1 // Test target codegen - host bc file has to be created first. 2 // RUN: %clang_cc1 -verify -fopenmp -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 -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 -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 -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 -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 8 #ifndef HEADER 9 #define HEADER 10 11 // CHECK: [[MEM_TY:%.+]] = type { [128 x i8] } 12 // CHECK-DAG: {{@__omp_offloading_.+}}_l20_exec_mode = weak constant i8 1 13 // CHECK-DAG: [[KERNEL_SIZE:@.+]] = internal unnamed_addr constant i{{64|32}} 4 14 // CHECK-DAG: [[KERNEL_SHARED:@.+]] = internal unnamed_addr constant i16 1 15 16 template<typename tx> 17 tx ftemplate(int n) { 18 int i; 19 20 #pragma omp target teams distribute 21 for (i = 0; i < 10; ++i) 22 { 23 #pragma omp parallel 24 ++i; 25 } 26 27 return i; 28 } 29 30 int bar(int n){ 31 int a = 0; 32 33 a += ftemplate<char>(n); 34 35 return a; 36 } 37 38 // CHECK-LABEL: define {{.*}}void {{@__omp_offloading_.+template.+l20}}_worker() 39 // CHECK: ret void 40 41 // CHECK: define {{.*}}void {{@__omp_offloading_.+template.+l20}}() 42 43 // CHECK-DAG: [[TID:%.+]] = call i32 @llvm.nvvm.read.ptx.sreg.tid.x() 44 // CHECK-DAG: [[NTH:%.+]] = call i32 @llvm.nvvm.read.ptx.sreg.ntid.x() 45 // CHECK-DAG: [[WS:%.+]] = call i32 @llvm.nvvm.read.ptx.sreg.warpsize() 46 // CHECK-DAG: [[TH_LIMIT:%.+]] = sub nuw i32 [[NTH]], [[WS]] 47 // CHECK: [[IS_WORKER:%.+]] = icmp ult i32 [[TID]], [[TH_LIMIT]] 48 // CHECK: br i1 [[IS_WORKER]], label {{%?}}[[WORKER:.+]], label {{%?}}[[CHECK_MASTER:.+]] 49 // 50 // CHECK: [[WORKER]] 51 // CHECK: {{call|invoke}} void {{@__omp_offloading_.+template.+l20}}_worker() 52 // CHECK: br label {{%?}}[[EXIT:.+]] 53 // 54 // CHECK: [[CHECK_MASTER]] 55 // CHECK-DAG: [[CMTID:%.+]] = call i32 @llvm.nvvm.read.ptx.sreg.tid.x() 56 // CHECK-DAG: [[CMNTH:%.+]] = call i32 @llvm.nvvm.read.ptx.sreg.ntid.x() 57 // CHECK-DAG: [[CMWS:%.+]] = call i32 @llvm.nvvm.read.ptx.sreg.warpsize() 58 // CHECK: [[IS_MASTER:%.+]] = icmp eq i32 [[CMTID]], 59 // CHECK: br i1 [[IS_MASTER]], label {{%?}}[[MASTER:.+]], label {{%?}}[[EXIT]] 60 // 61 // CHECK: [[MASTER]] 62 // CHECK-DAG: [[MNTH:%.+]] = call i32 @llvm.nvvm.read.ptx.sreg.ntid.x() 63 // CHECK-DAG: [[MWS:%.+]] = call i32 @llvm.nvvm.read.ptx.sreg.warpsize() 64 // CHECK: [[MTMP1:%.+]] = sub nuw i32 [[MNTH]], [[MWS]] 65 // CHECK: call void @__kmpc_kernel_init(i32 [[MTMP1]] 66 // CHECK: call void [[PARALLEL:@.+]](i32* %{{.+}}, i32* %{{.+}}) 67 // CHECK: br label {{%?}}[[TERMINATE:.+]] 68 // 69 // CHECK: [[TERMINATE]] 70 // CHECK: call void @__kmpc_kernel_deinit( 71 // CHECK: call void @__kmpc_barrier_simple_spmd(%struct.ident_t* null, i32 0) 72 // CHECK: br label {{%?}}[[EXIT]] 73 // 74 // CHECK: [[EXIT]] 75 // CHECK: ret void 76 77 // CHECK: define internal void [[PARALLEL]](i32* noalias %{{.+}}, i32* noalias %{{.+}}) 78 // CHECK: [[SHARED:%.+]] = load i16, i16* [[KERNEL_SHARED]], 79 // CHECK: [[SIZE:%.+]] = load i{{64|32}}, i{{64|32}}* [[KERNEL_SIZE]], 80 // CHECK: call void @__kmpc_get_team_static_memory(i16 0, i8* addrspacecast (i8 addrspace(3)* getelementptr inbounds ([[MEM_TY]], [[MEM_TY]] addrspace(3)* @{{.+}}, i32 0, i32 0, i32 0) to i8*), i{{64|32}} [[SIZE]], i16 [[SHARED]], i8** addrspacecast (i8* addrspace(3)* [[BUF:@.+]] to i8**)) 81 // CHECK: [[PTR:%.+]] = load i8*, i8* addrspace(3)* [[BUF]], 82 // CHECK: [[ADDR:%.+]] = getelementptr inbounds i8, i8* [[PTR]], i{{64|32}} 0 83 // CHECK: [[RD:%.+]] = bitcast i8* [[ADDR]] to [[GLOB_TY:%.+]]* 84 // CHECK: [[I_ADDR:%.+]] = getelementptr inbounds [[GLOB_TY]], [[GLOB_TY]]* [[RD]], i32 0, i32 0 85 // 86 // CHECK: call void @__kmpc_for_static_init_4( 87 // CHECK: call void @__kmpc_kernel_prepare_parallel(i8* bitcast (void (i16, i32)* @{{.+}} to i8*), i16 1) 88 // CHECK: call void @__kmpc_begin_sharing_variables(i8*** [[SHARED_VARS_PTR:%.+]], i{{64|32}} 1) 89 // CHECK: [[SHARED_VARS_BUF:%.+]] = load i8**, i8*** [[SHARED_VARS_PTR]], 90 // CHECK: [[VARS_BUF:%.+]] = getelementptr inbounds i8*, i8** [[SHARED_VARS_BUF]], i{{64|32}} 0 91 // CHECK: [[I_ADDR_BC:%.+]] = bitcast i32* [[I_ADDR]] to i8* 92 // CHECK: store i8* [[I_ADDR_BC]], i8** [[VARS_BUF]], 93 // CHECK: call void @__kmpc_barrier_simple_spmd(%struct.ident_t* null, i32 0) 94 // CHECK: call void @__kmpc_barrier_simple_spmd(%struct.ident_t* null, i32 0) 95 // CHECK: call void @__kmpc_end_sharing_variables() 96 // CHECK: call void @__kmpc_for_static_fini( 97 #endif 98