Skip to content

Commit 6043e16

Browse files
steffenlarsensteffenlarsen
committed
Merge from 'main' to 'sycl-web' (#4)
2 parents e67c1ac + 59510c4 commit 6043e16

File tree

592 files changed

+23623
-8130
lines changed

Some content is hidden

Large Commits have some content hidden by default. Use the searchbox below for content that may be hidden.

592 files changed

+23623
-8130
lines changed

‎clang/include/clang/AST/ComparisonCategories.h

Lines changed: 1 addition & 2 deletions
Original file line numberDiff line numberDiff line change
@@ -115,8 +115,7 @@ class ComparisonCategoryInfo {
115115
public:
116116
/// The declaration for the comparison category type from the
117117
/// standard library.
118-
// FIXME: Make this const
119-
CXXRecordDecl *Record = nullptr;
118+
const CXXRecordDecl *Record = nullptr;
120119

121120
/// The Kind of the comparison category type
122121
ComparisonCategoryType Kind;

‎clang/include/clang/AST/Type.h

Lines changed: 1 addition & 6 deletions
Original file line numberDiff line numberDiff line change
@@ -495,12 +495,7 @@ class Qualifiers {
495495
(A == LangAS::Default &&
496496
(B == LangAS::sycl_private || B == LangAS::sycl_local ||
497497
B == LangAS::sycl_global || B == LangAS::sycl_global_device ||
498-
B == LangAS::sycl_global_host)) ||
499-
// In HIP device compilation, any cuda address space is allowed
500-
// to implicitly cast into the default address space.
501-
(A == LangAS::Default &&
502-
(B == LangAS::cuda_constant || B == LangAS::cuda_device ||
503-
B == LangAS::cuda_shared));
498+
B == LangAS::sycl_global_host));
504499
}
505500

506501
/// Returns true if the address space in these qualifiers is equal to or

‎clang/include/clang/Basic/BuiltinsAArch64.def

Lines changed: 3 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -243,6 +243,9 @@ TARGET_HEADER_BUILTIN(_ReadStatusReg, "LLii", "nh", "intrin.h", ALL_MS_LANGUAG
243243
TARGET_HEADER_BUILTIN(_WriteStatusReg, "viLLi", "nh", "intrin.h", ALL_MS_LANGUAGES, "")
244244
TARGET_HEADER_BUILTIN(_AddressOfReturnAddress, "v*", "nh", "intrin.h", ALL_MS_LANGUAGES, "")
245245

246+
TARGET_HEADER_BUILTIN(__mulh, "SLLiSLLiSLLi", "nh", "intrin.h", ALL_MS_LANGUAGES, "")
247+
TARGET_HEADER_BUILTIN(__umulh, "ULLiULLiULLi", "nh", "intrin.h", ALL_MS_LANGUAGES, "")
248+
246249
#undef BUILTIN
247250
#undef LANGBUILTIN
248251
#undef TARGET_HEADER_BUILTIN

‎clang/include/clang/Basic/BuiltinsWebAssembly.def

Lines changed: 4 additions & 4 deletions
Original file line numberDiff line numberDiff line change
@@ -119,10 +119,10 @@ TARGET_BUILTIN(__builtin_wasm_all_true_i16x8, "iV8s", "nc", "simd128")
119119
TARGET_BUILTIN(__builtin_wasm_all_true_i32x4, "iV4i", "nc", "simd128")
120120
TARGET_BUILTIN(__builtin_wasm_all_true_i64x2, "iV2LLi", "nc", "simd128")
121121

122-
TARGET_BUILTIN(__builtin_wasm_bitmask_i8x16, "iV16Sc", "nc", "simd128")
123-
TARGET_BUILTIN(__builtin_wasm_bitmask_i16x8, "iV8s", "nc", "simd128")
124-
TARGET_BUILTIN(__builtin_wasm_bitmask_i32x4, "iV4i", "nc", "simd128")
125-
TARGET_BUILTIN(__builtin_wasm_bitmask_i64x2, "iV2LLi", "nc", "simd128")
122+
TARGET_BUILTIN(__builtin_wasm_bitmask_i8x16, "UiV16Sc", "nc", "simd128")
123+
TARGET_BUILTIN(__builtin_wasm_bitmask_i16x8, "UiV8s", "nc", "simd128")
124+
TARGET_BUILTIN(__builtin_wasm_bitmask_i32x4, "UiV4i", "nc", "simd128")
125+
TARGET_BUILTIN(__builtin_wasm_bitmask_i64x2, "UiV2LLi", "nc", "simd128")
126126

127127
TARGET_BUILTIN(__builtin_wasm_abs_f32x4, "V4fV4f", "nc", "simd128")
128128
TARGET_BUILTIN(__builtin_wasm_abs_f64x2, "V2dV2d", "nc", "simd128")

‎clang/include/clang/Basic/TargetInfo.h

Lines changed: 5 additions & 5 deletions
Original file line numberDiff line numberDiff line change
@@ -210,8 +210,8 @@ class TargetInfo : public virtual TransferrableTargetInfo,
210210
unsigned char RegParmMax, SSERegParmMax;
211211
TargetCXXABI TheCXXABI;
212212
const LangASMap *AddrSpaceMap;
213-
const unsigned *GridValues =
214-
nullptr; // Array of target-specific GPU grid values that must be
213+
const llvm::omp::GV *GridValues =
214+
nullptr; // target-specific GPU grid values that must be
215215
// consistent between host RTL (plugin), device RTL, and clang.
216216

217217
mutable StringRef PlatformName;
@@ -1410,10 +1410,10 @@ class TargetInfo : public virtual TransferrableTargetInfo,
14101410
return LangAS::Default;
14111411
}
14121412

1413-
/// Return a target-specific GPU grid value based on the GVIDX enum \p gv
1414-
unsigned getGridValue(llvm::omp::GVIDX gv) const {
1413+
/// Return a target-specific GPU grid values
1414+
const llvm::omp::GV &getGridValue() const {
14151415
assert(GridValues != nullptr && "GridValues not initialized");
1416-
return GridValues[gv];
1416+
return *GridValues;
14171417
}
14181418

14191419
/// Retrieve the name of the platform as it is used in the

‎clang/include/clang/Driver/Options.td

Lines changed: 1 addition & 1 deletion
Original file line numberDiff line numberDiff line change
@@ -2584,7 +2584,7 @@ def ftrivial_auto_var_init : Joined<["-"], "ftrivial-auto-var-init=">, Group<f_G
25842584
NormalizedValues<["Uninitialized", "Zero", "Pattern"]>,
25852585
MarshallingInfoEnum<LangOpts<"TrivialAutoVarInit">, "Uninitialized">;
25862586
def enable_trivial_var_init_zero : Flag<["-"], "enable-trivial-auto-var-init-zero-knowing-it-will-be-removed-from-clang">,
2587-
Flags<[CC1Option, CoreOption]>,
2587+
Flags<[CC1Option, CoreOption, NoArgumentUnused]>,
25882588
HelpText<"Trivial automatic variable initialization to zero is only here for benchmarks, it'll eventually be removed, and I'm OK with that because I'm only using it to benchmark">;
25892589
def ftrivial_auto_var_init_stop_after : Joined<["-"], "ftrivial-auto-var-init-stop-after=">, Group<f_Group>,
25902590
Flags<[CC1Option, CoreOption]>, HelpText<"Stop initializing trivial automatic stack variables after the specified number of instances">,

‎clang/include/clang/Sema/ParsedAttr.h

Lines changed: 2 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -70,6 +70,8 @@ struct ParsedAttrInfo {
7070
const char *NormalizedFullName;
7171
};
7272
ArrayRef<Spelling> Spellings;
73+
// The names of the known arguments of this attribute.
74+
ArrayRef<const char *> ArgNames;
7375

7476
ParsedAttrInfo(AttributeCommonInfo::Kind AttrKind =
7577
AttributeCommonInfo::NoSemaHandlerAttribute)

‎clang/include/clang/Sema/Sema.h

Lines changed: 8 additions & 9 deletions
Original file line numberDiff line numberDiff line change
@@ -11597,15 +11597,14 @@ class Sema final {
1159711597
SourceLocation ModifierLoc,
1159811598
SourceLocation EndLoc);
1159911599
/// Called on well-formed 'map' clause.
11600-
OMPClause *
11601-
ActOnOpenMPMapClause(ArrayRef<OpenMPMapModifierKind> MapTypeModifiers,
11602-
ArrayRef<SourceLocation> MapTypeModifiersLoc,
11603-
CXXScopeSpec &MapperIdScopeSpec,
11604-
DeclarationNameInfo &MapperId,
11605-
OpenMPMapClauseKind MapType, bool IsMapTypeImplicit,
11606-
SourceLocation MapLoc, SourceLocation ColonLoc,
11607-
ArrayRef<Expr *> VarList, const OMPVarListLocTy &Locs,
11608-
ArrayRef<Expr *> UnresolvedMappers = llvm::None);
11600+
OMPClause *ActOnOpenMPMapClause(
11601+
ArrayRef<OpenMPMapModifierKind> MapTypeModifiers,
11602+
ArrayRef<SourceLocation> MapTypeModifiersLoc,
11603+
CXXScopeSpec &MapperIdScopeSpec, DeclarationNameInfo &MapperId,
11604+
OpenMPMapClauseKind MapType, bool IsMapTypeImplicit,
11605+
SourceLocation MapLoc, SourceLocation ColonLoc, ArrayRef<Expr *> VarList,
11606+
const OMPVarListLocTy &Locs, bool NoDiagnose = false,
11607+
ArrayRef<Expr *> UnresolvedMappers = llvm::None);
1160911608
/// Called on well-formed 'num_teams' clause.
1161011609
OMPClause *ActOnOpenMPNumTeamsClause(Expr *NumTeams, SourceLocation StartLoc,
1161111610
SourceLocation LParenLoc,

‎clang/lib/Basic/Targets/AMDGPU.cpp

Lines changed: 1 addition & 1 deletion
Original file line numberDiff line numberDiff line change
@@ -335,7 +335,7 @@ AMDGPUTargetInfo::AMDGPUTargetInfo(const llvm::Triple &Triple,
335335
llvm::AMDGPU::getArchAttrR600(GPUKind)) {
336336
resetDataLayout(isAMDGCN(getTriple()) ? DataLayoutStringAMDGCN
337337
: DataLayoutStringR600);
338-
GridValues = llvm::omp::AMDGPUGpuGridValues;
338+
GridValues = &llvm::omp::AMDGPUGridValues;
339339

340340
setAddressSpaceMap(Triple.getOS() == llvm::Triple::Mesa3D ||
341341
!isAMDGCN(Triple));

‎clang/lib/Basic/Targets/AMDGPU.h

Lines changed: 1 addition & 10 deletions
Original file line numberDiff line numberDiff line change
@@ -352,16 +352,7 @@ class LLVM_LIBRARY_VISIBILITY AMDGPUTargetInfo final : public TargetInfo {
352352
}
353353

354354
LangAS getCUDABuiltinAddressSpace(unsigned AS) const override {
355-
switch (AS) {
356-
case 1:
357-
return LangAS::cuda_device;
358-
case 3:
359-
return LangAS::cuda_shared;
360-
case 4:
361-
return LangAS::cuda_constant;
362-
default:
363-
return getLangASFromTargetAS(AS);
364-
}
355+
return LangAS::Default;
365356
}
366357

367358
llvm::Optional<LangAS> getConstantAddressSpace() const override {

‎clang/lib/Basic/Targets/NVPTX.cpp

Lines changed: 1 addition & 1 deletion
Original file line numberDiff line numberDiff line change
@@ -65,7 +65,7 @@ NVPTXTargetInfo::NVPTXTargetInfo(const llvm::Triple &Triple,
6565
TLSSupported = false;
6666
VLASupported = false;
6767
AddrSpaceMap = &NVPTXAddrSpaceMap;
68-
GridValues = llvm::omp::NVPTXGpuGridValues;
68+
GridValues = &llvm::omp::NVPTXGridValues;
6969
UseAddrSpaceMapMangling = true;
7070
HasLegalHalfType = true;
7171
HasFloat16 = true;

‎clang/lib/Basic/Targets/PPC.cpp

Lines changed: 1 addition & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -237,6 +237,7 @@ static void defineXLCompatMacros(MacroBuilder &Builder) {
237237
Builder.defineMacro("__fsqrt", "__builtin_ppc_fsqrt");
238238
Builder.defineMacro("__fsqrts", "__builtin_ppc_fsqrts");
239239
Builder.defineMacro("__addex", "__builtin_ppc_addex");
240+
Builder.defineMacro("__cmplxl", "__builtin_complex");
240241
}
241242

242243
/// PPCTargetInfo::getTargetDefines - Return a set of the PowerPC-specific

‎clang/lib/CodeGen/BackendUtil.cpp

Lines changed: 5 additions & 3 deletions
Original file line numberDiff line numberDiff line change
@@ -1189,7 +1189,7 @@ static void addSanitizers(const Triple &TargetTriple,
11891189
CompileKernel, Recover, ModuleUseAfterScope, UseOdrIndicator,
11901190
DestructorKind));
11911191
MPM.addPass(createModuleToFunctionPassAdaptor(AddressSanitizerPass(
1192-
CompileKernel, Recover, UseAfterScope, UseAfterReturn)));
1192+
{CompileKernel, Recover, UseAfterScope, UseAfterReturn})));
11931193
}
11941194
};
11951195
ASanPass(SanitizerKind::Address, false);
@@ -1199,8 +1199,8 @@ static void addSanitizers(const Triple &TargetTriple,
11991199
if (LangOpts.Sanitize.has(Mask)) {
12001200
bool Recover = CodeGenOpts.SanitizeRecover.has(Mask);
12011201
MPM.addPass(HWAddressSanitizerPass(
1202-
CompileKernel, Recover,
1203-
/*DisableOptimization=*/CodeGenOpts.OptimizationLevel == 0));
1202+
{CompileKernel, Recover,
1203+
/*DisableOptimization=*/CodeGenOpts.OptimizationLevel == 0}));
12041204
}
12051205
};
12061206
HWASanPass(SanitizerKind::HWAddress, false);
@@ -1289,6 +1289,8 @@ void EmitAssemblyHelper::EmitAssemblyWithNewPassManager(
12891289
"", PGOOptions::NoAction, PGOOptions::CSIRInstr,
12901290
CodeGenOpts.DebugInfoForProfiling);
12911291
}
1292+
if (TM)
1293+
TM->setPGOOption(PGOOpt);
12921294

12931295
PipelineTuningOptions PTO;
12941296
PTO.LoopUnrolling = CodeGenOpts.UnrollLoops;

‎clang/lib/CodeGen/CGBuiltin.cpp

Lines changed: 23 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -9718,6 +9718,29 @@ Value *CodeGenFunction::EmitAArch64BuiltinExpr(unsigned BuiltinID,
97189718
return Builder.CreateCall(F);
97199719
}
97209720

9721+
if (BuiltinID == AArch64::BI__mulh || BuiltinID == AArch64::BI__umulh) {
9722+
llvm::Type *ResType = ConvertType(E->getType());
9723+
llvm::Type *Int128Ty = llvm::IntegerType::get(getLLVMContext(), 128);
9724+
9725+
bool IsSigned = BuiltinID == AArch64::BI__mulh;
9726+
Value *LHS =
9727+
Builder.CreateIntCast(EmitScalarExpr(E->getArg(0)), Int128Ty, IsSigned);
9728+
Value *RHS =
9729+
Builder.CreateIntCast(EmitScalarExpr(E->getArg(1)), Int128Ty, IsSigned);
9730+
9731+
Value *MulResult, *HigherBits;
9732+
if (IsSigned) {
9733+
MulResult = Builder.CreateNSWMul(LHS, RHS);
9734+
HigherBits = Builder.CreateAShr(MulResult, 64);
9735+
} else {
9736+
MulResult = Builder.CreateNUWMul(LHS, RHS);
9737+
HigherBits = Builder.CreateLShr(MulResult, 64);
9738+
}
9739+
HigherBits = Builder.CreateIntCast(HigherBits, ResType, IsSigned);
9740+
9741+
return HigherBits;
9742+
}
9743+
97219744
// Handle MSVC intrinsics before argument evaluation to prevent double
97229745
// evaluation.
97239746
if (Optional<MSVCIntrin> MsvcIntId = translateAarch64ToMsvcIntrin(BuiltinID))

‎clang/lib/CodeGen/CGDebugInfo.cpp

Lines changed: 16 additions & 1 deletion
Original file line numberDiff line numberDiff line change
@@ -2070,6 +2070,17 @@ llvm::DINodeArray CGDebugInfo::CollectCXXTemplateParams(
20702070
return CollectTemplateParams(TPList, TAList.asArray(), Unit);
20712071
}
20722072

2073+
llvm::DINodeArray CGDebugInfo::CollectBTFTagAnnotations(const Decl *D) {
2074+
SmallVector<llvm::Metadata *, 4> Annotations;
2075+
for (const auto *I : D->specific_attrs<BTFTagAttr>()) {
2076+
llvm::Metadata *Ops[2] = {
2077+
llvm::MDString::get(CGM.getLLVMContext(), StringRef("btf_tag")),
2078+
llvm::MDString::get(CGM.getLLVMContext(), I->getBTFTag())};
2079+
Annotations.push_back(llvm::MDNode::get(CGM.getLLVMContext(), Ops));
2080+
}
2081+
return DBuilder.getOrCreateArray(Annotations);
2082+
}
2083+
20732084
llvm::DIType *CGDebugInfo::getOrCreateVTablePtrType(llvm::DIFile *Unit) {
20742085
if (VTablePtrType)
20752086
return VTablePtrType;
@@ -3442,9 +3453,13 @@ llvm::DICompositeType *CGDebugInfo::CreateLimitedType(const RecordType *Ty) {
34423453
Flags |= llvm::DINode::FlagExportSymbols;
34433454
}
34443455

3456+
llvm::DINodeArray Annotations = nullptr;
3457+
if (D->hasAttr<BTFTagAttr>())
3458+
Annotations = CollectBTFTagAnnotations(D);
3459+
34453460
llvm::DICompositeType *RealDecl = DBuilder.createReplaceableCompositeType(
34463461
getTagForRecord(RD), RDName, RDContext, DefUnit, Line, 0, Size, Align,
3447-
Flags, Identifier);
3462+
Flags, Identifier, Annotations);
34483463

34493464
// Elements of composite types usually have back to the type, creating
34503465
// uniquing cycles. Distinct nodes are more efficient.

‎clang/lib/CodeGen/CGDebugInfo.h

Lines changed: 3 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -297,6 +297,9 @@ class CGDebugInfo {
297297
CollectCXXTemplateParams(const ClassTemplateSpecializationDecl *TS,
298298
llvm::DIFile *F);
299299

300+
/// A helper function to collect debug info for btf_tag annotations.
301+
llvm::DINodeArray CollectBTFTagAnnotations(const Decl *D);
302+
300303
llvm::DIType *createFieldType(StringRef name, QualType type,
301304
SourceLocation loc, AccessSpecifier AS,
302305
uint64_t offsetInBits, uint32_t AlignInBits,

‎clang/lib/CodeGen/CGExprScalar.cpp

Lines changed: 2 additions & 5 deletions
Original file line numberDiff line numberDiff line change
@@ -4806,11 +4806,8 @@ Value *ScalarExprEmitter::VisitAsTypeExpr(AsTypeExpr *E) {
48064806
// vector to get a vec4, then a bitcast if the target type is different.
48074807
if (NumElementsSrc == 3 && NumElementsDst != 3) {
48084808
Src = ConvertVec3AndVec4(Builder, CGF, Src, 4);
4809-
4810-
if (!CGF.CGM.getCodeGenOpts().PreserveVec3Type) {
4811-
Src = createCastsForTypeOfSameSize(Builder, CGF.CGM.getDataLayout(), Src,
4812-
DstTy);
4813-
}
4809+
Src = createCastsForTypeOfSameSize(Builder, CGF.CGM.getDataLayout(), Src,
4810+
DstTy);
48144811

48154812
Src->setName("astype");
48164813
return Src;

‎clang/lib/CodeGen/CGOpenMPRuntimeAMDGCN.cpp

Lines changed: 2 additions & 1 deletion
Original file line numberDiff line numberDiff line change
@@ -20,6 +20,7 @@
2020
#include "clang/AST/StmtVisitor.h"
2121
#include "clang/Basic/Cuda.h"
2222
#include "llvm/ADT/SmallPtrSet.h"
23+
#include "llvm/Frontend/OpenMP/OMPGridValues.h"
2324
#include "llvm/IR/IntrinsicsAMDGPU.h"
2425

2526
using namespace clang;
@@ -35,7 +36,7 @@ CGOpenMPRuntimeAMDGCN::CGOpenMPRuntimeAMDGCN(CodeGenModule &CGM)
3536
llvm::Value *CGOpenMPRuntimeAMDGCN::getGPUWarpSize(CodeGenFunction &CGF) {
3637
CGBuilderTy &Bld = CGF.Builder;
3738
// return constant compile-time target-specific warp size
38-
unsigned WarpSize = CGF.getTarget().getGridValue(llvm::omp::GV_Warp_Size);
39+
unsigned WarpSize = CGF.getTarget().getGridValue().GV_Warp_Size;
3940
return Bld.getInt32(WarpSize);
4041
}
4142

‎clang/lib/CodeGen/CGOpenMPRuntimeGPU.cpp

Lines changed: 6 additions & 7 deletions
Original file line numberDiff line numberDiff line change
@@ -339,7 +339,7 @@ class CheckVarsEscapingDeclContext final
339339
assert(!GlobalizedRD &&
340340
"Record for globalized variables is built already.");
341341
ArrayRef<const ValueDecl *> EscapedDeclsForParallel, EscapedDeclsForTeams;
342-
unsigned WarpSize = CGF.getTarget().getGridValue(llvm::omp::GV_Warp_Size);
342+
unsigned WarpSize = CGF.getTarget().getGridValue().GV_Warp_Size;
343343
if (IsInTTDRegion)
344344
EscapedDeclsForTeams = EscapedDecls.getArrayRef();
345345
else
@@ -535,8 +535,7 @@ class CheckVarsEscapingDeclContext final
535535
/// on the NVPTX device, to generate more efficient code.
536536
static llvm::Value *getNVPTXWarpID(CodeGenFunction &CGF) {
537537
CGBuilderTy &Bld = CGF.Builder;
538-
unsigned LaneIDBits =
539-
CGF.getTarget().getGridValue(llvm::omp::GV_Warp_Size_Log2);
538+
unsigned LaneIDBits = CGF.getTarget().getGridValue().GV_Warp_Size_Log2;
540539
auto &RT = static_cast<CGOpenMPRuntimeGPU &>(CGF.CGM.getOpenMPRuntime());
541540
return Bld.CreateAShr(RT.getGPUThreadID(CGF), LaneIDBits, "nvptx_warp_id");
542541
}
@@ -546,8 +545,8 @@ static llvm::Value *getNVPTXWarpID(CodeGenFunction &CGF) {
546545
/// on the NVPTX device, to generate more efficient code.
547546
static llvm::Value *getNVPTXLaneID(CodeGenFunction &CGF) {
548547
CGBuilderTy &Bld = CGF.Builder;
549-
unsigned LaneIDMask = CGF.getContext().getTargetInfo().getGridValue(
550-
llvm::omp::GV_Warp_Size_Log2_Mask);
548+
unsigned LaneIDMask =
549+
CGF.getContext().getTargetInfo().getGridValue().GV_Warp_Size_Log2_Mask;
551550
auto &RT = static_cast<CGOpenMPRuntimeGPU &>(CGF.CGM.getOpenMPRuntime());
552551
return Bld.CreateAnd(RT.getGPUThreadID(CGF), Bld.getInt32(LaneIDMask),
553552
"nvptx_lane_id");
@@ -1308,7 +1307,7 @@ llvm::Function *CGOpenMPRuntimeGPU::emitTeamsOutlinedFunction(
13081307
const RecordDecl *GlobalizedRD = nullptr;
13091308
llvm::SmallVector<const ValueDecl *, 4> LastPrivatesReductions;
13101309
llvm::SmallDenseMap<const ValueDecl *, const FieldDecl *> MappedDeclsFields;
1311-
unsigned WarpSize = CGM.getTarget().getGridValue(llvm::omp::GV_Warp_Size);
1310+
unsigned WarpSize = CGM.getTarget().getGridValue().GV_Warp_Size;
13121311
// Globalize team reductions variable unconditionally in all modes.
13131312
if (getExecutionMode() != CGOpenMPRuntimeGPU::EM_SPMD)
13141313
getTeamsReductionVars(CGM.getContext(), D, LastPrivatesReductions);
@@ -2089,7 +2088,7 @@ static llvm::Value *emitInterWarpCopyFunction(CodeGenModule &CGM,
20892088
"__openmp_nvptx_data_transfer_temporary_storage";
20902089
llvm::GlobalVariable *TransferMedium =
20912090
M.getGlobalVariable(TransferMediumName);
2092-
unsigned WarpSize = CGF.getTarget().getGridValue(llvm::omp::GV_Warp_Size);
2091+
unsigned WarpSize = CGF.getTarget().getGridValue().GV_Warp_Size;
20932092
if (!TransferMedium) {
20942093
auto *Ty = llvm::ArrayType::get(CGM.Int32Ty, WarpSize);
20952094
unsigned SharedAddressSpace = C.getTargetAddressSpace(LangAS::cuda_shared);

‎clang/lib/CodeGen/CGOpenMPRuntimeGPU.h

Lines changed: 0 additions & 1 deletion
Original file line numberDiff line numberDiff line change
@@ -17,7 +17,6 @@
1717
#include "CGOpenMPRuntime.h"
1818
#include "CodeGenFunction.h"
1919
#include "clang/AST/StmtOpenMP.h"
20-
#include "llvm/Frontend/OpenMP/OMPGridValues.h"
2120

2221
namespace clang {
2322
namespace CodeGen {

0 commit comments

Comments
 (0)