diff options
Diffstat (limited to 'mesalib/src')
45 files changed, 31565 insertions, 0 deletions
diff --git a/mesalib/src/mesa/program/Makefile b/mesalib/src/mesa/program/Makefile new file mode 100644 index 000000000..400a543bd --- /dev/null +++ b/mesalib/src/mesa/program/Makefile @@ -0,0 +1,7 @@ +all: program_parse.tab.c lex.yy.c + +program_parse.tab.c program_parse.tab.h: program_parse.y + bison -v -d $< + +lex.yy.c: program_lexer.l + flex --never-interactive $< diff --git a/mesalib/src/mesa/program/arbprogparse.c b/mesalib/src/mesa/program/arbprogparse.c new file mode 100644 index 000000000..f834aaf56 --- /dev/null +++ b/mesalib/src/mesa/program/arbprogparse.c @@ -0,0 +1,219 @@ +/* + * Mesa 3-D graphics library + * Version: 7.1 + * + * Copyright (C) 1999-2008 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. + */ + +#define DEBUG_PARSING 0 + +/** + * \file arbprogparse.c + * ARB_*_program parser core + * \author Karl Rasche + */ + +/** +Notes on program parameters, etc. + +The instructions we emit will use six kinds of source registers: + + PROGRAM_INPUT - input registers + PROGRAM_TEMPORARY - temp registers + PROGRAM_ADDRESS - address/indirect register + PROGRAM_SAMPLER - texture sampler + PROGRAM_CONSTANT - indexes into program->Parameters, a known constant/literal + PROGRAM_STATE_VAR - indexes into program->Parameters, and may actually be: + + a state variable, like "state.fog.color", or + + a pointer to a "program.local[k]" parameter, or + + a pointer to a "program.env[k]" parameter + +Basically, all the program.local[] and program.env[] values will get mapped +into the unified gl_program->Parameters array. This solves the problem of +having three separate program parameter arrays. +*/ + + +#include "main/glheader.h" +#include "main/imports.h" +#include "main/context.h" +#include "main/mtypes.h" +#include "arbprogparse.h" +#include "programopt.h" +#include "prog_parameter.h" +#include "prog_statevars.h" +#include "prog_instruction.h" +#include "program_parser.h" + + +void +_mesa_parse_arb_fragment_program(GLcontext* ctx, GLenum target, + const GLvoid *str, GLsizei len, + struct gl_fragment_program *program) +{ + struct gl_program prog; + struct asm_parser_state state; + GLuint i; + + ASSERT(target == GL_FRAGMENT_PROGRAM_ARB); + + memset(&prog, 0, sizeof(prog)); + memset(&state, 0, sizeof(state)); + state.prog = &prog; + + if (!_mesa_parse_arb_program(ctx, target, (const GLubyte*) str, len, + &state)) { + /* Error in the program. Just return. */ + return; + } + + if (program->Base.String != NULL) + free(program->Base.String); + + /* Copy the relevant contents of the arb_program struct into the + * fragment_program struct. + */ + program->Base.String = prog.String; + program->Base.NumInstructions = prog.NumInstructions; + program->Base.NumTemporaries = prog.NumTemporaries; + program->Base.NumParameters = prog.NumParameters; + program->Base.NumAttributes = prog.NumAttributes; + program->Base.NumAddressRegs = prog.NumAddressRegs; + program->Base.NumNativeInstructions = prog.NumNativeInstructions; + program->Base.NumNativeTemporaries = prog.NumNativeTemporaries; + program->Base.NumNativeParameters = prog.NumNativeParameters; + program->Base.NumNativeAttributes = prog.NumNativeAttributes; + program->Base.NumNativeAddressRegs = prog.NumNativeAddressRegs; + program->Base.NumAluInstructions = prog.NumAluInstructions; + program->Base.NumTexInstructions = prog.NumTexInstructions; + program->Base.NumTexIndirections = prog.NumTexIndirections; + program->Base.NumNativeAluInstructions = prog.NumAluInstructions; + program->Base.NumNativeTexInstructions = prog.NumTexInstructions; + program->Base.NumNativeTexIndirections = prog.NumTexIndirections; + program->Base.InputsRead = prog.InputsRead; + program->Base.OutputsWritten = prog.OutputsWritten; + program->Base.IndirectRegisterFiles = prog.IndirectRegisterFiles; + for (i = 0; i < MAX_TEXTURE_IMAGE_UNITS; i++) { + program->Base.TexturesUsed[i] = prog.TexturesUsed[i]; + if (prog.TexturesUsed[i]) + program->Base.SamplersUsed |= (1 << i); + } + program->Base.ShadowSamplers = prog.ShadowSamplers; + switch (state.option.Fog) { + case OPTION_FOG_EXP: program->FogOption = GL_EXP; break; + case OPTION_FOG_EXP2: program->FogOption = GL_EXP2; break; + case OPTION_FOG_LINEAR: program->FogOption = GL_LINEAR; break; + default: program->FogOption = GL_NONE; break; + } + program->OriginUpperLeft = state.option.OriginUpperLeft; + program->PixelCenterInteger = state.option.PixelCenterInteger; + + program->UsesKill = state.fragment.UsesKill; + + if (program->FogOption) + program->Base.InputsRead |= FRAG_BIT_FOGC; + + if (program->Base.Instructions) + free(program->Base.Instructions); + program->Base.Instructions = prog.Instructions; + + if (program->Base.Parameters) + _mesa_free_parameter_list(program->Base.Parameters); + program->Base.Parameters = prog.Parameters; + + /* Append fog instructions now if the program has "OPTION ARB_fog_exp" + * or similar. We used to leave this up to drivers, but it appears + * there's no hardware that wants to do fog in a discrete stage separate + * from the fragment shader. + */ + if (program->FogOption != GL_NONE) { + _mesa_append_fog_code(ctx, program); + program->FogOption = GL_NONE; + } + +#if DEBUG_FP + printf("____________Fragment program %u ________\n", program->Base.Id); + _mesa_print_program(&program->Base); +#endif +} + + + +/** + * Parse the vertex program string. If success, update the given + * vertex_program object with the new program. Else, leave the vertex_program + * object unchanged. + */ +void +_mesa_parse_arb_vertex_program(GLcontext *ctx, GLenum target, + const GLvoid *str, GLsizei len, + struct gl_vertex_program *program) +{ + struct gl_program prog; + struct asm_parser_state state; + + ASSERT(target == GL_VERTEX_PROGRAM_ARB); + + memset(&prog, 0, sizeof(prog)); + memset(&state, 0, sizeof(state)); + state.prog = &prog; + + if (!_mesa_parse_arb_program(ctx, target, (const GLubyte*) str, len, + &state)) { + _mesa_error(ctx, GL_INVALID_OPERATION, "glProgramString(bad program)"); + return; + } + + if (program->Base.String != NULL) + free(program->Base.String); + + /* Copy the relevant contents of the arb_program struct into the + * vertex_program struct. + */ + program->Base.String = prog.String; + program->Base.NumInstructions = prog.NumInstructions; + program->Base.NumTemporaries = prog.NumTemporaries; + program->Base.NumParameters = prog.NumParameters; + program->Base.NumAttributes = prog.NumAttributes; + program->Base.NumAddressRegs = prog.NumAddressRegs; + program->Base.NumNativeInstructions = prog.NumNativeInstructions; + program->Base.NumNativeTemporaries = prog.NumNativeTemporaries; + program->Base.NumNativeParameters = prog.NumNativeParameters; + program->Base.NumNativeAttributes = prog.NumNativeAttributes; + program->Base.NumNativeAddressRegs = prog.NumNativeAddressRegs; + program->Base.InputsRead = prog.InputsRead; + program->Base.OutputsWritten = prog.OutputsWritten; + program->Base.IndirectRegisterFiles = prog.IndirectRegisterFiles; + program->IsPositionInvariant = (state.option.PositionInvariant) + ? GL_TRUE : GL_FALSE; + + if (program->Base.Instructions) + free(program->Base.Instructions); + program->Base.Instructions = prog.Instructions; + + if (program->Base.Parameters) + _mesa_free_parameter_list(program->Base.Parameters); + program->Base.Parameters = prog.Parameters; + +#if DEBUG_VP + printf("____________Vertex program %u __________\n", program->Base.Id); + _mesa_print_program(&program->Base); +#endif +} diff --git a/mesalib/src/mesa/program/arbprogparse.h b/mesalib/src/mesa/program/arbprogparse.h new file mode 100644 index 000000000..980d39fb9 --- /dev/null +++ b/mesalib/src/mesa/program/arbprogparse.h @@ -0,0 +1,41 @@ +/* + * Mesa 3-D graphics library + * Version: 6.5 + * + * Copyright (C) 1999-2005 Brian Paul All Rights Reserved. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included + * in all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS + * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * BRIAN PAUL BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN + * AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN + * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. + */ + + +#ifndef ARBPROGPARSE_H +#define ARBPROGPARSE_H + +#include "main/mtypes.h" + +extern void +_mesa_parse_arb_vertex_program(GLcontext *ctx, GLenum target, + const GLvoid *str, GLsizei len, + struct gl_vertex_program *program); + +extern void +_mesa_parse_arb_fragment_program(GLcontext *ctx, GLenum target, + const GLvoid *str, GLsizei len, + struct gl_fragment_program *program); + +#endif diff --git a/mesalib/src/mesa/program/descrip.mms b/mesalib/src/mesa/program/descrip.mms new file mode 100644 index 000000000..59730020d --- /dev/null +++ b/mesalib/src/mesa/program/descrip.mms @@ -0,0 +1,93 @@ +# Makefile for core library for VMS +# contributed by Jouk Jansen joukj@hrem.nano.tudelft.nl +# Last revision : 29 September 2008 +.first + define gl [---.include.gl] + define math [-.math] + define swrast [-.swrast] + define array_cache [-.array_cache] + define glapi [-.glapi] + define main [-.main] + define shader [-.shader] + +.include [---]mms-config. + +##### MACROS ##### + +VPATH = RCS + +INCDIR = [---.include],[-.main],[-.glapi],[.slang] +LIBDIR = [---.lib] +CFLAGS = /include=($(INCDIR),[])/define=(PTHREADS=1,"__extension__=")/name=(as_is,short)/float=ieee/ieee=denorm + +SOURCES = \ + atifragshader.c \ + arbprogparse.c \ + arbprogram.c \ + nvfragparse.c \ + nvprogram.c \ + nvvertparse.c \ + program.c \ + programopt.c \ + prog_debug.c \ + prog_execute.c \ + prog_instruction.c \ + prog_parameter.c \ + prog_print.c \ + prog_cache.c \ + prog_statevars.c \ + shader_api.c prog_uniform.c + +OBJECTS = \ + atifragshader.obj,\ + arbprogparse.obj,\ + arbprogram.obj,\ + nvfragparse.obj,\ + nvprogram.obj,\ + nvvertparse.obj,\ + program.obj,\ + programopt.obj,\ + prog_debug.obj,\ + prog_execute.obj,\ + prog_instruction.obj,\ + prog_parameter.obj,\ + prog_print.obj,\ + prog_statevars.obj,\ + shader_api.obj,prog_uniform.obj,prog_cache.obj + +##### RULES ##### + +VERSION=Mesa V3.4 + +##### TARGETS ##### +all : + $(MMS)$(MMSQUALIFIERS) $(LIBDIR)$(GL_LIB) + set def [.slang] + $(MMS)$(MMSQUALIFIERS) + set def [-] + +# Make the library +$(LIBDIR)$(GL_LIB) : $(OBJECTS) + @ library $(LIBDIR)$(GL_LIB) $(OBJECTS) + +clean : + purge + delete *.obj;* + +atifragshader.obj : atifragshader.c +arbprogparse.obj : arbprogparse.c +arbprogram.obj : arbprogram.c +nvfragparse.obj : nvfragparse.c +nvprogram.obj : nvprogram.c +nvvertparse.obj : nvvertparse.c +program.obj : program.c +programopt. obj : programopt.c +prog_debug.obj : prog_debug.c +prog_execute.obj : prog_execute.c +prog_instruction.obj : prog_instruction.c +prog_parameter.obj : prog_parameter.c +prog_print.obj : prog_print.c +prog_statevars.obj : prog_statevars.c +shader_api.obj : shader_api.c +prog_uniform.obj : prog_uniform.c +prog_cache.obj : prog_cache.c diff --git a/mesalib/src/mesa/program/hash_table.c b/mesalib/src/mesa/program/hash_table.c new file mode 100644 index 000000000..f7ef366c1 --- /dev/null +++ b/mesalib/src/mesa/program/hash_table.c @@ -0,0 +1,190 @@ +/* + * Copyright © 2008 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 hash_table.c + * \brief Implementation of a generic, opaque hash table data type. + * + * \author Ian Romanick <ian.d.romanick@intel.com> + */ + +#include "main/imports.h" +#include "main/simple_list.h" +#include "hash_table.h" + +struct node { + struct node *next; + struct node *prev; +}; + +struct hash_table { + hash_func_t hash; + hash_compare_func_t compare; + + unsigned num_buckets; + struct node buckets[1]; +}; + + +struct hash_node { + struct node link; + const void *key; + void *data; +}; + + +struct hash_table * +hash_table_ctor(unsigned num_buckets, hash_func_t hash, + hash_compare_func_t compare) +{ + struct hash_table *ht; + unsigned i; + + + if (num_buckets < 16) { + num_buckets = 16; + } + + ht = malloc(sizeof(*ht) + ((num_buckets - 1) + * sizeof(ht->buckets[0]))); + if (ht != NULL) { + ht->hash = hash; + ht->compare = compare; + ht->num_buckets = num_buckets; + + for (i = 0; i < num_buckets; i++) { + make_empty_list(& ht->buckets[i]); + } + } + + return ht; +} + + +void +hash_table_dtor(struct hash_table *ht) +{ + hash_table_clear(ht); + free(ht); +} + + +void +hash_table_clear(struct hash_table *ht) +{ + struct node *node; + struct node *temp; + unsigned i; + + + for (i = 0; i < ht->num_buckets; i++) { + foreach_s(node, temp, & ht->buckets[i]) { + remove_from_list(node); + free(node); + } + + assert(is_empty_list(& ht->buckets[i])); + } +} + + +void * +hash_table_find(struct hash_table *ht, const void *key) +{ + const unsigned hash_value = (*ht->hash)(key); + const unsigned bucket = hash_value % ht->num_buckets; + struct node *node; + + foreach(node, & ht->buckets[bucket]) { + struct hash_node *hn = (struct hash_node *) node; + + if ((*ht->compare)(hn->key, key) == 0) { + return hn->data; + } + } + + return NULL; +} + + +void +hash_table_insert(struct hash_table *ht, void *data, const void *key) +{ + const unsigned hash_value = (*ht->hash)(key); + const unsigned bucket = hash_value % ht->num_buckets; + struct hash_node *node; + + node = calloc(1, sizeof(*node)); + + node->data = data; + node->key = key; + + insert_at_head(& ht->buckets[bucket], & node->link); +} + +void +hash_table_remove(struct hash_table *ht, const void *key) +{ + const unsigned hash_value = (*ht->hash)(key); + const unsigned bucket = hash_value % ht->num_buckets; + struct node *node; + + foreach(node, & ht->buckets[bucket]) { + struct hash_node *hn = (struct hash_node *) node; + + if ((*ht->compare)(hn->key, key) == 0) { + remove_from_list(node); + free(node); + return; + } + } +} + +unsigned +hash_table_string_hash(const void *key) +{ + const char *str = (const char *) key; + unsigned hash = 5381; + + + while (*str != '\0') { + hash = (hash * 33) + *str; + str++; + } + + return hash; +} + + +unsigned +hash_table_pointer_hash(const void *key) +{ + return (unsigned)((uintptr_t) key / sizeof(void *)); +} + + +int +hash_table_pointer_compare(const void *key1, const void *key2) +{ + return key1 == key2 ? 0 : 1; +} diff --git a/mesalib/src/mesa/program/hash_table.h b/mesalib/src/mesa/program/hash_table.h new file mode 100644 index 000000000..f1c4fdcd1 --- /dev/null +++ b/mesalib/src/mesa/program/hash_table.h @@ -0,0 +1,150 @@ +/* + * Copyright © 2008 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 hash_table.h + * \brief Implementation of a generic, opaque hash table data type. + * + * \author Ian Romanick <ian.d.romanick@intel.com> + */ + +#ifndef HASH_TABLE_H +#define HASH_TABLE_H + +struct hash_table; + +typedef unsigned (*hash_func_t)(const void *key); +typedef int (*hash_compare_func_t)(const void *key1, const void *key2); + +#ifdef __cplusplus +extern "C" { +#endif + +/** + * Hash table constructor + * + * Creates a hash table with the specified number of buckets. The supplied + * \c hash and \c compare routines are used when adding elements to the table + * and when searching for elements in the table. + * + * \param num_buckets Number of buckets (bins) in the hash table. + * \param hash Function used to compute hash value of input keys. + * \param compare Function used to compare keys. + */ +extern struct hash_table *hash_table_ctor(unsigned num_buckets, + hash_func_t hash, hash_compare_func_t compare); + + +/** + * Release all memory associated with a hash table + * + * \warning + * This function cannot release memory occupied either by keys or data. + */ +extern void hash_table_dtor(struct hash_table *ht); + + +/** + * Flush all entries from a hash table + * + * \param ht Table to be cleared of its entries. + */ +extern void hash_table_clear(struct hash_table *ht); + + +/** + * Search a hash table for a specific element + * + * \param ht Table to be searched + * \param key Key of the desired element + * + * \return + * The \c data value supplied to \c hash_table_insert when the element with + * the matching key was added. If no matching key exists in the table, + * \c NULL is returned. + */ +extern void *hash_table_find(struct hash_table *ht, const void *key); + + +/** + * Add an element to a hash table + */ +extern void hash_table_insert(struct hash_table *ht, void *data, + const void *key); + +/** + * Remove a specific element from a hash table. + */ +extern void hash_table_remove(struct hash_table *ht, const void *key); + +/** + * Compute hash value of a string + * + * Computes the hash value of a string using the DJB2 algorithm developed by + * Professor Daniel J. Bernstein. It was published on comp.lang.c once upon + * a time. I was unable to find the original posting in the archives. + * + * \param key Pointer to a NUL terminated string to be hashed. + * + * \sa hash_table_string_compare + */ +extern unsigned hash_table_string_hash(const void *key); + + +/** + * Compare two strings used as keys + * + * This is just a macro wrapper around \c strcmp. + * + * \sa hash_table_string_hash + */ +#define hash_table_string_compare ((hash_compare_func_t) strcmp) + + +/** + * Compute hash value of a pointer + * + * \param key Pointer to be used as a hash key + * + * \note + * The memory pointed to by \c key is \b never accessed. The value of \c key + * itself is used as the hash key + * + * \sa hash_table_pointer_compare + */ +unsigned +hash_table_pointer_hash(const void *key); + + +/** + * Compare two pointers used as keys + * + * \sa hash_table_pointer_hash + */ +int +hash_table_pointer_compare(const void *key1, const void *key2); + +#ifdef __cplusplus +} +#endif +#endif /* HASH_TABLE_H */ diff --git a/mesalib/src/mesa/program/ir_to_mesa.cpp b/mesalib/src/mesa/program/ir_to_mesa.cpp new file mode 100644 index 000000000..93b6c305f --- /dev/null +++ b/mesalib/src/mesa/program/ir_to_mesa.cpp @@ -0,0 +1,2965 @@ +/* + * 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 + * + * Translates the IR to ARB_fragment_program text if possible, + * printing the result + */ + +#include <stdio.h> +#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" +} + +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 = 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; + } + + int 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 talloc-based new need not call delete. It's + * easier to just talloc_free 'ctx' (or any of its ancestors). */ + static void* operator new(size_t size, void *ctx) + { + void *node; + + node = talloc_zero_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; + 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, int file, int index) + : file(file), index(index), var(var) + { + /* empty */ + } + + int 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; + + GLcontext *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); + + 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); + + GLboolean try_emit_mad(ir_expression *ir, + int mul_operand); + + int get_sampler_uniform_value(ir_dereference *deref); + + 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); + prog->InfoLog = talloc_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), + }; + + 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); +} + +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 (!(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); +} + +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: + 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; +} + +struct statevar_element { + const char *field; + int tokens[STATE_LENGTH]; + int swizzle; +}; + +static struct statevar_element gl_DepthRange_elements[] = { + {"near", {STATE_DEPTH_RANGE, 0, 0}, SWIZZLE_XXXX}, + {"far", {STATE_DEPTH_RANGE, 0, 0}, SWIZZLE_YYYY}, + {"diff", {STATE_DEPTH_RANGE, 0, 0}, SWIZZLE_ZZZZ}, +}; + +static struct statevar_element gl_ClipPlane_elements[] = { + {NULL, {STATE_CLIPPLANE, 0, 0}, SWIZZLE_XYZW} +}; + +static struct statevar_element gl_Point_elements[] = { + {"size", {STATE_POINT_SIZE}, SWIZZLE_XXXX}, + {"sizeMin", {STATE_POINT_SIZE}, SWIZZLE_YYYY}, + {"sizeMax", {STATE_POINT_SIZE}, SWIZZLE_ZZZZ}, + {"fadeThresholdSize", {STATE_POINT_SIZE}, SWIZZLE_WWWW}, + {"distanceConstantAttenuation", {STATE_POINT_ATTENUATION}, SWIZZLE_XXXX}, + {"distanceLinearAttenuation", {STATE_POINT_ATTENUATION}, SWIZZLE_YYYY}, + {"distanceQuadraticAttenuation", {STATE_POINT_ATTENUATION}, SWIZZLE_ZZZZ}, +}; + +static struct statevar_element gl_FrontMaterial_elements[] = { + {"emission", {STATE_MATERIAL, 0, STATE_EMISSION}, SWIZZLE_XYZW}, + {"ambient", {STATE_MATERIAL, 0, STATE_AMBIENT}, SWIZZLE_XYZW}, + {"diffuse", {STATE_MATERIAL, 0, STATE_DIFFUSE}, SWIZZLE_XYZW}, + {"specular", {STATE_MATERIAL, 0, STATE_SPECULAR}, SWIZZLE_XYZW}, + {"shininess", {STATE_MATERIAL, 0, STATE_SHININESS}, SWIZZLE_XXXX}, +}; + +static struct statevar_element gl_BackMaterial_elements[] = { + {"emission", {STATE_MATERIAL, 1, STATE_EMISSION}, SWIZZLE_XYZW}, + {"ambient", {STATE_MATERIAL, 1, STATE_AMBIENT}, SWIZZLE_XYZW}, + {"diffuse", {STATE_MATERIAL, 1, STATE_DIFFUSE}, SWIZZLE_XYZW}, + {"specular", {STATE_MATERIAL, 1, STATE_SPECULAR}, SWIZZLE_XYZW}, + {"shininess", {STATE_MATERIAL, 1, STATE_SHININESS}, SWIZZLE_XXXX}, +}; + +static struct statevar_element gl_LightSource_elements[] = { + {"ambient", {STATE_LIGHT, 0, STATE_AMBIENT}, SWIZZLE_XYZW}, + {"diffuse", {STATE_LIGHT, 0, STATE_DIFFUSE}, SWIZZLE_XYZW}, + {"specular", {STATE_LIGHT, 0, STATE_SPECULAR}, SWIZZLE_XYZW}, + {"position", {STATE_LIGHT, 0, STATE_POSITION}, SWIZZLE_XYZW}, + {"halfVector", {STATE_LIGHT, 0, STATE_HALF_VECTOR}, SWIZZLE_XYZW}, + {"spotDirection", {STATE_LIGHT, 0, STATE_SPOT_DIRECTION}, SWIZZLE_XYZW}, + {"spotCosCutoff", {STATE_LIGHT, 0, STATE_SPOT_DIRECTION}, SWIZZLE_WWWW}, + {"spotCutoff", {STATE_LIGHT, 0, STATE_SPOT_CUTOFF}, SWIZZLE_XXXX}, + {"spotExponent", {STATE_LIGHT, 0, STATE_ATTENUATION}, SWIZZLE_WWWW}, + {"constantAttenuation", {STATE_LIGHT, 0, STATE_ATTENUATION}, SWIZZLE_XXXX}, + {"linearAttenuation", {STATE_LIGHT, 0, STATE_ATTENUATION}, SWIZZLE_YYYY}, + {"quadraticAttenuation", {STATE_LIGHT, 0, STATE_ATTENUATION}, SWIZZLE_ZZZZ}, +}; + +static struct statevar_element gl_LightModel_elements[] = { + {"ambient", {STATE_LIGHTMODEL_AMBIENT, 0}, SWIZZLE_XYZW}, +}; + +static struct statevar_element gl_FrontLightModelProduct_elements[] = { + {"sceneColor", {STATE_LIGHTMODEL_SCENECOLOR, 0}, SWIZZLE_XYZW}, +}; + +static struct statevar_element gl_BackLightModelProduct_elements[] = { + {"sceneColor", {STATE_LIGHTMODEL_SCENECOLOR, 1}, SWIZZLE_XYZW}, +}; + +static struct statevar_element gl_FrontLightProduct_elements[] = { + {"ambient", {STATE_LIGHTPROD, 0, 0, STATE_AMBIENT}, SWIZZLE_XYZW}, + {"diffuse", {STATE_LIGHTPROD, 0, 0, STATE_DIFFUSE}, SWIZZLE_XYZW}, + {"specular", {STATE_LIGHTPROD, 0, 0, STATE_SPECULAR}, SWIZZLE_XYZW}, +}; + +static struct statevar_element gl_BackLightProduct_elements[] = { + {"ambient", {STATE_LIGHTPROD, 0, 1, STATE_AMBIENT}, SWIZZLE_XYZW}, + {"diffuse", {STATE_LIGHTPROD, 0, 1, STATE_DIFFUSE}, SWIZZLE_XYZW}, + {"specular", {STATE_LIGHTPROD, 0, 1, STATE_SPECULAR}, SWIZZLE_XYZW}, +}; + +static struct statevar_element gl_TextureEnvColor_elements[] = { + {NULL, {STATE_TEXENV_COLOR, 0}, SWIZZLE_XYZW}, +}; + +static struct statevar_element gl_EyePlaneS_elements[] = { + {NULL, {STATE_TEXGEN, 0, STATE_TEXGEN_EYE_S}, SWIZZLE_XYZW}, +}; + +static struct statevar_element gl_EyePlaneT_elements[] = { + {NULL, {STATE_TEXGEN, 0, STATE_TEXGEN_EYE_T}, SWIZZLE_XYZW}, +}; + +static struct statevar_element gl_EyePlaneR_elements[] = { + {NULL, {STATE_TEXGEN, 0, STATE_TEXGEN_EYE_R}, SWIZZLE_XYZW}, +}; + +static struct statevar_element gl_EyePlaneQ_elements[] = { + {NULL, {STATE_TEXGEN, 0, STATE_TEXGEN_EYE_Q}, SWIZZLE_XYZW}, +}; + +static struct statevar_element gl_ObjectPlaneS_elements[] = { + {NULL, {STATE_TEXGEN, 0, STATE_TEXGEN_OBJECT_S}, SWIZZLE_XYZW}, +}; + +static struct statevar_element gl_ObjectPlaneT_elements[] = { + {NULL, {STATE_TEXGEN, 0, STATE_TEXGEN_OBJECT_T}, SWIZZLE_XYZW}, +}; + +static struct statevar_element gl_ObjectPlaneR_elements[] = { + {NULL, {STATE_TEXGEN, 0, STATE_TEXGEN_OBJECT_R}, SWIZZLE_XYZW}, +}; + +static struct statevar_element gl_ObjectPlaneQ_elements[] = { + {NULL, {STATE_TEXGEN, 0, STATE_TEXGEN_OBJECT_Q}, SWIZZLE_XYZW}, +}; + +static struct statevar_element gl_Fog_elements[] = { + {"color", {STATE_FOG_COLOR}, SWIZZLE_XYZW}, + {"density", {STATE_FOG_PARAMS}, SWIZZLE_XXXX}, + {"start", {STATE_FOG_PARAMS}, SWIZZLE_YYYY}, + {"end", {STATE_FOG_PARAMS}, SWIZZLE_ZZZZ}, + {"scale", {STATE_FOG_PARAMS}, SWIZZLE_WWWW}, +}; + +static struct statevar_element gl_NormalScale_elements[] = { + {NULL, {STATE_NORMAL_SCALE}, SWIZZLE_XXXX}, +}; + +#define MATRIX(name, statevar, modifier) \ + static struct statevar_element name ## _elements[] = { \ + { NULL, { statevar, 0, 0, 0, modifier}, SWIZZLE_XYZW }, \ + { NULL, { statevar, 0, 1, 1, modifier}, SWIZZLE_XYZW }, \ + { NULL, { statevar, 0, 2, 2, modifier}, SWIZZLE_XYZW }, \ + { NULL, { statevar, 0, 3, 3, modifier}, SWIZZLE_XYZW }, \ + } + +MATRIX(gl_ModelViewMatrix, + STATE_MODELVIEW_MATRIX, STATE_MATRIX_TRANSPOSE); +MATRIX(gl_ModelViewMatrixInverse, + STATE_MODELVIEW_MATRIX, STATE_MATRIX_INVTRANS); +MATRIX(gl_ModelViewMatrixTranspose, + STATE_MODELVIEW_MATRIX, 0); +MATRIX(gl_ModelViewMatrixInverseTranspose, + STATE_MODELVIEW_MATRIX, STATE_MATRIX_INVERSE); + +MATRIX(gl_ProjectionMatrix, + STATE_PROJECTION_MATRIX, STATE_MATRIX_TRANSPOSE); +MATRIX(gl_ProjectionMatrixInverse, + STATE_PROJECTION_MATRIX, STATE_MATRIX_INVTRANS); +MATRIX(gl_ProjectionMatrixTranspose, + STATE_PROJECTION_MATRIX, 0); +MATRIX(gl_ProjectionMatrixInverseTranspose, + STATE_PROJECTION_MATRIX, STATE_MATRIX_INVERSE); + +MATRIX(gl_ModelViewProjectionMatrix, + STATE_MVP_MATRIX, STATE_MATRIX_TRANSPOSE); +MATRIX(gl_ModelViewProjectionMatrixInverse, + STATE_MVP_MATRIX, STATE_MATRIX_INVTRANS); +MATRIX(gl_ModelViewProjectionMatrixTranspose, + STATE_MVP_MATRIX, 0); +MATRIX(gl_ModelViewProjectionMatrixInverseTranspose, + STATE_MVP_MATRIX, STATE_MATRIX_INVERSE); + +MATRIX(gl_TextureMatrix, + STATE_TEXTURE_MATRIX, STATE_MATRIX_TRANSPOSE); +MATRIX(gl_TextureMatrixInverse, + STATE_TEXTURE_MATRIX, STATE_MATRIX_INVTRANS); +MATRIX(gl_TextureMatrixTranspose, + STATE_TEXTURE_MATRIX, 0); +MATRIX(gl_TextureMatrixInverseTranspose, + STATE_TEXTURE_MATRIX, STATE_MATRIX_INVERSE); + +static struct statevar_element gl_NormalMatrix_elements[] = { + { NULL, { STATE_MODELVIEW_MATRIX, 0, 0, 0, STATE_MATRIX_INVERSE}, + SWIZZLE_XYZW }, + { NULL, { STATE_MODELVIEW_MATRIX, 0, 1, 1, STATE_MATRIX_INVERSE}, + SWIZZLE_XYZW }, + { NULL, { STATE_MODELVIEW_MATRIX, 0, 2, 2, STATE_MATRIX_INVERSE}, + SWIZZLE_XYZW }, +}; + +#undef MATRIX + +#define STATEVAR(name) {#name, name ## _elements, Elements(name ## _elements)} + +static const struct statevar { + const char *name; + struct statevar_element *elements; + unsigned int num_elements; +} statevars[] = { + STATEVAR(gl_DepthRange), + STATEVAR(gl_ClipPlane), + STATEVAR(gl_Point), + STATEVAR(gl_FrontMaterial), + STATEVAR(gl_BackMaterial), + STATEVAR(gl_LightSource), + STATEVAR(gl_LightModel), + STATEVAR(gl_FrontLightModelProduct), + STATEVAR(gl_BackLightModelProduct), + STATEVAR(gl_FrontLightProduct), + STATEVAR(gl_BackLightProduct), + STATEVAR(gl_TextureEnvColor), + STATEVAR(gl_EyePlaneS), + STATEVAR(gl_EyePlaneT), + STATEVAR(gl_EyePlaneR), + STATEVAR(gl_EyePlaneQ), + STATEVAR(gl_ObjectPlaneS), + STATEVAR(gl_ObjectPlaneT), + STATEVAR(gl_ObjectPlaneR), + STATEVAR(gl_ObjectPlaneQ), + STATEVAR(gl_Fog), + + STATEVAR(gl_ModelViewMatrix), + STATEVAR(gl_ModelViewMatrixInverse), + STATEVAR(gl_ModelViewMatrixTranspose), + STATEVAR(gl_ModelViewMatrixInverseTranspose), + + STATEVAR(gl_ProjectionMatrix), + STATEVAR(gl_ProjectionMatrixInverse), + STATEVAR(gl_ProjectionMatrixTranspose), + STATEVAR(gl_ProjectionMatrixInverseTranspose), + + STATEVAR(gl_ModelViewProjectionMatrix), + STATEVAR(gl_ModelViewProjectionMatrixInverse), + STATEVAR(gl_ModelViewProjectionMatrixTranspose), + STATEVAR(gl_ModelViewProjectionMatrixInverseTranspose), + + STATEVAR(gl_TextureMatrix), + STATEVAR(gl_TextureMatrixInverse), + STATEVAR(gl_TextureMatrixTranspose), + STATEVAR(gl_TextureMatrixInverseTranspose), + + STATEVAR(gl_NormalMatrix), + STATEVAR(gl_NormalScale), +}; + +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; + } + + if (ir->mode == ir_var_uniform && strncmp(ir->name, "gl_", 3) == 0) { + unsigned int i; + + for (i = 0; i < Elements(statevars); i++) { + if (strcmp(ir->name, statevars[i].name) == 0) + break; + } + + if (i == Elements(statevars)) { + fail_link(this->shader_program, + "Failed to find builtin uniform `%s'\n", ir->name); + return; + } + + const struct statevar *statevar = &statevars[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 statevar_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; +} + +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::visit(ir_expression *ir) +{ + unsigned int operand; + struct ir_to_mesa_src_reg op[2]; + struct ir_to_mesa_src_reg result_src; + struct ir_to_mesa_dst_reg result_dst; + const glsl_type *vec4_type = glsl_type::get_instance(GLSL_TYPE_FLOAT, 4, 1); + const glsl_type *vec3_type = glsl_type::get_instance(GLSL_TYPE_FLOAT, 3, 1); + const glsl_type *vec2_type = glsl_type::get_instance(GLSL_TYPE_FLOAT, 2, 1); + + /* 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; + } + + 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_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]); + if (vector_elements == 4) + ir_to_mesa_emit_op2(ir, OPCODE_DP4, result_dst, temp, temp); + else if (vector_elements == 3) + ir_to_mesa_emit_op2(ir, OPCODE_DP3, result_dst, temp, temp); + else + ir_to_mesa_emit_op2(ir, OPCODE_DP2, result_dst, temp, temp); + 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]); + if (vector_elements == 4) + ir_to_mesa_emit_op2(ir, OPCODE_DP4, result_dst, temp, temp); + else if (vector_elements == 3) + ir_to_mesa_emit_op2(ir, OPCODE_DP3, result_dst, temp, temp); + else + ir_to_mesa_emit_op2(ir, OPCODE_DP2, result_dst, temp, temp); + 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: + switch (ir->operands[0]->type->vector_elements) { + case 4: + ir_to_mesa_emit_op2(ir, OPCODE_DP4, result_dst, op[0], op[0]); + break; + case 3: + ir_to_mesa_emit_op2(ir, OPCODE_DP3, result_dst, op[0], op[0]); + break; + case 2: + ir_to_mesa_emit_op2(ir, OPCODE_DP2, result_dst, op[0], op[0]); + break; + default: + assert(!"unreached: ir_unop_any of non-bvec"); + break; + } + 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: + if (ir->operands[0]->type == vec4_type) { + assert(ir->operands[1]->type == vec4_type); + ir_to_mesa_emit_op2(ir, OPCODE_DP4, + result_dst, + op[0], op[1]); + } else if (ir->operands[0]->type == vec3_type) { + assert(ir->operands[1]->type == vec3_type); + ir_to_mesa_emit_op2(ir, OPCODE_DP3, + result_dst, + op[0], op[1]); + } else if (ir->operands[0]->type == vec2_type) { + assert(ir->operands[1]->type == vec2_type); + ir_to_mesa_emit_op2(ir, OPCODE_DP2, + result_dst, + op[0], op[1]); + } + break; + + case ir_binop_cross: + ir_to_mesa_emit_op2(ir, OPCODE_XPD, result_dst, op[0], op[1]); + 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: + assert(!"GLSL 1.30 features unsupported"); + 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); + + if (!entry) { + switch (ir->var->mode) { + case ir_var_uniform: + entry = new(mem_ctx) variable_storage(ir->var, PROGRAM_UNIFORM, + ir->var->location); + this->variables.push_tail(entry); + break; + case ir_var_in: + case ir_var_out: + 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(ir->var->location != -1); + if (ir->var->mode == ir_var_in || + ir->var->mode == ir_var_inout) { + entry = new(mem_ctx) variable_storage(ir->var, + PROGRAM_INPUT, + ir->var->location); + + if (this->prog->Target == GL_VERTEX_PROGRAM_ARB && + ir->var->location >= VERT_ATTRIB_GENERIC0) { + _mesa_add_attribute(prog->Attributes, + ir->var->name, + _mesa_sizeof_glsl_type(ir->var->type->gl_type), + ir->var->type->gl_type, + ir->var->location - VERT_ATTRIB_GENERIC0); + } + } else { + entry = new(mem_ctx) variable_storage(ir->var, + PROGRAM_OUTPUT, + ir->var->location); + } + + break; + case ir_var_auto: + case ir_var_temporary: + entry = new(mem_ctx) variable_storage(ir->var, PROGRAM_TEMPORARY, + this->next_temp); + this->variables.push_tail(entry); + + next_temp += type_size(ir->var->type); + break; + } + + if (!entry) { + printf("Failed to make storage for %s\n", ir->var->name); + exit(1); + } + } + + this->result = ir_to_mesa_src_reg(entry->file, entry->index, ir->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 = talloc(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); + } + this->result.swizzle = swizzle_for_size(ir->type->vector_elements); + 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); +} + +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) { + ir_to_mesa_src_reg condition; + + ir->condition->accept(this); + condition = this->result; + + /* 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 computing the condition. + */ + condition.negate = ~condition.negate; + for (i = 0; i < type_size(ir->lhs->type); i++) { + 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 = talloc(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; +} + +class get_sampler_name : public ir_hierarchical_visitor +{ +public: + get_sampler_name(ir_to_mesa_visitor *mesa, ir_dereference *last) + { + this->mem_ctx = mesa->mem_ctx; + this->mesa = mesa; + this->name = NULL; + this->offset = 0; + this->last = last; + } + + virtual ir_visitor_status visit(ir_dereference_variable *ir) + { + this->name = ir->var->name; + return visit_continue; + } + + virtual ir_visitor_status visit_leave(ir_dereference_record *ir) + { + this->name = talloc_asprintf(mem_ctx, "%s.%s", name, ir->field); + return visit_continue; + } + + virtual ir_visitor_status visit_leave(ir_dereference_array *ir) + { + ir_constant *index = ir->array_index->as_constant(); + int i; + + if (index) { + i = index->value.i[0]; + } else { + /* GLSL 1.10 and 1.20 allowed variable sampler array indices, + * while GLSL 1.30 requires that the array indices be + * constant integer expressions. We don't expect any driver + * to actually work with a really variable array index, so + * all that would work would be an unrolled loop counter that ends + * up being constant above. + */ + mesa->shader_program->InfoLog = + talloc_asprintf_append(mesa->shader_program->InfoLog, + "warning: Variable sampler array index " + "unsupported.\nThis feature of the language " + "was removed in GLSL 1.20 and is unlikely " + "to be supported for 1.10 in Mesa.\n"); + i = 0; + } + if (ir != last) { + this->name = talloc_asprintf(mem_ctx, "%s[%d]", name, i); + } else { + offset = i; + } + return visit_continue; + } + + ir_to_mesa_visitor *mesa; + const char *name; + void *mem_ctx; + int offset; + ir_dereference *last; +}; + +int +ir_to_mesa_visitor::get_sampler_uniform_value(ir_dereference *sampler) +{ + get_sampler_name getname(this, sampler); + + sampler->accept(&getname); + + GLint index = _mesa_lookup_parameter_index(prog->Parameters, -1, + getname.name); + + if (index < 0) { + fail_link(this->shader_program, + "failed to find sampler named %s.\n", getname.name); + return 0; + } + + index += getname.offset; + + return this->prog->Parameters->ParameterValues[index][0]; +} + +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 = get_sampler_uniform_value(ir->sampler); + + 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; + + assert(ir->condition == NULL); /* FINISHME */ + + 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 = talloc_new(NULL); +} + +ir_to_mesa_visitor::~ir_to_mesa_visitor() +{ + talloc_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) - 1); + 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 = talloc_zero_array(v->mem_ctx, int, if_count); + loop_stack = talloc_zero_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 = talloc_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); + } + } + } + + talloc_free(sorted_uniforms); +} + +static void +set_uniform_initializer(GLcontext *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 = talloc_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 = talloc_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(GLcontext *ctx, + struct gl_shader_program *shader_program) +{ + void *mem_ctx = NULL; + + for (unsigned int i = 0; i < shader_program->_NumLinkedShaders; i++) { + struct gl_shader *shader = shader_program->_LinkedShaders[i]; + 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 = talloc_new(NULL); + + set_uniform_initializer(ctx, mem_ctx, shader_program, var->name, + var->type, var->constant_value); + } + } + + talloc_free(mem_ctx); +} + +struct gl_program * +get_mesa_program(GLcontext *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; + 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 = talloc_array(v.mem_ctx, ir_instruction *, + num_instructions); + + mesa_inst = mesa_instructions; + i = 0; + foreach_iter(exec_list_iterator, iter, v.instructions) { + ir_to_mesa_instruction *inst = (ir_to_mesa_instruction *)iter.get(); + + mesa_inst->Opcode = inst->op; + mesa_inst->CondUpdate = inst->cond_update; + 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; + + 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++; + } + + 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" { +GLboolean +_mesa_ir_compile_shader(GLcontext *ctx, struct gl_shader *shader) +{ + assert(shader->CompileStatus); + (void) ctx; + + return GL_TRUE; +} + +GLboolean +_mesa_ir_link_shader(GLcontext *ctx, struct gl_shader_program *prog) +{ + assert(prog->LinkStatus); + + for (unsigned i = 0; i < prog->_NumLinkedShaders; i++) { + bool progress; + exec_list *ir = prog->_LinkedShaders[i]->ir; + 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); + do_mod_to_fract(ir); + do_div_to_mul_rcp(ir); + do_explog_to_explog2(ir); + + progress = do_lower_jumps(ir, true, true, options->EmitNoMainReturn, options->EmitNoCont, options->EmitNoLoops) || progress; + + progress = do_common_optimization(ir, true, options->MaxUnrollIterations) || progress; + + if (options->EmitNoIfs) + progress = do_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 < prog->_NumLinkedShaders; i++) { + struct gl_program *linked_prog; + bool ok = true; + + linked_prog = get_mesa_program(ctx, prog, prog->_LinkedShaders[i]); + + 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; + } + if (!ok) { + return GL_FALSE; + } + _mesa_reference_program(ctx, &linked_prog, NULL); + } + + return GL_TRUE; +} + +void +_mesa_glsl_compile_shader(GLcontext *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); + } + + talloc_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); + + talloc_free(state); + + if (shader->CompileStatus) { + if (!ctx->Driver.CompileShader(ctx, shader)) + shader->CompileStatus = GL_FALSE; + } +} + +void +_mesa_glsl_link_shader(GLcontext *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); + + 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/program/ir_to_mesa.h b/mesalib/src/mesa/program/ir_to_mesa.h new file mode 100644 index 000000000..ecaacde4b --- /dev/null +++ b/mesalib/src/mesa/program/ir_to_mesa.h @@ -0,0 +1,38 @@ +/* + * 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. + */ + +#ifdef __cplusplus +extern "C" { +#endif + +#include "main/config.h" +#include "main/mtypes.h" + +void _mesa_glsl_compile_shader(GLcontext *ctx, struct gl_shader *sh); +void _mesa_glsl_link_shader(GLcontext *ctx, struct gl_shader_program *prog); +GLboolean _mesa_ir_compile_shader(GLcontext *ctx, struct gl_shader *shader); +GLboolean _mesa_ir_link_shader(GLcontext *ctx, struct gl_shader_program *prog); + +#ifdef __cplusplus +} +#endif diff --git a/mesalib/src/mesa/program/lex.yy.c b/mesalib/src/mesa/program/lex.yy.c new file mode 100644 index 000000000..135eca6fd --- /dev/null +++ b/mesalib/src/mesa/program/lex.yy.c @@ -0,0 +1,3681 @@ + +#line 3 "lex.yy.c" + +#define YY_INT_ALIGNED short int + +/* A lexical scanner generated by flex */ + +#define FLEX_SCANNER +#define YY_FLEX_MAJOR_VERSION 2 +#define YY_FLEX_MINOR_VERSION 5 +#define YY_FLEX_SUBMINOR_VERSION 35 +#if YY_FLEX_SUBMINOR_VERSION > 0 +#define FLEX_BETA +#endif + +/* First, we deal with platform-specific or compiler-specific issues. */ + +/* begin standard C headers. */ +#include <stdio.h> +#include <string.h> +#include <errno.h> +#include <stdlib.h> + +/* end standard C headers. */ + +/* flex integer type definitions */ + +#ifndef FLEXINT_H +#define FLEXINT_H + +/* C99 systems have <inttypes.h>. Non-C99 systems may or may not. */ + +#if defined (__STDC_VERSION__) && __STDC_VERSION__ >= 199901L + +/* C99 says to define __STDC_LIMIT_MACROS before including stdint.h, + * if you want the limit (max/min) macros for int types. + */ +#ifndef __STDC_LIMIT_MACROS +#define __STDC_LIMIT_MACROS 1 +#endif + +#include <inttypes.h> +typedef int8_t flex_int8_t; +typedef uint8_t flex_uint8_t; +typedef int16_t flex_int16_t; +typedef uint16_t flex_uint16_t; +typedef int32_t flex_int32_t; +typedef uint32_t flex_uint32_t; +#else +typedef signed char flex_int8_t; +typedef short int flex_int16_t; +typedef int flex_int32_t; +typedef unsigned char flex_uint8_t; +typedef unsigned short int flex_uint16_t; +typedef unsigned int flex_uint32_t; + +/* Limits of integral types. */ +#ifndef INT8_MIN +#define INT8_MIN (-128) +#endif +#ifndef INT16_MIN +#define INT16_MIN (-32767-1) +#endif +#ifndef INT32_MIN +#define INT32_MIN (-2147483647-1) +#endif +#ifndef INT8_MAX +#define INT8_MAX (127) +#endif +#ifndef INT16_MAX +#define INT16_MAX (32767) +#endif +#ifndef INT32_MAX +#define INT32_MAX (2147483647) +#endif +#ifndef UINT8_MAX +#define UINT8_MAX (255U) +#endif +#ifndef UINT16_MAX +#define UINT16_MAX (65535U) +#endif +#ifndef UINT32_MAX +#define UINT32_MAX (4294967295U) +#endif + +#endif /* ! C99 */ + +#endif /* ! FLEXINT_H */ + +#ifdef __cplusplus + +/* The "const" storage-class-modifier is valid. */ +#define YY_USE_CONST + +#else /* ! __cplusplus */ + +/* C99 requires __STDC__ to be defined as 1. */ +#if defined (__STDC__) + +#define YY_USE_CONST + +#endif /* defined (__STDC__) */ +#endif /* ! __cplusplus */ + +#ifdef YY_USE_CONST +#define yyconst const +#else +#define yyconst +#endif + +/* Returned upon end-of-file. */ +#define YY_NULL 0 + +/* Promotes a possibly negative, possibly signed char to an unsigned + * integer for use as an array index. If the signed char is negative, + * we want to instead treat it as an 8-bit unsigned char, hence the + * double cast. + */ +#define YY_SC_TO_UI(c) ((unsigned int) (unsigned char) c) + +/* An opaque pointer. */ +#ifndef YY_TYPEDEF_YY_SCANNER_T +#define YY_TYPEDEF_YY_SCANNER_T +typedef void* yyscan_t; +#endif + +/* For convenience, these vars (plus the bison vars far below) + are macros in the reentrant scanner. */ +#define yyin yyg->yyin_r +#define yyout yyg->yyout_r +#define yyextra yyg->yyextra_r +#define yyleng yyg->yyleng_r +#define yytext yyg->yytext_r +#define yylineno (YY_CURRENT_BUFFER_LVALUE->yy_bs_lineno) +#define yycolumn (YY_CURRENT_BUFFER_LVALUE->yy_bs_column) +#define yy_flex_debug yyg->yy_flex_debug_r + +/* Enter a start condition. This macro really ought to take a parameter, + * but we do it the disgusting crufty way forced on us by the ()-less + * definition of BEGIN. + */ +#define BEGIN yyg->yy_start = 1 + 2 * + +/* Translate the current start state into a value that can be later handed + * to BEGIN to return to the state. The YYSTATE alias is for lex + * compatibility. + */ +#define YY_START ((yyg->yy_start - 1) / 2) +#define YYSTATE YY_START + +/* Action number for EOF rule of a given start state. */ +#define YY_STATE_EOF(state) (YY_END_OF_BUFFER + state + 1) + +/* Special action meaning "start processing a new file". */ +#define YY_NEW_FILE yyrestart(yyin ,yyscanner ) + +#define YY_END_OF_BUFFER_CHAR 0 + +/* Size of default input buffer. */ +#ifndef YY_BUF_SIZE +#ifdef __ia64__ +/* On IA-64, the buffer size is 16k, not 8k. + * Moreover, YY_BUF_SIZE is 2*YY_READ_BUF_SIZE in the general case. + * Ditto for the __ia64__ case accordingly. + */ +#define YY_BUF_SIZE 32768 +#else +#define YY_BUF_SIZE 16384 +#endif /* __ia64__ */ +#endif + +/* The state buf must be large enough to hold one state per character in the main buffer. + */ +#define YY_STATE_BUF_SIZE ((YY_BUF_SIZE + 2) * sizeof(yy_state_type)) + +#ifndef YY_TYPEDEF_YY_BUFFER_STATE +#define YY_TYPEDEF_YY_BUFFER_STATE +typedef struct yy_buffer_state *YY_BUFFER_STATE; +#endif + +#define EOB_ACT_CONTINUE_SCAN 0 +#define EOB_ACT_END_OF_FILE 1 +#define EOB_ACT_LAST_MATCH 2 + + #define YY_LESS_LINENO(n) + +/* Return all but the first "n" matched characters back to the input stream. */ +#define yyless(n) \ + do \ + { \ + /* Undo effects of setting up yytext. */ \ + int yyless_macro_arg = (n); \ + YY_LESS_LINENO(yyless_macro_arg);\ + *yy_cp = yyg->yy_hold_char; \ + YY_RESTORE_YY_MORE_OFFSET \ + yyg->yy_c_buf_p = yy_cp = yy_bp + yyless_macro_arg - YY_MORE_ADJ; \ + YY_DO_BEFORE_ACTION; /* set up yytext again */ \ + } \ + while ( 0 ) + +#define unput(c) yyunput( c, yyg->yytext_ptr , yyscanner ) + +#ifndef YY_TYPEDEF_YY_SIZE_T +#define YY_TYPEDEF_YY_SIZE_T +typedef size_t yy_size_t; +#endif + +#ifndef YY_STRUCT_YY_BUFFER_STATE +#define YY_STRUCT_YY_BUFFER_STATE +struct yy_buffer_state + { + FILE *yy_input_file; + + char *yy_ch_buf; /* input buffer */ + char *yy_buf_pos; /* current position in input buffer */ + + /* Size of input buffer in bytes, not including room for EOB + * characters. + */ + yy_size_t yy_buf_size; + + /* Number of characters read into yy_ch_buf, not including EOB + * characters. + */ + int yy_n_chars; + + /* Whether we "own" the buffer - i.e., we know we created it, + * and can realloc() it to grow it, and should free() it to + * delete it. + */ + int yy_is_our_buffer; + + /* Whether this is an "interactive" input source; if so, and + * if we're using stdio for input, then we want to use getc() + * instead of fread(), to make sure we stop fetching input after + * each newline. + */ + int yy_is_interactive; + + /* Whether we're considered to be at the beginning of a line. + * If so, '^' rules will be active on the next match, otherwise + * not. + */ + int yy_at_bol; + + int yy_bs_lineno; /**< The line count. */ + int yy_bs_column; /**< The column count. */ + + /* Whether to try to fill the input buffer when we reach the + * end of it. + */ + int yy_fill_buffer; + + int yy_buffer_status; + +#define YY_BUFFER_NEW 0 +#define YY_BUFFER_NORMAL 1 + /* When an EOF's been seen but there's still some text to process + * then we mark the buffer as YY_EOF_PENDING, to indicate that we + * shouldn't try reading from the input source any more. We might + * still have a bunch of tokens to match, though, because of + * possible backing-up. + * + * When we actually see the EOF, we change the status to "new" + * (via yyrestart()), so that the user can continue scanning by + * just pointing yyin at a new input file. + */ +#define YY_BUFFER_EOF_PENDING 2 + + }; +#endif /* !YY_STRUCT_YY_BUFFER_STATE */ + +/* We provide macros for accessing buffer states in case in the + * future we want to put the buffer states in a more general + * "scanner state". + * + * Returns the top of the stack, or NULL. + */ +#define YY_CURRENT_BUFFER ( yyg->yy_buffer_stack \ + ? yyg->yy_buffer_stack[yyg->yy_buffer_stack_top] \ + : NULL) + +/* Same as previous macro, but useful when we know that the buffer stack is not + * NULL or when we need an lvalue. For internal use only. + */ +#define YY_CURRENT_BUFFER_LVALUE yyg->yy_buffer_stack[yyg->yy_buffer_stack_top] + +void yyrestart (FILE *input_file ,yyscan_t yyscanner ); +void yy_switch_to_buffer (YY_BUFFER_STATE new_buffer ,yyscan_t yyscanner ); +YY_BUFFER_STATE yy_create_buffer (FILE *file,int size ,yyscan_t yyscanner ); +void yy_delete_buffer (YY_BUFFER_STATE b ,yyscan_t yyscanner ); +void yy_flush_buffer (YY_BUFFER_STATE b ,yyscan_t yyscanner ); +void yypush_buffer_state (YY_BUFFER_STATE new_buffer ,yyscan_t yyscanner ); +void yypop_buffer_state (yyscan_t yyscanner ); + +static void yyensure_buffer_stack (yyscan_t yyscanner ); +static void yy_load_buffer_state (yyscan_t yyscanner ); +static void yy_init_buffer (YY_BUFFER_STATE b,FILE *file ,yyscan_t yyscanner ); + +#define YY_FLUSH_BUFFER yy_flush_buffer(YY_CURRENT_BUFFER ,yyscanner) + +YY_BUFFER_STATE yy_scan_buffer (char *base,yy_size_t size ,yyscan_t yyscanner ); +YY_BUFFER_STATE yy_scan_string (yyconst char *yy_str ,yyscan_t yyscanner ); +YY_BUFFER_STATE yy_scan_bytes (yyconst char *bytes,int len ,yyscan_t yyscanner ); + +void *yyalloc (yy_size_t ,yyscan_t yyscanner ); +void *yyrealloc (void *,yy_size_t ,yyscan_t yyscanner ); +void yyfree (void * ,yyscan_t yyscanner ); + +#define yy_new_buffer yy_create_buffer + +#define yy_set_interactive(is_interactive) \ + { \ + if ( ! YY_CURRENT_BUFFER ){ \ + yyensure_buffer_stack (yyscanner); \ + YY_CURRENT_BUFFER_LVALUE = \ + yy_create_buffer(yyin,YY_BUF_SIZE ,yyscanner); \ + } \ + YY_CURRENT_BUFFER_LVALUE->yy_is_interactive = is_interactive; \ + } + +#define yy_set_bol(at_bol) \ + { \ + if ( ! YY_CURRENT_BUFFER ){\ + yyensure_buffer_stack (yyscanner); \ + YY_CURRENT_BUFFER_LVALUE = \ + yy_create_buffer(yyin,YY_BUF_SIZE ,yyscanner); \ + } \ + YY_CURRENT_BUFFER_LVALUE->yy_at_bol = at_bol; \ + } + +#define YY_AT_BOL() (YY_CURRENT_BUFFER_LVALUE->yy_at_bol) + +/* Begin user sect3 */ + +#define yywrap(n) 1 +#define YY_SKIP_YYWRAP + +typedef unsigned char YY_CHAR; + +typedef int yy_state_type; + +#define yytext_ptr yytext_r + +static yy_state_type yy_get_previous_state (yyscan_t yyscanner ); +static yy_state_type yy_try_NUL_trans (yy_state_type current_state ,yyscan_t yyscanner); +static int yy_get_next_buffer (yyscan_t yyscanner ); +static void yy_fatal_error (yyconst char msg[] ,yyscan_t yyscanner ); + +/* Done after the current pattern has been matched and before the + * corresponding action - sets up yytext. + */ +#define YY_DO_BEFORE_ACTION \ + yyg->yytext_ptr = yy_bp; \ + yyleng = (size_t) (yy_cp - yy_bp); \ + yyg->yy_hold_char = *yy_cp; \ + *yy_cp = '\0'; \ + yyg->yy_c_buf_p = yy_cp; + +#define YY_NUM_RULES 170 +#define YY_END_OF_BUFFER 171 +/* This struct is not used in this scanner, + but its presence is necessary. */ +struct yy_trans_info + { + flex_int32_t yy_verify; + flex_int32_t yy_nxt; + }; +static yyconst flex_int16_t yy_accept[850] = + { 0, + 0, 0, 171, 169, 167, 166, 169, 169, 139, 165, + 141, 141, 141, 141, 139, 139, 139, 139, 139, 139, + 139, 139, 139, 139, 139, 139, 139, 139, 139, 139, + 139, 139, 139, 139, 139, 167, 0, 0, 168, 139, + 0, 140, 142, 162, 162, 0, 0, 0, 0, 162, + 0, 0, 0, 0, 0, 0, 0, 119, 163, 120, + 121, 153, 153, 153, 153, 0, 141, 0, 127, 128, + 129, 139, 139, 139, 139, 139, 139, 139, 139, 139, + 139, 139, 139, 139, 139, 139, 139, 139, 139, 139, + 139, 139, 139, 139, 139, 139, 139, 139, 139, 139, + + 139, 139, 139, 139, 139, 139, 139, 139, 139, 139, + 139, 139, 139, 139, 139, 139, 139, 139, 139, 139, + 139, 139, 139, 139, 139, 139, 0, 0, 0, 0, + 0, 0, 0, 0, 0, 161, 0, 0, 0, 0, + 0, 0, 0, 0, 0, 160, 160, 0, 0, 0, + 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, + 159, 159, 159, 0, 0, 0, 0, 0, 0, 0, + 0, 0, 0, 150, 150, 150, 151, 151, 152, 143, + 142, 143, 0, 144, 11, 12, 139, 13, 139, 139, + 14, 15, 139, 16, 17, 18, 19, 20, 21, 6, + + 22, 23, 24, 25, 26, 28, 27, 29, 30, 31, + 32, 33, 34, 35, 139, 139, 139, 139, 139, 40, + 41, 139, 42, 43, 44, 45, 46, 47, 48, 139, + 49, 50, 51, 52, 53, 54, 55, 139, 56, 57, + 58, 59, 139, 139, 64, 65, 139, 139, 139, 139, + 139, 139, 0, 0, 0, 0, 142, 0, 0, 0, + 0, 0, 0, 0, 0, 0, 0, 80, 81, 83, + 0, 158, 0, 0, 0, 0, 0, 0, 97, 0, + 0, 0, 0, 0, 0, 0, 0, 0, 0, 157, + 156, 156, 109, 0, 0, 0, 0, 0, 0, 0, + + 0, 0, 0, 147, 147, 148, 149, 0, 145, 11, + 11, 139, 12, 12, 12, 139, 139, 139, 139, 139, + 15, 15, 139, 130, 16, 16, 139, 17, 17, 139, + 18, 18, 139, 19, 19, 139, 20, 20, 139, 21, + 21, 139, 22, 22, 139, 24, 24, 139, 25, 25, + 139, 28, 28, 139, 27, 27, 139, 30, 30, 139, + 31, 31, 139, 32, 32, 139, 33, 33, 139, 34, + 34, 139, 35, 35, 139, 139, 139, 139, 36, 139, + 38, 139, 40, 40, 139, 41, 41, 139, 131, 42, + 42, 139, 43, 43, 139, 139, 45, 45, 139, 46, + + 46, 139, 47, 47, 139, 48, 48, 139, 139, 49, + 49, 139, 50, 50, 139, 51, 51, 139, 52, 52, + 139, 53, 53, 139, 54, 54, 139, 139, 10, 56, + 139, 57, 139, 58, 139, 59, 139, 60, 139, 62, + 139, 64, 64, 139, 139, 139, 139, 139, 139, 139, + 139, 0, 164, 0, 0, 0, 73, 74, 0, 0, + 0, 0, 0, 0, 0, 85, 0, 0, 0, 0, + 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, + 0, 0, 0, 0, 155, 0, 0, 0, 113, 0, + 115, 0, 0, 0, 0, 0, 0, 154, 146, 139, + + 139, 139, 4, 139, 139, 139, 139, 139, 139, 139, + 139, 139, 139, 139, 139, 139, 139, 139, 139, 139, + 139, 139, 139, 139, 139, 139, 9, 37, 39, 139, + 139, 139, 139, 139, 139, 139, 139, 139, 139, 139, + 139, 139, 139, 139, 139, 139, 139, 139, 139, 139, + 60, 139, 61, 62, 139, 63, 139, 139, 139, 139, + 139, 69, 139, 139, 0, 0, 0, 0, 0, 75, + 76, 0, 0, 0, 0, 84, 0, 0, 88, 91, + 0, 0, 0, 0, 0, 0, 0, 102, 103, 0, + 0, 0, 0, 108, 0, 0, 0, 0, 0, 0, + + 0, 0, 0, 0, 139, 139, 139, 139, 139, 139, + 5, 139, 139, 139, 139, 139, 139, 139, 139, 139, + 139, 139, 139, 139, 139, 139, 139, 139, 139, 139, + 7, 8, 139, 139, 139, 139, 139, 139, 139, 139, + 139, 139, 139, 139, 139, 139, 139, 139, 139, 139, + 139, 139, 139, 139, 61, 139, 139, 63, 139, 139, + 139, 139, 139, 70, 139, 66, 0, 0, 0, 0, + 124, 0, 0, 0, 0, 0, 0, 0, 0, 0, + 94, 0, 98, 99, 0, 101, 0, 0, 0, 0, + 0, 0, 0, 0, 0, 0, 117, 118, 0, 0, + + 125, 11, 3, 12, 135, 136, 139, 14, 15, 16, + 17, 18, 19, 20, 21, 22, 24, 25, 28, 27, + 30, 31, 32, 33, 34, 35, 40, 41, 42, 43, + 44, 45, 46, 47, 48, 139, 139, 139, 49, 50, + 51, 52, 53, 54, 55, 56, 57, 58, 59, 139, + 139, 139, 139, 64, 65, 139, 68, 126, 0, 0, + 71, 0, 77, 0, 0, 0, 86, 0, 0, 0, + 0, 0, 0, 100, 0, 0, 106, 93, 0, 0, + 0, 0, 0, 0, 122, 0, 139, 132, 133, 139, + 60, 139, 62, 139, 67, 0, 0, 0, 0, 79, + + 82, 87, 0, 0, 92, 0, 0, 0, 105, 0, + 0, 0, 0, 114, 116, 0, 139, 139, 61, 63, + 2, 1, 0, 78, 0, 90, 0, 96, 104, 0, + 0, 111, 112, 123, 139, 134, 0, 89, 0, 107, + 110, 139, 72, 95, 139, 139, 137, 138, 0 + } ; + +static yyconst flex_int32_t yy_ec[256] = + { 0, + 1, 1, 1, 1, 1, 1, 1, 1, 2, 3, + 1, 1, 4, 1, 1, 1, 1, 1, 1, 1, + 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, + 1, 2, 5, 1, 6, 7, 1, 1, 1, 1, + 1, 1, 8, 1, 8, 9, 1, 10, 11, 12, + 13, 14, 15, 15, 15, 15, 15, 1, 1, 1, + 1, 1, 1, 1, 16, 17, 18, 19, 20, 21, + 22, 23, 24, 7, 25, 26, 27, 28, 29, 30, + 31, 32, 33, 34, 35, 36, 37, 38, 39, 40, + 1, 1, 1, 1, 41, 1, 42, 43, 44, 45, + + 46, 47, 48, 49, 50, 51, 52, 53, 54, 55, + 56, 57, 58, 59, 60, 61, 62, 63, 64, 65, + 66, 67, 1, 1, 1, 1, 1, 1, 1, 1, + 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, + 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, + 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, + 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, + 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, + 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, + 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, + + 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, + 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, + 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, + 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, + 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, + 1, 1, 1, 1, 1 + } ; + +static yyconst flex_int32_t yy_meta[68] = + { 0, + 1, 1, 1, 1, 1, 1, 2, 1, 3, 2, + 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, + 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, + 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, + 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, + 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, + 2, 2, 2, 2, 2, 2, 2 + } ; + +static yyconst flex_int16_t yy_base[853] = + { 0, + 0, 0, 1299, 1300, 66, 1300, 1293, 1294, 0, 69, + 85, 128, 140, 152, 151, 58, 56, 63, 76, 1272, + 158, 160, 39, 163, 173, 189, 52, 1265, 76, 1235, + 1234, 1246, 1230, 1244, 1243, 105, 1272, 1284, 1300, 0, + 225, 1300, 218, 160, 157, 20, 123, 66, 119, 192, + 1244, 1230, 54, 162, 1228, 1240, 194, 1300, 200, 195, + 98, 227, 196, 231, 235, 293, 305, 316, 1300, 1300, + 1300, 1249, 1262, 1256, 223, 1245, 1248, 1244, 1259, 107, + 298, 1241, 1255, 246, 1241, 1254, 1245, 1258, 1235, 1246, + 1237, 182, 1238, 1229, 1238, 1229, 1228, 1229, 144, 1223, + + 1229, 1240, 1231, 1225, 1222, 1223, 1227, 289, 1236, 1223, + 302, 1230, 1217, 1231, 1207, 65, 315, 276, 1227, 1226, + 1202, 1187, 1182, 1199, 1175, 1180, 1206, 279, 1195, 293, + 1190, 342, 299, 1192, 1173, 317, 1183, 1179, 1174, 207, + 1180, 1166, 1182, 1179, 1170, 320, 324, 1172, 1161, 1175, + 1178, 1160, 1175, 1162, 1159, 1166, 284, 1174, 227, 288, + 327, 342, 345, 1151, 1168, 1169, 1162, 1144, 318, 1145, + 1167, 1158, 330, 341, 345, 349, 353, 357, 361, 1300, + 419, 430, 436, 442, 440, 441, 1191, 0, 1190, 1173, + 1163, 443, 1183, 444, 451, 468, 470, 472, 471, 0, + + 496, 0, 497, 498, 0, 499, 500, 0, 524, 525, + 526, 536, 537, 553, 1178, 1171, 1184, 354, 356, 561, + 563, 1165, 564, 565, 1157, 580, 590, 591, 592, 1178, + 593, 617, 618, 619, 629, 630, 1155, 1165, 330, 362, + 419, 483, 445, 364, 646, 1153, 1145, 1144, 1129, 1129, + 1128, 1127, 1170, 1142, 1130, 662, 669, 643, 1134, 487, + 1131, 1125, 1125, 1119, 1132, 1132, 1117, 1300, 1300, 1132, + 1120, 646, 1127, 135, 1124, 1130, 561, 1125, 1300, 1116, + 1123, 1122, 1125, 1111, 1110, 1114, 1109, 448, 1114, 650, + 653, 665, 1300, 1106, 1104, 1104, 1112, 1113, 1095, 670, + + 1100, 1106, 486, 579, 655, 661, 668, 726, 732, 1112, + 682, 1119, 1110, 688, 730, 1117, 1116, 1109, 1123, 1113, + 1104, 712, 1111, 0, 1102, 731, 1109, 1100, 733, 1107, + 1098, 734, 1105, 1096, 736, 1103, 1094, 737, 1101, 1092, + 738, 1099, 1090, 739, 1097, 1088, 740, 1095, 1086, 741, + 1093, 1084, 742, 1091, 1082, 743, 1089, 1080, 744, 1087, + 1078, 745, 1085, 1076, 746, 1083, 1074, 747, 1081, 1072, + 748, 1079, 1070, 749, 1077, 1080, 1073, 1080, 0, 1073, + 0, 1088, 1063, 750, 1070, 1061, 751, 1068, 0, 1059, + 752, 1066, 1057, 755, 1064, 1063, 1054, 758, 1061, 1052, + + 776, 1059, 1050, 777, 1057, 1048, 779, 1055, 1058, 1045, + 780, 1052, 1043, 782, 1050, 1041, 783, 1048, 1039, 784, + 1046, 1037, 785, 1044, 1035, 786, 1042, 1041, 0, 1032, + 1039, 1030, 1037, 1028, 1035, 1026, 1033, 787, 1032, 788, + 1047, 1022, 789, 1029, 1028, 1006, 1000, 1005, 1011, 994, + 1009, 424, 1300, 1008, 998, 1002, 1300, 1300, 992, 1001, + 987, 1004, 987, 990, 984, 1300, 985, 984, 981, 988, + 981, 989, 985, 995, 992, 974, 980, 987, 971, 970, + 988, 970, 982, 981, 1300, 980, 970, 974, 1300, 961, + 1300, 966, 966, 974, 957, 958, 968, 1300, 1300, 1000, + + 982, 998, 0, 798, 996, 996, 995, 994, 993, 992, + 991, 990, 989, 988, 987, 986, 985, 984, 983, 982, + 981, 980, 979, 978, 965, 958, 0, 0, 0, 975, + 974, 973, 972, 971, 970, 969, 968, 967, 945, 965, + 964, 963, 962, 961, 960, 959, 958, 957, 956, 955, + 929, 936, 793, 927, 934, 794, 950, 949, 918, 921, + 901, 0, 902, 895, 902, 901, 902, 894, 912, 1300, + 1300, 894, 892, 902, 895, 1300, 890, 907, 516, 1300, + 898, 882, 883, 892, 883, 882, 882, 1300, 881, 890, + 880, 896, 893, 1300, 892, 890, 879, 880, 876, 868, + + 875, 870, 871, 866, 892, 892, 890, 904, 903, 898, + 0, 886, 885, 884, 883, 882, 881, 880, 879, 878, + 877, 876, 875, 874, 873, 872, 871, 870, 869, 868, + 0, 0, 867, 866, 865, 864, 863, 862, 861, 860, + 859, 804, 858, 857, 856, 855, 854, 853, 852, 851, + 850, 849, 848, 865, 839, 846, 862, 836, 843, 841, + 840, 818, 818, 0, 825, 0, 859, 858, 807, 825, + 1300, 820, 815, 808, 804, 816, 806, 804, 800, 816, + 807, 806, 1300, 1300, 809, 1300, 804, 797, 786, 797, + 789, 793, 806, 801, 804, 786, 1300, 1300, 798, 787, + + 1300, 0, 0, 0, 0, 0, 826, 0, 0, 0, + 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, + 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, + 0, 0, 0, 0, 0, 814, 813, 802, 0, 0, + 0, 0, 0, 0, 0, 0, 0, 0, 0, 785, + 798, 779, 792, 0, 0, 656, 0, 0, 706, 702, + 1300, 649, 1300, 648, 648, 654, 1300, 637, 645, 610, + 612, 608, 608, 1300, 572, 583, 1300, 1300, 577, 573, + 560, 557, 542, 555, 1300, 539, 573, 0, 0, 572, + 0, 555, 0, 546, 0, 562, 551, 495, 479, 1300, + + 1300, 1300, 481, 481, 1300, 480, 443, 31, 1300, 141, + 166, 171, 186, 1300, 1300, 211, 236, 276, 0, 0, + 1300, 1300, 290, 1300, 325, 1300, 346, 1300, 1300, 343, + 341, 1300, 1300, 1300, 365, 0, 380, 1300, 371, 1300, + 1300, 486, 1300, 1300, 451, 458, 0, 0, 1300, 836, + 503, 839 + } ; + +static yyconst flex_int16_t yy_def[853] = + { 0, + 849, 1, 849, 849, 849, 849, 849, 850, 851, 849, + 849, 849, 849, 849, 851, 851, 851, 851, 851, 851, + 851, 851, 851, 851, 851, 851, 851, 851, 851, 851, + 851, 851, 851, 851, 851, 849, 849, 850, 849, 851, + 849, 849, 849, 849, 849, 849, 849, 849, 849, 849, + 849, 849, 849, 849, 849, 849, 849, 849, 849, 849, + 849, 849, 849, 849, 849, 852, 849, 849, 849, 849, + 849, 851, 851, 851, 851, 851, 851, 851, 851, 851, + 851, 851, 851, 851, 851, 851, 851, 851, 851, 851, + 851, 851, 851, 851, 851, 851, 851, 851, 851, 851, + + 851, 851, 851, 851, 851, 851, 851, 851, 851, 851, + 851, 851, 851, 851, 851, 851, 851, 851, 851, 851, + 851, 851, 851, 851, 851, 851, 849, 849, 849, 849, + 849, 849, 849, 849, 849, 849, 849, 849, 849, 849, + 849, 849, 849, 849, 849, 849, 849, 849, 849, 849, + 849, 849, 849, 849, 849, 849, 849, 849, 849, 849, + 849, 849, 849, 849, 849, 849, 849, 849, 849, 849, + 849, 849, 849, 849, 849, 849, 849, 849, 849, 849, + 849, 849, 849, 849, 851, 851, 851, 851, 851, 851, + 851, 851, 851, 851, 851, 851, 851, 851, 851, 851, + + 851, 851, 851, 851, 851, 851, 851, 851, 851, 851, + 851, 851, 851, 851, 851, 851, 851, 851, 851, 851, + 851, 851, 851, 851, 851, 851, 851, 851, 851, 851, + 851, 851, 851, 851, 851, 851, 851, 851, 851, 851, + 851, 851, 851, 851, 851, 851, 851, 851, 851, 851, + 851, 851, 849, 849, 849, 849, 849, 849, 849, 849, + 849, 849, 849, 849, 849, 849, 849, 849, 849, 849, + 849, 849, 849, 849, 849, 849, 849, 849, 849, 849, + 849, 849, 849, 849, 849, 849, 849, 849, 849, 849, + 849, 849, 849, 849, 849, 849, 849, 849, 849, 849, + + 849, 849, 849, 849, 849, 849, 849, 849, 849, 851, + 851, 851, 851, 851, 851, 851, 851, 851, 851, 851, + 851, 851, 851, 851, 851, 851, 851, 851, 851, 851, + 851, 851, 851, 851, 851, 851, 851, 851, 851, 851, + 851, 851, 851, 851, 851, 851, 851, 851, 851, 851, + 851, 851, 851, 851, 851, 851, 851, 851, 851, 851, + 851, 851, 851, 851, 851, 851, 851, 851, 851, 851, + 851, 851, 851, 851, 851, 851, 851, 851, 851, 851, + 851, 851, 851, 851, 851, 851, 851, 851, 851, 851, + 851, 851, 851, 851, 851, 851, 851, 851, 851, 851, + + 851, 851, 851, 851, 851, 851, 851, 851, 851, 851, + 851, 851, 851, 851, 851, 851, 851, 851, 851, 851, + 851, 851, 851, 851, 851, 851, 851, 851, 851, 851, + 851, 851, 851, 851, 851, 851, 851, 851, 851, 851, + 851, 851, 851, 851, 851, 851, 851, 851, 851, 851, + 851, 849, 849, 849, 849, 849, 849, 849, 849, 849, + 849, 849, 849, 849, 849, 849, 849, 849, 849, 849, + 849, 849, 849, 849, 849, 849, 849, 849, 849, 849, + 849, 849, 849, 849, 849, 849, 849, 849, 849, 849, + 849, 849, 849, 849, 849, 849, 849, 849, 849, 851, + + 851, 851, 851, 851, 851, 851, 851, 851, 851, 851, + 851, 851, 851, 851, 851, 851, 851, 851, 851, 851, + 851, 851, 851, 851, 851, 851, 851, 851, 851, 851, + 851, 851, 851, 851, 851, 851, 851, 851, 851, 851, + 851, 851, 851, 851, 851, 851, 851, 851, 851, 851, + 851, 851, 851, 851, 851, 851, 851, 851, 851, 851, + 851, 851, 851, 851, 849, 849, 849, 849, 849, 849, + 849, 849, 849, 849, 849, 849, 849, 849, 849, 849, + 849, 849, 849, 849, 849, 849, 849, 849, 849, 849, + 849, 849, 849, 849, 849, 849, 849, 849, 849, 849, + + 849, 849, 849, 849, 851, 851, 851, 851, 851, 851, + 851, 851, 851, 851, 851, 851, 851, 851, 851, 851, + 851, 851, 851, 851, 851, 851, 851, 851, 851, 851, + 851, 851, 851, 851, 851, 851, 851, 851, 851, 851, + 851, 851, 851, 851, 851, 851, 851, 851, 851, 851, + 851, 851, 851, 851, 851, 851, 851, 851, 851, 851, + 851, 851, 851, 851, 851, 851, 849, 849, 849, 849, + 849, 849, 849, 849, 849, 849, 849, 849, 849, 849, + 849, 849, 849, 849, 849, 849, 849, 849, 849, 849, + 849, 849, 849, 849, 849, 849, 849, 849, 849, 849, + + 849, 851, 851, 851, 851, 851, 851, 851, 851, 851, + 851, 851, 851, 851, 851, 851, 851, 851, 851, 851, + 851, 851, 851, 851, 851, 851, 851, 851, 851, 851, + 851, 851, 851, 851, 851, 851, 851, 851, 851, 851, + 851, 851, 851, 851, 851, 851, 851, 851, 851, 851, + 851, 851, 851, 851, 851, 851, 851, 851, 849, 849, + 849, 849, 849, 849, 849, 849, 849, 849, 849, 849, + 849, 849, 849, 849, 849, 849, 849, 849, 849, 849, + 849, 849, 849, 849, 849, 849, 851, 851, 851, 851, + 851, 851, 851, 851, 851, 849, 849, 849, 849, 849, + + 849, 849, 849, 849, 849, 849, 849, 849, 849, 849, + 849, 849, 849, 849, 849, 849, 851, 851, 851, 851, + 849, 849, 849, 849, 849, 849, 849, 849, 849, 849, + 849, 849, 849, 849, 851, 851, 849, 849, 849, 849, + 849, 851, 849, 849, 851, 851, 851, 851, 0, 849, + 849, 849 + } ; + +static yyconst flex_int16_t yy_nxt[1368] = + { 0, + 4, 5, 6, 5, 7, 8, 9, 4, 10, 11, + 12, 13, 14, 11, 11, 15, 9, 16, 17, 18, + 19, 9, 9, 9, 20, 21, 22, 9, 23, 24, + 9, 25, 26, 27, 28, 9, 9, 29, 9, 9, + 9, 9, 9, 9, 9, 9, 30, 9, 9, 9, + 9, 9, 9, 9, 9, 9, 31, 9, 32, 33, + 34, 9, 35, 9, 9, 9, 9, 36, 96, 36, + 41, 116, 137, 97, 80, 138, 829, 42, 43, 43, + 43, 43, 43, 43, 77, 81, 78, 119, 82, 117, + 83, 238, 79, 66, 67, 67, 67, 67, 67, 67, + + 84, 85, 239, 150, 68, 120, 36, 86, 36, 151, + 44, 45, 46, 47, 48, 49, 50, 51, 52, 141, + 142, 53, 54, 55, 56, 57, 58, 59, 60, 61, + 68, 143, 62, 63, 64, 65, 66, 67, 67, 67, + 67, 67, 67, 170, 194, 195, 69, 68, 66, 67, + 67, 67, 67, 67, 67, 218, 171, 219, 70, 68, + 66, 67, 67, 67, 67, 67, 67, 72, 139, 73, + 71, 68, 140, 68, 144, 92, 74, 145, 98, 88, + 467, 89, 75, 93, 76, 68, 90, 99, 94, 91, + 101, 100, 102, 103, 95, 468, 830, 68, 136, 133, + + 210, 133, 133, 152, 133, 104, 105, 133, 106, 107, + 108, 109, 110, 134, 111, 133, 112, 153, 133, 211, + 135, 831, 113, 114, 154, 115, 41, 43, 43, 43, + 43, 43, 43, 146, 147, 157, 832, 132, 165, 133, + 166, 161, 162, 167, 168, 833, 158, 163, 188, 159, + 133, 169, 160, 265, 189, 164, 834, 201, 133, 174, + 173, 175, 176, 132, 835, 266, 128, 129, 46, 47, + 48, 49, 172, 51, 52, 202, 285, 53, 54, 55, + 56, 57, 58, 130, 60, 61, 286, 243, 131, 244, + 173, 173, 173, 173, 177, 173, 173, 178, 179, 173, + + 173, 173, 181, 181, 181, 181, 181, 181, 228, 836, + 196, 197, 182, 66, 67, 67, 67, 67, 67, 67, + 198, 232, 229, 183, 68, 184, 184, 184, 184, 184, + 184, 240, 134, 241, 255, 233, 282, 287, 182, 135, + 258, 258, 283, 288, 242, 837, 258, 430, 164, 256, + 68, 257, 257, 257, 257, 257, 257, 258, 258, 258, + 261, 258, 258, 298, 258, 272, 258, 258, 258, 258, + 431, 258, 381, 299, 258, 258, 379, 838, 258, 432, + 440, 289, 258, 290, 258, 258, 291, 292, 380, 258, + 382, 839, 258, 303, 303, 303, 303, 840, 441, 841, + + 258, 842, 433, 258, 303, 303, 303, 303, 304, 303, + 303, 305, 306, 303, 303, 303, 303, 303, 303, 303, + 307, 303, 303, 303, 303, 303, 303, 303, 43, 43, + 43, 43, 43, 43, 843, 844, 434, 308, 132, 309, + 309, 309, 309, 309, 309, 184, 184, 184, 184, 184, + 184, 184, 184, 184, 184, 184, 184, 310, 313, 435, + 321, 325, 311, 314, 132, 322, 326, 438, 328, 847, + 565, 311, 315, 329, 322, 326, 848, 311, 314, 439, + 312, 316, 329, 323, 327, 331, 566, 334, 340, 337, + 332, 330, 335, 341, 338, 482, 845, 846, 483, 332, + + 436, 335, 341, 338, 40, 332, 828, 335, 333, 338, + 336, 342, 339, 343, 346, 349, 352, 355, 344, 347, + 350, 353, 356, 437, 827, 826, 825, 344, 347, 350, + 353, 356, 455, 824, 347, 350, 345, 348, 351, 354, + 357, 358, 361, 364, 823, 456, 359, 362, 365, 498, + 498, 498, 498, 367, 370, 359, 362, 365, 368, 371, + 822, 359, 362, 365, 360, 363, 366, 368, 371, 678, + 373, 821, 679, 368, 371, 374, 369, 372, 383, 820, + 386, 390, 393, 384, 374, 387, 391, 394, 819, 818, + 374, 817, 384, 375, 387, 391, 394, 397, 816, 815, + + 814, 385, 398, 388, 392, 395, 471, 400, 403, 406, + 410, 398, 401, 404, 407, 411, 813, 398, 812, 472, + 399, 401, 404, 407, 411, 811, 810, 401, 404, 407, + 402, 405, 408, 412, 413, 416, 419, 809, 808, 414, + 417, 420, 498, 498, 498, 498, 422, 425, 414, 417, + 420, 423, 426, 807, 414, 417, 420, 415, 418, 421, + 423, 426, 806, 442, 805, 804, 423, 426, 443, 424, + 427, 257, 257, 257, 257, 257, 257, 443, 257, 257, + 257, 257, 257, 257, 453, 453, 444, 453, 453, 803, + 453, 453, 453, 453, 453, 453, 802, 453, 801, 310, + + 453, 453, 800, 799, 453, 313, 485, 453, 453, 798, + 797, 453, 453, 492, 796, 493, 795, 494, 499, 498, + 498, 498, 312, 453, 498, 498, 498, 498, 316, 321, + 495, 498, 498, 498, 498, 309, 309, 309, 309, 309, + 309, 309, 309, 309, 309, 309, 309, 313, 325, 501, + 328, 331, 323, 334, 337, 340, 343, 346, 349, 352, + 355, 358, 361, 364, 367, 370, 373, 383, 386, 390, + 316, 327, 393, 330, 333, 397, 336, 339, 342, 345, + 348, 351, 354, 357, 360, 363, 366, 369, 372, 375, + 385, 388, 392, 400, 403, 395, 406, 410, 399, 413, + + 416, 419, 422, 425, 551, 554, 442, 794, 608, 609, + 655, 658, 793, 792, 736, 737, 402, 405, 791, 408, + 412, 790, 415, 418, 421, 424, 427, 552, 555, 444, + 610, 789, 788, 656, 659, 738, 38, 38, 38, 180, + 180, 787, 786, 785, 784, 783, 782, 781, 780, 779, + 778, 777, 776, 775, 774, 773, 772, 771, 770, 769, + 768, 767, 766, 765, 764, 763, 762, 761, 760, 759, + 758, 757, 756, 755, 754, 753, 659, 752, 751, 656, + 750, 749, 748, 747, 746, 745, 744, 743, 742, 741, + 740, 739, 735, 734, 733, 732, 731, 730, 729, 728, + + 727, 726, 725, 724, 723, 722, 721, 720, 719, 718, + 717, 716, 715, 714, 713, 712, 711, 710, 709, 708, + 707, 706, 705, 704, 703, 702, 701, 700, 699, 698, + 697, 696, 695, 694, 693, 692, 691, 690, 689, 688, + 687, 686, 685, 684, 683, 682, 681, 680, 677, 676, + 675, 674, 673, 672, 671, 670, 669, 668, 667, 666, + 665, 664, 663, 662, 661, 660, 657, 555, 654, 552, + 653, 652, 651, 650, 649, 648, 647, 646, 645, 644, + 643, 642, 641, 640, 639, 638, 637, 636, 635, 634, + 633, 632, 631, 630, 629, 628, 627, 626, 625, 624, + + 623, 622, 621, 620, 619, 618, 617, 616, 615, 614, + 613, 612, 611, 607, 606, 605, 604, 603, 602, 601, + 600, 599, 598, 597, 596, 595, 594, 593, 592, 591, + 590, 589, 588, 587, 586, 585, 584, 583, 582, 581, + 580, 579, 578, 577, 576, 575, 574, 573, 572, 571, + 570, 569, 568, 567, 564, 563, 562, 561, 560, 559, + 558, 557, 444, 556, 553, 550, 437, 549, 435, 548, + 433, 547, 431, 546, 545, 427, 544, 424, 543, 421, + 542, 418, 541, 415, 540, 412, 539, 538, 408, 537, + 405, 536, 402, 535, 399, 534, 533, 395, 532, 392, + + 531, 388, 530, 385, 529, 528, 527, 526, 525, 524, + 375, 523, 372, 522, 369, 521, 366, 520, 363, 519, + 360, 518, 357, 517, 354, 516, 351, 515, 348, 514, + 345, 513, 342, 512, 339, 511, 336, 510, 333, 509, + 330, 508, 327, 507, 323, 506, 505, 504, 503, 502, + 316, 500, 312, 497, 496, 491, 490, 489, 488, 487, + 486, 484, 481, 480, 479, 478, 477, 476, 475, 474, + 473, 470, 469, 466, 465, 464, 463, 462, 461, 460, + 459, 458, 457, 454, 289, 261, 452, 451, 450, 449, + 448, 447, 446, 445, 429, 428, 409, 396, 389, 378, + + 377, 376, 324, 320, 319, 318, 317, 302, 301, 300, + 297, 296, 295, 294, 293, 284, 281, 280, 279, 278, + 277, 276, 275, 274, 273, 271, 270, 269, 268, 267, + 264, 263, 262, 260, 259, 172, 254, 253, 252, 251, + 250, 249, 248, 247, 246, 245, 237, 236, 235, 234, + 231, 230, 227, 226, 225, 224, 223, 222, 221, 220, + 217, 216, 215, 214, 213, 212, 209, 208, 207, 206, + 205, 204, 203, 200, 199, 193, 192, 191, 190, 187, + 186, 185, 156, 155, 149, 148, 39, 127, 126, 125, + 124, 123, 122, 121, 118, 87, 39, 37, 849, 3, + + 849, 849, 849, 849, 849, 849, 849, 849, 849, 849, + 849, 849, 849, 849, 849, 849, 849, 849, 849, 849, + 849, 849, 849, 849, 849, 849, 849, 849, 849, 849, + 849, 849, 849, 849, 849, 849, 849, 849, 849, 849, + 849, 849, 849, 849, 849, 849, 849, 849, 849, 849, + 849, 849, 849, 849, 849, 849, 849, 849, 849, 849, + 849, 849, 849, 849, 849, 849, 849 + } ; + +static yyconst flex_int16_t yy_chk[1368] = + { 0, + 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, + 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, + 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, + 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, + 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, + 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, + 1, 1, 1, 1, 1, 1, 1, 5, 23, 5, + 10, 27, 46, 23, 17, 46, 808, 10, 10, 10, + 10, 10, 10, 10, 16, 17, 16, 29, 17, 27, + 18, 116, 16, 11, 11, 11, 11, 11, 11, 11, + + 18, 19, 116, 53, 11, 29, 36, 19, 36, 53, + 10, 10, 10, 10, 10, 10, 10, 10, 10, 48, + 48, 10, 10, 10, 10, 10, 10, 10, 10, 10, + 11, 48, 10, 10, 10, 10, 12, 12, 12, 12, + 12, 12, 12, 61, 80, 80, 12, 12, 13, 13, + 13, 13, 13, 13, 13, 99, 61, 99, 13, 13, + 14, 14, 14, 14, 14, 14, 14, 15, 47, 15, + 14, 14, 47, 12, 49, 22, 15, 49, 24, 21, + 274, 21, 15, 22, 15, 13, 21, 24, 22, 21, + 25, 24, 25, 25, 22, 274, 810, 14, 45, 45, + + 92, 44, 44, 54, 45, 25, 26, 44, 26, 26, + 26, 26, 26, 44, 26, 45, 26, 54, 44, 92, + 44, 811, 26, 26, 54, 26, 41, 43, 43, 43, + 43, 43, 43, 50, 50, 57, 812, 43, 60, 50, + 60, 59, 59, 60, 60, 813, 57, 59, 75, 57, + 50, 60, 57, 140, 75, 59, 816, 84, 59, 63, + 63, 63, 63, 43, 817, 140, 41, 41, 41, 41, + 41, 41, 62, 41, 41, 84, 159, 41, 41, 41, + 41, 41, 41, 41, 41, 41, 159, 118, 41, 118, + 62, 62, 62, 62, 64, 64, 64, 64, 65, 65, + + 65, 65, 66, 66, 66, 66, 66, 66, 108, 818, + 81, 81, 66, 67, 67, 67, 67, 67, 67, 67, + 81, 111, 108, 68, 67, 68, 68, 68, 68, 68, + 68, 117, 128, 117, 130, 111, 157, 160, 66, 128, + 133, 133, 157, 160, 117, 823, 133, 239, 130, 132, + 67, 132, 132, 132, 132, 132, 132, 133, 136, 136, + 136, 146, 146, 169, 136, 147, 147, 146, 161, 161, + 239, 147, 219, 169, 161, 136, 218, 825, 146, 240, + 244, 161, 147, 162, 162, 161, 163, 163, 218, 162, + 219, 827, 163, 173, 173, 173, 173, 830, 244, 831, + + 162, 835, 240, 163, 174, 174, 174, 174, 175, 175, + 175, 175, 176, 176, 176, 176, 177, 177, 177, 177, + 178, 178, 178, 178, 179, 179, 179, 179, 181, 181, + 181, 181, 181, 181, 837, 839, 241, 182, 181, 182, + 182, 182, 182, 182, 182, 183, 183, 183, 183, 183, + 183, 184, 184, 184, 184, 184, 184, 185, 186, 241, + 192, 194, 185, 186, 181, 192, 194, 243, 195, 845, + 452, 185, 186, 195, 192, 194, 846, 185, 186, 243, + 185, 186, 195, 192, 194, 196, 452, 197, 199, 198, + 196, 195, 197, 199, 198, 288, 842, 842, 288, 196, + + 242, 197, 199, 198, 851, 196, 807, 197, 196, 198, + 197, 199, 198, 201, 203, 204, 206, 207, 201, 203, + 204, 206, 207, 242, 806, 804, 803, 201, 203, 204, + 206, 207, 260, 799, 203, 204, 201, 203, 204, 206, + 207, 209, 210, 211, 798, 260, 209, 210, 211, 303, + 303, 303, 303, 212, 213, 209, 210, 211, 212, 213, + 797, 209, 210, 211, 209, 210, 211, 212, 213, 579, + 214, 796, 579, 212, 213, 214, 212, 213, 220, 794, + 221, 223, 224, 220, 214, 221, 223, 224, 792, 790, + 214, 787, 220, 214, 221, 223, 224, 226, 786, 784, + + 783, 220, 226, 221, 223, 224, 277, 227, 228, 229, + 231, 226, 227, 228, 229, 231, 782, 226, 781, 277, + 226, 227, 228, 229, 231, 780, 779, 227, 228, 229, + 227, 228, 229, 231, 232, 233, 234, 776, 775, 232, + 233, 234, 304, 304, 304, 304, 235, 236, 232, 233, + 234, 235, 236, 773, 232, 233, 234, 232, 233, 234, + 235, 236, 772, 245, 771, 770, 235, 236, 245, 235, + 236, 256, 256, 256, 256, 256, 256, 245, 257, 257, + 257, 257, 257, 257, 258, 258, 245, 272, 272, 769, + 258, 290, 290, 272, 291, 291, 768, 290, 766, 311, + + 291, 258, 765, 764, 272, 314, 292, 292, 290, 762, + 760, 291, 292, 300, 759, 300, 756, 300, 305, 305, + 305, 305, 311, 292, 306, 306, 306, 306, 314, 322, + 300, 307, 307, 307, 307, 308, 308, 308, 308, 308, + 308, 309, 309, 309, 309, 309, 309, 315, 326, 315, + 329, 332, 322, 335, 338, 341, 344, 347, 350, 353, + 356, 359, 362, 365, 368, 371, 374, 384, 387, 391, + 315, 326, 394, 329, 332, 398, 335, 338, 341, 344, + 347, 350, 353, 356, 359, 362, 365, 368, 371, 374, + 384, 387, 391, 401, 404, 394, 407, 411, 398, 414, + + 417, 420, 423, 426, 438, 440, 443, 753, 504, 504, + 553, 556, 752, 751, 642, 642, 401, 404, 750, 407, + 411, 738, 414, 417, 420, 423, 426, 438, 440, 443, + 504, 737, 736, 553, 556, 642, 850, 850, 850, 852, + 852, 707, 700, 699, 696, 695, 694, 693, 692, 691, + 690, 689, 688, 687, 685, 682, 681, 680, 679, 678, + 677, 676, 675, 674, 673, 672, 670, 669, 668, 667, + 665, 663, 662, 661, 660, 659, 658, 657, 656, 655, + 654, 653, 652, 651, 650, 649, 648, 647, 646, 645, + 644, 643, 641, 640, 639, 638, 637, 636, 635, 634, + + 633, 630, 629, 628, 627, 626, 625, 624, 623, 622, + 621, 620, 619, 618, 617, 616, 615, 614, 613, 612, + 610, 609, 608, 607, 606, 605, 604, 603, 602, 601, + 600, 599, 598, 597, 596, 595, 593, 592, 591, 590, + 589, 587, 586, 585, 584, 583, 582, 581, 578, 577, + 575, 574, 573, 572, 569, 568, 567, 566, 565, 564, + 563, 561, 560, 559, 558, 557, 555, 554, 552, 551, + 550, 549, 548, 547, 546, 545, 544, 543, 542, 541, + 540, 539, 538, 537, 536, 535, 534, 533, 532, 531, + 530, 526, 525, 524, 523, 522, 521, 520, 519, 518, + + 517, 516, 515, 514, 513, 512, 511, 510, 509, 508, + 507, 506, 505, 502, 501, 500, 497, 496, 495, 494, + 493, 492, 490, 488, 487, 486, 484, 483, 482, 481, + 480, 479, 478, 477, 476, 475, 474, 473, 472, 471, + 470, 469, 468, 467, 465, 464, 463, 462, 461, 460, + 459, 456, 455, 454, 451, 450, 449, 448, 447, 446, + 445, 444, 442, 441, 439, 437, 436, 435, 434, 433, + 432, 431, 430, 428, 427, 425, 424, 422, 421, 419, + 418, 416, 415, 413, 412, 410, 409, 408, 406, 405, + 403, 402, 400, 399, 397, 396, 395, 393, 392, 390, + + 388, 386, 385, 383, 382, 380, 378, 377, 376, 375, + 373, 372, 370, 369, 367, 366, 364, 363, 361, 360, + 358, 357, 355, 354, 352, 351, 349, 348, 346, 345, + 343, 342, 340, 339, 337, 336, 334, 333, 331, 330, + 328, 327, 325, 323, 321, 320, 319, 318, 317, 316, + 313, 312, 310, 302, 301, 299, 298, 297, 296, 295, + 294, 289, 287, 286, 285, 284, 283, 282, 281, 280, + 278, 276, 275, 273, 271, 270, 267, 266, 265, 264, + 263, 262, 261, 259, 255, 254, 253, 252, 251, 250, + 249, 248, 247, 246, 238, 237, 230, 225, 222, 217, + + 216, 215, 193, 191, 190, 189, 187, 172, 171, 170, + 168, 167, 166, 165, 164, 158, 156, 155, 154, 153, + 152, 151, 150, 149, 148, 145, 144, 143, 142, 141, + 139, 138, 137, 135, 134, 131, 129, 127, 126, 125, + 124, 123, 122, 121, 120, 119, 115, 114, 113, 112, + 110, 109, 107, 106, 105, 104, 103, 102, 101, 100, + 98, 97, 96, 95, 94, 93, 91, 90, 89, 88, + 87, 86, 85, 83, 82, 79, 78, 77, 76, 74, + 73, 72, 56, 55, 52, 51, 38, 37, 35, 34, + 33, 32, 31, 30, 28, 20, 8, 7, 3, 849, + + 849, 849, 849, 849, 849, 849, 849, 849, 849, 849, + 849, 849, 849, 849, 849, 849, 849, 849, 849, 849, + 849, 849, 849, 849, 849, 849, 849, 849, 849, 849, + 849, 849, 849, 849, 849, 849, 849, 849, 849, 849, + 849, 849, 849, 849, 849, 849, 849, 849, 849, 849, + 849, 849, 849, 849, 849, 849, 849, 849, 849, 849, + 849, 849, 849, 849, 849, 849, 849 + } ; + +/* The intent behind this definition is that it'll catch + * any uses of REJECT which flex missed. + */ +#define REJECT reject_used_but_not_detected +#define yymore() yymore_used_but_not_detected +#define YY_MORE_ADJ 0 +#define YY_RESTORE_YY_MORE_OFFSET +#line 1 "program_lexer.l" +#line 2 "program_lexer.l" +/* + * Copyright © 2009 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. + */ +#include "main/glheader.h" +#include "main/imports.h" +#include "program/prog_instruction.h" +#include "program/prog_statevars.h" +#include "program/symbol_table.h" +#include "program/program_parser.h" +#include "program/program_parse.tab.h" + +#define require_ARB_vp (yyextra->mode == ARB_vertex) +#define require_ARB_fp (yyextra->mode == ARB_fragment) +#define require_NV_fp (yyextra->option.NV_fragment) +#define require_shadow (yyextra->option.Shadow) +#define require_rect (yyextra->option.TexRect) +#define require_texarray (yyextra->option.TexArray) + +#ifndef HAVE_UNISTD_H +#define YY_NO_UNISTD_H +#endif + +#define return_token_or_IDENTIFIER(condition, token) \ + do { \ + if (condition) { \ + return token; \ + } else { \ + return handle_ident(yyextra, yytext, yylval); \ + } \ + } while (0) + +#define return_token_or_DOT(condition, token) \ + do { \ + if (condition) { \ + return token; \ + } else { \ + yyless(1); \ + return DOT; \ + } \ + } while (0) + + +#define return_opcode(condition, token, opcode, len) \ + do { \ + if (condition && \ + _mesa_parse_instruction_suffix(yyextra, \ + yytext + len, \ + & yylval->temp_inst)) { \ + yylval->temp_inst.Opcode = OPCODE_ ## opcode; \ + return token; \ + } else { \ + return handle_ident(yyextra, yytext, yylval); \ + } \ + } while (0) + +#define SWIZZLE_INVAL MAKE_SWIZZLE4(SWIZZLE_NIL, SWIZZLE_NIL, \ + SWIZZLE_NIL, SWIZZLE_NIL) + +static unsigned +mask_from_char(char c) +{ + switch (c) { + case 'x': + case 'r': + return WRITEMASK_X; + case 'y': + case 'g': + return WRITEMASK_Y; + case 'z': + case 'b': + return WRITEMASK_Z; + case 'w': + case 'a': + return WRITEMASK_W; + } + + return 0; +} + +static unsigned +swiz_from_char(char c) +{ + switch (c) { + case 'x': + case 'r': + return SWIZZLE_X; + case 'y': + case 'g': + return SWIZZLE_Y; + case 'z': + case 'b': + return SWIZZLE_Z; + case 'w': + case 'a': + return SWIZZLE_W; + } + + return 0; +} + +static int +handle_ident(struct asm_parser_state *state, const char *text, YYSTYPE *lval) +{ + lval->string = strdup(text); + + return (_mesa_symbol_table_find_symbol(state->st, 0, text) == NULL) + ? IDENTIFIER : USED_IDENTIFIER; +} + +#define YY_USER_ACTION \ + do { \ + yylloc->first_column = yylloc->last_column; \ + yylloc->last_column += yyleng; \ + if ((yylloc->first_line == 1) \ + && (yylloc->first_column == 1)) { \ + yylloc->position = 1; \ + } else { \ + yylloc->position += yylloc->last_column - yylloc->first_column; \ + } \ + } while(0); + +#define YY_NO_INPUT + +/* Yes, this is intentionally doing nothing. We have this line of code +here only to avoid the compiler complaining about an unput function +that is defined, but never called. */ +#define YY_USER_INIT while (0) { unput(0); } + +#define YY_EXTRA_TYPE struct asm_parser_state * + +/* Flex defines a couple of functions with no declarations nor the +static keyword. Declare them here to avoid a compiler warning. */ +int yyget_column (yyscan_t yyscanner); +void yyset_column (int column_no , yyscan_t yyscanner); + +#line 1177 "lex.yy.c" + +#define INITIAL 0 + +#ifndef YY_NO_UNISTD_H +/* Special case for "unistd.h", since it is non-ANSI. We include it way + * down here because we want the user's section 1 to have been scanned first. + * The user has a chance to override it with an option. + */ +#include <unistd.h> +#endif + +#ifndef YY_EXTRA_TYPE +#define YY_EXTRA_TYPE void * +#endif + +/* Holds the entire state of the reentrant scanner. */ +struct yyguts_t + { + + /* User-defined. Not touched by flex. */ + YY_EXTRA_TYPE yyextra_r; + + /* The rest are the same as the globals declared in the non-reentrant scanner. */ + FILE *yyin_r, *yyout_r; + size_t yy_buffer_stack_top; /**< index of top of stack. */ + size_t yy_buffer_stack_max; /**< capacity of stack. */ + YY_BUFFER_STATE * yy_buffer_stack; /**< Stack as an array. */ + char yy_hold_char; + int yy_n_chars; + int yyleng_r; + char *yy_c_buf_p; + int yy_init; + int yy_start; + int yy_did_buffer_switch_on_eof; + int yy_start_stack_ptr; + int yy_start_stack_depth; + int *yy_start_stack; + yy_state_type yy_last_accepting_state; + char* yy_last_accepting_cpos; + + int yylineno_r; + int yy_flex_debug_r; + + char *yytext_r; + int yy_more_flag; + int yy_more_len; + + YYSTYPE * yylval_r; + + YYLTYPE * yylloc_r; + + }; /* end struct yyguts_t */ + +static int yy_init_globals (yyscan_t yyscanner ); + + /* This must go here because YYSTYPE and YYLTYPE are included + * from bison output in section 1.*/ + # define yylval yyg->yylval_r + + # define yylloc yyg->yylloc_r + +int yylex_init (yyscan_t* scanner); + +int yylex_init_extra (YY_EXTRA_TYPE user_defined,yyscan_t* scanner); + +/* Accessor methods to globals. + These are made visible to non-reentrant scanners for convenience. */ + +int yylex_destroy (yyscan_t yyscanner ); + +int yyget_debug (yyscan_t yyscanner ); + +void yyset_debug (int debug_flag ,yyscan_t yyscanner ); + +YY_EXTRA_TYPE yyget_extra (yyscan_t yyscanner ); + +void yyset_extra (YY_EXTRA_TYPE user_defined ,yyscan_t yyscanner ); + +FILE *yyget_in (yyscan_t yyscanner ); + +void yyset_in (FILE * in_str ,yyscan_t yyscanner ); + +FILE *yyget_out (yyscan_t yyscanner ); + +void yyset_out (FILE * out_str ,yyscan_t yyscanner ); + +int yyget_leng (yyscan_t yyscanner ); + +char *yyget_text (yyscan_t yyscanner ); + +int yyget_lineno (yyscan_t yyscanner ); + +void yyset_lineno (int line_number ,yyscan_t yyscanner ); + +YYSTYPE * yyget_lval (yyscan_t yyscanner ); + +void yyset_lval (YYSTYPE * yylval_param ,yyscan_t yyscanner ); + + YYLTYPE *yyget_lloc (yyscan_t yyscanner ); + + void yyset_lloc (YYLTYPE * yylloc_param ,yyscan_t yyscanner ); + +/* Macros after this point can all be overridden by user definitions in + * section 1. + */ + +#ifndef YY_SKIP_YYWRAP +#ifdef __cplusplus +extern "C" int yywrap (yyscan_t yyscanner ); +#else +extern int yywrap (yyscan_t yyscanner ); +#endif +#endif + + static void yyunput (int c,char *buf_ptr ,yyscan_t yyscanner); + +#ifndef yytext_ptr +static void yy_flex_strncpy (char *,yyconst char *,int ,yyscan_t yyscanner); +#endif + +#ifdef YY_NEED_STRLEN +static int yy_flex_strlen (yyconst char * ,yyscan_t yyscanner); +#endif + +#ifndef YY_NO_INPUT + +#ifdef __cplusplus +static int yyinput (yyscan_t yyscanner ); +#else +static int input (yyscan_t yyscanner ); +#endif + +#endif + +/* Amount of stuff to slurp up with each read. */ +#ifndef YY_READ_BUF_SIZE +#ifdef __ia64__ +/* On IA-64, the buffer size is 16k, not 8k */ +#define YY_READ_BUF_SIZE 16384 +#else +#define YY_READ_BUF_SIZE 8192 +#endif /* __ia64__ */ +#endif + +/* Copy whatever the last rule matched to the standard output. */ +#ifndef ECHO +/* This used to be an fputs(), but since the string might contain NUL's, + * we now use fwrite(). + */ +#define ECHO do { if (fwrite( yytext, yyleng, 1, yyout )) {} } while (0) +#endif + +/* Gets input and stuffs it into "buf". number of characters read, or YY_NULL, + * is returned in "result". + */ +#ifndef YY_INPUT +#define YY_INPUT(buf,result,max_size) \ + if ( YY_CURRENT_BUFFER_LVALUE->yy_is_interactive ) \ + { \ + int c = '*'; \ + size_t n; \ + for ( n = 0; n < max_size && \ + (c = getc( yyin )) != EOF && c != '\n'; ++n ) \ + buf[n] = (char) c; \ + if ( c == '\n' ) \ + buf[n++] = (char) c; \ + if ( c == EOF && ferror( yyin ) ) \ + YY_FATAL_ERROR( "input in flex scanner failed" ); \ + result = n; \ + } \ + else \ + { \ + errno=0; \ + while ( (result = fread(buf, 1, max_size, yyin))==0 && ferror(yyin)) \ + { \ + if( errno != EINTR) \ + { \ + YY_FATAL_ERROR( "input in flex scanner failed" ); \ + break; \ + } \ + errno=0; \ + clearerr(yyin); \ + } \ + }\ +\ + +#endif + +/* No semi-colon after return; correct usage is to write "yyterminate();" - + * we don't want an extra ';' after the "return" because that will cause + * some compilers to complain about unreachable statements. + */ +#ifndef yyterminate +#define yyterminate() return YY_NULL +#endif + +/* Number of entries by which start-condition stack grows. */ +#ifndef YY_START_STACK_INCR +#define YY_START_STACK_INCR 25 +#endif + +/* Report a fatal error. */ +#ifndef YY_FATAL_ERROR +#define YY_FATAL_ERROR(msg) yy_fatal_error( msg , yyscanner) +#endif + +/* end tables serialization structures and prototypes */ + +/* Default declaration of generated scanner - a define so the user can + * easily add parameters. + */ +#ifndef YY_DECL +#define YY_DECL_IS_OURS 1 + +extern int yylex \ + (YYSTYPE * yylval_param,YYLTYPE * yylloc_param ,yyscan_t yyscanner); + +#define YY_DECL int yylex \ + (YYSTYPE * yylval_param, YYLTYPE * yylloc_param , yyscan_t yyscanner) +#endif /* !YY_DECL */ + +/* Code executed at the beginning of each rule, after yytext and yyleng + * have been set up. + */ +#ifndef YY_USER_ACTION +#define YY_USER_ACTION +#endif + +/* Code executed at the end of each rule. */ +#ifndef YY_BREAK +#define YY_BREAK break; +#endif + +#define YY_RULE_SETUP \ + YY_USER_ACTION + +/** The main scanner function which does all the work. + */ +YY_DECL +{ + register yy_state_type yy_current_state; + register char *yy_cp, *yy_bp; + register int yy_act; + struct yyguts_t * yyg = (struct yyguts_t*)yyscanner; + +#line 169 "program_lexer.l" + + +#line 1426 "lex.yy.c" + + yylval = yylval_param; + + yylloc = yylloc_param; + + if ( !yyg->yy_init ) + { + yyg->yy_init = 1; + +#ifdef YY_USER_INIT + YY_USER_INIT; +#endif + + if ( ! yyg->yy_start ) + yyg->yy_start = 1; /* first start state */ + + if ( ! yyin ) + yyin = stdin; + + if ( ! yyout ) + yyout = stdout; + + if ( ! YY_CURRENT_BUFFER ) { + yyensure_buffer_stack (yyscanner); + YY_CURRENT_BUFFER_LVALUE = + yy_create_buffer(yyin,YY_BUF_SIZE ,yyscanner); + } + + yy_load_buffer_state(yyscanner ); + } + + while ( 1 ) /* loops until end-of-file is reached */ + { + yy_cp = yyg->yy_c_buf_p; + + /* Support of yytext. */ + *yy_cp = yyg->yy_hold_char; + + /* yy_bp points to the position in yy_ch_buf of the start of + * the current run. + */ + yy_bp = yy_cp; + + yy_current_state = yyg->yy_start; +yy_match: + do + { + register YY_CHAR yy_c = yy_ec[YY_SC_TO_UI(*yy_cp)]; + if ( yy_accept[yy_current_state] ) + { + yyg->yy_last_accepting_state = yy_current_state; + yyg->yy_last_accepting_cpos = yy_cp; + } + while ( yy_chk[yy_base[yy_current_state] + yy_c] != yy_current_state ) + { + yy_current_state = (int) yy_def[yy_current_state]; + if ( yy_current_state >= 850 ) + yy_c = yy_meta[(unsigned int) yy_c]; + } + yy_current_state = yy_nxt[yy_base[yy_current_state] + (unsigned int) yy_c]; + ++yy_cp; + } + while ( yy_base[yy_current_state] != 1300 ); + +yy_find_action: + yy_act = yy_accept[yy_current_state]; + if ( yy_act == 0 ) + { /* have to back up */ + yy_cp = yyg->yy_last_accepting_cpos; + yy_current_state = yyg->yy_last_accepting_state; + yy_act = yy_accept[yy_current_state]; + } + + YY_DO_BEFORE_ACTION; + +do_action: /* This label is used only to access EOF actions. */ + + switch ( yy_act ) + { /* beginning of action switch */ + case 0: /* must back up */ + /* undo the effects of YY_DO_BEFORE_ACTION */ + *yy_cp = yyg->yy_hold_char; + yy_cp = yyg->yy_last_accepting_cpos; + yy_current_state = yyg->yy_last_accepting_state; + goto yy_find_action; + +case 1: +YY_RULE_SETUP +#line 171 "program_lexer.l" +{ return ARBvp_10; } + YY_BREAK +case 2: +YY_RULE_SETUP +#line 172 "program_lexer.l" +{ return ARBfp_10; } + YY_BREAK +case 3: +YY_RULE_SETUP +#line 173 "program_lexer.l" +{ + yylval->integer = at_address; + return_token_or_IDENTIFIER(require_ARB_vp, ADDRESS); +} + YY_BREAK +case 4: +YY_RULE_SETUP +#line 177 "program_lexer.l" +{ return ALIAS; } + YY_BREAK +case 5: +YY_RULE_SETUP +#line 178 "program_lexer.l" +{ return ATTRIB; } + YY_BREAK +case 6: +YY_RULE_SETUP +#line 179 "program_lexer.l" +{ return END; } + YY_BREAK +case 7: +YY_RULE_SETUP +#line 180 "program_lexer.l" +{ return OPTION; } + YY_BREAK +case 8: +YY_RULE_SETUP +#line 181 "program_lexer.l" +{ return OUTPUT; } + YY_BREAK +case 9: +YY_RULE_SETUP +#line 182 "program_lexer.l" +{ return PARAM; } + YY_BREAK +case 10: +YY_RULE_SETUP +#line 183 "program_lexer.l" +{ yylval->integer = at_temp; return TEMP; } + YY_BREAK +case 11: +YY_RULE_SETUP +#line 185 "program_lexer.l" +{ return_opcode( 1, VECTOR_OP, ABS, 3); } + YY_BREAK +case 12: +YY_RULE_SETUP +#line 186 "program_lexer.l" +{ return_opcode( 1, BIN_OP, ADD, 3); } + YY_BREAK +case 13: +YY_RULE_SETUP +#line 187 "program_lexer.l" +{ return_opcode(require_ARB_vp, ARL, ARL, 3); } + YY_BREAK +case 14: +YY_RULE_SETUP +#line 189 "program_lexer.l" +{ return_opcode(require_ARB_fp, TRI_OP, CMP, 3); } + YY_BREAK +case 15: +YY_RULE_SETUP +#line 190 "program_lexer.l" +{ return_opcode(require_ARB_fp, SCALAR_OP, COS, 3); } + YY_BREAK +case 16: +YY_RULE_SETUP +#line 192 "program_lexer.l" +{ return_opcode(require_NV_fp, VECTOR_OP, DDX, 3); } + YY_BREAK +case 17: +YY_RULE_SETUP +#line 193 "program_lexer.l" +{ return_opcode(require_NV_fp, VECTOR_OP, DDY, 3); } + YY_BREAK +case 18: +YY_RULE_SETUP +#line 194 "program_lexer.l" +{ return_opcode( 1, BIN_OP, DP3, 3); } + YY_BREAK +case 19: +YY_RULE_SETUP +#line 195 "program_lexer.l" +{ return_opcode( 1, BIN_OP, DP4, 3); } + YY_BREAK +case 20: +YY_RULE_SETUP +#line 196 "program_lexer.l" +{ return_opcode( 1, BIN_OP, DPH, 3); } + YY_BREAK +case 21: +YY_RULE_SETUP +#line 197 "program_lexer.l" +{ return_opcode( 1, BIN_OP, DST, 3); } + YY_BREAK +case 22: +YY_RULE_SETUP +#line 199 "program_lexer.l" +{ return_opcode( 1, SCALAR_OP, EX2, 3); } + YY_BREAK +case 23: +YY_RULE_SETUP +#line 200 "program_lexer.l" +{ return_opcode(require_ARB_vp, SCALAR_OP, EXP, 3); } + YY_BREAK +case 24: +YY_RULE_SETUP +#line 202 "program_lexer.l" +{ return_opcode( 1, VECTOR_OP, FLR, 3); } + YY_BREAK +case 25: +YY_RULE_SETUP +#line 203 "program_lexer.l" +{ return_opcode( 1, VECTOR_OP, FRC, 3); } + YY_BREAK +case 26: +YY_RULE_SETUP +#line 205 "program_lexer.l" +{ return_opcode(require_ARB_fp, KIL, KIL, 3); } + YY_BREAK +case 27: +YY_RULE_SETUP +#line 207 "program_lexer.l" +{ return_opcode( 1, VECTOR_OP, LIT, 3); } + YY_BREAK +case 28: +YY_RULE_SETUP +#line 208 "program_lexer.l" +{ return_opcode( 1, SCALAR_OP, LG2, 3); } + YY_BREAK +case 29: +YY_RULE_SETUP +#line 209 "program_lexer.l" +{ return_opcode(require_ARB_vp, SCALAR_OP, LOG, 3); } + YY_BREAK +case 30: +YY_RULE_SETUP +#line 210 "program_lexer.l" +{ return_opcode(require_ARB_fp, TRI_OP, LRP, 3); } + YY_BREAK +case 31: +YY_RULE_SETUP +#line 212 "program_lexer.l" +{ return_opcode( 1, TRI_OP, MAD, 3); } + YY_BREAK +case 32: +YY_RULE_SETUP +#line 213 "program_lexer.l" +{ return_opcode( 1, BIN_OP, MAX, 3); } + YY_BREAK +case 33: +YY_RULE_SETUP +#line 214 "program_lexer.l" +{ return_opcode( 1, BIN_OP, MIN, 3); } + YY_BREAK +case 34: +YY_RULE_SETUP +#line 215 "program_lexer.l" +{ return_opcode( 1, VECTOR_OP, MOV, 3); } + YY_BREAK +case 35: +YY_RULE_SETUP +#line 216 "program_lexer.l" +{ return_opcode( 1, BIN_OP, MUL, 3); } + YY_BREAK +case 36: +YY_RULE_SETUP +#line 218 "program_lexer.l" +{ return_opcode(require_NV_fp, VECTOR_OP, PK2H, 4); } + YY_BREAK +case 37: +YY_RULE_SETUP +#line 219 "program_lexer.l" +{ return_opcode(require_NV_fp, VECTOR_OP, PK2US, 5); } + YY_BREAK +case 38: +YY_RULE_SETUP +#line 220 "program_lexer.l" +{ return_opcode(require_NV_fp, VECTOR_OP, PK4B, 4); } + YY_BREAK +case 39: +YY_RULE_SETUP +#line 221 "program_lexer.l" +{ return_opcode(require_NV_fp, VECTOR_OP, PK4UB, 5); } + YY_BREAK +case 40: +YY_RULE_SETUP +#line 222 "program_lexer.l" +{ return_opcode( 1, BINSC_OP, POW, 3); } + YY_BREAK +case 41: +YY_RULE_SETUP +#line 224 "program_lexer.l" +{ return_opcode( 1, SCALAR_OP, RCP, 3); } + YY_BREAK +case 42: +YY_RULE_SETUP +#line 225 "program_lexer.l" +{ return_opcode(require_NV_fp, BIN_OP, RFL, 3); } + YY_BREAK +case 43: +YY_RULE_SETUP +#line 226 "program_lexer.l" +{ return_opcode( 1, SCALAR_OP, RSQ, 3); } + YY_BREAK +case 44: +YY_RULE_SETUP +#line 228 "program_lexer.l" +{ return_opcode(require_ARB_fp, SCALAR_OP, SCS, 3); } + YY_BREAK +case 45: +YY_RULE_SETUP +#line 229 "program_lexer.l" +{ return_opcode(require_NV_fp, BIN_OP, SEQ, 3); } + YY_BREAK +case 46: +YY_RULE_SETUP +#line 230 "program_lexer.l" +{ return_opcode(require_NV_fp, BIN_OP, SFL, 3); } + YY_BREAK +case 47: +YY_RULE_SETUP +#line 231 "program_lexer.l" +{ return_opcode( 1, BIN_OP, SGE, 3); } + YY_BREAK +case 48: +YY_RULE_SETUP +#line 232 "program_lexer.l" +{ return_opcode(require_NV_fp, BIN_OP, SGT, 3); } + YY_BREAK +case 49: +YY_RULE_SETUP +#line 233 "program_lexer.l" +{ return_opcode(require_ARB_fp, SCALAR_OP, SIN, 3); } + YY_BREAK +case 50: +YY_RULE_SETUP +#line 234 "program_lexer.l" +{ return_opcode(require_NV_fp, BIN_OP, SLE, 3); } + YY_BREAK +case 51: +YY_RULE_SETUP +#line 235 "program_lexer.l" +{ return_opcode( 1, BIN_OP, SLT, 3); } + YY_BREAK +case 52: +YY_RULE_SETUP +#line 236 "program_lexer.l" +{ return_opcode(require_NV_fp, BIN_OP, SNE, 3); } + YY_BREAK +case 53: +YY_RULE_SETUP +#line 237 "program_lexer.l" +{ return_opcode(require_NV_fp, BIN_OP, STR, 3); } + YY_BREAK +case 54: +YY_RULE_SETUP +#line 238 "program_lexer.l" +{ return_opcode( 1, BIN_OP, SUB, 3); } + YY_BREAK +case 55: +YY_RULE_SETUP +#line 239 "program_lexer.l" +{ return_opcode( 1, SWZ, SWZ, 3); } + YY_BREAK +case 56: +YY_RULE_SETUP +#line 241 "program_lexer.l" +{ return_opcode(require_ARB_fp, SAMPLE_OP, TEX, 3); } + YY_BREAK +case 57: +YY_RULE_SETUP +#line 242 "program_lexer.l" +{ return_opcode(require_ARB_fp, SAMPLE_OP, TXB, 3); } + YY_BREAK +case 58: +YY_RULE_SETUP +#line 243 "program_lexer.l" +{ return_opcode(require_NV_fp, TXD_OP, TXD, 3); } + YY_BREAK +case 59: +YY_RULE_SETUP +#line 244 "program_lexer.l" +{ return_opcode(require_ARB_fp, SAMPLE_OP, TXP, 3); } + YY_BREAK +case 60: +YY_RULE_SETUP +#line 246 "program_lexer.l" +{ return_opcode(require_NV_fp, SCALAR_OP, UP2H, 4); } + YY_BREAK +case 61: +YY_RULE_SETUP +#line 247 "program_lexer.l" +{ return_opcode(require_NV_fp, SCALAR_OP, UP2US, 5); } + YY_BREAK +case 62: +YY_RULE_SETUP +#line 248 "program_lexer.l" +{ return_opcode(require_NV_fp, SCALAR_OP, UP4B, 4); } + YY_BREAK +case 63: +YY_RULE_SETUP +#line 249 "program_lexer.l" +{ return_opcode(require_NV_fp, SCALAR_OP, UP4UB, 5); } + YY_BREAK +case 64: +YY_RULE_SETUP +#line 251 "program_lexer.l" +{ return_opcode(require_NV_fp, TRI_OP, X2D, 3); } + YY_BREAK +case 65: +YY_RULE_SETUP +#line 252 "program_lexer.l" +{ return_opcode( 1, BIN_OP, XPD, 3); } + YY_BREAK +case 66: +YY_RULE_SETUP +#line 254 "program_lexer.l" +{ return_token_or_IDENTIFIER(require_ARB_vp, VERTEX); } + YY_BREAK +case 67: +YY_RULE_SETUP +#line 255 "program_lexer.l" +{ return_token_or_IDENTIFIER(require_ARB_fp, FRAGMENT); } + YY_BREAK +case 68: +YY_RULE_SETUP +#line 256 "program_lexer.l" +{ return PROGRAM; } + YY_BREAK +case 69: +YY_RULE_SETUP +#line 257 "program_lexer.l" +{ return STATE; } + YY_BREAK +case 70: +YY_RULE_SETUP +#line 258 "program_lexer.l" +{ return RESULT; } + YY_BREAK +case 71: +YY_RULE_SETUP +#line 260 "program_lexer.l" +{ return AMBIENT; } + YY_BREAK +case 72: +YY_RULE_SETUP +#line 261 "program_lexer.l" +{ return ATTENUATION; } + YY_BREAK +case 73: +YY_RULE_SETUP +#line 262 "program_lexer.l" +{ return BACK; } + YY_BREAK +case 74: +YY_RULE_SETUP +#line 263 "program_lexer.l" +{ return_token_or_DOT(require_ARB_vp, CLIP); } + YY_BREAK +case 75: +YY_RULE_SETUP +#line 264 "program_lexer.l" +{ return COLOR; } + YY_BREAK +case 76: +YY_RULE_SETUP +#line 265 "program_lexer.l" +{ return_token_or_DOT(require_ARB_fp, DEPTH); } + YY_BREAK +case 77: +YY_RULE_SETUP +#line 266 "program_lexer.l" +{ return DIFFUSE; } + YY_BREAK +case 78: +YY_RULE_SETUP +#line 267 "program_lexer.l" +{ return DIRECTION; } + YY_BREAK +case 79: +YY_RULE_SETUP +#line 268 "program_lexer.l" +{ return EMISSION; } + YY_BREAK +case 80: +YY_RULE_SETUP +#line 269 "program_lexer.l" +{ return ENV; } + YY_BREAK +case 81: +YY_RULE_SETUP +#line 270 "program_lexer.l" +{ return EYE; } + YY_BREAK +case 82: +YY_RULE_SETUP +#line 271 "program_lexer.l" +{ return FOGCOORD; } + YY_BREAK +case 83: +YY_RULE_SETUP +#line 272 "program_lexer.l" +{ return FOG; } + YY_BREAK +case 84: +YY_RULE_SETUP +#line 273 "program_lexer.l" +{ return FRONT; } + YY_BREAK +case 85: +YY_RULE_SETUP +#line 274 "program_lexer.l" +{ return HALF; } + YY_BREAK +case 86: +YY_RULE_SETUP +#line 275 "program_lexer.l" +{ return INVERSE; } + YY_BREAK +case 87: +YY_RULE_SETUP +#line 276 "program_lexer.l" +{ return INVTRANS; } + YY_BREAK +case 88: +YY_RULE_SETUP +#line 277 "program_lexer.l" +{ return LIGHT; } + YY_BREAK +case 89: +YY_RULE_SETUP +#line 278 "program_lexer.l" +{ return LIGHTMODEL; } + YY_BREAK +case 90: +YY_RULE_SETUP +#line 279 "program_lexer.l" +{ return LIGHTPROD; } + YY_BREAK +case 91: +YY_RULE_SETUP +#line 280 "program_lexer.l" +{ return LOCAL; } + YY_BREAK +case 92: +YY_RULE_SETUP +#line 281 "program_lexer.l" +{ return MATERIAL; } + YY_BREAK +case 93: +YY_RULE_SETUP +#line 282 "program_lexer.l" +{ return MAT_PROGRAM; } + YY_BREAK +case 94: +YY_RULE_SETUP +#line 283 "program_lexer.l" +{ return MATRIX; } + YY_BREAK +case 95: +YY_RULE_SETUP +#line 284 "program_lexer.l" +{ return_token_or_DOT(require_ARB_vp, MATRIXINDEX); } + YY_BREAK +case 96: +YY_RULE_SETUP +#line 285 "program_lexer.l" +{ return MODELVIEW; } + YY_BREAK +case 97: +YY_RULE_SETUP +#line 286 "program_lexer.l" +{ return MVP; } + YY_BREAK +case 98: +YY_RULE_SETUP +#line 287 "program_lexer.l" +{ return_token_or_DOT(require_ARB_vp, NORMAL); } + YY_BREAK +case 99: +YY_RULE_SETUP +#line 288 "program_lexer.l" +{ return OBJECT; } + YY_BREAK +case 100: +YY_RULE_SETUP +#line 289 "program_lexer.l" +{ return PALETTE; } + YY_BREAK +case 101: +YY_RULE_SETUP +#line 290 "program_lexer.l" +{ return PARAMS; } + YY_BREAK +case 102: +YY_RULE_SETUP +#line 291 "program_lexer.l" +{ return PLANE; } + YY_BREAK +case 103: +YY_RULE_SETUP +#line 292 "program_lexer.l" +{ return_token_or_DOT(require_ARB_vp, POINT_TOK); } + YY_BREAK +case 104: +YY_RULE_SETUP +#line 293 "program_lexer.l" +{ return_token_or_DOT(require_ARB_vp, POINTSIZE); } + YY_BREAK +case 105: +YY_RULE_SETUP +#line 294 "program_lexer.l" +{ return POSITION; } + YY_BREAK +case 106: +YY_RULE_SETUP +#line 295 "program_lexer.l" +{ return PRIMARY; } + YY_BREAK +case 107: +YY_RULE_SETUP +#line 296 "program_lexer.l" +{ return PROJECTION; } + YY_BREAK +case 108: +YY_RULE_SETUP +#line 297 "program_lexer.l" +{ return_token_or_DOT(require_ARB_fp, RANGE); } + YY_BREAK +case 109: +YY_RULE_SETUP +#line 298 "program_lexer.l" +{ return ROW; } + YY_BREAK +case 110: +YY_RULE_SETUP +#line 299 "program_lexer.l" +{ return SCENECOLOR; } + YY_BREAK +case 111: +YY_RULE_SETUP +#line 300 "program_lexer.l" +{ return SECONDARY; } + YY_BREAK +case 112: +YY_RULE_SETUP +#line 301 "program_lexer.l" +{ return SHININESS; } + YY_BREAK +case 113: +YY_RULE_SETUP +#line 302 "program_lexer.l" +{ return_token_or_DOT(require_ARB_vp, SIZE_TOK); } + YY_BREAK +case 114: +YY_RULE_SETUP +#line 303 "program_lexer.l" +{ return SPECULAR; } + YY_BREAK +case 115: +YY_RULE_SETUP +#line 304 "program_lexer.l" +{ return SPOT; } + YY_BREAK +case 116: +YY_RULE_SETUP +#line 305 "program_lexer.l" +{ return TEXCOORD; } + YY_BREAK +case 117: +YY_RULE_SETUP +#line 306 "program_lexer.l" +{ return_token_or_DOT(require_ARB_fp, TEXENV); } + YY_BREAK +case 118: +YY_RULE_SETUP +#line 307 "program_lexer.l" +{ return_token_or_DOT(require_ARB_vp, TEXGEN); } + YY_BREAK +case 119: +YY_RULE_SETUP +#line 308 "program_lexer.l" +{ return_token_or_DOT(require_ARB_vp, TEXGEN_Q); } + YY_BREAK +case 120: +YY_RULE_SETUP +#line 309 "program_lexer.l" +{ return_token_or_DOT(require_ARB_vp, TEXGEN_S); } + YY_BREAK +case 121: +YY_RULE_SETUP +#line 310 "program_lexer.l" +{ return_token_or_DOT(require_ARB_vp, TEXGEN_T); } + YY_BREAK +case 122: +YY_RULE_SETUP +#line 311 "program_lexer.l" +{ return TEXTURE; } + YY_BREAK +case 123: +YY_RULE_SETUP +#line 312 "program_lexer.l" +{ return TRANSPOSE; } + YY_BREAK +case 124: +YY_RULE_SETUP +#line 313 "program_lexer.l" +{ return_token_or_DOT(require_ARB_vp, VTXATTRIB); } + YY_BREAK +case 125: +YY_RULE_SETUP +#line 314 "program_lexer.l" +{ return_token_or_DOT(require_ARB_vp, WEIGHT); } + YY_BREAK +case 126: +YY_RULE_SETUP +#line 316 "program_lexer.l" +{ return_token_or_IDENTIFIER(require_ARB_fp, TEXTURE_UNIT); } + YY_BREAK +case 127: +YY_RULE_SETUP +#line 317 "program_lexer.l" +{ return_token_or_IDENTIFIER(require_ARB_fp, TEX_1D); } + YY_BREAK +case 128: +YY_RULE_SETUP +#line 318 "program_lexer.l" +{ return_token_or_IDENTIFIER(require_ARB_fp, TEX_2D); } + YY_BREAK +case 129: +YY_RULE_SETUP +#line 319 "program_lexer.l" +{ return_token_or_IDENTIFIER(require_ARB_fp, TEX_3D); } + YY_BREAK +case 130: +YY_RULE_SETUP +#line 320 "program_lexer.l" +{ return_token_or_IDENTIFIER(require_ARB_fp, TEX_CUBE); } + YY_BREAK +case 131: +YY_RULE_SETUP +#line 321 "program_lexer.l" +{ return_token_or_IDENTIFIER(require_ARB_fp && require_rect, TEX_RECT); } + YY_BREAK +case 132: +YY_RULE_SETUP +#line 322 "program_lexer.l" +{ return_token_or_IDENTIFIER(require_ARB_fp && require_shadow, TEX_SHADOW1D); } + YY_BREAK +case 133: +YY_RULE_SETUP +#line 323 "program_lexer.l" +{ return_token_or_IDENTIFIER(require_ARB_fp && require_shadow, TEX_SHADOW2D); } + YY_BREAK +case 134: +YY_RULE_SETUP +#line 324 "program_lexer.l" +{ return_token_or_IDENTIFIER(require_ARB_fp && require_shadow && require_rect, TEX_SHADOWRECT); } + YY_BREAK +case 135: +YY_RULE_SETUP +#line 325 "program_lexer.l" +{ return_token_or_IDENTIFIER(require_ARB_fp && require_texarray, TEX_ARRAY1D); } + YY_BREAK +case 136: +YY_RULE_SETUP +#line 326 "program_lexer.l" +{ return_token_or_IDENTIFIER(require_ARB_fp && require_texarray, TEX_ARRAY2D); } + YY_BREAK +case 137: +YY_RULE_SETUP +#line 327 "program_lexer.l" +{ return_token_or_IDENTIFIER(require_ARB_fp && require_shadow && require_texarray, TEX_ARRAYSHADOW1D); } + YY_BREAK +case 138: +YY_RULE_SETUP +#line 328 "program_lexer.l" +{ return_token_or_IDENTIFIER(require_ARB_fp && require_shadow && require_texarray, TEX_ARRAYSHADOW2D); } + YY_BREAK +case 139: +YY_RULE_SETUP +#line 330 "program_lexer.l" +{ return handle_ident(yyextra, yytext, yylval); } + YY_BREAK +case 140: +YY_RULE_SETUP +#line 332 "program_lexer.l" +{ return DOT_DOT; } + YY_BREAK +case 141: +YY_RULE_SETUP +#line 334 "program_lexer.l" +{ + yylval->integer = strtol(yytext, NULL, 10); + return INTEGER; +} + YY_BREAK +case 142: +YY_RULE_SETUP +#line 338 "program_lexer.l" +{ + yylval->real = _mesa_strtof(yytext, NULL); + return REAL; +} + YY_BREAK +case 143: +/* rule 143 can match eol */ +*yy_cp = yyg->yy_hold_char; /* undo effects of setting up yytext */ +yyg->yy_c_buf_p = yy_cp -= 1; +YY_DO_BEFORE_ACTION; /* set up yytext again */ +YY_RULE_SETUP +#line 342 "program_lexer.l" +{ + yylval->real = _mesa_strtof(yytext, NULL); + return REAL; +} + YY_BREAK +case 144: +YY_RULE_SETUP +#line 346 "program_lexer.l" +{ + yylval->real = _mesa_strtof(yytext, NULL); + return REAL; +} + YY_BREAK +case 145: +YY_RULE_SETUP +#line 350 "program_lexer.l" +{ + yylval->real = _mesa_strtof(yytext, NULL); + return REAL; +} + YY_BREAK +case 146: +YY_RULE_SETUP +#line 355 "program_lexer.l" +{ + yylval->swiz_mask.swizzle = SWIZZLE_NOOP; + yylval->swiz_mask.mask = WRITEMASK_XYZW; + return MASK4; +} + YY_BREAK +case 147: +YY_RULE_SETUP +#line 361 "program_lexer.l" +{ + yylval->swiz_mask.swizzle = SWIZZLE_INVAL; + yylval->swiz_mask.mask = WRITEMASK_XY + | mask_from_char(yytext[3]); + return MASK3; +} + YY_BREAK +case 148: +YY_RULE_SETUP +#line 367 "program_lexer.l" +{ + yylval->swiz_mask.swizzle = SWIZZLE_INVAL; + yylval->swiz_mask.mask = WRITEMASK_XZW; + return MASK3; +} + YY_BREAK +case 149: +YY_RULE_SETUP +#line 372 "program_lexer.l" +{ + yylval->swiz_mask.swizzle = SWIZZLE_INVAL; + yylval->swiz_mask.mask = WRITEMASK_YZW; + return MASK3; +} + YY_BREAK +case 150: +YY_RULE_SETUP +#line 378 "program_lexer.l" +{ + yylval->swiz_mask.swizzle = SWIZZLE_INVAL; + yylval->swiz_mask.mask = WRITEMASK_X + | mask_from_char(yytext[2]); + return MASK2; +} + YY_BREAK +case 151: +YY_RULE_SETUP +#line 384 "program_lexer.l" +{ + yylval->swiz_mask.swizzle = SWIZZLE_INVAL; + yylval->swiz_mask.mask = WRITEMASK_Y + | mask_from_char(yytext[2]); + return MASK2; +} + YY_BREAK +case 152: +YY_RULE_SETUP +#line 390 "program_lexer.l" +{ + yylval->swiz_mask.swizzle = SWIZZLE_INVAL; + yylval->swiz_mask.mask = WRITEMASK_ZW; + return MASK2; +} + YY_BREAK +case 153: +YY_RULE_SETUP +#line 396 "program_lexer.l" +{ + const unsigned s = swiz_from_char(yytext[1]); + yylval->swiz_mask.swizzle = MAKE_SWIZZLE4(s, s, s, s); + yylval->swiz_mask.mask = mask_from_char(yytext[1]); + return MASK1; +} + YY_BREAK +case 154: +YY_RULE_SETUP +#line 403 "program_lexer.l" +{ + yylval->swiz_mask.swizzle = MAKE_SWIZZLE4(swiz_from_char(yytext[1]), + swiz_from_char(yytext[2]), + swiz_from_char(yytext[3]), + swiz_from_char(yytext[4])); + yylval->swiz_mask.mask = 0; + return SWIZZLE; +} + YY_BREAK +case 155: +YY_RULE_SETUP +#line 412 "program_lexer.l" +{ + yylval->swiz_mask.swizzle = SWIZZLE_NOOP; + yylval->swiz_mask.mask = WRITEMASK_XYZW; + return_token_or_DOT(require_ARB_fp, MASK4); +} + YY_BREAK +case 156: +YY_RULE_SETUP +#line 418 "program_lexer.l" +{ + yylval->swiz_mask.swizzle = SWIZZLE_INVAL; + yylval->swiz_mask.mask = WRITEMASK_XY + | mask_from_char(yytext[3]); + return_token_or_DOT(require_ARB_fp, MASK3); +} + YY_BREAK +case 157: +YY_RULE_SETUP +#line 424 "program_lexer.l" +{ + yylval->swiz_mask.swizzle = SWIZZLE_INVAL; + yylval->swiz_mask.mask = WRITEMASK_XZW; + return_token_or_DOT(require_ARB_fp, MASK3); +} + YY_BREAK +case 158: +YY_RULE_SETUP +#line 429 "program_lexer.l" +{ + yylval->swiz_mask.swizzle = SWIZZLE_INVAL; + yylval->swiz_mask.mask = WRITEMASK_YZW; + return_token_or_DOT(require_ARB_fp, MASK3); +} + YY_BREAK +case 159: +YY_RULE_SETUP +#line 435 "program_lexer.l" +{ + yylval->swiz_mask.swizzle = SWIZZLE_INVAL; + yylval->swiz_mask.mask = WRITEMASK_X + | mask_from_char(yytext[2]); + return_token_or_DOT(require_ARB_fp, MASK2); +} + YY_BREAK +case 160: +YY_RULE_SETUP +#line 441 "program_lexer.l" +{ + yylval->swiz_mask.swizzle = SWIZZLE_INVAL; + yylval->swiz_mask.mask = WRITEMASK_Y + | mask_from_char(yytext[2]); + return_token_or_DOT(require_ARB_fp, MASK2); +} + YY_BREAK +case 161: +YY_RULE_SETUP +#line 447 "program_lexer.l" +{ + yylval->swiz_mask.swizzle = SWIZZLE_INVAL; + yylval->swiz_mask.mask = WRITEMASK_ZW; + return_token_or_DOT(require_ARB_fp, MASK2); +} + YY_BREAK +case 162: +YY_RULE_SETUP +#line 453 "program_lexer.l" +{ + const unsigned s = swiz_from_char(yytext[1]); + yylval->swiz_mask.swizzle = MAKE_SWIZZLE4(s, s, s, s); + yylval->swiz_mask.mask = mask_from_char(yytext[1]); + return_token_or_DOT(require_ARB_fp, MASK1); +} + YY_BREAK +case 163: +YY_RULE_SETUP +#line 461 "program_lexer.l" +{ + if (require_ARB_vp) { + return TEXGEN_R; + } else { + yylval->swiz_mask.swizzle = MAKE_SWIZZLE4(SWIZZLE_X, SWIZZLE_X, + SWIZZLE_X, SWIZZLE_X); + yylval->swiz_mask.mask = WRITEMASK_X; + return MASK1; + } +} + YY_BREAK +case 164: +YY_RULE_SETUP +#line 472 "program_lexer.l" +{ + yylval->swiz_mask.swizzle = MAKE_SWIZZLE4(swiz_from_char(yytext[1]), + swiz_from_char(yytext[2]), + swiz_from_char(yytext[3]), + swiz_from_char(yytext[4])); + yylval->swiz_mask.mask = 0; + return_token_or_DOT(require_ARB_fp, SWIZZLE); +} + YY_BREAK +case 165: +YY_RULE_SETUP +#line 481 "program_lexer.l" +{ return DOT; } + YY_BREAK +case 166: +/* rule 166 can match eol */ +YY_RULE_SETUP +#line 483 "program_lexer.l" +{ + yylloc->first_line++; + yylloc->first_column = 1; + yylloc->last_line++; + yylloc->last_column = 1; + yylloc->position++; +} + YY_BREAK +case 167: +YY_RULE_SETUP +#line 490 "program_lexer.l" +/* eat whitespace */ ; + YY_BREAK +case 168: +*yy_cp = yyg->yy_hold_char; /* undo effects of setting up yytext */ +yyg->yy_c_buf_p = yy_cp -= 1; +YY_DO_BEFORE_ACTION; /* set up yytext again */ +YY_RULE_SETUP +#line 491 "program_lexer.l" +/* eat comments */ ; + YY_BREAK +case 169: +YY_RULE_SETUP +#line 492 "program_lexer.l" +{ return yytext[0]; } + YY_BREAK +case 170: +YY_RULE_SETUP +#line 493 "program_lexer.l" +ECHO; + YY_BREAK +#line 2490 "lex.yy.c" +case YY_STATE_EOF(INITIAL): + yyterminate(); + + case YY_END_OF_BUFFER: + { + /* Amount of text matched not including the EOB char. */ + int yy_amount_of_matched_text = (int) (yy_cp - yyg->yytext_ptr) - 1; + + /* Undo the effects of YY_DO_BEFORE_ACTION. */ + *yy_cp = yyg->yy_hold_char; + YY_RESTORE_YY_MORE_OFFSET + + if ( YY_CURRENT_BUFFER_LVALUE->yy_buffer_status == YY_BUFFER_NEW ) + { + /* We're scanning a new file or input source. It's + * possible that this happened because the user + * just pointed yyin at a new source and called + * yylex(). If so, then we have to assure + * consistency between YY_CURRENT_BUFFER and our + * globals. Here is the right place to do so, because + * this is the first action (other than possibly a + * back-up) that will match for the new input source. + */ + yyg->yy_n_chars = YY_CURRENT_BUFFER_LVALUE->yy_n_chars; + YY_CURRENT_BUFFER_LVALUE->yy_input_file = yyin; + YY_CURRENT_BUFFER_LVALUE->yy_buffer_status = YY_BUFFER_NORMAL; + } + + /* Note that here we test for yy_c_buf_p "<=" to the position + * of the first EOB in the buffer, since yy_c_buf_p will + * already have been incremented past the NUL character + * (since all states make transitions on EOB to the + * end-of-buffer state). Contrast this with the test + * in input(). + */ + if ( yyg->yy_c_buf_p <= &YY_CURRENT_BUFFER_LVALUE->yy_ch_buf[yyg->yy_n_chars] ) + { /* This was really a NUL. */ + yy_state_type yy_next_state; + + yyg->yy_c_buf_p = yyg->yytext_ptr + yy_amount_of_matched_text; + + yy_current_state = yy_get_previous_state( yyscanner ); + + /* Okay, we're now positioned to make the NUL + * transition. We couldn't have + * yy_get_previous_state() go ahead and do it + * for us because it doesn't know how to deal + * with the possibility of jamming (and we don't + * want to build jamming into it because then it + * will run more slowly). + */ + + yy_next_state = yy_try_NUL_trans( yy_current_state , yyscanner); + + yy_bp = yyg->yytext_ptr + YY_MORE_ADJ; + + if ( yy_next_state ) + { + /* Consume the NUL. */ + yy_cp = ++yyg->yy_c_buf_p; + yy_current_state = yy_next_state; + goto yy_match; + } + + else + { + yy_cp = yyg->yy_c_buf_p; + goto yy_find_action; + } + } + + else switch ( yy_get_next_buffer( yyscanner ) ) + { + case EOB_ACT_END_OF_FILE: + { + yyg->yy_did_buffer_switch_on_eof = 0; + + if ( yywrap(yyscanner ) ) + { + /* Note: because we've taken care in + * yy_get_next_buffer() to have set up + * yytext, we can now set up + * yy_c_buf_p so that if some total + * hoser (like flex itself) wants to + * call the scanner after we return the + * YY_NULL, it'll still work - another + * YY_NULL will get returned. + */ + yyg->yy_c_buf_p = yyg->yytext_ptr + YY_MORE_ADJ; + + yy_act = YY_STATE_EOF(YY_START); + goto do_action; + } + + else + { + if ( ! yyg->yy_did_buffer_switch_on_eof ) + YY_NEW_FILE; + } + break; + } + + case EOB_ACT_CONTINUE_SCAN: + yyg->yy_c_buf_p = + yyg->yytext_ptr + yy_amount_of_matched_text; + + yy_current_state = yy_get_previous_state( yyscanner ); + + yy_cp = yyg->yy_c_buf_p; + yy_bp = yyg->yytext_ptr + YY_MORE_ADJ; + goto yy_match; + + case EOB_ACT_LAST_MATCH: + yyg->yy_c_buf_p = + &YY_CURRENT_BUFFER_LVALUE->yy_ch_buf[yyg->yy_n_chars]; + + yy_current_state = yy_get_previous_state( yyscanner ); + + yy_cp = yyg->yy_c_buf_p; + yy_bp = yyg->yytext_ptr + YY_MORE_ADJ; + goto yy_find_action; + } + break; + } + + default: + YY_FATAL_ERROR( + "fatal flex scanner internal error--no action found" ); + } /* end of action switch */ + } /* end of scanning one token */ +} /* end of yylex */ + +/* yy_get_next_buffer - try to read in a new buffer + * + * Returns a code representing an action: + * EOB_ACT_LAST_MATCH - + * EOB_ACT_CONTINUE_SCAN - continue scanning from current position + * EOB_ACT_END_OF_FILE - end of file + */ +static int yy_get_next_buffer (yyscan_t yyscanner) +{ + struct yyguts_t * yyg = (struct yyguts_t*)yyscanner; + register char *dest = YY_CURRENT_BUFFER_LVALUE->yy_ch_buf; + register char *source = yyg->yytext_ptr; + register int number_to_move, i; + int ret_val; + + if ( yyg->yy_c_buf_p > &YY_CURRENT_BUFFER_LVALUE->yy_ch_buf[yyg->yy_n_chars + 1] ) + YY_FATAL_ERROR( + "fatal flex scanner internal error--end of buffer missed" ); + + if ( YY_CURRENT_BUFFER_LVALUE->yy_fill_buffer == 0 ) + { /* Don't try to fill the buffer, so this is an EOF. */ + if ( yyg->yy_c_buf_p - yyg->yytext_ptr - YY_MORE_ADJ == 1 ) + { + /* We matched a single character, the EOB, so + * treat this as a final EOF. + */ + return EOB_ACT_END_OF_FILE; + } + + else + { + /* We matched some text prior to the EOB, first + * process it. + */ + return EOB_ACT_LAST_MATCH; + } + } + + /* Try to read more data. */ + + /* First move last chars to start of buffer. */ + number_to_move = (int) (yyg->yy_c_buf_p - yyg->yytext_ptr) - 1; + + for ( i = 0; i < number_to_move; ++i ) + *(dest++) = *(source++); + + if ( YY_CURRENT_BUFFER_LVALUE->yy_buffer_status == YY_BUFFER_EOF_PENDING ) + /* don't do the read, it's not guaranteed to return an EOF, + * just force an EOF + */ + YY_CURRENT_BUFFER_LVALUE->yy_n_chars = yyg->yy_n_chars = 0; + + else + { + int num_to_read = + YY_CURRENT_BUFFER_LVALUE->yy_buf_size - number_to_move - 1; + + while ( num_to_read <= 0 ) + { /* Not enough room in the buffer - grow it. */ + + /* just a shorter name for the current buffer */ + YY_BUFFER_STATE b = YY_CURRENT_BUFFER; + + int yy_c_buf_p_offset = + (int) (yyg->yy_c_buf_p - b->yy_ch_buf); + + if ( b->yy_is_our_buffer ) + { + int new_size = b->yy_buf_size * 2; + + if ( new_size <= 0 ) + b->yy_buf_size += b->yy_buf_size / 8; + else + b->yy_buf_size *= 2; + + b->yy_ch_buf = (char *) + /* Include room in for 2 EOB chars. */ + yyrealloc((void *) b->yy_ch_buf,b->yy_buf_size + 2 ,yyscanner ); + } + else + /* Can't grow it, we don't own it. */ + b->yy_ch_buf = 0; + + if ( ! b->yy_ch_buf ) + YY_FATAL_ERROR( + "fatal error - scanner input buffer overflow" ); + + yyg->yy_c_buf_p = &b->yy_ch_buf[yy_c_buf_p_offset]; + + num_to_read = YY_CURRENT_BUFFER_LVALUE->yy_buf_size - + number_to_move - 1; + + } + + if ( num_to_read > YY_READ_BUF_SIZE ) + num_to_read = YY_READ_BUF_SIZE; + + /* Read in more data. */ + YY_INPUT( (&YY_CURRENT_BUFFER_LVALUE->yy_ch_buf[number_to_move]), + yyg->yy_n_chars, (size_t) num_to_read ); + + YY_CURRENT_BUFFER_LVALUE->yy_n_chars = yyg->yy_n_chars; + } + + if ( yyg->yy_n_chars == 0 ) + { + if ( number_to_move == YY_MORE_ADJ ) + { + ret_val = EOB_ACT_END_OF_FILE; + yyrestart(yyin ,yyscanner); + } + + else + { + ret_val = EOB_ACT_LAST_MATCH; + YY_CURRENT_BUFFER_LVALUE->yy_buffer_status = + YY_BUFFER_EOF_PENDING; + } + } + + else + ret_val = EOB_ACT_CONTINUE_SCAN; + + if ((yy_size_t) (yyg->yy_n_chars + number_to_move) > YY_CURRENT_BUFFER_LVALUE->yy_buf_size) { + /* Extend the array by 50%, plus the number we really need. */ + yy_size_t new_size = yyg->yy_n_chars + number_to_move + (yyg->yy_n_chars >> 1); + YY_CURRENT_BUFFER_LVALUE->yy_ch_buf = (char *) yyrealloc((void *) YY_CURRENT_BUFFER_LVALUE->yy_ch_buf,new_size ,yyscanner ); + if ( ! YY_CURRENT_BUFFER_LVALUE->yy_ch_buf ) + YY_FATAL_ERROR( "out of dynamic memory in yy_get_next_buffer()" ); + } + + yyg->yy_n_chars += number_to_move; + YY_CURRENT_BUFFER_LVALUE->yy_ch_buf[yyg->yy_n_chars] = YY_END_OF_BUFFER_CHAR; + YY_CURRENT_BUFFER_LVALUE->yy_ch_buf[yyg->yy_n_chars + 1] = YY_END_OF_BUFFER_CHAR; + + yyg->yytext_ptr = &YY_CURRENT_BUFFER_LVALUE->yy_ch_buf[0]; + + return ret_val; +} + +/* yy_get_previous_state - get the state just before the EOB char was reached */ + + static yy_state_type yy_get_previous_state (yyscan_t yyscanner) +{ + register yy_state_type yy_current_state; + register char *yy_cp; + struct yyguts_t * yyg = (struct yyguts_t*)yyscanner; + + yy_current_state = yyg->yy_start; + + for ( yy_cp = yyg->yytext_ptr + YY_MORE_ADJ; yy_cp < yyg->yy_c_buf_p; ++yy_cp ) + { + register YY_CHAR yy_c = (*yy_cp ? yy_ec[YY_SC_TO_UI(*yy_cp)] : 1); + if ( yy_accept[yy_current_state] ) + { + yyg->yy_last_accepting_state = yy_current_state; + yyg->yy_last_accepting_cpos = yy_cp; + } + while ( yy_chk[yy_base[yy_current_state] + yy_c] != yy_current_state ) + { + yy_current_state = (int) yy_def[yy_current_state]; + if ( yy_current_state >= 850 ) + yy_c = yy_meta[(unsigned int) yy_c]; + } + yy_current_state = yy_nxt[yy_base[yy_current_state] + (unsigned int) yy_c]; + } + + return yy_current_state; +} + +/* yy_try_NUL_trans - try to make a transition on the NUL character + * + * synopsis + * next_state = yy_try_NUL_trans( current_state ); + */ + static yy_state_type yy_try_NUL_trans (yy_state_type yy_current_state , yyscan_t yyscanner) +{ + register int yy_is_jam; + struct yyguts_t * yyg = (struct yyguts_t*)yyscanner; /* This var may be unused depending upon options. */ + register char *yy_cp = yyg->yy_c_buf_p; + + register YY_CHAR yy_c = 1; + if ( yy_accept[yy_current_state] ) + { + yyg->yy_last_accepting_state = yy_current_state; + yyg->yy_last_accepting_cpos = yy_cp; + } + while ( yy_chk[yy_base[yy_current_state] + yy_c] != yy_current_state ) + { + yy_current_state = (int) yy_def[yy_current_state]; + if ( yy_current_state >= 850 ) + yy_c = yy_meta[(unsigned int) yy_c]; + } + yy_current_state = yy_nxt[yy_base[yy_current_state] + (unsigned int) yy_c]; + yy_is_jam = (yy_current_state == 849); + + return yy_is_jam ? 0 : yy_current_state; +} + + static void yyunput (int c, register char * yy_bp , yyscan_t yyscanner) +{ + register char *yy_cp; + struct yyguts_t * yyg = (struct yyguts_t*)yyscanner; + + yy_cp = yyg->yy_c_buf_p; + + /* undo effects of setting up yytext */ + *yy_cp = yyg->yy_hold_char; + + if ( yy_cp < YY_CURRENT_BUFFER_LVALUE->yy_ch_buf + 2 ) + { /* need to shift things up to make room */ + /* +2 for EOB chars. */ + register int number_to_move = yyg->yy_n_chars + 2; + register char *dest = &YY_CURRENT_BUFFER_LVALUE->yy_ch_buf[ + YY_CURRENT_BUFFER_LVALUE->yy_buf_size + 2]; + register char *source = + &YY_CURRENT_BUFFER_LVALUE->yy_ch_buf[number_to_move]; + + while ( source > YY_CURRENT_BUFFER_LVALUE->yy_ch_buf ) + *--dest = *--source; + + yy_cp += (int) (dest - source); + yy_bp += (int) (dest - source); + YY_CURRENT_BUFFER_LVALUE->yy_n_chars = + yyg->yy_n_chars = YY_CURRENT_BUFFER_LVALUE->yy_buf_size; + + if ( yy_cp < YY_CURRENT_BUFFER_LVALUE->yy_ch_buf + 2 ) + YY_FATAL_ERROR( "flex scanner push-back overflow" ); + } + + *--yy_cp = (char) c; + + yyg->yytext_ptr = yy_bp; + yyg->yy_hold_char = *yy_cp; + yyg->yy_c_buf_p = yy_cp; +} + +#ifndef YY_NO_INPUT +#ifdef __cplusplus + static int yyinput (yyscan_t yyscanner) +#else + static int input (yyscan_t yyscanner) +#endif + +{ + int c; + struct yyguts_t * yyg = (struct yyguts_t*)yyscanner; + + *yyg->yy_c_buf_p = yyg->yy_hold_char; + + if ( *yyg->yy_c_buf_p == YY_END_OF_BUFFER_CHAR ) + { + /* yy_c_buf_p now points to the character we want to return. + * If this occurs *before* the EOB characters, then it's a + * valid NUL; if not, then we've hit the end of the buffer. + */ + if ( yyg->yy_c_buf_p < &YY_CURRENT_BUFFER_LVALUE->yy_ch_buf[yyg->yy_n_chars] ) + /* This was really a NUL. */ + *yyg->yy_c_buf_p = '\0'; + + else + { /* need more input */ + int offset = yyg->yy_c_buf_p - yyg->yytext_ptr; + ++yyg->yy_c_buf_p; + + switch ( yy_get_next_buffer( yyscanner ) ) + { + case EOB_ACT_LAST_MATCH: + /* This happens because yy_g_n_b() + * sees that we've accumulated a + * token and flags that we need to + * try matching the token before + * proceeding. But for input(), + * there's no matching to consider. + * So convert the EOB_ACT_LAST_MATCH + * to EOB_ACT_END_OF_FILE. + */ + + /* Reset buffer status. */ + yyrestart(yyin ,yyscanner); + + /*FALLTHROUGH*/ + + case EOB_ACT_END_OF_FILE: + { + if ( yywrap(yyscanner ) ) + return EOF; + + if ( ! yyg->yy_did_buffer_switch_on_eof ) + YY_NEW_FILE; +#ifdef __cplusplus + return yyinput(yyscanner); +#else + return input(yyscanner); +#endif + } + + case EOB_ACT_CONTINUE_SCAN: + yyg->yy_c_buf_p = yyg->yytext_ptr + offset; + break; + } + } + } + + c = *(unsigned char *) yyg->yy_c_buf_p; /* cast for 8-bit char's */ + *yyg->yy_c_buf_p = '\0'; /* preserve yytext */ + yyg->yy_hold_char = *++yyg->yy_c_buf_p; + + return c; +} +#endif /* ifndef YY_NO_INPUT */ + +/** Immediately switch to a different input stream. + * @param input_file A readable stream. + * @param yyscanner The scanner object. + * @note This function does not reset the start condition to @c INITIAL . + */ + void yyrestart (FILE * input_file , yyscan_t yyscanner) +{ + struct yyguts_t * yyg = (struct yyguts_t*)yyscanner; + + if ( ! YY_CURRENT_BUFFER ){ + yyensure_buffer_stack (yyscanner); + YY_CURRENT_BUFFER_LVALUE = + yy_create_buffer(yyin,YY_BUF_SIZE ,yyscanner); + } + + yy_init_buffer(YY_CURRENT_BUFFER,input_file ,yyscanner); + yy_load_buffer_state(yyscanner ); +} + +/** Switch to a different input buffer. + * @param new_buffer The new input buffer. + * @param yyscanner The scanner object. + */ + void yy_switch_to_buffer (YY_BUFFER_STATE new_buffer , yyscan_t yyscanner) +{ + struct yyguts_t * yyg = (struct yyguts_t*)yyscanner; + + /* TODO. We should be able to replace this entire function body + * with + * yypop_buffer_state(); + * yypush_buffer_state(new_buffer); + */ + yyensure_buffer_stack (yyscanner); + if ( YY_CURRENT_BUFFER == new_buffer ) + return; + + if ( YY_CURRENT_BUFFER ) + { + /* Flush out information for old buffer. */ + *yyg->yy_c_buf_p = yyg->yy_hold_char; + YY_CURRENT_BUFFER_LVALUE->yy_buf_pos = yyg->yy_c_buf_p; + YY_CURRENT_BUFFER_LVALUE->yy_n_chars = yyg->yy_n_chars; + } + + YY_CURRENT_BUFFER_LVALUE = new_buffer; + yy_load_buffer_state(yyscanner ); + + /* We don't actually know whether we did this switch during + * EOF (yywrap()) processing, but the only time this flag + * is looked at is after yywrap() is called, so it's safe + * to go ahead and always set it. + */ + yyg->yy_did_buffer_switch_on_eof = 1; +} + +static void yy_load_buffer_state (yyscan_t yyscanner) +{ + struct yyguts_t * yyg = (struct yyguts_t*)yyscanner; + yyg->yy_n_chars = YY_CURRENT_BUFFER_LVALUE->yy_n_chars; + yyg->yytext_ptr = yyg->yy_c_buf_p = YY_CURRENT_BUFFER_LVALUE->yy_buf_pos; + yyin = YY_CURRENT_BUFFER_LVALUE->yy_input_file; + yyg->yy_hold_char = *yyg->yy_c_buf_p; +} + +/** Allocate and initialize an input buffer state. + * @param file A readable stream. + * @param size The character buffer size in bytes. When in doubt, use @c YY_BUF_SIZE. + * @param yyscanner The scanner object. + * @return the allocated buffer state. + */ + YY_BUFFER_STATE yy_create_buffer (FILE * file, int size , yyscan_t yyscanner) +{ + YY_BUFFER_STATE b; + + b = (YY_BUFFER_STATE) yyalloc(sizeof( struct yy_buffer_state ) ,yyscanner ); + if ( ! b ) + YY_FATAL_ERROR( "out of dynamic memory in yy_create_buffer()" ); + + b->yy_buf_size = size; + + /* yy_ch_buf has to be 2 characters longer than the size given because + * we need to put in 2 end-of-buffer characters. + */ + b->yy_ch_buf = (char *) yyalloc(b->yy_buf_size + 2 ,yyscanner ); + if ( ! b->yy_ch_buf ) + YY_FATAL_ERROR( "out of dynamic memory in yy_create_buffer()" ); + + b->yy_is_our_buffer = 1; + + yy_init_buffer(b,file ,yyscanner); + + return b; +} + +/** Destroy the buffer. + * @param b a buffer created with yy_create_buffer() + * @param yyscanner The scanner object. + */ + void yy_delete_buffer (YY_BUFFER_STATE b , yyscan_t yyscanner) +{ + struct yyguts_t * yyg = (struct yyguts_t*)yyscanner; + + if ( ! b ) + return; + + if ( b == YY_CURRENT_BUFFER ) /* Not sure if we should pop here. */ + YY_CURRENT_BUFFER_LVALUE = (YY_BUFFER_STATE) 0; + + if ( b->yy_is_our_buffer ) + yyfree((void *) b->yy_ch_buf ,yyscanner ); + + yyfree((void *) b ,yyscanner ); +} + +/* Initializes or reinitializes a buffer. + * This function is sometimes called more than once on the same buffer, + * such as during a yyrestart() or at EOF. + */ + static void yy_init_buffer (YY_BUFFER_STATE b, FILE * file , yyscan_t yyscanner) + +{ + int oerrno = errno; + struct yyguts_t * yyg = (struct yyguts_t*)yyscanner; + + yy_flush_buffer(b ,yyscanner); + + b->yy_input_file = file; + b->yy_fill_buffer = 1; + + /* If b is the current buffer, then yy_init_buffer was _probably_ + * called from yyrestart() or through yy_get_next_buffer. + * In that case, we don't want to reset the lineno or column. + */ + if (b != YY_CURRENT_BUFFER){ + b->yy_bs_lineno = 1; + b->yy_bs_column = 0; + } + + b->yy_is_interactive = 0; + + errno = oerrno; +} + +/** Discard all buffered characters. On the next scan, YY_INPUT will be called. + * @param b the buffer state to be flushed, usually @c YY_CURRENT_BUFFER. + * @param yyscanner The scanner object. + */ + void yy_flush_buffer (YY_BUFFER_STATE b , yyscan_t yyscanner) +{ + struct yyguts_t * yyg = (struct yyguts_t*)yyscanner; + if ( ! b ) + return; + + b->yy_n_chars = 0; + + /* We always need two end-of-buffer characters. The first causes + * a transition to the end-of-buffer state. The second causes + * a jam in that state. + */ + b->yy_ch_buf[0] = YY_END_OF_BUFFER_CHAR; + b->yy_ch_buf[1] = YY_END_OF_BUFFER_CHAR; + + b->yy_buf_pos = &b->yy_ch_buf[0]; + + b->yy_at_bol = 1; + b->yy_buffer_status = YY_BUFFER_NEW; + + if ( b == YY_CURRENT_BUFFER ) + yy_load_buffer_state(yyscanner ); +} + +/** Pushes the new state onto the stack. The new state becomes + * the current state. This function will allocate the stack + * if necessary. + * @param new_buffer The new state. + * @param yyscanner The scanner object. + */ +void yypush_buffer_state (YY_BUFFER_STATE new_buffer , yyscan_t yyscanner) +{ + struct yyguts_t * yyg = (struct yyguts_t*)yyscanner; + if (new_buffer == NULL) + return; + + yyensure_buffer_stack(yyscanner); + + /* This block is copied from yy_switch_to_buffer. */ + if ( YY_CURRENT_BUFFER ) + { + /* Flush out information for old buffer. */ + *yyg->yy_c_buf_p = yyg->yy_hold_char; + YY_CURRENT_BUFFER_LVALUE->yy_buf_pos = yyg->yy_c_buf_p; + YY_CURRENT_BUFFER_LVALUE->yy_n_chars = yyg->yy_n_chars; + } + + /* Only push if top exists. Otherwise, replace top. */ + if (YY_CURRENT_BUFFER) + yyg->yy_buffer_stack_top++; + YY_CURRENT_BUFFER_LVALUE = new_buffer; + + /* copied from yy_switch_to_buffer. */ + yy_load_buffer_state(yyscanner ); + yyg->yy_did_buffer_switch_on_eof = 1; +} + +/** Removes and deletes the top of the stack, if present. + * The next element becomes the new top. + * @param yyscanner The scanner object. + */ +void yypop_buffer_state (yyscan_t yyscanner) +{ + struct yyguts_t * yyg = (struct yyguts_t*)yyscanner; + if (!YY_CURRENT_BUFFER) + return; + + yy_delete_buffer(YY_CURRENT_BUFFER ,yyscanner); + YY_CURRENT_BUFFER_LVALUE = NULL; + if (yyg->yy_buffer_stack_top > 0) + --yyg->yy_buffer_stack_top; + + if (YY_CURRENT_BUFFER) { + yy_load_buffer_state(yyscanner ); + yyg->yy_did_buffer_switch_on_eof = 1; + } +} + +/* Allocates the stack if it does not exist. + * Guarantees space for at least one push. + */ +static void yyensure_buffer_stack (yyscan_t yyscanner) +{ + int num_to_alloc; + struct yyguts_t * yyg = (struct yyguts_t*)yyscanner; + + if (!yyg->yy_buffer_stack) { + + /* First allocation is just for 2 elements, since we don't know if this + * scanner will even need a stack. We use 2 instead of 1 to avoid an + * immediate realloc on the next call. + */ + num_to_alloc = 1; + yyg->yy_buffer_stack = (struct yy_buffer_state**)yyalloc + (num_to_alloc * sizeof(struct yy_buffer_state*) + , yyscanner); + if ( ! yyg->yy_buffer_stack ) + YY_FATAL_ERROR( "out of dynamic memory in yyensure_buffer_stack()" ); + + memset(yyg->yy_buffer_stack, 0, num_to_alloc * sizeof(struct yy_buffer_state*)); + + yyg->yy_buffer_stack_max = num_to_alloc; + yyg->yy_buffer_stack_top = 0; + return; + } + + if (yyg->yy_buffer_stack_top >= (yyg->yy_buffer_stack_max) - 1){ + + /* Increase the buffer to prepare for a possible push. */ + int grow_size = 8 /* arbitrary grow size */; + + num_to_alloc = yyg->yy_buffer_stack_max + grow_size; + yyg->yy_buffer_stack = (struct yy_buffer_state**)yyrealloc + (yyg->yy_buffer_stack, + num_to_alloc * sizeof(struct yy_buffer_state*) + , yyscanner); + if ( ! yyg->yy_buffer_stack ) + YY_FATAL_ERROR( "out of dynamic memory in yyensure_buffer_stack()" ); + + /* zero only the new slots.*/ + memset(yyg->yy_buffer_stack + yyg->yy_buffer_stack_max, 0, grow_size * sizeof(struct yy_buffer_state*)); + yyg->yy_buffer_stack_max = num_to_alloc; + } +} + +/** Setup the input buffer state to scan directly from a user-specified character buffer. + * @param base the character buffer + * @param size the size in bytes of the character buffer + * @param yyscanner The scanner object. + * @return the newly allocated buffer state object. + */ +YY_BUFFER_STATE yy_scan_buffer (char * base, yy_size_t size , yyscan_t yyscanner) +{ + YY_BUFFER_STATE b; + + if ( size < 2 || + base[size-2] != YY_END_OF_BUFFER_CHAR || + base[size-1] != YY_END_OF_BUFFER_CHAR ) + /* They forgot to leave room for the EOB's. */ + return 0; + + b = (YY_BUFFER_STATE) yyalloc(sizeof( struct yy_buffer_state ) ,yyscanner ); + if ( ! b ) + YY_FATAL_ERROR( "out of dynamic memory in yy_scan_buffer()" ); + + b->yy_buf_size = size - 2; /* "- 2" to take care of EOB's */ + b->yy_buf_pos = b->yy_ch_buf = base; + b->yy_is_our_buffer = 0; + b->yy_input_file = 0; + b->yy_n_chars = b->yy_buf_size; + b->yy_is_interactive = 0; + b->yy_at_bol = 1; + b->yy_fill_buffer = 0; + b->yy_buffer_status = YY_BUFFER_NEW; + + yy_switch_to_buffer(b ,yyscanner ); + + return b; +} + +/** Setup the input buffer state to scan a string. The next call to yylex() will + * scan from a @e copy of @a str. + * @param yystr a NUL-terminated string to scan + * @param yyscanner The scanner object. + * @return the newly allocated buffer state object. + * @note If you want to scan bytes that may contain NUL values, then use + * yy_scan_bytes() instead. + */ +YY_BUFFER_STATE yy_scan_string (yyconst char * yystr , yyscan_t yyscanner) +{ + + return yy_scan_bytes(yystr,strlen(yystr) ,yyscanner); +} + +/** Setup the input buffer state to scan the given bytes. The next call to yylex() will + * scan from a @e copy of @a bytes. + * @param yybytes the byte buffer to scan + * @param _yybytes_len the number of bytes in the buffer pointed to by @a bytes. + * @param yyscanner The scanner object. + * @return the newly allocated buffer state object. + */ +YY_BUFFER_STATE yy_scan_bytes (yyconst char * yybytes, int _yybytes_len , yyscan_t yyscanner) +{ + YY_BUFFER_STATE b; + char *buf; + yy_size_t n; + int i; + + /* Get memory for full buffer, including space for trailing EOB's. */ + n = _yybytes_len + 2; + buf = (char *) yyalloc(n ,yyscanner ); + if ( ! buf ) + YY_FATAL_ERROR( "out of dynamic memory in yy_scan_bytes()" ); + + for ( i = 0; i < _yybytes_len; ++i ) + buf[i] = yybytes[i]; + + buf[_yybytes_len] = buf[_yybytes_len+1] = YY_END_OF_BUFFER_CHAR; + + b = yy_scan_buffer(buf,n ,yyscanner); + if ( ! b ) + YY_FATAL_ERROR( "bad buffer in yy_scan_bytes()" ); + + /* It's okay to grow etc. this buffer, and we should throw it + * away when we're done. + */ + b->yy_is_our_buffer = 1; + + return b; +} + +#ifndef YY_EXIT_FAILURE +#define YY_EXIT_FAILURE 2 +#endif + +static void yy_fatal_error (yyconst char* msg , yyscan_t yyscanner) +{ + (void) fprintf( stderr, "%s\n", msg ); + exit( YY_EXIT_FAILURE ); +} + +/* Redefine yyless() so it works in section 3 code. */ + +#undef yyless +#define yyless(n) \ + do \ + { \ + /* Undo effects of setting up yytext. */ \ + int yyless_macro_arg = (n); \ + YY_LESS_LINENO(yyless_macro_arg);\ + yytext[yyleng] = yyg->yy_hold_char; \ + yyg->yy_c_buf_p = yytext + yyless_macro_arg; \ + yyg->yy_hold_char = *yyg->yy_c_buf_p; \ + *yyg->yy_c_buf_p = '\0'; \ + yyleng = yyless_macro_arg; \ + } \ + while ( 0 ) + +/* Accessor methods (get/set functions) to struct members. */ + +/** Get the user-defined data for this scanner. + * @param yyscanner The scanner object. + */ +YY_EXTRA_TYPE yyget_extra (yyscan_t yyscanner) +{ + struct yyguts_t * yyg = (struct yyguts_t*)yyscanner; + return yyextra; +} + +/** Get the current line number. + * @param yyscanner The scanner object. + */ +int yyget_lineno (yyscan_t yyscanner) +{ + struct yyguts_t * yyg = (struct yyguts_t*)yyscanner; + + if (! YY_CURRENT_BUFFER) + return 0; + + return yylineno; +} + +/** Get the current column number. + * @param yyscanner The scanner object. + */ +int yyget_column (yyscan_t yyscanner) +{ + struct yyguts_t * yyg = (struct yyguts_t*)yyscanner; + + if (! YY_CURRENT_BUFFER) + return 0; + + return yycolumn; +} + +/** Get the input stream. + * @param yyscanner The scanner object. + */ +FILE *yyget_in (yyscan_t yyscanner) +{ + struct yyguts_t * yyg = (struct yyguts_t*)yyscanner; + return yyin; +} + +/** Get the output stream. + * @param yyscanner The scanner object. + */ +FILE *yyget_out (yyscan_t yyscanner) +{ + struct yyguts_t * yyg = (struct yyguts_t*)yyscanner; + return yyout; +} + +/** Get the length of the current token. + * @param yyscanner The scanner object. + */ +int yyget_leng (yyscan_t yyscanner) +{ + struct yyguts_t * yyg = (struct yyguts_t*)yyscanner; + return yyleng; +} + +/** Get the current token. + * @param yyscanner The scanner object. + */ + +char *yyget_text (yyscan_t yyscanner) +{ + struct yyguts_t * yyg = (struct yyguts_t*)yyscanner; + return yytext; +} + +/** Set the user-defined data. This data is never touched by the scanner. + * @param user_defined The data to be associated with this scanner. + * @param yyscanner The scanner object. + */ +void yyset_extra (YY_EXTRA_TYPE user_defined , yyscan_t yyscanner) +{ + struct yyguts_t * yyg = (struct yyguts_t*)yyscanner; + yyextra = user_defined ; +} + +/** Set the current line number. + * @param line_number + * @param yyscanner The scanner object. + */ +void yyset_lineno (int line_number , yyscan_t yyscanner) +{ + struct yyguts_t * yyg = (struct yyguts_t*)yyscanner; + + /* lineno is only valid if an input buffer exists. */ + if (! YY_CURRENT_BUFFER ) + yy_fatal_error( "yyset_lineno called with no buffer" , yyscanner); + + yylineno = line_number; +} + +/** Set the current column. + * @param line_number + * @param yyscanner The scanner object. + */ +void yyset_column (int column_no , yyscan_t yyscanner) +{ + struct yyguts_t * yyg = (struct yyguts_t*)yyscanner; + + /* column is only valid if an input buffer exists. */ + if (! YY_CURRENT_BUFFER ) + yy_fatal_error( "yyset_column called with no buffer" , yyscanner); + + yycolumn = column_no; +} + +/** Set the input stream. This does not discard the current + * input buffer. + * @param in_str A readable stream. + * @param yyscanner The scanner object. + * @see yy_switch_to_buffer + */ +void yyset_in (FILE * in_str , yyscan_t yyscanner) +{ + struct yyguts_t * yyg = (struct yyguts_t*)yyscanner; + yyin = in_str ; +} + +void yyset_out (FILE * out_str , yyscan_t yyscanner) +{ + struct yyguts_t * yyg = (struct yyguts_t*)yyscanner; + yyout = out_str ; +} + +int yyget_debug (yyscan_t yyscanner) +{ + struct yyguts_t * yyg = (struct yyguts_t*)yyscanner; + return yy_flex_debug; +} + +void yyset_debug (int bdebug , yyscan_t yyscanner) +{ + struct yyguts_t * yyg = (struct yyguts_t*)yyscanner; + yy_flex_debug = bdebug ; +} + +/* Accessor methods for yylval and yylloc */ + +YYSTYPE * yyget_lval (yyscan_t yyscanner) +{ + struct yyguts_t * yyg = (struct yyguts_t*)yyscanner; + return yylval; +} + +void yyset_lval (YYSTYPE * yylval_param , yyscan_t yyscanner) +{ + struct yyguts_t * yyg = (struct yyguts_t*)yyscanner; + yylval = yylval_param; +} + +YYLTYPE *yyget_lloc (yyscan_t yyscanner) +{ + struct yyguts_t * yyg = (struct yyguts_t*)yyscanner; + return yylloc; +} + +void yyset_lloc (YYLTYPE * yylloc_param , yyscan_t yyscanner) +{ + struct yyguts_t * yyg = (struct yyguts_t*)yyscanner; + yylloc = yylloc_param; +} + +/* User-visible API */ + +/* yylex_init is special because it creates the scanner itself, so it is + * the ONLY reentrant function that doesn't take the scanner as the last argument. + * That's why we explicitly handle the declaration, instead of using our macros. + */ + +int yylex_init(yyscan_t* ptr_yy_globals) + +{ + if (ptr_yy_globals == NULL){ + errno = EINVAL; + return 1; + } + + *ptr_yy_globals = (yyscan_t) yyalloc ( sizeof( struct yyguts_t ), NULL ); + + if (*ptr_yy_globals == NULL){ + errno = ENOMEM; + return 1; + } + + /* By setting to 0xAA, we expose bugs in yy_init_globals. Leave at 0x00 for releases. */ + memset(*ptr_yy_globals,0x00,sizeof(struct yyguts_t)); + + return yy_init_globals ( *ptr_yy_globals ); +} + +/* yylex_init_extra has the same functionality as yylex_init, but follows the + * convention of taking the scanner as the last argument. Note however, that + * this is a *pointer* to a scanner, as it will be allocated by this call (and + * is the reason, too, why this function also must handle its own declaration). + * The user defined value in the first argument will be available to yyalloc in + * the yyextra field. + */ + +int yylex_init_extra(YY_EXTRA_TYPE yy_user_defined,yyscan_t* ptr_yy_globals ) + +{ + struct yyguts_t dummy_yyguts; + + yyset_extra (yy_user_defined, &dummy_yyguts); + + if (ptr_yy_globals == NULL){ + errno = EINVAL; + return 1; + } + + *ptr_yy_globals = (yyscan_t) yyalloc ( sizeof( struct yyguts_t ), &dummy_yyguts ); + + if (*ptr_yy_globals == NULL){ + errno = ENOMEM; + return 1; + } + + /* By setting to 0xAA, we expose bugs in + yy_init_globals. Leave at 0x00 for releases. */ + memset(*ptr_yy_globals,0x00,sizeof(struct yyguts_t)); + + yyset_extra (yy_user_defined, *ptr_yy_globals); + + return yy_init_globals ( *ptr_yy_globals ); +} + +static int yy_init_globals (yyscan_t yyscanner) +{ + struct yyguts_t * yyg = (struct yyguts_t*)yyscanner; + /* Initialization is the same as for the non-reentrant scanner. + * This function is called from yylex_destroy(), so don't allocate here. + */ + + yyg->yy_buffer_stack = 0; + yyg->yy_buffer_stack_top = 0; + yyg->yy_buffer_stack_max = 0; + yyg->yy_c_buf_p = (char *) 0; + yyg->yy_init = 0; + yyg->yy_start = 0; + + yyg->yy_start_stack_ptr = 0; + yyg->yy_start_stack_depth = 0; + yyg->yy_start_stack = NULL; + +/* Defined in main.c */ +#ifdef YY_STDINIT + yyin = stdin; + yyout = stdout; +#else + yyin = (FILE *) 0; + yyout = (FILE *) 0; +#endif + + /* For future reference: Set errno on error, since we are called by + * yylex_init() + */ + return 0; +} + +/* yylex_destroy is for both reentrant and non-reentrant scanners. */ +int yylex_destroy (yyscan_t yyscanner) +{ + struct yyguts_t * yyg = (struct yyguts_t*)yyscanner; + + /* Pop the buffer stack, destroying each element. */ + while(YY_CURRENT_BUFFER){ + yy_delete_buffer(YY_CURRENT_BUFFER ,yyscanner ); + YY_CURRENT_BUFFER_LVALUE = NULL; + yypop_buffer_state(yyscanner); + } + + /* Destroy the stack itself. */ + yyfree(yyg->yy_buffer_stack ,yyscanner); + yyg->yy_buffer_stack = NULL; + + /* Destroy the start condition stack. */ + yyfree(yyg->yy_start_stack ,yyscanner ); + yyg->yy_start_stack = NULL; + + /* Reset the globals. This is important in a non-reentrant scanner so the next time + * yylex() is called, initialization will occur. */ + yy_init_globals( yyscanner); + + /* Destroy the main struct (reentrant only). */ + yyfree ( yyscanner , yyscanner ); + yyscanner = NULL; + return 0; +} + +/* + * Internal utility routines. + */ + +#ifndef yytext_ptr +static void yy_flex_strncpy (char* s1, yyconst char * s2, int n , yyscan_t yyscanner) +{ + register int i; + for ( i = 0; i < n; ++i ) + s1[i] = s2[i]; +} +#endif + +#ifdef YY_NEED_STRLEN +static int yy_flex_strlen (yyconst char * s , yyscan_t yyscanner) +{ + register int n; + for ( n = 0; s[n]; ++n ) + ; + + return n; +} +#endif + +void *yyalloc (yy_size_t size , yyscan_t yyscanner) +{ + return (void *) malloc( size ); +} + +void *yyrealloc (void * ptr, yy_size_t size , yyscan_t yyscanner) +{ + /* The cast to (char *) in the following accommodates both + * implementations that use char* generic pointers, and those + * that use void* generic pointers. It works with the latter + * because both ANSI C and C++ allow castless assignment from + * any pointer type to void*, and deal with argument conversions + * as though doing an assignment. + */ + return (void *) realloc( (char *) ptr, size ); +} + +void yyfree (void * ptr , yyscan_t yyscanner) +{ + free( (char *) ptr ); /* see yyrealloc() for (char *) cast */ +} + +#define YYTABLES_NAME "yytables" + +#line 493 "program_lexer.l" + + + +void +_mesa_program_lexer_ctor(void **scanner, struct asm_parser_state *state, + const char *string, size_t len) +{ + yylex_init_extra(state,scanner); + yy_scan_bytes(string,len,*scanner); +} + +void +_mesa_program_lexer_dtor(void *scanner) +{ + yylex_destroy(scanner); +} + diff --git a/mesalib/src/mesa/program/nvfragparse.c b/mesalib/src/mesa/program/nvfragparse.c new file mode 100644 index 000000000..0de3c5804 --- /dev/null +++ b/mesalib/src/mesa/program/nvfragparse.c @@ -0,0 +1,1588 @@ +/* + * Mesa 3-D graphics library + * Version: 6.5 + * + * Copyright (C) 1999-2005 Brian Paul All Rights Reserved. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included + * in all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS + * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * BRIAN PAUL BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN + * AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN + * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. + */ + +/** + * \file nvfragparse.c + * NVIDIA fragment program parser. + * \author Brian Paul + */ + +/* + * Regarding GL_NV_fragment_program: + * + * Portions of this software may use or implement intellectual + * property owned and licensed by NVIDIA Corporation. NVIDIA disclaims + * any and all warranties with respect to such intellectual property, + * including any use thereof or modifications thereto. + */ + +#include "main/glheader.h" +#include "main/context.h" +#include "main/imports.h" +#include "main/macros.h" +#include "program.h" +#include "prog_parameter.h" +#include "prog_print.h" +#include "prog_instruction.h" +#include "nvfragparse.h" + + +#define INPUT_1V 1 +#define INPUT_2V 2 +#define INPUT_3V 3 +#define INPUT_1S 4 +#define INPUT_2S 5 +#define INPUT_CC 6 +#define INPUT_1V_T 7 /* one source vector, plus textureId */ +#define INPUT_3V_T 8 /* one source vector, plus textureId */ +#define INPUT_NONE 9 +#define INPUT_1V_S 10 /* a string and a vector register */ +#define OUTPUT_V 20 +#define OUTPUT_S 21 +#define OUTPUT_NONE 22 + +/* IRIX defines some of these */ +#undef _R +#undef _H +#undef _X +#undef _C +#undef _S + +/* Optional suffixes */ +#define _R FLOAT32 /* float */ +#define _H FLOAT16 /* half-float */ +#define _X FIXED12 /* fixed */ +#define _C 0x08 /* set cond codes */ +#define _S 0x10 /* saturate, clamp result to [0,1] */ + +struct instruction_pattern { + const char *name; + enum prog_opcode opcode; + GLuint inputs; + GLuint outputs; + GLuint suffixes; +}; + +static const struct instruction_pattern Instructions[] = { + { "ADD", OPCODE_ADD, INPUT_2V, OUTPUT_V, _R | _H | _X | _C | _S }, + { "COS", OPCODE_COS, INPUT_1S, OUTPUT_S, _R | _H | _C | _S }, + { "DDX", OPCODE_DDX, INPUT_1V, OUTPUT_V, _R | _H | _C | _S }, + { "DDY", OPCODE_DDY, INPUT_1V, OUTPUT_V, _R | _H | _C | _S }, + { "DP3", OPCODE_DP3, INPUT_2V, OUTPUT_S, _R | _H | _X | _C | _S }, + { "DP4", OPCODE_DP4, INPUT_2V, OUTPUT_S, _R | _H | _X | _C | _S }, + { "DST", OPCODE_DP4, INPUT_2V, OUTPUT_V, _R | _H | _C | _S }, + { "EX2", OPCODE_DP4, INPUT_1S, OUTPUT_S, _R | _H | _C | _S }, + { "FLR", OPCODE_FLR, INPUT_1V, OUTPUT_V, _R | _H | _X | _C | _S }, + { "FRC", OPCODE_FRC, INPUT_1V, OUTPUT_V, _R | _H | _X | _C | _S }, + { "KIL", OPCODE_KIL_NV, INPUT_CC, OUTPUT_NONE, 0 }, + { "LG2", OPCODE_LG2, INPUT_1S, OUTPUT_S, _R | _H | _C | _S }, + { "LIT", OPCODE_LIT, INPUT_1V, OUTPUT_V, _R | _H | _C | _S }, + { "LRP", OPCODE_LRP, INPUT_3V, OUTPUT_V, _R | _H | _X | _C | _S }, + { "MAD", OPCODE_MAD, INPUT_3V, OUTPUT_V, _R | _H | _X | _C | _S }, + { "MAX", OPCODE_MAX, INPUT_2V, OUTPUT_V, _R | _H | _X | _C | _S }, + { "MIN", OPCODE_MIN, INPUT_2V, OUTPUT_V, _R | _H | _X | _C | _S }, + { "MOV", OPCODE_MOV, INPUT_1V, OUTPUT_V, _R | _H | _X | _C | _S }, + { "MUL", OPCODE_MUL, INPUT_2V, OUTPUT_V, _R | _H | _X | _C | _S }, + { "PK2H", OPCODE_PK2H, INPUT_1V, OUTPUT_S, 0 }, + { "PK2US", OPCODE_PK2US, INPUT_1V, OUTPUT_S, 0 }, + { "PK4B", OPCODE_PK4B, INPUT_1V, OUTPUT_S, 0 }, + { "PK4UB", OPCODE_PK4UB, INPUT_1V, OUTPUT_S, 0 }, + { "POW", OPCODE_POW, INPUT_2S, OUTPUT_S, _R | _H | _C | _S }, + { "RCP", OPCODE_RCP, INPUT_1S, OUTPUT_S, _R | _H | _C | _S }, + { "RFL", OPCODE_RFL, INPUT_2V, OUTPUT_V, _R | _H | _C | _S }, + { "RSQ", OPCODE_RSQ, INPUT_1S, OUTPUT_S, _R | _H | _C | _S }, + { "SEQ", OPCODE_SEQ, INPUT_2V, OUTPUT_V, _R | _H | _X | _C | _S }, + { "SFL", OPCODE_SFL, INPUT_2V, OUTPUT_V, _R | _H | _X | _C | _S }, + { "SGE", OPCODE_SGE, INPUT_2V, OUTPUT_V, _R | _H | _X | _C | _S }, + { "SGT", OPCODE_SGT, INPUT_2V, OUTPUT_V, _R | _H | _X | _C | _S }, + { "SIN", OPCODE_SIN, INPUT_1S, OUTPUT_S, _R | _H | _C | _S }, + { "SLE", OPCODE_SLE, INPUT_2V, OUTPUT_V, _R | _H | _X | _C | _S }, + { "SLT", OPCODE_SLT, INPUT_2V, OUTPUT_V, _R | _H | _X | _C | _S }, + { "SNE", OPCODE_SNE, INPUT_2V, OUTPUT_V, _R | _H | _X | _C | _S }, + { "STR", OPCODE_STR, INPUT_2V, OUTPUT_V, _R | _H | _X | _C | _S }, + { "SUB", OPCODE_SUB, INPUT_2V, OUTPUT_V, _R | _H | _X | _C | _S }, + { "TEX", OPCODE_TEX, INPUT_1V_T, OUTPUT_V, _C | _S }, + { "TXD", OPCODE_TXD, INPUT_3V_T, OUTPUT_V, _C | _S }, + { "TXP", OPCODE_TXP_NV, INPUT_1V_T, OUTPUT_V, _C | _S }, + { "UP2H", OPCODE_UP2H, INPUT_1S, OUTPUT_V, _C | _S }, + { "UP2US", OPCODE_UP2US, INPUT_1S, OUTPUT_V, _C | _S }, + { "UP4B", OPCODE_UP4B, INPUT_1S, OUTPUT_V, _C | _S }, + { "UP4UB", OPCODE_UP4UB, INPUT_1S, OUTPUT_V, _C | _S }, + { "X2D", OPCODE_X2D, INPUT_3V, OUTPUT_V, _R | _H | _C | _S }, + { "PRINT", OPCODE_PRINT, INPUT_1V_S, OUTPUT_NONE, 0 }, + { NULL, (enum prog_opcode) -1, 0, 0, 0 } +}; + + +/* + * Information needed or computed during parsing. + * Remember, we can't modify the target program object until we've + * _successfully_ parsed the program text. + */ +struct parse_state { + GLcontext *ctx; + const GLubyte *start; /* start of program string */ + const GLubyte *pos; /* current position */ + const GLubyte *curLine; + struct gl_fragment_program *program; /* current program */ + + struct gl_program_parameter_list *parameters; + + GLuint numInst; /* number of instructions parsed */ + GLuint inputsRead; /* bitmask of input registers used */ + GLuint outputsWritten; /* bitmask of 1 << FRAG_OUTPUT_* bits */ + GLuint texturesUsed[MAX_TEXTURE_IMAGE_UNITS]; +}; + + + +/* + * Called whenever we find an error during parsing. + */ +static void +record_error(struct parse_state *parseState, const char *msg, int lineNo) +{ +#ifdef DEBUG + GLint line, column; + const GLubyte *lineStr; + lineStr = _mesa_find_line_column(parseState->start, + parseState->pos, &line, &column); + _mesa_debug(parseState->ctx, + "nvfragparse.c(%d): line %d, column %d:%s (%s)\n", + lineNo, line, column, (char *) lineStr, msg); + free((void *) lineStr); +#else + (void) lineNo; +#endif + + /* Check that no error was already recorded. Only record the first one. */ + if (parseState->ctx->Program.ErrorString[0] == 0) { + _mesa_set_program_error(parseState->ctx, + parseState->pos - parseState->start, + msg); + } +} + + +#define RETURN_ERROR \ +do { \ + record_error(parseState, "Unexpected end of input.", __LINE__); \ + return GL_FALSE; \ +} while(0) + +#define RETURN_ERROR1(msg) \ +do { \ + record_error(parseState, msg, __LINE__); \ + return GL_FALSE; \ +} while(0) + +#define RETURN_ERROR2(msg1, msg2) \ +do { \ + char err[1000]; \ + sprintf(err, "%s %s", msg1, msg2); \ + record_error(parseState, err, __LINE__); \ + return GL_FALSE; \ +} while(0) + + + + +/* + * Search a list of instruction structures for a match. + */ +static struct instruction_pattern +MatchInstruction(const GLubyte *token) +{ + const struct instruction_pattern *inst; + struct instruction_pattern result; + + result.name = NULL; + result.opcode = MAX_OPCODE; /* i.e. invalid instruction */ + result.inputs = 0; + result.outputs = 0; + result.suffixes = 0; + + for (inst = Instructions; inst->name; inst++) { + if (strncmp((const char *) token, inst->name, 3) == 0) { + /* matched! */ + int i = 3; + result = *inst; + result.suffixes = 0; + /* look at suffix */ + if (token[i] == 'R') { + result.suffixes |= _R; + i++; + } + else if (token[i] == 'H') { + result.suffixes |= _H; + i++; + } + else if (token[i] == 'X') { + result.suffixes |= _X; + i++; + } + if (token[i] == 'C') { + result.suffixes |= _C; + i++; + } + if (token[i] == '_' && token[i+1] == 'S' && + token[i+2] == 'A' && token[i+3] == 'T') { + result.suffixes |= _S; + } + return result; + } + } + + return result; +} + + + + +/**********************************************************************/ + + +static GLboolean IsLetter(GLubyte b) +{ + return (b >= 'a' && b <= 'z') || + (b >= 'A' && b <= 'Z') || + (b == '_') || + (b == '$'); +} + + +static GLboolean IsDigit(GLubyte b) +{ + return b >= '0' && b <= '9'; +} + + +static GLboolean IsWhitespace(GLubyte b) +{ + return b == ' ' || b == '\t' || b == '\n' || b == '\r'; +} + + +/** + * Starting at 'str' find the next token. A token can be an integer, + * an identifier or punctuation symbol. + * \return <= 0 we found an error, else, return number of characters parsed. + */ +static GLint +GetToken(struct parse_state *parseState, GLubyte *token) +{ + const GLubyte *str = parseState->pos; + GLint i = 0, j = 0; + + token[0] = 0; + + /* skip whitespace and comments */ + while (str[i] && (IsWhitespace(str[i]) || str[i] == '#')) { + if (str[i] == '#') { + /* skip comment */ + while (str[i] && (str[i] != '\n' && str[i] != '\r')) { + i++; + } + if (str[i] == '\n' || str[i] == '\r') + parseState->curLine = str + i + 1; + } + else { + /* skip whitespace */ + if (str[i] == '\n' || str[i] == '\r') + parseState->curLine = str + i + 1; + i++; + } + } + + if (str[i] == 0) + return -i; + + /* try matching an integer */ + while (str[i] && IsDigit(str[i])) { + token[j++] = str[i++]; + } + if (j > 0 || !str[i]) { + token[j] = 0; + return i; + } + + /* try matching an identifier */ + if (IsLetter(str[i])) { + while (str[i] && (IsLetter(str[i]) || IsDigit(str[i]))) { + token[j++] = str[i++]; + } + token[j] = 0; + return i; + } + + /* punctuation character */ + if (str[i]) { + token[0] = str[i++]; + token[1] = 0; + return i; + } + + /* end of input */ + token[0] = 0; + return i; +} + + +/** + * Get next token from input stream and increment stream pointer past token. + */ +static GLboolean +Parse_Token(struct parse_state *parseState, GLubyte *token) +{ + GLint i; + i = GetToken(parseState, token); + if (i <= 0) { + parseState->pos += (-i); + return GL_FALSE; + } + parseState->pos += i; + return GL_TRUE; +} + + +/** + * Get next token from input stream but don't increment stream pointer. + */ +static GLboolean +Peek_Token(struct parse_state *parseState, GLubyte *token) +{ + GLint i, len; + i = GetToken(parseState, token); + if (i <= 0) { + parseState->pos += (-i); + return GL_FALSE; + } + len = (GLint) strlen((const char *) token); + parseState->pos += (i - len); + return GL_TRUE; +} + + +/**********************************************************************/ + +static const char *InputRegisters[MAX_NV_FRAGMENT_PROGRAM_INPUTS + 1] = { + "WPOS", "COL0", "COL1", "FOGC", + "TEX0", "TEX1", "TEX2", "TEX3", "TEX4", "TEX5", "TEX6", "TEX7", NULL +}; + + + +/**********************************************************************/ + +/** + * Try to match 'pattern' as the next token after any whitespace/comments. + */ +static GLboolean +Parse_String(struct parse_state *parseState, const char *pattern) +{ + const GLubyte *m; + GLint i; + + /* skip whitespace and comments */ + while (IsWhitespace(*parseState->pos) || *parseState->pos == '#') { + if (*parseState->pos == '#') { + while (*parseState->pos && (*parseState->pos != '\n' && *parseState->pos != '\r')) { + parseState->pos += 1; + } + if (*parseState->pos == '\n' || *parseState->pos == '\r') + parseState->curLine = parseState->pos + 1; + } + else { + /* skip whitespace */ + if (*parseState->pos == '\n' || *parseState->pos == '\r') + parseState->curLine = parseState->pos + 1; + parseState->pos += 1; + } + } + + /* Try to match the pattern */ + m = parseState->pos; + for (i = 0; pattern[i]; i++) { + if (*m != (GLubyte) pattern[i]) + return GL_FALSE; + m += 1; + } + parseState->pos = m; + + return GL_TRUE; /* success */ +} + + +static GLboolean +Parse_Identifier(struct parse_state *parseState, GLubyte *ident) +{ + if (!Parse_Token(parseState, ident)) + RETURN_ERROR; + if (IsLetter(ident[0])) + return GL_TRUE; + else + RETURN_ERROR1("Expected an identfier"); +} + + +/** + * Parse a floating point constant, or a defined symbol name. + * [+/-]N[.N[eN]] + * Output: number[0 .. 3] will get the value. + */ +static GLboolean +Parse_ScalarConstant(struct parse_state *parseState, GLfloat *number) +{ + char *end = NULL; + + *number = (GLfloat) _mesa_strtof((const char *) parseState->pos, &end); + + if (end && end > (char *) parseState->pos) { + /* got a number */ + parseState->pos = (GLubyte *) end; + number[1] = *number; + number[2] = *number; + number[3] = *number; + return GL_TRUE; + } + else { + /* should be an identifier */ + GLubyte ident[100]; + const GLfloat *constant; + if (!Parse_Identifier(parseState, ident)) + RETURN_ERROR1("Expected an identifier"); + constant = _mesa_lookup_parameter_value(parseState->parameters, + -1, (const char *) ident); + /* XXX Check that it's a constant and not a parameter */ + if (!constant) { + RETURN_ERROR1("Undefined symbol"); + } + else { + COPY_4V(number, constant); + return GL_TRUE; + } + } +} + + + +/** + * Parse a vector constant, one of: + * { float } + * { float, float } + * { float, float, float } + * { float, float, float, float } + */ +static GLboolean +Parse_VectorConstant(struct parse_state *parseState, GLfloat *vec) +{ + /* "{" was already consumed */ + + ASSIGN_4V(vec, 0.0, 0.0, 0.0, 1.0); + + if (!Parse_ScalarConstant(parseState, vec+0)) /* X */ + return GL_FALSE; + + if (Parse_String(parseState, "}")) { + return GL_TRUE; + } + + if (!Parse_String(parseState, ",")) + RETURN_ERROR1("Expected comma in vector constant"); + + if (!Parse_ScalarConstant(parseState, vec+1)) /* Y */ + return GL_FALSE; + + if (Parse_String(parseState, "}")) { + return GL_TRUE; + } + + if (!Parse_String(parseState, ",")) + RETURN_ERROR1("Expected comma in vector constant"); + + if (!Parse_ScalarConstant(parseState, vec+2)) /* Z */ + return GL_FALSE; + + if (Parse_String(parseState, "}")) { + return GL_TRUE; + } + + if (!Parse_String(parseState, ",")) + RETURN_ERROR1("Expected comma in vector constant"); + + if (!Parse_ScalarConstant(parseState, vec+3)) /* W */ + return GL_FALSE; + + if (!Parse_String(parseState, "}")) + RETURN_ERROR1("Expected closing brace in vector constant"); + + return GL_TRUE; +} + + +/** + * Parse <number>, <varname> or {a, b, c, d}. + * Return number of values in the vector or scalar, or zero if parse error. + */ +static GLuint +Parse_VectorOrScalarConstant(struct parse_state *parseState, GLfloat *vec) +{ + if (Parse_String(parseState, "{")) { + return Parse_VectorConstant(parseState, vec); + } + else { + GLboolean b = Parse_ScalarConstant(parseState, vec); + if (b) { + vec[1] = vec[2] = vec[3] = vec[0]; + } + return b; + } +} + + +/** + * Parse a texture image source: + * [TEX0 | TEX1 | .. | TEX15] , [1D | 2D | 3D | CUBE | RECT] + */ +static GLboolean +Parse_TextureImageId(struct parse_state *parseState, + GLubyte *texUnit, GLubyte *texTargetBit) +{ + GLubyte imageSrc[100]; + GLint unit; + + if (!Parse_Token(parseState, imageSrc)) + RETURN_ERROR; + + if (imageSrc[0] != 'T' || + imageSrc[1] != 'E' || + imageSrc[2] != 'X') { + RETURN_ERROR1("Expected TEX# source"); + } + unit = atoi((const char *) imageSrc + 3); + if ((unit < 0 || unit > MAX_TEXTURE_IMAGE_UNITS) || + (unit == 0 && (imageSrc[3] != '0' || imageSrc[4] != 0))) { + RETURN_ERROR1("Invalied TEX# source index"); + } + *texUnit = unit; + + if (!Parse_String(parseState, ",")) + RETURN_ERROR1("Expected ,"); + + if (Parse_String(parseState, "1D")) { + *texTargetBit = TEXTURE_1D_BIT; + } + else if (Parse_String(parseState, "2D")) { + *texTargetBit = TEXTURE_2D_BIT; + } + else if (Parse_String(parseState, "3D")) { + *texTargetBit = TEXTURE_3D_BIT; + } + else if (Parse_String(parseState, "CUBE")) { + *texTargetBit = TEXTURE_CUBE_BIT; + } + else if (Parse_String(parseState, "RECT")) { + *texTargetBit = TEXTURE_RECT_BIT; + } + else { + RETURN_ERROR1("Invalid texture target token"); + } + + /* update record of referenced texture units */ + parseState->texturesUsed[*texUnit] |= *texTargetBit; + if (_mesa_bitcount(parseState->texturesUsed[*texUnit]) > 1) { + RETURN_ERROR1("Only one texture target can be used per texture unit."); + } + + return GL_TRUE; +} + + +/** + * Parse a scalar suffix like .x, .y, .z or .w or parse a swizzle suffix + * like .wxyz, .xxyy, etc and return the swizzle indexes. + */ +static GLboolean +Parse_SwizzleSuffix(const GLubyte *token, GLuint swizzle[4]) +{ + if (token[1] == 0) { + /* single letter swizzle (scalar) */ + if (token[0] == 'x') + ASSIGN_4V(swizzle, 0, 0, 0, 0); + else if (token[0] == 'y') + ASSIGN_4V(swizzle, 1, 1, 1, 1); + else if (token[0] == 'z') + ASSIGN_4V(swizzle, 2, 2, 2, 2); + else if (token[0] == 'w') + ASSIGN_4V(swizzle, 3, 3, 3, 3); + else + return GL_FALSE; + } + else { + /* 4-component swizzle (vector) */ + GLint k; + for (k = 0; k < 4 && token[k]; k++) { + if (token[k] == 'x') + swizzle[k] = 0; + else if (token[k] == 'y') + swizzle[k] = 1; + else if (token[k] == 'z') + swizzle[k] = 2; + else if (token[k] == 'w') + swizzle[k] = 3; + else + return GL_FALSE; + } + if (k != 4) + return GL_FALSE; + } + return GL_TRUE; +} + + +static GLboolean +Parse_CondCodeMask(struct parse_state *parseState, + struct prog_dst_register *dstReg) +{ + if (Parse_String(parseState, "EQ")) + dstReg->CondMask = COND_EQ; + else if (Parse_String(parseState, "GE")) + dstReg->CondMask = COND_GE; + else if (Parse_String(parseState, "GT")) + dstReg->CondMask = COND_GT; + else if (Parse_String(parseState, "LE")) + dstReg->CondMask = COND_LE; + else if (Parse_String(parseState, "LT")) + dstReg->CondMask = COND_LT; + else if (Parse_String(parseState, "NE")) + dstReg->CondMask = COND_NE; + else if (Parse_String(parseState, "TR")) + dstReg->CondMask = COND_TR; + else if (Parse_String(parseState, "FL")) + dstReg->CondMask = COND_FL; + else + RETURN_ERROR1("Invalid condition code mask"); + + /* look for optional .xyzw swizzle */ + if (Parse_String(parseState, ".")) { + GLubyte token[100]; + GLuint swz[4]; + + if (!Parse_Token(parseState, token)) /* get xyzw suffix */ + RETURN_ERROR; + + if (!Parse_SwizzleSuffix(token, swz)) + RETURN_ERROR1("Invalid swizzle suffix"); + + dstReg->CondSwizzle = MAKE_SWIZZLE4(swz[0], swz[1], swz[2], swz[3]); + } + + return GL_TRUE; +} + + +/** + * Parse a temporary register: Rnn or Hnn + */ +static GLboolean +Parse_TempReg(struct parse_state *parseState, GLint *tempRegNum) +{ + GLubyte token[100]; + + /* Should be 'R##' or 'H##' */ + if (!Parse_Token(parseState, token)) + RETURN_ERROR; + if (token[0] != 'R' && token[0] != 'H') + RETURN_ERROR1("Expected R## or H##"); + + if (IsDigit(token[1])) { + GLint reg = atoi((const char *) (token + 1)); + if (token[0] == 'H') + reg += 32; + if (reg >= MAX_NV_FRAGMENT_PROGRAM_TEMPS) + RETURN_ERROR1("Invalid temporary register name"); + *tempRegNum = reg; + } + else { + RETURN_ERROR1("Invalid temporary register name"); + } + + return GL_TRUE; +} + + +/** + * Parse a write-only dummy register: RC or HC. + */ +static GLboolean +Parse_DummyReg(struct parse_state *parseState, GLint *regNum) +{ + if (Parse_String(parseState, "RC")) { + *regNum = 0; + } + else if (Parse_String(parseState, "HC")) { + *regNum = 1; + } + else { + RETURN_ERROR1("Invalid write-only register name"); + } + + return GL_TRUE; +} + + +/** + * Parse a program local parameter register "p[##]" + */ +static GLboolean +Parse_ProgramParamReg(struct parse_state *parseState, GLint *regNum) +{ + GLubyte token[100]; + + if (!Parse_String(parseState, "p[")) + RETURN_ERROR1("Expected p["); + + if (!Parse_Token(parseState, token)) + RETURN_ERROR; + + if (IsDigit(token[0])) { + /* a numbered program parameter register */ + GLint reg = atoi((const char *) token); + if (reg >= MAX_NV_FRAGMENT_PROGRAM_PARAMS) + RETURN_ERROR1("Invalid constant program number"); + *regNum = reg; + } + else { + RETURN_ERROR; + } + + if (!Parse_String(parseState, "]")) + RETURN_ERROR1("Expected ]"); + + return GL_TRUE; +} + + +/** + * Parse f[name] - fragment input register + */ +static GLboolean +Parse_FragReg(struct parse_state *parseState, GLint *tempRegNum) +{ + GLubyte token[100]; + GLint j; + + /* Match 'f[' */ + if (!Parse_String(parseState, "f[")) + RETURN_ERROR1("Expected f["); + + /* get <name> and look for match */ + if (!Parse_Token(parseState, token)) { + RETURN_ERROR; + } + for (j = 0; InputRegisters[j]; j++) { + if (strcmp((const char *) token, InputRegisters[j]) == 0) { + *tempRegNum = j; + parseState->inputsRead |= (1 << j); + break; + } + } + if (!InputRegisters[j]) { + /* unknown input register label */ + RETURN_ERROR2("Invalid register name", token); + } + + /* Match '[' */ + if (!Parse_String(parseState, "]")) + RETURN_ERROR1("Expected ]"); + + return GL_TRUE; +} + + +static GLboolean +Parse_OutputReg(struct parse_state *parseState, GLint *outputRegNum) +{ + GLubyte token[100]; + + /* Match "o[" */ + if (!Parse_String(parseState, "o[")) + RETURN_ERROR1("Expected o["); + + /* Get output reg name */ + if (!Parse_Token(parseState, token)) + RETURN_ERROR; + + /* try to match an output register name */ + if (strcmp((char *) token, "COLR") == 0 || + strcmp((char *) token, "COLH") == 0) { + /* note that we don't distinguish between COLR and COLH */ + *outputRegNum = FRAG_RESULT_COLOR; + parseState->outputsWritten |= (1 << FRAG_RESULT_COLOR); + } + else if (strcmp((char *) token, "DEPR") == 0) { + *outputRegNum = FRAG_RESULT_DEPTH; + parseState->outputsWritten |= (1 << FRAG_RESULT_DEPTH); + } + else { + RETURN_ERROR1("Invalid output register name"); + } + + /* Match ']' */ + if (!Parse_String(parseState, "]")) + RETURN_ERROR1("Expected ]"); + + return GL_TRUE; +} + + +static GLboolean +Parse_MaskedDstReg(struct parse_state *parseState, + struct prog_dst_register *dstReg) +{ + GLubyte token[100]; + GLint idx; + + /* Dst reg can be R<n>, H<n>, o[n], RC or HC */ + if (!Peek_Token(parseState, token)) + RETURN_ERROR; + + if (strcmp((const char *) token, "RC") == 0 || + strcmp((const char *) token, "HC") == 0) { + /* a write-only register */ + dstReg->File = PROGRAM_WRITE_ONLY; + if (!Parse_DummyReg(parseState, &idx)) + RETURN_ERROR; + dstReg->Index = idx; + } + else if (token[0] == 'R' || token[0] == 'H') { + /* a temporary register */ + dstReg->File = PROGRAM_TEMPORARY; + if (!Parse_TempReg(parseState, &idx)) + RETURN_ERROR; + dstReg->Index = idx; + } + else if (token[0] == 'o') { + /* an output register */ + dstReg->File = PROGRAM_OUTPUT; + if (!Parse_OutputReg(parseState, &idx)) + RETURN_ERROR; + dstReg->Index = idx; + } + else { + RETURN_ERROR1("Invalid destination register name"); + } + + /* Parse optional write mask */ + if (Parse_String(parseState, ".")) { + /* got a mask */ + GLint k = 0; + + if (!Parse_Token(parseState, token)) /* get xyzw writemask */ + RETURN_ERROR; + + dstReg->WriteMask = 0; + + if (token[k] == 'x') { + dstReg->WriteMask |= WRITEMASK_X; + k++; + } + if (token[k] == 'y') { + dstReg->WriteMask |= WRITEMASK_Y; + k++; + } + if (token[k] == 'z') { + dstReg->WriteMask |= WRITEMASK_Z; + k++; + } + if (token[k] == 'w') { + dstReg->WriteMask |= WRITEMASK_W; + k++; + } + if (k == 0) { + RETURN_ERROR1("Invalid writemask character"); + } + + } + else { + dstReg->WriteMask = WRITEMASK_XYZW; + } + + /* optional condition code mask */ + if (Parse_String(parseState, "(")) { + /* ("EQ" | "GE" | "GT" | "LE" | "LT" | "NE" | "TR" | "FL".x|y|z|w) */ + /* ("EQ" | "GE" | "GT" | "LE" | "LT" | "NE" | "TR" | "FL".[xyzw]) */ + if (!Parse_CondCodeMask(parseState, dstReg)) + RETURN_ERROR; + + if (!Parse_String(parseState, ")")) /* consume ")" */ + RETURN_ERROR1("Expected )"); + + return GL_TRUE; + } + else { + /* no cond code mask */ + dstReg->CondMask = COND_TR; + dstReg->CondSwizzle = SWIZZLE_NOOP; + return GL_TRUE; + } +} + + +/** + * Parse a vector source (register, constant, etc): + * <vectorSrc> ::= <absVectorSrc> + * | <baseVectorSrc> + * <absVectorSrc> ::= <negate> "|" <baseVectorSrc> "|" + */ +static GLboolean +Parse_VectorSrc(struct parse_state *parseState, + struct prog_src_register *srcReg) +{ + GLfloat sign = 1.0F; + GLubyte token[100]; + GLint idx; + GLuint negateBase, negateAbs; + + /* + * First, take care of +/- and absolute value stuff. + */ + if (Parse_String(parseState, "-")) + sign = -1.0F; + else if (Parse_String(parseState, "+")) + sign = +1.0F; + + if (Parse_String(parseState, "|")) { + srcReg->Abs = GL_TRUE; + negateAbs = (sign < 0.0F) ? NEGATE_XYZW : NEGATE_NONE; + + if (Parse_String(parseState, "-")) + negateBase = NEGATE_XYZW; + else if (Parse_String(parseState, "+")) + negateBase = NEGATE_NONE; + else + negateBase = NEGATE_NONE; + } + else { + srcReg->Abs = GL_FALSE; + negateAbs = NEGATE_NONE; + negateBase = (sign < 0.0F) ? NEGATE_XYZW : NEGATE_NONE; + } + + srcReg->Negate = srcReg->Abs ? negateAbs : negateBase; + + /* This should be the real src vector/register name */ + if (!Peek_Token(parseState, token)) + RETURN_ERROR; + + /* Src reg can be Rn, Hn, f[n], p[n], a named parameter, a scalar + * literal or vector literal. + */ + if (token[0] == 'R' || token[0] == 'H') { + srcReg->File = PROGRAM_TEMPORARY; + if (!Parse_TempReg(parseState, &idx)) + RETURN_ERROR; + srcReg->Index = idx; + } + else if (token[0] == 'f') { + /* XXX this might be an identifier! */ + srcReg->File = PROGRAM_INPUT; + if (!Parse_FragReg(parseState, &idx)) + RETURN_ERROR; + srcReg->Index = idx; + } + else if (token[0] == 'p') { + /* XXX this might be an identifier! */ + srcReg->File = PROGRAM_LOCAL_PARAM; + if (!Parse_ProgramParamReg(parseState, &idx)) + RETURN_ERROR; + srcReg->Index = idx; + } + else if (IsLetter(token[0])){ + GLubyte ident[100]; + GLint paramIndex; + if (!Parse_Identifier(parseState, ident)) + RETURN_ERROR; + paramIndex = _mesa_lookup_parameter_index(parseState->parameters, + -1, (const char *) ident); + if (paramIndex < 0) { + RETURN_ERROR2("Undefined constant or parameter: ", ident); + } + srcReg->File = PROGRAM_NAMED_PARAM; + srcReg->Index = paramIndex; + } + else if (IsDigit(token[0]) || token[0] == '-' || token[0] == '+' || token[0] == '.'){ + /* literal scalar constant */ + GLfloat values[4]; + GLuint paramIndex; + if (!Parse_ScalarConstant(parseState, values)) + RETURN_ERROR; + paramIndex = _mesa_add_unnamed_constant(parseState->parameters, + values, 4, NULL); + srcReg->File = PROGRAM_NAMED_PARAM; + srcReg->Index = paramIndex; + } + else if (token[0] == '{'){ + /* literal vector constant */ + GLfloat values[4]; + GLuint paramIndex; + (void) Parse_String(parseState, "{"); + if (!Parse_VectorConstant(parseState, values)) + RETURN_ERROR; + paramIndex = _mesa_add_unnamed_constant(parseState->parameters, + values, 4, NULL); + srcReg->File = PROGRAM_NAMED_PARAM; + srcReg->Index = paramIndex; + } + else { + RETURN_ERROR2("Invalid source register name", token); + } + + /* init swizzle fields */ + srcReg->Swizzle = SWIZZLE_NOOP; + + /* Look for optional swizzle suffix */ + if (Parse_String(parseState, ".")) { + GLuint swz[4]; + + if (!Parse_Token(parseState, token)) + RETURN_ERROR; + + if (!Parse_SwizzleSuffix(token, swz)) + RETURN_ERROR1("Invalid swizzle suffix"); + + srcReg->Swizzle = MAKE_SWIZZLE4(swz[0], swz[1], swz[2], swz[3]); + } + + /* Finish absolute value */ + if (srcReg->Abs && !Parse_String(parseState, "|")) { + RETURN_ERROR1("Expected |"); + } + + return GL_TRUE; +} + + +static GLboolean +Parse_ScalarSrcReg(struct parse_state *parseState, + struct prog_src_register *srcReg) +{ + GLubyte token[100]; + GLfloat sign = 1.0F; + GLboolean needSuffix = GL_TRUE; + GLint idx; + GLuint negateBase, negateAbs; + + /* + * First, take care of +/- and absolute value stuff. + */ + if (Parse_String(parseState, "-")) + sign = -1.0F; + else if (Parse_String(parseState, "+")) + sign = +1.0F; + + if (Parse_String(parseState, "|")) { + srcReg->Abs = GL_TRUE; + negateAbs = (sign < 0.0F) ? NEGATE_XYZW : NEGATE_NONE; + + if (Parse_String(parseState, "-")) + negateBase = NEGATE_XYZW; + else if (Parse_String(parseState, "+")) + negateBase = NEGATE_NONE; + else + negateBase = NEGATE_NONE; + } + else { + srcReg->Abs = GL_FALSE; + negateAbs = NEGATE_NONE; + negateBase = (sign < 0.0F) ? NEGATE_XYZW : NEGATE_NONE; + } + + srcReg->Negate = srcReg->Abs ? negateAbs : negateBase; + + if (!Peek_Token(parseState, token)) + RETURN_ERROR; + + /* Src reg can be R<n>, H<n> or a named fragment attrib */ + if (token[0] == 'R' || token[0] == 'H') { + srcReg->File = PROGRAM_TEMPORARY; + if (!Parse_TempReg(parseState, &idx)) + RETURN_ERROR; + srcReg->Index = idx; + } + else if (token[0] == 'f') { + srcReg->File = PROGRAM_INPUT; + if (!Parse_FragReg(parseState, &idx)) + RETURN_ERROR; + srcReg->Index = idx; + } + else if (token[0] == '{') { + /* vector literal */ + GLfloat values[4]; + GLuint paramIndex; + (void) Parse_String(parseState, "{"); + if (!Parse_VectorConstant(parseState, values)) + RETURN_ERROR; + paramIndex = _mesa_add_unnamed_constant(parseState->parameters, + values, 4, NULL); + srcReg->File = PROGRAM_NAMED_PARAM; + srcReg->Index = paramIndex; + } + else if (IsLetter(token[0])){ + /* named param/constant */ + GLubyte ident[100]; + GLint paramIndex; + if (!Parse_Identifier(parseState, ident)) + RETURN_ERROR; + paramIndex = _mesa_lookup_parameter_index(parseState->parameters, + -1, (const char *) ident); + if (paramIndex < 0) { + RETURN_ERROR2("Undefined constant or parameter: ", ident); + } + srcReg->File = PROGRAM_NAMED_PARAM; + srcReg->Index = paramIndex; + } + else if (IsDigit(token[0])) { + /* scalar literal */ + GLfloat values[4]; + GLuint paramIndex; + if (!Parse_ScalarConstant(parseState, values)) + RETURN_ERROR; + paramIndex = _mesa_add_unnamed_constant(parseState->parameters, + values, 4, NULL); + srcReg->Index = paramIndex; + srcReg->File = PROGRAM_NAMED_PARAM; + needSuffix = GL_FALSE; + } + else { + RETURN_ERROR2("Invalid scalar source argument", token); + } + + srcReg->Swizzle = 0; + if (needSuffix) { + /* parse .[xyzw] suffix */ + if (!Parse_String(parseState, ".")) + RETURN_ERROR1("Expected ."); + + if (!Parse_Token(parseState, token)) + RETURN_ERROR; + + if (token[0] == 'x' && token[1] == 0) { + srcReg->Swizzle = 0; + } + else if (token[0] == 'y' && token[1] == 0) { + srcReg->Swizzle = 1; + } + else if (token[0] == 'z' && token[1] == 0) { + srcReg->Swizzle = 2; + } + else if (token[0] == 'w' && token[1] == 0) { + srcReg->Swizzle = 3; + } + else { + RETURN_ERROR1("Invalid scalar source suffix"); + } + } + + /* Finish absolute value */ + if (srcReg->Abs && !Parse_String(parseState, "|")) { + RETURN_ERROR1("Expected |"); + } + + return GL_TRUE; +} + + +static GLboolean +Parse_PrintInstruction(struct parse_state *parseState, + struct prog_instruction *inst) +{ + const GLubyte *str; + GLubyte *msg; + GLuint len; + GLint idx; + + /* The first argument is a literal string 'just like this' */ + if (!Parse_String(parseState, "'")) + RETURN_ERROR1("Expected '"); + + str = parseState->pos; + for (len = 0; str[len] != '\''; len++) /* find closing quote */ + ; + parseState->pos += len + 1; + msg = (GLubyte*) malloc(len + 1); + + memcpy(msg, str, len); + msg[len] = 0; + inst->Data = msg; + + if (Parse_String(parseState, ",")) { + /* got an optional register to print */ + GLubyte token[100]; + GetToken(parseState, token); + if (token[0] == 'o') { + /* dst reg */ + if (!Parse_OutputReg(parseState, &idx)) + RETURN_ERROR; + inst->SrcReg[0].Index = idx; + inst->SrcReg[0].File = PROGRAM_OUTPUT; + } + else { + /* src reg */ + if (!Parse_VectorSrc(parseState, &inst->SrcReg[0])) + RETURN_ERROR; + } + } + else { + inst->SrcReg[0].File = PROGRAM_UNDEFINED; + } + + inst->SrcReg[0].Swizzle = SWIZZLE_NOOP; + inst->SrcReg[0].Abs = GL_FALSE; + inst->SrcReg[0].Negate = NEGATE_NONE; + + return GL_TRUE; +} + + +static GLboolean +Parse_InstructionSequence(struct parse_state *parseState, + struct prog_instruction program[]) +{ + while (1) { + struct prog_instruction *inst = program + parseState->numInst; + struct instruction_pattern instMatch; + GLubyte token[100]; + + /* Initialize the instruction */ + _mesa_init_instructions(inst, 1); + + /* special instructions */ + if (Parse_String(parseState, "DEFINE")) { + GLubyte id[100]; + GLfloat value[7]; /* yes, 7 to be safe */ + if (!Parse_Identifier(parseState, id)) + RETURN_ERROR; + /* XXX make sure id is not a reserved identifer, like R9 */ + if (!Parse_String(parseState, "=")) + RETURN_ERROR1("Expected ="); + if (!Parse_VectorOrScalarConstant(parseState, value)) + RETURN_ERROR; + if (!Parse_String(parseState, ";")) + RETURN_ERROR1("Expected ;"); + if (_mesa_lookup_parameter_index(parseState->parameters, + -1, (const char *) id) >= 0) { + RETURN_ERROR2(id, "already defined"); + } + _mesa_add_named_parameter(parseState->parameters, + (const char *) id, value); + } + else if (Parse_String(parseState, "DECLARE")) { + GLubyte id[100]; + GLfloat value[7] = {0, 0, 0, 0, 0, 0, 0}; /* yes, to be safe */ + if (!Parse_Identifier(parseState, id)) + RETURN_ERROR; + /* XXX make sure id is not a reserved identifer, like R9 */ + if (Parse_String(parseState, "=")) { + if (!Parse_VectorOrScalarConstant(parseState, value)) + RETURN_ERROR; + } + if (!Parse_String(parseState, ";")) + RETURN_ERROR1("Expected ;"); + if (_mesa_lookup_parameter_index(parseState->parameters, + -1, (const char *) id) >= 0) { + RETURN_ERROR2(id, "already declared"); + } + _mesa_add_named_parameter(parseState->parameters, + (const char *) id, value); + } + else if (Parse_String(parseState, "END")) { + inst->Opcode = OPCODE_END; + parseState->numInst++; + if (Parse_Token(parseState, token)) { + RETURN_ERROR1("Code after END opcode."); + } + break; + } + else { + /* general/arithmetic instruction */ + + /* get token */ + if (!Parse_Token(parseState, token)) { + RETURN_ERROR1("Missing END instruction."); + } + + /* try to find matching instuction */ + instMatch = MatchInstruction(token); + if (instMatch.opcode >= MAX_OPCODE) { + /* bad instruction name */ + RETURN_ERROR2("Unexpected token: ", token); + } + + inst->Opcode = instMatch.opcode; + inst->Precision = instMatch.suffixes & (_R | _H | _X); + inst->SaturateMode = (instMatch.suffixes & (_S)) + ? SATURATE_ZERO_ONE : SATURATE_OFF; + inst->CondUpdate = (instMatch.suffixes & (_C)) ? GL_TRUE : GL_FALSE; + + /* + * parse the input and output operands + */ + if (instMatch.outputs == OUTPUT_S || instMatch.outputs == OUTPUT_V) { + if (!Parse_MaskedDstReg(parseState, &inst->DstReg)) + RETURN_ERROR; + if (!Parse_String(parseState, ",")) + RETURN_ERROR1("Expected ,"); + } + else if (instMatch.outputs == OUTPUT_NONE) { + if (instMatch.opcode == OPCODE_KIL_NV) { + /* This is a little weird, the cond code info is in + * the dest register. + */ + if (!Parse_CondCodeMask(parseState, &inst->DstReg)) + RETURN_ERROR; + } + else { + ASSERT(instMatch.opcode == OPCODE_PRINT); + } + } + + if (instMatch.inputs == INPUT_1V) { + if (!Parse_VectorSrc(parseState, &inst->SrcReg[0])) + RETURN_ERROR; + } + else if (instMatch.inputs == INPUT_2V) { + if (!Parse_VectorSrc(parseState, &inst->SrcReg[0])) + RETURN_ERROR; + if (!Parse_String(parseState, ",")) + RETURN_ERROR1("Expected ,"); + if (!Parse_VectorSrc(parseState, &inst->SrcReg[1])) + RETURN_ERROR; + } + else if (instMatch.inputs == INPUT_3V) { + if (!Parse_VectorSrc(parseState, &inst->SrcReg[0])) + RETURN_ERROR; + if (!Parse_String(parseState, ",")) + RETURN_ERROR1("Expected ,"); + if (!Parse_VectorSrc(parseState, &inst->SrcReg[1])) + RETURN_ERROR; + if (!Parse_String(parseState, ",")) + RETURN_ERROR1("Expected ,"); + if (!Parse_VectorSrc(parseState, &inst->SrcReg[2])) + RETURN_ERROR; + } + else if (instMatch.inputs == INPUT_1S) { + if (!Parse_ScalarSrcReg(parseState, &inst->SrcReg[0])) + RETURN_ERROR; + } + else if (instMatch.inputs == INPUT_2S) { + if (!Parse_ScalarSrcReg(parseState, &inst->SrcReg[0])) + RETURN_ERROR; + if (!Parse_String(parseState, ",")) + RETURN_ERROR1("Expected ,"); + if (!Parse_ScalarSrcReg(parseState, &inst->SrcReg[1])) + RETURN_ERROR; + } + else if (instMatch.inputs == INPUT_CC) { + /* XXX to-do */ + } + else if (instMatch.inputs == INPUT_1V_T) { + GLubyte unit, idx; + if (!Parse_VectorSrc(parseState, &inst->SrcReg[0])) + RETURN_ERROR; + if (!Parse_String(parseState, ",")) + RETURN_ERROR1("Expected ,"); + if (!Parse_TextureImageId(parseState, &unit, &idx)) + RETURN_ERROR; + inst->TexSrcUnit = unit; + inst->TexSrcTarget = idx; + } + else if (instMatch.inputs == INPUT_3V_T) { + GLubyte unit, idx; + if (!Parse_VectorSrc(parseState, &inst->SrcReg[0])) + RETURN_ERROR; + if (!Parse_String(parseState, ",")) + RETURN_ERROR1("Expected ,"); + if (!Parse_VectorSrc(parseState, &inst->SrcReg[1])) + RETURN_ERROR; + if (!Parse_String(parseState, ",")) + RETURN_ERROR1("Expected ,"); + if (!Parse_VectorSrc(parseState, &inst->SrcReg[2])) + RETURN_ERROR; + if (!Parse_String(parseState, ",")) + RETURN_ERROR1("Expected ,"); + if (!Parse_TextureImageId(parseState, &unit, &idx)) + RETURN_ERROR; + inst->TexSrcUnit = unit; + inst->TexSrcTarget = idx; + } + else if (instMatch.inputs == INPUT_1V_S) { + if (!Parse_PrintInstruction(parseState, inst)) + RETURN_ERROR; + } + + /* end of statement semicolon */ + if (!Parse_String(parseState, ";")) + RETURN_ERROR1("Expected ;"); + + parseState->numInst++; + + if (parseState->numInst >= MAX_NV_FRAGMENT_PROGRAM_INSTRUCTIONS) + RETURN_ERROR1("Program too long"); + } + } + return GL_TRUE; +} + + + +/** + * Parse/compile the 'str' returning the compiled 'program'. + * ctx->Program.ErrorPos will be -1 if successful. Otherwise, ErrorPos + * indicates the position of the error in 'str'. + */ +void +_mesa_parse_nv_fragment_program(GLcontext *ctx, GLenum dstTarget, + const GLubyte *str, GLsizei len, + struct gl_fragment_program *program) +{ + struct parse_state parseState; + struct prog_instruction instBuffer[MAX_NV_FRAGMENT_PROGRAM_INSTRUCTIONS]; + struct prog_instruction *newInst; + GLenum target; + GLubyte *programString; + + /* Make a null-terminated copy of the program string */ + programString = (GLubyte *) MALLOC(len + 1); + if (!programString) { + _mesa_error(ctx, GL_OUT_OF_MEMORY, "glLoadProgramNV"); + return; + } + memcpy(programString, str, len); + programString[len] = 0; + + /* Get ready to parse */ + memset(&parseState, 0, sizeof(struct parse_state)); + parseState.ctx = ctx; + parseState.start = programString; + parseState.program = program; + parseState.numInst = 0; + parseState.curLine = programString; + parseState.parameters = _mesa_new_parameter_list(); + + /* Reset error state */ + _mesa_set_program_error(ctx, -1, NULL); + + /* check the program header */ + if (strncmp((const char *) programString, "!!FP1.0", 7) == 0) { + target = GL_FRAGMENT_PROGRAM_NV; + parseState.pos = programString + 7; + } + else if (strncmp((const char *) programString, "!!FCP1.0", 8) == 0) { + /* fragment / register combiner program - not supported */ + _mesa_set_program_error(ctx, 0, "Invalid fragment program header"); + _mesa_error(ctx, GL_INVALID_OPERATION, "glLoadProgramNV(bad header)"); + return; + } + else { + /* invalid header */ + _mesa_set_program_error(ctx, 0, "Invalid fragment program header"); + _mesa_error(ctx, GL_INVALID_OPERATION, "glLoadProgramNV(bad header)"); + return; + } + + /* make sure target and header match */ + if (target != dstTarget) { + _mesa_error(ctx, GL_INVALID_OPERATION, + "glLoadProgramNV(target mismatch 0x%x != 0x%x)", + target, dstTarget); + return; + } + + if (Parse_InstructionSequence(&parseState, instBuffer)) { + GLuint u; + /* successful parse! */ + + if (parseState.outputsWritten == 0) { + /* must write at least one output! */ + _mesa_error(ctx, GL_INVALID_OPERATION, + "Invalid fragment program - no outputs written."); + return; + } + + /* copy the compiled instructions */ + assert(parseState.numInst <= MAX_NV_FRAGMENT_PROGRAM_INSTRUCTIONS); + newInst = _mesa_alloc_instructions(parseState.numInst); + if (!newInst) { + _mesa_error(ctx, GL_OUT_OF_MEMORY, "glLoadProgramNV"); + return; /* out of memory */ + } + _mesa_copy_instructions(newInst, instBuffer, parseState.numInst); + + /* install the program */ + program->Base.Target = target; + if (program->Base.String) { + FREE(program->Base.String); + } + program->Base.String = programString; + program->Base.Format = GL_PROGRAM_FORMAT_ASCII_ARB; + if (program->Base.Instructions) { + free(program->Base.Instructions); + } + program->Base.Instructions = newInst; + program->Base.NumInstructions = parseState.numInst; + program->Base.InputsRead = parseState.inputsRead; + program->Base.OutputsWritten = parseState.outputsWritten; + for (u = 0; u < ctx->Const.MaxTextureImageUnits; u++) + program->Base.TexturesUsed[u] = parseState.texturesUsed[u]; + + /* save program parameters */ + program->Base.Parameters = parseState.parameters; + + /* allocate registers for declared program parameters */ +#if 00 + _mesa_assign_program_registers(&(program->SymbolTable)); +#endif + +#ifdef DEBUG_foo + printf("--- glLoadProgramNV(%d) result ---\n", program->Base.Id); + _mesa_fprint_program_opt(stdout, &program->Base, PROG_PRINT_NV, 0); + printf("----------------------------------\n"); +#endif + } + else { + /* Error! */ + _mesa_error(ctx, GL_INVALID_OPERATION, "glLoadProgramNV"); + /* NOTE: _mesa_set_program_error would have been called already */ + } +} + + +const char * +_mesa_nv_fragment_input_register_name(GLuint i) +{ + ASSERT(i < MAX_NV_FRAGMENT_PROGRAM_INPUTS); + return InputRegisters[i]; +} + diff --git a/mesalib/src/mesa/program/nvfragparse.h b/mesalib/src/mesa/program/nvfragparse.h new file mode 100644 index 000000000..e28a6c493 --- /dev/null +++ b/mesalib/src/mesa/program/nvfragparse.h @@ -0,0 +1,44 @@ + +/* + * Mesa 3-D graphics library + * Version: 5.1 + * + * Copyright (C) 1999-2002 Brian Paul All Rights Reserved. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included + * in all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS + * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * BRIAN PAUL BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN + * AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN + * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. + * + * Authors: + * Brian Paul + */ + + +#ifndef NVFRAGPARSE_H +#define NVFRAGPARSE_H + +#include "main/mtypes.h" + +extern void +_mesa_parse_nv_fragment_program(GLcontext *ctx, GLenum target, + const GLubyte *str, GLsizei len, + struct gl_fragment_program *program); + + +extern const char * +_mesa_nv_fragment_input_register_name(GLuint i); + +#endif diff --git a/mesalib/src/mesa/program/nvvertparse.c b/mesalib/src/mesa/program/nvvertparse.c new file mode 100644 index 000000000..1ac83d0e5 --- /dev/null +++ b/mesalib/src/mesa/program/nvvertparse.c @@ -0,0 +1,1459 @@ +/* + * Mesa 3-D graphics library + * Version: 6.5.2 + * + * Copyright (C) 1999-2006 Brian Paul All Rights Reserved. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included + * in all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS + * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * BRIAN PAUL BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN + * AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN + * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. + */ + +/** + * \file nvvertparse.c + * NVIDIA vertex program parser. + * \author Brian Paul + */ + +/* + * Regarding GL_NV_vertex_program, GL_NV_vertex_program1_1: + * + * Portions of this software may use or implement intellectual + * property owned and licensed by NVIDIA Corporation. NVIDIA disclaims + * any and all warranties with respect to such intellectual property, + * including any use thereof or modifications thereto. + */ + +#include "main/glheader.h" +#include "main/context.h" +#include "main/imports.h" +#include "main/nvprogram.h" +#include "nvvertparse.h" +#include "prog_instruction.h" +#include "prog_parameter.h" +#include "prog_print.h" +#include "program.h" + + +/** + * Current parsing state. This structure is passed among the parsing + * functions and keeps track of the current parser position and various + * program attributes. + */ +struct parse_state { + GLcontext *ctx; + const GLubyte *start; + const GLubyte *pos; + const GLubyte *curLine; + GLboolean isStateProgram; + GLboolean isPositionInvariant; + GLboolean isVersion1_1; + GLbitfield inputsRead; + GLbitfield outputsWritten; + GLboolean anyProgRegsWritten; + GLboolean indirectRegisterFiles; + GLuint numInst; /* number of instructions parsed */ +}; + + +/* + * Called whenever we find an error during parsing. + */ +static void +record_error(struct parse_state *parseState, const char *msg, int lineNo) +{ +#ifdef DEBUG + GLint line, column; + const GLubyte *lineStr; + lineStr = _mesa_find_line_column(parseState->start, + parseState->pos, &line, &column); + _mesa_debug(parseState->ctx, + "nvfragparse.c(%d): line %d, column %d:%s (%s)\n", + lineNo, line, column, (char *) lineStr, msg); + free((void *) lineStr); +#else + (void) lineNo; +#endif + + /* Check that no error was already recorded. Only record the first one. */ + if (parseState->ctx->Program.ErrorString[0] == 0) { + _mesa_set_program_error(parseState->ctx, + parseState->pos - parseState->start, + msg); + } +} + + +#define RETURN_ERROR \ +do { \ + record_error(parseState, "Unexpected end of input.", __LINE__); \ + return GL_FALSE; \ +} while(0) + +#define RETURN_ERROR1(msg) \ +do { \ + record_error(parseState, msg, __LINE__); \ + return GL_FALSE; \ +} while(0) + +#define RETURN_ERROR2(msg1, msg2) \ +do { \ + char err[1000]; \ + sprintf(err, "%s %s", msg1, msg2); \ + record_error(parseState, err, __LINE__); \ + return GL_FALSE; \ +} while(0) + + + + + +static GLboolean IsLetter(GLubyte b) +{ + return (b >= 'a' && b <= 'z') || (b >= 'A' && b <= 'Z'); +} + + +static GLboolean IsDigit(GLubyte b) +{ + return b >= '0' && b <= '9'; +} + + +static GLboolean IsWhitespace(GLubyte b) +{ + return b == ' ' || b == '\t' || b == '\n' || b == '\r'; +} + + +/** + * Starting at 'str' find the next token. A token can be an integer, + * an identifier or punctuation symbol. + * \return <= 0 we found an error, else, return number of characters parsed. + */ +static GLint +GetToken(struct parse_state *parseState, GLubyte *token) +{ + const GLubyte *str = parseState->pos; + GLint i = 0, j = 0; + + token[0] = 0; + + /* skip whitespace and comments */ + while (str[i] && (IsWhitespace(str[i]) || str[i] == '#')) { + if (str[i] == '#') { + /* skip comment */ + while (str[i] && (str[i] != '\n' && str[i] != '\r')) { + i++; + } + if (str[i] == '\n' || str[i] == '\r') + parseState->curLine = str + i + 1; + } + else { + /* skip whitespace */ + if (str[i] == '\n' || str[i] == '\r') + parseState->curLine = str + i + 1; + i++; + } + } + + if (str[i] == 0) + return -i; + + /* try matching an integer */ + while (str[i] && IsDigit(str[i])) { + token[j++] = str[i++]; + } + if (j > 0 || !str[i]) { + token[j] = 0; + return i; + } + + /* try matching an identifier */ + if (IsLetter(str[i])) { + while (str[i] && (IsLetter(str[i]) || IsDigit(str[i]))) { + token[j++] = str[i++]; + } + token[j] = 0; + return i; + } + + /* punctuation character */ + if (str[i]) { + token[0] = str[i++]; + token[1] = 0; + return i; + } + + /* end of input */ + token[0] = 0; + return i; +} + + +/** + * Get next token from input stream and increment stream pointer past token. + */ +static GLboolean +Parse_Token(struct parse_state *parseState, GLubyte *token) +{ + GLint i; + i = GetToken(parseState, token); + if (i <= 0) { + parseState->pos += (-i); + return GL_FALSE; + } + parseState->pos += i; + return GL_TRUE; +} + + +/** + * Get next token from input stream but don't increment stream pointer. + */ +static GLboolean +Peek_Token(struct parse_state *parseState, GLubyte *token) +{ + GLint i, len; + i = GetToken(parseState, token); + if (i <= 0) { + parseState->pos += (-i); + return GL_FALSE; + } + len = (GLint) strlen((const char *) token); + parseState->pos += (i - len); + return GL_TRUE; +} + + +/** + * Try to match 'pattern' as the next token after any whitespace/comments. + * Advance the current parsing position only if we match the pattern. + * \return GL_TRUE if pattern is matched, GL_FALSE otherwise. + */ +static GLboolean +Parse_String(struct parse_state *parseState, const char *pattern) +{ + const GLubyte *m; + GLint i; + + /* skip whitespace and comments */ + while (IsWhitespace(*parseState->pos) || *parseState->pos == '#') { + if (*parseState->pos == '#') { + while (*parseState->pos && (*parseState->pos != '\n' && *parseState->pos != '\r')) { + parseState->pos += 1; + } + if (*parseState->pos == '\n' || *parseState->pos == '\r') + parseState->curLine = parseState->pos + 1; + } + else { + /* skip whitespace */ + if (*parseState->pos == '\n' || *parseState->pos == '\r') + parseState->curLine = parseState->pos + 1; + parseState->pos += 1; + } + } + + /* Try to match the pattern */ + m = parseState->pos; + for (i = 0; pattern[i]; i++) { + if (*m != (GLubyte) pattern[i]) + return GL_FALSE; + m += 1; + } + parseState->pos = m; + + return GL_TRUE; /* success */ +} + + +/**********************************************************************/ + +static const char *InputRegisters[MAX_NV_VERTEX_PROGRAM_INPUTS + 1] = { + "OPOS", "WGHT", "NRML", "COL0", "COL1", "FOGC", "6", "7", + "TEX0", "TEX1", "TEX2", "TEX3", "TEX4", "TEX5", "TEX6", "TEX7", NULL +}; + +static const char *OutputRegisters[MAX_NV_VERTEX_PROGRAM_OUTPUTS + 1] = { + "HPOS", "COL0", "COL1", "FOGC", + "TEX0", "TEX1", "TEX2", "TEX3", "TEX4", "TEX5", "TEX6", "TEX7", + "PSIZ", "BFC0", "BFC1", NULL +}; + + + +/** + * Parse a temporary register: Rnn + */ +static GLboolean +Parse_TempReg(struct parse_state *parseState, GLint *tempRegNum) +{ + GLubyte token[100]; + + /* Should be 'R##' */ + if (!Parse_Token(parseState, token)) + RETURN_ERROR; + if (token[0] != 'R') + RETURN_ERROR1("Expected R##"); + + if (IsDigit(token[1])) { + GLint reg = atoi((char *) (token + 1)); + if (reg >= MAX_NV_VERTEX_PROGRAM_TEMPS) + RETURN_ERROR1("Bad temporary register name"); + *tempRegNum = reg; + } + else { + RETURN_ERROR1("Bad temporary register name"); + } + + return GL_TRUE; +} + + +/** + * Parse address register "A0.x" + */ +static GLboolean +Parse_AddrReg(struct parse_state *parseState) +{ + /* match 'A0' */ + if (!Parse_String(parseState, "A0")) + RETURN_ERROR; + + /* match '.' */ + if (!Parse_String(parseState, ".")) + RETURN_ERROR; + + /* match 'x' */ + if (!Parse_String(parseState, "x")) + RETURN_ERROR; + + return GL_TRUE; +} + + +/** + * Parse absolute program parameter register "c[##]" + */ +static GLboolean +Parse_AbsParamReg(struct parse_state *parseState, GLint *regNum) +{ + GLubyte token[100]; + + if (!Parse_String(parseState, "c")) + RETURN_ERROR; + + if (!Parse_String(parseState, "[")) + RETURN_ERROR; + + if (!Parse_Token(parseState, token)) + RETURN_ERROR; + + if (IsDigit(token[0])) { + /* a numbered program parameter register */ + GLint reg = atoi((char *) token); + if (reg >= MAX_NV_VERTEX_PROGRAM_PARAMS) + RETURN_ERROR1("Bad program parameter number"); + *regNum = reg; + } + else { + RETURN_ERROR; + } + + if (!Parse_String(parseState, "]")) + RETURN_ERROR; + + return GL_TRUE; +} + + +static GLboolean +Parse_ParamReg(struct parse_state *parseState, struct prog_src_register *srcReg) +{ + GLubyte token[100]; + + if (!Parse_String(parseState, "c")) + RETURN_ERROR; + + if (!Parse_String(parseState, "[")) + RETURN_ERROR; + + if (!Peek_Token(parseState, token)) + RETURN_ERROR; + + if (IsDigit(token[0])) { + /* a numbered program parameter register */ + GLint reg; + (void) Parse_Token(parseState, token); + reg = atoi((char *) token); + if (reg >= MAX_NV_VERTEX_PROGRAM_PARAMS) + RETURN_ERROR1("Bad program parameter number"); + srcReg->File = PROGRAM_ENV_PARAM; + srcReg->Index = reg; + } + else if (strcmp((const char *) token, "A0") == 0) { + /* address register "A0.x" */ + if (!Parse_AddrReg(parseState)) + RETURN_ERROR; + + srcReg->RelAddr = GL_TRUE; + srcReg->File = PROGRAM_ENV_PARAM; + parseState->indirectRegisterFiles |= (1 << srcReg->File); + /* Look for +/-N offset */ + if (!Peek_Token(parseState, token)) + RETURN_ERROR; + + if (token[0] == '-' || token[0] == '+') { + const GLubyte sign = token[0]; + (void) Parse_Token(parseState, token); /* consume +/- */ + + /* an integer should be next */ + if (!Parse_Token(parseState, token)) + RETURN_ERROR; + + if (IsDigit(token[0])) { + const GLint k = atoi((char *) token); + if (sign == '-') { + if (k > 64) + RETURN_ERROR1("Bad address offset"); + srcReg->Index = -k; + } + else { + if (k > 63) + RETURN_ERROR1("Bad address offset"); + srcReg->Index = k; + } + } + else { + RETURN_ERROR; + } + } + else { + /* probably got a ']', catch it below */ + } + } + else { + RETURN_ERROR; + } + + /* Match closing ']' */ + if (!Parse_String(parseState, "]")) + RETURN_ERROR; + + return GL_TRUE; +} + + +/** + * Parse v[#] or v[<name>] + */ +static GLboolean +Parse_AttribReg(struct parse_state *parseState, GLint *tempRegNum) +{ + GLubyte token[100]; + GLint j; + + /* Match 'v' */ + if (!Parse_String(parseState, "v")) + RETURN_ERROR; + + /* Match '[' */ + if (!Parse_String(parseState, "[")) + RETURN_ERROR; + + /* match number or named register */ + if (!Parse_Token(parseState, token)) + RETURN_ERROR; + + if (parseState->isStateProgram && token[0] != '0') + RETURN_ERROR1("Only v[0] accessible in vertex state programs"); + + if (IsDigit(token[0])) { + GLint reg = atoi((char *) token); + if (reg >= MAX_NV_VERTEX_PROGRAM_INPUTS) + RETURN_ERROR1("Bad vertex attribute register name"); + *tempRegNum = reg; + } + else { + for (j = 0; InputRegisters[j]; j++) { + if (strcmp((const char *) token, InputRegisters[j]) == 0) { + *tempRegNum = j; + break; + } + } + if (!InputRegisters[j]) { + /* unknown input register label */ + RETURN_ERROR2("Bad register name", token); + } + } + + /* Match '[' */ + if (!Parse_String(parseState, "]")) + RETURN_ERROR; + + return GL_TRUE; +} + + +static GLboolean +Parse_OutputReg(struct parse_state *parseState, GLint *outputRegNum) +{ + GLubyte token[100]; + GLint start, j; + + /* Match 'o' */ + if (!Parse_String(parseState, "o")) + RETURN_ERROR; + + /* Match '[' */ + if (!Parse_String(parseState, "[")) + RETURN_ERROR; + + /* Get output reg name */ + if (!Parse_Token(parseState, token)) + RETURN_ERROR; + + if (parseState->isPositionInvariant) + start = 1; /* skip HPOS register name */ + else + start = 0; + + /* try to match an output register name */ + for (j = start; OutputRegisters[j]; j++) { + if (strcmp((const char *) token, OutputRegisters[j]) == 0) { + *outputRegNum = j; + break; + } + } + if (!OutputRegisters[j]) + RETURN_ERROR1("Unrecognized output register name"); + + /* Match ']' */ + if (!Parse_String(parseState, "]")) + RETURN_ERROR1("Expected ]"); + + return GL_TRUE; +} + + +static GLboolean +Parse_MaskedDstReg(struct parse_state *parseState, struct prog_dst_register *dstReg) +{ + GLubyte token[100]; + GLint idx; + + /* Dst reg can be R<n> or o[n] */ + if (!Peek_Token(parseState, token)) + RETURN_ERROR; + + if (token[0] == 'R') { + /* a temporary register */ + dstReg->File = PROGRAM_TEMPORARY; + if (!Parse_TempReg(parseState, &idx)) + RETURN_ERROR; + dstReg->Index = idx; + } + else if (!parseState->isStateProgram && token[0] == 'o') { + /* an output register */ + dstReg->File = PROGRAM_OUTPUT; + if (!Parse_OutputReg(parseState, &idx)) + RETURN_ERROR; + dstReg->Index = idx; + } + else if (parseState->isStateProgram && token[0] == 'c' && + parseState->isStateProgram) { + /* absolute program parameter register */ + /* Only valid for vertex state programs */ + dstReg->File = PROGRAM_ENV_PARAM; + if (!Parse_AbsParamReg(parseState, &idx)) + RETURN_ERROR; + dstReg->Index = idx; + } + else { + RETURN_ERROR1("Bad destination register name"); + } + + /* Parse optional write mask */ + if (!Peek_Token(parseState, token)) + RETURN_ERROR; + + if (token[0] == '.') { + /* got a mask */ + GLint k = 0; + + if (!Parse_String(parseState, ".")) + RETURN_ERROR; + + if (!Parse_Token(parseState, token)) + RETURN_ERROR; + + dstReg->WriteMask = 0; + + if (token[k] == 'x') { + dstReg->WriteMask |= WRITEMASK_X; + k++; + } + if (token[k] == 'y') { + dstReg->WriteMask |= WRITEMASK_Y; + k++; + } + if (token[k] == 'z') { + dstReg->WriteMask |= WRITEMASK_Z; + k++; + } + if (token[k] == 'w') { + dstReg->WriteMask |= WRITEMASK_W; + k++; + } + if (k == 0) { + RETURN_ERROR1("Bad writemask character"); + } + return GL_TRUE; + } + else { + dstReg->WriteMask = WRITEMASK_XYZW; + return GL_TRUE; + } +} + + +static GLboolean +Parse_SwizzleSrcReg(struct parse_state *parseState, struct prog_src_register *srcReg) +{ + GLubyte token[100]; + GLint idx; + + srcReg->RelAddr = GL_FALSE; + + /* check for '-' */ + if (!Peek_Token(parseState, token)) + RETURN_ERROR; + if (token[0] == '-') { + (void) Parse_String(parseState, "-"); + srcReg->Negate = NEGATE_XYZW; + if (!Peek_Token(parseState, token)) + RETURN_ERROR; + } + else { + srcReg->Negate = NEGATE_NONE; + } + + /* Src reg can be R<n>, c[n], c[n +/- offset], or a named vertex attrib */ + if (token[0] == 'R') { + srcReg->File = PROGRAM_TEMPORARY; + if (!Parse_TempReg(parseState, &idx)) + RETURN_ERROR; + srcReg->Index = idx; + } + else if (token[0] == 'c') { + if (!Parse_ParamReg(parseState, srcReg)) + RETURN_ERROR; + } + else if (token[0] == 'v') { + srcReg->File = PROGRAM_INPUT; + if (!Parse_AttribReg(parseState, &idx)) + RETURN_ERROR; + srcReg->Index = idx; + } + else { + RETURN_ERROR2("Bad source register name", token); + } + + /* init swizzle fields */ + srcReg->Swizzle = SWIZZLE_NOOP; + + /* Look for optional swizzle suffix */ + if (!Peek_Token(parseState, token)) + RETURN_ERROR; + if (token[0] == '.') { + (void) Parse_String(parseState, "."); /* consume . */ + + if (!Parse_Token(parseState, token)) + RETURN_ERROR; + + if (token[1] == 0) { + /* single letter swizzle */ + if (token[0] == 'x') + srcReg->Swizzle = SWIZZLE_XXXX; + else if (token[0] == 'y') + srcReg->Swizzle = SWIZZLE_YYYY; + else if (token[0] == 'z') + srcReg->Swizzle = SWIZZLE_ZZZZ; + else if (token[0] == 'w') + srcReg->Swizzle = SWIZZLE_WWWW; + else + RETURN_ERROR1("Expected x, y, z, or w"); + } + else { + /* 2, 3 or 4-component swizzle */ + GLint k; + + srcReg->Swizzle = 0; + + for (k = 0; token[k] && k < 5; k++) { + if (token[k] == 'x') + srcReg->Swizzle |= 0 << (k*3); + else if (token[k] == 'y') + srcReg->Swizzle |= 1 << (k*3); + else if (token[k] == 'z') + srcReg->Swizzle |= 2 << (k*3); + else if (token[k] == 'w') + srcReg->Swizzle |= 3 << (k*3); + else + RETURN_ERROR; + } + if (k >= 5) + RETURN_ERROR; + } + } + + return GL_TRUE; +} + + +static GLboolean +Parse_ScalarSrcReg(struct parse_state *parseState, struct prog_src_register *srcReg) +{ + GLubyte token[100]; + GLint idx; + + srcReg->RelAddr = GL_FALSE; + + /* check for '-' */ + if (!Peek_Token(parseState, token)) + RETURN_ERROR; + if (token[0] == '-') { + srcReg->Negate = NEGATE_XYZW; + (void) Parse_String(parseState, "-"); /* consume '-' */ + if (!Peek_Token(parseState, token)) + RETURN_ERROR; + } + else { + srcReg->Negate = NEGATE_NONE; + } + + /* Src reg can be R<n>, c[n], c[n +/- offset], or a named vertex attrib */ + if (token[0] == 'R') { + srcReg->File = PROGRAM_TEMPORARY; + if (!Parse_TempReg(parseState, &idx)) + RETURN_ERROR; + srcReg->Index = idx; + } + else if (token[0] == 'c') { + if (!Parse_ParamReg(parseState, srcReg)) + RETURN_ERROR; + } + else if (token[0] == 'v') { + srcReg->File = PROGRAM_INPUT; + if (!Parse_AttribReg(parseState, &idx)) + RETURN_ERROR; + srcReg->Index = idx; + } + else { + RETURN_ERROR2("Bad source register name", token); + } + + /* Look for .[xyzw] suffix */ + if (!Parse_String(parseState, ".")) + RETURN_ERROR; + + if (!Parse_Token(parseState, token)) + RETURN_ERROR; + + if (token[0] == 'x' && token[1] == 0) { + srcReg->Swizzle = 0; + } + else if (token[0] == 'y' && token[1] == 0) { + srcReg->Swizzle = 1; + } + else if (token[0] == 'z' && token[1] == 0) { + srcReg->Swizzle = 2; + } + else if (token[0] == 'w' && token[1] == 0) { + srcReg->Swizzle = 3; + } + else { + RETURN_ERROR1("Bad scalar source suffix"); + } + + return GL_TRUE; +} + + +static GLint +Parse_UnaryOpInstruction(struct parse_state *parseState, + struct prog_instruction *inst, + enum prog_opcode opcode) +{ + if (opcode == OPCODE_ABS && !parseState->isVersion1_1) + RETURN_ERROR1("ABS illegal for vertex program 1.0"); + + inst->Opcode = opcode; + + /* dest reg */ + if (!Parse_MaskedDstReg(parseState, &inst->DstReg)) + RETURN_ERROR; + + /* comma */ + if (!Parse_String(parseState, ",")) + RETURN_ERROR; + + /* src arg */ + if (!Parse_SwizzleSrcReg(parseState, &inst->SrcReg[0])) + RETURN_ERROR; + + /* semicolon */ + if (!Parse_String(parseState, ";")) + RETURN_ERROR; + + return GL_TRUE; +} + + +static GLboolean +Parse_BiOpInstruction(struct parse_state *parseState, + struct prog_instruction *inst, + enum prog_opcode opcode) +{ + if (opcode == OPCODE_DPH && !parseState->isVersion1_1) + RETURN_ERROR1("DPH illegal for vertex program 1.0"); + if (opcode == OPCODE_SUB && !parseState->isVersion1_1) + RETURN_ERROR1("SUB illegal for vertex program 1.0"); + + inst->Opcode = opcode; + + /* dest reg */ + if (!Parse_MaskedDstReg(parseState, &inst->DstReg)) + RETURN_ERROR; + + /* comma */ + if (!Parse_String(parseState, ",")) + RETURN_ERROR; + + /* first src arg */ + if (!Parse_SwizzleSrcReg(parseState, &inst->SrcReg[0])) + RETURN_ERROR; + + /* comma */ + if (!Parse_String(parseState, ",")) + RETURN_ERROR; + + /* second src arg */ + if (!Parse_SwizzleSrcReg(parseState, &inst->SrcReg[1])) + RETURN_ERROR; + + /* semicolon */ + if (!Parse_String(parseState, ";")) + RETURN_ERROR; + + /* make sure we don't reference more than one program parameter register */ + if (inst->SrcReg[0].File == PROGRAM_ENV_PARAM && + inst->SrcReg[1].File == PROGRAM_ENV_PARAM && + inst->SrcReg[0].Index != inst->SrcReg[1].Index) + RETURN_ERROR1("Can't reference two program parameter registers"); + + /* make sure we don't reference more than one vertex attribute register */ + if (inst->SrcReg[0].File == PROGRAM_INPUT && + inst->SrcReg[1].File == PROGRAM_INPUT && + inst->SrcReg[0].Index != inst->SrcReg[1].Index) + RETURN_ERROR1("Can't reference two vertex attribute registers"); + + return GL_TRUE; +} + + +static GLboolean +Parse_TriOpInstruction(struct parse_state *parseState, + struct prog_instruction *inst, + enum prog_opcode opcode) +{ + inst->Opcode = opcode; + + /* dest reg */ + if (!Parse_MaskedDstReg(parseState, &inst->DstReg)) + RETURN_ERROR; + + /* comma */ + if (!Parse_String(parseState, ",")) + RETURN_ERROR; + + /* first src arg */ + if (!Parse_SwizzleSrcReg(parseState, &inst->SrcReg[0])) + RETURN_ERROR; + + /* comma */ + if (!Parse_String(parseState, ",")) + RETURN_ERROR; + + /* second src arg */ + if (!Parse_SwizzleSrcReg(parseState, &inst->SrcReg[1])) + RETURN_ERROR; + + /* comma */ + if (!Parse_String(parseState, ",")) + RETURN_ERROR; + + /* third src arg */ + if (!Parse_SwizzleSrcReg(parseState, &inst->SrcReg[2])) + RETURN_ERROR; + + /* semicolon */ + if (!Parse_String(parseState, ";")) + RETURN_ERROR; + + /* make sure we don't reference more than one program parameter register */ + if ((inst->SrcReg[0].File == PROGRAM_ENV_PARAM && + inst->SrcReg[1].File == PROGRAM_ENV_PARAM && + inst->SrcReg[0].Index != inst->SrcReg[1].Index) || + (inst->SrcReg[0].File == PROGRAM_ENV_PARAM && + inst->SrcReg[2].File == PROGRAM_ENV_PARAM && + inst->SrcReg[0].Index != inst->SrcReg[2].Index) || + (inst->SrcReg[1].File == PROGRAM_ENV_PARAM && + inst->SrcReg[2].File == PROGRAM_ENV_PARAM && + inst->SrcReg[1].Index != inst->SrcReg[2].Index)) + RETURN_ERROR1("Can only reference one program register"); + + /* make sure we don't reference more than one vertex attribute register */ + if ((inst->SrcReg[0].File == PROGRAM_INPUT && + inst->SrcReg[1].File == PROGRAM_INPUT && + inst->SrcReg[0].Index != inst->SrcReg[1].Index) || + (inst->SrcReg[0].File == PROGRAM_INPUT && + inst->SrcReg[2].File == PROGRAM_INPUT && + inst->SrcReg[0].Index != inst->SrcReg[2].Index) || + (inst->SrcReg[1].File == PROGRAM_INPUT && + inst->SrcReg[2].File == PROGRAM_INPUT && + inst->SrcReg[1].Index != inst->SrcReg[2].Index)) + RETURN_ERROR1("Can only reference one input register"); + + return GL_TRUE; +} + + +static GLboolean +Parse_ScalarInstruction(struct parse_state *parseState, + struct prog_instruction *inst, + enum prog_opcode opcode) +{ + if (opcode == OPCODE_RCC && !parseState->isVersion1_1) + RETURN_ERROR1("RCC illegal for vertex program 1.0"); + + inst->Opcode = opcode; + + /* dest reg */ + if (!Parse_MaskedDstReg(parseState, &inst->DstReg)) + RETURN_ERROR; + + /* comma */ + if (!Parse_String(parseState, ",")) + RETURN_ERROR; + + /* first src arg */ + if (!Parse_ScalarSrcReg(parseState, &inst->SrcReg[0])) + RETURN_ERROR; + + /* semicolon */ + if (!Parse_String(parseState, ";")) + RETURN_ERROR; + + return GL_TRUE; +} + + +static GLboolean +Parse_AddressInstruction(struct parse_state *parseState, struct prog_instruction *inst) +{ + inst->Opcode = OPCODE_ARL; + + /* Make ARB_vp backends happy */ + inst->DstReg.File = PROGRAM_ADDRESS; + inst->DstReg.WriteMask = WRITEMASK_X; + inst->DstReg.Index = 0; + + /* dest A0 reg */ + if (!Parse_AddrReg(parseState)) + RETURN_ERROR; + + /* comma */ + if (!Parse_String(parseState, ",")) + RETURN_ERROR; + + /* parse src reg */ + if (!Parse_ScalarSrcReg(parseState, &inst->SrcReg[0])) + RETURN_ERROR; + + /* semicolon */ + if (!Parse_String(parseState, ";")) + RETURN_ERROR; + + return GL_TRUE; +} + + +static GLboolean +Parse_EndInstruction(struct parse_state *parseState, struct prog_instruction *inst) +{ + GLubyte token[100]; + + inst->Opcode = OPCODE_END; + + /* this should fail! */ + if (Parse_Token(parseState, token)) + RETURN_ERROR2("Unexpected token after END:", token); + else + return GL_TRUE; +} + + +/** + * The PRINT instruction is Mesa-specific and is meant as a debugging aid for + * the vertex program developer. + * The NV_vertex_program extension grammar is modified as follows: + * + * <instruction> ::= <ARL-instruction> + * | ... + * | <PRINT-instruction> + * + * <PRINT-instruction> ::= "PRINT" <string literal> + * | "PRINT" <string literal> "," <srcReg> + * | "PRINT" <string literal> "," <dstReg> + */ +static GLboolean +Parse_PrintInstruction(struct parse_state *parseState, struct prog_instruction *inst) +{ + const GLubyte *str; + GLubyte *msg; + GLuint len; + GLubyte token[100]; + struct prog_src_register *srcReg = &inst->SrcReg[0]; + GLint idx; + + inst->Opcode = OPCODE_PRINT; + + /* The first argument is a literal string 'just like this' */ + if (!Parse_String(parseState, "'")) + RETURN_ERROR; + + str = parseState->pos; + for (len = 0; str[len] != '\''; len++) /* find closing quote */ + ; + parseState->pos += len + 1; + msg = (GLubyte*) malloc(len + 1); + + memcpy(msg, str, len); + msg[len] = 0; + inst->Data = msg; + + /* comma */ + if (Parse_String(parseState, ",")) { + + /* The second argument is a register name */ + if (!Peek_Token(parseState, token)) + RETURN_ERROR; + + srcReg->RelAddr = GL_FALSE; + srcReg->Negate = NEGATE_NONE; + srcReg->Swizzle = SWIZZLE_NOOP; + + /* Register can be R<n>, c[n], c[n +/- offset], a named vertex attrib, + * or an o[n] output register. + */ + if (token[0] == 'R') { + srcReg->File = PROGRAM_TEMPORARY; + if (!Parse_TempReg(parseState, &idx)) + RETURN_ERROR; + srcReg->Index = idx; + } + else if (token[0] == 'c') { + srcReg->File = PROGRAM_ENV_PARAM; + if (!Parse_ParamReg(parseState, srcReg)) + RETURN_ERROR; + } + else if (token[0] == 'v') { + srcReg->File = PROGRAM_INPUT; + if (!Parse_AttribReg(parseState, &idx)) + RETURN_ERROR; + srcReg->Index = idx; + } + else if (token[0] == 'o') { + srcReg->File = PROGRAM_OUTPUT; + if (!Parse_OutputReg(parseState, &idx)) + RETURN_ERROR; + srcReg->Index = idx; + } + else { + RETURN_ERROR2("Bad source register name", token); + } + } + else { + srcReg->File = PROGRAM_UNDEFINED; + } + + /* semicolon */ + if (!Parse_String(parseState, ";")) + RETURN_ERROR; + + return GL_TRUE; +} + + +static GLboolean +Parse_OptionSequence(struct parse_state *parseState, + struct prog_instruction program[]) +{ + (void) program; + while (1) { + if (!Parse_String(parseState, "OPTION")) + return GL_TRUE; /* ok, not an OPTION statement */ + if (Parse_String(parseState, "NV_position_invariant")) { + parseState->isPositionInvariant = GL_TRUE; + } + else { + RETURN_ERROR1("unexpected OPTION statement"); + } + if (!Parse_String(parseState, ";")) + return GL_FALSE; + } +} + + +static GLboolean +Parse_InstructionSequence(struct parse_state *parseState, + struct prog_instruction program[]) +{ + while (1) { + struct prog_instruction *inst = program + parseState->numInst; + + /* Initialize the instruction */ + _mesa_init_instructions(inst, 1); + + if (Parse_String(parseState, "MOV")) { + if (!Parse_UnaryOpInstruction(parseState, inst, OPCODE_MOV)) + RETURN_ERROR; + } + else if (Parse_String(parseState, "LIT")) { + if (!Parse_UnaryOpInstruction(parseState, inst, OPCODE_LIT)) + RETURN_ERROR; + } + else if (Parse_String(parseState, "ABS")) { + if (!Parse_UnaryOpInstruction(parseState, inst, OPCODE_ABS)) + RETURN_ERROR; + } + else if (Parse_String(parseState, "MUL")) { + if (!Parse_BiOpInstruction(parseState, inst, OPCODE_MUL)) + RETURN_ERROR; + } + else if (Parse_String(parseState, "ADD")) { + if (!Parse_BiOpInstruction(parseState, inst, OPCODE_ADD)) + RETURN_ERROR; + } + else if (Parse_String(parseState, "DP3")) { + if (!Parse_BiOpInstruction(parseState, inst, OPCODE_DP3)) + RETURN_ERROR; + } + else if (Parse_String(parseState, "DP4")) { + if (!Parse_BiOpInstruction(parseState, inst, OPCODE_DP4)) + RETURN_ERROR; + } + else if (Parse_String(parseState, "DST")) { + if (!Parse_BiOpInstruction(parseState, inst, OPCODE_DST)) + RETURN_ERROR; + } + else if (Parse_String(parseState, "MIN")) { + if (!Parse_BiOpInstruction(parseState, inst, OPCODE_MIN)) + RETURN_ERROR; + } + else if (Parse_String(parseState, "MAX")) { + if (!Parse_BiOpInstruction(parseState, inst, OPCODE_MAX)) + RETURN_ERROR; + } + else if (Parse_String(parseState, "SLT")) { + if (!Parse_BiOpInstruction(parseState, inst, OPCODE_SLT)) + RETURN_ERROR; + } + else if (Parse_String(parseState, "SGE")) { + if (!Parse_BiOpInstruction(parseState, inst, OPCODE_SGE)) + RETURN_ERROR; + } + else if (Parse_String(parseState, "DPH")) { + if (!Parse_BiOpInstruction(parseState, inst, OPCODE_DPH)) + RETURN_ERROR; + } + else if (Parse_String(parseState, "SUB")) { + if (!Parse_BiOpInstruction(parseState, inst, OPCODE_SUB)) + RETURN_ERROR; + } + else if (Parse_String(parseState, "MAD")) { + if (!Parse_TriOpInstruction(parseState, inst, OPCODE_MAD)) + RETURN_ERROR; + } + else if (Parse_String(parseState, "RCP")) { + if (!Parse_ScalarInstruction(parseState, inst, OPCODE_RCP)) + RETURN_ERROR; + } + else if (Parse_String(parseState, "RSQ")) { + if (!Parse_ScalarInstruction(parseState, inst, OPCODE_RSQ)) + RETURN_ERROR; + } + else if (Parse_String(parseState, "EXP")) { + if (!Parse_ScalarInstruction(parseState, inst, OPCODE_EXP)) + RETURN_ERROR; + } + else if (Parse_String(parseState, "LOG")) { + if (!Parse_ScalarInstruction(parseState, inst, OPCODE_LOG)) + RETURN_ERROR; + } + else if (Parse_String(parseState, "RCC")) { + if (!Parse_ScalarInstruction(parseState, inst, OPCODE_RCC)) + RETURN_ERROR; + } + else if (Parse_String(parseState, "ARL")) { + if (!Parse_AddressInstruction(parseState, inst)) + RETURN_ERROR; + } + else if (Parse_String(parseState, "PRINT")) { + if (!Parse_PrintInstruction(parseState, inst)) + RETURN_ERROR; + } + else if (Parse_String(parseState, "END")) { + if (!Parse_EndInstruction(parseState, inst)) + RETURN_ERROR; + else { + parseState->numInst++; + return GL_TRUE; /* all done */ + } + } + else { + /* bad instruction name */ + RETURN_ERROR1("Unexpected token"); + } + + /* examine input/output registers */ + if (inst->DstReg.File == PROGRAM_OUTPUT) + parseState->outputsWritten |= (1 << inst->DstReg.Index); + else if (inst->DstReg.File == PROGRAM_ENV_PARAM) + parseState->anyProgRegsWritten = GL_TRUE; + + if (inst->SrcReg[0].File == PROGRAM_INPUT) + parseState->inputsRead |= (1 << inst->SrcReg[0].Index); + if (inst->SrcReg[1].File == PROGRAM_INPUT) + parseState->inputsRead |= (1 << inst->SrcReg[1].Index); + if (inst->SrcReg[2].File == PROGRAM_INPUT) + parseState->inputsRead |= (1 << inst->SrcReg[2].Index); + + parseState->numInst++; + + if (parseState->numInst >= MAX_NV_VERTEX_PROGRAM_INSTRUCTIONS) + RETURN_ERROR1("Program too long"); + } + + RETURN_ERROR; +} + + +static GLboolean +Parse_Program(struct parse_state *parseState, + struct prog_instruction instBuffer[]) +{ + if (parseState->isVersion1_1) { + if (!Parse_OptionSequence(parseState, instBuffer)) { + return GL_FALSE; + } + } + return Parse_InstructionSequence(parseState, instBuffer); +} + + +/** + * Parse/compile the 'str' returning the compiled 'program'. + * ctx->Program.ErrorPos will be -1 if successful. Otherwise, ErrorPos + * indicates the position of the error in 'str'. + */ +void +_mesa_parse_nv_vertex_program(GLcontext *ctx, GLenum dstTarget, + const GLubyte *str, GLsizei len, + struct gl_vertex_program *program) +{ + struct parse_state parseState; + struct prog_instruction instBuffer[MAX_NV_VERTEX_PROGRAM_INSTRUCTIONS]; + struct prog_instruction *newInst; + GLenum target; + GLubyte *programString; + + /* Make a null-terminated copy of the program string */ + programString = (GLubyte *) MALLOC(len + 1); + if (!programString) { + _mesa_error(ctx, GL_OUT_OF_MEMORY, "glLoadProgramNV"); + return; + } + memcpy(programString, str, len); + programString[len] = 0; + + /* Get ready to parse */ + parseState.ctx = ctx; + parseState.start = programString; + parseState.isPositionInvariant = GL_FALSE; + parseState.isVersion1_1 = GL_FALSE; + parseState.numInst = 0; + parseState.inputsRead = 0; + parseState.outputsWritten = 0; + parseState.anyProgRegsWritten = GL_FALSE; + parseState.indirectRegisterFiles = 0x0; + + /* Reset error state */ + _mesa_set_program_error(ctx, -1, NULL); + + /* check the program header */ + if (strncmp((const char *) programString, "!!VP1.0", 7) == 0) { + target = GL_VERTEX_PROGRAM_NV; + parseState.pos = programString + 7; + parseState.isStateProgram = GL_FALSE; + } + else if (strncmp((const char *) programString, "!!VP1.1", 7) == 0) { + target = GL_VERTEX_PROGRAM_NV; + parseState.pos = programString + 7; + parseState.isStateProgram = GL_FALSE; + parseState.isVersion1_1 = GL_TRUE; + } + else if (strncmp((const char *) programString, "!!VSP1.0", 8) == 0) { + target = GL_VERTEX_STATE_PROGRAM_NV; + parseState.pos = programString + 8; + parseState.isStateProgram = GL_TRUE; + } + else { + /* invalid header */ + ctx->Program.ErrorPos = 0; + _mesa_error(ctx, GL_INVALID_OPERATION, "glLoadProgramNV(bad header)"); + return; + } + + /* make sure target and header match */ + if (target != dstTarget) { + _mesa_error(ctx, GL_INVALID_OPERATION, + "glLoadProgramNV(target mismatch)"); + return; + } + + + if (Parse_Program(&parseState, instBuffer)) { + gl_state_index state_tokens[STATE_LENGTH] = {0, 0, 0, 0, 0}; + int i; + + /* successful parse! */ + + if (parseState.isStateProgram) { + if (!parseState.anyProgRegsWritten) { + _mesa_error(ctx, GL_INVALID_OPERATION, + "glLoadProgramNV(c[#] not written)"); + return; + } + } + else { + if (!parseState.isPositionInvariant && + !(parseState.outputsWritten & (1 << VERT_RESULT_HPOS))) { + /* bit 1 = HPOS register */ + _mesa_error(ctx, GL_INVALID_OPERATION, + "glLoadProgramNV(HPOS not written)"); + return; + } + } + + /* copy the compiled instructions */ + assert(parseState.numInst <= MAX_NV_VERTEX_PROGRAM_INSTRUCTIONS); + newInst = _mesa_alloc_instructions(parseState.numInst); + if (!newInst) { + _mesa_error(ctx, GL_OUT_OF_MEMORY, "glLoadProgramNV"); + free(programString); + return; /* out of memory */ + } + _mesa_copy_instructions(newInst, instBuffer, parseState.numInst); + + /* install the program */ + program->Base.Target = target; + if (program->Base.String) { + free(program->Base.String); + } + program->Base.String = programString; + program->Base.Format = GL_PROGRAM_FORMAT_ASCII_ARB; + if (program->Base.Instructions) { + free(program->Base.Instructions); + } + program->Base.Instructions = newInst; + program->Base.InputsRead = parseState.inputsRead; + if (parseState.isPositionInvariant) + program->Base.InputsRead |= VERT_BIT_POS; + program->Base.NumInstructions = parseState.numInst; + program->Base.OutputsWritten = parseState.outputsWritten; + program->IsPositionInvariant = parseState.isPositionInvariant; + program->IsNVProgram = GL_TRUE; + +#ifdef DEBUG_foo + printf("--- glLoadProgramNV result ---\n"); + _mesa_fprint_program_opt(stdout, &program->Base, PROG_PRINT_NV, 0); + printf("------------------------------\n"); +#endif + + if (program->Base.Parameters) + _mesa_free_parameter_list(program->Base.Parameters); + + program->Base.Parameters = _mesa_new_parameter_list (); + program->Base.NumParameters = 0; + + program->Base.IndirectRegisterFiles = parseState.indirectRegisterFiles; + + state_tokens[0] = STATE_VERTEX_PROGRAM; + state_tokens[1] = STATE_ENV; + /* Add refs to all of the potential params, in order. If we want to not + * upload everything, _mesa_layout_parameters is the answer. + */ + for (i = 0; i < MAX_NV_VERTEX_PROGRAM_PARAMS; i++) { + GLint index; + state_tokens[2] = i; + index = _mesa_add_state_reference(program->Base.Parameters, + state_tokens); + assert(index == i); + } + program->Base.NumParameters = program->Base.Parameters->NumParameters; + + _mesa_setup_nv_temporary_count(ctx, &program->Base); + _mesa_emit_nv_temp_initialization(ctx, &program->Base); + } + else { + /* Error! */ + _mesa_error(ctx, GL_INVALID_OPERATION, "glLoadProgramNV"); + /* NOTE: _mesa_set_program_error would have been called already */ + /* GL_NV_vertex_program isn't supposed to set the error string + * so we reset it here. + */ + _mesa_set_program_error(ctx, ctx->Program.ErrorPos, NULL); + } +} + + +const char * +_mesa_nv_vertex_input_register_name(GLuint i) +{ + ASSERT(i < MAX_NV_VERTEX_PROGRAM_INPUTS); + return InputRegisters[i]; +} + + +const char * +_mesa_nv_vertex_output_register_name(GLuint i) +{ + ASSERT(i < MAX_NV_VERTEX_PROGRAM_OUTPUTS); + return OutputRegisters[i]; +} + diff --git a/mesalib/src/mesa/program/nvvertparse.h b/mesalib/src/mesa/program/nvvertparse.h new file mode 100644 index 000000000..91ef79e6c --- /dev/null +++ b/mesalib/src/mesa/program/nvvertparse.h @@ -0,0 +1,46 @@ +/* + * Mesa 3-D graphics library + * Version: 5.1 + * + * Copyright (C) 1999-2003 Brian Paul All Rights Reserved. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included + * in all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS + * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * BRIAN PAUL BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN + * AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN + * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. + * + * Authors: + * Brian Paul + */ + + +#ifndef NVVERTPARSE_H +#define NVVERTPARSE_H + +#include "main/mtypes.h" + +extern void +_mesa_parse_nv_vertex_program(GLcontext *ctx, GLenum target, + const GLubyte *str, GLsizei len, + struct gl_vertex_program *program); + + +extern const char * +_mesa_nv_vertex_input_register_name(GLuint i); + +extern const char * +_mesa_nv_vertex_output_register_name(GLuint i); + +#endif diff --git a/mesalib/src/mesa/program/prog_cache.c b/mesalib/src/mesa/program/prog_cache.c new file mode 100644 index 000000000..8af689754 --- /dev/null +++ b/mesalib/src/mesa/program/prog_cache.c @@ -0,0 +1,206 @@ +/************************************************************************** + * + * Copyright 2003 Tungsten Graphics, Inc., Cedar Park, Texas. + * All Rights Reserved. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the + * "Software"), to deal in the Software without restriction, including + * without limitation the rights to use, copy, modify, merge, publish, + * distribute, sub license, and/or sell copies of the Software, and to + * permit persons to whom the Software is furnished to do so, subject to + * the following conditions: + * + * The above copyright notice and this permission notice (including the + * next paragraph) shall be included in all copies or substantial portions + * of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS + * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF + * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. + * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR + * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, + * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE + * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. + * + **************************************************************************/ + + +#include "main/glheader.h" +#include "main/mtypes.h" +#include "main/imports.h" +#include "program/prog_cache.h" +#include "program/program.h" + + +struct cache_item +{ + GLuint hash; + void *key; + struct gl_program *program; + struct cache_item *next; +}; + +struct gl_program_cache +{ + struct cache_item **items; + struct cache_item *last; + GLuint size, n_items; +}; + + + +/** + * Compute hash index from state key. + */ +static GLuint +hash_key(const void *key, GLuint key_size) +{ + const GLuint *ikey = (const GLuint *) key; + GLuint hash = 0, i; + + assert(key_size >= 4); + + /* Make a slightly better attempt at a hash function: + */ + for (i = 0; i < key_size / sizeof(*ikey); i++) + { + hash += ikey[i]; + hash += (hash << 10); + hash ^= (hash >> 6); + } + + return hash; +} + + +/** + * Rebuild/expand the hash table to accomodate more entries + */ +static void +rehash(struct gl_program_cache *cache) +{ + struct cache_item **items; + struct cache_item *c, *next; + GLuint size, i; + + cache->last = NULL; + + size = cache->size * 3; + items = (struct cache_item**) malloc(size * sizeof(*items)); + memset(items, 0, size * sizeof(*items)); + + for (i = 0; i < cache->size; i++) + for (c = cache->items[i]; c; c = next) { + next = c->next; + c->next = items[c->hash % size]; + items[c->hash % size] = c; + } + + free(cache->items); + cache->items = items; + cache->size = size; +} + + +static void +clear_cache(GLcontext *ctx, struct gl_program_cache *cache) +{ + struct cache_item *c, *next; + GLuint i; + + cache->last = NULL; + + for (i = 0; i < cache->size; i++) { + for (c = cache->items[i]; c; c = next) { + next = c->next; + free(c->key); + _mesa_reference_program(ctx, &c->program, NULL); + free(c); + } + cache->items[i] = NULL; + } + + + cache->n_items = 0; +} + + + +struct gl_program_cache * +_mesa_new_program_cache(void) +{ + struct gl_program_cache *cache = CALLOC_STRUCT(gl_program_cache); + if (cache) { + cache->size = 17; + cache->items = (struct cache_item **) + calloc(1, cache->size * sizeof(struct cache_item)); + if (!cache->items) { + free(cache); + return NULL; + } + } + return cache; +} + + +void +_mesa_delete_program_cache(GLcontext *ctx, struct gl_program_cache *cache) +{ + clear_cache(ctx, cache); + free(cache->items); + free(cache); +} + + +struct gl_program * +_mesa_search_program_cache(struct gl_program_cache *cache, + const void *key, GLuint keysize) +{ + if (cache->last && + memcmp(cache->last->key, key, keysize) == 0) { + return cache->last->program; + } + else { + const GLuint hash = hash_key(key, keysize); + struct cache_item *c; + + for (c = cache->items[hash % cache->size]; c; c = c->next) { + if (c->hash == hash && memcmp(c->key, key, keysize) == 0) { + cache->last = c; + return c->program; + } + } + + return NULL; + } +} + + +void +_mesa_program_cache_insert(GLcontext *ctx, + struct gl_program_cache *cache, + const void *key, GLuint keysize, + struct gl_program *program) +{ + const GLuint hash = hash_key(key, keysize); + struct cache_item *c = CALLOC_STRUCT(cache_item); + + c->hash = hash; + + c->key = malloc(keysize); + memcpy(c->key, key, keysize); + + c->program = program; /* no refcount change */ + + if (cache->n_items > cache->size * 1.5) { + if (cache->size < 1000) + rehash(cache); + else + clear_cache(ctx, cache); + } + + cache->n_items++; + c->next = cache->items[hash % cache->size]; + cache->items[hash % cache->size] = c; +} diff --git a/mesalib/src/mesa/program/prog_cache.h b/mesalib/src/mesa/program/prog_cache.h new file mode 100644 index 000000000..bfe8f99d4 --- /dev/null +++ b/mesalib/src/mesa/program/prog_cache.h @@ -0,0 +1,58 @@ +/************************************************************************** + * + * Copyright 2003 Tungsten Graphics, Inc., Cedar Park, Texas. + * All Rights Reserved. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the + * "Software"), to deal in the Software without restriction, including + * without limitation the rights to use, copy, modify, merge, publish, + * distribute, sub license, and/or sell copies of the Software, and to + * permit persons to whom the Software is furnished to do so, subject to + * the following conditions: + * + * The above copyright notice and this permission notice (including the + * next paragraph) shall be included in all copies or substantial portions + * of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS + * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF + * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. + * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR + * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, + * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE + * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. + * + **************************************************************************/ + + +#ifndef PROG_CACHE_H +#define PROG_CACHE_H + + +#include "main/mtypes.h" + + +/** Opaque type */ +struct gl_program_cache; + + +extern struct gl_program_cache * +_mesa_new_program_cache(void); + +extern void +_mesa_delete_program_cache(GLcontext *ctx, struct gl_program_cache *pc); + + +extern struct gl_program * +_mesa_search_program_cache(struct gl_program_cache *cache, + const void *key, GLuint keysize); + +extern void +_mesa_program_cache_insert(GLcontext *ctx, + struct gl_program_cache *cache, + const void *key, GLuint keysize, + struct gl_program *program); + + +#endif /* PROG_CACHE_H */ diff --git a/mesalib/src/mesa/program/prog_execute.c b/mesalib/src/mesa/program/prog_execute.c new file mode 100644 index 000000000..2ae5bc572 --- /dev/null +++ b/mesalib/src/mesa/program/prog_execute.c @@ -0,0 +1,1879 @@ +/* + * Mesa 3-D graphics library + * Version: 7.3 + * + * Copyright (C) 1999-2008 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. + */ + +/** + * \file prog_execute.c + * Software interpreter for vertex/fragment programs. + * \author Brian Paul + */ + +/* + * NOTE: we do everything in single-precision floating point; we don't + * currently observe the single/half/fixed-precision qualifiers. + * + */ + + +#include "main/glheader.h" +#include "main/colormac.h" +#include "main/macros.h" +#include "prog_execute.h" +#include "prog_instruction.h" +#include "prog_parameter.h" +#include "prog_print.h" +#include "prog_noise.h" + + +/* debug predicate */ +#define DEBUG_PROG 0 + + +/** + * Set x to positive or negative infinity. + */ +#if defined(USE_IEEE) || defined(_WIN32) +#define SET_POS_INFINITY(x) \ + do { \ + fi_type fi; \ + fi.i = 0x7F800000; \ + x = fi.f; \ + } while (0) +#define SET_NEG_INFINITY(x) \ + do { \ + fi_type fi; \ + fi.i = 0xFF800000; \ + x = fi.f; \ + } while (0) +#elif defined(VMS) +#define SET_POS_INFINITY(x) x = __MAXFLOAT +#define SET_NEG_INFINITY(x) x = -__MAXFLOAT +#else +#define SET_POS_INFINITY(x) x = (GLfloat) HUGE_VAL +#define SET_NEG_INFINITY(x) x = (GLfloat) -HUGE_VAL +#endif + +#define SET_FLOAT_BITS(x, bits) ((fi_type *) (void *) &(x))->i = bits + + +static const GLfloat ZeroVec[4] = { 0.0F, 0.0F, 0.0F, 0.0F }; + + + +/** + * Return TRUE for +0 and other positive values, FALSE otherwise. + * Used for RCC opcode. + */ +static INLINE GLboolean +positive(float x) +{ + fi_type fi; + fi.f = x; + if (fi.i & 0x80000000) + return GL_FALSE; + return GL_TRUE; +} + + + +/** + * Return a pointer to the 4-element float vector specified by the given + * source register. + */ +static INLINE const GLfloat * +get_src_register_pointer(const struct prog_src_register *source, + const struct gl_program_machine *machine) +{ + const struct gl_program *prog = machine->CurProgram; + GLint reg = source->Index; + + if (source->RelAddr) { + /* add address register value to src index/offset */ + reg += machine->AddressReg[0][0]; + if (reg < 0) { + return ZeroVec; + } + } + + switch (source->File) { + case PROGRAM_TEMPORARY: + if (reg >= MAX_PROGRAM_TEMPS) + return ZeroVec; + return machine->Temporaries[reg]; + + case PROGRAM_INPUT: + if (prog->Target == GL_VERTEX_PROGRAM_ARB) { + if (reg >= VERT_ATTRIB_MAX) + return ZeroVec; + return machine->VertAttribs[reg]; + } + else { + if (reg >= FRAG_ATTRIB_MAX) + return ZeroVec; + return machine->Attribs[reg][machine->CurElement]; + } + + case PROGRAM_OUTPUT: + if (reg >= MAX_PROGRAM_OUTPUTS) + return ZeroVec; + return machine->Outputs[reg]; + + case PROGRAM_LOCAL_PARAM: + if (reg >= MAX_PROGRAM_LOCAL_PARAMS) + return ZeroVec; + return machine->CurProgram->LocalParams[reg]; + + case PROGRAM_ENV_PARAM: + if (reg >= MAX_PROGRAM_ENV_PARAMS) + return ZeroVec; + return machine->EnvParams[reg]; + + case PROGRAM_STATE_VAR: + /* Fallthrough */ + case PROGRAM_CONSTANT: + /* Fallthrough */ + case PROGRAM_UNIFORM: + /* Fallthrough */ + case PROGRAM_NAMED_PARAM: + if (reg >= (GLint) prog->Parameters->NumParameters) + return ZeroVec; + return prog->Parameters->ParameterValues[reg]; + + default: + _mesa_problem(NULL, + "Invalid src register file %d in get_src_register_pointer()", + source->File); + return NULL; + } +} + + +/** + * Return a pointer to the 4-element float vector specified by the given + * destination register. + */ +static INLINE GLfloat * +get_dst_register_pointer(const struct prog_dst_register *dest, + struct gl_program_machine *machine) +{ + static GLfloat dummyReg[4]; + GLint reg = dest->Index; + + if (dest->RelAddr) { + /* add address register value to src index/offset */ + reg += machine->AddressReg[0][0]; + if (reg < 0) { + return dummyReg; + } + } + + switch (dest->File) { + case PROGRAM_TEMPORARY: + if (reg >= MAX_PROGRAM_TEMPS) + return dummyReg; + return machine->Temporaries[reg]; + + case PROGRAM_OUTPUT: + if (reg >= MAX_PROGRAM_OUTPUTS) + return dummyReg; + return machine->Outputs[reg]; + + case PROGRAM_WRITE_ONLY: + return dummyReg; + + default: + _mesa_problem(NULL, + "Invalid dest register file %d in get_dst_register_pointer()", + dest->File); + return NULL; + } +} + + + +/** + * Fetch a 4-element float vector from the given source register. + * Apply swizzling and negating as needed. + */ +static void +fetch_vector4(const struct prog_src_register *source, + const struct gl_program_machine *machine, GLfloat result[4]) +{ + const GLfloat *src = get_src_register_pointer(source, machine); + ASSERT(src); + + if (source->Swizzle == SWIZZLE_NOOP) { + /* no swizzling */ + COPY_4V(result, src); + } + else { + ASSERT(GET_SWZ(source->Swizzle, 0) <= 3); + ASSERT(GET_SWZ(source->Swizzle, 1) <= 3); + ASSERT(GET_SWZ(source->Swizzle, 2) <= 3); + ASSERT(GET_SWZ(source->Swizzle, 3) <= 3); + result[0] = src[GET_SWZ(source->Swizzle, 0)]; + result[1] = src[GET_SWZ(source->Swizzle, 1)]; + result[2] = src[GET_SWZ(source->Swizzle, 2)]; + result[3] = src[GET_SWZ(source->Swizzle, 3)]; + } + + if (source->Abs) { + result[0] = FABSF(result[0]); + result[1] = FABSF(result[1]); + result[2] = FABSF(result[2]); + result[3] = FABSF(result[3]); + } + if (source->Negate) { + ASSERT(source->Negate == NEGATE_XYZW); + result[0] = -result[0]; + result[1] = -result[1]; + result[2] = -result[2]; + result[3] = -result[3]; + } + +#ifdef NAN_CHECK + assert(!IS_INF_OR_NAN(result[0])); + assert(!IS_INF_OR_NAN(result[0])); + assert(!IS_INF_OR_NAN(result[0])); + assert(!IS_INF_OR_NAN(result[0])); +#endif +} + + +/** + * Fetch a 4-element uint vector from the given source register. + * Apply swizzling but not negation/abs. + */ +static void +fetch_vector4ui(const struct prog_src_register *source, + const struct gl_program_machine *machine, GLuint result[4]) +{ + const GLuint *src = (GLuint *) get_src_register_pointer(source, machine); + ASSERT(src); + + if (source->Swizzle == SWIZZLE_NOOP) { + /* no swizzling */ + COPY_4V(result, src); + } + else { + ASSERT(GET_SWZ(source->Swizzle, 0) <= 3); + ASSERT(GET_SWZ(source->Swizzle, 1) <= 3); + ASSERT(GET_SWZ(source->Swizzle, 2) <= 3); + ASSERT(GET_SWZ(source->Swizzle, 3) <= 3); + result[0] = src[GET_SWZ(source->Swizzle, 0)]; + result[1] = src[GET_SWZ(source->Swizzle, 1)]; + result[2] = src[GET_SWZ(source->Swizzle, 2)]; + result[3] = src[GET_SWZ(source->Swizzle, 3)]; + } + + /* Note: no Negate or Abs here */ +} + + + +/** + * Fetch the derivative with respect to X or Y for the given register. + * XXX this currently only works for fragment program input attribs. + */ +static void +fetch_vector4_deriv(GLcontext * ctx, + const struct prog_src_register *source, + const struct gl_program_machine *machine, + char xOrY, GLfloat result[4]) +{ + if (source->File == PROGRAM_INPUT && + source->Index < (GLint) machine->NumDeriv) { + const GLint col = machine->CurElement; + const GLfloat w = machine->Attribs[FRAG_ATTRIB_WPOS][col][3]; + const GLfloat invQ = 1.0f / w; + GLfloat deriv[4]; + + if (xOrY == 'X') { + deriv[0] = machine->DerivX[source->Index][0] * invQ; + deriv[1] = machine->DerivX[source->Index][1] * invQ; + deriv[2] = machine->DerivX[source->Index][2] * invQ; + deriv[3] = machine->DerivX[source->Index][3] * invQ; + } + else { + deriv[0] = machine->DerivY[source->Index][0] * invQ; + deriv[1] = machine->DerivY[source->Index][1] * invQ; + deriv[2] = machine->DerivY[source->Index][2] * invQ; + deriv[3] = machine->DerivY[source->Index][3] * invQ; + } + + result[0] = deriv[GET_SWZ(source->Swizzle, 0)]; + result[1] = deriv[GET_SWZ(source->Swizzle, 1)]; + result[2] = deriv[GET_SWZ(source->Swizzle, 2)]; + result[3] = deriv[GET_SWZ(source->Swizzle, 3)]; + + if (source->Abs) { + result[0] = FABSF(result[0]); + result[1] = FABSF(result[1]); + result[2] = FABSF(result[2]); + result[3] = FABSF(result[3]); + } + if (source->Negate) { + ASSERT(source->Negate == NEGATE_XYZW); + result[0] = -result[0]; + result[1] = -result[1]; + result[2] = -result[2]; + result[3] = -result[3]; + } + } + else { + ASSIGN_4V(result, 0.0, 0.0, 0.0, 0.0); + } +} + + +/** + * As above, but only return result[0] element. + */ +static void +fetch_vector1(const struct prog_src_register *source, + const struct gl_program_machine *machine, GLfloat result[4]) +{ + const GLfloat *src = get_src_register_pointer(source, machine); + ASSERT(src); + + result[0] = src[GET_SWZ(source->Swizzle, 0)]; + + if (source->Abs) { + result[0] = FABSF(result[0]); + } + if (source->Negate) { + result[0] = -result[0]; + } +} + + +static GLuint +fetch_vector1ui(const struct prog_src_register *source, + const struct gl_program_machine *machine) +{ + const GLuint *src = (GLuint *) get_src_register_pointer(source, machine); + return src[GET_SWZ(source->Swizzle, 0)]; +} + + +/** + * Fetch texel from texture. Use partial derivatives when possible. + */ +static INLINE void +fetch_texel(GLcontext *ctx, + const struct gl_program_machine *machine, + const struct prog_instruction *inst, + const GLfloat texcoord[4], GLfloat lodBias, + GLfloat color[4]) +{ + const GLuint unit = machine->Samplers[inst->TexSrcUnit]; + + /* Note: we only have the right derivatives for fragment input attribs. + */ + if (machine->NumDeriv > 0 && + inst->SrcReg[0].File == PROGRAM_INPUT && + inst->SrcReg[0].Index == FRAG_ATTRIB_TEX0 + inst->TexSrcUnit) { + /* simple texture fetch for which we should have derivatives */ + GLuint attr = inst->SrcReg[0].Index; + machine->FetchTexelDeriv(ctx, texcoord, + machine->DerivX[attr], + machine->DerivY[attr], + lodBias, unit, color); + } + else { + machine->FetchTexelLod(ctx, texcoord, lodBias, unit, color); + } +} + + +/** + * Test value against zero and return GT, LT, EQ or UN if NaN. + */ +static INLINE GLuint +generate_cc(float value) +{ + if (value != value) + return COND_UN; /* NaN */ + if (value > 0.0F) + return COND_GT; + if (value < 0.0F) + return COND_LT; + return COND_EQ; +} + + +/** + * Test if the ccMaskRule is satisfied by the given condition code. + * Used to mask destination writes according to the current condition code. + */ +static INLINE GLboolean +test_cc(GLuint condCode, GLuint ccMaskRule) +{ + switch (ccMaskRule) { + case COND_EQ: return (condCode == COND_EQ); + case COND_NE: return (condCode != COND_EQ); + case COND_LT: return (condCode == COND_LT); + case COND_GE: return (condCode == COND_GT || condCode == COND_EQ); + case COND_LE: return (condCode == COND_LT || condCode == COND_EQ); + case COND_GT: return (condCode == COND_GT); + case COND_TR: return GL_TRUE; + case COND_FL: return GL_FALSE; + default: return GL_TRUE; + } +} + + +/** + * Evaluate the 4 condition codes against a predicate and return GL_TRUE + * or GL_FALSE to indicate result. + */ +static INLINE GLboolean +eval_condition(const struct gl_program_machine *machine, + const struct prog_instruction *inst) +{ + const GLuint swizzle = inst->DstReg.CondSwizzle; + const GLuint condMask = inst->DstReg.CondMask; + if (test_cc(machine->CondCodes[GET_SWZ(swizzle, 0)], condMask) || + test_cc(machine->CondCodes[GET_SWZ(swizzle, 1)], condMask) || + test_cc(machine->CondCodes[GET_SWZ(swizzle, 2)], condMask) || + test_cc(machine->CondCodes[GET_SWZ(swizzle, 3)], condMask)) { + return GL_TRUE; + } + else { + return GL_FALSE; + } +} + + + +/** + * Store 4 floats into a register. Observe the instructions saturate and + * set-condition-code flags. + */ +static void +store_vector4(const struct prog_instruction *inst, + struct gl_program_machine *machine, const GLfloat value[4]) +{ + const struct prog_dst_register *dstReg = &(inst->DstReg); + const GLboolean clamp = inst->SaturateMode == SATURATE_ZERO_ONE; + GLuint writeMask = dstReg->WriteMask; + GLfloat clampedValue[4]; + GLfloat *dst = get_dst_register_pointer(dstReg, machine); + +#if 0 + if (value[0] > 1.0e10 || + IS_INF_OR_NAN(value[0]) || + IS_INF_OR_NAN(value[1]) || + IS_INF_OR_NAN(value[2]) || IS_INF_OR_NAN(value[3])) + printf("store %g %g %g %g\n", value[0], value[1], value[2], value[3]); +#endif + + if (clamp) { + clampedValue[0] = CLAMP(value[0], 0.0F, 1.0F); + clampedValue[1] = CLAMP(value[1], 0.0F, 1.0F); + clampedValue[2] = CLAMP(value[2], 0.0F, 1.0F); + clampedValue[3] = CLAMP(value[3], 0.0F, 1.0F); + value = clampedValue; + } + + if (dstReg->CondMask != COND_TR) { + /* condition codes may turn off some writes */ + if (writeMask & WRITEMASK_X) { + if (!test_cc(machine->CondCodes[GET_SWZ(dstReg->CondSwizzle, 0)], + dstReg->CondMask)) + writeMask &= ~WRITEMASK_X; + } + if (writeMask & WRITEMASK_Y) { + if (!test_cc(machine->CondCodes[GET_SWZ(dstReg->CondSwizzle, 1)], + dstReg->CondMask)) + writeMask &= ~WRITEMASK_Y; + } + if (writeMask & WRITEMASK_Z) { + if (!test_cc(machine->CondCodes[GET_SWZ(dstReg->CondSwizzle, 2)], + dstReg->CondMask)) + writeMask &= ~WRITEMASK_Z; + } + if (writeMask & WRITEMASK_W) { + if (!test_cc(machine->CondCodes[GET_SWZ(dstReg->CondSwizzle, 3)], + dstReg->CondMask)) + writeMask &= ~WRITEMASK_W; + } + } + +#ifdef NAN_CHECK + assert(!IS_INF_OR_NAN(value[0])); + assert(!IS_INF_OR_NAN(value[0])); + assert(!IS_INF_OR_NAN(value[0])); + assert(!IS_INF_OR_NAN(value[0])); +#endif + + if (writeMask & WRITEMASK_X) + dst[0] = value[0]; + if (writeMask & WRITEMASK_Y) + dst[1] = value[1]; + if (writeMask & WRITEMASK_Z) + dst[2] = value[2]; + if (writeMask & WRITEMASK_W) + dst[3] = value[3]; + + if (inst->CondUpdate) { + if (writeMask & WRITEMASK_X) + machine->CondCodes[0] = generate_cc(value[0]); + if (writeMask & WRITEMASK_Y) + machine->CondCodes[1] = generate_cc(value[1]); + if (writeMask & WRITEMASK_Z) + machine->CondCodes[2] = generate_cc(value[2]); + if (writeMask & WRITEMASK_W) + machine->CondCodes[3] = generate_cc(value[3]); +#if DEBUG_PROG + printf("CondCodes=(%s,%s,%s,%s) for:\n", + _mesa_condcode_string(machine->CondCodes[0]), + _mesa_condcode_string(machine->CondCodes[1]), + _mesa_condcode_string(machine->CondCodes[2]), + _mesa_condcode_string(machine->CondCodes[3])); +#endif + } +} + + +/** + * Store 4 uints into a register. Observe the set-condition-code flags. + */ +static void +store_vector4ui(const struct prog_instruction *inst, + struct gl_program_machine *machine, const GLuint value[4]) +{ + const struct prog_dst_register *dstReg = &(inst->DstReg); + GLuint writeMask = dstReg->WriteMask; + GLuint *dst = (GLuint *) get_dst_register_pointer(dstReg, machine); + + if (dstReg->CondMask != COND_TR) { + /* condition codes may turn off some writes */ + if (writeMask & WRITEMASK_X) { + if (!test_cc(machine->CondCodes[GET_SWZ(dstReg->CondSwizzle, 0)], + dstReg->CondMask)) + writeMask &= ~WRITEMASK_X; + } + if (writeMask & WRITEMASK_Y) { + if (!test_cc(machine->CondCodes[GET_SWZ(dstReg->CondSwizzle, 1)], + dstReg->CondMask)) + writeMask &= ~WRITEMASK_Y; + } + if (writeMask & WRITEMASK_Z) { + if (!test_cc(machine->CondCodes[GET_SWZ(dstReg->CondSwizzle, 2)], + dstReg->CondMask)) + writeMask &= ~WRITEMASK_Z; + } + if (writeMask & WRITEMASK_W) { + if (!test_cc(machine->CondCodes[GET_SWZ(dstReg->CondSwizzle, 3)], + dstReg->CondMask)) + writeMask &= ~WRITEMASK_W; + } + } + + if (writeMask & WRITEMASK_X) + dst[0] = value[0]; + if (writeMask & WRITEMASK_Y) + dst[1] = value[1]; + if (writeMask & WRITEMASK_Z) + dst[2] = value[2]; + if (writeMask & WRITEMASK_W) + dst[3] = value[3]; + + if (inst->CondUpdate) { + if (writeMask & WRITEMASK_X) + machine->CondCodes[0] = generate_cc((float)value[0]); + if (writeMask & WRITEMASK_Y) + machine->CondCodes[1] = generate_cc((float)value[1]); + if (writeMask & WRITEMASK_Z) + machine->CondCodes[2] = generate_cc((float)value[2]); + if (writeMask & WRITEMASK_W) + machine->CondCodes[3] = generate_cc((float)value[3]); +#if DEBUG_PROG + printf("CondCodes=(%s,%s,%s,%s) for:\n", + _mesa_condcode_string(machine->CondCodes[0]), + _mesa_condcode_string(machine->CondCodes[1]), + _mesa_condcode_string(machine->CondCodes[2]), + _mesa_condcode_string(machine->CondCodes[3])); +#endif + } +} + + + +/** + * Execute the given vertex/fragment program. + * + * \param ctx rendering context + * \param program the program to execute + * \param machine machine state (must be initialized) + * \return GL_TRUE if program completed or GL_FALSE if program executed KIL. + */ +GLboolean +_mesa_execute_program(GLcontext * ctx, + const struct gl_program *program, + struct gl_program_machine *machine) +{ + const GLuint numInst = program->NumInstructions; + const GLuint maxExec = 10000; + GLuint pc, numExec = 0; + + machine->CurProgram = program; + + if (DEBUG_PROG) { + printf("execute program %u --------------------\n", program->Id); + } + + if (program->Target == GL_VERTEX_PROGRAM_ARB) { + machine->EnvParams = ctx->VertexProgram.Parameters; + } + else { + machine->EnvParams = ctx->FragmentProgram.Parameters; + } + + for (pc = 0; pc < numInst; pc++) { + const struct prog_instruction *inst = program->Instructions + pc; + + if (DEBUG_PROG) { + _mesa_print_instruction(inst); + } + + switch (inst->Opcode) { + case OPCODE_ABS: + { + GLfloat a[4], result[4]; + fetch_vector4(&inst->SrcReg[0], machine, a); + result[0] = FABSF(a[0]); + result[1] = FABSF(a[1]); + result[2] = FABSF(a[2]); + result[3] = FABSF(a[3]); + store_vector4(inst, machine, result); + } + break; + case OPCODE_ADD: + { + GLfloat a[4], b[4], result[4]; + fetch_vector4(&inst->SrcReg[0], machine, a); + fetch_vector4(&inst->SrcReg[1], machine, b); + result[0] = a[0] + b[0]; + result[1] = a[1] + b[1]; + result[2] = a[2] + b[2]; + result[3] = a[3] + b[3]; + store_vector4(inst, machine, result); + if (DEBUG_PROG) { + printf("ADD (%g %g %g %g) = (%g %g %g %g) + (%g %g %g %g)\n", + result[0], result[1], result[2], result[3], + a[0], a[1], a[2], a[3], b[0], b[1], b[2], b[3]); + } + } + break; + case OPCODE_AND: /* bitwise AND */ + { + GLuint a[4], b[4], result[4]; + fetch_vector4ui(&inst->SrcReg[0], machine, a); + fetch_vector4ui(&inst->SrcReg[1], machine, b); + result[0] = a[0] & b[0]; + result[1] = a[1] & b[1]; + result[2] = a[2] & b[2]; + result[3] = a[3] & b[3]; + store_vector4ui(inst, machine, result); + } + break; + case OPCODE_ARL: + { + GLfloat t[4]; + fetch_vector4(&inst->SrcReg[0], machine, t); + machine->AddressReg[0][0] = IFLOOR(t[0]); + if (DEBUG_PROG) { + printf("ARL %d\n", machine->AddressReg[0][0]); + } + } + break; + case OPCODE_BGNLOOP: + /* no-op */ + ASSERT(program->Instructions[inst->BranchTarget].Opcode + == OPCODE_ENDLOOP); + break; + case OPCODE_ENDLOOP: + /* subtract 1 here since pc is incremented by for(pc) loop */ + ASSERT(program->Instructions[inst->BranchTarget].Opcode + == OPCODE_BGNLOOP); + pc = inst->BranchTarget - 1; /* go to matching BNGLOOP */ + break; + case OPCODE_BGNSUB: /* begin subroutine */ + break; + case OPCODE_ENDSUB: /* end subroutine */ + break; + case OPCODE_BRA: /* branch (conditional) */ + if (eval_condition(machine, inst)) { + /* take branch */ + /* Subtract 1 here since we'll do pc++ below */ + pc = inst->BranchTarget - 1; + } + break; + case OPCODE_BRK: /* break out of loop (conditional) */ + ASSERT(program->Instructions[inst->BranchTarget].Opcode + == OPCODE_ENDLOOP); + if (eval_condition(machine, inst)) { + /* break out of loop */ + /* pc++ at end of for-loop will put us after the ENDLOOP inst */ + pc = inst->BranchTarget; + } + break; + case OPCODE_CONT: /* continue loop (conditional) */ + ASSERT(program->Instructions[inst->BranchTarget].Opcode + == OPCODE_ENDLOOP); + if (eval_condition(machine, inst)) { + /* continue at ENDLOOP */ + /* Subtract 1 here since we'll do pc++ at end of for-loop */ + pc = inst->BranchTarget - 1; + } + break; + case OPCODE_CAL: /* Call subroutine (conditional) */ + if (eval_condition(machine, inst)) { + /* call the subroutine */ + if (machine->StackDepth >= MAX_PROGRAM_CALL_DEPTH) { + return GL_TRUE; /* Per GL_NV_vertex_program2 spec */ + } + machine->CallStack[machine->StackDepth++] = pc + 1; /* next inst */ + /* Subtract 1 here since we'll do pc++ at end of for-loop */ + pc = inst->BranchTarget - 1; + } + break; + case OPCODE_CMP: + { + GLfloat a[4], b[4], c[4], result[4]; + fetch_vector4(&inst->SrcReg[0], machine, a); + fetch_vector4(&inst->SrcReg[1], machine, b); + fetch_vector4(&inst->SrcReg[2], machine, c); + result[0] = a[0] < 0.0F ? b[0] : c[0]; + result[1] = a[1] < 0.0F ? b[1] : c[1]; + result[2] = a[2] < 0.0F ? b[2] : c[2]; + result[3] = a[3] < 0.0F ? b[3] : c[3]; + store_vector4(inst, machine, result); + if (DEBUG_PROG) { + printf("CMP (%g %g %g %g) = (%g %g %g %g) < 0 ? (%g %g %g %g) : (%g %g %g %g)\n", + result[0], result[1], result[2], result[3], + a[0], a[1], a[2], a[3], + b[0], b[1], b[2], b[3], + c[0], c[1], c[2], c[3]); + } + } + break; + case OPCODE_COS: + { + GLfloat a[4], result[4]; + fetch_vector1(&inst->SrcReg[0], machine, a); + result[0] = result[1] = result[2] = result[3] + = (GLfloat) cos(a[0]); + store_vector4(inst, machine, result); + } + break; + case OPCODE_DDX: /* Partial derivative with respect to X */ + { + GLfloat result[4]; + fetch_vector4_deriv(ctx, &inst->SrcReg[0], machine, + 'X', result); + store_vector4(inst, machine, result); + } + break; + case OPCODE_DDY: /* Partial derivative with respect to Y */ + { + GLfloat result[4]; + fetch_vector4_deriv(ctx, &inst->SrcReg[0], machine, + 'Y', result); + store_vector4(inst, machine, result); + } + break; + case OPCODE_DP2: + { + GLfloat a[4], b[4], result[4]; + fetch_vector4(&inst->SrcReg[0], machine, a); + fetch_vector4(&inst->SrcReg[1], machine, b); + result[0] = result[1] = result[2] = result[3] = DOT2(a, b); + store_vector4(inst, machine, result); + if (DEBUG_PROG) { + printf("DP2 %g = (%g %g) . (%g %g)\n", + result[0], a[0], a[1], b[0], b[1]); + } + } + break; + case OPCODE_DP2A: + { + GLfloat a[4], b[4], c, result[4]; + fetch_vector4(&inst->SrcReg[0], machine, a); + fetch_vector4(&inst->SrcReg[1], machine, b); + fetch_vector1(&inst->SrcReg[1], machine, &c); + result[0] = result[1] = result[2] = result[3] = DOT2(a, b) + c; + store_vector4(inst, machine, result); + if (DEBUG_PROG) { + printf("DP2A %g = (%g %g) . (%g %g) + %g\n", + result[0], a[0], a[1], b[0], b[1], c); + } + } + break; + case OPCODE_DP3: + { + GLfloat a[4], b[4], result[4]; + fetch_vector4(&inst->SrcReg[0], machine, a); + fetch_vector4(&inst->SrcReg[1], machine, b); + result[0] = result[1] = result[2] = result[3] = DOT3(a, b); + store_vector4(inst, machine, result); + if (DEBUG_PROG) { + printf("DP3 %g = (%g %g %g) . (%g %g %g)\n", + result[0], a[0], a[1], a[2], b[0], b[1], b[2]); + } + } + break; + case OPCODE_DP4: + { + GLfloat a[4], b[4], result[4]; + fetch_vector4(&inst->SrcReg[0], machine, a); + fetch_vector4(&inst->SrcReg[1], machine, b); + result[0] = result[1] = result[2] = result[3] = DOT4(a, b); + store_vector4(inst, machine, result); + if (DEBUG_PROG) { + printf("DP4 %g = (%g, %g %g %g) . (%g, %g %g %g)\n", + result[0], a[0], a[1], a[2], a[3], + b[0], b[1], b[2], b[3]); + } + } + break; + case OPCODE_DPH: + { + GLfloat a[4], b[4], result[4]; + fetch_vector4(&inst->SrcReg[0], machine, a); + fetch_vector4(&inst->SrcReg[1], machine, b); + result[0] = result[1] = result[2] = result[3] = DOT3(a, b) + b[3]; + store_vector4(inst, machine, result); + } + break; + case OPCODE_DST: /* Distance vector */ + { + GLfloat a[4], b[4], result[4]; + fetch_vector4(&inst->SrcReg[0], machine, a); + fetch_vector4(&inst->SrcReg[1], machine, b); + result[0] = 1.0F; + result[1] = a[1] * b[1]; + result[2] = a[2]; + result[3] = b[3]; + store_vector4(inst, machine, result); + } + break; + case OPCODE_EXP: + { + GLfloat t[4], q[4], floor_t0; + fetch_vector1(&inst->SrcReg[0], machine, t); + floor_t0 = FLOORF(t[0]); + if (floor_t0 > FLT_MAX_EXP) { + SET_POS_INFINITY(q[0]); + SET_POS_INFINITY(q[2]); + } + else if (floor_t0 < FLT_MIN_EXP) { + q[0] = 0.0F; + q[2] = 0.0F; + } + else { + q[0] = LDEXPF(1.0, (int) floor_t0); + /* Note: GL_NV_vertex_program expects + * result.z = result.x * APPX(result.y) + * We do what the ARB extension says. + */ + q[2] = (GLfloat) pow(2.0, t[0]); + } + q[1] = t[0] - floor_t0; + q[3] = 1.0F; + store_vector4( inst, machine, q ); + } + break; + case OPCODE_EX2: /* Exponential base 2 */ + { + GLfloat a[4], result[4], val; + fetch_vector1(&inst->SrcReg[0], machine, a); + val = (GLfloat) pow(2.0, a[0]); + /* + if (IS_INF_OR_NAN(val)) + val = 1.0e10; + */ + result[0] = result[1] = result[2] = result[3] = val; + store_vector4(inst, machine, result); + } + break; + case OPCODE_FLR: + { + GLfloat a[4], result[4]; + fetch_vector4(&inst->SrcReg[0], machine, a); + result[0] = FLOORF(a[0]); + result[1] = FLOORF(a[1]); + result[2] = FLOORF(a[2]); + result[3] = FLOORF(a[3]); + store_vector4(inst, machine, result); + } + break; + case OPCODE_FRC: + { + GLfloat a[4], result[4]; + fetch_vector4(&inst->SrcReg[0], machine, a); + result[0] = a[0] - FLOORF(a[0]); + result[1] = a[1] - FLOORF(a[1]); + result[2] = a[2] - FLOORF(a[2]); + result[3] = a[3] - FLOORF(a[3]); + store_vector4(inst, machine, result); + } + break; + case OPCODE_IF: + { + GLboolean cond; + ASSERT(program->Instructions[inst->BranchTarget].Opcode + == OPCODE_ELSE || + program->Instructions[inst->BranchTarget].Opcode + == OPCODE_ENDIF); + /* eval condition */ + if (inst->SrcReg[0].File != PROGRAM_UNDEFINED) { + GLfloat a[4]; + fetch_vector1(&inst->SrcReg[0], machine, a); + cond = (a[0] != 0.0); + } + else { + cond = eval_condition(machine, inst); + } + if (DEBUG_PROG) { + printf("IF: %d\n", cond); + } + /* do if/else */ + if (cond) { + /* do if-clause (just continue execution) */ + } + else { + /* go to the instruction after ELSE or ENDIF */ + assert(inst->BranchTarget >= 0); + pc = inst->BranchTarget; + } + } + break; + case OPCODE_ELSE: + /* goto ENDIF */ + ASSERT(program->Instructions[inst->BranchTarget].Opcode + == OPCODE_ENDIF); + assert(inst->BranchTarget >= 0); + pc = inst->BranchTarget; + break; + case OPCODE_ENDIF: + /* nothing */ + break; + case OPCODE_KIL_NV: /* NV_f_p only (conditional) */ + if (eval_condition(machine, inst)) { + return GL_FALSE; + } + break; + case OPCODE_KIL: /* ARB_f_p only */ + { + GLfloat a[4]; + fetch_vector4(&inst->SrcReg[0], machine, a); + if (DEBUG_PROG) { + printf("KIL if (%g %g %g %g) <= 0.0\n", + a[0], a[1], a[2], a[3]); + } + + if (a[0] < 0.0F || a[1] < 0.0F || a[2] < 0.0F || a[3] < 0.0F) { + return GL_FALSE; + } + } + break; + case OPCODE_LG2: /* log base 2 */ + { + GLfloat a[4], result[4], val; + fetch_vector1(&inst->SrcReg[0], machine, a); + /* The fast LOG2 macro doesn't meet the precision requirements. + */ + if (a[0] == 0.0F) { + val = -FLT_MAX; + } + else { + val = (float)(log(a[0]) * 1.442695F); + } + result[0] = result[1] = result[2] = result[3] = val; + store_vector4(inst, machine, result); + } + break; + case OPCODE_LIT: + { + const GLfloat epsilon = 1.0F / 256.0F; /* from NV VP spec */ + GLfloat a[4], result[4]; + fetch_vector4(&inst->SrcReg[0], machine, a); + a[0] = MAX2(a[0], 0.0F); + a[1] = MAX2(a[1], 0.0F); + /* XXX ARB version clamps a[3], NV version doesn't */ + a[3] = CLAMP(a[3], -(128.0F - epsilon), (128.0F - epsilon)); + result[0] = 1.0F; + result[1] = a[0]; + /* XXX we could probably just use pow() here */ + if (a[0] > 0.0F) { + if (a[1] == 0.0 && a[3] == 0.0) + result[2] = 1.0F; + else + result[2] = (GLfloat) pow(a[1], a[3]); + } + else { + result[2] = 0.0F; + } + result[3] = 1.0F; + store_vector4(inst, machine, result); + if (DEBUG_PROG) { + printf("LIT (%g %g %g %g) : (%g %g %g %g)\n", + result[0], result[1], result[2], result[3], + a[0], a[1], a[2], a[3]); + } + } + break; + case OPCODE_LOG: + { + GLfloat t[4], q[4], abs_t0; + fetch_vector1(&inst->SrcReg[0], machine, t); + abs_t0 = FABSF(t[0]); + if (abs_t0 != 0.0F) { + /* Since we really can't handle infinite values on VMS + * like other OSes we'll use __MAXFLOAT to represent + * infinity. This may need some tweaking. + */ +#ifdef VMS + if (abs_t0 == __MAXFLOAT) +#else + if (IS_INF_OR_NAN(abs_t0)) +#endif + { + SET_POS_INFINITY(q[0]); + q[1] = 1.0F; + SET_POS_INFINITY(q[2]); + } + else { + int exponent; + GLfloat mantissa = FREXPF(t[0], &exponent); + q[0] = (GLfloat) (exponent - 1); + q[1] = (GLfloat) (2.0 * mantissa); /* map [.5, 1) -> [1, 2) */ + + /* The fast LOG2 macro doesn't meet the precision + * requirements. + */ + q[2] = (float)(log(t[0]) * 1.442695F); + } + } + else { + SET_NEG_INFINITY(q[0]); + q[1] = 1.0F; + SET_NEG_INFINITY(q[2]); + } + q[3] = 1.0; + store_vector4(inst, machine, q); + } + break; + case OPCODE_LRP: + { + GLfloat a[4], b[4], c[4], result[4]; + fetch_vector4(&inst->SrcReg[0], machine, a); + fetch_vector4(&inst->SrcReg[1], machine, b); + fetch_vector4(&inst->SrcReg[2], machine, c); + result[0] = a[0] * b[0] + (1.0F - a[0]) * c[0]; + result[1] = a[1] * b[1] + (1.0F - a[1]) * c[1]; + result[2] = a[2] * b[2] + (1.0F - a[2]) * c[2]; + result[3] = a[3] * b[3] + (1.0F - a[3]) * c[3]; + store_vector4(inst, machine, result); + if (DEBUG_PROG) { + printf("LRP (%g %g %g %g) = (%g %g %g %g), " + "(%g %g %g %g), (%g %g %g %g)\n", + result[0], result[1], result[2], result[3], + a[0], a[1], a[2], a[3], + b[0], b[1], b[2], b[3], c[0], c[1], c[2], c[3]); + } + } + break; + case OPCODE_MAD: + { + GLfloat a[4], b[4], c[4], result[4]; + fetch_vector4(&inst->SrcReg[0], machine, a); + fetch_vector4(&inst->SrcReg[1], machine, b); + fetch_vector4(&inst->SrcReg[2], machine, c); + result[0] = a[0] * b[0] + c[0]; + result[1] = a[1] * b[1] + c[1]; + result[2] = a[2] * b[2] + c[2]; + result[3] = a[3] * b[3] + c[3]; + store_vector4(inst, machine, result); + if (DEBUG_PROG) { + printf("MAD (%g %g %g %g) = (%g %g %g %g) * " + "(%g %g %g %g) + (%g %g %g %g)\n", + result[0], result[1], result[2], result[3], + a[0], a[1], a[2], a[3], + b[0], b[1], b[2], b[3], c[0], c[1], c[2], c[3]); + } + } + break; + case OPCODE_MAX: + { + GLfloat a[4], b[4], result[4]; + fetch_vector4(&inst->SrcReg[0], machine, a); + fetch_vector4(&inst->SrcReg[1], machine, b); + result[0] = MAX2(a[0], b[0]); + result[1] = MAX2(a[1], b[1]); + result[2] = MAX2(a[2], b[2]); + result[3] = MAX2(a[3], b[3]); + store_vector4(inst, machine, result); + if (DEBUG_PROG) { + printf("MAX (%g %g %g %g) = (%g %g %g %g), (%g %g %g %g)\n", + result[0], result[1], result[2], result[3], + a[0], a[1], a[2], a[3], b[0], b[1], b[2], b[3]); + } + } + break; + case OPCODE_MIN: + { + GLfloat a[4], b[4], result[4]; + fetch_vector4(&inst->SrcReg[0], machine, a); + fetch_vector4(&inst->SrcReg[1], machine, b); + result[0] = MIN2(a[0], b[0]); + result[1] = MIN2(a[1], b[1]); + result[2] = MIN2(a[2], b[2]); + result[3] = MIN2(a[3], b[3]); + store_vector4(inst, machine, result); + } + break; + case OPCODE_MOV: + { + GLfloat result[4]; + fetch_vector4(&inst->SrcReg[0], machine, result); + store_vector4(inst, machine, result); + if (DEBUG_PROG) { + printf("MOV (%g %g %g %g)\n", + result[0], result[1], result[2], result[3]); + } + } + break; + case OPCODE_MUL: + { + GLfloat a[4], b[4], result[4]; + fetch_vector4(&inst->SrcReg[0], machine, a); + fetch_vector4(&inst->SrcReg[1], machine, b); + result[0] = a[0] * b[0]; + result[1] = a[1] * b[1]; + result[2] = a[2] * b[2]; + result[3] = a[3] * b[3]; + store_vector4(inst, machine, result); + if (DEBUG_PROG) { + printf("MUL (%g %g %g %g) = (%g %g %g %g) * (%g %g %g %g)\n", + result[0], result[1], result[2], result[3], + a[0], a[1], a[2], a[3], b[0], b[1], b[2], b[3]); + } + } + break; + case OPCODE_NOISE1: + { + GLfloat a[4], result[4]; + fetch_vector1(&inst->SrcReg[0], machine, a); + result[0] = + result[1] = + result[2] = + result[3] = _mesa_noise1(a[0]); + store_vector4(inst, machine, result); + } + break; + case OPCODE_NOISE2: + { + GLfloat a[4], result[4]; + fetch_vector4(&inst->SrcReg[0], machine, a); + result[0] = + result[1] = + result[2] = result[3] = _mesa_noise2(a[0], a[1]); + store_vector4(inst, machine, result); + } + break; + case OPCODE_NOISE3: + { + GLfloat a[4], result[4]; + fetch_vector4(&inst->SrcReg[0], machine, a); + result[0] = + result[1] = + result[2] = + result[3] = _mesa_noise3(a[0], a[1], a[2]); + store_vector4(inst, machine, result); + } + break; + case OPCODE_NOISE4: + { + GLfloat a[4], result[4]; + fetch_vector4(&inst->SrcReg[0], machine, a); + result[0] = + result[1] = + result[2] = + result[3] = _mesa_noise4(a[0], a[1], a[2], a[3]); + store_vector4(inst, machine, result); + } + break; + case OPCODE_NOP: + break; + case OPCODE_NOT: /* bitwise NOT */ + { + GLuint a[4], result[4]; + fetch_vector4ui(&inst->SrcReg[0], machine, a); + result[0] = ~a[0]; + result[1] = ~a[1]; + result[2] = ~a[2]; + result[3] = ~a[3]; + store_vector4ui(inst, machine, result); + } + break; + case OPCODE_NRM3: /* 3-component normalization */ + { + GLfloat a[4], result[4]; + GLfloat tmp; + fetch_vector4(&inst->SrcReg[0], machine, a); + tmp = a[0] * a[0] + a[1] * a[1] + a[2] * a[2]; + if (tmp != 0.0F) + tmp = INV_SQRTF(tmp); + result[0] = tmp * a[0]; + result[1] = tmp * a[1]; + result[2] = tmp * a[2]; + result[3] = 0.0; /* undefined, but prevent valgrind warnings */ + store_vector4(inst, machine, result); + } + break; + case OPCODE_NRM4: /* 4-component normalization */ + { + GLfloat a[4], result[4]; + GLfloat tmp; + fetch_vector4(&inst->SrcReg[0], machine, a); + tmp = a[0] * a[0] + a[1] * a[1] + a[2] * a[2] + a[3] * a[3]; + if (tmp != 0.0F) + tmp = INV_SQRTF(tmp); + result[0] = tmp * a[0]; + result[1] = tmp * a[1]; + result[2] = tmp * a[2]; + result[3] = tmp * a[3]; + store_vector4(inst, machine, result); + } + break; + case OPCODE_OR: /* bitwise OR */ + { + GLuint a[4], b[4], result[4]; + fetch_vector4ui(&inst->SrcReg[0], machine, a); + fetch_vector4ui(&inst->SrcReg[1], machine, b); + result[0] = a[0] | b[0]; + result[1] = a[1] | b[1]; + result[2] = a[2] | b[2]; + result[3] = a[3] | b[3]; + store_vector4ui(inst, machine, result); + } + break; + case OPCODE_PK2H: /* pack two 16-bit floats in one 32-bit float */ + { + GLfloat a[4]; + GLuint result[4]; + GLhalfNV hx, hy; + fetch_vector4(&inst->SrcReg[0], machine, a); + hx = _mesa_float_to_half(a[0]); + hy = _mesa_float_to_half(a[1]); + result[0] = + result[1] = + result[2] = + result[3] = hx | (hy << 16); + store_vector4ui(inst, machine, result); + } + break; + case OPCODE_PK2US: /* pack two GLushorts into one 32-bit float */ + { + GLfloat a[4]; + GLuint result[4], usx, usy; + fetch_vector4(&inst->SrcReg[0], machine, a); + a[0] = CLAMP(a[0], 0.0F, 1.0F); + a[1] = CLAMP(a[1], 0.0F, 1.0F); + usx = IROUND(a[0] * 65535.0F); + usy = IROUND(a[1] * 65535.0F); + result[0] = + result[1] = + result[2] = + result[3] = usx | (usy << 16); + store_vector4ui(inst, machine, result); + } + break; + case OPCODE_PK4B: /* pack four GLbytes into one 32-bit float */ + { + GLfloat a[4]; + GLuint result[4], ubx, uby, ubz, ubw; + fetch_vector4(&inst->SrcReg[0], machine, a); + a[0] = CLAMP(a[0], -128.0F / 127.0F, 1.0F); + a[1] = CLAMP(a[1], -128.0F / 127.0F, 1.0F); + a[2] = CLAMP(a[2], -128.0F / 127.0F, 1.0F); + a[3] = CLAMP(a[3], -128.0F / 127.0F, 1.0F); + ubx = IROUND(127.0F * a[0] + 128.0F); + uby = IROUND(127.0F * a[1] + 128.0F); + ubz = IROUND(127.0F * a[2] + 128.0F); + ubw = IROUND(127.0F * a[3] + 128.0F); + result[0] = + result[1] = + result[2] = + result[3] = ubx | (uby << 8) | (ubz << 16) | (ubw << 24); + store_vector4ui(inst, machine, result); + } + break; + case OPCODE_PK4UB: /* pack four GLubytes into one 32-bit float */ + { + GLfloat a[4]; + GLuint result[4], ubx, uby, ubz, ubw; + fetch_vector4(&inst->SrcReg[0], machine, a); + a[0] = CLAMP(a[0], 0.0F, 1.0F); + a[1] = CLAMP(a[1], 0.0F, 1.0F); + a[2] = CLAMP(a[2], 0.0F, 1.0F); + a[3] = CLAMP(a[3], 0.0F, 1.0F); + ubx = IROUND(255.0F * a[0]); + uby = IROUND(255.0F * a[1]); + ubz = IROUND(255.0F * a[2]); + ubw = IROUND(255.0F * a[3]); + result[0] = + result[1] = + result[2] = + result[3] = ubx | (uby << 8) | (ubz << 16) | (ubw << 24); + store_vector4ui(inst, machine, result); + } + break; + case OPCODE_POW: + { + GLfloat a[4], b[4], result[4]; + fetch_vector1(&inst->SrcReg[0], machine, a); + fetch_vector1(&inst->SrcReg[1], machine, b); + result[0] = result[1] = result[2] = result[3] + = (GLfloat) pow(a[0], b[0]); + store_vector4(inst, machine, result); + } + break; + case OPCODE_RCC: /* clamped riciprocal */ + { + const float largest = 1.884467e+19, smallest = 5.42101e-20; + GLfloat a[4], r, result[4]; + fetch_vector1(&inst->SrcReg[0], machine, a); + if (DEBUG_PROG) { + if (a[0] == 0) + printf("RCC(0)\n"); + else if (IS_INF_OR_NAN(a[0])) + printf("RCC(inf)\n"); + } + if (a[0] == 1.0F) { + r = 1.0F; + } + else { + r = 1.0F / a[0]; + } + if (positive(r)) { + if (r > largest) { + r = largest; + } + else if (r < smallest) { + r = smallest; + } + } + else { + if (r < -largest) { + r = -largest; + } + else if (r > -smallest) { + r = -smallest; + } + } + result[0] = result[1] = result[2] = result[3] = r; + store_vector4(inst, machine, result); + } + break; + + case OPCODE_RCP: + { + GLfloat a[4], result[4]; + fetch_vector1(&inst->SrcReg[0], machine, a); + if (DEBUG_PROG) { + if (a[0] == 0) + printf("RCP(0)\n"); + else if (IS_INF_OR_NAN(a[0])) + printf("RCP(inf)\n"); + } + result[0] = result[1] = result[2] = result[3] = 1.0F / a[0]; + store_vector4(inst, machine, result); + } + break; + case OPCODE_RET: /* return from subroutine (conditional) */ + if (eval_condition(machine, inst)) { + if (machine->StackDepth == 0) { + return GL_TRUE; /* Per GL_NV_vertex_program2 spec */ + } + /* subtract one because of pc++ in the for loop */ + pc = machine->CallStack[--machine->StackDepth] - 1; + } + break; + case OPCODE_RFL: /* reflection vector */ + { + GLfloat axis[4], dir[4], result[4], tmpX, tmpW; + fetch_vector4(&inst->SrcReg[0], machine, axis); + fetch_vector4(&inst->SrcReg[1], machine, dir); + tmpW = DOT3(axis, axis); + tmpX = (2.0F * DOT3(axis, dir)) / tmpW; + result[0] = tmpX * axis[0] - dir[0]; + result[1] = tmpX * axis[1] - dir[1]; + result[2] = tmpX * axis[2] - dir[2]; + /* result[3] is never written! XXX enforce in parser! */ + store_vector4(inst, machine, result); + } + break; + case OPCODE_RSQ: /* 1 / sqrt() */ + { + GLfloat a[4], result[4]; + fetch_vector1(&inst->SrcReg[0], machine, a); + a[0] = FABSF(a[0]); + result[0] = result[1] = result[2] = result[3] = INV_SQRTF(a[0]); + store_vector4(inst, machine, result); + if (DEBUG_PROG) { + printf("RSQ %g = 1/sqrt(|%g|)\n", result[0], a[0]); + } + } + break; + case OPCODE_SCS: /* sine and cos */ + { + GLfloat a[4], result[4]; + fetch_vector1(&inst->SrcReg[0], machine, a); + result[0] = (GLfloat) cos(a[0]); + result[1] = (GLfloat) sin(a[0]); + result[2] = 0.0; /* undefined! */ + result[3] = 0.0; /* undefined! */ + store_vector4(inst, machine, result); + } + break; + case OPCODE_SEQ: /* set on equal */ + { + GLfloat a[4], b[4], result[4]; + fetch_vector4(&inst->SrcReg[0], machine, a); + fetch_vector4(&inst->SrcReg[1], machine, b); + result[0] = (a[0] == b[0]) ? 1.0F : 0.0F; + result[1] = (a[1] == b[1]) ? 1.0F : 0.0F; + result[2] = (a[2] == b[2]) ? 1.0F : 0.0F; + result[3] = (a[3] == b[3]) ? 1.0F : 0.0F; + store_vector4(inst, machine, result); + if (DEBUG_PROG) { + printf("SEQ (%g %g %g %g) = (%g %g %g %g) == (%g %g %g %g)\n", + result[0], result[1], result[2], result[3], + a[0], a[1], a[2], a[3], + b[0], b[1], b[2], b[3]); + } + } + break; + case OPCODE_SFL: /* set false, operands ignored */ + { + static const GLfloat result[4] = { 0.0F, 0.0F, 0.0F, 0.0F }; + store_vector4(inst, machine, result); + } + break; + case OPCODE_SGE: /* set on greater or equal */ + { + GLfloat a[4], b[4], result[4]; + fetch_vector4(&inst->SrcReg[0], machine, a); + fetch_vector4(&inst->SrcReg[1], machine, b); + result[0] = (a[0] >= b[0]) ? 1.0F : 0.0F; + result[1] = (a[1] >= b[1]) ? 1.0F : 0.0F; + result[2] = (a[2] >= b[2]) ? 1.0F : 0.0F; + result[3] = (a[3] >= b[3]) ? 1.0F : 0.0F; + store_vector4(inst, machine, result); + if (DEBUG_PROG) { + printf("SGE (%g %g %g %g) = (%g %g %g %g) >= (%g %g %g %g)\n", + result[0], result[1], result[2], result[3], + a[0], a[1], a[2], a[3], + b[0], b[1], b[2], b[3]); + } + } + break; + case OPCODE_SGT: /* set on greater */ + { + GLfloat a[4], b[4], result[4]; + fetch_vector4(&inst->SrcReg[0], machine, a); + fetch_vector4(&inst->SrcReg[1], machine, b); + result[0] = (a[0] > b[0]) ? 1.0F : 0.0F; + result[1] = (a[1] > b[1]) ? 1.0F : 0.0F; + result[2] = (a[2] > b[2]) ? 1.0F : 0.0F; + result[3] = (a[3] > b[3]) ? 1.0F : 0.0F; + store_vector4(inst, machine, result); + if (DEBUG_PROG) { + printf("SGT (%g %g %g %g) = (%g %g %g %g) > (%g %g %g %g)\n", + result[0], result[1], result[2], result[3], + a[0], a[1], a[2], a[3], + b[0], b[1], b[2], b[3]); + } + } + break; + case OPCODE_SIN: + { + GLfloat a[4], result[4]; + fetch_vector1(&inst->SrcReg[0], machine, a); + result[0] = result[1] = result[2] = result[3] + = (GLfloat) sin(a[0]); + store_vector4(inst, machine, result); + } + break; + case OPCODE_SLE: /* set on less or equal */ + { + GLfloat a[4], b[4], result[4]; + fetch_vector4(&inst->SrcReg[0], machine, a); + fetch_vector4(&inst->SrcReg[1], machine, b); + result[0] = (a[0] <= b[0]) ? 1.0F : 0.0F; + result[1] = (a[1] <= b[1]) ? 1.0F : 0.0F; + result[2] = (a[2] <= b[2]) ? 1.0F : 0.0F; + result[3] = (a[3] <= b[3]) ? 1.0F : 0.0F; + store_vector4(inst, machine, result); + if (DEBUG_PROG) { + printf("SLE (%g %g %g %g) = (%g %g %g %g) <= (%g %g %g %g)\n", + result[0], result[1], result[2], result[3], + a[0], a[1], a[2], a[3], + b[0], b[1], b[2], b[3]); + } + } + break; + case OPCODE_SLT: /* set on less */ + { + GLfloat a[4], b[4], result[4]; + fetch_vector4(&inst->SrcReg[0], machine, a); + fetch_vector4(&inst->SrcReg[1], machine, b); + result[0] = (a[0] < b[0]) ? 1.0F : 0.0F; + result[1] = (a[1] < b[1]) ? 1.0F : 0.0F; + result[2] = (a[2] < b[2]) ? 1.0F : 0.0F; + result[3] = (a[3] < b[3]) ? 1.0F : 0.0F; + store_vector4(inst, machine, result); + if (DEBUG_PROG) { + printf("SLT (%g %g %g %g) = (%g %g %g %g) < (%g %g %g %g)\n", + result[0], result[1], result[2], result[3], + a[0], a[1], a[2], a[3], + b[0], b[1], b[2], b[3]); + } + } + break; + case OPCODE_SNE: /* set on not equal */ + { + GLfloat a[4], b[4], result[4]; + fetch_vector4(&inst->SrcReg[0], machine, a); + fetch_vector4(&inst->SrcReg[1], machine, b); + result[0] = (a[0] != b[0]) ? 1.0F : 0.0F; + result[1] = (a[1] != b[1]) ? 1.0F : 0.0F; + result[2] = (a[2] != b[2]) ? 1.0F : 0.0F; + result[3] = (a[3] != b[3]) ? 1.0F : 0.0F; + store_vector4(inst, machine, result); + if (DEBUG_PROG) { + printf("SNE (%g %g %g %g) = (%g %g %g %g) != (%g %g %g %g)\n", + result[0], result[1], result[2], result[3], + a[0], a[1], a[2], a[3], + b[0], b[1], b[2], b[3]); + } + } + break; + case OPCODE_SSG: /* set sign (-1, 0 or +1) */ + { + GLfloat a[4], result[4]; + fetch_vector4(&inst->SrcReg[0], machine, a); + result[0] = (GLfloat) ((a[0] > 0.0F) - (a[0] < 0.0F)); + result[1] = (GLfloat) ((a[1] > 0.0F) - (a[1] < 0.0F)); + result[2] = (GLfloat) ((a[2] > 0.0F) - (a[2] < 0.0F)); + result[3] = (GLfloat) ((a[3] > 0.0F) - (a[3] < 0.0F)); + store_vector4(inst, machine, result); + } + break; + case OPCODE_STR: /* set true, operands ignored */ + { + static const GLfloat result[4] = { 1.0F, 1.0F, 1.0F, 1.0F }; + store_vector4(inst, machine, result); + } + break; + case OPCODE_SUB: + { + GLfloat a[4], b[4], result[4]; + fetch_vector4(&inst->SrcReg[0], machine, a); + fetch_vector4(&inst->SrcReg[1], machine, b); + result[0] = a[0] - b[0]; + result[1] = a[1] - b[1]; + result[2] = a[2] - b[2]; + result[3] = a[3] - b[3]; + store_vector4(inst, machine, result); + if (DEBUG_PROG) { + printf("SUB (%g %g %g %g) = (%g %g %g %g) - (%g %g %g %g)\n", + result[0], result[1], result[2], result[3], + a[0], a[1], a[2], a[3], b[0], b[1], b[2], b[3]); + } + } + break; + case OPCODE_SWZ: /* extended swizzle */ + { + const struct prog_src_register *source = &inst->SrcReg[0]; + const GLfloat *src = get_src_register_pointer(source, machine); + GLfloat result[4]; + GLuint i; + for (i = 0; i < 4; i++) { + const GLuint swz = GET_SWZ(source->Swizzle, i); + if (swz == SWIZZLE_ZERO) + result[i] = 0.0; + else if (swz == SWIZZLE_ONE) + result[i] = 1.0; + else { + ASSERT(swz >= 0); + ASSERT(swz <= 3); + result[i] = src[swz]; + } + if (source->Negate & (1 << i)) + result[i] = -result[i]; + } + store_vector4(inst, machine, result); + } + break; + case OPCODE_TEX: /* Both ARB and NV frag prog */ + /* Simple texel lookup */ + { + GLfloat texcoord[4], color[4]; + fetch_vector4(&inst->SrcReg[0], machine, texcoord); + + fetch_texel(ctx, machine, inst, texcoord, 0.0, color); + + if (DEBUG_PROG) { + printf("TEX (%g, %g, %g, %g) = texture[%d][%g, %g, %g, %g]\n", + color[0], color[1], color[2], color[3], + inst->TexSrcUnit, + texcoord[0], texcoord[1], texcoord[2], texcoord[3]); + } + store_vector4(inst, machine, color); + } + break; + case OPCODE_TXB: /* GL_ARB_fragment_program only */ + /* Texel lookup with LOD bias */ + { + GLfloat texcoord[4], color[4], lodBias; + + fetch_vector4(&inst->SrcReg[0], machine, texcoord); + + /* texcoord[3] is the bias to add to lambda */ + lodBias = texcoord[3]; + + fetch_texel(ctx, machine, inst, texcoord, lodBias, color); + + store_vector4(inst, machine, color); + } + break; + case OPCODE_TXD: /* GL_NV_fragment_program only */ + /* Texture lookup w/ partial derivatives for LOD */ + { + GLfloat texcoord[4], dtdx[4], dtdy[4], color[4]; + fetch_vector4(&inst->SrcReg[0], machine, texcoord); + fetch_vector4(&inst->SrcReg[1], machine, dtdx); + fetch_vector4(&inst->SrcReg[2], machine, dtdy); + machine->FetchTexelDeriv(ctx, texcoord, dtdx, dtdy, + 0.0, /* lodBias */ + inst->TexSrcUnit, color); + store_vector4(inst, machine, color); + } + break; + case OPCODE_TXL: + /* Texel lookup with explicit LOD */ + { + GLfloat texcoord[4], color[4], lod; + + fetch_vector4(&inst->SrcReg[0], machine, texcoord); + + /* texcoord[3] is the LOD */ + lod = texcoord[3]; + + machine->FetchTexelLod(ctx, texcoord, lod, + machine->Samplers[inst->TexSrcUnit], color); + + store_vector4(inst, machine, color); + } + break; + case OPCODE_TXP: /* GL_ARB_fragment_program only */ + /* Texture lookup w/ projective divide */ + { + GLfloat texcoord[4], color[4]; + + fetch_vector4(&inst->SrcReg[0], machine, texcoord); + /* Not so sure about this test - if texcoord[3] is + * zero, we'd probably be fine except for an ASSERT in + * IROUND_POS() which gets triggered by the inf values created. + */ + if (texcoord[3] != 0.0) { + texcoord[0] /= texcoord[3]; + texcoord[1] /= texcoord[3]; + texcoord[2] /= texcoord[3]; + } + + fetch_texel(ctx, machine, inst, texcoord, 0.0, color); + + store_vector4(inst, machine, color); + } + break; + case OPCODE_TXP_NV: /* GL_NV_fragment_program only */ + /* Texture lookup w/ projective divide, as above, but do not + * do the divide by w if sampling from a cube map. + */ + { + GLfloat texcoord[4], color[4]; + + fetch_vector4(&inst->SrcReg[0], machine, texcoord); + if (inst->TexSrcTarget != TEXTURE_CUBE_INDEX && + texcoord[3] != 0.0) { + texcoord[0] /= texcoord[3]; + texcoord[1] /= texcoord[3]; + texcoord[2] /= texcoord[3]; + } + + fetch_texel(ctx, machine, inst, texcoord, 0.0, color); + + store_vector4(inst, machine, color); + } + break; + case OPCODE_TRUNC: /* truncate toward zero */ + { + GLfloat a[4], result[4]; + fetch_vector4(&inst->SrcReg[0], machine, a); + result[0] = (GLfloat) (GLint) a[0]; + result[1] = (GLfloat) (GLint) a[1]; + result[2] = (GLfloat) (GLint) a[2]; + result[3] = (GLfloat) (GLint) a[3]; + store_vector4(inst, machine, result); + } + break; + case OPCODE_UP2H: /* unpack two 16-bit floats */ + { + const GLuint raw = fetch_vector1ui(&inst->SrcReg[0], machine); + GLfloat result[4]; + GLushort hx, hy; + hx = raw & 0xffff; + hy = raw >> 16; + result[0] = result[2] = _mesa_half_to_float(hx); + result[1] = result[3] = _mesa_half_to_float(hy); + store_vector4(inst, machine, result); + } + break; + case OPCODE_UP2US: /* unpack two GLushorts */ + { + const GLuint raw = fetch_vector1ui(&inst->SrcReg[0], machine); + GLfloat result[4]; + GLushort usx, usy; + usx = raw & 0xffff; + usy = raw >> 16; + result[0] = result[2] = usx * (1.0f / 65535.0f); + result[1] = result[3] = usy * (1.0f / 65535.0f); + store_vector4(inst, machine, result); + } + break; + case OPCODE_UP4B: /* unpack four GLbytes */ + { + const GLuint raw = fetch_vector1ui(&inst->SrcReg[0], machine); + GLfloat result[4]; + result[0] = (((raw >> 0) & 0xff) - 128) / 127.0F; + result[1] = (((raw >> 8) & 0xff) - 128) / 127.0F; + result[2] = (((raw >> 16) & 0xff) - 128) / 127.0F; + result[3] = (((raw >> 24) & 0xff) - 128) / 127.0F; + store_vector4(inst, machine, result); + } + break; + case OPCODE_UP4UB: /* unpack four GLubytes */ + { + const GLuint raw = fetch_vector1ui(&inst->SrcReg[0], machine); + GLfloat result[4]; + result[0] = ((raw >> 0) & 0xff) / 255.0F; + result[1] = ((raw >> 8) & 0xff) / 255.0F; + result[2] = ((raw >> 16) & 0xff) / 255.0F; + result[3] = ((raw >> 24) & 0xff) / 255.0F; + store_vector4(inst, machine, result); + } + break; + case OPCODE_XOR: /* bitwise XOR */ + { + GLuint a[4], b[4], result[4]; + fetch_vector4ui(&inst->SrcReg[0], machine, a); + fetch_vector4ui(&inst->SrcReg[1], machine, b); + result[0] = a[0] ^ b[0]; + result[1] = a[1] ^ b[1]; + result[2] = a[2] ^ b[2]; + result[3] = a[3] ^ b[3]; + store_vector4ui(inst, machine, result); + } + break; + case OPCODE_XPD: /* cross product */ + { + GLfloat a[4], b[4], result[4]; + fetch_vector4(&inst->SrcReg[0], machine, a); + fetch_vector4(&inst->SrcReg[1], machine, b); + result[0] = a[1] * b[2] - a[2] * b[1]; + result[1] = a[2] * b[0] - a[0] * b[2]; + result[2] = a[0] * b[1] - a[1] * b[0]; + result[3] = 1.0; + store_vector4(inst, machine, result); + if (DEBUG_PROG) { + printf("XPD (%g %g %g %g) = (%g %g %g) X (%g %g %g)\n", + result[0], result[1], result[2], result[3], + a[0], a[1], a[2], b[0], b[1], b[2]); + } + } + break; + case OPCODE_X2D: /* 2-D matrix transform */ + { + GLfloat a[4], b[4], c[4], result[4]; + fetch_vector4(&inst->SrcReg[0], machine, a); + fetch_vector4(&inst->SrcReg[1], machine, b); + fetch_vector4(&inst->SrcReg[2], machine, c); + result[0] = a[0] + b[0] * c[0] + b[1] * c[1]; + result[1] = a[1] + b[0] * c[2] + b[1] * c[3]; + result[2] = a[2] + b[0] * c[0] + b[1] * c[1]; + result[3] = a[3] + b[0] * c[2] + b[1] * c[3]; + store_vector4(inst, machine, result); + } + break; + case OPCODE_PRINT: + { + if (inst->SrcReg[0].File != PROGRAM_UNDEFINED) { + GLfloat a[4]; + fetch_vector4(&inst->SrcReg[0], machine, a); + printf("%s%g, %g, %g, %g\n", (const char *) inst->Data, + a[0], a[1], a[2], a[3]); + } + else { + printf("%s\n", (const char *) inst->Data); + } + } + break; + case OPCODE_END: + return GL_TRUE; + default: + _mesa_problem(ctx, "Bad opcode %d in _mesa_execute_program", + inst->Opcode); + return GL_TRUE; /* return value doesn't matter */ + } + + numExec++; + if (numExec > maxExec) { + static GLboolean reported = GL_FALSE; + if (!reported) { + _mesa_problem(ctx, "Infinite loop detected in fragment program"); + reported = GL_TRUE; + } + return GL_TRUE; + } + + } /* for pc */ + + return GL_TRUE; +} diff --git a/mesalib/src/mesa/program/prog_execute.h b/mesalib/src/mesa/program/prog_execute.h new file mode 100644 index 000000000..f59b65176 --- /dev/null +++ b/mesalib/src/mesa/program/prog_execute.h @@ -0,0 +1,86 @@ +/* + * Mesa 3-D graphics library + * Version: 7.0.3 + * + * Copyright (C) 1999-2007 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 PROG_EXECUTE_H +#define PROG_EXECUTE_H + +#include "main/config.h" +#include "main/mtypes.h" + + +typedef void (*FetchTexelLodFunc)(GLcontext *ctx, const GLfloat texcoord[4], + GLfloat lambda, GLuint unit, GLfloat color[4]); + +typedef void (*FetchTexelDerivFunc)(GLcontext *ctx, const GLfloat texcoord[4], + const GLfloat texdx[4], + const GLfloat texdy[4], + GLfloat lodBias, + GLuint unit, GLfloat color[4]); + + +/** + * Virtual machine state used during execution of vertex/fragment programs. + */ +struct gl_program_machine +{ + const struct gl_program *CurProgram; + + /** Fragment Input attributes */ + GLfloat (*Attribs)[MAX_WIDTH][4]; + GLfloat (*DerivX)[4]; + GLfloat (*DerivY)[4]; + GLuint NumDeriv; /**< Max index into DerivX/Y arrays */ + GLuint CurElement; /**< Index into Attribs arrays */ + + /** Vertex Input attribs */ + GLfloat VertAttribs[VERT_ATTRIB_MAX][4]; + + GLfloat Temporaries[MAX_PROGRAM_TEMPS][4]; + GLfloat Outputs[MAX_PROGRAM_OUTPUTS][4]; + GLfloat (*EnvParams)[4]; /**< Vertex or Fragment env parameters */ + GLuint CondCodes[4]; /**< COND_* value for x/y/z/w */ + GLint AddressReg[MAX_PROGRAM_ADDRESS_REGS][4]; + + const GLubyte *Samplers; /** Array mapping sampler var to tex unit */ + + GLuint CallStack[MAX_PROGRAM_CALL_DEPTH]; /**< For CAL/RET instructions */ + GLuint StackDepth; /**< Index/ptr to top of CallStack[] */ + + /** Texture fetch functions */ + FetchTexelLodFunc FetchTexelLod; + FetchTexelDerivFunc FetchTexelDeriv; +}; + + +extern void +_mesa_get_program_register(GLcontext *ctx, gl_register_file file, + GLuint index, GLfloat val[4]); + +extern GLboolean +_mesa_execute_program(GLcontext *ctx, + const struct gl_program *program, + struct gl_program_machine *machine); + + +#endif /* PROG_EXECUTE_H */ diff --git a/mesalib/src/mesa/program/prog_instruction.c b/mesalib/src/mesa/program/prog_instruction.c new file mode 100644 index 000000000..5d6cb476c --- /dev/null +++ b/mesalib/src/mesa/program/prog_instruction.c @@ -0,0 +1,354 @@ +/* + * Mesa 3-D graphics library + * Version: 7.3 + * + * Copyright (C) 1999-2008 Brian Paul All Rights Reserved. + * Copyright (C) 1999-2009 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. + */ + + +#include "main/glheader.h" +#include "main/imports.h" +#include "main/mtypes.h" +#include "prog_instruction.h" + + +/** + * Initialize program instruction fields to defaults. + * \param inst first instruction to initialize + * \param count number of instructions to initialize + */ +void +_mesa_init_instructions(struct prog_instruction *inst, GLuint count) +{ + GLuint i; + + memset(inst, 0, count * sizeof(struct prog_instruction)); + + for (i = 0; i < count; i++) { + inst[i].SrcReg[0].File = PROGRAM_UNDEFINED; + inst[i].SrcReg[0].Swizzle = SWIZZLE_NOOP; + inst[i].SrcReg[1].File = PROGRAM_UNDEFINED; + inst[i].SrcReg[1].Swizzle = SWIZZLE_NOOP; + inst[i].SrcReg[2].File = PROGRAM_UNDEFINED; + inst[i].SrcReg[2].Swizzle = SWIZZLE_NOOP; + + inst[i].DstReg.File = PROGRAM_UNDEFINED; + inst[i].DstReg.WriteMask = WRITEMASK_XYZW; + inst[i].DstReg.CondMask = COND_TR; + inst[i].DstReg.CondSwizzle = SWIZZLE_NOOP; + + inst[i].SaturateMode = SATURATE_OFF; + inst[i].Precision = FLOAT32; + } +} + + +/** + * Allocate an array of program instructions. + * \param numInst number of instructions + * \return pointer to instruction memory + */ +struct prog_instruction * +_mesa_alloc_instructions(GLuint numInst) +{ + return (struct prog_instruction *) + calloc(1, numInst * sizeof(struct prog_instruction)); +} + + +/** + * Reallocate memory storing an array of program instructions. + * This is used when we need to append additional instructions onto an + * program. + * \param oldInst pointer to first of old/src instructions + * \param numOldInst number of instructions at <oldInst> + * \param numNewInst desired size of new instruction array. + * \return pointer to start of new instruction array. + */ +struct prog_instruction * +_mesa_realloc_instructions(struct prog_instruction *oldInst, + GLuint numOldInst, GLuint numNewInst) +{ + struct prog_instruction *newInst; + + newInst = (struct prog_instruction *) + _mesa_realloc(oldInst, + numOldInst * sizeof(struct prog_instruction), + numNewInst * sizeof(struct prog_instruction)); + + return newInst; +} + + +/** + * Copy an array of program instructions. + * \param dest pointer to destination. + * \param src pointer to source. + * \param n number of instructions to copy. + * \return pointer to destination. + */ +struct prog_instruction * +_mesa_copy_instructions(struct prog_instruction *dest, + const struct prog_instruction *src, GLuint n) +{ + GLuint i; + memcpy(dest, src, n * sizeof(struct prog_instruction)); + for (i = 0; i < n; i++) { + if (src[i].Comment) + dest[i].Comment = _mesa_strdup(src[i].Comment); + } + return dest; +} + + +/** + * Free an array of instructions + */ +void +_mesa_free_instructions(struct prog_instruction *inst, GLuint count) +{ + GLuint i; + for (i = 0; i < count; i++) { + if (inst[i].Data) + free(inst[i].Data); + if (inst[i].Comment) + free((char *) inst[i].Comment); + } + free(inst); +} + + +/** + * Basic info about each instruction + */ +struct instruction_info +{ + gl_inst_opcode Opcode; + const char *Name; + GLuint NumSrcRegs; + GLuint NumDstRegs; +}; + +/** + * Instruction info + * \note Opcode should equal array index! + */ +static const struct instruction_info InstInfo[MAX_OPCODE] = { + { OPCODE_NOP, "NOP", 0, 0 }, + { OPCODE_ABS, "ABS", 1, 1 }, + { OPCODE_ADD, "ADD", 2, 1 }, + { OPCODE_AND, "AND", 2, 1 }, + { OPCODE_ARA, "ARA", 1, 1 }, + { OPCODE_ARL, "ARL", 1, 1 }, + { OPCODE_ARL_NV, "ARL_NV", 1, 1 }, + { OPCODE_ARR, "ARL", 1, 1 }, + { OPCODE_BGNLOOP,"BGNLOOP", 0, 0 }, + { OPCODE_BGNSUB, "BGNSUB", 0, 0 }, + { OPCODE_BRA, "BRA", 0, 0 }, + { OPCODE_BRK, "BRK", 0, 0 }, + { OPCODE_CAL, "CAL", 0, 0 }, + { OPCODE_CMP, "CMP", 3, 1 }, + { OPCODE_CONT, "CONT", 0, 0 }, + { OPCODE_COS, "COS", 1, 1 }, + { OPCODE_DDX, "DDX", 1, 1 }, + { OPCODE_DDY, "DDY", 1, 1 }, + { OPCODE_DP2, "DP2", 2, 1 }, + { OPCODE_DP2A, "DP2A", 3, 1 }, + { OPCODE_DP3, "DP3", 2, 1 }, + { OPCODE_DP4, "DP4", 2, 1 }, + { OPCODE_DPH, "DPH", 2, 1 }, + { OPCODE_DST, "DST", 2, 1 }, + { OPCODE_ELSE, "ELSE", 0, 0 }, + { OPCODE_EMIT_VERTEX, "EMIT_VERTEX", 0, 0 }, + { OPCODE_END, "END", 0, 0 }, + { OPCODE_END_PRIMITIVE, "END_PRIMITIVE", 0, 0 }, + { OPCODE_ENDIF, "ENDIF", 0, 0 }, + { OPCODE_ENDLOOP,"ENDLOOP", 0, 0 }, + { OPCODE_ENDSUB, "ENDSUB", 0, 0 }, + { OPCODE_EX2, "EX2", 1, 1 }, + { OPCODE_EXP, "EXP", 1, 1 }, + { OPCODE_FLR, "FLR", 1, 1 }, + { OPCODE_FRC, "FRC", 1, 1 }, + { OPCODE_IF, "IF", 1, 0 }, + { OPCODE_KIL, "KIL", 1, 0 }, + { OPCODE_KIL_NV, "KIL_NV", 0, 0 }, + { OPCODE_LG2, "LG2", 1, 1 }, + { OPCODE_LIT, "LIT", 1, 1 }, + { OPCODE_LOG, "LOG", 1, 1 }, + { OPCODE_LRP, "LRP", 3, 1 }, + { OPCODE_MAD, "MAD", 3, 1 }, + { OPCODE_MAX, "MAX", 2, 1 }, + { OPCODE_MIN, "MIN", 2, 1 }, + { OPCODE_MOV, "MOV", 1, 1 }, + { OPCODE_MUL, "MUL", 2, 1 }, + { OPCODE_NOISE1, "NOISE1", 1, 1 }, + { OPCODE_NOISE2, "NOISE2", 1, 1 }, + { OPCODE_NOISE3, "NOISE3", 1, 1 }, + { OPCODE_NOISE4, "NOISE4", 1, 1 }, + { OPCODE_NOT, "NOT", 1, 1 }, + { OPCODE_NRM3, "NRM3", 1, 1 }, + { OPCODE_NRM4, "NRM4", 1, 1 }, + { OPCODE_OR, "OR", 2, 1 }, + { OPCODE_PK2H, "PK2H", 1, 1 }, + { OPCODE_PK2US, "PK2US", 1, 1 }, + { OPCODE_PK4B, "PK4B", 1, 1 }, + { OPCODE_PK4UB, "PK4UB", 1, 1 }, + { OPCODE_POW, "POW", 2, 1 }, + { OPCODE_POPA, "POPA", 0, 0 }, + { OPCODE_PRINT, "PRINT", 1, 0 }, + { OPCODE_PUSHA, "PUSHA", 0, 0 }, + { OPCODE_RCC, "RCC", 1, 1 }, + { OPCODE_RCP, "RCP", 1, 1 }, + { OPCODE_RET, "RET", 0, 0 }, + { OPCODE_RFL, "RFL", 1, 1 }, + { OPCODE_RSQ, "RSQ", 1, 1 }, + { OPCODE_SCS, "SCS", 1, 1 }, + { OPCODE_SEQ, "SEQ", 2, 1 }, + { OPCODE_SFL, "SFL", 0, 1 }, + { OPCODE_SGE, "SGE", 2, 1 }, + { OPCODE_SGT, "SGT", 2, 1 }, + { OPCODE_SIN, "SIN", 1, 1 }, + { OPCODE_SLE, "SLE", 2, 1 }, + { OPCODE_SLT, "SLT", 2, 1 }, + { OPCODE_SNE, "SNE", 2, 1 }, + { OPCODE_SSG, "SSG", 1, 1 }, + { OPCODE_STR, "STR", 0, 1 }, + { OPCODE_SUB, "SUB", 2, 1 }, + { OPCODE_SWZ, "SWZ", 1, 1 }, + { OPCODE_TEX, "TEX", 1, 1 }, + { OPCODE_TXB, "TXB", 1, 1 }, + { OPCODE_TXD, "TXD", 3, 1 }, + { OPCODE_TXL, "TXL", 1, 1 }, + { OPCODE_TXP, "TXP", 1, 1 }, + { OPCODE_TXP_NV, "TXP_NV", 1, 1 }, + { OPCODE_TRUNC, "TRUNC", 1, 1 }, + { OPCODE_UP2H, "UP2H", 1, 1 }, + { OPCODE_UP2US, "UP2US", 1, 1 }, + { OPCODE_UP4B, "UP4B", 1, 1 }, + { OPCODE_UP4UB, "UP4UB", 1, 1 }, + { OPCODE_X2D, "X2D", 3, 1 }, + { OPCODE_XOR, "XOR", 2, 1 }, + { OPCODE_XPD, "XPD", 2, 1 } +}; + + +/** + * Return the number of src registers for the given instruction/opcode. + */ +GLuint +_mesa_num_inst_src_regs(gl_inst_opcode opcode) +{ + ASSERT(opcode < MAX_OPCODE); + ASSERT(opcode == InstInfo[opcode].Opcode); + ASSERT(OPCODE_XPD == InstInfo[OPCODE_XPD].Opcode); + return InstInfo[opcode].NumSrcRegs; +} + + +/** + * Return the number of dst registers for the given instruction/opcode. + */ +GLuint +_mesa_num_inst_dst_regs(gl_inst_opcode opcode) +{ + ASSERT(opcode < MAX_OPCODE); + ASSERT(opcode == InstInfo[opcode].Opcode); + ASSERT(OPCODE_XPD == InstInfo[OPCODE_XPD].Opcode); + return InstInfo[opcode].NumDstRegs; +} + + +GLboolean +_mesa_is_tex_instruction(gl_inst_opcode opcode) +{ + return (opcode == OPCODE_TEX || + opcode == OPCODE_TXB || + opcode == OPCODE_TXD || + opcode == OPCODE_TXL || + opcode == OPCODE_TXP); +} + + +/** + * Check if there's a potential src/dst register data dependency when + * using SOA execution. + * Example: + * MOV T, T.yxwz; + * This would expand into: + * MOV t0, t1; + * MOV t1, t0; + * MOV t2, t3; + * MOV t3, t2; + * The second instruction will have the wrong value for t0 if executed as-is. + */ +GLboolean +_mesa_check_soa_dependencies(const struct prog_instruction *inst) +{ + GLuint i, chan; + + if (inst->DstReg.WriteMask == WRITEMASK_X || + inst->DstReg.WriteMask == WRITEMASK_Y || + inst->DstReg.WriteMask == WRITEMASK_Z || + inst->DstReg.WriteMask == WRITEMASK_W || + inst->DstReg.WriteMask == 0x0) { + /* no chance of data dependency */ + return GL_FALSE; + } + + /* loop over src regs */ + for (i = 0; i < 3; i++) { + if (inst->SrcReg[i].File == inst->DstReg.File && + inst->SrcReg[i].Index == inst->DstReg.Index) { + /* loop over dest channels */ + GLuint channelsWritten = 0x0; + for (chan = 0; chan < 4; chan++) { + if (inst->DstReg.WriteMask & (1 << chan)) { + /* check if we're reading a channel that's been written */ + GLuint swizzle = GET_SWZ(inst->SrcReg[i].Swizzle, chan); + if (swizzle <= SWIZZLE_W && + (channelsWritten & (1 << swizzle))) { + return GL_TRUE; + } + + channelsWritten |= (1 << chan); + } + } + } + } + return GL_FALSE; +} + + +/** + * Return string name for given program opcode. + */ +const char * +_mesa_opcode_string(gl_inst_opcode opcode) +{ + if (opcode < MAX_OPCODE) + return InstInfo[opcode].Name; + else { + static char s[20]; + _mesa_snprintf(s, sizeof(s), "OP%u", opcode); + return s; + } +} + diff --git a/mesalib/src/mesa/program/prog_instruction.h b/mesalib/src/mesa/program/prog_instruction.h new file mode 100644 index 000000000..ca90de7ce --- /dev/null +++ b/mesalib/src/mesa/program/prog_instruction.h @@ -0,0 +1,454 @@ +/* + * Mesa 3-D graphics library + * Version: 7.3 + * + * Copyright (C) 1999-2008 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. + */ + + +/** + * \file prog_instruction.h + * + * Vertex/fragment program instruction datatypes and constants. + * + * \author Brian Paul + * \author Keith Whitwell + * \author Ian Romanick <idr@us.ibm.com> + */ + + +#ifndef PROG_INSTRUCTION_H +#define PROG_INSTRUCTION_H + + +#include "main/glheader.h" + + +/** + * Swizzle indexes. + * Do not change! + */ +/*@{*/ +#define SWIZZLE_X 0 +#define SWIZZLE_Y 1 +#define SWIZZLE_Z 2 +#define SWIZZLE_W 3 +#define SWIZZLE_ZERO 4 /**< For SWZ instruction only */ +#define SWIZZLE_ONE 5 /**< For SWZ instruction only */ +#define SWIZZLE_NIL 7 /**< used during shader code gen (undefined value) */ +/*@}*/ + +#define MAKE_SWIZZLE4(a,b,c,d) (((a)<<0) | ((b)<<3) | ((c)<<6) | ((d)<<9)) +#define SWIZZLE_NOOP MAKE_SWIZZLE4(0,1,2,3) +#define GET_SWZ(swz, idx) (((swz) >> ((idx)*3)) & 0x7) +#define GET_BIT(msk, idx) (((msk) >> (idx)) & 0x1) + +#define SWIZZLE_XYZW MAKE_SWIZZLE4(SWIZZLE_X, SWIZZLE_Y, SWIZZLE_Z, SWIZZLE_W) +#define SWIZZLE_XXXX MAKE_SWIZZLE4(SWIZZLE_X, SWIZZLE_X, SWIZZLE_X, SWIZZLE_X) +#define SWIZZLE_YYYY MAKE_SWIZZLE4(SWIZZLE_Y, SWIZZLE_Y, SWIZZLE_Y, SWIZZLE_Y) +#define SWIZZLE_ZZZZ MAKE_SWIZZLE4(SWIZZLE_Z, SWIZZLE_Z, SWIZZLE_Z, SWIZZLE_Z) +#define SWIZZLE_WWWW MAKE_SWIZZLE4(SWIZZLE_W, SWIZZLE_W, SWIZZLE_W, SWIZZLE_W) + + +/** + * Writemask values, 1 bit per component. + */ +/*@{*/ +#define WRITEMASK_X 0x1 +#define WRITEMASK_Y 0x2 +#define WRITEMASK_XY 0x3 +#define WRITEMASK_Z 0x4 +#define WRITEMASK_XZ 0x5 +#define WRITEMASK_YZ 0x6 +#define WRITEMASK_XYZ 0x7 +#define WRITEMASK_W 0x8 +#define WRITEMASK_XW 0x9 +#define WRITEMASK_YW 0xa +#define WRITEMASK_XYW 0xb +#define WRITEMASK_ZW 0xc +#define WRITEMASK_XZW 0xd +#define WRITEMASK_YZW 0xe +#define WRITEMASK_XYZW 0xf +/*@}*/ + + +/** + * Condition codes + */ +/*@{*/ +#define COND_GT 1 /**< greater than zero */ +#define COND_EQ 2 /**< equal to zero */ +#define COND_LT 3 /**< less than zero */ +#define COND_UN 4 /**< unordered (NaN) */ +#define COND_GE 5 /**< greater than or equal to zero */ +#define COND_LE 6 /**< less than or equal to zero */ +#define COND_NE 7 /**< not equal to zero */ +#define COND_TR 8 /**< always true */ +#define COND_FL 9 /**< always false */ +/*@}*/ + + +/** + * Instruction precision for GL_NV_fragment_program + */ +/*@{*/ +#define FLOAT32 0x1 +#define FLOAT16 0x2 +#define FIXED12 0x4 +/*@}*/ + + +/** + * Saturation modes when storing values. + */ +/*@{*/ +#define SATURATE_OFF 0 +#define SATURATE_ZERO_ONE 1 +/*@}*/ + + +/** + * Per-component negation masks + */ +/*@{*/ +#define NEGATE_X 0x1 +#define NEGATE_Y 0x2 +#define NEGATE_Z 0x4 +#define NEGATE_W 0x8 +#define NEGATE_XYZ 0x7 +#define NEGATE_XYZW 0xf +#define NEGATE_NONE 0x0 +/*@}*/ + + +/** + * Program instruction opcodes for vertex, fragment and geometry programs. + */ +typedef enum prog_opcode { + /* ARB_vp ARB_fp NV_vp NV_fp GLSL */ + /*------------------------------------------*/ + OPCODE_NOP = 0, /* X */ + OPCODE_ABS, /* X X 1.1 X */ + OPCODE_ADD, /* X X X X X */ + OPCODE_AND, /* */ + OPCODE_ARA, /* 2 */ + OPCODE_ARL, /* X X X */ + OPCODE_ARL_NV, /* 2 */ + OPCODE_ARR, /* 2 */ + OPCODE_BGNLOOP, /* opt */ + OPCODE_BGNSUB, /* opt */ + OPCODE_BRA, /* 2 X */ + OPCODE_BRK, /* 2 opt */ + OPCODE_CAL, /* 2 2 X */ + OPCODE_CMP, /* X X */ + OPCODE_CONT, /* opt */ + OPCODE_COS, /* X 2 X X */ + OPCODE_DDX, /* X X */ + OPCODE_DDY, /* X X */ + OPCODE_DP2, /* 2 X */ + OPCODE_DP2A, /* 2 */ + OPCODE_DP3, /* X X X X X */ + OPCODE_DP4, /* X X X X X */ + OPCODE_DPH, /* X X 1.1 */ + OPCODE_DST, /* X X X X */ + OPCODE_ELSE, /* X */ + OPCODE_EMIT_VERTEX,/* X */ + OPCODE_END, /* X X X X opt */ + OPCODE_END_PRIMITIVE,/* X */ + OPCODE_ENDIF, /* opt */ + OPCODE_ENDLOOP, /* opt */ + OPCODE_ENDSUB, /* opt */ + OPCODE_EX2, /* X X 2 X X */ + OPCODE_EXP, /* X X X */ + OPCODE_FLR, /* X X 2 X X */ + OPCODE_FRC, /* X X 2 X X */ + OPCODE_IF, /* opt */ + OPCODE_KIL, /* X */ + OPCODE_KIL_NV, /* X X */ + OPCODE_LG2, /* X X 2 X X */ + OPCODE_LIT, /* X X X X */ + OPCODE_LOG, /* X X X */ + OPCODE_LRP, /* X X X */ + OPCODE_MAD, /* X X X X X */ + OPCODE_MAX, /* X X X X X */ + OPCODE_MIN, /* X X X X X */ + OPCODE_MOV, /* X X X X X */ + OPCODE_MUL, /* X X X X X */ + OPCODE_NOISE1, /* X */ + OPCODE_NOISE2, /* X */ + OPCODE_NOISE3, /* X */ + OPCODE_NOISE4, /* X */ + OPCODE_NOT, /* */ + OPCODE_NRM3, /* X */ + OPCODE_NRM4, /* X */ + OPCODE_OR, /* */ + OPCODE_PK2H, /* X */ + OPCODE_PK2US, /* X */ + OPCODE_PK4B, /* X */ + OPCODE_PK4UB, /* X */ + OPCODE_POW, /* X X X X */ + OPCODE_POPA, /* 3 */ + OPCODE_PRINT, /* X X */ + OPCODE_PUSHA, /* 3 */ + OPCODE_RCC, /* 1.1 */ + OPCODE_RCP, /* X X X X X */ + OPCODE_RET, /* 2 2 X */ + OPCODE_RFL, /* X X */ + OPCODE_RSQ, /* X X X X X */ + OPCODE_SCS, /* X */ + OPCODE_SEQ, /* 2 X X */ + OPCODE_SFL, /* 2 X */ + OPCODE_SGE, /* X X X X X */ + OPCODE_SGT, /* 2 X X */ + OPCODE_SIN, /* X 2 X X */ + OPCODE_SLE, /* 2 X X */ + OPCODE_SLT, /* X X X X X */ + OPCODE_SNE, /* 2 X X */ + OPCODE_SSG, /* 2 */ + OPCODE_STR, /* 2 X */ + OPCODE_SUB, /* X X 1.1 X X */ + OPCODE_SWZ, /* X X */ + OPCODE_TEX, /* X 3 X X */ + OPCODE_TXB, /* X 3 X */ + OPCODE_TXD, /* X X */ + OPCODE_TXL, /* 3 2 X */ + OPCODE_TXP, /* X X */ + OPCODE_TXP_NV, /* 3 X */ + OPCODE_TRUNC, /* X */ + OPCODE_UP2H, /* X */ + OPCODE_UP2US, /* X */ + OPCODE_UP4B, /* X */ + OPCODE_UP4UB, /* X */ + OPCODE_X2D, /* X */ + OPCODE_XOR, /* */ + OPCODE_XPD, /* X X X */ + MAX_OPCODE +} gl_inst_opcode; + + +/** + * Number of bits for the src/dst register Index field. + * This limits the size of temp/uniform register files. + */ +#define INST_INDEX_BITS 10 + + +/** + * Instruction source register. + */ +struct prog_src_register +{ + GLuint File:4; /**< One of the PROGRAM_* register file values. */ + GLint Index:(INST_INDEX_BITS+1); /**< Extra bit here for sign bit. + * May be negative for relative addressing. + */ + GLuint Swizzle:12; + GLuint RelAddr:1; + + /** Take the component-wise absolute value */ + GLuint Abs:1; + + /** + * Post-Abs negation. + * This will either be NEGATE_NONE or NEGATE_XYZW, except for the SWZ + * instruction which allows per-component negation. + */ + GLuint Negate:4; + + /** + * Is the register two-dimensional. + * Two dimensional registers are of the + * REGISTER[index][index2] format. + * They are used by the geometry shaders where + * the first index is the index within an array + * and the second index is the semantic of the + * array, e.g. gl_PositionIn[index] would become + * INPUT[index][gl_PositionIn] + */ + GLuint HasIndex2:1; + GLuint RelAddr2:1; + GLint Index2:(INST_INDEX_BITS+1); /**< Extra bit here for sign bit. + * May be negative for relative + * addressing. */ +}; + + +/** + * Instruction destination register. + */ +struct prog_dst_register +{ + GLuint File:4; /**< One of the PROGRAM_* register file values */ + GLuint Index:INST_INDEX_BITS; /**< Unsigned, never negative */ + GLuint WriteMask:4; + GLuint RelAddr:1; + + /** + * \name Conditional destination update control. + * + * \since + * NV_fragment_program, NV_fragment_program_option, NV_vertex_program2, + * NV_vertex_program2_option. + */ + /*@{*/ + /** + * Takes one of the 9 possible condition values (EQ, FL, GT, GE, LE, LT, + * NE, TR, or UN). Dest reg is only written to if the matching + * (swizzled) condition code value passes. When a conditional update mask + * is not specified, this will be \c COND_TR. + */ + GLuint CondMask:4; + + /** + * Condition code swizzle value. + */ + GLuint CondSwizzle:12; + + /** + * Selects the condition code register to use for conditional destination + * update masking. In NV_fragmnet_program or NV_vertex_program2 mode, only + * condition code register 0 is available. In NV_vertex_program3 mode, + * condition code registers 0 and 1 are available. + */ + GLuint CondSrc:1; + /*@}*/ +}; + + +/** + * Vertex/fragment program instruction. + */ +struct prog_instruction +{ + gl_inst_opcode Opcode; + struct prog_src_register SrcReg[3]; + struct prog_dst_register DstReg; + + /** + * Indicates that the instruction should update the condition code + * register. + * + * \since + * NV_fragment_program, NV_fragment_program_option, NV_vertex_program2, + * NV_vertex_program2_option. + */ + GLuint CondUpdate:1; + + /** + * If prog_instruction::CondUpdate is \c GL_TRUE, this value selects the + * condition code register that is to be updated. + * + * In GL_NV_fragment_program or GL_NV_vertex_program2 mode, only condition + * code register 0 is available. In GL_NV_vertex_program3 mode, condition + * code registers 0 and 1 are available. + * + * \since + * NV_fragment_program, NV_fragment_program_option, NV_vertex_program2, + * NV_vertex_program2_option. + */ + GLuint CondDst:1; + + /** + * Saturate each value of the vectored result to the range [0,1] or the + * range [-1,1]. \c SSAT mode (i.e., saturation to the range [-1,1]) is + * only available in NV_fragment_program2 mode. + * Value is one of the SATURATE_* tokens. + * + * \since + * NV_fragment_program, NV_fragment_program_option, NV_vertex_program3. + */ + GLuint SaturateMode:2; + + /** + * Per-instruction selectable precision: FLOAT32, FLOAT16, FIXED12. + * + * \since + * NV_fragment_program, NV_fragment_program_option. + */ + GLuint Precision:3; + + /** + * \name Extra fields for TEX, TXB, TXD, TXL, TXP instructions. + */ + /*@{*/ + /** Source texture unit. */ + GLuint TexSrcUnit:5; + + /** Source texture target, one of TEXTURE_{1D,2D,3D,CUBE,RECT}_INDEX */ + GLuint TexSrcTarget:3; + + /** True if tex instruction should do shadow comparison */ + GLuint TexShadow:1; + /*@}*/ + + /** + * For BRA and CAL instructions, the location to jump to. + * For BGNLOOP, points to ENDLOOP (and vice-versa). + * For BRK, points to ENDLOOP + * For IF, points to ELSE or ENDIF. + * For ELSE, points to ENDIF. + */ + GLint BranchTarget; + + /** for debugging purposes */ + const char *Comment; + + /** Arbitrary data. Used for OPCODE_PRINT and some drivers */ + void *Data; + + /** for driver use (try to remove someday) */ + GLint Aux; +}; + + +extern void +_mesa_init_instructions(struct prog_instruction *inst, GLuint count); + +extern struct prog_instruction * +_mesa_alloc_instructions(GLuint numInst); + +extern struct prog_instruction * +_mesa_realloc_instructions(struct prog_instruction *oldInst, + GLuint numOldInst, GLuint numNewInst); + +extern struct prog_instruction * +_mesa_copy_instructions(struct prog_instruction *dest, + const struct prog_instruction *src, GLuint n); + +extern void +_mesa_free_instructions(struct prog_instruction *inst, GLuint count); + +extern GLuint +_mesa_num_inst_src_regs(gl_inst_opcode opcode); + +extern GLuint +_mesa_num_inst_dst_regs(gl_inst_opcode opcode); + +extern GLboolean +_mesa_is_tex_instruction(gl_inst_opcode opcode); + +extern GLboolean +_mesa_check_soa_dependencies(const struct prog_instruction *inst); + +extern const char * +_mesa_opcode_string(gl_inst_opcode opcode); + + +#endif /* PROG_INSTRUCTION_H */ diff --git a/mesalib/src/mesa/program/prog_noise.c b/mesalib/src/mesa/program/prog_noise.c new file mode 100644 index 000000000..1713ddb5f --- /dev/null +++ b/mesalib/src/mesa/program/prog_noise.c @@ -0,0 +1,638 @@ +/* + * Mesa 3-D graphics library + * Version: 6.5 + * + * Copyright (C) 2006 Brian Paul All Rights Reserved. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included + * in all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS + * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * BRIAN PAUL BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN + * AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN + * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. + */ + +/* + * SimplexNoise1234 + * Copyright (c) 2003-2005, Stefan Gustavson + * + * Contact: stegu@itn.liu.se + */ + +/** + * \file + * \brief C implementation of Perlin Simplex Noise over 1, 2, 3 and 4 dims. + * \author Stefan Gustavson (stegu@itn.liu.se) + * + * + * This implementation is "Simplex Noise" as presented by + * Ken Perlin at a relatively obscure and not often cited course + * session "Real-Time Shading" at Siggraph 2001 (before real + * time shading actually took on), under the title "hardware noise". + * The 3D function is numerically equivalent to his Java reference + * code available in the PDF course notes, although I re-implemented + * it from scratch to get more readable code. The 1D, 2D and 4D cases + * were implemented from scratch by me from Ken Perlin's text. + * + * This file has no dependencies on any other file, not even its own + * header file. The header file is made for use by external code only. + */ + + +#include "main/imports.h" +#include "prog_noise.h" + +#define FASTFLOOR(x) ( ((x)>0) ? ((int)x) : (((int)x)-1) ) + +/* + * --------------------------------------------------------------------- + * Static data + */ + +/** + * Permutation table. This is just a random jumble of all numbers 0-255, + * repeated twice to avoid wrapping the index at 255 for each lookup. + * This needs to be exactly the same for all instances on all platforms, + * so it's easiest to just keep it as static explicit data. + * This also removes the need for any initialisation of this class. + * + * Note that making this an int[] instead of a char[] might make the + * code run faster on platforms with a high penalty for unaligned single + * byte addressing. Intel x86 is generally single-byte-friendly, but + * some other CPUs are faster with 4-aligned reads. + * However, a char[] is smaller, which avoids cache trashing, and that + * is probably the most important aspect on most architectures. + * This array is accessed a *lot* by the noise functions. + * A vector-valued noise over 3D accesses it 96 times, and a + * float-valued 4D noise 64 times. We want this to fit in the cache! + */ +unsigned char perm[512] = { 151, 160, 137, 91, 90, 15, + 131, 13, 201, 95, 96, 53, 194, 233, 7, 225, 140, 36, 103, 30, 69, 142, 8, + 99, 37, 240, 21, 10, 23, + 190, 6, 148, 247, 120, 234, 75, 0, 26, 197, 62, 94, 252, 219, 203, 117, 35, + 11, 32, 57, 177, 33, + 88, 237, 149, 56, 87, 174, 20, 125, 136, 171, 168, 68, 175, 74, 165, 71, + 134, 139, 48, 27, 166, + 77, 146, 158, 231, 83, 111, 229, 122, 60, 211, 133, 230, 220, 105, 92, 41, + 55, 46, 245, 40, 244, + 102, 143, 54, 65, 25, 63, 161, 1, 216, 80, 73, 209, 76, 132, 187, 208, 89, + 18, 169, 200, 196, + 135, 130, 116, 188, 159, 86, 164, 100, 109, 198, 173, 186, 3, 64, 52, 217, + 226, 250, 124, 123, + 5, 202, 38, 147, 118, 126, 255, 82, 85, 212, 207, 206, 59, 227, 47, 16, 58, + 17, 182, 189, 28, 42, + 223, 183, 170, 213, 119, 248, 152, 2, 44, 154, 163, 70, 221, 153, 101, 155, + 167, 43, 172, 9, + 129, 22, 39, 253, 19, 98, 108, 110, 79, 113, 224, 232, 178, 185, 112, 104, + 218, 246, 97, 228, + 251, 34, 242, 193, 238, 210, 144, 12, 191, 179, 162, 241, 81, 51, 145, 235, + 249, 14, 239, 107, + 49, 192, 214, 31, 181, 199, 106, 157, 184, 84, 204, 176, 115, 121, 50, 45, + 127, 4, 150, 254, + 138, 236, 205, 93, 222, 114, 67, 29, 24, 72, 243, 141, 128, 195, 78, 66, + 215, 61, 156, 180, + 151, 160, 137, 91, 90, 15, + 131, 13, 201, 95, 96, 53, 194, 233, 7, 225, 140, 36, 103, 30, 69, 142, 8, + 99, 37, 240, 21, 10, 23, + 190, 6, 148, 247, 120, 234, 75, 0, 26, 197, 62, 94, 252, 219, 203, 117, 35, + 11, 32, 57, 177, 33, + 88, 237, 149, 56, 87, 174, 20, 125, 136, 171, 168, 68, 175, 74, 165, 71, + 134, 139, 48, 27, 166, + 77, 146, 158, 231, 83, 111, 229, 122, 60, 211, 133, 230, 220, 105, 92, 41, + 55, 46, 245, 40, 244, + 102, 143, 54, 65, 25, 63, 161, 1, 216, 80, 73, 209, 76, 132, 187, 208, 89, + 18, 169, 200, 196, + 135, 130, 116, 188, 159, 86, 164, 100, 109, 198, 173, 186, 3, 64, 52, 217, + 226, 250, 124, 123, + 5, 202, 38, 147, 118, 126, 255, 82, 85, 212, 207, 206, 59, 227, 47, 16, 58, + 17, 182, 189, 28, 42, + 223, 183, 170, 213, 119, 248, 152, 2, 44, 154, 163, 70, 221, 153, 101, 155, + 167, 43, 172, 9, + 129, 22, 39, 253, 19, 98, 108, 110, 79, 113, 224, 232, 178, 185, 112, 104, + 218, 246, 97, 228, + 251, 34, 242, 193, 238, 210, 144, 12, 191, 179, 162, 241, 81, 51, 145, 235, + 249, 14, 239, 107, + 49, 192, 214, 31, 181, 199, 106, 157, 184, 84, 204, 176, 115, 121, 50, 45, + 127, 4, 150, 254, + 138, 236, 205, 93, 222, 114, 67, 29, 24, 72, 243, 141, 128, 195, 78, 66, + 215, 61, 156, 180 +}; + +/* + * --------------------------------------------------------------------- + */ + +/* + * Helper functions to compute gradients-dot-residualvectors (1D to 4D) + * Note that these generate gradients of more than unit length. To make + * a close match with the value range of classic Perlin noise, the final + * noise values need to be rescaled to fit nicely within [-1,1]. + * (The simplex noise functions as such also have different scaling.) + * Note also that these noise functions are the most practical and useful + * signed version of Perlin noise. To return values according to the + * RenderMan specification from the SL noise() and pnoise() functions, + * the noise values need to be scaled and offset to [0,1], like this: + * float SLnoise = (SimplexNoise1234::noise(x,y,z) + 1.0) * 0.5; + */ + +static float +grad1(int hash, float x) +{ + int h = hash & 15; + float grad = 1.0f + (h & 7); /* Gradient value 1.0, 2.0, ..., 8.0 */ + if (h & 8) + grad = -grad; /* Set a random sign for the gradient */ + return (grad * x); /* Multiply the gradient with the distance */ +} + +static float +grad2(int hash, float x, float y) +{ + int h = hash & 7; /* Convert low 3 bits of hash code */ + float u = h < 4 ? x : y; /* into 8 simple gradient directions, */ + float v = h < 4 ? y : x; /* and compute the dot product with (x,y). */ + return ((h & 1) ? -u : u) + ((h & 2) ? -2.0f * v : 2.0f * v); +} + +static float +grad3(int hash, float x, float y, float z) +{ + int h = hash & 15; /* Convert low 4 bits of hash code into 12 simple */ + float u = h < 8 ? x : y; /* gradient directions, and compute dot product. */ + float v = h < 4 ? y : h == 12 || h == 14 ? x : z; /* Fix repeats at h = 12 to 15 */ + return ((h & 1) ? -u : u) + ((h & 2) ? -v : v); +} + +static float +grad4(int hash, float x, float y, float z, float t) +{ + int h = hash & 31; /* Convert low 5 bits of hash code into 32 simple */ + float u = h < 24 ? x : y; /* gradient directions, and compute dot product. */ + float v = h < 16 ? y : z; + float w = h < 8 ? z : t; + return ((h & 1) ? -u : u) + ((h & 2) ? -v : v) + ((h & 4) ? -w : w); +} + +/** + * A lookup table to traverse the simplex around a given point in 4D. + * Details can be found where this table is used, in the 4D noise method. + * TODO: This should not be required, backport it from Bill's GLSL code! + */ +static unsigned char simplex[64][4] = { + {0, 1, 2, 3}, {0, 1, 3, 2}, {0, 0, 0, 0}, {0, 2, 3, 1}, + {0, 0, 0, 0}, {0, 0, 0, 0}, {0, 0, 0, 0}, {1, 2, 3, 0}, + {0, 2, 1, 3}, {0, 0, 0, 0}, {0, 3, 1, 2}, {0, 3, 2, 1}, + {0, 0, 0, 0}, {0, 0, 0, 0}, {0, 0, 0, 0}, {1, 3, 2, 0}, + {0, 0, 0, 0}, {0, 0, 0, 0}, {0, 0, 0, 0}, {0, 0, 0, 0}, + {0, 0, 0, 0}, {0, 0, 0, 0}, {0, 0, 0, 0}, {0, 0, 0, 0}, + {1, 2, 0, 3}, {0, 0, 0, 0}, {1, 3, 0, 2}, {0, 0, 0, 0}, + {0, 0, 0, 0}, {0, 0, 0, 0}, {2, 3, 0, 1}, {2, 3, 1, 0}, + {1, 0, 2, 3}, {1, 0, 3, 2}, {0, 0, 0, 0}, {0, 0, 0, 0}, + {0, 0, 0, 0}, {2, 0, 3, 1}, {0, 0, 0, 0}, {2, 1, 3, 0}, + {0, 0, 0, 0}, {0, 0, 0, 0}, {0, 0, 0, 0}, {0, 0, 0, 0}, + {0, 0, 0, 0}, {0, 0, 0, 0}, {0, 0, 0, 0}, {0, 0, 0, 0}, + {2, 0, 1, 3}, {0, 0, 0, 0}, {0, 0, 0, 0}, {0, 0, 0, 0}, + {3, 0, 1, 2}, {3, 0, 2, 1}, {0, 0, 0, 0}, {3, 1, 2, 0}, + {2, 1, 0, 3}, {0, 0, 0, 0}, {0, 0, 0, 0}, {0, 0, 0, 0}, + {3, 1, 0, 2}, {0, 0, 0, 0}, {3, 2, 0, 1}, {3, 2, 1, 0} +}; + + +/** 1D simplex noise */ +GLfloat +_mesa_noise1(GLfloat x) +{ + int i0 = FASTFLOOR(x); + int i1 = i0 + 1; + float x0 = x - i0; + float x1 = x0 - 1.0f; + float t1 = 1.0f - x1 * x1; + float n0, n1; + + float t0 = 1.0f - x0 * x0; +/* if(t0 < 0.0f) t0 = 0.0f; // this never happens for the 1D case */ + t0 *= t0; + n0 = t0 * t0 * grad1(perm[i0 & 0xff], x0); + +/* if(t1 < 0.0f) t1 = 0.0f; // this never happens for the 1D case */ + t1 *= t1; + n1 = t1 * t1 * grad1(perm[i1 & 0xff], x1); + /* The maximum value of this noise is 8*(3/4)^4 = 2.53125 */ + /* A factor of 0.395 would scale to fit exactly within [-1,1], but */ + /* we want to match PRMan's 1D noise, so we scale it down some more. */ + return 0.25f * (n0 + n1); +} + + +/** 2D simplex noise */ +GLfloat +_mesa_noise2(GLfloat x, GLfloat y) +{ +#define F2 0.366025403f /* F2 = 0.5*(sqrt(3.0)-1.0) */ +#define G2 0.211324865f /* G2 = (3.0-Math.sqrt(3.0))/6.0 */ + + float n0, n1, n2; /* Noise contributions from the three corners */ + + /* Skew the input space to determine which simplex cell we're in */ + float s = (x + y) * F2; /* Hairy factor for 2D */ + float xs = x + s; + float ys = y + s; + int i = FASTFLOOR(xs); + int j = FASTFLOOR(ys); + + float t = (float) (i + j) * G2; + float X0 = i - t; /* Unskew the cell origin back to (x,y) space */ + float Y0 = j - t; + float x0 = x - X0; /* The x,y distances from the cell origin */ + float y0 = y - Y0; + + float x1, y1, x2, y2; + int ii, jj; + float t0, t1, t2; + + /* For the 2D case, the simplex shape is an equilateral triangle. */ + /* Determine which simplex we are in. */ + int i1, j1; /* Offsets for second (middle) corner of simplex in (i,j) coords */ + if (x0 > y0) { + i1 = 1; + j1 = 0; + } /* lower triangle, XY order: (0,0)->(1,0)->(1,1) */ + else { + i1 = 0; + j1 = 1; + } /* upper triangle, YX order: (0,0)->(0,1)->(1,1) */ + + /* A step of (1,0) in (i,j) means a step of (1-c,-c) in (x,y), and */ + /* a step of (0,1) in (i,j) means a step of (-c,1-c) in (x,y), where */ + /* c = (3-sqrt(3))/6 */ + + x1 = x0 - i1 + G2; /* Offsets for middle corner in (x,y) unskewed coords */ + y1 = y0 - j1 + G2; + x2 = x0 - 1.0f + 2.0f * G2; /* Offsets for last corner in (x,y) unskewed coords */ + y2 = y0 - 1.0f + 2.0f * G2; + + /* Wrap the integer indices at 256, to avoid indexing perm[] out of bounds */ + ii = i % 256; + jj = j % 256; + + /* Calculate the contribution from the three corners */ + t0 = 0.5f - x0 * x0 - y0 * y0; + if (t0 < 0.0f) + n0 = 0.0f; + else { + t0 *= t0; + n0 = t0 * t0 * grad2(perm[ii + perm[jj]], x0, y0); + } + + t1 = 0.5f - x1 * x1 - y1 * y1; + if (t1 < 0.0f) + n1 = 0.0f; + else { + t1 *= t1; + n1 = t1 * t1 * grad2(perm[ii + i1 + perm[jj + j1]], x1, y1); + } + + t2 = 0.5f - x2 * x2 - y2 * y2; + if (t2 < 0.0f) + n2 = 0.0f; + else { + t2 *= t2; + n2 = t2 * t2 * grad2(perm[ii + 1 + perm[jj + 1]], x2, y2); + } + + /* Add contributions from each corner to get the final noise value. */ + /* The result is scaled to return values in the interval [-1,1]. */ + return 40.0f * (n0 + n1 + n2); /* TODO: The scale factor is preliminary! */ +} + + +/** 3D simplex noise */ +GLfloat +_mesa_noise3(GLfloat x, GLfloat y, GLfloat z) +{ +/* Simple skewing factors for the 3D case */ +#define F3 0.333333333f +#define G3 0.166666667f + + float n0, n1, n2, n3; /* Noise contributions from the four corners */ + + /* Skew the input space to determine which simplex cell we're in */ + float s = (x + y + z) * F3; /* Very nice and simple skew factor for 3D */ + float xs = x + s; + float ys = y + s; + float zs = z + s; + int i = FASTFLOOR(xs); + int j = FASTFLOOR(ys); + int k = FASTFLOOR(zs); + + float t = (float) (i + j + k) * G3; + float X0 = i - t; /* Unskew the cell origin back to (x,y,z) space */ + float Y0 = j - t; + float Z0 = k - t; + float x0 = x - X0; /* The x,y,z distances from the cell origin */ + float y0 = y - Y0; + float z0 = z - Z0; + + float x1, y1, z1, x2, y2, z2, x3, y3, z3; + int ii, jj, kk; + float t0, t1, t2, t3; + + /* For the 3D case, the simplex shape is a slightly irregular tetrahedron. */ + /* Determine which simplex we are in. */ + int i1, j1, k1; /* Offsets for second corner of simplex in (i,j,k) coords */ + int i2, j2, k2; /* Offsets for third corner of simplex in (i,j,k) coords */ + +/* This code would benefit from a backport from the GLSL version! */ + if (x0 >= y0) { + if (y0 >= z0) { + i1 = 1; + j1 = 0; + k1 = 0; + i2 = 1; + j2 = 1; + k2 = 0; + } /* X Y Z order */ + else if (x0 >= z0) { + i1 = 1; + j1 = 0; + k1 = 0; + i2 = 1; + j2 = 0; + k2 = 1; + } /* X Z Y order */ + else { + i1 = 0; + j1 = 0; + k1 = 1; + i2 = 1; + j2 = 0; + k2 = 1; + } /* Z X Y order */ + } + else { /* x0<y0 */ + if (y0 < z0) { + i1 = 0; + j1 = 0; + k1 = 1; + i2 = 0; + j2 = 1; + k2 = 1; + } /* Z Y X order */ + else if (x0 < z0) { + i1 = 0; + j1 = 1; + k1 = 0; + i2 = 0; + j2 = 1; + k2 = 1; + } /* Y Z X order */ + else { + i1 = 0; + j1 = 1; + k1 = 0; + i2 = 1; + j2 = 1; + k2 = 0; + } /* Y X Z order */ + } + + /* A step of (1,0,0) in (i,j,k) means a step of (1-c,-c,-c) in + * (x,y,z), a step of (0,1,0) in (i,j,k) means a step of + * (-c,1-c,-c) in (x,y,z), and a step of (0,0,1) in (i,j,k) means a + * step of (-c,-c,1-c) in (x,y,z), where c = 1/6. + */ + + x1 = x0 - i1 + G3; /* Offsets for second corner in (x,y,z) coords */ + y1 = y0 - j1 + G3; + z1 = z0 - k1 + G3; + x2 = x0 - i2 + 2.0f * G3; /* Offsets for third corner in (x,y,z) coords */ + y2 = y0 - j2 + 2.0f * G3; + z2 = z0 - k2 + 2.0f * G3; + x3 = x0 - 1.0f + 3.0f * G3;/* Offsets for last corner in (x,y,z) coords */ + y3 = y0 - 1.0f + 3.0f * G3; + z3 = z0 - 1.0f + 3.0f * G3; + + /* Wrap the integer indices at 256 to avoid indexing perm[] out of bounds */ + ii = i % 256; + jj = j % 256; + kk = k % 256; + + /* Calculate the contribution from the four corners */ + t0 = 0.6f - x0 * x0 - y0 * y0 - z0 * z0; + if (t0 < 0.0f) + n0 = 0.0f; + else { + t0 *= t0; + n0 = t0 * t0 * grad3(perm[ii + perm[jj + perm[kk]]], x0, y0, z0); + } + + t1 = 0.6f - x1 * x1 - y1 * y1 - z1 * z1; + if (t1 < 0.0f) + n1 = 0.0f; + else { + t1 *= t1; + n1 = + t1 * t1 * grad3(perm[ii + i1 + perm[jj + j1 + perm[kk + k1]]], x1, + y1, z1); + } + + t2 = 0.6f - x2 * x2 - y2 * y2 - z2 * z2; + if (t2 < 0.0f) + n2 = 0.0f; + else { + t2 *= t2; + n2 = + t2 * t2 * grad3(perm[ii + i2 + perm[jj + j2 + perm[kk + k2]]], x2, + y2, z2); + } + + t3 = 0.6f - x3 * x3 - y3 * y3 - z3 * z3; + if (t3 < 0.0f) + n3 = 0.0f; + else { + t3 *= t3; + n3 = + t3 * t3 * grad3(perm[ii + 1 + perm[jj + 1 + perm[kk + 1]]], x3, y3, + z3); + } + + /* Add contributions from each corner to get the final noise value. + * The result is scaled to stay just inside [-1,1] + */ + return 32.0f * (n0 + n1 + n2 + n3); /* TODO: The scale factor is preliminary! */ +} + + +/** 4D simplex noise */ +GLfloat +_mesa_noise4(GLfloat x, GLfloat y, GLfloat z, GLfloat w) +{ + /* The skewing and unskewing factors are hairy again for the 4D case */ +#define F4 0.309016994f /* F4 = (Math.sqrt(5.0)-1.0)/4.0 */ +#define G4 0.138196601f /* G4 = (5.0-Math.sqrt(5.0))/20.0 */ + + float n0, n1, n2, n3, n4; /* Noise contributions from the five corners */ + + /* Skew the (x,y,z,w) space to determine which cell of 24 simplices we're in */ + float s = (x + y + z + w) * F4; /* Factor for 4D skewing */ + float xs = x + s; + float ys = y + s; + float zs = z + s; + float ws = w + s; + int i = FASTFLOOR(xs); + int j = FASTFLOOR(ys); + int k = FASTFLOOR(zs); + int l = FASTFLOOR(ws); + + float t = (i + j + k + l) * G4; /* Factor for 4D unskewing */ + float X0 = i - t; /* Unskew the cell origin back to (x,y,z,w) space */ + float Y0 = j - t; + float Z0 = k - t; + float W0 = l - t; + + float x0 = x - X0; /* The x,y,z,w distances from the cell origin */ + float y0 = y - Y0; + float z0 = z - Z0; + float w0 = w - W0; + + /* For the 4D case, the simplex is a 4D shape I won't even try to describe. + * To find out which of the 24 possible simplices we're in, we need to + * determine the magnitude ordering of x0, y0, z0 and w0. + * The method below is a good way of finding the ordering of x,y,z,w and + * then find the correct traversal order for the simplex we're in. + * First, six pair-wise comparisons are performed between each possible pair + * of the four coordinates, and the results are used to add up binary bits + * for an integer index. + */ + int c1 = (x0 > y0) ? 32 : 0; + int c2 = (x0 > z0) ? 16 : 0; + int c3 = (y0 > z0) ? 8 : 0; + int c4 = (x0 > w0) ? 4 : 0; + int c5 = (y0 > w0) ? 2 : 0; + int c6 = (z0 > w0) ? 1 : 0; + int c = c1 + c2 + c3 + c4 + c5 + c6; + + int i1, j1, k1, l1; /* The integer offsets for the second simplex corner */ + int i2, j2, k2, l2; /* The integer offsets for the third simplex corner */ + int i3, j3, k3, l3; /* The integer offsets for the fourth simplex corner */ + + float x1, y1, z1, w1, x2, y2, z2, w2, x3, y3, z3, w3, x4, y4, z4, w4; + int ii, jj, kk, ll; + float t0, t1, t2, t3, t4; + + /* + * simplex[c] is a 4-vector with the numbers 0, 1, 2 and 3 in some + * order. Many values of c will never occur, since e.g. x>y>z>w + * makes x<z, y<w and x<w impossible. Only the 24 indices which + * have non-zero entries make any sense. We use a thresholding to + * set the coordinates in turn from the largest magnitude. The + * number 3 in the "simplex" array is at the position of the + * largest coordinate. + */ + i1 = simplex[c][0] >= 3 ? 1 : 0; + j1 = simplex[c][1] >= 3 ? 1 : 0; + k1 = simplex[c][2] >= 3 ? 1 : 0; + l1 = simplex[c][3] >= 3 ? 1 : 0; + /* The number 2 in the "simplex" array is at the second largest coordinate. */ + i2 = simplex[c][0] >= 2 ? 1 : 0; + j2 = simplex[c][1] >= 2 ? 1 : 0; + k2 = simplex[c][2] >= 2 ? 1 : 0; + l2 = simplex[c][3] >= 2 ? 1 : 0; + /* The number 1 in the "simplex" array is at the second smallest coordinate. */ + i3 = simplex[c][0] >= 1 ? 1 : 0; + j3 = simplex[c][1] >= 1 ? 1 : 0; + k3 = simplex[c][2] >= 1 ? 1 : 0; + l3 = simplex[c][3] >= 1 ? 1 : 0; + /* The fifth corner has all coordinate offsets = 1, so no need to look that up. */ + + x1 = x0 - i1 + G4; /* Offsets for second corner in (x,y,z,w) coords */ + y1 = y0 - j1 + G4; + z1 = z0 - k1 + G4; + w1 = w0 - l1 + G4; + x2 = x0 - i2 + 2.0f * G4; /* Offsets for third corner in (x,y,z,w) coords */ + y2 = y0 - j2 + 2.0f * G4; + z2 = z0 - k2 + 2.0f * G4; + w2 = w0 - l2 + 2.0f * G4; + x3 = x0 - i3 + 3.0f * G4; /* Offsets for fourth corner in (x,y,z,w) coords */ + y3 = y0 - j3 + 3.0f * G4; + z3 = z0 - k3 + 3.0f * G4; + w3 = w0 - l3 + 3.0f * G4; + x4 = x0 - 1.0f + 4.0f * G4; /* Offsets for last corner in (x,y,z,w) coords */ + y4 = y0 - 1.0f + 4.0f * G4; + z4 = z0 - 1.0f + 4.0f * G4; + w4 = w0 - 1.0f + 4.0f * G4; + + /* Wrap the integer indices at 256, to avoid indexing perm[] out of bounds */ + ii = i % 256; + jj = j % 256; + kk = k % 256; + ll = l % 256; + + /* Calculate the contribution from the five corners */ + t0 = 0.6f - x0 * x0 - y0 * y0 - z0 * z0 - w0 * w0; + if (t0 < 0.0f) + n0 = 0.0f; + else { + t0 *= t0; + n0 = + t0 * t0 * grad4(perm[ii + perm[jj + perm[kk + perm[ll]]]], x0, y0, + z0, w0); + } + + t1 = 0.6f - x1 * x1 - y1 * y1 - z1 * z1 - w1 * w1; + if (t1 < 0.0f) + n1 = 0.0f; + else { + t1 *= t1; + n1 = + t1 * t1 * + grad4(perm[ii + i1 + perm[jj + j1 + perm[kk + k1 + perm[ll + l1]]]], + x1, y1, z1, w1); + } + + t2 = 0.6f - x2 * x2 - y2 * y2 - z2 * z2 - w2 * w2; + if (t2 < 0.0f) + n2 = 0.0f; + else { + t2 *= t2; + n2 = + t2 * t2 * + grad4(perm[ii + i2 + perm[jj + j2 + perm[kk + k2 + perm[ll + l2]]]], + x2, y2, z2, w2); + } + + t3 = 0.6f - x3 * x3 - y3 * y3 - z3 * z3 - w3 * w3; + if (t3 < 0.0f) + n3 = 0.0f; + else { + t3 *= t3; + n3 = + t3 * t3 * + grad4(perm[ii + i3 + perm[jj + j3 + perm[kk + k3 + perm[ll + l3]]]], + x3, y3, z3, w3); + } + + t4 = 0.6f - x4 * x4 - y4 * y4 - z4 * z4 - w4 * w4; + if (t4 < 0.0f) + n4 = 0.0f; + else { + t4 *= t4; + n4 = + t4 * t4 * + grad4(perm[ii + 1 + perm[jj + 1 + perm[kk + 1 + perm[ll + 1]]]], x4, + y4, z4, w4); + } + + /* Sum up and scale the result to cover the range [-1,1] */ + return 27.0f * (n0 + n1 + n2 + n3 + n4); /* TODO: The scale factor is preliminary! */ +} diff --git a/mesalib/src/mesa/program/prog_noise.h b/mesalib/src/mesa/program/prog_noise.h new file mode 100644 index 000000000..dd7986efc --- /dev/null +++ b/mesalib/src/mesa/program/prog_noise.h @@ -0,0 +1,36 @@ +/* + * Mesa 3-D graphics library + * Version: 6.5 + * + * Copyright (C) 2006 Brian Paul All Rights Reserved. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included + * in all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS + * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * BRIAN PAUL BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN + * AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN + * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. + */ + +#ifndef PROG_NOISE +#define PROG_NOISE + +#include "main/glheader.h" + +extern GLfloat _mesa_noise1(GLfloat); +extern GLfloat _mesa_noise2(GLfloat, GLfloat); +extern GLfloat _mesa_noise3(GLfloat, GLfloat, GLfloat); +extern GLfloat _mesa_noise4(GLfloat, GLfloat, GLfloat, GLfloat); + +#endif + diff --git a/mesalib/src/mesa/program/prog_optimize.c b/mesalib/src/mesa/program/prog_optimize.c new file mode 100644 index 000000000..0dc779073 --- /dev/null +++ b/mesalib/src/mesa/program/prog_optimize.c @@ -0,0 +1,1250 @@ +/* + * Mesa 3-D graphics library + * Version: 7.5 + * + * Copyright (C) 2009 VMware, Inc. All Rights Reserved. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included + * in all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS + * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * VMWARE BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN + * AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN + * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. + */ + + + +#include "main/glheader.h" +#include "main/context.h" +#include "main/macros.h" +#include "program.h" +#include "prog_instruction.h" +#include "prog_optimize.h" +#include "prog_print.h" + + +#define MAX_LOOP_NESTING 50 +/* MAX_PROGRAM_TEMPS is a low number (256), and we want to be able to + * register allocate many temporary values into that small number of + * temps. So allow large temporary indices coming into the register + * allocator. + */ +#define REG_ALLOCATE_MAX_PROGRAM_TEMPS ((1 << INST_INDEX_BITS) - 1) + +static GLboolean dbg = GL_FALSE; + +#define NO_MASK 0xf + +/** + * Returns the mask of channels (bitmask of WRITEMASK_X,Y,Z,W) which + * are read from the given src in this instruction, We also provide + * one optional masks which may mask other components in the dst + * register + */ +static GLuint +get_src_arg_mask(const struct prog_instruction *inst, + GLuint arg, GLuint dst_mask) +{ + GLuint read_mask, channel_mask; + GLuint comp; + + ASSERT(arg < _mesa_num_inst_src_regs(inst->Opcode)); + + /* Form the dst register, find the written channels */ + if (inst->CondUpdate) { + channel_mask = WRITEMASK_XYZW; + } + else { + switch (inst->Opcode) { + case OPCODE_MOV: + case OPCODE_MIN: + case OPCODE_MAX: + case OPCODE_ABS: + case OPCODE_ADD: + case OPCODE_MAD: + case OPCODE_MUL: + case OPCODE_SUB: + channel_mask = inst->DstReg.WriteMask & dst_mask; + break; + case OPCODE_RCP: + case OPCODE_SIN: + case OPCODE_COS: + case OPCODE_RSQ: + case OPCODE_POW: + case OPCODE_EX2: + case OPCODE_LOG: + channel_mask = WRITEMASK_X; + break; + case OPCODE_DP2: + channel_mask = WRITEMASK_XY; + break; + case OPCODE_DP3: + case OPCODE_XPD: + channel_mask = WRITEMASK_XYZ; + break; + default: + channel_mask = WRITEMASK_XYZW; + break; + } + } + + /* Now, given the src swizzle and the written channels, find which + * components are actually read + */ + read_mask = 0x0; + for (comp = 0; comp < 4; ++comp) { + const GLuint coord = GET_SWZ(inst->SrcReg[arg].Swizzle, comp); + ASSERT(coord < 4); + if (channel_mask & (1 << comp) && coord <= SWIZZLE_W) + read_mask |= 1 << coord; + } + + return read_mask; +} + + +/** + * For a MOV instruction, compute a write mask when src register also has + * a mask + */ +static GLuint +get_dst_mask_for_mov(const struct prog_instruction *mov, GLuint src_mask) +{ + const GLuint mask = mov->DstReg.WriteMask; + GLuint comp; + GLuint updated_mask = 0x0; + + ASSERT(mov->Opcode == OPCODE_MOV); + + for (comp = 0; comp < 4; ++comp) { + GLuint src_comp; + if ((mask & (1 << comp)) == 0) + continue; + src_comp = GET_SWZ(mov->SrcReg[0].Swizzle, comp); + if ((src_mask & (1 << src_comp)) == 0) + continue; + updated_mask |= 1 << comp; + } + + return updated_mask; +} + + +/** + * Ensure that the swizzle is regular. That is, all of the swizzle + * terms are SWIZZLE_X,Y,Z,W and not SWIZZLE_ZERO or SWIZZLE_ONE. + */ +static GLboolean +is_swizzle_regular(GLuint swz) +{ + return GET_SWZ(swz,0) <= SWIZZLE_W && + GET_SWZ(swz,1) <= SWIZZLE_W && + GET_SWZ(swz,2) <= SWIZZLE_W && + GET_SWZ(swz,3) <= SWIZZLE_W; +} + + +/** + * In 'prog' remove instruction[i] if removeFlags[i] == TRUE. + * \return number of instructions removed + */ +static GLuint +remove_instructions(struct gl_program *prog, const GLboolean *removeFlags) +{ + GLint i, removeEnd = 0, removeCount = 0; + GLuint totalRemoved = 0; + + /* go backward */ + for (i = prog->NumInstructions - 1; i >= 0; i--) { + if (removeFlags[i]) { + totalRemoved++; + if (removeCount == 0) { + /* begin a run of instructions to remove */ + removeEnd = i; + removeCount = 1; + } + else { + /* extend the run of instructions to remove */ + removeCount++; + } + } + else { + /* don't remove this instruction, but check if the preceeding + * instructions are to be removed. + */ + if (removeCount > 0) { + GLint removeStart = removeEnd - removeCount + 1; + _mesa_delete_instructions(prog, removeStart, removeCount); + removeStart = removeCount = 0; /* reset removal info */ + } + } + } + /* Finish removing if the first instruction was to be removed. */ + if (removeCount > 0) { + GLint removeStart = removeEnd - removeCount + 1; + _mesa_delete_instructions(prog, removeStart, removeCount); + } + return totalRemoved; +} + + +/** + * Remap register indexes according to map. + * \param prog the program to search/replace + * \param file the type of register file to search/replace + * \param map maps old register indexes to new indexes + */ +static void +replace_regs(struct gl_program *prog, gl_register_file file, const GLint map[]) +{ + GLuint i; + + for (i = 0; i < prog->NumInstructions; i++) { + struct prog_instruction *inst = prog->Instructions + i; + const GLuint numSrc = _mesa_num_inst_src_regs(inst->Opcode); + GLuint j; + for (j = 0; j < numSrc; j++) { + if (inst->SrcReg[j].File == file) { + GLuint index = inst->SrcReg[j].Index; + ASSERT(map[index] >= 0); + inst->SrcReg[j].Index = map[index]; + } + } + if (inst->DstReg.File == file) { + const GLuint index = inst->DstReg.Index; + ASSERT(map[index] >= 0); + inst->DstReg.Index = map[index]; + } + } +} + + +/** + * Remove dead instructions from the given program. + * This is very primitive for now. Basically look for temp registers + * that are written to but never read. Remove any instructions that + * write to such registers. Be careful with condition code setters. + */ +static GLboolean +_mesa_remove_dead_code_global(struct gl_program *prog) +{ + GLboolean tempRead[REG_ALLOCATE_MAX_PROGRAM_TEMPS][4]; + GLboolean *removeInst; /* per-instruction removal flag */ + GLuint i, rem = 0, comp; + + memset(tempRead, 0, sizeof(tempRead)); + + if (dbg) { + printf("Optimize: Begin dead code removal\n"); + /*_mesa_print_program(prog);*/ + } + + removeInst = (GLboolean *) + calloc(1, prog->NumInstructions * sizeof(GLboolean)); + + /* Determine which temps are read and written */ + for (i = 0; i < prog->NumInstructions; i++) { + const struct prog_instruction *inst = prog->Instructions + i; + const GLuint numSrc = _mesa_num_inst_src_regs(inst->Opcode); + GLuint j; + + /* check src regs */ + for (j = 0; j < numSrc; j++) { + if (inst->SrcReg[j].File == PROGRAM_TEMPORARY) { + const GLuint index = inst->SrcReg[j].Index; + GLuint read_mask; + ASSERT(index < REG_ALLOCATE_MAX_PROGRAM_TEMPS); + read_mask = get_src_arg_mask(inst, j, NO_MASK); + + if (inst->SrcReg[j].RelAddr) { + if (dbg) + printf("abort remove dead code (indirect temp)\n"); + goto done; + } + + for (comp = 0; comp < 4; comp++) { + const GLuint swz = GET_SWZ(inst->SrcReg[j].Swizzle, comp); + ASSERT(swz < 4); + if ((read_mask & (1 << swz)) == 0) + continue; + if (swz <= SWIZZLE_W) + tempRead[index][swz] = GL_TRUE; + } + } + } + + /* check dst reg */ + if (inst->DstReg.File == PROGRAM_TEMPORARY) { + const GLuint index = inst->DstReg.Index; + ASSERT(index < REG_ALLOCATE_MAX_PROGRAM_TEMPS); + + if (inst->DstReg.RelAddr) { + if (dbg) + printf("abort remove dead code (indirect temp)\n"); + goto done; + } + + if (inst->CondUpdate) { + /* If we're writing to this register and setting condition + * codes we cannot remove the instruction. Prevent removal + * by setting the 'read' flag. + */ + tempRead[index][0] = GL_TRUE; + tempRead[index][1] = GL_TRUE; + tempRead[index][2] = GL_TRUE; + tempRead[index][3] = GL_TRUE; + } + } + } + + /* find instructions that write to dead registers, flag for removal */ + for (i = 0; i < prog->NumInstructions; i++) { + struct prog_instruction *inst = prog->Instructions + i; + const GLuint numDst = _mesa_num_inst_dst_regs(inst->Opcode); + + if (numDst != 0 && inst->DstReg.File == PROGRAM_TEMPORARY) { + GLint chan, index = inst->DstReg.Index; + + for (chan = 0; chan < 4; chan++) { + if (!tempRead[index][chan] && + inst->DstReg.WriteMask & (1 << chan)) { + if (dbg) { + printf("Remove writemask on %u.%c\n", i, + chan == 3 ? 'w' : 'x' + chan); + } + inst->DstReg.WriteMask &= ~(1 << chan); + rem++; + } + } + + if (inst->DstReg.WriteMask == 0) { + /* If we cleared all writes, the instruction can be removed. */ + if (dbg) + printf("Remove instruction %u: \n", i); + removeInst[i] = GL_TRUE; + } + } + } + + /* now remove the instructions which aren't needed */ + rem = remove_instructions(prog, removeInst); + + if (dbg) { + printf("Optimize: End dead code removal.\n"); + printf(" %u channel writes removed\n", rem); + printf(" %u instructions removed\n", rem); + /*_mesa_print_program(prog);*/ + } + +done: + free(removeInst); + return rem != 0; +} + + +enum inst_use +{ + READ, + WRITE, + FLOW, + END +}; + + +/** + * Scan forward in program from 'start' for the next occurances of TEMP[index]. + * We look if an instruction reads the component given by the masks and if they + * are overwritten. + * Return READ, WRITE, FLOW or END to indicate the next usage or an indicator + * that we can't look further. + */ +static enum inst_use +find_next_use(const struct gl_program *prog, + GLuint start, + GLuint index, + GLuint mask) +{ + GLuint i; + + for (i = start; i < prog->NumInstructions; i++) { + const struct prog_instruction *inst = prog->Instructions + i; + switch (inst->Opcode) { + case OPCODE_BGNLOOP: + case OPCODE_BGNSUB: + case OPCODE_BRA: + case OPCODE_CAL: + case OPCODE_CONT: + case OPCODE_IF: + case OPCODE_ELSE: + case OPCODE_ENDIF: + case OPCODE_ENDLOOP: + case OPCODE_ENDSUB: + case OPCODE_RET: + return FLOW; + case OPCODE_END: + return END; + default: + { + const GLuint numSrc = _mesa_num_inst_src_regs(inst->Opcode); + GLuint j; + for (j = 0; j < numSrc; j++) { + if (inst->SrcReg[j].RelAddr || + (inst->SrcReg[j].File == PROGRAM_TEMPORARY && + inst->SrcReg[j].Index == index && + (get_src_arg_mask(inst,j,NO_MASK) & mask))) + return READ; + } + if (_mesa_num_inst_dst_regs(inst->Opcode) == 1 && + inst->DstReg.File == PROGRAM_TEMPORARY && + inst->DstReg.Index == index) { + mask &= ~inst->DstReg.WriteMask; + if (mask == 0) + return WRITE; + } + } + } + } + return END; +} + + +/** + * Is the given instruction opcode a flow-control opcode? + * XXX maybe move this into prog_instruction.[ch] + */ +static GLboolean +_mesa_is_flow_control_opcode(enum prog_opcode opcode) +{ + switch (opcode) { + case OPCODE_BGNLOOP: + case OPCODE_BGNSUB: + case OPCODE_BRA: + case OPCODE_CAL: + case OPCODE_CONT: + case OPCODE_IF: + case OPCODE_ELSE: + case OPCODE_END: + case OPCODE_ENDIF: + case OPCODE_ENDLOOP: + case OPCODE_ENDSUB: + case OPCODE_RET: + return GL_TRUE; + default: + return GL_FALSE; + } +} + + +/** + * Test if the given instruction is a simple MOV (no conditional updating, + * not relative addressing, no negation/abs, etc). + */ +static GLboolean +can_downward_mov_be_modifed(const struct prog_instruction *mov) +{ + return + mov->Opcode == OPCODE_MOV && + mov->CondUpdate == GL_FALSE && + mov->SrcReg[0].RelAddr == 0 && + mov->SrcReg[0].Negate == 0 && + mov->SrcReg[0].Abs == 0 && + mov->SrcReg[0].HasIndex2 == 0 && + mov->SrcReg[0].RelAddr2 == 0 && + mov->DstReg.RelAddr == 0 && + mov->DstReg.CondMask == COND_TR && + mov->SaturateMode == SATURATE_OFF; +} + + +static GLboolean +can_upward_mov_be_modifed(const struct prog_instruction *mov) +{ + return + can_downward_mov_be_modifed(mov) && + mov->DstReg.File == PROGRAM_TEMPORARY; +} + + +/** + * Try to remove use of extraneous MOV instructions, to free them up for dead + * code removal. + */ +static void +_mesa_remove_extra_move_use(struct gl_program *prog) +{ + GLuint i, j; + + if (dbg) { + printf("Optimize: Begin remove extra move use\n"); + _mesa_print_program(prog); + } + + /* + * Look for sequences such as this: + * MOV tmpX, arg0; + * ... + * FOO tmpY, tmpX, arg1; + * and convert into: + * MOV tmpX, arg0; + * ... + * FOO tmpY, arg0, arg1; + */ + + for (i = 0; i + 1 < prog->NumInstructions; i++) { + const struct prog_instruction *mov = prog->Instructions + i; + GLuint dst_mask, src_mask; + if (can_upward_mov_be_modifed(mov) == GL_FALSE) + continue; + + /* Scanning the code, we maintain the components which are still active in + * these two masks + */ + dst_mask = mov->DstReg.WriteMask; + src_mask = get_src_arg_mask(mov, 0, NO_MASK); + + /* Walk through remaining instructions until the or src reg gets + * rewritten or we get into some flow-control, eliminating the use of + * this MOV. + */ + for (j = i + 1; j < prog->NumInstructions; j++) { + struct prog_instruction *inst2 = prog->Instructions + j; + GLuint arg; + + if (_mesa_is_flow_control_opcode(inst2->Opcode)) + break; + + /* First rewrite this instruction's args if appropriate. */ + for (arg = 0; arg < _mesa_num_inst_src_regs(inst2->Opcode); arg++) { + GLuint comp, read_mask; + + if (inst2->SrcReg[arg].File != mov->DstReg.File || + inst2->SrcReg[arg].Index != mov->DstReg.Index || + inst2->SrcReg[arg].RelAddr || + inst2->SrcReg[arg].Abs) + continue; + read_mask = get_src_arg_mask(inst2, arg, NO_MASK); + + /* Adjust the swizzles of inst2 to point at MOV's source if ALL the + * components read still come from the mov instructions + */ + if (is_swizzle_regular(inst2->SrcReg[arg].Swizzle) && + (read_mask & dst_mask) == read_mask) { + for (comp = 0; comp < 4; comp++) { + const GLuint inst2_swz = + GET_SWZ(inst2->SrcReg[arg].Swizzle, comp); + const GLuint s = GET_SWZ(mov->SrcReg[0].Swizzle, inst2_swz); + inst2->SrcReg[arg].Swizzle &= ~(7 << (3 * comp)); + inst2->SrcReg[arg].Swizzle |= s << (3 * comp); + inst2->SrcReg[arg].Negate ^= (((mov->SrcReg[0].Negate >> + inst2_swz) & 0x1) << comp); + } + inst2->SrcReg[arg].File = mov->SrcReg[0].File; + inst2->SrcReg[arg].Index = mov->SrcReg[0].Index; + } + } + + /* The source of MOV is written. This potentially deactivates some + * components from the src and dst of the MOV instruction + */ + if (inst2->DstReg.File == mov->DstReg.File && + (inst2->DstReg.RelAddr || + inst2->DstReg.Index == mov->DstReg.Index)) { + dst_mask &= ~inst2->DstReg.WriteMask; + src_mask = get_src_arg_mask(mov, 0, dst_mask); + } + + /* Idem when the destination of mov is written */ + if (inst2->DstReg.File == mov->SrcReg[0].File && + (inst2->DstReg.RelAddr || + inst2->DstReg.Index == mov->SrcReg[0].Index)) { + src_mask &= ~inst2->DstReg.WriteMask; + dst_mask &= get_dst_mask_for_mov(mov, src_mask); + } + if (dst_mask == 0) + break; + } + } + + if (dbg) { + printf("Optimize: End remove extra move use.\n"); + /*_mesa_print_program(prog);*/ + } +} + + +/** + * Complements dead_code_global. Try to remove code in block of code by + * carefully monitoring the swizzles. Both functions should be merged into one + * with a proper control flow graph + */ +static GLboolean +_mesa_remove_dead_code_local(struct gl_program *prog) +{ + GLboolean *removeInst; + GLuint i, arg, rem = 0; + + removeInst = (GLboolean *) + calloc(1, prog->NumInstructions * sizeof(GLboolean)); + + for (i = 0; i < prog->NumInstructions; i++) { + const struct prog_instruction *inst = prog->Instructions + i; + const GLuint index = inst->DstReg.Index; + const GLuint mask = inst->DstReg.WriteMask; + enum inst_use use; + + /* We must deactivate the pass as soon as some indirection is used */ + if (inst->DstReg.RelAddr) + goto done; + for (arg = 0; arg < _mesa_num_inst_src_regs(inst->Opcode); arg++) + if (inst->SrcReg[arg].RelAddr) + goto done; + + if (_mesa_is_flow_control_opcode(inst->Opcode) || + _mesa_num_inst_dst_regs(inst->Opcode) == 0 || + inst->DstReg.File != PROGRAM_TEMPORARY || + inst->DstReg.RelAddr) + continue; + + use = find_next_use(prog, i+1, index, mask); + if (use == WRITE || use == END) + removeInst[i] = GL_TRUE; + } + + rem = remove_instructions(prog, removeInst); + +done: + free(removeInst); + return rem != 0; +} + + +/** + * Try to inject the destination of mov as the destination of inst and recompute + * the swizzles operators for the sources of inst if required. Return GL_TRUE + * of the substitution was possible, GL_FALSE otherwise + */ +static GLboolean +_mesa_merge_mov_into_inst(struct prog_instruction *inst, + const struct prog_instruction *mov) +{ + /* Indirection table which associates destination and source components for + * the mov instruction + */ + const GLuint mask = get_src_arg_mask(mov, 0, NO_MASK); + + /* Some components are not written by inst. We cannot remove the mov */ + if (mask != (inst->DstReg.WriteMask & mask)) + return GL_FALSE; + + /* Depending on the instruction, we may need to recompute the swizzles. + * Also, some other instructions (like TEX) are not linear. We will only + * consider completely active sources and destinations + */ + switch (inst->Opcode) { + + /* Carstesian instructions: we compute the swizzle */ + case OPCODE_MOV: + case OPCODE_MIN: + case OPCODE_MAX: + case OPCODE_ABS: + case OPCODE_ADD: + case OPCODE_MAD: + case OPCODE_MUL: + case OPCODE_SUB: + { + GLuint dst_to_src_comp[4] = {0,0,0,0}; + GLuint dst_comp, arg; + for (dst_comp = 0; dst_comp < 4; ++dst_comp) { + if (mov->DstReg.WriteMask & (1 << dst_comp)) { + const GLuint src_comp = GET_SWZ(mov->SrcReg[0].Swizzle, dst_comp); + ASSERT(src_comp < 4); + dst_to_src_comp[dst_comp] = src_comp; + } + } + + /* Patch each source of the instruction */ + for (arg = 0; arg < _mesa_num_inst_src_regs(inst->Opcode); arg++) { + const GLuint arg_swz = inst->SrcReg[arg].Swizzle; + inst->SrcReg[arg].Swizzle = 0; + + /* Reset each active component of the swizzle */ + for (dst_comp = 0; dst_comp < 4; ++dst_comp) { + GLuint src_comp, arg_comp; + if ((mov->DstReg.WriteMask & (1 << dst_comp)) == 0) + continue; + src_comp = dst_to_src_comp[dst_comp]; + ASSERT(src_comp < 4); + arg_comp = GET_SWZ(arg_swz, src_comp); + ASSERT(arg_comp < 4); + inst->SrcReg[arg].Swizzle |= arg_comp << (3*dst_comp); + } + } + inst->DstReg = mov->DstReg; + return GL_TRUE; + } + + /* Dot products and scalar instructions: we only change the destination */ + case OPCODE_RCP: + case OPCODE_SIN: + case OPCODE_COS: + case OPCODE_RSQ: + case OPCODE_POW: + case OPCODE_EX2: + case OPCODE_LOG: + case OPCODE_DP2: + case OPCODE_DP3: + case OPCODE_DP4: + inst->DstReg = mov->DstReg; + return GL_TRUE; + + /* All other instructions require fully active components with no swizzle */ + default: + if (mov->SrcReg[0].Swizzle != SWIZZLE_XYZW || + inst->DstReg.WriteMask != WRITEMASK_XYZW) + return GL_FALSE; + inst->DstReg = mov->DstReg; + return GL_TRUE; + } +} + + +/** + * Try to remove extraneous MOV instructions from the given program. + */ +static GLboolean +_mesa_remove_extra_moves(struct gl_program *prog) +{ + GLboolean *removeInst; /* per-instruction removal flag */ + GLuint i, rem = 0, nesting = 0; + + if (dbg) { + printf("Optimize: Begin remove extra moves\n"); + _mesa_print_program(prog); + } + + removeInst = (GLboolean *) + calloc(1, prog->NumInstructions * sizeof(GLboolean)); + + /* + * Look for sequences such as this: + * FOO tmpX, arg0, arg1; + * MOV tmpY, tmpX; + * and convert into: + * FOO tmpY, arg0, arg1; + */ + + for (i = 0; i < prog->NumInstructions; i++) { + const struct prog_instruction *mov = prog->Instructions + i; + + switch (mov->Opcode) { + case OPCODE_BGNLOOP: + case OPCODE_BGNSUB: + case OPCODE_IF: + nesting++; + break; + case OPCODE_ENDLOOP: + case OPCODE_ENDSUB: + case OPCODE_ENDIF: + nesting--; + break; + case OPCODE_MOV: + if (i > 0 && + can_downward_mov_be_modifed(mov) && + mov->SrcReg[0].File == PROGRAM_TEMPORARY && + nesting == 0) + { + + /* see if this MOV can be removed */ + const GLuint id = mov->SrcReg[0].Index; + struct prog_instruction *prevInst; + GLuint prevI; + + /* get pointer to previous instruction */ + prevI = i - 1; + while (prevI > 0 && removeInst[prevI]) + prevI--; + prevInst = prog->Instructions + prevI; + + if (prevInst->DstReg.File == PROGRAM_TEMPORARY && + prevInst->DstReg.Index == id && + prevInst->DstReg.RelAddr == 0 && + prevInst->DstReg.CondSrc == 0 && + prevInst->DstReg.CondMask == COND_TR) { + + const GLuint dst_mask = prevInst->DstReg.WriteMask; + enum inst_use next_use = find_next_use(prog, i+1, id, dst_mask); + + if (next_use == WRITE || next_use == END) { + /* OK, we can safely remove this MOV instruction. + * Transform: + * prevI: FOO tempIndex, x, y; + * i: MOV z, tempIndex; + * Into: + * prevI: FOO z, x, y; + */ + if (_mesa_merge_mov_into_inst(prevInst, mov)) { + removeInst[i] = GL_TRUE; + if (dbg) { + printf("Remove MOV at %u\n", i); + printf("new prev inst %u: ", prevI); + _mesa_print_instruction(prevInst); + } + } + } + } + } + break; + default: + ; /* nothing */ + } + } + + /* now remove the instructions which aren't needed */ + rem = remove_instructions(prog, removeInst); + + free(removeInst); + + if (dbg) { + printf("Optimize: End remove extra moves. %u instructions removed\n", rem); + /*_mesa_print_program(prog);*/ + } + + return rem != 0; +} + + +/** A live register interval */ +struct interval +{ + GLuint Reg; /** The temporary register index */ + GLuint Start, End; /** Start/end instruction numbers */ +}; + + +/** A list of register intervals */ +struct interval_list +{ + GLuint Num; + struct interval Intervals[REG_ALLOCATE_MAX_PROGRAM_TEMPS]; +}; + + +static void +append_interval(struct interval_list *list, const struct interval *inv) +{ + list->Intervals[list->Num++] = *inv; +} + + +/** Insert interval inv into list, sorted by interval end */ +static void +insert_interval_by_end(struct interval_list *list, const struct interval *inv) +{ + /* XXX we could do a binary search insertion here since list is sorted */ + GLint i = list->Num - 1; + while (i >= 0 && list->Intervals[i].End > inv->End) { + list->Intervals[i + 1] = list->Intervals[i]; + i--; + } + list->Intervals[i + 1] = *inv; + list->Num++; + +#ifdef DEBUG + { + GLuint i; + for (i = 0; i + 1 < list->Num; i++) { + ASSERT(list->Intervals[i].End <= list->Intervals[i + 1].End); + } + } +#endif +} + + +/** Remove the given interval from the interval list */ +static void +remove_interval(struct interval_list *list, const struct interval *inv) +{ + /* XXX we could binary search since list is sorted */ + GLuint k; + for (k = 0; k < list->Num; k++) { + if (list->Intervals[k].Reg == inv->Reg) { + /* found, remove it */ + ASSERT(list->Intervals[k].Start == inv->Start); + ASSERT(list->Intervals[k].End == inv->End); + while (k < list->Num - 1) { + list->Intervals[k] = list->Intervals[k + 1]; + k++; + } + list->Num--; + return; + } + } +} + + +/** called by qsort() */ +static int +compare_start(const void *a, const void *b) +{ + const struct interval *ia = (const struct interval *) a; + const struct interval *ib = (const struct interval *) b; + if (ia->Start < ib->Start) + return -1; + else if (ia->Start > ib->Start) + return +1; + else + return 0; +} + + +/** sort the interval list according to interval starts */ +static void +sort_interval_list_by_start(struct interval_list *list) +{ + qsort(list->Intervals, list->Num, sizeof(struct interval), compare_start); +#ifdef DEBUG + { + GLuint i; + for (i = 0; i + 1 < list->Num; i++) { + ASSERT(list->Intervals[i].Start <= list->Intervals[i + 1].Start); + } + } +#endif +} + +struct loop_info +{ + GLuint Start, End; /**< Start, end instructions of loop */ +}; + +/** + * Update the intermediate interval info for register 'index' and + * instruction 'ic'. + */ +static void +update_interval(GLint intBegin[], GLint intEnd[], + struct loop_info *loopStack, GLuint loopStackDepth, + GLuint index, GLuint ic) +{ + int i; + + /* If the register is used in a loop, extend its lifetime through the end + * of the outermost loop that doesn't contain its definition. + */ + for (i = 0; i < loopStackDepth; i++) { + if (intBegin[index] < loopStack[i].Start) { + ic = loopStack[i].End; + break; + } + } + + ASSERT(index < REG_ALLOCATE_MAX_PROGRAM_TEMPS); + if (intBegin[index] == -1) { + ASSERT(intEnd[index] == -1); + intBegin[index] = intEnd[index] = ic; + } + else { + intEnd[index] = ic; + } +} + + +/** + * Find first/last instruction that references each temporary register. + */ +GLboolean +_mesa_find_temp_intervals(const struct prog_instruction *instructions, + GLuint numInstructions, + GLint intBegin[REG_ALLOCATE_MAX_PROGRAM_TEMPS], + GLint intEnd[REG_ALLOCATE_MAX_PROGRAM_TEMPS]) +{ + struct loop_info loopStack[MAX_LOOP_NESTING]; + GLuint loopStackDepth = 0; + GLuint i; + + for (i = 0; i < REG_ALLOCATE_MAX_PROGRAM_TEMPS; i++){ + intBegin[i] = intEnd[i] = -1; + } + + /* Scan instructions looking for temporary registers */ + for (i = 0; i < numInstructions; i++) { + const struct prog_instruction *inst = instructions + i; + if (inst->Opcode == OPCODE_BGNLOOP) { + loopStack[loopStackDepth].Start = i; + loopStack[loopStackDepth].End = inst->BranchTarget; + loopStackDepth++; + } + else if (inst->Opcode == OPCODE_ENDLOOP) { + loopStackDepth--; + } + else if (inst->Opcode == OPCODE_CAL) { + return GL_FALSE; + } + else { + const GLuint numSrc = 3;/*_mesa_num_inst_src_regs(inst->Opcode);*/ + GLuint j; + for (j = 0; j < numSrc; j++) { + if (inst->SrcReg[j].File == PROGRAM_TEMPORARY) { + const GLuint index = inst->SrcReg[j].Index; + if (inst->SrcReg[j].RelAddr) + return GL_FALSE; + update_interval(intBegin, intEnd, loopStack, loopStackDepth, + index, i); + } + } + if (inst->DstReg.File == PROGRAM_TEMPORARY) { + const GLuint index = inst->DstReg.Index; + if (inst->DstReg.RelAddr) + return GL_FALSE; + update_interval(intBegin, intEnd, loopStack, loopStackDepth, + index, i); + } + } + } + + return GL_TRUE; +} + + +/** + * Find the live intervals for each temporary register in the program. + * For register R, the interval [A,B] indicates that R is referenced + * from instruction A through instruction B. + * Special consideration is needed for loops and subroutines. + * \return GL_TRUE if success, GL_FALSE if we cannot proceed for some reason + */ +static GLboolean +find_live_intervals(struct gl_program *prog, + struct interval_list *liveIntervals) +{ + GLint intBegin[REG_ALLOCATE_MAX_PROGRAM_TEMPS]; + GLint intEnd[REG_ALLOCATE_MAX_PROGRAM_TEMPS]; + GLuint i; + + /* + * Note: we'll return GL_FALSE below if we find relative indexing + * into the TEMP register file. We can't handle that yet. + * We also give up on subroutines for now. + */ + + if (dbg) { + printf("Optimize: Begin find intervals\n"); + } + + /* build intermediate arrays */ + if (!_mesa_find_temp_intervals(prog->Instructions, prog->NumInstructions, + intBegin, intEnd)) + return GL_FALSE; + + /* Build live intervals list from intermediate arrays */ + liveIntervals->Num = 0; + for (i = 0; i < REG_ALLOCATE_MAX_PROGRAM_TEMPS; i++) { + if (intBegin[i] >= 0) { + struct interval inv; + inv.Reg = i; + inv.Start = intBegin[i]; + inv.End = intEnd[i]; + append_interval(liveIntervals, &inv); + } + } + + /* Sort the list according to interval starts */ + sort_interval_list_by_start(liveIntervals); + + if (dbg) { + /* print interval info */ + for (i = 0; i < liveIntervals->Num; i++) { + const struct interval *inv = liveIntervals->Intervals + i; + printf("Reg[%d] live [%d, %d]:", + inv->Reg, inv->Start, inv->End); + if (1) { + GLuint j; + for (j = 0; j < inv->Start; j++) + printf(" "); + for (j = inv->Start; j <= inv->End; j++) + printf("x"); + } + printf("\n"); + } + } + + return GL_TRUE; +} + + +/** Scan the array of used register flags to find free entry */ +static GLint +alloc_register(GLboolean usedRegs[REG_ALLOCATE_MAX_PROGRAM_TEMPS]) +{ + GLuint k; + for (k = 0; k < REG_ALLOCATE_MAX_PROGRAM_TEMPS; k++) { + if (!usedRegs[k]) { + usedRegs[k] = GL_TRUE; + return k; + } + } + return -1; +} + + +/** + * This function implements "Linear Scan Register Allocation" to reduce + * the number of temporary registers used by the program. + * + * We compute the "live interval" for all temporary registers then + * examine the overlap of the intervals to allocate new registers. + * Basically, if two intervals do not overlap, they can use the same register. + */ +static void +_mesa_reallocate_registers(struct gl_program *prog) +{ + struct interval_list liveIntervals; + GLint registerMap[REG_ALLOCATE_MAX_PROGRAM_TEMPS]; + GLboolean usedRegs[REG_ALLOCATE_MAX_PROGRAM_TEMPS]; + GLuint i; + GLint maxTemp = -1; + + if (dbg) { + printf("Optimize: Begin live-interval register reallocation\n"); + _mesa_print_program(prog); + } + + for (i = 0; i < REG_ALLOCATE_MAX_PROGRAM_TEMPS; i++){ + registerMap[i] = -1; + usedRegs[i] = GL_FALSE; + } + + if (!find_live_intervals(prog, &liveIntervals)) { + if (dbg) + printf("Aborting register reallocation\n"); + return; + } + + { + struct interval_list activeIntervals; + activeIntervals.Num = 0; + + /* loop over live intervals, allocating a new register for each */ + for (i = 0; i < liveIntervals.Num; i++) { + const struct interval *live = liveIntervals.Intervals + i; + + if (dbg) + printf("Consider register %u\n", live->Reg); + + /* Expire old intervals. Intervals which have ended with respect + * to the live interval can have their remapped registers freed. + */ + { + GLint j; + for (j = 0; j < (GLint) activeIntervals.Num; j++) { + const struct interval *inv = activeIntervals.Intervals + j; + if (inv->End >= live->Start) { + /* Stop now. Since the activeInterval list is sorted + * we know we don't have to go further. + */ + break; + } + else { + /* Interval 'inv' has expired */ + const GLint regNew = registerMap[inv->Reg]; + ASSERT(regNew >= 0); + + if (dbg) + printf(" expire interval for reg %u\n", inv->Reg); + + /* remove interval j from active list */ + remove_interval(&activeIntervals, inv); + j--; /* counter-act j++ in for-loop above */ + + /* return register regNew to the free pool */ + if (dbg) + printf(" free reg %d\n", regNew); + ASSERT(usedRegs[regNew] == GL_TRUE); + usedRegs[regNew] = GL_FALSE; + } + } + } + + /* find a free register for this live interval */ + { + const GLint k = alloc_register(usedRegs); + if (k < 0) { + /* out of registers, give up */ + return; + } + registerMap[live->Reg] = k; + maxTemp = MAX2(maxTemp, k); + if (dbg) + printf(" remap register %u -> %d\n", live->Reg, k); + } + + /* Insert this live interval into the active list which is sorted + * by increasing end points. + */ + insert_interval_by_end(&activeIntervals, live); + } + } + + if (maxTemp + 1 < (GLint) liveIntervals.Num) { + /* OK, we've reduced the number of registers needed. + * Scan the program and replace all the old temporary register + * indexes with the new indexes. + */ + replace_regs(prog, PROGRAM_TEMPORARY, registerMap); + + prog->NumTemporaries = maxTemp + 1; + } + + if (dbg) { + printf("Optimize: End live-interval register reallocation\n"); + printf("Num temp regs before: %u after: %u\n", + liveIntervals.Num, maxTemp + 1); + _mesa_print_program(prog); + } +} + + +#if 0 +static void +print_it(GLcontext *ctx, struct gl_program *program, const char *txt) { + fprintf(stderr, "%s (%u inst):\n", txt, program->NumInstructions); + _mesa_print_program(program); + _mesa_print_program_parameters(ctx, program); + fprintf(stderr, "\n\n"); +} +#endif + + +/** + * Apply optimizations to the given program to eliminate unnecessary + * instructions, temp regs, etc. + */ +void +_mesa_optimize_program(GLcontext *ctx, struct gl_program *program) +{ + GLboolean any_change; + + /* Stop when no modifications were output */ + do { + any_change = GL_FALSE; + _mesa_remove_extra_move_use(program); + if (_mesa_remove_dead_code_global(program)) + any_change = GL_TRUE; + if (_mesa_remove_extra_moves(program)) + any_change = GL_TRUE; + if (_mesa_remove_dead_code_local(program)) + any_change = GL_TRUE; + _mesa_reallocate_registers(program); + } while (any_change); +} + diff --git a/mesalib/src/mesa/program/prog_optimize.h b/mesalib/src/mesa/program/prog_optimize.h new file mode 100644 index 000000000..06cd9cb2c --- /dev/null +++ b/mesalib/src/mesa/program/prog_optimize.h @@ -0,0 +1,46 @@ +/* + * Mesa 3-D graphics library + * Version: 7.5 + * + * Copyright (C) 2009 VMware, Inc. All Rights Reserved. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included + * in all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS + * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * VMWARE BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN + * AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN + * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. + */ + +#ifndef PROG_OPT_H +#define PROG_OPT_H + + +#include "main/config.h" +#include "main/mtypes.h" + + +struct gl_program; +struct prog_instruction; + + +extern GLboolean +_mesa_find_temp_intervals(const struct prog_instruction *instructions, + GLuint numInstructions, + GLint intBegin[MAX_PROGRAM_TEMPS], + GLint intEnd[MAX_PROGRAM_TEMPS]); + +extern void +_mesa_optimize_program(GLcontext *ctx, struct gl_program *program); + +#endif diff --git a/mesalib/src/mesa/program/prog_parameter.c b/mesalib/src/mesa/program/prog_parameter.c new file mode 100644 index 000000000..6bf8a081b --- /dev/null +++ b/mesalib/src/mesa/program/prog_parameter.c @@ -0,0 +1,665 @@ +/* + * Mesa 3-D graphics library + * Version: 7.3 + * + * Copyright (C) 1999-2008 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. + */ + +/** + * \file prog_parameter.c + * Program parameter lists and functions. + * \author Brian Paul + */ + + +#include "main/glheader.h" +#include "main/imports.h" +#include "main/macros.h" +#include "prog_instruction.h" +#include "prog_parameter.h" +#include "prog_statevars.h" + + +struct gl_program_parameter_list * +_mesa_new_parameter_list(void) +{ + return CALLOC_STRUCT(gl_program_parameter_list); +} + + +struct gl_program_parameter_list * +_mesa_new_parameter_list_sized(unsigned size) +{ + struct gl_program_parameter_list *p = _mesa_new_parameter_list(); + + if ((p != NULL) && (size != 0)) { + p->Size = size; + + /* alloc arrays */ + p->Parameters = (struct gl_program_parameter *) + calloc(1, size * sizeof(struct gl_program_parameter)); + + p->ParameterValues = (GLfloat (*)[4]) + _mesa_align_malloc(size * 4 *sizeof(GLfloat), 16); + + + if ((p->Parameters == NULL) || (p->ParameterValues == NULL)) { + free(p->Parameters); + _mesa_align_free(p->ParameterValues); + free(p); + p = NULL; + } + } + + return p; +} + + +/** + * Free a parameter list and all its parameters + */ +void +_mesa_free_parameter_list(struct gl_program_parameter_list *paramList) +{ + GLuint i; + for (i = 0; i < paramList->NumParameters; i++) { + if (paramList->Parameters[i].Name) + free((void *) paramList->Parameters[i].Name); + } + free(paramList->Parameters); + if (paramList->ParameterValues) + _mesa_align_free(paramList->ParameterValues); + free(paramList); +} + + +/** + * Add a new parameter to a parameter list. + * Note that parameter values are usually 4-element GLfloat vectors. + * When size > 4 we'll allocate a sequential block of parameters to + * store all the values (in blocks of 4). + * + * \param paramList the list to add the parameter to + * \param type type of parameter, such as + * \param name the parameter name, will be duplicated/copied! + * \param size number of elements in 'values' vector (1..4, or more) + * \param datatype GL_FLOAT, GL_FLOAT_VECx, GL_INT, GL_INT_VECx or GL_NONE. + * \param values initial parameter value, up to 4 GLfloats, or NULL + * \param state state indexes, or NULL + * \return index of new parameter in the list, or -1 if error (out of mem) + */ +GLint +_mesa_add_parameter(struct gl_program_parameter_list *paramList, + gl_register_file type, const char *name, + GLuint size, GLenum datatype, const GLfloat *values, + const gl_state_index state[STATE_LENGTH], + GLbitfield flags) +{ + const GLuint oldNum = paramList->NumParameters; + const GLuint sz4 = (size + 3) / 4; /* no. of new param slots needed */ + + assert(size > 0); + + if (oldNum + sz4 > paramList->Size) { + /* Need to grow the parameter list array (alloc some extra) */ + paramList->Size = paramList->Size + 4 * sz4; + + /* realloc arrays */ + paramList->Parameters = (struct gl_program_parameter *) + _mesa_realloc(paramList->Parameters, + oldNum * sizeof(struct gl_program_parameter), + paramList->Size * sizeof(struct gl_program_parameter)); + + paramList->ParameterValues = (GLfloat (*)[4]) + _mesa_align_realloc(paramList->ParameterValues, /* old buf */ + oldNum * 4 * sizeof(GLfloat), /* old size */ + paramList->Size * 4 *sizeof(GLfloat), /* new sz */ + 16); + } + + if (!paramList->Parameters || + !paramList->ParameterValues) { + /* out of memory */ + paramList->NumParameters = 0; + paramList->Size = 0; + return -1; + } + else { + GLuint i; + + paramList->NumParameters = oldNum + sz4; + + memset(¶mList->Parameters[oldNum], 0, + sz4 * sizeof(struct gl_program_parameter)); + + for (i = 0; i < sz4; i++) { + struct gl_program_parameter *p = paramList->Parameters + oldNum + i; + p->Name = name ? _mesa_strdup(name) : NULL; + p->Type = type; + p->Size = size; + p->DataType = datatype; + p->Flags = flags; + if (values) { + COPY_4V(paramList->ParameterValues[oldNum + i], values); + values += 4; + p->Initialized = GL_TRUE; + } + else { + /* silence valgrind */ + ASSIGN_4V(paramList->ParameterValues[oldNum + i], 0, 0, 0, 0); + } + size -= 4; + } + + if (state) { + for (i = 0; i < STATE_LENGTH; i++) + paramList->Parameters[oldNum].StateIndexes[i] = state[i]; + } + + return (GLint) oldNum; + } +} + + +/** + * Add a new named program parameter (Ex: NV_fragment_program DEFINE statement) + * \return index of the new entry in the parameter list + */ +GLint +_mesa_add_named_parameter(struct gl_program_parameter_list *paramList, + const char *name, const GLfloat values[4]) +{ + return _mesa_add_parameter(paramList, PROGRAM_NAMED_PARAM, name, + 4, GL_NONE, values, NULL, 0x0); + +} + + +/** + * Add a new named constant to the parameter list. + * This will be used when the program contains something like this: + * PARAM myVals = { 0, 1, 2, 3 }; + * + * \param paramList the parameter list + * \param name the name for the constant + * \param values four float values + * \return index/position of the new parameter in the parameter list + */ +GLint +_mesa_add_named_constant(struct gl_program_parameter_list *paramList, + const char *name, const GLfloat values[4], + GLuint size) +{ + /* first check if this is a duplicate constant */ + GLint pos; + for (pos = 0; pos < (GLint)paramList->NumParameters; pos++) { + const GLfloat *pvals = paramList->ParameterValues[pos]; + if (pvals[0] == values[0] && + pvals[1] == values[1] && + pvals[2] == values[2] && + pvals[3] == values[3] && + strcmp(paramList->Parameters[pos].Name, name) == 0) { + /* Same name and value is already in the param list - reuse it */ + return pos; + } + } + /* not found, add new parameter */ + return _mesa_add_parameter(paramList, PROGRAM_CONSTANT, name, + size, GL_NONE, values, NULL, 0x0); +} + + +/** + * Add a new unnamed constant to the parameter list. This will be used + * when a fragment/vertex program contains something like this: + * MOV r, { 0, 1, 2, 3 }; + * If swizzleOut is non-null we'll search the parameter list for an + * existing instance of the constant which matches with a swizzle. + * + * \param paramList the parameter list + * \param values four float values + * \param swizzleOut returns swizzle mask for accessing the constant + * \return index/position of the new parameter in the parameter list. + */ +GLint +_mesa_add_unnamed_constant(struct gl_program_parameter_list *paramList, + const GLfloat values[4], GLuint size, + GLuint *swizzleOut) +{ + GLint pos; + ASSERT(size >= 1); + ASSERT(size <= 4); + + if (swizzleOut && + _mesa_lookup_parameter_constant(paramList, values, + size, &pos, swizzleOut)) { + return pos; + } + + /* Look for empty space in an already unnamed constant parameter + * to add this constant. This will only work for single-element + * constants because we rely on smearing (i.e. .yyyy or .zzzz). + */ + if (size == 1 && swizzleOut) { + for (pos = 0; pos < (GLint) paramList->NumParameters; pos++) { + struct gl_program_parameter *p = paramList->Parameters + pos; + if (p->Type == PROGRAM_CONSTANT && p->Size + size <= 4) { + /* ok, found room */ + GLfloat *pVal = paramList->ParameterValues[pos]; + GLuint swz = p->Size; /* 1, 2 or 3 for Y, Z, W */ + pVal[p->Size] = values[0]; + p->Size++; + *swizzleOut = MAKE_SWIZZLE4(swz, swz, swz, swz); + return pos; + } + } + } + + /* add a new parameter to store this constant */ + pos = _mesa_add_parameter(paramList, PROGRAM_CONSTANT, NULL, + size, GL_NONE, values, NULL, 0x0); + if (pos >= 0 && swizzleOut) { + if (size == 1) + *swizzleOut = SWIZZLE_XXXX; + else + *swizzleOut = SWIZZLE_NOOP; + } + return pos; +} + +/** + * Add parameter representing a varying variable. + */ +GLint +_mesa_add_varying(struct gl_program_parameter_list *paramList, + const char *name, GLuint size, GLenum datatype, + GLbitfield flags) +{ + GLint i = _mesa_lookup_parameter_index(paramList, -1, name); + if (i >= 0 && paramList->Parameters[i].Type == PROGRAM_VARYING) { + /* already in list */ + return i; + } + else { + /*assert(size == 4);*/ + i = _mesa_add_parameter(paramList, PROGRAM_VARYING, name, + size, datatype, NULL, NULL, flags); + return i; + } +} + + +/** + * Add parameter representing a vertex program attribute. + * \param size size of attribute (in floats), may be -1 if unknown + * \param attrib the attribute index, or -1 if unknown + */ +GLint +_mesa_add_attribute(struct gl_program_parameter_list *paramList, + const char *name, GLint size, GLenum datatype, GLint attrib) +{ + GLint i = _mesa_lookup_parameter_index(paramList, -1, name); + if (i >= 0) { + /* replace */ + if (attrib < 0) + attrib = i; + paramList->Parameters[i].StateIndexes[0] = attrib; + } + else { + /* add */ + gl_state_index state[STATE_LENGTH]; + state[0] = (gl_state_index) attrib; + if (size < 0) + size = 4; + i = _mesa_add_parameter(paramList, PROGRAM_INPUT, name, + size, datatype, NULL, state, 0x0); + } + return i; +} + + + +#if 0 /* not used yet */ +/** + * Returns the number of 4-component registers needed to store a piece + * of GL state. For matrices this may be as many as 4 registers, + * everything else needs + * just 1 register. + */ +static GLuint +sizeof_state_reference(const GLint *stateTokens) +{ + if (stateTokens[0] == STATE_MATRIX) { + GLuint rows = stateTokens[4] - stateTokens[3] + 1; + assert(rows >= 1); + assert(rows <= 4); + return rows; + } + else { + return 1; + } +} +#endif + + +/** + * Add a new state reference to the parameter list. + * This will be used when the program contains something like this: + * PARAM ambient = state.material.front.ambient; + * + * \param paramList the parameter list + * \param stateTokens an array of 5 (STATE_LENGTH) state tokens + * \return index of the new parameter. + */ +GLint +_mesa_add_state_reference(struct gl_program_parameter_list *paramList, + const gl_state_index stateTokens[STATE_LENGTH]) +{ + const GLuint size = 4; /* XXX fix */ + char *name; + GLint index; + + /* Check if the state reference is already in the list */ + for (index = 0; index < (GLint) paramList->NumParameters; index++) { + GLuint i, match = 0; + for (i = 0; i < STATE_LENGTH; i++) { + if (paramList->Parameters[index].StateIndexes[i] == stateTokens[i]) { + match++; + } + else { + break; + } + } + if (match == STATE_LENGTH) { + /* this state reference is already in the parameter list */ + return index; + } + } + + name = _mesa_program_state_string(stateTokens); + index = _mesa_add_parameter(paramList, PROGRAM_STATE_VAR, name, + size, GL_NONE, + NULL, (gl_state_index *) stateTokens, 0x0); + paramList->StateFlags |= _mesa_program_state_flags(stateTokens); + + /* free name string here since we duplicated it in add_parameter() */ + free(name); + + return index; +} + + +/** + * Lookup a parameter value by name in the given parameter list. + * \return pointer to the float[4] values. + */ +GLfloat * +_mesa_lookup_parameter_value(const struct gl_program_parameter_list *paramList, + GLsizei nameLen, const char *name) +{ + GLint i = _mesa_lookup_parameter_index(paramList, nameLen, name); + if (i < 0) + return NULL; + else + return paramList->ParameterValues[i]; +} + + +/** + * Given a program parameter name, find its position in the list of parameters. + * \param paramList the parameter list to search + * \param nameLen length of name (in chars). + * If length is negative, assume that name is null-terminated. + * \param name the name to search for + * \return index of parameter in the list. + */ +GLint +_mesa_lookup_parameter_index(const struct gl_program_parameter_list *paramList, + GLsizei nameLen, const char *name) +{ + GLint i; + + if (!paramList) + return -1; + + if (nameLen == -1) { + /* name is null-terminated */ + for (i = 0; i < (GLint) paramList->NumParameters; i++) { + if (paramList->Parameters[i].Name && + strcmp(paramList->Parameters[i].Name, name) == 0) + return i; + } + } + else { + /* name is not null-terminated, use nameLen */ + for (i = 0; i < (GLint) paramList->NumParameters; i++) { + if (paramList->Parameters[i].Name && + strncmp(paramList->Parameters[i].Name, name, nameLen) == 0 + && strlen(paramList->Parameters[i].Name) == (size_t)nameLen) + return i; + } + } + return -1; +} + + +/** + * Look for a float vector in the given parameter list. The float vector + * may be of length 1, 2, 3 or 4. If swizzleOut is non-null, we'll try + * swizzling to find a match. + * \param list the parameter list to search + * \param v the float vector to search for + * \param vSize number of element in v + * \param posOut returns the position of the constant, if found + * \param swizzleOut returns a swizzle mask describing location of the + * vector elements if found. + * \return GL_TRUE if found, GL_FALSE if not found + */ +GLboolean +_mesa_lookup_parameter_constant(const struct gl_program_parameter_list *list, + const GLfloat v[], GLuint vSize, + GLint *posOut, GLuint *swizzleOut) +{ + GLuint i; + + assert(vSize >= 1); + assert(vSize <= 4); + + if (!list) { + *posOut = -1; + return GL_FALSE; + } + + for (i = 0; i < list->NumParameters; i++) { + if (list->Parameters[i].Type == PROGRAM_CONSTANT) { + if (!swizzleOut) { + /* swizzle not allowed */ + GLuint j, match = 0; + for (j = 0; j < vSize; j++) { + if (v[j] == list->ParameterValues[i][j]) + match++; + } + if (match == vSize) { + *posOut = i; + return GL_TRUE; + } + } + else { + /* try matching w/ swizzle */ + if (vSize == 1) { + /* look for v[0] anywhere within float[4] value */ + GLuint j; + for (j = 0; j < list->Parameters[i].Size; j++) { + if (list->ParameterValues[i][j] == v[0]) { + /* found it */ + *posOut = i; + *swizzleOut = MAKE_SWIZZLE4(j, j, j, j); + return GL_TRUE; + } + } + } + else if (vSize <= list->Parameters[i].Size) { + /* see if we can match this constant (with a swizzle) */ + GLuint swz[4]; + GLuint match = 0, j, k; + for (j = 0; j < vSize; j++) { + if (v[j] == list->ParameterValues[i][j]) { + swz[j] = j; + match++; + } + else { + for (k = 0; k < list->Parameters[i].Size; k++) { + if (v[j] == list->ParameterValues[i][k]) { + swz[j] = k; + match++; + break; + } + } + } + } + /* smear last value to remaining positions */ + for (; j < 4; j++) + swz[j] = swz[j-1]; + + if (match == vSize) { + *posOut = i; + *swizzleOut = MAKE_SWIZZLE4(swz[0], swz[1], swz[2], swz[3]); + return GL_TRUE; + } + } + } + } + } + + *posOut = -1; + return GL_FALSE; +} + + +struct gl_program_parameter_list * +_mesa_clone_parameter_list(const struct gl_program_parameter_list *list) +{ + struct gl_program_parameter_list *clone; + GLuint i; + + clone = _mesa_new_parameter_list(); + if (!clone) + return NULL; + + /** Not too efficient, but correct */ + for (i = 0; i < list->NumParameters; i++) { + struct gl_program_parameter *p = list->Parameters + i; + struct gl_program_parameter *pCopy; + GLuint size = MIN2(p->Size, 4); + GLint j = _mesa_add_parameter(clone, p->Type, p->Name, size, p->DataType, + list->ParameterValues[i], NULL, 0x0); + ASSERT(j >= 0); + pCopy = clone->Parameters + j; + pCopy->Flags = p->Flags; + /* copy state indexes */ + if (p->Type == PROGRAM_STATE_VAR) { + GLint k; + for (k = 0; k < STATE_LENGTH; k++) { + pCopy->StateIndexes[k] = p->StateIndexes[k]; + } + } + else { + clone->Parameters[j].Size = p->Size; + } + + } + + clone->StateFlags = list->StateFlags; + + return clone; +} + + +/** + * Return a new parameter list which is listA + listB. + */ +struct gl_program_parameter_list * +_mesa_combine_parameter_lists(const struct gl_program_parameter_list *listA, + const struct gl_program_parameter_list *listB) +{ + struct gl_program_parameter_list *list; + + if (listA) { + list = _mesa_clone_parameter_list(listA); + if (list && listB) { + GLuint i; + for (i = 0; i < listB->NumParameters; i++) { + struct gl_program_parameter *param = listB->Parameters + i; + _mesa_add_parameter(list, param->Type, param->Name, param->Size, + param->DataType, + listB->ParameterValues[i], + param->StateIndexes, + param->Flags); + } + } + } + else if (listB) { + list = _mesa_clone_parameter_list(listB); + } + else { + list = NULL; + } + return list; +} + + + +/** + * Find longest name of all uniform parameters in list. + */ +GLuint +_mesa_longest_parameter_name(const struct gl_program_parameter_list *list, + gl_register_file type) +{ + GLuint i, maxLen = 0; + if (!list) + return 0; + for (i = 0; i < list->NumParameters; i++) { + if (list->Parameters[i].Type == type) { + GLuint len = strlen(list->Parameters[i].Name); + if (len > maxLen) + maxLen = len; + } + } + return maxLen; +} + + +/** + * Count the number of parameters in the last that match the given type. + */ +GLuint +_mesa_num_parameters_of_type(const struct gl_program_parameter_list *list, + gl_register_file type) +{ + GLuint i, count = 0; + if (list) { + for (i = 0; i < list->NumParameters; i++) { + if (list->Parameters[i].Type == type) + count++; + } + } + return count; +} diff --git a/mesalib/src/mesa/program/prog_parameter.h b/mesalib/src/mesa/program/prog_parameter.h new file mode 100644 index 000000000..10cbbe57a --- /dev/null +++ b/mesalib/src/mesa/program/prog_parameter.h @@ -0,0 +1,168 @@ +/* + * Mesa 3-D graphics library + * Version: 7.3 + * + * Copyright (C) 1999-2008 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. + */ + +/** + * \file prog_parameter.c + * Program parameter lists and functions. + * \author Brian Paul + */ + +#ifndef PROG_PARAMETER_H +#define PROG_PARAMETER_H + +#include "main/mtypes.h" +#include "prog_statevars.h" + + +/** + * Program parameter flags + */ +/*@{*/ +#define PROG_PARAM_BIT_CENTROID 0x1 /**< for varying vars (GLSL 1.20) */ +#define PROG_PARAM_BIT_INVARIANT 0x2 /**< for varying vars (GLSL 1.20) */ +#define PROG_PARAM_BIT_FLAT 0x4 /**< for varying vars (GLSL 1.30) */ +#define PROG_PARAM_BIT_LINEAR 0x8 /**< for varying vars (GLSL 1.30) */ +#define PROG_PARAM_BIT_CYL_WRAP 0x10 /**< XXX gallium debug */ +/*@}*/ + + + +/** + * Program parameter. + * Used by shaders/programs for uniforms, constants, varying vars, etc. + */ +struct gl_program_parameter +{ + const char *Name; /**< Null-terminated string */ + gl_register_file Type; /**< PROGRAM_NAMED_PARAM, CONSTANT or STATE_VAR */ + GLenum DataType; /**< GL_FLOAT, GL_FLOAT_VEC2, etc */ + /** + * Number of components (1..4), or more. + * If the number of components is greater than 4, + * this parameter is part of a larger uniform like a GLSL matrix or array. + * The next program parameter's Size will be Size-4 of this parameter. + */ + GLuint Size; + GLboolean Initialized; /**< debug: Has the ParameterValue[] been set? */ + GLbitfield Flags; /**< Bitmask of PROG_PARAM_*_BIT */ + /** + * A sequence of STATE_* tokens and integers to identify GL state. + */ + gl_state_index StateIndexes[STATE_LENGTH]; +}; + + +/** + * List of gl_program_parameter instances. + */ +struct gl_program_parameter_list +{ + GLuint Size; /**< allocated size of Parameters, ParameterValues */ + GLuint NumParameters; /**< number of parameters in arrays */ + struct gl_program_parameter *Parameters; /**< Array [Size] */ + GLfloat (*ParameterValues)[4]; /**< Array [Size] of GLfloat[4] */ + GLbitfield StateFlags; /**< _NEW_* flags indicating which state changes + might invalidate ParameterValues[] */ +}; + + +extern struct gl_program_parameter_list * +_mesa_new_parameter_list(void); + +extern struct gl_program_parameter_list * +_mesa_new_parameter_list_sized(unsigned size); + +extern void +_mesa_free_parameter_list(struct gl_program_parameter_list *paramList); + +extern struct gl_program_parameter_list * +_mesa_clone_parameter_list(const struct gl_program_parameter_list *list); + +extern struct gl_program_parameter_list * +_mesa_combine_parameter_lists(const struct gl_program_parameter_list *a, + const struct gl_program_parameter_list *b); + +static INLINE GLuint +_mesa_num_parameters(const struct gl_program_parameter_list *list) +{ + return list ? list->NumParameters : 0; +} + +extern GLint +_mesa_add_parameter(struct gl_program_parameter_list *paramList, + gl_register_file type, const char *name, + GLuint size, GLenum datatype, const GLfloat *values, + const gl_state_index state[STATE_LENGTH], + GLbitfield flags); + +extern GLint +_mesa_add_named_parameter(struct gl_program_parameter_list *paramList, + const char *name, const GLfloat values[4]); + +extern GLint +_mesa_add_named_constant(struct gl_program_parameter_list *paramList, + const char *name, const GLfloat values[4], + GLuint size); + +extern GLint +_mesa_add_unnamed_constant(struct gl_program_parameter_list *paramList, + const GLfloat values[4], GLuint size, + GLuint *swizzleOut); + +extern GLint +_mesa_add_varying(struct gl_program_parameter_list *paramList, + const char *name, GLuint size, GLenum datatype, + GLbitfield flags); + +extern GLint +_mesa_add_attribute(struct gl_program_parameter_list *paramList, + const char *name, GLint size, GLenum datatype, GLint attrib); + +extern GLint +_mesa_add_state_reference(struct gl_program_parameter_list *paramList, + const gl_state_index stateTokens[STATE_LENGTH]); + +extern GLfloat * +_mesa_lookup_parameter_value(const struct gl_program_parameter_list *paramList, + GLsizei nameLen, const char *name); + +extern GLint +_mesa_lookup_parameter_index(const struct gl_program_parameter_list *paramList, + GLsizei nameLen, const char *name); + +extern GLboolean +_mesa_lookup_parameter_constant(const struct gl_program_parameter_list *list, + const GLfloat v[], GLuint vSize, + GLint *posOut, GLuint *swizzleOut); + +extern GLuint +_mesa_longest_parameter_name(const struct gl_program_parameter_list *list, + gl_register_file type); + +extern GLuint +_mesa_num_parameters_of_type(const struct gl_program_parameter_list *list, + gl_register_file type); + + +#endif /* PROG_PARAMETER_H */ diff --git a/mesalib/src/mesa/program/prog_parameter_layout.c b/mesalib/src/mesa/program/prog_parameter_layout.c new file mode 100644 index 000000000..d7dc97edb --- /dev/null +++ b/mesalib/src/mesa/program/prog_parameter_layout.c @@ -0,0 +1,214 @@ +/* + * Copyright © 2009 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 prog_parameter_layout.c + * \brief Helper functions to layout storage for program parameters + * + * \author Ian Romanick <ian.d.romanick@intel.com> + */ + +#include "main/compiler.h" +#include "main/mtypes.h" +#include "prog_parameter.h" +#include "prog_parameter_layout.h" +#include "prog_instruction.h" +#include "program_parser.h" + +unsigned +_mesa_combine_swizzles(unsigned base, unsigned applied) +{ + unsigned swiz = 0; + unsigned i; + + for (i = 0; i < 4; i++) { + const unsigned s = GET_SWZ(applied, i); + + swiz |= ((s <= SWIZZLE_W) ? GET_SWZ(base, s) : s) << (i * 3); + } + + return swiz; +} + + +/** + * Copy indirect access array from one parameter list to another + * + * \param src Parameter array copied from + * \param dst Parameter array copied to + * \param first Index of first element in \c src to copy + * \param count Number of elements to copy + * + * \return + * The location in \c dst of the first element copied from \c src on + * success. -1 on failure. + * + * \warning + * This function assumes that there is already enough space available in + * \c dst to hold all of the elements that will be copied over. + */ +static int +copy_indirect_accessed_array(struct gl_program_parameter_list *src, + struct gl_program_parameter_list *dst, + unsigned first, unsigned count) +{ + const int base = dst->NumParameters; + unsigned i, j; + + for (i = first; i < (first + count); i++) { + struct gl_program_parameter *curr = & src->Parameters[i]; + + if (curr->Type == PROGRAM_CONSTANT) { + j = dst->NumParameters; + } else { + for (j = 0; j < dst->NumParameters; j++) { + if (memcmp(dst->Parameters[j].StateIndexes, curr->StateIndexes, + sizeof(curr->StateIndexes)) == 0) { + return -1; + } + } + } + + assert(j == dst->NumParameters); + + /* copy src parameter [i] to dest parameter [j] */ + memcpy(& dst->Parameters[j], curr, + sizeof(dst->Parameters[j])); + memcpy(dst->ParameterValues[j], src->ParameterValues[i], + sizeof(GLfloat) * 4); + + /* Pointer to the string name was copied. Null-out src param name + * to prevent double free later. + */ + curr->Name = NULL; + + dst->NumParameters++; + } + + return base; +} + + +/** + * XXX description??? + * \return GL_TRUE for success, GL_FALSE for failure + */ +GLboolean +_mesa_layout_parameters(struct asm_parser_state *state) +{ + struct gl_program_parameter_list *layout; + struct asm_instruction *inst; + unsigned i; + + layout = + _mesa_new_parameter_list_sized(state->prog->Parameters->NumParameters); + + /* PASS 1: Move any parameters that are accessed indirectly from the + * original parameter list to the new parameter list. + */ + for (inst = state->inst_head; inst != NULL; inst = inst->next) { + for (i = 0; i < 3; i++) { + if (inst->SrcReg[i].Base.RelAddr) { + /* Only attempt to add the to the new parameter list once. + */ + if (!inst->SrcReg[i].Symbol->pass1_done) { + const int new_begin = + copy_indirect_accessed_array(state->prog->Parameters, layout, + inst->SrcReg[i].Symbol->param_binding_begin, + inst->SrcReg[i].Symbol->param_binding_length); + + if (new_begin < 0) { + return GL_FALSE; + } + + inst->SrcReg[i].Symbol->param_binding_begin = new_begin; + inst->SrcReg[i].Symbol->pass1_done = 1; + } + + /* Previously the Index was just the offset from the parameter + * array. Now that the base of the parameter array is known, the + * index can be updated to its actual value. + */ + inst->Base.SrcReg[i] = inst->SrcReg[i].Base; + inst->Base.SrcReg[i].Index += + inst->SrcReg[i].Symbol->param_binding_begin; + } + } + } + + /* PASS 2: Move any parameters that are not accessed indirectly from the + * original parameter list to the new parameter list. + */ + for (inst = state->inst_head; inst != NULL; inst = inst->next) { + for (i = 0; i < 3; i++) { + const struct gl_program_parameter *p; + const int idx = inst->SrcReg[i].Base.Index; + unsigned swizzle = SWIZZLE_NOOP; + + /* All relative addressed operands were processed on the first + * pass. Just skip them here. + */ + if (inst->SrcReg[i].Base.RelAddr) { + continue; + } + + if ((inst->SrcReg[i].Base.File <= PROGRAM_VARYING ) + || (inst->SrcReg[i].Base.File >= PROGRAM_WRITE_ONLY)) { + continue; + } + + inst->Base.SrcReg[i] = inst->SrcReg[i].Base; + p = & state->prog->Parameters->Parameters[idx]; + + switch (p->Type) { + case PROGRAM_CONSTANT: { + const float *const v = + state->prog->Parameters->ParameterValues[idx]; + + inst->Base.SrcReg[i].Index = + _mesa_add_unnamed_constant(layout, v, p->Size, & swizzle); + + inst->Base.SrcReg[i].Swizzle = + _mesa_combine_swizzles(swizzle, inst->Base.SrcReg[i].Swizzle); + break; + } + + case PROGRAM_STATE_VAR: + inst->Base.SrcReg[i].Index = + _mesa_add_state_reference(layout, p->StateIndexes); + break; + + default: + break; + } + + inst->SrcReg[i].Base.File = p->Type; + inst->Base.SrcReg[i].File = p->Type; + } + } + + _mesa_free_parameter_list(state->prog->Parameters); + state->prog->Parameters = layout; + + return GL_TRUE; +} diff --git a/mesalib/src/mesa/program/prog_parameter_layout.h b/mesalib/src/mesa/program/prog_parameter_layout.h new file mode 100644 index 000000000..99a7b6c72 --- /dev/null +++ b/mesalib/src/mesa/program/prog_parameter_layout.h @@ -0,0 +1,42 @@ +/* + * Copyright © 2009 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 prog_parameter_layout.h + * \brief Helper functions to layout storage for program parameters + * + * \author Ian Romanick <ian.d.romanick@intel.com> + */ + +#pragma once + +#ifndef PROG_PARAMETER_LAYOUT_H +#define PROG_PARAMETER_LAYOUT_H + +extern unsigned _mesa_combine_swizzles(unsigned base, unsigned applied); + +struct asm_parser_state; + +extern GLboolean _mesa_layout_parameters(struct asm_parser_state *state); + +#endif /* PROG_PARAMETER_LAYOUT_H */ diff --git a/mesalib/src/mesa/program/prog_print.c b/mesalib/src/mesa/program/prog_print.c new file mode 100644 index 000000000..00aa6de96 --- /dev/null +++ b/mesalib/src/mesa/program/prog_print.c @@ -0,0 +1,1086 @@ +/* + * Mesa 3-D graphics library + * Version: 7.3 + * + * 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 prog_print.c + * Print vertex/fragment programs - for debugging. + * \author Brian Paul + */ + +#include "main/glheader.h" +#include "main/context.h" +#include "main/imports.h" +#include "prog_instruction.h" +#include "prog_parameter.h" +#include "prog_print.h" +#include "prog_statevars.h" + + + +/** + * Return string name for given program/register file. + */ +static const char * +file_string(gl_register_file f, gl_prog_print_mode mode) +{ + switch (f) { + case PROGRAM_TEMPORARY: + return "TEMP"; + case PROGRAM_LOCAL_PARAM: + return "LOCAL"; + case PROGRAM_ENV_PARAM: + return "ENV"; + case PROGRAM_STATE_VAR: + return "STATE"; + case PROGRAM_INPUT: + return "INPUT"; + case PROGRAM_OUTPUT: + return "OUTPUT"; + case PROGRAM_NAMED_PARAM: + return "NAMED"; + case PROGRAM_CONSTANT: + return "CONST"; + case PROGRAM_UNIFORM: + return "UNIFORM"; + case PROGRAM_VARYING: + return "VARYING"; + case PROGRAM_WRITE_ONLY: + return "WRITE_ONLY"; + case PROGRAM_ADDRESS: + return "ADDR"; + case PROGRAM_SAMPLER: + return "SAMPLER"; + case PROGRAM_UNDEFINED: + return "UNDEFINED"; + default: + { + static char s[20]; + _mesa_snprintf(s, sizeof(s), "FILE%u", f); + return s; + } + } +} + + +/** + * Return ARB_v/f_prog-style input attrib string. + */ +static const char * +arb_input_attrib_string(GLint index, GLenum progType) +{ + /* + * These strings should match the VERT_ATTRIB_x and FRAG_ATTRIB_x tokens. + */ + const char *vertAttribs[] = { + "vertex.position", + "vertex.weight", + "vertex.normal", + "vertex.color.primary", + "vertex.color.secondary", + "vertex.fogcoord", + "vertex.(six)", + "vertex.(seven)", + "vertex.texcoord[0]", + "vertex.texcoord[1]", + "vertex.texcoord[2]", + "vertex.texcoord[3]", + "vertex.texcoord[4]", + "vertex.texcoord[5]", + "vertex.texcoord[6]", + "vertex.texcoord[7]", + "vertex.attrib[0]", + "vertex.attrib[1]", + "vertex.attrib[2]", + "vertex.attrib[3]", + "vertex.attrib[4]", + "vertex.attrib[5]", + "vertex.attrib[6]", + "vertex.attrib[7]", + "vertex.attrib[8]", + "vertex.attrib[9]", + "vertex.attrib[10]", + "vertex.attrib[11]", + "vertex.attrib[12]", + "vertex.attrib[13]", + "vertex.attrib[14]", + "vertex.attrib[15]" + }; + const char *fragAttribs[] = { + "fragment.position", + "fragment.color.primary", + "fragment.color.secondary", + "fragment.fogcoord", + "fragment.texcoord[0]", + "fragment.texcoord[1]", + "fragment.texcoord[2]", + "fragment.texcoord[3]", + "fragment.texcoord[4]", + "fragment.texcoord[5]", + "fragment.texcoord[6]", + "fragment.texcoord[7]", + "fragment.varying[0]", + "fragment.varying[1]", + "fragment.varying[2]", + "fragment.varying[3]", + "fragment.varying[4]", + "fragment.varying[5]", + "fragment.varying[6]", + "fragment.varying[7]" + }; + + /* sanity checks */ + assert(strcmp(vertAttribs[VERT_ATTRIB_TEX0], "vertex.texcoord[0]") == 0); + assert(strcmp(vertAttribs[VERT_ATTRIB_GENERIC15], "vertex.attrib[15]") == 0); + + if (progType == GL_VERTEX_PROGRAM_ARB) { + assert(index < sizeof(vertAttribs) / sizeof(vertAttribs[0])); + return vertAttribs[index]; + } + else { + assert(index < sizeof(fragAttribs) / sizeof(fragAttribs[0])); + return fragAttribs[index]; + } +} + + +/** + * Print a vertex program's InputsRead field in human-readable format. + * For debugging. + */ +void +_mesa_print_vp_inputs(GLbitfield inputs) +{ + printf("VP Inputs 0x%x: \n", inputs); + while (inputs) { + GLint attr = _mesa_ffs(inputs) - 1; + const char *name = arb_input_attrib_string(attr, + GL_VERTEX_PROGRAM_ARB); + printf(" %d: %s\n", attr, name); + inputs &= ~(1 << attr); + } +} + + +/** + * Print a fragment program's InputsRead field in human-readable format. + * For debugging. + */ +void +_mesa_print_fp_inputs(GLbitfield inputs) +{ + printf("FP Inputs 0x%x: \n", inputs); + while (inputs) { + GLint attr = _mesa_ffs(inputs) - 1; + const char *name = arb_input_attrib_string(attr, + GL_FRAGMENT_PROGRAM_ARB); + printf(" %d: %s\n", attr, name); + inputs &= ~(1 << attr); + } +} + + + +/** + * Return ARB_v/f_prog-style output attrib string. + */ +static const char * +arb_output_attrib_string(GLint index, GLenum progType) +{ + /* + * These strings should match the VERT_RESULT_x and FRAG_RESULT_x tokens. + */ + const char *vertResults[] = { + "result.position", + "result.color.primary", + "result.color.secondary", + "result.fogcoord", + "result.texcoord[0]", + "result.texcoord[1]", + "result.texcoord[2]", + "result.texcoord[3]", + "result.texcoord[4]", + "result.texcoord[5]", + "result.texcoord[6]", + "result.texcoord[7]", + "result.varying[0]", + "result.varying[1]", + "result.varying[2]", + "result.varying[3]", + "result.varying[4]", + "result.varying[5]", + "result.varying[6]", + "result.varying[7]" + }; + const char *fragResults[] = { + "result.color", + "result.color(half)", + "result.depth", + "result.color[0]", + "result.color[1]", + "result.color[2]", + "result.color[3]" + }; + + if (progType == GL_VERTEX_PROGRAM_ARB) { + assert(index < sizeof(vertResults) / sizeof(vertResults[0])); + return vertResults[index]; + } + else { + assert(index < sizeof(fragResults) / sizeof(fragResults[0])); + return fragResults[index]; + } +} + + +/** + * Return string representation of the given register. + * Note that some types of registers (like PROGRAM_UNIFORM) aren't defined + * by the ARB/NV program languages so we've taken some liberties here. + * \param f the register file (PROGRAM_INPUT, PROGRAM_TEMPORARY, etc) + * \param index number of the register in the register file + * \param mode the output format/mode/style + * \param prog pointer to containing program + */ +static const char * +reg_string(gl_register_file f, GLint index, gl_prog_print_mode mode, + GLboolean relAddr, const struct gl_program *prog, + GLboolean hasIndex2, GLboolean relAddr2, GLint index2) +{ + static char str[100]; + const char *addr = relAddr ? "ADDR+" : ""; + + str[0] = 0; + + switch (mode) { + case PROG_PRINT_DEBUG: + sprintf(str, "%s[%s%d]", file_string(f, mode), addr, index); + if (hasIndex2) { + int offset = strlen(str); + const char *addr2 = relAddr2 ? "ADDR+" : ""; + sprintf(str+offset, "[%s%d]", addr2, index2); + } + break; + + case PROG_PRINT_ARB: + switch (f) { + case PROGRAM_INPUT: + sprintf(str, "%s", arb_input_attrib_string(index, prog->Target)); + break; + case PROGRAM_OUTPUT: + sprintf(str, "%s", arb_output_attrib_string(index, prog->Target)); + break; + case PROGRAM_TEMPORARY: + sprintf(str, "temp%d", index); + break; + case PROGRAM_ENV_PARAM: + sprintf(str, "program.env[%s%d]", addr, index); + break; + case PROGRAM_LOCAL_PARAM: + sprintf(str, "program.local[%s%d]", addr, index); + break; + case PROGRAM_VARYING: /* extension */ + sprintf(str, "varying[%s%d]", addr, index); + break; + case PROGRAM_CONSTANT: /* extension */ + sprintf(str, "constant[%s%d]", addr, index); + break; + case PROGRAM_UNIFORM: /* extension */ + sprintf(str, "uniform[%s%d]", addr, index); + break; + case PROGRAM_STATE_VAR: + { + struct gl_program_parameter *param + = prog->Parameters->Parameters + index; + char *state = _mesa_program_state_string(param->StateIndexes); + sprintf(str, "%s", state); + free(state); + } + break; + case PROGRAM_ADDRESS: + sprintf(str, "A%d", index); + break; + default: + _mesa_problem(NULL, "bad file in reg_string()"); + } + break; + + case PROG_PRINT_NV: + switch (f) { + case PROGRAM_INPUT: + if (prog->Target == GL_VERTEX_PROGRAM_ARB) + sprintf(str, "v[%d]", index); + else + sprintf(str, "f[%d]", index); + break; + case PROGRAM_OUTPUT: + sprintf(str, "o[%d]", index); + break; + case PROGRAM_TEMPORARY: + sprintf(str, "R%d", index); + break; + case PROGRAM_ENV_PARAM: + sprintf(str, "c[%d]", index); + break; + case PROGRAM_VARYING: /* extension */ + sprintf(str, "varying[%s%d]", addr, index); + break; + case PROGRAM_UNIFORM: /* extension */ + sprintf(str, "uniform[%s%d]", addr, index); + break; + case PROGRAM_CONSTANT: /* extension */ + sprintf(str, "constant[%s%d]", addr, index); + break; + case PROGRAM_STATE_VAR: /* extension */ + sprintf(str, "state[%s%d]", addr, index); + break; + default: + _mesa_problem(NULL, "bad file in reg_string()"); + } + break; + + default: + _mesa_problem(NULL, "bad mode in reg_string()"); + } + + return str; +} + + +/** + * Return a string representation of the given swizzle word. + * If extended is true, use extended (comma-separated) format. + * \param swizzle the swizzle field + * \param negateBase 4-bit negation vector + * \param extended if true, also allow 0, 1 values + */ +const char * +_mesa_swizzle_string(GLuint swizzle, GLuint negateMask, GLboolean extended) +{ + static const char swz[] = "xyzw01!?"; /* See SWIZZLE_x definitions */ + static char s[20]; + GLuint i = 0; + + if (!extended && swizzle == SWIZZLE_NOOP && negateMask == 0) + return ""; /* no swizzle/negation */ + + if (!extended) + s[i++] = '.'; + + if (negateMask & NEGATE_X) + s[i++] = '-'; + s[i++] = swz[GET_SWZ(swizzle, 0)]; + + if (extended) { + s[i++] = ','; + } + + if (negateMask & NEGATE_Y) + s[i++] = '-'; + s[i++] = swz[GET_SWZ(swizzle, 1)]; + + if (extended) { + s[i++] = ','; + } + + if (negateMask & NEGATE_Z) + s[i++] = '-'; + s[i++] = swz[GET_SWZ(swizzle, 2)]; + + if (extended) { + s[i++] = ','; + } + + if (negateMask & NEGATE_W) + s[i++] = '-'; + s[i++] = swz[GET_SWZ(swizzle, 3)]; + + s[i] = 0; + return s; +} + + +void +_mesa_print_swizzle(GLuint swizzle) +{ + if (swizzle == SWIZZLE_XYZW) { + printf(".xyzw\n"); + } + else { + const char *s = _mesa_swizzle_string(swizzle, 0, 0); + printf("%s\n", s); + } +} + + +const char * +_mesa_writemask_string(GLuint writeMask) +{ + static char s[10]; + GLuint i = 0; + + if (writeMask == WRITEMASK_XYZW) + return ""; + + s[i++] = '.'; + if (writeMask & WRITEMASK_X) + s[i++] = 'x'; + if (writeMask & WRITEMASK_Y) + s[i++] = 'y'; + if (writeMask & WRITEMASK_Z) + s[i++] = 'z'; + if (writeMask & WRITEMASK_W) + s[i++] = 'w'; + + s[i] = 0; + return s; +} + + +const char * +_mesa_condcode_string(GLuint condcode) +{ + switch (condcode) { + case COND_GT: return "GT"; + case COND_EQ: return "EQ"; + case COND_LT: return "LT"; + case COND_UN: return "UN"; + case COND_GE: return "GE"; + case COND_LE: return "LE"; + case COND_NE: return "NE"; + case COND_TR: return "TR"; + case COND_FL: return "FL"; + default: return "cond???"; + } +} + + +static void +fprint_dst_reg(FILE * f, + const struct prog_dst_register *dstReg, + gl_prog_print_mode mode, + const struct gl_program *prog) +{ + fprintf(f, "%s%s", + reg_string((gl_register_file) dstReg->File, + dstReg->Index, mode, dstReg->RelAddr, prog, + GL_FALSE, GL_FALSE, 0), + _mesa_writemask_string(dstReg->WriteMask)); + + if (dstReg->CondMask != COND_TR) { + fprintf(f, " (%s.%s)", + _mesa_condcode_string(dstReg->CondMask), + _mesa_swizzle_string(dstReg->CondSwizzle, + GL_FALSE, GL_FALSE)); + } + +#if 0 + fprintf(f, "%s[%d]%s", + file_string((gl_register_file) dstReg->File, mode), + dstReg->Index, + _mesa_writemask_string(dstReg->WriteMask)); +#endif +} + + +static void +fprint_src_reg(FILE *f, + const struct prog_src_register *srcReg, + gl_prog_print_mode mode, + const struct gl_program *prog) +{ + const char *abs = srcReg->Abs ? "|" : ""; + + fprintf(f, "%s%s%s%s", + abs, + reg_string((gl_register_file) srcReg->File, + srcReg->Index, mode, srcReg->RelAddr, prog, + srcReg->HasIndex2, srcReg->RelAddr2, srcReg->Index2), + _mesa_swizzle_string(srcReg->Swizzle, + srcReg->Negate, GL_FALSE), + abs); +#if 0 + fprintf(f, "%s[%d]%s", + file_string((gl_register_file) srcReg->File, mode), + srcReg->Index, + _mesa_swizzle_string(srcReg->Swizzle, + srcReg->Negate, GL_FALSE)); +#endif +} + + +static void +fprint_comment(FILE *f, const struct prog_instruction *inst) +{ + if (inst->Comment) + fprintf(f, "; # %s\n", inst->Comment); + else + fprintf(f, ";\n"); +} + + +void +_mesa_fprint_alu_instruction(FILE *f, + const struct prog_instruction *inst, + const char *opcode_string, GLuint numRegs, + gl_prog_print_mode mode, + const struct gl_program *prog) +{ + GLuint j; + + fprintf(f, "%s", opcode_string); + if (inst->CondUpdate) + fprintf(f, ".C"); + + /* frag prog only */ + if (inst->SaturateMode == SATURATE_ZERO_ONE) + fprintf(f, "_SAT"); + + fprintf(f, " "); + if (inst->DstReg.File != PROGRAM_UNDEFINED) { + fprint_dst_reg(f, &inst->DstReg, mode, prog); + } + else { + fprintf(f, " ???"); + } + + if (numRegs > 0) + fprintf(f, ", "); + + for (j = 0; j < numRegs; j++) { + fprint_src_reg(f, inst->SrcReg + j, mode, prog); + if (j + 1 < numRegs) + fprintf(f, ", "); + } + + fprint_comment(f, inst); +} + + +void +_mesa_print_alu_instruction(const struct prog_instruction *inst, + const char *opcode_string, GLuint numRegs) +{ + _mesa_fprint_alu_instruction(stderr, inst, opcode_string, + numRegs, PROG_PRINT_DEBUG, NULL); +} + + +/** + * Print a single vertex/fragment program instruction. + */ +GLint +_mesa_fprint_instruction_opt(FILE *f, + const struct prog_instruction *inst, + GLint indent, + gl_prog_print_mode mode, + const struct gl_program *prog) +{ + GLint i; + + if (inst->Opcode == OPCODE_ELSE || + inst->Opcode == OPCODE_ENDIF || + inst->Opcode == OPCODE_ENDLOOP || + inst->Opcode == OPCODE_ENDSUB) { + indent -= 3; + } + for (i = 0; i < indent; i++) { + fprintf(f, " "); + } + + switch (inst->Opcode) { + case OPCODE_PRINT: + fprintf(f, "PRINT '%s'", (char *) inst->Data); + if (inst->SrcReg[0].File != PROGRAM_UNDEFINED) { + fprintf(f, ", "); + fprintf(f, "%s[%d]%s", + file_string((gl_register_file) inst->SrcReg[0].File, + mode), + inst->SrcReg[0].Index, + _mesa_swizzle_string(inst->SrcReg[0].Swizzle, + inst->SrcReg[0].Negate, GL_FALSE)); + } + if (inst->Comment) + fprintf(f, " # %s", inst->Comment); + fprint_comment(f, inst); + break; + case OPCODE_SWZ: + fprintf(f, "SWZ"); + if (inst->SaturateMode == SATURATE_ZERO_ONE) + fprintf(f, "_SAT"); + fprintf(f, " "); + fprint_dst_reg(f, &inst->DstReg, mode, prog); + fprintf(f, ", %s[%d], %s", + file_string((gl_register_file) inst->SrcReg[0].File, + mode), + inst->SrcReg[0].Index, + _mesa_swizzle_string(inst->SrcReg[0].Swizzle, + inst->SrcReg[0].Negate, GL_TRUE)); + fprint_comment(f, inst); + break; + case OPCODE_TEX: + case OPCODE_TXP: + case OPCODE_TXL: + case OPCODE_TXB: + fprintf(f, "%s", _mesa_opcode_string(inst->Opcode)); + if (inst->SaturateMode == SATURATE_ZERO_ONE) + fprintf(f, "_SAT"); + fprintf(f, " "); + fprint_dst_reg(f, &inst->DstReg, mode, prog); + fprintf(f, ", "); + fprint_src_reg(f, &inst->SrcReg[0], mode, prog); + fprintf(f, ", texture[%d], ", inst->TexSrcUnit); + switch (inst->TexSrcTarget) { + case TEXTURE_1D_INDEX: fprintf(f, "1D"); break; + case TEXTURE_2D_INDEX: fprintf(f, "2D"); break; + case TEXTURE_3D_INDEX: fprintf(f, "3D"); break; + case TEXTURE_CUBE_INDEX: fprintf(f, "CUBE"); break; + case TEXTURE_RECT_INDEX: fprintf(f, "RECT"); break; + case TEXTURE_1D_ARRAY_INDEX: fprintf(f, "1D_ARRAY"); break; + case TEXTURE_2D_ARRAY_INDEX: fprintf(f, "2D_ARRAY"); break; + default: + ; + } + if (inst->TexShadow) + fprintf(f, " SHADOW"); + fprint_comment(f, inst); + break; + + case OPCODE_KIL: + fprintf(f, "%s", _mesa_opcode_string(inst->Opcode)); + fprintf(f, " "); + fprint_src_reg(f, &inst->SrcReg[0], mode, prog); + fprint_comment(f, inst); + break; + case OPCODE_KIL_NV: + fprintf(f, "%s", _mesa_opcode_string(inst->Opcode)); + fprintf(f, " "); + fprintf(f, "%s.%s", + _mesa_condcode_string(inst->DstReg.CondMask), + _mesa_swizzle_string(inst->DstReg.CondSwizzle, + GL_FALSE, GL_FALSE)); + fprint_comment(f, inst); + break; + + case OPCODE_ARL: + fprintf(f, "ARL "); + fprint_dst_reg(f, &inst->DstReg, mode, prog); + fprintf(f, ", "); + fprint_src_reg(f, &inst->SrcReg[0], mode, prog); + fprint_comment(f, inst); + break; + case OPCODE_BRA: + fprintf(f, "BRA %d (%s%s)", + inst->BranchTarget, + _mesa_condcode_string(inst->DstReg.CondMask), + _mesa_swizzle_string(inst->DstReg.CondSwizzle, 0, GL_FALSE)); + fprint_comment(f, inst); + break; + case OPCODE_IF: + if (inst->SrcReg[0].File != PROGRAM_UNDEFINED) { + /* Use ordinary register */ + fprintf(f, "IF "); + fprint_src_reg(f, &inst->SrcReg[0], mode, prog); + fprintf(f, "; "); + } + else { + /* Use cond codes */ + fprintf(f, "IF (%s%s);", + _mesa_condcode_string(inst->DstReg.CondMask), + _mesa_swizzle_string(inst->DstReg.CondSwizzle, + 0, GL_FALSE)); + } + fprintf(f, " # (if false, goto %d)", inst->BranchTarget); + fprint_comment(f, inst); + return indent + 3; + case OPCODE_ELSE: + fprintf(f, "ELSE; # (goto %d)\n", inst->BranchTarget); + return indent + 3; + case OPCODE_ENDIF: + fprintf(f, "ENDIF;\n"); + break; + case OPCODE_BGNLOOP: + fprintf(f, "BGNLOOP; # (end at %d)\n", inst->BranchTarget); + return indent + 3; + case OPCODE_ENDLOOP: + fprintf(f, "ENDLOOP; # (goto %d)\n", inst->BranchTarget); + break; + case OPCODE_BRK: + case OPCODE_CONT: + fprintf(f, "%s (%s%s); # (goto %d)", + _mesa_opcode_string(inst->Opcode), + _mesa_condcode_string(inst->DstReg.CondMask), + _mesa_swizzle_string(inst->DstReg.CondSwizzle, 0, GL_FALSE), + inst->BranchTarget); + fprint_comment(f, inst); + break; + + case OPCODE_BGNSUB: + if (mode == PROG_PRINT_NV) { + fprintf(f, "%s:\n", inst->Comment); /* comment is label */ + return indent; + } + else { + fprintf(f, "BGNSUB"); + fprint_comment(f, inst); + return indent + 3; + } + case OPCODE_ENDSUB: + if (mode == PROG_PRINT_DEBUG) { + fprintf(f, "ENDSUB"); + fprint_comment(f, inst); + } + break; + case OPCODE_CAL: + if (mode == PROG_PRINT_NV) { + fprintf(f, "CAL %s; # (goto %d)\n", inst->Comment, inst->BranchTarget); + } + else { + fprintf(f, "CAL %u", inst->BranchTarget); + fprint_comment(f, inst); + } + break; + case OPCODE_RET: + fprintf(f, "RET (%s%s)", + _mesa_condcode_string(inst->DstReg.CondMask), + _mesa_swizzle_string(inst->DstReg.CondSwizzle, 0, GL_FALSE)); + fprint_comment(f, inst); + break; + + case OPCODE_END: + fprintf(f, "END\n"); + break; + case OPCODE_NOP: + if (mode == PROG_PRINT_DEBUG) { + fprintf(f, "NOP"); + fprint_comment(f, inst); + } + else if (inst->Comment) { + /* ARB/NV extensions don't have NOP instruction */ + fprintf(f, "# %s\n", inst->Comment); + } + break; + case OPCODE_EMIT_VERTEX: + fprintf(f, "EMIT_VERTEX\n"); + break; + case OPCODE_END_PRIMITIVE: + fprintf(f, "END_PRIMITIVE\n"); + break; + /* XXX may need other special-case instructions */ + default: + if (inst->Opcode < MAX_OPCODE) { + /* typical alu instruction */ + _mesa_fprint_alu_instruction(f, inst, + _mesa_opcode_string(inst->Opcode), + _mesa_num_inst_src_regs(inst->Opcode), + mode, prog); + } + else { + _mesa_fprint_alu_instruction(f, inst, + _mesa_opcode_string(inst->Opcode), + 3/*_mesa_num_inst_src_regs(inst->Opcode)*/, + mode, prog); + } + break; + } + return indent; +} + + +GLint +_mesa_print_instruction_opt(const struct prog_instruction *inst, + GLint indent, + gl_prog_print_mode mode, + const struct gl_program *prog) +{ + return _mesa_fprint_instruction_opt(stderr, inst, indent, mode, prog); +} + + +void +_mesa_print_instruction(const struct prog_instruction *inst) +{ + /* note: 4th param should be ignored for PROG_PRINT_DEBUG */ + _mesa_fprint_instruction_opt(stderr, inst, 0, PROG_PRINT_DEBUG, NULL); +} + + + +/** + * Print program, with options. + */ +void +_mesa_fprint_program_opt(FILE *f, + const struct gl_program *prog, + gl_prog_print_mode mode, + GLboolean lineNumbers) +{ + GLuint i, indent = 0; + + switch (prog->Target) { + case GL_VERTEX_PROGRAM_ARB: + if (mode == PROG_PRINT_ARB) + fprintf(f, "!!ARBvp1.0\n"); + else if (mode == PROG_PRINT_NV) + fprintf(f, "!!VP1.0\n"); + else + fprintf(f, "# Vertex Program/Shader %u\n", prog->Id); + break; + case GL_FRAGMENT_PROGRAM_ARB: + case GL_FRAGMENT_PROGRAM_NV: + if (mode == PROG_PRINT_ARB) + fprintf(f, "!!ARBfp1.0\n"); + else if (mode == PROG_PRINT_NV) + fprintf(f, "!!FP1.0\n"); + else + fprintf(f, "# Fragment Program/Shader %u\n", prog->Id); + break; + case MESA_GEOMETRY_PROGRAM: + fprintf(f, "# Geometry Shader\n"); + } + + for (i = 0; i < prog->NumInstructions; i++) { + if (lineNumbers) + fprintf(f, "%3d: ", i); + indent = _mesa_fprint_instruction_opt(f, prog->Instructions + i, + indent, mode, prog); + } +} + + +/** + * Print program to stderr, default options. + */ +void +_mesa_print_program(const struct gl_program *prog) +{ + _mesa_fprint_program_opt(stderr, prog, PROG_PRINT_DEBUG, GL_TRUE); +} + + +/** + * Return binary representation of 64-bit value (as a string). + * Insert a comma to separate each group of 8 bits. + * Note we return a pointer to local static storage so this is not + * re-entrant, etc. + * XXX move to imports.[ch] if useful elsewhere. + */ +static const char * +binary(GLbitfield64 val) +{ + static char buf[80]; + GLint i, len = 0; + for (i = 63; i >= 0; --i) { + if (val & (BITFIELD64_BIT(i))) + buf[len++] = '1'; + else if (len > 0 || i == 0) + buf[len++] = '0'; + if (len > 0 && ((i-1) % 8) == 7) + buf[len++] = ','; + } + buf[len] = '\0'; + return buf; +} + + +/** + * Print all of a program's parameters/fields to given file. + */ +static void +_mesa_fprint_program_parameters(FILE *f, + GLcontext *ctx, + const struct gl_program *prog) +{ + GLuint i; + + fprintf(f, "InputsRead: 0x%x (0b%s)\n", + prog->InputsRead, binary(prog->InputsRead)); + fprintf(f, "OutputsWritten: 0x%llx (0b%s)\n", + (unsigned long long)prog->OutputsWritten, + binary(prog->OutputsWritten)); + fprintf(f, "NumInstructions=%d\n", prog->NumInstructions); + fprintf(f, "NumTemporaries=%d\n", prog->NumTemporaries); + fprintf(f, "NumParameters=%d\n", prog->NumParameters); + fprintf(f, "NumAttributes=%d\n", prog->NumAttributes); + fprintf(f, "NumAddressRegs=%d\n", prog->NumAddressRegs); + fprintf(f, "IndirectRegisterFiles: 0x%x (0b%s)\n", + prog->IndirectRegisterFiles, binary(prog->IndirectRegisterFiles)); + fprintf(f, "SamplersUsed: 0x%x (0b%s)\n", + prog->SamplersUsed, binary(prog->SamplersUsed)); + fprintf(f, "Samplers=[ "); + for (i = 0; i < MAX_SAMPLERS; i++) { + fprintf(f, "%d ", prog->SamplerUnits[i]); + } + fprintf(f, "]\n"); + + _mesa_load_state_parameters(ctx, prog->Parameters); + +#if 0 + fprintf(f, "Local Params:\n"); + for (i = 0; i < MAX_PROGRAM_LOCAL_PARAMS; i++){ + const GLfloat *p = prog->LocalParams[i]; + fprintf(f, "%2d: %f, %f, %f, %f\n", i, p[0], p[1], p[2], p[3]); + } +#endif + _mesa_print_parameter_list(prog->Parameters); +} + + +/** + * Print all of a program's parameters/fields to stderr. + */ +void +_mesa_print_program_parameters(GLcontext *ctx, const struct gl_program *prog) +{ + _mesa_fprint_program_parameters(stderr, ctx, prog); +} + + +/** + * Print a program parameter list to given file. + */ +static void +_mesa_fprint_parameter_list(FILE *f, + const struct gl_program_parameter_list *list) +{ + const gl_prog_print_mode mode = PROG_PRINT_DEBUG; + GLuint i; + + if (!list) + return; + + if (0) + fprintf(f, "param list %p\n", (void *) list); + fprintf(f, "dirty state flags: 0x%x\n", list->StateFlags); + for (i = 0; i < list->NumParameters; i++){ + struct gl_program_parameter *param = list->Parameters + i; + const GLfloat *v = list->ParameterValues[i]; + fprintf(f, "param[%d] sz=%d %s %s = {%.3g, %.3g, %.3g, %.3g}", + i, param->Size, + file_string(list->Parameters[i].Type, mode), + param->Name, v[0], v[1], v[2], v[3]); + if (param->Flags & PROG_PARAM_BIT_CENTROID) + fprintf(f, " Centroid"); + if (param->Flags & PROG_PARAM_BIT_INVARIANT) + fprintf(f, " Invariant"); + if (param->Flags & PROG_PARAM_BIT_FLAT) + fprintf(f, " Flat"); + if (param->Flags & PROG_PARAM_BIT_LINEAR) + fprintf(f, " Linear"); + fprintf(f, "\n"); + } +} + + +/** + * Print a program parameter list to stderr. + */ +void +_mesa_print_parameter_list(const struct gl_program_parameter_list *list) +{ + _mesa_fprint_parameter_list(stderr, list); +} + + +/** + * Write shader and associated info to a file. + */ +void +_mesa_write_shader_to_file(const struct gl_shader *shader) +{ + const char *type; + char filename[100]; + FILE *f; + + if (shader->Type == GL_FRAGMENT_SHADER) + type = "frag"; + else if (shader->Type == GL_VERTEX_SHADER) + type = "vert"; + else + type = "geom"; + + _mesa_snprintf(filename, sizeof(filename), "shader_%u.%s", shader->Name, type); + f = fopen(filename, "w"); + if (!f) { + fprintf(stderr, "Unable to open %s for writing\n", filename); + return; + } + + fprintf(f, "/* Shader %u source, checksum %u */\n", shader->Name, shader->SourceChecksum); + fputs(shader->Source, f); + fprintf(f, "\n"); + + fprintf(f, "/* Compile status: %s */\n", + shader->CompileStatus ? "ok" : "fail"); + fprintf(f, "/* Log Info: */\n"); + if (shader->InfoLog) { + fputs(shader->InfoLog, f); + } + if (shader->CompileStatus && shader->Program) { + fprintf(f, "/* GPU code */\n"); + fprintf(f, "/*\n"); + _mesa_fprint_program_opt(f, shader->Program, PROG_PRINT_DEBUG, GL_TRUE); + fprintf(f, "*/\n"); + fprintf(f, "/* Parameters / constants */\n"); + fprintf(f, "/*\n"); + _mesa_fprint_parameter_list(f, shader->Program->Parameters); + fprintf(f, "*/\n"); + } + + fclose(f); +} + + +/** + * Append the shader's uniform info/values to the shader log file. + * The log file will typically have been created by the + * _mesa_write_shader_to_file function. + */ +void +_mesa_append_uniforms_to_file(const struct gl_shader *shader, + const struct gl_program *prog) +{ + const char *type; + char filename[100]; + FILE *f; + + if (shader->Type == GL_FRAGMENT_SHADER) + type = "frag"; + else + type = "vert"; + + _mesa_snprintf(filename, sizeof(filename), "shader_%u.%s", shader->Name, type); + f = fopen(filename, "a"); /* append */ + if (!f) { + fprintf(stderr, "Unable to open %s for appending\n", filename); + return; + } + + fprintf(f, "/* First-draw parameters / constants */\n"); + fprintf(f, "/*\n"); + _mesa_fprint_parameter_list(f, prog->Parameters); + fprintf(f, "*/\n"); + + fclose(f); +} diff --git a/mesalib/src/mesa/program/prog_print.h b/mesalib/src/mesa/program/prog_print.h new file mode 100644 index 000000000..78b90aeb4 --- /dev/null +++ b/mesalib/src/mesa/program/prog_print.h @@ -0,0 +1,117 @@ +/* + * Mesa 3-D graphics library + * Version: 6.5.3 + * + * Copyright (C) 1999-2007 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 PROG_PRINT_H +#define PROG_PRINT_H + +#include <stdio.h> + +#include "main/glheader.h" +#include "main/mtypes.h" + +struct gl_program; +struct gl_program_parameter_list; +struct gl_shader; +struct prog_instruction; + + +/** + * The output style to use when printing programs. + */ +typedef enum { + PROG_PRINT_ARB, + PROG_PRINT_NV, + PROG_PRINT_DEBUG +} gl_prog_print_mode; + + +extern void +_mesa_print_vp_inputs(GLbitfield inputs); + +extern void +_mesa_print_fp_inputs(GLbitfield inputs); + +extern const char * +_mesa_condcode_string(GLuint condcode); + +extern const char * +_mesa_swizzle_string(GLuint swizzle, GLuint negateBase, GLboolean extended); + +const char * +_mesa_writemask_string(GLuint writeMask); + +extern void +_mesa_print_swizzle(GLuint swizzle); + +extern void +_mesa_fprint_alu_instruction(FILE *f, + const struct prog_instruction *inst, + const char *opcode_string, GLuint numRegs, + gl_prog_print_mode mode, + const struct gl_program *prog); + +extern void +_mesa_print_alu_instruction(const struct prog_instruction *inst, + const char *opcode_string, GLuint numRegs); + +extern void +_mesa_print_instruction(const struct prog_instruction *inst); + +extern GLint +_mesa_fprint_instruction_opt(FILE *f, + const struct prog_instruction *inst, + GLint indent, + gl_prog_print_mode mode, + const struct gl_program *prog); + +extern GLint +_mesa_print_instruction_opt(const struct prog_instruction *inst, GLint indent, + gl_prog_print_mode mode, + const struct gl_program *prog); + +extern void +_mesa_print_program(const struct gl_program *prog); + +extern void +_mesa_fprint_program_opt(FILE *f, + const struct gl_program *prog, gl_prog_print_mode mode, + GLboolean lineNumbers); + +extern void +_mesa_print_program_parameters(GLcontext *ctx, const struct gl_program *prog); + +extern void +_mesa_print_parameter_list(const struct gl_program_parameter_list *list); + + +extern void +_mesa_write_shader_to_file(const struct gl_shader *shader); + +extern void +_mesa_append_uniforms_to_file(const struct gl_shader *shader, + const struct gl_program *prog); + + +#endif /* PROG_PRINT_H */ diff --git a/mesalib/src/mesa/program/prog_statevars.c b/mesalib/src/mesa/program/prog_statevars.c new file mode 100644 index 000000000..2687f8ae2 --- /dev/null +++ b/mesalib/src/mesa/program/prog_statevars.c @@ -0,0 +1,1189 @@ +/* + * Mesa 3-D graphics library + * Version: 7.1 + * + * Copyright (C) 1999-2007 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. + */ + +/** + * \file prog_statevars.c + * Program state variable management. + * \author Brian Paul + */ + + +#include "main/glheader.h" +#include "main/context.h" +#include "main/imports.h" +#include "main/macros.h" +#include "main/mtypes.h" +#include "prog_statevars.h" +#include "prog_parameter.h" + + +/** + * Use the list of tokens in the state[] array to find global GL state + * and return it in <value>. Usually, four values are returned in <value> + * but matrix queries may return as many as 16 values. + * This function is used for ARB vertex/fragment programs. + * The program parser will produce the state[] values. + */ +static void +_mesa_fetch_state(GLcontext *ctx, const gl_state_index state[], + GLfloat *value) +{ + switch (state[0]) { + case STATE_MATERIAL: + { + /* state[1] is either 0=front or 1=back side */ + const GLuint face = (GLuint) state[1]; + const struct gl_material *mat = &ctx->Light.Material; + ASSERT(face == 0 || face == 1); + /* we rely on tokens numbered so that _BACK_ == _FRONT_+ 1 */ + ASSERT(MAT_ATTRIB_FRONT_AMBIENT + 1 == MAT_ATTRIB_BACK_AMBIENT); + /* XXX we could get rid of this switch entirely with a little + * work in arbprogparse.c's parse_state_single_item(). + */ + /* state[2] is the material attribute */ + switch (state[2]) { + case STATE_AMBIENT: + COPY_4V(value, mat->Attrib[MAT_ATTRIB_FRONT_AMBIENT + face]); + return; + case STATE_DIFFUSE: + COPY_4V(value, mat->Attrib[MAT_ATTRIB_FRONT_DIFFUSE + face]); + return; + case STATE_SPECULAR: + COPY_4V(value, mat->Attrib[MAT_ATTRIB_FRONT_SPECULAR + face]); + return; + case STATE_EMISSION: + COPY_4V(value, mat->Attrib[MAT_ATTRIB_FRONT_EMISSION + face]); + return; + case STATE_SHININESS: + value[0] = mat->Attrib[MAT_ATTRIB_FRONT_SHININESS + face][0]; + value[1] = 0.0F; + value[2] = 0.0F; + value[3] = 1.0F; + return; + default: + _mesa_problem(ctx, "Invalid material state in fetch_state"); + return; + } + } + case STATE_LIGHT: + { + /* state[1] is the light number */ + const GLuint ln = (GLuint) state[1]; + /* state[2] is the light attribute */ + switch (state[2]) { + case STATE_AMBIENT: + COPY_4V(value, ctx->Light.Light[ln].Ambient); + return; + case STATE_DIFFUSE: + COPY_4V(value, ctx->Light.Light[ln].Diffuse); + return; + case STATE_SPECULAR: + COPY_4V(value, ctx->Light.Light[ln].Specular); + return; + case STATE_POSITION: + COPY_4V(value, ctx->Light.Light[ln].EyePosition); + return; + case STATE_ATTENUATION: + value[0] = ctx->Light.Light[ln].ConstantAttenuation; + value[1] = ctx->Light.Light[ln].LinearAttenuation; + value[2] = ctx->Light.Light[ln].QuadraticAttenuation; + value[3] = ctx->Light.Light[ln].SpotExponent; + return; + case STATE_SPOT_DIRECTION: + COPY_3V(value, ctx->Light.Light[ln].SpotDirection); + value[3] = ctx->Light.Light[ln]._CosCutoff; + return; + case STATE_SPOT_CUTOFF: + value[0] = ctx->Light.Light[ln].SpotCutoff; + return; + case STATE_HALF_VECTOR: + { + static const GLfloat eye_z[] = {0, 0, 1}; + GLfloat p[3]; + /* Compute infinite half angle vector: + * halfVector = normalize(normalize(lightPos) + (0, 0, 1)) + * light.EyePosition.w should be 0 for infinite lights. + */ + COPY_3V(p, ctx->Light.Light[ln].EyePosition); + NORMALIZE_3FV(p); + ADD_3V(value, p, eye_z); + NORMALIZE_3FV(value); + value[3] = 1.0; + } + return; + default: + _mesa_problem(ctx, "Invalid light state in fetch_state"); + return; + } + } + case STATE_LIGHTMODEL_AMBIENT: + COPY_4V(value, ctx->Light.Model.Ambient); + return; + case STATE_LIGHTMODEL_SCENECOLOR: + if (state[1] == 0) { + /* front */ + GLint i; + for (i = 0; i < 3; i++) { + value[i] = ctx->Light.Model.Ambient[i] + * ctx->Light.Material.Attrib[MAT_ATTRIB_FRONT_AMBIENT][i] + + ctx->Light.Material.Attrib[MAT_ATTRIB_FRONT_EMISSION][i]; + } + value[3] = ctx->Light.Material.Attrib[MAT_ATTRIB_FRONT_DIFFUSE][3]; + } + else { + /* back */ + GLint i; + for (i = 0; i < 3; i++) { + value[i] = ctx->Light.Model.Ambient[i] + * ctx->Light.Material.Attrib[MAT_ATTRIB_BACK_AMBIENT][i] + + ctx->Light.Material.Attrib[MAT_ATTRIB_BACK_EMISSION][i]; + } + value[3] = ctx->Light.Material.Attrib[MAT_ATTRIB_BACK_DIFFUSE][3]; + } + return; + case STATE_LIGHTPROD: + { + const GLuint ln = (GLuint) state[1]; + const GLuint face = (GLuint) state[2]; + GLint i; + ASSERT(face == 0 || face == 1); + switch (state[3]) { + case STATE_AMBIENT: + for (i = 0; i < 3; i++) { + value[i] = ctx->Light.Light[ln].Ambient[i] * + ctx->Light.Material.Attrib[MAT_ATTRIB_FRONT_AMBIENT+face][i]; + } + /* [3] = material alpha */ + value[3] = ctx->Light.Material.Attrib[MAT_ATTRIB_FRONT_AMBIENT+face][3]; + return; + case STATE_DIFFUSE: + for (i = 0; i < 3; i++) { + value[i] = ctx->Light.Light[ln].Diffuse[i] * + ctx->Light.Material.Attrib[MAT_ATTRIB_FRONT_DIFFUSE+face][i]; + } + /* [3] = material alpha */ + value[3] = ctx->Light.Material.Attrib[MAT_ATTRIB_FRONT_DIFFUSE+face][3]; + return; + case STATE_SPECULAR: + for (i = 0; i < 3; i++) { + value[i] = ctx->Light.Light[ln].Specular[i] * + ctx->Light.Material.Attrib[MAT_ATTRIB_FRONT_SPECULAR+face][i]; + } + /* [3] = material alpha */ + value[3] = ctx->Light.Material.Attrib[MAT_ATTRIB_FRONT_SPECULAR+face][3]; + return; + default: + _mesa_problem(ctx, "Invalid lightprod state in fetch_state"); + return; + } + } + case STATE_TEXGEN: + { + /* state[1] is the texture unit */ + const GLuint unit = (GLuint) state[1]; + /* state[2] is the texgen attribute */ + switch (state[2]) { + case STATE_TEXGEN_EYE_S: + COPY_4V(value, ctx->Texture.Unit[unit].GenS.EyePlane); + return; + case STATE_TEXGEN_EYE_T: + COPY_4V(value, ctx->Texture.Unit[unit].GenT.EyePlane); + return; + case STATE_TEXGEN_EYE_R: + COPY_4V(value, ctx->Texture.Unit[unit].GenR.EyePlane); + return; + case STATE_TEXGEN_EYE_Q: + COPY_4V(value, ctx->Texture.Unit[unit].GenQ.EyePlane); + return; + case STATE_TEXGEN_OBJECT_S: + COPY_4V(value, ctx->Texture.Unit[unit].GenS.ObjectPlane); + return; + case STATE_TEXGEN_OBJECT_T: + COPY_4V(value, ctx->Texture.Unit[unit].GenT.ObjectPlane); + return; + case STATE_TEXGEN_OBJECT_R: + COPY_4V(value, ctx->Texture.Unit[unit].GenR.ObjectPlane); + return; + case STATE_TEXGEN_OBJECT_Q: + COPY_4V(value, ctx->Texture.Unit[unit].GenQ.ObjectPlane); + return; + default: + _mesa_problem(ctx, "Invalid texgen state in fetch_state"); + return; + } + } + case STATE_TEXENV_COLOR: + { + /* state[1] is the texture unit */ + const GLuint unit = (GLuint) state[1]; + COPY_4V(value, ctx->Texture.Unit[unit].EnvColor); + } + return; + case STATE_FOG_COLOR: + COPY_4V(value, ctx->Fog.Color); + return; + case STATE_FOG_PARAMS: + value[0] = ctx->Fog.Density; + value[1] = ctx->Fog.Start; + value[2] = ctx->Fog.End; + value[3] = (ctx->Fog.End == ctx->Fog.Start) + ? 1.0f : (GLfloat)(1.0 / (ctx->Fog.End - ctx->Fog.Start)); + return; + case STATE_CLIPPLANE: + { + const GLuint plane = (GLuint) state[1]; + COPY_4V(value, ctx->Transform.EyeUserPlane[plane]); + } + return; + case STATE_POINT_SIZE: + value[0] = ctx->Point.Size; + value[1] = ctx->Point.MinSize; + value[2] = ctx->Point.MaxSize; + value[3] = ctx->Point.Threshold; + return; + case STATE_POINT_ATTENUATION: + value[0] = ctx->Point.Params[0]; + value[1] = ctx->Point.Params[1]; + value[2] = ctx->Point.Params[2]; + value[3] = 1.0F; + return; + case STATE_MODELVIEW_MATRIX: + case STATE_PROJECTION_MATRIX: + case STATE_MVP_MATRIX: + case STATE_TEXTURE_MATRIX: + case STATE_PROGRAM_MATRIX: + case STATE_COLOR_MATRIX: + { + /* state[0] = modelview, projection, texture, etc. */ + /* state[1] = which texture matrix or program matrix */ + /* state[2] = first row to fetch */ + /* state[3] = last row to fetch */ + /* state[4] = transpose, inverse or invtrans */ + const GLmatrix *matrix; + const gl_state_index mat = state[0]; + const GLuint index = (GLuint) state[1]; + const GLuint firstRow = (GLuint) state[2]; + const GLuint lastRow = (GLuint) state[3]; + const gl_state_index modifier = state[4]; + const GLfloat *m; + GLuint row, i; + ASSERT(firstRow >= 0); + ASSERT(firstRow < 4); + ASSERT(lastRow >= 0); + ASSERT(lastRow < 4); + if (mat == STATE_MODELVIEW_MATRIX) { + matrix = ctx->ModelviewMatrixStack.Top; + } + else if (mat == STATE_PROJECTION_MATRIX) { + matrix = ctx->ProjectionMatrixStack.Top; + } + else if (mat == STATE_MVP_MATRIX) { + matrix = &ctx->_ModelProjectMatrix; + } + else if (mat == STATE_TEXTURE_MATRIX) { + ASSERT(index < Elements(ctx->TextureMatrixStack)); + matrix = ctx->TextureMatrixStack[index].Top; + } + else if (mat == STATE_PROGRAM_MATRIX) { + ASSERT(index < Elements(ctx->ProgramMatrixStack)); + matrix = ctx->ProgramMatrixStack[index].Top; + } + else if (mat == STATE_COLOR_MATRIX) { + matrix = ctx->ColorMatrixStack.Top; + } + else { + _mesa_problem(ctx, "Bad matrix name in _mesa_fetch_state()"); + return; + } + if (modifier == STATE_MATRIX_INVERSE || + modifier == STATE_MATRIX_INVTRANS) { + /* Be sure inverse is up to date: + */ + _math_matrix_alloc_inv( (GLmatrix *) matrix ); + _math_matrix_analyse( (GLmatrix*) matrix ); + m = matrix->inv; + } + else { + m = matrix->m; + } + if (modifier == STATE_MATRIX_TRANSPOSE || + modifier == STATE_MATRIX_INVTRANS) { + for (i = 0, row = firstRow; row <= lastRow; row++) { + value[i++] = m[row * 4 + 0]; + value[i++] = m[row * 4 + 1]; + value[i++] = m[row * 4 + 2]; + value[i++] = m[row * 4 + 3]; + } + } + else { + for (i = 0, row = firstRow; row <= lastRow; row++) { + value[i++] = m[row + 0]; + value[i++] = m[row + 4]; + value[i++] = m[row + 8]; + value[i++] = m[row + 12]; + } + } + } + return; + case STATE_DEPTH_RANGE: + value[0] = ctx->Viewport.Near; /* near */ + value[1] = ctx->Viewport.Far; /* far */ + value[2] = ctx->Viewport.Far - ctx->Viewport.Near; /* far - near */ + value[3] = 1.0; + return; + case STATE_FRAGMENT_PROGRAM: + { + /* state[1] = {STATE_ENV, STATE_LOCAL} */ + /* state[2] = parameter index */ + const int idx = (int) state[2]; + switch (state[1]) { + case STATE_ENV: + COPY_4V(value, ctx->FragmentProgram.Parameters[idx]); + return; + case STATE_LOCAL: + COPY_4V(value, ctx->FragmentProgram.Current->Base.LocalParams[idx]); + return; + default: + _mesa_problem(ctx, "Bad state switch in _mesa_fetch_state()"); + return; + } + } + return; + + case STATE_VERTEX_PROGRAM: + { + /* state[1] = {STATE_ENV, STATE_LOCAL} */ + /* state[2] = parameter index */ + const int idx = (int) state[2]; + switch (state[1]) { + case STATE_ENV: + COPY_4V(value, ctx->VertexProgram.Parameters[idx]); + return; + case STATE_LOCAL: + COPY_4V(value, ctx->VertexProgram.Current->Base.LocalParams[idx]); + return; + default: + _mesa_problem(ctx, "Bad state switch in _mesa_fetch_state()"); + return; + } + } + return; + + case STATE_NORMAL_SCALE: + ASSIGN_4V(value, ctx->_ModelViewInvScale, 0, 0, 1); + return; + + case STATE_INTERNAL: + switch (state[1]) { + case STATE_CURRENT_ATTRIB: + { + const GLuint idx = (GLuint) state[2]; + COPY_4V(value, ctx->Current.Attrib[idx]); + } + return; + + case STATE_NORMAL_SCALE: + ASSIGN_4V(value, + ctx->_ModelViewInvScale, + ctx->_ModelViewInvScale, + ctx->_ModelViewInvScale, + 1); + return; + + case STATE_TEXRECT_SCALE: + /* Value = { 1/texWidth, 1/texHeight, 0, 1 }. + * Used to convert unnormalized texcoords to normalized texcoords. + */ + { + const int unit = (int) state[2]; + const struct gl_texture_object *texObj + = ctx->Texture.Unit[unit]._Current; + if (texObj) { + struct gl_texture_image *texImage = texObj->Image[0][0]; + ASSIGN_4V(value, + (GLfloat) (1.0 / texImage->Width), + (GLfloat) (1.0 / texImage->Height), + 0.0f, 1.0f); + } + } + return; + + case STATE_FOG_PARAMS_OPTIMIZED: + /* for simpler per-vertex/pixel fog calcs. POW (for EXP/EXP2 fog) + * might be more expensive than EX2 on some hw, plus it needs + * another constant (e) anyway. Linear fog can now be done with a + * single MAD. + * linear: fogcoord * -1/(end-start) + end/(end-start) + * exp: 2^-(density/ln(2) * fogcoord) + * exp2: 2^-((density/(ln(2)^2) * fogcoord)^2) + */ + value[0] = (ctx->Fog.End == ctx->Fog.Start) + ? 1.0f : (GLfloat)(-1.0F / (ctx->Fog.End - ctx->Fog.Start)); + value[1] = ctx->Fog.End * -value[0]; + value[2] = (GLfloat)(ctx->Fog.Density * ONE_DIV_LN2); + value[3] = (GLfloat)(ctx->Fog.Density * ONE_DIV_SQRT_LN2); + return; + + case STATE_POINT_SIZE_CLAMPED: + { + /* this includes implementation dependent limits, to avoid + * another potentially necessary clamp. + * Note: for sprites, point smooth (point AA) is ignored + * and we'll clamp to MinPointSizeAA and MaxPointSize, because we + * expect drivers will want to say their minimum for AA size is 0.0 + * but for non-AA it's 1.0 (because normal points with size below 1.0 + * need to get rounded up to 1.0, hence never disappear). GL does + * not specify max clamp size for sprites, other than it needs to be + * at least as large as max AA size, hence use non-AA size there. + */ + GLfloat minImplSize; + GLfloat maxImplSize; + if (ctx->Point.PointSprite) { + minImplSize = ctx->Const.MinPointSizeAA; + maxImplSize = ctx->Const.MaxPointSize; + } + else if (ctx->Point.SmoothFlag || ctx->Multisample._Enabled) { + minImplSize = ctx->Const.MinPointSizeAA; + maxImplSize = ctx->Const.MaxPointSizeAA; + } + else { + minImplSize = ctx->Const.MinPointSize; + maxImplSize = ctx->Const.MaxPointSize; + } + value[0] = ctx->Point.Size; + value[1] = ctx->Point.MinSize >= minImplSize ? ctx->Point.MinSize : minImplSize; + value[2] = ctx->Point.MaxSize <= maxImplSize ? ctx->Point.MaxSize : maxImplSize; + value[3] = ctx->Point.Threshold; + } + return; + case STATE_POINT_SIZE_IMPL_CLAMP: + { + /* for implementation clamp only in vs */ + GLfloat minImplSize; + GLfloat maxImplSize; + if (ctx->Point.PointSprite) { + minImplSize = ctx->Const.MinPointSizeAA; + maxImplSize = ctx->Const.MaxPointSize; + } + else if (ctx->Point.SmoothFlag || ctx->Multisample._Enabled) { + minImplSize = ctx->Const.MinPointSizeAA; + maxImplSize = ctx->Const.MaxPointSizeAA; + } + else { + minImplSize = ctx->Const.MinPointSize; + maxImplSize = ctx->Const.MaxPointSize; + } + value[0] = ctx->Point.Size; + value[1] = minImplSize; + value[2] = maxImplSize; + value[3] = ctx->Point.Threshold; + } + return; + case STATE_LIGHT_SPOT_DIR_NORMALIZED: + { + /* here, state[2] is the light number */ + /* pre-normalize spot dir */ + const GLuint ln = (GLuint) state[2]; + COPY_3V(value, ctx->Light.Light[ln]._NormSpotDirection); + value[3] = ctx->Light.Light[ln]._CosCutoff; + } + return; + + case STATE_LIGHT_POSITION: + { + const GLuint ln = (GLuint) state[2]; + COPY_4V(value, ctx->Light.Light[ln]._Position); + } + return; + + case STATE_LIGHT_POSITION_NORMALIZED: + { + const GLuint ln = (GLuint) state[2]; + COPY_4V(value, ctx->Light.Light[ln]._Position); + NORMALIZE_3FV( value ); + } + return; + + case STATE_LIGHT_HALF_VECTOR: + { + const GLuint ln = (GLuint) state[2]; + GLfloat p[3]; + /* Compute infinite half angle vector: + * halfVector = normalize(normalize(lightPos) + (0, 0, 1)) + * light.EyePosition.w should be 0 for infinite lights. + */ + COPY_3V(p, ctx->Light.Light[ln]._Position); + NORMALIZE_3FV(p); + ADD_3V(value, p, ctx->_EyeZDir); + NORMALIZE_3FV(value); + value[3] = 1.0; + } + return; + + case STATE_PT_SCALE: + value[0] = ctx->Pixel.RedScale; + value[1] = ctx->Pixel.GreenScale; + value[2] = ctx->Pixel.BlueScale; + value[3] = ctx->Pixel.AlphaScale; + return; + + case STATE_PT_BIAS: + value[0] = ctx->Pixel.RedBias; + value[1] = ctx->Pixel.GreenBias; + value[2] = ctx->Pixel.BlueBias; + value[3] = ctx->Pixel.AlphaBias; + return; + + case STATE_PCM_SCALE: + COPY_4V(value, ctx->Pixel.PostColorMatrixScale); + return; + + case STATE_PCM_BIAS: + COPY_4V(value, ctx->Pixel.PostColorMatrixBias); + return; + + case STATE_SHADOW_AMBIENT: + { + const int unit = (int) state[2]; + const struct gl_texture_object *texObj + = ctx->Texture.Unit[unit]._Current; + if (texObj) { + value[0] = + value[1] = + value[2] = + value[3] = texObj->CompareFailValue; + } + } + return; + + case STATE_FB_SIZE: + value[0] = (GLfloat) (ctx->DrawBuffer->Width - 1); + value[1] = (GLfloat) (ctx->DrawBuffer->Height - 1); + value[2] = 0.0F; + value[3] = 0.0F; + return; + + case STATE_ROT_MATRIX_0: + { + const int unit = (int) state[2]; + GLfloat *rotMat22 = ctx->Texture.Unit[unit].RotMatrix; + value[0] = rotMat22[0]; + value[1] = rotMat22[2]; + value[2] = 0.0; + value[3] = 0.0; + } + return; + + case STATE_ROT_MATRIX_1: + { + const int unit = (int) state[2]; + GLfloat *rotMat22 = ctx->Texture.Unit[unit].RotMatrix; + value[0] = rotMat22[1]; + value[1] = rotMat22[3]; + value[2] = 0.0; + value[3] = 0.0; + } + return; + + /* XXX: make sure new tokens added here are also handled in the + * _mesa_program_state_flags() switch, below. + */ + default: + /* Unknown state indexes are silently ignored here. + * Drivers may do something special. + */ + return; + } + return; + + default: + _mesa_problem(ctx, "Invalid state in _mesa_fetch_state"); + return; + } +} + + +/** + * Return a bitmask of the Mesa state flags (_NEW_* values) which would + * indicate that the given context state may have changed. + * The bitmask is used during validation to determine if we need to update + * vertex/fragment program parameters (like "state.material.color") when + * some GL state has changed. + */ +GLbitfield +_mesa_program_state_flags(const gl_state_index state[STATE_LENGTH]) +{ + switch (state[0]) { + case STATE_MATERIAL: + case STATE_LIGHT: + case STATE_LIGHTMODEL_AMBIENT: + case STATE_LIGHTMODEL_SCENECOLOR: + case STATE_LIGHTPROD: + return _NEW_LIGHT; + + case STATE_TEXGEN: + case STATE_TEXENV_COLOR: + return _NEW_TEXTURE; + + case STATE_FOG_COLOR: + case STATE_FOG_PARAMS: + return _NEW_FOG; + + case STATE_CLIPPLANE: + return _NEW_TRANSFORM; + + case STATE_POINT_SIZE: + case STATE_POINT_ATTENUATION: + return _NEW_POINT; + + case STATE_MODELVIEW_MATRIX: + return _NEW_MODELVIEW; + case STATE_PROJECTION_MATRIX: + return _NEW_PROJECTION; + case STATE_MVP_MATRIX: + return _NEW_MODELVIEW | _NEW_PROJECTION; + case STATE_TEXTURE_MATRIX: + return _NEW_TEXTURE_MATRIX; + case STATE_PROGRAM_MATRIX: + return _NEW_TRACK_MATRIX; + case STATE_COLOR_MATRIX: + return _NEW_COLOR_MATRIX; + + case STATE_DEPTH_RANGE: + return _NEW_VIEWPORT; + + case STATE_FRAGMENT_PROGRAM: + case STATE_VERTEX_PROGRAM: + return _NEW_PROGRAM; + + case STATE_NORMAL_SCALE: + return _NEW_MODELVIEW; + + case STATE_INTERNAL: + switch (state[1]) { + case STATE_CURRENT_ATTRIB: + return _NEW_CURRENT_ATTRIB; + + case STATE_NORMAL_SCALE: + return _NEW_MODELVIEW; + + case STATE_TEXRECT_SCALE: + case STATE_SHADOW_AMBIENT: + case STATE_ROT_MATRIX_0: + case STATE_ROT_MATRIX_1: + return _NEW_TEXTURE; + case STATE_FOG_PARAMS_OPTIMIZED: + return _NEW_FOG; + case STATE_POINT_SIZE_CLAMPED: + case STATE_POINT_SIZE_IMPL_CLAMP: + return _NEW_POINT | _NEW_MULTISAMPLE; + case STATE_LIGHT_SPOT_DIR_NORMALIZED: + case STATE_LIGHT_POSITION: + case STATE_LIGHT_POSITION_NORMALIZED: + case STATE_LIGHT_HALF_VECTOR: + return _NEW_LIGHT; + + case STATE_PT_SCALE: + case STATE_PT_BIAS: + case STATE_PCM_SCALE: + case STATE_PCM_BIAS: + return _NEW_PIXEL; + + case STATE_FB_SIZE: + return _NEW_BUFFERS; + + default: + /* unknown state indexes are silently ignored and + * no flag set, since it is handled by the driver. + */ + return 0; + } + + default: + _mesa_problem(NULL, "unexpected state[0] in make_state_flags()"); + return 0; + } +} + + +static void +append(char *dst, const char *src) +{ + while (*dst) + dst++; + while (*src) + *dst++ = *src++; + *dst = 0; +} + + +/** + * Convert token 'k' to a string, append it onto 'dst' string. + */ +static void +append_token(char *dst, gl_state_index k) +{ + switch (k) { + case STATE_MATERIAL: + append(dst, "material"); + break; + case STATE_LIGHT: + append(dst, "light"); + break; + case STATE_LIGHTMODEL_AMBIENT: + append(dst, "lightmodel.ambient"); + break; + case STATE_LIGHTMODEL_SCENECOLOR: + break; + case STATE_LIGHTPROD: + append(dst, "lightprod"); + break; + case STATE_TEXGEN: + append(dst, "texgen"); + break; + case STATE_FOG_COLOR: + append(dst, "fog.color"); + break; + case STATE_FOG_PARAMS: + append(dst, "fog.params"); + break; + case STATE_CLIPPLANE: + append(dst, "clip"); + break; + case STATE_POINT_SIZE: + append(dst, "point.size"); + break; + case STATE_POINT_ATTENUATION: + append(dst, "point.attenuation"); + break; + case STATE_MODELVIEW_MATRIX: + append(dst, "matrix.modelview"); + break; + case STATE_PROJECTION_MATRIX: + append(dst, "matrix.projection"); + break; + case STATE_MVP_MATRIX: + append(dst, "matrix.mvp"); + break; + case STATE_TEXTURE_MATRIX: + append(dst, "matrix.texture"); + break; + case STATE_PROGRAM_MATRIX: + append(dst, "matrix.program"); + break; + case STATE_COLOR_MATRIX: + append(dst, "matrix.color"); + break; + case STATE_MATRIX_INVERSE: + append(dst, ".inverse"); + break; + case STATE_MATRIX_TRANSPOSE: + append(dst, ".transpose"); + break; + case STATE_MATRIX_INVTRANS: + append(dst, ".invtrans"); + break; + case STATE_AMBIENT: + append(dst, ".ambient"); + break; + case STATE_DIFFUSE: + append(dst, ".diffuse"); + break; + case STATE_SPECULAR: + append(dst, ".specular"); + break; + case STATE_EMISSION: + append(dst, ".emission"); + break; + case STATE_SHININESS: + append(dst, "lshininess"); + break; + case STATE_HALF_VECTOR: + append(dst, ".half"); + break; + case STATE_POSITION: + append(dst, ".position"); + break; + case STATE_ATTENUATION: + append(dst, ".attenuation"); + break; + case STATE_SPOT_DIRECTION: + append(dst, ".spot.direction"); + break; + case STATE_SPOT_CUTOFF: + append(dst, ".spot.cutoff"); + break; + case STATE_TEXGEN_EYE_S: + append(dst, ".eye.s"); + break; + case STATE_TEXGEN_EYE_T: + append(dst, ".eye.t"); + break; + case STATE_TEXGEN_EYE_R: + append(dst, ".eye.r"); + break; + case STATE_TEXGEN_EYE_Q: + append(dst, ".eye.q"); + break; + case STATE_TEXGEN_OBJECT_S: + append(dst, ".object.s"); + break; + case STATE_TEXGEN_OBJECT_T: + append(dst, ".object.t"); + break; + case STATE_TEXGEN_OBJECT_R: + append(dst, ".object.r"); + break; + case STATE_TEXGEN_OBJECT_Q: + append(dst, ".object.q"); + break; + case STATE_TEXENV_COLOR: + append(dst, "texenv"); + break; + case STATE_DEPTH_RANGE: + append(dst, "depth.range"); + break; + case STATE_VERTEX_PROGRAM: + case STATE_FRAGMENT_PROGRAM: + break; + case STATE_ENV: + append(dst, "env"); + break; + case STATE_LOCAL: + append(dst, "local"); + break; + /* BEGIN internal state vars */ + case STATE_INTERNAL: + append(dst, ".internal."); + break; + case STATE_CURRENT_ATTRIB: + append(dst, "current"); + break; + case STATE_NORMAL_SCALE: + append(dst, "normalScale"); + break; + case STATE_TEXRECT_SCALE: + append(dst, "texrectScale"); + break; + case STATE_FOG_PARAMS_OPTIMIZED: + append(dst, "fogParamsOptimized"); + break; + case STATE_POINT_SIZE_CLAMPED: + append(dst, "pointSizeClamped"); + break; + case STATE_POINT_SIZE_IMPL_CLAMP: + append(dst, "pointSizeImplClamp"); + break; + case STATE_LIGHT_SPOT_DIR_NORMALIZED: + append(dst, "lightSpotDirNormalized"); + break; + case STATE_LIGHT_POSITION: + append(dst, "lightPosition"); + break; + case STATE_LIGHT_POSITION_NORMALIZED: + append(dst, "light.position.normalized"); + break; + case STATE_LIGHT_HALF_VECTOR: + append(dst, "lightHalfVector"); + break; + case STATE_PT_SCALE: + append(dst, "PTscale"); + break; + case STATE_PT_BIAS: + append(dst, "PTbias"); + break; + case STATE_PCM_SCALE: + append(dst, "PCMscale"); + break; + case STATE_PCM_BIAS: + append(dst, "PCMbias"); + break; + case STATE_SHADOW_AMBIENT: + append(dst, "CompareFailValue"); + break; + case STATE_FB_SIZE: + append(dst, "FbSize"); + break; + case STATE_ROT_MATRIX_0: + append(dst, "rotMatrixRow0"); + break; + case STATE_ROT_MATRIX_1: + append(dst, "rotMatrixRow1"); + break; + default: + /* probably STATE_INTERNAL_DRIVER+i (driver private state) */ + append(dst, "driverState"); + } +} + +static void +append_face(char *dst, GLint face) +{ + if (face == 0) + append(dst, "front."); + else + append(dst, "back."); +} + +static void +append_index(char *dst, GLint index) +{ + char s[20]; + sprintf(s, "[%d]", index); + append(dst, s); +} + +/** + * Make a string from the given state vector. + * For example, return "state.matrix.texture[2].inverse". + * Use free() to deallocate the string. + */ +char * +_mesa_program_state_string(const gl_state_index state[STATE_LENGTH]) +{ + char str[1000] = ""; + char tmp[30]; + + append(str, "state."); + append_token(str, state[0]); + + switch (state[0]) { + case STATE_MATERIAL: + append_face(str, state[1]); + append_token(str, state[2]); + break; + case STATE_LIGHT: + append_index(str, state[1]); /* light number [i]. */ + append_token(str, state[2]); /* coefficients */ + break; + case STATE_LIGHTMODEL_AMBIENT: + append(str, "lightmodel.ambient"); + break; + case STATE_LIGHTMODEL_SCENECOLOR: + if (state[1] == 0) { + append(str, "lightmodel.front.scenecolor"); + } + else { + append(str, "lightmodel.back.scenecolor"); + } + break; + case STATE_LIGHTPROD: + append_index(str, state[1]); /* light number [i]. */ + append_face(str, state[2]); + append_token(str, state[3]); + break; + case STATE_TEXGEN: + append_index(str, state[1]); /* tex unit [i] */ + append_token(str, state[2]); /* plane coef */ + break; + case STATE_TEXENV_COLOR: + append_index(str, state[1]); /* tex unit [i] */ + append(str, "color"); + break; + case STATE_CLIPPLANE: + append_index(str, state[1]); /* plane [i] */ + append(str, ".plane"); + break; + case STATE_MODELVIEW_MATRIX: + case STATE_PROJECTION_MATRIX: + case STATE_MVP_MATRIX: + case STATE_TEXTURE_MATRIX: + case STATE_PROGRAM_MATRIX: + case STATE_COLOR_MATRIX: + { + /* state[0] = modelview, projection, texture, etc. */ + /* state[1] = which texture matrix or program matrix */ + /* state[2] = first row to fetch */ + /* state[3] = last row to fetch */ + /* state[4] = transpose, inverse or invtrans */ + const gl_state_index mat = state[0]; + const GLuint index = (GLuint) state[1]; + const GLuint firstRow = (GLuint) state[2]; + const GLuint lastRow = (GLuint) state[3]; + const gl_state_index modifier = state[4]; + if (index || + mat == STATE_TEXTURE_MATRIX || + mat == STATE_PROGRAM_MATRIX) + append_index(str, index); + if (modifier) + append_token(str, modifier); + if (firstRow == lastRow) + sprintf(tmp, ".row[%d]", firstRow); + else + sprintf(tmp, ".row[%d..%d]", firstRow, lastRow); + append(str, tmp); + } + break; + case STATE_POINT_SIZE: + break; + case STATE_POINT_ATTENUATION: + break; + case STATE_FOG_PARAMS: + break; + case STATE_FOG_COLOR: + break; + case STATE_DEPTH_RANGE: + break; + case STATE_FRAGMENT_PROGRAM: + case STATE_VERTEX_PROGRAM: + /* state[1] = {STATE_ENV, STATE_LOCAL} */ + /* state[2] = parameter index */ + append_token(str, state[1]); + append_index(str, state[2]); + break; + case STATE_NORMAL_SCALE: + break; + case STATE_INTERNAL: + append_token(str, state[1]); + if (state[1] == STATE_CURRENT_ATTRIB) + append_index(str, state[2]); + break; + default: + _mesa_problem(NULL, "Invalid state in _mesa_program_state_string"); + break; + } + + return _mesa_strdup(str); +} + + +/** + * Loop over all the parameters in a parameter list. If the parameter + * is a GL state reference, look up the current value of that state + * variable and put it into the parameter's Value[4] array. + * This would be called at glBegin time when using a fragment program. + */ +void +_mesa_load_state_parameters(GLcontext *ctx, + struct gl_program_parameter_list *paramList) +{ + GLuint i; + + if (!paramList) + return; + + /*assert(ctx->Driver.NeedFlush == 0);*/ + + for (i = 0; i < paramList->NumParameters; i++) { + if (paramList->Parameters[i].Type == PROGRAM_STATE_VAR) { + _mesa_fetch_state(ctx, + (gl_state_index *) paramList->Parameters[i].StateIndexes, + paramList->ParameterValues[i]); + } + } +} + + +/** + * Copy the 16 elements of a matrix into four consecutive program + * registers starting at 'pos'. + */ +static void +load_matrix(GLfloat registers[][4], GLuint pos, const GLfloat mat[16]) +{ + GLuint i; + for (i = 0; i < 4; i++) { + registers[pos + i][0] = mat[0 + i]; + registers[pos + i][1] = mat[4 + i]; + registers[pos + i][2] = mat[8 + i]; + registers[pos + i][3] = mat[12 + i]; + } +} + + +/** + * As above, but transpose the matrix. + */ +static void +load_transpose_matrix(GLfloat registers[][4], GLuint pos, + const GLfloat mat[16]) +{ + memcpy(registers[pos], mat, 16 * sizeof(GLfloat)); +} + + +/** + * Load current vertex program's parameter registers with tracked + * matrices (if NV program). This only needs to be done per + * glBegin/glEnd, not per-vertex. + */ +void +_mesa_load_tracked_matrices(GLcontext *ctx) +{ + GLuint i; + + for (i = 0; i < MAX_NV_VERTEX_PROGRAM_PARAMS / 4; i++) { + /* point 'mat' at source matrix */ + GLmatrix *mat; + if (ctx->VertexProgram.TrackMatrix[i] == GL_MODELVIEW) { + mat = ctx->ModelviewMatrixStack.Top; + } + else if (ctx->VertexProgram.TrackMatrix[i] == GL_PROJECTION) { + mat = ctx->ProjectionMatrixStack.Top; + } + else if (ctx->VertexProgram.TrackMatrix[i] == GL_TEXTURE) { + GLuint unit = MIN2(ctx->Texture.CurrentUnit, + Elements(ctx->TextureMatrixStack) - 1); + mat = ctx->TextureMatrixStack[unit].Top; + } + else if (ctx->VertexProgram.TrackMatrix[i] == GL_COLOR) { + mat = ctx->ColorMatrixStack.Top; + } + else if (ctx->VertexProgram.TrackMatrix[i]==GL_MODELVIEW_PROJECTION_NV) { + /* XXX verify the combined matrix is up to date */ + mat = &ctx->_ModelProjectMatrix; + } + else if (ctx->VertexProgram.TrackMatrix[i] >= GL_MATRIX0_NV && + ctx->VertexProgram.TrackMatrix[i] <= GL_MATRIX7_NV) { + GLuint n = ctx->VertexProgram.TrackMatrix[i] - GL_MATRIX0_NV; + ASSERT(n < Elements(ctx->ProgramMatrixStack)); + mat = ctx->ProgramMatrixStack[n].Top; + } + else { + /* no matrix is tracked, but we leave the register values as-is */ + assert(ctx->VertexProgram.TrackMatrix[i] == GL_NONE); + continue; + } + + /* load the matrix values into sequential registers */ + if (ctx->VertexProgram.TrackMatrixTransform[i] == GL_IDENTITY_NV) { + load_matrix(ctx->VertexProgram.Parameters, i*4, mat->m); + } + else if (ctx->VertexProgram.TrackMatrixTransform[i] == GL_INVERSE_NV) { + _math_matrix_analyse(mat); /* update the inverse */ + ASSERT(!_math_matrix_is_dirty(mat)); + load_matrix(ctx->VertexProgram.Parameters, i*4, mat->inv); + } + else if (ctx->VertexProgram.TrackMatrixTransform[i] == GL_TRANSPOSE_NV) { + load_transpose_matrix(ctx->VertexProgram.Parameters, i*4, mat->m); + } + else { + assert(ctx->VertexProgram.TrackMatrixTransform[i] + == GL_INVERSE_TRANSPOSE_NV); + _math_matrix_analyse(mat); /* update the inverse */ + ASSERT(!_math_matrix_is_dirty(mat)); + load_transpose_matrix(ctx->VertexProgram.Parameters, i*4, mat->inv); + } + } +} diff --git a/mesalib/src/mesa/program/prog_statevars.h b/mesalib/src/mesa/program/prog_statevars.h new file mode 100644 index 000000000..1753471ff --- /dev/null +++ b/mesalib/src/mesa/program/prog_statevars.h @@ -0,0 +1,147 @@ +/* + * Mesa 3-D graphics library + * Version: 7.1 + * + * Copyright (C) 1999-2007 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 PROG_STATEVARS_H +#define PROG_STATEVARS_H + +#include "main/mtypes.h" + + +/** + * Number of STATE_* values we need to address any GL state. + * Used to dimension arrays. + */ +#define STATE_LENGTH 5 + + +/** + * Used for describing GL state referenced from inside ARB vertex and + * fragment programs. + * A string such as "state.light[0].ambient" gets translated into a + * sequence of tokens such as [ STATE_LIGHT, 0, STATE_AMBIENT ]. + * + * For state that's an array, like STATE_CLIPPLANE, the 2nd token [1] should + * always be the array index. + */ +typedef enum gl_state_index_ { + STATE_MATERIAL = 100, /* start at 100 so small ints are seen as ints */ + + STATE_LIGHT, + STATE_LIGHTMODEL_AMBIENT, + STATE_LIGHTMODEL_SCENECOLOR, + STATE_LIGHTPROD, + + STATE_TEXGEN, + + STATE_FOG_COLOR, + STATE_FOG_PARAMS, + + STATE_CLIPPLANE, + + STATE_POINT_SIZE, + STATE_POINT_ATTENUATION, + + STATE_MODELVIEW_MATRIX, + STATE_PROJECTION_MATRIX, + STATE_MVP_MATRIX, + STATE_TEXTURE_MATRIX, + STATE_PROGRAM_MATRIX, + STATE_COLOR_MATRIX, + STATE_MATRIX_INVERSE, + STATE_MATRIX_TRANSPOSE, + STATE_MATRIX_INVTRANS, + + STATE_AMBIENT, + STATE_DIFFUSE, + STATE_SPECULAR, + STATE_EMISSION, + STATE_SHININESS, + STATE_HALF_VECTOR, + + STATE_POSITION, /**< xyzw = position */ + STATE_ATTENUATION, /**< xyz = attenuation, w = spot exponent */ + STATE_SPOT_DIRECTION, /**< xyz = direction, w = cos(cutoff) */ + STATE_SPOT_CUTOFF, /**< x = cutoff, yzw = undefined */ + + STATE_TEXGEN_EYE_S, + STATE_TEXGEN_EYE_T, + STATE_TEXGEN_EYE_R, + STATE_TEXGEN_EYE_Q, + STATE_TEXGEN_OBJECT_S, + STATE_TEXGEN_OBJECT_T, + STATE_TEXGEN_OBJECT_R, + STATE_TEXGEN_OBJECT_Q, + + STATE_TEXENV_COLOR, + + STATE_DEPTH_RANGE, + + STATE_VERTEX_PROGRAM, + STATE_FRAGMENT_PROGRAM, + + STATE_ENV, + STATE_LOCAL, + + STATE_INTERNAL, /* Mesa additions */ + STATE_CURRENT_ATTRIB, /* ctx->Current vertex attrib value */ + STATE_NORMAL_SCALE, + STATE_TEXRECT_SCALE, + STATE_FOG_PARAMS_OPTIMIZED, /* for faster fog calc */ + STATE_POINT_SIZE_CLAMPED, /* includes implementation dependent size clamp */ + STATE_POINT_SIZE_IMPL_CLAMP, /* for implementation clamp only in vs */ + STATE_LIGHT_SPOT_DIR_NORMALIZED, /* pre-normalized spot dir */ + STATE_LIGHT_POSITION, /* object vs eye space */ + STATE_LIGHT_POSITION_NORMALIZED, /* object vs eye space */ + STATE_LIGHT_HALF_VECTOR, /* object vs eye space */ + STATE_PT_SCALE, /**< Pixel transfer RGBA scale */ + STATE_PT_BIAS, /**< Pixel transfer RGBA bias */ + STATE_PCM_SCALE, /**< Post color matrix RGBA scale */ + STATE_PCM_BIAS, /**< Post color matrix RGBA bias */ + STATE_SHADOW_AMBIENT, /**< ARB_shadow_ambient fail value; token[2] is texture unit index */ + STATE_FB_SIZE, /**< (width-1, height-1, 0, 0) */ + STATE_ROT_MATRIX_0, /**< ATI_envmap_bumpmap, rot matrix row 0 */ + STATE_ROT_MATRIX_1, /**< ATI_envmap_bumpmap, rot matrix row 1 */ + STATE_INTERNAL_DRIVER /* first available state index for drivers (must be last) */ +} gl_state_index; + + + +extern void +_mesa_load_state_parameters(GLcontext *ctx, + struct gl_program_parameter_list *paramList); + + +extern GLbitfield +_mesa_program_state_flags(const gl_state_index state[STATE_LENGTH]); + + +extern char * +_mesa_program_state_string(const gl_state_index state[STATE_LENGTH]); + + +extern void +_mesa_load_tracked_matrices(GLcontext *ctx); + + +#endif /* PROG_STATEVARS_H */ diff --git a/mesalib/src/mesa/program/prog_uniform.c b/mesalib/src/mesa/program/prog_uniform.c new file mode 100644 index 000000000..28acb8871 --- /dev/null +++ b/mesalib/src/mesa/program/prog_uniform.c @@ -0,0 +1,177 @@ +/* + * Mesa 3-D graphics library + * Version: 7.1 + * + * Copyright (C) 1999-2008 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. + */ + +/** + * \file prog_uniform.c + * Shader uniform functions. + * \author Brian Paul + */ + +#include "main/imports.h" +#include "main/mtypes.h" +#include "prog_uniform.h" + + +struct gl_uniform_list * +_mesa_new_uniform_list(void) +{ + return CALLOC_STRUCT(gl_uniform_list); +} + + +void +_mesa_free_uniform_list(struct gl_uniform_list *list) +{ + GLuint i; + + if (!list) + return; + + for (i = 0; i < list->NumUniforms; i++) { + free((void *) list->Uniforms[i].Name); + } + free(list->Uniforms); + free(list); +} + + +struct gl_uniform * +_mesa_append_uniform(struct gl_uniform_list *list, + const char *name, GLenum target, GLuint progPos) +{ + const GLuint oldNum = list->NumUniforms; + struct gl_uniform *uniform; + GLint index; + + assert(target == GL_VERTEX_PROGRAM_ARB || + target == GL_FRAGMENT_PROGRAM_ARB || + target == MESA_GEOMETRY_PROGRAM); + + index = _mesa_lookup_uniform(list, name); + if (index < 0) { + /* not found - append to list */ + + if (oldNum + 1 > list->Size) { + /* Need to grow the list array (alloc some extra) */ + list->Size += 4; + + /* realloc arrays */ + list->Uniforms = (struct gl_uniform *) + _mesa_realloc(list->Uniforms, + oldNum * sizeof(struct gl_uniform), + list->Size * sizeof(struct gl_uniform)); + } + + if (!list->Uniforms) { + /* out of memory */ + list->NumUniforms = 0; + list->Size = 0; + return GL_FALSE; + } + + uniform = list->Uniforms + oldNum; + + uniform->Name = _mesa_strdup(name); + uniform->VertPos = -1; + uniform->FragPos = -1; + uniform->GeomPos = -1; + uniform->Initialized = GL_FALSE; + + list->NumUniforms++; + } + else { + /* found */ + uniform = list->Uniforms + index; + } + + /* update position for the vertex or fragment program */ + if (target == GL_VERTEX_PROGRAM_ARB) { + if (uniform->VertPos != -1) { + /* this uniform is already in the list - that shouldn't happen */ + return GL_FALSE; + } + uniform->VertPos = progPos; + } else if (target == GL_FRAGMENT_PROGRAM_ARB) { + if (uniform->FragPos != -1) { + /* this uniform is already in the list - that shouldn't happen */ + return GL_FALSE; + } + uniform->FragPos = progPos; + } else { + if (uniform->GeomPos != -1) { + /* this uniform is already in the list - that shouldn't happen */ + return GL_FALSE; + } + uniform->GeomPos = progPos; + } + + return uniform; +} + + +/** + * Return the location/index of the named uniform in the uniform list, + * or -1 if not found. + */ +GLint +_mesa_lookup_uniform(const struct gl_uniform_list *list, const char *name) +{ + GLuint i; + for (i = 0; list && i < list->NumUniforms; i++) { + if (!strcmp(list->Uniforms[i].Name, name)) { + return i; + } + } + return -1; +} + + +GLint +_mesa_longest_uniform_name(const struct gl_uniform_list *list) +{ + GLint max = 0; + GLuint i; + for (i = 0; list && i < list->NumUniforms; i++) { + GLint len = (GLint) strlen(list->Uniforms[i].Name); + if (len > max) + max = len; + } + return max; +} + + +void +_mesa_print_uniforms(const struct gl_uniform_list *list) +{ + GLuint i; + printf("Uniform list %p:\n", (void *) list); + for (i = 0; i < list->NumUniforms; i++) { + printf("%d: %s %d %d %d\n", + i, + list->Uniforms[i].Name, + list->Uniforms[i].VertPos, + list->Uniforms[i].FragPos, + list->Uniforms[i].GeomPos); + } +} diff --git a/mesalib/src/mesa/program/prog_uniform.h b/mesalib/src/mesa/program/prog_uniform.h new file mode 100644 index 000000000..67f78006e --- /dev/null +++ b/mesalib/src/mesa/program/prog_uniform.h @@ -0,0 +1,89 @@ +/* + * Mesa 3-D graphics library + * Version: 7.1 + * + * Copyright (C) 1999-2008 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. + */ + +/** + * \file prog_uniform.c + * Shader uniform functions. + * \author Brian Paul + */ + +#ifndef PROG_UNIFORM_H +#define PROG_UNIFORM_H + +#include "main/glheader.h" + + +/** + * Shader program uniform variable. + * The glGetUniformLocation() and glUniform() commands will use this + * information. + * Note that a uniform such as "binormal" might be used in both the + * vertex shader and the fragment shader. When glUniform() is called to + * set the uniform's value, it must be updated in both the vertex and + * fragment shaders. The uniform may be in different locations in the + * two shaders so we keep track of that here. + */ +struct gl_uniform +{ + const char *Name; /**< Null-terminated string */ + GLint VertPos; + GLint FragPos; + GLint GeomPos; + GLboolean Initialized; /**< For debug. Has this uniform been set? */ + const struct glsl_type *Type; +}; + + +/** + * List of gl_uniforms + */ +struct gl_uniform_list +{ + GLuint Size; /**< allocated size of Uniforms array */ + GLuint NumUniforms; /**< number of uniforms in the array */ + struct gl_uniform *Uniforms; /**< Array [Size] */ +}; + + +extern struct gl_uniform_list * +_mesa_new_uniform_list(void); + +extern void +_mesa_free_uniform_list(struct gl_uniform_list *list); + +extern struct gl_uniform * +_mesa_append_uniform(struct gl_uniform_list *list, + const char *name, GLenum target, GLuint progPos); + +extern GLint +_mesa_lookup_uniform(const struct gl_uniform_list *list, const char *name); + +extern GLint +_mesa_longest_uniform_name(const struct gl_uniform_list *list); + +extern void +_mesa_print_uniforms(const struct gl_uniform_list *list); + + +#endif /* PROG_UNIFORM_H */ diff --git a/mesalib/src/mesa/program/program.c b/mesalib/src/mesa/program/program.c new file mode 100644 index 000000000..06b9539bd --- /dev/null +++ b/mesalib/src/mesa/program/program.c @@ -0,0 +1,976 @@ +/* + * Mesa 3-D graphics library + * Version: 6.5.3 + * + * Copyright (C) 1999-2007 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. + */ + +/** + * \file program.c + * Vertex and fragment program support functions. + * \author Brian Paul + */ + + +#include "main/glheader.h" +#include "main/context.h" +#include "main/hash.h" +#include "program.h" +#include "prog_cache.h" +#include "prog_parameter.h" +#include "prog_instruction.h" + + +/** + * A pointer to this dummy program is put into the hash table when + * glGenPrograms is called. + */ +struct gl_program _mesa_DummyProgram; + + +/** + * Init context's vertex/fragment program state + */ +void +_mesa_init_program(GLcontext *ctx) +{ + GLuint i; + + /* + * If this assertion fails, we need to increase the field + * size for register indexes (see INST_INDEX_BITS). + */ + ASSERT(ctx->Const.VertexProgram.MaxUniformComponents / 4 + <= (1 << INST_INDEX_BITS)); + ASSERT(ctx->Const.FragmentProgram.MaxUniformComponents / 4 + <= (1 << INST_INDEX_BITS)); + + ASSERT(ctx->Const.VertexProgram.MaxTemps <= (1 << INST_INDEX_BITS)); + ASSERT(ctx->Const.VertexProgram.MaxLocalParams <= (1 << INST_INDEX_BITS)); + ASSERT(ctx->Const.FragmentProgram.MaxTemps <= (1 << INST_INDEX_BITS)); + ASSERT(ctx->Const.FragmentProgram.MaxLocalParams <= (1 << INST_INDEX_BITS)); + + ASSERT(ctx->Const.VertexProgram.MaxUniformComponents <= 4 * MAX_UNIFORMS); + ASSERT(ctx->Const.FragmentProgram.MaxUniformComponents <= 4 * MAX_UNIFORMS); + + /* If this fails, increase prog_instruction::TexSrcUnit size */ + ASSERT(MAX_TEXTURE_UNITS < (1 << 5)); + + /* If this fails, increase prog_instruction::TexSrcTarget size */ + ASSERT(NUM_TEXTURE_TARGETS < (1 << 3)); + + ctx->Program.ErrorPos = -1; + ctx->Program.ErrorString = _mesa_strdup(""); + +#if FEATURE_NV_vertex_program || FEATURE_ARB_vertex_program + ctx->VertexProgram.Enabled = GL_FALSE; +#if FEATURE_es2_glsl + ctx->VertexProgram.PointSizeEnabled = + (ctx->API == API_OPENGLES2) ? GL_TRUE : GL_FALSE; +#else + ctx->VertexProgram.PointSizeEnabled = GL_FALSE; +#endif + ctx->VertexProgram.TwoSideEnabled = GL_FALSE; + _mesa_reference_vertprog(ctx, &ctx->VertexProgram.Current, + ctx->Shared->DefaultVertexProgram); + assert(ctx->VertexProgram.Current); + for (i = 0; i < MAX_NV_VERTEX_PROGRAM_PARAMS / 4; i++) { + ctx->VertexProgram.TrackMatrix[i] = GL_NONE; + ctx->VertexProgram.TrackMatrixTransform[i] = GL_IDENTITY_NV; + } + ctx->VertexProgram.Cache = _mesa_new_program_cache(); +#endif + +#if FEATURE_NV_fragment_program || FEATURE_ARB_fragment_program + ctx->FragmentProgram.Enabled = GL_FALSE; + _mesa_reference_fragprog(ctx, &ctx->FragmentProgram.Current, + ctx->Shared->DefaultFragmentProgram); + assert(ctx->FragmentProgram.Current); + ctx->FragmentProgram.Cache = _mesa_new_program_cache(); +#endif + +#if FEATURE_ARB_geometry_shader4 + ctx->GeometryProgram.Enabled = GL_FALSE; + /* right now by default we don't have a geometry program */ + _mesa_reference_geomprog(ctx, &ctx->GeometryProgram.Current, + NULL); + ctx->GeometryProgram.Cache = _mesa_new_program_cache(); +#endif + + /* XXX probably move this stuff */ +#if FEATURE_ATI_fragment_shader + ctx->ATIFragmentShader.Enabled = GL_FALSE; + ctx->ATIFragmentShader.Current = ctx->Shared->DefaultFragmentShader; + assert(ctx->ATIFragmentShader.Current); + ctx->ATIFragmentShader.Current->RefCount++; +#endif +} + + +/** + * Free a context's vertex/fragment program state + */ +void +_mesa_free_program_data(GLcontext *ctx) +{ +#if FEATURE_NV_vertex_program || FEATURE_ARB_vertex_program + _mesa_reference_vertprog(ctx, &ctx->VertexProgram.Current, NULL); + _mesa_delete_program_cache(ctx, ctx->VertexProgram.Cache); +#endif +#if FEATURE_NV_fragment_program || FEATURE_ARB_fragment_program + _mesa_reference_fragprog(ctx, &ctx->FragmentProgram.Current, NULL); + _mesa_delete_program_cache(ctx, ctx->FragmentProgram.Cache); +#endif +#if FEATURE_ARB_geometry_shader4 + _mesa_reference_geomprog(ctx, &ctx->GeometryProgram.Current, NULL); + _mesa_delete_program_cache(ctx, ctx->GeometryProgram.Cache); +#endif + /* XXX probably move this stuff */ +#if FEATURE_ATI_fragment_shader + if (ctx->ATIFragmentShader.Current) { + ctx->ATIFragmentShader.Current->RefCount--; + if (ctx->ATIFragmentShader.Current->RefCount <= 0) { + free(ctx->ATIFragmentShader.Current); + } + } +#endif + free((void *) ctx->Program.ErrorString); +} + + +/** + * Update the default program objects in the given context to reference those + * specified in the shared state and release those referencing the old + * shared state. + */ +void +_mesa_update_default_objects_program(GLcontext *ctx) +{ +#if FEATURE_NV_vertex_program || FEATURE_ARB_vertex_program + _mesa_reference_vertprog(ctx, &ctx->VertexProgram.Current, + (struct gl_vertex_program *) + ctx->Shared->DefaultVertexProgram); + assert(ctx->VertexProgram.Current); +#endif + +#if FEATURE_NV_fragment_program || FEATURE_ARB_fragment_program + _mesa_reference_fragprog(ctx, &ctx->FragmentProgram.Current, + (struct gl_fragment_program *) + ctx->Shared->DefaultFragmentProgram); + assert(ctx->FragmentProgram.Current); +#endif + +#if FEATURE_ARB_geometry_shader4 + _mesa_reference_geomprog(ctx, &ctx->GeometryProgram.Current, + (struct gl_geometry_program *) + ctx->Shared->DefaultGeometryProgram); +#endif + + /* XXX probably move this stuff */ +#if FEATURE_ATI_fragment_shader + if (ctx->ATIFragmentShader.Current) { + ctx->ATIFragmentShader.Current->RefCount--; + if (ctx->ATIFragmentShader.Current->RefCount <= 0) { + free(ctx->ATIFragmentShader.Current); + } + } + ctx->ATIFragmentShader.Current = (struct ati_fragment_shader *) ctx->Shared->DefaultFragmentShader; + assert(ctx->ATIFragmentShader.Current); + ctx->ATIFragmentShader.Current->RefCount++; +#endif +} + + +/** + * Set the vertex/fragment program error state (position and error string). + * This is generally called from within the parsers. + */ +void +_mesa_set_program_error(GLcontext *ctx, GLint pos, const char *string) +{ + ctx->Program.ErrorPos = pos; + free((void *) ctx->Program.ErrorString); + if (!string) + string = ""; + ctx->Program.ErrorString = _mesa_strdup(string); +} + + +/** + * Find the line number and column for 'pos' within 'string'. + * Return a copy of the line which contains 'pos'. Free the line with + * free(). + * \param string the program string + * \param pos the position within the string + * \param line returns the line number corresponding to 'pos'. + * \param col returns the column number corresponding to 'pos'. + * \return copy of the line containing 'pos'. + */ +const GLubyte * +_mesa_find_line_column(const GLubyte *string, const GLubyte *pos, + GLint *line, GLint *col) +{ + const GLubyte *lineStart = string; + const GLubyte *p = string; + GLubyte *s; + int len; + + *line = 1; + + while (p != pos) { + if (*p == (GLubyte) '\n') { + (*line)++; + lineStart = p + 1; + } + p++; + } + + *col = (pos - lineStart) + 1; + + /* return copy of this line */ + while (*p != 0 && *p != '\n') + p++; + len = p - lineStart; + s = (GLubyte *) malloc(len + 1); + memcpy(s, lineStart, len); + s[len] = 0; + + return s; +} + + +/** + * Initialize a new vertex/fragment program object. + */ +static struct gl_program * +_mesa_init_program_struct( GLcontext *ctx, struct gl_program *prog, + GLenum target, GLuint id) +{ + (void) ctx; + if (prog) { + GLuint i; + memset(prog, 0, sizeof(*prog)); + prog->Id = id; + prog->Target = target; + prog->Resident = GL_TRUE; + prog->RefCount = 1; + prog->Format = GL_PROGRAM_FORMAT_ASCII_ARB; + + /* default mapping from samplers to texture units */ + for (i = 0; i < MAX_SAMPLERS; i++) + prog->SamplerUnits[i] = i; + } + + return prog; +} + + +/** + * Initialize a new fragment program object. + */ +struct gl_program * +_mesa_init_fragment_program( GLcontext *ctx, struct gl_fragment_program *prog, + GLenum target, GLuint id) +{ + if (prog) + return _mesa_init_program_struct( ctx, &prog->Base, target, id ); + else + return NULL; +} + + +/** + * Initialize a new vertex program object. + */ +struct gl_program * +_mesa_init_vertex_program( GLcontext *ctx, struct gl_vertex_program *prog, + GLenum target, GLuint id) +{ + if (prog) + return _mesa_init_program_struct( ctx, &prog->Base, target, id ); + else + return NULL; +} + + +/** + * Initialize a new geometry program object. + */ +struct gl_program * +_mesa_init_geometry_program( GLcontext *ctx, struct gl_geometry_program *prog, + GLenum target, GLuint id) +{ + if (prog) + return _mesa_init_program_struct( ctx, &prog->Base, target, id ); + else + return NULL; +} + + +/** + * Allocate and initialize a new fragment/vertex program object but + * don't put it into the program hash table. Called via + * ctx->Driver.NewProgram. May be overridden (ie. replaced) by a + * device driver function to implement OO deriviation with additional + * types not understood by this function. + * + * \param ctx context + * \param id program id/number + * \param target program target/type + * \return pointer to new program object + */ +struct gl_program * +_mesa_new_program(GLcontext *ctx, GLenum target, GLuint id) +{ + struct gl_program *prog; + switch (target) { + case GL_VERTEX_PROGRAM_ARB: /* == GL_VERTEX_PROGRAM_NV */ + case GL_VERTEX_STATE_PROGRAM_NV: + prog = _mesa_init_vertex_program(ctx, CALLOC_STRUCT(gl_vertex_program), + target, id ); + break; + case GL_FRAGMENT_PROGRAM_NV: + case GL_FRAGMENT_PROGRAM_ARB: + prog =_mesa_init_fragment_program(ctx, + CALLOC_STRUCT(gl_fragment_program), + target, id ); + break; + case MESA_GEOMETRY_PROGRAM: + prog = _mesa_init_geometry_program(ctx, + CALLOC_STRUCT(gl_geometry_program), + target, id); + break; + default: + _mesa_problem(ctx, "bad target in _mesa_new_program"); + prog = NULL; + } + return prog; +} + + +/** + * Delete a program and remove it from the hash table, ignoring the + * reference count. + * Called via ctx->Driver.DeleteProgram. May be wrapped (OO deriviation) + * by a device driver function. + */ +void +_mesa_delete_program(GLcontext *ctx, struct gl_program *prog) +{ + (void) ctx; + ASSERT(prog); + ASSERT(prog->RefCount==0); + + if (prog == &_mesa_DummyProgram) + return; + + if (prog->String) + free(prog->String); + + _mesa_free_instructions(prog->Instructions, prog->NumInstructions); + + if (prog->Parameters) { + _mesa_free_parameter_list(prog->Parameters); + } + if (prog->Varying) { + _mesa_free_parameter_list(prog->Varying); + } + if (prog->Attributes) { + _mesa_free_parameter_list(prog->Attributes); + } + + free(prog); +} + + +/** + * Return the gl_program object for a given ID. + * Basically just a wrapper for _mesa_HashLookup() to avoid a lot of + * casts elsewhere. + */ +struct gl_program * +_mesa_lookup_program(GLcontext *ctx, GLuint id) +{ + if (id) + return (struct gl_program *) _mesa_HashLookup(ctx->Shared->Programs, id); + else + return NULL; +} + + +/** + * Reference counting for vertex/fragment programs + */ +void +_mesa_reference_program(GLcontext *ctx, + struct gl_program **ptr, + struct gl_program *prog) +{ + assert(ptr); + if (*ptr && prog) { + /* sanity check */ + if ((*ptr)->Target == GL_VERTEX_PROGRAM_ARB) + ASSERT(prog->Target == GL_VERTEX_PROGRAM_ARB); + else if ((*ptr)->Target == GL_FRAGMENT_PROGRAM_ARB) + ASSERT(prog->Target == GL_FRAGMENT_PROGRAM_ARB || + prog->Target == GL_FRAGMENT_PROGRAM_NV); + else if ((*ptr)->Target == MESA_GEOMETRY_PROGRAM) + ASSERT(prog->Target == MESA_GEOMETRY_PROGRAM); + } + if (*ptr == prog) { + return; /* no change */ + } + if (*ptr) { + GLboolean deleteFlag; + + /*_glthread_LOCK_MUTEX((*ptr)->Mutex);*/ +#if 0 + printf("Program %p ID=%u Target=%s Refcount-- to %d\n", + *ptr, (*ptr)->Id, + ((*ptr)->Target == GL_VERTEX_PROGRAM_ARB ? "VP" : + ((*ptr)->Target == MESA_GEOMETRY_PROGRAM ? "GP" : "FP")), + (*ptr)->RefCount - 1); +#endif + ASSERT((*ptr)->RefCount > 0); + (*ptr)->RefCount--; + + deleteFlag = ((*ptr)->RefCount == 0); + /*_glthread_UNLOCK_MUTEX((*ptr)->Mutex);*/ + + if (deleteFlag) { + ASSERT(ctx); + ctx->Driver.DeleteProgram(ctx, *ptr); + } + + *ptr = NULL; + } + + assert(!*ptr); + if (prog) { + /*_glthread_LOCK_MUTEX(prog->Mutex);*/ + prog->RefCount++; +#if 0 + printf("Program %p ID=%u Target=%s Refcount++ to %d\n", + prog, prog->Id, + (prog->Target == GL_VERTEX_PROGRAM_ARB ? "VP" : + (prog->Target == MESA_GEOMETRY_PROGRAM ? "GP" : "FP")), + prog->RefCount); +#endif + /*_glthread_UNLOCK_MUTEX(prog->Mutex);*/ + } + + *ptr = prog; +} + + +/** + * Return a copy of a program. + * XXX Problem here if the program object is actually OO-derivation + * made by a device driver. + */ +struct gl_program * +_mesa_clone_program(GLcontext *ctx, const struct gl_program *prog) +{ + struct gl_program *clone; + + clone = ctx->Driver.NewProgram(ctx, prog->Target, prog->Id); + if (!clone) + return NULL; + + assert(clone->Target == prog->Target); + assert(clone->RefCount == 1); + + clone->String = (GLubyte *) _mesa_strdup((char *) prog->String); + clone->Format = prog->Format; + clone->Instructions = _mesa_alloc_instructions(prog->NumInstructions); + if (!clone->Instructions) { + _mesa_reference_program(ctx, &clone, NULL); + return NULL; + } + _mesa_copy_instructions(clone->Instructions, prog->Instructions, + prog->NumInstructions); + clone->InputsRead = prog->InputsRead; + clone->OutputsWritten = prog->OutputsWritten; + clone->SamplersUsed = prog->SamplersUsed; + clone->ShadowSamplers = prog->ShadowSamplers; + memcpy(clone->TexturesUsed, prog->TexturesUsed, sizeof(prog->TexturesUsed)); + + if (prog->Parameters) + clone->Parameters = _mesa_clone_parameter_list(prog->Parameters); + memcpy(clone->LocalParams, prog->LocalParams, sizeof(clone->LocalParams)); + if (prog->Varying) + clone->Varying = _mesa_clone_parameter_list(prog->Varying); + if (prog->Attributes) + clone->Attributes = _mesa_clone_parameter_list(prog->Attributes); + memcpy(clone->LocalParams, prog->LocalParams, sizeof(clone->LocalParams)); + clone->IndirectRegisterFiles = prog->IndirectRegisterFiles; + clone->NumInstructions = prog->NumInstructions; + clone->NumTemporaries = prog->NumTemporaries; + clone->NumParameters = prog->NumParameters; + clone->NumAttributes = prog->NumAttributes; + clone->NumAddressRegs = prog->NumAddressRegs; + clone->NumNativeInstructions = prog->NumNativeInstructions; + clone->NumNativeTemporaries = prog->NumNativeTemporaries; + clone->NumNativeParameters = prog->NumNativeParameters; + clone->NumNativeAttributes = prog->NumNativeAttributes; + clone->NumNativeAddressRegs = prog->NumNativeAddressRegs; + clone->NumAluInstructions = prog->NumAluInstructions; + clone->NumTexInstructions = prog->NumTexInstructions; + clone->NumTexIndirections = prog->NumTexIndirections; + clone->NumNativeAluInstructions = prog->NumNativeAluInstructions; + clone->NumNativeTexInstructions = prog->NumNativeTexInstructions; + clone->NumNativeTexIndirections = prog->NumNativeTexIndirections; + + switch (prog->Target) { + case GL_VERTEX_PROGRAM_ARB: + { + const struct gl_vertex_program *vp + = (const struct gl_vertex_program *) prog; + struct gl_vertex_program *vpc = (struct gl_vertex_program *) clone; + vpc->IsPositionInvariant = vp->IsPositionInvariant; + vpc->IsNVProgram = vp->IsNVProgram; + } + break; + case GL_FRAGMENT_PROGRAM_ARB: + { + const struct gl_fragment_program *fp + = (const struct gl_fragment_program *) prog; + struct gl_fragment_program *fpc = (struct gl_fragment_program *) clone; + fpc->FogOption = fp->FogOption; + fpc->UsesKill = fp->UsesKill; + fpc->OriginUpperLeft = fp->OriginUpperLeft; + fpc->PixelCenterInteger = fp->PixelCenterInteger; + } + break; + case MESA_GEOMETRY_PROGRAM: + { + const struct gl_geometry_program *gp + = (const struct gl_geometry_program *) prog; + struct gl_geometry_program *gpc = (struct gl_geometry_program *) clone; + gpc->VerticesOut = gp->VerticesOut; + gpc->InputType = gp->InputType; + gpc->OutputType = gp->OutputType; + } + break; + default: + _mesa_problem(NULL, "Unexpected target in _mesa_clone_program"); + } + + return clone; +} + + +/** + * Insert 'count' NOP instructions at 'start' in the given program. + * Adjust branch targets accordingly. + */ +GLboolean +_mesa_insert_instructions(struct gl_program *prog, GLuint start, GLuint count) +{ + const GLuint origLen = prog->NumInstructions; + const GLuint newLen = origLen + count; + struct prog_instruction *newInst; + GLuint i; + + /* adjust branches */ + for (i = 0; i < prog->NumInstructions; i++) { + struct prog_instruction *inst = prog->Instructions + i; + if (inst->BranchTarget > 0) { + if ((GLuint)inst->BranchTarget >= start) { + inst->BranchTarget += count; + } + } + } + + /* Alloc storage for new instructions */ + newInst = _mesa_alloc_instructions(newLen); + if (!newInst) { + return GL_FALSE; + } + + /* Copy 'start' instructions into new instruction buffer */ + _mesa_copy_instructions(newInst, prog->Instructions, start); + + /* init the new instructions */ + _mesa_init_instructions(newInst + start, count); + + /* Copy the remaining/tail instructions to new inst buffer */ + _mesa_copy_instructions(newInst + start + count, + prog->Instructions + start, + origLen - start); + + /* free old instructions */ + _mesa_free_instructions(prog->Instructions, origLen); + + /* install new instructions */ + prog->Instructions = newInst; + prog->NumInstructions = newLen; + + return GL_TRUE; +} + +/** + * Delete 'count' instructions at 'start' in the given program. + * Adjust branch targets accordingly. + */ +GLboolean +_mesa_delete_instructions(struct gl_program *prog, GLuint start, GLuint count) +{ + const GLuint origLen = prog->NumInstructions; + const GLuint newLen = origLen - count; + struct prog_instruction *newInst; + GLuint i; + + /* adjust branches */ + for (i = 0; i < prog->NumInstructions; i++) { + struct prog_instruction *inst = prog->Instructions + i; + if (inst->BranchTarget > 0) { + if (inst->BranchTarget > (GLint) start) { + inst->BranchTarget -= count; + } + } + } + + /* Alloc storage for new instructions */ + newInst = _mesa_alloc_instructions(newLen); + if (!newInst) { + return GL_FALSE; + } + + /* Copy 'start' instructions into new instruction buffer */ + _mesa_copy_instructions(newInst, prog->Instructions, start); + + /* Copy the remaining/tail instructions to new inst buffer */ + _mesa_copy_instructions(newInst + start, + prog->Instructions + start + count, + newLen - start); + + /* free old instructions */ + _mesa_free_instructions(prog->Instructions, origLen); + + /* install new instructions */ + prog->Instructions = newInst; + prog->NumInstructions = newLen; + + return GL_TRUE; +} + + +/** + * Search instructions for registers that match (oldFile, oldIndex), + * replacing them with (newFile, newIndex). + */ +static void +replace_registers(struct prog_instruction *inst, GLuint numInst, + GLuint oldFile, GLuint oldIndex, + GLuint newFile, GLuint newIndex) +{ + GLuint i, j; + for (i = 0; i < numInst; i++) { + /* src regs */ + for (j = 0; j < _mesa_num_inst_src_regs(inst[i].Opcode); j++) { + if (inst[i].SrcReg[j].File == oldFile && + inst[i].SrcReg[j].Index == oldIndex) { + inst[i].SrcReg[j].File = newFile; + inst[i].SrcReg[j].Index = newIndex; + } + } + /* dst reg */ + if (inst[i].DstReg.File == oldFile && inst[i].DstReg.Index == oldIndex) { + inst[i].DstReg.File = newFile; + inst[i].DstReg.Index = newIndex; + } + } +} + + +/** + * Search instructions for references to program parameters. When found, + * increment the parameter index by 'offset'. + * Used when combining programs. + */ +static void +adjust_param_indexes(struct prog_instruction *inst, GLuint numInst, + GLuint offset) +{ + GLuint i, j; + for (i = 0; i < numInst; i++) { + for (j = 0; j < _mesa_num_inst_src_regs(inst[i].Opcode); j++) { + GLuint f = inst[i].SrcReg[j].File; + if (f == PROGRAM_CONSTANT || + f == PROGRAM_UNIFORM || + f == PROGRAM_STATE_VAR) { + inst[i].SrcReg[j].Index += offset; + } + } + } +} + + +/** + * Combine two programs into one. Fix instructions so the outputs of + * the first program go to the inputs of the second program. + */ +struct gl_program * +_mesa_combine_programs(GLcontext *ctx, + const struct gl_program *progA, + const struct gl_program *progB) +{ + struct prog_instruction *newInst; + struct gl_program *newProg; + const GLuint lenA = progA->NumInstructions - 1; /* omit END instr */ + const GLuint lenB = progB->NumInstructions; + const GLuint numParamsA = _mesa_num_parameters(progA->Parameters); + const GLuint newLength = lenA + lenB; + GLboolean usedTemps[MAX_PROGRAM_TEMPS]; + GLuint firstTemp = 0; + GLbitfield inputsB; + GLuint i; + + ASSERT(progA->Target == progB->Target); + + newInst = _mesa_alloc_instructions(newLength); + if (!newInst) + return GL_FALSE; + + _mesa_copy_instructions(newInst, progA->Instructions, lenA); + _mesa_copy_instructions(newInst + lenA, progB->Instructions, lenB); + + /* adjust branch / instruction addresses for B's instructions */ + for (i = 0; i < lenB; i++) { + newInst[lenA + i].BranchTarget += lenA; + } + + newProg = ctx->Driver.NewProgram(ctx, progA->Target, 0); + newProg->Instructions = newInst; + newProg->NumInstructions = newLength; + + /* find used temp regs (we may need new temps below) */ + _mesa_find_used_registers(newProg, PROGRAM_TEMPORARY, + usedTemps, MAX_PROGRAM_TEMPS); + + if (newProg->Target == GL_FRAGMENT_PROGRAM_ARB) { + struct gl_fragment_program *fprogA, *fprogB, *newFprog; + GLbitfield progB_inputsRead = progB->InputsRead; + GLint progB_colorFile, progB_colorIndex; + + fprogA = (struct gl_fragment_program *) progA; + fprogB = (struct gl_fragment_program *) progB; + newFprog = (struct gl_fragment_program *) newProg; + + newFprog->UsesKill = fprogA->UsesKill || fprogB->UsesKill; + + /* We'll do a search and replace for instances + * of progB_colorFile/progB_colorIndex below... + */ + progB_colorFile = PROGRAM_INPUT; + progB_colorIndex = FRAG_ATTRIB_COL0; + + /* + * The fragment program may get color from a state var rather than + * a fragment input (vertex output) if it's constant. + * See the texenvprogram.c code. + * So, search the program's parameter list now to see if the program + * gets color from a state var instead of a conventional fragment + * input register. + */ + for (i = 0; i < progB->Parameters->NumParameters; i++) { + struct gl_program_parameter *p = &progB->Parameters->Parameters[i]; + if (p->Type == PROGRAM_STATE_VAR && + p->StateIndexes[0] == STATE_INTERNAL && + p->StateIndexes[1] == STATE_CURRENT_ATTRIB && + p->StateIndexes[2] == VERT_ATTRIB_COLOR0) { + progB_inputsRead |= FRAG_BIT_COL0; + progB_colorFile = PROGRAM_STATE_VAR; + progB_colorIndex = i; + break; + } + } + + /* Connect color outputs of fprogA to color inputs of fprogB, via a + * new temporary register. + */ + if ((progA->OutputsWritten & (1 << FRAG_RESULT_COLOR)) && + (progB_inputsRead & FRAG_BIT_COL0)) { + GLint tempReg = _mesa_find_free_register(usedTemps, MAX_PROGRAM_TEMPS, + firstTemp); + if (tempReg < 0) { + _mesa_problem(ctx, "No free temp regs found in " + "_mesa_combine_programs(), using 31"); + tempReg = 31; + } + firstTemp = tempReg + 1; + + /* replace writes to result.color[0] with tempReg */ + replace_registers(newInst, lenA, + PROGRAM_OUTPUT, FRAG_RESULT_COLOR, + PROGRAM_TEMPORARY, tempReg); + /* replace reads from the input color with tempReg */ + replace_registers(newInst + lenA, lenB, + progB_colorFile, progB_colorIndex, /* search for */ + PROGRAM_TEMPORARY, tempReg /* replace with */ ); + } + + /* compute combined program's InputsRead */ + inputsB = progB_inputsRead; + if (progA->OutputsWritten & (1 << FRAG_RESULT_COLOR)) { + inputsB &= ~(1 << FRAG_ATTRIB_COL0); + } + newProg->InputsRead = progA->InputsRead | inputsB; + newProg->OutputsWritten = progB->OutputsWritten; + newProg->SamplersUsed = progA->SamplersUsed | progB->SamplersUsed; + } + else { + /* vertex program */ + assert(0); /* XXX todo */ + } + + /* + * Merge parameters (uniforms, constants, etc) + */ + newProg->Parameters = _mesa_combine_parameter_lists(progA->Parameters, + progB->Parameters); + + adjust_param_indexes(newInst + lenA, lenB, numParamsA); + + + return newProg; +} + + +/** + * Populate the 'used' array with flags indicating which registers (TEMPs, + * INPUTs, OUTPUTs, etc, are used by the given program. + * \param file type of register to scan for + * \param used returns true/false flags for in use / free + * \param usedSize size of the 'used' array + */ +void +_mesa_find_used_registers(const struct gl_program *prog, + gl_register_file file, + GLboolean used[], GLuint usedSize) +{ + GLuint i, j; + + memset(used, 0, usedSize); + + for (i = 0; i < prog->NumInstructions; i++) { + const struct prog_instruction *inst = prog->Instructions + i; + const GLuint n = _mesa_num_inst_src_regs(inst->Opcode); + + if (inst->DstReg.File == file) { + ASSERT(inst->DstReg.Index < usedSize); + if(inst->DstReg.Index < usedSize) + used[inst->DstReg.Index] = GL_TRUE; + } + + for (j = 0; j < n; j++) { + if (inst->SrcReg[j].File == file) { + ASSERT(inst->SrcReg[j].Index < usedSize); + if(inst->SrcReg[j].Index < usedSize) + used[inst->SrcReg[j].Index] = GL_TRUE; + } + } + } +} + + +/** + * Scan the given 'used' register flag array for the first entry + * that's >= firstReg. + * \param used vector of flags indicating registers in use (as returned + * by _mesa_find_used_registers()) + * \param usedSize size of the 'used' array + * \param firstReg first register to start searching at + * \return index of unused register, or -1 if none. + */ +GLint +_mesa_find_free_register(const GLboolean used[], + GLuint usedSize, GLuint firstReg) +{ + GLuint i; + + assert(firstReg < usedSize); + + for (i = firstReg; i < usedSize; i++) + if (!used[i]) + return i; + + return -1; +} + + +/** + * "Post-process" a GPU program. This is intended to be used for debugging. + * Example actions include no-op'ing instructions or changing instruction + * behaviour. + */ +void +_mesa_postprocess_program(GLcontext *ctx, struct gl_program *prog) +{ + static const GLfloat white[4] = { 0.5, 0.5, 0.5, 0.5 }; + GLuint i; + GLuint whiteSwizzle; + GLint whiteIndex = _mesa_add_unnamed_constant(prog->Parameters, + white, 4, &whiteSwizzle); + + (void) whiteIndex; + + for (i = 0; i < prog->NumInstructions; i++) { + struct prog_instruction *inst = prog->Instructions + i; + const GLuint n = _mesa_num_inst_src_regs(inst->Opcode); + + (void) n; + + if (_mesa_is_tex_instruction(inst->Opcode)) { +#if 0 + /* replace TEX/TXP/TXB with MOV */ + inst->Opcode = OPCODE_MOV; + inst->DstReg.WriteMask = WRITEMASK_XYZW; + inst->SrcReg[0].Swizzle = SWIZZLE_XYZW; + inst->SrcReg[0].Negate = NEGATE_NONE; +#endif + +#if 0 + /* disable shadow texture mode */ + inst->TexShadow = 0; +#endif + } + + if (inst->Opcode == OPCODE_TXP) { +#if 0 + inst->Opcode = OPCODE_MOV; + inst->DstReg.WriteMask = WRITEMASK_XYZW; + inst->SrcReg[0].File = PROGRAM_CONSTANT; + inst->SrcReg[0].Index = whiteIndex; + inst->SrcReg[0].Swizzle = SWIZZLE_XYZW; + inst->SrcReg[0].Negate = NEGATE_NONE; +#endif +#if 0 + inst->TexShadow = 0; +#endif +#if 0 + inst->Opcode = OPCODE_TEX; + inst->TexShadow = 0; +#endif + } + + } +} diff --git a/mesalib/src/mesa/program/program.h b/mesalib/src/mesa/program/program.h new file mode 100644 index 000000000..f8f379808 --- /dev/null +++ b/mesalib/src/mesa/program/program.h @@ -0,0 +1,203 @@ +/* + * Mesa 3-D graphics library + * Version: 6.5.3 + * + * Copyright (C) 1999-2007 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. + */ + +/** + * \file program.c + * Vertex and fragment program support functions. + * \author Brian Paul + */ + + +/** + * \mainpage Mesa vertex and fragment program module + * + * This module or directory contains most of the code for vertex and + * fragment programs and shaders, including state management, parsers, + * and (some) software routines for executing programs + */ + +#ifndef PROGRAM_H +#define PROGRAM_H + +#include "main/mtypes.h" + + +extern struct gl_program _mesa_DummyProgram; + + +extern void +_mesa_init_program(GLcontext *ctx); + +extern void +_mesa_free_program_data(GLcontext *ctx); + +extern void +_mesa_update_default_objects_program(GLcontext *ctx); + +extern void +_mesa_set_program_error(GLcontext *ctx, GLint pos, const char *string); + +extern const GLubyte * +_mesa_find_line_column(const GLubyte *string, const GLubyte *pos, + GLint *line, GLint *col); + + +extern struct gl_program * +_mesa_init_vertex_program(GLcontext *ctx, + struct gl_vertex_program *prog, + GLenum target, GLuint id); + +extern struct gl_program * +_mesa_init_fragment_program(GLcontext *ctx, + struct gl_fragment_program *prog, + GLenum target, GLuint id); + +extern struct gl_program * +_mesa_init_geometry_program(GLcontext *ctx, + struct gl_geometry_program *prog, + GLenum target, GLuint id); + +extern struct gl_program * +_mesa_new_program(GLcontext *ctx, GLenum target, GLuint id); + +extern void +_mesa_delete_program(GLcontext *ctx, struct gl_program *prog); + +extern struct gl_program * +_mesa_lookup_program(GLcontext *ctx, GLuint id); + +extern void +_mesa_reference_program(GLcontext *ctx, + struct gl_program **ptr, + struct gl_program *prog); + +static INLINE void +_mesa_reference_vertprog(GLcontext *ctx, + struct gl_vertex_program **ptr, + struct gl_vertex_program *prog) +{ + _mesa_reference_program(ctx, (struct gl_program **) ptr, + (struct gl_program *) prog); +} + +static INLINE void +_mesa_reference_fragprog(GLcontext *ctx, + struct gl_fragment_program **ptr, + struct gl_fragment_program *prog) +{ + _mesa_reference_program(ctx, (struct gl_program **) ptr, + (struct gl_program *) prog); +} + +static INLINE void +_mesa_reference_geomprog(GLcontext *ctx, + struct gl_geometry_program **ptr, + struct gl_geometry_program *prog) +{ + _mesa_reference_program(ctx, (struct gl_program **) ptr, + (struct gl_program *) prog); +} + +extern struct gl_program * +_mesa_clone_program(GLcontext *ctx, const struct gl_program *prog); + +static INLINE struct gl_vertex_program * +_mesa_clone_vertex_program(GLcontext *ctx, + const struct gl_vertex_program *prog) +{ + return (struct gl_vertex_program *) _mesa_clone_program(ctx, &prog->Base); +} + +static INLINE struct gl_geometry_program * +_mesa_clone_geometry_program(GLcontext *ctx, + const struct gl_geometry_program *prog) +{ + return (struct gl_geometry_program *) _mesa_clone_program(ctx, &prog->Base); +} + +static INLINE struct gl_fragment_program * +_mesa_clone_fragment_program(GLcontext *ctx, + const struct gl_fragment_program *prog) +{ + return (struct gl_fragment_program *) _mesa_clone_program(ctx, &prog->Base); +} + + +extern GLboolean +_mesa_insert_instructions(struct gl_program *prog, GLuint start, GLuint count); + +extern GLboolean +_mesa_delete_instructions(struct gl_program *prog, GLuint start, GLuint count); + +extern struct gl_program * +_mesa_combine_programs(GLcontext *ctx, + const struct gl_program *progA, + const struct gl_program *progB); + +extern void +_mesa_find_used_registers(const struct gl_program *prog, + gl_register_file file, + GLboolean used[], GLuint usedSize); + +extern GLint +_mesa_find_free_register(const GLboolean used[], + GLuint maxRegs, GLuint firstReg); + +extern void +_mesa_postprocess_program(GLcontext *ctx, struct gl_program *prog); + +/* keep these in the same order as TGSI_PROCESSOR_* */ + +static INLINE GLuint +_mesa_program_target_to_index(GLenum v) +{ + switch(v) + { + case GL_VERTEX_PROGRAM_ARB: + return MESA_SHADER_VERTEX; + case GL_FRAGMENT_PROGRAM_ARB: + return MESA_SHADER_FRAGMENT; + case GL_GEOMETRY_PROGRAM_NV: + return MESA_SHADER_GEOMETRY; + default: + ASSERT(0); + return ~0; + } +} + +static INLINE GLenum +_mesa_program_index_to_target(GLuint i) +{ + GLenum enums[MESA_SHADER_TYPES] = { + GL_VERTEX_PROGRAM_ARB, + GL_FRAGMENT_PROGRAM_ARB, + GL_GEOMETRY_PROGRAM_NV, + }; + if(i >= MESA_SHADER_TYPES) + return 0; + else + return enums[i]; +} + +#endif /* PROGRAM_H */ diff --git a/mesalib/src/mesa/program/program_lexer.l b/mesalib/src/mesa/program/program_lexer.l new file mode 100644 index 000000000..0a50dab97 --- /dev/null +++ b/mesalib/src/mesa/program/program_lexer.l @@ -0,0 +1,507 @@ +%{ +/* + * Copyright © 2009 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. + */ +#include "main/glheader.h" +#include "main/imports.h" +#include "program/prog_instruction.h" +#include "program/prog_statevars.h" +#include "program/symbol_table.h" +#include "program/program_parser.h" +#include "program/program_parse.tab.h" + +#define require_ARB_vp (yyextra->mode == ARB_vertex) +#define require_ARB_fp (yyextra->mode == ARB_fragment) +#define require_NV_fp (yyextra->option.NV_fragment) +#define require_shadow (yyextra->option.Shadow) +#define require_rect (yyextra->option.TexRect) +#define require_texarray (yyextra->option.TexArray) + +#ifndef HAVE_UNISTD_H +#define YY_NO_UNISTD_H +#endif + +#define return_token_or_IDENTIFIER(condition, token) \ + do { \ + if (condition) { \ + return token; \ + } else { \ + return handle_ident(yyextra, yytext, yylval); \ + } \ + } while (0) + +#define return_token_or_DOT(condition, token) \ + do { \ + if (condition) { \ + return token; \ + } else { \ + yyless(1); \ + return DOT; \ + } \ + } while (0) + + +#define return_opcode(condition, token, opcode, len) \ + do { \ + if (condition && \ + _mesa_parse_instruction_suffix(yyextra, \ + yytext + len, \ + & yylval->temp_inst)) { \ + yylval->temp_inst.Opcode = OPCODE_ ## opcode; \ + return token; \ + } else { \ + return handle_ident(yyextra, yytext, yylval); \ + } \ + } while (0) + +#define SWIZZLE_INVAL MAKE_SWIZZLE4(SWIZZLE_NIL, SWIZZLE_NIL, \ + SWIZZLE_NIL, SWIZZLE_NIL) + +static unsigned +mask_from_char(char c) +{ + switch (c) { + case 'x': + case 'r': + return WRITEMASK_X; + case 'y': + case 'g': + return WRITEMASK_Y; + case 'z': + case 'b': + return WRITEMASK_Z; + case 'w': + case 'a': + return WRITEMASK_W; + } + + return 0; +} + +static unsigned +swiz_from_char(char c) +{ + switch (c) { + case 'x': + case 'r': + return SWIZZLE_X; + case 'y': + case 'g': + return SWIZZLE_Y; + case 'z': + case 'b': + return SWIZZLE_Z; + case 'w': + case 'a': + return SWIZZLE_W; + } + + return 0; +} + +static int +handle_ident(struct asm_parser_state *state, const char *text, YYSTYPE *lval) +{ + lval->string = strdup(text); + + return (_mesa_symbol_table_find_symbol(state->st, 0, text) == NULL) + ? IDENTIFIER : USED_IDENTIFIER; +} + +#define YY_USER_ACTION \ + do { \ + yylloc->first_column = yylloc->last_column; \ + yylloc->last_column += yyleng; \ + if ((yylloc->first_line == 1) \ + && (yylloc->first_column == 1)) { \ + yylloc->position = 1; \ + } else { \ + yylloc->position += yylloc->last_column - yylloc->first_column; \ + } \ + } while(0); + +#define YY_NO_INPUT + +/* Yes, this is intentionally doing nothing. We have this line of code +here only to avoid the compiler complaining about an unput function +that is defined, but never called. */ +#define YY_USER_INIT while (0) { unput(0); } + +#define YY_EXTRA_TYPE struct asm_parser_state * + +/* Flex defines a couple of functions with no declarations nor the +static keyword. Declare them here to avoid a compiler warning. */ +int yyget_column (yyscan_t yyscanner); +void yyset_column (int column_no , yyscan_t yyscanner); + +%} + +num [0-9]+ +exp [Ee][-+]?[0-9]+ +frac "."[0-9]+ +dot "."[ \t]* + +sz [HRX]? +szf [HR]? +cc C? +sat (_SAT)? + +%option bison-bridge bison-locations reentrant noyywrap +%% + +"!!ARBvp1.0" { return ARBvp_10; } +"!!ARBfp1.0" { return ARBfp_10; } +ADDRESS { + yylval->integer = at_address; + return_token_or_IDENTIFIER(require_ARB_vp, ADDRESS); +} +ALIAS { return ALIAS; } +ATTRIB { return ATTRIB; } +END { return END; } +OPTION { return OPTION; } +OUTPUT { return OUTPUT; } +PARAM { return PARAM; } +TEMP { yylval->integer = at_temp; return TEMP; } + +ABS{sz}{cc}{sat} { return_opcode( 1, VECTOR_OP, ABS, 3); } +ADD{sz}{cc}{sat} { return_opcode( 1, BIN_OP, ADD, 3); } +ARL { return_opcode(require_ARB_vp, ARL, ARL, 3); } + +CMP{sat} { return_opcode(require_ARB_fp, TRI_OP, CMP, 3); } +COS{szf}{cc}{sat} { return_opcode(require_ARB_fp, SCALAR_OP, COS, 3); } + +DDX{szf}{cc}{sat} { return_opcode(require_NV_fp, VECTOR_OP, DDX, 3); } +DDY{szf}{cc}{sat} { return_opcode(require_NV_fp, VECTOR_OP, DDY, 3); } +DP3{sz}{cc}{sat} { return_opcode( 1, BIN_OP, DP3, 3); } +DP4{sz}{cc}{sat} { return_opcode( 1, BIN_OP, DP4, 3); } +DPH{sz}{cc}{sat} { return_opcode( 1, BIN_OP, DPH, 3); } +DST{szf}{cc}{sat} { return_opcode( 1, BIN_OP, DST, 3); } + +EX2{szf}{cc}{sat} { return_opcode( 1, SCALAR_OP, EX2, 3); } +EXP { return_opcode(require_ARB_vp, SCALAR_OP, EXP, 3); } + +FLR{sz}{cc}{sat} { return_opcode( 1, VECTOR_OP, FLR, 3); } +FRC{sz}{cc}{sat} { return_opcode( 1, VECTOR_OP, FRC, 3); } + +KIL { return_opcode(require_ARB_fp, KIL, KIL, 3); } + +LIT{szf}{cc}{sat} { return_opcode( 1, VECTOR_OP, LIT, 3); } +LG2{szf}{cc}{sat} { return_opcode( 1, SCALAR_OP, LG2, 3); } +LOG { return_opcode(require_ARB_vp, SCALAR_OP, LOG, 3); } +LRP{sz}{cc}{sat} { return_opcode(require_ARB_fp, TRI_OP, LRP, 3); } + +MAD{sz}{cc}{sat} { return_opcode( 1, TRI_OP, MAD, 3); } +MAX{sz}{cc}{sat} { return_opcode( 1, BIN_OP, MAX, 3); } +MIN{sz}{cc}{sat} { return_opcode( 1, BIN_OP, MIN, 3); } +MOV{sz}{cc}{sat} { return_opcode( 1, VECTOR_OP, MOV, 3); } +MUL{sz}{cc}{sat} { return_opcode( 1, BIN_OP, MUL, 3); } + +PK2H { return_opcode(require_NV_fp, VECTOR_OP, PK2H, 4); } +PK2US { return_opcode(require_NV_fp, VECTOR_OP, PK2US, 5); } +PK4B { return_opcode(require_NV_fp, VECTOR_OP, PK4B, 4); } +PK4UB { return_opcode(require_NV_fp, VECTOR_OP, PK4UB, 5); } +POW{szf}{cc}{sat} { return_opcode( 1, BINSC_OP, POW, 3); } + +RCP{szf}{cc}{sat} { return_opcode( 1, SCALAR_OP, RCP, 3); } +RFL{szf}{cc}{sat} { return_opcode(require_NV_fp, BIN_OP, RFL, 3); } +RSQ{szf}{cc}{sat} { return_opcode( 1, SCALAR_OP, RSQ, 3); } + +SCS{sat} { return_opcode(require_ARB_fp, SCALAR_OP, SCS, 3); } +SEQ{sz}{cc}{sat} { return_opcode(require_NV_fp, BIN_OP, SEQ, 3); } +SFL{sz}{cc}{sat} { return_opcode(require_NV_fp, BIN_OP, SFL, 3); } +SGE{sz}{cc}{sat} { return_opcode( 1, BIN_OP, SGE, 3); } +SGT{sz}{cc}{sat} { return_opcode(require_NV_fp, BIN_OP, SGT, 3); } +SIN{szf}{cc}{sat} { return_opcode(require_ARB_fp, SCALAR_OP, SIN, 3); } +SLE{sz}{cc}{sat} { return_opcode(require_NV_fp, BIN_OP, SLE, 3); } +SLT{sz}{cc}{sat} { return_opcode( 1, BIN_OP, SLT, 3); } +SNE{sz}{cc}{sat} { return_opcode(require_NV_fp, BIN_OP, SNE, 3); } +STR{sz}{cc}{sat} { return_opcode(require_NV_fp, BIN_OP, STR, 3); } +SUB{sz}{cc}{sat} { return_opcode( 1, BIN_OP, SUB, 3); } +SWZ{sat} { return_opcode( 1, SWZ, SWZ, 3); } + +TEX{cc}{sat} { return_opcode(require_ARB_fp, SAMPLE_OP, TEX, 3); } +TXB{cc}{sat} { return_opcode(require_ARB_fp, SAMPLE_OP, TXB, 3); } +TXD{cc}{sat} { return_opcode(require_NV_fp, TXD_OP, TXD, 3); } +TXP{cc}{sat} { return_opcode(require_ARB_fp, SAMPLE_OP, TXP, 3); } + +UP2H{cc}{sat} { return_opcode(require_NV_fp, SCALAR_OP, UP2H, 4); } +UP2US{cc}{sat} { return_opcode(require_NV_fp, SCALAR_OP, UP2US, 5); } +UP4B{cc}{sat} { return_opcode(require_NV_fp, SCALAR_OP, UP4B, 4); } +UP4UB{cc}{sat} { return_opcode(require_NV_fp, SCALAR_OP, UP4UB, 5); } + +X2D{szf}{cc}{sat} { return_opcode(require_NV_fp, TRI_OP, X2D, 3); } +XPD{sat} { return_opcode( 1, BIN_OP, XPD, 3); } + +vertex { return_token_or_IDENTIFIER(require_ARB_vp, VERTEX); } +fragment { return_token_or_IDENTIFIER(require_ARB_fp, FRAGMENT); } +program { return PROGRAM; } +state { return STATE; } +result { return RESULT; } + +{dot}ambient { return AMBIENT; } +{dot}attenuation { return ATTENUATION; } +{dot}back { return BACK; } +{dot}clip { return_token_or_DOT(require_ARB_vp, CLIP); } +{dot}color { return COLOR; } +{dot}depth { return_token_or_DOT(require_ARB_fp, DEPTH); } +{dot}diffuse { return DIFFUSE; } +{dot}direction { return DIRECTION; } +{dot}emission { return EMISSION; } +{dot}env { return ENV; } +{dot}eye { return EYE; } +{dot}fogcoord { return FOGCOORD; } +{dot}fog { return FOG; } +{dot}front { return FRONT; } +{dot}half { return HALF; } +{dot}inverse { return INVERSE; } +{dot}invtrans { return INVTRANS; } +{dot}light { return LIGHT; } +{dot}lightmodel { return LIGHTMODEL; } +{dot}lightprod { return LIGHTPROD; } +{dot}local { return LOCAL; } +{dot}material { return MATERIAL; } +{dot}program { return MAT_PROGRAM; } +{dot}matrix { return MATRIX; } +{dot}matrixindex { return_token_or_DOT(require_ARB_vp, MATRIXINDEX); } +{dot}modelview { return MODELVIEW; } +{dot}mvp { return MVP; } +{dot}normal { return_token_or_DOT(require_ARB_vp, NORMAL); } +{dot}object { return OBJECT; } +{dot}palette { return PALETTE; } +{dot}params { return PARAMS; } +{dot}plane { return PLANE; } +{dot}point { return_token_or_DOT(require_ARB_vp, POINT_TOK); } +{dot}pointsize { return_token_or_DOT(require_ARB_vp, POINTSIZE); } +{dot}position { return POSITION; } +{dot}primary { return PRIMARY; } +{dot}projection { return PROJECTION; } +{dot}range { return_token_or_DOT(require_ARB_fp, RANGE); } +{dot}row { return ROW; } +{dot}scenecolor { return SCENECOLOR; } +{dot}secondary { return SECONDARY; } +{dot}shininess { return SHININESS; } +{dot}size { return_token_or_DOT(require_ARB_vp, SIZE_TOK); } +{dot}specular { return SPECULAR; } +{dot}spot { return SPOT; } +{dot}texcoord { return TEXCOORD; } +{dot}texenv { return_token_or_DOT(require_ARB_fp, TEXENV); } +{dot}texgen { return_token_or_DOT(require_ARB_vp, TEXGEN); } +{dot}q { return_token_or_DOT(require_ARB_vp, TEXGEN_Q); } +{dot}s { return_token_or_DOT(require_ARB_vp, TEXGEN_S); } +{dot}t { return_token_or_DOT(require_ARB_vp, TEXGEN_T); } +{dot}texture { return TEXTURE; } +{dot}transpose { return TRANSPOSE; } +{dot}attrib { return_token_or_DOT(require_ARB_vp, VTXATTRIB); } +{dot}weight { return_token_or_DOT(require_ARB_vp, WEIGHT); } + +texture { return_token_or_IDENTIFIER(require_ARB_fp, TEXTURE_UNIT); } +1D { return_token_or_IDENTIFIER(require_ARB_fp, TEX_1D); } +2D { return_token_or_IDENTIFIER(require_ARB_fp, TEX_2D); } +3D { return_token_or_IDENTIFIER(require_ARB_fp, TEX_3D); } +CUBE { return_token_or_IDENTIFIER(require_ARB_fp, TEX_CUBE); } +RECT { return_token_or_IDENTIFIER(require_ARB_fp && require_rect, TEX_RECT); } +SHADOW1D { return_token_or_IDENTIFIER(require_ARB_fp && require_shadow, TEX_SHADOW1D); } +SHADOW2D { return_token_or_IDENTIFIER(require_ARB_fp && require_shadow, TEX_SHADOW2D); } +SHADOWRECT { return_token_or_IDENTIFIER(require_ARB_fp && require_shadow && require_rect, TEX_SHADOWRECT); } +ARRAY1D { return_token_or_IDENTIFIER(require_ARB_fp && require_texarray, TEX_ARRAY1D); } +ARRAY2D { return_token_or_IDENTIFIER(require_ARB_fp && require_texarray, TEX_ARRAY2D); } +ARRAYSHADOW1D { return_token_or_IDENTIFIER(require_ARB_fp && require_shadow && require_texarray, TEX_ARRAYSHADOW1D); } +ARRAYSHADOW2D { return_token_or_IDENTIFIER(require_ARB_fp && require_shadow && require_texarray, TEX_ARRAYSHADOW2D); } + +[_a-zA-Z$][_a-zA-Z0-9$]* { return handle_ident(yyextra, yytext, yylval); } + +".." { return DOT_DOT; } + +{num} { + yylval->integer = strtol(yytext, NULL, 10); + return INTEGER; +} +{num}?{frac}{exp}? { + yylval->real = _mesa_strtof(yytext, NULL); + return REAL; +} +{num}"."/[^.] { + yylval->real = _mesa_strtof(yytext, NULL); + return REAL; +} +{num}{exp} { + yylval->real = _mesa_strtof(yytext, NULL); + return REAL; +} +{num}"."{exp} { + yylval->real = _mesa_strtof(yytext, NULL); + return REAL; +} + +".xyzw" { + yylval->swiz_mask.swizzle = SWIZZLE_NOOP; + yylval->swiz_mask.mask = WRITEMASK_XYZW; + return MASK4; +} + +".xy"[zw] { + yylval->swiz_mask.swizzle = SWIZZLE_INVAL; + yylval->swiz_mask.mask = WRITEMASK_XY + | mask_from_char(yytext[3]); + return MASK3; +} +".xzw" { + yylval->swiz_mask.swizzle = SWIZZLE_INVAL; + yylval->swiz_mask.mask = WRITEMASK_XZW; + return MASK3; +} +".yzw" { + yylval->swiz_mask.swizzle = SWIZZLE_INVAL; + yylval->swiz_mask.mask = WRITEMASK_YZW; + return MASK3; +} + +".x"[yzw] { + yylval->swiz_mask.swizzle = SWIZZLE_INVAL; + yylval->swiz_mask.mask = WRITEMASK_X + | mask_from_char(yytext[2]); + return MASK2; +} +".y"[zw] { + yylval->swiz_mask.swizzle = SWIZZLE_INVAL; + yylval->swiz_mask.mask = WRITEMASK_Y + | mask_from_char(yytext[2]); + return MASK2; +} +".zw" { + yylval->swiz_mask.swizzle = SWIZZLE_INVAL; + yylval->swiz_mask.mask = WRITEMASK_ZW; + return MASK2; +} + +"."[xyzw] { + const unsigned s = swiz_from_char(yytext[1]); + yylval->swiz_mask.swizzle = MAKE_SWIZZLE4(s, s, s, s); + yylval->swiz_mask.mask = mask_from_char(yytext[1]); + return MASK1; +} + +"."[xyzw]{4} { + yylval->swiz_mask.swizzle = MAKE_SWIZZLE4(swiz_from_char(yytext[1]), + swiz_from_char(yytext[2]), + swiz_from_char(yytext[3]), + swiz_from_char(yytext[4])); + yylval->swiz_mask.mask = 0; + return SWIZZLE; +} + +".rgba" { + yylval->swiz_mask.swizzle = SWIZZLE_NOOP; + yylval->swiz_mask.mask = WRITEMASK_XYZW; + return_token_or_DOT(require_ARB_fp, MASK4); +} + +".rg"[ba] { + yylval->swiz_mask.swizzle = SWIZZLE_INVAL; + yylval->swiz_mask.mask = WRITEMASK_XY + | mask_from_char(yytext[3]); + return_token_or_DOT(require_ARB_fp, MASK3); +} +".rba" { + yylval->swiz_mask.swizzle = SWIZZLE_INVAL; + yylval->swiz_mask.mask = WRITEMASK_XZW; + return_token_or_DOT(require_ARB_fp, MASK3); +} +".gba" { + yylval->swiz_mask.swizzle = SWIZZLE_INVAL; + yylval->swiz_mask.mask = WRITEMASK_YZW; + return_token_or_DOT(require_ARB_fp, MASK3); +} + +".r"[gba] { + yylval->swiz_mask.swizzle = SWIZZLE_INVAL; + yylval->swiz_mask.mask = WRITEMASK_X + | mask_from_char(yytext[2]); + return_token_or_DOT(require_ARB_fp, MASK2); +} +".g"[ba] { + yylval->swiz_mask.swizzle = SWIZZLE_INVAL; + yylval->swiz_mask.mask = WRITEMASK_Y + | mask_from_char(yytext[2]); + return_token_or_DOT(require_ARB_fp, MASK2); +} +".ba" { + yylval->swiz_mask.swizzle = SWIZZLE_INVAL; + yylval->swiz_mask.mask = WRITEMASK_ZW; + return_token_or_DOT(require_ARB_fp, MASK2); +} + +"."[gba] { + const unsigned s = swiz_from_char(yytext[1]); + yylval->swiz_mask.swizzle = MAKE_SWIZZLE4(s, s, s, s); + yylval->swiz_mask.mask = mask_from_char(yytext[1]); + return_token_or_DOT(require_ARB_fp, MASK1); +} + + +".r" { + if (require_ARB_vp) { + return TEXGEN_R; + } else { + yylval->swiz_mask.swizzle = MAKE_SWIZZLE4(SWIZZLE_X, SWIZZLE_X, + SWIZZLE_X, SWIZZLE_X); + yylval->swiz_mask.mask = WRITEMASK_X; + return MASK1; + } +} + +"."[rgba]{4} { + yylval->swiz_mask.swizzle = MAKE_SWIZZLE4(swiz_from_char(yytext[1]), + swiz_from_char(yytext[2]), + swiz_from_char(yytext[3]), + swiz_from_char(yytext[4])); + yylval->swiz_mask.mask = 0; + return_token_or_DOT(require_ARB_fp, SWIZZLE); +} + +"." { return DOT; } + +\n { + yylloc->first_line++; + yylloc->first_column = 1; + yylloc->last_line++; + yylloc->last_column = 1; + yylloc->position++; +} +[ \t\r]+ /* eat whitespace */ ; +#.*$ /* eat comments */ ; +. { return yytext[0]; } +%% + +void +_mesa_program_lexer_ctor(void **scanner, struct asm_parser_state *state, + const char *string, size_t len) +{ + yylex_init_extra(state, scanner); + yy_scan_bytes(string, len, *scanner); +} + +void +_mesa_program_lexer_dtor(void *scanner) +{ + yylex_destroy(scanner); +} diff --git a/mesalib/src/mesa/program/program_parse.tab.c b/mesalib/src/mesa/program/program_parse.tab.c new file mode 100644 index 000000000..08ead30de --- /dev/null +++ b/mesalib/src/mesa/program/program_parse.tab.c @@ -0,0 +1,5730 @@ + +/* A Bison parser, made by GNU Bison 2.4.1. */ + +/* Skeleton implementation for Bison's Yacc-like parsers in C + + Copyright (C) 1984, 1989, 1990, 2000, 2001, 2002, 2003, 2004, 2005, 2006 + Free Software Foundation, Inc. + + This program is free software: you can redistribute it and/or modify + it under the terms of the GNU General Public License as published by + the Free Software Foundation, either version 3 of the License, or + (at your option) any later version. + + This program is distributed in the hope that it will be useful, + but WITHOUT ANY WARRANTY; without even the implied warranty of + MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the + GNU General Public License for more details. + + You should have received a copy of the GNU General Public License + along with this program. If not, see <http://www.gnu.org/licenses/>. */ + +/* As a special exception, you may create a larger work that contains + part or all of the Bison parser skeleton and distribute that work + under terms of your choice, so long as that work isn't itself a + parser generator using the skeleton or a modified version thereof + as a parser skeleton. Alternatively, if you modify or redistribute + the parser skeleton itself, you may (at your option) remove this + special exception, which will cause the skeleton and the resulting + Bison output files to be licensed under the GNU General Public + License without this special exception. + + This special exception was added by the Free Software Foundation in + version 2.2 of Bison. */ + +/* C LALR(1) parser skeleton written by Richard Stallman, by + simplifying the original so-called "semantic" parser. */ + +/* All symbols defined below should begin with yy or YY, to avoid + infringing on user name space. This should be done even for local + variables, as they might otherwise be expanded by user macros. + There are some unavoidable exceptions within include files to + define necessary library symbols; they are noted "INFRINGES ON + USER NAME SPACE" below. */ + +/* Identify Bison output. */ +#define YYBISON 1 + +/* Bison version. */ +#define YYBISON_VERSION "2.4.1" + +/* Skeleton name. */ +#define YYSKELETON_NAME "yacc.c" + +/* Pure parsers. */ +#define YYPURE 1 + +/* Push parsers. */ +#define YYPUSH 0 + +/* Pull parsers. */ +#define YYPULL 1 + +/* Using locations. */ +#define YYLSP_NEEDED 1 + + + +/* Copy the first part of user declarations. */ + +/* Line 189 of yacc.c */ +#line 1 "program_parse.y" + +/* + * Copyright © 2009 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. + */ +#include <stdio.h> +#include <stdlib.h> +#include <string.h> + +#include "main/mtypes.h" +#include "main/imports.h" +#include "program/program.h" +#include "program/prog_parameter.h" +#include "program/prog_parameter_layout.h" +#include "program/prog_statevars.h" +#include "program/prog_instruction.h" + +#include "program/symbol_table.h" +#include "program/program_parser.h" + +extern void *yy_scan_string(char *); +extern void yy_delete_buffer(void *); + +static struct asm_symbol *declare_variable(struct asm_parser_state *state, + char *name, enum asm_type t, struct YYLTYPE *locp); + +static int add_state_reference(struct gl_program_parameter_list *param_list, + const gl_state_index tokens[STATE_LENGTH]); + +static int initialize_symbol_from_state(struct gl_program *prog, + struct asm_symbol *param_var, const gl_state_index tokens[STATE_LENGTH]); + +static int initialize_symbol_from_param(struct gl_program *prog, + struct asm_symbol *param_var, const gl_state_index tokens[STATE_LENGTH]); + +static int initialize_symbol_from_const(struct gl_program *prog, + struct asm_symbol *param_var, const struct asm_vector *vec, + GLboolean allowSwizzle); + +static int yyparse(struct asm_parser_state *state); + +static char *make_error_string(const char *fmt, ...); + +static void yyerror(struct YYLTYPE *locp, struct asm_parser_state *state, + const char *s); + +static int validate_inputs(struct YYLTYPE *locp, + struct asm_parser_state *state); + +static void init_dst_reg(struct prog_dst_register *r); + +static void set_dst_reg(struct prog_dst_register *r, + gl_register_file file, GLint index); + +static void init_src_reg(struct asm_src_register *r); + +static void set_src_reg(struct asm_src_register *r, + gl_register_file file, GLint index); + +static void set_src_reg_swz(struct asm_src_register *r, + gl_register_file file, GLint index, GLuint swizzle); + +static void asm_instruction_set_operands(struct asm_instruction *inst, + const struct prog_dst_register *dst, const struct asm_src_register *src0, + const struct asm_src_register *src1, const struct asm_src_register *src2); + +static struct asm_instruction *asm_instruction_ctor(gl_inst_opcode op, + const struct prog_dst_register *dst, const struct asm_src_register *src0, + const struct asm_src_register *src1, const struct asm_src_register *src2); + +static struct asm_instruction *asm_instruction_copy_ctor( + const struct prog_instruction *base, const struct prog_dst_register *dst, + const struct asm_src_register *src0, const struct asm_src_register *src1, + const struct asm_src_register *src2); + +#ifndef FALSE +#define FALSE 0 +#define TRUE (!FALSE) +#endif + +#define YYLLOC_DEFAULT(Current, Rhs, N) \ + do { \ + if (YYID(N)) { \ + (Current).first_line = YYRHSLOC(Rhs, 1).first_line; \ + (Current).first_column = YYRHSLOC(Rhs, 1).first_column; \ + (Current).position = YYRHSLOC(Rhs, 1).position; \ + (Current).last_line = YYRHSLOC(Rhs, N).last_line; \ + (Current).last_column = YYRHSLOC(Rhs, N).last_column; \ + } else { \ + (Current).first_line = YYRHSLOC(Rhs, 0).last_line; \ + (Current).last_line = (Current).first_line; \ + (Current).first_column = YYRHSLOC(Rhs, 0).last_column; \ + (Current).last_column = (Current).first_column; \ + (Current).position = YYRHSLOC(Rhs, 0).position \ + + (Current).first_column; \ + } \ + } while(YYID(0)) + +#define YYLEX_PARAM state->scanner + + +/* Line 189 of yacc.c */ +#line 193 "program_parse.tab.c" + +/* Enabling traces. */ +#ifndef YYDEBUG +# define YYDEBUG 0 +#endif + +/* Enabling verbose error messages. */ +#ifdef YYERROR_VERBOSE +# undef YYERROR_VERBOSE +# define YYERROR_VERBOSE 1 +#else +# define YYERROR_VERBOSE 1 +#endif + +/* Enabling the token table. */ +#ifndef YYTOKEN_TABLE +# define YYTOKEN_TABLE 0 +#endif + + +/* Tokens. */ +#ifndef YYTOKENTYPE +# define YYTOKENTYPE + /* Put the tokens into the symbol table, so that GDB and other debuggers + know about them. */ + enum yytokentype { + ARBvp_10 = 258, + ARBfp_10 = 259, + ADDRESS = 260, + ALIAS = 261, + ATTRIB = 262, + OPTION = 263, + OUTPUT = 264, + PARAM = 265, + TEMP = 266, + END = 267, + BIN_OP = 268, + BINSC_OP = 269, + SAMPLE_OP = 270, + SCALAR_OP = 271, + TRI_OP = 272, + VECTOR_OP = 273, + ARL = 274, + KIL = 275, + SWZ = 276, + TXD_OP = 277, + INTEGER = 278, + REAL = 279, + AMBIENT = 280, + ATTENUATION = 281, + BACK = 282, + CLIP = 283, + COLOR = 284, + DEPTH = 285, + DIFFUSE = 286, + DIRECTION = 287, + EMISSION = 288, + ENV = 289, + EYE = 290, + FOG = 291, + FOGCOORD = 292, + FRAGMENT = 293, + FRONT = 294, + HALF = 295, + INVERSE = 296, + INVTRANS = 297, + LIGHT = 298, + LIGHTMODEL = 299, + LIGHTPROD = 300, + LOCAL = 301, + MATERIAL = 302, + MAT_PROGRAM = 303, + MATRIX = 304, + MATRIXINDEX = 305, + MODELVIEW = 306, + MVP = 307, + NORMAL = 308, + OBJECT = 309, + PALETTE = 310, + PARAMS = 311, + PLANE = 312, + POINT_TOK = 313, + POINTSIZE = 314, + POSITION = 315, + PRIMARY = 316, + PROGRAM = 317, + PROJECTION = 318, + RANGE = 319, + RESULT = 320, + ROW = 321, + SCENECOLOR = 322, + SECONDARY = 323, + SHININESS = 324, + SIZE_TOK = 325, + SPECULAR = 326, + SPOT = 327, + STATE = 328, + TEXCOORD = 329, + TEXENV = 330, + TEXGEN = 331, + TEXGEN_Q = 332, + TEXGEN_R = 333, + TEXGEN_S = 334, + TEXGEN_T = 335, + TEXTURE = 336, + TRANSPOSE = 337, + TEXTURE_UNIT = 338, + TEX_1D = 339, + TEX_2D = 340, + TEX_3D = 341, + TEX_CUBE = 342, + TEX_RECT = 343, + TEX_SHADOW1D = 344, + TEX_SHADOW2D = 345, + TEX_SHADOWRECT = 346, + TEX_ARRAY1D = 347, + TEX_ARRAY2D = 348, + TEX_ARRAYSHADOW1D = 349, + TEX_ARRAYSHADOW2D = 350, + VERTEX = 351, + VTXATTRIB = 352, + WEIGHT = 353, + IDENTIFIER = 354, + USED_IDENTIFIER = 355, + MASK4 = 356, + MASK3 = 357, + MASK2 = 358, + MASK1 = 359, + SWIZZLE = 360, + DOT_DOT = 361, + DOT = 362 + }; +#endif + + + +#if ! defined YYSTYPE && ! defined YYSTYPE_IS_DECLARED +typedef union YYSTYPE +{ + +/* Line 214 of yacc.c */ +#line 126 "program_parse.y" + + struct asm_instruction *inst; + struct asm_symbol *sym; + struct asm_symbol temp_sym; + struct asm_swizzle_mask swiz_mask; + struct asm_src_register src_reg; + struct prog_dst_register dst_reg; + struct prog_instruction temp_inst; + char *string; + unsigned result; + unsigned attrib; + int integer; + float real; + gl_state_index state[STATE_LENGTH]; + int negate; + struct asm_vector vector; + gl_inst_opcode opcode; + + struct { + unsigned swz; + unsigned rgba_valid:1; + unsigned xyzw_valid:1; + unsigned negate:1; + } ext_swizzle; + + + +/* Line 214 of yacc.c */ +#line 364 "program_parse.tab.c" +} YYSTYPE; +# define YYSTYPE_IS_TRIVIAL 1 +# define yystype YYSTYPE /* obsolescent; will be withdrawn */ +# define YYSTYPE_IS_DECLARED 1 +#endif + +#if ! defined YYLTYPE && ! defined YYLTYPE_IS_DECLARED +typedef struct YYLTYPE +{ + int first_line; + int first_column; + int last_line; + int last_column; +} YYLTYPE; +# define yyltype YYLTYPE /* obsolescent; will be withdrawn */ +# define YYLTYPE_IS_DECLARED 1 +# define YYLTYPE_IS_TRIVIAL 1 +#endif + + +/* Copy the second part of user declarations. */ + +/* Line 264 of yacc.c */ +#line 271 "program_parse.y" + +extern int yylex(YYSTYPE *yylval_param, YYLTYPE *yylloc_param, + void *yyscanner); + + +/* Line 264 of yacc.c */ +#line 395 "program_parse.tab.c" + +#ifdef short +# undef short +#endif + +#ifdef YYTYPE_UINT8 +typedef YYTYPE_UINT8 yytype_uint8; +#else +typedef unsigned char yytype_uint8; +#endif + +#ifdef YYTYPE_INT8 +typedef YYTYPE_INT8 yytype_int8; +#elif (defined __STDC__ || defined __C99__FUNC__ \ + || defined __cplusplus || defined _MSC_VER) +typedef signed char yytype_int8; +#else +typedef short int yytype_int8; +#endif + +#ifdef YYTYPE_UINT16 +typedef YYTYPE_UINT16 yytype_uint16; +#else +typedef unsigned short int yytype_uint16; +#endif + +#ifdef YYTYPE_INT16 +typedef YYTYPE_INT16 yytype_int16; +#else +typedef short int yytype_int16; +#endif + +#ifndef YYSIZE_T +# ifdef __SIZE_TYPE__ +# define YYSIZE_T __SIZE_TYPE__ +# elif defined size_t +# define YYSIZE_T size_t +# elif ! defined YYSIZE_T && (defined __STDC__ || defined __C99__FUNC__ \ + || defined __cplusplus || defined _MSC_VER) +# include <stddef.h> /* INFRINGES ON USER NAME SPACE */ +# define YYSIZE_T size_t +# else +# define YYSIZE_T unsigned int +# endif +#endif + +#define YYSIZE_MAXIMUM ((YYSIZE_T) -1) + +#ifndef YY_ +# if YYENABLE_NLS +# if ENABLE_NLS +# include <libintl.h> /* INFRINGES ON USER NAME SPACE */ +# define YY_(msgid) dgettext ("bison-runtime", msgid) +# endif +# endif +# ifndef YY_ +# define YY_(msgid) msgid +# endif +#endif + +/* Suppress unused-variable warnings by "using" E. */ +#if ! defined lint || defined __GNUC__ +# define YYUSE(e) ((void) (e)) +#else +# define YYUSE(e) /* empty */ +#endif + +/* Identity function, used to suppress warnings about constant conditions. */ +#ifndef lint +# define YYID(n) (n) +#else +#if (defined __STDC__ || defined __C99__FUNC__ \ + || defined __cplusplus || defined _MSC_VER) +static int +YYID (int yyi) +#else +static int +YYID (yyi) + int yyi; +#endif +{ + return yyi; +} +#endif + +#if ! defined yyoverflow || YYERROR_VERBOSE + +/* The parser invokes alloca or malloc; define the necessary symbols. */ + +# ifdef YYSTACK_USE_ALLOCA +# if YYSTACK_USE_ALLOCA +# ifdef __GNUC__ +# define YYSTACK_ALLOC __builtin_alloca +# elif defined __BUILTIN_VA_ARG_INCR +# include <alloca.h> /* INFRINGES ON USER NAME SPACE */ +# elif defined _AIX +# define YYSTACK_ALLOC __alloca +# elif defined _MSC_VER +# include <malloc.h> /* INFRINGES ON USER NAME SPACE */ +# define alloca _alloca +# else +# define YYSTACK_ALLOC alloca +# if ! defined _ALLOCA_H && ! defined _STDLIB_H && (defined __STDC__ || defined __C99__FUNC__ \ + || defined __cplusplus || defined _MSC_VER) +# include <stdlib.h> /* INFRINGES ON USER NAME SPACE */ +# ifndef _STDLIB_H +# define _STDLIB_H 1 +# endif +# endif +# endif +# endif +# endif + +# ifdef YYSTACK_ALLOC + /* Pacify GCC's `empty if-body' warning. */ +# define YYSTACK_FREE(Ptr) do { /* empty */; } while (YYID (0)) +# ifndef YYSTACK_ALLOC_MAXIMUM + /* The OS might guarantee only one guard page at the bottom of the stack, + and a page size can be as small as 4096 bytes. So we cannot safely + invoke alloca (N) if N exceeds 4096. Use a slightly smaller number + to allow for a few compiler-allocated temporary stack slots. */ +# define YYSTACK_ALLOC_MAXIMUM 4032 /* reasonable circa 2006 */ +# endif +# else +# define YYSTACK_ALLOC YYMALLOC +# define YYSTACK_FREE YYFREE +# ifndef YYSTACK_ALLOC_MAXIMUM +# define YYSTACK_ALLOC_MAXIMUM YYSIZE_MAXIMUM +# endif +# if (defined __cplusplus && ! defined _STDLIB_H \ + && ! ((defined YYMALLOC || defined malloc) \ + && (defined YYFREE || defined free))) +# include <stdlib.h> /* INFRINGES ON USER NAME SPACE */ +# ifndef _STDLIB_H +# define _STDLIB_H 1 +# endif +# endif +# ifndef YYMALLOC +# define YYMALLOC malloc +# if ! defined malloc && ! defined _STDLIB_H && (defined __STDC__ || defined __C99__FUNC__ \ + || defined __cplusplus || defined _MSC_VER) +void *malloc (YYSIZE_T); /* INFRINGES ON USER NAME SPACE */ +# endif +# endif +# ifndef YYFREE +# define YYFREE free +# if ! defined free && ! defined _STDLIB_H && (defined __STDC__ || defined __C99__FUNC__ \ + || defined __cplusplus || defined _MSC_VER) +void free (void *); /* INFRINGES ON USER NAME SPACE */ +# endif +# endif +# endif +#endif /* ! defined yyoverflow || YYERROR_VERBOSE */ + + +#if (! defined yyoverflow \ + && (! defined __cplusplus \ + || (defined YYLTYPE_IS_TRIVIAL && YYLTYPE_IS_TRIVIAL \ + && defined YYSTYPE_IS_TRIVIAL && YYSTYPE_IS_TRIVIAL))) + +/* A type that is properly aligned for any stack member. */ +union yyalloc +{ + yytype_int16 yyss_alloc; + YYSTYPE yyvs_alloc; + YYLTYPE yyls_alloc; +}; + +/* The size of the maximum gap between one aligned stack and the next. */ +# define YYSTACK_GAP_MAXIMUM (sizeof (union yyalloc) - 1) + +/* The size of an array large to enough to hold all stacks, each with + N elements. */ +# define YYSTACK_BYTES(N) \ + ((N) * (sizeof (yytype_int16) + sizeof (YYSTYPE) + sizeof (YYLTYPE)) \ + + 2 * YYSTACK_GAP_MAXIMUM) + +/* Copy COUNT objects from FROM to TO. The source and destination do + not overlap. */ +# ifndef YYCOPY +# if defined __GNUC__ && 1 < __GNUC__ +# define YYCOPY(To, From, Count) \ + __builtin_memcpy (To, From, (Count) * sizeof (*(From))) +# else +# define YYCOPY(To, From, Count) \ + do \ + { \ + YYSIZE_T yyi; \ + for (yyi = 0; yyi < (Count); yyi++) \ + (To)[yyi] = (From)[yyi]; \ + } \ + while (YYID (0)) +# endif +# endif + +/* Relocate STACK from its old location to the new one. The + local variables YYSIZE and YYSTACKSIZE give the old and new number of + elements in the stack, and YYPTR gives the new location of the + stack. Advance YYPTR to a properly aligned location for the next + stack. */ +# define YYSTACK_RELOCATE(Stack_alloc, Stack) \ + do \ + { \ + YYSIZE_T yynewbytes; \ + YYCOPY (&yyptr->Stack_alloc, Stack, yysize); \ + Stack = &yyptr->Stack_alloc; \ + yynewbytes = yystacksize * sizeof (*Stack) + YYSTACK_GAP_MAXIMUM; \ + yyptr += yynewbytes / sizeof (*yyptr); \ + } \ + while (YYID (0)) + +#endif + +/* YYFINAL -- State number of the termination state. */ +#define YYFINAL 5 +/* YYLAST -- Last index in YYTABLE. */ +#define YYLAST 396 + +/* YYNTOKENS -- Number of terminals. */ +#define YYNTOKENS 120 +/* YYNNTS -- Number of nonterminals. */ +#define YYNNTS 143 +/* YYNRULES -- Number of rules. */ +#define YYNRULES 282 +/* YYNRULES -- Number of states. */ +#define YYNSTATES 475 + +/* YYTRANSLATE(YYLEX) -- Bison symbol number corresponding to YYLEX. */ +#define YYUNDEFTOK 2 +#define YYMAXUTOK 362 + +#define YYTRANSLATE(YYX) \ + ((unsigned int) (YYX) <= YYMAXUTOK ? yytranslate[YYX] : YYUNDEFTOK) + +/* YYTRANSLATE[YYLEX] -- Bison symbol number corresponding to YYLEX. */ +static const yytype_uint8 yytranslate[] = +{ + 0, 2, 2, 2, 2, 2, 2, 2, 2, 2, + 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, + 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, + 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, + 115, 116, 2, 113, 109, 114, 2, 2, 2, 2, + 2, 2, 2, 2, 2, 2, 2, 2, 2, 108, + 2, 117, 2, 2, 2, 2, 2, 2, 2, 2, + 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, + 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, + 2, 111, 2, 112, 2, 2, 2, 2, 2, 2, + 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, + 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, + 2, 2, 2, 118, 110, 119, 2, 2, 2, 2, + 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, + 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, + 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, + 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, + 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, + 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, + 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, + 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, + 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, + 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, + 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, + 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, + 2, 2, 2, 2, 2, 2, 1, 2, 3, 4, + 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, + 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, + 25, 26, 27, 28, 29, 30, 31, 32, 33, 34, + 35, 36, 37, 38, 39, 40, 41, 42, 43, 44, + 45, 46, 47, 48, 49, 50, 51, 52, 53, 54, + 55, 56, 57, 58, 59, 60, 61, 62, 63, 64, + 65, 66, 67, 68, 69, 70, 71, 72, 73, 74, + 75, 76, 77, 78, 79, 80, 81, 82, 83, 84, + 85, 86, 87, 88, 89, 90, 91, 92, 93, 94, + 95, 96, 97, 98, 99, 100, 101, 102, 103, 104, + 105, 106, 107 +}; + +#if YYDEBUG +/* YYPRHS[YYN] -- Index of the first RHS symbol of rule number YYN in + YYRHS. */ +static const yytype_uint16 yyprhs[] = +{ + 0, 0, 3, 8, 10, 12, 15, 16, 20, 23, + 24, 27, 30, 32, 34, 36, 38, 40, 42, 44, + 46, 48, 50, 52, 54, 59, 64, 69, 76, 83, + 92, 101, 104, 107, 120, 123, 125, 127, 129, 131, + 133, 135, 137, 139, 141, 143, 145, 147, 154, 157, + 162, 165, 167, 171, 177, 181, 184, 192, 195, 197, + 199, 201, 203, 208, 210, 212, 214, 216, 218, 220, + 222, 226, 227, 230, 233, 235, 237, 239, 241, 243, + 245, 247, 249, 251, 252, 254, 256, 258, 260, 261, + 265, 269, 270, 273, 276, 278, 280, 282, 284, 286, + 288, 290, 292, 297, 300, 303, 305, 308, 310, 313, + 315, 318, 323, 328, 330, 331, 335, 337, 339, 342, + 344, 347, 349, 351, 355, 362, 363, 365, 368, 373, + 375, 379, 381, 383, 385, 387, 389, 391, 393, 395, + 397, 399, 402, 405, 408, 411, 414, 417, 420, 423, + 426, 429, 432, 435, 439, 441, 443, 445, 451, 453, + 455, 457, 460, 462, 464, 467, 469, 472, 479, 481, + 485, 487, 489, 491, 493, 495, 500, 502, 504, 506, + 508, 510, 512, 515, 517, 519, 525, 527, 530, 532, + 534, 540, 543, 544, 551, 555, 556, 558, 560, 562, + 564, 566, 569, 571, 573, 576, 581, 586, 587, 591, + 593, 595, 597, 600, 602, 604, 606, 608, 614, 616, + 620, 626, 632, 634, 638, 644, 646, 648, 650, 652, + 654, 656, 658, 660, 662, 666, 672, 680, 690, 693, + 696, 698, 700, 701, 702, 707, 709, 710, 711, 715, + 719, 721, 727, 730, 733, 736, 739, 743, 746, 750, + 751, 753, 755, 756, 758, 760, 761, 763, 765, 766, + 768, 770, 771, 775, 776, 780, 781, 785, 787, 789, + 791, 796, 798 +}; + +/* YYRHS -- A `-1'-separated list of the rules' RHS. */ +static const yytype_int16 yyrhs[] = +{ + 121, 0, -1, 122, 123, 125, 12, -1, 3, -1, + 4, -1, 123, 124, -1, -1, 8, 262, 108, -1, + 125, 126, -1, -1, 127, 108, -1, 170, 108, -1, + 128, -1, 129, -1, 130, -1, 131, -1, 132, -1, + 133, -1, 134, -1, 135, -1, 141, -1, 136, -1, + 137, -1, 138, -1, 19, 146, 109, 142, -1, 18, + 145, 109, 144, -1, 16, 145, 109, 142, -1, 14, + 145, 109, 142, 109, 142, -1, 13, 145, 109, 144, + 109, 144, -1, 17, 145, 109, 144, 109, 144, 109, + 144, -1, 15, 145, 109, 144, 109, 139, 109, 140, + -1, 20, 144, -1, 20, 166, -1, 22, 145, 109, + 144, 109, 144, 109, 144, 109, 139, 109, 140, -1, + 83, 256, -1, 84, -1, 85, -1, 86, -1, 87, + -1, 88, -1, 89, -1, 90, -1, 91, -1, 92, + -1, 93, -1, 94, -1, 95, -1, 21, 145, 109, + 150, 109, 147, -1, 241, 143, -1, 241, 110, 143, + 110, -1, 150, 162, -1, 238, -1, 241, 150, 163, + -1, 241, 110, 150, 163, 110, -1, 151, 164, 165, + -1, 159, 161, -1, 148, 109, 148, 109, 148, 109, + 148, -1, 241, 149, -1, 23, -1, 262, -1, 100, + -1, 172, -1, 152, 111, 153, 112, -1, 186, -1, + 249, -1, 100, -1, 100, -1, 154, -1, 155, -1, + 23, -1, 159, 160, 156, -1, -1, 113, 157, -1, + 114, 158, -1, 23, -1, 23, -1, 100, -1, 104, + -1, 104, -1, 104, -1, 104, -1, 101, -1, 105, + -1, -1, 101, -1, 102, -1, 103, -1, 104, -1, + -1, 115, 166, 116, -1, 115, 167, 116, -1, -1, + 168, 163, -1, 169, 163, -1, 99, -1, 100, -1, + 171, -1, 178, -1, 242, -1, 245, -1, 248, -1, + 261, -1, 7, 99, 117, 172, -1, 96, 173, -1, + 38, 177, -1, 60, -1, 98, 175, -1, 53, -1, + 29, 254, -1, 37, -1, 74, 255, -1, 50, 111, + 176, 112, -1, 97, 111, 174, 112, -1, 23, -1, + -1, 111, 176, 112, -1, 23, -1, 60, -1, 29, + 254, -1, 37, -1, 74, 255, -1, 179, -1, 180, + -1, 10, 99, 182, -1, 10, 99, 111, 181, 112, + 183, -1, -1, 23, -1, 117, 185, -1, 117, 118, + 184, 119, -1, 187, -1, 184, 109, 187, -1, 189, + -1, 225, -1, 235, -1, 189, -1, 225, -1, 236, + -1, 188, -1, 226, -1, 235, -1, 189, -1, 73, + 213, -1, 73, 190, -1, 73, 192, -1, 73, 195, + -1, 73, 197, -1, 73, 203, -1, 73, 199, -1, + 73, 206, -1, 73, 208, -1, 73, 210, -1, 73, + 212, -1, 73, 224, -1, 47, 253, 191, -1, 201, + -1, 33, -1, 69, -1, 43, 111, 202, 112, 193, + -1, 201, -1, 60, -1, 26, -1, 72, 194, -1, + 40, -1, 32, -1, 44, 196, -1, 25, -1, 253, + 67, -1, 45, 111, 202, 112, 253, 198, -1, 201, + -1, 75, 257, 200, -1, 29, -1, 25, -1, 31, + -1, 71, -1, 23, -1, 76, 255, 204, 205, -1, + 35, -1, 54, -1, 79, -1, 80, -1, 78, -1, + 77, -1, 36, 207, -1, 29, -1, 56, -1, 28, + 111, 209, 112, 57, -1, 23, -1, 58, 211, -1, + 70, -1, 26, -1, 215, 66, 111, 218, 112, -1, + 215, 214, -1, -1, 66, 111, 218, 106, 218, 112, + -1, 49, 219, 216, -1, -1, 217, -1, 41, -1, + 82, -1, 42, -1, 23, -1, 51, 220, -1, 63, + -1, 52, -1, 81, 255, -1, 55, 111, 222, 112, + -1, 48, 111, 223, 112, -1, -1, 111, 221, 112, + -1, 23, -1, 23, -1, 23, -1, 30, 64, -1, + 229, -1, 232, -1, 227, -1, 230, -1, 62, 34, + 111, 228, 112, -1, 233, -1, 233, 106, 233, -1, + 62, 34, 111, 233, 112, -1, 62, 46, 111, 231, + 112, -1, 234, -1, 234, 106, 234, -1, 62, 46, + 111, 234, 112, -1, 23, -1, 23, -1, 237, -1, + 239, -1, 238, -1, 239, -1, 240, -1, 24, -1, + 23, -1, 118, 240, 119, -1, 118, 240, 109, 240, + 119, -1, 118, 240, 109, 240, 109, 240, 119, -1, + 118, 240, 109, 240, 109, 240, 109, 240, 119, -1, + 241, 24, -1, 241, 23, -1, 113, -1, 114, -1, + -1, -1, 244, 11, 243, 247, -1, 262, -1, -1, + -1, 5, 246, 247, -1, 247, 109, 99, -1, 99, + -1, 244, 9, 99, 117, 249, -1, 65, 60, -1, + 65, 37, -1, 65, 250, -1, 65, 59, -1, 65, + 74, 255, -1, 65, 30, -1, 29, 251, 252, -1, + -1, 39, -1, 27, -1, -1, 61, -1, 68, -1, + -1, 39, -1, 27, -1, -1, 61, -1, 68, -1, + -1, 111, 258, 112, -1, -1, 111, 259, 112, -1, + -1, 111, 260, 112, -1, 23, -1, 23, -1, 23, + -1, 6, 99, 117, 100, -1, 99, -1, 100, -1 +}; + +/* YYRLINE[YYN] -- source line where rule number YYN was defined. */ +static const yytype_uint16 yyrline[] = +{ + 0, 278, 278, 281, 289, 301, 302, 305, 329, 330, + 333, 348, 351, 356, 363, 364, 365, 366, 367, 368, + 369, 372, 373, 374, 377, 383, 389, 395, 402, 408, + 415, 459, 464, 474, 518, 524, 525, 526, 527, 528, + 529, 530, 531, 532, 533, 534, 535, 538, 550, 558, + 575, 582, 601, 612, 632, 657, 664, 697, 704, 719, + 774, 817, 826, 848, 858, 862, 891, 910, 910, 912, + 919, 931, 932, 933, 936, 950, 964, 984, 995, 1007, + 1009, 1010, 1011, 1012, 1015, 1015, 1015, 1015, 1016, 1019, + 1023, 1028, 1035, 1042, 1049, 1072, 1095, 1096, 1097, 1098, + 1099, 1100, 1103, 1122, 1126, 1132, 1136, 1140, 1144, 1153, + 1162, 1166, 1171, 1177, 1188, 1188, 1189, 1191, 1195, 1199, + 1203, 1209, 1209, 1211, 1229, 1255, 1258, 1269, 1275, 1281, + 1282, 1289, 1295, 1301, 1309, 1315, 1321, 1329, 1335, 1341, + 1349, 1350, 1353, 1354, 1355, 1356, 1357, 1358, 1359, 1360, + 1361, 1362, 1363, 1366, 1375, 1379, 1383, 1389, 1398, 1402, + 1406, 1415, 1419, 1425, 1431, 1438, 1443, 1451, 1461, 1463, + 1471, 1477, 1481, 1485, 1491, 1502, 1511, 1515, 1520, 1524, + 1528, 1532, 1538, 1545, 1549, 1555, 1563, 1574, 1581, 1585, + 1591, 1601, 1612, 1616, 1634, 1643, 1646, 1652, 1656, 1660, + 1666, 1677, 1682, 1687, 1692, 1697, 1702, 1710, 1713, 1718, + 1731, 1739, 1750, 1758, 1758, 1760, 1760, 1762, 1772, 1777, + 1784, 1794, 1803, 1808, 1815, 1825, 1835, 1847, 1847, 1848, + 1848, 1850, 1860, 1868, 1878, 1886, 1894, 1903, 1914, 1918, + 1924, 1925, 1926, 1929, 1929, 1932, 1967, 1971, 1971, 1974, + 1981, 1990, 2004, 2013, 2022, 2026, 2035, 2044, 2055, 2062, + 2067, 2076, 2088, 2091, 2100, 2111, 2112, 2113, 2116, 2117, + 2118, 2121, 2122, 2125, 2126, 2129, 2130, 2133, 2144, 2155, + 2166, 2192, 2193 +}; +#endif + +#if YYDEBUG || YYERROR_VERBOSE || YYTOKEN_TABLE +/* YYTNAME[SYMBOL-NUM] -- String name of the symbol SYMBOL-NUM. + First, the terminals, then, starting at YYNTOKENS, nonterminals. */ +static const char *const yytname[] = +{ + "$end", "error", "$undefined", "ARBvp_10", "ARBfp_10", "ADDRESS", + "ALIAS", "ATTRIB", "OPTION", "OUTPUT", "PARAM", "TEMP", "END", "BIN_OP", + "BINSC_OP", "SAMPLE_OP", "SCALAR_OP", "TRI_OP", "VECTOR_OP", "ARL", + "KIL", "SWZ", "TXD_OP", "INTEGER", "REAL", "AMBIENT", "ATTENUATION", + "BACK", "CLIP", "COLOR", "DEPTH", "DIFFUSE", "DIRECTION", "EMISSION", + "ENV", "EYE", "FOG", "FOGCOORD", "FRAGMENT", "FRONT", "HALF", "INVERSE", + "INVTRANS", "LIGHT", "LIGHTMODEL", "LIGHTPROD", "LOCAL", "MATERIAL", + "MAT_PROGRAM", "MATRIX", "MATRIXINDEX", "MODELVIEW", "MVP", "NORMAL", + "OBJECT", "PALETTE", "PARAMS", "PLANE", "POINT_TOK", "POINTSIZE", + "POSITION", "PRIMARY", "PROGRAM", "PROJECTION", "RANGE", "RESULT", "ROW", + "SCENECOLOR", "SECONDARY", "SHININESS", "SIZE_TOK", "SPECULAR", "SPOT", + "STATE", "TEXCOORD", "TEXENV", "TEXGEN", "TEXGEN_Q", "TEXGEN_R", + "TEXGEN_S", "TEXGEN_T", "TEXTURE", "TRANSPOSE", "TEXTURE_UNIT", "TEX_1D", + "TEX_2D", "TEX_3D", "TEX_CUBE", "TEX_RECT", "TEX_SHADOW1D", + "TEX_SHADOW2D", "TEX_SHADOWRECT", "TEX_ARRAY1D", "TEX_ARRAY2D", + "TEX_ARRAYSHADOW1D", "TEX_ARRAYSHADOW2D", "VERTEX", "VTXATTRIB", + "WEIGHT", "IDENTIFIER", "USED_IDENTIFIER", "MASK4", "MASK3", "MASK2", + "MASK1", "SWIZZLE", "DOT_DOT", "DOT", "';'", "','", "'|'", "'['", "']'", + "'+'", "'-'", "'('", "')'", "'='", "'{'", "'}'", "$accept", "program", + "language", "optionSequence", "option", "statementSequence", "statement", + "instruction", "ALU_instruction", "TexInstruction", "ARL_instruction", + "VECTORop_instruction", "SCALARop_instruction", "BINSCop_instruction", + "BINop_instruction", "TRIop_instruction", "SAMPLE_instruction", + "KIL_instruction", "TXD_instruction", "texImageUnit", "texTarget", + "SWZ_instruction", "scalarSrcReg", "scalarUse", "swizzleSrcReg", + "maskedDstReg", "maskedAddrReg", "extendedSwizzle", "extSwizComp", + "extSwizSel", "srcReg", "dstReg", "progParamArray", "progParamArrayMem", + "progParamArrayAbs", "progParamArrayRel", "addrRegRelOffset", + "addrRegPosOffset", "addrRegNegOffset", "addrReg", "addrComponent", + "addrWriteMask", "scalarSuffix", "swizzleSuffix", "optionalMask", + "optionalCcMask", "ccTest", "ccTest2", "ccMaskRule", "ccMaskRule2", + "namingStatement", "ATTRIB_statement", "attribBinding", "vtxAttribItem", + "vtxAttribNum", "vtxOptWeightNum", "vtxWeightNum", "fragAttribItem", + "PARAM_statement", "PARAM_singleStmt", "PARAM_multipleStmt", + "optArraySize", "paramSingleInit", "paramMultipleInit", + "paramMultInitList", "paramSingleItemDecl", "paramSingleItemUse", + "paramMultipleItem", "stateMultipleItem", "stateSingleItem", + "stateMaterialItem", "stateMatProperty", "stateLightItem", + "stateLightProperty", "stateSpotProperty", "stateLightModelItem", + "stateLModProperty", "stateLightProdItem", "stateLProdProperty", + "stateTexEnvItem", "stateTexEnvProperty", "ambDiffSpecProperty", + "stateLightNumber", "stateTexGenItem", "stateTexGenType", + "stateTexGenCoord", "stateFogItem", "stateFogProperty", + "stateClipPlaneItem", "stateClipPlaneNum", "statePointItem", + "statePointProperty", "stateMatrixRow", "stateMatrixRows", + "optMatrixRows", "stateMatrixItem", "stateOptMatModifier", + "stateMatModifier", "stateMatrixRowNum", "stateMatrixName", + "stateOptModMatNum", "stateModMatNum", "statePaletteMatNum", + "stateProgramMatNum", "stateDepthItem", "programSingleItem", + "programMultipleItem", "progEnvParams", "progEnvParamNums", + "progEnvParam", "progLocalParams", "progLocalParamNums", + "progLocalParam", "progEnvParamNum", "progLocalParamNum", + "paramConstDecl", "paramConstUse", "paramConstScalarDecl", + "paramConstScalarUse", "paramConstVector", "signedFloatConstant", + "optionalSign", "TEMP_statement", "@1", "optVarSize", + "ADDRESS_statement", "@2", "varNameList", "OUTPUT_statement", + "resultBinding", "resultColBinding", "optResultFaceType", + "optResultColorType", "optFaceType", "optColorType", + "optTexCoordUnitNum", "optTexImageUnitNum", "optLegacyTexUnitNum", + "texCoordUnitNum", "texImageUnitNum", "legacyTexUnitNum", + "ALIAS_statement", "string", 0 +}; +#endif + +# ifdef YYPRINT +/* YYTOKNUM[YYLEX-NUM] -- Internal token number corresponding to + token YYLEX-NUM. */ +static const yytype_uint16 yytoknum[] = +{ + 0, 256, 257, 258, 259, 260, 261, 262, 263, 264, + 265, 266, 267, 268, 269, 270, 271, 272, 273, 274, + 275, 276, 277, 278, 279, 280, 281, 282, 283, 284, + 285, 286, 287, 288, 289, 290, 291, 292, 293, 294, + 295, 296, 297, 298, 299, 300, 301, 302, 303, 304, + 305, 306, 307, 308, 309, 310, 311, 312, 313, 314, + 315, 316, 317, 318, 319, 320, 321, 322, 323, 324, + 325, 326, 327, 328, 329, 330, 331, 332, 333, 334, + 335, 336, 337, 338, 339, 340, 341, 342, 343, 344, + 345, 346, 347, 348, 349, 350, 351, 352, 353, 354, + 355, 356, 357, 358, 359, 360, 361, 362, 59, 44, + 124, 91, 93, 43, 45, 40, 41, 61, 123, 125 +}; +# endif + +/* YYR1[YYN] -- Symbol number of symbol that rule YYN derives. */ +static const yytype_uint16 yyr1[] = +{ + 0, 120, 121, 122, 122, 123, 123, 124, 125, 125, + 126, 126, 127, 127, 128, 128, 128, 128, 128, 128, + 128, 129, 129, 129, 130, 131, 132, 133, 134, 135, + 136, 137, 137, 138, 139, 140, 140, 140, 140, 140, + 140, 140, 140, 140, 140, 140, 140, 141, 142, 142, + 143, 143, 144, 144, 145, 146, 147, 148, 149, 149, + 150, 150, 150, 150, 151, 151, 152, 153, 153, 154, + 155, 156, 156, 156, 157, 158, 159, 160, 161, 162, + 163, 163, 163, 163, 164, 164, 164, 164, 164, 165, + 165, 165, 166, 167, 168, 169, 170, 170, 170, 170, + 170, 170, 171, 172, 172, 173, 173, 173, 173, 173, + 173, 173, 173, 174, 175, 175, 176, 177, 177, 177, + 177, 178, 178, 179, 180, 181, 181, 182, 183, 184, + 184, 185, 185, 185, 186, 186, 186, 187, 187, 187, + 188, 188, 189, 189, 189, 189, 189, 189, 189, 189, + 189, 189, 189, 190, 191, 191, 191, 192, 193, 193, + 193, 193, 193, 194, 195, 196, 196, 197, 198, 199, + 200, 201, 201, 201, 202, 203, 204, 204, 205, 205, + 205, 205, 206, 207, 207, 208, 209, 210, 211, 211, + 212, 213, 214, 214, 215, 216, 216, 217, 217, 217, + 218, 219, 219, 219, 219, 219, 219, 220, 220, 221, + 222, 223, 224, 225, 225, 226, 226, 227, 228, 228, + 229, 230, 231, 231, 232, 233, 234, 235, 235, 236, + 236, 237, 238, 238, 239, 239, 239, 239, 240, 240, + 241, 241, 241, 243, 242, 244, 244, 246, 245, 247, + 247, 248, 249, 249, 249, 249, 249, 249, 250, 251, + 251, 251, 252, 252, 252, 253, 253, 253, 254, 254, + 254, 255, 255, 256, 256, 257, 257, 258, 259, 260, + 261, 262, 262 +}; + +/* YYR2[YYN] -- Number of symbols composing right hand side of rule YYN. */ +static const yytype_uint8 yyr2[] = +{ + 0, 2, 4, 1, 1, 2, 0, 3, 2, 0, + 2, 2, 1, 1, 1, 1, 1, 1, 1, 1, + 1, 1, 1, 1, 4, 4, 4, 6, 6, 8, + 8, 2, 2, 12, 2, 1, 1, 1, 1, 1, + 1, 1, 1, 1, 1, 1, 1, 6, 2, 4, + 2, 1, 3, 5, 3, 2, 7, 2, 1, 1, + 1, 1, 4, 1, 1, 1, 1, 1, 1, 1, + 3, 0, 2, 2, 1, 1, 1, 1, 1, 1, + 1, 1, 1, 0, 1, 1, 1, 1, 0, 3, + 3, 0, 2, 2, 1, 1, 1, 1, 1, 1, + 1, 1, 4, 2, 2, 1, 2, 1, 2, 1, + 2, 4, 4, 1, 0, 3, 1, 1, 2, 1, + 2, 1, 1, 3, 6, 0, 1, 2, 4, 1, + 3, 1, 1, 1, 1, 1, 1, 1, 1, 1, + 1, 2, 2, 2, 2, 2, 2, 2, 2, 2, + 2, 2, 2, 3, 1, 1, 1, 5, 1, 1, + 1, 2, 1, 1, 2, 1, 2, 6, 1, 3, + 1, 1, 1, 1, 1, 4, 1, 1, 1, 1, + 1, 1, 2, 1, 1, 5, 1, 2, 1, 1, + 5, 2, 0, 6, 3, 0, 1, 1, 1, 1, + 1, 2, 1, 1, 2, 4, 4, 0, 3, 1, + 1, 1, 2, 1, 1, 1, 1, 5, 1, 3, + 5, 5, 1, 3, 5, 1, 1, 1, 1, 1, + 1, 1, 1, 1, 3, 5, 7, 9, 2, 2, + 1, 1, 0, 0, 4, 1, 0, 0, 3, 3, + 1, 5, 2, 2, 2, 2, 3, 2, 3, 0, + 1, 1, 0, 1, 1, 0, 1, 1, 0, 1, + 1, 0, 3, 0, 3, 0, 3, 1, 1, 1, + 4, 1, 1 +}; + +/* YYDEFACT[STATE-NAME] -- Default rule to reduce with in state + STATE-NUM when YYTABLE doesn't specify something else to do. Zero + means the default is an error. */ +static const yytype_uint16 yydefact[] = +{ + 0, 3, 4, 0, 6, 1, 9, 0, 5, 246, + 281, 282, 0, 247, 0, 0, 0, 2, 0, 0, + 0, 0, 0, 0, 0, 242, 0, 0, 8, 0, + 12, 13, 14, 15, 16, 17, 18, 19, 21, 22, + 23, 20, 0, 96, 97, 121, 122, 98, 0, 99, + 100, 101, 245, 7, 0, 0, 0, 0, 0, 65, + 0, 88, 64, 0, 0, 0, 0, 0, 76, 0, + 0, 94, 240, 241, 31, 32, 83, 0, 0, 0, + 10, 11, 0, 243, 250, 248, 0, 0, 125, 242, + 123, 259, 257, 253, 255, 252, 271, 254, 242, 84, + 85, 86, 87, 91, 242, 242, 242, 242, 242, 242, + 78, 55, 81, 80, 82, 92, 233, 232, 0, 0, + 0, 0, 60, 0, 242, 83, 0, 61, 63, 134, + 135, 213, 214, 136, 229, 230, 0, 242, 0, 0, + 0, 280, 102, 126, 0, 127, 131, 132, 133, 227, + 228, 231, 0, 261, 260, 262, 0, 256, 0, 0, + 54, 0, 0, 0, 26, 0, 25, 24, 268, 119, + 117, 271, 104, 0, 0, 0, 0, 0, 0, 265, + 0, 265, 0, 0, 275, 271, 142, 143, 144, 145, + 147, 146, 148, 149, 150, 151, 0, 152, 268, 109, + 0, 107, 105, 271, 0, 114, 103, 83, 0, 52, + 0, 0, 0, 0, 244, 249, 0, 239, 238, 263, + 264, 258, 277, 0, 242, 95, 0, 0, 83, 242, + 0, 48, 0, 51, 0, 242, 269, 270, 118, 120, + 0, 0, 0, 212, 183, 184, 182, 0, 165, 267, + 266, 164, 0, 0, 0, 0, 207, 203, 0, 202, + 271, 195, 189, 188, 187, 0, 0, 0, 0, 108, + 0, 110, 0, 0, 106, 0, 242, 234, 69, 0, + 67, 68, 0, 242, 242, 251, 0, 124, 272, 28, + 89, 90, 93, 27, 0, 79, 50, 273, 0, 0, + 225, 0, 226, 0, 186, 0, 174, 0, 166, 0, + 171, 172, 155, 156, 173, 153, 154, 0, 0, 201, + 0, 204, 197, 199, 198, 194, 196, 279, 0, 170, + 169, 176, 177, 0, 0, 116, 0, 113, 0, 0, + 53, 0, 62, 77, 71, 47, 0, 0, 0, 242, + 49, 0, 34, 0, 242, 220, 224, 0, 0, 265, + 211, 0, 209, 0, 210, 0, 276, 181, 180, 178, + 179, 175, 200, 0, 111, 112, 115, 242, 235, 0, + 0, 70, 242, 58, 57, 59, 242, 0, 0, 0, + 129, 137, 140, 138, 215, 216, 139, 278, 0, 35, + 36, 37, 38, 39, 40, 41, 42, 43, 44, 45, + 46, 30, 29, 185, 160, 162, 159, 0, 157, 158, + 0, 206, 208, 205, 190, 0, 74, 72, 75, 73, + 0, 0, 0, 0, 141, 192, 242, 128, 274, 163, + 161, 167, 168, 242, 236, 242, 0, 0, 0, 0, + 191, 130, 0, 0, 0, 0, 218, 0, 222, 0, + 237, 242, 0, 217, 0, 221, 0, 0, 56, 33, + 219, 223, 0, 0, 193 +}; + +/* YYDEFGOTO[NTERM-NUM]. */ +static const yytype_int16 yydefgoto[] = +{ + -1, 3, 4, 6, 8, 9, 28, 29, 30, 31, + 32, 33, 34, 35, 36, 37, 38, 39, 40, 298, + 411, 41, 161, 231, 74, 60, 69, 345, 346, 384, + 232, 61, 126, 279, 280, 281, 381, 427, 429, 70, + 344, 111, 296, 115, 103, 160, 75, 227, 76, 228, + 42, 43, 127, 206, 338, 274, 336, 172, 44, 45, + 46, 144, 90, 287, 389, 145, 128, 390, 391, 129, + 186, 315, 187, 418, 440, 188, 251, 189, 441, 190, + 330, 316, 307, 191, 333, 371, 192, 246, 193, 305, + 194, 264, 195, 434, 450, 196, 325, 326, 373, 261, + 319, 363, 365, 361, 197, 130, 393, 394, 455, 131, + 395, 457, 132, 301, 303, 396, 133, 149, 134, 135, + 151, 77, 47, 139, 48, 49, 54, 85, 50, 62, + 97, 155, 221, 252, 238, 157, 352, 266, 223, 398, + 328, 51, 12 +}; + +/* YYPACT[STATE-NUM] -- Index in YYTABLE of the portion describing + STATE-NUM. */ +#define YYPACT_NINF -401 +static const yytype_int16 yypact[] = +{ + 193, -401, -401, 27, -401, -401, 62, 143, -401, 24, + -401, -401, -30, -401, -18, 12, 83, -401, 15, 15, + 15, 15, 15, 15, 67, 61, 15, 15, -401, 127, + -401, -401, -401, -401, -401, -401, -401, -401, -401, -401, + -401, -401, 144, -401, -401, -401, -401, -401, 204, -401, + -401, -401, -401, -401, 155, 136, 138, 34, 140, -401, + 147, 108, -401, 150, 156, 157, 158, 160, -401, 162, + 159, -401, -401, -401, -401, -401, 102, -13, 163, 164, + -401, -401, 165, -401, -401, 166, 170, 10, 235, 0, + -401, 141, -401, -401, -401, -401, 167, -401, 131, -401, + -401, -401, -401, 168, 131, 131, 131, 131, 131, 131, + -401, -401, -401, -401, -401, -401, -401, -401, 104, 97, + 114, 38, 169, 30, 131, 102, 171, -401, -401, -401, + -401, -401, -401, -401, -401, -401, 30, 131, 172, 155, + 175, -401, -401, -401, 173, -401, -401, -401, -401, -401, + -401, -401, 223, -401, -401, 123, 253, -401, 177, 149, + -401, 178, -10, 181, -401, 182, -401, -401, 134, -401, + -401, 167, -401, 183, 184, 185, 213, 99, 186, 154, + 187, 146, 153, 7, 188, 167, -401, -401, -401, -401, + -401, -401, -401, -401, -401, -401, 215, -401, 134, -401, + 190, -401, -401, 167, 191, 192, -401, 102, -48, -401, + 1, 195, 196, 214, 166, -401, 189, -401, -401, -401, + -401, -401, -401, 180, 131, -401, 194, 197, 102, 131, + 30, -401, 203, 205, 201, 131, -401, -401, -401, -401, + 285, 288, 289, -401, -401, -401, -401, 291, -401, -401, + -401, -401, 248, 291, 33, 206, 207, -401, 208, -401, + 167, 14, -401, -401, -401, 293, 292, 92, 209, -401, + 299, -401, 301, 299, -401, 216, 131, -401, -401, 217, + -401, -401, 221, 131, 131, -401, 212, -401, -401, -401, + -401, -401, -401, -401, 218, -401, -401, 220, 224, 225, + -401, 226, -401, 227, -401, 228, -401, 230, -401, 231, + -401, -401, -401, -401, -401, -401, -401, 304, 309, -401, + 312, -401, -401, -401, -401, -401, -401, -401, 232, -401, + -401, -401, -401, 161, 313, -401, 233, -401, 234, 238, + -401, 13, -401, -401, 137, -401, 242, -15, 243, 3, + -401, 314, -401, 133, 131, -401, -401, 296, 94, 146, + -401, 245, -401, 246, -401, 247, -401, -401, -401, -401, + -401, -401, -401, 249, -401, -401, -401, 131, -401, 332, + 337, -401, 131, -401, -401, -401, 131, 142, 114, 28, + -401, -401, -401, -401, -401, -401, -401, -401, 250, -401, + -401, -401, -401, -401, -401, -401, -401, -401, -401, -401, + -401, -401, -401, -401, -401, -401, -401, 331, -401, -401, + 68, -401, -401, -401, -401, 43, -401, -401, -401, -401, + 255, 256, 257, 258, -401, 300, 3, -401, -401, -401, + -401, -401, -401, 131, -401, 131, 201, 285, 288, 259, + -401, -401, 252, 264, 265, 263, 261, 266, 270, 313, + -401, 131, 133, -401, 285, -401, 288, 80, -401, -401, + -401, -401, 313, 267, -401 +}; + +/* YYPGOTO[NTERM-NUM]. */ +static const yytype_int16 yypgoto[] = +{ + -401, -401, -401, -401, -401, -401, -401, -401, -401, -401, + -401, -401, -401, -401, -401, -401, -401, -401, -401, -69, + -82, -401, -100, 151, -86, 210, -401, -401, -366, -401, + -54, -401, -401, -401, -401, -401, -401, -401, -401, 174, + -401, -401, -401, -118, -401, -401, 229, -401, -401, -401, + -401, -401, 295, -401, -401, -401, 110, -401, -401, -401, + -401, -401, -401, -401, -401, -401, -401, -51, -401, -88, + -401, -401, -401, -401, -401, -401, -401, -401, -401, -401, + -401, -311, 139, -401, -401, -401, -401, -401, -401, -401, + -401, -401, -401, -401, -401, -2, -401, -401, -400, -401, + -401, -401, -401, -401, -401, 298, -401, -401, -401, -401, + -401, -401, -401, -390, -295, 302, -401, -401, -136, -87, + -120, -89, -401, -401, -401, -401, -401, 251, -401, 176, + -401, -401, -401, -176, 198, -153, -401, -401, -401, -401, + -401, -401, -6 +}; + +/* YYTABLE[YYPACT[STATE-NUM]]. What to do in state STATE-NUM. If + positive, shift that token. If negative, reduce the rule which + number is the opposite. If zero, do what YYDEFACT says. + If YYTABLE_NINF, syntax error. */ +#define YYTABLE_NINF -230 +static const yytype_int16 yytable[] = +{ + 152, 146, 150, 52, 208, 254, 164, 209, 383, 167, + 116, 117, 158, 116, 117, 162, 430, 162, 239, 163, + 162, 165, 166, 125, 278, 118, 233, 5, 118, 13, + 14, 15, 267, 262, 16, 152, 17, 18, 19, 20, + 21, 22, 23, 24, 25, 26, 27, 419, 118, 119, + 271, 212, 119, 116, 117, 322, 323, 456, 310, 467, + 120, 276, 119, 120, 311, 387, 312, 198, 118, 207, + 7, 277, 473, 120, 470, 199, 388, 263, 53, 453, + 58, 55, 211, 121, 10, 11, 121, 122, 200, 275, + 122, 201, 119, 310, 233, 468, 324, 123, 202, 311, + 230, 68, 313, 120, 314, 124, 121, 321, 124, 442, + 292, 56, 203, 72, 73, 59, 72, 73, 124, 310, + 414, 124, 377, 10, 11, 311, 121, 331, 244, 293, + 122, 173, 378, 168, 415, 204, 205, 436, 289, 314, + 162, 169, 175, 174, 176, 88, 332, 437, 124, 299, + 177, 89, 443, 458, 416, 245, 341, 178, 179, 180, + 71, 181, 444, 182, 170, 314, 417, 68, 153, 91, + 92, 471, 183, 249, 72, 73, 432, 93, 171, 248, + 154, 249, 57, 420, 219, 250, 472, 152, 433, 184, + 185, 220, 424, 250, 347, 236, 1, 2, 348, 94, + 95, 255, 237, 112, 256, 257, 113, 114, 258, 99, + 100, 101, 102, 82, 96, 83, 259, 399, 400, 401, + 402, 403, 404, 405, 406, 407, 408, 409, 410, 63, + 64, 65, 66, 67, 260, 80, 78, 79, 367, 368, + 369, 370, 10, 11, 72, 73, 217, 218, 71, 225, + 379, 380, 81, 86, 84, 87, 98, 425, 143, 104, + 152, 392, 150, 110, 138, 105, 106, 107, 412, 108, + 141, 109, 136, 137, 215, 140, 222, 243, 156, 58, + -66, 268, 210, 159, 297, 216, 224, 229, 152, 213, + 234, 235, 288, 347, 240, 241, 242, 247, 253, 265, + 431, 270, 272, 273, 283, 284, 286, 295, 300, -229, + 290, 302, 304, 291, 306, 308, 327, 317, 318, 320, + 334, 329, 335, 452, 337, 343, 340, 360, 350, 342, + 349, 351, 362, 353, 354, 364, 372, 397, 355, 356, + 357, 385, 358, 359, 366, 374, 375, 152, 392, 150, + 376, 382, 386, 413, 152, 426, 347, 421, 422, 423, + 428, 424, 438, 439, 445, 446, 449, 464, 447, 448, + 459, 460, 347, 461, 462, 463, 466, 454, 465, 474, + 469, 294, 142, 339, 282, 451, 435, 147, 226, 285, + 214, 148, 309, 0, 0, 0, 269 +}; + +static const yytype_int16 yycheck[] = +{ + 89, 89, 89, 9, 124, 181, 106, 125, 23, 109, + 23, 24, 98, 23, 24, 104, 382, 106, 171, 105, + 109, 107, 108, 77, 23, 38, 162, 0, 38, 5, + 6, 7, 185, 26, 10, 124, 12, 13, 14, 15, + 16, 17, 18, 19, 20, 21, 22, 358, 38, 62, + 203, 137, 62, 23, 24, 41, 42, 447, 25, 459, + 73, 109, 62, 73, 31, 62, 33, 29, 38, 123, + 8, 119, 472, 73, 464, 37, 73, 70, 108, 445, + 65, 99, 136, 96, 99, 100, 96, 100, 50, 207, + 100, 53, 62, 25, 230, 461, 82, 110, 60, 31, + 110, 100, 69, 73, 71, 118, 96, 260, 118, 420, + 228, 99, 74, 113, 114, 100, 113, 114, 118, 25, + 26, 118, 109, 99, 100, 31, 96, 35, 29, 229, + 100, 34, 119, 29, 40, 97, 98, 109, 224, 71, + 229, 37, 28, 46, 30, 111, 54, 119, 118, 235, + 36, 117, 109, 448, 60, 56, 276, 43, 44, 45, + 99, 47, 119, 49, 60, 71, 72, 100, 27, 29, + 30, 466, 58, 27, 113, 114, 34, 37, 74, 25, + 39, 27, 99, 359, 61, 39, 106, 276, 46, 75, + 76, 68, 112, 39, 283, 61, 3, 4, 284, 59, + 60, 48, 68, 101, 51, 52, 104, 105, 55, 101, + 102, 103, 104, 9, 74, 11, 63, 84, 85, 86, + 87, 88, 89, 90, 91, 92, 93, 94, 95, 19, + 20, 21, 22, 23, 81, 108, 26, 27, 77, 78, + 79, 80, 99, 100, 113, 114, 23, 24, 99, 100, + 113, 114, 108, 117, 99, 117, 109, 377, 23, 109, + 349, 349, 349, 104, 99, 109, 109, 109, 354, 109, + 100, 109, 109, 109, 99, 109, 23, 64, 111, 65, + 111, 66, 111, 115, 83, 112, 109, 109, 377, 117, + 109, 109, 112, 382, 111, 111, 111, 111, 111, 111, + 386, 111, 111, 111, 109, 109, 117, 104, 23, 104, + 116, 23, 23, 116, 23, 67, 23, 111, 111, 111, + 111, 29, 23, 443, 23, 104, 110, 23, 110, 112, + 118, 111, 23, 109, 109, 23, 23, 23, 112, 112, + 112, 347, 112, 112, 112, 112, 112, 436, 436, 436, + 112, 109, 109, 57, 443, 23, 445, 112, 112, 112, + 23, 112, 112, 32, 109, 109, 66, 106, 111, 111, + 111, 119, 461, 109, 109, 112, 106, 446, 112, 112, + 462, 230, 87, 273, 210, 436, 388, 89, 159, 213, + 139, 89, 253, -1, -1, -1, 198 +}; + +/* YYSTOS[STATE-NUM] -- The (internal number of the) accessing + symbol of state STATE-NUM. */ +static const yytype_uint16 yystos[] = +{ + 0, 3, 4, 121, 122, 0, 123, 8, 124, 125, + 99, 100, 262, 5, 6, 7, 10, 12, 13, 14, + 15, 16, 17, 18, 19, 20, 21, 22, 126, 127, + 128, 129, 130, 131, 132, 133, 134, 135, 136, 137, + 138, 141, 170, 171, 178, 179, 180, 242, 244, 245, + 248, 261, 262, 108, 246, 99, 99, 99, 65, 100, + 145, 151, 249, 145, 145, 145, 145, 145, 100, 146, + 159, 99, 113, 114, 144, 166, 168, 241, 145, 145, + 108, 108, 9, 11, 99, 247, 117, 117, 111, 117, + 182, 29, 30, 37, 59, 60, 74, 250, 109, 101, + 102, 103, 104, 164, 109, 109, 109, 109, 109, 109, + 104, 161, 101, 104, 105, 163, 23, 24, 38, 62, + 73, 96, 100, 110, 118, 150, 152, 172, 186, 189, + 225, 229, 232, 236, 238, 239, 109, 109, 99, 243, + 109, 100, 172, 23, 181, 185, 189, 225, 235, 237, + 239, 240, 241, 27, 39, 251, 111, 255, 144, 115, + 165, 142, 241, 144, 142, 144, 144, 142, 29, 37, + 60, 74, 177, 34, 46, 28, 30, 36, 43, 44, + 45, 47, 49, 58, 75, 76, 190, 192, 195, 197, + 199, 203, 206, 208, 210, 212, 215, 224, 29, 37, + 50, 53, 60, 74, 97, 98, 173, 150, 240, 163, + 111, 150, 144, 117, 247, 99, 112, 23, 24, 61, + 68, 252, 23, 258, 109, 100, 166, 167, 169, 109, + 110, 143, 150, 238, 109, 109, 61, 68, 254, 255, + 111, 111, 111, 64, 29, 56, 207, 111, 25, 27, + 39, 196, 253, 111, 253, 48, 51, 52, 55, 63, + 81, 219, 26, 70, 211, 111, 257, 255, 66, 254, + 111, 255, 111, 111, 175, 163, 109, 119, 23, 153, + 154, 155, 159, 109, 109, 249, 117, 183, 112, 144, + 116, 116, 163, 142, 143, 104, 162, 83, 139, 144, + 23, 233, 23, 234, 23, 209, 23, 202, 67, 202, + 25, 31, 33, 69, 71, 191, 201, 111, 111, 220, + 111, 255, 41, 42, 82, 216, 217, 23, 260, 29, + 200, 35, 54, 204, 111, 23, 176, 23, 174, 176, + 110, 240, 112, 104, 160, 147, 148, 241, 144, 118, + 110, 111, 256, 109, 109, 112, 112, 112, 112, 112, + 23, 223, 23, 221, 23, 222, 112, 77, 78, 79, + 80, 205, 23, 218, 112, 112, 112, 109, 119, 113, + 114, 156, 109, 23, 149, 262, 109, 62, 73, 184, + 187, 188, 189, 226, 227, 230, 235, 23, 259, 84, + 85, 86, 87, 88, 89, 90, 91, 92, 93, 94, + 95, 140, 144, 57, 26, 40, 60, 72, 193, 201, + 253, 112, 112, 112, 112, 240, 23, 157, 23, 158, + 148, 144, 34, 46, 213, 215, 109, 119, 112, 32, + 194, 198, 201, 109, 119, 109, 109, 111, 111, 66, + 214, 187, 240, 148, 139, 228, 233, 231, 234, 111, + 119, 109, 109, 112, 106, 112, 106, 218, 148, 140, + 233, 234, 106, 218, 112 +}; + +#define yyerrok (yyerrstatus = 0) +#define yyclearin (yychar = YYEMPTY) +#define YYEMPTY (-2) +#define YYEOF 0 + +#define YYACCEPT goto yyacceptlab +#define YYABORT goto yyabortlab +#define YYERROR goto yyerrorlab + + +/* Like YYERROR except do call yyerror. This remains here temporarily + to ease the transition to the new meaning of YYERROR, for GCC. + Once GCC version 2 has supplanted version 1, this can go. */ + +#define YYFAIL goto yyerrlab + +#define YYRECOVERING() (!!yyerrstatus) + +#define YYBACKUP(Token, Value) \ +do \ + if (yychar == YYEMPTY && yylen == 1) \ + { \ + yychar = (Token); \ + yylval = (Value); \ + yytoken = YYTRANSLATE (yychar); \ + YYPOPSTACK (1); \ + goto yybackup; \ + } \ + else \ + { \ + yyerror (&yylloc, state, YY_("syntax error: cannot back up")); \ + YYERROR; \ + } \ +while (YYID (0)) + + +#define YYTERROR 1 +#define YYERRCODE 256 + + +/* YYLLOC_DEFAULT -- Set CURRENT to span from RHS[1] to RHS[N]. + If N is 0, then set CURRENT to the empty location which ends + the previous symbol: RHS[0] (always defined). */ + +#define YYRHSLOC(Rhs, K) ((Rhs)[K]) +#ifndef YYLLOC_DEFAULT +# define YYLLOC_DEFAULT(Current, Rhs, N) \ + do \ + if (YYID (N)) \ + { \ + (Current).first_line = YYRHSLOC (Rhs, 1).first_line; \ + (Current).first_column = YYRHSLOC (Rhs, 1).first_column; \ + (Current).last_line = YYRHSLOC (Rhs, N).last_line; \ + (Current).last_column = YYRHSLOC (Rhs, N).last_column; \ + } \ + else \ + { \ + (Current).first_line = (Current).last_line = \ + YYRHSLOC (Rhs, 0).last_line; \ + (Current).first_column = (Current).last_column = \ + YYRHSLOC (Rhs, 0).last_column; \ + } \ + while (YYID (0)) +#endif + + +/* YY_LOCATION_PRINT -- Print the location on the stream. + This macro was not mandated originally: define only if we know + we won't break user code: when these are the locations we know. */ + +#ifndef YY_LOCATION_PRINT +# if YYLTYPE_IS_TRIVIAL +# define YY_LOCATION_PRINT(File, Loc) \ + fprintf (File, "%d.%d-%d.%d", \ + (Loc).first_line, (Loc).first_column, \ + (Loc).last_line, (Loc).last_column) +# else +# define YY_LOCATION_PRINT(File, Loc) ((void) 0) +# endif +#endif + + +/* YYLEX -- calling `yylex' with the right arguments. */ + +#ifdef YYLEX_PARAM +# define YYLEX yylex (&yylval, &yylloc, YYLEX_PARAM) +#else +# define YYLEX yylex (&yylval, &yylloc, scanner) +#endif + +/* Enable debugging if requested. */ +#if YYDEBUG + +# ifndef YYFPRINTF +# include <stdio.h> /* INFRINGES ON USER NAME SPACE */ +# define YYFPRINTF fprintf +# endif + +# define YYDPRINTF(Args) \ +do { \ + if (yydebug) \ + YYFPRINTF Args; \ +} while (YYID (0)) + +# define YY_SYMBOL_PRINT(Title, Type, Value, Location) \ +do { \ + if (yydebug) \ + { \ + YYFPRINTF (stderr, "%s ", Title); \ + yy_symbol_print (stderr, \ + Type, Value, Location, state); \ + YYFPRINTF (stderr, "\n"); \ + } \ +} while (YYID (0)) + + +/*--------------------------------. +| Print this symbol on YYOUTPUT. | +`--------------------------------*/ + +/*ARGSUSED*/ +#if (defined __STDC__ || defined __C99__FUNC__ \ + || defined __cplusplus || defined _MSC_VER) +static void +yy_symbol_value_print (FILE *yyoutput, int yytype, YYSTYPE const * const yyvaluep, YYLTYPE const * const yylocationp, struct asm_parser_state *state) +#else +static void +yy_symbol_value_print (yyoutput, yytype, yyvaluep, yylocationp, state) + FILE *yyoutput; + int yytype; + YYSTYPE const * const yyvaluep; + YYLTYPE const * const yylocationp; + struct asm_parser_state *state; +#endif +{ + if (!yyvaluep) + return; + YYUSE (yylocationp); + YYUSE (state); +# ifdef YYPRINT + if (yytype < YYNTOKENS) + YYPRINT (yyoutput, yytoknum[yytype], *yyvaluep); +# else + YYUSE (yyoutput); +# endif + switch (yytype) + { + default: + break; + } +} + + +/*--------------------------------. +| Print this symbol on YYOUTPUT. | +`--------------------------------*/ + +#if (defined __STDC__ || defined __C99__FUNC__ \ + || defined __cplusplus || defined _MSC_VER) +static void +yy_symbol_print (FILE *yyoutput, int yytype, YYSTYPE const * const yyvaluep, YYLTYPE const * const yylocationp, struct asm_parser_state *state) +#else +static void +yy_symbol_print (yyoutput, yytype, yyvaluep, yylocationp, state) + FILE *yyoutput; + int yytype; + YYSTYPE const * const yyvaluep; + YYLTYPE const * const yylocationp; + struct asm_parser_state *state; +#endif +{ + if (yytype < YYNTOKENS) + YYFPRINTF (yyoutput, "token %s (", yytname[yytype]); + else + YYFPRINTF (yyoutput, "nterm %s (", yytname[yytype]); + + YY_LOCATION_PRINT (yyoutput, *yylocationp); + YYFPRINTF (yyoutput, ": "); + yy_symbol_value_print (yyoutput, yytype, yyvaluep, yylocationp, state); + YYFPRINTF (yyoutput, ")"); +} + +/*------------------------------------------------------------------. +| yy_stack_print -- Print the state stack from its BOTTOM up to its | +| TOP (included). | +`------------------------------------------------------------------*/ + +#if (defined __STDC__ || defined __C99__FUNC__ \ + || defined __cplusplus || defined _MSC_VER) +static void +yy_stack_print (yytype_int16 *yybottom, yytype_int16 *yytop) +#else +static void +yy_stack_print (yybottom, yytop) + yytype_int16 *yybottom; + yytype_int16 *yytop; +#endif +{ + YYFPRINTF (stderr, "Stack now"); + for (; yybottom <= yytop; yybottom++) + { + int yybot = *yybottom; + YYFPRINTF (stderr, " %d", yybot); + } + YYFPRINTF (stderr, "\n"); +} + +# define YY_STACK_PRINT(Bottom, Top) \ +do { \ + if (yydebug) \ + yy_stack_print ((Bottom), (Top)); \ +} while (YYID (0)) + + +/*------------------------------------------------. +| Report that the YYRULE is going to be reduced. | +`------------------------------------------------*/ + +#if (defined __STDC__ || defined __C99__FUNC__ \ + || defined __cplusplus || defined _MSC_VER) +static void +yy_reduce_print (YYSTYPE *yyvsp, YYLTYPE *yylsp, int yyrule, struct asm_parser_state *state) +#else +static void +yy_reduce_print (yyvsp, yylsp, yyrule, state) + YYSTYPE *yyvsp; + YYLTYPE *yylsp; + int yyrule; + struct asm_parser_state *state; +#endif +{ + int yynrhs = yyr2[yyrule]; + int yyi; + unsigned long int yylno = yyrline[yyrule]; + YYFPRINTF (stderr, "Reducing stack by rule %d (line %lu):\n", + yyrule - 1, yylno); + /* The symbols being reduced. */ + for (yyi = 0; yyi < yynrhs; yyi++) + { + YYFPRINTF (stderr, " $%d = ", yyi + 1); + yy_symbol_print (stderr, yyrhs[yyprhs[yyrule] + yyi], + &(yyvsp[(yyi + 1) - (yynrhs)]) + , &(yylsp[(yyi + 1) - (yynrhs)]) , state); + YYFPRINTF (stderr, "\n"); + } +} + +# define YY_REDUCE_PRINT(Rule) \ +do { \ + if (yydebug) \ + yy_reduce_print (yyvsp, yylsp, Rule, state); \ +} while (YYID (0)) + +/* Nonzero means print parse trace. It is left uninitialized so that + multiple parsers can coexist. */ +int yydebug; +#else /* !YYDEBUG */ +# define YYDPRINTF(Args) +# define YY_SYMBOL_PRINT(Title, Type, Value, Location) +# define YY_STACK_PRINT(Bottom, Top) +# define YY_REDUCE_PRINT(Rule) +#endif /* !YYDEBUG */ + + +/* YYINITDEPTH -- initial size of the parser's stacks. */ +#ifndef YYINITDEPTH +# define YYINITDEPTH 200 +#endif + +/* YYMAXDEPTH -- maximum size the stacks can grow to (effective only + if the built-in stack extension method is used). + + Do not make this value too large; the results are undefined if + YYSTACK_ALLOC_MAXIMUM < YYSTACK_BYTES (YYMAXDEPTH) + evaluated with infinite-precision integer arithmetic. */ + +#ifndef YYMAXDEPTH +# define YYMAXDEPTH 10000 +#endif + + + +#if YYERROR_VERBOSE + +# ifndef yystrlen +# if defined __GLIBC__ && defined _STRING_H +# define yystrlen strlen +# else +/* Return the length of YYSTR. */ +#if (defined __STDC__ || defined __C99__FUNC__ \ + || defined __cplusplus || defined _MSC_VER) +static YYSIZE_T +yystrlen (const char *yystr) +#else +static YYSIZE_T +yystrlen (yystr) + const char *yystr; +#endif +{ + YYSIZE_T yylen; + for (yylen = 0; yystr[yylen]; yylen++) + continue; + return yylen; +} +# endif +# endif + +# ifndef yystpcpy +# if defined __GLIBC__ && defined _STRING_H && defined _GNU_SOURCE +# define yystpcpy stpcpy +# else +/* Copy YYSRC to YYDEST, returning the address of the terminating '\0' in + YYDEST. */ +#if (defined __STDC__ || defined __C99__FUNC__ \ + || defined __cplusplus || defined _MSC_VER) +static char * +yystpcpy (char *yydest, const char *yysrc) +#else +static char * +yystpcpy (yydest, yysrc) + char *yydest; + const char *yysrc; +#endif +{ + char *yyd = yydest; + const char *yys = yysrc; + + while ((*yyd++ = *yys++) != '\0') + continue; + + return yyd - 1; +} +# endif +# endif + +# ifndef yytnamerr +/* Copy to YYRES the contents of YYSTR after stripping away unnecessary + quotes and backslashes, so that it's suitable for yyerror. The + heuristic is that double-quoting is unnecessary unless the string + contains an apostrophe, a comma, or backslash (other than + backslash-backslash). YYSTR is taken from yytname. If YYRES is + null, do not copy; instead, return the length of what the result + would have been. */ +static YYSIZE_T +yytnamerr (char *yyres, const char *yystr) +{ + if (*yystr == '"') + { + YYSIZE_T yyn = 0; + char const *yyp = yystr; + + for (;;) + switch (*++yyp) + { + case '\'': + case ',': + goto do_not_strip_quotes; + + case '\\': + if (*++yyp != '\\') + goto do_not_strip_quotes; + /* Fall through. */ + default: + if (yyres) + yyres[yyn] = *yyp; + yyn++; + break; + + case '"': + if (yyres) + yyres[yyn] = '\0'; + return yyn; + } + do_not_strip_quotes: ; + } + + if (! yyres) + return yystrlen (yystr); + + return yystpcpy (yyres, yystr) - yyres; +} +# endif + +/* Copy into YYRESULT an error message about the unexpected token + YYCHAR while in state YYSTATE. Return the number of bytes copied, + including the terminating null byte. If YYRESULT is null, do not + copy anything; just return the number of bytes that would be + copied. As a special case, return 0 if an ordinary "syntax error" + message will do. Return YYSIZE_MAXIMUM if overflow occurs during + size calculation. */ +static YYSIZE_T +yysyntax_error (char *yyresult, int yystate, int yychar) +{ + int yyn = yypact[yystate]; + + if (! (YYPACT_NINF < yyn && yyn <= YYLAST)) + return 0; + else + { + int yytype = YYTRANSLATE (yychar); + YYSIZE_T yysize0 = yytnamerr (0, yytname[yytype]); + YYSIZE_T yysize = yysize0; + YYSIZE_T yysize1; + int yysize_overflow = 0; + enum { YYERROR_VERBOSE_ARGS_MAXIMUM = 5 }; + char const *yyarg[YYERROR_VERBOSE_ARGS_MAXIMUM]; + int yyx; + +# if 0 + /* This is so xgettext sees the translatable formats that are + constructed on the fly. */ + YY_("syntax error, unexpected %s"); + YY_("syntax error, unexpected %s, expecting %s"); + YY_("syntax error, unexpected %s, expecting %s or %s"); + YY_("syntax error, unexpected %s, expecting %s or %s or %s"); + YY_("syntax error, unexpected %s, expecting %s or %s or %s or %s"); +# endif + char *yyfmt; + char const *yyf; + static char const yyunexpected[] = "syntax error, unexpected %s"; + static char const yyexpecting[] = ", expecting %s"; + static char const yyor[] = " or %s"; + char yyformat[sizeof yyunexpected + + sizeof yyexpecting - 1 + + ((YYERROR_VERBOSE_ARGS_MAXIMUM - 2) + * (sizeof yyor - 1))]; + char const *yyprefix = yyexpecting; + + /* Start YYX at -YYN if negative to avoid negative indexes in + YYCHECK. */ + int yyxbegin = yyn < 0 ? -yyn : 0; + + /* Stay within bounds of both yycheck and yytname. */ + int yychecklim = YYLAST - yyn + 1; + int yyxend = yychecklim < YYNTOKENS ? yychecklim : YYNTOKENS; + int yycount = 1; + + yyarg[0] = yytname[yytype]; + yyfmt = yystpcpy (yyformat, yyunexpected); + + for (yyx = yyxbegin; yyx < yyxend; ++yyx) + if (yycheck[yyx + yyn] == yyx && yyx != YYTERROR) + { + if (yycount == YYERROR_VERBOSE_ARGS_MAXIMUM) + { + yycount = 1; + yysize = yysize0; + yyformat[sizeof yyunexpected - 1] = '\0'; + break; + } + yyarg[yycount++] = yytname[yyx]; + yysize1 = yysize + yytnamerr (0, yytname[yyx]); + yysize_overflow |= (yysize1 < yysize); + yysize = yysize1; + yyfmt = yystpcpy (yyfmt, yyprefix); + yyprefix = yyor; + } + + yyf = YY_(yyformat); + yysize1 = yysize + yystrlen (yyf); + yysize_overflow |= (yysize1 < yysize); + yysize = yysize1; + + if (yysize_overflow) + return YYSIZE_MAXIMUM; + + if (yyresult) + { + /* Avoid sprintf, as that infringes on the user's name space. + Don't have undefined behavior even if the translation + produced a string with the wrong number of "%s"s. */ + char *yyp = yyresult; + int yyi = 0; + while ((*yyp = *yyf) != '\0') + { + if (*yyp == '%' && yyf[1] == 's' && yyi < yycount) + { + yyp += yytnamerr (yyp, yyarg[yyi++]); + yyf += 2; + } + else + { + yyp++; + yyf++; + } + } + } + return yysize; + } +} +#endif /* YYERROR_VERBOSE */ + + +/*-----------------------------------------------. +| Release the memory associated to this symbol. | +`-----------------------------------------------*/ + +/*ARGSUSED*/ +#if (defined __STDC__ || defined __C99__FUNC__ \ + || defined __cplusplus || defined _MSC_VER) +static void +yydestruct (const char *yymsg, int yytype, YYSTYPE *yyvaluep, YYLTYPE *yylocationp, struct asm_parser_state *state) +#else +static void +yydestruct (yymsg, yytype, yyvaluep, yylocationp, state) + const char *yymsg; + int yytype; + YYSTYPE *yyvaluep; + YYLTYPE *yylocationp; + struct asm_parser_state *state; +#endif +{ + YYUSE (yyvaluep); + YYUSE (yylocationp); + YYUSE (state); + + if (!yymsg) + yymsg = "Deleting"; + YY_SYMBOL_PRINT (yymsg, yytype, yyvaluep, yylocationp); + + switch (yytype) + { + + default: + break; + } +} + +/* Prevent warnings from -Wmissing-prototypes. */ +#ifdef YYPARSE_PARAM +#if defined __STDC__ || defined __cplusplus +int yyparse (void *YYPARSE_PARAM); +#else +int yyparse (); +#endif +#else /* ! YYPARSE_PARAM */ +#if defined __STDC__ || defined __cplusplus +int yyparse (struct asm_parser_state *state); +#else +int yyparse (); +#endif +#endif /* ! YYPARSE_PARAM */ + + + + + +/*-------------------------. +| yyparse or yypush_parse. | +`-------------------------*/ + +#ifdef YYPARSE_PARAM +#if (defined __STDC__ || defined __C99__FUNC__ \ + || defined __cplusplus || defined _MSC_VER) +int +yyparse (void *YYPARSE_PARAM) +#else +int +yyparse (YYPARSE_PARAM) + void *YYPARSE_PARAM; +#endif +#else /* ! YYPARSE_PARAM */ +#if (defined __STDC__ || defined __C99__FUNC__ \ + || defined __cplusplus || defined _MSC_VER) +int +yyparse (struct asm_parser_state *state) +#else +int +yyparse (state) + struct asm_parser_state *state; +#endif +#endif +{ +/* The lookahead symbol. */ +int yychar; + +/* The semantic value of the lookahead symbol. */ +YYSTYPE yylval; + +/* Location data for the lookahead symbol. */ +YYLTYPE yylloc; + + /* Number of syntax errors so far. */ + int yynerrs; + + int yystate; + /* Number of tokens to shift before error messages enabled. */ + int yyerrstatus; + + /* The stacks and their tools: + `yyss': related to states. + `yyvs': related to semantic values. + `yyls': related to locations. + + Refer to the stacks thru separate pointers, to allow yyoverflow + to reallocate them elsewhere. */ + + /* The state stack. */ + yytype_int16 yyssa[YYINITDEPTH]; + yytype_int16 *yyss; + yytype_int16 *yyssp; + + /* The semantic value stack. */ + YYSTYPE yyvsa[YYINITDEPTH]; + YYSTYPE *yyvs; + YYSTYPE *yyvsp; + + /* The location stack. */ + YYLTYPE yylsa[YYINITDEPTH]; + YYLTYPE *yyls; + YYLTYPE *yylsp; + + /* The locations where the error started and ended. */ + YYLTYPE yyerror_range[2]; + + YYSIZE_T yystacksize; + + int yyn; + int yyresult; + /* Lookahead token as an internal (translated) token number. */ + int yytoken; + /* The variables used to return semantic value and location from the + action routines. */ + YYSTYPE yyval; + YYLTYPE yyloc; + +#if YYERROR_VERBOSE + /* Buffer for error messages, and its allocated size. */ + char yymsgbuf[128]; + char *yymsg = yymsgbuf; + YYSIZE_T yymsg_alloc = sizeof yymsgbuf; +#endif + +#define YYPOPSTACK(N) (yyvsp -= (N), yyssp -= (N), yylsp -= (N)) + + /* The number of symbols on the RHS of the reduced rule. + Keep to zero when no symbol should be popped. */ + int yylen = 0; + + yytoken = 0; + yyss = yyssa; + yyvs = yyvsa; + yyls = yylsa; + yystacksize = YYINITDEPTH; + + YYDPRINTF ((stderr, "Starting parse\n")); + + yystate = 0; + yyerrstatus = 0; + yynerrs = 0; + yychar = YYEMPTY; /* Cause a token to be read. */ + + /* Initialize stack pointers. + Waste one element of value and location stack + so that they stay on the same level as the state stack. + The wasted elements are never initialized. */ + yyssp = yyss; + yyvsp = yyvs; + yylsp = yyls; + +#if YYLTYPE_IS_TRIVIAL + /* Initialize the default location before parsing starts. */ + yylloc.first_line = yylloc.last_line = 1; + yylloc.first_column = yylloc.last_column = 1; +#endif + + goto yysetstate; + +/*------------------------------------------------------------. +| yynewstate -- Push a new state, which is found in yystate. | +`------------------------------------------------------------*/ + yynewstate: + /* In all cases, when you get here, the value and location stacks + have just been pushed. So pushing a state here evens the stacks. */ + yyssp++; + + yysetstate: + *yyssp = yystate; + + if (yyss + yystacksize - 1 <= yyssp) + { + /* Get the current used size of the three stacks, in elements. */ + YYSIZE_T yysize = yyssp - yyss + 1; + +#ifdef yyoverflow + { + /* Give user a chance to reallocate the stack. Use copies of + these so that the &'s don't force the real ones into + memory. */ + YYSTYPE *yyvs1 = yyvs; + yytype_int16 *yyss1 = yyss; + YYLTYPE *yyls1 = yyls; + + /* Each stack pointer address is followed by the size of the + data in use in that stack, in bytes. This used to be a + conditional around just the two extra args, but that might + be undefined if yyoverflow is a macro. */ + yyoverflow (YY_("memory exhausted"), + &yyss1, yysize * sizeof (*yyssp), + &yyvs1, yysize * sizeof (*yyvsp), + &yyls1, yysize * sizeof (*yylsp), + &yystacksize); + + yyls = yyls1; + yyss = yyss1; + yyvs = yyvs1; + } +#else /* no yyoverflow */ +# ifndef YYSTACK_RELOCATE + goto yyexhaustedlab; +# else + /* Extend the stack our own way. */ + if (YYMAXDEPTH <= yystacksize) + goto yyexhaustedlab; + yystacksize *= 2; + if (YYMAXDEPTH < yystacksize) + yystacksize = YYMAXDEPTH; + + { + yytype_int16 *yyss1 = yyss; + union yyalloc *yyptr = + (union yyalloc *) YYSTACK_ALLOC (YYSTACK_BYTES (yystacksize)); + if (! yyptr) + goto yyexhaustedlab; + YYSTACK_RELOCATE (yyss_alloc, yyss); + YYSTACK_RELOCATE (yyvs_alloc, yyvs); + YYSTACK_RELOCATE (yyls_alloc, yyls); +# undef YYSTACK_RELOCATE + if (yyss1 != yyssa) + YYSTACK_FREE (yyss1); + } +# endif +#endif /* no yyoverflow */ + + yyssp = yyss + yysize - 1; + yyvsp = yyvs + yysize - 1; + yylsp = yyls + yysize - 1; + + YYDPRINTF ((stderr, "Stack size increased to %lu\n", + (unsigned long int) yystacksize)); + + if (yyss + yystacksize - 1 <= yyssp) + YYABORT; + } + + YYDPRINTF ((stderr, "Entering state %d\n", yystate)); + + if (yystate == YYFINAL) + YYACCEPT; + + goto yybackup; + +/*-----------. +| yybackup. | +`-----------*/ +yybackup: + + /* Do appropriate processing given the current state. Read a + lookahead token if we need one and don't already have one. */ + + /* First try to decide what to do without reference to lookahead token. */ + yyn = yypact[yystate]; + if (yyn == YYPACT_NINF) + goto yydefault; + + /* Not known => get a lookahead token if don't already have one. */ + + /* YYCHAR is either YYEMPTY or YYEOF or a valid lookahead symbol. */ + if (yychar == YYEMPTY) + { + YYDPRINTF ((stderr, "Reading a token: ")); + yychar = YYLEX; + } + + if (yychar <= YYEOF) + { + yychar = yytoken = YYEOF; + YYDPRINTF ((stderr, "Now at end of input.\n")); + } + else + { + yytoken = YYTRANSLATE (yychar); + YY_SYMBOL_PRINT ("Next token is", yytoken, &yylval, &yylloc); + } + + /* If the proper action on seeing token YYTOKEN is to reduce or to + detect an error, take that action. */ + yyn += yytoken; + if (yyn < 0 || YYLAST < yyn || yycheck[yyn] != yytoken) + goto yydefault; + yyn = yytable[yyn]; + if (yyn <= 0) + { + if (yyn == 0 || yyn == YYTABLE_NINF) + goto yyerrlab; + yyn = -yyn; + goto yyreduce; + } + + /* Count tokens shifted since error; after three, turn off error + status. */ + if (yyerrstatus) + yyerrstatus--; + + /* Shift the lookahead token. */ + YY_SYMBOL_PRINT ("Shifting", yytoken, &yylval, &yylloc); + + /* Discard the shifted token. */ + yychar = YYEMPTY; + + yystate = yyn; + *++yyvsp = yylval; + *++yylsp = yylloc; + goto yynewstate; + + +/*-----------------------------------------------------------. +| yydefault -- do the default action for the current state. | +`-----------------------------------------------------------*/ +yydefault: + yyn = yydefact[yystate]; + if (yyn == 0) + goto yyerrlab; + goto yyreduce; + + +/*-----------------------------. +| yyreduce -- Do a reduction. | +`-----------------------------*/ +yyreduce: + /* yyn is the number of a rule to reduce with. */ + yylen = yyr2[yyn]; + + /* If YYLEN is nonzero, implement the default value of the action: + `$$ = $1'. + + Otherwise, the following line sets YYVAL to garbage. + This behavior is undocumented and Bison + users should not rely upon it. Assigning to YYVAL + unconditionally makes the parser a bit smaller, and it avoids a + GCC warning that YYVAL may be used uninitialized. */ + yyval = yyvsp[1-yylen]; + + /* Default location. */ + YYLLOC_DEFAULT (yyloc, (yylsp - yylen), yylen); + YY_REDUCE_PRINT (yyn); + switch (yyn) + { + case 3: + +/* Line 1455 of yacc.c */ +#line 282 "program_parse.y" + { + if (state->prog->Target != GL_VERTEX_PROGRAM_ARB) { + yyerror(& (yylsp[(1) - (1)]), state, "invalid fragment program header"); + + } + state->mode = ARB_vertex; + ;} + break; + + case 4: + +/* Line 1455 of yacc.c */ +#line 290 "program_parse.y" + { + if (state->prog->Target != GL_FRAGMENT_PROGRAM_ARB) { + yyerror(& (yylsp[(1) - (1)]), state, "invalid vertex program header"); + } + state->mode = ARB_fragment; + + state->option.TexRect = + (state->ctx->Extensions.NV_texture_rectangle != GL_FALSE); + ;} + break; + + case 7: + +/* Line 1455 of yacc.c */ +#line 306 "program_parse.y" + { + int valid = 0; + + if (state->mode == ARB_vertex) { + valid = _mesa_ARBvp_parse_option(state, (yyvsp[(2) - (3)].string)); + } else if (state->mode == ARB_fragment) { + valid = _mesa_ARBfp_parse_option(state, (yyvsp[(2) - (3)].string)); + } + + + free((yyvsp[(2) - (3)].string)); + + if (!valid) { + const char *const err_str = (state->mode == ARB_vertex) + ? "invalid ARB vertex program option" + : "invalid ARB fragment program option"; + + yyerror(& (yylsp[(2) - (3)]), state, err_str); + YYERROR; + } + ;} + break; + + case 10: + +/* Line 1455 of yacc.c */ +#line 334 "program_parse.y" + { + if ((yyvsp[(1) - (2)].inst) != NULL) { + if (state->inst_tail == NULL) { + state->inst_head = (yyvsp[(1) - (2)].inst); + } else { + state->inst_tail->next = (yyvsp[(1) - (2)].inst); + } + + state->inst_tail = (yyvsp[(1) - (2)].inst); + (yyvsp[(1) - (2)].inst)->next = NULL; + + state->prog->NumInstructions++; + } + ;} + break; + + case 12: + +/* Line 1455 of yacc.c */ +#line 352 "program_parse.y" + { + (yyval.inst) = (yyvsp[(1) - (1)].inst); + state->prog->NumAluInstructions++; + ;} + break; + + case 13: + +/* Line 1455 of yacc.c */ +#line 357 "program_parse.y" + { + (yyval.inst) = (yyvsp[(1) - (1)].inst); + state->prog->NumTexInstructions++; + ;} + break; + + case 24: + +/* Line 1455 of yacc.c */ +#line 378 "program_parse.y" + { + (yyval.inst) = asm_instruction_ctor(OPCODE_ARL, & (yyvsp[(2) - (4)].dst_reg), & (yyvsp[(4) - (4)].src_reg), NULL, NULL); + ;} + break; + + case 25: + +/* Line 1455 of yacc.c */ +#line 384 "program_parse.y" + { + (yyval.inst) = asm_instruction_copy_ctor(& (yyvsp[(1) - (4)].temp_inst), & (yyvsp[(2) - (4)].dst_reg), & (yyvsp[(4) - (4)].src_reg), NULL, NULL); + ;} + break; + + case 26: + +/* Line 1455 of yacc.c */ +#line 390 "program_parse.y" + { + (yyval.inst) = asm_instruction_copy_ctor(& (yyvsp[(1) - (4)].temp_inst), & (yyvsp[(2) - (4)].dst_reg), & (yyvsp[(4) - (4)].src_reg), NULL, NULL); + ;} + break; + + case 27: + +/* Line 1455 of yacc.c */ +#line 396 "program_parse.y" + { + (yyval.inst) = asm_instruction_copy_ctor(& (yyvsp[(1) - (6)].temp_inst), & (yyvsp[(2) - (6)].dst_reg), & (yyvsp[(4) - (6)].src_reg), & (yyvsp[(6) - (6)].src_reg), NULL); + ;} + break; + + case 28: + +/* Line 1455 of yacc.c */ +#line 403 "program_parse.y" + { + (yyval.inst) = asm_instruction_copy_ctor(& (yyvsp[(1) - (6)].temp_inst), & (yyvsp[(2) - (6)].dst_reg), & (yyvsp[(4) - (6)].src_reg), & (yyvsp[(6) - (6)].src_reg), NULL); + ;} + break; + + case 29: + +/* Line 1455 of yacc.c */ +#line 410 "program_parse.y" + { + (yyval.inst) = asm_instruction_copy_ctor(& (yyvsp[(1) - (8)].temp_inst), & (yyvsp[(2) - (8)].dst_reg), & (yyvsp[(4) - (8)].src_reg), & (yyvsp[(6) - (8)].src_reg), & (yyvsp[(8) - (8)].src_reg)); + ;} + break; + + case 30: + +/* Line 1455 of yacc.c */ +#line 416 "program_parse.y" + { + (yyval.inst) = asm_instruction_copy_ctor(& (yyvsp[(1) - (8)].temp_inst), & (yyvsp[(2) - (8)].dst_reg), & (yyvsp[(4) - (8)].src_reg), NULL, NULL); + if ((yyval.inst) != NULL) { + const GLbitfield tex_mask = (1U << (yyvsp[(6) - (8)].integer)); + GLbitfield shadow_tex = 0; + GLbitfield target_mask = 0; + + + (yyval.inst)->Base.TexSrcUnit = (yyvsp[(6) - (8)].integer); + + if ((yyvsp[(8) - (8)].integer) < 0) { + shadow_tex = tex_mask; + + (yyval.inst)->Base.TexSrcTarget = -(yyvsp[(8) - (8)].integer); + (yyval.inst)->Base.TexShadow = 1; + } else { + (yyval.inst)->Base.TexSrcTarget = (yyvsp[(8) - (8)].integer); + } + + target_mask = (1U << (yyval.inst)->Base.TexSrcTarget); + + /* If this texture unit was previously accessed and that access + * had a different texture target, generate an error. + * + * If this texture unit was previously accessed and that access + * had a different shadow mode, generate an error. + */ + if ((state->prog->TexturesUsed[(yyvsp[(6) - (8)].integer)] != 0) + && ((state->prog->TexturesUsed[(yyvsp[(6) - (8)].integer)] != target_mask) + || ((state->prog->ShadowSamplers & tex_mask) + != shadow_tex))) { + yyerror(& (yylsp[(8) - (8)]), state, + "multiple targets used on one texture image unit"); + YYERROR; + } + + + state->prog->TexturesUsed[(yyvsp[(6) - (8)].integer)] |= target_mask; + state->prog->ShadowSamplers |= shadow_tex; + } + ;} + break; + + case 31: + +/* Line 1455 of yacc.c */ +#line 460 "program_parse.y" + { + (yyval.inst) = asm_instruction_ctor(OPCODE_KIL, NULL, & (yyvsp[(2) - (2)].src_reg), NULL, NULL); + state->fragment.UsesKill = 1; + ;} + break; + + case 32: + +/* Line 1455 of yacc.c */ +#line 465 "program_parse.y" + { + (yyval.inst) = asm_instruction_ctor(OPCODE_KIL_NV, NULL, NULL, NULL, NULL); + (yyval.inst)->Base.DstReg.CondMask = (yyvsp[(2) - (2)].dst_reg).CondMask; + (yyval.inst)->Base.DstReg.CondSwizzle = (yyvsp[(2) - (2)].dst_reg).CondSwizzle; + (yyval.inst)->Base.DstReg.CondSrc = (yyvsp[(2) - (2)].dst_reg).CondSrc; + state->fragment.UsesKill = 1; + ;} + break; + + case 33: + +/* Line 1455 of yacc.c */ +#line 475 "program_parse.y" + { + (yyval.inst) = asm_instruction_copy_ctor(& (yyvsp[(1) - (12)].temp_inst), & (yyvsp[(2) - (12)].dst_reg), & (yyvsp[(4) - (12)].src_reg), & (yyvsp[(6) - (12)].src_reg), & (yyvsp[(8) - (12)].src_reg)); + if ((yyval.inst) != NULL) { + const GLbitfield tex_mask = (1U << (yyvsp[(10) - (12)].integer)); + GLbitfield shadow_tex = 0; + GLbitfield target_mask = 0; + + + (yyval.inst)->Base.TexSrcUnit = (yyvsp[(10) - (12)].integer); + + if ((yyvsp[(12) - (12)].integer) < 0) { + shadow_tex = tex_mask; + + (yyval.inst)->Base.TexSrcTarget = -(yyvsp[(12) - (12)].integer); + (yyval.inst)->Base.TexShadow = 1; + } else { + (yyval.inst)->Base.TexSrcTarget = (yyvsp[(12) - (12)].integer); + } + + target_mask = (1U << (yyval.inst)->Base.TexSrcTarget); + + /* If this texture unit was previously accessed and that access + * had a different texture target, generate an error. + * + * If this texture unit was previously accessed and that access + * had a different shadow mode, generate an error. + */ + if ((state->prog->TexturesUsed[(yyvsp[(10) - (12)].integer)] != 0) + && ((state->prog->TexturesUsed[(yyvsp[(10) - (12)].integer)] != target_mask) + || ((state->prog->ShadowSamplers & tex_mask) + != shadow_tex))) { + yyerror(& (yylsp[(12) - (12)]), state, + "multiple targets used on one texture image unit"); + YYERROR; + } + + + state->prog->TexturesUsed[(yyvsp[(10) - (12)].integer)] |= target_mask; + state->prog->ShadowSamplers |= shadow_tex; + } + ;} + break; + + case 34: + +/* Line 1455 of yacc.c */ +#line 519 "program_parse.y" + { + (yyval.integer) = (yyvsp[(2) - (2)].integer); + ;} + break; + + case 35: + +/* Line 1455 of yacc.c */ +#line 524 "program_parse.y" + { (yyval.integer) = TEXTURE_1D_INDEX; ;} + break; + + case 36: + +/* Line 1455 of yacc.c */ +#line 525 "program_parse.y" + { (yyval.integer) = TEXTURE_2D_INDEX; ;} + break; + + case 37: + +/* Line 1455 of yacc.c */ +#line 526 "program_parse.y" + { (yyval.integer) = TEXTURE_3D_INDEX; ;} + break; + + case 38: + +/* Line 1455 of yacc.c */ +#line 527 "program_parse.y" + { (yyval.integer) = TEXTURE_CUBE_INDEX; ;} + break; + + case 39: + +/* Line 1455 of yacc.c */ +#line 528 "program_parse.y" + { (yyval.integer) = TEXTURE_RECT_INDEX; ;} + break; + + case 40: + +/* Line 1455 of yacc.c */ +#line 529 "program_parse.y" + { (yyval.integer) = -TEXTURE_1D_INDEX; ;} + break; + + case 41: + +/* Line 1455 of yacc.c */ +#line 530 "program_parse.y" + { (yyval.integer) = -TEXTURE_2D_INDEX; ;} + break; + + case 42: + +/* Line 1455 of yacc.c */ +#line 531 "program_parse.y" + { (yyval.integer) = -TEXTURE_RECT_INDEX; ;} + break; + + case 43: + +/* Line 1455 of yacc.c */ +#line 532 "program_parse.y" + { (yyval.integer) = TEXTURE_1D_ARRAY_INDEX; ;} + break; + + case 44: + +/* Line 1455 of yacc.c */ +#line 533 "program_parse.y" + { (yyval.integer) = TEXTURE_2D_ARRAY_INDEX; ;} + break; + + case 45: + +/* Line 1455 of yacc.c */ +#line 534 "program_parse.y" + { (yyval.integer) = -TEXTURE_1D_ARRAY_INDEX; ;} + break; + + case 46: + +/* Line 1455 of yacc.c */ +#line 535 "program_parse.y" + { (yyval.integer) = -TEXTURE_2D_ARRAY_INDEX; ;} + break; + + case 47: + +/* Line 1455 of yacc.c */ +#line 539 "program_parse.y" + { + /* FIXME: Is this correct? Should the extenedSwizzle be applied + * FIXME: to the existing swizzle? + */ + (yyvsp[(4) - (6)].src_reg).Base.Swizzle = (yyvsp[(6) - (6)].swiz_mask).swizzle; + (yyvsp[(4) - (6)].src_reg).Base.Negate = (yyvsp[(6) - (6)].swiz_mask).mask; + + (yyval.inst) = asm_instruction_copy_ctor(& (yyvsp[(1) - (6)].temp_inst), & (yyvsp[(2) - (6)].dst_reg), & (yyvsp[(4) - (6)].src_reg), NULL, NULL); + ;} + break; + + case 48: + +/* Line 1455 of yacc.c */ +#line 551 "program_parse.y" + { + (yyval.src_reg) = (yyvsp[(2) - (2)].src_reg); + + if ((yyvsp[(1) - (2)].negate)) { + (yyval.src_reg).Base.Negate = ~(yyval.src_reg).Base.Negate; + } + ;} + break; + + case 49: + +/* Line 1455 of yacc.c */ +#line 559 "program_parse.y" + { + (yyval.src_reg) = (yyvsp[(3) - (4)].src_reg); + + if (!state->option.NV_fragment) { + yyerror(& (yylsp[(2) - (4)]), state, "unexpected character '|'"); + YYERROR; + } + + if ((yyvsp[(1) - (4)].negate)) { + (yyval.src_reg).Base.Negate = ~(yyval.src_reg).Base.Negate; + } + + (yyval.src_reg).Base.Abs = 1; + ;} + break; + + case 50: + +/* Line 1455 of yacc.c */ +#line 576 "program_parse.y" + { + (yyval.src_reg) = (yyvsp[(1) - (2)].src_reg); + + (yyval.src_reg).Base.Swizzle = _mesa_combine_swizzles((yyval.src_reg).Base.Swizzle, + (yyvsp[(2) - (2)].swiz_mask).swizzle); + ;} + break; + + case 51: + +/* Line 1455 of yacc.c */ +#line 583 "program_parse.y" + { + struct asm_symbol temp_sym; + + if (!state->option.NV_fragment) { + yyerror(& (yylsp[(1) - (1)]), state, "expected scalar suffix"); + YYERROR; + } + + memset(& temp_sym, 0, sizeof(temp_sym)); + temp_sym.param_binding_begin = ~0; + initialize_symbol_from_const(state->prog, & temp_sym, & (yyvsp[(1) - (1)].vector), GL_TRUE); + + set_src_reg_swz(& (yyval.src_reg), PROGRAM_CONSTANT, + temp_sym.param_binding_begin, + temp_sym.param_binding_swizzle); + ;} + break; + + case 52: + +/* Line 1455 of yacc.c */ +#line 602 "program_parse.y" + { + (yyval.src_reg) = (yyvsp[(2) - (3)].src_reg); + + if ((yyvsp[(1) - (3)].negate)) { + (yyval.src_reg).Base.Negate = ~(yyval.src_reg).Base.Negate; + } + + (yyval.src_reg).Base.Swizzle = _mesa_combine_swizzles((yyval.src_reg).Base.Swizzle, + (yyvsp[(3) - (3)].swiz_mask).swizzle); + ;} + break; + + case 53: + +/* Line 1455 of yacc.c */ +#line 613 "program_parse.y" + { + (yyval.src_reg) = (yyvsp[(3) - (5)].src_reg); + + if (!state->option.NV_fragment) { + yyerror(& (yylsp[(2) - (5)]), state, "unexpected character '|'"); + YYERROR; + } + + if ((yyvsp[(1) - (5)].negate)) { + (yyval.src_reg).Base.Negate = ~(yyval.src_reg).Base.Negate; + } + + (yyval.src_reg).Base.Abs = 1; + (yyval.src_reg).Base.Swizzle = _mesa_combine_swizzles((yyval.src_reg).Base.Swizzle, + (yyvsp[(4) - (5)].swiz_mask).swizzle); + ;} + break; + + case 54: + +/* Line 1455 of yacc.c */ +#line 633 "program_parse.y" + { + (yyval.dst_reg) = (yyvsp[(1) - (3)].dst_reg); + (yyval.dst_reg).WriteMask = (yyvsp[(2) - (3)].swiz_mask).mask; + (yyval.dst_reg).CondMask = (yyvsp[(3) - (3)].dst_reg).CondMask; + (yyval.dst_reg).CondSwizzle = (yyvsp[(3) - (3)].dst_reg).CondSwizzle; + (yyval.dst_reg).CondSrc = (yyvsp[(3) - (3)].dst_reg).CondSrc; + + if ((yyval.dst_reg).File == PROGRAM_OUTPUT) { + /* Technically speaking, this should check that it is in + * vertex program mode. However, PositionInvariant can never be + * set in fragment program mode, so it is somewhat irrelevant. + */ + if (state->option.PositionInvariant + && ((yyval.dst_reg).Index == VERT_RESULT_HPOS)) { + yyerror(& (yylsp[(1) - (3)]), state, "position-invariant programs cannot " + "write position"); + YYERROR; + } + + state->prog->OutputsWritten |= BITFIELD64_BIT((yyval.dst_reg).Index); + } + ;} + break; + + case 55: + +/* Line 1455 of yacc.c */ +#line 658 "program_parse.y" + { + set_dst_reg(& (yyval.dst_reg), PROGRAM_ADDRESS, 0); + (yyval.dst_reg).WriteMask = (yyvsp[(2) - (2)].swiz_mask).mask; + ;} + break; + + case 56: + +/* Line 1455 of yacc.c */ +#line 665 "program_parse.y" + { + const unsigned xyzw_valid = + ((yyvsp[(1) - (7)].ext_swizzle).xyzw_valid << 0) + | ((yyvsp[(3) - (7)].ext_swizzle).xyzw_valid << 1) + | ((yyvsp[(5) - (7)].ext_swizzle).xyzw_valid << 2) + | ((yyvsp[(7) - (7)].ext_swizzle).xyzw_valid << 3); + const unsigned rgba_valid = + ((yyvsp[(1) - (7)].ext_swizzle).rgba_valid << 0) + | ((yyvsp[(3) - (7)].ext_swizzle).rgba_valid << 1) + | ((yyvsp[(5) - (7)].ext_swizzle).rgba_valid << 2) + | ((yyvsp[(7) - (7)].ext_swizzle).rgba_valid << 3); + + /* All of the swizzle components have to be valid in either RGBA + * or XYZW. Note that 0 and 1 are valid in both, so both masks + * can have some bits set. + * + * We somewhat deviate from the spec here. It would be really hard + * to figure out which component is the error, and there probably + * isn't a lot of benefit. + */ + if ((rgba_valid != 0x0f) && (xyzw_valid != 0x0f)) { + yyerror(& (yylsp[(1) - (7)]), state, "cannot combine RGBA and XYZW swizzle " + "components"); + YYERROR; + } + + (yyval.swiz_mask).swizzle = MAKE_SWIZZLE4((yyvsp[(1) - (7)].ext_swizzle).swz, (yyvsp[(3) - (7)].ext_swizzle).swz, (yyvsp[(5) - (7)].ext_swizzle).swz, (yyvsp[(7) - (7)].ext_swizzle).swz); + (yyval.swiz_mask).mask = ((yyvsp[(1) - (7)].ext_swizzle).negate) | ((yyvsp[(3) - (7)].ext_swizzle).negate << 1) | ((yyvsp[(5) - (7)].ext_swizzle).negate << 2) + | ((yyvsp[(7) - (7)].ext_swizzle).negate << 3); + ;} + break; + + case 57: + +/* Line 1455 of yacc.c */ +#line 698 "program_parse.y" + { + (yyval.ext_swizzle) = (yyvsp[(2) - (2)].ext_swizzle); + (yyval.ext_swizzle).negate = ((yyvsp[(1) - (2)].negate)) ? 1 : 0; + ;} + break; + + case 58: + +/* Line 1455 of yacc.c */ +#line 705 "program_parse.y" + { + if (((yyvsp[(1) - (1)].integer) != 0) && ((yyvsp[(1) - (1)].integer) != 1)) { + yyerror(& (yylsp[(1) - (1)]), state, "invalid extended swizzle selector"); + YYERROR; + } + + (yyval.ext_swizzle).swz = ((yyvsp[(1) - (1)].integer) == 0) ? SWIZZLE_ZERO : SWIZZLE_ONE; + + /* 0 and 1 are valid for both RGBA swizzle names and XYZW + * swizzle names. + */ + (yyval.ext_swizzle).xyzw_valid = 1; + (yyval.ext_swizzle).rgba_valid = 1; + ;} + break; + + case 59: + +/* Line 1455 of yacc.c */ +#line 720 "program_parse.y" + { + char s; + + if (strlen((yyvsp[(1) - (1)].string)) > 1) { + yyerror(& (yylsp[(1) - (1)]), state, "invalid extended swizzle selector"); + YYERROR; + } + + s = (yyvsp[(1) - (1)].string)[0]; + free((yyvsp[(1) - (1)].string)); + + switch (s) { + case 'x': + (yyval.ext_swizzle).swz = SWIZZLE_X; + (yyval.ext_swizzle).xyzw_valid = 1; + break; + case 'y': + (yyval.ext_swizzle).swz = SWIZZLE_Y; + (yyval.ext_swizzle).xyzw_valid = 1; + break; + case 'z': + (yyval.ext_swizzle).swz = SWIZZLE_Z; + (yyval.ext_swizzle).xyzw_valid = 1; + break; + case 'w': + (yyval.ext_swizzle).swz = SWIZZLE_W; + (yyval.ext_swizzle).xyzw_valid = 1; + break; + + case 'r': + (yyval.ext_swizzle).swz = SWIZZLE_X; + (yyval.ext_swizzle).rgba_valid = 1; + break; + case 'g': + (yyval.ext_swizzle).swz = SWIZZLE_Y; + (yyval.ext_swizzle).rgba_valid = 1; + break; + case 'b': + (yyval.ext_swizzle).swz = SWIZZLE_Z; + (yyval.ext_swizzle).rgba_valid = 1; + break; + case 'a': + (yyval.ext_swizzle).swz = SWIZZLE_W; + (yyval.ext_swizzle).rgba_valid = 1; + break; + + default: + yyerror(& (yylsp[(1) - (1)]), state, "invalid extended swizzle selector"); + YYERROR; + break; + } + ;} + break; + + case 60: + +/* Line 1455 of yacc.c */ +#line 775 "program_parse.y" + { + struct asm_symbol *const s = (struct asm_symbol *) + _mesa_symbol_table_find_symbol(state->st, 0, (yyvsp[(1) - (1)].string)); + + free((yyvsp[(1) - (1)].string)); + + if (s == NULL) { + yyerror(& (yylsp[(1) - (1)]), state, "invalid operand variable"); + YYERROR; + } else if ((s->type != at_param) && (s->type != at_temp) + && (s->type != at_attrib)) { + yyerror(& (yylsp[(1) - (1)]), state, "invalid operand variable"); + YYERROR; + } else if ((s->type == at_param) && s->param_is_array) { + yyerror(& (yylsp[(1) - (1)]), state, "non-array access to array PARAM"); + YYERROR; + } + + init_src_reg(& (yyval.src_reg)); + switch (s->type) { + case at_temp: + set_src_reg(& (yyval.src_reg), PROGRAM_TEMPORARY, s->temp_binding); + break; + case at_param: + set_src_reg_swz(& (yyval.src_reg), s->param_binding_type, + s->param_binding_begin, + s->param_binding_swizzle); + break; + case at_attrib: + set_src_reg(& (yyval.src_reg), PROGRAM_INPUT, s->attrib_binding); + state->prog->InputsRead |= (1U << (yyval.src_reg).Base.Index); + + if (!validate_inputs(& (yylsp[(1) - (1)]), state)) { + YYERROR; + } + break; + + default: + YYERROR; + break; + } + ;} + break; + + case 61: + +/* Line 1455 of yacc.c */ +#line 818 "program_parse.y" + { + set_src_reg(& (yyval.src_reg), PROGRAM_INPUT, (yyvsp[(1) - (1)].attrib)); + state->prog->InputsRead |= (1U << (yyval.src_reg).Base.Index); + + if (!validate_inputs(& (yylsp[(1) - (1)]), state)) { + YYERROR; + } + ;} + break; + + case 62: + +/* Line 1455 of yacc.c */ +#line 827 "program_parse.y" + { + if (! (yyvsp[(3) - (4)].src_reg).Base.RelAddr + && ((unsigned) (yyvsp[(3) - (4)].src_reg).Base.Index >= (yyvsp[(1) - (4)].sym)->param_binding_length)) { + yyerror(& (yylsp[(3) - (4)]), state, "out of bounds array access"); + YYERROR; + } + + init_src_reg(& (yyval.src_reg)); + (yyval.src_reg).Base.File = (yyvsp[(1) - (4)].sym)->param_binding_type; + + if ((yyvsp[(3) - (4)].src_reg).Base.RelAddr) { + state->prog->IndirectRegisterFiles |= (1 << (yyval.src_reg).Base.File); + (yyvsp[(1) - (4)].sym)->param_accessed_indirectly = 1; + + (yyval.src_reg).Base.RelAddr = 1; + (yyval.src_reg).Base.Index = (yyvsp[(3) - (4)].src_reg).Base.Index; + (yyval.src_reg).Symbol = (yyvsp[(1) - (4)].sym); + } else { + (yyval.src_reg).Base.Index = (yyvsp[(1) - (4)].sym)->param_binding_begin + (yyvsp[(3) - (4)].src_reg).Base.Index; + } + ;} + break; + + case 63: + +/* Line 1455 of yacc.c */ +#line 849 "program_parse.y" + { + gl_register_file file = ((yyvsp[(1) - (1)].temp_sym).name != NULL) + ? (yyvsp[(1) - (1)].temp_sym).param_binding_type + : PROGRAM_CONSTANT; + set_src_reg_swz(& (yyval.src_reg), file, (yyvsp[(1) - (1)].temp_sym).param_binding_begin, + (yyvsp[(1) - (1)].temp_sym).param_binding_swizzle); + ;} + break; + + case 64: + +/* Line 1455 of yacc.c */ +#line 859 "program_parse.y" + { + set_dst_reg(& (yyval.dst_reg), PROGRAM_OUTPUT, (yyvsp[(1) - (1)].result)); + ;} + break; + + case 65: + +/* Line 1455 of yacc.c */ +#line 863 "program_parse.y" + { + struct asm_symbol *const s = (struct asm_symbol *) + _mesa_symbol_table_find_symbol(state->st, 0, (yyvsp[(1) - (1)].string)); + + free((yyvsp[(1) - (1)].string)); + + if (s == NULL) { + yyerror(& (yylsp[(1) - (1)]), state, "invalid operand variable"); + YYERROR; + } else if ((s->type != at_output) && (s->type != at_temp)) { + yyerror(& (yylsp[(1) - (1)]), state, "invalid operand variable"); + YYERROR; + } + + switch (s->type) { + case at_temp: + set_dst_reg(& (yyval.dst_reg), PROGRAM_TEMPORARY, s->temp_binding); + break; + case at_output: + set_dst_reg(& (yyval.dst_reg), PROGRAM_OUTPUT, s->output_binding); + break; + default: + set_dst_reg(& (yyval.dst_reg), s->param_binding_type, s->param_binding_begin); + break; + } + ;} + break; + + case 66: + +/* Line 1455 of yacc.c */ +#line 892 "program_parse.y" + { + struct asm_symbol *const s = (struct asm_symbol *) + _mesa_symbol_table_find_symbol(state->st, 0, (yyvsp[(1) - (1)].string)); + + free((yyvsp[(1) - (1)].string)); + + if (s == NULL) { + yyerror(& (yylsp[(1) - (1)]), state, "invalid operand variable"); + YYERROR; + } else if ((s->type != at_param) || !s->param_is_array) { + yyerror(& (yylsp[(1) - (1)]), state, "array access to non-PARAM variable"); + YYERROR; + } else { + (yyval.sym) = s; + } + ;} + break; + + case 69: + +/* Line 1455 of yacc.c */ +#line 913 "program_parse.y" + { + init_src_reg(& (yyval.src_reg)); + (yyval.src_reg).Base.Index = (yyvsp[(1) - (1)].integer); + ;} + break; + + case 70: + +/* Line 1455 of yacc.c */ +#line 920 "program_parse.y" + { + /* FINISHME: Add support for multiple address registers. + */ + /* FINISHME: Add support for 4-component address registers. + */ + init_src_reg(& (yyval.src_reg)); + (yyval.src_reg).Base.RelAddr = 1; + (yyval.src_reg).Base.Index = (yyvsp[(3) - (3)].integer); + ;} + break; + + case 71: + +/* Line 1455 of yacc.c */ +#line 931 "program_parse.y" + { (yyval.integer) = 0; ;} + break; + + case 72: + +/* Line 1455 of yacc.c */ +#line 932 "program_parse.y" + { (yyval.integer) = (yyvsp[(2) - (2)].integer); ;} + break; + + case 73: + +/* Line 1455 of yacc.c */ +#line 933 "program_parse.y" + { (yyval.integer) = -(yyvsp[(2) - (2)].integer); ;} + break; + + case 74: + +/* Line 1455 of yacc.c */ +#line 937 "program_parse.y" + { + if (((yyvsp[(1) - (1)].integer) < 0) || ((yyvsp[(1) - (1)].integer) > 4095)) { + char s[100]; + _mesa_snprintf(s, sizeof(s), + "relative address offset too large (%d)", (yyvsp[(1) - (1)].integer)); + yyerror(& (yylsp[(1) - (1)]), state, s); + YYERROR; + } else { + (yyval.integer) = (yyvsp[(1) - (1)].integer); + } + ;} + break; + + case 75: + +/* Line 1455 of yacc.c */ +#line 951 "program_parse.y" + { + if (((yyvsp[(1) - (1)].integer) < 0) || ((yyvsp[(1) - (1)].integer) > 4096)) { + char s[100]; + _mesa_snprintf(s, sizeof(s), + "relative address offset too large (%d)", (yyvsp[(1) - (1)].integer)); + yyerror(& (yylsp[(1) - (1)]), state, s); + YYERROR; + } else { + (yyval.integer) = (yyvsp[(1) - (1)].integer); + } + ;} + break; + + case 76: + +/* Line 1455 of yacc.c */ +#line 965 "program_parse.y" + { + struct asm_symbol *const s = (struct asm_symbol *) + _mesa_symbol_table_find_symbol(state->st, 0, (yyvsp[(1) - (1)].string)); + + free((yyvsp[(1) - (1)].string)); + + if (s == NULL) { + yyerror(& (yylsp[(1) - (1)]), state, "invalid array member"); + YYERROR; + } else if (s->type != at_address) { + yyerror(& (yylsp[(1) - (1)]), state, + "invalid variable for indexed array access"); + YYERROR; + } else { + (yyval.sym) = s; + } + ;} + break; + + case 77: + +/* Line 1455 of yacc.c */ +#line 985 "program_parse.y" + { + if ((yyvsp[(1) - (1)].swiz_mask).mask != WRITEMASK_X) { + yyerror(& (yylsp[(1) - (1)]), state, "invalid address component selector"); + YYERROR; + } else { + (yyval.swiz_mask) = (yyvsp[(1) - (1)].swiz_mask); + } + ;} + break; + + case 78: + +/* Line 1455 of yacc.c */ +#line 996 "program_parse.y" + { + if ((yyvsp[(1) - (1)].swiz_mask).mask != WRITEMASK_X) { + yyerror(& (yylsp[(1) - (1)]), state, + "address register write mask must be \".x\""); + YYERROR; + } else { + (yyval.swiz_mask) = (yyvsp[(1) - (1)].swiz_mask); + } + ;} + break; + + case 83: + +/* Line 1455 of yacc.c */ +#line 1012 "program_parse.y" + { (yyval.swiz_mask).swizzle = SWIZZLE_NOOP; (yyval.swiz_mask).mask = WRITEMASK_XYZW; ;} + break; + + case 88: + +/* Line 1455 of yacc.c */ +#line 1016 "program_parse.y" + { (yyval.swiz_mask).swizzle = SWIZZLE_NOOP; (yyval.swiz_mask).mask = WRITEMASK_XYZW; ;} + break; + + case 89: + +/* Line 1455 of yacc.c */ +#line 1020 "program_parse.y" + { + (yyval.dst_reg) = (yyvsp[(2) - (3)].dst_reg); + ;} + break; + + case 90: + +/* Line 1455 of yacc.c */ +#line 1024 "program_parse.y" + { + (yyval.dst_reg) = (yyvsp[(2) - (3)].dst_reg); + ;} + break; + + case 91: + +/* Line 1455 of yacc.c */ +#line 1028 "program_parse.y" + { + (yyval.dst_reg).CondMask = COND_TR; + (yyval.dst_reg).CondSwizzle = SWIZZLE_NOOP; + (yyval.dst_reg).CondSrc = 0; + ;} + break; + + case 92: + +/* Line 1455 of yacc.c */ +#line 1036 "program_parse.y" + { + (yyval.dst_reg) = (yyvsp[(1) - (2)].dst_reg); + (yyval.dst_reg).CondSwizzle = (yyvsp[(2) - (2)].swiz_mask).swizzle; + ;} + break; + + case 93: + +/* Line 1455 of yacc.c */ +#line 1043 "program_parse.y" + { + (yyval.dst_reg) = (yyvsp[(1) - (2)].dst_reg); + (yyval.dst_reg).CondSwizzle = (yyvsp[(2) - (2)].swiz_mask).swizzle; + ;} + break; + + case 94: + +/* Line 1455 of yacc.c */ +#line 1050 "program_parse.y" + { + const int cond = _mesa_parse_cc((yyvsp[(1) - (1)].string)); + if ((cond == 0) || ((yyvsp[(1) - (1)].string)[2] != '\0')) { + char *const err_str = + make_error_string("invalid condition code \"%s\"", (yyvsp[(1) - (1)].string)); + + yyerror(& (yylsp[(1) - (1)]), state, (err_str != NULL) + ? err_str : "invalid condition code"); + + if (err_str != NULL) { + free(err_str); + } + + YYERROR; + } + + (yyval.dst_reg).CondMask = cond; + (yyval.dst_reg).CondSwizzle = SWIZZLE_NOOP; + (yyval.dst_reg).CondSrc = 0; + ;} + break; + + case 95: + +/* Line 1455 of yacc.c */ +#line 1073 "program_parse.y" + { + const int cond = _mesa_parse_cc((yyvsp[(1) - (1)].string)); + if ((cond == 0) || ((yyvsp[(1) - (1)].string)[2] != '\0')) { + char *const err_str = + make_error_string("invalid condition code \"%s\"", (yyvsp[(1) - (1)].string)); + + yyerror(& (yylsp[(1) - (1)]), state, (err_str != NULL) + ? err_str : "invalid condition code"); + + if (err_str != NULL) { + free(err_str); + } + + YYERROR; + } + + (yyval.dst_reg).CondMask = cond; + (yyval.dst_reg).CondSwizzle = SWIZZLE_NOOP; + (yyval.dst_reg).CondSrc = 0; + ;} + break; + + case 102: + +/* Line 1455 of yacc.c */ +#line 1104 "program_parse.y" + { + struct asm_symbol *const s = + declare_variable(state, (yyvsp[(2) - (4)].string), at_attrib, & (yylsp[(2) - (4)])); + + if (s == NULL) { + free((yyvsp[(2) - (4)].string)); + YYERROR; + } else { + s->attrib_binding = (yyvsp[(4) - (4)].attrib); + state->InputsBound |= (1U << s->attrib_binding); + + if (!validate_inputs(& (yylsp[(4) - (4)]), state)) { + YYERROR; + } + } + ;} + break; + + case 103: + +/* Line 1455 of yacc.c */ +#line 1123 "program_parse.y" + { + (yyval.attrib) = (yyvsp[(2) - (2)].attrib); + ;} + break; + + case 104: + +/* Line 1455 of yacc.c */ +#line 1127 "program_parse.y" + { + (yyval.attrib) = (yyvsp[(2) - (2)].attrib); + ;} + break; + + case 105: + +/* Line 1455 of yacc.c */ +#line 1133 "program_parse.y" + { + (yyval.attrib) = VERT_ATTRIB_POS; + ;} + break; + + case 106: + +/* Line 1455 of yacc.c */ +#line 1137 "program_parse.y" + { + (yyval.attrib) = VERT_ATTRIB_WEIGHT; + ;} + break; + + case 107: + +/* Line 1455 of yacc.c */ +#line 1141 "program_parse.y" + { + (yyval.attrib) = VERT_ATTRIB_NORMAL; + ;} + break; + + case 108: + +/* Line 1455 of yacc.c */ +#line 1145 "program_parse.y" + { + if (!state->ctx->Extensions.EXT_secondary_color) { + yyerror(& (yylsp[(2) - (2)]), state, "GL_EXT_secondary_color not supported"); + YYERROR; + } + + (yyval.attrib) = VERT_ATTRIB_COLOR0 + (yyvsp[(2) - (2)].integer); + ;} + break; + + case 109: + +/* Line 1455 of yacc.c */ +#line 1154 "program_parse.y" + { + if (!state->ctx->Extensions.EXT_fog_coord) { + yyerror(& (yylsp[(1) - (1)]), state, "GL_EXT_fog_coord not supported"); + YYERROR; + } + + (yyval.attrib) = VERT_ATTRIB_FOG; + ;} + break; + + case 110: + +/* Line 1455 of yacc.c */ +#line 1163 "program_parse.y" + { + (yyval.attrib) = VERT_ATTRIB_TEX0 + (yyvsp[(2) - (2)].integer); + ;} + break; + + case 111: + +/* Line 1455 of yacc.c */ +#line 1167 "program_parse.y" + { + yyerror(& (yylsp[(1) - (4)]), state, "GL_ARB_matrix_palette not supported"); + YYERROR; + ;} + break; + + case 112: + +/* Line 1455 of yacc.c */ +#line 1172 "program_parse.y" + { + (yyval.attrib) = VERT_ATTRIB_GENERIC0 + (yyvsp[(3) - (4)].integer); + ;} + break; + + case 113: + +/* Line 1455 of yacc.c */ +#line 1178 "program_parse.y" + { + if ((unsigned) (yyvsp[(1) - (1)].integer) >= state->limits->MaxAttribs) { + yyerror(& (yylsp[(1) - (1)]), state, "invalid vertex attribute reference"); + YYERROR; + } + + (yyval.integer) = (yyvsp[(1) - (1)].integer); + ;} + break; + + case 117: + +/* Line 1455 of yacc.c */ +#line 1192 "program_parse.y" + { + (yyval.attrib) = FRAG_ATTRIB_WPOS; + ;} + break; + + case 118: + +/* Line 1455 of yacc.c */ +#line 1196 "program_parse.y" + { + (yyval.attrib) = FRAG_ATTRIB_COL0 + (yyvsp[(2) - (2)].integer); + ;} + break; + + case 119: + +/* Line 1455 of yacc.c */ +#line 1200 "program_parse.y" + { + (yyval.attrib) = FRAG_ATTRIB_FOGC; + ;} + break; + + case 120: + +/* Line 1455 of yacc.c */ +#line 1204 "program_parse.y" + { + (yyval.attrib) = FRAG_ATTRIB_TEX0 + (yyvsp[(2) - (2)].integer); + ;} + break; + + case 123: + +/* Line 1455 of yacc.c */ +#line 1212 "program_parse.y" + { + struct asm_symbol *const s = + declare_variable(state, (yyvsp[(2) - (3)].string), at_param, & (yylsp[(2) - (3)])); + + if (s == NULL) { + free((yyvsp[(2) - (3)].string)); + YYERROR; + } else { + s->param_binding_type = (yyvsp[(3) - (3)].temp_sym).param_binding_type; + s->param_binding_begin = (yyvsp[(3) - (3)].temp_sym).param_binding_begin; + s->param_binding_length = (yyvsp[(3) - (3)].temp_sym).param_binding_length; + s->param_binding_swizzle = (yyvsp[(3) - (3)].temp_sym).param_binding_swizzle; + s->param_is_array = 0; + } + ;} + break; + + case 124: + +/* Line 1455 of yacc.c */ +#line 1230 "program_parse.y" + { + if (((yyvsp[(4) - (6)].integer) != 0) && ((unsigned) (yyvsp[(4) - (6)].integer) != (yyvsp[(6) - (6)].temp_sym).param_binding_length)) { + free((yyvsp[(2) - (6)].string)); + yyerror(& (yylsp[(4) - (6)]), state, + "parameter array size and number of bindings must match"); + YYERROR; + } else { + struct asm_symbol *const s = + declare_variable(state, (yyvsp[(2) - (6)].string), (yyvsp[(6) - (6)].temp_sym).type, & (yylsp[(2) - (6)])); + + if (s == NULL) { + free((yyvsp[(2) - (6)].string)); + YYERROR; + } else { + s->param_binding_type = (yyvsp[(6) - (6)].temp_sym).param_binding_type; + s->param_binding_begin = (yyvsp[(6) - (6)].temp_sym).param_binding_begin; + s->param_binding_length = (yyvsp[(6) - (6)].temp_sym).param_binding_length; + s->param_binding_swizzle = SWIZZLE_XYZW; + s->param_is_array = 1; + } + } + ;} + break; + + case 125: + +/* Line 1455 of yacc.c */ +#line 1255 "program_parse.y" + { + (yyval.integer) = 0; + ;} + break; + + case 126: + +/* Line 1455 of yacc.c */ +#line 1259 "program_parse.y" + { + if (((yyvsp[(1) - (1)].integer) < 1) || ((unsigned) (yyvsp[(1) - (1)].integer) > state->limits->MaxParameters)) { + yyerror(& (yylsp[(1) - (1)]), state, "invalid parameter array size"); + YYERROR; + } else { + (yyval.integer) = (yyvsp[(1) - (1)].integer); + } + ;} + break; + + case 127: + +/* Line 1455 of yacc.c */ +#line 1270 "program_parse.y" + { + (yyval.temp_sym) = (yyvsp[(2) - (2)].temp_sym); + ;} + break; + + case 128: + +/* Line 1455 of yacc.c */ +#line 1276 "program_parse.y" + { + (yyval.temp_sym) = (yyvsp[(3) - (4)].temp_sym); + ;} + break; + + case 130: + +/* Line 1455 of yacc.c */ +#line 1283 "program_parse.y" + { + (yyvsp[(1) - (3)].temp_sym).param_binding_length += (yyvsp[(3) - (3)].temp_sym).param_binding_length; + (yyval.temp_sym) = (yyvsp[(1) - (3)].temp_sym); + ;} + break; + + case 131: + +/* Line 1455 of yacc.c */ +#line 1290 "program_parse.y" + { + memset(& (yyval.temp_sym), 0, sizeof((yyval.temp_sym))); + (yyval.temp_sym).param_binding_begin = ~0; + initialize_symbol_from_state(state->prog, & (yyval.temp_sym), (yyvsp[(1) - (1)].state)); + ;} + break; + + case 132: + +/* Line 1455 of yacc.c */ +#line 1296 "program_parse.y" + { + memset(& (yyval.temp_sym), 0, sizeof((yyval.temp_sym))); + (yyval.temp_sym).param_binding_begin = ~0; + initialize_symbol_from_param(state->prog, & (yyval.temp_sym), (yyvsp[(1) - (1)].state)); + ;} + break; + + case 133: + +/* Line 1455 of yacc.c */ +#line 1302 "program_parse.y" + { + memset(& (yyval.temp_sym), 0, sizeof((yyval.temp_sym))); + (yyval.temp_sym).param_binding_begin = ~0; + initialize_symbol_from_const(state->prog, & (yyval.temp_sym), & (yyvsp[(1) - (1)].vector), GL_TRUE); + ;} + break; + + case 134: + +/* Line 1455 of yacc.c */ +#line 1310 "program_parse.y" + { + memset(& (yyval.temp_sym), 0, sizeof((yyval.temp_sym))); + (yyval.temp_sym).param_binding_begin = ~0; + initialize_symbol_from_state(state->prog, & (yyval.temp_sym), (yyvsp[(1) - (1)].state)); + ;} + break; + + case 135: + +/* Line 1455 of yacc.c */ +#line 1316 "program_parse.y" + { + memset(& (yyval.temp_sym), 0, sizeof((yyval.temp_sym))); + (yyval.temp_sym).param_binding_begin = ~0; + initialize_symbol_from_param(state->prog, & (yyval.temp_sym), (yyvsp[(1) - (1)].state)); + ;} + break; + + case 136: + +/* Line 1455 of yacc.c */ +#line 1322 "program_parse.y" + { + memset(& (yyval.temp_sym), 0, sizeof((yyval.temp_sym))); + (yyval.temp_sym).param_binding_begin = ~0; + initialize_symbol_from_const(state->prog, & (yyval.temp_sym), & (yyvsp[(1) - (1)].vector), GL_TRUE); + ;} + break; + + case 137: + +/* Line 1455 of yacc.c */ +#line 1330 "program_parse.y" + { + memset(& (yyval.temp_sym), 0, sizeof((yyval.temp_sym))); + (yyval.temp_sym).param_binding_begin = ~0; + initialize_symbol_from_state(state->prog, & (yyval.temp_sym), (yyvsp[(1) - (1)].state)); + ;} + break; + + case 138: + +/* Line 1455 of yacc.c */ +#line 1336 "program_parse.y" + { + memset(& (yyval.temp_sym), 0, sizeof((yyval.temp_sym))); + (yyval.temp_sym).param_binding_begin = ~0; + initialize_symbol_from_param(state->prog, & (yyval.temp_sym), (yyvsp[(1) - (1)].state)); + ;} + break; + + case 139: + +/* Line 1455 of yacc.c */ +#line 1342 "program_parse.y" + { + memset(& (yyval.temp_sym), 0, sizeof((yyval.temp_sym))); + (yyval.temp_sym).param_binding_begin = ~0; + initialize_symbol_from_const(state->prog, & (yyval.temp_sym), & (yyvsp[(1) - (1)].vector), GL_FALSE); + ;} + break; + + case 140: + +/* Line 1455 of yacc.c */ +#line 1349 "program_parse.y" + { memcpy((yyval.state), (yyvsp[(1) - (1)].state), sizeof((yyval.state))); ;} + break; + + case 141: + +/* Line 1455 of yacc.c */ +#line 1350 "program_parse.y" + { memcpy((yyval.state), (yyvsp[(2) - (2)].state), sizeof((yyval.state))); ;} + break; + + case 142: + +/* Line 1455 of yacc.c */ +#line 1353 "program_parse.y" + { memcpy((yyval.state), (yyvsp[(2) - (2)].state), sizeof((yyval.state))); ;} + break; + + case 143: + +/* Line 1455 of yacc.c */ +#line 1354 "program_parse.y" + { memcpy((yyval.state), (yyvsp[(2) - (2)].state), sizeof((yyval.state))); ;} + break; + + case 144: + +/* Line 1455 of yacc.c */ +#line 1355 "program_parse.y" + { memcpy((yyval.state), (yyvsp[(2) - (2)].state), sizeof((yyval.state))); ;} + break; + + case 145: + +/* Line 1455 of yacc.c */ +#line 1356 "program_parse.y" + { memcpy((yyval.state), (yyvsp[(2) - (2)].state), sizeof((yyval.state))); ;} + break; + + case 146: + +/* Line 1455 of yacc.c */ +#line 1357 "program_parse.y" + { memcpy((yyval.state), (yyvsp[(2) - (2)].state), sizeof((yyval.state))); ;} + break; + + case 147: + +/* Line 1455 of yacc.c */ +#line 1358 "program_parse.y" + { memcpy((yyval.state), (yyvsp[(2) - (2)].state), sizeof((yyval.state))); ;} + break; + + case 148: + +/* Line 1455 of yacc.c */ +#line 1359 "program_parse.y" + { memcpy((yyval.state), (yyvsp[(2) - (2)].state), sizeof((yyval.state))); ;} + break; + + case 149: + +/* Line 1455 of yacc.c */ +#line 1360 "program_parse.y" + { memcpy((yyval.state), (yyvsp[(2) - (2)].state), sizeof((yyval.state))); ;} + break; + + case 150: + +/* Line 1455 of yacc.c */ +#line 1361 "program_parse.y" + { memcpy((yyval.state), (yyvsp[(2) - (2)].state), sizeof((yyval.state))); ;} + break; + + case 151: + +/* Line 1455 of yacc.c */ +#line 1362 "program_parse.y" + { memcpy((yyval.state), (yyvsp[(2) - (2)].state), sizeof((yyval.state))); ;} + break; + + case 152: + +/* Line 1455 of yacc.c */ +#line 1363 "program_parse.y" + { memcpy((yyval.state), (yyvsp[(2) - (2)].state), sizeof((yyval.state))); ;} + break; + + case 153: + +/* Line 1455 of yacc.c */ +#line 1367 "program_parse.y" + { + memset((yyval.state), 0, sizeof((yyval.state))); + (yyval.state)[0] = STATE_MATERIAL; + (yyval.state)[1] = (yyvsp[(2) - (3)].integer); + (yyval.state)[2] = (yyvsp[(3) - (3)].integer); + ;} + break; + + case 154: + +/* Line 1455 of yacc.c */ +#line 1376 "program_parse.y" + { + (yyval.integer) = (yyvsp[(1) - (1)].integer); + ;} + break; + + case 155: + +/* Line 1455 of yacc.c */ +#line 1380 "program_parse.y" + { + (yyval.integer) = STATE_EMISSION; + ;} + break; + + case 156: + +/* Line 1455 of yacc.c */ +#line 1384 "program_parse.y" + { + (yyval.integer) = STATE_SHININESS; + ;} + break; + + case 157: + +/* Line 1455 of yacc.c */ +#line 1390 "program_parse.y" + { + memset((yyval.state), 0, sizeof((yyval.state))); + (yyval.state)[0] = STATE_LIGHT; + (yyval.state)[1] = (yyvsp[(3) - (5)].integer); + (yyval.state)[2] = (yyvsp[(5) - (5)].integer); + ;} + break; + + case 158: + +/* Line 1455 of yacc.c */ +#line 1399 "program_parse.y" + { + (yyval.integer) = (yyvsp[(1) - (1)].integer); + ;} + break; + + case 159: + +/* Line 1455 of yacc.c */ +#line 1403 "program_parse.y" + { + (yyval.integer) = STATE_POSITION; + ;} + break; + + case 160: + +/* Line 1455 of yacc.c */ +#line 1407 "program_parse.y" + { + if (!state->ctx->Extensions.EXT_point_parameters) { + yyerror(& (yylsp[(1) - (1)]), state, "GL_ARB_point_parameters not supported"); + YYERROR; + } + + (yyval.integer) = STATE_ATTENUATION; + ;} + break; + + case 161: + +/* Line 1455 of yacc.c */ +#line 1416 "program_parse.y" + { + (yyval.integer) = (yyvsp[(2) - (2)].integer); + ;} + break; + + case 162: + +/* Line 1455 of yacc.c */ +#line 1420 "program_parse.y" + { + (yyval.integer) = STATE_HALF_VECTOR; + ;} + break; + + case 163: + +/* Line 1455 of yacc.c */ +#line 1426 "program_parse.y" + { + (yyval.integer) = STATE_SPOT_DIRECTION; + ;} + break; + + case 164: + +/* Line 1455 of yacc.c */ +#line 1432 "program_parse.y" + { + (yyval.state)[0] = (yyvsp[(2) - (2)].state)[0]; + (yyval.state)[1] = (yyvsp[(2) - (2)].state)[1]; + ;} + break; + + case 165: + +/* Line 1455 of yacc.c */ +#line 1439 "program_parse.y" + { + memset((yyval.state), 0, sizeof((yyval.state))); + (yyval.state)[0] = STATE_LIGHTMODEL_AMBIENT; + ;} + break; + + case 166: + +/* Line 1455 of yacc.c */ +#line 1444 "program_parse.y" + { + memset((yyval.state), 0, sizeof((yyval.state))); + (yyval.state)[0] = STATE_LIGHTMODEL_SCENECOLOR; + (yyval.state)[1] = (yyvsp[(1) - (2)].integer); + ;} + break; + + case 167: + +/* Line 1455 of yacc.c */ +#line 1452 "program_parse.y" + { + memset((yyval.state), 0, sizeof((yyval.state))); + (yyval.state)[0] = STATE_LIGHTPROD; + (yyval.state)[1] = (yyvsp[(3) - (6)].integer); + (yyval.state)[2] = (yyvsp[(5) - (6)].integer); + (yyval.state)[3] = (yyvsp[(6) - (6)].integer); + ;} + break; + + case 169: + +/* Line 1455 of yacc.c */ +#line 1464 "program_parse.y" + { + memset((yyval.state), 0, sizeof((yyval.state))); + (yyval.state)[0] = (yyvsp[(3) - (3)].integer); + (yyval.state)[1] = (yyvsp[(2) - (3)].integer); + ;} + break; + + case 170: + +/* Line 1455 of yacc.c */ +#line 1472 "program_parse.y" + { + (yyval.integer) = STATE_TEXENV_COLOR; + ;} + break; + + case 171: + +/* Line 1455 of yacc.c */ +#line 1478 "program_parse.y" + { + (yyval.integer) = STATE_AMBIENT; + ;} + break; + + case 172: + +/* Line 1455 of yacc.c */ +#line 1482 "program_parse.y" + { + (yyval.integer) = STATE_DIFFUSE; + ;} + break; + + case 173: + +/* Line 1455 of yacc.c */ +#line 1486 "program_parse.y" + { + (yyval.integer) = STATE_SPECULAR; + ;} + break; + + case 174: + +/* Line 1455 of yacc.c */ +#line 1492 "program_parse.y" + { + if ((unsigned) (yyvsp[(1) - (1)].integer) >= state->MaxLights) { + yyerror(& (yylsp[(1) - (1)]), state, "invalid light selector"); + YYERROR; + } + + (yyval.integer) = (yyvsp[(1) - (1)].integer); + ;} + break; + + case 175: + +/* Line 1455 of yacc.c */ +#line 1503 "program_parse.y" + { + memset((yyval.state), 0, sizeof((yyval.state))); + (yyval.state)[0] = STATE_TEXGEN; + (yyval.state)[1] = (yyvsp[(2) - (4)].integer); + (yyval.state)[2] = (yyvsp[(3) - (4)].integer) + (yyvsp[(4) - (4)].integer); + ;} + break; + + case 176: + +/* Line 1455 of yacc.c */ +#line 1512 "program_parse.y" + { + (yyval.integer) = STATE_TEXGEN_EYE_S; + ;} + break; + + case 177: + +/* Line 1455 of yacc.c */ +#line 1516 "program_parse.y" + { + (yyval.integer) = STATE_TEXGEN_OBJECT_S; + ;} + break; + + case 178: + +/* Line 1455 of yacc.c */ +#line 1521 "program_parse.y" + { + (yyval.integer) = STATE_TEXGEN_EYE_S - STATE_TEXGEN_EYE_S; + ;} + break; + + case 179: + +/* Line 1455 of yacc.c */ +#line 1525 "program_parse.y" + { + (yyval.integer) = STATE_TEXGEN_EYE_T - STATE_TEXGEN_EYE_S; + ;} + break; + + case 180: + +/* Line 1455 of yacc.c */ +#line 1529 "program_parse.y" + { + (yyval.integer) = STATE_TEXGEN_EYE_R - STATE_TEXGEN_EYE_S; + ;} + break; + + case 181: + +/* Line 1455 of yacc.c */ +#line 1533 "program_parse.y" + { + (yyval.integer) = STATE_TEXGEN_EYE_Q - STATE_TEXGEN_EYE_S; + ;} + break; + + case 182: + +/* Line 1455 of yacc.c */ +#line 1539 "program_parse.y" + { + memset((yyval.state), 0, sizeof((yyval.state))); + (yyval.state)[0] = (yyvsp[(2) - (2)].integer); + ;} + break; + + case 183: + +/* Line 1455 of yacc.c */ +#line 1546 "program_parse.y" + { + (yyval.integer) = STATE_FOG_COLOR; + ;} + break; + + case 184: + +/* Line 1455 of yacc.c */ +#line 1550 "program_parse.y" + { + (yyval.integer) = STATE_FOG_PARAMS; + ;} + break; + + case 185: + +/* Line 1455 of yacc.c */ +#line 1556 "program_parse.y" + { + memset((yyval.state), 0, sizeof((yyval.state))); + (yyval.state)[0] = STATE_CLIPPLANE; + (yyval.state)[1] = (yyvsp[(3) - (5)].integer); + ;} + break; + + case 186: + +/* Line 1455 of yacc.c */ +#line 1564 "program_parse.y" + { + if ((unsigned) (yyvsp[(1) - (1)].integer) >= state->MaxClipPlanes) { + yyerror(& (yylsp[(1) - (1)]), state, "invalid clip plane selector"); + YYERROR; + } + + (yyval.integer) = (yyvsp[(1) - (1)].integer); + ;} + break; + + case 187: + +/* Line 1455 of yacc.c */ +#line 1575 "program_parse.y" + { + memset((yyval.state), 0, sizeof((yyval.state))); + (yyval.state)[0] = (yyvsp[(2) - (2)].integer); + ;} + break; + + case 188: + +/* Line 1455 of yacc.c */ +#line 1582 "program_parse.y" + { + (yyval.integer) = STATE_POINT_SIZE; + ;} + break; + + case 189: + +/* Line 1455 of yacc.c */ +#line 1586 "program_parse.y" + { + (yyval.integer) = STATE_POINT_ATTENUATION; + ;} + break; + + case 190: + +/* Line 1455 of yacc.c */ +#line 1592 "program_parse.y" + { + (yyval.state)[0] = (yyvsp[(1) - (5)].state)[0]; + (yyval.state)[1] = (yyvsp[(1) - (5)].state)[1]; + (yyval.state)[2] = (yyvsp[(4) - (5)].integer); + (yyval.state)[3] = (yyvsp[(4) - (5)].integer); + (yyval.state)[4] = (yyvsp[(1) - (5)].state)[2]; + ;} + break; + + case 191: + +/* Line 1455 of yacc.c */ +#line 1602 "program_parse.y" + { + (yyval.state)[0] = (yyvsp[(1) - (2)].state)[0]; + (yyval.state)[1] = (yyvsp[(1) - (2)].state)[1]; + (yyval.state)[2] = (yyvsp[(2) - (2)].state)[2]; + (yyval.state)[3] = (yyvsp[(2) - (2)].state)[3]; + (yyval.state)[4] = (yyvsp[(1) - (2)].state)[2]; + ;} + break; + + case 192: + +/* Line 1455 of yacc.c */ +#line 1612 "program_parse.y" + { + (yyval.state)[2] = 0; + (yyval.state)[3] = 3; + ;} + break; + + case 193: + +/* Line 1455 of yacc.c */ +#line 1617 "program_parse.y" + { + /* It seems logical that the matrix row range specifier would have + * to specify a range or more than one row (i.e., $5 > $3). + * However, the ARB_vertex_program spec says "a program will fail + * to load if <a> is greater than <b>." This means that $3 == $5 + * is valid. + */ + if ((yyvsp[(3) - (6)].integer) > (yyvsp[(5) - (6)].integer)) { + yyerror(& (yylsp[(3) - (6)]), state, "invalid matrix row range"); + YYERROR; + } + + (yyval.state)[2] = (yyvsp[(3) - (6)].integer); + (yyval.state)[3] = (yyvsp[(5) - (6)].integer); + ;} + break; + + case 194: + +/* Line 1455 of yacc.c */ +#line 1635 "program_parse.y" + { + (yyval.state)[0] = (yyvsp[(2) - (3)].state)[0]; + (yyval.state)[1] = (yyvsp[(2) - (3)].state)[1]; + (yyval.state)[2] = (yyvsp[(3) - (3)].integer); + ;} + break; + + case 195: + +/* Line 1455 of yacc.c */ +#line 1643 "program_parse.y" + { + (yyval.integer) = 0; + ;} + break; + + case 196: + +/* Line 1455 of yacc.c */ +#line 1647 "program_parse.y" + { + (yyval.integer) = (yyvsp[(1) - (1)].integer); + ;} + break; + + case 197: + +/* Line 1455 of yacc.c */ +#line 1653 "program_parse.y" + { + (yyval.integer) = STATE_MATRIX_INVERSE; + ;} + break; + + case 198: + +/* Line 1455 of yacc.c */ +#line 1657 "program_parse.y" + { + (yyval.integer) = STATE_MATRIX_TRANSPOSE; + ;} + break; + + case 199: + +/* Line 1455 of yacc.c */ +#line 1661 "program_parse.y" + { + (yyval.integer) = STATE_MATRIX_INVTRANS; + ;} + break; + + case 200: + +/* Line 1455 of yacc.c */ +#line 1667 "program_parse.y" + { + if ((yyvsp[(1) - (1)].integer) > 3) { + yyerror(& (yylsp[(1) - (1)]), state, "invalid matrix row reference"); + YYERROR; + } + + (yyval.integer) = (yyvsp[(1) - (1)].integer); + ;} + break; + + case 201: + +/* Line 1455 of yacc.c */ +#line 1678 "program_parse.y" + { + (yyval.state)[0] = STATE_MODELVIEW_MATRIX; + (yyval.state)[1] = (yyvsp[(2) - (2)].integer); + ;} + break; + + case 202: + +/* Line 1455 of yacc.c */ +#line 1683 "program_parse.y" + { + (yyval.state)[0] = STATE_PROJECTION_MATRIX; + (yyval.state)[1] = 0; + ;} + break; + + case 203: + +/* Line 1455 of yacc.c */ +#line 1688 "program_parse.y" + { + (yyval.state)[0] = STATE_MVP_MATRIX; + (yyval.state)[1] = 0; + ;} + break; + + case 204: + +/* Line 1455 of yacc.c */ +#line 1693 "program_parse.y" + { + (yyval.state)[0] = STATE_TEXTURE_MATRIX; + (yyval.state)[1] = (yyvsp[(2) - (2)].integer); + ;} + break; + + case 205: + +/* Line 1455 of yacc.c */ +#line 1698 "program_parse.y" + { + yyerror(& (yylsp[(1) - (4)]), state, "GL_ARB_matrix_palette not supported"); + YYERROR; + ;} + break; + + case 206: + +/* Line 1455 of yacc.c */ +#line 1703 "program_parse.y" + { + (yyval.state)[0] = STATE_PROGRAM_MATRIX; + (yyval.state)[1] = (yyvsp[(3) - (4)].integer); + ;} + break; + + case 207: + +/* Line 1455 of yacc.c */ +#line 1710 "program_parse.y" + { + (yyval.integer) = 0; + ;} + break; + + case 208: + +/* Line 1455 of yacc.c */ +#line 1714 "program_parse.y" + { + (yyval.integer) = (yyvsp[(2) - (3)].integer); + ;} + break; + + case 209: + +/* Line 1455 of yacc.c */ +#line 1719 "program_parse.y" + { + /* Since GL_ARB_vertex_blend isn't supported, only modelview matrix + * zero is valid. + */ + if ((yyvsp[(1) - (1)].integer) != 0) { + yyerror(& (yylsp[(1) - (1)]), state, "invalid modelview matrix index"); + YYERROR; + } + + (yyval.integer) = (yyvsp[(1) - (1)].integer); + ;} + break; + + case 210: + +/* Line 1455 of yacc.c */ +#line 1732 "program_parse.y" + { + /* Since GL_ARB_matrix_palette isn't supported, just let any value + * through here. The error will be generated later. + */ + (yyval.integer) = (yyvsp[(1) - (1)].integer); + ;} + break; + + case 211: + +/* Line 1455 of yacc.c */ +#line 1740 "program_parse.y" + { + if ((unsigned) (yyvsp[(1) - (1)].integer) >= state->MaxProgramMatrices) { + yyerror(& (yylsp[(1) - (1)]), state, "invalid program matrix selector"); + YYERROR; + } + + (yyval.integer) = (yyvsp[(1) - (1)].integer); + ;} + break; + + case 212: + +/* Line 1455 of yacc.c */ +#line 1751 "program_parse.y" + { + memset((yyval.state), 0, sizeof((yyval.state))); + (yyval.state)[0] = STATE_DEPTH_RANGE; + ;} + break; + + case 217: + +/* Line 1455 of yacc.c */ +#line 1763 "program_parse.y" + { + memset((yyval.state), 0, sizeof((yyval.state))); + (yyval.state)[0] = state->state_param_enum; + (yyval.state)[1] = STATE_ENV; + (yyval.state)[2] = (yyvsp[(4) - (5)].state)[0]; + (yyval.state)[3] = (yyvsp[(4) - (5)].state)[1]; + ;} + break; + + case 218: + +/* Line 1455 of yacc.c */ +#line 1773 "program_parse.y" + { + (yyval.state)[0] = (yyvsp[(1) - (1)].integer); + (yyval.state)[1] = (yyvsp[(1) - (1)].integer); + ;} + break; + + case 219: + +/* Line 1455 of yacc.c */ +#line 1778 "program_parse.y" + { + (yyval.state)[0] = (yyvsp[(1) - (3)].integer); + (yyval.state)[1] = (yyvsp[(3) - (3)].integer); + ;} + break; + + case 220: + +/* Line 1455 of yacc.c */ +#line 1785 "program_parse.y" + { + memset((yyval.state), 0, sizeof((yyval.state))); + (yyval.state)[0] = state->state_param_enum; + (yyval.state)[1] = STATE_ENV; + (yyval.state)[2] = (yyvsp[(4) - (5)].integer); + (yyval.state)[3] = (yyvsp[(4) - (5)].integer); + ;} + break; + + case 221: + +/* Line 1455 of yacc.c */ +#line 1795 "program_parse.y" + { + memset((yyval.state), 0, sizeof((yyval.state))); + (yyval.state)[0] = state->state_param_enum; + (yyval.state)[1] = STATE_LOCAL; + (yyval.state)[2] = (yyvsp[(4) - (5)].state)[0]; + (yyval.state)[3] = (yyvsp[(4) - (5)].state)[1]; + ;} + break; + + case 222: + +/* Line 1455 of yacc.c */ +#line 1804 "program_parse.y" + { + (yyval.state)[0] = (yyvsp[(1) - (1)].integer); + (yyval.state)[1] = (yyvsp[(1) - (1)].integer); + ;} + break; + + case 223: + +/* Line 1455 of yacc.c */ +#line 1809 "program_parse.y" + { + (yyval.state)[0] = (yyvsp[(1) - (3)].integer); + (yyval.state)[1] = (yyvsp[(3) - (3)].integer); + ;} + break; + + case 224: + +/* Line 1455 of yacc.c */ +#line 1816 "program_parse.y" + { + memset((yyval.state), 0, sizeof((yyval.state))); + (yyval.state)[0] = state->state_param_enum; + (yyval.state)[1] = STATE_LOCAL; + (yyval.state)[2] = (yyvsp[(4) - (5)].integer); + (yyval.state)[3] = (yyvsp[(4) - (5)].integer); + ;} + break; + + case 225: + +/* Line 1455 of yacc.c */ +#line 1826 "program_parse.y" + { + if ((unsigned) (yyvsp[(1) - (1)].integer) >= state->limits->MaxEnvParams) { + yyerror(& (yylsp[(1) - (1)]), state, "invalid environment parameter reference"); + YYERROR; + } + (yyval.integer) = (yyvsp[(1) - (1)].integer); + ;} + break; + + case 226: + +/* Line 1455 of yacc.c */ +#line 1836 "program_parse.y" + { + if ((unsigned) (yyvsp[(1) - (1)].integer) >= state->limits->MaxLocalParams) { + yyerror(& (yylsp[(1) - (1)]), state, "invalid local parameter reference"); + YYERROR; + } + (yyval.integer) = (yyvsp[(1) - (1)].integer); + ;} + break; + + case 231: + +/* Line 1455 of yacc.c */ +#line 1851 "program_parse.y" + { + (yyval.vector).count = 4; + (yyval.vector).data[0] = (yyvsp[(1) - (1)].real); + (yyval.vector).data[1] = (yyvsp[(1) - (1)].real); + (yyval.vector).data[2] = (yyvsp[(1) - (1)].real); + (yyval.vector).data[3] = (yyvsp[(1) - (1)].real); + ;} + break; + + case 232: + +/* Line 1455 of yacc.c */ +#line 1861 "program_parse.y" + { + (yyval.vector).count = 1; + (yyval.vector).data[0] = (yyvsp[(1) - (1)].real); + (yyval.vector).data[1] = (yyvsp[(1) - (1)].real); + (yyval.vector).data[2] = (yyvsp[(1) - (1)].real); + (yyval.vector).data[3] = (yyvsp[(1) - (1)].real); + ;} + break; + + case 233: + +/* Line 1455 of yacc.c */ +#line 1869 "program_parse.y" + { + (yyval.vector).count = 1; + (yyval.vector).data[0] = (float) (yyvsp[(1) - (1)].integer); + (yyval.vector).data[1] = (float) (yyvsp[(1) - (1)].integer); + (yyval.vector).data[2] = (float) (yyvsp[(1) - (1)].integer); + (yyval.vector).data[3] = (float) (yyvsp[(1) - (1)].integer); + ;} + break; + + case 234: + +/* Line 1455 of yacc.c */ +#line 1879 "program_parse.y" + { + (yyval.vector).count = 4; + (yyval.vector).data[0] = (yyvsp[(2) - (3)].real); + (yyval.vector).data[1] = 0.0f; + (yyval.vector).data[2] = 0.0f; + (yyval.vector).data[3] = 1.0f; + ;} + break; + + case 235: + +/* Line 1455 of yacc.c */ +#line 1887 "program_parse.y" + { + (yyval.vector).count = 4; + (yyval.vector).data[0] = (yyvsp[(2) - (5)].real); + (yyval.vector).data[1] = (yyvsp[(4) - (5)].real); + (yyval.vector).data[2] = 0.0f; + (yyval.vector).data[3] = 1.0f; + ;} + break; + + case 236: + +/* Line 1455 of yacc.c */ +#line 1896 "program_parse.y" + { + (yyval.vector).count = 4; + (yyval.vector).data[0] = (yyvsp[(2) - (7)].real); + (yyval.vector).data[1] = (yyvsp[(4) - (7)].real); + (yyval.vector).data[2] = (yyvsp[(6) - (7)].real); + (yyval.vector).data[3] = 1.0f; + ;} + break; + + case 237: + +/* Line 1455 of yacc.c */ +#line 1905 "program_parse.y" + { + (yyval.vector).count = 4; + (yyval.vector).data[0] = (yyvsp[(2) - (9)].real); + (yyval.vector).data[1] = (yyvsp[(4) - (9)].real); + (yyval.vector).data[2] = (yyvsp[(6) - (9)].real); + (yyval.vector).data[3] = (yyvsp[(8) - (9)].real); + ;} + break; + + case 238: + +/* Line 1455 of yacc.c */ +#line 1915 "program_parse.y" + { + (yyval.real) = ((yyvsp[(1) - (2)].negate)) ? -(yyvsp[(2) - (2)].real) : (yyvsp[(2) - (2)].real); + ;} + break; + + case 239: + +/* Line 1455 of yacc.c */ +#line 1919 "program_parse.y" + { + (yyval.real) = (float)(((yyvsp[(1) - (2)].negate)) ? -(yyvsp[(2) - (2)].integer) : (yyvsp[(2) - (2)].integer)); + ;} + break; + + case 240: + +/* Line 1455 of yacc.c */ +#line 1924 "program_parse.y" + { (yyval.negate) = FALSE; ;} + break; + + case 241: + +/* Line 1455 of yacc.c */ +#line 1925 "program_parse.y" + { (yyval.negate) = TRUE; ;} + break; + + case 242: + +/* Line 1455 of yacc.c */ +#line 1926 "program_parse.y" + { (yyval.negate) = FALSE; ;} + break; + + case 243: + +/* Line 1455 of yacc.c */ +#line 1929 "program_parse.y" + { (yyval.integer) = (yyvsp[(2) - (2)].integer); ;} + break; + + case 245: + +/* Line 1455 of yacc.c */ +#line 1933 "program_parse.y" + { + /* NV_fragment_program_option defines the size qualifiers in a + * fairly broken way. "SHORT" or "LONG" can optionally be used + * before TEMP or OUTPUT. However, neither is a reserved word! + * This means that we have to parse it as an identifier, then check + * to make sure it's one of the valid values. *sigh* + * + * In addition, the grammar in the extension spec does *not* allow + * the size specifier to be optional, but all known implementations + * do. + */ + if (!state->option.NV_fragment) { + yyerror(& (yylsp[(1) - (1)]), state, "unexpected IDENTIFIER"); + YYERROR; + } + + if (strcmp("SHORT", (yyvsp[(1) - (1)].string)) == 0) { + } else if (strcmp("LONG", (yyvsp[(1) - (1)].string)) == 0) { + } else { + char *const err_str = + make_error_string("invalid storage size specifier \"%s\"", + (yyvsp[(1) - (1)].string)); + + yyerror(& (yylsp[(1) - (1)]), state, (err_str != NULL) + ? err_str : "invalid storage size specifier"); + + if (err_str != NULL) { + free(err_str); + } + + YYERROR; + } + ;} + break; + + case 246: + +/* Line 1455 of yacc.c */ +#line 1967 "program_parse.y" + { + ;} + break; + + case 247: + +/* Line 1455 of yacc.c */ +#line 1971 "program_parse.y" + { (yyval.integer) = (yyvsp[(1) - (1)].integer); ;} + break; + + case 249: + +/* Line 1455 of yacc.c */ +#line 1975 "program_parse.y" + { + if (!declare_variable(state, (yyvsp[(3) - (3)].string), (yyvsp[(0) - (3)].integer), & (yylsp[(3) - (3)]))) { + free((yyvsp[(3) - (3)].string)); + YYERROR; + } + ;} + break; + + case 250: + +/* Line 1455 of yacc.c */ +#line 1982 "program_parse.y" + { + if (!declare_variable(state, (yyvsp[(1) - (1)].string), (yyvsp[(0) - (1)].integer), & (yylsp[(1) - (1)]))) { + free((yyvsp[(1) - (1)].string)); + YYERROR; + } + ;} + break; + + case 251: + +/* Line 1455 of yacc.c */ +#line 1991 "program_parse.y" + { + struct asm_symbol *const s = + declare_variable(state, (yyvsp[(3) - (5)].string), at_output, & (yylsp[(3) - (5)])); + + if (s == NULL) { + free((yyvsp[(3) - (5)].string)); + YYERROR; + } else { + s->output_binding = (yyvsp[(5) - (5)].result); + } + ;} + break; + + case 252: + +/* Line 1455 of yacc.c */ +#line 2005 "program_parse.y" + { + if (state->mode == ARB_vertex) { + (yyval.result) = VERT_RESULT_HPOS; + } else { + yyerror(& (yylsp[(2) - (2)]), state, "invalid program result name"); + YYERROR; + } + ;} + break; + + case 253: + +/* Line 1455 of yacc.c */ +#line 2014 "program_parse.y" + { + if (state->mode == ARB_vertex) { + (yyval.result) = VERT_RESULT_FOGC; + } else { + yyerror(& (yylsp[(2) - (2)]), state, "invalid program result name"); + YYERROR; + } + ;} + break; + + case 254: + +/* Line 1455 of yacc.c */ +#line 2023 "program_parse.y" + { + (yyval.result) = (yyvsp[(2) - (2)].result); + ;} + break; + + case 255: + +/* Line 1455 of yacc.c */ +#line 2027 "program_parse.y" + { + if (state->mode == ARB_vertex) { + (yyval.result) = VERT_RESULT_PSIZ; + } else { + yyerror(& (yylsp[(2) - (2)]), state, "invalid program result name"); + YYERROR; + } + ;} + break; + + case 256: + +/* Line 1455 of yacc.c */ +#line 2036 "program_parse.y" + { + if (state->mode == ARB_vertex) { + (yyval.result) = VERT_RESULT_TEX0 + (yyvsp[(3) - (3)].integer); + } else { + yyerror(& (yylsp[(2) - (3)]), state, "invalid program result name"); + YYERROR; + } + ;} + break; + + case 257: + +/* Line 1455 of yacc.c */ +#line 2045 "program_parse.y" + { + if (state->mode == ARB_fragment) { + (yyval.result) = FRAG_RESULT_DEPTH; + } else { + yyerror(& (yylsp[(2) - (2)]), state, "invalid program result name"); + YYERROR; + } + ;} + break; + + case 258: + +/* Line 1455 of yacc.c */ +#line 2056 "program_parse.y" + { + (yyval.result) = (yyvsp[(2) - (3)].integer) + (yyvsp[(3) - (3)].integer); + ;} + break; + + case 259: + +/* Line 1455 of yacc.c */ +#line 2062 "program_parse.y" + { + (yyval.integer) = (state->mode == ARB_vertex) + ? VERT_RESULT_COL0 + : FRAG_RESULT_COLOR; + ;} + break; + + case 260: + +/* Line 1455 of yacc.c */ +#line 2068 "program_parse.y" + { + if (state->mode == ARB_vertex) { + (yyval.integer) = VERT_RESULT_COL0; + } else { + yyerror(& (yylsp[(1) - (1)]), state, "invalid program result name"); + YYERROR; + } + ;} + break; + + case 261: + +/* Line 1455 of yacc.c */ +#line 2077 "program_parse.y" + { + if (state->mode == ARB_vertex) { + (yyval.integer) = VERT_RESULT_BFC0; + } else { + yyerror(& (yylsp[(1) - (1)]), state, "invalid program result name"); + YYERROR; + } + ;} + break; + + case 262: + +/* Line 1455 of yacc.c */ +#line 2088 "program_parse.y" + { + (yyval.integer) = 0; + ;} + break; + + case 263: + +/* Line 1455 of yacc.c */ +#line 2092 "program_parse.y" + { + if (state->mode == ARB_vertex) { + (yyval.integer) = 0; + } else { + yyerror(& (yylsp[(1) - (1)]), state, "invalid program result name"); + YYERROR; + } + ;} + break; + + case 264: + +/* Line 1455 of yacc.c */ +#line 2101 "program_parse.y" + { + if (state->mode == ARB_vertex) { + (yyval.integer) = 1; + } else { + yyerror(& (yylsp[(1) - (1)]), state, "invalid program result name"); + YYERROR; + } + ;} + break; + + case 265: + +/* Line 1455 of yacc.c */ +#line 2111 "program_parse.y" + { (yyval.integer) = 0; ;} + break; + + case 266: + +/* Line 1455 of yacc.c */ +#line 2112 "program_parse.y" + { (yyval.integer) = 0; ;} + break; + + case 267: + +/* Line 1455 of yacc.c */ +#line 2113 "program_parse.y" + { (yyval.integer) = 1; ;} + break; + + case 268: + +/* Line 1455 of yacc.c */ +#line 2116 "program_parse.y" + { (yyval.integer) = 0; ;} + break; + + case 269: + +/* Line 1455 of yacc.c */ +#line 2117 "program_parse.y" + { (yyval.integer) = 0; ;} + break; + + case 270: + +/* Line 1455 of yacc.c */ +#line 2118 "program_parse.y" + { (yyval.integer) = 1; ;} + break; + + case 271: + +/* Line 1455 of yacc.c */ +#line 2121 "program_parse.y" + { (yyval.integer) = 0; ;} + break; + + case 272: + +/* Line 1455 of yacc.c */ +#line 2122 "program_parse.y" + { (yyval.integer) = (yyvsp[(2) - (3)].integer); ;} + break; + + case 273: + +/* Line 1455 of yacc.c */ +#line 2125 "program_parse.y" + { (yyval.integer) = 0; ;} + break; + + case 274: + +/* Line 1455 of yacc.c */ +#line 2126 "program_parse.y" + { (yyval.integer) = (yyvsp[(2) - (3)].integer); ;} + break; + + case 275: + +/* Line 1455 of yacc.c */ +#line 2129 "program_parse.y" + { (yyval.integer) = 0; ;} + break; + + case 276: + +/* Line 1455 of yacc.c */ +#line 2130 "program_parse.y" + { (yyval.integer) = (yyvsp[(2) - (3)].integer); ;} + break; + + case 277: + +/* Line 1455 of yacc.c */ +#line 2134 "program_parse.y" + { + if ((unsigned) (yyvsp[(1) - (1)].integer) >= state->MaxTextureCoordUnits) { + yyerror(& (yylsp[(1) - (1)]), state, "invalid texture coordinate unit selector"); + YYERROR; + } + + (yyval.integer) = (yyvsp[(1) - (1)].integer); + ;} + break; + + case 278: + +/* Line 1455 of yacc.c */ +#line 2145 "program_parse.y" + { + if ((unsigned) (yyvsp[(1) - (1)].integer) >= state->MaxTextureImageUnits) { + yyerror(& (yylsp[(1) - (1)]), state, "invalid texture image unit selector"); + YYERROR; + } + + (yyval.integer) = (yyvsp[(1) - (1)].integer); + ;} + break; + + case 279: + +/* Line 1455 of yacc.c */ +#line 2156 "program_parse.y" + { + if ((unsigned) (yyvsp[(1) - (1)].integer) >= state->MaxTextureUnits) { + yyerror(& (yylsp[(1) - (1)]), state, "invalid texture unit selector"); + YYERROR; + } + + (yyval.integer) = (yyvsp[(1) - (1)].integer); + ;} + break; + + case 280: + +/* Line 1455 of yacc.c */ +#line 2167 "program_parse.y" + { + struct asm_symbol *exist = (struct asm_symbol *) + _mesa_symbol_table_find_symbol(state->st, 0, (yyvsp[(2) - (4)].string)); + struct asm_symbol *target = (struct asm_symbol *) + _mesa_symbol_table_find_symbol(state->st, 0, (yyvsp[(4) - (4)].string)); + + free((yyvsp[(4) - (4)].string)); + + if (exist != NULL) { + char m[1000]; + _mesa_snprintf(m, sizeof(m), "redeclared identifier: %s", (yyvsp[(2) - (4)].string)); + free((yyvsp[(2) - (4)].string)); + yyerror(& (yylsp[(2) - (4)]), state, m); + YYERROR; + } else if (target == NULL) { + free((yyvsp[(2) - (4)].string)); + yyerror(& (yylsp[(4) - (4)]), state, + "undefined variable binding in ALIAS statement"); + YYERROR; + } else { + _mesa_symbol_table_add_symbol(state->st, 0, (yyvsp[(2) - (4)].string), target); + } + ;} + break; + + + +/* Line 1455 of yacc.c */ +#line 4938 "program_parse.tab.c" + default: break; + } + YY_SYMBOL_PRINT ("-> $$ =", yyr1[yyn], &yyval, &yyloc); + + YYPOPSTACK (yylen); + yylen = 0; + YY_STACK_PRINT (yyss, yyssp); + + *++yyvsp = yyval; + *++yylsp = yyloc; + + /* Now `shift' the result of the reduction. Determine what state + that goes to, based on the state we popped back to and the rule + number reduced by. */ + + yyn = yyr1[yyn]; + + yystate = yypgoto[yyn - YYNTOKENS] + *yyssp; + if (0 <= yystate && yystate <= YYLAST && yycheck[yystate] == *yyssp) + yystate = yytable[yystate]; + else + yystate = yydefgoto[yyn - YYNTOKENS]; + + goto yynewstate; + + +/*------------------------------------. +| yyerrlab -- here on detecting error | +`------------------------------------*/ +yyerrlab: + /* If not already recovering from an error, report this error. */ + if (!yyerrstatus) + { + ++yynerrs; +#if ! YYERROR_VERBOSE + yyerror (&yylloc, state, YY_("syntax error")); +#else + { + YYSIZE_T yysize = yysyntax_error (0, yystate, yychar); + if (yymsg_alloc < yysize && yymsg_alloc < YYSTACK_ALLOC_MAXIMUM) + { + YYSIZE_T yyalloc = 2 * yysize; + if (! (yysize <= yyalloc && yyalloc <= YYSTACK_ALLOC_MAXIMUM)) + yyalloc = YYSTACK_ALLOC_MAXIMUM; + if (yymsg != yymsgbuf) + YYSTACK_FREE (yymsg); + yymsg = (char *) YYSTACK_ALLOC (yyalloc); + if (yymsg) + yymsg_alloc = yyalloc; + else + { + yymsg = yymsgbuf; + yymsg_alloc = sizeof yymsgbuf; + } + } + + if (0 < yysize && yysize <= yymsg_alloc) + { + (void) yysyntax_error (yymsg, yystate, yychar); + yyerror (&yylloc, state, yymsg); + } + else + { + yyerror (&yylloc, state, YY_("syntax error")); + if (yysize != 0) + goto yyexhaustedlab; + } + } +#endif + } + + yyerror_range[0] = yylloc; + + if (yyerrstatus == 3) + { + /* If just tried and failed to reuse lookahead token after an + error, discard it. */ + + if (yychar <= YYEOF) + { + /* Return failure if at end of input. */ + if (yychar == YYEOF) + YYABORT; + } + else + { + yydestruct ("Error: discarding", + yytoken, &yylval, &yylloc, state); + yychar = YYEMPTY; + } + } + + /* Else will try to reuse lookahead token after shifting the error + token. */ + goto yyerrlab1; + + +/*---------------------------------------------------. +| yyerrorlab -- error raised explicitly by YYERROR. | +`---------------------------------------------------*/ +yyerrorlab: + + /* Pacify compilers like GCC when the user code never invokes + YYERROR and the label yyerrorlab therefore never appears in user + code. */ + if (/*CONSTCOND*/ 0) + goto yyerrorlab; + + yyerror_range[0] = yylsp[1-yylen]; + /* Do not reclaim the symbols of the rule which action triggered + this YYERROR. */ + YYPOPSTACK (yylen); + yylen = 0; + YY_STACK_PRINT (yyss, yyssp); + yystate = *yyssp; + goto yyerrlab1; + + +/*-------------------------------------------------------------. +| yyerrlab1 -- common code for both syntax error and YYERROR. | +`-------------------------------------------------------------*/ +yyerrlab1: + yyerrstatus = 3; /* Each real token shifted decrements this. */ + + for (;;) + { + yyn = yypact[yystate]; + if (yyn != YYPACT_NINF) + { + yyn += YYTERROR; + if (0 <= yyn && yyn <= YYLAST && yycheck[yyn] == YYTERROR) + { + yyn = yytable[yyn]; + if (0 < yyn) + break; + } + } + + /* Pop the current state because it cannot handle the error token. */ + if (yyssp == yyss) + YYABORT; + + yyerror_range[0] = *yylsp; + yydestruct ("Error: popping", + yystos[yystate], yyvsp, yylsp, state); + YYPOPSTACK (1); + yystate = *yyssp; + YY_STACK_PRINT (yyss, yyssp); + } + + *++yyvsp = yylval; + + yyerror_range[1] = yylloc; + /* Using YYLLOC is tempting, but would change the location of + the lookahead. YYLOC is available though. */ + YYLLOC_DEFAULT (yyloc, (yyerror_range - 1), 2); + *++yylsp = yyloc; + + /* Shift the error token. */ + YY_SYMBOL_PRINT ("Shifting", yystos[yyn], yyvsp, yylsp); + + yystate = yyn; + goto yynewstate; + + +/*-------------------------------------. +| yyacceptlab -- YYACCEPT comes here. | +`-------------------------------------*/ +yyacceptlab: + yyresult = 0; + goto yyreturn; + +/*-----------------------------------. +| yyabortlab -- YYABORT comes here. | +`-----------------------------------*/ +yyabortlab: + yyresult = 1; + goto yyreturn; + +#if !defined(yyoverflow) || YYERROR_VERBOSE +/*-------------------------------------------------. +| yyexhaustedlab -- memory exhaustion comes here. | +`-------------------------------------------------*/ +yyexhaustedlab: + yyerror (&yylloc, state, YY_("memory exhausted")); + yyresult = 2; + /* Fall through. */ +#endif + +yyreturn: + if (yychar != YYEMPTY) + yydestruct ("Cleanup: discarding lookahead", + yytoken, &yylval, &yylloc, state); + /* Do not reclaim the symbols of the rule which action triggered + this YYABORT or YYACCEPT. */ + YYPOPSTACK (yylen); + YY_STACK_PRINT (yyss, yyssp); + while (yyssp != yyss) + { + yydestruct ("Cleanup: popping", + yystos[*yyssp], yyvsp, yylsp, state); + YYPOPSTACK (1); + } +#ifndef yyoverflow + if (yyss != yyssa) + YYSTACK_FREE (yyss); +#endif +#if YYERROR_VERBOSE + if (yymsg != yymsgbuf) + YYSTACK_FREE (yymsg); +#endif + /* Make sure YYID is used. */ + return YYID (yyresult); +} + + + +/* Line 1675 of yacc.c */ +#line 2196 "program_parse.y" + + +void +asm_instruction_set_operands(struct asm_instruction *inst, + const struct prog_dst_register *dst, + const struct asm_src_register *src0, + const struct asm_src_register *src1, + const struct asm_src_register *src2) +{ + /* In the core ARB extensions only the KIL instruction doesn't have a + * destination register. + */ + if (dst == NULL) { + init_dst_reg(& inst->Base.DstReg); + } else { + inst->Base.DstReg = *dst; + } + + /* The only instruction that doesn't have any source registers is the + * condition-code based KIL instruction added by NV_fragment_program_option. + */ + if (src0 != NULL) { + inst->Base.SrcReg[0] = src0->Base; + inst->SrcReg[0] = *src0; + } else { + init_src_reg(& inst->SrcReg[0]); + } + + if (src1 != NULL) { + inst->Base.SrcReg[1] = src1->Base; + inst->SrcReg[1] = *src1; + } else { + init_src_reg(& inst->SrcReg[1]); + } + + if (src2 != NULL) { + inst->Base.SrcReg[2] = src2->Base; + inst->SrcReg[2] = *src2; + } else { + init_src_reg(& inst->SrcReg[2]); + } +} + + +struct asm_instruction * +asm_instruction_ctor(gl_inst_opcode op, + const struct prog_dst_register *dst, + const struct asm_src_register *src0, + const struct asm_src_register *src1, + const struct asm_src_register *src2) +{ + struct asm_instruction *inst = CALLOC_STRUCT(asm_instruction); + + if (inst) { + _mesa_init_instructions(& inst->Base, 1); + inst->Base.Opcode = op; + + asm_instruction_set_operands(inst, dst, src0, src1, src2); + } + + return inst; +} + + +struct asm_instruction * +asm_instruction_copy_ctor(const struct prog_instruction *base, + const struct prog_dst_register *dst, + const struct asm_src_register *src0, + const struct asm_src_register *src1, + const struct asm_src_register *src2) +{ + struct asm_instruction *inst = CALLOC_STRUCT(asm_instruction); + + if (inst) { + _mesa_init_instructions(& inst->Base, 1); + inst->Base.Opcode = base->Opcode; + inst->Base.CondUpdate = base->CondUpdate; + inst->Base.CondDst = base->CondDst; + inst->Base.SaturateMode = base->SaturateMode; + inst->Base.Precision = base->Precision; + + asm_instruction_set_operands(inst, dst, src0, src1, src2); + } + + return inst; +} + + +void +init_dst_reg(struct prog_dst_register *r) +{ + memset(r, 0, sizeof(*r)); + r->File = PROGRAM_UNDEFINED; + r->WriteMask = WRITEMASK_XYZW; + r->CondMask = COND_TR; + r->CondSwizzle = SWIZZLE_NOOP; +} + + +/** Like init_dst_reg() but set the File and Index fields. */ +void +set_dst_reg(struct prog_dst_register *r, gl_register_file file, GLint index) +{ + const GLint maxIndex = 1 << INST_INDEX_BITS; + const GLint minIndex = 0; + ASSERT(index >= minIndex); + (void) minIndex; + ASSERT(index <= maxIndex); + (void) maxIndex; + ASSERT(file == PROGRAM_TEMPORARY || + file == PROGRAM_ADDRESS || + file == PROGRAM_OUTPUT); + memset(r, 0, sizeof(*r)); + r->File = file; + r->Index = index; + r->WriteMask = WRITEMASK_XYZW; + r->CondMask = COND_TR; + r->CondSwizzle = SWIZZLE_NOOP; +} + + +void +init_src_reg(struct asm_src_register *r) +{ + memset(r, 0, sizeof(*r)); + r->Base.File = PROGRAM_UNDEFINED; + r->Base.Swizzle = SWIZZLE_NOOP; + r->Symbol = NULL; +} + + +/** Like init_src_reg() but set the File and Index fields. + * \return GL_TRUE if a valid src register, GL_FALSE otherwise + */ +void +set_src_reg(struct asm_src_register *r, gl_register_file file, GLint index) +{ + set_src_reg_swz(r, file, index, SWIZZLE_XYZW); +} + + +void +set_src_reg_swz(struct asm_src_register *r, gl_register_file file, GLint index, + GLuint swizzle) +{ + const GLint maxIndex = (1 << INST_INDEX_BITS) - 1; + const GLint minIndex = -(1 << INST_INDEX_BITS); + ASSERT(file < PROGRAM_FILE_MAX); + ASSERT(index >= minIndex); + (void) minIndex; + ASSERT(index <= maxIndex); + (void) maxIndex; + memset(r, 0, sizeof(*r)); + r->Base.File = file; + r->Base.Index = index; + r->Base.Swizzle = swizzle; + r->Symbol = NULL; +} + + +/** + * Validate the set of inputs used by a program + * + * Validates that legal sets of inputs are used by the program. In this case + * "used" included both reading the input or binding the input to a name using + * the \c ATTRIB command. + * + * \return + * \c TRUE if the combination of inputs used is valid, \c FALSE otherwise. + */ +int +validate_inputs(struct YYLTYPE *locp, struct asm_parser_state *state) +{ + const int inputs = state->prog->InputsRead | state->InputsBound; + + if (((inputs & 0x0ffff) & (inputs >> 16)) != 0) { + yyerror(locp, state, "illegal use of generic attribute and name attribute"); + return 0; + } + + return 1; +} + + +struct asm_symbol * +declare_variable(struct asm_parser_state *state, char *name, enum asm_type t, + struct YYLTYPE *locp) +{ + struct asm_symbol *s = NULL; + struct asm_symbol *exist = (struct asm_symbol *) + _mesa_symbol_table_find_symbol(state->st, 0, name); + + + if (exist != NULL) { + yyerror(locp, state, "redeclared identifier"); + } else { + s = calloc(1, sizeof(struct asm_symbol)); + s->name = name; + s->type = t; + + switch (t) { + case at_temp: + if (state->prog->NumTemporaries >= state->limits->MaxTemps) { + yyerror(locp, state, "too many temporaries declared"); + free(s); + return NULL; + } + + s->temp_binding = state->prog->NumTemporaries; + state->prog->NumTemporaries++; + break; + + case at_address: + if (state->prog->NumAddressRegs >= state->limits->MaxAddressRegs) { + yyerror(locp, state, "too many address registers declared"); + free(s); + return NULL; + } + + /* FINISHME: Add support for multiple address registers. + */ + state->prog->NumAddressRegs++; + break; + + default: + break; + } + + _mesa_symbol_table_add_symbol(state->st, 0, s->name, s); + s->next = state->sym; + state->sym = s; + } + + return s; +} + + +int add_state_reference(struct gl_program_parameter_list *param_list, + const gl_state_index tokens[STATE_LENGTH]) +{ + const GLuint size = 4; /* XXX fix */ + char *name; + GLint index; + + name = _mesa_program_state_string(tokens); + index = _mesa_add_parameter(param_list, PROGRAM_STATE_VAR, name, + size, GL_NONE, NULL, tokens, 0x0); + param_list->StateFlags |= _mesa_program_state_flags(tokens); + + /* free name string here since we duplicated it in add_parameter() */ + free(name); + + return index; +} + + +int +initialize_symbol_from_state(struct gl_program *prog, + struct asm_symbol *param_var, + const gl_state_index tokens[STATE_LENGTH]) +{ + int idx = -1; + gl_state_index state_tokens[STATE_LENGTH]; + + + memcpy(state_tokens, tokens, sizeof(state_tokens)); + + param_var->type = at_param; + param_var->param_binding_type = PROGRAM_STATE_VAR; + + /* If we are adding a STATE_MATRIX that has multiple rows, we need to + * unroll it and call add_state_reference() for each row + */ + if ((state_tokens[0] == STATE_MODELVIEW_MATRIX || + state_tokens[0] == STATE_PROJECTION_MATRIX || + state_tokens[0] == STATE_MVP_MATRIX || + state_tokens[0] == STATE_TEXTURE_MATRIX || + state_tokens[0] == STATE_PROGRAM_MATRIX) + && (state_tokens[2] != state_tokens[3])) { + int row; + const int first_row = state_tokens[2]; + const int last_row = state_tokens[3]; + + for (row = first_row; row <= last_row; row++) { + state_tokens[2] = state_tokens[3] = row; + + idx = add_state_reference(prog->Parameters, state_tokens); + if (param_var->param_binding_begin == ~0U) { + param_var->param_binding_begin = idx; + param_var->param_binding_swizzle = SWIZZLE_XYZW; + } + + param_var->param_binding_length++; + } + } + else { + idx = add_state_reference(prog->Parameters, state_tokens); + if (param_var->param_binding_begin == ~0U) { + param_var->param_binding_begin = idx; + param_var->param_binding_swizzle = SWIZZLE_XYZW; + } + param_var->param_binding_length++; + } + + return idx; +} + + +int +initialize_symbol_from_param(struct gl_program *prog, + struct asm_symbol *param_var, + const gl_state_index tokens[STATE_LENGTH]) +{ + int idx = -1; + gl_state_index state_tokens[STATE_LENGTH]; + + + memcpy(state_tokens, tokens, sizeof(state_tokens)); + + assert((state_tokens[0] == STATE_VERTEX_PROGRAM) + || (state_tokens[0] == STATE_FRAGMENT_PROGRAM)); + assert((state_tokens[1] == STATE_ENV) + || (state_tokens[1] == STATE_LOCAL)); + + /* + * The param type is STATE_VAR. The program parameter entry will + * effectively be a pointer into the LOCAL or ENV parameter array. + */ + param_var->type = at_param; + param_var->param_binding_type = PROGRAM_STATE_VAR; + + /* If we are adding a STATE_ENV or STATE_LOCAL that has multiple elements, + * we need to unroll it and call add_state_reference() for each row + */ + if (state_tokens[2] != state_tokens[3]) { + int row; + const int first_row = state_tokens[2]; + const int last_row = state_tokens[3]; + + for (row = first_row; row <= last_row; row++) { + state_tokens[2] = state_tokens[3] = row; + + idx = add_state_reference(prog->Parameters, state_tokens); + if (param_var->param_binding_begin == ~0U) { + param_var->param_binding_begin = idx; + param_var->param_binding_swizzle = SWIZZLE_XYZW; + } + param_var->param_binding_length++; + } + } + else { + idx = add_state_reference(prog->Parameters, state_tokens); + if (param_var->param_binding_begin == ~0U) { + param_var->param_binding_begin = idx; + param_var->param_binding_swizzle = SWIZZLE_XYZW; + } + param_var->param_binding_length++; + } + + return idx; +} + + +/** + * Put a float/vector constant/literal into the parameter list. + * \param param_var returns info about the parameter/constant's location, + * binding, type, etc. + * \param vec the vector/constant to add + * \param allowSwizzle if true, try to consolidate constants which only differ + * by a swizzle. We don't want to do this when building + * arrays of constants that may be indexed indirectly. + * \return index of the constant in the parameter list. + */ +int +initialize_symbol_from_const(struct gl_program *prog, + struct asm_symbol *param_var, + const struct asm_vector *vec, + GLboolean allowSwizzle) +{ + unsigned swizzle; + const int idx = _mesa_add_unnamed_constant(prog->Parameters, + vec->data, vec->count, + allowSwizzle ? &swizzle : NULL); + + param_var->type = at_param; + param_var->param_binding_type = PROGRAM_CONSTANT; + + if (param_var->param_binding_begin == ~0U) { + param_var->param_binding_begin = idx; + param_var->param_binding_swizzle = allowSwizzle ? swizzle : SWIZZLE_XYZW; + } + param_var->param_binding_length++; + + return idx; +} + + +char * +make_error_string(const char *fmt, ...) +{ + int length; + char *str; + va_list args; + + + /* Call vsnprintf once to determine how large the final string is. Call it + * again to do the actual formatting. from the vsnprintf manual page: + * + * Upon successful return, these functions return the number of + * characters printed (not including the trailing '\0' used to end + * output to strings). + */ + va_start(args, fmt); + length = 1 + vsnprintf(NULL, 0, fmt, args); + va_end(args); + + str = malloc(length); + if (str) { + va_start(args, fmt); + vsnprintf(str, length, fmt, args); + va_end(args); + } + + return str; +} + + +void +yyerror(YYLTYPE *locp, struct asm_parser_state *state, const char *s) +{ + char *err_str; + + + err_str = make_error_string("glProgramStringARB(%s)\n", s); + if (err_str) { + _mesa_error(state->ctx, GL_INVALID_OPERATION, "%s", err_str); + free(err_str); + } + + err_str = make_error_string("line %u, char %u: error: %s\n", + locp->first_line, locp->first_column, s); + _mesa_set_program_error(state->ctx, locp->position, err_str); + + if (err_str) { + free(err_str); + } +} + + +GLboolean +_mesa_parse_arb_program(GLcontext *ctx, GLenum target, const GLubyte *str, + GLsizei len, struct asm_parser_state *state) +{ + struct asm_instruction *inst; + unsigned i; + GLubyte *strz; + GLboolean result = GL_FALSE; + void *temp; + struct asm_symbol *sym; + + state->ctx = ctx; + state->prog->Target = target; + state->prog->Parameters = _mesa_new_parameter_list(); + + /* Make a copy of the program string and force it to be NUL-terminated. + */ + strz = (GLubyte *) malloc(len + 1); + if (strz == NULL) { + _mesa_error(ctx, GL_OUT_OF_MEMORY, "glProgramStringARB"); + return GL_FALSE; + } + memcpy (strz, str, len); + strz[len] = '\0'; + + state->prog->String = strz; + + state->st = _mesa_symbol_table_ctor(); + + state->limits = (target == GL_VERTEX_PROGRAM_ARB) + ? & ctx->Const.VertexProgram + : & ctx->Const.FragmentProgram; + + state->MaxTextureImageUnits = ctx->Const.MaxTextureImageUnits; + state->MaxTextureCoordUnits = ctx->Const.MaxTextureCoordUnits; + state->MaxTextureUnits = ctx->Const.MaxTextureUnits; + state->MaxClipPlanes = ctx->Const.MaxClipPlanes; + state->MaxLights = ctx->Const.MaxLights; + state->MaxProgramMatrices = ctx->Const.MaxProgramMatrices; + + state->state_param_enum = (target == GL_VERTEX_PROGRAM_ARB) + ? STATE_VERTEX_PROGRAM : STATE_FRAGMENT_PROGRAM; + + _mesa_set_program_error(ctx, -1, NULL); + + _mesa_program_lexer_ctor(& state->scanner, state, (const char *) str, len); + yyparse(state); + _mesa_program_lexer_dtor(state->scanner); + + + if (ctx->Program.ErrorPos != -1) { + goto error; + } + + if (! _mesa_layout_parameters(state)) { + struct YYLTYPE loc; + + loc.first_line = 0; + loc.first_column = 0; + loc.position = len; + + yyerror(& loc, state, "invalid PARAM usage"); + goto error; + } + + + + /* Add one instruction to store the "END" instruction. + */ + state->prog->Instructions = + _mesa_alloc_instructions(state->prog->NumInstructions + 1); + inst = state->inst_head; + for (i = 0; i < state->prog->NumInstructions; i++) { + struct asm_instruction *const temp = inst->next; + + state->prog->Instructions[i] = inst->Base; + inst = temp; + } + + /* Finally, tag on an OPCODE_END instruction */ + { + const GLuint numInst = state->prog->NumInstructions; + _mesa_init_instructions(state->prog->Instructions + numInst, 1); + state->prog->Instructions[numInst].Opcode = OPCODE_END; + } + state->prog->NumInstructions++; + + state->prog->NumParameters = state->prog->Parameters->NumParameters; + state->prog->NumAttributes = _mesa_bitcount(state->prog->InputsRead); + + /* + * Initialize native counts to logical counts. The device driver may + * change them if program is translated into a hardware program. + */ + state->prog->NumNativeInstructions = state->prog->NumInstructions; + state->prog->NumNativeTemporaries = state->prog->NumTemporaries; + state->prog->NumNativeParameters = state->prog->NumParameters; + state->prog->NumNativeAttributes = state->prog->NumAttributes; + state->prog->NumNativeAddressRegs = state->prog->NumAddressRegs; + + result = GL_TRUE; + +error: + for (inst = state->inst_head; inst != NULL; inst = temp) { + temp = inst->next; + free(inst); + } + + state->inst_head = NULL; + state->inst_tail = NULL; + + for (sym = state->sym; sym != NULL; sym = temp) { + temp = sym->next; + + free((void *) sym->name); + free(sym); + } + state->sym = NULL; + + _mesa_symbol_table_dtor(state->st); + state->st = NULL; + + return result; +} + diff --git a/mesalib/src/mesa/program/program_parse.tab.h b/mesalib/src/mesa/program/program_parse.tab.h new file mode 100644 index 000000000..045241d9e --- /dev/null +++ b/mesalib/src/mesa/program/program_parse.tab.h @@ -0,0 +1,209 @@ + +/* A Bison parser, made by GNU Bison 2.4.1. */ + +/* Skeleton interface for Bison's Yacc-like parsers in C + + Copyright (C) 1984, 1989, 1990, 2000, 2001, 2002, 2003, 2004, 2005, 2006 + Free Software Foundation, Inc. + + This program is free software: you can redistribute it and/or modify + it under the terms of the GNU General Public License as published by + the Free Software Foundation, either version 3 of the License, or + (at your option) any later version. + + This program is distributed in the hope that it will be useful, + but WITHOUT ANY WARRANTY; without even the implied warranty of + MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the + GNU General Public License for more details. + + You should have received a copy of the GNU General Public License + along with this program. If not, see <http://www.gnu.org/licenses/>. */ + +/* As a special exception, you may create a larger work that contains + part or all of the Bison parser skeleton and distribute that work + under terms of your choice, so long as that work isn't itself a + parser generator using the skeleton or a modified version thereof + as a parser skeleton. Alternatively, if you modify or redistribute + the parser skeleton itself, you may (at your option) remove this + special exception, which will cause the skeleton and the resulting + Bison output files to be licensed under the GNU General Public + License without this special exception. + + This special exception was added by the Free Software Foundation in + version 2.2 of Bison. */ + + +/* Tokens. */ +#ifndef YYTOKENTYPE +# define YYTOKENTYPE + /* Put the tokens into the symbol table, so that GDB and other debuggers + know about them. */ + enum yytokentype { + ARBvp_10 = 258, + ARBfp_10 = 259, + ADDRESS = 260, + ALIAS = 261, + ATTRIB = 262, + OPTION = 263, + OUTPUT = 264, + PARAM = 265, + TEMP = 266, + END = 267, + BIN_OP = 268, + BINSC_OP = 269, + SAMPLE_OP = 270, + SCALAR_OP = 271, + TRI_OP = 272, + VECTOR_OP = 273, + ARL = 274, + KIL = 275, + SWZ = 276, + TXD_OP = 277, + INTEGER = 278, + REAL = 279, + AMBIENT = 280, + ATTENUATION = 281, + BACK = 282, + CLIP = 283, + COLOR = 284, + DEPTH = 285, + DIFFUSE = 286, + DIRECTION = 287, + EMISSION = 288, + ENV = 289, + EYE = 290, + FOG = 291, + FOGCOORD = 292, + FRAGMENT = 293, + FRONT = 294, + HALF = 295, + INVERSE = 296, + INVTRANS = 297, + LIGHT = 298, + LIGHTMODEL = 299, + LIGHTPROD = 300, + LOCAL = 301, + MATERIAL = 302, + MAT_PROGRAM = 303, + MATRIX = 304, + MATRIXINDEX = 305, + MODELVIEW = 306, + MVP = 307, + NORMAL = 308, + OBJECT = 309, + PALETTE = 310, + PARAMS = 311, + PLANE = 312, + POINT_TOK = 313, + POINTSIZE = 314, + POSITION = 315, + PRIMARY = 316, + PROGRAM = 317, + PROJECTION = 318, + RANGE = 319, + RESULT = 320, + ROW = 321, + SCENECOLOR = 322, + SECONDARY = 323, + SHININESS = 324, + SIZE_TOK = 325, + SPECULAR = 326, + SPOT = 327, + STATE = 328, + TEXCOORD = 329, + TEXENV = 330, + TEXGEN = 331, + TEXGEN_Q = 332, + TEXGEN_R = 333, + TEXGEN_S = 334, + TEXGEN_T = 335, + TEXTURE = 336, + TRANSPOSE = 337, + TEXTURE_UNIT = 338, + TEX_1D = 339, + TEX_2D = 340, + TEX_3D = 341, + TEX_CUBE = 342, + TEX_RECT = 343, + TEX_SHADOW1D = 344, + TEX_SHADOW2D = 345, + TEX_SHADOWRECT = 346, + TEX_ARRAY1D = 347, + TEX_ARRAY2D = 348, + TEX_ARRAYSHADOW1D = 349, + TEX_ARRAYSHADOW2D = 350, + VERTEX = 351, + VTXATTRIB = 352, + WEIGHT = 353, + IDENTIFIER = 354, + USED_IDENTIFIER = 355, + MASK4 = 356, + MASK3 = 357, + MASK2 = 358, + MASK1 = 359, + SWIZZLE = 360, + DOT_DOT = 361, + DOT = 362 + }; +#endif + + + +#if ! defined YYSTYPE && ! defined YYSTYPE_IS_DECLARED +typedef union YYSTYPE +{ + +/* Line 1676 of yacc.c */ +#line 126 "program_parse.y" + + struct asm_instruction *inst; + struct asm_symbol *sym; + struct asm_symbol temp_sym; + struct asm_swizzle_mask swiz_mask; + struct asm_src_register src_reg; + struct prog_dst_register dst_reg; + struct prog_instruction temp_inst; + char *string; + unsigned result; + unsigned attrib; + int integer; + float real; + gl_state_index state[STATE_LENGTH]; + int negate; + struct asm_vector vector; + gl_inst_opcode opcode; + + struct { + unsigned swz; + unsigned rgba_valid:1; + unsigned xyzw_valid:1; + unsigned negate:1; + } ext_swizzle; + + + +/* Line 1676 of yacc.c */ +#line 187 "program_parse.tab.h" +} YYSTYPE; +# define YYSTYPE_IS_TRIVIAL 1 +# define yystype YYSTYPE /* obsolescent; will be withdrawn */ +# define YYSTYPE_IS_DECLARED 1 +#endif + + + +#if ! defined YYLTYPE && ! defined YYLTYPE_IS_DECLARED +typedef struct YYLTYPE +{ + int first_line; + int first_column; + int last_line; + int last_column; +} YYLTYPE; +# define yyltype YYLTYPE /* obsolescent; will be withdrawn */ +# define YYLTYPE_IS_DECLARED 1 +# define YYLTYPE_IS_TRIVIAL 1 +#endif + + + diff --git a/mesalib/src/mesa/program/program_parse.y b/mesalib/src/mesa/program/program_parse.y new file mode 100644 index 000000000..cf621ae42 --- /dev/null +++ b/mesalib/src/mesa/program/program_parse.y @@ -0,0 +1,2768 @@ +%{ +/* + * Copyright © 2009 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. + */ +#include <stdio.h> +#include <stdlib.h> +#include <string.h> + +#include "main/mtypes.h" +#include "main/imports.h" +#include "program/program.h" +#include "program/prog_parameter.h" +#include "program/prog_parameter_layout.h" +#include "program/prog_statevars.h" +#include "program/prog_instruction.h" + +#include "program/symbol_table.h" +#include "program/program_parser.h" + +extern void *yy_scan_string(char *); +extern void yy_delete_buffer(void *); + +static struct asm_symbol *declare_variable(struct asm_parser_state *state, + char *name, enum asm_type t, struct YYLTYPE *locp); + +static int add_state_reference(struct gl_program_parameter_list *param_list, + const gl_state_index tokens[STATE_LENGTH]); + +static int initialize_symbol_from_state(struct gl_program *prog, + struct asm_symbol *param_var, const gl_state_index tokens[STATE_LENGTH]); + +static int initialize_symbol_from_param(struct gl_program *prog, + struct asm_symbol *param_var, const gl_state_index tokens[STATE_LENGTH]); + +static int initialize_symbol_from_const(struct gl_program *prog, + struct asm_symbol *param_var, const struct asm_vector *vec, + GLboolean allowSwizzle); + +static int yyparse(struct asm_parser_state *state); + +static char *make_error_string(const char *fmt, ...); + +static void yyerror(struct YYLTYPE *locp, struct asm_parser_state *state, + const char *s); + +static int validate_inputs(struct YYLTYPE *locp, + struct asm_parser_state *state); + +static void init_dst_reg(struct prog_dst_register *r); + +static void set_dst_reg(struct prog_dst_register *r, + gl_register_file file, GLint index); + +static void init_src_reg(struct asm_src_register *r); + +static void set_src_reg(struct asm_src_register *r, + gl_register_file file, GLint index); + +static void set_src_reg_swz(struct asm_src_register *r, + gl_register_file file, GLint index, GLuint swizzle); + +static void asm_instruction_set_operands(struct asm_instruction *inst, + const struct prog_dst_register *dst, const struct asm_src_register *src0, + const struct asm_src_register *src1, const struct asm_src_register *src2); + +static struct asm_instruction *asm_instruction_ctor(gl_inst_opcode op, + const struct prog_dst_register *dst, const struct asm_src_register *src0, + const struct asm_src_register *src1, const struct asm_src_register *src2); + +static struct asm_instruction *asm_instruction_copy_ctor( + const struct prog_instruction *base, const struct prog_dst_register *dst, + const struct asm_src_register *src0, const struct asm_src_register *src1, + const struct asm_src_register *src2); + +#ifndef FALSE +#define FALSE 0 +#define TRUE (!FALSE) +#endif + +#define YYLLOC_DEFAULT(Current, Rhs, N) \ + do { \ + if (YYID(N)) { \ + (Current).first_line = YYRHSLOC(Rhs, 1).first_line; \ + (Current).first_column = YYRHSLOC(Rhs, 1).first_column; \ + (Current).position = YYRHSLOC(Rhs, 1).position; \ + (Current).last_line = YYRHSLOC(Rhs, N).last_line; \ + (Current).last_column = YYRHSLOC(Rhs, N).last_column; \ + } else { \ + (Current).first_line = YYRHSLOC(Rhs, 0).last_line; \ + (Current).last_line = (Current).first_line; \ + (Current).first_column = YYRHSLOC(Rhs, 0).last_column; \ + (Current).last_column = (Current).first_column; \ + (Current).position = YYRHSLOC(Rhs, 0).position \ + + (Current).first_column; \ + } \ + } while(YYID(0)) + +#define YYLEX_PARAM state->scanner +%} + +%pure-parser +%locations +%parse-param { struct asm_parser_state *state } +%error-verbose +%lex-param { void *scanner } + +%union { + struct asm_instruction *inst; + struct asm_symbol *sym; + struct asm_symbol temp_sym; + struct asm_swizzle_mask swiz_mask; + struct asm_src_register src_reg; + struct prog_dst_register dst_reg; + struct prog_instruction temp_inst; + char *string; + unsigned result; + unsigned attrib; + int integer; + float real; + gl_state_index state[STATE_LENGTH]; + int negate; + struct asm_vector vector; + gl_inst_opcode opcode; + + struct { + unsigned swz; + unsigned rgba_valid:1; + unsigned xyzw_valid:1; + unsigned negate:1; + } ext_swizzle; +} + +%token ARBvp_10 ARBfp_10 + +/* Tokens for assembler pseudo-ops */ +%token <integer> ADDRESS +%token ALIAS ATTRIB +%token OPTION OUTPUT +%token PARAM +%token <integer> TEMP +%token END + + /* Tokens for instructions */ +%token <temp_inst> BIN_OP BINSC_OP SAMPLE_OP SCALAR_OP TRI_OP VECTOR_OP +%token <temp_inst> ARL KIL SWZ TXD_OP + +%token <integer> INTEGER +%token <real> REAL + +%token AMBIENT ATTENUATION +%token BACK +%token CLIP COLOR +%token DEPTH DIFFUSE DIRECTION +%token EMISSION ENV EYE +%token FOG FOGCOORD FRAGMENT FRONT +%token HALF +%token INVERSE INVTRANS +%token LIGHT LIGHTMODEL LIGHTPROD LOCAL +%token MATERIAL MAT_PROGRAM MATRIX MATRIXINDEX MODELVIEW MVP +%token NORMAL +%token OBJECT +%token PALETTE PARAMS PLANE POINT_TOK POINTSIZE POSITION PRIMARY PROGRAM PROJECTION +%token RANGE RESULT ROW +%token SCENECOLOR SECONDARY SHININESS SIZE_TOK SPECULAR SPOT STATE +%token TEXCOORD TEXENV TEXGEN TEXGEN_Q TEXGEN_R TEXGEN_S TEXGEN_T TEXTURE TRANSPOSE +%token TEXTURE_UNIT TEX_1D TEX_2D TEX_3D TEX_CUBE TEX_RECT +%token TEX_SHADOW1D TEX_SHADOW2D TEX_SHADOWRECT +%token TEX_ARRAY1D TEX_ARRAY2D TEX_ARRAYSHADOW1D TEX_ARRAYSHADOW2D +%token VERTEX VTXATTRIB +%token WEIGHT + +%token <string> IDENTIFIER USED_IDENTIFIER +%type <string> string +%token <swiz_mask> MASK4 MASK3 MASK2 MASK1 SWIZZLE +%token DOT_DOT +%token DOT + +%type <inst> instruction ALU_instruction TexInstruction +%type <inst> ARL_instruction VECTORop_instruction +%type <inst> SCALARop_instruction BINSCop_instruction BINop_instruction +%type <inst> TRIop_instruction TXD_instruction SWZ_instruction SAMPLE_instruction +%type <inst> KIL_instruction + +%type <dst_reg> dstReg maskedDstReg maskedAddrReg +%type <src_reg> srcReg scalarUse scalarSrcReg swizzleSrcReg +%type <swiz_mask> scalarSuffix swizzleSuffix extendedSwizzle +%type <ext_swizzle> extSwizComp extSwizSel +%type <swiz_mask> optionalMask + +%type <sym> progParamArray +%type <integer> addrRegRelOffset addrRegPosOffset addrRegNegOffset +%type <src_reg> progParamArrayMem progParamArrayAbs progParamArrayRel +%type <sym> addrReg +%type <swiz_mask> addrComponent addrWriteMask + +%type <dst_reg> ccMaskRule ccTest ccMaskRule2 ccTest2 optionalCcMask + +%type <result> resultBinding resultColBinding +%type <integer> optFaceType optColorType +%type <integer> optResultFaceType optResultColorType + +%type <integer> optTexImageUnitNum texImageUnitNum +%type <integer> optTexCoordUnitNum texCoordUnitNum +%type <integer> optLegacyTexUnitNum legacyTexUnitNum +%type <integer> texImageUnit texTarget +%type <integer> vtxAttribNum + +%type <attrib> attribBinding vtxAttribItem fragAttribItem + +%type <temp_sym> paramSingleInit paramSingleItemDecl +%type <integer> optArraySize + +%type <state> stateSingleItem stateMultipleItem +%type <state> stateMaterialItem +%type <state> stateLightItem stateLightModelItem stateLightProdItem +%type <state> stateTexGenItem stateFogItem stateClipPlaneItem statePointItem +%type <state> stateMatrixItem stateMatrixRow stateMatrixRows +%type <state> stateTexEnvItem stateDepthItem + +%type <state> stateLModProperty +%type <state> stateMatrixName optMatrixRows + +%type <integer> stateMatProperty +%type <integer> stateLightProperty stateSpotProperty +%type <integer> stateLightNumber stateLProdProperty +%type <integer> stateTexGenType stateTexGenCoord +%type <integer> stateTexEnvProperty +%type <integer> stateFogProperty +%type <integer> stateClipPlaneNum +%type <integer> statePointProperty + +%type <integer> stateOptMatModifier stateMatModifier stateMatrixRowNum +%type <integer> stateOptModMatNum stateModMatNum statePaletteMatNum +%type <integer> stateProgramMatNum + +%type <integer> ambDiffSpecProperty + +%type <state> programSingleItem progEnvParam progLocalParam +%type <state> programMultipleItem progEnvParams progLocalParams + +%type <temp_sym> paramMultipleInit paramMultInitList paramMultipleItem +%type <temp_sym> paramSingleItemUse + +%type <integer> progEnvParamNum progLocalParamNum +%type <state> progEnvParamNums progLocalParamNums + +%type <vector> paramConstDecl paramConstUse +%type <vector> paramConstScalarDecl paramConstScalarUse paramConstVector +%type <real> signedFloatConstant +%type <negate> optionalSign + +%{ +extern int yylex(YYSTYPE *yylval_param, YYLTYPE *yylloc_param, + void *yyscanner); +%} + +%% + +program: language optionSequence statementSequence END + ; + +language: ARBvp_10 + { + if (state->prog->Target != GL_VERTEX_PROGRAM_ARB) { + yyerror(& @1, state, "invalid fragment program header"); + + } + state->mode = ARB_vertex; + } + | ARBfp_10 + { + if (state->prog->Target != GL_FRAGMENT_PROGRAM_ARB) { + yyerror(& @1, state, "invalid vertex program header"); + } + state->mode = ARB_fragment; + + state->option.TexRect = + (state->ctx->Extensions.NV_texture_rectangle != GL_FALSE); + } + ; + +optionSequence: optionSequence option + | + ; + +option: OPTION string ';' + { + int valid = 0; + + if (state->mode == ARB_vertex) { + valid = _mesa_ARBvp_parse_option(state, $2); + } else if (state->mode == ARB_fragment) { + valid = _mesa_ARBfp_parse_option(state, $2); + } + + + free($2); + + if (!valid) { + const char *const err_str = (state->mode == ARB_vertex) + ? "invalid ARB vertex program option" + : "invalid ARB fragment program option"; + + yyerror(& @2, state, err_str); + YYERROR; + } + } + ; + +statementSequence: statementSequence statement + | + ; + +statement: instruction ';' + { + if ($1 != NULL) { + if (state->inst_tail == NULL) { + state->inst_head = $1; + } else { + state->inst_tail->next = $1; + } + + state->inst_tail = $1; + $1->next = NULL; + + state->prog->NumInstructions++; + } + } + | namingStatement ';' + ; + +instruction: ALU_instruction + { + $$ = $1; + state->prog->NumAluInstructions++; + } + | TexInstruction + { + $$ = $1; + state->prog->NumTexInstructions++; + } + ; + +ALU_instruction: ARL_instruction + | VECTORop_instruction + | SCALARop_instruction + | BINSCop_instruction + | BINop_instruction + | TRIop_instruction + | SWZ_instruction + ; + +TexInstruction: SAMPLE_instruction + | KIL_instruction + | TXD_instruction + ; + +ARL_instruction: ARL maskedAddrReg ',' scalarSrcReg + { + $$ = asm_instruction_ctor(OPCODE_ARL, & $2, & $4, NULL, NULL); + } + ; + +VECTORop_instruction: VECTOR_OP maskedDstReg ',' swizzleSrcReg + { + $$ = asm_instruction_copy_ctor(& $1, & $2, & $4, NULL, NULL); + } + ; + +SCALARop_instruction: SCALAR_OP maskedDstReg ',' scalarSrcReg + { + $$ = asm_instruction_copy_ctor(& $1, & $2, & $4, NULL, NULL); + } + ; + +BINSCop_instruction: BINSC_OP maskedDstReg ',' scalarSrcReg ',' scalarSrcReg + { + $$ = asm_instruction_copy_ctor(& $1, & $2, & $4, & $6, NULL); + } + ; + + +BINop_instruction: BIN_OP maskedDstReg ',' swizzleSrcReg ',' swizzleSrcReg + { + $$ = asm_instruction_copy_ctor(& $1, & $2, & $4, & $6, NULL); + } + ; + +TRIop_instruction: TRI_OP maskedDstReg ',' + swizzleSrcReg ',' swizzleSrcReg ',' swizzleSrcReg + { + $$ = asm_instruction_copy_ctor(& $1, & $2, & $4, & $6, & $8); + } + ; + +SAMPLE_instruction: SAMPLE_OP maskedDstReg ',' swizzleSrcReg ',' texImageUnit ',' texTarget + { + $$ = asm_instruction_copy_ctor(& $1, & $2, & $4, NULL, NULL); + if ($$ != NULL) { + const GLbitfield tex_mask = (1U << $6); + GLbitfield shadow_tex = 0; + GLbitfield target_mask = 0; + + + $$->Base.TexSrcUnit = $6; + + if ($8 < 0) { + shadow_tex = tex_mask; + + $$->Base.TexSrcTarget = -$8; + $$->Base.TexShadow = 1; + } else { + $$->Base.TexSrcTarget = $8; + } + + target_mask = (1U << $$->Base.TexSrcTarget); + + /* If this texture unit was previously accessed and that access + * had a different texture target, generate an error. + * + * If this texture unit was previously accessed and that access + * had a different shadow mode, generate an error. + */ + if ((state->prog->TexturesUsed[$6] != 0) + && ((state->prog->TexturesUsed[$6] != target_mask) + || ((state->prog->ShadowSamplers & tex_mask) + != shadow_tex))) { + yyerror(& @8, state, + "multiple targets used on one texture image unit"); + YYERROR; + } + + + state->prog->TexturesUsed[$6] |= target_mask; + state->prog->ShadowSamplers |= shadow_tex; + } + } + ; + +KIL_instruction: KIL swizzleSrcReg + { + $$ = asm_instruction_ctor(OPCODE_KIL, NULL, & $2, NULL, NULL); + state->fragment.UsesKill = 1; + } + | KIL ccTest + { + $$ = asm_instruction_ctor(OPCODE_KIL_NV, NULL, NULL, NULL, NULL); + $$->Base.DstReg.CondMask = $2.CondMask; + $$->Base.DstReg.CondSwizzle = $2.CondSwizzle; + $$->Base.DstReg.CondSrc = $2.CondSrc; + state->fragment.UsesKill = 1; + } + ; + +TXD_instruction: TXD_OP maskedDstReg ',' swizzleSrcReg ',' swizzleSrcReg ',' swizzleSrcReg ',' texImageUnit ',' texTarget + { + $$ = asm_instruction_copy_ctor(& $1, & $2, & $4, & $6, & $8); + if ($$ != NULL) { + const GLbitfield tex_mask = (1U << $10); + GLbitfield shadow_tex = 0; + GLbitfield target_mask = 0; + + + $$->Base.TexSrcUnit = $10; + + if ($12 < 0) { + shadow_tex = tex_mask; + + $$->Base.TexSrcTarget = -$12; + $$->Base.TexShadow = 1; + } else { + $$->Base.TexSrcTarget = $12; + } + + target_mask = (1U << $$->Base.TexSrcTarget); + + /* If this texture unit was previously accessed and that access + * had a different texture target, generate an error. + * + * If this texture unit was previously accessed and that access + * had a different shadow mode, generate an error. + */ + if ((state->prog->TexturesUsed[$10] != 0) + && ((state->prog->TexturesUsed[$10] != target_mask) + || ((state->prog->ShadowSamplers & tex_mask) + != shadow_tex))) { + yyerror(& @12, state, + "multiple targets used on one texture image unit"); + YYERROR; + } + + + state->prog->TexturesUsed[$10] |= target_mask; + state->prog->ShadowSamplers |= shadow_tex; + } + } + ; + +texImageUnit: TEXTURE_UNIT optTexImageUnitNum + { + $$ = $2; + } + ; + +texTarget: TEX_1D { $$ = TEXTURE_1D_INDEX; } + | TEX_2D { $$ = TEXTURE_2D_INDEX; } + | TEX_3D { $$ = TEXTURE_3D_INDEX; } + | TEX_CUBE { $$ = TEXTURE_CUBE_INDEX; } + | TEX_RECT { $$ = TEXTURE_RECT_INDEX; } + | TEX_SHADOW1D { $$ = -TEXTURE_1D_INDEX; } + | TEX_SHADOW2D { $$ = -TEXTURE_2D_INDEX; } + | TEX_SHADOWRECT { $$ = -TEXTURE_RECT_INDEX; } + | TEX_ARRAY1D { $$ = TEXTURE_1D_ARRAY_INDEX; } + | TEX_ARRAY2D { $$ = TEXTURE_2D_ARRAY_INDEX; } + | TEX_ARRAYSHADOW1D { $$ = -TEXTURE_1D_ARRAY_INDEX; } + | TEX_ARRAYSHADOW2D { $$ = -TEXTURE_2D_ARRAY_INDEX; } + ; + +SWZ_instruction: SWZ maskedDstReg ',' srcReg ',' extendedSwizzle + { + /* FIXME: Is this correct? Should the extenedSwizzle be applied + * FIXME: to the existing swizzle? + */ + $4.Base.Swizzle = $6.swizzle; + $4.Base.Negate = $6.mask; + + $$ = asm_instruction_copy_ctor(& $1, & $2, & $4, NULL, NULL); + } + ; + +scalarSrcReg: optionalSign scalarUse + { + $$ = $2; + + if ($1) { + $$.Base.Negate = ~$$.Base.Negate; + } + } + | optionalSign '|' scalarUse '|' + { + $$ = $3; + + if (!state->option.NV_fragment) { + yyerror(& @2, state, "unexpected character '|'"); + YYERROR; + } + + if ($1) { + $$.Base.Negate = ~$$.Base.Negate; + } + + $$.Base.Abs = 1; + } + ; + +scalarUse: srcReg scalarSuffix + { + $$ = $1; + + $$.Base.Swizzle = _mesa_combine_swizzles($$.Base.Swizzle, + $2.swizzle); + } + | paramConstScalarUse + { + struct asm_symbol temp_sym; + + if (!state->option.NV_fragment) { + yyerror(& @1, state, "expected scalar suffix"); + YYERROR; + } + + memset(& temp_sym, 0, sizeof(temp_sym)); + temp_sym.param_binding_begin = ~0; + initialize_symbol_from_const(state->prog, & temp_sym, & $1, GL_TRUE); + + set_src_reg_swz(& $$, PROGRAM_CONSTANT, + temp_sym.param_binding_begin, + temp_sym.param_binding_swizzle); + } + ; + +swizzleSrcReg: optionalSign srcReg swizzleSuffix + { + $$ = $2; + + if ($1) { + $$.Base.Negate = ~$$.Base.Negate; + } + + $$.Base.Swizzle = _mesa_combine_swizzles($$.Base.Swizzle, + $3.swizzle); + } + | optionalSign '|' srcReg swizzleSuffix '|' + { + $$ = $3; + + if (!state->option.NV_fragment) { + yyerror(& @2, state, "unexpected character '|'"); + YYERROR; + } + + if ($1) { + $$.Base.Negate = ~$$.Base.Negate; + } + + $$.Base.Abs = 1; + $$.Base.Swizzle = _mesa_combine_swizzles($$.Base.Swizzle, + $4.swizzle); + } + + ; + +maskedDstReg: dstReg optionalMask optionalCcMask + { + $$ = $1; + $$.WriteMask = $2.mask; + $$.CondMask = $3.CondMask; + $$.CondSwizzle = $3.CondSwizzle; + $$.CondSrc = $3.CondSrc; + + if ($$.File == PROGRAM_OUTPUT) { + /* Technically speaking, this should check that it is in + * vertex program mode. However, PositionInvariant can never be + * set in fragment program mode, so it is somewhat irrelevant. + */ + if (state->option.PositionInvariant + && ($$.Index == VERT_RESULT_HPOS)) { + yyerror(& @1, state, "position-invariant programs cannot " + "write position"); + YYERROR; + } + + state->prog->OutputsWritten |= BITFIELD64_BIT($$.Index); + } + } + ; + +maskedAddrReg: addrReg addrWriteMask + { + set_dst_reg(& $$, PROGRAM_ADDRESS, 0); + $$.WriteMask = $2.mask; + } + ; + +extendedSwizzle: extSwizComp ',' extSwizComp ',' extSwizComp ',' extSwizComp + { + const unsigned xyzw_valid = + ($1.xyzw_valid << 0) + | ($3.xyzw_valid << 1) + | ($5.xyzw_valid << 2) + | ($7.xyzw_valid << 3); + const unsigned rgba_valid = + ($1.rgba_valid << 0) + | ($3.rgba_valid << 1) + | ($5.rgba_valid << 2) + | ($7.rgba_valid << 3); + + /* All of the swizzle components have to be valid in either RGBA + * or XYZW. Note that 0 and 1 are valid in both, so both masks + * can have some bits set. + * + * We somewhat deviate from the spec here. It would be really hard + * to figure out which component is the error, and there probably + * isn't a lot of benefit. + */ + if ((rgba_valid != 0x0f) && (xyzw_valid != 0x0f)) { + yyerror(& @1, state, "cannot combine RGBA and XYZW swizzle " + "components"); + YYERROR; + } + + $$.swizzle = MAKE_SWIZZLE4($1.swz, $3.swz, $5.swz, $7.swz); + $$.mask = ($1.negate) | ($3.negate << 1) | ($5.negate << 2) + | ($7.negate << 3); + } + ; + +extSwizComp: optionalSign extSwizSel + { + $$ = $2; + $$.negate = ($1) ? 1 : 0; + } + ; + +extSwizSel: INTEGER + { + if (($1 != 0) && ($1 != 1)) { + yyerror(& @1, state, "invalid extended swizzle selector"); + YYERROR; + } + + $$.swz = ($1 == 0) ? SWIZZLE_ZERO : SWIZZLE_ONE; + + /* 0 and 1 are valid for both RGBA swizzle names and XYZW + * swizzle names. + */ + $$.xyzw_valid = 1; + $$.rgba_valid = 1; + } + | string + { + char s; + + if (strlen($1) > 1) { + yyerror(& @1, state, "invalid extended swizzle selector"); + YYERROR; + } + + s = $1[0]; + free($1); + + switch (s) { + case 'x': + $$.swz = SWIZZLE_X; + $$.xyzw_valid = 1; + break; + case 'y': + $$.swz = SWIZZLE_Y; + $$.xyzw_valid = 1; + break; + case 'z': + $$.swz = SWIZZLE_Z; + $$.xyzw_valid = 1; + break; + case 'w': + $$.swz = SWIZZLE_W; + $$.xyzw_valid = 1; + break; + + case 'r': + $$.swz = SWIZZLE_X; + $$.rgba_valid = 1; + break; + case 'g': + $$.swz = SWIZZLE_Y; + $$.rgba_valid = 1; + break; + case 'b': + $$.swz = SWIZZLE_Z; + $$.rgba_valid = 1; + break; + case 'a': + $$.swz = SWIZZLE_W; + $$.rgba_valid = 1; + break; + + default: + yyerror(& @1, state, "invalid extended swizzle selector"); + YYERROR; + break; + } + } + ; + +srcReg: USED_IDENTIFIER /* temporaryReg | progParamSingle */ + { + struct asm_symbol *const s = (struct asm_symbol *) + _mesa_symbol_table_find_symbol(state->st, 0, $1); + + free($1); + + if (s == NULL) { + yyerror(& @1, state, "invalid operand variable"); + YYERROR; + } else if ((s->type != at_param) && (s->type != at_temp) + && (s->type != at_attrib)) { + yyerror(& @1, state, "invalid operand variable"); + YYERROR; + } else if ((s->type == at_param) && s->param_is_array) { + yyerror(& @1, state, "non-array access to array PARAM"); + YYERROR; + } + + init_src_reg(& $$); + switch (s->type) { + case at_temp: + set_src_reg(& $$, PROGRAM_TEMPORARY, s->temp_binding); + break; + case at_param: + set_src_reg_swz(& $$, s->param_binding_type, + s->param_binding_begin, + s->param_binding_swizzle); + break; + case at_attrib: + set_src_reg(& $$, PROGRAM_INPUT, s->attrib_binding); + state->prog->InputsRead |= (1U << $$.Base.Index); + + if (!validate_inputs(& @1, state)) { + YYERROR; + } + break; + + default: + YYERROR; + break; + } + } + | attribBinding + { + set_src_reg(& $$, PROGRAM_INPUT, $1); + state->prog->InputsRead |= (1U << $$.Base.Index); + + if (!validate_inputs(& @1, state)) { + YYERROR; + } + } + | progParamArray '[' progParamArrayMem ']' + { + if (! $3.Base.RelAddr + && ((unsigned) $3.Base.Index >= $1->param_binding_length)) { + yyerror(& @3, state, "out of bounds array access"); + YYERROR; + } + + init_src_reg(& $$); + $$.Base.File = $1->param_binding_type; + + if ($3.Base.RelAddr) { + state->prog->IndirectRegisterFiles |= (1 << $$.Base.File); + $1->param_accessed_indirectly = 1; + + $$.Base.RelAddr = 1; + $$.Base.Index = $3.Base.Index; + $$.Symbol = $1; + } else { + $$.Base.Index = $1->param_binding_begin + $3.Base.Index; + } + } + | paramSingleItemUse + { + gl_register_file file = ($1.name != NULL) + ? $1.param_binding_type + : PROGRAM_CONSTANT; + set_src_reg_swz(& $$, file, $1.param_binding_begin, + $1.param_binding_swizzle); + } + ; + +dstReg: resultBinding + { + set_dst_reg(& $$, PROGRAM_OUTPUT, $1); + } + | USED_IDENTIFIER /* temporaryReg | vertexResultReg */ + { + struct asm_symbol *const s = (struct asm_symbol *) + _mesa_symbol_table_find_symbol(state->st, 0, $1); + + free($1); + + if (s == NULL) { + yyerror(& @1, state, "invalid operand variable"); + YYERROR; + } else if ((s->type != at_output) && (s->type != at_temp)) { + yyerror(& @1, state, "invalid operand variable"); + YYERROR; + } + + switch (s->type) { + case at_temp: + set_dst_reg(& $$, PROGRAM_TEMPORARY, s->temp_binding); + break; + case at_output: + set_dst_reg(& $$, PROGRAM_OUTPUT, s->output_binding); + break; + default: + set_dst_reg(& $$, s->param_binding_type, s->param_binding_begin); + break; + } + } + ; + +progParamArray: USED_IDENTIFIER + { + struct asm_symbol *const s = (struct asm_symbol *) + _mesa_symbol_table_find_symbol(state->st, 0, $1); + + free($1); + + if (s == NULL) { + yyerror(& @1, state, "invalid operand variable"); + YYERROR; + } else if ((s->type != at_param) || !s->param_is_array) { + yyerror(& @1, state, "array access to non-PARAM variable"); + YYERROR; + } else { + $$ = s; + } + } + ; + +progParamArrayMem: progParamArrayAbs | progParamArrayRel; + +progParamArrayAbs: INTEGER + { + init_src_reg(& $$); + $$.Base.Index = $1; + } + ; + +progParamArrayRel: addrReg addrComponent addrRegRelOffset + { + /* FINISHME: Add support for multiple address registers. + */ + /* FINISHME: Add support for 4-component address registers. + */ + init_src_reg(& $$); + $$.Base.RelAddr = 1; + $$.Base.Index = $3; + } + ; + +addrRegRelOffset: { $$ = 0; } + | '+' addrRegPosOffset { $$ = $2; } + | '-' addrRegNegOffset { $$ = -$2; } + ; + +addrRegPosOffset: INTEGER + { + if (($1 < 0) || ($1 > 4095)) { + char s[100]; + _mesa_snprintf(s, sizeof(s), + "relative address offset too large (%d)", $1); + yyerror(& @1, state, s); + YYERROR; + } else { + $$ = $1; + } + } + ; + +addrRegNegOffset: INTEGER + { + if (($1 < 0) || ($1 > 4096)) { + char s[100]; + _mesa_snprintf(s, sizeof(s), + "relative address offset too large (%d)", $1); + yyerror(& @1, state, s); + YYERROR; + } else { + $$ = $1; + } + } + ; + +addrReg: USED_IDENTIFIER + { + struct asm_symbol *const s = (struct asm_symbol *) + _mesa_symbol_table_find_symbol(state->st, 0, $1); + + free($1); + + if (s == NULL) { + yyerror(& @1, state, "invalid array member"); + YYERROR; + } else if (s->type != at_address) { + yyerror(& @1, state, + "invalid variable for indexed array access"); + YYERROR; + } else { + $$ = s; + } + } + ; + +addrComponent: MASK1 + { + if ($1.mask != WRITEMASK_X) { + yyerror(& @1, state, "invalid address component selector"); + YYERROR; + } else { + $$ = $1; + } + } + ; + +addrWriteMask: MASK1 + { + if ($1.mask != WRITEMASK_X) { + yyerror(& @1, state, + "address register write mask must be \".x\""); + YYERROR; + } else { + $$ = $1; + } + } + ; + +scalarSuffix: MASK1; + +swizzleSuffix: MASK1 + | MASK4 + | SWIZZLE + | { $$.swizzle = SWIZZLE_NOOP; $$.mask = WRITEMASK_XYZW; } + ; + +optionalMask: MASK4 | MASK3 | MASK2 | MASK1 + | { $$.swizzle = SWIZZLE_NOOP; $$.mask = WRITEMASK_XYZW; } + ; + +optionalCcMask: '(' ccTest ')' + { + $$ = $2; + } + | '(' ccTest2 ')' + { + $$ = $2; + } + | + { + $$.CondMask = COND_TR; + $$.CondSwizzle = SWIZZLE_NOOP; + $$.CondSrc = 0; + } + ; + +ccTest: ccMaskRule swizzleSuffix + { + $$ = $1; + $$.CondSwizzle = $2.swizzle; + } + ; + +ccTest2: ccMaskRule2 swizzleSuffix + { + $$ = $1; + $$.CondSwizzle = $2.swizzle; + } + ; + +ccMaskRule: IDENTIFIER + { + const int cond = _mesa_parse_cc($1); + if ((cond == 0) || ($1[2] != '\0')) { + char *const err_str = + make_error_string("invalid condition code \"%s\"", $1); + + yyerror(& @1, state, (err_str != NULL) + ? err_str : "invalid condition code"); + + if (err_str != NULL) { + free(err_str); + } + + YYERROR; + } + + $$.CondMask = cond; + $$.CondSwizzle = SWIZZLE_NOOP; + $$.CondSrc = 0; + } + ; + +ccMaskRule2: USED_IDENTIFIER + { + const int cond = _mesa_parse_cc($1); + if ((cond == 0) || ($1[2] != '\0')) { + char *const err_str = + make_error_string("invalid condition code \"%s\"", $1); + + yyerror(& @1, state, (err_str != NULL) + ? err_str : "invalid condition code"); + + if (err_str != NULL) { + free(err_str); + } + + YYERROR; + } + + $$.CondMask = cond; + $$.CondSwizzle = SWIZZLE_NOOP; + $$.CondSrc = 0; + } + ; + +namingStatement: ATTRIB_statement + | PARAM_statement + | TEMP_statement + | ADDRESS_statement + | OUTPUT_statement + | ALIAS_statement + ; + +ATTRIB_statement: ATTRIB IDENTIFIER '=' attribBinding + { + struct asm_symbol *const s = + declare_variable(state, $2, at_attrib, & @2); + + if (s == NULL) { + free($2); + YYERROR; + } else { + s->attrib_binding = $4; + state->InputsBound |= (1U << s->attrib_binding); + + if (!validate_inputs(& @4, state)) { + YYERROR; + } + } + } + ; + +attribBinding: VERTEX vtxAttribItem + { + $$ = $2; + } + | FRAGMENT fragAttribItem + { + $$ = $2; + } + ; + +vtxAttribItem: POSITION + { + $$ = VERT_ATTRIB_POS; + } + | WEIGHT vtxOptWeightNum + { + $$ = VERT_ATTRIB_WEIGHT; + } + | NORMAL + { + $$ = VERT_ATTRIB_NORMAL; + } + | COLOR optColorType + { + if (!state->ctx->Extensions.EXT_secondary_color) { + yyerror(& @2, state, "GL_EXT_secondary_color not supported"); + YYERROR; + } + + $$ = VERT_ATTRIB_COLOR0 + $2; + } + | FOGCOORD + { + if (!state->ctx->Extensions.EXT_fog_coord) { + yyerror(& @1, state, "GL_EXT_fog_coord not supported"); + YYERROR; + } + + $$ = VERT_ATTRIB_FOG; + } + | TEXCOORD optTexCoordUnitNum + { + $$ = VERT_ATTRIB_TEX0 + $2; + } + | MATRIXINDEX '[' vtxWeightNum ']' + { + yyerror(& @1, state, "GL_ARB_matrix_palette not supported"); + YYERROR; + } + | VTXATTRIB '[' vtxAttribNum ']' + { + $$ = VERT_ATTRIB_GENERIC0 + $3; + } + ; + +vtxAttribNum: INTEGER + { + if ((unsigned) $1 >= state->limits->MaxAttribs) { + yyerror(& @1, state, "invalid vertex attribute reference"); + YYERROR; + } + + $$ = $1; + } + ; + +vtxOptWeightNum: | '[' vtxWeightNum ']'; +vtxWeightNum: INTEGER; + +fragAttribItem: POSITION + { + $$ = FRAG_ATTRIB_WPOS; + } + | COLOR optColorType + { + $$ = FRAG_ATTRIB_COL0 + $2; + } + | FOGCOORD + { + $$ = FRAG_ATTRIB_FOGC; + } + | TEXCOORD optTexCoordUnitNum + { + $$ = FRAG_ATTRIB_TEX0 + $2; + } + ; + +PARAM_statement: PARAM_singleStmt | PARAM_multipleStmt; + +PARAM_singleStmt: PARAM IDENTIFIER paramSingleInit + { + struct asm_symbol *const s = + declare_variable(state, $2, at_param, & @2); + + if (s == NULL) { + free($2); + YYERROR; + } else { + s->param_binding_type = $3.param_binding_type; + s->param_binding_begin = $3.param_binding_begin; + s->param_binding_length = $3.param_binding_length; + s->param_binding_swizzle = $3.param_binding_swizzle; + s->param_is_array = 0; + } + } + ; + +PARAM_multipleStmt: PARAM IDENTIFIER '[' optArraySize ']' paramMultipleInit + { + if (($4 != 0) && ((unsigned) $4 != $6.param_binding_length)) { + free($2); + yyerror(& @4, state, + "parameter array size and number of bindings must match"); + YYERROR; + } else { + struct asm_symbol *const s = + declare_variable(state, $2, $6.type, & @2); + + if (s == NULL) { + free($2); + YYERROR; + } else { + s->param_binding_type = $6.param_binding_type; + s->param_binding_begin = $6.param_binding_begin; + s->param_binding_length = $6.param_binding_length; + s->param_binding_swizzle = SWIZZLE_XYZW; + s->param_is_array = 1; + } + } + } + ; + +optArraySize: + { + $$ = 0; + } + | INTEGER + { + if (($1 < 1) || ((unsigned) $1 > state->limits->MaxParameters)) { + yyerror(& @1, state, "invalid parameter array size"); + YYERROR; + } else { + $$ = $1; + } + } + ; + +paramSingleInit: '=' paramSingleItemDecl + { + $$ = $2; + } + ; + +paramMultipleInit: '=' '{' paramMultInitList '}' + { + $$ = $3; + } + ; + +paramMultInitList: paramMultipleItem + | paramMultInitList ',' paramMultipleItem + { + $1.param_binding_length += $3.param_binding_length; + $$ = $1; + } + ; + +paramSingleItemDecl: stateSingleItem + { + memset(& $$, 0, sizeof($$)); + $$.param_binding_begin = ~0; + initialize_symbol_from_state(state->prog, & $$, $1); + } + | programSingleItem + { + memset(& $$, 0, sizeof($$)); + $$.param_binding_begin = ~0; + initialize_symbol_from_param(state->prog, & $$, $1); + } + | paramConstDecl + { + memset(& $$, 0, sizeof($$)); + $$.param_binding_begin = ~0; + initialize_symbol_from_const(state->prog, & $$, & $1, GL_TRUE); + } + ; + +paramSingleItemUse: stateSingleItem + { + memset(& $$, 0, sizeof($$)); + $$.param_binding_begin = ~0; + initialize_symbol_from_state(state->prog, & $$, $1); + } + | programSingleItem + { + memset(& $$, 0, sizeof($$)); + $$.param_binding_begin = ~0; + initialize_symbol_from_param(state->prog, & $$, $1); + } + | paramConstUse + { + memset(& $$, 0, sizeof($$)); + $$.param_binding_begin = ~0; + initialize_symbol_from_const(state->prog, & $$, & $1, GL_TRUE); + } + ; + +paramMultipleItem: stateMultipleItem + { + memset(& $$, 0, sizeof($$)); + $$.param_binding_begin = ~0; + initialize_symbol_from_state(state->prog, & $$, $1); + } + | programMultipleItem + { + memset(& $$, 0, sizeof($$)); + $$.param_binding_begin = ~0; + initialize_symbol_from_param(state->prog, & $$, $1); + } + | paramConstDecl + { + memset(& $$, 0, sizeof($$)); + $$.param_binding_begin = ~0; + initialize_symbol_from_const(state->prog, & $$, & $1, GL_FALSE); + } + ; + +stateMultipleItem: stateSingleItem { memcpy($$, $1, sizeof($$)); } + | STATE stateMatrixRows { memcpy($$, $2, sizeof($$)); } + ; + +stateSingleItem: STATE stateMaterialItem { memcpy($$, $2, sizeof($$)); } + | STATE stateLightItem { memcpy($$, $2, sizeof($$)); } + | STATE stateLightModelItem { memcpy($$, $2, sizeof($$)); } + | STATE stateLightProdItem { memcpy($$, $2, sizeof($$)); } + | STATE stateTexGenItem { memcpy($$, $2, sizeof($$)); } + | STATE stateTexEnvItem { memcpy($$, $2, sizeof($$)); } + | STATE stateFogItem { memcpy($$, $2, sizeof($$)); } + | STATE stateClipPlaneItem { memcpy($$, $2, sizeof($$)); } + | STATE statePointItem { memcpy($$, $2, sizeof($$)); } + | STATE stateMatrixRow { memcpy($$, $2, sizeof($$)); } + | STATE stateDepthItem { memcpy($$, $2, sizeof($$)); } + ; + +stateMaterialItem: MATERIAL optFaceType stateMatProperty + { + memset($$, 0, sizeof($$)); + $$[0] = STATE_MATERIAL; + $$[1] = $2; + $$[2] = $3; + } + ; + +stateMatProperty: ambDiffSpecProperty + { + $$ = $1; + } + | EMISSION + { + $$ = STATE_EMISSION; + } + | SHININESS + { + $$ = STATE_SHININESS; + } + ; + +stateLightItem: LIGHT '[' stateLightNumber ']' stateLightProperty + { + memset($$, 0, sizeof($$)); + $$[0] = STATE_LIGHT; + $$[1] = $3; + $$[2] = $5; + } + ; + +stateLightProperty: ambDiffSpecProperty + { + $$ = $1; + } + | POSITION + { + $$ = STATE_POSITION; + } + | ATTENUATION + { + if (!state->ctx->Extensions.EXT_point_parameters) { + yyerror(& @1, state, "GL_ARB_point_parameters not supported"); + YYERROR; + } + + $$ = STATE_ATTENUATION; + } + | SPOT stateSpotProperty + { + $$ = $2; + } + | HALF + { + $$ = STATE_HALF_VECTOR; + } + ; + +stateSpotProperty: DIRECTION + { + $$ = STATE_SPOT_DIRECTION; + } + ; + +stateLightModelItem: LIGHTMODEL stateLModProperty + { + $$[0] = $2[0]; + $$[1] = $2[1]; + } + ; + +stateLModProperty: AMBIENT + { + memset($$, 0, sizeof($$)); + $$[0] = STATE_LIGHTMODEL_AMBIENT; + } + | optFaceType SCENECOLOR + { + memset($$, 0, sizeof($$)); + $$[0] = STATE_LIGHTMODEL_SCENECOLOR; + $$[1] = $1; + } + ; + +stateLightProdItem: LIGHTPROD '[' stateLightNumber ']' optFaceType stateLProdProperty + { + memset($$, 0, sizeof($$)); + $$[0] = STATE_LIGHTPROD; + $$[1] = $3; + $$[2] = $5; + $$[3] = $6; + } + ; + +stateLProdProperty: ambDiffSpecProperty; + +stateTexEnvItem: TEXENV optLegacyTexUnitNum stateTexEnvProperty + { + memset($$, 0, sizeof($$)); + $$[0] = $3; + $$[1] = $2; + } + ; + +stateTexEnvProperty: COLOR + { + $$ = STATE_TEXENV_COLOR; + } + ; + +ambDiffSpecProperty: AMBIENT + { + $$ = STATE_AMBIENT; + } + | DIFFUSE + { + $$ = STATE_DIFFUSE; + } + | SPECULAR + { + $$ = STATE_SPECULAR; + } + ; + +stateLightNumber: INTEGER + { + if ((unsigned) $1 >= state->MaxLights) { + yyerror(& @1, state, "invalid light selector"); + YYERROR; + } + + $$ = $1; + } + ; + +stateTexGenItem: TEXGEN optTexCoordUnitNum stateTexGenType stateTexGenCoord + { + memset($$, 0, sizeof($$)); + $$[0] = STATE_TEXGEN; + $$[1] = $2; + $$[2] = $3 + $4; + } + ; + +stateTexGenType: EYE + { + $$ = STATE_TEXGEN_EYE_S; + } + | OBJECT + { + $$ = STATE_TEXGEN_OBJECT_S; + } + ; +stateTexGenCoord: TEXGEN_S + { + $$ = STATE_TEXGEN_EYE_S - STATE_TEXGEN_EYE_S; + } + | TEXGEN_T + { + $$ = STATE_TEXGEN_EYE_T - STATE_TEXGEN_EYE_S; + } + | TEXGEN_R + { + $$ = STATE_TEXGEN_EYE_R - STATE_TEXGEN_EYE_S; + } + | TEXGEN_Q + { + $$ = STATE_TEXGEN_EYE_Q - STATE_TEXGEN_EYE_S; + } + ; + +stateFogItem: FOG stateFogProperty + { + memset($$, 0, sizeof($$)); + $$[0] = $2; + } + ; + +stateFogProperty: COLOR + { + $$ = STATE_FOG_COLOR; + } + | PARAMS + { + $$ = STATE_FOG_PARAMS; + } + ; + +stateClipPlaneItem: CLIP '[' stateClipPlaneNum ']' PLANE + { + memset($$, 0, sizeof($$)); + $$[0] = STATE_CLIPPLANE; + $$[1] = $3; + } + ; + +stateClipPlaneNum: INTEGER + { + if ((unsigned) $1 >= state->MaxClipPlanes) { + yyerror(& @1, state, "invalid clip plane selector"); + YYERROR; + } + + $$ = $1; + } + ; + +statePointItem: POINT_TOK statePointProperty + { + memset($$, 0, sizeof($$)); + $$[0] = $2; + } + ; + +statePointProperty: SIZE_TOK + { + $$ = STATE_POINT_SIZE; + } + | ATTENUATION + { + $$ = STATE_POINT_ATTENUATION; + } + ; + +stateMatrixRow: stateMatrixItem ROW '[' stateMatrixRowNum ']' + { + $$[0] = $1[0]; + $$[1] = $1[1]; + $$[2] = $4; + $$[3] = $4; + $$[4] = $1[2]; + } + ; + +stateMatrixRows: stateMatrixItem optMatrixRows + { + $$[0] = $1[0]; + $$[1] = $1[1]; + $$[2] = $2[2]; + $$[3] = $2[3]; + $$[4] = $1[2]; + } + ; + +optMatrixRows: + { + $$[2] = 0; + $$[3] = 3; + } + | ROW '[' stateMatrixRowNum DOT_DOT stateMatrixRowNum ']' + { + /* It seems logical that the matrix row range specifier would have + * to specify a range or more than one row (i.e., $5 > $3). + * However, the ARB_vertex_program spec says "a program will fail + * to load if <a> is greater than <b>." This means that $3 == $5 + * is valid. + */ + if ($3 > $5) { + yyerror(& @3, state, "invalid matrix row range"); + YYERROR; + } + + $$[2] = $3; + $$[3] = $5; + } + ; + +stateMatrixItem: MATRIX stateMatrixName stateOptMatModifier + { + $$[0] = $2[0]; + $$[1] = $2[1]; + $$[2] = $3; + } + ; + +stateOptMatModifier: + { + $$ = 0; + } + | stateMatModifier + { + $$ = $1; + } + ; + +stateMatModifier: INVERSE + { + $$ = STATE_MATRIX_INVERSE; + } + | TRANSPOSE + { + $$ = STATE_MATRIX_TRANSPOSE; + } + | INVTRANS + { + $$ = STATE_MATRIX_INVTRANS; + } + ; + +stateMatrixRowNum: INTEGER + { + if ($1 > 3) { + yyerror(& @1, state, "invalid matrix row reference"); + YYERROR; + } + + $$ = $1; + } + ; + +stateMatrixName: MODELVIEW stateOptModMatNum + { + $$[0] = STATE_MODELVIEW_MATRIX; + $$[1] = $2; + } + | PROJECTION + { + $$[0] = STATE_PROJECTION_MATRIX; + $$[1] = 0; + } + | MVP + { + $$[0] = STATE_MVP_MATRIX; + $$[1] = 0; + } + | TEXTURE optTexCoordUnitNum + { + $$[0] = STATE_TEXTURE_MATRIX; + $$[1] = $2; + } + | PALETTE '[' statePaletteMatNum ']' + { + yyerror(& @1, state, "GL_ARB_matrix_palette not supported"); + YYERROR; + } + | MAT_PROGRAM '[' stateProgramMatNum ']' + { + $$[0] = STATE_PROGRAM_MATRIX; + $$[1] = $3; + } + ; + +stateOptModMatNum: + { + $$ = 0; + } + | '[' stateModMatNum ']' + { + $$ = $2; + } + ; +stateModMatNum: INTEGER + { + /* Since GL_ARB_vertex_blend isn't supported, only modelview matrix + * zero is valid. + */ + if ($1 != 0) { + yyerror(& @1, state, "invalid modelview matrix index"); + YYERROR; + } + + $$ = $1; + } + ; +statePaletteMatNum: INTEGER + { + /* Since GL_ARB_matrix_palette isn't supported, just let any value + * through here. The error will be generated later. + */ + $$ = $1; + } + ; +stateProgramMatNum: INTEGER + { + if ((unsigned) $1 >= state->MaxProgramMatrices) { + yyerror(& @1, state, "invalid program matrix selector"); + YYERROR; + } + + $$ = $1; + } + ; + +stateDepthItem: DEPTH RANGE + { + memset($$, 0, sizeof($$)); + $$[0] = STATE_DEPTH_RANGE; + } + ; + + +programSingleItem: progEnvParam | progLocalParam; + +programMultipleItem: progEnvParams | progLocalParams; + +progEnvParams: PROGRAM ENV '[' progEnvParamNums ']' + { + memset($$, 0, sizeof($$)); + $$[0] = state->state_param_enum; + $$[1] = STATE_ENV; + $$[2] = $4[0]; + $$[3] = $4[1]; + } + ; + +progEnvParamNums: progEnvParamNum + { + $$[0] = $1; + $$[1] = $1; + } + | progEnvParamNum DOT_DOT progEnvParamNum + { + $$[0] = $1; + $$[1] = $3; + } + ; + +progEnvParam: PROGRAM ENV '[' progEnvParamNum ']' + { + memset($$, 0, sizeof($$)); + $$[0] = state->state_param_enum; + $$[1] = STATE_ENV; + $$[2] = $4; + $$[3] = $4; + } + ; + +progLocalParams: PROGRAM LOCAL '[' progLocalParamNums ']' + { + memset($$, 0, sizeof($$)); + $$[0] = state->state_param_enum; + $$[1] = STATE_LOCAL; + $$[2] = $4[0]; + $$[3] = $4[1]; + } + +progLocalParamNums: progLocalParamNum + { + $$[0] = $1; + $$[1] = $1; + } + | progLocalParamNum DOT_DOT progLocalParamNum + { + $$[0] = $1; + $$[1] = $3; + } + ; + +progLocalParam: PROGRAM LOCAL '[' progLocalParamNum ']' + { + memset($$, 0, sizeof($$)); + $$[0] = state->state_param_enum; + $$[1] = STATE_LOCAL; + $$[2] = $4; + $$[3] = $4; + } + ; + +progEnvParamNum: INTEGER + { + if ((unsigned) $1 >= state->limits->MaxEnvParams) { + yyerror(& @1, state, "invalid environment parameter reference"); + YYERROR; + } + $$ = $1; + } + ; + +progLocalParamNum: INTEGER + { + if ((unsigned) $1 >= state->limits->MaxLocalParams) { + yyerror(& @1, state, "invalid local parameter reference"); + YYERROR; + } + $$ = $1; + } + ; + + + +paramConstDecl: paramConstScalarDecl | paramConstVector; +paramConstUse: paramConstScalarUse | paramConstVector; + +paramConstScalarDecl: signedFloatConstant + { + $$.count = 4; + $$.data[0] = $1; + $$.data[1] = $1; + $$.data[2] = $1; + $$.data[3] = $1; + } + ; + +paramConstScalarUse: REAL + { + $$.count = 1; + $$.data[0] = $1; + $$.data[1] = $1; + $$.data[2] = $1; + $$.data[3] = $1; + } + | INTEGER + { + $$.count = 1; + $$.data[0] = (float) $1; + $$.data[1] = (float) $1; + $$.data[2] = (float) $1; + $$.data[3] = (float) $1; + } + ; + +paramConstVector: '{' signedFloatConstant '}' + { + $$.count = 4; + $$.data[0] = $2; + $$.data[1] = 0.0f; + $$.data[2] = 0.0f; + $$.data[3] = 1.0f; + } + | '{' signedFloatConstant ',' signedFloatConstant '}' + { + $$.count = 4; + $$.data[0] = $2; + $$.data[1] = $4; + $$.data[2] = 0.0f; + $$.data[3] = 1.0f; + } + | '{' signedFloatConstant ',' signedFloatConstant ',' + signedFloatConstant '}' + { + $$.count = 4; + $$.data[0] = $2; + $$.data[1] = $4; + $$.data[2] = $6; + $$.data[3] = 1.0f; + } + | '{' signedFloatConstant ',' signedFloatConstant ',' + signedFloatConstant ',' signedFloatConstant '}' + { + $$.count = 4; + $$.data[0] = $2; + $$.data[1] = $4; + $$.data[2] = $6; + $$.data[3] = $8; + } + ; + +signedFloatConstant: optionalSign REAL + { + $$ = ($1) ? -$2 : $2; + } + | optionalSign INTEGER + { + $$ = (float)(($1) ? -$2 : $2); + } + ; + +optionalSign: '+' { $$ = FALSE; } + | '-' { $$ = TRUE; } + | { $$ = FALSE; } + ; + +TEMP_statement: optVarSize TEMP { $<integer>$ = $2; } varNameList + ; + +optVarSize: string + { + /* NV_fragment_program_option defines the size qualifiers in a + * fairly broken way. "SHORT" or "LONG" can optionally be used + * before TEMP or OUTPUT. However, neither is a reserved word! + * This means that we have to parse it as an identifier, then check + * to make sure it's one of the valid values. *sigh* + * + * In addition, the grammar in the extension spec does *not* allow + * the size specifier to be optional, but all known implementations + * do. + */ + if (!state->option.NV_fragment) { + yyerror(& @1, state, "unexpected IDENTIFIER"); + YYERROR; + } + + if (strcmp("SHORT", $1) == 0) { + } else if (strcmp("LONG", $1) == 0) { + } else { + char *const err_str = + make_error_string("invalid storage size specifier \"%s\"", + $1); + + yyerror(& @1, state, (err_str != NULL) + ? err_str : "invalid storage size specifier"); + + if (err_str != NULL) { + free(err_str); + } + + YYERROR; + } + } + | + { + } + ; + +ADDRESS_statement: ADDRESS { $<integer>$ = $1; } varNameList + ; + +varNameList: varNameList ',' IDENTIFIER + { + if (!declare_variable(state, $3, $<integer>0, & @3)) { + free($3); + YYERROR; + } + } + | IDENTIFIER + { + if (!declare_variable(state, $1, $<integer>0, & @1)) { + free($1); + YYERROR; + } + } + ; + +OUTPUT_statement: optVarSize OUTPUT IDENTIFIER '=' resultBinding + { + struct asm_symbol *const s = + declare_variable(state, $3, at_output, & @3); + + if (s == NULL) { + free($3); + YYERROR; + } else { + s->output_binding = $5; + } + } + ; + +resultBinding: RESULT POSITION + { + if (state->mode == ARB_vertex) { + $$ = VERT_RESULT_HPOS; + } else { + yyerror(& @2, state, "invalid program result name"); + YYERROR; + } + } + | RESULT FOGCOORD + { + if (state->mode == ARB_vertex) { + $$ = VERT_RESULT_FOGC; + } else { + yyerror(& @2, state, "invalid program result name"); + YYERROR; + } + } + | RESULT resultColBinding + { + $$ = $2; + } + | RESULT POINTSIZE + { + if (state->mode == ARB_vertex) { + $$ = VERT_RESULT_PSIZ; + } else { + yyerror(& @2, state, "invalid program result name"); + YYERROR; + } + } + | RESULT TEXCOORD optTexCoordUnitNum + { + if (state->mode == ARB_vertex) { + $$ = VERT_RESULT_TEX0 + $3; + } else { + yyerror(& @2, state, "invalid program result name"); + YYERROR; + } + } + | RESULT DEPTH + { + if (state->mode == ARB_fragment) { + $$ = FRAG_RESULT_DEPTH; + } else { + yyerror(& @2, state, "invalid program result name"); + YYERROR; + } + } + ; + +resultColBinding: COLOR optResultFaceType optResultColorType + { + $$ = $2 + $3; + } + ; + +optResultFaceType: + { + $$ = (state->mode == ARB_vertex) + ? VERT_RESULT_COL0 + : FRAG_RESULT_COLOR; + } + | FRONT + { + if (state->mode == ARB_vertex) { + $$ = VERT_RESULT_COL0; + } else { + yyerror(& @1, state, "invalid program result name"); + YYERROR; + } + } + | BACK + { + if (state->mode == ARB_vertex) { + $$ = VERT_RESULT_BFC0; + } else { + yyerror(& @1, state, "invalid program result name"); + YYERROR; + } + } + ; + +optResultColorType: + { + $$ = 0; + } + | PRIMARY + { + if (state->mode == ARB_vertex) { + $$ = 0; + } else { + yyerror(& @1, state, "invalid program result name"); + YYERROR; + } + } + | SECONDARY + { + if (state->mode == ARB_vertex) { + $$ = 1; + } else { + yyerror(& @1, state, "invalid program result name"); + YYERROR; + } + } + ; + +optFaceType: { $$ = 0; } + | FRONT { $$ = 0; } + | BACK { $$ = 1; } + ; + +optColorType: { $$ = 0; } + | PRIMARY { $$ = 0; } + | SECONDARY { $$ = 1; } + ; + +optTexCoordUnitNum: { $$ = 0; } + | '[' texCoordUnitNum ']' { $$ = $2; } + ; + +optTexImageUnitNum: { $$ = 0; } + | '[' texImageUnitNum ']' { $$ = $2; } + ; + +optLegacyTexUnitNum: { $$ = 0; } + | '[' legacyTexUnitNum ']' { $$ = $2; } + ; + +texCoordUnitNum: INTEGER + { + if ((unsigned) $1 >= state->MaxTextureCoordUnits) { + yyerror(& @1, state, "invalid texture coordinate unit selector"); + YYERROR; + } + + $$ = $1; + } + ; + +texImageUnitNum: INTEGER + { + if ((unsigned) $1 >= state->MaxTextureImageUnits) { + yyerror(& @1, state, "invalid texture image unit selector"); + YYERROR; + } + + $$ = $1; + } + ; + +legacyTexUnitNum: INTEGER + { + if ((unsigned) $1 >= state->MaxTextureUnits) { + yyerror(& @1, state, "invalid texture unit selector"); + YYERROR; + } + + $$ = $1; + } + ; + +ALIAS_statement: ALIAS IDENTIFIER '=' USED_IDENTIFIER + { + struct asm_symbol *exist = (struct asm_symbol *) + _mesa_symbol_table_find_symbol(state->st, 0, $2); + struct asm_symbol *target = (struct asm_symbol *) + _mesa_symbol_table_find_symbol(state->st, 0, $4); + + free($4); + + if (exist != NULL) { + char m[1000]; + _mesa_snprintf(m, sizeof(m), "redeclared identifier: %s", $2); + free($2); + yyerror(& @2, state, m); + YYERROR; + } else if (target == NULL) { + free($2); + yyerror(& @4, state, + "undefined variable binding in ALIAS statement"); + YYERROR; + } else { + _mesa_symbol_table_add_symbol(state->st, 0, $2, target); + } + } + ; + +string: IDENTIFIER + | USED_IDENTIFIER + ; + +%% + +void +asm_instruction_set_operands(struct asm_instruction *inst, + const struct prog_dst_register *dst, + const struct asm_src_register *src0, + const struct asm_src_register *src1, + const struct asm_src_register *src2) +{ + /* In the core ARB extensions only the KIL instruction doesn't have a + * destination register. + */ + if (dst == NULL) { + init_dst_reg(& inst->Base.DstReg); + } else { + inst->Base.DstReg = *dst; + } + + /* The only instruction that doesn't have any source registers is the + * condition-code based KIL instruction added by NV_fragment_program_option. + */ + if (src0 != NULL) { + inst->Base.SrcReg[0] = src0->Base; + inst->SrcReg[0] = *src0; + } else { + init_src_reg(& inst->SrcReg[0]); + } + + if (src1 != NULL) { + inst->Base.SrcReg[1] = src1->Base; + inst->SrcReg[1] = *src1; + } else { + init_src_reg(& inst->SrcReg[1]); + } + + if (src2 != NULL) { + inst->Base.SrcReg[2] = src2->Base; + inst->SrcReg[2] = *src2; + } else { + init_src_reg(& inst->SrcReg[2]); + } +} + + +struct asm_instruction * +asm_instruction_ctor(gl_inst_opcode op, + const struct prog_dst_register *dst, + const struct asm_src_register *src0, + const struct asm_src_register *src1, + const struct asm_src_register *src2) +{ + struct asm_instruction *inst = CALLOC_STRUCT(asm_instruction); + + if (inst) { + _mesa_init_instructions(& inst->Base, 1); + inst->Base.Opcode = op; + + asm_instruction_set_operands(inst, dst, src0, src1, src2); + } + + return inst; +} + + +struct asm_instruction * +asm_instruction_copy_ctor(const struct prog_instruction *base, + const struct prog_dst_register *dst, + const struct asm_src_register *src0, + const struct asm_src_register *src1, + const struct asm_src_register *src2) +{ + struct asm_instruction *inst = CALLOC_STRUCT(asm_instruction); + + if (inst) { + _mesa_init_instructions(& inst->Base, 1); + inst->Base.Opcode = base->Opcode; + inst->Base.CondUpdate = base->CondUpdate; + inst->Base.CondDst = base->CondDst; + inst->Base.SaturateMode = base->SaturateMode; + inst->Base.Precision = base->Precision; + + asm_instruction_set_operands(inst, dst, src0, src1, src2); + } + + return inst; +} + + +void +init_dst_reg(struct prog_dst_register *r) +{ + memset(r, 0, sizeof(*r)); + r->File = PROGRAM_UNDEFINED; + r->WriteMask = WRITEMASK_XYZW; + r->CondMask = COND_TR; + r->CondSwizzle = SWIZZLE_NOOP; +} + + +/** Like init_dst_reg() but set the File and Index fields. */ +void +set_dst_reg(struct prog_dst_register *r, gl_register_file file, GLint index) +{ + const GLint maxIndex = 1 << INST_INDEX_BITS; + const GLint minIndex = 0; + ASSERT(index >= minIndex); + (void) minIndex; + ASSERT(index <= maxIndex); + (void) maxIndex; + ASSERT(file == PROGRAM_TEMPORARY || + file == PROGRAM_ADDRESS || + file == PROGRAM_OUTPUT); + memset(r, 0, sizeof(*r)); + r->File = file; + r->Index = index; + r->WriteMask = WRITEMASK_XYZW; + r->CondMask = COND_TR; + r->CondSwizzle = SWIZZLE_NOOP; +} + + +void +init_src_reg(struct asm_src_register *r) +{ + memset(r, 0, sizeof(*r)); + r->Base.File = PROGRAM_UNDEFINED; + r->Base.Swizzle = SWIZZLE_NOOP; + r->Symbol = NULL; +} + + +/** Like init_src_reg() but set the File and Index fields. + * \return GL_TRUE if a valid src register, GL_FALSE otherwise + */ +void +set_src_reg(struct asm_src_register *r, gl_register_file file, GLint index) +{ + set_src_reg_swz(r, file, index, SWIZZLE_XYZW); +} + + +void +set_src_reg_swz(struct asm_src_register *r, gl_register_file file, GLint index, + GLuint swizzle) +{ + const GLint maxIndex = (1 << INST_INDEX_BITS) - 1; + const GLint minIndex = -(1 << INST_INDEX_BITS); + ASSERT(file < PROGRAM_FILE_MAX); + ASSERT(index >= minIndex); + (void) minIndex; + ASSERT(index <= maxIndex); + (void) maxIndex; + memset(r, 0, sizeof(*r)); + r->Base.File = file; + r->Base.Index = index; + r->Base.Swizzle = swizzle; + r->Symbol = NULL; +} + + +/** + * Validate the set of inputs used by a program + * + * Validates that legal sets of inputs are used by the program. In this case + * "used" included both reading the input or binding the input to a name using + * the \c ATTRIB command. + * + * \return + * \c TRUE if the combination of inputs used is valid, \c FALSE otherwise. + */ +int +validate_inputs(struct YYLTYPE *locp, struct asm_parser_state *state) +{ + const int inputs = state->prog->InputsRead | state->InputsBound; + + if (((inputs & 0x0ffff) & (inputs >> 16)) != 0) { + yyerror(locp, state, "illegal use of generic attribute and name attribute"); + return 0; + } + + return 1; +} + + +struct asm_symbol * +declare_variable(struct asm_parser_state *state, char *name, enum asm_type t, + struct YYLTYPE *locp) +{ + struct asm_symbol *s = NULL; + struct asm_symbol *exist = (struct asm_symbol *) + _mesa_symbol_table_find_symbol(state->st, 0, name); + + + if (exist != NULL) { + yyerror(locp, state, "redeclared identifier"); + } else { + s = calloc(1, sizeof(struct asm_symbol)); + s->name = name; + s->type = t; + + switch (t) { + case at_temp: + if (state->prog->NumTemporaries >= state->limits->MaxTemps) { + yyerror(locp, state, "too many temporaries declared"); + free(s); + return NULL; + } + + s->temp_binding = state->prog->NumTemporaries; + state->prog->NumTemporaries++; + break; + + case at_address: + if (state->prog->NumAddressRegs >= state->limits->MaxAddressRegs) { + yyerror(locp, state, "too many address registers declared"); + free(s); + return NULL; + } + + /* FINISHME: Add support for multiple address registers. + */ + state->prog->NumAddressRegs++; + break; + + default: + break; + } + + _mesa_symbol_table_add_symbol(state->st, 0, s->name, s); + s->next = state->sym; + state->sym = s; + } + + return s; +} + + +int add_state_reference(struct gl_program_parameter_list *param_list, + const gl_state_index tokens[STATE_LENGTH]) +{ + const GLuint size = 4; /* XXX fix */ + char *name; + GLint index; + + name = _mesa_program_state_string(tokens); + index = _mesa_add_parameter(param_list, PROGRAM_STATE_VAR, name, + size, GL_NONE, NULL, tokens, 0x0); + param_list->StateFlags |= _mesa_program_state_flags(tokens); + + /* free name string here since we duplicated it in add_parameter() */ + free(name); + + return index; +} + + +int +initialize_symbol_from_state(struct gl_program *prog, + struct asm_symbol *param_var, + const gl_state_index tokens[STATE_LENGTH]) +{ + int idx = -1; + gl_state_index state_tokens[STATE_LENGTH]; + + + memcpy(state_tokens, tokens, sizeof(state_tokens)); + + param_var->type = at_param; + param_var->param_binding_type = PROGRAM_STATE_VAR; + + /* If we are adding a STATE_MATRIX that has multiple rows, we need to + * unroll it and call add_state_reference() for each row + */ + if ((state_tokens[0] == STATE_MODELVIEW_MATRIX || + state_tokens[0] == STATE_PROJECTION_MATRIX || + state_tokens[0] == STATE_MVP_MATRIX || + state_tokens[0] == STATE_TEXTURE_MATRIX || + state_tokens[0] == STATE_PROGRAM_MATRIX) + && (state_tokens[2] != state_tokens[3])) { + int row; + const int first_row = state_tokens[2]; + const int last_row = state_tokens[3]; + + for (row = first_row; row <= last_row; row++) { + state_tokens[2] = state_tokens[3] = row; + + idx = add_state_reference(prog->Parameters, state_tokens); + if (param_var->param_binding_begin == ~0U) { + param_var->param_binding_begin = idx; + param_var->param_binding_swizzle = SWIZZLE_XYZW; + } + + param_var->param_binding_length++; + } + } + else { + idx = add_state_reference(prog->Parameters, state_tokens); + if (param_var->param_binding_begin == ~0U) { + param_var->param_binding_begin = idx; + param_var->param_binding_swizzle = SWIZZLE_XYZW; + } + param_var->param_binding_length++; + } + + return idx; +} + + +int +initialize_symbol_from_param(struct gl_program *prog, + struct asm_symbol *param_var, + const gl_state_index tokens[STATE_LENGTH]) +{ + int idx = -1; + gl_state_index state_tokens[STATE_LENGTH]; + + + memcpy(state_tokens, tokens, sizeof(state_tokens)); + + assert((state_tokens[0] == STATE_VERTEX_PROGRAM) + || (state_tokens[0] == STATE_FRAGMENT_PROGRAM)); + assert((state_tokens[1] == STATE_ENV) + || (state_tokens[1] == STATE_LOCAL)); + + /* + * The param type is STATE_VAR. The program parameter entry will + * effectively be a pointer into the LOCAL or ENV parameter array. + */ + param_var->type = at_param; + param_var->param_binding_type = PROGRAM_STATE_VAR; + + /* If we are adding a STATE_ENV or STATE_LOCAL that has multiple elements, + * we need to unroll it and call add_state_reference() for each row + */ + if (state_tokens[2] != state_tokens[3]) { + int row; + const int first_row = state_tokens[2]; + const int last_row = state_tokens[3]; + + for (row = first_row; row <= last_row; row++) { + state_tokens[2] = state_tokens[3] = row; + + idx = add_state_reference(prog->Parameters, state_tokens); + if (param_var->param_binding_begin == ~0U) { + param_var->param_binding_begin = idx; + param_var->param_binding_swizzle = SWIZZLE_XYZW; + } + param_var->param_binding_length++; + } + } + else { + idx = add_state_reference(prog->Parameters, state_tokens); + if (param_var->param_binding_begin == ~0U) { + param_var->param_binding_begin = idx; + param_var->param_binding_swizzle = SWIZZLE_XYZW; + } + param_var->param_binding_length++; + } + + return idx; +} + + +/** + * Put a float/vector constant/literal into the parameter list. + * \param param_var returns info about the parameter/constant's location, + * binding, type, etc. + * \param vec the vector/constant to add + * \param allowSwizzle if true, try to consolidate constants which only differ + * by a swizzle. We don't want to do this when building + * arrays of constants that may be indexed indirectly. + * \return index of the constant in the parameter list. + */ +int +initialize_symbol_from_const(struct gl_program *prog, + struct asm_symbol *param_var, + const struct asm_vector *vec, + GLboolean allowSwizzle) +{ + unsigned swizzle; + const int idx = _mesa_add_unnamed_constant(prog->Parameters, + vec->data, vec->count, + allowSwizzle ? &swizzle : NULL); + + param_var->type = at_param; + param_var->param_binding_type = PROGRAM_CONSTANT; + + if (param_var->param_binding_begin == ~0U) { + param_var->param_binding_begin = idx; + param_var->param_binding_swizzle = allowSwizzle ? swizzle : SWIZZLE_XYZW; + } + param_var->param_binding_length++; + + return idx; +} + + +char * +make_error_string(const char *fmt, ...) +{ + int length; + char *str; + va_list args; + + + /* Call vsnprintf once to determine how large the final string is. Call it + * again to do the actual formatting. from the vsnprintf manual page: + * + * Upon successful return, these functions return the number of + * characters printed (not including the trailing '\0' used to end + * output to strings). + */ + va_start(args, fmt); + length = 1 + vsnprintf(NULL, 0, fmt, args); + va_end(args); + + str = malloc(length); + if (str) { + va_start(args, fmt); + vsnprintf(str, length, fmt, args); + va_end(args); + } + + return str; +} + + +void +yyerror(YYLTYPE *locp, struct asm_parser_state *state, const char *s) +{ + char *err_str; + + + err_str = make_error_string("glProgramStringARB(%s)\n", s); + if (err_str) { + _mesa_error(state->ctx, GL_INVALID_OPERATION, "%s", err_str); + free(err_str); + } + + err_str = make_error_string("line %u, char %u: error: %s\n", + locp->first_line, locp->first_column, s); + _mesa_set_program_error(state->ctx, locp->position, err_str); + + if (err_str) { + free(err_str); + } +} + + +GLboolean +_mesa_parse_arb_program(GLcontext *ctx, GLenum target, const GLubyte *str, + GLsizei len, struct asm_parser_state *state) +{ + struct asm_instruction *inst; + unsigned i; + GLubyte *strz; + GLboolean result = GL_FALSE; + void *temp; + struct asm_symbol *sym; + + state->ctx = ctx; + state->prog->Target = target; + state->prog->Parameters = _mesa_new_parameter_list(); + + /* Make a copy of the program string and force it to be NUL-terminated. + */ + strz = (GLubyte *) malloc(len + 1); + if (strz == NULL) { + _mesa_error(ctx, GL_OUT_OF_MEMORY, "glProgramStringARB"); + return GL_FALSE; + } + memcpy (strz, str, len); + strz[len] = '\0'; + + state->prog->String = strz; + + state->st = _mesa_symbol_table_ctor(); + + state->limits = (target == GL_VERTEX_PROGRAM_ARB) + ? & ctx->Const.VertexProgram + : & ctx->Const.FragmentProgram; + + state->MaxTextureImageUnits = ctx->Const.MaxTextureImageUnits; + state->MaxTextureCoordUnits = ctx->Const.MaxTextureCoordUnits; + state->MaxTextureUnits = ctx->Const.MaxTextureUnits; + state->MaxClipPlanes = ctx->Const.MaxClipPlanes; + state->MaxLights = ctx->Const.MaxLights; + state->MaxProgramMatrices = ctx->Const.MaxProgramMatrices; + + state->state_param_enum = (target == GL_VERTEX_PROGRAM_ARB) + ? STATE_VERTEX_PROGRAM : STATE_FRAGMENT_PROGRAM; + + _mesa_set_program_error(ctx, -1, NULL); + + _mesa_program_lexer_ctor(& state->scanner, state, (const char *) str, len); + yyparse(state); + _mesa_program_lexer_dtor(state->scanner); + + + if (ctx->Program.ErrorPos != -1) { + goto error; + } + + if (! _mesa_layout_parameters(state)) { + struct YYLTYPE loc; + + loc.first_line = 0; + loc.first_column = 0; + loc.position = len; + + yyerror(& loc, state, "invalid PARAM usage"); + goto error; + } + + + + /* Add one instruction to store the "END" instruction. + */ + state->prog->Instructions = + _mesa_alloc_instructions(state->prog->NumInstructions + 1); + inst = state->inst_head; + for (i = 0; i < state->prog->NumInstructions; i++) { + struct asm_instruction *const temp = inst->next; + + state->prog->Instructions[i] = inst->Base; + inst = temp; + } + + /* Finally, tag on an OPCODE_END instruction */ + { + const GLuint numInst = state->prog->NumInstructions; + _mesa_init_instructions(state->prog->Instructions + numInst, 1); + state->prog->Instructions[numInst].Opcode = OPCODE_END; + } + state->prog->NumInstructions++; + + state->prog->NumParameters = state->prog->Parameters->NumParameters; + state->prog->NumAttributes = _mesa_bitcount(state->prog->InputsRead); + + /* + * Initialize native counts to logical counts. The device driver may + * change them if program is translated into a hardware program. + */ + state->prog->NumNativeInstructions = state->prog->NumInstructions; + state->prog->NumNativeTemporaries = state->prog->NumTemporaries; + state->prog->NumNativeParameters = state->prog->NumParameters; + state->prog->NumNativeAttributes = state->prog->NumAttributes; + state->prog->NumNativeAddressRegs = state->prog->NumAddressRegs; + + result = GL_TRUE; + +error: + for (inst = state->inst_head; inst != NULL; inst = temp) { + temp = inst->next; + free(inst); + } + + state->inst_head = NULL; + state->inst_tail = NULL; + + for (sym = state->sym; sym != NULL; sym = temp) { + temp = sym->next; + + free((void *) sym->name); + free(sym); + } + state->sym = NULL; + + _mesa_symbol_table_dtor(state->st); + state->st = NULL; + + return result; +} diff --git a/mesalib/src/mesa/program/program_parse_extra.c b/mesalib/src/mesa/program/program_parse_extra.c new file mode 100644 index 000000000..ae98b782b --- /dev/null +++ b/mesalib/src/mesa/program/program_parse_extra.c @@ -0,0 +1,255 @@ +/* + * Copyright © 2009 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. + */ + +#include <string.h> +#include "main/mtypes.h" +#include "prog_instruction.h" +#include "program_parser.h" + + +/** + * Extra assembly-level parser routines + * + * \author Ian Romanick <ian.d.romanick@intel.com> + */ + +int +_mesa_parse_instruction_suffix(const struct asm_parser_state *state, + const char *suffix, + struct prog_instruction *inst) +{ + inst->CondUpdate = 0; + inst->CondDst = 0; + inst->SaturateMode = SATURATE_OFF; + inst->Precision = FLOAT32; + + + /* The first possible suffix element is the precision specifier from + * NV_fragment_program_option. + */ + if (state->option.NV_fragment) { + switch (suffix[0]) { + case 'H': + inst->Precision = FLOAT16; + suffix++; + break; + case 'R': + inst->Precision = FLOAT32; + suffix++; + break; + case 'X': + inst->Precision = FIXED12; + suffix++; + break; + default: + break; + } + } + + /* The next possible suffix element is the condition code modifier selection + * from NV_fragment_program_option. + */ + if (state->option.NV_fragment) { + if (suffix[0] == 'C') { + inst->CondUpdate = 1; + suffix++; + } + } + + + /* The final possible suffix element is the saturation selector from + * ARB_fragment_program. + */ + if (state->mode == ARB_fragment) { + if (strcmp(suffix, "_SAT") == 0) { + inst->SaturateMode = SATURATE_ZERO_ONE; + suffix += 4; + } + } + + + /* It is an error for all of the suffix string not to be consumed. + */ + return suffix[0] == '\0'; +} + + +int +_mesa_parse_cc(const char *s) +{ + int cond = 0; + + switch (s[0]) { + case 'E': + if (s[1] == 'Q') { + cond = COND_EQ; + } + break; + + case 'F': + if (s[1] == 'L') { + cond = COND_FL; + } + break; + + case 'G': + if (s[1] == 'E') { + cond = COND_GE; + } else if (s[1] == 'T') { + cond = COND_GT; + } + break; + + case 'L': + if (s[1] == 'E') { + cond = COND_LE; + } else if (s[1] == 'T') { + cond = COND_LT; + } + break; + + case 'N': + if (s[1] == 'E') { + cond = COND_NE; + } + break; + + case 'T': + if (s[1] == 'R') { + cond = COND_TR; + } + break; + + default: + break; + } + + return ((cond == 0) || (s[2] != '\0')) ? 0 : cond; +} + + +int +_mesa_ARBvp_parse_option(struct asm_parser_state *state, const char *option) +{ + if (strcmp(option, "ARB_position_invariant") == 0) { + state->option.PositionInvariant = 1; + return 1; + } + + return 0; +} + + +int +_mesa_ARBfp_parse_option(struct asm_parser_state *state, const char *option) +{ + /* All of the options currently supported start with "ARB_". The code is + * currently structured with nested if-statements because eventually options + * that start with "NV_" will be supported. This structure will result in + * less churn when those options are added. + */ + if (strncmp(option, "ARB_", 4) == 0) { + /* Advance the pointer past the "ARB_" prefix. + */ + option += 4; + + + if (strncmp(option, "fog_", 4) == 0) { + option += 4; + + if (state->option.Fog == OPTION_NONE) { + if (strcmp(option, "exp") == 0) { + state->option.Fog = OPTION_FOG_EXP; + return 1; + } else if (strcmp(option, "exp2") == 0) { + state->option.Fog = OPTION_FOG_EXP2; + return 1; + } else if (strcmp(option, "linear") == 0) { + state->option.Fog = OPTION_FOG_LINEAR; + return 1; + } + } + + return 0; + } else if (strncmp(option, "precision_hint_", 15) == 0) { + option += 15; + + if (state->option.PrecisionHint == OPTION_NONE) { + if (strcmp(option, "nicest") == 0) { + state->option.PrecisionHint = OPTION_NICEST; + return 1; + } else if (strcmp(option, "fastest") == 0) { + state->option.PrecisionHint = OPTION_FASTEST; + return 1; + } + } + + return 0; + } else if (strcmp(option, "draw_buffers") == 0) { + /* Don't need to check extension availability because all Mesa-based + * drivers support GL_ARB_draw_buffers. + */ + state->option.DrawBuffers = 1; + return 1; + } else if (strcmp(option, "fragment_program_shadow") == 0) { + if (state->ctx->Extensions.ARB_fragment_program_shadow) { + state->option.Shadow = 1; + return 1; + } + } else if (strncmp(option, "fragment_coord_", 15) == 0) { + option += 15; + if (state->ctx->Extensions.ARB_fragment_coord_conventions) { + if (strcmp(option, "origin_upper_left") == 0) { + state->option.OriginUpperLeft = 1; + return 1; + } + else if (strcmp(option, "pixel_center_integer") == 0) { + state->option.PixelCenterInteger = 1; + return 1; + } + } + } + } else if (strncmp(option, "NV_fragment_program", 19) == 0) { + option += 19; + + /* Other NV_fragment_program strings may be supported later. + */ + if (option[0] == '\0') { + if (state->ctx->Extensions.NV_fragment_program_option) { + state->option.NV_fragment = 1; + return 1; + } + } + } else if (strncmp(option, "MESA_", 5) == 0) { + option += 5; + + if (strcmp(option, "texture_array") == 0) { + if (state->ctx->Extensions.MESA_texture_array) { + state->option.TexArray = 1; + return 1; + } + } + } + + return 0; +} diff --git a/mesalib/src/mesa/program/program_parser.h b/mesalib/src/mesa/program/program_parser.h new file mode 100644 index 000000000..be952d4b9 --- /dev/null +++ b/mesalib/src/mesa/program/program_parser.h @@ -0,0 +1,302 @@ +/* + * Copyright © 2009 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. + */ +#pragma once + +#include "main/config.h" + +#ifndef MTYPES_H +struct __GLcontextRec; +typedef struct __GLcontextRec GLcontext; +#endif + +enum asm_type { + at_none, + at_address, + at_attrib, + at_param, + at_temp, + at_output +}; + +struct asm_symbol { + struct asm_symbol *next; /**< List linkage for freeing. */ + const char *name; + enum asm_type type; + unsigned attrib_binding; + unsigned output_binding; /**< Output / result register number. */ + + /** + * One of PROGRAM_STATE_VAR, PROGRAM_LOCAL_PARAM, or PROGRAM_ENV_PARAM. + */ + unsigned param_binding_type; + + /** + * Offset into the program_parameter_list where the tokens representing our + * bound state (or constants) start. + */ + unsigned param_binding_begin; + + /** + * Constants put into the parameter list may be swizzled. This + * field contain's the symbol's swizzle. (SWIZZLE_X/Y/Z/W) + */ + unsigned param_binding_swizzle; + + /* This is how many entries in the program_parameter_list we take up + * with our state tokens or constants. Note that this is _not_ the same as + * the number of param registers we eventually use. + */ + unsigned param_binding_length; + + /** + * Index of the temp register assigned to this variable. + */ + unsigned temp_binding; + + /** + * Flag whether or not a PARAM is an array + */ + unsigned param_is_array:1; + + + /** + * Flag whether or not a PARAM array is accessed indirectly + */ + unsigned param_accessed_indirectly:1; + + + /** + * \brief Is first pass of parameter layout done with this variable? + * + * The parameter layout routine operates in two passes. This flag tracks + * whether or not the first pass has handled this variable. + * + * \sa _mesa_layout_parameters + */ + unsigned pass1_done:1; +}; + + +struct asm_vector { + unsigned count; + float data[4]; +}; + + +struct asm_swizzle_mask { + unsigned swizzle:12; + unsigned mask:4; +}; + + +struct asm_src_register { + struct prog_src_register Base; + + /** + * Symbol associated with indirect access to parameter arrays. + * + * If \c Base::RelAddr is 1, this will point to the symbol for the parameter + * that is being dereferenced. Further, \c Base::Index will be the offset + * from the address register being used. + */ + struct asm_symbol *Symbol; +}; + + +struct asm_instruction { + struct prog_instruction Base; + struct asm_instruction *next; + struct asm_src_register SrcReg[3]; +}; + + +struct asm_parser_state { + GLcontext *ctx; + struct gl_program *prog; + + /** + * Per-program target limits + */ + struct gl_program_constants *limits; + + struct _mesa_symbol_table *st; + + /** + * Linked list of symbols + * + * This list is \b only used when cleaning up compiler state and freeing + * memory. + */ + struct asm_symbol *sym; + + /** + * State for the lexer. + */ + void *scanner; + + /** + * Linked list of instructions generated during parsing. + */ + /*@{*/ + struct asm_instruction *inst_head; + struct asm_instruction *inst_tail; + /*@}*/ + + + /** + * Selected limits copied from gl_constants + * + * These are limits from the GL context, but various bits in the program + * must be validated against these values. + */ + /*@{*/ + unsigned MaxTextureCoordUnits; + unsigned MaxTextureImageUnits; + unsigned MaxTextureUnits; + unsigned MaxClipPlanes; + unsigned MaxLights; + unsigned MaxProgramMatrices; + /*@}*/ + + /** + * Value to use in state vector accessors for environment and local + * parameters + */ + unsigned state_param_enum; + + + /** + * Input attributes bound to specific names + * + * This is only needed so that errors can be properly produced when + * multiple ATTRIB statements bind illegal combinations of vertex + * attributes. + */ + unsigned InputsBound; + + enum { + invalid_mode = 0, + ARB_vertex, + ARB_fragment + } mode; + + struct { + unsigned PositionInvariant:1; + unsigned Fog:2; + unsigned PrecisionHint:2; + unsigned DrawBuffers:1; + unsigned Shadow:1; + unsigned TexRect:1; + unsigned TexArray:1; + unsigned NV_fragment:1; + unsigned OriginUpperLeft:1; + unsigned PixelCenterInteger:1; + } option; + + struct { + unsigned UsesKill:1; + } fragment; +}; + +#define OPTION_NONE 0 +#define OPTION_FOG_EXP 1 +#define OPTION_FOG_EXP2 2 +#define OPTION_FOG_LINEAR 3 +#define OPTION_NICEST 1 +#define OPTION_FASTEST 2 + +typedef struct YYLTYPE { + int first_line; + int first_column; + int last_line; + int last_column; + int position; +} YYLTYPE; + +#define YYLTYPE_IS_DECLARED 1 +#define YYLTYPE_IS_TRIVIAL 1 + + +extern GLboolean _mesa_parse_arb_program(GLcontext *ctx, GLenum target, + const GLubyte *str, GLsizei len, struct asm_parser_state *state); + + + +/* From program_lexer.l. */ +extern void _mesa_program_lexer_dtor(void *scanner); + +extern void _mesa_program_lexer_ctor(void **scanner, + struct asm_parser_state *state, const char *string, size_t len); + + +/** + *\name From program_parse_extra.c + */ +/*@{*/ + +/** + * Parses and processes an option string to an ARB vertex program + * + * \return + * Non-zero on success, zero on failure. + */ +extern int _mesa_ARBvp_parse_option(struct asm_parser_state *state, + const char *option); + +/** + * Parses and processes an option string to an ARB fragment program + * + * \return + * Non-zero on success, zero on failure. + */ +extern int _mesa_ARBfp_parse_option(struct asm_parser_state *state, + const char *option); + +/** + * Parses and processes instruction suffixes + * + * Instruction suffixes, such as \c _SAT, are processed. The relevant bits + * are set in \c inst. If suffixes are encountered that are either not known + * or not supported by the modes and options set in \c state, zero will be + * returned. + * + * \return + * Non-zero on success, zero on failure. + */ +extern int _mesa_parse_instruction_suffix(const struct asm_parser_state *state, + const char *suffix, struct prog_instruction *inst); + +/** + * Parses a condition code name + * + * The condition code names (e.g., \c LT, \c GT, \c NE) were added to assembly + * shaders with the \c GL_NV_fragment_program_option extension. This function + * converts a string representation into one of the \c COND_ macros. + * + * \return + * One of the \c COND_ macros defined in prog_instruction.h on success or zero + * on failure. + */ +extern int _mesa_parse_cc(const char *s); + +/*@}*/ diff --git a/mesalib/src/mesa/program/programopt.c b/mesalib/src/mesa/program/programopt.c new file mode 100644 index 000000000..fb2ebe633 --- /dev/null +++ b/mesalib/src/mesa/program/programopt.c @@ -0,0 +1,669 @@ +/* + * Mesa 3-D graphics library + * Version: 6.5.3 + * + * Copyright (C) 1999-2007 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. + */ + +/** + * \file programopt.c + * Vertex/Fragment program optimizations and transformations for program + * options, etc. + * + * \author Brian Paul + */ + + +#include "main/glheader.h" +#include "main/context.h" +#include "prog_parameter.h" +#include "prog_statevars.h" +#include "program.h" +#include "programopt.h" +#include "prog_instruction.h" + + +/** + * This function inserts instructions for coordinate modelview * projection + * into a vertex program. + * May be used to implement the position_invariant option. + */ +static void +_mesa_insert_mvp_dp4_code(GLcontext *ctx, struct gl_vertex_program *vprog) +{ + struct prog_instruction *newInst; + const GLuint origLen = vprog->Base.NumInstructions; + const GLuint newLen = origLen + 4; + GLuint i; + + /* + * Setup state references for the modelview/projection matrix. + * XXX we should check if these state vars are already declared. + */ + static const gl_state_index mvpState[4][STATE_LENGTH] = { + { STATE_MVP_MATRIX, 0, 0, 0, 0 }, /* state.matrix.mvp.row[0] */ + { STATE_MVP_MATRIX, 0, 1, 1, 0 }, /* state.matrix.mvp.row[1] */ + { STATE_MVP_MATRIX, 0, 2, 2, 0 }, /* state.matrix.mvp.row[2] */ + { STATE_MVP_MATRIX, 0, 3, 3, 0 }, /* state.matrix.mvp.row[3] */ + }; + GLint mvpRef[4]; + + for (i = 0; i < 4; i++) { + mvpRef[i] = _mesa_add_state_reference(vprog->Base.Parameters, + mvpState[i]); + } + + /* Alloc storage for new instructions */ + newInst = _mesa_alloc_instructions(newLen); + if (!newInst) { + _mesa_error(ctx, GL_OUT_OF_MEMORY, + "glProgramString(inserting position_invariant code)"); + return; + } + + /* + * Generated instructions: + * newInst[0] = DP4 result.position.x, mvp.row[0], vertex.position; + * newInst[1] = DP4 result.position.y, mvp.row[1], vertex.position; + * newInst[2] = DP4 result.position.z, mvp.row[2], vertex.position; + * newInst[3] = DP4 result.position.w, mvp.row[3], vertex.position; + */ + _mesa_init_instructions(newInst, 4); + for (i = 0; i < 4; i++) { + newInst[i].Opcode = OPCODE_DP4; + newInst[i].DstReg.File = PROGRAM_OUTPUT; + newInst[i].DstReg.Index = VERT_RESULT_HPOS; + newInst[i].DstReg.WriteMask = (WRITEMASK_X << i); + newInst[i].SrcReg[0].File = PROGRAM_STATE_VAR; + newInst[i].SrcReg[0].Index = mvpRef[i]; + newInst[i].SrcReg[0].Swizzle = SWIZZLE_NOOP; + newInst[i].SrcReg[1].File = PROGRAM_INPUT; + newInst[i].SrcReg[1].Index = VERT_ATTRIB_POS; + newInst[i].SrcReg[1].Swizzle = SWIZZLE_NOOP; + } + + /* Append original instructions after new instructions */ + _mesa_copy_instructions (newInst + 4, vprog->Base.Instructions, origLen); + + /* free old instructions */ + _mesa_free_instructions(vprog->Base.Instructions, origLen); + + /* install new instructions */ + vprog->Base.Instructions = newInst; + vprog->Base.NumInstructions = newLen; + vprog->Base.InputsRead |= VERT_BIT_POS; + vprog->Base.OutputsWritten |= BITFIELD64_BIT(VERT_RESULT_HPOS); +} + + +static void +_mesa_insert_mvp_mad_code(GLcontext *ctx, struct gl_vertex_program *vprog) +{ + struct prog_instruction *newInst; + const GLuint origLen = vprog->Base.NumInstructions; + const GLuint newLen = origLen + 4; + GLuint hposTemp; + GLuint i; + + /* + * Setup state references for the modelview/projection matrix. + * XXX we should check if these state vars are already declared. + */ + static const gl_state_index mvpState[4][STATE_LENGTH] = { + { STATE_MVP_MATRIX, 0, 0, 0, STATE_MATRIX_TRANSPOSE }, + { STATE_MVP_MATRIX, 0, 1, 1, STATE_MATRIX_TRANSPOSE }, + { STATE_MVP_MATRIX, 0, 2, 2, STATE_MATRIX_TRANSPOSE }, + { STATE_MVP_MATRIX, 0, 3, 3, STATE_MATRIX_TRANSPOSE }, + }; + GLint mvpRef[4]; + + for (i = 0; i < 4; i++) { + mvpRef[i] = _mesa_add_state_reference(vprog->Base.Parameters, + mvpState[i]); + } + + /* Alloc storage for new instructions */ + newInst = _mesa_alloc_instructions(newLen); + if (!newInst) { + _mesa_error(ctx, GL_OUT_OF_MEMORY, + "glProgramString(inserting position_invariant code)"); + return; + } + + /* TEMP hposTemp; */ + hposTemp = vprog->Base.NumTemporaries++; + + /* + * Generated instructions: + * emit_op2(p, OPCODE_MUL, tmp, 0, swizzle1(src,X), mat[0]); + * emit_op3(p, OPCODE_MAD, tmp, 0, swizzle1(src,Y), mat[1], tmp); + * emit_op3(p, OPCODE_MAD, tmp, 0, swizzle1(src,Z), mat[2], tmp); + * emit_op3(p, OPCODE_MAD, dest, 0, swizzle1(src,W), mat[3], tmp); + */ + _mesa_init_instructions(newInst, 4); + + newInst[0].Opcode = OPCODE_MUL; + newInst[0].DstReg.File = PROGRAM_TEMPORARY; + newInst[0].DstReg.Index = hposTemp; + newInst[0].DstReg.WriteMask = WRITEMASK_XYZW; + newInst[0].SrcReg[0].File = PROGRAM_INPUT; + newInst[0].SrcReg[0].Index = VERT_ATTRIB_POS; + newInst[0].SrcReg[0].Swizzle = SWIZZLE_XXXX; + newInst[0].SrcReg[1].File = PROGRAM_STATE_VAR; + newInst[0].SrcReg[1].Index = mvpRef[0]; + newInst[0].SrcReg[1].Swizzle = SWIZZLE_NOOP; + + for (i = 1; i <= 2; i++) { + newInst[i].Opcode = OPCODE_MAD; + newInst[i].DstReg.File = PROGRAM_TEMPORARY; + newInst[i].DstReg.Index = hposTemp; + newInst[i].DstReg.WriteMask = WRITEMASK_XYZW; + newInst[i].SrcReg[0].File = PROGRAM_INPUT; + newInst[i].SrcReg[0].Index = VERT_ATTRIB_POS; + newInst[i].SrcReg[0].Swizzle = MAKE_SWIZZLE4(i,i,i,i); + newInst[i].SrcReg[1].File = PROGRAM_STATE_VAR; + newInst[i].SrcReg[1].Index = mvpRef[i]; + newInst[i].SrcReg[1].Swizzle = SWIZZLE_NOOP; + newInst[i].SrcReg[2].File = PROGRAM_TEMPORARY; + newInst[i].SrcReg[2].Index = hposTemp; + newInst[1].SrcReg[2].Swizzle = SWIZZLE_NOOP; + } + + newInst[3].Opcode = OPCODE_MAD; + newInst[3].DstReg.File = PROGRAM_OUTPUT; + newInst[3].DstReg.Index = VERT_RESULT_HPOS; + newInst[3].DstReg.WriteMask = WRITEMASK_XYZW; + newInst[3].SrcReg[0].File = PROGRAM_INPUT; + newInst[3].SrcReg[0].Index = VERT_ATTRIB_POS; + newInst[3].SrcReg[0].Swizzle = SWIZZLE_WWWW; + newInst[3].SrcReg[1].File = PROGRAM_STATE_VAR; + newInst[3].SrcReg[1].Index = mvpRef[3]; + newInst[3].SrcReg[1].Swizzle = SWIZZLE_NOOP; + newInst[3].SrcReg[2].File = PROGRAM_TEMPORARY; + newInst[3].SrcReg[2].Index = hposTemp; + newInst[3].SrcReg[2].Swizzle = SWIZZLE_NOOP; + + + /* Append original instructions after new instructions */ + _mesa_copy_instructions (newInst + 4, vprog->Base.Instructions, origLen); + + /* free old instructions */ + _mesa_free_instructions(vprog->Base.Instructions, origLen); + + /* install new instructions */ + vprog->Base.Instructions = newInst; + vprog->Base.NumInstructions = newLen; + vprog->Base.InputsRead |= VERT_BIT_POS; + vprog->Base.OutputsWritten |= BITFIELD64_BIT(VERT_RESULT_HPOS); +} + + +void +_mesa_insert_mvp_code(GLcontext *ctx, struct gl_vertex_program *vprog) +{ + if (ctx->mvp_with_dp4) + _mesa_insert_mvp_dp4_code( ctx, vprog ); + else + _mesa_insert_mvp_mad_code( ctx, vprog ); +} + + + + + + +/** + * Append extra instructions onto the given fragment program to implement + * the fog mode specified by fprog->FogOption. + * The fragment.fogcoord input is used to compute the fog blend factor. + * + * XXX with a little work, this function could be adapted to add fog code + * to vertex programs too. + */ +void +_mesa_append_fog_code(GLcontext *ctx, struct gl_fragment_program *fprog) +{ + static const gl_state_index fogPStateOpt[STATE_LENGTH] + = { STATE_INTERNAL, STATE_FOG_PARAMS_OPTIMIZED, 0, 0, 0 }; + static const gl_state_index fogColorState[STATE_LENGTH] + = { STATE_FOG_COLOR, 0, 0, 0, 0}; + struct prog_instruction *newInst, *inst; + const GLuint origLen = fprog->Base.NumInstructions; + const GLuint newLen = origLen + 5; + GLuint i; + GLint fogPRefOpt, fogColorRef; /* state references */ + GLuint colorTemp, fogFactorTemp; /* temporary registerss */ + + if (fprog->FogOption == GL_NONE) { + _mesa_problem(ctx, "_mesa_append_fog_code() called for fragment program" + " with FogOption == GL_NONE"); + return; + } + + /* Alloc storage for new instructions */ + newInst = _mesa_alloc_instructions(newLen); + if (!newInst) { + _mesa_error(ctx, GL_OUT_OF_MEMORY, + "glProgramString(inserting fog_option code)"); + return; + } + + /* Copy orig instructions into new instruction buffer */ + _mesa_copy_instructions(newInst, fprog->Base.Instructions, origLen); + + /* PARAM fogParamsRefOpt = internal optimized fog params; */ + fogPRefOpt + = _mesa_add_state_reference(fprog->Base.Parameters, fogPStateOpt); + /* PARAM fogColorRef = state.fog.color; */ + fogColorRef + = _mesa_add_state_reference(fprog->Base.Parameters, fogColorState); + + /* TEMP colorTemp; */ + colorTemp = fprog->Base.NumTemporaries++; + /* TEMP fogFactorTemp; */ + fogFactorTemp = fprog->Base.NumTemporaries++; + + /* Scan program to find where result.color is written */ + inst = newInst; + for (i = 0; i < fprog->Base.NumInstructions; i++) { + if (inst->Opcode == OPCODE_END) + break; + if (inst->DstReg.File == PROGRAM_OUTPUT && + inst->DstReg.Index == FRAG_RESULT_COLOR) { + /* change the instruction to write to colorTemp w/ clamping */ + inst->DstReg.File = PROGRAM_TEMPORARY; + inst->DstReg.Index = colorTemp; + inst->SaturateMode = SATURATE_ZERO_ONE; + /* don't break (may be several writes to result.color) */ + } + inst++; + } + assert(inst->Opcode == OPCODE_END); /* we'll overwrite this inst */ + + _mesa_init_instructions(inst, 5); + + /* emit instructions to compute fog blending factor */ + if (fprog->FogOption == GL_LINEAR) { + /* MAD fogFactorTemp.x, fragment.fogcoord.x, fogPRefOpt.x, fogPRefOpt.y; */ + inst->Opcode = OPCODE_MAD; + inst->DstReg.File = PROGRAM_TEMPORARY; + inst->DstReg.Index = fogFactorTemp; + inst->DstReg.WriteMask = WRITEMASK_X; + inst->SrcReg[0].File = PROGRAM_INPUT; + inst->SrcReg[0].Index = FRAG_ATTRIB_FOGC; + inst->SrcReg[0].Swizzle = SWIZZLE_XXXX; + inst->SrcReg[1].File = PROGRAM_STATE_VAR; + inst->SrcReg[1].Index = fogPRefOpt; + inst->SrcReg[1].Swizzle = SWIZZLE_XXXX; + inst->SrcReg[2].File = PROGRAM_STATE_VAR; + inst->SrcReg[2].Index = fogPRefOpt; + inst->SrcReg[2].Swizzle = SWIZZLE_YYYY; + inst->SaturateMode = SATURATE_ZERO_ONE; + inst++; + } + else { + ASSERT(fprog->FogOption == GL_EXP || fprog->FogOption == GL_EXP2); + /* fogPRefOpt.z = d/ln(2), fogPRefOpt.w = d/sqrt(ln(2) */ + /* EXP: MUL fogFactorTemp.x, fogPRefOpt.z, fragment.fogcoord.x; */ + /* EXP2: MUL fogFactorTemp.x, fogPRefOpt.w, fragment.fogcoord.x; */ + inst->Opcode = OPCODE_MUL; + inst->DstReg.File = PROGRAM_TEMPORARY; + inst->DstReg.Index = fogFactorTemp; + inst->DstReg.WriteMask = WRITEMASK_X; + inst->SrcReg[0].File = PROGRAM_STATE_VAR; + inst->SrcReg[0].Index = fogPRefOpt; + inst->SrcReg[0].Swizzle + = (fprog->FogOption == GL_EXP) ? SWIZZLE_ZZZZ : SWIZZLE_WWWW; + inst->SrcReg[1].File = PROGRAM_INPUT; + inst->SrcReg[1].Index = FRAG_ATTRIB_FOGC; + inst->SrcReg[1].Swizzle = SWIZZLE_XXXX; + inst++; + if (fprog->FogOption == GL_EXP2) { + /* MUL fogFactorTemp.x, fogFactorTemp.x, fogFactorTemp.x; */ + inst->Opcode = OPCODE_MUL; + inst->DstReg.File = PROGRAM_TEMPORARY; + inst->DstReg.Index = fogFactorTemp; + inst->DstReg.WriteMask = WRITEMASK_X; + inst->SrcReg[0].File = PROGRAM_TEMPORARY; + inst->SrcReg[0].Index = fogFactorTemp; + inst->SrcReg[0].Swizzle = SWIZZLE_XXXX; + inst->SrcReg[1].File = PROGRAM_TEMPORARY; + inst->SrcReg[1].Index = fogFactorTemp; + inst->SrcReg[1].Swizzle = SWIZZLE_XXXX; + inst++; + } + /* EX2_SAT fogFactorTemp.x, -fogFactorTemp.x; */ + inst->Opcode = OPCODE_EX2; + inst->DstReg.File = PROGRAM_TEMPORARY; + inst->DstReg.Index = fogFactorTemp; + inst->DstReg.WriteMask = WRITEMASK_X; + inst->SrcReg[0].File = PROGRAM_TEMPORARY; + inst->SrcReg[0].Index = fogFactorTemp; + inst->SrcReg[0].Negate = NEGATE_XYZW; + inst->SrcReg[0].Swizzle = SWIZZLE_XXXX; + inst->SaturateMode = SATURATE_ZERO_ONE; + inst++; + } + /* LRP result.color.xyz, fogFactorTemp.xxxx, colorTemp, fogColorRef; */ + inst->Opcode = OPCODE_LRP; + inst->DstReg.File = PROGRAM_OUTPUT; + inst->DstReg.Index = FRAG_RESULT_COLOR; + inst->DstReg.WriteMask = WRITEMASK_XYZ; + inst->SrcReg[0].File = PROGRAM_TEMPORARY; + inst->SrcReg[0].Index = fogFactorTemp; + inst->SrcReg[0].Swizzle = SWIZZLE_XXXX; + inst->SrcReg[1].File = PROGRAM_TEMPORARY; + inst->SrcReg[1].Index = colorTemp; + inst->SrcReg[1].Swizzle = SWIZZLE_NOOP; + inst->SrcReg[2].File = PROGRAM_STATE_VAR; + inst->SrcReg[2].Index = fogColorRef; + inst->SrcReg[2].Swizzle = SWIZZLE_NOOP; + inst++; + /* MOV result.color.w, colorTemp.x; # copy alpha */ + inst->Opcode = OPCODE_MOV; + inst->DstReg.File = PROGRAM_OUTPUT; + inst->DstReg.Index = FRAG_RESULT_COLOR; + inst->DstReg.WriteMask = WRITEMASK_W; + inst->SrcReg[0].File = PROGRAM_TEMPORARY; + inst->SrcReg[0].Index = colorTemp; + inst->SrcReg[0].Swizzle = SWIZZLE_NOOP; + inst++; + /* END; */ + inst->Opcode = OPCODE_END; + inst++; + + /* free old instructions */ + _mesa_free_instructions(fprog->Base.Instructions, origLen); + + /* install new instructions */ + fprog->Base.Instructions = newInst; + fprog->Base.NumInstructions = inst - newInst; + fprog->Base.InputsRead |= FRAG_BIT_FOGC; + /* XXX do this? fprog->FogOption = GL_NONE; */ +} + + + +static GLboolean +is_texture_instruction(const struct prog_instruction *inst) +{ + switch (inst->Opcode) { + case OPCODE_TEX: + case OPCODE_TXB: + case OPCODE_TXD: + case OPCODE_TXL: + case OPCODE_TXP: + case OPCODE_TXP_NV: + return GL_TRUE; + default: + return GL_FALSE; + } +} + + +/** + * Count the number of texure indirections in the given program. + * The program's NumTexIndirections field will be updated. + * See the GL_ARB_fragment_program spec (issue 24) for details. + * XXX we count texture indirections in texenvprogram.c (maybe use this code + * instead and elsewhere). + */ +void +_mesa_count_texture_indirections(struct gl_program *prog) +{ + GLuint indirections = 1; + GLbitfield tempsOutput = 0x0; + GLbitfield aluTemps = 0x0; + GLuint i; + + for (i = 0; i < prog->NumInstructions; i++) { + const struct prog_instruction *inst = prog->Instructions + i; + + if (is_texture_instruction(inst)) { + if (((inst->SrcReg[0].File == PROGRAM_TEMPORARY) && + (tempsOutput & (1 << inst->SrcReg[0].Index))) || + ((inst->Opcode != OPCODE_KIL) && + (inst->DstReg.File == PROGRAM_TEMPORARY) && + (aluTemps & (1 << inst->DstReg.Index)))) + { + indirections++; + tempsOutput = 0x0; + aluTemps = 0x0; + } + } + else { + GLuint j; + for (j = 0; j < 3; j++) { + if (inst->SrcReg[j].File == PROGRAM_TEMPORARY) + aluTemps |= (1 << inst->SrcReg[j].Index); + } + if (inst->DstReg.File == PROGRAM_TEMPORARY) + aluTemps |= (1 << inst->DstReg.Index); + } + + if ((inst->Opcode != OPCODE_KIL) && (inst->DstReg.File == PROGRAM_TEMPORARY)) + tempsOutput |= (1 << inst->DstReg.Index); + } + + prog->NumTexIndirections = indirections; +} + + +/** + * Count number of texture instructions in given program and update the + * program's NumTexInstructions field. + */ +void +_mesa_count_texture_instructions(struct gl_program *prog) +{ + GLuint i; + prog->NumTexInstructions = 0; + for (i = 0; i < prog->NumInstructions; i++) { + prog->NumTexInstructions += is_texture_instruction(prog->Instructions + i); + } +} + + +/** + * Scan/rewrite program to remove reads of custom (output) registers. + * The passed type has to be either PROGRAM_OUTPUT or PROGRAM_VARYING + * (for vertex shaders). + * In GLSL shaders, varying vars can be read and written. + * On some hardware, trying to read an output register causes trouble. + * So, rewrite the program to use a temporary register in this case. + */ +void +_mesa_remove_output_reads(struct gl_program *prog, gl_register_file type) +{ + GLuint i; + GLint outputMap[VERT_RESULT_MAX]; + GLuint numVaryingReads = 0; + GLboolean usedTemps[MAX_PROGRAM_TEMPS]; + GLuint firstTemp = 0; + + _mesa_find_used_registers(prog, PROGRAM_TEMPORARY, + usedTemps, MAX_PROGRAM_TEMPS); + + assert(type == PROGRAM_VARYING || type == PROGRAM_OUTPUT); + assert(prog->Target == GL_VERTEX_PROGRAM_ARB || type != PROGRAM_VARYING); + + for (i = 0; i < VERT_RESULT_MAX; i++) + outputMap[i] = -1; + + /* look for instructions which read from varying vars */ + for (i = 0; i < prog->NumInstructions; i++) { + struct prog_instruction *inst = prog->Instructions + i; + const GLuint numSrc = _mesa_num_inst_src_regs(inst->Opcode); + GLuint j; + for (j = 0; j < numSrc; j++) { + if (inst->SrcReg[j].File == type) { + /* replace the read with a temp reg */ + const GLuint var = inst->SrcReg[j].Index; + if (outputMap[var] == -1) { + numVaryingReads++; + outputMap[var] = _mesa_find_free_register(usedTemps, + MAX_PROGRAM_TEMPS, + firstTemp); + firstTemp = outputMap[var] + 1; + } + inst->SrcReg[j].File = PROGRAM_TEMPORARY; + inst->SrcReg[j].Index = outputMap[var]; + } + } + } + + if (numVaryingReads == 0) + return; /* nothing to be done */ + + /* look for instructions which write to the varying vars identified above */ + for (i = 0; i < prog->NumInstructions; i++) { + struct prog_instruction *inst = prog->Instructions + i; + if (inst->DstReg.File == type && + outputMap[inst->DstReg.Index] >= 0) { + /* change inst to write to the temp reg, instead of the varying */ + inst->DstReg.File = PROGRAM_TEMPORARY; + inst->DstReg.Index = outputMap[inst->DstReg.Index]; + } + } + + /* insert new instructions to copy the temp vars to the varying vars */ + { + struct prog_instruction *inst; + GLint endPos, var; + + /* Look for END instruction and insert the new varying writes */ + endPos = -1; + for (i = 0; i < prog->NumInstructions; i++) { + struct prog_instruction *inst = prog->Instructions + i; + if (inst->Opcode == OPCODE_END) { + endPos = i; + _mesa_insert_instructions(prog, i, numVaryingReads); + break; + } + } + + assert(endPos >= 0); + + /* insert new MOV instructions here */ + inst = prog->Instructions + endPos; + for (var = 0; var < VERT_RESULT_MAX; var++) { + if (outputMap[var] >= 0) { + /* MOV VAR[var], TEMP[tmp]; */ + inst->Opcode = OPCODE_MOV; + inst->DstReg.File = type; + inst->DstReg.Index = var; + inst->SrcReg[0].File = PROGRAM_TEMPORARY; + inst->SrcReg[0].Index = outputMap[var]; + inst++; + } + } + } +} + + +/** + * Make the given fragment program into a "no-op" shader. + * Actually, just copy the incoming fragment color (or texcoord) + * to the output color. + * This is for debug/test purposes. + */ +void +_mesa_nop_fragment_program(GLcontext *ctx, struct gl_fragment_program *prog) +{ + struct prog_instruction *inst; + GLuint inputAttr; + + inst = _mesa_alloc_instructions(2); + if (!inst) { + _mesa_error(ctx, GL_OUT_OF_MEMORY, "_mesa_nop_fragment_program"); + return; + } + + _mesa_init_instructions(inst, 2); + + inst[0].Opcode = OPCODE_MOV; + inst[0].DstReg.File = PROGRAM_OUTPUT; + inst[0].DstReg.Index = FRAG_RESULT_COLOR; + inst[0].SrcReg[0].File = PROGRAM_INPUT; + if (prog->Base.InputsRead & FRAG_BIT_COL0) + inputAttr = FRAG_ATTRIB_COL0; + else + inputAttr = FRAG_ATTRIB_TEX0; + inst[0].SrcReg[0].Index = inputAttr; + + inst[1].Opcode = OPCODE_END; + + _mesa_free_instructions(prog->Base.Instructions, + prog->Base.NumInstructions); + + prog->Base.Instructions = inst; + prog->Base.NumInstructions = 2; + prog->Base.InputsRead = 1 << inputAttr; + prog->Base.OutputsWritten = BITFIELD64_BIT(FRAG_RESULT_COLOR); +} + + +/** + * \sa _mesa_nop_fragment_program + * Replace the given vertex program with a "no-op" program that just + * transforms vertex position and emits color. + */ +void +_mesa_nop_vertex_program(GLcontext *ctx, struct gl_vertex_program *prog) +{ + struct prog_instruction *inst; + GLuint inputAttr; + + /* + * Start with a simple vertex program that emits color. + */ + inst = _mesa_alloc_instructions(2); + if (!inst) { + _mesa_error(ctx, GL_OUT_OF_MEMORY, "_mesa_nop_vertex_program"); + return; + } + + _mesa_init_instructions(inst, 2); + + inst[0].Opcode = OPCODE_MOV; + inst[0].DstReg.File = PROGRAM_OUTPUT; + inst[0].DstReg.Index = VERT_RESULT_COL0; + inst[0].SrcReg[0].File = PROGRAM_INPUT; + if (prog->Base.InputsRead & VERT_BIT_COLOR0) + inputAttr = VERT_ATTRIB_COLOR0; + else + inputAttr = VERT_ATTRIB_TEX0; + inst[0].SrcReg[0].Index = inputAttr; + + inst[1].Opcode = OPCODE_END; + + _mesa_free_instructions(prog->Base.Instructions, + prog->Base.NumInstructions); + + prog->Base.Instructions = inst; + prog->Base.NumInstructions = 2; + prog->Base.InputsRead = 1 << inputAttr; + prog->Base.OutputsWritten = BITFIELD64_BIT(VERT_RESULT_COL0); + + /* + * Now insert code to do standard modelview/projection transformation. + */ + _mesa_insert_mvp_code(ctx, prog); +} diff --git a/mesalib/src/mesa/program/programopt.h b/mesalib/src/mesa/program/programopt.h new file mode 100644 index 000000000..4af6357f9 --- /dev/null +++ b/mesalib/src/mesa/program/programopt.h @@ -0,0 +1,53 @@ +/* + * Mesa 3-D graphics library + * Version: 6.5.3 + * + * Copyright (C) 1999-2007 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 PROGRAMOPT_H +#define PROGRAMOPT_H 1 + +#include "main/mtypes.h" + +extern void +_mesa_insert_mvp_code(GLcontext *ctx, struct gl_vertex_program *vprog); + +extern void +_mesa_append_fog_code(GLcontext *ctx, struct gl_fragment_program *fprog); + +extern void +_mesa_count_texture_indirections(struct gl_program *prog); + +extern void +_mesa_count_texture_instructions(struct gl_program *prog); + +extern void +_mesa_remove_output_reads(struct gl_program *prog, gl_register_file type); + +extern void +_mesa_nop_fragment_program(GLcontext *ctx, struct gl_fragment_program *prog); + +extern void +_mesa_nop_vertex_program(GLcontext *ctx, struct gl_vertex_program *prog); + + +#endif /* PROGRAMOPT_H */ diff --git a/mesalib/src/mesa/program/symbol_table.c b/mesalib/src/mesa/program/symbol_table.c new file mode 100644 index 000000000..09e7cb44e --- /dev/null +++ b/mesalib/src/mesa/program/symbol_table.c @@ -0,0 +1,413 @@ +/* + * Copyright © 2008 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. + */ + +#include "main/imports.h" +#include "symbol_table.h" +#include "hash_table.h" + +struct symbol { + /** + * Link to the next symbol in the table with the same name + * + * The linked list of symbols with the same name is ordered by scope + * from inner-most to outer-most. + */ + struct symbol *next_with_same_name; + + + /** + * Link to the next symbol in the table with the same scope + * + * The linked list of symbols with the same scope is unordered. Symbols + * in this list my have unique names. + */ + struct symbol *next_with_same_scope; + + + /** + * Header information for the list of symbols with the same name. + */ + struct symbol_header *hdr; + + + /** + * Name space of the symbol + * + * Name space are arbitrary user assigned integers. No two symbols can + * exist in the same name space at the same scope level. + */ + int name_space; + + /** Scope depth where this symbol was defined. */ + unsigned depth; + + /** + * Arbitrary user supplied data. + */ + void *data; +}; + + +/** + */ +struct symbol_header { + /** Linkage in list of all headers in a given symbol table. */ + struct symbol_header *next; + + /** Symbol name. */ + char *name; + + /** Linked list of symbols with the same name. */ + struct symbol *symbols; +}; + + +/** + * Element of the scope stack. + */ +struct scope_level { + /** Link to next (inner) scope level. */ + struct scope_level *next; + + /** Linked list of symbols with the same scope. */ + struct symbol *symbols; +}; + + +/** + * + */ +struct _mesa_symbol_table { + /** Hash table containing all symbols in the symbol table. */ + struct hash_table *ht; + + /** Top of scope stack. */ + struct scope_level *current_scope; + + /** List of all symbol headers in the table. */ + struct symbol_header *hdr; + + /** Current scope depth. */ + unsigned depth; +}; + + +struct _mesa_symbol_table_iterator { + /** + * Name space of symbols returned by this iterator. + */ + int name_space; + + + /** + * Currently iterated symbol + * + * The next call to \c _mesa_symbol_table_iterator_get will return this + * value. It will also update this value to the value that should be + * returned by the next call. + */ + struct symbol *curr; +}; + + +static void +check_symbol_table(struct _mesa_symbol_table *table) +{ +#if 1 + struct scope_level *scope; + + for (scope = table->current_scope; scope != NULL; scope = scope->next) { + struct symbol *sym; + + for (sym = scope->symbols + ; sym != NULL + ; sym = sym->next_with_same_name) { + const struct symbol_header *const hdr = sym->hdr; + struct symbol *sym2; + + for (sym2 = hdr->symbols + ; sym2 != NULL + ; sym2 = sym2->next_with_same_name) { + assert(sym2->hdr == hdr); + } + } + } +#endif +} + +void +_mesa_symbol_table_pop_scope(struct _mesa_symbol_table *table) +{ + struct scope_level *const scope = table->current_scope; + struct symbol *sym = scope->symbols; + + table->current_scope = scope->next; + table->depth--; + + free(scope); + + while (sym != NULL) { + struct symbol *const next = sym->next_with_same_scope; + struct symbol_header *const hdr = sym->hdr; + + assert(hdr->symbols == sym); + + hdr->symbols = sym->next_with_same_name; + + free(sym); + + sym = next; + } + + check_symbol_table(table); +} + + +void +_mesa_symbol_table_push_scope(struct _mesa_symbol_table *table) +{ + struct scope_level *const scope = calloc(1, sizeof(*scope)); + + scope->next = table->current_scope; + table->current_scope = scope; + table->depth++; +} + + +static struct symbol_header * +find_symbol(struct _mesa_symbol_table *table, const char *name) +{ + return (struct symbol_header *) hash_table_find(table->ht, name); +} + + +struct _mesa_symbol_table_iterator * +_mesa_symbol_table_iterator_ctor(struct _mesa_symbol_table *table, + int name_space, const char *name) +{ + struct _mesa_symbol_table_iterator *iter = calloc(1, sizeof(*iter)); + struct symbol_header *const hdr = find_symbol(table, name); + + iter->name_space = name_space; + + if (hdr != NULL) { + struct symbol *sym; + + for (sym = hdr->symbols; sym != NULL; sym = sym->next_with_same_name) { + assert(sym->hdr == hdr); + + if ((name_space == -1) || (sym->name_space == name_space)) { + iter->curr = sym; + break; + } + } + } + + return iter; +} + + +void +_mesa_symbol_table_iterator_dtor(struct _mesa_symbol_table_iterator *iter) +{ + free(iter); +} + + +void * +_mesa_symbol_table_iterator_get(struct _mesa_symbol_table_iterator *iter) +{ + return (iter->curr == NULL) ? NULL : iter->curr->data; +} + + +int +_mesa_symbol_table_iterator_next(struct _mesa_symbol_table_iterator *iter) +{ + struct symbol_header *hdr; + + if (iter->curr == NULL) { + return 0; + } + + hdr = iter->curr->hdr; + iter->curr = iter->curr->next_with_same_name; + + while (iter->curr != NULL) { + assert(iter->curr->hdr == hdr); + + if ((iter->name_space == -1) + || (iter->curr->name_space == iter->name_space)) { + return 1; + } + + iter->curr = iter->curr->next_with_same_name; + } + + return 0; +} + + +/** + * Determine the scope "distance" of a symbol from the current scope + * + * \return + * A non-negative number for the number of scopes between the current scope + * and the scope where a symbol was defined. A value of zero means the current + * scope. A negative number if the symbol does not exist. + */ +int +_mesa_symbol_table_symbol_scope(struct _mesa_symbol_table *table, + int name_space, const char *name) +{ + struct symbol_header *const hdr = find_symbol(table, name); + struct symbol *sym; + + if (hdr != NULL) { + for (sym = hdr->symbols; sym != NULL; sym = sym->next_with_same_name) { + assert(sym->hdr == hdr); + + if ((name_space == -1) || (sym->name_space == name_space)) { + assert(sym->depth <= table->depth); + return sym->depth - table->depth; + } + } + } + + return -1; +} + + +void * +_mesa_symbol_table_find_symbol(struct _mesa_symbol_table *table, + int name_space, const char *name) +{ + struct symbol_header *const hdr = find_symbol(table, name); + + if (hdr != NULL) { + struct symbol *sym; + + + for (sym = hdr->symbols; sym != NULL; sym = sym->next_with_same_name) { + assert(sym->hdr == hdr); + + if ((name_space == -1) || (sym->name_space == name_space)) { + return sym->data; + } + } + } + + return NULL; +} + + +int +_mesa_symbol_table_add_symbol(struct _mesa_symbol_table *table, + int name_space, const char *name, + void *declaration) +{ + struct symbol_header *hdr; + struct symbol *sym; + + check_symbol_table(table); + + hdr = find_symbol(table, name); + + check_symbol_table(table); + + if (hdr == NULL) { + hdr = calloc(1, sizeof(*hdr)); + hdr->name = strdup(name); + + hash_table_insert(table->ht, hdr, hdr->name); + hdr->next = table->hdr; + table->hdr = hdr; + } + + check_symbol_table(table); + + /* If the symbol already exists in this namespace at this scope, it cannot + * be added to the table. + */ + for (sym = hdr->symbols + ; (sym != NULL) && (sym->name_space != name_space) + ; sym = sym->next_with_same_name) { + /* empty */ + } + + if (sym && (sym->depth == table->depth)) + return -1; + + sym = calloc(1, sizeof(*sym)); + sym->next_with_same_name = hdr->symbols; + sym->next_with_same_scope = table->current_scope->symbols; + sym->hdr = hdr; + sym->name_space = name_space; + sym->data = declaration; + sym->depth = table->depth; + + assert(sym->hdr == hdr); + + hdr->symbols = sym; + table->current_scope->symbols = sym; + + check_symbol_table(table); + return 0; +} + + +struct _mesa_symbol_table * +_mesa_symbol_table_ctor(void) +{ + struct _mesa_symbol_table *table = calloc(1, sizeof(*table)); + + if (table != NULL) { + table->ht = hash_table_ctor(32, hash_table_string_hash, + hash_table_string_compare); + + _mesa_symbol_table_push_scope(table); + } + + return table; +} + + +void +_mesa_symbol_table_dtor(struct _mesa_symbol_table *table) +{ + struct symbol_header *hdr; + struct symbol_header *next; + + while (table->current_scope != NULL) { + _mesa_symbol_table_pop_scope(table); + } + + for (hdr = table->hdr; hdr != NULL; hdr = next) { + next = hdr->next; + free(hdr->name); + free(hdr); + } + + hash_table_dtor(table->ht); + free(table); +} diff --git a/mesalib/src/mesa/program/symbol_table.h b/mesalib/src/mesa/program/symbol_table.h new file mode 100644 index 000000000..1d570fc1a --- /dev/null +++ b/mesalib/src/mesa/program/symbol_table.h @@ -0,0 +1,58 @@ +/* + * Copyright © 2008 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. + */ +#ifndef MESA_SYMBOL_TABLE_H +#define MESA_SYMBOL_TABLE_H + +struct _mesa_symbol_table; +struct _mesa_symbol_table_iterator; + +extern void _mesa_symbol_table_push_scope(struct _mesa_symbol_table *table); + +extern void _mesa_symbol_table_pop_scope(struct _mesa_symbol_table *table); + +extern int _mesa_symbol_table_add_symbol(struct _mesa_symbol_table *symtab, + int name_space, const char *name, void *declaration); + +extern int _mesa_symbol_table_symbol_scope(struct _mesa_symbol_table *table, + int name_space, const char *name); + +extern void *_mesa_symbol_table_find_symbol( + struct _mesa_symbol_table *symtab, int name_space, const char *name); + +extern struct _mesa_symbol_table *_mesa_symbol_table_ctor(void); + +extern void _mesa_symbol_table_dtor(struct _mesa_symbol_table *); + +extern struct _mesa_symbol_table_iterator *_mesa_symbol_table_iterator_ctor( + struct _mesa_symbol_table *table, int name_space, const char *name); + +extern void _mesa_symbol_table_iterator_dtor( + struct _mesa_symbol_table_iterator *); + +extern void *_mesa_symbol_table_iterator_get( + struct _mesa_symbol_table_iterator *iter); + +extern int _mesa_symbol_table_iterator_next( + struct _mesa_symbol_table_iterator *iter); + +#endif /* MESA_SYMBOL_TABLE_H */ |