].negate; + result_src = op[0]; + break; + case ir_unop_abs: + ir_to_mesa_emit_op1(ir, OPCODE_ABS, result_dst, op[0]); + break; + case ir_unop_sign: + ir_to_mesa_emit_op1(ir, OPCODE_SSG, result_dst, op[0]); + break; + case ir_unop_rcp: + ir_to_mesa_emit_scalar_op1(ir, OPCODE_RCP, result_dst, op[0]); + break; + + case ir_unop_exp2: + ir_to_mesa_emit_scalar_op1(ir, OPCODE_EX2, result_dst, op[0]); + break; + case ir_unop_exp: + case ir_unop_log: + assert(!"not reached: should be handled by ir_explog_to_explog2"); + break; + case ir_unop_log2: + ir_to_mesa_emit_scalar_op1(ir, OPCODE_LG2, result_dst, op[0]); + break; + case ir_unop_sin: + ir_to_mesa_emit_scalar_op1(ir, OPCODE_SIN, result_dst, op[0]); + break; + case ir_unop_cos: + ir_to_mesa_emit_scalar_op1(ir, OPCODE_COS, result_dst, op[0]); + break; + case ir_unop_sin_reduced: + emit_scs(ir, OPCODE_SIN, result_dst, op[0]); + break; + case ir_unop_cos_reduced: + emit_scs(ir, OPCODE_COS, result_dst, op[0]); + break; + + case ir_unop_dFdx: + ir_to_mesa_emit_op1(ir, OPCODE_DDX, result_dst, op[0]); + break; + case ir_unop_dFdy: + ir_to_mesa_emit_op1(ir, OPCODE_DDY, result_dst, op[0]); + break; + + case ir_unop_noise: { + const enum prog_opcode opcode = + prog_opcode(OPCODE_NOISE1 + + (ir->operands[0]->type->vector_elements) - 1); + assert((opcode >= OPCODE_NOISE1) && (opcode <= OPCODE_NOISE4)); + + ir_to_mesa_emit_op1(ir, opcode, result_dst, op[0]); + break; + } + + case ir_binop_add: + ir_to_mesa_emit_op2(ir, OPCODE_ADD, result_dst, op[0], op[1]); + break; + case ir_binop_sub: + ir_to_mesa_emit_op2(ir, OPCODE_SUB, result_dst, op[0], op[1]); + break; + + case ir_binop_mul: + ir_to_mesa_emit_op2(ir, OPCODE_MUL, result_dst, op[0], op[1]); + break; + case ir_binop_div: + assert(!"not reached: should be handled by ir_div_to_mul_rcp"); + case ir_binop_mod: + assert(!"ir_binop_mod should have been converted to b * fract(a/b)"); + break; + + case ir_binop_less: + ir_to_mesa_emit_op2(ir, OPCODE_SLT, result_dst, op[0], op[1]); + break; + case ir_binop_greater: + ir_to_mesa_emit_op2(ir, OPCODE_SGT, result_dst, op[0], op[1]); + break; + case ir_binop_lequal: + ir_to_mesa_emit_op2(ir, OPCODE_SLE, result_dst, op[0], op[1]); + break; + case ir_binop_gequal: + ir_to_mesa_emit_op2(ir, OPCODE_SGE, result_dst, op[0], op[1]); + break; + case ir_binop_equal: + ir_to_mesa_emit_op2(ir, OPCODE_SEQ, result_dst, op[0], op[1]); + break; + case ir_binop_nequal: + ir_to_mesa_emit_op2(ir, OPCODE_SNE, result_dst, op[0], op[1]); + break; + case ir_binop_all_equal: + /* "==" operator producing a scalar boolean. */ + if (ir->operands[0]->type->is_vector() || + ir->operands[1]->type->is_vector()) { + ir_to_mesa_src_reg temp = get_temp(glsl_type::vec4_type); + ir_to_mesa_emit_op2(ir, OPCODE_SNE, + ir_to_mesa_dst_reg_from_src(temp), op[0], op[1]); + ir_to_mesa_emit_dp(ir, result_dst, temp, temp, vector_elements); + ir_to_mesa_emit_op2(ir, OPCODE_SEQ, + result_dst, result_src, src_reg_for_float(0.0)); + } else { + ir_to_mesa_emit_op2(ir, OPCODE_SEQ, result_dst, op[0], op[1]); + } + break; + case ir_binop_any_nequal: + /* "!=" operator producing a scalar boolean. */ + if (ir->operands[0]->type->is_vector() || + ir->operands[1]->type->is_vector()) { + ir_to_mesa_src_reg temp = get_temp(glsl_type::vec4_type); + ir_to_mesa_emit_op2(ir, OPCODE_SNE, + ir_to_mesa_dst_reg_from_src(temp), op[0], op[1]); + ir_to_mesa_emit_dp(ir, result_dst, temp, temp, vector_elements); + ir_to_mesa_emit_op2(ir, OPCODE_SNE, + result_dst, result_src, src_reg_for_float(0.0)); + } else { + ir_to_mesa_emit_op2(ir, OPCODE_SNE, result_dst, op[0], op[1]); + } + break; + + case ir_unop_any: + assert(ir->operands[0]->type->is_vector()); + ir_to_mesa_emit_dp(ir, result_dst, op[0], op[0], + ir->operands[0]->type->vector_elements); + ir_to_mesa_emit_op2(ir, OPCODE_SNE, + result_dst, result_src, src_reg_for_float(0.0)); + break; + + case ir_binop_logic_xor: + ir_to_mesa_emit_op2(ir, OPCODE_SNE, result_dst, op[0], op[1]); + break; + + case ir_binop_logic_or: + /* This could be a saturated add and skip the SNE. */ + ir_to_mesa_emit_op2(ir, OPCODE_ADD, + result_dst, + op[0], op[1]); + + ir_to_mesa_emit_op2(ir, OPCODE_SNE, + result_dst, + result_src, src_reg_for_float(0.0)); + break; + + case ir_binop_logic_and: + /* the bool args are stored as float 0.0 or 1.0, so "mul" gives us "and". */ + ir_to_mesa_emit_op2(ir, OPCODE_MUL, + result_dst, + op[0], op[1]); + break; + + case ir_binop_dot: + assert(ir->operands[0]->type->is_vector()); + assert(ir->operands[0]->type == ir->operands[1]->type); + ir_to_mesa_emit_dp(ir, result_dst, op[0], op[1], + ir->operands[0]->type->vector_elements); + break; + + case ir_unop_sqrt: + /* sqrt(x) = x * rsq(x). */ + ir_to_mesa_emit_scalar_op1(ir, OPCODE_RSQ, result_dst, op[0]); + ir_to_mesa_emit_op2(ir, OPCODE_MUL, result_dst, result_src, op[0]); + /* For incoming channels <= 0, set the result to 0. */ + op[0].negate = ~op[0].negate; + ir_to_mesa_emit_op3(ir, OPCODE_CMP, result_dst, + op[0], result_src, src_reg_for_float(0.0)); + break; + case ir_unop_rsq: + ir_to_mesa_emit_scalar_op1(ir, OPCODE_RSQ, result_dst, op[0]); + break; + case ir_unop_i2f: + case ir_unop_b2f: + case ir_unop_b2i: + /* Mesa IR lacks types, ints are stored as truncated floats. */ + result_src = op[0]; + break; + case ir_unop_f2i: + ir_to_mesa_emit_op1(ir, OPCODE_TRUNC, result_dst, op[0]); + break; + case ir_unop_f2b: + case ir_unop_i2b: + ir_to_mesa_emit_op2(ir, OPCODE_SNE, result_dst, + op[0], src_reg_for_float(0.0)); + break; + case ir_unop_trunc: + ir_to_mesa_emit_op1(ir, OPCODE_TRUNC, result_dst, op[0]); + break; + case ir_unop_ceil: + op[0].negate = ~op[0].negate; + ir_to_mesa_emit_op1(ir, OPCODE_FLR, result_dst, op[0]); + result_src.negate = ~result_src.negate; + break; + case ir_unop_floor: + ir_to_mesa_emit_op1(ir, OPCODE_FLR, result_dst, op[0]); + break; + case ir_unop_fract: + ir_to_mesa_emit_op1(ir, OPCODE_FRC, result_dst, op[0]); + break; + + case ir_binop_min: + ir_to_mesa_emit_op2(ir, OPCODE_MIN, result_dst, op[0], op[1]); + break; + case ir_binop_max: + ir_to_mesa_emit_op2(ir, OPCODE_MAX, result_dst, op[0], op[1]); + break; + case ir_binop_pow: + ir_to_mesa_emit_scalar_op2(ir, OPCODE_POW, result_dst, op[0], op[1]); + break; + + case ir_unop_bit_not: + case ir_unop_u2f: + case ir_binop_lshift: + case ir_binop_rshift: + case ir_binop_bit_and: + case ir_binop_bit_xor: + case ir_binop_bit_or: + case ir_unop_round_even: + assert(!"GLSL 1.30 features unsupported"); + break; + + case ir_quadop_vector: + /* This operation should have already been handled. + */ + assert(!"Should not get here."); + break; + } + + this->result = result_src; +} + + +void +ir_to_mesa_visitor::visit(ir_swizzle *ir) +{ + ir_to_mesa_src_reg src_reg; + int i; + int swizzle[4]; + + /* Note that this is only swizzles in expressions, not those on the left + * hand side of an assignment, which do write masking. See ir_assignment + * for that. + */ + + ir->val->accept(this); + src_reg = this->result; + assert(src_reg.file != PROGRAM_UNDEFINED); + + for (i = 0; i < 4; i++) { + if (i < ir->type->vector_elements) { + switch (i) { + case 0: + swizzle[i] = GET_SWZ(src_reg.swizzle, ir->mask.x); + break; + case 1: + swizzle[i] = GET_SWZ(src_reg.swizzle, ir->mask.y); + break; + case 2: + swizzle[i] = GET_SWZ(src_reg.swizzle, ir->mask.z); + break; + case 3: + swizzle[i] = GET_SWZ(src_reg.swizzle, ir->mask.w); + break; + } + } else { + /* If the type is smaller than a vec4, replicate the last + * channel out. + */ + swizzle[i] = swizzle[ir->type->vector_elements - 1]; + } + } + + src_reg.swizzle = MAKE_SWIZZLE4(swizzle[0], + swizzle[1], + swizzle[2], + swizzle[3]); + + this->result = src_reg; +} + +void +ir_to_mesa_visitor::visit(ir_dereference_variable *ir) +{ + variable_storage *entry = find_variable_storage(ir->var); + ir_variable *var = ir->var; + + if (!entry) { + switch (var->mode) { + case ir_var_uniform: + entry = new(mem_ctx) variable_storage(var, PROGRAM_UNIFORM, + var->location); + this->variables.push_tail(entry); + break; + case ir_var_in: + case ir_var_inout: + /* The linker assigns locations for varyings and attributes, + * including deprecated builtins (like gl_Color), user-assign + * generic attributes (glBindVertexLocation), and + * user-defined varyings. + * + * FINISHME: We would hit this path for function arguments. Fix! + */ + assert(var->location != -1); + entry = new(mem_ctx) variable_storage(var, + PROGRAM_INPUT, + var->location); + if (this->prog->Target == GL_VERTEX_PROGRAM_ARB && + var->location >= VERT_ATTRIB_GENERIC0) { + _mesa_add_attribute(this->prog->Attributes, + var->name, + _mesa_sizeof_glsl_type(var->type->gl_type), + var->type->gl_type, + var->location - VERT_ATTRIB_GENERIC0); + } + break; + case ir_var_out: + assert(var->location != -1); + entry = new(mem_ctx) variable_storage(var, + PROGRAM_OUTPUT, + var->location); + break; + case ir_var_system_value: + entry = new(mem_ctx) variable_storage(var, + PROGRAM_SYSTEM_VALUE, + var->location); + break; + case ir_var_auto: + case ir_var_temporary: + entry = new(mem_ctx) variable_storage(var, PROGRAM_TEMPORARY, + this->next_temp); + this->variables.push_tail(entry); + + next_temp += type_size(var->type); + break; + } + + if (!entry) { + printf("Failed to make storage for %s\n", var->name); + exit(1); + } + } + + this->result = ir_to_mesa_src_reg(entry->file, entry->index, var->type); +} + +void +ir_to_mesa_visitor::visit(ir_dereference_array *ir) +{ + ir_constant *index; + ir_to_mesa_src_reg src_reg; + int element_size = type_size(ir->type); + + index = ir->array_index->constant_expression_value(); + + ir->array->accept(this); + src_reg = this->result; + + if (index) { + src_reg.index += index->value.i[0] * element_size; + } else { + ir_to_mesa_src_reg array_base = this->result; + /* Variable index array dereference. It eats the "vec4" of the + * base of the array and an index that offsets the Mesa register + * index. + */ + ir->array_index->accept(this); + + ir_to_mesa_src_reg index_reg; + + if (element_size == 1) { + index_reg = this->result; + } else { + index_reg = get_temp(glsl_type::float_type); + + ir_to_mesa_emit_op2(ir, OPCODE_MUL, + ir_to_mesa_dst_reg_from_src(index_reg), + this->result, src_reg_for_float(element_size)); + } + + src_reg.reladdr = ralloc(mem_ctx, ir_to_mesa_src_reg); + memcpy(src_reg.reladdr, &index_reg, sizeof(index_reg)); + } + + /* If the type is smaller than a vec4, replicate the last channel out. */ + if (ir->type->is_scalar() || ir->type->is_vector()) + src_reg.swizzle = swizzle_for_size(ir->type->vector_elements); + else + src_reg.swizzle = SWIZZLE_NOOP; + + this->result = src_reg; +} + +void +ir_to_mesa_visitor::visit(ir_dereference_record *ir) +{ + unsigned int i; + const glsl_type *struct_type = ir->record->type; + int offset = 0; + + ir->record->accept(this); + + for (i = 0; i < struct_type->length; i++) { + if (strcmp(struct_type->fields.structure[i].name, ir->field) == 0) + break; + offset += type_size(struct_type->fields.structure[i].type); + } + + /* If the type is smaller than a vec4, replicate the last channel out. */ + if (ir->type->is_scalar() || ir->type->is_vector()) + this->result.swizzle = swizzle_for_size(ir->type->vector_elements); + else + this->result.swizzle = SWIZZLE_NOOP; + + this->result.index += offset; +} + +/** + * We want to be careful in assignment setup to hit the actual storage + * instead of potentially using a temporary like we might with the + * ir_dereference handler. + */ +static struct ir_to_mesa_dst_reg +get_assignment_lhs(ir_dereference *ir, ir_to_mesa_visitor *v) +{ + /* The LHS must be a dereference. If the LHS is a variable indexed array + * access of a vector, it must be separated into a series conditional moves + * before reaching this point (see ir_vec_index_to_cond_assign). + */ + assert(ir->as_dereference()); + ir_dereference_array *deref_array = ir->as_dereference_array(); + if (deref_array) { + assert(!deref_array->array->type->is_vector()); + } + + /* Use the rvalue deref handler for the most part. We'll ignore + * swizzles in it and write swizzles using writemask, though. + */ + ir->accept(v); + return ir_to_mesa_dst_reg_from_src(v->result); +} + +/** + * Process the condition of a conditional assignment + * + * Examines the condition of a conditional assignment to generate the optimal + * first operand of a \c CMP instruction. If the condition is a relational + * operator with 0 (e.g., \c ir_binop_less), the value being compared will be + * used as the source for the \c CMP instruction. Otherwise the comparison + * is processed to a boolean result, and the boolean result is used as the + * operand to the CMP instruction. + */ +bool +ir_to_mesa_visitor::process_move_condition(ir_rvalue *ir) +{ + ir_rvalue *src_ir = ir; + bool negate = true; + bool switch_order = false; + + ir_expression *const expr = ir->as_expression(); + if ((expr != NULL) && (expr->get_num_operands() == 2)) { + bool zero_on_left = false; + + if (expr->operands[0]->is_zero()) { + src_ir = expr->operands[1]; + zero_on_left = true; + } else if (expr->operands[1]->is_zero()) { + src_ir = expr->operands[0]; + zero_on_left = false; + } + + /* a is - 0 + - 0 + + * (a < 0) T F F ( a < 0) T F F + * (0 < a) F F T (-a < 0) F F T + * (a <= 0) T T F (-a < 0) F F T (swap order of other operands) + * (0 <= a) F T T ( a < 0) T F F (swap order of other operands) + * (a > 0) F F T (-a < 0) F F T + * (0 > a) T F F ( a < 0) T F F + * (a >= 0) F T T ( a < 0) T F F (swap order of other operands) + * (0 >= a) T T F (-a < 0) F F T (swap order of other operands) + * + * Note that exchanging the order of 0 and 'a' in the comparison simply + * means that the value of 'a' should be negated. + */ + if (src_ir != ir) { + switch (expr->operation) { + case ir_binop_less: + switch_order = false; + negate = zero_on_left; + break; + + case ir_binop_greater: + switch_order = false; + negate = !zero_on_left; + break; + + case ir_binop_lequal: + switch_order = true; + negate = !zero_on_left; + break; + + case ir_binop_gequal: + switch_order = true; + negate = zero_on_left; + break; + + default: + /* This isn't the right kind of comparison afterall, so make sure + * the whole condition is visited. + */ + src_ir = ir; + break; + } + } + } + + src_ir->accept(this); + + /* We use the OPCODE_CMP (a < 0 ? b : c) for conditional moves, and the + * condition we produced is 0.0 or 1.0. By flipping the sign, we can + * choose which value OPCODE_CMP produces without an extra instruction + * computing the condition. + */ + if (negate) + this->result.negate = ~this->result.negate; + + return switch_order; +} + +void +ir_to_mesa_visitor::visit(ir_assignment *ir) +{ + struct ir_to_mesa_dst_reg l; + struct ir_to_mesa_src_reg r; + int i; + + ir->rhs->accept(this); + r = this->result; + + l = get_assignment_lhs(ir->lhs, this); + + /* FINISHME: This should really set to the correct maximal writemask for each + * FINISHME: component written (in the loops below). This case can only + * FINISHME: occur for matrices, arrays, and structures. + */ + if (ir->write_mask == 0) { + assert(!ir->lhs->type->is_scalar() && !ir->lhs->type->is_vector()); + l.writemask = WRITEMASK_XYZW; + } else if (ir->lhs->type->is_scalar()) { + /* FINISHME: This hack makes writing to gl_FragDepth, which lives in the + * FINISHME: W component of fragment shader output zero, work correctly. + */ + l.writemask = WRITEMASK_XYZW; + } else { + int swizzles[4]; + int first_enabled_chan = 0; + int rhs_chan = 0; + + assert(ir->lhs->type->is_vector()); + l.writemask = ir->write_mask; + + for (int i = 0; i < 4; i++) { + if (l.writemask & (1 << i)) { + first_enabled_chan = GET_SWZ(r.swizzle, i); + break; + } + } + + /* Swizzle a small RHS vector into the channels being written. + * + * glsl ir treats write_mask as dictating how many channels are + * present on the RHS while Mesa IR treats write_mask as just + * showing which channels of the vec4 RHS get written. + */ + for (int i = 0; i < 4; i++) { + if (l.writemask & (1 << i)) + swizzles[i] = GET_SWZ(r.swizzle, rhs_chan++); + else + swizzles[i] = first_enabled_chan; + } + r.swizzle = MAKE_SWIZZLE4(swizzles[0], swizzles[1], + swizzles[2], swizzles[3]); + } + + assert(l.file != PROGRAM_UNDEFINED); + assert(r.file != PROGRAM_UNDEFINED); + + if (ir->condition) { + const bool switch_order = this->process_move_condition(ir->condition); + ir_to_mesa_src_reg condition = this->result; + + for (i = 0; i < type_size(ir->lhs->type); i++) { + if (switch_order) { + ir_to_mesa_emit_op3(ir, OPCODE_CMP, l, + condition, ir_to_mesa_src_reg_from_dst(l), r); + } else { + ir_to_mesa_emit_op3(ir, OPCODE_CMP, l, + condition, r, ir_to_mesa_src_reg_from_dst(l)); + } + + l.index++; + r.index++; + } + } else { + for (i = 0; i < type_size(ir->lhs->type); i++) { + ir_to_mesa_emit_op1(ir, OPCODE_MOV, l, r); + l.index++; + r.index++; + } + } +} + + +void +ir_to_mesa_visitor::visit(ir_constant *ir) +{ + ir_to_mesa_src_reg src_reg; + GLfloat stack_vals[4] = { 0 }; + GLfloat *values = stack_vals; + unsigned int i; + + /* Unfortunately, 4 floats is all we can get into + * _mesa_add_unnamed_constant. So, make a temp to store an + * aggregate constant and move each constant value into it. If we + * get lucky, copy propagation will eliminate the extra moves. + */ + + if (ir->type->base_type == GLSL_TYPE_STRUCT) { + ir_to_mesa_src_reg temp_base = get_temp(ir->type); + ir_to_mesa_dst_reg temp = ir_to_mesa_dst_reg_from_src(temp_base); + + foreach_iter(exec_list_iterator, iter, ir->components) { + ir_constant *field_value = (ir_constant *)iter.get(); + int size = type_size(field_value->type); + + assert(size > 0); + + field_value->accept(this); + src_reg = this->result; + + for (i = 0; i < (unsigned int)size; i++) { + ir_to_mesa_emit_op1(ir, OPCODE_MOV, temp, src_reg); + + src_reg.index++; + temp.index++; + } + } + this->result = temp_base; + return; + } + + if (ir->type->is_array()) { + ir_to_mesa_src_reg temp_base = get_temp(ir->type); + ir_to_mesa_dst_reg temp = ir_to_mesa_dst_reg_from_src(temp_base); + int size = type_size(ir->type->fields.array); + + assert(size > 0); + + for (i = 0; i < ir->type->length; i++) { + ir->array_elements[i]->accept(this); + src_reg = this->result; + for (int j = 0; j < size; j++) { + ir_to_mesa_emit_op1(ir, OPCODE_MOV, temp, src_reg); + + src_reg.index++; + temp.index++; + } + } + this->result = temp_base; + return; + } + + if (ir->type->is_matrix()) { + ir_to_mesa_src_reg mat = get_temp(ir->type); + ir_to_mesa_dst_reg mat_column = ir_to_mesa_dst_reg_from_src(mat); + + for (i = 0; i < ir->type->matrix_columns; i++) { + assert(ir->type->base_type == GLSL_TYPE_FLOAT); + values = &ir->value.f[i * ir->type->vector_elements]; + + src_reg = ir_to_mesa_src_reg(PROGRAM_CONSTANT, -1, NULL); + src_reg.index = _mesa_add_unnamed_constant(this->prog->Parameters, + values, + ir->type->vector_elements, + &src_reg.swizzle); + ir_to_mesa_emit_op1(ir, OPCODE_MOV, mat_column, src_reg); + + mat_column.index++; + } + + this->result = mat; + return; + } + + src_reg.file = PROGRAM_CONSTANT; + switch (ir->type->base_type) { + case GLSL_TYPE_FLOAT: + values = &ir->value.f[0]; + break; + case GLSL_TYPE_UINT: + for (i = 0; i < ir->type->vector_elements; i++) { + values[i] = ir->value.u[i]; + } + break; + case GLSL_TYPE_INT: + for (i = 0; i < ir->type->vector_elements; i++) { + values[i] = ir->value.i[i]; + } + break; + case GLSL_TYPE_BOOL: + for (i = 0; i < ir->type->vector_elements; i++) { + values[i] = ir->value.b[i]; + } + break; + default: + assert(!"Non-float/uint/int/bool constant"); + } + + this->result = ir_to_mesa_src_reg(PROGRAM_CONSTANT, -1, ir->type); + this->result.index = _mesa_add_unnamed_constant(this->prog->Parameters, + values, + ir->type->vector_elements, + &this->result.swizzle); +} + +function_entry * +ir_to_mesa_visitor::get_function_signature(ir_function_signature *sig) +{ + function_entry *entry; + + foreach_iter(exec_list_iterator, iter, this->function_signatures) { + entry = (function_entry *)iter.get(); + + if (entry->sig == sig) + return entry; + } + + entry = ralloc(mem_ctx, function_entry); + entry->sig = sig; + entry->sig_id = this->next_signature_id++; + entry->bgn_inst = NULL; + + /* Allocate storage for all the parameters. */ + foreach_iter(exec_list_iterator, iter, sig->parameters) { + ir_variable *param = (ir_variable *)iter.get(); + variable_storage *storage; + + storage = find_variable_storage(param); + assert(!storage); + + storage = new(mem_ctx) variable_storage(param, PROGRAM_TEMPORARY, + this->next_temp); + this->variables.push_tail(storage); + + this->next_temp += type_size(param->type); + } + + if (!sig->return_type->is_void()) { + entry->return_reg = get_temp(sig->return_type); + } else { + entry->return_reg = ir_to_mesa_undef; + } + + this->function_signatures.push_tail(entry); + return entry; +} + +void +ir_to_mesa_visitor::visit(ir_call *ir) +{ + ir_to_mesa_instruction *call_inst; + ir_function_signature *sig = ir->get_callee(); + function_entry *entry = get_function_signature(sig); + int i; + + /* Process in parameters. */ + exec_list_iterator sig_iter = sig->parameters.iterator(); + foreach_iter(exec_list_iterator, iter, *ir) { + ir_rvalue *param_rval = (ir_rvalue *)iter.get(); + ir_variable *param = (ir_variable *)sig_iter.get(); + + if (param->mode == ir_var_in || + param->mode == ir_var_inout) { + variable_storage *storage = find_variable_storage(param); + assert(storage); + + param_rval->accept(this); + ir_to_mesa_src_reg r = this->result; + + ir_to_mesa_dst_reg l; + l.file = storage->file; + l.index = storage->index; + l.reladdr = NULL; + l.writemask = WRITEMASK_XYZW; + l.cond_mask = COND_TR; + + for (i = 0; i < type_size(param->type); i++) { + ir_to_mesa_emit_op1(ir, OPCODE_MOV, l, r); + l.index++; + r.index++; + } + } + + sig_iter.next(); + } + assert(!sig_iter.has_next()); + + /* Emit call instruction */ + call_inst = ir_to_mesa_emit_op1(ir, OPCODE_CAL, + ir_to_mesa_undef_dst, ir_to_mesa_undef); + call_inst->function = entry; + + /* Process out parameters. */ + sig_iter = sig->parameters.iterator(); + foreach_iter(exec_list_iterator, iter, *ir) { + ir_rvalue *param_rval = (ir_rvalue *)iter.get(); + ir_variable *param = (ir_variable *)sig_iter.get(); + + if (param->mode == ir_var_out || + param->mode == ir_var_inout) { + variable_storage *storage = find_variable_storage(param); + assert(storage); + + ir_to_mesa_src_reg r; + r.file = storage->file; + r.index = storage->index; + r.reladdr = NULL; + r.swizzle = SWIZZLE_NOOP; + r.negate = 0; + + param_rval->accept(this); + ir_to_mesa_dst_reg l = ir_to_mesa_dst_reg_from_src(this->result); + + for (i = 0; i < type_size(param->type); i++) { + ir_to_mesa_emit_op1(ir, OPCODE_MOV, l, r); + l.index++; + r.index++; + } + } + + sig_iter.next(); + } + assert(!sig_iter.has_next()); + + /* Process return value. */ + this->result = entry->return_reg; +} + +void +ir_to_mesa_visitor::visit(ir_texture *ir) +{ + ir_to_mesa_src_reg result_src, coord, lod_info, projector; + ir_to_mesa_dst_reg result_dst, coord_dst; + ir_to_mesa_instruction *inst = NULL; + prog_opcode opcode = OPCODE_NOP; + + ir->coordinate->accept(this); + + /* Put our coords in a temp. We'll need to modify them for shadow, + * projection, or LOD, so the only case we'd use it as is is if + * we're doing plain old texturing. Mesa IR optimization should + * handle cleaning up our mess in that case. + */ + coord = get_temp(glsl_type::vec4_type); + coord_dst = ir_to_mesa_dst_reg_from_src(coord); + ir_to_mesa_emit_op1(ir, OPCODE_MOV, coord_dst, + this->result); + + if (ir->projector) { + ir->projector->accept(this); + projector = this->result; + } + + /* Storage for our result. Ideally for an assignment we'd be using + * the actual storage for the result here, instead. + */ + result_src = get_temp(glsl_type::vec4_type); + result_dst = ir_to_mesa_dst_reg_from_src(result_src); + + switch (ir->op) { + case ir_tex: + opcode = OPCODE_TEX; + break; + case ir_txb: + opcode = OPCODE_TXB; + ir->lod_info.bias->accept(this); + lod_info = this->result; + break; + case ir_txl: + opcode = OPCODE_TXL; + ir->lod_info.lod->accept(this); + lod_info = this->result; + break; + case ir_txd: + case ir_txf: + assert(!"GLSL 1.30 features unsupported"); + break; + } + + if (ir->projector) { + if (opcode == OPCODE_TEX) { + /* Slot the projector in as the last component of the coord. */ + coord_dst.writemask = WRITEMASK_W; + ir_to_mesa_emit_op1(ir, OPCODE_MOV, coord_dst, projector); + coord_dst.writemask = WRITEMASK_XYZW; + opcode = OPCODE_TXP; + } else { + ir_to_mesa_src_reg coord_w = coord; + coord_w.swizzle = SWIZZLE_WWWW; + + /* For the other TEX opcodes there's no projective version + * since the last slot is taken up by lod info. Do the + * projective divide now. + */ + coord_dst.writemask = WRITEMASK_W; + ir_to_mesa_emit_op1(ir, OPCODE_RCP, coord_dst, projector); + + coord_dst.writemask = WRITEMASK_XYZ; + ir_to_mesa_emit_op2(ir, OPCODE_MUL, coord_dst, coord, coord_w); + + coord_dst.writemask = WRITEMASK_XYZW; + coord.swizzle = SWIZZLE_XYZW; + } + } + + if (ir->shadow_comparitor) { + /* Slot the shadow value in as the second to last component of the + * coord. + */ + ir->shadow_comparitor->accept(this); + coord_dst.writemask = WRITEMASK_Z; + ir_to_mesa_emit_op1(ir, OPCODE_MOV, coord_dst, this->result); + coord_dst.writemask = WRITEMASK_XYZW; + } + + if (opcode == OPCODE_TXL || opcode == OPCODE_TXB) { + /* Mesa IR stores lod or lod bias in the last channel of the coords. */ + coord_dst.writemask = WRITEMASK_W; + ir_to_mesa_emit_op1(ir, OPCODE_MOV, coord_dst, lod_info); + coord_dst.writemask = WRITEMASK_XYZW; + } + + inst = ir_to_mesa_emit_op1(ir, opcode, result_dst, coord); + + if (ir->shadow_comparitor) + inst->tex_shadow = GL_TRUE; + + inst->sampler = _mesa_get_sampler_uniform_value(ir->sampler, + this->shader_program, + this->prog); + + const glsl_type *sampler_type = ir->sampler->type; + + switch (sampler_type->sampler_dimensionality) { + case GLSL_SAMPLER_DIM_1D: + inst->tex_target = (sampler_type->sampler_array) + ? TEXTURE_1D_ARRAY_INDEX : TEXTURE_1D_INDEX; + break; + case GLSL_SAMPLER_DIM_2D: + inst->tex_target = (sampler_type->sampler_array) + ? TEXTURE_2D_ARRAY_INDEX : TEXTURE_2D_INDEX; + break; + case GLSL_SAMPLER_DIM_3D: + inst->tex_target = TEXTURE_3D_INDEX; + break; + case GLSL_SAMPLER_DIM_CUBE: + inst->tex_target = TEXTURE_CUBE_INDEX; + break; + case GLSL_SAMPLER_DIM_RECT: + inst->tex_target = TEXTURE_RECT_INDEX; + break; + case GLSL_SAMPLER_DIM_BUF: + assert(!"FINISHME: Implement ARB_texture_buffer_object"); + break; + default: + assert(!"Should not get here."); + } + + this->result = result_src; +} + +void +ir_to_mesa_visitor::visit(ir_return *ir) +{ + if (ir->get_value()) { + ir_to_mesa_dst_reg l; + int i; + + assert(current_function); + + ir->get_value()->accept(this); + ir_to_mesa_src_reg r = this->result; + + l = ir_to_mesa_dst_reg_from_src(current_function->return_reg); + + for (i = 0; i < type_size(current_function->sig->return_type); i++) { + ir_to_mesa_emit_op1(ir, OPCODE_MOV, l, r); + l.index++; + r.index++; + } + } + + ir_to_mesa_emit_op0(ir, OPCODE_RET); +} + +void +ir_to_mesa_visitor::visit(ir_discard *ir) +{ + struct gl_fragment_program *fp = (struct gl_fragment_program *)this->prog; + + if (ir->condition) { + ir->condition->accept(this); + this->result.negate = ~this->result.negate; + ir_to_mesa_emit_op1(ir, OPCODE_KIL, ir_to_mesa_undef_dst, this->result); + } else { + ir_to_mesa_emit_op0(ir, OPCODE_KIL_NV); + } + + fp->UsesKill = GL_TRUE; +} + +void +ir_to_mesa_visitor::visit(ir_if *ir) +{ + ir_to_mesa_instruction *cond_inst, *if_inst, *else_inst = NULL; + ir_to_mesa_instruction *prev_inst; + + prev_inst = (ir_to_mesa_instruction *)this->instructions.get_tail(); + + ir->condition->accept(this); + assert(this->result.file != PROGRAM_UNDEFINED); + + if (this->options->EmitCondCodes) { + cond_inst = (ir_to_mesa_instruction *)this->instructions.get_tail(); + + /* See if we actually generated any instruction for generating + * the condition. If not, then cook up a move to a temp so we + * have something to set cond_update on. + */ + if (cond_inst == prev_inst) { + ir_to_mesa_src_reg temp = get_temp(glsl_type::bool_type); + cond_inst = ir_to_mesa_emit_op1(ir->condition, OPCODE_MOV, + ir_to_mesa_dst_reg_from_src(temp), + result); + } + cond_inst->cond_update = GL_TRUE; + + if_inst = ir_to_mesa_emit_op0(ir->condition, OPCODE_IF); + if_inst->dst_reg.cond_mask = COND_NE; + } else { + if_inst = ir_to_mesa_emit_op1(ir->condition, + OPCODE_IF, ir_to_mesa_undef_dst, + this->result); + } + + this->instructions.push_tail(if_inst); + + visit_exec_list(&ir->then_instructions, this); + + if (!ir->else_instructions.is_empty()) { + else_inst = ir_to_mesa_emit_op0(ir->condition, OPCODE_ELSE); + visit_exec_list(&ir->else_instructions, this); + } + + if_inst = ir_to_mesa_emit_op1(ir->condition, OPCODE_ENDIF, + ir_to_mesa_undef_dst, ir_to_mesa_undef); +} + +ir_to_mesa_visitor::ir_to_mesa_visitor() +{ + result.file = PROGRAM_UNDEFINED; + next_temp = 1; + next_signature_id = 1; + current_function = NULL; + mem_ctx = ralloc_context(NULL); +} + +ir_to_mesa_visitor::~ir_to_mesa_visitor() +{ + ralloc_free(mem_ctx); +} + +static struct prog_src_register +mesa_src_reg_from_ir_src_reg(ir_to_mesa_src_reg reg) +{ + struct prog_src_register mesa_reg; + + mesa_reg.File = reg.file; + assert(reg.index < (1 << INST_INDEX_BITS)); + mesa_reg.Index = reg.index; + mesa_reg.Swizzle = reg.swizzle; + mesa_reg.RelAddr = reg.reladdr != NULL; + mesa_reg.Negate = reg.negate; + mesa_reg.Abs = 0; + mesa_reg.HasIndex2 = GL_FALSE; + mesa_reg.RelAddr2 = 0; + mesa_reg.Index2 = 0; + + return mesa_reg; +} + +static void +set_branchtargets(ir_to_mesa_visitor *v, + struct prog_instruction *mesa_instructions, + int num_instructions) +{ + int if_count = 0, loop_count = 0; + int *if_stack, *loop_stack; + int if_stack_pos = 0, loop_stack_pos = 0; + int i, j; + + for (i = 0; i < num_instructions; i++) { + switch (mesa_instructions[i].Opcode) { + case OPCODE_IF: + if_count++; + break; + case OPCODE_BGNLOOP: + loop_count++; + break; + case OPCODE_BRK: + case OPCODE_CONT: + mesa_instructions[i].BranchTarget = -1; + break; + default: + break; + } + } + + if_stack = rzalloc_array(v->mem_ctx, int, if_count); + loop_stack = rzalloc_array(v->mem_ctx, int, loop_count); + + for (i = 0; i < num_instructions; i++) { + switch (mesa_instructions[i].Opcode) { + case OPCODE_IF: + if_stack[if_stack_pos] = i; + if_stack_pos++; + break; + case OPCODE_ELSE: + mesa_instructions[if_stack[if_stack_pos - 1]].BranchTarget = i; + if_stack[if_stack_pos - 1] = i; + break; + case OPCODE_ENDIF: + mesa_instructions[if_stack[if_stack_pos - 1]].BranchTarget = i; + if_stack_pos--; + break; + case OPCODE_BGNLOOP: + loop_stack[loop_stack_pos] = i; + loop_stack_pos++; + break; + case OPCODE_ENDLOOP: + loop_stack_pos--; + /* Rewrite any breaks/conts at this nesting level (haven't + * already had a BranchTarget assigned) to point to the end + * of the loop. + */ + for (j = loop_stack[loop_stack_pos]; j < i; j++) { + if (mesa_instructions[j].Opcode == OPCODE_BRK || + mesa_instructions[j].Opcode == OPCODE_CONT) { + if (mesa_instructions[j].BranchTarget == -1) { + mesa_instructions[j].BranchTarget = i; + } + } + } + /* The loop ends point at each other. */ + mesa_instructions[i].BranchTarget = loop_stack[loop_stack_pos]; + mesa_instructions[loop_stack[loop_stack_pos]].BranchTarget = i; + break; + case OPCODE_CAL: + foreach_iter(exec_list_iterator, iter, v->function_signatures) { + function_entry *entry = (function_entry *)iter.get(); + + if (entry->sig_id == mesa_instructions[i].BranchTarget) { + mesa_instructions[i].BranchTarget = entry->inst; + break; + } + } + break; + default: + break; + } + } +} + +static void +print_program(struct prog_instruction *mesa_instructions, + ir_instruction **mesa_instruction_annotation, + int num_instructions) +{ + ir_instruction *last_ir = NULL; + int i; + int indent = 0; + + for (i = 0; i < num_instructions; i++) { + struct prog_instruction *mesa_inst = mesa_instructions + i; + ir_instruction *ir = mesa_instruction_annotation[i]; + + fprintf(stdout, "%3d: ", i); + + if (last_ir != ir && ir) { + int j; + + for (j = 0; j < indent; j++) { + fprintf(stdout, " "); + } + ir->print(); + printf("\n"); + last_ir = ir; + + fprintf(stdout, " "); /* line number spacing. */ + } + + indent = _mesa_fprint_instruction_opt(stdout, mesa_inst, indent, + PROG_PRINT_DEBUG, NULL); + } +} + +static void +count_resources(struct gl_program *prog) +{ + unsigned int i; + + prog->SamplersUsed = 0; + + for (i = 0; i < prog->NumInstructions; i++) { + struct prog_instruction *inst = &prog->Instructions[i]; + + if (_mesa_is_tex_instruction(inst->Opcode)) { + prog->SamplerTargets[inst->TexSrcUnit] = + (gl_texture_index)inst->TexSrcTarget; + prog->SamplersUsed |= 1 << inst->TexSrcUnit; + if (inst->TexShadow) { + prog->ShadowSamplers |= 1 << inst->TexSrcUnit; + } + } + } + + _mesa_update_shader_textures_used(prog); +} + +struct uniform_sort { + struct gl_uniform *u; + int pos; +}; + +/* The shader_program->Uniforms list is almost sorted in increasing + * uniform->{Frag,Vert}Pos locations, but not quite when there are + * uniforms shared between targets. We need to add parameters in + * increasing order for the targets. + */ +static int +sort_uniforms(const void *a, const void *b) +{ + struct uniform_sort *u1 = (struct uniform_sort *)a; + struct uniform_sort *u2 = (struct uniform_sort *)b; + + return u1->pos - u2->pos; +} + +/* Add the uniforms to the parameters. The linker chose locations + * in our parameters lists (which weren't created yet), which the + * uniforms code will use to poke values into our parameters list + * when uniforms are updated. + */ +static void +add_uniforms_to_parameters_list(struct gl_shader_program *shader_program, + struct gl_shader *shader, + struct gl_program *prog) +{ + unsigned int i; + unsigned int next_sampler = 0, num_uniforms = 0; + struct uniform_sort *sorted_uniforms; + + sorted_uniforms = ralloc_array(NULL, struct uniform_sort, + shader_program->Uniforms->NumUniforms); + + for (i = 0; i < shader_program->Uniforms->NumUniforms; i++) { + struct gl_uniform *uniform = shader_program->Uniforms->Uniforms + i; + int parameter_index = -1; + + switch (shader->Type) { + case GL_VERTEX_SHADER: + parameter_index = uniform->VertPos; + break; + case GL_FRAGMENT_SHADER: + parameter_index = uniform->FragPos; + break; + case GL_GEOMETRY_SHADER: + parameter_index = uniform->GeomPos; + break; + } + + /* Only add uniforms used in our target. */ + if (parameter_index != -1) { + sorted_uniforms[num_uniforms].pos = parameter_index; + sorted_uniforms[num_uniforms].u = uniform; + num_uniforms++; + } + } + + qsort(sorted_uniforms, num_uniforms, sizeof(struct uniform_sort), + sort_uniforms); + + for (i = 0; i < num_uniforms; i++) { + struct gl_uniform *uniform = sorted_uniforms[i].u; + int parameter_index = sorted_uniforms[i].pos; + const glsl_type *type = uniform->Type; + unsigned int size; + + if (type->is_vector() || + type->is_scalar()) { + size = type->vector_elements; + } else { + size = type_size(type) * 4; + } + + gl_register_file file; + if (type->is_sampler() || + (type->is_array() && type->fields.array->is_sampler())) { + file = PROGRAM_SAMPLER; + } else { + file = PROGRAM_UNIFORM; + } + + GLint index = _mesa_lookup_parameter_index(prog->Parameters, -1, + uniform->Name); + + if (index < 0) { + index = _mesa_add_parameter(prog->Parameters, file, + uniform->Name, size, type->gl_type, + NULL, NULL, 0x0); + + /* Sampler uniform values are stored in prog->SamplerUnits, + * and the entry in that array is selected by this index we + * store in ParameterValues[]. + */ + if (file == PROGRAM_SAMPLER) { + for (unsigned int j = 0; j < size / 4; j++) + prog->Parameters->ParameterValues[index + j][0] = next_sampler++; + } + + /* The location chosen in the Parameters list here (returned + * from _mesa_add_uniform) has to match what the linker chose. + */ + if (index != parameter_index) { + fail_link(shader_program, "Allocation of uniform `%s' to target " + "failed (%d vs %d)\n", + uniform->Name, index, parameter_index); + } + } + } + + ralloc_free(sorted_uniforms); +} + +static void +set_uniform_initializer(struct gl_context *ctx, void *mem_ctx, + struct gl_shader_program *shader_program, + const char *name, const glsl_type *type, + ir_constant *val) +{ + if (type->is_record()) { + ir_constant *field_constant; + + field_constant = (ir_constant *)val->components.get_head(); + + for (unsigned int i = 0; i < type->length; i++) { + const glsl_type *field_type = type->fields.structure[i].type; + const char *field_name = ralloc_asprintf(mem_ctx, "%s.%s", name, + type->fields.structure[i].name); + set_uniform_initializer(ctx, mem_ctx, shader_program, field_name, + field_type, field_constant); + field_constant = (ir_constant *)field_constant->next; + } + return; + } + + int loc = _mesa_get_uniform_location(ctx, shader_program, name); + + if (loc == -1) { + fail_link(shader_program, + "Couldn't find uniform for initializer %s\n", name); + return; + } + + for (unsigned int i = 0; i < (type->is_array() ? type->length : 1); i++) { + ir_constant *element; + const glsl_type *element_type; + if (type->is_array()) { + element = val->array_elements[i]; + element_type = type->fields.array; + } else { + element = val; + element_type = type; + } + + void *values; + + if (element_type->base_type == GLSL_TYPE_BOOL) { + int *conv = ralloc_array(mem_ctx, int, element_type->components()); + for (unsigned int j = 0; j < element_type->components(); j++) { + conv[j] = element->value.b[j]; + } + values = (void *)conv; + element_type = glsl_type::get_instance(GLSL_TYPE_INT, + element_type->vector_elements, + 1); + } else { + values = &element->value; + } + + if (element_type->is_matrix()) { + _mesa_uniform_matrix(ctx, shader_program, + element_type->matrix_columns, + element_type->vector_elements, + loc, 1, GL_FALSE, (GLfloat *)values); + loc += element_type->matrix_columns; + } else { + _mesa_uniform(ctx, shader_program, loc, element_type->matrix_columns, + values, element_type->gl_type); + loc += type_size(element_type); + } + } +} + +static void +set_uniform_initializers(struct gl_context *ctx, + struct gl_shader_program *shader_program) +{ + void *mem_ctx = NULL; + + for (unsigned int i = 0; i < MESA_SHADER_TYPES; i++) { + struct gl_shader *shader = shader_program->_LinkedShaders[i]; + + if (shader == NULL) + continue; + + foreach_iter(exec_list_iterator, iter, *shader->ir) { + ir_instruction *ir = (ir_instruction *)iter.get(); + ir_variable *var = ir->as_variable(); + + if (!var || var->mode != ir_var_uniform || !var->constant_value) + continue; + + if (!mem_ctx) + mem_ctx = ralloc_context(NULL); + + set_uniform_initializer(ctx, mem_ctx, shader_program, var->name, + var->type, var->constant_value); + } + } + + ralloc_free(mem_ctx); +} + +/* + * On a basic block basis, tracks available PROGRAM_TEMPORARY register + * channels for copy propagation and updates following instructions to + * use the original versions. + * + * The ir_to_mesa_visitor lazily produces code assuming that this pass + * will occur. As an example, a TXP production before this pass: + * + * 0: MOV TEMP[1], INPUT[4].xyyy; + * 1: MOV TEMP[1].w, INPUT[4].wwww; + * 2: TXP TEMP[2], TEMP[1], texture[0], 2D; + * + * and after: + * + * 0: MOV TEMP[1], INPUT[4].xyyy; + * 1: MOV TEMP[1].w, INPUT[4].wwww; + * 2: TXP TEMP[2], INPUT[4].xyyw, texture[0], 2D; + * + * which allows for dead code elimination on TEMP[1]'s writes. + */ +void +ir_to_mesa_visitor::copy_propagate(void) +{ + ir_to_mesa_instruction **acp = rzalloc_array(mem_ctx, + ir_to_mesa_instruction *, + this->next_temp * 4); + + foreach_iter(exec_list_iterator, iter, this->instructions) { + ir_to_mesa_instruction *inst = (ir_to_mesa_instruction *)iter.get(); + + assert(inst->dst_reg.file != PROGRAM_TEMPORARY + || inst->dst_reg.index < this->next_temp); + + /* First, do any copy propagation possible into the src regs. */ + for (int r = 0; r < 3; r++) { + ir_to_mesa_instruction *first = NULL; + bool good = true; + int acp_base = inst->src_reg[r].index * 4; + + if (inst->src_reg[r].file != PROGRAM_TEMPORARY || + inst->src_reg[r].reladdr) + continue; + + /* See if we can find entries in the ACP consisting of MOVs + * from the same src register for all the swizzled channels + * of this src register reference. + */ + for (int i = 0; i < 4; i++) { + int src_chan = GET_SWZ(inst->src_reg[r].swizzle, i); + ir_to_mesa_instruction *copy_chan = acp[acp_base + src_chan]; + + if (!copy_chan) { + good = false; + break; + } + + if (!first) { + first = copy_chan; + } else { + if (first->src_reg[0].file != copy_chan->src_reg[0].file || + first->src_reg[0].index != copy_chan->src_reg[0].index) { + good = false; + break; + } + } + } + + if (good) { + /* We've now validated that we can copy-propagate to + * replace this src register reference. Do it. + */ + inst->src_reg[r].file = first->src_reg[0].file; + inst->src_reg[r].index = first->src_reg[0].index; + + int swizzle = 0; + for (int i = 0; i < 4; i++) { + int src_chan = GET_SWZ(inst->src_reg[r].swizzle, i); + ir_to_mesa_instruction *copy_inst = acp[acp_base + src_chan]; + swizzle |= (GET_SWZ(copy_inst->src_reg[0].swizzle, src_chan) << + (3 * i)); + } + inst->src_reg[r].swizzle = swizzle; + } + } + + switch (inst->op) { + case OPCODE_BGNLOOP: + case OPCODE_ENDLOOP: + case OPCODE_ELSE: + case OPCODE_ENDIF: + /* End of a basic block, clear the ACP entirely. */ + memset(acp, 0, sizeof(*acp) * this->next_temp * 4); + break; + + default: + /* Continuing the block, clear any written channels from + * the ACP. + */ + if (inst->dst_reg.file == PROGRAM_TEMPORARY && inst->dst_reg.reladdr) { + /* Any temporary might be written, so no copy propagation + * across this instruction. + */ + memset(acp, 0, sizeof(*acp) * this->next_temp * 4); + } else if (inst->dst_reg.file == PROGRAM_OUTPUT && + inst->dst_reg.reladdr) { + /* Any output might be written, so no copy propagation + * from outputs across this instruction. + */ + for (int r = 0; r < this->next_temp; r++) { + for (int c = 0; c < 4; c++) { + if (!acp[4 * r + c]) + continue; + + if (acp[4 * r + c]->src_reg[0].file == PROGRAM_OUTPUT) + acp[4 * r + c] = NULL; + } + } + } else if (inst->dst_reg.file == PROGRAM_TEMPORARY || + inst->dst_reg.file == PROGRAM_OUTPUT) { + /* Clear where it's used as dst. */ + if (inst->dst_reg.file == PROGRAM_TEMPORARY) { + for (int c = 0; c < 4; c++) { + if (inst->dst_reg.writemask & (1 << c)) { + acp[4 * inst->dst_reg.index + c] = NULL; + } + } + } + + /* Clear where it's used as src. */ + for (int r = 0; r < this->next_temp; r++) { + for (int c = 0; c < 4; c++) { + if (!acp[4 * r + c]) + continue; + + int src_chan = GET_SWZ(acp[4 * r + c]->src_reg[0].swizzle, c); + + if (acp[4 * r + c]->src_reg[0].file == inst->dst_reg.file && + acp[4 * r + c]->src_reg[0].index == inst->dst_reg.index && + inst->dst_reg.writemask & (1 << src_chan)) + { + acp[4 * r + c] = NULL; + } + } + } + } + break; + } + + /* If this is a copy, add it to the ACP. */ + if (inst->op == OPCODE_MOV && + inst->dst_reg.file == PROGRAM_TEMPORARY && + !inst->dst_reg.reladdr && + !inst->saturate && + !inst->src_reg[0].reladdr && + !inst->src_reg[0].negate) { + for (int i = 0; i < 4; i++) { + if (inst->dst_reg.writemask & (1 << i)) { + acp[4 * inst->dst_reg.index + i] = inst; + } + } + } + } + + ralloc_free(acp); +} + + +/** + * Convert a shader's GLSL IR into a Mesa gl_program. + */ +static struct gl_program * +get_mesa_program(struct gl_context *ctx, + struct gl_shader_program *shader_program, + struct gl_shader *shader) +{ + ir_to_mesa_visitor v; + struct prog_instruction *mesa_instructions, *mesa_inst; + ir_instruction **mesa_instruction_annotation; + int i; + struct gl_program *prog; + GLenum target; + const char *target_string; + GLboolean progress; + struct gl_shader_compiler_options *options = + &ctx->ShaderCompilerOptions[_mesa_shader_type_to_index(shader->Type)]; + + switch (shader->Type) { + case GL_VERTEX_SHADER: + target = GL_VERTEX_PROGRAM_ARB; + target_string = "vertex"; + break; + case GL_FRAGMENT_SHADER: + target = GL_FRAGMENT_PROGRAM_ARB; + target_string = "fragment"; + break; + case GL_GEOMETRY_SHADER: + target = GL_GEOMETRY_PROGRAM_NV; + target_string = "geometry"; + break; + default: + assert(!"should not be reached"); + return NULL; + } + + validate_ir_tree(shader->ir); + + prog = ctx->Driver.NewProgram(ctx, target, shader_program->Name); + if (!prog) + return NULL; + prog->Parameters = _mesa_new_parameter_list(); + prog->Varying = _mesa_new_parameter_list(); + prog->Attributes = _mesa_new_parameter_list(); + v.ctx = ctx; + v.prog = prog; + v.shader_program = shader_program; + v.options = options; + + add_uniforms_to_parameters_list(shader_program, shader, prog); + + /* Emit Mesa IR for main(). */ + visit_exec_list(shader->ir, &v); + v.ir_to_mesa_emit_op0(NULL, OPCODE_END); + + /* Now emit bodies for any functions that were used. */ + do { + progress = GL_FALSE; + + foreach_iter(exec_list_iterator, iter, v.function_signatures) { + function_entry *entry = (function_entry *)iter.get(); + + if (!entry->bgn_inst) { + v.current_function = entry; + + entry->bgn_inst = v.ir_to_mesa_emit_op0(NULL, OPCODE_BGNSUB); + entry->bgn_inst->function = entry; + + visit_exec_list(&entry->sig->body, &v); + + ir_to_mesa_instruction *last; + last = (ir_to_mesa_instruction *)v.instructions.get_tail(); + if (last->op != OPCODE_RET) + v.ir_to_mesa_emit_op0(NULL, OPCODE_RET); + + ir_to_mesa_instruction *end; + end = v.ir_to_mesa_emit_op0(NULL, OPCODE_ENDSUB); + end->function = entry; + + progress = GL_TRUE; + } + } + } while (progress); + + prog->NumTemporaries = v.next_temp; + + int num_instructions = 0; + foreach_iter(exec_list_iterator, iter, v.instructions) { + num_instructions++; + } + + mesa_instructions = + (struct prog_instruction *)calloc(num_instructions, + sizeof(*mesa_instructions)); + mesa_instruction_annotation = ralloc_array(v.mem_ctx, ir_instruction *, + num_instructions); + + v.copy_propagate(); + + /* Convert ir_mesa_instructions into prog_instructions. + */ + mesa_inst = mesa_instructions; + i = 0; + foreach_iter(exec_list_iterator, iter, v.instructions) { + const ir_to_mesa_instruction *inst = (ir_to_mesa_instruction *)iter.get(); + + mesa_inst->Opcode = inst->op; + mesa_inst->CondUpdate = inst->cond_update; + if (inst->saturate) + mesa_inst->SaturateMode = SATURATE_ZERO_ONE; + mesa_inst->DstReg.File = inst->dst_reg.file; + mesa_inst->DstReg.Index = inst->dst_reg.index; + mesa_inst->DstReg.CondMask = inst->dst_reg.cond_mask; + mesa_inst->DstReg.WriteMask = inst->dst_reg.writemask; + mesa_inst->DstReg.RelAddr = inst->dst_reg.reladdr != NULL; + mesa_inst->SrcReg[0] = mesa_src_reg_from_ir_src_reg(inst->src_reg[0]); + mesa_inst->SrcReg[1] = mesa_src_reg_from_ir_src_reg(inst->src_reg[1]); + mesa_inst->SrcReg[2] = mesa_src_reg_from_ir_src_reg(inst->src_reg[2]); + mesa_inst->TexSrcUnit = inst->sampler; + mesa_inst->TexSrcTarget = inst->tex_target; + mesa_inst->TexShadow = inst->tex_shadow; + mesa_instruction_annotation[i] = inst->ir; + + /* Set IndirectRegisterFiles. */ + if (mesa_inst->DstReg.RelAddr) + prog->IndirectRegisterFiles |= 1 << mesa_inst->DstReg.File; + + /* Update program's bitmask of indirectly accessed register files */ + for (unsigned src = 0; src < 3; src++) + if (mesa_inst->SrcReg[src].RelAddr) + prog->IndirectRegisterFiles |= 1 << mesa_inst->SrcReg[src].File; + + if (options->EmitNoIfs && mesa_inst->Opcode == OPCODE_IF) { + fail_link(shader_program, "Couldn't flatten if statement\n"); + } + + switch (mesa_inst->Opcode) { + case OPCODE_BGNSUB: + inst->function->inst = i; + mesa_inst->Comment = strdup(inst->function->sig->function_name()); + break; + case OPCODE_ENDSUB: + mesa_inst->Comment = strdup(inst->function->sig->function_name()); + break; + case OPCODE_CAL: + mesa_inst->BranchTarget = inst->function->sig_id; /* rewritten later */ + break; + case OPCODE_ARL: + prog->NumAddressRegs = 1; + break; + default: + break; + } + + mesa_inst++; + i++; + + if (!shader_program->LinkStatus) + break; + } + + if (!shader_program->LinkStatus) { + free(mesa_instructions); + _mesa_reference_program(ctx, &shader->Program, NULL); + return NULL; + } + + set_branchtargets(&v, mesa_instructions, num_instructions); + + if (ctx->Shader.Flags & GLSL_DUMP) { + printf("\n"); + printf("GLSL IR for linked %s program %d:\n", target_string, + shader_program->Name); + _mesa_print_ir(shader->ir, NULL); + printf("\n"); + printf("\n"); + printf("Mesa IR for linked %s program %d:\n", target_string, + shader_program->Name); + print_program(mesa_instructions, mesa_instruction_annotation, + num_instructions); + } + + prog->Instructions = mesa_instructions; + prog->NumInstructions = num_instructions; + + do_set_program_inouts(shader->ir, prog); + count_resources(prog); + + _mesa_reference_program(ctx, &shader->Program, prog); + + if ((ctx->Shader.Flags & GLSL_NO_OPT) == 0) { + _mesa_optimize_program(ctx, prog); + } + + return prog; +} + +extern "C" { + +/** + * Called via ctx->Driver.CompilerShader(). + * This is a no-op. + * XXX can we remove the ctx->Driver.CompileShader() hook? + */ +GLboolean +_mesa_ir_compile_shader(struct gl_context *ctx, struct gl_shader *shader) +{ + assert(shader->CompileStatus); + (void) ctx; + + return GL_TRUE; +} + + +/** + * Link a shader. + * Called via ctx->Driver.LinkShader() + * This actually involves converting GLSL IR into Mesa gl_programs with + * code lowering and other optimizations. + */ +GLboolean +_mesa_ir_link_shader(struct gl_context *ctx, struct gl_shader_program *prog) +{ + assert(prog->LinkStatus); + + for (unsigned i = 0; i < MESA_SHADER_TYPES; i++) { + if (prog->_LinkedShaders[i] == NULL) + continue; + + bool progress; + exec_list *ir = prog->_LinkedShaders[i]->ir; + const struct gl_shader_compiler_options *options = + &ctx->ShaderCompilerOptions[_mesa_shader_type_to_index(prog->_LinkedShaders[i]->Type)]; + + do { + progress = false; + + /* Lowering */ + do_mat_op_to_vec(ir); + lower_instructions(ir, (MOD_TO_FRACT | DIV_TO_MUL_RCP | EXP_TO_EXP2 + | LOG_TO_LOG2 + | ((options->EmitNoPow) ? POW_TO_EXP2 : 0))); + + progress = do_lower_jumps(ir, true, true, options->EmitNoMainReturn, options->EmitNoCont, options->EmitNoLoops) || progress; + + progress = do_common_optimization(ir, true, options->MaxUnrollIterations) || progress; + + progress = lower_quadop_vector(ir, true) || progress; + + if (options->EmitNoIfs) { + progress = lower_discard(ir) || progress; + progress = lower_if_to_cond_assign(ir) || progress; + } + + if (options->EmitNoNoise) + progress = lower_noise(ir) || progress; + + /* If there are forms of indirect addressing that the driver + * cannot handle, perform the lowering pass. + */ + if (options->EmitNoIndirectInput || options->EmitNoIndirectOutput + || options->EmitNoIndirectTemp || options->EmitNoIndirectUniform) + progress = + lower_variable_index_to_cond_assign(ir, + options->EmitNoIndirectInput, + options->EmitNoIndirectOutput, + options->EmitNoIndirectTemp, + options->EmitNoIndirectUniform) + || progress; + + progress = do_vec_index_to_cond_assign(ir) || progress; + } while (progress); + + validate_ir_tree(ir); + } + + for (unsigned i = 0; i < MESA_SHADER_TYPES; i++) { + struct gl_program *linked_prog; + + if (prog->_LinkedShaders[i] == NULL) + continue; + + linked_prog = get_mesa_program(ctx, prog, prog->_LinkedShaders[i]); + + if (linked_prog) { + bool ok = true; + + switch (prog->_LinkedShaders[i]->Type) { + case GL_VERTEX_SHADER: + _mesa_reference_vertprog(ctx, &prog->VertexProgram, + (struct gl_vertex_program *)linked_prog); + ok = ctx->Driver.ProgramStringNotify(ctx, GL_VERTEX_PROGRAM_ARB, + linked_prog); + break; + case GL_FRAGMENT_SHADER: + _mesa_reference_fragprog(ctx, &prog->FragmentProgram, + (struct gl_fragment_program *)linked_prog); + ok = ctx->Driver.ProgramStringNotify(ctx, GL_FRAGMENT_PROGRAM_ARB, + linked_prog); + break; + case GL_GEOMETRY_SHADER: + _mesa_reference_geomprog(ctx, &prog->GeometryProgram, + (struct gl_geometry_program *)linked_prog); + ok = ctx->Driver.ProgramStringNotify(ctx, GL_GEOMETRY_PROGRAM_NV, + linked_prog); + break; + } + if (!ok) { + return GL_FALSE; + } + } + + _mesa_reference_program(ctx, &linked_prog, NULL); + } + + return GL_TRUE; +} + + +/** + * Compile a GLSL shader. Called via glCompileShader(). + */ +void +_mesa_glsl_compile_shader(struct gl_context *ctx, struct gl_shader *shader) +{ + struct _mesa_glsl_parse_state *state = + new(shader) _mesa_glsl_parse_state(ctx, shader->Type, shader); + + const char *source = shader->Source; + /* Check if the user called glCompileShader without first calling + * glShaderSource. This should fail to compile, but not raise a GL_ERROR. + */ + if (source == NULL) { + shader->CompileStatus = GL_FALSE; + return; + } + + state->error = preprocess(state, &source, &state->info_log, + &ctx->Extensions, ctx->API); + + if (ctx->Shader.Flags & GLSL_DUMP) { + printf("GLSL source for shader %d:\n", shader->Name); + printf("%s\n", shader->Source); + } + + if (!state->error) { + _mesa_glsl_lexer_ctor(state, source); + _mesa_glsl_parse(state); + _mesa_glsl_lexer_dtor(state); + } + + ralloc_free(shader->ir); + shader->ir = new(shader) exec_list; + if (!state->error && !state->translation_unit.is_empty()) + _mesa_ast_to_hir(shader->ir, state); + + if (!state->error && !shader->ir->is_empty()) { + validate_ir_tree(shader->ir); + + /* Do some optimization at compile time to reduce shader IR size + * and reduce later work if the same shader is linked multiple times + */ + while (do_common_optimization(shader->ir, false, 32)) + ; + + validate_ir_tree(shader->ir); + } + + shader->symbols = state->symbols; + + shader->CompileStatus = !state->error; + shader->InfoLog = state->info_log; + shader->Version = state->language_version; + memcpy(shader->builtins_to_link, state->builtins_to_link, + sizeof(shader->builtins_to_link[0]) * state->num_builtins_to_link); + shader->num_builtins_to_link = state->num_builtins_to_link; + + if (ctx->Shader.Flags & GLSL_LOG) { + _mesa_write_shader_to_file(shader); + } + + if (ctx->Shader.Flags & GLSL_DUMP) { + if (shader->CompileStatus) { + printf("GLSL IR for shader %d:\n", shader->Name); + _mesa_print_ir(shader->ir, NULL); + printf("\n\n"); + } else { + printf("GLSL shader %d failed to compile.\n", shader->Name); + } + if (shader->InfoLog && shader->InfoLog[0] != 0) { + printf("GLSL shader %d info log:\n", shader->Name); + printf("%s\n", shader->InfoLog); + } + } + + /* Retain any live IR, but trash the rest. */ + reparent_ir(shader->ir, shader->ir); + + ralloc_free(state); + + if (shader->CompileStatus) { + if (!ctx->Driver.CompileShader(ctx, shader)) + shader->CompileStatus = GL_FALSE; + } +} + + +/** + * Link a GLSL shader program. Called via glLinkProgram(). + */ +void +_mesa_glsl_link_shader(struct gl_context *ctx, struct gl_shader_program *prog) +{ + unsigned int i; + + _mesa_clear_shader_program_data(ctx, prog); + + prog->LinkStatus = GL_TRUE; + + for (i = 0; i < prog->NumShaders; i++) { + if (!prog->Shaders[i]->CompileStatus) { + fail_link(prog, "linking with uncompiled shader"); + prog->LinkStatus = GL_FALSE; + } + } + + prog->Varying = _mesa_new_parameter_list(); + _mesa_reference_vertprog(ctx, &prog->VertexProgram, NULL); + _mesa_reference_fragprog(ctx, &prog->FragmentProgram, NULL); + _mesa_reference_geomprog(ctx, &prog->GeometryProgram, NULL); + + if (prog->LinkStatus) { + link_shaders(ctx, prog); + } + + if (prog->LinkStatus) { + if (!ctx->Driver.LinkShader(ctx, prog)) { + prog->LinkStatus = GL_FALSE; + } + } + + set_uniform_initializers(ctx, prog); + + if (ctx->Shader.Flags & GLSL_DUMP) { + if (!prog->LinkStatus) { + printf("GLSL shader program %d failed to link\n", prog->Name); + } + + if (prog->InfoLog && prog->InfoLog[0] != 0) { + printf("GLSL shader program %d info log:\n", prog->Name); + printf("%s\n", prog->InfoLog); + } + } +} + +} /* extern "C" */ diff --git a/mesalib/src/mesa/state_tracker/st_cb_fbo.c b/mesalib/src/mesa/state_tracker/st_cb_fbo.c index 398e32a34..2934f9ffb 100644 --- a/mesalib/src/mesa/state_tracker/st_cb_fbo.c +++ b/mesalib/src/mesa/state_tracker/st_cb_fbo.c @@ -1,620 +1,639 @@ -/************************************************************************** - * - * 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. - * - **************************************************************************/ - - -/** - * Framebuffer/renderbuffer functions. - * - * \author Brian Paul - */ - - -#include "main/imports.h" -#include "main/context.h" -#include "main/fbobject.h" -#include "main/framebuffer.h" -#include "main/macros.h" -#include "main/mfeatures.h" -#include "main/renderbuffer.h" - -#include "pipe/p_context.h" -#include "pipe/p_defines.h" -#include "pipe/p_screen.h" -#include "st_context.h" -#include "st_cb_fbo.h" -#include "st_cb_flush.h" -#include "st_format.h" -#include "st_texture.h" -#include "st_manager.h" - -#include "util/u_format.h" -#include "util/u_inlines.h" -#include "util/u_surface.h" - - -/** - * gl_renderbuffer::AllocStorage() - * This is called to allocate the original drawing surface, and - * during window resize. - */ -static GLboolean -st_renderbuffer_alloc_storage(struct gl_context * ctx, - struct gl_renderbuffer *rb, - GLenum internalFormat, - GLuint width, GLuint height) -{ - struct st_context *st = st_context(ctx); - struct pipe_context *pipe = st->pipe; - struct pipe_screen *screen = st->pipe->screen; - struct st_renderbuffer *strb = st_renderbuffer(rb); - enum pipe_format format; - struct pipe_surface surf_tmpl; - - if (strb->format != PIPE_FORMAT_NONE) - format = strb->format; - else - format = st_choose_renderbuffer_format(screen, internalFormat, - rb->NumSamples); - - /* init renderbuffer fields */ - strb->Base.Width = width; - strb->Base.Height = height; - strb->Base.Format = st_pipe_format_to_mesa_format(format); - strb->Base.DataType = st_format_datatype(format); - - strb->defined = GL_FALSE; /* undefined contents now */ - - if (strb->software) { - size_t size; - - free(strb->data); - - assert(strb->format != PIPE_FORMAT_NONE); - - strb->stride = util_format_get_stride(strb->format, width); - size = util_format_get_2d_size(strb->format, strb->stride, height); - - strb->data = malloc(size); - - return strb->data != NULL; - } - else { - struct pipe_resource template; - - /* Free the old surface and texture - */ - pipe_surface_reference( &strb->surface, NULL ); - pipe_resource_reference( &strb->texture, NULL ); - pipe_sampler_view_reference(&strb->sampler_view, NULL); - - /* Setup new texture template. - */ - memset(&template, 0, sizeof(template)); - template.target = st->internal_target; - template.format = format; - template.width0 = width; - template.height0 = height; - template.depth0 = 1; - template.array_size = 1; - template.last_level = 0; - template.nr_samples = rb->NumSamples; - if (util_format_is_depth_or_stencil(format)) { - template.bind = PIPE_BIND_DEPTH_STENCIL; - } - else { - template.bind = (PIPE_BIND_DISPLAY_TARGET | - PIPE_BIND_RENDER_TARGET); - } - - strb->texture = screen->resource_create(screen, &template); - - if (!strb->texture) - return FALSE; - - memset(&surf_tmpl, 0, sizeof(surf_tmpl)); - u_surface_default_template(&surf_tmpl, strb->texture, template.bind); - strb->surface = pipe->create_surface(pipe, - strb->texture, - &surf_tmpl); - if (strb->surface) { - assert(strb->surface->texture); - assert(strb->surface->format); - assert(strb->surface->width == width); - assert(strb->surface->height == height); - } - - return strb->surface != NULL; - } -} - - -/** - * gl_renderbuffer::Delete() - */ -static void -st_renderbuffer_delete(struct gl_renderbuffer *rb) -{ - struct st_renderbuffer *strb = st_renderbuffer(rb); - ASSERT(strb); - pipe_surface_reference(&strb->surface, NULL); - pipe_resource_reference(&strb->texture, NULL); - pipe_sampler_view_reference(&strb->sampler_view, NULL); - free(strb->data); - free(strb); -} - - -/** - * gl_renderbuffer::GetPointer() - */ -static void * -null_get_pointer(struct gl_context * ctx, struct gl_renderbuffer *rb, - GLint x, GLint y) -{ - /* By returning NULL we force all software rendering to go through - * the span routines. - */ -#if 0 - assert(0); /* Should never get called with softpipe */ -#endif - return NULL; -} - - -/** - * Called via ctx->Driver.NewFramebuffer() - */ -static struct gl_framebuffer * -st_new_framebuffer(struct gl_context *ctx, GLuint name) -{ - /* XXX not sure we need to subclass gl_framebuffer for pipe */ - return _mesa_new_framebuffer(ctx, name); -} - - -/** - * Called via ctx->Driver.NewRenderbuffer() - */ -static struct gl_renderbuffer * -st_new_renderbuffer(struct gl_context *ctx, GLuint name) -{ - struct st_renderbuffer *strb = ST_CALLOC_STRUCT(st_renderbuffer); - if (strb) { - _mesa_init_renderbuffer(&strb->Base, name); - strb->Base.Delete = st_renderbuffer_delete; - strb->Base.AllocStorage = st_renderbuffer_alloc_storage; - strb->Base.GetPointer = null_get_pointer; - strb->format = PIPE_FORMAT_NONE; - return &strb->Base; - } - return NULL; -} - - -/** - * Allocate a renderbuffer for a an on-screen window (not a user-created - * renderbuffer). The window system code determines the format. - */ -struct gl_renderbuffer * -st_new_renderbuffer_fb(enum pipe_format format, int samples, boolean sw) -{ - struct st_renderbuffer *strb; - - strb = ST_CALLOC_STRUCT(st_renderbuffer); - if (!strb) { - _mesa_error(NULL, GL_OUT_OF_MEMORY, "creating renderbuffer"); - return NULL; - } - - _mesa_init_renderbuffer(&strb->Base, 0); - strb->Base.ClassID = 0x4242; /* just a unique value */ - strb->Base.NumSamples = samples; - strb->Base.Format = st_pipe_format_to_mesa_format(format); - strb->Base.DataType = st_format_datatype(format); - strb->format = format; - strb->software = sw; - - switch (format) { - case PIPE_FORMAT_R8G8B8A8_UNORM: - case PIPE_FORMAT_B8G8R8A8_UNORM: - case PIPE_FORMAT_A8R8G8B8_UNORM: - case PIPE_FORMAT_R8G8B8X8_UNORM: - case PIPE_FORMAT_B8G8R8X8_UNORM: - case PIPE_FORMAT_X8R8G8B8_UNORM: - case PIPE_FORMAT_B5G5R5A1_UNORM: - case PIPE_FORMAT_B4G4R4A4_UNORM: - case PIPE_FORMAT_B5G6R5_UNORM: - strb->Base.InternalFormat = GL_RGBA; - break; - case PIPE_FORMAT_Z16_UNORM: - strb->Base.InternalFormat = GL_DEPTH_COMPONENT16; - break; - case PIPE_FORMAT_Z32_UNORM: - strb->Base.InternalFormat = GL_DEPTH_COMPONENT32; - break; - case PIPE_FORMAT_Z24_UNORM_S8_USCALED: - case PIPE_FORMAT_S8_USCALED_Z24_UNORM: - case PIPE_FORMAT_Z24X8_UNORM: - case PIPE_FORMAT_X8Z24_UNORM: - strb->Base.InternalFormat = GL_DEPTH24_STENCIL8_EXT; - break; - case PIPE_FORMAT_S8_USCALED: - strb->Base.InternalFormat = GL_STENCIL_INDEX8_EXT; - break; - case PIPE_FORMAT_R16G16B16A16_SNORM: - strb->Base.InternalFormat = GL_RGBA16; - break; - case PIPE_FORMAT_R8_UNORM: - strb->Base.InternalFormat = GL_R8; - break; - case PIPE_FORMAT_R8G8_UNORM: - strb->Base.InternalFormat = GL_RG8; - break; - case PIPE_FORMAT_R16_UNORM: - strb->Base.InternalFormat = GL_R16; - break; - case PIPE_FORMAT_R16G16_UNORM: - strb->Base.InternalFormat = GL_RG16; - break; - default: - _mesa_problem(NULL, - "Unexpected format in st_new_renderbuffer_fb"); - free(strb); - return NULL; - } - - /* st-specific methods */ - strb->Base.Delete = st_renderbuffer_delete; - strb->Base.AllocStorage = st_renderbuffer_alloc_storage; - strb->Base.GetPointer = null_get_pointer; - - /* surface is allocated in st_renderbuffer_alloc_storage() */ - strb->surface = NULL; - - return &strb->Base; -} - - - - -/** - * Called via ctx->Driver.BindFramebufferEXT(). - */ -static void -st_bind_framebuffer(struct gl_context *ctx, GLenum target, - struct gl_framebuffer *fb, struct gl_framebuffer *fbread) -{ - -} - -/** - * Called by ctx->Driver.FramebufferRenderbuffer - */ -static void -st_framebuffer_renderbuffer(struct gl_context *ctx, - struct gl_framebuffer *fb, - GLenum attachment, - struct gl_renderbuffer *rb) -{ - /* XXX no need for derivation? */ - _mesa_framebuffer_renderbuffer(ctx, fb, attachment, rb); -} - - -/** - * Called by ctx->Driver.RenderTexture - */ -static void -st_render_texture(struct gl_context *ctx, - struct gl_framebuffer *fb, - struct gl_renderbuffer_attachment *att) -{ - struct st_context *st = st_context(ctx); - struct pipe_context *pipe = st->pipe; - struct st_renderbuffer *strb; - struct gl_renderbuffer *rb; - struct pipe_resource *pt = st_get_texobj_resource(att->Texture); - struct st_texture_object *stObj; - const struct gl_texture_image *texImage; - struct pipe_surface surf_tmpl; - - /* When would this fail? Perhaps assert? */ - if (!pt) - return; - - /* get pointer to texture image we're rendeing to */ - texImage = att->Texture->Image[att->CubeMapFace][att->TextureLevel]; - - /* create new renderbuffer which wraps the texture image */ - rb = st_new_renderbuffer(ctx, 0); - if (!rb) { - _mesa_error(ctx, GL_OUT_OF_MEMORY, "glFramebufferTexture()"); - return; - } - - _mesa_reference_renderbuffer(&att->Renderbuffer, rb); - assert(rb->RefCount == 1); - rb->AllocStorage = NULL; /* should not get called */ - strb = st_renderbuffer(rb); - - assert(strb->Base.RefCount > 0); - - /* get the texture for the texture object */ - stObj = st_texture_object(att->Texture); - - /* point renderbuffer at texobject */ - strb->rtt = stObj; - strb->rtt_level = att->TextureLevel; - strb->rtt_face = att->CubeMapFace; - strb->rtt_slice = att->Zoffset; - - rb->Width = texImage->Width2; - rb->Height = texImage->Height2; - rb->_BaseFormat = texImage->_BaseFormat; - /*printf("***** render to texture level %d: %d x %d\n", att->TextureLevel, rb->Width, rb->Height);*/ - - /*printf("***** pipe texture %d x %d\n", pt->width0, pt->height0);*/ - - pipe_resource_reference( &strb->texture, pt ); - - pipe_surface_reference(&strb->surface, NULL); - - pipe_sampler_view_reference(&strb->sampler_view, - st_get_texture_sampler_view(stObj, pipe)); - - assert(strb->rtt_level <= strb->texture->last_level); - - /* new surface for rendering into the texture */ - memset(&surf_tmpl, 0, sizeof(surf_tmpl)); - surf_tmpl.format = ctx->Color.sRGBEnabled ? strb->texture->format : util_format_linear(strb->texture->format); - surf_tmpl.usage = PIPE_BIND_RENDER_TARGET; - surf_tmpl.u.tex.level = strb->rtt_level; - surf_tmpl.u.tex.first_layer = strb->rtt_face + strb->rtt_slice; - surf_tmpl.u.tex.last_layer = strb->rtt_face + strb->rtt_slice; - strb->surface = pipe->create_surface(pipe, - strb->texture, - &surf_tmpl); - - strb->format = pt->format; - - strb->Base.Format = st_pipe_format_to_mesa_format(pt->format); - strb->Base.DataType = st_format_datatype(pt->format); - - /* - printf("RENDER TO TEXTURE obj=%p pt=%p surf=%p %d x %d\n", - att->Texture, pt, strb->surface, rb->Width, rb->Height); - */ - - /* Invalidate buffer state so that the pipe's framebuffer state - * gets updated. - * That's where the new renderbuffer (which we just created) gets - * passed to the pipe as a (color/depth) render target. - */ - st_invalidate_state(ctx, _NEW_BUFFERS); -} - - -/** - * Called via ctx->Driver.FinishRenderTexture. - */ -static void -st_finish_render_texture(struct gl_context *ctx, - struct gl_renderbuffer_attachment *att) -{ - struct st_context *st = st_context(ctx); - struct st_renderbuffer *strb = st_renderbuffer(att->Renderbuffer); - - if (!strb) - return; - - st_flush(st, PIPE_FLUSH_RENDER_CACHE, NULL); - - strb->rtt = NULL; - - /* - printf("FINISH RENDER TO TEXTURE surf=%p\n", strb->surface); - */ - - /* restore previous framebuffer state */ - st_invalidate_state(ctx, _NEW_BUFFERS); -} - - -/** - * Validate a renderbuffer attachment for a particular set of bindings. - */ -static GLboolean -st_validate_attachment(struct pipe_screen *screen, - const struct gl_renderbuffer_attachment *att, - unsigned bindings) -{ - const struct st_texture_object *stObj = st_texture_object(att->Texture); - - /* Only validate texture attachments for now, since - * st_renderbuffer_alloc_storage makes sure that - * the format is supported. - */ - if (att->Type != GL_TEXTURE) - return GL_TRUE; - - if (!stObj) - return GL_FALSE; - - return screen->is_format_supported(screen, stObj->pt->format, - PIPE_TEXTURE_2D, - stObj->pt->nr_samples, bindings, 0); -} - - -/** - * Check if two renderbuffer attachments name a combined depth/stencil - * renderbuffer. - */ -GLboolean -st_is_depth_stencil_combined(const struct gl_renderbuffer_attachment *depth, - const struct gl_renderbuffer_attachment *stencil) -{ - assert(depth && stencil); - - if (depth->Type == stencil->Type) { - if (depth->Type == GL_RENDERBUFFER_EXT && - depth->Renderbuffer == stencil->Renderbuffer) - return GL_TRUE; - - if (depth->Type == GL_TEXTURE && - depth->Texture == stencil->Texture) - return GL_TRUE; - } - - return GL_FALSE; -} - - -/** - * Check that the framebuffer configuration is valid in terms of what - * the driver can support. - * - * For Gallium we only supports combined Z+stencil, not separate buffers. - */ -static void -st_validate_framebuffer(struct gl_context *ctx, struct gl_framebuffer *fb) -{ - struct st_context *st = st_context(ctx); - struct pipe_screen *screen = st->pipe->screen; - const struct gl_renderbuffer_attachment *depth = - &fb->Attachment[BUFFER_DEPTH]; - const struct gl_renderbuffer_attachment *stencil = - &fb->Attachment[BUFFER_STENCIL]; - GLuint i; - - if (depth->Type && stencil->Type && depth->Type != stencil->Type) { - fb->_Status = GL_FRAMEBUFFER_UNSUPPORTED_EXT; - return; - } - if (depth->Type == GL_RENDERBUFFER_EXT && - stencil->Type == GL_RENDERBUFFER_EXT && - depth->Renderbuffer != stencil->Renderbuffer) { - fb->_Status = GL_FRAMEBUFFER_UNSUPPORTED_EXT; - return; - } - if (depth->Type == GL_TEXTURE && - stencil->Type == GL_TEXTURE && - depth->Texture != stencil->Texture) { - fb->_Status = GL_FRAMEBUFFER_UNSUPPORTED_EXT; - return; - } - - if (!st_validate_attachment(screen, - depth, - PIPE_BIND_DEPTH_STENCIL)) { - fb->_Status = GL_FRAMEBUFFER_UNSUPPORTED_EXT; - return; - } - if (!st_validate_attachment(screen, - stencil, - PIPE_BIND_DEPTH_STENCIL)) { - fb->_Status = GL_FRAMEBUFFER_UNSUPPORTED_EXT; - return; - } - for (i = 0; i < ctx->Const.MaxColorAttachments; i++) { - if (!st_validate_attachment(screen, - &fb->Attachment[BUFFER_COLOR0 + i], - PIPE_BIND_RENDER_TARGET)) { - fb->_Status = GL_FRAMEBUFFER_UNSUPPORTED_EXT; - return; - } - } -} - - -/** - * Called via glDrawBuffer. - */ -static void -st_DrawBuffers(struct gl_context *ctx, GLsizei count, const GLenum *buffers) -{ - struct st_context *st = st_context(ctx); - struct gl_framebuffer *fb = ctx->DrawBuffer; - GLuint i; - - (void) count; - (void) buffers; - - /* add the renderbuffers on demand */ - for (i = 0; i < fb->_NumColorDrawBuffers; i++) { - gl_buffer_index idx = fb->_ColorDrawBufferIndexes[i]; - st_manager_add_color_renderbuffer(st, fb, idx); - } -} - - -/** - * Called via glReadBuffer. - */ -static void -st_ReadBuffer(struct gl_context *ctx, GLenum buffer) -{ - struct st_context *st = st_context(ctx); - struct gl_framebuffer *fb = ctx->ReadBuffer; - - (void) buffer; - - /* add the renderbuffer on demand */ - st_manager_add_color_renderbuffer(st, fb, fb->_ColorReadBufferIndex); -} - - -void st_init_fbo_functions(struct dd_function_table *functions) -{ -#if FEATURE_EXT_framebuffer_object - functions->NewFramebuffer = st_new_framebuffer; - functions->NewRenderbuffer = st_new_renderbuffer; - functions->BindFramebuffer = st_bind_framebuffer; - functions->FramebufferRenderbuffer = st_framebuffer_renderbuffer; - functions->RenderTexture = st_render_texture; - functions->FinishRenderTexture = st_finish_render_texture; - functions->ValidateFramebuffer = st_validate_framebuffer; -#endif - /* no longer needed by core Mesa, drivers handle resizes... - functions->ResizeBuffers = st_resize_buffers; - */ - - functions->DrawBuffers = st_DrawBuffers; - functions->ReadBuffer = st_ReadBuffer; -} - -/* XXX unused ? */ -struct pipe_sampler_view * -st_get_renderbuffer_sampler_view(struct st_renderbuffer *rb, - struct pipe_context *pipe) -{ - if (!rb->sampler_view) { - rb->sampler_view = st_create_texture_sampler_view(pipe, rb->texture); - } - - return rb->sampler_view; -} +/************************************************************************** + * + * 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. + * + **************************************************************************/ + + +/** + * Framebuffer/renderbuffer functions. + * + * \author Brian Paul + */ + + +#include "main/imports.h" +#include "main/context.h" +#include "main/fbobject.h" +#include "main/framebuffer.h" +#include "main/macros.h" +#include "main/mfeatures.h" +#include "main/renderbuffer.h" + +#include "pipe/p_context.h" +#include "pipe/p_defines.h" +#include "pipe/p_screen.h" +#include "st_context.h" +#include "st_cb_fbo.h" +#include "st_cb_flush.h" +#include "st_format.h" +#include "st_texture.h" +#include "st_manager.h" + +#include "util/u_format.h" +#include "util/u_inlines.h" +#include "util/u_surface.h" + + +/** + * gl_renderbuffer::AllocStorage() + * This is called to allocate the original drawing surface, and + * during window resize. + */ +static GLboolean +st_renderbuffer_alloc_storage(struct gl_context * ctx, + struct gl_renderbuffer *rb, + GLenum internalFormat, + GLuint width, GLuint height) +{ + struct st_context *st = st_context(ctx); + struct pipe_context *pipe = st->pipe; + struct pipe_screen *screen = st->pipe->screen; + struct st_renderbuffer *strb = st_renderbuffer(rb); + enum pipe_format format; + struct pipe_surface surf_tmpl; + + if (strb->format != PIPE_FORMAT_NONE) + format = strb->format; + else + format = st_choose_renderbuffer_format(screen, internalFormat, + rb->NumSamples); + + /* init renderbuffer fields */ + strb->Base.Width = width; + strb->Base.Height = height; + strb->Base.Format = st_pipe_format_to_mesa_format(format); + strb->Base.DataType = st_format_datatype(format); + + strb->defined = GL_FALSE; /* undefined contents now */ + + if (strb->software) { + size_t size; + + free(strb->data); + + assert(strb->format != PIPE_FORMAT_NONE); + + strb->stride = util_format_get_stride(strb->format, width); + size = util_format_get_2d_size(strb->format, strb->stride, height); + + strb->data = malloc(size); + + return strb->data != NULL; + } + else { + struct pipe_resource template; + + /* Free the old surface and texture + */ + pipe_surface_reference( &strb->surface, NULL ); + pipe_resource_reference( &strb->texture, NULL ); + pipe_sampler_view_reference(&strb->sampler_view, NULL); + + /* Setup new texture template. + */ + memset(&template, 0, sizeof(template)); + template.target = st->internal_target; + template.format = format; + template.width0 = width; + template.height0 = height; + template.depth0 = 1; + template.array_size = 1; + template.last_level = 0; + template.nr_samples = rb->NumSamples; + if (util_format_is_depth_or_stencil(format)) { + template.bind = PIPE_BIND_DEPTH_STENCIL; + } + else { + template.bind = (PIPE_BIND_DISPLAY_TARGET | + PIPE_BIND_RENDER_TARGET); + } + + strb->texture = screen->resource_create(screen, &template); + + if (!strb->texture) + return FALSE; + + memset(&surf_tmpl, 0, sizeof(surf_tmpl)); + u_surface_default_template(&surf_tmpl, strb->texture, template.bind); + strb->surface = pipe->create_surface(pipe, + strb->texture, + &surf_tmpl); + if (strb->surface) { + assert(strb->surface->texture); + assert(strb->surface->format); + assert(strb->surface->width == width); + assert(strb->surface->height == height); + } + + return strb->surface != NULL; + } +} + + +/** + * gl_renderbuffer::Delete() + */ +static void +st_renderbuffer_delete(struct gl_renderbuffer *rb) +{ + struct st_renderbuffer *strb = st_renderbuffer(rb); + ASSERT(strb); + pipe_surface_reference(&strb->surface, NULL); + pipe_resource_reference(&strb->texture, NULL); + pipe_sampler_view_reference(&strb->sampler_view, NULL); + free(strb->data); + free(strb); +} + + +/** + * gl_renderbuffer::GetPointer() + */ +static void * +null_get_pointer(struct gl_context * ctx, struct gl_renderbuffer *rb, + GLint x, GLint y) +{ + /* By returning NULL we force all software rendering to go through + * the span routines. + */ +#if 0 + assert(0); /* Should never get called with softpipe */ +#endif + return NULL; +} + + +/** + * Called via ctx->Driver.NewFramebuffer() + */ +static struct gl_framebuffer * +st_new_framebuffer(struct gl_context *ctx, GLuint name) +{ + /* XXX not sure we need to subclass gl_framebuffer for pipe */ + return _mesa_new_framebuffer(ctx, name); +} + + +/** + * Called via ctx->Driver.NewRenderbuffer() + */ +static struct gl_renderbuffer * +st_new_renderbuffer(struct gl_context *ctx, GLuint name) +{ + struct st_renderbuffer *strb = ST_CALLOC_STRUCT(st_renderbuffer); + if (strb) { + _mesa_init_renderbuffer(&strb->Base, name); + strb->Base.Delete = st_renderbuffer_delete; + strb->Base.AllocStorage = st_renderbuffer_alloc_storage; + strb->Base.GetPointer = null_get_pointer; + strb->format = PIPE_FORMAT_NONE; + return &strb->Base; + } + return NULL; +} + + +/** + * Allocate a renderbuffer for a an on-screen window (not a user-created + * renderbuffer). The window system code determines the format. + */ +struct gl_renderbuffer * +st_new_renderbuffer_fb(enum pipe_format format, int samples, boolean sw) +{ + struct st_renderbuffer *strb; + + strb = ST_CALLOC_STRUCT(st_renderbuffer); + if (!strb) { + _mesa_error(NULL, GL_OUT_OF_MEMORY, "creating renderbuffer"); + return NULL; + } + + _mesa_init_renderbuffer(&strb->Base, 0); + strb->Base.ClassID = 0x4242; /* just a unique value */ + strb->Base.NumSamples = samples; + strb->Base.Format = st_pipe_format_to_mesa_format(format); + strb->Base.DataType = st_format_datatype(format); + strb->format = format; + strb->software = sw; + + switch (format) { + case PIPE_FORMAT_R8G8B8A8_UNORM: + case PIPE_FORMAT_B8G8R8A8_UNORM: + case PIPE_FORMAT_A8R8G8B8_UNORM: + case PIPE_FORMAT_R8G8B8X8_UNORM: + case PIPE_FORMAT_B8G8R8X8_UNORM: + case PIPE_FORMAT_X8R8G8B8_UNORM: + case PIPE_FORMAT_B5G5R5A1_UNORM: + case PIPE_FORMAT_B4G4R4A4_UNORM: + case PIPE_FORMAT_B5G6R5_UNORM: + strb->Base.InternalFormat = GL_RGBA; + break; + case PIPE_FORMAT_Z16_UNORM: + strb->Base.InternalFormat = GL_DEPTH_COMPONENT16; + break; + case PIPE_FORMAT_Z32_UNORM: + strb->Base.InternalFormat = GL_DEPTH_COMPONENT32; + break; + case PIPE_FORMAT_Z24_UNORM_S8_USCALED: + case PIPE_FORMAT_S8_USCALED_Z24_UNORM: + case PIPE_FORMAT_Z24X8_UNORM: + case PIPE_FORMAT_X8Z24_UNORM: + strb->Base.InternalFormat = GL_DEPTH24_STENCIL8_EXT; + break; + case PIPE_FORMAT_S8_USCALED: + strb->Base.InternalFormat = GL_STENCIL_INDEX8_EXT; + break; + case PIPE_FORMAT_R16G16B16A16_SNORM: + strb->Base.InternalFormat = GL_RGBA16; + break; + case PIPE_FORMAT_R8_UNORM: + strb->Base.InternalFormat = GL_R8; + break; + case PIPE_FORMAT_R8G8_UNORM: + strb->Base.InternalFormat = GL_RG8; + break; + case PIPE_FORMAT_R16_UNORM: + strb->Base.InternalFormat = GL_R16; + break; + case PIPE_FORMAT_R16G16_UNORM: + strb->Base.InternalFormat = GL_RG16; + break; + default: + _mesa_problem(NULL, + "Unexpected format in st_new_renderbuffer_fb"); + free(strb); + return NULL; + } + + /* st-specific methods */ + strb->Base.Delete = st_renderbuffer_delete; + strb->Base.AllocStorage = st_renderbuffer_alloc_storage; + strb->Base.GetPointer = null_get_pointer; + + /* surface is allocated in st_renderbuffer_alloc_storage() */ + strb->surface = NULL; + + return &strb->Base; +} + + + + +/** + * Called via ctx->Driver.BindFramebufferEXT(). + */ +static void +st_bind_framebuffer(struct gl_context *ctx, GLenum target, + struct gl_framebuffer *fb, struct gl_framebuffer *fbread) +{ + +} + +/** + * Called by ctx->Driver.FramebufferRenderbuffer + */ +static void +st_framebuffer_renderbuffer(struct gl_context *ctx, + struct gl_framebuffer *fb, + GLenum attachment, + struct gl_renderbuffer *rb) +{ + /* XXX no need for derivation? */ + _mesa_framebuffer_renderbuffer(ctx, fb, attachment, rb); +} + + +/** + * Called by ctx->Driver.RenderTexture + */ +static void +st_render_texture(struct gl_context *ctx, + struct gl_framebuffer *fb, + struct gl_renderbuffer_attachment *att) +{ + struct st_context *st = st_context(ctx); + struct pipe_context *pipe = st->pipe; + struct st_renderbuffer *strb; + struct gl_renderbuffer *rb; + struct pipe_resource *pt = st_get_texobj_resource(att->Texture); + struct st_texture_object *stObj; + const struct gl_texture_image *texImage; + struct pipe_surface surf_tmpl; + + /* When would this fail? Perhaps assert? */ + if (!pt) + return; + + /* get pointer to texture image we're rendeing to */ + texImage = att->Texture->Image[att->CubeMapFace][att->TextureLevel]; + + /* create new renderbuffer which wraps the texture image */ + rb = st_new_renderbuffer(ctx, 0); + if (!rb) { + _mesa_error(ctx, GL_OUT_OF_MEMORY, "glFramebufferTexture()"); + return; + } + + _mesa_reference_renderbuffer(&att->Renderbuffer, rb); + assert(rb->RefCount == 1); + rb->AllocStorage = NULL; /* should not get called */ + strb = st_renderbuffer(rb); + + assert(strb->Base.RefCount > 0); + + /* get the texture for the texture object */ + stObj = st_texture_object(att->Texture); + + /* point renderbuffer at texobject */ + strb->rtt = stObj; + strb->rtt_level = att->TextureLevel; + strb->rtt_face = att->CubeMapFace; + strb->rtt_slice = att->Zoffset; + + rb->Width = texImage->Width2; + rb->Height = texImage->Height2; + rb->_BaseFormat = texImage->_BaseFormat; + /*printf("***** render to texture level %d: %d x %d\n", att->TextureLevel, rb->Width, rb->Height);*/ + + /*printf("***** pipe texture %d x %d\n", pt->width0, pt->height0);*/ + + pipe_resource_reference( &strb->texture, pt ); + + pipe_surface_reference(&strb->surface, NULL); + + pipe_sampler_view_reference(&strb->sampler_view, + st_get_texture_sampler_view(stObj, pipe)); + + assert(strb->rtt_level <= strb->texture->last_level); + + /* new surface for rendering into the texture */ + memset(&surf_tmpl, 0, sizeof(surf_tmpl)); + surf_tmpl.format = ctx->Color.sRGBEnabled ? strb->texture->format : util_format_linear(strb->texture->format); + surf_tmpl.usage = PIPE_BIND_RENDER_TARGET; + surf_tmpl.u.tex.level = strb->rtt_level; + surf_tmpl.u.tex.first_layer = strb->rtt_face + strb->rtt_slice; + surf_tmpl.u.tex.last_layer = strb->rtt_face + strb->rtt_slice; + strb->surface = pipe->create_surface(pipe, + strb->texture, + &surf_tmpl); + + strb->format = pt->format; + + strb->Base.Format = st_pipe_format_to_mesa_format(pt->format); + strb->Base.DataType = st_format_datatype(pt->format); + + /* + printf("RENDER TO TEXTURE obj=%p pt=%p surf=%p %d x %d\n", + att->Texture, pt, strb->surface, rb->Width, rb->Height); + */ + + /* Invalidate buffer state so that the pipe's framebuffer state + * gets updated. + * That's where the new renderbuffer (which we just created) gets + * passed to the pipe as a (color/depth) render target. + */ + st_invalidate_state(ctx, _NEW_BUFFERS); +} + + +/** + * Called via ctx->Driver.FinishRenderTexture. + */ +static void +st_finish_render_texture(struct gl_context *ctx, + struct gl_renderbuffer_attachment *att) +{ + struct st_context *st = st_context(ctx); + struct st_renderbuffer *strb = st_renderbuffer(att->Renderbuffer); + + if (!strb) + return; + + st_flush(st, PIPE_FLUSH_RENDER_CACHE, NULL); + + strb->rtt = NULL; + + /* + printf("FINISH RENDER TO TEXTURE surf=%p\n", strb->surface); + */ + + /* restore previous framebuffer state */ + st_invalidate_state(ctx, _NEW_BUFFERS); +} + + +/** + * Validate a renderbuffer attachment for a particular set of bindings. + */ +static GLboolean +st_validate_attachment(struct gl_context *ctx, + struct pipe_screen *screen, + const struct gl_renderbuffer_attachment *att, + unsigned bindings) +{ + const struct st_texture_object *stObj = st_texture_object(att->Texture); + enum pipe_format format; + gl_format texFormat; + + /* Only validate texture attachments for now, since + * st_renderbuffer_alloc_storage makes sure that + * the format is supported. + */ + if (att->Type != GL_TEXTURE) + return GL_TRUE; + + if (!stObj) + return GL_FALSE; + + format = stObj->pt->format; + texFormat = + stObj->base.Image[att->CubeMapFace][att->TextureLevel]->TexFormat; + + /* If the encoding is sRGB and sRGB rendering cannot be enabled, + * check for linear format support instead. + * Later when we create a surface, we change the format to a linear one. */ + if (!ctx->Const.sRGBCapable && + _mesa_get_format_color_encoding(texFormat) == GL_SRGB) { + const gl_format linearFormat = _mesa_get_srgb_format_linear(texFormat); + format = st_mesa_format_to_pipe_format(linearFormat); + } + + return screen->is_format_supported(screen, format, + PIPE_TEXTURE_2D, + stObj->pt->nr_samples, bindings, 0); +} + + +/** + * Check if two renderbuffer attachments name a combined depth/stencil + * renderbuffer. + */ +GLboolean +st_is_depth_stencil_combined(const struct gl_renderbuffer_attachment *depth, + const struct gl_renderbuffer_attachment *stencil) +{ + assert(depth && stencil); + + if (depth->Type == stencil->Type) { + if (depth->Type == GL_RENDERBUFFER_EXT && + depth->Renderbuffer == stencil->Renderbuffer) + return GL_TRUE; + + if (depth->Type == GL_TEXTURE && + depth->Texture == stencil->Texture) + return GL_TRUE; + } + + return GL_FALSE; +} + + +/** + * Check that the framebuffer configuration is valid in terms of what + * the driver can support. + * + * For Gallium we only supports combined Z+stencil, not separate buffers. + */ +static void +st_validate_framebuffer(struct gl_context *ctx, struct gl_framebuffer *fb) +{ + struct st_context *st = st_context(ctx); + struct pipe_screen *screen = st->pipe->screen; + const struct gl_renderbuffer_attachment *depth = + &fb->Attachment[BUFFER_DEPTH]; + const struct gl_renderbuffer_attachment *stencil = + &fb->Attachment[BUFFER_STENCIL]; + GLuint i; + + if (depth->Type && stencil->Type && depth->Type != stencil->Type) { + fb->_Status = GL_FRAMEBUFFER_UNSUPPORTED_EXT; + return; + } + if (depth->Type == GL_RENDERBUFFER_EXT && + stencil->Type == GL_RENDERBUFFER_EXT && + depth->Renderbuffer != stencil->Renderbuffer) { + fb->_Status = GL_FRAMEBUFFER_UNSUPPORTED_EXT; + return; + } + if (depth->Type == GL_TEXTURE && + stencil->Type == GL_TEXTURE && + depth->Texture != stencil->Texture) { + fb->_Status = GL_FRAMEBUFFER_UNSUPPORTED_EXT; + return; + } + + if (!st_validate_attachment(ctx, + screen, + depth, + PIPE_BIND_DEPTH_STENCIL)) { + fb->_Status = GL_FRAMEBUFFER_UNSUPPORTED_EXT; + return; + } + if (!st_validate_attachment(ctx, + screen, + stencil, + PIPE_BIND_DEPTH_STENCIL)) { + fb->_Status = GL_FRAMEBUFFER_UNSUPPORTED_EXT; + return; + } + for (i = 0; i < ctx->Const.MaxColorAttachments; i++) { + if (!st_validate_attachment(ctx, + screen, + &fb->Attachment[BUFFER_COLOR0 + i], + PIPE_BIND_RENDER_TARGET)) { + fb->_Status = GL_FRAMEBUFFER_UNSUPPORTED_EXT; + return; + } + } +} + + +/** + * Called via glDrawBuffer. + */ +static void +st_DrawBuffers(struct gl_context *ctx, GLsizei count, const GLenum *buffers) +{ + struct st_context *st = st_context(ctx); + struct gl_framebuffer *fb = ctx->DrawBuffer; + GLuint i; + + (void) count; + (void) buffers; + + /* add the renderbuffers on demand */ + for (i = 0; i < fb->_NumColorDrawBuffers; i++) { + gl_buffer_index idx = fb->_ColorDrawBufferIndexes[i]; + st_manager_add_color_renderbuffer(st, fb, idx); + } +} + + +/** + * Called via glReadBuffer. + */ +static void +st_ReadBuffer(struct gl_context *ctx, GLenum buffer) +{ + struct st_context *st = st_context(ctx); + struct gl_framebuffer *fb = ctx->ReadBuffer; + + (void) buffer; + + /* add the renderbuffer on demand */ + st_manager_add_color_renderbuffer(st, fb, fb->_ColorReadBufferIndex); +} + + +void st_init_fbo_functions(struct dd_function_table *functions) +{ +#if FEATURE_EXT_framebuffer_object + functions->NewFramebuffer = st_new_framebuffer; + functions->NewRenderbuffer = st_new_renderbuffer; + functions->BindFramebuffer = st_bind_framebuffer; + functions->FramebufferRenderbuffer = st_framebuffer_renderbuffer; + functions->RenderTexture = st_render_texture; + functions->FinishRenderTexture = st_finish_render_texture; + functions->ValidateFramebuffer = st_validate_framebuffer; +#endif + /* no longer needed by core Mesa, drivers handle resizes... + functions->ResizeBuffers = st_resize_buffers; + */ + + functions->DrawBuffers = st_DrawBuffers; + functions->ReadBuffer = st_ReadBuffer; +} + +/* XXX unused ? */ +struct pipe_sampler_view * +st_get_renderbuffer_sampler_view(struct st_renderbuffer *rb, + struct pipe_context *pipe) +{ + if (!rb->sampler_view) { + rb->sampler_view = st_create_texture_sampler_view(pipe, rb->texture); + } + + return rb->sampler_view; +} diff --git a/mesalib/src/mesa/state_tracker/st_manager.c b/mesalib/src/mesa/state_tracker/st_manager.c index 3b54eb7b8..5684e9768 100644 --- a/mesalib/src/mesa/state_tracker/st_manager.c +++ b/mesalib/src/mesa/state_tracker/st_manager.c @@ -426,19 +426,13 @@ st_framebuffer_create(struct st_framebuffer_iface *stfbi) struct gl_config mode; gl_buffer_index idx; + if (!stfbi) + return NULL; + stfb = CALLOC_STRUCT(st_framebuffer); if (!stfb) return NULL; - /* for FBO-only context */ - if (!stfbi) { - struct gl_framebuffer *base = _mesa_get_incomplete_framebuffer(); - - stfb->Base = *base; - - return stfb; - } - st_visual_to_context_mode(stfbi->visual, &mode); _mesa_initialize_window_framebuffer(&stfb->Base, &mode); @@ -764,7 +758,8 @@ st_api_make_current(struct st_api *stapi, struct st_context_iface *stctxi, ret = _mesa_make_current(st->ctx, &stdraw->Base, &stread->Base); } else { - ret = FALSE; + struct gl_framebuffer *incomplete = _mesa_get_incomplete_framebuffer(); + ret = _mesa_make_current(st->ctx, incomplete, incomplete); } st_framebuffer_reference(&stdraw, NULL); diff --git a/mesalib/src/mesa/swrast/s_trispan.h b/mesalib/src/mesa/swrast/s_trispan.h deleted file mode 100644 index 15207e863..000000000 --- a/mesalib/src/mesa/swrast/s_trispan.h +++ /dev/null @@ -1,31 +0,0 @@ - -/* - * Mesa 3-D graphics library - * Version: 3.5 - * - * Copyright (C) 1999-2001 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. - */ - - -#ifndef S_TRISPAN_H -#define S_TRISPAN_H - - -#endif /* S_TRISPAN_H */ diff --git a/mesalib/src/mesa/x86/read_rgba_span_x86.S b/mesalib/src/mesa/x86/read_rgba_span_x86.S index 817729973..04571afb7 100644 --- a/mesalib/src/mesa/x86/read_rgba_span_x86.S +++ b/mesalib/src/mesa/x86/read_rgba_span_x86.S @@ -1,678 +1,686 @@ -/* - * (C) Copyright IBM Corporation 2004 - * 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 - * on 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 - * IBM AND/OR THEIR 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. - */ - -/** - * \file read_rgba_span_x86.S - * Optimized routines to transfer pixel data from the framebuffer to a - * buffer in main memory. - * - * \author Ian Romanick - */ - - .file "read_rgba_span_x86.S" -#if !defined(__DJGPP__) && !defined(__MINGW32__) && !defined(__APPLE__) /* this one cries for assyntax.h */ -/* Kevin F. Quinn 2nd July 2006 - * Replaced data segment constants with text-segment instructions. - */ -#define LOAD_MASK(mvins,m1,m2) \ - pushl $0xff00ff00 ;\ - pushl $0xff00ff00 ;\ - pushl $0xff00ff00 ;\ - pushl $0xff00ff00 ;\ - mvins (%esp), m1 ;\ - pushl $0x00ff0000 ;\ - pushl $0x00ff0000 ;\ - pushl $0x00ff0000 ;\ - pushl $0x00ff0000 ;\ - mvins (%esp), m2 ;\ - addl $32, %esp - -/* I implemented these as macros because they appear in several places, - * and I've tweaked them a number of times. I got tired of changing every - * place they appear. :) - */ - -#define DO_ONE_PIXEL() \ - movl (%ebx), %eax ; \ - addl $4, %ebx ; \ - bswap %eax /* ARGB -> BGRA */ ; \ - rorl $8, %eax /* BGRA -> ABGR */ ; \ - movl %eax, (%ecx) /* ABGR -> R, G, B, A */ ; \ - addl $4, %ecx - -#define DO_ONE_LAST_PIXEL() \ - movl (%ebx), %eax ; \ - bswap %eax /* ARGB -> BGRA */ ; \ - rorl $8, %eax /* BGRA -> ABGR */ ; \ - movl %eax, (%ecx) /* ABGR -> R, G, B, A */ ; \ - - -/** - * MMX optimized version of the BGRA8888_REV to RGBA copy routine. - * - * \warning - * This function assumes that the caller will issue the EMMS instruction - * at the correct places. - */ - -.globl _generic_read_RGBA_span_BGRA8888_REV_MMX -.hidden _generic_read_RGBA_span_BGRA8888_REV_MMX - .type _generic_read_RGBA_span_BGRA8888_REV_MMX, @function -_generic_read_RGBA_span_BGRA8888_REV_MMX: - pushl %ebx - -#ifdef USE_INNER_EMMS - emms -#endif - LOAD_MASK(movq,%mm1,%mm2) - - movl 8(%esp), %ebx /* source pointer */ - movl 16(%esp), %edx /* number of pixels to copy */ - movl 12(%esp), %ecx /* destination pointer */ - - testl %edx, %edx - jle .L20 /* Bail if there's nothing to do. */ - - movl %ebx, %eax - - negl %eax - sarl $2, %eax - andl $1, %eax - je .L17 - - subl %eax, %edx - DO_ONE_PIXEL() -.L17: - - /* Would it be faster to unroll this loop once and process 4 pixels - * per pass, instead of just two? - */ - - movl %edx, %eax - shrl %eax - jmp .L18 -.L19: - movq (%ebx), %mm0 - addl $8, %ebx - - /* These 9 instructions do what PSHUFB (if there were such an - * instruction) could do in 1. :( - */ - - movq %mm0, %mm3 - movq %mm0, %mm4 - - pand %mm2, %mm3 - psllq $16, %mm4 - psrlq $16, %mm3 - pand %mm2, %mm4 - - pand %mm1, %mm0 - por %mm4, %mm3 - por %mm3, %mm0 - - movq %mm0, (%ecx) - addl $8, %ecx - subl $1, %eax -.L18: - jne .L19 - -#ifdef USE_INNER_EMMS - emms -#endif - - /* At this point there are either 1 or 0 pixels remaining to be - * converted. Convert the last pixel, if needed. - */ - - testl $1, %edx - je .L20 - - DO_ONE_LAST_PIXEL() - -.L20: - popl %ebx - ret - .size _generic_read_RGBA_span_BGRA8888_REV_MMX, .-_generic_read_RGBA_span_BGRA8888_REV_MMX - - -/** - * SSE optimized version of the BGRA8888_REV to RGBA copy routine. SSE - * instructions are only actually used to read data from the framebuffer. - * In practice, the speed-up is pretty small. - * - * \todo - * Do some more testing and determine if there's any reason to have this - * function in addition to the MMX version. - * - * \warning - * This function assumes that the caller will issue the EMMS instruction - * at the correct places. - */ - -.globl _generic_read_RGBA_span_BGRA8888_REV_SSE -.hidden _generic_read_RGBA_span_BGRA8888_REV_SSE - .type _generic_read_RGBA_span_BGRA8888_REV_SSE, @function -_generic_read_RGBA_span_BGRA8888_REV_SSE: - pushl %esi - pushl %ebx - pushl %ebp - -#ifdef USE_INNER_EMMS - emms -#endif - - LOAD_MASK(movq,%mm1,%mm2) - - movl 16(%esp), %ebx /* source pointer */ - movl 24(%esp), %edx /* number of pixels to copy */ - movl 20(%esp), %ecx /* destination pointer */ - - testl %edx, %edx - jle .L35 /* Bail if there's nothing to do. */ - - movl %esp, %ebp - subl $16, %esp - andl $0xfffffff0, %esp - - movl %ebx, %eax - movl %edx, %esi - - negl %eax - andl $15, %eax - sarl $2, %eax - cmpl %edx, %eax - cmovle %eax, %esi - - subl %esi, %edx - - testl $1, %esi - je .L32 - - DO_ONE_PIXEL() -.L32: - - testl $2, %esi - je .L31 - - movq (%ebx), %mm0 - addl $8, %ebx - - movq %mm0, %mm3 - movq %mm0, %mm4 - - pand %mm2, %mm3 - psllq $16, %mm4 - psrlq $16, %mm3 - pand %mm2, %mm4 - - pand %mm1, %mm0 - por %mm4, %mm3 - por %mm3, %mm0 - - movq %mm0, (%ecx) - addl $8, %ecx -.L31: - - movl %edx, %eax - shrl $2, %eax - jmp .L33 -.L34: - movaps (%ebx), %xmm0 - addl $16, %ebx - - /* This would be so much better if we could just move directly from - * an SSE register to an MMX register. Unfortunately, that - * functionality wasn't introduced until SSE2 with the MOVDQ2Q - * instruction. - */ - - movaps %xmm0, (%esp) - movq (%esp), %mm0 - movq 8(%esp), %mm5 - - movq %mm0, %mm3 - movq %mm0, %mm4 - movq %mm5, %mm6 - movq %mm5, %mm7 - - pand %mm2, %mm3 - pand %mm2, %mm6 - - psllq $16, %mm4 - psllq $16, %mm7 - - psrlq $16, %mm3 - psrlq $16, %mm6 - - pand %mm2, %mm4 - pand %mm2, %mm7 - - pand %mm1, %mm0 - pand %mm1, %mm5 - - por %mm4, %mm3 - por %mm7, %mm6 - - por %mm3, %mm0 - por %mm6, %mm5 - - movq %mm0, (%ecx) - movq %mm5, 8(%ecx) - addl $16, %ecx - - subl $1, %eax -.L33: - jne .L34 - -#ifdef USE_INNER_EMMS - emms -#endif - movl %ebp, %esp - - /* At this point there are either [0, 3] pixels remaining to be - * converted. - */ - - testl $2, %edx - je .L36 - - movq (%ebx), %mm0 - addl $8, %ebx - - movq %mm0, %mm3 - movq %mm0, %mm4 - - pand %mm2, %mm3 - psllq $16, %mm4 - psrlq $16, %mm3 - pand %mm2, %mm4 - - pand %mm1, %mm0 - por %mm4, %mm3 - por %mm3, %mm0 - - movq %mm0, (%ecx) - addl $8, %ecx -.L36: - - testl $1, %edx - je .L35 - - DO_ONE_LAST_PIXEL() -.L35: - popl %ebp - popl %ebx - popl %esi - ret - .size _generic_read_RGBA_span_BGRA8888_REV_SSE, .-_generic_read_RGBA_span_BGRA8888_REV_SSE - - -/** - * SSE2 optimized version of the BGRA8888_REV to RGBA copy routine. - */ - - .text -.globl _generic_read_RGBA_span_BGRA8888_REV_SSE2 -.hidden _generic_read_RGBA_span_BGRA8888_REV_SSE2 - .type _generic_read_RGBA_span_BGRA8888_REV_SSE2, @function -_generic_read_RGBA_span_BGRA8888_REV_SSE2: - pushl %esi - pushl %ebx - - LOAD_MASK(movdqu,%xmm1,%xmm2) - - movl 12(%esp), %ebx /* source pointer */ - movl 20(%esp), %edx /* number of pixels to copy */ - movl 16(%esp), %ecx /* destination pointer */ - - movl %ebx, %eax - movl %edx, %esi - - testl %edx, %edx - jle .L46 /* Bail if there's nothing to do. */ - - /* If the source pointer isn't a multiple of 16 we have to process - * a few pixels the "slow" way to get the address aligned for - * the SSE fetch intsructions. - */ - - negl %eax - andl $15, %eax - sarl $2, %eax - - cmpl %edx, %eax - cmovbe %eax, %esi - subl %esi, %edx - - testl $1, %esi - je .L41 - - DO_ONE_PIXEL() -.L41: - testl $2, %esi - je .L40 - - movq (%ebx), %xmm0 - addl $8, %ebx - - movdqa %xmm0, %xmm3 - movdqa %xmm0, %xmm4 - andps %xmm1, %xmm0 - - andps %xmm2, %xmm3 - pslldq $2, %xmm4 - psrldq $2, %xmm3 - andps %xmm2, %xmm4 - - orps %xmm4, %xmm3 - orps %xmm3, %xmm0 - - movq %xmm0, (%ecx) - addl $8, %ecx -.L40: - - /* Would it be worth having a specialized version of this loop for - * the case where the destination is 16-byte aligned? That version - * would be identical except that it could use movedqa instead of - * movdqu. - */ - - movl %edx, %eax - shrl $2, %eax - jmp .L42 -.L43: - movdqa (%ebx), %xmm0 - addl $16, %ebx - - movdqa %xmm0, %xmm3 - movdqa %xmm0, %xmm4 - andps %xmm1, %xmm0 - - andps %xmm2, %xmm3 - pslldq $2, %xmm4 - psrldq $2, %xmm3 - andps %xmm2, %xmm4 - - orps %xmm4, %xmm3 - orps %xmm3, %xmm0 - - movdqu %xmm0, (%ecx) - addl $16, %ecx - subl $1, %eax -.L42: - jne .L43 - - - /* There may be upto 3 pixels remaining to be copied. Take care - * of them now. We do the 2 pixel case first because the data - * will be aligned. - */ - - testl $2, %edx - je .L47 - - movq (%ebx), %xmm0 - addl $8, %ebx - - movdqa %xmm0, %xmm3 - movdqa %xmm0, %xmm4 - andps %xmm1, %xmm0 - - andps %xmm2, %xmm3 - pslldq $2, %xmm4 - psrldq $2, %xmm3 - andps %xmm2, %xmm4 - - orps %xmm4, %xmm3 - orps %xmm3, %xmm0 - - movq %xmm0, (%ecx) - addl $8, %ecx -.L47: - - testl $1, %edx - je .L46 - - DO_ONE_LAST_PIXEL() -.L46: - - popl %ebx - popl %esi - ret - .size _generic_read_RGBA_span_BGRA8888_REV_SSE2, .-_generic_read_RGBA_span_BGRA8888_REV_SSE2 - - - -#define MASK_565_L 0x07e0f800 -#define MASK_565_H 0x0000001f -/* Setting SCALE_ADJUST to 5 gives a perfect match with the - * classic C implementation in Mesa. Setting SCALE_ADJUST - * to 0 is slightly faster but at a small cost to accuracy. - */ -#define SCALE_ADJUST 5 -#if SCALE_ADJUST == 5 -#define PRESCALE_L 0x00100001 -#define PRESCALE_H 0x00000200 -#define SCALE_L 0x40C620E8 -#define SCALE_H 0x0000839d -#elif SCALE_ADJUST == 0 -#define PRESCALE_L 0x00200001 -#define PRESCALE_H 0x00000800 -#define SCALE_L 0x01040108 -#define SCALE_H 0x00000108 -#else -#error SCALE_ADJUST must either be 5 or 0. -#endif -#define ALPHA_L 0x00000000 -#define ALPHA_H 0x00ff0000 - -/** - * MMX optimized version of the RGB565 to RGBA copy routine. - */ - - .text - .globl _generic_read_RGBA_span_RGB565_MMX - .hidden _generic_read_RGBA_span_RGB565_MMX - .type _generic_read_RGBA_span_RGB565_MMX, @function - -_generic_read_RGBA_span_RGB565_MMX: - -#ifdef USE_INNER_EMMS - emms -#endif - - movl 4(%esp), %eax /* source pointer */ - movl 8(%esp), %edx /* destination pointer */ - movl 12(%esp), %ecx /* number of pixels to copy */ - - pushl $MASK_565_H - pushl $MASK_565_L - movq (%esp), %mm5 - pushl $PRESCALE_H - pushl $PRESCALE_L - movq (%esp), %mm6 - pushl $SCALE_H - pushl $SCALE_L - movq (%esp), %mm7 - pushl $ALPHA_H - pushl $ALPHA_L - movq (%esp), %mm3 - addl $32,%esp - - sarl $2, %ecx - jl .L01 /* Bail early if the count is negative. */ - jmp .L02 - -.L03: - /* Fetch 4 RGB565 pixels into %mm4. Distribute the first and - * second pixels into the four words of %mm0 and %mm2. - */ - - movq (%eax), %mm4 - addl $8, %eax - - pshufw $0x00, %mm4, %mm0 - pshufw $0x55, %mm4, %mm2 - - - /* Mask the pixels so that each word of each register contains only - * one color component. - */ - - pand %mm5, %mm0 - pand %mm5, %mm2 - - - /* Adjust the component values so that they are as small as possible, - * but large enough so that we can multiply them by an unsigned 16-bit - * number and get a value as large as 0x00ff0000. - */ - - pmullw %mm6, %mm0 - pmullw %mm6, %mm2 -#if SCALE_ADJUST > 0 - psrlw $SCALE_ADJUST, %mm0 - psrlw $SCALE_ADJUST, %mm2 -#endif - - /* Scale the input component values to be on the range - * [0, 0x00ff0000]. This it the real magic of the whole routine. - */ - - pmulhuw %mm7, %mm0 - pmulhuw %mm7, %mm2 - - - /* Always set the alpha value to 0xff. - */ - - por %mm3, %mm0 - por %mm3, %mm2 - - - /* Pack the 16-bit values to 8-bit values and store the converted - * pixel data. - */ - - packuswb %mm2, %mm0 - movq %mm0, (%edx) - addl $8, %edx - - pshufw $0xaa, %mm4, %mm0 - pshufw $0xff, %mm4, %mm2 - - pand %mm5, %mm0 - pand %mm5, %mm2 - pmullw %mm6, %mm0 - pmullw %mm6, %mm2 -#if SCALE_ADJUST > 0 - psrlw $SCALE_ADJUST, %mm0 - psrlw $SCALE_ADJUST, %mm2 -#endif - pmulhuw %mm7, %mm0 - pmulhuw %mm7, %mm2 - - por %mm3, %mm0 - por %mm3, %mm2 - - packuswb %mm2, %mm0 - - movq %mm0, (%edx) - addl $8, %edx - - subl $1, %ecx -.L02: - jne .L03 - - - /* At this point there can be at most 3 pixels left to process. If - * there is either 2 or 3 left, process 2. - */ - - movl 12(%esp), %ecx - testl $0x02, %ecx - je .L04 - - movd (%eax), %mm4 - addl $4, %eax - - pshufw $0x00, %mm4, %mm0 - pshufw $0x55, %mm4, %mm2 - - pand %mm5, %mm0 - pand %mm5, %mm2 - pmullw %mm6, %mm0 - pmullw %mm6, %mm2 -#if SCALE_ADJUST > 0 - psrlw $SCALE_ADJUST, %mm0 - psrlw $SCALE_ADJUST, %mm2 -#endif - pmulhuw %mm7, %mm0 - pmulhuw %mm7, %mm2 - - por %mm3, %mm0 - por %mm3, %mm2 - - packuswb %mm2, %mm0 - - movq %mm0, (%edx) - addl $8, %edx - -.L04: - /* At this point there can be at most 1 pixel left to process. - * Process it if needed. - */ - - testl $0x01, %ecx - je .L01 - - movzwl (%eax), %ecx - movd %ecx, %mm4 - - pshufw $0x00, %mm4, %mm0 - - pand %mm5, %mm0 - pmullw %mm6, %mm0 -#if SCALE_ADJUST > 0 - psrlw $SCALE_ADJUST, %mm0 -#endif - pmulhuw %mm7, %mm0 - - por %mm3, %mm0 - - packuswb %mm0, %mm0 - - movd %mm0, (%edx) - -.L01: -#ifdef USE_INNER_EMMS - emms -#endif - ret -#endif /* !defined(__DJGPP__) && !defined(__MINGW32__) && !defined(__APPLE__) */ - -#if defined (__ELF__) && defined (__linux__) - .section .note.GNU-stack,"",%progbits -#endif +/* + * (C) Copyright IBM Corporation 2004 + * 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 + * on 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 + * IBM AND/OR THEIR 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. + */ + +/** + * \file read_rgba_span_x86.S + * Optimized routines to transfer pixel data from the framebuffer to a + * buffer in main memory. + * + * \author Ian Romanick + */ + + .file "read_rgba_span_x86.S" +#if !defined(__DJGPP__) && !defined(__MINGW32__) && !defined(__APPLE__) /* this one cries for assyntax.h */ +/* Kevin F. Quinn 2nd July 2006 + * Replaced data segment constants with text-segment instructions. + */ +#define LOAD_MASK(mvins,m1,m2) \ + pushl $0xff00ff00 ;\ + pushl $0xff00ff00 ;\ + pushl $0xff00ff00 ;\ + pushl $0xff00ff00 ;\ + mvins (%esp), m1 ;\ + pushl $0x00ff0000 ;\ + pushl $0x00ff0000 ;\ + pushl $0x00ff0000 ;\ + pushl $0x00ff0000 ;\ + mvins (%esp), m2 ;\ + addl $32, %esp + +/* I implemented these as macros because they appear in several places, + * and I've tweaked them a number of times. I got tired of changing every + * place they appear. :) + */ + +#define DO_ONE_PIXEL() \ + movl (%ebx), %eax ; \ + addl $4, %ebx ; \ + bswap %eax /* ARGB -> BGRA */ ; \ + rorl $8, %eax /* BGRA -> ABGR */ ; \ + movl %eax, (%ecx) /* ABGR -> R, G, B, A */ ; \ + addl $4, %ecx + +#define DO_ONE_LAST_PIXEL() \ + movl (%ebx), %eax ; \ + bswap %eax /* ARGB -> BGRA */ ; \ + rorl $8, %eax /* BGRA -> ABGR */ ; \ + movl %eax, (%ecx) /* ABGR -> R, G, B, A */ ; \ + + +/** + * MMX optimized version of the BGRA8888_REV to RGBA copy routine. + * + * \warning + * This function assumes that the caller will issue the EMMS instruction + * at the correct places. + */ + +.globl _generic_read_RGBA_span_BGRA8888_REV_MMX +#ifndef USE_DRICORE +.hidden _generic_read_RGBA_span_BGRA8888_REV_MMX +#endif + .type _generic_read_RGBA_span_BGRA8888_REV_MMX, @function +_generic_read_RGBA_span_BGRA8888_REV_MMX: + pushl %ebx + +#ifdef USE_INNER_EMMS + emms +#endif + LOAD_MASK(movq,%mm1,%mm2) + + movl 8(%esp), %ebx /* source pointer */ + movl 16(%esp), %edx /* number of pixels to copy */ + movl 12(%esp), %ecx /* destination pointer */ + + testl %edx, %edx + jle .L20 /* Bail if there's nothing to do. */ + + movl %ebx, %eax + + negl %eax + sarl $2, %eax + andl $1, %eax + je .L17 + + subl %eax, %edx + DO_ONE_PIXEL() +.L17: + + /* Would it be faster to unroll this loop once and process 4 pixels + * per pass, instead of just two? + */ + + movl %edx, %eax + shrl %eax + jmp .L18 +.L19: + movq (%ebx), %mm0 + addl $8, %ebx + + /* These 9 instructions do what PSHUFB (if there were such an + * instruction) could do in 1. :( + */ + + movq %mm0, %mm3 + movq %mm0, %mm4 + + pand %mm2, %mm3 + psllq $16, %mm4 + psrlq $16, %mm3 + pand %mm2, %mm4 + + pand %mm1, %mm0 + por %mm4, %mm3 + por %mm3, %mm0 + + movq %mm0, (%ecx) + addl $8, %ecx + subl $1, %eax +.L18: + jne .L19 + +#ifdef USE_INNER_EMMS + emms +#endif + + /* At this point there are either 1 or 0 pixels remaining to be + * converted. Convert the last pixel, if needed. + */ + + testl $1, %edx + je .L20 + + DO_ONE_LAST_PIXEL() + +.L20: + popl %ebx + ret + .size _generic_read_RGBA_span_BGRA8888_REV_MMX, .-_generic_read_RGBA_span_BGRA8888_REV_MMX + + +/** + * SSE optimized version of the BGRA8888_REV to RGBA copy routine. SSE + * instructions are only actually used to read data from the framebuffer. + * In practice, the speed-up is pretty small. + * + * \todo + * Do some more testing and determine if there's any reason to have this + * function in addition to the MMX version. + * + * \warning + * This function assumes that the caller will issue the EMMS instruction + * at the correct places. + */ + +.globl _generic_read_RGBA_span_BGRA8888_REV_SSE +#ifndef USE_DRICORE +.hidden _generic_read_RGBA_span_BGRA8888_REV_SSE +#endif + .type _generic_read_RGBA_span_BGRA8888_REV_SSE, @function +_generic_read_RGBA_span_BGRA8888_REV_SSE: + pushl %esi + pushl %ebx + pushl %ebp + +#ifdef USE_INNER_EMMS + emms +#endif + + LOAD_MASK(movq,%mm1,%mm2) + + movl 16(%esp), %ebx /* source pointer */ + movl 24(%esp), %edx /* number of pixels to copy */ + movl 20(%esp), %ecx /* destination pointer */ + + testl %edx, %edx + jle .L35 /* Bail if there's nothing to do. */ + + movl %esp, %ebp + subl $16, %esp + andl $0xfffffff0, %esp + + movl %ebx, %eax + movl %edx, %esi + + negl %eax + andl $15, %eax + sarl $2, %eax + cmpl %edx, %eax + cmovle %eax, %esi + + subl %esi, %edx + + testl $1, %esi + je .L32 + + DO_ONE_PIXEL() +.L32: + + testl $2, %esi + je .L31 + + movq (%ebx), %mm0 + addl $8, %ebx + + movq %mm0, %mm3 + movq %mm0, %mm4 + + pand %mm2, %mm3 + psllq $16, %mm4 + psrlq $16, %mm3 + pand %mm2, %mm4 + + pand %mm1, %mm0 + por %mm4, %mm3 + por %mm3, %mm0 + + movq %mm0, (%ecx) + addl $8, %ecx +.L31: + + movl %edx, %eax + shrl $2, %eax + jmp .L33 +.L34: + movaps (%ebx), %xmm0 + addl $16, %ebx + + /* This would be so much better if we could just move directly from + * an SSE register to an MMX register. Unfortunately, that + * functionality wasn't introduced until SSE2 with the MOVDQ2Q + * instruction. + */ + + movaps %xmm0, (%esp) + movq (%esp), %mm0 + movq 8(%esp), %mm5 + + movq %mm0, %mm3 + movq %mm0, %mm4 + movq %mm5, %mm6 + movq %mm5, %mm7 + + pand %mm2, %mm3 + pand %mm2, %mm6 + + psllq $16, %mm4 + psllq $16, %mm7 + + psrlq $16, %mm3 + psrlq $16, %mm6 + + pand %mm2, %mm4 + pand %mm2, %mm7 + + pand %mm1, %mm0 + pand %mm1, %mm5 + + por %mm4, %mm3 + por %mm7, %mm6 + + por %mm3, %mm0 + por %mm6, %mm5 + + movq %mm0, (%ecx) + movq %mm5, 8(%ecx) + addl $16, %ecx + + subl $1, %eax +.L33: + jne .L34 + +#ifdef USE_INNER_EMMS + emms +#endif + movl %ebp, %esp + + /* At this point there are either [0, 3] pixels remaining to be + * converted. + */ + + testl $2, %edx + je .L36 + + movq (%ebx), %mm0 + addl $8, %ebx + + movq %mm0, %mm3 + movq %mm0, %mm4 + + pand %mm2, %mm3 + psllq $16, %mm4 + psrlq $16, %mm3 + pand %mm2, %mm4 + + pand %mm1, %mm0 + por %mm4, %mm3 + por %mm3, %mm0 + + movq %mm0, (%ecx) + addl $8, %ecx +.L36: + + testl $1, %edx + je .L35 + + DO_ONE_LAST_PIXEL() +.L35: + popl %ebp + popl %ebx + popl %esi + ret + .size _generic_read_RGBA_span_BGRA8888_REV_SSE, .-_generic_read_RGBA_span_BGRA8888_REV_SSE + + +/** + * SSE2 optimized version of the BGRA8888_REV to RGBA copy routine. + */ + + .text +.globl _generic_read_RGBA_span_BGRA8888_REV_SSE2 +#ifndef USE_DRICORE +.hidden _generic_read_RGBA_span_BGRA8888_REV_SSE2 +#endif + .type _generic_read_RGBA_span_BGRA8888_REV_SSE2, @function +_generic_read_RGBA_span_BGRA8888_REV_SSE2: + pushl %esi + pushl %ebx + + LOAD_MASK(movdqu,%xmm1,%xmm2) + + movl 12(%esp), %ebx /* source pointer */ + movl 20(%esp), %edx /* number of pixels to copy */ + movl 16(%esp), %ecx /* destination pointer */ + + movl %ebx, %eax + movl %edx, %esi + + testl %edx, %edx + jle .L46 /* Bail if there's nothing to do. */ + + /* If the source pointer isn't a multiple of 16 we have to process + * a few pixels the "slow" way to get the address aligned for + * the SSE fetch intsructions. + */ + + negl %eax + andl $15, %eax + sarl $2, %eax + + cmpl %edx, %eax + cmovbe %eax, %esi + subl %esi, %edx + + testl $1, %esi + je .L41 + + DO_ONE_PIXEL() +.L41: + testl $2, %esi + je .L40 + + movq (%ebx), %xmm0 + addl $8, %ebx + + movdqa %xmm0, %xmm3 + movdqa %xmm0, %xmm4 + andps %xmm1, %xmm0 + + andps %xmm2, %xmm3 + pslldq $2, %xmm4 + psrldq $2, %xmm3 + andps %xmm2, %xmm4 + + orps %xmm4, %xmm3 + orps %xmm3, %xmm0 + + movq %xmm0, (%ecx) + addl $8, %ecx +.L40: + + /* Would it be worth having a specialized version of this loop for + * the case where the destination is 16-byte aligned? That version + * would be identical except that it could use movedqa instead of + * movdqu. + */ + + movl %edx, %eax + shrl $2, %eax + jmp .L42 +.L43: + movdqa (%ebx), %xmm0 + addl $16, %ebx + + movdqa %xmm0, %xmm3 + movdqa %xmm0, %xmm4 + andps %xmm1, %xmm0 + + andps %xmm2, %xmm3 + pslldq $2, %xmm4 + psrldq $2, %xmm3 + andps %xmm2, %xmm4 + + orps %xmm4, %xmm3 + orps %xmm3, %xmm0 + + movdqu %xmm0, (%ecx) + addl $16, %ecx + subl $1, %eax +.L42: + jne .L43 + + + /* There may be upto 3 pixels remaining to be copied. Take care + * of them now. We do the 2 pixel case first because the data + * will be aligned. + */ + + testl $2, %edx + je .L47 + + movq (%ebx), %xmm0 + addl $8, %ebx + + movdqa %xmm0, %xmm3 + movdqa %xmm0, %xmm4 + andps %xmm1, %xmm0 + + andps %xmm2, %xmm3 + pslldq $2, %xmm4 + psrldq $2, %xmm3 + andps %xmm2, %xmm4 + + orps %xmm4, %xmm3 + orps %xmm3, %xmm0 + + movq %xmm0, (%ecx) + addl $8, %ecx +.L47: + + testl $1, %edx + je .L46 + + DO_ONE_LAST_PIXEL() +.L46: + + popl %ebx + popl %esi + ret + .size _generic_read_RGBA_span_BGRA8888_REV_SSE2, .-_generic_read_RGBA_span_BGRA8888_REV_SSE2 + + + +#define MASK_565_L 0x07e0f800 +#define MASK_565_H 0x0000001f +/* Setting SCALE_ADJUST to 5 gives a perfect match with the + * classic C implementation in Mesa. Setting SCALE_ADJUST + * to 0 is slightly faster but at a small cost to accuracy. + */ +#define SCALE_ADJUST 5 +#if SCALE_ADJUST == 5 +#define PRESCALE_L 0x00100001 +#define PRESCALE_H 0x00000200 +#define SCALE_L 0x40C620E8 +#define SCALE_H 0x0000839d +#elif SCALE_ADJUST == 0 +#define PRESCALE_L 0x00200001 +#define PRESCALE_H 0x00000800 +#define SCALE_L 0x01040108 +#define SCALE_H 0x00000108 +#else +#error SCALE_ADJUST must either be 5 or 0. +#endif +#define ALPHA_L 0x00000000 +#define ALPHA_H 0x00ff0000 + +/** + * MMX optimized version of the RGB565 to RGBA copy routine. + */ + + .text + .globl _generic_read_RGBA_span_RGB565_MMX +#ifndef USE_DRICORE + .hidden _generic_read_RGBA_span_RGB565_MMX +#endif + .type _generic_read_RGBA_span_RGB565_MMX, @function + +_generic_read_RGBA_span_RGB565_MMX: + +#ifdef USE_INNER_EMMS + emms +#endif + + movl 4(%esp), %eax /* source pointer */ + movl 8(%esp), %edx /* destination pointer */ + movl 12(%esp), %ecx /* number of pixels to copy */ + + pushl $MASK_565_H + pushl $MASK_565_L + movq (%esp), %mm5 + pushl $PRESCALE_H + pushl $PRESCALE_L + movq (%esp), %mm6 + pushl $SCALE_H + pushl $SCALE_L + movq (%esp), %mm7 + pushl $ALPHA_H + pushl $ALPHA_L + movq (%esp), %mm3 + addl $32,%esp + + sarl $2, %ecx + jl .L01 /* Bail early if the count is negative. */ + jmp .L02 + +.L03: + /* Fetch 4 RGB565 pixels into %mm4. Distribute the first and + * second pixels into the four words of %mm0 and %mm2. + */ + + movq (%eax), %mm4 + addl $8, %eax + + pshufw $0x00, %mm4, %mm0 + pshufw $0x55, %mm4, %mm2 + + + /* Mask the pixels so that each word of each register contains only + * one color component. + */ + + pand %mm5, %mm0 + pand %mm5, %mm2 + + + /* Adjust the component values so that they are as small as possible, + * but large enough so that we can multiply them by an unsigned 16-bit + * number and get a value as large as 0x00ff0000. + */ + + pmullw %mm6, %mm0 + pmullw %mm6, %mm2 +#if SCALE_ADJUST > 0 + psrlw $SCALE_ADJUST, %mm0 + psrlw $SCALE_ADJUST, %mm2 +#endif + + /* Scale the input component values to be on the range + * [0, 0x00ff0000]. This it the real magic of the whole routine. + */ + + pmulhuw %mm7, %mm0 + pmulhuw %mm7, %mm2 + + + /* Always set the alpha value to 0xff. + */ + + por %mm3, %mm0 + por %mm3, %mm2 + + + /* Pack the 16-bit values to 8-bit values and store the converted + * pixel data. + */ + + packuswb %mm2, %mm0 + movq %mm0, (%edx) + addl $8, %edx + + pshufw $0xaa, %mm4, %mm0 + pshufw $0xff, %mm4, %mm2 + + pand %mm5, %mm0 + pand %mm5, %mm2 + pmullw %mm6, %mm0 + pmullw %mm6, %mm2 +#if SCALE_ADJUST > 0 + psrlw $SCALE_ADJUST, %mm0 + psrlw $SCALE_ADJUST, %mm2 +#endif + pmulhuw %mm7, %mm0 + pmulhuw %mm7, %mm2 + + por %mm3, %mm0 + por %mm3, %mm2 + + packuswb %mm2, %mm0 + + movq %mm0, (%edx) + addl $8, %edx + + subl $1, %ecx +.L02: + jne .L03 + + + /* At this point there can be at most 3 pixels left to process. If + * there is either 2 or 3 left, process 2. + */ + + movl 12(%esp), %ecx + testl $0x02, %ecx + je .L04 + + movd (%eax), %mm4 + addl $4, %eax + + pshufw $0x00, %mm4, %mm0 + pshufw $0x55, %mm4, %mm2 + + pand %mm5, %mm0 + pand %mm5, %mm2 + pmullw %mm6, %mm0 + pmullw %mm6, %mm2 +#if SCALE_ADJUST > 0 + psrlw $SCALE_ADJUST, %mm0 + psrlw $SCALE_ADJUST, %mm2 +#endif + pmulhuw %mm7, %mm0 + pmulhuw %mm7, %mm2 + + por %mm3, %mm0 + por %mm3, %mm2 + + packuswb %mm2, %mm0 + + movq %mm0, (%edx) + addl $8, %edx + +.L04: + /* At this point there can be at most 1 pixel left to process. + * Process it if needed. + */ + + testl $0x01, %ecx + je .L01 + + movzwl (%eax), %ecx + movd %ecx, %mm4 + + pshufw $0x00, %mm4, %mm0 + + pand %mm5, %mm0 + pmullw %mm6, %mm0 +#if SCALE_ADJUST > 0 + psrlw $SCALE_ADJUST, %mm0 +#endif + pmulhuw %mm7, %mm0 + + por %mm3, %mm0 + + packuswb %mm0, %mm0 + + movd %mm0, (%edx) + +.L01: +#ifdef USE_INNER_EMMS + emms +#endif + ret +#endif /* !defined(__DJGPP__) && !defined(__MINGW32__) && !defined(__APPLE__) */ + +#if defined (__ELF__) && defined (__linux__) + .section .note.GNU-stack,"",%progbits +#endif -- cgit v1.2.3