Rename to axslcc [release 1.9.6]

.github/workflows/build.yml

@@ -14,12 +14,12 @@ jobs:
       shell: pwsh
       run: |
         cmake -B build
-        cmake --build build --config Release --target glslcc
+        cmake --build build --config Release --target axslcc
     - name: Upload
       uses: actions/upload-artifact@v4
       with:
-        path: ./build/src/Release/glslcc.exe
-        name: glslcc-win64
+        path: ./build/src/Release/axslcc.exe
+        name: axslcc-win64
   linux:
     runs-on: ubuntu-latest
     steps:
@@ -28,31 +28,31 @@ jobs:
       shell: pwsh
       run: |
         cmake -B build -DCMAKE_BUILD_TYPE=Release
-        cmake --build build --config Release --target glslcc
+        cmake --build build --config Release --target axslcc
     - name: Upload
       uses: actions/upload-artifact@v4
       with:
-        path: ./build/src/glslcc
-        name: glslcc-linux
+        path: ./build/src/axslcc
+        name: axslcc-linux
   osx-arm64:
     runs-on: macos-latest
     steps:
     - uses: actions/checkout@v4
     - name: Setup Xcode version
       uses: maxim-lobanov/setup-xcode@v1
       with:
-        xcode-version: '14.3'
+        xcode-version: '15.0'
     - name: build
       shell: pwsh
       run: |
         cmake -B build -DCMAKE_BUILD_TYPE=Release "-DCMAKE_OSX_DEPLOYMENT_TARGET=10.15"
-        cmake --build build --config Release --target glslcc
-        lipo -info ./build/src/glslcc
+        cmake --build build --config Release --target axslcc
+        lipo -info ./build/src/axslcc
     - name: Upload
       uses: actions/upload-artifact@v4
       with:
-        path: ./build/src/glslcc
-        name: glslcc-osx-arm64
+        path: ./build/src/axslcc
+        name: axslcc-osx-arm64
 
   osx-x64:
     runs-on: macos-13
@@ -61,15 +61,15 @@ jobs:
     - name: Setup Xcode version
       uses: maxim-lobanov/setup-xcode@v1
       with:
-        xcode-version: '14.2'
+        xcode-version: '15.0'
     - name: build
       shell: pwsh
       run: |
         cmake -B build -DCMAKE_BUILD_TYPE=Release "-DCMAKE_OSX_DEPLOYMENT_TARGET=10.15"
-        cmake --build build --config Release --target glslcc
-        lipo -info ./build/src/glslcc
+        cmake --build build --config Release --target axslcc
+        lipo -info ./build/src/axslcc
     - name: Upload
       uses: actions/upload-artifact@v4
       with:
-        path: ./build/src/glslcc
-        name: glslcc-osx-x64
+        path: ./build/src/axslcc
+        name: axslcc-osx-x64

.github/workflows/dist.yml

@@ -55,23 +55,23 @@ jobs:
     - name: Create packages
       if: ${{ steps.check_ver.outputs.release_ver != '' }}
       run: |
-        curl -L https://github.com/simdsoft/1kiss/releases/download/devtools/d3dcompiler_47.dll -o ./glslcc-win64/d3dcompiler_47.dll
-        ls -l ./glslcc-win64
+        curl -L https://github.com/simdsoft/1kiss/releases/download/devtools/d3dcompiler_47.dll -o ./axslcc-win64/d3dcompiler_47.dll
+        ls -l ./axslcc-win64
         mkdir -p pkg_dir
-        prefix=glslcc-${{ steps.check_ver.outputs.release_ver }}
-        ls -l glslcc-linux
-        ls -l glslcc-osx-x64
-        ls -l glslcc-osx-arm64
-        sudo chmod u+x ./glslcc-linux/glslcc
-        sudo chmod u+x ./glslcc-osx-x64/glslcc
-        sudo chmod u+x ./glslcc-osx-arm64/glslcc
-        ls -l glslcc-linux
-        ls -l glslcc-osx-x64
-        ls -l glslcc-osx-arm64
-        tar -czvf ./pkg_dir/$prefix-linux.tar.gz -C ./glslcc-linux glslcc
-        tar -czvf ./pkg_dir/$prefix-osx-x64.tar.gz -C ./glslcc-osx-x64 glslcc
-        tar -czvf ./pkg_dir/$prefix-osx-arm64.tar.gz -C ./glslcc-osx-arm64 glslcc
-        sh -c "cd ./glslcc-win64; zip ../pkg_dir/$prefix-win64.zip glslcc.exe d3dcompiler_47.dll"
+        prefix=axslcc-${{ steps.check_ver.outputs.release_ver }}
+        ls -l axslcc-linux
+        ls -l axslcc-osx-x64
+        ls -l axslcc-osx-arm64
+        sudo chmod u+x ./axslcc-linux/axslcc
+        sudo chmod u+x ./axslcc-osx-x64/axslcc
+        sudo chmod u+x ./axslcc-osx-arm64/axslcc
+        ls -l axslcc-linux
+        ls -l axslcc-osx-x64
+        ls -l axslcc-osx-arm64
+        tar -czvf ./pkg_dir/$prefix-linux.tar.gz -C ./axslcc-linux axslcc
+        tar -czvf ./pkg_dir/$prefix-osx-x64.tar.gz -C ./axslcc-osx-x64 axslcc
+        tar -czvf ./pkg_dir/$prefix-osx-arm64.tar.gz -C ./axslcc-osx-arm64 axslcc
+        sh -c "cd ./axslcc-win64; zip ../pkg_dir/$prefix-win64.zip axslcc.exe d3dcompiler_47.dll"
     - name: Publish to github release page
       if: ${{ steps.check_ver.outputs.release_ver != '' }}
       uses: softprops/action-gh-release@v2
@@ -82,4 +82,4 @@ jobs:
           ./pkg_dir/*
       env:
         GITHUB_TOKEN: ${{ secrets.GITHUB_TOKEN }}
-        GITHUB_REPOSITORY: axmolengine/glslcc
+        GITHUB_REPOSITORY: axmolengine/axslcc

.gitignore

@@ -58,3 +58,4 @@ install
 *.obj
 *.zip
 *.gz
+3rdparty/glslang

3rdparty/spirv-cross/CMakeLists.txt

@@ -19,7 +19,7 @@
 #  2. The MIT License, found at <http://opensource.org/licenses/MIT>.
 #
 
-cmake_minimum_required(VERSION 3.0)
+cmake_minimum_required(VERSION 3.6)
 set(CMAKE_CXX_STANDARD 11)
 set(CMAKE_CXX_EXTENSIONS OFF)
 
@@ -28,6 +28,9 @@ if (${CMAKE_VERSION} VERSION_GREATER "3.0.1")
 	cmake_policy(SET CMP0048 NEW)
 endif()
 
+# Avoid error on Android NDK 27+'s CMake files
+cmake_policy(SET CMP0057 NEW)
+
 project(SPIRV-Cross LANGUAGES CXX C)
 enable_testing()
 
@@ -77,7 +80,7 @@ set(spirv-cross-build-version "unknown")
 find_package(Git)
 if (GIT_FOUND)
 	execute_process(
-		COMMAND ${GIT_EXECUTABLE} rev-parse --short HEAD
+		COMMAND ${GIT_EXECUTABLE} describe --always --tags --dirty=+
 		WORKING_DIRECTORY ${CMAKE_CURRENT_SOURCE_DIR}
 		OUTPUT_VARIABLE spirv-cross-build-version
 		ERROR_QUIET
@@ -243,7 +246,7 @@ set(spirv-cross-util-sources
 		${CMAKE_CURRENT_SOURCE_DIR}/spirv_cross_util.hpp)
 
 set(spirv-cross-abi-major 0)
-set(spirv-cross-abi-minor 58)
+set(spirv-cross-abi-minor 64)
 set(spirv-cross-abi-patch 0)
 set(SPIRV_CROSS_VERSION ${spirv-cross-abi-major}.${spirv-cross-abi-minor}.${spirv-cross-abi-patch})
 
@@ -465,7 +468,19 @@ if (SPIRV_CROSS_CLI)
 		#  - Update the reference files
 		#  - Get cycle counts from malisc
 		#  - Keep failing outputs
-		find_package(PythonInterp)
+		if (${CMAKE_VERSION} VERSION_GREATER "3.12")
+			find_package(Python3)
+			if (${PYTHON3_FOUND})
+				set(PYTHONINTERP_FOUND ON)
+				set(PYTHON_VERSION_MAJOR 3)
+				set(PYTHON_EXECUTABLE ${Python3_EXECUTABLE})
+			else()
+				set(PYTHONINTERP_FOUND OFF)
+			endif()
+		else()
+			find_package(PythonInterp)
+		endif()
+
 		find_program(spirv-cross-glslang NAMES glslangValidator
 				PATHS ${CMAKE_CURRENT_SOURCE_DIR}/external/glslang-build/output/bin
 				NO_DEFAULT_PATH)

3rdparty/spirv-cross/main.cpp

@@ -675,11 +675,13 @@ struct CLIArguments
 	bool msl_force_sample_rate_shading = false;
 	bool msl_manual_helper_invocation_updates = true;
 	bool msl_check_discarded_frag_stores = false;
+	bool msl_force_fragment_with_side_effects_execution = false;
 	bool msl_sample_dref_lod_array_as_grad = false;
 	bool msl_runtime_array_rich_descriptor = false;
 	bool msl_replace_recursive_inputs = false;
 	bool msl_readwrite_texture_fences = true;
 	bool msl_agx_manual_cube_grad_fixup = false;
+	bool msl_input_attachment_is_ds_attachment = false;
 	const char *msl_combined_sampler_suffix = nullptr;
 	bool glsl_emit_push_constant_as_ubo = false;
 	bool glsl_emit_ubo_as_plain_uniforms = false;
@@ -872,6 +874,10 @@ static void print_help_msl()
 	                "\t[--msl-runtime-array-rich-descriptor]:\n\t\tWhen declaring a runtime array of SSBOs, declare an array of {ptr, len} pairs to support OpArrayLength.\n"
 	                "\t[--msl-replace-recursive-inputs]:\n\t\tWorks around a Metal 3.1 regression bug, which causes an infinite recursion crash during Metal's analysis of an entry point input structure that itself contains internal recursion.\n"
 	                "\t[--msl-texture-buffer-native]:\n\t\tEnable native support for texel buffers. Otherwise, it is emulated as a normal texture.\n"
+	                "\t[--msl-input-attachment-is-ds-attachment]:\n\t\tAdds a simple depth passthrough in fragment shaders when they do not modify the depth value.\n"
+	                "\t\tRequired to force Metal to write to the depth/stencil attachment post fragment execution.\n"
+	                "\t\tOtherwise, Metal may optimize the write to pre fragment execution which goes against the Vulkan spec.\n"
+	                "\t\tOnly required if an input attachment and depth/stencil attachment reference the same resource.\n"
 	                "\t[--msl-framebuffer-fetch]:\n\t\tImplement subpass inputs with frame buffer fetch.\n"
 	                "\t\tEmits [[color(N)]] inputs in fragment stage.\n"
 	                "\t\tRequires an Apple GPU.\n"
@@ -956,6 +962,15 @@ static void print_help_msl()
 	                "\t\tSome Metal devices have a bug where stores to resources from a fragment shader\n"
 	                "\t\tcontinue to execute, even when the fragment is discarded. These checks\n"
 	                "\t\tprevent these stores from executing.\n"
+	                "\t[--msl-force-frag-execution]:\n\t\tEnforces fragment execution to avoid early discard by Metal\n"
+	                "\t\tMetal will prematurely discard fragments before execution when side effects are present.\n"
+	                "\t\tThis condition is triggered under the following conditions (side effect operations happen before discard):\n"
+	                "\t\t\t1. Pre fragment depth test fails.\n"
+	                "\t\t\t2. Modify depth value in fragment shader to constant value known at compile time.\n"
+	                "\t\t\t3. Constant value will not pass post fragment depth test.\n"
+	                "\t\t\t4. Fragment is always discarded in fragment execution.\n"
+	                "\t\tHowever, Vulkan expects fragment shader to be executed since it cannot be discarded until the discard\n"
+	                "\t\tpresent in the fragment execution, which would also execute the operations with side effects.\n"
 	                "\t[--msl-sample-dref-lod-array-as-grad]:\n\t\tUse a gradient instead of a level argument.\n"
 	                "\t\tSome Metal devices have a bug where the level() argument to\n"
 	                "\t\tdepth2d_array<T>::sample_compare() in a fragment shader is biased by some\n"
@@ -1242,10 +1257,12 @@ static string compile_iteration(const CLIArguments &args, std::vector<uint32_t>
 		msl_opts.force_sample_rate_shading = args.msl_force_sample_rate_shading;
 		msl_opts.manual_helper_invocation_updates = args.msl_manual_helper_invocation_updates;
 		msl_opts.check_discarded_frag_stores = args.msl_check_discarded_frag_stores;
+		msl_opts.force_fragment_with_side_effects_execution = args.msl_force_fragment_with_side_effects_execution;
 		msl_opts.sample_dref_lod_array_as_grad = args.msl_sample_dref_lod_array_as_grad;
 		msl_opts.ios_support_base_vertex_instance = true;
 		msl_opts.runtime_array_rich_descriptor = args.msl_runtime_array_rich_descriptor;
 		msl_opts.replace_recursive_inputs = args.msl_replace_recursive_inputs;
+		msl_opts.input_attachment_is_ds_attachment = args.msl_input_attachment_is_ds_attachment;
 		msl_opts.readwrite_texture_fences = args.msl_readwrite_texture_fences;
 		msl_opts.agx_manual_cube_grad_fixup = args.msl_agx_manual_cube_grad_fixup;
 		msl_comp->set_msl_options(msl_opts);
@@ -1800,6 +1817,7 @@ static int main_inner(int argc, char *argv[])
 	cbs.add("--msl-no-manual-helper-invocation-updates",
 	        [&args](CLIParser &) { args.msl_manual_helper_invocation_updates = false; });
 	cbs.add("--msl-check-discarded-frag-stores", [&args](CLIParser &) { args.msl_check_discarded_frag_stores = true; });
+	cbs.add("--msl-force-frag-with-side-effects-execution", [&args](CLIParser &) { args.msl_force_fragment_with_side_effects_execution = true; });
 	cbs.add("--msl-sample-dref-lod-array-as-grad",
 	        [&args](CLIParser &) { args.msl_sample_dref_lod_array_as_grad = true; });
 	cbs.add("--msl-no-readwrite-texture-fences", [&args](CLIParser &) { args.msl_readwrite_texture_fences = false; });
@@ -1811,6 +1829,7 @@ static int main_inner(int argc, char *argv[])
 	        [&args](CLIParser &) { args.msl_runtime_array_rich_descriptor = true; });
 	cbs.add("--msl-replace-recursive-inputs",
 	        [&args](CLIParser &) { args.msl_replace_recursive_inputs = true; });
+	cbs.add("--msl-input-attachment-is-ds-attachment", [&args](CLIParser &) { args.msl_input_attachment_is_ds_attachment = true; });
 	cbs.add("--extension", [&args](CLIParser &parser) { args.extensions.push_back(parser.next_string()); });
 	cbs.add("--rename-entry-point", [&args](CLIParser &parser) {
 		auto old_name = parser.next_string();

3rdparty/spirv-cross/spirv.h

@@ -1,5 +1,5 @@
 /*
-** Copyright (c) 2014-2020 The Khronos Group Inc.
+** Copyright (c) 2014-2024 The Khronos Group Inc.
 ** 
 ** Permission is hereby granted, free of charge, to any person obtaining a copy
 ** of this software and/or associated documentation files (the "Materials"),
@@ -509,6 +509,9 @@ typedef enum SpvDecoration_ {
     SpvDecorationMaxByteOffsetId = 47,
     SpvDecorationNoSignedWrap = 4469,
     SpvDecorationNoUnsignedWrap = 4470,
+    SpvDecorationWeightTextureQCOM = 4487,
+    SpvDecorationBlockMatchTextureQCOM = 4488,
+    SpvDecorationBlockMatchSamplerQCOM = 4499,
     SpvDecorationExplicitInterpAMD = 4999,
     SpvDecorationOverrideCoverageNV = 5248,
     SpvDecorationPassthroughNV = 5250,
@@ -991,6 +994,10 @@ typedef enum SpvCapability_ {
     SpvCapabilityRayQueryKHR = 4472,
     SpvCapabilityRayTraversalPrimitiveCullingKHR = 4478,
     SpvCapabilityRayTracingKHR = 4479,
+    SpvCapabilityTextureSampleWeightedQCOM = 4484,
+    SpvCapabilityTextureBoxFilterQCOM = 4485,
+    SpvCapabilityTextureBlockMatchQCOM = 4486,
+    SpvCapabilityTextureBlockMatch2QCOM = 4498,
     SpvCapabilityFloat16ImageAMD = 5008,
     SpvCapabilityImageGatherBiasLodAMD = 5009,
     SpvCapabilityFragmentMaskAMD = 5010,
@@ -1596,6 +1603,14 @@ typedef enum SpvOp_ {
     SpvOpRayQueryConfirmIntersectionKHR = 4476,
     SpvOpRayQueryProceedKHR = 4477,
     SpvOpRayQueryGetIntersectionTypeKHR = 4479,
+    SpvOpImageSampleWeightedQCOM = 4480,
+    SpvOpImageBoxFilterQCOM = 4481,
+    SpvOpImageBlockMatchSSDQCOM = 4482,
+    SpvOpImageBlockMatchSADQCOM = 4483,
+    SpvOpImageBlockMatchWindowSSDQCOM = 4500,
+    SpvOpImageBlockMatchWindowSADQCOM = 4501,
+    SpvOpImageBlockMatchGatherSSDQCOM = 4502,
+    SpvOpImageBlockMatchGatherSADQCOM = 4503,
     SpvOpGroupIAddNonUniformAMD = 5000,
     SpvOpGroupFAddNonUniformAMD = 5001,
     SpvOpGroupFMinNonUniformAMD = 5002,
@@ -2271,6 +2286,14 @@ inline void SpvHasResultAndType(SpvOp opcode, bool *hasResult, bool *hasResultTy
     case SpvOpRayQueryConfirmIntersectionKHR: *hasResult = false; *hasResultType = false; break;
     case SpvOpRayQueryProceedKHR: *hasResult = true; *hasResultType = true; break;
     case SpvOpRayQueryGetIntersectionTypeKHR: *hasResult = true; *hasResultType = true; break;
+    case SpvOpImageSampleWeightedQCOM: *hasResult = true; *hasResultType = true; break;
+    case SpvOpImageBoxFilterQCOM: *hasResult = true; *hasResultType = true; break;
+    case SpvOpImageBlockMatchSSDQCOM: *hasResult = true; *hasResultType = true; break;
+    case SpvOpImageBlockMatchSADQCOM: *hasResult = true; *hasResultType = true; break;
+    case SpvOpImageBlockMatchWindowSSDQCOM: *hasResult = true; *hasResultType = true; break;
+    case SpvOpImageBlockMatchWindowSADQCOM: *hasResult = true; *hasResultType = true; break;
+    case SpvOpImageBlockMatchGatherSSDQCOM: *hasResult = true; *hasResultType = true; break;
+    case SpvOpImageBlockMatchGatherSADQCOM: *hasResult = true; *hasResultType = true; break;
     case SpvOpGroupIAddNonUniformAMD: *hasResult = true; *hasResultType = true; break;
     case SpvOpGroupFAddNonUniformAMD: *hasResult = true; *hasResultType = true; break;
     case SpvOpGroupFMinNonUniformAMD: *hasResult = true; *hasResultType = true; break;

3rdparty/spirv-cross/spirv.hpp

@@ -1,4 +1,4 @@
-// Copyright (c) 2014-2020 The Khronos Group Inc.
+// Copyright (c) 2014-2024 The Khronos Group Inc.
 // 
 // Permission is hereby granted, free of charge, to any person obtaining a copy
 // of this software and/or associated documentation files (the "Materials"),
@@ -505,6 +505,9 @@ enum Decoration {
     DecorationMaxByteOffsetId = 47,
     DecorationNoSignedWrap = 4469,
     DecorationNoUnsignedWrap = 4470,
+    DecorationWeightTextureQCOM = 4487,
+    DecorationBlockMatchTextureQCOM = 4488,
+    DecorationBlockMatchSamplerQCOM = 4499,
     DecorationExplicitInterpAMD = 4999,
     DecorationOverrideCoverageNV = 5248,
     DecorationPassthroughNV = 5250,
@@ -987,6 +990,10 @@ enum Capability {
     CapabilityRayQueryKHR = 4472,
     CapabilityRayTraversalPrimitiveCullingKHR = 4478,
     CapabilityRayTracingKHR = 4479,
+    CapabilityTextureSampleWeightedQCOM = 4484,
+    CapabilityTextureBoxFilterQCOM = 4485,
+    CapabilityTextureBlockMatchQCOM = 4486,
+    CapabilityTextureBlockMatch2QCOM = 4498,
     CapabilityFloat16ImageAMD = 5008,
     CapabilityImageGatherBiasLodAMD = 5009,
     CapabilityFragmentMaskAMD = 5010,
@@ -1592,6 +1599,14 @@ enum Op {
     OpRayQueryConfirmIntersectionKHR = 4476,
     OpRayQueryProceedKHR = 4477,
     OpRayQueryGetIntersectionTypeKHR = 4479,
+    OpImageSampleWeightedQCOM = 4480,
+    OpImageBoxFilterQCOM = 4481,
+    OpImageBlockMatchSSDQCOM = 4482,
+    OpImageBlockMatchSADQCOM = 4483,
+    OpImageBlockMatchWindowSSDQCOM = 4500,
+    OpImageBlockMatchWindowSADQCOM = 4501,
+    OpImageBlockMatchGatherSSDQCOM = 4502,
+    OpImageBlockMatchGatherSADQCOM = 4503,
     OpGroupIAddNonUniformAMD = 5000,
     OpGroupFAddNonUniformAMD = 5001,
     OpGroupFMinNonUniformAMD = 5002,
@@ -2267,6 +2282,14 @@ inline void HasResultAndType(Op opcode, bool *hasResult, bool *hasResultType) {
     case OpRayQueryConfirmIntersectionKHR: *hasResult = false; *hasResultType = false; break;
     case OpRayQueryProceedKHR: *hasResult = true; *hasResultType = true; break;
     case OpRayQueryGetIntersectionTypeKHR: *hasResult = true; *hasResultType = true; break;
+    case OpImageSampleWeightedQCOM: *hasResult = true; *hasResultType = true; break;
+    case OpImageBoxFilterQCOM: *hasResult = true; *hasResultType = true; break;
+    case OpImageBlockMatchSSDQCOM: *hasResult = true; *hasResultType = true; break;
+    case OpImageBlockMatchSADQCOM: *hasResult = true; *hasResultType = true; break;
+    case OpImageBlockMatchWindowSSDQCOM: *hasResult = true; *hasResultType = true; break;
+    case OpImageBlockMatchWindowSADQCOM: *hasResult = true; *hasResultType = true; break;
+    case OpImageBlockMatchGatherSSDQCOM: *hasResult = true; *hasResultType = true; break;
+    case OpImageBlockMatchGatherSADQCOM: *hasResult = true; *hasResultType = true; break;
     case OpGroupIAddNonUniformAMD: *hasResult = true; *hasResultType = true; break;
     case OpGroupFAddNonUniformAMD: *hasResult = true; *hasResultType = true; break;
     case OpGroupFMinNonUniformAMD: *hasResult = true; *hasResultType = true; break;