| Alexey Bataev | 88202be | 2017-07-27 13:20:36 +0000 | [diff] [blame] | 1 | // RUN: %clang_cc1 -verify -triple x86_64-apple-darwin10 -fopenmp -x c++ -emit-llvm %s -o - | FileCheck %s |
| 2 | // RUN: %clang_cc1 -fopenmp -x c++ -triple x86_64-apple-darwin10 -emit-pch -o %t %s |
| 3 | // RUN: %clang_cc1 -fopenmp -x c++ -triple x86_64-apple-darwin10 -include-pch %t -verify %s -emit-llvm -o - | FileCheck %s |
| 4 | // expected-no-diagnostics |
| 5 | #ifndef HEADER |
| 6 | #define HEADER |
| 7 | |
| 8 | // CHECK: [[PRIVATES:%.+]] = type { i8*, i8* } |
| 9 | |
| 10 | struct S { |
| 11 | int a; |
| 12 | S() : a(0) {} |
| 13 | S(const S&) {} |
| 14 | S& operator=(const S&) {return *this;} |
| 15 | ~S() {} |
| 16 | friend S operator+(const S&a, const S&b) {return a;} |
| 17 | }; |
| 18 | |
| 19 | |
| 20 | int main(int argc, char **argv) { |
| 21 | int a; |
| 22 | float b; |
| 23 | S c[5]; |
| 24 | short d[argc]; |
| 25 | #pragma omp taskgroup task_reduction(+: a, b, argc) |
| 26 | { |
| 27 | #pragma omp taskgroup task_reduction(-:c, d) |
| 28 | #pragma omp parallel |
| 29 | #pragma omp task in_reduction(+:a) in_reduction(-:d) |
| 30 | a += d[a]; |
| 31 | } |
| 32 | return 0; |
| 33 | } |
| 34 | |
| 35 | // CHECK-LABEL: @main |
| 36 | // CHECK: void @__kmpc_taskgroup(%ident_t* @0, i32 [[GTID:%.+]]) |
| 37 | // CHECK: [[TD1:%.+]] = call i8* @__kmpc_task_reduction_init(i32 [[GTID]], i32 3, i8* % |
| 38 | // CHECK-NEXT: store i8* [[TD1]], i8** [[TD1_ADDR:%[^,]+]], |
| 39 | // CHECK-NEXT: call void @__kmpc_taskgroup(%ident_t* @0, i32 [[GTID]]) |
| 40 | // CHECK: [[TD2:%.+]] = call i8* @__kmpc_task_reduction_init(i32 [[GTID]], i32 2, i8* % |
| 41 | // CHECK-NEXT: store i8* [[TD2]], i8** [[TD2_ADDR:%[^,]+]], |
| 42 | // CHECK-NEXT: call void (%ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%ident_t* @0, i32 5, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32*, i64, i16*, i8**, i8**)* [[OMP_PARALLEL:@.+]] to void (i32*, i32*, ...)*), i32* %{{.+}}, i64 %{{.+}}, i16* %{{.+}}, i8** [[TD1_ADDR]], i8** [[TD2_ADDR]]) |
| 43 | // CHECK-NEXT: call void @__kmpc_end_taskgroup(%ident_t* @0, i32 [[GTID]]) |
| 44 | // CHECK-NEXT: call void @__kmpc_end_taskgroup(%ident_t* @0, i32 [[GTID]]) |
| 45 | |
| 46 | // CHECK: define internal void [[OMP_PARALLEL]]( |
| 47 | // CHECK: [[TASK_T:%.+]] = call i8* @__kmpc_omp_task_alloc(%ident_t* @0, i32 [[GTID:%.+]], i32 1, i64 56, i64 40, i32 (i32, i8*)* bitcast (i32 (i32, [[T:%.+]]*)* [[OMP_TASK:@.+]] to i32 (i32, i8*)*)) |
| 48 | // CHECK-NEXT: [[TASK_T_WITH_PRIVS:%.+]] = bitcast i8* [[TASK_T]] to [[T]]* |
| 49 | // CHECK: [[PRIVS:%.+]] = getelementptr inbounds [[T]], [[T]]* [[TASK_T_WITH_PRIVS]], i32 0, i32 1 |
| 50 | // CHECK: [[TD1_REF:%.+]] = getelementptr inbounds [[PRIVATES]], [[PRIVATES]]* [[PRIVS]], i32 0, i32 0 |
| 51 | // CHECK-NEXT: [[TD1_SHAR:%.+]] = getelementptr inbounds % |
| 52 | // CHECK-NEXT: [[TD1_ADDR:%.+]] = load i8**, i8*** [[TD1_SHAR]], |
| 53 | // CHECK-NEXT: [[TD1:%.+]] = load i8*, i8** [[TD1_ADDR]], |
| 54 | // CHECK-NEXT: store i8* [[TD1]], i8** [[TD1_REF]], |
| 55 | // CHECK-NEXT: [[TD2_REF:%.+]] = getelementptr inbounds [[PRIVATES]], [[PRIVATES]]* [[PRIVS]], i32 0, i32 1 |
| 56 | // CHECK-NEXT: [[TD2_SHAR:%.+]] = getelementptr inbounds % |
| 57 | // CHECK-NEXT: [[TD2_ADDR:%.+]] = load i8**, i8*** [[TD2_SHAR]], |
| 58 | // CHECK-NEXT: [[TD2:%.+]] = load i8*, i8** [[TD2_ADDR]], |
| 59 | // CHECK-NEXT: store i8* [[TD2]], i8** [[TD2_REF]], |
| 60 | // CHECK-NEXT: call i32 @__kmpc_omp_task(%ident_t* @0, i32 [[GTID]], i8* [[TASK_T]]) |
| 61 | // CHECK-NEXT: ret void |
| 62 | // CHECK-NEXT: } |
| 63 | |
| 64 | // CHECK: define internal {{.*}} [[OMP_TASK]]( |
| 65 | // CHECK: call void (i8*, ...) %{{[^(]+}}(i8* %{{.+}}, i8*** [[TD1_REF:%[^,]+]], i8*** [[TD2_REF:%[^,]+]]) |
| 66 | // CHECK-NEXT: [[TD1_ADDR:%.+]] = load i8**, i8*** [[TD1_REF]], |
| 67 | // CHECK-NEXT: [[TD2_ADDR:%.+]] = load i8**, i8*** [[TD2_REF]], |
| 68 | // CHECK-NEXT: [[A_REF:%.+]] = getelementptr inbounds % |
| 69 | // CHECK-NEXT: [[A_ADDR:%.+]] = load i32*, i32** [[A_REF]], |
| 70 | // CHECK-NEXT: [[TD1:%.+]] = load i8*, i8** [[TD1_ADDR]], |
| 71 | // CHECK-NEXT: [[GTID:%.+]] = load i32, i32* % |
| 72 | // CHECK-NEXT: [[A_PTR:%.+]] = bitcast i32* [[A_ADDR]] to i8* |
| 73 | // CHECK-NEXT: call i8* @__kmpc_task_reduction_get_th_data(i32 [[GTID]], i8* [[TD1]], i8* [[A_PTR]]) |
| 74 | // CHECK: [[D_REF:%.+]] = getelementptr inbounds % |
| 75 | // CHECK-NEXT: [[D_ADDR:%.+]] = load i16*, i16** [[D_REF]], |
| 76 | // CHECK: [[TD2:%.+]] = load i8*, i8** [[TD2_ADDR]], |
| 77 | // CHECK-NEXT: [[D_PTR:%.+]] = bitcast i16* [[D_ADDR]] to i8* |
| 78 | // CHECK-NEXT: call i8* @__kmpc_task_reduction_get_th_data(i32 [[GTID]], i8* [[TD2]], i8* [[D_PTR]]) |
| 79 | // CHECK: add nsw i32 |
| 80 | // CHECK: store i32 % |
| 81 | #endif |