llvmbot wrote:
<!--LLVM PR SUMMARY COMMENT--> @llvm/pr-subscribers-hlsl @llvm/pr-subscribers-clang-codegen Author: None (joaosaffran) <details> <summary>Changes</summary> This patch removes optnone from HLSL entry functions and instead uses `dx.disable_optimizations` module flag to know when to enable `DisableOptimization` Shader Flag. This is part of: #<!-- -->167936 --- Full diff: https://github.com/llvm/llvm-project/pull/180069.diff 6 Files Affected: - (modified) clang/lib/CodeGen/CGHLSLRuntime.cpp (+8-1) - (modified) clang/test/CodeGenHLSL/inline-functions.hlsl (+5-9) - (modified) llvm/lib/Target/DirectX/DXILShaderFlags.cpp (+8-12) - (modified) llvm/test/CodeGen/DirectX/ShaderFlags/disable-opt-cs.ll (+6-2) - (modified) llvm/test/CodeGen/DirectX/ShaderFlags/disable-opt-lib.ll (+10-5) - (removed) llvm/test/CodeGen/DirectX/ShaderFlags/lib-entry-attr-error.ll (-30) ``````````diff diff --git a/clang/lib/CodeGen/CGHLSLRuntime.cpp b/clang/lib/CodeGen/CGHLSLRuntime.cpp index 985bc8640118e..006861a1efe43 100644 --- a/clang/lib/CodeGen/CGHLSLRuntime.cpp +++ b/clang/lib/CodeGen/CGHLSLRuntime.cpp @@ -267,6 +267,13 @@ static std::optional<llvm::Value *> initializeLocalResourceArray( return Index; } +static void addDisableOptimizations(llvm::Module &M) { + if (M.getModuleFlag("dx.disable_optimizations")) + return; + StringRef Key = "dx.disable_optimizations"; + M.addModuleFlag(llvm::Module::ModFlagBehavior::Override, Key, 1); +} + } // namespace llvm::Type * @@ -523,7 +530,7 @@ void clang::CodeGen::CGHLSLRuntime::setHLSLEntryAttributes( // hence not able to set attributes of the newly materialized entry functions. // So, set attributes of entry function here, as appropriate. if (CGM.getCodeGenOpts().OptimizationLevel == 0) - Fn->addFnAttr(llvm::Attribute::OptimizeNone); + addDisableOptimizations(CGM.getModule()); Fn->addFnAttr(llvm::Attribute::NoInline); if (CGM.getLangOpts().HLSLSpvEnableMaximalReconvergence) { diff --git a/clang/test/CodeGenHLSL/inline-functions.hlsl b/clang/test/CodeGenHLSL/inline-functions.hlsl index 0c7467e2f972e..d24c07cf5224f 100644 --- a/clang/test/CodeGenHLSL/inline-functions.hlsl +++ b/clang/test/CodeGenHLSL/inline-functions.hlsl @@ -1,9 +1,9 @@ -// RUN: %clang_cc1 -triple dxil-pc-shadermodel6.3-library %s -emit-llvm -disable-llvm-passes -o - | FileCheck %s --check-prefixes=CHECK,NOINLINE,OPT_ATTR -// RUN: %clang_cc1 -triple dxil-pc-shadermodel6.3-library %s -emit-llvm -O0 -o - | FileCheck %s --check-prefixes=CHECK,INLINE,OPT_ATTR -// RUN: %clang_cc1 -triple dxil-pc-shadermodel6.3-library %s -emit-llvm -O1 -o - | FileCheck %s --check-prefixes=CHECK,INLINE,NOOPT_ATTR +// RUN: %clang_cc1 -triple dxil-pc-shadermodel6.3-library %s -emit-llvm -disable-llvm-passes -o - | FileCheck %s --check-prefixes=CHECK,NOINLINE +// RUN: %clang_cc1 -triple dxil-pc-shadermodel6.3-library %s -emit-llvm -O0 -o - | FileCheck %s --check-prefixes=CHECK,INLINE +// RUN: %clang_cc1 -triple dxil-pc-shadermodel6.3-library %s -emit-llvm -O1 -o - | FileCheck %s --check-prefixes=CHECK,INLINE // RUN: %clang_cc1 -triple dxil-pc-shadermodel6.0-compute %s -emit-llvm -disable-llvm-passes -o - | FileCheck %s --check-prefixes=CHECK,NOINLINE -// RUN: %clang_cc1 -triple dxil-pc-shadermodel6.0-compute %s -emit-llvm -O0 -o - | FileCheck %s --check-prefixes=CHECK,INLINE,OPT_ATTR -// RUN: %clang_cc1 -triple dxil-pc-shadermodel6.0-compute %s -emit-llvm -O1 -o - | FileCheck %s --check-prefixes=CHECK,INLINE,NOOPT_ATTR +// RUN: %clang_cc1 -triple dxil-pc-shadermodel6.0-compute %s -emit-llvm -O0 -o - | FileCheck %s --check-prefixes=CHECK,INLINE +// RUN: %clang_cc1 -triple dxil-pc-shadermodel6.0-compute %s -emit-llvm -O1 -o - | FileCheck %s --check-prefixes=CHECK,INLINE // Tests that user functions will always be inlined. // This includes exported functions and mangled entry point implementation functions. @@ -71,8 +71,6 @@ RWBuffer<unsigned> Indices; // NOINLINE: ret void // The unmangled version is not inlined, EntryAttr reflects that -// OPT_ATTR: Function Attrs: {{.*}}optnone -// NOOPT_ATTR-NOT: Function Attrs: {{.*}}optnone // CHECK: define void @main() {{[a-z_ ]*}}[[EntryAttr:\#[0-9]+]] // Make sure function calls are inlined when AlwaysInline is run // This only leaves calls to llvm. intrinsics @@ -99,8 +97,6 @@ void main(unsigned int GI : SV_GroupIndex) { // NOINLINE: ret void // The unmangled version is not inlined, EntryAttr reflects that -// OPT_ATTR: Function Attrs: {{.*}}optnone -// NOOPT_ATTR-NOT: Function Attrs: {{.*}}optnone // CHECK: define void @main10() {{[a-z_ ]*}}[[EntryAttr]] // Make sure function calls are inlined when AlwaysInline is run // This only leaves calls to llvm. intrinsics diff --git a/llvm/lib/Target/DirectX/DXILShaderFlags.cpp b/llvm/lib/Target/DirectX/DXILShaderFlags.cpp index 54cf56f92277b..87d531489b757 100644 --- a/llvm/lib/Target/DirectX/DXILShaderFlags.cpp +++ b/llvm/lib/Target/DirectX/DXILShaderFlags.cpp @@ -104,6 +104,13 @@ static bool checkWaveOps(Intrinsic::ID IID) { } } +static bool isOptimizationDisabled(const Module &M) { + const StringRef Key = "dx.disable_optimizations"; + if (auto *Flag = mdconst::extract_or_null<ConstantInt>(M.getModuleFlag(Key))) + return Flag->getValue().getBoolValue(); + return false; +} + // Checks to see if the status bit from a load with status // instruction is ever extracted. If it is, the module needs // to have the TiledResources shader flag set. @@ -249,18 +256,7 @@ ModuleShaderFlags::gatherGlobalModuleFlags(const Module &M, ComputedShaderFlags CSF; - // Set DisableOptimizations flag based on the presence of OptimizeNone - // attribute of entry functions. - if (MMDI.EntryPropertyVec.size() > 0) { - CSF.DisableOptimizations = MMDI.EntryPropertyVec[0].Entry->hasFnAttribute( - llvm::Attribute::OptimizeNone); - // Ensure all entry functions have the same optimization attribute - for (const auto &EntryFunProps : MMDI.EntryPropertyVec) - if (CSF.DisableOptimizations != - EntryFunProps.Entry->hasFnAttribute(llvm::Attribute::OptimizeNone)) - EntryFunProps.Entry->getContext().diagnose(DiagnosticInfoUnsupported( - *(EntryFunProps.Entry), "Inconsistent optnone attribute ")); - } + CSF.DisableOptimizations = isOptimizationDisabled(M); CSF.UAVsAtEveryStage = hasUAVsAtEveryStage(DRM, MMDI); diff --git a/llvm/test/CodeGen/DirectX/ShaderFlags/disable-opt-cs.ll b/llvm/test/CodeGen/DirectX/ShaderFlags/disable-opt-cs.ll index 4bdb7ec50f6f4..4cb46043b41fc 100644 --- a/llvm/test/CodeGen/DirectX/ShaderFlags/disable-opt-cs.ll +++ b/llvm/test/CodeGen/DirectX/ShaderFlags/disable-opt-cs.ll @@ -18,7 +18,7 @@ target triple = "dxilv1.0-pc-shadermodel6.0-compute" -; Function Attrs: convergent noinline norecurse optnone +; Function Attrs: convergent noinline norecurse define void @main() #0 { entry: ret void @@ -30,5 +30,9 @@ entry: ret i32 0 } -attributes #0 = { convergent noinline norecurse optnone "hlsl.numthreads"="1,1,1" "hlsl.shader"="compute" "no-infs-fp-math"="true" "no-nans-fp-math"="true" "no-signed-zeros-fp-math"="true" "no-trapping-math"="true" "stack-protector-buffer-size"="8" } +attributes #0 = { convergent noinline norecurse "hlsl.numthreads"="1,1,1" "hlsl.shader"="compute" "no-infs-fp-math"="true" "no-nans-fp-math"="true" "no-signed-zeros-fp-math"="true" "no-trapping-math"="true" "stack-protector-buffer-size"="8" } attributes #1 = { alwaysinline convergent mustprogress norecurse nounwind "no-infs-fp-math"="true" "no-nans-fp-math"="true" "no-signed-zeros-fp-math"="true" "no-trapping-math"="true" "stack-protector-buffer-size"="8" } + +!llvm.module.flags = !{!0} + +!0 = !{i32 4, !"dx.disable_optimizations", i32 1} diff --git a/llvm/test/CodeGen/DirectX/ShaderFlags/disable-opt-lib.ll b/llvm/test/CodeGen/DirectX/ShaderFlags/disable-opt-lib.ll index 03756710adc3a..27d1e2a0910c3 100644 --- a/llvm/test/CodeGen/DirectX/ShaderFlags/disable-opt-lib.ll +++ b/llvm/test/CodeGen/DirectX/ShaderFlags/disable-opt-lib.ll @@ -21,24 +21,29 @@ target triple = "dxilv1.3-pc-shadermodel6.3-library" -; Function Attrs: convergent mustprogress noinline norecurse nounwind optnone +; Function Attrs: convergent mustprogress noinline norecurse nounwind define internal void @_Z4mainv() #0 { entry: ret void } -; Function Attrs: convergent noinline norecurse optnone +; Function Attrs: convergent noinline norecurse define void @main() #1 { entry: call void @_Z4mainv() ret void } -; Function Attrs: convergent mustprogress noinline norecurse nounwind optnone +; Function Attrs: convergent mustprogress noinline norecurse nounwind define noundef i32 @_Z3foov() #0 { entry: ret i32 0 } -attributes #0 = { convergent mustprogress noinline norecurse nounwind optnone "no-infs-fp-math"="true" "no-nans-fp-math"="true" "no-signed-zeros-fp-math"="true" "no-trapping-math"="true" "stack-protector-buffer-size"="8" } -attributes #1 = { convergent noinline norecurse optnone "hlsl.numthreads"="1,1,1" "hlsl.shader"="compute" "no-infs-fp-math"="true" "no-nans-fp-math"="true" "no-signed-zeros-fp-math"="true" "no-trapping-math"="true" "stack-protector-buffer-size"="8" } +attributes #0 = { convergent mustprogress noinline norecurse nounwind "no-infs-fp-math"="true" "no-nans-fp-math"="true" "no-signed-zeros-fp-math"="true" "no-trapping-math"="true" "stack-protector-buffer-size"="8" } +attributes #1 = { convergent noinline norecurse "hlsl.numthreads"="1,1,1" "hlsl.shader"="compute" "no-infs-fp-math"="true" "no-nans-fp-math"="true" "no-signed-zeros-fp-math"="true" "no-trapping-math"="true" "stack-protector-buffer-size"="8" } + + +!llvm.module.flags = !{!0} + +!0 = !{i32 4, !"dx.disable_optimizations", i32 1} diff --git a/llvm/test/CodeGen/DirectX/ShaderFlags/lib-entry-attr-error.ll b/llvm/test/CodeGen/DirectX/ShaderFlags/lib-entry-attr-error.ll deleted file mode 100644 index ce35c03d6bcfa..0000000000000 --- a/llvm/test/CodeGen/DirectX/ShaderFlags/lib-entry-attr-error.ll +++ /dev/null @@ -1,30 +0,0 @@ -; RUN: not opt -S --passes="print-dx-shader-flags" 2>&1 %s | FileCheck %s - -target triple = "dxilv1.3-pc-shadermodel6.3-library" - -; All entry functions of a library shader need to either have optnone -; or not have the attribute -; CHECK: error: -; CHECK-SAME: in function entry_two -; CHECK-SAME: Inconsistent optnone attribute -; Function Attrs: convergent noinline norecurse optnone -define void @entry_one() #0 { -entry: - ret void -} - -; Function Attrs: convergent noinline norecurse -define void @entry_two() #1 { -entry: - ret void -} - -attributes #0 = { convergent noinline norecurse optnone "hlsl.numthreads"="1,1,1" "hlsl.shader"="compute" "no-infs-fp-math"="true" "no-nans-fp-math"="true" "no-signed-zeros-fp-math"="true" "no-trapping-math"="true" "stack-protector-buffer-size"="8" } -attributes #1 = { convergent noinline norecurse "hlsl.numthreads"="1,1,1" "hlsl.shader"="compute" "no-infs-fp-math"="true" "no-nans-fp-math"="true" "no-signed-zeros-fp-math"="true" "no-trapping-math"="true" "stack-protector-buffer-size"="8" } - -!llvm.module.flags = !{!0, !1} -!dx.valver = !{!2} - -!0 = !{i32 1, !"wchar_size", i32 4} -!1 = !{i32 4, !"dx.disable_optimizations", i32 1} -!2 = !{i32 1, i32 8} `````````` </details> https://github.com/llvm/llvm-project/pull/180069 _______________________________________________ cfe-commits mailing list [email protected] https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
