Skip to content

[Clang][OpenCL][AMDGPU] OpenCL Kernel stubs should be assigned alwaysinline attribute #137769

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
merged 1 commit into from
May 7, 2025
+35 −35
Merged

[Clang][OpenCL][AMDGPU] OpenCL Kernel stubs should be assigned alwaysinline attribute #137769

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
16 changes: 16 additions & 0 deletions clang/lib/CodeGen/CodeGenModule.cpp
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -6174,6 +6174,22 @@ void CodeGenModule::EmitGlobalFunctionDefinition(GlobalDecl GD,
CodeGenFunction(*this).GenerateCode(GD, Fn, FI);

setNonAliasAttributes(GD, Fn);

bool ShouldAddOptNone = !CodeGenOpts.DisableO0ImplyOptNone &&
(CodeGenOpts.OptimizationLevel == 0) &&
!D->hasAttr<MinSizeAttr>();
Comment on lines +6178 to +6180
Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Really hate optnone being used to implement -O0

Copy link
Contributor Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

I am following the logic for ShouldAddOptNone used later in SetLLVMFunctionAttributesForDefinition(D, Fn); here https://github.com/llvm/llvm-project/blob/main/clang/lib/CodeGen/CodeGenModule.cpp#L2552 to decide if optnone and noinline attribute should be added.

If we don't do the same check we end up adding alwaysinline to the stub in cases where ShouldAddOptNone would be true. This prevents adding of optnone and noinline later in https://github.com/llvm/llvm-project/blob/main/clang/lib/CodeGen/CodeGenModule.cpp#L2552

Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Yes, I know why you did this. I'm objecting to the fact that you need to


if (D->hasAttr<OpenCLKernelAttr>()) {
if (GD.getKernelReferenceKind() == KernelReferenceKind::Stub &&
!D->hasAttr<NoInlineAttr>() &&
!Fn->hasFnAttribute(llvm::Attribute::NoInline) &&
!D->hasAttr<OptimizeNoneAttr>() &&
!Fn->hasFnAttribute(llvm::Attribute::OptimizeNone) &&
!ShouldAddOptNone) {
Fn->addFnAttr(llvm::Attribute::AlwaysInline);
}
}

SetLLVMFunctionAttributesForDefinition(D, Fn);

if (const ConstructorAttr *CA = D->getAttr<ConstructorAttr>())
Expand Down
6 changes: 3 additions & 3 deletions clang/test/CodeGenOpenCL/amdgpu-enqueue-kernel.cl
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -492,7 +492,7 @@ kernel void test_target_features_kernel(global int *i) {
// GFX900-NEXT: ret void
//
//
// GFX900: Function Attrs: convergent norecurse nounwind
// GFX900: Function Attrs: alwaysinline convergent norecurse nounwind
// GFX900-LABEL: define dso_local void @__clang_ocl_kern_imp_test(
// GFX900-SAME: ptr addrspace(1) noundef align 1 [[A:%.*]], i8 noundef signext [[B:%.*]], ptr addrspace(1) noundef align 8 [[C:%.*]], i64 noundef [[D:%.*]]) #[[ATTR3:[0-9]+]] !kernel_arg_addr_space [[META10]] !kernel_arg_access_qual [[META11]] !kernel_arg_type [[META12]] !kernel_arg_base_type [[META12]] !kernel_arg_type_qual [[META13]] {
// GFX900-NEXT: [[ENTRY:.*:]]
Expand Down Expand Up @@ -640,7 +640,7 @@ kernel void test_target_features_kernel(global int *i) {
// GFX900-NEXT: ret void
//
//
// GFX900: Function Attrs: convergent norecurse nounwind
// GFX900: Function Attrs: alwaysinline convergent norecurse nounwind
// GFX900-LABEL: define dso_local void @__clang_ocl_kern_imp_test_target_features_kernel(
// GFX900-SAME: ptr addrspace(1) noundef align 4 [[I:%.*]]) #[[ATTR3]] !kernel_arg_addr_space [[META22]] !kernel_arg_access_qual [[META23]] !kernel_arg_type [[META24]] !kernel_arg_base_type [[META24]] !kernel_arg_type_qual [[META25]] {
// GFX900-NEXT: [[ENTRY:.*:]]
Expand Down Expand Up @@ -832,7 +832,7 @@ kernel void test_target_features_kernel(global int *i) {
// GFX900: attributes #[[ATTR0:[0-9]+]] = { "objc_arc_inert" }
// GFX900: attributes #[[ATTR1]] = { convergent norecurse nounwind "denormal-fp-math-f32"="preserve-sign,preserve-sign" "no-trapping-math"="true" "stack-protector-buffer-size"="8" "target-cpu"="gfx900" "target-features"="+16-bit-insts,+ci-insts,+dpp,+gfx8-insts,+gfx9-insts,+s-memrealtime,+s-memtime-inst,+wavefrontsize64,-sram-ecc" }
// GFX900: attributes #[[ATTR2]] = { convergent norecurse nounwind "amdgpu-flat-work-group-size"="1,256" "denormal-fp-math-f32"="preserve-sign,preserve-sign" "no-trapping-math"="true" "stack-protector-buffer-size"="8" "target-cpu"="gfx900" "target-features"="+16-bit-insts,+ci-insts,+dpp,+gfx8-insts,+gfx9-insts,+s-memrealtime,+s-memtime-inst,+wavefrontsize64,-sram-ecc" "uniform-work-group-size"="false" }
// GFX900: attributes #[[ATTR3]] = { convergent norecurse nounwind "amdgpu-flat-work-group-size"="1,256" "denormal-fp-math-f32"="preserve-sign,preserve-sign" "no-trapping-math"="true" "stack-protector-buffer-size"="8" "target-cpu"="gfx900" "target-features"="+16-bit-insts,+ci-insts,+dpp,+gfx8-insts,+gfx9-insts,+s-memrealtime,+s-memtime-inst,+wavefrontsize64,-sram-ecc" }
// GFX900: attributes #[[ATTR3]] = { alwaysinline convergent norecurse nounwind "amdgpu-flat-work-group-size"="1,256" "denormal-fp-math-f32"="preserve-sign,preserve-sign" "no-trapping-math"="true" "stack-protector-buffer-size"="8" "target-cpu"="gfx900" "target-features"="+16-bit-insts,+ci-insts,+dpp,+gfx8-insts,+gfx9-insts,+s-memrealtime,+s-memtime-inst,+wavefrontsize64,-sram-ecc" }
// GFX900: attributes #[[ATTR4:[0-9]+]] = { nocallback nofree nosync nounwind willreturn memory(argmem: readwrite) }
// GFX900: attributes #[[ATTR5:[0-9]+]] = { nocallback nofree nounwind willreturn memory(argmem: readwrite) }
// GFX900: attributes #[[ATTR6]] = { convergent nounwind "denormal-fp-math-f32"="preserve-sign,preserve-sign" "no-trapping-math"="true" "stack-protector-buffer-size"="8" "target-cpu"="gfx900" "target-features"="+16-bit-insts,+ci-insts,+dpp,+gfx8-insts,+gfx9-insts,+s-memrealtime,+s-memtime-inst,+wavefrontsize64,-sram-ecc" }
Expand Down
4 changes: 0 additions & 4 deletions clang/test/CodeGenOpenCL/cl-uniform-wg-size.cl
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -5,7 +5,6 @@

kernel void ker() {};
// CHECK: define{{.*}}@ker() #[[ATTR0:[0-9]+]]
// CHECK: call void @__clang_ocl_kern_imp_ker() #[[ATTR2:[0-9]+]]

// CHECK: define{{.*}}@__clang_ocl_kern_imp_ker() #[[ATTR1:[0-9]+]]

Expand All @@ -18,6 +17,3 @@ void foo() {};

// CHECK: attributes #[[ATTR1]]
// CHECK-NOT: uniform-work-group-size

// CHECK: attributes #[[ATTR2]]
// CHECK-NOT: uniform-work-group-size
24 changes: 9 additions & 15 deletions clang/test/CodeGenOpenCL/cl20-device-side-enqueue.cl
View file
Open in desktop
Original file line number Diff line number Diff line change
@@ -1,12 +1,12 @@
// RUN: %clang_cc1 -no-enable-noundef-analysis %s -cl-std=CL2.0 -ffake-address-space-map -O0 -emit-llvm -o - -triple "spir-unknown-unknown" | FileCheck %s --check-prefixes=COMMON,B32,SPIR,TRIPLESPIR
// RUN: %clang_cc1 -no-enable-noundef-analysis %s -cl-std=CL2.0 -ffake-address-space-map -O0 -emit-llvm -o - -triple "spir64-unknown-unknown" | FileCheck %s --check-prefixes=COMMON,B64,SPIR,TRIPLESPIR
// RUN: %clang_cc1 -no-enable-noundef-analysis %s -cl-std=CL2.0 -ffake-address-space-map -O1 -emit-llvm -o - -triple "spir64-unknown-unknown" | FileCheck %s --check-prefixes=CHECK-LIFETIMES,TRIPLESPIR
// RUN: %clang_cc1 -no-enable-noundef-analysis %s -cl-std=CL3.0 -ffake-address-space-map -O0 -emit-llvm -o - -triple "spir-unknown-unknown" | FileCheck %s --check-prefixes=COMMON,B32,SPIR,TRIPLESPIR
// RUN: %clang_cc1 -no-enable-noundef-analysis %s -cl-std=CL3.0 -ffake-address-space-map -O0 -emit-llvm -o - -triple "spir64-unknown-unknown" | FileCheck %s --check-prefixes=COMMON,B64,SPIR,TRIPLESPIR
// RUN: %clang_cc1 -no-enable-noundef-analysis %s -cl-std=CL3.0 -ffake-address-space-map -O1 -emit-llvm -o - -triple "spir64-unknown-unknown" | FileCheck %s --check-prefixes=CHECK-LIFETIMES,TRIPLESPIR
// RUN: %clang_cc1 -no-enable-noundef-analysis %s -cl-std=CL2.0 -ffake-address-space-map -O0 -emit-llvm -o - -triple "x86_64-unknown-linux-gnu" | FileCheck %s --check-prefixes=COMMON,B64,X86,TRIPLEX86
// RUN: %clang_cc1 -no-enable-noundef-analysis %s -cl-std=CL3.0 -ffake-address-space-map -O0 -emit-llvm -o - -triple "x86_64-unknown-linux-gnu" | FileCheck %s --check-prefixes=COMMON,B64,X86,TRIPLEX86
// RUN: %clang_cc1 -no-enable-noundef-analysis %s -cl-std=CL3.0 -ffake-address-space-map -O1 -emit-llvm -o - -triple "x86_64-unknown-linux-gnu" | FileCheck %s --check-prefixes=CHECK-LIFETIMES,TRIPLEX86
// RUN: %clang_cc1 -no-enable-noundef-analysis %s -cl-std=CL2.0 -ffake-address-space-map -O0 -emit-llvm -o - -triple "spir-unknown-unknown" | FileCheck %s --check-prefixes=COMMON,B32,SPIR
// RUN: %clang_cc1 -no-enable-noundef-analysis %s -cl-std=CL2.0 -ffake-address-space-map -O0 -emit-llvm -o - -triple "spir64-unknown-unknown" | FileCheck %s --check-prefixes=COMMON,B64,SPIR
// RUN: %clang_cc1 -no-enable-noundef-analysis %s -cl-std=CL2.0 -ffake-address-space-map -O1 -emit-llvm -o - -triple "spir64-unknown-unknown" | FileCheck %s --check-prefix=CHECK-LIFETIMES
// RUN: %clang_cc1 -no-enable-noundef-analysis %s -cl-std=CL3.0 -ffake-address-space-map -O0 -emit-llvm -o - -triple "spir-unknown-unknown" | FileCheck %s --check-prefixes=COMMON,B32,SPIR
// RUN: %clang_cc1 -no-enable-noundef-analysis %s -cl-std=CL3.0 -ffake-address-space-map -O0 -emit-llvm -o - -triple "spir64-unknown-unknown" | FileCheck %s --check-prefixes=COMMON,B64,SPIR
// RUN: %clang_cc1 -no-enable-noundef-analysis %s -cl-std=CL3.0 -ffake-address-space-map -O1 -emit-llvm -o - -triple "spir64-unknown-unknown" | FileCheck %s --check-prefix=CHECK-LIFETIMES
// RUN: %clang_cc1 -no-enable-noundef-analysis %s -cl-std=CL2.0 -ffake-address-space-map -O0 -emit-llvm -o - -triple "x86_64-unknown-linux-gnu" | FileCheck %s --check-prefixes=COMMON,B64,X86
// RUN: %clang_cc1 -no-enable-noundef-analysis %s -cl-std=CL3.0 -ffake-address-space-map -O0 -emit-llvm -o - -triple "x86_64-unknown-linux-gnu" | FileCheck %s --check-prefixes=COMMON,B64,X86
// RUN: %clang_cc1 -no-enable-noundef-analysis %s -cl-std=CL3.0 -ffake-address-space-map -O1 -emit-llvm -o - -triple "x86_64-unknown-linux-gnu" | FileCheck %s --check-prefix=CHECK-LIFETIMES

#pragma OPENCL EXTENSION cl_khr_subgroups : enable

Expand Down Expand Up @@ -39,12 +39,6 @@ void callee(int id, __global int *out) {
out[id] = id;
}

// TRIPLESPIR: define{{.*}} void @device_side_enqueue(ptr addrspace(1) align 4 %{{.*}}, ptr addrspace(1) align 4 %b, i32 %i)
// TRIPLESPIR: call spir_func void @__clang_ocl_kern_imp_device_side_enqueue({{.*}})

// TRIPLEX86: define{{.*}} void @device_side_enqueue(ptr addrspace(1) align 4 %{{.*}}, ptr addrspace(1) align 4 %b, i32 %i)
// TRIPLEX86: call void @__clang_ocl_kern_imp_device_side_enqueue({{.*}})

// COMMON-LABEL: define{{.*}} void @__clang_ocl_kern_imp_device_side_enqueue(ptr addrspace(1) align 4 %{{.*}}, ptr addrspace(1) align 4 %b, i32 %i)
kernel void device_side_enqueue(global int *a, global int *b, int i) {
// SPIR: %default_queue = alloca target("spirv.Queue")
Expand Down
9 changes: 5 additions & 4 deletions clang/test/CodeGenOpenCL/convergent.cl
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -127,7 +127,7 @@ void test_not_unroll() {
// CHECK: declare spir_func void @nodupfun(){{[^#]*}} #[[attr3:[0-9]+]]

// CHECK-LABEL: @assume_convergent_asm
// CHECK: tail call void asm sideeffect "s_barrier", ""() #5
// CHECK: tail call void asm sideeffect "s_barrier", ""() #6
kernel void assume_convergent_asm()
{
__asm__ volatile("s_barrier");
Expand All @@ -138,6 +138,7 @@ kernel void assume_convergent_asm()
// CHECK: attributes #2 = { {{[^}]*}}convergent{{[^}]*}} }
// CHECK: attributes #3 = { {{[^}]*}}convergent noduplicate{{[^}]*}} }
// CHECK: attributes #4 = { {{[^}]*}}convergent{{[^}]*}} }
// CHECK: attributes #5 = { {{[^}]*}}convergent{{[^}]*}} }
// CHECK: attributes #6 = { {{[^}]*}}nounwind{{[^}]*}} }
// CHECK: attributes #7 = { {{[^}]*}}convergent noduplicate nounwind{{[^}]*}} }
// CHECK: attributes #5 = { {{[^}]*}}alwaysinline convergent{{[^}]*}} }
// CHECK: attributes #6 = { {{[^}]*}}convergent{{[^}]*}} }
// CHECK: attributes #7 = { {{[^}]*}}nounwind{{[^}]*}} }
// CHECK: attributes #8 = { {{[^}]*}}convergent noduplicate nounwind{{[^}]*}} }
11 changes: 2 additions & 9 deletions clang/test/CodeGenOpenCL/enqueue-kernel-non-entry-block.cl
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -9,15 +9,8 @@
typedef struct {int a;} ndrange_t;

kernel void test(int i) {

// AMDGPU-LABEL: define {{.*}} amdgpu_kernel void @test
// AMDGPU-LABEL: call void @__clang_ocl_kern_imp_test(i32 noundef %0)

// SPIR-LABEL: define {{.*}} spir_kernel void @test
// SPIR-LABEL: call spir_func void @__clang_ocl_kern_imp_test(i32 noundef %0)

// AMDGPU-LABEL: define {{.*}} void @__clang_ocl_kern_imp_test
// SPIR-LABEL: define {{.*}} spir_func void @__clang_ocl_kern_imp_test

// COMMON-LABEL: entry:
// AMDGPU: %block_sizes = alloca [1 x i64]
Expand All @@ -44,5 +37,5 @@ kernel void test(int i) {

// CHECK-DEBUG: ![[TESTFILE:[0-9]+]] = !DIFile(filename: "<stdin>"
// CHECK-DEBUG: ![[TESTSCOPE:[0-9]+]] = distinct !DISubprogram(name: "test", linkageName: "__clang_ocl_kern_imp_test", {{.*}} file: ![[TESTFILE]]
// CHECK-DEBUG: ![[IFSCOPE:[0-9]+]] = distinct !DILexicalBlock(scope: ![[TESTSCOPE]], file: ![[TESTFILE]], line: 33)
// CHECK-DEBUG: ![[TEMPLOCATION]] = !DILocation(line: 34, scope: ![[IFSCOPE]])
// CHECK-DEBUG: ![[IFSCOPE:[0-9]+]] = distinct !DILexicalBlock(scope: ![[TESTSCOPE]], file: ![[TESTFILE]], line: 26)
// CHECK-DEBUG: ![[TEMPLOCATION]] = !DILocation(line: 27, scope: ![[IFSCOPE]])