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 %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 %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 %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 %s --check-prefix CHECK --check-prefix CHECK-32 8 9 // Test target codegen - host bc file has to be created first. 10 // 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 11 // 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 %s --check-prefix TCHECK --check-prefix TCHECK-64 12 // 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 13 // 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 %s --check-prefix TCHECK --check-prefix TCHECK-64 14 // 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 15 // 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 %s --check-prefix TCHECK --check-prefix TCHECK-32 16 // 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 17 // 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 %s --check-prefix TCHECK --check-prefix TCHECK-32 18 19 // expected-no-diagnostics 20 #ifndef HEADER 21 #define HEADER 22 // CHECK-DAG: %ident_t = type { i32, i32, i32, i32, i8* } 23 // CHECK-DAG: [[STR:@.+]] = private unnamed_addr constant [23 x i8] c";unknown;unknown;0;0;;\00" 24 // CHECK-DAG: [[DEF_LOC:@.+]] = private unnamed_addr constant %ident_t { i32 0, i32 2, i32 0, i32 0, i8* getelementptr inbounds ([23 x i8], [23 x i8]* [[STR]], i32 0, i32 0) } 25 26 // CHECK-DAG: [[TT:%.+]] = type { i64, i8 } 27 // CHECK-DAG: [[S1:%.+]] = type { double } 28 // CHECK-DAG: [[ENTTY:%.+]] = type { i8*, i8*, i[[SZ:32|64]], i32, i32 } 29 // CHECK-DAG: [[DEVTY:%.+]] = type { i8*, i8*, [[ENTTY]]*, [[ENTTY]]* } 30 // CHECK-DAG: [[DSCTY:%.+]] = type { i32, [[DEVTY]]*, [[ENTTY]]*, [[ENTTY]]* } 31 32 // TCHECK: [[ENTTY:%.+]] = type { i8*, i8*, i{{32|64}}, i32, i32 } 33 34 // CHECK-DAG: $[[REGFN:\.omp_offloading\..+]] = comdat 35 36 // We have 8 target regions, but only 7 that actually will generate offloading 37 // code, only 6 will have mapped arguments, and only 4 have all-constant map 38 // sizes. 39 40 // CHECK-DAG: [[SIZET2:@.+]] = private unnamed_addr constant [3 x i[[SZ]]] [i[[SZ]] 2, i[[SZ]] 4, i[[SZ]] 4] 41 // CHECK-DAG: [[MAPT2:@.+]] = private unnamed_addr constant [3 x i32] [i32 288, i32 288, i32 288] 42 // CHECK-DAG: [[SIZET3:@.+]] = private unnamed_addr constant [2 x i[[SZ]]] [i[[SZ]] 4, i[[SZ]] 2] 43 // CHECK-DAG: [[MAPT3:@.+]] = private unnamed_addr constant [2 x i32] [i32 288, i32 288] 44 // CHECK-DAG: [[MAPT4:@.+]] = private unnamed_addr constant [10 x i32] [i32 288, i32 547, i32 288, i32 547, i32 547, i32 288, i32 288, i32 547, i32 547, i32 288] 45 // CHECK-DAG: [[SIZET5:@.+]] = private unnamed_addr constant [3 x i[[SZ]]] [i[[SZ]] 4, i[[SZ]] 2, i[[SZ]] 40] 46 // CHECK-DAG: [[MAPT5:@.+]] = private unnamed_addr constant [3 x i32] [i32 288, i32 288, i32 547] 47 // CHECK-DAG: [[SIZET6:@.+]] = private unnamed_addr constant [4 x i[[SZ]]] [i[[SZ]] 4, i[[SZ]] 2, i[[SZ]] 1, i[[SZ]] 40] 48 // CHECK-DAG: [[MAPT6:@.+]] = private unnamed_addr constant [4 x i32] [i32 288, i32 288, i32 288, i32 547] 49 // CHECK-DAG: [[MAPT7:@.+]] = private unnamed_addr constant [5 x i32] [i32 547, i32 288, i32 288, i32 288, i32 547] 50 // CHECK-DAG: @{{.*}} = private constant i8 0 51 // CHECK-DAG: @{{.*}} = private constant i8 0 52 // CHECK-DAG: @{{.*}} = private constant i8 0 53 // CHECK-DAG: @{{.*}} = private constant i8 0 54 // CHECK-DAG: @{{.*}} = private constant i8 0 55 // CHECK-DAG: @{{.*}} = private constant i8 0 56 // CHECK-DAG: @{{.*}} = private constant i8 0 57 58 // TCHECK: @{{.+}} = constant [[ENTTY]] 59 // TCHECK: @{{.+}} = constant [[ENTTY]] 60 // TCHECK: @{{.+}} = constant [[ENTTY]] 61 // TCHECK: @{{.+}} = constant [[ENTTY]] 62 // TCHECK: @{{.+}} = constant [[ENTTY]] 63 // TCHECK: @{{.+}} = constant [[ENTTY]] 64 // TCHECK: @{{.+}} = constant [[ENTTY]] 65 // TCHECK-NOT: @{{.+}} = constant [[ENTTY]] 66 67 // Check if offloading descriptor is created. 68 // CHECK: [[ENTBEGIN:@.+]] = external constant [[ENTTY]] 69 // CHECK: [[ENTEND:@.+]] = external constant [[ENTTY]] 70 // CHECK: [[DEVBEGIN:@.+]] = external constant i8 71 // CHECK: [[DEVEND:@.+]] = external constant i8 72 // CHECK: [[IMAGES:@.+]] = internal unnamed_addr constant [1 x [[DEVTY]]] [{{.+}} { i8* [[DEVBEGIN]], i8* [[DEVEND]], [[ENTTY]]* [[ENTBEGIN]], [[ENTTY]]* [[ENTEND]] }], comdat($[[REGFN]]) 73 // 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]]) 74 75 // Check target registration is registered as a Ctor. 76 // CHECK: appending global [1 x { i32, void ()*, i8* }] [{ i32, void ()*, i8* } { i32 0, void ()* bitcast (void (i8*)* @[[REGFN]] to void ()*), i8* bitcast (void (i8*)* @[[REGFN]] to i8*) }] 77 78 79 template<typename tx, typename ty> 80 struct TT{ 81 tx X; 82 ty Y; 83 }; 84 85 // CHECK-LABEL: get_val 86 long long get_val() { return 0; } 87 88 // CHECK: define {{.*}}[[FOO:@.+]]( 89 int foo(int n) { 90 int a = 0; 91 short aa = 0; 92 float b[10]; 93 float bn[n]; 94 double c[5][10]; 95 double cn[5][n]; 96 TT<long long, char> d; 97 98 // CHECK: [[RET:%.+]] = call i32 @__tgt_target_teams(i32 -1, i8* @{{[^,]+}}, i32 0, i8** null, i8** null, i[[SZ]]* null, i32* null, i32 1, i32 0) 99 // CHECK-NEXT: [[ERROR:%.+]] = icmp ne i32 [[RET]], 0 100 // CHECK-NEXT: br i1 [[ERROR]], label %[[FAIL:[^,]+]], label %[[END:[^,]+]] 101 // CHECK: [[FAIL]] 102 // CHECK: call void [[HVT0:@.+]]() 103 // CHECK-NEXT: br label %[[END]] 104 // CHECK: [[END]] 105 #pragma omp target parallel for simd 106 for (int i = 3; i < 32; i += 5) { 107 } 108 109 // CHECK: call void [[HVT1:@.+]](i[[SZ]] {{[^,]+}}, i{{32|64}}{{[*]*}} {{[^)]+}}) 110 long long k = get_val(); 111 #pragma omp target parallel for simd if(target: 0) linear(k : 3) schedule(dynamic) 112 for (int i = 10; i > 1; i--) { 113 a += 1; 114 } 115 116 // CHECK-DAG: [[RET:%.+]] = call i32 @__tgt_target_teams(i32 -1, i8* @{{[^,]+}}, i32 3, i8** [[BP:%[^,]+]], i8** [[P:%[^,]+]], i[[SZ]]* getelementptr inbounds ([3 x i[[SZ]]], [3 x i[[SZ]]]* [[SIZET2]], i32 0, i32 0), i32* getelementptr inbounds ([3 x i32], [3 x i32]* [[MAPT2]], i32 0, i32 0), i32 1, i32 0) 117 // CHECK-DAG: [[BP]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[BPR:%[^,]+]], i32 0, i32 0 118 // CHECK-DAG: [[P]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[PR:%[^,]+]], i32 0, i32 0 119 // CHECK-DAG: [[BPADDR0:%.+]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[BPR]], i32 0, i32 0 120 // CHECK-DAG: [[PADDR0:%.+]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[PR]], i32 0, i32 0 121 // CHECK-DAG: [[CBPADDR0:%.+]] = bitcast i8** [[BPADDR0]] to i[[SZ]]* 122 // CHECK-DAG: [[CPADDR0:%.+]] = bitcast i8** [[PADDR0]] to i[[SZ]]* 123 // CHECK-DAG: store i[[SZ]] [[VAL0:%.+]], i[[SZ]]* [[CBPADDR0]], 124 // CHECK-DAG: store i[[SZ]] [[VAL0]], i[[SZ]]* [[CPADDR0]], 125 // CHECK-DAG: [[BPADDR1:%.+]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[BPR]], i32 0, i32 1 126 // CHECK-DAG: [[PADDR1:%.+]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[PR]], i32 0, i32 1 127 // CHECK-DAG: [[CBPADDR1:%.+]] = bitcast i8** [[BPADDR1]] to i[[SZ]]* 128 // CHECK-DAG: [[CPADDR1:%.+]] = bitcast i8** [[PADDR1]] to i[[SZ]]* 129 // CHECK-DAG: store i[[SZ]] [[VAL1:%.+]], i[[SZ]]* [[CBPADDR1]], 130 // CHECK-DAG: store i[[SZ]] [[VAL1]], i[[SZ]]* [[CPADDR1]], 131 // CHECK-DAG: [[BPADDR2:%.+]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[BPR]], i32 0, i32 1 132 // CHECK-DAG: [[PADDR2:%.+]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[PR]], i32 0, i32 1 133 // CHECK-DAG: [[CBPADDR2:%.+]] = bitcast i8** [[BPADDR2]] to i[[SZ]]* 134 // CHECK-DAG: [[CPADDR2:%.+]] = bitcast i8** [[PADDR2]] to i[[SZ]]* 135 // CHECK-DAG: store i[[SZ]] [[VAL2:%.+]], i[[SZ]]* [[CBPADDR2]], 136 // CHECK-DAG: store i[[SZ]] [[VAL2]], i[[SZ]]* [[CPADDR2]], 137 138 // CHECK-NEXT: [[ERROR:%.+]] = icmp ne i32 [[RET]], 0 139 // CHECK-NEXT: br i1 [[ERROR]], label %[[FAIL:[^,]+]], label %[[END:[^,]+]] 140 // CHECK: [[FAIL]] 141 // CHECK: call void [[HVT2:@.+]](i[[SZ]] {{[^,]+}}, i[[SZ]] {{[^)]+}}) 142 // CHECK-NEXT: br label %[[END]] 143 // CHECK: [[END]] 144 int lin = 12; 145 #pragma omp target parallel for simd if(target: 1) linear(lin, a : get_val()) 146 for (unsigned long long it = 2000; it >= 600; it-=400) { 147 aa += 1; 148 } 149 150 // CHECK: [[IF:%.+]] = icmp sgt i32 {{[^,]+}}, 10 151 // CHECK: br i1 [[IF]], label %[[IFTHEN:[^,]+]], label %[[IFELSE:[^,]+]] 152 // CHECK: [[IFTHEN]] 153 // CHECK-DAG: [[RET:%.+]] = call i32 @__tgt_target_teams(i32 -1, i8* @{{[^,]+}}, i32 2, i8** [[BPR:%[^,]+]], i8** [[PR:%[^,]+]], i[[SZ]]* getelementptr inbounds ([2 x i[[SZ]]], [2 x i[[SZ]]]* [[SIZET3]], i32 0, i32 0), i32* getelementptr inbounds ([2 x i32], [2 x i32]* [[MAPT3]], i32 0, i32 0), i32 1, i32 0) 154 // CHECK-DAG: [[BPR]] = getelementptr inbounds [2 x i8*], [2 x i8*]* [[BP:%[^,]+]], i32 0, i32 0 155 // CHECK-DAG: [[PR]] = getelementptr inbounds [2 x i8*], [2 x i8*]* [[P:%[^,]+]], i32 0, i32 0 156 157 // CHECK-DAG: [[BPADDR0:%.+]] = getelementptr inbounds [2 x i8*], [2 x i8*]* [[BP]], i32 0, i32 0 158 // CHECK-DAG: [[PADDR0:%.+]] = getelementptr inbounds [2 x i8*], [2 x i8*]* [[P]], i32 0, i32 0 159 // CHECK-DAG: [[CBPADDR0:%.+]] = bitcast i8** [[BPADDR0]] to i[[SZ]]* 160 // CHECK-DAG: [[CPADDR0:%.+]] = bitcast i8** [[PADDR0]] to i[[SZ]]* 161 // CHECK-DAG: store i[[SZ]] [[VAL0:%.+]], i[[SZ]]* [[CBPADDR0]], 162 // CHECK-DAG: store i[[SZ]] [[VAL0]], i[[SZ]]* [[CPADDR0]], 163 164 // CHECK-DAG: [[BPADDR1:%.+]] = getelementptr inbounds [2 x i8*], [2 x i8*]* [[BP]], i32 0, i32 1 165 // CHECK-DAG: [[PADDR1:%.+]] = getelementptr inbounds [2 x i8*], [2 x i8*]* [[P]], i32 0, i32 1 166 // CHECK-DAG: [[CBPADDR1:%.+]] = bitcast i8** [[BPADDR1]] to i[[SZ]]* 167 // CHECK-DAG: [[CPADDR1:%.+]] = bitcast i8** [[PADDR1]] to i[[SZ]]* 168 // CHECK-DAG: store i[[SZ]] [[VAL1:%.+]], i[[SZ]]* [[CBPADDR1]], 169 // CHECK-DAG: store i[[SZ]] [[VAL1]], i[[SZ]]* [[CPADDR1]], 170 // CHECK: [[ERROR:%.+]] = icmp ne i32 [[RET]], 0 171 // CHECK-NEXT: br i1 [[ERROR]], label %[[FAIL:.+]], label %[[END:[^,]+]] 172 // CHECK: [[FAIL]] 173 // CHECK: call void [[HVT3:@.+]]({{[^,]+}}, {{[^,]+}}) 174 // CHECK-NEXT: br label %[[END]] 175 // CHECK: [[END]] 176 // CHECK-NEXT: br label %[[IFEND:.+]] 177 // CHECK: [[IFELSE]] 178 // CHECK: call void [[HVT3]]({{[^,]+}}, {{[^,]+}}) 179 // CHECK-NEXT: br label %[[IFEND]] 180 // CHECK: [[IFEND]] 181 182 #pragma omp target parallel for simd if(target: n>10) 183 for (short it = 6; it <= 20; it-=-4) { 184 a += 1; 185 aa += 1; 186 } 187 188 // We capture 3 VLA sizes in this target region 189 // CHECK: [[A_VAL:%.+]] = load i32, i32* %{{.+}}, 190 // CHECK: store i32 [[A_VAL]], i32* [[A_CADDR:%.+]], 191 // CHECK-64: [[A_VAL:%.+]] = load i32, i32* %{{.+}}, 192 // CHECK-64: [[A_ADDR:%.+]] = bitcast i[[SZ]]* [[A_CADDR:%.+]] to i32* 193 // CHECK-64: store i32 [[A_VAL]], i32* [[A_ADDR]], 194 // CHECK-64: [[A_CVAL:%.+]] = load i[[SZ]], i[[SZ]]* [[A_CADDR]], 195 196 // CHECK-32: [[A_VAL:%.+]] = load i32, i32* %{{.+}}, 197 // CHECK-32: store i32 [[A_VAL]], i32* [[A_CADDR:%.+]], 198 // CHECK-32: [[A_CVAL:%.+]] = load i[[SZ]], i[[SZ]]* [[A_CADDR]], 199 200 // CHECK: [[BNSIZE:%.+]] = mul nuw i[[SZ]] [[VLA0:%.+]], 4 201 // CHECK: [[CNELEMSIZE2:%.+]] = mul nuw i[[SZ]] 5, [[VLA1:%.+]] 202 // CHECK: [[CNSIZE:%.+]] = mul nuw i[[SZ]] [[CNELEMSIZE2]], 8 203 204 // CHECK: [[IF:%.+]] = icmp sgt i32 {{[^,]+}}, 20 205 // CHECK: br i1 [[IF]], label %[[TRY:[^,]+]], label %[[FAIL:[^,]+]] 206 // CHECK: [[TRY]] 207 // CHECK-DAG: [[RET:%.+]] = call i32 @__tgt_target_teams(i32 -1, i8* @{{[^,]+}}, i32 10, i8** [[BPR:%[^,]+]], i8** [[PR:%[^,]+]], i[[SZ]]* [[SR:%[^,]+]], i32* getelementptr inbounds ([10 x i32], [10 x i32]* [[MAPT4]], i32 0, i32 0), i32 1, i32 0) 208 // CHECK-DAG: [[BPR]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[BP:%[^,]+]], i32 0, i32 0 209 // CHECK-DAG: [[PR]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[P:%[^,]+]], i32 0, i32 0 210 // CHECK-DAG: [[SR]] = getelementptr inbounds [10 x i[[SZ]]], [10 x i[[SZ]]]* [[S:%[^,]+]], i32 0, i32 0 211 212 // CHECK-DAG: [[SADDR0:%.+]] = getelementptr inbounds [10 x i[[SZ]]], [10 x i[[SZ]]]* [[S]], i32 0, i32 [[IDX0:[0-9]+]] 213 // CHECK-DAG: [[BPADDR0:%.+]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[BP]], i32 0, i32 [[IDX0]] 214 // CHECK-DAG: [[PADDR0:%.+]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[P]], i32 0, i32 [[IDX0]] 215 // CHECK-DAG: [[SADDR1:%.+]] = getelementptr inbounds [10 x i[[SZ]]], [10 x i[[SZ]]]* [[S]], i32 0, i32 [[IDX1:[0-9]+]] 216 // CHECK-DAG: [[BPADDR1:%.+]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[BP]], i32 0, i32 [[IDX1]] 217 // CHECK-DAG: [[PADDR1:%.+]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[P]], i32 0, i32 [[IDX1]] 218 // CHECK-DAG: [[SADDR2:%.+]] = getelementptr inbounds [10 x i[[SZ]]], [10 x i[[SZ]]]* [[S]], i32 0, i32 [[IDX2:[0-9]+]] 219 // CHECK-DAG: [[BPADDR2:%.+]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[BP]], i32 0, i32 [[IDX2]] 220 // CHECK-DAG: [[PADDR2:%.+]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[P]], i32 0, i32 [[IDX2]] 221 // CHECK-DAG: [[SADDR3:%.+]] = getelementptr inbounds [10 x i[[SZ]]], [10 x i[[SZ]]]* [[S]], i32 0, i32 [[IDX3:[0-9]+]] 222 // CHECK-DAG: [[BPADDR3:%.+]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[BP]], i32 0, i32 [[IDX3]] 223 // CHECK-DAG: [[PADDR3:%.+]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[P]], i32 0, i32 [[IDX3]] 224 // CHECK-DAG: [[SADDR4:%.+]] = getelementptr inbounds [10 x i[[SZ]]], [10 x i[[SZ]]]* [[S]], i32 0, i32 [[IDX4:[0-9]+]] 225 // CHECK-DAG: [[BPADDR4:%.+]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[BP]], i32 0, i32 [[IDX4]] 226 // CHECK-DAG: [[PADDR4:%.+]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[P]], i32 0, i32 [[IDX4]] 227 // CHECK-DAG: [[SADDR5:%.+]] = getelementptr inbounds [10 x i[[SZ]]], [10 x i[[SZ]]]* [[S]], i32 0, i32 [[IDX5:[0-9]+]] 228 // CHECK-DAG: [[BPADDR5:%.+]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[BP]], i32 0, i32 [[IDX5]] 229 // CHECK-DAG: [[PADDR5:%.+]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[P]], i32 0, i32 [[IDX5]] 230 // CHECK-DAG: [[SADDR6:%.+]] = getelementptr inbounds [10 x i[[SZ]]], [10 x i[[SZ]]]* [[S]], i32 0, i32 [[IDX6:[0-9]+]] 231 // CHECK-DAG: [[BPADDR6:%.+]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[BP]], i32 0, i32 [[IDX6]] 232 // CHECK-DAG: [[PADDR6:%.+]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[P]], i32 0, i32 [[IDX6]] 233 // CHECK-DAG: [[SADDR7:%.+]] = getelementptr inbounds [10 x i[[SZ]]], [10 x i[[SZ]]]* [[S]], i32 0, i32 [[IDX7:[0-9]+]] 234 // CHECK-DAG: [[BPADDR7:%.+]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[BP]], i32 0, i32 [[IDX7]] 235 // CHECK-DAG: [[PADDR7:%.+]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[P]], i32 0, i32 [[IDX7]] 236 // CHECK-DAG: [[SADDR8:%.+]] = getelementptr inbounds [10 x i[[SZ]]], [10 x i[[SZ]]]* [[S]], i32 0, i32 [[IDX8:[0-9]+]] 237 // CHECK-DAG: [[BPADDR8:%.+]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[BP]], i32 0, i32 [[IDX8]] 238 // CHECK-DAG: [[PADDR8:%.+]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[P]], i32 0, i32 [[IDX8]] 239 240 // The names below are not necessarily consistent with the names used for the 241 // addresses above as some are repeated. 242 // CHECK-DAG: store i[[SZ]] [[VLA0]], i[[SZ]]* [[CBPADDR0:%.+]], 243 // CHECK-DAG: store i[[SZ]] [[VLA0]], i[[SZ]]* [[CPADDR0:%.+]], 244 // CHECK-DAG: [[CBPADDR0]] = bitcast i8** {{%[^,]+}} to i[[SZ]]* 245 // CHECK-DAG: [[CPADDR0]] = bitcast i8** {{%[^,]+}} to i[[SZ]]* 246 // CHECK-DAG: store i[[SZ]] {{4|8}}, i[[SZ]]* {{%[^,]+}} 247 248 // CHECK-DAG: store i[[SZ]] [[VLA1]], i[[SZ]]* [[CBPADDR1:%.+]], 249 // CHECK-DAG: store i[[SZ]] [[VLA1]], i[[SZ]]* [[CPADDR1:%.+]], 250 // CHECK-DAG: [[CBPADDR1]] = bitcast i8** {{%[^,]+}} to i[[SZ]]* 251 // CHECK-DAG: [[CPADDR1]] = bitcast i8** {{%[^,]+}} to i[[SZ]]* 252 // CHECK-DAG: store i[[SZ]] {{4|8}}, i[[SZ]]* {{%[^,]+}} 253 254 // CHECK-DAG: store i[[SZ]] 5, i[[SZ]]* [[CBPADDR2:%.+]], 255 // CHECK-DAG: store i[[SZ]] 5, i[[SZ]]* [[CPADDR2:%.+]], 256 // CHECK-DAG: [[CBPADDR2]] = bitcast i8** {{%[^,]+}} to i[[SZ]]* 257 // CHECK-DAG: [[CPADDR2]] = bitcast i8** {{%[^,]+}} to i[[SZ]]* 258 // CHECK-DAG: store i[[SZ]] {{4|8}}, i[[SZ]]* {{%[^,]+}} 259 260 // CHECK-DAG: store i[[SZ]] [[A_CVAL]], i[[SZ]]* [[CBPADDR3:%.+]], 261 // CHECK-DAG: store i[[SZ]] [[A_CVAL]], i[[SZ]]* [[CPADDR3:%.+]], 262 // CHECK-DAG: [[CBPADDR3]] = bitcast i8** {{%[^,]+}} to i[[SZ]]* 263 // CHECK-DAG: [[CPADDR3]] = bitcast i8** {{%[^,]+}} to i[[SZ]]* 264 // CHECK-DAG: store i[[SZ]] 4, i[[SZ]]* {{%[^,]+}} 265 266 // CHECK-DAG: store [10 x float]* %{{.+}}, [10 x float]** [[CBPADDR4:%.+]], 267 // CHECK-DAG: store [10 x float]* %{{.+}}, [10 x float]** [[CPADDR4:%.+]], 268 // CHECK-DAG: [[CBPADDR4]] = bitcast i8** {{%[^,]+}} to [10 x float]** 269 // CHECK-DAG: [[CPADDR4]] = bitcast i8** {{%[^,]+}} to [10 x float]** 270 // CHECK-DAG: store i[[SZ]] 40, i[[SZ]]* {{%[^,]+}} 271 272 // CHECK-DAG: store float* %{{.+}}, float** [[CBPADDR5:%.+]], 273 // CHECK-DAG: store float* %{{.+}}, float** [[CPADDR5:%.+]], 274 // CHECK-DAG: [[CBPADDR5]] = bitcast i8** {{%[^,]+}} to float** 275 // CHECK-DAG: [[CPADDR5]] = bitcast i8** {{%[^,]+}} to float** 276 // CHECK-DAG: store i[[SZ]] [[BNSIZE]], i[[SZ]]* {{%[^,]+}} 277 278 // CHECK-DAG: store [5 x [10 x double]]* %{{.+}}, [5 x [10 x double]]** [[CBPADDR6:%.+]], 279 // CHECK-DAG: store [5 x [10 x double]]* %{{.+}}, [5 x [10 x double]]** [[CPADDR6:%.+]], 280 // CHECK-DAG: [[CBPADDR6]] = bitcast i8** {{%[^,]+}} to [5 x [10 x double]]** 281 // CHECK-DAG: [[CPADDR6]] = bitcast i8** {{%[^,]+}} to [5 x [10 x double]]** 282 // CHECK-DAG: store i[[SZ]] 400, i[[SZ]]* {{%[^,]+}} 283 284 // CHECK-DAG: store double* %{{.+}}, double** [[CBPADDR7:%.+]], 285 // CHECK-DAG: store double* %{{.+}}, double** [[CPADDR7:%.+]], 286 // CHECK-DAG: [[CBPADDR7]] = bitcast i8** {{%[^,]+}} to double** 287 // CHECK-DAG: [[CPADDR7]] = bitcast i8** {{%[^,]+}} to double** 288 // CHECK-DAG: store i[[SZ]] [[CNSIZE]], i[[SZ]]* {{%[^,]+}} 289 290 // CHECK-DAG: store [[TT]]* %{{.+}}, [[TT]]** [[CBPADDR8:%.+]], 291 // CHECK-DAG: store [[TT]]* %{{.+}}, [[TT]]** [[CPADDR8:%.+]], 292 // CHECK-DAG: [[CBPADDR8]] = bitcast i8** {{%[^,]+}} to [[TT]]** 293 // CHECK-DAG: [[CPADDR8]] = bitcast i8** {{%[^,]+}} to [[TT]]** 294 // CHECK-DAG: store i[[SZ]] {{12|16}}, i[[SZ]]* {{%[^,]+}} 295 296 // CHECK-NEXT: [[ERROR:%.+]] = icmp ne i32 [[RET]], 0 297 // CHECK-NEXT: br i1 [[ERROR]], label %[[FAIL:[^,]+]], label %[[END:[^,]+]] 298 299 // CHECK: [[FAIL]] 300 // CHECK: call void [[HVT4:@.+]]({{[^,]+}}, {{[^,]+}}, {{[^,]+}}, {{[^,]+}}, {{[^,]+}}, {{[^,]+}}, {{[^,]+}}, {{[^,]+}}, {{[^,]+}}, {{[^)]+}}) 301 // CHECK-NEXT: br label %[[END]] 302 // CHECK: [[END]] 303 #pragma omp target parallel for simd if(target: n>20) schedule(static, a) 304 for (unsigned char it = 'z'; it >= 'a'; it+=-1) { 305 a += 1; 306 b[2] += 1.0; 307 bn[3] += 1.0; 308 c[1][2] += 1.0; 309 cn[1][3] += 1.0; 310 d.X += 1; 311 d.Y += 1; 312 } 313 314 return a; 315 } 316 317 // Check that the offloading functions are emitted and that the arguments are 318 // correct and loaded correctly for the target regions in foo(). 319 320 // CHECK: define internal void [[HVT0]]() 321 // CHECK: call {{.*}}void (%ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%ident_t* [[DEF_LOC]], i32 0, void (i32*, i32*, ...)* bitcast (void (i32*, i32*)* [[OMP_OUTLINED:@.+]] to void (i32*, i32*, ...)*)) 322 // 323 // 324 // CHECK: define internal {{.*}}void [[OMP_OUTLINED]](i32* noalias %.global_tid., i32* noalias %.bound_tid.) 325 // CHECK: !llvm.loop 326 // CHECK: ret void 327 // CHECK-NEXT: } 328 329 330 // CHECK: define internal void [[HVT1]](i[[SZ]] %{{.+}}, i{{32|64}}{{[*]*.*}} %{{.+}}) 331 // Create stack storage and store argument in there. 332 // CHECK: [[AA_ADDR:%.+]] = alloca i[[SZ]], align 333 // CHECK: alloca i{{32|64}}{{[*]*}}, align 334 // CHECK: [[AA_CASTED:%.+]] = alloca i[[SZ]], align 335 // CHECK: store i[[SZ]] %{{.+}}, i[[SZ]]* [[AA_ADDR]], align 336 // CHECK-64: [[AA_CADDR:%.+]] = bitcast i[[SZ]]* [[AA_ADDR]] to i32* 337 // CHECK-64: [[AA:%.+]] = load i32, i32* [[AA_CADDR]], align 338 // CHECK-32: [[AA:%.+]] = load i32, i32* [[AA_ADDR]], align 339 // CHECK-64: [[AA_C:%.+]] = bitcast i[[SZ]]* [[AA_CASTED]] to i32* 340 // CHECK-64: store i32 [[AA]], i32* [[AA_C]], align 341 // CHECK-32: store i32 [[AA]], i32* [[AA_CASTED]], align 342 // CHECK: [[PARAM:%.+]] = load i[[SZ]], i[[SZ]]* [[AA_CASTED]], align 343 // CHECK: call {{.*}}void (%ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%ident_t* [[DEF_LOC]], i32 2, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i[[SZ]], i{{32|64}}{{[*]*}})* [[OMP_OUTLINED1:@.+]] to void (i32*, i32*, ...)*), i[[SZ]] [[PARAM]], i{{32|64}}{{[*]*}} %{{.+}}) 344 // 345 // 346 // CHECK: define internal {{.*}}void [[OMP_OUTLINED1]](i32* noalias %.global_tid., i32* noalias %.bound_tid., i[[SZ]] %{{.+}}) 347 // CHECK: [[AA_ADDR:%.+]] = alloca i[[SZ]], align 348 // CHECK: store i[[SZ]] %{{.+}}, i[[SZ]]* [[AA_ADDR]], align 349 // CHECK-64: [[AA_CADDR:%.+]] = bitcast i[[SZ]]* [[AA_ADDR]] to i32* 350 // CHECK-64: [[AA:%.+]] = load i32, i32* [[AA_CADDR]], align 351 // CHECK-32: [[AA:%.+]] = load i32, i32* [[AA_ADDR]], align 352 // CHECK: !llvm.mem.parallel_loop_access 353 // CHECK: !llvm.loop 354 // CHECK: ret void 355 // CHECK-NEXT: } 356 357 // CHECK: define internal void [[HVT2]](i[[SZ]] %{{.+}}, i[[SZ]] %{{.+}}, i[[SZ]] %{{.+}}) 358 // Create stack storage and store argument in there. 359 // CHECK: [[AA_ADDR:%.+]] = alloca i[[SZ]], align 360 // CHECK: alloca i[[SZ]], align 361 // CHECK: alloca i[[SZ]], align 362 // CHECK: [[AA_CASTED:%.+]] = alloca i[[SZ]], align 363 // CHECK: store i[[SZ]] %{{.+}}, i[[SZ]]* [[AA_ADDR]], align 364 // CHECK: [[AA_CADDR:%.+]] = bitcast i[[SZ]]* [[AA_ADDR]] to i16* 365 // CHECK: [[AA:%.+]] = load i16, i16* [[AA_CADDR]], align 366 // CHECK: [[AA_C:%.+]] = bitcast i[[SZ]]* [[AA_CASTED]] to i16* 367 // CHECK: store i16 [[AA]], i16* [[AA_C]], align 368 // CHECK: [[PARAM:%.+]] = load i[[SZ]], i[[SZ]]* [[AA_CASTED]], align 369 // CHECK: call {{.*}}void (%ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%ident_t* [[DEF_LOC]], i32 3, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i[[SZ]], i[[SZ]], i[[SZ]])* [[OMP_OUTLINED2:@.+]] to void (i32*, i32*, ...)*), i[[SZ]] [[PARAM]], i[[SZ]] {{.+}}, i[[SZ]] {{.+}}) 370 // 371 // 372 // CHECK: define internal {{.*}}void [[OMP_OUTLINED2]](i32* noalias %.global_tid., i32* noalias %.bound_tid., i[[SZ]] %{{.+}}, i[[SZ]] %{{.+}}, i[[SZ]] %{{.+}}) 373 // CHECK: [[AA_ADDR:%.+]] = alloca i[[SZ]], align 374 // CHECK: store i[[SZ]] %{{.+}}, i[[SZ]]* [[AA_ADDR]], align 375 // CHECK: [[AA_CADDR:%.+]] = bitcast i[[SZ]]* [[AA_ADDR]] to i16* 376 // CHECK: [[AA:%.+]] = load i16, i16* [[AA_CADDR]], align 377 // CHECK: !llvm.loop 378 // CHECK: ret void 379 // CHECK-NEXT: } 380 381 // CHECK: define internal void [[HVT3]] 382 // Create stack storage and store argument in there. 383 // CHECK: [[A_ADDR:%.+]] = alloca i[[SZ]], align 384 // CHECK: [[AA_ADDR:%.+]] = alloca i[[SZ]], align 385 // CHECK: [[A_CASTED:%.+]] = alloca i[[SZ]], align 386 // CHECK: [[AA_CASTED:%.+]] = alloca i[[SZ]], align 387 // CHECK-DAG: store i[[SZ]] %{{.+}}, i[[SZ]]* [[A_ADDR]], align 388 // CHECK-DAG: store i[[SZ]] %{{.+}}, i[[SZ]]* [[AA_ADDR]], align 389 // CHECK-64-DAG:[[A_CADDR:%.+]] = bitcast i[[SZ]]* [[A_ADDR]] to i32* 390 // CHECK-DAG: [[AA_CADDR:%.+]] = bitcast i[[SZ]]* [[AA_ADDR]] to i16* 391 // CHECK-64-DAG:[[A:%.+]] = load i32, i32* [[A_CADDR]], align 392 // CHECK-32-DAG:[[A:%.+]] = load i32, i32* [[A_ADDR]], align 393 // CHECK-64-DAG:[[A_C:%.+]] = bitcast i[[SZ]]* [[A_CASTED]] to i32* 394 // CHECK-64-DAG:store i32 [[A]], i32* [[A_C]], align 395 // CHECK-32-DAG:store i32 [[A]], i32* [[A_CASTED]], align 396 // CHECK-DAG: [[AA:%.+]] = load i16, i16* [[AA_CADDR]], align 397 // CHECK-DAG: [[AA_C:%.+]] = bitcast i[[SZ]]* [[AA_CASTED]] to i16* 398 // CHECK-DAG: store i16 [[AA]], i16* [[AA_C]], align 399 // CHECK-DAG: [[PARAM1:%.+]] = load i[[SZ]], i[[SZ]]* [[A_CASTED]], align 400 // CHECK-DAG: [[PARAM2:%.+]] = load i[[SZ]], i[[SZ]]* [[AA_CASTED]], align 401 // CHECK-DAG: call {{.*}}void (%ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%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]]) 402 // 403 // 404 // CHECK: define internal {{.*}}void [[OMP_OUTLINED3]](i32* noalias %.global_tid., i32* noalias %.bound_tid., i[[SZ]] %{{.+}}, i[[SZ]] %{{.+}}) 405 // CHECK: [[A_ADDR:%.+]] = alloca i[[SZ]], align 406 // CHECK: [[AA_ADDR:%.+]] = alloca i[[SZ]], align 407 // CHECK-DAG: store i[[SZ]] %{{.+}}, i[[SZ]]* [[A_ADDR]], align 408 // CHECK-DAG: store i[[SZ]] %{{.+}}, i[[SZ]]* [[AA_ADDR]], align 409 // CHECK-64-DAG:[[A_CADDR:%.+]] = bitcast i[[SZ]]* [[A_ADDR]] to i32* 410 // CHECK-DAG: [[AA_CADDR:%.+]] = bitcast i[[SZ]]* [[AA_ADDR]] to i16* 411 // CHECK: !llvm.loop 412 // CHECK: ret void 413 // CHECK-NEXT: } 414 415 // CHECK: define internal void [[HVT4]] 416 // Create local storage for each capture. 417 // CHECK: [[LOCAL_A:%.+]] = alloca i[[SZ]] 418 // CHECK: [[LOCAL_B:%.+]] = alloca [10 x float]* 419 // CHECK: [[LOCAL_VLA1:%.+]] = alloca i[[SZ]] 420 // CHECK: [[LOCAL_BN:%.+]] = alloca float* 421 // CHECK: [[LOCAL_C:%.+]] = alloca [5 x [10 x double]]* 422 // CHECK: [[LOCAL_VLA2:%.+]] = alloca i[[SZ]] 423 // CHECK: [[LOCAL_VLA3:%.+]] = alloca i[[SZ]] 424 // CHECK: [[LOCAL_CN:%.+]] = alloca double* 425 // CHECK: [[LOCAL_D:%.+]] = alloca [[TT]]* 426 // CHECK: alloca i[[SZ]] 427 // CHECK: [[LOCAL_A_CASTED:%.+]] = alloca i[[SZ]] 428 // CHECK-DAG: store i[[SZ]] [[ARG_A:%.+]], i[[SZ]]* [[LOCAL_A]] 429 // CHECK-DAG: store [10 x float]* [[ARG_B:%.+]], [10 x float]** [[LOCAL_B]] 430 // CHECK-DAG: store i[[SZ]] [[ARG_VLA1:%.+]], i[[SZ]]* [[LOCAL_VLA1]] 431 // CHECK-DAG: store float* [[ARG_BN:%.+]], float** [[LOCAL_BN]] 432 // CHECK-DAG: store [5 x [10 x double]]* [[ARG_C:%.+]], [5 x [10 x double]]** [[LOCAL_C]] 433 // CHECK-DAG: store i[[SZ]] [[ARG_VLA2:%.+]], i[[SZ]]* [[LOCAL_VLA2]] 434 // CHECK-DAG: store i[[SZ]] [[ARG_VLA3:%.+]], i[[SZ]]* [[LOCAL_VLA3]] 435 // CHECK-DAG: store double* [[ARG_CN:%.+]], double** [[LOCAL_CN]] 436 // CHECK-DAG: store [[TT]]* [[ARG_D:%.+]], [[TT]]** [[LOCAL_D]] 437 438 // CHECK-64-DAG:[[CONV_AP:%.+]] = bitcast i[[SZ]]* [[LOCAL_A]] to i32* 439 // CHECK-DAG: [[REF_B:%.+]] = load [10 x float]*, [10 x float]** [[LOCAL_B]], 440 // CHECK-DAG: [[VAL_VLA1:%.+]] = load i[[SZ]], i[[SZ]]* [[LOCAL_VLA1]], 441 // CHECK-DAG: [[REF_BN:%.+]] = load float*, float** [[LOCAL_BN]], 442 // CHECK-DAG: [[REF_C:%.+]] = load [5 x [10 x double]]*, [5 x [10 x double]]** [[LOCAL_C]], 443 // CHECK-DAG: [[VAL_VLA2:%.+]] = load i[[SZ]], i[[SZ]]* [[LOCAL_VLA2]], 444 // CHECK-DAG: [[VAL_VLA3:%.+]] = load i[[SZ]], i[[SZ]]* [[LOCAL_VLA3]], 445 // CHECK-DAG: [[REF_CN:%.+]] = load double*, double** [[LOCAL_CN]], 446 // CHECK-DAG: [[REF_D:%.+]] = load [[TT]]*, [[TT]]** [[LOCAL_D]], 447 448 // CHECK-64-DAG:[[CONV_A:%.+]] = load i32, i32* [[CONV_AP]] 449 // CHECK-64-DAG:[[CONV:%.+]] = bitcast i[[SZ]]* [[LOCAL_A_CASTED]] to i32* 450 // CHECK-64-DAG:store i32 [[CONV_A]], i32* [[CONV]], align 451 // CHECK-32-DAG:[[LOCAL_AV:%.+]] = load i32, i32* [[LOCAL_A]] 452 // CHECK-32-DAG:store i32 [[LOCAL_AV]], i32* [[LOCAL_A_CASTED]], align 453 // CHECK-DAG: [[REF_A:%.+]] = load i[[SZ]], i[[SZ]]* [[LOCAL_A_CASTED]], 454 455 // CHECK: call {{.*}}void (%ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%ident_t* [[DEF_LOC]], i32 10, 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]]*, i[[SZ]])* [[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]], i[[SZ]] %{{.+}}) 456 // 457 // 458 // 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]]* {{.+}}) 459 // To reduce complexity, we're only going as far as validating the signature of the outlined parallel function. 460 461 template<typename tx> 462 tx ftemplate(int n) { 463 tx a = 0; 464 short aa = 0; 465 tx b[10]; 466 467 #pragma omp target parallel for simd if(target: n>40) 468 for (long long i = -10; i < 10; i += 3) { 469 a += 1; 470 aa += 1; 471 b[2] += 1; 472 } 473 474 return a; 475 } 476 477 static 478 int fstatic(int n) { 479 int a = 0; 480 short aa = 0; 481 char aaa = 0; 482 int b[10]; 483 484 #pragma omp target parallel for simd if(target: n>50) 485 for (unsigned i=100; i<10; i+=10) { 486 a += 1; 487 aa += 1; 488 aaa += 1; 489 b[2] += 1; 490 } 491 492 return a; 493 } 494 495 struct S1 { 496 double a; 497 498 int r1(int n){ 499 int b = n+1; 500 short int c[2][n]; 501 502 #pragma omp target parallel for simd if(target: n>60) 503 for (unsigned long long it = 2000; it >= 600; it -= 400) { 504 this->a = (double)b + 1.5; 505 c[1][1] = ++a; 506 } 507 508 return c[1][1] + (int)b; 509 } 510 }; 511 512 // CHECK: define {{.*}}@{{.*}}bar{{.*}} 513 int bar(int n){ 514 int a = 0; 515 516 // CHECK: call {{.*}}i32 [[FOO]](i32 {{.*}}) 517 a += foo(n); 518 519 S1 S; 520 // CHECK: call {{.*}}i32 [[FS1:@.+]]([[S1]]* {{.*}}, i32 {{.*}}) 521 a += S.r1(n); 522 523 // CHECK: call {{.*}}i32 [[FSTATIC:@.+]](i32 {{.*}}) 524 a += fstatic(n); 525 526 // CHECK: call {{.*}}i32 [[FTEMPLATE:@.+]](i32 {{.*}}) 527 a += ftemplate<int>(n); 528 529 return a; 530 } 531 532 // 533 // CHECK: define {{.*}}[[FS1]] 534 // 535 // CHECK: i8* @llvm.stacksave() 536 // CHECK-64: [[B_ADDR:%.+]] = bitcast i[[SZ]]* [[B_CADDR:%.+]] to i32* 537 // CHECK-64: store i32 %{{.+}}, i32* [[B_ADDR]], 538 // CHECK-64: [[B_CVAL:%.+]] = load i[[SZ]], i[[SZ]]* [[B_CADDR]], 539 540 // CHECK-32: store i32 %{{.+}}, i32* [[B_ADDR:%.+]], 541 // CHECK-32: [[B_CVAL:%.+]] = load i[[SZ]], i[[SZ]]* [[B_ADDR]], 542 543 // We capture 2 VLA sizes in this target region 544 // CHECK: [[CELEMSIZE2:%.+]] = mul nuw i[[SZ]] 2, [[VLA0:%.+]] 545 // CHECK: [[CSIZE:%.+]] = mul nuw i[[SZ]] [[CELEMSIZE2]], 2 546 547 // CHECK: [[IF:%.+]] = icmp sgt i32 {{[^,]+}}, 60 548 // CHECK: br i1 [[IF]], label %[[TRY:[^,]+]], label %[[FAIL:[^,]+]] 549 // CHECK: [[TRY]] 550 // CHECK-DAG: [[RET:%.+]] = call i32 @__tgt_target_teams(i32 -1, i8* @{{[^,]+}}, i32 5, i8** [[BPR:%[^,]+]], i8** [[PR:%[^,]+]], i[[SZ]]* [[SR:%[^,]+]], i32* getelementptr inbounds ([5 x i32], [5 x i32]* [[MAPT7]], i32 0, i32 0), i32 1, i32 0) 551 // CHECK-DAG: [[BPR]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[BP:%.+]], i32 0, i32 0 552 // CHECK-DAG: [[PR]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[P:%.+]], i32 0, i32 0 553 // CHECK-DAG: [[SR]] = getelementptr inbounds [5 x i[[SZ]]], [5 x i[[SZ]]]* [[S:%.+]], i32 0, i32 0 554 // CHECK-DAG: [[SADDR0:%.+]] = getelementptr inbounds [5 x i[[SZ]]], [5 x i[[SZ]]]* [[S]], i32 [[IDX0:[0-9]+]] 555 // CHECK-DAG: [[BPADDR0:%.+]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[BP]], i32 [[IDX0]] 556 // CHECK-DAG: [[PADDR0:%.+]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[P]], i32 [[IDX0]] 557 // CHECK-DAG: [[SADDR1:%.+]] = getelementptr inbounds [5 x i[[SZ]]], [5 x i[[SZ]]]* [[S]], i32 [[IDX1:[0-9]+]] 558 // CHECK-DAG: [[BPADDR1:%.+]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[BP]], i32 [[IDX1]] 559 // CHECK-DAG: [[PADDR1:%.+]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[P]], i32 [[IDX1]] 560 // CHECK-DAG: [[SADDR2:%.+]] = getelementptr inbounds [5 x i[[SZ]]], [5 x i[[SZ]]]* [[S]], i32 [[IDX2:[0-9]+]] 561 // CHECK-DAG: [[BPADDR2:%.+]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[BP]], i32 [[IDX2]] 562 // CHECK-DAG: [[PADDR2:%.+]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[P]], i32 [[IDX2]] 563 // CHECK-DAG: [[SADDR3:%.+]] = getelementptr inbounds [5 x i[[SZ]]], [5 x i[[SZ]]]* [[S]], i32 [[IDX3:[0-9]+]] 564 // CHECK-DAG: [[BPADDR3:%.+]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[BP]], i32 [[IDX3]] 565 // CHECK-DAG: [[PADDR3:%.+]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[P]], i32 [[IDX3]] 566 567 // The names below are not necessarily consistent with the names used for the 568 // addresses above as some are repeated. 569 // CHECK-DAG: store i[[SZ]] [[VLA0]], i[[SZ]]* [[CBPADDR0:%.+]], 570 // CHECK-DAG: store i[[SZ]] [[VLA0]], i[[SZ]]* [[CPADDR0:%.+]], 571 // CHECK-DAG: [[CBPADDR0]] = bitcast i8** {{%[^,]+}} to i[[SZ]]* 572 // CHECK-DAG: [[CPADDR0]] = bitcast i8** {{%[^,]+}} to i[[SZ]]* 573 // CHECK-DAG: store i[[SZ]] {{4|8}}, i[[SZ]]* {{%[^,]+}} 574 575 // CHECK-DAG: store i[[SZ]] 2, i[[SZ]]* [[CBPADDR1:%.+]], 576 // CHECK-DAG: store i[[SZ]] 2, i[[SZ]]* [[CPADDR1:%.+]], 577 // CHECK-DAG: [[CBPADDR1]] = bitcast i8** {{%[^,]+}} to i[[SZ]]* 578 // CHECK-DAG: [[CPADDR1]] = bitcast i8** {{%[^,]+}} to i[[SZ]]* 579 // CHECK-DAG: store i[[SZ]] {{4|8}}, i[[SZ]]* {{%[^,]+}} 580 581 // CHECK-DAG: store i[[SZ]] [[B_CVAL]], i[[SZ]]* [[CBPADDR2:%.+]], 582 // CHECK-DAG: store i[[SZ]] [[B_CVAL]], i[[SZ]]* [[CPADDR2:%.+]], 583 // CHECK-DAG: [[CBPADDR2]] = bitcast i8** {{%[^,]+}} to i[[SZ]]* 584 // CHECK-DAG: [[CPADDR2]] = bitcast i8** {{%[^,]+}} to i[[SZ]]* 585 // CHECK-DAG: store i[[SZ]] 4, i[[SZ]]* {{%[^,]+}} 586 587 // CHECK-DAG: store [[S1]]* %{{.+}}, [[S1]]** [[CBPADDR3:%.+]], 588 // CHECK-DAG: store [[S1]]* %{{.+}}, [[S1]]** [[CPADDR3:%.+]], 589 // CHECK-DAG: [[CBPADDR3]] = bitcast i8** {{%[^,]+}} to [[S1]]** 590 // CHECK-DAG: [[CPADDR3]] = bitcast i8** {{%[^,]+}} to [[S1]]** 591 // CHECK-DAG: store i[[SZ]] 8, i[[SZ]]* {{%[^,]+}} 592 593 // CHECK-DAG: store i16* %{{.+}}, i16** [[CBPADDR4:%.+]], 594 // CHECK-DAG: store i16* %{{.+}}, i16** [[CPADDR4:%.+]], 595 // CHECK-DAG: [[CBPADDR4]] = bitcast i8** {{%[^,]+}} to i16** 596 // CHECK-DAG: [[CPADDR4]] = bitcast i8** {{%[^,]+}} to i16** 597 // CHECK-DAG: store i[[SZ]] [[CSIZE]], i[[SZ]]* {{%[^,]+}} 598 599 // CHECK-NEXT: [[ERROR:%.+]] = icmp ne i32 [[RET]], 0 600 // CHECK-NEXT: br i1 [[ERROR]], label %[[FAIL:[^,]+]], label %[[END:[^,]+]] 601 602 // CHECK: [[FAIL]] 603 // CHECK: call void [[HVT7:@.+]]({{[^,]+}}, {{[^,]+}}, {{[^,]+}}, {{[^,]+}}, {{[^,]+}}) 604 // CHECK-NEXT: br label %[[END]] 605 // CHECK: [[END]] 606 607 // 608 // CHECK: define {{.*}}[[FSTATIC]] 609 // 610 // CHECK: [[IF:%.+]] = icmp sgt i32 {{[^,]+}}, 50 611 // CHECK: br i1 [[IF]], label %[[IFTHEN:[^,]+]], label %[[IFELSE:[^,]+]] 612 // CHECK: [[IFTHEN]] 613 // CHECK-DAG: [[RET:%.+]] = call i32 @__tgt_target_teams(i32 -1, i8* @{{[^,]+}}, i32 4, i8** [[BPR:%[^,]+]], i8** [[PR:%[^,]+]], i[[SZ]]* getelementptr inbounds ([4 x i[[SZ]]], [4 x i[[SZ]]]* [[SIZET6]], i32 0, i32 0), i32* getelementptr inbounds ([4 x i32], [4 x i32]* [[MAPT6]], i32 0, i32 0), i32 1, i32 0) 614 // CHECK-DAG: [[BPR]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[BP:%.+]], i32 0, i32 0 615 // CHECK-DAG: [[PR]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[P:%.+]], i32 0, i32 0 616 617 // CHECK-DAG: [[BPADDR0:%.+]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[BP]], i32 0, i32 0 618 // CHECK-DAG: [[PADDR0:%.+]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[P]], i32 0, i32 0 619 // CHECK-DAG: [[CBPADDR0:%.+]] = bitcast i8** [[BPADDR0]] to i[[SZ]]* 620 // CHECK-DAG: [[CPADDR0:%.+]] = bitcast i8** [[PADDR0]] to i[[SZ]]* 621 // CHECK-DAG: store i[[SZ]] [[VAL0:%.+]], i[[SZ]]* [[CBPADDR0]], 622 // CHECK-DAG: store i[[SZ]] [[VAL0]], i[[SZ]]* [[CPADDR0]], 623 624 // CHECK-DAG: [[BPADDR1:%.+]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[BP]], i32 0, i32 1 625 // CHECK-DAG: [[PADDR1:%.+]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[P]], i32 0, i32 1 626 // CHECK-DAG: [[CBPADDR1:%.+]] = bitcast i8** [[BPADDR1]] to i[[SZ]]* 627 // CHECK-DAG: [[CPADDR1:%.+]] = bitcast i8** [[PADDR1]] to i[[SZ]]* 628 // CHECK-DAG: store i[[SZ]] [[VAL1:%.+]], i[[SZ]]* [[CBPADDR1]], 629 // CHECK-DAG: store i[[SZ]] [[VAL1]], i[[SZ]]* [[CPADDR1]], 630 631 // CHECK-DAG: [[BPADDR2:%.+]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[BP]], i32 0, i32 2 632 // CHECK-DAG: [[PADDR2:%.+]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[P]], i32 0, i32 2 633 // CHECK-DAG: [[CBPADDR2:%.+]] = bitcast i8** [[BPADDR2]] to i[[SZ]]* 634 // CHECK-DAG: [[CPADDR2:%.+]] = bitcast i8** [[PADDR2]] to i[[SZ]]* 635 // CHECK-DAG: store i[[SZ]] [[VAL2:%.+]], i[[SZ]]* [[CBPADDR2]], 636 // CHECK-DAG: store i[[SZ]] [[VAL2]], i[[SZ]]* [[CPADDR2]], 637 638 // CHECK-DAG: [[BPADDR3:%.+]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[BP]], i32 0, i32 3 639 // CHECK-DAG: [[PADDR3:%.+]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[P]], i32 0, i32 3 640 // CHECK-DAG: [[CBPADDR3:%.+]] = bitcast i8** [[BPADDR3]] to [10 x i32]** 641 // CHECK-DAG: [[CPADDR3:%.+]] = bitcast i8** [[PADDR3]] to [10 x i32]** 642 // CHECK-DAG: store [10 x i32]* [[VAL3:%.+]], [10 x i32]** [[CBPADDR3]], 643 // CHECK-DAG: store [10 x i32]* [[VAL3]], [10 x i32]** [[CPADDR3]], 644 645 // CHECK: [[ERROR:%.+]] = icmp ne i32 [[RET]], 0 646 // CHECK-NEXT: br i1 [[ERROR]], label %[[FAIL:.+]], label %[[END:[^,]+]] 647 // CHECK: [[FAIL]] 648 // CHECK: call void [[HVT6:@.+]]({{[^,]+}}, {{[^,]+}}, {{[^,]+}}, {{[^,]+}}) 649 // CHECK-NEXT: br label %[[END]] 650 // CHECK: [[END]] 651 // CHECK-NEXT: br label %[[IFEND:.+]] 652 // CHECK: [[IFELSE]] 653 // CHECK: call void [[HVT6]]({{[^,]+}}, {{[^,]+}}, {{[^,]+}}, {{[^,]+}}) 654 // CHECK-NEXT: br label %[[IFEND]] 655 // CHECK: [[IFEND]] 656 657 // 658 // CHECK: define {{.*}}[[FTEMPLATE]] 659 // 660 // CHECK: [[IF:%.+]] = icmp sgt i32 {{[^,]+}}, 40 661 // CHECK: br i1 [[IF]], label %[[IFTHEN:[^,]+]], label %[[IFELSE:[^,]+]] 662 // CHECK: [[IFTHEN]] 663 // CHECK-DAG: [[RET:%.+]] = call i32 @__tgt_target_teams(i32 -1, i8* @{{[^,]+}}, i32 3, i8** [[BPR:%[^,]+]], i8** [[PR:%[^,]+]], i[[SZ]]* getelementptr inbounds ([3 x i[[SZ]]], [3 x i[[SZ]]]* [[SIZET5]], i32 0, i32 0), i32* getelementptr inbounds ([3 x i32], [3 x i32]* [[MAPT5]], i32 0, i32 0), i32 1, i32 0) 664 // CHECK-DAG: [[BPR]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[BP:%.+]], i32 0, i32 0 665 // CHECK-DAG: [[PR]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[P:%.+]], i32 0, i32 0 666 667 // CHECK-DAG: [[BPADDR0:%.+]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[BP]], i32 0, i32 0 668 // CHECK-DAG: [[PADDR0:%.+]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[P]], i32 0, i32 0 669 // CHECK-DAG: [[CBPADDR0:%.+]] = bitcast i8** [[BPADDR0]] to i[[SZ]]* 670 // CHECK-DAG: [[CPADDR0:%.+]] = bitcast i8** [[PADDR0]] to i[[SZ]]* 671 // CHECK-DAG: store i[[SZ]] [[VAL0:%.+]], i[[SZ]]* [[CBPADDR0]], 672 // CHECK-DAG: store i[[SZ]] [[VAL0]], i[[SZ]]* [[CPADDR0]], 673 674 // CHECK-DAG: [[BPADDR1:%.+]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[BP]], i32 0, i32 1 675 // CHECK-DAG: [[PADDR1:%.+]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[P]], i32 0, i32 1 676 // CHECK-DAG: [[CBPADDR1:%.+]] = bitcast i8** [[BPADDR1]] to i[[SZ]]* 677 // CHECK-DAG: [[CPADDR1:%.+]] = bitcast i8** [[PADDR1]] to i[[SZ]]* 678 // CHECK-DAG: store i[[SZ]] [[VAL1:%.+]], i[[SZ]]* [[CBPADDR1]], 679 // CHECK-DAG: store i[[SZ]] [[VAL1]], i[[SZ]]* [[CPADDR1]], 680 681 // CHECK-DAG: [[BPADDR2:%.+]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[BP]], i32 0, i32 2 682 // CHECK-DAG: [[PADDR2:%.+]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[P]], i32 0, i32 2 683 // CHECK-DAG: [[CBPADDR2:%.+]] = bitcast i8** [[BPADDR2]] to [10 x i32]** 684 // CHECK-DAG: [[CPADDR2:%.+]] = bitcast i8** [[PADDR2]] to [10 x i32]** 685 // CHECK-DAG: store [10 x i32]* [[VAL2:%.+]], [10 x i32]** [[CBPADDR2]], 686 // CHECK-DAG: store [10 x i32]* [[VAL2]], [10 x i32]** [[CPADDR2]], 687 688 // CHECK: [[ERROR:%.+]] = icmp ne i32 [[RET]], 0 689 // CHECK-NEXT: br i1 [[ERROR]], label %[[FAIL:.+]], label %[[END:[^,]+]] 690 // CHECK: [[FAIL]] 691 // CHECK: call void [[HVT5:@.+]]({{[^,]+}}, {{[^,]+}}, {{[^,]+}}) 692 // CHECK-NEXT: br label %[[END]] 693 // CHECK: [[END]] 694 // CHECK-NEXT: br label %[[IFEND:.+]] 695 // CHECK: [[IFELSE]] 696 // CHECK: call void [[HVT:@.+]]({{[^,]+}}, {{[^,]+}}, {{[^,]+}}) 697 // CHECK-NEXT: br label %[[IFEND]] 698 // CHECK: [[IFEND]] 699 700 // Check that the offloading functions are emitted and that the arguments are 701 // correct and loaded correctly for the target regions of the callees of bar(). 702 703 // CHECK: define internal void [[HVT7]] 704 // Create local storage for each capture. 705 // CHECK: [[LOCAL_THIS:%.+]] = alloca [[S1]]* 706 // CHECK: [[LOCAL_B:%.+]] = alloca i[[SZ]] 707 // CHECK: [[LOCAL_VLA1:%.+]] = alloca i[[SZ]] 708 // CHECK: [[LOCAL_VLA2:%.+]] = alloca i[[SZ]] 709 // CHECK: [[LOCAL_C:%.+]] = alloca i16* 710 // CHECK: [[LOCAL_B_CASTED:%.+]] = alloca i[[SZ]] 711 // CHECK-DAG: store [[S1]]* [[ARG_THIS:%.+]], [[S1]]** [[LOCAL_THIS]] 712 // CHECK-DAG: store i[[SZ]] [[ARG_B:%.+]], i[[SZ]]* [[LOCAL_B]] 713 // CHECK-DAG: store i[[SZ]] [[ARG_VLA1:%.+]], i[[SZ]]* [[LOCAL_VLA1]] 714 // CHECK-DAG: store i[[SZ]] [[ARG_VLA2:%.+]], i[[SZ]]* [[LOCAL_VLA2]] 715 // CHECK-DAG: store i16* [[ARG_C:%.+]], i16** [[LOCAL_C]] 716 // Store captures in the context. 717 // CHECK-DAG: [[REF_THIS:%.+]] = load [[S1]]*, [[S1]]** [[LOCAL_THIS]], 718 // CHECK-64-DAG:[[CONV_BP:%.+]] = bitcast i[[SZ]]* [[LOCAL_B]] to i32* 719 // CHECK-DAG: [[VAL_VLA1:%.+]] = load i[[SZ]], i[[SZ]]* [[LOCAL_VLA1]], 720 // CHECK-DAG: [[VAL_VLA2:%.+]] = load i[[SZ]], i[[SZ]]* [[LOCAL_VLA2]], 721 // CHECK-DAG: [[REF_C:%.+]] = load i16*, i16** [[LOCAL_C]], 722 723 // CHECK-64-DAG:[[CONV_B:%.+]] = load i32, i32* [[CONV_BP]] 724 // CHECK-64-DAG:[[CONV:%.+]] = bitcast i[[SZ]]* [[LOCAL_B_CASTED]] to i32* 725 // CHECK-64-DAG:store i32 [[CONV_B]], i32* [[CONV]], align 726 // CHECK-32-DAG:[[LOCAL_BV:%.+]] = load i32, i32* [[LOCAL_B]] 727 // CHECK-32-DAG:store i32 [[LOCAL_BV]], i32* [[LOCAL_B_CASTED]], align 728 // CHECK-DAG: [[REF_B:%.+]] = load i[[SZ]], i[[SZ]]* [[LOCAL_B_CASTED]], 729 730 // CHECK: call {{.*}}void (%ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%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]]) 731 // 732 // 733 // CHECK: define internal {{.*}}void [[OMP_OUTLINED5]](i32* noalias %.global_tid., i32* noalias %.bound_tid., [[S1]]* %{{.+}}, i[[SZ]] %{{.+}}, i[[SZ]] %{{.+}}, i[[SZ]] %{{.+}}, i16* {{.+}}) 734 // To reduce complexity, we're only going as far as validating the signature of the outlined parallel function. 735 736 737 // CHECK: define internal void [[HVT6]] 738 // Create local storage for each capture. 739 // CHECK: [[LOCAL_A:%.+]] = alloca i[[SZ]] 740 // CHECK: [[LOCAL_AA:%.+]] = alloca i[[SZ]] 741 // CHECK: [[LOCAL_AAA:%.+]] = alloca i[[SZ]] 742 // CHECK: [[LOCAL_B:%.+]] = alloca [10 x i32]* 743 // CHECK: [[LOCAL_A_CASTED:%.+]] = alloca i[[SZ]] 744 // CHECK: [[LOCAL_AA_CASTED:%.+]] = alloca i[[SZ]] 745 // CHECK: [[LOCAL_AAA_CASTED:%.+]] = alloca i[[SZ]] 746 // CHECK-DAG: store i[[SZ]] [[ARG_A:%.+]], i[[SZ]]* [[LOCAL_A]] 747 // CHECK-DAG: store i[[SZ]] [[ARG_AA:%.+]], i[[SZ]]* [[LOCAL_AA]] 748 // CHECK-DAG: store i[[SZ]] [[ARG_AAA:%.+]], i[[SZ]]* [[LOCAL_AAA]] 749 // CHECK-DAG: store [10 x i32]* [[ARG_B:%.+]], [10 x i32]** [[LOCAL_B]] 750 // Store captures in the context. 751 // CHECK-64-DAG:[[CONV_AP:%.+]] = bitcast i[[SZ]]* [[LOCAL_A]] to i32* 752 // CHECK-DAG: [[CONV_AAP:%.+]] = bitcast i[[SZ]]* [[LOCAL_AA]] to i16* 753 // CHECK-DAG: [[CONV_AAAP:%.+]] = bitcast i[[SZ]]* [[LOCAL_AAA]] to i8* 754 // CHECK-DAG: [[REF_B:%.+]] = load [10 x i32]*, [10 x i32]** [[LOCAL_B]], 755 756 // CHECK-64-DAG:[[CONV_A:%.+]] = load i32, i32* [[CONV_AP]] 757 // CHECK-64-DAG:[[CONV:%.+]] = bitcast i[[SZ]]* [[LOCAL_A_CASTED]] to i32* 758 // CHECK-64-DAG:store i32 [[CONV_A]], i32* [[CONV]], align 759 // CHECK-32-DAG:[[LOCAL_AV:%.+]] = load i32, i32* [[LOCAL_A]] 760 // CHECK-32-DAG:store i32 [[LOCAL_AV]], i32* [[LOCAL_A_CASTED]], align 761 // CHECK-DAG: [[REF_A:%.+]] = load i[[SZ]], i[[SZ]]* [[LOCAL_A_CASTED]], 762 763 // CHECK-DAG: [[CONV_AA:%.+]] = load i16, i16* [[CONV_AAP]] 764 // CHECK-DAG: [[CONV:%.+]] = bitcast i[[SZ]]* [[LOCAL_AA_CASTED]] to i16* 765 // CHECK-DAG: store i16 [[CONV_AA]], i16* [[CONV]], align 766 // CHECK-DAG: [[REF_AA:%.+]] = load i[[SZ]], i[[SZ]]* [[LOCAL_AA_CASTED]], 767 768 // CHECK-DAG: [[CONV_AAA:%.+]] = load i8, i8* [[CONV_AAAP]] 769 // CHECK-DAG: [[CONV:%.+]] = bitcast i[[SZ]]* [[LOCAL_AAA_CASTED]] to i8* 770 // CHECK-DAG: store i8 [[CONV_AAA]], i8* [[CONV]], align 771 // CHECK-DAG: [[REF_AAA:%.+]] = load i[[SZ]], i[[SZ]]* [[LOCAL_AAA_CASTED]], 772 773 // CHECK: call {{.*}}void (%ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%ident_t* [[DEF_LOC]], i32 4, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i[[SZ]], i[[SZ]], i[[SZ]], [10 x i32]*)* [[OMP_OUTLINED6:@.+]] to void (i32*, i32*, ...)*), i[[SZ]] [[REF_A]], i[[SZ]] [[REF_AA]], i[[SZ]] [[REF_AAA]], [10 x i32]* [[REF_B]]) 774 // 775 // 776 // CHECK: define internal {{.*}}void [[OMP_OUTLINED6]](i32* noalias %.global_tid., i32* noalias %.bound_tid., i[[SZ]] %{{.+}}, i[[SZ]] %{{.+}}, i[[SZ]] %{{.+}}, [10 x i32]* {{.+}}) 777 // To reduce complexity, we're only going as far as validating the signature of the outlined parallel function. 778 779 // CHECK: define internal void [[HVT5]] 780 // Create local storage for each capture. 781 // CHECK: [[LOCAL_A:%.+]] = alloca i[[SZ]] 782 // CHECK: [[LOCAL_AA:%.+]] = alloca i[[SZ]] 783 // CHECK: [[LOCAL_B:%.+]] = alloca [10 x i32]* 784 // CHECK: [[LOCAL_A_CASTED:%.+]] = alloca i[[SZ]] 785 // CHECK: [[LOCAL_AA_CASTED:%.+]] = alloca i[[SZ]] 786 // CHECK-DAG: store i[[SZ]] [[ARG_A:%.+]], i[[SZ]]* [[LOCAL_A]] 787 // CHECK-DAG: store i[[SZ]] [[ARG_AA:%.+]], i[[SZ]]* [[LOCAL_AA]] 788 // CHECK-DAG: store [10 x i32]* [[ARG_B:%.+]], [10 x i32]** [[LOCAL_B]] 789 // Store captures in the context. 790 // CHECK-64-DAG:[[CONV_AP:%.+]] = bitcast i[[SZ]]* [[LOCAL_A]] to i32* 791 // CHECK-DAG: [[CONV_AAP:%.+]] = bitcast i[[SZ]]* [[LOCAL_AA]] to i16* 792 // CHECK-DAG: [[REF_B:%.+]] = load [10 x i32]*, [10 x i32]** [[LOCAL_B]], 793 794 // CHECK-64-DAG:[[CONV_A:%.+]] = load i32, i32* [[CONV_AP]] 795 // CHECK-64-DAG:[[CONV:%.+]] = bitcast i[[SZ]]* [[LOCAL_A_CASTED]] to i32* 796 // CHECK-64-DAG:store i32 [[CONV_A]], i32* [[CONV]], align 797 // CHECK-32-DAG:[[LOCAL_AV:%.+]] = load i32, i32* [[LOCAL_A]] 798 // CHECK-32-DAG:store i32 [[LOCAL_AV]], i32* [[LOCAL_A_CASTED]], align 799 // CHECK-DAG: [[REF_A:%.+]] = load i[[SZ]], i[[SZ]]* [[LOCAL_A_CASTED]], 800 801 // CHECK-DAG: [[CONV_AA:%.+]] = load i16, i16* [[CONV_AAP]] 802 // CHECK-DAG: [[CONV:%.+]] = bitcast i[[SZ]]* [[LOCAL_AA_CASTED]] to i16* 803 // CHECK-DAG: store i16 [[CONV_AA]], i16* [[CONV]], align 804 // CHECK-DAG: [[REF_AA:%.+]] = load i[[SZ]], i[[SZ]]* [[LOCAL_AA_CASTED]], 805 806 // CHECK: call {{.*}}void (%ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%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]]) 807 // 808 // 809 // CHECK: define internal {{.*}}void [[OMP_OUTLINED7]](i32* noalias %.global_tid., i32* noalias %.bound_tid., i[[SZ]] %{{.+}}, i[[SZ]] %{{.+}}, [10 x i32]* {{.+}}) 810 // To reduce complexity, we're only going as far as validating the signature of the outlined parallel function. 811 812 #endif 813