1 // RUN: %clang_cc1 -verify -fopenmp -x c++ -triple x86_64-apple-darwin10 -emit-llvm %s -o - | FileCheck %s 2 // RUN: %clang_cc1 -fopenmp -x c++ -std=c++11 -triple x86_64-apple-darwin10 -emit-pch -o %t %s 3 // RUN: %clang_cc1 -fopenmp -x c++ -triple x86_64-apple-darwin10 -std=c++11 -include-pch %t -verify %s -emit-llvm -o - | FileCheck %s 4 // RUN: %clang_cc1 -verify -fopenmp -x c++ -std=c++11 -DLAMBDA -triple x86_64-apple-darwin10 -emit-llvm %s -o - | FileCheck -check-prefix=LAMBDA %s 5 // RUN: %clang_cc1 -verify -fopenmp -x c++ -fblocks -DBLOCKS -triple x86_64-apple-darwin10 -emit-llvm %s -o - | FileCheck -check-prefix=BLOCKS %s 6 7 // RUN: %clang_cc1 -verify -fopenmp-simd -x c++ -triple x86_64-apple-darwin10 -emit-llvm %s -o - | FileCheck --check-prefix SIMD-ONLY0 %s 8 // RUN: %clang_cc1 -fopenmp-simd -x c++ -std=c++11 -triple x86_64-apple-darwin10 -emit-pch -o %t %s 9 // RUN: %clang_cc1 -fopenmp-simd -x c++ -triple x86_64-apple-darwin10 -std=c++11 -include-pch %t -verify %s -emit-llvm -o - | FileCheck --check-prefix SIMD-ONLY0 %s 10 // RUN: %clang_cc1 -verify -fopenmp-simd -x c++ -std=c++11 -DLAMBDA -triple x86_64-apple-darwin10 -emit-llvm %s -o - | FileCheck --check-prefix SIMD-ONLY0 %s 11 // RUN: %clang_cc1 -verify -fopenmp-simd -x c++ -fblocks -DBLOCKS -triple x86_64-apple-darwin10 -emit-llvm %s -o - | FileCheck --check-prefix SIMD-ONLY0 %s 12 // SIMD-ONLY0-NOT: {{__kmpc|__tgt}} 13 // expected-no-diagnostics 14 #ifndef HEADER 15 #define HEADER 16 17 template <class T> 18 struct S { 19 T f; 20 S(T a) : f(a) {} 21 S() : f() {} 22 S<T> &operator=(const S<T> &); 23 operator T() { return T(); } 24 ~S() {} 25 }; 26 27 volatile int g = 1212; 28 float f; 29 char cnt; 30 31 // CHECK: [[S_FLOAT_TY:%.+]] = type { float } 32 // CHECK: [[S_INT_TY:%.+]] = type { i32 } 33 // CHECK-DAG: [[F:@.+]] = global float 0.0 34 // CHECK-DAG: [[CNT:@.+]] = global i8 0 35 template <typename T> 36 T tmain() { 37 S<T> test; 38 T *pvar = &test.f; 39 T lvar = T(); 40 #pragma omp parallel for linear(pvar, lvar) 41 for (int i = 0; i < 2; ++i) { 42 ++pvar, ++lvar; 43 } 44 return T(); 45 } 46 47 int main() { 48 #ifdef LAMBDA 49 // LAMBDA: [[G:@.+]] = global i{{[0-9]+}} 1212, 50 // LAMBDA-LABEL: @main 51 // LAMBDA: call void [[OUTER_LAMBDA:@.+]]( 52 [&]() { 53 // LAMBDA: define{{.*}} internal{{.*}} void [[OUTER_LAMBDA]]( 54 // LAMBDA: call void {{.+}} @__kmpc_fork_call({{.+}}, i32 1, {{.+}}* [[OMP_REGION:@.+]] to {{.+}}, i32* [[G]]) 55 #pragma omp parallel for linear(g:5) 56 for (int i = 0; i < 2; ++i) { 57 // LAMBDA: define{{.*}} internal{{.*}} void [[OMP_REGION]](i32* noalias %{{.+}}, i32* noalias %{{.+}}, i32* nonnull align 4 dereferenceable(4) %{{.+}}) 58 // LAMBDA: alloca i{{[0-9]+}}, 59 // LAMBDA: alloca i{{[0-9]+}}, 60 // LAMBDA: [[G_START_ADDR:%.+]] = alloca i{{[0-9]+}}, 61 // LAMBDA: alloca i{{[0-9]+}}, 62 // LAMBDA: alloca i{{[0-9]+}}, 63 // LAMBDA: alloca i{{[0-9]+}}, 64 // LAMBDA: alloca i{{[0-9]+}}, 65 // LAMBDA: alloca i{{[0-9]+}}, 66 // LAMBDA: [[G_PRIVATE_ADDR:%.+]] = alloca i{{[0-9]+}}, 67 // LAMBDA: store i32 0, 68 // LAMBDA: [[GTID_REF:%.+]] = load i{{[0-9]+}}*, i{{[0-9]+}}** %{{.+}} 69 // LAMBDA: [[GTID:%.+]] = load i{{[0-9]+}}, i{{[0-9]+}}* [[GTID_REF]] 70 // LAMBDA: call {{.+}} @__kmpc_for_static_init_4(%{{.+}}* @{{.+}}, i32 [[GTID]], i32 34, i32* [[IS_LAST_ADDR:%.+]], i32* %{{.+}}, i32* %{{.+}}, i32* %{{.+}}, i32 1, i32 1) 71 // LAMBDA: [[VAL:%.+]] = load i32, i32* [[G_START_ADDR]] 72 // LAMBDA: [[CNT:%.+]] = load i32, i32* 73 // LAMBDA: [[MUL:%.+]] = mul nsw i32 [[CNT]], 5 74 // LAMBDA: [[ADD:%.+]] = add nsw i32 [[VAL]], [[MUL]] 75 // LAMBDA: store i32 [[ADD]], i32* [[G_PRIVATE_ADDR]], 76 // LAMBDA: [[VAL:%.+]] = load i32, i32* [[G_PRIVATE_ADDR]], 77 // LAMBDA: [[ADD:%.+]] = add nsw i32 [[VAL]], 5 78 // LAMBDA: store i32 [[ADD]], i32* [[G_PRIVATE_ADDR]], 79 // LAMBDA: [[G_PRIVATE_ADDR_REF:%.+]] = getelementptr inbounds %{{.+}}, %{{.+}}* [[ARG:%.+]], i{{[0-9]+}} 0, i{{[0-9]+}} 0 80 // LAMBDA: store i{{[0-9]+}}* [[G_PRIVATE_ADDR]], i{{[0-9]+}}** [[G_PRIVATE_ADDR_REF]] 81 // LAMBDA: call void [[INNER_LAMBDA:@.+]](%{{.+}}* [[ARG]]) 82 // LAMBDA: call void @__kmpc_for_static_fini(%{{.+}}* @{{.+}}, i32 [[GTID]]) 83 g += 5; 84 [&]() { 85 // LAMBDA: define {{.+}} void [[INNER_LAMBDA]](%{{.+}}* [[ARG_PTR:%.+]]) 86 // LAMBDA: store %{{.+}}* [[ARG_PTR]], %{{.+}}** [[ARG_PTR_REF:%.+]], 87 g = 2; 88 // LAMBDA: [[ARG_PTR:%.+]] = load %{{.+}}*, %{{.+}}** [[ARG_PTR_REF]] 89 // LAMBDA: [[G_PTR_REF:%.+]] = getelementptr inbounds %{{.+}}, %{{.+}}* [[ARG_PTR]], i{{[0-9]+}} 0, i{{[0-9]+}} 0 90 // LAMBDA: [[G_REF:%.+]] = load i{{[0-9]+}}*, i{{[0-9]+}}** [[G_PTR_REF]] 91 // LAMBDA: store i{{[0-9]+}} 2, i{{[0-9]+}}* [[G_REF]] 92 }(); 93 } 94 }(); 95 return 0; 96 #elif defined(BLOCKS) 97 // BLOCKS: [[G:@.+]] = global i{{[0-9]+}} 1212, 98 // BLOCKS-LABEL: @main 99 // BLOCKS: call void {{%.+}}(i8 100 ^{ 101 // BLOCKS: define{{.*}} internal{{.*}} void {{.+}}(i8* 102 // BLOCKS: call void {{.+}} @__kmpc_fork_call({{.+}}, i32 1, {{.+}}* [[OMP_REGION:@.+]] to {{.+}}, i32* [[G]]) 103 #pragma omp parallel for linear(g:5) 104 for (int i = 0; i < 2; ++i) { 105 // BLOCKS: define{{.*}} internal{{.*}} void [[OMP_REGION]](i32* noalias %{{.+}}, i32* noalias %{{.+}}, i32* nonnull align 4 dereferenceable(4) %{{.+}}) 106 // BLOCKS: alloca i{{[0-9]+}}, 107 // BLOCKS: alloca i{{[0-9]+}}, 108 // BLOCKS: [[G_START_ADDR:%.+]] = alloca i{{[0-9]+}}, 109 // BLOCKS: alloca i{{[0-9]+}}, 110 // BLOCKS: alloca i{{[0-9]+}}, 111 // BLOCKS: alloca i{{[0-9]+}}, 112 // BLOCKS: alloca i{{[0-9]+}}, 113 // BLOCKS: alloca i{{[0-9]+}}, 114 // BLOCKS: [[G_PRIVATE_ADDR:%.+]] = alloca i{{[0-9]+}}, 115 // BLOCKS: store i32 0, 116 // BLOCKS: [[GTID_REF:%.+]] = load i{{[0-9]+}}*, i{{[0-9]+}}** %{{.+}} 117 // BLOCKS: [[GTID:%.+]] = load i{{[0-9]+}}, i{{[0-9]+}}* [[GTID_REF]] 118 // BLOCKS: call {{.+}} @__kmpc_for_static_init_4(%{{.+}}* @{{.+}}, i32 [[GTID]], i32 34, i32* [[IS_LAST_ADDR:%.+]], i32* %{{.+}}, i32* %{{.+}}, i32* %{{.+}}, i32 1, i32 1) 119 // BLOCKS: [[VAL:%.+]] = load i32, i32* [[G_START_ADDR]] 120 // BLOCKS: [[CNT:%.+]] = load i32, i32* 121 // BLOCKS: [[MUL:%.+]] = mul nsw i32 [[CNT]], 5 122 // BLOCKS: [[ADD:%.+]] = add nsw i32 [[VAL]], [[MUL]] 123 // BLOCKS: store i32 [[ADD]], i32* [[G_PRIVATE_ADDR]], 124 // BLOCKS: [[VAL:%.+]] = load i32, i32* [[G_PRIVATE_ADDR]], 125 // BLOCKS: [[ADD:%.+]] = add nsw i32 [[VAL]], 5 126 // BLOCKS: store i32 [[ADD]], i32* [[G_PRIVATE_ADDR]], 127 // BLOCKS-NOT: [[G]]{{[[^:word:]]}} 128 // BLOCKS: i{{[0-9]+}}* [[G_PRIVATE_ADDR]] 129 // BLOCKS-NOT: [[G]]{{[[^:word:]]}} 130 // BLOCKS: call void {{%.+}}(i8 131 // BLOCKS: call void @__kmpc_for_static_fini(%{{.+}}* @{{.+}}, i32 [[GTID]]) 132 g += 5; 133 g = 1; 134 ^{ 135 // BLOCKS: define {{.+}} void {{@.+}}(i8* 136 g = 2; 137 // BLOCKS-NOT: [[G]]{{[[^:word:]]}} 138 // BLOCKS: store i{{[0-9]+}} 2, i{{[0-9]+}}* 139 // BLOCKS-NOT: [[G]]{{[[^:word:]]}} 140 // BLOCKS: ret 141 }(); 142 } 143 }(); 144 return 0; 145 #else 146 S<float> test; 147 float *pvar = &test.f; 148 long long lvar = 0; 149 #pragma omp parallel for linear(pvar, lvar : 3) 150 for (int i = 0; i < 2; ++i) { 151 pvar += 3, lvar += 3; 152 } 153 return tmain<int>(); 154 #endif 155 } 156 157 // CHECK: define i{{[0-9]+}} @main() 158 // CHECK: [[TEST:%.+]] = alloca [[S_FLOAT_TY]], 159 // CHECK: call {{.*}} [[S_FLOAT_TY_DEF_CONSTR:@.+]]([[S_FLOAT_TY]]* [[TEST]]) 160 // CHECK: call void (%{{.+}}*, i{{[0-9]+}}, void (i{{[0-9]+}}*, i{{[0-9]+}}*, ...)*, ...) @__kmpc_fork_call(%{{.+}}* @{{.+}}, i{{[0-9]+}} 2, void (i{{[0-9]+}}*, i{{[0-9]+}}*, ...)* bitcast (void (i{{[0-9]+}}*, i{{[0-9]+}}*, float**, i64*)* [[MAIN_MICROTASK:@.+]] to void 161 // CHECK: = call {{.+}} [[TMAIN_INT:@.+]]() 162 // CHECK: call void [[S_FLOAT_TY_DESTR:@.+]]([[S_FLOAT_TY]]* 163 // CHECK: ret 164 165 // CHECK: define internal void [[MAIN_MICROTASK]](i{{[0-9]+}}* noalias [[GTID_ADDR:%.+]], i{{[0-9]+}}* noalias %{{.+}}, float** nonnull align 8 dereferenceable(8) %{{.+}}, i64* nonnull align 8 dereferenceable(8) %{{.+}}) 166 // CHECK: alloca i{{[0-9]+}}, 167 // CHECK: alloca i{{[0-9]+}}, 168 // CHECK: [[PVAR_START:%.+]] = alloca float*, 169 // CHECK: [[LVAR_START:%.+]] = alloca i64, 170 // CHECK: alloca i{{[0-9]+}}, 171 // CHECK: alloca i{{[0-9]+}}, 172 // CHECK: alloca i{{[0-9]+}}, 173 // CHECK: alloca i{{[0-9]+}}, 174 // CHECK: [[PVAR_PRIV:%.+]] = alloca float*, 175 // CHECK: [[LVAR_PRIV:%.+]] = alloca i64, 176 // CHECK: store i{{[0-9]+}}* [[GTID_ADDR]], i{{[0-9]+}}** [[GTID_ADDR_REF:%.+]] 177 178 // CHECK: [[PVAR_REF:%.+]] = load float**, float*** % 179 // CHECK: [[LVAR_REF:%.+]] = load i64*, i64** % 180 181 // Check for default initialization. 182 // CHECK: [[PVAR_VAL:%.+]] = load float*, float** [[PVAR_REF]], 183 // CHECK: store float* [[PVAR_VAL]], float** [[PVAR_START]], 184 // CHECK: [[LVAR_VAL:%.+]] = load i64, i64* [[LVAR_REF]], 185 // CHECK: store i64 [[LVAR_VAL]], i64* [[LVAR_START]], 186 // CHECK: call {{.+}} @__kmpc_for_static_init_4(%{{.+}}* @{{.+}}, i32 [[GTID:%.+]], i32 34, i32* [[IS_LAST_ADDR:%.+]], i32* %{{.+}}, i32* %{{.+}}, i32* %{{.+}}, i32 1, i32 1) 187 // CHECK: [[PVAR_VAL:%.+]] = load float*, float** [[PVAR_START]], 188 // CHECK: [[CNT:%.+]] = load i32, i32* 189 // CHECK: [[MUL:%.+]] = mul nsw i32 [[CNT]], 3 190 // CHECK: [[IDX:%.+]] = sext i32 [[MUL]] to i64 191 // CHECK: [[PTR:%.+]] = getelementptr inbounds float, float* [[PVAR_VAL]], i64 [[IDX]] 192 // CHECK: store float* [[PTR]], float** [[PVAR_PRIV]], 193 // CHECK: [[LVAR_VAL:%.+]] = load i64, i64* [[LVAR_START]], 194 // CHECK: [[CNT:%.+]] = load i32, i32* 195 // CHECK: [[MUL:%.+]] = mul nsw i32 [[CNT]], 3 196 // CHECK: [[CONV:%.+]] = sext i32 [[MUL]] to i64 197 // CHECK: [[VAL:%.+]] = add nsw i64 [[LVAR_VAL]], [[CONV]] 198 // CHECK: store i64 [[VAL]], i64* [[LVAR_PRIV]], 199 // CHECK: [[PVAR_VAL:%.+]] = load float*, float** [[PVAR_PRIV]] 200 // CHECK: [[PTR:%.+]] = getelementptr inbounds float, float* [[PVAR_VAL]], i64 3 201 // CHECK: store float* [[PTR]], float** [[PVAR_PRIV]], 202 // CHECK: [[LVAR_VAL:%.+]] = load i64, i64* [[LVAR_PRIV]], 203 // CHECK: [[ADD:%.+]] = add nsw i64 [[LVAR_VAL]], 3 204 // CHECK: store i64 [[ADD]], i64* [[LVAR_PRIV]], 205 // CHECK: call void @__kmpc_for_static_fini(%{{.+}}* @{{.+}}, i32 %{{.+}}) 206 // CHECK: ret void 207 208 // CHECK: define {{.*}} i{{[0-9]+}} [[TMAIN_INT]]() 209 // CHECK: [[TEST:%.+]] = alloca [[S_INT_TY]], 210 // CHECK: call {{.*}} [[S_INT_TY_DEF_CONSTR:@.+]]([[S_INT_TY]]* [[TEST]]) 211 // CHECK: call void (%{{.+}}*, i{{[0-9]+}}, void (i{{[0-9]+}}*, i{{[0-9]+}}*, ...)*, ...) @__kmpc_fork_call(%{{.+}}* @{{.+}}, i{{[0-9]+}} 2, void (i{{[0-9]+}}*, i{{[0-9]+}}*, ...)* bitcast (void (i{{[0-9]+}}*, i{{[0-9]+}}*, i32**, i32*)* [[TMAIN_MICROTASK:@.+]] to void 212 // CHECK: call void [[S_INT_TY_DESTR:@.+]]([[S_INT_TY]]* 213 // CHECK: ret 214 // 215 // CHECK: define internal void [[TMAIN_MICROTASK]](i{{[0-9]+}}* noalias [[GTID_ADDR:%.+]], i{{[0-9]+}}* noalias %{{.+}}, i32** nonnull align 8 dereferenceable(8) %{{.+}}, i32* nonnull align 4 dereferenceable(4) %{{.+}}) 216 // CHECK: alloca i{{[0-9]+}}, 217 // CHECK: alloca i{{[0-9]+}}, 218 // CHECK: [[PVAR_START:%.+]] = alloca i32*, 219 // CHECK: [[LVAR_START:%.+]] = alloca i32, 220 // CHECK: alloca i{{[0-9]+}}, 221 // CHECK: alloca i{{[0-9]+}}, 222 // CHECK: alloca i{{[0-9]+}}, 223 // CHECK: alloca i{{[0-9]+}}, 224 // CHECK: [[PVAR_PRIV:%.+]] = alloca i32*, 225 // CHECK: [[LVAR_PRIV:%.+]] = alloca i32, 226 // CHECK: store i{{[0-9]+}}* [[GTID_ADDR]], i{{[0-9]+}}** [[GTID_ADDR_REF:%.+]] 227 228 // CHECK: [[PVAR_REF:%.+]] = load i32**, i32*** % 229 // CHECK: [[LVAR_REF:%.+]] = load i32*, i32** % 230 231 // Check for default initialization. 232 // CHECK: [[PVAR_VAL:%.+]] = load i32*, i32** [[PVAR_REF]], 233 // CHECK: store i32* [[PVAR_VAL]], i32** [[PVAR_START]], 234 // CHECK: [[LVAR_VAL:%.+]] = load i32, i32* [[LVAR_REF]], 235 // CHECK: store i32 [[LVAR_VAL]], i32* [[LVAR_START]], 236 // CHECK: call {{.+}} @__kmpc_for_static_init_4(%{{.+}}* @{{.+}}, i32 [[GTID:%.+]], i32 34, i32* [[IS_LAST_ADDR:%.+]], i32* %{{.+}}, i32* %{{.+}}, i32* %{{.+}}, i32 1, i32 1) 237 // CHECK: [[PVAR_VAL:%.+]] = load i32*, i32** [[PVAR_START]], 238 // CHECK: [[CNT:%.+]] = load i32, i32* 239 // CHECK: [[MUL:%.+]] = mul nsw i32 [[CNT]], 1 240 // CHECK: [[IDX:%.+]] = sext i32 [[MUL]] to i64 241 // CHECK: [[PTR:%.+]] = getelementptr inbounds i32, i32* [[PVAR_VAL]], i64 [[IDX]] 242 // CHECK: store i32* [[PTR]], i32** [[PVAR_PRIV]], 243 // CHECK: [[LVAR_VAL:%.+]] = load i32, i32* [[LVAR_START]], 244 // CHECK: [[CNT:%.+]] = load i32, i32* 245 // CHECK: [[MUL:%.+]] = mul nsw i32 [[CNT]], 1 246 // CHECK: [[VAL:%.+]] = add nsw i32 [[LVAR_VAL]], [[MUL]] 247 // CHECK: store i32 [[VAL]], i32* [[LVAR_PRIV]], 248 // CHECK: [[PVAR_VAL:%.+]] = load i32*, i32** [[PVAR_PRIV]] 249 // CHECK: [[PTR:%.+]] = getelementptr inbounds i32, i32* [[PVAR_VAL]], i32 1 250 // CHECK: store i32* [[PTR]], i32** [[PVAR_PRIV]], 251 // CHECK: [[LVAR_VAL:%.+]] = load i32, i32* [[LVAR_PRIV]], 252 // CHECK: [[ADD:%.+]] = add nsw i32 [[LVAR_VAL]], 1 253 // CHECK: store i32 [[ADD]], i32* [[LVAR_PRIV]], 254 // CHECK: call void @__kmpc_for_static_fini(%{{.+}}* @{{.+}}, i32 %{{.+}}) 255 // CHECK: ret void 256 #endif 257 258