1 // RUN: %clang_cc1 -DCHECK -verify -fopenmp -x c++ -triple powerpc64le-unknown-unknown -fopenmp-targets=powerpc64le-ibm-linux-gnu -emit-llvm %s -o - | FileCheck %s --check-prefix CHECK --check-prefix CHECK-64
2 // RUN: %clang_cc1 -DCHECK -fopenmp -x c++ -std=c++11 -triple powerpc64le-unknown-unknown -fopenmp-targets=powerpc64le-ibm-linux-gnu -emit-pch -o %t %s
3 // RUN: %clang_cc1 -DCHECK -fopenmp -x c++ -triple powerpc64le-unknown-unknown -fopenmp-targets=powerpc64le-ibm-linux-gnu -std=c++11 -include-pch %t -verify %s -emit-llvm -o - | FileCheck %s --check-prefix CHECK --check-prefix CHECK-64
4 // RUN: %clang_cc1 -DCHECK -verify -fopenmp -x c++ -triple i386-unknown-unknown -fopenmp-targets=i386-pc-linux-gnu -emit-llvm %s -o - | FileCheck %s --check-prefix CHECK --check-prefix CHECK-32
5 // RUN: %clang_cc1 -DCHECK -fopenmp -x c++ -std=c++11 -triple i386-unknown-unknown -fopenmp-targets=i386-pc-linux-gnu -emit-pch -o %t %s
6 // RUN: %clang_cc1 -DCHECK -fopenmp -x c++ -triple i386-unknown-unknown -fopenmp-targets=i386-pc-linux-gnu -std=c++11 -include-pch %t -verify %s -emit-llvm -o - | FileCheck %s --check-prefix CHECK --check-prefix CHECK-32
7 
8 // 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 - | FileCheck %s --check-prefix LAMBDA --check-prefix LAMBDA-64
9 // 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
10 // RUN: %clang_cc1 -DLAMBDA -fopenmp -x c++  -std=c++11 -triple powerpc64le-unknown-unknown -fopenmp-targets=powerpc64le-ibm-linux-gnu -std=c++11 -include-pch %t -verify %s -emit-llvm -o - | FileCheck %s --check-prefix LAMBDA --check-prefix LAMBDA-64
11 
12 // expected-no-diagnostics
13 #ifndef HEADER
14 #define HEADER
15 
16 struct St {
17   int a, b;
18   St() : a(0), b(0) {}
19   St(const St &st) : a(st.a + st.b), b(0) {}
20   ~St() {}
21 };
22 
23 volatile int g = 1212;
24 volatile int &g1 = g;
25 
26 template <class T>
27 struct S {
28   T f;
29   S(T a) : f(a + g) {}
30   S() : f(g) {}
31   S(const S &s, St t = St()) : f(s.f + t.a) {}
32   operator T() { return T(); }
33   ~S() {}
34 };
35 
36 // CHECK-DAG: [[S_FLOAT_TY:%.+]] = type { float }
37 // CHECK-DAG: [[S_INT_TY:%.+]] = type { i{{[0-9]+}} }
38 // CHECK-DAG: [[ST_TY:%.+]] = type { i{{[0-9]+}}, i{{[0-9]+}} }
39 
40 template <typename T>
41 T tmain() {
42   S<T> test;
43   T t_var = T();
44   T vec[] = {1, 2};
45   S<T> s_arr[] = {1, 2};
46   S<T> &var = test;
47 #pragma omp target
48 #pragma omp teams distribute firstprivate(t_var, vec, s_arr, var)
49   for (int i = 0; i < 2; ++i) {
50     vec[i] = t_var;
51     s_arr[i] = var;
52   }
53   return T();
54 }
55 
56 // CHECK-DAG: [[TEST:@.+]] = global [[S_FLOAT_TY]] zeroinitializer,
57 S<float> test;
58 // CHECK-DAG: [[T_VAR:@.+]] = global i{{[0-9]+}} 333,
59 int t_var = 333;
60 // CHECK-DAG: [[VEC:@.+]] = global [2 x i{{[0-9]+}}] [i{{[0-9]+}} 1, i{{[0-9]+}} 2],
61 int vec[] = {1, 2};
62 // CHECK-DAG: [[S_ARR:@.+]] = global [2 x [[S_FLOAT_TY]]] zeroinitializer,
63 S<float> s_arr[] = {1, 2};
64 // CHECK-DAG: [[VAR:@.+]] = global [[S_FLOAT_TY]] zeroinitializer,
65 S<float> var(3);
66 // CHECK-DAG: [[SIVAR:@.+]] = internal global i{{[0-9]+}} 0,
67 
68 int main() {
69   static int sivar;
70 #ifdef LAMBDA
71   // LAMBDA: [[G:@.+]] = global i{{[0-9]+}} 1212,
72   // LAMBDA-LABEL: @main
73   // LAMBDA: call void [[OUTER_LAMBDA:@.+]](
74   [&]() {
75     // LAMBDA: define{{.*}} internal{{.*}} void [[OUTER_LAMBDA]](
76     // LAMBDA: call i32 @__tgt_target(i32 -1, i8* @{{[^,]+}}, i32 3, i8** %{{[^,]+}}, i8** %{{[^,]+}}, i{{64|32}}* {{.+}}@{{[^,]+}}, i32 0, i32 0), i32* {{.+}}@{{[^,]+}}, i32 0, i32 0))
77     // LAMBDA: call void @[[LOFFL1:.+]](i{{64|32}} %{{.+}})
78     // LAMBDA:  ret
79 #pragma omp target
80 #pragma omp teams distribute firstprivate(g, g1, sivar)
81   for (int i = 0; i < 2; ++i) {
82     // LAMBDA: define{{.*}} internal{{.*}} void @[[LOFFL1]](i{{64|32}} {{%.+}}, i{{64|32}} {{%.+}})
83     // LAMBDA: {{%.+}} = alloca i{{[0-9]+}},
84     // LAMBDA: {{%.+}} = alloca i{{[0-9]+}},
85     // LAMBDA: {{%.+}} = alloca i{{[0-9]+}},
86     // LAMBDA: [[G_CAST:%.+]] = alloca i{{[0-9]+}},
87     // LAMBDA: [[G1_CAST:%.+]] = alloca i{{[0-9]+}},
88     // LAMBDA: [[SIVAR_CAST:%.+]] = alloca i{{[0-9]+}},
89     // LAMBDA-DAG: [[G_CAST_VAL:%.+]] = load{{.+}} [[G_CAST]],
90     // LAMBDA-DAG: [[G1_CAST_VAL:%.+]] = load{{.+}} [[G1_CAST]],
91     // LAMBDA-DAG: [[SIVAR_CAST_VAL:%.+]] = load{{.+}} [[SIVAR_CAST]],
92     // LAMBDA: call void {{.+}} @__kmpc_fork_teams({{.+}}, i32 3, {{.+}} @[[LOUTL1:.+]] to {{.+}}, {{.+}} [[G_CAST_VAL]], {{.+}} [[G1_CAST_VAL]], {{.+}} [[SIVAR_CAST_VAL]])
93     // LAMBDA: ret void
94 
95     // LAMBDA: define internal void @[[LOUTL1]]({{.+}})
96     // Skip global and bound tid vars
97     // LAMBDA: {{.+}} = alloca i32*,
98     // LAMBDA: {{.+}} = alloca i32*,
99     // LAMBDA: [[G_ADDR:%.+]] = alloca i{{[0-9]+}},
100     // LAMBDA: [[G1_ADDR:%.+]] = alloca i{{[0-9]+}},
101     // LAMBDA: [[SIVAR_ADDR:%.+]] = alloca i{{[0-9]+}},
102     // LAMBDA: [[G1_TMP:%.+]] = alloca i32*,
103     // skip loop vars
104     // LAMBDA-DAG: store {{.+}}, {{.+}} [[G_ADDR]],
105     // LAMBDA-DAG: store {{.+}}, {{.+}} [[G1_ADDR]],
106     // LAMBDA-DAG: store {{.+}}, {{.+}} [[SIVAR_ADDR]],
107     // LAMBDA-DAG: [[G_CONV:%.+]] = bitcast {{.+}} [[G_ADDR]] to
108     // LAMBDA-DAG: [[G1_CONV:%.+]] = bitcast {{.+}} [[G1_ADDR]] to
109     // LAMBDA-DAG: [[SIVAR_CONV:%.+]] = bitcast {{.+}} [[SIVAR_ADDR]] to
110     // LAMBDA-DAG: store{{.+}} [[G1_CONV]], {{.+}} [[G1_TMP]],
111     g = 1;
112     g1 = 1;
113     sivar = 2;
114     // LAMBDA: call void @__kmpc_for_static_init_4(
115     // LAMBDA-DAG: store{{.+}} 1, {{.+}} [[G_CONV]],
116     // LAMBDA-DAG: [[G1:%.+]] = load{{.+}}, {{.+}}* [[G1_TMP]]
117     // LAMBDA-DAG: store{{.+}} 1, {{.+}} [[G1]],
118     // LAMBDA-DAG: store{{.+}} 2, {{.+}} [[SIVAR_CONV]],
119     // LAMBDA-DAG: [[G1_REF:%.+]] = load{{.+}}, {{.+}} [[G1_TMP]],
120     // LAMBDA-DAG: store{{.+}} 1, {{.+}} [[G1_REF]],
121     // LAMBDA: call void [[INNER_LAMBDA:@.+]](
122     // LAMBDA: call void @__kmpc_for_static_fini(
123     [&]() {
124       // LAMBDA: define {{.+}} void [[INNER_LAMBDA]](%{{.+}}* [[ARG_PTR:%.+]])
125       // LAMBDA: store %{{.+}}* [[ARG_PTR]], %{{.+}}** [[ARG_PTR_REF:%.+]],
126       g = 2;
127       g1 = 2;
128       sivar = 4;
129       // LAMBDA: [[ARG_PTR:%.+]] = load %{{.+}}*, %{{.+}}** [[ARG_PTR_REF]]
130 
131       // LAMBDA: [[G_PTR_REF:%.+]] = getelementptr inbounds %{{.+}}, %{{.+}}* [[ARG_PTR]], i{{[0-9]+}} 0, i{{[0-9]+}} 0
132       // LAMBDA: [[G_REF:%.+]] = load i{{[0-9]+}}*, i{{[0-9]+}}** [[G_PTR_REF]]
133       // LAMBDA: store i{{[0-9]+}} 2, i{{[0-9]+}}* [[G_REF]]
134       // LAMBDA: [[G1_PTR_REF:%.+]] = getelementptr inbounds %{{.+}}, %{{.+}}* [[ARG_PTR]], i{{[0-9]+}} 0, i{{[0-9]+}} 1
135       // LAMBDA: [[G1_REF:%.+]] = load i{{[0-9]+}}*, i{{[0-9]+}}** [[G1_PTR_REF]]
136       // LAMBDA: store i{{[0-9]+}} 2, i{{[0-9]+}}* [[G1_REF]]
137       // LAMBDA: [[SIVAR_PTR_REF:%.+]] = getelementptr inbounds %{{.+}}, %{{.+}}* [[ARG_PTR]], i{{[0-9]+}} 0, i{{[0-9]+}} 2
138       // LAMBDA: [[SIVAR_REF:%.+]] = load i{{[0-9]+}}*, i{{[0-9]+}}** [[SIVAR_PTR_REF]]
139       // LAMBDA: store i{{[0-9]+}} 4, i{{[0-9]+}}* [[SIVAR_REF]]
140     }();
141   }
142   }();
143   return 0;
144 #else
145 #pragma omp target
146 #pragma omp teams distribute firstprivate(t_var, vec, s_arr, var, sivar)
147   for (int i = 0; i < 2; ++i) {
148     vec[i] = t_var;
149     s_arr[i] = var;
150     sivar += i;
151   }
152   return tmain<int>();
153 #endif
154 }
155 
156 // CHECK: define {{.*}}i{{[0-9]+}} @main()
157 // CHECK: call i32 @__tgt_target(i32 -1, i8* @{{[^,]+}}, i32 5, i8** %{{[^,]+}}, i8** %{{[^,]+}}, i{{64|32}}* {{.+}}@{{[^,]+}}, i32 0, i32 0), i32* {{.+}}@{{[^,]+}}, i32 0, i32 0))
158 // CHECK: call void @[[OFFL1:.+]](i{{64|32}} %{{.+}})
159 // CHECK: {{%.+}} = call{{.*}} i32 @[[TMAIN_INT:.+]]()
160 // CHECK:  ret
161 
162 // CHECK: define{{.*}} void @[[OFFL1]]({{.+}})
163 // CHECK: [[T_VAR_PRIV:%.+]] = alloca i{{[0-9]+}},
164 // CHECK: [[VEC_PRIV:%.+]] = alloca [2 x i{{[0-9]+}}]*,
165 // CHECK: [[S_ARR_PRIV:%.+]] = alloca [2 x [[S_FLOAT_TY]]]*,
166 // CHECK: [[VAR_PRIV:%.+]] = alloca [[S_FLOAT_TY]]*,
167 // CHECK: [[SIVAR_PRIV:%.+]] = alloca i{{[0-9]+}},
168 // CHECK: [[T_VAR_CAST:%.+]] = alloca i{{[0-9]+}},
169 // CHECK: [[SIVAR_CAST:%.+]] = alloca i{{[0-9]+}},
170 
171 // CHECK-DAG: [[VEC_TE_PAR:%.+]] = load [2 x i{{[0-9]+}}]*, [2 x i{{[0-9]+}}]** [[VEC_PRIV]],
172 // CHECK-DAG: [[T_VAR_TE_PAR:%.+]] = load i{{[0-9]+}}, i{{[0-9]+}}* [[T_VAR_CAST]],
173 // CHECK-DAG: [[S_ARR_TE_PAR:%.+]] = load [2 x [[S_FLOAT_TY]]]*, [2 x [[S_FLOAT_TY]]]** [[S_ARR_PRIV]],
174 // CHECK-DAG: [[VAR_TE_PAR:%.+]] = load [[S_FLOAT_TY]]*, [[S_FLOAT_TY]]** [[VAR_PRIV]],
175 // CHECK-DAG: [[SIVAR_TE_PAR:%.+]] = load i{{[0-9]+}}, i{{[0-9]+}}* [[SIVAR_CAST]],
176 
177 // CHECK: call void {{.+}} @__kmpc_fork_teams({{.+}}, i32 5, {{.+}} @[[OUTL1:.+]] to {{.+}}, [2 x i{{[0-9]+}}]* [[VEC_TE_PAR]], i{{[0-9]+}} [[T_VAR_TE_PAR]], [2 x [[S_FLOAT_TY]]]* [[S_ARR_TE_PAR]], [[S_FLOAT_TY]]* [[VAR_TE_PAR]], i{{[0-9]+}} [[SIVAR_TE_PAR]])
178 // CHECK: ret void
179 
180 // CHECK: define internal void @[[OUTL1]]({{.+}})
181 // Skip global and bound tid vars
182 // CHECK: {{.+}} = alloca i32*,
183 // CHECK: {{.+}} = alloca i32*,
184 // CHECK: [[VEC_ADDR:%.+]] = alloca [2 x i{{[0-9]+}}]*,
185 // CHECK: [[T_VAR_ADDR:%.+]] = alloca i{{[0-9]+}},
186 // CHECK: [[S_ARR_ADDR:%.+]] = alloca [2 x [[S_FLOAT_TY]]]*,
187 // CHECK: [[VAR_ADDR:%.+]] = alloca [[S_FLOAT_TY]]*,
188 // CHECK: [[SIVAR_ADDR:%.+]] = alloca i{{[0-9]+}},
189 // Skip temp vars for loop
190 // CHECK: alloca i{{[0-9]+}},
191 // CHECK: alloca i{{[0-9]+}},
192 // CHECK: alloca i{{[0-9]+}},
193 // CHECK: alloca i{{[0-9]+}},
194 // CHECK: alloca i{{[0-9]+}},
195 // CHECK: [[VEC_PRIV:%.+]] = alloca [2 x i{{[0-9]+}}],
196 // CHECK: [[S_ARR_PRIV:%.+]] = alloca [2 x [[S_FLOAT_TY]]],
197 // CHECK: [[AGG_TMP1:%.+]] = alloca [[ST_TY]],
198 // CHECK: [[VAR_PRIV:%.+]] = alloca [[S_FLOAT_TY]],
199 // CHECK: [[AGG_TMP2:%.+]] = alloca [[ST_TY]],
200 
201 // param copy
202 // CHECK: store [2 x i{{[0-9]+}}]* {{.+}}, [2 x i{{[0-9]+}}]** [[VEC_ADDR]],
203 // CHECK: store i{{[0-9]+}} {{.+}}, i{{[0-9]+}}* [[T_VAR_ADDR]],
204 // CHECK: store [2 x [[S_FLOAT_TY]]]* {{.+}}, [2 x [[S_FLOAT_TY]]]** [[S_ARR_ADDR]],
205 // CHECK: store [[S_FLOAT_TY]]* {{.+}}, [[S_FLOAT_TY]]** [[VAR_ADDR]],
206 // CHECK: store i{{[0-9]+}} {{.+}}, i{{[0-9]+}}* [[SIVAR_ADDR]],
207 
208 // T_VAR and SIVAR
209 // CHECK-DAG-64: [[CONV_TVAR:%.+]] = bitcast i64* [[T_VAR_ADDR]] to i32*
210 // CHECK-DAG-64: [[CONV_SIVAR:%.+]] = bitcast i64* [[SIVAR_ADDR]] to i32*
211 
212 // preparation vars
213 // CHECK-DAG: [[VEC_ADDR_VAL:%.+]] = load [2 x i{{[0-9]+}}]*, [2 x i{{[0-9]+}}]** [[VEC_ADDR]],
214 // CHECK-DAG: [[S_ARR_ADDR_REF:%.+]] = load [2 x [[S_FLOAT_TY]]]*, [2 x [[S_FLOAT_TY]]]** [[S_ARR_ADDR]],
215 // CHECK-DAG: [[VAR_ADDR_REF:%.+]] = load{{.+}} [[VAR_ADDR]],
216 
217 // firstprivate vec(vec): copy from *_addr into priv1 and then from priv1 into priv2
218 // CHECK-DAG: [[VEC_DEST_PRIV:%.+]] = bitcast [2 x i{{[0-9]+}}]* [[VEC_PRIV]] to i8*
219 // CHECK-DAG: [[VEC_SRC:%.+]] =  bitcast [2 x i{{[0-9]+}}]* [[VEC_ADDR_VAL]] to i8*
220 // CHECK: call void @llvm.memcpy.{{.+}}(i8* [[VEC_DEST_PRIV]], i8* [[VEC_SRC]], {{.+}})
221 
222 // firstprivate(s_arr)
223 // CHECK-DAG: [[S_ARR_PRIV_BGN:%.+]] = getelementptr{{.*}} [2 x [[S_FLOAT_TY]]], [2 x [[S_FLOAT_TY]]]* [[S_ARR_PRIV]],
224 // CHECK-DAG: [[S_ARR_ADDR_BGN:%.+]] = bitcast [2 x [[S_FLOAT_TY]]]* [[S_ARR_ADDR_REF]] to
225 // CHECK-DAG: [[S_ARR_FIN:%.+]] = icmp{{.+}} [[S_ARR_PRIV_BGN]],
226 // CHECK-DAG: [[S_ARR_SRC_COPY:%.+]] = phi{{.+}} [ [[S_ARR_ADDR_BGN]], {{.+}} ], [ [[S_ARR_SRC:%.+]], {{.+}} ]
227 // CHECK-DAG: [[S_ARR_DST_COPY:%.+]] = phi{{.+}} [ [[S_ARR_PRIV_BGN]], {{.+}}], [ [[S_ARR_DST:%.+]], {{.+}} ]
228 // CHECK-DAG: call void @{{.+}}({{.+}} [[AGG_TMP1]])
229 // CHECK-DAG: call void @{{.+}}({{.+}} [[S_ARR_DST_COPY]], {{.+}} [[S_ARR_SRC_COPY]], {{.+}} [[AGG_TMP1]])
230 // CHECK-DAG: call void @{{.+}}({{.+}} [[AGG_TMP1]])
231 // CHECK-DAG: [[S_ARR_DST]] = getelementptr {{.+}} [[S_ARR_DST_COPY]],
232 // CHECK-DAG: [[S_ARR_SRC]] = getelementptr {{.+}} [[S_ARR_SRC_COPY]],
233 
234 // firstprivate(var)
235 // CHECK-DAG: call void @{{.+}}({{.+}} [[AGG_TMP2]])
236 // CHECK-DAG: call void @{{.+}}({{.+}} [[VAR_PRIV]], {{.+}} [[VAR_ADDR_REF]], {{.+}} [[AGG_TMP2]])
237 // CHECK-DAG: call void @{{.+}}({{.+}} [[AGG_TMP2]])
238 
239 // CHECK: call void @__kmpc_for_static_init_4(
240 // CHECK-DAG-32: {{.+}} = {{.+}} [[T_VAR_ADDR]]
241 // CHECK-DAG-64: {{.+}} = {{.+}} [[CONV_TVAR]]
242 // CHECK-DAG: {{.+}} = {{.+}} [[VEC_PRIV]]
243 // CHECK-DAG: {{.+}} = {{.+}} [[S_ARR_PRIV]]
244 // CHECK-DAG: {{.+}} = {{.+}} [[VAR_PRIV]]
245 // CHECK-DAG-32: {{.+}} = {{.+}} [[SIVAR_ADDR]]
246 // CHECK-DAG-64: {{.+}} = {{.+}} [[CONV_SIVAR]]
247 // CHECK: call void @__kmpc_for_static_fini(
248 // CHECK: ret void
249 
250 // CHECK: define{{.*}} i{{[0-9]+}} @[[TMAIN_INT]]()
251 // CHECK: call i32 @__tgt_target(i32 -1, i8* @{{[^,]+}}, i32 4, i8** %{{[^,]+}}, i8** %{{[^,]+}}, i{{64|32}}* {{.+}}@{{[^,]+}}, i32 0, i32 0), i32* {{.+}}@{{[^,]+}}, i32 0, i32 0))
252 // CHECK: call void @[[TOFFL1:.+]](i{{64|32}} %{{.+}})
253 // CHECK:  ret
254 
255 // CHECK: define {{.*}}void @[[TOFFL1]]({{.+}})
256 // CHECK: [[TT_VAR_PRIV:%.+]] = alloca i{{[0-9]+}},
257 // CHECK: [[TVEC_PRIV:%.+]] = alloca [2 x i{{[0-9]+}}]*,
258 // CHECK: [[TS_ARR_PRIV:%.+]] = alloca [2 x [[S_INT_TY]]]*,
259 // CHECK: [[TVAR_PRIV:%.+]] = alloca [[S_INT_TY]]*,
260 // CHECK: [[TT_VAR_CAST:%.+]] = alloca i{{[0-9]+}},
261 
262 // CHECK-DAG: [[TVEC_TE_PAR:%.+]] = load [2 x i{{[0-9]+}}]*, [2 x i{{[0-9]+}}]** [[TVEC_PRIV]],
263 // CHECK-DAG: [[TT_VAR_TE_PAR:%.+]] = load i{{[0-9]+}}, i{{[0-9]+}}* [[TT_VAR_CAST]],
264 // CHECK-DAG: [[TS_ARR_TE_PAR:%.+]] = load [2 x [[S_INT_TY]]]*, [2 x [[S_INT_TY]]]** [[TS_ARR_PRIV]],
265 // CHECK-DAG: [[TVAR_TE_PAR:%.+]] = load [[S_INT_TY]]*, [[S_INT_TY]]** [[TVAR_PRIV]],
266 
267 // CHECK: call void {{.+}} @__kmpc_fork_teams({{.+}}, i32 4, {{.+}} @[[TOUTL1:.+]] to {{.+}}, [2 x i{{[0-9]+}}]* [[TVEC_TE_PAR]], i{{[0-9]+}} [[TT_VAR_TE_PAR]], [2 x [[S_INT_TY]]]* [[TS_ARR_TE_PAR]], [[S_INT_TY]]* [[TVAR_TE_PAR]])
268 // CHECK: ret void
269 
270 // CHECK: define internal void @[[TOUTL1]]({{.+}})
271 // Skip global and bound tid vars
272 // CHECK: {{.+}} = alloca i32*,
273 // CHECK: {{.+}} = alloca i32*,
274 // CHECK: [[VEC_ADDR:%.+]] = alloca [2 x i{{[0-9]+}}]*,
275 // CHECK: [[T_VAR_ADDR:%.+]] = alloca i{{[0-9]+}},
276 // CHECK: [[S_ARR_ADDR:%.+]] = alloca [2 x [[S_INT_TY]]]*,
277 // CHECK: [[VAR_ADDR:%.+]] = alloca [[S_INT_TY]]*,
278 // Skip temp vars for loop
279 // CHECK: alloca i{{[0-9]+}},
280 // CHECK: alloca i{{[0-9]+}},
281 // CHECK: alloca i{{[0-9]+}},
282 // CHECK: alloca i{{[0-9]+}},
283 // CHECK: alloca i{{[0-9]+}},
284 // CHECK: [[VEC_PRIV:%.+]] = alloca [2 x i{{[0-9]+}}],
285 // CHECK: [[S_ARR_PRIV:%.+]] = alloca [2 x [[S_INT_TY]]],
286 // CHECK: [[AGG_TMP1:%.+]] = alloca [[ST_TY]],
287 // CHECK: [[VAR_PRIV:%.+]] = alloca [[S_INT_TY]],
288 // CHECK: [[AGG_TMP2:%.+]] = alloca [[ST_TY]],
289 // CHECK: [[TMP:%.+]] = alloca [[S_INT_TY]]*,
290 
291 // param copy
292 // CHECK: store [2 x i{{[0-9]+}}]* {{.+}}, [2 x i{{[0-9]+}}]** [[VEC_ADDR]],
293 // CHECK: store i{{[0-9]+}} {{.+}}, i{{[0-9]+}}* [[T_VAR_ADDR]],
294 // CHECK: store [2 x [[S_INT_TY]]]* {{.+}}, [2 x [[S_INT_TY]]]** [[S_ARR_ADDR]],
295 // CHECK: store [[S_INT_TY]]* {{.+}}, [[S_INT_TY]]** [[VAR_ADDR]],
296 
297 
298 // T_VAR and preparation variables
299 // CHECK: [[VEC_ADDR_VAL:%.+]] = load [2 x i{{[0-9]+}}]*, [2 x i{{[0-9]+}}]** [[VEC_ADDR]],
300 // CHECK-64: [[CONV_TVAR:%.+]] = bitcast i64* [[T_VAR_ADDR]] to i32*
301 // CHECK: [[S_ARR_ADDR_REF:%.+]] = load [2 x [[S_INT_TY]]]*, [2 x [[S_INT_TY]]]** [[S_ARR_ADDR]],
302 
303 // firstprivate vec(vec): copy from *_addr into priv1 and then from priv1 into priv2
304 // CHECK-DAG: [[VEC_DEST_PRIV:%.+]] = bitcast [2 x i{{[0-9]+}}]* [[VEC_PRIV]] to i8*
305 // CHECK-DAG: [[VEC_SRC:%.+]] = bitcast [2 x i{{[0-9]+}}]* [[VEC_ADDR_VAL]] to i8*
306 // CHECK: call void @llvm.memcpy.{{.+}}(i8* [[VEC_DEST_PRIV]], i8* [[VEC_SRC]], {{.+}})
307 
308 // firstprivate(s_arr)
309 // CHECK-DAG: [[S_ARR_PRIV_BGN:%.+]] = getelementptr{{.*}} [2 x [[S_INT_TY]]], [2 x [[S_INT_TY]]]* [[S_ARR_PRIV]],
310 // CHECK-DAG: [[S_ARR_ADDR_BGN:%.+]] = bitcast [2 x [[S_INT_TY]]]* [[S_ARR_ADDR_REF]] to
311 // CHECK-DAG: [[S_ARR_FIN:%.+]] = icmp{{.+}} [[S_ARR_PRIV_BGN]],
312 // CHECK-DAG: [[S_ARR_SRC_COPY:%.+]] = phi{{.+}} [ [[S_ARR_ADDR_BGN]], {{.+}} ], [ [[S_ARR_SRC:%.+]], {{.+}} ]
313 // CHECK-DAG: [[S_ARR_DST_COPY:%.+]] = phi{{.+}} [ [[S_ARR_PRIV_BGN]], {{.+}} ], [ [[S_ARR_DST:%.+]], {{.+}} ]
314 // CHECK-DAG: call void @{{.+}}({{.+}} [[AGG_TMP1]])
315 // CHECK-DAG: call void @{{.+}}({{.+}} [[S_ARR_DST_COPY]], {{.+}} [[S_ARR_SRC_COPY]], {{.+}} [[AGG_TMP1]])
316 // CHECK-DAG: call void @{{.+}}({{.+}} [[AGG_TMP1]])
317 // CHECK-DAG: [[S_ARR_DST]] = getelementptr {{.+}} [[S_ARR_DST_COPY]],
318 // CHECK-DAG: [[S_ARR_SRC]] = getelementptr {{.+}} [[S_ARR_SRC_COPY]],
319 
320 // firstprivate(var)
321 // CHECK-DAG: [[VAR_ADDR_REF:%.+]] = load{{.+}} [[VAR_ADDR]],
322 // CHECK-DAG: call void @{{.+}}({{.+}} [[AGG_TMP2]])
323 // CHECK-DAG: call void @{{.+}}({{.+}} [[VAR_PRIV]], {{.+}} [[VAR_ADDR_REF]], {{.+}} [[AGG_TMP2]])
324 // CHECK-DAG: call void @{{.+}}({{.+}} [[AGG_TMP2]])
325 // CHECK-DAG: store [[S_INT_TY]]* [[VAR_PRIV]], [[S_INT_TY]]** [[TMP]],
326 
327 // CHECK: call void @__kmpc_for_static_init_4(
328 // CHECK-DAG-32: {{.+}} = {{.+}} [[T_VAR_ADDR]]
329 // CHECK-DAG-64: {{.+}} = {{.+}} [[CONV_TVAR]]
330 // CHECK-DAG: {{.+}} = {{.+}} [[VEC_PRIV]]
331 // CHECK-DAG: {{.+}} = {{.+}} [[TMP]]
332 // CHECK-DAG: {{.+}} = {{.+}} [[S_ARR_PRIV]]
333 // CHECK: call void @__kmpc_for_static_fini(
334 // CHECK: ret void
335 
336 #endif
337