xref: /llvm-project/clang/test/OpenMP/parallel_if_codegen.cpp (revision 956cae2f09b21429dbcb02066c99e35a239aa4bf)
1 // NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --function-signature --include-generated-funcs --replace-function-regex "__omp_offloading_[0-9a-z]+_[0-9a-z]+" --prefix-filecheck-ir-name _
2 // RUN: %clang_cc1 -verify -fopenmp -fopenmp-version=45 -x c++ -triple x86_64-unknown-linux -emit-llvm %s -o - | FileCheck %s --check-prefix=CHECK1
3 // RUN: %clang_cc1 -fopenmp -fopenmp-version=45 -x c++ -std=c++11 -triple x86_64-unknown-linux -emit-pch -o %t %s
4 // RUN: %clang_cc1 -fopenmp -fopenmp-version=45 -x c++ -triple x86_64-unknown-linux -std=c++11 -include-pch %t -verify %s -emit-llvm -o - | FileCheck %s --check-prefix=CHECK2
5 
6 // RUN: %clang_cc1 -verify -fopenmp-simd -fopenmp-version=45 -x c++ -triple x86_64-unknown-linux -emit-llvm %s -o - | FileCheck %s --check-prefix=CHECK3
7 // RUN: %clang_cc1 -fopenmp-simd -fopenmp-version=45 -x c++ -std=c++11 -triple x86_64-unknown-linux -emit-pch -o %t %s
8 // RUN: %clang_cc1 -fopenmp-simd -fopenmp-version=45 -x c++ -triple x86_64-unknown-linux -std=c++11 -include-pch %t -verify %s -emit-llvm -o - | FileCheck %s --check-prefix=CHECK4
9 
10 // RUN: %clang_cc1 -verify -fopenmp -x c++ -triple x86_64-unknown-linux -emit-llvm %s -disable-O0-optnone -o - | FileCheck %s --check-prefix=CHECK5
11 // RUN: %clang_cc1 -verify -fopenmp -x c++ -triple x86_64-unknown-linux -emit-llvm %s -o - | FileCheck %s --check-prefix=CHECK6
12 // RUN: %clang_cc1 -fopenmp -x c++ -std=c++11 -triple x86_64-unknown-linux -emit-pch -o %t %s
13 // RUN: %clang_cc1 -fopenmp -x c++ -triple x86_64-unknown-linux -std=c++11 -include-pch %t -verify %s -emit-llvm -o - | FileCheck %s --check-prefix=CHECK7
14 
15 // RUN: %clang_cc1 -verify -fopenmp-simd -x c++ -triple x86_64-unknown-linux -emit-llvm %s -o - | FileCheck %s --check-prefix=CHECK8
16 // RUN: %clang_cc1 -fopenmp-simd -x c++ -std=c++11 -triple x86_64-unknown-linux -emit-pch -o %t %s
17 // RUN: %clang_cc1 -fopenmp-simd -x c++ -triple x86_64-unknown-linux -std=c++11 -include-pch %t -verify %s -emit-llvm -o - | FileCheck %s --check-prefix=CHECK9
18 
19 // expected-no-diagnostics
20 #ifndef HEADER
21 #define HEADER
22 
23 void fn1();
24 void fn2();
25 void fn3();
26 void fn4();
27 void fn5();
28 void fn6();
29 
30 int Arg;
31 
32 void gtid_test() {
33 #pragma omp parallel
34 #pragma omp parallel if (parallel: false)
35   gtid_test();
36 }
37 
38 
39 
40 template <typename T>
41 int tmain(T Arg) {
42 #pragma omp parallel if (true)
43   fn1();
44 #pragma omp parallel if (false)
45   fn2();
46 #pragma omp parallel if (parallel: Arg)
47   fn3();
48   return 0;
49 }
50 
51 int main() {
52 #pragma omp parallel if (true)
53   fn4();
54 #pragma omp parallel if (false)
55   fn5();
56 
57 #pragma omp parallel if (Arg)
58   fn6();
59   return tmain(Arg);
60 }
61 
62 
63 
64 
65 
66 
67 
68 
69 #endif
70 // CHECK1-LABEL: define {{[^@]+}}@_Z9gtid_testv
71 // CHECK1-SAME: () #[[ATTR0:[0-9]+]] {
72 // CHECK1-NEXT:  entry:
73 // CHECK1-NEXT:    call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB1:[0-9]+]], i32 0, void (i32*, i32*, ...)* bitcast (void (i32*, i32*)* @.omp_outlined. to void (i32*, i32*, ...)*))
74 // CHECK1-NEXT:    ret void
75 //
76 //
77 // CHECK1-LABEL: define {{[^@]+}}@.omp_outlined.
78 // CHECK1-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]]) #[[ATTR1:[0-9]+]] {
79 // CHECK1-NEXT:  entry:
80 // CHECK1-NEXT:    [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
81 // CHECK1-NEXT:    [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
82 // CHECK1-NEXT:    [[DOTBOUND_ZERO_ADDR:%.*]] = alloca i32, align 4
83 // CHECK1-NEXT:    store i32 0, i32* [[DOTBOUND_ZERO_ADDR]], align 4
84 // CHECK1-NEXT:    store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
85 // CHECK1-NEXT:    store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
86 // CHECK1-NEXT:    [[TMP0:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
87 // CHECK1-NEXT:    [[TMP1:%.*]] = load i32, i32* [[TMP0]], align 4
88 // CHECK1-NEXT:    call void @__kmpc_serialized_parallel(%struct.ident_t* @[[GLOB1]], i32 [[TMP1]])
89 // CHECK1-NEXT:    [[TMP2:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
90 // CHECK1-NEXT:    call void @.omp_outlined..1(i32* [[TMP2]], i32* [[DOTBOUND_ZERO_ADDR]]) #[[ATTR2:[0-9]+]]
91 // CHECK1-NEXT:    call void @__kmpc_end_serialized_parallel(%struct.ident_t* @[[GLOB1]], i32 [[TMP1]])
92 // CHECK1-NEXT:    ret void
93 //
94 //
95 // CHECK1-LABEL: define {{[^@]+}}@.omp_outlined..1
96 // CHECK1-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]]) #[[ATTR1]] {
97 // CHECK1-NEXT:  entry:
98 // CHECK1-NEXT:    [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
99 // CHECK1-NEXT:    [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
100 // CHECK1-NEXT:    store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
101 // CHECK1-NEXT:    store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
102 // CHECK1-NEXT:    call void @_Z9gtid_testv()
103 // CHECK1-NEXT:    ret void
104 //
105 //
106 // CHECK1-LABEL: define {{[^@]+}}@main
107 // CHECK1-SAME: () #[[ATTR3:[0-9]+]] {
108 // CHECK1-NEXT:  entry:
109 // CHECK1-NEXT:    [[RETVAL:%.*]] = alloca i32, align 4
110 // CHECK1-NEXT:    [[DOTTHREADID_TEMP_:%.*]] = alloca i32, align 4
111 // CHECK1-NEXT:    [[DOTBOUND_ZERO_ADDR:%.*]] = alloca i32, align 4
112 // CHECK1-NEXT:    [[DOTTHREADID_TEMP_1:%.*]] = alloca i32, align 4
113 // CHECK1-NEXT:    [[DOTBOUND_ZERO_ADDR2:%.*]] = alloca i32, align 4
114 // CHECK1-NEXT:    store i32 0, i32* [[DOTBOUND_ZERO_ADDR2]], align 4
115 // CHECK1-NEXT:    store i32 0, i32* [[DOTBOUND_ZERO_ADDR]], align 4
116 // CHECK1-NEXT:    [[TMP0:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]])
117 // CHECK1-NEXT:    store i32 0, i32* [[RETVAL]], align 4
118 // CHECK1-NEXT:    call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB1]], i32 0, void (i32*, i32*, ...)* bitcast (void (i32*, i32*)* @.omp_outlined..2 to void (i32*, i32*, ...)*))
119 // CHECK1-NEXT:    call void @__kmpc_serialized_parallel(%struct.ident_t* @[[GLOB1]], i32 [[TMP0]])
120 // CHECK1-NEXT:    store i32 [[TMP0]], i32* [[DOTTHREADID_TEMP_]], align 4
121 // CHECK1-NEXT:    call void @.omp_outlined..3(i32* [[DOTTHREADID_TEMP_]], i32* [[DOTBOUND_ZERO_ADDR]]) #[[ATTR2]]
122 // CHECK1-NEXT:    call void @__kmpc_end_serialized_parallel(%struct.ident_t* @[[GLOB1]], i32 [[TMP0]])
123 // CHECK1-NEXT:    [[TMP1:%.*]] = load i32, i32* @Arg, align 4
124 // CHECK1-NEXT:    [[TOBOOL:%.*]] = icmp ne i32 [[TMP1]], 0
125 // CHECK1-NEXT:    br i1 [[TOBOOL]], label [[OMP_IF_THEN:%.*]], label [[OMP_IF_ELSE:%.*]]
126 // CHECK1:       omp_if.then:
127 // CHECK1-NEXT:    call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB1]], i32 0, void (i32*, i32*, ...)* bitcast (void (i32*, i32*)* @.omp_outlined..4 to void (i32*, i32*, ...)*))
128 // CHECK1-NEXT:    br label [[OMP_IF_END:%.*]]
129 // CHECK1:       omp_if.else:
130 // CHECK1-NEXT:    call void @__kmpc_serialized_parallel(%struct.ident_t* @[[GLOB1]], i32 [[TMP0]])
131 // CHECK1-NEXT:    store i32 [[TMP0]], i32* [[DOTTHREADID_TEMP_1]], align 4
132 // CHECK1-NEXT:    call void @.omp_outlined..4(i32* [[DOTTHREADID_TEMP_1]], i32* [[DOTBOUND_ZERO_ADDR2]]) #[[ATTR2]]
133 // CHECK1-NEXT:    call void @__kmpc_end_serialized_parallel(%struct.ident_t* @[[GLOB1]], i32 [[TMP0]])
134 // CHECK1-NEXT:    br label [[OMP_IF_END]]
135 // CHECK1:       omp_if.end:
136 // CHECK1-NEXT:    [[TMP2:%.*]] = load i32, i32* @Arg, align 4
137 // CHECK1-NEXT:    [[CALL:%.*]] = call i32 @_Z5tmainIiEiT_(i32 [[TMP2]])
138 // CHECK1-NEXT:    ret i32 [[CALL]]
139 //
140 //
141 // CHECK1-LABEL: define {{[^@]+}}@.omp_outlined..2
142 // CHECK1-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]]) #[[ATTR1]] {
143 // CHECK1-NEXT:  entry:
144 // CHECK1-NEXT:    [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
145 // CHECK1-NEXT:    [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
146 // CHECK1-NEXT:    store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
147 // CHECK1-NEXT:    store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
148 // CHECK1-NEXT:    call void @_Z3fn4v()
149 // CHECK1-NEXT:    ret void
150 //
151 //
152 // CHECK1-LABEL: define {{[^@]+}}@.omp_outlined..3
153 // CHECK1-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]]) #[[ATTR1]] {
154 // CHECK1-NEXT:  entry:
155 // CHECK1-NEXT:    [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
156 // CHECK1-NEXT:    [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
157 // CHECK1-NEXT:    store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
158 // CHECK1-NEXT:    store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
159 // CHECK1-NEXT:    call void @_Z3fn5v()
160 // CHECK1-NEXT:    ret void
161 //
162 //
163 // CHECK1-LABEL: define {{[^@]+}}@.omp_outlined..4
164 // CHECK1-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]]) #[[ATTR1]] {
165 // CHECK1-NEXT:  entry:
166 // CHECK1-NEXT:    [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
167 // CHECK1-NEXT:    [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
168 // CHECK1-NEXT:    store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
169 // CHECK1-NEXT:    store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
170 // CHECK1-NEXT:    call void @_Z3fn6v()
171 // CHECK1-NEXT:    ret void
172 //
173 //
174 // CHECK1-LABEL: define {{[^@]+}}@_Z5tmainIiEiT_
175 // CHECK1-SAME: (i32 [[ARG:%.*]]) #[[ATTR0]] comdat {
176 // CHECK1-NEXT:  entry:
177 // CHECK1-NEXT:    [[ARG_ADDR:%.*]] = alloca i32, align 4
178 // CHECK1-NEXT:    [[DOTTHREADID_TEMP_:%.*]] = alloca i32, align 4
179 // CHECK1-NEXT:    [[DOTBOUND_ZERO_ADDR:%.*]] = alloca i32, align 4
180 // CHECK1-NEXT:    [[DOTTHREADID_TEMP_1:%.*]] = alloca i32, align 4
181 // CHECK1-NEXT:    [[DOTBOUND_ZERO_ADDR2:%.*]] = alloca i32, align 4
182 // CHECK1-NEXT:    store i32 0, i32* [[DOTBOUND_ZERO_ADDR2]], align 4
183 // CHECK1-NEXT:    store i32 0, i32* [[DOTBOUND_ZERO_ADDR]], align 4
184 // CHECK1-NEXT:    [[TMP0:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]])
185 // CHECK1-NEXT:    store i32 [[ARG]], i32* [[ARG_ADDR]], align 4
186 // CHECK1-NEXT:    call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB1]], i32 0, void (i32*, i32*, ...)* bitcast (void (i32*, i32*)* @.omp_outlined..5 to void (i32*, i32*, ...)*))
187 // CHECK1-NEXT:    call void @__kmpc_serialized_parallel(%struct.ident_t* @[[GLOB1]], i32 [[TMP0]])
188 // CHECK1-NEXT:    store i32 [[TMP0]], i32* [[DOTTHREADID_TEMP_]], align 4
189 // CHECK1-NEXT:    call void @.omp_outlined..6(i32* [[DOTTHREADID_TEMP_]], i32* [[DOTBOUND_ZERO_ADDR]]) #[[ATTR2]]
190 // CHECK1-NEXT:    call void @__kmpc_end_serialized_parallel(%struct.ident_t* @[[GLOB1]], i32 [[TMP0]])
191 // CHECK1-NEXT:    [[TMP1:%.*]] = load i32, i32* [[ARG_ADDR]], align 4
192 // CHECK1-NEXT:    [[TOBOOL:%.*]] = icmp ne i32 [[TMP1]], 0
193 // CHECK1-NEXT:    br i1 [[TOBOOL]], label [[OMP_IF_THEN:%.*]], label [[OMP_IF_ELSE:%.*]]
194 // CHECK1:       omp_if.then:
195 // CHECK1-NEXT:    call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB1]], i32 0, void (i32*, i32*, ...)* bitcast (void (i32*, i32*)* @.omp_outlined..7 to void (i32*, i32*, ...)*))
196 // CHECK1-NEXT:    br label [[OMP_IF_END:%.*]]
197 // CHECK1:       omp_if.else:
198 // CHECK1-NEXT:    call void @__kmpc_serialized_parallel(%struct.ident_t* @[[GLOB1]], i32 [[TMP0]])
199 // CHECK1-NEXT:    store i32 [[TMP0]], i32* [[DOTTHREADID_TEMP_1]], align 4
200 // CHECK1-NEXT:    call void @.omp_outlined..7(i32* [[DOTTHREADID_TEMP_1]], i32* [[DOTBOUND_ZERO_ADDR2]]) #[[ATTR2]]
201 // CHECK1-NEXT:    call void @__kmpc_end_serialized_parallel(%struct.ident_t* @[[GLOB1]], i32 [[TMP0]])
202 // CHECK1-NEXT:    br label [[OMP_IF_END]]
203 // CHECK1:       omp_if.end:
204 // CHECK1-NEXT:    ret i32 0
205 //
206 //
207 // CHECK1-LABEL: define {{[^@]+}}@.omp_outlined..5
208 // CHECK1-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]]) #[[ATTR1]] {
209 // CHECK1-NEXT:  entry:
210 // CHECK1-NEXT:    [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
211 // CHECK1-NEXT:    [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
212 // CHECK1-NEXT:    store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
213 // CHECK1-NEXT:    store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
214 // CHECK1-NEXT:    call void @_Z3fn1v()
215 // CHECK1-NEXT:    ret void
216 //
217 //
218 // CHECK1-LABEL: define {{[^@]+}}@.omp_outlined..6
219 // CHECK1-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]]) #[[ATTR1]] {
220 // CHECK1-NEXT:  entry:
221 // CHECK1-NEXT:    [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
222 // CHECK1-NEXT:    [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
223 // CHECK1-NEXT:    store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
224 // CHECK1-NEXT:    store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
225 // CHECK1-NEXT:    call void @_Z3fn2v()
226 // CHECK1-NEXT:    ret void
227 //
228 //
229 // CHECK1-LABEL: define {{[^@]+}}@.omp_outlined..7
230 // CHECK1-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]]) #[[ATTR1]] {
231 // CHECK1-NEXT:  entry:
232 // CHECK1-NEXT:    [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
233 // CHECK1-NEXT:    [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
234 // CHECK1-NEXT:    store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
235 // CHECK1-NEXT:    store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
236 // CHECK1-NEXT:    call void @_Z3fn3v()
237 // CHECK1-NEXT:    ret void
238 //
239 //
240 // CHECK2-LABEL: define {{[^@]+}}@_Z9gtid_testv
241 // CHECK2-SAME: () #[[ATTR0:[0-9]+]] {
242 // CHECK2-NEXT:  entry:
243 // CHECK2-NEXT:    call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB1:[0-9]+]], i32 0, void (i32*, i32*, ...)* bitcast (void (i32*, i32*)* @.omp_outlined. to void (i32*, i32*, ...)*))
244 // CHECK2-NEXT:    ret void
245 //
246 //
247 // CHECK2-LABEL: define {{[^@]+}}@.omp_outlined.
248 // CHECK2-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]]) #[[ATTR1:[0-9]+]] {
249 // CHECK2-NEXT:  entry:
250 // CHECK2-NEXT:    [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
251 // CHECK2-NEXT:    [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
252 // CHECK2-NEXT:    [[DOTBOUND_ZERO_ADDR:%.*]] = alloca i32, align 4
253 // CHECK2-NEXT:    store i32 0, i32* [[DOTBOUND_ZERO_ADDR]], align 4
254 // CHECK2-NEXT:    store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
255 // CHECK2-NEXT:    store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
256 // CHECK2-NEXT:    [[TMP0:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
257 // CHECK2-NEXT:    [[TMP1:%.*]] = load i32, i32* [[TMP0]], align 4
258 // CHECK2-NEXT:    call void @__kmpc_serialized_parallel(%struct.ident_t* @[[GLOB1]], i32 [[TMP1]])
259 // CHECK2-NEXT:    [[TMP2:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
260 // CHECK2-NEXT:    call void @.omp_outlined..1(i32* [[TMP2]], i32* [[DOTBOUND_ZERO_ADDR]]) #[[ATTR2:[0-9]+]]
261 // CHECK2-NEXT:    call void @__kmpc_end_serialized_parallel(%struct.ident_t* @[[GLOB1]], i32 [[TMP1]])
262 // CHECK2-NEXT:    ret void
263 //
264 //
265 // CHECK2-LABEL: define {{[^@]+}}@.omp_outlined..1
266 // CHECK2-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]]) #[[ATTR1]] {
267 // CHECK2-NEXT:  entry:
268 // CHECK2-NEXT:    [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
269 // CHECK2-NEXT:    [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
270 // CHECK2-NEXT:    store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
271 // CHECK2-NEXT:    store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
272 // CHECK2-NEXT:    call void @_Z9gtid_testv()
273 // CHECK2-NEXT:    ret void
274 //
275 //
276 // CHECK2-LABEL: define {{[^@]+}}@main
277 // CHECK2-SAME: () #[[ATTR3:[0-9]+]] {
278 // CHECK2-NEXT:  entry:
279 // CHECK2-NEXT:    [[RETVAL:%.*]] = alloca i32, align 4
280 // CHECK2-NEXT:    [[DOTTHREADID_TEMP_:%.*]] = alloca i32, align 4
281 // CHECK2-NEXT:    [[DOTBOUND_ZERO_ADDR:%.*]] = alloca i32, align 4
282 // CHECK2-NEXT:    [[DOTTHREADID_TEMP_1:%.*]] = alloca i32, align 4
283 // CHECK2-NEXT:    [[DOTBOUND_ZERO_ADDR2:%.*]] = alloca i32, align 4
284 // CHECK2-NEXT:    store i32 0, i32* [[DOTBOUND_ZERO_ADDR2]], align 4
285 // CHECK2-NEXT:    store i32 0, i32* [[DOTBOUND_ZERO_ADDR]], align 4
286 // CHECK2-NEXT:    [[TMP0:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]])
287 // CHECK2-NEXT:    store i32 0, i32* [[RETVAL]], align 4
288 // CHECK2-NEXT:    call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB1]], i32 0, void (i32*, i32*, ...)* bitcast (void (i32*, i32*)* @.omp_outlined..2 to void (i32*, i32*, ...)*))
289 // CHECK2-NEXT:    call void @__kmpc_serialized_parallel(%struct.ident_t* @[[GLOB1]], i32 [[TMP0]])
290 // CHECK2-NEXT:    store i32 [[TMP0]], i32* [[DOTTHREADID_TEMP_]], align 4
291 // CHECK2-NEXT:    call void @.omp_outlined..3(i32* [[DOTTHREADID_TEMP_]], i32* [[DOTBOUND_ZERO_ADDR]]) #[[ATTR2]]
292 // CHECK2-NEXT:    call void @__kmpc_end_serialized_parallel(%struct.ident_t* @[[GLOB1]], i32 [[TMP0]])
293 // CHECK2-NEXT:    [[TMP1:%.*]] = load i32, i32* @Arg, align 4
294 // CHECK2-NEXT:    [[TOBOOL:%.*]] = icmp ne i32 [[TMP1]], 0
295 // CHECK2-NEXT:    br i1 [[TOBOOL]], label [[OMP_IF_THEN:%.*]], label [[OMP_IF_ELSE:%.*]]
296 // CHECK2:       omp_if.then:
297 // CHECK2-NEXT:    call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB1]], i32 0, void (i32*, i32*, ...)* bitcast (void (i32*, i32*)* @.omp_outlined..4 to void (i32*, i32*, ...)*))
298 // CHECK2-NEXT:    br label [[OMP_IF_END:%.*]]
299 // CHECK2:       omp_if.else:
300 // CHECK2-NEXT:    call void @__kmpc_serialized_parallel(%struct.ident_t* @[[GLOB1]], i32 [[TMP0]])
301 // CHECK2-NEXT:    store i32 [[TMP0]], i32* [[DOTTHREADID_TEMP_1]], align 4
302 // CHECK2-NEXT:    call void @.omp_outlined..4(i32* [[DOTTHREADID_TEMP_1]], i32* [[DOTBOUND_ZERO_ADDR2]]) #[[ATTR2]]
303 // CHECK2-NEXT:    call void @__kmpc_end_serialized_parallel(%struct.ident_t* @[[GLOB1]], i32 [[TMP0]])
304 // CHECK2-NEXT:    br label [[OMP_IF_END]]
305 // CHECK2:       omp_if.end:
306 // CHECK2-NEXT:    [[TMP2:%.*]] = load i32, i32* @Arg, align 4
307 // CHECK2-NEXT:    [[CALL:%.*]] = call i32 @_Z5tmainIiEiT_(i32 [[TMP2]])
308 // CHECK2-NEXT:    ret i32 [[CALL]]
309 //
310 //
311 // CHECK2-LABEL: define {{[^@]+}}@.omp_outlined..2
312 // CHECK2-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]]) #[[ATTR1]] {
313 // CHECK2-NEXT:  entry:
314 // CHECK2-NEXT:    [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
315 // CHECK2-NEXT:    [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
316 // CHECK2-NEXT:    store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
317 // CHECK2-NEXT:    store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
318 // CHECK2-NEXT:    call void @_Z3fn4v()
319 // CHECK2-NEXT:    ret void
320 //
321 //
322 // CHECK2-LABEL: define {{[^@]+}}@.omp_outlined..3
323 // CHECK2-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]]) #[[ATTR1]] {
324 // CHECK2-NEXT:  entry:
325 // CHECK2-NEXT:    [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
326 // CHECK2-NEXT:    [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
327 // CHECK2-NEXT:    store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
328 // CHECK2-NEXT:    store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
329 // CHECK2-NEXT:    call void @_Z3fn5v()
330 // CHECK2-NEXT:    ret void
331 //
332 //
333 // CHECK2-LABEL: define {{[^@]+}}@.omp_outlined..4
334 // CHECK2-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]]) #[[ATTR1]] {
335 // CHECK2-NEXT:  entry:
336 // CHECK2-NEXT:    [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
337 // CHECK2-NEXT:    [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
338 // CHECK2-NEXT:    store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
339 // CHECK2-NEXT:    store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
340 // CHECK2-NEXT:    call void @_Z3fn6v()
341 // CHECK2-NEXT:    ret void
342 //
343 //
344 // CHECK2-LABEL: define {{[^@]+}}@_Z5tmainIiEiT_
345 // CHECK2-SAME: (i32 [[ARG:%.*]]) #[[ATTR0]] comdat {
346 // CHECK2-NEXT:  entry:
347 // CHECK2-NEXT:    [[ARG_ADDR:%.*]] = alloca i32, align 4
348 // CHECK2-NEXT:    [[DOTTHREADID_TEMP_:%.*]] = alloca i32, align 4
349 // CHECK2-NEXT:    [[DOTBOUND_ZERO_ADDR:%.*]] = alloca i32, align 4
350 // CHECK2-NEXT:    [[DOTTHREADID_TEMP_1:%.*]] = alloca i32, align 4
351 // CHECK2-NEXT:    [[DOTBOUND_ZERO_ADDR2:%.*]] = alloca i32, align 4
352 // CHECK2-NEXT:    store i32 0, i32* [[DOTBOUND_ZERO_ADDR2]], align 4
353 // CHECK2-NEXT:    store i32 0, i32* [[DOTBOUND_ZERO_ADDR]], align 4
354 // CHECK2-NEXT:    [[TMP0:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]])
355 // CHECK2-NEXT:    store i32 [[ARG]], i32* [[ARG_ADDR]], align 4
356 // CHECK2-NEXT:    call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB1]], i32 0, void (i32*, i32*, ...)* bitcast (void (i32*, i32*)* @.omp_outlined..5 to void (i32*, i32*, ...)*))
357 // CHECK2-NEXT:    call void @__kmpc_serialized_parallel(%struct.ident_t* @[[GLOB1]], i32 [[TMP0]])
358 // CHECK2-NEXT:    store i32 [[TMP0]], i32* [[DOTTHREADID_TEMP_]], align 4
359 // CHECK2-NEXT:    call void @.omp_outlined..6(i32* [[DOTTHREADID_TEMP_]], i32* [[DOTBOUND_ZERO_ADDR]]) #[[ATTR2]]
360 // CHECK2-NEXT:    call void @__kmpc_end_serialized_parallel(%struct.ident_t* @[[GLOB1]], i32 [[TMP0]])
361 // CHECK2-NEXT:    [[TMP1:%.*]] = load i32, i32* [[ARG_ADDR]], align 4
362 // CHECK2-NEXT:    [[TOBOOL:%.*]] = icmp ne i32 [[TMP1]], 0
363 // CHECK2-NEXT:    br i1 [[TOBOOL]], label [[OMP_IF_THEN:%.*]], label [[OMP_IF_ELSE:%.*]]
364 // CHECK2:       omp_if.then:
365 // CHECK2-NEXT:    call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB1]], i32 0, void (i32*, i32*, ...)* bitcast (void (i32*, i32*)* @.omp_outlined..7 to void (i32*, i32*, ...)*))
366 // CHECK2-NEXT:    br label [[OMP_IF_END:%.*]]
367 // CHECK2:       omp_if.else:
368 // CHECK2-NEXT:    call void @__kmpc_serialized_parallel(%struct.ident_t* @[[GLOB1]], i32 [[TMP0]])
369 // CHECK2-NEXT:    store i32 [[TMP0]], i32* [[DOTTHREADID_TEMP_1]], align 4
370 // CHECK2-NEXT:    call void @.omp_outlined..7(i32* [[DOTTHREADID_TEMP_1]], i32* [[DOTBOUND_ZERO_ADDR2]]) #[[ATTR2]]
371 // CHECK2-NEXT:    call void @__kmpc_end_serialized_parallel(%struct.ident_t* @[[GLOB1]], i32 [[TMP0]])
372 // CHECK2-NEXT:    br label [[OMP_IF_END]]
373 // CHECK2:       omp_if.end:
374 // CHECK2-NEXT:    ret i32 0
375 //
376 //
377 // CHECK2-LABEL: define {{[^@]+}}@.omp_outlined..5
378 // CHECK2-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]]) #[[ATTR1]] {
379 // CHECK2-NEXT:  entry:
380 // CHECK2-NEXT:    [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
381 // CHECK2-NEXT:    [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
382 // CHECK2-NEXT:    store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
383 // CHECK2-NEXT:    store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
384 // CHECK2-NEXT:    call void @_Z3fn1v()
385 // CHECK2-NEXT:    ret void
386 //
387 //
388 // CHECK2-LABEL: define {{[^@]+}}@.omp_outlined..6
389 // CHECK2-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]]) #[[ATTR1]] {
390 // CHECK2-NEXT:  entry:
391 // CHECK2-NEXT:    [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
392 // CHECK2-NEXT:    [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
393 // CHECK2-NEXT:    store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
394 // CHECK2-NEXT:    store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
395 // CHECK2-NEXT:    call void @_Z3fn2v()
396 // CHECK2-NEXT:    ret void
397 //
398 //
399 // CHECK2-LABEL: define {{[^@]+}}@.omp_outlined..7
400 // CHECK2-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]]) #[[ATTR1]] {
401 // CHECK2-NEXT:  entry:
402 // CHECK2-NEXT:    [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
403 // CHECK2-NEXT:    [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
404 // CHECK2-NEXT:    store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
405 // CHECK2-NEXT:    store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
406 // CHECK2-NEXT:    call void @_Z3fn3v()
407 // CHECK2-NEXT:    ret void
408 //
409 //
410 // CHECK3-LABEL: define {{[^@]+}}@_Z9gtid_testv
411 // CHECK3-SAME: () #[[ATTR0:[0-9]+]] {
412 // CHECK3-NEXT:  entry:
413 // CHECK3-NEXT:    call void @_Z9gtid_testv()
414 // CHECK3-NEXT:    ret void
415 //
416 //
417 // CHECK3-LABEL: define {{[^@]+}}@main
418 // CHECK3-SAME: () #[[ATTR1:[0-9]+]] {
419 // CHECK3-NEXT:  entry:
420 // CHECK3-NEXT:    [[RETVAL:%.*]] = alloca i32, align 4
421 // CHECK3-NEXT:    store i32 0, i32* [[RETVAL]], align 4
422 // CHECK3-NEXT:    call void @_Z3fn4v()
423 // CHECK3-NEXT:    call void @_Z3fn5v()
424 // CHECK3-NEXT:    call void @_Z3fn6v()
425 // CHECK3-NEXT:    [[TMP0:%.*]] = load i32, i32* @Arg, align 4
426 // CHECK3-NEXT:    [[CALL:%.*]] = call i32 @_Z5tmainIiEiT_(i32 [[TMP0]])
427 // CHECK3-NEXT:    ret i32 [[CALL]]
428 //
429 //
430 // CHECK3-LABEL: define {{[^@]+}}@_Z5tmainIiEiT_
431 // CHECK3-SAME: (i32 [[ARG:%.*]]) #[[ATTR0]] comdat {
432 // CHECK3-NEXT:  entry:
433 // CHECK3-NEXT:    [[ARG_ADDR:%.*]] = alloca i32, align 4
434 // CHECK3-NEXT:    store i32 [[ARG]], i32* [[ARG_ADDR]], align 4
435 // CHECK3-NEXT:    call void @_Z3fn1v()
436 // CHECK3-NEXT:    call void @_Z3fn2v()
437 // CHECK3-NEXT:    call void @_Z3fn3v()
438 // CHECK3-NEXT:    ret i32 0
439 //
440 //
441 // CHECK4-LABEL: define {{[^@]+}}@_Z9gtid_testv
442 // CHECK4-SAME: () #[[ATTR0:[0-9]+]] {
443 // CHECK4-NEXT:  entry:
444 // CHECK4-NEXT:    call void @_Z9gtid_testv()
445 // CHECK4-NEXT:    ret void
446 //
447 //
448 // CHECK4-LABEL: define {{[^@]+}}@main
449 // CHECK4-SAME: () #[[ATTR1:[0-9]+]] {
450 // CHECK4-NEXT:  entry:
451 // CHECK4-NEXT:    [[RETVAL:%.*]] = alloca i32, align 4
452 // CHECK4-NEXT:    store i32 0, i32* [[RETVAL]], align 4
453 // CHECK4-NEXT:    call void @_Z3fn4v()
454 // CHECK4-NEXT:    call void @_Z3fn5v()
455 // CHECK4-NEXT:    call void @_Z3fn6v()
456 // CHECK4-NEXT:    [[TMP0:%.*]] = load i32, i32* @Arg, align 4
457 // CHECK4-NEXT:    [[CALL:%.*]] = call i32 @_Z5tmainIiEiT_(i32 [[TMP0]])
458 // CHECK4-NEXT:    ret i32 [[CALL]]
459 //
460 //
461 // CHECK4-LABEL: define {{[^@]+}}@_Z5tmainIiEiT_
462 // CHECK4-SAME: (i32 [[ARG:%.*]]) #[[ATTR0]] comdat {
463 // CHECK4-NEXT:  entry:
464 // CHECK4-NEXT:    [[ARG_ADDR:%.*]] = alloca i32, align 4
465 // CHECK4-NEXT:    store i32 [[ARG]], i32* [[ARG_ADDR]], align 4
466 // CHECK4-NEXT:    call void @_Z3fn1v()
467 // CHECK4-NEXT:    call void @_Z3fn2v()
468 // CHECK4-NEXT:    call void @_Z3fn3v()
469 // CHECK4-NEXT:    ret i32 0
470 //
471 //
472 // CHECK5-LABEL: define {{[^@]+}}@_Z9gtid_testv
473 // CHECK5-SAME: () #[[ATTR0:[0-9]+]] {
474 // CHECK5-NEXT:  entry:
475 // CHECK5-NEXT:    call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB1:[0-9]+]], i32 0, void (i32*, i32*, ...)* bitcast (void (i32*, i32*)* @.omp_outlined. to void (i32*, i32*, ...)*))
476 // CHECK5-NEXT:    ret void
477 //
478 //
479 // CHECK5-LABEL: define {{[^@]+}}@.omp_outlined.
480 // CHECK5-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]]) #[[ATTR1:[0-9]+]] {
481 // CHECK5-NEXT:  entry:
482 // CHECK5-NEXT:    [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
483 // CHECK5-NEXT:    [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
484 // CHECK5-NEXT:    [[DOTBOUND_ZERO_ADDR:%.*]] = alloca i32, align 4
485 // CHECK5-NEXT:    store i32 0, i32* [[DOTBOUND_ZERO_ADDR]], align 4
486 // CHECK5-NEXT:    store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
487 // CHECK5-NEXT:    store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
488 // CHECK5-NEXT:    [[TMP0:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
489 // CHECK5-NEXT:    [[TMP1:%.*]] = load i32, i32* [[TMP0]], align 4
490 // CHECK5-NEXT:    call void @__kmpc_serialized_parallel(%struct.ident_t* @[[GLOB1]], i32 [[TMP1]])
491 // CHECK5-NEXT:    [[TMP2:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
492 // CHECK5-NEXT:    call void @.omp_outlined..1(i32* [[TMP2]], i32* [[DOTBOUND_ZERO_ADDR]]) #[[ATTR2:[0-9]+]]
493 // CHECK5-NEXT:    call void @__kmpc_end_serialized_parallel(%struct.ident_t* @[[GLOB1]], i32 [[TMP1]])
494 // CHECK5-NEXT:    ret void
495 //
496 //
497 // CHECK5-LABEL: define {{[^@]+}}@.omp_outlined..1
498 // CHECK5-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]]) #[[ATTR1]] {
499 // CHECK5-NEXT:  entry:
500 // CHECK5-NEXT:    [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
501 // CHECK5-NEXT:    [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
502 // CHECK5-NEXT:    store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
503 // CHECK5-NEXT:    store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
504 // CHECK5-NEXT:    call void @_Z9gtid_testv()
505 // CHECK5-NEXT:    ret void
506 //
507 //
508 // CHECK5-LABEL: define {{[^@]+}}@main
509 // CHECK5-SAME: () #[[ATTR3:[0-9]+]] {
510 // CHECK5-NEXT:  entry:
511 // CHECK5-NEXT:    [[RETVAL:%.*]] = alloca i32, align 4
512 // CHECK5-NEXT:    [[DOTTHREADID_TEMP_:%.*]] = alloca i32, align 4
513 // CHECK5-NEXT:    [[DOTBOUND_ZERO_ADDR:%.*]] = alloca i32, align 4
514 // CHECK5-NEXT:    [[DOTTHREADID_TEMP_1:%.*]] = alloca i32, align 4
515 // CHECK5-NEXT:    [[DOTBOUND_ZERO_ADDR2:%.*]] = alloca i32, align 4
516 // CHECK5-NEXT:    store i32 0, i32* [[DOTBOUND_ZERO_ADDR2]], align 4
517 // CHECK5-NEXT:    store i32 0, i32* [[DOTBOUND_ZERO_ADDR]], align 4
518 // CHECK5-NEXT:    [[TMP0:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]])
519 // CHECK5-NEXT:    store i32 0, i32* [[RETVAL]], align 4
520 // CHECK5-NEXT:    call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB1]], i32 0, void (i32*, i32*, ...)* bitcast (void (i32*, i32*)* @.omp_outlined..2 to void (i32*, i32*, ...)*))
521 // CHECK5-NEXT:    call void @__kmpc_serialized_parallel(%struct.ident_t* @[[GLOB1]], i32 [[TMP0]])
522 // CHECK5-NEXT:    store i32 [[TMP0]], i32* [[DOTTHREADID_TEMP_]], align 4
523 // CHECK5-NEXT:    call void @.omp_outlined..3(i32* [[DOTTHREADID_TEMP_]], i32* [[DOTBOUND_ZERO_ADDR]]) #[[ATTR2]]
524 // CHECK5-NEXT:    call void @__kmpc_end_serialized_parallel(%struct.ident_t* @[[GLOB1]], i32 [[TMP0]])
525 // CHECK5-NEXT:    [[TMP1:%.*]] = load i32, i32* @Arg, align 4
526 // CHECK5-NEXT:    [[TOBOOL:%.*]] = icmp ne i32 [[TMP1]], 0
527 // CHECK5-NEXT:    br i1 [[TOBOOL]], label [[OMP_IF_THEN:%.*]], label [[OMP_IF_ELSE:%.*]]
528 // CHECK5:       omp_if.then:
529 // CHECK5-NEXT:    call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB1]], i32 0, void (i32*, i32*, ...)* bitcast (void (i32*, i32*)* @.omp_outlined..4 to void (i32*, i32*, ...)*))
530 // CHECK5-NEXT:    br label [[OMP_IF_END:%.*]]
531 // CHECK5:       omp_if.else:
532 // CHECK5-NEXT:    call void @__kmpc_serialized_parallel(%struct.ident_t* @[[GLOB1]], i32 [[TMP0]])
533 // CHECK5-NEXT:    store i32 [[TMP0]], i32* [[DOTTHREADID_TEMP_1]], align 4
534 // CHECK5-NEXT:    call void @.omp_outlined..4(i32* [[DOTTHREADID_TEMP_1]], i32* [[DOTBOUND_ZERO_ADDR2]]) #[[ATTR2]]
535 // CHECK5-NEXT:    call void @__kmpc_end_serialized_parallel(%struct.ident_t* @[[GLOB1]], i32 [[TMP0]])
536 // CHECK5-NEXT:    br label [[OMP_IF_END]]
537 // CHECK5:       omp_if.end:
538 // CHECK5-NEXT:    [[TMP2:%.*]] = load i32, i32* @Arg, align 4
539 // CHECK5-NEXT:    [[CALL:%.*]] = call i32 @_Z5tmainIiEiT_(i32 [[TMP2]])
540 // CHECK5-NEXT:    ret i32 [[CALL]]
541 //
542 //
543 // CHECK5-LABEL: define {{[^@]+}}@.omp_outlined..2
544 // CHECK5-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]]) #[[ATTR1]] {
545 // CHECK5-NEXT:  entry:
546 // CHECK5-NEXT:    [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
547 // CHECK5-NEXT:    [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
548 // CHECK5-NEXT:    store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
549 // CHECK5-NEXT:    store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
550 // CHECK5-NEXT:    call void @_Z3fn4v()
551 // CHECK5-NEXT:    ret void
552 //
553 //
554 // CHECK5-LABEL: define {{[^@]+}}@.omp_outlined..3
555 // CHECK5-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]]) #[[ATTR1]] {
556 // CHECK5-NEXT:  entry:
557 // CHECK5-NEXT:    [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
558 // CHECK5-NEXT:    [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
559 // CHECK5-NEXT:    store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
560 // CHECK5-NEXT:    store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
561 // CHECK5-NEXT:    call void @_Z3fn5v()
562 // CHECK5-NEXT:    ret void
563 //
564 //
565 // CHECK5-LABEL: define {{[^@]+}}@.omp_outlined..4
566 // CHECK5-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]]) #[[ATTR1]] {
567 // CHECK5-NEXT:  entry:
568 // CHECK5-NEXT:    [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
569 // CHECK5-NEXT:    [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
570 // CHECK5-NEXT:    store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
571 // CHECK5-NEXT:    store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
572 // CHECK5-NEXT:    call void @_Z3fn6v()
573 // CHECK5-NEXT:    ret void
574 //
575 //
576 // CHECK5-LABEL: define {{[^@]+}}@_Z5tmainIiEiT_
577 // CHECK5-SAME: (i32 [[ARG:%.*]]) #[[ATTR0]] comdat {
578 // CHECK5-NEXT:  entry:
579 // CHECK5-NEXT:    [[ARG_ADDR:%.*]] = alloca i32, align 4
580 // CHECK5-NEXT:    [[DOTTHREADID_TEMP_:%.*]] = alloca i32, align 4
581 // CHECK5-NEXT:    [[DOTBOUND_ZERO_ADDR:%.*]] = alloca i32, align 4
582 // CHECK5-NEXT:    [[DOTTHREADID_TEMP_1:%.*]] = alloca i32, align 4
583 // CHECK5-NEXT:    [[DOTBOUND_ZERO_ADDR2:%.*]] = alloca i32, align 4
584 // CHECK5-NEXT:    store i32 0, i32* [[DOTBOUND_ZERO_ADDR2]], align 4
585 // CHECK5-NEXT:    store i32 0, i32* [[DOTBOUND_ZERO_ADDR]], align 4
586 // CHECK5-NEXT:    [[TMP0:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]])
587 // CHECK5-NEXT:    store i32 [[ARG]], i32* [[ARG_ADDR]], align 4
588 // CHECK5-NEXT:    call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB1]], i32 0, void (i32*, i32*, ...)* bitcast (void (i32*, i32*)* @.omp_outlined..5 to void (i32*, i32*, ...)*))
589 // CHECK5-NEXT:    call void @__kmpc_serialized_parallel(%struct.ident_t* @[[GLOB1]], i32 [[TMP0]])
590 // CHECK5-NEXT:    store i32 [[TMP0]], i32* [[DOTTHREADID_TEMP_]], align 4
591 // CHECK5-NEXT:    call void @.omp_outlined..6(i32* [[DOTTHREADID_TEMP_]], i32* [[DOTBOUND_ZERO_ADDR]]) #[[ATTR2]]
592 // CHECK5-NEXT:    call void @__kmpc_end_serialized_parallel(%struct.ident_t* @[[GLOB1]], i32 [[TMP0]])
593 // CHECK5-NEXT:    [[TMP1:%.*]] = load i32, i32* [[ARG_ADDR]], align 4
594 // CHECK5-NEXT:    [[TOBOOL:%.*]] = icmp ne i32 [[TMP1]], 0
595 // CHECK5-NEXT:    br i1 [[TOBOOL]], label [[OMP_IF_THEN:%.*]], label [[OMP_IF_ELSE:%.*]]
596 // CHECK5:       omp_if.then:
597 // CHECK5-NEXT:    call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB1]], i32 0, void (i32*, i32*, ...)* bitcast (void (i32*, i32*)* @.omp_outlined..7 to void (i32*, i32*, ...)*))
598 // CHECK5-NEXT:    br label [[OMP_IF_END:%.*]]
599 // CHECK5:       omp_if.else:
600 // CHECK5-NEXT:    call void @__kmpc_serialized_parallel(%struct.ident_t* @[[GLOB1]], i32 [[TMP0]])
601 // CHECK5-NEXT:    store i32 [[TMP0]], i32* [[DOTTHREADID_TEMP_1]], align 4
602 // CHECK5-NEXT:    call void @.omp_outlined..7(i32* [[DOTTHREADID_TEMP_1]], i32* [[DOTBOUND_ZERO_ADDR2]]) #[[ATTR2]]
603 // CHECK5-NEXT:    call void @__kmpc_end_serialized_parallel(%struct.ident_t* @[[GLOB1]], i32 [[TMP0]])
604 // CHECK5-NEXT:    br label [[OMP_IF_END]]
605 // CHECK5:       omp_if.end:
606 // CHECK5-NEXT:    ret i32 0
607 //
608 //
609 // CHECK5-LABEL: define {{[^@]+}}@.omp_outlined..5
610 // CHECK5-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]]) #[[ATTR1]] {
611 // CHECK5-NEXT:  entry:
612 // CHECK5-NEXT:    [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
613 // CHECK5-NEXT:    [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
614 // CHECK5-NEXT:    store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
615 // CHECK5-NEXT:    store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
616 // CHECK5-NEXT:    call void @_Z3fn1v()
617 // CHECK5-NEXT:    ret void
618 //
619 //
620 // CHECK5-LABEL: define {{[^@]+}}@.omp_outlined..6
621 // CHECK5-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]]) #[[ATTR1]] {
622 // CHECK5-NEXT:  entry:
623 // CHECK5-NEXT:    [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
624 // CHECK5-NEXT:    [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
625 // CHECK5-NEXT:    store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
626 // CHECK5-NEXT:    store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
627 // CHECK5-NEXT:    call void @_Z3fn2v()
628 // CHECK5-NEXT:    ret void
629 //
630 //
631 // CHECK5-LABEL: define {{[^@]+}}@.omp_outlined..7
632 // CHECK5-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]]) #[[ATTR1]] {
633 // CHECK5-NEXT:  entry:
634 // CHECK5-NEXT:    [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
635 // CHECK5-NEXT:    [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
636 // CHECK5-NEXT:    store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
637 // CHECK5-NEXT:    store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
638 // CHECK5-NEXT:    call void @_Z3fn3v()
639 // CHECK5-NEXT:    ret void
640 //
641 //
642 // CHECK6-LABEL: define {{[^@]+}}@_Z9gtid_testv
643 // CHECK6-SAME: () #[[ATTR0:[0-9]+]] {
644 // CHECK6-NEXT:  entry:
645 // CHECK6-NEXT:    call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB1:[0-9]+]], i32 0, void (i32*, i32*, ...)* bitcast (void (i32*, i32*)* @.omp_outlined. to void (i32*, i32*, ...)*))
646 // CHECK6-NEXT:    ret void
647 //
648 //
649 // CHECK6-LABEL: define {{[^@]+}}@.omp_outlined.
650 // CHECK6-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]]) #[[ATTR1:[0-9]+]] {
651 // CHECK6-NEXT:  entry:
652 // CHECK6-NEXT:    [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
653 // CHECK6-NEXT:    [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
654 // CHECK6-NEXT:    [[DOTBOUND_ZERO_ADDR:%.*]] = alloca i32, align 4
655 // CHECK6-NEXT:    store i32 0, i32* [[DOTBOUND_ZERO_ADDR]], align 4
656 // CHECK6-NEXT:    store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
657 // CHECK6-NEXT:    store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
658 // CHECK6-NEXT:    [[TMP0:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
659 // CHECK6-NEXT:    [[TMP1:%.*]] = load i32, i32* [[TMP0]], align 4
660 // CHECK6-NEXT:    call void @__kmpc_serialized_parallel(%struct.ident_t* @[[GLOB1]], i32 [[TMP1]])
661 // CHECK6-NEXT:    [[TMP2:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
662 // CHECK6-NEXT:    call void @.omp_outlined..1(i32* [[TMP2]], i32* [[DOTBOUND_ZERO_ADDR]]) #[[ATTR2:[0-9]+]]
663 // CHECK6-NEXT:    call void @__kmpc_end_serialized_parallel(%struct.ident_t* @[[GLOB1]], i32 [[TMP1]])
664 // CHECK6-NEXT:    ret void
665 //
666 //
667 // CHECK6-LABEL: define {{[^@]+}}@.omp_outlined..1
668 // CHECK6-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]]) #[[ATTR1]] {
669 // CHECK6-NEXT:  entry:
670 // CHECK6-NEXT:    [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
671 // CHECK6-NEXT:    [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
672 // CHECK6-NEXT:    store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
673 // CHECK6-NEXT:    store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
674 // CHECK6-NEXT:    call void @_Z9gtid_testv()
675 // CHECK6-NEXT:    ret void
676 //
677 //
678 // CHECK6-LABEL: define {{[^@]+}}@main
679 // CHECK6-SAME: () #[[ATTR3:[0-9]+]] {
680 // CHECK6-NEXT:  entry:
681 // CHECK6-NEXT:    [[RETVAL:%.*]] = alloca i32, align 4
682 // CHECK6-NEXT:    [[DOTTHREADID_TEMP_:%.*]] = alloca i32, align 4
683 // CHECK6-NEXT:    [[DOTBOUND_ZERO_ADDR:%.*]] = alloca i32, align 4
684 // CHECK6-NEXT:    [[DOTTHREADID_TEMP_1:%.*]] = alloca i32, align 4
685 // CHECK6-NEXT:    [[DOTBOUND_ZERO_ADDR2:%.*]] = alloca i32, align 4
686 // CHECK6-NEXT:    store i32 0, i32* [[DOTBOUND_ZERO_ADDR2]], align 4
687 // CHECK6-NEXT:    store i32 0, i32* [[DOTBOUND_ZERO_ADDR]], align 4
688 // CHECK6-NEXT:    [[TMP0:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]])
689 // CHECK6-NEXT:    store i32 0, i32* [[RETVAL]], align 4
690 // CHECK6-NEXT:    call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB1]], i32 0, void (i32*, i32*, ...)* bitcast (void (i32*, i32*)* @.omp_outlined..2 to void (i32*, i32*, ...)*))
691 // CHECK6-NEXT:    call void @__kmpc_serialized_parallel(%struct.ident_t* @[[GLOB1]], i32 [[TMP0]])
692 // CHECK6-NEXT:    store i32 [[TMP0]], i32* [[DOTTHREADID_TEMP_]], align 4
693 // CHECK6-NEXT:    call void @.omp_outlined..3(i32* [[DOTTHREADID_TEMP_]], i32* [[DOTBOUND_ZERO_ADDR]]) #[[ATTR2]]
694 // CHECK6-NEXT:    call void @__kmpc_end_serialized_parallel(%struct.ident_t* @[[GLOB1]], i32 [[TMP0]])
695 // CHECK6-NEXT:    [[TMP1:%.*]] = load i32, i32* @Arg, align 4
696 // CHECK6-NEXT:    [[TOBOOL:%.*]] = icmp ne i32 [[TMP1]], 0
697 // CHECK6-NEXT:    br i1 [[TOBOOL]], label [[OMP_IF_THEN:%.*]], label [[OMP_IF_ELSE:%.*]]
698 // CHECK6:       omp_if.then:
699 // CHECK6-NEXT:    call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB1]], i32 0, void (i32*, i32*, ...)* bitcast (void (i32*, i32*)* @.omp_outlined..4 to void (i32*, i32*, ...)*))
700 // CHECK6-NEXT:    br label [[OMP_IF_END:%.*]]
701 // CHECK6:       omp_if.else:
702 // CHECK6-NEXT:    call void @__kmpc_serialized_parallel(%struct.ident_t* @[[GLOB1]], i32 [[TMP0]])
703 // CHECK6-NEXT:    store i32 [[TMP0]], i32* [[DOTTHREADID_TEMP_1]], align 4
704 // CHECK6-NEXT:    call void @.omp_outlined..4(i32* [[DOTTHREADID_TEMP_1]], i32* [[DOTBOUND_ZERO_ADDR2]]) #[[ATTR2]]
705 // CHECK6-NEXT:    call void @__kmpc_end_serialized_parallel(%struct.ident_t* @[[GLOB1]], i32 [[TMP0]])
706 // CHECK6-NEXT:    br label [[OMP_IF_END]]
707 // CHECK6:       omp_if.end:
708 // CHECK6-NEXT:    [[TMP2:%.*]] = load i32, i32* @Arg, align 4
709 // CHECK6-NEXT:    [[CALL:%.*]] = call i32 @_Z5tmainIiEiT_(i32 [[TMP2]])
710 // CHECK6-NEXT:    ret i32 [[CALL]]
711 //
712 //
713 // CHECK6-LABEL: define {{[^@]+}}@.omp_outlined..2
714 // CHECK6-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]]) #[[ATTR1]] {
715 // CHECK6-NEXT:  entry:
716 // CHECK6-NEXT:    [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
717 // CHECK6-NEXT:    [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
718 // CHECK6-NEXT:    store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
719 // CHECK6-NEXT:    store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
720 // CHECK6-NEXT:    call void @_Z3fn4v()
721 // CHECK6-NEXT:    ret void
722 //
723 //
724 // CHECK6-LABEL: define {{[^@]+}}@.omp_outlined..3
725 // CHECK6-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]]) #[[ATTR1]] {
726 // CHECK6-NEXT:  entry:
727 // CHECK6-NEXT:    [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
728 // CHECK6-NEXT:    [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
729 // CHECK6-NEXT:    store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
730 // CHECK6-NEXT:    store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
731 // CHECK6-NEXT:    call void @_Z3fn5v()
732 // CHECK6-NEXT:    ret void
733 //
734 //
735 // CHECK6-LABEL: define {{[^@]+}}@.omp_outlined..4
736 // CHECK6-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]]) #[[ATTR1]] {
737 // CHECK6-NEXT:  entry:
738 // CHECK6-NEXT:    [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
739 // CHECK6-NEXT:    [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
740 // CHECK6-NEXT:    store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
741 // CHECK6-NEXT:    store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
742 // CHECK6-NEXT:    call void @_Z3fn6v()
743 // CHECK6-NEXT:    ret void
744 //
745 //
746 // CHECK6-LABEL: define {{[^@]+}}@_Z5tmainIiEiT_
747 // CHECK6-SAME: (i32 [[ARG:%.*]]) #[[ATTR0]] comdat {
748 // CHECK6-NEXT:  entry:
749 // CHECK6-NEXT:    [[ARG_ADDR:%.*]] = alloca i32, align 4
750 // CHECK6-NEXT:    [[DOTTHREADID_TEMP_:%.*]] = alloca i32, align 4
751 // CHECK6-NEXT:    [[DOTBOUND_ZERO_ADDR:%.*]] = alloca i32, align 4
752 // CHECK6-NEXT:    [[DOTTHREADID_TEMP_1:%.*]] = alloca i32, align 4
753 // CHECK6-NEXT:    [[DOTBOUND_ZERO_ADDR2:%.*]] = alloca i32, align 4
754 // CHECK6-NEXT:    store i32 0, i32* [[DOTBOUND_ZERO_ADDR2]], align 4
755 // CHECK6-NEXT:    store i32 0, i32* [[DOTBOUND_ZERO_ADDR]], align 4
756 // CHECK6-NEXT:    [[TMP0:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]])
757 // CHECK6-NEXT:    store i32 [[ARG]], i32* [[ARG_ADDR]], align 4
758 // CHECK6-NEXT:    call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB1]], i32 0, void (i32*, i32*, ...)* bitcast (void (i32*, i32*)* @.omp_outlined..5 to void (i32*, i32*, ...)*))
759 // CHECK6-NEXT:    call void @__kmpc_serialized_parallel(%struct.ident_t* @[[GLOB1]], i32 [[TMP0]])
760 // CHECK6-NEXT:    store i32 [[TMP0]], i32* [[DOTTHREADID_TEMP_]], align 4
761 // CHECK6-NEXT:    call void @.omp_outlined..6(i32* [[DOTTHREADID_TEMP_]], i32* [[DOTBOUND_ZERO_ADDR]]) #[[ATTR2]]
762 // CHECK6-NEXT:    call void @__kmpc_end_serialized_parallel(%struct.ident_t* @[[GLOB1]], i32 [[TMP0]])
763 // CHECK6-NEXT:    [[TMP1:%.*]] = load i32, i32* [[ARG_ADDR]], align 4
764 // CHECK6-NEXT:    [[TOBOOL:%.*]] = icmp ne i32 [[TMP1]], 0
765 // CHECK6-NEXT:    br i1 [[TOBOOL]], label [[OMP_IF_THEN:%.*]], label [[OMP_IF_ELSE:%.*]]
766 // CHECK6:       omp_if.then:
767 // CHECK6-NEXT:    call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB1]], i32 0, void (i32*, i32*, ...)* bitcast (void (i32*, i32*)* @.omp_outlined..7 to void (i32*, i32*, ...)*))
768 // CHECK6-NEXT:    br label [[OMP_IF_END:%.*]]
769 // CHECK6:       omp_if.else:
770 // CHECK6-NEXT:    call void @__kmpc_serialized_parallel(%struct.ident_t* @[[GLOB1]], i32 [[TMP0]])
771 // CHECK6-NEXT:    store i32 [[TMP0]], i32* [[DOTTHREADID_TEMP_1]], align 4
772 // CHECK6-NEXT:    call void @.omp_outlined..7(i32* [[DOTTHREADID_TEMP_1]], i32* [[DOTBOUND_ZERO_ADDR2]]) #[[ATTR2]]
773 // CHECK6-NEXT:    call void @__kmpc_end_serialized_parallel(%struct.ident_t* @[[GLOB1]], i32 [[TMP0]])
774 // CHECK6-NEXT:    br label [[OMP_IF_END]]
775 // CHECK6:       omp_if.end:
776 // CHECK6-NEXT:    ret i32 0
777 //
778 //
779 // CHECK6-LABEL: define {{[^@]+}}@.omp_outlined..5
780 // CHECK6-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]]) #[[ATTR1]] {
781 // CHECK6-NEXT:  entry:
782 // CHECK6-NEXT:    [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
783 // CHECK6-NEXT:    [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
784 // CHECK6-NEXT:    store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
785 // CHECK6-NEXT:    store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
786 // CHECK6-NEXT:    call void @_Z3fn1v()
787 // CHECK6-NEXT:    ret void
788 //
789 //
790 // CHECK6-LABEL: define {{[^@]+}}@.omp_outlined..6
791 // CHECK6-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]]) #[[ATTR1]] {
792 // CHECK6-NEXT:  entry:
793 // CHECK6-NEXT:    [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
794 // CHECK6-NEXT:    [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
795 // CHECK6-NEXT:    store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
796 // CHECK6-NEXT:    store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
797 // CHECK6-NEXT:    call void @_Z3fn2v()
798 // CHECK6-NEXT:    ret void
799 //
800 //
801 // CHECK6-LABEL: define {{[^@]+}}@.omp_outlined..7
802 // CHECK6-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]]) #[[ATTR1]] {
803 // CHECK6-NEXT:  entry:
804 // CHECK6-NEXT:    [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
805 // CHECK6-NEXT:    [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
806 // CHECK6-NEXT:    store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
807 // CHECK6-NEXT:    store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
808 // CHECK6-NEXT:    call void @_Z3fn3v()
809 // CHECK6-NEXT:    ret void
810 //
811 //
812 // CHECK7-LABEL: define {{[^@]+}}@_Z9gtid_testv
813 // CHECK7-SAME: () #[[ATTR0:[0-9]+]] {
814 // CHECK7-NEXT:  entry:
815 // CHECK7-NEXT:    call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB1:[0-9]+]], i32 0, void (i32*, i32*, ...)* bitcast (void (i32*, i32*)* @.omp_outlined. to void (i32*, i32*, ...)*))
816 // CHECK7-NEXT:    ret void
817 //
818 //
819 // CHECK7-LABEL: define {{[^@]+}}@.omp_outlined.
820 // CHECK7-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]]) #[[ATTR1:[0-9]+]] {
821 // CHECK7-NEXT:  entry:
822 // CHECK7-NEXT:    [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
823 // CHECK7-NEXT:    [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
824 // CHECK7-NEXT:    [[DOTBOUND_ZERO_ADDR:%.*]] = alloca i32, align 4
825 // CHECK7-NEXT:    store i32 0, i32* [[DOTBOUND_ZERO_ADDR]], align 4
826 // CHECK7-NEXT:    store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
827 // CHECK7-NEXT:    store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
828 // CHECK7-NEXT:    [[TMP0:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
829 // CHECK7-NEXT:    [[TMP1:%.*]] = load i32, i32* [[TMP0]], align 4
830 // CHECK7-NEXT:    call void @__kmpc_serialized_parallel(%struct.ident_t* @[[GLOB1]], i32 [[TMP1]])
831 // CHECK7-NEXT:    [[TMP2:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
832 // CHECK7-NEXT:    call void @.omp_outlined..1(i32* [[TMP2]], i32* [[DOTBOUND_ZERO_ADDR]]) #[[ATTR2:[0-9]+]]
833 // CHECK7-NEXT:    call void @__kmpc_end_serialized_parallel(%struct.ident_t* @[[GLOB1]], i32 [[TMP1]])
834 // CHECK7-NEXT:    ret void
835 //
836 //
837 // CHECK7-LABEL: define {{[^@]+}}@.omp_outlined..1
838 // CHECK7-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]]) #[[ATTR1]] {
839 // CHECK7-NEXT:  entry:
840 // CHECK7-NEXT:    [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
841 // CHECK7-NEXT:    [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
842 // CHECK7-NEXT:    store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
843 // CHECK7-NEXT:    store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
844 // CHECK7-NEXT:    call void @_Z9gtid_testv()
845 // CHECK7-NEXT:    ret void
846 //
847 //
848 // CHECK7-LABEL: define {{[^@]+}}@main
849 // CHECK7-SAME: () #[[ATTR3:[0-9]+]] {
850 // CHECK7-NEXT:  entry:
851 // CHECK7-NEXT:    [[RETVAL:%.*]] = alloca i32, align 4
852 // CHECK7-NEXT:    [[DOTTHREADID_TEMP_:%.*]] = alloca i32, align 4
853 // CHECK7-NEXT:    [[DOTBOUND_ZERO_ADDR:%.*]] = alloca i32, align 4
854 // CHECK7-NEXT:    [[DOTTHREADID_TEMP_1:%.*]] = alloca i32, align 4
855 // CHECK7-NEXT:    [[DOTBOUND_ZERO_ADDR2:%.*]] = alloca i32, align 4
856 // CHECK7-NEXT:    store i32 0, i32* [[DOTBOUND_ZERO_ADDR2]], align 4
857 // CHECK7-NEXT:    store i32 0, i32* [[DOTBOUND_ZERO_ADDR]], align 4
858 // CHECK7-NEXT:    [[TMP0:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]])
859 // CHECK7-NEXT:    store i32 0, i32* [[RETVAL]], align 4
860 // CHECK7-NEXT:    call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB1]], i32 0, void (i32*, i32*, ...)* bitcast (void (i32*, i32*)* @.omp_outlined..2 to void (i32*, i32*, ...)*))
861 // CHECK7-NEXT:    call void @__kmpc_serialized_parallel(%struct.ident_t* @[[GLOB1]], i32 [[TMP0]])
862 // CHECK7-NEXT:    store i32 [[TMP0]], i32* [[DOTTHREADID_TEMP_]], align 4
863 // CHECK7-NEXT:    call void @.omp_outlined..3(i32* [[DOTTHREADID_TEMP_]], i32* [[DOTBOUND_ZERO_ADDR]]) #[[ATTR2]]
864 // CHECK7-NEXT:    call void @__kmpc_end_serialized_parallel(%struct.ident_t* @[[GLOB1]], i32 [[TMP0]])
865 // CHECK7-NEXT:    [[TMP1:%.*]] = load i32, i32* @Arg, align 4
866 // CHECK7-NEXT:    [[TOBOOL:%.*]] = icmp ne i32 [[TMP1]], 0
867 // CHECK7-NEXT:    br i1 [[TOBOOL]], label [[OMP_IF_THEN:%.*]], label [[OMP_IF_ELSE:%.*]]
868 // CHECK7:       omp_if.then:
869 // CHECK7-NEXT:    call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB1]], i32 0, void (i32*, i32*, ...)* bitcast (void (i32*, i32*)* @.omp_outlined..4 to void (i32*, i32*, ...)*))
870 // CHECK7-NEXT:    br label [[OMP_IF_END:%.*]]
871 // CHECK7:       omp_if.else:
872 // CHECK7-NEXT:    call void @__kmpc_serialized_parallel(%struct.ident_t* @[[GLOB1]], i32 [[TMP0]])
873 // CHECK7-NEXT:    store i32 [[TMP0]], i32* [[DOTTHREADID_TEMP_1]], align 4
874 // CHECK7-NEXT:    call void @.omp_outlined..4(i32* [[DOTTHREADID_TEMP_1]], i32* [[DOTBOUND_ZERO_ADDR2]]) #[[ATTR2]]
875 // CHECK7-NEXT:    call void @__kmpc_end_serialized_parallel(%struct.ident_t* @[[GLOB1]], i32 [[TMP0]])
876 // CHECK7-NEXT:    br label [[OMP_IF_END]]
877 // CHECK7:       omp_if.end:
878 // CHECK7-NEXT:    [[TMP2:%.*]] = load i32, i32* @Arg, align 4
879 // CHECK7-NEXT:    [[CALL:%.*]] = call i32 @_Z5tmainIiEiT_(i32 [[TMP2]])
880 // CHECK7-NEXT:    ret i32 [[CALL]]
881 //
882 //
883 // CHECK7-LABEL: define {{[^@]+}}@.omp_outlined..2
884 // CHECK7-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]]) #[[ATTR1]] {
885 // CHECK7-NEXT:  entry:
886 // CHECK7-NEXT:    [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
887 // CHECK7-NEXT:    [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
888 // CHECK7-NEXT:    store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
889 // CHECK7-NEXT:    store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
890 // CHECK7-NEXT:    call void @_Z3fn4v()
891 // CHECK7-NEXT:    ret void
892 //
893 //
894 // CHECK7-LABEL: define {{[^@]+}}@.omp_outlined..3
895 // CHECK7-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]]) #[[ATTR1]] {
896 // CHECK7-NEXT:  entry:
897 // CHECK7-NEXT:    [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
898 // CHECK7-NEXT:    [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
899 // CHECK7-NEXT:    store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
900 // CHECK7-NEXT:    store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
901 // CHECK7-NEXT:    call void @_Z3fn5v()
902 // CHECK7-NEXT:    ret void
903 //
904 //
905 // CHECK7-LABEL: define {{[^@]+}}@.omp_outlined..4
906 // CHECK7-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]]) #[[ATTR1]] {
907 // CHECK7-NEXT:  entry:
908 // CHECK7-NEXT:    [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
909 // CHECK7-NEXT:    [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
910 // CHECK7-NEXT:    store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
911 // CHECK7-NEXT:    store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
912 // CHECK7-NEXT:    call void @_Z3fn6v()
913 // CHECK7-NEXT:    ret void
914 //
915 //
916 // CHECK7-LABEL: define {{[^@]+}}@_Z5tmainIiEiT_
917 // CHECK7-SAME: (i32 [[ARG:%.*]]) #[[ATTR0]] comdat {
918 // CHECK7-NEXT:  entry:
919 // CHECK7-NEXT:    [[ARG_ADDR:%.*]] = alloca i32, align 4
920 // CHECK7-NEXT:    [[DOTTHREADID_TEMP_:%.*]] = alloca i32, align 4
921 // CHECK7-NEXT:    [[DOTBOUND_ZERO_ADDR:%.*]] = alloca i32, align 4
922 // CHECK7-NEXT:    [[DOTTHREADID_TEMP_1:%.*]] = alloca i32, align 4
923 // CHECK7-NEXT:    [[DOTBOUND_ZERO_ADDR2:%.*]] = alloca i32, align 4
924 // CHECK7-NEXT:    store i32 0, i32* [[DOTBOUND_ZERO_ADDR2]], align 4
925 // CHECK7-NEXT:    store i32 0, i32* [[DOTBOUND_ZERO_ADDR]], align 4
926 // CHECK7-NEXT:    [[TMP0:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]])
927 // CHECK7-NEXT:    store i32 [[ARG]], i32* [[ARG_ADDR]], align 4
928 // CHECK7-NEXT:    call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB1]], i32 0, void (i32*, i32*, ...)* bitcast (void (i32*, i32*)* @.omp_outlined..5 to void (i32*, i32*, ...)*))
929 // CHECK7-NEXT:    call void @__kmpc_serialized_parallel(%struct.ident_t* @[[GLOB1]], i32 [[TMP0]])
930 // CHECK7-NEXT:    store i32 [[TMP0]], i32* [[DOTTHREADID_TEMP_]], align 4
931 // CHECK7-NEXT:    call void @.omp_outlined..6(i32* [[DOTTHREADID_TEMP_]], i32* [[DOTBOUND_ZERO_ADDR]]) #[[ATTR2]]
932 // CHECK7-NEXT:    call void @__kmpc_end_serialized_parallel(%struct.ident_t* @[[GLOB1]], i32 [[TMP0]])
933 // CHECK7-NEXT:    [[TMP1:%.*]] = load i32, i32* [[ARG_ADDR]], align 4
934 // CHECK7-NEXT:    [[TOBOOL:%.*]] = icmp ne i32 [[TMP1]], 0
935 // CHECK7-NEXT:    br i1 [[TOBOOL]], label [[OMP_IF_THEN:%.*]], label [[OMP_IF_ELSE:%.*]]
936 // CHECK7:       omp_if.then:
937 // CHECK7-NEXT:    call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB1]], i32 0, void (i32*, i32*, ...)* bitcast (void (i32*, i32*)* @.omp_outlined..7 to void (i32*, i32*, ...)*))
938 // CHECK7-NEXT:    br label [[OMP_IF_END:%.*]]
939 // CHECK7:       omp_if.else:
940 // CHECK7-NEXT:    call void @__kmpc_serialized_parallel(%struct.ident_t* @[[GLOB1]], i32 [[TMP0]])
941 // CHECK7-NEXT:    store i32 [[TMP0]], i32* [[DOTTHREADID_TEMP_1]], align 4
942 // CHECK7-NEXT:    call void @.omp_outlined..7(i32* [[DOTTHREADID_TEMP_1]], i32* [[DOTBOUND_ZERO_ADDR2]]) #[[ATTR2]]
943 // CHECK7-NEXT:    call void @__kmpc_end_serialized_parallel(%struct.ident_t* @[[GLOB1]], i32 [[TMP0]])
944 // CHECK7-NEXT:    br label [[OMP_IF_END]]
945 // CHECK7:       omp_if.end:
946 // CHECK7-NEXT:    ret i32 0
947 //
948 //
949 // CHECK7-LABEL: define {{[^@]+}}@.omp_outlined..5
950 // CHECK7-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]]) #[[ATTR1]] {
951 // CHECK7-NEXT:  entry:
952 // CHECK7-NEXT:    [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
953 // CHECK7-NEXT:    [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
954 // CHECK7-NEXT:    store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
955 // CHECK7-NEXT:    store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
956 // CHECK7-NEXT:    call void @_Z3fn1v()
957 // CHECK7-NEXT:    ret void
958 //
959 //
960 // CHECK7-LABEL: define {{[^@]+}}@.omp_outlined..6
961 // CHECK7-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]]) #[[ATTR1]] {
962 // CHECK7-NEXT:  entry:
963 // CHECK7-NEXT:    [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
964 // CHECK7-NEXT:    [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
965 // CHECK7-NEXT:    store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
966 // CHECK7-NEXT:    store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
967 // CHECK7-NEXT:    call void @_Z3fn2v()
968 // CHECK7-NEXT:    ret void
969 //
970 //
971 // CHECK7-LABEL: define {{[^@]+}}@.omp_outlined..7
972 // CHECK7-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]]) #[[ATTR1]] {
973 // CHECK7-NEXT:  entry:
974 // CHECK7-NEXT:    [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
975 // CHECK7-NEXT:    [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
976 // CHECK7-NEXT:    store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
977 // CHECK7-NEXT:    store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
978 // CHECK7-NEXT:    call void @_Z3fn3v()
979 // CHECK7-NEXT:    ret void
980 //
981 //
982 // CHECK8-LABEL: define {{[^@]+}}@_Z9gtid_testv
983 // CHECK8-SAME: () #[[ATTR0:[0-9]+]] {
984 // CHECK8-NEXT:  entry:
985 // CHECK8-NEXT:    call void @_Z9gtid_testv()
986 // CHECK8-NEXT:    ret void
987 //
988 //
989 // CHECK8-LABEL: define {{[^@]+}}@main
990 // CHECK8-SAME: () #[[ATTR1:[0-9]+]] {
991 // CHECK8-NEXT:  entry:
992 // CHECK8-NEXT:    [[RETVAL:%.*]] = alloca i32, align 4
993 // CHECK8-NEXT:    store i32 0, i32* [[RETVAL]], align 4
994 // CHECK8-NEXT:    call void @_Z3fn4v()
995 // CHECK8-NEXT:    call void @_Z3fn5v()
996 // CHECK8-NEXT:    call void @_Z3fn6v()
997 // CHECK8-NEXT:    [[TMP0:%.*]] = load i32, i32* @Arg, align 4
998 // CHECK8-NEXT:    [[CALL:%.*]] = call i32 @_Z5tmainIiEiT_(i32 [[TMP0]])
999 // CHECK8-NEXT:    ret i32 [[CALL]]
1000 //
1001 //
1002 // CHECK8-LABEL: define {{[^@]+}}@_Z5tmainIiEiT_
1003 // CHECK8-SAME: (i32 [[ARG:%.*]]) #[[ATTR0]] comdat {
1004 // CHECK8-NEXT:  entry:
1005 // CHECK8-NEXT:    [[ARG_ADDR:%.*]] = alloca i32, align 4
1006 // CHECK8-NEXT:    store i32 [[ARG]], i32* [[ARG_ADDR]], align 4
1007 // CHECK8-NEXT:    call void @_Z3fn1v()
1008 // CHECK8-NEXT:    call void @_Z3fn2v()
1009 // CHECK8-NEXT:    call void @_Z3fn3v()
1010 // CHECK8-NEXT:    ret i32 0
1011 //
1012 //
1013 // CHECK9-LABEL: define {{[^@]+}}@_Z9gtid_testv
1014 // CHECK9-SAME: () #[[ATTR0:[0-9]+]] {
1015 // CHECK9-NEXT:  entry:
1016 // CHECK9-NEXT:    call void @_Z9gtid_testv()
1017 // CHECK9-NEXT:    ret void
1018 //
1019 //
1020 // CHECK9-LABEL: define {{[^@]+}}@main
1021 // CHECK9-SAME: () #[[ATTR1:[0-9]+]] {
1022 // CHECK9-NEXT:  entry:
1023 // CHECK9-NEXT:    [[RETVAL:%.*]] = alloca i32, align 4
1024 // CHECK9-NEXT:    store i32 0, i32* [[RETVAL]], align 4
1025 // CHECK9-NEXT:    call void @_Z3fn4v()
1026 // CHECK9-NEXT:    call void @_Z3fn5v()
1027 // CHECK9-NEXT:    call void @_Z3fn6v()
1028 // CHECK9-NEXT:    [[TMP0:%.*]] = load i32, i32* @Arg, align 4
1029 // CHECK9-NEXT:    [[CALL:%.*]] = call i32 @_Z5tmainIiEiT_(i32 [[TMP0]])
1030 // CHECK9-NEXT:    ret i32 [[CALL]]
1031 //
1032 //
1033 // CHECK9-LABEL: define {{[^@]+}}@_Z5tmainIiEiT_
1034 // CHECK9-SAME: (i32 [[ARG:%.*]]) #[[ATTR0]] comdat {
1035 // CHECK9-NEXT:  entry:
1036 // CHECK9-NEXT:    [[ARG_ADDR:%.*]] = alloca i32, align 4
1037 // CHECK9-NEXT:    store i32 [[ARG]], i32* [[ARG_ADDR]], align 4
1038 // CHECK9-NEXT:    call void @_Z3fn1v()
1039 // CHECK9-NEXT:    call void @_Z3fn2v()
1040 // CHECK9-NEXT:    call void @_Z3fn3v()
1041 // CHECK9-NEXT:    ret i32 0
1042 //
1043