diff options
author | marha <marha@users.sourceforge.net> | 2014-05-11 17:43:25 +0200 |
---|---|---|
committer | marha <marha@users.sourceforge.net> | 2014-05-11 17:43:25 +0200 |
commit | 26dfcdba6c306d6312c598e4191449302329be68 (patch) | |
tree | 2bf56b5a0eb2cc554833fdc6e08f2b59f1bfdabd /mesalib/src/glsl | |
parent | 2dc2880eacae3d52f4ab4bb5ec43c5bbf50c5c3f (diff) | |
parent | ae06feae7876db47ff0e1fde40cf4a324a412037 (diff) | |
download | vcxsrv-26dfcdba6c306d6312c598e4191449302329be68.tar.gz vcxsrv-26dfcdba6c306d6312c598e4191449302329be68.tar.bz2 vcxsrv-26dfcdba6c306d6312c598e4191449302329be68.zip |
Merge remote-tracking branch 'origin/released'
Conflicts:
mesalib/src/mesa/main/imports.h
Diffstat (limited to 'mesalib/src/glsl')
-rw-r--r-- | mesalib/src/glsl/Makefile.am | 11 | ||||
-rw-r--r-- | mesalib/src/glsl/ast_to_hir.cpp | 80 | ||||
-rw-r--r-- | mesalib/src/glsl/builtin_types.cpp | 2 | ||||
-rw-r--r-- | mesalib/src/glsl/builtin_variables.cpp | 63 | ||||
-rw-r--r-- | mesalib/src/glsl/glcpp/glcpp-parse.y | 5 | ||||
-rw-r--r-- | mesalib/src/glsl/glsl_parser.yy | 15 | ||||
-rw-r--r-- | mesalib/src/glsl/glsl_parser_extras.cpp | 82 | ||||
-rw-r--r-- | mesalib/src/glsl/glsl_parser_extras.h | 116 | ||||
-rw-r--r-- | mesalib/src/glsl/ir.cpp | 2 | ||||
-rw-r--r-- | mesalib/src/glsl/ir_optimization.h | 4 | ||||
-rw-r--r-- | mesalib/src/glsl/link_varyings.cpp | 374 | ||||
-rw-r--r-- | mesalib/src/glsl/linker.cpp | 190 | ||||
-rw-r--r-- | mesalib/src/glsl/lower_instructions.cpp | 58 | ||||
-rw-r--r-- | mesalib/src/glsl/lower_packed_varyings.cpp | 39 | ||||
-rw-r--r-- | mesalib/src/glsl/main.cpp | 6 | ||||
-rw-r--r-- | mesalib/src/glsl/opt_dead_builtin_varyings.cpp | 7 |
16 files changed, 813 insertions, 241 deletions
diff --git a/mesalib/src/glsl/Makefile.am b/mesalib/src/glsl/Makefile.am index 534eaa385..fd0e837d1 100644 --- a/mesalib/src/glsl/Makefile.am +++ b/mesalib/src/glsl/Makefile.am @@ -61,7 +61,9 @@ tests_general_ir_test_SOURCES = \ $(GLSL_SRCDIR)/standalone_scaffolding.cpp \ tests/builtin_variable_test.cpp \ tests/invalidate_locations_test.cpp \ - tests/general_ir_test.cpp + tests/general_ir_test.cpp \ + tests/varyings_test.cpp \ + tests/common.c tests_general_ir_test_CFLAGS = \ $(PTHREAD_CFLAGS) tests_general_ir_test_LDADD = \ @@ -76,7 +78,8 @@ tests_uniform_initializer_test_SOURCES = \ $(top_srcdir)/src/mesa/program/symbol_table.c \ tests/copy_constant_to_storage_tests.cpp \ tests/set_uniform_initializer_tests.cpp \ - tests/uniform_initializer_utils.cpp + tests/uniform_initializer_utils.cpp \ + tests/common.c tests_uniform_initializer_test_CFLAGS = \ $(PTHREAD_CFLAGS) tests_uniform_initializer_test_LDADD = \ @@ -95,7 +98,8 @@ tests_ralloc_test_LDADD = \ tests_sampler_types_test_SOURCES = \ $(top_srcdir)/src/mesa/program/prog_hash_table.c\ $(top_srcdir)/src/mesa/program/symbol_table.c \ - tests/sampler_types_test.cpp + tests/sampler_types_test.cpp \ + tests/common.c tests_sampler_types_test_CFLAGS = \ $(PTHREAD_CFLAGS) tests_sampler_types_test_LDADD = \ @@ -138,6 +142,7 @@ glsl_test_SOURCES = \ $(top_srcdir)/src/mesa/program/prog_hash_table.c \ $(top_srcdir)/src/mesa/program/symbol_table.c \ $(GLSL_SRCDIR)/standalone_scaffolding.cpp \ + tests/common.c \ test.cpp \ test_optpass.cpp diff --git a/mesalib/src/glsl/ast_to_hir.cpp b/mesalib/src/glsl/ast_to_hir.cpp index 0411befa9..7516c33e1 100644 --- a/mesalib/src/glsl/ast_to_hir.cpp +++ b/mesalib/src/glsl/ast_to_hir.cpp @@ -123,6 +123,11 @@ _mesa_ast_to_hir(exec_list *instructions, struct _mesa_glsl_parse_state *state) instructions->push_head(var); } + /* Figure out if gl_FragCoord is actually used in fragment shader */ + ir_variable *const var = state->symbols->get_variable("gl_FragCoord"); + if (var != NULL) + state->fs_uses_gl_fragcoord = var->data.used; + /* From section 7.1 (Built-In Language Variables) of the GLSL 4.10 spec: * * If multiple shaders using members of a built-in block belonging to @@ -2341,6 +2346,34 @@ apply_image_qualifier_to_variable(const struct ast_type_qualifier *qual, } } +static inline const char* +get_layout_qualifier_string(bool origin_upper_left, bool pixel_center_integer) +{ + if (origin_upper_left && pixel_center_integer) + return "origin_upper_left, pixel_center_integer"; + else if (origin_upper_left) + return "origin_upper_left"; + else if (pixel_center_integer) + return "pixel_center_integer"; + else + return " "; +} + +static inline bool +is_conflicting_fragcoord_redeclaration(struct _mesa_glsl_parse_state *state, + const struct ast_type_qualifier *qual) +{ + /* If gl_FragCoord was previously declared, and the qualifiers were + * different in any way, return true. + */ + if (state->fs_redeclares_gl_fragcoord) { + return (state->fs_pixel_center_integer != qual->flags.q.pixel_center_integer + || state->fs_origin_upper_left != qual->flags.q.origin_upper_left); + } + + return false; +} + static void apply_type_qualifier_to_variable(const struct ast_type_qualifier *qual, ir_variable *var, @@ -2505,6 +2538,53 @@ apply_type_qualifier_to_variable(const struct ast_type_qualifier *qual, qual_string); } + if (var->name != NULL && strcmp(var->name, "gl_FragCoord") == 0) { + + /* Section 4.3.8.1, page 39 of GLSL 1.50 spec says: + * + * "Within any shader, the first redeclarations of gl_FragCoord + * must appear before any use of gl_FragCoord." + * + * Generate a compiler error if above condition is not met by the + * fragment shader. + */ + ir_variable *earlier = state->symbols->get_variable("gl_FragCoord"); + if (earlier != NULL && + earlier->data.used && + !state->fs_redeclares_gl_fragcoord) { + _mesa_glsl_error(loc, state, + "gl_FragCoord used before its first redeclaration " + "in fragment shader"); + } + + /* Make sure all gl_FragCoord redeclarations specify the same layout + * qualifiers. + */ + if (is_conflicting_fragcoord_redeclaration(state, qual)) { + const char *const qual_string = + get_layout_qualifier_string(qual->flags.q.origin_upper_left, + qual->flags.q.pixel_center_integer); + + const char *const state_string = + get_layout_qualifier_string(state->fs_origin_upper_left, + state->fs_pixel_center_integer); + + _mesa_glsl_error(loc, state, + "gl_FragCoord redeclared with different layout " + "qualifiers (%s) and (%s) ", + state_string, + qual_string); + } + state->fs_origin_upper_left = qual->flags.q.origin_upper_left; + state->fs_pixel_center_integer = qual->flags.q.pixel_center_integer; + state->fs_redeclares_gl_fragcoord_with_no_layout_qualifiers = + !qual->flags.q.origin_upper_left && !qual->flags.q.pixel_center_integer; + state->fs_redeclares_gl_fragcoord = + state->fs_origin_upper_left || + state->fs_pixel_center_integer || + state->fs_redeclares_gl_fragcoord_with_no_layout_qualifiers; + } + if (qual->flags.q.explicit_location) { validate_explicit_location(qual, var, state, loc); } else if (qual->flags.q.explicit_index) { diff --git a/mesalib/src/glsl/builtin_types.cpp b/mesalib/src/glsl/builtin_types.cpp index dd42ecb05..0a0fa8cd3 100644 --- a/mesalib/src/glsl/builtin_types.cpp +++ b/mesalib/src/glsl/builtin_types.cpp @@ -241,7 +241,7 @@ const static struct builtin_type_versions { T(atomic_uint, 420, 999) }; -const glsl_type *const deprecated_types[] = { +static const glsl_type *const deprecated_types[] = { glsl_type::struct_gl_PointParameters_type, glsl_type::struct_gl_MaterialParameters_type, glsl_type::struct_gl_LightSourceParameters_type, diff --git a/mesalib/src/glsl/builtin_variables.cpp b/mesalib/src/glsl/builtin_variables.cpp index 4176ae6e6..9b35850ee 100644 --- a/mesalib/src/glsl/builtin_variables.cpp +++ b/mesalib/src/glsl/builtin_variables.cpp @@ -30,21 +30,21 @@ #include "program/prog_statevars.h" #include "program/prog_instruction.h" -static struct gl_builtin_uniform_element gl_NumSamples_elements[] = { +static const struct gl_builtin_uniform_element gl_NumSamples_elements[] = { {NULL, {STATE_NUM_SAMPLES, 0, 0}, SWIZZLE_XXXX} }; -static struct gl_builtin_uniform_element gl_DepthRange_elements[] = { +static const struct gl_builtin_uniform_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 gl_builtin_uniform_element gl_ClipPlane_elements[] = { +static const struct gl_builtin_uniform_element gl_ClipPlane_elements[] = { {NULL, {STATE_CLIPPLANE, 0, 0}, SWIZZLE_XYZW} }; -static struct gl_builtin_uniform_element gl_Point_elements[] = { +static const struct gl_builtin_uniform_element gl_Point_elements[] = { {"size", {STATE_POINT_SIZE}, SWIZZLE_XXXX}, {"sizeMin", {STATE_POINT_SIZE}, SWIZZLE_YYYY}, {"sizeMax", {STATE_POINT_SIZE}, SWIZZLE_ZZZZ}, @@ -54,7 +54,7 @@ static struct gl_builtin_uniform_element gl_Point_elements[] = { {"distanceQuadraticAttenuation", {STATE_POINT_ATTENUATION}, SWIZZLE_ZZZZ}, }; -static struct gl_builtin_uniform_element gl_FrontMaterial_elements[] = { +static const struct gl_builtin_uniform_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}, @@ -62,7 +62,7 @@ static struct gl_builtin_uniform_element gl_FrontMaterial_elements[] = { {"shininess", {STATE_MATERIAL, 0, STATE_SHININESS}, SWIZZLE_XXXX}, }; -static struct gl_builtin_uniform_element gl_BackMaterial_elements[] = { +static const struct gl_builtin_uniform_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}, @@ -70,7 +70,7 @@ static struct gl_builtin_uniform_element gl_BackMaterial_elements[] = { {"shininess", {STATE_MATERIAL, 1, STATE_SHININESS}, SWIZZLE_XXXX}, }; -static struct gl_builtin_uniform_element gl_LightSource_elements[] = { +static const struct gl_builtin_uniform_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}, @@ -89,67 +89,67 @@ static struct gl_builtin_uniform_element gl_LightSource_elements[] = { {"quadraticAttenuation", {STATE_LIGHT, 0, STATE_ATTENUATION}, SWIZZLE_ZZZZ}, }; -static struct gl_builtin_uniform_element gl_LightModel_elements[] = { +static const struct gl_builtin_uniform_element gl_LightModel_elements[] = { {"ambient", {STATE_LIGHTMODEL_AMBIENT, 0}, SWIZZLE_XYZW}, }; -static struct gl_builtin_uniform_element gl_FrontLightModelProduct_elements[] = { +static const struct gl_builtin_uniform_element gl_FrontLightModelProduct_elements[] = { {"sceneColor", {STATE_LIGHTMODEL_SCENECOLOR, 0}, SWIZZLE_XYZW}, }; -static struct gl_builtin_uniform_element gl_BackLightModelProduct_elements[] = { +static const struct gl_builtin_uniform_element gl_BackLightModelProduct_elements[] = { {"sceneColor", {STATE_LIGHTMODEL_SCENECOLOR, 1}, SWIZZLE_XYZW}, }; -static struct gl_builtin_uniform_element gl_FrontLightProduct_elements[] = { +static const struct gl_builtin_uniform_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 gl_builtin_uniform_element gl_BackLightProduct_elements[] = { +static const struct gl_builtin_uniform_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 gl_builtin_uniform_element gl_TextureEnvColor_elements[] = { +static const struct gl_builtin_uniform_element gl_TextureEnvColor_elements[] = { {NULL, {STATE_TEXENV_COLOR, 0}, SWIZZLE_XYZW}, }; -static struct gl_builtin_uniform_element gl_EyePlaneS_elements[] = { +static const struct gl_builtin_uniform_element gl_EyePlaneS_elements[] = { {NULL, {STATE_TEXGEN, 0, STATE_TEXGEN_EYE_S}, SWIZZLE_XYZW}, }; -static struct gl_builtin_uniform_element gl_EyePlaneT_elements[] = { +static const struct gl_builtin_uniform_element gl_EyePlaneT_elements[] = { {NULL, {STATE_TEXGEN, 0, STATE_TEXGEN_EYE_T}, SWIZZLE_XYZW}, }; -static struct gl_builtin_uniform_element gl_EyePlaneR_elements[] = { +static const struct gl_builtin_uniform_element gl_EyePlaneR_elements[] = { {NULL, {STATE_TEXGEN, 0, STATE_TEXGEN_EYE_R}, SWIZZLE_XYZW}, }; -static struct gl_builtin_uniform_element gl_EyePlaneQ_elements[] = { +static const struct gl_builtin_uniform_element gl_EyePlaneQ_elements[] = { {NULL, {STATE_TEXGEN, 0, STATE_TEXGEN_EYE_Q}, SWIZZLE_XYZW}, }; -static struct gl_builtin_uniform_element gl_ObjectPlaneS_elements[] = { +static const struct gl_builtin_uniform_element gl_ObjectPlaneS_elements[] = { {NULL, {STATE_TEXGEN, 0, STATE_TEXGEN_OBJECT_S}, SWIZZLE_XYZW}, }; -static struct gl_builtin_uniform_element gl_ObjectPlaneT_elements[] = { +static const struct gl_builtin_uniform_element gl_ObjectPlaneT_elements[] = { {NULL, {STATE_TEXGEN, 0, STATE_TEXGEN_OBJECT_T}, SWIZZLE_XYZW}, }; -static struct gl_builtin_uniform_element gl_ObjectPlaneR_elements[] = { +static const struct gl_builtin_uniform_element gl_ObjectPlaneR_elements[] = { {NULL, {STATE_TEXGEN, 0, STATE_TEXGEN_OBJECT_R}, SWIZZLE_XYZW}, }; -static struct gl_builtin_uniform_element gl_ObjectPlaneQ_elements[] = { +static const struct gl_builtin_uniform_element gl_ObjectPlaneQ_elements[] = { {NULL, {STATE_TEXGEN, 0, STATE_TEXGEN_OBJECT_Q}, SWIZZLE_XYZW}, }; -static struct gl_builtin_uniform_element gl_Fog_elements[] = { +static const struct gl_builtin_uniform_element gl_Fog_elements[] = { {"color", {STATE_FOG_COLOR}, SWIZZLE_XYZW}, {"density", {STATE_FOG_PARAMS}, SWIZZLE_XXXX}, {"start", {STATE_FOG_PARAMS}, SWIZZLE_YYYY}, @@ -157,32 +157,32 @@ static struct gl_builtin_uniform_element gl_Fog_elements[] = { {"scale", {STATE_FOG_PARAMS}, SWIZZLE_WWWW}, }; -static struct gl_builtin_uniform_element gl_NormalScale_elements[] = { +static const struct gl_builtin_uniform_element gl_NormalScale_elements[] = { {NULL, {STATE_NORMAL_SCALE}, SWIZZLE_XXXX}, }; -static struct gl_builtin_uniform_element gl_BumpRotMatrix0MESA_elements[] = { +static const struct gl_builtin_uniform_element gl_BumpRotMatrix0MESA_elements[] = { {NULL, {STATE_INTERNAL, STATE_ROT_MATRIX_0}, SWIZZLE_XYZW}, }; -static struct gl_builtin_uniform_element gl_BumpRotMatrix1MESA_elements[] = { +static const struct gl_builtin_uniform_element gl_BumpRotMatrix1MESA_elements[] = { {NULL, {STATE_INTERNAL, STATE_ROT_MATRIX_1}, SWIZZLE_XYZW}, }; -static struct gl_builtin_uniform_element gl_FogParamsOptimizedMESA_elements[] = { +static const struct gl_builtin_uniform_element gl_FogParamsOptimizedMESA_elements[] = { {NULL, {STATE_INTERNAL, STATE_FOG_PARAMS_OPTIMIZED}, SWIZZLE_XYZW}, }; -static struct gl_builtin_uniform_element gl_CurrentAttribVertMESA_elements[] = { +static const struct gl_builtin_uniform_element gl_CurrentAttribVertMESA_elements[] = { {NULL, {STATE_INTERNAL, STATE_CURRENT_ATTRIB, 0}, SWIZZLE_XYZW}, }; -static struct gl_builtin_uniform_element gl_CurrentAttribFragMESA_elements[] = { +static const struct gl_builtin_uniform_element gl_CurrentAttribFragMESA_elements[] = { {NULL, {STATE_INTERNAL, STATE_CURRENT_ATTRIB_MAYBE_VP_CLAMPED, 0}, SWIZZLE_XYZW}, }; #define MATRIX(name, statevar, modifier) \ - static struct gl_builtin_uniform_element name ## _elements[] = { \ + static const struct gl_builtin_uniform_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 }, \ @@ -225,7 +225,7 @@ MATRIX(gl_TextureMatrixTranspose, MATRIX(gl_TextureMatrixInverseTranspose, STATE_TEXTURE_MATRIX, STATE_MATRIX_INVERSE); -static struct gl_builtin_uniform_element gl_NormalMatrix_elements[] = { +static const struct gl_builtin_uniform_element gl_NormalMatrix_elements[] = { { NULL, { STATE_MODELVIEW_MATRIX, 0, 0, 0, STATE_MATRIX_INVERSE}, MAKE_SWIZZLE4(SWIZZLE_X, SWIZZLE_Y, SWIZZLE_Z, SWIZZLE_Z) }, { NULL, { STATE_MODELVIEW_MATRIX, 0, 1, 1, STATE_MATRIX_INVERSE}, @@ -498,7 +498,8 @@ builtin_variable_generator::add_uniform(const glsl_type *type, for (unsigned a = 0; a < array_count; a++) { for (unsigned j = 0; j < statevar->num_elements; j++) { - struct gl_builtin_uniform_element *element = &statevar->elements[j]; + const struct gl_builtin_uniform_element *element = + &statevar->elements[j]; memcpy(slots->tokens, element->tokens, sizeof(element->tokens)); if (type->is_array()) { diff --git a/mesalib/src/glsl/glcpp/glcpp-parse.y b/mesalib/src/glsl/glcpp/glcpp-parse.y index f28d8531e..98875837c 100644 --- a/mesalib/src/glsl/glcpp/glcpp-parse.y +++ b/mesalib/src/glsl/glcpp/glcpp-parse.y @@ -2062,6 +2062,7 @@ _glcpp_parser_handle_version_declaration(glcpp_parser_t *parser, intmax_t versio /* Add pre-defined macros. */ if (parser->is_gles) { add_builtin_define(parser, "GL_ES", 1); + add_builtin_define(parser, "GL_EXT_separate_shader_objects", 1); if (extensions != NULL) { if (extensions->OES_EGL_image_external) @@ -2069,6 +2070,7 @@ _glcpp_parser_handle_version_declaration(glcpp_parser_t *parser, intmax_t versio } } else { add_builtin_define(parser, "GL_ARB_draw_buffers", 1); + add_builtin_define(parser, "GL_ARB_separate_shader_objects", 1); add_builtin_define(parser, "GL_ARB_texture_rectangle", 1); add_builtin_define(parser, "GL_AMD_shader_trinary_minmax", 1); @@ -2134,9 +2136,6 @@ _glcpp_parser_handle_version_declaration(glcpp_parser_t *parser, intmax_t versio if (extensions->ARB_texture_gather) add_builtin_define(parser, "GL_ARB_texture_gather", 1); - if (extensions->ARB_separate_shader_objects) - add_builtin_define(parser, "GL_ARB_separate_shader_objects", 1); - if (extensions->ARB_shader_atomic_counters) add_builtin_define(parser, "GL_ARB_shader_atomic_counters", 1); diff --git a/mesalib/src/glsl/glsl_parser.yy b/mesalib/src/glsl/glsl_parser.yy index 531ae394b..b71b240de 100644 --- a/mesalib/src/glsl/glsl_parser.yy +++ b/mesalib/src/glsl/glsl_parser.yy @@ -1215,7 +1215,7 @@ layout_qualifier_id: /* Layout qualifiers for GLSL 1.50 geometry shaders. */ if (!$$.flags.i) { - struct { + static const struct { const char *s; GLenum e; } map[] = { @@ -1320,6 +1320,13 @@ layout_qualifier_id: if (match_layout_qualifier("location", $1, state) == 0) { $$.flags.q.explicit_location = 1; + if ($$.flags.q.attribute == 1 && + state->ARB_explicit_attrib_location_warn) { + _mesa_glsl_warning(& @1, state, + "GL_ARB_explicit_attrib_location layout " + "identifier `%s' used", $1); + } + if ($3 >= 0) { $$.location = $3; } else { @@ -1369,7 +1376,7 @@ layout_qualifier_id: } } - static const char *local_size_qualifiers[3] = { + static const char * const local_size_qualifiers[3] = { "local_size_x", "local_size_y", "local_size_z", @@ -1427,10 +1434,6 @@ layout_qualifier_id: _mesa_glsl_error(& @1, state, "unrecognized layout identifier " "`%s'", $1); YYERROR; - } else if (state->ARB_explicit_attrib_location_warn) { - _mesa_glsl_warning(& @1, state, - "GL_ARB_explicit_attrib_location layout " - "identifier `%s' used", $1); } } | interface_block_layout_qualifier diff --git a/mesalib/src/glsl/glsl_parser_extras.cpp b/mesalib/src/glsl/glsl_parser_extras.cpp index 03c2a972a..d3339e779 100644 --- a/mesalib/src/glsl/glsl_parser_extras.cpp +++ b/mesalib/src/glsl/glsl_parser_extras.cpp @@ -49,7 +49,7 @@ glsl_compute_version_string(void *mem_ctx, bool is_es, unsigned version) } -static unsigned known_desktop_glsl_versions[] = +static const unsigned known_desktop_glsl_versions[] = { 110, 120, 130, 140, 150, 330, 400, 410, 420, 430, 440 }; @@ -197,6 +197,12 @@ _mesa_glsl_parse_state::_mesa_glsl_parse_state(struct gl_context *_ctx, this->default_uniform_qualifier->flags.q.shared = 1; this->default_uniform_qualifier->flags.q.column_major = 1; + this->fs_uses_gl_fragcoord = false; + this->fs_redeclares_gl_fragcoord = false; + this->fs_origin_upper_left = false; + this->fs_pixel_center_integer = false; + this->fs_redeclares_gl_fragcoord_with_no_layout_qualifiers = false; + this->gs_input_prim_type_specified = false; this->gs_input_size = 0; this->in_qualifier = new(this) ast_type_qualifier(); @@ -500,40 +506,53 @@ struct _mesa_glsl_extension { static const _mesa_glsl_extension _mesa_glsl_supported_extensions[] = { /* API availability */ /* name GL ES supported flag */ + + /* ARB extensions go here, sorted alphabetically. + */ EXT(ARB_arrays_of_arrays, true, false, ARB_arrays_of_arrays), + EXT(ARB_compute_shader, true, false, ARB_compute_shader), EXT(ARB_conservative_depth, true, false, ARB_conservative_depth), EXT(ARB_draw_buffers, true, false, dummy_true), EXT(ARB_draw_instanced, true, false, ARB_draw_instanced), EXT(ARB_explicit_attrib_location, true, false, ARB_explicit_attrib_location), EXT(ARB_fragment_coord_conventions, true, false, ARB_fragment_coord_conventions), - EXT(ARB_texture_rectangle, true, false, dummy_true), - EXT(EXT_texture_array, true, false, EXT_texture_array), - EXT(ARB_separate_shader_objects, true, false, ARB_separate_shader_objects), - EXT(ARB_shader_texture_lod, true, false, ARB_shader_texture_lod), - EXT(ARB_shader_stencil_export, true, false, ARB_shader_stencil_export), - EXT(AMD_conservative_depth, true, false, ARB_conservative_depth), - EXT(AMD_shader_stencil_export, true, false, ARB_shader_stencil_export), - EXT(OES_texture_3D, false, true, EXT_texture3D), - EXT(OES_EGL_image_external, false, true, OES_EGL_image_external), + EXT(ARB_gpu_shader5, true, false, ARB_gpu_shader5), + EXT(ARB_sample_shading, true, false, ARB_sample_shading), + EXT(ARB_separate_shader_objects, true, false, dummy_true), + EXT(ARB_shader_atomic_counters, true, false, ARB_shader_atomic_counters), EXT(ARB_shader_bit_encoding, true, false, ARB_shader_bit_encoding), - EXT(ARB_uniform_buffer_object, true, false, ARB_uniform_buffer_object), - EXT(OES_standard_derivatives, false, true, OES_standard_derivatives), - EXT(ARB_texture_cube_map_array, true, false, ARB_texture_cube_map_array), - EXT(ARB_shading_language_packing, true, false, ARB_shading_language_packing), + EXT(ARB_shader_image_load_store, true, false, ARB_shader_image_load_store), + EXT(ARB_shader_stencil_export, true, false, ARB_shader_stencil_export), + EXT(ARB_shader_texture_lod, true, false, ARB_shader_texture_lod), EXT(ARB_shading_language_420pack, true, false, ARB_shading_language_420pack), + EXT(ARB_shading_language_packing, true, false, ARB_shading_language_packing), + EXT(ARB_texture_cube_map_array, true, false, ARB_texture_cube_map_array), + EXT(ARB_texture_gather, true, false, ARB_texture_gather), EXT(ARB_texture_multisample, true, false, ARB_texture_multisample), EXT(ARB_texture_query_levels, true, false, ARB_texture_query_levels), EXT(ARB_texture_query_lod, true, false, ARB_texture_query_lod), - EXT(ARB_gpu_shader5, true, false, ARB_gpu_shader5), + EXT(ARB_texture_rectangle, true, false, dummy_true), + EXT(ARB_uniform_buffer_object, true, false, ARB_uniform_buffer_object), + EXT(ARB_viewport_array, true, false, ARB_viewport_array), + + /* KHR extensions go here, sorted alphabetically. + */ + + /* OES extensions go here, sorted alphabetically. + */ + EXT(OES_EGL_image_external, false, true, OES_EGL_image_external), + EXT(OES_standard_derivatives, false, true, OES_standard_derivatives), + EXT(OES_texture_3D, false, true, EXT_texture3D), + + /* All other extensions go here, sorted alphabetically. + */ + EXT(AMD_conservative_depth, true, false, ARB_conservative_depth), + EXT(AMD_shader_stencil_export, true, false, ARB_shader_stencil_export), + EXT(AMD_shader_trinary_minmax, true, false, dummy_true), EXT(AMD_vertex_shader_layer, true, false, AMD_vertex_shader_layer), + EXT(EXT_separate_shader_objects, false, true, dummy_true), EXT(EXT_shader_integer_mix, true, true, EXT_shader_integer_mix), - EXT(ARB_texture_gather, true, false, ARB_texture_gather), - EXT(ARB_shader_atomic_counters, true, false, ARB_shader_atomic_counters), - EXT(ARB_sample_shading, true, false, ARB_sample_shading), - EXT(AMD_shader_trinary_minmax, true, false, dummy_true), - EXT(ARB_viewport_array, true, false, ARB_viewport_array), - EXT(ARB_compute_shader, true, false, ARB_compute_shader), - EXT(ARB_shader_image_load_store, true, false, ARB_shader_image_load_store), + EXT(EXT_texture_array, true, false, EXT_texture_array), }; #undef EXT @@ -637,7 +656,7 @@ _mesa_glsl_process_extension(const char *name, YYLTYPE *name_locp, if (extension && extension->compatible_with_state(state)) { extension->set_flags(state, behavior); } else { - static const char *const fmt = "extension `%s' unsupported in %s shader"; + static const char fmt[] = "extension `%s' unsupported in %s shader"; if (behavior == extension_require) { _mesa_glsl_error(name_locp, state, fmt, @@ -1356,6 +1375,14 @@ set_shader_inout_layout(struct gl_shader *shader, assert(!state->cs_input_local_size_specified); } + if (shader->Stage != MESA_SHADER_FRAGMENT) { + /* Should have been prevented by the parser. */ + assert(!state->fs_uses_gl_fragcoord); + assert(!state->fs_redeclares_gl_fragcoord); + assert(!state->fs_pixel_center_integer); + assert(!state->fs_origin_upper_left); + } + switch (shader->Stage) { case MESA_SHADER_GEOMETRY: shader->Geom.VerticesOut = 0; @@ -1389,6 +1416,15 @@ set_shader_inout_layout(struct gl_shader *shader, } break; + case MESA_SHADER_FRAGMENT: + shader->redeclares_gl_fragcoord = state->fs_redeclares_gl_fragcoord; + shader->uses_gl_fragcoord = state->fs_uses_gl_fragcoord; + shader->pixel_center_integer = state->fs_pixel_center_integer; + shader->origin_upper_left = state->fs_origin_upper_left; + shader->ARB_fragment_coord_conventions_enable = + state->ARB_fragment_coord_conventions_enable; + break; + default: /* Nothing to do. */ break; diff --git a/mesalib/src/glsl/glsl_parser_extras.h b/mesalib/src/glsl/glsl_parser_extras.h index 0fd778dc3..49402fa21 100644 --- a/mesalib/src/glsl/glsl_parser_extras.h +++ b/mesalib/src/glsl/glsl_parser_extras.h @@ -144,8 +144,7 @@ struct _mesa_glsl_parse_state { { if (!this->has_separate_shader_objects()) { const char *const requirement = this->es_shader - ? "GL_EXT_separate_shader_objects (not supported by this " - "implementation)" + ? "GL_EXT_separate_shader_objects extension" : "GL_ARB_separate_shader_objects extension or GLSL 420"; _mesa_glsl_error(locp, this, "%s explicit location requires %s", @@ -168,7 +167,8 @@ struct _mesa_glsl_parse_state { bool has_separate_shader_objects() const { - return ARB_separate_shader_objects_enable || is_version(410, 0); + return ARB_separate_shader_objects_enable || is_version(410, 0) + || EXT_separate_shader_objects_enable; } void process_version_directive(YYLTYPE *locp, int version, @@ -204,6 +204,18 @@ struct _mesa_glsl_parse_state { struct ast_type_qualifier *default_uniform_qualifier; /** + * Variables to track different cases if a fragment shader redeclares + * built-in variable gl_FragCoord. + * + * Note: These values are computed at ast_to_hir time rather than at parse + * time. + */ + bool fs_redeclares_gl_fragcoord; + bool fs_origin_upper_left; + bool fs_pixel_center_integer; + bool fs_redeclares_gl_fragcoord_with_no_layout_qualifiers; + + /** * True if a geometry shader input primitive type was specified using a * layout directive. * @@ -341,8 +353,14 @@ struct _mesa_glsl_parse_state { * \name Enable bits for GLSL extensions */ /*@{*/ + /* ARB extensions go here, sorted alphabetically. + */ bool ARB_arrays_of_arrays_enable; bool ARB_arrays_of_arrays_warn; + bool ARB_compute_shader_enable; + bool ARB_compute_shader_warn; + bool ARB_conservative_depth_enable; + bool ARB_conservative_depth_warn; bool ARB_draw_buffers_enable; bool ARB_draw_buffers_warn; bool ARB_draw_instanced_enable; @@ -351,70 +369,78 @@ struct _mesa_glsl_parse_state { bool ARB_explicit_attrib_location_warn; bool ARB_fragment_coord_conventions_enable; bool ARB_fragment_coord_conventions_warn; - bool ARB_texture_rectangle_enable; - bool ARB_texture_rectangle_warn; - bool ARB_texture_gather_enable; - bool ARB_texture_gather_warn; - bool EXT_texture_array_enable; - bool EXT_texture_array_warn; + bool ARB_gpu_shader5_enable; + bool ARB_gpu_shader5_warn; + bool ARB_sample_shading_enable; + bool ARB_sample_shading_warn; bool ARB_separate_shader_objects_enable; bool ARB_separate_shader_objects_warn; - bool ARB_shader_texture_lod_enable; - bool ARB_shader_texture_lod_warn; - bool ARB_shader_stencil_export_enable; - bool ARB_shader_stencil_export_warn; - bool AMD_conservative_depth_enable; - bool AMD_conservative_depth_warn; - bool ARB_conservative_depth_enable; - bool ARB_conservative_depth_warn; - bool AMD_shader_stencil_export_enable; - bool AMD_shader_stencil_export_warn; - bool OES_texture_3D_enable; - bool OES_texture_3D_warn; - bool OES_EGL_image_external_enable; - bool OES_EGL_image_external_warn; + bool ARB_shader_atomic_counters_enable; + bool ARB_shader_atomic_counters_warn; bool ARB_shader_bit_encoding_enable; bool ARB_shader_bit_encoding_warn; - bool ARB_uniform_buffer_object_enable; - bool ARB_uniform_buffer_object_warn; - bool OES_standard_derivatives_enable; - bool OES_standard_derivatives_warn; - bool ARB_texture_cube_map_array_enable; - bool ARB_texture_cube_map_array_warn; + bool ARB_shader_image_load_store_enable; + bool ARB_shader_image_load_store_warn; + bool ARB_shader_stencil_export_enable; + bool ARB_shader_stencil_export_warn; + bool ARB_shader_texture_lod_enable; + bool ARB_shader_texture_lod_warn; + bool ARB_shading_language_420pack_enable; + bool ARB_shading_language_420pack_warn; bool ARB_shading_language_packing_enable; bool ARB_shading_language_packing_warn; + bool ARB_texture_cube_map_array_enable; + bool ARB_texture_cube_map_array_warn; + bool ARB_texture_gather_enable; + bool ARB_texture_gather_warn; bool ARB_texture_multisample_enable; bool ARB_texture_multisample_warn; bool ARB_texture_query_levels_enable; bool ARB_texture_query_levels_warn; bool ARB_texture_query_lod_enable; bool ARB_texture_query_lod_warn; - bool ARB_gpu_shader5_enable; - bool ARB_gpu_shader5_warn; + bool ARB_texture_rectangle_enable; + bool ARB_texture_rectangle_warn; + bool ARB_uniform_buffer_object_enable; + bool ARB_uniform_buffer_object_warn; + bool ARB_viewport_array_enable; + bool ARB_viewport_array_warn; + + /* KHR extensions go here, sorted alphabetically. + */ + + /* OES extensions go here, sorted alphabetically. + */ + bool OES_EGL_image_external_enable; + bool OES_EGL_image_external_warn; + bool OES_standard_derivatives_enable; + bool OES_standard_derivatives_warn; + bool OES_texture_3D_enable; + bool OES_texture_3D_warn; + + /* All other extensions go here, sorted alphabetically. + */ + bool AMD_conservative_depth_enable; + bool AMD_conservative_depth_warn; + bool AMD_shader_stencil_export_enable; + bool AMD_shader_stencil_export_warn; + bool AMD_shader_trinary_minmax_enable; + bool AMD_shader_trinary_minmax_warn; bool AMD_vertex_shader_layer_enable; bool AMD_vertex_shader_layer_warn; - bool ARB_shading_language_420pack_enable; - bool ARB_shading_language_420pack_warn; - bool ARB_sample_shading_enable; - bool ARB_sample_shading_warn; + bool EXT_separate_shader_objects_enable; + bool EXT_separate_shader_objects_warn; bool EXT_shader_integer_mix_enable; bool EXT_shader_integer_mix_warn; - bool ARB_shader_atomic_counters_enable; - bool ARB_shader_atomic_counters_warn; - bool AMD_shader_trinary_minmax_enable; - bool AMD_shader_trinary_minmax_warn; - bool ARB_viewport_array_enable; - bool ARB_viewport_array_warn; - bool ARB_compute_shader_enable; - bool ARB_compute_shader_warn; - bool ARB_shader_image_load_store_enable; - bool ARB_shader_image_load_store_warn; + bool EXT_texture_array_enable; + bool EXT_texture_array_warn; /*@}*/ /** Extensions supported by the OpenGL implementation. */ const struct gl_extensions *extensions; bool uses_builtin_functions; + bool fs_uses_gl_fragcoord; /** * For geometry shaders, size of the most recently seen input declaration diff --git a/mesalib/src/glsl/ir.cpp b/mesalib/src/glsl/ir.cpp index f8f661d5e..dec49bd64 100644 --- a/mesalib/src/glsl/ir.cpp +++ b/mesalib/src/glsl/ir.cpp @@ -1333,7 +1333,7 @@ ir_dereference::is_lvalue() const } -static const char *tex_opcode_strs[] = { "tex", "txb", "txl", "txd", "txf", "txf_ms", "txs", "lod", "tg4", "query_levels" }; +static const char * const tex_opcode_strs[] = { "tex", "txb", "txl", "txd", "txf", "txf_ms", "txs", "lod", "tg4", "query_levels" }; const char *ir_texture::opcode_string() { diff --git a/mesalib/src/glsl/ir_optimization.h b/mesalib/src/glsl/ir_optimization.h index 40bb61392..c63921c26 100644 --- a/mesalib/src/glsl/ir_optimization.h +++ b/mesalib/src/glsl/ir_optimization.h @@ -38,6 +38,8 @@ #define INT_DIV_TO_MUL_RCP 0x40 #define BITFIELD_INSERT_TO_BFM_BFI 0x80 #define LDEXP_TO_ARITH 0x100 +#define CARRY_TO_ARITH 0x200 +#define BORROW_TO_ARITH 0x400 /** * \see class lower_packing_builtins_visitor @@ -112,7 +114,7 @@ bool lower_clip_distance(gl_shader *shader); void lower_output_reads(exec_list *instructions); bool lower_packing_builtins(exec_list *instructions, int op_mask); void lower_ubo_reference(struct gl_shader *shader, exec_list *instructions); -void lower_packed_varyings(void *mem_ctx, unsigned location_base, +void lower_packed_varyings(void *mem_ctx, unsigned locations_used, ir_variable_mode mode, unsigned gs_input_vertices, gl_shader *shader); bool lower_vector_insert(exec_list *instructions, bool lower_nonconstant_index); diff --git a/mesalib/src/glsl/link_varyings.cpp b/mesalib/src/glsl/link_varyings.cpp index c925c00e3..71998dfa9 100644 --- a/mesalib/src/glsl/link_varyings.cpp +++ b/mesalib/src/glsl/link_varyings.cpp @@ -172,6 +172,7 @@ cross_validate_outputs_to_inputs(struct gl_shader_program *prog, gl_shader *producer, gl_shader *consumer) { glsl_symbol_table parameters; + ir_variable *explicit_locations[MAX_VARYING] = { NULL, }; /* Find all shader outputs in the "producer" stage. */ @@ -181,7 +182,26 @@ cross_validate_outputs_to_inputs(struct gl_shader_program *prog, if ((var == NULL) || (var->data.mode != ir_var_shader_out)) continue; - parameters.add_variable(var); + if (!var->data.explicit_location + || var->data.location < VARYING_SLOT_VAR0) + parameters.add_variable(var); + else { + /* User-defined varyings with explicit locations are handled + * differently because they do not need to have matching names. + */ + const unsigned idx = var->data.location - VARYING_SLOT_VAR0; + + if (explicit_locations[idx] != NULL) { + linker_error(prog, + "%s shader has multiple outputs explicitly " + "assigned to location %d\n", + _mesa_shader_stage_to_string(producer->Stage), + idx); + return; + } + + explicit_locations[idx] = var; + } } @@ -220,7 +240,27 @@ cross_validate_outputs_to_inputs(struct gl_shader_program *prog, front_color, back_color, consumer->Stage, producer->Stage); } else { - ir_variable *const output = parameters.get_variable(input->name); + /* The rules for connecting inputs and outputs change in the presence + * of explicit locations. In this case, we no longer care about the + * names of the variables. Instead, we care only about the + * explicitly assigned location. + */ + ir_variable *output = NULL; + if (input->data.explicit_location + && input->data.location >= VARYING_SLOT_VAR0) { + output = explicit_locations[input->data.location - VARYING_SLOT_VAR0]; + + if (output == NULL) { + linker_error(prog, + "%s shader input `%s' with explicit location " + "has no matching output\n", + _mesa_shader_stage_to_string(consumer->Stage), + input->name); + } + } else { + output = parameters.get_variable(input->name); + } + if (output != NULL) { cross_validate_types_and_qualifiers(prog, input, output, consumer->Stage, producer->Stage); @@ -622,7 +662,7 @@ public: ~varying_matches(); void record(ir_variable *producer_var, ir_variable *consumer_var); unsigned assign_locations(); - void store_locations(unsigned producer_base, unsigned consumer_base) const; + void store_locations() const; private: /** @@ -648,8 +688,8 @@ private: PACKING_ORDER_VEC3, }; - static unsigned compute_packing_class(ir_variable *var); - static packing_order_enum compute_packing_order(ir_variable *var); + static unsigned compute_packing_class(const ir_variable *var); + static packing_order_enum compute_packing_order(const ir_variable *var); static int match_comparator(const void *x_generic, const void *y_generic); /** @@ -746,7 +786,10 @@ varying_matches::~varying_matches() void varying_matches::record(ir_variable *producer_var, ir_variable *consumer_var) { - if (!producer_var->data.is_unmatched_generic_inout) { + assert(producer_var != NULL || consumer_var != NULL); + + if ((producer_var && !producer_var->data.is_unmatched_generic_inout) + || (consumer_var && !consumer_var->data.is_unmatched_generic_inout)) { /* Either a location already exists for this variable (since it is part * of fixed functionality), or it has already been recorded as part of a * previous match. @@ -781,24 +824,28 @@ varying_matches::record(ir_variable *producer_var, ir_variable *consumer_var) realloc(this->matches, sizeof(*this->matches) * this->matches_capacity); } + + const ir_variable *const var = (producer_var != NULL) + ? producer_var : consumer_var; + this->matches[this->num_matches].packing_class - = this->compute_packing_class(producer_var); + = this->compute_packing_class(var); this->matches[this->num_matches].packing_order - = this->compute_packing_order(producer_var); + = this->compute_packing_order(var); if (this->disable_varying_packing) { - unsigned slots = producer_var->type->is_array() - ? (producer_var->type->length - * producer_var->type->fields.array->matrix_columns) - : producer_var->type->matrix_columns; + unsigned slots = var->type->is_array() + ? (var->type->length * var->type->fields.array->matrix_columns) + : var->type->matrix_columns; this->matches[this->num_matches].num_components = 4 * slots; } else { this->matches[this->num_matches].num_components - = producer_var->type->component_slots(); + = var->type->component_slots(); } this->matches[this->num_matches].producer_var = producer_var; this->matches[this->num_matches].consumer_var = consumer_var; this->num_matches++; - producer_var->data.is_unmatched_generic_inout = 0; + if (producer_var) + producer_var->data.is_unmatched_generic_inout = 0; if (consumer_var) consumer_var->data.is_unmatched_generic_inout = 0; } @@ -842,8 +889,7 @@ varying_matches::assign_locations() * assignments that were made by varying_matches::assign_locations(). */ void -varying_matches::store_locations(unsigned producer_base, - unsigned consumer_base) const +varying_matches::store_locations() const { for (unsigned i = 0; i < this->num_matches; i++) { ir_variable *producer_var = this->matches[i].producer_var; @@ -852,11 +898,14 @@ varying_matches::store_locations(unsigned producer_base, unsigned slot = generic_location / 4; unsigned offset = generic_location % 4; - producer_var->data.location = producer_base + slot; - producer_var->data.location_frac = offset; + if (producer_var) { + producer_var->data.location = VARYING_SLOT_VAR0 + slot; + producer_var->data.location_frac = offset; + } + if (consumer_var) { assert(consumer_var->data.location == -1); - consumer_var->data.location = consumer_base + slot; + consumer_var->data.location = VARYING_SLOT_VAR0 + slot; consumer_var->data.location_frac = offset; } } @@ -869,7 +918,7 @@ varying_matches::store_locations(unsigned producer_base, * be safely backed into the same vec4. */ unsigned -varying_matches::compute_packing_class(ir_variable *var) +varying_matches::compute_packing_class(const ir_variable *var) { /* Without help from the back-end, there is no way to pack together * variables with different interpolation types, because @@ -900,7 +949,7 @@ varying_matches::compute_packing_class(ir_variable *var) * other varyings in the same packing class. */ varying_matches::packing_order_enum -varying_matches::compute_packing_order(ir_variable *var) +varying_matches::compute_packing_order(const ir_variable *var) { const glsl_type *element_type = var->type; @@ -943,7 +992,7 @@ varying_matches::match_comparator(const void *x_generic, const void *y_generic) * varyings, but excludes variables such as gl_FrontFacing and gl_FragCoord. */ static bool -is_varying_var(gl_shader_stage stage, const ir_variable *var) +var_counts_against_varying_limit(gl_shader_stage stage, const ir_variable *var) { /* Only fragment shaders will take a varying variable as an input */ if (stage == MESA_SHADER_FRAGMENT && @@ -1037,6 +1086,157 @@ private: }; +namespace linker { + +bool +populate_consumer_input_sets(void *mem_ctx, exec_list *ir, + hash_table *consumer_inputs, + hash_table *consumer_interface_inputs, + ir_variable *consumer_inputs_with_locations[VARYING_SLOT_MAX]) +{ + memset(consumer_inputs_with_locations, + 0, + sizeof(consumer_inputs_with_locations[0]) * VARYING_SLOT_MAX); + + foreach_list(node, ir) { + ir_variable *const input_var = ((ir_instruction *) node)->as_variable(); + + if ((input_var != NULL) && (input_var->data.mode == ir_var_shader_in)) { + if (input_var->type->is_interface()) + return false; + + if (input_var->data.explicit_location) { + /* assign_varying_locations only cares about finding the + * ir_variable at the start of a contiguous location block. + * + * - For !producer, consumer_inputs_with_locations isn't used. + * + * - For !consumer, consumer_inputs_with_locations is empty. + * + * For consumer && producer, if you were trying to set some + * ir_variable to the middle of a location block on the other side + * of producer/consumer, cross_validate_outputs_to_inputs() should + * be link-erroring due to either type mismatch or location + * overlaps. If the variables do match up, then they've got a + * matching data.location and you only looked at + * consumer_inputs_with_locations[var->data.location], not any + * following entries for the array/structure. + */ + consumer_inputs_with_locations[input_var->data.location] = + input_var; + } else if (input_var->get_interface_type() != NULL) { + char *const iface_field_name = + ralloc_asprintf(mem_ctx, "%s.%s", + input_var->get_interface_type()->name, + input_var->name); + hash_table_insert(consumer_interface_inputs, input_var, + iface_field_name); + } else { + hash_table_insert(consumer_inputs, input_var, + ralloc_strdup(mem_ctx, input_var->name)); + } + } + } + + return true; +} + +/** + * Find a variable from the consumer that "matches" the specified variable + * + * This function only finds inputs with names that match. There is no + * validation (here) that the types, etc. are compatible. + */ +ir_variable * +get_matching_input(void *mem_ctx, + const ir_variable *output_var, + hash_table *consumer_inputs, + hash_table *consumer_interface_inputs, + ir_variable *consumer_inputs_with_locations[VARYING_SLOT_MAX]) +{ + ir_variable *input_var; + + if (output_var->data.explicit_location) { + input_var = consumer_inputs_with_locations[output_var->data.location]; + } else if (output_var->get_interface_type() != NULL) { + char *const iface_field_name = + ralloc_asprintf(mem_ctx, "%s.%s", + output_var->get_interface_type()->name, + output_var->name); + input_var = + (ir_variable *) hash_table_find(consumer_interface_inputs, + iface_field_name); + } else { + input_var = + (ir_variable *) hash_table_find(consumer_inputs, output_var->name); + } + + return (input_var == NULL || input_var->data.mode != ir_var_shader_in) + ? NULL : input_var; +} + +} + +static int +io_variable_cmp(const void *_a, const void *_b) +{ + const ir_variable *const a = *(const ir_variable **) _a; + const ir_variable *const b = *(const ir_variable **) _b; + + if (a->data.explicit_location && b->data.explicit_location) + return b->data.location - a->data.location; + + if (a->data.explicit_location && !b->data.explicit_location) + return 1; + + if (!a->data.explicit_location && b->data.explicit_location) + return -1; + + return -strcmp(a->name, b->name); +} + +/** + * Sort the shader IO variables into canonical order + */ +static void +canonicalize_shader_io(exec_list *ir, enum ir_variable_mode io_mode) +{ + ir_variable *var_table[MAX_PROGRAM_OUTPUTS * 4]; + unsigned num_variables = 0; + + foreach_list(node, ir) { + ir_variable *const var = ((ir_instruction *) node)->as_variable(); + + if (var == NULL || var->data.mode != io_mode) + continue; + + /* If we have already encountered more I/O variables that could + * successfully link, bail. + */ + if (num_variables == ARRAY_SIZE(var_table)) + return; + + var_table[num_variables++] = var; + } + + if (num_variables == 0) + return; + + /* Sort the list in reverse order (io_variable_cmp handles this). Later + * we're going to push the variables on to the IR list as a stack, so we + * want the last variable (in canonical order) to be first in the list. + */ + qsort(var_table, num_variables, sizeof(var_table[0]), io_variable_cmp); + + /* Remove the variable from it's current location in the IR, and put it at + * the front. + */ + for (unsigned i = 0; i < num_variables; i++) { + var_table[i]->remove(); + ir->push_head(var_table[i]); + } +} + /** * Assign locations for all variables that are produced in one pipeline stage * (the "producer") and consumed in the next stage (the "consumer"). @@ -1069,8 +1269,6 @@ assign_varying_locations(struct gl_context *ctx, tfeedback_decl *tfeedback_decls, unsigned gs_input_vertices) { - const unsigned producer_base = VARYING_SLOT_VAR0; - const unsigned consumer_base = VARYING_SLOT_VAR0; varying_matches matches(ctx->Const.DisableVaryingPacking, consumer && consumer->Stage == MESA_SHADER_FRAGMENT); hash_table *tfeedback_candidates @@ -1079,67 +1277,85 @@ assign_varying_locations(struct gl_context *ctx, = hash_table_ctor(0, hash_table_string_hash, hash_table_string_compare); hash_table *consumer_interface_inputs = hash_table_ctor(0, hash_table_string_hash, hash_table_string_compare); + ir_variable *consumer_inputs_with_locations[VARYING_SLOT_MAX] = { + NULL, + }; - /* Operate in a total of three passes. + /* Operate in a total of four passes. + * + * 1. Sort inputs / outputs into a canonical order. This is necessary so + * that inputs / outputs of separable shaders will be assigned + * predictable locations regardless of the order in which declarations + * appeared in the shader source. * - * 1. Assign locations for any matching inputs and outputs. + * 2. Assign locations for any matching inputs and outputs. * - * 2. Mark output variables in the producer that do not have locations as + * 3. Mark output variables in the producer that do not have locations as * not being outputs. This lets the optimizer eliminate them. * - * 3. Mark input variables in the consumer that do not have locations as + * 4. Mark input variables in the consumer that do not have locations as * not being inputs. This lets the optimizer eliminate them. */ + if (consumer) + canonicalize_shader_io(consumer->ir, ir_var_shader_in); + + if (producer) + canonicalize_shader_io(producer->ir, ir_var_shader_out); + + if (consumer + && !linker::populate_consumer_input_sets(mem_ctx, + consumer->ir, + consumer_inputs, + consumer_interface_inputs, + consumer_inputs_with_locations)) { + assert(!"populate_consumer_input_sets failed"); + hash_table_dtor(tfeedback_candidates); + hash_table_dtor(consumer_inputs); + hash_table_dtor(consumer_interface_inputs); + return false; + } - if (consumer) { - foreach_list(node, consumer->ir) { - ir_variable *const input_var = + if (producer) { + foreach_list(node, producer->ir) { + ir_variable *const output_var = ((ir_instruction *) node)->as_variable(); - if ((input_var != NULL) && (input_var->data.mode == ir_var_shader_in)) { - if (input_var->get_interface_type() != NULL) { - char *const iface_field_name = - ralloc_asprintf(mem_ctx, "%s.%s", - input_var->get_interface_type()->name, - input_var->name); - hash_table_insert(consumer_interface_inputs, input_var, - iface_field_name); - } else { - hash_table_insert(consumer_inputs, input_var, - ralloc_strdup(mem_ctx, input_var->name)); - } - } - } - } - - foreach_list(node, producer->ir) { - ir_variable *const output_var = ((ir_instruction *) node)->as_variable(); + if ((output_var == NULL) || + (output_var->data.mode != ir_var_shader_out)) + continue; - if ((output_var == NULL) || (output_var->data.mode != ir_var_shader_out)) - continue; + tfeedback_candidate_generator g(mem_ctx, tfeedback_candidates); + g.process(output_var); - tfeedback_candidate_generator g(mem_ctx, tfeedback_candidates); - g.process(output_var); - - ir_variable *input_var; - if (output_var->get_interface_type() != NULL) { - char *const iface_field_name = - ralloc_asprintf(mem_ctx, "%s.%s", - output_var->get_interface_type()->name, - output_var->name); - input_var = - (ir_variable *) hash_table_find(consumer_interface_inputs, - iface_field_name); - } else { - input_var = - (ir_variable *) hash_table_find(consumer_inputs, output_var->name); + ir_variable *const input_var = + linker::get_matching_input(mem_ctx, output_var, consumer_inputs, + consumer_interface_inputs, + consumer_inputs_with_locations); + + /* If a matching input variable was found, add this ouptut (and the + * input) to the set. If this is a separable program and there is no + * consumer stage, add the output. + */ + if (input_var || (prog->SeparateShader && consumer == NULL)) { + matches.record(output_var, input_var); + } } + } else { + /* If there's no producer stage, then this must be a separable program. + * For example, we may have a program that has just a fragment shader. + * Later this program will be used with some arbitrary vertex (or + * geometry) shader program. This means that locations must be assigned + * for all the inputs. + */ + foreach_list(node, consumer->ir) { + ir_variable *const input_var = + ((ir_instruction *) node)->as_variable(); - if (input_var && input_var->data.mode != ir_var_shader_in) - input_var = NULL; + if ((input_var == NULL) || + (input_var->data.mode != ir_var_shader_in)) + continue; - if (input_var) { - matches.record(output_var, input_var); + matches.record(NULL, input_var); } } @@ -1162,7 +1378,7 @@ assign_varying_locations(struct gl_context *ctx, } const unsigned slots_used = matches.assign_locations(); - matches.store_locations(producer_base, consumer_base); + matches.store_locations(); for (unsigned i = 0; i < num_tfeedback_decls; ++i) { if (!tfeedback_decls[i].is_varying()) @@ -1187,15 +1403,17 @@ assign_varying_locations(struct gl_context *ctx, */ assert(!ctx->Extensions.EXT_transform_feedback); } else { - lower_packed_varyings(mem_ctx, producer_base, slots_used, - ir_var_shader_out, 0, producer); + if (producer) { + lower_packed_varyings(mem_ctx, slots_used, ir_var_shader_out, + 0, producer); + } if (consumer) { - lower_packed_varyings(mem_ctx, consumer_base, slots_used, - ir_var_shader_in, gs_input_vertices, consumer); + lower_packed_varyings(mem_ctx, slots_used, ir_var_shader_in, + gs_input_vertices, consumer); } } - if (consumer) { + if (consumer && producer) { foreach_list(node, consumer->ir) { ir_variable *const var = ((ir_instruction *) node)->as_variable(); @@ -1244,7 +1462,7 @@ check_against_output_limit(struct gl_context *ctx, ir_variable *const var = ((ir_instruction *) node)->as_variable(); if (var && var->data.mode == ir_var_shader_out && - is_varying_var(producer->Stage, var)) { + var_counts_against_varying_limit(producer->Stage, var)) { output_vectors += var->type->count_attribute_slots(); } } @@ -1283,7 +1501,7 @@ check_against_input_limit(struct gl_context *ctx, ir_variable *const var = ((ir_instruction *) node)->as_variable(); if (var && var->data.mode == ir_var_shader_in && - is_varying_var(consumer->Stage, var)) { + var_counts_against_varying_limit(consumer->Stage, var)) { input_vectors += var->type->count_attribute_slots(); } } diff --git a/mesalib/src/glsl/linker.cpp b/mesalib/src/glsl/linker.cpp index c2f7f4863..a43d23082 100644 --- a/mesalib/src/glsl/linker.cpp +++ b/mesalib/src/glsl/linker.cpp @@ -1195,6 +1195,83 @@ private: }; /** + * Performs the cross-validation of layout qualifiers specified in + * redeclaration of gl_FragCoord for the attached fragment shaders, + * and propagates them to the linked FS and linked shader program. + */ +static void +link_fs_input_layout_qualifiers(struct gl_shader_program *prog, + struct gl_shader *linked_shader, + struct gl_shader **shader_list, + unsigned num_shaders) +{ + linked_shader->redeclares_gl_fragcoord = false; + linked_shader->uses_gl_fragcoord = false; + linked_shader->origin_upper_left = false; + linked_shader->pixel_center_integer = false; + + if (linked_shader->Stage != MESA_SHADER_FRAGMENT || + (prog->Version < 150 && !prog->ARB_fragment_coord_conventions_enable)) + return; + + for (unsigned i = 0; i < num_shaders; i++) { + struct gl_shader *shader = shader_list[i]; + /* From the GLSL 1.50 spec, page 39: + * + * "If gl_FragCoord is redeclared in any fragment shader in a program, + * it must be redeclared in all the fragment shaders in that program + * that have a static use gl_FragCoord." + * + * Exclude the case when one of the 'linked_shader' or 'shader' redeclares + * gl_FragCoord with no layout qualifiers but the other one doesn't + * redeclare it. If we strictly follow GLSL 1.50 spec's language, it + * should be a link error. But, generating link error for this case will + * be a wrong behaviour which spec didn't intend to do and it could also + * break some applications. + */ + if ((linked_shader->redeclares_gl_fragcoord + && !shader->redeclares_gl_fragcoord + && shader->uses_gl_fragcoord + && (linked_shader->origin_upper_left + || linked_shader->pixel_center_integer)) + || (shader->redeclares_gl_fragcoord + && !linked_shader->redeclares_gl_fragcoord + && linked_shader->uses_gl_fragcoord + && (shader->origin_upper_left + || shader->pixel_center_integer))) { + linker_error(prog, "fragment shader defined with conflicting " + "layout qualifiers for gl_FragCoord\n"); + } + + /* From the GLSL 1.50 spec, page 39: + * + * "All redeclarations of gl_FragCoord in all fragment shaders in a + * single program must have the same set of qualifiers." + */ + if (linked_shader->redeclares_gl_fragcoord && shader->redeclares_gl_fragcoord + && (shader->origin_upper_left != linked_shader->origin_upper_left + || shader->pixel_center_integer != linked_shader->pixel_center_integer)) { + linker_error(prog, "fragment shader defined with conflicting " + "layout qualifiers for gl_FragCoord\n"); + } + + /* Update the linked shader state. Note that uses_gl_fragcoord should + * accumulate the results. The other values should replace. If there + * are multiple redeclarations, all the fields except uses_gl_fragcoord + * are already known to be the same. + */ + if (shader->redeclares_gl_fragcoord || shader->uses_gl_fragcoord) { + linked_shader->redeclares_gl_fragcoord = + shader->redeclares_gl_fragcoord; + linked_shader->uses_gl_fragcoord = linked_shader->uses_gl_fragcoord + || shader->uses_gl_fragcoord; + linked_shader->origin_upper_left = shader->origin_upper_left; + linked_shader->pixel_center_integer = shader->pixel_center_integer; + } + } +} + +/** * Performs the cross-validation of geometry shader max_vertices and * primitive type layout qualifiers for the attached geometry shaders, * and propagates them to the linked GS and linked shader program. @@ -1471,6 +1548,7 @@ link_intrastage_shaders(void *mem_ctx, linked->NumUniformBlocks = num_uniform_blocks; ralloc_steal(linked, linked->UniformBlocks); + link_fs_input_layout_qualifiers(prog, linked, shader_list, num_shaders); link_gs_inout_layout_qualifiers(prog, linked, shader_list, num_shaders); link_cs_input_layout_qualifiers(prog, linked, shader_list, num_shaders); @@ -1798,10 +1876,12 @@ assign_attribute_or_color_locations(gl_shader_program *prog, * active attribute array, both of which require multiple * contiguous generic attributes." * - * Previous versions of the spec contain similar language but omit - * the bit about attribute arrays. + * I think above text prohibits the aliasing of explicit and + * automatic assignments. But, aliasing is allowed in manual + * assignments of attribute locations. See below comments for + * the details. * - * Page 61 of the OpenGL 4.0 spec also says: + * From OpenGL 4.0 spec, page 61: * * "It is possible for an application to bind more than one * attribute name to the same location. This is referred to as @@ -1814,29 +1894,84 @@ assign_attribute_or_color_locations(gl_shader_program *prog, * but implementations are not required to generate an error * in this case." * - * These two paragraphs are either somewhat contradictory, or I - * don't fully understand one or both of them. - */ - /* FINISHME: The code as currently written does not support - * FINISHME: attribute location aliasing (see comment above). + * From GLSL 4.30 spec, page 54: + * + * "A program will fail to link if any two non-vertex shader + * input variables are assigned to the same location. For + * vertex shaders, multiple input variables may be assigned + * to the same location using either layout qualifiers or via + * the OpenGL API. However, such aliasing is intended only to + * support vertex shaders where each execution path accesses + * at most one input per each location. Implementations are + * permitted, but not required, to generate link-time errors + * if they detect that every path through the vertex shader + * executable accesses multiple inputs assigned to any single + * location. For all shader types, a program will fail to link + * if explicit location assignments leave the linker unable + * to find space for other variables without explicit + * assignments." + * + * From OpenGL ES 3.0 spec, page 56: + * + * "Binding more than one attribute name to the same location + * is referred to as aliasing, and is not permitted in OpenGL + * ES Shading Language 3.00 vertex shaders. LinkProgram will + * fail when this condition exists. However, aliasing is + * possible in OpenGL ES Shading Language 1.00 vertex shaders. + * This will only work if only one of the aliased attributes + * is active in the executable program, or if no path through + * the shader consumes more than one attribute of a set of + * attributes aliased to the same location. A link error can + * occur if the linker determines that every path through the + * shader consumes multiple aliased attributes, but implemen- + * tations are not required to generate an error in this case." + * + * After looking at above references from OpenGL, OpenGL ES and + * GLSL specifications, we allow aliasing of vertex input variables + * in: OpenGL 2.0 (and above) and OpenGL ES 2.0. + * + * NOTE: This is not required by the spec but its worth mentioning + * here that we're not doing anything to make sure that no path + * through the vertex shader executable accesses multiple inputs + * assigned to any single location. */ + /* Mask representing the contiguous slots that will be used by * this attribute. */ const unsigned attr = var->data.location - generic_base; const unsigned use_mask = (1 << slots) - 1; + const char *const string = (target_index == MESA_SHADER_VERTEX) + ? "vertex shader input" : "fragment shader output"; + + /* Generate a link error if the requested locations for this + * attribute exceed the maximum allowed attribute location. + */ + if (attr + slots > max_index) { + linker_error(prog, + "insufficient contiguous locations " + "available for %s `%s' %d %d %d", string, + var->name, used_locations, use_mask, attr); + return false; + } /* Generate a link error if the set of bits requested for this * attribute overlaps any previously allocated bits. */ if ((~(use_mask << attr) & used_locations) != used_locations) { - const char *const string = (target_index == MESA_SHADER_VERTEX) - ? "vertex shader input" : "fragment shader output"; - linker_error(prog, - "insufficient contiguous locations " - "available for %s `%s' %d %d %d", string, - var->name, used_locations, use_mask, attr); - return false; + if (target_index == MESA_SHADER_FRAGMENT || + (prog->IsES && prog->Version >= 300)) { + linker_error(prog, + "overlapping location is assigned " + "to %s `%s' %d %d %d\n", string, + var->name, used_locations, use_mask, attr); + return false; + } else { + linker_warning(prog, + "overlapping location is assigned " + "to %s `%s' %d %d %d\n", string, + var->name, used_locations, use_mask, attr); + } } used_locations |= (use_mask << attr); @@ -2115,6 +2250,7 @@ link_shaders(struct gl_context *ctx, struct gl_shader_program *prog) ralloc_free(prog->AtomicBuffers); prog->AtomicBuffers = NULL; prog->NumAtomicBuffers = 0; + prog->ARB_fragment_coord_conventions_enable = false; /* Separate the shaders into groups based on their type. */ @@ -2141,6 +2277,9 @@ link_shaders(struct gl_context *ctx, struct gl_shader_program *prog) goto done; } + prog->ARB_fragment_coord_conventions_enable |= + prog->Shaders[i]->ARB_fragment_coord_conventions_enable; + gl_shader_stage shader_type = prog->Shaders[i]->Stage; shader_list[shader_type][num_shaders[shader_type]] = prog->Shaders[i]; num_shaders[shader_type]++; @@ -2161,7 +2300,8 @@ link_shaders(struct gl_context *ctx, struct gl_shader_program *prog) /* Geometry shaders have to be linked with vertex shaders. */ if (num_shaders[MESA_SHADER_GEOMETRY] > 0 && - num_shaders[MESA_SHADER_VERTEX] == 0) { + num_shaders[MESA_SHADER_VERTEX] == 0 && + !prog->SeparateShader) { linker_error(prog, "Geometry shader must be linked with " "vertex shader\n"); goto done; @@ -2363,7 +2503,7 @@ link_shaders(struct gl_context *ctx, struct gl_shader_program *prog) if (last >= 0 && last < MESA_SHADER_FRAGMENT) { gl_shader *const sh = prog->_LinkedShaders[last]; - if (num_tfeedback_decls != 0) { + if (num_tfeedback_decls != 0 || prog->SeparateShader) { /* There was no fragment shader, but we still have to assign varying * locations for use by transform feedback. */ @@ -2377,7 +2517,8 @@ link_shaders(struct gl_context *ctx, struct gl_shader_program *prog) do_dead_builtin_varyings(ctx, sh, NULL, num_tfeedback_decls, tfeedback_decls); - demote_shader_inputs_and_outputs(sh, ir_var_shader_out); + if (!prog->SeparateShader) + demote_shader_inputs_and_outputs(sh, ir_var_shader_out); /* Eliminate code that is now dead due to unused outputs being demoted. */ @@ -2392,7 +2533,16 @@ link_shaders(struct gl_context *ctx, struct gl_shader_program *prog) do_dead_builtin_varyings(ctx, NULL, sh, num_tfeedback_decls, tfeedback_decls); - demote_shader_inputs_and_outputs(sh, ir_var_shader_in); + if (prog->SeparateShader) { + if (!assign_varying_locations(ctx, mem_ctx, prog, + NULL /* producer */, + sh /* consumer */, + 0 /* num_tfeedback_decls */, + NULL /* tfeedback_decls */, + 0 /* gs_input_vertices */)) + goto done; + } else + demote_shader_inputs_and_outputs(sh, ir_var_shader_in); while (do_dead_code(sh->ir, false)) ; @@ -2457,7 +2607,7 @@ link_shaders(struct gl_context *ctx, struct gl_shader_program *prog) * fragment shader) is absent. So, the extension shouldn't change the * behavior specified in GLSL specification. */ - if (!prog->InternalSeparateShader && ctx->API == API_OPENGLES2) { + if (!prog->SeparateShader && ctx->API == API_OPENGLES2) { if (prog->_LinkedShaders[MESA_SHADER_VERTEX] == NULL) { linker_error(prog, "program lacks a vertex shader\n"); } else if (prog->_LinkedShaders[MESA_SHADER_FRAGMENT] == NULL) { diff --git a/mesalib/src/glsl/lower_instructions.cpp b/mesalib/src/glsl/lower_instructions.cpp index 49316d002..176070c87 100644 --- a/mesalib/src/glsl/lower_instructions.cpp +++ b/mesalib/src/glsl/lower_instructions.cpp @@ -39,6 +39,8 @@ * - MOD_TO_FRACT * - LDEXP_TO_ARITH * - BITFIELD_INSERT_TO_BFM_BFI + * - CARRY_TO_ARITH + * - BORROW_TO_ARITH * * SUB_TO_ADD_NEG: * --------------- @@ -94,6 +96,14 @@ * Many GPUs implement the bitfieldInsert() built-in from ARB_gpu_shader_5 * with a pair of instructions. * + * CARRY_TO_ARITH: + * --------------- + * Converts ir_carry into (x + y) < x. + * + * BORROW_TO_ARITH: + * ---------------- + * Converts ir_borrow into (x < y). + * */ #include "main/core.h" /* for M_LOG2E */ @@ -127,6 +137,8 @@ private: void log_to_log2(ir_expression *); void bitfield_insert_to_bfm_bfi(ir_expression *); void ldexp_to_arith(ir_expression *); + void carry_to_arith(ir_expression *); + void borrow_to_arith(ir_expression *); }; } /* anonymous namespace */ @@ -436,6 +448,42 @@ lower_instructions_visitor::ldexp_to_arith(ir_expression *ir) this->progress = true; } +void +lower_instructions_visitor::carry_to_arith(ir_expression *ir) +{ + /* Translates + * ir_binop_carry x y + * into + * sum = ir_binop_add x y + * bcarry = ir_binop_less sum x + * carry = ir_unop_b2i bcarry + */ + + ir_rvalue *x_clone = ir->operands[0]->clone(ir, NULL); + ir->operation = ir_unop_i2u; + ir->operands[0] = b2i(less(add(ir->operands[0], ir->operands[1]), x_clone)); + ir->operands[1] = NULL; + + this->progress = true; +} + +void +lower_instructions_visitor::borrow_to_arith(ir_expression *ir) +{ + /* Translates + * ir_binop_borrow x y + * into + * bcarry = ir_binop_less x y + * carry = ir_unop_b2i bcarry + */ + + ir->operation = ir_unop_i2u; + ir->operands[0] = b2i(less(ir->operands[0], ir->operands[1])); + ir->operands[1] = NULL; + + this->progress = true; +} + ir_visitor_status lower_instructions_visitor::visit_leave(ir_expression *ir) { @@ -482,6 +530,16 @@ lower_instructions_visitor::visit_leave(ir_expression *ir) ldexp_to_arith(ir); break; + case ir_binop_carry: + if (lowering(CARRY_TO_ARITH)) + carry_to_arith(ir); + break; + + case ir_binop_borrow: + if (lowering(BORROW_TO_ARITH)) + borrow_to_arith(ir); + break; + default: return visit_continue; } diff --git a/mesalib/src/glsl/lower_packed_varyings.cpp b/mesalib/src/glsl/lower_packed_varyings.cpp index 8c1b8850b..e8654748f 100644 --- a/mesalib/src/glsl/lower_packed_varyings.cpp +++ b/mesalib/src/glsl/lower_packed_varyings.cpp @@ -160,8 +160,7 @@ namespace { class lower_packed_varyings_visitor { public: - lower_packed_varyings_visitor(void *mem_ctx, unsigned location_base, - unsigned locations_used, + lower_packed_varyings_visitor(void *mem_ctx, unsigned locations_used, ir_variable_mode mode, unsigned gs_input_vertices, exec_list *out_instructions); @@ -190,18 +189,10 @@ private: void * const mem_ctx; /** - * Location representing the first generic varying slot for this shader - * stage (e.g. VARYING_SLOT_VAR0 if we are packing vertex shader outputs). - * Varyings whose location is less than this value are assumed to - * correspond to special fixed function hardware, so they are not lowered. - */ - const unsigned location_base; - - /** * Number of generic varying slots which are used by this shader. This is * used to allocate temporary intermediate data structures. If any varying * used by this shader has a location greater than or equal to - * location_base + locations_used, an assertion will fire. + * VARYING_SLOT_VAR0 + locations_used, an assertion will fire. */ const unsigned locations_used; @@ -235,11 +226,9 @@ private: } /* anonymous namespace */ lower_packed_varyings_visitor::lower_packed_varyings_visitor( - void *mem_ctx, unsigned location_base, unsigned locations_used, - ir_variable_mode mode, unsigned gs_input_vertices, - exec_list *out_instructions) + void *mem_ctx, unsigned locations_used, ir_variable_mode mode, + unsigned gs_input_vertices, exec_list *out_instructions) : mem_ctx(mem_ctx), - location_base(location_base), locations_used(locations_used), packed_varyings((ir_variable **) rzalloc_array_size(mem_ctx, sizeof(*packed_varyings), @@ -259,7 +248,7 @@ lower_packed_varyings_visitor::run(exec_list *instructions) continue; if (var->data.mode != this->mode || - var->data.location < (int) this->location_base || + var->data.location < VARYING_SLOT_VAR0 || !this->needs_lowering(var)) continue; @@ -542,7 +531,7 @@ lower_packed_varyings_visitor::get_packed_varying_deref( unsigned location, ir_variable *unpacked_var, const char *name, unsigned vertex_index) { - unsigned slot = location - this->location_base; + unsigned slot = location - VARYING_SLOT_VAR0; assert(slot < locations_used); if (this->packed_varyings[slot] == NULL) { char *packed_name = ralloc_asprintf(this->mem_ctx, "packed:%s", name); @@ -595,7 +584,12 @@ lower_packed_varyings_visitor::get_packed_varying_deref( bool lower_packed_varyings_visitor::needs_lowering(ir_variable *var) { - /* Things composed of vec4's don't need lowering. Everything else does. */ + /* Things composed of vec4's and varyings with explicitly assigned + * locations don't need lowering. Everything else does. + */ + if (var->data.explicit_location) + return false; + const glsl_type *type = var->type; if (this->gs_input_vertices != 0) { assert(type->is_array()); @@ -654,9 +648,9 @@ lower_packed_varyings_gs_splicer::visit(ir_emit_vertex *ev) void -lower_packed_varyings(void *mem_ctx, unsigned location_base, - unsigned locations_used, ir_variable_mode mode, - unsigned gs_input_vertices, gl_shader *shader) +lower_packed_varyings(void *mem_ctx, unsigned locations_used, + ir_variable_mode mode, unsigned gs_input_vertices, + gl_shader *shader) { exec_list *instructions = shader->ir; ir_function *main_func = shader->symbols->get_function("main"); @@ -664,8 +658,7 @@ lower_packed_varyings(void *mem_ctx, unsigned location_base, ir_function_signature *main_func_sig = main_func->matching_signature(NULL, &void_parameters); exec_list new_instructions; - lower_packed_varyings_visitor visitor(mem_ctx, location_base, - locations_used, mode, + lower_packed_varyings_visitor visitor(mem_ctx, locations_used, mode, gs_input_vertices, &new_instructions); visitor.run(instructions); if (mode == ir_var_shader_out) { diff --git a/mesalib/src/glsl/main.cpp b/mesalib/src/glsl/main.cpp index 0d8c01f6a..6c229465c 100644 --- a/mesalib/src/glsl/main.cpp +++ b/mesalib/src/glsl/main.cpp @@ -51,6 +51,12 @@ static int glsl_version = 330; +extern "C" void +_mesa_error_no_memory(const char *caller) +{ + fprintf(stderr, "Mesa error: out of memory in %s", caller); +} + static void initialize_context(struct gl_context *ctx, gl_api api) { diff --git a/mesalib/src/glsl/opt_dead_builtin_varyings.cpp b/mesalib/src/glsl/opt_dead_builtin_varyings.cpp index c2a306e7b..6612592aa 100644 --- a/mesalib/src/glsl/opt_dead_builtin_varyings.cpp +++ b/mesalib/src/glsl/opt_dead_builtin_varyings.cpp @@ -518,14 +518,9 @@ do_dead_builtin_varyings(struct gl_context *ctx, /* Lowering of built-in varyings has no effect with the core context and * GLES2, because they are not available there. - * - * EXT_separate_shader_objects doesn't allow this optimization, - * because a program object can be bound partially (e.g. only one - * stage of a program object can be bound). */ if (ctx->API == API_OPENGL_CORE || - ctx->API == API_OPENGLES2 || - ctx->Extensions.EXT_separate_shader_objects) { + ctx->API == API_OPENGLES2) { return; } |