[clang] [clang-tools-extra] [CudaSPIRV] Allow using integral non-type template parameters as attribute args (PR #131546)
Alexander Shaposhnikov via cfe-commits
cfe-commits at lists.llvm.org
Mon Mar 17 14:26:33 PDT 2025
https://github.com/alexander-shaposhnikov updated https://github.com/llvm/llvm-project/pull/131546
>From 6f00315409719797e353854a7e73f6eb75674dae Mon Sep 17 00:00:00 2001
From: Alexander Shaposhnikov <ashaposhnikov at google.com>
Date: Mon, 17 Mar 2025 00:39:24 +0000
Subject: [PATCH] [CudaSPIRV] Allow using integral non-type template parameters
as reqd_work_group_size arguments
---
.../altera/SingleWorkItemBarrierCheck.cpp | 8 +-
clang/include/clang/Basic/Attr.td | 7 +-
clang/lib/CodeGen/CodeGenFunction.cpp | 18 +++--
clang/lib/CodeGen/Targets/AMDGPU.cpp | 10 ++-
clang/lib/CodeGen/Targets/TCE.cpp | 27 +++----
clang/lib/Sema/SemaDeclAttr.cpp | 79 ++++++++++++++++---
.../lib/Sema/SemaTemplateInstantiateDecl.cpp | 32 ++++++++
clang/test/CodeGenCUDASPIRV/spirv-attrs.cu | 7 ++
clang/test/SemaCUDA/spirv-attrs-diag.cu | 34 ++++++++
clang/test/SemaCUDA/spirv-attrs.cu | 16 +++-
10 files changed, 197 insertions(+), 41 deletions(-)
create mode 100644 clang/test/SemaCUDA/spirv-attrs-diag.cu
diff --git a/clang-tools-extra/clang-tidy/altera/SingleWorkItemBarrierCheck.cpp b/clang-tools-extra/clang-tidy/altera/SingleWorkItemBarrierCheck.cpp
index df21c425ea956..c5da66a1f28b6 100644
--- a/clang-tools-extra/clang-tidy/altera/SingleWorkItemBarrierCheck.cpp
+++ b/clang-tools-extra/clang-tidy/altera/SingleWorkItemBarrierCheck.cpp
@@ -54,8 +54,12 @@ void SingleWorkItemBarrierCheck::check(const MatchFinder::MatchResult &Result) {
bool IsNDRange = false;
if (MatchedDecl->hasAttr<ReqdWorkGroupSizeAttr>()) {
const auto *Attribute = MatchedDecl->getAttr<ReqdWorkGroupSizeAttr>();
- if (Attribute->getXDim() > 1 || Attribute->getYDim() > 1 ||
- Attribute->getZDim() > 1)
+ auto Eval = [&](Expr *E) {
+ return E->EvaluateKnownConstInt(MatchedDecl->getASTContext())
+ .getExtValue();
+ };
+ if (Eval(Attribute->getXDim()) > 1 || Eval(Attribute->getYDim()) > 1 ||
+ Eval(Attribute->getZDim()) > 1)
IsNDRange = true;
}
if (IsNDRange) // No warning if kernel is treated as an NDRange.
diff --git a/clang/include/clang/Basic/Attr.td b/clang/include/clang/Basic/Attr.td
index 4d34346460561..cceb4085d523d 100644
--- a/clang/include/clang/Basic/Attr.td
+++ b/clang/include/clang/Basic/Attr.td
@@ -3044,8 +3044,7 @@ def NoDeref : TypeAttr {
def ReqdWorkGroupSize : InheritableAttr {
// Does not have a [[]] spelling because it is an OpenCL-related attribute.
let Spellings = [GNU<"reqd_work_group_size">];
- let Args = [UnsignedArgument<"XDim">, UnsignedArgument<"YDim">,
- UnsignedArgument<"ZDim">];
+ let Args = [ExprArgument<"XDim">, ExprArgument<"YDim">, ExprArgument<"ZDim">];
let Subjects = SubjectList<[Function], ErrorDiag>;
let Documentation = [Undocumented];
}
@@ -3053,9 +3052,7 @@ def ReqdWorkGroupSize : InheritableAttr {
def WorkGroupSizeHint : InheritableAttr {
// Does not have a [[]] spelling because it is an OpenCL-related attribute.
let Spellings = [GNU<"work_group_size_hint">];
- let Args = [UnsignedArgument<"XDim">,
- UnsignedArgument<"YDim">,
- UnsignedArgument<"ZDim">];
+ let Args = [ExprArgument<"XDim">, ExprArgument<"YDim">, ExprArgument<"ZDim">];
let Subjects = SubjectList<[Function], ErrorDiag>;
let Documentation = [Undocumented];
}
diff --git a/clang/lib/CodeGen/CodeGenFunction.cpp b/clang/lib/CodeGen/CodeGenFunction.cpp
index 447192bc7f60c..de1e04a7982fa 100644
--- a/clang/lib/CodeGen/CodeGenFunction.cpp
+++ b/clang/lib/CodeGen/CodeGenFunction.cpp
@@ -649,18 +649,24 @@ void CodeGenFunction::EmitKernelMetadata(const FunctionDecl *FD,
}
if (const WorkGroupSizeHintAttr *A = FD->getAttr<WorkGroupSizeHintAttr>()) {
+ auto Eval = [&](Expr *E) {
+ return E->EvaluateKnownConstInt(FD->getASTContext()).getExtValue();
+ };
llvm::Metadata *AttrMDArgs[] = {
- llvm::ConstantAsMetadata::get(Builder.getInt32(A->getXDim())),
- llvm::ConstantAsMetadata::get(Builder.getInt32(A->getYDim())),
- llvm::ConstantAsMetadata::get(Builder.getInt32(A->getZDim()))};
+ llvm::ConstantAsMetadata::get(Builder.getInt32(Eval(A->getXDim()))),
+ llvm::ConstantAsMetadata::get(Builder.getInt32(Eval(A->getYDim()))),
+ llvm::ConstantAsMetadata::get(Builder.getInt32(Eval(A->getZDim())))};
Fn->setMetadata("work_group_size_hint", llvm::MDNode::get(Context, AttrMDArgs));
}
if (const ReqdWorkGroupSizeAttr *A = FD->getAttr<ReqdWorkGroupSizeAttr>()) {
+ auto Eval = [&](Expr *E) {
+ return E->EvaluateKnownConstInt(FD->getASTContext()).getExtValue();
+ };
llvm::Metadata *AttrMDArgs[] = {
- llvm::ConstantAsMetadata::get(Builder.getInt32(A->getXDim())),
- llvm::ConstantAsMetadata::get(Builder.getInt32(A->getYDim())),
- llvm::ConstantAsMetadata::get(Builder.getInt32(A->getZDim()))};
+ llvm::ConstantAsMetadata::get(Builder.getInt32(Eval(A->getXDim()))),
+ llvm::ConstantAsMetadata::get(Builder.getInt32(Eval(A->getYDim()))),
+ llvm::ConstantAsMetadata::get(Builder.getInt32(Eval(A->getZDim())))};
Fn->setMetadata("reqd_work_group_size", llvm::MDNode::get(Context, AttrMDArgs));
}
diff --git a/clang/lib/CodeGen/Targets/AMDGPU.cpp b/clang/lib/CodeGen/Targets/AMDGPU.cpp
index 7d84310ba0bf5..e3c8243cbb0b5 100644
--- a/clang/lib/CodeGen/Targets/AMDGPU.cpp
+++ b/clang/lib/CodeGen/Targets/AMDGPU.cpp
@@ -753,12 +753,16 @@ void CodeGenModule::handleAMDGPUFlatWorkGroupSizeAttr(
int32_t *MaxThreadsVal) {
unsigned Min = 0;
unsigned Max = 0;
+ auto Eval = [&](Expr *E) {
+ return E->EvaluateKnownConstInt(getContext()).getExtValue();
+ };
if (FlatWGS) {
- Min = FlatWGS->getMin()->EvaluateKnownConstInt(getContext()).getExtValue();
- Max = FlatWGS->getMax()->EvaluateKnownConstInt(getContext()).getExtValue();
+ Min = Eval(FlatWGS->getMin());
+ Max = Eval(FlatWGS->getMax());
}
if (ReqdWGS && Min == 0 && Max == 0)
- Min = Max = ReqdWGS->getXDim() * ReqdWGS->getYDim() * ReqdWGS->getZDim();
+ Min = Max = Eval(ReqdWGS->getXDim()) * Eval(ReqdWGS->getYDim()) *
+ Eval(ReqdWGS->getZDim());
if (Min != 0) {
assert(Min <= Max && "Min must be less than or equal Max");
diff --git a/clang/lib/CodeGen/Targets/TCE.cpp b/clang/lib/CodeGen/Targets/TCE.cpp
index d7178b4b8a949..f3685ccd9825a 100644
--- a/clang/lib/CodeGen/Targets/TCE.cpp
+++ b/clang/lib/CodeGen/Targets/TCE.cpp
@@ -50,24 +50,21 @@ void TCETargetCodeGenInfo::setTargetAttributes(
M.getModule().getOrInsertNamedMetadata(
"opencl.kernel_wg_size_info");
- SmallVector<llvm::Metadata *, 5> Operands;
- Operands.push_back(llvm::ConstantAsMetadata::get(F));
-
- Operands.push_back(
+ auto Eval = [&](Expr *E) {
+ return E->EvaluateKnownConstInt(FD->getASTContext());
+ };
+ SmallVector<llvm::Metadata *, 5> Operands{
+ llvm::ConstantAsMetadata::get(F),
llvm::ConstantAsMetadata::get(llvm::Constant::getIntegerValue(
- M.Int32Ty, llvm::APInt(32, Attr->getXDim()))));
- Operands.push_back(
+ M.Int32Ty, Eval(Attr->getXDim()))),
llvm::ConstantAsMetadata::get(llvm::Constant::getIntegerValue(
- M.Int32Ty, llvm::APInt(32, Attr->getYDim()))));
- Operands.push_back(
+ M.Int32Ty, Eval(Attr->getYDim()))),
llvm::ConstantAsMetadata::get(llvm::Constant::getIntegerValue(
- M.Int32Ty, llvm::APInt(32, Attr->getZDim()))));
-
- // Add a boolean constant operand for "required" (true) or "hint"
- // (false) for implementing the work_group_size_hint attr later.
- // Currently always true as the hint is not yet implemented.
- Operands.push_back(
- llvm::ConstantAsMetadata::get(llvm::ConstantInt::getTrue(Context)));
+ M.Int32Ty, Eval(Attr->getZDim()))),
+ // Add a boolean constant operand for "required" (true) or "hint"
+ // (false) for implementing the work_group_size_hint attr later.
+ // Currently always true as the hint is not yet implemented.
+ llvm::ConstantAsMetadata::get(llvm::ConstantInt::getTrue(Context))};
OpenCLMetadata->addOperand(llvm::MDNode::get(Context, Operands));
}
}
diff --git a/clang/lib/Sema/SemaDeclAttr.cpp b/clang/lib/Sema/SemaDeclAttr.cpp
index 769f5316ed934..0a8a3e1c49414 100644
--- a/clang/lib/Sema/SemaDeclAttr.cpp
+++ b/clang/lib/Sema/SemaDeclAttr.cpp
@@ -2914,21 +2914,70 @@ static void handleWeakImportAttr(Sema &S, Decl *D, const ParsedAttr &AL) {
D->addAttr(::new (S.Context) WeakImportAttr(S.Context, AL));
}
+// Checks whether an argument of launch_bounds-like attribute is
+// acceptable, performs implicit conversion to Rvalue, and returns
+// non-nullptr Expr result on success. Otherwise, it returns nullptr
+// and may output an error.
+template <class Attribute>
+static Expr *makeAttributeArgExpr(Sema &S, Expr *E, const Attribute &Attr,
+ const unsigned Idx) {
+ if (S.DiagnoseUnexpandedParameterPack(E))
+ return nullptr;
+
+ // Accept template arguments for now as they depend on something else.
+ // We'll get to check them when they eventually get instantiated.
+ if (E->isValueDependent())
+ return E;
+
+ std::optional<llvm::APSInt> I = llvm::APSInt(64);
+ if (!(I = E->getIntegerConstantExpr(S.Context))) {
+ S.Diag(E->getExprLoc(), diag::err_attribute_argument_n_type)
+ << &Attr << Idx << AANT_ArgumentIntegerConstant << E->getSourceRange();
+ return nullptr;
+ }
+ // Make sure we can fit it in 32 bits.
+ if (!I->isIntN(32)) {
+ S.Diag(E->getExprLoc(), diag::err_ice_too_large)
+ << toString(*I, 10, false) << 32 << /* Unsigned */ 1;
+ return nullptr;
+ }
+ if (*I < 0)
+ S.Diag(E->getExprLoc(), diag::err_attribute_requires_positive_integer)
+ << &Attr << /*non-negative*/ 1 << E->getSourceRange();
+
+ // We may need to perform implicit conversion of the argument.
+ InitializedEntity Entity = InitializedEntity::InitializeParameter(
+ S.Context, S.Context.getConstType(S.Context.IntTy), /*consume*/ false);
+ ExprResult ValArg = S.PerformCopyInitialization(Entity, SourceLocation(), E);
+ assert(!ValArg.isInvalid() &&
+ "Unexpected PerformCopyInitialization() failure.");
+
+ return ValArg.getAs<Expr>();
+}
+
// Handles reqd_work_group_size and work_group_size_hint.
template <typename WorkGroupAttr>
static void handleWorkGroupSize(Sema &S, Decl *D, const ParsedAttr &AL) {
- uint32_t WGSize[3];
+ Expr *WGSize[3];
for (unsigned i = 0; i < 3; ++i) {
- const Expr *E = AL.getArgAsExpr(i);
- if (!S.checkUInt32Argument(AL, E, WGSize[i], i,
- /*StrictlyUnsigned=*/true))
+ if (Expr *E = makeAttributeArgExpr(S, AL.getArgAsExpr(i), AL, i))
+ WGSize[i] = E;
+ else
return;
}
- if (!llvm::all_of(WGSize, [](uint32_t Size) { return Size == 0; })) {
+ auto IsZero = [&](Expr *E) {
+ if (E->isValueDependent())
+ return false;
+ std::optional<llvm::APSInt> I = E->getIntegerConstantExpr(S.Context);
+ assert(I && "Non-integer constant expr");
+ return I->isZero();
+ };
+
+ if (!llvm::all_of(WGSize, IsZero)) {
for (unsigned i = 0; i < 3; ++i) {
const Expr *E = AL.getArgAsExpr(i);
- if (WGSize[i] == 0) {
+ if (IsZero(WGSize[i])) {
S.Diag(AL.getLoc(), diag::err_attribute_argument_is_zero)
<< AL << E->getSourceRange();
return;
@@ -2936,10 +2985,22 @@ static void handleWorkGroupSize(Sema &S, Decl *D, const ParsedAttr &AL) {
}
}
+ auto Equal = [&](Expr *LHS, Expr *RHS) {
+ if (LHS->isValueDependent() || RHS->isValueDependent())
+ return true;
+ std::optional<llvm::APSInt> L = LHS->getIntegerConstantExpr(S.Context);
+ assert(L && "Non-integer constant expr");
+ std::optional<llvm::APSInt> R = RHS->getIntegerConstantExpr(S.Context);
+ assert(L && "Non-integer constant expr");
+ return L == R;
+ };
+
WorkGroupAttr *Existing = D->getAttr<WorkGroupAttr>();
- if (Existing && !(Existing->getXDim() == WGSize[0] &&
- Existing->getYDim() == WGSize[1] &&
- Existing->getZDim() == WGSize[2]))
+ if (Existing &&
+ !llvm::equal(std::initializer_list<Expr *>{Existing->getXDim(),
+ Existing->getYDim(),
+ Existing->getZDim()},
+ WGSize, Equal))
S.Diag(AL.getLoc(), diag::warn_duplicate_attribute) << AL;
D->addAttr(::new (S.Context)
diff --git a/clang/lib/Sema/SemaTemplateInstantiateDecl.cpp b/clang/lib/Sema/SemaTemplateInstantiateDecl.cpp
index 170c0b0d39f86..d5330430a3b73 100644
--- a/clang/lib/Sema/SemaTemplateInstantiateDecl.cpp
+++ b/clang/lib/Sema/SemaTemplateInstantiateDecl.cpp
@@ -572,6 +572,32 @@ static void instantiateDependentAMDGPUFlatWorkGroupSizeAttr(
S.AMDGPU().addAMDGPUFlatWorkGroupSizeAttr(New, Attr, MinExpr, MaxExpr);
}
+static void instantiateDependentReqdWorkGroupSizeAttr(
+ Sema &S, const MultiLevelTemplateArgumentList &TemplateArgs,
+ const ReqdWorkGroupSizeAttr &Attr, Decl *New) {
+ // Both min and max expression are constant expressions.
+ EnterExpressionEvaluationContext Unevaluated(
+ S, Sema::ExpressionEvaluationContext::ConstantEvaluated);
+
+ ExprResult Result = S.SubstExpr(Attr.getXDim(), TemplateArgs);
+ if (Result.isInvalid())
+ return;
+ Expr *X = Result.getAs<Expr>();
+
+ Result = S.SubstExpr(Attr.getYDim(), TemplateArgs);
+ if (Result.isInvalid())
+ return;
+ Expr *Y = Result.getAs<Expr>();
+
+ Result = S.SubstExpr(Attr.getZDim(), TemplateArgs);
+ if (Result.isInvalid())
+ return;
+ Expr *Z = Result.getAs<Expr>();
+
+ ASTContext &Context = S.getASTContext();
+ New->addAttr(::new (Context) ReqdWorkGroupSizeAttr(Context, Attr, X, Y, Z));
+}
+
ExplicitSpecifier Sema::instantiateExplicitSpecifier(
const MultiLevelTemplateArgumentList &TemplateArgs, ExplicitSpecifier ES) {
if (!ES.getExpr())
@@ -812,6 +838,12 @@ void Sema::InstantiateAttrs(const MultiLevelTemplateArgumentList &TemplateArgs,
continue;
}
+ if (const auto *ReqdWorkGroupSize =
+ dyn_cast<ReqdWorkGroupSizeAttr>(TmplAttr)) {
+ instantiateDependentReqdWorkGroupSizeAttr(*this, TemplateArgs,
+ *ReqdWorkGroupSize, New);
+ }
+
if (const auto *AMDGPUFlatWorkGroupSize =
dyn_cast<AMDGPUFlatWorkGroupSizeAttr>(TmplAttr)) {
instantiateDependentAMDGPUFlatWorkGroupSizeAttr(
diff --git a/clang/test/CodeGenCUDASPIRV/spirv-attrs.cu b/clang/test/CodeGenCUDASPIRV/spirv-attrs.cu
index 727e0e233329c..9dfe09e5d9a35 100644
--- a/clang/test/CodeGenCUDASPIRV/spirv-attrs.cu
+++ b/clang/test/CodeGenCUDASPIRV/spirv-attrs.cu
@@ -18,12 +18,18 @@ __global__ void vec_type_hint_int() {}
__attribute__((intel_reqd_sub_group_size(64)))
__global__ void intel_reqd_sub_group_size_64() {}
+template <unsigned a, unsigned b, unsigned c>
+__attribute__((reqd_work_group_size(a, b, c)))
+__global__ void reqd_work_group_size_a_b_c() {}
+
+template __global__ void reqd_work_group_size_a_b_c<256,1,1>(void);
// CHECK: define spir_kernel void @_Z26reqd_work_group_size_0_0_0v() #[[ATTR:[0-9]+]] !reqd_work_group_size ![[WG_SIZE_ZEROS:[0-9]+]]
// CHECK: define spir_kernel void @_Z28reqd_work_group_size_128_1_1v() #[[ATTR:[0-9]+]] !reqd_work_group_size ![[WG_SIZE:[0-9]+]]
// CHECK: define spir_kernel void @_Z26work_group_size_hint_2_2_2v() #[[ATTR]] !work_group_size_hint ![[WG_HINT:[0-9]+]]
// CHECK: define spir_kernel void @_Z17vec_type_hint_intv() #[[ATTR]] !vec_type_hint ![[VEC_HINT:[0-9]+]]
// CHECK: define spir_kernel void @_Z28intel_reqd_sub_group_size_64v() #[[ATTR]] !intel_reqd_sub_group_size ![[SUB_GRP:[0-9]+]]
+// CHECK: define spir_kernel void @_Z26reqd_work_group_size_a_b_cILj256ELj1ELj1EEvv() #[[ATTR]] comdat !reqd_work_group_size ![[WG_SIZE_TMPL:[0-9]+]]
// CHECK: attributes #[[ATTR]] = { {{.*}} }
@@ -32,3 +38,4 @@ __global__ void intel_reqd_sub_group_size_64() {}
// CHECK: ![[WG_HINT]] = !{i32 2, i32 2, i32 2}
// CHECK: ![[VEC_HINT]] = !{i32 undef, i32 1}
// CHECK: ![[SUB_GRP]] = !{i32 64}
+// CHECK: ![[WG_SIZE_TMPL]] = !{i32 256, i32 1, i32 1}
diff --git a/clang/test/SemaCUDA/spirv-attrs-diag.cu b/clang/test/SemaCUDA/spirv-attrs-diag.cu
new file mode 100644
index 0000000000000..ddd41527172d6
--- /dev/null
+++ b/clang/test/SemaCUDA/spirv-attrs-diag.cu
@@ -0,0 +1,34 @@
+// RUN: %clang_cc1 -triple spirv64 -aux-triple x86_64-unknown-linux-gnu \
+// RUN: -fcuda-is-device -verify -fsyntax-only %s
+
+#define __global__ __attribute__((global))
+
+__attribute__((reqd_work_group_size(0x100000000, 1, 1))) // expected-error {{integer constant expression evaluates to value 4294967296 that cannot be represented in a 32-bit unsigned integer type}}
+__global__ void TestTooBigArg1(void);
+
+__attribute__((work_group_size_hint(0x100000000, 1, 1))) // expected-error {{integer constant expression evaluates to value 4294967296 that cannot be represented in a 32-bit unsigned integer type}}
+__global__ void TestTooBigArg2(void);
+
+template <int... Args>
+__attribute__((reqd_work_group_size(Args))) // expected-error {{expression contains unexpanded parameter pack 'Args'}}
+__global__ void TestTemplateVariadicArgs1(void) {}
+
+template <int... Args>
+__attribute__((work_group_size_hint(Args))) // expected-error {{expression contains unexpanded parameter pack 'Args'}}
+__global__ void TestTemplateVariadicArgs2(void) {}
+
+template <class a> // expected-note {{declared here}}
+__attribute__((reqd_work_group_size(a, 1, 1))) // expected-error {{'a' does not refer to a value}}
+__global__ void TestTemplateArgClass1(void) {}
+
+template <class a> // expected-note {{declared here}}
+__attribute__((work_group_size_hint(a, 1, 1))) // expected-error {{'a' does not refer to a value}}
+__global__ void TestTemplateArgClass2(void) {}
+
+constexpr int A = 512;
+
+__attribute__((reqd_work_group_size(A, A, A)))
+__global__ void TestConstIntArg1(void) {}
+
+__attribute__((work_group_size_hint(A, A, A)))
+__global__ void TestConstIntArg2(void) {}
diff --git a/clang/test/SemaCUDA/spirv-attrs.cu b/clang/test/SemaCUDA/spirv-attrs.cu
index 6539421423ee1..3aabe85a9b803 100644
--- a/clang/test/SemaCUDA/spirv-attrs.cu
+++ b/clang/test/SemaCUDA/spirv-attrs.cu
@@ -3,14 +3,28 @@
// RUN: %clang_cc1 -triple spirv64 -aux-triple x86_64-unknown-linux-gnu \
// RUN: -fcuda-is-device -verify -fsyntax-only %s
-#include "Inputs/cuda.h"
+#define __global__ __attribute__((global))
__attribute__((reqd_work_group_size(128, 1, 1)))
__global__ void reqd_work_group_size_128_1_1() {}
+template <unsigned a, unsigned b, unsigned c>
+__attribute__((reqd_work_group_size(a, b, c)))
+__global__ void reqd_work_group_size_a_b_c() {}
+
+template <>
+__global__ void reqd_work_group_size_a_b_c<128,1,1>(void);
+
__attribute__((work_group_size_hint(2, 2, 2)))
__global__ void work_group_size_hint_2_2_2() {}
+template <unsigned a, unsigned b, unsigned c>
+__attribute__((work_group_size_hint(a, b, c)))
+__global__ void work_group_size_hint_a_b_c() {}
+
+template <>
+__global__ void work_group_size_hint_a_b_c<128,1,1>(void);
+
__attribute__((vec_type_hint(int)))
__global__ void vec_type_hint_int() {}
More information about the cfe-commits
mailing list