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 // RUN: %clang_cc1 -verify -fopenmp -x c++ -std=c++11 -DARRAY -triple x86_64-apple-darwin10 -emit-llvm %s -o - | FileCheck -check-prefix=ARRAY %s
7
8 // RUN: %clang_cc1 -verify -fopenmp-simd -x c++ -triple x86_64-apple-darwin10 -emit-llvm %s -o - | FileCheck --check-prefix SIMD-ONLY0 %s
9 // RUN: %clang_cc1 -fopenmp-simd -x c++ -std=c++11 -triple x86_64-apple-darwin10 -emit-pch -o %t %s
10 // 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
11 // 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
12 // 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
13 // RUN: %clang_cc1 -verify -fopenmp-simd -x c++ -std=c++11 -DARRAY -triple x86_64-apple-darwin10 -emit-llvm %s -o - | FileCheck --check-prefix SIMD-ONLY0 %s
14 // SIMD-ONLY0-NOT: {{__kmpc|__tgt}}
15 // expected-no-diagnostics
16
17 #ifndef ARRAY
18 #ifndef HEADER
19 #define HEADER
20
21 template <class T>
22 struct S {
23 T f;
SS24 S(T a) : f(a) {}
SS25 S() : f() {}
operator TS26 operator T() { return T(); }
~SS27 ~S() {}
28 };
29
30 volatile double g;
31
32 // CHECK-DAG: [[KMP_TASK_T_TY:%.+]] = type { i8*, i32 (i32, i8*)*, i32, %union{{.+}}, %union{{.+}}, i64, i64, i64, i32, i8* }
33 // CHECK-DAG: [[S_DOUBLE_TY:%.+]] = type { double }
34 // CHECK-DAG: [[CAP_MAIN_TY:%.+]] = type { i8 }
35 // CHECK-DAG: [[PRIVATES_MAIN_TY:%.+]] = type {{.?}}{ [2 x [[S_DOUBLE_TY]]], [[S_DOUBLE_TY]], i32, [2 x i32]
36 // CHECK-DAG: [[KMP_TASK_MAIN_TY:%.+]] = type { [[KMP_TASK_T_TY]], [[PRIVATES_MAIN_TY]] }
37 // CHECK-DAG: [[S_INT_TY:%.+]] = type { i32 }
38 // CHECK-DAG: [[CAP_TMAIN_TY:%.+]] = type { i8 }
39 // CHECK-DAG: [[PRIVATES_TMAIN_TY:%.+]] = type { i32, [2 x i32], [2 x [[S_INT_TY]]], [[S_INT_TY]], [104 x i8] }
40 // CHECK-DAG: [[KMP_TASK_TMAIN_TY:%.+]] = type { [[KMP_TASK_T_TY]], [{{[0-9]+}} x i8], [[PRIVATES_TMAIN_TY]] }
41 template <typename T>
tmain()42 T tmain() {
43 S<T> test;
44 T t_var __attribute__((aligned(128))) = T();
45 T vec[] = {1, 2};
46 S<T> s_arr[] = {1, 2};
47 S<T> var(3);
48 #pragma omp parallel master taskloop simd private(t_var, vec, s_arr, s_arr, var, var)
49 for (int i = 0; i < 10; ++i) {
50 vec[0] = t_var;
51 s_arr[0] = var;
52 }
53 return T();
54 }
55
main()56 int main() {
57 static int sivar;
58 #ifdef LAMBDA
59 // LAMBDA: [[G:@.+]] ={{.*}} global double
60 // LAMBDA-LABEL: @main
61 // LAMBDA: call{{( x86_thiscallcc)?}} void [[OUTER_LAMBDA:@.+]](
62 [&]() {
63 // LAMBDA: define{{.*}} internal{{.*}} void [[OUTER_LAMBDA]](
64 // LAMBDA: [[RES:%.+]] = call i8* @__kmpc_omp_task_alloc(%{{[^ ]+}} @{{[^,]+}}, i32 %{{[^,]+}}, i32 1, i64 96, i64 1, i32 (i32, i8*)* bitcast (i32 (i32, %{{[^*]+}}*)* [[TASK_ENTRY:@[^ ]+]] to i32 (i32, i8*)*))
65 // LAMBDA: [[PRIVATES:%.+]] = getelementptr inbounds %{{.+}}, %{{.+}}* %{{.+}}, i{{.+}} 0, i{{.+}} 1
66 // LAMBDA: call void @__kmpc_taskloop(%{{.+}}* @{{.+}}, i32 %{{.+}}, i8* [[RES]], i32 1, i64* %{{.+}}, i64* %{{.+}}, i64 %{{.+}}, i32 1, i32 0, i64 0, i8* null)
67 // LAMBDA: ret
68 #pragma omp parallel master taskloop simd private(g, sivar)
69 for (int i = 0; i < 10; ++i) {
70 // LAMBDA: define {{.+}} void [[INNER_LAMBDA:@.+]](%{{.+}}* {{[^,]*}} [[ARG_PTR:%.+]])
71 // LAMBDA: store %{{.+}}* [[ARG_PTR]], %{{.+}}** [[ARG_PTR_REF:%.+]],
72 // LAMBDA: [[ARG_PTR:%.+]] = load %{{.+}}*, %{{.+}}** [[ARG_PTR_REF]]
73 // LAMBDA: [[G_PTR_REF:%.+]] = getelementptr inbounds %{{.+}}, %{{.+}}* [[ARG_PTR]], i{{[0-9]+}} 0, i{{[0-9]+}} 0
74 // LAMBDA: [[G_REF:%.+]] = load double*, double** [[G_PTR_REF]]
75 // LAMBDA: store double 2.0{{.+}}, double* [[G_REF]]
76 // LAMBDA: [[SIVAR_PTR_REF:%.+]] = getelementptr inbounds %{{.+}}, %{{.+}}* [[ARG_PTR]], i{{[0-9]+}} 0, i{{[0-9]+}} 1
77 // LAMBDA: [[SIVAR_REF:%.+]] = load i{{[0-9]+}}*, i{{[0-9]+}}** [[SIVAR_PTR_REF]]
78 // LAMBDA: store i{{[0-9]+}} 3, i{{[0-9]+}}* [[SIVAR_REF]]
79
80 // LAMBDA: define internal i32 [[TASK_ENTRY]](i32 %0, %{{.+}}* noalias %1)
81 g = 1;
82 sivar = 2;
83 // LAMBDA: store double 1.0{{.+}}, double* %{{.+}},
84 // LAMBDA: store i{{[0-9]+}} 2, i{{[0-9]+}}* %{{.+}},
85 // LAMBDA: call void [[INNER_LAMBDA]](%
86 // LAMBDA: ret
87 [&]() {
88 g = 2;
89 sivar = 3;
90 }();
91 }
92 }();
93 return 0;
94 #elif defined(BLOCKS)
95 // BLOCKS: [[G:@.+]] ={{.*}} global double
96 // BLOCKS: [[SIVAR:@.+]] = internal global i{{[0-9]+}} 0,
97 // BLOCKS-LABEL: @main
98 // BLOCKS: call void {{%.+}}(i8
99 ^{
100 // BLOCKS: define{{.*}} internal{{.*}} void {{.+}}(i8*
101 // BLOCKS: [[RES:%.+]] = call i8* @__kmpc_omp_task_alloc(%{{[^ ]+}} @{{[^,]+}}, i32 %{{[^,]+}}, i32 1, i64 96, i64 1, i32 (i32, i8*)* bitcast (i32 (i32, %{{[^*]+}}*)* [[TASK_ENTRY:@[^ ]+]] to i32 (i32, i8*)*))
102 // BLOCKS: [[PRIVATES:%.+]] = getelementptr inbounds %{{.+}}, %{{.+}}* %{{.+}}, i{{.+}} 0, i{{.+}} 1
103 // BLOCKS: call void @__kmpc_taskloop(%{{.+}}* @{{.+}}, i32 %{{.+}}, i8* [[RES]], i32 1, i64* %{{.+}}, i64* %{{.+}}, i64 %{{.+}}, i32 1, i32 0, i64 0, i8* null)
104 // BLOCKS: ret
105 #pragma omp parallel master taskloop simd private(g, sivar)
106 for (int i = 0; i < 10; ++i) {
107 // BLOCKS: define {{.+}} void {{@.+}}(i8*
108 // BLOCKS-NOT: [[G]]{{[[^:word:]]}}
109 // BLOCKS: store double 2.0{{.+}}, double*
110 // BLOCKS-NOT: [[G]]{{[[^:word:]]}}
111 // BLOCKS-NOT: [[SIVAR]]{{[[^:word:]]}}
112 // BLOCKS: store i{{[0-9]+}} 4, i{{[0-9]+}}*
113 // BLOCKS-NOT: [[SIVAR]]{{[[^:word:]]}}
114 // BLOCKS: ret
115
116 // BLOCKS: define internal i32 [[TASK_ENTRY]](i32 %0, %{{.+}}* noalias %1)
117 g = 1;
118 sivar = 3;
119 // BLOCKS: store double 1.0{{.+}}, double* %{{.+}},
120 // BLOCKS-NOT: [[G]]{{[[^:word:]]}}
121 // BLOCKS: store i{{[0-9]+}} 3, i{{[0-9]+}}* %{{.+}},
122 // BLOCKS-NOT: [[SIVAR]]{{[[^:word:]]}}
123 // BLOCKS: call void {{%.+}}(i8
124 ^{
125 g = 2;
126 sivar = 4;
127 }();
128 }
129 }();
130 return 0;
131 #else
132 S<double> test;
133 int t_var = 0;
134 int vec[] = {1, 2};
135 S<double> s_arr[] = {1, 2};
136 S<double> var(3);
137 #pragma omp parallel master taskloop simd private(var, t_var, s_arr, vec, s_arr, var, sivar)
138 for (int i = 0; i < 10; ++i) {
139 vec[0] = t_var;
140 s_arr[0] = var;
141 sivar = 8;
142 }
143 #pragma omp task
144 g+=1;
145 return tmain<int>();
146 #endif
147 }
148
149 // CHECK: define{{.*}} i{{[0-9]+}} @main()
150 // CHECK: [[TEST:%.+]] = alloca [[S_DOUBLE_TY]],
151 // CHECK: [[T_VAR_ADDR:%.+]] = alloca i32,
152 // CHECK: [[VEC_ADDR:%.+]] = alloca [2 x i32],
153 // CHECK: [[S_ARR_ADDR:%.+]] = alloca [2 x [[S_DOUBLE_TY]]],
154 // CHECK: [[VAR_ADDR:%.+]] = alloca [[S_DOUBLE_TY]],
155
156 // CHECK: call {{.*}} [[S_DOUBLE_TY_DEF_CONSTR:@.+]]([[S_DOUBLE_TY]]* {{[^,]*}} [[TEST]])
157
158 // CHECK: [[RES:%.+]] = call {{.*}}i32 @__kmpc_master(
159 // CHECK-NEXT: [[IS_MASTER:%.+]] = icmp ne i32 [[RES]], 0
160 // CHECK-NEXT: br i1 [[IS_MASTER]], label {{%?}}[[THEN:.+]], label {{%?}}[[EXIT:.+]]
161 // CHECK: [[THEN]]
162 // Do not store original variables in capture struct.
163 // CHECK-NOT: getelementptr inbounds [[CAP_MAIN_TY]],
164
165 // Allocate task.
166 // Returns struct kmp_task_t {
167 // [[KMP_TASK_T_TY]] task_data;
168 // [[KMP_TASK_MAIN_TY]] privates;
169 // };
170 // CHECK: [[RES:%.+]] = call i8* @__kmpc_omp_task_alloc([[LOC:%.+]], i32 [[GTID:%.+]], i32 9, i64 120, i64 1, i32 (i32, i8*)* bitcast (i32 (i32, [[KMP_TASK_MAIN_TY]]*)* [[TASK_ENTRY:@[^ ]+]] to i32 (i32, i8*)*))
171 // CHECK: [[RES_KMP_TASK:%.+]] = bitcast i8* [[RES]] to [[KMP_TASK_MAIN_TY]]*
172
173 // CHECK: [[TASK:%.+]] = getelementptr inbounds [[KMP_TASK_MAIN_TY]], [[KMP_TASK_MAIN_TY]]* [[RES_KMP_TASK]], i{{[0-9]+}} 0, i{{[0-9]+}} 0
174 // Initialize kmp_task_t->privates with default values (no init for simple types, default constructors for classes).
175 // Also copy address of private copy to the corresponding shareds reference.
176 // CHECK: [[PRIVATES:%.+]] = getelementptr inbounds [[KMP_TASK_MAIN_TY]], [[KMP_TASK_MAIN_TY]]* [[RES_KMP_TASK]], i{{[0-9]+}} 0, i{{[0-9]+}} 1
177
178 // Constructors for s_arr and var.
179 // a_arr;
180 // CHECK: [[PRIVATE_S_ARR_REF:%.+]] = getelementptr inbounds [[PRIVATES_MAIN_TY]], [[PRIVATES_MAIN_TY]]* [[PRIVATES]], i{{[0-9]+}} 0, i{{[0-9]+}} 0
181 // CHECK: getelementptr inbounds [2 x [[S_DOUBLE_TY]]], [2 x [[S_DOUBLE_TY]]]* [[PRIVATE_S_ARR_REF]], i{{.+}} 0, i{{.+}} 0
182 // CHECK: getelementptr inbounds [[S_DOUBLE_TY]], [[S_DOUBLE_TY]]* %{{.+}}, i{{.+}} 2
183 // CHECK: call void [[S_DOUBLE_TY_DEF_CONSTR]]([[S_DOUBLE_TY]]* {{[^,]*}} [[S_ARR_CUR:%.+]])
184 // CHECK: getelementptr inbounds [[S_DOUBLE_TY]], [[S_DOUBLE_TY]]* [[S_ARR_CUR]], i{{.+}} 1
185 // CHECK: icmp eq
186 // CHECK: br i1
187
188 // var;
189 // CHECK: [[PRIVATE_VAR_REF:%.+]] = getelementptr inbounds [[PRIVATES_MAIN_TY]], [[PRIVATES_MAIN_TY]]* [[PRIVATES]], i{{.+}} 0, i{{.+}} 1
190 // CHECK: call void [[S_DOUBLE_TY_DEF_CONSTR]]([[S_DOUBLE_TY]]* {{[^,]*}} [[PRIVATE_VAR_REF:%.+]])
191
192 // Provide pointer to destructor function, which will destroy private variables at the end of the task.
193 // CHECK: [[DESTRUCTORS_REF:%.+]] = getelementptr inbounds [[KMP_TASK_T_TY]], [[KMP_TASK_T_TY]]* [[TASK]], i{{.+}} 0, i{{.+}} 3
194 // CHECK: [[DESTRUCTORS_PTR:%.+]] = bitcast %union{{.+}}* [[DESTRUCTORS_REF]] to i32 (i32, i8*)**
195 // CHECK: store i32 (i32, i8*)* bitcast (i32 (i32, [[KMP_TASK_MAIN_TY]]*)* [[DESTRUCTORS:@.+]] to i32 (i32, i8*)*), i32 (i32, i8*)** [[DESTRUCTORS_PTR]],
196
197 // Start task.
198 // CHECK: call void @__kmpc_taskloop([[LOC]], i32 [[GTID]], i8* [[RES]], i32 1, i64* %{{.+}}, i64* %{{.+}}, i64 %{{.+}}, i32 1, i32 0, i64 0, i8* bitcast (void ([[KMP_TASK_MAIN_TY]]*, [[KMP_TASK_MAIN_TY]]*, i32)* [[MAIN_DUP:@.+]] to i8*))
199 // CHECK: call {{.*}}void @__kmpc_end_master(
200 // CHECK-NEXT: br label {{%?}}[[EXIT]]
201 // CHECK: [[EXIT]]
202
203 // No destructors must be called for private copies of s_arr and var.
204 // CHECK-NOT: getelementptr inbounds [[PRIVATES_MAIN_TY]], [[PRIVATES_MAIN_TY]]* [[PRIVATES]], i{{.+}} 0, i{{.+}} 2
205 // CHECK-NOT: getelementptr inbounds [[PRIVATES_MAIN_TY]], [[PRIVATES_MAIN_TY]]* [[PRIVATES]], i{{.+}} 0, i{{.+}} 3
206 //
207
208 // CHECK: define internal void [[PRIVATES_MAP_FN:@.+]]([[PRIVATES_MAIN_TY]]* noalias %0, [[S_DOUBLE_TY]]** noalias %1, i32** noalias %2, [2 x [[S_DOUBLE_TY]]]** noalias %3, [2 x i32]** noalias %4, i32** noalias %5)
209 // CHECK: [[PRIVATES:%.+]] = load [[PRIVATES_MAIN_TY]]*, [[PRIVATES_MAIN_TY]]**
210 // CHECK: [[PRIV_S_VAR:%.+]] = getelementptr inbounds [[PRIVATES_MAIN_TY]], [[PRIVATES_MAIN_TY]]* [[PRIVATES]], i32 0, i32 0
211 // CHECK: [[ARG3:%.+]] = load [2 x [[S_DOUBLE_TY]]]**, [2 x [[S_DOUBLE_TY]]]*** %{{.+}},
212 // CHECK: store [2 x [[S_DOUBLE_TY]]]* [[PRIV_S_VAR]], [2 x [[S_DOUBLE_TY]]]** [[ARG3]],
213 // CHECK: [[PRIV_VAR:%.+]] = getelementptr inbounds [[PRIVATES_MAIN_TY]], [[PRIVATES_MAIN_TY]]* [[PRIVATES]], i32 0, i32 1
214 // CHECK: [[ARG1:%.+]] = load [[S_DOUBLE_TY]]**, [[S_DOUBLE_TY]]*** {{.+}},
215 // CHECK: store [[S_DOUBLE_TY]]* [[PRIV_VAR]], [[S_DOUBLE_TY]]** [[ARG1]],
216 // CHECK: [[PRIV_T_VAR:%.+]] = getelementptr inbounds [[PRIVATES_MAIN_TY]], [[PRIVATES_MAIN_TY]]* [[PRIVATES]], i32 0, i32 2
217 // CHECK: [[ARG2:%.+]] = load i32**, i32*** %{{.+}},
218 // CHECK: store i32* [[PRIV_T_VAR]], i32** [[ARG2]],
219 // CHECK: [[PRIV_VEC:%.+]] = getelementptr inbounds [[PRIVATES_MAIN_TY]], [[PRIVATES_MAIN_TY]]* [[PRIVATES]], i32 0, i32 3
220 // CHECK: [[ARG4:%.+]] = load [2 x i32]**, [2 x i32]*** %{{.+}},
221 // CHECK: store [2 x i32]* [[PRIV_VEC]], [2 x i32]** [[ARG4]],
222 // CHECK: ret void
223
224 // CHECK: define internal i32 [[TASK_ENTRY]](i32 %0, [[KMP_TASK_MAIN_TY]]* noalias %1)
225
226 // CHECK: [[PRIV_VAR_ADDR:%.+]] = alloca [[S_DOUBLE_TY]]*,
227 // CHECK: [[PRIV_T_VAR_ADDR:%.+]] = alloca i32*,
228 // CHECK: [[PRIV_S_ARR_ADDR:%.+]] = alloca [2 x [[S_DOUBLE_TY]]]*,
229 // CHECK: [[PRIV_VEC_ADDR:%.+]] = alloca [2 x i32]*,
230 // CHECK: [[PRIV_SIVAR_ADDR:%.+]] = alloca i32*,
231 // CHECK: store void (i8*, ...)* bitcast (void ([[PRIVATES_MAIN_TY]]*, [[S_DOUBLE_TY]]**, i32**, [2 x [[S_DOUBLE_TY]]]**, [2 x i32]**, i32**)* [[PRIVATES_MAP_FN]] to void (i8*, ...)*), void (i8*, ...)** [[MAP_FN_ADDR:%.+]],
232 // CHECK: [[MAP_FN:%.+]] = load void (i8*, ...)*, void (i8*, ...)** [[MAP_FN_ADDR]],
233 // CHECK: [[FN:%.+]] = bitcast void (i8*, ...)* [[MAP_FN]] to void (i8*,
234 // CHECK: call void [[FN]](i8* %{{.+}}, [[S_DOUBLE_TY]]** [[PRIV_VAR_ADDR]], i32** [[PRIV_T_VAR_ADDR]], [2 x [[S_DOUBLE_TY]]]** [[PRIV_S_ARR_ADDR]], [2 x i32]** [[PRIV_VEC_ADDR]], i32** [[PRIV_SIVAR_ADDR]])
235 // CHECK: [[PRIV_VAR:%.+]] = load [[S_DOUBLE_TY]]*, [[S_DOUBLE_TY]]** [[PRIV_VAR_ADDR]],
236 // CHECK: [[PRIV_T_VAR:%.+]] = load i32*, i32** [[PRIV_T_VAR_ADDR]],
237 // CHECK: [[PRIV_S_ARR:%.+]] = load [2 x [[S_DOUBLE_TY]]]*, [2 x [[S_DOUBLE_TY]]]** [[PRIV_S_ARR_ADDR]],
238 // CHECK: [[PRIV_VEC:%.+]] = load [2 x i32]*, [2 x i32]** [[PRIV_VEC_ADDR]],
239 // CHECK: [[PRIV_SIVAR:%.+]] = load i32*, i32** [[PRIV_SIVAR_ADDR]],
240
241 // Privates actually are used.
242 // CHECK-DAG: [[PRIV_VAR]]
243 // CHECK-DAG: [[PRIV_T_VAR]]
244 // CHECK-DAG: [[PRIV_S_ARR]]
245 // CHECK-DAG: [[PRIV_VEC]]
246 // CHECK_DAG: [[PRIV_SIVAR]]
247
248 // CHECK: ret
249
250 // CHECK: define internal void [[MAIN_DUP]]([[KMP_TASK_MAIN_TY]]* %0, [[KMP_TASK_MAIN_TY]]* %1, i32 %2)
251 // CHECK: getelementptr inbounds [[KMP_TASK_MAIN_TY]], [[KMP_TASK_MAIN_TY]]* %{{.+}}, i32 0, i32 1
252 // CHECK: getelementptr inbounds [[PRIVATES_MAIN_TY]], [[PRIVATES_MAIN_TY]]* %{{.+}}, i32 0, i32 0
253 // CHECK: getelementptr inbounds [2 x [[S_DOUBLE_TY]]], [2 x [[S_DOUBLE_TY]]]* %{{.+}}, i32 0, i32 0
254 // CHECK: getelementptr inbounds [[S_DOUBLE_TY]], [[S_DOUBLE_TY]]* %{{.+}}, i64 2
255 // CHECK: br label %
256
257 // CHECK: phi [[S_DOUBLE_TY]]*
258 // CHECK: call {{.*}} [[S_DOUBLE_TY_DEF_CONSTR]]([[S_DOUBLE_TY]]*
259 // CHECK: getelementptr inbounds [[S_DOUBLE_TY]], [[S_DOUBLE_TY]]* %{{.+}}, i64 1
260 // CHECK: icmp eq [[S_DOUBLE_TY]]* %
261 // CHECK: br i1 %
262
263 // CHECK: getelementptr inbounds [[PRIVATES_MAIN_TY]], [[PRIVATES_MAIN_TY]]* %{{.+}}, i32 0, i32 1
264 // CHECK: call {{.*}} [[S_DOUBLE_TY_DEF_CONSTR]]([[S_DOUBLE_TY]]*
265 // CHECK: ret void
266
267 // CHECK: define internal i32 [[DESTRUCTORS]](i32 %0, [[KMP_TASK_MAIN_TY]]* noalias %1)
268 // CHECK: [[PRIVATES:%.+]] = getelementptr inbounds [[KMP_TASK_MAIN_TY]], [[KMP_TASK_MAIN_TY]]* [[RES_KMP_TASK:%.+]], i{{[0-9]+}} 0, i{{[0-9]+}} 1
269 // CHECK: [[PRIVATE_S_ARR_REF:%.+]] = getelementptr inbounds [[PRIVATES_MAIN_TY]], [[PRIVATES_MAIN_TY]]* [[PRIVATES]], i{{.+}} 0, i{{.+}} 0
270 // CHECK: [[PRIVATE_VAR_REF:%.+]] = getelementptr inbounds [[PRIVATES_MAIN_TY]], [[PRIVATES_MAIN_TY]]* [[PRIVATES]], i{{.+}} 0, i{{.+}} 1
271 // CHECK: call void @_ZN1SIdED1Ev([[S_DOUBLE_TY]]* {{[^,]*}} [[PRIVATE_VAR_REF]])
272 // CHECK: getelementptr inbounds [2 x [[S_DOUBLE_TY]]], [2 x [[S_DOUBLE_TY]]]* [[PRIVATE_S_ARR_REF]], i{{.+}} 0, i{{.+}} 0
273 // CHECK: getelementptr inbounds [[S_DOUBLE_TY]], [[S_DOUBLE_TY]]* %{{.+}}, i{{.+}} 2
274 // CHECK: [[PRIVATE_S_ARR_ELEM_REF:%.+]] = getelementptr inbounds [[S_DOUBLE_TY]], [[S_DOUBLE_TY]]* %{{.+}}, i{{.+}} -1
275 // CHECK: call void @_ZN1SIdED1Ev([[S_DOUBLE_TY]]* {{[^,]*}} [[PRIVATE_S_ARR_ELEM_REF]])
276 // CHECK: icmp eq
277 // CHECK: br i1
278 // CHECK: ret i32
279
280 // CHECK: [[TEST:%.+]] = alloca [[S_INT_TY]],
281 // CHECK: [[T_VAR_ADDR:%.+]] = alloca i32,
282 // CHECK: [[VEC_ADDR:%.+]] = alloca [2 x i32],
283 // CHECK: [[S_ARR_ADDR:%.+]] = alloca [2 x [[S_INT_TY]]],
284 // CHECK: [[VAR_ADDR:%.+]] = alloca [[S_INT_TY]],
285
286 // CHECK: call {{.*}} [[S_INT_TY_DEF_CONSTR:@.+]]([[S_INT_TY]]* {{[^,]*}} [[TEST]])
287
288 // Do not store original variables in capture struct.
289 // CHECK-NOT: getelementptr inbounds [[CAP_TMAIN_TY]],
290
291 // Allocate task.
292 // Returns struct kmp_task_t {
293 // [[KMP_TASK_T_TY]] task_data;
294 // [[KMP_TASK_TMAIN_TY]] privates;
295 // };
296 // CHECK: [[RES:%.+]] = call i8* @__kmpc_omp_task_alloc([[LOC]], i32 [[GTID:%.+]], i32 9, i64 256, i64 1, i32 (i32, i8*)* bitcast (i32 (i32, [[KMP_TASK_TMAIN_TY]]*)* [[TASK_ENTRY:@[^ ]+]] to i32 (i32, i8*)*))
297 // CHECK: [[RES_KMP_TASK:%.+]] = bitcast i8* [[RES]] to [[KMP_TASK_TMAIN_TY]]*
298
299 // CHECK: [[TASK:%.+]] = getelementptr inbounds [[KMP_TASK_TMAIN_TY]], [[KMP_TASK_TMAIN_TY]]* [[RES_KMP_TASK]], i{{[0-9]+}} 0, i{{[0-9]+}} 0
300
301 // Initialize kmp_task_t->privates with default values (no init for simple types, default constructors for classes).
302 // CHECK: [[PRIVATES:%.+]] = getelementptr inbounds [[KMP_TASK_TMAIN_TY]], [[KMP_TASK_TMAIN_TY]]* [[RES_KMP_TASK]], i{{[0-9]+}} 0, i{{[0-9]+}} 2
303
304 // Constructors for s_arr and var.
305 // a_arr;
306 // CHECK: [[PRIVATE_S_ARR_REF:%.+]] = getelementptr inbounds [[PRIVATES_TMAIN_TY]], [[PRIVATES_TMAIN_TY]]* [[PRIVATES]], i{{[0-9]+}} 0, i{{[0-9]+}} 2
307 // CHECK: getelementptr inbounds [2 x [[S_INT_TY]]], [2 x [[S_INT_TY]]]* [[PRIVATE_S_ARR_REF]], i{{.+}} 0, i{{.+}} 0
308 // CHECK: getelementptr inbounds [[S_INT_TY]], [[S_INT_TY]]* %{{.+}}, i{{.+}} 2
309 // CHECK: call void [[S_INT_TY_DEF_CONSTR]]([[S_INT_TY]]* {{[^,]*}} [[S_ARR_CUR:%.+]])
310 // CHECK: getelementptr inbounds [[S_INT_TY]], [[S_INT_TY]]* [[S_ARR_CUR]], i{{.+}} 1
311 // CHECK: icmp eq
312 // CHECK: br i1
313
314 // var;
315 // CHECK: [[PRIVATE_VAR_REF:%.+]] = getelementptr inbounds [[PRIVATES_TMAIN_TY]], [[PRIVATES_TMAIN_TY]]* [[PRIVATES]], i{{.+}} 0, i{{.+}} 3
316 // CHECK: call void [[S_INT_TY_DEF_CONSTR]]([[S_INT_TY]]* {{[^,]*}} [[PRIVATE_VAR_REF:%.+]])
317
318 // Provide pointer to destructor function, which will destroy private variables at the end of the task.
319 // CHECK: [[DESTRUCTORS_REF:%.+]] = getelementptr inbounds [[KMP_TASK_T_TY]], [[KMP_TASK_T_TY]]* [[TASK]], i{{.+}} 0, i{{.+}} 3
320 // CHECK: [[DESTRUCTORS_PTR:%.+]] = bitcast %union{{.+}}* [[DESTRUCTORS_REF]] to i32 (i32, i8*)**
321 // CHECK: store i32 (i32, i8*)* bitcast (i32 (i32, [[KMP_TASK_TMAIN_TY]]*)* [[DESTRUCTORS:@.+]] to i32 (i32, i8*)*), i32 (i32, i8*)** [[DESTRUCTORS_PTR]],
322
323 // Start task.
324 // CHECK: call void @__kmpc_taskloop([[LOC]], i32 [[GTID]], i8* [[RES]], i32 1, i64* %{{.+}}, i64* %{{.+}}, i64 %{{.+}}, i32 1, i32 0, i64 0, i8* bitcast (void ([[KMP_TASK_TMAIN_TY]]*, [[KMP_TASK_TMAIN_TY]]*, i32)* [[TMAIN_DUP:@.+]] to i8*))
325
326 // No destructors must be called for private copies of s_arr and var.
327 // CHECK-NOT: getelementptr inbounds [[PRIVATES_TMAIN_TY]], [[PRIVATES_TMAIN_TY]]* [[PRIVATES]], i{{.+}} 0, i{{.+}} 2
328 // CHECK-NOT: getelementptr inbounds [[PRIVATES_TMAIN_TY]], [[PRIVATES_TMAIN_TY]]* [[PRIVATES]], i{{.+}} 0, i{{.+}} 3
329 //
330
331 // CHECK: define internal void [[PRIVATES_MAP_FN:@.+]]([[PRIVATES_TMAIN_TY]]* noalias %0, i32** noalias %1, [2 x i32]** noalias %2, [2 x [[S_INT_TY]]]** noalias %3, [[S_INT_TY]]** noalias %4)
332 // CHECK: [[PRIVATES:%.+]] = load [[PRIVATES_TMAIN_TY]]*, [[PRIVATES_TMAIN_TY]]**
333 // CHECK: [[PRIV_T_VAR:%.+]] = getelementptr inbounds [[PRIVATES_TMAIN_TY]], [[PRIVATES_TMAIN_TY]]* [[PRIVATES]], i32 0, i32 0
334 // CHECK: [[ARG1:%.+]] = load i32**, i32*** %{{.+}},
335 // CHECK: store i32* [[PRIV_T_VAR]], i32** [[ARG1]],
336 // CHECK: [[PRIV_VEC:%.+]] = getelementptr inbounds [[PRIVATES_TMAIN_TY]], [[PRIVATES_TMAIN_TY]]* [[PRIVATES]], i32 0, i32 1
337 // CHECK: [[ARG2:%.+]] = load [2 x i32]**, [2 x i32]*** %{{.+}},
338 // CHECK: store [2 x i32]* [[PRIV_VEC]], [2 x i32]** [[ARG2]],
339 // CHECK: [[PRIV_S_VAR:%.+]] = getelementptr inbounds [[PRIVATES_TMAIN_TY]], [[PRIVATES_TMAIN_TY]]* [[PRIVATES]], i32 0, i32 2
340 // CHECK: [[ARG3:%.+]] = load [2 x [[S_INT_TY]]]**, [2 x [[S_INT_TY]]]*** %{{.+}},
341 // CHECK: store [2 x [[S_INT_TY]]]* [[PRIV_S_VAR]], [2 x [[S_INT_TY]]]** [[ARG3]],
342 // CHECK: [[PRIV_VAR:%.+]] = getelementptr inbounds [[PRIVATES_TMAIN_TY]], [[PRIVATES_TMAIN_TY]]* [[PRIVATES]], i32 0, i32 3
343 // CHECK: [[ARG4:%.+]] = load [[S_INT_TY]]**, [[S_INT_TY]]*** {{.+}},
344 // CHECK: store [[S_INT_TY]]* [[PRIV_VAR]], [[S_INT_TY]]** [[ARG4]],
345 // CHECK: ret void
346
347 // CHECK: define internal i32 [[TASK_ENTRY]](i32 %0, [[KMP_TASK_TMAIN_TY]]* noalias %1)
348
349 // CHECK: alloca i32*,
350 // CHECK-DAG: [[PRIV_T_VAR_ADDR:%.+]] = alloca i32*,
351 // CHECK-DAG: [[PRIV_VEC_ADDR:%.+]] = alloca [2 x i32]*,
352 // CHECK-DAG: [[PRIV_S_ARR_ADDR:%.+]] = alloca [2 x [[S_INT_TY]]]*,
353 // CHECK-DAG: [[PRIV_VAR_ADDR:%.+]] = alloca [[S_INT_TY]]*,
354 // CHECK: store void (i8*, ...)* bitcast (void ([[PRIVATES_TMAIN_TY]]*, i32**, [2 x i32]**, [2 x [[S_INT_TY]]]**, [[S_INT_TY]]**)* [[PRIVATES_MAP_FN]] to void (i8*, ...)*), void (i8*, ...)** [[MAP_FN_ADDR:%.+]],
355 // CHECK: [[MAP_FN:%.+]] = load void (i8*, ...)*, void (i8*, ...)** [[MAP_FN_ADDR]],
356 // CHECK: [[FN:%.+]] = bitcast void (i8*, ...)* [[MAP_FN]] to void (i8*,
357 // CHECK: call void [[FN]](i8* %{{.+}}, i32** [[PRIV_T_VAR_ADDR]], [2 x i32]** [[PRIV_VEC_ADDR]], [2 x [[S_INT_TY]]]** [[PRIV_S_ARR_ADDR]], [[S_INT_TY]]** [[PRIV_VAR_ADDR]])
358 // CHECK: [[PRIV_T_VAR:%.+]] = load i32*, i32** [[PRIV_T_VAR_ADDR]],
359 // CHECK: [[PRIV_VEC:%.+]] = load [2 x i32]*, [2 x i32]** [[PRIV_VEC_ADDR]],
360 // CHECK: [[PRIV_S_ARR:%.+]] = load [2 x [[S_INT_TY]]]*, [2 x [[S_INT_TY]]]** [[PRIV_S_ARR_ADDR]],
361 // CHECK: [[PRIV_VAR:%.+]] = load [[S_INT_TY]]*, [[S_INT_TY]]** [[PRIV_VAR_ADDR]],
362
363 // Privates actually are used.
364 // CHECK-DAG: [[PRIV_VAR]]
365 // CHECK-DAG: [[PRIV_T_VAR]]
366 // CHECK-DAG: [[PRIV_S_ARR]]
367 // CHECK-DAG: [[PRIV_VEC]]
368
369 // CHECK: ret
370
371 // CHECK: define internal void [[TMAIN_DUP]]([[KMP_TASK_TMAIN_TY]]* %0, [[KMP_TASK_TMAIN_TY]]* %1, i32 %2)
372 // CHECK: getelementptr inbounds [[KMP_TASK_TMAIN_TY]], [[KMP_TASK_TMAIN_TY]]* %{{.+}}, i32 0, i32 2
373 // CHECK: getelementptr inbounds [[PRIVATES_TMAIN_TY]], [[PRIVATES_TMAIN_TY]]* %{{.+}}, i32 0, i32 2
374 // CHECK: getelementptr inbounds [2 x [[S_INT_TY]]], [2 x [[S_INT_TY]]]* %{{.+}}, i32 0, i32 0
375 // CHECK: getelementptr inbounds [[S_INT_TY]], [[S_INT_TY]]* %{{.+}}, i64 2
376 // CHECK: br label %
377
378 // CHECK: phi [[S_INT_TY]]*
379 // CHECK: call {{.*}} [[S_INT_TY_DEF_CONSTR]]([[S_INT_TY]]*
380 // CHECK: getelementptr inbounds [[S_INT_TY]], [[S_INT_TY]]* %{{.+}}, i64 1
381 // CHECK: icmp eq [[S_INT_TY]]* %
382 // CHECK: br i1 %
383
384 // CHECK: getelementptr inbounds [[PRIVATES_TMAIN_TY]], [[PRIVATES_TMAIN_TY]]* %{{.+}}, i32 0, i32 3
385 // CHECK: call {{.*}} [[S_INT_TY_DEF_CONSTR]]([[S_INT_TY]]*
386 // CHECK: ret void
387
388 // CHECK: define internal i32 [[DESTRUCTORS]](i32 %0, [[KMP_TASK_TMAIN_TY]]* noalias %1)
389 // CHECK: [[PRIVATES:%.+]] = getelementptr inbounds [[KMP_TASK_TMAIN_TY]], [[KMP_TASK_TMAIN_TY]]* [[RES_KMP_TASK:%.+]], i{{[0-9]+}} 0, i{{[0-9]+}} 2
390 // CHECK: [[PRIVATE_S_ARR_REF:%.+]] = getelementptr inbounds [[PRIVATES_TMAIN_TY]], [[PRIVATES_TMAIN_TY]]* [[PRIVATES]], i{{.+}} 0, i{{.+}} 2
391 // CHECK: [[PRIVATE_VAR_REF:%.+]] = getelementptr inbounds [[PRIVATES_TMAIN_TY]], [[PRIVATES_TMAIN_TY]]* [[PRIVATES]], i{{.+}} 0, i{{.+}} 3
392 // CHECK: call void @_ZN1SIiED1Ev([[S_INT_TY]]* {{[^,]*}} [[PRIVATE_VAR_REF]])
393 // CHECK: getelementptr inbounds [2 x [[S_INT_TY]]], [2 x [[S_INT_TY]]]* [[PRIVATE_S_ARR_REF]], i{{.+}} 0, i{{.+}} 0
394 // CHECK: getelementptr inbounds [[S_INT_TY]], [[S_INT_TY]]* %{{.+}}, i{{.+}} 2
395 // CHECK: [[PRIVATE_S_ARR_ELEM_REF:%.+]] = getelementptr inbounds [[S_INT_TY]], [[S_INT_TY]]* %{{.+}}, i{{.+}} -1
396 // CHECK: call void @_ZN1SIiED1Ev([[S_INT_TY]]* {{[^,]*}} [[PRIVATE_S_ARR_ELEM_REF]])
397 // CHECK: icmp eq
398 // CHECK: br i1
399 // CHECK: ret i32
400
401 #endif
402 #else
403 // ARRAY-LABEL: array_func
404 struct St {
405 int a, b;
StSt406 St() : a(0), b(0) {}
operator =St407 St &operator=(const St &) { return *this; };
~StSt408 ~St() {}
409 };
410
array_func(int n,float a[n],St s[2])411 void array_func(int n, float a[n], St s[2]) {
412 // ARRAY: call i8* @__kmpc_omp_task_alloc(
413 // ARRAY: call void @__kmpc_taskloop(
414 // ARRAY: store float** %{{.+}}, float*** %{{.+}},
415 // ARRAY: store %struct.St** %{{.+}}, %struct.St*** %{{.+}},
416 #pragma omp parallel master taskloop simd private(a, s)
417 for (int i = 0; i < 10; ++i)
418 ;
419 }
420 #endif
421
422