[Clang][OpenMP][OMPBuilder] (2/4) Privatize `parallel` for `OMPBuilder`

[fghanim]

Summary

adding lit-test for usage of 'private' in the OMPBuilder implementation
of omp parallel

note: this was originally a part of D79677

clang/test/OpenMP/parallel_private_codegen.cpp

-// RUN: %clang_cc1 -verify -fopenmp -x c++ -triple x86_64-unknown-unknown -emit-llvm %s -o - | FileCheck %s
+// RUN: %clang_cc1 -verify -fopenmp -x c++ -triple x86_64-unknown-unknown -emit-llvm %s -o - | FileCheck %s -check-prefixes=ALL,CHECK
 // RUN: %clang_cc1 -fopenmp -x c++ -std=c++11 -triple x86_64-unknown-unknown -emit-pch -o %t %s
-// RUN: %clang_cc1 -fopenmp -x c++ -triple x86_64-unknown-unknown -std=c++11 -include-pch %t -verify %s -emit-llvm -o - | FileCheck %s
+// RUN: %clang_cc1 -fopenmp -x c++ -triple x86_64-unknown-unknown -std=c++11 -include-pch %t -verify %s -emit-llvm -o - | FileCheck %s -check-prefixes=ALL,CHECK
 // RUN: %clang_cc1 -verify -fopenmp -x c++ -std=c++11 -DLAMBDA -triple %itanium_abi_triple -emit-llvm %s -o - | FileCheck -check-prefix=LAMBDA %s
 // RUN: %clang_cc1 -verify -fopenmp -x c++ -fblocks -DBLOCKS -triple %itanium_abi_triple -emit-llvm %s -o - | FileCheck -check-prefix=BLOCKS %s
+// RUN: %clang_cc1 -verify -fopenmp -fopenmp-enable-irbuilder -x c++ -triple x86_64-unknown-unknown -emit-llvm %s -o - | FileCheck %s -check-prefixes=ALL,IRBUILDER
 
 // RUN: %clang_cc1 -verify -fopenmp-simd -x c++ -triple x86_64-unknown-unknown -emit-llvm %s -o - | FileCheck --check-prefix SIMD-ONLY0 %s
 // RUN: %clang_cc1 -fopenmp-simd -x c++ -std=c++11 -triple x86_64-unknown-unknown -emit-pch -o %t %s
 // RUN: %clang_cc1 -fopenmp-simd -x c++ -triple x86_64-unknown-unknown -std=c++11 -include-pch %t -verify %s -emit-llvm -o - | FileCheck --check-prefix SIMD-ONLY0 %s
 // RUN: %clang_cc1 -verify -fopenmp-simd -x c++ -std=c++11 -DLAMBDA -triple %itanium_abi_triple -emit-llvm %s -o - | FileCheck --check-prefix SIMD-ONLY0 %s
 // RUN: %clang_cc1 -verify -fopenmp-simd -x c++ -fblocks -DBLOCKS -triple %itanium_abi_triple -emit-llvm %s -o - | FileCheck --check-prefix SIMD-ONLY0 %s
 // SIMD-ONLY0-NOT: {{__kmpc|__tgt}}
 // expected-no-diagnostics
#pragma omp parallel private(a)
       }();
     }();
 #else
     ++(this)->a;
 #endif
   }
 };
 
-// CHECK: [[SS_TY:%.+]] = type { i{{[0-9]+}}, i8
+// ALL: [[SS_TY:%.+]] = type { i{{[0-9]+}}, i8
 // LAMBDA: [[SS_TY:%.+]] = type { i{{[0-9]+}}, i8
 // BLOCKS: [[SS_TY:%.+]] = type { i{{[0-9]+}}, i8
-// CHECK: [[S_FLOAT_TY:%.+]] = type { float }
-// CHECK: [[S_INT_TY:%.+]] = type { i{{[0-9]+}} }
-// CHECK: [[SST_TY:%.+]] = type { i{{[0-9]+}} }
+// ALL: [[S_FLOAT_TY:%.+]] = type { float }
+// ALL: [[S_INT_TY:%.+]] = type { i{{[0-9]+}} }
+// ALL: [[SST_TY:%.+]] = type { i{{[0-9]+}} }
 template <typename T>
 T tmain() {
   S<T> test;
   SST<T> sst;
   T t_var __attribute__((aligned(128))) = T();
   T vec[] __attribute__((aligned(128))) = {1, 2};
   S<T> s_arr[] __attribute__((aligned(128))) = {1, 2};
   S<T> var __attribute__((aligned(128))) (3);
#pragma omp parallel private(t_var, vec, s_arr, var, sivar)
     vec[0] = t_var;
     s_arr[0] = var;
     sivar = 3;
   }
   return tmain<int>();
 #endif
 }
 
-// CHECK: define i{{[0-9]+}} @main()
-// CHECK: [[TEST:%.+]] = alloca [[S_FLOAT_TY]],
-// CHECK: call {{.*}} [[S_FLOAT_TY_DEF_CONSTR:@.+]]([[S_FLOAT_TY]]* [[TEST]])
-// CHECK: call void (%{{.+}}*, i{{[0-9]+}}, void (i{{[0-9]+}}*, i{{[0-9]+}}*, ...)*, ...) @__kmpc_fork_call(%{{.+}}* @{{.+}}, i{{[0-9]+}} 0, void (i{{[0-9]+}}*, i{{[0-9]+}}*, ...)* bitcast (void (i{{[0-9]+}}*, i{{[0-9]+}}*)* [[MAIN_MICROTASK:@.+]] to void
-// CHECK: = call i{{.+}} [[TMAIN_INT:@.+]]()
-// CHECK: call void [[S_FLOAT_TY_DESTR:@.+]]([[S_FLOAT_TY]]*
-// CHECK: ret
+// ALL: define i{{[0-9]+}} @main()
+// ALL: [[TEST:%.+]] = alloca [[S_FLOAT_TY]],
+// ALL: call {{.*}} [[S_FLOAT_TY_DEF_CONSTR:@.+]]([[S_FLOAT_TY]]* [[TEST]])
+// ALL: call void (%{{.+}}*, i{{[0-9]+}}, void (i{{[0-9]+}}*, i{{[0-9]+}}*, ...)*, ...) @__kmpc_fork_call(%{{.+}}* @{{.+}}, i{{[0-9]+}} 0, void (i{{[0-9]+}}*, i{{[0-9]+}}*, ...)* bitcast (void (i{{[0-9]+}}*, i{{[0-9]+}}*)* [[MAIN_MICROTASK:@.+]] to void
+// ALL: = call i{{.+}} [[TMAIN_INT:@.+]]()
+// ALL: call void [[S_FLOAT_TY_DESTR:@.+]]([[S_FLOAT_TY]]*
+// ALL: ret
 //
-// CHECK: define internal void [[MAIN_MICROTASK]](i{{[0-9]+}}* noalias [[GTID_ADDR:%.+]], i{{[0-9]+}}* noalias %{{.+}})
-// CHECK: [[T_VAR_PRIV:%.+]] = alloca i{{[0-9]+}},
-// CHECK: [[VEC_PRIV:%.+]] = alloca [2 x i{{[0-9]+}}],
-// CHECK: [[S_ARR_PRIV:%.+]] = alloca [2 x [[S_FLOAT_TY]]],
-// CHECK: [[VAR_PRIV:%.+]] = alloca [[S_FLOAT_TY]],
-// CHECK: [[SIVAR_PRIV:%.+]] = alloca i{{[0-9]+}},
+// ALL: define internal void [[MAIN_MICROTASK]](i{{[0-9]+}}* noalias [[GTID_ADDR:%.+]], i{{[0-9]+}}* noalias %{{.+}})
+// ALL: [[T_VAR_PRIV:%.+]] = alloca i{{[0-9]+}},
+// ALL: [[VEC_PRIV:%.+]] = alloca [2 x i{{[0-9]+}}],
+// ALL: [[S_ARR_PRIV:%.+]] = alloca [2 x [[S_FLOAT_TY]]],
+// ALL: [[VAR_PRIV:%.+]] = alloca [[S_FLOAT_TY]],
+// ALL: [[SIVAR_PRIV:%.+]] = alloca i{{[0-9]+}},
 // CHECK: store i{{[0-9]+}}* [[GTID_ADDR]], i{{[0-9]+}}** [[GTID_ADDR_REF:%.+]]
-// CHECK-NOT: [[T_VAR_PRIV]]
-// CHECK-NOT: [[VEC_PRIV]]
-// CHECK: {{.+}}:
+// IRBUILDER: %{{[0-9]+}} = load i{{[0-9]+}}, i{{[0-9]+}}* [[GTID_ADDR]]
+// IRBUILDER: store i{{[0-9]+}} %{{[0-9]+}}, i{{[0-9]+}}* [[GTID_ADDR_REF:%.+]]
+// ALL-NOT: [[T_VAR_PRIV]]
+// ALL-NOT: [[VEC_PRIV]]
+// ALL: {{.+}}:
 // CHECK: [[S_ARR_PRIV_ITEM:%.+]] = phi [[S_FLOAT_TY]]*
-// CHECK: call {{.*}} [[S_FLOAT_TY_DEF_CONSTR]]([[S_FLOAT_TY]]* [[S_ARR_PRIV_ITEM]])
-// CHECK-NOT: [[T_VAR_PRIV]]
-// CHECK-NOT: [[VEC_PRIV]]
-// CHECK: call {{.*}} [[S_FLOAT_TY_DEF_CONSTR]]([[S_FLOAT_TY]]* [[VAR_PRIV]])
-// CHECK-DAG: call void [[S_FLOAT_TY_DESTR]]([[S_FLOAT_TY]]* [[VAR_PRIV]])
-// CHECK-DAG: call void [[S_FLOAT_TY_DESTR]]([[S_FLOAT_TY]]*
-// CHECK: ret void
-
-// CHECK: define {{.*}} i{{[0-9]+}} [[TMAIN_INT]]()
-// CHECK: [[TEST:%.+]] = alloca [[S_INT_TY]],
-// CHECK: call {{.*}} [[S_INT_TY_DEF_CONSTR:@.+]]([[S_INT_TY]]* [[TEST]])
-// CHECK: call void (%{{.+}}*, i{{[0-9]+}}, void (i{{[0-9]+}}*, i{{[0-9]+}}*, ...)*, ...) @__kmpc_fork_call(%{{.+}}* @{{.+}}, i{{[0-9]+}} 0, void (i{{[0-9]+}}*, i{{[0-9]+}}*, ...)* bitcast (void (i{{[0-9]+}}*, i{{[0-9]+}}*)* [[TMAIN_MICROTASK:@.+]] to void
-// CHECK: call void [[S_INT_TY_DESTR:@.+]]([[S_INT_TY]]*
-// CHECK: ret
-//
-// CHECK: define {{.+}} @{{.+}}([[SS_TY]]*
-// CHECK: store i{{[0-9]+}} 0, i{{[0-9]+}}* %
-// CHECK: store i8
+// IRBUILDER: {{.+}}:
+// IRBUILDER: [[S_ARR_PRIV_ITEM:%.+]] = phi [[S_FLOAT_TY]]*
+// ALL: call {{.*}} [[S_FLOAT_TY_DEF_CONSTR]]([[S_FLOAT_TY]]* [[S_ARR_PRIV_ITEM]])
+// ALL-NOT: [[T_VAR_PRIV]]
+// ALL-NOT: [[VEC_PRIV]]
+// ALL: call {{.*}} [[S_FLOAT_TY_DEF_CONSTR]]([[S_FLOAT_TY]]* [[VAR_PRIV]])
+// ALL-DAG: call void [[S_FLOAT_TY_DESTR]]([[S_FLOAT_TY]]* [[VAR_PRIV]])
+// ALL-DAG: call void [[S_FLOAT_TY_DESTR]]([[S_FLOAT_TY]]*
+// ALL: ret void
+
+// ALL: define {{.*}} i{{[0-9]+}} [[TMAIN_INT]]()
+// ALL: [[TEST:%.+]] = alloca [[S_INT_TY]],
+// ALL: call {{.*}} [[S_INT_TY_DEF_CONSTR:@.+]]([[S_INT_TY]]* [[TEST]])
+// ALL: call void (%{{.+}}*, i{{[0-9]+}}, void (i{{[0-9]+}}*, i{{[0-9]+}}*, ...)*, ...) @__kmpc_fork_call(%{{.+}}* @{{.+}}, i{{[0-9]+}} 0, void (i{{[0-9]+}}*, i{{[0-9]+}}*, ...)* bitcast (void (i{{[0-9]+}}*, i{{[0-9]+}}*)* [[TMAIN_MICROTASK:@.+]] to void
+// ALL: call void [[S_INT_TY_DESTR:@.+]]([[S_INT_TY]]*
+// ALL: ret
+
+// IRBUILDER: define internal void [[TMAIN_MICROTASK]](i{{[0-9]+}}* noalias [[GTID_ADDR:%.+]], i{{[0-9]+}}* noalias %{{.+}})
+// IRBUILDER: [[T_VAR_PRIV:%.+]] = alloca i{{[0-9]+}}, align 128
+// IRBUILDER: [[VEC_PRIV:%.+]] = alloca [2 x i{{[0-9]+}}], align 128
+// IRBUILDER: [[S_ARR_PRIV:%.+]] = alloca [2 x [[S_INT_TY]]], align 128
+// IRBUILDER: [[VAR_PRIV:%.+]] = alloca [[S_INT_TY]], align 128
+// IRBUILDER: %{{[0-9]+}} = load i{{[0-9]+}}, i{{[0-9]+}}* [[GTID_ADDR]]
+// IRBUILDER: store i{{[0-9]+}} %{{[0-9]+}}, i{{[0-9]+}}* [[GTID_ADDR_REF:%.+]]
+// IRBUILDER-NOT: [[T_VAR_PRIV]]
+// IRBUILDER-NOT: [[VEC_PRIV]]
+// IRBUILDER-NOT: [[SIVAR_PRIV]]
+// IRBUILDER: {{.+}}:
+// IRBUILDER: ret void
+// IRBUILDER: {{.+}}:
+// IRBUILDER: [[S_ARR_PRIV_ITEM:%.+]] = phi [[S_INT_TY]]*
+// IRBUILDER: call {{.*}} [[S_INT_TY_DEF_CONSTR]]([[S_INT_TY]]* [[S_ARR_PRIV_ITEM]])
+// IRBUILDER-NOT: [[T_VAR_PRIV]]
+// IRBUILDER-NOT: [[VEC_PRIV]]
+// IRBUILDER: call {{.*}} [[S_INT_TY_DEF_CONSTR]]([[S_INT_TY]]* [[VAR_PRIV]])
+// IRBUILDER-DAG: call void [[S_INT_TY_DESTR]]([[S_INT_TY]]* [[VAR_PRIV]])
+// IRBUILDER-DAG: call void [[S_INT_TY_DESTR]]([[S_INT_TY]]*
+
+// ALL: define {{.+}} @{{.+}}([[SS_TY]]*
+// ALL: store i{{[0-9]+}} 0, i{{[0-9]+}}* %
+// ALL: store i8
 // CHECK: call void (%{{.+}}*, i{{[0-9]+}}, void (i{{[0-9]+}}*, i{{[0-9]+}}*, ...)*, ...) @__kmpc_fork_call(%{{.+}}* @{{.+}}, i{{[0-9]+}} 1, void (i{{[0-9]+}}*, i{{[0-9]+}}*, ...)* bitcast (void (i{{[0-9]+}}*, i{{[0-9]+}}*, [[SS_TY]]*)* [[SS_MICROTASK:@.+]] to void
-// CHECK: ret
+// IRBUILDER: call void (%{{.+}}*, i{{[0-9]+}}, void (i{{[0-9]+}}*, i{{[0-9]+}}*, ...)*, ...) @__kmpc_fork_call(%{{.+}}* @{{.+}}, i{{[0-9]+}} 0, void (i{{[0-9]+}}*, i{{[0-9]+}}*, ...)* bitcast (void (i{{[0-9]+}}*, i{{[0-9]+}}*)* [[SS_MICROTASK:@.+]] to void
+// ALL: ret
 
 // CHECK: define internal void [[SS_MICROTASK]](i{{[0-9]+}}* noalias [[GTID_ADDR:%.+]], i{{[0-9]+}}* noalias %{{.+}}, [[SS_TY]]* %{{.+}})
-// CHECK: [[A_PRIV:%.+]] = alloca i{{[0-9]+}},
-// CHECK: [[B_PRIV:%.+]] = alloca i{{[0-9]+}},
-// CHECK: [[C_PRIV:%.+]] = alloca i{{[0-9]+}},
-// CHECK: store i{{[0-9]+}}* [[A_PRIV]], i{{[0-9]+}}** [[REFA:%.+]],
-// CHECK: store i{{[0-9]+}}* [[C_PRIV]], i{{[0-9]+}}** [[REFC:%.+]],
+// IRBUILDER: define internal void [[SS_MICROTASK]](i{{[0-9]+}}* noalias [[GTID_ADDR:%.+]], i{{[0-9]+}}* noalias %{{.+}})
+// ALL: [[A_PRIV:%.+]] = alloca i{{[0-9]+}},
+// ALL: [[B_PRIV:%.+]] = alloca i{{[0-9]+}},
+// ALL: [[C_PRIV:%.+]] = alloca i{{[0-9]+}},
+// ALL: store i{{[0-9]+}}* [[A_PRIV]], i{{[0-9]+}}** [[REFA:%.+]],
+// ALL: store i{{[0-9]+}}* [[C_PRIV]], i{{[0-9]+}}** [[REFC:%.+]],
+// IRBUILDER: {{.+}}:
+// IRBUILDER: ret void
 // CHECK-NEXT: [[A_PRIV:%.+]] = load i{{[0-9]+}}*, i{{[0-9]+}}** [[REFA]],
-// CHECK-NEXT: [[A_VAL:%.+]] = load i{{[0-9]+}}, i{{[0-9]+}}* [[A_PRIV]],
-// CHECK-NEXT: [[INC:%.+]] = add nsw i{{[0-9]+}} [[A_VAL]], 1
-// CHECK-NEXT: store i{{[0-9]+}} [[INC]], i{{[0-9]+}}* [[A_PRIV]],
-// CHECK-NEXT: [[B_VAL:%.+]] = load i{{[0-9]+}}, i{{[0-9]+}}* [[B_PRIV]],
-// CHECK-NEXT: [[DEC:%.+]] = add nsw i{{[0-9]+}} [[B_VAL]], -1
-// CHECK-NEXT: store i{{[0-9]+}} [[DEC]], i{{[0-9]+}}* [[B_PRIV]],
-// CHECK-NEXT: [[C_PRIV:%.+]] = load i{{[0-9]+}}*, i{{[0-9]+}}** [[REFC]],
-// CHECK-NEXT: [[C_VAL:%.+]] = load i{{[0-9]+}}, i{{[0-9]+}}* [[C_PRIV]],
-// CHECK-NEXT: [[DIV:%.+]] = sdiv i{{[0-9]+}} [[C_VAL]], 1
-// CHECK-NEXT: store i{{[0-9]+}} [[DIV]], i{{[0-9]+}}* [[C_PRIV]],
+// IRBUILDER: [[A_PRIV:%.+]] = load i{{[0-9]+}}*, i{{[0-9]+}}** [[REFA]],
+// ALL-NEXT: [[A_VAL:%.+]] = load i{{[0-9]+}}, i{{[0-9]+}}* [[A_PRIV]],
+// ALL-NEXT: [[INC:%.+]] = add nsw i{{[0-9]+}} [[A_VAL]], 1
+// ALL-NEXT: store i{{[0-9]+}} [[INC]], i{{[0-9]+}}* [[A_PRIV]],
+// ALL-NEXT: [[B_VAL:%.+]] = load i{{[0-9]+}}, i{{[0-9]+}}* [[B_PRIV]],
+// ALL-NEXT: [[DEC:%.+]] = add nsw i{{[0-9]+}} [[B_VAL]], -1
+// ALL-NEXT: store i{{[0-9]+}} [[DEC]], i{{[0-9]+}}* [[B_PRIV]],
+// ALL-NEXT: [[C_PRIV:%.+]] = load i{{[0-9]+}}*, i{{[0-9]+}}** [[REFC]],
+// ALL-NEXT: [[C_VAL:%.+]] = load i{{[0-9]+}}, i{{[0-9]+}}* [[C_PRIV]],
+// ALL-NEXT: [[DIV:%.+]] = sdiv i{{[0-9]+}} [[C_VAL]], 1
+// ALL-NEXT: store i{{[0-9]+}} [[DIV]], i{{[0-9]+}}* [[C_PRIV]],
 // CHECK-NEXT: ret void
 
 // CHECK: define internal void [[TMAIN_MICROTASK]](i{{[0-9]+}}* noalias [[GTID_ADDR:%.+]], i{{[0-9]+}}* noalias %{{.+}})
 // CHECK: [[T_VAR_PRIV:%.+]] = alloca i{{[0-9]+}}, align 128
 // CHECK: [[VEC_PRIV:%.+]] = alloca [2 x i{{[0-9]+}}], align 128
 // CHECK: [[S_ARR_PRIV:%.+]] = alloca [2 x [[S_INT_TY]]], align 128
 // CHECK: [[VAR_PRIV:%.+]] = alloca [[S_INT_TY]], align 128
 // CHECK: store i{{[0-9]+}}* [[GTID_ADDR]], i{{[0-9]+}}** [[GTID_ADDR_REF:%.+]]
 // CHECK-NOT: [[T_VAR_PRIV]]
 // CHECK-NOT: [[VEC_PRIV]]
 // CHECK-NOT: [[SIVAR_PRIV]]
 // CHECK: {{.+}}:
 // CHECK: [[S_ARR_PRIV_ITEM:%.+]] = phi [[S_INT_TY]]*
 // CHECK: call {{.*}} [[S_INT_TY_DEF_CONSTR]]([[S_INT_TY]]* [[S_ARR_PRIV_ITEM]])
 // CHECK-NOT: [[T_VAR_PRIV]]
 // CHECK-NOT: [[VEC_PRIV]]
 // CHECK: call {{.*}} [[S_INT_TY_DEF_CONSTR]]([[S_INT_TY]]* [[VAR_PRIV]])
 // CHECK-DAG: call void [[S_INT_TY_DESTR]]([[S_INT_TY]]* [[VAR_PRIV]])
 // CHECK-DAG: call void [[S_INT_TY_DESTR]]([[S_INT_TY]]*
 // CHECK: ret void
 
-// CHECK: define {{.+}} @{{.+}}([[SST_TY]]* %
-// CHECK: store i{{[0-9]+}} 0, i{{[0-9]+}}* %
+// ALL: define {{.+}} @{{.+}}([[SST_TY]]* %
+// ALL: store i{{[0-9]+}} 0, i{{[0-9]+}}* %
 // CHECK: call void (%{{.+}}*, i{{[0-9]+}}, void (i{{[0-9]+}}*, i{{[0-9]+}}*, ...)*, ...) @__kmpc_fork_call(%{{.+}}* @{{.+}}, i{{[0-9]+}} 1, void (i{{[0-9]+}}*, i{{[0-9]+}}*, ...)* bitcast (void (i{{[0-9]+}}*, i{{[0-9]+}}*, [[SST_TY]]*)* [[SST_MICROTASK:@.+]] to void
-// CHECK: ret
+// IRBUILDER: call void (%{{.+}}*, i{{[0-9]+}}, void (i{{[0-9]+}}*, i{{[0-9]+}}*, ...)*, ...) @__kmpc_fork_call(%{{.+}}* @{{.+}}, i{{[0-9]+}} 0, void (i{{[0-9]+}}*, i{{[0-9]+}}*, ...)* bitcast (void (i{{[0-9]+}}*, i{{[0-9]+}}*)* [[SST_MICROTASK:@.+]] to void
+// ALL: ret
 
 // CHECK: define internal void [[SST_MICROTASK]](i{{[0-9]+}}* noalias [[GTID_ADDR:%.+]], i{{[0-9]+}}* noalias %{{.+}}, [[SST_TY]]* %{{.+}})
+// IRBUILDER: define internal void [[SST_MICROTASK]](i{{[0-9]+}}* noalias [[GTID_ADDR:%.+]], i{{[0-9]+}}* noalias %{{.+}})
 // CHECK: [[GTID_ADDR_PTR:%.+]] = alloca i32*,
 // CHECK: [[GTID_ADDR:%.+]] = load i32*, i32** [[GTID_ADDR_PTR]],
 // CHECK: [[GTID:%.+]] = load i32, i32* [[GTID_ADDR]],
-// CHECK: [[A_VOID_PTR:%.+]] = call i8* @__kmpc_alloc(i32 [[GTID]], i64 4, i8* inttoptr (i64 2 to i8*))
-// CHECK: [[A_PRIV:%.+]] = bitcast i8* [[A_VOID_PTR]] to i32*
-// CHECK: store i{{[0-9]+}}* [[A_PRIV]], i{{[0-9]+}}** [[REF:%.+]],
+// IRBUILDER: [[GTID:%.+]] = call i32 @__kmpc_global_thread_num(%{{.+}}* @{{.+}})
+// ALL: [[A_VOID_PTR:%.+]] = call i8* @__kmpc_alloc(i32 [[GTID]], i64 4, i8* inttoptr (i64 2 to i8*))
+// IRBUILDER: [[GTID2:%.+]] = call i32 @__kmpc_global_thread_num(%{{.+}}* @{{.+}})
+// ALL: [[A_PRIV:%.+]] = bitcast i8* [[A_VOID_PTR]] to i32*
+// ALL: store i{{[0-9]+}}* [[A_PRIV]], i{{[0-9]+}}** [[REF:%.+]],
+// IRBUILDER: ret void
 // CHECK-NEXT: [[A_PRIV:%.+]] = load i{{[0-9]+}}*, i{{[0-9]+}}** [[REF]],
-// CHECK-NEXT: [[A_VAL:%.+]] = load i{{[0-9]+}}, i{{[0-9]+}}* [[A_PRIV]],
-// CHECK-NEXT: [[INC:%.+]] = add nsw i{{[0-9]+}} [[A_VAL]], 1
-// CHECK-NEXT: store i{{[0-9]+}} [[INC]], i{{[0-9]+}}* [[A_PRIV]],
+// IRBUILDER: [[A_PRIV:%.+]] = load i{{[0-9]+}}*, i{{[0-9]+}}** [[REF]],
+// ALL-NEXT: [[A_VAL:%.+]] = load i{{[0-9]+}}, i{{[0-9]+}}* [[A_PRIV]],
+// ALL-NEXT: [[INC:%.+]] = add nsw i{{[0-9]+}} [[A_VAL]], 1
+// ALL-NEXT: store i{{[0-9]+}} [[INC]], i{{[0-9]+}}* [[A_PRIV]],
 // CHECK-NEXT: call void @__kmpc_free(i32 [[GTID]], i8* [[A_VOID_PTR]], i8* inttoptr (i64 2 to i8*))
+// IRBUILDER: call void @__kmpc_free(i32 [[GTID2]], i8* [[A_VOID_PTR]], i8* inttoptr (i64 2 to i8*))
 // CHECK-NEXT: ret void
 
 #endif