1 // Test target codegen - host bc file has to be created first.
2 // RUN: %clang_cc1 -verify -fopenmp -fopenmp-version=45 -x c++ -triple powerpc64le-unknown-unknown -fopenmp-targets=nvptx64-nvidia-cuda -emit-llvm-bc %s -o %t-ppc-host.bc
3 // RUN: %clang_cc1 -verify -fopenmp -fopenmp-version=45 -x c++ -triple nvptx64-unknown-unknown -fopenmp-targets=nvptx64-nvidia-cuda -emit-llvm %s -fopenmp-is-device -fopenmp-host-ir-file-path %t-ppc-host.bc -o - | FileCheck %s --check-prefix CHECK --check-prefix CHECK-64
4 // RUN: %clang_cc1 -verify -fopenmp -fopenmp-version=45 -x c++ -triple i386-unknown-unknown -fopenmp-targets=nvptx-nvidia-cuda -emit-llvm-bc %s -o %t-x86-host.bc
5 // RUN: %clang_cc1 -verify -fopenmp -fopenmp-version=45 -x c++ -triple nvptx-unknown-unknown -fopenmp-targets=nvptx-nvidia-cuda -emit-llvm %s -fopenmp-is-device -fopenmp-host-ir-file-path %t-x86-host.bc -o - | FileCheck %s --check-prefix CHECK --check-prefix CHECK-32
6 // RUN: %clang_cc1 -verify -fopenmp -fopenmp-version=45 -fexceptions -fcxx-exceptions -x c++ -triple nvptx-unknown-unknown -fopenmp-targets=nvptx-nvidia-cuda -emit-llvm %s -fopenmp-is-device -fopenmp-host-ir-file-path %t-x86-host.bc -o - | FileCheck %s --check-prefix CHECK --check-prefix CHECK-32
7 // expected-no-diagnostics
15 int main(int argc, char **argv) {
16 int b[10], c[10], d[10];
17 #pragma omp target teams map(tofrom:a)
18 #pragma omp distribute parallel for firstprivate(b) lastprivate(c) if(a)
19 for (int i= 0; i < argc; ++i)
20 a = foo(&i) + foo(&a) + foo(&b[i]) + foo(&c[i]) + foo(&d[i]);
24 // CHECK: @__omp_offloading_{{.*}}_main_l17_exec_mode = weak constant i8 1
26 // CHECK-LABEL: define internal void @__omp_offloading_{{.*}}_main_l17_worker(
28 // CHECK: define weak void @__omp_offloading_{{.*}}_main_l17([10 x i32]* dereferenceable(40) %{{.+}}, [10 x i32]* dereferenceable(40) %{{.+}}, i32* dereferenceable(4) %{{.+}}, i{{64|32}} %{{.+}}, [10 x i32]* dereferenceable(40) %{{.+}})
29 // CHECK: [[PTR:%.+]] = call i8* @__kmpc_data_sharing_push_stack(i{{64|32}} 84, i16 0)
30 // CHECK: [[STACK:%.+]] = bitcast i8* [[PTR]] to %struct._globalized_locals_ty*
31 // CHECK: [[ARGC:%.+]] = load i32, i32* %{{.+}}, align
32 // CHECK: [[ARGC_ADDR:%.+]] = getelementptr inbounds %struct._globalized_locals_ty, %struct._globalized_locals_ty* [[STACK]], i{{32|64}} 0, i{{32|64}} 0
33 // CHECK: store i32 [[ARGC]], i32* [[ARGC_ADDR]],
34 // CHECK: getelementptr inbounds %struct._globalized_locals_ty, %struct._globalized_locals_ty* [[STACK]], i{{32|64}} 0, i{{32|64}} 1
35 // CHECK: getelementptr inbounds %struct._globalized_locals_ty, %struct._globalized_locals_ty* [[STACK]], i{{32|64}} 0, i{{32|64}} 2
36 // CHECK: call void @__kmpc_for_static_init_4(
38 // CHECK: call void @__kmpc_serialized_parallel(
39 // CHECK: call void [[PARALLEL:@.+]](
40 // CHECK: call void @__kmpc_end_serialized_parallel(
42 // CHECK: call void @__kmpc_for_static_fini(%struct.ident_t* @
44 // CHECK: call void @__kmpc_data_sharing_pop_stack(i8* [[PTR]])
46 // CHECK: define internal void [[PARALLEL]](
47 // CHECK-NOT: call i8* @__kmpc_data_sharing_push_stack(
49 // CHECK-NOT: call void @__kmpc_data_sharing_pop_stack(