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