1 // RUN: %clang_cc1 -DLAMBDA -verify -fopenmp -x c++ -std=c++11 -triple powerpc64le-unknown-unknown -fopenmp-targets=powerpc64le-ibm-linux-gnu -emit-llvm %s -o - -Wno-openmp-mapping | FileCheck %s --check-prefix LAMBDA --check-prefix LAMBDA-64
2 // RUN: %clang_cc1 -DLAMBDA -fopenmp -x c++ -std=c++11 -triple powerpc64le-unknown-unknown -fopenmp-targets=powerpc64le-ibm-linux-gnu -emit-pch -o %t %s -Wno-openmp-mapping
3 // RUN: %clang_cc1 -DLAMBDA -fopenmp -x c++ -triple powerpc64le-unknown-unknown -fopenmp-targets=powerpc64le-ibm-linux-gnu -std=c++11 -include-pch %t -verify %s -emit-llvm -o - -Wno-openmp-mapping | FileCheck %s --check-prefix LAMBDA --check-prefix LAMBDA-64
4 // RUN: %clang_cc1 -DLAMBDA -verify -fopenmp -x c++ -std=c++11 -triple i386-unknown-unknown -fopenmp-targets=i386-pc-linux-gnu -emit-llvm %s -o - -Wno-openmp-mapping | FileCheck %s --check-prefix LAMBDA --check-prefix LAMBDA-32
5 // RUN: %clang_cc1 -DLAMBDA -fopenmp -x c++ -std=c++11 -triple i386-unknown-unknown -fopenmp-targets=i386-pc-linux-gnu -emit-pch -o %t %s -Wno-openmp-mapping
6 // RUN: %clang_cc1 -DLAMBDA -fopenmp -x c++ -std=c++11 -triple i386-unknown-unknown -fopenmp-targets=i386-pc-linux-gnu -std=c++11 -include-pch %t -verify %s -emit-llvm -o - -Wno-openmp-mapping | FileCheck %s --check-prefix LAMBDA --check-prefix LAMBDA-32
7 
8 // RUN: %clang_cc1 -DLAMBDA -verify -fopenmp-simd -x c++ -std=c++11 -triple powerpc64le-unknown-unknown -fopenmp-targets=powerpc64le-ibm-linux-gnu -emit-llvm %s -o - -Wno-openmp-mapping | FileCheck --check-prefix SIMD-ONLY0 %s
9 // RUN: %clang_cc1 -DLAMBDA -fopenmp-simd -x c++ -std=c++11 -triple powerpc64le-unknown-unknown -fopenmp-targets=powerpc64le-ibm-linux-gnu -emit-pch -o %t %s -Wno-openmp-mapping
10 // RUN: %clang_cc1 -DLAMBDA -fopenmp-simd -x c++ -triple powerpc64le-unknown-unknown -fopenmp-targets=powerpc64le-ibm-linux-gnu -std=c++11 -include-pch %t -verify %s -emit-llvm -o - -Wno-openmp-mapping | FileCheck --check-prefix SIMD-ONLY0 %s
11 // RUN: %clang_cc1 -DLAMBDA -verify -fopenmp-simd -x c++ -std=c++11 -triple i386-unknown-unknown -fopenmp-targets=i386-pc-linux-gnu -emit-llvm %s -o - -Wno-openmp-mapping | FileCheck --check-prefix SIMD-ONLY0 %s
12 // RUN: %clang_cc1 -DLAMBDA -fopenmp-simd -x c++ -std=c++11 -triple i386-unknown-unknown -fopenmp-targets=i386-pc-linux-gnu -emit-pch -o %t %s -Wno-openmp-mapping
13 // RUN: %clang_cc1 -DLAMBDA -fopenmp-simd -x c++ -std=c++11 -triple i386-unknown-unknown -fopenmp-targets=i386-pc-linux-gnu -std=c++11 -include-pch %t -verify %s -emit-llvm -o - -Wno-openmp-mapping | FileCheck --check-prefix SIMD-ONLY0 %s
14 // SIMD-ONLY0-NOT: {{__kmpc|__tgt}}
15 
16 // RUN: %clang_cc1  -verify -fopenmp -x c++ -std=c++11 -triple powerpc64le-unknown-unknown -fopenmp-targets=powerpc64le-ibm-linux-gnu -emit-llvm %s -o - -Wno-openmp-mapping | FileCheck %s --check-prefix CHECK --check-prefix CHECK-64
17 // RUN: %clang_cc1  -fopenmp -x c++ -std=c++11 -triple powerpc64le-unknown-unknown -fopenmp-targets=powerpc64le-ibm-linux-gnu -emit-pch -o %t %s -Wno-openmp-mapping
18 // RUN: %clang_cc1  -fopenmp -x c++ -triple powerpc64le-unknown-unknown -fopenmp-targets=powerpc64le-ibm-linux-gnu -std=c++11 -include-pch %t -verify %s -emit-llvm -o - -Wno-openmp-mapping| FileCheck %s --check-prefix CHECK --check-prefix CHECK-64
19 // RUN: %clang_cc1  -verify -fopenmp -x c++ -std=c++11 -triple i386-unknown-unknown -fopenmp-targets=i386-pc-linux-gnu -emit-llvm %s -o - -Wno-openmp-mapping | FileCheck %s --check-prefix CHECK --check-prefix CHECK-32
20 // RUN: %clang_cc1  -fopenmp -x c++ -std=c++11 -triple i386-unknown-unknown -fopenmp-targets=i386-pc-linux-gnu -emit-pch -o %t %s -Wno-openmp-mapping
21 // RUN: %clang_cc1  -fopenmp -x c++ -std=c++11 -triple i386-unknown-unknown -fopenmp-targets=i386-pc-linux-gnu -std=c++11 -include-pch %t -verify %s -emit-llvm -o - -Wno-openmp-mapping | FileCheck %s --check-prefix CHECK --check-prefix CHECK-32
22 
23 // RUN: %clang_cc1  -verify -fopenmp-simd -x c++ -std=c++11 -triple powerpc64le-unknown-unknown -fopenmp-targets=powerpc64le-ibm-linux-gnu -emit-llvm %s -o - -Wno-openmp-mapping | FileCheck --check-prefix SIMD-ONLY1 %s
24 // RUN: %clang_cc1  -fopenmp-simd -x c++ -std=c++11 -triple powerpc64le-unknown-unknown -fopenmp-targets=powerpc64le-ibm-linux-gnu -emit-pch -o %t %s -Wno-openmp-mapping
25 // RUN: %clang_cc1  -fopenmp-simd -x c++ -triple powerpc64le-unknown-unknown -fopenmp-targets=powerpc64le-ibm-linux-gnu -std=c++11 -include-pch %t -verify %s -emit-llvm -o - -Wno-openmp-mapping | FileCheck --check-prefix SIMD-ONLY1 %s
26 // RUN: %clang_cc1  -verify -fopenmp-simd -x c++ -std=c++11 -triple i386-unknown-unknown -fopenmp-targets=i386-pc-linux-gnu -emit-llvm %s -o - -Wno-openmp-mapping | FileCheck --check-prefix SIMD-ONLY1 %s
27 // RUN: %clang_cc1  -fopenmp-simd -x c++ -std=c++11 -triple i386-unknown-unknown -fopenmp-targets=i386-pc-linux-gnu -emit-pch -o %t %s -Wno-openmp-mapping
28 // RUN: %clang_cc1  -fopenmp-simd -x c++ -std=c++11 -triple i386-unknown-unknown -fopenmp-targets=i386-pc-linux-gnu -std=c++11 -include-pch %t -verify %s -emit-llvm -o - -Wno-openmp-mapping | FileCheck --check-prefix SIMD-ONLY1 %s
29 // SIMD-ONLY1-NOT: {{__kmpc|__tgt}}
30 // expected-no-diagnostics
31 #ifndef HEADER
32 #define HEADER
33 
34 template <class T>
35 struct S {
36   T f;
SS37   S(T a) : f(a) {}
SS38   S() : f() {}
operator TS39   operator T() { return T(); }
~SS40   ~S() {}
41 };
42 
43 // CHECK: [[S_FLOAT_TY:%.+]] = type { float }
44 // CHECK: [[S_INT_TY:%.+]] = type { i{{[0-9]+}} }
45 template <typename T>
tmain()46 T tmain() {
47   S<T> test;
48   T t_var = T();
49   T vec[] = {1, 2};
50   S<T> s_arr[] = {1, 2};
51   S<T> &var = test;
52   #pragma omp target
53   #pragma omp teams distribute simd lastprivate(t_var, vec, s_arr, s_arr, var, var)
54   for (int i = 0; i < 2; ++i) {
55     vec[i] = t_var;
56     s_arr[i] = var;
57   }
58   return T();
59 }
60 
main()61 int main() {
62   static int svar;
63   volatile double g;
64   volatile double &g1 = g;
65 
66   #ifdef LAMBDA
67   // LAMBDA-LABEL: @main
68   // LAMBDA: call{{.*}} void [[OUTER_LAMBDA:@.+]](
69   [&]() {
70     static float sfvar;
71     // LAMBDA: define{{.*}} internal{{.*}} void [[OUTER_LAMBDA]](
72     // LAMBDA: call i{{[0-9]+}} @__tgt_target_teams_mapper(%struct.ident_t* @{{.+}},
73     // LAMBDA: call void [[OFFLOADING_FUN:@.+]](
74 
75     // LAMBDA: define{{.+}} void [[OFFLOADING_FUN]](
76     // LAMBDA: call {{.*}}void {{.+}} @__kmpc_fork_teams({{.+}}, i32 4, {{.+}}* [[OMP_OUTLINED:@.+]] to {{.+}})
77     #pragma omp target
78     #pragma omp teams distribute simd lastprivate(g, g1, svar, sfvar)
79     for (int i = 0; i < 2; ++i) {
80       // LAMBDA: define{{.*}} internal{{.*}} void [[OMP_OUTLINED]](i32* noalias %{{.+}}, i32* noalias %{{.+}}, double*{{.+}} [[G_IN:%.+]], double*{{.+}} [[G1_IN:%.+]], i{{[0-9]+}}*{{.+}} [[SVAR_IN:%.+]], float*{{.+}} [[SFVAR_IN:%.+]])
81       // LAMBDA: [[G_PRIVATE_ADDR:%.+]] = alloca double*,
82       // LAMBDA: [[G1_PRIVATE_ADDR:%.+]] = alloca double*,
83       // LAMBDA: [[SVAR_PRIVATE_ADDR:%.+]] = alloca i{{[0-9]+}}*,
84       // LAMBDA: [[SFVAR_PRIVATE_ADDR:%.+]] = alloca float*,
85       // loop variables
86       // LAMBDA: {{.+}} = alloca i{{[0-9]+}},
87       // LAMBDA: {{.+}} = alloca i{{[0-9]+}},
88       // LAMBDA: {{.+}} = alloca i{{[0-9]+}},
89       // LAMBDA: {{.+}} = alloca i{{[0-9]+}},
90       // LAMBDA: {{.+}} = alloca i{{[0-9]+}},
91       // LAMBDA: [[OMP_IS_LAST:%.+]] = alloca i{{[0-9]+}},
92       // LAMBDA: [[G_PRIVATE:%.+]] = alloca double,
93       // LAMBDA: [[G1_PRIVATE:%.+]] = alloca double,
94       // LAMBDA: [[TMP_G1_PRIVATE:%.+]] = alloca double*,
95       // LAMBDA: [[SVAR_PRIVATE:%.+]] = alloca i{{[0-9]+}},
96       // LAMBDA: [[SFVAR_PRIVATE:%.+]] = alloca float,
97       // LAMBDA: store double* [[G_IN]], double** [[G_PRIVATE_ADDR]],
98       // LAMBDA: store double* [[G1_IN]], double** [[G1_PRIVATE_ADDR]],
99       // LAMBDA: store i{{[0-9]+}}* [[SVAR_IN]], i{{[0-9]+}}** [[SVAR_PRIVATE_ADDR]],
100       // LAMBDA: store float* [[SFVAR_IN]], float** [[SFVAR_PRIVATE_ADDR]],
101 
102       // init private variables
103       // LAMBDA: [[G_IN_REF:%.+]] = load double*, double** [[G_PRIVATE_ADDR]],
104       // LAMBDA: [[G1_IN_REF:%.+]] = load double*, double** [[G1_PRIVATE_ADDR]],
105       // LAMBDA: [[SVAR_IN_REF:%.+]] = load i{{[0-9]+}}*, i{{[0-9]+}}** [[SVAR_PRIVATE_ADDR]],
106       // LAMBDA: [[SFVAR_IN_REF:%.+]] = load float*, float** [[SFVAR_PRIVATE_ADDR]],
107       // LAMBDA: store double* [[G1_IN_REF]], double** [[TMP:%.+]],
108       // LAMBDA: [[G1_IN_REF:%.+]] = load double*, double** [[TMP]],
109       // LAMBDA: store double* [[G1_PRIVATE]], double** [[TMP_G1_PRIVATE]],
110       g = 1;
111       g1 = 1;
112       svar = 3;
113       sfvar = 4.0;
114       // LAMBDA: call {{.*}}void @__kmpc_for_static_init_4(
115       // LAMBDA: store double 1.0{{.+}}, double* [[G_PRIVATE]],
116       // LAMBDA: [[TMP_G1_REF:%.+]] = load double*, double** [[TMP_G1_PRIVATE]],
117       // LAMBDA: store{{.+}} double 1.0{{.+}}, double* [[TMP_G1_REF]],
118       // LAMBDA: store i{{[0-9]+}} 3, i{{[0-9]+}}* [[SVAR_PRIVATE]],
119       // LAMBDA: store float 4.0{{.+}}, float* [[SFVAR_PRIVATE]],
120       // LAMBDA: [[G_PRIVATE_ADDR_REF:%.+]] = getelementptr inbounds %{{.+}}, %{{.+}}* [[ARG:%.+]], i{{[0-9]+}} 0, i{{[0-9]+}} 0
121       // LAMBDA: store double* [[G_PRIVATE]], double** [[G_PRIVATE_ADDR_REF]],
122       // LAMBDA: [[TMP_PRIVATE_ADDR_REF:%.+]] = getelementptr inbounds %{{.+}}, %{{.+}}* [[ARG:%.+]], i{{[0-9]+}} 0, i{{[0-9]+}} 1
123       // LAMBDA: [[G1_PRIVATE_ADDR_FROM_TMP:%.+]] = load double*, double** [[TMP_G1_PRIVATE]],
124       // LAMBDA: store double* [[G1_PRIVATE_ADDR_FROM_TMP]], double** [[TMP_PRIVATE_ADDR_REF]],
125       // LAMBDA: [[SVAR_PRIVATE_ADDR_REF:%.+]] = getelementptr inbounds %{{.+}}, %{{.+}}* [[ARG:%.+]], i{{[0-9]+}} 0, i{{[0-9]+}} 2
126       // LAMBDA: store i{{[0-9]+}}* [[SVAR_PRIVATE]], i{{[0-9]+}}** [[SVAR_PRIVATE_ADDR_REF]]
127       // LAMBDA: [[SFVAR_PRIVATE_ADDR_REF:%.+]] = getelementptr inbounds %{{.+}}, %{{.+}}* [[ARG:%.+]], i{{[0-9]+}} 0, i{{[0-9]+}} 3
128       // LAMBDA: store float* [[SFVAR_PRIVATE]], float** [[SFVAR_PRIVATE_ADDR_REF]]
129       // LAMBDA: call{{.*}} void [[INNER_LAMBDA:@.+]](%{{.+}}* {{[^,]*}} [[ARG]])
130       // LAMBDA: call {{.*}}void @__kmpc_for_static_fini(
131       // LAMBDA: store i32 2, i32* %
132       // LAMBDA: [[OMP_IS_LAST_VAL:%.+]] = load i{{[0-9]+}}, i{{[0-9]+}}* [[OMP_IS_LAST]],
133       // LAMBDA: [[IS_LAST_IT:%.+]] = icmp ne i{{[0-9]+}} [[OMP_IS_LAST_VAL]], 0
134       // LAMBDA: br i1 [[IS_LAST_IT]], label %[[OMP_LASTPRIV_BLOCK:.+]], label %[[OMP_LASTPRIV_DONE:.+]]
135 
136       // LAMBDA: [[OMP_LASTPRIV_BLOCK]]:
137       // LAMBDA: [[G_PRIV_VAL:%.+]] = load double, double* [[G_PRIVATE]],
138       // LAMBDA: store{{.*}} double [[G_PRIV_VAL]], double* [[G_IN_REF]],
139       // LAMBDA: [[TMP_G1_PRIV_REF:%.+]] = load double*, double** [[TMP_G1_PRIVATE]],
140       // LAMBDA: [[TMP_G1_PRIV_VAL:%.+]] = load double, double* [[TMP_G1_PRIV_REF]],
141       // LAMBDA: store{{.*}} double [[TMP_G1_PRIV_VAL]], double* [[G1_IN_REF]],
142 
143       // LAMBDA: [[SVAR_PRIV_VAL:%.+]] = load i{{[0-9]+}}, i{{[0-9]+}}* [[SVAR_PRIVATE]],
144       // LAMBDA: store i{{[0-9]+}} [[SVAR_PRIV_VAL]], i{{[0-9]+}}* [[SVAR_IN_REF]],
145       // LAMBDA: [[SFVAR_PRIV_VAL:%.+]] = load float, float* [[SFVAR_PRIVATE]],
146       // LAMBDA: store float [[SFVAR_PRIV_VAL]], float* [[SFVAR_IN_REF]],
147       // LAMBDA: br label %[[OMP_LASTPRIV_DONE]]
148       // LAMBDA: [[OMP_LASTPRIV_DONE]]:
149       // LAMBDA: ret
150       [&]() {
151         // LAMBDA: define {{.+}} void [[INNER_LAMBDA]](%{{.+}}* {{[^,]*}} [[ARG_PTR:%.+]])
152         // LAMBDA: store %{{.+}}* [[ARG_PTR]], %{{.+}}** [[ARG_PTR_REF:%.+]],
153         g = 2;
154         g1 = 2;
155         svar = 4;
156         sfvar = 8.0;
157         // LAMBDA: [[ARG_PTR:%.+]] = load %{{.+}}*, %{{.+}}** [[ARG_PTR_REF]]
158         // LAMBDA: [[G_PTR_REF:%.+]] = getelementptr inbounds %{{.+}}, %{{.+}}* [[ARG_PTR]], i{{[0-9]+}} 0, i{{[0-9]+}} 0
159         // LAMBDA: [[G_REF:%.+]] = load double*, double** [[G_PTR_REF]]
160         // LAMBDA: store double 2.0{{.+}}, double* [[G_REF]]
161 
162         // LAMBDA: [[TMP_PTR_REF:%.+]] = getelementptr inbounds %{{.+}}, %{{.+}}* [[ARG_PTR]], i{{[0-9]+}} 0, i{{[0-9]+}} 1
163         // LAMBDA: [[G1_REF:%.+]] = load double*, double** [[TMP_PTR_REF]]
164         // LAMBDA: store double 2.0{{.+}}, double* [[G1_REF]],
165         // LAMBDA: [[SVAR_PTR_REF:%.+]] = getelementptr inbounds %{{.+}}, %{{.+}}* [[ARG_PTR]], i{{[0-9]+}} 0, i{{[0-9]+}} 2
166         // LAMBDA: [[SVAR_REF:%.+]] = load i{{[0-9]+}}*, i{{[0-9]+}}** [[SVAR_PTR_REF]]
167         // LAMBDA: store i{{[0-9]+}} 4, i{{[0-9]+}}* [[SVAR_REF]]
168         // LAMBDA: [[SFVAR_PTR_REF:%.+]] = getelementptr inbounds %{{.+}}, %{{.+}}* [[ARG_PTR]], i{{[0-9]+}} 0, i{{[0-9]+}} 3
169         // LAMBDA: [[SFVAR_REF:%.+]] = load float*, float** [[SFVAR_PTR_REF]]
170         // LAMBDA: store float 8.0{{.+}}, float* [[SFVAR_REF]]
171       }();
172     }
173   }();
174   return 0;
175   #else
176   S<float> test;
177   int t_var = 0;
178   int vec[] = {1, 2};
179   S<float> s_arr[] = {1, 2};
180   S<float> &var = test;
181 
182   #pragma omp target
183   #pragma omp teams distribute simd lastprivate(t_var, vec, s_arr, s_arr, var, var, svar)
184   for (int i = 0; i < 2; ++i) {
185     vec[i] = t_var;
186     s_arr[i] = var;
187   }
188   int i;
189 
190   return tmain<int>();
191   #endif
192 }
193 
194 // CHECK: define{{.*}} i{{[0-9]+}} @main()
195 // CHECK: [[TEST:%.+]] = alloca [[S_FLOAT_TY]],
196 // CHECK: call {{.*}} [[S_FLOAT_TY_DEF_CONSTR:@.+]]([[S_FLOAT_TY]]* {{[^,]*}} [[TEST]])
197 // CHECK: call i{{[0-9]+}} @__tgt_target_teams_mapper(%struct.ident_t* @{{.+}},
198 // CHECK: call void [[OFFLOAD_FUN:@.+]](i{{[0-9]+}} {{.+}}, [2 x i{{[0-9]+}}]* {{.+}}, [2 x [[S_FLOAT_TY]]]* {{.+}}, [[S_FLOAT_TY]]* {{.+}}, i{{[0-9]+}} {{.+}})
199 // CHECK: ret
200 
201 // CHECK: define{{.+}} [[OFFLOAD_FUN]](i{{[0-9]+}} {{.+}}, [2 x i{{[0-9]+}}]*{{.+}} {{.+}}, [2 x [[S_FLOAT_TY]]]*{{.+}} {{.+}}, [[S_FLOAT_TY]]*{{.+}} {{.+}}, i{{[0-9]+}} {{.+}})
202 // CHECK: call void (%{{.+}}*, i{{[0-9]+}}, void (i{{[0-9]+}}*, i{{[0-9]+}}*, ...)*, ...) @__kmpc_fork_teams(
203 // CHECK: ret
204 //
205 // CHECK: define internal void [[OMP_OUTLINED:@.+]](i{{[0-9]+}}* noalias [[GTID_ADDR:%.+]], i{{[0-9]+}}* noalias %{{.+}}, [2 x i{{[0-9]+}}]*{{.+}} [[VEC_IN:%.+]], i{{[0-9]+}}*{{.+}} [[T_VAR_IN:%.+]], [2 x [[S_FLOAT_TY]]]*{{.+}} [[S_ARR_IN:%.+]], [[S_FLOAT_TY]]*{{.+}} [[VAR_IN:%.+]], i{{[0-9]+}}*{{.*}} [[S_VAR_IN:%.+]])
206 // CHECK: {{.+}} = alloca i{{[0-9]+}}*,
207 // CHECK: {{.+}} = alloca i{{[0-9]+}}*,
208 // CHECK: [[VEC_ADDR:%.+]] = alloca [2 x i{{[0-9]+}}]*,
209 // CHECK: [[T_VAR_ADDR:%.+]] = alloca i{{[0-9]+}}*,
210 // CHECK: [[S_ARR_ADDR:%.+]] = alloca [2 x [[S_FLOAT_TY]]]*,
211 // CHECK: [[VAR_ADDR:%.+]] = alloca [[S_FLOAT_TY]]*,
212 // CHECK: [[SVAR_ADDR:%.+]] = alloca i{{[0-9]+}}*,
213 // skip loop variables
214 // CHECK: {{.+}} = alloca i{{[0-9]+}},
215 // CHECK: {{.+}} = alloca i{{[0-9]+}},
216 // CHECK: {{.+}} = alloca i{{[0-9]+}},
217 // CHECK: {{.+}} = alloca i{{[0-9]+}},
218 // CHECK: {{.+}} = alloca i{{[0-9]+}},
219 // CHECK: [[OMP_IS_LAST:%.+]] = alloca i{{[0-9]+}},
220 // CHECK: [[T_VAR_PRIV:%.+]] = alloca i{{[0-9]+}},
221 // CHECK: [[VEC_PRIV:%.+]] = alloca [2 x i{{[0-9]+}}],
222 // CHECK: [[S_ARR_PRIV:%.+]] = alloca [2 x [[S_FLOAT_TY]]],
223 // CHECK: [[VAR_PRIV:%.+]] = alloca [[S_FLOAT_TY]],
224 // CHECK: [[TMP_PRIV:%.+]] = alloca [[S_FLOAT_TY]]*,
225 // CHECK: [[S_VAR_PRIV:%.+]] = alloca i{{[0-9]+}},
226 
227 // copy from parameters to local address variables
228 // CHECK: store [2 x i{{[0-9]+}}]* [[VEC_IN]], [2 x i{{[0-9]+}}]** [[VEC_ADDR]],
229 // CHECK: store i{{[0-9]+}}* [[T_VAR_IN]], i{{[0-9]+}}** [[T_VAR_ADDR]],
230 // CHECK: store [2 x [[S_FLOAT_TY]]]* [[S_ARR_IN]], [2 x [[S_FLOAT_TY]]]** [[S_ARR_ADDR]],
231 // CHECK: store [[S_FLOAT_TY]]* [[VAR_IN]], [[S_FLOAT_TY]]** [[VAR_ADDR]],
232 // CHECK: store i{{[0-9]+}}* [[S_VAR_IN]], i{{[0-9]+}}** [[SVAR_ADDR]],
233 
234 // load content of local address variables
235 // CHECK: [[VEC_ADDR_REF:%.+]] = load [2 x i{{[0-9]+}}]*, [2 x i{{[0-9]+}}]** [[VEC_ADDR]],
236 // CHECK: [[T_VAR_ADDR_REF:%.+]] = load i{{[0-9]+}}*, i{{[0-9]+}}** [[T_VAR_ADDR]],
237 // CHECK: [[S_ARR_ADDR_REF:%.+]] = load [2 x [[S_FLOAT_TY]]]*, [2 x [[S_FLOAT_TY]]]** [[S_ARR_ADDR]],
238 // CHECK: [[VAR_ADDR_REF:%.+]] = load [[S_FLOAT_TY]]*, [[S_FLOAT_TY]]** [[VAR_ADDR]],
239 // CHECK: [[SVAR_ADDR_REF:%.+]] = load i{{[0-9]+}}*, i{{[0-9]+}}** [[SVAR_ADDR]],
240 // CHECK: store [[S_FLOAT_TY]]* [[VAR_ADDR_REF]], [[S_FLOAT_TY]]** [[TMP:%.+]],
241 // CHECK: store i{{[0-9]+}} 0, i{{[0-9]+}}* [[OMP_IS_LAST]],
242 // CHECK:  [[VAR_ADDR_REF:%.+]] = load {{.+}}, {{.+}} [[TMP]],
243 // the distribute loop
244 // CHECK: call void @__kmpc_for_static_init_4(
245 // assignment: vec[i] = t_var;
246 // CHECK: [[T_VAR_PRIV_VAL:%.+]] = load i{{[0-9]+}}, i{{[0-9]+}}* [[T_VAR_PRIV]],
247 // CHECK: [[VEC_PTR:%.+]] = getelementptr inbounds [2 x i{{[0-9]+}}], [2 x i{{[0-9]+}}]* [[VEC_PRIV]], i{{[0-9]+}} 0, i{{[0-9]+}} {{.+}}
248 // CHECK:  store i{{[0-9]+}} [[T_VAR_PRIV_VAL]], i{{[0-9]+}}* [[VEC_PTR]],
249 
250 // assignment: s_arr[i] = var;
251 // CHECK-DAG: [[S_ARR_PTR:%.+]] = getelementptr inbounds [2 x [[S_FLOAT_TY]]], [2 x [[S_FLOAT_TY]]]* [[S_ARR_PRIV]],
252 // CHECK-DAG: [[TMP_VAL:%.+]] = load [[S_FLOAT_TY]]*, [[S_FLOAT_TY]]** [[TMP_PRIV]],
253 // CHECK-DAG: [[S_ARR_PTR_BCAST:%.+]] = bitcast [[S_FLOAT_TY]]* [[S_ARR_PTR]] to i8*
254 // CHECK-DAG: [[TMP_VAL_BCAST:%.+]] = bitcast [[S_FLOAT_TY]]* [[TMP_VAL]] to i8*
255 // CHECK: call void @llvm.memcpy.{{.+}}(i8* align {{[0-9]+}} [[S_ARR_PTR_BCAST]], i8* align {{[0-9]+}} [[TMP_VAL_BCAST]],
256 // CHECK: call void @__kmpc_for_static_fini(
257 
258 // lastprivates
259 // CHECK: [[OMP_IS_LAST_VAL:%.+]] = load i{{[0-9]+}}, i{{[0-9]+}}* [[OMP_IS_LAST]],
260 // CHECK: [[IS_LAST_IT:%.+]] = icmp ne i{{[0-9]+}} [[OMP_IS_LAST_VAL]], 0
261 // CHECK: br i1 [[IS_LAST_IT]], label %[[OMP_LASTPRIV_BLOCK:.+]], label %[[OMP_LASTPRIV_DONE:.+]]
262 
263 // CHECK: [[OMP_LASTPRIV_BLOCK]]:
264 // CHECK: [[T_VAR_VAL:%.+]] = load i{{[0-9]+}}, i{{[0-9]+}}* [[T_VAR_PRIV]],
265 // CHECK: store i{{[0-9]+}} [[T_VAR_VAL]], i{{[0-9]+}}* [[T_VAR_ADDR_REF]],
266 // CHECK: [[BCAST_VEC_ADDR_REF:%.+]] = bitcast [2 x i{{[0-9]+}}]* [[VEC_ADDR_REF]] to i8*
267 // CHECK: [[BCAST_VEC_PRIV:%.+]] = bitcast [2 x i{{[0-9]+}}]* [[VEC_PRIV]] to i8*
268 // CHECK: call void @llvm.memcpy.{{.+}}(i8* align {{[0-9]+}} [[BCAST_VEC_ADDR_REF]], i8* align {{[0-9]+}} [[BCAST_VEC_PRIV]],
269 // CHECK: [[S_ARR_BEGIN:%.+]] = getelementptr inbounds [2 x [[S_FLOAT_TY]]], [2 x [[S_FLOAT_TY]]]* [[S_ARR_ADDR_REF]], i{{[0-9]+}} 0, i{{[0-9]+}} 0
270 // CHECK: [[S_ARR_PRIV_BCAST:%.+]] = bitcast [2 x [[S_FLOAT_TY]]]* [[S_ARR_PRIV]] to [[S_FLOAT_TY]]*
271 // CHECK: [[S_ARR_BEGIN_GEP:%.+]] = getelementptr [[S_FLOAT_TY]], [[S_FLOAT_TY]]* [[S_ARR_BEGIN]], i{{[0-9]+}} 2
272 // CHECK: [[S_ARR_IS_EMPTY:%.+]] = icmp eq [[S_FLOAT_TY]]* [[S_ARR_BEGIN]], [[S_ARR_BEGIN_GEP]]
273 // CHECK: br i1 [[S_ARR_IS_EMPTY]], label %[[S_ARR_COPY_DONE:.+]], label %[[S_ARR_COPY_BLOCK:.+]]
274 // CHECK: [[S_ARR_COPY_BLOCK]]:
275 // CHECK: [[S_ARR_SRC_EL:%.+]] = phi [[S_FLOAT_TY]]*{{.+}}
276 // CHECK: [[S_ARR_DST_EL:%.+]] = phi [[S_FLOAT_TY]]*{{.+}}
277 // CHECK: [[S_ARR_DST_BCAST:%.+]] = bitcast [[S_FLOAT_TY]]* [[S_ARR_DST_EL]] to i8*
278 // CHECK: [[S_ARR_SRC_BCAST:%.+]] = bitcast [[S_FLOAT_TY]]* [[S_ARR_SRC_EL]] to i8*
279 // CHECK: call void @llvm.memcpy.{{.+}}(i8* align {{[0-9]+}} [[S_ARR_DST_BCAST]], i8* align {{[0-9]+}} [[S_ARR_SRC_BCAST]]{{.+}})
280 // CHECK: [[S_ARR_DST_NEXT:%.+]] = getelementptr [[S_FLOAT_TY]], [[S_FLOAT_TY]]* [[S_ARR_DST_EL]], i{{[0-9]+}} 1
281 // CHECK: [[S_ARR_SRC_NEXT:%.+]] = getelementptr{{.+}}
282 // CHECK: [[CPY_IS_FINISHED:%.+]] = icmp eq [[S_FLOAT_TY]]* [[S_ARR_DST_NEXT]], [[S_ARR_BEGIN_GEP]]
283 // CHECK: br i1 [[CPY_IS_FINISHED]], label %[[S_ARR_COPY_DONE]], label %[[S_ARR_COPY_BLOCK]]
284 // CHECK: [[S_ARR_COPY_DONE]]:
285 // CHECK: [[TMP_VAL1:%.+]] = load [[S_FLOAT_TY]]*, [[S_FLOAT_TY]]** [[TMP_PRIV]],
286 // CHECK: [[VAR_ADDR_REF_BCAST:%.+]] = bitcast [[S_FLOAT_TY]]* [[VAR_ADDR_REF]] to i8*
287 // CHECK: [[TMP_VAL1_BCAST:%.+]] = bitcast [[S_FLOAT_TY]]* [[TMP_VAL1]] to i8*
288 // CHECK: call void @llvm.memcpy.{{.+}}(i8* align {{[0-9]+}} [[VAR_ADDR_REF_BCAST]], i8* align {{[0-9]+}} [[TMP_VAL1_BCAST]],{{.+}})
289 // CHECK: [[SVAR_VAL:%.+]] = load i{{[0-9]+}}, i{{[0-9]+}}* [[S_VAR_PRIV]],
290 // CHECK: store i{{[0-9]+}} [[SVAR_VAL]], i{{[0-9]+}}* [[SVAR_ADDR_REF]],
291 // CHECK: ret void
292 
293 // template tmain
294 // CHECK: define{{.*}} i{{[0-9]+}} [[TMAIN_INT:@.+]]()
295 // CHECK: [[TEST:%.+]] = alloca [[S_INT_TY]],
296 // CHECK: call {{.*}} [[S_INT_TY_DEF_CONSTR:@.+]]([[S_INT_TY]]* {{[^,]*}} [[TEST]])
297 // CHECK: call i{{[0-9]+}} @__tgt_target_teams_mapper(%struct.ident_t* @{{.+}},
298 // CHECK: call void [[OFFLOAD_FUN_1:@.+]](i{{[0-9]+}} {{.+}}, [2 x i{{[0-9]+}}]* {{.+}}, [2 x [[S_INT_TY]]]* {{.+}}, [[S_INT_TY]]* {{.+}})
299 // CHECK: ret
300 
301 
302 // CHECK: define internal void [[OFFLOAD_FUN_1]](
303 // CHECK: call void (%{{.+}}*, i{{[0-9]+}}, void (i{{[0-9]+}}*, i{{[0-9]+}}*, ...)*, ...) @__kmpc_fork_teams(%{{.+}}* @{{.+}}, i{{[0-9]+}} 4,
304 // CHECK: ret
305 
306 // CHECK: define internal void [[OMP_OUTLINED_1:@.+]](i{{[0-9]+}}* noalias [[GTID_ADDR1:%.+]], i{{[0-9]+}}* noalias %{{.+}}, [2 x i{{[0-9]+}}]*{{.+}} [[VEC_IN1:%.+]], i{{[0-9]+}}*{{.+}} [[T_VAR_IN1:%.+]], [2 x [[S_INT_TY]]]*{{.+}} [[S_ARR_IN1:%.+]], [[S_INT_TY]]*{{.+}} [[VAR_IN1:%.+]])
307 // skip alloca of global_tid and bound_tid
308 // CHECK: {{.+}} = alloca i{{[0-9]+}}*,
309 // CHECK: {{.+}} = alloca i{{[0-9]+}}*,
310 // CHECK: [[VEC_ADDR1:%.+]] = alloca [2 x i{{[0-9]+}}]*,
311 // CHECK: [[T_VAR_ADDR1:%.+]] = alloca i{{[0-9]+}}*,
312 // CHECK: [[S_ARR_ADDR1:%.+]] = alloca [2 x [[S_INT_TY]]]*,
313 // CHECK: [[VAR_ADDR1:%.+]] = alloca [[S_INT_TY]]*,
314 // CHECK: [[TMP1:%.+]] = alloca [[S_INT_TY]]*,
315 // skip loop variables
316 // CHECK: {{.+}} = alloca i{{[0-9]+}},
317 // CHECK: {{.+}} = alloca i{{[0-9]+}},
318 // CHECK: {{.+}} = alloca i{{[0-9]+}},
319 // CHECK: {{.+}} = alloca i{{[0-9]+}},
320 // CHECK: {{.+}} = alloca i{{[0-9]+}},
321 // CHECK: [[OMP_IS_LAST1:%.+]] = alloca i{{[0-9]+}},
322 // CHECK: [[T_VAR_PRIV1:%.+]] = alloca i{{[0-9]+}},
323 // CHECK: [[VEC_PRIV1:%.+]] = alloca [2 x i{{[0-9]+}}],
324 // CHECK: [[S_ARR_PRIV1:%.+]] = alloca [2 x [[S_INT_TY]]],
325 // CHECK: [[VAR_PRIV1:%.+]] = alloca [[S_INT_TY]],
326 // CHECK: [[TMP_PRIV1:%.+]] = alloca [[S_INT_TY]]*,
327 
328 // skip init of bound and global tid
329 // CHECK: store i{{[0-9]+}}* {{.*}},
330 // CHECK: store i{{[0-9]+}}* {{.*}},
331 // copy from parameters to local address variables
332 // CHECK: store [2 x i{{[0-9]+}}]* [[VEC_IN1]], [2 x i{{[0-9]+}}]** [[VEC_ADDR1]],
333 // CHECK: store i{{[0-9]+}}* [[T_VAR_IN1]], i{{[0-9]+}}** [[T_VAR_ADDR1]],
334 // CHECK: store [2 x [[S_INT_TY]]]* [[S_ARR_IN1]], [2 x [[S_INT_TY]]]** [[S_ARR_ADDR1]],
335 // CHECK: store [[S_INT_TY]]* [[VAR_IN1]], [[S_INT_TY]]** [[VAR_ADDR1]],
336 
337 // load content of local address variables
338 // CHECK: [[VEC_ADDR_REF1:%.+]] = load [2 x i{{[0-9]+}}]*, [2 x i{{[0-9]+}}]** [[VEC_ADDR1]],
339 // CHECK: [[T_VAR_ADDR_REF1:%.+]] = load i{{[0-9]+}}*, i{{[0-9]+}}** [[T_VAR_ADDR1]],
340 // CHECK: [[S_ARR_ADDR_REF1:%.+]] = load [2 x [[S_INT_TY]]]*, [2 x [[S_INT_TY]]]** [[S_ARR_ADDR1]],
341 // CHECK: [[VAR_ADDR1_REF:%.+]] = load [[S_INT_TY]]*, [[S_INT_TY]]** [[VAR_ADDR1]],
342 // CHECK: store [[S_INT_TY]]* [[VAR_ADDR1_REF]], [[S_INT_TY]]** [[TMP1]],
343 // CHECK-DAG: store i{{[0-9]+}} 0, i{{[0-9]+}}* [[OMP_IS_LAST1]],
344 // CHECK: [[VAR_ADDR1_REF:%.+]] = load [[S_INT_TY]]*, [[S_INT_TY]]** [[TMP1]],
345 // CHECK-DAG: store [[S_INT_TY]]* [[VAR_PRIV1]], [[S_INT_TY]]** [[TMP_PRIV1]],
346 // CHECK: call void @__kmpc_for_static_init_4(
347 // assignment: vec[i] = t_var;
348 // CHECK: [[IV_VAL1:%.+]] =
349 // CHECK: [[T_VAR_PRIV_VAL1:%.+]] = load i{{[0-9]+}}, i{{[0-9]+}}* [[T_VAR_PRIV1]],
350 // CHECK: [[VEC_PTR1:%.+]] = getelementptr inbounds [2 x i{{[0-9]+}}], [2 x i{{[0-9]+}}]* [[VEC_PRIV1]], i{{[0-9]+}} 0, i{{[0-9]+}} {{.+}}
351 // CHECK:  store i{{[0-9]+}} [[T_VAR_PRIV_VAL1]], i{{[0-9]+}}* [[VEC_PTR1]],
352 
353 // assignment: s_arr[i] = var;
354 // CHECK-DAG: [[S_ARR_PTR1:%.+]] = getelementptr inbounds [2 x [[S_INT_TY]]], [2 x [[S_INT_TY]]]* [[S_ARR_PRIV1]],
355 // CHECK-DAG: [[TMP_VAL1:%.+]] = load [[S_INT_TY]]*, [[S_INT_TY]]** [[TMP_PRIV1]],
356 // CHECK-DAG: [[S_ARR_PTR_BCAST1:%.+]] = bitcast [[S_INT_TY]]* [[S_ARR_PTR1]] to i8*
357 // CHECK-DAG: [[TMP_VAL_BCAST1:%.+]] = bitcast [[S_INT_TY]]* [[TMP_VAL1]] to i8*
358 // CHECK-DAG: call void @llvm.memcpy.{{.+}}(i8* align {{[0-9]+}} [[S_ARR_PTR_BCAST1]], i8* align {{[0-9]+}} [[TMP_VAL_BCAST1]],
359 // CHECK: call void @__kmpc_for_static_fini(
360 
361 // lastprivates
362 // CHECK: [[OMP_IS_LAST_VAL:%.+]] = load i{{[0-9]+}}, i{{[0-9]+}}* [[OMP_IS_LAST1]],
363 // CHECK: [[IS_LAST_IT:%.+]] = icmp ne i{{[0-9]+}} [[OMP_IS_LAST_VAL]], 0
364 // CHECK: br i1 [[IS_LAST_IT]], label %[[OMP_LASTPRIV_BLOCK:.+]], label %[[OMP_LASTPRIV_DONE:.+]]
365 
366 // CHECK: [[OMP_LASTPRIV_BLOCK]]:
367 // CHECK: [[T_VAR_VAL:%.+]] = load i{{[0-9]+}}, i{{[0-9]+}}* [[T_VAR_PRIV1]],
368 // CHECK: store i{{[0-9]+}} [[T_VAR_VAL]], i{{[0-9]+}}* [[T_VAR_ADDR_REF1]],
369 // CHECK: [[BCAST_VEC_ADDR_REF:%.+]] = bitcast [2 x i{{[0-9]+}}]* [[VEC_ADDR_REF1]] to i8*
370 // CHECK: [[BCAST_VEC_PRIV:%.+]] = bitcast [2 x i{{[0-9]+}}]* [[VEC_PRIV1]] to i8*
371 // CHECK: call void @llvm.memcpy.{{.+}}(i8* align {{[0-9]+}} [[BCAST_VEC_ADDR_REF]], i8* align {{[0-9]+}} [[BCAST_VEC_PRIV]],
372 // CHECK: [[S_ARR_BEGIN:%.+]] = getelementptr inbounds [2 x [[S_INT_TY]]], [2 x [[S_INT_TY]]]* [[S_ARR_ADDR_REF]], i{{[0-9]+}} 0, i{{[0-9]+}} 0
373 // CHECK: [[S_ARR_PRIV_BCAST:%.+]] = bitcast [2 x [[S_INT_TY]]]* [[S_ARR_PRIV1]] to [[S_INT_TY]]*
374 // CHECK: [[S_ARR_BEGIN_GEP:%.+]] = getelementptr [[S_INT_TY]], [[S_INT_TY]]* [[S_ARR_BEGIN]], i{{[0-9]+}} 2
375 // CHECK: [[S_ARR_IS_EMPTY:%.+]] = icmp eq [[S_INT_TY]]* [[S_ARR_BEGIN]], [[S_ARR_BEGIN_GEP]]
376 // CHECK: br i1 [[S_ARR_IS_EMPTY]], label %[[S_ARR_COPY_DONE:.+]], label %[[S_ARR_COPY_BLOCK:.+]]
377 // CHECK: [[S_ARR_COPY_BLOCK]]:
378 // CHECK: [[S_ARR_SRC_EL:%.+]] = phi [[S_INT_TY]]*{{.+}}
379 // CHECK: [[S_ARR_DST_EL:%.+]] = phi [[S_INT_TY]]*{{.+}}
380 // CHECK: [[S_ARR_DST_BCAST:%.+]] = bitcast [[S_INT_TY]]* [[S_ARR_DST_EL]] to i8*
381 // CHECK: [[S_ARR_SRC_BCAST:%.+]] = bitcast [[S_INT_TY]]* [[S_ARR_SRC_EL]] to i8*
382 // CHECK: call void @llvm.memcpy.{{.+}}(i8* align {{[0-9]+}} [[S_ARR_DST_BCAST]], i8* align {{[0-9]+}} [[S_ARR_SRC_BCAST]]{{.+}})
383 // CHECK: [[S_ARR_DST_NEXT:%.+]] = getelementptr [[S_INT_TY]], [[S_INT_TY]]* [[S_ARR_DST_EL]], i{{[0-9]+}} 1
384 // CHECK: [[S_ARR_SRC_NEXT:%.+]] = getelementptr{{.+}}
385 // CHECK: [[CPY_IS_FINISHED:%.+]] = icmp eq [[S_INT_TY]]* [[S_ARR_DST_NEXT]], [[S_ARR_BEGIN_GEP]]
386 // CHECK: br i1 [[CPY_IS_FINISHED]], label %[[S_ARR_COPY_DONE]], label %[[S_ARR_COPY_BLOCK]]
387 // CHECK: [[S_ARR_COPY_DONE]]:
388 // CHECK: [[TMP_VAL1:%.+]] = load [[S_INT_TY]]*, [[S_INT_TY]]** [[TMP_PRIV1]],
389 // CHECK: [[VAR_ADDR_REF_BCAST:%.+]] = bitcast [[S_INT_TY]]* [[VAR_ADDR1_REF]] to i8*
390 // CHECK: [[TMP_VAL1_BCAST:%.+]] = bitcast [[S_INT_TY]]* [[TMP_VAL1]] to i8*
391 // CHECK: call void @llvm.memcpy.{{.+}}(i8* align {{[0-9]+}} [[VAR_ADDR_REF_BCAST]], i8* align {{[0-9]+}} [[TMP_VAL1_BCAST]],{{.+}})
392 // CHECK: ret void
393 // CHECK: !{!"llvm.loop.vectorize.enable", i1 true}
394 #endif
395