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 // REQUIRES: x86-registered-target 8 #ifndef HEADER 9 #define HEADER 10 11 template <class T> 12 struct S { 13 T f; 14 S(T a) : f(a) {} 15 S() : f() {} 16 S<T> &operator=(const S<T> &); 17 operator T() { return T(); } 18 ~S() {} 19 }; 20 21 volatile int g = 1212; 22 float f; 23 char cnt; 24 25 // CHECK: [[S_FLOAT_TY:%.+]] = type { float } 26 // CHECK: [[S_INT_TY:%.+]] = type { i32 } 27 // CHECK-DAG: [[F:@.+]] = global float 0.0 28 // CHECK-DAG: [[CNT:@.+]] = global i8 0 29 template <typename T> 30 T tmain() { 31 S<T> test; 32 T *pvar = &test.f; 33 T lvar = T(); 34 #pragma omp parallel for linear(pvar, lvar) 35 for (int i = 0; i < 2; ++i) { 36 ++pvar, ++lvar; 37 } 38 return T(); 39 } 40 41 int main() { 42 #ifdef LAMBDA 43 // LAMBDA: [[G:@.+]] = global i{{[0-9]+}} 1212, 44 // LAMBDA-LABEL: @main 45 // LAMBDA: call void [[OUTER_LAMBDA:@.+]]( 46 [&]() { 47 // LAMBDA: define{{.*}} internal{{.*}} void [[OUTER_LAMBDA]]( 48 // LAMBDA: call void {{.+}} @__kmpc_fork_call({{.+}}, i32 1, {{.+}}* [[OMP_REGION:@.+]] to {{.+}}, i32* [[G]]) 49 #pragma omp parallel for linear(g:5) 50 for (int i = 0; i < 2; ++i) { 51 // LAMBDA: define{{.*}} internal{{.*}} void [[OMP_REGION]](i32* noalias %{{.+}}, i32* noalias %{{.+}}, i32* dereferenceable(4) %{{.+}}) 52 // LAMBDA: alloca i{{[0-9]+}}, 53 // LAMBDA: [[G_START_ADDR:%.+]] = 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: alloca i{{[0-9]+}}, 59 // LAMBDA: [[G_PRIVATE_ADDR:%.+]] = alloca i{{[0-9]+}}, 60 // LAMBDA: store i32 0, 61 // LAMBDA: [[GTID_REF:%.+]] = load i{{[0-9]+}}*, i{{[0-9]+}}** %{{.+}} 62 // LAMBDA: [[GTID:%.+]] = load i{{[0-9]+}}, i{{[0-9]+}}* [[GTID_REF]] 63 // LAMBDA: call {{.+}} @__kmpc_for_static_init_4(%{{.+}}* @{{.+}}, i32 [[GTID]], i32 34, i32* [[IS_LAST_ADDR:%.+]], i32* %{{.+}}, i32* %{{.+}}, i32* %{{.+}}, i32 1, i32 1) 64 // LAMBDA: [[VAL:%.+]] = load i32, i32* [[G_START_ADDR]] 65 // LAMBDA: [[CNT:%.+]] = load i32, i32* 66 // LAMBDA: [[MUL:%.+]] = mul nsw i32 [[CNT]], 5 67 // LAMBDA: [[ADD:%.+]] = add nsw i32 [[VAL]], [[MUL]] 68 // LAMBDA: store i32 [[ADD]], i32* [[G_PRIVATE_ADDR]], 69 // LAMBDA: [[VAL:%.+]] = load i32, i32* [[G_PRIVATE_ADDR]], 70 // LAMBDA: [[ADD:%.+]] = add nsw i32 [[VAL]], 5 71 // LAMBDA: store i32 [[ADD]], i32* [[G_PRIVATE_ADDR]], 72 // LAMBDA: [[G_PRIVATE_ADDR_REF:%.+]] = getelementptr inbounds %{{.+}}, %{{.+}}* [[ARG:%.+]], i{{[0-9]+}} 0, i{{[0-9]+}} 0 73 // LAMBDA: store i{{[0-9]+}}* [[G_PRIVATE_ADDR]], i{{[0-9]+}}** [[G_PRIVATE_ADDR_REF]] 74 // LAMBDA: call void [[INNER_LAMBDA:@.+]](%{{.+}}* [[ARG]]) 75 // LAMBDA: call void @__kmpc_for_static_fini(%{{.+}}* @{{.+}}, i32 [[GTID]]) 76 g += 5; 77 [&]() { 78 // LAMBDA: define {{.+}} void [[INNER_LAMBDA]](%{{.+}}* [[ARG_PTR:%.+]]) 79 // LAMBDA: store %{{.+}}* [[ARG_PTR]], %{{.+}}** [[ARG_PTR_REF:%.+]], 80 g = 2; 81 // LAMBDA: [[ARG_PTR:%.+]] = load %{{.+}}*, %{{.+}}** [[ARG_PTR_REF]] 82 // LAMBDA: [[G_PTR_REF:%.+]] = getelementptr inbounds %{{.+}}, %{{.+}}* [[ARG_PTR]], i{{[0-9]+}} 0, i{{[0-9]+}} 0 83 // LAMBDA: [[G_REF:%.+]] = load i{{[0-9]+}}*, i{{[0-9]+}}** [[G_PTR_REF]] 84 // LAMBDA: store i{{[0-9]+}} 2, i{{[0-9]+}}* [[G_REF]] 85 }(); 86 } 87 }(); 88 return 0; 89 #elif defined(BLOCKS) 90 // BLOCKS: [[G:@.+]] = global i{{[0-9]+}} 1212, 91 // BLOCKS-LABEL: @main 92 // BLOCKS: call void {{%.+}}(i8 93 ^{ 94 // BLOCKS: define{{.*}} internal{{.*}} void {{.+}}(i8* 95 // BLOCKS: call void {{.+}} @__kmpc_fork_call({{.+}}, i32 1, {{.+}}* [[OMP_REGION:@.+]] to {{.+}}, i32* [[G]]) 96 #pragma omp parallel for linear(g:5) 97 for (int i = 0; i < 2; ++i) { 98 // BLOCKS: define{{.*}} internal{{.*}} void [[OMP_REGION]](i32* noalias %{{.+}}, i32* noalias %{{.+}}, i32* dereferenceable(4) %{{.+}}) 99 // BLOCKS: alloca i{{[0-9]+}}, 100 // BLOCKS: [[G_START_ADDR:%.+]] = 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: alloca i{{[0-9]+}}, 106 // BLOCKS: [[G_PRIVATE_ADDR:%.+]] = alloca i{{[0-9]+}}, 107 // BLOCKS: store i32 0, 108 // BLOCKS: [[GTID_REF:%.+]] = load i{{[0-9]+}}*, i{{[0-9]+}}** %{{.+}} 109 // BLOCKS: [[GTID:%.+]] = load i{{[0-9]+}}, i{{[0-9]+}}* [[GTID_REF]] 110 // BLOCKS: call {{.+}} @__kmpc_for_static_init_4(%{{.+}}* @{{.+}}, i32 [[GTID]], i32 34, i32* [[IS_LAST_ADDR:%.+]], i32* %{{.+}}, i32* %{{.+}}, i32* %{{.+}}, i32 1, i32 1) 111 // BLOCKS: [[VAL:%.+]] = load i32, i32* [[G_START_ADDR]] 112 // BLOCKS: [[CNT:%.+]] = load i32, i32* 113 // BLOCKS: [[MUL:%.+]] = mul nsw i32 [[CNT]], 5 114 // BLOCKS: [[ADD:%.+]] = add nsw i32 [[VAL]], [[MUL]] 115 // BLOCKS: store i32 [[ADD]], i32* [[G_PRIVATE_ADDR]], 116 // BLOCKS: [[VAL:%.+]] = load i32, i32* [[G_PRIVATE_ADDR]], 117 // BLOCKS: [[ADD:%.+]] = add nsw i32 [[VAL]], 5 118 // BLOCKS: store i32 [[ADD]], i32* [[G_PRIVATE_ADDR]], 119 // BLOCKS-NOT: [[G]]{{[[^:word:]]}} 120 // BLOCKS: i{{[0-9]+}}* [[G_PRIVATE_ADDR]] 121 // BLOCKS-NOT: [[G]]{{[[^:word:]]}} 122 // BLOCKS: call void {{%.+}}(i8 123 // BLOCKS: call void @__kmpc_for_static_fini(%{{.+}}* @{{.+}}, i32 [[GTID]]) 124 g += 5; 125 g = 1; 126 ^{ 127 // BLOCKS: define {{.+}} void {{@.+}}(i8* 128 g = 2; 129 // BLOCKS-NOT: [[G]]{{[[^:word:]]}} 130 // BLOCKS: store i{{[0-9]+}} 2, i{{[0-9]+}}* 131 // BLOCKS-NOT: [[G]]{{[[^:word:]]}} 132 // BLOCKS: ret 133 }(); 134 } 135 }(); 136 return 0; 137 #else 138 S<float> test; 139 float *pvar = &test.f; 140 long long lvar = 0; 141 #pragma omp parallel for linear(pvar, lvar : 3) 142 for (int i = 0; i < 2; ++i) { 143 pvar += 3, lvar += 3; 144 } 145 return tmain<int>(); 146 #endif 147 } 148 149 // CHECK: define i{{[0-9]+}} @main() 150 // CHECK: [[TEST:%.+]] = alloca [[S_FLOAT_TY]], 151 // CHECK: call {{.*}} [[S_FLOAT_TY_DEF_CONSTR:@.+]]([[S_FLOAT_TY]]* [[TEST]]) 152 // 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 153 // CHECK: = call {{.+}} [[TMAIN_INT:@.+]]() 154 // CHECK: call void [[S_FLOAT_TY_DESTR:@.+]]([[S_FLOAT_TY]]* 155 // CHECK: ret 156 157 // CHECK: define internal void [[MAIN_MICROTASK]](i{{[0-9]+}}* noalias [[GTID_ADDR:%.+]], i{{[0-9]+}}* noalias %{{.+}}, float** dereferenceable(8) %{{.+}}, i64* dereferenceable(8) %{{.+}}) 158 // CHECK: alloca i{{[0-9]+}}, 159 // CHECK: [[PVAR_START:%.+]] = alloca float*, 160 // CHECK: [[LVAR_START:%.+]] = alloca i64, 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: alloca i{{[0-9]+}}, 166 // CHECK: [[PVAR_PRIV:%.+]] = alloca float*, 167 // CHECK: [[LVAR_PRIV:%.+]] = alloca i64, 168 // CHECK: store i{{[0-9]+}}* [[GTID_ADDR]], i{{[0-9]+}}** [[GTID_ADDR_REF:%.+]] 169 170 // CHECK: [[PVAR_REF:%.+]] = load float**, float*** % 171 // CHECK: [[LVAR_REF:%.+]] = load i64*, i64** % 172 173 // Check for default initialization. 174 // CHECK: [[PVAR_VAL:%.+]] = load float*, float** [[PVAR_REF]], 175 // CHECK: store float* [[PVAR_VAL]], float** [[PVAR_START]], 176 // CHECK: [[LVAR_VAL:%.+]] = load i64, i64* [[LVAR_REF]], 177 // CHECK: store i64 [[LVAR_VAL]], i64* [[LVAR_START]], 178 // CHECK: call {{.+}} @__kmpc_for_static_init_4(%{{.+}}* @{{.+}}, i32 [[GTID:%.+]], i32 34, i32* [[IS_LAST_ADDR:%.+]], i32* %{{.+}}, i32* %{{.+}}, i32* %{{.+}}, i32 1, i32 1) 179 // CHECK: [[PVAR_VAL:%.+]] = load float*, float** [[PVAR_START]], 180 // CHECK: [[CNT:%.+]] = load i32, i32* 181 // CHECK: [[MUL:%.+]] = mul nsw i32 [[CNT]], 3 182 // CHECK: [[IDX:%.+]] = sext i32 [[MUL]] to i64 183 // CHECK: [[PTR:%.+]] = getelementptr inbounds float, float* [[PVAR_VAL]], i64 [[IDX]] 184 // CHECK: store float* [[PTR]], float** [[PVAR_PRIV]], 185 // CHECK: [[LVAR_VAL:%.+]] = load i64, i64* [[LVAR_START]], 186 // CHECK: [[CNT:%.+]] = load i32, i32* 187 // CHECK: [[MUL:%.+]] = mul nsw i32 [[CNT]], 3 188 // CHECK: [[CONV:%.+]] = sext i32 [[MUL]] to i64 189 // CHECK: [[VAL:%.+]] = add nsw i64 [[LVAR_VAL]], [[CONV]] 190 // CHECK: store i64 [[VAL]], i64* [[LVAR_PRIV]], 191 // CHECK: [[PVAR_VAL:%.+]] = load float*, float** [[PVAR_PRIV]] 192 // CHECK: [[PTR:%.+]] = getelementptr inbounds float, float* [[PVAR_VAL]], i64 3 193 // CHECK: store float* [[PTR]], float** [[PVAR_PRIV]], 194 // CHECK: [[LVAR_VAL:%.+]] = load i64, i64* [[LVAR_PRIV]], 195 // CHECK: [[ADD:%.+]] = add nsw i64 [[LVAR_VAL]], 3 196 // CHECK: store i64 [[ADD]], i64* [[LVAR_PRIV]], 197 // CHECK: call void @__kmpc_for_static_fini(%{{.+}}* @{{.+}}, i32 %{{.+}}) 198 // CHECK: ret void 199 200 // CHECK: define {{.*}} i{{[0-9]+}} [[TMAIN_INT]]() 201 // CHECK: [[TEST:%.+]] = alloca [[S_INT_TY]], 202 // CHECK: call {{.*}} [[S_INT_TY_DEF_CONSTR:@.+]]([[S_INT_TY]]* [[TEST]]) 203 // 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 204 // CHECK: call void [[S_INT_TY_DESTR:@.+]]([[S_INT_TY]]* 205 // CHECK: ret 206 // 207 // CHECK: define internal void [[TMAIN_MICROTASK]](i{{[0-9]+}}* noalias [[GTID_ADDR:%.+]], i{{[0-9]+}}* noalias %{{.+}}, i32** dereferenceable(8) %{{.+}}, i32* dereferenceable(4) %{{.+}}) 208 // CHECK: alloca i{{[0-9]+}}, 209 // CHECK: [[PVAR_START:%.+]] = alloca i32*, 210 // CHECK: [[LVAR_START:%.+]] = alloca i32, 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: alloca i{{[0-9]+}}, 216 // CHECK: [[PVAR_PRIV:%.+]] = alloca i32*, 217 // CHECK: [[LVAR_PRIV:%.+]] = alloca i32, 218 // CHECK: store i{{[0-9]+}}* [[GTID_ADDR]], i{{[0-9]+}}** [[GTID_ADDR_REF:%.+]] 219 220 // CHECK: [[PVAR_REF:%.+]] = load i32**, i32*** % 221 // CHECK: [[LVAR_REF:%.+]] = load i32*, i32** % 222 223 // Check for default initialization. 224 // CHECK: [[PVAR_VAL:%.+]] = load i32*, i32** [[PVAR_REF]], 225 // CHECK: store i32* [[PVAR_VAL]], i32** [[PVAR_START]], 226 // CHECK: [[LVAR_VAL:%.+]] = load i32, i32* [[LVAR_REF]], 227 // CHECK: store i32 [[LVAR_VAL]], i32* [[LVAR_START]], 228 // CHECK: call {{.+}} @__kmpc_for_static_init_4(%{{.+}}* @{{.+}}, i32 [[GTID:%.+]], i32 34, i32* [[IS_LAST_ADDR:%.+]], i32* %{{.+}}, i32* %{{.+}}, i32* %{{.+}}, i32 1, i32 1) 229 // CHECK: [[PVAR_VAL:%.+]] = load i32*, i32** [[PVAR_START]], 230 // CHECK: [[CNT:%.+]] = load i32, i32* 231 // CHECK: [[MUL:%.+]] = mul nsw i32 [[CNT]], 1 232 // CHECK: [[IDX:%.+]] = sext i32 [[MUL]] to i64 233 // CHECK: [[PTR:%.+]] = getelementptr inbounds i32, i32* [[PVAR_VAL]], i64 [[IDX]] 234 // CHECK: store i32* [[PTR]], i32** [[PVAR_PRIV]], 235 // CHECK: [[LVAR_VAL:%.+]] = load i32, i32* [[LVAR_START]], 236 // CHECK: [[CNT:%.+]] = load i32, i32* 237 // CHECK: [[MUL:%.+]] = mul nsw i32 [[CNT]], 1 238 // CHECK: [[VAL:%.+]] = add nsw i32 [[LVAR_VAL]], [[MUL]] 239 // CHECK: store i32 [[VAL]], i32* [[LVAR_PRIV]], 240 // CHECK: [[PVAR_VAL:%.+]] = load i32*, i32** [[PVAR_PRIV]] 241 // CHECK: [[PTR:%.+]] = getelementptr inbounds i32, i32* [[PVAR_VAL]], i32 1 242 // CHECK: store i32* [[PTR]], i32** [[PVAR_PRIV]], 243 // CHECK: [[LVAR_VAL:%.+]] = load i32, i32* [[LVAR_PRIV]], 244 // CHECK: [[ADD:%.+]] = add nsw i32 [[LVAR_VAL]], 1 245 // CHECK: store i32 [[ADD]], i32* [[LVAR_PRIV]], 246 // CHECK: call void @__kmpc_for_static_fini(%{{.+}}* @{{.+}}, i32 %{{.+}}) 247 // CHECK: ret void 248 #endif 249 250