1 | // RUN: %clang_cc1 -verify -fopenmp -x c++ -triple i386-pc-linux-gnu -emit-llvm %s -o - | FileCheck %s -check-prefix=CHECK -check-prefix=CHECK-32 |
2 | // RUN: %clang_cc1 -fopenmp -x c++ -std=c++11 -triple i386-pc-linux-gnu -emit-pch -o %t %s |
3 | // RUN: %clang_cc1 -fopenmp -x c++ -triple i386-pc-linux-gnu -std=c++11 -include-pch %t -verify %s -emit-llvm -o - | FileCheck %s -check-prefix=CHECK -check-prefix=CHECK-32 |
4 | // RUN: %clang_cc1 -verify -fopenmp -x c++ -std=c++11 -DLAMBDA -triple i386-pc-linux-gnu -emit-llvm %s -o - | FileCheck -check-prefix=LAMBDA -check-prefix=LAMBDA-32 %s |
5 | // RUN: %clang_cc1 -verify -fopenmp -x c++ -fblocks -DBLOCKS -triple i386-pc-linux-gnu -emit-llvm %s -o - | FileCheck -check-prefix=BLOCKS -check-prefix=BLOCKS-32 %s |
6 | |
7 | // RUN: %clang_cc1 -verify -fopenmp-simd -x c++ -triple i386-pc-linux-gnu -emit-llvm %s -o - | FileCheck --check-prefix SIMD-ONLY0 %s |
8 | // RUN: %clang_cc1 -fopenmp-simd -x c++ -std=c++11 -triple i386-pc-linux-gnu -emit-pch -o %t %s |
9 | // RUN: %clang_cc1 -fopenmp-simd -x c++ -triple i386-pc-linux-gnu -std=c++11 -include-pch %t -verify %s -emit-llvm -o - | FileCheck --check-prefix SIMD-ONLY0 %s |
10 | // RUN: %clang_cc1 -verify -fopenmp-simd -x c++ -std=c++11 -DLAMBDA -triple i386-pc-linux-gnu -emit-llvm %s -o - | FileCheck --check-prefix SIMD-ONLY0 %s |
11 | // RUN: %clang_cc1 -verify -fopenmp-simd -x c++ -fblocks -DBLOCKS -triple i386-pc-linux-gnu -emit-llvm %s -o - | FileCheck --check-prefix SIMD-ONLY0 %s |
12 | // SIMD-ONLY0-NOT: {{__kmpc|__tgt}} |
13 | |
14 | // RUN: %clang_cc1 -verify -fopenmp -x c++ -triple x86_64-pc-linux-gnu -emit-llvm %s -o - | FileCheck %s -check-prefix=CHECK -check-prefix=CHECK-64 |
15 | // RUN: %clang_cc1 -fopenmp -x c++ -std=c++11 -triple x86_64-pc-linux-gnu -emit-pch -o %t %s |
16 | // RUN: %clang_cc1 -fopenmp -x c++ -triple x86_64-pc-linux-gnu -std=c++11 -include-pch %t -verify %s -emit-llvm -o - | FileCheck %s -check-prefix=CHECK -check-prefix=CHECK-64 |
17 | // RUN: %clang_cc1 -verify -fopenmp -x c++ -std=c++11 -DLAMBDA -triple x86_64-pc-linux-gnu -emit-llvm %s -o - | FileCheck -check-prefix=LAMBDA -check-prefix=LAMBDA-64 %s |
18 | // RUN: %clang_cc1 -verify -fopenmp -x c++ -fblocks -DBLOCKS -triple x86_64-pc-linux-gnu -emit-llvm %s -o - | FileCheck -check-prefix=BLOCKS -check-prefix=BLOCKS-64 %s |
19 | |
20 | // RUN: %clang_cc1 -verify -fopenmp-simd -x c++ -triple x86_64-pc-linux-gnu -emit-llvm %s -o - | FileCheck --check-prefix SIMD-ONLY1 %s |
21 | // RUN: %clang_cc1 -fopenmp-simd -x c++ -std=c++11 -triple x86_64-pc-linux-gnu -emit-pch -o %t %s |
22 | // RUN: %clang_cc1 -fopenmp-simd -x c++ -triple x86_64-pc-linux-gnu -std=c++11 -include-pch %t -verify %s -emit-llvm -o - | FileCheck --check-prefix SIMD-ONLY1 %s |
23 | // RUN: %clang_cc1 -verify -fopenmp-simd -x c++ -std=c++11 -DLAMBDA -triple x86_64-pc-linux-gnu -emit-llvm %s -o - | FileCheck --check-prefix SIMD-ONLY1 %s |
24 | // RUN: %clang_cc1 -verify -fopenmp-simd -x c++ -fblocks -DBLOCKS -triple x86_64-pc-linux-gnu -emit-llvm %s -o - | FileCheck --check-prefix SIMD-ONLY1 %s |
25 | // SIMD-ONLY1-NOT: {{__kmpc|__tgt}} |
26 | |
27 | // 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 |
28 | |
29 | // 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-ONLY2 %s |
30 | // SIMD-ONLY2-NOT: {{__kmpc|__tgt}} |
31 | // expected-no-diagnostics |
32 | #ifndef ARRAY |
33 | #ifndef HEADER |
34 | #define HEADER |
35 | |
36 | struct St { |
37 | int a, b; |
38 | St() : a(0), b(0) {} |
39 | St(const St &st) : a(st.a + st.b), b(0) {} |
40 | ~St() {} |
41 | }; |
42 | |
43 | volatile int g __attribute__((aligned(128))) = 1212; |
44 | |
45 | struct SS { |
46 | int a; |
47 | int b : 4; |
48 | int &c; |
49 | int e[4]; |
50 | SS(int &d) : a(0), b(0), c(d) { |
51 | #pragma omp parallel firstprivate(a, b, c, e) |
52 | #ifdef LAMBDA |
53 | [&]() { |
54 | ++this->a, --b, (this)->c /= 1; |
55 | #pragma omp parallel firstprivate(a, b, c) |
56 | ++(this)->a, --b, this->c /= 1; |
57 | }(); |
58 | #elif defined(BLOCKS) |
59 | ^{ |
60 | ++a; |
61 | --this->b; |
62 | (this)->c /= 1; |
63 | #pragma omp parallel firstprivate(a, b, c) |
64 | ++(this)->a, --b, this->c /= 1; |
65 | }(); |
66 | #else |
67 | ++this->a, --b, c /= 1, e[2] = 1111; |
68 | #endif |
69 | } |
70 | }; |
71 | |
72 | template<typename T> |
73 | struct SST { |
74 | T a; |
75 | SST() : a(T()) { |
76 | #pragma omp parallel firstprivate(a) |
77 | #ifdef LAMBDA |
78 | [&]() { |
79 | [&]() { |
80 | ++this->a; |
81 | #pragma omp parallel firstprivate(a) |
82 | ++(this)->a; |
83 | }(); |
84 | }(); |
85 | #elif defined(BLOCKS) |
86 | ^{ |
87 | ^{ |
88 | ++a; |
89 | #pragma omp parallel firstprivate(a) |
90 | ++(this)->a; |
91 | }(); |
92 | }(); |
93 | #else |
94 | ++(this)->a; |
95 | #endif |
96 | } |
97 | }; |
98 | |
99 | template <class T> |
100 | struct S { |
101 | T f; |
102 | S(T a) : f(a + g) {} |
103 | S() : f(g) {} |
104 | S(const S &s, St t = St()) : f(s.f + t.a) {} |
105 | operator T() { return T(); } |
106 | ~S() {} |
107 | }; |
108 | |
109 | // CHECK: [[SS_TY:%.+]] = type { i{{[0-9]+}}, i8 |
110 | // LAMBDA: [[SS_TY:%.+]] = type { i{{[0-9]+}}, i8 |
111 | // BLOCKS: [[SS_TY:%.+]] = type { i{{[0-9]+}}, i8 |
112 | // CHECK-DAG: [[S_FLOAT_TY:%.+]] = type { float } |
113 | // CHECK-DAG: [[S_INT_TY:%.+]] = type { i{{[0-9]+}} } |
114 | // CHECK-DAG: [[ST_TY:%.+]] = type { i{{[0-9]+}}, i{{[0-9]+}} } |
115 | |
116 | template <typename T> |
117 | T tmain() { |
118 | S<T> test; |
119 | SST<T> sst; |
120 | T t_var __attribute__((aligned(128))) = T(); |
121 | T vec[] __attribute__((aligned(128))) = {1, 2}; |
122 | S<T> s_arr[] __attribute__((aligned(128))) = {1, 2}; |
123 | S<T> var __attribute__((aligned(128))) (3); |
124 | #pragma omp parallel firstprivate(t_var, vec, s_arr, var) |
125 | { |
126 | vec[0] = t_var; |
127 | s_arr[0] = var; |
128 | } |
129 | #pragma omp parallel firstprivate(t_var) |
130 | {} |
131 | return T(); |
132 | } |
133 | |
134 | int main() { |
135 | static int sivar; |
136 | SS ss(sivar); |
137 | #ifdef LAMBDA |
138 | // LAMBDA: [[G:@.+]] = global i{{[0-9]+}} 1212, |
139 | // LAMBDA-LABEL: @main |
140 | // LAMBDA: alloca [[SS_TY]], |
141 | // LAMBDA: alloca [[CAP_TY:%.+]], |
142 | // LAMBDA: call{{.*}} void [[OUTER_LAMBDA:@[^(]+]]([[CAP_TY]]* |
143 | [&]() { |
144 | // LAMBDA: define{{.*}} internal{{.*}} void [[OUTER_LAMBDA]]( |
145 | // LAMBDA: call {{.*}}void {{.+}} @__kmpc_fork_call({{.+}}, i32 2, {{.+}}* [[OMP_REGION:@.+]] to {{.+}}, i32* [[G]], {{.+}}) |
146 | #pragma omp parallel firstprivate(g, sivar) |
147 | { |
148 | // LAMBDA: define {{.+}} @{{.+}}([[SS_TY]]* |
149 | // LAMBDA: getelementptr inbounds [[SS_TY]], [[SS_TY]]* %{{.+}}, i32 0, i32 0 |
150 | // LAMBDA: store i{{[0-9]+}} 0, i{{[0-9]+}}* % |
151 | // LAMBDA: getelementptr inbounds [[SS_TY]], [[SS_TY]]* %{{.+}}, i32 0, i32 1 |
152 | // LAMBDA: store i8 |
153 | // LAMBDA: getelementptr inbounds [[SS_TY]], [[SS_TY]]* %{{.+}}, i32 0, i32 2 |
154 | // LAMBDA: getelementptr inbounds [[SS_TY]], [[SS_TY]]* %{{.+}}, i32 0, i32 0 |
155 | // LAMBDA: getelementptr inbounds [[SS_TY]], [[SS_TY]]* %{{.+}}, i32 0, i32 1 |
156 | // LAMBDA: getelementptr inbounds [[SS_TY]], [[SS_TY]]* %{{.+}}, i32 0, i32 2 |
157 | // LAMBDA: call void (%{{.+}}*, i{{[0-9]+}}, void (i{{[0-9]+}}*, i{{[0-9]+}}*, ...)*, ...) @__kmpc_fork_call(%{{.+}}* @{{.+}}, i{{[0-9]+}} 5, void (i{{[0-9]+}}*, i{{[0-9]+}}*, ...)* bitcast (void (i{{[0-9]+}}*, i{{[0-9]+}}*, [[SS_TY]]*, [[iz:i64|i32]], {{i64|i32}}, {{i64|i32}}, [4 x i{{[0-9]+}}]*)* [[SS_MICROTASK:@.+]] to void |
158 | // LAMBDA: ret |
159 | |
160 | // LAMBDA: define internal void [[SS_MICROTASK]](i{{[0-9]+}}* noalias [[GTID_ADDR:%.+]], i{{[0-9]+}}* noalias %{{.+}}, [[SS_TY]]* %{{.+}}, [[iz]] {{.+}}, [[iz]] {{.+}}, [[iz]] {{.+}}, [4 x i{{[0-9]+}}]* {{.+}}) |
161 | // LAMBDA-NOT: getelementptr {{.*}}[[SS_TY]], [[SS_TY]]* % |
162 | // LAMBDA: call{{.*}} void |
163 | // LAMBDA: ret void |
164 | |
165 | // LAMBDA: define internal void @{{.+}}(i{{[0-9]+}}* noalias [[GTID_ADDR:%.+]], i{{[0-9]+}}* noalias %{{.+}}, [[SS_TY]]* %{{.+}}, [[iz]] {{.+}}, [[iz]] {{.+}}, [[iz]] {{.+}}) |
166 | // LAMBDA: [[A_PRIV:%.+]] = alloca i{{[0-9]+}}, |
167 | // LAMBDA: [[B_PRIV:%.+]] = alloca i{{[0-9]+}}, |
168 | // LAMBDA: [[C_PRIV:%.+]] = alloca i{{[0-9]+}}, |
169 | // LAMBDA-64: [[A_CONV:%.+]] = bitcast i64* [[A_PRIV]] to i32* |
170 | // LAMBDA-64: store i32* [[A_CONV]], i32** [[REFA:%.+]], |
171 | // LAMBDA-32: store i32* [[A_PRIV]], i32** [[REFA:%.+]], |
172 | // LAMBDA-64: [[B_CONV:%.+]] = bitcast i64* [[B_PRIV]] to i32* |
173 | // LAMBDA-64: [[C_CONV:%.+]] = bitcast i64* [[C_PRIV]] to i32* |
174 | // LAMBDA-64: store i32* [[C_CONV]], i32** [[REFC:%.+]], |
175 | // LAMBDA-32: store i32* [[C_PRIV]], i32** [[REFC:%.+]], |
176 | // LAMBDA-NEXT: [[A_PRIV:%.+]] = load i{{[0-9]+}}*, i{{[0-9]+}}** [[REFA]], |
177 | // LAMBDA-NEXT: [[A_VAL:%.+]] = load i{{[0-9]+}}, i{{[0-9]+}}* [[A_PRIV]], |
178 | // LAMBDA-NEXT: [[INC:%.+]] = add nsw i{{[0-9]+}} [[A_VAL]], 1 |
179 | // LAMBDA-NEXT: store i{{[0-9]+}} [[INC]], i{{[0-9]+}}* [[A_PRIV]], |
180 | // LAMBDA-64-NEXT: [[B_VAL:%.+]] = load i{{[0-9]+}}, i{{[0-9]+}}* [[B_CONV]], |
181 | // LAMBDA-32-NEXT: [[B_VAL:%.+]] = load i{{[0-9]+}}, i{{[0-9]+}}* [[B_PRIV]], |
182 | // LAMBDA-NEXT: [[DEC:%.+]] = add nsw i{{[0-9]+}} [[B_VAL]], -1 |
183 | // LAMBDA-64-NEXT: store i{{[0-9]+}} [[DEC]], i{{[0-9]+}}* [[B_CONV]], |
184 | // LAMBDA-32-NEXT: store i{{[0-9]+}} [[DEC]], i{{[0-9]+}}* [[B_PRIV]], |
185 | // LAMBDA-NEXT: [[C_PRIV:%.+]] = load i{{[0-9]+}}*, i{{[0-9]+}}** [[REFC]], |
186 | // LAMBDA-NEXT: [[C_VAL:%.+]] = load i{{[0-9]+}}, i{{[0-9]+}}* [[C_PRIV]], |
187 | // LAMBDA-NEXT: [[DIV:%.+]] = sdiv i{{[0-9]+}} [[C_VAL]], 1 |
188 | // LAMBDA-NEXT: store i{{[0-9]+}} [[DIV]], i{{[0-9]+}}* [[C_PRIV]], |
189 | // LAMBDA-NEXT: ret void |
190 | |
191 | // LAMBDA: define{{.*}} internal{{.*}} void [[OMP_REGION]](i32* noalias %{{.+}}, i32* noalias %{{.+}}, i32* dereferenceable(4) %{{.+}}, [[iz]] {{.*}}%{{.+}}) |
192 | // LAMBDA: [[SIVAR_PRIVATE_ADDR:%.+]] = alloca i{{[0-9]+}}, |
193 | // LAMBDA: [[G_PRIVATE_ADDR:%.+]] = alloca i{{[0-9]+}}, align 128 |
194 | // LAMBDA: [[G_REF:%.+]] = load i{{[0-9]+}}*, i{{[0-9]+}}** [[G_REF_ADDR:%.+]] |
195 | // LAMBDA-64: [[SIVAR_PRIVATE_CONV:%.+]] = bitcast i64* [[SIVAR_PRIVATE_ADDR]] to i32* |
196 | // LAMBDA: [[G_VAL:%.+]] = load volatile i{{[0-9]+}}, i{{[0-9]+}}* [[G_REF]], align 128 |
197 | // LAMBDA: store i{{[0-9]+}} [[G_VAL]], i{{[0-9]+}}* [[G_PRIVATE_ADDR]], align 128 |
198 | // LAMBDA-NOT: call {{.*}}void @__kmpc_barrier( |
199 | g = 1; |
200 | sivar = 2; |
201 | // LAMBDA: store i{{[0-9]+}} 1, i{{[0-9]+}}* [[G_PRIVATE_ADDR]], |
202 | // LAMBDA-64: store i{{[0-9]+}} 2, i{{[0-9]+}}* [[SIVAR_PRIVATE_CONV]], |
203 | // LAMBDA-32: store i{{[0-9]+}} 2, i{{[0-9]+}}* [[SIVAR_PRIVATE_ADDR]], |
204 | // LAMBDA: [[G_PRIVATE_ADDR_REF:%.+]] = getelementptr inbounds %{{.+}}, %{{.+}}* [[ARG:%.+]], i{{[0-9]+}} 0, i{{[0-9]+}} 0 |
205 | // LAMBDA: store i{{[0-9]+}}* [[G_PRIVATE_ADDR]], i{{[0-9]+}}** [[G_PRIVATE_ADDR_REF]] |
206 | // LAMBDA: [[SIVAR_PRIVATE_ADDR_REF:%.+]] = getelementptr inbounds %{{.+}}, %{{.+}}* [[ARG:%.+]], i{{[0-9]+}} 0, i{{[0-9]+}} 1 |
207 | // LAMBDA-64: store i{{[0-9]+}}* [[SIVAR_PRIVATE_CONV]], i{{[0-9]+}}** [[SIVAR_PRIVATE_ADDR_REF]] |
208 | // LAMBDA-32: store i{{[0-9]+}}* [[SIVAR_PRIVATE_ADDR]], i{{[0-9]+}}** [[SIVAR_PRIVATE_ADDR_REF]] |
209 | // LAMBDA: call{{.*}} void [[INNER_LAMBDA:@.+]](%{{.+}}* [[ARG]]) |
210 | [&]() { |
211 | // LAMBDA: define {{.+}} void [[INNER_LAMBDA]](%{{.+}}* [[ARG_PTR:%.+]]) |
212 | // LAMBDA: store %{{.+}}* [[ARG_PTR]], %{{.+}}** [[ARG_PTR_REF:%.+]], |
213 | g = 2; |
214 | sivar = 4; |
215 | // LAMBDA: [[ARG_PTR:%.+]] = load %{{.+}}*, %{{.+}}** [[ARG_PTR_REF]] |
216 | // LAMBDA: [[G_PTR_REF:%.+]] = getelementptr inbounds %{{.+}}, %{{.+}}* [[ARG_PTR]], i{{[0-9]+}} 0, i{{[0-9]+}} 0 |
217 | // LAMBDA: [[G_REF:%.+]] = load i{{[0-9]+}}*, i{{[0-9]+}}** [[G_PTR_REF]] |
218 | // LAMBDA: [[SIVAR_PTR_REF:%.+]] = getelementptr inbounds %{{.+}}, %{{.+}}* [[ARG_PTR]], i{{[0-9]+}} 0, i{{[0-9]+}} 1 |
219 | // LAMBDA: [[SIVAR_REF:%.+]] = load i{{[0-9]+}}*, i{{[0-9]+}}** [[SIVAR_PTR_REF]] |
220 | // LAMBDA: store i{{[0-9]+}} 4, i{{[0-9]+}}* [[SIVAR_REF]] |
221 | }(); |
222 | } |
223 | }(); |
224 | return 0; |
225 | #elif defined(BLOCKS) |
226 | // BLOCKS: [[G:@.+]] = global i{{[0-9]+}} 1212, |
227 | // BLOCKS-LABEL: @main |
228 | // BLOCKS: call |
229 | // BLOCKS: call {{.*}}void {{%.+}}(i8 |
230 | ^{ |
231 | // BLOCKS: define{{.*}} internal{{.*}} void {{.+}}(i8* |
232 | // BLOCKS: call {{.*}}void {{.+}} @__kmpc_fork_call({{.+}}, i32 2, {{.+}}* [[OMP_REGION:@.+]] to {{.+}}, i32* [[G]], {{.+}}) |
233 | #pragma omp parallel firstprivate(g, sivar) |
234 | { |
235 | // BLOCKS: define{{.*}} internal{{.*}} void [[OMP_REGION]](i32* noalias %{{.+}}, i32* noalias %{{.+}}, i32* dereferenceable(4) %{{.+}}, [[iz:i64|i32]] {{.*}}%{{.+}}) |
236 | // BLOCKS: [[SIVAR_PRIVATE_ADDR:%.+]] = alloca i{{[0-9]+}}, |
237 | // BLOCKS: [[G_PRIVATE_ADDR:%.+]] = alloca i{{[0-9]+}}, align 128 |
238 | // BLOCKS: [[G_REF:%.+]] = load i{{[0-9]+}}*, i{{[0-9]+}}** [[G_REF_ADDR:%.+]] |
239 | // BLOCKS-64: [[SIVAR_PRIVATE_CONV:%.+]] = bitcast i64* [[SIVAR_PRIVATE_ADDR]] to i32* |
240 | // BLOCKS: [[G_VAL:%.+]] = load volatile i{{[0-9]+}}, i{{[0-9]+}}* [[G_REF]], align 128 |
241 | // BLOCKS: store i{{[0-9]+}} [[G_VAL]], i{{[0-9]+}}* [[G_PRIVATE_ADDR]], align 128 |
242 | // BLOCKS-NOT: call {{.*}}void @__kmpc_barrier( |
243 | g = 1; |
244 | sivar = 2; |
245 | // BLOCKS: store i{{[0-9]+}} 1, i{{[0-9]+}}* [[G_PRIVATE_ADDR]], |
246 | // BLOCKS-64: store i{{[0-9]+}} 2, i{{[0-9]+}}* [[SIVAR_PRIVATE_CONV]], |
247 | // BLOCKS-32: store i{{[0-9]+}} 2, i{{[0-9]+}}* [[SIVAR_PRIVATE_ADDR]], |
248 | // BLOCKS-NOT: [[G]]{{[[^:word:]]}} |
249 | // BLOCKS: i{{[0-9]+}}* [[G_PRIVATE_ADDR]] |
250 | // BLOCKS-NOT: [[G]]{{[[^:word:]]}} |
251 | // BLOCKS-NOT: [[SIVAR]]{{[[^:word:]]}} |
252 | // BLOCKS-64: i{{[0-9]+}}* [[SIVAR_PRIVATE_CONV]] |
253 | // BLOCKS-32: i{{[0-9]+}}* [[SIVAR_PRIVATE_ADDR]] |
254 | // BLOCKS-NOT: [[SIVAR]]{{[[^:word:]]}} |
255 | // BLOCKS: call {{.*}}void {{%.+}}(i8 |
256 | ^{ |
257 | // BLOCKS: define {{.+}} void {{@.+}}(i8* |
258 | g = 2; |
259 | sivar = 4; |
260 | // BLOCKS-NOT: [[G]]{{[[^:word:]]}} |
261 | // BLOCKS: store i{{[0-9]+}} 2, i{{[0-9]+}}* |
262 | // BLOCKS-NOT: [[G]]{{[[^:word:]]}} |
263 | // BLOCKS-NOT: [[SIVAR]]{{[[^:word:]]}} |
264 | // BLOCKS: store i{{[0-9]+}} 4, i{{[0-9]+}}* |
265 | // BLOCKS-NOT: [[SIVAR]]{{[[^:word:]]}} |
266 | // BLOCKS: ret |
267 | }(); |
268 | } |
269 | }(); |
270 | return 0; |
271 | // BLOCKS: define {{.+}} @{{.+}}([[SS_TY]]* |
272 | // BLOCKS: getelementptr inbounds [[SS_TY]], [[SS_TY]]* %{{.+}}, i32 0, i32 0 |
273 | // BLOCKS: store i{{[0-9]+}} 0, i{{[0-9]+}}* % |
274 | // BLOCKS: getelementptr inbounds [[SS_TY]], [[SS_TY]]* %{{.+}}, i32 0, i32 1 |
275 | // BLOCKS: store i8 |
276 | // BLOCKS: getelementptr inbounds [[SS_TY]], [[SS_TY]]* %{{.+}}, i32 0, i32 2 |
277 | // BLOCKS: getelementptr inbounds [[SS_TY]], [[SS_TY]]* %{{.+}}, i32 0, i32 0 |
278 | // BLOCKS: getelementptr inbounds [[SS_TY]], [[SS_TY]]* %{{.+}}, i32 0, i32 1 |
279 | // BLOCKS: getelementptr inbounds [[SS_TY]], [[SS_TY]]* %{{.+}}, i32 0, i32 2 |
280 | // BLOCKS: call void (%{{.+}}*, i{{[0-9]+}}, void (i{{[0-9]+}}*, i{{[0-9]+}}*, ...)*, ...) @__kmpc_fork_call(%{{.+}}* @{{.+}}, i{{[0-9]+}} 5, void (i{{[0-9]+}}*, i{{[0-9]+}}*, ...)* bitcast (void (i{{[0-9]+}}*, i{{[0-9]+}}*, [[SS_TY]]*, [[iz]], [[iz]], [[iz]], [4 x i{{[0-9]+}}]*)* [[SS_MICROTASK:@.+]] to void |
281 | // BLOCKS: ret |
282 | |
283 | // BLOCKS: define internal void [[SS_MICROTASK]](i{{[0-9]+}}* noalias [[GTID_ADDR:%.+]], i{{[0-9]+}}* noalias %{{.+}}, [[SS_TY]]* %{{.+}}, [[iz]] {{.+}}, [[iz]] {{.+}}, [[iz]] {{.+}}, [4 x i{{[0-9]+}}]* {{.+}}) |
284 | // BLOCKS-NOT: getelementptr {{.*}}[[SS_TY]], [[SS_TY]]* % |
285 | // BLOCKS: call{{.*}} void |
286 | // BLOCKS: ret void |
287 | |
288 | // BLOCKS: define internal void @{{.+}}(i{{[0-9]+}}* noalias [[GTID_ADDR:%.+]], i{{[0-9]+}}* noalias %{{.+}}, [[SS_TY]]* %{{.+}}, [[iz]] {{.+}}, [[iz]] {{.+}}, [[iz]] {{.+}}) |
289 | // BLOCKS: [[A_PRIV:%.+]] = alloca i{{[0-9]+}}, |
290 | // BLOCKS: [[B_PRIV:%.+]] = alloca i{{[0-9]+}}, |
291 | // BLOCKS: [[C_PRIV:%.+]] = alloca i{{[0-9]+}}, |
292 | // BLOCKS-64: [[A_CONV:%.+]] = bitcast i64* [[A_PRIV]] to i32* |
293 | // BLOCKS-64: store i32* [[A_CONV]], i32** [[REFA:%.+]], |
294 | // BLOCKS-32: store i32* [[A_PRIV]], i32** [[REFA:%.+]], |
295 | // BLOCKS-64: [[B_CONV:%.+]] = bitcast i64* [[B_PRIV]] to i32* |
296 | // BLOCKS-64: [[C_CONV:%.+]] = bitcast i64* [[C_PRIV]] to i32* |
297 | // BLOCKS-64: store i32* [[C_CONV]], i32** [[REFC:%.+]], |
298 | // BLOCKS-32: store i32* [[C_PRIV]], i32** [[REFC:%.+]], |
299 | // BLOCKS-NEXT: [[A_PRIV:%.+]] = load i{{[0-9]+}}*, i{{[0-9]+}}** [[REFA]], |
300 | // BLOCKS-NEXT: [[A_VAL:%.+]] = load i{{[0-9]+}}, i{{[0-9]+}}* [[A_PRIV]], |
301 | // BLOCKS-NEXT: [[INC:%.+]] = add nsw i{{[0-9]+}} [[A_VAL]], 1 |
302 | // BLOCKS-NEXT: store i{{[0-9]+}} [[INC]], i{{[0-9]+}}* [[A_PRIV]], |
303 | // BLOCKS-64-NEXT: [[B_VAL:%.+]] = load i{{[0-9]+}}, i{{[0-9]+}}* [[B_CONV]], |
304 | // BLOCKS-32-NEXT: [[B_VAL:%.+]] = load i{{[0-9]+}}, i{{[0-9]+}}* [[B_PRIV]], |
305 | // BLOCKS-NEXT: [[DEC:%.+]] = add nsw i{{[0-9]+}} [[B_VAL]], -1 |
306 | // BLOCKS-64-NEXT: store i{{[0-9]+}} [[DEC]], i{{[0-9]+}}* [[B_CONV]], |
307 | // BLOCKS-32-NEXT: store i{{[0-9]+}} [[DEC]], i{{[0-9]+}}* [[B_PRIV]], |
308 | // BLOCKS-NEXT: [[C_PRIV:%.+]] = load i{{[0-9]+}}*, i{{[0-9]+}}** [[REFC]], |
309 | // BLOCKS-NEXT: [[C_VAL:%.+]] = load i{{[0-9]+}}, i{{[0-9]+}}* [[C_PRIV]], |
310 | // BLOCKS-NEXT: [[DIV:%.+]] = sdiv i{{[0-9]+}} [[C_VAL]], 1 |
311 | // BLOCKS-NEXT: store i{{[0-9]+}} [[DIV]], i{{[0-9]+}}* [[C_PRIV]], |
312 | // BLOCKS-NEXT: ret void |
313 | #else |
314 | S<float> test; |
315 | int t_var = 0; |
316 | int vec[] = {1, 2}; |
317 | S<float> s_arr[] = {1, 2}; |
318 | S<float> var(3); |
319 | #pragma omp parallel firstprivate(t_var, vec, s_arr, var, sivar) |
320 | { |
321 | vec[0] = t_var; |
322 | s_arr[0] = var; |
323 | sivar = 2; |
324 | } |
325 | #pragma omp parallel firstprivate(t_var) |
326 | {} |
327 | return tmain<int>(); |
328 | #endif |
329 | } |
330 | |
331 | // CHECK: define {{.*}}i{{[0-9]+}} @main() |
332 | // CHECK: [[TEST:%.+]] = alloca [[S_FLOAT_TY]], |
333 | // CHECK: [[T_VAR:%.+]] = alloca i32, |
334 | // CHECK: [[T_VARCAST:%.+]] = alloca [[iz:i64|i32]], |
335 | // CHECK: [[SIVARCAST:%.+]] = alloca [[iz]], |
336 | // CHECK: call {{.*}} [[S_FLOAT_TY_DEF_CONSTR:@.+]]([[S_FLOAT_TY]]* [[TEST]]) |
337 | // CHECK: [[T_VARVAL:%.+]] = load i32, i32* [[T_VAR]], |
338 | // CHECK-64: [[T_VARCONV:%.+]] = bitcast i64* [[T_VARCAST]] to i32* |
339 | // CHECK-64: store i32 [[T_VARVAL]], i32* [[T_VARCONV]], |
340 | // CHECK-32: store i32 [[T_VARVAL]], i32* [[T_VARCAST]], |
341 | // CHECK: [[T_VARPVT:%.+]] = load [[iz]], [[iz]]* [[T_VARCAST]], |
342 | // CHECK: [[SIVARVAL:%.+]] = load i32, i32* @{{.+}}, |
343 | // CHECK-64: [[SIVARCONV:%.+]] = bitcast i64* [[SIVARCAST]] to i32* |
344 | // CHECK-64: store i32 [[SIVARVAL]], i32* [[SIVARCONV]], |
345 | // CHECK-32: store i32 [[SIVARVAL]], i32* [[SIVARCAST]], |
346 | // CHECK: [[SIVARPVT:%.+]] = load [[iz]], [[iz]]* [[SIVARCAST]], |
347 | // CHECK: call {{.*}}void (%{{.+}}*, i{{[0-9]+}}, void (i{{[0-9]+}}*, i{{[0-9]+}}*, ...)*, ...) @__kmpc_fork_call(%{{.+}}* @{{.+}}, i{{[0-9]+}} 5, void (i{{[0-9]+}}*, i{{[0-9]+}}*, ...)* bitcast (void (i{{[0-9]+}}*, i{{[0-9]+}}*, [2 x i32]*, [[iz]], [2 x [[S_FLOAT_TY]]]*, [[S_FLOAT_TY]]*, i{{[0-9]+}})* [[MAIN_MICROTASK:@.+]] to void {{.*}}[[iz]] [[T_VARPVT]],{{.*}}[[iz]] [[SIVARPVT]] |
348 | // CHECK: = call {{.*}}i{{.+}} [[TMAIN_INT:@.+]]() |
349 | // CHECK: call {{.*}} [[S_FLOAT_TY_DESTR:@.+]]([[S_FLOAT_TY]]* |
350 | // CHECK: ret |
351 | // |
352 | // CHECK: define internal {{.*}}void [[MAIN_MICROTASK]](i{{[0-9]+}}* noalias [[GTID_ADDR:%.+]], i{{[0-9]+}}* noalias %{{.+}}, [2 x i32]* dereferenceable(8) %{{.+}}, [[iz]] {{.*}}%{{.+}}, [2 x [[S_FLOAT_TY]]]* dereferenceable(8) %{{.+}}, [[S_FLOAT_TY]]* dereferenceable(4) %{{.+}}, [[iz]] {{.*}}[[SIVAR:%.+]]) |
353 | // CHECK: [[T_VAR_PRIV:%.+]] = alloca i{{[0-9]+}}, |
354 | // CHECK: [[SIVAR7_PRIV:%.+]] = alloca i{{[0-9]+}}, |
355 | // CHECK: [[VEC_PRIV:%.+]] = alloca [2 x i{{[0-9]+}}], |
356 | // CHECK: [[S_ARR_PRIV:%.+]] = alloca [2 x [[S_FLOAT_TY]]], |
357 | // CHECK: [[VAR_PRIV:%.+]] = alloca [[S_FLOAT_TY]], |
358 | // CHECK: store i{{[0-9]+}}* [[GTID_ADDR]], i{{[0-9]+}}** [[GTID_ADDR_ADDR:%.+]], |
359 | |
360 | // CHECK: [[VEC_REF:%.+]] = load [2 x i{{[0-9]+}}]*, [2 x i{{[0-9]+}}]** % |
361 | // CHECK-NOT: load i{{[0-9]+}}*, i{{[0-9]+}}** % |
362 | // CHECK-64: [[T_VAR_CONV:%.+]] = bitcast i64* [[T_VAR_PRIV]] to i32* |
363 | // CHECK: [[S_ARR_REF:%.+]] = load [2 x [[S_FLOAT_TY]]]*, [2 x [[S_FLOAT_TY]]]** % |
364 | // CHECK: [[VAR_REF:%.+]] = load [[S_FLOAT_TY]]*, [[S_FLOAT_TY]]** % |
365 | // CHECK-NOT: load i{{[0-9]+}}*, i{{[0-9]+}}** % |
366 | // CHECK-64: [[SIVAR7_CONV:%.+]] = bitcast i64* [[SIVAR7_PRIV]] to i32* |
367 | // CHECK: [[VEC_DEST:%.+]] = bitcast [2 x i{{[0-9]+}}]* [[VEC_PRIV]] to i8* |
368 | // CHECK: [[VEC_SRC:%.+]] = bitcast [2 x i{{[0-9]+}}]* [[VEC_REF]] to i8* |
369 | // CHECK: call void @llvm.memcpy.{{.+}}(i8* align {{[0-9]+}} [[VEC_DEST]], i8* align {{[0-9]+}} [[VEC_SRC]], |
370 | // CHECK: [[S_ARR_PRIV_BEGIN:%.+]] = getelementptr inbounds [2 x [[S_FLOAT_TY]]], [2 x [[S_FLOAT_TY]]]* [[S_ARR_PRIV]], i{{[0-9]+}} 0, i{{[0-9]+}} 0 |
371 | // CHECK: [[S_ARR_BEGIN:%.+]] = bitcast [2 x [[S_FLOAT_TY]]]* [[S_ARR_REF]] to [[S_FLOAT_TY]]* |
372 | // CHECK: [[S_ARR_PRIV_END:%.+]] = getelementptr [[S_FLOAT_TY]], [[S_FLOAT_TY]]* [[S_ARR_PRIV_BEGIN]], i{{[0-9]+}} 2 |
373 | // CHECK: [[IS_EMPTY:%.+]] = icmp eq [[S_FLOAT_TY]]* [[S_ARR_PRIV_BEGIN]], [[S_ARR_PRIV_END]] |
374 | // CHECK: br i1 [[IS_EMPTY]], label %[[S_ARR_BODY_DONE:.+]], label %[[S_ARR_BODY:.+]] |
375 | // CHECK: [[S_ARR_BODY]] |
376 | // CHECK: call {{.*}} [[ST_TY_DEFAULT_CONSTR:@.+]]([[ST_TY]]* [[ST_TY_TEMP:%.+]]) |
377 | // CHECK: call {{.*}} [[S_FLOAT_TY_COPY_CONSTR:@.+]]([[S_FLOAT_TY]]* {{.+}}, [[S_FLOAT_TY]]* {{.+}}, [[ST_TY]]* [[ST_TY_TEMP]]) |
378 | // CHECK: call {{.*}} [[ST_TY_DESTR:@.+]]([[ST_TY]]* [[ST_TY_TEMP]]) |
379 | // CHECK: br i1 {{.+}}, label %{{.+}}, label %[[S_ARR_BODY]] |
380 | // CHECK: call {{.*}} [[ST_TY_DEFAULT_CONSTR]]([[ST_TY]]* [[ST_TY_TEMP:%.+]]) |
381 | // CHECK: call {{.*}} [[S_FLOAT_TY_COPY_CONSTR]]([[S_FLOAT_TY]]* [[VAR_PRIV]], [[S_FLOAT_TY]]* {{.*}} [[VAR_REF]], [[ST_TY]]* [[ST_TY_TEMP]]) |
382 | // CHECK: call {{.*}} [[ST_TY_DESTR]]([[ST_TY]]* [[ST_TY_TEMP]]) |
383 | |
384 | // CHECK-64: store i{{[0-9]+}} 2, i{{[0-9]+}}* [[SIVAR7_CONV]], |
385 | // CHECK-32: store i{{[0-9]+}} 2, i{{[0-9]+}}* [[SIVAR7_PRIV]], |
386 | |
387 | // CHECK-DAG: call {{.*}} [[S_FLOAT_TY_DESTR]]([[S_FLOAT_TY]]* [[VAR_PRIV]]) |
388 | // CHECK-DAG: call {{.*}} [[S_FLOAT_TY_DESTR]]([[S_FLOAT_TY]]* |
389 | // CHECK: ret void |
390 | // CHECK: define {{.*}} i{{[0-9]+}} [[TMAIN_INT]]() |
391 | // CHECK: [[TEST:%.+]] = alloca [[S_INT_TY]], |
392 | // CHECK: call {{.*}} [[S_INT_TY_DEF_CONSTR:@.+]]([[S_INT_TY]]* [[TEST]]) |
393 | // CHECK: call {{.*}}void (%{{.+}}*, i{{[0-9]+}}, void (i{{[0-9]+}}*, i{{[0-9]+}}*, ...)*, ...) @__kmpc_fork_call(%{{.+}}* @{{.+}}, i{{[0-9]+}} 4, void (i{{[0-9]+}}*, i{{[0-9]+}}*, ...)* bitcast (void (i{{[0-9]+}}*, i{{[0-9]+}}*, [2 x i32]*, i32*, [2 x [[S_INT_TY]]]*, [[S_INT_TY]]*)* [[TMAIN_MICROTASK:@.+]] to void |
394 | // CHECK: call {{.*}} [[S_INT_TY_DESTR:@.+]]([[S_INT_TY]]* |
395 | // CHECK: ret |
396 | // |
397 | // CHECK: define {{.+}} @{{.+}}([[SS_TY]]* |
398 | // CHECK: getelementptr inbounds [[SS_TY]], [[SS_TY]]* %{{.+}}, i32 0, i32 0 |
399 | // CHECK: store i{{[0-9]+}} 0, i{{[0-9]+}}* % |
400 | // CHECK: getelementptr inbounds [[SS_TY]], [[SS_TY]]* %{{.+}}, i32 0, i32 1 |
401 | // CHECK: store i8 |
402 | // CHECK: getelementptr inbounds [[SS_TY]], [[SS_TY]]* %{{.+}}, i32 0, i32 2 |
403 | // CHECK: getelementptr inbounds [[SS_TY]], [[SS_TY]]* %{{.+}}, i32 0, i32 0 |
404 | // CHECK: getelementptr inbounds [[SS_TY]], [[SS_TY]]* %{{.+}}, i32 0, i32 1 |
405 | // CHECK: getelementptr inbounds [[SS_TY]], [[SS_TY]]* %{{.+}}, i32 0, i32 2 |
406 | // CHECK: call void (%{{.+}}*, i{{[0-9]+}}, void (i{{[0-9]+}}*, i{{[0-9]+}}*, ...)*, ...) @__kmpc_fork_call(%{{.+}}* @{{.+}}, i{{[0-9]+}} 5, void (i{{[0-9]+}}*, i{{[0-9]+}}*, ...)* bitcast (void (i{{[0-9]+}}*, i{{[0-9]+}}*, [[SS_TY]]*, [[iz]], [[iz]], [[iz]], [4 x i32]*)* [[SS_MICROTASK:@.+]] to void |
407 | // CHECK: ret |
408 | |
409 | // CHECK: define internal void [[SS_MICROTASK]](i{{[0-9]+}}* noalias [[GTID_ADDR:%.+]], i{{[0-9]+}}* noalias %{{.+}}, [[SS_TY]]* %{{.+}}, [[iz]] {{.+}}, [[iz]] {{.+}}, [[iz]] {{.+}}, [4 x i{{[0-9]+}}]* {{.+}}) |
410 | // CHECK: [[A_PRIV:%.+]] = alloca i{{[0-9]+}}, |
411 | // CHECK: [[B_PRIV:%.+]] = alloca i{{[0-9]+}}, |
412 | // CHECK: [[C_PRIV:%.+]] = alloca i{{[0-9]+}}, |
413 | // CHECK: [[E_PRIV:%.+]] = alloca [4 x i{{[0-9]+}}], |
414 | // CHECK: store i{{[0-9]+}} {{.+}}, i{{[0-9]+}}* [[A_PRIV]] |
415 | // CHECK: store i{{[0-9]+}} {{.+}}, i{{[0-9]+}}* [[B_PRIV]] |
416 | // CHECK: store i{{[0-9]+}} {{.+}}, i{{[0-9]+}}* [[C_PRIV]] |
417 | // CHECK-64: [[A_CONV:%.+]] = bitcast i64* [[A_PRIV:%.+]] to i32* |
418 | // CHECK-64: store i32* [[A_CONV]], i32** [[REFA:%.+]], |
419 | // CHECK-32: store i32* [[A_PRIV]], i32** [[REFA:%.+]], |
420 | // CHECK-64: [[B_CONV:%.+]] = bitcast i64* [[B_PRIV:%.+]] to i32* |
421 | // CHECK-64: [[C_CONV:%.+]] = bitcast i64* [[C_PRIV:%.+]] to i32* |
422 | // CHECK-64: store i32* [[C_CONV]], i32** [[REFC:%.+]], |
423 | // CHECK-32: store i32* [[C_PRIV]], i32** [[REFC:%.+]], |
424 | // CHECK: bitcast [4 x i{{[0-9]+}}]* [[E_PRIV]] to i8* |
425 | // CHECK: bitcast [4 x i{{[0-9]+}}]* %{{.+}} to i8* |
426 | // CHECK: call void @llvm.memcpy |
427 | // CHECK: store [4 x i{{[0-9]+}}]* [[E_PRIV]], [4 x i{{[0-9]+}}]** [[REFE:%.+]], |
428 | // CHECK-NEXT: [[A_PRIV:%.+]] = load i{{[0-9]+}}*, i{{[0-9]+}}** [[REFA]], |
429 | // CHECK-NEXT: [[A_VAL:%.+]] = load i{{[0-9]+}}, i{{[0-9]+}}* [[A_PRIV]], |
430 | // CHECK-NEXT: [[INC:%.+]] = add nsw i{{[0-9]+}} [[A_VAL]], 1 |
431 | // CHECK-NEXT: store i{{[0-9]+}} [[INC]], i{{[0-9]+}}* [[A_PRIV]], |
432 | // CHECK-64-NEXT: [[B_VAL:%.+]] = load i{{[0-9]+}}, i{{[0-9]+}}* [[B_CONV]], |
433 | // CHECK-32-NEXT: [[B_VAL:%.+]] = load i{{[0-9]+}}, i{{[0-9]+}}* [[B_PRIV]], |
434 | // CHECK-NEXT: [[DEC:%.+]] = add nsw i{{[0-9]+}} [[B_VAL]], -1 |
435 | // CHECK-64-NEXT: store i{{[0-9]+}} [[DEC]], i{{[0-9]+}}* [[B_CONV]], |
436 | // CHECK-32-NEXT: store i{{[0-9]+}} [[DEC]], i{{[0-9]+}}* [[B_PRIV]], |
437 | // CHECK-NEXT: [[C_PRIV:%.+]] = load i{{[0-9]+}}*, i{{[0-9]+}}** [[REFC]], |
438 | // CHECK-NEXT: [[C_VAL:%.+]] = load i{{[0-9]+}}, i{{[0-9]+}}* [[C_PRIV]], |
439 | // CHECK-NEXT: [[DIV:%.+]] = sdiv i{{[0-9]+}} [[C_VAL]], 1 |
440 | // CHECK-NEXT: store i{{[0-9]+}} [[DIV]], i{{[0-9]+}}* [[C_PRIV]], |
441 | // CHECK-NEXT: [[E_PRIV:%.+]] = load [4 x i{{[0-9]+}}]*, [4 x i{{[0-9]+}}]** [[REFE]], |
442 | // CHECK-NEXT: [[E_PRIV_2:%.+]] = getelementptr inbounds [4 x i{{[0-9]+}}], [4 x i{{[0-9]+}}]* [[E_PRIV]], i{{[0-9]+}} 0, i{{[0-9]+}} 2 |
443 | // CHECK-NEXT: store i32 1111, i32* [[E_PRIV_2]], |
444 | // CHECK-NEXT: ret void |
445 | |
446 | // CHECK: define internal {{.*}}void [[TMAIN_MICROTASK]](i{{[0-9]+}}* noalias [[GTID_ADDR:%.+]], i{{[0-9]+}}* noalias %{{.+}}, [2 x i32]* dereferenceable(8) %{{.+}}, i32* dereferenceable(4) %{{.+}}, [2 x [[S_INT_TY]]]* dereferenceable(8) %{{.+}}, [[S_INT_TY]]* dereferenceable(4) %{{.+}}) |
447 | // CHECK: [[T_VAR_PRIV:%.+]] = alloca i{{[0-9]+}}, align 128 |
448 | // CHECK: [[VEC_PRIV:%.+]] = alloca [2 x i{{[0-9]+}}], align 128 |
449 | // CHECK: [[S_ARR_PRIV:%.+]] = alloca [2 x [[S_INT_TY]]], align 128 |
450 | // CHECK: [[VAR_PRIV:%.+]] = alloca [[S_INT_TY]], align 128 |
451 | // CHECK: store i{{[0-9]+}}* [[GTID_ADDR]], i{{[0-9]+}}** [[GTID_ADDR_ADDR:%.+]], |
452 | |
453 | // CHECK: [[VEC_REF:%.+]] = load [2 x i{{[0-9]+}}]*, [2 x i{{[0-9]+}}]** % |
454 | // CHECK: [[T_VAR_REF:%.+]] = load i{{[0-9]+}}*, i{{[0-9]+}}** % |
455 | // CHECK: [[S_ARR_REF:%.+]] = load [2 x [[S_INT_TY]]]*, [2 x [[S_INT_TY]]]** % |
456 | // CHECK: [[VAR_REF:%.+]] = load [[S_INT_TY]]*, [[S_INT_TY]]** % |
457 | |
458 | // CHECK: [[T_VAR_VAL:%.+]] = load i{{[0-9]+}}, i{{[0-9]+}}* [[T_VAR_REF]], align 128 |
459 | // CHECK: store i{{[0-9]+}} [[T_VAR_VAL]], i{{[0-9]+}}* [[T_VAR_PRIV]], align 128 |
460 | // CHECK: [[VEC_DEST:%.+]] = bitcast [2 x i{{[0-9]+}}]* [[VEC_PRIV]] to i8* |
461 | // CHECK: [[VEC_SRC:%.+]] = bitcast [2 x i{{[0-9]+}}]* [[VEC_REF]] to i8* |
462 | // CHECK: call void @llvm.memcpy.{{.+}}(i8* align 128 [[VEC_DEST]], i8* align 128 [[VEC_SRC]], i{{[0-9]+}} {{[0-9]+}}, i1 |
463 | // CHECK: [[S_ARR_PRIV_BEGIN:%.+]] = getelementptr inbounds [2 x [[S_INT_TY]]], [2 x [[S_INT_TY]]]* [[S_ARR_PRIV]], i{{[0-9]+}} 0, i{{[0-9]+}} 0 |
464 | // CHECK: [[S_ARR_BEGIN:%.+]] = bitcast [2 x [[S_INT_TY]]]* [[S_ARR_REF]] to [[S_INT_TY]]* |
465 | // CHECK: [[S_ARR_PRIV_END:%.+]] = getelementptr [[S_INT_TY]], [[S_INT_TY]]* [[S_ARR_PRIV_BEGIN]], i{{[0-9]+}} 2 |
466 | // CHECK: [[IS_EMPTY:%.+]] = icmp eq [[S_INT_TY]]* [[S_ARR_PRIV_BEGIN]], [[S_ARR_PRIV_END]] |
467 | // CHECK: br i1 [[IS_EMPTY]], label %[[S_ARR_BODY_DONE:.+]], label %[[S_ARR_BODY:.+]] |
468 | // CHECK: [[S_ARR_BODY]] |
469 | // CHECK: call {{.*}} [[ST_TY_DEFAULT_CONSTR]]([[ST_TY]]* [[ST_TY_TEMP:%.+]]) |
470 | // CHECK: call {{.*}} [[S_INT_TY_COPY_CONSTR:@.+]]([[S_INT_TY]]* {{.+}}, [[S_INT_TY]]* {{.+}}, [[ST_TY]]* [[ST_TY_TEMP]]) |
471 | // CHECK: call {{.*}} [[ST_TY_DESTR]]([[ST_TY]]* [[ST_TY_TEMP]]) |
472 | // CHECK: br i1 {{.+}}, label %{{.+}}, label %[[S_ARR_BODY]] |
473 | // CHECK: call {{.*}} [[ST_TY_DEFAULT_CONSTR]]([[ST_TY]]* [[ST_TY_TEMP:%.+]]) |
474 | // CHECK: call {{.*}} [[S_INT_TY_COPY_CONSTR]]([[S_INT_TY]]* [[VAR_PRIV]], [[S_INT_TY]]* {{.*}} [[VAR_REF]], [[ST_TY]]* [[ST_TY_TEMP]]) |
475 | // CHECK: call {{.*}} [[ST_TY_DESTR]]([[ST_TY]]* [[ST_TY_TEMP]]) |
476 | // CHECK-NOT: call {{.*}}void @__kmpc_barrier( |
477 | // CHECK-DAG: call {{.*}} [[S_INT_TY_DESTR]]([[S_INT_TY]]* [[VAR_PRIV]]) |
478 | // CHECK-DAG: call {{.*}} [[S_INT_TY_DESTR]]([[S_INT_TY]]* |
479 | // CHECK: ret void |
480 | |
481 | #endif |
482 | #else |
483 | struct St { |
484 | int a, b; |
485 | St() : a(0), b(0) {} |
486 | St(const St &) { } |
487 | ~St() {} |
488 | void St_func(St s[2], int n, long double vla1[n]) { |
489 | double vla2[n][n] __attribute__((aligned(128))); |
490 | a = b; |
491 | #pragma omp parallel firstprivate(s, vla1, vla2) |
492 | vla1[b] = vla2[1][n - 1] = a = b; |
493 | } |
494 | }; |
495 | |
496 | // ARRAY-LABEL: array_func |
497 | void array_func(float a[3], St s[2], int n, long double vla1[n]) { |
498 | double vla2[n][n] __attribute__((aligned(128))); |
499 | // ARRAY: @__kmpc_fork_call( |
500 | // ARRAY-DAG: [[PRIV_S:%.+]] = alloca %struct.St*, |
501 | // ARRAY-DAG: [[PRIV_VLA1:%.+]] = alloca x86_fp80*, |
502 | // ARRAY-DAG: [[PRIV_A:%.+]] = alloca float*, |
503 | // ARRAY-DAG: [[PRIV_VLA2:%.+]] = alloca double*, |
504 | // ARRAY-DAG: store %struct.St* %{{.+}}, %struct.St** [[PRIV_S]], |
505 | // ARRAY-DAG: store x86_fp80* %{{.+}}, x86_fp80** [[PRIV_VLA1]], |
506 | // ARRAY-DAG: store float* %{{.+}}, float** [[PRIV_A]], |
507 | // ARRAY-DAG: store double* %{{.+}}, double** [[PRIV_VLA2]], |
508 | // ARRAY: call i8* @llvm.stacksave() |
509 | // ARRAY: [[SIZE:%.+]] = mul nuw i64 %{{.+}}, 8 |
510 | // ARRAY: call void @llvm.memcpy.p0i8.p0i8.i64(i8* align 128 %{{.+}}, i8* align 128 %{{.+}}, i64 [[SIZE]], i1 false) |
511 | #pragma omp parallel firstprivate(a, s, vla1, vla2) |
512 | s[0].St_func(s, n, vla1); |
513 | ; |
514 | } |
515 | |
516 | // ARRAY-LABEL: St_func |
517 | // ARRAY: @__kmpc_fork_call( |
518 | // ARRAY-DAG: [[PRIV_VLA1:%.+]] = alloca x86_fp80*, |
519 | // ARRAY-DAG: [[PRIV_S:%.+]] = alloca %struct.St*, |
520 | // ARRAY-DAG: [[PRIV_VLA2:%.+]] = alloca double*, |
521 | // ARRAY-DAG: store %struct.St* %{{.+}}, %struct.St** [[PRIV_S]], |
522 | // ARRAY-DAG: store x86_fp80* %{{.+}}, x86_fp80** [[PRIV_VLA1]], |
523 | // ARRAY-DAG: store double* %{{.+}}, double** [[PRIV_VLA2]], |
524 | // ARRAY: call i8* @llvm.stacksave() |
525 | // ARRAY: [[SIZE:%.+]] = mul nuw i64 %{{.+}}, 8 |
526 | // ARRAY: call void @llvm.memcpy.p0i8.p0i8.i64(i8* align 128 %{{.+}}, i8* align 128 %{{.+}}, i64 [[SIZE]], i1 false) |
527 | #endif |
528 | |
529 | |
530 | |