xref: /minix3/external/bsd/llvm/dist/clang/test/OpenMP/parallel_private_codegen.cpp (revision 0a6a1f1d05b60e214de2f05a7310ddd1f0e590e7)
1 // RUN: %clang_cc1 -verify -fopenmp=libiomp5 -x c++ -triple x86_64-unknown-unknown -emit-llvm %s -o - | FileCheck %s
2 // RUN: %clang_cc1 -fopenmp=libiomp5 -x c++ -std=c++11 -triple x86_64-unknown-unknown -emit-pch -o %t %s
3 // RUN: %clang_cc1 -fopenmp=libiomp5 -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=libiomp5 -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=libiomp5 -x c++ -fblocks -DBLOCKS -triple %itanium_abi_triple -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;
SS13   S(T a) : f(a) {}
SS14   S() : f() {}
operator TS15   operator T() { return T(); }
~SS16   ~S() {}
17 };
18 
19 volatile int g = 1212;
20 
21 // CHECK: [[S_FLOAT_TY:%.+]] = type { float }
22 // CHECK: [[CAP_MAIN_TY:%.+]] = type { [2 x i{{[0-9]+}}]*, i{{[0-9]+}}*, [2 x [[S_FLOAT_TY]]]*, [[S_FLOAT_TY]]* }
23 // CHECK: [[S_INT_TY:%.+]] = type { i{{[0-9]+}} }
24 // CHECK: [[CAP_TMAIN_TY:%.+]] = type { [2 x i{{[0-9]+}}]*, i{{[0-9]+}}*, [2 x [[S_INT_TY]]]*, [[S_INT_TY]]* }
25 // CHECK: [[IMPLICIT_BARRIER_LOC:@.+]] = private unnamed_addr constant %{{.+}} { i32 0, i32 66, i32 0, i32 0, i8*
26 template <typename T>
tmain()27 T tmain() {
28   S<T> test;
29   T t_var = T();
30   T vec[] = {1, 2};
31   S<T> s_arr[] = {1, 2};
32   S<T> var(3);
33 #pragma omp parallel private(t_var, vec, s_arr, var)
34   {
35     vec[0] = t_var;
36     s_arr[0] = var;
37   }
38   return T();
39 }
40 
main()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: [[G_LOCAL_REF:%.+]] = getelementptr inbounds %{{.+}}* [[AGG_CAPTURED:%.+]], i{{[0-9]+}} 0, i{{[0-9]+}} 0
49   // LAMBDA: store i{{[0-9]+}}* [[G]], i{{[0-9]+}}** [[G_LOCAL_REF]]
50   // LAMBDA: [[ARG:%.+]] = bitcast %{{.+}}* [[AGG_CAPTURED]] to i8*
51   // LAMBDA: call void {{.+}}* @__kmpc_fork_call({{.+}}, i32 1, {{.+}}* [[OMP_REGION:@.+]] to {{.+}}, i8* [[ARG]])
52 #pragma omp parallel private(g)
53   {
54     // LAMBDA: define{{.*}} internal{{.*}} void [[OMP_REGION]](i32* %{{.+}}, i32* %{{.+}}, %{{.+}}* [[ARG:%.+]])
55     // LAMBDA: [[G_PRIVATE_ADDR:%.+]] = alloca i{{[0-9]+}},
56     // LAMBDA: store %{{.+}}* [[ARG]], %{{.+}}** [[ARG_REF:%.+]],
57     // LAMBDA: call i32 @__kmpc_cancel_barrier(
58     g = 1;
59     // LAMBDA: store volatile i{{[0-9]+}} 1, i{{[0-9]+}}* [[G_PRIVATE_ADDR]],
60     // LAMBDA: [[G_PRIVATE_ADDR_REF:%.+]] = getelementptr inbounds %{{.+}}* [[ARG:%.+]], i{{[0-9]+}} 0, i{{[0-9]+}} 0
61     // LAMBDA: store i{{[0-9]+}}* [[G_PRIVATE_ADDR]], i{{[0-9]+}}** [[G_PRIVATE_ADDR_REF]]
62     // LAMBDA: call void [[INNER_LAMBDA:@.+]](%{{.+}}* [[ARG]])
63     [&]() {
64       // LAMBDA: define {{.+}} void [[INNER_LAMBDA]](%{{.+}}* [[ARG_PTR:%.+]])
65       // LAMBDA: store %{{.+}}* [[ARG_PTR]], %{{.+}}** [[ARG_PTR_REF:%.+]],
66       g = 2;
67       // LAMBDA: [[ARG_PTR:%.+]] = load %{{.+}}** [[ARG_PTR_REF]]
68       // LAMBDA: [[G_PTR_REF:%.+]] = getelementptr inbounds %{{.+}}* [[ARG_PTR]], i{{[0-9]+}} 0, i{{[0-9]+}} 0
69       // LAMBDA: [[G_REF:%.+]] = load i{{[0-9]+}}** [[G_PTR_REF]]
70       // LAMBDA: store volatile i{{[0-9]+}} 2, i{{[0-9]+}}* [[G_REF]]
71     }();
72   }
73   }();
74   return 0;
75 #elif defined(BLOCKS)
76   // BLOCKS: [[G:@.+]] = global i{{[0-9]+}} 1212,
77   // BLOCKS-LABEL: @main
78   // BLOCKS: call void {{%.+}}(i8*
79   ^{
80   // BLOCKS: define{{.*}} internal{{.*}} void {{.+}}(i8*
81   // BLOCKS: [[G_LOCAL_REF:%.+]] = getelementptr inbounds %{{.+}}* [[AGG_CAPTURED:%.+]], i{{[0-9]+}} 0, i{{[0-9]+}} 0
82   // BLOCKS: store i{{[0-9]+}}* [[G]], i{{[0-9]+}}** [[G_LOCAL_REF]]
83   // BLOCKS: [[ARG:%.+]] = bitcast %{{.+}}* [[AGG_CAPTURED]] to i8*
84   // BLOCKS: call void {{.+}}* @__kmpc_fork_call({{.+}}, i32 1, {{.+}}* [[OMP_REGION:@.+]] to {{.+}}, i8* [[ARG]])
85 #pragma omp parallel private(g)
86   {
87     // BLOCKS: define{{.*}} internal{{.*}} void [[OMP_REGION]](i32* %{{.+}}, i32* %{{.+}}, %{{.+}}* [[ARG:%.+]])
88     // BLOCKS: [[G_PRIVATE_ADDR:%.+]] = alloca i{{[0-9]+}},
89     // BLOCKS: store %{{.+}}* [[ARG]], %{{.+}}** [[ARG_REF:%.+]],
90     // BLOCKS: call i32 @__kmpc_cancel_barrier(
91     g = 1;
92     // BLOCKS: store volatile i{{[0-9]+}} 1, i{{[0-9]+}}* [[G_PRIVATE_ADDR]],
93     // BLOCKS-NOT: [[G]]{{[[^:word:]]}}
94     // BLOCKS: i{{[0-9]+}}* [[G_PRIVATE_ADDR]]
95     // BLOCKS-NOT: [[G]]{{[[^:word:]]}}
96     // BLOCKS: call void {{%.+}}(i8*
97     ^{
98       // BLOCKS: define {{.+}} void {{@.+}}(i8*
99       g = 2;
100       // BLOCKS-NOT: [[G]]{{[[^:word:]]}}
101       // BLOCKS: store volatile i{{[0-9]+}} 2, i{{[0-9]+}}*
102       // BLOCKS-NOT: [[G]]{{[[^:word:]]}}
103       // BLOCKS: ret
104     }();
105   }
106   }();
107   return 0;
108 #else
109   S<float> test;
110   int t_var = 0;
111   int vec[] = {1, 2};
112   S<float> s_arr[] = {1, 2};
113   S<float> var(3);
114 #pragma omp parallel private(t_var, vec, s_arr, var)
115   {
116     vec[0] = t_var;
117     s_arr[0] = var;
118   }
119   return tmain<int>();
120 #endif
121 }
122 
123 // CHECK: define i{{[0-9]+}} @main()
124 // CHECK: [[TEST:%.+]] = alloca [[S_FLOAT_TY]],
125 // CHECK: call {{.*}} [[S_FLOAT_TY_DEF_CONSTR:@.+]]([[S_FLOAT_TY]]* [[TEST]])
126 // CHECK: %{{.+}} = bitcast [[CAP_MAIN_TY]]*
127 // CHECK: call void (%{{.+}}*, i{{[0-9]+}}, void (i{{[0-9]+}}*, i{{[0-9]+}}*, ...)*, ...)* @__kmpc_fork_call(%{{.+}}* @{{.+}}, i{{[0-9]+}} 1, void (i{{[0-9]+}}*, i{{[0-9]+}}*, ...)* bitcast (void (i{{[0-9]+}}*, i{{[0-9]+}}*, [[CAP_MAIN_TY]]*)* [[MAIN_MICROTASK:@.+]] to void
128 // CHECK: = call i{{.+}} [[TMAIN_INT:@.+]]()
129 // CHECK: call void [[S_FLOAT_TY_DESTR:@.+]]([[S_FLOAT_TY]]*
130 // CHECK: ret
131 //
132 // CHECK: define internal void [[MAIN_MICROTASK]](i{{[0-9]+}}* [[GTID_ADDR:%.+]], i{{[0-9]+}}* %{{.+}}, [[CAP_MAIN_TY]]* %{{.+}})
133 // CHECK: [[T_VAR_PRIV:%.+]] = alloca i{{[0-9]+}},
134 // CHECK: [[VEC_PRIV:%.+]] = alloca [2 x i{{[0-9]+}}],
135 // CHECK: [[S_ARR_PRIV:%.+]] = alloca [2 x [[S_FLOAT_TY]]],
136 // CHECK: [[VAR_PRIV:%.+]] = alloca [[S_FLOAT_TY]],
137 // CHECK: store i{{[0-9]+}}* [[GTID_ADDR]], i{{[0-9]+}}** [[GTID_ADDR_REF:%.+]]
138 // CHECK-NOT: [[T_VAR_PRIV]]
139 // CHECK-NOT: [[VEC_PRIV]]
140 // CHECK: {{.+}}:
141 // CHECK: [[S_ARR_PRIV_ITEM:%.+]] = phi [[S_FLOAT_TY]]*
142 // CHECK: call {{.*}} [[S_FLOAT_TY_DEF_CONSTR]]([[S_FLOAT_TY]]* [[S_ARR_PRIV_ITEM]])
143 // CHECK-NOT: [[T_VAR_PRIV]]
144 // CHECK-NOT: [[VEC_PRIV]]
145 // CHECK: call {{.*}} [[S_FLOAT_TY_DEF_CONSTR]]([[S_FLOAT_TY]]* [[VAR_PRIV]])
146 // CHECK: [[GTID_REF:%.+]] = load i{{[0-9]+}}** [[GTID_ADDR_REF]]
147 // CHECK: [[GTID:%.+]] = load i{{[0-9]+}}* [[GTID_REF]]
148 // CHECK: call i32 @__kmpc_cancel_barrier(%{{.+}}* [[IMPLICIT_BARRIER_LOC]], i{{[0-9]+}} [[GTID]])
149 // CHECK-DAG: call void [[S_FLOAT_TY_DESTR]]([[S_FLOAT_TY]]* [[VAR_PRIV]])
150 // CHECK-DAG: call void [[S_FLOAT_TY_DESTR]]([[S_FLOAT_TY]]*
151 // CHECK: ret void
152 
153 // CHECK: define {{.*}} i{{[0-9]+}} [[TMAIN_INT]]()
154 // CHECK: [[TEST:%.+]] = alloca [[S_INT_TY]],
155 // CHECK: call {{.*}} [[S_INT_TY_DEF_CONSTR:@.+]]([[S_INT_TY]]* [[TEST]])
156 // CHECK: call void (%{{.+}}*, i{{[0-9]+}}, void (i{{[0-9]+}}*, i{{[0-9]+}}*, ...)*, ...)* @__kmpc_fork_call(%{{.+}}* @{{.+}}, i{{[0-9]+}} 1, void (i{{[0-9]+}}*, i{{[0-9]+}}*, ...)* bitcast (void (i{{[0-9]+}}*, i{{[0-9]+}}*, [[CAP_TMAIN_TY]]*)* [[TMAIN_MICROTASK:@.+]] to void
157 // CHECK: call void [[S_INT_TY_DESTR:@.+]]([[S_INT_TY]]*
158 // CHECK: ret
159 //
160 // CHECK: define internal void [[TMAIN_MICROTASK]](i{{[0-9]+}}* [[GTID_ADDR:%.+]], i{{[0-9]+}}* %{{.+}}, [[CAP_TMAIN_TY]]* %{{.+}})
161 // CHECK: [[T_VAR_PRIV:%.+]] = alloca i{{[0-9]+}},
162 // CHECK: [[VEC_PRIV:%.+]] = alloca [2 x i{{[0-9]+}}],
163 // CHECK: [[S_ARR_PRIV:%.+]] = alloca [2 x [[S_INT_TY]]],
164 // CHECK: [[VAR_PRIV:%.+]] = alloca [[S_INT_TY]],
165 // CHECK: store i{{[0-9]+}}* [[GTID_ADDR]], i{{[0-9]+}}** [[GTID_ADDR_REF:%.+]]
166 // CHECK-NOT: [[T_VAR_PRIV]]
167 // CHECK-NOT: [[VEC_PRIV]]
168 // CHECK: {{.+}}:
169 // CHECK: [[S_ARR_PRIV_ITEM:%.+]] = phi [[S_INT_TY]]*
170 // CHECK: call {{.*}} [[S_INT_TY_DEF_CONSTR]]([[S_INT_TY]]* [[S_ARR_PRIV_ITEM]])
171 // CHECK-NOT: [[T_VAR_PRIV]]
172 // CHECK-NOT: [[VEC_PRIV]]
173 // CHECK: call {{.*}} [[S_INT_TY_DEF_CONSTR]]([[S_INT_TY]]* [[VAR_PRIV]])
174 // CHECK: [[GTID_REF:%.+]] = load i{{[0-9]+}}** [[GTID_ADDR_REF]]
175 // CHECK: [[GTID:%.+]] = load i{{[0-9]+}}* [[GTID_REF]]
176 // CHECK: call i32 @__kmpc_cancel_barrier(%{{.+}}* [[IMPLICIT_BARRIER_LOC]], i{{[0-9]+}} [[GTID]])
177 // CHECK-DAG: call void [[S_INT_TY_DESTR]]([[S_INT_TY]]* [[VAR_PRIV]])
178 // CHECK-DAG: call void [[S_INT_TY_DESTR]]([[S_INT_TY]]*
179 // CHECK: ret void
180 #endif
181 
182