| // NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --function-signature --include-generated-funcs --replace-value-regex "__omp_offloading_[0-9a-z]+_[0-9a-z]+" "reduction_size[.].+[.]" "pl_cond[.].+[.|,]" --prefix-filecheck-ir-name _ |
| // Test target codegen - host bc file has to be created first. |
| // 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 |
| // RUN: %clang_cc1 -verify -fopenmp -x c++ -triple nvptx64-unknown-unknown -fopenmp-targets=nvptx64-nvidia-cuda -emit-llvm %s -fopenmp-is-target-device -fopenmp-host-ir-file-path %t-ppc-host.bc -o - | FileCheck %s --check-prefix=CHECK1 |
| // 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 |
| // RUN: %clang_cc1 -verify -fopenmp -x c++ -triple nvptx-unknown-unknown -fopenmp-targets=nvptx-nvidia-cuda -emit-llvm %s -fopenmp-is-target-device -fopenmp-host-ir-file-path %t-x86-host.bc -o - | FileCheck %s --check-prefix=CHECK2 |
| // RUN: %clang_cc1 -verify -fopenmp -fexceptions -fcxx-exceptions -x c++ -triple nvptx-unknown-unknown -fopenmp-targets=nvptx-nvidia-cuda -emit-llvm %s -fopenmp-is-target-device -fopenmp-host-ir-file-path %t-x86-host.bc -o - | FileCheck %s --check-prefix=CHECK2 |
| // expected-no-diagnostics |
| #ifndef HEADER |
| #define HEADER |
| |
| template<typename tx> |
| tx ftemplate(int n) { |
| tx a = 0; |
| short aa = 0; |
| tx b[10]; |
| |
| #pragma omp target teams if(0) |
| { |
| b[2] += 1; |
| } |
| |
| #pragma omp target teams if(1) |
| { |
| a = '1'; |
| } |
| |
| #pragma omp target teams if(n>40) |
| { |
| aa = 1; |
| } |
| |
| #pragma omp target teams |
| { |
| #pragma omp parallel |
| #pragma omp parallel |
| aa = 1; |
| } |
| |
| return a; |
| } |
| |
| int bar(int n){ |
| int a = 0; |
| |
| a += ftemplate<char>(n); |
| |
| return a; |
| } |
| |
| #endif |
| // CHECK1-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9ftemplateIcET_i_l23 |
| // CHECK1-SAME: (ptr noalias noundef [[DYN_PTR:%.*]], i64 noundef [[A:%.*]]) #[[ATTR0:[0-9]+]] { |
| // CHECK1-NEXT: entry: |
| // CHECK1-NEXT: [[DYN_PTR_ADDR:%.*]] = alloca ptr, align 8 |
| // CHECK1-NEXT: [[A_ADDR:%.*]] = alloca i64, align 8 |
| // CHECK1-NEXT: [[A_CASTED:%.*]] = alloca i64, align 8 |
| // CHECK1-NEXT: [[DOTZERO_ADDR:%.*]] = alloca i32, align 4 |
| // CHECK1-NEXT: [[DOTTHREADID_TEMP_:%.*]] = alloca i32, align 4 |
| // CHECK1-NEXT: store ptr [[DYN_PTR]], ptr [[DYN_PTR_ADDR]], align 8 |
| // CHECK1-NEXT: store i64 [[A]], ptr [[A_ADDR]], align 8 |
| // CHECK1-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_target_init(ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9ftemplateIcET_i_l23_kernel_environment, ptr [[DYN_PTR]]) |
| // CHECK1-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP0]], -1 |
| // CHECK1-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[WORKER_EXIT:%.*]] |
| // CHECK1: user_code.entry: |
| // CHECK1-NEXT: [[TMP1:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB1:[0-9]+]]) |
| // CHECK1-NEXT: [[TMP2:%.*]] = load i8, ptr [[A_ADDR]], align 1 |
| // CHECK1-NEXT: store i8 [[TMP2]], ptr [[A_CASTED]], align 1 |
| // CHECK1-NEXT: [[TMP3:%.*]] = load i64, ptr [[A_CASTED]], align 8 |
| // CHECK1-NEXT: store i32 0, ptr [[DOTZERO_ADDR]], align 4 |
| // CHECK1-NEXT: store i32 [[TMP1]], ptr [[DOTTHREADID_TEMP_]], align 4 |
| // CHECK1-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9ftemplateIcET_i_l23_omp_outlined(ptr [[DOTTHREADID_TEMP_]], ptr [[DOTZERO_ADDR]], i64 [[TMP3]]) #[[ATTR2:[0-9]+]] |
| // CHECK1-NEXT: call void @__kmpc_target_deinit() |
| // CHECK1-NEXT: ret void |
| // CHECK1: worker.exit: |
| // CHECK1-NEXT: ret void |
| // |
| // |
| // CHECK1-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9ftemplateIcET_i_l23_omp_outlined |
| // CHECK1-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], i64 noundef [[A:%.*]]) #[[ATTR1:[0-9]+]] { |
| // CHECK1-NEXT: entry: |
| // CHECK1-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 8 |
| // CHECK1-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 8 |
| // CHECK1-NEXT: [[A_ADDR:%.*]] = alloca i64, align 8 |
| // CHECK1-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 8 |
| // CHECK1-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 8 |
| // CHECK1-NEXT: store i64 [[A]], ptr [[A_ADDR]], align 8 |
| // CHECK1-NEXT: store i8 49, ptr [[A_ADDR]], align 1 |
| // CHECK1-NEXT: ret void |
| // |
| // |
| // CHECK1-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9ftemplateIcET_i_l28 |
| // CHECK1-SAME: (ptr noalias noundef [[DYN_PTR:%.*]], i64 noundef [[AA:%.*]]) #[[ATTR0]] { |
| // CHECK1-NEXT: entry: |
| // CHECK1-NEXT: [[DYN_PTR_ADDR:%.*]] = alloca ptr, align 8 |
| // CHECK1-NEXT: [[AA_ADDR:%.*]] = alloca i64, align 8 |
| // CHECK1-NEXT: [[AA_CASTED:%.*]] = alloca i64, align 8 |
| // CHECK1-NEXT: [[DOTZERO_ADDR:%.*]] = alloca i32, align 4 |
| // CHECK1-NEXT: [[DOTTHREADID_TEMP_:%.*]] = alloca i32, align 4 |
| // CHECK1-NEXT: store ptr [[DYN_PTR]], ptr [[DYN_PTR_ADDR]], align 8 |
| // CHECK1-NEXT: store i64 [[AA]], ptr [[AA_ADDR]], align 8 |
| // CHECK1-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_target_init(ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9ftemplateIcET_i_l28_kernel_environment, ptr [[DYN_PTR]]) |
| // CHECK1-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP0]], -1 |
| // CHECK1-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[WORKER_EXIT:%.*]] |
| // CHECK1: user_code.entry: |
| // CHECK1-NEXT: [[TMP1:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB1]]) |
| // CHECK1-NEXT: [[TMP2:%.*]] = load i16, ptr [[AA_ADDR]], align 2 |
| // CHECK1-NEXT: store i16 [[TMP2]], ptr [[AA_CASTED]], align 2 |
| // CHECK1-NEXT: [[TMP3:%.*]] = load i64, ptr [[AA_CASTED]], align 8 |
| // CHECK1-NEXT: store i32 0, ptr [[DOTZERO_ADDR]], align 4 |
| // CHECK1-NEXT: store i32 [[TMP1]], ptr [[DOTTHREADID_TEMP_]], align 4 |
| // CHECK1-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9ftemplateIcET_i_l28_omp_outlined(ptr [[DOTTHREADID_TEMP_]], ptr [[DOTZERO_ADDR]], i64 [[TMP3]]) #[[ATTR2]] |
| // CHECK1-NEXT: call void @__kmpc_target_deinit() |
| // CHECK1-NEXT: ret void |
| // CHECK1: worker.exit: |
| // CHECK1-NEXT: ret void |
| // |
| // |
| // CHECK1-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9ftemplateIcET_i_l28_omp_outlined |
| // CHECK1-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], i64 noundef [[AA:%.*]]) #[[ATTR1]] { |
| // CHECK1-NEXT: entry: |
| // CHECK1-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 8 |
| // CHECK1-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 8 |
| // CHECK1-NEXT: [[AA_ADDR:%.*]] = alloca i64, align 8 |
| // CHECK1-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 8 |
| // CHECK1-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 8 |
| // CHECK1-NEXT: store i64 [[AA]], ptr [[AA_ADDR]], align 8 |
| // CHECK1-NEXT: store i16 1, ptr [[AA_ADDR]], align 2 |
| // CHECK1-NEXT: ret void |
| // |
| // |
| // CHECK1-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9ftemplateIcET_i_l33 |
| // CHECK1-SAME: (ptr noalias noundef [[DYN_PTR:%.*]], i64 noundef [[AA:%.*]]) #[[ATTR0]] { |
| // CHECK1-NEXT: entry: |
| // CHECK1-NEXT: [[DYN_PTR_ADDR:%.*]] = alloca ptr, align 8 |
| // CHECK1-NEXT: [[AA_ADDR:%.*]] = alloca i64, align 8 |
| // CHECK1-NEXT: [[AA_CASTED:%.*]] = alloca i64, align 8 |
| // CHECK1-NEXT: [[DOTZERO_ADDR:%.*]] = alloca i32, align 4 |
| // CHECK1-NEXT: [[DOTTHREADID_TEMP_:%.*]] = alloca i32, align 4 |
| // CHECK1-NEXT: store ptr [[DYN_PTR]], ptr [[DYN_PTR_ADDR]], align 8 |
| // CHECK1-NEXT: store i64 [[AA]], ptr [[AA_ADDR]], align 8 |
| // CHECK1-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_target_init(ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9ftemplateIcET_i_l33_kernel_environment, ptr [[DYN_PTR]]) |
| // CHECK1-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP0]], -1 |
| // CHECK1-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[WORKER_EXIT:%.*]] |
| // CHECK1: user_code.entry: |
| // CHECK1-NEXT: [[TMP1:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB1]]) |
| // CHECK1-NEXT: [[TMP2:%.*]] = load i16, ptr [[AA_ADDR]], align 2 |
| // CHECK1-NEXT: store i16 [[TMP2]], ptr [[AA_CASTED]], align 2 |
| // CHECK1-NEXT: [[TMP3:%.*]] = load i64, ptr [[AA_CASTED]], align 8 |
| // CHECK1-NEXT: store i32 0, ptr [[DOTZERO_ADDR]], align 4 |
| // CHECK1-NEXT: store i32 [[TMP1]], ptr [[DOTTHREADID_TEMP_]], align 4 |
| // CHECK1-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9ftemplateIcET_i_l33_omp_outlined(ptr [[DOTTHREADID_TEMP_]], ptr [[DOTZERO_ADDR]], i64 [[TMP3]]) #[[ATTR2]] |
| // CHECK1-NEXT: call void @__kmpc_target_deinit() |
| // CHECK1-NEXT: ret void |
| // CHECK1: worker.exit: |
| // CHECK1-NEXT: ret void |
| // |
| // |
| // CHECK1-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9ftemplateIcET_i_l33_omp_outlined |
| // CHECK1-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], i64 noundef [[AA:%.*]]) #[[ATTR1]] { |
| // CHECK1-NEXT: entry: |
| // CHECK1-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 8 |
| // CHECK1-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 8 |
| // CHECK1-NEXT: [[AA_ADDR:%.*]] = alloca i64, align 8 |
| // CHECK1-NEXT: [[CAPTURED_VARS_ADDRS:%.*]] = alloca [1 x ptr], align 8 |
| // CHECK1-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 8 |
| // CHECK1-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 8 |
| // CHECK1-NEXT: store i64 [[AA]], ptr [[AA_ADDR]], align 8 |
| // CHECK1-NEXT: [[TMP0:%.*]] = getelementptr inbounds [1 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 0 |
| // CHECK1-NEXT: store ptr [[AA_ADDR]], ptr [[TMP0]], align 8 |
| // CHECK1-NEXT: [[TMP1:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8 |
| // CHECK1-NEXT: [[TMP2:%.*]] = load i32, ptr [[TMP1]], align 4 |
| // CHECK1-NEXT: call void @__kmpc_parallel_51(ptr @[[GLOB1]], i32 [[TMP2]], i32 1, i32 -1, i32 -1, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9ftemplateIcET_i_l33_omp_outlined_omp_outlined, ptr null, ptr [[CAPTURED_VARS_ADDRS]], i64 1) |
| // CHECK1-NEXT: ret void |
| // |
| // |
| // CHECK1-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9ftemplateIcET_i_l33_omp_outlined_omp_outlined |
| // CHECK1-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], ptr noundef nonnull align 2 dereferenceable(2) [[AA:%.*]]) #[[ATTR1]] { |
| // CHECK1-NEXT: entry: |
| // CHECK1-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 8 |
| // CHECK1-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 8 |
| // CHECK1-NEXT: [[AA_ADDR:%.*]] = alloca ptr, align 8 |
| // CHECK1-NEXT: [[CAPTURED_VARS_ADDRS:%.*]] = alloca [1 x ptr], align 8 |
| // CHECK1-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 8 |
| // CHECK1-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 8 |
| // CHECK1-NEXT: store ptr [[AA]], ptr [[AA_ADDR]], align 8 |
| // CHECK1-NEXT: [[TMP0:%.*]] = load ptr, ptr [[AA_ADDR]], align 8 |
| // CHECK1-NEXT: [[TMP1:%.*]] = getelementptr inbounds [1 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 0 |
| // CHECK1-NEXT: store ptr [[TMP0]], ptr [[TMP1]], align 8 |
| // CHECK1-NEXT: [[TMP2:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8 |
| // CHECK1-NEXT: [[TMP3:%.*]] = load i32, ptr [[TMP2]], align 4 |
| // CHECK1-NEXT: call void @__kmpc_parallel_51(ptr @[[GLOB1]], i32 [[TMP3]], i32 1, i32 -1, i32 -1, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9ftemplateIcET_i_l33_omp_outlined_omp_outlined_omp_outlined, ptr null, ptr [[CAPTURED_VARS_ADDRS]], i64 1) |
| // CHECK1-NEXT: ret void |
| // |
| // |
| // CHECK1-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9ftemplateIcET_i_l33_omp_outlined_omp_outlined_omp_outlined |
| // CHECK1-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], ptr noundef nonnull align 2 dereferenceable(2) [[AA:%.*]]) #[[ATTR1]] { |
| // CHECK1-NEXT: entry: |
| // CHECK1-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 8 |
| // CHECK1-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 8 |
| // CHECK1-NEXT: [[AA_ADDR:%.*]] = alloca ptr, align 8 |
| // CHECK1-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 8 |
| // CHECK1-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 8 |
| // CHECK1-NEXT: store ptr [[AA]], ptr [[AA_ADDR]], align 8 |
| // CHECK1-NEXT: [[TMP0:%.*]] = load ptr, ptr [[AA_ADDR]], align 8 |
| // CHECK1-NEXT: store i16 1, ptr [[TMP0]], align 2 |
| // CHECK1-NEXT: ret void |
| // |
| // |
| // CHECK2-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9ftemplateIcET_i_l23 |
| // CHECK2-SAME: (ptr noalias noundef [[DYN_PTR:%.*]], i32 noundef [[A:%.*]]) #[[ATTR0:[0-9]+]] { |
| // CHECK2-NEXT: entry: |
| // CHECK2-NEXT: [[DYN_PTR_ADDR:%.*]] = alloca ptr, align 4 |
| // CHECK2-NEXT: [[A_ADDR:%.*]] = alloca i32, align 4 |
| // CHECK2-NEXT: [[A_CASTED:%.*]] = alloca i32, align 4 |
| // CHECK2-NEXT: [[DOTZERO_ADDR:%.*]] = alloca i32, align 4 |
| // CHECK2-NEXT: [[DOTTHREADID_TEMP_:%.*]] = alloca i32, align 4 |
| // CHECK2-NEXT: store ptr [[DYN_PTR]], ptr [[DYN_PTR_ADDR]], align 4 |
| // CHECK2-NEXT: store i32 [[A]], ptr [[A_ADDR]], align 4 |
| // CHECK2-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_target_init(ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9ftemplateIcET_i_l23_kernel_environment, ptr [[DYN_PTR]]) |
| // CHECK2-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP0]], -1 |
| // CHECK2-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[WORKER_EXIT:%.*]] |
| // CHECK2: user_code.entry: |
| // CHECK2-NEXT: [[TMP1:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB1:[0-9]+]]) |
| // CHECK2-NEXT: [[TMP2:%.*]] = load i8, ptr [[A_ADDR]], align 1 |
| // CHECK2-NEXT: store i8 [[TMP2]], ptr [[A_CASTED]], align 1 |
| // CHECK2-NEXT: [[TMP3:%.*]] = load i32, ptr [[A_CASTED]], align 4 |
| // CHECK2-NEXT: store i32 0, ptr [[DOTZERO_ADDR]], align 4 |
| // CHECK2-NEXT: store i32 [[TMP1]], ptr [[DOTTHREADID_TEMP_]], align 4 |
| // CHECK2-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9ftemplateIcET_i_l23_omp_outlined(ptr [[DOTTHREADID_TEMP_]], ptr [[DOTZERO_ADDR]], i32 [[TMP3]]) #[[ATTR2:[0-9]+]] |
| // CHECK2-NEXT: call void @__kmpc_target_deinit() |
| // CHECK2-NEXT: ret void |
| // CHECK2: worker.exit: |
| // CHECK2-NEXT: ret void |
| // |
| // |
| // CHECK2-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9ftemplateIcET_i_l23_omp_outlined |
| // CHECK2-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], i32 noundef [[A:%.*]]) #[[ATTR1:[0-9]+]] { |
| // CHECK2-NEXT: entry: |
| // CHECK2-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 4 |
| // CHECK2-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 4 |
| // CHECK2-NEXT: [[A_ADDR:%.*]] = alloca i32, align 4 |
| // CHECK2-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 4 |
| // CHECK2-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 4 |
| // CHECK2-NEXT: store i32 [[A]], ptr [[A_ADDR]], align 4 |
| // CHECK2-NEXT: store i8 49, ptr [[A_ADDR]], align 1 |
| // CHECK2-NEXT: ret void |
| // |
| // |
| // CHECK2-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9ftemplateIcET_i_l28 |
| // CHECK2-SAME: (ptr noalias noundef [[DYN_PTR:%.*]], i32 noundef [[AA:%.*]]) #[[ATTR0]] { |
| // CHECK2-NEXT: entry: |
| // CHECK2-NEXT: [[DYN_PTR_ADDR:%.*]] = alloca ptr, align 4 |
| // CHECK2-NEXT: [[AA_ADDR:%.*]] = alloca i32, align 4 |
| // CHECK2-NEXT: [[AA_CASTED:%.*]] = alloca i32, align 4 |
| // CHECK2-NEXT: [[DOTZERO_ADDR:%.*]] = alloca i32, align 4 |
| // CHECK2-NEXT: [[DOTTHREADID_TEMP_:%.*]] = alloca i32, align 4 |
| // CHECK2-NEXT: store ptr [[DYN_PTR]], ptr [[DYN_PTR_ADDR]], align 4 |
| // CHECK2-NEXT: store i32 [[AA]], ptr [[AA_ADDR]], align 4 |
| // CHECK2-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_target_init(ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9ftemplateIcET_i_l28_kernel_environment, ptr [[DYN_PTR]]) |
| // CHECK2-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP0]], -1 |
| // CHECK2-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[WORKER_EXIT:%.*]] |
| // CHECK2: user_code.entry: |
| // CHECK2-NEXT: [[TMP1:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB1]]) |
| // CHECK2-NEXT: [[TMP2:%.*]] = load i16, ptr [[AA_ADDR]], align 2 |
| // CHECK2-NEXT: store i16 [[TMP2]], ptr [[AA_CASTED]], align 2 |
| // CHECK2-NEXT: [[TMP3:%.*]] = load i32, ptr [[AA_CASTED]], align 4 |
| // CHECK2-NEXT: store i32 0, ptr [[DOTZERO_ADDR]], align 4 |
| // CHECK2-NEXT: store i32 [[TMP1]], ptr [[DOTTHREADID_TEMP_]], align 4 |
| // CHECK2-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9ftemplateIcET_i_l28_omp_outlined(ptr [[DOTTHREADID_TEMP_]], ptr [[DOTZERO_ADDR]], i32 [[TMP3]]) #[[ATTR2]] |
| // CHECK2-NEXT: call void @__kmpc_target_deinit() |
| // CHECK2-NEXT: ret void |
| // CHECK2: worker.exit: |
| // CHECK2-NEXT: ret void |
| // |
| // |
| // CHECK2-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9ftemplateIcET_i_l28_omp_outlined |
| // CHECK2-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], i32 noundef [[AA:%.*]]) #[[ATTR1]] { |
| // CHECK2-NEXT: entry: |
| // CHECK2-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 4 |
| // CHECK2-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 4 |
| // CHECK2-NEXT: [[AA_ADDR:%.*]] = alloca i32, align 4 |
| // CHECK2-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 4 |
| // CHECK2-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 4 |
| // CHECK2-NEXT: store i32 [[AA]], ptr [[AA_ADDR]], align 4 |
| // CHECK2-NEXT: store i16 1, ptr [[AA_ADDR]], align 2 |
| // CHECK2-NEXT: ret void |
| // |
| // |
| // CHECK2-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9ftemplateIcET_i_l33 |
| // CHECK2-SAME: (ptr noalias noundef [[DYN_PTR:%.*]], i32 noundef [[AA:%.*]]) #[[ATTR0]] { |
| // CHECK2-NEXT: entry: |
| // CHECK2-NEXT: [[DYN_PTR_ADDR:%.*]] = alloca ptr, align 4 |
| // CHECK2-NEXT: [[AA_ADDR:%.*]] = alloca i32, align 4 |
| // CHECK2-NEXT: [[AA_CASTED:%.*]] = alloca i32, align 4 |
| // CHECK2-NEXT: [[DOTZERO_ADDR:%.*]] = alloca i32, align 4 |
| // CHECK2-NEXT: [[DOTTHREADID_TEMP_:%.*]] = alloca i32, align 4 |
| // CHECK2-NEXT: store ptr [[DYN_PTR]], ptr [[DYN_PTR_ADDR]], align 4 |
| // CHECK2-NEXT: store i32 [[AA]], ptr [[AA_ADDR]], align 4 |
| // CHECK2-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_target_init(ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9ftemplateIcET_i_l33_kernel_environment, ptr [[DYN_PTR]]) |
| // CHECK2-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP0]], -1 |
| // CHECK2-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[WORKER_EXIT:%.*]] |
| // CHECK2: user_code.entry: |
| // CHECK2-NEXT: [[TMP1:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB1]]) |
| // CHECK2-NEXT: [[TMP2:%.*]] = load i16, ptr [[AA_ADDR]], align 2 |
| // CHECK2-NEXT: store i16 [[TMP2]], ptr [[AA_CASTED]], align 2 |
| // CHECK2-NEXT: [[TMP3:%.*]] = load i32, ptr [[AA_CASTED]], align 4 |
| // CHECK2-NEXT: store i32 0, ptr [[DOTZERO_ADDR]], align 4 |
| // CHECK2-NEXT: store i32 [[TMP1]], ptr [[DOTTHREADID_TEMP_]], align 4 |
| // CHECK2-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9ftemplateIcET_i_l33_omp_outlined(ptr [[DOTTHREADID_TEMP_]], ptr [[DOTZERO_ADDR]], i32 [[TMP3]]) #[[ATTR2]] |
| // CHECK2-NEXT: call void @__kmpc_target_deinit() |
| // CHECK2-NEXT: ret void |
| // CHECK2: worker.exit: |
| // CHECK2-NEXT: ret void |
| // |
| // |
| // CHECK2-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9ftemplateIcET_i_l33_omp_outlined |
| // CHECK2-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], i32 noundef [[AA:%.*]]) #[[ATTR1]] { |
| // CHECK2-NEXT: entry: |
| // CHECK2-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 4 |
| // CHECK2-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 4 |
| // CHECK2-NEXT: [[AA_ADDR:%.*]] = alloca i32, align 4 |
| // CHECK2-NEXT: [[CAPTURED_VARS_ADDRS:%.*]] = alloca [1 x ptr], align 4 |
| // CHECK2-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 4 |
| // CHECK2-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 4 |
| // CHECK2-NEXT: store i32 [[AA]], ptr [[AA_ADDR]], align 4 |
| // CHECK2-NEXT: [[TMP0:%.*]] = getelementptr inbounds [1 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i32 0, i32 0 |
| // CHECK2-NEXT: store ptr [[AA_ADDR]], ptr [[TMP0]], align 4 |
| // CHECK2-NEXT: [[TMP1:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4 |
| // CHECK2-NEXT: [[TMP2:%.*]] = load i32, ptr [[TMP1]], align 4 |
| // CHECK2-NEXT: call void @__kmpc_parallel_51(ptr @[[GLOB1]], i32 [[TMP2]], i32 1, i32 -1, i32 -1, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9ftemplateIcET_i_l33_omp_outlined_omp_outlined, ptr null, ptr [[CAPTURED_VARS_ADDRS]], i32 1) |
| // CHECK2-NEXT: ret void |
| // |
| // |
| // CHECK2-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9ftemplateIcET_i_l33_omp_outlined_omp_outlined |
| // CHECK2-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], ptr noundef nonnull align 2 dereferenceable(2) [[AA:%.*]]) #[[ATTR1]] { |
| // CHECK2-NEXT: entry: |
| // CHECK2-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 4 |
| // CHECK2-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 4 |
| // CHECK2-NEXT: [[AA_ADDR:%.*]] = alloca ptr, align 4 |
| // CHECK2-NEXT: [[CAPTURED_VARS_ADDRS:%.*]] = alloca [1 x ptr], align 4 |
| // CHECK2-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 4 |
| // CHECK2-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 4 |
| // CHECK2-NEXT: store ptr [[AA]], ptr [[AA_ADDR]], align 4 |
| // CHECK2-NEXT: [[TMP0:%.*]] = load ptr, ptr [[AA_ADDR]], align 4 |
| // CHECK2-NEXT: [[TMP1:%.*]] = getelementptr inbounds [1 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i32 0, i32 0 |
| // CHECK2-NEXT: store ptr [[TMP0]], ptr [[TMP1]], align 4 |
| // CHECK2-NEXT: [[TMP2:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4 |
| // CHECK2-NEXT: [[TMP3:%.*]] = load i32, ptr [[TMP2]], align 4 |
| // CHECK2-NEXT: call void @__kmpc_parallel_51(ptr @[[GLOB1]], i32 [[TMP3]], i32 1, i32 -1, i32 -1, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9ftemplateIcET_i_l33_omp_outlined_omp_outlined_omp_outlined, ptr null, ptr [[CAPTURED_VARS_ADDRS]], i32 1) |
| // CHECK2-NEXT: ret void |
| // |
| // |
| // CHECK2-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9ftemplateIcET_i_l33_omp_outlined_omp_outlined_omp_outlined |
| // CHECK2-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], ptr noundef nonnull align 2 dereferenceable(2) [[AA:%.*]]) #[[ATTR1]] { |
| // CHECK2-NEXT: entry: |
| // CHECK2-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 4 |
| // CHECK2-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 4 |
| // CHECK2-NEXT: [[AA_ADDR:%.*]] = alloca ptr, align 4 |
| // CHECK2-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 4 |
| // CHECK2-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 4 |
| // CHECK2-NEXT: store ptr [[AA]], ptr [[AA_ADDR]], align 4 |
| // CHECK2-NEXT: [[TMP0:%.*]] = load ptr, ptr [[AA_ADDR]], align 4 |
| // CHECK2-NEXT: store i16 1, ptr [[TMP0]], align 2 |
| // CHECK2-NEXT: ret void |
| // |
