1 // Test host codegen. 2 // RUN: %clang_cc1 -verify -fopenmp -fopenmp-version=45 -x c++ -triple powerpc64le-unknown-unknown -fopenmp-targets=powerpc64le-ibm-linux-gnu -emit-llvm %s -o - | FileCheck -allow-deprecated-dag-overlap %s --check-prefix CHECK --check-prefix CHECK-64 3 // RUN: %clang_cc1 -fopenmp -fopenmp-version=45 -x c++ -std=c++11 -triple powerpc64le-unknown-unknown -fopenmp-targets=powerpc64le-ibm-linux-gnu -emit-pch -o %t %s 4 // RUN: %clang_cc1 -fopenmp -fopenmp-version=45 -x c++ -triple powerpc64le-unknown-unknown -fopenmp-targets=powerpc64le-ibm-linux-gnu -std=c++11 -include-pch %t -verify %s -emit-llvm -o - | FileCheck -allow-deprecated-dag-overlap %s --check-prefix CHECK --check-prefix CHECK-64 5 // RUN: %clang_cc1 -verify -fopenmp -fopenmp-version=45 -x c++ -triple i386-unknown-unknown -fopenmp-targets=i386-pc-linux-gnu -emit-llvm %s -o - | FileCheck -allow-deprecated-dag-overlap %s --check-prefix CHECK --check-prefix CHECK-32 6 // RUN: %clang_cc1 -fopenmp -fopenmp-version=45 -x c++ -std=c++11 -triple i386-unknown-unknown -fopenmp-targets=i386-pc-linux-gnu -emit-pch -o %t %s 7 // RUN: %clang_cc1 -fopenmp -fopenmp-version=45 -x c++ -triple i386-unknown-unknown -fopenmp-targets=i386-pc-linux-gnu -std=c++11 -include-pch %t -verify %s -emit-llvm -o - | FileCheck -allow-deprecated-dag-overlap %s --check-prefix CHECK --check-prefix CHECK-32 8 9 // RUN: %clang_cc1 -verify -fopenmp-simd -fopenmp-version=45 -x c++ -triple powerpc64le-unknown-unknown -fopenmp-targets=powerpc64le-ibm-linux-gnu -emit-llvm %s -o - | FileCheck -allow-deprecated-dag-overlap --check-prefix SIMD-ONLY0 %s 10 // RUN: %clang_cc1 -fopenmp-simd -fopenmp-version=45 -x c++ -std=c++11 -triple powerpc64le-unknown-unknown -fopenmp-targets=powerpc64le-ibm-linux-gnu -emit-pch -o %t %s 11 // RUN: %clang_cc1 -fopenmp-simd -fopenmp-version=45 -x c++ -triple powerpc64le-unknown-unknown -fopenmp-targets=powerpc64le-ibm-linux-gnu -std=c++11 -include-pch %t -verify %s -emit-llvm -o - | FileCheck -allow-deprecated-dag-overlap --check-prefix SIMD-ONLY0 %s 12 // RUN: %clang_cc1 -verify -fopenmp-simd -fopenmp-version=45 -x c++ -triple i386-unknown-unknown -fopenmp-targets=i386-pc-linux-gnu -emit-llvm %s -o - | FileCheck -allow-deprecated-dag-overlap --check-prefix SIMD-ONLY0 %s 13 // RUN: %clang_cc1 -fopenmp-simd -fopenmp-version=45 -x c++ -std=c++11 -triple i386-unknown-unknown -fopenmp-targets=i386-pc-linux-gnu -emit-pch -o %t %s 14 // RUN: %clang_cc1 -fopenmp-simd -fopenmp-version=45 -x c++ -triple i386-unknown-unknown -fopenmp-targets=i386-pc-linux-gnu -std=c++11 -include-pch %t -verify %s -emit-llvm -o - | FileCheck -allow-deprecated-dag-overlap --check-prefix SIMD-ONLY0 %s 15 // SIMD-ONLY0-NOT: {{__kmpc|__tgt}} 16 17 // Test target codegen - host bc file has to be created first. 18 // RUN: %clang_cc1 -verify -fopenmp -fopenmp-version=45 -x c++ -triple powerpc64le-unknown-unknown -fopenmp-targets=powerpc64le-ibm-linux-gnu -emit-llvm-bc %s -o %t-ppc-host.bc 19 // RUN: %clang_cc1 -verify -fopenmp -fopenmp-version=45 -x c++ -triple powerpc64le-unknown-unknown -fopenmp-targets=powerpc64le-ibm-linux-gnu -emit-llvm %s -fopenmp-is-device -fopenmp-host-ir-file-path %t-ppc-host.bc -o - | FileCheck -allow-deprecated-dag-overlap %s --check-prefix TCHECK --check-prefix TCHECK-64 20 // RUN: %clang_cc1 -fopenmp -fopenmp-version=45 -x c++ -std=c++11 -triple powerpc64le-unknown-unknown -fopenmp-targets=powerpc64le-ibm-linux-gnu -emit-pch -fopenmp-is-device -fopenmp-host-ir-file-path %t-ppc-host.bc -o %t %s 21 // RUN: %clang_cc1 -fopenmp -fopenmp-version=45 -x c++ -triple powerpc64le-unknown-unknown -fopenmp-targets=powerpc64le-ibm-linux-gnu -std=c++11 -fopenmp-is-device -fopenmp-host-ir-file-path %t-ppc-host.bc -include-pch %t -verify %s -emit-llvm -o - | FileCheck -allow-deprecated-dag-overlap %s --check-prefix TCHECK --check-prefix TCHECK-64 22 // RUN: %clang_cc1 -verify -fopenmp -fopenmp-version=45 -x c++ -triple i386-unknown-unknown -fopenmp-targets=i386-pc-linux-gnu -emit-llvm-bc %s -o %t-x86-host.bc 23 // RUN: %clang_cc1 -verify -fopenmp -fopenmp-version=45 -x c++ -triple i386-unknown-unknown -fopenmp-targets=i386-pc-linux-gnu -emit-llvm %s -fopenmp-is-device -fopenmp-host-ir-file-path %t-x86-host.bc -o - | FileCheck -allow-deprecated-dag-overlap %s --check-prefix TCHECK --check-prefix TCHECK-32 24 // RUN: %clang_cc1 -fopenmp -fopenmp-version=45 -x c++ -std=c++11 -triple i386-unknown-unknown -fopenmp-targets=i386-pc-linux-gnu -emit-pch -fopenmp-is-device -fopenmp-host-ir-file-path %t-x86-host.bc -o %t %s 25 // RUN: %clang_cc1 -fopenmp -fopenmp-version=45 -x c++ -triple i386-unknown-unknown -fopenmp-targets=i386-pc-linux-gnu -std=c++11 -fopenmp-is-device -fopenmp-host-ir-file-path %t-x86-host.bc -include-pch %t -verify %s -emit-llvm -o - | FileCheck -allow-deprecated-dag-overlap %s --check-prefix TCHECK --check-prefix TCHECK-32 26 27 // RUN: %clang_cc1 -verify -fopenmp-simd -fopenmp-version=45 -x c++ -triple powerpc64le-unknown-unknown -fopenmp-targets=powerpc64le-ibm-linux-gnu -emit-llvm-bc %s -o %t-ppc-host.bc 28 // RUN: %clang_cc1 -verify -fopenmp-simd -fopenmp-version=45 -x c++ -triple powerpc64le-unknown-unknown -fopenmp-targets=powerpc64le-ibm-linux-gnu -emit-llvm %s -fopenmp-is-device -fopenmp-host-ir-file-path %t-ppc-host.bc -o - | FileCheck -allow-deprecated-dag-overlap --check-prefix SIMD-ONLY1 %s 29 // RUN: %clang_cc1 -fopenmp-simd -fopenmp-version=45 -x c++ -std=c++11 -triple powerpc64le-unknown-unknown -fopenmp-targets=powerpc64le-ibm-linux-gnu -emit-pch -fopenmp-is-device -fopenmp-host-ir-file-path %t-ppc-host.bc -o %t %s 30 // RUN: %clang_cc1 -fopenmp-simd -fopenmp-version=45 -x c++ -triple powerpc64le-unknown-unknown -fopenmp-targets=powerpc64le-ibm-linux-gnu -std=c++11 -fopenmp-is-device -fopenmp-host-ir-file-path %t-ppc-host.bc -include-pch %t -verify %s -emit-llvm -o - | FileCheck -allow-deprecated-dag-overlap --check-prefix SIMD-ONLY1 %s 31 // RUN: %clang_cc1 -verify -fopenmp-simd -fopenmp-version=45 -x c++ -triple i386-unknown-unknown -fopenmp-targets=i386-pc-linux-gnu -emit-llvm-bc %s -o %t-x86-host.bc 32 // RUN: %clang_cc1 -verify -fopenmp-simd -fopenmp-version=45 -x c++ -triple i386-unknown-unknown -fopenmp-targets=i386-pc-linux-gnu -emit-llvm %s -fopenmp-is-device -fopenmp-host-ir-file-path %t-x86-host.bc -o - | FileCheck -allow-deprecated-dag-overlap --check-prefix SIMD-ONLY1 %s 33 // RUN: %clang_cc1 -fopenmp-simd -fopenmp-version=45 -x c++ -std=c++11 -triple i386-unknown-unknown -fopenmp-targets=i386-pc-linux-gnu -emit-pch -fopenmp-is-device -fopenmp-host-ir-file-path %t-x86-host.bc -o %t %s 34 // RUN: %clang_cc1 -fopenmp-simd -fopenmp-version=45 -x c++ -triple i386-unknown-unknown -fopenmp-targets=i386-pc-linux-gnu -std=c++11 -fopenmp-is-device -fopenmp-host-ir-file-path %t-x86-host.bc -include-pch %t -verify %s -emit-llvm -o - | FileCheck -allow-deprecated-dag-overlap --check-prefix SIMD-ONLY1 %s 35 // SIMD-ONLY1-NOT: {{__kmpc|__tgt}} 36 // expected-no-diagnostics 37 38 #ifndef HEADER 39 #define HEADER 40 41 // CHECK-DAG: %struct.ident_t = type { i32, i32, i32, i32, i8* } 42 // CHECK-DAG: [[STR:@.+]] = private unnamed_addr constant [23 x i8] c";unknown;unknown;0;0;;\00" 43 // CHECK-DAG: [[DEF_LOC:@.+]] = private unnamed_addr global %struct.ident_t { i32 0, i32 2, i32 0, i32 0, i8* getelementptr inbounds ([23 x i8], [23 x i8]* [[STR]], i32 0, i32 0) } 44 45 // CHECK-DAG: [[TT:%.+]] = type { i64, i8 } 46 // CHECK-DAG: [[S1:%.+]] = type { double } 47 // CHECK-DAG: [[ENTTY:%.+]] = type { i8*, i8*, i[[SZ:32|64]], i32, i32 } 48 // CHECK-DAG: [[DEVTY:%.+]] = type { i8*, i8*, [[ENTTY]]*, [[ENTTY]]* } 49 // CHECK-DAG: [[DSCTY:%.+]] = type { i32, [[DEVTY]]*, [[ENTTY]]*, [[ENTTY]]* } 50 51 // TCHECK: [[ENTTY:%.+]] = type { i8*, i8*, i{{32|64}}, i32, i32 } 52 53 // CHECK-DAG: $[[REGFN:\.omp_offloading\..+]] = comdat 54 55 // We have 8 target regions, but only 7 that actually will generate offloading 56 // code, only 6 will have mapped arguments, and only 4 have all-constant map 57 // sizes. 58 59 // CHECK-DAG: [[SIZET:@.+]] = private unnamed_addr constant [3 x i64] [i64 2, i64 4, i64 4] 60 // CHECK-DAG: [[MAPT:@.+]] = private unnamed_addr constant [3 x i64] [i64 288, i64 800, i64 800] 61 // CHECK-DAG: [[SIZET2:@.+]] = private unnamed_addr constant [1 x i64] [i64 2] 62 // CHECK-DAG: [[MAPT2:@.+]] = private unnamed_addr constant [1 x i64] [i64 800] 63 // CHECK-DAG: [[SIZET3:@.+]] = private unnamed_addr constant [2 x i64] [i64 4, i64 2] 64 // CHECK-DAG: [[MAPT3:@.+]] = private unnamed_addr constant [2 x i64] [i64 800, i64 800] 65 // CHECK-DAG: [[MAPT4:@.+]] = private unnamed_addr constant [9 x i64] [i64 800, i64 547, i64 800, i64 547, i64 547, i64 800, i64 800, i64 547, i64 547] 66 // CHECK-DAG: [[MAPT5:@.+]] = private unnamed_addr constant [6 x i64] [i64 32, i64 281474976711171, i64 800, i64 800, i64 800, i64 547] 67 // CHECK-DAG: [[SIZET6:@.+]] = private unnamed_addr constant [5 x i64] [i64 4, i64 4, i64 2, i64 1, i64 40] 68 // CHECK-DAG: [[MAPT6:@.+]] = private unnamed_addr constant [5 x i64] [i64 800, i64 800, i64 800, i64 800, i64 547] 69 // CHECK-DAG: [[SIZET7:@.+]] = private unnamed_addr constant [3 x i64] [i64 4, i64 2, i64 40] 70 // CHECK-DAG: [[MAPT7:@.+]] = private unnamed_addr constant [3 x i64] [i64 800, i64 800, i64 547] 71 // CHECK-DAG: @{{.*}} = weak constant i8 0 72 // CHECK-DAG: @{{.*}} = weak constant i8 0 73 // CHECK-DAG: @{{.*}} = weak constant i8 0 74 // CHECK-DAG: @{{.*}} = weak constant i8 0 75 // CHECK-DAG: @{{.*}} = weak constant i8 0 76 // CHECK-DAG: @{{.*}} = weak constant i8 0 77 // CHECK-DAG: @{{.*}} = weak constant i8 0 78 79 // TCHECK: @{{.+}} = weak constant [[ENTTY]] 80 // TCHECK: @{{.+}} = weak constant [[ENTTY]] 81 // TCHECK: @{{.+}} = weak constant [[ENTTY]] 82 // TCHECK: @{{.+}} = weak constant [[ENTTY]] 83 // TCHECK: @{{.+}} = weak constant [[ENTTY]] 84 // TCHECK: @{{.+}} = weak constant [[ENTTY]] 85 // TCHECK: @{{.+}} = weak constant [[ENTTY]] 86 // TCHECK-NOT: @{{.+}} = weak constant [[ENTTY]] 87 88 // Check if offloading descriptor is created. 89 // CHECK: [[ENTBEGIN:@.+]] = external constant [[ENTTY]] 90 // CHECK: [[ENTEND:@.+]] = external constant [[ENTTY]] 91 // CHECK: [[DEVBEGIN:@.+]] = extern_weak constant i8 92 // CHECK: [[DEVEND:@.+]] = extern_weak constant i8 93 // CHECK: [[IMAGES:@.+]] = internal unnamed_addr constant [1 x [[DEVTY]]] [{{.+}} { i8* [[DEVBEGIN]], i8* [[DEVEND]], [[ENTTY]]* [[ENTBEGIN]], [[ENTTY]]* [[ENTEND]] }], comdat($[[REGFN]]) 94 // CHECK: [[DESC:@.+]] = internal constant [[DSCTY]] { i32 1, [[DEVTY]]* getelementptr inbounds ([1 x [[DEVTY]]], [1 x [[DEVTY]]]* [[IMAGES]], i32 0, i32 0), [[ENTTY]]* [[ENTBEGIN]], [[ENTTY]]* [[ENTEND]] }, comdat($[[REGFN]]) 95 96 // Check target registration is registered as a Ctor. 97 // CHECK: appending global [2 x { i32, void ()*, i8* }] [{ i32, void ()*, i8* } { i32 0, void ()* @.omp_offloading.requires_reg, i8* null }, { i32, void ()*, i8* } { i32 0, void ()* @[[REGFN]], i8* bitcast (void ()* @[[REGFN]] to i8*) }] 98 99 100 template<typename tx, typename ty> 101 struct TT{ 102 tx X; 103 ty Y; 104 }; 105 106 int global; 107 108 // CHECK: define {{.*}}[[FOO:@.+]]( 109 int foo(int n) { 110 int a = 0; 111 short aa = 0; 112 float b[10]; 113 float bn[n]; 114 double c[5][10]; 115 double cn[5][n]; 116 TT<long long, char> d; 117 118 // CHECK-DAG: [[RET:%.+]] = call i32 @__tgt_target_teams_nowait(i64 -1, i8* @{{[^,]+}}, i32 3, i8** [[BP:%[^,]+]], i8** [[P:%[^,]+]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* [[SIZET]], i32 0, i32 0), i64* getelementptr inbounds ([3 x i64], [3 x i64]* [[MAPT]], i32 0, i32 0), i32 {{[^,]+}}, i32 {{[^)]+}}) 119 // CHECK-DAG: [[BP]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[BPR:%[^,]+]], i32 0, i32 0 120 // CHECK-DAG: [[P]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[PR:%[^,]+]], i32 0, i32 0 121 // CHECK-DAG: [[BPADDR0:%.+]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[BPR]], i32 0, i32 [[IDX0:[0-9]+]] 122 // CHECK-DAG: [[PADDR0:%.+]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[PR]], i32 0, i32 [[IDX0]] 123 // CHECK-DAG: [[CBPADDR0:%.+]] = bitcast i8** [[BPADDR0]] to i[[SZ]]* 124 // CHECK-DAG: [[CPADDR0:%.+]] = bitcast i8** [[PADDR0]] to i[[SZ]]* 125 // CHECK-DAG: store i[[SZ]] %{{.+}}, i[[SZ]]* [[CBPADDR0]] 126 // CHECK-DAG: store i[[SZ]] %{{.+}}, i[[SZ]]* [[CPADDR0]] 127 // CHECK-DAG: [[BPADDR1:%.+]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[BPR]], i32 0, i32 [[IDX1:[0-9]+]] 128 // CHECK-DAG: [[PADDR1:%.+]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[PR]], i32 0, i32 [[IDX1]] 129 // CHECK-DAG: [[CBPADDR1:%.+]] = bitcast i8** [[BPADDR1]] to i[[SZ]]* 130 // CHECK-DAG: [[CPADDR1:%.+]] = bitcast i8** [[PADDR1]] to i[[SZ]]* 131 // CHECK-DAG: store i[[SZ]] %{{.+}}, i[[SZ]]* [[CBPADDR1]] 132 // CHECK-DAG: store i[[SZ]] %{{.+}}, i[[SZ]]* [[CPADDR1]] 133 // CHECK-DAG: [[BPADDR2:%.+]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[BPR]], i32 0, i32 [[IDX1:[0-9]+]] 134 // CHECK-DAG: [[PADDR2:%.+]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[PR]], i32 0, i32 [[IDX1]] 135 // CHECK-DAG: [[CBPADDR2:%.+]] = bitcast i8** [[BPADDR2]] to i[[SZ]]* 136 // CHECK-DAG: [[CPADDR2:%.+]] = bitcast i8** [[PADDR2]] to i[[SZ]]* 137 // CHECK-DAG: store i[[SZ]] %{{.+}}, i[[SZ]]* [[CBPADDR2]] 138 // CHECK-DAG: store i[[SZ]] %{{.+}}, i[[SZ]]* [[CPADDR2]] 139 // CHECK-NEXT: [[ERROR:%.+]] = icmp ne i32 [[RET]], 0 140 // CHECK-NEXT: br i1 [[ERROR]], label %[[FAIL:[^,]+]], label %[[END:[^,]+]] 141 // CHECK: [[FAIL]] 142 // CHECK: call void [[HVT0:@.+]](i[[SZ]] {{[^,]+}}, i[[SZ]] {{[^,]+}}, i[[SZ]] {{[^)]+}}) 143 // CHECK-NEXT: br label %[[END]] 144 // CHECK: [[END]] 145 #pragma omp target teams distribute simd num_teams(a) thread_limit(a) firstprivate(aa) simdlen(16) nowait 146 for (int i = 0; i < 10; ++i) { 147 } 148 149 // CHECK: call void [[HVT1:@.+]](i[[SZ]] {{[^,]+}}) 150 #pragma omp target teams distribute simd if(target: 0) safelen(32) linear(a) 151 for (a = 0; a < 10; ++a) { 152 a += 1; 153 } 154 155 // CHECK-DAG: [[RET:%.+]] = call i32 @__tgt_target_teams(i64 -1, i8* @{{[^,]+}}, i32 1, i8** [[BP:%[^,]+]], i8** [[P:%[^,]+]], i64* getelementptr inbounds ([1 x i64], [1 x i64]* [[SIZET2]], i32 0, i32 0), i64* getelementptr inbounds ([1 x i64], [1 x i64]* [[MAPT2]], i32 0, i32 0), i32 0, i32 1) 156 // CHECK-DAG: [[BP]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[BPR:%[^,]+]], i32 0, i32 0 157 // CHECK-DAG: [[P]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[PR:%[^,]+]], i32 0, i32 0 158 // CHECK-DAG: [[BPADDR0:%.+]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[BPR]], i32 0, i32 [[IDX0:[0-9]+]] 159 // CHECK-DAG: [[PADDR0:%.+]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[PR]], i32 0, i32 [[IDX0]] 160 // CHECK-DAG: [[CBPADDR0:%.+]] = bitcast i8** [[BPADDR0]] to i[[SZ]]* 161 // CHECK-DAG: [[CPADDR0:%.+]] = bitcast i8** [[PADDR0]] to i[[SZ]]* 162 // CHECK-DAG: store i[[SZ]] %{{.+}}, i[[SZ]]* [[CBPADDR0]] 163 // CHECK-DAG: store i[[SZ]] %{{.+}}, i[[SZ]]* [[CPADDR0]] 164 165 // CHECK-NEXT: [[ERROR:%.+]] = icmp ne i32 [[RET]], 0 166 // CHECK-NEXT: br i1 [[ERROR]], label %[[FAIL:[^,]+]], label %[[END:[^,]+]] 167 // CHECK: [[FAIL]] 168 // CHECK: call void [[HVT2:@.+]](i[[SZ]] {{[^,]+}}) 169 // CHECK-NEXT: br label %[[END]] 170 // CHECK: [[END]] 171 #pragma omp target teams distribute simd if(target: 1) 172 for (int i = 0; i < 10; ++i) { 173 aa += 1; 174 } 175 176 // CHECK: [[IF:%.+]] = icmp sgt i32 {{[^,]+}}, 10 177 // CHECK: br i1 [[IF]], label %[[IFTHEN:[^,]+]], label %[[IFELSE:[^,]+]] 178 // CHECK: [[IFTHEN]] 179 // CHECK-DAG: [[RET:%.+]] = call i32 @__tgt_target_teams(i64 -1, i8* @{{[^,]+}}, i32 2, i8** [[BPR:%[^,]+]], i8** [[PR:%[^,]+]], i64* getelementptr inbounds ([2 x i64], [2 x i64]* [[SIZET3]], i32 0, i32 0), i64* getelementptr inbounds ([2 x i64], [2 x i64]* [[MAPT3]], i32 0, i32 0), i32 0, i32 1) 180 // CHECK-DAG: [[BPR]] = getelementptr inbounds [2 x i8*], [2 x i8*]* [[BP:%[^,]+]], i32 0, i32 0 181 // CHECK-DAG: [[PR]] = getelementptr inbounds [2 x i8*], [2 x i8*]* [[P:%[^,]+]], i32 0, i32 0 182 183 // CHECK-DAG: [[BPADDR0:%.+]] = getelementptr inbounds [2 x i8*], [2 x i8*]* [[BP]], i32 0, i32 0 184 // CHECK-DAG: [[PADDR0:%.+]] = getelementptr inbounds [2 x i8*], [2 x i8*]* [[P]], i32 0, i32 0 185 // CHECK-DAG: [[CBPADDR0:%.+]] = bitcast i8** [[BPADDR0]] to i[[SZ]]* 186 // CHECK-DAG: [[CPADDR0:%.+]] = bitcast i8** [[PADDR0]] to i[[SZ]]* 187 // CHECK-DAG: store i[[SZ]] %{{.+}}, i[[SZ]]* [[CBPADDR0]] 188 // CHECK-DAG: store i[[SZ]] %{{.+}}, i[[SZ]]* [[CPADDR0]] 189 190 // CHECK-DAG: [[BPADDR1:%.+]] = getelementptr inbounds [2 x i8*], [2 x i8*]* [[BP]], i32 0, i32 1 191 // CHECK-DAG: [[PADDR1:%.+]] = getelementptr inbounds [2 x i8*], [2 x i8*]* [[P]], i32 0, i32 1 192 // CHECK-DAG: [[CBPADDR1:%.+]] = bitcast i8** [[BPADDR1]] to i[[SZ]]* 193 // CHECK-DAG: [[CPADDR1:%.+]] = bitcast i8** [[PADDR1]] to i[[SZ]]* 194 // CHECK-DAG: store i[[SZ]] %{{.+}}, i[[SZ]]* [[CBPADDR1]] 195 // CHECK-DAG: store i[[SZ]] %{{.+}}, i[[SZ]]* [[CPADDR1]] 196 // CHECK: [[ERROR:%.+]] = icmp ne i32 [[RET]], 0 197 // CHECK-NEXT: br i1 [[ERROR]], label %[[FAIL:.+]], label %[[END:[^,]+]] 198 // CHECK: [[FAIL]] 199 // CHECK: call void [[HVT3:@.+]]({{[^,]+}}, {{[^,]+}}) 200 // CHECK-NEXT: br label %[[END]] 201 // CHECK: [[END]] 202 // CHECK-NEXT: br label %[[IFEND:.+]] 203 // CHECK: [[IFELSE]] 204 // CHECK: call void [[HVT3]]({{[^,]+}}, {{[^,]+}}) 205 // CHECK-NEXT: br label %[[IFEND]] 206 // CHECK: [[IFEND]] 207 #pragma omp target teams distribute simd if(target: n>10) 208 for (int i = 0; i < 10; ++i) { 209 a += 1; 210 aa += 1; 211 } 212 213 // We capture 3 VLA sizes in this target region 214 // CHECK-64: [[A_VAL:%.+]] = load i32, i32* %{{.+}}, 215 // CHECK-64: [[A_ADDR:%.+]] = bitcast i[[SZ]]* [[A_CADDR:%.+]] to i32* 216 // CHECK-64: store i32 [[A_VAL]], i32* [[A_ADDR]], 217 // CHECK-64: [[A_CVAL:%.+]] = load i[[SZ]], i[[SZ]]* [[A_CADDR]], 218 219 // CHECK-32: [[A_VAL:%.+]] = load i32, i32* %{{.+}}, 220 // CHECK-32: store i32 [[A_VAL]], i32* [[A_CADDR:%.+]], 221 // CHECK-32: [[A_CVAL:%.+]] = load i[[SZ]], i[[SZ]]* [[A_CADDR]], 222 223 // CHECK: [[IF:%.+]] = icmp sgt i32 {{[^,]+}}, 20 224 // CHECK: br i1 [[IF]], label %[[TRY:[^,]+]], label %[[FAIL:[^,]+]] 225 // CHECK: [[TRY]] 226 // CHECK-64: [[BNSIZE:%.+]] = mul nuw i64 [[VLA0:%.+]], 4 227 // CHECK-32: [[BNSZSIZE:%.+]] = mul nuw i32 [[VLA0:%.+]], 4 228 // CHECK-32: [[BNSIZE:%.+]] = sext i32 [[BNSZSIZE]] to i64 229 // CHECK: [[CNELEMSIZE2:%.+]] = mul nuw i[[SZ]] 5, [[VLA1:%.+]] 230 // CHECK-64: [[CNSIZE:%.+]] = mul nuw i64 [[CNELEMSIZE2]], 8 231 // CHECK-32: [[CNSZSIZE:%.+]] = mul nuw i32 [[CNELEMSIZE2]], 8 232 // CHECK-32: [[CNSIZE:%.+]] = sext i32 [[CNSZSIZE]] to i64 233 234 // CHECK-DAG: [[RET:%.+]] = call i32 @__tgt_target_teams(i64 -1, i8* @{{[^,]+}}, i32 9, i8** [[BPR:%[^,]+]], i8** [[PR:%[^,]+]], i64* [[SR:%[^,]+]], i64* getelementptr inbounds ([9 x i64], [9 x i64]* [[MAPT4]], i32 0, i32 0), i32 0, i32 1) 235 // CHECK-DAG: [[BPR]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[BP:%[^,]+]], i32 0, i32 0 236 // CHECK-DAG: [[PR]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[P:%[^,]+]], i32 0, i32 0 237 // CHECK-DAG: [[SR]] = getelementptr inbounds [9 x i64], [9 x i64]* [[S:%[^,]+]], i32 0, i32 0 238 239 // CHECK-DAG: [[SADDR0:%.+]] = getelementptr inbounds [9 x i64], [9 x i64]* [[S]], i32 0, i32 [[IDX0:[0-9]+]] 240 // CHECK-DAG: [[BPADDR0:%.+]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[BP]], i32 0, i32 [[IDX0]] 241 // CHECK-DAG: [[PADDR0:%.+]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[P]], i32 0, i32 [[IDX0]] 242 // CHECK-DAG: [[SADDR1:%.+]] = getelementptr inbounds [9 x i64], [9 x i64]* [[S]], i32 0, i32 [[IDX1:[0-9]+]] 243 // CHECK-DAG: [[BPADDR1:%.+]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[BP]], i32 0, i32 [[IDX1]] 244 // CHECK-DAG: [[PADDR1:%.+]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[P]], i32 0, i32 [[IDX1]] 245 // CHECK-DAG: [[SADDR2:%.+]] = getelementptr inbounds [9 x i64], [9 x i64]* [[S]], i32 0, i32 [[IDX2:[0-9]+]] 246 // CHECK-DAG: [[BPADDR2:%.+]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[BP]], i32 0, i32 [[IDX2]] 247 // CHECK-DAG: [[PADDR2:%.+]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[P]], i32 0, i32 [[IDX2]] 248 // CHECK-DAG: [[SADDR3:%.+]] = getelementptr inbounds [9 x i64], [9 x i64]* [[S]], i32 0, i32 [[IDX3:[0-9]+]] 249 // CHECK-DAG: [[BPADDR3:%.+]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[BP]], i32 0, i32 [[IDX3]] 250 // CHECK-DAG: [[PADDR3:%.+]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[P]], i32 0, i32 [[IDX3]] 251 // CHECK-DAG: [[SADDR4:%.+]] = getelementptr inbounds [9 x i64], [9 x i64]* [[S]], i32 0, i32 [[IDX4:[0-9]+]] 252 // CHECK-DAG: [[BPADDR4:%.+]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[BP]], i32 0, i32 [[IDX4]] 253 // CHECK-DAG: [[PADDR4:%.+]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[P]], i32 0, i32 [[IDX4]] 254 // CHECK-DAG: [[SADDR5:%.+]] = getelementptr inbounds [9 x i64], [9 x i64]* [[S]], i32 0, i32 [[IDX5:[0-9]+]] 255 // CHECK-DAG: [[BPADDR5:%.+]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[BP]], i32 0, i32 [[IDX5]] 256 // CHECK-DAG: [[PADDR5:%.+]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[P]], i32 0, i32 [[IDX5]] 257 // CHECK-DAG: [[SADDR6:%.+]] = getelementptr inbounds [9 x i64], [9 x i64]* [[S]], i32 0, i32 [[IDX6:[0-9]+]] 258 // CHECK-DAG: [[BPADDR6:%.+]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[BP]], i32 0, i32 [[IDX6]] 259 // CHECK-DAG: [[PADDR6:%.+]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[P]], i32 0, i32 [[IDX6]] 260 // CHECK-DAG: [[SADDR7:%.+]] = getelementptr inbounds [9 x i64], [9 x i64]* [[S]], i32 0, i32 [[IDX7:[0-9]+]] 261 // CHECK-DAG: [[BPADDR7:%.+]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[BP]], i32 0, i32 [[IDX7]] 262 // CHECK-DAG: [[PADDR7:%.+]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[P]], i32 0, i32 [[IDX7]] 263 // CHECK-DAG: [[SADDR8:%.+]] = getelementptr inbounds [9 x i64], [9 x i64]* [[S]], i32 0, i32 [[IDX8:[0-9]+]] 264 // CHECK-DAG: [[BPADDR8:%.+]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[BP]], i32 0, i32 [[IDX8]] 265 // CHECK-DAG: [[PADDR8:%.+]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[P]], i32 0, i32 [[IDX8]] 266 267 // The names below are not necessarily consistent with the names used for the 268 // addresses above as some are repeated. 269 // CHECK-DAG: store i[[SZ]] [[VLA0]], i[[SZ]]* [[CBPADDR0:%.+]], 270 // CHECK-DAG: store i[[SZ]] [[VLA0]], i[[SZ]]* [[CPADDR0:%.+]], 271 // CHECK-DAG: [[CBPADDR0]] = bitcast i8** {{%[^,]+}} to i[[SZ]]* 272 // CHECK-DAG: [[CPADDR0]] = bitcast i8** {{%[^,]+}} to i[[SZ]]* 273 // CHECK-DAG: store i64 {{4|8}}, i64* {{%[^,]+}} 274 275 // CHECK-DAG: store i[[SZ]] [[VLA1]], i[[SZ]]* [[CBPADDR1:%.+]], 276 // CHECK-DAG: store i[[SZ]] [[VLA1]], i[[SZ]]* [[CPADDR1:%.+]], 277 // CHECK-DAG: [[CBPADDR1]] = bitcast i8** {{%[^,]+}} to i[[SZ]]* 278 // CHECK-DAG: [[CPADDR1]] = bitcast i8** {{%[^,]+}} to i[[SZ]]* 279 // CHECK-DAG: store i64 {{4|8}}, i64* {{%[^,]+}} 280 281 // CHECK-DAG: store i[[SZ]] 5, i[[SZ]]* [[CBPADDR2:%.+]], 282 // CHECK-DAG: store i[[SZ]] 5, i[[SZ]]* [[CPADDR2:%.+]], 283 // CHECK-DAG: [[CBPADDR2]] = bitcast i8** {{%[^,]+}} to i[[SZ]]* 284 // CHECK-DAG: [[CPADDR2]] = bitcast i8** {{%[^,]+}} to i[[SZ]]* 285 // CHECK-DAG: store i64 {{4|8}}, i64* {{%[^,]+}} 286 287 // CHECK-DAG: store i[[SZ]] [[A_CVAL]], i[[SZ]]* [[CBPADDR3:%.+]], 288 // CHECK-DAG: store i[[SZ]] [[A_CVAL]], i[[SZ]]* [[CPADDR3:%.+]], 289 // CHECK-DAG: [[CBPADDR3]] = bitcast i8** {{%[^,]+}} to i[[SZ]]* 290 // CHECK-DAG: [[CPADDR3]] = bitcast i8** {{%[^,]+}} to i[[SZ]]* 291 // CHECK-DAG: store i64 4, i64* {{%[^,]+}} 292 293 // CHECK-DAG: store [10 x float]* %{{.+}}, [10 x float]** [[CBPADDR4:%.+]], 294 // CHECK-DAG: store [10 x float]* %{{.+}}, [10 x float]** [[CPADDR4:%.+]], 295 // CHECK-DAG: [[CBPADDR4]] = bitcast i8** {{%[^,]+}} to [10 x float]** 296 // CHECK-DAG: [[CPADDR4]] = bitcast i8** {{%[^,]+}} to [10 x float]** 297 // CHECK-DAG: store i64 40, i64* {{%[^,]+}} 298 299 // CHECK-DAG: store float* %{{.+}}, float** [[CBPADDR5:%.+]], 300 // CHECK-DAG: store float* %{{.+}}, float** [[CPADDR5:%.+]], 301 // CHECK-DAG: [[CBPADDR5]] = bitcast i8** {{%[^,]+}} to float** 302 // CHECK-DAG: [[CPADDR5]] = bitcast i8** {{%[^,]+}} to float** 303 // CHECK-DAG: store i64 [[BNSIZE]], i64* {{%[^,]+}} 304 305 // CHECK-DAG: store [5 x [10 x double]]* %{{.+}}, [5 x [10 x double]]** [[CBPADDR6:%.+]], 306 // CHECK-DAG: store [5 x [10 x double]]* %{{.+}}, [5 x [10 x double]]** [[CPADDR6:%.+]], 307 // CHECK-DAG: [[CBPADDR6]] = bitcast i8** {{%[^,]+}} to [5 x [10 x double]]** 308 // CHECK-DAG: [[CPADDR6]] = bitcast i8** {{%[^,]+}} to [5 x [10 x double]]** 309 // CHECK-DAG: store i64 400, i64* {{%[^,]+}} 310 311 // CHECK-DAG: store double* %{{.+}}, double** [[CBPADDR7:%.+]], 312 // CHECK-DAG: store double* %{{.+}}, double** [[CPADDR7:%.+]], 313 // CHECK-DAG: [[CBPADDR7]] = bitcast i8** {{%[^,]+}} to double** 314 // CHECK-DAG: [[CPADDR7]] = bitcast i8** {{%[^,]+}} to double** 315 // CHECK-DAG: store i64 [[CNSIZE]], i64* {{%[^,]+}} 316 317 // CHECK-DAG: store [[TT]]* %{{.+}}, [[TT]]** [[CBPADDR8:%.+]], 318 // CHECK-DAG: store [[TT]]* %{{.+}}, [[TT]]** [[CPADDR8:%.+]], 319 // CHECK-DAG: [[CBPADDR8]] = bitcast i8** {{%[^,]+}} to [[TT]]** 320 // CHECK-DAG: [[CPADDR8]] = bitcast i8** {{%[^,]+}} to [[TT]]** 321 // CHECK-DAG: store i64 {{12|16}}, i64* {{%[^,]+}} 322 323 // CHECK-NEXT: [[ERROR:%.+]] = icmp ne i32 [[RET]], 0 324 // CHECK-NEXT: br i1 [[ERROR]], label %[[FAIL:[^,]+]], label %[[END:[^,]+]] 325 326 // CHECK: [[FAIL]] 327 // CHECK: call void [[HVT4:@.+]]({{[^,]+}}, {{[^,]+}}, {{[^,]+}}, {{[^,]+}}, {{[^,]+}}, {{[^,]+}}, {{[^,]+}}, {{[^,]+}}, {{[^,]+}}) 328 // CHECK-NEXT: br label %[[END]] 329 // CHECK: [[END]] 330 #pragma omp target teams distribute simd if(target: n>20) aligned(b) 331 for (int i = 0; i < 10; ++i) { 332 a += 1; 333 b[2] += 1.0; 334 bn[3] += 1.0; 335 c[1][2] += 1.0; 336 cn[1][3] += 1.0; 337 d.X += 1; 338 d.Y += 1; 339 } 340 341 return a; 342 } 343 344 // Check that the offloading functions are emitted and that the arguments are 345 // correct and loaded correctly for the target regions in foo(). 346 347 // CHECK: define internal void [[HVT0]](i[[SZ]] {{[^,]+}}, i[[SZ]] {{[^,]+}}, i[[SZ]] {{[^)]+}}) 348 // CHECK: call {{.*}}void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* [[DEF_LOC]], i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i[[SZ]])* [[OMP_OUTLINED:@.+]] to void (i32*, i32*, ...)*), i[[SZ]] {{[^)]+}}) 349 // 350 // 351 // CHECK: define internal {{.*}}void [[OMP_OUTLINED]](i32* noalias %.global_tid., i32* noalias %.bound_tid., i[[SZ]] {{[^)]+}}) 352 // CHECK: alloca i[[SZ]], 353 // CHECK: bitcast i[[SZ]]* {{.+}} to i16* 354 // CHECK: ret void 355 // CHECK-NEXT: } 356 357 358 // CHECK: define internal void [[HVT1]](i[[SZ]] %{{.+}}) 359 // Create stack storage and store argument in there. 360 // CHECK: [[AA_ADDR:%.+]] = alloca i[[SZ]], align 361 // CHECK: [[AA_CASTED:%.+]] = alloca i[[SZ]], align 362 // CHECK: store i[[SZ]] %{{.+}}, i[[SZ]]* [[AA_ADDR]], align 363 // CHECK-64: [[AA_CADDR:%.+]] = bitcast i[[SZ]]* [[AA_ADDR]] to i32* 364 // CHECK-64: [[AA:%.+]] = load i32, i32* [[AA_CADDR]], align 365 // CHECK-32: [[AA:%.+]] = load i32, i32* [[AA_ADDR]], align 366 // CHECK-64: [[AA_C:%.+]] = bitcast i[[SZ]]* [[AA_CASTED]] to i32* 367 // CHECK-64: store i32 [[AA]], i32* [[AA_C]], align 368 // CHECK-32: store i32 [[AA]], i32* [[AA_CASTED]], align 369 // CHECK: [[PARAM:%.+]] = load i[[SZ]], i[[SZ]]* [[AA_CASTED]], align 370 // CHECK: call {{.*}}void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* [[DEF_LOC]], i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i[[SZ]])* [[OMP_OUTLINED1:@.+]] to void (i32*, i32*, ...)*), i[[SZ]] [[PARAM]]) 371 // 372 // 373 // CHECK: define internal {{.*}}void [[OMP_OUTLINED1]](i32* noalias %.global_tid., i32* noalias %.bound_tid., i[[SZ]] %{{.+}}) 374 // CHECK: [[AA_ADDR:%.+]] = alloca i[[SZ]], align 375 // CHECK: store i[[SZ]] %{{.+}}, i[[SZ]]* [[AA_ADDR]], align 376 // CHECK-64: [[AA_CADDR:%.+]] = bitcast i[[SZ]]* [[AA_ADDR]] to i32* 377 // CHECK-64: store i32 10, i32* [[AA_CADDR]], align 378 // CHECK-32: store i32 10, i32* [[AA_ADDR]], align 379 // CHECK: ret void 380 // CHECK-NEXT: } 381 382 // CHECK: define internal void [[HVT2]](i[[SZ]] %{{.+}}) 383 // Create stack storage and store argument in there. 384 // CHECK: [[AA_ADDR:%.+]] = alloca i[[SZ]], align 385 // CHECK: [[AA_CASTED:%.+]] = alloca i[[SZ]], align 386 // CHECK: store i[[SZ]] %{{.+}}, i[[SZ]]* [[AA_ADDR]], align 387 // CHECK: [[AA_CADDR:%.+]] = bitcast i[[SZ]]* [[AA_ADDR]] to i16* 388 // CHECK: [[AA:%.+]] = load i16, i16* [[AA_CADDR]], align 389 // CHECK: [[AA_C:%.+]] = bitcast i[[SZ]]* [[AA_CASTED]] to i16* 390 // CHECK: store i16 [[AA]], i16* [[AA_C]], align 391 // CHECK: [[PARAM:%.+]] = load i[[SZ]], i[[SZ]]* [[AA_CASTED]], align 392 // CHECK: call {{.*}}void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* [[DEF_LOC]], i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i[[SZ]])* [[OMP_OUTLINED2:@.+]] to void (i32*, i32*, ...)*), i[[SZ]] [[PARAM]]) 393 // 394 // 395 // CHECK: define internal {{.*}}void [[OMP_OUTLINED2]](i32* noalias %.global_tid., i32* noalias %.bound_tid., i[[SZ]] %{{.+}}) 396 // CHECK: [[AA_ADDR:%.+]] = alloca i[[SZ]], align 397 // CHECK: store i[[SZ]] %{{.+}}, i[[SZ]]* [[AA_ADDR]], align 398 // CHECK: [[AA_CADDR:%.+]] = bitcast i[[SZ]]* [[AA_ADDR]] to i16* 399 // CHECK: [[AA:%.+]] = load i16, i16* [[AA_CADDR]], align 400 // CHECK: ret void 401 // CHECK-NEXT: } 402 403 // CHECK: define internal void [[HVT3]] 404 // Create stack storage and store argument in there. 405 // CHECK: [[A_ADDR:%.+]] = alloca i[[SZ]], align 406 // CHECK: [[AA_ADDR:%.+]] = alloca i[[SZ]], align 407 // CHECK: [[A_CASTED:%.+]] = alloca i[[SZ]], align 408 // CHECK: [[AA_CASTED:%.+]] = alloca i[[SZ]], align 409 // CHECK-DAG: store i[[SZ]] %{{.+}}, i[[SZ]]* [[A_ADDR]], align 410 // CHECK-DAG: store i[[SZ]] %{{.+}}, i[[SZ]]* [[AA_ADDR]], align 411 // CHECK-64-DAG:[[A_CADDR:%.+]] = bitcast i[[SZ]]* [[A_ADDR]] to i32* 412 // CHECK-DAG: [[AA_CADDR:%.+]] = bitcast i[[SZ]]* [[AA_ADDR]] to i16* 413 // CHECK-64-DAG:[[A:%.+]] = load i32, i32* [[A_CADDR]], align 414 // CHECK-32-DAG:[[A:%.+]] = load i32, i32* [[A_ADDR]], align 415 // CHECK-64-DAG:[[A_C:%.+]] = bitcast i[[SZ]]* [[A_CASTED]] to i32* 416 // CHECK-64-DAG:store i32 [[A]], i32* [[A_C]], align 417 // CHECK-32-DAG:store i32 [[A]], i32* [[A_CASTED]], align 418 // CHECK-DAG: [[AA:%.+]] = load i16, i16* [[AA_CADDR]], align 419 // CHECK-DAG: [[AA_C:%.+]] = bitcast i[[SZ]]* [[AA_CASTED]] to i16* 420 // CHECK-DAG: store i16 [[AA]], i16* [[AA_C]], align 421 // CHECK-DAG: [[PARAM1:%.+]] = load i[[SZ]], i[[SZ]]* [[A_CASTED]], align 422 // CHECK-DAG: [[PARAM2:%.+]] = load i[[SZ]], i[[SZ]]* [[AA_CASTED]], align 423 // CHECK-DAG: call {{.*}}void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* [[DEF_LOC]], i32 2, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i[[SZ]], i[[SZ]])* [[OMP_OUTLINED3:@.+]] to void (i32*, i32*, ...)*), i[[SZ]] [[PARAM1]], i[[SZ]] [[PARAM2]]) 424 // 425 // 426 // CHECK: define internal {{.*}}void [[OMP_OUTLINED3]](i32* noalias %.global_tid., i32* noalias %.bound_tid., i[[SZ]] %{{.+}}, i[[SZ]] %{{.+}}) 427 // CHECK: [[A_ADDR:%.+]] = alloca i[[SZ]], align 428 // CHECK: [[AA_ADDR:%.+]] = alloca i[[SZ]], align 429 // CHECK-DAG: store i[[SZ]] %{{.+}}, i[[SZ]]* [[A_ADDR]], align 430 // CHECK-DAG: store i[[SZ]] %{{.+}}, i[[SZ]]* [[AA_ADDR]], align 431 // CHECK-64-DAG:[[A_CADDR:%.+]] = bitcast i[[SZ]]* [[A_ADDR]] to i32* 432 // CHECK-DAG: [[AA_CADDR:%.+]] = bitcast i[[SZ]]* [[AA_ADDR]] to i16* 433 // CHECK: ret void 434 // CHECK-NEXT: } 435 436 // CHECK: define internal void [[HVT4]] 437 // Create local storage for each capture. 438 // CHECK: [[LOCAL_A:%.+]] = alloca i[[SZ]] 439 // CHECK: [[LOCAL_B:%.+]] = alloca [10 x float]* 440 // CHECK: [[LOCAL_VLA1:%.+]] = alloca i[[SZ]] 441 // CHECK: [[LOCAL_BN:%.+]] = alloca float* 442 // CHECK: [[LOCAL_C:%.+]] = alloca [5 x [10 x double]]* 443 // CHECK: [[LOCAL_VLA2:%.+]] = alloca i[[SZ]] 444 // CHECK: [[LOCAL_VLA3:%.+]] = alloca i[[SZ]] 445 // CHECK: [[LOCAL_CN:%.+]] = alloca double* 446 // CHECK: [[LOCAL_D:%.+]] = alloca [[TT]]* 447 // CHECK: [[LOCAL_A_CASTED:%.+]] = alloca i[[SZ]] 448 // CHECK-DAG: store i[[SZ]] [[ARG_A:%.+]], i[[SZ]]* [[LOCAL_A]] 449 // CHECK-DAG: store [10 x float]* [[ARG_B:%.+]], [10 x float]** [[LOCAL_B]] 450 // CHECK-DAG: store i[[SZ]] [[ARG_VLA1:%.+]], i[[SZ]]* [[LOCAL_VLA1]] 451 // CHECK-DAG: store float* [[ARG_BN:%.+]], float** [[LOCAL_BN]] 452 // CHECK-DAG: store [5 x [10 x double]]* [[ARG_C:%.+]], [5 x [10 x double]]** [[LOCAL_C]] 453 // CHECK-DAG: store i[[SZ]] [[ARG_VLA2:%.+]], i[[SZ]]* [[LOCAL_VLA2]] 454 // CHECK-DAG: store i[[SZ]] [[ARG_VLA3:%.+]], i[[SZ]]* [[LOCAL_VLA3]] 455 // CHECK-DAG: store double* [[ARG_CN:%.+]], double** [[LOCAL_CN]] 456 // CHECK-DAG: store [[TT]]* [[ARG_D:%.+]], [[TT]]** [[LOCAL_D]] 457 458 // CHECK-64-DAG:[[CONV_AP:%.+]] = bitcast i[[SZ]]* [[LOCAL_A]] to i32* 459 // CHECK-DAG: [[REF_B:%.+]] = load [10 x float]*, [10 x float]** [[LOCAL_B]], 460 // CHECK-DAG: [[VAL_VLA1:%.+]] = load i[[SZ]], i[[SZ]]* [[LOCAL_VLA1]], 461 // CHECK-DAG: [[REF_BN:%.+]] = load float*, float** [[LOCAL_BN]], 462 // CHECK-DAG: [[REF_C:%.+]] = load [5 x [10 x double]]*, [5 x [10 x double]]** [[LOCAL_C]], 463 // CHECK-DAG: [[VAL_VLA2:%.+]] = load i[[SZ]], i[[SZ]]* [[LOCAL_VLA2]], 464 // CHECK-DAG: [[VAL_VLA3:%.+]] = load i[[SZ]], i[[SZ]]* [[LOCAL_VLA3]], 465 // CHECK-DAG: [[REF_CN:%.+]] = load double*, double** [[LOCAL_CN]], 466 // CHECK-DAG: [[REF_D:%.+]] = load [[TT]]*, [[TT]]** [[LOCAL_D]], 467 468 // CHECK-64-DAG:[[CONV_A:%.+]] = load i32, i32* [[CONV_AP]] 469 // CHECK-64-DAG:[[CONV:%.+]] = bitcast i[[SZ]]* [[LOCAL_A_CASTED]] to i32* 470 // CHECK-64-DAG:store i32 [[CONV_A]], i32* [[CONV]], align 471 // CHECK-32-DAG:[[LOCAL_AV:%.+]] = load i32, i32* [[LOCAL_A]] 472 // CHECK-32-DAG:store i32 [[LOCAL_AV]], i32* [[LOCAL_A_CASTED]], align 473 // CHECK-DAG: [[REF_A:%.+]] = load i[[SZ]], i[[SZ]]* [[LOCAL_A_CASTED]], 474 475 // CHECK: call {{.*}}void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* [[DEF_LOC]], i32 9, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i[[SZ]], [10 x float]*, i[[SZ]], float*, [5 x [10 x double]]*, i[[SZ]], i[[SZ]], double*, [[TT]]*)* [[OMP_OUTLINED4:@.+]] to void (i32*, i32*, ...)*), i[[SZ]] [[REF_A]], [10 x float]* [[REF_B]], i[[SZ]] [[VAL_VLA1]], float* [[REF_BN]], [5 x [10 x double]]* [[REF_C]], i[[SZ]] [[VAL_VLA2]], i[[SZ]] [[VAL_VLA3]], double* [[REF_CN]], [[TT]]* [[REF_D]]) 476 // 477 // 478 // CHECK: define internal {{.*}}void [[OMP_OUTLINED4]](i32* noalias %.global_tid., i32* noalias %.bound_tid., i[[SZ]] %{{.+}}, [10 x float]* {{.+}}, i[[SZ]] %{{.+}}, float* {{.+}}, [5 x [10 x double]]* {{.+}}, i[[SZ]] %{{.+}}, i[[SZ]] %{{.+}}, double* {{.+}}, [[TT]]* {{.+}}) 479 // To reduce complexity, we're only going as far as validating the signature of the outlined parallel function. 480 481 template<typename tx> 482 tx ftemplate(int n) { 483 tx a = 0; 484 short aa = 0; 485 tx b[10]; 486 487 #pragma omp target teams distribute simd if(target: n>40) 488 for (int i = 0; i < 10; ++i) { 489 a += 1; 490 aa += 1; 491 b[2] += 1; 492 } 493 494 return a; 495 } 496 497 static 498 int fstatic(int n) { 499 int a = 0; 500 short aa = 0; 501 char aaa = 0; 502 int b[10]; 503 504 #pragma omp target teams distribute simd if(target: n>50) 505 for (int i = a; i < n; ++i) { 506 a += 1; 507 aa += 1; 508 aaa += 1; 509 b[2] += 1; 510 } 511 512 return a; 513 } 514 515 struct S1 { 516 double a; 517 518 int r1(int n){ 519 int b = n+1; 520 short int c[2][n]; 521 522 #pragma omp target teams distribute simd if(target: n>60) 523 for (int i = 0; i < 10; ++i) { 524 this->a = (double)b + 1.5; 525 c[1][1] = ++a; 526 } 527 528 return c[1][1] + (int)b; 529 } 530 }; 531 532 // CHECK: define {{.*}}@{{.*}}bar{{.*}} 533 int bar(int n){ 534 int a = 0; 535 536 // CHECK: call {{.*}}i32 [[FOO]](i32 {{.*}}) 537 a += foo(n); 538 539 S1 S; 540 // CHECK: call {{.*}}i32 [[FS1:@.+]]([[S1]]* {{.*}}, i32 {{.*}}) 541 a += S.r1(n); 542 543 // CHECK: call {{.*}}i32 [[FSTATIC:@.+]](i32 {{.*}}) 544 a += fstatic(n); 545 546 // CHECK: call {{.*}}i32 [[FTEMPLATE:@.+]](i32 {{.*}}) 547 a += ftemplate<int>(n); 548 549 return a; 550 } 551 552 // 553 // CHECK: define {{.*}}[[FS1]] 554 // 555 // CHECK: i8* @llvm.stacksave() 556 // CHECK-64: [[B_ADDR:%.+]] = bitcast i[[SZ]]* [[B_CADDR:%.+]] to i32* 557 // CHECK-64: store i32 %{{.+}}, i32* [[B_ADDR]], 558 // CHECK-64: [[B_CVAL:%.+]] = load i[[SZ]], i[[SZ]]* [[B_CADDR]], 559 560 // CHECK-32: store i32 %{{.+}}, i32* %__vla_expr 561 // CHECK-32: store i32 %{{.+}}, i32* [[B_ADDR:%.+]], 562 // CHECK-32: [[B_CVAL:%.+]] = load i[[SZ]], i[[SZ]]* [[B_ADDR]], 563 564 // CHECK: [[IF:%.+]] = icmp sgt i32 {{[^,]+}}, 60 565 // CHECK: br i1 [[IF]], label %[[TRY:[^,]+]], label %[[FAIL:[^,]+]] 566 // CHECK: [[TRY]] 567 // We capture 2 VLA sizes in this target region 568 // CHECK: [[CELEMSIZE2:%.+]] = mul nuw i[[SZ]] 2, [[VLA0:%.+]] 569 // CHECK-64: [[CSIZE:%.+]] = mul nuw i64 [[CELEMSIZE2]], 2 570 // CHECK-32: [[CSZSIZE:%.+]] = mul nuw i32 [[CELEMSIZE2]], 2 571 // CHECK-32: [[CSIZE:%.+]] = sext i32 [[CSZSIZE]] to i64 572 573 // CHECK-DAG: [[RET:%.+]] = call i32 @__tgt_target_teams(i64 -1, i8* @{{[^,]+}}, i32 6, i8** [[BPR:%[^,]+]], i8** [[PR:%[^,]+]], i64* [[SR:%[^,]+]], i64* getelementptr inbounds ([6 x i64], [6 x i64]* [[MAPT5]], i32 0, i32 0), i32 0, i32 1) 574 // CHECK-DAG: [[BPR]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[BP:%.+]], i32 0, i32 0 575 // CHECK-DAG: [[PR]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[P:%.+]], i32 0, i32 0 576 // CHECK-DAG: [[SR]] = getelementptr inbounds [6 x i64], [6 x i64]* [[S:%.+]], i32 0, i32 0 577 // CHECK-DAG: [[SADDR0:%.+]] = getelementptr inbounds [6 x i64], [6 x i64]* [[S]], i32 [[IDX0:[0-9]+]] 578 // CHECK-DAG: [[BPADDR0:%.+]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[BP]], i32 [[IDX0]] 579 // CHECK-DAG: [[PADDR0:%.+]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[P]], i32 [[IDX0]] 580 // CHECK-DAG: [[SADDR1:%.+]] = getelementptr inbounds [6 x i64], [6 x i64]* [[S]], i32 [[IDX1:[0-9]+]] 581 // CHECK-DAG: [[BPADDR1:%.+]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[BP]], i32 [[IDX1]] 582 // CHECK-DAG: [[PADDR1:%.+]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[P]], i32 [[IDX1]] 583 // CHECK-DAG: [[SADDR2:%.+]] = getelementptr inbounds [6 x i64], [6 x i64]* [[S]], i32 [[IDX2:[0-9]+]] 584 // CHECK-DAG: [[BPADDR2:%.+]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[BP]], i32 [[IDX2]] 585 // CHECK-DAG: [[PADDR2:%.+]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[P]], i32 [[IDX2]] 586 // CHECK-DAG: [[SADDR3:%.+]] = getelementptr inbounds [6 x i64], [6 x i64]* [[S]], i32 [[IDX3:[0-9]+]] 587 // CHECK-DAG: [[BPADDR3:%.+]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[BP]], i32 [[IDX3]] 588 // CHECK-DAG: [[PADDR3:%.+]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[P]], i32 [[IDX3]] 589 // CHECK-DAG: [[SADDR4:%.+]] = getelementptr inbounds [6 x i64], [6 x i64]* [[S]], i32 [[IDX3:[0-9]+]] 590 // CHECK-DAG: [[BPADDR4:%.+]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[BP]], i32 [[IDX3]] 591 // CHECK-DAG: [[PADDR4:%.+]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[P]], i32 [[IDX3]] 592 // CHECK-DAG: [[SADDR5:%.+]] = getelementptr inbounds [6 x i64], [6 x i64]* [[S]], i32 [[IDX3:[0-9]+]] 593 // CHECK-DAG: [[BPADDR5:%.+]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[BP]], i32 [[IDX3]] 594 // CHECK-DAG: [[PADDR5:%.+]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[P]], i32 [[IDX3]] 595 596 // The names below are not necessarily consistent with the names used for the 597 // addresses above as some are repeated. 598 // CHECK-DAG: store i[[SZ]] [[VLA0]], i[[SZ]]* [[CBPADDR0:%.+]], 599 // CHECK-DAG: store i[[SZ]] [[VLA0]], i[[SZ]]* [[CPADDR0:%.+]], 600 // CHECK-DAG: [[CBPADDR0]] = bitcast i8** {{%[^,]+}} to i[[SZ]]* 601 // CHECK-DAG: [[CPADDR0]] = bitcast i8** {{%[^,]+}} to i[[SZ]]* 602 // CHECK-DAG: store i64 {{4|8}}, i64* {{%[^,]+}} 603 604 // CHECK-DAG: store i[[SZ]] 2, i[[SZ]]* [[CBPADDR1:%.+]], 605 // CHECK-DAG: store i[[SZ]] 2, i[[SZ]]* [[CPADDR1:%.+]], 606 // CHECK-DAG: [[CBPADDR1]] = bitcast i8** {{%[^,]+}} to i[[SZ]]* 607 // CHECK-DAG: [[CPADDR1]] = bitcast i8** {{%[^,]+}} to i[[SZ]]* 608 // CHECK-DAG: store i64 {{4|8}}, i64* {{%[^,]+}} 609 610 // CHECK-DAG: store i[[SZ]] [[B_CVAL]], i[[SZ]]* [[CBPADDR2:%.+]], 611 // CHECK-DAG: store i[[SZ]] [[B_CVAL]], i[[SZ]]* [[CPADDR2:%.+]], 612 // CHECK-DAG: [[CBPADDR2]] = bitcast i8** {{%[^,]+}} to i[[SZ]]* 613 // CHECK-DAG: [[CPADDR2]] = bitcast i8** {{%[^,]+}} to i[[SZ]]* 614 // CHECK-DAG: store i64 4, i64* {{%[^,]+}} 615 616 // CHECK-DAG: store [[S1]]* %{{.+}}, [[S1]]** [[CBPADDR3:%.+]], 617 // CHECK-DAG: store double* %{{.+}}, double** [[CPADDR3:%.+]], 618 // CHECK-DAG: [[CBPADDR3]] = bitcast i8** {{%[^,]+}} to [[S1]]** 619 // CHECK-DAG: [[CPADDR3]] = bitcast i8** {{%[^,]+}} to double** 620 // CHECK-DAG: store i64 8, i64* {{%[^,]+}} 621 622 // CHECK-DAG: store [[S1]]* %{{.+}}, [[S1]]** [[CBPADDR4:%.+]], 623 // CHECK-DAG: store double* %{{.+}}, double** [[CPADDR4:%.+]], 624 // CHECK-DAG: [[CBPADDR4]] = bitcast i8** {{%[^,]+}} to [[S1]]** 625 // CHECK-DAG: [[CPADDR4]] = bitcast i8** {{%[^,]+}} to double** 626 // CHECK-DAG: store i64 8, i64* {{%[^,]+}} 627 628 // CHECK-DAG: store i16* %{{.+}}, i16** [[CBPADDR5:%.+]], 629 // CHECK-DAG: store i16* %{{.+}}, i16** [[CPADDR5:%.+]], 630 // CHECK-DAG: [[CBPADDR5]] = bitcast i8** {{%[^,]+}} to i16** 631 // CHECK-DAG: [[CPADDR5]] = bitcast i8** {{%[^,]+}} to i16** 632 // CHECK-DAG: store i64 [[CSIZE]], i64* {{%[^,]+}} 633 634 // CHECK-NEXT: [[ERROR:%.+]] = icmp ne i32 [[RET]], 0 635 // CHECK-NEXT: br i1 [[ERROR]], label %[[FAIL:[^,]+]], label %[[END:[^,]+]] 636 637 // CHECK: [[FAIL]] 638 // CHECK: call void [[HVT7:@.+]]({{[^,]+}}, {{[^,]+}}, {{[^,]+}}, {{[^,]+}}, {{[^,]+}}) 639 // CHECK-NEXT: br label %[[END]] 640 // CHECK: [[END]] 641 642 // 643 // CHECK: define {{.*}}[[FSTATIC]] 644 // 645 // CHECK: [[IF:%.+]] = icmp sgt i32 {{[^,]+}}, 50 646 // CHECK: br i1 [[IF]], label %[[IFTHEN:[^,]+]], label %[[IFELSE:[^,]+]] 647 // CHECK: [[IFTHEN]] 648 // CHECK-DAG: [[RET:%.+]] = call i32 @__tgt_target_teams(i64 -1, i8* @{{[^,]+}}, i32 5, i8** [[BPR:%[^,]+]], i8** [[PR:%[^,]+]], i64* getelementptr inbounds ([5 x i64], [5 x i64]* [[SIZET6]], i32 0, i32 0), i64* getelementptr inbounds ([5 x i64], [5 x i64]* [[MAPT6]], i32 0, i32 0), i32 0, i32 1) 649 // CHECK-DAG: [[BPR]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[BP:%.+]], i32 0, i32 0 650 // CHECK-DAG: [[PR]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[P:%.+]], i32 0, i32 0 651 652 // CHECK-DAG: [[BPADDR0:%.+]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[BP]], i32 0, i32 0 653 // CHECK-DAG: [[PADDR0:%.+]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[P]], i32 0, i32 0 654 // CHECK-DAG: store i[[SZ]] [[VAL0:%.+]], i[[SZ]]* [[CBPADDR0:%.+]], 655 // CHECK-DAG: store i[[SZ]] [[VAL0]], i[[SZ]]* [[CPADDR0:%.+]], 656 // CHECK-DAG: [[CBPADDR0]] = bitcast i8** {{%[^,]+}} to i[[SZ]]* 657 // CHECK-DAG: [[CPADDR0]] = bitcast i8** {{%[^,]+}} to i[[SZ]]* 658 659 // CHECK-DAG: [[BPADDR1:%.+]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[BP]], i32 0, i32 1 660 // CHECK-DAG: [[PADDR1:%.+]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[P]], i32 0, i32 1 661 // CHECK-DAG: store i[[SZ]] [[VAL1:%.+]], i[[SZ]]* [[CBPADDR1:%.+]], 662 // CHECK-DAG: store i[[SZ]] [[VAL1]], i[[SZ]]* [[CPADDR1:%.+]], 663 // CHECK-DAG: [[CBPADDR1]] = bitcast i8** {{%[^,]+}} to i[[SZ]]* 664 // CHECK-DAG: [[CPADDR1]] = bitcast i8** {{%[^,]+}} to i[[SZ]]* 665 666 // CHECK-DAG: [[BPADDR2:%.+]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[BP]], i32 0, i32 2 667 // CHECK-DAG: [[PADDR2:%.+]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[P]], i32 0, i32 2 668 // CHECK-DAG: store i[[SZ]] [[VAL2:%.+]], i[[SZ]]* [[CBPADDR2:%.+]], 669 // CHECK-DAG: store i[[SZ]] [[VAL2]], i[[SZ]]* [[CPADDR2:%.+]], 670 // CHECK-DAG: [[CBPADDR2]] = bitcast i8** {{%[^,]+}} to i[[SZ]]* 671 // CHECK-DAG: [[CPADDR2]] = bitcast i8** {{%[^,]+}} to i[[SZ]]* 672 673 // CHECK-DAG: [[BPADDR3:%.+]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[BP]], i32 0, i32 3 674 // CHECK-DAG: [[PADDR3:%.+]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[P]], i32 0, i32 3 675 // CHECK-DAG: store i[[SZ]] [[VAL3:%.+]], i[[SZ]]* [[CBPADDR3:%.+]], 676 // CHECK-DAG: store i[[SZ]] [[VAL3]], i[[SZ]]* [[CPADDR3:%.+]], 677 // CHECK-DAG: [[CBPADDR3]] = bitcast i8** {{%[^,]+}} to i[[SZ]]* 678 // CHECK-DAG: [[CPADDR3]] = bitcast i8** {{%[^,]+}} to i[[SZ]]* 679 680 // CHECK-DAG: [[BPADDR4:%.+]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[BP]], i32 0, i32 4 681 // CHECK-DAG: [[PADDR4:%.+]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[P]], i32 0, i32 4 682 // CHECK-DAG: store [10 x i32]* %{{.+}}, [10 x i32]** [[CBPADDR4:%.+]], 683 // CHECK-DAG: store [10 x i32]* %{{.+}}, [10 x i32]** [[CPADDR4:%.+]], 684 // CHECK-DAG: [[CBPADDR4]] = bitcast i8** {{%[^,]+}} to [10 x i32]** 685 // CHECK-DAG: [[CPADDR4]] = bitcast i8** {{%[^,]+}} to [10 x i32]** 686 687 // CHECK: [[ERROR:%.+]] = icmp ne i32 [[RET]], 0 688 // CHECK-NEXT: br i1 [[ERROR]], label %[[FAIL:.+]], label %[[END:[^,]+]] 689 // CHECK: [[FAIL]] 690 // CHECK: call void [[HVT6:@.+]]({{[^,]+}}, {{[^,]+}}, {{[^,]+}}, {{[^,]+}}, {{[^,]+}}) 691 // CHECK-NEXT: br label %[[END]] 692 // CHECK: [[END]] 693 // CHECK-NEXT: br label %[[IFEND:.+]] 694 // CHECK: [[IFELSE]] 695 // CHECK: call void [[HVT6]]({{[^,]+}}, {{[^,]+}}, {{[^,]+}}, {{[^,]+}}, {{[^,]+}}) 696 // CHECK-NEXT: br label %[[IFEND]] 697 // CHECK: [[IFEND]] 698 699 // 700 // CHECK: define {{.*}}[[FTEMPLATE]] 701 // 702 // CHECK: [[IF:%.+]] = icmp sgt i32 {{[^,]+}}, 40 703 // CHECK: br i1 [[IF]], label %[[IFTHEN:[^,]+]], label %[[IFELSE:[^,]+]] 704 // CHECK: [[IFTHEN]] 705 // CHECK-DAG: [[RET:%.+]] = call i32 @__tgt_target_teams(i64 -1, i8* @{{[^,]+}}, i32 3, i8** [[BPR:%[^,]+]], i8** [[PR:%[^,]+]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* [[SIZET7]], i32 0, i32 0), i64* getelementptr inbounds ([3 x i64], [3 x i64]* [[MAPT7]], i32 0, i32 0), i32 0, i32 1) 706 // CHECK-DAG: [[BPR]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[BP:%.+]], i32 0, i32 0 707 // CHECK-DAG: [[PR]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[P:%.+]], i32 0, i32 0 708 709 // CHECK-DAG: [[BPADDR0:%.+]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[BP]], i32 0, i32 0 710 // CHECK-DAG: [[PADDR0:%.+]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[P]], i32 0, i32 0 711 // CHECK-DAG: store i[[SZ]] [[VAL0:%.+]], i[[SZ]]* [[CBPADDR0:%.+]], 712 // CHECK-DAG: store i[[SZ]] [[VAL0]], i[[SZ]]* [[CPADDR0:%.+]], 713 // CHECK-DAG: [[CBPADDR0]] = bitcast i8** {{%[^,]+}} to i[[SZ]]* 714 // CHECK-DAG: [[CPADDR0]] = bitcast i8** {{%[^,]+}} to i[[SZ]]* 715 716 // CHECK-DAG: [[BPADDR1:%.+]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[BP]], i32 0, i32 1 717 // CHECK-DAG: [[PADDR1:%.+]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[P]], i32 0, i32 1 718 // CHECK-DAG: store i[[SZ]] [[VAL1:%.+]], i[[SZ]]* [[CBPADDR1:%.+]], 719 // CHECK-DAG: store i[[SZ]] [[VAL1]], i[[SZ]]* [[CPADDR1:%.+]], 720 // CHECK-DAG: [[CBPADDR1]] = bitcast i8** {{%[^,]+}} to i[[SZ]]* 721 // CHECK-DAG: [[CPADDR1]] = bitcast i8** {{%[^,]+}} to i[[SZ]]* 722 723 // CHECK-DAG: [[BPADDR2:%.+]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[BP]], i32 0, i32 2 724 // CHECK-DAG: [[PADDR2:%.+]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[P]], i32 0, i32 2 725 // CHECK-DAG: store i[[SZ]] [[VAL2:%.+]], i[[SZ]]* [[CBPADDR2:%.+]], 726 // CHECK-DAG: store i[[SZ]] [[VAL2]], i[[SZ]]* [[CPADDR2:%.+]], 727 // CHECK-DAG: [[CBPADDR2]] = bitcast i8** {{%[^,]+}} to i[[SZ]]* 728 // CHECK-DAG: [[CPADDR2]] = bitcast i8** {{%[^,]+}} to i[[SZ]]* 729 730 // CHECK: [[ERROR:%.+]] = icmp ne i32 [[RET]], 0 731 // CHECK-NEXT: br i1 [[ERROR]], label %[[FAIL:.+]], label %[[END:[^,]+]] 732 // CHECK: [[FAIL]] 733 // CHECK: call void [[HVT5:@.+]]({{[^,]+}}, {{[^,]+}}, {{[^,]+}}) 734 // CHECK-NEXT: br label %[[END]] 735 // CHECK: [[END]] 736 // CHECK-NEXT: br label %[[IFEND:.+]] 737 // CHECK: [[IFELSE]] 738 // CHECK: call void [[HVT5]]({{[^,]+}}, {{[^,]+}}, {{[^,]+}}) 739 // CHECK-NEXT: br label %[[IFEND]] 740 // CHECK: [[IFEND]] 741 742 743 744 // Check that the offloading functions are emitted and that the arguments are 745 // correct and loaded correctly for the target regions of the callees of bar(). 746 747 // CHECK: define internal void [[HVT7]] 748 // Create local storage for each capture. 749 // CHECK: [[LOCAL_THIS:%.+]] = alloca [[S1]]* 750 // CHECK: [[LOCAL_B:%.+]] = alloca i[[SZ]] 751 // CHECK: [[LOCAL_VLA1:%.+]] = alloca i[[SZ]] 752 // CHECK: [[LOCAL_VLA2:%.+]] = alloca i[[SZ]] 753 // CHECK: [[LOCAL_C:%.+]] = alloca i16* 754 // CHECK: [[LOCAL_B_CASTED:%.+]] = alloca i[[SZ]] 755 // CHECK-DAG: store [[S1]]* [[ARG_THIS:%.+]], [[S1]]** [[LOCAL_THIS]] 756 // CHECK-DAG: store i[[SZ]] [[ARG_B:%.+]], i[[SZ]]* [[LOCAL_B]] 757 // CHECK-DAG: store i[[SZ]] [[ARG_VLA1:%.+]], i[[SZ]]* [[LOCAL_VLA1]] 758 // CHECK-DAG: store i[[SZ]] [[ARG_VLA2:%.+]], i[[SZ]]* [[LOCAL_VLA2]] 759 // CHECK-DAG: store i16* [[ARG_C:%.+]], i16** [[LOCAL_C]] 760 // Store captures in the context. 761 // CHECK-DAG: [[REF_THIS:%.+]] = load [[S1]]*, [[S1]]** [[LOCAL_THIS]], 762 // CHECK-64-DAG:[[CONV_BP:%.+]] = bitcast i[[SZ]]* [[LOCAL_B]] to i32* 763 // CHECK-DAG: [[VAL_VLA1:%.+]] = load i[[SZ]], i[[SZ]]* [[LOCAL_VLA1]], 764 // CHECK-DAG: [[VAL_VLA2:%.+]] = load i[[SZ]], i[[SZ]]* [[LOCAL_VLA2]], 765 // CHECK-DAG: [[REF_C:%.+]] = load i16*, i16** [[LOCAL_C]], 766 767 // CHECK-64-DAG:[[CONV_B:%.+]] = load i32, i32* [[CONV_BP]] 768 // CHECK-64-DAG:[[CONV:%.+]] = bitcast i[[SZ]]* [[LOCAL_B_CASTED]] to i32* 769 // CHECK-64-DAG:store i32 [[CONV_B]], i32* [[CONV]], align 770 // CHECK-32-DAG:[[LOCAL_BV:%.+]] = load i32, i32* [[LOCAL_B]] 771 // CHECK-32-DAG:store i32 [[LOCAL_BV]], i32* [[LOCAL_B_CASTED]], align 772 // CHECK-DAG: [[REF_B:%.+]] = load i[[SZ]], i[[SZ]]* [[LOCAL_B_CASTED]], 773 774 // CHECK: call {{.*}}void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* [[DEF_LOC]], i32 5, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, [[S1]]*, i[[SZ]], i[[SZ]], i[[SZ]], i16*)* [[OMP_OUTLINED5:@.+]] to void (i32*, i32*, ...)*), [[S1]]* [[REF_THIS]], i[[SZ]] [[REF_B]], i[[SZ]] [[VAL_VLA1]], i[[SZ]] [[VAL_VLA2]], i16* [[REF_C]]) 775 // 776 // 777 // CHECK: define internal {{.*}}void [[OMP_OUTLINED5]](i32* noalias %.global_tid., i32* noalias %.bound_tid., [[S1]]* %{{.+}}, i[[SZ]] %{{.+}}, i[[SZ]] %{{.+}}, i[[SZ]] %{{.+}}, i16* {{.+}}) 778 // To reduce complexity, we're only going as far as validating the signature of the outlined parallel function. 779 780 781 // CHECK: define internal void [[HVT6]] 782 // Create local storage for each capture. 783 // CHECK: [[LOCAL_A:%.+]] = alloca i[[SZ]] 784 // CHECK: alloca i[[SZ]], 785 // CHECK: [[LOCAL_AA:%.+]] = alloca i[[SZ]] 786 // CHECK: [[LOCAL_AAA:%.+]] = alloca i[[SZ]] 787 // CHECK: [[LOCAL_B:%.+]] = alloca [10 x i32]* 788 // CHECK: [[LOCAL_A_CASTED:%.+]] = alloca i[[SZ]] 789 // CHECK: alloca i[[SZ]], 790 // CHECK: [[LOCAL_AA_CASTED:%.+]] = alloca i[[SZ]] 791 // CHECK: [[LOCAL_AAA_CASTED:%.+]] = alloca i[[SZ]] 792 // CHECK-DAG: store i[[SZ]] [[ARG_A:%.+]], i[[SZ]]* [[LOCAL_A]] 793 // CHECK-DAG: store i[[SZ]] [[ARG_AA:%.+]], i[[SZ]]* [[LOCAL_AA]] 794 // CHECK-DAG: store i[[SZ]] [[ARG_AAA:%.+]], i[[SZ]]* [[LOCAL_AAA]] 795 // CHECK-DAG: store [10 x i32]* [[ARG_B:%.+]], [10 x i32]** [[LOCAL_B]] 796 // Store captures in the context. 797 // CHECK-64-DAG:[[CONV_AP:%.+]] = bitcast i[[SZ]]* [[LOCAL_A]] to i32* 798 // CHECK-DAG: [[CONV_AAP:%.+]] = bitcast i[[SZ]]* [[LOCAL_AA]] to i16* 799 // CHECK-DAG: [[CONV_AAAP:%.+]] = bitcast i[[SZ]]* [[LOCAL_AAA]] to i8* 800 // CHECK-DAG: [[REF_B:%.+]] = load [10 x i32]*, [10 x i32]** [[LOCAL_B]], 801 802 // CHECK-64-DAG:[[CONV_A:%.+]] = load i32, i32* [[CONV_AP]] 803 // CHECK-64-DAG:[[CONV:%.+]] = bitcast i[[SZ]]* [[LOCAL_A_CASTED]] to i32* 804 // CHECK-64-DAG:store i32 [[CONV_A]], i32* [[CONV]], align 805 // CHECK-32-DAG:[[LOCAL_AV:%.+]] = load i32, i32* [[LOCAL_A]] 806 // CHECK-32-DAG:store i32 [[LOCAL_AV]], i32* [[LOCAL_A_CASTED]], align 807 // CHECK-DAG: [[REF_A:%.+]] = load i[[SZ]], i[[SZ]]* [[LOCAL_A_CASTED]], 808 809 // CHECK-DAG: [[CONV_AA:%.+]] = load i16, i16* [[CONV_AAP]] 810 // CHECK-DAG: [[CONV:%.+]] = bitcast i[[SZ]]* [[LOCAL_AA_CASTED]] to i16* 811 // CHECK-DAG: store i16 [[CONV_AA]], i16* [[CONV]], align 812 // CHECK-DAG: [[REF_AA:%.+]] = load i[[SZ]], i[[SZ]]* [[LOCAL_AA_CASTED]], 813 814 // CHECK-DAG: [[CONV_AAA:%.+]] = load i8, i8* [[CONV_AAAP]] 815 // CHECK-DAG: [[CONV:%.+]] = bitcast i[[SZ]]* [[LOCAL_AAA_CASTED]] to i8* 816 // CHECK-DAG: store i8 [[CONV_AAA]], i8* [[CONV]], align 817 // CHECK-DAG: [[REF_AAA:%.+]] = load i[[SZ]], i[[SZ]]* [[LOCAL_AAA_CASTED]], 818 819 // CHECK: call {{.*}}void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* [[DEF_LOC]], i32 5, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i[[SZ]], i[[SZ]], i[[SZ]], i[[SZ]], [10 x i32]*)* [[OMP_OUTLINED6:@.+]] to void (i32*, i32*, ...)*), i[[SZ]] [[REF_A]], i[[SZ]] {{.+}}, i[[SZ]] [[REF_AA]], i[[SZ]] [[REF_AAA]], [10 x i32]* [[REF_B]]) 820 // 821 // 822 // CHECK: define internal {{.*}}void [[OMP_OUTLINED6]](i32* noalias %.global_tid., i32* noalias %.bound_tid., i[[SZ]] %{{.+}}, i[[SZ]] %{{.+}}, i[[SZ]] %{{.+}}, [10 x i32]* {{.+}}) 823 // To reduce complexity, we're only going as far as validating the signature of the outlined parallel function. 824 825 // CHECK: define internal void [[HVT5]] 826 // Create local storage for each capture. 827 // CHECK: [[LOCAL_A:%.+]] = alloca i[[SZ]] 828 // CHECK: [[LOCAL_AA:%.+]] = alloca i[[SZ]] 829 // CHECK: [[LOCAL_B:%.+]] = alloca [10 x i32]* 830 // CHECK: [[LOCAL_A_CASTED:%.+]] = alloca i[[SZ]] 831 // CHECK: [[LOCAL_AA_CASTED:%.+]] = alloca i[[SZ]] 832 // CHECK-DAG: store i[[SZ]] [[ARG_A:%.+]], i[[SZ]]* [[LOCAL_A]] 833 // CHECK-DAG: store i[[SZ]] [[ARG_AA:%.+]], i[[SZ]]* [[LOCAL_AA]] 834 // CHECK-DAG: store [10 x i32]* [[ARG_B:%.+]], [10 x i32]** [[LOCAL_B]] 835 // Store captures in the context. 836 // CHECK-64-DAG:[[CONV_AP:%.+]] = bitcast i[[SZ]]* [[LOCAL_A]] to i32* 837 // CHECK-DAG: [[CONV_AAP:%.+]] = bitcast i[[SZ]]* [[LOCAL_AA]] to i16* 838 // CHECK-DAG: [[REF_B:%.+]] = load [10 x i32]*, [10 x i32]** [[LOCAL_B]], 839 840 // CHECK-64-DAG:[[CONV_A:%.+]] = load i32, i32* [[CONV_AP]] 841 // CHECK-64-DAG:[[CONV:%.+]] = bitcast i[[SZ]]* [[LOCAL_A_CASTED]] to i32* 842 // CHECK-64-DAG:store i32 [[CONV_A]], i32* [[CONV]], align 843 // CHECK-32-DAG:[[LOCAL_AV:%.+]] = load i32, i32* [[LOCAL_A]] 844 // CHECK-32-DAG:store i32 [[LOCAL_AV]], i32* [[LOCAL_A_CASTED]], align 845 // CHECK-DAG: [[REF_A:%.+]] = load i[[SZ]], i[[SZ]]* [[LOCAL_A_CASTED]], 846 847 // CHECK-DAG: [[CONV_AA:%.+]] = load i16, i16* [[CONV_AAP]] 848 // CHECK-DAG: [[CONV:%.+]] = bitcast i[[SZ]]* [[LOCAL_AA_CASTED]] to i16* 849 // CHECK-DAG: store i16 [[CONV_AA]], i16* [[CONV]], align 850 // CHECK-DAG: [[REF_AA:%.+]] = load i[[SZ]], i[[SZ]]* [[LOCAL_AA_CASTED]], 851 852 // CHECK: call {{.*}}void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* [[DEF_LOC]], i32 3, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i[[SZ]], i[[SZ]], [10 x i32]*)* [[OMP_OUTLINED7:@.+]] to void (i32*, i32*, ...)*), i[[SZ]] [[REF_A]], i[[SZ]] [[REF_AA]], [10 x i32]* [[REF_B]]) 853 // 854 // 855 // CHECK: define internal {{.*}}void [[OMP_OUTLINED7]](i32* noalias %.global_tid., i32* noalias %.bound_tid., i[[SZ]] %{{.+}}, i[[SZ]] %{{.+}}, [10 x i32]* {{.+}}) 856 // To reduce complexity, we're only going as far as validating the signature of the outlined parallel function. 857 858 // CHECK-DAG: !{!"llvm.loop.vectorize.width", i32 16} 859 // CHECK-DAG: !{!"llvm.loop.vectorize.enable", i1 true} 860 // CHECK-DAG: !{!"llvm.loop.vectorize.width", i32 32} 861 862 #endif 863