r286944 - [OPENMP] Fixed codegen for 'omp cancel' construct.
Alexey Bataev via cfe-commits
cfe-commits at lists.llvm.org
Tue Nov 15 01:11:51 PST 2016
Author: abataev
Date: Tue Nov 15 03:11:50 2016
New Revision: 286944
URL: http://llvm.org/viewvc/llvm-project?rev=286944&view=rev
Log:
[OPENMP] Fixed codegen for 'omp cancel' construct.
If 'omp cancel' construct is used in a worksharing construct it may cause
hanging of the software in case if reduction clause is used. Patch fixes
this problem by avoiding extra reduction processing for branches that
were canceled.
Modified:
cfe/trunk/lib/CodeGen/CGStmtOpenMP.cpp
cfe/trunk/lib/CodeGen/CodeGenFunction.h
cfe/trunk/test/OpenMP/cancel_codegen.cpp
Modified: cfe/trunk/lib/CodeGen/CGStmtOpenMP.cpp
URL: http://llvm.org/viewvc/llvm-project/cfe/trunk/lib/CodeGen/CGStmtOpenMP.cpp?rev=286944&r1=286943&r2=286944&view=diff
==============================================================================
--- cfe/trunk/lib/CodeGen/CGStmtOpenMP.cpp (original)
+++ cfe/trunk/lib/CodeGen/CGStmtOpenMP.cpp Tue Nov 15 03:11:50 2016
@@ -1781,9 +1781,17 @@ void CodeGenFunction::EmitOMPOuterLoop(b
EmitBlock(LoopExit.getBlock());
// Tell the runtime we are done.
- if (!DynamicOrOrdered)
- RT.emitForStaticFinish(*this, S.getLocEnd());
+ SourceLocation ELoc = S.getLocEnd();
+ auto &&CodeGen = [DynamicOrOrdered, ELoc](CodeGenFunction &CGF) {
+ if (!DynamicOrOrdered)
+ CGF.CGM.getOpenMPRuntime().emitForStaticFinish(CGF, ELoc);
+ };
+ CodeGen(*this);
+ OpenMPDirectiveKind DKind = S.getDirectiveKind();
+ if (DKind == OMPD_for || DKind == OMPD_parallel_for ||
+ DKind == OMPD_distribute_parallel_for)
+ OMPCancelStack.back().CodeGen = CodeGen;
}
void CodeGenFunction::EmitOMPForOuterLoop(
@@ -1895,6 +1903,7 @@ void CodeGenFunction::EmitOMPDistributeO
void CodeGenFunction::EmitOMPDistributeParallelForDirective(
const OMPDistributeParallelForDirective &S) {
OMPLexicalScope Scope(*this, S, /*AsInlined=*/true);
+ OMPCancelStackRAII CancelRegion(*this);
CGM.getOpenMPRuntime().emitInlinedDirective(
*this, OMPD_distribute_parallel_for,
[&S](CodeGenFunction &CGF, PrePostActionTy &) {
@@ -2123,7 +2132,15 @@ bool CodeGenFunction::EmitOMPWorksharing
[](CodeGenFunction &) {});
EmitBlock(LoopExit.getBlock());
// Tell the runtime we are done.
- RT.emitForStaticFinish(*this, S.getLocStart());
+ SourceLocation ELoc = S.getLocEnd();
+ auto &&CodeGen = [ELoc](CodeGenFunction &CGF) {
+ CGF.CGM.getOpenMPRuntime().emitForStaticFinish(CGF, ELoc);
+ };
+ CodeGen(*this);
+ OpenMPDirectiveKind DKind = S.getDirectiveKind();
+ if (DKind == OMPD_for || DKind == OMPD_parallel_for ||
+ DKind == OMPD_distribute_parallel_for)
+ OMPCancelStack.back().CodeGen = CodeGen;
} else {
const bool IsMonotonic =
Ordered || ScheduleKind.Schedule == OMPC_SCHEDULE_static ||
@@ -2177,6 +2194,7 @@ void CodeGenFunction::EmitOMPForDirectiv
};
{
OMPLexicalScope Scope(*this, S, /*AsInlined=*/true);
+ OMPCancelStackRAII CancelRegion(*this);
CGM.getOpenMPRuntime().emitInlinedDirective(*this, OMPD_for, CodeGen,
S.hasCancel());
}
@@ -2313,7 +2331,12 @@ void CodeGenFunction::EmitSections(const
CGF.EmitOMPInnerLoop(S, /*RequiresCleanup=*/false, &Cond, &Inc, BodyGen,
[](CodeGenFunction &) {});
// Tell the runtime we are done.
- CGF.CGM.getOpenMPRuntime().emitForStaticFinish(CGF, S.getLocStart());
+ SourceLocation ELoc = S.getLocEnd();
+ auto &&CodeGen = [ELoc](CodeGenFunction &CGF) {
+ CGF.CGM.getOpenMPRuntime().emitForStaticFinish(CGF, ELoc);
+ };
+ CodeGen(CGF);
+ CGF.OMPCancelStack.back().CodeGen = CodeGen;
CGF.EmitOMPReductionClauseFinal(S);
// Emit post-update of the reduction variables if IsLastIter != 0.
emitPostUpdateForReductionClause(
@@ -2351,6 +2374,7 @@ void CodeGenFunction::EmitSections(const
void CodeGenFunction::EmitOMPSectionsDirective(const OMPSectionsDirective &S) {
{
OMPLexicalScope Scope(*this, S, /*AsInlined=*/true);
+ OMPCancelStackRAII CancelRegion(*this);
EmitSections(S);
}
// Emit an implicit barrier at the end.
@@ -2438,6 +2462,7 @@ void CodeGenFunction::EmitOMPParallelFor
// Emit directive as a combined directive that consists of two implicit
// directives: 'parallel' with 'for' directive.
auto &&CodeGen = [&S](CodeGenFunction &CGF, PrePostActionTy &) {
+ OMPCancelStackRAII CancelRegion(CGF);
CGF.EmitOMPWorksharingLoop(S);
};
emitCommonOMPParallelDirective(*this, S, OMPD_for, CodeGen);
@@ -2458,6 +2483,7 @@ void CodeGenFunction::EmitOMPParallelSec
// Emit directive as a combined directive that consists of two implicit
// directives: 'parallel' with 'sections' directive.
auto &&CodeGen = [&S](CodeGenFunction &CGF, PrePostActionTy &) {
+ OMPCancelStackRAII CancelRegion(CGF);
CGF.EmitSections(S);
};
emitCommonOMPParallelDirective(*this, S, OMPD_sections, CodeGen);
@@ -3438,8 +3464,11 @@ CodeGenFunction::getOMPCancelDestination
if (Kind == OMPD_parallel || Kind == OMPD_task)
return ReturnBlock;
assert(Kind == OMPD_for || Kind == OMPD_section || Kind == OMPD_sections ||
- Kind == OMPD_parallel_sections || Kind == OMPD_parallel_for);
- return BreakContinueStack.back().BreakBlock;
+ Kind == OMPD_parallel_sections || Kind == OMPD_parallel_for ||
+ Kind == OMPD_distribute_parallel_for);
+ if (!OMPCancelStack.back().ExitBlock.isValid())
+ OMPCancelStack.back().ExitBlock = getJumpDestInCurrentScope("cancel.exit");
+ return OMPCancelStack.back().ExitBlock;
}
void CodeGenFunction::EmitOMPUseDevicePtrClause(
Modified: cfe/trunk/lib/CodeGen/CodeGenFunction.h
URL: http://llvm.org/viewvc/llvm-project/cfe/trunk/lib/CodeGen/CodeGenFunction.h?rev=286944&r1=286943&r2=286944&view=diff
==============================================================================
--- cfe/trunk/lib/CodeGen/CodeGenFunction.h (original)
+++ cfe/trunk/lib/CodeGen/CodeGenFunction.h Tue Nov 15 03:11:50 2016
@@ -982,6 +982,35 @@ private:
};
SmallVector<BreakContinue, 8> BreakContinueStack;
+ /// Data for exit block for proper support of OpenMP cancellation constructs.
+ struct OMPCancel {
+ JumpDest ExitBlock;
+ llvm::function_ref<void(CodeGenFunction &CGF)> CodeGen;
+ OMPCancel() : CodeGen([](CodeGenFunction &CGF) {}) {}
+ };
+ SmallVector<OMPCancel, 8> OMPCancelStack;
+
+ /// Controls insertion of cancellation exit blocks in worksharing constructs.
+ class OMPCancelStackRAII {
+ CodeGenFunction &CGF;
+
+ public:
+ OMPCancelStackRAII(CodeGenFunction &CGF) : CGF(CGF) {
+ CGF.OMPCancelStack.push_back({});
+ }
+ ~OMPCancelStackRAII() {
+ if (CGF.HaveInsertPoint() &&
+ CGF.OMPCancelStack.back().ExitBlock.isValid()) {
+ auto CJD = CGF.getJumpDestInCurrentScope("cancel.cont");
+ CGF.EmitBranchThroughCleanup(CJD);
+ CGF.EmitBlock(CGF.OMPCancelStack.back().ExitBlock.getBlock());
+ CGF.OMPCancelStack.back().CodeGen(CGF);
+ CGF.EmitBranchThroughCleanup(CJD);
+ CGF.EmitBlock(CJD.getBlock());
+ }
+ }
+ };
+
CodeGenPGO PGO;
/// Calculate branch weights appropriate for PGO data
Modified: cfe/trunk/test/OpenMP/cancel_codegen.cpp
URL: http://llvm.org/viewvc/llvm-project/cfe/trunk/test/OpenMP/cancel_codegen.cpp?rev=286944&r1=286943&r2=286944&view=diff
==============================================================================
--- cfe/trunk/test/OpenMP/cancel_codegen.cpp (original)
+++ cfe/trunk/test/OpenMP/cancel_codegen.cpp Tue Nov 15 03:11:50 2016
@@ -90,9 +90,11 @@ for (int i = 0; i < argc; ++i) {
}
}
// CHECK: call void (%ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(
-#pragma omp parallel for
+int r = 0;
+#pragma omp parallel for reduction(+:r)
for (int i = 0; i < argc; ++i) {
#pragma omp cancel for
+ r += i;
}
// CHECK: call void (%ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(
return argc;
@@ -163,6 +165,9 @@ for (int i = 0; i < argc; ++i) {
// CHECK: [[CONTINUE]]
// CHECK: br label
// CHECK: call void @__kmpc_for_static_fini(
+// CHECK: call i32 @__kmpc_reduce_nowait(
+// CHECK: call void @__kmpc_end_reduce_nowait(
+// CHECK: call void @__kmpc_for_static_fini(
// CHECK: ret void
#endif
More information about the cfe-commits
mailing list