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
5 // RUN: %clang_cc1 -verify -triple x86_64-apple-darwin10 -fopenmp-simd -x c++ -emit-llvm %s -o - | FileCheck --check-prefix SIMD-ONLY0 %s
6 // RUN: %clang_cc1 -fopenmp-simd -x c++ -triple x86_64-apple-darwin10 -emit-pch -o %t %s
7 // RUN: %clang_cc1 -fopenmp-simd -x c++ -triple x86_64-apple-darwin10 -include-pch %t -verify %s -emit-llvm -o - | FileCheck --check-prefix SIMD-ONLY0 %s
8 // SIMD-ONLY0-NOT: {{__kmpc|__tgt}}
9 // expected-no-diagnostics
10 #ifndef HEADER
11 #define HEADER
12
13 // CHECK: [[PRIVATES:%.+]] = type { i8*, i8* }
14
15 struct S {
16 int a;
SS17 S() : a(0) {}
SS18 S(const S&) {}
operator =S19 S& operator=(const S&) {return *this;}
~SS20 ~S() {}
operator +(const S & a,const S & b)21 friend S operator+(const S&a, const S&b) {return a;}
22 };
23
24
main(int argc,char ** argv)25 int main(int argc, char **argv) {
26 int a;
27 float b;
28 S c[5];
29 short d[argc];
30 #pragma omp taskgroup task_reduction(+: a, b, argc)
31 {
32 #pragma omp taskgroup task_reduction(-:c, d)
33 #pragma omp parallel
34 #pragma omp taskloop in_reduction(+:a) in_reduction(-:d)
35 for (int i = 0; i < 5; ++i)
36 a += d[a];
37 }
38 return 0;
39 }
40
41 // CHECK-LABEL: @main
42 // CHECK: void @__kmpc_taskgroup(%struct.ident_t* @0, i32 [[GTID:%.+]])
43 // CHECK: [[TD1:%.+]] = call i8* @__kmpc_task_reduction_init(i32 [[GTID]], i32 3, i8* %
44 // CHECK-NEXT: store i8* [[TD1]], i8** [[TD1_ADDR:%[^,]+]],
45 // CHECK-NEXT: call void @__kmpc_taskgroup(%struct.ident_t* @0, i32 [[GTID]])
46 // CHECK: [[TD2:%.+]] = call i8* @__kmpc_task_reduction_init(i32 [[GTID]], i32 2, i8* %
47 // CHECK-NEXT: store i8* [[TD2]], i8** [[TD2_ADDR:%[^,]+]],
48 // CHECK-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.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]])
49 // CHECK-NEXT: call void @__kmpc_end_taskgroup(%struct.ident_t* @0, i32 [[GTID]])
50 // CHECK-NEXT: call void @__kmpc_end_taskgroup(%struct.ident_t* @0, i32 [[GTID]])
51
52 // CHECK: define internal void [[OMP_PARALLEL]](
53 // CHECK: [[TASK_T:%.+]] = call i8* @__kmpc_omp_task_alloc(%struct.ident_t* @0, i32 [[GTID:%.+]], i32 1, i64 96, i64 40, i32 (i32, i8*)* bitcast (i32 (i32, [[T:%.+]]*)* [[OMP_TASK:@.+]] to i32 (i32, i8*)*))
54 // CHECK-NEXT: [[TASK_T_WITH_PRIVS:%.+]] = bitcast i8* [[TASK_T]] to [[T]]*
55 // CHECK: [[PRIVS:%.+]] = getelementptr inbounds [[T]], [[T]]* [[TASK_T_WITH_PRIVS]], i32 0, i32 1
56 // CHECK: [[TD1_REF:%.+]] = getelementptr inbounds [[PRIVATES]], [[PRIVATES]]* [[PRIVS]], i32 0, i32 0
57 // CHECK-NEXT: [[TD1_SHAR:%.+]] = getelementptr inbounds %
58 // CHECK-NEXT: [[TD1_ADDR:%.+]] = load i8**, i8*** [[TD1_SHAR]],
59 // CHECK-NEXT: [[TD1:%.+]] = load i8*, i8** [[TD1_ADDR]],
60 // CHECK-NEXT: store i8* [[TD1]], i8** [[TD1_REF]],
61 // CHECK-NEXT: [[TD2_REF:%.+]] = getelementptr inbounds [[PRIVATES]], [[PRIVATES]]* [[PRIVS]], i32 0, i32 1
62 // CHECK-NEXT: [[TD2_SHAR:%.+]] = getelementptr inbounds %
63 // CHECK-NEXT: [[TD2_ADDR:%.+]] = load i8**, i8*** [[TD2_SHAR]],
64 // CHECK-NEXT: [[TD2:%.+]] = load i8*, i8** [[TD2_ADDR]],
65 // CHECK-NEXT: store i8* [[TD2]], i8** [[TD2_REF]],
66 // CHECK: call void @__kmpc_taskloop(%struct.ident_t* @0, i32 [[GTID]], i8* [[TASK_T]], i32 1,
67 // CHECK: ret void
68 // CHECK-NEXT: }
69
70 // CHECK: define internal {{.*}} [[OMP_TASK]](
71 // CHECK: call void (i8*, ...) %{{[^(]+}}(i8* %{{.+}}, i8*** [[TD1_REF:%[^,]+]], i8*** [[TD2_REF:%[^,]+]])
72 // CHECK-NEXT: [[TD1_ADDR:%.+]] = load i8**, i8*** [[TD1_REF]],
73 // CHECK-NEXT: [[TD2_ADDR:%.+]] = load i8**, i8*** [[TD2_REF]],
74 // CHECK-NEXT: [[A_REF:%.+]] = getelementptr inbounds %
75 // CHECK-NEXT: [[A_ADDR:%.+]] = load i32*, i32** [[A_REF]],
76 // CHECK-NEXT: [[TD1:%.+]] = load i8*, i8** [[TD1_ADDR]],
77 // CHECK-NEXT: [[GTID:%.+]] = load i32, i32* %
78 // CHECK-NEXT: [[A_PTR:%.+]] = bitcast i32* [[A_ADDR]] to i8*
79 // CHECK-NEXT: call i8* @__kmpc_task_reduction_get_th_data(i32 [[GTID]], i8* [[TD1]], i8* [[A_PTR]])
80 // CHECK: [[D_REF:%.+]] = getelementptr inbounds %
81 // CHECK-NEXT: [[D_ADDR:%.+]] = load i16*, i16** [[D_REF]],
82 // CHECK: [[TD2:%.+]] = load i8*, i8** [[TD2_ADDR]],
83 // CHECK-NEXT: [[D_PTR:%.+]] = bitcast i16* [[D_ADDR]] to i8*
84 // CHECK-NEXT: call i8* @__kmpc_task_reduction_get_th_data(i32 [[GTID]], i8* [[TD2]], i8* [[D_PTR]])
85 // CHECK: add nsw i32
86 // CHECK: store i32 %
87 #endif
88