Home | History | Annotate | Download | only in OpenMP
      1 // RUN: %clang_cc1 -verify -fopenmp -x c++ -triple x86_64-unknown-unknown -emit-llvm %s -o - | FileCheck %s
      2 // RUN: %clang_cc1 -fopenmp -x c++ -std=c++11 -triple x86_64-unknown-unknown -emit-pch -o %t %s
      3 // RUN: %clang_cc1 -fopenmp -x c++ -triple x86_64-unknown-unknown -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 %itanium_abi_triple -emit-llvm %s -o - | FileCheck -check-prefix=LAMBDA %s
      5 // RUN: %clang_cc1 -verify -fopenmp -x c++ -fblocks -DBLOCKS -triple %itanium_abi_triple -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   operator T() { return T(); }
     17   ~S() {}
     18 };
     19 
     20 volatile double g;
     21 volatile double &g1 = g;
     22 
     23 // CHECK: [[S_FLOAT_TY:%.+]] = type { float }
     24 // CHECK: [[S_INT_TY:%.+]] = type { i{{[0-9]+}} }
     25 template <typename T>
     26 T tmain() {
     27   S<T> test;
     28   T t_var = T();
     29   T vec[] = {1, 2};
     30   S<T> s_arr[] = {1, 2};
     31   S<T> &var = test;
     32 #pragma omp parallel
     33 #pragma omp for private(t_var, vec, s_arr, s_arr, var, var)
     34   for (int i = 0; i < 2; ++i) {
     35     vec[i] = t_var;
     36     s_arr[i] = var;
     37   }
     38   return T();
     39 }
     40 
     41 int main() {
     42   static int svar;
     43 #ifdef LAMBDA
     44   // LAMBDA: [[G:@.+]] = global double
     45   // LAMBDA-LABEL: @main
     46   // LAMBDA: call{{.*}} void [[OUTER_LAMBDA:@.+]](
     47   [&]() {
     48   static float sfvar;
     49   // LAMBDA: define{{.*}} internal{{.*}} void [[OUTER_LAMBDA]](
     50   // LAMBDA: call {{.*}}void {{.+}} @__kmpc_fork_call({{.+}}, i32 0, {{.+}}* [[OMP_REGION:@.+]] to {{.+}})
     51 #pragma omp parallel
     52 #pragma omp for private(g, g1, svar, sfvar)
     53   for (int i = 0; i < 2; ++i) {
     54     // LAMBDA: define{{.*}} internal{{.*}} void [[OMP_REGION]](i32* noalias %{{.+}}, i32* noalias %{{.+}})
     55     // LAMBDA: [[G_PRIVATE_ADDR:%.+]] = alloca double,
     56     // LAMBDA: [[SVAR_PRIVATE_ADDR:%.+]] = alloca i{{[0-9]+}},
     57     // LAMBDA: [[SFVAR_PRIVATE_ADDR:%.+]] = alloca float,
     58     g = 1;
     59     g1 = 1;
     60     svar = 3;
     61     sfvar = 4.0;
     62     // LAMBDA: call {{.*}}void @__kmpc_for_static_init_4(
     63     // LAMBDA: store double 1.0{{.+}}, double* [[G_PRIVATE_ADDR]],
     64     // LAMBDA: store i{{[0-9]+}} 3, i{{[0-9]+}}* [[SVAR_PRIVATE_ADDR]],
     65     // LAMBDA: store float 4.0{{.+}}, float* [[SFVAR_PRIVATE_ADDR]],
     66     // LAMBDA: [[G_PRIVATE_ADDR_REF:%.+]] = getelementptr inbounds %{{.+}}, %{{.+}}* [[ARG:%.+]], i{{[0-9]+}} 0, i{{[0-9]+}} 0
     67     // LAMBDA: store double* [[G_PRIVATE_ADDR]], double** [[G_PRIVATE_ADDR_REF]]
     68     // LAMBDA: [[SVAR_PRIVATE_ADDR_REF:%.+]] = getelementptr inbounds %{{.+}}, %{{.+}}* [[ARG:%.+]], i{{[0-9]+}} 0, i{{[0-9]+}} 1
     69     // LAMBDA: store i{{[0-9]+}}* [[SVAR_PRIVATE_ADDR]], i{{[0-9]+}}** [[SVAR_PRIVATE_ADDR_REF]]
     70     // LAMBDA: [[SFVAR_PRIVATE_ADDR_REF:%.+]] = getelementptr inbounds %{{.+}}, %{{.+}}* [[ARG:%.+]], i{{[0-9]+}} 0, i{{[0-9]+}} 2
     71     // LAMBDA: store float* [[SFVAR_PRIVATE_ADDR]], float** [[SFVAR_PRIVATE_ADDR_REF]]
     72     // LAMBDA: call{{.*}} void [[INNER_LAMBDA:@.+]](%{{.+}}* [[ARG]])
     73     // LAMBDA: call {{.*}}void @__kmpc_for_static_fini(
     74     [&]() {
     75       // LAMBDA: define {{.+}} void [[INNER_LAMBDA]](%{{.+}}* [[ARG_PTR:%.+]])
     76       // LAMBDA: store %{{.+}}* [[ARG_PTR]], %{{.+}}** [[ARG_PTR_REF:%.+]],
     77       g = 2;
     78       g1 = 2;
     79       svar = 4;
     80       sfvar = 8.0;
     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 double*, double** [[G_PTR_REF]]
     84       // LAMBDA: store double 2.0{{.+}}, double* [[G_REF]]
     85       // LAMBDA: [[SVAR_PTR_REF:%.+]] = getelementptr inbounds %{{.+}}, %{{.+}}* [[ARG_PTR]], i{{[0-9]+}} 0, i{{[0-9]+}} 1
     86       // LAMBDA: [[SVAR_REF:%.+]] = load i{{[0-9]+}}*, i{{[0-9]+}}** [[SVAR_PTR_REF]]
     87       // LAMBDA: store i{{[0-9]+}} 4, i{{[0-9]+}}* [[SVAR_REF]]
     88       // LAMBDA: [[SFVAR_PTR_REF:%.+]] = getelementptr inbounds %{{.+}}, %{{.+}}* [[ARG_PTR]], i{{[0-9]+}} 0, i{{[0-9]+}} 2
     89       // LAMBDA: [[SFVAR_REF:%.+]] = load float*, float** [[SFVAR_PTR_REF]]
     90       // LAMBDA: store float 8.0{{.+}}, float* [[SFVAR_REF]]
     91     }();
     92   }
     93   }();
     94   return 0;
     95 #elif defined(BLOCKS)
     96   // BLOCKS: [[G:@.+]] = global double
     97   // BLOCKS-LABEL: @main
     98   // BLOCKS: call {{.*}}void {{%.+}}(i8
     99   ^{
    100   static float sfvar;
    101   // BLOCKS: define{{.*}} internal{{.*}} void {{.+}}(i8*
    102   // BLOCKS: call {{.*}}void {{.+}} @__kmpc_fork_call({{.+}}, i32 0, {{.+}}* [[OMP_REGION:@.+]] to {{.+}})
    103 #pragma omp parallel
    104 #pragma omp for private(g, g1, svar, sfvar)
    105   for (int i = 0; i < 2; ++i) {
    106     // BLOCKS: define{{.*}} internal{{.*}} void [[OMP_REGION]](i32* noalias %{{.+}}, i32* noalias %{{.+}})
    107     // BLOCKS: [[G_PRIVATE_ADDR:%.+]] = alloca double,
    108     // BLOCKS: [[SVAR_PRIVATE_ADDR:%.+]] = alloca i{{[0-9]+}},
    109     // BLOCKS: [[SFVAR_PRIVATE_ADDR:%.+]] = alloca float,
    110     g = 1;
    111     g1 = 1;
    112     svar = 2;
    113     sfvar = 3.0;
    114     // BLOCKS: call {{.*}}void @__kmpc_for_static_init_4(
    115     // BLOCKS: store double 1.0{{.+}}, double* [[G_PRIVATE_ADDR]],
    116     // BLOCKS-NOT: [[G]]{{[[^:word:]]}}
    117     // BLOCKS: store i{{[0-9]+}} 2, i{{[0-9]+}}* [[SVAR_PRIVATE_ADDR]],
    118     // BLOCKS-NOT: [[SVAR]]{{[[^:word:]]}}
    119     // BLOCKS: store float 3.0{{.+}}, float* [[SFVAR_PRIVATE_ADDR]],
    120     // BLOCKS-NOT: [[SFVAR]]{{[[^:word:]]}}
    121     // BLOCKS: double* [[G_PRIVATE_ADDR]]
    122     // BLOCKS-NOT: [[G]]{{[[^:word:]]}}
    123     // BLOCKS: i{{[0-9]+}}* [[SVAR_PRIVATE_ADDR]]
    124     // BLOCKS-NOT: [[SVAR]]{{[[^:word:]]}}
    125     // BLOCKS: float* [[SFVAR_PRIVATE_ADDR]]
    126     // BLOCKS-NOT: [[SFVAR]]{{[[^:word:]]}}
    127     // BLOCKS: call {{.*}}void {{%.+}}(i8
    128     // BLOCKS: call {{.*}}void @__kmpc_for_static_fini(
    129     ^{
    130       // BLOCKS: define {{.+}} void {{@.+}}(i8*
    131       g = 2;
    132       g1 = 2;
    133       svar = 4;
    134       sfvar = 9.0;
    135       // BLOCKS-NOT: [[G]]{{[[^:word:]]}}
    136       // BLOCKS: store double 2.0{{.+}}, double*
    137       // BLOCKS-NOT: [[G]]{{[[^:word:]]}}
    138       // BLOCKS-NOT: [[SVAR]]{{[[^:word:]]}}
    139       // BLOCKS: store i{{[0-9]+}} 4, i{{[0-9]+}}*
    140       // BLOCKS-NOT: [[SVAR]]{{[[^:word:]]}}
    141       // BLOCKS-NOT: [[SFVAR]]{{[[^:word:]]}}
    142       // BLOCKS: store float 9.0{{.+}}, float*
    143       // BLOCKS-NOT: [[SFVAR]]{{[[^:word:]]}}
    144       // BLOCKS: ret
    145     }();
    146   }
    147   }();
    148   return 0;
    149 #else
    150   S<float> test;
    151   int t_var = 0;
    152   int vec[] = {1, 2};
    153   S<float> s_arr[] = {1, 2};
    154   S<float> &var = test;
    155 #pragma omp parallel
    156 #pragma omp for private(t_var, vec, s_arr, s_arr, var, var, svar)
    157   for (int i = 0; i < 2; ++i) {
    158     vec[i] = t_var;
    159     s_arr[i] = var;
    160   }
    161   int i;
    162 #pragma omp parallel
    163 #pragma omp for private(i)
    164   for (i = 0; i < 2; ++i) {
    165     ;
    166   }
    167   return tmain<int>();
    168 #endif
    169 }
    170 
    171 // CHECK: define i{{[0-9]+}} @main()
    172 // CHECK: [[TEST:%.+]] = alloca [[S_FLOAT_TY]],
    173 // CHECK: call {{.*}} [[S_FLOAT_TY_DEF_CONSTR:@.+]]([[S_FLOAT_TY]]* [[TEST]])
    174 // CHECK: call void (%{{.+}}*, i{{[0-9]+}}, void (i{{[0-9]+}}*, i{{[0-9]+}}*, ...)*, ...) @__kmpc_fork_call(%{{.+}}* @{{.+}}, i{{[0-9]+}} 0, void (i{{[0-9]+}}*, i{{[0-9]+}}*, ...)* bitcast (void (i{{[0-9]+}}*, i{{[0-9]+}}*)* [[MAIN_MICROTASK:@.+]] to void
    175 // CHECK: = call i{{.+}} [[TMAIN_INT:@.+]]()
    176 // CHECK: call void [[S_FLOAT_TY_DESTR:@.+]]([[S_FLOAT_TY]]*
    177 // CHECK: ret
    178 //
    179 // CHECK: define internal void [[MAIN_MICROTASK]](i{{[0-9]+}}* noalias [[GTID_ADDR:%.+]], i{{[0-9]+}}* noalias %{{.+}})
    180 // CHECK: [[T_VAR_PRIV:%.+]] = alloca i{{[0-9]+}},
    181 // CHECK: [[VEC_PRIV:%.+]] = alloca [2 x i{{[0-9]+}}],
    182 // CHECK: [[S_ARR_PRIV:%.+]] = alloca [2 x [[S_FLOAT_TY]]],
    183 // CHECK-NOT: alloca [2 x [[S_FLOAT_TY]]],
    184 // CHECK: [[VAR_PRIV:%.+]] = alloca [[S_FLOAT_TY]],
    185 // CHECK-NOT: alloca [[S_FLOAT_TY]],
    186 // CHECK: [[S_VAR_PRIV:%.+]] = alloca i{{[0-9]+}},
    187 // CHECK: store i{{[0-9]+}}* [[GTID_ADDR]], i{{[0-9]+}}** [[GTID_ADDR_REF:%.+]]
    188 // CHECK-NOT: [[T_VAR_PRIV]]
    189 // CHECK-NOT: [[VEC_PRIV]]
    190 // CHECK: {{.+}}:
    191 // CHECK: [[S_ARR_PRIV_ITEM:%.+]] = phi [[S_FLOAT_TY]]*
    192 // CHECK: call {{.*}} [[S_FLOAT_TY_DEF_CONSTR]]([[S_FLOAT_TY]]* [[S_ARR_PRIV_ITEM]])
    193 // CHECK-NOT: [[T_VAR_PRIV]]
    194 // CHECK-NOT: [[VEC_PRIV]]
    195 // CHECK: call {{.*}} [[S_FLOAT_TY_DEF_CONSTR]]([[S_FLOAT_TY]]* [[VAR_PRIV]])
    196 // CHECK: call void @__kmpc_for_static_init_4(
    197 // CHECK: call void @__kmpc_for_static_fini(
    198 // CHECK-DAG: call void [[S_FLOAT_TY_DESTR]]([[S_FLOAT_TY]]* [[VAR_PRIV]])
    199 // CHECK-DAG: call void [[S_FLOAT_TY_DESTR]]([[S_FLOAT_TY]]*
    200 // CHECK: ret void
    201 
    202 // CHECK: define {{.*}} i{{[0-9]+}} [[TMAIN_INT]]()
    203 // CHECK: [[TEST:%.+]] = alloca [[S_INT_TY]],
    204 // CHECK: call {{.*}} [[S_INT_TY_DEF_CONSTR:@.+]]([[S_INT_TY]]* [[TEST]])
    205 // CHECK: call void (%{{.+}}*, i{{[0-9]+}}, void (i{{[0-9]+}}*, i{{[0-9]+}}*, ...)*, ...) @__kmpc_fork_call(%{{.+}}* @{{.+}}, i{{[0-9]+}} 0, void (i{{[0-9]+}}*, i{{[0-9]+}}*, ...)* bitcast (void (i{{[0-9]+}}*, i{{[0-9]+}}*)* [[TMAIN_MICROTASK:@.+]] to void
    206 // CHECK: call void [[S_INT_TY_DESTR:@.+]]([[S_INT_TY]]*
    207 // CHECK: ret
    208 //
    209 // CHECK: define internal void [[TMAIN_MICROTASK]](i{{[0-9]+}}* noalias [[GTID_ADDR:%.+]], i{{[0-9]+}}* noalias %{{.+}})
    210 // CHECK: [[T_VAR_PRIV:%.+]] = alloca i{{[0-9]+}},
    211 // CHECK: [[VEC_PRIV:%.+]] = alloca [2 x i{{[0-9]+}}],
    212 // CHECK: [[S_ARR_PRIV:%.+]] = alloca [2 x [[S_INT_TY]]],
    213 // CHECK-NOT: alloca [2 x [[S_INT_TY]]],
    214 // CHECK: [[VAR_PRIV:%.+]] = alloca [[S_INT_TY]],
    215 // CHECK-NOT: alloca [[S_INT_TY]],
    216 // CHECK: store i{{[0-9]+}}* [[GTID_ADDR]], i{{[0-9]+}}** [[GTID_ADDR_REF:%.+]]
    217 // CHECK-NOT: [[T_VAR_PRIV]]
    218 // CHECK-NOT: [[VEC_PRIV]]
    219 // CHECK: {{.+}}:
    220 // CHECK: [[S_ARR_PRIV_ITEM:%.+]] = phi [[S_INT_TY]]*
    221 // CHECK: call {{.*}} [[S_INT_TY_DEF_CONSTR]]([[S_INT_TY]]* [[S_ARR_PRIV_ITEM]])
    222 // CHECK-NOT: [[T_VAR_PRIV]]
    223 // CHECK-NOT: [[VEC_PRIV]]
    224 // CHECK: call {{.*}} [[S_INT_TY_DEF_CONSTR]]([[S_INT_TY]]* [[VAR_PRIV]])
    225 // CHECK: call void @__kmpc_for_static_init_4(
    226 // CHECK: call void @__kmpc_for_static_fini(
    227 // CHECK-DAG: call void [[S_INT_TY_DESTR]]([[S_INT_TY]]* [[VAR_PRIV]])
    228 // CHECK-DAG: call void [[S_INT_TY_DESTR]]([[S_INT_TY]]*
    229 // CHECK: ret void
    230 #endif
    231 
    232