[OpenMP] Change __kmpc_free_shared to include the paired allocation…

… size This patch changes `__kmpc_free_shared` to take an additional argument corresponding to the associated allocation's size. This makes it easier to implement the allocator in the runtime. Reviewed By: jdoerfert Differential Revision: https://reviews.llvm.org/D106496
llvm · Jul 22, 2021 · 754eb1c · 754eb1c
1 parent 402b681
commit 754eb1c
Show file tree

Hide file tree

Showing 23 changed files with 100 additions and 90 deletions.
diff --git a/clang/lib/CodeGen/CGOpenMPRuntimeGPU.cpp b/clang/lib/CodeGen/CGOpenMPRuntimeGPU.cpp
@@ -1437,7 +1437,9 @@ void CGOpenMPRuntimeGPU::emitGenericVarsProlog(CodeGenFunction &CGF,
                                 CGM.getModule(), OMPRTL___kmpc_alloc_shared),
                             AllocArgs, VD->getName());
 
-    I->getSecond().EscapedVariableLengthDeclsAddrs.emplace_back(VoidPtr);
+    I->getSecond().EscapedVariableLengthDeclsAddrs.emplace_back(
+        std::pair<llvm::Value *, llvm::Value *>(
+            {VoidPtr, CGF.getTypeSize(VD->getType())}));
     LValue Base = CGF.MakeAddrLValue(VoidPtr, VD->getType(),
                                      CGM.getContext().getDeclAlign(VD),
                                      AlignmentSource::Decl);
@@ -1456,19 +1458,22 @@ void CGOpenMPRuntimeGPU::emitGenericVarsEpilog(CodeGenFunction &CGF,
   const auto I = FunctionGlobalizedDecls.find(CGF.CurFn);
   if (I != FunctionGlobalizedDecls.end()) {
     // Deallocate the memory for each globalized VLA object
-    for (llvm::Value *Addr :
+    for (auto AddrSizePair :
          llvm::reverse(I->getSecond().EscapedVariableLengthDeclsAddrs)) {
       CGF.EmitRuntimeCall(OMPBuilder.getOrCreateRuntimeFunction(
                               CGM.getModule(), OMPRTL___kmpc_free_shared),
-                          Addr);
+                          {AddrSizePair.first, AddrSizePair.second});
     }
     // Deallocate the memory for each globalized value
     for (auto &Rec : llvm::reverse(I->getSecond().LocalVarData)) {
+      const auto *VD = cast<VarDecl>(Rec.first);
       I->getSecond().MappedParams->restore(CGF);
 
+      llvm::Value *FreeArgs[] = {Rec.second.GlobalizedVal,
+                                 CGF.getTypeSize(VD->getType())};
       CGF.EmitRuntimeCall(OMPBuilder.getOrCreateRuntimeFunction(
                               CGM.getModule(), OMPRTL___kmpc_free_shared),
-                          {Rec.second.GlobalizedVal});
+                          FreeArgs);
     }
   }
 }

diff --git a/clang/lib/CodeGen/CGOpenMPRuntimeGPU.h b/clang/lib/CodeGen/CGOpenMPRuntimeGPU.h
@@ -423,7 +423,8 @@ class CGOpenMPRuntimeGPU : public CGOpenMPRuntime {
     llvm::Optional<DeclToAddrMapTy> SecondaryLocalVarData = llvm::None;
     EscapedParamsTy EscapedParameters;
     llvm::SmallVector<const ValueDecl*, 4> EscapedVariableLengthDecls;
-    llvm::SmallVector<llvm::Value *, 4> EscapedVariableLengthDeclsAddrs;
+    llvm::SmallVector<std::pair<llvm::Value *, llvm::Value *>, 4>
+        EscapedVariableLengthDeclsAddrs;
     llvm::Value *IsInSPMDModeFlag = nullptr;
     std::unique_ptr<CodeGenFunction::OMPMapVars> MappedParams;
   };

diff --git a/clang/test/OpenMP/declare_target_codegen_globalization.cpp b/clang/test/OpenMP/declare_target_codegen_globalization.cpp
@@ -81,6 +81,6 @@ int maini1() {
 // CHECK1-NEXT:    [[A:%.*]] = call i8* @__kmpc_alloc_shared(i64 4)
 // CHECK1-NEXT:    [[A_ON_STACK:%.*]] = bitcast i8* [[A]] to i32*
 // CHECK1-NEXT:    [[CALL:%.*]] = call i32 @_Z3fooRi(i32* nonnull align 4 dereferenceable(4) [[A_ON_STACK]]) #[[ATTR3]]
-// CHECK1-NEXT:    call void @__kmpc_free_shared(i8* [[A]])
+// CHECK1-NEXT:    call void @__kmpc_free_shared(i8* [[A]], i64 4)
 // CHECK1-NEXT:    ret i32 [[CALL]]
 //
diff --git a/clang/test/OpenMP/nvptx_data_sharing.cpp b/clang/test/OpenMP/nvptx_data_sharing.cpp
@@ -418,8 +418,8 @@ void test_ds(){
 // CHECK-NEXT:    store i8* [[TMP8]], i8** [[TMP7]], align 8
 // CHECK-NEXT:    [[TMP9:%.*]] = bitcast [2 x i8*]* [[CAPTURED_VARS_ADDRS1]] to i8**
 // CHECK-NEXT:    call void @__kmpc_parallel_51(%struct.ident_t* @[[GLOB1]], i32 [[TMP1]], i32 1, i32 -1, i32 -1, i8* bitcast (void (i32*, i32*, i32*, i32*)* @__omp_outlined__1 to i8*), i8* bitcast (void (i16, i32)* @__omp_outlined__1_wrapper to i8*), i8** [[TMP9]], i64 2)
-// CHECK-NEXT:    call void @__kmpc_free_shared(i8* [[B]])
-// CHECK-NEXT:    call void @__kmpc_free_shared(i8* [[A]])
+// CHECK-NEXT:    call void @__kmpc_free_shared(i8* [[B]], i64 4)
+// CHECK-NEXT:    call void @__kmpc_free_shared(i8* [[A]], i64 4)
 // CHECK-NEXT:    call void @__kmpc_target_deinit(%struct.ident_t* @[[GLOB1]], i1 false, i1 true)
 // CHECK-NEXT:    ret void
 // CHECK:       worker.exit:

diff --git a/clang/test/OpenMP/nvptx_distribute_parallel_generic_mode_codegen.cpp b/clang/test/OpenMP/nvptx_distribute_parallel_generic_mode_codegen.cpp
@@ -3174,7 +3174,7 @@ int main(int argc, char **argv) {
 // CHECK4:       .omp.lastprivate.done:
 // CHECK4-NEXT:    br label [[OMP_PRECOND_END]]
 // CHECK4:       omp.precond.end:
-// CHECK4-NEXT:    call void @__kmpc_free_shared(i8* [[C1]])
+// CHECK4-NEXT:    call void @__kmpc_free_shared(i8* [[C1]], i64 40)
 // CHECK4-NEXT:    ret void
 //
 //
@@ -3503,7 +3503,7 @@ int main(int argc, char **argv) {
 // CHECK5:       .omp.lastprivate.done:
 // CHECK5-NEXT:    br label [[OMP_PRECOND_END]]
 // CHECK5:       omp.precond.end:
-// CHECK5-NEXT:    call void @__kmpc_free_shared(i8* [[C1]])
+// CHECK5-NEXT:    call void @__kmpc_free_shared(i8* [[C1]], i32 40)
 // CHECK5-NEXT:    ret void
 //
 //
@@ -3826,7 +3826,7 @@ int main(int argc, char **argv) {
 // CHECK6:       .omp.lastprivate.done:
 // CHECK6-NEXT:    br label [[OMP_PRECOND_END]]
 // CHECK6:       omp.precond.end:
-// CHECK6-NEXT:    call void @__kmpc_free_shared(i8* [[C1]])
+// CHECK6-NEXT:    call void @__kmpc_free_shared(i8* [[C1]], i32 40)
 // CHECK6-NEXT:    ret void
 //
 //

diff --git a/clang/test/OpenMP/nvptx_parallel_codegen.cpp b/clang/test/OpenMP/nvptx_parallel_codegen.cpp
@@ -1645,7 +1645,7 @@ int bar(int n){
 // CHECK1-NEXT:    [[TMP6:%.*]] = load i32, i32* [[A_ON_STACK]], align 4
 // CHECK1-NEXT:    [[INC:%.*]] = add nsw i32 [[TMP6]], 1
 // CHECK1-NEXT:    store i32 [[INC]], i32* [[A_ON_STACK]], align 4
-// CHECK1-NEXT:    call void @__kmpc_free_shared(i8* [[A1]])
+// CHECK1-NEXT:    call void @__kmpc_free_shared(i8* [[A1]], i64 4)
 // CHECK1-NEXT:    call void @__kmpc_target_deinit(%struct.ident_t* @[[GLOB1]], i1 false, i1 true)
 // CHECK1-NEXT:    ret void
 // CHECK1:       worker.exit:
@@ -1917,7 +1917,7 @@ int bar(int n){
 // CHECK2-NEXT:    [[TMP6:%.*]] = load i32, i32* [[A_ON_STACK]], align 4
 // CHECK2-NEXT:    [[INC:%.*]] = add nsw i32 [[TMP6]], 1
 // CHECK2-NEXT:    store i32 [[INC]], i32* [[A_ON_STACK]], align 4
-// CHECK2-NEXT:    call void @__kmpc_free_shared(i8* [[A1]])
+// CHECK2-NEXT:    call void @__kmpc_free_shared(i8* [[A1]], i32 4)
 // CHECK2-NEXT:    call void @__kmpc_target_deinit(%struct.ident_t* @[[GLOB1]], i1 false, i1 true)
 // CHECK2-NEXT:    ret void
 // CHECK2:       worker.exit:

diff --git a/clang/test/OpenMP/nvptx_parallel_for_codegen.cpp b/clang/test/OpenMP/nvptx_parallel_for_codegen.cpp
@@ -486,7 +486,7 @@ int bar(int n){
 // CHECK-NEXT:    [[TMP9:%.*]] = load i32, i32* [[ARRAYIDX]], align 4
 // CHECK-NEXT:    [[ADD:%.*]] = add nsw i32 [[TMP9]], 1
 // CHECK-NEXT:    store i32 [[ADD]], i32* [[ARRAYIDX]], align 4
-// CHECK-NEXT:    call void @__kmpc_free_shared(i8* [[D]])
+// CHECK-NEXT:    call void @__kmpc_free_shared(i8* [[D]], i64 4)
 // CHECK-NEXT:    call void @__kmpc_target_deinit(%struct.ident_t* @[[GLOB1]], i1 false, i1 true)
 // CHECK-NEXT:    ret void
 // CHECK:       worker.exit:

diff --git a/clang/test/OpenMP/nvptx_target_codegen.cpp b/clang/test/OpenMP/nvptx_target_codegen.cpp
@@ -432,7 +432,7 @@ void unreachable_call() {
 // CHECK1-NEXT:    [[TMP6:%.*]] = bitcast [2 x i8*]* [[CAPTURED_VARS_ADDRS]] to i8**
 // CHECK1-NEXT:    call void @__kmpc_parallel_51(%struct.ident_t* @[[GLOB3]], i32 [[TMP0]], i32 1, i32 -1, i32 -1, i8* bitcast (void (i32*, i32*, i32*, double*)* @__omp_outlined__1 to i8*), i8* bitcast (void (i16, i32)* @__omp_outlined__1_wrapper to i8*), i8** [[TMP6]], i64 2)
 // CHECK1-NEXT:    [[TMP7:%.*]] = load i32, i32* [[F_ON_STACK]], align 4
-// CHECK1-NEXT:    call void @__kmpc_free_shared(i8* [[F]])
+// CHECK1-NEXT:    call void @__kmpc_free_shared(i8* [[F]], i64 4)
 // CHECK1-NEXT:    ret i32 [[TMP7]]
 //
 //
@@ -816,7 +816,7 @@ void unreachable_call() {
 // CHECK2-NEXT:    [[TMP6:%.*]] = bitcast [2 x i8*]* [[CAPTURED_VARS_ADDRS]] to i8**
 // CHECK2-NEXT:    call void @__kmpc_parallel_51(%struct.ident_t* @[[GLOB3]], i32 [[TMP0]], i32 1, i32 -1, i32 -1, i8* bitcast (void (i32*, i32*, i32*, double*)* @__omp_outlined__1 to i8*), i8* bitcast (void (i16, i32)* @__omp_outlined__1_wrapper to i8*), i8** [[TMP6]], i32 2)
 // CHECK2-NEXT:    [[TMP7:%.*]] = load i32, i32* [[F_ON_STACK]], align 4
-// CHECK2-NEXT:    call void @__kmpc_free_shared(i8* [[F]])
+// CHECK2-NEXT:    call void @__kmpc_free_shared(i8* [[F]], i32 4)
 // CHECK2-NEXT:    ret i32 [[TMP7]]
 //
 //
@@ -1199,7 +1199,7 @@ void unreachable_call() {
 // CHECK3-NEXT:    [[TMP6:%.*]] = bitcast [2 x i8*]* [[CAPTURED_VARS_ADDRS]] to i8**
 // CHECK3-NEXT:    call void @__kmpc_parallel_51(%struct.ident_t* @[[GLOB3]], i32 [[TMP0]], i32 1, i32 -1, i32 -1, i8* bitcast (void (i32*, i32*, i32*, double*)* @__omp_outlined__1 to i8*), i8* bitcast (void (i16, i32)* @__omp_outlined__1_wrapper to i8*), i8** [[TMP6]], i32 2)
 // CHECK3-NEXT:    [[TMP7:%.*]] = load i32, i32* [[F_ON_STACK]], align 4
-// CHECK3-NEXT:    call void @__kmpc_free_shared(i8* [[F]])
+// CHECK3-NEXT:    call void @__kmpc_free_shared(i8* [[F]], i32 4)
 // CHECK3-NEXT:    ret i32 [[TMP7]]
 //
 //

diff --git a/clang/test/OpenMP/nvptx_target_parallel_reduction_codegen_tbaa_PR46146.cpp b/clang/test/OpenMP/nvptx_target_parallel_reduction_codegen_tbaa_PR46146.cpp
@@ -173,9 +173,9 @@ void test() {
 // CHECK1-NEXT:    call void @llvm.lifetime.end.p0i8(i64 4, i8* [[TMP32]]) #[[ATTR1]]
 // CHECK1-NEXT:    [[TMP33:%.*]] = bitcast i32* [[DOTOMP_IV]] to i8*
 // CHECK1-NEXT:    call void @llvm.lifetime.end.p0i8(i64 4, i8* [[TMP33]]) #[[ATTR1]]
-// CHECK1-NEXT:    call void @__kmpc_free_shared(i8* [[PARTIAL_SUM]])
-// CHECK1-NEXT:    call void @__kmpc_free_shared(i8* [[IEND]])
-// CHECK1-NEXT:    call void @__kmpc_free_shared(i8* [[ISTART]])
+// CHECK1-NEXT:    call void @__kmpc_free_shared(i8* [[PARTIAL_SUM]], i64 8)
+// CHECK1-NEXT:    call void @__kmpc_free_shared(i8* [[IEND]], i64 4)
+// CHECK1-NEXT:    call void @__kmpc_free_shared(i8* [[ISTART]], i64 4)
 // CHECK1-NEXT:    ret void
 //
 //
@@ -740,9 +740,9 @@ void test() {
 // CHECK1-NEXT:    call void @llvm.lifetime.end.p0i8(i64 4, i8* [[TMP32]]) #[[ATTR1]]
 // CHECK1-NEXT:    [[TMP33:%.*]] = bitcast i32* [[DOTOMP_IV]] to i8*
 // CHECK1-NEXT:    call void @llvm.lifetime.end.p0i8(i64 4, i8* [[TMP33]]) #[[ATTR1]]
-// CHECK1-NEXT:    call void @__kmpc_free_shared(i8* [[PARTIAL_SUM]])
-// CHECK1-NEXT:    call void @__kmpc_free_shared(i8* [[IEND]])
-// CHECK1-NEXT:    call void @__kmpc_free_shared(i8* [[ISTART]])
+// CHECK1-NEXT:    call void @__kmpc_free_shared(i8* [[PARTIAL_SUM]], i64 16)
+// CHECK1-NEXT:    call void @__kmpc_free_shared(i8* [[IEND]], i64 4)
+// CHECK1-NEXT:    call void @__kmpc_free_shared(i8* [[ISTART]], i64 4)
 // CHECK1-NEXT:    ret void
 //
 //
@@ -1407,9 +1407,9 @@ void test() {
 // CHECK2-NEXT:    call void @llvm.lifetime.end.p0i8(i64 4, i8* [[TMP32]]) #[[ATTR1]]
 // CHECK2-NEXT:    [[TMP33:%.*]] = bitcast i32* [[DOTOMP_IV]] to i8*
 // CHECK2-NEXT:    call void @llvm.lifetime.end.p0i8(i64 4, i8* [[TMP33]]) #[[ATTR1]]
-// CHECK2-NEXT:    call void @__kmpc_free_shared(i8* [[PARTIAL_SUM]])
-// CHECK2-NEXT:    call void @__kmpc_free_shared(i8* [[IEND]])
-// CHECK2-NEXT:    call void @__kmpc_free_shared(i8* [[ISTART]])
+// CHECK2-NEXT:    call void @__kmpc_free_shared(i8* [[PARTIAL_SUM]], i64 8)
+// CHECK2-NEXT:    call void @__kmpc_free_shared(i8* [[IEND]], i64 4)
+// CHECK2-NEXT:    call void @__kmpc_free_shared(i8* [[ISTART]], i64 4)
 // CHECK2-NEXT:    ret void
 //
 //
@@ -1974,9 +1974,9 @@ void test() {
 // CHECK2-NEXT:    call void @llvm.lifetime.end.p0i8(i64 4, i8* [[TMP32]]) #[[ATTR1]]
 // CHECK2-NEXT:    [[TMP33:%.*]] = bitcast i32* [[DOTOMP_IV]] to i8*
 // CHECK2-NEXT:    call void @llvm.lifetime.end.p0i8(i64 4, i8* [[TMP33]]) #[[ATTR1]]
-// CHECK2-NEXT:    call void @__kmpc_free_shared(i8* [[PARTIAL_SUM]])
-// CHECK2-NEXT:    call void @__kmpc_free_shared(i8* [[IEND]])
-// CHECK2-NEXT:    call void @__kmpc_free_shared(i8* [[ISTART]])
+// CHECK2-NEXT:    call void @__kmpc_free_shared(i8* [[PARTIAL_SUM]], i64 16)
+// CHECK2-NEXT:    call void @__kmpc_free_shared(i8* [[IEND]], i64 4)
+// CHECK2-NEXT:    call void @__kmpc_free_shared(i8* [[ISTART]], i64 4)
 // CHECK2-NEXT:    ret void
 //
 //
@@ -2641,9 +2641,9 @@ void test() {
 // CHECK3-NEXT:    call void @llvm.lifetime.end.p0i8(i64 4, i8* [[TMP32]]) #[[ATTR1]]
 // CHECK3-NEXT:    [[TMP33:%.*]] = bitcast i32* [[DOTOMP_IV]] to i8*
 // CHECK3-NEXT:    call void @llvm.lifetime.end.p0i8(i64 4, i8* [[TMP33]]) #[[ATTR1]]
-// CHECK3-NEXT:    call void @__kmpc_free_shared(i8* [[PARTIAL_SUM]])
-// CHECK3-NEXT:    call void @__kmpc_free_shared(i8* [[IEND]])
-// CHECK3-NEXT:    call void @__kmpc_free_shared(i8* [[ISTART]])
+// CHECK3-NEXT:    call void @__kmpc_free_shared(i8* [[PARTIAL_SUM]], i64 8)
+// CHECK3-NEXT:    call void @__kmpc_free_shared(i8* [[IEND]], i64 4)
+// CHECK3-NEXT:    call void @__kmpc_free_shared(i8* [[ISTART]], i64 4)
 // CHECK3-NEXT:    ret void
 //
 //
@@ -3208,9 +3208,9 @@ void test() {
 // CHECK3-NEXT:    call void @llvm.lifetime.end.p0i8(i64 4, i8* [[TMP32]]) #[[ATTR1]]
 // CHECK3-NEXT:    [[TMP33:%.*]] = bitcast i32* [[DOTOMP_IV]] to i8*
 // CHECK3-NEXT:    call void @llvm.lifetime.end.p0i8(i64 4, i8* [[TMP33]]) #[[ATTR1]]
-// CHECK3-NEXT:    call void @__kmpc_free_shared(i8* [[PARTIAL_SUM]])
-// CHECK3-NEXT:    call void @__kmpc_free_shared(i8* [[IEND]])
-// CHECK3-NEXT:    call void @__kmpc_free_shared(i8* [[ISTART]])
+// CHECK3-NEXT:    call void @__kmpc_free_shared(i8* [[PARTIAL_SUM]], i64 16)
+// CHECK3-NEXT:    call void @__kmpc_free_shared(i8* [[IEND]], i64 4)
+// CHECK3-NEXT:    call void @__kmpc_free_shared(i8* [[ISTART]], i64 4)
 // CHECK3-NEXT:    ret void
 //
 //

diff --git a/clang/test/OpenMP/nvptx_target_teams_distribute_codegen.cpp b/clang/test/OpenMP/nvptx_target_teams_distribute_codegen.cpp
@@ -657,7 +657,7 @@ int bar(int n){
 // CHECK1-NEXT:    br label [[OMP_LOOP_EXIT:%.*]]
 // CHECK1:       omp.loop.exit:
 // CHECK1-NEXT:    call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB2]], i32 [[TMP1]])
-// CHECK1-NEXT:    call void @__kmpc_free_shared(i8* [[I]])
+// CHECK1-NEXT:    call void @__kmpc_free_shared(i8* [[I]], i64 4)
 // CHECK1-NEXT:    ret void
 //
 //
@@ -779,7 +779,7 @@ int bar(int n){
 // CHECK2-NEXT:    br label [[OMP_LOOP_EXIT:%.*]]
 // CHECK2:       omp.loop.exit:
 // CHECK2-NEXT:    call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB2]], i32 [[TMP1]])
-// CHECK2-NEXT:    call void @__kmpc_free_shared(i8* [[I]])
+// CHECK2-NEXT:    call void @__kmpc_free_shared(i8* [[I]], i32 4)
 // CHECK2-NEXT:    ret void
 //
 //
@@ -901,7 +901,7 @@ int bar(int n){
 // CHECK3-NEXT:    br label [[OMP_LOOP_EXIT:%.*]]
 // CHECK3:       omp.loop.exit:
 // CHECK3-NEXT:    call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB2]], i32 [[TMP1]])
-// CHECK3-NEXT:    call void @__kmpc_free_shared(i8* [[I]])
+// CHECK3-NEXT:    call void @__kmpc_free_shared(i8* [[I]], i32 4)
 // CHECK3-NEXT:    ret void
 //
 //

diff --git a/clang/test/OpenMP/nvptx_target_teams_distribute_parallel_for_codegen.cpp b/clang/test/OpenMP/nvptx_target_teams_distribute_parallel_for_codegen.cpp
@@ -18656,7 +18656,7 @@ int bar(int n){
 // CHECK1:       .omp.lastprivate.done:
 // CHECK1-NEXT:    br label [[OMP_PRECOND_END]]
 // CHECK1:       omp.precond.end:
-// CHECK1-NEXT:    call void @__kmpc_free_shared(i8* [[L2]])
+// CHECK1-NEXT:    call void @__kmpc_free_shared(i8* [[L2]], i64 4)
 // CHECK1-NEXT:    ret void
 //
 //
@@ -20250,7 +20250,7 @@ int bar(int n){
 // CHECK2:       .omp.lastprivate.done:
 // CHECK2-NEXT:    br label [[OMP_PRECOND_END]]
 // CHECK2:       omp.precond.end:
-// CHECK2-NEXT:    call void @__kmpc_free_shared(i8* [[L2]])
+// CHECK2-NEXT:    call void @__kmpc_free_shared(i8* [[L2]], i64 4)
 // CHECK2-NEXT:    ret void
 //
 //
@@ -21829,7 +21829,7 @@ int bar(int n){
 // CHECK3:       .omp.lastprivate.done:
 // CHECK3-NEXT:    br label [[OMP_PRECOND_END]]
 // CHECK3:       omp.precond.end:
-// CHECK3-NEXT:    call void @__kmpc_free_shared(i8* [[L1]])
+// CHECK3-NEXT:    call void @__kmpc_free_shared(i8* [[L1]], i32 4)
 // CHECK3-NEXT:    ret void
 //
 //
@@ -23363,7 +23363,7 @@ int bar(int n){
 // CHECK4:       .omp.lastprivate.done:
 // CHECK4-NEXT:    br label [[OMP_PRECOND_END]]
 // CHECK4:       omp.precond.end:
-// CHECK4-NEXT:    call void @__kmpc_free_shared(i8* [[L1]])
+// CHECK4-NEXT:    call void @__kmpc_free_shared(i8* [[L1]], i32 4)
 // CHECK4-NEXT:    ret void
 //
 //

diff --git a/clang/test/OpenMP/nvptx_target_teams_distribute_parallel_for_simd_codegen.cpp b/clang/test/OpenMP/nvptx_target_teams_distribute_parallel_for_simd_codegen.cpp
@@ -9581,7 +9581,7 @@ int bar(int n){
 // CHECK1:       .omp.lastprivate.done:
 // CHECK1-NEXT:    br label [[OMP_PRECOND_END]]
 // CHECK1:       omp.precond.end:
-// CHECK1-NEXT:    call void @__kmpc_free_shared(i8* [[L2]])
+// CHECK1-NEXT:    call void @__kmpc_free_shared(i8* [[L2]], i64 4)
 // CHECK1-NEXT:    ret void
 //
 //
@@ -10655,7 +10655,7 @@ int bar(int n){
 // CHECK2:       .omp.lastprivate.done:
 // CHECK2-NEXT:    br label [[OMP_PRECOND_END]]
 // CHECK2:       omp.precond.end:
-// CHECK2-NEXT:    call void @__kmpc_free_shared(i8* [[L1]])
+// CHECK2-NEXT:    call void @__kmpc_free_shared(i8* [[L1]], i32 4)
 // CHECK2-NEXT:    ret void
 //
 //
@@ -11693,7 +11693,7 @@ int bar(int n){
 // CHECK3:       .omp.lastprivate.done:
 // CHECK3-NEXT:    br label [[OMP_PRECOND_END]]
 // CHECK3:       omp.precond.end:
-// CHECK3-NEXT:    call void @__kmpc_free_shared(i8* [[L1]])
+// CHECK3-NEXT:    call void @__kmpc_free_shared(i8* [[L1]], i32 4)
 // CHECK3-NEXT:    ret void
 //
 //