jyu2 created this revision.
jyu2 added reviewers: ABataev, mikerice, jdoerfert.
jyu2 added a project: OpenMP.
Herald added subscribers: sunshaoce, guansong, yaxunl.
Herald added a project: All.
jyu2 requested review of this revision.
Herald added subscribers: cfe-commits, jplehr, sstefan1.
Herald added a project: clang.
This is following the ordered depend clause.
1> Using OMPClause instead OMPDependClause in emitDoacrossOrdered to hand both
OMPDependClause and OMPDoacrossClause.
2> Add new utility function emitRestoreIP which used for both clause.
Repository:
rG LLVM Github Monorepo
https://reviews.llvm.org/D154180
Files:
clang/lib/CodeGen/CGOpenMPRuntime.cpp
clang/lib/CodeGen/CGOpenMPRuntime.h
clang/lib/CodeGen/CGStmtOpenMP.cpp
clang/test/OpenMP/ordered_doacross_codegen.c
Index: clang/test/OpenMP/ordered_doacross_codegen.c
===================================================================
--- clang/test/OpenMP/ordered_doacross_codegen.c
+++ clang/test/OpenMP/ordered_doacross_codegen.c
@@ -2,13 +2,21 @@
// RUN: %clang_cc1 -fopenmp -triple x86_64-unknown-unknown -emit-pch -o %t %s
// RUN: %clang_cc1 -fopenmp -triple x86_64-unknown-unknown -include-pch %t -verify %s -emit-llvm -o - | FileCheck %s --check-prefixes=CHECK,CHECK-NORMAL
+// RUN: %clang_cc1 -verify -fopenmp -triple x86_64-unknown-unknown -emit-llvm %s -o - -fopenmp-version=52 | FileCheck %s --check-prefixes=CHECK,CHECK-NORMAL
// RUN: %clang_cc1 -verify -fopenmp -fopenmp-enable-irbuilder -triple x86_64-unknown-unknown -emit-llvm %s -o - | FileCheck %s --check-prefixes=CHECK,CHECK-IRBUILDER
+// RUN: %clang_cc1 -verify -fopenmp -fopenmp-enable-irbuilder -triple x86_64-unknown-unknown -fopenmp-version=52 -emit-llvm %s -o - | FileCheck %s --check-prefixes=CHECK,CHECK-IRBUILDER
// RUN: %clang_cc1 -fopenmp -fopenmp-enable-irbuilder -triple x86_64-unknown-unknown -emit-pch -o %t %s
// RUN: %clang_cc1 -fopenmp -fopenmp-enable-irbuilder -triple x86_64-unknown-unknown -include-pch %t -verify %s -emit-llvm -o - | FileCheck %s --check-prefixes=CHECK,CHECK-IRBUILDER
+// RUN: %clang_cc1 -fopenmp -fopenmp-version=52 -fopenmp-enable-irbuilder -triple x86_64-unknown-unknown -emit-pch -o %t %s
+// RUN: %clang_cc1 -fopenmp -fopenmp-version=52 -fopenmp-enable-irbuilder -triple x86_64-unknown-unknown -include-pch %t -verify %s -emit-llvm -o - | FileCheck %s --check-prefixes=CHECK,CHECK-IRBUILDER
+
// RUN: %clang_cc1 -verify -fopenmp-simd -triple x86_64-unknown-unknown -emit-llvm %s -o - | FileCheck --check-prefix SIMD-ONLY0 %s
+// RUN: %clang_cc1 -verify -fopenmp-simd -fopenmp-version=52 -triple x86_64-unknown-unknown -emit-llvm %s -o - | FileCheck --check-prefix SIMD-ONLY0 %s
// RUN: %clang_cc1 -fopenmp-simd -triple x86_64-unknown-unknown -emit-pch -o %t %s
// RUN: %clang_cc1 -fopenmp-simd -triple x86_64-unknown-unknown -include-pch %t -verify %s -emit-llvm -o - | FileCheck --check-prefix SIMD-ONLY0 %s
+// RUN: %clang_cc1 -fopenmp-simd -fopenmp-version=52 -triple x86_64-unknown-unknown -emit-pch -o %t %s
+// RUN: %clang_cc1 -fopenmp-simd -fopenmp-version=52 -triple x86_64-unknown-unknown -include-pch %t -verify %s -emit-llvm -o - | FileCheck --check-prefix SIMD-ONLY0 %s
// SIMD-ONLY0-NOT: {{__kmpc|__tgt}}
// expected-no-diagnostics
@@ -51,7 +59,11 @@
// CHECK-NORMAL-NEXT: call void @__kmpc_doacross_post(ptr [[IDENT]], i32 [[GTID]], ptr [[TMP]])
// CHECK-IRBUILDER-NEXT: [[GTID1:%.+]] = call i32 @__kmpc_global_thread_num(ptr [[IDENT:@.+]])
// CHECK-IRBUILDER-NEXT: call void @__kmpc_doacross_post(ptr [[IDENT]], i32 [[GTID1]], ptr [[TMP]])
+#if _OPENMP >= 202111
+#pragma omp ordered doacross(source:)
+#else
#pragma omp ordered depend(source)
+#endif
c[i] = c[i] + 1;
foo();
// CHECK: call void @foo()
@@ -66,7 +78,11 @@
// CHECK-NORMAL-NEXT: call void @__kmpc_doacross_wait(ptr [[IDENT]], i32 [[GTID]], ptr [[TMP]])
// CHECK-IRBUILDER-NEXT: [[GTID2:%.+]] = call i32 @__kmpc_global_thread_num(ptr [[IDENT:@.+]])
// CHECK-IRBUILDER-NEXT: call void @__kmpc_doacross_wait(ptr [[IDENT]], i32 [[GTID2]], ptr [[TMP]])
+#if _OPENMP >= 202111
+#pragma omp ordered doacross(sink : i - 2)
+#else
#pragma omp ordered depend(sink : i - 2)
+#endif
d[i] = a[i - 2];
}
// CHECK: call void @__kmpc_for_static_fini(
Index: clang/lib/CodeGen/CGStmtOpenMP.cpp
===================================================================
--- clang/lib/CodeGen/CGStmtOpenMP.cpp
+++ clang/lib/CodeGen/CGStmtOpenMP.cpp
@@ -5839,6 +5839,33 @@
return Fn;
}
+static void emitRestoreIP(CodeGenFunction &CGF, const OMPClause *C,
+ llvm::OpenMPIRBuilder::InsertPointTy AllocaIP,
+ llvm::OpenMPIRBuilder &OMPBuilder) {
+ auto DOC = dyn_cast<OMPDoacrossClause>(C);
+ auto DC = dyn_cast<OMPDependClause>(C);
+
+ unsigned NumLoops = DC ? DC->getNumLoops() : DOC->getNumLoops();
+ QualType Int64Ty = CGF.CGM.getContext().getIntTypeForBitwidth(
+ /*DestWidth=*/64, /*Signed=*/1);
+ llvm::SmallVector<llvm::Value *> StoreValues;
+ for (unsigned I = 0; I < NumLoops; I++) {
+ const Expr *CounterVal = DC ? DC->getLoopData(I) : DOC->getLoopData(I);
+ assert(CounterVal);
+ llvm::Value *StoreValue = CGF.EmitScalarConversion(
+ CGF.EmitScalarExpr(CounterVal), CounterVal->getType(), Int64Ty,
+ CounterVal->getExprLoc());
+ StoreValues.emplace_back(StoreValue);
+ }
+ bool IsDependSource = false;
+ if ((DC && DC->getDependencyKind() == OMPC_DEPEND_source) ||
+ (DOC && DOC->getDependenceType() == OMPC_DOACROSS_source))
+ IsDependSource = true;
+ CGF.Builder.restoreIP(
+ OMPBuilder.createOrderedDepend(CGF.Builder, AllocaIP, NumLoops,
+ StoreValues, ".cnt.addr", IsDependSource));
+}
+
void CodeGenFunction::EmitOMPOrderedDirective(const OMPOrderedDirective &S) {
if (CGM.getLangOpts().OpenMPIRBuilder) {
llvm::OpenMPIRBuilder &OMPBuilder = CGM.getOpenMPRuntime().getOMPBuilder();
@@ -5850,26 +5877,16 @@
"No associated statement must be in ordered depend construct.");
InsertPointTy AllocaIP(AllocaInsertPt->getParent(),
AllocaInsertPt->getIterator());
- for (const auto *DC : S.getClausesOfKind<OMPDependClause>()) {
- unsigned NumLoops = DC->getNumLoops();
- QualType Int64Ty = CGM.getContext().getIntTypeForBitwidth(
- /*DestWidth=*/64, /*Signed=*/1);
- llvm::SmallVector<llvm::Value *> StoreValues;
- for (unsigned I = 0; I < NumLoops; I++) {
- const Expr *CounterVal = DC->getLoopData(I);
- assert(CounterVal);
- llvm::Value *StoreValue = EmitScalarConversion(
- EmitScalarExpr(CounterVal), CounterVal->getType(), Int64Ty,
- CounterVal->getExprLoc());
- StoreValues.emplace_back(StoreValue);
- }
- bool IsDependSource = false;
- if (DC->getDependencyKind() == OMPC_DEPEND_source)
- IsDependSource = true;
- Builder.restoreIP(OMPBuilder.createOrderedDepend(
- Builder, AllocaIP, NumLoops, StoreValues, ".cnt.addr",
- IsDependSource));
- }
+ for (const auto *DC : S.getClausesOfKind<OMPDependClause>())
+ emitRestoreIP(*this, dyn_cast<OMPClause>(DC), AllocaIP, OMPBuilder);
+ } else if (S.hasClausesOfKind<OMPDoacrossClause>()) {
+ // The ordered directive with doacross clause.
+ assert(!S.hasAssociatedStmt() &&
+ "No associated statement must be in ordered doacross construct.");
+ InsertPointTy AllocaIP(AllocaInsertPt->getParent(),
+ AllocaInsertPt->getIterator());
+ for (const auto *DC : S.getClausesOfKind<OMPDoacrossClause>())
+ emitRestoreIP(*this, dyn_cast<OMPClause>(DC), AllocaIP, OMPBuilder);
} else {
// The ordered directive with threads or simd clause, or without clause.
// Without clause, it behaves as if the threads clause is specified.
@@ -5916,6 +5933,13 @@
CGM.getOpenMPRuntime().emitDoacrossOrdered(*this, DC);
return;
}
+ if (S.hasClausesOfKind<OMPDoacrossClause>()) {
+ assert(!S.hasAssociatedStmt() &&
+ "No associated statement must be in ordered doacross construct.");
+ for (const auto *DC : S.getClausesOfKind<OMPDoacrossClause>())
+ CGM.getOpenMPRuntime().emitDoacrossOrdered(*this, DC);
+ return;
+ }
const auto *C = S.getSingleClause<OMPSIMDClause>();
auto &&CodeGen = [&S, C, this](CodeGenFunction &CGF,
PrePostActionTy &Action) {
Index: clang/lib/CodeGen/CGOpenMPRuntime.h
===================================================================
--- clang/lib/CodeGen/CGOpenMPRuntime.h
+++ clang/lib/CodeGen/CGOpenMPRuntime.h
@@ -1482,10 +1482,9 @@
virtual void emitDoacrossInit(CodeGenFunction &CGF, const OMPLoopDirective &D,
ArrayRef<Expr *> NumIterations);
- /// Emit code for doacross ordered directive with 'depend' clause.
- /// \param C 'depend' clause with 'sink|source' dependency kind.
- virtual void emitDoacrossOrdered(CodeGenFunction &CGF,
- const OMPDependClause *C);
+ /// Emit code for doacross ordered directive with 'doacross | depend' clause.
+ /// \param C 'doacross | depend' clause with 'sink|source' dependence type.
+ virtual void emitDoacrossOrdered(CodeGenFunction &CGF, const OMPClause *C);
/// Translates the native parameter of outlined function if this is required
/// for target.
@@ -2235,10 +2234,9 @@
void emitDoacrossInit(CodeGenFunction &CGF, const OMPLoopDirective &D,
ArrayRef<Expr *> NumIterations) override;
- /// Emit code for doacross ordered directive with 'depend' clause.
- /// \param C 'depend' clause with 'sink|source' dependency kind.
- void emitDoacrossOrdered(CodeGenFunction &CGF,
- const OMPDependClause *C) override;
+ /// Emit code for doacross ordered directive with 'doacross | depend' clause.
+ /// \param C 'doacross | depend' clause with 'sink|source' dependence type.
+ void emitDoacrossOrdered(CodeGenFunction &CGF, const OMPClause *C) override;
/// Translates the native parameter of outlined function if this is required
/// for target.
Index: clang/lib/CodeGen/CGOpenMPRuntime.cpp
===================================================================
--- clang/lib/CodeGen/CGOpenMPRuntime.cpp
+++ clang/lib/CodeGen/CGOpenMPRuntime.cpp
@@ -11380,15 +11380,21 @@
}
void CGOpenMPRuntime::emitDoacrossOrdered(CodeGenFunction &CGF,
- const OMPDependClause *C) {
+ const OMPClause *CL) {
+ const OMPDependClause *DC =
+ isa<OMPDependClause>(CL) ? dyn_cast<OMPDependClause>(CL) : nullptr;
+ const OMPDoacrossClause *DoC =
+ isa<OMPDoacrossClause>(CL) ? dyn_cast<OMPDoacrossClause>(CL) : nullptr;
QualType Int64Ty =
CGM.getContext().getIntTypeForBitwidth(/*DestWidth=*/64, /*Signed=*/1);
- llvm::APInt Size(/*numBits=*/32, C->getNumLoops());
+ llvm::APInt Size(/*numBits=*/32,
+ DoC ? DoC->getNumLoops() : DC->getNumLoops());
QualType ArrayTy = CGM.getContext().getConstantArrayType(
Int64Ty, Size, nullptr, ArrayType::Normal, 0);
Address CntAddr = CGF.CreateMemTemp(ArrayTy, ".cnt.addr");
- for (unsigned I = 0, E = C->getNumLoops(); I < E; ++I) {
- const Expr *CounterVal = C->getLoopData(I);
+ for (unsigned I = 0, E = DoC ? DoC->getNumLoops() : DC->getNumLoops(); I < E;
+ ++I) {
+ const Expr *CounterVal = DoC ? DoC->getLoopData(I) : DC->getLoopData(I);
assert(CounterVal);
llvm::Value *CntVal = CGF.EmitScalarConversion(
CGF.EmitScalarExpr(CounterVal), CounterVal->getType(), Int64Ty,
@@ -11397,17 +11403,29 @@
/*Volatile=*/false, Int64Ty);
}
llvm::Value *Args[] = {
- emitUpdateLocation(CGF, C->getBeginLoc()),
- getThreadID(CGF, C->getBeginLoc()),
+ emitUpdateLocation(CGF, DoC ? DoC->getBeginLoc() : DC->getBeginLoc()),
+ getThreadID(CGF, DoC ? DoC->getBeginLoc() : DC->getBeginLoc()),
CGF.Builder.CreateConstArrayGEP(CntAddr, 0).getPointer()};
llvm::FunctionCallee RTLFn;
- if (C->getDependencyKind() == OMPC_DEPEND_source) {
- RTLFn = OMPBuilder.getOrCreateRuntimeFunction(CGM.getModule(),
- OMPRTL___kmpc_doacross_post);
- } else {
- assert(C->getDependencyKind() == OMPC_DEPEND_sink);
- RTLFn = OMPBuilder.getOrCreateRuntimeFunction(CGM.getModule(),
- OMPRTL___kmpc_doacross_wait);
+ if (DoC) {
+ if (DoC->getDependenceType() == OMPC_DOACROSS_source) {
+ RTLFn = OMPBuilder.getOrCreateRuntimeFunction(
+ CGM.getModule(), OMPRTL___kmpc_doacross_post);
+ } else {
+ assert(DoC->getDependenceType() == OMPC_DOACROSS_sink);
+ RTLFn = OMPBuilder.getOrCreateRuntimeFunction(
+ CGM.getModule(), OMPRTL___kmpc_doacross_wait);
+ }
+ }
+ if (DC) {
+ if (DC->getDependencyKind() == OMPC_DEPEND_source) {
+ RTLFn = OMPBuilder.getOrCreateRuntimeFunction(
+ CGM.getModule(), OMPRTL___kmpc_doacross_post);
+ } else {
+ assert(DC->getDependencyKind() == OMPC_DEPEND_sink);
+ RTLFn = OMPBuilder.getOrCreateRuntimeFunction(
+ CGM.getModule(), OMPRTL___kmpc_doacross_wait);
+ }
}
CGF.EmitRuntimeCall(RTLFn, Args);
}
@@ -12394,7 +12412,7 @@
}
void CGOpenMPSIMDRuntime::emitDoacrossOrdered(CodeGenFunction &CGF,
- const OMPDependClause *C) {
+ const OMPClause *C) {
llvm_unreachable("Not supported in SIMD-only mode");
}
_______________________________________________
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
