diff options
Diffstat (limited to 'src/glsl/nir')
| -rw-r--r-- | src/glsl/nir/builtin_type_macros.h | 172 | ||||
| -rw-r--r-- | src/glsl/nir/glsl_to_nir.cpp | 98 | ||||
| -rw-r--r-- | src/glsl/nir/glsl_types.cpp | 1850 | ||||
| -rw-r--r-- | src/glsl/nir/glsl_types.h | 890 | ||||
| -rw-r--r-- | src/glsl/nir/nir.c | 108 | ||||
| -rw-r--r-- | src/glsl/nir/nir.h | 59 | ||||
| -rw-r--r-- | src/glsl/nir/nir_constant_expressions.py | 1 | ||||
| -rw-r--r-- | src/glsl/nir/nir_instr_set.c | 519 | ||||
| -rw-r--r-- | src/glsl/nir/nir_instr_set.h | 62 | ||||
| -rw-r--r-- | src/glsl/nir/nir_intrinsics.h | 6 | ||||
| -rw-r--r-- | src/glsl/nir/nir_lower_atomics.c | 22 | ||||
| -rw-r--r-- | src/glsl/nir/nir_opt_cse.c | 293 | ||||
| -rw-r--r-- | src/glsl/nir/nir_sweep.c | 2 | ||||
| -rw-r--r-- | src/glsl/nir/nir_types.cpp | 6 | ||||
| -rw-r--r-- | src/glsl/nir/nir_types.h | 4 | ||||
| -rw-r--r-- | src/glsl/nir/shader_enums.c | 213 | ||||
| -rw-r--r-- | src/glsl/nir/shader_enums.h | 500 |
17 files changed, 4453 insertions, 352 deletions
diff --git a/src/glsl/nir/builtin_type_macros.h b/src/glsl/nir/builtin_type_macros.h new file mode 100644 index 00000000000..8e16ae45489 --- /dev/null +++ b/src/glsl/nir/builtin_type_macros.h @@ -0,0 +1,172 @@ +/* + * Copyright © 2013 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 builtin_type_macros.h + * + * This contains definitions for all GLSL built-in types, regardless of what + * language version or extension might provide them. + */ + +#include "glsl_types.h" + +DECL_TYPE(error, GL_INVALID_ENUM, GLSL_TYPE_ERROR, 0, 0) +DECL_TYPE(void, GL_INVALID_ENUM, GLSL_TYPE_VOID, 0, 0) + +DECL_TYPE(bool, GL_BOOL, GLSL_TYPE_BOOL, 1, 1) +DECL_TYPE(bvec2, GL_BOOL_VEC2, GLSL_TYPE_BOOL, 2, 1) +DECL_TYPE(bvec3, GL_BOOL_VEC3, GLSL_TYPE_BOOL, 3, 1) +DECL_TYPE(bvec4, GL_BOOL_VEC4, GLSL_TYPE_BOOL, 4, 1) + +DECL_TYPE(int, GL_INT, GLSL_TYPE_INT, 1, 1) +DECL_TYPE(ivec2, GL_INT_VEC2, GLSL_TYPE_INT, 2, 1) +DECL_TYPE(ivec3, GL_INT_VEC3, GLSL_TYPE_INT, 3, 1) +DECL_TYPE(ivec4, GL_INT_VEC4, GLSL_TYPE_INT, 4, 1) + +DECL_TYPE(uint, GL_UNSIGNED_INT, GLSL_TYPE_UINT, 1, 1) +DECL_TYPE(uvec2, GL_UNSIGNED_INT_VEC2, GLSL_TYPE_UINT, 2, 1) +DECL_TYPE(uvec3, GL_UNSIGNED_INT_VEC3, GLSL_TYPE_UINT, 3, 1) +DECL_TYPE(uvec4, GL_UNSIGNED_INT_VEC4, GLSL_TYPE_UINT, 4, 1) + +DECL_TYPE(float, GL_FLOAT, GLSL_TYPE_FLOAT, 1, 1) +DECL_TYPE(vec2, GL_FLOAT_VEC2, GLSL_TYPE_FLOAT, 2, 1) +DECL_TYPE(vec3, GL_FLOAT_VEC3, GLSL_TYPE_FLOAT, 3, 1) +DECL_TYPE(vec4, GL_FLOAT_VEC4, GLSL_TYPE_FLOAT, 4, 1) + +DECL_TYPE(mat2, GL_FLOAT_MAT2, GLSL_TYPE_FLOAT, 2, 2) +DECL_TYPE(mat3, GL_FLOAT_MAT3, GLSL_TYPE_FLOAT, 3, 3) +DECL_TYPE(mat4, GL_FLOAT_MAT4, GLSL_TYPE_FLOAT, 4, 4) + +DECL_TYPE(mat2x3, GL_FLOAT_MAT2x3, GLSL_TYPE_FLOAT, 3, 2) +DECL_TYPE(mat2x4, GL_FLOAT_MAT2x4, GLSL_TYPE_FLOAT, 4, 2) +DECL_TYPE(mat3x2, GL_FLOAT_MAT3x2, GLSL_TYPE_FLOAT, 2, 3) +DECL_TYPE(mat3x4, GL_FLOAT_MAT3x4, GLSL_TYPE_FLOAT, 4, 3) +DECL_TYPE(mat4x2, GL_FLOAT_MAT4x2, GLSL_TYPE_FLOAT, 2, 4) +DECL_TYPE(mat4x3, GL_FLOAT_MAT4x3, GLSL_TYPE_FLOAT, 3, 4) + +DECL_TYPE(double, GL_DOUBLE, GLSL_TYPE_DOUBLE, 1, 1) +DECL_TYPE(dvec2, GL_DOUBLE_VEC2, GLSL_TYPE_DOUBLE, 2, 1) +DECL_TYPE(dvec3, GL_DOUBLE_VEC3, GLSL_TYPE_DOUBLE, 3, 1) +DECL_TYPE(dvec4, GL_DOUBLE_VEC4, GLSL_TYPE_DOUBLE, 4, 1) + +DECL_TYPE(dmat2, GL_DOUBLE_MAT2, GLSL_TYPE_DOUBLE, 2, 2) +DECL_TYPE(dmat3, GL_DOUBLE_MAT3, GLSL_TYPE_DOUBLE, 3, 3) +DECL_TYPE(dmat4, GL_DOUBLE_MAT4, GLSL_TYPE_DOUBLE, 4, 4) + +DECL_TYPE(dmat2x3, GL_DOUBLE_MAT2x3, GLSL_TYPE_DOUBLE, 3, 2) +DECL_TYPE(dmat2x4, GL_DOUBLE_MAT2x4, GLSL_TYPE_DOUBLE, 4, 2) +DECL_TYPE(dmat3x2, GL_DOUBLE_MAT3x2, GLSL_TYPE_DOUBLE, 2, 3) +DECL_TYPE(dmat3x4, GL_DOUBLE_MAT3x4, GLSL_TYPE_DOUBLE, 4, 3) +DECL_TYPE(dmat4x2, GL_DOUBLE_MAT4x2, GLSL_TYPE_DOUBLE, 2, 4) +DECL_TYPE(dmat4x3, GL_DOUBLE_MAT4x3, GLSL_TYPE_DOUBLE, 3, 4) + +DECL_TYPE(sampler1D, GL_SAMPLER_1D, GLSL_TYPE_SAMPLER, GLSL_SAMPLER_DIM_1D, 0, 0, GLSL_TYPE_FLOAT) +DECL_TYPE(sampler2D, GL_SAMPLER_2D, GLSL_TYPE_SAMPLER, GLSL_SAMPLER_DIM_2D, 0, 0, GLSL_TYPE_FLOAT) +DECL_TYPE(sampler3D, GL_SAMPLER_3D, GLSL_TYPE_SAMPLER, GLSL_SAMPLER_DIM_3D, 0, 0, GLSL_TYPE_FLOAT) +DECL_TYPE(samplerCube, GL_SAMPLER_CUBE, GLSL_TYPE_SAMPLER, GLSL_SAMPLER_DIM_CUBE, 0, 0, GLSL_TYPE_FLOAT) +DECL_TYPE(sampler1DArray, GL_SAMPLER_1D_ARRAY, GLSL_TYPE_SAMPLER, GLSL_SAMPLER_DIM_1D, 0, 1, GLSL_TYPE_FLOAT) +DECL_TYPE(sampler2DArray, GL_SAMPLER_2D_ARRAY, GLSL_TYPE_SAMPLER, GLSL_SAMPLER_DIM_2D, 0, 1, GLSL_TYPE_FLOAT) +DECL_TYPE(samplerCubeArray, GL_SAMPLER_CUBE_MAP_ARRAY, GLSL_TYPE_SAMPLER, GLSL_SAMPLER_DIM_CUBE, 0, 1, GLSL_TYPE_FLOAT) +DECL_TYPE(sampler2DRect, GL_SAMPLER_2D_RECT, GLSL_TYPE_SAMPLER, GLSL_SAMPLER_DIM_RECT, 0, 0, GLSL_TYPE_FLOAT) +DECL_TYPE(samplerBuffer, GL_SAMPLER_BUFFER, GLSL_TYPE_SAMPLER, GLSL_SAMPLER_DIM_BUF, 0, 0, GLSL_TYPE_FLOAT) +DECL_TYPE(sampler2DMS, GL_SAMPLER_2D_MULTISAMPLE, GLSL_TYPE_SAMPLER, GLSL_SAMPLER_DIM_MS, 0, 0, GLSL_TYPE_FLOAT) +DECL_TYPE(sampler2DMSArray, GL_SAMPLER_2D_MULTISAMPLE_ARRAY, GLSL_TYPE_SAMPLER, GLSL_SAMPLER_DIM_MS, 0, 1, GLSL_TYPE_FLOAT) + +DECL_TYPE(isampler1D, GL_INT_SAMPLER_1D, GLSL_TYPE_SAMPLER, GLSL_SAMPLER_DIM_1D, 0, 0, GLSL_TYPE_INT) +DECL_TYPE(isampler2D, GL_INT_SAMPLER_2D, GLSL_TYPE_SAMPLER, GLSL_SAMPLER_DIM_2D, 0, 0, GLSL_TYPE_INT) +DECL_TYPE(isampler3D, GL_INT_SAMPLER_3D, GLSL_TYPE_SAMPLER, GLSL_SAMPLER_DIM_3D, 0, 0, GLSL_TYPE_INT) +DECL_TYPE(isamplerCube, GL_INT_SAMPLER_CUBE, GLSL_TYPE_SAMPLER, GLSL_SAMPLER_DIM_CUBE, 0, 0, GLSL_TYPE_INT) +DECL_TYPE(isampler1DArray, GL_INT_SAMPLER_1D_ARRAY, GLSL_TYPE_SAMPLER, GLSL_SAMPLER_DIM_1D, 0, 1, GLSL_TYPE_INT) +DECL_TYPE(isampler2DArray, GL_INT_SAMPLER_2D_ARRAY, GLSL_TYPE_SAMPLER, GLSL_SAMPLER_DIM_2D, 0, 1, GLSL_TYPE_INT) +DECL_TYPE(isamplerCubeArray, GL_INT_SAMPLER_CUBE_MAP_ARRAY, GLSL_TYPE_SAMPLER, GLSL_SAMPLER_DIM_CUBE, 0, 1, GLSL_TYPE_INT) +DECL_TYPE(isampler2DRect, GL_INT_SAMPLER_2D_RECT, GLSL_TYPE_SAMPLER, GLSL_SAMPLER_DIM_RECT, 0, 0, GLSL_TYPE_INT) +DECL_TYPE(isamplerBuffer, GL_INT_SAMPLER_BUFFER, GLSL_TYPE_SAMPLER, GLSL_SAMPLER_DIM_BUF, 0, 0, GLSL_TYPE_INT) +DECL_TYPE(isampler2DMS, GL_INT_SAMPLER_2D_MULTISAMPLE, GLSL_TYPE_SAMPLER, GLSL_SAMPLER_DIM_MS, 0, 0, GLSL_TYPE_INT) +DECL_TYPE(isampler2DMSArray, GL_INT_SAMPLER_2D_MULTISAMPLE_ARRAY, GLSL_TYPE_SAMPLER, GLSL_SAMPLER_DIM_MS, 0, 1, GLSL_TYPE_INT) + +DECL_TYPE(usampler1D, GL_UNSIGNED_INT_SAMPLER_1D, GLSL_TYPE_SAMPLER, GLSL_SAMPLER_DIM_1D, 0, 0, GLSL_TYPE_UINT) +DECL_TYPE(usampler2D, GL_UNSIGNED_INT_SAMPLER_2D, GLSL_TYPE_SAMPLER, GLSL_SAMPLER_DIM_2D, 0, 0, GLSL_TYPE_UINT) +DECL_TYPE(usampler3D, GL_UNSIGNED_INT_SAMPLER_3D, GLSL_TYPE_SAMPLER, GLSL_SAMPLER_DIM_3D, 0, 0, GLSL_TYPE_UINT) +DECL_TYPE(usamplerCube, GL_UNSIGNED_INT_SAMPLER_CUBE, GLSL_TYPE_SAMPLER, GLSL_SAMPLER_DIM_CUBE, 0, 0, GLSL_TYPE_UINT) +DECL_TYPE(usampler1DArray, GL_UNSIGNED_INT_SAMPLER_1D_ARRAY, GLSL_TYPE_SAMPLER, GLSL_SAMPLER_DIM_1D, 0, 1, GLSL_TYPE_UINT) +DECL_TYPE(usampler2DArray, GL_UNSIGNED_INT_SAMPLER_2D_ARRAY, GLSL_TYPE_SAMPLER, GLSL_SAMPLER_DIM_2D, 0, 1, GLSL_TYPE_UINT) +DECL_TYPE(usamplerCubeArray, GL_UNSIGNED_INT_SAMPLER_CUBE_MAP_ARRAY, GLSL_TYPE_SAMPLER, GLSL_SAMPLER_DIM_CUBE, 0, 1, GLSL_TYPE_UINT) +DECL_TYPE(usampler2DRect, GL_UNSIGNED_INT_SAMPLER_2D_RECT, GLSL_TYPE_SAMPLER, GLSL_SAMPLER_DIM_RECT, 0, 0, GLSL_TYPE_UINT) +DECL_TYPE(usamplerBuffer, GL_UNSIGNED_INT_SAMPLER_BUFFER, GLSL_TYPE_SAMPLER, GLSL_SAMPLER_DIM_BUF, 0, 0, GLSL_TYPE_UINT) +DECL_TYPE(usampler2DMS, GL_UNSIGNED_INT_SAMPLER_2D_MULTISAMPLE, GLSL_TYPE_SAMPLER, GLSL_SAMPLER_DIM_MS, 0, 0, GLSL_TYPE_UINT) +DECL_TYPE(usampler2DMSArray, GL_UNSIGNED_INT_SAMPLER_2D_MULTISAMPLE_ARRAY, GLSL_TYPE_SAMPLER, GLSL_SAMPLER_DIM_MS, 0, 1, GLSL_TYPE_UINT) + +DECL_TYPE(sampler1DShadow, GL_SAMPLER_1D_SHADOW, GLSL_TYPE_SAMPLER, GLSL_SAMPLER_DIM_1D, 1, 0, GLSL_TYPE_FLOAT) +DECL_TYPE(sampler2DShadow, GL_SAMPLER_2D_SHADOW, GLSL_TYPE_SAMPLER, GLSL_SAMPLER_DIM_2D, 1, 0, GLSL_TYPE_FLOAT) +DECL_TYPE(samplerCubeShadow, GL_SAMPLER_CUBE_SHADOW, GLSL_TYPE_SAMPLER, GLSL_SAMPLER_DIM_CUBE, 1, 0, GLSL_TYPE_FLOAT) +DECL_TYPE(sampler1DArrayShadow, GL_SAMPLER_1D_ARRAY_SHADOW, GLSL_TYPE_SAMPLER, GLSL_SAMPLER_DIM_1D, 1, 1, GLSL_TYPE_FLOAT) +DECL_TYPE(sampler2DArrayShadow, GL_SAMPLER_2D_ARRAY_SHADOW, GLSL_TYPE_SAMPLER, GLSL_SAMPLER_DIM_2D, 1, 1, GLSL_TYPE_FLOAT) +DECL_TYPE(samplerCubeArrayShadow, GL_SAMPLER_CUBE_MAP_ARRAY_SHADOW, GLSL_TYPE_SAMPLER, GLSL_SAMPLER_DIM_CUBE, 1, 1, GLSL_TYPE_FLOAT) +DECL_TYPE(sampler2DRectShadow, GL_SAMPLER_2D_RECT_SHADOW, GLSL_TYPE_SAMPLER, GLSL_SAMPLER_DIM_RECT, 1, 0, GLSL_TYPE_FLOAT) + +DECL_TYPE(samplerExternalOES, GL_SAMPLER_EXTERNAL_OES, GLSL_TYPE_SAMPLER, GLSL_SAMPLER_DIM_EXTERNAL, 0, 0, GLSL_TYPE_FLOAT) + +DECL_TYPE(image1D, GL_IMAGE_1D, GLSL_TYPE_IMAGE, GLSL_SAMPLER_DIM_1D, 0, 0, GLSL_TYPE_FLOAT) +DECL_TYPE(image2D, GL_IMAGE_2D, GLSL_TYPE_IMAGE, GLSL_SAMPLER_DIM_2D, 0, 0, GLSL_TYPE_FLOAT) +DECL_TYPE(image3D, GL_IMAGE_3D, GLSL_TYPE_IMAGE, GLSL_SAMPLER_DIM_3D, 0, 0, GLSL_TYPE_FLOAT) +DECL_TYPE(image2DRect, GL_IMAGE_2D_RECT, GLSL_TYPE_IMAGE, GLSL_SAMPLER_DIM_RECT, 0, 0, GLSL_TYPE_FLOAT) +DECL_TYPE(imageCube, GL_IMAGE_CUBE, GLSL_TYPE_IMAGE, GLSL_SAMPLER_DIM_CUBE, 0, 0, GLSL_TYPE_FLOAT) +DECL_TYPE(imageBuffer, GL_IMAGE_BUFFER, GLSL_TYPE_IMAGE, GLSL_SAMPLER_DIM_BUF, 0, 0, GLSL_TYPE_FLOAT) +DECL_TYPE(image1DArray, GL_IMAGE_1D_ARRAY, GLSL_TYPE_IMAGE, GLSL_SAMPLER_DIM_1D, 0, 1, GLSL_TYPE_FLOAT) +DECL_TYPE(image2DArray, GL_IMAGE_2D_ARRAY, GLSL_TYPE_IMAGE, GLSL_SAMPLER_DIM_2D, 0, 1, GLSL_TYPE_FLOAT) +DECL_TYPE(imageCubeArray, GL_IMAGE_CUBE_MAP_ARRAY, GLSL_TYPE_IMAGE, GLSL_SAMPLER_DIM_CUBE, 0, 1, GLSL_TYPE_FLOAT) +DECL_TYPE(image2DMS, GL_IMAGE_2D_MULTISAMPLE, GLSL_TYPE_IMAGE, GLSL_SAMPLER_DIM_MS, 0, 0, GLSL_TYPE_FLOAT) +DECL_TYPE(image2DMSArray, GL_IMAGE_2D_MULTISAMPLE_ARRAY, GLSL_TYPE_IMAGE, GLSL_SAMPLER_DIM_MS, 0, 1, GLSL_TYPE_FLOAT) +DECL_TYPE(iimage1D, GL_INT_IMAGE_1D, GLSL_TYPE_IMAGE, GLSL_SAMPLER_DIM_1D, 0, 0, GLSL_TYPE_INT) +DECL_TYPE(iimage2D, GL_INT_IMAGE_2D, GLSL_TYPE_IMAGE, GLSL_SAMPLER_DIM_2D, 0, 0, GLSL_TYPE_INT) +DECL_TYPE(iimage3D, GL_INT_IMAGE_3D, GLSL_TYPE_IMAGE, GLSL_SAMPLER_DIM_3D, 0, 0, GLSL_TYPE_INT) +DECL_TYPE(iimage2DRect, GL_INT_IMAGE_2D_RECT, GLSL_TYPE_IMAGE, GLSL_SAMPLER_DIM_RECT, 0, 0, GLSL_TYPE_INT) +DECL_TYPE(iimageCube, GL_INT_IMAGE_CUBE, GLSL_TYPE_IMAGE, GLSL_SAMPLER_DIM_CUBE, 0, 0, GLSL_TYPE_INT) +DECL_TYPE(iimageBuffer, GL_INT_IMAGE_BUFFER, GLSL_TYPE_IMAGE, GLSL_SAMPLER_DIM_BUF, 0, 0, GLSL_TYPE_INT) +DECL_TYPE(iimage1DArray, GL_INT_IMAGE_1D_ARRAY, GLSL_TYPE_IMAGE, GLSL_SAMPLER_DIM_1D, 0, 1, GLSL_TYPE_INT) +DECL_TYPE(iimage2DArray, GL_INT_IMAGE_2D_ARRAY, GLSL_TYPE_IMAGE, GLSL_SAMPLER_DIM_2D, 0, 1, GLSL_TYPE_INT) +DECL_TYPE(iimageCubeArray, GL_INT_IMAGE_CUBE_MAP_ARRAY, GLSL_TYPE_IMAGE, GLSL_SAMPLER_DIM_CUBE, 0, 1, GLSL_TYPE_INT) +DECL_TYPE(iimage2DMS, GL_INT_IMAGE_2D_MULTISAMPLE, GLSL_TYPE_IMAGE, GLSL_SAMPLER_DIM_MS, 0, 0, GLSL_TYPE_INT) +DECL_TYPE(iimage2DMSArray, GL_INT_IMAGE_2D_MULTISAMPLE_ARRAY, GLSL_TYPE_IMAGE, GLSL_SAMPLER_DIM_MS, 0, 1, GLSL_TYPE_INT) +DECL_TYPE(uimage1D, GL_UNSIGNED_INT_IMAGE_1D, GLSL_TYPE_IMAGE, GLSL_SAMPLER_DIM_1D, 0, 0, GLSL_TYPE_UINT) +DECL_TYPE(uimage2D, GL_UNSIGNED_INT_IMAGE_2D, GLSL_TYPE_IMAGE, GLSL_SAMPLER_DIM_2D, 0, 0, GLSL_TYPE_UINT) +DECL_TYPE(uimage3D, GL_UNSIGNED_INT_IMAGE_3D, GLSL_TYPE_IMAGE, GLSL_SAMPLER_DIM_3D, 0, 0, GLSL_TYPE_UINT) +DECL_TYPE(uimage2DRect, GL_UNSIGNED_INT_IMAGE_2D_RECT, GLSL_TYPE_IMAGE, GLSL_SAMPLER_DIM_RECT, 0, 0, GLSL_TYPE_UINT) +DECL_TYPE(uimageCube, GL_UNSIGNED_INT_IMAGE_CUBE, GLSL_TYPE_IMAGE, GLSL_SAMPLER_DIM_CUBE, 0, 0, GLSL_TYPE_UINT) +DECL_TYPE(uimageBuffer, GL_UNSIGNED_INT_IMAGE_BUFFER, GLSL_TYPE_IMAGE, GLSL_SAMPLER_DIM_BUF, 0, 0, GLSL_TYPE_UINT) +DECL_TYPE(uimage1DArray, GL_UNSIGNED_INT_IMAGE_1D_ARRAY, GLSL_TYPE_IMAGE, GLSL_SAMPLER_DIM_1D, 0, 1, GLSL_TYPE_UINT) +DECL_TYPE(uimage2DArray, GL_UNSIGNED_INT_IMAGE_2D_ARRAY, GLSL_TYPE_IMAGE, GLSL_SAMPLER_DIM_2D, 0, 1, GLSL_TYPE_UINT) +DECL_TYPE(uimageCubeArray, GL_UNSIGNED_INT_IMAGE_CUBE_MAP_ARRAY, GLSL_TYPE_IMAGE, GLSL_SAMPLER_DIM_CUBE, 0, 1, GLSL_TYPE_UINT) +DECL_TYPE(uimage2DMS, GL_UNSIGNED_INT_IMAGE_2D_MULTISAMPLE, GLSL_TYPE_IMAGE, GLSL_SAMPLER_DIM_MS, 0, 0, GLSL_TYPE_UINT) +DECL_TYPE(uimage2DMSArray, GL_UNSIGNED_INT_IMAGE_2D_MULTISAMPLE_ARRAY, GLSL_TYPE_IMAGE, GLSL_SAMPLER_DIM_MS, 0, 1, GLSL_TYPE_UINT) + +DECL_TYPE(atomic_uint, GL_UNSIGNED_INT_ATOMIC_COUNTER, GLSL_TYPE_ATOMIC_UINT, 1, 1) + +STRUCT_TYPE(gl_DepthRangeParameters) +STRUCT_TYPE(gl_PointParameters) +STRUCT_TYPE(gl_MaterialParameters) +STRUCT_TYPE(gl_LightSourceParameters) +STRUCT_TYPE(gl_LightModelParameters) +STRUCT_TYPE(gl_LightModelProducts) +STRUCT_TYPE(gl_LightProducts) +STRUCT_TYPE(gl_FogParameters) diff --git a/src/glsl/nir/glsl_to_nir.cpp b/src/glsl/nir/glsl_to_nir.cpp index 6bedb4eb8e6..e57e834d948 100644 --- a/src/glsl/nir/glsl_to_nir.cpp +++ b/src/glsl/nir/glsl_to_nir.cpp @@ -152,11 +152,13 @@ glsl_to_nir(const struct gl_shader_program *shader_prog, if (sh->Program->SamplersUsed & (1 << i)) num_textures = i; - shader->info.name = ralloc_asprintf(shader, "GLSL%d", sh->Name); + shader->info.name = ralloc_asprintf(shader, "GLSL%d", shader_prog->Name); + if (shader_prog->Label) + shader->info.label = ralloc_strdup(shader, shader_prog->Label); shader->info.num_textures = num_textures; shader->info.num_ubos = sh->NumUniformBlocks; shader->info.num_abos = shader_prog->NumAtomicBuffers; - shader->info.num_ssbos = shader_prog->NumBufferInterfaceBlocks; + shader->info.num_ssbos = sh->NumShaderStorageBlocks; shader->info.num_images = sh->NumImages; shader->info.inputs_read = sh->Program->InputsRead; shader->info.outputs_written = sh->Program->OutputsWritten; @@ -164,11 +166,37 @@ glsl_to_nir(const struct gl_shader_program *shader_prog, shader->info.uses_texture_gather = sh->Program->UsesGather; shader->info.uses_clip_distance_out = sh->Program->UsesClipDistanceOut; shader->info.separate_shader = shader_prog->SeparateShader; - shader->info.gs.vertices_out = sh->Geom.VerticesOut; - shader->info.gs.invocations = sh->Geom.Invocations; shader->info.has_transform_feedback_varyings = shader_prog->TransformFeedback.NumVarying > 0; + switch (stage) { + case MESA_SHADER_GEOMETRY: + shader->info.gs.vertices_out = sh->Geom.VerticesOut; + shader->info.gs.invocations = sh->Geom.Invocations; + break; + + case MESA_SHADER_FRAGMENT: { + struct gl_fragment_program *fp = + (struct gl_fragment_program *)sh->Program; + + shader->info.fs.uses_discard = fp->UsesKill; + shader->info.fs.early_fragment_tests = sh->EarlyFragmentTests; + shader->info.fs.depth_layout = fp->FragDepthLayout; + break; + } + + case MESA_SHADER_COMPUTE: { + struct gl_compute_program *cp = (struct gl_compute_program *)sh->Program; + shader->info.cs.local_size[0] = cp->LocalSize[0]; + shader->info.cs.local_size[1] = cp->LocalSize[1]; + shader->info.cs.local_size[2] = cp->LocalSize[2]; + break; + } + + default: + break; /* No stage-specific info */ + } + return shader; } @@ -393,35 +421,10 @@ nir_visitor::visit(ir_variable *ir) var->interface_type = ir->get_interface_type(); - switch (var->data.mode) { - case nir_var_local: - exec_list_push_tail(&impl->locals, &var->node); - break; - - case nir_var_global: - exec_list_push_tail(&shader->globals, &var->node); - break; - - case nir_var_shader_in: - exec_list_push_tail(&shader->inputs, &var->node); - break; - - case nir_var_shader_out: - exec_list_push_tail(&shader->outputs, &var->node); - break; - - case nir_var_uniform: - case nir_var_shader_storage: - exec_list_push_tail(&shader->uniforms, &var->node); - break; - - case nir_var_system_value: - exec_list_push_tail(&shader->system_values, &var->node); - break; - - default: - unreachable("not reached"); - } + if (var->data.mode == nir_var_local) + nir_function_impl_add_variable(impl, var); + else + nir_shader_add_variable(shader, var); _mesa_hash_table_insert(var_table, ir, var); this->var = var; @@ -695,9 +698,21 @@ nir_visitor::visit(ir_call *ir) } else if (strcmp(ir->callee_name(), "__intrinsic_ssbo_atomic_xor_internal") == 0) { op = nir_intrinsic_ssbo_atomic_xor; } else if (strcmp(ir->callee_name(), "__intrinsic_ssbo_atomic_min_internal") == 0) { - op = nir_intrinsic_ssbo_atomic_min; + assert(ir->return_deref); + if (ir->return_deref->type == glsl_type::int_type) + op = nir_intrinsic_ssbo_atomic_imin; + else if (ir->return_deref->type == glsl_type::uint_type) + op = nir_intrinsic_ssbo_atomic_umin; + else + unreachable("Invalid type"); } else if (strcmp(ir->callee_name(), "__intrinsic_ssbo_atomic_max_internal") == 0) { - op = nir_intrinsic_ssbo_atomic_max; + assert(ir->return_deref); + if (ir->return_deref->type == glsl_type::int_type) + op = nir_intrinsic_ssbo_atomic_imax; + else if (ir->return_deref->type == glsl_type::uint_type) + op = nir_intrinsic_ssbo_atomic_umax; + else + unreachable("Invalid type"); } else if (strcmp(ir->callee_name(), "__intrinsic_ssbo_atomic_exchange_internal") == 0) { op = nir_intrinsic_ssbo_atomic_exchange; } else if (strcmp(ir->callee_name(), "__intrinsic_ssbo_atomic_comp_swap_internal") == 0) { @@ -906,8 +921,10 @@ nir_visitor::visit(ir_call *ir) break; } case nir_intrinsic_ssbo_atomic_add: - case nir_intrinsic_ssbo_atomic_min: - case nir_intrinsic_ssbo_atomic_max: + case nir_intrinsic_ssbo_atomic_imin: + case nir_intrinsic_ssbo_atomic_umin: + case nir_intrinsic_ssbo_atomic_imax: + case nir_intrinsic_ssbo_atomic_umax: case nir_intrinsic_ssbo_atomic_and: case nir_intrinsic_ssbo_atomic_or: case nir_intrinsic_ssbo_atomic_xor: @@ -2065,13 +2082,10 @@ nir_visitor::visit(ir_constant *ir) * constant initializer and return a dereference. */ - nir_variable *var = ralloc(this->shader, nir_variable); - var->name = ralloc_strdup(var, "const_temp"); - var->type = ir->type; - var->data.mode = nir_var_local; + nir_variable *var = + nir_local_variable_create(this->impl, ir->type, "const_temp"); var->data.read_only = true; var->constant_initializer = constant_copy(ir, var); - exec_list_push_tail(&this->impl->locals, &var->node); this->deref_head = nir_deref_var_create(this->shader, var); this->deref_tail = &this->deref_head->deref; diff --git a/src/glsl/nir/glsl_types.cpp b/src/glsl/nir/glsl_types.cpp new file mode 100644 index 00000000000..309f9dca61e --- /dev/null +++ b/src/glsl/nir/glsl_types.cpp @@ -0,0 +1,1850 @@ +/* + * 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 "main/core.h" /* for Elements, MAX2 */ +#include "glsl_parser_extras.h" +#include "glsl_types.h" +#include "util/hash_table.h" + + +mtx_t glsl_type::mutex = _MTX_INITIALIZER_NP; +hash_table *glsl_type::array_types = NULL; +hash_table *glsl_type::record_types = NULL; +hash_table *glsl_type::interface_types = NULL; +hash_table *glsl_type::function_types = NULL; +hash_table *glsl_type::subroutine_types = NULL; +void *glsl_type::mem_ctx = NULL; + +void +glsl_type::init_ralloc_type_ctx(void) +{ + if (glsl_type::mem_ctx == NULL) { + glsl_type::mem_ctx = ralloc_autofree_context(); + assert(glsl_type::mem_ctx != NULL); + } +} + +glsl_type::glsl_type(GLenum gl_type, + glsl_base_type base_type, unsigned vector_elements, + unsigned matrix_columns, const char *name) : + gl_type(gl_type), + base_type(base_type), + sampler_dimensionality(0), sampler_shadow(0), sampler_array(0), + sampler_type(0), interface_packing(0), + vector_elements(vector_elements), matrix_columns(matrix_columns), + length(0) +{ + mtx_lock(&glsl_type::mutex); + + init_ralloc_type_ctx(); + assert(name != NULL); + this->name = ralloc_strdup(this->mem_ctx, name); + + mtx_unlock(&glsl_type::mutex); + + /* Neither dimension is zero or both dimensions are zero. + */ + assert((vector_elements == 0) == (matrix_columns == 0)); + memset(& fields, 0, sizeof(fields)); +} + +glsl_type::glsl_type(GLenum gl_type, glsl_base_type base_type, + enum glsl_sampler_dim dim, bool shadow, bool array, + unsigned type, const char *name) : + gl_type(gl_type), + base_type(base_type), + sampler_dimensionality(dim), sampler_shadow(shadow), + sampler_array(array), sampler_type(type), interface_packing(0), + length(0) +{ + mtx_lock(&glsl_type::mutex); + + init_ralloc_type_ctx(); + assert(name != NULL); + this->name = ralloc_strdup(this->mem_ctx, name); + + mtx_unlock(&glsl_type::mutex); + + memset(& fields, 0, sizeof(fields)); + + if (base_type == GLSL_TYPE_SAMPLER) { + /* Samplers take no storage whatsoever. */ + matrix_columns = vector_elements = 0; + } else { + matrix_columns = vector_elements = 1; + } +} + +glsl_type::glsl_type(const glsl_struct_field *fields, unsigned num_fields, + const char *name) : + gl_type(0), + base_type(GLSL_TYPE_STRUCT), + sampler_dimensionality(0), sampler_shadow(0), sampler_array(0), + sampler_type(0), interface_packing(0), + vector_elements(0), matrix_columns(0), + length(num_fields) +{ + unsigned int i; + + mtx_lock(&glsl_type::mutex); + + init_ralloc_type_ctx(); + assert(name != NULL); + this->name = ralloc_strdup(this->mem_ctx, name); + this->fields.structure = ralloc_array(this->mem_ctx, + glsl_struct_field, length); + + for (i = 0; i < length; i++) { + this->fields.structure[i].type = fields[i].type; + this->fields.structure[i].name = ralloc_strdup(this->fields.structure, + fields[i].name); + this->fields.structure[i].location = fields[i].location; + this->fields.structure[i].interpolation = fields[i].interpolation; + this->fields.structure[i].centroid = fields[i].centroid; + this->fields.structure[i].sample = fields[i].sample; + this->fields.structure[i].matrix_layout = fields[i].matrix_layout; + this->fields.structure[i].patch = fields[i].patch; + this->fields.structure[i].image_read_only = fields[i].image_read_only; + this->fields.structure[i].image_write_only = fields[i].image_write_only; + this->fields.structure[i].image_coherent = fields[i].image_coherent; + this->fields.structure[i].image_volatile = fields[i].image_volatile; + this->fields.structure[i].image_restrict = fields[i].image_restrict; + } + + mtx_unlock(&glsl_type::mutex); +} + +glsl_type::glsl_type(const glsl_struct_field *fields, unsigned num_fields, + enum glsl_interface_packing packing, const char *name) : + gl_type(0), + base_type(GLSL_TYPE_INTERFACE), + sampler_dimensionality(0), sampler_shadow(0), sampler_array(0), + sampler_type(0), interface_packing((unsigned) packing), + vector_elements(0), matrix_columns(0), + length(num_fields) +{ + unsigned int i; + + mtx_lock(&glsl_type::mutex); + + init_ralloc_type_ctx(); + assert(name != NULL); + this->name = ralloc_strdup(this->mem_ctx, name); + this->fields.structure = ralloc_array(this->mem_ctx, + glsl_struct_field, length); + for (i = 0; i < length; i++) { + this->fields.structure[i].type = fields[i].type; + this->fields.structure[i].name = ralloc_strdup(this->fields.structure, + fields[i].name); + this->fields.structure[i].location = fields[i].location; + this->fields.structure[i].interpolation = fields[i].interpolation; + this->fields.structure[i].centroid = fields[i].centroid; + this->fields.structure[i].sample = fields[i].sample; + this->fields.structure[i].matrix_layout = fields[i].matrix_layout; + this->fields.structure[i].patch = fields[i].patch; + } + + mtx_unlock(&glsl_type::mutex); +} + +glsl_type::glsl_type(const glsl_type *return_type, + const glsl_function_param *params, unsigned num_params) : + gl_type(0), + base_type(GLSL_TYPE_FUNCTION), + sampler_dimensionality(0), sampler_shadow(0), sampler_array(0), + sampler_type(0), interface_packing(0), + vector_elements(0), matrix_columns(0), + length(num_params) +{ + unsigned int i; + + mtx_lock(&glsl_type::mutex); + + init_ralloc_type_ctx(); + + this->fields.parameters = rzalloc_array(this->mem_ctx, + glsl_function_param, num_params + 1); + + /* We store the return type as the first parameter */ + this->fields.parameters[0].type = return_type; + this->fields.parameters[0].in = false; + this->fields.parameters[0].out = true; + + /* We store the i'th parameter in slot i+1 */ + for (i = 0; i < length; i++) { + this->fields.parameters[i + 1].type = params[i].type; + this->fields.parameters[i + 1].in = params[i].in; + this->fields.parameters[i + 1].out = params[i].out; + } + + mtx_unlock(&glsl_type::mutex); +} + +glsl_type::glsl_type(const char *subroutine_name) : + gl_type(0), + base_type(GLSL_TYPE_SUBROUTINE), + sampler_dimensionality(0), sampler_shadow(0), sampler_array(0), + sampler_type(0), interface_packing(0), + vector_elements(1), matrix_columns(1), + length(0) +{ + mtx_lock(&glsl_type::mutex); + + init_ralloc_type_ctx(); + assert(subroutine_name != NULL); + this->name = ralloc_strdup(this->mem_ctx, subroutine_name); + mtx_unlock(&glsl_type::mutex); +} + +bool +glsl_type::contains_sampler() const +{ + if (this->is_array()) { + return this->fields.array->contains_sampler(); + } else if (this->is_record()) { + for (unsigned int i = 0; i < this->length; i++) { + if (this->fields.structure[i].type->contains_sampler()) + return true; + } + return false; + } else { + return this->is_sampler(); + } +} + + +bool +glsl_type::contains_integer() const +{ + if (this->is_array()) { + return this->fields.array->contains_integer(); + } else if (this->is_record()) { + for (unsigned int i = 0; i < this->length; i++) { + if (this->fields.structure[i].type->contains_integer()) + return true; + } + return false; + } else { + return this->is_integer(); + } +} + +bool +glsl_type::contains_double() const +{ + if (this->is_array()) { + return this->fields.array->contains_double(); + } else if (this->is_record()) { + for (unsigned int i = 0; i < this->length; i++) { + if (this->fields.structure[i].type->contains_double()) + return true; + } + return false; + } else { + return this->is_double(); + } +} + +bool +glsl_type::contains_opaque() const { + switch (base_type) { + case GLSL_TYPE_SAMPLER: + case GLSL_TYPE_IMAGE: + case GLSL_TYPE_ATOMIC_UINT: + return true; + case GLSL_TYPE_ARRAY: + return fields.array->contains_opaque(); + case GLSL_TYPE_STRUCT: + for (unsigned int i = 0; i < length; i++) { + if (fields.structure[i].type->contains_opaque()) + return true; + } + return false; + default: + return false; + } +} + +bool +glsl_type::contains_subroutine() const +{ + if (this->is_array()) { + return this->fields.array->contains_subroutine(); + } else if (this->is_record()) { + for (unsigned int i = 0; i < this->length; i++) { + if (this->fields.structure[i].type->contains_subroutine()) + return true; + } + return false; + } else { + return this->is_subroutine(); + } +} + +gl_texture_index +glsl_type::sampler_index() const +{ + const glsl_type *const t = (this->is_array()) ? this->fields.array : this; + + assert(t->is_sampler()); + + switch (t->sampler_dimensionality) { + case GLSL_SAMPLER_DIM_1D: + return (t->sampler_array) ? TEXTURE_1D_ARRAY_INDEX : TEXTURE_1D_INDEX; + case GLSL_SAMPLER_DIM_2D: + return (t->sampler_array) ? TEXTURE_2D_ARRAY_INDEX : TEXTURE_2D_INDEX; + case GLSL_SAMPLER_DIM_3D: + return TEXTURE_3D_INDEX; + case GLSL_SAMPLER_DIM_CUBE: + return (t->sampler_array) ? TEXTURE_CUBE_ARRAY_INDEX : TEXTURE_CUBE_INDEX; + case GLSL_SAMPLER_DIM_RECT: + return TEXTURE_RECT_INDEX; + case GLSL_SAMPLER_DIM_BUF: + return TEXTURE_BUFFER_INDEX; + case GLSL_SAMPLER_DIM_EXTERNAL: + return TEXTURE_EXTERNAL_INDEX; + case GLSL_SAMPLER_DIM_MS: + return (t->sampler_array) ? TEXTURE_2D_MULTISAMPLE_ARRAY_INDEX : TEXTURE_2D_MULTISAMPLE_INDEX; + default: + assert(!"Should not get here."); + return TEXTURE_BUFFER_INDEX; + } +} + +bool +glsl_type::contains_image() const +{ + if (this->is_array()) { + return this->fields.array->contains_image(); + } else if (this->is_record()) { + for (unsigned int i = 0; i < this->length; i++) { + if (this->fields.structure[i].type->contains_image()) + return true; + } + return false; + } else { + return this->is_image(); + } +} + +const glsl_type *glsl_type::get_base_type() const +{ + switch (base_type) { + case GLSL_TYPE_UINT: + return uint_type; + case GLSL_TYPE_INT: + return int_type; + case GLSL_TYPE_FLOAT: + return float_type; + case GLSL_TYPE_DOUBLE: + return double_type; + case GLSL_TYPE_BOOL: + return bool_type; + default: + return error_type; + } +} + + +const glsl_type *glsl_type::get_scalar_type() const +{ + const glsl_type *type = this; + + /* Handle arrays */ + while (type->base_type == GLSL_TYPE_ARRAY) + type = type->fields.array; + + /* Handle vectors and matrices */ + switch (type->base_type) { + case GLSL_TYPE_UINT: + return uint_type; + case GLSL_TYPE_INT: + return int_type; + case GLSL_TYPE_FLOAT: + return float_type; + case GLSL_TYPE_DOUBLE: + return double_type; + case GLSL_TYPE_BOOL: + return bool_type; + default: + /* Handle everything else */ + return type; + } +} + + +void +_mesa_glsl_release_types(void) +{ + /* Should only be called during atexit (either when unloading shared + * object, or if process terminates), so no mutex-locking should be + * necessary. + */ + if (glsl_type::array_types != NULL) { + _mesa_hash_table_destroy(glsl_type::array_types, NULL); + glsl_type::array_types = NULL; + } + + if (glsl_type::record_types != NULL) { + _mesa_hash_table_destroy(glsl_type::record_types, NULL); + glsl_type::record_types = NULL; + } + + if (glsl_type::interface_types != NULL) { + _mesa_hash_table_destroy(glsl_type::interface_types, NULL); + glsl_type::interface_types = NULL; + } +} + + +glsl_type::glsl_type(const glsl_type *array, unsigned length) : + base_type(GLSL_TYPE_ARRAY), + sampler_dimensionality(0), sampler_shadow(0), sampler_array(0), + sampler_type(0), interface_packing(0), + vector_elements(0), matrix_columns(0), + length(length), name(NULL) +{ + this->fields.array = array; + /* Inherit the gl type of the base. The GL type is used for + * uniform/statevar handling in Mesa and the arrayness of the type + * is represented by the size rather than the type. + */ + this->gl_type = array->gl_type; + + /* Allow a maximum of 10 characters for the array size. This is enough + * for 32-bits of ~0. The extra 3 are for the '[', ']', and terminating + * NUL. + */ + const unsigned name_length = strlen(array->name) + 10 + 3; + + mtx_lock(&glsl_type::mutex); + char *const n = (char *) ralloc_size(this->mem_ctx, name_length); + mtx_unlock(&glsl_type::mutex); + + if (length == 0) + snprintf(n, name_length, "%s[]", array->name); + else { + /* insert outermost dimensions in the correct spot + * otherwise the dimension order will be backwards + */ + const char *pos = strchr(array->name, '['); + if (pos) { + int idx = pos - array->name; + snprintf(n, idx+1, "%s", array->name); + snprintf(n + idx, name_length - idx, "[%u]%s", + length, array->name + idx); + } else { + snprintf(n, name_length, "%s[%u]", array->name, length); + } + } + + this->name = n; +} + + +const glsl_type * +glsl_type::vec(unsigned components) +{ + if (components == 0 || components > 4) + return error_type; + + static const glsl_type *const ts[] = { + float_type, vec2_type, vec3_type, vec4_type + }; + return ts[components - 1]; +} + +const glsl_type * +glsl_type::dvec(unsigned components) +{ + if (components == 0 || components > 4) + return error_type; + + static const glsl_type *const ts[] = { + double_type, dvec2_type, dvec3_type, dvec4_type + }; + return ts[components - 1]; +} + +const glsl_type * +glsl_type::ivec(unsigned components) +{ + if (components == 0 || components > 4) + return error_type; + + static const glsl_type *const ts[] = { + int_type, ivec2_type, ivec3_type, ivec4_type + }; + return ts[components - 1]; +} + + +const glsl_type * +glsl_type::uvec(unsigned components) +{ + if (components == 0 || components > 4) + return error_type; + + static const glsl_type *const ts[] = { + uint_type, uvec2_type, uvec3_type, uvec4_type + }; + return ts[components - 1]; +} + + +const glsl_type * +glsl_type::bvec(unsigned components) +{ + if (components == 0 || components > 4) + return error_type; + + static const glsl_type *const ts[] = { + bool_type, bvec2_type, bvec3_type, bvec4_type + }; + return ts[components - 1]; +} + + +const glsl_type * +glsl_type::get_instance(unsigned base_type, unsigned rows, unsigned columns) +{ + if (base_type == GLSL_TYPE_VOID) + return void_type; + + if ((rows < 1) || (rows > 4) || (columns < 1) || (columns > 4)) + return error_type; + + /* Treat GLSL vectors as Nx1 matrices. + */ + if (columns == 1) { + switch (base_type) { + case GLSL_TYPE_UINT: + return uvec(rows); + case GLSL_TYPE_INT: + return ivec(rows); + case GLSL_TYPE_FLOAT: + return vec(rows); + case GLSL_TYPE_DOUBLE: + return dvec(rows); + case GLSL_TYPE_BOOL: + return bvec(rows); + default: + return error_type; + } + } else { + if ((base_type != GLSL_TYPE_FLOAT && base_type != GLSL_TYPE_DOUBLE) || (rows == 1)) + return error_type; + + /* GLSL matrix types are named mat{COLUMNS}x{ROWS}. Only the following + * combinations are valid: + * + * 1 2 3 4 + * 1 + * 2 x x x + * 3 x x x + * 4 x x x + */ +#define IDX(c,r) (((c-1)*3) + (r-1)) + + if (base_type == GLSL_TYPE_DOUBLE) { + switch (IDX(columns, rows)) { + case IDX(2,2): return dmat2_type; + case IDX(2,3): return dmat2x3_type; + case IDX(2,4): return dmat2x4_type; + case IDX(3,2): return dmat3x2_type; + case IDX(3,3): return dmat3_type; + case IDX(3,4): return dmat3x4_type; + case IDX(4,2): return dmat4x2_type; + case IDX(4,3): return dmat4x3_type; + case IDX(4,4): return dmat4_type; + default: return error_type; + } + } else { + switch (IDX(columns, rows)) { + case IDX(2,2): return mat2_type; + case IDX(2,3): return mat2x3_type; + case IDX(2,4): return mat2x4_type; + case IDX(3,2): return mat3x2_type; + case IDX(3,3): return mat3_type; + case IDX(3,4): return mat3x4_type; + case IDX(4,2): return mat4x2_type; + case IDX(4,3): return mat4x3_type; + case IDX(4,4): return mat4_type; + default: return error_type; + } + } + } + + assert(!"Should not get here."); + return error_type; +} + +const glsl_type * +glsl_type::get_sampler_instance(enum glsl_sampler_dim dim, + bool shadow, + bool array, + glsl_base_type type) +{ + switch (type) { + case GLSL_TYPE_FLOAT: + switch (dim) { + case GLSL_SAMPLER_DIM_1D: + if (shadow) + return (array ? sampler1DArrayShadow_type : sampler1DShadow_type); + else + return (array ? sampler1DArray_type : sampler1D_type); + case GLSL_SAMPLER_DIM_2D: + if (shadow) + return (array ? sampler2DArrayShadow_type : sampler2DShadow_type); + else + return (array ? sampler2DArray_type : sampler2D_type); + case GLSL_SAMPLER_DIM_3D: + if (shadow || array) + return error_type; + else + return sampler3D_type; + case GLSL_SAMPLER_DIM_CUBE: + if (shadow) + return (array ? samplerCubeArrayShadow_type : samplerCubeShadow_type); + else + return (array ? samplerCubeArray_type : samplerCube_type); + case GLSL_SAMPLER_DIM_RECT: + if (array) + return error_type; + if (shadow) + return sampler2DRectShadow_type; + else + return sampler2DRect_type; + case GLSL_SAMPLER_DIM_BUF: + if (shadow || array) + return error_type; + else + return samplerBuffer_type; + case GLSL_SAMPLER_DIM_MS: + if (shadow) + return error_type; + return (array ? sampler2DMSArray_type : sampler2DMS_type); + case GLSL_SAMPLER_DIM_EXTERNAL: + if (shadow || array) + return error_type; + else + return samplerExternalOES_type; + } + case GLSL_TYPE_INT: + if (shadow) + return error_type; + switch (dim) { + case GLSL_SAMPLER_DIM_1D: + return (array ? isampler1DArray_type : isampler1D_type); + case GLSL_SAMPLER_DIM_2D: + return (array ? isampler2DArray_type : isampler2D_type); + case GLSL_SAMPLER_DIM_3D: + if (array) + return error_type; + return isampler3D_type; + case GLSL_SAMPLER_DIM_CUBE: + return (array ? isamplerCubeArray_type : isamplerCube_type); + case GLSL_SAMPLER_DIM_RECT: + if (array) + return error_type; + return isampler2DRect_type; + case GLSL_SAMPLER_DIM_BUF: + if (array) + return error_type; + return isamplerBuffer_type; + case GLSL_SAMPLER_DIM_MS: + return (array ? isampler2DMSArray_type : isampler2DMS_type); + case GLSL_SAMPLER_DIM_EXTERNAL: + return error_type; + } + case GLSL_TYPE_UINT: + if (shadow) + return error_type; + switch (dim) { + case GLSL_SAMPLER_DIM_1D: + return (array ? usampler1DArray_type : usampler1D_type); + case GLSL_SAMPLER_DIM_2D: + return (array ? usampler2DArray_type : usampler2D_type); + case GLSL_SAMPLER_DIM_3D: + if (array) + return error_type; + return usampler3D_type; + case GLSL_SAMPLER_DIM_CUBE: + return (array ? usamplerCubeArray_type : usamplerCube_type); + case GLSL_SAMPLER_DIM_RECT: + if (array) + return error_type; + return usampler2DRect_type; + case GLSL_SAMPLER_DIM_BUF: + if (array) + return error_type; + return usamplerBuffer_type; + case GLSL_SAMPLER_DIM_MS: + return (array ? usampler2DMSArray_type : usampler2DMS_type); + case GLSL_SAMPLER_DIM_EXTERNAL: + return error_type; + } + default: + return error_type; + } + + unreachable("switch statement above should be complete"); +} + +const glsl_type * +glsl_type::get_array_instance(const glsl_type *base, unsigned array_size) +{ + /* Generate a name using the base type pointer in the key. This is + * done because the name of the base type may not be unique across + * shaders. For example, two shaders may have different record types + * named 'foo'. + */ + char key[128]; + snprintf(key, sizeof(key), "%p[%u]", (void *) base, array_size); + + mtx_lock(&glsl_type::mutex); + + if (array_types == NULL) { + array_types = _mesa_hash_table_create(NULL, _mesa_key_hash_string, + _mesa_key_string_equal); + } + + const struct hash_entry *entry = _mesa_hash_table_search(array_types, key); + if (entry == NULL) { + mtx_unlock(&glsl_type::mutex); + const glsl_type *t = new glsl_type(base, array_size); + mtx_lock(&glsl_type::mutex); + + entry = _mesa_hash_table_insert(array_types, + ralloc_strdup(mem_ctx, key), + (void *) t); + } + + assert(((glsl_type *) entry->data)->base_type == GLSL_TYPE_ARRAY); + assert(((glsl_type *) entry->data)->length == array_size); + assert(((glsl_type *) entry->data)->fields.array == base); + + mtx_unlock(&glsl_type::mutex); + + return (glsl_type *) entry->data; +} + + +bool +glsl_type::record_compare(const glsl_type *b) const +{ + if (this->length != b->length) + return false; + + if (this->interface_packing != b->interface_packing) + return false; + + /* From the GLSL 4.20 specification (Sec 4.2): + * + * "Structures must have the same name, sequence of type names, and + * type definitions, and field names to be considered the same type." + * + * GLSL ES behaves the same (Ver 1.00 Sec 4.2.4, Ver 3.00 Sec 4.2.5). + * + * Note that we cannot force type name check when comparing unnamed + * structure types, these have a unique name assigned during parsing. + */ + if (!this->is_anonymous() && !b->is_anonymous()) + if (strcmp(this->name, b->name) != 0) + return false; + + for (unsigned i = 0; i < this->length; i++) { + if (this->fields.structure[i].type != b->fields.structure[i].type) + return false; + if (strcmp(this->fields.structure[i].name, + b->fields.structure[i].name) != 0) + return false; + if (this->fields.structure[i].matrix_layout + != b->fields.structure[i].matrix_layout) + return false; + if (this->fields.structure[i].location + != b->fields.structure[i].location) + return false; + if (this->fields.structure[i].interpolation + != b->fields.structure[i].interpolation) + return false; + if (this->fields.structure[i].centroid + != b->fields.structure[i].centroid) + return false; + if (this->fields.structure[i].sample + != b->fields.structure[i].sample) + return false; + if (this->fields.structure[i].patch + != b->fields.structure[i].patch) + return false; + if (this->fields.structure[i].image_read_only + != b->fields.structure[i].image_read_only) + return false; + if (this->fields.structure[i].image_write_only + != b->fields.structure[i].image_write_only) + return false; + if (this->fields.structure[i].image_coherent + != b->fields.structure[i].image_coherent) + return false; + if (this->fields.structure[i].image_volatile + != b->fields.structure[i].image_volatile) + return false; + if (this->fields.structure[i].image_restrict + != b->fields.structure[i].image_restrict) + return false; + } + + return true; +} + + +bool +glsl_type::record_key_compare(const void *a, const void *b) +{ + const glsl_type *const key1 = (glsl_type *) a; + const glsl_type *const key2 = (glsl_type *) b; + + return strcmp(key1->name, key2->name) == 0 && key1->record_compare(key2); +} + + +/** + * Generate an integer hash value for a glsl_type structure type. + */ +unsigned +glsl_type::record_key_hash(const void *a) +{ + const glsl_type *const key = (glsl_type *) a; + uintptr_t hash = key->length; + unsigned retval; + + for (unsigned i = 0; i < key->length; i++) { + /* casting pointer to uintptr_t */ + hash = (hash * 13 ) + (uintptr_t) key->fields.structure[i].type; + } + + if (sizeof(hash) == 8) + retval = (hash & 0xffffffff) ^ ((uint64_t) hash >> 32); + else + retval = hash; + + return retval; +} + + +const glsl_type * +glsl_type::get_record_instance(const glsl_struct_field *fields, + unsigned num_fields, + const char *name) +{ + const glsl_type key(fields, num_fields, name); + + mtx_lock(&glsl_type::mutex); + + if (record_types == NULL) { + record_types = _mesa_hash_table_create(NULL, record_key_hash, + record_key_compare); + } + + const struct hash_entry *entry = _mesa_hash_table_search(record_types, + &key); + if (entry == NULL) { + mtx_unlock(&glsl_type::mutex); + const glsl_type *t = new glsl_type(fields, num_fields, name); + mtx_lock(&glsl_type::mutex); + + entry = _mesa_hash_table_insert(record_types, t, (void *) t); + } + + assert(((glsl_type *) entry->data)->base_type == GLSL_TYPE_STRUCT); + assert(((glsl_type *) entry->data)->length == num_fields); + assert(strcmp(((glsl_type *) entry->data)->name, name) == 0); + + mtx_unlock(&glsl_type::mutex); + + return (glsl_type *) entry->data; +} + + +const glsl_type * +glsl_type::get_interface_instance(const glsl_struct_field *fields, + unsigned num_fields, + enum glsl_interface_packing packing, + const char *block_name) +{ + const glsl_type key(fields, num_fields, packing, block_name); + + mtx_lock(&glsl_type::mutex); + + if (interface_types == NULL) { + interface_types = _mesa_hash_table_create(NULL, record_key_hash, + record_key_compare); + } + + const struct hash_entry *entry = _mesa_hash_table_search(interface_types, + &key); + if (entry == NULL) { + mtx_unlock(&glsl_type::mutex); + const glsl_type *t = new glsl_type(fields, num_fields, + packing, block_name); + mtx_lock(&glsl_type::mutex); + + entry = _mesa_hash_table_insert(interface_types, t, (void *) t); + } + + assert(((glsl_type *) entry->data)->base_type == GLSL_TYPE_INTERFACE); + assert(((glsl_type *) entry->data)->length == num_fields); + assert(strcmp(((glsl_type *) entry->data)->name, block_name) == 0); + + mtx_unlock(&glsl_type::mutex); + + return (glsl_type *) entry->data; +} + +const glsl_type * +glsl_type::get_subroutine_instance(const char *subroutine_name) +{ + const glsl_type key(subroutine_name); + + mtx_lock(&glsl_type::mutex); + + if (subroutine_types == NULL) { + subroutine_types = _mesa_hash_table_create(NULL, record_key_hash, + record_key_compare); + } + + const struct hash_entry *entry = _mesa_hash_table_search(subroutine_types, + &key); + if (entry == NULL) { + mtx_unlock(&glsl_type::mutex); + const glsl_type *t = new glsl_type(subroutine_name); + mtx_lock(&glsl_type::mutex); + + entry = _mesa_hash_table_insert(subroutine_types, t, (void *) t); + } + + assert(((glsl_type *) entry->data)->base_type == GLSL_TYPE_SUBROUTINE); + assert(strcmp(((glsl_type *) entry->data)->name, subroutine_name) == 0); + + mtx_unlock(&glsl_type::mutex); + + return (glsl_type *) entry->data; +} + + +static bool +function_key_compare(const void *a, const void *b) +{ + const glsl_type *const key1 = (glsl_type *) a; + const glsl_type *const key2 = (glsl_type *) b; + + if (key1->length != key2->length) + return 1; + + return memcmp(key1->fields.parameters, key2->fields.parameters, + (key1->length + 1) * sizeof(*key1->fields.parameters)); +} + + +static uint32_t +function_key_hash(const void *a) +{ + const glsl_type *const key = (glsl_type *) a; + char hash_key[128]; + unsigned size = 0; + + size = snprintf(hash_key, sizeof(hash_key), "%08x", key->length); + + for (unsigned i = 0; i < key->length; i++) { + if (size >= sizeof(hash_key)) + break; + + size += snprintf(& hash_key[size], sizeof(hash_key) - size, + "%p", (void *) key->fields.structure[i].type); + } + + return _mesa_hash_string(hash_key); +} + +const glsl_type * +glsl_type::get_function_instance(const glsl_type *return_type, + const glsl_function_param *params, + unsigned num_params) +{ + const glsl_type key(return_type, params, num_params); + + mtx_lock(&glsl_type::mutex); + + if (function_types == NULL) { + function_types = _mesa_hash_table_create(NULL, function_key_hash, + function_key_compare); + } + + struct hash_entry *entry = _mesa_hash_table_search(function_types, &key); + if (entry == NULL) { + mtx_unlock(&glsl_type::mutex); + const glsl_type *t = new glsl_type(return_type, params, num_params); + mtx_lock(&glsl_type::mutex); + + entry = _mesa_hash_table_insert(function_types, t, (void *) t); + } + + const glsl_type *t = (const glsl_type *)entry->data; + + assert(t->base_type == GLSL_TYPE_FUNCTION); + assert(t->length == num_params); + + mtx_unlock(&glsl_type::mutex); + + return t; +} + + +const glsl_type * +glsl_type::get_mul_type(const glsl_type *type_a, const glsl_type *type_b) +{ + if (type_a == type_b) { + return type_a; + } else if (type_a->is_matrix() && type_b->is_matrix()) { + /* Matrix multiply. The columns of A must match the rows of B. Given + * the other previously tested constraints, this means the vector type + * of a row from A must be the same as the vector type of a column from + * B. + */ + if (type_a->row_type() == type_b->column_type()) { + /* The resulting matrix has the number of columns of matrix B and + * the number of rows of matrix A. We get the row count of A by + * looking at the size of a vector that makes up a column. The + * transpose (size of a row) is done for B. + */ + const glsl_type *const type = + get_instance(type_a->base_type, + type_a->column_type()->vector_elements, + type_b->row_type()->vector_elements); + assert(type != error_type); + + return type; + } + } else if (type_a->is_matrix()) { + /* A is a matrix and B is a column vector. Columns of A must match + * rows of B. Given the other previously tested constraints, this + * means the vector type of a row from A must be the same as the + * vector the type of B. + */ + if (type_a->row_type() == type_b) { + /* The resulting vector has a number of elements equal to + * the number of rows of matrix A. */ + const glsl_type *const type = + get_instance(type_a->base_type, + type_a->column_type()->vector_elements, + 1); + assert(type != error_type); + + return type; + } + } else { + assert(type_b->is_matrix()); + + /* A is a row vector and B is a matrix. Columns of A must match rows + * of B. Given the other previously tested constraints, this means + * the type of A must be the same as the vector type of a column from + * B. + */ + if (type_a == type_b->column_type()) { + /* The resulting vector has a number of elements equal to + * the number of columns of matrix B. */ + const glsl_type *const type = + get_instance(type_a->base_type, + type_b->row_type()->vector_elements, + 1); + assert(type != error_type); + + return type; + } + } + + return error_type; +} + + +const glsl_type * +glsl_type::field_type(const char *name) const +{ + if (this->base_type != GLSL_TYPE_STRUCT + && this->base_type != GLSL_TYPE_INTERFACE) + return error_type; + + for (unsigned i = 0; i < this->length; i++) { + if (strcmp(name, this->fields.structure[i].name) == 0) + return this->fields.structure[i].type; + } + + return error_type; +} + + +int +glsl_type::field_index(const char *name) const +{ + if (this->base_type != GLSL_TYPE_STRUCT + && this->base_type != GLSL_TYPE_INTERFACE) + return -1; + + for (unsigned i = 0; i < this->length; i++) { + if (strcmp(name, this->fields.structure[i].name) == 0) + return i; + } + + return -1; +} + + +unsigned +glsl_type::component_slots() const +{ + switch (this->base_type) { + case GLSL_TYPE_UINT: + case GLSL_TYPE_INT: + case GLSL_TYPE_FLOAT: + case GLSL_TYPE_BOOL: + return this->components(); + + case GLSL_TYPE_DOUBLE: + return 2 * this->components(); + + case GLSL_TYPE_STRUCT: + case GLSL_TYPE_INTERFACE: { + unsigned size = 0; + + for (unsigned i = 0; i < this->length; i++) + size += this->fields.structure[i].type->component_slots(); + + return size; + } + + case GLSL_TYPE_ARRAY: + return this->length * this->fields.array->component_slots(); + + case GLSL_TYPE_IMAGE: + return 1; + case GLSL_TYPE_SUBROUTINE: + return 1; + + case GLSL_TYPE_FUNCTION: + case GLSL_TYPE_SAMPLER: + case GLSL_TYPE_ATOMIC_UINT: + case GLSL_TYPE_VOID: + case GLSL_TYPE_ERROR: + break; + } + + return 0; +} + +unsigned +glsl_type::record_location_offset(unsigned length) const +{ + unsigned offset = 0; + const glsl_type *t = this->without_array(); + if (t->is_record()) { + assert(length <= t->length); + + for (unsigned i = 0; i < length; i++) { + const glsl_type *st = t->fields.structure[i].type; + const glsl_type *wa = st->without_array(); + if (wa->is_record()) { + unsigned r_offset = wa->record_location_offset(wa->length); + offset += st->is_array() ? + st->arrays_of_arrays_size() * r_offset : r_offset; + } else if (st->is_array() && st->fields.array->is_array()) { + unsigned outer_array_size = st->length; + const glsl_type *base_type = st->fields.array; + + /* For arrays of arrays the outer arrays take up a uniform + * slot for each element. The innermost array elements share a + * single slot so we ignore the innermost array when calculating + * the offset. + */ + while (base_type->fields.array->is_array()) { + outer_array_size = outer_array_size * base_type->length; + base_type = base_type->fields.array; + } + offset += outer_array_size; + } else { + /* We dont worry about arrays here because unless the array + * contains a structure or another array it only takes up a single + * uniform slot. + */ + offset += 1; + } + } + } + return offset; +} + +unsigned +glsl_type::uniform_locations() const +{ + unsigned size = 0; + + switch (this->base_type) { + case GLSL_TYPE_UINT: + case GLSL_TYPE_INT: + case GLSL_TYPE_FLOAT: + case GLSL_TYPE_DOUBLE: + case GLSL_TYPE_BOOL: + case GLSL_TYPE_SAMPLER: + case GLSL_TYPE_IMAGE: + case GLSL_TYPE_SUBROUTINE: + return 1; + + case GLSL_TYPE_STRUCT: + case GLSL_TYPE_INTERFACE: + for (unsigned i = 0; i < this->length; i++) + size += this->fields.structure[i].type->uniform_locations(); + return size; + case GLSL_TYPE_ARRAY: + return this->length * this->fields.array->uniform_locations(); + default: + return 0; + } +} + +bool +glsl_type::can_implicitly_convert_to(const glsl_type *desired, + _mesa_glsl_parse_state *state) const +{ + if (this == desired) + return true; + + /* There is no conversion among matrix types. */ + if (this->matrix_columns > 1 || desired->matrix_columns > 1) + return false; + + /* Vector size must match. */ + if (this->vector_elements != desired->vector_elements) + return false; + + /* int and uint can be converted to float. */ + if (desired->is_float() && this->is_integer()) + return true; + + /* With GLSL 4.0 / ARB_gpu_shader5, int can be converted to uint. + * Note that state may be NULL here, when resolving function calls in the + * linker. By this time, all the state-dependent checks have already + * happened though, so allow anything that's allowed in any shader version. */ + if ((!state || state->is_version(400, 0) || state->ARB_gpu_shader5_enable) && + desired->base_type == GLSL_TYPE_UINT && this->base_type == GLSL_TYPE_INT) + return true; + + /* No implicit conversions from double. */ + if ((!state || state->has_double()) && this->is_double()) + return false; + + /* Conversions from different types to double. */ + if ((!state || state->has_double()) && desired->is_double()) { + if (this->is_float()) + return true; + if (this->is_integer()) + return true; + } + + return false; +} + +unsigned +glsl_type::std140_base_alignment(bool row_major) const +{ + unsigned N = is_double() ? 8 : 4; + + /* (1) If the member is a scalar consuming <N> basic machine units, the + * base alignment is <N>. + * + * (2) If the member is a two- or four-component vector with components + * consuming <N> basic machine units, the base alignment is 2<N> or + * 4<N>, respectively. + * + * (3) If the member is a three-component vector with components consuming + * <N> basic machine units, the base alignment is 4<N>. + */ + if (this->is_scalar() || this->is_vector()) { + switch (this->vector_elements) { + case 1: + return N; + case 2: + return 2 * N; + case 3: + case 4: + return 4 * N; + } + } + + /* (4) If the member is an array of scalars or vectors, the base alignment + * and array stride are set to match the base alignment of a single + * array element, according to rules (1), (2), and (3), and rounded up + * to the base alignment of a vec4. The array may have padding at the + * end; the base offset of the member following the array is rounded up + * to the next multiple of the base alignment. + * + * (6) If the member is an array of <S> column-major matrices with <C> + * columns and <R> rows, the matrix is stored identically to a row of + * <S>*<C> column vectors with <R> components each, according to rule + * (4). + * + * (8) If the member is an array of <S> row-major matrices with <C> columns + * and <R> rows, the matrix is stored identically to a row of <S>*<R> + * row vectors with <C> components each, according to rule (4). + * + * (10) If the member is an array of <S> structures, the <S> elements of + * the array are laid out in order, according to rule (9). + */ + if (this->is_array()) { + if (this->fields.array->is_scalar() || + this->fields.array->is_vector() || + this->fields.array->is_matrix()) { + return MAX2(this->fields.array->std140_base_alignment(row_major), 16); + } else { + assert(this->fields.array->is_record() || + this->fields.array->is_array()); + return this->fields.array->std140_base_alignment(row_major); + } + } + + /* (5) If the member is a column-major matrix with <C> columns and + * <R> rows, the matrix is stored identically to an array of + * <C> column vectors with <R> components each, according to + * rule (4). + * + * (7) If the member is a row-major matrix with <C> columns and <R> + * rows, the matrix is stored identically to an array of <R> + * row vectors with <C> components each, according to rule (4). + */ + if (this->is_matrix()) { + const struct glsl_type *vec_type, *array_type; + int c = this->matrix_columns; + int r = this->vector_elements; + + if (row_major) { + vec_type = get_instance(base_type, c, 1); + array_type = glsl_type::get_array_instance(vec_type, r); + } else { + vec_type = get_instance(base_type, r, 1); + array_type = glsl_type::get_array_instance(vec_type, c); + } + + return array_type->std140_base_alignment(false); + } + + /* (9) If the member is a structure, the base alignment of the + * structure is <N>, where <N> is the largest base alignment + * value of any of its members, and rounded up to the base + * alignment of a vec4. The individual members of this + * sub-structure are then assigned offsets by applying this set + * of rules recursively, where the base offset of the first + * member of the sub-structure is equal to the aligned offset + * of the structure. The structure may have padding at the end; + * the base offset of the member following the sub-structure is + * rounded up to the next multiple of the base alignment of the + * structure. + */ + if (this->is_record()) { + unsigned base_alignment = 16; + for (unsigned i = 0; i < this->length; i++) { + bool field_row_major = row_major; + const enum glsl_matrix_layout matrix_layout = + glsl_matrix_layout(this->fields.structure[i].matrix_layout); + if (matrix_layout == GLSL_MATRIX_LAYOUT_ROW_MAJOR) { + field_row_major = true; + } else if (matrix_layout == GLSL_MATRIX_LAYOUT_COLUMN_MAJOR) { + field_row_major = false; + } + + const struct glsl_type *field_type = this->fields.structure[i].type; + base_alignment = MAX2(base_alignment, + field_type->std140_base_alignment(field_row_major)); + } + return base_alignment; + } + + assert(!"not reached"); + return -1; +} + +unsigned +glsl_type::std140_size(bool row_major) const +{ + unsigned N = is_double() ? 8 : 4; + + /* (1) If the member is a scalar consuming <N> basic machine units, the + * base alignment is <N>. + * + * (2) If the member is a two- or four-component vector with components + * consuming <N> basic machine units, the base alignment is 2<N> or + * 4<N>, respectively. + * + * (3) If the member is a three-component vector with components consuming + * <N> basic machine units, the base alignment is 4<N>. + */ + if (this->is_scalar() || this->is_vector()) { + return this->vector_elements * N; + } + + /* (5) If the member is a column-major matrix with <C> columns and + * <R> rows, the matrix is stored identically to an array of + * <C> column vectors with <R> components each, according to + * rule (4). + * + * (6) If the member is an array of <S> column-major matrices with <C> + * columns and <R> rows, the matrix is stored identically to a row of + * <S>*<C> column vectors with <R> components each, according to rule + * (4). + * + * (7) If the member is a row-major matrix with <C> columns and <R> + * rows, the matrix is stored identically to an array of <R> + * row vectors with <C> components each, according to rule (4). + * + * (8) If the member is an array of <S> row-major matrices with <C> columns + * and <R> rows, the matrix is stored identically to a row of <S>*<R> + * row vectors with <C> components each, according to rule (4). + */ + if (this->without_array()->is_matrix()) { + const struct glsl_type *element_type; + const struct glsl_type *vec_type; + unsigned int array_len; + + if (this->is_array()) { + element_type = this->without_array(); + array_len = this->arrays_of_arrays_size(); + } else { + element_type = this; + array_len = 1; + } + + if (row_major) { + vec_type = get_instance(element_type->base_type, + element_type->matrix_columns, 1); + + array_len *= element_type->vector_elements; + } else { + vec_type = get_instance(element_type->base_type, + element_type->vector_elements, 1); + array_len *= element_type->matrix_columns; + } + const glsl_type *array_type = glsl_type::get_array_instance(vec_type, + array_len); + + return array_type->std140_size(false); + } + + /* (4) If the member is an array of scalars or vectors, the base alignment + * and array stride are set to match the base alignment of a single + * array element, according to rules (1), (2), and (3), and rounded up + * to the base alignment of a vec4. The array may have padding at the + * end; the base offset of the member following the array is rounded up + * to the next multiple of the base alignment. + * + * (10) If the member is an array of <S> structures, the <S> elements of + * the array are laid out in order, according to rule (9). + */ + if (this->is_array()) { + if (this->without_array()->is_record()) { + return this->arrays_of_arrays_size() * + this->without_array()->std140_size(row_major); + } else { + unsigned element_base_align = + this->without_array()->std140_base_alignment(row_major); + return this->arrays_of_arrays_size() * MAX2(element_base_align, 16); + } + } + + /* (9) If the member is a structure, the base alignment of the + * structure is <N>, where <N> is the largest base alignment + * value of any of its members, and rounded up to the base + * alignment of a vec4. The individual members of this + * sub-structure are then assigned offsets by applying this set + * of rules recursively, where the base offset of the first + * member of the sub-structure is equal to the aligned offset + * of the structure. The structure may have padding at the end; + * the base offset of the member following the sub-structure is + * rounded up to the next multiple of the base alignment of the + * structure. + */ + if (this->is_record() || this->is_interface()) { + unsigned size = 0; + unsigned max_align = 0; + + for (unsigned i = 0; i < this->length; i++) { + bool field_row_major = row_major; + const enum glsl_matrix_layout matrix_layout = + glsl_matrix_layout(this->fields.structure[i].matrix_layout); + if (matrix_layout == GLSL_MATRIX_LAYOUT_ROW_MAJOR) { + field_row_major = true; + } else if (matrix_layout == GLSL_MATRIX_LAYOUT_COLUMN_MAJOR) { + field_row_major = false; + } + + const struct glsl_type *field_type = this->fields.structure[i].type; + unsigned align = field_type->std140_base_alignment(field_row_major); + + /* Ignore unsized arrays when calculating size */ + if (field_type->is_unsized_array()) + continue; + + size = glsl_align(size, align); + size += field_type->std140_size(field_row_major); + + max_align = MAX2(align, max_align); + + if (field_type->is_record() && (i + 1 < this->length)) + size = glsl_align(size, 16); + } + size = glsl_align(size, MAX2(max_align, 16)); + return size; + } + + assert(!"not reached"); + return -1; +} + +unsigned +glsl_type::std430_base_alignment(bool row_major) const +{ + + unsigned N = is_double() ? 8 : 4; + + /* (1) If the member is a scalar consuming <N> basic machine units, the + * base alignment is <N>. + * + * (2) If the member is a two- or four-component vector with components + * consuming <N> basic machine units, the base alignment is 2<N> or + * 4<N>, respectively. + * + * (3) If the member is a three-component vector with components consuming + * <N> basic machine units, the base alignment is 4<N>. + */ + if (this->is_scalar() || this->is_vector()) { + switch (this->vector_elements) { + case 1: + return N; + case 2: + return 2 * N; + case 3: + case 4: + return 4 * N; + } + } + + /* OpenGL 4.30 spec, section 7.6.2.2 "Standard Uniform Block Layout": + * + * "When using the std430 storage layout, shader storage blocks will be + * laid out in buffer storage identically to uniform and shader storage + * blocks using the std140 layout, except that the base alignment and + * stride of arrays of scalars and vectors in rule 4 and of structures + * in rule 9 are not rounded up a multiple of the base alignment of a vec4. + */ + + /* (1) If the member is a scalar consuming <N> basic machine units, the + * base alignment is <N>. + * + * (2) If the member is a two- or four-component vector with components + * consuming <N> basic machine units, the base alignment is 2<N> or + * 4<N>, respectively. + * + * (3) If the member is a three-component vector with components consuming + * <N> basic machine units, the base alignment is 4<N>. + */ + if (this->is_array()) + return this->fields.array->std430_base_alignment(row_major); + + /* (5) If the member is a column-major matrix with <C> columns and + * <R> rows, the matrix is stored identically to an array of + * <C> column vectors with <R> components each, according to + * rule (4). + * + * (7) If the member is a row-major matrix with <C> columns and <R> + * rows, the matrix is stored identically to an array of <R> + * row vectors with <C> components each, according to rule (4). + */ + if (this->is_matrix()) { + const struct glsl_type *vec_type, *array_type; + int c = this->matrix_columns; + int r = this->vector_elements; + + if (row_major) { + vec_type = get_instance(base_type, c, 1); + array_type = glsl_type::get_array_instance(vec_type, r); + } else { + vec_type = get_instance(base_type, r, 1); + array_type = glsl_type::get_array_instance(vec_type, c); + } + + return array_type->std430_base_alignment(false); + } + + /* (9) If the member is a structure, the base alignment of the + * structure is <N>, where <N> is the largest base alignment + * value of any of its members, and rounded up to the base + * alignment of a vec4. The individual members of this + * sub-structure are then assigned offsets by applying this set + * of rules recursively, where the base offset of the first + * member of the sub-structure is equal to the aligned offset + * of the structure. The structure may have padding at the end; + * the base offset of the member following the sub-structure is + * rounded up to the next multiple of the base alignment of the + * structure. + */ + if (this->is_record()) { + unsigned base_alignment = 0; + for (unsigned i = 0; i < this->length; i++) { + bool field_row_major = row_major; + const enum glsl_matrix_layout matrix_layout = + glsl_matrix_layout(this->fields.structure[i].matrix_layout); + if (matrix_layout == GLSL_MATRIX_LAYOUT_ROW_MAJOR) { + field_row_major = true; + } else if (matrix_layout == GLSL_MATRIX_LAYOUT_COLUMN_MAJOR) { + field_row_major = false; + } + + const struct glsl_type *field_type = this->fields.structure[i].type; + base_alignment = MAX2(base_alignment, + field_type->std430_base_alignment(field_row_major)); + } + assert(base_alignment > 0); + return base_alignment; + } + assert(!"not reached"); + return -1; +} + +unsigned +glsl_type::std430_array_stride(bool row_major) const +{ + unsigned N = is_double() ? 8 : 4; + + /* Notice that the array stride of a vec3 is not 3 * N but 4 * N. + * See OpenGL 4.30 spec, section 7.6.2.2 "Standard Uniform Block Layout" + * + * (3) If the member is a three-component vector with components consuming + * <N> basic machine units, the base alignment is 4<N>. + */ + if (this->is_vector() && this->vector_elements == 3) + return 4 * N; + + /* By default use std430_size(row_major) */ + return this->std430_size(row_major); +} + +unsigned +glsl_type::std430_size(bool row_major) const +{ + unsigned N = is_double() ? 8 : 4; + + /* OpenGL 4.30 spec, section 7.6.2.2 "Standard Uniform Block Layout": + * + * "When using the std430 storage layout, shader storage blocks will be + * laid out in buffer storage identically to uniform and shader storage + * blocks using the std140 layout, except that the base alignment and + * stride of arrays of scalars and vectors in rule 4 and of structures + * in rule 9 are not rounded up a multiple of the base alignment of a vec4. + */ + if (this->is_scalar() || this->is_vector()) + return this->vector_elements * N; + + if (this->without_array()->is_matrix()) { + const struct glsl_type *element_type; + const struct glsl_type *vec_type; + unsigned int array_len; + + if (this->is_array()) { + element_type = this->without_array(); + array_len = this->arrays_of_arrays_size(); + } else { + element_type = this; + array_len = 1; + } + + if (row_major) { + vec_type = get_instance(element_type->base_type, + element_type->matrix_columns, 1); + + array_len *= element_type->vector_elements; + } else { + vec_type = get_instance(element_type->base_type, + element_type->vector_elements, 1); + array_len *= element_type->matrix_columns; + } + const glsl_type *array_type = glsl_type::get_array_instance(vec_type, + array_len); + + return array_type->std430_size(false); + } + + if (this->is_array()) { + if (this->without_array()->is_record()) + return this->arrays_of_arrays_size() * + this->without_array()->std430_size(row_major); + else + return this->arrays_of_arrays_size() * + this->without_array()->std430_base_alignment(row_major); + } + + if (this->is_record() || this->is_interface()) { + unsigned size = 0; + unsigned max_align = 0; + + for (unsigned i = 0; i < this->length; i++) { + bool field_row_major = row_major; + const enum glsl_matrix_layout matrix_layout = + glsl_matrix_layout(this->fields.structure[i].matrix_layout); + if (matrix_layout == GLSL_MATRIX_LAYOUT_ROW_MAJOR) { + field_row_major = true; + } else if (matrix_layout == GLSL_MATRIX_LAYOUT_COLUMN_MAJOR) { + field_row_major = false; + } + + const struct glsl_type *field_type = this->fields.structure[i].type; + unsigned align = field_type->std430_base_alignment(field_row_major); + size = glsl_align(size, align); + size += field_type->std430_size(field_row_major); + + max_align = MAX2(align, max_align); + } + size = glsl_align(size, max_align); + return size; + } + + assert(!"not reached"); + return -1; +} + +unsigned +glsl_type::count_attribute_slots() const +{ + /* From page 31 (page 37 of the PDF) of the GLSL 1.50 spec: + * + * "A scalar input counts the same amount against this limit as a vec4, + * so applications may want to consider packing groups of four + * unrelated float inputs together into a vector to better utilize the + * capabilities of the underlying hardware. A matrix input will use up + * multiple locations. The number of locations used will equal the + * number of columns in the matrix." + * + * The spec does not explicitly say how arrays are counted. However, it + * should be safe to assume the total number of slots consumed by an array + * is the number of entries in the array multiplied by the number of slots + * consumed by a single element of the array. + * + * The spec says nothing about how structs are counted, because vertex + * attributes are not allowed to be (or contain) structs. However, Mesa + * allows varying structs, the number of varying slots taken up by a + * varying struct is simply equal to the sum of the number of slots taken + * up by each element. + */ + switch (this->base_type) { + case GLSL_TYPE_UINT: + case GLSL_TYPE_INT: + case GLSL_TYPE_FLOAT: + case GLSL_TYPE_BOOL: + case GLSL_TYPE_DOUBLE: + return this->matrix_columns; + + case GLSL_TYPE_STRUCT: + case GLSL_TYPE_INTERFACE: { + unsigned size = 0; + + for (unsigned i = 0; i < this->length; i++) + size += this->fields.structure[i].type->count_attribute_slots(); + + return size; + } + + case GLSL_TYPE_ARRAY: + return this->length * this->fields.array->count_attribute_slots(); + + case GLSL_TYPE_FUNCTION: + case GLSL_TYPE_SAMPLER: + case GLSL_TYPE_IMAGE: + case GLSL_TYPE_ATOMIC_UINT: + case GLSL_TYPE_VOID: + case GLSL_TYPE_SUBROUTINE: + case GLSL_TYPE_ERROR: + break; + } + + assert(!"Unexpected type in count_attribute_slots()"); + + return 0; +} + +int +glsl_type::coordinate_components() const +{ + int size; + + switch (sampler_dimensionality) { + case GLSL_SAMPLER_DIM_1D: + case GLSL_SAMPLER_DIM_BUF: + size = 1; + break; + case GLSL_SAMPLER_DIM_2D: + case GLSL_SAMPLER_DIM_RECT: + case GLSL_SAMPLER_DIM_MS: + case GLSL_SAMPLER_DIM_EXTERNAL: + size = 2; + break; + case GLSL_SAMPLER_DIM_3D: + case GLSL_SAMPLER_DIM_CUBE: + size = 3; + break; + default: + assert(!"Should not get here."); + size = 1; + break; + } + + /* Array textures need an additional component for the array index, except + * for cubemap array images that behave like a 2D array of interleaved + * cubemap faces. + */ + if (sampler_array && + !(base_type == GLSL_TYPE_IMAGE && + sampler_dimensionality == GLSL_SAMPLER_DIM_CUBE)) + size += 1; + + return size; +} + +/** + * Declarations of type flyweights (glsl_type::_foo_type) and + * convenience pointers (glsl_type::foo_type). + * @{ + */ +#define DECL_TYPE(NAME, ...) \ + const glsl_type glsl_type::_##NAME##_type = glsl_type(__VA_ARGS__, #NAME); \ + const glsl_type *const glsl_type::NAME##_type = &glsl_type::_##NAME##_type; + +#define STRUCT_TYPE(NAME) + +#include "builtin_type_macros.h" +/** @} */ diff --git a/src/glsl/nir/glsl_types.h b/src/glsl/nir/glsl_types.h new file mode 100644 index 00000000000..b83e1ca3d2c --- /dev/null +++ b/src/glsl/nir/glsl_types.h @@ -0,0 +1,890 @@ +/* -*- c++ -*- */ +/* + * 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 +#ifndef GLSL_TYPES_H +#define GLSL_TYPES_H + +#include <string.h> +#include <assert.h> + +#ifdef __cplusplus +extern "C" { +#endif + +struct _mesa_glsl_parse_state; +struct glsl_symbol_table; + +extern void +_mesa_glsl_initialize_types(struct _mesa_glsl_parse_state *state); + +extern void +_mesa_glsl_release_types(void); + +#ifdef __cplusplus +} +#endif + +enum glsl_base_type { + GLSL_TYPE_UINT = 0, + GLSL_TYPE_INT, + GLSL_TYPE_FLOAT, + GLSL_TYPE_DOUBLE, + GLSL_TYPE_BOOL, + GLSL_TYPE_SAMPLER, + GLSL_TYPE_IMAGE, + GLSL_TYPE_ATOMIC_UINT, + GLSL_TYPE_STRUCT, + GLSL_TYPE_FUNCTION, + GLSL_TYPE_INTERFACE, + GLSL_TYPE_ARRAY, + GLSL_TYPE_VOID, + GLSL_TYPE_SUBROUTINE, + GLSL_TYPE_ERROR +}; + +enum glsl_sampler_dim { + GLSL_SAMPLER_DIM_1D = 0, + GLSL_SAMPLER_DIM_2D, + GLSL_SAMPLER_DIM_3D, + GLSL_SAMPLER_DIM_CUBE, + GLSL_SAMPLER_DIM_RECT, + GLSL_SAMPLER_DIM_BUF, + GLSL_SAMPLER_DIM_EXTERNAL, + GLSL_SAMPLER_DIM_MS +}; + +enum glsl_interface_packing { + GLSL_INTERFACE_PACKING_STD140, + GLSL_INTERFACE_PACKING_SHARED, + GLSL_INTERFACE_PACKING_PACKED, + GLSL_INTERFACE_PACKING_STD430 +}; + +enum glsl_matrix_layout { + /** + * The layout of the matrix is inherited from the object containing the + * matrix (the top level structure or the uniform block). + */ + GLSL_MATRIX_LAYOUT_INHERITED, + + /** + * Explicit column-major layout + * + * If a uniform block doesn't have an explicit layout set, it will default + * to this layout. + */ + GLSL_MATRIX_LAYOUT_COLUMN_MAJOR, + + /** + * Row-major layout + */ + GLSL_MATRIX_LAYOUT_ROW_MAJOR +}; + +#ifdef __cplusplus +#include "GL/gl.h" +#include "util/ralloc.h" +#include "main/mtypes.h" /* for gl_texture_index, C++'s enum rules are broken */ + +struct glsl_type { + GLenum gl_type; + glsl_base_type base_type; + + unsigned sampler_dimensionality:3; /**< \see glsl_sampler_dim */ + unsigned sampler_shadow:1; + unsigned sampler_array:1; + unsigned sampler_type:2; /**< Type of data returned using this + * sampler or image. Only \c + * GLSL_TYPE_FLOAT, \c GLSL_TYPE_INT, + * and \c GLSL_TYPE_UINT are valid. + */ + unsigned interface_packing:2; + + /* Callers of this ralloc-based new need not call delete. It's + * easier to just ralloc_free 'mem_ctx' (or any of its ancestors). */ + static void* operator new(size_t size) + { + mtx_lock(&glsl_type::mutex); + + /* mem_ctx should have been created by the static members */ + assert(glsl_type::mem_ctx != NULL); + + void *type; + + type = ralloc_size(glsl_type::mem_ctx, size); + assert(type != NULL); + + mtx_unlock(&glsl_type::mutex); + + return type; + } + + /* If the user *does* call delete, that's OK, we will just + * ralloc_free in that case. */ + static void operator delete(void *type) + { + mtx_lock(&glsl_type::mutex); + ralloc_free(type); + mtx_unlock(&glsl_type::mutex); + } + + /** + * \name Vector and matrix element counts + * + * For scalars, each of these values will be 1. For non-numeric types + * these will be 0. + */ + /*@{*/ + uint8_t vector_elements; /**< 1, 2, 3, or 4 vector elements. */ + uint8_t matrix_columns; /**< 1, 2, 3, or 4 matrix columns. */ + /*@}*/ + + /** + * For \c GLSL_TYPE_ARRAY, this is the length of the array. For + * \c GLSL_TYPE_STRUCT or \c GLSL_TYPE_INTERFACE, it is the number of + * elements in the structure and the number of values pointed to by + * \c fields.structure (below). + */ + unsigned length; + + /** + * Name of the data type + * + * Will never be \c NULL. + */ + const char *name; + + /** + * Subtype of composite data types. + */ + union { + const struct glsl_type *array; /**< Type of array elements. */ + struct glsl_function_param *parameters; /**< Parameters to function. */ + struct glsl_struct_field *structure; /**< List of struct fields. */ + } fields; + + /** + * \name Pointers to various public type singletons + */ + /*@{*/ +#undef DECL_TYPE +#define DECL_TYPE(NAME, ...) \ + static const glsl_type *const NAME##_type; +#undef STRUCT_TYPE +#define STRUCT_TYPE(NAME) \ + static const glsl_type *const struct_##NAME##_type; +#include "builtin_type_macros.h" + /*@}*/ + + /** + * Convenience accessors for vector types (shorter than get_instance()). + * @{ + */ + static const glsl_type *vec(unsigned components); + static const glsl_type *dvec(unsigned components); + static const glsl_type *ivec(unsigned components); + static const glsl_type *uvec(unsigned components); + static const glsl_type *bvec(unsigned components); + /**@}*/ + + /** + * For numeric and boolean derived types returns the basic scalar type + * + * If the type is a numeric or boolean scalar, vector, or matrix type, + * this function gets the scalar type of the individual components. For + * all other types, including arrays of numeric or boolean types, the + * error type is returned. + */ + const glsl_type *get_base_type() const; + + /** + * Get the basic scalar type which this type aggregates. + * + * If the type is a numeric or boolean scalar, vector, or matrix, or an + * array of any of those, this function gets the scalar type of the + * individual components. For structs and arrays of structs, this function + * returns the struct type. For samplers and arrays of samplers, this + * function returns the sampler type. + */ + const glsl_type *get_scalar_type() const; + + /** + * Get the instance of a built-in scalar, vector, or matrix type + */ + static const glsl_type *get_instance(unsigned base_type, unsigned rows, + unsigned columns); + + /** + * Get the instance of a sampler type + */ + static const glsl_type *get_sampler_instance(enum glsl_sampler_dim dim, + bool shadow, + bool array, + glsl_base_type type); + + + /** + * Get the instance of an array type + */ + static const glsl_type *get_array_instance(const glsl_type *base, + unsigned elements); + + /** + * Get the instance of a record type + */ + static const glsl_type *get_record_instance(const glsl_struct_field *fields, + unsigned num_fields, + const char *name); + + /** + * Get the instance of an interface block type + */ + static const glsl_type *get_interface_instance(const glsl_struct_field *fields, + unsigned num_fields, + enum glsl_interface_packing packing, + const char *block_name); + + /** + * Get the instance of an subroutine type + */ + static const glsl_type *get_subroutine_instance(const char *subroutine_name); + + /** + * Get the instance of a function type + */ + static const glsl_type *get_function_instance(const struct glsl_type *return_type, + const glsl_function_param *parameters, + unsigned num_params); + + /** + * Get the type resulting from a multiplication of \p type_a * \p type_b + */ + static const glsl_type *get_mul_type(const glsl_type *type_a, + const glsl_type *type_b); + + /** + * Query the total number of scalars that make up a scalar, vector or matrix + */ + unsigned components() const + { + return vector_elements * matrix_columns; + } + + /** + * Calculate the number of components slots required to hold this type + * + * This is used to determine how many uniform or varying locations a type + * might occupy. + */ + unsigned component_slots() const; + + /** + * Calculate offset between the base location of the struct in + * uniform storage and a struct member. + * For the initial call, length is the index of the member to find the + * offset for. + */ + unsigned record_location_offset(unsigned length) const; + + /** + * Calculate the number of unique values from glGetUniformLocation for the + * elements of the type. + * + * This is used to allocate slots in the UniformRemapTable, the amount of + * locations may not match with actual used storage space by the driver. + */ + unsigned uniform_locations() const; + + /** + * Calculate the number of attribute slots required to hold this type + * + * This implements the language rules of GLSL 1.50 for counting the number + * of slots used by a vertex attribute. It also determines the number of + * varying slots the type will use up in the absence of varying packing + * (and thus, it can be used to measure the number of varying slots used by + * the varyings that are generated by lower_packed_varyings). + */ + unsigned count_attribute_slots() const; + + + /** + * Alignment in bytes of the start of this type in a std140 uniform + * block. + */ + unsigned std140_base_alignment(bool row_major) const; + + /** Size in bytes of this type in a std140 uniform block. + * + * Note that this is not GL_UNIFORM_SIZE (which is the number of + * elements in the array) + */ + unsigned std140_size(bool row_major) const; + + /** + * Alignment in bytes of the start of this type in a std430 shader + * storage block. + */ + unsigned std430_base_alignment(bool row_major) const; + + /** + * Calculate array stride in bytes of this type in a std430 shader storage + * block. + */ + unsigned std430_array_stride(bool row_major) const; + + /** + * Size in bytes of this type in a std430 shader storage block. + * + * Note that this is not GL_BUFFER_SIZE + */ + unsigned std430_size(bool row_major) const; + + /** + * \brief Can this type be implicitly converted to another? + * + * \return True if the types are identical or if this type can be converted + * to \c desired according to Section 4.1.10 of the GLSL spec. + * + * \verbatim + * From page 25 (31 of the pdf) of the GLSL 1.50 spec, Section 4.1.10 + * Implicit Conversions: + * + * In some situations, an expression and its type will be implicitly + * converted to a different type. The following table shows all allowed + * implicit conversions: + * + * Type of expression | Can be implicitly converted to + * -------------------------------------------------- + * int float + * uint + * + * ivec2 vec2 + * uvec2 + * + * ivec3 vec3 + * uvec3 + * + * ivec4 vec4 + * uvec4 + * + * There are no implicit array or structure conversions. For example, + * an array of int cannot be implicitly converted to an array of float. + * There are no implicit conversions between signed and unsigned + * integers. + * \endverbatim + */ + bool can_implicitly_convert_to(const glsl_type *desired, + _mesa_glsl_parse_state *state) const; + + /** + * Query whether or not a type is a scalar (non-vector and non-matrix). + */ + bool is_scalar() const + { + return (vector_elements == 1) + && (base_type >= GLSL_TYPE_UINT) + && (base_type <= GLSL_TYPE_BOOL); + } + + /** + * Query whether or not a type is a vector + */ + bool is_vector() const + { + return (vector_elements > 1) + && (matrix_columns == 1) + && (base_type >= GLSL_TYPE_UINT) + && (base_type <= GLSL_TYPE_BOOL); + } + + /** + * Query whether or not a type is a matrix + */ + bool is_matrix() const + { + /* GLSL only has float matrices. */ + return (matrix_columns > 1) && (base_type == GLSL_TYPE_FLOAT || base_type == GLSL_TYPE_DOUBLE); + } + + /** + * Query whether or not a type is a non-array numeric type + */ + bool is_numeric() const + { + return (base_type >= GLSL_TYPE_UINT) && (base_type <= GLSL_TYPE_DOUBLE); + } + + /** + * Query whether or not a type is an integral type + */ + bool is_integer() const + { + return (base_type == GLSL_TYPE_UINT) || (base_type == GLSL_TYPE_INT); + } + + /** + * Query whether or not type is an integral type, or for struct and array + * types, contains an integral type. + */ + bool contains_integer() const; + + /** + * Query whether or not type is a double type, or for struct and array + * types, contains a double type. + */ + bool contains_double() const; + + /** + * Query whether or not a type is a float type + */ + bool is_float() const + { + return base_type == GLSL_TYPE_FLOAT; + } + + /** + * Query whether or not a type is a double type + */ + bool is_double() const + { + return base_type == GLSL_TYPE_DOUBLE; + } + + /** + * Query whether or not a type is a non-array boolean type + */ + bool is_boolean() const + { + return base_type == GLSL_TYPE_BOOL; + } + + /** + * Query whether or not a type is a sampler + */ + bool is_sampler() const + { + return base_type == GLSL_TYPE_SAMPLER; + } + + /** + * Query whether or not type is a sampler, or for struct and array + * types, contains a sampler. + */ + bool contains_sampler() const; + + /** + * Get the Mesa texture target index for a sampler type. + */ + gl_texture_index sampler_index() const; + + /** + * Query whether or not type is an image, or for struct and array + * types, contains an image. + */ + bool contains_image() const; + + /** + * Query whether or not a type is an image + */ + bool is_image() const + { + return base_type == GLSL_TYPE_IMAGE; + } + + /** + * Query whether or not a type is an array + */ + bool is_array() const + { + return base_type == GLSL_TYPE_ARRAY; + } + + /** + * Query whether or not a type is a record + */ + bool is_record() const + { + return base_type == GLSL_TYPE_STRUCT; + } + + /** + * Query whether or not a type is an interface + */ + bool is_interface() const + { + return base_type == GLSL_TYPE_INTERFACE; + } + + /** + * Query whether or not a type is the void type singleton. + */ + bool is_void() const + { + return base_type == GLSL_TYPE_VOID; + } + + /** + * Query whether or not a type is the error type singleton. + */ + bool is_error() const + { + return base_type == GLSL_TYPE_ERROR; + } + + /** + * Query if a type is unnamed/anonymous (named by the parser) + */ + + bool is_subroutine() const + { + return base_type == GLSL_TYPE_SUBROUTINE; + } + bool contains_subroutine() const; + + bool is_anonymous() const + { + return !strncmp(name, "#anon", 5); + } + + /** + * Get the type stripped of any arrays + * + * \return + * Pointer to the type of elements of the first non-array type for array + * types, or pointer to itself for non-array types. + */ + const glsl_type *without_array() const + { + const glsl_type *t = this; + + while (t->is_array()) + t = t->fields.array; + + return t; + } + + /** + * Return the total number of elements in an array including the elements + * in arrays of arrays. + */ + unsigned arrays_of_arrays_size() const + { + if (!is_array()) + return 0; + + unsigned size = length; + const glsl_type *base_type = fields.array; + + while (base_type->is_array()) { + size = size * base_type->length; + base_type = base_type->fields.array; + } + return size; + } + + /** + * Return the amount of atomic counter storage required for a type. + */ + unsigned atomic_size() const + { + if (base_type == GLSL_TYPE_ATOMIC_UINT) + return ATOMIC_COUNTER_SIZE; + else if (is_array()) + return length * fields.array->atomic_size(); + else + return 0; + } + + /** + * Return whether a type contains any atomic counters. + */ + bool contains_atomic() const + { + return atomic_size() > 0; + } + + /** + * Return whether a type contains any opaque types. + */ + bool contains_opaque() const; + + /** + * Query the full type of a matrix row + * + * \return + * If the type is not a matrix, \c glsl_type::error_type is returned. + * Otherwise a type matching the rows of the matrix is returned. + */ + const glsl_type *row_type() const + { + return is_matrix() + ? get_instance(base_type, matrix_columns, 1) + : error_type; + } + + /** + * Query the full type of a matrix column + * + * \return + * If the type is not a matrix, \c glsl_type::error_type is returned. + * Otherwise a type matching the columns of the matrix is returned. + */ + const glsl_type *column_type() const + { + return is_matrix() + ? get_instance(base_type, vector_elements, 1) + : error_type; + } + + /** + * Get the type of a structure field + * + * \return + * Pointer to the type of the named field. If the type is not a structure + * or the named field does not exist, \c glsl_type::error_type is returned. + */ + const glsl_type *field_type(const char *name) const; + + /** + * Get the location of a field within a record type + */ + int field_index(const char *name) const; + + /** + * Query the number of elements in an array type + * + * \return + * The number of elements in the array for array types or -1 for non-array + * types. If the number of elements in the array has not yet been declared, + * zero is returned. + */ + int array_size() const + { + return is_array() ? length : -1; + } + + /** + * Query whether the array size for all dimensions has been declared. + */ + bool is_unsized_array() const + { + return is_array() && length == 0; + } + + /** + * Return the number of coordinate components needed for this + * sampler or image type. + * + * This is based purely on the sampler's dimensionality. For example, this + * returns 1 for sampler1D, and 3 for sampler2DArray. + * + * Note that this is often different than actual coordinate type used in + * a texturing built-in function, since those pack additional values (such + * as the shadow comparitor or projector) into the coordinate type. + */ + int coordinate_components() const; + + /** + * Compare a record type against another record type. + * + * This is useful for matching record types declared across shader stages. + */ + bool record_compare(const glsl_type *b) const; + +private: + + static mtx_t mutex; + + /** + * ralloc context for all glsl_type allocations + * + * Set on the first call to \c glsl_type::new. + */ + static void *mem_ctx; + + void init_ralloc_type_ctx(void); + + /** Constructor for vector and matrix types */ + glsl_type(GLenum gl_type, + glsl_base_type base_type, unsigned vector_elements, + unsigned matrix_columns, const char *name); + + /** Constructor for sampler or image types */ + glsl_type(GLenum gl_type, glsl_base_type base_type, + enum glsl_sampler_dim dim, bool shadow, bool array, + unsigned type, const char *name); + + /** Constructor for record types */ + glsl_type(const glsl_struct_field *fields, unsigned num_fields, + const char *name); + + /** Constructor for interface types */ + glsl_type(const glsl_struct_field *fields, unsigned num_fields, + enum glsl_interface_packing packing, const char *name); + + /** Constructor for interface types */ + glsl_type(const glsl_type *return_type, + const glsl_function_param *params, unsigned num_params); + + /** Constructor for array types */ + glsl_type(const glsl_type *array, unsigned length); + + /** Constructor for subroutine types */ + glsl_type(const char *name); + + /** Hash table containing the known array types. */ + static struct hash_table *array_types; + + /** Hash table containing the known record types. */ + static struct hash_table *record_types; + + /** Hash table containing the known interface types. */ + static struct hash_table *interface_types; + + /** Hash table containing the known subroutine types. */ + static struct hash_table *subroutine_types; + + /** Hash table containing the known function types. */ + static struct hash_table *function_types; + + static bool record_key_compare(const void *a, const void *b); + static unsigned record_key_hash(const void *key); + + /** + * \name Built-in type flyweights + */ + /*@{*/ +#undef DECL_TYPE +#define DECL_TYPE(NAME, ...) static const glsl_type _##NAME##_type; +#undef STRUCT_TYPE +#define STRUCT_TYPE(NAME) static const glsl_type _struct_##NAME##_type; +#include "builtin_type_macros.h" + /*@}*/ + + /** + * \name Friend functions. + * + * These functions are friends because they must have C linkage and the + * need to call various private methods or access various private static + * data. + */ + /*@{*/ + friend void _mesa_glsl_initialize_types(struct _mesa_glsl_parse_state *); + friend void _mesa_glsl_release_types(void); + /*@}*/ +}; + +#undef DECL_TYPE +#undef STRUCT_TYPE +#endif /* __cplusplus */ + +struct glsl_struct_field { + const struct glsl_type *type; + const char *name; + + /** + * For interface blocks, gl_varying_slot corresponding to the input/output + * if this is a built-in input/output (i.e. a member of the built-in + * gl_PerVertex interface block); -1 otherwise. + * + * Ignored for structs. + */ + int location; + + /** + * For interface blocks, the interpolation mode (as in + * ir_variable::interpolation). 0 otherwise. + */ + unsigned interpolation:2; + + /** + * For interface blocks, 1 if this variable uses centroid interpolation (as + * in ir_variable::centroid). 0 otherwise. + */ + unsigned centroid:1; + + /** + * For interface blocks, 1 if this variable uses sample interpolation (as + * in ir_variable::sample). 0 otherwise. + */ + unsigned sample:1; + + /** + * Layout of the matrix. Uses glsl_matrix_layout values. + */ + unsigned matrix_layout:2; + + /** + * For interface blocks, 1 if this variable is a per-patch input or output + * (as in ir_variable::patch). 0 otherwise. + */ + unsigned patch:1; + + /** + * For interface blocks, it has a value if this variable uses multiple vertex + * streams (as in ir_variable::stream). -1 otherwise. + */ + int stream; + + /** + * Image qualifiers, applicable to buffer variables defined in shader + * storage buffer objects (SSBOs) + */ + unsigned image_read_only:1; + unsigned image_write_only:1; + unsigned image_coherent:1; + unsigned image_volatile:1; + unsigned image_restrict:1; + +#ifdef __cplusplus + glsl_struct_field(const struct glsl_type *_type, const char *_name) + : type(_type), name(_name), location(-1), interpolation(0), centroid(0), + sample(0), matrix_layout(GLSL_MATRIX_LAYOUT_INHERITED), patch(0), + stream(-1) + { + /* empty */ + } + + glsl_struct_field() + { + /* empty */ + } +#endif +}; + +struct glsl_function_param { + const struct glsl_type *type; + + bool in; + bool out; +}; + +static inline unsigned int +glsl_align(unsigned int a, unsigned int align) +{ + return (a + align - 1) / align * align; +} + +#endif /* GLSL_TYPES_H */ diff --git a/src/glsl/nir/nir.c b/src/glsl/nir/nir.c index e12da805281..793bdafb54b 100644 --- a/src/glsl/nir/nir.c +++ b/src/glsl/nir/nir.c @@ -103,6 +103,72 @@ nir_reg_remove(nir_register *reg) exec_node_remove(®->node); } +void +nir_shader_add_variable(nir_shader *shader, nir_variable *var) +{ + switch (var->data.mode) { + case nir_var_local: + assert(!"nir_shader_add_variable cannot be used for local variables"); + break; + + case nir_var_global: + exec_list_push_tail(&shader->globals, &var->node); + break; + + case nir_var_shader_in: + exec_list_push_tail(&shader->inputs, &var->node); + break; + + case nir_var_shader_out: + exec_list_push_tail(&shader->outputs, &var->node); + break; + + case nir_var_uniform: + case nir_var_shader_storage: + exec_list_push_tail(&shader->uniforms, &var->node); + break; + + case nir_var_system_value: + exec_list_push_tail(&shader->system_values, &var->node); + break; + } +} + +nir_variable * +nir_variable_create(nir_shader *shader, nir_variable_mode mode, + const struct glsl_type *type, const char *name) +{ + nir_variable *var = rzalloc(shader, nir_variable); + var->name = ralloc_strdup(var, name); + var->type = type; + var->data.mode = mode; + + if ((mode == nir_var_shader_in && shader->stage != MESA_SHADER_VERTEX) || + (mode == nir_var_shader_out && shader->stage != MESA_SHADER_FRAGMENT)) + var->data.interpolation = INTERP_QUALIFIER_SMOOTH; + + if (mode == nir_var_shader_in || mode == nir_var_uniform) + var->data.read_only = true; + + nir_shader_add_variable(shader, var); + + return var; +} + +nir_variable * +nir_local_variable_create(nir_function_impl *impl, + const struct glsl_type *type, const char *name) +{ + nir_variable *var = rzalloc(impl->overload->function->shader, nir_variable); + var->name = ralloc_strdup(var, name); + var->type = type; + var->data.mode = nir_var_local; + + nir_function_impl_add_variable(impl, var); + + return var; +} + nir_function * nir_function_create(nir_shader *shader, const char *name) { @@ -1080,31 +1146,33 @@ nir_src_as_const_value(nir_src src) return &load->value; } +/** + * Returns true if the source is known to be dynamically uniform. Otherwise it + * returns false which means it may or may not be dynamically uniform but it + * can't be determined. + */ bool -nir_srcs_equal(nir_src src1, nir_src src2) +nir_src_is_dynamically_uniform(nir_src src) { - if (src1.is_ssa) { - if (src2.is_ssa) { - return src1.ssa == src2.ssa; - } else { - return false; - } - } else { - if (src2.is_ssa) { - return false; - } else { - if ((src1.reg.indirect == NULL) != (src2.reg.indirect == NULL)) - return false; + if (!src.is_ssa) + return false; - if (src1.reg.indirect) { - if (!nir_srcs_equal(*src1.reg.indirect, *src2.reg.indirect)) - return false; - } + /* Constants are trivially dynamically uniform */ + if (src.ssa->parent_instr->type == nir_instr_type_load_const) + return true; - return src1.reg.reg == src2.reg.reg && - src1.reg.base_offset == src2.reg.base_offset; - } + /* As are uniform variables */ + if (src.ssa->parent_instr->type == nir_instr_type_intrinsic) { + nir_intrinsic_instr *intr = nir_instr_as_intrinsic(src.ssa->parent_instr); + + if (intr->intrinsic == nir_intrinsic_load_uniform) + return true; } + + /* XXX: this could have many more tests, such as when a sampler function is + * called with dynamically uniform arguments. + */ + return false; } static void diff --git a/src/glsl/nir/nir.h b/src/glsl/nir/nir.h index f7b9483d74a..825c34805c4 100644 --- a/src/glsl/nir/nir.h +++ b/src/glsl/nir/nir.h @@ -35,7 +35,7 @@ #include "util/set.h" #include "util/bitset.h" #include "nir_types.h" -#include "glsl/shader_enums.h" +#include "shader_enums.h" #include <stdio.h> #include "nir_opcodes.h" @@ -738,7 +738,7 @@ nir_alu_instr_channel_used(nir_alu_instr *instr, unsigned src, unsigned channel) * used for a source */ static inline unsigned -nir_ssa_alu_instr_src_components(nir_alu_instr *instr, unsigned src) +nir_ssa_alu_instr_src_components(const nir_alu_instr *instr, unsigned src) { assert(instr->dest.dest.is_ssa); @@ -1486,6 +1486,9 @@ typedef struct nir_shader_compiler_options { typedef struct nir_shader_info { const char *name; + /* Descriptive name provided by the client; may be NULL */ + const char *label; + /* Number of textures used by this shader */ unsigned num_textures; /* Number of uniform buffers used by this shader */ @@ -1516,13 +1519,32 @@ typedef struct nir_shader_info { /** Was this shader linked with any transform feedback varyings? */ bool has_transform_feedback_varyings; - struct { - /** The maximum number of vertices the geometry shader might write. */ - unsigned vertices_out; + union { + struct { + /** The maximum number of vertices the geometry shader might write. */ + unsigned vertices_out; + + /** 1 .. MAX_GEOMETRY_SHADER_INVOCATIONS */ + unsigned invocations; + } gs; + + struct { + bool uses_discard; + + /** + * Whether early fragment tests are enabled as defined by + * ARB_shader_image_load_store. + */ + bool early_fragment_tests; + + /** gl_FragDepth layout for ARB_conservative_depth. */ + enum gl_frag_depth_layout depth_layout; + } fs; - /** 1 .. MAX_GEOMETRY_SHADER_INVOCATIONS */ - unsigned invocations; - } gs; + struct { + unsigned local_size[3]; + } cs; + }; } nir_shader_info; typedef struct nir_shader { @@ -1585,6 +1607,26 @@ nir_register *nir_local_reg_create(nir_function_impl *impl); void nir_reg_remove(nir_register *reg); +/** Adds a variable to the appropreate list in nir_shader */ +void nir_shader_add_variable(nir_shader *shader, nir_variable *var); + +static inline void +nir_function_impl_add_variable(nir_function_impl *impl, nir_variable *var) +{ + assert(var->data.mode == nir_var_local); + exec_list_push_tail(&impl->locals, &var->node); +} + +/** creates a variable, sets a few defaults, and adds it to the list */ +nir_variable *nir_variable_create(nir_shader *shader, + nir_variable_mode mode, + const struct glsl_type *type, + const char *name); +/** creates a local variable and adds it to the list */ +nir_variable *nir_local_variable_create(nir_function_impl *impl, + const struct glsl_type *type, + const char *name); + /** creates a function and adds it to the shader's list of functions */ nir_function *nir_function_create(nir_shader *shader, const char *name); @@ -1821,6 +1863,7 @@ bool nir_foreach_dest(nir_instr *instr, nir_foreach_dest_cb cb, void *state); bool nir_foreach_src(nir_instr *instr, nir_foreach_src_cb cb, void *state); nir_const_value *nir_src_as_const_value(nir_src src); +bool nir_src_is_dynamically_uniform(nir_src src); bool nir_srcs_equal(nir_src src1, nir_src src2); void nir_instr_rewrite_src(nir_instr *instr, nir_src *src, nir_src new_src); void nir_instr_move_src(nir_instr *dest_instr, nir_src *dest, nir_src *src); diff --git a/src/glsl/nir/nir_constant_expressions.py b/src/glsl/nir/nir_constant_expressions.py index 8fd9b1039a7..2ba8554645d 100644 --- a/src/glsl/nir/nir_constant_expressions.py +++ b/src/glsl/nir/nir_constant_expressions.py @@ -29,6 +29,7 @@ template = """\ #include <math.h> #include "main/core.h" #include "util/rounding.h" /* for _mesa_roundeven */ +#include "util/half_float.h" #include "nir_constant_expressions.h" #if defined(__SUNPRO_CC) diff --git a/src/glsl/nir/nir_instr_set.c b/src/glsl/nir/nir_instr_set.c new file mode 100644 index 00000000000..d3f939fe805 --- /dev/null +++ b/src/glsl/nir/nir_instr_set.c @@ -0,0 +1,519 @@ +/* + * Copyright © 2014 Connor Abbott + * + * 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 "nir_instr_set.h" +#include "nir_vla.h" + +#define HASH(hash, data) _mesa_fnv32_1a_accumulate((hash), (data)) + +static uint32_t +hash_src(uint32_t hash, const nir_src *src) +{ + assert(src->is_ssa); + hash = HASH(hash, src->ssa); + return hash; +} + +static uint32_t +hash_alu_src(uint32_t hash, const nir_alu_src *src, unsigned num_components) +{ + hash = HASH(hash, src->abs); + hash = HASH(hash, src->negate); + + for (unsigned i = 0; i < num_components; i++) + hash = HASH(hash, src->swizzle[i]); + + hash = hash_src(hash, &src->src); + return hash; +} + +static uint32_t +hash_alu(uint32_t hash, const nir_alu_instr *instr) +{ + hash = HASH(hash, instr->op); + hash = HASH(hash, instr->dest.dest.ssa.num_components); + + if (nir_op_infos[instr->op].algebraic_properties & NIR_OP_IS_COMMUTATIVE) { + assert(nir_op_infos[instr->op].num_inputs == 2); + uint32_t hash0 = hash_alu_src(hash, &instr->src[0], + nir_ssa_alu_instr_src_components(instr, 0)); + uint32_t hash1 = hash_alu_src(hash, &instr->src[1], + nir_ssa_alu_instr_src_components(instr, 1)); + /* For commutative operations, we need some commutative way of + * combining the hashes. One option would be to XOR them but that + * means that anything with two identical sources will hash to 0 and + * that's common enough we probably don't want the guaranteed + * collision. Either addition or multiplication will also work. + */ + hash = hash0 * hash1; + } else { + for (unsigned i = 0; i < nir_op_infos[instr->op].num_inputs; i++) { + hash = hash_alu_src(hash, &instr->src[i], + nir_ssa_alu_instr_src_components(instr, i)); + } + } + + return hash; +} + +static uint32_t +hash_load_const(uint32_t hash, const nir_load_const_instr *instr) +{ + hash = HASH(hash, instr->def.num_components); + + hash = _mesa_fnv32_1a_accumulate_block(hash, instr->value.f, + instr->def.num_components + * sizeof(instr->value.f[0])); + + return hash; +} + +static int +cmp_phi_src(const void *data1, const void *data2) +{ + nir_phi_src *src1 = *(nir_phi_src **)data1; + nir_phi_src *src2 = *(nir_phi_src **)data2; + return src1->pred - src2->pred; +} + +static uint32_t +hash_phi(uint32_t hash, const nir_phi_instr *instr) +{ + hash = HASH(hash, instr->instr.block); + + /* sort sources by predecessor, since the order shouldn't matter */ + unsigned num_preds = instr->instr.block->predecessors->entries; + NIR_VLA(nir_phi_src *, srcs, num_preds); + unsigned i = 0; + nir_foreach_phi_src(instr, src) { + srcs[i++] = src; + } + + qsort(srcs, num_preds, sizeof(nir_phi_src *), cmp_phi_src); + + for (i = 0; i < num_preds; i++) { + hash = hash_src(hash, &srcs[i]->src); + hash = HASH(hash, srcs[i]->pred); + } + + return hash; +} + +static uint32_t +hash_intrinsic(uint32_t hash, const nir_intrinsic_instr *instr) +{ + const nir_intrinsic_info *info = &nir_intrinsic_infos[instr->intrinsic]; + hash = HASH(hash, instr->intrinsic); + + if (info->has_dest) + hash = HASH(hash, instr->dest.ssa.num_components); + + assert(info->num_variables == 0); + + hash = _mesa_fnv32_1a_accumulate_block(hash, instr->const_index, + info->num_indices + * sizeof(instr->const_index[0])); + return hash; +} + +static uint32_t +hash_tex(uint32_t hash, const nir_tex_instr *instr) +{ + hash = HASH(hash, instr->op); + hash = HASH(hash, instr->num_srcs); + + for (unsigned i = 0; i < instr->num_srcs; i++) { + hash = HASH(hash, instr->src[i].src_type); + hash = hash_src(hash, &instr->src[i].src); + } + + hash = HASH(hash, instr->coord_components); + hash = HASH(hash, instr->sampler_dim); + hash = HASH(hash, instr->is_array); + hash = HASH(hash, instr->is_shadow); + hash = HASH(hash, instr->is_new_style_shadow); + hash = HASH(hash, instr->const_offset); + unsigned component = instr->component; + hash = HASH(hash, component); + hash = HASH(hash, instr->sampler_index); + hash = HASH(hash, instr->sampler_array_size); + + assert(!instr->sampler); + + return hash; +} + +/* Computes a hash of an instruction for use in a hash table. Note that this + * will only work for instructions where instr_can_rewrite() returns true, and + * it should return identical hashes for two instructions that are the same + * according nir_instrs_equal(). + */ + +static uint32_t +hash_instr(const void *data) +{ + const nir_instr *instr = data; + uint32_t hash = _mesa_fnv32_1a_offset_bias; + + switch (instr->type) { + case nir_instr_type_alu: + hash = hash_alu(hash, nir_instr_as_alu(instr)); + break; + case nir_instr_type_load_const: + hash = hash_load_const(hash, nir_instr_as_load_const(instr)); + break; + case nir_instr_type_phi: + hash = hash_phi(hash, nir_instr_as_phi(instr)); + break; + case nir_instr_type_intrinsic: + hash = hash_intrinsic(hash, nir_instr_as_intrinsic(instr)); + break; + case nir_instr_type_tex: + hash = hash_tex(hash, nir_instr_as_tex(instr)); + break; + default: + unreachable("Invalid instruction type"); + } + + return hash; +} + +bool +nir_srcs_equal(nir_src src1, nir_src src2) +{ + if (src1.is_ssa) { + if (src2.is_ssa) { + return src1.ssa == src2.ssa; + } else { + return false; + } + } else { + if (src2.is_ssa) { + return false; + } else { + if ((src1.reg.indirect == NULL) != (src2.reg.indirect == NULL)) + return false; + + if (src1.reg.indirect) { + if (!nir_srcs_equal(*src1.reg.indirect, *src2.reg.indirect)) + return false; + } + + return src1.reg.reg == src2.reg.reg && + src1.reg.base_offset == src2.reg.base_offset; + } + } +} + +static bool +nir_alu_srcs_equal(const nir_alu_instr *alu1, const nir_alu_instr *alu2, + unsigned src1, unsigned src2) +{ + if (alu1->src[src1].abs != alu2->src[src2].abs || + alu1->src[src1].negate != alu2->src[src2].negate) + return false; + + for (unsigned i = 0; i < nir_ssa_alu_instr_src_components(alu1, src1); i++) { + if (alu1->src[src1].swizzle[i] != alu2->src[src2].swizzle[i]) + return false; + } + + return nir_srcs_equal(alu1->src[src1].src, alu2->src[src2].src); +} + +/* Returns "true" if two instructions are equal. Note that this will only + * work for the subset of instructions defined by instr_can_rewrite(). Also, + * it should only return "true" for instructions that hash_instr() will return + * the same hash for (ignoring collisions, of course). + */ + +static bool +nir_instrs_equal(const nir_instr *instr1, const nir_instr *instr2) +{ + if (instr1->type != instr2->type) + return false; + + switch (instr1->type) { + case nir_instr_type_alu: { + nir_alu_instr *alu1 = nir_instr_as_alu(instr1); + nir_alu_instr *alu2 = nir_instr_as_alu(instr2); + + if (alu1->op != alu2->op) + return false; + + /* TODO: We can probably acutally do something more inteligent such + * as allowing different numbers and taking a maximum or something + * here */ + if (alu1->dest.dest.ssa.num_components != alu2->dest.dest.ssa.num_components) + return false; + + if (nir_op_infos[alu1->op].algebraic_properties & NIR_OP_IS_COMMUTATIVE) { + assert(nir_op_infos[alu1->op].num_inputs == 2); + return (nir_alu_srcs_equal(alu1, alu2, 0, 0) && + nir_alu_srcs_equal(alu1, alu2, 1, 1)) || + (nir_alu_srcs_equal(alu1, alu2, 0, 1) && + nir_alu_srcs_equal(alu1, alu2, 1, 0)); + } else { + for (unsigned i = 0; i < nir_op_infos[alu1->op].num_inputs; i++) { + if (!nir_alu_srcs_equal(alu1, alu2, i, i)) + return false; + } + } + return true; + } + case nir_instr_type_tex: { + nir_tex_instr *tex1 = nir_instr_as_tex(instr1); + nir_tex_instr *tex2 = nir_instr_as_tex(instr2); + + if (tex1->op != tex2->op) + return false; + + if (tex1->num_srcs != tex2->num_srcs) + return false; + for (unsigned i = 0; i < tex1->num_srcs; i++) { + if (tex1->src[i].src_type != tex2->src[i].src_type || + !nir_srcs_equal(tex1->src[i].src, tex2->src[i].src)) { + return false; + } + } + + if (tex1->coord_components != tex2->coord_components || + tex1->sampler_dim != tex2->sampler_dim || + tex1->is_array != tex2->is_array || + tex1->is_shadow != tex2->is_shadow || + tex1->is_new_style_shadow != tex2->is_new_style_shadow || + memcmp(tex1->const_offset, tex2->const_offset, + sizeof(tex1->const_offset)) != 0 || + tex1->component != tex2->component || + tex1->sampler_index != tex2->sampler_index || + tex1->sampler_array_size != tex2->sampler_array_size) { + return false; + } + + /* Don't support un-lowered sampler derefs currently. */ + assert(!tex1->sampler && !tex2->sampler); + + return true; + } + case nir_instr_type_load_const: { + nir_load_const_instr *load1 = nir_instr_as_load_const(instr1); + nir_load_const_instr *load2 = nir_instr_as_load_const(instr2); + + if (load1->def.num_components != load2->def.num_components) + return false; + + return memcmp(load1->value.f, load2->value.f, + load1->def.num_components * sizeof(*load2->value.f)) == 0; + } + case nir_instr_type_phi: { + nir_phi_instr *phi1 = nir_instr_as_phi(instr1); + nir_phi_instr *phi2 = nir_instr_as_phi(instr2); + + if (phi1->instr.block != phi2->instr.block) + return false; + + nir_foreach_phi_src(phi1, src1) { + nir_foreach_phi_src(phi2, src2) { + if (src1->pred == src2->pred) { + if (!nir_srcs_equal(src1->src, src2->src)) + return false; + + break; + } + } + } + + return true; + } + case nir_instr_type_intrinsic: { + nir_intrinsic_instr *intrinsic1 = nir_instr_as_intrinsic(instr1); + nir_intrinsic_instr *intrinsic2 = nir_instr_as_intrinsic(instr2); + const nir_intrinsic_info *info = + &nir_intrinsic_infos[intrinsic1->intrinsic]; + + if (intrinsic1->intrinsic != intrinsic2->intrinsic || + intrinsic1->num_components != intrinsic2->num_components) + return false; + + if (info->has_dest && intrinsic1->dest.ssa.num_components != + intrinsic2->dest.ssa.num_components) + return false; + + for (unsigned i = 0; i < info->num_srcs; i++) { + if (!nir_srcs_equal(intrinsic1->src[i], intrinsic2->src[i])) + return false; + } + + assert(info->num_variables == 0); + + for (unsigned i = 0; i < info->num_indices; i++) { + if (intrinsic1->const_index[i] != intrinsic2->const_index[i]) + return false; + } + + return true; + } + case nir_instr_type_call: + case nir_instr_type_jump: + case nir_instr_type_ssa_undef: + case nir_instr_type_parallel_copy: + default: + unreachable("Invalid instruction type"); + } + + return false; +} + +static bool +src_is_ssa(nir_src *src, void *data) +{ + (void) data; + return src->is_ssa; +} + +static bool +dest_is_ssa(nir_dest *dest, void *data) +{ + (void) data; + return dest->is_ssa; +} + +/* This function determines if uses of an instruction can safely be rewritten + * to use another identical instruction instead. Note that this function must + * be kept in sync with hash_instr() and nir_instrs_equal() -- only + * instructions that pass this test will be handed on to those functions, and + * conversely they must handle everything that this function returns true for. + */ + +static bool +instr_can_rewrite(nir_instr *instr) +{ + /* We only handle SSA. */ + if (!nir_foreach_dest(instr, dest_is_ssa, NULL) || + !nir_foreach_src(instr, src_is_ssa, NULL)) + return false; + + switch (instr->type) { + case nir_instr_type_alu: + case nir_instr_type_load_const: + case nir_instr_type_phi: + return true; + case nir_instr_type_tex: { + nir_tex_instr *tex = nir_instr_as_tex(instr); + + /* Don't support un-lowered sampler derefs currently. */ + if (tex->sampler) + return false; + + return true; + } + case nir_instr_type_intrinsic: { + const nir_intrinsic_info *info = + &nir_intrinsic_infos[nir_instr_as_intrinsic(instr)->intrinsic]; + return (info->flags & NIR_INTRINSIC_CAN_ELIMINATE) && + (info->flags & NIR_INTRINSIC_CAN_REORDER) && + info->num_variables == 0; /* not implemented yet */ + } + case nir_instr_type_call: + case nir_instr_type_jump: + case nir_instr_type_ssa_undef: + return false; + case nir_instr_type_parallel_copy: + default: + unreachable("Invalid instruction type"); + } + + return false; +} + +static nir_ssa_def * +nir_instr_get_dest_ssa_def(nir_instr *instr) +{ + switch (instr->type) { + case nir_instr_type_alu: + assert(nir_instr_as_alu(instr)->dest.dest.is_ssa); + return &nir_instr_as_alu(instr)->dest.dest.ssa; + case nir_instr_type_load_const: + return &nir_instr_as_load_const(instr)->def; + case nir_instr_type_phi: + assert(nir_instr_as_phi(instr)->dest.is_ssa); + return &nir_instr_as_phi(instr)->dest.ssa; + case nir_instr_type_intrinsic: + assert(nir_instr_as_intrinsic(instr)->dest.is_ssa); + return &nir_instr_as_intrinsic(instr)->dest.ssa; + case nir_instr_type_tex: + assert(nir_instr_as_tex(instr)->dest.is_ssa); + return &nir_instr_as_tex(instr)->dest.ssa; + default: + unreachable("We never ask for any of these"); + } +} + +static bool +cmp_func(const void *data1, const void *data2) +{ + return nir_instrs_equal(data1, data2); +} + +struct set * +nir_instr_set_create(void *mem_ctx) +{ + return _mesa_set_create(mem_ctx, hash_instr, cmp_func); +} + +void +nir_instr_set_destroy(struct set *instr_set) +{ + _mesa_set_destroy(instr_set, NULL); +} + +bool +nir_instr_set_add_or_rewrite(struct set *instr_set, nir_instr *instr) +{ + if (!instr_can_rewrite(instr)) + return false; + + struct set_entry *entry = _mesa_set_search(instr_set, instr); + if (entry) { + nir_ssa_def *def = nir_instr_get_dest_ssa_def(instr); + nir_ssa_def *new_def = + nir_instr_get_dest_ssa_def((nir_instr *) entry->key); + nir_ssa_def_rewrite_uses(def, nir_src_for_ssa(new_def)); + return true; + } + + _mesa_set_add(instr_set, instr); + return false; +} + +void +nir_instr_set_remove(struct set *instr_set, nir_instr *instr) +{ + if (!instr_can_rewrite(instr)) + return; + + struct set_entry *entry = _mesa_set_search(instr_set, instr); + if (entry) + _mesa_set_remove(instr_set, entry); +} + diff --git a/src/glsl/nir/nir_instr_set.h b/src/glsl/nir/nir_instr_set.h new file mode 100644 index 00000000000..939e8ddbf58 --- /dev/null +++ b/src/glsl/nir/nir_instr_set.h @@ -0,0 +1,62 @@ +/* + * Copyright © 2014 Connor Abbott + * + * 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 "nir.h" + +/** + * This file defines functions for creating, destroying, and manipulating an + * "instruction set," which is an abstraction for finding duplicate + * instructions using a hash set. Note that the question of whether an + * instruction is actually a duplicate (e.g. whether it has any side effects) + * is handled transparently. The user can pass any instruction to + * nir_instr_set_add_or_rewrite() and nir_instr_set_remove(), and if the + * instruction isn't safe to rewrite or isn't supported, it's silently + * removed. + */ + +/*@{*/ + +/** Creates an instruction set, using a given ralloc mem_ctx */ +struct set *nir_instr_set_create(void *mem_ctx); + +/** Destroys an instruction set. */ +void nir_instr_set_destroy(struct set *instr_set); + +/** + * Adds an instruction to an instruction set if it doesn't exist, or if it + * does already exist, rewrites all uses of it to point to the other + * already-inserted instruction. Returns 'true' if the uses of the instruction + * were rewritten. + */ +bool nir_instr_set_add_or_rewrite(struct set *instr_set, nir_instr *instr); + +/** + * Removes an instruction from an instruction set, so that other instructions + * won't be merged with it. + */ +void nir_instr_set_remove(struct set *instr_set, nir_instr *instr); + +/*@}*/ + diff --git a/src/glsl/nir/nir_intrinsics.h b/src/glsl/nir/nir_intrinsics.h index b5a0d715aa3..68a18b9c11a 100644 --- a/src/glsl/nir/nir_intrinsics.h +++ b/src/glsl/nir/nir_intrinsics.h @@ -174,8 +174,10 @@ INTRINSIC(image_samples, 0, ARR(), true, 1, 1, 0, * 3: For CompSwap only: the second data parameter. */ INTRINSIC(ssbo_atomic_add, 3, ARR(1, 1, 1), true, 1, 0, 0, 0) -INTRINSIC(ssbo_atomic_min, 3, ARR(1, 1, 1), true, 1, 0, 0, 0) -INTRINSIC(ssbo_atomic_max, 3, ARR(1, 1, 1), true, 1, 0, 0, 0) +INTRINSIC(ssbo_atomic_imin, 3, ARR(1, 1, 1), true, 1, 0, 0, 0) +INTRINSIC(ssbo_atomic_umin, 3, ARR(1, 1, 1), true, 1, 0, 0, 0) +INTRINSIC(ssbo_atomic_imax, 3, ARR(1, 1, 1), true, 1, 0, 0, 0) +INTRINSIC(ssbo_atomic_umax, 3, ARR(1, 1, 1), true, 1, 0, 0, 0) INTRINSIC(ssbo_atomic_and, 3, ARR(1, 1, 1), true, 1, 0, 0, 0) INTRINSIC(ssbo_atomic_or, 3, ARR(1, 1, 1), true, 1, 0, 0, 0) INTRINSIC(ssbo_atomic_xor, 3, ARR(1, 1, 1), true, 1, 0, 0, 0) diff --git a/src/glsl/nir/nir_lower_atomics.c b/src/glsl/nir/nir_lower_atomics.c index 6f9ecc019ec..46e137652a1 100644 --- a/src/glsl/nir/nir_lower_atomics.c +++ b/src/glsl/nir/nir_lower_atomics.c @@ -72,20 +72,22 @@ lower_instr(nir_intrinsic_instr *instr, nir_function_impl *impl) nir_ssa_def *offset_def = &offset_const->def; - if (instr->variables[0]->deref.child != NULL) { - assert(instr->variables[0]->deref.child->deref_type == - nir_deref_type_array); - nir_deref_array *deref_array = - nir_deref_as_array(instr->variables[0]->deref.child); - assert(deref_array->deref.child == NULL); + nir_deref *tail = &instr->variables[0]->deref; + while (tail->child != NULL) { + assert(tail->child->deref_type == nir_deref_type_array); + nir_deref_array *deref_array = nir_deref_as_array(tail->child); + tail = tail->child; - offset_const->value.u[0] += - deref_array->base_offset * ATOMIC_COUNTER_SIZE; + unsigned child_array_elements = tail->child != NULL ? + glsl_get_aoa_size(tail->type) : 1; + + offset_const->value.u[0] += deref_array->base_offset * + child_array_elements * ATOMIC_COUNTER_SIZE; if (deref_array->deref_array_type == nir_deref_array_type_indirect) { nir_load_const_instr *atomic_counter_size = nir_load_const_instr_create(mem_ctx, 1); - atomic_counter_size->value.u[0] = ATOMIC_COUNTER_SIZE; + atomic_counter_size->value.u[0] = child_array_elements * ATOMIC_COUNTER_SIZE; nir_instr_insert_before(&instr->instr, &atomic_counter_size->instr); nir_alu_instr *mul = nir_alu_instr_create(mem_ctx, nir_op_imul); @@ -102,7 +104,7 @@ lower_instr(nir_intrinsic_instr *instr, nir_function_impl *impl) add->src[0].src.is_ssa = true; add->src[0].src.ssa = &mul->dest.dest.ssa; add->src[1].src.is_ssa = true; - add->src[1].src.ssa = &offset_const->def; + add->src[1].src.ssa = offset_def; nir_instr_insert_before(&instr->instr, &add->instr); offset_def = &add->dest.dest.ssa; diff --git a/src/glsl/nir/nir_opt_cse.c b/src/glsl/nir/nir_opt_cse.c index 64c94afd480..93a6635337a 100644 --- a/src/glsl/nir/nir_opt_cse.c +++ b/src/glsl/nir/nir_opt_cse.c @@ -22,306 +22,60 @@ * * Authors: * Jason Ekstrand ([email protected]) + * Connor Abbott ([email protected]) * */ -#include "nir.h" +#include "nir_instr_set.h" /* * Implements common subexpression elimination */ -struct cse_state { - void *mem_ctx; - bool progress; -}; - -static bool -nir_alu_srcs_equal(nir_alu_instr *alu1, nir_alu_instr *alu2, unsigned src1, - unsigned src2) -{ - if (alu1->src[src1].abs != alu2->src[src2].abs || - alu1->src[src1].negate != alu2->src[src2].negate) - return false; - - for (unsigned i = 0; i < nir_ssa_alu_instr_src_components(alu1, src1); i++) { - if (alu1->src[src1].swizzle[i] != alu2->src[src2].swizzle[i]) - return false; - } - - return nir_srcs_equal(alu1->src[src1].src, alu2->src[src2].src); -} - -static bool -nir_instrs_equal(nir_instr *instr1, nir_instr *instr2) -{ - if (instr1->type != instr2->type) - return false; - - switch (instr1->type) { - case nir_instr_type_alu: { - nir_alu_instr *alu1 = nir_instr_as_alu(instr1); - nir_alu_instr *alu2 = nir_instr_as_alu(instr2); - - if (alu1->op != alu2->op) - return false; - - /* TODO: We can probably acutally do something more inteligent such - * as allowing different numbers and taking a maximum or something - * here */ - if (alu1->dest.dest.ssa.num_components != alu2->dest.dest.ssa.num_components) - return false; - - if (nir_op_infos[alu1->op].algebraic_properties & NIR_OP_IS_COMMUTATIVE) { - assert(nir_op_infos[alu1->op].num_inputs == 2); - return (nir_alu_srcs_equal(alu1, alu2, 0, 0) && - nir_alu_srcs_equal(alu1, alu2, 1, 1)) || - (nir_alu_srcs_equal(alu1, alu2, 0, 1) && - nir_alu_srcs_equal(alu1, alu2, 1, 0)); - } else { - for (unsigned i = 0; i < nir_op_infos[alu1->op].num_inputs; i++) { - if (!nir_alu_srcs_equal(alu1, alu2, i, i)) - return false; - } - } - return true; - } - case nir_instr_type_tex: { - nir_tex_instr *tex1 = nir_instr_as_tex(instr1); - nir_tex_instr *tex2 = nir_instr_as_tex(instr2); - - if (tex1->op != tex2->op) - return false; - - if (tex1->num_srcs != tex2->num_srcs) - return false; - for (unsigned i = 0; i < tex1->num_srcs; i++) { - if (tex1->src[i].src_type != tex2->src[i].src_type || - !nir_srcs_equal(tex1->src[i].src, tex2->src[i].src)) { - return false; - } - } - - if (tex1->coord_components != tex2->coord_components || - tex1->sampler_dim != tex2->sampler_dim || - tex1->is_array != tex2->is_array || - tex1->is_shadow != tex2->is_shadow || - tex1->is_new_style_shadow != tex2->is_new_style_shadow || - memcmp(tex1->const_offset, tex2->const_offset, - sizeof(tex1->const_offset)) != 0 || - tex1->component != tex2->component || - tex1->sampler_index != tex2->sampler_index || - tex1->sampler_array_size != tex2->sampler_array_size) { - return false; - } - - /* Don't support un-lowered sampler derefs currently. */ - if (tex1->sampler || tex2->sampler) - return false; - - return true; - } - case nir_instr_type_load_const: { - nir_load_const_instr *load1 = nir_instr_as_load_const(instr1); - nir_load_const_instr *load2 = nir_instr_as_load_const(instr2); - - if (load1->def.num_components != load2->def.num_components) - return false; - - return memcmp(load1->value.f, load2->value.f, - load1->def.num_components * sizeof(*load2->value.f)) == 0; - } - case nir_instr_type_phi: { - nir_phi_instr *phi1 = nir_instr_as_phi(instr1); - nir_phi_instr *phi2 = nir_instr_as_phi(instr2); - - if (phi1->instr.block != phi2->instr.block) - return false; - - nir_foreach_phi_src(phi1, src1) { - nir_foreach_phi_src(phi2, src2) { - if (src1->pred == src2->pred) { - if (!nir_srcs_equal(src1->src, src2->src)) - return false; - - break; - } - } - } - - return true; - } - case nir_instr_type_intrinsic: { - nir_intrinsic_instr *intrinsic1 = nir_instr_as_intrinsic(instr1); - nir_intrinsic_instr *intrinsic2 = nir_instr_as_intrinsic(instr2); - const nir_intrinsic_info *info = - &nir_intrinsic_infos[intrinsic1->intrinsic]; - - if (intrinsic1->intrinsic != intrinsic2->intrinsic || - intrinsic1->num_components != intrinsic2->num_components) - return false; - - if (info->has_dest && intrinsic1->dest.ssa.num_components != - intrinsic2->dest.ssa.num_components) - return false; - - for (unsigned i = 0; i < info->num_srcs; i++) { - if (!nir_srcs_equal(intrinsic1->src[i], intrinsic2->src[i])) - return false; - } - - assert(info->num_variables == 0); - - for (unsigned i = 0; i < info->num_indices; i++) { - if (intrinsic1->const_index[i] != intrinsic2->const_index[i]) - return false; - } - - return true; - } - case nir_instr_type_call: - case nir_instr_type_jump: - case nir_instr_type_ssa_undef: - case nir_instr_type_parallel_copy: - default: - unreachable("Invalid instruction type"); - } - - return false; -} - -static bool -src_is_ssa(nir_src *src, void *data) -{ - (void) data; - return src->is_ssa; -} - -static bool -dest_is_ssa(nir_dest *dest, void *data) -{ - (void) data; - return dest->is_ssa; -} +/* + * Visits and CSE's the given block and all its descendants in the dominance + * tree recursively. Note that the instr_set is guaranteed to only ever + * contain instructions that dominate the current block. + */ static bool -nir_instr_can_cse(nir_instr *instr) -{ - /* We only handle SSA. */ - if (!nir_foreach_dest(instr, dest_is_ssa, NULL) || - !nir_foreach_src(instr, src_is_ssa, NULL)) - return false; - - switch (instr->type) { - case nir_instr_type_alu: - case nir_instr_type_tex: - case nir_instr_type_load_const: - case nir_instr_type_phi: - return true; - case nir_instr_type_intrinsic: { - const nir_intrinsic_info *info = - &nir_intrinsic_infos[nir_instr_as_intrinsic(instr)->intrinsic]; - return (info->flags & NIR_INTRINSIC_CAN_ELIMINATE) && - (info->flags & NIR_INTRINSIC_CAN_REORDER) && - info->num_variables == 0; /* not implemented yet */ - } - case nir_instr_type_call: - case nir_instr_type_jump: - case nir_instr_type_ssa_undef: - return false; - case nir_instr_type_parallel_copy: - default: - unreachable("Invalid instruction type"); - } - - return false; -} - -static nir_ssa_def * -nir_instr_get_dest_ssa_def(nir_instr *instr) +cse_block(nir_block *block, struct set *instr_set) { - switch (instr->type) { - case nir_instr_type_alu: - assert(nir_instr_as_alu(instr)->dest.dest.is_ssa); - return &nir_instr_as_alu(instr)->dest.dest.ssa; - case nir_instr_type_tex: - assert(nir_instr_as_tex(instr)->dest.is_ssa); - return &nir_instr_as_tex(instr)->dest.ssa; - case nir_instr_type_load_const: - return &nir_instr_as_load_const(instr)->def; - case nir_instr_type_phi: - assert(nir_instr_as_phi(instr)->dest.is_ssa); - return &nir_instr_as_phi(instr)->dest.ssa; - case nir_instr_type_intrinsic: - assert(nir_instr_as_intrinsic(instr)->dest.is_ssa); - return &nir_instr_as_intrinsic(instr)->dest.ssa; - default: - unreachable("We never ask for any of these"); - } -} - -static void -nir_opt_cse_instr(nir_instr *instr, struct cse_state *state) -{ - if (!nir_instr_can_cse(instr)) - return; + bool progress = false; - for (struct exec_node *node = instr->node.prev; - !exec_node_is_head_sentinel(node); node = node->prev) { - nir_instr *other = exec_node_data(nir_instr, node, node); - if (nir_instrs_equal(instr, other)) { - nir_ssa_def *other_def = nir_instr_get_dest_ssa_def(other); - nir_ssa_def_rewrite_uses(nir_instr_get_dest_ssa_def(instr), - nir_src_for_ssa(other_def)); + nir_foreach_instr_safe(block, instr) { + if (nir_instr_set_add_or_rewrite(instr_set, instr)) { + progress = true; nir_instr_remove(instr); - state->progress = true; - return; } } - for (nir_block *block = instr->block->imm_dom; - block != NULL; block = block->imm_dom) { - nir_foreach_instr_reverse(block, other) { - if (nir_instrs_equal(instr, other)) { - nir_ssa_def *other_def = nir_instr_get_dest_ssa_def(other); - nir_ssa_def_rewrite_uses(nir_instr_get_dest_ssa_def(instr), - nir_src_for_ssa(other_def)); - nir_instr_remove(instr); - state->progress = true; - return; - } - } + for (unsigned i = 0; i < block->num_dom_children; i++) { + nir_block *child = block->dom_children[i]; + progress |= cse_block(child, instr_set); } -} - -static bool -nir_opt_cse_block(nir_block *block, void *void_state) -{ - struct cse_state *state = void_state; - nir_foreach_instr_safe(block, instr) - nir_opt_cse_instr(instr, state); + nir_foreach_instr(block, instr) + nir_instr_set_remove(instr_set, instr); - return true; + return progress; } static bool nir_opt_cse_impl(nir_function_impl *impl) { - struct cse_state state; - - state.mem_ctx = ralloc_parent(impl); - state.progress = false; + struct set *instr_set = nir_instr_set_create(NULL); nir_metadata_require(impl, nir_metadata_dominance); - nir_foreach_block(impl, nir_opt_cse_block, &state); + bool progress = cse_block(nir_start_block(impl), instr_set); - if (state.progress) + if (progress) nir_metadata_preserve(impl, nir_metadata_block_index | nir_metadata_dominance); - return state.progress; + nir_instr_set_destroy(instr_set); + return progress; } bool @@ -336,3 +90,4 @@ nir_opt_cse(nir_shader *shader) return progress; } + diff --git a/src/glsl/nir/nir_sweep.c b/src/glsl/nir/nir_sweep.c index b6ce43b5224..5a22f509f50 100644 --- a/src/glsl/nir/nir_sweep.c +++ b/src/glsl/nir/nir_sweep.c @@ -155,6 +155,8 @@ nir_sweep(nir_shader *nir) ralloc_adopt(rubbish, nir); ralloc_steal(nir, (char *)nir->info.name); + if (nir->info.label) + ralloc_steal(nir, (char *)nir->info.label); /* Variables and registers are not dead. Steal them back. */ steal_list(nir, nir_variable, &nir->uniforms); diff --git a/src/glsl/nir/nir_types.cpp b/src/glsl/nir/nir_types.cpp index 01f0e9b5abc..4a1250e546c 100644 --- a/src/glsl/nir/nir_types.cpp +++ b/src/glsl/nir/nir_types.cpp @@ -118,6 +118,12 @@ glsl_get_length(const struct glsl_type *type) return type->is_matrix() ? type->matrix_columns : type->length; } +unsigned +glsl_get_aoa_size(const struct glsl_type *type) +{ + return type->arrays_of_arrays_size(); +} + const char * glsl_get_struct_elem_name(const struct glsl_type *type, unsigned index) { diff --git a/src/glsl/nir/nir_types.h b/src/glsl/nir/nir_types.h index 1a0cb1fb774..a61af6cba75 100644 --- a/src/glsl/nir/nir_types.h +++ b/src/glsl/nir/nir_types.h @@ -31,7 +31,7 @@ /* C wrapper around glsl_types.h */ -#include "../glsl_types.h" +#include "glsl_types.h" #ifdef __cplusplus extern "C" { @@ -65,6 +65,8 @@ unsigned glsl_get_matrix_columns(const struct glsl_type *type); unsigned glsl_get_length(const struct glsl_type *type); +unsigned glsl_get_aoa_size(const struct glsl_type *type); + const char *glsl_get_struct_elem_name(const struct glsl_type *type, unsigned index); diff --git a/src/glsl/nir/shader_enums.c b/src/glsl/nir/shader_enums.c new file mode 100644 index 00000000000..66a25e72344 --- /dev/null +++ b/src/glsl/nir/shader_enums.c @@ -0,0 +1,213 @@ +/* + * Mesa 3-D graphics library + * + * Copyright © 2015 Red Hat + * + * 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. + * + * Authors: + * Rob Clark <[email protected]> + */ + +#include "shader_enums.h" +#include "util/macros.h" +#include "mesa/main/config.h" + +#define ENUM(x) [x] = #x +#define NAME(val) ((((val) < ARRAY_SIZE(names)) && names[(val)]) ? names[(val)] : "UNKNOWN") + +const char * gl_shader_stage_name(gl_shader_stage stage) +{ + static const char *names[] = { + ENUM(MESA_SHADER_VERTEX), + ENUM(MESA_SHADER_TESS_CTRL), + ENUM(MESA_SHADER_TESS_EVAL), + ENUM(MESA_SHADER_GEOMETRY), + ENUM(MESA_SHADER_FRAGMENT), + ENUM(MESA_SHADER_COMPUTE), + }; + STATIC_ASSERT(ARRAY_SIZE(names) == MESA_SHADER_STAGES); + return NAME(stage); +} + +const char * gl_vert_attrib_name(gl_vert_attrib attrib) +{ + static const char *names[] = { + ENUM(VERT_ATTRIB_POS), + ENUM(VERT_ATTRIB_WEIGHT), + ENUM(VERT_ATTRIB_NORMAL), + ENUM(VERT_ATTRIB_COLOR0), + ENUM(VERT_ATTRIB_COLOR1), + ENUM(VERT_ATTRIB_FOG), + ENUM(VERT_ATTRIB_COLOR_INDEX), + ENUM(VERT_ATTRIB_EDGEFLAG), + ENUM(VERT_ATTRIB_TEX0), + ENUM(VERT_ATTRIB_TEX1), + ENUM(VERT_ATTRIB_TEX2), + ENUM(VERT_ATTRIB_TEX3), + ENUM(VERT_ATTRIB_TEX4), + ENUM(VERT_ATTRIB_TEX5), + ENUM(VERT_ATTRIB_TEX6), + ENUM(VERT_ATTRIB_TEX7), + ENUM(VERT_ATTRIB_POINT_SIZE), + ENUM(VERT_ATTRIB_GENERIC0), + ENUM(VERT_ATTRIB_GENERIC1), + ENUM(VERT_ATTRIB_GENERIC2), + ENUM(VERT_ATTRIB_GENERIC3), + ENUM(VERT_ATTRIB_GENERIC4), + ENUM(VERT_ATTRIB_GENERIC5), + ENUM(VERT_ATTRIB_GENERIC6), + ENUM(VERT_ATTRIB_GENERIC7), + ENUM(VERT_ATTRIB_GENERIC8), + ENUM(VERT_ATTRIB_GENERIC9), + ENUM(VERT_ATTRIB_GENERIC10), + ENUM(VERT_ATTRIB_GENERIC11), + ENUM(VERT_ATTRIB_GENERIC12), + ENUM(VERT_ATTRIB_GENERIC13), + ENUM(VERT_ATTRIB_GENERIC14), + ENUM(VERT_ATTRIB_GENERIC15), + }; + STATIC_ASSERT(ARRAY_SIZE(names) == VERT_ATTRIB_MAX); + return NAME(attrib); +} + +const char * gl_varying_slot_name(gl_varying_slot slot) +{ + static const char *names[] = { + ENUM(VARYING_SLOT_POS), + ENUM(VARYING_SLOT_COL0), + ENUM(VARYING_SLOT_COL1), + ENUM(VARYING_SLOT_FOGC), + ENUM(VARYING_SLOT_TEX0), + ENUM(VARYING_SLOT_TEX1), + ENUM(VARYING_SLOT_TEX2), + ENUM(VARYING_SLOT_TEX3), + ENUM(VARYING_SLOT_TEX4), + ENUM(VARYING_SLOT_TEX5), + ENUM(VARYING_SLOT_TEX6), + ENUM(VARYING_SLOT_TEX7), + ENUM(VARYING_SLOT_PSIZ), + ENUM(VARYING_SLOT_BFC0), + ENUM(VARYING_SLOT_BFC1), + ENUM(VARYING_SLOT_EDGE), + ENUM(VARYING_SLOT_CLIP_VERTEX), + ENUM(VARYING_SLOT_CLIP_DIST0), + ENUM(VARYING_SLOT_CLIP_DIST1), + ENUM(VARYING_SLOT_PRIMITIVE_ID), + ENUM(VARYING_SLOT_LAYER), + ENUM(VARYING_SLOT_VIEWPORT), + ENUM(VARYING_SLOT_FACE), + ENUM(VARYING_SLOT_PNTC), + ENUM(VARYING_SLOT_TESS_LEVEL_OUTER), + ENUM(VARYING_SLOT_TESS_LEVEL_INNER), + ENUM(VARYING_SLOT_VAR0), + ENUM(VARYING_SLOT_VAR1), + ENUM(VARYING_SLOT_VAR2), + ENUM(VARYING_SLOT_VAR3), + ENUM(VARYING_SLOT_VAR4), + ENUM(VARYING_SLOT_VAR5), + ENUM(VARYING_SLOT_VAR6), + ENUM(VARYING_SLOT_VAR7), + ENUM(VARYING_SLOT_VAR8), + ENUM(VARYING_SLOT_VAR9), + ENUM(VARYING_SLOT_VAR10), + ENUM(VARYING_SLOT_VAR11), + ENUM(VARYING_SLOT_VAR12), + ENUM(VARYING_SLOT_VAR13), + ENUM(VARYING_SLOT_VAR14), + ENUM(VARYING_SLOT_VAR15), + ENUM(VARYING_SLOT_VAR16), + ENUM(VARYING_SLOT_VAR17), + ENUM(VARYING_SLOT_VAR18), + ENUM(VARYING_SLOT_VAR19), + ENUM(VARYING_SLOT_VAR20), + ENUM(VARYING_SLOT_VAR21), + ENUM(VARYING_SLOT_VAR22), + ENUM(VARYING_SLOT_VAR23), + ENUM(VARYING_SLOT_VAR24), + ENUM(VARYING_SLOT_VAR25), + ENUM(VARYING_SLOT_VAR26), + ENUM(VARYING_SLOT_VAR27), + ENUM(VARYING_SLOT_VAR28), + ENUM(VARYING_SLOT_VAR29), + ENUM(VARYING_SLOT_VAR30), + ENUM(VARYING_SLOT_VAR31), + }; + STATIC_ASSERT(ARRAY_SIZE(names) == VARYING_SLOT_MAX); + return NAME(slot); +} + +const char * gl_system_value_name(gl_system_value sysval) +{ + static const char *names[] = { + ENUM(SYSTEM_VALUE_VERTEX_ID), + ENUM(SYSTEM_VALUE_INSTANCE_ID), + ENUM(SYSTEM_VALUE_VERTEX_ID_ZERO_BASE), + ENUM(SYSTEM_VALUE_BASE_VERTEX), + ENUM(SYSTEM_VALUE_INVOCATION_ID), + ENUM(SYSTEM_VALUE_FRONT_FACE), + ENUM(SYSTEM_VALUE_SAMPLE_ID), + ENUM(SYSTEM_VALUE_SAMPLE_POS), + ENUM(SYSTEM_VALUE_SAMPLE_MASK_IN), + ENUM(SYSTEM_VALUE_TESS_COORD), + ENUM(SYSTEM_VALUE_VERTICES_IN), + ENUM(SYSTEM_VALUE_PRIMITIVE_ID), + ENUM(SYSTEM_VALUE_TESS_LEVEL_OUTER), + ENUM(SYSTEM_VALUE_TESS_LEVEL_INNER), + ENUM(SYSTEM_VALUE_LOCAL_INVOCATION_ID), + ENUM(SYSTEM_VALUE_WORK_GROUP_ID), + ENUM(SYSTEM_VALUE_NUM_WORK_GROUPS), + ENUM(SYSTEM_VALUE_VERTEX_CNT), + }; + STATIC_ASSERT(ARRAY_SIZE(names) == SYSTEM_VALUE_MAX); + return NAME(sysval); +} + +const char * glsl_interp_qualifier_name(enum glsl_interp_qualifier qual) +{ + static const char *names[] = { + ENUM(INTERP_QUALIFIER_NONE), + ENUM(INTERP_QUALIFIER_SMOOTH), + ENUM(INTERP_QUALIFIER_FLAT), + ENUM(INTERP_QUALIFIER_NOPERSPECTIVE), + }; + STATIC_ASSERT(ARRAY_SIZE(names) == INTERP_QUALIFIER_COUNT); + return NAME(qual); +} + +const char * gl_frag_result_name(gl_frag_result result) +{ + static const char *names[] = { + ENUM(FRAG_RESULT_DEPTH), + ENUM(FRAG_RESULT_STENCIL), + ENUM(FRAG_RESULT_COLOR), + ENUM(FRAG_RESULT_SAMPLE_MASK), + ENUM(FRAG_RESULT_DATA0), + ENUM(FRAG_RESULT_DATA1), + ENUM(FRAG_RESULT_DATA2), + ENUM(FRAG_RESULT_DATA3), + ENUM(FRAG_RESULT_DATA4), + ENUM(FRAG_RESULT_DATA5), + ENUM(FRAG_RESULT_DATA6), + ENUM(FRAG_RESULT_DATA7), + }; + STATIC_ASSERT(ARRAY_SIZE(names) == FRAG_RESULT_MAX); + return NAME(result); +} diff --git a/src/glsl/nir/shader_enums.h b/src/glsl/nir/shader_enums.h new file mode 100644 index 00000000000..d1cf7ca04cc --- /dev/null +++ b/src/glsl/nir/shader_enums.h @@ -0,0 +1,500 @@ +/* + * Mesa 3-D graphics library + * + * 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 + * 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 SHADER_ENUMS_H +#define SHADER_ENUMS_H + +/** + * Shader stages. Note that these will become 5 with tessellation. + * + * The order must match how shaders are ordered in the pipeline. + * The GLSL linker assumes that if i<j, then the j-th shader is + * executed later than the i-th shader. + */ +typedef enum +{ + MESA_SHADER_VERTEX = 0, + MESA_SHADER_TESS_CTRL = 1, + MESA_SHADER_TESS_EVAL = 2, + MESA_SHADER_GEOMETRY = 3, + MESA_SHADER_FRAGMENT = 4, + MESA_SHADER_COMPUTE = 5, +} gl_shader_stage; + +const char * gl_shader_stage_name(gl_shader_stage stage); + +#define MESA_SHADER_STAGES (MESA_SHADER_COMPUTE + 1) + + +/** + * Indexes for vertex program attributes. + * GL_NV_vertex_program aliases generic attributes over the conventional + * attributes. In GL_ARB_vertex_program shader the aliasing is optional. + * In GL_ARB_vertex_shader / OpenGL 2.0 the aliasing is disallowed (the + * generic attributes are distinct/separate). + */ +typedef enum +{ + VERT_ATTRIB_POS = 0, + VERT_ATTRIB_WEIGHT = 1, + VERT_ATTRIB_NORMAL = 2, + VERT_ATTRIB_COLOR0 = 3, + VERT_ATTRIB_COLOR1 = 4, + VERT_ATTRIB_FOG = 5, + VERT_ATTRIB_COLOR_INDEX = 6, + VERT_ATTRIB_EDGEFLAG = 7, + VERT_ATTRIB_TEX0 = 8, + VERT_ATTRIB_TEX1 = 9, + VERT_ATTRIB_TEX2 = 10, + VERT_ATTRIB_TEX3 = 11, + VERT_ATTRIB_TEX4 = 12, + VERT_ATTRIB_TEX5 = 13, + VERT_ATTRIB_TEX6 = 14, + VERT_ATTRIB_TEX7 = 15, + VERT_ATTRIB_POINT_SIZE = 16, + VERT_ATTRIB_GENERIC0 = 17, + VERT_ATTRIB_GENERIC1 = 18, + VERT_ATTRIB_GENERIC2 = 19, + VERT_ATTRIB_GENERIC3 = 20, + VERT_ATTRIB_GENERIC4 = 21, + VERT_ATTRIB_GENERIC5 = 22, + VERT_ATTRIB_GENERIC6 = 23, + VERT_ATTRIB_GENERIC7 = 24, + VERT_ATTRIB_GENERIC8 = 25, + VERT_ATTRIB_GENERIC9 = 26, + VERT_ATTRIB_GENERIC10 = 27, + VERT_ATTRIB_GENERIC11 = 28, + VERT_ATTRIB_GENERIC12 = 29, + VERT_ATTRIB_GENERIC13 = 30, + VERT_ATTRIB_GENERIC14 = 31, + VERT_ATTRIB_GENERIC15 = 32, + VERT_ATTRIB_MAX = 33 +} gl_vert_attrib; + +const char * gl_vert_attrib_name(gl_vert_attrib attrib); + +/** + * Symbolic constats to help iterating over + * specific blocks of vertex attributes. + * + * VERT_ATTRIB_FF + * includes all fixed function attributes as well as + * the aliased GL_NV_vertex_program shader attributes. + * VERT_ATTRIB_TEX + * include the classic texture coordinate attributes. + * Is a subset of VERT_ATTRIB_FF. + * VERT_ATTRIB_GENERIC + * include the OpenGL 2.0+ GLSL generic shader attributes. + * These alias the generic GL_ARB_vertex_shader attributes. + */ +#define VERT_ATTRIB_FF(i) (VERT_ATTRIB_POS + (i)) +#define VERT_ATTRIB_FF_MAX VERT_ATTRIB_GENERIC0 + +#define VERT_ATTRIB_TEX(i) (VERT_ATTRIB_TEX0 + (i)) +#define VERT_ATTRIB_TEX_MAX MAX_TEXTURE_COORD_UNITS + +#define VERT_ATTRIB_GENERIC(i) (VERT_ATTRIB_GENERIC0 + (i)) +#define VERT_ATTRIB_GENERIC_MAX MAX_VERTEX_GENERIC_ATTRIBS + +/** + * Bitflags for vertex attributes. + * These are used in bitfields in many places. + */ +/*@{*/ +#define VERT_BIT_POS BITFIELD64_BIT(VERT_ATTRIB_POS) +#define VERT_BIT_WEIGHT BITFIELD64_BIT(VERT_ATTRIB_WEIGHT) +#define VERT_BIT_NORMAL BITFIELD64_BIT(VERT_ATTRIB_NORMAL) +#define VERT_BIT_COLOR0 BITFIELD64_BIT(VERT_ATTRIB_COLOR0) +#define VERT_BIT_COLOR1 BITFIELD64_BIT(VERT_ATTRIB_COLOR1) +#define VERT_BIT_FOG BITFIELD64_BIT(VERT_ATTRIB_FOG) +#define VERT_BIT_COLOR_INDEX BITFIELD64_BIT(VERT_ATTRIB_COLOR_INDEX) +#define VERT_BIT_EDGEFLAG BITFIELD64_BIT(VERT_ATTRIB_EDGEFLAG) +#define VERT_BIT_TEX0 BITFIELD64_BIT(VERT_ATTRIB_TEX0) +#define VERT_BIT_TEX1 BITFIELD64_BIT(VERT_ATTRIB_TEX1) +#define VERT_BIT_TEX2 BITFIELD64_BIT(VERT_ATTRIB_TEX2) +#define VERT_BIT_TEX3 BITFIELD64_BIT(VERT_ATTRIB_TEX3) +#define VERT_BIT_TEX4 BITFIELD64_BIT(VERT_ATTRIB_TEX4) +#define VERT_BIT_TEX5 BITFIELD64_BIT(VERT_ATTRIB_TEX5) +#define VERT_BIT_TEX6 BITFIELD64_BIT(VERT_ATTRIB_TEX6) +#define VERT_BIT_TEX7 BITFIELD64_BIT(VERT_ATTRIB_TEX7) +#define VERT_BIT_POINT_SIZE BITFIELD64_BIT(VERT_ATTRIB_POINT_SIZE) +#define VERT_BIT_GENERIC0 BITFIELD64_BIT(VERT_ATTRIB_GENERIC0) + +#define VERT_BIT(i) BITFIELD64_BIT(i) +#define VERT_BIT_ALL BITFIELD64_RANGE(0, VERT_ATTRIB_MAX) + +#define VERT_BIT_FF(i) VERT_BIT(i) +#define VERT_BIT_FF_ALL BITFIELD64_RANGE(0, VERT_ATTRIB_FF_MAX) +#define VERT_BIT_TEX(i) VERT_BIT(VERT_ATTRIB_TEX(i)) +#define VERT_BIT_TEX_ALL \ + BITFIELD64_RANGE(VERT_ATTRIB_TEX(0), VERT_ATTRIB_TEX_MAX) + +#define VERT_BIT_GENERIC(i) VERT_BIT(VERT_ATTRIB_GENERIC(i)) +#define VERT_BIT_GENERIC_ALL \ + BITFIELD64_RANGE(VERT_ATTRIB_GENERIC(0), VERT_ATTRIB_GENERIC_MAX) +/*@}*/ + + +/** + * Indexes for vertex shader outputs, geometry shader inputs/outputs, and + * fragment shader inputs. + * + * Note that some of these values are not available to all pipeline stages. + * + * When this enum is updated, the following code must be updated too: + * - vertResults (in prog_print.c's arb_output_attrib_string()) + * - fragAttribs (in prog_print.c's arb_input_attrib_string()) + * - _mesa_varying_slot_in_fs() + */ +typedef enum +{ + VARYING_SLOT_POS, + VARYING_SLOT_COL0, /* COL0 and COL1 must be contiguous */ + VARYING_SLOT_COL1, + VARYING_SLOT_FOGC, + VARYING_SLOT_TEX0, /* TEX0-TEX7 must be contiguous */ + VARYING_SLOT_TEX1, + VARYING_SLOT_TEX2, + VARYING_SLOT_TEX3, + VARYING_SLOT_TEX4, + VARYING_SLOT_TEX5, + VARYING_SLOT_TEX6, + VARYING_SLOT_TEX7, + VARYING_SLOT_PSIZ, /* Does not appear in FS */ + VARYING_SLOT_BFC0, /* Does not appear in FS */ + VARYING_SLOT_BFC1, /* Does not appear in FS */ + VARYING_SLOT_EDGE, /* Does not appear in FS */ + VARYING_SLOT_CLIP_VERTEX, /* Does not appear in FS */ + VARYING_SLOT_CLIP_DIST0, + VARYING_SLOT_CLIP_DIST1, + VARYING_SLOT_PRIMITIVE_ID, /* Does not appear in VS */ + VARYING_SLOT_LAYER, /* Appears as VS or GS output */ + VARYING_SLOT_VIEWPORT, /* Appears as VS or GS output */ + VARYING_SLOT_FACE, /* FS only */ + VARYING_SLOT_PNTC, /* FS only */ + VARYING_SLOT_TESS_LEVEL_OUTER, /* Only appears as TCS output. */ + VARYING_SLOT_TESS_LEVEL_INNER, /* Only appears as TCS output. */ + VARYING_SLOT_VAR0, /* First generic varying slot */ + /* the remaining are simply for the benefit of gl_varying_slot_name() + * and not to be construed as an upper bound: + */ + VARYING_SLOT_VAR1, + VARYING_SLOT_VAR2, + VARYING_SLOT_VAR3, + VARYING_SLOT_VAR4, + VARYING_SLOT_VAR5, + VARYING_SLOT_VAR6, + VARYING_SLOT_VAR7, + VARYING_SLOT_VAR8, + VARYING_SLOT_VAR9, + VARYING_SLOT_VAR10, + VARYING_SLOT_VAR11, + VARYING_SLOT_VAR12, + VARYING_SLOT_VAR13, + VARYING_SLOT_VAR14, + VARYING_SLOT_VAR15, + VARYING_SLOT_VAR16, + VARYING_SLOT_VAR17, + VARYING_SLOT_VAR18, + VARYING_SLOT_VAR19, + VARYING_SLOT_VAR20, + VARYING_SLOT_VAR21, + VARYING_SLOT_VAR22, + VARYING_SLOT_VAR23, + VARYING_SLOT_VAR24, + VARYING_SLOT_VAR25, + VARYING_SLOT_VAR26, + VARYING_SLOT_VAR27, + VARYING_SLOT_VAR28, + VARYING_SLOT_VAR29, + VARYING_SLOT_VAR30, + VARYING_SLOT_VAR31, +} gl_varying_slot; + + +#define VARYING_SLOT_MAX (VARYING_SLOT_VAR0 + MAX_VARYING) +#define VARYING_SLOT_PATCH0 (VARYING_SLOT_MAX) +#define VARYING_SLOT_TESS_MAX (VARYING_SLOT_PATCH0 + MAX_VARYING) + +const char * gl_varying_slot_name(gl_varying_slot slot); + +/** + * Bitflags for varying slots. + */ +/*@{*/ +#define VARYING_BIT_POS BITFIELD64_BIT(VARYING_SLOT_POS) +#define VARYING_BIT_COL0 BITFIELD64_BIT(VARYING_SLOT_COL0) +#define VARYING_BIT_COL1 BITFIELD64_BIT(VARYING_SLOT_COL1) +#define VARYING_BIT_FOGC BITFIELD64_BIT(VARYING_SLOT_FOGC) +#define VARYING_BIT_TEX0 BITFIELD64_BIT(VARYING_SLOT_TEX0) +#define VARYING_BIT_TEX1 BITFIELD64_BIT(VARYING_SLOT_TEX1) +#define VARYING_BIT_TEX2 BITFIELD64_BIT(VARYING_SLOT_TEX2) +#define VARYING_BIT_TEX3 BITFIELD64_BIT(VARYING_SLOT_TEX3) +#define VARYING_BIT_TEX4 BITFIELD64_BIT(VARYING_SLOT_TEX4) +#define VARYING_BIT_TEX5 BITFIELD64_BIT(VARYING_SLOT_TEX5) +#define VARYING_BIT_TEX6 BITFIELD64_BIT(VARYING_SLOT_TEX6) +#define VARYING_BIT_TEX7 BITFIELD64_BIT(VARYING_SLOT_TEX7) +#define VARYING_BIT_TEX(U) BITFIELD64_BIT(VARYING_SLOT_TEX0 + (U)) +#define VARYING_BITS_TEX_ANY BITFIELD64_RANGE(VARYING_SLOT_TEX0, \ + MAX_TEXTURE_COORD_UNITS) +#define VARYING_BIT_PSIZ BITFIELD64_BIT(VARYING_SLOT_PSIZ) +#define VARYING_BIT_BFC0 BITFIELD64_BIT(VARYING_SLOT_BFC0) +#define VARYING_BIT_BFC1 BITFIELD64_BIT(VARYING_SLOT_BFC1) +#define VARYING_BIT_EDGE BITFIELD64_BIT(VARYING_SLOT_EDGE) +#define VARYING_BIT_CLIP_VERTEX BITFIELD64_BIT(VARYING_SLOT_CLIP_VERTEX) +#define VARYING_BIT_CLIP_DIST0 BITFIELD64_BIT(VARYING_SLOT_CLIP_DIST0) +#define VARYING_BIT_CLIP_DIST1 BITFIELD64_BIT(VARYING_SLOT_CLIP_DIST1) +#define VARYING_BIT_PRIMITIVE_ID BITFIELD64_BIT(VARYING_SLOT_PRIMITIVE_ID) +#define VARYING_BIT_LAYER BITFIELD64_BIT(VARYING_SLOT_LAYER) +#define VARYING_BIT_VIEWPORT BITFIELD64_BIT(VARYING_SLOT_VIEWPORT) +#define VARYING_BIT_FACE BITFIELD64_BIT(VARYING_SLOT_FACE) +#define VARYING_BIT_PNTC BITFIELD64_BIT(VARYING_SLOT_PNTC) +#define VARYING_BIT_TESS_LEVEL_OUTER BITFIELD64_BIT(VARYING_SLOT_TESS_LEVEL_OUTER) +#define VARYING_BIT_TESS_LEVEL_INNER BITFIELD64_BIT(VARYING_SLOT_TESS_LEVEL_INNER) +#define VARYING_BIT_VAR(V) BITFIELD64_BIT(VARYING_SLOT_VAR0 + (V)) +/*@}*/ + +/** + * Bitflags for system values. + */ +#define SYSTEM_BIT_SAMPLE_ID ((uint64_t)1 << SYSTEM_VALUE_SAMPLE_ID) +#define SYSTEM_BIT_SAMPLE_POS ((uint64_t)1 << SYSTEM_VALUE_SAMPLE_POS) +#define SYSTEM_BIT_SAMPLE_MASK_IN ((uint64_t)1 << SYSTEM_VALUE_SAMPLE_MASK_IN) +#define SYSTEM_BIT_LOCAL_INVOCATION_ID ((uint64_t)1 << SYSTEM_VALUE_LOCAL_INVOCATION_ID) + +/** + * If the gl_register_file is PROGRAM_SYSTEM_VALUE, the register index will be + * one of these values. If a NIR variable's mode is nir_var_system_value, it + * will be one of these values. + */ +typedef enum +{ + /** + * \name Vertex shader system values + */ + /*@{*/ + /** + * OpenGL-style vertex ID. + * + * Section 2.11.7 (Shader Execution), subsection Shader Inputs, of the + * OpenGL 3.3 core profile spec says: + * + * "gl_VertexID holds the integer index i implicitly passed by + * DrawArrays or one of the other drawing commands defined in section + * 2.8.3." + * + * Section 2.8.3 (Drawing Commands) of the same spec says: + * + * "The commands....are equivalent to the commands with the same base + * name (without the BaseVertex suffix), except that the ith element + * transferred by the corresponding draw call will be taken from + * element indices[i] + basevertex of each enabled array." + * + * Additionally, the overview in the GL_ARB_shader_draw_parameters spec + * says: + * + * "In unextended GL, vertex shaders have inputs named gl_VertexID and + * gl_InstanceID, which contain, respectively the index of the vertex + * and instance. The value of gl_VertexID is the implicitly passed + * index of the vertex being processed, which includes the value of + * baseVertex, for those commands that accept it." + * + * gl_VertexID gets basevertex added in. This differs from DirectX where + * SV_VertexID does \b not get basevertex added in. + * + * \note + * If all system values are available, \c SYSTEM_VALUE_VERTEX_ID will be + * equal to \c SYSTEM_VALUE_VERTEX_ID_ZERO_BASE plus + * \c SYSTEM_VALUE_BASE_VERTEX. + * + * \sa SYSTEM_VALUE_VERTEX_ID_ZERO_BASE, SYSTEM_VALUE_BASE_VERTEX + */ + SYSTEM_VALUE_VERTEX_ID, + + /** + * Instanced ID as supplied to gl_InstanceID + * + * Values assigned to gl_InstanceID always begin with zero, regardless of + * the value of baseinstance. + * + * Section 11.1.3.9 (Shader Inputs) of the OpenGL 4.4 core profile spec + * says: + * + * "gl_InstanceID holds the integer instance number of the current + * primitive in an instanced draw call (see section 10.5)." + * + * Through a big chain of pseudocode, section 10.5 describes that + * baseinstance is not counted by gl_InstanceID. In that section, notice + * + * "If an enabled vertex attribute array is instanced (it has a + * non-zero divisor as specified by VertexAttribDivisor), the element + * index that is transferred to the GL, for all vertices, is given by + * + * floor(instance/divisor) + baseinstance + * + * If an array corresponding to an attribute required by a vertex + * shader is not enabled, then the corresponding element is taken from + * the current attribute state (see section 10.2)." + * + * Note that baseinstance is \b not included in the value of instance. + */ + SYSTEM_VALUE_INSTANCE_ID, + + /** + * DirectX-style vertex ID. + * + * Unlike \c SYSTEM_VALUE_VERTEX_ID, this system value does \b not include + * the value of basevertex. + * + * \sa SYSTEM_VALUE_VERTEX_ID, SYSTEM_VALUE_BASE_VERTEX + */ + SYSTEM_VALUE_VERTEX_ID_ZERO_BASE, + + /** + * Value of \c basevertex passed to \c glDrawElementsBaseVertex and similar + * functions. + * + * \sa SYSTEM_VALUE_VERTEX_ID, SYSTEM_VALUE_VERTEX_ID_ZERO_BASE + */ + SYSTEM_VALUE_BASE_VERTEX, + /*@}*/ + + /** + * \name Geometry shader system values + */ + /*@{*/ + SYSTEM_VALUE_INVOCATION_ID, /**< (Also in Tessellation Control shader) */ + /*@}*/ + + /** + * \name Fragment shader system values + */ + /*@{*/ + SYSTEM_VALUE_FRONT_FACE, /**< (not done yet) */ + SYSTEM_VALUE_SAMPLE_ID, + SYSTEM_VALUE_SAMPLE_POS, + SYSTEM_VALUE_SAMPLE_MASK_IN, + /*@}*/ + + /** + * \name Tessellation Evaluation shader system values + */ + /*@{*/ + SYSTEM_VALUE_TESS_COORD, + SYSTEM_VALUE_VERTICES_IN, /**< Tessellation vertices in input patch */ + SYSTEM_VALUE_PRIMITIVE_ID, + SYSTEM_VALUE_TESS_LEVEL_OUTER, /**< TES input */ + SYSTEM_VALUE_TESS_LEVEL_INNER, /**< TES input */ + /*@}*/ + + /** + * \name Compute shader system values + */ + /*@{*/ + SYSTEM_VALUE_LOCAL_INVOCATION_ID, + SYSTEM_VALUE_WORK_GROUP_ID, + SYSTEM_VALUE_NUM_WORK_GROUPS, + /*@}*/ + + /** + * Driver internal vertex-count, used (for example) for drivers to + * calculate stride for stream-out outputs. Not externally visible. + */ + SYSTEM_VALUE_VERTEX_CNT, + + SYSTEM_VALUE_MAX /**< Number of values */ +} gl_system_value; + +const char * gl_system_value_name(gl_system_value sysval); + +/** + * The possible interpolation qualifiers that can be applied to a fragment + * shader input in GLSL. + * + * Note: INTERP_QUALIFIER_NONE must be 0 so that memsetting the + * gl_fragment_program data structure to 0 causes the default behavior. + */ +enum glsl_interp_qualifier +{ + INTERP_QUALIFIER_NONE = 0, + INTERP_QUALIFIER_SMOOTH, + INTERP_QUALIFIER_FLAT, + INTERP_QUALIFIER_NOPERSPECTIVE, + INTERP_QUALIFIER_COUNT /**< Number of interpolation qualifiers */ +}; + +const char * glsl_interp_qualifier_name(enum glsl_interp_qualifier qual); + +/** + * Fragment program results + */ +typedef enum +{ + FRAG_RESULT_DEPTH = 0, + FRAG_RESULT_STENCIL = 1, + /* If a single color should be written to all render targets, this + * register is written. No FRAG_RESULT_DATAn will be written. + */ + FRAG_RESULT_COLOR = 2, + FRAG_RESULT_SAMPLE_MASK = 3, + + /* FRAG_RESULT_DATAn are the per-render-target (GLSL gl_FragData[n] + * or ARB_fragment_program fragment.color[n]) color results. If + * any are written, FRAG_RESULT_COLOR will not be written. + * FRAG_RESULT_DATA1 and up are simply for the benefit of + * gl_frag_result_name() and not to be construed as an upper bound + */ + FRAG_RESULT_DATA0 = 4, + FRAG_RESULT_DATA1, + FRAG_RESULT_DATA2, + FRAG_RESULT_DATA3, + FRAG_RESULT_DATA4, + FRAG_RESULT_DATA5, + FRAG_RESULT_DATA6, + FRAG_RESULT_DATA7, +} gl_frag_result; + +const char * gl_frag_result_name(gl_frag_result result); + +#define FRAG_RESULT_MAX (FRAG_RESULT_DATA0 + MAX_DRAW_BUFFERS) + +/** + * \brief Layout qualifiers for gl_FragDepth. + * + * Extension AMD_conservative_depth allows gl_FragDepth to be redeclared with + * a layout qualifier. + * + * \see enum ir_depth_layout + */ +enum gl_frag_depth_layout +{ + FRAG_DEPTH_LAYOUT_NONE, /**< No layout is specified. */ + FRAG_DEPTH_LAYOUT_ANY, + FRAG_DEPTH_LAYOUT_GREATER, + FRAG_DEPTH_LAYOUT_LESS, + FRAG_DEPTH_LAYOUT_UNCHANGED +}; + +#endif /* SHADER_ENUMS_H */ |