[Clang][OpenMP] Bug fix Default clause variable category #165276
Merged
Conversation
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
variable category.
2) Adding a new test case
clang/test/OpenMP/parallel_default_variableCategory_codegen.cpp
3) Taking care of new comments in
llvm#157063
llvmbot
added
clang
Member
|
@llvm/pr-subscribers-clang Author: None (SunilKuravinakop) ChangesIn the default clause, check for allocatable type in the variable category has been made more robust. Taking care of new comments in the previous "Support for Default clause variable category" 157063. Full diff: https://github.com/llvm/llvm-project/pull/165276.diff 2 Files Affected:
diff --git a/clang/lib/Sema/SemaOpenMP.cpp b/clang/lib/Sema/SemaOpenMP.cpp
index 5b5b1b685e153..f677be7f02583 100644
--- a/clang/lib/Sema/SemaOpenMP.cpp
+++ b/clang/lib/Sema/SemaOpenMP.cpp
@@ -1314,6 +1314,22 @@ static std::string getOpenMPClauseNameForDiag(OpenMPClauseKind C) {
return getOpenMPClauseName(C).str();
}
+bool isAllocatableType(QualType QT) {
+ if (QT->isPointerType())
+ return true;
+ QT = QT.getCanonicalType().getUnqualifiedType();
+ if (const CXXRecordDecl *RD = QT->getAsCXXRecordDecl()) {
+ if (isa<ClassTemplateSpecializationDecl>(RD)) {
+ std::string QName = RD->getQualifiedNameAsString();
+ return (QName == "std::vector" || QName == "vector" ||
+ QName == "std::unique_ptr" || QName == "unique_ptr" ||
+ QName == "std::shared_ptr" || QName == "shared_ptr" ||
+ QName == "llvm::SmallVector" || QName == "SmallVector");
+ }
+ }
+ return false;
+}
+
DSAStackTy::DSAVarData DSAStackTy::getDSA(const_iterator &Iter,
ValueDecl *D) const {
D = getCanonicalDecl(D);
@@ -1370,20 +1386,19 @@ DSAStackTy::DSAVarData DSAStackTy::getDSA(const_iterator &Iter,
DefaultDataSharingAttributes IterDA = Iter->DefaultAttr;
switch (Iter->DefaultVCAttr) {
case DSA_VC_aggregate:
- if (!VD->getType()->isAggregateType())
+ if (!D->getType()->isAggregateType())
IterDA = DSA_none;
break;
case DSA_VC_allocatable:
- if (!(VD->getType()->isPointerType() ||
- VD->getType()->isVariableArrayType()))
+ if (!isAllocatableType(D->getType()))
IterDA = DSA_none;
break;
case DSA_VC_pointer:
- if (!VD->getType()->isPointerType())
+ if (!D->getType()->isPointerType())
IterDA = DSA_none;
break;
case DSA_VC_scalar:
- if (!VD->getType()->isScalarType())
+ if (!D->getType()->isScalarType())
IterDA = DSA_none;
break;
case DSA_VC_all:
diff --git a/clang/test/OpenMP/parallel_default_variableCategory_codegen.cpp b/clang/test/OpenMP/parallel_default_variableCategory_codegen.cpp
new file mode 100644
index 0000000000000..b0674158f57e5
--- /dev/null
+++ b/clang/test/OpenMP/parallel_default_variableCategory_codegen.cpp
@@ -0,0 +1,117 @@
+// RUN: %clangxx -DOMP60 -Xclang -verify -Wno-vla -fopenmp -fopenmp-version=60 -x c++ -S -emit-llvm %s -o - | FileCheck --check-prefixes=OMP60 %s
+// expected-no-diagnostics
+#ifndef HEADER
+#define HEADER
+
+#include <vector>
+
+int global;
+#define VECTOR_SIZE 4
+int main (int argc, char **argv) {
+ int i,n;
+ int x;
+
+ n = VECTOR_SIZE;
+
+ #pragma omp parallel masked firstprivate(x) num_threads(2)
+ {
+ int *xPtr = nullptr;
+ // scalar
+ #pragma omp task default(shared:scalar)
+ {
+ xPtr = &x;
+ }
+ #pragma omp taskwait
+
+ // pointer
+ #pragma omp task default(shared:pointer) shared(x)
+ {
+ xPtr = &x;
+ }
+ #pragma omp taskwait
+ }
+
+ int *aggregate[VECTOR_SIZE] = {0,0,0,0};
+ std::vector<int *> arr(VECTOR_SIZE,0);
+
+ #pragma omp parallel masked num_threads(2)
+ {
+ // aggregate
+ #pragma omp task default(shared:aggregate)
+ for(i=0;i<n;i++) {
+ aggregate[i] = &x;
+ }
+ #pragma omp taskwait
+
+ #pragma omp task default(shared:aggregate) shared(x)
+ for(i=0;i<n;i++) {
+ aggregate[i] = &x;
+ }
+ #pragma omp taskwait
+
+ // allocatable
+ #pragma omp task default(shared:allocatable)
+ for(i=0;i<n;i++) {
+ arr[i] = &x;
+ }
+ #pragma omp taskwait
+
+ #pragma omp task default(shared:allocatable) shared(x)
+ for(i=0;i<n;i++) {
+ arr[i] = &x;
+ }
+ #pragma omp taskwait
+
+ // all
+ #pragma omp task default(shared:all)
+ for(i=0;i<n;i++) {
+ aggregate[i] = &x;
+ }
+ #pragma omp taskwait
+ }
+}
+
+#endif
+
+// OMP60-LABEL: define {{.*}}main.omp_outlined{{.*}}
+// OMP60-NEXT: entry:
+// OMP60-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 8
+// OMP60-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 8
+// OMP60: [[X_ADDR:%.*]] = alloca{{.*}}
+// OMP60: [[xPTR:%.*]] = alloca{{.*}}
+// OMP60: store ptr null, ptr [[xPTR]]{{.*}}
+// OMP60: store ptr [[xPTR]]{{.*}}
+// OMP60: store ptr [[X_ADDR]]{{.*}}
+// OMP60-NEXT: {{.*}}call{{.*}}__kmpc_omp_task_alloc{{.*}}
+// OMP60: ret void
+//
+// OMP60: define {{.*}}main.omp_outlined{{.*}}
+// OMP60-NEXT: entry:
+// OMP60-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 8
+// OMP60-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 8
+// OMP60: [[I_ADDR:%.*]] = alloca{{.*}}
+// OMP60-NEXT: [[N_ADDR:%.*]] = alloca{{.*}}
+// OMP60-NEXT: [[AGGREGATE_ADDR:%.*]] = alloca{{.*}}
+// OMP60-NEXT: [[X_ADDR:%.*]] = alloca{{.*}}
+// OMP60-NEXT: [[ARR_ADDR:%.*]] = alloca{{.*}}
+// OMP60: [[TMP0:%.*]] = load{{.*}}[[I_ADDR]]
+// OMP60-NEXT: [[TMP1:%.*]] = load{{.*}}[[N_ADDR]]
+// OMP60-NEXT: [[TMP2:%.*]] = load{{.*}}[[AGGREGATE_ADDR]]
+// OMP60-NEXT: [[TMP3:%.*]] = load{{.*}}[[X_ADDR]]
+// OMP60-NEXT: [[TMP4:%.*]] = load{{.*}}[[ARR_ADDR]]
+// OMP60: store ptr [[TMP2]]{{.*}}
+// OMP60-NEXT: {{.*}}call{{.*}}__kmpc_omp_task_alloc{{.*}}
+// OMP60: store ptr [[TMP2]]{{.*}}
+// OMP60: store ptr [[TMP3]]{{.*}}
+// OMP60-NEXT: {{.*}}call{{.*}}__kmpc_omp_task_alloc{{.*}}
+// OMP60: store ptr [[TMP4]]{{.*}}
+// OMP60-NEXT: {{.*}}call{{.*}}__kmpc_omp_task_alloc{{.*}}
+// OMP60: store ptr [[TMP4]]{{.*}}
+// OMP60: store ptr [[TMP3]]{{.*}}
+// OMP60-NEXT: {{.*}}call{{.*}}__kmpc_omp_task_alloc{{.*}}
+// OMP60: store ptr [[TMP0]]{{.*}}
+// OMP60: store ptr [[TMP1]]{{.*}}
+// OMP60: store ptr [[TMP2]]{{.*}}
+// OMP60: store ptr [[TMP3]]{{.*}}
+// OMP60-NEXT: {{.*}}call{{.*}}__kmpc_omp_task_alloc{{.*}}
+// OMP60: ret void
|
clang/lib/Sema/SemaOpenMP.cpp
Outdated
Comment on lines
1317
to
1331
| bool isAllocatableType(QualType QT) { | ||
| if (QT->isPointerType()) | ||
| return true; | ||
| QT = QT.getCanonicalType().getUnqualifiedType(); | ||
| if (const CXXRecordDecl *RD = QT->getAsCXXRecordDecl()) { | ||
| if (isa<ClassTemplateSpecializationDecl>(RD)) { | ||
| std::string QName = RD->getQualifiedNameAsString(); | ||
| return (QName == "std::vector" || QName == "vector" || | ||
| QName == "std::unique_ptr" || QName == "unique_ptr" || | ||
| QName == "std::shared_ptr" || QName == "shared_ptr" || | ||
| QName == "llvm::SmallVector" || QName == "SmallVector"); | ||
| } | ||
| } | ||
| return false; | ||
| } |
Member
There was a problem hiding this comment.
Poor check, it should not check the names, should check the members
added 6 commits
November 5, 2025 23:41
…allel_default_variableCategory_codegen.cpp. 2) Taking care of the feedback (from Alexey) in SemaOpenMP.cpp.
clang/test/OpenMP/parallel_default_variableCategory_codegen.cpp
alexey-bataev
requested changes
Nov 11, 2025
clang/lib/Sema/SemaOpenMP.cpp
Outdated
Comment on lines
1329
to
1336
| if (NS->isStdNamespace()) | ||
| if (Name == "vector" || Name == "unique_ptr" || | ||
| Name == "shared_ptr" || Name == "array") | ||
| return true; | ||
|
|
||
| if (NS->getIdentifier()->getName() == "llvm") | ||
| if (Name == "SmallVector") | ||
| return true; |
Member
There was a problem hiding this comment.
There should not be anything like this. Please, remove!
Contributor
Author
There was a problem hiding this comment.
A similar kind of code exists in file clang/lib/StaticAnalyzer/Checkers/SmartPtrModeling.cpp, function isStdSmartPtr( ). Without this I cannot check if the variable is of type Allocatable.
Member
There was a problem hiding this comment.
Analayser is the analysis tool, not a compiler, it may produce wrong results.
Contributor
Author
There was a problem hiding this comment.
Code related to allocatable variable category is now removed by checkin #167735 of David Pagan. These changes have been pulled into the current bug fix.
alexey-bataev
approved these changes
Nov 14, 2025
Collaborator
|
LLVM Buildbot has detected a new failure on builder Full details are available at: https://lab.llvm.org/buildbot/#/builders/144/builds/40161 Here is the relevant piece of the build log for the reference |
Collaborator
|
LLVM Buildbot has detected a new failure on builder Full details are available at: https://lab.llvm.org/buildbot/#/builders/46/builds/26374 Here is the relevant piece of the build log for the reference |
SunilKuravinakop
added a commit
to SunilKuravinakop/llvm-project
that referenced
this pull request
Nov 14, 2025
Collaborator
|
LLVM Buildbot has detected a new failure on builder Full details are available at: https://lab.llvm.org/buildbot/#/builders/11/builds/28052 Here is the relevant piece of the build log for the reference |
llvm-sync bot
pushed a commit
to arm/arm-toolchain
that referenced
this pull request
Nov 14, 2025
…tegory" (#168083) Reverts llvm/llvm-project#165276 The newly added test failed on a number of buildbots.
SunilKuravinakop
pushed a commit
to SunilKuravinakop/llvm-project
that referenced
this pull request
Nov 14, 2025
… test to restore Ubuntu build. "allocate". This is not needed as allocatable modifier is not applicable to the default clause in C/C++.
chichunchen
pushed a commit
that referenced
this pull request
Nov 14, 2025
Same changes as in fix for [165276](#165276) except for remove unnecessary <vector> include in test to restore Ubuntu build. This is not needed as allocatable modifier is not applicable to the default clause in C/C++. Co-authored-by: Sunil Kuravinakop <[email protected]>
llvm-sync bot
pushed a commit
to arm/arm-toolchain
that referenced
this pull request
Nov 14, 2025
…#168112) Same changes as in fix for [165276](llvm/llvm-project#165276) except for remove unnecessary <vector> include in test to restore Ubuntu build. This is not needed as allocatable modifier is not applicable to the default clause in C/C++. Co-authored-by: Sunil Kuravinakop <[email protected]>
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
5 participants
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
In the default clause taking care of new comments in the previous "Support for Default clause variable category" 157063 and adding a new test case.