1 // NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --function-signature --include-generated-funcs --replace-value-regex "__omp_offloading_[0-9a-z]+_[0-9a-z]+" "reduction_size[.].+[.]" "pl_cond[.].+[.|,]" --prefix-filecheck-ir-name _
2 // RUN: %clang_cc1 -verify -fopenmp -fopenmp-version=45 -fopenmp-targets=powerpc64le-ibm-linux-gnu -x c++ -triple x86_64-unknown-unknown -emit-llvm %s -o - | FileCheck %s --check-prefix=CHECK1
3 // RUN: %clang_cc1 -fopenmp -fopenmp-version=45 -fopenmp-targets=powerpc64le-ibm-linux-gnu -x c++ -std=c++11 -triple x86_64-unknown-unknown -emit-pch -o %t %s
4 // RUN: %clang_cc1 -fopenmp -fopenmp-version=45 -fopenmp-targets=powerpc64le-ibm-linux-gnu -x c++ -triple x86_64-unknown-unknown -std=c++11 -include-pch %t -verify %s -emit-llvm -o - | FileCheck %s --check-prefix=CHECK2
5 // RUN: %clang_cc1 -verify -fopenmp -fopenmp-version=50 -DOMP5 -fopenmp-targets=powerpc64le-ibm-linux-gnu -x c++ -triple x86_64-unknown-unknown -emit-llvm %s -o - | FileCheck %s --check-prefix=CHECK3
6 // RUN: %clang_cc1 -fopenmp -fopenmp-version=50 -DOMP5 -fopenmp-targets=powerpc64le-ibm-linux-gnu -x c++ -std=c++11 -triple x86_64-unknown-unknown -emit-pch -o %t %s
7 // RUN: %clang_cc1 -fopenmp -fopenmp-version=50 -DOMP5 -fopenmp-targets=powerpc64le-ibm-linux-gnu -x c++ -triple x86_64-unknown-unknown -std=c++11 -include-pch %t -verify %s -emit-llvm -o - | FileCheck %s --check-prefix=CHECK4
8
9 // RUN: %clang_cc1 -verify -fopenmp-simd -fopenmp-version=45 -fopenmp-targets=powerpc64le-ibm-linux-gnu -x c++ -triple x86_64-unknown-unknown -emit-llvm %s -o - | FileCheck %s --check-prefix=CHECK5
10 // RUN: %clang_cc1 -fopenmp-simd -fopenmp-version=45 -fopenmp-targets=powerpc64le-ibm-linux-gnu -x c++ -std=c++11 -triple x86_64-unknown-unknown -emit-pch -o %t %s
11 // RUN: %clang_cc1 -fopenmp-simd -fopenmp-version=45 -fopenmp-targets=powerpc64le-ibm-linux-gnu -x c++ -triple x86_64-unknown-unknown -std=c++11 -include-pch %t -verify %s -emit-llvm -o - | FileCheck %s --check-prefix=CHECK6
12 // RUN: %clang_cc1 -verify -fopenmp-simd -fopenmp-version=50 -DOMP5 -fopenmp-targets=powerpc64le-ibm-linux-gnu -x c++ -triple x86_64-unknown-unknown -emit-llvm %s -o - | FileCheck %s --check-prefix=CHECK7
13 // RUN: %clang_cc1 -fopenmp-simd -fopenmp-version=50 -DOMP5 -fopenmp-targets=powerpc64le-ibm-linux-gnu -x c++ -std=c++11 -triple x86_64-unknown-unknown -emit-pch -o %t %s
14 // RUN: %clang_cc1 -fopenmp-simd -fopenmp-version=50 -DOMP5 -fopenmp-targets=powerpc64le-ibm-linux-gnu -x c++ -triple x86_64-unknown-unknown -std=c++11 -include-pch %t -verify %s -emit-llvm -o - | FileCheck %s --check-prefix=CHECK8
15
16 // RUN: %clang_cc1 -verify -fopenmp -fopenmp-version=45 -fopenmp-targets=powerpc64le-ibm-linux-gnu -x c++ -triple aarch64-unknown-unknown -emit-llvm %s -o - | FileCheck %s --check-prefix=CHECK9
17 // RUN: %clang_cc1 -fopenmp -fopenmp-version=45 -fopenmp-targets=powerpc64le-ibm-linux-gnu -x c++ -std=c++11 -triple aarch64-unknown-unknown -emit-pch -o %t %s
18 // RUN: %clang_cc1 -fopenmp -fopenmp-version=45 -fopenmp-targets=powerpc64le-ibm-linux-gnu -x c++ -triple aarch64-unknown-unknown -std=c++11 -include-pch %t -verify %s -emit-llvm -o - | FileCheck %s --check-prefix=CHECK10
19 // RUN: %clang_cc1 -verify -fopenmp -fopenmp-version=50 -DOMP5 -fopenmp-targets=powerpc64le-ibm-linux-gnu -x c++ -triple aarch64-unknown-unknown -emit-llvm %s -o - | FileCheck %s --check-prefix=CHECK11
20 // RUN: %clang_cc1 -fopenmp -fopenmp-version=50 -DOMP5 -fopenmp-targets=powerpc64le-ibm-linux-gnu -x c++ -std=c++11 -triple aarch64-unknown-unknown -emit-pch -o %t %s
21 // RUN: %clang_cc1 -fopenmp -fopenmp-version=50 -DOMP5 -fopenmp-targets=powerpc64le-ibm-linux-gnu -x c++ -triple aarch64-unknown-unknown -std=c++11 -include-pch %t -verify %s -emit-llvm -o - | FileCheck %s --check-prefix=CHECK12
22
23 // RUN: %clang_cc1 -verify -fopenmp-simd -fopenmp-version=45 -fopenmp-targets=powerpc64le-ibm-linux-gnu -x c++ -triple aarch64-unknown-unknown -emit-llvm %s -o - | FileCheck %s --check-prefix=CHECK13
24 // RUN: %clang_cc1 -fopenmp-simd -fopenmp-version=45 -fopenmp-targets=powerpc64le-ibm-linux-gnu -x c++ -std=c++11 -triple aarch64-unknown-unknown -emit-pch -o %t %s
25 // RUN: %clang_cc1 -fopenmp-simd -fopenmp-version=45 -fopenmp-targets=powerpc64le-ibm-linux-gnu -x c++ -triple aarch64-unknown-unknown -std=c++11 -include-pch %t -verify %s -emit-llvm -o - | FileCheck %s --check-prefix=CHECK14
26 // RUN: %clang_cc1 -verify -fopenmp-simd -fopenmp-version=50 -DOMP5 -fopenmp-targets=powerpc64le-ibm-linux-gnu -x c++ -triple aarch64-unknown-unknown -emit-llvm %s -o - | FileCheck %s --check-prefix=CHECK15
27 // RUN: %clang_cc1 -fopenmp-simd -fopenmp-version=50 -DOMP5 -fopenmp-targets=powerpc64le-ibm-linux-gnu -x c++ -std=c++11 -triple aarch64-unknown-unknown -emit-pch -o %t %s
28 // RUN: %clang_cc1 -fopenmp-simd -fopenmp-version=50 -DOMP5 -fopenmp-targets=powerpc64le-ibm-linux-gnu -x c++ -triple aarch64-unknown-unknown -std=c++11 -include-pch %t -verify %s -emit-llvm -o - | FileCheck %s --check-prefix=CHECK16
29 // expected-no-diagnostics
30 #ifndef HEADER
31 #define HEADER
32
33 void fn1();
34 void fn2();
35 void fn3();
36 void fn4();
37 void fn5();
38 void fn6();
39
40 int Arg;
41
gtid_test()42 void gtid_test() {
43 #pragma omp target
44 #pragma omp teams distribute parallel for simd
45 for(int i = 0 ; i < 100; i++) {}
46
47 #pragma omp target
48 #pragma omp teams distribute parallel for simd if (parallel: false)
49 for(int i = 0 ; i < 100; i++) {
50 gtid_test();
51 }
52 }
53
54
55 template <typename T>
tmain(T Arg)56 int tmain(T Arg) {
57 #pragma omp target
58 #pragma omp teams distribute parallel for simd if (true)
59 for(int i = 0 ; i < 100; i++) {
60 fn1();
61 }
62 #pragma omp target
63 #pragma omp teams distribute parallel for simd if (false)
64 for(int i = 0 ; i < 100; i++) {
65 fn2();
66 }
67 #pragma omp target
68 #pragma omp teams distribute parallel for simd if (parallel: Arg)
69 for(int i = 0 ; i < 100; i++) {
70 fn3();
71 }
72 return 0;
73 }
74
main()75 int main() {
76 #pragma omp target
77 #pragma omp teams distribute parallel for simd if (true)
78 for(int i = 0 ; i < 100; i++) {
79
80
81 fn4();
82 }
83
84 #pragma omp target
85 #pragma omp teams distribute parallel for simd if (false)
86 for(int i = 0 ; i < 100; i++) {
87
88
89 fn5();
90 }
91
92 #pragma omp target
93 #pragma omp teams distribute parallel for simd if (Arg)
94 for(int i = 0 ; i < 100; i++) {
95
96
97 fn6();
98 }
99
100 return tmain(Arg);
101 }
102
103
104
105
106
107
108 // call void [[T_OUTLINE_FUN_3:@.+]](
109
110
111
112 #endif
113 // CHECK1-LABEL: define {{[^@]+}}@_Z9gtid_testv
114 // CHECK1-SAME: () #[[ATTR0:[0-9]+]] {
115 // CHECK1-NEXT: entry:
116 // CHECK1-NEXT: [[TMP:%.*]] = alloca i32, align 4
117 // CHECK1-NEXT: [[_TMP1:%.*]] = alloca i32, align 4
118 // CHECK1-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3:[0-9]+]], i64 -1, i64 100)
119 // CHECK1-NEXT: [[TMP0:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9gtid_testv_l43.region_id, i32 0, i8** null, i8** null, i64* null, i64* null, i8** null, i8** null, i32 0, i32 0)
120 // CHECK1-NEXT: [[TMP1:%.*]] = icmp ne i32 [[TMP0]], 0
121 // CHECK1-NEXT: br i1 [[TMP1]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
122 // CHECK1: omp_offload.failed:
123 // CHECK1-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9gtid_testv_l43() #[[ATTR2:[0-9]+]]
124 // CHECK1-NEXT: br label [[OMP_OFFLOAD_CONT]]
125 // CHECK1: omp_offload.cont:
126 // CHECK1-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i64 100)
127 // CHECK1-NEXT: [[TMP2:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9gtid_testv_l47.region_id, i32 0, i8** null, i8** null, i64* null, i64* null, i8** null, i8** null, i32 0, i32 1)
128 // CHECK1-NEXT: [[TMP3:%.*]] = icmp ne i32 [[TMP2]], 0
129 // CHECK1-NEXT: br i1 [[TMP3]], label [[OMP_OFFLOAD_FAILED2:%.*]], label [[OMP_OFFLOAD_CONT3:%.*]]
130 // CHECK1: omp_offload.failed2:
131 // CHECK1-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9gtid_testv_l47() #[[ATTR2]]
132 // CHECK1-NEXT: br label [[OMP_OFFLOAD_CONT3]]
133 // CHECK1: omp_offload.cont3:
134 // CHECK1-NEXT: ret void
135 //
136 //
137 // CHECK1-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9gtid_testv_l43
138 // CHECK1-SAME: () #[[ATTR1:[0-9]+]] {
139 // CHECK1-NEXT: entry:
140 // CHECK1-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 0, void (i32*, i32*, ...)* bitcast (void (i32*, i32*)* @.omp_outlined. to void (i32*, i32*, ...)*))
141 // CHECK1-NEXT: ret void
142 //
143 //
144 // CHECK1-LABEL: define {{[^@]+}}@.omp_outlined.
145 // CHECK1-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]]) #[[ATTR1]] {
146 // CHECK1-NEXT: entry:
147 // CHECK1-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
148 // CHECK1-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
149 // CHECK1-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
150 // CHECK1-NEXT: [[TMP:%.*]] = alloca i32, align 4
151 // CHECK1-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4
152 // CHECK1-NEXT: [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4
153 // CHECK1-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
154 // CHECK1-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
155 // CHECK1-NEXT: [[I:%.*]] = alloca i32, align 4
156 // CHECK1-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
157 // CHECK1-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
158 // CHECK1-NEXT: store i32 0, i32* [[DOTOMP_COMB_LB]], align 4
159 // CHECK1-NEXT: store i32 99, i32* [[DOTOMP_COMB_UB]], align 4
160 // CHECK1-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4
161 // CHECK1-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4
162 // CHECK1-NEXT: [[TMP0:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
163 // CHECK1-NEXT: [[TMP1:%.*]] = load i32, i32* [[TMP0]], align 4
164 // CHECK1-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB1:[0-9]+]], i32 [[TMP1]], i32 92, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_COMB_LB]], i32* [[DOTOMP_COMB_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1)
165 // CHECK1-NEXT: [[TMP2:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
166 // CHECK1-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP2]], 99
167 // CHECK1-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
168 // CHECK1: cond.true:
169 // CHECK1-NEXT: br label [[COND_END:%.*]]
170 // CHECK1: cond.false:
171 // CHECK1-NEXT: [[TMP3:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
172 // CHECK1-NEXT: br label [[COND_END]]
173 // CHECK1: cond.end:
174 // CHECK1-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP3]], [[COND_FALSE]] ]
175 // CHECK1-NEXT: store i32 [[COND]], i32* [[DOTOMP_COMB_UB]], align 4
176 // CHECK1-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4
177 // CHECK1-NEXT: store i32 [[TMP4]], i32* [[DOTOMP_IV]], align 4
178 // CHECK1-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
179 // CHECK1: omp.inner.for.cond:
180 // CHECK1-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
181 // CHECK1-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
182 // CHECK1-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP5]], [[TMP6]]
183 // CHECK1-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
184 // CHECK1: omp.inner.for.body:
185 // CHECK1-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4
186 // CHECK1-NEXT: [[TMP8:%.*]] = zext i32 [[TMP7]] to i64
187 // CHECK1-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
188 // CHECK1-NEXT: [[TMP10:%.*]] = zext i32 [[TMP9]] to i64
189 // CHECK1-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB3]], i32 2, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64, i64)* @.omp_outlined..1 to void (i32*, i32*, ...)*), i64 [[TMP8]], i64 [[TMP10]])
190 // CHECK1-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
191 // CHECK1: omp.inner.for.inc:
192 // CHECK1-NEXT: [[TMP11:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
193 // CHECK1-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4
194 // CHECK1-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP11]], [[TMP12]]
195 // CHECK1-NEXT: store i32 [[ADD]], i32* [[DOTOMP_IV]], align 4
196 // CHECK1-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP10:![0-9]+]]
197 // CHECK1: omp.inner.for.end:
198 // CHECK1-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
199 // CHECK1: omp.loop.exit:
200 // CHECK1-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP1]])
201 // CHECK1-NEXT: [[TMP13:%.*]] = load i32, i32* [[DOTOMP_IS_LAST]], align 4
202 // CHECK1-NEXT: [[TMP14:%.*]] = icmp ne i32 [[TMP13]], 0
203 // CHECK1-NEXT: br i1 [[TMP14]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
204 // CHECK1: .omp.final.then:
205 // CHECK1-NEXT: store i32 100, i32* [[I]], align 4
206 // CHECK1-NEXT: br label [[DOTOMP_FINAL_DONE]]
207 // CHECK1: .omp.final.done:
208 // CHECK1-NEXT: ret void
209 //
210 //
211 // CHECK1-LABEL: define {{[^@]+}}@.omp_outlined..1
212 // CHECK1-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[DOTPREVIOUS_LB_:%.*]], i64 [[DOTPREVIOUS_UB_:%.*]]) #[[ATTR1]] {
213 // CHECK1-NEXT: entry:
214 // CHECK1-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
215 // CHECK1-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
216 // CHECK1-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i64, align 8
217 // CHECK1-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i64, align 8
218 // CHECK1-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
219 // CHECK1-NEXT: [[TMP:%.*]] = alloca i32, align 4
220 // CHECK1-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4
221 // CHECK1-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4
222 // CHECK1-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
223 // CHECK1-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
224 // CHECK1-NEXT: [[I:%.*]] = alloca i32, align 4
225 // CHECK1-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
226 // CHECK1-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
227 // CHECK1-NEXT: store i64 [[DOTPREVIOUS_LB_]], i64* [[DOTPREVIOUS_LB__ADDR]], align 8
228 // CHECK1-NEXT: store i64 [[DOTPREVIOUS_UB_]], i64* [[DOTPREVIOUS_UB__ADDR]], align 8
229 // CHECK1-NEXT: store i32 0, i32* [[DOTOMP_LB]], align 4
230 // CHECK1-NEXT: store i32 99, i32* [[DOTOMP_UB]], align 4
231 // CHECK1-NEXT: [[TMP0:%.*]] = load i64, i64* [[DOTPREVIOUS_LB__ADDR]], align 8
232 // CHECK1-NEXT: [[CONV:%.*]] = trunc i64 [[TMP0]] to i32
233 // CHECK1-NEXT: [[TMP1:%.*]] = load i64, i64* [[DOTPREVIOUS_UB__ADDR]], align 8
234 // CHECK1-NEXT: [[CONV1:%.*]] = trunc i64 [[TMP1]] to i32
235 // CHECK1-NEXT: store i32 [[CONV]], i32* [[DOTOMP_LB]], align 4
236 // CHECK1-NEXT: store i32 [[CONV1]], i32* [[DOTOMP_UB]], align 4
237 // CHECK1-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4
238 // CHECK1-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4
239 // CHECK1-NEXT: [[TMP2:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
240 // CHECK1-NEXT: [[TMP3:%.*]] = load i32, i32* [[TMP2]], align 4
241 // CHECK1-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB2:[0-9]+]], i32 [[TMP3]], i32 34, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_LB]], i32* [[DOTOMP_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1)
242 // CHECK1-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
243 // CHECK1-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP4]], 99
244 // CHECK1-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
245 // CHECK1: cond.true:
246 // CHECK1-NEXT: br label [[COND_END:%.*]]
247 // CHECK1: cond.false:
248 // CHECK1-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
249 // CHECK1-NEXT: br label [[COND_END]]
250 // CHECK1: cond.end:
251 // CHECK1-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP5]], [[COND_FALSE]] ]
252 // CHECK1-NEXT: store i32 [[COND]], i32* [[DOTOMP_UB]], align 4
253 // CHECK1-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4
254 // CHECK1-NEXT: store i32 [[TMP6]], i32* [[DOTOMP_IV]], align 4
255 // CHECK1-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
256 // CHECK1: omp.inner.for.cond:
257 // CHECK1-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
258 // CHECK1-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
259 // CHECK1-NEXT: [[CMP2:%.*]] = icmp sle i32 [[TMP7]], [[TMP8]]
260 // CHECK1-NEXT: br i1 [[CMP2]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
261 // CHECK1: omp.inner.for.body:
262 // CHECK1-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
263 // CHECK1-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP9]], 1
264 // CHECK1-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
265 // CHECK1-NEXT: store i32 [[ADD]], i32* [[I]], align 4
266 // CHECK1-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
267 // CHECK1: omp.body.continue:
268 // CHECK1-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
269 // CHECK1: omp.inner.for.inc:
270 // CHECK1-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
271 // CHECK1-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP10]], 1
272 // CHECK1-NEXT: store i32 [[ADD3]], i32* [[DOTOMP_IV]], align 4
273 // CHECK1-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP12:![0-9]+]]
274 // CHECK1: omp.inner.for.end:
275 // CHECK1-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
276 // CHECK1: omp.loop.exit:
277 // CHECK1-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP3]])
278 // CHECK1-NEXT: [[TMP11:%.*]] = load i32, i32* [[DOTOMP_IS_LAST]], align 4
279 // CHECK1-NEXT: [[TMP12:%.*]] = icmp ne i32 [[TMP11]], 0
280 // CHECK1-NEXT: br i1 [[TMP12]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
281 // CHECK1: .omp.final.then:
282 // CHECK1-NEXT: store i32 100, i32* [[I]], align 4
283 // CHECK1-NEXT: br label [[DOTOMP_FINAL_DONE]]
284 // CHECK1: .omp.final.done:
285 // CHECK1-NEXT: ret void
286 //
287 //
288 // CHECK1-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9gtid_testv_l47
289 // CHECK1-SAME: () #[[ATTR1]] {
290 // CHECK1-NEXT: entry:
291 // CHECK1-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 0, void (i32*, i32*, ...)* bitcast (void (i32*, i32*)* @.omp_outlined..2 to void (i32*, i32*, ...)*))
292 // CHECK1-NEXT: ret void
293 //
294 //
295 // CHECK1-LABEL: define {{[^@]+}}@.omp_outlined..2
296 // CHECK1-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]]) #[[ATTR1]] {
297 // CHECK1-NEXT: entry:
298 // CHECK1-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
299 // CHECK1-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
300 // CHECK1-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
301 // CHECK1-NEXT: [[TMP:%.*]] = alloca i32, align 4
302 // CHECK1-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4
303 // CHECK1-NEXT: [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4
304 // CHECK1-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
305 // CHECK1-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
306 // CHECK1-NEXT: [[I:%.*]] = alloca i32, align 4
307 // CHECK1-NEXT: [[DOTBOUND_ZERO_ADDR:%.*]] = alloca i32, align 4
308 // CHECK1-NEXT: store i32 0, i32* [[DOTBOUND_ZERO_ADDR]], align 4
309 // CHECK1-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
310 // CHECK1-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
311 // CHECK1-NEXT: store i32 0, i32* [[DOTOMP_COMB_LB]], align 4
312 // CHECK1-NEXT: store i32 99, i32* [[DOTOMP_COMB_UB]], align 4
313 // CHECK1-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4
314 // CHECK1-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4
315 // CHECK1-NEXT: [[TMP0:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
316 // CHECK1-NEXT: [[TMP1:%.*]] = load i32, i32* [[TMP0]], align 4
317 // CHECK1-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB1]], i32 [[TMP1]], i32 92, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_COMB_LB]], i32* [[DOTOMP_COMB_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1)
318 // CHECK1-NEXT: [[TMP2:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
319 // CHECK1-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP2]], 99
320 // CHECK1-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
321 // CHECK1: cond.true:
322 // CHECK1-NEXT: br label [[COND_END:%.*]]
323 // CHECK1: cond.false:
324 // CHECK1-NEXT: [[TMP3:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
325 // CHECK1-NEXT: br label [[COND_END]]
326 // CHECK1: cond.end:
327 // CHECK1-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP3]], [[COND_FALSE]] ]
328 // CHECK1-NEXT: store i32 [[COND]], i32* [[DOTOMP_COMB_UB]], align 4
329 // CHECK1-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4
330 // CHECK1-NEXT: store i32 [[TMP4]], i32* [[DOTOMP_IV]], align 4
331 // CHECK1-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
332 // CHECK1: omp.inner.for.cond:
333 // CHECK1-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
334 // CHECK1-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
335 // CHECK1-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP5]], [[TMP6]]
336 // CHECK1-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
337 // CHECK1: omp.inner.for.body:
338 // CHECK1-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4
339 // CHECK1-NEXT: [[TMP8:%.*]] = zext i32 [[TMP7]] to i64
340 // CHECK1-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
341 // CHECK1-NEXT: [[TMP10:%.*]] = zext i32 [[TMP9]] to i64
342 // CHECK1-NEXT: call void @__kmpc_serialized_parallel(%struct.ident_t* @[[GLOB3]], i32 [[TMP1]])
343 // CHECK1-NEXT: [[TMP11:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
344 // CHECK1-NEXT: call void @.omp_outlined..3(i32* [[TMP11]], i32* [[DOTBOUND_ZERO_ADDR]], i64 [[TMP8]], i64 [[TMP10]]) #[[ATTR2]]
345 // CHECK1-NEXT: call void @__kmpc_end_serialized_parallel(%struct.ident_t* @[[GLOB3]], i32 [[TMP1]])
346 // CHECK1-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
347 // CHECK1: omp.inner.for.inc:
348 // CHECK1-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
349 // CHECK1-NEXT: [[TMP13:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4
350 // CHECK1-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP12]], [[TMP13]]
351 // CHECK1-NEXT: store i32 [[ADD]], i32* [[DOTOMP_IV]], align 4
352 // CHECK1-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP15:![0-9]+]]
353 // CHECK1: omp.inner.for.end:
354 // CHECK1-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
355 // CHECK1: omp.loop.exit:
356 // CHECK1-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP1]])
357 // CHECK1-NEXT: [[TMP14:%.*]] = load i32, i32* [[DOTOMP_IS_LAST]], align 4
358 // CHECK1-NEXT: [[TMP15:%.*]] = icmp ne i32 [[TMP14]], 0
359 // CHECK1-NEXT: br i1 [[TMP15]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
360 // CHECK1: .omp.final.then:
361 // CHECK1-NEXT: store i32 100, i32* [[I]], align 4
362 // CHECK1-NEXT: br label [[DOTOMP_FINAL_DONE]]
363 // CHECK1: .omp.final.done:
364 // CHECK1-NEXT: ret void
365 //
366 //
367 // CHECK1-LABEL: define {{[^@]+}}@.omp_outlined..3
368 // CHECK1-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[DOTPREVIOUS_LB_:%.*]], i64 [[DOTPREVIOUS_UB_:%.*]]) #[[ATTR1]] {
369 // CHECK1-NEXT: entry:
370 // CHECK1-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
371 // CHECK1-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
372 // CHECK1-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i64, align 8
373 // CHECK1-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i64, align 8
374 // CHECK1-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
375 // CHECK1-NEXT: [[TMP:%.*]] = alloca i32, align 4
376 // CHECK1-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4
377 // CHECK1-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4
378 // CHECK1-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
379 // CHECK1-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
380 // CHECK1-NEXT: [[I:%.*]] = alloca i32, align 4
381 // CHECK1-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
382 // CHECK1-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
383 // CHECK1-NEXT: store i64 [[DOTPREVIOUS_LB_]], i64* [[DOTPREVIOUS_LB__ADDR]], align 8
384 // CHECK1-NEXT: store i64 [[DOTPREVIOUS_UB_]], i64* [[DOTPREVIOUS_UB__ADDR]], align 8
385 // CHECK1-NEXT: store i32 0, i32* [[DOTOMP_LB]], align 4
386 // CHECK1-NEXT: store i32 99, i32* [[DOTOMP_UB]], align 4
387 // CHECK1-NEXT: [[TMP0:%.*]] = load i64, i64* [[DOTPREVIOUS_LB__ADDR]], align 8
388 // CHECK1-NEXT: [[CONV:%.*]] = trunc i64 [[TMP0]] to i32
389 // CHECK1-NEXT: [[TMP1:%.*]] = load i64, i64* [[DOTPREVIOUS_UB__ADDR]], align 8
390 // CHECK1-NEXT: [[CONV1:%.*]] = trunc i64 [[TMP1]] to i32
391 // CHECK1-NEXT: store i32 [[CONV]], i32* [[DOTOMP_LB]], align 4
392 // CHECK1-NEXT: store i32 [[CONV1]], i32* [[DOTOMP_UB]], align 4
393 // CHECK1-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4
394 // CHECK1-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4
395 // CHECK1-NEXT: [[TMP2:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
396 // CHECK1-NEXT: [[TMP3:%.*]] = load i32, i32* [[TMP2]], align 4
397 // CHECK1-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB2]], i32 [[TMP3]], i32 34, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_LB]], i32* [[DOTOMP_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1)
398 // CHECK1-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
399 // CHECK1-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP4]], 99
400 // CHECK1-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
401 // CHECK1: cond.true:
402 // CHECK1-NEXT: br label [[COND_END:%.*]]
403 // CHECK1: cond.false:
404 // CHECK1-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
405 // CHECK1-NEXT: br label [[COND_END]]
406 // CHECK1: cond.end:
407 // CHECK1-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP5]], [[COND_FALSE]] ]
408 // CHECK1-NEXT: store i32 [[COND]], i32* [[DOTOMP_UB]], align 4
409 // CHECK1-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4
410 // CHECK1-NEXT: store i32 [[TMP6]], i32* [[DOTOMP_IV]], align 4
411 // CHECK1-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
412 // CHECK1: omp.inner.for.cond:
413 // CHECK1-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
414 // CHECK1-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
415 // CHECK1-NEXT: [[CMP2:%.*]] = icmp sle i32 [[TMP7]], [[TMP8]]
416 // CHECK1-NEXT: br i1 [[CMP2]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
417 // CHECK1: omp.inner.for.body:
418 // CHECK1-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
419 // CHECK1-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP9]], 1
420 // CHECK1-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
421 // CHECK1-NEXT: store i32 [[ADD]], i32* [[I]], align 4
422 // CHECK1-NEXT: call void @_Z9gtid_testv()
423 // CHECK1-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
424 // CHECK1: omp.body.continue:
425 // CHECK1-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
426 // CHECK1: omp.inner.for.inc:
427 // CHECK1-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
428 // CHECK1-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP10]], 1
429 // CHECK1-NEXT: store i32 [[ADD3]], i32* [[DOTOMP_IV]], align 4
430 // CHECK1-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP16:![0-9]+]]
431 // CHECK1: omp.inner.for.end:
432 // CHECK1-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
433 // CHECK1: omp.loop.exit:
434 // CHECK1-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP3]])
435 // CHECK1-NEXT: [[TMP11:%.*]] = load i32, i32* [[DOTOMP_IS_LAST]], align 4
436 // CHECK1-NEXT: [[TMP12:%.*]] = icmp ne i32 [[TMP11]], 0
437 // CHECK1-NEXT: br i1 [[TMP12]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
438 // CHECK1: .omp.final.then:
439 // CHECK1-NEXT: store i32 100, i32* [[I]], align 4
440 // CHECK1-NEXT: br label [[DOTOMP_FINAL_DONE]]
441 // CHECK1: .omp.final.done:
442 // CHECK1-NEXT: ret void
443 //
444 //
445 // CHECK1-LABEL: define {{[^@]+}}@main
446 // CHECK1-SAME: () #[[ATTR3:[0-9]+]] {
447 // CHECK1-NEXT: entry:
448 // CHECK1-NEXT: [[RETVAL:%.*]] = alloca i32, align 4
449 // CHECK1-NEXT: [[TMP:%.*]] = alloca i32, align 4
450 // CHECK1-NEXT: [[_TMP1:%.*]] = alloca i32, align 4
451 // CHECK1-NEXT: [[ARG_CASTED:%.*]] = alloca i64, align 8
452 // CHECK1-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [1 x i8*], align 8
453 // CHECK1-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [1 x i8*], align 8
454 // CHECK1-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [1 x i8*], align 8
455 // CHECK1-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i8, align 1
456 // CHECK1-NEXT: [[_TMP5:%.*]] = alloca i32, align 4
457 // CHECK1-NEXT: store i32 0, i32* [[RETVAL]], align 4
458 // CHECK1-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i64 100)
459 // CHECK1-NEXT: [[TMP0:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l76.region_id, i32 0, i8** null, i8** null, i64* null, i64* null, i8** null, i8** null, i32 0, i32 0)
460 // CHECK1-NEXT: [[TMP1:%.*]] = icmp ne i32 [[TMP0]], 0
461 // CHECK1-NEXT: br i1 [[TMP1]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
462 // CHECK1: omp_offload.failed:
463 // CHECK1-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l76() #[[ATTR2]]
464 // CHECK1-NEXT: br label [[OMP_OFFLOAD_CONT]]
465 // CHECK1: omp_offload.cont:
466 // CHECK1-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i64 100)
467 // CHECK1-NEXT: [[TMP2:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l84.region_id, i32 0, i8** null, i8** null, i64* null, i64* null, i8** null, i8** null, i32 0, i32 1)
468 // CHECK1-NEXT: [[TMP3:%.*]] = icmp ne i32 [[TMP2]], 0
469 // CHECK1-NEXT: br i1 [[TMP3]], label [[OMP_OFFLOAD_FAILED2:%.*]], label [[OMP_OFFLOAD_CONT3:%.*]]
470 // CHECK1: omp_offload.failed2:
471 // CHECK1-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l84() #[[ATTR2]]
472 // CHECK1-NEXT: br label [[OMP_OFFLOAD_CONT3]]
473 // CHECK1: omp_offload.cont3:
474 // CHECK1-NEXT: [[TMP4:%.*]] = load i32, i32* @Arg, align 4
475 // CHECK1-NEXT: [[CONV:%.*]] = bitcast i64* [[ARG_CASTED]] to i32*
476 // CHECK1-NEXT: store i32 [[TMP4]], i32* [[CONV]], align 4
477 // CHECK1-NEXT: [[TMP5:%.*]] = load i64, i64* [[ARG_CASTED]], align 8
478 // CHECK1-NEXT: [[TMP6:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
479 // CHECK1-NEXT: [[TMP7:%.*]] = bitcast i8** [[TMP6]] to i64*
480 // CHECK1-NEXT: store i64 [[TMP5]], i64* [[TMP7]], align 8
481 // CHECK1-NEXT: [[TMP8:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
482 // CHECK1-NEXT: [[TMP9:%.*]] = bitcast i8** [[TMP8]] to i64*
483 // CHECK1-NEXT: store i64 [[TMP5]], i64* [[TMP9]], align 8
484 // CHECK1-NEXT: [[TMP10:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0
485 // CHECK1-NEXT: store i8* null, i8** [[TMP10]], align 8
486 // CHECK1-NEXT: [[TMP11:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
487 // CHECK1-NEXT: [[TMP12:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
488 // CHECK1-NEXT: [[TMP13:%.*]] = load i32, i32* @Arg, align 4
489 // CHECK1-NEXT: [[TOBOOL:%.*]] = icmp ne i32 [[TMP13]], 0
490 // CHECK1-NEXT: [[FROMBOOL:%.*]] = zext i1 [[TOBOOL]] to i8
491 // CHECK1-NEXT: store i8 [[FROMBOOL]], i8* [[DOTCAPTURE_EXPR_]], align 1
492 // CHECK1-NEXT: [[TMP14:%.*]] = load i8, i8* [[DOTCAPTURE_EXPR_]], align 1
493 // CHECK1-NEXT: [[TOBOOL4:%.*]] = trunc i8 [[TMP14]] to i1
494 // CHECK1-NEXT: [[TMP15:%.*]] = select i1 [[TOBOOL4]], i32 0, i32 1
495 // CHECK1-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i64 100)
496 // CHECK1-NEXT: [[TMP16:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l92.region_id, i32 1, i8** [[TMP11]], i8** [[TMP12]], i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_sizes, i32 0, i32 0), i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_maptypes, i32 0, i32 0), i8** null, i8** null, i32 0, i32 [[TMP15]])
497 // CHECK1-NEXT: [[TMP17:%.*]] = icmp ne i32 [[TMP16]], 0
498 // CHECK1-NEXT: br i1 [[TMP17]], label [[OMP_OFFLOAD_FAILED6:%.*]], label [[OMP_OFFLOAD_CONT7:%.*]]
499 // CHECK1: omp_offload.failed6:
500 // CHECK1-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l92(i64 [[TMP5]]) #[[ATTR2]]
501 // CHECK1-NEXT: br label [[OMP_OFFLOAD_CONT7]]
502 // CHECK1: omp_offload.cont7:
503 // CHECK1-NEXT: [[TMP18:%.*]] = load i32, i32* @Arg, align 4
504 // CHECK1-NEXT: [[CALL:%.*]] = call i32 @_Z5tmainIiEiT_(i32 [[TMP18]])
505 // CHECK1-NEXT: ret i32 [[CALL]]
506 //
507 //
508 // CHECK1-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l76
509 // CHECK1-SAME: () #[[ATTR1]] {
510 // CHECK1-NEXT: entry:
511 // CHECK1-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 0, void (i32*, i32*, ...)* bitcast (void (i32*, i32*)* @.omp_outlined..4 to void (i32*, i32*, ...)*))
512 // CHECK1-NEXT: ret void
513 //
514 //
515 // CHECK1-LABEL: define {{[^@]+}}@.omp_outlined..4
516 // CHECK1-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]]) #[[ATTR1]] {
517 // CHECK1-NEXT: entry:
518 // CHECK1-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
519 // CHECK1-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
520 // CHECK1-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
521 // CHECK1-NEXT: [[TMP:%.*]] = alloca i32, align 4
522 // CHECK1-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4
523 // CHECK1-NEXT: [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4
524 // CHECK1-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
525 // CHECK1-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
526 // CHECK1-NEXT: [[I:%.*]] = alloca i32, align 4
527 // CHECK1-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
528 // CHECK1-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
529 // CHECK1-NEXT: store i32 0, i32* [[DOTOMP_COMB_LB]], align 4
530 // CHECK1-NEXT: store i32 99, i32* [[DOTOMP_COMB_UB]], align 4
531 // CHECK1-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4
532 // CHECK1-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4
533 // CHECK1-NEXT: [[TMP0:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
534 // CHECK1-NEXT: [[TMP1:%.*]] = load i32, i32* [[TMP0]], align 4
535 // CHECK1-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB1]], i32 [[TMP1]], i32 92, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_COMB_LB]], i32* [[DOTOMP_COMB_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1)
536 // CHECK1-NEXT: [[TMP2:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
537 // CHECK1-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP2]], 99
538 // CHECK1-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
539 // CHECK1: cond.true:
540 // CHECK1-NEXT: br label [[COND_END:%.*]]
541 // CHECK1: cond.false:
542 // CHECK1-NEXT: [[TMP3:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
543 // CHECK1-NEXT: br label [[COND_END]]
544 // CHECK1: cond.end:
545 // CHECK1-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP3]], [[COND_FALSE]] ]
546 // CHECK1-NEXT: store i32 [[COND]], i32* [[DOTOMP_COMB_UB]], align 4
547 // CHECK1-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4
548 // CHECK1-NEXT: store i32 [[TMP4]], i32* [[DOTOMP_IV]], align 4
549 // CHECK1-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
550 // CHECK1: omp.inner.for.cond:
551 // CHECK1-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
552 // CHECK1-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
553 // CHECK1-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP5]], [[TMP6]]
554 // CHECK1-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
555 // CHECK1: omp.inner.for.body:
556 // CHECK1-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4
557 // CHECK1-NEXT: [[TMP8:%.*]] = zext i32 [[TMP7]] to i64
558 // CHECK1-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
559 // CHECK1-NEXT: [[TMP10:%.*]] = zext i32 [[TMP9]] to i64
560 // CHECK1-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB3]], i32 2, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64, i64)* @.omp_outlined..5 to void (i32*, i32*, ...)*), i64 [[TMP8]], i64 [[TMP10]])
561 // CHECK1-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
562 // CHECK1: omp.inner.for.inc:
563 // CHECK1-NEXT: [[TMP11:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
564 // CHECK1-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4
565 // CHECK1-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP11]], [[TMP12]]
566 // CHECK1-NEXT: store i32 [[ADD]], i32* [[DOTOMP_IV]], align 4
567 // CHECK1-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP17:![0-9]+]]
568 // CHECK1: omp.inner.for.end:
569 // CHECK1-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
570 // CHECK1: omp.loop.exit:
571 // CHECK1-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP1]])
572 // CHECK1-NEXT: [[TMP13:%.*]] = load i32, i32* [[DOTOMP_IS_LAST]], align 4
573 // CHECK1-NEXT: [[TMP14:%.*]] = icmp ne i32 [[TMP13]], 0
574 // CHECK1-NEXT: br i1 [[TMP14]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
575 // CHECK1: .omp.final.then:
576 // CHECK1-NEXT: store i32 100, i32* [[I]], align 4
577 // CHECK1-NEXT: br label [[DOTOMP_FINAL_DONE]]
578 // CHECK1: .omp.final.done:
579 // CHECK1-NEXT: ret void
580 //
581 //
582 // CHECK1-LABEL: define {{[^@]+}}@.omp_outlined..5
583 // CHECK1-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[DOTPREVIOUS_LB_:%.*]], i64 [[DOTPREVIOUS_UB_:%.*]]) #[[ATTR1]] {
584 // CHECK1-NEXT: entry:
585 // CHECK1-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
586 // CHECK1-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
587 // CHECK1-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i64, align 8
588 // CHECK1-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i64, align 8
589 // CHECK1-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
590 // CHECK1-NEXT: [[TMP:%.*]] = alloca i32, align 4
591 // CHECK1-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4
592 // CHECK1-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4
593 // CHECK1-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
594 // CHECK1-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
595 // CHECK1-NEXT: [[I:%.*]] = alloca i32, align 4
596 // CHECK1-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
597 // CHECK1-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
598 // CHECK1-NEXT: store i64 [[DOTPREVIOUS_LB_]], i64* [[DOTPREVIOUS_LB__ADDR]], align 8
599 // CHECK1-NEXT: store i64 [[DOTPREVIOUS_UB_]], i64* [[DOTPREVIOUS_UB__ADDR]], align 8
600 // CHECK1-NEXT: store i32 0, i32* [[DOTOMP_LB]], align 4
601 // CHECK1-NEXT: store i32 99, i32* [[DOTOMP_UB]], align 4
602 // CHECK1-NEXT: [[TMP0:%.*]] = load i64, i64* [[DOTPREVIOUS_LB__ADDR]], align 8
603 // CHECK1-NEXT: [[CONV:%.*]] = trunc i64 [[TMP0]] to i32
604 // CHECK1-NEXT: [[TMP1:%.*]] = load i64, i64* [[DOTPREVIOUS_UB__ADDR]], align 8
605 // CHECK1-NEXT: [[CONV1:%.*]] = trunc i64 [[TMP1]] to i32
606 // CHECK1-NEXT: store i32 [[CONV]], i32* [[DOTOMP_LB]], align 4
607 // CHECK1-NEXT: store i32 [[CONV1]], i32* [[DOTOMP_UB]], align 4
608 // CHECK1-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4
609 // CHECK1-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4
610 // CHECK1-NEXT: [[TMP2:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
611 // CHECK1-NEXT: [[TMP3:%.*]] = load i32, i32* [[TMP2]], align 4
612 // CHECK1-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB2]], i32 [[TMP3]], i32 34, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_LB]], i32* [[DOTOMP_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1)
613 // CHECK1-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
614 // CHECK1-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP4]], 99
615 // CHECK1-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
616 // CHECK1: cond.true:
617 // CHECK1-NEXT: br label [[COND_END:%.*]]
618 // CHECK1: cond.false:
619 // CHECK1-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
620 // CHECK1-NEXT: br label [[COND_END]]
621 // CHECK1: cond.end:
622 // CHECK1-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP5]], [[COND_FALSE]] ]
623 // CHECK1-NEXT: store i32 [[COND]], i32* [[DOTOMP_UB]], align 4
624 // CHECK1-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4
625 // CHECK1-NEXT: store i32 [[TMP6]], i32* [[DOTOMP_IV]], align 4
626 // CHECK1-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
627 // CHECK1: omp.inner.for.cond:
628 // CHECK1-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
629 // CHECK1-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
630 // CHECK1-NEXT: [[CMP2:%.*]] = icmp sle i32 [[TMP7]], [[TMP8]]
631 // CHECK1-NEXT: br i1 [[CMP2]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
632 // CHECK1: omp.inner.for.body:
633 // CHECK1-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
634 // CHECK1-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP9]], 1
635 // CHECK1-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
636 // CHECK1-NEXT: store i32 [[ADD]], i32* [[I]], align 4
637 // CHECK1-NEXT: call void @_Z3fn4v()
638 // CHECK1-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
639 // CHECK1: omp.body.continue:
640 // CHECK1-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
641 // CHECK1: omp.inner.for.inc:
642 // CHECK1-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
643 // CHECK1-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP10]], 1
644 // CHECK1-NEXT: store i32 [[ADD3]], i32* [[DOTOMP_IV]], align 4
645 // CHECK1-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP18:![0-9]+]]
646 // CHECK1: omp.inner.for.end:
647 // CHECK1-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
648 // CHECK1: omp.loop.exit:
649 // CHECK1-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP3]])
650 // CHECK1-NEXT: [[TMP11:%.*]] = load i32, i32* [[DOTOMP_IS_LAST]], align 4
651 // CHECK1-NEXT: [[TMP12:%.*]] = icmp ne i32 [[TMP11]], 0
652 // CHECK1-NEXT: br i1 [[TMP12]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
653 // CHECK1: .omp.final.then:
654 // CHECK1-NEXT: store i32 100, i32* [[I]], align 4
655 // CHECK1-NEXT: br label [[DOTOMP_FINAL_DONE]]
656 // CHECK1: .omp.final.done:
657 // CHECK1-NEXT: ret void
658 //
659 //
660 // CHECK1-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l84
661 // CHECK1-SAME: () #[[ATTR1]] {
662 // CHECK1-NEXT: entry:
663 // CHECK1-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 0, void (i32*, i32*, ...)* bitcast (void (i32*, i32*)* @.omp_outlined..6 to void (i32*, i32*, ...)*))
664 // CHECK1-NEXT: ret void
665 //
666 //
667 // CHECK1-LABEL: define {{[^@]+}}@.omp_outlined..6
668 // CHECK1-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]]) #[[ATTR1]] {
669 // CHECK1-NEXT: entry:
670 // CHECK1-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
671 // CHECK1-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
672 // CHECK1-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
673 // CHECK1-NEXT: [[TMP:%.*]] = alloca i32, align 4
674 // CHECK1-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4
675 // CHECK1-NEXT: [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4
676 // CHECK1-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
677 // CHECK1-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
678 // CHECK1-NEXT: [[I:%.*]] = alloca i32, align 4
679 // CHECK1-NEXT: [[DOTBOUND_ZERO_ADDR:%.*]] = alloca i32, align 4
680 // CHECK1-NEXT: store i32 0, i32* [[DOTBOUND_ZERO_ADDR]], align 4
681 // CHECK1-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
682 // CHECK1-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
683 // CHECK1-NEXT: store i32 0, i32* [[DOTOMP_COMB_LB]], align 4
684 // CHECK1-NEXT: store i32 99, i32* [[DOTOMP_COMB_UB]], align 4
685 // CHECK1-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4
686 // CHECK1-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4
687 // CHECK1-NEXT: [[TMP0:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
688 // CHECK1-NEXT: [[TMP1:%.*]] = load i32, i32* [[TMP0]], align 4
689 // CHECK1-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB1]], i32 [[TMP1]], i32 92, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_COMB_LB]], i32* [[DOTOMP_COMB_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1)
690 // CHECK1-NEXT: [[TMP2:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
691 // CHECK1-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP2]], 99
692 // CHECK1-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
693 // CHECK1: cond.true:
694 // CHECK1-NEXT: br label [[COND_END:%.*]]
695 // CHECK1: cond.false:
696 // CHECK1-NEXT: [[TMP3:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
697 // CHECK1-NEXT: br label [[COND_END]]
698 // CHECK1: cond.end:
699 // CHECK1-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP3]], [[COND_FALSE]] ]
700 // CHECK1-NEXT: store i32 [[COND]], i32* [[DOTOMP_COMB_UB]], align 4
701 // CHECK1-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4
702 // CHECK1-NEXT: store i32 [[TMP4]], i32* [[DOTOMP_IV]], align 4
703 // CHECK1-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
704 // CHECK1: omp.inner.for.cond:
705 // CHECK1-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
706 // CHECK1-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
707 // CHECK1-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP5]], [[TMP6]]
708 // CHECK1-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
709 // CHECK1: omp.inner.for.body:
710 // CHECK1-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4
711 // CHECK1-NEXT: [[TMP8:%.*]] = zext i32 [[TMP7]] to i64
712 // CHECK1-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
713 // CHECK1-NEXT: [[TMP10:%.*]] = zext i32 [[TMP9]] to i64
714 // CHECK1-NEXT: call void @__kmpc_serialized_parallel(%struct.ident_t* @[[GLOB3]], i32 [[TMP1]])
715 // CHECK1-NEXT: [[TMP11:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
716 // CHECK1-NEXT: call void @.omp_outlined..7(i32* [[TMP11]], i32* [[DOTBOUND_ZERO_ADDR]], i64 [[TMP8]], i64 [[TMP10]]) #[[ATTR2]]
717 // CHECK1-NEXT: call void @__kmpc_end_serialized_parallel(%struct.ident_t* @[[GLOB3]], i32 [[TMP1]])
718 // CHECK1-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
719 // CHECK1: omp.inner.for.inc:
720 // CHECK1-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
721 // CHECK1-NEXT: [[TMP13:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4
722 // CHECK1-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP12]], [[TMP13]]
723 // CHECK1-NEXT: store i32 [[ADD]], i32* [[DOTOMP_IV]], align 4
724 // CHECK1-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP19:![0-9]+]]
725 // CHECK1: omp.inner.for.end:
726 // CHECK1-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
727 // CHECK1: omp.loop.exit:
728 // CHECK1-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP1]])
729 // CHECK1-NEXT: [[TMP14:%.*]] = load i32, i32* [[DOTOMP_IS_LAST]], align 4
730 // CHECK1-NEXT: [[TMP15:%.*]] = icmp ne i32 [[TMP14]], 0
731 // CHECK1-NEXT: br i1 [[TMP15]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
732 // CHECK1: .omp.final.then:
733 // CHECK1-NEXT: store i32 100, i32* [[I]], align 4
734 // CHECK1-NEXT: br label [[DOTOMP_FINAL_DONE]]
735 // CHECK1: .omp.final.done:
736 // CHECK1-NEXT: ret void
737 //
738 //
739 // CHECK1-LABEL: define {{[^@]+}}@.omp_outlined..7
740 // CHECK1-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[DOTPREVIOUS_LB_:%.*]], i64 [[DOTPREVIOUS_UB_:%.*]]) #[[ATTR1]] {
741 // CHECK1-NEXT: entry:
742 // CHECK1-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
743 // CHECK1-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
744 // CHECK1-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i64, align 8
745 // CHECK1-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i64, align 8
746 // CHECK1-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
747 // CHECK1-NEXT: [[TMP:%.*]] = alloca i32, align 4
748 // CHECK1-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4
749 // CHECK1-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4
750 // CHECK1-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
751 // CHECK1-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
752 // CHECK1-NEXT: [[I:%.*]] = alloca i32, align 4
753 // CHECK1-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
754 // CHECK1-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
755 // CHECK1-NEXT: store i64 [[DOTPREVIOUS_LB_]], i64* [[DOTPREVIOUS_LB__ADDR]], align 8
756 // CHECK1-NEXT: store i64 [[DOTPREVIOUS_UB_]], i64* [[DOTPREVIOUS_UB__ADDR]], align 8
757 // CHECK1-NEXT: store i32 0, i32* [[DOTOMP_LB]], align 4
758 // CHECK1-NEXT: store i32 99, i32* [[DOTOMP_UB]], align 4
759 // CHECK1-NEXT: [[TMP0:%.*]] = load i64, i64* [[DOTPREVIOUS_LB__ADDR]], align 8
760 // CHECK1-NEXT: [[CONV:%.*]] = trunc i64 [[TMP0]] to i32
761 // CHECK1-NEXT: [[TMP1:%.*]] = load i64, i64* [[DOTPREVIOUS_UB__ADDR]], align 8
762 // CHECK1-NEXT: [[CONV1:%.*]] = trunc i64 [[TMP1]] to i32
763 // CHECK1-NEXT: store i32 [[CONV]], i32* [[DOTOMP_LB]], align 4
764 // CHECK1-NEXT: store i32 [[CONV1]], i32* [[DOTOMP_UB]], align 4
765 // CHECK1-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4
766 // CHECK1-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4
767 // CHECK1-NEXT: [[TMP2:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
768 // CHECK1-NEXT: [[TMP3:%.*]] = load i32, i32* [[TMP2]], align 4
769 // CHECK1-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB2]], i32 [[TMP3]], i32 34, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_LB]], i32* [[DOTOMP_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1)
770 // CHECK1-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
771 // CHECK1-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP4]], 99
772 // CHECK1-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
773 // CHECK1: cond.true:
774 // CHECK1-NEXT: br label [[COND_END:%.*]]
775 // CHECK1: cond.false:
776 // CHECK1-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
777 // CHECK1-NEXT: br label [[COND_END]]
778 // CHECK1: cond.end:
779 // CHECK1-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP5]], [[COND_FALSE]] ]
780 // CHECK1-NEXT: store i32 [[COND]], i32* [[DOTOMP_UB]], align 4
781 // CHECK1-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4
782 // CHECK1-NEXT: store i32 [[TMP6]], i32* [[DOTOMP_IV]], align 4
783 // CHECK1-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
784 // CHECK1: omp.inner.for.cond:
785 // CHECK1-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
786 // CHECK1-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
787 // CHECK1-NEXT: [[CMP2:%.*]] = icmp sle i32 [[TMP7]], [[TMP8]]
788 // CHECK1-NEXT: br i1 [[CMP2]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
789 // CHECK1: omp.inner.for.body:
790 // CHECK1-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
791 // CHECK1-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP9]], 1
792 // CHECK1-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
793 // CHECK1-NEXT: store i32 [[ADD]], i32* [[I]], align 4
794 // CHECK1-NEXT: call void @_Z3fn5v()
795 // CHECK1-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
796 // CHECK1: omp.body.continue:
797 // CHECK1-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
798 // CHECK1: omp.inner.for.inc:
799 // CHECK1-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
800 // CHECK1-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP10]], 1
801 // CHECK1-NEXT: store i32 [[ADD3]], i32* [[DOTOMP_IV]], align 4
802 // CHECK1-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP20:![0-9]+]]
803 // CHECK1: omp.inner.for.end:
804 // CHECK1-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
805 // CHECK1: omp.loop.exit:
806 // CHECK1-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP3]])
807 // CHECK1-NEXT: [[TMP11:%.*]] = load i32, i32* [[DOTOMP_IS_LAST]], align 4
808 // CHECK1-NEXT: [[TMP12:%.*]] = icmp ne i32 [[TMP11]], 0
809 // CHECK1-NEXT: br i1 [[TMP12]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
810 // CHECK1: .omp.final.then:
811 // CHECK1-NEXT: store i32 100, i32* [[I]], align 4
812 // CHECK1-NEXT: br label [[DOTOMP_FINAL_DONE]]
813 // CHECK1: .omp.final.done:
814 // CHECK1-NEXT: ret void
815 //
816 //
817 // CHECK1-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l92
818 // CHECK1-SAME: (i64 [[ARG:%.*]]) #[[ATTR1]] {
819 // CHECK1-NEXT: entry:
820 // CHECK1-NEXT: [[ARG_ADDR:%.*]] = alloca i64, align 8
821 // CHECK1-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i8, align 1
822 // CHECK1-NEXT: [[DOTCAPTURE_EXPR__CASTED:%.*]] = alloca i64, align 8
823 // CHECK1-NEXT: store i64 [[ARG]], i64* [[ARG_ADDR]], align 8
824 // CHECK1-NEXT: [[CONV:%.*]] = bitcast i64* [[ARG_ADDR]] to i32*
825 // CHECK1-NEXT: [[TMP0:%.*]] = load i32, i32* [[CONV]], align 8
826 // CHECK1-NEXT: [[TOBOOL:%.*]] = icmp ne i32 [[TMP0]], 0
827 // CHECK1-NEXT: [[FROMBOOL:%.*]] = zext i1 [[TOBOOL]] to i8
828 // CHECK1-NEXT: store i8 [[FROMBOOL]], i8* [[DOTCAPTURE_EXPR_]], align 1
829 // CHECK1-NEXT: [[TMP1:%.*]] = load i8, i8* [[DOTCAPTURE_EXPR_]], align 1
830 // CHECK1-NEXT: [[TOBOOL1:%.*]] = trunc i8 [[TMP1]] to i1
831 // CHECK1-NEXT: [[CONV2:%.*]] = bitcast i64* [[DOTCAPTURE_EXPR__CASTED]] to i8*
832 // CHECK1-NEXT: [[FROMBOOL3:%.*]] = zext i1 [[TOBOOL1]] to i8
833 // CHECK1-NEXT: store i8 [[FROMBOOL3]], i8* [[CONV2]], align 1
834 // CHECK1-NEXT: [[TMP2:%.*]] = load i64, i64* [[DOTCAPTURE_EXPR__CASTED]], align 8
835 // CHECK1-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64)* @.omp_outlined..8 to void (i32*, i32*, ...)*), i64 [[TMP2]])
836 // CHECK1-NEXT: ret void
837 //
838 //
839 // CHECK1-LABEL: define {{[^@]+}}@.omp_outlined..8
840 // CHECK1-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[DOTCAPTURE_EXPR_:%.*]]) #[[ATTR1]] {
841 // CHECK1-NEXT: entry:
842 // CHECK1-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
843 // CHECK1-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
844 // CHECK1-NEXT: [[DOTCAPTURE_EXPR__ADDR:%.*]] = alloca i64, align 8
845 // CHECK1-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
846 // CHECK1-NEXT: [[TMP:%.*]] = alloca i32, align 4
847 // CHECK1-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4
848 // CHECK1-NEXT: [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4
849 // CHECK1-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
850 // CHECK1-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
851 // CHECK1-NEXT: [[I:%.*]] = alloca i32, align 4
852 // CHECK1-NEXT: [[DOTBOUND_ZERO_ADDR:%.*]] = alloca i32, align 4
853 // CHECK1-NEXT: store i32 0, i32* [[DOTBOUND_ZERO_ADDR]], align 4
854 // CHECK1-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
855 // CHECK1-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
856 // CHECK1-NEXT: store i64 [[DOTCAPTURE_EXPR_]], i64* [[DOTCAPTURE_EXPR__ADDR]], align 8
857 // CHECK1-NEXT: [[CONV:%.*]] = bitcast i64* [[DOTCAPTURE_EXPR__ADDR]] to i8*
858 // CHECK1-NEXT: store i32 0, i32* [[DOTOMP_COMB_LB]], align 4
859 // CHECK1-NEXT: store i32 99, i32* [[DOTOMP_COMB_UB]], align 4
860 // CHECK1-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4
861 // CHECK1-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4
862 // CHECK1-NEXT: [[TMP0:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
863 // CHECK1-NEXT: [[TMP1:%.*]] = load i32, i32* [[TMP0]], align 4
864 // CHECK1-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB1]], i32 [[TMP1]], i32 92, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_COMB_LB]], i32* [[DOTOMP_COMB_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1)
865 // CHECK1-NEXT: [[TMP2:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
866 // CHECK1-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP2]], 99
867 // CHECK1-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
868 // CHECK1: cond.true:
869 // CHECK1-NEXT: br label [[COND_END:%.*]]
870 // CHECK1: cond.false:
871 // CHECK1-NEXT: [[TMP3:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
872 // CHECK1-NEXT: br label [[COND_END]]
873 // CHECK1: cond.end:
874 // CHECK1-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP3]], [[COND_FALSE]] ]
875 // CHECK1-NEXT: store i32 [[COND]], i32* [[DOTOMP_COMB_UB]], align 4
876 // CHECK1-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4
877 // CHECK1-NEXT: store i32 [[TMP4]], i32* [[DOTOMP_IV]], align 4
878 // CHECK1-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
879 // CHECK1: omp.inner.for.cond:
880 // CHECK1-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
881 // CHECK1-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
882 // CHECK1-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP5]], [[TMP6]]
883 // CHECK1-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
884 // CHECK1: omp.inner.for.body:
885 // CHECK1-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4
886 // CHECK1-NEXT: [[TMP8:%.*]] = zext i32 [[TMP7]] to i64
887 // CHECK1-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
888 // CHECK1-NEXT: [[TMP10:%.*]] = zext i32 [[TMP9]] to i64
889 // CHECK1-NEXT: [[TMP11:%.*]] = load i8, i8* [[CONV]], align 8
890 // CHECK1-NEXT: [[TOBOOL:%.*]] = trunc i8 [[TMP11]] to i1
891 // CHECK1-NEXT: br i1 [[TOBOOL]], label [[OMP_IF_THEN:%.*]], label [[OMP_IF_ELSE:%.*]]
892 // CHECK1: omp_if.then:
893 // CHECK1-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB3]], i32 2, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64, i64)* @.omp_outlined..9 to void (i32*, i32*, ...)*), i64 [[TMP8]], i64 [[TMP10]])
894 // CHECK1-NEXT: br label [[OMP_IF_END:%.*]]
895 // CHECK1: omp_if.else:
896 // CHECK1-NEXT: call void @__kmpc_serialized_parallel(%struct.ident_t* @[[GLOB3]], i32 [[TMP1]])
897 // CHECK1-NEXT: [[TMP12:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
898 // CHECK1-NEXT: call void @.omp_outlined..9(i32* [[TMP12]], i32* [[DOTBOUND_ZERO_ADDR]], i64 [[TMP8]], i64 [[TMP10]]) #[[ATTR2]]
899 // CHECK1-NEXT: call void @__kmpc_end_serialized_parallel(%struct.ident_t* @[[GLOB3]], i32 [[TMP1]])
900 // CHECK1-NEXT: br label [[OMP_IF_END]]
901 // CHECK1: omp_if.end:
902 // CHECK1-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
903 // CHECK1: omp.inner.for.inc:
904 // CHECK1-NEXT: [[TMP13:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
905 // CHECK1-NEXT: [[TMP14:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4
906 // CHECK1-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP13]], [[TMP14]]
907 // CHECK1-NEXT: store i32 [[ADD]], i32* [[DOTOMP_IV]], align 4
908 // CHECK1-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP21:![0-9]+]]
909 // CHECK1: omp.inner.for.end:
910 // CHECK1-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
911 // CHECK1: omp.loop.exit:
912 // CHECK1-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP1]])
913 // CHECK1-NEXT: [[TMP15:%.*]] = load i32, i32* [[DOTOMP_IS_LAST]], align 4
914 // CHECK1-NEXT: [[TMP16:%.*]] = icmp ne i32 [[TMP15]], 0
915 // CHECK1-NEXT: br i1 [[TMP16]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
916 // CHECK1: .omp.final.then:
917 // CHECK1-NEXT: store i32 100, i32* [[I]], align 4
918 // CHECK1-NEXT: br label [[DOTOMP_FINAL_DONE]]
919 // CHECK1: .omp.final.done:
920 // CHECK1-NEXT: ret void
921 //
922 //
923 // CHECK1-LABEL: define {{[^@]+}}@.omp_outlined..9
924 // CHECK1-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[DOTPREVIOUS_LB_:%.*]], i64 [[DOTPREVIOUS_UB_:%.*]]) #[[ATTR1]] {
925 // CHECK1-NEXT: entry:
926 // CHECK1-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
927 // CHECK1-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
928 // CHECK1-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i64, align 8
929 // CHECK1-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i64, align 8
930 // CHECK1-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
931 // CHECK1-NEXT: [[TMP:%.*]] = alloca i32, align 4
932 // CHECK1-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4
933 // CHECK1-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4
934 // CHECK1-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
935 // CHECK1-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
936 // CHECK1-NEXT: [[I:%.*]] = alloca i32, align 4
937 // CHECK1-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
938 // CHECK1-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
939 // CHECK1-NEXT: store i64 [[DOTPREVIOUS_LB_]], i64* [[DOTPREVIOUS_LB__ADDR]], align 8
940 // CHECK1-NEXT: store i64 [[DOTPREVIOUS_UB_]], i64* [[DOTPREVIOUS_UB__ADDR]], align 8
941 // CHECK1-NEXT: store i32 0, i32* [[DOTOMP_LB]], align 4
942 // CHECK1-NEXT: store i32 99, i32* [[DOTOMP_UB]], align 4
943 // CHECK1-NEXT: [[TMP0:%.*]] = load i64, i64* [[DOTPREVIOUS_LB__ADDR]], align 8
944 // CHECK1-NEXT: [[CONV:%.*]] = trunc i64 [[TMP0]] to i32
945 // CHECK1-NEXT: [[TMP1:%.*]] = load i64, i64* [[DOTPREVIOUS_UB__ADDR]], align 8
946 // CHECK1-NEXT: [[CONV1:%.*]] = trunc i64 [[TMP1]] to i32
947 // CHECK1-NEXT: store i32 [[CONV]], i32* [[DOTOMP_LB]], align 4
948 // CHECK1-NEXT: store i32 [[CONV1]], i32* [[DOTOMP_UB]], align 4
949 // CHECK1-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4
950 // CHECK1-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4
951 // CHECK1-NEXT: [[TMP2:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
952 // CHECK1-NEXT: [[TMP3:%.*]] = load i32, i32* [[TMP2]], align 4
953 // CHECK1-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB2]], i32 [[TMP3]], i32 34, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_LB]], i32* [[DOTOMP_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1)
954 // CHECK1-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
955 // CHECK1-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP4]], 99
956 // CHECK1-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
957 // CHECK1: cond.true:
958 // CHECK1-NEXT: br label [[COND_END:%.*]]
959 // CHECK1: cond.false:
960 // CHECK1-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
961 // CHECK1-NEXT: br label [[COND_END]]
962 // CHECK1: cond.end:
963 // CHECK1-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP5]], [[COND_FALSE]] ]
964 // CHECK1-NEXT: store i32 [[COND]], i32* [[DOTOMP_UB]], align 4
965 // CHECK1-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4
966 // CHECK1-NEXT: store i32 [[TMP6]], i32* [[DOTOMP_IV]], align 4
967 // CHECK1-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
968 // CHECK1: omp.inner.for.cond:
969 // CHECK1-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
970 // CHECK1-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
971 // CHECK1-NEXT: [[CMP2:%.*]] = icmp sle i32 [[TMP7]], [[TMP8]]
972 // CHECK1-NEXT: br i1 [[CMP2]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
973 // CHECK1: omp.inner.for.body:
974 // CHECK1-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
975 // CHECK1-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP9]], 1
976 // CHECK1-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
977 // CHECK1-NEXT: store i32 [[ADD]], i32* [[I]], align 4
978 // CHECK1-NEXT: call void @_Z3fn6v()
979 // CHECK1-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
980 // CHECK1: omp.body.continue:
981 // CHECK1-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
982 // CHECK1: omp.inner.for.inc:
983 // CHECK1-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
984 // CHECK1-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP10]], 1
985 // CHECK1-NEXT: store i32 [[ADD3]], i32* [[DOTOMP_IV]], align 4
986 // CHECK1-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP22:![0-9]+]]
987 // CHECK1: omp.inner.for.end:
988 // CHECK1-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
989 // CHECK1: omp.loop.exit:
990 // CHECK1-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP3]])
991 // CHECK1-NEXT: [[TMP11:%.*]] = load i32, i32* [[DOTOMP_IS_LAST]], align 4
992 // CHECK1-NEXT: [[TMP12:%.*]] = icmp ne i32 [[TMP11]], 0
993 // CHECK1-NEXT: br i1 [[TMP12]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
994 // CHECK1: .omp.final.then:
995 // CHECK1-NEXT: store i32 100, i32* [[I]], align 4
996 // CHECK1-NEXT: br label [[DOTOMP_FINAL_DONE]]
997 // CHECK1: .omp.final.done:
998 // CHECK1-NEXT: ret void
999 //
1000 //
1001 // CHECK1-LABEL: define {{[^@]+}}@_Z5tmainIiEiT_
1002 // CHECK1-SAME: (i32 [[ARG:%.*]]) #[[ATTR0]] comdat {
1003 // CHECK1-NEXT: entry:
1004 // CHECK1-NEXT: [[ARG_ADDR:%.*]] = alloca i32, align 4
1005 // CHECK1-NEXT: [[TMP:%.*]] = alloca i32, align 4
1006 // CHECK1-NEXT: [[_TMP1:%.*]] = alloca i32, align 4
1007 // CHECK1-NEXT: [[ARG_CASTED:%.*]] = alloca i64, align 8
1008 // CHECK1-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [1 x i8*], align 8
1009 // CHECK1-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [1 x i8*], align 8
1010 // CHECK1-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [1 x i8*], align 8
1011 // CHECK1-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i8, align 1
1012 // CHECK1-NEXT: [[_TMP5:%.*]] = alloca i32, align 4
1013 // CHECK1-NEXT: store i32 [[ARG]], i32* [[ARG_ADDR]], align 4
1014 // CHECK1-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i64 100)
1015 // CHECK1-NEXT: [[TMP0:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiEiT__l57.region_id, i32 0, i8** null, i8** null, i64* null, i64* null, i8** null, i8** null, i32 0, i32 0)
1016 // CHECK1-NEXT: [[TMP1:%.*]] = icmp ne i32 [[TMP0]], 0
1017 // CHECK1-NEXT: br i1 [[TMP1]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
1018 // CHECK1: omp_offload.failed:
1019 // CHECK1-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiEiT__l57() #[[ATTR2]]
1020 // CHECK1-NEXT: br label [[OMP_OFFLOAD_CONT]]
1021 // CHECK1: omp_offload.cont:
1022 // CHECK1-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i64 100)
1023 // CHECK1-NEXT: [[TMP2:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiEiT__l62.region_id, i32 0, i8** null, i8** null, i64* null, i64* null, i8** null, i8** null, i32 0, i32 1)
1024 // CHECK1-NEXT: [[TMP3:%.*]] = icmp ne i32 [[TMP2]], 0
1025 // CHECK1-NEXT: br i1 [[TMP3]], label [[OMP_OFFLOAD_FAILED2:%.*]], label [[OMP_OFFLOAD_CONT3:%.*]]
1026 // CHECK1: omp_offload.failed2:
1027 // CHECK1-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiEiT__l62() #[[ATTR2]]
1028 // CHECK1-NEXT: br label [[OMP_OFFLOAD_CONT3]]
1029 // CHECK1: omp_offload.cont3:
1030 // CHECK1-NEXT: [[TMP4:%.*]] = load i32, i32* [[ARG_ADDR]], align 4
1031 // CHECK1-NEXT: [[CONV:%.*]] = bitcast i64* [[ARG_CASTED]] to i32*
1032 // CHECK1-NEXT: store i32 [[TMP4]], i32* [[CONV]], align 4
1033 // CHECK1-NEXT: [[TMP5:%.*]] = load i64, i64* [[ARG_CASTED]], align 8
1034 // CHECK1-NEXT: [[TMP6:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
1035 // CHECK1-NEXT: [[TMP7:%.*]] = bitcast i8** [[TMP6]] to i64*
1036 // CHECK1-NEXT: store i64 [[TMP5]], i64* [[TMP7]], align 8
1037 // CHECK1-NEXT: [[TMP8:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
1038 // CHECK1-NEXT: [[TMP9:%.*]] = bitcast i8** [[TMP8]] to i64*
1039 // CHECK1-NEXT: store i64 [[TMP5]], i64* [[TMP9]], align 8
1040 // CHECK1-NEXT: [[TMP10:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0
1041 // CHECK1-NEXT: store i8* null, i8** [[TMP10]], align 8
1042 // CHECK1-NEXT: [[TMP11:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
1043 // CHECK1-NEXT: [[TMP12:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
1044 // CHECK1-NEXT: [[TMP13:%.*]] = load i32, i32* [[ARG_ADDR]], align 4
1045 // CHECK1-NEXT: [[TOBOOL:%.*]] = icmp ne i32 [[TMP13]], 0
1046 // CHECK1-NEXT: [[FROMBOOL:%.*]] = zext i1 [[TOBOOL]] to i8
1047 // CHECK1-NEXT: store i8 [[FROMBOOL]], i8* [[DOTCAPTURE_EXPR_]], align 1
1048 // CHECK1-NEXT: [[TMP14:%.*]] = load i8, i8* [[DOTCAPTURE_EXPR_]], align 1
1049 // CHECK1-NEXT: [[TOBOOL4:%.*]] = trunc i8 [[TMP14]] to i1
1050 // CHECK1-NEXT: [[TMP15:%.*]] = select i1 [[TOBOOL4]], i32 0, i32 1
1051 // CHECK1-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i64 100)
1052 // CHECK1-NEXT: [[TMP16:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiEiT__l67.region_id, i32 1, i8** [[TMP11]], i8** [[TMP12]], i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_sizes.16, i32 0, i32 0), i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_maptypes.17, i32 0, i32 0), i8** null, i8** null, i32 0, i32 [[TMP15]])
1053 // CHECK1-NEXT: [[TMP17:%.*]] = icmp ne i32 [[TMP16]], 0
1054 // CHECK1-NEXT: br i1 [[TMP17]], label [[OMP_OFFLOAD_FAILED6:%.*]], label [[OMP_OFFLOAD_CONT7:%.*]]
1055 // CHECK1: omp_offload.failed6:
1056 // CHECK1-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiEiT__l67(i64 [[TMP5]]) #[[ATTR2]]
1057 // CHECK1-NEXT: br label [[OMP_OFFLOAD_CONT7]]
1058 // CHECK1: omp_offload.cont7:
1059 // CHECK1-NEXT: ret i32 0
1060 //
1061 //
1062 // CHECK1-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiEiT__l57
1063 // CHECK1-SAME: () #[[ATTR1]] {
1064 // CHECK1-NEXT: entry:
1065 // CHECK1-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 0, void (i32*, i32*, ...)* bitcast (void (i32*, i32*)* @.omp_outlined..10 to void (i32*, i32*, ...)*))
1066 // CHECK1-NEXT: ret void
1067 //
1068 //
1069 // CHECK1-LABEL: define {{[^@]+}}@.omp_outlined..10
1070 // CHECK1-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]]) #[[ATTR1]] {
1071 // CHECK1-NEXT: entry:
1072 // CHECK1-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
1073 // CHECK1-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
1074 // CHECK1-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
1075 // CHECK1-NEXT: [[TMP:%.*]] = alloca i32, align 4
1076 // CHECK1-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4
1077 // CHECK1-NEXT: [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4
1078 // CHECK1-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
1079 // CHECK1-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
1080 // CHECK1-NEXT: [[I:%.*]] = alloca i32, align 4
1081 // CHECK1-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
1082 // CHECK1-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
1083 // CHECK1-NEXT: store i32 0, i32* [[DOTOMP_COMB_LB]], align 4
1084 // CHECK1-NEXT: store i32 99, i32* [[DOTOMP_COMB_UB]], align 4
1085 // CHECK1-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4
1086 // CHECK1-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4
1087 // CHECK1-NEXT: [[TMP0:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
1088 // CHECK1-NEXT: [[TMP1:%.*]] = load i32, i32* [[TMP0]], align 4
1089 // CHECK1-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB1]], i32 [[TMP1]], i32 92, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_COMB_LB]], i32* [[DOTOMP_COMB_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1)
1090 // CHECK1-NEXT: [[TMP2:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
1091 // CHECK1-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP2]], 99
1092 // CHECK1-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
1093 // CHECK1: cond.true:
1094 // CHECK1-NEXT: br label [[COND_END:%.*]]
1095 // CHECK1: cond.false:
1096 // CHECK1-NEXT: [[TMP3:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
1097 // CHECK1-NEXT: br label [[COND_END]]
1098 // CHECK1: cond.end:
1099 // CHECK1-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP3]], [[COND_FALSE]] ]
1100 // CHECK1-NEXT: store i32 [[COND]], i32* [[DOTOMP_COMB_UB]], align 4
1101 // CHECK1-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4
1102 // CHECK1-NEXT: store i32 [[TMP4]], i32* [[DOTOMP_IV]], align 4
1103 // CHECK1-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
1104 // CHECK1: omp.inner.for.cond:
1105 // CHECK1-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
1106 // CHECK1-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
1107 // CHECK1-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP5]], [[TMP6]]
1108 // CHECK1-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
1109 // CHECK1: omp.inner.for.body:
1110 // CHECK1-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4
1111 // CHECK1-NEXT: [[TMP8:%.*]] = zext i32 [[TMP7]] to i64
1112 // CHECK1-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
1113 // CHECK1-NEXT: [[TMP10:%.*]] = zext i32 [[TMP9]] to i64
1114 // CHECK1-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB3]], i32 2, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64, i64)* @.omp_outlined..11 to void (i32*, i32*, ...)*), i64 [[TMP8]], i64 [[TMP10]])
1115 // CHECK1-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
1116 // CHECK1: omp.inner.for.inc:
1117 // CHECK1-NEXT: [[TMP11:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
1118 // CHECK1-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4
1119 // CHECK1-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP11]], [[TMP12]]
1120 // CHECK1-NEXT: store i32 [[ADD]], i32* [[DOTOMP_IV]], align 4
1121 // CHECK1-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP23:![0-9]+]]
1122 // CHECK1: omp.inner.for.end:
1123 // CHECK1-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
1124 // CHECK1: omp.loop.exit:
1125 // CHECK1-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP1]])
1126 // CHECK1-NEXT: [[TMP13:%.*]] = load i32, i32* [[DOTOMP_IS_LAST]], align 4
1127 // CHECK1-NEXT: [[TMP14:%.*]] = icmp ne i32 [[TMP13]], 0
1128 // CHECK1-NEXT: br i1 [[TMP14]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
1129 // CHECK1: .omp.final.then:
1130 // CHECK1-NEXT: store i32 100, i32* [[I]], align 4
1131 // CHECK1-NEXT: br label [[DOTOMP_FINAL_DONE]]
1132 // CHECK1: .omp.final.done:
1133 // CHECK1-NEXT: ret void
1134 //
1135 //
1136 // CHECK1-LABEL: define {{[^@]+}}@.omp_outlined..11
1137 // CHECK1-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[DOTPREVIOUS_LB_:%.*]], i64 [[DOTPREVIOUS_UB_:%.*]]) #[[ATTR1]] {
1138 // CHECK1-NEXT: entry:
1139 // CHECK1-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
1140 // CHECK1-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
1141 // CHECK1-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i64, align 8
1142 // CHECK1-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i64, align 8
1143 // CHECK1-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
1144 // CHECK1-NEXT: [[TMP:%.*]] = alloca i32, align 4
1145 // CHECK1-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4
1146 // CHECK1-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4
1147 // CHECK1-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
1148 // CHECK1-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
1149 // CHECK1-NEXT: [[I:%.*]] = alloca i32, align 4
1150 // CHECK1-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
1151 // CHECK1-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
1152 // CHECK1-NEXT: store i64 [[DOTPREVIOUS_LB_]], i64* [[DOTPREVIOUS_LB__ADDR]], align 8
1153 // CHECK1-NEXT: store i64 [[DOTPREVIOUS_UB_]], i64* [[DOTPREVIOUS_UB__ADDR]], align 8
1154 // CHECK1-NEXT: store i32 0, i32* [[DOTOMP_LB]], align 4
1155 // CHECK1-NEXT: store i32 99, i32* [[DOTOMP_UB]], align 4
1156 // CHECK1-NEXT: [[TMP0:%.*]] = load i64, i64* [[DOTPREVIOUS_LB__ADDR]], align 8
1157 // CHECK1-NEXT: [[CONV:%.*]] = trunc i64 [[TMP0]] to i32
1158 // CHECK1-NEXT: [[TMP1:%.*]] = load i64, i64* [[DOTPREVIOUS_UB__ADDR]], align 8
1159 // CHECK1-NEXT: [[CONV1:%.*]] = trunc i64 [[TMP1]] to i32
1160 // CHECK1-NEXT: store i32 [[CONV]], i32* [[DOTOMP_LB]], align 4
1161 // CHECK1-NEXT: store i32 [[CONV1]], i32* [[DOTOMP_UB]], align 4
1162 // CHECK1-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4
1163 // CHECK1-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4
1164 // CHECK1-NEXT: [[TMP2:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
1165 // CHECK1-NEXT: [[TMP3:%.*]] = load i32, i32* [[TMP2]], align 4
1166 // CHECK1-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB2]], i32 [[TMP3]], i32 34, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_LB]], i32* [[DOTOMP_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1)
1167 // CHECK1-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
1168 // CHECK1-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP4]], 99
1169 // CHECK1-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
1170 // CHECK1: cond.true:
1171 // CHECK1-NEXT: br label [[COND_END:%.*]]
1172 // CHECK1: cond.false:
1173 // CHECK1-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
1174 // CHECK1-NEXT: br label [[COND_END]]
1175 // CHECK1: cond.end:
1176 // CHECK1-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP5]], [[COND_FALSE]] ]
1177 // CHECK1-NEXT: store i32 [[COND]], i32* [[DOTOMP_UB]], align 4
1178 // CHECK1-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4
1179 // CHECK1-NEXT: store i32 [[TMP6]], i32* [[DOTOMP_IV]], align 4
1180 // CHECK1-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
1181 // CHECK1: omp.inner.for.cond:
1182 // CHECK1-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
1183 // CHECK1-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
1184 // CHECK1-NEXT: [[CMP2:%.*]] = icmp sle i32 [[TMP7]], [[TMP8]]
1185 // CHECK1-NEXT: br i1 [[CMP2]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
1186 // CHECK1: omp.inner.for.body:
1187 // CHECK1-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
1188 // CHECK1-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP9]], 1
1189 // CHECK1-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
1190 // CHECK1-NEXT: store i32 [[ADD]], i32* [[I]], align 4
1191 // CHECK1-NEXT: call void @_Z3fn1v()
1192 // CHECK1-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
1193 // CHECK1: omp.body.continue:
1194 // CHECK1-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
1195 // CHECK1: omp.inner.for.inc:
1196 // CHECK1-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
1197 // CHECK1-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP10]], 1
1198 // CHECK1-NEXT: store i32 [[ADD3]], i32* [[DOTOMP_IV]], align 4
1199 // CHECK1-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP24:![0-9]+]]
1200 // CHECK1: omp.inner.for.end:
1201 // CHECK1-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
1202 // CHECK1: omp.loop.exit:
1203 // CHECK1-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP3]])
1204 // CHECK1-NEXT: [[TMP11:%.*]] = load i32, i32* [[DOTOMP_IS_LAST]], align 4
1205 // CHECK1-NEXT: [[TMP12:%.*]] = icmp ne i32 [[TMP11]], 0
1206 // CHECK1-NEXT: br i1 [[TMP12]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
1207 // CHECK1: .omp.final.then:
1208 // CHECK1-NEXT: store i32 100, i32* [[I]], align 4
1209 // CHECK1-NEXT: br label [[DOTOMP_FINAL_DONE]]
1210 // CHECK1: .omp.final.done:
1211 // CHECK1-NEXT: ret void
1212 //
1213 //
1214 // CHECK1-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiEiT__l62
1215 // CHECK1-SAME: () #[[ATTR1]] {
1216 // CHECK1-NEXT: entry:
1217 // CHECK1-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 0, void (i32*, i32*, ...)* bitcast (void (i32*, i32*)* @.omp_outlined..12 to void (i32*, i32*, ...)*))
1218 // CHECK1-NEXT: ret void
1219 //
1220 //
1221 // CHECK1-LABEL: define {{[^@]+}}@.omp_outlined..12
1222 // CHECK1-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]]) #[[ATTR1]] {
1223 // CHECK1-NEXT: entry:
1224 // CHECK1-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
1225 // CHECK1-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
1226 // CHECK1-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
1227 // CHECK1-NEXT: [[TMP:%.*]] = alloca i32, align 4
1228 // CHECK1-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4
1229 // CHECK1-NEXT: [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4
1230 // CHECK1-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
1231 // CHECK1-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
1232 // CHECK1-NEXT: [[I:%.*]] = alloca i32, align 4
1233 // CHECK1-NEXT: [[DOTBOUND_ZERO_ADDR:%.*]] = alloca i32, align 4
1234 // CHECK1-NEXT: store i32 0, i32* [[DOTBOUND_ZERO_ADDR]], align 4
1235 // CHECK1-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
1236 // CHECK1-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
1237 // CHECK1-NEXT: store i32 0, i32* [[DOTOMP_COMB_LB]], align 4
1238 // CHECK1-NEXT: store i32 99, i32* [[DOTOMP_COMB_UB]], align 4
1239 // CHECK1-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4
1240 // CHECK1-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4
1241 // CHECK1-NEXT: [[TMP0:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
1242 // CHECK1-NEXT: [[TMP1:%.*]] = load i32, i32* [[TMP0]], align 4
1243 // CHECK1-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB1]], i32 [[TMP1]], i32 92, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_COMB_LB]], i32* [[DOTOMP_COMB_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1)
1244 // CHECK1-NEXT: [[TMP2:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
1245 // CHECK1-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP2]], 99
1246 // CHECK1-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
1247 // CHECK1: cond.true:
1248 // CHECK1-NEXT: br label [[COND_END:%.*]]
1249 // CHECK1: cond.false:
1250 // CHECK1-NEXT: [[TMP3:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
1251 // CHECK1-NEXT: br label [[COND_END]]
1252 // CHECK1: cond.end:
1253 // CHECK1-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP3]], [[COND_FALSE]] ]
1254 // CHECK1-NEXT: store i32 [[COND]], i32* [[DOTOMP_COMB_UB]], align 4
1255 // CHECK1-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4
1256 // CHECK1-NEXT: store i32 [[TMP4]], i32* [[DOTOMP_IV]], align 4
1257 // CHECK1-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
1258 // CHECK1: omp.inner.for.cond:
1259 // CHECK1-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
1260 // CHECK1-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
1261 // CHECK1-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP5]], [[TMP6]]
1262 // CHECK1-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
1263 // CHECK1: omp.inner.for.body:
1264 // CHECK1-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4
1265 // CHECK1-NEXT: [[TMP8:%.*]] = zext i32 [[TMP7]] to i64
1266 // CHECK1-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
1267 // CHECK1-NEXT: [[TMP10:%.*]] = zext i32 [[TMP9]] to i64
1268 // CHECK1-NEXT: call void @__kmpc_serialized_parallel(%struct.ident_t* @[[GLOB3]], i32 [[TMP1]])
1269 // CHECK1-NEXT: [[TMP11:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
1270 // CHECK1-NEXT: call void @.omp_outlined..13(i32* [[TMP11]], i32* [[DOTBOUND_ZERO_ADDR]], i64 [[TMP8]], i64 [[TMP10]]) #[[ATTR2]]
1271 // CHECK1-NEXT: call void @__kmpc_end_serialized_parallel(%struct.ident_t* @[[GLOB3]], i32 [[TMP1]])
1272 // CHECK1-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
1273 // CHECK1: omp.inner.for.inc:
1274 // CHECK1-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
1275 // CHECK1-NEXT: [[TMP13:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4
1276 // CHECK1-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP12]], [[TMP13]]
1277 // CHECK1-NEXT: store i32 [[ADD]], i32* [[DOTOMP_IV]], align 4
1278 // CHECK1-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP25:![0-9]+]]
1279 // CHECK1: omp.inner.for.end:
1280 // CHECK1-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
1281 // CHECK1: omp.loop.exit:
1282 // CHECK1-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP1]])
1283 // CHECK1-NEXT: [[TMP14:%.*]] = load i32, i32* [[DOTOMP_IS_LAST]], align 4
1284 // CHECK1-NEXT: [[TMP15:%.*]] = icmp ne i32 [[TMP14]], 0
1285 // CHECK1-NEXT: br i1 [[TMP15]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
1286 // CHECK1: .omp.final.then:
1287 // CHECK1-NEXT: store i32 100, i32* [[I]], align 4
1288 // CHECK1-NEXT: br label [[DOTOMP_FINAL_DONE]]
1289 // CHECK1: .omp.final.done:
1290 // CHECK1-NEXT: ret void
1291 //
1292 //
1293 // CHECK1-LABEL: define {{[^@]+}}@.omp_outlined..13
1294 // CHECK1-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[DOTPREVIOUS_LB_:%.*]], i64 [[DOTPREVIOUS_UB_:%.*]]) #[[ATTR1]] {
1295 // CHECK1-NEXT: entry:
1296 // CHECK1-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
1297 // CHECK1-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
1298 // CHECK1-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i64, align 8
1299 // CHECK1-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i64, align 8
1300 // CHECK1-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
1301 // CHECK1-NEXT: [[TMP:%.*]] = alloca i32, align 4
1302 // CHECK1-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4
1303 // CHECK1-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4
1304 // CHECK1-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
1305 // CHECK1-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
1306 // CHECK1-NEXT: [[I:%.*]] = alloca i32, align 4
1307 // CHECK1-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
1308 // CHECK1-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
1309 // CHECK1-NEXT: store i64 [[DOTPREVIOUS_LB_]], i64* [[DOTPREVIOUS_LB__ADDR]], align 8
1310 // CHECK1-NEXT: store i64 [[DOTPREVIOUS_UB_]], i64* [[DOTPREVIOUS_UB__ADDR]], align 8
1311 // CHECK1-NEXT: store i32 0, i32* [[DOTOMP_LB]], align 4
1312 // CHECK1-NEXT: store i32 99, i32* [[DOTOMP_UB]], align 4
1313 // CHECK1-NEXT: [[TMP0:%.*]] = load i64, i64* [[DOTPREVIOUS_LB__ADDR]], align 8
1314 // CHECK1-NEXT: [[CONV:%.*]] = trunc i64 [[TMP0]] to i32
1315 // CHECK1-NEXT: [[TMP1:%.*]] = load i64, i64* [[DOTPREVIOUS_UB__ADDR]], align 8
1316 // CHECK1-NEXT: [[CONV1:%.*]] = trunc i64 [[TMP1]] to i32
1317 // CHECK1-NEXT: store i32 [[CONV]], i32* [[DOTOMP_LB]], align 4
1318 // CHECK1-NEXT: store i32 [[CONV1]], i32* [[DOTOMP_UB]], align 4
1319 // CHECK1-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4
1320 // CHECK1-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4
1321 // CHECK1-NEXT: [[TMP2:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
1322 // CHECK1-NEXT: [[TMP3:%.*]] = load i32, i32* [[TMP2]], align 4
1323 // CHECK1-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB2]], i32 [[TMP3]], i32 34, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_LB]], i32* [[DOTOMP_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1)
1324 // CHECK1-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
1325 // CHECK1-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP4]], 99
1326 // CHECK1-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
1327 // CHECK1: cond.true:
1328 // CHECK1-NEXT: br label [[COND_END:%.*]]
1329 // CHECK1: cond.false:
1330 // CHECK1-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
1331 // CHECK1-NEXT: br label [[COND_END]]
1332 // CHECK1: cond.end:
1333 // CHECK1-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP5]], [[COND_FALSE]] ]
1334 // CHECK1-NEXT: store i32 [[COND]], i32* [[DOTOMP_UB]], align 4
1335 // CHECK1-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4
1336 // CHECK1-NEXT: store i32 [[TMP6]], i32* [[DOTOMP_IV]], align 4
1337 // CHECK1-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
1338 // CHECK1: omp.inner.for.cond:
1339 // CHECK1-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
1340 // CHECK1-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
1341 // CHECK1-NEXT: [[CMP2:%.*]] = icmp sle i32 [[TMP7]], [[TMP8]]
1342 // CHECK1-NEXT: br i1 [[CMP2]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
1343 // CHECK1: omp.inner.for.body:
1344 // CHECK1-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
1345 // CHECK1-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP9]], 1
1346 // CHECK1-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
1347 // CHECK1-NEXT: store i32 [[ADD]], i32* [[I]], align 4
1348 // CHECK1-NEXT: call void @_Z3fn2v()
1349 // CHECK1-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
1350 // CHECK1: omp.body.continue:
1351 // CHECK1-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
1352 // CHECK1: omp.inner.for.inc:
1353 // CHECK1-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
1354 // CHECK1-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP10]], 1
1355 // CHECK1-NEXT: store i32 [[ADD3]], i32* [[DOTOMP_IV]], align 4
1356 // CHECK1-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP26:![0-9]+]]
1357 // CHECK1: omp.inner.for.end:
1358 // CHECK1-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
1359 // CHECK1: omp.loop.exit:
1360 // CHECK1-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP3]])
1361 // CHECK1-NEXT: [[TMP11:%.*]] = load i32, i32* [[DOTOMP_IS_LAST]], align 4
1362 // CHECK1-NEXT: [[TMP12:%.*]] = icmp ne i32 [[TMP11]], 0
1363 // CHECK1-NEXT: br i1 [[TMP12]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
1364 // CHECK1: .omp.final.then:
1365 // CHECK1-NEXT: store i32 100, i32* [[I]], align 4
1366 // CHECK1-NEXT: br label [[DOTOMP_FINAL_DONE]]
1367 // CHECK1: .omp.final.done:
1368 // CHECK1-NEXT: ret void
1369 //
1370 //
1371 // CHECK1-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiEiT__l67
1372 // CHECK1-SAME: (i64 [[ARG:%.*]]) #[[ATTR1]] {
1373 // CHECK1-NEXT: entry:
1374 // CHECK1-NEXT: [[ARG_ADDR:%.*]] = alloca i64, align 8
1375 // CHECK1-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i8, align 1
1376 // CHECK1-NEXT: [[DOTCAPTURE_EXPR__CASTED:%.*]] = alloca i64, align 8
1377 // CHECK1-NEXT: store i64 [[ARG]], i64* [[ARG_ADDR]], align 8
1378 // CHECK1-NEXT: [[CONV:%.*]] = bitcast i64* [[ARG_ADDR]] to i32*
1379 // CHECK1-NEXT: [[TMP0:%.*]] = load i32, i32* [[CONV]], align 8
1380 // CHECK1-NEXT: [[TOBOOL:%.*]] = icmp ne i32 [[TMP0]], 0
1381 // CHECK1-NEXT: [[FROMBOOL:%.*]] = zext i1 [[TOBOOL]] to i8
1382 // CHECK1-NEXT: store i8 [[FROMBOOL]], i8* [[DOTCAPTURE_EXPR_]], align 1
1383 // CHECK1-NEXT: [[TMP1:%.*]] = load i8, i8* [[DOTCAPTURE_EXPR_]], align 1
1384 // CHECK1-NEXT: [[TOBOOL1:%.*]] = trunc i8 [[TMP1]] to i1
1385 // CHECK1-NEXT: [[CONV2:%.*]] = bitcast i64* [[DOTCAPTURE_EXPR__CASTED]] to i8*
1386 // CHECK1-NEXT: [[FROMBOOL3:%.*]] = zext i1 [[TOBOOL1]] to i8
1387 // CHECK1-NEXT: store i8 [[FROMBOOL3]], i8* [[CONV2]], align 1
1388 // CHECK1-NEXT: [[TMP2:%.*]] = load i64, i64* [[DOTCAPTURE_EXPR__CASTED]], align 8
1389 // CHECK1-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64)* @.omp_outlined..14 to void (i32*, i32*, ...)*), i64 [[TMP2]])
1390 // CHECK1-NEXT: ret void
1391 //
1392 //
1393 // CHECK1-LABEL: define {{[^@]+}}@.omp_outlined..14
1394 // CHECK1-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[DOTCAPTURE_EXPR_:%.*]]) #[[ATTR1]] {
1395 // CHECK1-NEXT: entry:
1396 // CHECK1-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
1397 // CHECK1-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
1398 // CHECK1-NEXT: [[DOTCAPTURE_EXPR__ADDR:%.*]] = alloca i64, align 8
1399 // CHECK1-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
1400 // CHECK1-NEXT: [[TMP:%.*]] = alloca i32, align 4
1401 // CHECK1-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4
1402 // CHECK1-NEXT: [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4
1403 // CHECK1-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
1404 // CHECK1-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
1405 // CHECK1-NEXT: [[I:%.*]] = alloca i32, align 4
1406 // CHECK1-NEXT: [[DOTBOUND_ZERO_ADDR:%.*]] = alloca i32, align 4
1407 // CHECK1-NEXT: store i32 0, i32* [[DOTBOUND_ZERO_ADDR]], align 4
1408 // CHECK1-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
1409 // CHECK1-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
1410 // CHECK1-NEXT: store i64 [[DOTCAPTURE_EXPR_]], i64* [[DOTCAPTURE_EXPR__ADDR]], align 8
1411 // CHECK1-NEXT: [[CONV:%.*]] = bitcast i64* [[DOTCAPTURE_EXPR__ADDR]] to i8*
1412 // CHECK1-NEXT: store i32 0, i32* [[DOTOMP_COMB_LB]], align 4
1413 // CHECK1-NEXT: store i32 99, i32* [[DOTOMP_COMB_UB]], align 4
1414 // CHECK1-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4
1415 // CHECK1-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4
1416 // CHECK1-NEXT: [[TMP0:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
1417 // CHECK1-NEXT: [[TMP1:%.*]] = load i32, i32* [[TMP0]], align 4
1418 // CHECK1-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB1]], i32 [[TMP1]], i32 92, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_COMB_LB]], i32* [[DOTOMP_COMB_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1)
1419 // CHECK1-NEXT: [[TMP2:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
1420 // CHECK1-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP2]], 99
1421 // CHECK1-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
1422 // CHECK1: cond.true:
1423 // CHECK1-NEXT: br label [[COND_END:%.*]]
1424 // CHECK1: cond.false:
1425 // CHECK1-NEXT: [[TMP3:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
1426 // CHECK1-NEXT: br label [[COND_END]]
1427 // CHECK1: cond.end:
1428 // CHECK1-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP3]], [[COND_FALSE]] ]
1429 // CHECK1-NEXT: store i32 [[COND]], i32* [[DOTOMP_COMB_UB]], align 4
1430 // CHECK1-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4
1431 // CHECK1-NEXT: store i32 [[TMP4]], i32* [[DOTOMP_IV]], align 4
1432 // CHECK1-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
1433 // CHECK1: omp.inner.for.cond:
1434 // CHECK1-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
1435 // CHECK1-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
1436 // CHECK1-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP5]], [[TMP6]]
1437 // CHECK1-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
1438 // CHECK1: omp.inner.for.body:
1439 // CHECK1-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4
1440 // CHECK1-NEXT: [[TMP8:%.*]] = zext i32 [[TMP7]] to i64
1441 // CHECK1-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
1442 // CHECK1-NEXT: [[TMP10:%.*]] = zext i32 [[TMP9]] to i64
1443 // CHECK1-NEXT: [[TMP11:%.*]] = load i8, i8* [[CONV]], align 8
1444 // CHECK1-NEXT: [[TOBOOL:%.*]] = trunc i8 [[TMP11]] to i1
1445 // CHECK1-NEXT: br i1 [[TOBOOL]], label [[OMP_IF_THEN:%.*]], label [[OMP_IF_ELSE:%.*]]
1446 // CHECK1: omp_if.then:
1447 // CHECK1-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB3]], i32 2, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64, i64)* @.omp_outlined..15 to void (i32*, i32*, ...)*), i64 [[TMP8]], i64 [[TMP10]])
1448 // CHECK1-NEXT: br label [[OMP_IF_END:%.*]]
1449 // CHECK1: omp_if.else:
1450 // CHECK1-NEXT: call void @__kmpc_serialized_parallel(%struct.ident_t* @[[GLOB3]], i32 [[TMP1]])
1451 // CHECK1-NEXT: [[TMP12:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
1452 // CHECK1-NEXT: call void @.omp_outlined..15(i32* [[TMP12]], i32* [[DOTBOUND_ZERO_ADDR]], i64 [[TMP8]], i64 [[TMP10]]) #[[ATTR2]]
1453 // CHECK1-NEXT: call void @__kmpc_end_serialized_parallel(%struct.ident_t* @[[GLOB3]], i32 [[TMP1]])
1454 // CHECK1-NEXT: br label [[OMP_IF_END]]
1455 // CHECK1: omp_if.end:
1456 // CHECK1-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
1457 // CHECK1: omp.inner.for.inc:
1458 // CHECK1-NEXT: [[TMP13:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
1459 // CHECK1-NEXT: [[TMP14:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4
1460 // CHECK1-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP13]], [[TMP14]]
1461 // CHECK1-NEXT: store i32 [[ADD]], i32* [[DOTOMP_IV]], align 4
1462 // CHECK1-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP27:![0-9]+]]
1463 // CHECK1: omp.inner.for.end:
1464 // CHECK1-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
1465 // CHECK1: omp.loop.exit:
1466 // CHECK1-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP1]])
1467 // CHECK1-NEXT: [[TMP15:%.*]] = load i32, i32* [[DOTOMP_IS_LAST]], align 4
1468 // CHECK1-NEXT: [[TMP16:%.*]] = icmp ne i32 [[TMP15]], 0
1469 // CHECK1-NEXT: br i1 [[TMP16]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
1470 // CHECK1: .omp.final.then:
1471 // CHECK1-NEXT: store i32 100, i32* [[I]], align 4
1472 // CHECK1-NEXT: br label [[DOTOMP_FINAL_DONE]]
1473 // CHECK1: .omp.final.done:
1474 // CHECK1-NEXT: ret void
1475 //
1476 //
1477 // CHECK1-LABEL: define {{[^@]+}}@.omp_outlined..15
1478 // CHECK1-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[DOTPREVIOUS_LB_:%.*]], i64 [[DOTPREVIOUS_UB_:%.*]]) #[[ATTR1]] {
1479 // CHECK1-NEXT: entry:
1480 // CHECK1-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
1481 // CHECK1-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
1482 // CHECK1-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i64, align 8
1483 // CHECK1-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i64, align 8
1484 // CHECK1-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
1485 // CHECK1-NEXT: [[TMP:%.*]] = alloca i32, align 4
1486 // CHECK1-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4
1487 // CHECK1-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4
1488 // CHECK1-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
1489 // CHECK1-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
1490 // CHECK1-NEXT: [[I:%.*]] = alloca i32, align 4
1491 // CHECK1-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
1492 // CHECK1-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
1493 // CHECK1-NEXT: store i64 [[DOTPREVIOUS_LB_]], i64* [[DOTPREVIOUS_LB__ADDR]], align 8
1494 // CHECK1-NEXT: store i64 [[DOTPREVIOUS_UB_]], i64* [[DOTPREVIOUS_UB__ADDR]], align 8
1495 // CHECK1-NEXT: store i32 0, i32* [[DOTOMP_LB]], align 4
1496 // CHECK1-NEXT: store i32 99, i32* [[DOTOMP_UB]], align 4
1497 // CHECK1-NEXT: [[TMP0:%.*]] = load i64, i64* [[DOTPREVIOUS_LB__ADDR]], align 8
1498 // CHECK1-NEXT: [[CONV:%.*]] = trunc i64 [[TMP0]] to i32
1499 // CHECK1-NEXT: [[TMP1:%.*]] = load i64, i64* [[DOTPREVIOUS_UB__ADDR]], align 8
1500 // CHECK1-NEXT: [[CONV1:%.*]] = trunc i64 [[TMP1]] to i32
1501 // CHECK1-NEXT: store i32 [[CONV]], i32* [[DOTOMP_LB]], align 4
1502 // CHECK1-NEXT: store i32 [[CONV1]], i32* [[DOTOMP_UB]], align 4
1503 // CHECK1-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4
1504 // CHECK1-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4
1505 // CHECK1-NEXT: [[TMP2:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
1506 // CHECK1-NEXT: [[TMP3:%.*]] = load i32, i32* [[TMP2]], align 4
1507 // CHECK1-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB2]], i32 [[TMP3]], i32 34, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_LB]], i32* [[DOTOMP_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1)
1508 // CHECK1-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
1509 // CHECK1-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP4]], 99
1510 // CHECK1-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
1511 // CHECK1: cond.true:
1512 // CHECK1-NEXT: br label [[COND_END:%.*]]
1513 // CHECK1: cond.false:
1514 // CHECK1-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
1515 // CHECK1-NEXT: br label [[COND_END]]
1516 // CHECK1: cond.end:
1517 // CHECK1-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP5]], [[COND_FALSE]] ]
1518 // CHECK1-NEXT: store i32 [[COND]], i32* [[DOTOMP_UB]], align 4
1519 // CHECK1-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4
1520 // CHECK1-NEXT: store i32 [[TMP6]], i32* [[DOTOMP_IV]], align 4
1521 // CHECK1-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
1522 // CHECK1: omp.inner.for.cond:
1523 // CHECK1-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
1524 // CHECK1-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
1525 // CHECK1-NEXT: [[CMP2:%.*]] = icmp sle i32 [[TMP7]], [[TMP8]]
1526 // CHECK1-NEXT: br i1 [[CMP2]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
1527 // CHECK1: omp.inner.for.body:
1528 // CHECK1-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
1529 // CHECK1-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP9]], 1
1530 // CHECK1-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
1531 // CHECK1-NEXT: store i32 [[ADD]], i32* [[I]], align 4
1532 // CHECK1-NEXT: call void @_Z3fn3v()
1533 // CHECK1-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
1534 // CHECK1: omp.body.continue:
1535 // CHECK1-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
1536 // CHECK1: omp.inner.for.inc:
1537 // CHECK1-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
1538 // CHECK1-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP10]], 1
1539 // CHECK1-NEXT: store i32 [[ADD3]], i32* [[DOTOMP_IV]], align 4
1540 // CHECK1-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP28:![0-9]+]]
1541 // CHECK1: omp.inner.for.end:
1542 // CHECK1-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
1543 // CHECK1: omp.loop.exit:
1544 // CHECK1-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP3]])
1545 // CHECK1-NEXT: [[TMP11:%.*]] = load i32, i32* [[DOTOMP_IS_LAST]], align 4
1546 // CHECK1-NEXT: [[TMP12:%.*]] = icmp ne i32 [[TMP11]], 0
1547 // CHECK1-NEXT: br i1 [[TMP12]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
1548 // CHECK1: .omp.final.then:
1549 // CHECK1-NEXT: store i32 100, i32* [[I]], align 4
1550 // CHECK1-NEXT: br label [[DOTOMP_FINAL_DONE]]
1551 // CHECK1: .omp.final.done:
1552 // CHECK1-NEXT: ret void
1553 //
1554 //
1555 // CHECK1-LABEL: define {{[^@]+}}@.omp_offloading.requires_reg
1556 // CHECK1-SAME: () #[[ATTR5:[0-9]+]] {
1557 // CHECK1-NEXT: entry:
1558 // CHECK1-NEXT: call void @__tgt_register_requires(i64 1)
1559 // CHECK1-NEXT: ret void
1560 //
1561 //
1562 // CHECK2-LABEL: define {{[^@]+}}@_Z9gtid_testv
1563 // CHECK2-SAME: () #[[ATTR0:[0-9]+]] {
1564 // CHECK2-NEXT: entry:
1565 // CHECK2-NEXT: [[TMP:%.*]] = alloca i32, align 4
1566 // CHECK2-NEXT: [[_TMP1:%.*]] = alloca i32, align 4
1567 // CHECK2-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3:[0-9]+]], i64 -1, i64 100)
1568 // CHECK2-NEXT: [[TMP0:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9gtid_testv_l43.region_id, i32 0, i8** null, i8** null, i64* null, i64* null, i8** null, i8** null, i32 0, i32 0)
1569 // CHECK2-NEXT: [[TMP1:%.*]] = icmp ne i32 [[TMP0]], 0
1570 // CHECK2-NEXT: br i1 [[TMP1]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
1571 // CHECK2: omp_offload.failed:
1572 // CHECK2-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9gtid_testv_l43() #[[ATTR2:[0-9]+]]
1573 // CHECK2-NEXT: br label [[OMP_OFFLOAD_CONT]]
1574 // CHECK2: omp_offload.cont:
1575 // CHECK2-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i64 100)
1576 // CHECK2-NEXT: [[TMP2:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9gtid_testv_l47.region_id, i32 0, i8** null, i8** null, i64* null, i64* null, i8** null, i8** null, i32 0, i32 1)
1577 // CHECK2-NEXT: [[TMP3:%.*]] = icmp ne i32 [[TMP2]], 0
1578 // CHECK2-NEXT: br i1 [[TMP3]], label [[OMP_OFFLOAD_FAILED2:%.*]], label [[OMP_OFFLOAD_CONT3:%.*]]
1579 // CHECK2: omp_offload.failed2:
1580 // CHECK2-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9gtid_testv_l47() #[[ATTR2]]
1581 // CHECK2-NEXT: br label [[OMP_OFFLOAD_CONT3]]
1582 // CHECK2: omp_offload.cont3:
1583 // CHECK2-NEXT: ret void
1584 //
1585 //
1586 // CHECK2-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9gtid_testv_l43
1587 // CHECK2-SAME: () #[[ATTR1:[0-9]+]] {
1588 // CHECK2-NEXT: entry:
1589 // CHECK2-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 0, void (i32*, i32*, ...)* bitcast (void (i32*, i32*)* @.omp_outlined. to void (i32*, i32*, ...)*))
1590 // CHECK2-NEXT: ret void
1591 //
1592 //
1593 // CHECK2-LABEL: define {{[^@]+}}@.omp_outlined.
1594 // CHECK2-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]]) #[[ATTR1]] {
1595 // CHECK2-NEXT: entry:
1596 // CHECK2-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
1597 // CHECK2-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
1598 // CHECK2-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
1599 // CHECK2-NEXT: [[TMP:%.*]] = alloca i32, align 4
1600 // CHECK2-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4
1601 // CHECK2-NEXT: [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4
1602 // CHECK2-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
1603 // CHECK2-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
1604 // CHECK2-NEXT: [[I:%.*]] = alloca i32, align 4
1605 // CHECK2-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
1606 // CHECK2-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
1607 // CHECK2-NEXT: store i32 0, i32* [[DOTOMP_COMB_LB]], align 4
1608 // CHECK2-NEXT: store i32 99, i32* [[DOTOMP_COMB_UB]], align 4
1609 // CHECK2-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4
1610 // CHECK2-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4
1611 // CHECK2-NEXT: [[TMP0:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
1612 // CHECK2-NEXT: [[TMP1:%.*]] = load i32, i32* [[TMP0]], align 4
1613 // CHECK2-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB1:[0-9]+]], i32 [[TMP1]], i32 92, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_COMB_LB]], i32* [[DOTOMP_COMB_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1)
1614 // CHECK2-NEXT: [[TMP2:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
1615 // CHECK2-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP2]], 99
1616 // CHECK2-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
1617 // CHECK2: cond.true:
1618 // CHECK2-NEXT: br label [[COND_END:%.*]]
1619 // CHECK2: cond.false:
1620 // CHECK2-NEXT: [[TMP3:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
1621 // CHECK2-NEXT: br label [[COND_END]]
1622 // CHECK2: cond.end:
1623 // CHECK2-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP3]], [[COND_FALSE]] ]
1624 // CHECK2-NEXT: store i32 [[COND]], i32* [[DOTOMP_COMB_UB]], align 4
1625 // CHECK2-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4
1626 // CHECK2-NEXT: store i32 [[TMP4]], i32* [[DOTOMP_IV]], align 4
1627 // CHECK2-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
1628 // CHECK2: omp.inner.for.cond:
1629 // CHECK2-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
1630 // CHECK2-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
1631 // CHECK2-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP5]], [[TMP6]]
1632 // CHECK2-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
1633 // CHECK2: omp.inner.for.body:
1634 // CHECK2-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4
1635 // CHECK2-NEXT: [[TMP8:%.*]] = zext i32 [[TMP7]] to i64
1636 // CHECK2-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
1637 // CHECK2-NEXT: [[TMP10:%.*]] = zext i32 [[TMP9]] to i64
1638 // CHECK2-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB3]], i32 2, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64, i64)* @.omp_outlined..1 to void (i32*, i32*, ...)*), i64 [[TMP8]], i64 [[TMP10]])
1639 // CHECK2-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
1640 // CHECK2: omp.inner.for.inc:
1641 // CHECK2-NEXT: [[TMP11:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
1642 // CHECK2-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4
1643 // CHECK2-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP11]], [[TMP12]]
1644 // CHECK2-NEXT: store i32 [[ADD]], i32* [[DOTOMP_IV]], align 4
1645 // CHECK2-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP10:![0-9]+]]
1646 // CHECK2: omp.inner.for.end:
1647 // CHECK2-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
1648 // CHECK2: omp.loop.exit:
1649 // CHECK2-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP1]])
1650 // CHECK2-NEXT: [[TMP13:%.*]] = load i32, i32* [[DOTOMP_IS_LAST]], align 4
1651 // CHECK2-NEXT: [[TMP14:%.*]] = icmp ne i32 [[TMP13]], 0
1652 // CHECK2-NEXT: br i1 [[TMP14]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
1653 // CHECK2: .omp.final.then:
1654 // CHECK2-NEXT: store i32 100, i32* [[I]], align 4
1655 // CHECK2-NEXT: br label [[DOTOMP_FINAL_DONE]]
1656 // CHECK2: .omp.final.done:
1657 // CHECK2-NEXT: ret void
1658 //
1659 //
1660 // CHECK2-LABEL: define {{[^@]+}}@.omp_outlined..1
1661 // CHECK2-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[DOTPREVIOUS_LB_:%.*]], i64 [[DOTPREVIOUS_UB_:%.*]]) #[[ATTR1]] {
1662 // CHECK2-NEXT: entry:
1663 // CHECK2-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
1664 // CHECK2-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
1665 // CHECK2-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i64, align 8
1666 // CHECK2-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i64, align 8
1667 // CHECK2-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
1668 // CHECK2-NEXT: [[TMP:%.*]] = alloca i32, align 4
1669 // CHECK2-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4
1670 // CHECK2-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4
1671 // CHECK2-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
1672 // CHECK2-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
1673 // CHECK2-NEXT: [[I:%.*]] = alloca i32, align 4
1674 // CHECK2-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
1675 // CHECK2-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
1676 // CHECK2-NEXT: store i64 [[DOTPREVIOUS_LB_]], i64* [[DOTPREVIOUS_LB__ADDR]], align 8
1677 // CHECK2-NEXT: store i64 [[DOTPREVIOUS_UB_]], i64* [[DOTPREVIOUS_UB__ADDR]], align 8
1678 // CHECK2-NEXT: store i32 0, i32* [[DOTOMP_LB]], align 4
1679 // CHECK2-NEXT: store i32 99, i32* [[DOTOMP_UB]], align 4
1680 // CHECK2-NEXT: [[TMP0:%.*]] = load i64, i64* [[DOTPREVIOUS_LB__ADDR]], align 8
1681 // CHECK2-NEXT: [[CONV:%.*]] = trunc i64 [[TMP0]] to i32
1682 // CHECK2-NEXT: [[TMP1:%.*]] = load i64, i64* [[DOTPREVIOUS_UB__ADDR]], align 8
1683 // CHECK2-NEXT: [[CONV1:%.*]] = trunc i64 [[TMP1]] to i32
1684 // CHECK2-NEXT: store i32 [[CONV]], i32* [[DOTOMP_LB]], align 4
1685 // CHECK2-NEXT: store i32 [[CONV1]], i32* [[DOTOMP_UB]], align 4
1686 // CHECK2-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4
1687 // CHECK2-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4
1688 // CHECK2-NEXT: [[TMP2:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
1689 // CHECK2-NEXT: [[TMP3:%.*]] = load i32, i32* [[TMP2]], align 4
1690 // CHECK2-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB2:[0-9]+]], i32 [[TMP3]], i32 34, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_LB]], i32* [[DOTOMP_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1)
1691 // CHECK2-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
1692 // CHECK2-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP4]], 99
1693 // CHECK2-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
1694 // CHECK2: cond.true:
1695 // CHECK2-NEXT: br label [[COND_END:%.*]]
1696 // CHECK2: cond.false:
1697 // CHECK2-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
1698 // CHECK2-NEXT: br label [[COND_END]]
1699 // CHECK2: cond.end:
1700 // CHECK2-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP5]], [[COND_FALSE]] ]
1701 // CHECK2-NEXT: store i32 [[COND]], i32* [[DOTOMP_UB]], align 4
1702 // CHECK2-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4
1703 // CHECK2-NEXT: store i32 [[TMP6]], i32* [[DOTOMP_IV]], align 4
1704 // CHECK2-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
1705 // CHECK2: omp.inner.for.cond:
1706 // CHECK2-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
1707 // CHECK2-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
1708 // CHECK2-NEXT: [[CMP2:%.*]] = icmp sle i32 [[TMP7]], [[TMP8]]
1709 // CHECK2-NEXT: br i1 [[CMP2]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
1710 // CHECK2: omp.inner.for.body:
1711 // CHECK2-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
1712 // CHECK2-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP9]], 1
1713 // CHECK2-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
1714 // CHECK2-NEXT: store i32 [[ADD]], i32* [[I]], align 4
1715 // CHECK2-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
1716 // CHECK2: omp.body.continue:
1717 // CHECK2-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
1718 // CHECK2: omp.inner.for.inc:
1719 // CHECK2-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
1720 // CHECK2-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP10]], 1
1721 // CHECK2-NEXT: store i32 [[ADD3]], i32* [[DOTOMP_IV]], align 4
1722 // CHECK2-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP12:![0-9]+]]
1723 // CHECK2: omp.inner.for.end:
1724 // CHECK2-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
1725 // CHECK2: omp.loop.exit:
1726 // CHECK2-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP3]])
1727 // CHECK2-NEXT: [[TMP11:%.*]] = load i32, i32* [[DOTOMP_IS_LAST]], align 4
1728 // CHECK2-NEXT: [[TMP12:%.*]] = icmp ne i32 [[TMP11]], 0
1729 // CHECK2-NEXT: br i1 [[TMP12]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
1730 // CHECK2: .omp.final.then:
1731 // CHECK2-NEXT: store i32 100, i32* [[I]], align 4
1732 // CHECK2-NEXT: br label [[DOTOMP_FINAL_DONE]]
1733 // CHECK2: .omp.final.done:
1734 // CHECK2-NEXT: ret void
1735 //
1736 //
1737 // CHECK2-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9gtid_testv_l47
1738 // CHECK2-SAME: () #[[ATTR1]] {
1739 // CHECK2-NEXT: entry:
1740 // CHECK2-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 0, void (i32*, i32*, ...)* bitcast (void (i32*, i32*)* @.omp_outlined..2 to void (i32*, i32*, ...)*))
1741 // CHECK2-NEXT: ret void
1742 //
1743 //
1744 // CHECK2-LABEL: define {{[^@]+}}@.omp_outlined..2
1745 // CHECK2-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]]) #[[ATTR1]] {
1746 // CHECK2-NEXT: entry:
1747 // CHECK2-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
1748 // CHECK2-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
1749 // CHECK2-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
1750 // CHECK2-NEXT: [[TMP:%.*]] = alloca i32, align 4
1751 // CHECK2-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4
1752 // CHECK2-NEXT: [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4
1753 // CHECK2-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
1754 // CHECK2-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
1755 // CHECK2-NEXT: [[I:%.*]] = alloca i32, align 4
1756 // CHECK2-NEXT: [[DOTBOUND_ZERO_ADDR:%.*]] = alloca i32, align 4
1757 // CHECK2-NEXT: store i32 0, i32* [[DOTBOUND_ZERO_ADDR]], align 4
1758 // CHECK2-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
1759 // CHECK2-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
1760 // CHECK2-NEXT: store i32 0, i32* [[DOTOMP_COMB_LB]], align 4
1761 // CHECK2-NEXT: store i32 99, i32* [[DOTOMP_COMB_UB]], align 4
1762 // CHECK2-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4
1763 // CHECK2-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4
1764 // CHECK2-NEXT: [[TMP0:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
1765 // CHECK2-NEXT: [[TMP1:%.*]] = load i32, i32* [[TMP0]], align 4
1766 // CHECK2-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB1]], i32 [[TMP1]], i32 92, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_COMB_LB]], i32* [[DOTOMP_COMB_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1)
1767 // CHECK2-NEXT: [[TMP2:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
1768 // CHECK2-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP2]], 99
1769 // CHECK2-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
1770 // CHECK2: cond.true:
1771 // CHECK2-NEXT: br label [[COND_END:%.*]]
1772 // CHECK2: cond.false:
1773 // CHECK2-NEXT: [[TMP3:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
1774 // CHECK2-NEXT: br label [[COND_END]]
1775 // CHECK2: cond.end:
1776 // CHECK2-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP3]], [[COND_FALSE]] ]
1777 // CHECK2-NEXT: store i32 [[COND]], i32* [[DOTOMP_COMB_UB]], align 4
1778 // CHECK2-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4
1779 // CHECK2-NEXT: store i32 [[TMP4]], i32* [[DOTOMP_IV]], align 4
1780 // CHECK2-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
1781 // CHECK2: omp.inner.for.cond:
1782 // CHECK2-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
1783 // CHECK2-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
1784 // CHECK2-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP5]], [[TMP6]]
1785 // CHECK2-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
1786 // CHECK2: omp.inner.for.body:
1787 // CHECK2-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4
1788 // CHECK2-NEXT: [[TMP8:%.*]] = zext i32 [[TMP7]] to i64
1789 // CHECK2-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
1790 // CHECK2-NEXT: [[TMP10:%.*]] = zext i32 [[TMP9]] to i64
1791 // CHECK2-NEXT: call void @__kmpc_serialized_parallel(%struct.ident_t* @[[GLOB3]], i32 [[TMP1]])
1792 // CHECK2-NEXT: [[TMP11:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
1793 // CHECK2-NEXT: call void @.omp_outlined..3(i32* [[TMP11]], i32* [[DOTBOUND_ZERO_ADDR]], i64 [[TMP8]], i64 [[TMP10]]) #[[ATTR2]]
1794 // CHECK2-NEXT: call void @__kmpc_end_serialized_parallel(%struct.ident_t* @[[GLOB3]], i32 [[TMP1]])
1795 // CHECK2-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
1796 // CHECK2: omp.inner.for.inc:
1797 // CHECK2-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
1798 // CHECK2-NEXT: [[TMP13:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4
1799 // CHECK2-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP12]], [[TMP13]]
1800 // CHECK2-NEXT: store i32 [[ADD]], i32* [[DOTOMP_IV]], align 4
1801 // CHECK2-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP15:![0-9]+]]
1802 // CHECK2: omp.inner.for.end:
1803 // CHECK2-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
1804 // CHECK2: omp.loop.exit:
1805 // CHECK2-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP1]])
1806 // CHECK2-NEXT: [[TMP14:%.*]] = load i32, i32* [[DOTOMP_IS_LAST]], align 4
1807 // CHECK2-NEXT: [[TMP15:%.*]] = icmp ne i32 [[TMP14]], 0
1808 // CHECK2-NEXT: br i1 [[TMP15]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
1809 // CHECK2: .omp.final.then:
1810 // CHECK2-NEXT: store i32 100, i32* [[I]], align 4
1811 // CHECK2-NEXT: br label [[DOTOMP_FINAL_DONE]]
1812 // CHECK2: .omp.final.done:
1813 // CHECK2-NEXT: ret void
1814 //
1815 //
1816 // CHECK2-LABEL: define {{[^@]+}}@.omp_outlined..3
1817 // CHECK2-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[DOTPREVIOUS_LB_:%.*]], i64 [[DOTPREVIOUS_UB_:%.*]]) #[[ATTR1]] {
1818 // CHECK2-NEXT: entry:
1819 // CHECK2-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
1820 // CHECK2-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
1821 // CHECK2-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i64, align 8
1822 // CHECK2-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i64, align 8
1823 // CHECK2-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
1824 // CHECK2-NEXT: [[TMP:%.*]] = alloca i32, align 4
1825 // CHECK2-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4
1826 // CHECK2-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4
1827 // CHECK2-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
1828 // CHECK2-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
1829 // CHECK2-NEXT: [[I:%.*]] = alloca i32, align 4
1830 // CHECK2-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
1831 // CHECK2-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
1832 // CHECK2-NEXT: store i64 [[DOTPREVIOUS_LB_]], i64* [[DOTPREVIOUS_LB__ADDR]], align 8
1833 // CHECK2-NEXT: store i64 [[DOTPREVIOUS_UB_]], i64* [[DOTPREVIOUS_UB__ADDR]], align 8
1834 // CHECK2-NEXT: store i32 0, i32* [[DOTOMP_LB]], align 4
1835 // CHECK2-NEXT: store i32 99, i32* [[DOTOMP_UB]], align 4
1836 // CHECK2-NEXT: [[TMP0:%.*]] = load i64, i64* [[DOTPREVIOUS_LB__ADDR]], align 8
1837 // CHECK2-NEXT: [[CONV:%.*]] = trunc i64 [[TMP0]] to i32
1838 // CHECK2-NEXT: [[TMP1:%.*]] = load i64, i64* [[DOTPREVIOUS_UB__ADDR]], align 8
1839 // CHECK2-NEXT: [[CONV1:%.*]] = trunc i64 [[TMP1]] to i32
1840 // CHECK2-NEXT: store i32 [[CONV]], i32* [[DOTOMP_LB]], align 4
1841 // CHECK2-NEXT: store i32 [[CONV1]], i32* [[DOTOMP_UB]], align 4
1842 // CHECK2-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4
1843 // CHECK2-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4
1844 // CHECK2-NEXT: [[TMP2:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
1845 // CHECK2-NEXT: [[TMP3:%.*]] = load i32, i32* [[TMP2]], align 4
1846 // CHECK2-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB2]], i32 [[TMP3]], i32 34, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_LB]], i32* [[DOTOMP_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1)
1847 // CHECK2-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
1848 // CHECK2-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP4]], 99
1849 // CHECK2-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
1850 // CHECK2: cond.true:
1851 // CHECK2-NEXT: br label [[COND_END:%.*]]
1852 // CHECK2: cond.false:
1853 // CHECK2-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
1854 // CHECK2-NEXT: br label [[COND_END]]
1855 // CHECK2: cond.end:
1856 // CHECK2-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP5]], [[COND_FALSE]] ]
1857 // CHECK2-NEXT: store i32 [[COND]], i32* [[DOTOMP_UB]], align 4
1858 // CHECK2-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4
1859 // CHECK2-NEXT: store i32 [[TMP6]], i32* [[DOTOMP_IV]], align 4
1860 // CHECK2-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
1861 // CHECK2: omp.inner.for.cond:
1862 // CHECK2-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
1863 // CHECK2-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
1864 // CHECK2-NEXT: [[CMP2:%.*]] = icmp sle i32 [[TMP7]], [[TMP8]]
1865 // CHECK2-NEXT: br i1 [[CMP2]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
1866 // CHECK2: omp.inner.for.body:
1867 // CHECK2-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
1868 // CHECK2-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP9]], 1
1869 // CHECK2-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
1870 // CHECK2-NEXT: store i32 [[ADD]], i32* [[I]], align 4
1871 // CHECK2-NEXT: call void @_Z9gtid_testv()
1872 // CHECK2-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
1873 // CHECK2: omp.body.continue:
1874 // CHECK2-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
1875 // CHECK2: omp.inner.for.inc:
1876 // CHECK2-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
1877 // CHECK2-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP10]], 1
1878 // CHECK2-NEXT: store i32 [[ADD3]], i32* [[DOTOMP_IV]], align 4
1879 // CHECK2-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP16:![0-9]+]]
1880 // CHECK2: omp.inner.for.end:
1881 // CHECK2-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
1882 // CHECK2: omp.loop.exit:
1883 // CHECK2-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP3]])
1884 // CHECK2-NEXT: [[TMP11:%.*]] = load i32, i32* [[DOTOMP_IS_LAST]], align 4
1885 // CHECK2-NEXT: [[TMP12:%.*]] = icmp ne i32 [[TMP11]], 0
1886 // CHECK2-NEXT: br i1 [[TMP12]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
1887 // CHECK2: .omp.final.then:
1888 // CHECK2-NEXT: store i32 100, i32* [[I]], align 4
1889 // CHECK2-NEXT: br label [[DOTOMP_FINAL_DONE]]
1890 // CHECK2: .omp.final.done:
1891 // CHECK2-NEXT: ret void
1892 //
1893 //
1894 // CHECK2-LABEL: define {{[^@]+}}@main
1895 // CHECK2-SAME: () #[[ATTR3:[0-9]+]] {
1896 // CHECK2-NEXT: entry:
1897 // CHECK2-NEXT: [[RETVAL:%.*]] = alloca i32, align 4
1898 // CHECK2-NEXT: [[TMP:%.*]] = alloca i32, align 4
1899 // CHECK2-NEXT: [[_TMP1:%.*]] = alloca i32, align 4
1900 // CHECK2-NEXT: [[ARG_CASTED:%.*]] = alloca i64, align 8
1901 // CHECK2-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [1 x i8*], align 8
1902 // CHECK2-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [1 x i8*], align 8
1903 // CHECK2-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [1 x i8*], align 8
1904 // CHECK2-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i8, align 1
1905 // CHECK2-NEXT: [[_TMP5:%.*]] = alloca i32, align 4
1906 // CHECK2-NEXT: store i32 0, i32* [[RETVAL]], align 4
1907 // CHECK2-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i64 100)
1908 // CHECK2-NEXT: [[TMP0:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l76.region_id, i32 0, i8** null, i8** null, i64* null, i64* null, i8** null, i8** null, i32 0, i32 0)
1909 // CHECK2-NEXT: [[TMP1:%.*]] = icmp ne i32 [[TMP0]], 0
1910 // CHECK2-NEXT: br i1 [[TMP1]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
1911 // CHECK2: omp_offload.failed:
1912 // CHECK2-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l76() #[[ATTR2]]
1913 // CHECK2-NEXT: br label [[OMP_OFFLOAD_CONT]]
1914 // CHECK2: omp_offload.cont:
1915 // CHECK2-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i64 100)
1916 // CHECK2-NEXT: [[TMP2:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l84.region_id, i32 0, i8** null, i8** null, i64* null, i64* null, i8** null, i8** null, i32 0, i32 1)
1917 // CHECK2-NEXT: [[TMP3:%.*]] = icmp ne i32 [[TMP2]], 0
1918 // CHECK2-NEXT: br i1 [[TMP3]], label [[OMP_OFFLOAD_FAILED2:%.*]], label [[OMP_OFFLOAD_CONT3:%.*]]
1919 // CHECK2: omp_offload.failed2:
1920 // CHECK2-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l84() #[[ATTR2]]
1921 // CHECK2-NEXT: br label [[OMP_OFFLOAD_CONT3]]
1922 // CHECK2: omp_offload.cont3:
1923 // CHECK2-NEXT: [[TMP4:%.*]] = load i32, i32* @Arg, align 4
1924 // CHECK2-NEXT: [[CONV:%.*]] = bitcast i64* [[ARG_CASTED]] to i32*
1925 // CHECK2-NEXT: store i32 [[TMP4]], i32* [[CONV]], align 4
1926 // CHECK2-NEXT: [[TMP5:%.*]] = load i64, i64* [[ARG_CASTED]], align 8
1927 // CHECK2-NEXT: [[TMP6:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
1928 // CHECK2-NEXT: [[TMP7:%.*]] = bitcast i8** [[TMP6]] to i64*
1929 // CHECK2-NEXT: store i64 [[TMP5]], i64* [[TMP7]], align 8
1930 // CHECK2-NEXT: [[TMP8:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
1931 // CHECK2-NEXT: [[TMP9:%.*]] = bitcast i8** [[TMP8]] to i64*
1932 // CHECK2-NEXT: store i64 [[TMP5]], i64* [[TMP9]], align 8
1933 // CHECK2-NEXT: [[TMP10:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0
1934 // CHECK2-NEXT: store i8* null, i8** [[TMP10]], align 8
1935 // CHECK2-NEXT: [[TMP11:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
1936 // CHECK2-NEXT: [[TMP12:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
1937 // CHECK2-NEXT: [[TMP13:%.*]] = load i32, i32* @Arg, align 4
1938 // CHECK2-NEXT: [[TOBOOL:%.*]] = icmp ne i32 [[TMP13]], 0
1939 // CHECK2-NEXT: [[FROMBOOL:%.*]] = zext i1 [[TOBOOL]] to i8
1940 // CHECK2-NEXT: store i8 [[FROMBOOL]], i8* [[DOTCAPTURE_EXPR_]], align 1
1941 // CHECK2-NEXT: [[TMP14:%.*]] = load i8, i8* [[DOTCAPTURE_EXPR_]], align 1
1942 // CHECK2-NEXT: [[TOBOOL4:%.*]] = trunc i8 [[TMP14]] to i1
1943 // CHECK2-NEXT: [[TMP15:%.*]] = select i1 [[TOBOOL4]], i32 0, i32 1
1944 // CHECK2-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i64 100)
1945 // CHECK2-NEXT: [[TMP16:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l92.region_id, i32 1, i8** [[TMP11]], i8** [[TMP12]], i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_sizes, i32 0, i32 0), i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_maptypes, i32 0, i32 0), i8** null, i8** null, i32 0, i32 [[TMP15]])
1946 // CHECK2-NEXT: [[TMP17:%.*]] = icmp ne i32 [[TMP16]], 0
1947 // CHECK2-NEXT: br i1 [[TMP17]], label [[OMP_OFFLOAD_FAILED6:%.*]], label [[OMP_OFFLOAD_CONT7:%.*]]
1948 // CHECK2: omp_offload.failed6:
1949 // CHECK2-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l92(i64 [[TMP5]]) #[[ATTR2]]
1950 // CHECK2-NEXT: br label [[OMP_OFFLOAD_CONT7]]
1951 // CHECK2: omp_offload.cont7:
1952 // CHECK2-NEXT: [[TMP18:%.*]] = load i32, i32* @Arg, align 4
1953 // CHECK2-NEXT: [[CALL:%.*]] = call i32 @_Z5tmainIiEiT_(i32 [[TMP18]])
1954 // CHECK2-NEXT: ret i32 [[CALL]]
1955 //
1956 //
1957 // CHECK2-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l76
1958 // CHECK2-SAME: () #[[ATTR1]] {
1959 // CHECK2-NEXT: entry:
1960 // CHECK2-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 0, void (i32*, i32*, ...)* bitcast (void (i32*, i32*)* @.omp_outlined..4 to void (i32*, i32*, ...)*))
1961 // CHECK2-NEXT: ret void
1962 //
1963 //
1964 // CHECK2-LABEL: define {{[^@]+}}@.omp_outlined..4
1965 // CHECK2-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]]) #[[ATTR1]] {
1966 // CHECK2-NEXT: entry:
1967 // CHECK2-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
1968 // CHECK2-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
1969 // CHECK2-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
1970 // CHECK2-NEXT: [[TMP:%.*]] = alloca i32, align 4
1971 // CHECK2-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4
1972 // CHECK2-NEXT: [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4
1973 // CHECK2-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
1974 // CHECK2-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
1975 // CHECK2-NEXT: [[I:%.*]] = alloca i32, align 4
1976 // CHECK2-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
1977 // CHECK2-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
1978 // CHECK2-NEXT: store i32 0, i32* [[DOTOMP_COMB_LB]], align 4
1979 // CHECK2-NEXT: store i32 99, i32* [[DOTOMP_COMB_UB]], align 4
1980 // CHECK2-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4
1981 // CHECK2-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4
1982 // CHECK2-NEXT: [[TMP0:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
1983 // CHECK2-NEXT: [[TMP1:%.*]] = load i32, i32* [[TMP0]], align 4
1984 // CHECK2-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB1]], i32 [[TMP1]], i32 92, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_COMB_LB]], i32* [[DOTOMP_COMB_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1)
1985 // CHECK2-NEXT: [[TMP2:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
1986 // CHECK2-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP2]], 99
1987 // CHECK2-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
1988 // CHECK2: cond.true:
1989 // CHECK2-NEXT: br label [[COND_END:%.*]]
1990 // CHECK2: cond.false:
1991 // CHECK2-NEXT: [[TMP3:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
1992 // CHECK2-NEXT: br label [[COND_END]]
1993 // CHECK2: cond.end:
1994 // CHECK2-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP3]], [[COND_FALSE]] ]
1995 // CHECK2-NEXT: store i32 [[COND]], i32* [[DOTOMP_COMB_UB]], align 4
1996 // CHECK2-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4
1997 // CHECK2-NEXT: store i32 [[TMP4]], i32* [[DOTOMP_IV]], align 4
1998 // CHECK2-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
1999 // CHECK2: omp.inner.for.cond:
2000 // CHECK2-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
2001 // CHECK2-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
2002 // CHECK2-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP5]], [[TMP6]]
2003 // CHECK2-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
2004 // CHECK2: omp.inner.for.body:
2005 // CHECK2-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4
2006 // CHECK2-NEXT: [[TMP8:%.*]] = zext i32 [[TMP7]] to i64
2007 // CHECK2-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
2008 // CHECK2-NEXT: [[TMP10:%.*]] = zext i32 [[TMP9]] to i64
2009 // CHECK2-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB3]], i32 2, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64, i64)* @.omp_outlined..5 to void (i32*, i32*, ...)*), i64 [[TMP8]], i64 [[TMP10]])
2010 // CHECK2-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
2011 // CHECK2: omp.inner.for.inc:
2012 // CHECK2-NEXT: [[TMP11:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
2013 // CHECK2-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4
2014 // CHECK2-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP11]], [[TMP12]]
2015 // CHECK2-NEXT: store i32 [[ADD]], i32* [[DOTOMP_IV]], align 4
2016 // CHECK2-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP17:![0-9]+]]
2017 // CHECK2: omp.inner.for.end:
2018 // CHECK2-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
2019 // CHECK2: omp.loop.exit:
2020 // CHECK2-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP1]])
2021 // CHECK2-NEXT: [[TMP13:%.*]] = load i32, i32* [[DOTOMP_IS_LAST]], align 4
2022 // CHECK2-NEXT: [[TMP14:%.*]] = icmp ne i32 [[TMP13]], 0
2023 // CHECK2-NEXT: br i1 [[TMP14]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
2024 // CHECK2: .omp.final.then:
2025 // CHECK2-NEXT: store i32 100, i32* [[I]], align 4
2026 // CHECK2-NEXT: br label [[DOTOMP_FINAL_DONE]]
2027 // CHECK2: .omp.final.done:
2028 // CHECK2-NEXT: ret void
2029 //
2030 //
2031 // CHECK2-LABEL: define {{[^@]+}}@.omp_outlined..5
2032 // CHECK2-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[DOTPREVIOUS_LB_:%.*]], i64 [[DOTPREVIOUS_UB_:%.*]]) #[[ATTR1]] {
2033 // CHECK2-NEXT: entry:
2034 // CHECK2-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
2035 // CHECK2-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
2036 // CHECK2-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i64, align 8
2037 // CHECK2-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i64, align 8
2038 // CHECK2-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
2039 // CHECK2-NEXT: [[TMP:%.*]] = alloca i32, align 4
2040 // CHECK2-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4
2041 // CHECK2-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4
2042 // CHECK2-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
2043 // CHECK2-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
2044 // CHECK2-NEXT: [[I:%.*]] = alloca i32, align 4
2045 // CHECK2-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
2046 // CHECK2-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
2047 // CHECK2-NEXT: store i64 [[DOTPREVIOUS_LB_]], i64* [[DOTPREVIOUS_LB__ADDR]], align 8
2048 // CHECK2-NEXT: store i64 [[DOTPREVIOUS_UB_]], i64* [[DOTPREVIOUS_UB__ADDR]], align 8
2049 // CHECK2-NEXT: store i32 0, i32* [[DOTOMP_LB]], align 4
2050 // CHECK2-NEXT: store i32 99, i32* [[DOTOMP_UB]], align 4
2051 // CHECK2-NEXT: [[TMP0:%.*]] = load i64, i64* [[DOTPREVIOUS_LB__ADDR]], align 8
2052 // CHECK2-NEXT: [[CONV:%.*]] = trunc i64 [[TMP0]] to i32
2053 // CHECK2-NEXT: [[TMP1:%.*]] = load i64, i64* [[DOTPREVIOUS_UB__ADDR]], align 8
2054 // CHECK2-NEXT: [[CONV1:%.*]] = trunc i64 [[TMP1]] to i32
2055 // CHECK2-NEXT: store i32 [[CONV]], i32* [[DOTOMP_LB]], align 4
2056 // CHECK2-NEXT: store i32 [[CONV1]], i32* [[DOTOMP_UB]], align 4
2057 // CHECK2-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4
2058 // CHECK2-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4
2059 // CHECK2-NEXT: [[TMP2:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
2060 // CHECK2-NEXT: [[TMP3:%.*]] = load i32, i32* [[TMP2]], align 4
2061 // CHECK2-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB2]], i32 [[TMP3]], i32 34, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_LB]], i32* [[DOTOMP_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1)
2062 // CHECK2-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
2063 // CHECK2-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP4]], 99
2064 // CHECK2-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
2065 // CHECK2: cond.true:
2066 // CHECK2-NEXT: br label [[COND_END:%.*]]
2067 // CHECK2: cond.false:
2068 // CHECK2-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
2069 // CHECK2-NEXT: br label [[COND_END]]
2070 // CHECK2: cond.end:
2071 // CHECK2-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP5]], [[COND_FALSE]] ]
2072 // CHECK2-NEXT: store i32 [[COND]], i32* [[DOTOMP_UB]], align 4
2073 // CHECK2-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4
2074 // CHECK2-NEXT: store i32 [[TMP6]], i32* [[DOTOMP_IV]], align 4
2075 // CHECK2-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
2076 // CHECK2: omp.inner.for.cond:
2077 // CHECK2-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
2078 // CHECK2-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
2079 // CHECK2-NEXT: [[CMP2:%.*]] = icmp sle i32 [[TMP7]], [[TMP8]]
2080 // CHECK2-NEXT: br i1 [[CMP2]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
2081 // CHECK2: omp.inner.for.body:
2082 // CHECK2-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
2083 // CHECK2-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP9]], 1
2084 // CHECK2-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
2085 // CHECK2-NEXT: store i32 [[ADD]], i32* [[I]], align 4
2086 // CHECK2-NEXT: call void @_Z3fn4v()
2087 // CHECK2-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
2088 // CHECK2: omp.body.continue:
2089 // CHECK2-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
2090 // CHECK2: omp.inner.for.inc:
2091 // CHECK2-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
2092 // CHECK2-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP10]], 1
2093 // CHECK2-NEXT: store i32 [[ADD3]], i32* [[DOTOMP_IV]], align 4
2094 // CHECK2-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP18:![0-9]+]]
2095 // CHECK2: omp.inner.for.end:
2096 // CHECK2-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
2097 // CHECK2: omp.loop.exit:
2098 // CHECK2-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP3]])
2099 // CHECK2-NEXT: [[TMP11:%.*]] = load i32, i32* [[DOTOMP_IS_LAST]], align 4
2100 // CHECK2-NEXT: [[TMP12:%.*]] = icmp ne i32 [[TMP11]], 0
2101 // CHECK2-NEXT: br i1 [[TMP12]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
2102 // CHECK2: .omp.final.then:
2103 // CHECK2-NEXT: store i32 100, i32* [[I]], align 4
2104 // CHECK2-NEXT: br label [[DOTOMP_FINAL_DONE]]
2105 // CHECK2: .omp.final.done:
2106 // CHECK2-NEXT: ret void
2107 //
2108 //
2109 // CHECK2-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l84
2110 // CHECK2-SAME: () #[[ATTR1]] {
2111 // CHECK2-NEXT: entry:
2112 // CHECK2-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 0, void (i32*, i32*, ...)* bitcast (void (i32*, i32*)* @.omp_outlined..6 to void (i32*, i32*, ...)*))
2113 // CHECK2-NEXT: ret void
2114 //
2115 //
2116 // CHECK2-LABEL: define {{[^@]+}}@.omp_outlined..6
2117 // CHECK2-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]]) #[[ATTR1]] {
2118 // CHECK2-NEXT: entry:
2119 // CHECK2-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
2120 // CHECK2-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
2121 // CHECK2-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
2122 // CHECK2-NEXT: [[TMP:%.*]] = alloca i32, align 4
2123 // CHECK2-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4
2124 // CHECK2-NEXT: [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4
2125 // CHECK2-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
2126 // CHECK2-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
2127 // CHECK2-NEXT: [[I:%.*]] = alloca i32, align 4
2128 // CHECK2-NEXT: [[DOTBOUND_ZERO_ADDR:%.*]] = alloca i32, align 4
2129 // CHECK2-NEXT: store i32 0, i32* [[DOTBOUND_ZERO_ADDR]], align 4
2130 // CHECK2-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
2131 // CHECK2-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
2132 // CHECK2-NEXT: store i32 0, i32* [[DOTOMP_COMB_LB]], align 4
2133 // CHECK2-NEXT: store i32 99, i32* [[DOTOMP_COMB_UB]], align 4
2134 // CHECK2-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4
2135 // CHECK2-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4
2136 // CHECK2-NEXT: [[TMP0:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
2137 // CHECK2-NEXT: [[TMP1:%.*]] = load i32, i32* [[TMP0]], align 4
2138 // CHECK2-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB1]], i32 [[TMP1]], i32 92, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_COMB_LB]], i32* [[DOTOMP_COMB_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1)
2139 // CHECK2-NEXT: [[TMP2:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
2140 // CHECK2-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP2]], 99
2141 // CHECK2-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
2142 // CHECK2: cond.true:
2143 // CHECK2-NEXT: br label [[COND_END:%.*]]
2144 // CHECK2: cond.false:
2145 // CHECK2-NEXT: [[TMP3:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
2146 // CHECK2-NEXT: br label [[COND_END]]
2147 // CHECK2: cond.end:
2148 // CHECK2-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP3]], [[COND_FALSE]] ]
2149 // CHECK2-NEXT: store i32 [[COND]], i32* [[DOTOMP_COMB_UB]], align 4
2150 // CHECK2-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4
2151 // CHECK2-NEXT: store i32 [[TMP4]], i32* [[DOTOMP_IV]], align 4
2152 // CHECK2-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
2153 // CHECK2: omp.inner.for.cond:
2154 // CHECK2-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
2155 // CHECK2-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
2156 // CHECK2-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP5]], [[TMP6]]
2157 // CHECK2-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
2158 // CHECK2: omp.inner.for.body:
2159 // CHECK2-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4
2160 // CHECK2-NEXT: [[TMP8:%.*]] = zext i32 [[TMP7]] to i64
2161 // CHECK2-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
2162 // CHECK2-NEXT: [[TMP10:%.*]] = zext i32 [[TMP9]] to i64
2163 // CHECK2-NEXT: call void @__kmpc_serialized_parallel(%struct.ident_t* @[[GLOB3]], i32 [[TMP1]])
2164 // CHECK2-NEXT: [[TMP11:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
2165 // CHECK2-NEXT: call void @.omp_outlined..7(i32* [[TMP11]], i32* [[DOTBOUND_ZERO_ADDR]], i64 [[TMP8]], i64 [[TMP10]]) #[[ATTR2]]
2166 // CHECK2-NEXT: call void @__kmpc_end_serialized_parallel(%struct.ident_t* @[[GLOB3]], i32 [[TMP1]])
2167 // CHECK2-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
2168 // CHECK2: omp.inner.for.inc:
2169 // CHECK2-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
2170 // CHECK2-NEXT: [[TMP13:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4
2171 // CHECK2-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP12]], [[TMP13]]
2172 // CHECK2-NEXT: store i32 [[ADD]], i32* [[DOTOMP_IV]], align 4
2173 // CHECK2-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP19:![0-9]+]]
2174 // CHECK2: omp.inner.for.end:
2175 // CHECK2-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
2176 // CHECK2: omp.loop.exit:
2177 // CHECK2-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP1]])
2178 // CHECK2-NEXT: [[TMP14:%.*]] = load i32, i32* [[DOTOMP_IS_LAST]], align 4
2179 // CHECK2-NEXT: [[TMP15:%.*]] = icmp ne i32 [[TMP14]], 0
2180 // CHECK2-NEXT: br i1 [[TMP15]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
2181 // CHECK2: .omp.final.then:
2182 // CHECK2-NEXT: store i32 100, i32* [[I]], align 4
2183 // CHECK2-NEXT: br label [[DOTOMP_FINAL_DONE]]
2184 // CHECK2: .omp.final.done:
2185 // CHECK2-NEXT: ret void
2186 //
2187 //
2188 // CHECK2-LABEL: define {{[^@]+}}@.omp_outlined..7
2189 // CHECK2-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[DOTPREVIOUS_LB_:%.*]], i64 [[DOTPREVIOUS_UB_:%.*]]) #[[ATTR1]] {
2190 // CHECK2-NEXT: entry:
2191 // CHECK2-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
2192 // CHECK2-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
2193 // CHECK2-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i64, align 8
2194 // CHECK2-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i64, align 8
2195 // CHECK2-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
2196 // CHECK2-NEXT: [[TMP:%.*]] = alloca i32, align 4
2197 // CHECK2-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4
2198 // CHECK2-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4
2199 // CHECK2-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
2200 // CHECK2-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
2201 // CHECK2-NEXT: [[I:%.*]] = alloca i32, align 4
2202 // CHECK2-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
2203 // CHECK2-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
2204 // CHECK2-NEXT: store i64 [[DOTPREVIOUS_LB_]], i64* [[DOTPREVIOUS_LB__ADDR]], align 8
2205 // CHECK2-NEXT: store i64 [[DOTPREVIOUS_UB_]], i64* [[DOTPREVIOUS_UB__ADDR]], align 8
2206 // CHECK2-NEXT: store i32 0, i32* [[DOTOMP_LB]], align 4
2207 // CHECK2-NEXT: store i32 99, i32* [[DOTOMP_UB]], align 4
2208 // CHECK2-NEXT: [[TMP0:%.*]] = load i64, i64* [[DOTPREVIOUS_LB__ADDR]], align 8
2209 // CHECK2-NEXT: [[CONV:%.*]] = trunc i64 [[TMP0]] to i32
2210 // CHECK2-NEXT: [[TMP1:%.*]] = load i64, i64* [[DOTPREVIOUS_UB__ADDR]], align 8
2211 // CHECK2-NEXT: [[CONV1:%.*]] = trunc i64 [[TMP1]] to i32
2212 // CHECK2-NEXT: store i32 [[CONV]], i32* [[DOTOMP_LB]], align 4
2213 // CHECK2-NEXT: store i32 [[CONV1]], i32* [[DOTOMP_UB]], align 4
2214 // CHECK2-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4
2215 // CHECK2-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4
2216 // CHECK2-NEXT: [[TMP2:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
2217 // CHECK2-NEXT: [[TMP3:%.*]] = load i32, i32* [[TMP2]], align 4
2218 // CHECK2-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB2]], i32 [[TMP3]], i32 34, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_LB]], i32* [[DOTOMP_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1)
2219 // CHECK2-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
2220 // CHECK2-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP4]], 99
2221 // CHECK2-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
2222 // CHECK2: cond.true:
2223 // CHECK2-NEXT: br label [[COND_END:%.*]]
2224 // CHECK2: cond.false:
2225 // CHECK2-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
2226 // CHECK2-NEXT: br label [[COND_END]]
2227 // CHECK2: cond.end:
2228 // CHECK2-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP5]], [[COND_FALSE]] ]
2229 // CHECK2-NEXT: store i32 [[COND]], i32* [[DOTOMP_UB]], align 4
2230 // CHECK2-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4
2231 // CHECK2-NEXT: store i32 [[TMP6]], i32* [[DOTOMP_IV]], align 4
2232 // CHECK2-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
2233 // CHECK2: omp.inner.for.cond:
2234 // CHECK2-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
2235 // CHECK2-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
2236 // CHECK2-NEXT: [[CMP2:%.*]] = icmp sle i32 [[TMP7]], [[TMP8]]
2237 // CHECK2-NEXT: br i1 [[CMP2]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
2238 // CHECK2: omp.inner.for.body:
2239 // CHECK2-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
2240 // CHECK2-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP9]], 1
2241 // CHECK2-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
2242 // CHECK2-NEXT: store i32 [[ADD]], i32* [[I]], align 4
2243 // CHECK2-NEXT: call void @_Z3fn5v()
2244 // CHECK2-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
2245 // CHECK2: omp.body.continue:
2246 // CHECK2-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
2247 // CHECK2: omp.inner.for.inc:
2248 // CHECK2-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
2249 // CHECK2-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP10]], 1
2250 // CHECK2-NEXT: store i32 [[ADD3]], i32* [[DOTOMP_IV]], align 4
2251 // CHECK2-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP20:![0-9]+]]
2252 // CHECK2: omp.inner.for.end:
2253 // CHECK2-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
2254 // CHECK2: omp.loop.exit:
2255 // CHECK2-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP3]])
2256 // CHECK2-NEXT: [[TMP11:%.*]] = load i32, i32* [[DOTOMP_IS_LAST]], align 4
2257 // CHECK2-NEXT: [[TMP12:%.*]] = icmp ne i32 [[TMP11]], 0
2258 // CHECK2-NEXT: br i1 [[TMP12]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
2259 // CHECK2: .omp.final.then:
2260 // CHECK2-NEXT: store i32 100, i32* [[I]], align 4
2261 // CHECK2-NEXT: br label [[DOTOMP_FINAL_DONE]]
2262 // CHECK2: .omp.final.done:
2263 // CHECK2-NEXT: ret void
2264 //
2265 //
2266 // CHECK2-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l92
2267 // CHECK2-SAME: (i64 [[ARG:%.*]]) #[[ATTR1]] {
2268 // CHECK2-NEXT: entry:
2269 // CHECK2-NEXT: [[ARG_ADDR:%.*]] = alloca i64, align 8
2270 // CHECK2-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i8, align 1
2271 // CHECK2-NEXT: [[DOTCAPTURE_EXPR__CASTED:%.*]] = alloca i64, align 8
2272 // CHECK2-NEXT: store i64 [[ARG]], i64* [[ARG_ADDR]], align 8
2273 // CHECK2-NEXT: [[CONV:%.*]] = bitcast i64* [[ARG_ADDR]] to i32*
2274 // CHECK2-NEXT: [[TMP0:%.*]] = load i32, i32* [[CONV]], align 8
2275 // CHECK2-NEXT: [[TOBOOL:%.*]] = icmp ne i32 [[TMP0]], 0
2276 // CHECK2-NEXT: [[FROMBOOL:%.*]] = zext i1 [[TOBOOL]] to i8
2277 // CHECK2-NEXT: store i8 [[FROMBOOL]], i8* [[DOTCAPTURE_EXPR_]], align 1
2278 // CHECK2-NEXT: [[TMP1:%.*]] = load i8, i8* [[DOTCAPTURE_EXPR_]], align 1
2279 // CHECK2-NEXT: [[TOBOOL1:%.*]] = trunc i8 [[TMP1]] to i1
2280 // CHECK2-NEXT: [[CONV2:%.*]] = bitcast i64* [[DOTCAPTURE_EXPR__CASTED]] to i8*
2281 // CHECK2-NEXT: [[FROMBOOL3:%.*]] = zext i1 [[TOBOOL1]] to i8
2282 // CHECK2-NEXT: store i8 [[FROMBOOL3]], i8* [[CONV2]], align 1
2283 // CHECK2-NEXT: [[TMP2:%.*]] = load i64, i64* [[DOTCAPTURE_EXPR__CASTED]], align 8
2284 // CHECK2-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64)* @.omp_outlined..8 to void (i32*, i32*, ...)*), i64 [[TMP2]])
2285 // CHECK2-NEXT: ret void
2286 //
2287 //
2288 // CHECK2-LABEL: define {{[^@]+}}@.omp_outlined..8
2289 // CHECK2-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[DOTCAPTURE_EXPR_:%.*]]) #[[ATTR1]] {
2290 // CHECK2-NEXT: entry:
2291 // CHECK2-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
2292 // CHECK2-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
2293 // CHECK2-NEXT: [[DOTCAPTURE_EXPR__ADDR:%.*]] = alloca i64, align 8
2294 // CHECK2-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
2295 // CHECK2-NEXT: [[TMP:%.*]] = alloca i32, align 4
2296 // CHECK2-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4
2297 // CHECK2-NEXT: [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4
2298 // CHECK2-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
2299 // CHECK2-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
2300 // CHECK2-NEXT: [[I:%.*]] = alloca i32, align 4
2301 // CHECK2-NEXT: [[DOTBOUND_ZERO_ADDR:%.*]] = alloca i32, align 4
2302 // CHECK2-NEXT: store i32 0, i32* [[DOTBOUND_ZERO_ADDR]], align 4
2303 // CHECK2-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
2304 // CHECK2-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
2305 // CHECK2-NEXT: store i64 [[DOTCAPTURE_EXPR_]], i64* [[DOTCAPTURE_EXPR__ADDR]], align 8
2306 // CHECK2-NEXT: [[CONV:%.*]] = bitcast i64* [[DOTCAPTURE_EXPR__ADDR]] to i8*
2307 // CHECK2-NEXT: store i32 0, i32* [[DOTOMP_COMB_LB]], align 4
2308 // CHECK2-NEXT: store i32 99, i32* [[DOTOMP_COMB_UB]], align 4
2309 // CHECK2-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4
2310 // CHECK2-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4
2311 // CHECK2-NEXT: [[TMP0:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
2312 // CHECK2-NEXT: [[TMP1:%.*]] = load i32, i32* [[TMP0]], align 4
2313 // CHECK2-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB1]], i32 [[TMP1]], i32 92, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_COMB_LB]], i32* [[DOTOMP_COMB_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1)
2314 // CHECK2-NEXT: [[TMP2:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
2315 // CHECK2-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP2]], 99
2316 // CHECK2-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
2317 // CHECK2: cond.true:
2318 // CHECK2-NEXT: br label [[COND_END:%.*]]
2319 // CHECK2: cond.false:
2320 // CHECK2-NEXT: [[TMP3:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
2321 // CHECK2-NEXT: br label [[COND_END]]
2322 // CHECK2: cond.end:
2323 // CHECK2-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP3]], [[COND_FALSE]] ]
2324 // CHECK2-NEXT: store i32 [[COND]], i32* [[DOTOMP_COMB_UB]], align 4
2325 // CHECK2-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4
2326 // CHECK2-NEXT: store i32 [[TMP4]], i32* [[DOTOMP_IV]], align 4
2327 // CHECK2-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
2328 // CHECK2: omp.inner.for.cond:
2329 // CHECK2-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
2330 // CHECK2-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
2331 // CHECK2-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP5]], [[TMP6]]
2332 // CHECK2-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
2333 // CHECK2: omp.inner.for.body:
2334 // CHECK2-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4
2335 // CHECK2-NEXT: [[TMP8:%.*]] = zext i32 [[TMP7]] to i64
2336 // CHECK2-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
2337 // CHECK2-NEXT: [[TMP10:%.*]] = zext i32 [[TMP9]] to i64
2338 // CHECK2-NEXT: [[TMP11:%.*]] = load i8, i8* [[CONV]], align 8
2339 // CHECK2-NEXT: [[TOBOOL:%.*]] = trunc i8 [[TMP11]] to i1
2340 // CHECK2-NEXT: br i1 [[TOBOOL]], label [[OMP_IF_THEN:%.*]], label [[OMP_IF_ELSE:%.*]]
2341 // CHECK2: omp_if.then:
2342 // CHECK2-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB3]], i32 2, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64, i64)* @.omp_outlined..9 to void (i32*, i32*, ...)*), i64 [[TMP8]], i64 [[TMP10]])
2343 // CHECK2-NEXT: br label [[OMP_IF_END:%.*]]
2344 // CHECK2: omp_if.else:
2345 // CHECK2-NEXT: call void @__kmpc_serialized_parallel(%struct.ident_t* @[[GLOB3]], i32 [[TMP1]])
2346 // CHECK2-NEXT: [[TMP12:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
2347 // CHECK2-NEXT: call void @.omp_outlined..9(i32* [[TMP12]], i32* [[DOTBOUND_ZERO_ADDR]], i64 [[TMP8]], i64 [[TMP10]]) #[[ATTR2]]
2348 // CHECK2-NEXT: call void @__kmpc_end_serialized_parallel(%struct.ident_t* @[[GLOB3]], i32 [[TMP1]])
2349 // CHECK2-NEXT: br label [[OMP_IF_END]]
2350 // CHECK2: omp_if.end:
2351 // CHECK2-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
2352 // CHECK2: omp.inner.for.inc:
2353 // CHECK2-NEXT: [[TMP13:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
2354 // CHECK2-NEXT: [[TMP14:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4
2355 // CHECK2-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP13]], [[TMP14]]
2356 // CHECK2-NEXT: store i32 [[ADD]], i32* [[DOTOMP_IV]], align 4
2357 // CHECK2-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP21:![0-9]+]]
2358 // CHECK2: omp.inner.for.end:
2359 // CHECK2-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
2360 // CHECK2: omp.loop.exit:
2361 // CHECK2-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP1]])
2362 // CHECK2-NEXT: [[TMP15:%.*]] = load i32, i32* [[DOTOMP_IS_LAST]], align 4
2363 // CHECK2-NEXT: [[TMP16:%.*]] = icmp ne i32 [[TMP15]], 0
2364 // CHECK2-NEXT: br i1 [[TMP16]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
2365 // CHECK2: .omp.final.then:
2366 // CHECK2-NEXT: store i32 100, i32* [[I]], align 4
2367 // CHECK2-NEXT: br label [[DOTOMP_FINAL_DONE]]
2368 // CHECK2: .omp.final.done:
2369 // CHECK2-NEXT: ret void
2370 //
2371 //
2372 // CHECK2-LABEL: define {{[^@]+}}@.omp_outlined..9
2373 // CHECK2-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[DOTPREVIOUS_LB_:%.*]], i64 [[DOTPREVIOUS_UB_:%.*]]) #[[ATTR1]] {
2374 // CHECK2-NEXT: entry:
2375 // CHECK2-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
2376 // CHECK2-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
2377 // CHECK2-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i64, align 8
2378 // CHECK2-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i64, align 8
2379 // CHECK2-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
2380 // CHECK2-NEXT: [[TMP:%.*]] = alloca i32, align 4
2381 // CHECK2-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4
2382 // CHECK2-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4
2383 // CHECK2-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
2384 // CHECK2-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
2385 // CHECK2-NEXT: [[I:%.*]] = alloca i32, align 4
2386 // CHECK2-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
2387 // CHECK2-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
2388 // CHECK2-NEXT: store i64 [[DOTPREVIOUS_LB_]], i64* [[DOTPREVIOUS_LB__ADDR]], align 8
2389 // CHECK2-NEXT: store i64 [[DOTPREVIOUS_UB_]], i64* [[DOTPREVIOUS_UB__ADDR]], align 8
2390 // CHECK2-NEXT: store i32 0, i32* [[DOTOMP_LB]], align 4
2391 // CHECK2-NEXT: store i32 99, i32* [[DOTOMP_UB]], align 4
2392 // CHECK2-NEXT: [[TMP0:%.*]] = load i64, i64* [[DOTPREVIOUS_LB__ADDR]], align 8
2393 // CHECK2-NEXT: [[CONV:%.*]] = trunc i64 [[TMP0]] to i32
2394 // CHECK2-NEXT: [[TMP1:%.*]] = load i64, i64* [[DOTPREVIOUS_UB__ADDR]], align 8
2395 // CHECK2-NEXT: [[CONV1:%.*]] = trunc i64 [[TMP1]] to i32
2396 // CHECK2-NEXT: store i32 [[CONV]], i32* [[DOTOMP_LB]], align 4
2397 // CHECK2-NEXT: store i32 [[CONV1]], i32* [[DOTOMP_UB]], align 4
2398 // CHECK2-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4
2399 // CHECK2-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4
2400 // CHECK2-NEXT: [[TMP2:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
2401 // CHECK2-NEXT: [[TMP3:%.*]] = load i32, i32* [[TMP2]], align 4
2402 // CHECK2-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB2]], i32 [[TMP3]], i32 34, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_LB]], i32* [[DOTOMP_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1)
2403 // CHECK2-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
2404 // CHECK2-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP4]], 99
2405 // CHECK2-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
2406 // CHECK2: cond.true:
2407 // CHECK2-NEXT: br label [[COND_END:%.*]]
2408 // CHECK2: cond.false:
2409 // CHECK2-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
2410 // CHECK2-NEXT: br label [[COND_END]]
2411 // CHECK2: cond.end:
2412 // CHECK2-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP5]], [[COND_FALSE]] ]
2413 // CHECK2-NEXT: store i32 [[COND]], i32* [[DOTOMP_UB]], align 4
2414 // CHECK2-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4
2415 // CHECK2-NEXT: store i32 [[TMP6]], i32* [[DOTOMP_IV]], align 4
2416 // CHECK2-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
2417 // CHECK2: omp.inner.for.cond:
2418 // CHECK2-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
2419 // CHECK2-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
2420 // CHECK2-NEXT: [[CMP2:%.*]] = icmp sle i32 [[TMP7]], [[TMP8]]
2421 // CHECK2-NEXT: br i1 [[CMP2]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
2422 // CHECK2: omp.inner.for.body:
2423 // CHECK2-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
2424 // CHECK2-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP9]], 1
2425 // CHECK2-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
2426 // CHECK2-NEXT: store i32 [[ADD]], i32* [[I]], align 4
2427 // CHECK2-NEXT: call void @_Z3fn6v()
2428 // CHECK2-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
2429 // CHECK2: omp.body.continue:
2430 // CHECK2-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
2431 // CHECK2: omp.inner.for.inc:
2432 // CHECK2-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
2433 // CHECK2-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP10]], 1
2434 // CHECK2-NEXT: store i32 [[ADD3]], i32* [[DOTOMP_IV]], align 4
2435 // CHECK2-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP22:![0-9]+]]
2436 // CHECK2: omp.inner.for.end:
2437 // CHECK2-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
2438 // CHECK2: omp.loop.exit:
2439 // CHECK2-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP3]])
2440 // CHECK2-NEXT: [[TMP11:%.*]] = load i32, i32* [[DOTOMP_IS_LAST]], align 4
2441 // CHECK2-NEXT: [[TMP12:%.*]] = icmp ne i32 [[TMP11]], 0
2442 // CHECK2-NEXT: br i1 [[TMP12]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
2443 // CHECK2: .omp.final.then:
2444 // CHECK2-NEXT: store i32 100, i32* [[I]], align 4
2445 // CHECK2-NEXT: br label [[DOTOMP_FINAL_DONE]]
2446 // CHECK2: .omp.final.done:
2447 // CHECK2-NEXT: ret void
2448 //
2449 //
2450 // CHECK2-LABEL: define {{[^@]+}}@_Z5tmainIiEiT_
2451 // CHECK2-SAME: (i32 [[ARG:%.*]]) #[[ATTR0]] comdat {
2452 // CHECK2-NEXT: entry:
2453 // CHECK2-NEXT: [[ARG_ADDR:%.*]] = alloca i32, align 4
2454 // CHECK2-NEXT: [[TMP:%.*]] = alloca i32, align 4
2455 // CHECK2-NEXT: [[_TMP1:%.*]] = alloca i32, align 4
2456 // CHECK2-NEXT: [[ARG_CASTED:%.*]] = alloca i64, align 8
2457 // CHECK2-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [1 x i8*], align 8
2458 // CHECK2-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [1 x i8*], align 8
2459 // CHECK2-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [1 x i8*], align 8
2460 // CHECK2-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i8, align 1
2461 // CHECK2-NEXT: [[_TMP5:%.*]] = alloca i32, align 4
2462 // CHECK2-NEXT: store i32 [[ARG]], i32* [[ARG_ADDR]], align 4
2463 // CHECK2-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i64 100)
2464 // CHECK2-NEXT: [[TMP0:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiEiT__l57.region_id, i32 0, i8** null, i8** null, i64* null, i64* null, i8** null, i8** null, i32 0, i32 0)
2465 // CHECK2-NEXT: [[TMP1:%.*]] = icmp ne i32 [[TMP0]], 0
2466 // CHECK2-NEXT: br i1 [[TMP1]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
2467 // CHECK2: omp_offload.failed:
2468 // CHECK2-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiEiT__l57() #[[ATTR2]]
2469 // CHECK2-NEXT: br label [[OMP_OFFLOAD_CONT]]
2470 // CHECK2: omp_offload.cont:
2471 // CHECK2-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i64 100)
2472 // CHECK2-NEXT: [[TMP2:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiEiT__l62.region_id, i32 0, i8** null, i8** null, i64* null, i64* null, i8** null, i8** null, i32 0, i32 1)
2473 // CHECK2-NEXT: [[TMP3:%.*]] = icmp ne i32 [[TMP2]], 0
2474 // CHECK2-NEXT: br i1 [[TMP3]], label [[OMP_OFFLOAD_FAILED2:%.*]], label [[OMP_OFFLOAD_CONT3:%.*]]
2475 // CHECK2: omp_offload.failed2:
2476 // CHECK2-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiEiT__l62() #[[ATTR2]]
2477 // CHECK2-NEXT: br label [[OMP_OFFLOAD_CONT3]]
2478 // CHECK2: omp_offload.cont3:
2479 // CHECK2-NEXT: [[TMP4:%.*]] = load i32, i32* [[ARG_ADDR]], align 4
2480 // CHECK2-NEXT: [[CONV:%.*]] = bitcast i64* [[ARG_CASTED]] to i32*
2481 // CHECK2-NEXT: store i32 [[TMP4]], i32* [[CONV]], align 4
2482 // CHECK2-NEXT: [[TMP5:%.*]] = load i64, i64* [[ARG_CASTED]], align 8
2483 // CHECK2-NEXT: [[TMP6:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
2484 // CHECK2-NEXT: [[TMP7:%.*]] = bitcast i8** [[TMP6]] to i64*
2485 // CHECK2-NEXT: store i64 [[TMP5]], i64* [[TMP7]], align 8
2486 // CHECK2-NEXT: [[TMP8:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
2487 // CHECK2-NEXT: [[TMP9:%.*]] = bitcast i8** [[TMP8]] to i64*
2488 // CHECK2-NEXT: store i64 [[TMP5]], i64* [[TMP9]], align 8
2489 // CHECK2-NEXT: [[TMP10:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0
2490 // CHECK2-NEXT: store i8* null, i8** [[TMP10]], align 8
2491 // CHECK2-NEXT: [[TMP11:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
2492 // CHECK2-NEXT: [[TMP12:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
2493 // CHECK2-NEXT: [[TMP13:%.*]] = load i32, i32* [[ARG_ADDR]], align 4
2494 // CHECK2-NEXT: [[TOBOOL:%.*]] = icmp ne i32 [[TMP13]], 0
2495 // CHECK2-NEXT: [[FROMBOOL:%.*]] = zext i1 [[TOBOOL]] to i8
2496 // CHECK2-NEXT: store i8 [[FROMBOOL]], i8* [[DOTCAPTURE_EXPR_]], align 1
2497 // CHECK2-NEXT: [[TMP14:%.*]] = load i8, i8* [[DOTCAPTURE_EXPR_]], align 1
2498 // CHECK2-NEXT: [[TOBOOL4:%.*]] = trunc i8 [[TMP14]] to i1
2499 // CHECK2-NEXT: [[TMP15:%.*]] = select i1 [[TOBOOL4]], i32 0, i32 1
2500 // CHECK2-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i64 100)
2501 // CHECK2-NEXT: [[TMP16:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiEiT__l67.region_id, i32 1, i8** [[TMP11]], i8** [[TMP12]], i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_sizes.16, i32 0, i32 0), i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_maptypes.17, i32 0, i32 0), i8** null, i8** null, i32 0, i32 [[TMP15]])
2502 // CHECK2-NEXT: [[TMP17:%.*]] = icmp ne i32 [[TMP16]], 0
2503 // CHECK2-NEXT: br i1 [[TMP17]], label [[OMP_OFFLOAD_FAILED6:%.*]], label [[OMP_OFFLOAD_CONT7:%.*]]
2504 // CHECK2: omp_offload.failed6:
2505 // CHECK2-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiEiT__l67(i64 [[TMP5]]) #[[ATTR2]]
2506 // CHECK2-NEXT: br label [[OMP_OFFLOAD_CONT7]]
2507 // CHECK2: omp_offload.cont7:
2508 // CHECK2-NEXT: ret i32 0
2509 //
2510 //
2511 // CHECK2-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiEiT__l57
2512 // CHECK2-SAME: () #[[ATTR1]] {
2513 // CHECK2-NEXT: entry:
2514 // CHECK2-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 0, void (i32*, i32*, ...)* bitcast (void (i32*, i32*)* @.omp_outlined..10 to void (i32*, i32*, ...)*))
2515 // CHECK2-NEXT: ret void
2516 //
2517 //
2518 // CHECK2-LABEL: define {{[^@]+}}@.omp_outlined..10
2519 // CHECK2-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]]) #[[ATTR1]] {
2520 // CHECK2-NEXT: entry:
2521 // CHECK2-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
2522 // CHECK2-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
2523 // CHECK2-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
2524 // CHECK2-NEXT: [[TMP:%.*]] = alloca i32, align 4
2525 // CHECK2-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4
2526 // CHECK2-NEXT: [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4
2527 // CHECK2-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
2528 // CHECK2-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
2529 // CHECK2-NEXT: [[I:%.*]] = alloca i32, align 4
2530 // CHECK2-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
2531 // CHECK2-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
2532 // CHECK2-NEXT: store i32 0, i32* [[DOTOMP_COMB_LB]], align 4
2533 // CHECK2-NEXT: store i32 99, i32* [[DOTOMP_COMB_UB]], align 4
2534 // CHECK2-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4
2535 // CHECK2-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4
2536 // CHECK2-NEXT: [[TMP0:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
2537 // CHECK2-NEXT: [[TMP1:%.*]] = load i32, i32* [[TMP0]], align 4
2538 // CHECK2-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB1]], i32 [[TMP1]], i32 92, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_COMB_LB]], i32* [[DOTOMP_COMB_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1)
2539 // CHECK2-NEXT: [[TMP2:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
2540 // CHECK2-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP2]], 99
2541 // CHECK2-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
2542 // CHECK2: cond.true:
2543 // CHECK2-NEXT: br label [[COND_END:%.*]]
2544 // CHECK2: cond.false:
2545 // CHECK2-NEXT: [[TMP3:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
2546 // CHECK2-NEXT: br label [[COND_END]]
2547 // CHECK2: cond.end:
2548 // CHECK2-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP3]], [[COND_FALSE]] ]
2549 // CHECK2-NEXT: store i32 [[COND]], i32* [[DOTOMP_COMB_UB]], align 4
2550 // CHECK2-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4
2551 // CHECK2-NEXT: store i32 [[TMP4]], i32* [[DOTOMP_IV]], align 4
2552 // CHECK2-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
2553 // CHECK2: omp.inner.for.cond:
2554 // CHECK2-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
2555 // CHECK2-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
2556 // CHECK2-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP5]], [[TMP6]]
2557 // CHECK2-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
2558 // CHECK2: omp.inner.for.body:
2559 // CHECK2-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4
2560 // CHECK2-NEXT: [[TMP8:%.*]] = zext i32 [[TMP7]] to i64
2561 // CHECK2-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
2562 // CHECK2-NEXT: [[TMP10:%.*]] = zext i32 [[TMP9]] to i64
2563 // CHECK2-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB3]], i32 2, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64, i64)* @.omp_outlined..11 to void (i32*, i32*, ...)*), i64 [[TMP8]], i64 [[TMP10]])
2564 // CHECK2-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
2565 // CHECK2: omp.inner.for.inc:
2566 // CHECK2-NEXT: [[TMP11:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
2567 // CHECK2-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4
2568 // CHECK2-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP11]], [[TMP12]]
2569 // CHECK2-NEXT: store i32 [[ADD]], i32* [[DOTOMP_IV]], align 4
2570 // CHECK2-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP23:![0-9]+]]
2571 // CHECK2: omp.inner.for.end:
2572 // CHECK2-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
2573 // CHECK2: omp.loop.exit:
2574 // CHECK2-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP1]])
2575 // CHECK2-NEXT: [[TMP13:%.*]] = load i32, i32* [[DOTOMP_IS_LAST]], align 4
2576 // CHECK2-NEXT: [[TMP14:%.*]] = icmp ne i32 [[TMP13]], 0
2577 // CHECK2-NEXT: br i1 [[TMP14]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
2578 // CHECK2: .omp.final.then:
2579 // CHECK2-NEXT: store i32 100, i32* [[I]], align 4
2580 // CHECK2-NEXT: br label [[DOTOMP_FINAL_DONE]]
2581 // CHECK2: .omp.final.done:
2582 // CHECK2-NEXT: ret void
2583 //
2584 //
2585 // CHECK2-LABEL: define {{[^@]+}}@.omp_outlined..11
2586 // CHECK2-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[DOTPREVIOUS_LB_:%.*]], i64 [[DOTPREVIOUS_UB_:%.*]]) #[[ATTR1]] {
2587 // CHECK2-NEXT: entry:
2588 // CHECK2-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
2589 // CHECK2-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
2590 // CHECK2-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i64, align 8
2591 // CHECK2-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i64, align 8
2592 // CHECK2-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
2593 // CHECK2-NEXT: [[TMP:%.*]] = alloca i32, align 4
2594 // CHECK2-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4
2595 // CHECK2-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4
2596 // CHECK2-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
2597 // CHECK2-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
2598 // CHECK2-NEXT: [[I:%.*]] = alloca i32, align 4
2599 // CHECK2-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
2600 // CHECK2-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
2601 // CHECK2-NEXT: store i64 [[DOTPREVIOUS_LB_]], i64* [[DOTPREVIOUS_LB__ADDR]], align 8
2602 // CHECK2-NEXT: store i64 [[DOTPREVIOUS_UB_]], i64* [[DOTPREVIOUS_UB__ADDR]], align 8
2603 // CHECK2-NEXT: store i32 0, i32* [[DOTOMP_LB]], align 4
2604 // CHECK2-NEXT: store i32 99, i32* [[DOTOMP_UB]], align 4
2605 // CHECK2-NEXT: [[TMP0:%.*]] = load i64, i64* [[DOTPREVIOUS_LB__ADDR]], align 8
2606 // CHECK2-NEXT: [[CONV:%.*]] = trunc i64 [[TMP0]] to i32
2607 // CHECK2-NEXT: [[TMP1:%.*]] = load i64, i64* [[DOTPREVIOUS_UB__ADDR]], align 8
2608 // CHECK2-NEXT: [[CONV1:%.*]] = trunc i64 [[TMP1]] to i32
2609 // CHECK2-NEXT: store i32 [[CONV]], i32* [[DOTOMP_LB]], align 4
2610 // CHECK2-NEXT: store i32 [[CONV1]], i32* [[DOTOMP_UB]], align 4
2611 // CHECK2-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4
2612 // CHECK2-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4
2613 // CHECK2-NEXT: [[TMP2:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
2614 // CHECK2-NEXT: [[TMP3:%.*]] = load i32, i32* [[TMP2]], align 4
2615 // CHECK2-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB2]], i32 [[TMP3]], i32 34, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_LB]], i32* [[DOTOMP_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1)
2616 // CHECK2-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
2617 // CHECK2-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP4]], 99
2618 // CHECK2-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
2619 // CHECK2: cond.true:
2620 // CHECK2-NEXT: br label [[COND_END:%.*]]
2621 // CHECK2: cond.false:
2622 // CHECK2-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
2623 // CHECK2-NEXT: br label [[COND_END]]
2624 // CHECK2: cond.end:
2625 // CHECK2-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP5]], [[COND_FALSE]] ]
2626 // CHECK2-NEXT: store i32 [[COND]], i32* [[DOTOMP_UB]], align 4
2627 // CHECK2-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4
2628 // CHECK2-NEXT: store i32 [[TMP6]], i32* [[DOTOMP_IV]], align 4
2629 // CHECK2-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
2630 // CHECK2: omp.inner.for.cond:
2631 // CHECK2-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
2632 // CHECK2-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
2633 // CHECK2-NEXT: [[CMP2:%.*]] = icmp sle i32 [[TMP7]], [[TMP8]]
2634 // CHECK2-NEXT: br i1 [[CMP2]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
2635 // CHECK2: omp.inner.for.body:
2636 // CHECK2-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
2637 // CHECK2-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP9]], 1
2638 // CHECK2-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
2639 // CHECK2-NEXT: store i32 [[ADD]], i32* [[I]], align 4
2640 // CHECK2-NEXT: call void @_Z3fn1v()
2641 // CHECK2-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
2642 // CHECK2: omp.body.continue:
2643 // CHECK2-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
2644 // CHECK2: omp.inner.for.inc:
2645 // CHECK2-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
2646 // CHECK2-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP10]], 1
2647 // CHECK2-NEXT: store i32 [[ADD3]], i32* [[DOTOMP_IV]], align 4
2648 // CHECK2-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP24:![0-9]+]]
2649 // CHECK2: omp.inner.for.end:
2650 // CHECK2-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
2651 // CHECK2: omp.loop.exit:
2652 // CHECK2-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP3]])
2653 // CHECK2-NEXT: [[TMP11:%.*]] = load i32, i32* [[DOTOMP_IS_LAST]], align 4
2654 // CHECK2-NEXT: [[TMP12:%.*]] = icmp ne i32 [[TMP11]], 0
2655 // CHECK2-NEXT: br i1 [[TMP12]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
2656 // CHECK2: .omp.final.then:
2657 // CHECK2-NEXT: store i32 100, i32* [[I]], align 4
2658 // CHECK2-NEXT: br label [[DOTOMP_FINAL_DONE]]
2659 // CHECK2: .omp.final.done:
2660 // CHECK2-NEXT: ret void
2661 //
2662 //
2663 // CHECK2-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiEiT__l62
2664 // CHECK2-SAME: () #[[ATTR1]] {
2665 // CHECK2-NEXT: entry:
2666 // CHECK2-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 0, void (i32*, i32*, ...)* bitcast (void (i32*, i32*)* @.omp_outlined..12 to void (i32*, i32*, ...)*))
2667 // CHECK2-NEXT: ret void
2668 //
2669 //
2670 // CHECK2-LABEL: define {{[^@]+}}@.omp_outlined..12
2671 // CHECK2-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]]) #[[ATTR1]] {
2672 // CHECK2-NEXT: entry:
2673 // CHECK2-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
2674 // CHECK2-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
2675 // CHECK2-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
2676 // CHECK2-NEXT: [[TMP:%.*]] = alloca i32, align 4
2677 // CHECK2-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4
2678 // CHECK2-NEXT: [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4
2679 // CHECK2-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
2680 // CHECK2-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
2681 // CHECK2-NEXT: [[I:%.*]] = alloca i32, align 4
2682 // CHECK2-NEXT: [[DOTBOUND_ZERO_ADDR:%.*]] = alloca i32, align 4
2683 // CHECK2-NEXT: store i32 0, i32* [[DOTBOUND_ZERO_ADDR]], align 4
2684 // CHECK2-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
2685 // CHECK2-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
2686 // CHECK2-NEXT: store i32 0, i32* [[DOTOMP_COMB_LB]], align 4
2687 // CHECK2-NEXT: store i32 99, i32* [[DOTOMP_COMB_UB]], align 4
2688 // CHECK2-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4
2689 // CHECK2-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4
2690 // CHECK2-NEXT: [[TMP0:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
2691 // CHECK2-NEXT: [[TMP1:%.*]] = load i32, i32* [[TMP0]], align 4
2692 // CHECK2-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB1]], i32 [[TMP1]], i32 92, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_COMB_LB]], i32* [[DOTOMP_COMB_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1)
2693 // CHECK2-NEXT: [[TMP2:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
2694 // CHECK2-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP2]], 99
2695 // CHECK2-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
2696 // CHECK2: cond.true:
2697 // CHECK2-NEXT: br label [[COND_END:%.*]]
2698 // CHECK2: cond.false:
2699 // CHECK2-NEXT: [[TMP3:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
2700 // CHECK2-NEXT: br label [[COND_END]]
2701 // CHECK2: cond.end:
2702 // CHECK2-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP3]], [[COND_FALSE]] ]
2703 // CHECK2-NEXT: store i32 [[COND]], i32* [[DOTOMP_COMB_UB]], align 4
2704 // CHECK2-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4
2705 // CHECK2-NEXT: store i32 [[TMP4]], i32* [[DOTOMP_IV]], align 4
2706 // CHECK2-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
2707 // CHECK2: omp.inner.for.cond:
2708 // CHECK2-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
2709 // CHECK2-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
2710 // CHECK2-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP5]], [[TMP6]]
2711 // CHECK2-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
2712 // CHECK2: omp.inner.for.body:
2713 // CHECK2-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4
2714 // CHECK2-NEXT: [[TMP8:%.*]] = zext i32 [[TMP7]] to i64
2715 // CHECK2-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
2716 // CHECK2-NEXT: [[TMP10:%.*]] = zext i32 [[TMP9]] to i64
2717 // CHECK2-NEXT: call void @__kmpc_serialized_parallel(%struct.ident_t* @[[GLOB3]], i32 [[TMP1]])
2718 // CHECK2-NEXT: [[TMP11:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
2719 // CHECK2-NEXT: call void @.omp_outlined..13(i32* [[TMP11]], i32* [[DOTBOUND_ZERO_ADDR]], i64 [[TMP8]], i64 [[TMP10]]) #[[ATTR2]]
2720 // CHECK2-NEXT: call void @__kmpc_end_serialized_parallel(%struct.ident_t* @[[GLOB3]], i32 [[TMP1]])
2721 // CHECK2-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
2722 // CHECK2: omp.inner.for.inc:
2723 // CHECK2-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
2724 // CHECK2-NEXT: [[TMP13:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4
2725 // CHECK2-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP12]], [[TMP13]]
2726 // CHECK2-NEXT: store i32 [[ADD]], i32* [[DOTOMP_IV]], align 4
2727 // CHECK2-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP25:![0-9]+]]
2728 // CHECK2: omp.inner.for.end:
2729 // CHECK2-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
2730 // CHECK2: omp.loop.exit:
2731 // CHECK2-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP1]])
2732 // CHECK2-NEXT: [[TMP14:%.*]] = load i32, i32* [[DOTOMP_IS_LAST]], align 4
2733 // CHECK2-NEXT: [[TMP15:%.*]] = icmp ne i32 [[TMP14]], 0
2734 // CHECK2-NEXT: br i1 [[TMP15]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
2735 // CHECK2: .omp.final.then:
2736 // CHECK2-NEXT: store i32 100, i32* [[I]], align 4
2737 // CHECK2-NEXT: br label [[DOTOMP_FINAL_DONE]]
2738 // CHECK2: .omp.final.done:
2739 // CHECK2-NEXT: ret void
2740 //
2741 //
2742 // CHECK2-LABEL: define {{[^@]+}}@.omp_outlined..13
2743 // CHECK2-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[DOTPREVIOUS_LB_:%.*]], i64 [[DOTPREVIOUS_UB_:%.*]]) #[[ATTR1]] {
2744 // CHECK2-NEXT: entry:
2745 // CHECK2-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
2746 // CHECK2-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
2747 // CHECK2-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i64, align 8
2748 // CHECK2-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i64, align 8
2749 // CHECK2-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
2750 // CHECK2-NEXT: [[TMP:%.*]] = alloca i32, align 4
2751 // CHECK2-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4
2752 // CHECK2-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4
2753 // CHECK2-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
2754 // CHECK2-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
2755 // CHECK2-NEXT: [[I:%.*]] = alloca i32, align 4
2756 // CHECK2-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
2757 // CHECK2-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
2758 // CHECK2-NEXT: store i64 [[DOTPREVIOUS_LB_]], i64* [[DOTPREVIOUS_LB__ADDR]], align 8
2759 // CHECK2-NEXT: store i64 [[DOTPREVIOUS_UB_]], i64* [[DOTPREVIOUS_UB__ADDR]], align 8
2760 // CHECK2-NEXT: store i32 0, i32* [[DOTOMP_LB]], align 4
2761 // CHECK2-NEXT: store i32 99, i32* [[DOTOMP_UB]], align 4
2762 // CHECK2-NEXT: [[TMP0:%.*]] = load i64, i64* [[DOTPREVIOUS_LB__ADDR]], align 8
2763 // CHECK2-NEXT: [[CONV:%.*]] = trunc i64 [[TMP0]] to i32
2764 // CHECK2-NEXT: [[TMP1:%.*]] = load i64, i64* [[DOTPREVIOUS_UB__ADDR]], align 8
2765 // CHECK2-NEXT: [[CONV1:%.*]] = trunc i64 [[TMP1]] to i32
2766 // CHECK2-NEXT: store i32 [[CONV]], i32* [[DOTOMP_LB]], align 4
2767 // CHECK2-NEXT: store i32 [[CONV1]], i32* [[DOTOMP_UB]], align 4
2768 // CHECK2-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4
2769 // CHECK2-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4
2770 // CHECK2-NEXT: [[TMP2:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
2771 // CHECK2-NEXT: [[TMP3:%.*]] = load i32, i32* [[TMP2]], align 4
2772 // CHECK2-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB2]], i32 [[TMP3]], i32 34, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_LB]], i32* [[DOTOMP_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1)
2773 // CHECK2-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
2774 // CHECK2-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP4]], 99
2775 // CHECK2-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
2776 // CHECK2: cond.true:
2777 // CHECK2-NEXT: br label [[COND_END:%.*]]
2778 // CHECK2: cond.false:
2779 // CHECK2-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
2780 // CHECK2-NEXT: br label [[COND_END]]
2781 // CHECK2: cond.end:
2782 // CHECK2-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP5]], [[COND_FALSE]] ]
2783 // CHECK2-NEXT: store i32 [[COND]], i32* [[DOTOMP_UB]], align 4
2784 // CHECK2-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4
2785 // CHECK2-NEXT: store i32 [[TMP6]], i32* [[DOTOMP_IV]], align 4
2786 // CHECK2-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
2787 // CHECK2: omp.inner.for.cond:
2788 // CHECK2-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
2789 // CHECK2-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
2790 // CHECK2-NEXT: [[CMP2:%.*]] = icmp sle i32 [[TMP7]], [[TMP8]]
2791 // CHECK2-NEXT: br i1 [[CMP2]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
2792 // CHECK2: omp.inner.for.body:
2793 // CHECK2-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
2794 // CHECK2-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP9]], 1
2795 // CHECK2-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
2796 // CHECK2-NEXT: store i32 [[ADD]], i32* [[I]], align 4
2797 // CHECK2-NEXT: call void @_Z3fn2v()
2798 // CHECK2-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
2799 // CHECK2: omp.body.continue:
2800 // CHECK2-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
2801 // CHECK2: omp.inner.for.inc:
2802 // CHECK2-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
2803 // CHECK2-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP10]], 1
2804 // CHECK2-NEXT: store i32 [[ADD3]], i32* [[DOTOMP_IV]], align 4
2805 // CHECK2-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP26:![0-9]+]]
2806 // CHECK2: omp.inner.for.end:
2807 // CHECK2-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
2808 // CHECK2: omp.loop.exit:
2809 // CHECK2-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP3]])
2810 // CHECK2-NEXT: [[TMP11:%.*]] = load i32, i32* [[DOTOMP_IS_LAST]], align 4
2811 // CHECK2-NEXT: [[TMP12:%.*]] = icmp ne i32 [[TMP11]], 0
2812 // CHECK2-NEXT: br i1 [[TMP12]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
2813 // CHECK2: .omp.final.then:
2814 // CHECK2-NEXT: store i32 100, i32* [[I]], align 4
2815 // CHECK2-NEXT: br label [[DOTOMP_FINAL_DONE]]
2816 // CHECK2: .omp.final.done:
2817 // CHECK2-NEXT: ret void
2818 //
2819 //
2820 // CHECK2-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiEiT__l67
2821 // CHECK2-SAME: (i64 [[ARG:%.*]]) #[[ATTR1]] {
2822 // CHECK2-NEXT: entry:
2823 // CHECK2-NEXT: [[ARG_ADDR:%.*]] = alloca i64, align 8
2824 // CHECK2-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i8, align 1
2825 // CHECK2-NEXT: [[DOTCAPTURE_EXPR__CASTED:%.*]] = alloca i64, align 8
2826 // CHECK2-NEXT: store i64 [[ARG]], i64* [[ARG_ADDR]], align 8
2827 // CHECK2-NEXT: [[CONV:%.*]] = bitcast i64* [[ARG_ADDR]] to i32*
2828 // CHECK2-NEXT: [[TMP0:%.*]] = load i32, i32* [[CONV]], align 8
2829 // CHECK2-NEXT: [[TOBOOL:%.*]] = icmp ne i32 [[TMP0]], 0
2830 // CHECK2-NEXT: [[FROMBOOL:%.*]] = zext i1 [[TOBOOL]] to i8
2831 // CHECK2-NEXT: store i8 [[FROMBOOL]], i8* [[DOTCAPTURE_EXPR_]], align 1
2832 // CHECK2-NEXT: [[TMP1:%.*]] = load i8, i8* [[DOTCAPTURE_EXPR_]], align 1
2833 // CHECK2-NEXT: [[TOBOOL1:%.*]] = trunc i8 [[TMP1]] to i1
2834 // CHECK2-NEXT: [[CONV2:%.*]] = bitcast i64* [[DOTCAPTURE_EXPR__CASTED]] to i8*
2835 // CHECK2-NEXT: [[FROMBOOL3:%.*]] = zext i1 [[TOBOOL1]] to i8
2836 // CHECK2-NEXT: store i8 [[FROMBOOL3]], i8* [[CONV2]], align 1
2837 // CHECK2-NEXT: [[TMP2:%.*]] = load i64, i64* [[DOTCAPTURE_EXPR__CASTED]], align 8
2838 // CHECK2-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64)* @.omp_outlined..14 to void (i32*, i32*, ...)*), i64 [[TMP2]])
2839 // CHECK2-NEXT: ret void
2840 //
2841 //
2842 // CHECK2-LABEL: define {{[^@]+}}@.omp_outlined..14
2843 // CHECK2-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[DOTCAPTURE_EXPR_:%.*]]) #[[ATTR1]] {
2844 // CHECK2-NEXT: entry:
2845 // CHECK2-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
2846 // CHECK2-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
2847 // CHECK2-NEXT: [[DOTCAPTURE_EXPR__ADDR:%.*]] = alloca i64, align 8
2848 // CHECK2-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
2849 // CHECK2-NEXT: [[TMP:%.*]] = alloca i32, align 4
2850 // CHECK2-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4
2851 // CHECK2-NEXT: [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4
2852 // CHECK2-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
2853 // CHECK2-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
2854 // CHECK2-NEXT: [[I:%.*]] = alloca i32, align 4
2855 // CHECK2-NEXT: [[DOTBOUND_ZERO_ADDR:%.*]] = alloca i32, align 4
2856 // CHECK2-NEXT: store i32 0, i32* [[DOTBOUND_ZERO_ADDR]], align 4
2857 // CHECK2-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
2858 // CHECK2-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
2859 // CHECK2-NEXT: store i64 [[DOTCAPTURE_EXPR_]], i64* [[DOTCAPTURE_EXPR__ADDR]], align 8
2860 // CHECK2-NEXT: [[CONV:%.*]] = bitcast i64* [[DOTCAPTURE_EXPR__ADDR]] to i8*
2861 // CHECK2-NEXT: store i32 0, i32* [[DOTOMP_COMB_LB]], align 4
2862 // CHECK2-NEXT: store i32 99, i32* [[DOTOMP_COMB_UB]], align 4
2863 // CHECK2-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4
2864 // CHECK2-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4
2865 // CHECK2-NEXT: [[TMP0:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
2866 // CHECK2-NEXT: [[TMP1:%.*]] = load i32, i32* [[TMP0]], align 4
2867 // CHECK2-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB1]], i32 [[TMP1]], i32 92, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_COMB_LB]], i32* [[DOTOMP_COMB_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1)
2868 // CHECK2-NEXT: [[TMP2:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
2869 // CHECK2-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP2]], 99
2870 // CHECK2-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
2871 // CHECK2: cond.true:
2872 // CHECK2-NEXT: br label [[COND_END:%.*]]
2873 // CHECK2: cond.false:
2874 // CHECK2-NEXT: [[TMP3:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
2875 // CHECK2-NEXT: br label [[COND_END]]
2876 // CHECK2: cond.end:
2877 // CHECK2-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP3]], [[COND_FALSE]] ]
2878 // CHECK2-NEXT: store i32 [[COND]], i32* [[DOTOMP_COMB_UB]], align 4
2879 // CHECK2-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4
2880 // CHECK2-NEXT: store i32 [[TMP4]], i32* [[DOTOMP_IV]], align 4
2881 // CHECK2-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
2882 // CHECK2: omp.inner.for.cond:
2883 // CHECK2-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
2884 // CHECK2-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
2885 // CHECK2-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP5]], [[TMP6]]
2886 // CHECK2-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
2887 // CHECK2: omp.inner.for.body:
2888 // CHECK2-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4
2889 // CHECK2-NEXT: [[TMP8:%.*]] = zext i32 [[TMP7]] to i64
2890 // CHECK2-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
2891 // CHECK2-NEXT: [[TMP10:%.*]] = zext i32 [[TMP9]] to i64
2892 // CHECK2-NEXT: [[TMP11:%.*]] = load i8, i8* [[CONV]], align 8
2893 // CHECK2-NEXT: [[TOBOOL:%.*]] = trunc i8 [[TMP11]] to i1
2894 // CHECK2-NEXT: br i1 [[TOBOOL]], label [[OMP_IF_THEN:%.*]], label [[OMP_IF_ELSE:%.*]]
2895 // CHECK2: omp_if.then:
2896 // CHECK2-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB3]], i32 2, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64, i64)* @.omp_outlined..15 to void (i32*, i32*, ...)*), i64 [[TMP8]], i64 [[TMP10]])
2897 // CHECK2-NEXT: br label [[OMP_IF_END:%.*]]
2898 // CHECK2: omp_if.else:
2899 // CHECK2-NEXT: call void @__kmpc_serialized_parallel(%struct.ident_t* @[[GLOB3]], i32 [[TMP1]])
2900 // CHECK2-NEXT: [[TMP12:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
2901 // CHECK2-NEXT: call void @.omp_outlined..15(i32* [[TMP12]], i32* [[DOTBOUND_ZERO_ADDR]], i64 [[TMP8]], i64 [[TMP10]]) #[[ATTR2]]
2902 // CHECK2-NEXT: call void @__kmpc_end_serialized_parallel(%struct.ident_t* @[[GLOB3]], i32 [[TMP1]])
2903 // CHECK2-NEXT: br label [[OMP_IF_END]]
2904 // CHECK2: omp_if.end:
2905 // CHECK2-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
2906 // CHECK2: omp.inner.for.inc:
2907 // CHECK2-NEXT: [[TMP13:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
2908 // CHECK2-NEXT: [[TMP14:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4
2909 // CHECK2-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP13]], [[TMP14]]
2910 // CHECK2-NEXT: store i32 [[ADD]], i32* [[DOTOMP_IV]], align 4
2911 // CHECK2-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP27:![0-9]+]]
2912 // CHECK2: omp.inner.for.end:
2913 // CHECK2-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
2914 // CHECK2: omp.loop.exit:
2915 // CHECK2-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP1]])
2916 // CHECK2-NEXT: [[TMP15:%.*]] = load i32, i32* [[DOTOMP_IS_LAST]], align 4
2917 // CHECK2-NEXT: [[TMP16:%.*]] = icmp ne i32 [[TMP15]], 0
2918 // CHECK2-NEXT: br i1 [[TMP16]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
2919 // CHECK2: .omp.final.then:
2920 // CHECK2-NEXT: store i32 100, i32* [[I]], align 4
2921 // CHECK2-NEXT: br label [[DOTOMP_FINAL_DONE]]
2922 // CHECK2: .omp.final.done:
2923 // CHECK2-NEXT: ret void
2924 //
2925 //
2926 // CHECK2-LABEL: define {{[^@]+}}@.omp_outlined..15
2927 // CHECK2-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[DOTPREVIOUS_LB_:%.*]], i64 [[DOTPREVIOUS_UB_:%.*]]) #[[ATTR1]] {
2928 // CHECK2-NEXT: entry:
2929 // CHECK2-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
2930 // CHECK2-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
2931 // CHECK2-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i64, align 8
2932 // CHECK2-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i64, align 8
2933 // CHECK2-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
2934 // CHECK2-NEXT: [[TMP:%.*]] = alloca i32, align 4
2935 // CHECK2-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4
2936 // CHECK2-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4
2937 // CHECK2-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
2938 // CHECK2-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
2939 // CHECK2-NEXT: [[I:%.*]] = alloca i32, align 4
2940 // CHECK2-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
2941 // CHECK2-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
2942 // CHECK2-NEXT: store i64 [[DOTPREVIOUS_LB_]], i64* [[DOTPREVIOUS_LB__ADDR]], align 8
2943 // CHECK2-NEXT: store i64 [[DOTPREVIOUS_UB_]], i64* [[DOTPREVIOUS_UB__ADDR]], align 8
2944 // CHECK2-NEXT: store i32 0, i32* [[DOTOMP_LB]], align 4
2945 // CHECK2-NEXT: store i32 99, i32* [[DOTOMP_UB]], align 4
2946 // CHECK2-NEXT: [[TMP0:%.*]] = load i64, i64* [[DOTPREVIOUS_LB__ADDR]], align 8
2947 // CHECK2-NEXT: [[CONV:%.*]] = trunc i64 [[TMP0]] to i32
2948 // CHECK2-NEXT: [[TMP1:%.*]] = load i64, i64* [[DOTPREVIOUS_UB__ADDR]], align 8
2949 // CHECK2-NEXT: [[CONV1:%.*]] = trunc i64 [[TMP1]] to i32
2950 // CHECK2-NEXT: store i32 [[CONV]], i32* [[DOTOMP_LB]], align 4
2951 // CHECK2-NEXT: store i32 [[CONV1]], i32* [[DOTOMP_UB]], align 4
2952 // CHECK2-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4
2953 // CHECK2-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4
2954 // CHECK2-NEXT: [[TMP2:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
2955 // CHECK2-NEXT: [[TMP3:%.*]] = load i32, i32* [[TMP2]], align 4
2956 // CHECK2-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB2]], i32 [[TMP3]], i32 34, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_LB]], i32* [[DOTOMP_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1)
2957 // CHECK2-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
2958 // CHECK2-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP4]], 99
2959 // CHECK2-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
2960 // CHECK2: cond.true:
2961 // CHECK2-NEXT: br label [[COND_END:%.*]]
2962 // CHECK2: cond.false:
2963 // CHECK2-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
2964 // CHECK2-NEXT: br label [[COND_END]]
2965 // CHECK2: cond.end:
2966 // CHECK2-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP5]], [[COND_FALSE]] ]
2967 // CHECK2-NEXT: store i32 [[COND]], i32* [[DOTOMP_UB]], align 4
2968 // CHECK2-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4
2969 // CHECK2-NEXT: store i32 [[TMP6]], i32* [[DOTOMP_IV]], align 4
2970 // CHECK2-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
2971 // CHECK2: omp.inner.for.cond:
2972 // CHECK2-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
2973 // CHECK2-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
2974 // CHECK2-NEXT: [[CMP2:%.*]] = icmp sle i32 [[TMP7]], [[TMP8]]
2975 // CHECK2-NEXT: br i1 [[CMP2]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
2976 // CHECK2: omp.inner.for.body:
2977 // CHECK2-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
2978 // CHECK2-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP9]], 1
2979 // CHECK2-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
2980 // CHECK2-NEXT: store i32 [[ADD]], i32* [[I]], align 4
2981 // CHECK2-NEXT: call void @_Z3fn3v()
2982 // CHECK2-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
2983 // CHECK2: omp.body.continue:
2984 // CHECK2-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
2985 // CHECK2: omp.inner.for.inc:
2986 // CHECK2-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
2987 // CHECK2-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP10]], 1
2988 // CHECK2-NEXT: store i32 [[ADD3]], i32* [[DOTOMP_IV]], align 4
2989 // CHECK2-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP28:![0-9]+]]
2990 // CHECK2: omp.inner.for.end:
2991 // CHECK2-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
2992 // CHECK2: omp.loop.exit:
2993 // CHECK2-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP3]])
2994 // CHECK2-NEXT: [[TMP11:%.*]] = load i32, i32* [[DOTOMP_IS_LAST]], align 4
2995 // CHECK2-NEXT: [[TMP12:%.*]] = icmp ne i32 [[TMP11]], 0
2996 // CHECK2-NEXT: br i1 [[TMP12]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
2997 // CHECK2: .omp.final.then:
2998 // CHECK2-NEXT: store i32 100, i32* [[I]], align 4
2999 // CHECK2-NEXT: br label [[DOTOMP_FINAL_DONE]]
3000 // CHECK2: .omp.final.done:
3001 // CHECK2-NEXT: ret void
3002 //
3003 //
3004 // CHECK2-LABEL: define {{[^@]+}}@.omp_offloading.requires_reg
3005 // CHECK2-SAME: () #[[ATTR5:[0-9]+]] {
3006 // CHECK2-NEXT: entry:
3007 // CHECK2-NEXT: call void @__tgt_register_requires(i64 1)
3008 // CHECK2-NEXT: ret void
3009 //
3010 //
3011 // CHECK3-LABEL: define {{[^@]+}}@_Z9gtid_testv
3012 // CHECK3-SAME: () #[[ATTR0:[0-9]+]] {
3013 // CHECK3-NEXT: entry:
3014 // CHECK3-NEXT: [[TMP:%.*]] = alloca i32, align 4
3015 // CHECK3-NEXT: [[_TMP1:%.*]] = alloca i32, align 4
3016 // CHECK3-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3:[0-9]+]], i64 -1, i64 100)
3017 // CHECK3-NEXT: [[TMP0:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9gtid_testv_l43.region_id, i32 0, i8** null, i8** null, i64* null, i64* null, i8** null, i8** null, i32 0, i32 0)
3018 // CHECK3-NEXT: [[TMP1:%.*]] = icmp ne i32 [[TMP0]], 0
3019 // CHECK3-NEXT: br i1 [[TMP1]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
3020 // CHECK3: omp_offload.failed:
3021 // CHECK3-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9gtid_testv_l43() #[[ATTR2:[0-9]+]]
3022 // CHECK3-NEXT: br label [[OMP_OFFLOAD_CONT]]
3023 // CHECK3: omp_offload.cont:
3024 // CHECK3-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i64 100)
3025 // CHECK3-NEXT: [[TMP2:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9gtid_testv_l47.region_id, i32 0, i8** null, i8** null, i64* null, i64* null, i8** null, i8** null, i32 0, i32 1)
3026 // CHECK3-NEXT: [[TMP3:%.*]] = icmp ne i32 [[TMP2]], 0
3027 // CHECK3-NEXT: br i1 [[TMP3]], label [[OMP_OFFLOAD_FAILED2:%.*]], label [[OMP_OFFLOAD_CONT3:%.*]]
3028 // CHECK3: omp_offload.failed2:
3029 // CHECK3-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9gtid_testv_l47() #[[ATTR2]]
3030 // CHECK3-NEXT: br label [[OMP_OFFLOAD_CONT3]]
3031 // CHECK3: omp_offload.cont3:
3032 // CHECK3-NEXT: ret void
3033 //
3034 //
3035 // CHECK3-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9gtid_testv_l43
3036 // CHECK3-SAME: () #[[ATTR1:[0-9]+]] {
3037 // CHECK3-NEXT: entry:
3038 // CHECK3-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 0, void (i32*, i32*, ...)* bitcast (void (i32*, i32*)* @.omp_outlined. to void (i32*, i32*, ...)*))
3039 // CHECK3-NEXT: ret void
3040 //
3041 //
3042 // CHECK3-LABEL: define {{[^@]+}}@.omp_outlined.
3043 // CHECK3-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]]) #[[ATTR1]] {
3044 // CHECK3-NEXT: entry:
3045 // CHECK3-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
3046 // CHECK3-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
3047 // CHECK3-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
3048 // CHECK3-NEXT: [[TMP:%.*]] = alloca i32, align 4
3049 // CHECK3-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4
3050 // CHECK3-NEXT: [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4
3051 // CHECK3-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
3052 // CHECK3-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
3053 // CHECK3-NEXT: [[I:%.*]] = alloca i32, align 4
3054 // CHECK3-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
3055 // CHECK3-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
3056 // CHECK3-NEXT: store i32 0, i32* [[DOTOMP_COMB_LB]], align 4
3057 // CHECK3-NEXT: store i32 99, i32* [[DOTOMP_COMB_UB]], align 4
3058 // CHECK3-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4
3059 // CHECK3-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4
3060 // CHECK3-NEXT: [[TMP0:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
3061 // CHECK3-NEXT: [[TMP1:%.*]] = load i32, i32* [[TMP0]], align 4
3062 // CHECK3-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB1:[0-9]+]], i32 [[TMP1]], i32 92, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_COMB_LB]], i32* [[DOTOMP_COMB_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1)
3063 // CHECK3-NEXT: [[TMP2:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
3064 // CHECK3-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP2]], 99
3065 // CHECK3-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
3066 // CHECK3: cond.true:
3067 // CHECK3-NEXT: br label [[COND_END:%.*]]
3068 // CHECK3: cond.false:
3069 // CHECK3-NEXT: [[TMP3:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
3070 // CHECK3-NEXT: br label [[COND_END]]
3071 // CHECK3: cond.end:
3072 // CHECK3-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP3]], [[COND_FALSE]] ]
3073 // CHECK3-NEXT: store i32 [[COND]], i32* [[DOTOMP_COMB_UB]], align 4
3074 // CHECK3-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4
3075 // CHECK3-NEXT: store i32 [[TMP4]], i32* [[DOTOMP_IV]], align 4
3076 // CHECK3-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
3077 // CHECK3: omp.inner.for.cond:
3078 // CHECK3-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
3079 // CHECK3-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
3080 // CHECK3-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP5]], [[TMP6]]
3081 // CHECK3-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
3082 // CHECK3: omp.inner.for.body:
3083 // CHECK3-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4
3084 // CHECK3-NEXT: [[TMP8:%.*]] = zext i32 [[TMP7]] to i64
3085 // CHECK3-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
3086 // CHECK3-NEXT: [[TMP10:%.*]] = zext i32 [[TMP9]] to i64
3087 // CHECK3-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB3]], i32 2, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64, i64)* @.omp_outlined..1 to void (i32*, i32*, ...)*), i64 [[TMP8]], i64 [[TMP10]])
3088 // CHECK3-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
3089 // CHECK3: omp.inner.for.inc:
3090 // CHECK3-NEXT: [[TMP11:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
3091 // CHECK3-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4
3092 // CHECK3-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP11]], [[TMP12]]
3093 // CHECK3-NEXT: store i32 [[ADD]], i32* [[DOTOMP_IV]], align 4
3094 // CHECK3-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP10:![0-9]+]]
3095 // CHECK3: omp.inner.for.end:
3096 // CHECK3-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
3097 // CHECK3: omp.loop.exit:
3098 // CHECK3-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP1]])
3099 // CHECK3-NEXT: [[TMP13:%.*]] = load i32, i32* [[DOTOMP_IS_LAST]], align 4
3100 // CHECK3-NEXT: [[TMP14:%.*]] = icmp ne i32 [[TMP13]], 0
3101 // CHECK3-NEXT: br i1 [[TMP14]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
3102 // CHECK3: .omp.final.then:
3103 // CHECK3-NEXT: store i32 100, i32* [[I]], align 4
3104 // CHECK3-NEXT: br label [[DOTOMP_FINAL_DONE]]
3105 // CHECK3: .omp.final.done:
3106 // CHECK3-NEXT: ret void
3107 //
3108 //
3109 // CHECK3-LABEL: define {{[^@]+}}@.omp_outlined..1
3110 // CHECK3-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[DOTPREVIOUS_LB_:%.*]], i64 [[DOTPREVIOUS_UB_:%.*]]) #[[ATTR1]] {
3111 // CHECK3-NEXT: entry:
3112 // CHECK3-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
3113 // CHECK3-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
3114 // CHECK3-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i64, align 8
3115 // CHECK3-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i64, align 8
3116 // CHECK3-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
3117 // CHECK3-NEXT: [[TMP:%.*]] = alloca i32, align 4
3118 // CHECK3-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4
3119 // CHECK3-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4
3120 // CHECK3-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
3121 // CHECK3-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
3122 // CHECK3-NEXT: [[I:%.*]] = alloca i32, align 4
3123 // CHECK3-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
3124 // CHECK3-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
3125 // CHECK3-NEXT: store i64 [[DOTPREVIOUS_LB_]], i64* [[DOTPREVIOUS_LB__ADDR]], align 8
3126 // CHECK3-NEXT: store i64 [[DOTPREVIOUS_UB_]], i64* [[DOTPREVIOUS_UB__ADDR]], align 8
3127 // CHECK3-NEXT: store i32 0, i32* [[DOTOMP_LB]], align 4
3128 // CHECK3-NEXT: store i32 99, i32* [[DOTOMP_UB]], align 4
3129 // CHECK3-NEXT: [[TMP0:%.*]] = load i64, i64* [[DOTPREVIOUS_LB__ADDR]], align 8
3130 // CHECK3-NEXT: [[CONV:%.*]] = trunc i64 [[TMP0]] to i32
3131 // CHECK3-NEXT: [[TMP1:%.*]] = load i64, i64* [[DOTPREVIOUS_UB__ADDR]], align 8
3132 // CHECK3-NEXT: [[CONV1:%.*]] = trunc i64 [[TMP1]] to i32
3133 // CHECK3-NEXT: store i32 [[CONV]], i32* [[DOTOMP_LB]], align 4
3134 // CHECK3-NEXT: store i32 [[CONV1]], i32* [[DOTOMP_UB]], align 4
3135 // CHECK3-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4
3136 // CHECK3-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4
3137 // CHECK3-NEXT: [[TMP2:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
3138 // CHECK3-NEXT: [[TMP3:%.*]] = load i32, i32* [[TMP2]], align 4
3139 // CHECK3-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB2:[0-9]+]], i32 [[TMP3]], i32 34, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_LB]], i32* [[DOTOMP_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1)
3140 // CHECK3-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
3141 // CHECK3-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP4]], 99
3142 // CHECK3-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
3143 // CHECK3: cond.true:
3144 // CHECK3-NEXT: br label [[COND_END:%.*]]
3145 // CHECK3: cond.false:
3146 // CHECK3-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
3147 // CHECK3-NEXT: br label [[COND_END]]
3148 // CHECK3: cond.end:
3149 // CHECK3-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP5]], [[COND_FALSE]] ]
3150 // CHECK3-NEXT: store i32 [[COND]], i32* [[DOTOMP_UB]], align 4
3151 // CHECK3-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4
3152 // CHECK3-NEXT: store i32 [[TMP6]], i32* [[DOTOMP_IV]], align 4
3153 // CHECK3-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
3154 // CHECK3: omp.inner.for.cond:
3155 // CHECK3-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
3156 // CHECK3-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
3157 // CHECK3-NEXT: [[CMP2:%.*]] = icmp sle i32 [[TMP7]], [[TMP8]]
3158 // CHECK3-NEXT: br i1 [[CMP2]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
3159 // CHECK3: omp.inner.for.body:
3160 // CHECK3-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
3161 // CHECK3-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP9]], 1
3162 // CHECK3-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
3163 // CHECK3-NEXT: store i32 [[ADD]], i32* [[I]], align 4
3164 // CHECK3-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
3165 // CHECK3: omp.body.continue:
3166 // CHECK3-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
3167 // CHECK3: omp.inner.for.inc:
3168 // CHECK3-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
3169 // CHECK3-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP10]], 1
3170 // CHECK3-NEXT: store i32 [[ADD3]], i32* [[DOTOMP_IV]], align 4
3171 // CHECK3-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP12:![0-9]+]]
3172 // CHECK3: omp.inner.for.end:
3173 // CHECK3-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
3174 // CHECK3: omp.loop.exit:
3175 // CHECK3-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP3]])
3176 // CHECK3-NEXT: [[TMP11:%.*]] = load i32, i32* [[DOTOMP_IS_LAST]], align 4
3177 // CHECK3-NEXT: [[TMP12:%.*]] = icmp ne i32 [[TMP11]], 0
3178 // CHECK3-NEXT: br i1 [[TMP12]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
3179 // CHECK3: .omp.final.then:
3180 // CHECK3-NEXT: store i32 100, i32* [[I]], align 4
3181 // CHECK3-NEXT: br label [[DOTOMP_FINAL_DONE]]
3182 // CHECK3: .omp.final.done:
3183 // CHECK3-NEXT: ret void
3184 //
3185 //
3186 // CHECK3-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9gtid_testv_l47
3187 // CHECK3-SAME: () #[[ATTR1]] {
3188 // CHECK3-NEXT: entry:
3189 // CHECK3-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 0, void (i32*, i32*, ...)* bitcast (void (i32*, i32*)* @.omp_outlined..2 to void (i32*, i32*, ...)*))
3190 // CHECK3-NEXT: ret void
3191 //
3192 //
3193 // CHECK3-LABEL: define {{[^@]+}}@.omp_outlined..2
3194 // CHECK3-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]]) #[[ATTR1]] {
3195 // CHECK3-NEXT: entry:
3196 // CHECK3-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
3197 // CHECK3-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
3198 // CHECK3-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
3199 // CHECK3-NEXT: [[TMP:%.*]] = alloca i32, align 4
3200 // CHECK3-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4
3201 // CHECK3-NEXT: [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4
3202 // CHECK3-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
3203 // CHECK3-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
3204 // CHECK3-NEXT: [[I:%.*]] = alloca i32, align 4
3205 // CHECK3-NEXT: [[DOTBOUND_ZERO_ADDR:%.*]] = alloca i32, align 4
3206 // CHECK3-NEXT: store i32 0, i32* [[DOTBOUND_ZERO_ADDR]], align 4
3207 // CHECK3-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
3208 // CHECK3-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
3209 // CHECK3-NEXT: store i32 0, i32* [[DOTOMP_COMB_LB]], align 4
3210 // CHECK3-NEXT: store i32 99, i32* [[DOTOMP_COMB_UB]], align 4
3211 // CHECK3-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4
3212 // CHECK3-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4
3213 // CHECK3-NEXT: [[TMP0:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
3214 // CHECK3-NEXT: [[TMP1:%.*]] = load i32, i32* [[TMP0]], align 4
3215 // CHECK3-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB1]], i32 [[TMP1]], i32 92, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_COMB_LB]], i32* [[DOTOMP_COMB_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1)
3216 // CHECK3-NEXT: [[TMP2:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
3217 // CHECK3-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP2]], 99
3218 // CHECK3-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
3219 // CHECK3: cond.true:
3220 // CHECK3-NEXT: br label [[COND_END:%.*]]
3221 // CHECK3: cond.false:
3222 // CHECK3-NEXT: [[TMP3:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
3223 // CHECK3-NEXT: br label [[COND_END]]
3224 // CHECK3: cond.end:
3225 // CHECK3-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP3]], [[COND_FALSE]] ]
3226 // CHECK3-NEXT: store i32 [[COND]], i32* [[DOTOMP_COMB_UB]], align 4
3227 // CHECK3-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4
3228 // CHECK3-NEXT: store i32 [[TMP4]], i32* [[DOTOMP_IV]], align 4
3229 // CHECK3-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
3230 // CHECK3: omp.inner.for.cond:
3231 // CHECK3-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
3232 // CHECK3-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
3233 // CHECK3-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP5]], [[TMP6]]
3234 // CHECK3-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
3235 // CHECK3: omp.inner.for.body:
3236 // CHECK3-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4
3237 // CHECK3-NEXT: [[TMP8:%.*]] = zext i32 [[TMP7]] to i64
3238 // CHECK3-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
3239 // CHECK3-NEXT: [[TMP10:%.*]] = zext i32 [[TMP9]] to i64
3240 // CHECK3-NEXT: call void @__kmpc_serialized_parallel(%struct.ident_t* @[[GLOB3]], i32 [[TMP1]])
3241 // CHECK3-NEXT: [[TMP11:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
3242 // CHECK3-NEXT: call void @.omp_outlined..3(i32* [[TMP11]], i32* [[DOTBOUND_ZERO_ADDR]], i64 [[TMP8]], i64 [[TMP10]]) #[[ATTR2]]
3243 // CHECK3-NEXT: call void @__kmpc_end_serialized_parallel(%struct.ident_t* @[[GLOB3]], i32 [[TMP1]])
3244 // CHECK3-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
3245 // CHECK3: omp.inner.for.inc:
3246 // CHECK3-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
3247 // CHECK3-NEXT: [[TMP13:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4
3248 // CHECK3-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP12]], [[TMP13]]
3249 // CHECK3-NEXT: store i32 [[ADD]], i32* [[DOTOMP_IV]], align 4
3250 // CHECK3-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP15:![0-9]+]]
3251 // CHECK3: omp.inner.for.end:
3252 // CHECK3-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
3253 // CHECK3: omp.loop.exit:
3254 // CHECK3-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP1]])
3255 // CHECK3-NEXT: [[TMP14:%.*]] = load i32, i32* [[DOTOMP_IS_LAST]], align 4
3256 // CHECK3-NEXT: [[TMP15:%.*]] = icmp ne i32 [[TMP14]], 0
3257 // CHECK3-NEXT: br i1 [[TMP15]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
3258 // CHECK3: .omp.final.then:
3259 // CHECK3-NEXT: store i32 100, i32* [[I]], align 4
3260 // CHECK3-NEXT: br label [[DOTOMP_FINAL_DONE]]
3261 // CHECK3: .omp.final.done:
3262 // CHECK3-NEXT: ret void
3263 //
3264 //
3265 // CHECK3-LABEL: define {{[^@]+}}@.omp_outlined..3
3266 // CHECK3-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[DOTPREVIOUS_LB_:%.*]], i64 [[DOTPREVIOUS_UB_:%.*]]) #[[ATTR1]] {
3267 // CHECK3-NEXT: entry:
3268 // CHECK3-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
3269 // CHECK3-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
3270 // CHECK3-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i64, align 8
3271 // CHECK3-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i64, align 8
3272 // CHECK3-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
3273 // CHECK3-NEXT: [[TMP:%.*]] = alloca i32, align 4
3274 // CHECK3-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4
3275 // CHECK3-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4
3276 // CHECK3-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
3277 // CHECK3-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
3278 // CHECK3-NEXT: [[I:%.*]] = alloca i32, align 4
3279 // CHECK3-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
3280 // CHECK3-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
3281 // CHECK3-NEXT: store i64 [[DOTPREVIOUS_LB_]], i64* [[DOTPREVIOUS_LB__ADDR]], align 8
3282 // CHECK3-NEXT: store i64 [[DOTPREVIOUS_UB_]], i64* [[DOTPREVIOUS_UB__ADDR]], align 8
3283 // CHECK3-NEXT: store i32 0, i32* [[DOTOMP_LB]], align 4
3284 // CHECK3-NEXT: store i32 99, i32* [[DOTOMP_UB]], align 4
3285 // CHECK3-NEXT: [[TMP0:%.*]] = load i64, i64* [[DOTPREVIOUS_LB__ADDR]], align 8
3286 // CHECK3-NEXT: [[CONV:%.*]] = trunc i64 [[TMP0]] to i32
3287 // CHECK3-NEXT: [[TMP1:%.*]] = load i64, i64* [[DOTPREVIOUS_UB__ADDR]], align 8
3288 // CHECK3-NEXT: [[CONV1:%.*]] = trunc i64 [[TMP1]] to i32
3289 // CHECK3-NEXT: store i32 [[CONV]], i32* [[DOTOMP_LB]], align 4
3290 // CHECK3-NEXT: store i32 [[CONV1]], i32* [[DOTOMP_UB]], align 4
3291 // CHECK3-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4
3292 // CHECK3-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4
3293 // CHECK3-NEXT: [[TMP2:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
3294 // CHECK3-NEXT: [[TMP3:%.*]] = load i32, i32* [[TMP2]], align 4
3295 // CHECK3-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB2]], i32 [[TMP3]], i32 34, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_LB]], i32* [[DOTOMP_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1)
3296 // CHECK3-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
3297 // CHECK3-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP4]], 99
3298 // CHECK3-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
3299 // CHECK3: cond.true:
3300 // CHECK3-NEXT: br label [[COND_END:%.*]]
3301 // CHECK3: cond.false:
3302 // CHECK3-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
3303 // CHECK3-NEXT: br label [[COND_END]]
3304 // CHECK3: cond.end:
3305 // CHECK3-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP5]], [[COND_FALSE]] ]
3306 // CHECK3-NEXT: store i32 [[COND]], i32* [[DOTOMP_UB]], align 4
3307 // CHECK3-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4
3308 // CHECK3-NEXT: store i32 [[TMP6]], i32* [[DOTOMP_IV]], align 4
3309 // CHECK3-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
3310 // CHECK3: omp.inner.for.cond:
3311 // CHECK3-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
3312 // CHECK3-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
3313 // CHECK3-NEXT: [[CMP2:%.*]] = icmp sle i32 [[TMP7]], [[TMP8]]
3314 // CHECK3-NEXT: br i1 [[CMP2]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
3315 // CHECK3: omp.inner.for.body:
3316 // CHECK3-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
3317 // CHECK3-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP9]], 1
3318 // CHECK3-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
3319 // CHECK3-NEXT: store i32 [[ADD]], i32* [[I]], align 4
3320 // CHECK3-NEXT: call void @_Z9gtid_testv()
3321 // CHECK3-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
3322 // CHECK3: omp.body.continue:
3323 // CHECK3-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
3324 // CHECK3: omp.inner.for.inc:
3325 // CHECK3-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
3326 // CHECK3-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP10]], 1
3327 // CHECK3-NEXT: store i32 [[ADD3]], i32* [[DOTOMP_IV]], align 4
3328 // CHECK3-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP16:![0-9]+]]
3329 // CHECK3: omp.inner.for.end:
3330 // CHECK3-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
3331 // CHECK3: omp.loop.exit:
3332 // CHECK3-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP3]])
3333 // CHECK3-NEXT: [[TMP11:%.*]] = load i32, i32* [[DOTOMP_IS_LAST]], align 4
3334 // CHECK3-NEXT: [[TMP12:%.*]] = icmp ne i32 [[TMP11]], 0
3335 // CHECK3-NEXT: br i1 [[TMP12]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
3336 // CHECK3: .omp.final.then:
3337 // CHECK3-NEXT: store i32 100, i32* [[I]], align 4
3338 // CHECK3-NEXT: br label [[DOTOMP_FINAL_DONE]]
3339 // CHECK3: .omp.final.done:
3340 // CHECK3-NEXT: ret void
3341 //
3342 //
3343 // CHECK3-LABEL: define {{[^@]+}}@main
3344 // CHECK3-SAME: () #[[ATTR3:[0-9]+]] {
3345 // CHECK3-NEXT: entry:
3346 // CHECK3-NEXT: [[RETVAL:%.*]] = alloca i32, align 4
3347 // CHECK3-NEXT: [[TMP:%.*]] = alloca i32, align 4
3348 // CHECK3-NEXT: [[_TMP1:%.*]] = alloca i32, align 4
3349 // CHECK3-NEXT: [[ARG_CASTED:%.*]] = alloca i64, align 8
3350 // CHECK3-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [1 x i8*], align 8
3351 // CHECK3-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [1 x i8*], align 8
3352 // CHECK3-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [1 x i8*], align 8
3353 // CHECK3-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i8, align 1
3354 // CHECK3-NEXT: [[_TMP5:%.*]] = alloca i32, align 4
3355 // CHECK3-NEXT: store i32 0, i32* [[RETVAL]], align 4
3356 // CHECK3-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i64 100)
3357 // CHECK3-NEXT: [[TMP0:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l76.region_id, i32 0, i8** null, i8** null, i64* null, i64* null, i8** null, i8** null, i32 0, i32 0)
3358 // CHECK3-NEXT: [[TMP1:%.*]] = icmp ne i32 [[TMP0]], 0
3359 // CHECK3-NEXT: br i1 [[TMP1]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
3360 // CHECK3: omp_offload.failed:
3361 // CHECK3-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l76() #[[ATTR2]]
3362 // CHECK3-NEXT: br label [[OMP_OFFLOAD_CONT]]
3363 // CHECK3: omp_offload.cont:
3364 // CHECK3-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i64 100)
3365 // CHECK3-NEXT: [[TMP2:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l84.region_id, i32 0, i8** null, i8** null, i64* null, i64* null, i8** null, i8** null, i32 0, i32 1)
3366 // CHECK3-NEXT: [[TMP3:%.*]] = icmp ne i32 [[TMP2]], 0
3367 // CHECK3-NEXT: br i1 [[TMP3]], label [[OMP_OFFLOAD_FAILED2:%.*]], label [[OMP_OFFLOAD_CONT3:%.*]]
3368 // CHECK3: omp_offload.failed2:
3369 // CHECK3-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l84() #[[ATTR2]]
3370 // CHECK3-NEXT: br label [[OMP_OFFLOAD_CONT3]]
3371 // CHECK3: omp_offload.cont3:
3372 // CHECK3-NEXT: [[TMP4:%.*]] = load i32, i32* @Arg, align 4
3373 // CHECK3-NEXT: [[CONV:%.*]] = bitcast i64* [[ARG_CASTED]] to i32*
3374 // CHECK3-NEXT: store i32 [[TMP4]], i32* [[CONV]], align 4
3375 // CHECK3-NEXT: [[TMP5:%.*]] = load i64, i64* [[ARG_CASTED]], align 8
3376 // CHECK3-NEXT: [[TMP6:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
3377 // CHECK3-NEXT: [[TMP7:%.*]] = bitcast i8** [[TMP6]] to i64*
3378 // CHECK3-NEXT: store i64 [[TMP5]], i64* [[TMP7]], align 8
3379 // CHECK3-NEXT: [[TMP8:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
3380 // CHECK3-NEXT: [[TMP9:%.*]] = bitcast i8** [[TMP8]] to i64*
3381 // CHECK3-NEXT: store i64 [[TMP5]], i64* [[TMP9]], align 8
3382 // CHECK3-NEXT: [[TMP10:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0
3383 // CHECK3-NEXT: store i8* null, i8** [[TMP10]], align 8
3384 // CHECK3-NEXT: [[TMP11:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
3385 // CHECK3-NEXT: [[TMP12:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
3386 // CHECK3-NEXT: [[TMP13:%.*]] = load i32, i32* @Arg, align 4
3387 // CHECK3-NEXT: [[TOBOOL:%.*]] = icmp ne i32 [[TMP13]], 0
3388 // CHECK3-NEXT: [[FROMBOOL:%.*]] = zext i1 [[TOBOOL]] to i8
3389 // CHECK3-NEXT: store i8 [[FROMBOOL]], i8* [[DOTCAPTURE_EXPR_]], align 1
3390 // CHECK3-NEXT: [[TMP14:%.*]] = load i8, i8* [[DOTCAPTURE_EXPR_]], align 1
3391 // CHECK3-NEXT: [[TOBOOL4:%.*]] = trunc i8 [[TMP14]] to i1
3392 // CHECK3-NEXT: [[TMP15:%.*]] = select i1 [[TOBOOL4]], i32 0, i32 1
3393 // CHECK3-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i64 100)
3394 // CHECK3-NEXT: [[TMP16:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l92.region_id, i32 1, i8** [[TMP11]], i8** [[TMP12]], i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_sizes, i32 0, i32 0), i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_maptypes, i32 0, i32 0), i8** null, i8** null, i32 0, i32 [[TMP15]])
3395 // CHECK3-NEXT: [[TMP17:%.*]] = icmp ne i32 [[TMP16]], 0
3396 // CHECK3-NEXT: br i1 [[TMP17]], label [[OMP_OFFLOAD_FAILED6:%.*]], label [[OMP_OFFLOAD_CONT7:%.*]]
3397 // CHECK3: omp_offload.failed6:
3398 // CHECK3-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l92(i64 [[TMP5]]) #[[ATTR2]]
3399 // CHECK3-NEXT: br label [[OMP_OFFLOAD_CONT7]]
3400 // CHECK3: omp_offload.cont7:
3401 // CHECK3-NEXT: [[TMP18:%.*]] = load i32, i32* @Arg, align 4
3402 // CHECK3-NEXT: [[CALL:%.*]] = call i32 @_Z5tmainIiEiT_(i32 [[TMP18]])
3403 // CHECK3-NEXT: ret i32 [[CALL]]
3404 //
3405 //
3406 // CHECK3-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l76
3407 // CHECK3-SAME: () #[[ATTR1]] {
3408 // CHECK3-NEXT: entry:
3409 // CHECK3-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 0, void (i32*, i32*, ...)* bitcast (void (i32*, i32*)* @.omp_outlined..4 to void (i32*, i32*, ...)*))
3410 // CHECK3-NEXT: ret void
3411 //
3412 //
3413 // CHECK3-LABEL: define {{[^@]+}}@.omp_outlined..4
3414 // CHECK3-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]]) #[[ATTR1]] {
3415 // CHECK3-NEXT: entry:
3416 // CHECK3-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
3417 // CHECK3-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
3418 // CHECK3-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
3419 // CHECK3-NEXT: [[TMP:%.*]] = alloca i32, align 4
3420 // CHECK3-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4
3421 // CHECK3-NEXT: [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4
3422 // CHECK3-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
3423 // CHECK3-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
3424 // CHECK3-NEXT: [[I:%.*]] = alloca i32, align 4
3425 // CHECK3-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
3426 // CHECK3-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
3427 // CHECK3-NEXT: store i32 0, i32* [[DOTOMP_COMB_LB]], align 4
3428 // CHECK3-NEXT: store i32 99, i32* [[DOTOMP_COMB_UB]], align 4
3429 // CHECK3-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4
3430 // CHECK3-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4
3431 // CHECK3-NEXT: [[TMP0:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
3432 // CHECK3-NEXT: [[TMP1:%.*]] = load i32, i32* [[TMP0]], align 4
3433 // CHECK3-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB1]], i32 [[TMP1]], i32 92, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_COMB_LB]], i32* [[DOTOMP_COMB_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1)
3434 // CHECK3-NEXT: [[TMP2:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
3435 // CHECK3-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP2]], 99
3436 // CHECK3-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
3437 // CHECK3: cond.true:
3438 // CHECK3-NEXT: br label [[COND_END:%.*]]
3439 // CHECK3: cond.false:
3440 // CHECK3-NEXT: [[TMP3:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
3441 // CHECK3-NEXT: br label [[COND_END]]
3442 // CHECK3: cond.end:
3443 // CHECK3-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP3]], [[COND_FALSE]] ]
3444 // CHECK3-NEXT: store i32 [[COND]], i32* [[DOTOMP_COMB_UB]], align 4
3445 // CHECK3-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4
3446 // CHECK3-NEXT: store i32 [[TMP4]], i32* [[DOTOMP_IV]], align 4
3447 // CHECK3-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
3448 // CHECK3: omp.inner.for.cond:
3449 // CHECK3-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
3450 // CHECK3-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
3451 // CHECK3-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP5]], [[TMP6]]
3452 // CHECK3-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
3453 // CHECK3: omp.inner.for.body:
3454 // CHECK3-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4
3455 // CHECK3-NEXT: [[TMP8:%.*]] = zext i32 [[TMP7]] to i64
3456 // CHECK3-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
3457 // CHECK3-NEXT: [[TMP10:%.*]] = zext i32 [[TMP9]] to i64
3458 // CHECK3-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB3]], i32 2, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64, i64)* @.omp_outlined..5 to void (i32*, i32*, ...)*), i64 [[TMP8]], i64 [[TMP10]])
3459 // CHECK3-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
3460 // CHECK3: omp.inner.for.inc:
3461 // CHECK3-NEXT: [[TMP11:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
3462 // CHECK3-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4
3463 // CHECK3-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP11]], [[TMP12]]
3464 // CHECK3-NEXT: store i32 [[ADD]], i32* [[DOTOMP_IV]], align 4
3465 // CHECK3-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP17:![0-9]+]]
3466 // CHECK3: omp.inner.for.end:
3467 // CHECK3-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
3468 // CHECK3: omp.loop.exit:
3469 // CHECK3-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP1]])
3470 // CHECK3-NEXT: [[TMP13:%.*]] = load i32, i32* [[DOTOMP_IS_LAST]], align 4
3471 // CHECK3-NEXT: [[TMP14:%.*]] = icmp ne i32 [[TMP13]], 0
3472 // CHECK3-NEXT: br i1 [[TMP14]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
3473 // CHECK3: .omp.final.then:
3474 // CHECK3-NEXT: store i32 100, i32* [[I]], align 4
3475 // CHECK3-NEXT: br label [[DOTOMP_FINAL_DONE]]
3476 // CHECK3: .omp.final.done:
3477 // CHECK3-NEXT: ret void
3478 //
3479 //
3480 // CHECK3-LABEL: define {{[^@]+}}@.omp_outlined..5
3481 // CHECK3-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[DOTPREVIOUS_LB_:%.*]], i64 [[DOTPREVIOUS_UB_:%.*]]) #[[ATTR1]] {
3482 // CHECK3-NEXT: entry:
3483 // CHECK3-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
3484 // CHECK3-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
3485 // CHECK3-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i64, align 8
3486 // CHECK3-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i64, align 8
3487 // CHECK3-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
3488 // CHECK3-NEXT: [[TMP:%.*]] = alloca i32, align 4
3489 // CHECK3-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4
3490 // CHECK3-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4
3491 // CHECK3-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
3492 // CHECK3-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
3493 // CHECK3-NEXT: [[I:%.*]] = alloca i32, align 4
3494 // CHECK3-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
3495 // CHECK3-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
3496 // CHECK3-NEXT: store i64 [[DOTPREVIOUS_LB_]], i64* [[DOTPREVIOUS_LB__ADDR]], align 8
3497 // CHECK3-NEXT: store i64 [[DOTPREVIOUS_UB_]], i64* [[DOTPREVIOUS_UB__ADDR]], align 8
3498 // CHECK3-NEXT: store i32 0, i32* [[DOTOMP_LB]], align 4
3499 // CHECK3-NEXT: store i32 99, i32* [[DOTOMP_UB]], align 4
3500 // CHECK3-NEXT: [[TMP0:%.*]] = load i64, i64* [[DOTPREVIOUS_LB__ADDR]], align 8
3501 // CHECK3-NEXT: [[CONV:%.*]] = trunc i64 [[TMP0]] to i32
3502 // CHECK3-NEXT: [[TMP1:%.*]] = load i64, i64* [[DOTPREVIOUS_UB__ADDR]], align 8
3503 // CHECK3-NEXT: [[CONV1:%.*]] = trunc i64 [[TMP1]] to i32
3504 // CHECK3-NEXT: store i32 [[CONV]], i32* [[DOTOMP_LB]], align 4
3505 // CHECK3-NEXT: store i32 [[CONV1]], i32* [[DOTOMP_UB]], align 4
3506 // CHECK3-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4
3507 // CHECK3-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4
3508 // CHECK3-NEXT: [[TMP2:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
3509 // CHECK3-NEXT: [[TMP3:%.*]] = load i32, i32* [[TMP2]], align 4
3510 // CHECK3-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB2]], i32 [[TMP3]], i32 34, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_LB]], i32* [[DOTOMP_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1)
3511 // CHECK3-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
3512 // CHECK3-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP4]], 99
3513 // CHECK3-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
3514 // CHECK3: cond.true:
3515 // CHECK3-NEXT: br label [[COND_END:%.*]]
3516 // CHECK3: cond.false:
3517 // CHECK3-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
3518 // CHECK3-NEXT: br label [[COND_END]]
3519 // CHECK3: cond.end:
3520 // CHECK3-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP5]], [[COND_FALSE]] ]
3521 // CHECK3-NEXT: store i32 [[COND]], i32* [[DOTOMP_UB]], align 4
3522 // CHECK3-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4
3523 // CHECK3-NEXT: store i32 [[TMP6]], i32* [[DOTOMP_IV]], align 4
3524 // CHECK3-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
3525 // CHECK3: omp.inner.for.cond:
3526 // CHECK3-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
3527 // CHECK3-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
3528 // CHECK3-NEXT: [[CMP2:%.*]] = icmp sle i32 [[TMP7]], [[TMP8]]
3529 // CHECK3-NEXT: br i1 [[CMP2]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
3530 // CHECK3: omp.inner.for.body:
3531 // CHECK3-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
3532 // CHECK3-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP9]], 1
3533 // CHECK3-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
3534 // CHECK3-NEXT: store i32 [[ADD]], i32* [[I]], align 4
3535 // CHECK3-NEXT: call void @_Z3fn4v()
3536 // CHECK3-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
3537 // CHECK3: omp.body.continue:
3538 // CHECK3-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
3539 // CHECK3: omp.inner.for.inc:
3540 // CHECK3-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
3541 // CHECK3-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP10]], 1
3542 // CHECK3-NEXT: store i32 [[ADD3]], i32* [[DOTOMP_IV]], align 4
3543 // CHECK3-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP18:![0-9]+]]
3544 // CHECK3: omp.inner.for.end:
3545 // CHECK3-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
3546 // CHECK3: omp.loop.exit:
3547 // CHECK3-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP3]])
3548 // CHECK3-NEXT: [[TMP11:%.*]] = load i32, i32* [[DOTOMP_IS_LAST]], align 4
3549 // CHECK3-NEXT: [[TMP12:%.*]] = icmp ne i32 [[TMP11]], 0
3550 // CHECK3-NEXT: br i1 [[TMP12]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
3551 // CHECK3: .omp.final.then:
3552 // CHECK3-NEXT: store i32 100, i32* [[I]], align 4
3553 // CHECK3-NEXT: br label [[DOTOMP_FINAL_DONE]]
3554 // CHECK3: .omp.final.done:
3555 // CHECK3-NEXT: ret void
3556 //
3557 //
3558 // CHECK3-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l84
3559 // CHECK3-SAME: () #[[ATTR1]] {
3560 // CHECK3-NEXT: entry:
3561 // CHECK3-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 0, void (i32*, i32*, ...)* bitcast (void (i32*, i32*)* @.omp_outlined..6 to void (i32*, i32*, ...)*))
3562 // CHECK3-NEXT: ret void
3563 //
3564 //
3565 // CHECK3-LABEL: define {{[^@]+}}@.omp_outlined..6
3566 // CHECK3-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]]) #[[ATTR1]] {
3567 // CHECK3-NEXT: entry:
3568 // CHECK3-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
3569 // CHECK3-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
3570 // CHECK3-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
3571 // CHECK3-NEXT: [[TMP:%.*]] = alloca i32, align 4
3572 // CHECK3-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4
3573 // CHECK3-NEXT: [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4
3574 // CHECK3-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
3575 // CHECK3-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
3576 // CHECK3-NEXT: [[I:%.*]] = alloca i32, align 4
3577 // CHECK3-NEXT: [[DOTBOUND_ZERO_ADDR:%.*]] = alloca i32, align 4
3578 // CHECK3-NEXT: store i32 0, i32* [[DOTBOUND_ZERO_ADDR]], align 4
3579 // CHECK3-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
3580 // CHECK3-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
3581 // CHECK3-NEXT: store i32 0, i32* [[DOTOMP_COMB_LB]], align 4
3582 // CHECK3-NEXT: store i32 99, i32* [[DOTOMP_COMB_UB]], align 4
3583 // CHECK3-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4
3584 // CHECK3-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4
3585 // CHECK3-NEXT: [[TMP0:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
3586 // CHECK3-NEXT: [[TMP1:%.*]] = load i32, i32* [[TMP0]], align 4
3587 // CHECK3-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB1]], i32 [[TMP1]], i32 92, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_COMB_LB]], i32* [[DOTOMP_COMB_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1)
3588 // CHECK3-NEXT: [[TMP2:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
3589 // CHECK3-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP2]], 99
3590 // CHECK3-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
3591 // CHECK3: cond.true:
3592 // CHECK3-NEXT: br label [[COND_END:%.*]]
3593 // CHECK3: cond.false:
3594 // CHECK3-NEXT: [[TMP3:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
3595 // CHECK3-NEXT: br label [[COND_END]]
3596 // CHECK3: cond.end:
3597 // CHECK3-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP3]], [[COND_FALSE]] ]
3598 // CHECK3-NEXT: store i32 [[COND]], i32* [[DOTOMP_COMB_UB]], align 4
3599 // CHECK3-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4
3600 // CHECK3-NEXT: store i32 [[TMP4]], i32* [[DOTOMP_IV]], align 4
3601 // CHECK3-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
3602 // CHECK3: omp.inner.for.cond:
3603 // CHECK3-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
3604 // CHECK3-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
3605 // CHECK3-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP5]], [[TMP6]]
3606 // CHECK3-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
3607 // CHECK3: omp.inner.for.body:
3608 // CHECK3-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4
3609 // CHECK3-NEXT: [[TMP8:%.*]] = zext i32 [[TMP7]] to i64
3610 // CHECK3-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
3611 // CHECK3-NEXT: [[TMP10:%.*]] = zext i32 [[TMP9]] to i64
3612 // CHECK3-NEXT: call void @__kmpc_serialized_parallel(%struct.ident_t* @[[GLOB3]], i32 [[TMP1]])
3613 // CHECK3-NEXT: [[TMP11:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
3614 // CHECK3-NEXT: call void @.omp_outlined..7(i32* [[TMP11]], i32* [[DOTBOUND_ZERO_ADDR]], i64 [[TMP8]], i64 [[TMP10]]) #[[ATTR2]]
3615 // CHECK3-NEXT: call void @__kmpc_end_serialized_parallel(%struct.ident_t* @[[GLOB3]], i32 [[TMP1]])
3616 // CHECK3-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
3617 // CHECK3: omp.inner.for.inc:
3618 // CHECK3-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
3619 // CHECK3-NEXT: [[TMP13:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4
3620 // CHECK3-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP12]], [[TMP13]]
3621 // CHECK3-NEXT: store i32 [[ADD]], i32* [[DOTOMP_IV]], align 4
3622 // CHECK3-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP19:![0-9]+]]
3623 // CHECK3: omp.inner.for.end:
3624 // CHECK3-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
3625 // CHECK3: omp.loop.exit:
3626 // CHECK3-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP1]])
3627 // CHECK3-NEXT: [[TMP14:%.*]] = load i32, i32* [[DOTOMP_IS_LAST]], align 4
3628 // CHECK3-NEXT: [[TMP15:%.*]] = icmp ne i32 [[TMP14]], 0
3629 // CHECK3-NEXT: br i1 [[TMP15]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
3630 // CHECK3: .omp.final.then:
3631 // CHECK3-NEXT: store i32 100, i32* [[I]], align 4
3632 // CHECK3-NEXT: br label [[DOTOMP_FINAL_DONE]]
3633 // CHECK3: .omp.final.done:
3634 // CHECK3-NEXT: ret void
3635 //
3636 //
3637 // CHECK3-LABEL: define {{[^@]+}}@.omp_outlined..7
3638 // CHECK3-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[DOTPREVIOUS_LB_:%.*]], i64 [[DOTPREVIOUS_UB_:%.*]]) #[[ATTR1]] {
3639 // CHECK3-NEXT: entry:
3640 // CHECK3-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
3641 // CHECK3-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
3642 // CHECK3-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i64, align 8
3643 // CHECK3-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i64, align 8
3644 // CHECK3-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
3645 // CHECK3-NEXT: [[TMP:%.*]] = alloca i32, align 4
3646 // CHECK3-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4
3647 // CHECK3-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4
3648 // CHECK3-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
3649 // CHECK3-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
3650 // CHECK3-NEXT: [[I:%.*]] = alloca i32, align 4
3651 // CHECK3-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
3652 // CHECK3-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
3653 // CHECK3-NEXT: store i64 [[DOTPREVIOUS_LB_]], i64* [[DOTPREVIOUS_LB__ADDR]], align 8
3654 // CHECK3-NEXT: store i64 [[DOTPREVIOUS_UB_]], i64* [[DOTPREVIOUS_UB__ADDR]], align 8
3655 // CHECK3-NEXT: store i32 0, i32* [[DOTOMP_LB]], align 4
3656 // CHECK3-NEXT: store i32 99, i32* [[DOTOMP_UB]], align 4
3657 // CHECK3-NEXT: [[TMP0:%.*]] = load i64, i64* [[DOTPREVIOUS_LB__ADDR]], align 8
3658 // CHECK3-NEXT: [[CONV:%.*]] = trunc i64 [[TMP0]] to i32
3659 // CHECK3-NEXT: [[TMP1:%.*]] = load i64, i64* [[DOTPREVIOUS_UB__ADDR]], align 8
3660 // CHECK3-NEXT: [[CONV1:%.*]] = trunc i64 [[TMP1]] to i32
3661 // CHECK3-NEXT: store i32 [[CONV]], i32* [[DOTOMP_LB]], align 4
3662 // CHECK3-NEXT: store i32 [[CONV1]], i32* [[DOTOMP_UB]], align 4
3663 // CHECK3-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4
3664 // CHECK3-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4
3665 // CHECK3-NEXT: [[TMP2:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
3666 // CHECK3-NEXT: [[TMP3:%.*]] = load i32, i32* [[TMP2]], align 4
3667 // CHECK3-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB2]], i32 [[TMP3]], i32 34, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_LB]], i32* [[DOTOMP_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1)
3668 // CHECK3-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
3669 // CHECK3-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP4]], 99
3670 // CHECK3-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
3671 // CHECK3: cond.true:
3672 // CHECK3-NEXT: br label [[COND_END:%.*]]
3673 // CHECK3: cond.false:
3674 // CHECK3-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
3675 // CHECK3-NEXT: br label [[COND_END]]
3676 // CHECK3: cond.end:
3677 // CHECK3-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP5]], [[COND_FALSE]] ]
3678 // CHECK3-NEXT: store i32 [[COND]], i32* [[DOTOMP_UB]], align 4
3679 // CHECK3-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4
3680 // CHECK3-NEXT: store i32 [[TMP6]], i32* [[DOTOMP_IV]], align 4
3681 // CHECK3-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
3682 // CHECK3: omp.inner.for.cond:
3683 // CHECK3-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
3684 // CHECK3-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
3685 // CHECK3-NEXT: [[CMP2:%.*]] = icmp sle i32 [[TMP7]], [[TMP8]]
3686 // CHECK3-NEXT: br i1 [[CMP2]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
3687 // CHECK3: omp.inner.for.body:
3688 // CHECK3-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
3689 // CHECK3-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP9]], 1
3690 // CHECK3-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
3691 // CHECK3-NEXT: store i32 [[ADD]], i32* [[I]], align 4
3692 // CHECK3-NEXT: call void @_Z3fn5v()
3693 // CHECK3-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
3694 // CHECK3: omp.body.continue:
3695 // CHECK3-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
3696 // CHECK3: omp.inner.for.inc:
3697 // CHECK3-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
3698 // CHECK3-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP10]], 1
3699 // CHECK3-NEXT: store i32 [[ADD3]], i32* [[DOTOMP_IV]], align 4
3700 // CHECK3-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP21:![0-9]+]]
3701 // CHECK3: omp.inner.for.end:
3702 // CHECK3-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
3703 // CHECK3: omp.loop.exit:
3704 // CHECK3-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP3]])
3705 // CHECK3-NEXT: [[TMP11:%.*]] = load i32, i32* [[DOTOMP_IS_LAST]], align 4
3706 // CHECK3-NEXT: [[TMP12:%.*]] = icmp ne i32 [[TMP11]], 0
3707 // CHECK3-NEXT: br i1 [[TMP12]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
3708 // CHECK3: .omp.final.then:
3709 // CHECK3-NEXT: store i32 100, i32* [[I]], align 4
3710 // CHECK3-NEXT: br label [[DOTOMP_FINAL_DONE]]
3711 // CHECK3: .omp.final.done:
3712 // CHECK3-NEXT: ret void
3713 //
3714 //
3715 // CHECK3-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l92
3716 // CHECK3-SAME: (i64 [[ARG:%.*]]) #[[ATTR1]] {
3717 // CHECK3-NEXT: entry:
3718 // CHECK3-NEXT: [[ARG_ADDR:%.*]] = alloca i64, align 8
3719 // CHECK3-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i8, align 1
3720 // CHECK3-NEXT: [[DOTCAPTURE_EXPR__CASTED:%.*]] = alloca i64, align 8
3721 // CHECK3-NEXT: store i64 [[ARG]], i64* [[ARG_ADDR]], align 8
3722 // CHECK3-NEXT: [[CONV:%.*]] = bitcast i64* [[ARG_ADDR]] to i32*
3723 // CHECK3-NEXT: [[TMP0:%.*]] = load i32, i32* [[CONV]], align 8
3724 // CHECK3-NEXT: [[TOBOOL:%.*]] = icmp ne i32 [[TMP0]], 0
3725 // CHECK3-NEXT: [[FROMBOOL:%.*]] = zext i1 [[TOBOOL]] to i8
3726 // CHECK3-NEXT: store i8 [[FROMBOOL]], i8* [[DOTCAPTURE_EXPR_]], align 1
3727 // CHECK3-NEXT: [[TMP1:%.*]] = load i8, i8* [[DOTCAPTURE_EXPR_]], align 1
3728 // CHECK3-NEXT: [[TOBOOL1:%.*]] = trunc i8 [[TMP1]] to i1
3729 // CHECK3-NEXT: [[CONV2:%.*]] = bitcast i64* [[DOTCAPTURE_EXPR__CASTED]] to i8*
3730 // CHECK3-NEXT: [[FROMBOOL3:%.*]] = zext i1 [[TOBOOL1]] to i8
3731 // CHECK3-NEXT: store i8 [[FROMBOOL3]], i8* [[CONV2]], align 1
3732 // CHECK3-NEXT: [[TMP2:%.*]] = load i64, i64* [[DOTCAPTURE_EXPR__CASTED]], align 8
3733 // CHECK3-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64)* @.omp_outlined..8 to void (i32*, i32*, ...)*), i64 [[TMP2]])
3734 // CHECK3-NEXT: ret void
3735 //
3736 //
3737 // CHECK3-LABEL: define {{[^@]+}}@.omp_outlined..8
3738 // CHECK3-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[DOTCAPTURE_EXPR_:%.*]]) #[[ATTR1]] {
3739 // CHECK3-NEXT: entry:
3740 // CHECK3-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
3741 // CHECK3-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
3742 // CHECK3-NEXT: [[DOTCAPTURE_EXPR__ADDR:%.*]] = alloca i64, align 8
3743 // CHECK3-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
3744 // CHECK3-NEXT: [[TMP:%.*]] = alloca i32, align 4
3745 // CHECK3-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4
3746 // CHECK3-NEXT: [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4
3747 // CHECK3-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
3748 // CHECK3-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
3749 // CHECK3-NEXT: [[I:%.*]] = alloca i32, align 4
3750 // CHECK3-NEXT: [[DOTCAPTURE_EXPR__CASTED:%.*]] = alloca i64, align 8
3751 // CHECK3-NEXT: [[DOTBOUND_ZERO_ADDR:%.*]] = alloca i32, align 4
3752 // CHECK3-NEXT: [[DOTCAPTURE_EXPR__CASTED11:%.*]] = alloca i64, align 8
3753 // CHECK3-NEXT: [[DOTBOUND_ZERO_ADDR17:%.*]] = alloca i32, align 4
3754 // CHECK3-NEXT: store i32 0, i32* [[DOTBOUND_ZERO_ADDR17]], align 4
3755 // CHECK3-NEXT: store i32 0, i32* [[DOTBOUND_ZERO_ADDR]], align 4
3756 // CHECK3-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
3757 // CHECK3-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
3758 // CHECK3-NEXT: store i64 [[DOTCAPTURE_EXPR_]], i64* [[DOTCAPTURE_EXPR__ADDR]], align 8
3759 // CHECK3-NEXT: [[CONV:%.*]] = bitcast i64* [[DOTCAPTURE_EXPR__ADDR]] to i8*
3760 // CHECK3-NEXT: store i32 0, i32* [[DOTOMP_COMB_LB]], align 4
3761 // CHECK3-NEXT: store i32 99, i32* [[DOTOMP_COMB_UB]], align 4
3762 // CHECK3-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4
3763 // CHECK3-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4
3764 // CHECK3-NEXT: [[TMP0:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
3765 // CHECK3-NEXT: [[TMP1:%.*]] = load i32, i32* [[TMP0]], align 4
3766 // CHECK3-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB1]], i32 [[TMP1]], i32 92, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_COMB_LB]], i32* [[DOTOMP_COMB_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1)
3767 // CHECK3-NEXT: [[TMP2:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
3768 // CHECK3-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP2]], 99
3769 // CHECK3-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
3770 // CHECK3: cond.true:
3771 // CHECK3-NEXT: br label [[COND_END:%.*]]
3772 // CHECK3: cond.false:
3773 // CHECK3-NEXT: [[TMP3:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
3774 // CHECK3-NEXT: br label [[COND_END]]
3775 // CHECK3: cond.end:
3776 // CHECK3-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP3]], [[COND_FALSE]] ]
3777 // CHECK3-NEXT: store i32 [[COND]], i32* [[DOTOMP_COMB_UB]], align 4
3778 // CHECK3-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4
3779 // CHECK3-NEXT: store i32 [[TMP4]], i32* [[DOTOMP_IV]], align 4
3780 // CHECK3-NEXT: [[TMP5:%.*]] = load i8, i8* [[CONV]], align 8
3781 // CHECK3-NEXT: [[TOBOOL:%.*]] = trunc i8 [[TMP5]] to i1
3782 // CHECK3-NEXT: br i1 [[TOBOOL]], label [[OMP_IF_THEN:%.*]], label [[OMP_IF_ELSE6:%.*]]
3783 // CHECK3: omp_if.then:
3784 // CHECK3-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
3785 // CHECK3: omp.inner.for.cond:
3786 // CHECK3-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
3787 // CHECK3-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
3788 // CHECK3-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP6]], [[TMP7]]
3789 // CHECK3-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
3790 // CHECK3: omp.inner.for.body:
3791 // CHECK3-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4
3792 // CHECK3-NEXT: [[TMP9:%.*]] = zext i32 [[TMP8]] to i64
3793 // CHECK3-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
3794 // CHECK3-NEXT: [[TMP11:%.*]] = zext i32 [[TMP10]] to i64
3795 // CHECK3-NEXT: [[TMP12:%.*]] = load i8, i8* [[CONV]], align 8
3796 // CHECK3-NEXT: [[TOBOOL2:%.*]] = trunc i8 [[TMP12]] to i1
3797 // CHECK3-NEXT: [[CONV3:%.*]] = bitcast i64* [[DOTCAPTURE_EXPR__CASTED]] to i8*
3798 // CHECK3-NEXT: [[FROMBOOL:%.*]] = zext i1 [[TOBOOL2]] to i8
3799 // CHECK3-NEXT: store i8 [[FROMBOOL]], i8* [[CONV3]], align 1
3800 // CHECK3-NEXT: [[TMP13:%.*]] = load i64, i64* [[DOTCAPTURE_EXPR__CASTED]], align 8
3801 // CHECK3-NEXT: [[TMP14:%.*]] = load i8, i8* [[CONV]], align 8
3802 // CHECK3-NEXT: [[TOBOOL4:%.*]] = trunc i8 [[TMP14]] to i1
3803 // CHECK3-NEXT: br i1 [[TOBOOL4]], label [[OMP_IF_THEN5:%.*]], label [[OMP_IF_ELSE:%.*]]
3804 // CHECK3: omp_if.then5:
3805 // CHECK3-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB3]], i32 3, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64, i64, i64)* @.omp_outlined..9 to void (i32*, i32*, ...)*), i64 [[TMP9]], i64 [[TMP11]], i64 [[TMP13]])
3806 // CHECK3-NEXT: br label [[OMP_IF_END:%.*]]
3807 // CHECK3: omp_if.else:
3808 // CHECK3-NEXT: call void @__kmpc_serialized_parallel(%struct.ident_t* @[[GLOB3]], i32 [[TMP1]])
3809 // CHECK3-NEXT: [[TMP15:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
3810 // CHECK3-NEXT: call void @.omp_outlined..9(i32* [[TMP15]], i32* [[DOTBOUND_ZERO_ADDR]], i64 [[TMP9]], i64 [[TMP11]], i64 [[TMP13]]) #[[ATTR2]]
3811 // CHECK3-NEXT: call void @__kmpc_end_serialized_parallel(%struct.ident_t* @[[GLOB3]], i32 [[TMP1]])
3812 // CHECK3-NEXT: br label [[OMP_IF_END]]
3813 // CHECK3: omp_if.end:
3814 // CHECK3-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
3815 // CHECK3: omp.inner.for.inc:
3816 // CHECK3-NEXT: [[TMP16:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
3817 // CHECK3-NEXT: [[TMP17:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4
3818 // CHECK3-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP16]], [[TMP17]]
3819 // CHECK3-NEXT: store i32 [[ADD]], i32* [[DOTOMP_IV]], align 4
3820 // CHECK3-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP22:![0-9]+]]
3821 // CHECK3: omp.inner.for.end:
3822 // CHECK3-NEXT: br label [[OMP_IF_END22:%.*]]
3823 // CHECK3: omp_if.else6:
3824 // CHECK3-NEXT: br label [[OMP_INNER_FOR_COND7:%.*]]
3825 // CHECK3: omp.inner.for.cond7:
3826 // CHECK3-NEXT: [[TMP18:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
3827 // CHECK3-NEXT: [[TMP19:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
3828 // CHECK3-NEXT: [[CMP8:%.*]] = icmp sle i32 [[TMP18]], [[TMP19]]
3829 // CHECK3-NEXT: br i1 [[CMP8]], label [[OMP_INNER_FOR_BODY9:%.*]], label [[OMP_INNER_FOR_END21:%.*]]
3830 // CHECK3: omp.inner.for.body9:
3831 // CHECK3-NEXT: [[TMP20:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4
3832 // CHECK3-NEXT: [[TMP21:%.*]] = zext i32 [[TMP20]] to i64
3833 // CHECK3-NEXT: [[TMP22:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
3834 // CHECK3-NEXT: [[TMP23:%.*]] = zext i32 [[TMP22]] to i64
3835 // CHECK3-NEXT: [[TMP24:%.*]] = load i8, i8* [[CONV]], align 8
3836 // CHECK3-NEXT: [[TOBOOL10:%.*]] = trunc i8 [[TMP24]] to i1
3837 // CHECK3-NEXT: [[CONV12:%.*]] = bitcast i64* [[DOTCAPTURE_EXPR__CASTED11]] to i8*
3838 // CHECK3-NEXT: [[FROMBOOL13:%.*]] = zext i1 [[TOBOOL10]] to i8
3839 // CHECK3-NEXT: store i8 [[FROMBOOL13]], i8* [[CONV12]], align 1
3840 // CHECK3-NEXT: [[TMP25:%.*]] = load i64, i64* [[DOTCAPTURE_EXPR__CASTED11]], align 8
3841 // CHECK3-NEXT: [[TMP26:%.*]] = load i8, i8* [[CONV]], align 8
3842 // CHECK3-NEXT: [[TOBOOL14:%.*]] = trunc i8 [[TMP26]] to i1
3843 // CHECK3-NEXT: br i1 [[TOBOOL14]], label [[OMP_IF_THEN15:%.*]], label [[OMP_IF_ELSE16:%.*]]
3844 // CHECK3: omp_if.then15:
3845 // CHECK3-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB3]], i32 3, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64, i64, i64)* @.omp_outlined..10 to void (i32*, i32*, ...)*), i64 [[TMP21]], i64 [[TMP23]], i64 [[TMP25]])
3846 // CHECK3-NEXT: br label [[OMP_IF_END18:%.*]]
3847 // CHECK3: omp_if.else16:
3848 // CHECK3-NEXT: call void @__kmpc_serialized_parallel(%struct.ident_t* @[[GLOB3]], i32 [[TMP1]])
3849 // CHECK3-NEXT: [[TMP27:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
3850 // CHECK3-NEXT: call void @.omp_outlined..10(i32* [[TMP27]], i32* [[DOTBOUND_ZERO_ADDR17]], i64 [[TMP21]], i64 [[TMP23]], i64 [[TMP25]]) #[[ATTR2]]
3851 // CHECK3-NEXT: call void @__kmpc_end_serialized_parallel(%struct.ident_t* @[[GLOB3]], i32 [[TMP1]])
3852 // CHECK3-NEXT: br label [[OMP_IF_END18]]
3853 // CHECK3: omp_if.end18:
3854 // CHECK3-NEXT: br label [[OMP_INNER_FOR_INC19:%.*]]
3855 // CHECK3: omp.inner.for.inc19:
3856 // CHECK3-NEXT: [[TMP28:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
3857 // CHECK3-NEXT: [[TMP29:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4
3858 // CHECK3-NEXT: [[ADD20:%.*]] = add nsw i32 [[TMP28]], [[TMP29]]
3859 // CHECK3-NEXT: store i32 [[ADD20]], i32* [[DOTOMP_IV]], align 4
3860 // CHECK3-NEXT: br label [[OMP_INNER_FOR_COND7]], !llvm.loop [[LOOP23:![0-9]+]]
3861 // CHECK3: omp.inner.for.end21:
3862 // CHECK3-NEXT: br label [[OMP_IF_END22]]
3863 // CHECK3: omp_if.end22:
3864 // CHECK3-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
3865 // CHECK3: omp.loop.exit:
3866 // CHECK3-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP1]])
3867 // CHECK3-NEXT: [[TMP30:%.*]] = load i32, i32* [[DOTOMP_IS_LAST]], align 4
3868 // CHECK3-NEXT: [[TMP31:%.*]] = icmp ne i32 [[TMP30]], 0
3869 // CHECK3-NEXT: br i1 [[TMP31]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
3870 // CHECK3: .omp.final.then:
3871 // CHECK3-NEXT: store i32 100, i32* [[I]], align 4
3872 // CHECK3-NEXT: br label [[DOTOMP_FINAL_DONE]]
3873 // CHECK3: .omp.final.done:
3874 // CHECK3-NEXT: ret void
3875 //
3876 //
3877 // CHECK3-LABEL: define {{[^@]+}}@.omp_outlined..9
3878 // CHECK3-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[DOTPREVIOUS_LB_:%.*]], i64 [[DOTPREVIOUS_UB_:%.*]], i64 [[DOTCAPTURE_EXPR_:%.*]]) #[[ATTR1]] {
3879 // CHECK3-NEXT: entry:
3880 // CHECK3-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
3881 // CHECK3-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
3882 // CHECK3-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i64, align 8
3883 // CHECK3-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i64, align 8
3884 // CHECK3-NEXT: [[DOTCAPTURE_EXPR__ADDR:%.*]] = alloca i64, align 8
3885 // CHECK3-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
3886 // CHECK3-NEXT: [[TMP:%.*]] = alloca i32, align 4
3887 // CHECK3-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4
3888 // CHECK3-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4
3889 // CHECK3-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
3890 // CHECK3-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
3891 // CHECK3-NEXT: [[I:%.*]] = alloca i32, align 4
3892 // CHECK3-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
3893 // CHECK3-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
3894 // CHECK3-NEXT: store i64 [[DOTPREVIOUS_LB_]], i64* [[DOTPREVIOUS_LB__ADDR]], align 8
3895 // CHECK3-NEXT: store i64 [[DOTPREVIOUS_UB_]], i64* [[DOTPREVIOUS_UB__ADDR]], align 8
3896 // CHECK3-NEXT: store i64 [[DOTCAPTURE_EXPR_]], i64* [[DOTCAPTURE_EXPR__ADDR]], align 8
3897 // CHECK3-NEXT: [[CONV:%.*]] = bitcast i64* [[DOTCAPTURE_EXPR__ADDR]] to i8*
3898 // CHECK3-NEXT: store i32 0, i32* [[DOTOMP_LB]], align 4
3899 // CHECK3-NEXT: store i32 99, i32* [[DOTOMP_UB]], align 4
3900 // CHECK3-NEXT: [[TMP0:%.*]] = load i64, i64* [[DOTPREVIOUS_LB__ADDR]], align 8
3901 // CHECK3-NEXT: [[CONV1:%.*]] = trunc i64 [[TMP0]] to i32
3902 // CHECK3-NEXT: [[TMP1:%.*]] = load i64, i64* [[DOTPREVIOUS_UB__ADDR]], align 8
3903 // CHECK3-NEXT: [[CONV2:%.*]] = trunc i64 [[TMP1]] to i32
3904 // CHECK3-NEXT: store i32 [[CONV1]], i32* [[DOTOMP_LB]], align 4
3905 // CHECK3-NEXT: store i32 [[CONV2]], i32* [[DOTOMP_UB]], align 4
3906 // CHECK3-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4
3907 // CHECK3-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4
3908 // CHECK3-NEXT: [[TMP2:%.*]] = load i8, i8* [[CONV]], align 8
3909 // CHECK3-NEXT: [[TOBOOL:%.*]] = trunc i8 [[TMP2]] to i1
3910 // CHECK3-NEXT: br i1 [[TOBOOL]], label [[OMP_IF_THEN:%.*]], label [[OMP_IF_ELSE:%.*]]
3911 // CHECK3: omp_if.then:
3912 // CHECK3-NEXT: [[TMP3:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
3913 // CHECK3-NEXT: [[TMP4:%.*]] = load i32, i32* [[TMP3]], align 4
3914 // CHECK3-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB2]], i32 [[TMP4]], i32 34, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_LB]], i32* [[DOTOMP_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1)
3915 // CHECK3-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
3916 // CHECK3-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP5]], 99
3917 // CHECK3-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
3918 // CHECK3: cond.true:
3919 // CHECK3-NEXT: br label [[COND_END:%.*]]
3920 // CHECK3: cond.false:
3921 // CHECK3-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
3922 // CHECK3-NEXT: br label [[COND_END]]
3923 // CHECK3: cond.end:
3924 // CHECK3-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP6]], [[COND_FALSE]] ]
3925 // CHECK3-NEXT: store i32 [[COND]], i32* [[DOTOMP_UB]], align 4
3926 // CHECK3-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4
3927 // CHECK3-NEXT: store i32 [[TMP7]], i32* [[DOTOMP_IV]], align 4
3928 // CHECK3-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
3929 // CHECK3: omp.inner.for.cond:
3930 // CHECK3-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
3931 // CHECK3-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
3932 // CHECK3-NEXT: [[CMP3:%.*]] = icmp sle i32 [[TMP8]], [[TMP9]]
3933 // CHECK3-NEXT: br i1 [[CMP3]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
3934 // CHECK3: omp.inner.for.body:
3935 // CHECK3-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
3936 // CHECK3-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP10]], 1
3937 // CHECK3-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
3938 // CHECK3-NEXT: store i32 [[ADD]], i32* [[I]], align 4
3939 // CHECK3-NEXT: call void @_Z3fn6v()
3940 // CHECK3-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
3941 // CHECK3: omp.body.continue:
3942 // CHECK3-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
3943 // CHECK3: omp.inner.for.inc:
3944 // CHECK3-NEXT: [[TMP11:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
3945 // CHECK3-NEXT: [[ADD4:%.*]] = add nsw i32 [[TMP11]], 1
3946 // CHECK3-NEXT: store i32 [[ADD4]], i32* [[DOTOMP_IV]], align 4
3947 // CHECK3-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP24:![0-9]+]]
3948 // CHECK3: omp.inner.for.end:
3949 // CHECK3-NEXT: br label [[OMP_IF_END:%.*]]
3950 // CHECK3: omp_if.else:
3951 // CHECK3-NEXT: [[TMP12:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
3952 // CHECK3-NEXT: [[TMP13:%.*]] = load i32, i32* [[TMP12]], align 4
3953 // CHECK3-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB2]], i32 [[TMP13]], i32 34, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_LB]], i32* [[DOTOMP_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1)
3954 // CHECK3-NEXT: [[TMP14:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
3955 // CHECK3-NEXT: [[CMP5:%.*]] = icmp sgt i32 [[TMP14]], 99
3956 // CHECK3-NEXT: br i1 [[CMP5]], label [[COND_TRUE6:%.*]], label [[COND_FALSE7:%.*]]
3957 // CHECK3: cond.true6:
3958 // CHECK3-NEXT: br label [[COND_END8:%.*]]
3959 // CHECK3: cond.false7:
3960 // CHECK3-NEXT: [[TMP15:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
3961 // CHECK3-NEXT: br label [[COND_END8]]
3962 // CHECK3: cond.end8:
3963 // CHECK3-NEXT: [[COND9:%.*]] = phi i32 [ 99, [[COND_TRUE6]] ], [ [[TMP15]], [[COND_FALSE7]] ]
3964 // CHECK3-NEXT: store i32 [[COND9]], i32* [[DOTOMP_UB]], align 4
3965 // CHECK3-NEXT: [[TMP16:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4
3966 // CHECK3-NEXT: store i32 [[TMP16]], i32* [[DOTOMP_IV]], align 4
3967 // CHECK3-NEXT: br label [[OMP_INNER_FOR_COND10:%.*]]
3968 // CHECK3: omp.inner.for.cond10:
3969 // CHECK3-NEXT: [[TMP17:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
3970 // CHECK3-NEXT: [[TMP18:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
3971 // CHECK3-NEXT: [[CMP11:%.*]] = icmp sle i32 [[TMP17]], [[TMP18]]
3972 // CHECK3-NEXT: br i1 [[CMP11]], label [[OMP_INNER_FOR_BODY12:%.*]], label [[OMP_INNER_FOR_END18:%.*]]
3973 // CHECK3: omp.inner.for.body12:
3974 // CHECK3-NEXT: [[TMP19:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
3975 // CHECK3-NEXT: [[MUL13:%.*]] = mul nsw i32 [[TMP19]], 1
3976 // CHECK3-NEXT: [[ADD14:%.*]] = add nsw i32 0, [[MUL13]]
3977 // CHECK3-NEXT: store i32 [[ADD14]], i32* [[I]], align 4
3978 // CHECK3-NEXT: call void @_Z3fn6v()
3979 // CHECK3-NEXT: br label [[OMP_BODY_CONTINUE15:%.*]]
3980 // CHECK3: omp.body.continue15:
3981 // CHECK3-NEXT: br label [[OMP_INNER_FOR_INC16:%.*]]
3982 // CHECK3: omp.inner.for.inc16:
3983 // CHECK3-NEXT: [[TMP20:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
3984 // CHECK3-NEXT: [[ADD17:%.*]] = add nsw i32 [[TMP20]], 1
3985 // CHECK3-NEXT: store i32 [[ADD17]], i32* [[DOTOMP_IV]], align 4
3986 // CHECK3-NEXT: br label [[OMP_INNER_FOR_COND10]], !llvm.loop [[LOOP25:![0-9]+]]
3987 // CHECK3: omp.inner.for.end18:
3988 // CHECK3-NEXT: br label [[OMP_IF_END]]
3989 // CHECK3: omp_if.end:
3990 // CHECK3-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
3991 // CHECK3: omp.loop.exit:
3992 // CHECK3-NEXT: [[TMP21:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
3993 // CHECK3-NEXT: [[TMP22:%.*]] = load i32, i32* [[TMP21]], align 4
3994 // CHECK3-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP22]])
3995 // CHECK3-NEXT: [[TMP23:%.*]] = load i32, i32* [[DOTOMP_IS_LAST]], align 4
3996 // CHECK3-NEXT: [[TMP24:%.*]] = icmp ne i32 [[TMP23]], 0
3997 // CHECK3-NEXT: br i1 [[TMP24]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
3998 // CHECK3: .omp.final.then:
3999 // CHECK3-NEXT: store i32 100, i32* [[I]], align 4
4000 // CHECK3-NEXT: br label [[DOTOMP_FINAL_DONE]]
4001 // CHECK3: .omp.final.done:
4002 // CHECK3-NEXT: ret void
4003 //
4004 //
4005 // CHECK3-LABEL: define {{[^@]+}}@.omp_outlined..10
4006 // CHECK3-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[DOTPREVIOUS_LB_:%.*]], i64 [[DOTPREVIOUS_UB_:%.*]], i64 [[DOTCAPTURE_EXPR_:%.*]]) #[[ATTR1]] {
4007 // CHECK3-NEXT: entry:
4008 // CHECK3-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
4009 // CHECK3-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
4010 // CHECK3-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i64, align 8
4011 // CHECK3-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i64, align 8
4012 // CHECK3-NEXT: [[DOTCAPTURE_EXPR__ADDR:%.*]] = alloca i64, align 8
4013 // CHECK3-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
4014 // CHECK3-NEXT: [[TMP:%.*]] = alloca i32, align 4
4015 // CHECK3-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4
4016 // CHECK3-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4
4017 // CHECK3-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
4018 // CHECK3-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
4019 // CHECK3-NEXT: [[I:%.*]] = alloca i32, align 4
4020 // CHECK3-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
4021 // CHECK3-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
4022 // CHECK3-NEXT: store i64 [[DOTPREVIOUS_LB_]], i64* [[DOTPREVIOUS_LB__ADDR]], align 8
4023 // CHECK3-NEXT: store i64 [[DOTPREVIOUS_UB_]], i64* [[DOTPREVIOUS_UB__ADDR]], align 8
4024 // CHECK3-NEXT: store i64 [[DOTCAPTURE_EXPR_]], i64* [[DOTCAPTURE_EXPR__ADDR]], align 8
4025 // CHECK3-NEXT: [[CONV:%.*]] = bitcast i64* [[DOTCAPTURE_EXPR__ADDR]] to i8*
4026 // CHECK3-NEXT: store i32 0, i32* [[DOTOMP_LB]], align 4
4027 // CHECK3-NEXT: store i32 99, i32* [[DOTOMP_UB]], align 4
4028 // CHECK3-NEXT: [[TMP0:%.*]] = load i64, i64* [[DOTPREVIOUS_LB__ADDR]], align 8
4029 // CHECK3-NEXT: [[CONV1:%.*]] = trunc i64 [[TMP0]] to i32
4030 // CHECK3-NEXT: [[TMP1:%.*]] = load i64, i64* [[DOTPREVIOUS_UB__ADDR]], align 8
4031 // CHECK3-NEXT: [[CONV2:%.*]] = trunc i64 [[TMP1]] to i32
4032 // CHECK3-NEXT: store i32 [[CONV1]], i32* [[DOTOMP_LB]], align 4
4033 // CHECK3-NEXT: store i32 [[CONV2]], i32* [[DOTOMP_UB]], align 4
4034 // CHECK3-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4
4035 // CHECK3-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4
4036 // CHECK3-NEXT: [[TMP2:%.*]] = load i8, i8* [[CONV]], align 8
4037 // CHECK3-NEXT: [[TOBOOL:%.*]] = trunc i8 [[TMP2]] to i1
4038 // CHECK3-NEXT: br i1 [[TOBOOL]], label [[OMP_IF_THEN:%.*]], label [[OMP_IF_ELSE:%.*]]
4039 // CHECK3: omp_if.then:
4040 // CHECK3-NEXT: [[TMP3:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
4041 // CHECK3-NEXT: [[TMP4:%.*]] = load i32, i32* [[TMP3]], align 4
4042 // CHECK3-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB2]], i32 [[TMP4]], i32 34, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_LB]], i32* [[DOTOMP_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1)
4043 // CHECK3-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
4044 // CHECK3-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP5]], 99
4045 // CHECK3-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
4046 // CHECK3: cond.true:
4047 // CHECK3-NEXT: br label [[COND_END:%.*]]
4048 // CHECK3: cond.false:
4049 // CHECK3-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
4050 // CHECK3-NEXT: br label [[COND_END]]
4051 // CHECK3: cond.end:
4052 // CHECK3-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP6]], [[COND_FALSE]] ]
4053 // CHECK3-NEXT: store i32 [[COND]], i32* [[DOTOMP_UB]], align 4
4054 // CHECK3-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4
4055 // CHECK3-NEXT: store i32 [[TMP7]], i32* [[DOTOMP_IV]], align 4
4056 // CHECK3-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
4057 // CHECK3: omp.inner.for.cond:
4058 // CHECK3-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
4059 // CHECK3-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
4060 // CHECK3-NEXT: [[CMP3:%.*]] = icmp sle i32 [[TMP8]], [[TMP9]]
4061 // CHECK3-NEXT: br i1 [[CMP3]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
4062 // CHECK3: omp.inner.for.body:
4063 // CHECK3-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
4064 // CHECK3-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP10]], 1
4065 // CHECK3-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
4066 // CHECK3-NEXT: store i32 [[ADD]], i32* [[I]], align 4
4067 // CHECK3-NEXT: call void @_Z3fn6v()
4068 // CHECK3-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
4069 // CHECK3: omp.body.continue:
4070 // CHECK3-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
4071 // CHECK3: omp.inner.for.inc:
4072 // CHECK3-NEXT: [[TMP11:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
4073 // CHECK3-NEXT: [[ADD4:%.*]] = add nsw i32 [[TMP11]], 1
4074 // CHECK3-NEXT: store i32 [[ADD4]], i32* [[DOTOMP_IV]], align 4
4075 // CHECK3-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP26:![0-9]+]]
4076 // CHECK3: omp.inner.for.end:
4077 // CHECK3-NEXT: br label [[OMP_IF_END:%.*]]
4078 // CHECK3: omp_if.else:
4079 // CHECK3-NEXT: [[TMP12:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
4080 // CHECK3-NEXT: [[TMP13:%.*]] = load i32, i32* [[TMP12]], align 4
4081 // CHECK3-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB2]], i32 [[TMP13]], i32 34, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_LB]], i32* [[DOTOMP_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1)
4082 // CHECK3-NEXT: [[TMP14:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
4083 // CHECK3-NEXT: [[CMP5:%.*]] = icmp sgt i32 [[TMP14]], 99
4084 // CHECK3-NEXT: br i1 [[CMP5]], label [[COND_TRUE6:%.*]], label [[COND_FALSE7:%.*]]
4085 // CHECK3: cond.true6:
4086 // CHECK3-NEXT: br label [[COND_END8:%.*]]
4087 // CHECK3: cond.false7:
4088 // CHECK3-NEXT: [[TMP15:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
4089 // CHECK3-NEXT: br label [[COND_END8]]
4090 // CHECK3: cond.end8:
4091 // CHECK3-NEXT: [[COND9:%.*]] = phi i32 [ 99, [[COND_TRUE6]] ], [ [[TMP15]], [[COND_FALSE7]] ]
4092 // CHECK3-NEXT: store i32 [[COND9]], i32* [[DOTOMP_UB]], align 4
4093 // CHECK3-NEXT: [[TMP16:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4
4094 // CHECK3-NEXT: store i32 [[TMP16]], i32* [[DOTOMP_IV]], align 4
4095 // CHECK3-NEXT: br label [[OMP_INNER_FOR_COND10:%.*]]
4096 // CHECK3: omp.inner.for.cond10:
4097 // CHECK3-NEXT: [[TMP17:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
4098 // CHECK3-NEXT: [[TMP18:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
4099 // CHECK3-NEXT: [[CMP11:%.*]] = icmp sle i32 [[TMP17]], [[TMP18]]
4100 // CHECK3-NEXT: br i1 [[CMP11]], label [[OMP_INNER_FOR_BODY12:%.*]], label [[OMP_INNER_FOR_END18:%.*]]
4101 // CHECK3: omp.inner.for.body12:
4102 // CHECK3-NEXT: [[TMP19:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
4103 // CHECK3-NEXT: [[MUL13:%.*]] = mul nsw i32 [[TMP19]], 1
4104 // CHECK3-NEXT: [[ADD14:%.*]] = add nsw i32 0, [[MUL13]]
4105 // CHECK3-NEXT: store i32 [[ADD14]], i32* [[I]], align 4
4106 // CHECK3-NEXT: call void @_Z3fn6v()
4107 // CHECK3-NEXT: br label [[OMP_BODY_CONTINUE15:%.*]]
4108 // CHECK3: omp.body.continue15:
4109 // CHECK3-NEXT: br label [[OMP_INNER_FOR_INC16:%.*]]
4110 // CHECK3: omp.inner.for.inc16:
4111 // CHECK3-NEXT: [[TMP20:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
4112 // CHECK3-NEXT: [[ADD17:%.*]] = add nsw i32 [[TMP20]], 1
4113 // CHECK3-NEXT: store i32 [[ADD17]], i32* [[DOTOMP_IV]], align 4
4114 // CHECK3-NEXT: br label [[OMP_INNER_FOR_COND10]], !llvm.loop [[LOOP27:![0-9]+]]
4115 // CHECK3: omp.inner.for.end18:
4116 // CHECK3-NEXT: br label [[OMP_IF_END]]
4117 // CHECK3: omp_if.end:
4118 // CHECK3-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
4119 // CHECK3: omp.loop.exit:
4120 // CHECK3-NEXT: [[TMP21:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
4121 // CHECK3-NEXT: [[TMP22:%.*]] = load i32, i32* [[TMP21]], align 4
4122 // CHECK3-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP22]])
4123 // CHECK3-NEXT: [[TMP23:%.*]] = load i32, i32* [[DOTOMP_IS_LAST]], align 4
4124 // CHECK3-NEXT: [[TMP24:%.*]] = icmp ne i32 [[TMP23]], 0
4125 // CHECK3-NEXT: br i1 [[TMP24]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
4126 // CHECK3: .omp.final.then:
4127 // CHECK3-NEXT: store i32 100, i32* [[I]], align 4
4128 // CHECK3-NEXT: br label [[DOTOMP_FINAL_DONE]]
4129 // CHECK3: .omp.final.done:
4130 // CHECK3-NEXT: ret void
4131 //
4132 //
4133 // CHECK3-LABEL: define {{[^@]+}}@_Z5tmainIiEiT_
4134 // CHECK3-SAME: (i32 [[ARG:%.*]]) #[[ATTR0]] comdat {
4135 // CHECK3-NEXT: entry:
4136 // CHECK3-NEXT: [[ARG_ADDR:%.*]] = alloca i32, align 4
4137 // CHECK3-NEXT: [[TMP:%.*]] = alloca i32, align 4
4138 // CHECK3-NEXT: [[_TMP1:%.*]] = alloca i32, align 4
4139 // CHECK3-NEXT: [[ARG_CASTED:%.*]] = alloca i64, align 8
4140 // CHECK3-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [1 x i8*], align 8
4141 // CHECK3-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [1 x i8*], align 8
4142 // CHECK3-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [1 x i8*], align 8
4143 // CHECK3-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i8, align 1
4144 // CHECK3-NEXT: [[_TMP5:%.*]] = alloca i32, align 4
4145 // CHECK3-NEXT: store i32 [[ARG]], i32* [[ARG_ADDR]], align 4
4146 // CHECK3-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i64 100)
4147 // CHECK3-NEXT: [[TMP0:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiEiT__l57.region_id, i32 0, i8** null, i8** null, i64* null, i64* null, i8** null, i8** null, i32 0, i32 0)
4148 // CHECK3-NEXT: [[TMP1:%.*]] = icmp ne i32 [[TMP0]], 0
4149 // CHECK3-NEXT: br i1 [[TMP1]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
4150 // CHECK3: omp_offload.failed:
4151 // CHECK3-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiEiT__l57() #[[ATTR2]]
4152 // CHECK3-NEXT: br label [[OMP_OFFLOAD_CONT]]
4153 // CHECK3: omp_offload.cont:
4154 // CHECK3-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i64 100)
4155 // CHECK3-NEXT: [[TMP2:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiEiT__l62.region_id, i32 0, i8** null, i8** null, i64* null, i64* null, i8** null, i8** null, i32 0, i32 1)
4156 // CHECK3-NEXT: [[TMP3:%.*]] = icmp ne i32 [[TMP2]], 0
4157 // CHECK3-NEXT: br i1 [[TMP3]], label [[OMP_OFFLOAD_FAILED2:%.*]], label [[OMP_OFFLOAD_CONT3:%.*]]
4158 // CHECK3: omp_offload.failed2:
4159 // CHECK3-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiEiT__l62() #[[ATTR2]]
4160 // CHECK3-NEXT: br label [[OMP_OFFLOAD_CONT3]]
4161 // CHECK3: omp_offload.cont3:
4162 // CHECK3-NEXT: [[TMP4:%.*]] = load i32, i32* [[ARG_ADDR]], align 4
4163 // CHECK3-NEXT: [[CONV:%.*]] = bitcast i64* [[ARG_CASTED]] to i32*
4164 // CHECK3-NEXT: store i32 [[TMP4]], i32* [[CONV]], align 4
4165 // CHECK3-NEXT: [[TMP5:%.*]] = load i64, i64* [[ARG_CASTED]], align 8
4166 // CHECK3-NEXT: [[TMP6:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
4167 // CHECK3-NEXT: [[TMP7:%.*]] = bitcast i8** [[TMP6]] to i64*
4168 // CHECK3-NEXT: store i64 [[TMP5]], i64* [[TMP7]], align 8
4169 // CHECK3-NEXT: [[TMP8:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
4170 // CHECK3-NEXT: [[TMP9:%.*]] = bitcast i8** [[TMP8]] to i64*
4171 // CHECK3-NEXT: store i64 [[TMP5]], i64* [[TMP9]], align 8
4172 // CHECK3-NEXT: [[TMP10:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0
4173 // CHECK3-NEXT: store i8* null, i8** [[TMP10]], align 8
4174 // CHECK3-NEXT: [[TMP11:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
4175 // CHECK3-NEXT: [[TMP12:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
4176 // CHECK3-NEXT: [[TMP13:%.*]] = load i32, i32* [[ARG_ADDR]], align 4
4177 // CHECK3-NEXT: [[TOBOOL:%.*]] = icmp ne i32 [[TMP13]], 0
4178 // CHECK3-NEXT: [[FROMBOOL:%.*]] = zext i1 [[TOBOOL]] to i8
4179 // CHECK3-NEXT: store i8 [[FROMBOOL]], i8* [[DOTCAPTURE_EXPR_]], align 1
4180 // CHECK3-NEXT: [[TMP14:%.*]] = load i8, i8* [[DOTCAPTURE_EXPR_]], align 1
4181 // CHECK3-NEXT: [[TOBOOL4:%.*]] = trunc i8 [[TMP14]] to i1
4182 // CHECK3-NEXT: [[TMP15:%.*]] = select i1 [[TOBOOL4]], i32 0, i32 1
4183 // CHECK3-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i64 100)
4184 // CHECK3-NEXT: [[TMP16:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiEiT__l67.region_id, i32 1, i8** [[TMP11]], i8** [[TMP12]], i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_sizes.17, i32 0, i32 0), i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_maptypes.18, i32 0, i32 0), i8** null, i8** null, i32 0, i32 [[TMP15]])
4185 // CHECK3-NEXT: [[TMP17:%.*]] = icmp ne i32 [[TMP16]], 0
4186 // CHECK3-NEXT: br i1 [[TMP17]], label [[OMP_OFFLOAD_FAILED6:%.*]], label [[OMP_OFFLOAD_CONT7:%.*]]
4187 // CHECK3: omp_offload.failed6:
4188 // CHECK3-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiEiT__l67(i64 [[TMP5]]) #[[ATTR2]]
4189 // CHECK3-NEXT: br label [[OMP_OFFLOAD_CONT7]]
4190 // CHECK3: omp_offload.cont7:
4191 // CHECK3-NEXT: ret i32 0
4192 //
4193 //
4194 // CHECK3-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiEiT__l57
4195 // CHECK3-SAME: () #[[ATTR1]] {
4196 // CHECK3-NEXT: entry:
4197 // CHECK3-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 0, void (i32*, i32*, ...)* bitcast (void (i32*, i32*)* @.omp_outlined..11 to void (i32*, i32*, ...)*))
4198 // CHECK3-NEXT: ret void
4199 //
4200 //
4201 // CHECK3-LABEL: define {{[^@]+}}@.omp_outlined..11
4202 // CHECK3-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]]) #[[ATTR1]] {
4203 // CHECK3-NEXT: entry:
4204 // CHECK3-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
4205 // CHECK3-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
4206 // CHECK3-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
4207 // CHECK3-NEXT: [[TMP:%.*]] = alloca i32, align 4
4208 // CHECK3-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4
4209 // CHECK3-NEXT: [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4
4210 // CHECK3-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
4211 // CHECK3-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
4212 // CHECK3-NEXT: [[I:%.*]] = alloca i32, align 4
4213 // CHECK3-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
4214 // CHECK3-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
4215 // CHECK3-NEXT: store i32 0, i32* [[DOTOMP_COMB_LB]], align 4
4216 // CHECK3-NEXT: store i32 99, i32* [[DOTOMP_COMB_UB]], align 4
4217 // CHECK3-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4
4218 // CHECK3-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4
4219 // CHECK3-NEXT: [[TMP0:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
4220 // CHECK3-NEXT: [[TMP1:%.*]] = load i32, i32* [[TMP0]], align 4
4221 // CHECK3-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB1]], i32 [[TMP1]], i32 92, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_COMB_LB]], i32* [[DOTOMP_COMB_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1)
4222 // CHECK3-NEXT: [[TMP2:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
4223 // CHECK3-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP2]], 99
4224 // CHECK3-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
4225 // CHECK3: cond.true:
4226 // CHECK3-NEXT: br label [[COND_END:%.*]]
4227 // CHECK3: cond.false:
4228 // CHECK3-NEXT: [[TMP3:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
4229 // CHECK3-NEXT: br label [[COND_END]]
4230 // CHECK3: cond.end:
4231 // CHECK3-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP3]], [[COND_FALSE]] ]
4232 // CHECK3-NEXT: store i32 [[COND]], i32* [[DOTOMP_COMB_UB]], align 4
4233 // CHECK3-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4
4234 // CHECK3-NEXT: store i32 [[TMP4]], i32* [[DOTOMP_IV]], align 4
4235 // CHECK3-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
4236 // CHECK3: omp.inner.for.cond:
4237 // CHECK3-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
4238 // CHECK3-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
4239 // CHECK3-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP5]], [[TMP6]]
4240 // CHECK3-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
4241 // CHECK3: omp.inner.for.body:
4242 // CHECK3-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4
4243 // CHECK3-NEXT: [[TMP8:%.*]] = zext i32 [[TMP7]] to i64
4244 // CHECK3-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
4245 // CHECK3-NEXT: [[TMP10:%.*]] = zext i32 [[TMP9]] to i64
4246 // CHECK3-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB3]], i32 2, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64, i64)* @.omp_outlined..12 to void (i32*, i32*, ...)*), i64 [[TMP8]], i64 [[TMP10]])
4247 // CHECK3-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
4248 // CHECK3: omp.inner.for.inc:
4249 // CHECK3-NEXT: [[TMP11:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
4250 // CHECK3-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4
4251 // CHECK3-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP11]], [[TMP12]]
4252 // CHECK3-NEXT: store i32 [[ADD]], i32* [[DOTOMP_IV]], align 4
4253 // CHECK3-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP28:![0-9]+]]
4254 // CHECK3: omp.inner.for.end:
4255 // CHECK3-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
4256 // CHECK3: omp.loop.exit:
4257 // CHECK3-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP1]])
4258 // CHECK3-NEXT: [[TMP13:%.*]] = load i32, i32* [[DOTOMP_IS_LAST]], align 4
4259 // CHECK3-NEXT: [[TMP14:%.*]] = icmp ne i32 [[TMP13]], 0
4260 // CHECK3-NEXT: br i1 [[TMP14]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
4261 // CHECK3: .omp.final.then:
4262 // CHECK3-NEXT: store i32 100, i32* [[I]], align 4
4263 // CHECK3-NEXT: br label [[DOTOMP_FINAL_DONE]]
4264 // CHECK3: .omp.final.done:
4265 // CHECK3-NEXT: ret void
4266 //
4267 //
4268 // CHECK3-LABEL: define {{[^@]+}}@.omp_outlined..12
4269 // CHECK3-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[DOTPREVIOUS_LB_:%.*]], i64 [[DOTPREVIOUS_UB_:%.*]]) #[[ATTR1]] {
4270 // CHECK3-NEXT: entry:
4271 // CHECK3-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
4272 // CHECK3-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
4273 // CHECK3-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i64, align 8
4274 // CHECK3-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i64, align 8
4275 // CHECK3-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
4276 // CHECK3-NEXT: [[TMP:%.*]] = alloca i32, align 4
4277 // CHECK3-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4
4278 // CHECK3-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4
4279 // CHECK3-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
4280 // CHECK3-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
4281 // CHECK3-NEXT: [[I:%.*]] = alloca i32, align 4
4282 // CHECK3-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
4283 // CHECK3-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
4284 // CHECK3-NEXT: store i64 [[DOTPREVIOUS_LB_]], i64* [[DOTPREVIOUS_LB__ADDR]], align 8
4285 // CHECK3-NEXT: store i64 [[DOTPREVIOUS_UB_]], i64* [[DOTPREVIOUS_UB__ADDR]], align 8
4286 // CHECK3-NEXT: store i32 0, i32* [[DOTOMP_LB]], align 4
4287 // CHECK3-NEXT: store i32 99, i32* [[DOTOMP_UB]], align 4
4288 // CHECK3-NEXT: [[TMP0:%.*]] = load i64, i64* [[DOTPREVIOUS_LB__ADDR]], align 8
4289 // CHECK3-NEXT: [[CONV:%.*]] = trunc i64 [[TMP0]] to i32
4290 // CHECK3-NEXT: [[TMP1:%.*]] = load i64, i64* [[DOTPREVIOUS_UB__ADDR]], align 8
4291 // CHECK3-NEXT: [[CONV1:%.*]] = trunc i64 [[TMP1]] to i32
4292 // CHECK3-NEXT: store i32 [[CONV]], i32* [[DOTOMP_LB]], align 4
4293 // CHECK3-NEXT: store i32 [[CONV1]], i32* [[DOTOMP_UB]], align 4
4294 // CHECK3-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4
4295 // CHECK3-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4
4296 // CHECK3-NEXT: [[TMP2:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
4297 // CHECK3-NEXT: [[TMP3:%.*]] = load i32, i32* [[TMP2]], align 4
4298 // CHECK3-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB2]], i32 [[TMP3]], i32 34, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_LB]], i32* [[DOTOMP_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1)
4299 // CHECK3-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
4300 // CHECK3-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP4]], 99
4301 // CHECK3-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
4302 // CHECK3: cond.true:
4303 // CHECK3-NEXT: br label [[COND_END:%.*]]
4304 // CHECK3: cond.false:
4305 // CHECK3-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
4306 // CHECK3-NEXT: br label [[COND_END]]
4307 // CHECK3: cond.end:
4308 // CHECK3-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP5]], [[COND_FALSE]] ]
4309 // CHECK3-NEXT: store i32 [[COND]], i32* [[DOTOMP_UB]], align 4
4310 // CHECK3-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4
4311 // CHECK3-NEXT: store i32 [[TMP6]], i32* [[DOTOMP_IV]], align 4
4312 // CHECK3-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
4313 // CHECK3: omp.inner.for.cond:
4314 // CHECK3-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
4315 // CHECK3-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
4316 // CHECK3-NEXT: [[CMP2:%.*]] = icmp sle i32 [[TMP7]], [[TMP8]]
4317 // CHECK3-NEXT: br i1 [[CMP2]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
4318 // CHECK3: omp.inner.for.body:
4319 // CHECK3-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
4320 // CHECK3-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP9]], 1
4321 // CHECK3-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
4322 // CHECK3-NEXT: store i32 [[ADD]], i32* [[I]], align 4
4323 // CHECK3-NEXT: call void @_Z3fn1v()
4324 // CHECK3-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
4325 // CHECK3: omp.body.continue:
4326 // CHECK3-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
4327 // CHECK3: omp.inner.for.inc:
4328 // CHECK3-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
4329 // CHECK3-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP10]], 1
4330 // CHECK3-NEXT: store i32 [[ADD3]], i32* [[DOTOMP_IV]], align 4
4331 // CHECK3-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP29:![0-9]+]]
4332 // CHECK3: omp.inner.for.end:
4333 // CHECK3-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
4334 // CHECK3: omp.loop.exit:
4335 // CHECK3-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP3]])
4336 // CHECK3-NEXT: [[TMP11:%.*]] = load i32, i32* [[DOTOMP_IS_LAST]], align 4
4337 // CHECK3-NEXT: [[TMP12:%.*]] = icmp ne i32 [[TMP11]], 0
4338 // CHECK3-NEXT: br i1 [[TMP12]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
4339 // CHECK3: .omp.final.then:
4340 // CHECK3-NEXT: store i32 100, i32* [[I]], align 4
4341 // CHECK3-NEXT: br label [[DOTOMP_FINAL_DONE]]
4342 // CHECK3: .omp.final.done:
4343 // CHECK3-NEXT: ret void
4344 //
4345 //
4346 // CHECK3-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiEiT__l62
4347 // CHECK3-SAME: () #[[ATTR1]] {
4348 // CHECK3-NEXT: entry:
4349 // CHECK3-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 0, void (i32*, i32*, ...)* bitcast (void (i32*, i32*)* @.omp_outlined..13 to void (i32*, i32*, ...)*))
4350 // CHECK3-NEXT: ret void
4351 //
4352 //
4353 // CHECK3-LABEL: define {{[^@]+}}@.omp_outlined..13
4354 // CHECK3-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]]) #[[ATTR1]] {
4355 // CHECK3-NEXT: entry:
4356 // CHECK3-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
4357 // CHECK3-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
4358 // CHECK3-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
4359 // CHECK3-NEXT: [[TMP:%.*]] = alloca i32, align 4
4360 // CHECK3-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4
4361 // CHECK3-NEXT: [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4
4362 // CHECK3-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
4363 // CHECK3-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
4364 // CHECK3-NEXT: [[I:%.*]] = alloca i32, align 4
4365 // CHECK3-NEXT: [[DOTBOUND_ZERO_ADDR:%.*]] = alloca i32, align 4
4366 // CHECK3-NEXT: store i32 0, i32* [[DOTBOUND_ZERO_ADDR]], align 4
4367 // CHECK3-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
4368 // CHECK3-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
4369 // CHECK3-NEXT: store i32 0, i32* [[DOTOMP_COMB_LB]], align 4
4370 // CHECK3-NEXT: store i32 99, i32* [[DOTOMP_COMB_UB]], align 4
4371 // CHECK3-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4
4372 // CHECK3-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4
4373 // CHECK3-NEXT: [[TMP0:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
4374 // CHECK3-NEXT: [[TMP1:%.*]] = load i32, i32* [[TMP0]], align 4
4375 // CHECK3-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB1]], i32 [[TMP1]], i32 92, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_COMB_LB]], i32* [[DOTOMP_COMB_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1)
4376 // CHECK3-NEXT: [[TMP2:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
4377 // CHECK3-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP2]], 99
4378 // CHECK3-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
4379 // CHECK3: cond.true:
4380 // CHECK3-NEXT: br label [[COND_END:%.*]]
4381 // CHECK3: cond.false:
4382 // CHECK3-NEXT: [[TMP3:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
4383 // CHECK3-NEXT: br label [[COND_END]]
4384 // CHECK3: cond.end:
4385 // CHECK3-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP3]], [[COND_FALSE]] ]
4386 // CHECK3-NEXT: store i32 [[COND]], i32* [[DOTOMP_COMB_UB]], align 4
4387 // CHECK3-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4
4388 // CHECK3-NEXT: store i32 [[TMP4]], i32* [[DOTOMP_IV]], align 4
4389 // CHECK3-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
4390 // CHECK3: omp.inner.for.cond:
4391 // CHECK3-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
4392 // CHECK3-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
4393 // CHECK3-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP5]], [[TMP6]]
4394 // CHECK3-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
4395 // CHECK3: omp.inner.for.body:
4396 // CHECK3-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4
4397 // CHECK3-NEXT: [[TMP8:%.*]] = zext i32 [[TMP7]] to i64
4398 // CHECK3-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
4399 // CHECK3-NEXT: [[TMP10:%.*]] = zext i32 [[TMP9]] to i64
4400 // CHECK3-NEXT: call void @__kmpc_serialized_parallel(%struct.ident_t* @[[GLOB3]], i32 [[TMP1]])
4401 // CHECK3-NEXT: [[TMP11:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
4402 // CHECK3-NEXT: call void @.omp_outlined..14(i32* [[TMP11]], i32* [[DOTBOUND_ZERO_ADDR]], i64 [[TMP8]], i64 [[TMP10]]) #[[ATTR2]]
4403 // CHECK3-NEXT: call void @__kmpc_end_serialized_parallel(%struct.ident_t* @[[GLOB3]], i32 [[TMP1]])
4404 // CHECK3-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
4405 // CHECK3: omp.inner.for.inc:
4406 // CHECK3-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
4407 // CHECK3-NEXT: [[TMP13:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4
4408 // CHECK3-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP12]], [[TMP13]]
4409 // CHECK3-NEXT: store i32 [[ADD]], i32* [[DOTOMP_IV]], align 4
4410 // CHECK3-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP30:![0-9]+]]
4411 // CHECK3: omp.inner.for.end:
4412 // CHECK3-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
4413 // CHECK3: omp.loop.exit:
4414 // CHECK3-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP1]])
4415 // CHECK3-NEXT: [[TMP14:%.*]] = load i32, i32* [[DOTOMP_IS_LAST]], align 4
4416 // CHECK3-NEXT: [[TMP15:%.*]] = icmp ne i32 [[TMP14]], 0
4417 // CHECK3-NEXT: br i1 [[TMP15]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
4418 // CHECK3: .omp.final.then:
4419 // CHECK3-NEXT: store i32 100, i32* [[I]], align 4
4420 // CHECK3-NEXT: br label [[DOTOMP_FINAL_DONE]]
4421 // CHECK3: .omp.final.done:
4422 // CHECK3-NEXT: ret void
4423 //
4424 //
4425 // CHECK3-LABEL: define {{[^@]+}}@.omp_outlined..14
4426 // CHECK3-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[DOTPREVIOUS_LB_:%.*]], i64 [[DOTPREVIOUS_UB_:%.*]]) #[[ATTR1]] {
4427 // CHECK3-NEXT: entry:
4428 // CHECK3-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
4429 // CHECK3-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
4430 // CHECK3-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i64, align 8
4431 // CHECK3-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i64, align 8
4432 // CHECK3-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
4433 // CHECK3-NEXT: [[TMP:%.*]] = alloca i32, align 4
4434 // CHECK3-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4
4435 // CHECK3-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4
4436 // CHECK3-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
4437 // CHECK3-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
4438 // CHECK3-NEXT: [[I:%.*]] = alloca i32, align 4
4439 // CHECK3-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
4440 // CHECK3-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
4441 // CHECK3-NEXT: store i64 [[DOTPREVIOUS_LB_]], i64* [[DOTPREVIOUS_LB__ADDR]], align 8
4442 // CHECK3-NEXT: store i64 [[DOTPREVIOUS_UB_]], i64* [[DOTPREVIOUS_UB__ADDR]], align 8
4443 // CHECK3-NEXT: store i32 0, i32* [[DOTOMP_LB]], align 4
4444 // CHECK3-NEXT: store i32 99, i32* [[DOTOMP_UB]], align 4
4445 // CHECK3-NEXT: [[TMP0:%.*]] = load i64, i64* [[DOTPREVIOUS_LB__ADDR]], align 8
4446 // CHECK3-NEXT: [[CONV:%.*]] = trunc i64 [[TMP0]] to i32
4447 // CHECK3-NEXT: [[TMP1:%.*]] = load i64, i64* [[DOTPREVIOUS_UB__ADDR]], align 8
4448 // CHECK3-NEXT: [[CONV1:%.*]] = trunc i64 [[TMP1]] to i32
4449 // CHECK3-NEXT: store i32 [[CONV]], i32* [[DOTOMP_LB]], align 4
4450 // CHECK3-NEXT: store i32 [[CONV1]], i32* [[DOTOMP_UB]], align 4
4451 // CHECK3-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4
4452 // CHECK3-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4
4453 // CHECK3-NEXT: [[TMP2:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
4454 // CHECK3-NEXT: [[TMP3:%.*]] = load i32, i32* [[TMP2]], align 4
4455 // CHECK3-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB2]], i32 [[TMP3]], i32 34, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_LB]], i32* [[DOTOMP_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1)
4456 // CHECK3-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
4457 // CHECK3-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP4]], 99
4458 // CHECK3-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
4459 // CHECK3: cond.true:
4460 // CHECK3-NEXT: br label [[COND_END:%.*]]
4461 // CHECK3: cond.false:
4462 // CHECK3-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
4463 // CHECK3-NEXT: br label [[COND_END]]
4464 // CHECK3: cond.end:
4465 // CHECK3-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP5]], [[COND_FALSE]] ]
4466 // CHECK3-NEXT: store i32 [[COND]], i32* [[DOTOMP_UB]], align 4
4467 // CHECK3-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4
4468 // CHECK3-NEXT: store i32 [[TMP6]], i32* [[DOTOMP_IV]], align 4
4469 // CHECK3-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
4470 // CHECK3: omp.inner.for.cond:
4471 // CHECK3-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
4472 // CHECK3-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
4473 // CHECK3-NEXT: [[CMP2:%.*]] = icmp sle i32 [[TMP7]], [[TMP8]]
4474 // CHECK3-NEXT: br i1 [[CMP2]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
4475 // CHECK3: omp.inner.for.body:
4476 // CHECK3-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
4477 // CHECK3-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP9]], 1
4478 // CHECK3-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
4479 // CHECK3-NEXT: store i32 [[ADD]], i32* [[I]], align 4
4480 // CHECK3-NEXT: call void @_Z3fn2v()
4481 // CHECK3-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
4482 // CHECK3: omp.body.continue:
4483 // CHECK3-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
4484 // CHECK3: omp.inner.for.inc:
4485 // CHECK3-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
4486 // CHECK3-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP10]], 1
4487 // CHECK3-NEXT: store i32 [[ADD3]], i32* [[DOTOMP_IV]], align 4
4488 // CHECK3-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP31:![0-9]+]]
4489 // CHECK3: omp.inner.for.end:
4490 // CHECK3-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
4491 // CHECK3: omp.loop.exit:
4492 // CHECK3-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP3]])
4493 // CHECK3-NEXT: [[TMP11:%.*]] = load i32, i32* [[DOTOMP_IS_LAST]], align 4
4494 // CHECK3-NEXT: [[TMP12:%.*]] = icmp ne i32 [[TMP11]], 0
4495 // CHECK3-NEXT: br i1 [[TMP12]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
4496 // CHECK3: .omp.final.then:
4497 // CHECK3-NEXT: store i32 100, i32* [[I]], align 4
4498 // CHECK3-NEXT: br label [[DOTOMP_FINAL_DONE]]
4499 // CHECK3: .omp.final.done:
4500 // CHECK3-NEXT: ret void
4501 //
4502 //
4503 // CHECK3-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiEiT__l67
4504 // CHECK3-SAME: (i64 [[ARG:%.*]]) #[[ATTR1]] {
4505 // CHECK3-NEXT: entry:
4506 // CHECK3-NEXT: [[ARG_ADDR:%.*]] = alloca i64, align 8
4507 // CHECK3-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i8, align 1
4508 // CHECK3-NEXT: [[DOTCAPTURE_EXPR__CASTED:%.*]] = alloca i64, align 8
4509 // CHECK3-NEXT: store i64 [[ARG]], i64* [[ARG_ADDR]], align 8
4510 // CHECK3-NEXT: [[CONV:%.*]] = bitcast i64* [[ARG_ADDR]] to i32*
4511 // CHECK3-NEXT: [[TMP0:%.*]] = load i32, i32* [[CONV]], align 8
4512 // CHECK3-NEXT: [[TOBOOL:%.*]] = icmp ne i32 [[TMP0]], 0
4513 // CHECK3-NEXT: [[FROMBOOL:%.*]] = zext i1 [[TOBOOL]] to i8
4514 // CHECK3-NEXT: store i8 [[FROMBOOL]], i8* [[DOTCAPTURE_EXPR_]], align 1
4515 // CHECK3-NEXT: [[TMP1:%.*]] = load i8, i8* [[DOTCAPTURE_EXPR_]], align 1
4516 // CHECK3-NEXT: [[TOBOOL1:%.*]] = trunc i8 [[TMP1]] to i1
4517 // CHECK3-NEXT: [[CONV2:%.*]] = bitcast i64* [[DOTCAPTURE_EXPR__CASTED]] to i8*
4518 // CHECK3-NEXT: [[FROMBOOL3:%.*]] = zext i1 [[TOBOOL1]] to i8
4519 // CHECK3-NEXT: store i8 [[FROMBOOL3]], i8* [[CONV2]], align 1
4520 // CHECK3-NEXT: [[TMP2:%.*]] = load i64, i64* [[DOTCAPTURE_EXPR__CASTED]], align 8
4521 // CHECK3-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64)* @.omp_outlined..15 to void (i32*, i32*, ...)*), i64 [[TMP2]])
4522 // CHECK3-NEXT: ret void
4523 //
4524 //
4525 // CHECK3-LABEL: define {{[^@]+}}@.omp_outlined..15
4526 // CHECK3-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[DOTCAPTURE_EXPR_:%.*]]) #[[ATTR1]] {
4527 // CHECK3-NEXT: entry:
4528 // CHECK3-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
4529 // CHECK3-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
4530 // CHECK3-NEXT: [[DOTCAPTURE_EXPR__ADDR:%.*]] = alloca i64, align 8
4531 // CHECK3-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
4532 // CHECK3-NEXT: [[TMP:%.*]] = alloca i32, align 4
4533 // CHECK3-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4
4534 // CHECK3-NEXT: [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4
4535 // CHECK3-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
4536 // CHECK3-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
4537 // CHECK3-NEXT: [[I:%.*]] = alloca i32, align 4
4538 // CHECK3-NEXT: [[DOTBOUND_ZERO_ADDR:%.*]] = alloca i32, align 4
4539 // CHECK3-NEXT: store i32 0, i32* [[DOTBOUND_ZERO_ADDR]], align 4
4540 // CHECK3-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
4541 // CHECK3-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
4542 // CHECK3-NEXT: store i64 [[DOTCAPTURE_EXPR_]], i64* [[DOTCAPTURE_EXPR__ADDR]], align 8
4543 // CHECK3-NEXT: [[CONV:%.*]] = bitcast i64* [[DOTCAPTURE_EXPR__ADDR]] to i8*
4544 // CHECK3-NEXT: store i32 0, i32* [[DOTOMP_COMB_LB]], align 4
4545 // CHECK3-NEXT: store i32 99, i32* [[DOTOMP_COMB_UB]], align 4
4546 // CHECK3-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4
4547 // CHECK3-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4
4548 // CHECK3-NEXT: [[TMP0:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
4549 // CHECK3-NEXT: [[TMP1:%.*]] = load i32, i32* [[TMP0]], align 4
4550 // CHECK3-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB1]], i32 [[TMP1]], i32 92, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_COMB_LB]], i32* [[DOTOMP_COMB_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1)
4551 // CHECK3-NEXT: [[TMP2:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
4552 // CHECK3-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP2]], 99
4553 // CHECK3-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
4554 // CHECK3: cond.true:
4555 // CHECK3-NEXT: br label [[COND_END:%.*]]
4556 // CHECK3: cond.false:
4557 // CHECK3-NEXT: [[TMP3:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
4558 // CHECK3-NEXT: br label [[COND_END]]
4559 // CHECK3: cond.end:
4560 // CHECK3-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP3]], [[COND_FALSE]] ]
4561 // CHECK3-NEXT: store i32 [[COND]], i32* [[DOTOMP_COMB_UB]], align 4
4562 // CHECK3-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4
4563 // CHECK3-NEXT: store i32 [[TMP4]], i32* [[DOTOMP_IV]], align 4
4564 // CHECK3-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
4565 // CHECK3: omp.inner.for.cond:
4566 // CHECK3-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
4567 // CHECK3-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
4568 // CHECK3-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP5]], [[TMP6]]
4569 // CHECK3-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
4570 // CHECK3: omp.inner.for.body:
4571 // CHECK3-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4
4572 // CHECK3-NEXT: [[TMP8:%.*]] = zext i32 [[TMP7]] to i64
4573 // CHECK3-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
4574 // CHECK3-NEXT: [[TMP10:%.*]] = zext i32 [[TMP9]] to i64
4575 // CHECK3-NEXT: [[TMP11:%.*]] = load i8, i8* [[CONV]], align 8
4576 // CHECK3-NEXT: [[TOBOOL:%.*]] = trunc i8 [[TMP11]] to i1
4577 // CHECK3-NEXT: br i1 [[TOBOOL]], label [[OMP_IF_THEN:%.*]], label [[OMP_IF_ELSE:%.*]]
4578 // CHECK3: omp_if.then:
4579 // CHECK3-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB3]], i32 2, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64, i64)* @.omp_outlined..16 to void (i32*, i32*, ...)*), i64 [[TMP8]], i64 [[TMP10]])
4580 // CHECK3-NEXT: br label [[OMP_IF_END:%.*]]
4581 // CHECK3: omp_if.else:
4582 // CHECK3-NEXT: call void @__kmpc_serialized_parallel(%struct.ident_t* @[[GLOB3]], i32 [[TMP1]])
4583 // CHECK3-NEXT: [[TMP12:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
4584 // CHECK3-NEXT: call void @.omp_outlined..16(i32* [[TMP12]], i32* [[DOTBOUND_ZERO_ADDR]], i64 [[TMP8]], i64 [[TMP10]]) #[[ATTR2]]
4585 // CHECK3-NEXT: call void @__kmpc_end_serialized_parallel(%struct.ident_t* @[[GLOB3]], i32 [[TMP1]])
4586 // CHECK3-NEXT: br label [[OMP_IF_END]]
4587 // CHECK3: omp_if.end:
4588 // CHECK3-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
4589 // CHECK3: omp.inner.for.inc:
4590 // CHECK3-NEXT: [[TMP13:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
4591 // CHECK3-NEXT: [[TMP14:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4
4592 // CHECK3-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP13]], [[TMP14]]
4593 // CHECK3-NEXT: store i32 [[ADD]], i32* [[DOTOMP_IV]], align 4
4594 // CHECK3-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP32:![0-9]+]]
4595 // CHECK3: omp.inner.for.end:
4596 // CHECK3-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
4597 // CHECK3: omp.loop.exit:
4598 // CHECK3-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP1]])
4599 // CHECK3-NEXT: [[TMP15:%.*]] = load i32, i32* [[DOTOMP_IS_LAST]], align 4
4600 // CHECK3-NEXT: [[TMP16:%.*]] = icmp ne i32 [[TMP15]], 0
4601 // CHECK3-NEXT: br i1 [[TMP16]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
4602 // CHECK3: .omp.final.then:
4603 // CHECK3-NEXT: store i32 100, i32* [[I]], align 4
4604 // CHECK3-NEXT: br label [[DOTOMP_FINAL_DONE]]
4605 // CHECK3: .omp.final.done:
4606 // CHECK3-NEXT: ret void
4607 //
4608 //
4609 // CHECK3-LABEL: define {{[^@]+}}@.omp_outlined..16
4610 // CHECK3-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[DOTPREVIOUS_LB_:%.*]], i64 [[DOTPREVIOUS_UB_:%.*]]) #[[ATTR1]] {
4611 // CHECK3-NEXT: entry:
4612 // CHECK3-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
4613 // CHECK3-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
4614 // CHECK3-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i64, align 8
4615 // CHECK3-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i64, align 8
4616 // CHECK3-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
4617 // CHECK3-NEXT: [[TMP:%.*]] = alloca i32, align 4
4618 // CHECK3-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4
4619 // CHECK3-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4
4620 // CHECK3-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
4621 // CHECK3-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
4622 // CHECK3-NEXT: [[I:%.*]] = alloca i32, align 4
4623 // CHECK3-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
4624 // CHECK3-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
4625 // CHECK3-NEXT: store i64 [[DOTPREVIOUS_LB_]], i64* [[DOTPREVIOUS_LB__ADDR]], align 8
4626 // CHECK3-NEXT: store i64 [[DOTPREVIOUS_UB_]], i64* [[DOTPREVIOUS_UB__ADDR]], align 8
4627 // CHECK3-NEXT: store i32 0, i32* [[DOTOMP_LB]], align 4
4628 // CHECK3-NEXT: store i32 99, i32* [[DOTOMP_UB]], align 4
4629 // CHECK3-NEXT: [[TMP0:%.*]] = load i64, i64* [[DOTPREVIOUS_LB__ADDR]], align 8
4630 // CHECK3-NEXT: [[CONV:%.*]] = trunc i64 [[TMP0]] to i32
4631 // CHECK3-NEXT: [[TMP1:%.*]] = load i64, i64* [[DOTPREVIOUS_UB__ADDR]], align 8
4632 // CHECK3-NEXT: [[CONV1:%.*]] = trunc i64 [[TMP1]] to i32
4633 // CHECK3-NEXT: store i32 [[CONV]], i32* [[DOTOMP_LB]], align 4
4634 // CHECK3-NEXT: store i32 [[CONV1]], i32* [[DOTOMP_UB]], align 4
4635 // CHECK3-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4
4636 // CHECK3-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4
4637 // CHECK3-NEXT: [[TMP2:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
4638 // CHECK3-NEXT: [[TMP3:%.*]] = load i32, i32* [[TMP2]], align 4
4639 // CHECK3-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB2]], i32 [[TMP3]], i32 34, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_LB]], i32* [[DOTOMP_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1)
4640 // CHECK3-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
4641 // CHECK3-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP4]], 99
4642 // CHECK3-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
4643 // CHECK3: cond.true:
4644 // CHECK3-NEXT: br label [[COND_END:%.*]]
4645 // CHECK3: cond.false:
4646 // CHECK3-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
4647 // CHECK3-NEXT: br label [[COND_END]]
4648 // CHECK3: cond.end:
4649 // CHECK3-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP5]], [[COND_FALSE]] ]
4650 // CHECK3-NEXT: store i32 [[COND]], i32* [[DOTOMP_UB]], align 4
4651 // CHECK3-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4
4652 // CHECK3-NEXT: store i32 [[TMP6]], i32* [[DOTOMP_IV]], align 4
4653 // CHECK3-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
4654 // CHECK3: omp.inner.for.cond:
4655 // CHECK3-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
4656 // CHECK3-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
4657 // CHECK3-NEXT: [[CMP2:%.*]] = icmp sle i32 [[TMP7]], [[TMP8]]
4658 // CHECK3-NEXT: br i1 [[CMP2]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
4659 // CHECK3: omp.inner.for.body:
4660 // CHECK3-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
4661 // CHECK3-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP9]], 1
4662 // CHECK3-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
4663 // CHECK3-NEXT: store i32 [[ADD]], i32* [[I]], align 4
4664 // CHECK3-NEXT: call void @_Z3fn3v()
4665 // CHECK3-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
4666 // CHECK3: omp.body.continue:
4667 // CHECK3-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
4668 // CHECK3: omp.inner.for.inc:
4669 // CHECK3-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
4670 // CHECK3-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP10]], 1
4671 // CHECK3-NEXT: store i32 [[ADD3]], i32* [[DOTOMP_IV]], align 4
4672 // CHECK3-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP33:![0-9]+]]
4673 // CHECK3: omp.inner.for.end:
4674 // CHECK3-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
4675 // CHECK3: omp.loop.exit:
4676 // CHECK3-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP3]])
4677 // CHECK3-NEXT: [[TMP11:%.*]] = load i32, i32* [[DOTOMP_IS_LAST]], align 4
4678 // CHECK3-NEXT: [[TMP12:%.*]] = icmp ne i32 [[TMP11]], 0
4679 // CHECK3-NEXT: br i1 [[TMP12]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
4680 // CHECK3: .omp.final.then:
4681 // CHECK3-NEXT: store i32 100, i32* [[I]], align 4
4682 // CHECK3-NEXT: br label [[DOTOMP_FINAL_DONE]]
4683 // CHECK3: .omp.final.done:
4684 // CHECK3-NEXT: ret void
4685 //
4686 //
4687 // CHECK3-LABEL: define {{[^@]+}}@.omp_offloading.requires_reg
4688 // CHECK3-SAME: () #[[ATTR5:[0-9]+]] {
4689 // CHECK3-NEXT: entry:
4690 // CHECK3-NEXT: call void @__tgt_register_requires(i64 1)
4691 // CHECK3-NEXT: ret void
4692 //
4693 //
4694 // CHECK4-LABEL: define {{[^@]+}}@_Z9gtid_testv
4695 // CHECK4-SAME: () #[[ATTR0:[0-9]+]] {
4696 // CHECK4-NEXT: entry:
4697 // CHECK4-NEXT: [[TMP:%.*]] = alloca i32, align 4
4698 // CHECK4-NEXT: [[_TMP1:%.*]] = alloca i32, align 4
4699 // CHECK4-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3:[0-9]+]], i64 -1, i64 100)
4700 // CHECK4-NEXT: [[TMP0:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9gtid_testv_l43.region_id, i32 0, i8** null, i8** null, i64* null, i64* null, i8** null, i8** null, i32 0, i32 0)
4701 // CHECK4-NEXT: [[TMP1:%.*]] = icmp ne i32 [[TMP0]], 0
4702 // CHECK4-NEXT: br i1 [[TMP1]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
4703 // CHECK4: omp_offload.failed:
4704 // CHECK4-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9gtid_testv_l43() #[[ATTR2:[0-9]+]]
4705 // CHECK4-NEXT: br label [[OMP_OFFLOAD_CONT]]
4706 // CHECK4: omp_offload.cont:
4707 // CHECK4-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i64 100)
4708 // CHECK4-NEXT: [[TMP2:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9gtid_testv_l47.region_id, i32 0, i8** null, i8** null, i64* null, i64* null, i8** null, i8** null, i32 0, i32 1)
4709 // CHECK4-NEXT: [[TMP3:%.*]] = icmp ne i32 [[TMP2]], 0
4710 // CHECK4-NEXT: br i1 [[TMP3]], label [[OMP_OFFLOAD_FAILED2:%.*]], label [[OMP_OFFLOAD_CONT3:%.*]]
4711 // CHECK4: omp_offload.failed2:
4712 // CHECK4-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9gtid_testv_l47() #[[ATTR2]]
4713 // CHECK4-NEXT: br label [[OMP_OFFLOAD_CONT3]]
4714 // CHECK4: omp_offload.cont3:
4715 // CHECK4-NEXT: ret void
4716 //
4717 //
4718 // CHECK4-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9gtid_testv_l43
4719 // CHECK4-SAME: () #[[ATTR1:[0-9]+]] {
4720 // CHECK4-NEXT: entry:
4721 // CHECK4-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 0, void (i32*, i32*, ...)* bitcast (void (i32*, i32*)* @.omp_outlined. to void (i32*, i32*, ...)*))
4722 // CHECK4-NEXT: ret void
4723 //
4724 //
4725 // CHECK4-LABEL: define {{[^@]+}}@.omp_outlined.
4726 // CHECK4-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]]) #[[ATTR1]] {
4727 // CHECK4-NEXT: entry:
4728 // CHECK4-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
4729 // CHECK4-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
4730 // CHECK4-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
4731 // CHECK4-NEXT: [[TMP:%.*]] = alloca i32, align 4
4732 // CHECK4-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4
4733 // CHECK4-NEXT: [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4
4734 // CHECK4-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
4735 // CHECK4-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
4736 // CHECK4-NEXT: [[I:%.*]] = alloca i32, align 4
4737 // CHECK4-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
4738 // CHECK4-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
4739 // CHECK4-NEXT: store i32 0, i32* [[DOTOMP_COMB_LB]], align 4
4740 // CHECK4-NEXT: store i32 99, i32* [[DOTOMP_COMB_UB]], align 4
4741 // CHECK4-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4
4742 // CHECK4-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4
4743 // CHECK4-NEXT: [[TMP0:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
4744 // CHECK4-NEXT: [[TMP1:%.*]] = load i32, i32* [[TMP0]], align 4
4745 // CHECK4-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB1:[0-9]+]], i32 [[TMP1]], i32 92, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_COMB_LB]], i32* [[DOTOMP_COMB_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1)
4746 // CHECK4-NEXT: [[TMP2:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
4747 // CHECK4-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP2]], 99
4748 // CHECK4-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
4749 // CHECK4: cond.true:
4750 // CHECK4-NEXT: br label [[COND_END:%.*]]
4751 // CHECK4: cond.false:
4752 // CHECK4-NEXT: [[TMP3:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
4753 // CHECK4-NEXT: br label [[COND_END]]
4754 // CHECK4: cond.end:
4755 // CHECK4-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP3]], [[COND_FALSE]] ]
4756 // CHECK4-NEXT: store i32 [[COND]], i32* [[DOTOMP_COMB_UB]], align 4
4757 // CHECK4-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4
4758 // CHECK4-NEXT: store i32 [[TMP4]], i32* [[DOTOMP_IV]], align 4
4759 // CHECK4-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
4760 // CHECK4: omp.inner.for.cond:
4761 // CHECK4-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
4762 // CHECK4-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
4763 // CHECK4-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP5]], [[TMP6]]
4764 // CHECK4-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
4765 // CHECK4: omp.inner.for.body:
4766 // CHECK4-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4
4767 // CHECK4-NEXT: [[TMP8:%.*]] = zext i32 [[TMP7]] to i64
4768 // CHECK4-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
4769 // CHECK4-NEXT: [[TMP10:%.*]] = zext i32 [[TMP9]] to i64
4770 // CHECK4-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB3]], i32 2, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64, i64)* @.omp_outlined..1 to void (i32*, i32*, ...)*), i64 [[TMP8]], i64 [[TMP10]])
4771 // CHECK4-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
4772 // CHECK4: omp.inner.for.inc:
4773 // CHECK4-NEXT: [[TMP11:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
4774 // CHECK4-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4
4775 // CHECK4-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP11]], [[TMP12]]
4776 // CHECK4-NEXT: store i32 [[ADD]], i32* [[DOTOMP_IV]], align 4
4777 // CHECK4-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP10:![0-9]+]]
4778 // CHECK4: omp.inner.for.end:
4779 // CHECK4-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
4780 // CHECK4: omp.loop.exit:
4781 // CHECK4-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP1]])
4782 // CHECK4-NEXT: [[TMP13:%.*]] = load i32, i32* [[DOTOMP_IS_LAST]], align 4
4783 // CHECK4-NEXT: [[TMP14:%.*]] = icmp ne i32 [[TMP13]], 0
4784 // CHECK4-NEXT: br i1 [[TMP14]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
4785 // CHECK4: .omp.final.then:
4786 // CHECK4-NEXT: store i32 100, i32* [[I]], align 4
4787 // CHECK4-NEXT: br label [[DOTOMP_FINAL_DONE]]
4788 // CHECK4: .omp.final.done:
4789 // CHECK4-NEXT: ret void
4790 //
4791 //
4792 // CHECK4-LABEL: define {{[^@]+}}@.omp_outlined..1
4793 // CHECK4-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[DOTPREVIOUS_LB_:%.*]], i64 [[DOTPREVIOUS_UB_:%.*]]) #[[ATTR1]] {
4794 // CHECK4-NEXT: entry:
4795 // CHECK4-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
4796 // CHECK4-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
4797 // CHECK4-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i64, align 8
4798 // CHECK4-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i64, align 8
4799 // CHECK4-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
4800 // CHECK4-NEXT: [[TMP:%.*]] = alloca i32, align 4
4801 // CHECK4-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4
4802 // CHECK4-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4
4803 // CHECK4-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
4804 // CHECK4-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
4805 // CHECK4-NEXT: [[I:%.*]] = alloca i32, align 4
4806 // CHECK4-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
4807 // CHECK4-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
4808 // CHECK4-NEXT: store i64 [[DOTPREVIOUS_LB_]], i64* [[DOTPREVIOUS_LB__ADDR]], align 8
4809 // CHECK4-NEXT: store i64 [[DOTPREVIOUS_UB_]], i64* [[DOTPREVIOUS_UB__ADDR]], align 8
4810 // CHECK4-NEXT: store i32 0, i32* [[DOTOMP_LB]], align 4
4811 // CHECK4-NEXT: store i32 99, i32* [[DOTOMP_UB]], align 4
4812 // CHECK4-NEXT: [[TMP0:%.*]] = load i64, i64* [[DOTPREVIOUS_LB__ADDR]], align 8
4813 // CHECK4-NEXT: [[CONV:%.*]] = trunc i64 [[TMP0]] to i32
4814 // CHECK4-NEXT: [[TMP1:%.*]] = load i64, i64* [[DOTPREVIOUS_UB__ADDR]], align 8
4815 // CHECK4-NEXT: [[CONV1:%.*]] = trunc i64 [[TMP1]] to i32
4816 // CHECK4-NEXT: store i32 [[CONV]], i32* [[DOTOMP_LB]], align 4
4817 // CHECK4-NEXT: store i32 [[CONV1]], i32* [[DOTOMP_UB]], align 4
4818 // CHECK4-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4
4819 // CHECK4-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4
4820 // CHECK4-NEXT: [[TMP2:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
4821 // CHECK4-NEXT: [[TMP3:%.*]] = load i32, i32* [[TMP2]], align 4
4822 // CHECK4-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB2:[0-9]+]], i32 [[TMP3]], i32 34, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_LB]], i32* [[DOTOMP_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1)
4823 // CHECK4-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
4824 // CHECK4-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP4]], 99
4825 // CHECK4-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
4826 // CHECK4: cond.true:
4827 // CHECK4-NEXT: br label [[COND_END:%.*]]
4828 // CHECK4: cond.false:
4829 // CHECK4-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
4830 // CHECK4-NEXT: br label [[COND_END]]
4831 // CHECK4: cond.end:
4832 // CHECK4-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP5]], [[COND_FALSE]] ]
4833 // CHECK4-NEXT: store i32 [[COND]], i32* [[DOTOMP_UB]], align 4
4834 // CHECK4-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4
4835 // CHECK4-NEXT: store i32 [[TMP6]], i32* [[DOTOMP_IV]], align 4
4836 // CHECK4-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
4837 // CHECK4: omp.inner.for.cond:
4838 // CHECK4-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
4839 // CHECK4-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
4840 // CHECK4-NEXT: [[CMP2:%.*]] = icmp sle i32 [[TMP7]], [[TMP8]]
4841 // CHECK4-NEXT: br i1 [[CMP2]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
4842 // CHECK4: omp.inner.for.body:
4843 // CHECK4-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
4844 // CHECK4-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP9]], 1
4845 // CHECK4-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
4846 // CHECK4-NEXT: store i32 [[ADD]], i32* [[I]], align 4
4847 // CHECK4-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
4848 // CHECK4: omp.body.continue:
4849 // CHECK4-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
4850 // CHECK4: omp.inner.for.inc:
4851 // CHECK4-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
4852 // CHECK4-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP10]], 1
4853 // CHECK4-NEXT: store i32 [[ADD3]], i32* [[DOTOMP_IV]], align 4
4854 // CHECK4-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP12:![0-9]+]]
4855 // CHECK4: omp.inner.for.end:
4856 // CHECK4-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
4857 // CHECK4: omp.loop.exit:
4858 // CHECK4-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP3]])
4859 // CHECK4-NEXT: [[TMP11:%.*]] = load i32, i32* [[DOTOMP_IS_LAST]], align 4
4860 // CHECK4-NEXT: [[TMP12:%.*]] = icmp ne i32 [[TMP11]], 0
4861 // CHECK4-NEXT: br i1 [[TMP12]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
4862 // CHECK4: .omp.final.then:
4863 // CHECK4-NEXT: store i32 100, i32* [[I]], align 4
4864 // CHECK4-NEXT: br label [[DOTOMP_FINAL_DONE]]
4865 // CHECK4: .omp.final.done:
4866 // CHECK4-NEXT: ret void
4867 //
4868 //
4869 // CHECK4-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9gtid_testv_l47
4870 // CHECK4-SAME: () #[[ATTR1]] {
4871 // CHECK4-NEXT: entry:
4872 // CHECK4-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 0, void (i32*, i32*, ...)* bitcast (void (i32*, i32*)* @.omp_outlined..2 to void (i32*, i32*, ...)*))
4873 // CHECK4-NEXT: ret void
4874 //
4875 //
4876 // CHECK4-LABEL: define {{[^@]+}}@.omp_outlined..2
4877 // CHECK4-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]]) #[[ATTR1]] {
4878 // CHECK4-NEXT: entry:
4879 // CHECK4-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
4880 // CHECK4-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
4881 // CHECK4-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
4882 // CHECK4-NEXT: [[TMP:%.*]] = alloca i32, align 4
4883 // CHECK4-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4
4884 // CHECK4-NEXT: [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4
4885 // CHECK4-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
4886 // CHECK4-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
4887 // CHECK4-NEXT: [[I:%.*]] = alloca i32, align 4
4888 // CHECK4-NEXT: [[DOTBOUND_ZERO_ADDR:%.*]] = alloca i32, align 4
4889 // CHECK4-NEXT: store i32 0, i32* [[DOTBOUND_ZERO_ADDR]], align 4
4890 // CHECK4-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
4891 // CHECK4-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
4892 // CHECK4-NEXT: store i32 0, i32* [[DOTOMP_COMB_LB]], align 4
4893 // CHECK4-NEXT: store i32 99, i32* [[DOTOMP_COMB_UB]], align 4
4894 // CHECK4-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4
4895 // CHECK4-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4
4896 // CHECK4-NEXT: [[TMP0:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
4897 // CHECK4-NEXT: [[TMP1:%.*]] = load i32, i32* [[TMP0]], align 4
4898 // CHECK4-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB1]], i32 [[TMP1]], i32 92, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_COMB_LB]], i32* [[DOTOMP_COMB_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1)
4899 // CHECK4-NEXT: [[TMP2:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
4900 // CHECK4-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP2]], 99
4901 // CHECK4-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
4902 // CHECK4: cond.true:
4903 // CHECK4-NEXT: br label [[COND_END:%.*]]
4904 // CHECK4: cond.false:
4905 // CHECK4-NEXT: [[TMP3:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
4906 // CHECK4-NEXT: br label [[COND_END]]
4907 // CHECK4: cond.end:
4908 // CHECK4-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP3]], [[COND_FALSE]] ]
4909 // CHECK4-NEXT: store i32 [[COND]], i32* [[DOTOMP_COMB_UB]], align 4
4910 // CHECK4-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4
4911 // CHECK4-NEXT: store i32 [[TMP4]], i32* [[DOTOMP_IV]], align 4
4912 // CHECK4-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
4913 // CHECK4: omp.inner.for.cond:
4914 // CHECK4-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
4915 // CHECK4-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
4916 // CHECK4-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP5]], [[TMP6]]
4917 // CHECK4-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
4918 // CHECK4: omp.inner.for.body:
4919 // CHECK4-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4
4920 // CHECK4-NEXT: [[TMP8:%.*]] = zext i32 [[TMP7]] to i64
4921 // CHECK4-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
4922 // CHECK4-NEXT: [[TMP10:%.*]] = zext i32 [[TMP9]] to i64
4923 // CHECK4-NEXT: call void @__kmpc_serialized_parallel(%struct.ident_t* @[[GLOB3]], i32 [[TMP1]])
4924 // CHECK4-NEXT: [[TMP11:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
4925 // CHECK4-NEXT: call void @.omp_outlined..3(i32* [[TMP11]], i32* [[DOTBOUND_ZERO_ADDR]], i64 [[TMP8]], i64 [[TMP10]]) #[[ATTR2]]
4926 // CHECK4-NEXT: call void @__kmpc_end_serialized_parallel(%struct.ident_t* @[[GLOB3]], i32 [[TMP1]])
4927 // CHECK4-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
4928 // CHECK4: omp.inner.for.inc:
4929 // CHECK4-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
4930 // CHECK4-NEXT: [[TMP13:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4
4931 // CHECK4-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP12]], [[TMP13]]
4932 // CHECK4-NEXT: store i32 [[ADD]], i32* [[DOTOMP_IV]], align 4
4933 // CHECK4-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP15:![0-9]+]]
4934 // CHECK4: omp.inner.for.end:
4935 // CHECK4-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
4936 // CHECK4: omp.loop.exit:
4937 // CHECK4-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP1]])
4938 // CHECK4-NEXT: [[TMP14:%.*]] = load i32, i32* [[DOTOMP_IS_LAST]], align 4
4939 // CHECK4-NEXT: [[TMP15:%.*]] = icmp ne i32 [[TMP14]], 0
4940 // CHECK4-NEXT: br i1 [[TMP15]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
4941 // CHECK4: .omp.final.then:
4942 // CHECK4-NEXT: store i32 100, i32* [[I]], align 4
4943 // CHECK4-NEXT: br label [[DOTOMP_FINAL_DONE]]
4944 // CHECK4: .omp.final.done:
4945 // CHECK4-NEXT: ret void
4946 //
4947 //
4948 // CHECK4-LABEL: define {{[^@]+}}@.omp_outlined..3
4949 // CHECK4-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[DOTPREVIOUS_LB_:%.*]], i64 [[DOTPREVIOUS_UB_:%.*]]) #[[ATTR1]] {
4950 // CHECK4-NEXT: entry:
4951 // CHECK4-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
4952 // CHECK4-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
4953 // CHECK4-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i64, align 8
4954 // CHECK4-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i64, align 8
4955 // CHECK4-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
4956 // CHECK4-NEXT: [[TMP:%.*]] = alloca i32, align 4
4957 // CHECK4-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4
4958 // CHECK4-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4
4959 // CHECK4-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
4960 // CHECK4-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
4961 // CHECK4-NEXT: [[I:%.*]] = alloca i32, align 4
4962 // CHECK4-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
4963 // CHECK4-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
4964 // CHECK4-NEXT: store i64 [[DOTPREVIOUS_LB_]], i64* [[DOTPREVIOUS_LB__ADDR]], align 8
4965 // CHECK4-NEXT: store i64 [[DOTPREVIOUS_UB_]], i64* [[DOTPREVIOUS_UB__ADDR]], align 8
4966 // CHECK4-NEXT: store i32 0, i32* [[DOTOMP_LB]], align 4
4967 // CHECK4-NEXT: store i32 99, i32* [[DOTOMP_UB]], align 4
4968 // CHECK4-NEXT: [[TMP0:%.*]] = load i64, i64* [[DOTPREVIOUS_LB__ADDR]], align 8
4969 // CHECK4-NEXT: [[CONV:%.*]] = trunc i64 [[TMP0]] to i32
4970 // CHECK4-NEXT: [[TMP1:%.*]] = load i64, i64* [[DOTPREVIOUS_UB__ADDR]], align 8
4971 // CHECK4-NEXT: [[CONV1:%.*]] = trunc i64 [[TMP1]] to i32
4972 // CHECK4-NEXT: store i32 [[CONV]], i32* [[DOTOMP_LB]], align 4
4973 // CHECK4-NEXT: store i32 [[CONV1]], i32* [[DOTOMP_UB]], align 4
4974 // CHECK4-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4
4975 // CHECK4-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4
4976 // CHECK4-NEXT: [[TMP2:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
4977 // CHECK4-NEXT: [[TMP3:%.*]] = load i32, i32* [[TMP2]], align 4
4978 // CHECK4-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB2]], i32 [[TMP3]], i32 34, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_LB]], i32* [[DOTOMP_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1)
4979 // CHECK4-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
4980 // CHECK4-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP4]], 99
4981 // CHECK4-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
4982 // CHECK4: cond.true:
4983 // CHECK4-NEXT: br label [[COND_END:%.*]]
4984 // CHECK4: cond.false:
4985 // CHECK4-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
4986 // CHECK4-NEXT: br label [[COND_END]]
4987 // CHECK4: cond.end:
4988 // CHECK4-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP5]], [[COND_FALSE]] ]
4989 // CHECK4-NEXT: store i32 [[COND]], i32* [[DOTOMP_UB]], align 4
4990 // CHECK4-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4
4991 // CHECK4-NEXT: store i32 [[TMP6]], i32* [[DOTOMP_IV]], align 4
4992 // CHECK4-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
4993 // CHECK4: omp.inner.for.cond:
4994 // CHECK4-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
4995 // CHECK4-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
4996 // CHECK4-NEXT: [[CMP2:%.*]] = icmp sle i32 [[TMP7]], [[TMP8]]
4997 // CHECK4-NEXT: br i1 [[CMP2]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
4998 // CHECK4: omp.inner.for.body:
4999 // CHECK4-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
5000 // CHECK4-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP9]], 1
5001 // CHECK4-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
5002 // CHECK4-NEXT: store i32 [[ADD]], i32* [[I]], align 4
5003 // CHECK4-NEXT: call void @_Z9gtid_testv()
5004 // CHECK4-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
5005 // CHECK4: omp.body.continue:
5006 // CHECK4-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
5007 // CHECK4: omp.inner.for.inc:
5008 // CHECK4-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
5009 // CHECK4-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP10]], 1
5010 // CHECK4-NEXT: store i32 [[ADD3]], i32* [[DOTOMP_IV]], align 4
5011 // CHECK4-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP16:![0-9]+]]
5012 // CHECK4: omp.inner.for.end:
5013 // CHECK4-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
5014 // CHECK4: omp.loop.exit:
5015 // CHECK4-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP3]])
5016 // CHECK4-NEXT: [[TMP11:%.*]] = load i32, i32* [[DOTOMP_IS_LAST]], align 4
5017 // CHECK4-NEXT: [[TMP12:%.*]] = icmp ne i32 [[TMP11]], 0
5018 // CHECK4-NEXT: br i1 [[TMP12]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
5019 // CHECK4: .omp.final.then:
5020 // CHECK4-NEXT: store i32 100, i32* [[I]], align 4
5021 // CHECK4-NEXT: br label [[DOTOMP_FINAL_DONE]]
5022 // CHECK4: .omp.final.done:
5023 // CHECK4-NEXT: ret void
5024 //
5025 //
5026 // CHECK4-LABEL: define {{[^@]+}}@main
5027 // CHECK4-SAME: () #[[ATTR3:[0-9]+]] {
5028 // CHECK4-NEXT: entry:
5029 // CHECK4-NEXT: [[RETVAL:%.*]] = alloca i32, align 4
5030 // CHECK4-NEXT: [[TMP:%.*]] = alloca i32, align 4
5031 // CHECK4-NEXT: [[_TMP1:%.*]] = alloca i32, align 4
5032 // CHECK4-NEXT: [[ARG_CASTED:%.*]] = alloca i64, align 8
5033 // CHECK4-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [1 x i8*], align 8
5034 // CHECK4-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [1 x i8*], align 8
5035 // CHECK4-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [1 x i8*], align 8
5036 // CHECK4-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i8, align 1
5037 // CHECK4-NEXT: [[_TMP5:%.*]] = alloca i32, align 4
5038 // CHECK4-NEXT: store i32 0, i32* [[RETVAL]], align 4
5039 // CHECK4-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i64 100)
5040 // CHECK4-NEXT: [[TMP0:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l76.region_id, i32 0, i8** null, i8** null, i64* null, i64* null, i8** null, i8** null, i32 0, i32 0)
5041 // CHECK4-NEXT: [[TMP1:%.*]] = icmp ne i32 [[TMP0]], 0
5042 // CHECK4-NEXT: br i1 [[TMP1]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
5043 // CHECK4: omp_offload.failed:
5044 // CHECK4-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l76() #[[ATTR2]]
5045 // CHECK4-NEXT: br label [[OMP_OFFLOAD_CONT]]
5046 // CHECK4: omp_offload.cont:
5047 // CHECK4-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i64 100)
5048 // CHECK4-NEXT: [[TMP2:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l84.region_id, i32 0, i8** null, i8** null, i64* null, i64* null, i8** null, i8** null, i32 0, i32 1)
5049 // CHECK4-NEXT: [[TMP3:%.*]] = icmp ne i32 [[TMP2]], 0
5050 // CHECK4-NEXT: br i1 [[TMP3]], label [[OMP_OFFLOAD_FAILED2:%.*]], label [[OMP_OFFLOAD_CONT3:%.*]]
5051 // CHECK4: omp_offload.failed2:
5052 // CHECK4-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l84() #[[ATTR2]]
5053 // CHECK4-NEXT: br label [[OMP_OFFLOAD_CONT3]]
5054 // CHECK4: omp_offload.cont3:
5055 // CHECK4-NEXT: [[TMP4:%.*]] = load i32, i32* @Arg, align 4
5056 // CHECK4-NEXT: [[CONV:%.*]] = bitcast i64* [[ARG_CASTED]] to i32*
5057 // CHECK4-NEXT: store i32 [[TMP4]], i32* [[CONV]], align 4
5058 // CHECK4-NEXT: [[TMP5:%.*]] = load i64, i64* [[ARG_CASTED]], align 8
5059 // CHECK4-NEXT: [[TMP6:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
5060 // CHECK4-NEXT: [[TMP7:%.*]] = bitcast i8** [[TMP6]] to i64*
5061 // CHECK4-NEXT: store i64 [[TMP5]], i64* [[TMP7]], align 8
5062 // CHECK4-NEXT: [[TMP8:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
5063 // CHECK4-NEXT: [[TMP9:%.*]] = bitcast i8** [[TMP8]] to i64*
5064 // CHECK4-NEXT: store i64 [[TMP5]], i64* [[TMP9]], align 8
5065 // CHECK4-NEXT: [[TMP10:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0
5066 // CHECK4-NEXT: store i8* null, i8** [[TMP10]], align 8
5067 // CHECK4-NEXT: [[TMP11:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
5068 // CHECK4-NEXT: [[TMP12:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
5069 // CHECK4-NEXT: [[TMP13:%.*]] = load i32, i32* @Arg, align 4
5070 // CHECK4-NEXT: [[TOBOOL:%.*]] = icmp ne i32 [[TMP13]], 0
5071 // CHECK4-NEXT: [[FROMBOOL:%.*]] = zext i1 [[TOBOOL]] to i8
5072 // CHECK4-NEXT: store i8 [[FROMBOOL]], i8* [[DOTCAPTURE_EXPR_]], align 1
5073 // CHECK4-NEXT: [[TMP14:%.*]] = load i8, i8* [[DOTCAPTURE_EXPR_]], align 1
5074 // CHECK4-NEXT: [[TOBOOL4:%.*]] = trunc i8 [[TMP14]] to i1
5075 // CHECK4-NEXT: [[TMP15:%.*]] = select i1 [[TOBOOL4]], i32 0, i32 1
5076 // CHECK4-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i64 100)
5077 // CHECK4-NEXT: [[TMP16:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l92.region_id, i32 1, i8** [[TMP11]], i8** [[TMP12]], i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_sizes, i32 0, i32 0), i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_maptypes, i32 0, i32 0), i8** null, i8** null, i32 0, i32 [[TMP15]])
5078 // CHECK4-NEXT: [[TMP17:%.*]] = icmp ne i32 [[TMP16]], 0
5079 // CHECK4-NEXT: br i1 [[TMP17]], label [[OMP_OFFLOAD_FAILED6:%.*]], label [[OMP_OFFLOAD_CONT7:%.*]]
5080 // CHECK4: omp_offload.failed6:
5081 // CHECK4-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l92(i64 [[TMP5]]) #[[ATTR2]]
5082 // CHECK4-NEXT: br label [[OMP_OFFLOAD_CONT7]]
5083 // CHECK4: omp_offload.cont7:
5084 // CHECK4-NEXT: [[TMP18:%.*]] = load i32, i32* @Arg, align 4
5085 // CHECK4-NEXT: [[CALL:%.*]] = call i32 @_Z5tmainIiEiT_(i32 [[TMP18]])
5086 // CHECK4-NEXT: ret i32 [[CALL]]
5087 //
5088 //
5089 // CHECK4-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l76
5090 // CHECK4-SAME: () #[[ATTR1]] {
5091 // CHECK4-NEXT: entry:
5092 // CHECK4-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 0, void (i32*, i32*, ...)* bitcast (void (i32*, i32*)* @.omp_outlined..4 to void (i32*, i32*, ...)*))
5093 // CHECK4-NEXT: ret void
5094 //
5095 //
5096 // CHECK4-LABEL: define {{[^@]+}}@.omp_outlined..4
5097 // CHECK4-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]]) #[[ATTR1]] {
5098 // CHECK4-NEXT: entry:
5099 // CHECK4-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
5100 // CHECK4-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
5101 // CHECK4-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
5102 // CHECK4-NEXT: [[TMP:%.*]] = alloca i32, align 4
5103 // CHECK4-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4
5104 // CHECK4-NEXT: [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4
5105 // CHECK4-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
5106 // CHECK4-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
5107 // CHECK4-NEXT: [[I:%.*]] = alloca i32, align 4
5108 // CHECK4-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
5109 // CHECK4-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
5110 // CHECK4-NEXT: store i32 0, i32* [[DOTOMP_COMB_LB]], align 4
5111 // CHECK4-NEXT: store i32 99, i32* [[DOTOMP_COMB_UB]], align 4
5112 // CHECK4-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4
5113 // CHECK4-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4
5114 // CHECK4-NEXT: [[TMP0:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
5115 // CHECK4-NEXT: [[TMP1:%.*]] = load i32, i32* [[TMP0]], align 4
5116 // CHECK4-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB1]], i32 [[TMP1]], i32 92, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_COMB_LB]], i32* [[DOTOMP_COMB_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1)
5117 // CHECK4-NEXT: [[TMP2:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
5118 // CHECK4-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP2]], 99
5119 // CHECK4-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
5120 // CHECK4: cond.true:
5121 // CHECK4-NEXT: br label [[COND_END:%.*]]
5122 // CHECK4: cond.false:
5123 // CHECK4-NEXT: [[TMP3:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
5124 // CHECK4-NEXT: br label [[COND_END]]
5125 // CHECK4: cond.end:
5126 // CHECK4-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP3]], [[COND_FALSE]] ]
5127 // CHECK4-NEXT: store i32 [[COND]], i32* [[DOTOMP_COMB_UB]], align 4
5128 // CHECK4-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4
5129 // CHECK4-NEXT: store i32 [[TMP4]], i32* [[DOTOMP_IV]], align 4
5130 // CHECK4-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
5131 // CHECK4: omp.inner.for.cond:
5132 // CHECK4-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
5133 // CHECK4-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
5134 // CHECK4-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP5]], [[TMP6]]
5135 // CHECK4-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
5136 // CHECK4: omp.inner.for.body:
5137 // CHECK4-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4
5138 // CHECK4-NEXT: [[TMP8:%.*]] = zext i32 [[TMP7]] to i64
5139 // CHECK4-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
5140 // CHECK4-NEXT: [[TMP10:%.*]] = zext i32 [[TMP9]] to i64
5141 // CHECK4-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB3]], i32 2, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64, i64)* @.omp_outlined..5 to void (i32*, i32*, ...)*), i64 [[TMP8]], i64 [[TMP10]])
5142 // CHECK4-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
5143 // CHECK4: omp.inner.for.inc:
5144 // CHECK4-NEXT: [[TMP11:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
5145 // CHECK4-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4
5146 // CHECK4-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP11]], [[TMP12]]
5147 // CHECK4-NEXT: store i32 [[ADD]], i32* [[DOTOMP_IV]], align 4
5148 // CHECK4-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP17:![0-9]+]]
5149 // CHECK4: omp.inner.for.end:
5150 // CHECK4-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
5151 // CHECK4: omp.loop.exit:
5152 // CHECK4-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP1]])
5153 // CHECK4-NEXT: [[TMP13:%.*]] = load i32, i32* [[DOTOMP_IS_LAST]], align 4
5154 // CHECK4-NEXT: [[TMP14:%.*]] = icmp ne i32 [[TMP13]], 0
5155 // CHECK4-NEXT: br i1 [[TMP14]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
5156 // CHECK4: .omp.final.then:
5157 // CHECK4-NEXT: store i32 100, i32* [[I]], align 4
5158 // CHECK4-NEXT: br label [[DOTOMP_FINAL_DONE]]
5159 // CHECK4: .omp.final.done:
5160 // CHECK4-NEXT: ret void
5161 //
5162 //
5163 // CHECK4-LABEL: define {{[^@]+}}@.omp_outlined..5
5164 // CHECK4-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[DOTPREVIOUS_LB_:%.*]], i64 [[DOTPREVIOUS_UB_:%.*]]) #[[ATTR1]] {
5165 // CHECK4-NEXT: entry:
5166 // CHECK4-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
5167 // CHECK4-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
5168 // CHECK4-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i64, align 8
5169 // CHECK4-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i64, align 8
5170 // CHECK4-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
5171 // CHECK4-NEXT: [[TMP:%.*]] = alloca i32, align 4
5172 // CHECK4-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4
5173 // CHECK4-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4
5174 // CHECK4-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
5175 // CHECK4-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
5176 // CHECK4-NEXT: [[I:%.*]] = alloca i32, align 4
5177 // CHECK4-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
5178 // CHECK4-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
5179 // CHECK4-NEXT: store i64 [[DOTPREVIOUS_LB_]], i64* [[DOTPREVIOUS_LB__ADDR]], align 8
5180 // CHECK4-NEXT: store i64 [[DOTPREVIOUS_UB_]], i64* [[DOTPREVIOUS_UB__ADDR]], align 8
5181 // CHECK4-NEXT: store i32 0, i32* [[DOTOMP_LB]], align 4
5182 // CHECK4-NEXT: store i32 99, i32* [[DOTOMP_UB]], align 4
5183 // CHECK4-NEXT: [[TMP0:%.*]] = load i64, i64* [[DOTPREVIOUS_LB__ADDR]], align 8
5184 // CHECK4-NEXT: [[CONV:%.*]] = trunc i64 [[TMP0]] to i32
5185 // CHECK4-NEXT: [[TMP1:%.*]] = load i64, i64* [[DOTPREVIOUS_UB__ADDR]], align 8
5186 // CHECK4-NEXT: [[CONV1:%.*]] = trunc i64 [[TMP1]] to i32
5187 // CHECK4-NEXT: store i32 [[CONV]], i32* [[DOTOMP_LB]], align 4
5188 // CHECK4-NEXT: store i32 [[CONV1]], i32* [[DOTOMP_UB]], align 4
5189 // CHECK4-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4
5190 // CHECK4-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4
5191 // CHECK4-NEXT: [[TMP2:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
5192 // CHECK4-NEXT: [[TMP3:%.*]] = load i32, i32* [[TMP2]], align 4
5193 // CHECK4-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB2]], i32 [[TMP3]], i32 34, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_LB]], i32* [[DOTOMP_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1)
5194 // CHECK4-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
5195 // CHECK4-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP4]], 99
5196 // CHECK4-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
5197 // CHECK4: cond.true:
5198 // CHECK4-NEXT: br label [[COND_END:%.*]]
5199 // CHECK4: cond.false:
5200 // CHECK4-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
5201 // CHECK4-NEXT: br label [[COND_END]]
5202 // CHECK4: cond.end:
5203 // CHECK4-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP5]], [[COND_FALSE]] ]
5204 // CHECK4-NEXT: store i32 [[COND]], i32* [[DOTOMP_UB]], align 4
5205 // CHECK4-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4
5206 // CHECK4-NEXT: store i32 [[TMP6]], i32* [[DOTOMP_IV]], align 4
5207 // CHECK4-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
5208 // CHECK4: omp.inner.for.cond:
5209 // CHECK4-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
5210 // CHECK4-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
5211 // CHECK4-NEXT: [[CMP2:%.*]] = icmp sle i32 [[TMP7]], [[TMP8]]
5212 // CHECK4-NEXT: br i1 [[CMP2]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
5213 // CHECK4: omp.inner.for.body:
5214 // CHECK4-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
5215 // CHECK4-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP9]], 1
5216 // CHECK4-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
5217 // CHECK4-NEXT: store i32 [[ADD]], i32* [[I]], align 4
5218 // CHECK4-NEXT: call void @_Z3fn4v()
5219 // CHECK4-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
5220 // CHECK4: omp.body.continue:
5221 // CHECK4-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
5222 // CHECK4: omp.inner.for.inc:
5223 // CHECK4-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
5224 // CHECK4-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP10]], 1
5225 // CHECK4-NEXT: store i32 [[ADD3]], i32* [[DOTOMP_IV]], align 4
5226 // CHECK4-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP18:![0-9]+]]
5227 // CHECK4: omp.inner.for.end:
5228 // CHECK4-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
5229 // CHECK4: omp.loop.exit:
5230 // CHECK4-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP3]])
5231 // CHECK4-NEXT: [[TMP11:%.*]] = load i32, i32* [[DOTOMP_IS_LAST]], align 4
5232 // CHECK4-NEXT: [[TMP12:%.*]] = icmp ne i32 [[TMP11]], 0
5233 // CHECK4-NEXT: br i1 [[TMP12]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
5234 // CHECK4: .omp.final.then:
5235 // CHECK4-NEXT: store i32 100, i32* [[I]], align 4
5236 // CHECK4-NEXT: br label [[DOTOMP_FINAL_DONE]]
5237 // CHECK4: .omp.final.done:
5238 // CHECK4-NEXT: ret void
5239 //
5240 //
5241 // CHECK4-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l84
5242 // CHECK4-SAME: () #[[ATTR1]] {
5243 // CHECK4-NEXT: entry:
5244 // CHECK4-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 0, void (i32*, i32*, ...)* bitcast (void (i32*, i32*)* @.omp_outlined..6 to void (i32*, i32*, ...)*))
5245 // CHECK4-NEXT: ret void
5246 //
5247 //
5248 // CHECK4-LABEL: define {{[^@]+}}@.omp_outlined..6
5249 // CHECK4-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]]) #[[ATTR1]] {
5250 // CHECK4-NEXT: entry:
5251 // CHECK4-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
5252 // CHECK4-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
5253 // CHECK4-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
5254 // CHECK4-NEXT: [[TMP:%.*]] = alloca i32, align 4
5255 // CHECK4-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4
5256 // CHECK4-NEXT: [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4
5257 // CHECK4-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
5258 // CHECK4-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
5259 // CHECK4-NEXT: [[I:%.*]] = alloca i32, align 4
5260 // CHECK4-NEXT: [[DOTBOUND_ZERO_ADDR:%.*]] = alloca i32, align 4
5261 // CHECK4-NEXT: store i32 0, i32* [[DOTBOUND_ZERO_ADDR]], align 4
5262 // CHECK4-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
5263 // CHECK4-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
5264 // CHECK4-NEXT: store i32 0, i32* [[DOTOMP_COMB_LB]], align 4
5265 // CHECK4-NEXT: store i32 99, i32* [[DOTOMP_COMB_UB]], align 4
5266 // CHECK4-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4
5267 // CHECK4-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4
5268 // CHECK4-NEXT: [[TMP0:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
5269 // CHECK4-NEXT: [[TMP1:%.*]] = load i32, i32* [[TMP0]], align 4
5270 // CHECK4-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB1]], i32 [[TMP1]], i32 92, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_COMB_LB]], i32* [[DOTOMP_COMB_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1)
5271 // CHECK4-NEXT: [[TMP2:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
5272 // CHECK4-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP2]], 99
5273 // CHECK4-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
5274 // CHECK4: cond.true:
5275 // CHECK4-NEXT: br label [[COND_END:%.*]]
5276 // CHECK4: cond.false:
5277 // CHECK4-NEXT: [[TMP3:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
5278 // CHECK4-NEXT: br label [[COND_END]]
5279 // CHECK4: cond.end:
5280 // CHECK4-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP3]], [[COND_FALSE]] ]
5281 // CHECK4-NEXT: store i32 [[COND]], i32* [[DOTOMP_COMB_UB]], align 4
5282 // CHECK4-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4
5283 // CHECK4-NEXT: store i32 [[TMP4]], i32* [[DOTOMP_IV]], align 4
5284 // CHECK4-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
5285 // CHECK4: omp.inner.for.cond:
5286 // CHECK4-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
5287 // CHECK4-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
5288 // CHECK4-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP5]], [[TMP6]]
5289 // CHECK4-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
5290 // CHECK4: omp.inner.for.body:
5291 // CHECK4-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4
5292 // CHECK4-NEXT: [[TMP8:%.*]] = zext i32 [[TMP7]] to i64
5293 // CHECK4-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
5294 // CHECK4-NEXT: [[TMP10:%.*]] = zext i32 [[TMP9]] to i64
5295 // CHECK4-NEXT: call void @__kmpc_serialized_parallel(%struct.ident_t* @[[GLOB3]], i32 [[TMP1]])
5296 // CHECK4-NEXT: [[TMP11:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
5297 // CHECK4-NEXT: call void @.omp_outlined..7(i32* [[TMP11]], i32* [[DOTBOUND_ZERO_ADDR]], i64 [[TMP8]], i64 [[TMP10]]) #[[ATTR2]]
5298 // CHECK4-NEXT: call void @__kmpc_end_serialized_parallel(%struct.ident_t* @[[GLOB3]], i32 [[TMP1]])
5299 // CHECK4-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
5300 // CHECK4: omp.inner.for.inc:
5301 // CHECK4-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
5302 // CHECK4-NEXT: [[TMP13:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4
5303 // CHECK4-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP12]], [[TMP13]]
5304 // CHECK4-NEXT: store i32 [[ADD]], i32* [[DOTOMP_IV]], align 4
5305 // CHECK4-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP19:![0-9]+]]
5306 // CHECK4: omp.inner.for.end:
5307 // CHECK4-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
5308 // CHECK4: omp.loop.exit:
5309 // CHECK4-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP1]])
5310 // CHECK4-NEXT: [[TMP14:%.*]] = load i32, i32* [[DOTOMP_IS_LAST]], align 4
5311 // CHECK4-NEXT: [[TMP15:%.*]] = icmp ne i32 [[TMP14]], 0
5312 // CHECK4-NEXT: br i1 [[TMP15]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
5313 // CHECK4: .omp.final.then:
5314 // CHECK4-NEXT: store i32 100, i32* [[I]], align 4
5315 // CHECK4-NEXT: br label [[DOTOMP_FINAL_DONE]]
5316 // CHECK4: .omp.final.done:
5317 // CHECK4-NEXT: ret void
5318 //
5319 //
5320 // CHECK4-LABEL: define {{[^@]+}}@.omp_outlined..7
5321 // CHECK4-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[DOTPREVIOUS_LB_:%.*]], i64 [[DOTPREVIOUS_UB_:%.*]]) #[[ATTR1]] {
5322 // CHECK4-NEXT: entry:
5323 // CHECK4-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
5324 // CHECK4-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
5325 // CHECK4-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i64, align 8
5326 // CHECK4-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i64, align 8
5327 // CHECK4-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
5328 // CHECK4-NEXT: [[TMP:%.*]] = alloca i32, align 4
5329 // CHECK4-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4
5330 // CHECK4-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4
5331 // CHECK4-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
5332 // CHECK4-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
5333 // CHECK4-NEXT: [[I:%.*]] = alloca i32, align 4
5334 // CHECK4-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
5335 // CHECK4-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
5336 // CHECK4-NEXT: store i64 [[DOTPREVIOUS_LB_]], i64* [[DOTPREVIOUS_LB__ADDR]], align 8
5337 // CHECK4-NEXT: store i64 [[DOTPREVIOUS_UB_]], i64* [[DOTPREVIOUS_UB__ADDR]], align 8
5338 // CHECK4-NEXT: store i32 0, i32* [[DOTOMP_LB]], align 4
5339 // CHECK4-NEXT: store i32 99, i32* [[DOTOMP_UB]], align 4
5340 // CHECK4-NEXT: [[TMP0:%.*]] = load i64, i64* [[DOTPREVIOUS_LB__ADDR]], align 8
5341 // CHECK4-NEXT: [[CONV:%.*]] = trunc i64 [[TMP0]] to i32
5342 // CHECK4-NEXT: [[TMP1:%.*]] = load i64, i64* [[DOTPREVIOUS_UB__ADDR]], align 8
5343 // CHECK4-NEXT: [[CONV1:%.*]] = trunc i64 [[TMP1]] to i32
5344 // CHECK4-NEXT: store i32 [[CONV]], i32* [[DOTOMP_LB]], align 4
5345 // CHECK4-NEXT: store i32 [[CONV1]], i32* [[DOTOMP_UB]], align 4
5346 // CHECK4-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4
5347 // CHECK4-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4
5348 // CHECK4-NEXT: [[TMP2:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
5349 // CHECK4-NEXT: [[TMP3:%.*]] = load i32, i32* [[TMP2]], align 4
5350 // CHECK4-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB2]], i32 [[TMP3]], i32 34, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_LB]], i32* [[DOTOMP_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1)
5351 // CHECK4-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
5352 // CHECK4-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP4]], 99
5353 // CHECK4-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
5354 // CHECK4: cond.true:
5355 // CHECK4-NEXT: br label [[COND_END:%.*]]
5356 // CHECK4: cond.false:
5357 // CHECK4-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
5358 // CHECK4-NEXT: br label [[COND_END]]
5359 // CHECK4: cond.end:
5360 // CHECK4-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP5]], [[COND_FALSE]] ]
5361 // CHECK4-NEXT: store i32 [[COND]], i32* [[DOTOMP_UB]], align 4
5362 // CHECK4-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4
5363 // CHECK4-NEXT: store i32 [[TMP6]], i32* [[DOTOMP_IV]], align 4
5364 // CHECK4-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
5365 // CHECK4: omp.inner.for.cond:
5366 // CHECK4-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
5367 // CHECK4-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
5368 // CHECK4-NEXT: [[CMP2:%.*]] = icmp sle i32 [[TMP7]], [[TMP8]]
5369 // CHECK4-NEXT: br i1 [[CMP2]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
5370 // CHECK4: omp.inner.for.body:
5371 // CHECK4-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
5372 // CHECK4-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP9]], 1
5373 // CHECK4-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
5374 // CHECK4-NEXT: store i32 [[ADD]], i32* [[I]], align 4
5375 // CHECK4-NEXT: call void @_Z3fn5v()
5376 // CHECK4-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
5377 // CHECK4: omp.body.continue:
5378 // CHECK4-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
5379 // CHECK4: omp.inner.for.inc:
5380 // CHECK4-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
5381 // CHECK4-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP10]], 1
5382 // CHECK4-NEXT: store i32 [[ADD3]], i32* [[DOTOMP_IV]], align 4
5383 // CHECK4-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP21:![0-9]+]]
5384 // CHECK4: omp.inner.for.end:
5385 // CHECK4-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
5386 // CHECK4: omp.loop.exit:
5387 // CHECK4-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP3]])
5388 // CHECK4-NEXT: [[TMP11:%.*]] = load i32, i32* [[DOTOMP_IS_LAST]], align 4
5389 // CHECK4-NEXT: [[TMP12:%.*]] = icmp ne i32 [[TMP11]], 0
5390 // CHECK4-NEXT: br i1 [[TMP12]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
5391 // CHECK4: .omp.final.then:
5392 // CHECK4-NEXT: store i32 100, i32* [[I]], align 4
5393 // CHECK4-NEXT: br label [[DOTOMP_FINAL_DONE]]
5394 // CHECK4: .omp.final.done:
5395 // CHECK4-NEXT: ret void
5396 //
5397 //
5398 // CHECK4-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l92
5399 // CHECK4-SAME: (i64 [[ARG:%.*]]) #[[ATTR1]] {
5400 // CHECK4-NEXT: entry:
5401 // CHECK4-NEXT: [[ARG_ADDR:%.*]] = alloca i64, align 8
5402 // CHECK4-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i8, align 1
5403 // CHECK4-NEXT: [[DOTCAPTURE_EXPR__CASTED:%.*]] = alloca i64, align 8
5404 // CHECK4-NEXT: store i64 [[ARG]], i64* [[ARG_ADDR]], align 8
5405 // CHECK4-NEXT: [[CONV:%.*]] = bitcast i64* [[ARG_ADDR]] to i32*
5406 // CHECK4-NEXT: [[TMP0:%.*]] = load i32, i32* [[CONV]], align 8
5407 // CHECK4-NEXT: [[TOBOOL:%.*]] = icmp ne i32 [[TMP0]], 0
5408 // CHECK4-NEXT: [[FROMBOOL:%.*]] = zext i1 [[TOBOOL]] to i8
5409 // CHECK4-NEXT: store i8 [[FROMBOOL]], i8* [[DOTCAPTURE_EXPR_]], align 1
5410 // CHECK4-NEXT: [[TMP1:%.*]] = load i8, i8* [[DOTCAPTURE_EXPR_]], align 1
5411 // CHECK4-NEXT: [[TOBOOL1:%.*]] = trunc i8 [[TMP1]] to i1
5412 // CHECK4-NEXT: [[CONV2:%.*]] = bitcast i64* [[DOTCAPTURE_EXPR__CASTED]] to i8*
5413 // CHECK4-NEXT: [[FROMBOOL3:%.*]] = zext i1 [[TOBOOL1]] to i8
5414 // CHECK4-NEXT: store i8 [[FROMBOOL3]], i8* [[CONV2]], align 1
5415 // CHECK4-NEXT: [[TMP2:%.*]] = load i64, i64* [[DOTCAPTURE_EXPR__CASTED]], align 8
5416 // CHECK4-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64)* @.omp_outlined..8 to void (i32*, i32*, ...)*), i64 [[TMP2]])
5417 // CHECK4-NEXT: ret void
5418 //
5419 //
5420 // CHECK4-LABEL: define {{[^@]+}}@.omp_outlined..8
5421 // CHECK4-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[DOTCAPTURE_EXPR_:%.*]]) #[[ATTR1]] {
5422 // CHECK4-NEXT: entry:
5423 // CHECK4-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
5424 // CHECK4-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
5425 // CHECK4-NEXT: [[DOTCAPTURE_EXPR__ADDR:%.*]] = alloca i64, align 8
5426 // CHECK4-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
5427 // CHECK4-NEXT: [[TMP:%.*]] = alloca i32, align 4
5428 // CHECK4-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4
5429 // CHECK4-NEXT: [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4
5430 // CHECK4-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
5431 // CHECK4-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
5432 // CHECK4-NEXT: [[I:%.*]] = alloca i32, align 4
5433 // CHECK4-NEXT: [[DOTCAPTURE_EXPR__CASTED:%.*]] = alloca i64, align 8
5434 // CHECK4-NEXT: [[DOTBOUND_ZERO_ADDR:%.*]] = alloca i32, align 4
5435 // CHECK4-NEXT: [[DOTCAPTURE_EXPR__CASTED11:%.*]] = alloca i64, align 8
5436 // CHECK4-NEXT: [[DOTBOUND_ZERO_ADDR17:%.*]] = alloca i32, align 4
5437 // CHECK4-NEXT: store i32 0, i32* [[DOTBOUND_ZERO_ADDR17]], align 4
5438 // CHECK4-NEXT: store i32 0, i32* [[DOTBOUND_ZERO_ADDR]], align 4
5439 // CHECK4-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
5440 // CHECK4-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
5441 // CHECK4-NEXT: store i64 [[DOTCAPTURE_EXPR_]], i64* [[DOTCAPTURE_EXPR__ADDR]], align 8
5442 // CHECK4-NEXT: [[CONV:%.*]] = bitcast i64* [[DOTCAPTURE_EXPR__ADDR]] to i8*
5443 // CHECK4-NEXT: store i32 0, i32* [[DOTOMP_COMB_LB]], align 4
5444 // CHECK4-NEXT: store i32 99, i32* [[DOTOMP_COMB_UB]], align 4
5445 // CHECK4-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4
5446 // CHECK4-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4
5447 // CHECK4-NEXT: [[TMP0:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
5448 // CHECK4-NEXT: [[TMP1:%.*]] = load i32, i32* [[TMP0]], align 4
5449 // CHECK4-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB1]], i32 [[TMP1]], i32 92, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_COMB_LB]], i32* [[DOTOMP_COMB_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1)
5450 // CHECK4-NEXT: [[TMP2:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
5451 // CHECK4-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP2]], 99
5452 // CHECK4-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
5453 // CHECK4: cond.true:
5454 // CHECK4-NEXT: br label [[COND_END:%.*]]
5455 // CHECK4: cond.false:
5456 // CHECK4-NEXT: [[TMP3:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
5457 // CHECK4-NEXT: br label [[COND_END]]
5458 // CHECK4: cond.end:
5459 // CHECK4-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP3]], [[COND_FALSE]] ]
5460 // CHECK4-NEXT: store i32 [[COND]], i32* [[DOTOMP_COMB_UB]], align 4
5461 // CHECK4-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4
5462 // CHECK4-NEXT: store i32 [[TMP4]], i32* [[DOTOMP_IV]], align 4
5463 // CHECK4-NEXT: [[TMP5:%.*]] = load i8, i8* [[CONV]], align 8
5464 // CHECK4-NEXT: [[TOBOOL:%.*]] = trunc i8 [[TMP5]] to i1
5465 // CHECK4-NEXT: br i1 [[TOBOOL]], label [[OMP_IF_THEN:%.*]], label [[OMP_IF_ELSE6:%.*]]
5466 // CHECK4: omp_if.then:
5467 // CHECK4-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
5468 // CHECK4: omp.inner.for.cond:
5469 // CHECK4-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
5470 // CHECK4-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
5471 // CHECK4-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP6]], [[TMP7]]
5472 // CHECK4-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
5473 // CHECK4: omp.inner.for.body:
5474 // CHECK4-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4
5475 // CHECK4-NEXT: [[TMP9:%.*]] = zext i32 [[TMP8]] to i64
5476 // CHECK4-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
5477 // CHECK4-NEXT: [[TMP11:%.*]] = zext i32 [[TMP10]] to i64
5478 // CHECK4-NEXT: [[TMP12:%.*]] = load i8, i8* [[CONV]], align 8
5479 // CHECK4-NEXT: [[TOBOOL2:%.*]] = trunc i8 [[TMP12]] to i1
5480 // CHECK4-NEXT: [[CONV3:%.*]] = bitcast i64* [[DOTCAPTURE_EXPR__CASTED]] to i8*
5481 // CHECK4-NEXT: [[FROMBOOL:%.*]] = zext i1 [[TOBOOL2]] to i8
5482 // CHECK4-NEXT: store i8 [[FROMBOOL]], i8* [[CONV3]], align 1
5483 // CHECK4-NEXT: [[TMP13:%.*]] = load i64, i64* [[DOTCAPTURE_EXPR__CASTED]], align 8
5484 // CHECK4-NEXT: [[TMP14:%.*]] = load i8, i8* [[CONV]], align 8
5485 // CHECK4-NEXT: [[TOBOOL4:%.*]] = trunc i8 [[TMP14]] to i1
5486 // CHECK4-NEXT: br i1 [[TOBOOL4]], label [[OMP_IF_THEN5:%.*]], label [[OMP_IF_ELSE:%.*]]
5487 // CHECK4: omp_if.then5:
5488 // CHECK4-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB3]], i32 3, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64, i64, i64)* @.omp_outlined..9 to void (i32*, i32*, ...)*), i64 [[TMP9]], i64 [[TMP11]], i64 [[TMP13]])
5489 // CHECK4-NEXT: br label [[OMP_IF_END:%.*]]
5490 // CHECK4: omp_if.else:
5491 // CHECK4-NEXT: call void @__kmpc_serialized_parallel(%struct.ident_t* @[[GLOB3]], i32 [[TMP1]])
5492 // CHECK4-NEXT: [[TMP15:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
5493 // CHECK4-NEXT: call void @.omp_outlined..9(i32* [[TMP15]], i32* [[DOTBOUND_ZERO_ADDR]], i64 [[TMP9]], i64 [[TMP11]], i64 [[TMP13]]) #[[ATTR2]]
5494 // CHECK4-NEXT: call void @__kmpc_end_serialized_parallel(%struct.ident_t* @[[GLOB3]], i32 [[TMP1]])
5495 // CHECK4-NEXT: br label [[OMP_IF_END]]
5496 // CHECK4: omp_if.end:
5497 // CHECK4-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
5498 // CHECK4: omp.inner.for.inc:
5499 // CHECK4-NEXT: [[TMP16:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
5500 // CHECK4-NEXT: [[TMP17:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4
5501 // CHECK4-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP16]], [[TMP17]]
5502 // CHECK4-NEXT: store i32 [[ADD]], i32* [[DOTOMP_IV]], align 4
5503 // CHECK4-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP22:![0-9]+]]
5504 // CHECK4: omp.inner.for.end:
5505 // CHECK4-NEXT: br label [[OMP_IF_END22:%.*]]
5506 // CHECK4: omp_if.else6:
5507 // CHECK4-NEXT: br label [[OMP_INNER_FOR_COND7:%.*]]
5508 // CHECK4: omp.inner.for.cond7:
5509 // CHECK4-NEXT: [[TMP18:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
5510 // CHECK4-NEXT: [[TMP19:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
5511 // CHECK4-NEXT: [[CMP8:%.*]] = icmp sle i32 [[TMP18]], [[TMP19]]
5512 // CHECK4-NEXT: br i1 [[CMP8]], label [[OMP_INNER_FOR_BODY9:%.*]], label [[OMP_INNER_FOR_END21:%.*]]
5513 // CHECK4: omp.inner.for.body9:
5514 // CHECK4-NEXT: [[TMP20:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4
5515 // CHECK4-NEXT: [[TMP21:%.*]] = zext i32 [[TMP20]] to i64
5516 // CHECK4-NEXT: [[TMP22:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
5517 // CHECK4-NEXT: [[TMP23:%.*]] = zext i32 [[TMP22]] to i64
5518 // CHECK4-NEXT: [[TMP24:%.*]] = load i8, i8* [[CONV]], align 8
5519 // CHECK4-NEXT: [[TOBOOL10:%.*]] = trunc i8 [[TMP24]] to i1
5520 // CHECK4-NEXT: [[CONV12:%.*]] = bitcast i64* [[DOTCAPTURE_EXPR__CASTED11]] to i8*
5521 // CHECK4-NEXT: [[FROMBOOL13:%.*]] = zext i1 [[TOBOOL10]] to i8
5522 // CHECK4-NEXT: store i8 [[FROMBOOL13]], i8* [[CONV12]], align 1
5523 // CHECK4-NEXT: [[TMP25:%.*]] = load i64, i64* [[DOTCAPTURE_EXPR__CASTED11]], align 8
5524 // CHECK4-NEXT: [[TMP26:%.*]] = load i8, i8* [[CONV]], align 8
5525 // CHECK4-NEXT: [[TOBOOL14:%.*]] = trunc i8 [[TMP26]] to i1
5526 // CHECK4-NEXT: br i1 [[TOBOOL14]], label [[OMP_IF_THEN15:%.*]], label [[OMP_IF_ELSE16:%.*]]
5527 // CHECK4: omp_if.then15:
5528 // CHECK4-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB3]], i32 3, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64, i64, i64)* @.omp_outlined..10 to void (i32*, i32*, ...)*), i64 [[TMP21]], i64 [[TMP23]], i64 [[TMP25]])
5529 // CHECK4-NEXT: br label [[OMP_IF_END18:%.*]]
5530 // CHECK4: omp_if.else16:
5531 // CHECK4-NEXT: call void @__kmpc_serialized_parallel(%struct.ident_t* @[[GLOB3]], i32 [[TMP1]])
5532 // CHECK4-NEXT: [[TMP27:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
5533 // CHECK4-NEXT: call void @.omp_outlined..10(i32* [[TMP27]], i32* [[DOTBOUND_ZERO_ADDR17]], i64 [[TMP21]], i64 [[TMP23]], i64 [[TMP25]]) #[[ATTR2]]
5534 // CHECK4-NEXT: call void @__kmpc_end_serialized_parallel(%struct.ident_t* @[[GLOB3]], i32 [[TMP1]])
5535 // CHECK4-NEXT: br label [[OMP_IF_END18]]
5536 // CHECK4: omp_if.end18:
5537 // CHECK4-NEXT: br label [[OMP_INNER_FOR_INC19:%.*]]
5538 // CHECK4: omp.inner.for.inc19:
5539 // CHECK4-NEXT: [[TMP28:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
5540 // CHECK4-NEXT: [[TMP29:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4
5541 // CHECK4-NEXT: [[ADD20:%.*]] = add nsw i32 [[TMP28]], [[TMP29]]
5542 // CHECK4-NEXT: store i32 [[ADD20]], i32* [[DOTOMP_IV]], align 4
5543 // CHECK4-NEXT: br label [[OMP_INNER_FOR_COND7]], !llvm.loop [[LOOP23:![0-9]+]]
5544 // CHECK4: omp.inner.for.end21:
5545 // CHECK4-NEXT: br label [[OMP_IF_END22]]
5546 // CHECK4: omp_if.end22:
5547 // CHECK4-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
5548 // CHECK4: omp.loop.exit:
5549 // CHECK4-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP1]])
5550 // CHECK4-NEXT: [[TMP30:%.*]] = load i32, i32* [[DOTOMP_IS_LAST]], align 4
5551 // CHECK4-NEXT: [[TMP31:%.*]] = icmp ne i32 [[TMP30]], 0
5552 // CHECK4-NEXT: br i1 [[TMP31]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
5553 // CHECK4: .omp.final.then:
5554 // CHECK4-NEXT: store i32 100, i32* [[I]], align 4
5555 // CHECK4-NEXT: br label [[DOTOMP_FINAL_DONE]]
5556 // CHECK4: .omp.final.done:
5557 // CHECK4-NEXT: ret void
5558 //
5559 //
5560 // CHECK4-LABEL: define {{[^@]+}}@.omp_outlined..9
5561 // CHECK4-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[DOTPREVIOUS_LB_:%.*]], i64 [[DOTPREVIOUS_UB_:%.*]], i64 [[DOTCAPTURE_EXPR_:%.*]]) #[[ATTR1]] {
5562 // CHECK4-NEXT: entry:
5563 // CHECK4-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
5564 // CHECK4-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
5565 // CHECK4-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i64, align 8
5566 // CHECK4-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i64, align 8
5567 // CHECK4-NEXT: [[DOTCAPTURE_EXPR__ADDR:%.*]] = alloca i64, align 8
5568 // CHECK4-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
5569 // CHECK4-NEXT: [[TMP:%.*]] = alloca i32, align 4
5570 // CHECK4-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4
5571 // CHECK4-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4
5572 // CHECK4-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
5573 // CHECK4-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
5574 // CHECK4-NEXT: [[I:%.*]] = alloca i32, align 4
5575 // CHECK4-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
5576 // CHECK4-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
5577 // CHECK4-NEXT: store i64 [[DOTPREVIOUS_LB_]], i64* [[DOTPREVIOUS_LB__ADDR]], align 8
5578 // CHECK4-NEXT: store i64 [[DOTPREVIOUS_UB_]], i64* [[DOTPREVIOUS_UB__ADDR]], align 8
5579 // CHECK4-NEXT: store i64 [[DOTCAPTURE_EXPR_]], i64* [[DOTCAPTURE_EXPR__ADDR]], align 8
5580 // CHECK4-NEXT: [[CONV:%.*]] = bitcast i64* [[DOTCAPTURE_EXPR__ADDR]] to i8*
5581 // CHECK4-NEXT: store i32 0, i32* [[DOTOMP_LB]], align 4
5582 // CHECK4-NEXT: store i32 99, i32* [[DOTOMP_UB]], align 4
5583 // CHECK4-NEXT: [[TMP0:%.*]] = load i64, i64* [[DOTPREVIOUS_LB__ADDR]], align 8
5584 // CHECK4-NEXT: [[CONV1:%.*]] = trunc i64 [[TMP0]] to i32
5585 // CHECK4-NEXT: [[TMP1:%.*]] = load i64, i64* [[DOTPREVIOUS_UB__ADDR]], align 8
5586 // CHECK4-NEXT: [[CONV2:%.*]] = trunc i64 [[TMP1]] to i32
5587 // CHECK4-NEXT: store i32 [[CONV1]], i32* [[DOTOMP_LB]], align 4
5588 // CHECK4-NEXT: store i32 [[CONV2]], i32* [[DOTOMP_UB]], align 4
5589 // CHECK4-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4
5590 // CHECK4-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4
5591 // CHECK4-NEXT: [[TMP2:%.*]] = load i8, i8* [[CONV]], align 8
5592 // CHECK4-NEXT: [[TOBOOL:%.*]] = trunc i8 [[TMP2]] to i1
5593 // CHECK4-NEXT: br i1 [[TOBOOL]], label [[OMP_IF_THEN:%.*]], label [[OMP_IF_ELSE:%.*]]
5594 // CHECK4: omp_if.then:
5595 // CHECK4-NEXT: [[TMP3:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
5596 // CHECK4-NEXT: [[TMP4:%.*]] = load i32, i32* [[TMP3]], align 4
5597 // CHECK4-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB2]], i32 [[TMP4]], i32 34, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_LB]], i32* [[DOTOMP_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1)
5598 // CHECK4-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
5599 // CHECK4-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP5]], 99
5600 // CHECK4-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
5601 // CHECK4: cond.true:
5602 // CHECK4-NEXT: br label [[COND_END:%.*]]
5603 // CHECK4: cond.false:
5604 // CHECK4-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
5605 // CHECK4-NEXT: br label [[COND_END]]
5606 // CHECK4: cond.end:
5607 // CHECK4-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP6]], [[COND_FALSE]] ]
5608 // CHECK4-NEXT: store i32 [[COND]], i32* [[DOTOMP_UB]], align 4
5609 // CHECK4-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4
5610 // CHECK4-NEXT: store i32 [[TMP7]], i32* [[DOTOMP_IV]], align 4
5611 // CHECK4-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
5612 // CHECK4: omp.inner.for.cond:
5613 // CHECK4-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
5614 // CHECK4-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
5615 // CHECK4-NEXT: [[CMP3:%.*]] = icmp sle i32 [[TMP8]], [[TMP9]]
5616 // CHECK4-NEXT: br i1 [[CMP3]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
5617 // CHECK4: omp.inner.for.body:
5618 // CHECK4-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
5619 // CHECK4-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP10]], 1
5620 // CHECK4-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
5621 // CHECK4-NEXT: store i32 [[ADD]], i32* [[I]], align 4
5622 // CHECK4-NEXT: call void @_Z3fn6v()
5623 // CHECK4-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
5624 // CHECK4: omp.body.continue:
5625 // CHECK4-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
5626 // CHECK4: omp.inner.for.inc:
5627 // CHECK4-NEXT: [[TMP11:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
5628 // CHECK4-NEXT: [[ADD4:%.*]] = add nsw i32 [[TMP11]], 1
5629 // CHECK4-NEXT: store i32 [[ADD4]], i32* [[DOTOMP_IV]], align 4
5630 // CHECK4-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP24:![0-9]+]]
5631 // CHECK4: omp.inner.for.end:
5632 // CHECK4-NEXT: br label [[OMP_IF_END:%.*]]
5633 // CHECK4: omp_if.else:
5634 // CHECK4-NEXT: [[TMP12:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
5635 // CHECK4-NEXT: [[TMP13:%.*]] = load i32, i32* [[TMP12]], align 4
5636 // CHECK4-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB2]], i32 [[TMP13]], i32 34, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_LB]], i32* [[DOTOMP_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1)
5637 // CHECK4-NEXT: [[TMP14:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
5638 // CHECK4-NEXT: [[CMP5:%.*]] = icmp sgt i32 [[TMP14]], 99
5639 // CHECK4-NEXT: br i1 [[CMP5]], label [[COND_TRUE6:%.*]], label [[COND_FALSE7:%.*]]
5640 // CHECK4: cond.true6:
5641 // CHECK4-NEXT: br label [[COND_END8:%.*]]
5642 // CHECK4: cond.false7:
5643 // CHECK4-NEXT: [[TMP15:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
5644 // CHECK4-NEXT: br label [[COND_END8]]
5645 // CHECK4: cond.end8:
5646 // CHECK4-NEXT: [[COND9:%.*]] = phi i32 [ 99, [[COND_TRUE6]] ], [ [[TMP15]], [[COND_FALSE7]] ]
5647 // CHECK4-NEXT: store i32 [[COND9]], i32* [[DOTOMP_UB]], align 4
5648 // CHECK4-NEXT: [[TMP16:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4
5649 // CHECK4-NEXT: store i32 [[TMP16]], i32* [[DOTOMP_IV]], align 4
5650 // CHECK4-NEXT: br label [[OMP_INNER_FOR_COND10:%.*]]
5651 // CHECK4: omp.inner.for.cond10:
5652 // CHECK4-NEXT: [[TMP17:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
5653 // CHECK4-NEXT: [[TMP18:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
5654 // CHECK4-NEXT: [[CMP11:%.*]] = icmp sle i32 [[TMP17]], [[TMP18]]
5655 // CHECK4-NEXT: br i1 [[CMP11]], label [[OMP_INNER_FOR_BODY12:%.*]], label [[OMP_INNER_FOR_END18:%.*]]
5656 // CHECK4: omp.inner.for.body12:
5657 // CHECK4-NEXT: [[TMP19:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
5658 // CHECK4-NEXT: [[MUL13:%.*]] = mul nsw i32 [[TMP19]], 1
5659 // CHECK4-NEXT: [[ADD14:%.*]] = add nsw i32 0, [[MUL13]]
5660 // CHECK4-NEXT: store i32 [[ADD14]], i32* [[I]], align 4
5661 // CHECK4-NEXT: call void @_Z3fn6v()
5662 // CHECK4-NEXT: br label [[OMP_BODY_CONTINUE15:%.*]]
5663 // CHECK4: omp.body.continue15:
5664 // CHECK4-NEXT: br label [[OMP_INNER_FOR_INC16:%.*]]
5665 // CHECK4: omp.inner.for.inc16:
5666 // CHECK4-NEXT: [[TMP20:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
5667 // CHECK4-NEXT: [[ADD17:%.*]] = add nsw i32 [[TMP20]], 1
5668 // CHECK4-NEXT: store i32 [[ADD17]], i32* [[DOTOMP_IV]], align 4
5669 // CHECK4-NEXT: br label [[OMP_INNER_FOR_COND10]], !llvm.loop [[LOOP25:![0-9]+]]
5670 // CHECK4: omp.inner.for.end18:
5671 // CHECK4-NEXT: br label [[OMP_IF_END]]
5672 // CHECK4: omp_if.end:
5673 // CHECK4-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
5674 // CHECK4: omp.loop.exit:
5675 // CHECK4-NEXT: [[TMP21:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
5676 // CHECK4-NEXT: [[TMP22:%.*]] = load i32, i32* [[TMP21]], align 4
5677 // CHECK4-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP22]])
5678 // CHECK4-NEXT: [[TMP23:%.*]] = load i32, i32* [[DOTOMP_IS_LAST]], align 4
5679 // CHECK4-NEXT: [[TMP24:%.*]] = icmp ne i32 [[TMP23]], 0
5680 // CHECK4-NEXT: br i1 [[TMP24]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
5681 // CHECK4: .omp.final.then:
5682 // CHECK4-NEXT: store i32 100, i32* [[I]], align 4
5683 // CHECK4-NEXT: br label [[DOTOMP_FINAL_DONE]]
5684 // CHECK4: .omp.final.done:
5685 // CHECK4-NEXT: ret void
5686 //
5687 //
5688 // CHECK4-LABEL: define {{[^@]+}}@.omp_outlined..10
5689 // CHECK4-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[DOTPREVIOUS_LB_:%.*]], i64 [[DOTPREVIOUS_UB_:%.*]], i64 [[DOTCAPTURE_EXPR_:%.*]]) #[[ATTR1]] {
5690 // CHECK4-NEXT: entry:
5691 // CHECK4-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
5692 // CHECK4-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
5693 // CHECK4-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i64, align 8
5694 // CHECK4-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i64, align 8
5695 // CHECK4-NEXT: [[DOTCAPTURE_EXPR__ADDR:%.*]] = alloca i64, align 8
5696 // CHECK4-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
5697 // CHECK4-NEXT: [[TMP:%.*]] = alloca i32, align 4
5698 // CHECK4-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4
5699 // CHECK4-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4
5700 // CHECK4-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
5701 // CHECK4-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
5702 // CHECK4-NEXT: [[I:%.*]] = alloca i32, align 4
5703 // CHECK4-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
5704 // CHECK4-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
5705 // CHECK4-NEXT: store i64 [[DOTPREVIOUS_LB_]], i64* [[DOTPREVIOUS_LB__ADDR]], align 8
5706 // CHECK4-NEXT: store i64 [[DOTPREVIOUS_UB_]], i64* [[DOTPREVIOUS_UB__ADDR]], align 8
5707 // CHECK4-NEXT: store i64 [[DOTCAPTURE_EXPR_]], i64* [[DOTCAPTURE_EXPR__ADDR]], align 8
5708 // CHECK4-NEXT: [[CONV:%.*]] = bitcast i64* [[DOTCAPTURE_EXPR__ADDR]] to i8*
5709 // CHECK4-NEXT: store i32 0, i32* [[DOTOMP_LB]], align 4
5710 // CHECK4-NEXT: store i32 99, i32* [[DOTOMP_UB]], align 4
5711 // CHECK4-NEXT: [[TMP0:%.*]] = load i64, i64* [[DOTPREVIOUS_LB__ADDR]], align 8
5712 // CHECK4-NEXT: [[CONV1:%.*]] = trunc i64 [[TMP0]] to i32
5713 // CHECK4-NEXT: [[TMP1:%.*]] = load i64, i64* [[DOTPREVIOUS_UB__ADDR]], align 8
5714 // CHECK4-NEXT: [[CONV2:%.*]] = trunc i64 [[TMP1]] to i32
5715 // CHECK4-NEXT: store i32 [[CONV1]], i32* [[DOTOMP_LB]], align 4
5716 // CHECK4-NEXT: store i32 [[CONV2]], i32* [[DOTOMP_UB]], align 4
5717 // CHECK4-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4
5718 // CHECK4-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4
5719 // CHECK4-NEXT: [[TMP2:%.*]] = load i8, i8* [[CONV]], align 8
5720 // CHECK4-NEXT: [[TOBOOL:%.*]] = trunc i8 [[TMP2]] to i1
5721 // CHECK4-NEXT: br i1 [[TOBOOL]], label [[OMP_IF_THEN:%.*]], label [[OMP_IF_ELSE:%.*]]
5722 // CHECK4: omp_if.then:
5723 // CHECK4-NEXT: [[TMP3:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
5724 // CHECK4-NEXT: [[TMP4:%.*]] = load i32, i32* [[TMP3]], align 4
5725 // CHECK4-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB2]], i32 [[TMP4]], i32 34, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_LB]], i32* [[DOTOMP_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1)
5726 // CHECK4-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
5727 // CHECK4-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP5]], 99
5728 // CHECK4-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
5729 // CHECK4: cond.true:
5730 // CHECK4-NEXT: br label [[COND_END:%.*]]
5731 // CHECK4: cond.false:
5732 // CHECK4-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
5733 // CHECK4-NEXT: br label [[COND_END]]
5734 // CHECK4: cond.end:
5735 // CHECK4-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP6]], [[COND_FALSE]] ]
5736 // CHECK4-NEXT: store i32 [[COND]], i32* [[DOTOMP_UB]], align 4
5737 // CHECK4-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4
5738 // CHECK4-NEXT: store i32 [[TMP7]], i32* [[DOTOMP_IV]], align 4
5739 // CHECK4-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
5740 // CHECK4: omp.inner.for.cond:
5741 // CHECK4-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
5742 // CHECK4-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
5743 // CHECK4-NEXT: [[CMP3:%.*]] = icmp sle i32 [[TMP8]], [[TMP9]]
5744 // CHECK4-NEXT: br i1 [[CMP3]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
5745 // CHECK4: omp.inner.for.body:
5746 // CHECK4-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
5747 // CHECK4-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP10]], 1
5748 // CHECK4-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
5749 // CHECK4-NEXT: store i32 [[ADD]], i32* [[I]], align 4
5750 // CHECK4-NEXT: call void @_Z3fn6v()
5751 // CHECK4-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
5752 // CHECK4: omp.body.continue:
5753 // CHECK4-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
5754 // CHECK4: omp.inner.for.inc:
5755 // CHECK4-NEXT: [[TMP11:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
5756 // CHECK4-NEXT: [[ADD4:%.*]] = add nsw i32 [[TMP11]], 1
5757 // CHECK4-NEXT: store i32 [[ADD4]], i32* [[DOTOMP_IV]], align 4
5758 // CHECK4-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP26:![0-9]+]]
5759 // CHECK4: omp.inner.for.end:
5760 // CHECK4-NEXT: br label [[OMP_IF_END:%.*]]
5761 // CHECK4: omp_if.else:
5762 // CHECK4-NEXT: [[TMP12:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
5763 // CHECK4-NEXT: [[TMP13:%.*]] = load i32, i32* [[TMP12]], align 4
5764 // CHECK4-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB2]], i32 [[TMP13]], i32 34, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_LB]], i32* [[DOTOMP_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1)
5765 // CHECK4-NEXT: [[TMP14:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
5766 // CHECK4-NEXT: [[CMP5:%.*]] = icmp sgt i32 [[TMP14]], 99
5767 // CHECK4-NEXT: br i1 [[CMP5]], label [[COND_TRUE6:%.*]], label [[COND_FALSE7:%.*]]
5768 // CHECK4: cond.true6:
5769 // CHECK4-NEXT: br label [[COND_END8:%.*]]
5770 // CHECK4: cond.false7:
5771 // CHECK4-NEXT: [[TMP15:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
5772 // CHECK4-NEXT: br label [[COND_END8]]
5773 // CHECK4: cond.end8:
5774 // CHECK4-NEXT: [[COND9:%.*]] = phi i32 [ 99, [[COND_TRUE6]] ], [ [[TMP15]], [[COND_FALSE7]] ]
5775 // CHECK4-NEXT: store i32 [[COND9]], i32* [[DOTOMP_UB]], align 4
5776 // CHECK4-NEXT: [[TMP16:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4
5777 // CHECK4-NEXT: store i32 [[TMP16]], i32* [[DOTOMP_IV]], align 4
5778 // CHECK4-NEXT: br label [[OMP_INNER_FOR_COND10:%.*]]
5779 // CHECK4: omp.inner.for.cond10:
5780 // CHECK4-NEXT: [[TMP17:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
5781 // CHECK4-NEXT: [[TMP18:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
5782 // CHECK4-NEXT: [[CMP11:%.*]] = icmp sle i32 [[TMP17]], [[TMP18]]
5783 // CHECK4-NEXT: br i1 [[CMP11]], label [[OMP_INNER_FOR_BODY12:%.*]], label [[OMP_INNER_FOR_END18:%.*]]
5784 // CHECK4: omp.inner.for.body12:
5785 // CHECK4-NEXT: [[TMP19:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
5786 // CHECK4-NEXT: [[MUL13:%.*]] = mul nsw i32 [[TMP19]], 1
5787 // CHECK4-NEXT: [[ADD14:%.*]] = add nsw i32 0, [[MUL13]]
5788 // CHECK4-NEXT: store i32 [[ADD14]], i32* [[I]], align 4
5789 // CHECK4-NEXT: call void @_Z3fn6v()
5790 // CHECK4-NEXT: br label [[OMP_BODY_CONTINUE15:%.*]]
5791 // CHECK4: omp.body.continue15:
5792 // CHECK4-NEXT: br label [[OMP_INNER_FOR_INC16:%.*]]
5793 // CHECK4: omp.inner.for.inc16:
5794 // CHECK4-NEXT: [[TMP20:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
5795 // CHECK4-NEXT: [[ADD17:%.*]] = add nsw i32 [[TMP20]], 1
5796 // CHECK4-NEXT: store i32 [[ADD17]], i32* [[DOTOMP_IV]], align 4
5797 // CHECK4-NEXT: br label [[OMP_INNER_FOR_COND10]], !llvm.loop [[LOOP27:![0-9]+]]
5798 // CHECK4: omp.inner.for.end18:
5799 // CHECK4-NEXT: br label [[OMP_IF_END]]
5800 // CHECK4: omp_if.end:
5801 // CHECK4-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
5802 // CHECK4: omp.loop.exit:
5803 // CHECK4-NEXT: [[TMP21:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
5804 // CHECK4-NEXT: [[TMP22:%.*]] = load i32, i32* [[TMP21]], align 4
5805 // CHECK4-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP22]])
5806 // CHECK4-NEXT: [[TMP23:%.*]] = load i32, i32* [[DOTOMP_IS_LAST]], align 4
5807 // CHECK4-NEXT: [[TMP24:%.*]] = icmp ne i32 [[TMP23]], 0
5808 // CHECK4-NEXT: br i1 [[TMP24]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
5809 // CHECK4: .omp.final.then:
5810 // CHECK4-NEXT: store i32 100, i32* [[I]], align 4
5811 // CHECK4-NEXT: br label [[DOTOMP_FINAL_DONE]]
5812 // CHECK4: .omp.final.done:
5813 // CHECK4-NEXT: ret void
5814 //
5815 //
5816 // CHECK4-LABEL: define {{[^@]+}}@_Z5tmainIiEiT_
5817 // CHECK4-SAME: (i32 [[ARG:%.*]]) #[[ATTR0]] comdat {
5818 // CHECK4-NEXT: entry:
5819 // CHECK4-NEXT: [[ARG_ADDR:%.*]] = alloca i32, align 4
5820 // CHECK4-NEXT: [[TMP:%.*]] = alloca i32, align 4
5821 // CHECK4-NEXT: [[_TMP1:%.*]] = alloca i32, align 4
5822 // CHECK4-NEXT: [[ARG_CASTED:%.*]] = alloca i64, align 8
5823 // CHECK4-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [1 x i8*], align 8
5824 // CHECK4-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [1 x i8*], align 8
5825 // CHECK4-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [1 x i8*], align 8
5826 // CHECK4-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i8, align 1
5827 // CHECK4-NEXT: [[_TMP5:%.*]] = alloca i32, align 4
5828 // CHECK4-NEXT: store i32 [[ARG]], i32* [[ARG_ADDR]], align 4
5829 // CHECK4-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i64 100)
5830 // CHECK4-NEXT: [[TMP0:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiEiT__l57.region_id, i32 0, i8** null, i8** null, i64* null, i64* null, i8** null, i8** null, i32 0, i32 0)
5831 // CHECK4-NEXT: [[TMP1:%.*]] = icmp ne i32 [[TMP0]], 0
5832 // CHECK4-NEXT: br i1 [[TMP1]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
5833 // CHECK4: omp_offload.failed:
5834 // CHECK4-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiEiT__l57() #[[ATTR2]]
5835 // CHECK4-NEXT: br label [[OMP_OFFLOAD_CONT]]
5836 // CHECK4: omp_offload.cont:
5837 // CHECK4-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i64 100)
5838 // CHECK4-NEXT: [[TMP2:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiEiT__l62.region_id, i32 0, i8** null, i8** null, i64* null, i64* null, i8** null, i8** null, i32 0, i32 1)
5839 // CHECK4-NEXT: [[TMP3:%.*]] = icmp ne i32 [[TMP2]], 0
5840 // CHECK4-NEXT: br i1 [[TMP3]], label [[OMP_OFFLOAD_FAILED2:%.*]], label [[OMP_OFFLOAD_CONT3:%.*]]
5841 // CHECK4: omp_offload.failed2:
5842 // CHECK4-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiEiT__l62() #[[ATTR2]]
5843 // CHECK4-NEXT: br label [[OMP_OFFLOAD_CONT3]]
5844 // CHECK4: omp_offload.cont3:
5845 // CHECK4-NEXT: [[TMP4:%.*]] = load i32, i32* [[ARG_ADDR]], align 4
5846 // CHECK4-NEXT: [[CONV:%.*]] = bitcast i64* [[ARG_CASTED]] to i32*
5847 // CHECK4-NEXT: store i32 [[TMP4]], i32* [[CONV]], align 4
5848 // CHECK4-NEXT: [[TMP5:%.*]] = load i64, i64* [[ARG_CASTED]], align 8
5849 // CHECK4-NEXT: [[TMP6:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
5850 // CHECK4-NEXT: [[TMP7:%.*]] = bitcast i8** [[TMP6]] to i64*
5851 // CHECK4-NEXT: store i64 [[TMP5]], i64* [[TMP7]], align 8
5852 // CHECK4-NEXT: [[TMP8:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
5853 // CHECK4-NEXT: [[TMP9:%.*]] = bitcast i8** [[TMP8]] to i64*
5854 // CHECK4-NEXT: store i64 [[TMP5]], i64* [[TMP9]], align 8
5855 // CHECK4-NEXT: [[TMP10:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0
5856 // CHECK4-NEXT: store i8* null, i8** [[TMP10]], align 8
5857 // CHECK4-NEXT: [[TMP11:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
5858 // CHECK4-NEXT: [[TMP12:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
5859 // CHECK4-NEXT: [[TMP13:%.*]] = load i32, i32* [[ARG_ADDR]], align 4
5860 // CHECK4-NEXT: [[TOBOOL:%.*]] = icmp ne i32 [[TMP13]], 0
5861 // CHECK4-NEXT: [[FROMBOOL:%.*]] = zext i1 [[TOBOOL]] to i8
5862 // CHECK4-NEXT: store i8 [[FROMBOOL]], i8* [[DOTCAPTURE_EXPR_]], align 1
5863 // CHECK4-NEXT: [[TMP14:%.*]] = load i8, i8* [[DOTCAPTURE_EXPR_]], align 1
5864 // CHECK4-NEXT: [[TOBOOL4:%.*]] = trunc i8 [[TMP14]] to i1
5865 // CHECK4-NEXT: [[TMP15:%.*]] = select i1 [[TOBOOL4]], i32 0, i32 1
5866 // CHECK4-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i64 100)
5867 // CHECK4-NEXT: [[TMP16:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiEiT__l67.region_id, i32 1, i8** [[TMP11]], i8** [[TMP12]], i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_sizes.17, i32 0, i32 0), i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_maptypes.18, i32 0, i32 0), i8** null, i8** null, i32 0, i32 [[TMP15]])
5868 // CHECK4-NEXT: [[TMP17:%.*]] = icmp ne i32 [[TMP16]], 0
5869 // CHECK4-NEXT: br i1 [[TMP17]], label [[OMP_OFFLOAD_FAILED6:%.*]], label [[OMP_OFFLOAD_CONT7:%.*]]
5870 // CHECK4: omp_offload.failed6:
5871 // CHECK4-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiEiT__l67(i64 [[TMP5]]) #[[ATTR2]]
5872 // CHECK4-NEXT: br label [[OMP_OFFLOAD_CONT7]]
5873 // CHECK4: omp_offload.cont7:
5874 // CHECK4-NEXT: ret i32 0
5875 //
5876 //
5877 // CHECK4-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiEiT__l57
5878 // CHECK4-SAME: () #[[ATTR1]] {
5879 // CHECK4-NEXT: entry:
5880 // CHECK4-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 0, void (i32*, i32*, ...)* bitcast (void (i32*, i32*)* @.omp_outlined..11 to void (i32*, i32*, ...)*))
5881 // CHECK4-NEXT: ret void
5882 //
5883 //
5884 // CHECK4-LABEL: define {{[^@]+}}@.omp_outlined..11
5885 // CHECK4-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]]) #[[ATTR1]] {
5886 // CHECK4-NEXT: entry:
5887 // CHECK4-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
5888 // CHECK4-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
5889 // CHECK4-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
5890 // CHECK4-NEXT: [[TMP:%.*]] = alloca i32, align 4
5891 // CHECK4-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4
5892 // CHECK4-NEXT: [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4
5893 // CHECK4-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
5894 // CHECK4-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
5895 // CHECK4-NEXT: [[I:%.*]] = alloca i32, align 4
5896 // CHECK4-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
5897 // CHECK4-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
5898 // CHECK4-NEXT: store i32 0, i32* [[DOTOMP_COMB_LB]], align 4
5899 // CHECK4-NEXT: store i32 99, i32* [[DOTOMP_COMB_UB]], align 4
5900 // CHECK4-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4
5901 // CHECK4-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4
5902 // CHECK4-NEXT: [[TMP0:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
5903 // CHECK4-NEXT: [[TMP1:%.*]] = load i32, i32* [[TMP0]], align 4
5904 // CHECK4-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB1]], i32 [[TMP1]], i32 92, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_COMB_LB]], i32* [[DOTOMP_COMB_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1)
5905 // CHECK4-NEXT: [[TMP2:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
5906 // CHECK4-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP2]], 99
5907 // CHECK4-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
5908 // CHECK4: cond.true:
5909 // CHECK4-NEXT: br label [[COND_END:%.*]]
5910 // CHECK4: cond.false:
5911 // CHECK4-NEXT: [[TMP3:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
5912 // CHECK4-NEXT: br label [[COND_END]]
5913 // CHECK4: cond.end:
5914 // CHECK4-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP3]], [[COND_FALSE]] ]
5915 // CHECK4-NEXT: store i32 [[COND]], i32* [[DOTOMP_COMB_UB]], align 4
5916 // CHECK4-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4
5917 // CHECK4-NEXT: store i32 [[TMP4]], i32* [[DOTOMP_IV]], align 4
5918 // CHECK4-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
5919 // CHECK4: omp.inner.for.cond:
5920 // CHECK4-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
5921 // CHECK4-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
5922 // CHECK4-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP5]], [[TMP6]]
5923 // CHECK4-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
5924 // CHECK4: omp.inner.for.body:
5925 // CHECK4-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4
5926 // CHECK4-NEXT: [[TMP8:%.*]] = zext i32 [[TMP7]] to i64
5927 // CHECK4-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
5928 // CHECK4-NEXT: [[TMP10:%.*]] = zext i32 [[TMP9]] to i64
5929 // CHECK4-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB3]], i32 2, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64, i64)* @.omp_outlined..12 to void (i32*, i32*, ...)*), i64 [[TMP8]], i64 [[TMP10]])
5930 // CHECK4-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
5931 // CHECK4: omp.inner.for.inc:
5932 // CHECK4-NEXT: [[TMP11:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
5933 // CHECK4-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4
5934 // CHECK4-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP11]], [[TMP12]]
5935 // CHECK4-NEXT: store i32 [[ADD]], i32* [[DOTOMP_IV]], align 4
5936 // CHECK4-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP28:![0-9]+]]
5937 // CHECK4: omp.inner.for.end:
5938 // CHECK4-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
5939 // CHECK4: omp.loop.exit:
5940 // CHECK4-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP1]])
5941 // CHECK4-NEXT: [[TMP13:%.*]] = load i32, i32* [[DOTOMP_IS_LAST]], align 4
5942 // CHECK4-NEXT: [[TMP14:%.*]] = icmp ne i32 [[TMP13]], 0
5943 // CHECK4-NEXT: br i1 [[TMP14]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
5944 // CHECK4: .omp.final.then:
5945 // CHECK4-NEXT: store i32 100, i32* [[I]], align 4
5946 // CHECK4-NEXT: br label [[DOTOMP_FINAL_DONE]]
5947 // CHECK4: .omp.final.done:
5948 // CHECK4-NEXT: ret void
5949 //
5950 //
5951 // CHECK4-LABEL: define {{[^@]+}}@.omp_outlined..12
5952 // CHECK4-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[DOTPREVIOUS_LB_:%.*]], i64 [[DOTPREVIOUS_UB_:%.*]]) #[[ATTR1]] {
5953 // CHECK4-NEXT: entry:
5954 // CHECK4-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
5955 // CHECK4-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
5956 // CHECK4-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i64, align 8
5957 // CHECK4-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i64, align 8
5958 // CHECK4-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
5959 // CHECK4-NEXT: [[TMP:%.*]] = alloca i32, align 4
5960 // CHECK4-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4
5961 // CHECK4-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4
5962 // CHECK4-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
5963 // CHECK4-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
5964 // CHECK4-NEXT: [[I:%.*]] = alloca i32, align 4
5965 // CHECK4-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
5966 // CHECK4-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
5967 // CHECK4-NEXT: store i64 [[DOTPREVIOUS_LB_]], i64* [[DOTPREVIOUS_LB__ADDR]], align 8
5968 // CHECK4-NEXT: store i64 [[DOTPREVIOUS_UB_]], i64* [[DOTPREVIOUS_UB__ADDR]], align 8
5969 // CHECK4-NEXT: store i32 0, i32* [[DOTOMP_LB]], align 4
5970 // CHECK4-NEXT: store i32 99, i32* [[DOTOMP_UB]], align 4
5971 // CHECK4-NEXT: [[TMP0:%.*]] = load i64, i64* [[DOTPREVIOUS_LB__ADDR]], align 8
5972 // CHECK4-NEXT: [[CONV:%.*]] = trunc i64 [[TMP0]] to i32
5973 // CHECK4-NEXT: [[TMP1:%.*]] = load i64, i64* [[DOTPREVIOUS_UB__ADDR]], align 8
5974 // CHECK4-NEXT: [[CONV1:%.*]] = trunc i64 [[TMP1]] to i32
5975 // CHECK4-NEXT: store i32 [[CONV]], i32* [[DOTOMP_LB]], align 4
5976 // CHECK4-NEXT: store i32 [[CONV1]], i32* [[DOTOMP_UB]], align 4
5977 // CHECK4-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4
5978 // CHECK4-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4
5979 // CHECK4-NEXT: [[TMP2:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
5980 // CHECK4-NEXT: [[TMP3:%.*]] = load i32, i32* [[TMP2]], align 4
5981 // CHECK4-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB2]], i32 [[TMP3]], i32 34, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_LB]], i32* [[DOTOMP_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1)
5982 // CHECK4-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
5983 // CHECK4-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP4]], 99
5984 // CHECK4-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
5985 // CHECK4: cond.true:
5986 // CHECK4-NEXT: br label [[COND_END:%.*]]
5987 // CHECK4: cond.false:
5988 // CHECK4-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
5989 // CHECK4-NEXT: br label [[COND_END]]
5990 // CHECK4: cond.end:
5991 // CHECK4-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP5]], [[COND_FALSE]] ]
5992 // CHECK4-NEXT: store i32 [[COND]], i32* [[DOTOMP_UB]], align 4
5993 // CHECK4-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4
5994 // CHECK4-NEXT: store i32 [[TMP6]], i32* [[DOTOMP_IV]], align 4
5995 // CHECK4-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
5996 // CHECK4: omp.inner.for.cond:
5997 // CHECK4-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
5998 // CHECK4-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
5999 // CHECK4-NEXT: [[CMP2:%.*]] = icmp sle i32 [[TMP7]], [[TMP8]]
6000 // CHECK4-NEXT: br i1 [[CMP2]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
6001 // CHECK4: omp.inner.for.body:
6002 // CHECK4-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
6003 // CHECK4-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP9]], 1
6004 // CHECK4-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
6005 // CHECK4-NEXT: store i32 [[ADD]], i32* [[I]], align 4
6006 // CHECK4-NEXT: call void @_Z3fn1v()
6007 // CHECK4-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
6008 // CHECK4: omp.body.continue:
6009 // CHECK4-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
6010 // CHECK4: omp.inner.for.inc:
6011 // CHECK4-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
6012 // CHECK4-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP10]], 1
6013 // CHECK4-NEXT: store i32 [[ADD3]], i32* [[DOTOMP_IV]], align 4
6014 // CHECK4-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP29:![0-9]+]]
6015 // CHECK4: omp.inner.for.end:
6016 // CHECK4-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
6017 // CHECK4: omp.loop.exit:
6018 // CHECK4-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP3]])
6019 // CHECK4-NEXT: [[TMP11:%.*]] = load i32, i32* [[DOTOMP_IS_LAST]], align 4
6020 // CHECK4-NEXT: [[TMP12:%.*]] = icmp ne i32 [[TMP11]], 0
6021 // CHECK4-NEXT: br i1 [[TMP12]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
6022 // CHECK4: .omp.final.then:
6023 // CHECK4-NEXT: store i32 100, i32* [[I]], align 4
6024 // CHECK4-NEXT: br label [[DOTOMP_FINAL_DONE]]
6025 // CHECK4: .omp.final.done:
6026 // CHECK4-NEXT: ret void
6027 //
6028 //
6029 // CHECK4-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiEiT__l62
6030 // CHECK4-SAME: () #[[ATTR1]] {
6031 // CHECK4-NEXT: entry:
6032 // CHECK4-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 0, void (i32*, i32*, ...)* bitcast (void (i32*, i32*)* @.omp_outlined..13 to void (i32*, i32*, ...)*))
6033 // CHECK4-NEXT: ret void
6034 //
6035 //
6036 // CHECK4-LABEL: define {{[^@]+}}@.omp_outlined..13
6037 // CHECK4-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]]) #[[ATTR1]] {
6038 // CHECK4-NEXT: entry:
6039 // CHECK4-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
6040 // CHECK4-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
6041 // CHECK4-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
6042 // CHECK4-NEXT: [[TMP:%.*]] = alloca i32, align 4
6043 // CHECK4-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4
6044 // CHECK4-NEXT: [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4
6045 // CHECK4-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
6046 // CHECK4-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
6047 // CHECK4-NEXT: [[I:%.*]] = alloca i32, align 4
6048 // CHECK4-NEXT: [[DOTBOUND_ZERO_ADDR:%.*]] = alloca i32, align 4
6049 // CHECK4-NEXT: store i32 0, i32* [[DOTBOUND_ZERO_ADDR]], align 4
6050 // CHECK4-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
6051 // CHECK4-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
6052 // CHECK4-NEXT: store i32 0, i32* [[DOTOMP_COMB_LB]], align 4
6053 // CHECK4-NEXT: store i32 99, i32* [[DOTOMP_COMB_UB]], align 4
6054 // CHECK4-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4
6055 // CHECK4-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4
6056 // CHECK4-NEXT: [[TMP0:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
6057 // CHECK4-NEXT: [[TMP1:%.*]] = load i32, i32* [[TMP0]], align 4
6058 // CHECK4-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB1]], i32 [[TMP1]], i32 92, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_COMB_LB]], i32* [[DOTOMP_COMB_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1)
6059 // CHECK4-NEXT: [[TMP2:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
6060 // CHECK4-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP2]], 99
6061 // CHECK4-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
6062 // CHECK4: cond.true:
6063 // CHECK4-NEXT: br label [[COND_END:%.*]]
6064 // CHECK4: cond.false:
6065 // CHECK4-NEXT: [[TMP3:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
6066 // CHECK4-NEXT: br label [[COND_END]]
6067 // CHECK4: cond.end:
6068 // CHECK4-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP3]], [[COND_FALSE]] ]
6069 // CHECK4-NEXT: store i32 [[COND]], i32* [[DOTOMP_COMB_UB]], align 4
6070 // CHECK4-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4
6071 // CHECK4-NEXT: store i32 [[TMP4]], i32* [[DOTOMP_IV]], align 4
6072 // CHECK4-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
6073 // CHECK4: omp.inner.for.cond:
6074 // CHECK4-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
6075 // CHECK4-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
6076 // CHECK4-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP5]], [[TMP6]]
6077 // CHECK4-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
6078 // CHECK4: omp.inner.for.body:
6079 // CHECK4-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4
6080 // CHECK4-NEXT: [[TMP8:%.*]] = zext i32 [[TMP7]] to i64
6081 // CHECK4-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
6082 // CHECK4-NEXT: [[TMP10:%.*]] = zext i32 [[TMP9]] to i64
6083 // CHECK4-NEXT: call void @__kmpc_serialized_parallel(%struct.ident_t* @[[GLOB3]], i32 [[TMP1]])
6084 // CHECK4-NEXT: [[TMP11:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
6085 // CHECK4-NEXT: call void @.omp_outlined..14(i32* [[TMP11]], i32* [[DOTBOUND_ZERO_ADDR]], i64 [[TMP8]], i64 [[TMP10]]) #[[ATTR2]]
6086 // CHECK4-NEXT: call void @__kmpc_end_serialized_parallel(%struct.ident_t* @[[GLOB3]], i32 [[TMP1]])
6087 // CHECK4-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
6088 // CHECK4: omp.inner.for.inc:
6089 // CHECK4-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
6090 // CHECK4-NEXT: [[TMP13:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4
6091 // CHECK4-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP12]], [[TMP13]]
6092 // CHECK4-NEXT: store i32 [[ADD]], i32* [[DOTOMP_IV]], align 4
6093 // CHECK4-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP30:![0-9]+]]
6094 // CHECK4: omp.inner.for.end:
6095 // CHECK4-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
6096 // CHECK4: omp.loop.exit:
6097 // CHECK4-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP1]])
6098 // CHECK4-NEXT: [[TMP14:%.*]] = load i32, i32* [[DOTOMP_IS_LAST]], align 4
6099 // CHECK4-NEXT: [[TMP15:%.*]] = icmp ne i32 [[TMP14]], 0
6100 // CHECK4-NEXT: br i1 [[TMP15]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
6101 // CHECK4: .omp.final.then:
6102 // CHECK4-NEXT: store i32 100, i32* [[I]], align 4
6103 // CHECK4-NEXT: br label [[DOTOMP_FINAL_DONE]]
6104 // CHECK4: .omp.final.done:
6105 // CHECK4-NEXT: ret void
6106 //
6107 //
6108 // CHECK4-LABEL: define {{[^@]+}}@.omp_outlined..14
6109 // CHECK4-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[DOTPREVIOUS_LB_:%.*]], i64 [[DOTPREVIOUS_UB_:%.*]]) #[[ATTR1]] {
6110 // CHECK4-NEXT: entry:
6111 // CHECK4-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
6112 // CHECK4-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
6113 // CHECK4-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i64, align 8
6114 // CHECK4-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i64, align 8
6115 // CHECK4-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
6116 // CHECK4-NEXT: [[TMP:%.*]] = alloca i32, align 4
6117 // CHECK4-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4
6118 // CHECK4-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4
6119 // CHECK4-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
6120 // CHECK4-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
6121 // CHECK4-NEXT: [[I:%.*]] = alloca i32, align 4
6122 // CHECK4-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
6123 // CHECK4-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
6124 // CHECK4-NEXT: store i64 [[DOTPREVIOUS_LB_]], i64* [[DOTPREVIOUS_LB__ADDR]], align 8
6125 // CHECK4-NEXT: store i64 [[DOTPREVIOUS_UB_]], i64* [[DOTPREVIOUS_UB__ADDR]], align 8
6126 // CHECK4-NEXT: store i32 0, i32* [[DOTOMP_LB]], align 4
6127 // CHECK4-NEXT: store i32 99, i32* [[DOTOMP_UB]], align 4
6128 // CHECK4-NEXT: [[TMP0:%.*]] = load i64, i64* [[DOTPREVIOUS_LB__ADDR]], align 8
6129 // CHECK4-NEXT: [[CONV:%.*]] = trunc i64 [[TMP0]] to i32
6130 // CHECK4-NEXT: [[TMP1:%.*]] = load i64, i64* [[DOTPREVIOUS_UB__ADDR]], align 8
6131 // CHECK4-NEXT: [[CONV1:%.*]] = trunc i64 [[TMP1]] to i32
6132 // CHECK4-NEXT: store i32 [[CONV]], i32* [[DOTOMP_LB]], align 4
6133 // CHECK4-NEXT: store i32 [[CONV1]], i32* [[DOTOMP_UB]], align 4
6134 // CHECK4-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4
6135 // CHECK4-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4
6136 // CHECK4-NEXT: [[TMP2:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
6137 // CHECK4-NEXT: [[TMP3:%.*]] = load i32, i32* [[TMP2]], align 4
6138 // CHECK4-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB2]], i32 [[TMP3]], i32 34, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_LB]], i32* [[DOTOMP_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1)
6139 // CHECK4-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
6140 // CHECK4-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP4]], 99
6141 // CHECK4-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
6142 // CHECK4: cond.true:
6143 // CHECK4-NEXT: br label [[COND_END:%.*]]
6144 // CHECK4: cond.false:
6145 // CHECK4-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
6146 // CHECK4-NEXT: br label [[COND_END]]
6147 // CHECK4: cond.end:
6148 // CHECK4-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP5]], [[COND_FALSE]] ]
6149 // CHECK4-NEXT: store i32 [[COND]], i32* [[DOTOMP_UB]], align 4
6150 // CHECK4-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4
6151 // CHECK4-NEXT: store i32 [[TMP6]], i32* [[DOTOMP_IV]], align 4
6152 // CHECK4-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
6153 // CHECK4: omp.inner.for.cond:
6154 // CHECK4-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
6155 // CHECK4-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
6156 // CHECK4-NEXT: [[CMP2:%.*]] = icmp sle i32 [[TMP7]], [[TMP8]]
6157 // CHECK4-NEXT: br i1 [[CMP2]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
6158 // CHECK4: omp.inner.for.body:
6159 // CHECK4-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
6160 // CHECK4-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP9]], 1
6161 // CHECK4-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
6162 // CHECK4-NEXT: store i32 [[ADD]], i32* [[I]], align 4
6163 // CHECK4-NEXT: call void @_Z3fn2v()
6164 // CHECK4-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
6165 // CHECK4: omp.body.continue:
6166 // CHECK4-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
6167 // CHECK4: omp.inner.for.inc:
6168 // CHECK4-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
6169 // CHECK4-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP10]], 1
6170 // CHECK4-NEXT: store i32 [[ADD3]], i32* [[DOTOMP_IV]], align 4
6171 // CHECK4-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP31:![0-9]+]]
6172 // CHECK4: omp.inner.for.end:
6173 // CHECK4-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
6174 // CHECK4: omp.loop.exit:
6175 // CHECK4-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP3]])
6176 // CHECK4-NEXT: [[TMP11:%.*]] = load i32, i32* [[DOTOMP_IS_LAST]], align 4
6177 // CHECK4-NEXT: [[TMP12:%.*]] = icmp ne i32 [[TMP11]], 0
6178 // CHECK4-NEXT: br i1 [[TMP12]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
6179 // CHECK4: .omp.final.then:
6180 // CHECK4-NEXT: store i32 100, i32* [[I]], align 4
6181 // CHECK4-NEXT: br label [[DOTOMP_FINAL_DONE]]
6182 // CHECK4: .omp.final.done:
6183 // CHECK4-NEXT: ret void
6184 //
6185 //
6186 // CHECK4-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiEiT__l67
6187 // CHECK4-SAME: (i64 [[ARG:%.*]]) #[[ATTR1]] {
6188 // CHECK4-NEXT: entry:
6189 // CHECK4-NEXT: [[ARG_ADDR:%.*]] = alloca i64, align 8
6190 // CHECK4-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i8, align 1
6191 // CHECK4-NEXT: [[DOTCAPTURE_EXPR__CASTED:%.*]] = alloca i64, align 8
6192 // CHECK4-NEXT: store i64 [[ARG]], i64* [[ARG_ADDR]], align 8
6193 // CHECK4-NEXT: [[CONV:%.*]] = bitcast i64* [[ARG_ADDR]] to i32*
6194 // CHECK4-NEXT: [[TMP0:%.*]] = load i32, i32* [[CONV]], align 8
6195 // CHECK4-NEXT: [[TOBOOL:%.*]] = icmp ne i32 [[TMP0]], 0
6196 // CHECK4-NEXT: [[FROMBOOL:%.*]] = zext i1 [[TOBOOL]] to i8
6197 // CHECK4-NEXT: store i8 [[FROMBOOL]], i8* [[DOTCAPTURE_EXPR_]], align 1
6198 // CHECK4-NEXT: [[TMP1:%.*]] = load i8, i8* [[DOTCAPTURE_EXPR_]], align 1
6199 // CHECK4-NEXT: [[TOBOOL1:%.*]] = trunc i8 [[TMP1]] to i1
6200 // CHECK4-NEXT: [[CONV2:%.*]] = bitcast i64* [[DOTCAPTURE_EXPR__CASTED]] to i8*
6201 // CHECK4-NEXT: [[FROMBOOL3:%.*]] = zext i1 [[TOBOOL1]] to i8
6202 // CHECK4-NEXT: store i8 [[FROMBOOL3]], i8* [[CONV2]], align 1
6203 // CHECK4-NEXT: [[TMP2:%.*]] = load i64, i64* [[DOTCAPTURE_EXPR__CASTED]], align 8
6204 // CHECK4-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64)* @.omp_outlined..15 to void (i32*, i32*, ...)*), i64 [[TMP2]])
6205 // CHECK4-NEXT: ret void
6206 //
6207 //
6208 // CHECK4-LABEL: define {{[^@]+}}@.omp_outlined..15
6209 // CHECK4-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[DOTCAPTURE_EXPR_:%.*]]) #[[ATTR1]] {
6210 // CHECK4-NEXT: entry:
6211 // CHECK4-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
6212 // CHECK4-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
6213 // CHECK4-NEXT: [[DOTCAPTURE_EXPR__ADDR:%.*]] = alloca i64, align 8
6214 // CHECK4-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
6215 // CHECK4-NEXT: [[TMP:%.*]] = alloca i32, align 4
6216 // CHECK4-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4
6217 // CHECK4-NEXT: [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4
6218 // CHECK4-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
6219 // CHECK4-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
6220 // CHECK4-NEXT: [[I:%.*]] = alloca i32, align 4
6221 // CHECK4-NEXT: [[DOTBOUND_ZERO_ADDR:%.*]] = alloca i32, align 4
6222 // CHECK4-NEXT: store i32 0, i32* [[DOTBOUND_ZERO_ADDR]], align 4
6223 // CHECK4-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
6224 // CHECK4-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
6225 // CHECK4-NEXT: store i64 [[DOTCAPTURE_EXPR_]], i64* [[DOTCAPTURE_EXPR__ADDR]], align 8
6226 // CHECK4-NEXT: [[CONV:%.*]] = bitcast i64* [[DOTCAPTURE_EXPR__ADDR]] to i8*
6227 // CHECK4-NEXT: store i32 0, i32* [[DOTOMP_COMB_LB]], align 4
6228 // CHECK4-NEXT: store i32 99, i32* [[DOTOMP_COMB_UB]], align 4
6229 // CHECK4-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4
6230 // CHECK4-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4
6231 // CHECK4-NEXT: [[TMP0:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
6232 // CHECK4-NEXT: [[TMP1:%.*]] = load i32, i32* [[TMP0]], align 4
6233 // CHECK4-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB1]], i32 [[TMP1]], i32 92, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_COMB_LB]], i32* [[DOTOMP_COMB_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1)
6234 // CHECK4-NEXT: [[TMP2:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
6235 // CHECK4-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP2]], 99
6236 // CHECK4-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
6237 // CHECK4: cond.true:
6238 // CHECK4-NEXT: br label [[COND_END:%.*]]
6239 // CHECK4: cond.false:
6240 // CHECK4-NEXT: [[TMP3:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
6241 // CHECK4-NEXT: br label [[COND_END]]
6242 // CHECK4: cond.end:
6243 // CHECK4-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP3]], [[COND_FALSE]] ]
6244 // CHECK4-NEXT: store i32 [[COND]], i32* [[DOTOMP_COMB_UB]], align 4
6245 // CHECK4-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4
6246 // CHECK4-NEXT: store i32 [[TMP4]], i32* [[DOTOMP_IV]], align 4
6247 // CHECK4-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
6248 // CHECK4: omp.inner.for.cond:
6249 // CHECK4-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
6250 // CHECK4-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
6251 // CHECK4-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP5]], [[TMP6]]
6252 // CHECK4-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
6253 // CHECK4: omp.inner.for.body:
6254 // CHECK4-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4
6255 // CHECK4-NEXT: [[TMP8:%.*]] = zext i32 [[TMP7]] to i64
6256 // CHECK4-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
6257 // CHECK4-NEXT: [[TMP10:%.*]] = zext i32 [[TMP9]] to i64
6258 // CHECK4-NEXT: [[TMP11:%.*]] = load i8, i8* [[CONV]], align 8
6259 // CHECK4-NEXT: [[TOBOOL:%.*]] = trunc i8 [[TMP11]] to i1
6260 // CHECK4-NEXT: br i1 [[TOBOOL]], label [[OMP_IF_THEN:%.*]], label [[OMP_IF_ELSE:%.*]]
6261 // CHECK4: omp_if.then:
6262 // CHECK4-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB3]], i32 2, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64, i64)* @.omp_outlined..16 to void (i32*, i32*, ...)*), i64 [[TMP8]], i64 [[TMP10]])
6263 // CHECK4-NEXT: br label [[OMP_IF_END:%.*]]
6264 // CHECK4: omp_if.else:
6265 // CHECK4-NEXT: call void @__kmpc_serialized_parallel(%struct.ident_t* @[[GLOB3]], i32 [[TMP1]])
6266 // CHECK4-NEXT: [[TMP12:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
6267 // CHECK4-NEXT: call void @.omp_outlined..16(i32* [[TMP12]], i32* [[DOTBOUND_ZERO_ADDR]], i64 [[TMP8]], i64 [[TMP10]]) #[[ATTR2]]
6268 // CHECK4-NEXT: call void @__kmpc_end_serialized_parallel(%struct.ident_t* @[[GLOB3]], i32 [[TMP1]])
6269 // CHECK4-NEXT: br label [[OMP_IF_END]]
6270 // CHECK4: omp_if.end:
6271 // CHECK4-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
6272 // CHECK4: omp.inner.for.inc:
6273 // CHECK4-NEXT: [[TMP13:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
6274 // CHECK4-NEXT: [[TMP14:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4
6275 // CHECK4-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP13]], [[TMP14]]
6276 // CHECK4-NEXT: store i32 [[ADD]], i32* [[DOTOMP_IV]], align 4
6277 // CHECK4-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP32:![0-9]+]]
6278 // CHECK4: omp.inner.for.end:
6279 // CHECK4-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
6280 // CHECK4: omp.loop.exit:
6281 // CHECK4-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP1]])
6282 // CHECK4-NEXT: [[TMP15:%.*]] = load i32, i32* [[DOTOMP_IS_LAST]], align 4
6283 // CHECK4-NEXT: [[TMP16:%.*]] = icmp ne i32 [[TMP15]], 0
6284 // CHECK4-NEXT: br i1 [[TMP16]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
6285 // CHECK4: .omp.final.then:
6286 // CHECK4-NEXT: store i32 100, i32* [[I]], align 4
6287 // CHECK4-NEXT: br label [[DOTOMP_FINAL_DONE]]
6288 // CHECK4: .omp.final.done:
6289 // CHECK4-NEXT: ret void
6290 //
6291 //
6292 // CHECK4-LABEL: define {{[^@]+}}@.omp_outlined..16
6293 // CHECK4-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[DOTPREVIOUS_LB_:%.*]], i64 [[DOTPREVIOUS_UB_:%.*]]) #[[ATTR1]] {
6294 // CHECK4-NEXT: entry:
6295 // CHECK4-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
6296 // CHECK4-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
6297 // CHECK4-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i64, align 8
6298 // CHECK4-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i64, align 8
6299 // CHECK4-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
6300 // CHECK4-NEXT: [[TMP:%.*]] = alloca i32, align 4
6301 // CHECK4-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4
6302 // CHECK4-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4
6303 // CHECK4-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
6304 // CHECK4-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
6305 // CHECK4-NEXT: [[I:%.*]] = alloca i32, align 4
6306 // CHECK4-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
6307 // CHECK4-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
6308 // CHECK4-NEXT: store i64 [[DOTPREVIOUS_LB_]], i64* [[DOTPREVIOUS_LB__ADDR]], align 8
6309 // CHECK4-NEXT: store i64 [[DOTPREVIOUS_UB_]], i64* [[DOTPREVIOUS_UB__ADDR]], align 8
6310 // CHECK4-NEXT: store i32 0, i32* [[DOTOMP_LB]], align 4
6311 // CHECK4-NEXT: store i32 99, i32* [[DOTOMP_UB]], align 4
6312 // CHECK4-NEXT: [[TMP0:%.*]] = load i64, i64* [[DOTPREVIOUS_LB__ADDR]], align 8
6313 // CHECK4-NEXT: [[CONV:%.*]] = trunc i64 [[TMP0]] to i32
6314 // CHECK4-NEXT: [[TMP1:%.*]] = load i64, i64* [[DOTPREVIOUS_UB__ADDR]], align 8
6315 // CHECK4-NEXT: [[CONV1:%.*]] = trunc i64 [[TMP1]] to i32
6316 // CHECK4-NEXT: store i32 [[CONV]], i32* [[DOTOMP_LB]], align 4
6317 // CHECK4-NEXT: store i32 [[CONV1]], i32* [[DOTOMP_UB]], align 4
6318 // CHECK4-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4
6319 // CHECK4-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4
6320 // CHECK4-NEXT: [[TMP2:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
6321 // CHECK4-NEXT: [[TMP3:%.*]] = load i32, i32* [[TMP2]], align 4
6322 // CHECK4-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB2]], i32 [[TMP3]], i32 34, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_LB]], i32* [[DOTOMP_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1)
6323 // CHECK4-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
6324 // CHECK4-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP4]], 99
6325 // CHECK4-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
6326 // CHECK4: cond.true:
6327 // CHECK4-NEXT: br label [[COND_END:%.*]]
6328 // CHECK4: cond.false:
6329 // CHECK4-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
6330 // CHECK4-NEXT: br label [[COND_END]]
6331 // CHECK4: cond.end:
6332 // CHECK4-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP5]], [[COND_FALSE]] ]
6333 // CHECK4-NEXT: store i32 [[COND]], i32* [[DOTOMP_UB]], align 4
6334 // CHECK4-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4
6335 // CHECK4-NEXT: store i32 [[TMP6]], i32* [[DOTOMP_IV]], align 4
6336 // CHECK4-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
6337 // CHECK4: omp.inner.for.cond:
6338 // CHECK4-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
6339 // CHECK4-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
6340 // CHECK4-NEXT: [[CMP2:%.*]] = icmp sle i32 [[TMP7]], [[TMP8]]
6341 // CHECK4-NEXT: br i1 [[CMP2]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
6342 // CHECK4: omp.inner.for.body:
6343 // CHECK4-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
6344 // CHECK4-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP9]], 1
6345 // CHECK4-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
6346 // CHECK4-NEXT: store i32 [[ADD]], i32* [[I]], align 4
6347 // CHECK4-NEXT: call void @_Z3fn3v()
6348 // CHECK4-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
6349 // CHECK4: omp.body.continue:
6350 // CHECK4-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
6351 // CHECK4: omp.inner.for.inc:
6352 // CHECK4-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
6353 // CHECK4-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP10]], 1
6354 // CHECK4-NEXT: store i32 [[ADD3]], i32* [[DOTOMP_IV]], align 4
6355 // CHECK4-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP33:![0-9]+]]
6356 // CHECK4: omp.inner.for.end:
6357 // CHECK4-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
6358 // CHECK4: omp.loop.exit:
6359 // CHECK4-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP3]])
6360 // CHECK4-NEXT: [[TMP11:%.*]] = load i32, i32* [[DOTOMP_IS_LAST]], align 4
6361 // CHECK4-NEXT: [[TMP12:%.*]] = icmp ne i32 [[TMP11]], 0
6362 // CHECK4-NEXT: br i1 [[TMP12]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
6363 // CHECK4: .omp.final.then:
6364 // CHECK4-NEXT: store i32 100, i32* [[I]], align 4
6365 // CHECK4-NEXT: br label [[DOTOMP_FINAL_DONE]]
6366 // CHECK4: .omp.final.done:
6367 // CHECK4-NEXT: ret void
6368 //
6369 //
6370 // CHECK4-LABEL: define {{[^@]+}}@.omp_offloading.requires_reg
6371 // CHECK4-SAME: () #[[ATTR5:[0-9]+]] {
6372 // CHECK4-NEXT: entry:
6373 // CHECK4-NEXT: call void @__tgt_register_requires(i64 1)
6374 // CHECK4-NEXT: ret void
6375 //
6376 //
6377 // CHECK5-LABEL: define {{[^@]+}}@_Z9gtid_testv
6378 // CHECK5-SAME: () #[[ATTR0:[0-9]+]] {
6379 // CHECK5-NEXT: entry:
6380 // CHECK5-NEXT: [[TMP:%.*]] = alloca i32, align 4
6381 // CHECK5-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4
6382 // CHECK5-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4
6383 // CHECK5-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
6384 // CHECK5-NEXT: [[I:%.*]] = alloca i32, align 4
6385 // CHECK5-NEXT: [[_TMP2:%.*]] = alloca i32, align 4
6386 // CHECK5-NEXT: [[DOTOMP_LB3:%.*]] = alloca i32, align 4
6387 // CHECK5-NEXT: [[DOTOMP_UB4:%.*]] = alloca i32, align 4
6388 // CHECK5-NEXT: [[DOTOMP_IV5:%.*]] = alloca i32, align 4
6389 // CHECK5-NEXT: [[I6:%.*]] = alloca i32, align 4
6390 // CHECK5-NEXT: store i32 0, i32* [[DOTOMP_LB]], align 4
6391 // CHECK5-NEXT: store i32 99, i32* [[DOTOMP_UB]], align 4
6392 // CHECK5-NEXT: [[TMP0:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4
6393 // CHECK5-NEXT: store i32 [[TMP0]], i32* [[DOTOMP_IV]], align 4
6394 // CHECK5-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
6395 // CHECK5: omp.inner.for.cond:
6396 // CHECK5-NEXT: [[TMP1:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !2
6397 // CHECK5-NEXT: [[TMP2:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4, !llvm.access.group !2
6398 // CHECK5-NEXT: [[CMP:%.*]] = icmp sle i32 [[TMP1]], [[TMP2]]
6399 // CHECK5-NEXT: br i1 [[CMP]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
6400 // CHECK5: omp.inner.for.body:
6401 // CHECK5-NEXT: [[TMP3:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !2
6402 // CHECK5-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP3]], 1
6403 // CHECK5-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
6404 // CHECK5-NEXT: store i32 [[ADD]], i32* [[I]], align 4, !llvm.access.group !2
6405 // CHECK5-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
6406 // CHECK5: omp.body.continue:
6407 // CHECK5-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
6408 // CHECK5: omp.inner.for.inc:
6409 // CHECK5-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !2
6410 // CHECK5-NEXT: [[ADD1:%.*]] = add nsw i32 [[TMP4]], 1
6411 // CHECK5-NEXT: store i32 [[ADD1]], i32* [[DOTOMP_IV]], align 4, !llvm.access.group !2
6412 // CHECK5-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP3:![0-9]+]]
6413 // CHECK5: omp.inner.for.end:
6414 // CHECK5-NEXT: store i32 100, i32* [[I]], align 4
6415 // CHECK5-NEXT: store i32 0, i32* [[DOTOMP_LB3]], align 4
6416 // CHECK5-NEXT: store i32 99, i32* [[DOTOMP_UB4]], align 4
6417 // CHECK5-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_LB3]], align 4
6418 // CHECK5-NEXT: store i32 [[TMP5]], i32* [[DOTOMP_IV5]], align 4
6419 // CHECK5-NEXT: br label [[OMP_INNER_FOR_COND7:%.*]]
6420 // CHECK5: omp.inner.for.cond7:
6421 // CHECK5-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_IV5]], align 4, !llvm.access.group !6
6422 // CHECK5-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_UB4]], align 4, !llvm.access.group !6
6423 // CHECK5-NEXT: [[CMP8:%.*]] = icmp sle i32 [[TMP6]], [[TMP7]]
6424 // CHECK5-NEXT: br i1 [[CMP8]], label [[OMP_INNER_FOR_BODY9:%.*]], label [[OMP_INNER_FOR_END15:%.*]]
6425 // CHECK5: omp.inner.for.body9:
6426 // CHECK5-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_IV5]], align 4, !llvm.access.group !6
6427 // CHECK5-NEXT: [[MUL10:%.*]] = mul nsw i32 [[TMP8]], 1
6428 // CHECK5-NEXT: [[ADD11:%.*]] = add nsw i32 0, [[MUL10]]
6429 // CHECK5-NEXT: store i32 [[ADD11]], i32* [[I6]], align 4, !llvm.access.group !6
6430 // CHECK5-NEXT: call void @_Z9gtid_testv(), !llvm.access.group !6
6431 // CHECK5-NEXT: br label [[OMP_BODY_CONTINUE12:%.*]]
6432 // CHECK5: omp.body.continue12:
6433 // CHECK5-NEXT: br label [[OMP_INNER_FOR_INC13:%.*]]
6434 // CHECK5: omp.inner.for.inc13:
6435 // CHECK5-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_IV5]], align 4, !llvm.access.group !6
6436 // CHECK5-NEXT: [[ADD14:%.*]] = add nsw i32 [[TMP9]], 1
6437 // CHECK5-NEXT: store i32 [[ADD14]], i32* [[DOTOMP_IV5]], align 4, !llvm.access.group !6
6438 // CHECK5-NEXT: br label [[OMP_INNER_FOR_COND7]], !llvm.loop [[LOOP7:![0-9]+]]
6439 // CHECK5: omp.inner.for.end15:
6440 // CHECK5-NEXT: store i32 100, i32* [[I6]], align 4
6441 // CHECK5-NEXT: ret void
6442 //
6443 //
6444 // CHECK5-LABEL: define {{[^@]+}}@main
6445 // CHECK5-SAME: () #[[ATTR1:[0-9]+]] {
6446 // CHECK5-NEXT: entry:
6447 // CHECK5-NEXT: [[RETVAL:%.*]] = alloca i32, align 4
6448 // CHECK5-NEXT: [[TMP:%.*]] = alloca i32, align 4
6449 // CHECK5-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4
6450 // CHECK5-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4
6451 // CHECK5-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
6452 // CHECK5-NEXT: [[I:%.*]] = alloca i32, align 4
6453 // CHECK5-NEXT: [[_TMP2:%.*]] = alloca i32, align 4
6454 // CHECK5-NEXT: [[DOTOMP_LB3:%.*]] = alloca i32, align 4
6455 // CHECK5-NEXT: [[DOTOMP_UB4:%.*]] = alloca i32, align 4
6456 // CHECK5-NEXT: [[DOTOMP_IV5:%.*]] = alloca i32, align 4
6457 // CHECK5-NEXT: [[I6:%.*]] = alloca i32, align 4
6458 // CHECK5-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i8, align 1
6459 // CHECK5-NEXT: [[_TMP16:%.*]] = alloca i32, align 4
6460 // CHECK5-NEXT: [[DOTOMP_LB17:%.*]] = alloca i32, align 4
6461 // CHECK5-NEXT: [[DOTOMP_UB18:%.*]] = alloca i32, align 4
6462 // CHECK5-NEXT: [[DOTOMP_IV19:%.*]] = alloca i32, align 4
6463 // CHECK5-NEXT: [[I20:%.*]] = alloca i32, align 4
6464 // CHECK5-NEXT: store i32 0, i32* [[RETVAL]], align 4
6465 // CHECK5-NEXT: store i32 0, i32* [[DOTOMP_LB]], align 4
6466 // CHECK5-NEXT: store i32 99, i32* [[DOTOMP_UB]], align 4
6467 // CHECK5-NEXT: [[TMP0:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4
6468 // CHECK5-NEXT: store i32 [[TMP0]], i32* [[DOTOMP_IV]], align 4
6469 // CHECK5-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
6470 // CHECK5: omp.inner.for.cond:
6471 // CHECK5-NEXT: [[TMP1:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !9
6472 // CHECK5-NEXT: [[TMP2:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4, !llvm.access.group !9
6473 // CHECK5-NEXT: [[CMP:%.*]] = icmp sle i32 [[TMP1]], [[TMP2]]
6474 // CHECK5-NEXT: br i1 [[CMP]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
6475 // CHECK5: omp.inner.for.body:
6476 // CHECK5-NEXT: [[TMP3:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !9
6477 // CHECK5-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP3]], 1
6478 // CHECK5-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
6479 // CHECK5-NEXT: store i32 [[ADD]], i32* [[I]], align 4, !llvm.access.group !9
6480 // CHECK5-NEXT: call void @_Z3fn4v(), !llvm.access.group !9
6481 // CHECK5-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
6482 // CHECK5: omp.body.continue:
6483 // CHECK5-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
6484 // CHECK5: omp.inner.for.inc:
6485 // CHECK5-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !9
6486 // CHECK5-NEXT: [[ADD1:%.*]] = add nsw i32 [[TMP4]], 1
6487 // CHECK5-NEXT: store i32 [[ADD1]], i32* [[DOTOMP_IV]], align 4, !llvm.access.group !9
6488 // CHECK5-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP10:![0-9]+]]
6489 // CHECK5: omp.inner.for.end:
6490 // CHECK5-NEXT: store i32 100, i32* [[I]], align 4
6491 // CHECK5-NEXT: store i32 0, i32* [[DOTOMP_LB3]], align 4
6492 // CHECK5-NEXT: store i32 99, i32* [[DOTOMP_UB4]], align 4
6493 // CHECK5-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_LB3]], align 4
6494 // CHECK5-NEXT: store i32 [[TMP5]], i32* [[DOTOMP_IV5]], align 4
6495 // CHECK5-NEXT: br label [[OMP_INNER_FOR_COND7:%.*]]
6496 // CHECK5: omp.inner.for.cond7:
6497 // CHECK5-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_IV5]], align 4, !llvm.access.group !12
6498 // CHECK5-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_UB4]], align 4, !llvm.access.group !12
6499 // CHECK5-NEXT: [[CMP8:%.*]] = icmp sle i32 [[TMP6]], [[TMP7]]
6500 // CHECK5-NEXT: br i1 [[CMP8]], label [[OMP_INNER_FOR_BODY9:%.*]], label [[OMP_INNER_FOR_END15:%.*]]
6501 // CHECK5: omp.inner.for.body9:
6502 // CHECK5-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_IV5]], align 4, !llvm.access.group !12
6503 // CHECK5-NEXT: [[MUL10:%.*]] = mul nsw i32 [[TMP8]], 1
6504 // CHECK5-NEXT: [[ADD11:%.*]] = add nsw i32 0, [[MUL10]]
6505 // CHECK5-NEXT: store i32 [[ADD11]], i32* [[I6]], align 4, !llvm.access.group !12
6506 // CHECK5-NEXT: call void @_Z3fn5v(), !llvm.access.group !12
6507 // CHECK5-NEXT: br label [[OMP_BODY_CONTINUE12:%.*]]
6508 // CHECK5: omp.body.continue12:
6509 // CHECK5-NEXT: br label [[OMP_INNER_FOR_INC13:%.*]]
6510 // CHECK5: omp.inner.for.inc13:
6511 // CHECK5-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_IV5]], align 4, !llvm.access.group !12
6512 // CHECK5-NEXT: [[ADD14:%.*]] = add nsw i32 [[TMP9]], 1
6513 // CHECK5-NEXT: store i32 [[ADD14]], i32* [[DOTOMP_IV5]], align 4, !llvm.access.group !12
6514 // CHECK5-NEXT: br label [[OMP_INNER_FOR_COND7]], !llvm.loop [[LOOP13:![0-9]+]]
6515 // CHECK5: omp.inner.for.end15:
6516 // CHECK5-NEXT: store i32 100, i32* [[I6]], align 4
6517 // CHECK5-NEXT: [[TMP10:%.*]] = load i32, i32* @Arg, align 4
6518 // CHECK5-NEXT: [[TOBOOL:%.*]] = icmp ne i32 [[TMP10]], 0
6519 // CHECK5-NEXT: [[FROMBOOL:%.*]] = zext i1 [[TOBOOL]] to i8
6520 // CHECK5-NEXT: store i8 [[FROMBOOL]], i8* [[DOTCAPTURE_EXPR_]], align 1
6521 // CHECK5-NEXT: store i32 0, i32* [[DOTOMP_LB17]], align 4
6522 // CHECK5-NEXT: store i32 99, i32* [[DOTOMP_UB18]], align 4
6523 // CHECK5-NEXT: [[TMP11:%.*]] = load i32, i32* [[DOTOMP_LB17]], align 4
6524 // CHECK5-NEXT: store i32 [[TMP11]], i32* [[DOTOMP_IV19]], align 4
6525 // CHECK5-NEXT: br label [[OMP_INNER_FOR_COND21:%.*]]
6526 // CHECK5: omp.inner.for.cond21:
6527 // CHECK5-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTOMP_IV19]], align 4, !llvm.access.group !15
6528 // CHECK5-NEXT: [[TMP13:%.*]] = load i32, i32* [[DOTOMP_UB18]], align 4, !llvm.access.group !15
6529 // CHECK5-NEXT: [[CMP22:%.*]] = icmp sle i32 [[TMP12]], [[TMP13]]
6530 // CHECK5-NEXT: br i1 [[CMP22]], label [[OMP_INNER_FOR_BODY23:%.*]], label [[OMP_INNER_FOR_END29:%.*]]
6531 // CHECK5: omp.inner.for.body23:
6532 // CHECK5-NEXT: [[TMP14:%.*]] = load i32, i32* [[DOTOMP_IV19]], align 4, !llvm.access.group !15
6533 // CHECK5-NEXT: [[MUL24:%.*]] = mul nsw i32 [[TMP14]], 1
6534 // CHECK5-NEXT: [[ADD25:%.*]] = add nsw i32 0, [[MUL24]]
6535 // CHECK5-NEXT: store i32 [[ADD25]], i32* [[I20]], align 4, !llvm.access.group !15
6536 // CHECK5-NEXT: call void @_Z3fn6v(), !llvm.access.group !15
6537 // CHECK5-NEXT: br label [[OMP_BODY_CONTINUE26:%.*]]
6538 // CHECK5: omp.body.continue26:
6539 // CHECK5-NEXT: br label [[OMP_INNER_FOR_INC27:%.*]]
6540 // CHECK5: omp.inner.for.inc27:
6541 // CHECK5-NEXT: [[TMP15:%.*]] = load i32, i32* [[DOTOMP_IV19]], align 4, !llvm.access.group !15
6542 // CHECK5-NEXT: [[ADD28:%.*]] = add nsw i32 [[TMP15]], 1
6543 // CHECK5-NEXT: store i32 [[ADD28]], i32* [[DOTOMP_IV19]], align 4, !llvm.access.group !15
6544 // CHECK5-NEXT: br label [[OMP_INNER_FOR_COND21]], !llvm.loop [[LOOP16:![0-9]+]]
6545 // CHECK5: omp.inner.for.end29:
6546 // CHECK5-NEXT: store i32 100, i32* [[I20]], align 4
6547 // CHECK5-NEXT: [[TMP16:%.*]] = load i32, i32* @Arg, align 4
6548 // CHECK5-NEXT: [[CALL:%.*]] = call i32 @_Z5tmainIiEiT_(i32 [[TMP16]])
6549 // CHECK5-NEXT: ret i32 [[CALL]]
6550 //
6551 //
6552 // CHECK5-LABEL: define {{[^@]+}}@_Z5tmainIiEiT_
6553 // CHECK5-SAME: (i32 [[ARG:%.*]]) #[[ATTR0]] comdat {
6554 // CHECK5-NEXT: entry:
6555 // CHECK5-NEXT: [[ARG_ADDR:%.*]] = alloca i32, align 4
6556 // CHECK5-NEXT: [[TMP:%.*]] = alloca i32, align 4
6557 // CHECK5-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4
6558 // CHECK5-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4
6559 // CHECK5-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
6560 // CHECK5-NEXT: [[I:%.*]] = alloca i32, align 4
6561 // CHECK5-NEXT: [[_TMP2:%.*]] = alloca i32, align 4
6562 // CHECK5-NEXT: [[DOTOMP_LB3:%.*]] = alloca i32, align 4
6563 // CHECK5-NEXT: [[DOTOMP_UB4:%.*]] = alloca i32, align 4
6564 // CHECK5-NEXT: [[DOTOMP_IV5:%.*]] = alloca i32, align 4
6565 // CHECK5-NEXT: [[I6:%.*]] = alloca i32, align 4
6566 // CHECK5-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i8, align 1
6567 // CHECK5-NEXT: [[_TMP16:%.*]] = alloca i32, align 4
6568 // CHECK5-NEXT: [[DOTOMP_LB17:%.*]] = alloca i32, align 4
6569 // CHECK5-NEXT: [[DOTOMP_UB18:%.*]] = alloca i32, align 4
6570 // CHECK5-NEXT: [[DOTOMP_IV19:%.*]] = alloca i32, align 4
6571 // CHECK5-NEXT: [[I20:%.*]] = alloca i32, align 4
6572 // CHECK5-NEXT: store i32 [[ARG]], i32* [[ARG_ADDR]], align 4
6573 // CHECK5-NEXT: store i32 0, i32* [[DOTOMP_LB]], align 4
6574 // CHECK5-NEXT: store i32 99, i32* [[DOTOMP_UB]], align 4
6575 // CHECK5-NEXT: [[TMP0:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4
6576 // CHECK5-NEXT: store i32 [[TMP0]], i32* [[DOTOMP_IV]], align 4
6577 // CHECK5-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
6578 // CHECK5: omp.inner.for.cond:
6579 // CHECK5-NEXT: [[TMP1:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !18
6580 // CHECK5-NEXT: [[TMP2:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4, !llvm.access.group !18
6581 // CHECK5-NEXT: [[CMP:%.*]] = icmp sle i32 [[TMP1]], [[TMP2]]
6582 // CHECK5-NEXT: br i1 [[CMP]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
6583 // CHECK5: omp.inner.for.body:
6584 // CHECK5-NEXT: [[TMP3:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !18
6585 // CHECK5-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP3]], 1
6586 // CHECK5-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
6587 // CHECK5-NEXT: store i32 [[ADD]], i32* [[I]], align 4, !llvm.access.group !18
6588 // CHECK5-NEXT: call void @_Z3fn1v(), !llvm.access.group !18
6589 // CHECK5-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
6590 // CHECK5: omp.body.continue:
6591 // CHECK5-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
6592 // CHECK5: omp.inner.for.inc:
6593 // CHECK5-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !18
6594 // CHECK5-NEXT: [[ADD1:%.*]] = add nsw i32 [[TMP4]], 1
6595 // CHECK5-NEXT: store i32 [[ADD1]], i32* [[DOTOMP_IV]], align 4, !llvm.access.group !18
6596 // CHECK5-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP19:![0-9]+]]
6597 // CHECK5: omp.inner.for.end:
6598 // CHECK5-NEXT: store i32 100, i32* [[I]], align 4
6599 // CHECK5-NEXT: store i32 0, i32* [[DOTOMP_LB3]], align 4
6600 // CHECK5-NEXT: store i32 99, i32* [[DOTOMP_UB4]], align 4
6601 // CHECK5-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_LB3]], align 4
6602 // CHECK5-NEXT: store i32 [[TMP5]], i32* [[DOTOMP_IV5]], align 4
6603 // CHECK5-NEXT: br label [[OMP_INNER_FOR_COND7:%.*]]
6604 // CHECK5: omp.inner.for.cond7:
6605 // CHECK5-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_IV5]], align 4, !llvm.access.group !21
6606 // CHECK5-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_UB4]], align 4, !llvm.access.group !21
6607 // CHECK5-NEXT: [[CMP8:%.*]] = icmp sle i32 [[TMP6]], [[TMP7]]
6608 // CHECK5-NEXT: br i1 [[CMP8]], label [[OMP_INNER_FOR_BODY9:%.*]], label [[OMP_INNER_FOR_END15:%.*]]
6609 // CHECK5: omp.inner.for.body9:
6610 // CHECK5-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_IV5]], align 4, !llvm.access.group !21
6611 // CHECK5-NEXT: [[MUL10:%.*]] = mul nsw i32 [[TMP8]], 1
6612 // CHECK5-NEXT: [[ADD11:%.*]] = add nsw i32 0, [[MUL10]]
6613 // CHECK5-NEXT: store i32 [[ADD11]], i32* [[I6]], align 4, !llvm.access.group !21
6614 // CHECK5-NEXT: call void @_Z3fn2v(), !llvm.access.group !21
6615 // CHECK5-NEXT: br label [[OMP_BODY_CONTINUE12:%.*]]
6616 // CHECK5: omp.body.continue12:
6617 // CHECK5-NEXT: br label [[OMP_INNER_FOR_INC13:%.*]]
6618 // CHECK5: omp.inner.for.inc13:
6619 // CHECK5-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_IV5]], align 4, !llvm.access.group !21
6620 // CHECK5-NEXT: [[ADD14:%.*]] = add nsw i32 [[TMP9]], 1
6621 // CHECK5-NEXT: store i32 [[ADD14]], i32* [[DOTOMP_IV5]], align 4, !llvm.access.group !21
6622 // CHECK5-NEXT: br label [[OMP_INNER_FOR_COND7]], !llvm.loop [[LOOP22:![0-9]+]]
6623 // CHECK5: omp.inner.for.end15:
6624 // CHECK5-NEXT: store i32 100, i32* [[I6]], align 4
6625 // CHECK5-NEXT: [[TMP10:%.*]] = load i32, i32* [[ARG_ADDR]], align 4
6626 // CHECK5-NEXT: [[TOBOOL:%.*]] = icmp ne i32 [[TMP10]], 0
6627 // CHECK5-NEXT: [[FROMBOOL:%.*]] = zext i1 [[TOBOOL]] to i8
6628 // CHECK5-NEXT: store i8 [[FROMBOOL]], i8* [[DOTCAPTURE_EXPR_]], align 1
6629 // CHECK5-NEXT: store i32 0, i32* [[DOTOMP_LB17]], align 4
6630 // CHECK5-NEXT: store i32 99, i32* [[DOTOMP_UB18]], align 4
6631 // CHECK5-NEXT: [[TMP11:%.*]] = load i32, i32* [[DOTOMP_LB17]], align 4
6632 // CHECK5-NEXT: store i32 [[TMP11]], i32* [[DOTOMP_IV19]], align 4
6633 // CHECK5-NEXT: br label [[OMP_INNER_FOR_COND21:%.*]]
6634 // CHECK5: omp.inner.for.cond21:
6635 // CHECK5-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTOMP_IV19]], align 4, !llvm.access.group !24
6636 // CHECK5-NEXT: [[TMP13:%.*]] = load i32, i32* [[DOTOMP_UB18]], align 4, !llvm.access.group !24
6637 // CHECK5-NEXT: [[CMP22:%.*]] = icmp sle i32 [[TMP12]], [[TMP13]]
6638 // CHECK5-NEXT: br i1 [[CMP22]], label [[OMP_INNER_FOR_BODY23:%.*]], label [[OMP_INNER_FOR_END29:%.*]]
6639 // CHECK5: omp.inner.for.body23:
6640 // CHECK5-NEXT: [[TMP14:%.*]] = load i32, i32* [[DOTOMP_IV19]], align 4, !llvm.access.group !24
6641 // CHECK5-NEXT: [[MUL24:%.*]] = mul nsw i32 [[TMP14]], 1
6642 // CHECK5-NEXT: [[ADD25:%.*]] = add nsw i32 0, [[MUL24]]
6643 // CHECK5-NEXT: store i32 [[ADD25]], i32* [[I20]], align 4, !llvm.access.group !24
6644 // CHECK5-NEXT: call void @_Z3fn3v(), !llvm.access.group !24
6645 // CHECK5-NEXT: br label [[OMP_BODY_CONTINUE26:%.*]]
6646 // CHECK5: omp.body.continue26:
6647 // CHECK5-NEXT: br label [[OMP_INNER_FOR_INC27:%.*]]
6648 // CHECK5: omp.inner.for.inc27:
6649 // CHECK5-NEXT: [[TMP15:%.*]] = load i32, i32* [[DOTOMP_IV19]], align 4, !llvm.access.group !24
6650 // CHECK5-NEXT: [[ADD28:%.*]] = add nsw i32 [[TMP15]], 1
6651 // CHECK5-NEXT: store i32 [[ADD28]], i32* [[DOTOMP_IV19]], align 4, !llvm.access.group !24
6652 // CHECK5-NEXT: br label [[OMP_INNER_FOR_COND21]], !llvm.loop [[LOOP25:![0-9]+]]
6653 // CHECK5: omp.inner.for.end29:
6654 // CHECK5-NEXT: store i32 100, i32* [[I20]], align 4
6655 // CHECK5-NEXT: ret i32 0
6656 //
6657 //
6658 // CHECK6-LABEL: define {{[^@]+}}@_Z9gtid_testv
6659 // CHECK6-SAME: () #[[ATTR0:[0-9]+]] {
6660 // CHECK6-NEXT: entry:
6661 // CHECK6-NEXT: [[TMP:%.*]] = alloca i32, align 4
6662 // CHECK6-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4
6663 // CHECK6-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4
6664 // CHECK6-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
6665 // CHECK6-NEXT: [[I:%.*]] = alloca i32, align 4
6666 // CHECK6-NEXT: [[_TMP2:%.*]] = alloca i32, align 4
6667 // CHECK6-NEXT: [[DOTOMP_LB3:%.*]] = alloca i32, align 4
6668 // CHECK6-NEXT: [[DOTOMP_UB4:%.*]] = alloca i32, align 4
6669 // CHECK6-NEXT: [[DOTOMP_IV5:%.*]] = alloca i32, align 4
6670 // CHECK6-NEXT: [[I6:%.*]] = alloca i32, align 4
6671 // CHECK6-NEXT: store i32 0, i32* [[DOTOMP_LB]], align 4
6672 // CHECK6-NEXT: store i32 99, i32* [[DOTOMP_UB]], align 4
6673 // CHECK6-NEXT: [[TMP0:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4
6674 // CHECK6-NEXT: store i32 [[TMP0]], i32* [[DOTOMP_IV]], align 4
6675 // CHECK6-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
6676 // CHECK6: omp.inner.for.cond:
6677 // CHECK6-NEXT: [[TMP1:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !2
6678 // CHECK6-NEXT: [[TMP2:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4, !llvm.access.group !2
6679 // CHECK6-NEXT: [[CMP:%.*]] = icmp sle i32 [[TMP1]], [[TMP2]]
6680 // CHECK6-NEXT: br i1 [[CMP]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
6681 // CHECK6: omp.inner.for.body:
6682 // CHECK6-NEXT: [[TMP3:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !2
6683 // CHECK6-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP3]], 1
6684 // CHECK6-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
6685 // CHECK6-NEXT: store i32 [[ADD]], i32* [[I]], align 4, !llvm.access.group !2
6686 // CHECK6-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
6687 // CHECK6: omp.body.continue:
6688 // CHECK6-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
6689 // CHECK6: omp.inner.for.inc:
6690 // CHECK6-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !2
6691 // CHECK6-NEXT: [[ADD1:%.*]] = add nsw i32 [[TMP4]], 1
6692 // CHECK6-NEXT: store i32 [[ADD1]], i32* [[DOTOMP_IV]], align 4, !llvm.access.group !2
6693 // CHECK6-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP3:![0-9]+]]
6694 // CHECK6: omp.inner.for.end:
6695 // CHECK6-NEXT: store i32 100, i32* [[I]], align 4
6696 // CHECK6-NEXT: store i32 0, i32* [[DOTOMP_LB3]], align 4
6697 // CHECK6-NEXT: store i32 99, i32* [[DOTOMP_UB4]], align 4
6698 // CHECK6-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_LB3]], align 4
6699 // CHECK6-NEXT: store i32 [[TMP5]], i32* [[DOTOMP_IV5]], align 4
6700 // CHECK6-NEXT: br label [[OMP_INNER_FOR_COND7:%.*]]
6701 // CHECK6: omp.inner.for.cond7:
6702 // CHECK6-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_IV5]], align 4, !llvm.access.group !6
6703 // CHECK6-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_UB4]], align 4, !llvm.access.group !6
6704 // CHECK6-NEXT: [[CMP8:%.*]] = icmp sle i32 [[TMP6]], [[TMP7]]
6705 // CHECK6-NEXT: br i1 [[CMP8]], label [[OMP_INNER_FOR_BODY9:%.*]], label [[OMP_INNER_FOR_END15:%.*]]
6706 // CHECK6: omp.inner.for.body9:
6707 // CHECK6-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_IV5]], align 4, !llvm.access.group !6
6708 // CHECK6-NEXT: [[MUL10:%.*]] = mul nsw i32 [[TMP8]], 1
6709 // CHECK6-NEXT: [[ADD11:%.*]] = add nsw i32 0, [[MUL10]]
6710 // CHECK6-NEXT: store i32 [[ADD11]], i32* [[I6]], align 4, !llvm.access.group !6
6711 // CHECK6-NEXT: call void @_Z9gtid_testv(), !llvm.access.group !6
6712 // CHECK6-NEXT: br label [[OMP_BODY_CONTINUE12:%.*]]
6713 // CHECK6: omp.body.continue12:
6714 // CHECK6-NEXT: br label [[OMP_INNER_FOR_INC13:%.*]]
6715 // CHECK6: omp.inner.for.inc13:
6716 // CHECK6-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_IV5]], align 4, !llvm.access.group !6
6717 // CHECK6-NEXT: [[ADD14:%.*]] = add nsw i32 [[TMP9]], 1
6718 // CHECK6-NEXT: store i32 [[ADD14]], i32* [[DOTOMP_IV5]], align 4, !llvm.access.group !6
6719 // CHECK6-NEXT: br label [[OMP_INNER_FOR_COND7]], !llvm.loop [[LOOP7:![0-9]+]]
6720 // CHECK6: omp.inner.for.end15:
6721 // CHECK6-NEXT: store i32 100, i32* [[I6]], align 4
6722 // CHECK6-NEXT: ret void
6723 //
6724 //
6725 // CHECK6-LABEL: define {{[^@]+}}@main
6726 // CHECK6-SAME: () #[[ATTR1:[0-9]+]] {
6727 // CHECK6-NEXT: entry:
6728 // CHECK6-NEXT: [[RETVAL:%.*]] = alloca i32, align 4
6729 // CHECK6-NEXT: [[TMP:%.*]] = alloca i32, align 4
6730 // CHECK6-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4
6731 // CHECK6-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4
6732 // CHECK6-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
6733 // CHECK6-NEXT: [[I:%.*]] = alloca i32, align 4
6734 // CHECK6-NEXT: [[_TMP2:%.*]] = alloca i32, align 4
6735 // CHECK6-NEXT: [[DOTOMP_LB3:%.*]] = alloca i32, align 4
6736 // CHECK6-NEXT: [[DOTOMP_UB4:%.*]] = alloca i32, align 4
6737 // CHECK6-NEXT: [[DOTOMP_IV5:%.*]] = alloca i32, align 4
6738 // CHECK6-NEXT: [[I6:%.*]] = alloca i32, align 4
6739 // CHECK6-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i8, align 1
6740 // CHECK6-NEXT: [[_TMP16:%.*]] = alloca i32, align 4
6741 // CHECK6-NEXT: [[DOTOMP_LB17:%.*]] = alloca i32, align 4
6742 // CHECK6-NEXT: [[DOTOMP_UB18:%.*]] = alloca i32, align 4
6743 // CHECK6-NEXT: [[DOTOMP_IV19:%.*]] = alloca i32, align 4
6744 // CHECK6-NEXT: [[I20:%.*]] = alloca i32, align 4
6745 // CHECK6-NEXT: store i32 0, i32* [[RETVAL]], align 4
6746 // CHECK6-NEXT: store i32 0, i32* [[DOTOMP_LB]], align 4
6747 // CHECK6-NEXT: store i32 99, i32* [[DOTOMP_UB]], align 4
6748 // CHECK6-NEXT: [[TMP0:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4
6749 // CHECK6-NEXT: store i32 [[TMP0]], i32* [[DOTOMP_IV]], align 4
6750 // CHECK6-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
6751 // CHECK6: omp.inner.for.cond:
6752 // CHECK6-NEXT: [[TMP1:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !9
6753 // CHECK6-NEXT: [[TMP2:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4, !llvm.access.group !9
6754 // CHECK6-NEXT: [[CMP:%.*]] = icmp sle i32 [[TMP1]], [[TMP2]]
6755 // CHECK6-NEXT: br i1 [[CMP]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
6756 // CHECK6: omp.inner.for.body:
6757 // CHECK6-NEXT: [[TMP3:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !9
6758 // CHECK6-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP3]], 1
6759 // CHECK6-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
6760 // CHECK6-NEXT: store i32 [[ADD]], i32* [[I]], align 4, !llvm.access.group !9
6761 // CHECK6-NEXT: call void @_Z3fn4v(), !llvm.access.group !9
6762 // CHECK6-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
6763 // CHECK6: omp.body.continue:
6764 // CHECK6-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
6765 // CHECK6: omp.inner.for.inc:
6766 // CHECK6-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !9
6767 // CHECK6-NEXT: [[ADD1:%.*]] = add nsw i32 [[TMP4]], 1
6768 // CHECK6-NEXT: store i32 [[ADD1]], i32* [[DOTOMP_IV]], align 4, !llvm.access.group !9
6769 // CHECK6-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP10:![0-9]+]]
6770 // CHECK6: omp.inner.for.end:
6771 // CHECK6-NEXT: store i32 100, i32* [[I]], align 4
6772 // CHECK6-NEXT: store i32 0, i32* [[DOTOMP_LB3]], align 4
6773 // CHECK6-NEXT: store i32 99, i32* [[DOTOMP_UB4]], align 4
6774 // CHECK6-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_LB3]], align 4
6775 // CHECK6-NEXT: store i32 [[TMP5]], i32* [[DOTOMP_IV5]], align 4
6776 // CHECK6-NEXT: br label [[OMP_INNER_FOR_COND7:%.*]]
6777 // CHECK6: omp.inner.for.cond7:
6778 // CHECK6-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_IV5]], align 4, !llvm.access.group !12
6779 // CHECK6-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_UB4]], align 4, !llvm.access.group !12
6780 // CHECK6-NEXT: [[CMP8:%.*]] = icmp sle i32 [[TMP6]], [[TMP7]]
6781 // CHECK6-NEXT: br i1 [[CMP8]], label [[OMP_INNER_FOR_BODY9:%.*]], label [[OMP_INNER_FOR_END15:%.*]]
6782 // CHECK6: omp.inner.for.body9:
6783 // CHECK6-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_IV5]], align 4, !llvm.access.group !12
6784 // CHECK6-NEXT: [[MUL10:%.*]] = mul nsw i32 [[TMP8]], 1
6785 // CHECK6-NEXT: [[ADD11:%.*]] = add nsw i32 0, [[MUL10]]
6786 // CHECK6-NEXT: store i32 [[ADD11]], i32* [[I6]], align 4, !llvm.access.group !12
6787 // CHECK6-NEXT: call void @_Z3fn5v(), !llvm.access.group !12
6788 // CHECK6-NEXT: br label [[OMP_BODY_CONTINUE12:%.*]]
6789 // CHECK6: omp.body.continue12:
6790 // CHECK6-NEXT: br label [[OMP_INNER_FOR_INC13:%.*]]
6791 // CHECK6: omp.inner.for.inc13:
6792 // CHECK6-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_IV5]], align 4, !llvm.access.group !12
6793 // CHECK6-NEXT: [[ADD14:%.*]] = add nsw i32 [[TMP9]], 1
6794 // CHECK6-NEXT: store i32 [[ADD14]], i32* [[DOTOMP_IV5]], align 4, !llvm.access.group !12
6795 // CHECK6-NEXT: br label [[OMP_INNER_FOR_COND7]], !llvm.loop [[LOOP13:![0-9]+]]
6796 // CHECK6: omp.inner.for.end15:
6797 // CHECK6-NEXT: store i32 100, i32* [[I6]], align 4
6798 // CHECK6-NEXT: [[TMP10:%.*]] = load i32, i32* @Arg, align 4
6799 // CHECK6-NEXT: [[TOBOOL:%.*]] = icmp ne i32 [[TMP10]], 0
6800 // CHECK6-NEXT: [[FROMBOOL:%.*]] = zext i1 [[TOBOOL]] to i8
6801 // CHECK6-NEXT: store i8 [[FROMBOOL]], i8* [[DOTCAPTURE_EXPR_]], align 1
6802 // CHECK6-NEXT: store i32 0, i32* [[DOTOMP_LB17]], align 4
6803 // CHECK6-NEXT: store i32 99, i32* [[DOTOMP_UB18]], align 4
6804 // CHECK6-NEXT: [[TMP11:%.*]] = load i32, i32* [[DOTOMP_LB17]], align 4
6805 // CHECK6-NEXT: store i32 [[TMP11]], i32* [[DOTOMP_IV19]], align 4
6806 // CHECK6-NEXT: br label [[OMP_INNER_FOR_COND21:%.*]]
6807 // CHECK6: omp.inner.for.cond21:
6808 // CHECK6-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTOMP_IV19]], align 4, !llvm.access.group !15
6809 // CHECK6-NEXT: [[TMP13:%.*]] = load i32, i32* [[DOTOMP_UB18]], align 4, !llvm.access.group !15
6810 // CHECK6-NEXT: [[CMP22:%.*]] = icmp sle i32 [[TMP12]], [[TMP13]]
6811 // CHECK6-NEXT: br i1 [[CMP22]], label [[OMP_INNER_FOR_BODY23:%.*]], label [[OMP_INNER_FOR_END29:%.*]]
6812 // CHECK6: omp.inner.for.body23:
6813 // CHECK6-NEXT: [[TMP14:%.*]] = load i32, i32* [[DOTOMP_IV19]], align 4, !llvm.access.group !15
6814 // CHECK6-NEXT: [[MUL24:%.*]] = mul nsw i32 [[TMP14]], 1
6815 // CHECK6-NEXT: [[ADD25:%.*]] = add nsw i32 0, [[MUL24]]
6816 // CHECK6-NEXT: store i32 [[ADD25]], i32* [[I20]], align 4, !llvm.access.group !15
6817 // CHECK6-NEXT: call void @_Z3fn6v(), !llvm.access.group !15
6818 // CHECK6-NEXT: br label [[OMP_BODY_CONTINUE26:%.*]]
6819 // CHECK6: omp.body.continue26:
6820 // CHECK6-NEXT: br label [[OMP_INNER_FOR_INC27:%.*]]
6821 // CHECK6: omp.inner.for.inc27:
6822 // CHECK6-NEXT: [[TMP15:%.*]] = load i32, i32* [[DOTOMP_IV19]], align 4, !llvm.access.group !15
6823 // CHECK6-NEXT: [[ADD28:%.*]] = add nsw i32 [[TMP15]], 1
6824 // CHECK6-NEXT: store i32 [[ADD28]], i32* [[DOTOMP_IV19]], align 4, !llvm.access.group !15
6825 // CHECK6-NEXT: br label [[OMP_INNER_FOR_COND21]], !llvm.loop [[LOOP16:![0-9]+]]
6826 // CHECK6: omp.inner.for.end29:
6827 // CHECK6-NEXT: store i32 100, i32* [[I20]], align 4
6828 // CHECK6-NEXT: [[TMP16:%.*]] = load i32, i32* @Arg, align 4
6829 // CHECK6-NEXT: [[CALL:%.*]] = call i32 @_Z5tmainIiEiT_(i32 [[TMP16]])
6830 // CHECK6-NEXT: ret i32 [[CALL]]
6831 //
6832 //
6833 // CHECK6-LABEL: define {{[^@]+}}@_Z5tmainIiEiT_
6834 // CHECK6-SAME: (i32 [[ARG:%.*]]) #[[ATTR0]] comdat {
6835 // CHECK6-NEXT: entry:
6836 // CHECK6-NEXT: [[ARG_ADDR:%.*]] = alloca i32, align 4
6837 // CHECK6-NEXT: [[TMP:%.*]] = alloca i32, align 4
6838 // CHECK6-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4
6839 // CHECK6-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4
6840 // CHECK6-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
6841 // CHECK6-NEXT: [[I:%.*]] = alloca i32, align 4
6842 // CHECK6-NEXT: [[_TMP2:%.*]] = alloca i32, align 4
6843 // CHECK6-NEXT: [[DOTOMP_LB3:%.*]] = alloca i32, align 4
6844 // CHECK6-NEXT: [[DOTOMP_UB4:%.*]] = alloca i32, align 4
6845 // CHECK6-NEXT: [[DOTOMP_IV5:%.*]] = alloca i32, align 4
6846 // CHECK6-NEXT: [[I6:%.*]] = alloca i32, align 4
6847 // CHECK6-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i8, align 1
6848 // CHECK6-NEXT: [[_TMP16:%.*]] = alloca i32, align 4
6849 // CHECK6-NEXT: [[DOTOMP_LB17:%.*]] = alloca i32, align 4
6850 // CHECK6-NEXT: [[DOTOMP_UB18:%.*]] = alloca i32, align 4
6851 // CHECK6-NEXT: [[DOTOMP_IV19:%.*]] = alloca i32, align 4
6852 // CHECK6-NEXT: [[I20:%.*]] = alloca i32, align 4
6853 // CHECK6-NEXT: store i32 [[ARG]], i32* [[ARG_ADDR]], align 4
6854 // CHECK6-NEXT: store i32 0, i32* [[DOTOMP_LB]], align 4
6855 // CHECK6-NEXT: store i32 99, i32* [[DOTOMP_UB]], align 4
6856 // CHECK6-NEXT: [[TMP0:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4
6857 // CHECK6-NEXT: store i32 [[TMP0]], i32* [[DOTOMP_IV]], align 4
6858 // CHECK6-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
6859 // CHECK6: omp.inner.for.cond:
6860 // CHECK6-NEXT: [[TMP1:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !18
6861 // CHECK6-NEXT: [[TMP2:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4, !llvm.access.group !18
6862 // CHECK6-NEXT: [[CMP:%.*]] = icmp sle i32 [[TMP1]], [[TMP2]]
6863 // CHECK6-NEXT: br i1 [[CMP]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
6864 // CHECK6: omp.inner.for.body:
6865 // CHECK6-NEXT: [[TMP3:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !18
6866 // CHECK6-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP3]], 1
6867 // CHECK6-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
6868 // CHECK6-NEXT: store i32 [[ADD]], i32* [[I]], align 4, !llvm.access.group !18
6869 // CHECK6-NEXT: call void @_Z3fn1v(), !llvm.access.group !18
6870 // CHECK6-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
6871 // CHECK6: omp.body.continue:
6872 // CHECK6-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
6873 // CHECK6: omp.inner.for.inc:
6874 // CHECK6-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !18
6875 // CHECK6-NEXT: [[ADD1:%.*]] = add nsw i32 [[TMP4]], 1
6876 // CHECK6-NEXT: store i32 [[ADD1]], i32* [[DOTOMP_IV]], align 4, !llvm.access.group !18
6877 // CHECK6-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP19:![0-9]+]]
6878 // CHECK6: omp.inner.for.end:
6879 // CHECK6-NEXT: store i32 100, i32* [[I]], align 4
6880 // CHECK6-NEXT: store i32 0, i32* [[DOTOMP_LB3]], align 4
6881 // CHECK6-NEXT: store i32 99, i32* [[DOTOMP_UB4]], align 4
6882 // CHECK6-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_LB3]], align 4
6883 // CHECK6-NEXT: store i32 [[TMP5]], i32* [[DOTOMP_IV5]], align 4
6884 // CHECK6-NEXT: br label [[OMP_INNER_FOR_COND7:%.*]]
6885 // CHECK6: omp.inner.for.cond7:
6886 // CHECK6-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_IV5]], align 4, !llvm.access.group !21
6887 // CHECK6-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_UB4]], align 4, !llvm.access.group !21
6888 // CHECK6-NEXT: [[CMP8:%.*]] = icmp sle i32 [[TMP6]], [[TMP7]]
6889 // CHECK6-NEXT: br i1 [[CMP8]], label [[OMP_INNER_FOR_BODY9:%.*]], label [[OMP_INNER_FOR_END15:%.*]]
6890 // CHECK6: omp.inner.for.body9:
6891 // CHECK6-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_IV5]], align 4, !llvm.access.group !21
6892 // CHECK6-NEXT: [[MUL10:%.*]] = mul nsw i32 [[TMP8]], 1
6893 // CHECK6-NEXT: [[ADD11:%.*]] = add nsw i32 0, [[MUL10]]
6894 // CHECK6-NEXT: store i32 [[ADD11]], i32* [[I6]], align 4, !llvm.access.group !21
6895 // CHECK6-NEXT: call void @_Z3fn2v(), !llvm.access.group !21
6896 // CHECK6-NEXT: br label [[OMP_BODY_CONTINUE12:%.*]]
6897 // CHECK6: omp.body.continue12:
6898 // CHECK6-NEXT: br label [[OMP_INNER_FOR_INC13:%.*]]
6899 // CHECK6: omp.inner.for.inc13:
6900 // CHECK6-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_IV5]], align 4, !llvm.access.group !21
6901 // CHECK6-NEXT: [[ADD14:%.*]] = add nsw i32 [[TMP9]], 1
6902 // CHECK6-NEXT: store i32 [[ADD14]], i32* [[DOTOMP_IV5]], align 4, !llvm.access.group !21
6903 // CHECK6-NEXT: br label [[OMP_INNER_FOR_COND7]], !llvm.loop [[LOOP22:![0-9]+]]
6904 // CHECK6: omp.inner.for.end15:
6905 // CHECK6-NEXT: store i32 100, i32* [[I6]], align 4
6906 // CHECK6-NEXT: [[TMP10:%.*]] = load i32, i32* [[ARG_ADDR]], align 4
6907 // CHECK6-NEXT: [[TOBOOL:%.*]] = icmp ne i32 [[TMP10]], 0
6908 // CHECK6-NEXT: [[FROMBOOL:%.*]] = zext i1 [[TOBOOL]] to i8
6909 // CHECK6-NEXT: store i8 [[FROMBOOL]], i8* [[DOTCAPTURE_EXPR_]], align 1
6910 // CHECK6-NEXT: store i32 0, i32* [[DOTOMP_LB17]], align 4
6911 // CHECK6-NEXT: store i32 99, i32* [[DOTOMP_UB18]], align 4
6912 // CHECK6-NEXT: [[TMP11:%.*]] = load i32, i32* [[DOTOMP_LB17]], align 4
6913 // CHECK6-NEXT: store i32 [[TMP11]], i32* [[DOTOMP_IV19]], align 4
6914 // CHECK6-NEXT: br label [[OMP_INNER_FOR_COND21:%.*]]
6915 // CHECK6: omp.inner.for.cond21:
6916 // CHECK6-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTOMP_IV19]], align 4, !llvm.access.group !24
6917 // CHECK6-NEXT: [[TMP13:%.*]] = load i32, i32* [[DOTOMP_UB18]], align 4, !llvm.access.group !24
6918 // CHECK6-NEXT: [[CMP22:%.*]] = icmp sle i32 [[TMP12]], [[TMP13]]
6919 // CHECK6-NEXT: br i1 [[CMP22]], label [[OMP_INNER_FOR_BODY23:%.*]], label [[OMP_INNER_FOR_END29:%.*]]
6920 // CHECK6: omp.inner.for.body23:
6921 // CHECK6-NEXT: [[TMP14:%.*]] = load i32, i32* [[DOTOMP_IV19]], align 4, !llvm.access.group !24
6922 // CHECK6-NEXT: [[MUL24:%.*]] = mul nsw i32 [[TMP14]], 1
6923 // CHECK6-NEXT: [[ADD25:%.*]] = add nsw i32 0, [[MUL24]]
6924 // CHECK6-NEXT: store i32 [[ADD25]], i32* [[I20]], align 4, !llvm.access.group !24
6925 // CHECK6-NEXT: call void @_Z3fn3v(), !llvm.access.group !24
6926 // CHECK6-NEXT: br label [[OMP_BODY_CONTINUE26:%.*]]
6927 // CHECK6: omp.body.continue26:
6928 // CHECK6-NEXT: br label [[OMP_INNER_FOR_INC27:%.*]]
6929 // CHECK6: omp.inner.for.inc27:
6930 // CHECK6-NEXT: [[TMP15:%.*]] = load i32, i32* [[DOTOMP_IV19]], align 4, !llvm.access.group !24
6931 // CHECK6-NEXT: [[ADD28:%.*]] = add nsw i32 [[TMP15]], 1
6932 // CHECK6-NEXT: store i32 [[ADD28]], i32* [[DOTOMP_IV19]], align 4, !llvm.access.group !24
6933 // CHECK6-NEXT: br label [[OMP_INNER_FOR_COND21]], !llvm.loop [[LOOP25:![0-9]+]]
6934 // CHECK6: omp.inner.for.end29:
6935 // CHECK6-NEXT: store i32 100, i32* [[I20]], align 4
6936 // CHECK6-NEXT: ret i32 0
6937 //
6938 //
6939 // CHECK7-LABEL: define {{[^@]+}}@_Z9gtid_testv
6940 // CHECK7-SAME: () #[[ATTR0:[0-9]+]] {
6941 // CHECK7-NEXT: entry:
6942 // CHECK7-NEXT: [[TMP:%.*]] = alloca i32, align 4
6943 // CHECK7-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4
6944 // CHECK7-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4
6945 // CHECK7-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
6946 // CHECK7-NEXT: [[I:%.*]] = alloca i32, align 4
6947 // CHECK7-NEXT: [[_TMP2:%.*]] = alloca i32, align 4
6948 // CHECK7-NEXT: [[DOTOMP_LB3:%.*]] = alloca i32, align 4
6949 // CHECK7-NEXT: [[DOTOMP_UB4:%.*]] = alloca i32, align 4
6950 // CHECK7-NEXT: [[DOTOMP_IV5:%.*]] = alloca i32, align 4
6951 // CHECK7-NEXT: [[I6:%.*]] = alloca i32, align 4
6952 // CHECK7-NEXT: store i32 0, i32* [[DOTOMP_LB]], align 4
6953 // CHECK7-NEXT: store i32 99, i32* [[DOTOMP_UB]], align 4
6954 // CHECK7-NEXT: [[TMP0:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4
6955 // CHECK7-NEXT: store i32 [[TMP0]], i32* [[DOTOMP_IV]], align 4
6956 // CHECK7-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
6957 // CHECK7: omp.inner.for.cond:
6958 // CHECK7-NEXT: [[TMP1:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !2
6959 // CHECK7-NEXT: [[TMP2:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4, !llvm.access.group !2
6960 // CHECK7-NEXT: [[CMP:%.*]] = icmp sle i32 [[TMP1]], [[TMP2]]
6961 // CHECK7-NEXT: br i1 [[CMP]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
6962 // CHECK7: omp.inner.for.body:
6963 // CHECK7-NEXT: [[TMP3:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !2
6964 // CHECK7-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP3]], 1
6965 // CHECK7-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
6966 // CHECK7-NEXT: store i32 [[ADD]], i32* [[I]], align 4, !llvm.access.group !2
6967 // CHECK7-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
6968 // CHECK7: omp.body.continue:
6969 // CHECK7-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
6970 // CHECK7: omp.inner.for.inc:
6971 // CHECK7-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !2
6972 // CHECK7-NEXT: [[ADD1:%.*]] = add nsw i32 [[TMP4]], 1
6973 // CHECK7-NEXT: store i32 [[ADD1]], i32* [[DOTOMP_IV]], align 4, !llvm.access.group !2
6974 // CHECK7-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP3:![0-9]+]]
6975 // CHECK7: omp.inner.for.end:
6976 // CHECK7-NEXT: store i32 100, i32* [[I]], align 4
6977 // CHECK7-NEXT: store i32 0, i32* [[DOTOMP_LB3]], align 4
6978 // CHECK7-NEXT: store i32 99, i32* [[DOTOMP_UB4]], align 4
6979 // CHECK7-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_LB3]], align 4
6980 // CHECK7-NEXT: store i32 [[TMP5]], i32* [[DOTOMP_IV5]], align 4
6981 // CHECK7-NEXT: br label [[OMP_INNER_FOR_COND7:%.*]]
6982 // CHECK7: omp.inner.for.cond7:
6983 // CHECK7-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_IV5]], align 4, !llvm.access.group !6
6984 // CHECK7-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_UB4]], align 4, !llvm.access.group !6
6985 // CHECK7-NEXT: [[CMP8:%.*]] = icmp sle i32 [[TMP6]], [[TMP7]]
6986 // CHECK7-NEXT: br i1 [[CMP8]], label [[OMP_INNER_FOR_BODY9:%.*]], label [[OMP_INNER_FOR_END15:%.*]]
6987 // CHECK7: omp.inner.for.body9:
6988 // CHECK7-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_IV5]], align 4, !llvm.access.group !6
6989 // CHECK7-NEXT: [[MUL10:%.*]] = mul nsw i32 [[TMP8]], 1
6990 // CHECK7-NEXT: [[ADD11:%.*]] = add nsw i32 0, [[MUL10]]
6991 // CHECK7-NEXT: store i32 [[ADD11]], i32* [[I6]], align 4, !llvm.access.group !6
6992 // CHECK7-NEXT: call void @_Z9gtid_testv(), !llvm.access.group !6
6993 // CHECK7-NEXT: br label [[OMP_BODY_CONTINUE12:%.*]]
6994 // CHECK7: omp.body.continue12:
6995 // CHECK7-NEXT: br label [[OMP_INNER_FOR_INC13:%.*]]
6996 // CHECK7: omp.inner.for.inc13:
6997 // CHECK7-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_IV5]], align 4, !llvm.access.group !6
6998 // CHECK7-NEXT: [[ADD14:%.*]] = add nsw i32 [[TMP9]], 1
6999 // CHECK7-NEXT: store i32 [[ADD14]], i32* [[DOTOMP_IV5]], align 4, !llvm.access.group !6
7000 // CHECK7-NEXT: br label [[OMP_INNER_FOR_COND7]], !llvm.loop [[LOOP7:![0-9]+]]
7001 // CHECK7: omp.inner.for.end15:
7002 // CHECK7-NEXT: store i32 100, i32* [[I6]], align 4
7003 // CHECK7-NEXT: ret void
7004 //
7005 //
7006 // CHECK7-LABEL: define {{[^@]+}}@main
7007 // CHECK7-SAME: () #[[ATTR1:[0-9]+]] {
7008 // CHECK7-NEXT: entry:
7009 // CHECK7-NEXT: [[RETVAL:%.*]] = alloca i32, align 4
7010 // CHECK7-NEXT: [[TMP:%.*]] = alloca i32, align 4
7011 // CHECK7-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4
7012 // CHECK7-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4
7013 // CHECK7-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
7014 // CHECK7-NEXT: [[I:%.*]] = alloca i32, align 4
7015 // CHECK7-NEXT: [[_TMP2:%.*]] = alloca i32, align 4
7016 // CHECK7-NEXT: [[DOTOMP_LB3:%.*]] = alloca i32, align 4
7017 // CHECK7-NEXT: [[DOTOMP_UB4:%.*]] = alloca i32, align 4
7018 // CHECK7-NEXT: [[DOTOMP_IV5:%.*]] = alloca i32, align 4
7019 // CHECK7-NEXT: [[I6:%.*]] = alloca i32, align 4
7020 // CHECK7-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i8, align 1
7021 // CHECK7-NEXT: [[_TMP16:%.*]] = alloca i32, align 4
7022 // CHECK7-NEXT: [[DOTOMP_LB17:%.*]] = alloca i32, align 4
7023 // CHECK7-NEXT: [[DOTOMP_UB18:%.*]] = alloca i32, align 4
7024 // CHECK7-NEXT: [[DOTOMP_IV19:%.*]] = alloca i32, align 4
7025 // CHECK7-NEXT: [[I20:%.*]] = alloca i32, align 4
7026 // CHECK7-NEXT: store i32 0, i32* [[RETVAL]], align 4
7027 // CHECK7-NEXT: store i32 0, i32* [[DOTOMP_LB]], align 4
7028 // CHECK7-NEXT: store i32 99, i32* [[DOTOMP_UB]], align 4
7029 // CHECK7-NEXT: [[TMP0:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4
7030 // CHECK7-NEXT: store i32 [[TMP0]], i32* [[DOTOMP_IV]], align 4
7031 // CHECK7-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
7032 // CHECK7: omp.inner.for.cond:
7033 // CHECK7-NEXT: [[TMP1:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !9
7034 // CHECK7-NEXT: [[TMP2:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4, !llvm.access.group !9
7035 // CHECK7-NEXT: [[CMP:%.*]] = icmp sle i32 [[TMP1]], [[TMP2]]
7036 // CHECK7-NEXT: br i1 [[CMP]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
7037 // CHECK7: omp.inner.for.body:
7038 // CHECK7-NEXT: [[TMP3:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !9
7039 // CHECK7-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP3]], 1
7040 // CHECK7-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
7041 // CHECK7-NEXT: store i32 [[ADD]], i32* [[I]], align 4, !llvm.access.group !9
7042 // CHECK7-NEXT: call void @_Z3fn4v(), !llvm.access.group !9
7043 // CHECK7-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
7044 // CHECK7: omp.body.continue:
7045 // CHECK7-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
7046 // CHECK7: omp.inner.for.inc:
7047 // CHECK7-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !9
7048 // CHECK7-NEXT: [[ADD1:%.*]] = add nsw i32 [[TMP4]], 1
7049 // CHECK7-NEXT: store i32 [[ADD1]], i32* [[DOTOMP_IV]], align 4, !llvm.access.group !9
7050 // CHECK7-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP10:![0-9]+]]
7051 // CHECK7: omp.inner.for.end:
7052 // CHECK7-NEXT: store i32 100, i32* [[I]], align 4
7053 // CHECK7-NEXT: store i32 0, i32* [[DOTOMP_LB3]], align 4
7054 // CHECK7-NEXT: store i32 99, i32* [[DOTOMP_UB4]], align 4
7055 // CHECK7-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_LB3]], align 4
7056 // CHECK7-NEXT: store i32 [[TMP5]], i32* [[DOTOMP_IV5]], align 4
7057 // CHECK7-NEXT: br label [[OMP_INNER_FOR_COND7:%.*]]
7058 // CHECK7: omp.inner.for.cond7:
7059 // CHECK7-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_IV5]], align 4
7060 // CHECK7-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_UB4]], align 4
7061 // CHECK7-NEXT: [[CMP8:%.*]] = icmp sle i32 [[TMP6]], [[TMP7]]
7062 // CHECK7-NEXT: br i1 [[CMP8]], label [[OMP_INNER_FOR_BODY9:%.*]], label [[OMP_INNER_FOR_END15:%.*]]
7063 // CHECK7: omp.inner.for.body9:
7064 // CHECK7-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_IV5]], align 4
7065 // CHECK7-NEXT: [[MUL10:%.*]] = mul nsw i32 [[TMP8]], 1
7066 // CHECK7-NEXT: [[ADD11:%.*]] = add nsw i32 0, [[MUL10]]
7067 // CHECK7-NEXT: store i32 [[ADD11]], i32* [[I6]], align 4
7068 // CHECK7-NEXT: call void @_Z3fn5v()
7069 // CHECK7-NEXT: br label [[OMP_BODY_CONTINUE12:%.*]]
7070 // CHECK7: omp.body.continue12:
7071 // CHECK7-NEXT: br label [[OMP_INNER_FOR_INC13:%.*]]
7072 // CHECK7: omp.inner.for.inc13:
7073 // CHECK7-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_IV5]], align 4
7074 // CHECK7-NEXT: [[ADD14:%.*]] = add nsw i32 [[TMP9]], 1
7075 // CHECK7-NEXT: store i32 [[ADD14]], i32* [[DOTOMP_IV5]], align 4
7076 // CHECK7-NEXT: br label [[OMP_INNER_FOR_COND7]], !llvm.loop [[LOOP12:![0-9]+]]
7077 // CHECK7: omp.inner.for.end15:
7078 // CHECK7-NEXT: store i32 100, i32* [[I6]], align 4
7079 // CHECK7-NEXT: [[TMP10:%.*]] = load i32, i32* @Arg, align 4
7080 // CHECK7-NEXT: [[TOBOOL:%.*]] = icmp ne i32 [[TMP10]], 0
7081 // CHECK7-NEXT: [[FROMBOOL:%.*]] = zext i1 [[TOBOOL]] to i8
7082 // CHECK7-NEXT: store i8 [[FROMBOOL]], i8* [[DOTCAPTURE_EXPR_]], align 1
7083 // CHECK7-NEXT: store i32 0, i32* [[DOTOMP_LB17]], align 4
7084 // CHECK7-NEXT: store i32 99, i32* [[DOTOMP_UB18]], align 4
7085 // CHECK7-NEXT: [[TMP11:%.*]] = load i32, i32* [[DOTOMP_LB17]], align 4
7086 // CHECK7-NEXT: store i32 [[TMP11]], i32* [[DOTOMP_IV19]], align 4
7087 // CHECK7-NEXT: [[TMP12:%.*]] = load i8, i8* [[DOTCAPTURE_EXPR_]], align 1
7088 // CHECK7-NEXT: [[TOBOOL21:%.*]] = trunc i8 [[TMP12]] to i1
7089 // CHECK7-NEXT: br i1 [[TOBOOL21]], label [[OMP_IF_THEN:%.*]], label [[OMP_IF_ELSE:%.*]]
7090 // CHECK7: omp_if.then:
7091 // CHECK7-NEXT: br label [[OMP_INNER_FOR_COND22:%.*]]
7092 // CHECK7: omp.inner.for.cond22:
7093 // CHECK7-NEXT: [[TMP13:%.*]] = load i32, i32* [[DOTOMP_IV19]], align 4, !llvm.access.group !14
7094 // CHECK7-NEXT: [[TMP14:%.*]] = load i32, i32* [[DOTOMP_UB18]], align 4, !llvm.access.group !14
7095 // CHECK7-NEXT: [[CMP23:%.*]] = icmp sle i32 [[TMP13]], [[TMP14]]
7096 // CHECK7-NEXT: br i1 [[CMP23]], label [[OMP_INNER_FOR_BODY24:%.*]], label [[OMP_INNER_FOR_END30:%.*]]
7097 // CHECK7: omp.inner.for.body24:
7098 // CHECK7-NEXT: [[TMP15:%.*]] = load i32, i32* [[DOTOMP_IV19]], align 4, !llvm.access.group !14
7099 // CHECK7-NEXT: [[MUL25:%.*]] = mul nsw i32 [[TMP15]], 1
7100 // CHECK7-NEXT: [[ADD26:%.*]] = add nsw i32 0, [[MUL25]]
7101 // CHECK7-NEXT: store i32 [[ADD26]], i32* [[I20]], align 4, !llvm.access.group !14
7102 // CHECK7-NEXT: call void @_Z3fn6v(), !llvm.access.group !14
7103 // CHECK7-NEXT: br label [[OMP_BODY_CONTINUE27:%.*]]
7104 // CHECK7: omp.body.continue27:
7105 // CHECK7-NEXT: br label [[OMP_INNER_FOR_INC28:%.*]]
7106 // CHECK7: omp.inner.for.inc28:
7107 // CHECK7-NEXT: [[TMP16:%.*]] = load i32, i32* [[DOTOMP_IV19]], align 4, !llvm.access.group !14
7108 // CHECK7-NEXT: [[ADD29:%.*]] = add nsw i32 [[TMP16]], 1
7109 // CHECK7-NEXT: store i32 [[ADD29]], i32* [[DOTOMP_IV19]], align 4, !llvm.access.group !14
7110 // CHECK7-NEXT: br label [[OMP_INNER_FOR_COND22]], !llvm.loop [[LOOP15:![0-9]+]]
7111 // CHECK7: omp.inner.for.end30:
7112 // CHECK7-NEXT: br label [[OMP_IF_END:%.*]]
7113 // CHECK7: omp_if.else:
7114 // CHECK7-NEXT: br label [[OMP_INNER_FOR_COND31:%.*]]
7115 // CHECK7: omp.inner.for.cond31:
7116 // CHECK7-NEXT: [[TMP17:%.*]] = load i32, i32* [[DOTOMP_IV19]], align 4
7117 // CHECK7-NEXT: [[TMP18:%.*]] = load i32, i32* [[DOTOMP_UB18]], align 4
7118 // CHECK7-NEXT: [[CMP32:%.*]] = icmp sle i32 [[TMP17]], [[TMP18]]
7119 // CHECK7-NEXT: br i1 [[CMP32]], label [[OMP_INNER_FOR_BODY33:%.*]], label [[OMP_INNER_FOR_END39:%.*]]
7120 // CHECK7: omp.inner.for.body33:
7121 // CHECK7-NEXT: [[TMP19:%.*]] = load i32, i32* [[DOTOMP_IV19]], align 4
7122 // CHECK7-NEXT: [[MUL34:%.*]] = mul nsw i32 [[TMP19]], 1
7123 // CHECK7-NEXT: [[ADD35:%.*]] = add nsw i32 0, [[MUL34]]
7124 // CHECK7-NEXT: store i32 [[ADD35]], i32* [[I20]], align 4
7125 // CHECK7-NEXT: call void @_Z3fn6v()
7126 // CHECK7-NEXT: br label [[OMP_BODY_CONTINUE36:%.*]]
7127 // CHECK7: omp.body.continue36:
7128 // CHECK7-NEXT: br label [[OMP_INNER_FOR_INC37:%.*]]
7129 // CHECK7: omp.inner.for.inc37:
7130 // CHECK7-NEXT: [[TMP20:%.*]] = load i32, i32* [[DOTOMP_IV19]], align 4
7131 // CHECK7-NEXT: [[ADD38:%.*]] = add nsw i32 [[TMP20]], 1
7132 // CHECK7-NEXT: store i32 [[ADD38]], i32* [[DOTOMP_IV19]], align 4
7133 // CHECK7-NEXT: br label [[OMP_INNER_FOR_COND31]], !llvm.loop [[LOOP17:![0-9]+]]
7134 // CHECK7: omp.inner.for.end39:
7135 // CHECK7-NEXT: br label [[OMP_IF_END]]
7136 // CHECK7: omp_if.end:
7137 // CHECK7-NEXT: store i32 100, i32* [[I20]], align 4
7138 // CHECK7-NEXT: [[TMP21:%.*]] = load i32, i32* @Arg, align 4
7139 // CHECK7-NEXT: [[CALL:%.*]] = call i32 @_Z5tmainIiEiT_(i32 [[TMP21]])
7140 // CHECK7-NEXT: ret i32 [[CALL]]
7141 //
7142 //
7143 // CHECK7-LABEL: define {{[^@]+}}@_Z5tmainIiEiT_
7144 // CHECK7-SAME: (i32 [[ARG:%.*]]) #[[ATTR0]] comdat {
7145 // CHECK7-NEXT: entry:
7146 // CHECK7-NEXT: [[ARG_ADDR:%.*]] = alloca i32, align 4
7147 // CHECK7-NEXT: [[TMP:%.*]] = alloca i32, align 4
7148 // CHECK7-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4
7149 // CHECK7-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4
7150 // CHECK7-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
7151 // CHECK7-NEXT: [[I:%.*]] = alloca i32, align 4
7152 // CHECK7-NEXT: [[_TMP2:%.*]] = alloca i32, align 4
7153 // CHECK7-NEXT: [[DOTOMP_LB3:%.*]] = alloca i32, align 4
7154 // CHECK7-NEXT: [[DOTOMP_UB4:%.*]] = alloca i32, align 4
7155 // CHECK7-NEXT: [[DOTOMP_IV5:%.*]] = alloca i32, align 4
7156 // CHECK7-NEXT: [[I6:%.*]] = alloca i32, align 4
7157 // CHECK7-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i8, align 1
7158 // CHECK7-NEXT: [[_TMP16:%.*]] = alloca i32, align 4
7159 // CHECK7-NEXT: [[DOTOMP_LB17:%.*]] = alloca i32, align 4
7160 // CHECK7-NEXT: [[DOTOMP_UB18:%.*]] = alloca i32, align 4
7161 // CHECK7-NEXT: [[DOTOMP_IV19:%.*]] = alloca i32, align 4
7162 // CHECK7-NEXT: [[I20:%.*]] = alloca i32, align 4
7163 // CHECK7-NEXT: store i32 [[ARG]], i32* [[ARG_ADDR]], align 4
7164 // CHECK7-NEXT: store i32 0, i32* [[DOTOMP_LB]], align 4
7165 // CHECK7-NEXT: store i32 99, i32* [[DOTOMP_UB]], align 4
7166 // CHECK7-NEXT: [[TMP0:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4
7167 // CHECK7-NEXT: store i32 [[TMP0]], i32* [[DOTOMP_IV]], align 4
7168 // CHECK7-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
7169 // CHECK7: omp.inner.for.cond:
7170 // CHECK7-NEXT: [[TMP1:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !18
7171 // CHECK7-NEXT: [[TMP2:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4, !llvm.access.group !18
7172 // CHECK7-NEXT: [[CMP:%.*]] = icmp sle i32 [[TMP1]], [[TMP2]]
7173 // CHECK7-NEXT: br i1 [[CMP]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
7174 // CHECK7: omp.inner.for.body:
7175 // CHECK7-NEXT: [[TMP3:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !18
7176 // CHECK7-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP3]], 1
7177 // CHECK7-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
7178 // CHECK7-NEXT: store i32 [[ADD]], i32* [[I]], align 4, !llvm.access.group !18
7179 // CHECK7-NEXT: call void @_Z3fn1v(), !llvm.access.group !18
7180 // CHECK7-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
7181 // CHECK7: omp.body.continue:
7182 // CHECK7-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
7183 // CHECK7: omp.inner.for.inc:
7184 // CHECK7-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !18
7185 // CHECK7-NEXT: [[ADD1:%.*]] = add nsw i32 [[TMP4]], 1
7186 // CHECK7-NEXT: store i32 [[ADD1]], i32* [[DOTOMP_IV]], align 4, !llvm.access.group !18
7187 // CHECK7-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP19:![0-9]+]]
7188 // CHECK7: omp.inner.for.end:
7189 // CHECK7-NEXT: store i32 100, i32* [[I]], align 4
7190 // CHECK7-NEXT: store i32 0, i32* [[DOTOMP_LB3]], align 4
7191 // CHECK7-NEXT: store i32 99, i32* [[DOTOMP_UB4]], align 4
7192 // CHECK7-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_LB3]], align 4
7193 // CHECK7-NEXT: store i32 [[TMP5]], i32* [[DOTOMP_IV5]], align 4
7194 // CHECK7-NEXT: br label [[OMP_INNER_FOR_COND7:%.*]]
7195 // CHECK7: omp.inner.for.cond7:
7196 // CHECK7-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_IV5]], align 4
7197 // CHECK7-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_UB4]], align 4
7198 // CHECK7-NEXT: [[CMP8:%.*]] = icmp sle i32 [[TMP6]], [[TMP7]]
7199 // CHECK7-NEXT: br i1 [[CMP8]], label [[OMP_INNER_FOR_BODY9:%.*]], label [[OMP_INNER_FOR_END15:%.*]]
7200 // CHECK7: omp.inner.for.body9:
7201 // CHECK7-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_IV5]], align 4
7202 // CHECK7-NEXT: [[MUL10:%.*]] = mul nsw i32 [[TMP8]], 1
7203 // CHECK7-NEXT: [[ADD11:%.*]] = add nsw i32 0, [[MUL10]]
7204 // CHECK7-NEXT: store i32 [[ADD11]], i32* [[I6]], align 4
7205 // CHECK7-NEXT: call void @_Z3fn2v()
7206 // CHECK7-NEXT: br label [[OMP_BODY_CONTINUE12:%.*]]
7207 // CHECK7: omp.body.continue12:
7208 // CHECK7-NEXT: br label [[OMP_INNER_FOR_INC13:%.*]]
7209 // CHECK7: omp.inner.for.inc13:
7210 // CHECK7-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_IV5]], align 4
7211 // CHECK7-NEXT: [[ADD14:%.*]] = add nsw i32 [[TMP9]], 1
7212 // CHECK7-NEXT: store i32 [[ADD14]], i32* [[DOTOMP_IV5]], align 4
7213 // CHECK7-NEXT: br label [[OMP_INNER_FOR_COND7]], !llvm.loop [[LOOP21:![0-9]+]]
7214 // CHECK7: omp.inner.for.end15:
7215 // CHECK7-NEXT: store i32 100, i32* [[I6]], align 4
7216 // CHECK7-NEXT: [[TMP10:%.*]] = load i32, i32* [[ARG_ADDR]], align 4
7217 // CHECK7-NEXT: [[TOBOOL:%.*]] = icmp ne i32 [[TMP10]], 0
7218 // CHECK7-NEXT: [[FROMBOOL:%.*]] = zext i1 [[TOBOOL]] to i8
7219 // CHECK7-NEXT: store i8 [[FROMBOOL]], i8* [[DOTCAPTURE_EXPR_]], align 1
7220 // CHECK7-NEXT: store i32 0, i32* [[DOTOMP_LB17]], align 4
7221 // CHECK7-NEXT: store i32 99, i32* [[DOTOMP_UB18]], align 4
7222 // CHECK7-NEXT: [[TMP11:%.*]] = load i32, i32* [[DOTOMP_LB17]], align 4
7223 // CHECK7-NEXT: store i32 [[TMP11]], i32* [[DOTOMP_IV19]], align 4
7224 // CHECK7-NEXT: br label [[OMP_INNER_FOR_COND21:%.*]]
7225 // CHECK7: omp.inner.for.cond21:
7226 // CHECK7-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTOMP_IV19]], align 4, !llvm.access.group !22
7227 // CHECK7-NEXT: [[TMP13:%.*]] = load i32, i32* [[DOTOMP_UB18]], align 4, !llvm.access.group !22
7228 // CHECK7-NEXT: [[CMP22:%.*]] = icmp sle i32 [[TMP12]], [[TMP13]]
7229 // CHECK7-NEXT: br i1 [[CMP22]], label [[OMP_INNER_FOR_BODY23:%.*]], label [[OMP_INNER_FOR_END29:%.*]]
7230 // CHECK7: omp.inner.for.body23:
7231 // CHECK7-NEXT: [[TMP14:%.*]] = load i32, i32* [[DOTOMP_IV19]], align 4, !llvm.access.group !22
7232 // CHECK7-NEXT: [[MUL24:%.*]] = mul nsw i32 [[TMP14]], 1
7233 // CHECK7-NEXT: [[ADD25:%.*]] = add nsw i32 0, [[MUL24]]
7234 // CHECK7-NEXT: store i32 [[ADD25]], i32* [[I20]], align 4, !llvm.access.group !22
7235 // CHECK7-NEXT: call void @_Z3fn3v(), !llvm.access.group !22
7236 // CHECK7-NEXT: br label [[OMP_BODY_CONTINUE26:%.*]]
7237 // CHECK7: omp.body.continue26:
7238 // CHECK7-NEXT: br label [[OMP_INNER_FOR_INC27:%.*]]
7239 // CHECK7: omp.inner.for.inc27:
7240 // CHECK7-NEXT: [[TMP15:%.*]] = load i32, i32* [[DOTOMP_IV19]], align 4, !llvm.access.group !22
7241 // CHECK7-NEXT: [[ADD28:%.*]] = add nsw i32 [[TMP15]], 1
7242 // CHECK7-NEXT: store i32 [[ADD28]], i32* [[DOTOMP_IV19]], align 4, !llvm.access.group !22
7243 // CHECK7-NEXT: br label [[OMP_INNER_FOR_COND21]], !llvm.loop [[LOOP23:![0-9]+]]
7244 // CHECK7: omp.inner.for.end29:
7245 // CHECK7-NEXT: store i32 100, i32* [[I20]], align 4
7246 // CHECK7-NEXT: ret i32 0
7247 //
7248 //
7249 // CHECK8-LABEL: define {{[^@]+}}@_Z9gtid_testv
7250 // CHECK8-SAME: () #[[ATTR0:[0-9]+]] {
7251 // CHECK8-NEXT: entry:
7252 // CHECK8-NEXT: [[TMP:%.*]] = alloca i32, align 4
7253 // CHECK8-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4
7254 // CHECK8-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4
7255 // CHECK8-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
7256 // CHECK8-NEXT: [[I:%.*]] = alloca i32, align 4
7257 // CHECK8-NEXT: [[_TMP2:%.*]] = alloca i32, align 4
7258 // CHECK8-NEXT: [[DOTOMP_LB3:%.*]] = alloca i32, align 4
7259 // CHECK8-NEXT: [[DOTOMP_UB4:%.*]] = alloca i32, align 4
7260 // CHECK8-NEXT: [[DOTOMP_IV5:%.*]] = alloca i32, align 4
7261 // CHECK8-NEXT: [[I6:%.*]] = alloca i32, align 4
7262 // CHECK8-NEXT: store i32 0, i32* [[DOTOMP_LB]], align 4
7263 // CHECK8-NEXT: store i32 99, i32* [[DOTOMP_UB]], align 4
7264 // CHECK8-NEXT: [[TMP0:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4
7265 // CHECK8-NEXT: store i32 [[TMP0]], i32* [[DOTOMP_IV]], align 4
7266 // CHECK8-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
7267 // CHECK8: omp.inner.for.cond:
7268 // CHECK8-NEXT: [[TMP1:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !2
7269 // CHECK8-NEXT: [[TMP2:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4, !llvm.access.group !2
7270 // CHECK8-NEXT: [[CMP:%.*]] = icmp sle i32 [[TMP1]], [[TMP2]]
7271 // CHECK8-NEXT: br i1 [[CMP]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
7272 // CHECK8: omp.inner.for.body:
7273 // CHECK8-NEXT: [[TMP3:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !2
7274 // CHECK8-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP3]], 1
7275 // CHECK8-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
7276 // CHECK8-NEXT: store i32 [[ADD]], i32* [[I]], align 4, !llvm.access.group !2
7277 // CHECK8-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
7278 // CHECK8: omp.body.continue:
7279 // CHECK8-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
7280 // CHECK8: omp.inner.for.inc:
7281 // CHECK8-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !2
7282 // CHECK8-NEXT: [[ADD1:%.*]] = add nsw i32 [[TMP4]], 1
7283 // CHECK8-NEXT: store i32 [[ADD1]], i32* [[DOTOMP_IV]], align 4, !llvm.access.group !2
7284 // CHECK8-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP3:![0-9]+]]
7285 // CHECK8: omp.inner.for.end:
7286 // CHECK8-NEXT: store i32 100, i32* [[I]], align 4
7287 // CHECK8-NEXT: store i32 0, i32* [[DOTOMP_LB3]], align 4
7288 // CHECK8-NEXT: store i32 99, i32* [[DOTOMP_UB4]], align 4
7289 // CHECK8-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_LB3]], align 4
7290 // CHECK8-NEXT: store i32 [[TMP5]], i32* [[DOTOMP_IV5]], align 4
7291 // CHECK8-NEXT: br label [[OMP_INNER_FOR_COND7:%.*]]
7292 // CHECK8: omp.inner.for.cond7:
7293 // CHECK8-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_IV5]], align 4, !llvm.access.group !6
7294 // CHECK8-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_UB4]], align 4, !llvm.access.group !6
7295 // CHECK8-NEXT: [[CMP8:%.*]] = icmp sle i32 [[TMP6]], [[TMP7]]
7296 // CHECK8-NEXT: br i1 [[CMP8]], label [[OMP_INNER_FOR_BODY9:%.*]], label [[OMP_INNER_FOR_END15:%.*]]
7297 // CHECK8: omp.inner.for.body9:
7298 // CHECK8-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_IV5]], align 4, !llvm.access.group !6
7299 // CHECK8-NEXT: [[MUL10:%.*]] = mul nsw i32 [[TMP8]], 1
7300 // CHECK8-NEXT: [[ADD11:%.*]] = add nsw i32 0, [[MUL10]]
7301 // CHECK8-NEXT: store i32 [[ADD11]], i32* [[I6]], align 4, !llvm.access.group !6
7302 // CHECK8-NEXT: call void @_Z9gtid_testv(), !llvm.access.group !6
7303 // CHECK8-NEXT: br label [[OMP_BODY_CONTINUE12:%.*]]
7304 // CHECK8: omp.body.continue12:
7305 // CHECK8-NEXT: br label [[OMP_INNER_FOR_INC13:%.*]]
7306 // CHECK8: omp.inner.for.inc13:
7307 // CHECK8-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_IV5]], align 4, !llvm.access.group !6
7308 // CHECK8-NEXT: [[ADD14:%.*]] = add nsw i32 [[TMP9]], 1
7309 // CHECK8-NEXT: store i32 [[ADD14]], i32* [[DOTOMP_IV5]], align 4, !llvm.access.group !6
7310 // CHECK8-NEXT: br label [[OMP_INNER_FOR_COND7]], !llvm.loop [[LOOP7:![0-9]+]]
7311 // CHECK8: omp.inner.for.end15:
7312 // CHECK8-NEXT: store i32 100, i32* [[I6]], align 4
7313 // CHECK8-NEXT: ret void
7314 //
7315 //
7316 // CHECK8-LABEL: define {{[^@]+}}@main
7317 // CHECK8-SAME: () #[[ATTR1:[0-9]+]] {
7318 // CHECK8-NEXT: entry:
7319 // CHECK8-NEXT: [[RETVAL:%.*]] = alloca i32, align 4
7320 // CHECK8-NEXT: [[TMP:%.*]] = alloca i32, align 4
7321 // CHECK8-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4
7322 // CHECK8-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4
7323 // CHECK8-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
7324 // CHECK8-NEXT: [[I:%.*]] = alloca i32, align 4
7325 // CHECK8-NEXT: [[_TMP2:%.*]] = alloca i32, align 4
7326 // CHECK8-NEXT: [[DOTOMP_LB3:%.*]] = alloca i32, align 4
7327 // CHECK8-NEXT: [[DOTOMP_UB4:%.*]] = alloca i32, align 4
7328 // CHECK8-NEXT: [[DOTOMP_IV5:%.*]] = alloca i32, align 4
7329 // CHECK8-NEXT: [[I6:%.*]] = alloca i32, align 4
7330 // CHECK8-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i8, align 1
7331 // CHECK8-NEXT: [[_TMP16:%.*]] = alloca i32, align 4
7332 // CHECK8-NEXT: [[DOTOMP_LB17:%.*]] = alloca i32, align 4
7333 // CHECK8-NEXT: [[DOTOMP_UB18:%.*]] = alloca i32, align 4
7334 // CHECK8-NEXT: [[DOTOMP_IV19:%.*]] = alloca i32, align 4
7335 // CHECK8-NEXT: [[I20:%.*]] = alloca i32, align 4
7336 // CHECK8-NEXT: store i32 0, i32* [[RETVAL]], align 4
7337 // CHECK8-NEXT: store i32 0, i32* [[DOTOMP_LB]], align 4
7338 // CHECK8-NEXT: store i32 99, i32* [[DOTOMP_UB]], align 4
7339 // CHECK8-NEXT: [[TMP0:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4
7340 // CHECK8-NEXT: store i32 [[TMP0]], i32* [[DOTOMP_IV]], align 4
7341 // CHECK8-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
7342 // CHECK8: omp.inner.for.cond:
7343 // CHECK8-NEXT: [[TMP1:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !9
7344 // CHECK8-NEXT: [[TMP2:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4, !llvm.access.group !9
7345 // CHECK8-NEXT: [[CMP:%.*]] = icmp sle i32 [[TMP1]], [[TMP2]]
7346 // CHECK8-NEXT: br i1 [[CMP]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
7347 // CHECK8: omp.inner.for.body:
7348 // CHECK8-NEXT: [[TMP3:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !9
7349 // CHECK8-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP3]], 1
7350 // CHECK8-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
7351 // CHECK8-NEXT: store i32 [[ADD]], i32* [[I]], align 4, !llvm.access.group !9
7352 // CHECK8-NEXT: call void @_Z3fn4v(), !llvm.access.group !9
7353 // CHECK8-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
7354 // CHECK8: omp.body.continue:
7355 // CHECK8-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
7356 // CHECK8: omp.inner.for.inc:
7357 // CHECK8-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !9
7358 // CHECK8-NEXT: [[ADD1:%.*]] = add nsw i32 [[TMP4]], 1
7359 // CHECK8-NEXT: store i32 [[ADD1]], i32* [[DOTOMP_IV]], align 4, !llvm.access.group !9
7360 // CHECK8-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP10:![0-9]+]]
7361 // CHECK8: omp.inner.for.end:
7362 // CHECK8-NEXT: store i32 100, i32* [[I]], align 4
7363 // CHECK8-NEXT: store i32 0, i32* [[DOTOMP_LB3]], align 4
7364 // CHECK8-NEXT: store i32 99, i32* [[DOTOMP_UB4]], align 4
7365 // CHECK8-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_LB3]], align 4
7366 // CHECK8-NEXT: store i32 [[TMP5]], i32* [[DOTOMP_IV5]], align 4
7367 // CHECK8-NEXT: br label [[OMP_INNER_FOR_COND7:%.*]]
7368 // CHECK8: omp.inner.for.cond7:
7369 // CHECK8-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_IV5]], align 4
7370 // CHECK8-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_UB4]], align 4
7371 // CHECK8-NEXT: [[CMP8:%.*]] = icmp sle i32 [[TMP6]], [[TMP7]]
7372 // CHECK8-NEXT: br i1 [[CMP8]], label [[OMP_INNER_FOR_BODY9:%.*]], label [[OMP_INNER_FOR_END15:%.*]]
7373 // CHECK8: omp.inner.for.body9:
7374 // CHECK8-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_IV5]], align 4
7375 // CHECK8-NEXT: [[MUL10:%.*]] = mul nsw i32 [[TMP8]], 1
7376 // CHECK8-NEXT: [[ADD11:%.*]] = add nsw i32 0, [[MUL10]]
7377 // CHECK8-NEXT: store i32 [[ADD11]], i32* [[I6]], align 4
7378 // CHECK8-NEXT: call void @_Z3fn5v()
7379 // CHECK8-NEXT: br label [[OMP_BODY_CONTINUE12:%.*]]
7380 // CHECK8: omp.body.continue12:
7381 // CHECK8-NEXT: br label [[OMP_INNER_FOR_INC13:%.*]]
7382 // CHECK8: omp.inner.for.inc13:
7383 // CHECK8-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_IV5]], align 4
7384 // CHECK8-NEXT: [[ADD14:%.*]] = add nsw i32 [[TMP9]], 1
7385 // CHECK8-NEXT: store i32 [[ADD14]], i32* [[DOTOMP_IV5]], align 4
7386 // CHECK8-NEXT: br label [[OMP_INNER_FOR_COND7]], !llvm.loop [[LOOP12:![0-9]+]]
7387 // CHECK8: omp.inner.for.end15:
7388 // CHECK8-NEXT: store i32 100, i32* [[I6]], align 4
7389 // CHECK8-NEXT: [[TMP10:%.*]] = load i32, i32* @Arg, align 4
7390 // CHECK8-NEXT: [[TOBOOL:%.*]] = icmp ne i32 [[TMP10]], 0
7391 // CHECK8-NEXT: [[FROMBOOL:%.*]] = zext i1 [[TOBOOL]] to i8
7392 // CHECK8-NEXT: store i8 [[FROMBOOL]], i8* [[DOTCAPTURE_EXPR_]], align 1
7393 // CHECK8-NEXT: store i32 0, i32* [[DOTOMP_LB17]], align 4
7394 // CHECK8-NEXT: store i32 99, i32* [[DOTOMP_UB18]], align 4
7395 // CHECK8-NEXT: [[TMP11:%.*]] = load i32, i32* [[DOTOMP_LB17]], align 4
7396 // CHECK8-NEXT: store i32 [[TMP11]], i32* [[DOTOMP_IV19]], align 4
7397 // CHECK8-NEXT: [[TMP12:%.*]] = load i8, i8* [[DOTCAPTURE_EXPR_]], align 1
7398 // CHECK8-NEXT: [[TOBOOL21:%.*]] = trunc i8 [[TMP12]] to i1
7399 // CHECK8-NEXT: br i1 [[TOBOOL21]], label [[OMP_IF_THEN:%.*]], label [[OMP_IF_ELSE:%.*]]
7400 // CHECK8: omp_if.then:
7401 // CHECK8-NEXT: br label [[OMP_INNER_FOR_COND22:%.*]]
7402 // CHECK8: omp.inner.for.cond22:
7403 // CHECK8-NEXT: [[TMP13:%.*]] = load i32, i32* [[DOTOMP_IV19]], align 4, !llvm.access.group !14
7404 // CHECK8-NEXT: [[TMP14:%.*]] = load i32, i32* [[DOTOMP_UB18]], align 4, !llvm.access.group !14
7405 // CHECK8-NEXT: [[CMP23:%.*]] = icmp sle i32 [[TMP13]], [[TMP14]]
7406 // CHECK8-NEXT: br i1 [[CMP23]], label [[OMP_INNER_FOR_BODY24:%.*]], label [[OMP_INNER_FOR_END30:%.*]]
7407 // CHECK8: omp.inner.for.body24:
7408 // CHECK8-NEXT: [[TMP15:%.*]] = load i32, i32* [[DOTOMP_IV19]], align 4, !llvm.access.group !14
7409 // CHECK8-NEXT: [[MUL25:%.*]] = mul nsw i32 [[TMP15]], 1
7410 // CHECK8-NEXT: [[ADD26:%.*]] = add nsw i32 0, [[MUL25]]
7411 // CHECK8-NEXT: store i32 [[ADD26]], i32* [[I20]], align 4, !llvm.access.group !14
7412 // CHECK8-NEXT: call void @_Z3fn6v(), !llvm.access.group !14
7413 // CHECK8-NEXT: br label [[OMP_BODY_CONTINUE27:%.*]]
7414 // CHECK8: omp.body.continue27:
7415 // CHECK8-NEXT: br label [[OMP_INNER_FOR_INC28:%.*]]
7416 // CHECK8: omp.inner.for.inc28:
7417 // CHECK8-NEXT: [[TMP16:%.*]] = load i32, i32* [[DOTOMP_IV19]], align 4, !llvm.access.group !14
7418 // CHECK8-NEXT: [[ADD29:%.*]] = add nsw i32 [[TMP16]], 1
7419 // CHECK8-NEXT: store i32 [[ADD29]], i32* [[DOTOMP_IV19]], align 4, !llvm.access.group !14
7420 // CHECK8-NEXT: br label [[OMP_INNER_FOR_COND22]], !llvm.loop [[LOOP15:![0-9]+]]
7421 // CHECK8: omp.inner.for.end30:
7422 // CHECK8-NEXT: br label [[OMP_IF_END:%.*]]
7423 // CHECK8: omp_if.else:
7424 // CHECK8-NEXT: br label [[OMP_INNER_FOR_COND31:%.*]]
7425 // CHECK8: omp.inner.for.cond31:
7426 // CHECK8-NEXT: [[TMP17:%.*]] = load i32, i32* [[DOTOMP_IV19]], align 4
7427 // CHECK8-NEXT: [[TMP18:%.*]] = load i32, i32* [[DOTOMP_UB18]], align 4
7428 // CHECK8-NEXT: [[CMP32:%.*]] = icmp sle i32 [[TMP17]], [[TMP18]]
7429 // CHECK8-NEXT: br i1 [[CMP32]], label [[OMP_INNER_FOR_BODY33:%.*]], label [[OMP_INNER_FOR_END39:%.*]]
7430 // CHECK8: omp.inner.for.body33:
7431 // CHECK8-NEXT: [[TMP19:%.*]] = load i32, i32* [[DOTOMP_IV19]], align 4
7432 // CHECK8-NEXT: [[MUL34:%.*]] = mul nsw i32 [[TMP19]], 1
7433 // CHECK8-NEXT: [[ADD35:%.*]] = add nsw i32 0, [[MUL34]]
7434 // CHECK8-NEXT: store i32 [[ADD35]], i32* [[I20]], align 4
7435 // CHECK8-NEXT: call void @_Z3fn6v()
7436 // CHECK8-NEXT: br label [[OMP_BODY_CONTINUE36:%.*]]
7437 // CHECK8: omp.body.continue36:
7438 // CHECK8-NEXT: br label [[OMP_INNER_FOR_INC37:%.*]]
7439 // CHECK8: omp.inner.for.inc37:
7440 // CHECK8-NEXT: [[TMP20:%.*]] = load i32, i32* [[DOTOMP_IV19]], align 4
7441 // CHECK8-NEXT: [[ADD38:%.*]] = add nsw i32 [[TMP20]], 1
7442 // CHECK8-NEXT: store i32 [[ADD38]], i32* [[DOTOMP_IV19]], align 4
7443 // CHECK8-NEXT: br label [[OMP_INNER_FOR_COND31]], !llvm.loop [[LOOP17:![0-9]+]]
7444 // CHECK8: omp.inner.for.end39:
7445 // CHECK8-NEXT: br label [[OMP_IF_END]]
7446 // CHECK8: omp_if.end:
7447 // CHECK8-NEXT: store i32 100, i32* [[I20]], align 4
7448 // CHECK8-NEXT: [[TMP21:%.*]] = load i32, i32* @Arg, align 4
7449 // CHECK8-NEXT: [[CALL:%.*]] = call i32 @_Z5tmainIiEiT_(i32 [[TMP21]])
7450 // CHECK8-NEXT: ret i32 [[CALL]]
7451 //
7452 //
7453 // CHECK8-LABEL: define {{[^@]+}}@_Z5tmainIiEiT_
7454 // CHECK8-SAME: (i32 [[ARG:%.*]]) #[[ATTR0]] comdat {
7455 // CHECK8-NEXT: entry:
7456 // CHECK8-NEXT: [[ARG_ADDR:%.*]] = alloca i32, align 4
7457 // CHECK8-NEXT: [[TMP:%.*]] = alloca i32, align 4
7458 // CHECK8-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4
7459 // CHECK8-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4
7460 // CHECK8-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
7461 // CHECK8-NEXT: [[I:%.*]] = alloca i32, align 4
7462 // CHECK8-NEXT: [[_TMP2:%.*]] = alloca i32, align 4
7463 // CHECK8-NEXT: [[DOTOMP_LB3:%.*]] = alloca i32, align 4
7464 // CHECK8-NEXT: [[DOTOMP_UB4:%.*]] = alloca i32, align 4
7465 // CHECK8-NEXT: [[DOTOMP_IV5:%.*]] = alloca i32, align 4
7466 // CHECK8-NEXT: [[I6:%.*]] = alloca i32, align 4
7467 // CHECK8-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i8, align 1
7468 // CHECK8-NEXT: [[_TMP16:%.*]] = alloca i32, align 4
7469 // CHECK8-NEXT: [[DOTOMP_LB17:%.*]] = alloca i32, align 4
7470 // CHECK8-NEXT: [[DOTOMP_UB18:%.*]] = alloca i32, align 4
7471 // CHECK8-NEXT: [[DOTOMP_IV19:%.*]] = alloca i32, align 4
7472 // CHECK8-NEXT: [[I20:%.*]] = alloca i32, align 4
7473 // CHECK8-NEXT: store i32 [[ARG]], i32* [[ARG_ADDR]], align 4
7474 // CHECK8-NEXT: store i32 0, i32* [[DOTOMP_LB]], align 4
7475 // CHECK8-NEXT: store i32 99, i32* [[DOTOMP_UB]], align 4
7476 // CHECK8-NEXT: [[TMP0:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4
7477 // CHECK8-NEXT: store i32 [[TMP0]], i32* [[DOTOMP_IV]], align 4
7478 // CHECK8-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
7479 // CHECK8: omp.inner.for.cond:
7480 // CHECK8-NEXT: [[TMP1:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !18
7481 // CHECK8-NEXT: [[TMP2:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4, !llvm.access.group !18
7482 // CHECK8-NEXT: [[CMP:%.*]] = icmp sle i32 [[TMP1]], [[TMP2]]
7483 // CHECK8-NEXT: br i1 [[CMP]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
7484 // CHECK8: omp.inner.for.body:
7485 // CHECK8-NEXT: [[TMP3:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !18
7486 // CHECK8-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP3]], 1
7487 // CHECK8-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
7488 // CHECK8-NEXT: store i32 [[ADD]], i32* [[I]], align 4, !llvm.access.group !18
7489 // CHECK8-NEXT: call void @_Z3fn1v(), !llvm.access.group !18
7490 // CHECK8-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
7491 // CHECK8: omp.body.continue:
7492 // CHECK8-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
7493 // CHECK8: omp.inner.for.inc:
7494 // CHECK8-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !18
7495 // CHECK8-NEXT: [[ADD1:%.*]] = add nsw i32 [[TMP4]], 1
7496 // CHECK8-NEXT: store i32 [[ADD1]], i32* [[DOTOMP_IV]], align 4, !llvm.access.group !18
7497 // CHECK8-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP19:![0-9]+]]
7498 // CHECK8: omp.inner.for.end:
7499 // CHECK8-NEXT: store i32 100, i32* [[I]], align 4
7500 // CHECK8-NEXT: store i32 0, i32* [[DOTOMP_LB3]], align 4
7501 // CHECK8-NEXT: store i32 99, i32* [[DOTOMP_UB4]], align 4
7502 // CHECK8-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_LB3]], align 4
7503 // CHECK8-NEXT: store i32 [[TMP5]], i32* [[DOTOMP_IV5]], align 4
7504 // CHECK8-NEXT: br label [[OMP_INNER_FOR_COND7:%.*]]
7505 // CHECK8: omp.inner.for.cond7:
7506 // CHECK8-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_IV5]], align 4
7507 // CHECK8-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_UB4]], align 4
7508 // CHECK8-NEXT: [[CMP8:%.*]] = icmp sle i32 [[TMP6]], [[TMP7]]
7509 // CHECK8-NEXT: br i1 [[CMP8]], label [[OMP_INNER_FOR_BODY9:%.*]], label [[OMP_INNER_FOR_END15:%.*]]
7510 // CHECK8: omp.inner.for.body9:
7511 // CHECK8-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_IV5]], align 4
7512 // CHECK8-NEXT: [[MUL10:%.*]] = mul nsw i32 [[TMP8]], 1
7513 // CHECK8-NEXT: [[ADD11:%.*]] = add nsw i32 0, [[MUL10]]
7514 // CHECK8-NEXT: store i32 [[ADD11]], i32* [[I6]], align 4
7515 // CHECK8-NEXT: call void @_Z3fn2v()
7516 // CHECK8-NEXT: br label [[OMP_BODY_CONTINUE12:%.*]]
7517 // CHECK8: omp.body.continue12:
7518 // CHECK8-NEXT: br label [[OMP_INNER_FOR_INC13:%.*]]
7519 // CHECK8: omp.inner.for.inc13:
7520 // CHECK8-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_IV5]], align 4
7521 // CHECK8-NEXT: [[ADD14:%.*]] = add nsw i32 [[TMP9]], 1
7522 // CHECK8-NEXT: store i32 [[ADD14]], i32* [[DOTOMP_IV5]], align 4
7523 // CHECK8-NEXT: br label [[OMP_INNER_FOR_COND7]], !llvm.loop [[LOOP21:![0-9]+]]
7524 // CHECK8: omp.inner.for.end15:
7525 // CHECK8-NEXT: store i32 100, i32* [[I6]], align 4
7526 // CHECK8-NEXT: [[TMP10:%.*]] = load i32, i32* [[ARG_ADDR]], align 4
7527 // CHECK8-NEXT: [[TOBOOL:%.*]] = icmp ne i32 [[TMP10]], 0
7528 // CHECK8-NEXT: [[FROMBOOL:%.*]] = zext i1 [[TOBOOL]] to i8
7529 // CHECK8-NEXT: store i8 [[FROMBOOL]], i8* [[DOTCAPTURE_EXPR_]], align 1
7530 // CHECK8-NEXT: store i32 0, i32* [[DOTOMP_LB17]], align 4
7531 // CHECK8-NEXT: store i32 99, i32* [[DOTOMP_UB18]], align 4
7532 // CHECK8-NEXT: [[TMP11:%.*]] = load i32, i32* [[DOTOMP_LB17]], align 4
7533 // CHECK8-NEXT: store i32 [[TMP11]], i32* [[DOTOMP_IV19]], align 4
7534 // CHECK8-NEXT: br label [[OMP_INNER_FOR_COND21:%.*]]
7535 // CHECK8: omp.inner.for.cond21:
7536 // CHECK8-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTOMP_IV19]], align 4, !llvm.access.group !22
7537 // CHECK8-NEXT: [[TMP13:%.*]] = load i32, i32* [[DOTOMP_UB18]], align 4, !llvm.access.group !22
7538 // CHECK8-NEXT: [[CMP22:%.*]] = icmp sle i32 [[TMP12]], [[TMP13]]
7539 // CHECK8-NEXT: br i1 [[CMP22]], label [[OMP_INNER_FOR_BODY23:%.*]], label [[OMP_INNER_FOR_END29:%.*]]
7540 // CHECK8: omp.inner.for.body23:
7541 // CHECK8-NEXT: [[TMP14:%.*]] = load i32, i32* [[DOTOMP_IV19]], align 4, !llvm.access.group !22
7542 // CHECK8-NEXT: [[MUL24:%.*]] = mul nsw i32 [[TMP14]], 1
7543 // CHECK8-NEXT: [[ADD25:%.*]] = add nsw i32 0, [[MUL24]]
7544 // CHECK8-NEXT: store i32 [[ADD25]], i32* [[I20]], align 4, !llvm.access.group !22
7545 // CHECK8-NEXT: call void @_Z3fn3v(), !llvm.access.group !22
7546 // CHECK8-NEXT: br label [[OMP_BODY_CONTINUE26:%.*]]
7547 // CHECK8: omp.body.continue26:
7548 // CHECK8-NEXT: br label [[OMP_INNER_FOR_INC27:%.*]]
7549 // CHECK8: omp.inner.for.inc27:
7550 // CHECK8-NEXT: [[TMP15:%.*]] = load i32, i32* [[DOTOMP_IV19]], align 4, !llvm.access.group !22
7551 // CHECK8-NEXT: [[ADD28:%.*]] = add nsw i32 [[TMP15]], 1
7552 // CHECK8-NEXT: store i32 [[ADD28]], i32* [[DOTOMP_IV19]], align 4, !llvm.access.group !22
7553 // CHECK8-NEXT: br label [[OMP_INNER_FOR_COND21]], !llvm.loop [[LOOP23:![0-9]+]]
7554 // CHECK8: omp.inner.for.end29:
7555 // CHECK8-NEXT: store i32 100, i32* [[I20]], align 4
7556 // CHECK8-NEXT: ret i32 0
7557 //
7558 //
7559 // CHECK9-LABEL: define {{[^@]+}}@_Z9gtid_testv
7560 // CHECK9-SAME: () #[[ATTR0:[0-9]+]] {
7561 // CHECK9-NEXT: entry:
7562 // CHECK9-NEXT: [[TMP:%.*]] = alloca i32, align 4
7563 // CHECK9-NEXT: [[_TMP1:%.*]] = alloca i32, align 4
7564 // CHECK9-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3:[0-9]+]], i64 -1, i64 100)
7565 // CHECK9-NEXT: [[TMP0:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9gtid_testv_l43.region_id, i32 0, i8** null, i8** null, i64* null, i64* null, i8** null, i8** null, i32 0, i32 0)
7566 // CHECK9-NEXT: [[TMP1:%.*]] = icmp ne i32 [[TMP0]], 0
7567 // CHECK9-NEXT: br i1 [[TMP1]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
7568 // CHECK9: omp_offload.failed:
7569 // CHECK9-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9gtid_testv_l43() #[[ATTR2:[0-9]+]]
7570 // CHECK9-NEXT: br label [[OMP_OFFLOAD_CONT]]
7571 // CHECK9: omp_offload.cont:
7572 // CHECK9-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i64 100)
7573 // CHECK9-NEXT: [[TMP2:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9gtid_testv_l47.region_id, i32 0, i8** null, i8** null, i64* null, i64* null, i8** null, i8** null, i32 0, i32 1)
7574 // CHECK9-NEXT: [[TMP3:%.*]] = icmp ne i32 [[TMP2]], 0
7575 // CHECK9-NEXT: br i1 [[TMP3]], label [[OMP_OFFLOAD_FAILED2:%.*]], label [[OMP_OFFLOAD_CONT3:%.*]]
7576 // CHECK9: omp_offload.failed2:
7577 // CHECK9-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9gtid_testv_l47() #[[ATTR2]]
7578 // CHECK9-NEXT: br label [[OMP_OFFLOAD_CONT3]]
7579 // CHECK9: omp_offload.cont3:
7580 // CHECK9-NEXT: ret void
7581 //
7582 //
7583 // CHECK9-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9gtid_testv_l43
7584 // CHECK9-SAME: () #[[ATTR1:[0-9]+]] {
7585 // CHECK9-NEXT: entry:
7586 // CHECK9-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 0, void (i32*, i32*, ...)* bitcast (void (i32*, i32*)* @.omp_outlined. to void (i32*, i32*, ...)*))
7587 // CHECK9-NEXT: ret void
7588 //
7589 //
7590 // CHECK9-LABEL: define {{[^@]+}}@.omp_outlined.
7591 // CHECK9-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]]) #[[ATTR1]] {
7592 // CHECK9-NEXT: entry:
7593 // CHECK9-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
7594 // CHECK9-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
7595 // CHECK9-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
7596 // CHECK9-NEXT: [[TMP:%.*]] = alloca i32, align 4
7597 // CHECK9-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4
7598 // CHECK9-NEXT: [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4
7599 // CHECK9-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
7600 // CHECK9-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
7601 // CHECK9-NEXT: [[I:%.*]] = alloca i32, align 4
7602 // CHECK9-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
7603 // CHECK9-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
7604 // CHECK9-NEXT: store i32 0, i32* [[DOTOMP_COMB_LB]], align 4
7605 // CHECK9-NEXT: store i32 99, i32* [[DOTOMP_COMB_UB]], align 4
7606 // CHECK9-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4
7607 // CHECK9-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4
7608 // CHECK9-NEXT: [[TMP0:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
7609 // CHECK9-NEXT: [[TMP1:%.*]] = load i32, i32* [[TMP0]], align 4
7610 // CHECK9-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB1:[0-9]+]], i32 [[TMP1]], i32 92, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_COMB_LB]], i32* [[DOTOMP_COMB_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1)
7611 // CHECK9-NEXT: [[TMP2:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
7612 // CHECK9-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP2]], 99
7613 // CHECK9-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
7614 // CHECK9: cond.true:
7615 // CHECK9-NEXT: br label [[COND_END:%.*]]
7616 // CHECK9: cond.false:
7617 // CHECK9-NEXT: [[TMP3:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
7618 // CHECK9-NEXT: br label [[COND_END]]
7619 // CHECK9: cond.end:
7620 // CHECK9-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP3]], [[COND_FALSE]] ]
7621 // CHECK9-NEXT: store i32 [[COND]], i32* [[DOTOMP_COMB_UB]], align 4
7622 // CHECK9-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4
7623 // CHECK9-NEXT: store i32 [[TMP4]], i32* [[DOTOMP_IV]], align 4
7624 // CHECK9-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
7625 // CHECK9: omp.inner.for.cond:
7626 // CHECK9-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
7627 // CHECK9-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
7628 // CHECK9-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP5]], [[TMP6]]
7629 // CHECK9-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
7630 // CHECK9: omp.inner.for.body:
7631 // CHECK9-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4
7632 // CHECK9-NEXT: [[TMP8:%.*]] = zext i32 [[TMP7]] to i64
7633 // CHECK9-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
7634 // CHECK9-NEXT: [[TMP10:%.*]] = zext i32 [[TMP9]] to i64
7635 // CHECK9-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB3]], i32 2, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64, i64)* @.omp_outlined..1 to void (i32*, i32*, ...)*), i64 [[TMP8]], i64 [[TMP10]])
7636 // CHECK9-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
7637 // CHECK9: omp.inner.for.inc:
7638 // CHECK9-NEXT: [[TMP11:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
7639 // CHECK9-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4
7640 // CHECK9-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP11]], [[TMP12]]
7641 // CHECK9-NEXT: store i32 [[ADD]], i32* [[DOTOMP_IV]], align 4
7642 // CHECK9-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP14:![0-9]+]]
7643 // CHECK9: omp.inner.for.end:
7644 // CHECK9-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
7645 // CHECK9: omp.loop.exit:
7646 // CHECK9-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP1]])
7647 // CHECK9-NEXT: [[TMP13:%.*]] = load i32, i32* [[DOTOMP_IS_LAST]], align 4
7648 // CHECK9-NEXT: [[TMP14:%.*]] = icmp ne i32 [[TMP13]], 0
7649 // CHECK9-NEXT: br i1 [[TMP14]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
7650 // CHECK9: .omp.final.then:
7651 // CHECK9-NEXT: store i32 100, i32* [[I]], align 4
7652 // CHECK9-NEXT: br label [[DOTOMP_FINAL_DONE]]
7653 // CHECK9: .omp.final.done:
7654 // CHECK9-NEXT: ret void
7655 //
7656 //
7657 // CHECK9-LABEL: define {{[^@]+}}@.omp_outlined..1
7658 // CHECK9-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[DOTPREVIOUS_LB_:%.*]], i64 [[DOTPREVIOUS_UB_:%.*]]) #[[ATTR1]] {
7659 // CHECK9-NEXT: entry:
7660 // CHECK9-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
7661 // CHECK9-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
7662 // CHECK9-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i64, align 8
7663 // CHECK9-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i64, align 8
7664 // CHECK9-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
7665 // CHECK9-NEXT: [[TMP:%.*]] = alloca i32, align 4
7666 // CHECK9-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4
7667 // CHECK9-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4
7668 // CHECK9-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
7669 // CHECK9-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
7670 // CHECK9-NEXT: [[I:%.*]] = alloca i32, align 4
7671 // CHECK9-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
7672 // CHECK9-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
7673 // CHECK9-NEXT: store i64 [[DOTPREVIOUS_LB_]], i64* [[DOTPREVIOUS_LB__ADDR]], align 8
7674 // CHECK9-NEXT: store i64 [[DOTPREVIOUS_UB_]], i64* [[DOTPREVIOUS_UB__ADDR]], align 8
7675 // CHECK9-NEXT: store i32 0, i32* [[DOTOMP_LB]], align 4
7676 // CHECK9-NEXT: store i32 99, i32* [[DOTOMP_UB]], align 4
7677 // CHECK9-NEXT: [[TMP0:%.*]] = load i64, i64* [[DOTPREVIOUS_LB__ADDR]], align 8
7678 // CHECK9-NEXT: [[CONV:%.*]] = trunc i64 [[TMP0]] to i32
7679 // CHECK9-NEXT: [[TMP1:%.*]] = load i64, i64* [[DOTPREVIOUS_UB__ADDR]], align 8
7680 // CHECK9-NEXT: [[CONV1:%.*]] = trunc i64 [[TMP1]] to i32
7681 // CHECK9-NEXT: store i32 [[CONV]], i32* [[DOTOMP_LB]], align 4
7682 // CHECK9-NEXT: store i32 [[CONV1]], i32* [[DOTOMP_UB]], align 4
7683 // CHECK9-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4
7684 // CHECK9-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4
7685 // CHECK9-NEXT: [[TMP2:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
7686 // CHECK9-NEXT: [[TMP3:%.*]] = load i32, i32* [[TMP2]], align 4
7687 // CHECK9-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB2:[0-9]+]], i32 [[TMP3]], i32 34, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_LB]], i32* [[DOTOMP_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1)
7688 // CHECK9-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
7689 // CHECK9-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP4]], 99
7690 // CHECK9-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
7691 // CHECK9: cond.true:
7692 // CHECK9-NEXT: br label [[COND_END:%.*]]
7693 // CHECK9: cond.false:
7694 // CHECK9-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
7695 // CHECK9-NEXT: br label [[COND_END]]
7696 // CHECK9: cond.end:
7697 // CHECK9-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP5]], [[COND_FALSE]] ]
7698 // CHECK9-NEXT: store i32 [[COND]], i32* [[DOTOMP_UB]], align 4
7699 // CHECK9-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4
7700 // CHECK9-NEXT: store i32 [[TMP6]], i32* [[DOTOMP_IV]], align 4
7701 // CHECK9-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
7702 // CHECK9: omp.inner.for.cond:
7703 // CHECK9-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
7704 // CHECK9-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
7705 // CHECK9-NEXT: [[CMP2:%.*]] = icmp sle i32 [[TMP7]], [[TMP8]]
7706 // CHECK9-NEXT: br i1 [[CMP2]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
7707 // CHECK9: omp.inner.for.body:
7708 // CHECK9-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
7709 // CHECK9-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP9]], 1
7710 // CHECK9-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
7711 // CHECK9-NEXT: store i32 [[ADD]], i32* [[I]], align 4
7712 // CHECK9-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
7713 // CHECK9: omp.body.continue:
7714 // CHECK9-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
7715 // CHECK9: omp.inner.for.inc:
7716 // CHECK9-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
7717 // CHECK9-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP10]], 1
7718 // CHECK9-NEXT: store i32 [[ADD3]], i32* [[DOTOMP_IV]], align 4
7719 // CHECK9-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP16:![0-9]+]]
7720 // CHECK9: omp.inner.for.end:
7721 // CHECK9-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
7722 // CHECK9: omp.loop.exit:
7723 // CHECK9-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP3]])
7724 // CHECK9-NEXT: [[TMP11:%.*]] = load i32, i32* [[DOTOMP_IS_LAST]], align 4
7725 // CHECK9-NEXT: [[TMP12:%.*]] = icmp ne i32 [[TMP11]], 0
7726 // CHECK9-NEXT: br i1 [[TMP12]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
7727 // CHECK9: .omp.final.then:
7728 // CHECK9-NEXT: store i32 100, i32* [[I]], align 4
7729 // CHECK9-NEXT: br label [[DOTOMP_FINAL_DONE]]
7730 // CHECK9: .omp.final.done:
7731 // CHECK9-NEXT: ret void
7732 //
7733 //
7734 // CHECK9-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9gtid_testv_l47
7735 // CHECK9-SAME: () #[[ATTR1]] {
7736 // CHECK9-NEXT: entry:
7737 // CHECK9-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 0, void (i32*, i32*, ...)* bitcast (void (i32*, i32*)* @.omp_outlined..2 to void (i32*, i32*, ...)*))
7738 // CHECK9-NEXT: ret void
7739 //
7740 //
7741 // CHECK9-LABEL: define {{[^@]+}}@.omp_outlined..2
7742 // CHECK9-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]]) #[[ATTR1]] {
7743 // CHECK9-NEXT: entry:
7744 // CHECK9-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
7745 // CHECK9-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
7746 // CHECK9-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
7747 // CHECK9-NEXT: [[TMP:%.*]] = alloca i32, align 4
7748 // CHECK9-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4
7749 // CHECK9-NEXT: [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4
7750 // CHECK9-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
7751 // CHECK9-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
7752 // CHECK9-NEXT: [[I:%.*]] = alloca i32, align 4
7753 // CHECK9-NEXT: [[DOTBOUND_ZERO_ADDR:%.*]] = alloca i32, align 4
7754 // CHECK9-NEXT: store i32 0, i32* [[DOTBOUND_ZERO_ADDR]], align 4
7755 // CHECK9-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
7756 // CHECK9-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
7757 // CHECK9-NEXT: store i32 0, i32* [[DOTOMP_COMB_LB]], align 4
7758 // CHECK9-NEXT: store i32 99, i32* [[DOTOMP_COMB_UB]], align 4
7759 // CHECK9-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4
7760 // CHECK9-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4
7761 // CHECK9-NEXT: [[TMP0:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
7762 // CHECK9-NEXT: [[TMP1:%.*]] = load i32, i32* [[TMP0]], align 4
7763 // CHECK9-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB1]], i32 [[TMP1]], i32 92, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_COMB_LB]], i32* [[DOTOMP_COMB_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1)
7764 // CHECK9-NEXT: [[TMP2:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
7765 // CHECK9-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP2]], 99
7766 // CHECK9-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
7767 // CHECK9: cond.true:
7768 // CHECK9-NEXT: br label [[COND_END:%.*]]
7769 // CHECK9: cond.false:
7770 // CHECK9-NEXT: [[TMP3:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
7771 // CHECK9-NEXT: br label [[COND_END]]
7772 // CHECK9: cond.end:
7773 // CHECK9-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP3]], [[COND_FALSE]] ]
7774 // CHECK9-NEXT: store i32 [[COND]], i32* [[DOTOMP_COMB_UB]], align 4
7775 // CHECK9-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4
7776 // CHECK9-NEXT: store i32 [[TMP4]], i32* [[DOTOMP_IV]], align 4
7777 // CHECK9-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
7778 // CHECK9: omp.inner.for.cond:
7779 // CHECK9-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
7780 // CHECK9-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
7781 // CHECK9-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP5]], [[TMP6]]
7782 // CHECK9-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
7783 // CHECK9: omp.inner.for.body:
7784 // CHECK9-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4
7785 // CHECK9-NEXT: [[TMP8:%.*]] = zext i32 [[TMP7]] to i64
7786 // CHECK9-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
7787 // CHECK9-NEXT: [[TMP10:%.*]] = zext i32 [[TMP9]] to i64
7788 // CHECK9-NEXT: call void @__kmpc_serialized_parallel(%struct.ident_t* @[[GLOB3]], i32 [[TMP1]])
7789 // CHECK9-NEXT: [[TMP11:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
7790 // CHECK9-NEXT: call void @.omp_outlined..3(i32* [[TMP11]], i32* [[DOTBOUND_ZERO_ADDR]], i64 [[TMP8]], i64 [[TMP10]]) #[[ATTR2]]
7791 // CHECK9-NEXT: call void @__kmpc_end_serialized_parallel(%struct.ident_t* @[[GLOB3]], i32 [[TMP1]])
7792 // CHECK9-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
7793 // CHECK9: omp.inner.for.inc:
7794 // CHECK9-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
7795 // CHECK9-NEXT: [[TMP13:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4
7796 // CHECK9-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP12]], [[TMP13]]
7797 // CHECK9-NEXT: store i32 [[ADD]], i32* [[DOTOMP_IV]], align 4
7798 // CHECK9-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP19:![0-9]+]]
7799 // CHECK9: omp.inner.for.end:
7800 // CHECK9-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
7801 // CHECK9: omp.loop.exit:
7802 // CHECK9-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP1]])
7803 // CHECK9-NEXT: [[TMP14:%.*]] = load i32, i32* [[DOTOMP_IS_LAST]], align 4
7804 // CHECK9-NEXT: [[TMP15:%.*]] = icmp ne i32 [[TMP14]], 0
7805 // CHECK9-NEXT: br i1 [[TMP15]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
7806 // CHECK9: .omp.final.then:
7807 // CHECK9-NEXT: store i32 100, i32* [[I]], align 4
7808 // CHECK9-NEXT: br label [[DOTOMP_FINAL_DONE]]
7809 // CHECK9: .omp.final.done:
7810 // CHECK9-NEXT: ret void
7811 //
7812 //
7813 // CHECK9-LABEL: define {{[^@]+}}@.omp_outlined..3
7814 // CHECK9-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[DOTPREVIOUS_LB_:%.*]], i64 [[DOTPREVIOUS_UB_:%.*]]) #[[ATTR1]] {
7815 // CHECK9-NEXT: entry:
7816 // CHECK9-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
7817 // CHECK9-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
7818 // CHECK9-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i64, align 8
7819 // CHECK9-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i64, align 8
7820 // CHECK9-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
7821 // CHECK9-NEXT: [[TMP:%.*]] = alloca i32, align 4
7822 // CHECK9-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4
7823 // CHECK9-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4
7824 // CHECK9-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
7825 // CHECK9-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
7826 // CHECK9-NEXT: [[I:%.*]] = alloca i32, align 4
7827 // CHECK9-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
7828 // CHECK9-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
7829 // CHECK9-NEXT: store i64 [[DOTPREVIOUS_LB_]], i64* [[DOTPREVIOUS_LB__ADDR]], align 8
7830 // CHECK9-NEXT: store i64 [[DOTPREVIOUS_UB_]], i64* [[DOTPREVIOUS_UB__ADDR]], align 8
7831 // CHECK9-NEXT: store i32 0, i32* [[DOTOMP_LB]], align 4
7832 // CHECK9-NEXT: store i32 99, i32* [[DOTOMP_UB]], align 4
7833 // CHECK9-NEXT: [[TMP0:%.*]] = load i64, i64* [[DOTPREVIOUS_LB__ADDR]], align 8
7834 // CHECK9-NEXT: [[CONV:%.*]] = trunc i64 [[TMP0]] to i32
7835 // CHECK9-NEXT: [[TMP1:%.*]] = load i64, i64* [[DOTPREVIOUS_UB__ADDR]], align 8
7836 // CHECK9-NEXT: [[CONV1:%.*]] = trunc i64 [[TMP1]] to i32
7837 // CHECK9-NEXT: store i32 [[CONV]], i32* [[DOTOMP_LB]], align 4
7838 // CHECK9-NEXT: store i32 [[CONV1]], i32* [[DOTOMP_UB]], align 4
7839 // CHECK9-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4
7840 // CHECK9-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4
7841 // CHECK9-NEXT: [[TMP2:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
7842 // CHECK9-NEXT: [[TMP3:%.*]] = load i32, i32* [[TMP2]], align 4
7843 // CHECK9-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB2]], i32 [[TMP3]], i32 34, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_LB]], i32* [[DOTOMP_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1)
7844 // CHECK9-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
7845 // CHECK9-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP4]], 99
7846 // CHECK9-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
7847 // CHECK9: cond.true:
7848 // CHECK9-NEXT: br label [[COND_END:%.*]]
7849 // CHECK9: cond.false:
7850 // CHECK9-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
7851 // CHECK9-NEXT: br label [[COND_END]]
7852 // CHECK9: cond.end:
7853 // CHECK9-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP5]], [[COND_FALSE]] ]
7854 // CHECK9-NEXT: store i32 [[COND]], i32* [[DOTOMP_UB]], align 4
7855 // CHECK9-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4
7856 // CHECK9-NEXT: store i32 [[TMP6]], i32* [[DOTOMP_IV]], align 4
7857 // CHECK9-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
7858 // CHECK9: omp.inner.for.cond:
7859 // CHECK9-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
7860 // CHECK9-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
7861 // CHECK9-NEXT: [[CMP2:%.*]] = icmp sle i32 [[TMP7]], [[TMP8]]
7862 // CHECK9-NEXT: br i1 [[CMP2]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
7863 // CHECK9: omp.inner.for.body:
7864 // CHECK9-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
7865 // CHECK9-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP9]], 1
7866 // CHECK9-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
7867 // CHECK9-NEXT: store i32 [[ADD]], i32* [[I]], align 4
7868 // CHECK9-NEXT: call void @_Z9gtid_testv()
7869 // CHECK9-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
7870 // CHECK9: omp.body.continue:
7871 // CHECK9-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
7872 // CHECK9: omp.inner.for.inc:
7873 // CHECK9-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
7874 // CHECK9-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP10]], 1
7875 // CHECK9-NEXT: store i32 [[ADD3]], i32* [[DOTOMP_IV]], align 4
7876 // CHECK9-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP20:![0-9]+]]
7877 // CHECK9: omp.inner.for.end:
7878 // CHECK9-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
7879 // CHECK9: omp.loop.exit:
7880 // CHECK9-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP3]])
7881 // CHECK9-NEXT: [[TMP11:%.*]] = load i32, i32* [[DOTOMP_IS_LAST]], align 4
7882 // CHECK9-NEXT: [[TMP12:%.*]] = icmp ne i32 [[TMP11]], 0
7883 // CHECK9-NEXT: br i1 [[TMP12]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
7884 // CHECK9: .omp.final.then:
7885 // CHECK9-NEXT: store i32 100, i32* [[I]], align 4
7886 // CHECK9-NEXT: br label [[DOTOMP_FINAL_DONE]]
7887 // CHECK9: .omp.final.done:
7888 // CHECK9-NEXT: ret void
7889 //
7890 //
7891 // CHECK9-LABEL: define {{[^@]+}}@main
7892 // CHECK9-SAME: () #[[ATTR3:[0-9]+]] {
7893 // CHECK9-NEXT: entry:
7894 // CHECK9-NEXT: [[RETVAL:%.*]] = alloca i32, align 4
7895 // CHECK9-NEXT: [[TMP:%.*]] = alloca i32, align 4
7896 // CHECK9-NEXT: [[_TMP1:%.*]] = alloca i32, align 4
7897 // CHECK9-NEXT: [[ARG_CASTED:%.*]] = alloca i64, align 8
7898 // CHECK9-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [1 x i8*], align 8
7899 // CHECK9-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [1 x i8*], align 8
7900 // CHECK9-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [1 x i8*], align 8
7901 // CHECK9-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i8, align 1
7902 // CHECK9-NEXT: [[_TMP5:%.*]] = alloca i32, align 4
7903 // CHECK9-NEXT: store i32 0, i32* [[RETVAL]], align 4
7904 // CHECK9-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i64 100)
7905 // CHECK9-NEXT: [[TMP0:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l76.region_id, i32 0, i8** null, i8** null, i64* null, i64* null, i8** null, i8** null, i32 0, i32 0)
7906 // CHECK9-NEXT: [[TMP1:%.*]] = icmp ne i32 [[TMP0]], 0
7907 // CHECK9-NEXT: br i1 [[TMP1]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
7908 // CHECK9: omp_offload.failed:
7909 // CHECK9-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l76() #[[ATTR2]]
7910 // CHECK9-NEXT: br label [[OMP_OFFLOAD_CONT]]
7911 // CHECK9: omp_offload.cont:
7912 // CHECK9-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i64 100)
7913 // CHECK9-NEXT: [[TMP2:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l84.region_id, i32 0, i8** null, i8** null, i64* null, i64* null, i8** null, i8** null, i32 0, i32 1)
7914 // CHECK9-NEXT: [[TMP3:%.*]] = icmp ne i32 [[TMP2]], 0
7915 // CHECK9-NEXT: br i1 [[TMP3]], label [[OMP_OFFLOAD_FAILED2:%.*]], label [[OMP_OFFLOAD_CONT3:%.*]]
7916 // CHECK9: omp_offload.failed2:
7917 // CHECK9-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l84() #[[ATTR2]]
7918 // CHECK9-NEXT: br label [[OMP_OFFLOAD_CONT3]]
7919 // CHECK9: omp_offload.cont3:
7920 // CHECK9-NEXT: [[TMP4:%.*]] = load i32, i32* @Arg, align 4
7921 // CHECK9-NEXT: [[CONV:%.*]] = bitcast i64* [[ARG_CASTED]] to i32*
7922 // CHECK9-NEXT: store i32 [[TMP4]], i32* [[CONV]], align 4
7923 // CHECK9-NEXT: [[TMP5:%.*]] = load i64, i64* [[ARG_CASTED]], align 8
7924 // CHECK9-NEXT: [[TMP6:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
7925 // CHECK9-NEXT: [[TMP7:%.*]] = bitcast i8** [[TMP6]] to i64*
7926 // CHECK9-NEXT: store i64 [[TMP5]], i64* [[TMP7]], align 8
7927 // CHECK9-NEXT: [[TMP8:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
7928 // CHECK9-NEXT: [[TMP9:%.*]] = bitcast i8** [[TMP8]] to i64*
7929 // CHECK9-NEXT: store i64 [[TMP5]], i64* [[TMP9]], align 8
7930 // CHECK9-NEXT: [[TMP10:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0
7931 // CHECK9-NEXT: store i8* null, i8** [[TMP10]], align 8
7932 // CHECK9-NEXT: [[TMP11:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
7933 // CHECK9-NEXT: [[TMP12:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
7934 // CHECK9-NEXT: [[TMP13:%.*]] = load i32, i32* @Arg, align 4
7935 // CHECK9-NEXT: [[TOBOOL:%.*]] = icmp ne i32 [[TMP13]], 0
7936 // CHECK9-NEXT: [[FROMBOOL:%.*]] = zext i1 [[TOBOOL]] to i8
7937 // CHECK9-NEXT: store i8 [[FROMBOOL]], i8* [[DOTCAPTURE_EXPR_]], align 1
7938 // CHECK9-NEXT: [[TMP14:%.*]] = load i8, i8* [[DOTCAPTURE_EXPR_]], align 1
7939 // CHECK9-NEXT: [[TOBOOL4:%.*]] = trunc i8 [[TMP14]] to i1
7940 // CHECK9-NEXT: [[TMP15:%.*]] = select i1 [[TOBOOL4]], i32 0, i32 1
7941 // CHECK9-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i64 100)
7942 // CHECK9-NEXT: [[TMP16:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l92.region_id, i32 1, i8** [[TMP11]], i8** [[TMP12]], i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_sizes, i32 0, i32 0), i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_maptypes, i32 0, i32 0), i8** null, i8** null, i32 0, i32 [[TMP15]])
7943 // CHECK9-NEXT: [[TMP17:%.*]] = icmp ne i32 [[TMP16]], 0
7944 // CHECK9-NEXT: br i1 [[TMP17]], label [[OMP_OFFLOAD_FAILED6:%.*]], label [[OMP_OFFLOAD_CONT7:%.*]]
7945 // CHECK9: omp_offload.failed6:
7946 // CHECK9-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l92(i64 [[TMP5]]) #[[ATTR2]]
7947 // CHECK9-NEXT: br label [[OMP_OFFLOAD_CONT7]]
7948 // CHECK9: omp_offload.cont7:
7949 // CHECK9-NEXT: [[TMP18:%.*]] = load i32, i32* @Arg, align 4
7950 // CHECK9-NEXT: [[CALL:%.*]] = call i32 @_Z5tmainIiEiT_(i32 [[TMP18]])
7951 // CHECK9-NEXT: ret i32 [[CALL]]
7952 //
7953 //
7954 // CHECK9-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l76
7955 // CHECK9-SAME: () #[[ATTR1]] {
7956 // CHECK9-NEXT: entry:
7957 // CHECK9-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 0, void (i32*, i32*, ...)* bitcast (void (i32*, i32*)* @.omp_outlined..4 to void (i32*, i32*, ...)*))
7958 // CHECK9-NEXT: ret void
7959 //
7960 //
7961 // CHECK9-LABEL: define {{[^@]+}}@.omp_outlined..4
7962 // CHECK9-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]]) #[[ATTR1]] {
7963 // CHECK9-NEXT: entry:
7964 // CHECK9-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
7965 // CHECK9-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
7966 // CHECK9-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
7967 // CHECK9-NEXT: [[TMP:%.*]] = alloca i32, align 4
7968 // CHECK9-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4
7969 // CHECK9-NEXT: [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4
7970 // CHECK9-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
7971 // CHECK9-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
7972 // CHECK9-NEXT: [[I:%.*]] = alloca i32, align 4
7973 // CHECK9-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
7974 // CHECK9-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
7975 // CHECK9-NEXT: store i32 0, i32* [[DOTOMP_COMB_LB]], align 4
7976 // CHECK9-NEXT: store i32 99, i32* [[DOTOMP_COMB_UB]], align 4
7977 // CHECK9-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4
7978 // CHECK9-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4
7979 // CHECK9-NEXT: [[TMP0:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
7980 // CHECK9-NEXT: [[TMP1:%.*]] = load i32, i32* [[TMP0]], align 4
7981 // CHECK9-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB1]], i32 [[TMP1]], i32 92, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_COMB_LB]], i32* [[DOTOMP_COMB_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1)
7982 // CHECK9-NEXT: [[TMP2:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
7983 // CHECK9-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP2]], 99
7984 // CHECK9-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
7985 // CHECK9: cond.true:
7986 // CHECK9-NEXT: br label [[COND_END:%.*]]
7987 // CHECK9: cond.false:
7988 // CHECK9-NEXT: [[TMP3:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
7989 // CHECK9-NEXT: br label [[COND_END]]
7990 // CHECK9: cond.end:
7991 // CHECK9-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP3]], [[COND_FALSE]] ]
7992 // CHECK9-NEXT: store i32 [[COND]], i32* [[DOTOMP_COMB_UB]], align 4
7993 // CHECK9-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4
7994 // CHECK9-NEXT: store i32 [[TMP4]], i32* [[DOTOMP_IV]], align 4
7995 // CHECK9-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
7996 // CHECK9: omp.inner.for.cond:
7997 // CHECK9-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
7998 // CHECK9-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
7999 // CHECK9-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP5]], [[TMP6]]
8000 // CHECK9-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
8001 // CHECK9: omp.inner.for.body:
8002 // CHECK9-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4
8003 // CHECK9-NEXT: [[TMP8:%.*]] = zext i32 [[TMP7]] to i64
8004 // CHECK9-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
8005 // CHECK9-NEXT: [[TMP10:%.*]] = zext i32 [[TMP9]] to i64
8006 // CHECK9-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB3]], i32 2, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64, i64)* @.omp_outlined..5 to void (i32*, i32*, ...)*), i64 [[TMP8]], i64 [[TMP10]])
8007 // CHECK9-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
8008 // CHECK9: omp.inner.for.inc:
8009 // CHECK9-NEXT: [[TMP11:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
8010 // CHECK9-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4
8011 // CHECK9-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP11]], [[TMP12]]
8012 // CHECK9-NEXT: store i32 [[ADD]], i32* [[DOTOMP_IV]], align 4
8013 // CHECK9-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP21:![0-9]+]]
8014 // CHECK9: omp.inner.for.end:
8015 // CHECK9-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
8016 // CHECK9: omp.loop.exit:
8017 // CHECK9-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP1]])
8018 // CHECK9-NEXT: [[TMP13:%.*]] = load i32, i32* [[DOTOMP_IS_LAST]], align 4
8019 // CHECK9-NEXT: [[TMP14:%.*]] = icmp ne i32 [[TMP13]], 0
8020 // CHECK9-NEXT: br i1 [[TMP14]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
8021 // CHECK9: .omp.final.then:
8022 // CHECK9-NEXT: store i32 100, i32* [[I]], align 4
8023 // CHECK9-NEXT: br label [[DOTOMP_FINAL_DONE]]
8024 // CHECK9: .omp.final.done:
8025 // CHECK9-NEXT: ret void
8026 //
8027 //
8028 // CHECK9-LABEL: define {{[^@]+}}@.omp_outlined..5
8029 // CHECK9-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[DOTPREVIOUS_LB_:%.*]], i64 [[DOTPREVIOUS_UB_:%.*]]) #[[ATTR1]] {
8030 // CHECK9-NEXT: entry:
8031 // CHECK9-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
8032 // CHECK9-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
8033 // CHECK9-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i64, align 8
8034 // CHECK9-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i64, align 8
8035 // CHECK9-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
8036 // CHECK9-NEXT: [[TMP:%.*]] = alloca i32, align 4
8037 // CHECK9-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4
8038 // CHECK9-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4
8039 // CHECK9-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
8040 // CHECK9-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
8041 // CHECK9-NEXT: [[I:%.*]] = alloca i32, align 4
8042 // CHECK9-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
8043 // CHECK9-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
8044 // CHECK9-NEXT: store i64 [[DOTPREVIOUS_LB_]], i64* [[DOTPREVIOUS_LB__ADDR]], align 8
8045 // CHECK9-NEXT: store i64 [[DOTPREVIOUS_UB_]], i64* [[DOTPREVIOUS_UB__ADDR]], align 8
8046 // CHECK9-NEXT: store i32 0, i32* [[DOTOMP_LB]], align 4
8047 // CHECK9-NEXT: store i32 99, i32* [[DOTOMP_UB]], align 4
8048 // CHECK9-NEXT: [[TMP0:%.*]] = load i64, i64* [[DOTPREVIOUS_LB__ADDR]], align 8
8049 // CHECK9-NEXT: [[CONV:%.*]] = trunc i64 [[TMP0]] to i32
8050 // CHECK9-NEXT: [[TMP1:%.*]] = load i64, i64* [[DOTPREVIOUS_UB__ADDR]], align 8
8051 // CHECK9-NEXT: [[CONV1:%.*]] = trunc i64 [[TMP1]] to i32
8052 // CHECK9-NEXT: store i32 [[CONV]], i32* [[DOTOMP_LB]], align 4
8053 // CHECK9-NEXT: store i32 [[CONV1]], i32* [[DOTOMP_UB]], align 4
8054 // CHECK9-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4
8055 // CHECK9-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4
8056 // CHECK9-NEXT: [[TMP2:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
8057 // CHECK9-NEXT: [[TMP3:%.*]] = load i32, i32* [[TMP2]], align 4
8058 // CHECK9-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB2]], i32 [[TMP3]], i32 34, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_LB]], i32* [[DOTOMP_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1)
8059 // CHECK9-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
8060 // CHECK9-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP4]], 99
8061 // CHECK9-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
8062 // CHECK9: cond.true:
8063 // CHECK9-NEXT: br label [[COND_END:%.*]]
8064 // CHECK9: cond.false:
8065 // CHECK9-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
8066 // CHECK9-NEXT: br label [[COND_END]]
8067 // CHECK9: cond.end:
8068 // CHECK9-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP5]], [[COND_FALSE]] ]
8069 // CHECK9-NEXT: store i32 [[COND]], i32* [[DOTOMP_UB]], align 4
8070 // CHECK9-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4
8071 // CHECK9-NEXT: store i32 [[TMP6]], i32* [[DOTOMP_IV]], align 4
8072 // CHECK9-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
8073 // CHECK9: omp.inner.for.cond:
8074 // CHECK9-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
8075 // CHECK9-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
8076 // CHECK9-NEXT: [[CMP2:%.*]] = icmp sle i32 [[TMP7]], [[TMP8]]
8077 // CHECK9-NEXT: br i1 [[CMP2]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
8078 // CHECK9: omp.inner.for.body:
8079 // CHECK9-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
8080 // CHECK9-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP9]], 1
8081 // CHECK9-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
8082 // CHECK9-NEXT: store i32 [[ADD]], i32* [[I]], align 4
8083 // CHECK9-NEXT: call void @_Z3fn4v()
8084 // CHECK9-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
8085 // CHECK9: omp.body.continue:
8086 // CHECK9-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
8087 // CHECK9: omp.inner.for.inc:
8088 // CHECK9-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
8089 // CHECK9-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP10]], 1
8090 // CHECK9-NEXT: store i32 [[ADD3]], i32* [[DOTOMP_IV]], align 4
8091 // CHECK9-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP22:![0-9]+]]
8092 // CHECK9: omp.inner.for.end:
8093 // CHECK9-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
8094 // CHECK9: omp.loop.exit:
8095 // CHECK9-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP3]])
8096 // CHECK9-NEXT: [[TMP11:%.*]] = load i32, i32* [[DOTOMP_IS_LAST]], align 4
8097 // CHECK9-NEXT: [[TMP12:%.*]] = icmp ne i32 [[TMP11]], 0
8098 // CHECK9-NEXT: br i1 [[TMP12]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
8099 // CHECK9: .omp.final.then:
8100 // CHECK9-NEXT: store i32 100, i32* [[I]], align 4
8101 // CHECK9-NEXT: br label [[DOTOMP_FINAL_DONE]]
8102 // CHECK9: .omp.final.done:
8103 // CHECK9-NEXT: ret void
8104 //
8105 //
8106 // CHECK9-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l84
8107 // CHECK9-SAME: () #[[ATTR1]] {
8108 // CHECK9-NEXT: entry:
8109 // CHECK9-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 0, void (i32*, i32*, ...)* bitcast (void (i32*, i32*)* @.omp_outlined..6 to void (i32*, i32*, ...)*))
8110 // CHECK9-NEXT: ret void
8111 //
8112 //
8113 // CHECK9-LABEL: define {{[^@]+}}@.omp_outlined..6
8114 // CHECK9-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]]) #[[ATTR1]] {
8115 // CHECK9-NEXT: entry:
8116 // CHECK9-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
8117 // CHECK9-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
8118 // CHECK9-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
8119 // CHECK9-NEXT: [[TMP:%.*]] = alloca i32, align 4
8120 // CHECK9-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4
8121 // CHECK9-NEXT: [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4
8122 // CHECK9-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
8123 // CHECK9-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
8124 // CHECK9-NEXT: [[I:%.*]] = alloca i32, align 4
8125 // CHECK9-NEXT: [[DOTBOUND_ZERO_ADDR:%.*]] = alloca i32, align 4
8126 // CHECK9-NEXT: store i32 0, i32* [[DOTBOUND_ZERO_ADDR]], align 4
8127 // CHECK9-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
8128 // CHECK9-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
8129 // CHECK9-NEXT: store i32 0, i32* [[DOTOMP_COMB_LB]], align 4
8130 // CHECK9-NEXT: store i32 99, i32* [[DOTOMP_COMB_UB]], align 4
8131 // CHECK9-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4
8132 // CHECK9-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4
8133 // CHECK9-NEXT: [[TMP0:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
8134 // CHECK9-NEXT: [[TMP1:%.*]] = load i32, i32* [[TMP0]], align 4
8135 // CHECK9-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB1]], i32 [[TMP1]], i32 92, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_COMB_LB]], i32* [[DOTOMP_COMB_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1)
8136 // CHECK9-NEXT: [[TMP2:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
8137 // CHECK9-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP2]], 99
8138 // CHECK9-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
8139 // CHECK9: cond.true:
8140 // CHECK9-NEXT: br label [[COND_END:%.*]]
8141 // CHECK9: cond.false:
8142 // CHECK9-NEXT: [[TMP3:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
8143 // CHECK9-NEXT: br label [[COND_END]]
8144 // CHECK9: cond.end:
8145 // CHECK9-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP3]], [[COND_FALSE]] ]
8146 // CHECK9-NEXT: store i32 [[COND]], i32* [[DOTOMP_COMB_UB]], align 4
8147 // CHECK9-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4
8148 // CHECK9-NEXT: store i32 [[TMP4]], i32* [[DOTOMP_IV]], align 4
8149 // CHECK9-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
8150 // CHECK9: omp.inner.for.cond:
8151 // CHECK9-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
8152 // CHECK9-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
8153 // CHECK9-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP5]], [[TMP6]]
8154 // CHECK9-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
8155 // CHECK9: omp.inner.for.body:
8156 // CHECK9-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4
8157 // CHECK9-NEXT: [[TMP8:%.*]] = zext i32 [[TMP7]] to i64
8158 // CHECK9-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
8159 // CHECK9-NEXT: [[TMP10:%.*]] = zext i32 [[TMP9]] to i64
8160 // CHECK9-NEXT: call void @__kmpc_serialized_parallel(%struct.ident_t* @[[GLOB3]], i32 [[TMP1]])
8161 // CHECK9-NEXT: [[TMP11:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
8162 // CHECK9-NEXT: call void @.omp_outlined..7(i32* [[TMP11]], i32* [[DOTBOUND_ZERO_ADDR]], i64 [[TMP8]], i64 [[TMP10]]) #[[ATTR2]]
8163 // CHECK9-NEXT: call void @__kmpc_end_serialized_parallel(%struct.ident_t* @[[GLOB3]], i32 [[TMP1]])
8164 // CHECK9-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
8165 // CHECK9: omp.inner.for.inc:
8166 // CHECK9-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
8167 // CHECK9-NEXT: [[TMP13:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4
8168 // CHECK9-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP12]], [[TMP13]]
8169 // CHECK9-NEXT: store i32 [[ADD]], i32* [[DOTOMP_IV]], align 4
8170 // CHECK9-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP23:![0-9]+]]
8171 // CHECK9: omp.inner.for.end:
8172 // CHECK9-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
8173 // CHECK9: omp.loop.exit:
8174 // CHECK9-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP1]])
8175 // CHECK9-NEXT: [[TMP14:%.*]] = load i32, i32* [[DOTOMP_IS_LAST]], align 4
8176 // CHECK9-NEXT: [[TMP15:%.*]] = icmp ne i32 [[TMP14]], 0
8177 // CHECK9-NEXT: br i1 [[TMP15]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
8178 // CHECK9: .omp.final.then:
8179 // CHECK9-NEXT: store i32 100, i32* [[I]], align 4
8180 // CHECK9-NEXT: br label [[DOTOMP_FINAL_DONE]]
8181 // CHECK9: .omp.final.done:
8182 // CHECK9-NEXT: ret void
8183 //
8184 //
8185 // CHECK9-LABEL: define {{[^@]+}}@.omp_outlined..7
8186 // CHECK9-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[DOTPREVIOUS_LB_:%.*]], i64 [[DOTPREVIOUS_UB_:%.*]]) #[[ATTR1]] {
8187 // CHECK9-NEXT: entry:
8188 // CHECK9-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
8189 // CHECK9-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
8190 // CHECK9-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i64, align 8
8191 // CHECK9-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i64, align 8
8192 // CHECK9-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
8193 // CHECK9-NEXT: [[TMP:%.*]] = alloca i32, align 4
8194 // CHECK9-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4
8195 // CHECK9-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4
8196 // CHECK9-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
8197 // CHECK9-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
8198 // CHECK9-NEXT: [[I:%.*]] = alloca i32, align 4
8199 // CHECK9-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
8200 // CHECK9-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
8201 // CHECK9-NEXT: store i64 [[DOTPREVIOUS_LB_]], i64* [[DOTPREVIOUS_LB__ADDR]], align 8
8202 // CHECK9-NEXT: store i64 [[DOTPREVIOUS_UB_]], i64* [[DOTPREVIOUS_UB__ADDR]], align 8
8203 // CHECK9-NEXT: store i32 0, i32* [[DOTOMP_LB]], align 4
8204 // CHECK9-NEXT: store i32 99, i32* [[DOTOMP_UB]], align 4
8205 // CHECK9-NEXT: [[TMP0:%.*]] = load i64, i64* [[DOTPREVIOUS_LB__ADDR]], align 8
8206 // CHECK9-NEXT: [[CONV:%.*]] = trunc i64 [[TMP0]] to i32
8207 // CHECK9-NEXT: [[TMP1:%.*]] = load i64, i64* [[DOTPREVIOUS_UB__ADDR]], align 8
8208 // CHECK9-NEXT: [[CONV1:%.*]] = trunc i64 [[TMP1]] to i32
8209 // CHECK9-NEXT: store i32 [[CONV]], i32* [[DOTOMP_LB]], align 4
8210 // CHECK9-NEXT: store i32 [[CONV1]], i32* [[DOTOMP_UB]], align 4
8211 // CHECK9-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4
8212 // CHECK9-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4
8213 // CHECK9-NEXT: [[TMP2:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
8214 // CHECK9-NEXT: [[TMP3:%.*]] = load i32, i32* [[TMP2]], align 4
8215 // CHECK9-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB2]], i32 [[TMP3]], i32 34, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_LB]], i32* [[DOTOMP_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1)
8216 // CHECK9-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
8217 // CHECK9-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP4]], 99
8218 // CHECK9-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
8219 // CHECK9: cond.true:
8220 // CHECK9-NEXT: br label [[COND_END:%.*]]
8221 // CHECK9: cond.false:
8222 // CHECK9-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
8223 // CHECK9-NEXT: br label [[COND_END]]
8224 // CHECK9: cond.end:
8225 // CHECK9-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP5]], [[COND_FALSE]] ]
8226 // CHECK9-NEXT: store i32 [[COND]], i32* [[DOTOMP_UB]], align 4
8227 // CHECK9-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4
8228 // CHECK9-NEXT: store i32 [[TMP6]], i32* [[DOTOMP_IV]], align 4
8229 // CHECK9-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
8230 // CHECK9: omp.inner.for.cond:
8231 // CHECK9-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
8232 // CHECK9-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
8233 // CHECK9-NEXT: [[CMP2:%.*]] = icmp sle i32 [[TMP7]], [[TMP8]]
8234 // CHECK9-NEXT: br i1 [[CMP2]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
8235 // CHECK9: omp.inner.for.body:
8236 // CHECK9-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
8237 // CHECK9-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP9]], 1
8238 // CHECK9-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
8239 // CHECK9-NEXT: store i32 [[ADD]], i32* [[I]], align 4
8240 // CHECK9-NEXT: call void @_Z3fn5v()
8241 // CHECK9-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
8242 // CHECK9: omp.body.continue:
8243 // CHECK9-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
8244 // CHECK9: omp.inner.for.inc:
8245 // CHECK9-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
8246 // CHECK9-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP10]], 1
8247 // CHECK9-NEXT: store i32 [[ADD3]], i32* [[DOTOMP_IV]], align 4
8248 // CHECK9-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP24:![0-9]+]]
8249 // CHECK9: omp.inner.for.end:
8250 // CHECK9-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
8251 // CHECK9: omp.loop.exit:
8252 // CHECK9-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP3]])
8253 // CHECK9-NEXT: [[TMP11:%.*]] = load i32, i32* [[DOTOMP_IS_LAST]], align 4
8254 // CHECK9-NEXT: [[TMP12:%.*]] = icmp ne i32 [[TMP11]], 0
8255 // CHECK9-NEXT: br i1 [[TMP12]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
8256 // CHECK9: .omp.final.then:
8257 // CHECK9-NEXT: store i32 100, i32* [[I]], align 4
8258 // CHECK9-NEXT: br label [[DOTOMP_FINAL_DONE]]
8259 // CHECK9: .omp.final.done:
8260 // CHECK9-NEXT: ret void
8261 //
8262 //
8263 // CHECK9-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l92
8264 // CHECK9-SAME: (i64 [[ARG:%.*]]) #[[ATTR1]] {
8265 // CHECK9-NEXT: entry:
8266 // CHECK9-NEXT: [[ARG_ADDR:%.*]] = alloca i64, align 8
8267 // CHECK9-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i8, align 1
8268 // CHECK9-NEXT: [[DOTCAPTURE_EXPR__CASTED:%.*]] = alloca i64, align 8
8269 // CHECK9-NEXT: store i64 [[ARG]], i64* [[ARG_ADDR]], align 8
8270 // CHECK9-NEXT: [[CONV:%.*]] = bitcast i64* [[ARG_ADDR]] to i32*
8271 // CHECK9-NEXT: [[TMP0:%.*]] = load i32, i32* [[CONV]], align 8
8272 // CHECK9-NEXT: [[TOBOOL:%.*]] = icmp ne i32 [[TMP0]], 0
8273 // CHECK9-NEXT: [[FROMBOOL:%.*]] = zext i1 [[TOBOOL]] to i8
8274 // CHECK9-NEXT: store i8 [[FROMBOOL]], i8* [[DOTCAPTURE_EXPR_]], align 1
8275 // CHECK9-NEXT: [[TMP1:%.*]] = load i8, i8* [[DOTCAPTURE_EXPR_]], align 1
8276 // CHECK9-NEXT: [[TOBOOL1:%.*]] = trunc i8 [[TMP1]] to i1
8277 // CHECK9-NEXT: [[CONV2:%.*]] = bitcast i64* [[DOTCAPTURE_EXPR__CASTED]] to i8*
8278 // CHECK9-NEXT: [[FROMBOOL3:%.*]] = zext i1 [[TOBOOL1]] to i8
8279 // CHECK9-NEXT: store i8 [[FROMBOOL3]], i8* [[CONV2]], align 1
8280 // CHECK9-NEXT: [[TMP2:%.*]] = load i64, i64* [[DOTCAPTURE_EXPR__CASTED]], align 8
8281 // CHECK9-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64)* @.omp_outlined..8 to void (i32*, i32*, ...)*), i64 [[TMP2]])
8282 // CHECK9-NEXT: ret void
8283 //
8284 //
8285 // CHECK9-LABEL: define {{[^@]+}}@.omp_outlined..8
8286 // CHECK9-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[DOTCAPTURE_EXPR_:%.*]]) #[[ATTR1]] {
8287 // CHECK9-NEXT: entry:
8288 // CHECK9-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
8289 // CHECK9-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
8290 // CHECK9-NEXT: [[DOTCAPTURE_EXPR__ADDR:%.*]] = alloca i64, align 8
8291 // CHECK9-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
8292 // CHECK9-NEXT: [[TMP:%.*]] = alloca i32, align 4
8293 // CHECK9-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4
8294 // CHECK9-NEXT: [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4
8295 // CHECK9-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
8296 // CHECK9-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
8297 // CHECK9-NEXT: [[I:%.*]] = alloca i32, align 4
8298 // CHECK9-NEXT: [[DOTBOUND_ZERO_ADDR:%.*]] = alloca i32, align 4
8299 // CHECK9-NEXT: store i32 0, i32* [[DOTBOUND_ZERO_ADDR]], align 4
8300 // CHECK9-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
8301 // CHECK9-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
8302 // CHECK9-NEXT: store i64 [[DOTCAPTURE_EXPR_]], i64* [[DOTCAPTURE_EXPR__ADDR]], align 8
8303 // CHECK9-NEXT: [[CONV:%.*]] = bitcast i64* [[DOTCAPTURE_EXPR__ADDR]] to i8*
8304 // CHECK9-NEXT: store i32 0, i32* [[DOTOMP_COMB_LB]], align 4
8305 // CHECK9-NEXT: store i32 99, i32* [[DOTOMP_COMB_UB]], align 4
8306 // CHECK9-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4
8307 // CHECK9-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4
8308 // CHECK9-NEXT: [[TMP0:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
8309 // CHECK9-NEXT: [[TMP1:%.*]] = load i32, i32* [[TMP0]], align 4
8310 // CHECK9-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB1]], i32 [[TMP1]], i32 92, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_COMB_LB]], i32* [[DOTOMP_COMB_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1)
8311 // CHECK9-NEXT: [[TMP2:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
8312 // CHECK9-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP2]], 99
8313 // CHECK9-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
8314 // CHECK9: cond.true:
8315 // CHECK9-NEXT: br label [[COND_END:%.*]]
8316 // CHECK9: cond.false:
8317 // CHECK9-NEXT: [[TMP3:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
8318 // CHECK9-NEXT: br label [[COND_END]]
8319 // CHECK9: cond.end:
8320 // CHECK9-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP3]], [[COND_FALSE]] ]
8321 // CHECK9-NEXT: store i32 [[COND]], i32* [[DOTOMP_COMB_UB]], align 4
8322 // CHECK9-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4
8323 // CHECK9-NEXT: store i32 [[TMP4]], i32* [[DOTOMP_IV]], align 4
8324 // CHECK9-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
8325 // CHECK9: omp.inner.for.cond:
8326 // CHECK9-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
8327 // CHECK9-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
8328 // CHECK9-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP5]], [[TMP6]]
8329 // CHECK9-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
8330 // CHECK9: omp.inner.for.body:
8331 // CHECK9-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4
8332 // CHECK9-NEXT: [[TMP8:%.*]] = zext i32 [[TMP7]] to i64
8333 // CHECK9-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
8334 // CHECK9-NEXT: [[TMP10:%.*]] = zext i32 [[TMP9]] to i64
8335 // CHECK9-NEXT: [[TMP11:%.*]] = load i8, i8* [[CONV]], align 8
8336 // CHECK9-NEXT: [[TOBOOL:%.*]] = trunc i8 [[TMP11]] to i1
8337 // CHECK9-NEXT: br i1 [[TOBOOL]], label [[OMP_IF_THEN:%.*]], label [[OMP_IF_ELSE:%.*]]
8338 // CHECK9: omp_if.then:
8339 // CHECK9-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB3]], i32 2, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64, i64)* @.omp_outlined..9 to void (i32*, i32*, ...)*), i64 [[TMP8]], i64 [[TMP10]])
8340 // CHECK9-NEXT: br label [[OMP_IF_END:%.*]]
8341 // CHECK9: omp_if.else:
8342 // CHECK9-NEXT: call void @__kmpc_serialized_parallel(%struct.ident_t* @[[GLOB3]], i32 [[TMP1]])
8343 // CHECK9-NEXT: [[TMP12:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
8344 // CHECK9-NEXT: call void @.omp_outlined..9(i32* [[TMP12]], i32* [[DOTBOUND_ZERO_ADDR]], i64 [[TMP8]], i64 [[TMP10]]) #[[ATTR2]]
8345 // CHECK9-NEXT: call void @__kmpc_end_serialized_parallel(%struct.ident_t* @[[GLOB3]], i32 [[TMP1]])
8346 // CHECK9-NEXT: br label [[OMP_IF_END]]
8347 // CHECK9: omp_if.end:
8348 // CHECK9-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
8349 // CHECK9: omp.inner.for.inc:
8350 // CHECK9-NEXT: [[TMP13:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
8351 // CHECK9-NEXT: [[TMP14:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4
8352 // CHECK9-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP13]], [[TMP14]]
8353 // CHECK9-NEXT: store i32 [[ADD]], i32* [[DOTOMP_IV]], align 4
8354 // CHECK9-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP25:![0-9]+]]
8355 // CHECK9: omp.inner.for.end:
8356 // CHECK9-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
8357 // CHECK9: omp.loop.exit:
8358 // CHECK9-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP1]])
8359 // CHECK9-NEXT: [[TMP15:%.*]] = load i32, i32* [[DOTOMP_IS_LAST]], align 4
8360 // CHECK9-NEXT: [[TMP16:%.*]] = icmp ne i32 [[TMP15]], 0
8361 // CHECK9-NEXT: br i1 [[TMP16]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
8362 // CHECK9: .omp.final.then:
8363 // CHECK9-NEXT: store i32 100, i32* [[I]], align 4
8364 // CHECK9-NEXT: br label [[DOTOMP_FINAL_DONE]]
8365 // CHECK9: .omp.final.done:
8366 // CHECK9-NEXT: ret void
8367 //
8368 //
8369 // CHECK9-LABEL: define {{[^@]+}}@.omp_outlined..9
8370 // CHECK9-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[DOTPREVIOUS_LB_:%.*]], i64 [[DOTPREVIOUS_UB_:%.*]]) #[[ATTR1]] {
8371 // CHECK9-NEXT: entry:
8372 // CHECK9-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
8373 // CHECK9-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
8374 // CHECK9-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i64, align 8
8375 // CHECK9-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i64, align 8
8376 // CHECK9-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
8377 // CHECK9-NEXT: [[TMP:%.*]] = alloca i32, align 4
8378 // CHECK9-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4
8379 // CHECK9-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4
8380 // CHECK9-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
8381 // CHECK9-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
8382 // CHECK9-NEXT: [[I:%.*]] = alloca i32, align 4
8383 // CHECK9-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
8384 // CHECK9-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
8385 // CHECK9-NEXT: store i64 [[DOTPREVIOUS_LB_]], i64* [[DOTPREVIOUS_LB__ADDR]], align 8
8386 // CHECK9-NEXT: store i64 [[DOTPREVIOUS_UB_]], i64* [[DOTPREVIOUS_UB__ADDR]], align 8
8387 // CHECK9-NEXT: store i32 0, i32* [[DOTOMP_LB]], align 4
8388 // CHECK9-NEXT: store i32 99, i32* [[DOTOMP_UB]], align 4
8389 // CHECK9-NEXT: [[TMP0:%.*]] = load i64, i64* [[DOTPREVIOUS_LB__ADDR]], align 8
8390 // CHECK9-NEXT: [[CONV:%.*]] = trunc i64 [[TMP0]] to i32
8391 // CHECK9-NEXT: [[TMP1:%.*]] = load i64, i64* [[DOTPREVIOUS_UB__ADDR]], align 8
8392 // CHECK9-NEXT: [[CONV1:%.*]] = trunc i64 [[TMP1]] to i32
8393 // CHECK9-NEXT: store i32 [[CONV]], i32* [[DOTOMP_LB]], align 4
8394 // CHECK9-NEXT: store i32 [[CONV1]], i32* [[DOTOMP_UB]], align 4
8395 // CHECK9-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4
8396 // CHECK9-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4
8397 // CHECK9-NEXT: [[TMP2:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
8398 // CHECK9-NEXT: [[TMP3:%.*]] = load i32, i32* [[TMP2]], align 4
8399 // CHECK9-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB2]], i32 [[TMP3]], i32 34, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_LB]], i32* [[DOTOMP_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1)
8400 // CHECK9-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
8401 // CHECK9-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP4]], 99
8402 // CHECK9-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
8403 // CHECK9: cond.true:
8404 // CHECK9-NEXT: br label [[COND_END:%.*]]
8405 // CHECK9: cond.false:
8406 // CHECK9-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
8407 // CHECK9-NEXT: br label [[COND_END]]
8408 // CHECK9: cond.end:
8409 // CHECK9-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP5]], [[COND_FALSE]] ]
8410 // CHECK9-NEXT: store i32 [[COND]], i32* [[DOTOMP_UB]], align 4
8411 // CHECK9-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4
8412 // CHECK9-NEXT: store i32 [[TMP6]], i32* [[DOTOMP_IV]], align 4
8413 // CHECK9-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
8414 // CHECK9: omp.inner.for.cond:
8415 // CHECK9-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
8416 // CHECK9-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
8417 // CHECK9-NEXT: [[CMP2:%.*]] = icmp sle i32 [[TMP7]], [[TMP8]]
8418 // CHECK9-NEXT: br i1 [[CMP2]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
8419 // CHECK9: omp.inner.for.body:
8420 // CHECK9-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
8421 // CHECK9-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP9]], 1
8422 // CHECK9-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
8423 // CHECK9-NEXT: store i32 [[ADD]], i32* [[I]], align 4
8424 // CHECK9-NEXT: call void @_Z3fn6v()
8425 // CHECK9-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
8426 // CHECK9: omp.body.continue:
8427 // CHECK9-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
8428 // CHECK9: omp.inner.for.inc:
8429 // CHECK9-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
8430 // CHECK9-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP10]], 1
8431 // CHECK9-NEXT: store i32 [[ADD3]], i32* [[DOTOMP_IV]], align 4
8432 // CHECK9-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP26:![0-9]+]]
8433 // CHECK9: omp.inner.for.end:
8434 // CHECK9-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
8435 // CHECK9: omp.loop.exit:
8436 // CHECK9-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP3]])
8437 // CHECK9-NEXT: [[TMP11:%.*]] = load i32, i32* [[DOTOMP_IS_LAST]], align 4
8438 // CHECK9-NEXT: [[TMP12:%.*]] = icmp ne i32 [[TMP11]], 0
8439 // CHECK9-NEXT: br i1 [[TMP12]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
8440 // CHECK9: .omp.final.then:
8441 // CHECK9-NEXT: store i32 100, i32* [[I]], align 4
8442 // CHECK9-NEXT: br label [[DOTOMP_FINAL_DONE]]
8443 // CHECK9: .omp.final.done:
8444 // CHECK9-NEXT: ret void
8445 //
8446 //
8447 // CHECK9-LABEL: define {{[^@]+}}@_Z5tmainIiEiT_
8448 // CHECK9-SAME: (i32 [[ARG:%.*]]) #[[ATTR0]] comdat {
8449 // CHECK9-NEXT: entry:
8450 // CHECK9-NEXT: [[ARG_ADDR:%.*]] = alloca i32, align 4
8451 // CHECK9-NEXT: [[TMP:%.*]] = alloca i32, align 4
8452 // CHECK9-NEXT: [[_TMP1:%.*]] = alloca i32, align 4
8453 // CHECK9-NEXT: [[ARG_CASTED:%.*]] = alloca i64, align 8
8454 // CHECK9-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [1 x i8*], align 8
8455 // CHECK9-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [1 x i8*], align 8
8456 // CHECK9-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [1 x i8*], align 8
8457 // CHECK9-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i8, align 1
8458 // CHECK9-NEXT: [[_TMP5:%.*]] = alloca i32, align 4
8459 // CHECK9-NEXT: store i32 [[ARG]], i32* [[ARG_ADDR]], align 4
8460 // CHECK9-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i64 100)
8461 // CHECK9-NEXT: [[TMP0:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiEiT__l57.region_id, i32 0, i8** null, i8** null, i64* null, i64* null, i8** null, i8** null, i32 0, i32 0)
8462 // CHECK9-NEXT: [[TMP1:%.*]] = icmp ne i32 [[TMP0]], 0
8463 // CHECK9-NEXT: br i1 [[TMP1]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
8464 // CHECK9: omp_offload.failed:
8465 // CHECK9-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiEiT__l57() #[[ATTR2]]
8466 // CHECK9-NEXT: br label [[OMP_OFFLOAD_CONT]]
8467 // CHECK9: omp_offload.cont:
8468 // CHECK9-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i64 100)
8469 // CHECK9-NEXT: [[TMP2:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiEiT__l62.region_id, i32 0, i8** null, i8** null, i64* null, i64* null, i8** null, i8** null, i32 0, i32 1)
8470 // CHECK9-NEXT: [[TMP3:%.*]] = icmp ne i32 [[TMP2]], 0
8471 // CHECK9-NEXT: br i1 [[TMP3]], label [[OMP_OFFLOAD_FAILED2:%.*]], label [[OMP_OFFLOAD_CONT3:%.*]]
8472 // CHECK9: omp_offload.failed2:
8473 // CHECK9-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiEiT__l62() #[[ATTR2]]
8474 // CHECK9-NEXT: br label [[OMP_OFFLOAD_CONT3]]
8475 // CHECK9: omp_offload.cont3:
8476 // CHECK9-NEXT: [[TMP4:%.*]] = load i32, i32* [[ARG_ADDR]], align 4
8477 // CHECK9-NEXT: [[CONV:%.*]] = bitcast i64* [[ARG_CASTED]] to i32*
8478 // CHECK9-NEXT: store i32 [[TMP4]], i32* [[CONV]], align 4
8479 // CHECK9-NEXT: [[TMP5:%.*]] = load i64, i64* [[ARG_CASTED]], align 8
8480 // CHECK9-NEXT: [[TMP6:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
8481 // CHECK9-NEXT: [[TMP7:%.*]] = bitcast i8** [[TMP6]] to i64*
8482 // CHECK9-NEXT: store i64 [[TMP5]], i64* [[TMP7]], align 8
8483 // CHECK9-NEXT: [[TMP8:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
8484 // CHECK9-NEXT: [[TMP9:%.*]] = bitcast i8** [[TMP8]] to i64*
8485 // CHECK9-NEXT: store i64 [[TMP5]], i64* [[TMP9]], align 8
8486 // CHECK9-NEXT: [[TMP10:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0
8487 // CHECK9-NEXT: store i8* null, i8** [[TMP10]], align 8
8488 // CHECK9-NEXT: [[TMP11:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
8489 // CHECK9-NEXT: [[TMP12:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
8490 // CHECK9-NEXT: [[TMP13:%.*]] = load i32, i32* [[ARG_ADDR]], align 4
8491 // CHECK9-NEXT: [[TOBOOL:%.*]] = icmp ne i32 [[TMP13]], 0
8492 // CHECK9-NEXT: [[FROMBOOL:%.*]] = zext i1 [[TOBOOL]] to i8
8493 // CHECK9-NEXT: store i8 [[FROMBOOL]], i8* [[DOTCAPTURE_EXPR_]], align 1
8494 // CHECK9-NEXT: [[TMP14:%.*]] = load i8, i8* [[DOTCAPTURE_EXPR_]], align 1
8495 // CHECK9-NEXT: [[TOBOOL4:%.*]] = trunc i8 [[TMP14]] to i1
8496 // CHECK9-NEXT: [[TMP15:%.*]] = select i1 [[TOBOOL4]], i32 0, i32 1
8497 // CHECK9-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i64 100)
8498 // CHECK9-NEXT: [[TMP16:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiEiT__l67.region_id, i32 1, i8** [[TMP11]], i8** [[TMP12]], i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_sizes.16, i32 0, i32 0), i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_maptypes.17, i32 0, i32 0), i8** null, i8** null, i32 0, i32 [[TMP15]])
8499 // CHECK9-NEXT: [[TMP17:%.*]] = icmp ne i32 [[TMP16]], 0
8500 // CHECK9-NEXT: br i1 [[TMP17]], label [[OMP_OFFLOAD_FAILED6:%.*]], label [[OMP_OFFLOAD_CONT7:%.*]]
8501 // CHECK9: omp_offload.failed6:
8502 // CHECK9-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiEiT__l67(i64 [[TMP5]]) #[[ATTR2]]
8503 // CHECK9-NEXT: br label [[OMP_OFFLOAD_CONT7]]
8504 // CHECK9: omp_offload.cont7:
8505 // CHECK9-NEXT: ret i32 0
8506 //
8507 //
8508 // CHECK9-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiEiT__l57
8509 // CHECK9-SAME: () #[[ATTR1]] {
8510 // CHECK9-NEXT: entry:
8511 // CHECK9-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 0, void (i32*, i32*, ...)* bitcast (void (i32*, i32*)* @.omp_outlined..10 to void (i32*, i32*, ...)*))
8512 // CHECK9-NEXT: ret void
8513 //
8514 //
8515 // CHECK9-LABEL: define {{[^@]+}}@.omp_outlined..10
8516 // CHECK9-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]]) #[[ATTR1]] {
8517 // CHECK9-NEXT: entry:
8518 // CHECK9-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
8519 // CHECK9-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
8520 // CHECK9-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
8521 // CHECK9-NEXT: [[TMP:%.*]] = alloca i32, align 4
8522 // CHECK9-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4
8523 // CHECK9-NEXT: [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4
8524 // CHECK9-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
8525 // CHECK9-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
8526 // CHECK9-NEXT: [[I:%.*]] = alloca i32, align 4
8527 // CHECK9-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
8528 // CHECK9-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
8529 // CHECK9-NEXT: store i32 0, i32* [[DOTOMP_COMB_LB]], align 4
8530 // CHECK9-NEXT: store i32 99, i32* [[DOTOMP_COMB_UB]], align 4
8531 // CHECK9-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4
8532 // CHECK9-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4
8533 // CHECK9-NEXT: [[TMP0:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
8534 // CHECK9-NEXT: [[TMP1:%.*]] = load i32, i32* [[TMP0]], align 4
8535 // CHECK9-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB1]], i32 [[TMP1]], i32 92, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_COMB_LB]], i32* [[DOTOMP_COMB_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1)
8536 // CHECK9-NEXT: [[TMP2:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
8537 // CHECK9-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP2]], 99
8538 // CHECK9-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
8539 // CHECK9: cond.true:
8540 // CHECK9-NEXT: br label [[COND_END:%.*]]
8541 // CHECK9: cond.false:
8542 // CHECK9-NEXT: [[TMP3:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
8543 // CHECK9-NEXT: br label [[COND_END]]
8544 // CHECK9: cond.end:
8545 // CHECK9-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP3]], [[COND_FALSE]] ]
8546 // CHECK9-NEXT: store i32 [[COND]], i32* [[DOTOMP_COMB_UB]], align 4
8547 // CHECK9-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4
8548 // CHECK9-NEXT: store i32 [[TMP4]], i32* [[DOTOMP_IV]], align 4
8549 // CHECK9-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
8550 // CHECK9: omp.inner.for.cond:
8551 // CHECK9-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
8552 // CHECK9-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
8553 // CHECK9-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP5]], [[TMP6]]
8554 // CHECK9-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
8555 // CHECK9: omp.inner.for.body:
8556 // CHECK9-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4
8557 // CHECK9-NEXT: [[TMP8:%.*]] = zext i32 [[TMP7]] to i64
8558 // CHECK9-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
8559 // CHECK9-NEXT: [[TMP10:%.*]] = zext i32 [[TMP9]] to i64
8560 // CHECK9-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB3]], i32 2, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64, i64)* @.omp_outlined..11 to void (i32*, i32*, ...)*), i64 [[TMP8]], i64 [[TMP10]])
8561 // CHECK9-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
8562 // CHECK9: omp.inner.for.inc:
8563 // CHECK9-NEXT: [[TMP11:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
8564 // CHECK9-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4
8565 // CHECK9-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP11]], [[TMP12]]
8566 // CHECK9-NEXT: store i32 [[ADD]], i32* [[DOTOMP_IV]], align 4
8567 // CHECK9-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP27:![0-9]+]]
8568 // CHECK9: omp.inner.for.end:
8569 // CHECK9-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
8570 // CHECK9: omp.loop.exit:
8571 // CHECK9-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP1]])
8572 // CHECK9-NEXT: [[TMP13:%.*]] = load i32, i32* [[DOTOMP_IS_LAST]], align 4
8573 // CHECK9-NEXT: [[TMP14:%.*]] = icmp ne i32 [[TMP13]], 0
8574 // CHECK9-NEXT: br i1 [[TMP14]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
8575 // CHECK9: .omp.final.then:
8576 // CHECK9-NEXT: store i32 100, i32* [[I]], align 4
8577 // CHECK9-NEXT: br label [[DOTOMP_FINAL_DONE]]
8578 // CHECK9: .omp.final.done:
8579 // CHECK9-NEXT: ret void
8580 //
8581 //
8582 // CHECK9-LABEL: define {{[^@]+}}@.omp_outlined..11
8583 // CHECK9-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[DOTPREVIOUS_LB_:%.*]], i64 [[DOTPREVIOUS_UB_:%.*]]) #[[ATTR1]] {
8584 // CHECK9-NEXT: entry:
8585 // CHECK9-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
8586 // CHECK9-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
8587 // CHECK9-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i64, align 8
8588 // CHECK9-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i64, align 8
8589 // CHECK9-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
8590 // CHECK9-NEXT: [[TMP:%.*]] = alloca i32, align 4
8591 // CHECK9-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4
8592 // CHECK9-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4
8593 // CHECK9-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
8594 // CHECK9-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
8595 // CHECK9-NEXT: [[I:%.*]] = alloca i32, align 4
8596 // CHECK9-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
8597 // CHECK9-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
8598 // CHECK9-NEXT: store i64 [[DOTPREVIOUS_LB_]], i64* [[DOTPREVIOUS_LB__ADDR]], align 8
8599 // CHECK9-NEXT: store i64 [[DOTPREVIOUS_UB_]], i64* [[DOTPREVIOUS_UB__ADDR]], align 8
8600 // CHECK9-NEXT: store i32 0, i32* [[DOTOMP_LB]], align 4
8601 // CHECK9-NEXT: store i32 99, i32* [[DOTOMP_UB]], align 4
8602 // CHECK9-NEXT: [[TMP0:%.*]] = load i64, i64* [[DOTPREVIOUS_LB__ADDR]], align 8
8603 // CHECK9-NEXT: [[CONV:%.*]] = trunc i64 [[TMP0]] to i32
8604 // CHECK9-NEXT: [[TMP1:%.*]] = load i64, i64* [[DOTPREVIOUS_UB__ADDR]], align 8
8605 // CHECK9-NEXT: [[CONV1:%.*]] = trunc i64 [[TMP1]] to i32
8606 // CHECK9-NEXT: store i32 [[CONV]], i32* [[DOTOMP_LB]], align 4
8607 // CHECK9-NEXT: store i32 [[CONV1]], i32* [[DOTOMP_UB]], align 4
8608 // CHECK9-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4
8609 // CHECK9-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4
8610 // CHECK9-NEXT: [[TMP2:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
8611 // CHECK9-NEXT: [[TMP3:%.*]] = load i32, i32* [[TMP2]], align 4
8612 // CHECK9-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB2]], i32 [[TMP3]], i32 34, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_LB]], i32* [[DOTOMP_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1)
8613 // CHECK9-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
8614 // CHECK9-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP4]], 99
8615 // CHECK9-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
8616 // CHECK9: cond.true:
8617 // CHECK9-NEXT: br label [[COND_END:%.*]]
8618 // CHECK9: cond.false:
8619 // CHECK9-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
8620 // CHECK9-NEXT: br label [[COND_END]]
8621 // CHECK9: cond.end:
8622 // CHECK9-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP5]], [[COND_FALSE]] ]
8623 // CHECK9-NEXT: store i32 [[COND]], i32* [[DOTOMP_UB]], align 4
8624 // CHECK9-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4
8625 // CHECK9-NEXT: store i32 [[TMP6]], i32* [[DOTOMP_IV]], align 4
8626 // CHECK9-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
8627 // CHECK9: omp.inner.for.cond:
8628 // CHECK9-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
8629 // CHECK9-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
8630 // CHECK9-NEXT: [[CMP2:%.*]] = icmp sle i32 [[TMP7]], [[TMP8]]
8631 // CHECK9-NEXT: br i1 [[CMP2]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
8632 // CHECK9: omp.inner.for.body:
8633 // CHECK9-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
8634 // CHECK9-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP9]], 1
8635 // CHECK9-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
8636 // CHECK9-NEXT: store i32 [[ADD]], i32* [[I]], align 4
8637 // CHECK9-NEXT: call void @_Z3fn1v()
8638 // CHECK9-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
8639 // CHECK9: omp.body.continue:
8640 // CHECK9-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
8641 // CHECK9: omp.inner.for.inc:
8642 // CHECK9-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
8643 // CHECK9-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP10]], 1
8644 // CHECK9-NEXT: store i32 [[ADD3]], i32* [[DOTOMP_IV]], align 4
8645 // CHECK9-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP28:![0-9]+]]
8646 // CHECK9: omp.inner.for.end:
8647 // CHECK9-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
8648 // CHECK9: omp.loop.exit:
8649 // CHECK9-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP3]])
8650 // CHECK9-NEXT: [[TMP11:%.*]] = load i32, i32* [[DOTOMP_IS_LAST]], align 4
8651 // CHECK9-NEXT: [[TMP12:%.*]] = icmp ne i32 [[TMP11]], 0
8652 // CHECK9-NEXT: br i1 [[TMP12]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
8653 // CHECK9: .omp.final.then:
8654 // CHECK9-NEXT: store i32 100, i32* [[I]], align 4
8655 // CHECK9-NEXT: br label [[DOTOMP_FINAL_DONE]]
8656 // CHECK9: .omp.final.done:
8657 // CHECK9-NEXT: ret void
8658 //
8659 //
8660 // CHECK9-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiEiT__l62
8661 // CHECK9-SAME: () #[[ATTR1]] {
8662 // CHECK9-NEXT: entry:
8663 // CHECK9-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 0, void (i32*, i32*, ...)* bitcast (void (i32*, i32*)* @.omp_outlined..12 to void (i32*, i32*, ...)*))
8664 // CHECK9-NEXT: ret void
8665 //
8666 //
8667 // CHECK9-LABEL: define {{[^@]+}}@.omp_outlined..12
8668 // CHECK9-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]]) #[[ATTR1]] {
8669 // CHECK9-NEXT: entry:
8670 // CHECK9-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
8671 // CHECK9-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
8672 // CHECK9-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
8673 // CHECK9-NEXT: [[TMP:%.*]] = alloca i32, align 4
8674 // CHECK9-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4
8675 // CHECK9-NEXT: [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4
8676 // CHECK9-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
8677 // CHECK9-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
8678 // CHECK9-NEXT: [[I:%.*]] = alloca i32, align 4
8679 // CHECK9-NEXT: [[DOTBOUND_ZERO_ADDR:%.*]] = alloca i32, align 4
8680 // CHECK9-NEXT: store i32 0, i32* [[DOTBOUND_ZERO_ADDR]], align 4
8681 // CHECK9-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
8682 // CHECK9-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
8683 // CHECK9-NEXT: store i32 0, i32* [[DOTOMP_COMB_LB]], align 4
8684 // CHECK9-NEXT: store i32 99, i32* [[DOTOMP_COMB_UB]], align 4
8685 // CHECK9-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4
8686 // CHECK9-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4
8687 // CHECK9-NEXT: [[TMP0:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
8688 // CHECK9-NEXT: [[TMP1:%.*]] = load i32, i32* [[TMP0]], align 4
8689 // CHECK9-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB1]], i32 [[TMP1]], i32 92, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_COMB_LB]], i32* [[DOTOMP_COMB_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1)
8690 // CHECK9-NEXT: [[TMP2:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
8691 // CHECK9-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP2]], 99
8692 // CHECK9-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
8693 // CHECK9: cond.true:
8694 // CHECK9-NEXT: br label [[COND_END:%.*]]
8695 // CHECK9: cond.false:
8696 // CHECK9-NEXT: [[TMP3:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
8697 // CHECK9-NEXT: br label [[COND_END]]
8698 // CHECK9: cond.end:
8699 // CHECK9-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP3]], [[COND_FALSE]] ]
8700 // CHECK9-NEXT: store i32 [[COND]], i32* [[DOTOMP_COMB_UB]], align 4
8701 // CHECK9-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4
8702 // CHECK9-NEXT: store i32 [[TMP4]], i32* [[DOTOMP_IV]], align 4
8703 // CHECK9-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
8704 // CHECK9: omp.inner.for.cond:
8705 // CHECK9-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
8706 // CHECK9-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
8707 // CHECK9-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP5]], [[TMP6]]
8708 // CHECK9-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
8709 // CHECK9: omp.inner.for.body:
8710 // CHECK9-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4
8711 // CHECK9-NEXT: [[TMP8:%.*]] = zext i32 [[TMP7]] to i64
8712 // CHECK9-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
8713 // CHECK9-NEXT: [[TMP10:%.*]] = zext i32 [[TMP9]] to i64
8714 // CHECK9-NEXT: call void @__kmpc_serialized_parallel(%struct.ident_t* @[[GLOB3]], i32 [[TMP1]])
8715 // CHECK9-NEXT: [[TMP11:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
8716 // CHECK9-NEXT: call void @.omp_outlined..13(i32* [[TMP11]], i32* [[DOTBOUND_ZERO_ADDR]], i64 [[TMP8]], i64 [[TMP10]]) #[[ATTR2]]
8717 // CHECK9-NEXT: call void @__kmpc_end_serialized_parallel(%struct.ident_t* @[[GLOB3]], i32 [[TMP1]])
8718 // CHECK9-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
8719 // CHECK9: omp.inner.for.inc:
8720 // CHECK9-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
8721 // CHECK9-NEXT: [[TMP13:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4
8722 // CHECK9-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP12]], [[TMP13]]
8723 // CHECK9-NEXT: store i32 [[ADD]], i32* [[DOTOMP_IV]], align 4
8724 // CHECK9-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP29:![0-9]+]]
8725 // CHECK9: omp.inner.for.end:
8726 // CHECK9-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
8727 // CHECK9: omp.loop.exit:
8728 // CHECK9-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP1]])
8729 // CHECK9-NEXT: [[TMP14:%.*]] = load i32, i32* [[DOTOMP_IS_LAST]], align 4
8730 // CHECK9-NEXT: [[TMP15:%.*]] = icmp ne i32 [[TMP14]], 0
8731 // CHECK9-NEXT: br i1 [[TMP15]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
8732 // CHECK9: .omp.final.then:
8733 // CHECK9-NEXT: store i32 100, i32* [[I]], align 4
8734 // CHECK9-NEXT: br label [[DOTOMP_FINAL_DONE]]
8735 // CHECK9: .omp.final.done:
8736 // CHECK9-NEXT: ret void
8737 //
8738 //
8739 // CHECK9-LABEL: define {{[^@]+}}@.omp_outlined..13
8740 // CHECK9-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[DOTPREVIOUS_LB_:%.*]], i64 [[DOTPREVIOUS_UB_:%.*]]) #[[ATTR1]] {
8741 // CHECK9-NEXT: entry:
8742 // CHECK9-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
8743 // CHECK9-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
8744 // CHECK9-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i64, align 8
8745 // CHECK9-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i64, align 8
8746 // CHECK9-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
8747 // CHECK9-NEXT: [[TMP:%.*]] = alloca i32, align 4
8748 // CHECK9-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4
8749 // CHECK9-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4
8750 // CHECK9-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
8751 // CHECK9-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
8752 // CHECK9-NEXT: [[I:%.*]] = alloca i32, align 4
8753 // CHECK9-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
8754 // CHECK9-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
8755 // CHECK9-NEXT: store i64 [[DOTPREVIOUS_LB_]], i64* [[DOTPREVIOUS_LB__ADDR]], align 8
8756 // CHECK9-NEXT: store i64 [[DOTPREVIOUS_UB_]], i64* [[DOTPREVIOUS_UB__ADDR]], align 8
8757 // CHECK9-NEXT: store i32 0, i32* [[DOTOMP_LB]], align 4
8758 // CHECK9-NEXT: store i32 99, i32* [[DOTOMP_UB]], align 4
8759 // CHECK9-NEXT: [[TMP0:%.*]] = load i64, i64* [[DOTPREVIOUS_LB__ADDR]], align 8
8760 // CHECK9-NEXT: [[CONV:%.*]] = trunc i64 [[TMP0]] to i32
8761 // CHECK9-NEXT: [[TMP1:%.*]] = load i64, i64* [[DOTPREVIOUS_UB__ADDR]], align 8
8762 // CHECK9-NEXT: [[CONV1:%.*]] = trunc i64 [[TMP1]] to i32
8763 // CHECK9-NEXT: store i32 [[CONV]], i32* [[DOTOMP_LB]], align 4
8764 // CHECK9-NEXT: store i32 [[CONV1]], i32* [[DOTOMP_UB]], align 4
8765 // CHECK9-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4
8766 // CHECK9-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4
8767 // CHECK9-NEXT: [[TMP2:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
8768 // CHECK9-NEXT: [[TMP3:%.*]] = load i32, i32* [[TMP2]], align 4
8769 // CHECK9-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB2]], i32 [[TMP3]], i32 34, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_LB]], i32* [[DOTOMP_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1)
8770 // CHECK9-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
8771 // CHECK9-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP4]], 99
8772 // CHECK9-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
8773 // CHECK9: cond.true:
8774 // CHECK9-NEXT: br label [[COND_END:%.*]]
8775 // CHECK9: cond.false:
8776 // CHECK9-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
8777 // CHECK9-NEXT: br label [[COND_END]]
8778 // CHECK9: cond.end:
8779 // CHECK9-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP5]], [[COND_FALSE]] ]
8780 // CHECK9-NEXT: store i32 [[COND]], i32* [[DOTOMP_UB]], align 4
8781 // CHECK9-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4
8782 // CHECK9-NEXT: store i32 [[TMP6]], i32* [[DOTOMP_IV]], align 4
8783 // CHECK9-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
8784 // CHECK9: omp.inner.for.cond:
8785 // CHECK9-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
8786 // CHECK9-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
8787 // CHECK9-NEXT: [[CMP2:%.*]] = icmp sle i32 [[TMP7]], [[TMP8]]
8788 // CHECK9-NEXT: br i1 [[CMP2]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
8789 // CHECK9: omp.inner.for.body:
8790 // CHECK9-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
8791 // CHECK9-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP9]], 1
8792 // CHECK9-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
8793 // CHECK9-NEXT: store i32 [[ADD]], i32* [[I]], align 4
8794 // CHECK9-NEXT: call void @_Z3fn2v()
8795 // CHECK9-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
8796 // CHECK9: omp.body.continue:
8797 // CHECK9-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
8798 // CHECK9: omp.inner.for.inc:
8799 // CHECK9-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
8800 // CHECK9-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP10]], 1
8801 // CHECK9-NEXT: store i32 [[ADD3]], i32* [[DOTOMP_IV]], align 4
8802 // CHECK9-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP30:![0-9]+]]
8803 // CHECK9: omp.inner.for.end:
8804 // CHECK9-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
8805 // CHECK9: omp.loop.exit:
8806 // CHECK9-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP3]])
8807 // CHECK9-NEXT: [[TMP11:%.*]] = load i32, i32* [[DOTOMP_IS_LAST]], align 4
8808 // CHECK9-NEXT: [[TMP12:%.*]] = icmp ne i32 [[TMP11]], 0
8809 // CHECK9-NEXT: br i1 [[TMP12]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
8810 // CHECK9: .omp.final.then:
8811 // CHECK9-NEXT: store i32 100, i32* [[I]], align 4
8812 // CHECK9-NEXT: br label [[DOTOMP_FINAL_DONE]]
8813 // CHECK9: .omp.final.done:
8814 // CHECK9-NEXT: ret void
8815 //
8816 //
8817 // CHECK9-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiEiT__l67
8818 // CHECK9-SAME: (i64 [[ARG:%.*]]) #[[ATTR1]] {
8819 // CHECK9-NEXT: entry:
8820 // CHECK9-NEXT: [[ARG_ADDR:%.*]] = alloca i64, align 8
8821 // CHECK9-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i8, align 1
8822 // CHECK9-NEXT: [[DOTCAPTURE_EXPR__CASTED:%.*]] = alloca i64, align 8
8823 // CHECK9-NEXT: store i64 [[ARG]], i64* [[ARG_ADDR]], align 8
8824 // CHECK9-NEXT: [[CONV:%.*]] = bitcast i64* [[ARG_ADDR]] to i32*
8825 // CHECK9-NEXT: [[TMP0:%.*]] = load i32, i32* [[CONV]], align 8
8826 // CHECK9-NEXT: [[TOBOOL:%.*]] = icmp ne i32 [[TMP0]], 0
8827 // CHECK9-NEXT: [[FROMBOOL:%.*]] = zext i1 [[TOBOOL]] to i8
8828 // CHECK9-NEXT: store i8 [[FROMBOOL]], i8* [[DOTCAPTURE_EXPR_]], align 1
8829 // CHECK9-NEXT: [[TMP1:%.*]] = load i8, i8* [[DOTCAPTURE_EXPR_]], align 1
8830 // CHECK9-NEXT: [[TOBOOL1:%.*]] = trunc i8 [[TMP1]] to i1
8831 // CHECK9-NEXT: [[CONV2:%.*]] = bitcast i64* [[DOTCAPTURE_EXPR__CASTED]] to i8*
8832 // CHECK9-NEXT: [[FROMBOOL3:%.*]] = zext i1 [[TOBOOL1]] to i8
8833 // CHECK9-NEXT: store i8 [[FROMBOOL3]], i8* [[CONV2]], align 1
8834 // CHECK9-NEXT: [[TMP2:%.*]] = load i64, i64* [[DOTCAPTURE_EXPR__CASTED]], align 8
8835 // CHECK9-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64)* @.omp_outlined..14 to void (i32*, i32*, ...)*), i64 [[TMP2]])
8836 // CHECK9-NEXT: ret void
8837 //
8838 //
8839 // CHECK9-LABEL: define {{[^@]+}}@.omp_outlined..14
8840 // CHECK9-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[DOTCAPTURE_EXPR_:%.*]]) #[[ATTR1]] {
8841 // CHECK9-NEXT: entry:
8842 // CHECK9-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
8843 // CHECK9-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
8844 // CHECK9-NEXT: [[DOTCAPTURE_EXPR__ADDR:%.*]] = alloca i64, align 8
8845 // CHECK9-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
8846 // CHECK9-NEXT: [[TMP:%.*]] = alloca i32, align 4
8847 // CHECK9-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4
8848 // CHECK9-NEXT: [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4
8849 // CHECK9-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
8850 // CHECK9-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
8851 // CHECK9-NEXT: [[I:%.*]] = alloca i32, align 4
8852 // CHECK9-NEXT: [[DOTBOUND_ZERO_ADDR:%.*]] = alloca i32, align 4
8853 // CHECK9-NEXT: store i32 0, i32* [[DOTBOUND_ZERO_ADDR]], align 4
8854 // CHECK9-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
8855 // CHECK9-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
8856 // CHECK9-NEXT: store i64 [[DOTCAPTURE_EXPR_]], i64* [[DOTCAPTURE_EXPR__ADDR]], align 8
8857 // CHECK9-NEXT: [[CONV:%.*]] = bitcast i64* [[DOTCAPTURE_EXPR__ADDR]] to i8*
8858 // CHECK9-NEXT: store i32 0, i32* [[DOTOMP_COMB_LB]], align 4
8859 // CHECK9-NEXT: store i32 99, i32* [[DOTOMP_COMB_UB]], align 4
8860 // CHECK9-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4
8861 // CHECK9-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4
8862 // CHECK9-NEXT: [[TMP0:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
8863 // CHECK9-NEXT: [[TMP1:%.*]] = load i32, i32* [[TMP0]], align 4
8864 // CHECK9-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB1]], i32 [[TMP1]], i32 92, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_COMB_LB]], i32* [[DOTOMP_COMB_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1)
8865 // CHECK9-NEXT: [[TMP2:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
8866 // CHECK9-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP2]], 99
8867 // CHECK9-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
8868 // CHECK9: cond.true:
8869 // CHECK9-NEXT: br label [[COND_END:%.*]]
8870 // CHECK9: cond.false:
8871 // CHECK9-NEXT: [[TMP3:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
8872 // CHECK9-NEXT: br label [[COND_END]]
8873 // CHECK9: cond.end:
8874 // CHECK9-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP3]], [[COND_FALSE]] ]
8875 // CHECK9-NEXT: store i32 [[COND]], i32* [[DOTOMP_COMB_UB]], align 4
8876 // CHECK9-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4
8877 // CHECK9-NEXT: store i32 [[TMP4]], i32* [[DOTOMP_IV]], align 4
8878 // CHECK9-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
8879 // CHECK9: omp.inner.for.cond:
8880 // CHECK9-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
8881 // CHECK9-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
8882 // CHECK9-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP5]], [[TMP6]]
8883 // CHECK9-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
8884 // CHECK9: omp.inner.for.body:
8885 // CHECK9-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4
8886 // CHECK9-NEXT: [[TMP8:%.*]] = zext i32 [[TMP7]] to i64
8887 // CHECK9-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
8888 // CHECK9-NEXT: [[TMP10:%.*]] = zext i32 [[TMP9]] to i64
8889 // CHECK9-NEXT: [[TMP11:%.*]] = load i8, i8* [[CONV]], align 8
8890 // CHECK9-NEXT: [[TOBOOL:%.*]] = trunc i8 [[TMP11]] to i1
8891 // CHECK9-NEXT: br i1 [[TOBOOL]], label [[OMP_IF_THEN:%.*]], label [[OMP_IF_ELSE:%.*]]
8892 // CHECK9: omp_if.then:
8893 // CHECK9-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB3]], i32 2, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64, i64)* @.omp_outlined..15 to void (i32*, i32*, ...)*), i64 [[TMP8]], i64 [[TMP10]])
8894 // CHECK9-NEXT: br label [[OMP_IF_END:%.*]]
8895 // CHECK9: omp_if.else:
8896 // CHECK9-NEXT: call void @__kmpc_serialized_parallel(%struct.ident_t* @[[GLOB3]], i32 [[TMP1]])
8897 // CHECK9-NEXT: [[TMP12:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
8898 // CHECK9-NEXT: call void @.omp_outlined..15(i32* [[TMP12]], i32* [[DOTBOUND_ZERO_ADDR]], i64 [[TMP8]], i64 [[TMP10]]) #[[ATTR2]]
8899 // CHECK9-NEXT: call void @__kmpc_end_serialized_parallel(%struct.ident_t* @[[GLOB3]], i32 [[TMP1]])
8900 // CHECK9-NEXT: br label [[OMP_IF_END]]
8901 // CHECK9: omp_if.end:
8902 // CHECK9-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
8903 // CHECK9: omp.inner.for.inc:
8904 // CHECK9-NEXT: [[TMP13:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
8905 // CHECK9-NEXT: [[TMP14:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4
8906 // CHECK9-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP13]], [[TMP14]]
8907 // CHECK9-NEXT: store i32 [[ADD]], i32* [[DOTOMP_IV]], align 4
8908 // CHECK9-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP31:![0-9]+]]
8909 // CHECK9: omp.inner.for.end:
8910 // CHECK9-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
8911 // CHECK9: omp.loop.exit:
8912 // CHECK9-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP1]])
8913 // CHECK9-NEXT: [[TMP15:%.*]] = load i32, i32* [[DOTOMP_IS_LAST]], align 4
8914 // CHECK9-NEXT: [[TMP16:%.*]] = icmp ne i32 [[TMP15]], 0
8915 // CHECK9-NEXT: br i1 [[TMP16]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
8916 // CHECK9: .omp.final.then:
8917 // CHECK9-NEXT: store i32 100, i32* [[I]], align 4
8918 // CHECK9-NEXT: br label [[DOTOMP_FINAL_DONE]]
8919 // CHECK9: .omp.final.done:
8920 // CHECK9-NEXT: ret void
8921 //
8922 //
8923 // CHECK9-LABEL: define {{[^@]+}}@.omp_outlined..15
8924 // CHECK9-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[DOTPREVIOUS_LB_:%.*]], i64 [[DOTPREVIOUS_UB_:%.*]]) #[[ATTR1]] {
8925 // CHECK9-NEXT: entry:
8926 // CHECK9-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
8927 // CHECK9-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
8928 // CHECK9-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i64, align 8
8929 // CHECK9-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i64, align 8
8930 // CHECK9-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
8931 // CHECK9-NEXT: [[TMP:%.*]] = alloca i32, align 4
8932 // CHECK9-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4
8933 // CHECK9-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4
8934 // CHECK9-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
8935 // CHECK9-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
8936 // CHECK9-NEXT: [[I:%.*]] = alloca i32, align 4
8937 // CHECK9-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
8938 // CHECK9-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
8939 // CHECK9-NEXT: store i64 [[DOTPREVIOUS_LB_]], i64* [[DOTPREVIOUS_LB__ADDR]], align 8
8940 // CHECK9-NEXT: store i64 [[DOTPREVIOUS_UB_]], i64* [[DOTPREVIOUS_UB__ADDR]], align 8
8941 // CHECK9-NEXT: store i32 0, i32* [[DOTOMP_LB]], align 4
8942 // CHECK9-NEXT: store i32 99, i32* [[DOTOMP_UB]], align 4
8943 // CHECK9-NEXT: [[TMP0:%.*]] = load i64, i64* [[DOTPREVIOUS_LB__ADDR]], align 8
8944 // CHECK9-NEXT: [[CONV:%.*]] = trunc i64 [[TMP0]] to i32
8945 // CHECK9-NEXT: [[TMP1:%.*]] = load i64, i64* [[DOTPREVIOUS_UB__ADDR]], align 8
8946 // CHECK9-NEXT: [[CONV1:%.*]] = trunc i64 [[TMP1]] to i32
8947 // CHECK9-NEXT: store i32 [[CONV]], i32* [[DOTOMP_LB]], align 4
8948 // CHECK9-NEXT: store i32 [[CONV1]], i32* [[DOTOMP_UB]], align 4
8949 // CHECK9-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4
8950 // CHECK9-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4
8951 // CHECK9-NEXT: [[TMP2:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
8952 // CHECK9-NEXT: [[TMP3:%.*]] = load i32, i32* [[TMP2]], align 4
8953 // CHECK9-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB2]], i32 [[TMP3]], i32 34, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_LB]], i32* [[DOTOMP_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1)
8954 // CHECK9-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
8955 // CHECK9-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP4]], 99
8956 // CHECK9-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
8957 // CHECK9: cond.true:
8958 // CHECK9-NEXT: br label [[COND_END:%.*]]
8959 // CHECK9: cond.false:
8960 // CHECK9-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
8961 // CHECK9-NEXT: br label [[COND_END]]
8962 // CHECK9: cond.end:
8963 // CHECK9-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP5]], [[COND_FALSE]] ]
8964 // CHECK9-NEXT: store i32 [[COND]], i32* [[DOTOMP_UB]], align 4
8965 // CHECK9-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4
8966 // CHECK9-NEXT: store i32 [[TMP6]], i32* [[DOTOMP_IV]], align 4
8967 // CHECK9-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
8968 // CHECK9: omp.inner.for.cond:
8969 // CHECK9-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
8970 // CHECK9-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
8971 // CHECK9-NEXT: [[CMP2:%.*]] = icmp sle i32 [[TMP7]], [[TMP8]]
8972 // CHECK9-NEXT: br i1 [[CMP2]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
8973 // CHECK9: omp.inner.for.body:
8974 // CHECK9-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
8975 // CHECK9-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP9]], 1
8976 // CHECK9-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
8977 // CHECK9-NEXT: store i32 [[ADD]], i32* [[I]], align 4
8978 // CHECK9-NEXT: call void @_Z3fn3v()
8979 // CHECK9-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
8980 // CHECK9: omp.body.continue:
8981 // CHECK9-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
8982 // CHECK9: omp.inner.for.inc:
8983 // CHECK9-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
8984 // CHECK9-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP10]], 1
8985 // CHECK9-NEXT: store i32 [[ADD3]], i32* [[DOTOMP_IV]], align 4
8986 // CHECK9-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP32:![0-9]+]]
8987 // CHECK9: omp.inner.for.end:
8988 // CHECK9-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
8989 // CHECK9: omp.loop.exit:
8990 // CHECK9-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP3]])
8991 // CHECK9-NEXT: [[TMP11:%.*]] = load i32, i32* [[DOTOMP_IS_LAST]], align 4
8992 // CHECK9-NEXT: [[TMP12:%.*]] = icmp ne i32 [[TMP11]], 0
8993 // CHECK9-NEXT: br i1 [[TMP12]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
8994 // CHECK9: .omp.final.then:
8995 // CHECK9-NEXT: store i32 100, i32* [[I]], align 4
8996 // CHECK9-NEXT: br label [[DOTOMP_FINAL_DONE]]
8997 // CHECK9: .omp.final.done:
8998 // CHECK9-NEXT: ret void
8999 //
9000 //
9001 // CHECK9-LABEL: define {{[^@]+}}@.omp_offloading.requires_reg
9002 // CHECK9-SAME: () #[[ATTR5:[0-9]+]] {
9003 // CHECK9-NEXT: entry:
9004 // CHECK9-NEXT: call void @__tgt_register_requires(i64 1)
9005 // CHECK9-NEXT: ret void
9006 //
9007 //
9008 // CHECK10-LABEL: define {{[^@]+}}@_Z9gtid_testv
9009 // CHECK10-SAME: () #[[ATTR0:[0-9]+]] {
9010 // CHECK10-NEXT: entry:
9011 // CHECK10-NEXT: [[TMP:%.*]] = alloca i32, align 4
9012 // CHECK10-NEXT: [[_TMP1:%.*]] = alloca i32, align 4
9013 // CHECK10-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3:[0-9]+]], i64 -1, i64 100)
9014 // CHECK10-NEXT: [[TMP0:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9gtid_testv_l43.region_id, i32 0, i8** null, i8** null, i64* null, i64* null, i8** null, i8** null, i32 0, i32 0)
9015 // CHECK10-NEXT: [[TMP1:%.*]] = icmp ne i32 [[TMP0]], 0
9016 // CHECK10-NEXT: br i1 [[TMP1]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
9017 // CHECK10: omp_offload.failed:
9018 // CHECK10-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9gtid_testv_l43() #[[ATTR2:[0-9]+]]
9019 // CHECK10-NEXT: br label [[OMP_OFFLOAD_CONT]]
9020 // CHECK10: omp_offload.cont:
9021 // CHECK10-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i64 100)
9022 // CHECK10-NEXT: [[TMP2:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9gtid_testv_l47.region_id, i32 0, i8** null, i8** null, i64* null, i64* null, i8** null, i8** null, i32 0, i32 1)
9023 // CHECK10-NEXT: [[TMP3:%.*]] = icmp ne i32 [[TMP2]], 0
9024 // CHECK10-NEXT: br i1 [[TMP3]], label [[OMP_OFFLOAD_FAILED2:%.*]], label [[OMP_OFFLOAD_CONT3:%.*]]
9025 // CHECK10: omp_offload.failed2:
9026 // CHECK10-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9gtid_testv_l47() #[[ATTR2]]
9027 // CHECK10-NEXT: br label [[OMP_OFFLOAD_CONT3]]
9028 // CHECK10: omp_offload.cont3:
9029 // CHECK10-NEXT: ret void
9030 //
9031 //
9032 // CHECK10-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9gtid_testv_l43
9033 // CHECK10-SAME: () #[[ATTR1:[0-9]+]] {
9034 // CHECK10-NEXT: entry:
9035 // CHECK10-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 0, void (i32*, i32*, ...)* bitcast (void (i32*, i32*)* @.omp_outlined. to void (i32*, i32*, ...)*))
9036 // CHECK10-NEXT: ret void
9037 //
9038 //
9039 // CHECK10-LABEL: define {{[^@]+}}@.omp_outlined.
9040 // CHECK10-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]]) #[[ATTR1]] {
9041 // CHECK10-NEXT: entry:
9042 // CHECK10-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
9043 // CHECK10-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
9044 // CHECK10-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
9045 // CHECK10-NEXT: [[TMP:%.*]] = alloca i32, align 4
9046 // CHECK10-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4
9047 // CHECK10-NEXT: [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4
9048 // CHECK10-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
9049 // CHECK10-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
9050 // CHECK10-NEXT: [[I:%.*]] = alloca i32, align 4
9051 // CHECK10-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
9052 // CHECK10-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
9053 // CHECK10-NEXT: store i32 0, i32* [[DOTOMP_COMB_LB]], align 4
9054 // CHECK10-NEXT: store i32 99, i32* [[DOTOMP_COMB_UB]], align 4
9055 // CHECK10-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4
9056 // CHECK10-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4
9057 // CHECK10-NEXT: [[TMP0:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
9058 // CHECK10-NEXT: [[TMP1:%.*]] = load i32, i32* [[TMP0]], align 4
9059 // CHECK10-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB1:[0-9]+]], i32 [[TMP1]], i32 92, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_COMB_LB]], i32* [[DOTOMP_COMB_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1)
9060 // CHECK10-NEXT: [[TMP2:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
9061 // CHECK10-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP2]], 99
9062 // CHECK10-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
9063 // CHECK10: cond.true:
9064 // CHECK10-NEXT: br label [[COND_END:%.*]]
9065 // CHECK10: cond.false:
9066 // CHECK10-NEXT: [[TMP3:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
9067 // CHECK10-NEXT: br label [[COND_END]]
9068 // CHECK10: cond.end:
9069 // CHECK10-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP3]], [[COND_FALSE]] ]
9070 // CHECK10-NEXT: store i32 [[COND]], i32* [[DOTOMP_COMB_UB]], align 4
9071 // CHECK10-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4
9072 // CHECK10-NEXT: store i32 [[TMP4]], i32* [[DOTOMP_IV]], align 4
9073 // CHECK10-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
9074 // CHECK10: omp.inner.for.cond:
9075 // CHECK10-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
9076 // CHECK10-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
9077 // CHECK10-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP5]], [[TMP6]]
9078 // CHECK10-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
9079 // CHECK10: omp.inner.for.body:
9080 // CHECK10-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4
9081 // CHECK10-NEXT: [[TMP8:%.*]] = zext i32 [[TMP7]] to i64
9082 // CHECK10-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
9083 // CHECK10-NEXT: [[TMP10:%.*]] = zext i32 [[TMP9]] to i64
9084 // CHECK10-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB3]], i32 2, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64, i64)* @.omp_outlined..1 to void (i32*, i32*, ...)*), i64 [[TMP8]], i64 [[TMP10]])
9085 // CHECK10-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
9086 // CHECK10: omp.inner.for.inc:
9087 // CHECK10-NEXT: [[TMP11:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
9088 // CHECK10-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4
9089 // CHECK10-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP11]], [[TMP12]]
9090 // CHECK10-NEXT: store i32 [[ADD]], i32* [[DOTOMP_IV]], align 4
9091 // CHECK10-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP14:![0-9]+]]
9092 // CHECK10: omp.inner.for.end:
9093 // CHECK10-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
9094 // CHECK10: omp.loop.exit:
9095 // CHECK10-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP1]])
9096 // CHECK10-NEXT: [[TMP13:%.*]] = load i32, i32* [[DOTOMP_IS_LAST]], align 4
9097 // CHECK10-NEXT: [[TMP14:%.*]] = icmp ne i32 [[TMP13]], 0
9098 // CHECK10-NEXT: br i1 [[TMP14]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
9099 // CHECK10: .omp.final.then:
9100 // CHECK10-NEXT: store i32 100, i32* [[I]], align 4
9101 // CHECK10-NEXT: br label [[DOTOMP_FINAL_DONE]]
9102 // CHECK10: .omp.final.done:
9103 // CHECK10-NEXT: ret void
9104 //
9105 //
9106 // CHECK10-LABEL: define {{[^@]+}}@.omp_outlined..1
9107 // CHECK10-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[DOTPREVIOUS_LB_:%.*]], i64 [[DOTPREVIOUS_UB_:%.*]]) #[[ATTR1]] {
9108 // CHECK10-NEXT: entry:
9109 // CHECK10-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
9110 // CHECK10-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
9111 // CHECK10-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i64, align 8
9112 // CHECK10-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i64, align 8
9113 // CHECK10-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
9114 // CHECK10-NEXT: [[TMP:%.*]] = alloca i32, align 4
9115 // CHECK10-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4
9116 // CHECK10-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4
9117 // CHECK10-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
9118 // CHECK10-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
9119 // CHECK10-NEXT: [[I:%.*]] = alloca i32, align 4
9120 // CHECK10-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
9121 // CHECK10-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
9122 // CHECK10-NEXT: store i64 [[DOTPREVIOUS_LB_]], i64* [[DOTPREVIOUS_LB__ADDR]], align 8
9123 // CHECK10-NEXT: store i64 [[DOTPREVIOUS_UB_]], i64* [[DOTPREVIOUS_UB__ADDR]], align 8
9124 // CHECK10-NEXT: store i32 0, i32* [[DOTOMP_LB]], align 4
9125 // CHECK10-NEXT: store i32 99, i32* [[DOTOMP_UB]], align 4
9126 // CHECK10-NEXT: [[TMP0:%.*]] = load i64, i64* [[DOTPREVIOUS_LB__ADDR]], align 8
9127 // CHECK10-NEXT: [[CONV:%.*]] = trunc i64 [[TMP0]] to i32
9128 // CHECK10-NEXT: [[TMP1:%.*]] = load i64, i64* [[DOTPREVIOUS_UB__ADDR]], align 8
9129 // CHECK10-NEXT: [[CONV1:%.*]] = trunc i64 [[TMP1]] to i32
9130 // CHECK10-NEXT: store i32 [[CONV]], i32* [[DOTOMP_LB]], align 4
9131 // CHECK10-NEXT: store i32 [[CONV1]], i32* [[DOTOMP_UB]], align 4
9132 // CHECK10-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4
9133 // CHECK10-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4
9134 // CHECK10-NEXT: [[TMP2:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
9135 // CHECK10-NEXT: [[TMP3:%.*]] = load i32, i32* [[TMP2]], align 4
9136 // CHECK10-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB2:[0-9]+]], i32 [[TMP3]], i32 34, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_LB]], i32* [[DOTOMP_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1)
9137 // CHECK10-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
9138 // CHECK10-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP4]], 99
9139 // CHECK10-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
9140 // CHECK10: cond.true:
9141 // CHECK10-NEXT: br label [[COND_END:%.*]]
9142 // CHECK10: cond.false:
9143 // CHECK10-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
9144 // CHECK10-NEXT: br label [[COND_END]]
9145 // CHECK10: cond.end:
9146 // CHECK10-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP5]], [[COND_FALSE]] ]
9147 // CHECK10-NEXT: store i32 [[COND]], i32* [[DOTOMP_UB]], align 4
9148 // CHECK10-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4
9149 // CHECK10-NEXT: store i32 [[TMP6]], i32* [[DOTOMP_IV]], align 4
9150 // CHECK10-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
9151 // CHECK10: omp.inner.for.cond:
9152 // CHECK10-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
9153 // CHECK10-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
9154 // CHECK10-NEXT: [[CMP2:%.*]] = icmp sle i32 [[TMP7]], [[TMP8]]
9155 // CHECK10-NEXT: br i1 [[CMP2]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
9156 // CHECK10: omp.inner.for.body:
9157 // CHECK10-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
9158 // CHECK10-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP9]], 1
9159 // CHECK10-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
9160 // CHECK10-NEXT: store i32 [[ADD]], i32* [[I]], align 4
9161 // CHECK10-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
9162 // CHECK10: omp.body.continue:
9163 // CHECK10-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
9164 // CHECK10: omp.inner.for.inc:
9165 // CHECK10-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
9166 // CHECK10-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP10]], 1
9167 // CHECK10-NEXT: store i32 [[ADD3]], i32* [[DOTOMP_IV]], align 4
9168 // CHECK10-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP16:![0-9]+]]
9169 // CHECK10: omp.inner.for.end:
9170 // CHECK10-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
9171 // CHECK10: omp.loop.exit:
9172 // CHECK10-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP3]])
9173 // CHECK10-NEXT: [[TMP11:%.*]] = load i32, i32* [[DOTOMP_IS_LAST]], align 4
9174 // CHECK10-NEXT: [[TMP12:%.*]] = icmp ne i32 [[TMP11]], 0
9175 // CHECK10-NEXT: br i1 [[TMP12]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
9176 // CHECK10: .omp.final.then:
9177 // CHECK10-NEXT: store i32 100, i32* [[I]], align 4
9178 // CHECK10-NEXT: br label [[DOTOMP_FINAL_DONE]]
9179 // CHECK10: .omp.final.done:
9180 // CHECK10-NEXT: ret void
9181 //
9182 //
9183 // CHECK10-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9gtid_testv_l47
9184 // CHECK10-SAME: () #[[ATTR1]] {
9185 // CHECK10-NEXT: entry:
9186 // CHECK10-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 0, void (i32*, i32*, ...)* bitcast (void (i32*, i32*)* @.omp_outlined..2 to void (i32*, i32*, ...)*))
9187 // CHECK10-NEXT: ret void
9188 //
9189 //
9190 // CHECK10-LABEL: define {{[^@]+}}@.omp_outlined..2
9191 // CHECK10-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]]) #[[ATTR1]] {
9192 // CHECK10-NEXT: entry:
9193 // CHECK10-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
9194 // CHECK10-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
9195 // CHECK10-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
9196 // CHECK10-NEXT: [[TMP:%.*]] = alloca i32, align 4
9197 // CHECK10-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4
9198 // CHECK10-NEXT: [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4
9199 // CHECK10-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
9200 // CHECK10-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
9201 // CHECK10-NEXT: [[I:%.*]] = alloca i32, align 4
9202 // CHECK10-NEXT: [[DOTBOUND_ZERO_ADDR:%.*]] = alloca i32, align 4
9203 // CHECK10-NEXT: store i32 0, i32* [[DOTBOUND_ZERO_ADDR]], align 4
9204 // CHECK10-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
9205 // CHECK10-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
9206 // CHECK10-NEXT: store i32 0, i32* [[DOTOMP_COMB_LB]], align 4
9207 // CHECK10-NEXT: store i32 99, i32* [[DOTOMP_COMB_UB]], align 4
9208 // CHECK10-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4
9209 // CHECK10-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4
9210 // CHECK10-NEXT: [[TMP0:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
9211 // CHECK10-NEXT: [[TMP1:%.*]] = load i32, i32* [[TMP0]], align 4
9212 // CHECK10-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB1]], i32 [[TMP1]], i32 92, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_COMB_LB]], i32* [[DOTOMP_COMB_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1)
9213 // CHECK10-NEXT: [[TMP2:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
9214 // CHECK10-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP2]], 99
9215 // CHECK10-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
9216 // CHECK10: cond.true:
9217 // CHECK10-NEXT: br label [[COND_END:%.*]]
9218 // CHECK10: cond.false:
9219 // CHECK10-NEXT: [[TMP3:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
9220 // CHECK10-NEXT: br label [[COND_END]]
9221 // CHECK10: cond.end:
9222 // CHECK10-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP3]], [[COND_FALSE]] ]
9223 // CHECK10-NEXT: store i32 [[COND]], i32* [[DOTOMP_COMB_UB]], align 4
9224 // CHECK10-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4
9225 // CHECK10-NEXT: store i32 [[TMP4]], i32* [[DOTOMP_IV]], align 4
9226 // CHECK10-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
9227 // CHECK10: omp.inner.for.cond:
9228 // CHECK10-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
9229 // CHECK10-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
9230 // CHECK10-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP5]], [[TMP6]]
9231 // CHECK10-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
9232 // CHECK10: omp.inner.for.body:
9233 // CHECK10-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4
9234 // CHECK10-NEXT: [[TMP8:%.*]] = zext i32 [[TMP7]] to i64
9235 // CHECK10-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
9236 // CHECK10-NEXT: [[TMP10:%.*]] = zext i32 [[TMP9]] to i64
9237 // CHECK10-NEXT: call void @__kmpc_serialized_parallel(%struct.ident_t* @[[GLOB3]], i32 [[TMP1]])
9238 // CHECK10-NEXT: [[TMP11:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
9239 // CHECK10-NEXT: call void @.omp_outlined..3(i32* [[TMP11]], i32* [[DOTBOUND_ZERO_ADDR]], i64 [[TMP8]], i64 [[TMP10]]) #[[ATTR2]]
9240 // CHECK10-NEXT: call void @__kmpc_end_serialized_parallel(%struct.ident_t* @[[GLOB3]], i32 [[TMP1]])
9241 // CHECK10-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
9242 // CHECK10: omp.inner.for.inc:
9243 // CHECK10-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
9244 // CHECK10-NEXT: [[TMP13:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4
9245 // CHECK10-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP12]], [[TMP13]]
9246 // CHECK10-NEXT: store i32 [[ADD]], i32* [[DOTOMP_IV]], align 4
9247 // CHECK10-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP19:![0-9]+]]
9248 // CHECK10: omp.inner.for.end:
9249 // CHECK10-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
9250 // CHECK10: omp.loop.exit:
9251 // CHECK10-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP1]])
9252 // CHECK10-NEXT: [[TMP14:%.*]] = load i32, i32* [[DOTOMP_IS_LAST]], align 4
9253 // CHECK10-NEXT: [[TMP15:%.*]] = icmp ne i32 [[TMP14]], 0
9254 // CHECK10-NEXT: br i1 [[TMP15]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
9255 // CHECK10: .omp.final.then:
9256 // CHECK10-NEXT: store i32 100, i32* [[I]], align 4
9257 // CHECK10-NEXT: br label [[DOTOMP_FINAL_DONE]]
9258 // CHECK10: .omp.final.done:
9259 // CHECK10-NEXT: ret void
9260 //
9261 //
9262 // CHECK10-LABEL: define {{[^@]+}}@.omp_outlined..3
9263 // CHECK10-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[DOTPREVIOUS_LB_:%.*]], i64 [[DOTPREVIOUS_UB_:%.*]]) #[[ATTR1]] {
9264 // CHECK10-NEXT: entry:
9265 // CHECK10-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
9266 // CHECK10-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
9267 // CHECK10-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i64, align 8
9268 // CHECK10-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i64, align 8
9269 // CHECK10-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
9270 // CHECK10-NEXT: [[TMP:%.*]] = alloca i32, align 4
9271 // CHECK10-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4
9272 // CHECK10-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4
9273 // CHECK10-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
9274 // CHECK10-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
9275 // CHECK10-NEXT: [[I:%.*]] = alloca i32, align 4
9276 // CHECK10-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
9277 // CHECK10-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
9278 // CHECK10-NEXT: store i64 [[DOTPREVIOUS_LB_]], i64* [[DOTPREVIOUS_LB__ADDR]], align 8
9279 // CHECK10-NEXT: store i64 [[DOTPREVIOUS_UB_]], i64* [[DOTPREVIOUS_UB__ADDR]], align 8
9280 // CHECK10-NEXT: store i32 0, i32* [[DOTOMP_LB]], align 4
9281 // CHECK10-NEXT: store i32 99, i32* [[DOTOMP_UB]], align 4
9282 // CHECK10-NEXT: [[TMP0:%.*]] = load i64, i64* [[DOTPREVIOUS_LB__ADDR]], align 8
9283 // CHECK10-NEXT: [[CONV:%.*]] = trunc i64 [[TMP0]] to i32
9284 // CHECK10-NEXT: [[TMP1:%.*]] = load i64, i64* [[DOTPREVIOUS_UB__ADDR]], align 8
9285 // CHECK10-NEXT: [[CONV1:%.*]] = trunc i64 [[TMP1]] to i32
9286 // CHECK10-NEXT: store i32 [[CONV]], i32* [[DOTOMP_LB]], align 4
9287 // CHECK10-NEXT: store i32 [[CONV1]], i32* [[DOTOMP_UB]], align 4
9288 // CHECK10-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4
9289 // CHECK10-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4
9290 // CHECK10-NEXT: [[TMP2:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
9291 // CHECK10-NEXT: [[TMP3:%.*]] = load i32, i32* [[TMP2]], align 4
9292 // CHECK10-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB2]], i32 [[TMP3]], i32 34, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_LB]], i32* [[DOTOMP_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1)
9293 // CHECK10-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
9294 // CHECK10-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP4]], 99
9295 // CHECK10-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
9296 // CHECK10: cond.true:
9297 // CHECK10-NEXT: br label [[COND_END:%.*]]
9298 // CHECK10: cond.false:
9299 // CHECK10-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
9300 // CHECK10-NEXT: br label [[COND_END]]
9301 // CHECK10: cond.end:
9302 // CHECK10-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP5]], [[COND_FALSE]] ]
9303 // CHECK10-NEXT: store i32 [[COND]], i32* [[DOTOMP_UB]], align 4
9304 // CHECK10-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4
9305 // CHECK10-NEXT: store i32 [[TMP6]], i32* [[DOTOMP_IV]], align 4
9306 // CHECK10-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
9307 // CHECK10: omp.inner.for.cond:
9308 // CHECK10-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
9309 // CHECK10-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
9310 // CHECK10-NEXT: [[CMP2:%.*]] = icmp sle i32 [[TMP7]], [[TMP8]]
9311 // CHECK10-NEXT: br i1 [[CMP2]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
9312 // CHECK10: omp.inner.for.body:
9313 // CHECK10-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
9314 // CHECK10-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP9]], 1
9315 // CHECK10-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
9316 // CHECK10-NEXT: store i32 [[ADD]], i32* [[I]], align 4
9317 // CHECK10-NEXT: call void @_Z9gtid_testv()
9318 // CHECK10-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
9319 // CHECK10: omp.body.continue:
9320 // CHECK10-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
9321 // CHECK10: omp.inner.for.inc:
9322 // CHECK10-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
9323 // CHECK10-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP10]], 1
9324 // CHECK10-NEXT: store i32 [[ADD3]], i32* [[DOTOMP_IV]], align 4
9325 // CHECK10-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP20:![0-9]+]]
9326 // CHECK10: omp.inner.for.end:
9327 // CHECK10-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
9328 // CHECK10: omp.loop.exit:
9329 // CHECK10-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP3]])
9330 // CHECK10-NEXT: [[TMP11:%.*]] = load i32, i32* [[DOTOMP_IS_LAST]], align 4
9331 // CHECK10-NEXT: [[TMP12:%.*]] = icmp ne i32 [[TMP11]], 0
9332 // CHECK10-NEXT: br i1 [[TMP12]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
9333 // CHECK10: .omp.final.then:
9334 // CHECK10-NEXT: store i32 100, i32* [[I]], align 4
9335 // CHECK10-NEXT: br label [[DOTOMP_FINAL_DONE]]
9336 // CHECK10: .omp.final.done:
9337 // CHECK10-NEXT: ret void
9338 //
9339 //
9340 // CHECK10-LABEL: define {{[^@]+}}@main
9341 // CHECK10-SAME: () #[[ATTR3:[0-9]+]] {
9342 // CHECK10-NEXT: entry:
9343 // CHECK10-NEXT: [[RETVAL:%.*]] = alloca i32, align 4
9344 // CHECK10-NEXT: [[TMP:%.*]] = alloca i32, align 4
9345 // CHECK10-NEXT: [[_TMP1:%.*]] = alloca i32, align 4
9346 // CHECK10-NEXT: [[ARG_CASTED:%.*]] = alloca i64, align 8
9347 // CHECK10-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [1 x i8*], align 8
9348 // CHECK10-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [1 x i8*], align 8
9349 // CHECK10-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [1 x i8*], align 8
9350 // CHECK10-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i8, align 1
9351 // CHECK10-NEXT: [[_TMP5:%.*]] = alloca i32, align 4
9352 // CHECK10-NEXT: store i32 0, i32* [[RETVAL]], align 4
9353 // CHECK10-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i64 100)
9354 // CHECK10-NEXT: [[TMP0:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l76.region_id, i32 0, i8** null, i8** null, i64* null, i64* null, i8** null, i8** null, i32 0, i32 0)
9355 // CHECK10-NEXT: [[TMP1:%.*]] = icmp ne i32 [[TMP0]], 0
9356 // CHECK10-NEXT: br i1 [[TMP1]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
9357 // CHECK10: omp_offload.failed:
9358 // CHECK10-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l76() #[[ATTR2]]
9359 // CHECK10-NEXT: br label [[OMP_OFFLOAD_CONT]]
9360 // CHECK10: omp_offload.cont:
9361 // CHECK10-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i64 100)
9362 // CHECK10-NEXT: [[TMP2:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l84.region_id, i32 0, i8** null, i8** null, i64* null, i64* null, i8** null, i8** null, i32 0, i32 1)
9363 // CHECK10-NEXT: [[TMP3:%.*]] = icmp ne i32 [[TMP2]], 0
9364 // CHECK10-NEXT: br i1 [[TMP3]], label [[OMP_OFFLOAD_FAILED2:%.*]], label [[OMP_OFFLOAD_CONT3:%.*]]
9365 // CHECK10: omp_offload.failed2:
9366 // CHECK10-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l84() #[[ATTR2]]
9367 // CHECK10-NEXT: br label [[OMP_OFFLOAD_CONT3]]
9368 // CHECK10: omp_offload.cont3:
9369 // CHECK10-NEXT: [[TMP4:%.*]] = load i32, i32* @Arg, align 4
9370 // CHECK10-NEXT: [[CONV:%.*]] = bitcast i64* [[ARG_CASTED]] to i32*
9371 // CHECK10-NEXT: store i32 [[TMP4]], i32* [[CONV]], align 4
9372 // CHECK10-NEXT: [[TMP5:%.*]] = load i64, i64* [[ARG_CASTED]], align 8
9373 // CHECK10-NEXT: [[TMP6:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
9374 // CHECK10-NEXT: [[TMP7:%.*]] = bitcast i8** [[TMP6]] to i64*
9375 // CHECK10-NEXT: store i64 [[TMP5]], i64* [[TMP7]], align 8
9376 // CHECK10-NEXT: [[TMP8:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
9377 // CHECK10-NEXT: [[TMP9:%.*]] = bitcast i8** [[TMP8]] to i64*
9378 // CHECK10-NEXT: store i64 [[TMP5]], i64* [[TMP9]], align 8
9379 // CHECK10-NEXT: [[TMP10:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0
9380 // CHECK10-NEXT: store i8* null, i8** [[TMP10]], align 8
9381 // CHECK10-NEXT: [[TMP11:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
9382 // CHECK10-NEXT: [[TMP12:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
9383 // CHECK10-NEXT: [[TMP13:%.*]] = load i32, i32* @Arg, align 4
9384 // CHECK10-NEXT: [[TOBOOL:%.*]] = icmp ne i32 [[TMP13]], 0
9385 // CHECK10-NEXT: [[FROMBOOL:%.*]] = zext i1 [[TOBOOL]] to i8
9386 // CHECK10-NEXT: store i8 [[FROMBOOL]], i8* [[DOTCAPTURE_EXPR_]], align 1
9387 // CHECK10-NEXT: [[TMP14:%.*]] = load i8, i8* [[DOTCAPTURE_EXPR_]], align 1
9388 // CHECK10-NEXT: [[TOBOOL4:%.*]] = trunc i8 [[TMP14]] to i1
9389 // CHECK10-NEXT: [[TMP15:%.*]] = select i1 [[TOBOOL4]], i32 0, i32 1
9390 // CHECK10-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i64 100)
9391 // CHECK10-NEXT: [[TMP16:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l92.region_id, i32 1, i8** [[TMP11]], i8** [[TMP12]], i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_sizes, i32 0, i32 0), i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_maptypes, i32 0, i32 0), i8** null, i8** null, i32 0, i32 [[TMP15]])
9392 // CHECK10-NEXT: [[TMP17:%.*]] = icmp ne i32 [[TMP16]], 0
9393 // CHECK10-NEXT: br i1 [[TMP17]], label [[OMP_OFFLOAD_FAILED6:%.*]], label [[OMP_OFFLOAD_CONT7:%.*]]
9394 // CHECK10: omp_offload.failed6:
9395 // CHECK10-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l92(i64 [[TMP5]]) #[[ATTR2]]
9396 // CHECK10-NEXT: br label [[OMP_OFFLOAD_CONT7]]
9397 // CHECK10: omp_offload.cont7:
9398 // CHECK10-NEXT: [[TMP18:%.*]] = load i32, i32* @Arg, align 4
9399 // CHECK10-NEXT: [[CALL:%.*]] = call i32 @_Z5tmainIiEiT_(i32 [[TMP18]])
9400 // CHECK10-NEXT: ret i32 [[CALL]]
9401 //
9402 //
9403 // CHECK10-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l76
9404 // CHECK10-SAME: () #[[ATTR1]] {
9405 // CHECK10-NEXT: entry:
9406 // CHECK10-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 0, void (i32*, i32*, ...)* bitcast (void (i32*, i32*)* @.omp_outlined..4 to void (i32*, i32*, ...)*))
9407 // CHECK10-NEXT: ret void
9408 //
9409 //
9410 // CHECK10-LABEL: define {{[^@]+}}@.omp_outlined..4
9411 // CHECK10-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]]) #[[ATTR1]] {
9412 // CHECK10-NEXT: entry:
9413 // CHECK10-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
9414 // CHECK10-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
9415 // CHECK10-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
9416 // CHECK10-NEXT: [[TMP:%.*]] = alloca i32, align 4
9417 // CHECK10-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4
9418 // CHECK10-NEXT: [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4
9419 // CHECK10-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
9420 // CHECK10-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
9421 // CHECK10-NEXT: [[I:%.*]] = alloca i32, align 4
9422 // CHECK10-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
9423 // CHECK10-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
9424 // CHECK10-NEXT: store i32 0, i32* [[DOTOMP_COMB_LB]], align 4
9425 // CHECK10-NEXT: store i32 99, i32* [[DOTOMP_COMB_UB]], align 4
9426 // CHECK10-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4
9427 // CHECK10-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4
9428 // CHECK10-NEXT: [[TMP0:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
9429 // CHECK10-NEXT: [[TMP1:%.*]] = load i32, i32* [[TMP0]], align 4
9430 // CHECK10-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB1]], i32 [[TMP1]], i32 92, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_COMB_LB]], i32* [[DOTOMP_COMB_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1)
9431 // CHECK10-NEXT: [[TMP2:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
9432 // CHECK10-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP2]], 99
9433 // CHECK10-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
9434 // CHECK10: cond.true:
9435 // CHECK10-NEXT: br label [[COND_END:%.*]]
9436 // CHECK10: cond.false:
9437 // CHECK10-NEXT: [[TMP3:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
9438 // CHECK10-NEXT: br label [[COND_END]]
9439 // CHECK10: cond.end:
9440 // CHECK10-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP3]], [[COND_FALSE]] ]
9441 // CHECK10-NEXT: store i32 [[COND]], i32* [[DOTOMP_COMB_UB]], align 4
9442 // CHECK10-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4
9443 // CHECK10-NEXT: store i32 [[TMP4]], i32* [[DOTOMP_IV]], align 4
9444 // CHECK10-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
9445 // CHECK10: omp.inner.for.cond:
9446 // CHECK10-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
9447 // CHECK10-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
9448 // CHECK10-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP5]], [[TMP6]]
9449 // CHECK10-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
9450 // CHECK10: omp.inner.for.body:
9451 // CHECK10-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4
9452 // CHECK10-NEXT: [[TMP8:%.*]] = zext i32 [[TMP7]] to i64
9453 // CHECK10-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
9454 // CHECK10-NEXT: [[TMP10:%.*]] = zext i32 [[TMP9]] to i64
9455 // CHECK10-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB3]], i32 2, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64, i64)* @.omp_outlined..5 to void (i32*, i32*, ...)*), i64 [[TMP8]], i64 [[TMP10]])
9456 // CHECK10-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
9457 // CHECK10: omp.inner.for.inc:
9458 // CHECK10-NEXT: [[TMP11:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
9459 // CHECK10-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4
9460 // CHECK10-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP11]], [[TMP12]]
9461 // CHECK10-NEXT: store i32 [[ADD]], i32* [[DOTOMP_IV]], align 4
9462 // CHECK10-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP21:![0-9]+]]
9463 // CHECK10: omp.inner.for.end:
9464 // CHECK10-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
9465 // CHECK10: omp.loop.exit:
9466 // CHECK10-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP1]])
9467 // CHECK10-NEXT: [[TMP13:%.*]] = load i32, i32* [[DOTOMP_IS_LAST]], align 4
9468 // CHECK10-NEXT: [[TMP14:%.*]] = icmp ne i32 [[TMP13]], 0
9469 // CHECK10-NEXT: br i1 [[TMP14]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
9470 // CHECK10: .omp.final.then:
9471 // CHECK10-NEXT: store i32 100, i32* [[I]], align 4
9472 // CHECK10-NEXT: br label [[DOTOMP_FINAL_DONE]]
9473 // CHECK10: .omp.final.done:
9474 // CHECK10-NEXT: ret void
9475 //
9476 //
9477 // CHECK10-LABEL: define {{[^@]+}}@.omp_outlined..5
9478 // CHECK10-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[DOTPREVIOUS_LB_:%.*]], i64 [[DOTPREVIOUS_UB_:%.*]]) #[[ATTR1]] {
9479 // CHECK10-NEXT: entry:
9480 // CHECK10-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
9481 // CHECK10-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
9482 // CHECK10-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i64, align 8
9483 // CHECK10-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i64, align 8
9484 // CHECK10-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
9485 // CHECK10-NEXT: [[TMP:%.*]] = alloca i32, align 4
9486 // CHECK10-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4
9487 // CHECK10-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4
9488 // CHECK10-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
9489 // CHECK10-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
9490 // CHECK10-NEXT: [[I:%.*]] = alloca i32, align 4
9491 // CHECK10-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
9492 // CHECK10-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
9493 // CHECK10-NEXT: store i64 [[DOTPREVIOUS_LB_]], i64* [[DOTPREVIOUS_LB__ADDR]], align 8
9494 // CHECK10-NEXT: store i64 [[DOTPREVIOUS_UB_]], i64* [[DOTPREVIOUS_UB__ADDR]], align 8
9495 // CHECK10-NEXT: store i32 0, i32* [[DOTOMP_LB]], align 4
9496 // CHECK10-NEXT: store i32 99, i32* [[DOTOMP_UB]], align 4
9497 // CHECK10-NEXT: [[TMP0:%.*]] = load i64, i64* [[DOTPREVIOUS_LB__ADDR]], align 8
9498 // CHECK10-NEXT: [[CONV:%.*]] = trunc i64 [[TMP0]] to i32
9499 // CHECK10-NEXT: [[TMP1:%.*]] = load i64, i64* [[DOTPREVIOUS_UB__ADDR]], align 8
9500 // CHECK10-NEXT: [[CONV1:%.*]] = trunc i64 [[TMP1]] to i32
9501 // CHECK10-NEXT: store i32 [[CONV]], i32* [[DOTOMP_LB]], align 4
9502 // CHECK10-NEXT: store i32 [[CONV1]], i32* [[DOTOMP_UB]], align 4
9503 // CHECK10-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4
9504 // CHECK10-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4
9505 // CHECK10-NEXT: [[TMP2:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
9506 // CHECK10-NEXT: [[TMP3:%.*]] = load i32, i32* [[TMP2]], align 4
9507 // CHECK10-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB2]], i32 [[TMP3]], i32 34, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_LB]], i32* [[DOTOMP_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1)
9508 // CHECK10-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
9509 // CHECK10-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP4]], 99
9510 // CHECK10-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
9511 // CHECK10: cond.true:
9512 // CHECK10-NEXT: br label [[COND_END:%.*]]
9513 // CHECK10: cond.false:
9514 // CHECK10-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
9515 // CHECK10-NEXT: br label [[COND_END]]
9516 // CHECK10: cond.end:
9517 // CHECK10-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP5]], [[COND_FALSE]] ]
9518 // CHECK10-NEXT: store i32 [[COND]], i32* [[DOTOMP_UB]], align 4
9519 // CHECK10-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4
9520 // CHECK10-NEXT: store i32 [[TMP6]], i32* [[DOTOMP_IV]], align 4
9521 // CHECK10-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
9522 // CHECK10: omp.inner.for.cond:
9523 // CHECK10-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
9524 // CHECK10-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
9525 // CHECK10-NEXT: [[CMP2:%.*]] = icmp sle i32 [[TMP7]], [[TMP8]]
9526 // CHECK10-NEXT: br i1 [[CMP2]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
9527 // CHECK10: omp.inner.for.body:
9528 // CHECK10-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
9529 // CHECK10-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP9]], 1
9530 // CHECK10-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
9531 // CHECK10-NEXT: store i32 [[ADD]], i32* [[I]], align 4
9532 // CHECK10-NEXT: call void @_Z3fn4v()
9533 // CHECK10-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
9534 // CHECK10: omp.body.continue:
9535 // CHECK10-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
9536 // CHECK10: omp.inner.for.inc:
9537 // CHECK10-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
9538 // CHECK10-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP10]], 1
9539 // CHECK10-NEXT: store i32 [[ADD3]], i32* [[DOTOMP_IV]], align 4
9540 // CHECK10-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP22:![0-9]+]]
9541 // CHECK10: omp.inner.for.end:
9542 // CHECK10-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
9543 // CHECK10: omp.loop.exit:
9544 // CHECK10-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP3]])
9545 // CHECK10-NEXT: [[TMP11:%.*]] = load i32, i32* [[DOTOMP_IS_LAST]], align 4
9546 // CHECK10-NEXT: [[TMP12:%.*]] = icmp ne i32 [[TMP11]], 0
9547 // CHECK10-NEXT: br i1 [[TMP12]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
9548 // CHECK10: .omp.final.then:
9549 // CHECK10-NEXT: store i32 100, i32* [[I]], align 4
9550 // CHECK10-NEXT: br label [[DOTOMP_FINAL_DONE]]
9551 // CHECK10: .omp.final.done:
9552 // CHECK10-NEXT: ret void
9553 //
9554 //
9555 // CHECK10-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l84
9556 // CHECK10-SAME: () #[[ATTR1]] {
9557 // CHECK10-NEXT: entry:
9558 // CHECK10-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 0, void (i32*, i32*, ...)* bitcast (void (i32*, i32*)* @.omp_outlined..6 to void (i32*, i32*, ...)*))
9559 // CHECK10-NEXT: ret void
9560 //
9561 //
9562 // CHECK10-LABEL: define {{[^@]+}}@.omp_outlined..6
9563 // CHECK10-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]]) #[[ATTR1]] {
9564 // CHECK10-NEXT: entry:
9565 // CHECK10-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
9566 // CHECK10-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
9567 // CHECK10-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
9568 // CHECK10-NEXT: [[TMP:%.*]] = alloca i32, align 4
9569 // CHECK10-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4
9570 // CHECK10-NEXT: [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4
9571 // CHECK10-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
9572 // CHECK10-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
9573 // CHECK10-NEXT: [[I:%.*]] = alloca i32, align 4
9574 // CHECK10-NEXT: [[DOTBOUND_ZERO_ADDR:%.*]] = alloca i32, align 4
9575 // CHECK10-NEXT: store i32 0, i32* [[DOTBOUND_ZERO_ADDR]], align 4
9576 // CHECK10-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
9577 // CHECK10-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
9578 // CHECK10-NEXT: store i32 0, i32* [[DOTOMP_COMB_LB]], align 4
9579 // CHECK10-NEXT: store i32 99, i32* [[DOTOMP_COMB_UB]], align 4
9580 // CHECK10-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4
9581 // CHECK10-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4
9582 // CHECK10-NEXT: [[TMP0:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
9583 // CHECK10-NEXT: [[TMP1:%.*]] = load i32, i32* [[TMP0]], align 4
9584 // CHECK10-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB1]], i32 [[TMP1]], i32 92, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_COMB_LB]], i32* [[DOTOMP_COMB_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1)
9585 // CHECK10-NEXT: [[TMP2:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
9586 // CHECK10-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP2]], 99
9587 // CHECK10-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
9588 // CHECK10: cond.true:
9589 // CHECK10-NEXT: br label [[COND_END:%.*]]
9590 // CHECK10: cond.false:
9591 // CHECK10-NEXT: [[TMP3:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
9592 // CHECK10-NEXT: br label [[COND_END]]
9593 // CHECK10: cond.end:
9594 // CHECK10-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP3]], [[COND_FALSE]] ]
9595 // CHECK10-NEXT: store i32 [[COND]], i32* [[DOTOMP_COMB_UB]], align 4
9596 // CHECK10-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4
9597 // CHECK10-NEXT: store i32 [[TMP4]], i32* [[DOTOMP_IV]], align 4
9598 // CHECK10-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
9599 // CHECK10: omp.inner.for.cond:
9600 // CHECK10-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
9601 // CHECK10-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
9602 // CHECK10-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP5]], [[TMP6]]
9603 // CHECK10-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
9604 // CHECK10: omp.inner.for.body:
9605 // CHECK10-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4
9606 // CHECK10-NEXT: [[TMP8:%.*]] = zext i32 [[TMP7]] to i64
9607 // CHECK10-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
9608 // CHECK10-NEXT: [[TMP10:%.*]] = zext i32 [[TMP9]] to i64
9609 // CHECK10-NEXT: call void @__kmpc_serialized_parallel(%struct.ident_t* @[[GLOB3]], i32 [[TMP1]])
9610 // CHECK10-NEXT: [[TMP11:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
9611 // CHECK10-NEXT: call void @.omp_outlined..7(i32* [[TMP11]], i32* [[DOTBOUND_ZERO_ADDR]], i64 [[TMP8]], i64 [[TMP10]]) #[[ATTR2]]
9612 // CHECK10-NEXT: call void @__kmpc_end_serialized_parallel(%struct.ident_t* @[[GLOB3]], i32 [[TMP1]])
9613 // CHECK10-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
9614 // CHECK10: omp.inner.for.inc:
9615 // CHECK10-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
9616 // CHECK10-NEXT: [[TMP13:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4
9617 // CHECK10-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP12]], [[TMP13]]
9618 // CHECK10-NEXT: store i32 [[ADD]], i32* [[DOTOMP_IV]], align 4
9619 // CHECK10-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP23:![0-9]+]]
9620 // CHECK10: omp.inner.for.end:
9621 // CHECK10-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
9622 // CHECK10: omp.loop.exit:
9623 // CHECK10-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP1]])
9624 // CHECK10-NEXT: [[TMP14:%.*]] = load i32, i32* [[DOTOMP_IS_LAST]], align 4
9625 // CHECK10-NEXT: [[TMP15:%.*]] = icmp ne i32 [[TMP14]], 0
9626 // CHECK10-NEXT: br i1 [[TMP15]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
9627 // CHECK10: .omp.final.then:
9628 // CHECK10-NEXT: store i32 100, i32* [[I]], align 4
9629 // CHECK10-NEXT: br label [[DOTOMP_FINAL_DONE]]
9630 // CHECK10: .omp.final.done:
9631 // CHECK10-NEXT: ret void
9632 //
9633 //
9634 // CHECK10-LABEL: define {{[^@]+}}@.omp_outlined..7
9635 // CHECK10-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[DOTPREVIOUS_LB_:%.*]], i64 [[DOTPREVIOUS_UB_:%.*]]) #[[ATTR1]] {
9636 // CHECK10-NEXT: entry:
9637 // CHECK10-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
9638 // CHECK10-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
9639 // CHECK10-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i64, align 8
9640 // CHECK10-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i64, align 8
9641 // CHECK10-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
9642 // CHECK10-NEXT: [[TMP:%.*]] = alloca i32, align 4
9643 // CHECK10-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4
9644 // CHECK10-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4
9645 // CHECK10-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
9646 // CHECK10-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
9647 // CHECK10-NEXT: [[I:%.*]] = alloca i32, align 4
9648 // CHECK10-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
9649 // CHECK10-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
9650 // CHECK10-NEXT: store i64 [[DOTPREVIOUS_LB_]], i64* [[DOTPREVIOUS_LB__ADDR]], align 8
9651 // CHECK10-NEXT: store i64 [[DOTPREVIOUS_UB_]], i64* [[DOTPREVIOUS_UB__ADDR]], align 8
9652 // CHECK10-NEXT: store i32 0, i32* [[DOTOMP_LB]], align 4
9653 // CHECK10-NEXT: store i32 99, i32* [[DOTOMP_UB]], align 4
9654 // CHECK10-NEXT: [[TMP0:%.*]] = load i64, i64* [[DOTPREVIOUS_LB__ADDR]], align 8
9655 // CHECK10-NEXT: [[CONV:%.*]] = trunc i64 [[TMP0]] to i32
9656 // CHECK10-NEXT: [[TMP1:%.*]] = load i64, i64* [[DOTPREVIOUS_UB__ADDR]], align 8
9657 // CHECK10-NEXT: [[CONV1:%.*]] = trunc i64 [[TMP1]] to i32
9658 // CHECK10-NEXT: store i32 [[CONV]], i32* [[DOTOMP_LB]], align 4
9659 // CHECK10-NEXT: store i32 [[CONV1]], i32* [[DOTOMP_UB]], align 4
9660 // CHECK10-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4
9661 // CHECK10-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4
9662 // CHECK10-NEXT: [[TMP2:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
9663 // CHECK10-NEXT: [[TMP3:%.*]] = load i32, i32* [[TMP2]], align 4
9664 // CHECK10-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB2]], i32 [[TMP3]], i32 34, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_LB]], i32* [[DOTOMP_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1)
9665 // CHECK10-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
9666 // CHECK10-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP4]], 99
9667 // CHECK10-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
9668 // CHECK10: cond.true:
9669 // CHECK10-NEXT: br label [[COND_END:%.*]]
9670 // CHECK10: cond.false:
9671 // CHECK10-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
9672 // CHECK10-NEXT: br label [[COND_END]]
9673 // CHECK10: cond.end:
9674 // CHECK10-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP5]], [[COND_FALSE]] ]
9675 // CHECK10-NEXT: store i32 [[COND]], i32* [[DOTOMP_UB]], align 4
9676 // CHECK10-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4
9677 // CHECK10-NEXT: store i32 [[TMP6]], i32* [[DOTOMP_IV]], align 4
9678 // CHECK10-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
9679 // CHECK10: omp.inner.for.cond:
9680 // CHECK10-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
9681 // CHECK10-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
9682 // CHECK10-NEXT: [[CMP2:%.*]] = icmp sle i32 [[TMP7]], [[TMP8]]
9683 // CHECK10-NEXT: br i1 [[CMP2]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
9684 // CHECK10: omp.inner.for.body:
9685 // CHECK10-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
9686 // CHECK10-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP9]], 1
9687 // CHECK10-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
9688 // CHECK10-NEXT: store i32 [[ADD]], i32* [[I]], align 4
9689 // CHECK10-NEXT: call void @_Z3fn5v()
9690 // CHECK10-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
9691 // CHECK10: omp.body.continue:
9692 // CHECK10-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
9693 // CHECK10: omp.inner.for.inc:
9694 // CHECK10-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
9695 // CHECK10-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP10]], 1
9696 // CHECK10-NEXT: store i32 [[ADD3]], i32* [[DOTOMP_IV]], align 4
9697 // CHECK10-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP24:![0-9]+]]
9698 // CHECK10: omp.inner.for.end:
9699 // CHECK10-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
9700 // CHECK10: omp.loop.exit:
9701 // CHECK10-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP3]])
9702 // CHECK10-NEXT: [[TMP11:%.*]] = load i32, i32* [[DOTOMP_IS_LAST]], align 4
9703 // CHECK10-NEXT: [[TMP12:%.*]] = icmp ne i32 [[TMP11]], 0
9704 // CHECK10-NEXT: br i1 [[TMP12]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
9705 // CHECK10: .omp.final.then:
9706 // CHECK10-NEXT: store i32 100, i32* [[I]], align 4
9707 // CHECK10-NEXT: br label [[DOTOMP_FINAL_DONE]]
9708 // CHECK10: .omp.final.done:
9709 // CHECK10-NEXT: ret void
9710 //
9711 //
9712 // CHECK10-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l92
9713 // CHECK10-SAME: (i64 [[ARG:%.*]]) #[[ATTR1]] {
9714 // CHECK10-NEXT: entry:
9715 // CHECK10-NEXT: [[ARG_ADDR:%.*]] = alloca i64, align 8
9716 // CHECK10-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i8, align 1
9717 // CHECK10-NEXT: [[DOTCAPTURE_EXPR__CASTED:%.*]] = alloca i64, align 8
9718 // CHECK10-NEXT: store i64 [[ARG]], i64* [[ARG_ADDR]], align 8
9719 // CHECK10-NEXT: [[CONV:%.*]] = bitcast i64* [[ARG_ADDR]] to i32*
9720 // CHECK10-NEXT: [[TMP0:%.*]] = load i32, i32* [[CONV]], align 8
9721 // CHECK10-NEXT: [[TOBOOL:%.*]] = icmp ne i32 [[TMP0]], 0
9722 // CHECK10-NEXT: [[FROMBOOL:%.*]] = zext i1 [[TOBOOL]] to i8
9723 // CHECK10-NEXT: store i8 [[FROMBOOL]], i8* [[DOTCAPTURE_EXPR_]], align 1
9724 // CHECK10-NEXT: [[TMP1:%.*]] = load i8, i8* [[DOTCAPTURE_EXPR_]], align 1
9725 // CHECK10-NEXT: [[TOBOOL1:%.*]] = trunc i8 [[TMP1]] to i1
9726 // CHECK10-NEXT: [[CONV2:%.*]] = bitcast i64* [[DOTCAPTURE_EXPR__CASTED]] to i8*
9727 // CHECK10-NEXT: [[FROMBOOL3:%.*]] = zext i1 [[TOBOOL1]] to i8
9728 // CHECK10-NEXT: store i8 [[FROMBOOL3]], i8* [[CONV2]], align 1
9729 // CHECK10-NEXT: [[TMP2:%.*]] = load i64, i64* [[DOTCAPTURE_EXPR__CASTED]], align 8
9730 // CHECK10-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64)* @.omp_outlined..8 to void (i32*, i32*, ...)*), i64 [[TMP2]])
9731 // CHECK10-NEXT: ret void
9732 //
9733 //
9734 // CHECK10-LABEL: define {{[^@]+}}@.omp_outlined..8
9735 // CHECK10-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[DOTCAPTURE_EXPR_:%.*]]) #[[ATTR1]] {
9736 // CHECK10-NEXT: entry:
9737 // CHECK10-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
9738 // CHECK10-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
9739 // CHECK10-NEXT: [[DOTCAPTURE_EXPR__ADDR:%.*]] = alloca i64, align 8
9740 // CHECK10-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
9741 // CHECK10-NEXT: [[TMP:%.*]] = alloca i32, align 4
9742 // CHECK10-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4
9743 // CHECK10-NEXT: [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4
9744 // CHECK10-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
9745 // CHECK10-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
9746 // CHECK10-NEXT: [[I:%.*]] = alloca i32, align 4
9747 // CHECK10-NEXT: [[DOTBOUND_ZERO_ADDR:%.*]] = alloca i32, align 4
9748 // CHECK10-NEXT: store i32 0, i32* [[DOTBOUND_ZERO_ADDR]], align 4
9749 // CHECK10-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
9750 // CHECK10-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
9751 // CHECK10-NEXT: store i64 [[DOTCAPTURE_EXPR_]], i64* [[DOTCAPTURE_EXPR__ADDR]], align 8
9752 // CHECK10-NEXT: [[CONV:%.*]] = bitcast i64* [[DOTCAPTURE_EXPR__ADDR]] to i8*
9753 // CHECK10-NEXT: store i32 0, i32* [[DOTOMP_COMB_LB]], align 4
9754 // CHECK10-NEXT: store i32 99, i32* [[DOTOMP_COMB_UB]], align 4
9755 // CHECK10-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4
9756 // CHECK10-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4
9757 // CHECK10-NEXT: [[TMP0:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
9758 // CHECK10-NEXT: [[TMP1:%.*]] = load i32, i32* [[TMP0]], align 4
9759 // CHECK10-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB1]], i32 [[TMP1]], i32 92, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_COMB_LB]], i32* [[DOTOMP_COMB_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1)
9760 // CHECK10-NEXT: [[TMP2:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
9761 // CHECK10-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP2]], 99
9762 // CHECK10-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
9763 // CHECK10: cond.true:
9764 // CHECK10-NEXT: br label [[COND_END:%.*]]
9765 // CHECK10: cond.false:
9766 // CHECK10-NEXT: [[TMP3:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
9767 // CHECK10-NEXT: br label [[COND_END]]
9768 // CHECK10: cond.end:
9769 // CHECK10-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP3]], [[COND_FALSE]] ]
9770 // CHECK10-NEXT: store i32 [[COND]], i32* [[DOTOMP_COMB_UB]], align 4
9771 // CHECK10-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4
9772 // CHECK10-NEXT: store i32 [[TMP4]], i32* [[DOTOMP_IV]], align 4
9773 // CHECK10-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
9774 // CHECK10: omp.inner.for.cond:
9775 // CHECK10-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
9776 // CHECK10-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
9777 // CHECK10-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP5]], [[TMP6]]
9778 // CHECK10-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
9779 // CHECK10: omp.inner.for.body:
9780 // CHECK10-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4
9781 // CHECK10-NEXT: [[TMP8:%.*]] = zext i32 [[TMP7]] to i64
9782 // CHECK10-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
9783 // CHECK10-NEXT: [[TMP10:%.*]] = zext i32 [[TMP9]] to i64
9784 // CHECK10-NEXT: [[TMP11:%.*]] = load i8, i8* [[CONV]], align 8
9785 // CHECK10-NEXT: [[TOBOOL:%.*]] = trunc i8 [[TMP11]] to i1
9786 // CHECK10-NEXT: br i1 [[TOBOOL]], label [[OMP_IF_THEN:%.*]], label [[OMP_IF_ELSE:%.*]]
9787 // CHECK10: omp_if.then:
9788 // CHECK10-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB3]], i32 2, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64, i64)* @.omp_outlined..9 to void (i32*, i32*, ...)*), i64 [[TMP8]], i64 [[TMP10]])
9789 // CHECK10-NEXT: br label [[OMP_IF_END:%.*]]
9790 // CHECK10: omp_if.else:
9791 // CHECK10-NEXT: call void @__kmpc_serialized_parallel(%struct.ident_t* @[[GLOB3]], i32 [[TMP1]])
9792 // CHECK10-NEXT: [[TMP12:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
9793 // CHECK10-NEXT: call void @.omp_outlined..9(i32* [[TMP12]], i32* [[DOTBOUND_ZERO_ADDR]], i64 [[TMP8]], i64 [[TMP10]]) #[[ATTR2]]
9794 // CHECK10-NEXT: call void @__kmpc_end_serialized_parallel(%struct.ident_t* @[[GLOB3]], i32 [[TMP1]])
9795 // CHECK10-NEXT: br label [[OMP_IF_END]]
9796 // CHECK10: omp_if.end:
9797 // CHECK10-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
9798 // CHECK10: omp.inner.for.inc:
9799 // CHECK10-NEXT: [[TMP13:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
9800 // CHECK10-NEXT: [[TMP14:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4
9801 // CHECK10-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP13]], [[TMP14]]
9802 // CHECK10-NEXT: store i32 [[ADD]], i32* [[DOTOMP_IV]], align 4
9803 // CHECK10-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP25:![0-9]+]]
9804 // CHECK10: omp.inner.for.end:
9805 // CHECK10-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
9806 // CHECK10: omp.loop.exit:
9807 // CHECK10-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP1]])
9808 // CHECK10-NEXT: [[TMP15:%.*]] = load i32, i32* [[DOTOMP_IS_LAST]], align 4
9809 // CHECK10-NEXT: [[TMP16:%.*]] = icmp ne i32 [[TMP15]], 0
9810 // CHECK10-NEXT: br i1 [[TMP16]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
9811 // CHECK10: .omp.final.then:
9812 // CHECK10-NEXT: store i32 100, i32* [[I]], align 4
9813 // CHECK10-NEXT: br label [[DOTOMP_FINAL_DONE]]
9814 // CHECK10: .omp.final.done:
9815 // CHECK10-NEXT: ret void
9816 //
9817 //
9818 // CHECK10-LABEL: define {{[^@]+}}@.omp_outlined..9
9819 // CHECK10-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[DOTPREVIOUS_LB_:%.*]], i64 [[DOTPREVIOUS_UB_:%.*]]) #[[ATTR1]] {
9820 // CHECK10-NEXT: entry:
9821 // CHECK10-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
9822 // CHECK10-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
9823 // CHECK10-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i64, align 8
9824 // CHECK10-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i64, align 8
9825 // CHECK10-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
9826 // CHECK10-NEXT: [[TMP:%.*]] = alloca i32, align 4
9827 // CHECK10-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4
9828 // CHECK10-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4
9829 // CHECK10-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
9830 // CHECK10-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
9831 // CHECK10-NEXT: [[I:%.*]] = alloca i32, align 4
9832 // CHECK10-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
9833 // CHECK10-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
9834 // CHECK10-NEXT: store i64 [[DOTPREVIOUS_LB_]], i64* [[DOTPREVIOUS_LB__ADDR]], align 8
9835 // CHECK10-NEXT: store i64 [[DOTPREVIOUS_UB_]], i64* [[DOTPREVIOUS_UB__ADDR]], align 8
9836 // CHECK10-NEXT: store i32 0, i32* [[DOTOMP_LB]], align 4
9837 // CHECK10-NEXT: store i32 99, i32* [[DOTOMP_UB]], align 4
9838 // CHECK10-NEXT: [[TMP0:%.*]] = load i64, i64* [[DOTPREVIOUS_LB__ADDR]], align 8
9839 // CHECK10-NEXT: [[CONV:%.*]] = trunc i64 [[TMP0]] to i32
9840 // CHECK10-NEXT: [[TMP1:%.*]] = load i64, i64* [[DOTPREVIOUS_UB__ADDR]], align 8
9841 // CHECK10-NEXT: [[CONV1:%.*]] = trunc i64 [[TMP1]] to i32
9842 // CHECK10-NEXT: store i32 [[CONV]], i32* [[DOTOMP_LB]], align 4
9843 // CHECK10-NEXT: store i32 [[CONV1]], i32* [[DOTOMP_UB]], align 4
9844 // CHECK10-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4
9845 // CHECK10-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4
9846 // CHECK10-NEXT: [[TMP2:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
9847 // CHECK10-NEXT: [[TMP3:%.*]] = load i32, i32* [[TMP2]], align 4
9848 // CHECK10-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB2]], i32 [[TMP3]], i32 34, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_LB]], i32* [[DOTOMP_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1)
9849 // CHECK10-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
9850 // CHECK10-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP4]], 99
9851 // CHECK10-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
9852 // CHECK10: cond.true:
9853 // CHECK10-NEXT: br label [[COND_END:%.*]]
9854 // CHECK10: cond.false:
9855 // CHECK10-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
9856 // CHECK10-NEXT: br label [[COND_END]]
9857 // CHECK10: cond.end:
9858 // CHECK10-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP5]], [[COND_FALSE]] ]
9859 // CHECK10-NEXT: store i32 [[COND]], i32* [[DOTOMP_UB]], align 4
9860 // CHECK10-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4
9861 // CHECK10-NEXT: store i32 [[TMP6]], i32* [[DOTOMP_IV]], align 4
9862 // CHECK10-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
9863 // CHECK10: omp.inner.for.cond:
9864 // CHECK10-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
9865 // CHECK10-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
9866 // CHECK10-NEXT: [[CMP2:%.*]] = icmp sle i32 [[TMP7]], [[TMP8]]
9867 // CHECK10-NEXT: br i1 [[CMP2]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
9868 // CHECK10: omp.inner.for.body:
9869 // CHECK10-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
9870 // CHECK10-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP9]], 1
9871 // CHECK10-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
9872 // CHECK10-NEXT: store i32 [[ADD]], i32* [[I]], align 4
9873 // CHECK10-NEXT: call void @_Z3fn6v()
9874 // CHECK10-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
9875 // CHECK10: omp.body.continue:
9876 // CHECK10-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
9877 // CHECK10: omp.inner.for.inc:
9878 // CHECK10-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
9879 // CHECK10-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP10]], 1
9880 // CHECK10-NEXT: store i32 [[ADD3]], i32* [[DOTOMP_IV]], align 4
9881 // CHECK10-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP26:![0-9]+]]
9882 // CHECK10: omp.inner.for.end:
9883 // CHECK10-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
9884 // CHECK10: omp.loop.exit:
9885 // CHECK10-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP3]])
9886 // CHECK10-NEXT: [[TMP11:%.*]] = load i32, i32* [[DOTOMP_IS_LAST]], align 4
9887 // CHECK10-NEXT: [[TMP12:%.*]] = icmp ne i32 [[TMP11]], 0
9888 // CHECK10-NEXT: br i1 [[TMP12]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
9889 // CHECK10: .omp.final.then:
9890 // CHECK10-NEXT: store i32 100, i32* [[I]], align 4
9891 // CHECK10-NEXT: br label [[DOTOMP_FINAL_DONE]]
9892 // CHECK10: .omp.final.done:
9893 // CHECK10-NEXT: ret void
9894 //
9895 //
9896 // CHECK10-LABEL: define {{[^@]+}}@_Z5tmainIiEiT_
9897 // CHECK10-SAME: (i32 [[ARG:%.*]]) #[[ATTR0]] comdat {
9898 // CHECK10-NEXT: entry:
9899 // CHECK10-NEXT: [[ARG_ADDR:%.*]] = alloca i32, align 4
9900 // CHECK10-NEXT: [[TMP:%.*]] = alloca i32, align 4
9901 // CHECK10-NEXT: [[_TMP1:%.*]] = alloca i32, align 4
9902 // CHECK10-NEXT: [[ARG_CASTED:%.*]] = alloca i64, align 8
9903 // CHECK10-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [1 x i8*], align 8
9904 // CHECK10-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [1 x i8*], align 8
9905 // CHECK10-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [1 x i8*], align 8
9906 // CHECK10-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i8, align 1
9907 // CHECK10-NEXT: [[_TMP5:%.*]] = alloca i32, align 4
9908 // CHECK10-NEXT: store i32 [[ARG]], i32* [[ARG_ADDR]], align 4
9909 // CHECK10-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i64 100)
9910 // CHECK10-NEXT: [[TMP0:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiEiT__l57.region_id, i32 0, i8** null, i8** null, i64* null, i64* null, i8** null, i8** null, i32 0, i32 0)
9911 // CHECK10-NEXT: [[TMP1:%.*]] = icmp ne i32 [[TMP0]], 0
9912 // CHECK10-NEXT: br i1 [[TMP1]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
9913 // CHECK10: omp_offload.failed:
9914 // CHECK10-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiEiT__l57() #[[ATTR2]]
9915 // CHECK10-NEXT: br label [[OMP_OFFLOAD_CONT]]
9916 // CHECK10: omp_offload.cont:
9917 // CHECK10-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i64 100)
9918 // CHECK10-NEXT: [[TMP2:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiEiT__l62.region_id, i32 0, i8** null, i8** null, i64* null, i64* null, i8** null, i8** null, i32 0, i32 1)
9919 // CHECK10-NEXT: [[TMP3:%.*]] = icmp ne i32 [[TMP2]], 0
9920 // CHECK10-NEXT: br i1 [[TMP3]], label [[OMP_OFFLOAD_FAILED2:%.*]], label [[OMP_OFFLOAD_CONT3:%.*]]
9921 // CHECK10: omp_offload.failed2:
9922 // CHECK10-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiEiT__l62() #[[ATTR2]]
9923 // CHECK10-NEXT: br label [[OMP_OFFLOAD_CONT3]]
9924 // CHECK10: omp_offload.cont3:
9925 // CHECK10-NEXT: [[TMP4:%.*]] = load i32, i32* [[ARG_ADDR]], align 4
9926 // CHECK10-NEXT: [[CONV:%.*]] = bitcast i64* [[ARG_CASTED]] to i32*
9927 // CHECK10-NEXT: store i32 [[TMP4]], i32* [[CONV]], align 4
9928 // CHECK10-NEXT: [[TMP5:%.*]] = load i64, i64* [[ARG_CASTED]], align 8
9929 // CHECK10-NEXT: [[TMP6:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
9930 // CHECK10-NEXT: [[TMP7:%.*]] = bitcast i8** [[TMP6]] to i64*
9931 // CHECK10-NEXT: store i64 [[TMP5]], i64* [[TMP7]], align 8
9932 // CHECK10-NEXT: [[TMP8:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
9933 // CHECK10-NEXT: [[TMP9:%.*]] = bitcast i8** [[TMP8]] to i64*
9934 // CHECK10-NEXT: store i64 [[TMP5]], i64* [[TMP9]], align 8
9935 // CHECK10-NEXT: [[TMP10:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0
9936 // CHECK10-NEXT: store i8* null, i8** [[TMP10]], align 8
9937 // CHECK10-NEXT: [[TMP11:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
9938 // CHECK10-NEXT: [[TMP12:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
9939 // CHECK10-NEXT: [[TMP13:%.*]] = load i32, i32* [[ARG_ADDR]], align 4
9940 // CHECK10-NEXT: [[TOBOOL:%.*]] = icmp ne i32 [[TMP13]], 0
9941 // CHECK10-NEXT: [[FROMBOOL:%.*]] = zext i1 [[TOBOOL]] to i8
9942 // CHECK10-NEXT: store i8 [[FROMBOOL]], i8* [[DOTCAPTURE_EXPR_]], align 1
9943 // CHECK10-NEXT: [[TMP14:%.*]] = load i8, i8* [[DOTCAPTURE_EXPR_]], align 1
9944 // CHECK10-NEXT: [[TOBOOL4:%.*]] = trunc i8 [[TMP14]] to i1
9945 // CHECK10-NEXT: [[TMP15:%.*]] = select i1 [[TOBOOL4]], i32 0, i32 1
9946 // CHECK10-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i64 100)
9947 // CHECK10-NEXT: [[TMP16:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiEiT__l67.region_id, i32 1, i8** [[TMP11]], i8** [[TMP12]], i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_sizes.16, i32 0, i32 0), i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_maptypes.17, i32 0, i32 0), i8** null, i8** null, i32 0, i32 [[TMP15]])
9948 // CHECK10-NEXT: [[TMP17:%.*]] = icmp ne i32 [[TMP16]], 0
9949 // CHECK10-NEXT: br i1 [[TMP17]], label [[OMP_OFFLOAD_FAILED6:%.*]], label [[OMP_OFFLOAD_CONT7:%.*]]
9950 // CHECK10: omp_offload.failed6:
9951 // CHECK10-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiEiT__l67(i64 [[TMP5]]) #[[ATTR2]]
9952 // CHECK10-NEXT: br label [[OMP_OFFLOAD_CONT7]]
9953 // CHECK10: omp_offload.cont7:
9954 // CHECK10-NEXT: ret i32 0
9955 //
9956 //
9957 // CHECK10-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiEiT__l57
9958 // CHECK10-SAME: () #[[ATTR1]] {
9959 // CHECK10-NEXT: entry:
9960 // CHECK10-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 0, void (i32*, i32*, ...)* bitcast (void (i32*, i32*)* @.omp_outlined..10 to void (i32*, i32*, ...)*))
9961 // CHECK10-NEXT: ret void
9962 //
9963 //
9964 // CHECK10-LABEL: define {{[^@]+}}@.omp_outlined..10
9965 // CHECK10-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]]) #[[ATTR1]] {
9966 // CHECK10-NEXT: entry:
9967 // CHECK10-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
9968 // CHECK10-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
9969 // CHECK10-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
9970 // CHECK10-NEXT: [[TMP:%.*]] = alloca i32, align 4
9971 // CHECK10-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4
9972 // CHECK10-NEXT: [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4
9973 // CHECK10-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
9974 // CHECK10-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
9975 // CHECK10-NEXT: [[I:%.*]] = alloca i32, align 4
9976 // CHECK10-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
9977 // CHECK10-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
9978 // CHECK10-NEXT: store i32 0, i32* [[DOTOMP_COMB_LB]], align 4
9979 // CHECK10-NEXT: store i32 99, i32* [[DOTOMP_COMB_UB]], align 4
9980 // CHECK10-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4
9981 // CHECK10-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4
9982 // CHECK10-NEXT: [[TMP0:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
9983 // CHECK10-NEXT: [[TMP1:%.*]] = load i32, i32* [[TMP0]], align 4
9984 // CHECK10-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB1]], i32 [[TMP1]], i32 92, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_COMB_LB]], i32* [[DOTOMP_COMB_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1)
9985 // CHECK10-NEXT: [[TMP2:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
9986 // CHECK10-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP2]], 99
9987 // CHECK10-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
9988 // CHECK10: cond.true:
9989 // CHECK10-NEXT: br label [[COND_END:%.*]]
9990 // CHECK10: cond.false:
9991 // CHECK10-NEXT: [[TMP3:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
9992 // CHECK10-NEXT: br label [[COND_END]]
9993 // CHECK10: cond.end:
9994 // CHECK10-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP3]], [[COND_FALSE]] ]
9995 // CHECK10-NEXT: store i32 [[COND]], i32* [[DOTOMP_COMB_UB]], align 4
9996 // CHECK10-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4
9997 // CHECK10-NEXT: store i32 [[TMP4]], i32* [[DOTOMP_IV]], align 4
9998 // CHECK10-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
9999 // CHECK10: omp.inner.for.cond:
10000 // CHECK10-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
10001 // CHECK10-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
10002 // CHECK10-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP5]], [[TMP6]]
10003 // CHECK10-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
10004 // CHECK10: omp.inner.for.body:
10005 // CHECK10-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4
10006 // CHECK10-NEXT: [[TMP8:%.*]] = zext i32 [[TMP7]] to i64
10007 // CHECK10-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
10008 // CHECK10-NEXT: [[TMP10:%.*]] = zext i32 [[TMP9]] to i64
10009 // CHECK10-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB3]], i32 2, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64, i64)* @.omp_outlined..11 to void (i32*, i32*, ...)*), i64 [[TMP8]], i64 [[TMP10]])
10010 // CHECK10-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
10011 // CHECK10: omp.inner.for.inc:
10012 // CHECK10-NEXT: [[TMP11:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
10013 // CHECK10-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4
10014 // CHECK10-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP11]], [[TMP12]]
10015 // CHECK10-NEXT: store i32 [[ADD]], i32* [[DOTOMP_IV]], align 4
10016 // CHECK10-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP27:![0-9]+]]
10017 // CHECK10: omp.inner.for.end:
10018 // CHECK10-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
10019 // CHECK10: omp.loop.exit:
10020 // CHECK10-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP1]])
10021 // CHECK10-NEXT: [[TMP13:%.*]] = load i32, i32* [[DOTOMP_IS_LAST]], align 4
10022 // CHECK10-NEXT: [[TMP14:%.*]] = icmp ne i32 [[TMP13]], 0
10023 // CHECK10-NEXT: br i1 [[TMP14]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
10024 // CHECK10: .omp.final.then:
10025 // CHECK10-NEXT: store i32 100, i32* [[I]], align 4
10026 // CHECK10-NEXT: br label [[DOTOMP_FINAL_DONE]]
10027 // CHECK10: .omp.final.done:
10028 // CHECK10-NEXT: ret void
10029 //
10030 //
10031 // CHECK10-LABEL: define {{[^@]+}}@.omp_outlined..11
10032 // CHECK10-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[DOTPREVIOUS_LB_:%.*]], i64 [[DOTPREVIOUS_UB_:%.*]]) #[[ATTR1]] {
10033 // CHECK10-NEXT: entry:
10034 // CHECK10-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
10035 // CHECK10-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
10036 // CHECK10-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i64, align 8
10037 // CHECK10-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i64, align 8
10038 // CHECK10-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
10039 // CHECK10-NEXT: [[TMP:%.*]] = alloca i32, align 4
10040 // CHECK10-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4
10041 // CHECK10-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4
10042 // CHECK10-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
10043 // CHECK10-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
10044 // CHECK10-NEXT: [[I:%.*]] = alloca i32, align 4
10045 // CHECK10-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
10046 // CHECK10-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
10047 // CHECK10-NEXT: store i64 [[DOTPREVIOUS_LB_]], i64* [[DOTPREVIOUS_LB__ADDR]], align 8
10048 // CHECK10-NEXT: store i64 [[DOTPREVIOUS_UB_]], i64* [[DOTPREVIOUS_UB__ADDR]], align 8
10049 // CHECK10-NEXT: store i32 0, i32* [[DOTOMP_LB]], align 4
10050 // CHECK10-NEXT: store i32 99, i32* [[DOTOMP_UB]], align 4
10051 // CHECK10-NEXT: [[TMP0:%.*]] = load i64, i64* [[DOTPREVIOUS_LB__ADDR]], align 8
10052 // CHECK10-NEXT: [[CONV:%.*]] = trunc i64 [[TMP0]] to i32
10053 // CHECK10-NEXT: [[TMP1:%.*]] = load i64, i64* [[DOTPREVIOUS_UB__ADDR]], align 8
10054 // CHECK10-NEXT: [[CONV1:%.*]] = trunc i64 [[TMP1]] to i32
10055 // CHECK10-NEXT: store i32 [[CONV]], i32* [[DOTOMP_LB]], align 4
10056 // CHECK10-NEXT: store i32 [[CONV1]], i32* [[DOTOMP_UB]], align 4
10057 // CHECK10-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4
10058 // CHECK10-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4
10059 // CHECK10-NEXT: [[TMP2:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
10060 // CHECK10-NEXT: [[TMP3:%.*]] = load i32, i32* [[TMP2]], align 4
10061 // CHECK10-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB2]], i32 [[TMP3]], i32 34, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_LB]], i32* [[DOTOMP_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1)
10062 // CHECK10-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
10063 // CHECK10-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP4]], 99
10064 // CHECK10-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
10065 // CHECK10: cond.true:
10066 // CHECK10-NEXT: br label [[COND_END:%.*]]
10067 // CHECK10: cond.false:
10068 // CHECK10-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
10069 // CHECK10-NEXT: br label [[COND_END]]
10070 // CHECK10: cond.end:
10071 // CHECK10-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP5]], [[COND_FALSE]] ]
10072 // CHECK10-NEXT: store i32 [[COND]], i32* [[DOTOMP_UB]], align 4
10073 // CHECK10-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4
10074 // CHECK10-NEXT: store i32 [[TMP6]], i32* [[DOTOMP_IV]], align 4
10075 // CHECK10-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
10076 // CHECK10: omp.inner.for.cond:
10077 // CHECK10-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
10078 // CHECK10-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
10079 // CHECK10-NEXT: [[CMP2:%.*]] = icmp sle i32 [[TMP7]], [[TMP8]]
10080 // CHECK10-NEXT: br i1 [[CMP2]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
10081 // CHECK10: omp.inner.for.body:
10082 // CHECK10-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
10083 // CHECK10-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP9]], 1
10084 // CHECK10-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
10085 // CHECK10-NEXT: store i32 [[ADD]], i32* [[I]], align 4
10086 // CHECK10-NEXT: call void @_Z3fn1v()
10087 // CHECK10-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
10088 // CHECK10: omp.body.continue:
10089 // CHECK10-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
10090 // CHECK10: omp.inner.for.inc:
10091 // CHECK10-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
10092 // CHECK10-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP10]], 1
10093 // CHECK10-NEXT: store i32 [[ADD3]], i32* [[DOTOMP_IV]], align 4
10094 // CHECK10-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP28:![0-9]+]]
10095 // CHECK10: omp.inner.for.end:
10096 // CHECK10-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
10097 // CHECK10: omp.loop.exit:
10098 // CHECK10-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP3]])
10099 // CHECK10-NEXT: [[TMP11:%.*]] = load i32, i32* [[DOTOMP_IS_LAST]], align 4
10100 // CHECK10-NEXT: [[TMP12:%.*]] = icmp ne i32 [[TMP11]], 0
10101 // CHECK10-NEXT: br i1 [[TMP12]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
10102 // CHECK10: .omp.final.then:
10103 // CHECK10-NEXT: store i32 100, i32* [[I]], align 4
10104 // CHECK10-NEXT: br label [[DOTOMP_FINAL_DONE]]
10105 // CHECK10: .omp.final.done:
10106 // CHECK10-NEXT: ret void
10107 //
10108 //
10109 // CHECK10-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiEiT__l62
10110 // CHECK10-SAME: () #[[ATTR1]] {
10111 // CHECK10-NEXT: entry:
10112 // CHECK10-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 0, void (i32*, i32*, ...)* bitcast (void (i32*, i32*)* @.omp_outlined..12 to void (i32*, i32*, ...)*))
10113 // CHECK10-NEXT: ret void
10114 //
10115 //
10116 // CHECK10-LABEL: define {{[^@]+}}@.omp_outlined..12
10117 // CHECK10-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]]) #[[ATTR1]] {
10118 // CHECK10-NEXT: entry:
10119 // CHECK10-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
10120 // CHECK10-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
10121 // CHECK10-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
10122 // CHECK10-NEXT: [[TMP:%.*]] = alloca i32, align 4
10123 // CHECK10-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4
10124 // CHECK10-NEXT: [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4
10125 // CHECK10-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
10126 // CHECK10-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
10127 // CHECK10-NEXT: [[I:%.*]] = alloca i32, align 4
10128 // CHECK10-NEXT: [[DOTBOUND_ZERO_ADDR:%.*]] = alloca i32, align 4
10129 // CHECK10-NEXT: store i32 0, i32* [[DOTBOUND_ZERO_ADDR]], align 4
10130 // CHECK10-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
10131 // CHECK10-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
10132 // CHECK10-NEXT: store i32 0, i32* [[DOTOMP_COMB_LB]], align 4
10133 // CHECK10-NEXT: store i32 99, i32* [[DOTOMP_COMB_UB]], align 4
10134 // CHECK10-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4
10135 // CHECK10-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4
10136 // CHECK10-NEXT: [[TMP0:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
10137 // CHECK10-NEXT: [[TMP1:%.*]] = load i32, i32* [[TMP0]], align 4
10138 // CHECK10-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB1]], i32 [[TMP1]], i32 92, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_COMB_LB]], i32* [[DOTOMP_COMB_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1)
10139 // CHECK10-NEXT: [[TMP2:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
10140 // CHECK10-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP2]], 99
10141 // CHECK10-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
10142 // CHECK10: cond.true:
10143 // CHECK10-NEXT: br label [[COND_END:%.*]]
10144 // CHECK10: cond.false:
10145 // CHECK10-NEXT: [[TMP3:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
10146 // CHECK10-NEXT: br label [[COND_END]]
10147 // CHECK10: cond.end:
10148 // CHECK10-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP3]], [[COND_FALSE]] ]
10149 // CHECK10-NEXT: store i32 [[COND]], i32* [[DOTOMP_COMB_UB]], align 4
10150 // CHECK10-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4
10151 // CHECK10-NEXT: store i32 [[TMP4]], i32* [[DOTOMP_IV]], align 4
10152 // CHECK10-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
10153 // CHECK10: omp.inner.for.cond:
10154 // CHECK10-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
10155 // CHECK10-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
10156 // CHECK10-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP5]], [[TMP6]]
10157 // CHECK10-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
10158 // CHECK10: omp.inner.for.body:
10159 // CHECK10-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4
10160 // CHECK10-NEXT: [[TMP8:%.*]] = zext i32 [[TMP7]] to i64
10161 // CHECK10-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
10162 // CHECK10-NEXT: [[TMP10:%.*]] = zext i32 [[TMP9]] to i64
10163 // CHECK10-NEXT: call void @__kmpc_serialized_parallel(%struct.ident_t* @[[GLOB3]], i32 [[TMP1]])
10164 // CHECK10-NEXT: [[TMP11:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
10165 // CHECK10-NEXT: call void @.omp_outlined..13(i32* [[TMP11]], i32* [[DOTBOUND_ZERO_ADDR]], i64 [[TMP8]], i64 [[TMP10]]) #[[ATTR2]]
10166 // CHECK10-NEXT: call void @__kmpc_end_serialized_parallel(%struct.ident_t* @[[GLOB3]], i32 [[TMP1]])
10167 // CHECK10-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
10168 // CHECK10: omp.inner.for.inc:
10169 // CHECK10-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
10170 // CHECK10-NEXT: [[TMP13:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4
10171 // CHECK10-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP12]], [[TMP13]]
10172 // CHECK10-NEXT: store i32 [[ADD]], i32* [[DOTOMP_IV]], align 4
10173 // CHECK10-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP29:![0-9]+]]
10174 // CHECK10: omp.inner.for.end:
10175 // CHECK10-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
10176 // CHECK10: omp.loop.exit:
10177 // CHECK10-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP1]])
10178 // CHECK10-NEXT: [[TMP14:%.*]] = load i32, i32* [[DOTOMP_IS_LAST]], align 4
10179 // CHECK10-NEXT: [[TMP15:%.*]] = icmp ne i32 [[TMP14]], 0
10180 // CHECK10-NEXT: br i1 [[TMP15]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
10181 // CHECK10: .omp.final.then:
10182 // CHECK10-NEXT: store i32 100, i32* [[I]], align 4
10183 // CHECK10-NEXT: br label [[DOTOMP_FINAL_DONE]]
10184 // CHECK10: .omp.final.done:
10185 // CHECK10-NEXT: ret void
10186 //
10187 //
10188 // CHECK10-LABEL: define {{[^@]+}}@.omp_outlined..13
10189 // CHECK10-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[DOTPREVIOUS_LB_:%.*]], i64 [[DOTPREVIOUS_UB_:%.*]]) #[[ATTR1]] {
10190 // CHECK10-NEXT: entry:
10191 // CHECK10-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
10192 // CHECK10-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
10193 // CHECK10-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i64, align 8
10194 // CHECK10-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i64, align 8
10195 // CHECK10-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
10196 // CHECK10-NEXT: [[TMP:%.*]] = alloca i32, align 4
10197 // CHECK10-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4
10198 // CHECK10-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4
10199 // CHECK10-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
10200 // CHECK10-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
10201 // CHECK10-NEXT: [[I:%.*]] = alloca i32, align 4
10202 // CHECK10-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
10203 // CHECK10-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
10204 // CHECK10-NEXT: store i64 [[DOTPREVIOUS_LB_]], i64* [[DOTPREVIOUS_LB__ADDR]], align 8
10205 // CHECK10-NEXT: store i64 [[DOTPREVIOUS_UB_]], i64* [[DOTPREVIOUS_UB__ADDR]], align 8
10206 // CHECK10-NEXT: store i32 0, i32* [[DOTOMP_LB]], align 4
10207 // CHECK10-NEXT: store i32 99, i32* [[DOTOMP_UB]], align 4
10208 // CHECK10-NEXT: [[TMP0:%.*]] = load i64, i64* [[DOTPREVIOUS_LB__ADDR]], align 8
10209 // CHECK10-NEXT: [[CONV:%.*]] = trunc i64 [[TMP0]] to i32
10210 // CHECK10-NEXT: [[TMP1:%.*]] = load i64, i64* [[DOTPREVIOUS_UB__ADDR]], align 8
10211 // CHECK10-NEXT: [[CONV1:%.*]] = trunc i64 [[TMP1]] to i32
10212 // CHECK10-NEXT: store i32 [[CONV]], i32* [[DOTOMP_LB]], align 4
10213 // CHECK10-NEXT: store i32 [[CONV1]], i32* [[DOTOMP_UB]], align 4
10214 // CHECK10-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4
10215 // CHECK10-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4
10216 // CHECK10-NEXT: [[TMP2:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
10217 // CHECK10-NEXT: [[TMP3:%.*]] = load i32, i32* [[TMP2]], align 4
10218 // CHECK10-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB2]], i32 [[TMP3]], i32 34, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_LB]], i32* [[DOTOMP_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1)
10219 // CHECK10-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
10220 // CHECK10-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP4]], 99
10221 // CHECK10-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
10222 // CHECK10: cond.true:
10223 // CHECK10-NEXT: br label [[COND_END:%.*]]
10224 // CHECK10: cond.false:
10225 // CHECK10-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
10226 // CHECK10-NEXT: br label [[COND_END]]
10227 // CHECK10: cond.end:
10228 // CHECK10-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP5]], [[COND_FALSE]] ]
10229 // CHECK10-NEXT: store i32 [[COND]], i32* [[DOTOMP_UB]], align 4
10230 // CHECK10-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4
10231 // CHECK10-NEXT: store i32 [[TMP6]], i32* [[DOTOMP_IV]], align 4
10232 // CHECK10-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
10233 // CHECK10: omp.inner.for.cond:
10234 // CHECK10-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
10235 // CHECK10-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
10236 // CHECK10-NEXT: [[CMP2:%.*]] = icmp sle i32 [[TMP7]], [[TMP8]]
10237 // CHECK10-NEXT: br i1 [[CMP2]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
10238 // CHECK10: omp.inner.for.body:
10239 // CHECK10-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
10240 // CHECK10-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP9]], 1
10241 // CHECK10-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
10242 // CHECK10-NEXT: store i32 [[ADD]], i32* [[I]], align 4
10243 // CHECK10-NEXT: call void @_Z3fn2v()
10244 // CHECK10-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
10245 // CHECK10: omp.body.continue:
10246 // CHECK10-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
10247 // CHECK10: omp.inner.for.inc:
10248 // CHECK10-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
10249 // CHECK10-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP10]], 1
10250 // CHECK10-NEXT: store i32 [[ADD3]], i32* [[DOTOMP_IV]], align 4
10251 // CHECK10-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP30:![0-9]+]]
10252 // CHECK10: omp.inner.for.end:
10253 // CHECK10-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
10254 // CHECK10: omp.loop.exit:
10255 // CHECK10-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP3]])
10256 // CHECK10-NEXT: [[TMP11:%.*]] = load i32, i32* [[DOTOMP_IS_LAST]], align 4
10257 // CHECK10-NEXT: [[TMP12:%.*]] = icmp ne i32 [[TMP11]], 0
10258 // CHECK10-NEXT: br i1 [[TMP12]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
10259 // CHECK10: .omp.final.then:
10260 // CHECK10-NEXT: store i32 100, i32* [[I]], align 4
10261 // CHECK10-NEXT: br label [[DOTOMP_FINAL_DONE]]
10262 // CHECK10: .omp.final.done:
10263 // CHECK10-NEXT: ret void
10264 //
10265 //
10266 // CHECK10-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiEiT__l67
10267 // CHECK10-SAME: (i64 [[ARG:%.*]]) #[[ATTR1]] {
10268 // CHECK10-NEXT: entry:
10269 // CHECK10-NEXT: [[ARG_ADDR:%.*]] = alloca i64, align 8
10270 // CHECK10-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i8, align 1
10271 // CHECK10-NEXT: [[DOTCAPTURE_EXPR__CASTED:%.*]] = alloca i64, align 8
10272 // CHECK10-NEXT: store i64 [[ARG]], i64* [[ARG_ADDR]], align 8
10273 // CHECK10-NEXT: [[CONV:%.*]] = bitcast i64* [[ARG_ADDR]] to i32*
10274 // CHECK10-NEXT: [[TMP0:%.*]] = load i32, i32* [[CONV]], align 8
10275 // CHECK10-NEXT: [[TOBOOL:%.*]] = icmp ne i32 [[TMP0]], 0
10276 // CHECK10-NEXT: [[FROMBOOL:%.*]] = zext i1 [[TOBOOL]] to i8
10277 // CHECK10-NEXT: store i8 [[FROMBOOL]], i8* [[DOTCAPTURE_EXPR_]], align 1
10278 // CHECK10-NEXT: [[TMP1:%.*]] = load i8, i8* [[DOTCAPTURE_EXPR_]], align 1
10279 // CHECK10-NEXT: [[TOBOOL1:%.*]] = trunc i8 [[TMP1]] to i1
10280 // CHECK10-NEXT: [[CONV2:%.*]] = bitcast i64* [[DOTCAPTURE_EXPR__CASTED]] to i8*
10281 // CHECK10-NEXT: [[FROMBOOL3:%.*]] = zext i1 [[TOBOOL1]] to i8
10282 // CHECK10-NEXT: store i8 [[FROMBOOL3]], i8* [[CONV2]], align 1
10283 // CHECK10-NEXT: [[TMP2:%.*]] = load i64, i64* [[DOTCAPTURE_EXPR__CASTED]], align 8
10284 // CHECK10-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64)* @.omp_outlined..14 to void (i32*, i32*, ...)*), i64 [[TMP2]])
10285 // CHECK10-NEXT: ret void
10286 //
10287 //
10288 // CHECK10-LABEL: define {{[^@]+}}@.omp_outlined..14
10289 // CHECK10-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[DOTCAPTURE_EXPR_:%.*]]) #[[ATTR1]] {
10290 // CHECK10-NEXT: entry:
10291 // CHECK10-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
10292 // CHECK10-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
10293 // CHECK10-NEXT: [[DOTCAPTURE_EXPR__ADDR:%.*]] = alloca i64, align 8
10294 // CHECK10-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
10295 // CHECK10-NEXT: [[TMP:%.*]] = alloca i32, align 4
10296 // CHECK10-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4
10297 // CHECK10-NEXT: [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4
10298 // CHECK10-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
10299 // CHECK10-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
10300 // CHECK10-NEXT: [[I:%.*]] = alloca i32, align 4
10301 // CHECK10-NEXT: [[DOTBOUND_ZERO_ADDR:%.*]] = alloca i32, align 4
10302 // CHECK10-NEXT: store i32 0, i32* [[DOTBOUND_ZERO_ADDR]], align 4
10303 // CHECK10-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
10304 // CHECK10-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
10305 // CHECK10-NEXT: store i64 [[DOTCAPTURE_EXPR_]], i64* [[DOTCAPTURE_EXPR__ADDR]], align 8
10306 // CHECK10-NEXT: [[CONV:%.*]] = bitcast i64* [[DOTCAPTURE_EXPR__ADDR]] to i8*
10307 // CHECK10-NEXT: store i32 0, i32* [[DOTOMP_COMB_LB]], align 4
10308 // CHECK10-NEXT: store i32 99, i32* [[DOTOMP_COMB_UB]], align 4
10309 // CHECK10-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4
10310 // CHECK10-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4
10311 // CHECK10-NEXT: [[TMP0:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
10312 // CHECK10-NEXT: [[TMP1:%.*]] = load i32, i32* [[TMP0]], align 4
10313 // CHECK10-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB1]], i32 [[TMP1]], i32 92, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_COMB_LB]], i32* [[DOTOMP_COMB_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1)
10314 // CHECK10-NEXT: [[TMP2:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
10315 // CHECK10-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP2]], 99
10316 // CHECK10-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
10317 // CHECK10: cond.true:
10318 // CHECK10-NEXT: br label [[COND_END:%.*]]
10319 // CHECK10: cond.false:
10320 // CHECK10-NEXT: [[TMP3:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
10321 // CHECK10-NEXT: br label [[COND_END]]
10322 // CHECK10: cond.end:
10323 // CHECK10-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP3]], [[COND_FALSE]] ]
10324 // CHECK10-NEXT: store i32 [[COND]], i32* [[DOTOMP_COMB_UB]], align 4
10325 // CHECK10-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4
10326 // CHECK10-NEXT: store i32 [[TMP4]], i32* [[DOTOMP_IV]], align 4
10327 // CHECK10-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
10328 // CHECK10: omp.inner.for.cond:
10329 // CHECK10-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
10330 // CHECK10-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
10331 // CHECK10-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP5]], [[TMP6]]
10332 // CHECK10-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
10333 // CHECK10: omp.inner.for.body:
10334 // CHECK10-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4
10335 // CHECK10-NEXT: [[TMP8:%.*]] = zext i32 [[TMP7]] to i64
10336 // CHECK10-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
10337 // CHECK10-NEXT: [[TMP10:%.*]] = zext i32 [[TMP9]] to i64
10338 // CHECK10-NEXT: [[TMP11:%.*]] = load i8, i8* [[CONV]], align 8
10339 // CHECK10-NEXT: [[TOBOOL:%.*]] = trunc i8 [[TMP11]] to i1
10340 // CHECK10-NEXT: br i1 [[TOBOOL]], label [[OMP_IF_THEN:%.*]], label [[OMP_IF_ELSE:%.*]]
10341 // CHECK10: omp_if.then:
10342 // CHECK10-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB3]], i32 2, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64, i64)* @.omp_outlined..15 to void (i32*, i32*, ...)*), i64 [[TMP8]], i64 [[TMP10]])
10343 // CHECK10-NEXT: br label [[OMP_IF_END:%.*]]
10344 // CHECK10: omp_if.else:
10345 // CHECK10-NEXT: call void @__kmpc_serialized_parallel(%struct.ident_t* @[[GLOB3]], i32 [[TMP1]])
10346 // CHECK10-NEXT: [[TMP12:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
10347 // CHECK10-NEXT: call void @.omp_outlined..15(i32* [[TMP12]], i32* [[DOTBOUND_ZERO_ADDR]], i64 [[TMP8]], i64 [[TMP10]]) #[[ATTR2]]
10348 // CHECK10-NEXT: call void @__kmpc_end_serialized_parallel(%struct.ident_t* @[[GLOB3]], i32 [[TMP1]])
10349 // CHECK10-NEXT: br label [[OMP_IF_END]]
10350 // CHECK10: omp_if.end:
10351 // CHECK10-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
10352 // CHECK10: omp.inner.for.inc:
10353 // CHECK10-NEXT: [[TMP13:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
10354 // CHECK10-NEXT: [[TMP14:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4
10355 // CHECK10-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP13]], [[TMP14]]
10356 // CHECK10-NEXT: store i32 [[ADD]], i32* [[DOTOMP_IV]], align 4
10357 // CHECK10-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP31:![0-9]+]]
10358 // CHECK10: omp.inner.for.end:
10359 // CHECK10-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
10360 // CHECK10: omp.loop.exit:
10361 // CHECK10-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP1]])
10362 // CHECK10-NEXT: [[TMP15:%.*]] = load i32, i32* [[DOTOMP_IS_LAST]], align 4
10363 // CHECK10-NEXT: [[TMP16:%.*]] = icmp ne i32 [[TMP15]], 0
10364 // CHECK10-NEXT: br i1 [[TMP16]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
10365 // CHECK10: .omp.final.then:
10366 // CHECK10-NEXT: store i32 100, i32* [[I]], align 4
10367 // CHECK10-NEXT: br label [[DOTOMP_FINAL_DONE]]
10368 // CHECK10: .omp.final.done:
10369 // CHECK10-NEXT: ret void
10370 //
10371 //
10372 // CHECK10-LABEL: define {{[^@]+}}@.omp_outlined..15
10373 // CHECK10-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[DOTPREVIOUS_LB_:%.*]], i64 [[DOTPREVIOUS_UB_:%.*]]) #[[ATTR1]] {
10374 // CHECK10-NEXT: entry:
10375 // CHECK10-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
10376 // CHECK10-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
10377 // CHECK10-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i64, align 8
10378 // CHECK10-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i64, align 8
10379 // CHECK10-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
10380 // CHECK10-NEXT: [[TMP:%.*]] = alloca i32, align 4
10381 // CHECK10-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4
10382 // CHECK10-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4
10383 // CHECK10-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
10384 // CHECK10-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
10385 // CHECK10-NEXT: [[I:%.*]] = alloca i32, align 4
10386 // CHECK10-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
10387 // CHECK10-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
10388 // CHECK10-NEXT: store i64 [[DOTPREVIOUS_LB_]], i64* [[DOTPREVIOUS_LB__ADDR]], align 8
10389 // CHECK10-NEXT: store i64 [[DOTPREVIOUS_UB_]], i64* [[DOTPREVIOUS_UB__ADDR]], align 8
10390 // CHECK10-NEXT: store i32 0, i32* [[DOTOMP_LB]], align 4
10391 // CHECK10-NEXT: store i32 99, i32* [[DOTOMP_UB]], align 4
10392 // CHECK10-NEXT: [[TMP0:%.*]] = load i64, i64* [[DOTPREVIOUS_LB__ADDR]], align 8
10393 // CHECK10-NEXT: [[CONV:%.*]] = trunc i64 [[TMP0]] to i32
10394 // CHECK10-NEXT: [[TMP1:%.*]] = load i64, i64* [[DOTPREVIOUS_UB__ADDR]], align 8
10395 // CHECK10-NEXT: [[CONV1:%.*]] = trunc i64 [[TMP1]] to i32
10396 // CHECK10-NEXT: store i32 [[CONV]], i32* [[DOTOMP_LB]], align 4
10397 // CHECK10-NEXT: store i32 [[CONV1]], i32* [[DOTOMP_UB]], align 4
10398 // CHECK10-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4
10399 // CHECK10-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4
10400 // CHECK10-NEXT: [[TMP2:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
10401 // CHECK10-NEXT: [[TMP3:%.*]] = load i32, i32* [[TMP2]], align 4
10402 // CHECK10-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB2]], i32 [[TMP3]], i32 34, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_LB]], i32* [[DOTOMP_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1)
10403 // CHECK10-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
10404 // CHECK10-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP4]], 99
10405 // CHECK10-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
10406 // CHECK10: cond.true:
10407 // CHECK10-NEXT: br label [[COND_END:%.*]]
10408 // CHECK10: cond.false:
10409 // CHECK10-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
10410 // CHECK10-NEXT: br label [[COND_END]]
10411 // CHECK10: cond.end:
10412 // CHECK10-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP5]], [[COND_FALSE]] ]
10413 // CHECK10-NEXT: store i32 [[COND]], i32* [[DOTOMP_UB]], align 4
10414 // CHECK10-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4
10415 // CHECK10-NEXT: store i32 [[TMP6]], i32* [[DOTOMP_IV]], align 4
10416 // CHECK10-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
10417 // CHECK10: omp.inner.for.cond:
10418 // CHECK10-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
10419 // CHECK10-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
10420 // CHECK10-NEXT: [[CMP2:%.*]] = icmp sle i32 [[TMP7]], [[TMP8]]
10421 // CHECK10-NEXT: br i1 [[CMP2]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
10422 // CHECK10: omp.inner.for.body:
10423 // CHECK10-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
10424 // CHECK10-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP9]], 1
10425 // CHECK10-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
10426 // CHECK10-NEXT: store i32 [[ADD]], i32* [[I]], align 4
10427 // CHECK10-NEXT: call void @_Z3fn3v()
10428 // CHECK10-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
10429 // CHECK10: omp.body.continue:
10430 // CHECK10-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
10431 // CHECK10: omp.inner.for.inc:
10432 // CHECK10-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
10433 // CHECK10-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP10]], 1
10434 // CHECK10-NEXT: store i32 [[ADD3]], i32* [[DOTOMP_IV]], align 4
10435 // CHECK10-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP32:![0-9]+]]
10436 // CHECK10: omp.inner.for.end:
10437 // CHECK10-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
10438 // CHECK10: omp.loop.exit:
10439 // CHECK10-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP3]])
10440 // CHECK10-NEXT: [[TMP11:%.*]] = load i32, i32* [[DOTOMP_IS_LAST]], align 4
10441 // CHECK10-NEXT: [[TMP12:%.*]] = icmp ne i32 [[TMP11]], 0
10442 // CHECK10-NEXT: br i1 [[TMP12]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
10443 // CHECK10: .omp.final.then:
10444 // CHECK10-NEXT: store i32 100, i32* [[I]], align 4
10445 // CHECK10-NEXT: br label [[DOTOMP_FINAL_DONE]]
10446 // CHECK10: .omp.final.done:
10447 // CHECK10-NEXT: ret void
10448 //
10449 //
10450 // CHECK10-LABEL: define {{[^@]+}}@.omp_offloading.requires_reg
10451 // CHECK10-SAME: () #[[ATTR5:[0-9]+]] {
10452 // CHECK10-NEXT: entry:
10453 // CHECK10-NEXT: call void @__tgt_register_requires(i64 1)
10454 // CHECK10-NEXT: ret void
10455 //
10456 //
10457 // CHECK11-LABEL: define {{[^@]+}}@_Z9gtid_testv
10458 // CHECK11-SAME: () #[[ATTR0:[0-9]+]] {
10459 // CHECK11-NEXT: entry:
10460 // CHECK11-NEXT: [[TMP:%.*]] = alloca i32, align 4
10461 // CHECK11-NEXT: [[_TMP1:%.*]] = alloca i32, align 4
10462 // CHECK11-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3:[0-9]+]], i64 -1, i64 100)
10463 // CHECK11-NEXT: [[TMP0:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9gtid_testv_l43.region_id, i32 0, i8** null, i8** null, i64* null, i64* null, i8** null, i8** null, i32 0, i32 0)
10464 // CHECK11-NEXT: [[TMP1:%.*]] = icmp ne i32 [[TMP0]], 0
10465 // CHECK11-NEXT: br i1 [[TMP1]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
10466 // CHECK11: omp_offload.failed:
10467 // CHECK11-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9gtid_testv_l43() #[[ATTR2:[0-9]+]]
10468 // CHECK11-NEXT: br label [[OMP_OFFLOAD_CONT]]
10469 // CHECK11: omp_offload.cont:
10470 // CHECK11-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i64 100)
10471 // CHECK11-NEXT: [[TMP2:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9gtid_testv_l47.region_id, i32 0, i8** null, i8** null, i64* null, i64* null, i8** null, i8** null, i32 0, i32 1)
10472 // CHECK11-NEXT: [[TMP3:%.*]] = icmp ne i32 [[TMP2]], 0
10473 // CHECK11-NEXT: br i1 [[TMP3]], label [[OMP_OFFLOAD_FAILED2:%.*]], label [[OMP_OFFLOAD_CONT3:%.*]]
10474 // CHECK11: omp_offload.failed2:
10475 // CHECK11-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9gtid_testv_l47() #[[ATTR2]]
10476 // CHECK11-NEXT: br label [[OMP_OFFLOAD_CONT3]]
10477 // CHECK11: omp_offload.cont3:
10478 // CHECK11-NEXT: ret void
10479 //
10480 //
10481 // CHECK11-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9gtid_testv_l43
10482 // CHECK11-SAME: () #[[ATTR1:[0-9]+]] {
10483 // CHECK11-NEXT: entry:
10484 // CHECK11-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 0, void (i32*, i32*, ...)* bitcast (void (i32*, i32*)* @.omp_outlined. to void (i32*, i32*, ...)*))
10485 // CHECK11-NEXT: ret void
10486 //
10487 //
10488 // CHECK11-LABEL: define {{[^@]+}}@.omp_outlined.
10489 // CHECK11-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]]) #[[ATTR1]] {
10490 // CHECK11-NEXT: entry:
10491 // CHECK11-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
10492 // CHECK11-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
10493 // CHECK11-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
10494 // CHECK11-NEXT: [[TMP:%.*]] = alloca i32, align 4
10495 // CHECK11-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4
10496 // CHECK11-NEXT: [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4
10497 // CHECK11-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
10498 // CHECK11-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
10499 // CHECK11-NEXT: [[I:%.*]] = alloca i32, align 4
10500 // CHECK11-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
10501 // CHECK11-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
10502 // CHECK11-NEXT: store i32 0, i32* [[DOTOMP_COMB_LB]], align 4
10503 // CHECK11-NEXT: store i32 99, i32* [[DOTOMP_COMB_UB]], align 4
10504 // CHECK11-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4
10505 // CHECK11-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4
10506 // CHECK11-NEXT: [[TMP0:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
10507 // CHECK11-NEXT: [[TMP1:%.*]] = load i32, i32* [[TMP0]], align 4
10508 // CHECK11-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB1:[0-9]+]], i32 [[TMP1]], i32 92, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_COMB_LB]], i32* [[DOTOMP_COMB_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1)
10509 // CHECK11-NEXT: [[TMP2:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
10510 // CHECK11-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP2]], 99
10511 // CHECK11-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
10512 // CHECK11: cond.true:
10513 // CHECK11-NEXT: br label [[COND_END:%.*]]
10514 // CHECK11: cond.false:
10515 // CHECK11-NEXT: [[TMP3:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
10516 // CHECK11-NEXT: br label [[COND_END]]
10517 // CHECK11: cond.end:
10518 // CHECK11-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP3]], [[COND_FALSE]] ]
10519 // CHECK11-NEXT: store i32 [[COND]], i32* [[DOTOMP_COMB_UB]], align 4
10520 // CHECK11-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4
10521 // CHECK11-NEXT: store i32 [[TMP4]], i32* [[DOTOMP_IV]], align 4
10522 // CHECK11-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
10523 // CHECK11: omp.inner.for.cond:
10524 // CHECK11-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
10525 // CHECK11-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
10526 // CHECK11-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP5]], [[TMP6]]
10527 // CHECK11-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
10528 // CHECK11: omp.inner.for.body:
10529 // CHECK11-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4
10530 // CHECK11-NEXT: [[TMP8:%.*]] = zext i32 [[TMP7]] to i64
10531 // CHECK11-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
10532 // CHECK11-NEXT: [[TMP10:%.*]] = zext i32 [[TMP9]] to i64
10533 // CHECK11-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB3]], i32 2, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64, i64)* @.omp_outlined..1 to void (i32*, i32*, ...)*), i64 [[TMP8]], i64 [[TMP10]])
10534 // CHECK11-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
10535 // CHECK11: omp.inner.for.inc:
10536 // CHECK11-NEXT: [[TMP11:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
10537 // CHECK11-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4
10538 // CHECK11-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP11]], [[TMP12]]
10539 // CHECK11-NEXT: store i32 [[ADD]], i32* [[DOTOMP_IV]], align 4
10540 // CHECK11-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP14:![0-9]+]]
10541 // CHECK11: omp.inner.for.end:
10542 // CHECK11-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
10543 // CHECK11: omp.loop.exit:
10544 // CHECK11-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP1]])
10545 // CHECK11-NEXT: [[TMP13:%.*]] = load i32, i32* [[DOTOMP_IS_LAST]], align 4
10546 // CHECK11-NEXT: [[TMP14:%.*]] = icmp ne i32 [[TMP13]], 0
10547 // CHECK11-NEXT: br i1 [[TMP14]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
10548 // CHECK11: .omp.final.then:
10549 // CHECK11-NEXT: store i32 100, i32* [[I]], align 4
10550 // CHECK11-NEXT: br label [[DOTOMP_FINAL_DONE]]
10551 // CHECK11: .omp.final.done:
10552 // CHECK11-NEXT: ret void
10553 //
10554 //
10555 // CHECK11-LABEL: define {{[^@]+}}@.omp_outlined..1
10556 // CHECK11-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[DOTPREVIOUS_LB_:%.*]], i64 [[DOTPREVIOUS_UB_:%.*]]) #[[ATTR1]] {
10557 // CHECK11-NEXT: entry:
10558 // CHECK11-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
10559 // CHECK11-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
10560 // CHECK11-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i64, align 8
10561 // CHECK11-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i64, align 8
10562 // CHECK11-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
10563 // CHECK11-NEXT: [[TMP:%.*]] = alloca i32, align 4
10564 // CHECK11-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4
10565 // CHECK11-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4
10566 // CHECK11-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
10567 // CHECK11-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
10568 // CHECK11-NEXT: [[I:%.*]] = alloca i32, align 4
10569 // CHECK11-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
10570 // CHECK11-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
10571 // CHECK11-NEXT: store i64 [[DOTPREVIOUS_LB_]], i64* [[DOTPREVIOUS_LB__ADDR]], align 8
10572 // CHECK11-NEXT: store i64 [[DOTPREVIOUS_UB_]], i64* [[DOTPREVIOUS_UB__ADDR]], align 8
10573 // CHECK11-NEXT: store i32 0, i32* [[DOTOMP_LB]], align 4
10574 // CHECK11-NEXT: store i32 99, i32* [[DOTOMP_UB]], align 4
10575 // CHECK11-NEXT: [[TMP0:%.*]] = load i64, i64* [[DOTPREVIOUS_LB__ADDR]], align 8
10576 // CHECK11-NEXT: [[CONV:%.*]] = trunc i64 [[TMP0]] to i32
10577 // CHECK11-NEXT: [[TMP1:%.*]] = load i64, i64* [[DOTPREVIOUS_UB__ADDR]], align 8
10578 // CHECK11-NEXT: [[CONV1:%.*]] = trunc i64 [[TMP1]] to i32
10579 // CHECK11-NEXT: store i32 [[CONV]], i32* [[DOTOMP_LB]], align 4
10580 // CHECK11-NEXT: store i32 [[CONV1]], i32* [[DOTOMP_UB]], align 4
10581 // CHECK11-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4
10582 // CHECK11-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4
10583 // CHECK11-NEXT: [[TMP2:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
10584 // CHECK11-NEXT: [[TMP3:%.*]] = load i32, i32* [[TMP2]], align 4
10585 // CHECK11-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB2:[0-9]+]], i32 [[TMP3]], i32 34, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_LB]], i32* [[DOTOMP_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1)
10586 // CHECK11-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
10587 // CHECK11-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP4]], 99
10588 // CHECK11-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
10589 // CHECK11: cond.true:
10590 // CHECK11-NEXT: br label [[COND_END:%.*]]
10591 // CHECK11: cond.false:
10592 // CHECK11-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
10593 // CHECK11-NEXT: br label [[COND_END]]
10594 // CHECK11: cond.end:
10595 // CHECK11-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP5]], [[COND_FALSE]] ]
10596 // CHECK11-NEXT: store i32 [[COND]], i32* [[DOTOMP_UB]], align 4
10597 // CHECK11-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4
10598 // CHECK11-NEXT: store i32 [[TMP6]], i32* [[DOTOMP_IV]], align 4
10599 // CHECK11-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
10600 // CHECK11: omp.inner.for.cond:
10601 // CHECK11-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
10602 // CHECK11-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
10603 // CHECK11-NEXT: [[CMP2:%.*]] = icmp sle i32 [[TMP7]], [[TMP8]]
10604 // CHECK11-NEXT: br i1 [[CMP2]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
10605 // CHECK11: omp.inner.for.body:
10606 // CHECK11-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
10607 // CHECK11-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP9]], 1
10608 // CHECK11-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
10609 // CHECK11-NEXT: store i32 [[ADD]], i32* [[I]], align 4
10610 // CHECK11-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
10611 // CHECK11: omp.body.continue:
10612 // CHECK11-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
10613 // CHECK11: omp.inner.for.inc:
10614 // CHECK11-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
10615 // CHECK11-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP10]], 1
10616 // CHECK11-NEXT: store i32 [[ADD3]], i32* [[DOTOMP_IV]], align 4
10617 // CHECK11-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP16:![0-9]+]]
10618 // CHECK11: omp.inner.for.end:
10619 // CHECK11-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
10620 // CHECK11: omp.loop.exit:
10621 // CHECK11-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP3]])
10622 // CHECK11-NEXT: [[TMP11:%.*]] = load i32, i32* [[DOTOMP_IS_LAST]], align 4
10623 // CHECK11-NEXT: [[TMP12:%.*]] = icmp ne i32 [[TMP11]], 0
10624 // CHECK11-NEXT: br i1 [[TMP12]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
10625 // CHECK11: .omp.final.then:
10626 // CHECK11-NEXT: store i32 100, i32* [[I]], align 4
10627 // CHECK11-NEXT: br label [[DOTOMP_FINAL_DONE]]
10628 // CHECK11: .omp.final.done:
10629 // CHECK11-NEXT: ret void
10630 //
10631 //
10632 // CHECK11-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9gtid_testv_l47
10633 // CHECK11-SAME: () #[[ATTR1]] {
10634 // CHECK11-NEXT: entry:
10635 // CHECK11-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 0, void (i32*, i32*, ...)* bitcast (void (i32*, i32*)* @.omp_outlined..2 to void (i32*, i32*, ...)*))
10636 // CHECK11-NEXT: ret void
10637 //
10638 //
10639 // CHECK11-LABEL: define {{[^@]+}}@.omp_outlined..2
10640 // CHECK11-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]]) #[[ATTR1]] {
10641 // CHECK11-NEXT: entry:
10642 // CHECK11-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
10643 // CHECK11-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
10644 // CHECK11-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
10645 // CHECK11-NEXT: [[TMP:%.*]] = alloca i32, align 4
10646 // CHECK11-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4
10647 // CHECK11-NEXT: [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4
10648 // CHECK11-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
10649 // CHECK11-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
10650 // CHECK11-NEXT: [[I:%.*]] = alloca i32, align 4
10651 // CHECK11-NEXT: [[DOTBOUND_ZERO_ADDR:%.*]] = alloca i32, align 4
10652 // CHECK11-NEXT: store i32 0, i32* [[DOTBOUND_ZERO_ADDR]], align 4
10653 // CHECK11-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
10654 // CHECK11-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
10655 // CHECK11-NEXT: store i32 0, i32* [[DOTOMP_COMB_LB]], align 4
10656 // CHECK11-NEXT: store i32 99, i32* [[DOTOMP_COMB_UB]], align 4
10657 // CHECK11-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4
10658 // CHECK11-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4
10659 // CHECK11-NEXT: [[TMP0:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
10660 // CHECK11-NEXT: [[TMP1:%.*]] = load i32, i32* [[TMP0]], align 4
10661 // CHECK11-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB1]], i32 [[TMP1]], i32 92, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_COMB_LB]], i32* [[DOTOMP_COMB_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1)
10662 // CHECK11-NEXT: [[TMP2:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
10663 // CHECK11-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP2]], 99
10664 // CHECK11-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
10665 // CHECK11: cond.true:
10666 // CHECK11-NEXT: br label [[COND_END:%.*]]
10667 // CHECK11: cond.false:
10668 // CHECK11-NEXT: [[TMP3:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
10669 // CHECK11-NEXT: br label [[COND_END]]
10670 // CHECK11: cond.end:
10671 // CHECK11-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP3]], [[COND_FALSE]] ]
10672 // CHECK11-NEXT: store i32 [[COND]], i32* [[DOTOMP_COMB_UB]], align 4
10673 // CHECK11-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4
10674 // CHECK11-NEXT: store i32 [[TMP4]], i32* [[DOTOMP_IV]], align 4
10675 // CHECK11-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
10676 // CHECK11: omp.inner.for.cond:
10677 // CHECK11-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
10678 // CHECK11-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
10679 // CHECK11-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP5]], [[TMP6]]
10680 // CHECK11-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
10681 // CHECK11: omp.inner.for.body:
10682 // CHECK11-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4
10683 // CHECK11-NEXT: [[TMP8:%.*]] = zext i32 [[TMP7]] to i64
10684 // CHECK11-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
10685 // CHECK11-NEXT: [[TMP10:%.*]] = zext i32 [[TMP9]] to i64
10686 // CHECK11-NEXT: call void @__kmpc_serialized_parallel(%struct.ident_t* @[[GLOB3]], i32 [[TMP1]])
10687 // CHECK11-NEXT: [[TMP11:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
10688 // CHECK11-NEXT: call void @.omp_outlined..3(i32* [[TMP11]], i32* [[DOTBOUND_ZERO_ADDR]], i64 [[TMP8]], i64 [[TMP10]]) #[[ATTR2]]
10689 // CHECK11-NEXT: call void @__kmpc_end_serialized_parallel(%struct.ident_t* @[[GLOB3]], i32 [[TMP1]])
10690 // CHECK11-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
10691 // CHECK11: omp.inner.for.inc:
10692 // CHECK11-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
10693 // CHECK11-NEXT: [[TMP13:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4
10694 // CHECK11-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP12]], [[TMP13]]
10695 // CHECK11-NEXT: store i32 [[ADD]], i32* [[DOTOMP_IV]], align 4
10696 // CHECK11-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP19:![0-9]+]]
10697 // CHECK11: omp.inner.for.end:
10698 // CHECK11-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
10699 // CHECK11: omp.loop.exit:
10700 // CHECK11-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP1]])
10701 // CHECK11-NEXT: [[TMP14:%.*]] = load i32, i32* [[DOTOMP_IS_LAST]], align 4
10702 // CHECK11-NEXT: [[TMP15:%.*]] = icmp ne i32 [[TMP14]], 0
10703 // CHECK11-NEXT: br i1 [[TMP15]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
10704 // CHECK11: .omp.final.then:
10705 // CHECK11-NEXT: store i32 100, i32* [[I]], align 4
10706 // CHECK11-NEXT: br label [[DOTOMP_FINAL_DONE]]
10707 // CHECK11: .omp.final.done:
10708 // CHECK11-NEXT: ret void
10709 //
10710 //
10711 // CHECK11-LABEL: define {{[^@]+}}@.omp_outlined..3
10712 // CHECK11-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[DOTPREVIOUS_LB_:%.*]], i64 [[DOTPREVIOUS_UB_:%.*]]) #[[ATTR1]] {
10713 // CHECK11-NEXT: entry:
10714 // CHECK11-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
10715 // CHECK11-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
10716 // CHECK11-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i64, align 8
10717 // CHECK11-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i64, align 8
10718 // CHECK11-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
10719 // CHECK11-NEXT: [[TMP:%.*]] = alloca i32, align 4
10720 // CHECK11-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4
10721 // CHECK11-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4
10722 // CHECK11-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
10723 // CHECK11-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
10724 // CHECK11-NEXT: [[I:%.*]] = alloca i32, align 4
10725 // CHECK11-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
10726 // CHECK11-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
10727 // CHECK11-NEXT: store i64 [[DOTPREVIOUS_LB_]], i64* [[DOTPREVIOUS_LB__ADDR]], align 8
10728 // CHECK11-NEXT: store i64 [[DOTPREVIOUS_UB_]], i64* [[DOTPREVIOUS_UB__ADDR]], align 8
10729 // CHECK11-NEXT: store i32 0, i32* [[DOTOMP_LB]], align 4
10730 // CHECK11-NEXT: store i32 99, i32* [[DOTOMP_UB]], align 4
10731 // CHECK11-NEXT: [[TMP0:%.*]] = load i64, i64* [[DOTPREVIOUS_LB__ADDR]], align 8
10732 // CHECK11-NEXT: [[CONV:%.*]] = trunc i64 [[TMP0]] to i32
10733 // CHECK11-NEXT: [[TMP1:%.*]] = load i64, i64* [[DOTPREVIOUS_UB__ADDR]], align 8
10734 // CHECK11-NEXT: [[CONV1:%.*]] = trunc i64 [[TMP1]] to i32
10735 // CHECK11-NEXT: store i32 [[CONV]], i32* [[DOTOMP_LB]], align 4
10736 // CHECK11-NEXT: store i32 [[CONV1]], i32* [[DOTOMP_UB]], align 4
10737 // CHECK11-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4
10738 // CHECK11-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4
10739 // CHECK11-NEXT: [[TMP2:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
10740 // CHECK11-NEXT: [[TMP3:%.*]] = load i32, i32* [[TMP2]], align 4
10741 // CHECK11-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB2]], i32 [[TMP3]], i32 34, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_LB]], i32* [[DOTOMP_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1)
10742 // CHECK11-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
10743 // CHECK11-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP4]], 99
10744 // CHECK11-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
10745 // CHECK11: cond.true:
10746 // CHECK11-NEXT: br label [[COND_END:%.*]]
10747 // CHECK11: cond.false:
10748 // CHECK11-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
10749 // CHECK11-NEXT: br label [[COND_END]]
10750 // CHECK11: cond.end:
10751 // CHECK11-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP5]], [[COND_FALSE]] ]
10752 // CHECK11-NEXT: store i32 [[COND]], i32* [[DOTOMP_UB]], align 4
10753 // CHECK11-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4
10754 // CHECK11-NEXT: store i32 [[TMP6]], i32* [[DOTOMP_IV]], align 4
10755 // CHECK11-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
10756 // CHECK11: omp.inner.for.cond:
10757 // CHECK11-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
10758 // CHECK11-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
10759 // CHECK11-NEXT: [[CMP2:%.*]] = icmp sle i32 [[TMP7]], [[TMP8]]
10760 // CHECK11-NEXT: br i1 [[CMP2]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
10761 // CHECK11: omp.inner.for.body:
10762 // CHECK11-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
10763 // CHECK11-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP9]], 1
10764 // CHECK11-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
10765 // CHECK11-NEXT: store i32 [[ADD]], i32* [[I]], align 4
10766 // CHECK11-NEXT: call void @_Z9gtid_testv()
10767 // CHECK11-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
10768 // CHECK11: omp.body.continue:
10769 // CHECK11-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
10770 // CHECK11: omp.inner.for.inc:
10771 // CHECK11-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
10772 // CHECK11-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP10]], 1
10773 // CHECK11-NEXT: store i32 [[ADD3]], i32* [[DOTOMP_IV]], align 4
10774 // CHECK11-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP20:![0-9]+]]
10775 // CHECK11: omp.inner.for.end:
10776 // CHECK11-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
10777 // CHECK11: omp.loop.exit:
10778 // CHECK11-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP3]])
10779 // CHECK11-NEXT: [[TMP11:%.*]] = load i32, i32* [[DOTOMP_IS_LAST]], align 4
10780 // CHECK11-NEXT: [[TMP12:%.*]] = icmp ne i32 [[TMP11]], 0
10781 // CHECK11-NEXT: br i1 [[TMP12]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
10782 // CHECK11: .omp.final.then:
10783 // CHECK11-NEXT: store i32 100, i32* [[I]], align 4
10784 // CHECK11-NEXT: br label [[DOTOMP_FINAL_DONE]]
10785 // CHECK11: .omp.final.done:
10786 // CHECK11-NEXT: ret void
10787 //
10788 //
10789 // CHECK11-LABEL: define {{[^@]+}}@main
10790 // CHECK11-SAME: () #[[ATTR3:[0-9]+]] {
10791 // CHECK11-NEXT: entry:
10792 // CHECK11-NEXT: [[RETVAL:%.*]] = alloca i32, align 4
10793 // CHECK11-NEXT: [[TMP:%.*]] = alloca i32, align 4
10794 // CHECK11-NEXT: [[_TMP1:%.*]] = alloca i32, align 4
10795 // CHECK11-NEXT: [[ARG_CASTED:%.*]] = alloca i64, align 8
10796 // CHECK11-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [1 x i8*], align 8
10797 // CHECK11-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [1 x i8*], align 8
10798 // CHECK11-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [1 x i8*], align 8
10799 // CHECK11-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i8, align 1
10800 // CHECK11-NEXT: [[_TMP5:%.*]] = alloca i32, align 4
10801 // CHECK11-NEXT: store i32 0, i32* [[RETVAL]], align 4
10802 // CHECK11-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i64 100)
10803 // CHECK11-NEXT: [[TMP0:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l76.region_id, i32 0, i8** null, i8** null, i64* null, i64* null, i8** null, i8** null, i32 0, i32 0)
10804 // CHECK11-NEXT: [[TMP1:%.*]] = icmp ne i32 [[TMP0]], 0
10805 // CHECK11-NEXT: br i1 [[TMP1]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
10806 // CHECK11: omp_offload.failed:
10807 // CHECK11-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l76() #[[ATTR2]]
10808 // CHECK11-NEXT: br label [[OMP_OFFLOAD_CONT]]
10809 // CHECK11: omp_offload.cont:
10810 // CHECK11-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i64 100)
10811 // CHECK11-NEXT: [[TMP2:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l84.region_id, i32 0, i8** null, i8** null, i64* null, i64* null, i8** null, i8** null, i32 0, i32 1)
10812 // CHECK11-NEXT: [[TMP3:%.*]] = icmp ne i32 [[TMP2]], 0
10813 // CHECK11-NEXT: br i1 [[TMP3]], label [[OMP_OFFLOAD_FAILED2:%.*]], label [[OMP_OFFLOAD_CONT3:%.*]]
10814 // CHECK11: omp_offload.failed2:
10815 // CHECK11-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l84() #[[ATTR2]]
10816 // CHECK11-NEXT: br label [[OMP_OFFLOAD_CONT3]]
10817 // CHECK11: omp_offload.cont3:
10818 // CHECK11-NEXT: [[TMP4:%.*]] = load i32, i32* @Arg, align 4
10819 // CHECK11-NEXT: [[CONV:%.*]] = bitcast i64* [[ARG_CASTED]] to i32*
10820 // CHECK11-NEXT: store i32 [[TMP4]], i32* [[CONV]], align 4
10821 // CHECK11-NEXT: [[TMP5:%.*]] = load i64, i64* [[ARG_CASTED]], align 8
10822 // CHECK11-NEXT: [[TMP6:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
10823 // CHECK11-NEXT: [[TMP7:%.*]] = bitcast i8** [[TMP6]] to i64*
10824 // CHECK11-NEXT: store i64 [[TMP5]], i64* [[TMP7]], align 8
10825 // CHECK11-NEXT: [[TMP8:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
10826 // CHECK11-NEXT: [[TMP9:%.*]] = bitcast i8** [[TMP8]] to i64*
10827 // CHECK11-NEXT: store i64 [[TMP5]], i64* [[TMP9]], align 8
10828 // CHECK11-NEXT: [[TMP10:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0
10829 // CHECK11-NEXT: store i8* null, i8** [[TMP10]], align 8
10830 // CHECK11-NEXT: [[TMP11:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
10831 // CHECK11-NEXT: [[TMP12:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
10832 // CHECK11-NEXT: [[TMP13:%.*]] = load i32, i32* @Arg, align 4
10833 // CHECK11-NEXT: [[TOBOOL:%.*]] = icmp ne i32 [[TMP13]], 0
10834 // CHECK11-NEXT: [[FROMBOOL:%.*]] = zext i1 [[TOBOOL]] to i8
10835 // CHECK11-NEXT: store i8 [[FROMBOOL]], i8* [[DOTCAPTURE_EXPR_]], align 1
10836 // CHECK11-NEXT: [[TMP14:%.*]] = load i8, i8* [[DOTCAPTURE_EXPR_]], align 1
10837 // CHECK11-NEXT: [[TOBOOL4:%.*]] = trunc i8 [[TMP14]] to i1
10838 // CHECK11-NEXT: [[TMP15:%.*]] = select i1 [[TOBOOL4]], i32 0, i32 1
10839 // CHECK11-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i64 100)
10840 // CHECK11-NEXT: [[TMP16:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l92.region_id, i32 1, i8** [[TMP11]], i8** [[TMP12]], i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_sizes, i32 0, i32 0), i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_maptypes, i32 0, i32 0), i8** null, i8** null, i32 0, i32 [[TMP15]])
10841 // CHECK11-NEXT: [[TMP17:%.*]] = icmp ne i32 [[TMP16]], 0
10842 // CHECK11-NEXT: br i1 [[TMP17]], label [[OMP_OFFLOAD_FAILED6:%.*]], label [[OMP_OFFLOAD_CONT7:%.*]]
10843 // CHECK11: omp_offload.failed6:
10844 // CHECK11-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l92(i64 [[TMP5]]) #[[ATTR2]]
10845 // CHECK11-NEXT: br label [[OMP_OFFLOAD_CONT7]]
10846 // CHECK11: omp_offload.cont7:
10847 // CHECK11-NEXT: [[TMP18:%.*]] = load i32, i32* @Arg, align 4
10848 // CHECK11-NEXT: [[CALL:%.*]] = call i32 @_Z5tmainIiEiT_(i32 [[TMP18]])
10849 // CHECK11-NEXT: ret i32 [[CALL]]
10850 //
10851 //
10852 // CHECK11-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l76
10853 // CHECK11-SAME: () #[[ATTR1]] {
10854 // CHECK11-NEXT: entry:
10855 // CHECK11-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 0, void (i32*, i32*, ...)* bitcast (void (i32*, i32*)* @.omp_outlined..4 to void (i32*, i32*, ...)*))
10856 // CHECK11-NEXT: ret void
10857 //
10858 //
10859 // CHECK11-LABEL: define {{[^@]+}}@.omp_outlined..4
10860 // CHECK11-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]]) #[[ATTR1]] {
10861 // CHECK11-NEXT: entry:
10862 // CHECK11-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
10863 // CHECK11-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
10864 // CHECK11-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
10865 // CHECK11-NEXT: [[TMP:%.*]] = alloca i32, align 4
10866 // CHECK11-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4
10867 // CHECK11-NEXT: [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4
10868 // CHECK11-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
10869 // CHECK11-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
10870 // CHECK11-NEXT: [[I:%.*]] = alloca i32, align 4
10871 // CHECK11-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
10872 // CHECK11-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
10873 // CHECK11-NEXT: store i32 0, i32* [[DOTOMP_COMB_LB]], align 4
10874 // CHECK11-NEXT: store i32 99, i32* [[DOTOMP_COMB_UB]], align 4
10875 // CHECK11-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4
10876 // CHECK11-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4
10877 // CHECK11-NEXT: [[TMP0:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
10878 // CHECK11-NEXT: [[TMP1:%.*]] = load i32, i32* [[TMP0]], align 4
10879 // CHECK11-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB1]], i32 [[TMP1]], i32 92, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_COMB_LB]], i32* [[DOTOMP_COMB_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1)
10880 // CHECK11-NEXT: [[TMP2:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
10881 // CHECK11-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP2]], 99
10882 // CHECK11-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
10883 // CHECK11: cond.true:
10884 // CHECK11-NEXT: br label [[COND_END:%.*]]
10885 // CHECK11: cond.false:
10886 // CHECK11-NEXT: [[TMP3:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
10887 // CHECK11-NEXT: br label [[COND_END]]
10888 // CHECK11: cond.end:
10889 // CHECK11-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP3]], [[COND_FALSE]] ]
10890 // CHECK11-NEXT: store i32 [[COND]], i32* [[DOTOMP_COMB_UB]], align 4
10891 // CHECK11-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4
10892 // CHECK11-NEXT: store i32 [[TMP4]], i32* [[DOTOMP_IV]], align 4
10893 // CHECK11-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
10894 // CHECK11: omp.inner.for.cond:
10895 // CHECK11-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
10896 // CHECK11-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
10897 // CHECK11-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP5]], [[TMP6]]
10898 // CHECK11-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
10899 // CHECK11: omp.inner.for.body:
10900 // CHECK11-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4
10901 // CHECK11-NEXT: [[TMP8:%.*]] = zext i32 [[TMP7]] to i64
10902 // CHECK11-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
10903 // CHECK11-NEXT: [[TMP10:%.*]] = zext i32 [[TMP9]] to i64
10904 // CHECK11-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB3]], i32 2, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64, i64)* @.omp_outlined..5 to void (i32*, i32*, ...)*), i64 [[TMP8]], i64 [[TMP10]])
10905 // CHECK11-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
10906 // CHECK11: omp.inner.for.inc:
10907 // CHECK11-NEXT: [[TMP11:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
10908 // CHECK11-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4
10909 // CHECK11-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP11]], [[TMP12]]
10910 // CHECK11-NEXT: store i32 [[ADD]], i32* [[DOTOMP_IV]], align 4
10911 // CHECK11-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP21:![0-9]+]]
10912 // CHECK11: omp.inner.for.end:
10913 // CHECK11-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
10914 // CHECK11: omp.loop.exit:
10915 // CHECK11-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP1]])
10916 // CHECK11-NEXT: [[TMP13:%.*]] = load i32, i32* [[DOTOMP_IS_LAST]], align 4
10917 // CHECK11-NEXT: [[TMP14:%.*]] = icmp ne i32 [[TMP13]], 0
10918 // CHECK11-NEXT: br i1 [[TMP14]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
10919 // CHECK11: .omp.final.then:
10920 // CHECK11-NEXT: store i32 100, i32* [[I]], align 4
10921 // CHECK11-NEXT: br label [[DOTOMP_FINAL_DONE]]
10922 // CHECK11: .omp.final.done:
10923 // CHECK11-NEXT: ret void
10924 //
10925 //
10926 // CHECK11-LABEL: define {{[^@]+}}@.omp_outlined..5
10927 // CHECK11-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[DOTPREVIOUS_LB_:%.*]], i64 [[DOTPREVIOUS_UB_:%.*]]) #[[ATTR1]] {
10928 // CHECK11-NEXT: entry:
10929 // CHECK11-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
10930 // CHECK11-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
10931 // CHECK11-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i64, align 8
10932 // CHECK11-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i64, align 8
10933 // CHECK11-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
10934 // CHECK11-NEXT: [[TMP:%.*]] = alloca i32, align 4
10935 // CHECK11-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4
10936 // CHECK11-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4
10937 // CHECK11-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
10938 // CHECK11-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
10939 // CHECK11-NEXT: [[I:%.*]] = alloca i32, align 4
10940 // CHECK11-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
10941 // CHECK11-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
10942 // CHECK11-NEXT: store i64 [[DOTPREVIOUS_LB_]], i64* [[DOTPREVIOUS_LB__ADDR]], align 8
10943 // CHECK11-NEXT: store i64 [[DOTPREVIOUS_UB_]], i64* [[DOTPREVIOUS_UB__ADDR]], align 8
10944 // CHECK11-NEXT: store i32 0, i32* [[DOTOMP_LB]], align 4
10945 // CHECK11-NEXT: store i32 99, i32* [[DOTOMP_UB]], align 4
10946 // CHECK11-NEXT: [[TMP0:%.*]] = load i64, i64* [[DOTPREVIOUS_LB__ADDR]], align 8
10947 // CHECK11-NEXT: [[CONV:%.*]] = trunc i64 [[TMP0]] to i32
10948 // CHECK11-NEXT: [[TMP1:%.*]] = load i64, i64* [[DOTPREVIOUS_UB__ADDR]], align 8
10949 // CHECK11-NEXT: [[CONV1:%.*]] = trunc i64 [[TMP1]] to i32
10950 // CHECK11-NEXT: store i32 [[CONV]], i32* [[DOTOMP_LB]], align 4
10951 // CHECK11-NEXT: store i32 [[CONV1]], i32* [[DOTOMP_UB]], align 4
10952 // CHECK11-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4
10953 // CHECK11-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4
10954 // CHECK11-NEXT: [[TMP2:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
10955 // CHECK11-NEXT: [[TMP3:%.*]] = load i32, i32* [[TMP2]], align 4
10956 // CHECK11-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB2]], i32 [[TMP3]], i32 34, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_LB]], i32* [[DOTOMP_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1)
10957 // CHECK11-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
10958 // CHECK11-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP4]], 99
10959 // CHECK11-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
10960 // CHECK11: cond.true:
10961 // CHECK11-NEXT: br label [[COND_END:%.*]]
10962 // CHECK11: cond.false:
10963 // CHECK11-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
10964 // CHECK11-NEXT: br label [[COND_END]]
10965 // CHECK11: cond.end:
10966 // CHECK11-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP5]], [[COND_FALSE]] ]
10967 // CHECK11-NEXT: store i32 [[COND]], i32* [[DOTOMP_UB]], align 4
10968 // CHECK11-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4
10969 // CHECK11-NEXT: store i32 [[TMP6]], i32* [[DOTOMP_IV]], align 4
10970 // CHECK11-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
10971 // CHECK11: omp.inner.for.cond:
10972 // CHECK11-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
10973 // CHECK11-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
10974 // CHECK11-NEXT: [[CMP2:%.*]] = icmp sle i32 [[TMP7]], [[TMP8]]
10975 // CHECK11-NEXT: br i1 [[CMP2]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
10976 // CHECK11: omp.inner.for.body:
10977 // CHECK11-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
10978 // CHECK11-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP9]], 1
10979 // CHECK11-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
10980 // CHECK11-NEXT: store i32 [[ADD]], i32* [[I]], align 4
10981 // CHECK11-NEXT: call void @_Z3fn4v()
10982 // CHECK11-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
10983 // CHECK11: omp.body.continue:
10984 // CHECK11-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
10985 // CHECK11: omp.inner.for.inc:
10986 // CHECK11-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
10987 // CHECK11-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP10]], 1
10988 // CHECK11-NEXT: store i32 [[ADD3]], i32* [[DOTOMP_IV]], align 4
10989 // CHECK11-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP22:![0-9]+]]
10990 // CHECK11: omp.inner.for.end:
10991 // CHECK11-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
10992 // CHECK11: omp.loop.exit:
10993 // CHECK11-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP3]])
10994 // CHECK11-NEXT: [[TMP11:%.*]] = load i32, i32* [[DOTOMP_IS_LAST]], align 4
10995 // CHECK11-NEXT: [[TMP12:%.*]] = icmp ne i32 [[TMP11]], 0
10996 // CHECK11-NEXT: br i1 [[TMP12]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
10997 // CHECK11: .omp.final.then:
10998 // CHECK11-NEXT: store i32 100, i32* [[I]], align 4
10999 // CHECK11-NEXT: br label [[DOTOMP_FINAL_DONE]]
11000 // CHECK11: .omp.final.done:
11001 // CHECK11-NEXT: ret void
11002 //
11003 //
11004 // CHECK11-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l84
11005 // CHECK11-SAME: () #[[ATTR1]] {
11006 // CHECK11-NEXT: entry:
11007 // CHECK11-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 0, void (i32*, i32*, ...)* bitcast (void (i32*, i32*)* @.omp_outlined..6 to void (i32*, i32*, ...)*))
11008 // CHECK11-NEXT: ret void
11009 //
11010 //
11011 // CHECK11-LABEL: define {{[^@]+}}@.omp_outlined..6
11012 // CHECK11-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]]) #[[ATTR1]] {
11013 // CHECK11-NEXT: entry:
11014 // CHECK11-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
11015 // CHECK11-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
11016 // CHECK11-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
11017 // CHECK11-NEXT: [[TMP:%.*]] = alloca i32, align 4
11018 // CHECK11-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4
11019 // CHECK11-NEXT: [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4
11020 // CHECK11-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
11021 // CHECK11-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
11022 // CHECK11-NEXT: [[I:%.*]] = alloca i32, align 4
11023 // CHECK11-NEXT: [[DOTBOUND_ZERO_ADDR:%.*]] = alloca i32, align 4
11024 // CHECK11-NEXT: store i32 0, i32* [[DOTBOUND_ZERO_ADDR]], align 4
11025 // CHECK11-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
11026 // CHECK11-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
11027 // CHECK11-NEXT: store i32 0, i32* [[DOTOMP_COMB_LB]], align 4
11028 // CHECK11-NEXT: store i32 99, i32* [[DOTOMP_COMB_UB]], align 4
11029 // CHECK11-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4
11030 // CHECK11-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4
11031 // CHECK11-NEXT: [[TMP0:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
11032 // CHECK11-NEXT: [[TMP1:%.*]] = load i32, i32* [[TMP0]], align 4
11033 // CHECK11-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB1]], i32 [[TMP1]], i32 92, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_COMB_LB]], i32* [[DOTOMP_COMB_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1)
11034 // CHECK11-NEXT: [[TMP2:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
11035 // CHECK11-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP2]], 99
11036 // CHECK11-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
11037 // CHECK11: cond.true:
11038 // CHECK11-NEXT: br label [[COND_END:%.*]]
11039 // CHECK11: cond.false:
11040 // CHECK11-NEXT: [[TMP3:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
11041 // CHECK11-NEXT: br label [[COND_END]]
11042 // CHECK11: cond.end:
11043 // CHECK11-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP3]], [[COND_FALSE]] ]
11044 // CHECK11-NEXT: store i32 [[COND]], i32* [[DOTOMP_COMB_UB]], align 4
11045 // CHECK11-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4
11046 // CHECK11-NEXT: store i32 [[TMP4]], i32* [[DOTOMP_IV]], align 4
11047 // CHECK11-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
11048 // CHECK11: omp.inner.for.cond:
11049 // CHECK11-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
11050 // CHECK11-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
11051 // CHECK11-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP5]], [[TMP6]]
11052 // CHECK11-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
11053 // CHECK11: omp.inner.for.body:
11054 // CHECK11-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4
11055 // CHECK11-NEXT: [[TMP8:%.*]] = zext i32 [[TMP7]] to i64
11056 // CHECK11-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
11057 // CHECK11-NEXT: [[TMP10:%.*]] = zext i32 [[TMP9]] to i64
11058 // CHECK11-NEXT: call void @__kmpc_serialized_parallel(%struct.ident_t* @[[GLOB3]], i32 [[TMP1]])
11059 // CHECK11-NEXT: [[TMP11:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
11060 // CHECK11-NEXT: call void @.omp_outlined..7(i32* [[TMP11]], i32* [[DOTBOUND_ZERO_ADDR]], i64 [[TMP8]], i64 [[TMP10]]) #[[ATTR2]]
11061 // CHECK11-NEXT: call void @__kmpc_end_serialized_parallel(%struct.ident_t* @[[GLOB3]], i32 [[TMP1]])
11062 // CHECK11-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
11063 // CHECK11: omp.inner.for.inc:
11064 // CHECK11-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
11065 // CHECK11-NEXT: [[TMP13:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4
11066 // CHECK11-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP12]], [[TMP13]]
11067 // CHECK11-NEXT: store i32 [[ADD]], i32* [[DOTOMP_IV]], align 4
11068 // CHECK11-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP23:![0-9]+]]
11069 // CHECK11: omp.inner.for.end:
11070 // CHECK11-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
11071 // CHECK11: omp.loop.exit:
11072 // CHECK11-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP1]])
11073 // CHECK11-NEXT: [[TMP14:%.*]] = load i32, i32* [[DOTOMP_IS_LAST]], align 4
11074 // CHECK11-NEXT: [[TMP15:%.*]] = icmp ne i32 [[TMP14]], 0
11075 // CHECK11-NEXT: br i1 [[TMP15]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
11076 // CHECK11: .omp.final.then:
11077 // CHECK11-NEXT: store i32 100, i32* [[I]], align 4
11078 // CHECK11-NEXT: br label [[DOTOMP_FINAL_DONE]]
11079 // CHECK11: .omp.final.done:
11080 // CHECK11-NEXT: ret void
11081 //
11082 //
11083 // CHECK11-LABEL: define {{[^@]+}}@.omp_outlined..7
11084 // CHECK11-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[DOTPREVIOUS_LB_:%.*]], i64 [[DOTPREVIOUS_UB_:%.*]]) #[[ATTR1]] {
11085 // CHECK11-NEXT: entry:
11086 // CHECK11-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
11087 // CHECK11-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
11088 // CHECK11-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i64, align 8
11089 // CHECK11-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i64, align 8
11090 // CHECK11-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
11091 // CHECK11-NEXT: [[TMP:%.*]] = alloca i32, align 4
11092 // CHECK11-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4
11093 // CHECK11-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4
11094 // CHECK11-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
11095 // CHECK11-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
11096 // CHECK11-NEXT: [[I:%.*]] = alloca i32, align 4
11097 // CHECK11-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
11098 // CHECK11-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
11099 // CHECK11-NEXT: store i64 [[DOTPREVIOUS_LB_]], i64* [[DOTPREVIOUS_LB__ADDR]], align 8
11100 // CHECK11-NEXT: store i64 [[DOTPREVIOUS_UB_]], i64* [[DOTPREVIOUS_UB__ADDR]], align 8
11101 // CHECK11-NEXT: store i32 0, i32* [[DOTOMP_LB]], align 4
11102 // CHECK11-NEXT: store i32 99, i32* [[DOTOMP_UB]], align 4
11103 // CHECK11-NEXT: [[TMP0:%.*]] = load i64, i64* [[DOTPREVIOUS_LB__ADDR]], align 8
11104 // CHECK11-NEXT: [[CONV:%.*]] = trunc i64 [[TMP0]] to i32
11105 // CHECK11-NEXT: [[TMP1:%.*]] = load i64, i64* [[DOTPREVIOUS_UB__ADDR]], align 8
11106 // CHECK11-NEXT: [[CONV1:%.*]] = trunc i64 [[TMP1]] to i32
11107 // CHECK11-NEXT: store i32 [[CONV]], i32* [[DOTOMP_LB]], align 4
11108 // CHECK11-NEXT: store i32 [[CONV1]], i32* [[DOTOMP_UB]], align 4
11109 // CHECK11-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4
11110 // CHECK11-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4
11111 // CHECK11-NEXT: [[TMP2:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
11112 // CHECK11-NEXT: [[TMP3:%.*]] = load i32, i32* [[TMP2]], align 4
11113 // CHECK11-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB2]], i32 [[TMP3]], i32 34, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_LB]], i32* [[DOTOMP_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1)
11114 // CHECK11-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
11115 // CHECK11-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP4]], 99
11116 // CHECK11-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
11117 // CHECK11: cond.true:
11118 // CHECK11-NEXT: br label [[COND_END:%.*]]
11119 // CHECK11: cond.false:
11120 // CHECK11-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
11121 // CHECK11-NEXT: br label [[COND_END]]
11122 // CHECK11: cond.end:
11123 // CHECK11-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP5]], [[COND_FALSE]] ]
11124 // CHECK11-NEXT: store i32 [[COND]], i32* [[DOTOMP_UB]], align 4
11125 // CHECK11-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4
11126 // CHECK11-NEXT: store i32 [[TMP6]], i32* [[DOTOMP_IV]], align 4
11127 // CHECK11-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
11128 // CHECK11: omp.inner.for.cond:
11129 // CHECK11-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
11130 // CHECK11-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
11131 // CHECK11-NEXT: [[CMP2:%.*]] = icmp sle i32 [[TMP7]], [[TMP8]]
11132 // CHECK11-NEXT: br i1 [[CMP2]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
11133 // CHECK11: omp.inner.for.body:
11134 // CHECK11-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
11135 // CHECK11-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP9]], 1
11136 // CHECK11-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
11137 // CHECK11-NEXT: store i32 [[ADD]], i32* [[I]], align 4
11138 // CHECK11-NEXT: call void @_Z3fn5v()
11139 // CHECK11-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
11140 // CHECK11: omp.body.continue:
11141 // CHECK11-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
11142 // CHECK11: omp.inner.for.inc:
11143 // CHECK11-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
11144 // CHECK11-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP10]], 1
11145 // CHECK11-NEXT: store i32 [[ADD3]], i32* [[DOTOMP_IV]], align 4
11146 // CHECK11-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP25:![0-9]+]]
11147 // CHECK11: omp.inner.for.end:
11148 // CHECK11-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
11149 // CHECK11: omp.loop.exit:
11150 // CHECK11-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP3]])
11151 // CHECK11-NEXT: [[TMP11:%.*]] = load i32, i32* [[DOTOMP_IS_LAST]], align 4
11152 // CHECK11-NEXT: [[TMP12:%.*]] = icmp ne i32 [[TMP11]], 0
11153 // CHECK11-NEXT: br i1 [[TMP12]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
11154 // CHECK11: .omp.final.then:
11155 // CHECK11-NEXT: store i32 100, i32* [[I]], align 4
11156 // CHECK11-NEXT: br label [[DOTOMP_FINAL_DONE]]
11157 // CHECK11: .omp.final.done:
11158 // CHECK11-NEXT: ret void
11159 //
11160 //
11161 // CHECK11-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l92
11162 // CHECK11-SAME: (i64 [[ARG:%.*]]) #[[ATTR1]] {
11163 // CHECK11-NEXT: entry:
11164 // CHECK11-NEXT: [[ARG_ADDR:%.*]] = alloca i64, align 8
11165 // CHECK11-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i8, align 1
11166 // CHECK11-NEXT: [[DOTCAPTURE_EXPR__CASTED:%.*]] = alloca i64, align 8
11167 // CHECK11-NEXT: store i64 [[ARG]], i64* [[ARG_ADDR]], align 8
11168 // CHECK11-NEXT: [[CONV:%.*]] = bitcast i64* [[ARG_ADDR]] to i32*
11169 // CHECK11-NEXT: [[TMP0:%.*]] = load i32, i32* [[CONV]], align 8
11170 // CHECK11-NEXT: [[TOBOOL:%.*]] = icmp ne i32 [[TMP0]], 0
11171 // CHECK11-NEXT: [[FROMBOOL:%.*]] = zext i1 [[TOBOOL]] to i8
11172 // CHECK11-NEXT: store i8 [[FROMBOOL]], i8* [[DOTCAPTURE_EXPR_]], align 1
11173 // CHECK11-NEXT: [[TMP1:%.*]] = load i8, i8* [[DOTCAPTURE_EXPR_]], align 1
11174 // CHECK11-NEXT: [[TOBOOL1:%.*]] = trunc i8 [[TMP1]] to i1
11175 // CHECK11-NEXT: [[CONV2:%.*]] = bitcast i64* [[DOTCAPTURE_EXPR__CASTED]] to i8*
11176 // CHECK11-NEXT: [[FROMBOOL3:%.*]] = zext i1 [[TOBOOL1]] to i8
11177 // CHECK11-NEXT: store i8 [[FROMBOOL3]], i8* [[CONV2]], align 1
11178 // CHECK11-NEXT: [[TMP2:%.*]] = load i64, i64* [[DOTCAPTURE_EXPR__CASTED]], align 8
11179 // CHECK11-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64)* @.omp_outlined..8 to void (i32*, i32*, ...)*), i64 [[TMP2]])
11180 // CHECK11-NEXT: ret void
11181 //
11182 //
11183 // CHECK11-LABEL: define {{[^@]+}}@.omp_outlined..8
11184 // CHECK11-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[DOTCAPTURE_EXPR_:%.*]]) #[[ATTR1]] {
11185 // CHECK11-NEXT: entry:
11186 // CHECK11-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
11187 // CHECK11-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
11188 // CHECK11-NEXT: [[DOTCAPTURE_EXPR__ADDR:%.*]] = alloca i64, align 8
11189 // CHECK11-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
11190 // CHECK11-NEXT: [[TMP:%.*]] = alloca i32, align 4
11191 // CHECK11-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4
11192 // CHECK11-NEXT: [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4
11193 // CHECK11-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
11194 // CHECK11-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
11195 // CHECK11-NEXT: [[I:%.*]] = alloca i32, align 4
11196 // CHECK11-NEXT: [[DOTCAPTURE_EXPR__CASTED:%.*]] = alloca i64, align 8
11197 // CHECK11-NEXT: [[DOTBOUND_ZERO_ADDR:%.*]] = alloca i32, align 4
11198 // CHECK11-NEXT: [[DOTCAPTURE_EXPR__CASTED11:%.*]] = alloca i64, align 8
11199 // CHECK11-NEXT: [[DOTBOUND_ZERO_ADDR17:%.*]] = alloca i32, align 4
11200 // CHECK11-NEXT: store i32 0, i32* [[DOTBOUND_ZERO_ADDR17]], align 4
11201 // CHECK11-NEXT: store i32 0, i32* [[DOTBOUND_ZERO_ADDR]], align 4
11202 // CHECK11-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
11203 // CHECK11-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
11204 // CHECK11-NEXT: store i64 [[DOTCAPTURE_EXPR_]], i64* [[DOTCAPTURE_EXPR__ADDR]], align 8
11205 // CHECK11-NEXT: [[CONV:%.*]] = bitcast i64* [[DOTCAPTURE_EXPR__ADDR]] to i8*
11206 // CHECK11-NEXT: store i32 0, i32* [[DOTOMP_COMB_LB]], align 4
11207 // CHECK11-NEXT: store i32 99, i32* [[DOTOMP_COMB_UB]], align 4
11208 // CHECK11-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4
11209 // CHECK11-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4
11210 // CHECK11-NEXT: [[TMP0:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
11211 // CHECK11-NEXT: [[TMP1:%.*]] = load i32, i32* [[TMP0]], align 4
11212 // CHECK11-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB1]], i32 [[TMP1]], i32 92, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_COMB_LB]], i32* [[DOTOMP_COMB_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1)
11213 // CHECK11-NEXT: [[TMP2:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
11214 // CHECK11-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP2]], 99
11215 // CHECK11-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
11216 // CHECK11: cond.true:
11217 // CHECK11-NEXT: br label [[COND_END:%.*]]
11218 // CHECK11: cond.false:
11219 // CHECK11-NEXT: [[TMP3:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
11220 // CHECK11-NEXT: br label [[COND_END]]
11221 // CHECK11: cond.end:
11222 // CHECK11-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP3]], [[COND_FALSE]] ]
11223 // CHECK11-NEXT: store i32 [[COND]], i32* [[DOTOMP_COMB_UB]], align 4
11224 // CHECK11-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4
11225 // CHECK11-NEXT: store i32 [[TMP4]], i32* [[DOTOMP_IV]], align 4
11226 // CHECK11-NEXT: [[TMP5:%.*]] = load i8, i8* [[CONV]], align 8
11227 // CHECK11-NEXT: [[TOBOOL:%.*]] = trunc i8 [[TMP5]] to i1
11228 // CHECK11-NEXT: br i1 [[TOBOOL]], label [[OMP_IF_THEN:%.*]], label [[OMP_IF_ELSE6:%.*]]
11229 // CHECK11: omp_if.then:
11230 // CHECK11-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
11231 // CHECK11: omp.inner.for.cond:
11232 // CHECK11-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
11233 // CHECK11-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
11234 // CHECK11-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP6]], [[TMP7]]
11235 // CHECK11-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
11236 // CHECK11: omp.inner.for.body:
11237 // CHECK11-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4
11238 // CHECK11-NEXT: [[TMP9:%.*]] = zext i32 [[TMP8]] to i64
11239 // CHECK11-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
11240 // CHECK11-NEXT: [[TMP11:%.*]] = zext i32 [[TMP10]] to i64
11241 // CHECK11-NEXT: [[TMP12:%.*]] = load i8, i8* [[CONV]], align 8
11242 // CHECK11-NEXT: [[TOBOOL2:%.*]] = trunc i8 [[TMP12]] to i1
11243 // CHECK11-NEXT: [[CONV3:%.*]] = bitcast i64* [[DOTCAPTURE_EXPR__CASTED]] to i8*
11244 // CHECK11-NEXT: [[FROMBOOL:%.*]] = zext i1 [[TOBOOL2]] to i8
11245 // CHECK11-NEXT: store i8 [[FROMBOOL]], i8* [[CONV3]], align 1
11246 // CHECK11-NEXT: [[TMP13:%.*]] = load i64, i64* [[DOTCAPTURE_EXPR__CASTED]], align 8
11247 // CHECK11-NEXT: [[TMP14:%.*]] = load i8, i8* [[CONV]], align 8
11248 // CHECK11-NEXT: [[TOBOOL4:%.*]] = trunc i8 [[TMP14]] to i1
11249 // CHECK11-NEXT: br i1 [[TOBOOL4]], label [[OMP_IF_THEN5:%.*]], label [[OMP_IF_ELSE:%.*]]
11250 // CHECK11: omp_if.then5:
11251 // CHECK11-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB3]], i32 3, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64, i64, i64)* @.omp_outlined..9 to void (i32*, i32*, ...)*), i64 [[TMP9]], i64 [[TMP11]], i64 [[TMP13]])
11252 // CHECK11-NEXT: br label [[OMP_IF_END:%.*]]
11253 // CHECK11: omp_if.else:
11254 // CHECK11-NEXT: call void @__kmpc_serialized_parallel(%struct.ident_t* @[[GLOB3]], i32 [[TMP1]])
11255 // CHECK11-NEXT: [[TMP15:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
11256 // CHECK11-NEXT: call void @.omp_outlined..9(i32* [[TMP15]], i32* [[DOTBOUND_ZERO_ADDR]], i64 [[TMP9]], i64 [[TMP11]], i64 [[TMP13]]) #[[ATTR2]]
11257 // CHECK11-NEXT: call void @__kmpc_end_serialized_parallel(%struct.ident_t* @[[GLOB3]], i32 [[TMP1]])
11258 // CHECK11-NEXT: br label [[OMP_IF_END]]
11259 // CHECK11: omp_if.end:
11260 // CHECK11-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
11261 // CHECK11: omp.inner.for.inc:
11262 // CHECK11-NEXT: [[TMP16:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
11263 // CHECK11-NEXT: [[TMP17:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4
11264 // CHECK11-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP16]], [[TMP17]]
11265 // CHECK11-NEXT: store i32 [[ADD]], i32* [[DOTOMP_IV]], align 4
11266 // CHECK11-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP26:![0-9]+]]
11267 // CHECK11: omp.inner.for.end:
11268 // CHECK11-NEXT: br label [[OMP_IF_END22:%.*]]
11269 // CHECK11: omp_if.else6:
11270 // CHECK11-NEXT: br label [[OMP_INNER_FOR_COND7:%.*]]
11271 // CHECK11: omp.inner.for.cond7:
11272 // CHECK11-NEXT: [[TMP18:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
11273 // CHECK11-NEXT: [[TMP19:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
11274 // CHECK11-NEXT: [[CMP8:%.*]] = icmp sle i32 [[TMP18]], [[TMP19]]
11275 // CHECK11-NEXT: br i1 [[CMP8]], label [[OMP_INNER_FOR_BODY9:%.*]], label [[OMP_INNER_FOR_END21:%.*]]
11276 // CHECK11: omp.inner.for.body9:
11277 // CHECK11-NEXT: [[TMP20:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4
11278 // CHECK11-NEXT: [[TMP21:%.*]] = zext i32 [[TMP20]] to i64
11279 // CHECK11-NEXT: [[TMP22:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
11280 // CHECK11-NEXT: [[TMP23:%.*]] = zext i32 [[TMP22]] to i64
11281 // CHECK11-NEXT: [[TMP24:%.*]] = load i8, i8* [[CONV]], align 8
11282 // CHECK11-NEXT: [[TOBOOL10:%.*]] = trunc i8 [[TMP24]] to i1
11283 // CHECK11-NEXT: [[CONV12:%.*]] = bitcast i64* [[DOTCAPTURE_EXPR__CASTED11]] to i8*
11284 // CHECK11-NEXT: [[FROMBOOL13:%.*]] = zext i1 [[TOBOOL10]] to i8
11285 // CHECK11-NEXT: store i8 [[FROMBOOL13]], i8* [[CONV12]], align 1
11286 // CHECK11-NEXT: [[TMP25:%.*]] = load i64, i64* [[DOTCAPTURE_EXPR__CASTED11]], align 8
11287 // CHECK11-NEXT: [[TMP26:%.*]] = load i8, i8* [[CONV]], align 8
11288 // CHECK11-NEXT: [[TOBOOL14:%.*]] = trunc i8 [[TMP26]] to i1
11289 // CHECK11-NEXT: br i1 [[TOBOOL14]], label [[OMP_IF_THEN15:%.*]], label [[OMP_IF_ELSE16:%.*]]
11290 // CHECK11: omp_if.then15:
11291 // CHECK11-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB3]], i32 3, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64, i64, i64)* @.omp_outlined..10 to void (i32*, i32*, ...)*), i64 [[TMP21]], i64 [[TMP23]], i64 [[TMP25]])
11292 // CHECK11-NEXT: br label [[OMP_IF_END18:%.*]]
11293 // CHECK11: omp_if.else16:
11294 // CHECK11-NEXT: call void @__kmpc_serialized_parallel(%struct.ident_t* @[[GLOB3]], i32 [[TMP1]])
11295 // CHECK11-NEXT: [[TMP27:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
11296 // CHECK11-NEXT: call void @.omp_outlined..10(i32* [[TMP27]], i32* [[DOTBOUND_ZERO_ADDR17]], i64 [[TMP21]], i64 [[TMP23]], i64 [[TMP25]]) #[[ATTR2]]
11297 // CHECK11-NEXT: call void @__kmpc_end_serialized_parallel(%struct.ident_t* @[[GLOB3]], i32 [[TMP1]])
11298 // CHECK11-NEXT: br label [[OMP_IF_END18]]
11299 // CHECK11: omp_if.end18:
11300 // CHECK11-NEXT: br label [[OMP_INNER_FOR_INC19:%.*]]
11301 // CHECK11: omp.inner.for.inc19:
11302 // CHECK11-NEXT: [[TMP28:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
11303 // CHECK11-NEXT: [[TMP29:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4
11304 // CHECK11-NEXT: [[ADD20:%.*]] = add nsw i32 [[TMP28]], [[TMP29]]
11305 // CHECK11-NEXT: store i32 [[ADD20]], i32* [[DOTOMP_IV]], align 4
11306 // CHECK11-NEXT: br label [[OMP_INNER_FOR_COND7]], !llvm.loop [[LOOP27:![0-9]+]]
11307 // CHECK11: omp.inner.for.end21:
11308 // CHECK11-NEXT: br label [[OMP_IF_END22]]
11309 // CHECK11: omp_if.end22:
11310 // CHECK11-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
11311 // CHECK11: omp.loop.exit:
11312 // CHECK11-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP1]])
11313 // CHECK11-NEXT: [[TMP30:%.*]] = load i32, i32* [[DOTOMP_IS_LAST]], align 4
11314 // CHECK11-NEXT: [[TMP31:%.*]] = icmp ne i32 [[TMP30]], 0
11315 // CHECK11-NEXT: br i1 [[TMP31]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
11316 // CHECK11: .omp.final.then:
11317 // CHECK11-NEXT: store i32 100, i32* [[I]], align 4
11318 // CHECK11-NEXT: br label [[DOTOMP_FINAL_DONE]]
11319 // CHECK11: .omp.final.done:
11320 // CHECK11-NEXT: ret void
11321 //
11322 //
11323 // CHECK11-LABEL: define {{[^@]+}}@.omp_outlined..9
11324 // CHECK11-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[DOTPREVIOUS_LB_:%.*]], i64 [[DOTPREVIOUS_UB_:%.*]], i64 [[DOTCAPTURE_EXPR_:%.*]]) #[[ATTR1]] {
11325 // CHECK11-NEXT: entry:
11326 // CHECK11-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
11327 // CHECK11-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
11328 // CHECK11-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i64, align 8
11329 // CHECK11-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i64, align 8
11330 // CHECK11-NEXT: [[DOTCAPTURE_EXPR__ADDR:%.*]] = alloca i64, align 8
11331 // CHECK11-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
11332 // CHECK11-NEXT: [[TMP:%.*]] = alloca i32, align 4
11333 // CHECK11-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4
11334 // CHECK11-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4
11335 // CHECK11-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
11336 // CHECK11-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
11337 // CHECK11-NEXT: [[I:%.*]] = alloca i32, align 4
11338 // CHECK11-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
11339 // CHECK11-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
11340 // CHECK11-NEXT: store i64 [[DOTPREVIOUS_LB_]], i64* [[DOTPREVIOUS_LB__ADDR]], align 8
11341 // CHECK11-NEXT: store i64 [[DOTPREVIOUS_UB_]], i64* [[DOTPREVIOUS_UB__ADDR]], align 8
11342 // CHECK11-NEXT: store i64 [[DOTCAPTURE_EXPR_]], i64* [[DOTCAPTURE_EXPR__ADDR]], align 8
11343 // CHECK11-NEXT: [[CONV:%.*]] = bitcast i64* [[DOTCAPTURE_EXPR__ADDR]] to i8*
11344 // CHECK11-NEXT: store i32 0, i32* [[DOTOMP_LB]], align 4
11345 // CHECK11-NEXT: store i32 99, i32* [[DOTOMP_UB]], align 4
11346 // CHECK11-NEXT: [[TMP0:%.*]] = load i64, i64* [[DOTPREVIOUS_LB__ADDR]], align 8
11347 // CHECK11-NEXT: [[CONV1:%.*]] = trunc i64 [[TMP0]] to i32
11348 // CHECK11-NEXT: [[TMP1:%.*]] = load i64, i64* [[DOTPREVIOUS_UB__ADDR]], align 8
11349 // CHECK11-NEXT: [[CONV2:%.*]] = trunc i64 [[TMP1]] to i32
11350 // CHECK11-NEXT: store i32 [[CONV1]], i32* [[DOTOMP_LB]], align 4
11351 // CHECK11-NEXT: store i32 [[CONV2]], i32* [[DOTOMP_UB]], align 4
11352 // CHECK11-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4
11353 // CHECK11-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4
11354 // CHECK11-NEXT: [[TMP2:%.*]] = load i8, i8* [[CONV]], align 8
11355 // CHECK11-NEXT: [[TOBOOL:%.*]] = trunc i8 [[TMP2]] to i1
11356 // CHECK11-NEXT: br i1 [[TOBOOL]], label [[OMP_IF_THEN:%.*]], label [[OMP_IF_ELSE:%.*]]
11357 // CHECK11: omp_if.then:
11358 // CHECK11-NEXT: [[TMP3:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
11359 // CHECK11-NEXT: [[TMP4:%.*]] = load i32, i32* [[TMP3]], align 4
11360 // CHECK11-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB2]], i32 [[TMP4]], i32 34, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_LB]], i32* [[DOTOMP_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1)
11361 // CHECK11-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
11362 // CHECK11-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP5]], 99
11363 // CHECK11-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
11364 // CHECK11: cond.true:
11365 // CHECK11-NEXT: br label [[COND_END:%.*]]
11366 // CHECK11: cond.false:
11367 // CHECK11-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
11368 // CHECK11-NEXT: br label [[COND_END]]
11369 // CHECK11: cond.end:
11370 // CHECK11-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP6]], [[COND_FALSE]] ]
11371 // CHECK11-NEXT: store i32 [[COND]], i32* [[DOTOMP_UB]], align 4
11372 // CHECK11-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4
11373 // CHECK11-NEXT: store i32 [[TMP7]], i32* [[DOTOMP_IV]], align 4
11374 // CHECK11-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
11375 // CHECK11: omp.inner.for.cond:
11376 // CHECK11-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
11377 // CHECK11-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
11378 // CHECK11-NEXT: [[CMP3:%.*]] = icmp sle i32 [[TMP8]], [[TMP9]]
11379 // CHECK11-NEXT: br i1 [[CMP3]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
11380 // CHECK11: omp.inner.for.body:
11381 // CHECK11-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
11382 // CHECK11-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP10]], 1
11383 // CHECK11-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
11384 // CHECK11-NEXT: store i32 [[ADD]], i32* [[I]], align 4
11385 // CHECK11-NEXT: call void @_Z3fn6v()
11386 // CHECK11-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
11387 // CHECK11: omp.body.continue:
11388 // CHECK11-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
11389 // CHECK11: omp.inner.for.inc:
11390 // CHECK11-NEXT: [[TMP11:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
11391 // CHECK11-NEXT: [[ADD4:%.*]] = add nsw i32 [[TMP11]], 1
11392 // CHECK11-NEXT: store i32 [[ADD4]], i32* [[DOTOMP_IV]], align 4
11393 // CHECK11-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP28:![0-9]+]]
11394 // CHECK11: omp.inner.for.end:
11395 // CHECK11-NEXT: br label [[OMP_IF_END:%.*]]
11396 // CHECK11: omp_if.else:
11397 // CHECK11-NEXT: [[TMP12:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
11398 // CHECK11-NEXT: [[TMP13:%.*]] = load i32, i32* [[TMP12]], align 4
11399 // CHECK11-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB2]], i32 [[TMP13]], i32 34, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_LB]], i32* [[DOTOMP_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1)
11400 // CHECK11-NEXT: [[TMP14:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
11401 // CHECK11-NEXT: [[CMP5:%.*]] = icmp sgt i32 [[TMP14]], 99
11402 // CHECK11-NEXT: br i1 [[CMP5]], label [[COND_TRUE6:%.*]], label [[COND_FALSE7:%.*]]
11403 // CHECK11: cond.true6:
11404 // CHECK11-NEXT: br label [[COND_END8:%.*]]
11405 // CHECK11: cond.false7:
11406 // CHECK11-NEXT: [[TMP15:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
11407 // CHECK11-NEXT: br label [[COND_END8]]
11408 // CHECK11: cond.end8:
11409 // CHECK11-NEXT: [[COND9:%.*]] = phi i32 [ 99, [[COND_TRUE6]] ], [ [[TMP15]], [[COND_FALSE7]] ]
11410 // CHECK11-NEXT: store i32 [[COND9]], i32* [[DOTOMP_UB]], align 4
11411 // CHECK11-NEXT: [[TMP16:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4
11412 // CHECK11-NEXT: store i32 [[TMP16]], i32* [[DOTOMP_IV]], align 4
11413 // CHECK11-NEXT: br label [[OMP_INNER_FOR_COND10:%.*]]
11414 // CHECK11: omp.inner.for.cond10:
11415 // CHECK11-NEXT: [[TMP17:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
11416 // CHECK11-NEXT: [[TMP18:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
11417 // CHECK11-NEXT: [[CMP11:%.*]] = icmp sle i32 [[TMP17]], [[TMP18]]
11418 // CHECK11-NEXT: br i1 [[CMP11]], label [[OMP_INNER_FOR_BODY12:%.*]], label [[OMP_INNER_FOR_END18:%.*]]
11419 // CHECK11: omp.inner.for.body12:
11420 // CHECK11-NEXT: [[TMP19:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
11421 // CHECK11-NEXT: [[MUL13:%.*]] = mul nsw i32 [[TMP19]], 1
11422 // CHECK11-NEXT: [[ADD14:%.*]] = add nsw i32 0, [[MUL13]]
11423 // CHECK11-NEXT: store i32 [[ADD14]], i32* [[I]], align 4
11424 // CHECK11-NEXT: call void @_Z3fn6v()
11425 // CHECK11-NEXT: br label [[OMP_BODY_CONTINUE15:%.*]]
11426 // CHECK11: omp.body.continue15:
11427 // CHECK11-NEXT: br label [[OMP_INNER_FOR_INC16:%.*]]
11428 // CHECK11: omp.inner.for.inc16:
11429 // CHECK11-NEXT: [[TMP20:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
11430 // CHECK11-NEXT: [[ADD17:%.*]] = add nsw i32 [[TMP20]], 1
11431 // CHECK11-NEXT: store i32 [[ADD17]], i32* [[DOTOMP_IV]], align 4
11432 // CHECK11-NEXT: br label [[OMP_INNER_FOR_COND10]], !llvm.loop [[LOOP29:![0-9]+]]
11433 // CHECK11: omp.inner.for.end18:
11434 // CHECK11-NEXT: br label [[OMP_IF_END]]
11435 // CHECK11: omp_if.end:
11436 // CHECK11-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
11437 // CHECK11: omp.loop.exit:
11438 // CHECK11-NEXT: [[TMP21:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
11439 // CHECK11-NEXT: [[TMP22:%.*]] = load i32, i32* [[TMP21]], align 4
11440 // CHECK11-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP22]])
11441 // CHECK11-NEXT: [[TMP23:%.*]] = load i32, i32* [[DOTOMP_IS_LAST]], align 4
11442 // CHECK11-NEXT: [[TMP24:%.*]] = icmp ne i32 [[TMP23]], 0
11443 // CHECK11-NEXT: br i1 [[TMP24]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
11444 // CHECK11: .omp.final.then:
11445 // CHECK11-NEXT: store i32 100, i32* [[I]], align 4
11446 // CHECK11-NEXT: br label [[DOTOMP_FINAL_DONE]]
11447 // CHECK11: .omp.final.done:
11448 // CHECK11-NEXT: ret void
11449 //
11450 //
11451 // CHECK11-LABEL: define {{[^@]+}}@.omp_outlined..10
11452 // CHECK11-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[DOTPREVIOUS_LB_:%.*]], i64 [[DOTPREVIOUS_UB_:%.*]], i64 [[DOTCAPTURE_EXPR_:%.*]]) #[[ATTR1]] {
11453 // CHECK11-NEXT: entry:
11454 // CHECK11-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
11455 // CHECK11-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
11456 // CHECK11-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i64, align 8
11457 // CHECK11-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i64, align 8
11458 // CHECK11-NEXT: [[DOTCAPTURE_EXPR__ADDR:%.*]] = alloca i64, align 8
11459 // CHECK11-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
11460 // CHECK11-NEXT: [[TMP:%.*]] = alloca i32, align 4
11461 // CHECK11-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4
11462 // CHECK11-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4
11463 // CHECK11-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
11464 // CHECK11-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
11465 // CHECK11-NEXT: [[I:%.*]] = alloca i32, align 4
11466 // CHECK11-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
11467 // CHECK11-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
11468 // CHECK11-NEXT: store i64 [[DOTPREVIOUS_LB_]], i64* [[DOTPREVIOUS_LB__ADDR]], align 8
11469 // CHECK11-NEXT: store i64 [[DOTPREVIOUS_UB_]], i64* [[DOTPREVIOUS_UB__ADDR]], align 8
11470 // CHECK11-NEXT: store i64 [[DOTCAPTURE_EXPR_]], i64* [[DOTCAPTURE_EXPR__ADDR]], align 8
11471 // CHECK11-NEXT: [[CONV:%.*]] = bitcast i64* [[DOTCAPTURE_EXPR__ADDR]] to i8*
11472 // CHECK11-NEXT: store i32 0, i32* [[DOTOMP_LB]], align 4
11473 // CHECK11-NEXT: store i32 99, i32* [[DOTOMP_UB]], align 4
11474 // CHECK11-NEXT: [[TMP0:%.*]] = load i64, i64* [[DOTPREVIOUS_LB__ADDR]], align 8
11475 // CHECK11-NEXT: [[CONV1:%.*]] = trunc i64 [[TMP0]] to i32
11476 // CHECK11-NEXT: [[TMP1:%.*]] = load i64, i64* [[DOTPREVIOUS_UB__ADDR]], align 8
11477 // CHECK11-NEXT: [[CONV2:%.*]] = trunc i64 [[TMP1]] to i32
11478 // CHECK11-NEXT: store i32 [[CONV1]], i32* [[DOTOMP_LB]], align 4
11479 // CHECK11-NEXT: store i32 [[CONV2]], i32* [[DOTOMP_UB]], align 4
11480 // CHECK11-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4
11481 // CHECK11-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4
11482 // CHECK11-NEXT: [[TMP2:%.*]] = load i8, i8* [[CONV]], align 8
11483 // CHECK11-NEXT: [[TOBOOL:%.*]] = trunc i8 [[TMP2]] to i1
11484 // CHECK11-NEXT: br i1 [[TOBOOL]], label [[OMP_IF_THEN:%.*]], label [[OMP_IF_ELSE:%.*]]
11485 // CHECK11: omp_if.then:
11486 // CHECK11-NEXT: [[TMP3:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
11487 // CHECK11-NEXT: [[TMP4:%.*]] = load i32, i32* [[TMP3]], align 4
11488 // CHECK11-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB2]], i32 [[TMP4]], i32 34, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_LB]], i32* [[DOTOMP_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1)
11489 // CHECK11-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
11490 // CHECK11-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP5]], 99
11491 // CHECK11-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
11492 // CHECK11: cond.true:
11493 // CHECK11-NEXT: br label [[COND_END:%.*]]
11494 // CHECK11: cond.false:
11495 // CHECK11-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
11496 // CHECK11-NEXT: br label [[COND_END]]
11497 // CHECK11: cond.end:
11498 // CHECK11-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP6]], [[COND_FALSE]] ]
11499 // CHECK11-NEXT: store i32 [[COND]], i32* [[DOTOMP_UB]], align 4
11500 // CHECK11-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4
11501 // CHECK11-NEXT: store i32 [[TMP7]], i32* [[DOTOMP_IV]], align 4
11502 // CHECK11-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
11503 // CHECK11: omp.inner.for.cond:
11504 // CHECK11-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
11505 // CHECK11-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
11506 // CHECK11-NEXT: [[CMP3:%.*]] = icmp sle i32 [[TMP8]], [[TMP9]]
11507 // CHECK11-NEXT: br i1 [[CMP3]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
11508 // CHECK11: omp.inner.for.body:
11509 // CHECK11-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
11510 // CHECK11-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP10]], 1
11511 // CHECK11-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
11512 // CHECK11-NEXT: store i32 [[ADD]], i32* [[I]], align 4
11513 // CHECK11-NEXT: call void @_Z3fn6v()
11514 // CHECK11-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
11515 // CHECK11: omp.body.continue:
11516 // CHECK11-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
11517 // CHECK11: omp.inner.for.inc:
11518 // CHECK11-NEXT: [[TMP11:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
11519 // CHECK11-NEXT: [[ADD4:%.*]] = add nsw i32 [[TMP11]], 1
11520 // CHECK11-NEXT: store i32 [[ADD4]], i32* [[DOTOMP_IV]], align 4
11521 // CHECK11-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP30:![0-9]+]]
11522 // CHECK11: omp.inner.for.end:
11523 // CHECK11-NEXT: br label [[OMP_IF_END:%.*]]
11524 // CHECK11: omp_if.else:
11525 // CHECK11-NEXT: [[TMP12:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
11526 // CHECK11-NEXT: [[TMP13:%.*]] = load i32, i32* [[TMP12]], align 4
11527 // CHECK11-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB2]], i32 [[TMP13]], i32 34, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_LB]], i32* [[DOTOMP_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1)
11528 // CHECK11-NEXT: [[TMP14:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
11529 // CHECK11-NEXT: [[CMP5:%.*]] = icmp sgt i32 [[TMP14]], 99
11530 // CHECK11-NEXT: br i1 [[CMP5]], label [[COND_TRUE6:%.*]], label [[COND_FALSE7:%.*]]
11531 // CHECK11: cond.true6:
11532 // CHECK11-NEXT: br label [[COND_END8:%.*]]
11533 // CHECK11: cond.false7:
11534 // CHECK11-NEXT: [[TMP15:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
11535 // CHECK11-NEXT: br label [[COND_END8]]
11536 // CHECK11: cond.end8:
11537 // CHECK11-NEXT: [[COND9:%.*]] = phi i32 [ 99, [[COND_TRUE6]] ], [ [[TMP15]], [[COND_FALSE7]] ]
11538 // CHECK11-NEXT: store i32 [[COND9]], i32* [[DOTOMP_UB]], align 4
11539 // CHECK11-NEXT: [[TMP16:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4
11540 // CHECK11-NEXT: store i32 [[TMP16]], i32* [[DOTOMP_IV]], align 4
11541 // CHECK11-NEXT: br label [[OMP_INNER_FOR_COND10:%.*]]
11542 // CHECK11: omp.inner.for.cond10:
11543 // CHECK11-NEXT: [[TMP17:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
11544 // CHECK11-NEXT: [[TMP18:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
11545 // CHECK11-NEXT: [[CMP11:%.*]] = icmp sle i32 [[TMP17]], [[TMP18]]
11546 // CHECK11-NEXT: br i1 [[CMP11]], label [[OMP_INNER_FOR_BODY12:%.*]], label [[OMP_INNER_FOR_END18:%.*]]
11547 // CHECK11: omp.inner.for.body12:
11548 // CHECK11-NEXT: [[TMP19:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
11549 // CHECK11-NEXT: [[MUL13:%.*]] = mul nsw i32 [[TMP19]], 1
11550 // CHECK11-NEXT: [[ADD14:%.*]] = add nsw i32 0, [[MUL13]]
11551 // CHECK11-NEXT: store i32 [[ADD14]], i32* [[I]], align 4
11552 // CHECK11-NEXT: call void @_Z3fn6v()
11553 // CHECK11-NEXT: br label [[OMP_BODY_CONTINUE15:%.*]]
11554 // CHECK11: omp.body.continue15:
11555 // CHECK11-NEXT: br label [[OMP_INNER_FOR_INC16:%.*]]
11556 // CHECK11: omp.inner.for.inc16:
11557 // CHECK11-NEXT: [[TMP20:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
11558 // CHECK11-NEXT: [[ADD17:%.*]] = add nsw i32 [[TMP20]], 1
11559 // CHECK11-NEXT: store i32 [[ADD17]], i32* [[DOTOMP_IV]], align 4
11560 // CHECK11-NEXT: br label [[OMP_INNER_FOR_COND10]], !llvm.loop [[LOOP31:![0-9]+]]
11561 // CHECK11: omp.inner.for.end18:
11562 // CHECK11-NEXT: br label [[OMP_IF_END]]
11563 // CHECK11: omp_if.end:
11564 // CHECK11-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
11565 // CHECK11: omp.loop.exit:
11566 // CHECK11-NEXT: [[TMP21:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
11567 // CHECK11-NEXT: [[TMP22:%.*]] = load i32, i32* [[TMP21]], align 4
11568 // CHECK11-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP22]])
11569 // CHECK11-NEXT: [[TMP23:%.*]] = load i32, i32* [[DOTOMP_IS_LAST]], align 4
11570 // CHECK11-NEXT: [[TMP24:%.*]] = icmp ne i32 [[TMP23]], 0
11571 // CHECK11-NEXT: br i1 [[TMP24]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
11572 // CHECK11: .omp.final.then:
11573 // CHECK11-NEXT: store i32 100, i32* [[I]], align 4
11574 // CHECK11-NEXT: br label [[DOTOMP_FINAL_DONE]]
11575 // CHECK11: .omp.final.done:
11576 // CHECK11-NEXT: ret void
11577 //
11578 //
11579 // CHECK11-LABEL: define {{[^@]+}}@_Z5tmainIiEiT_
11580 // CHECK11-SAME: (i32 [[ARG:%.*]]) #[[ATTR0]] comdat {
11581 // CHECK11-NEXT: entry:
11582 // CHECK11-NEXT: [[ARG_ADDR:%.*]] = alloca i32, align 4
11583 // CHECK11-NEXT: [[TMP:%.*]] = alloca i32, align 4
11584 // CHECK11-NEXT: [[_TMP1:%.*]] = alloca i32, align 4
11585 // CHECK11-NEXT: [[ARG_CASTED:%.*]] = alloca i64, align 8
11586 // CHECK11-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [1 x i8*], align 8
11587 // CHECK11-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [1 x i8*], align 8
11588 // CHECK11-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [1 x i8*], align 8
11589 // CHECK11-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i8, align 1
11590 // CHECK11-NEXT: [[_TMP5:%.*]] = alloca i32, align 4
11591 // CHECK11-NEXT: store i32 [[ARG]], i32* [[ARG_ADDR]], align 4
11592 // CHECK11-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i64 100)
11593 // CHECK11-NEXT: [[TMP0:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiEiT__l57.region_id, i32 0, i8** null, i8** null, i64* null, i64* null, i8** null, i8** null, i32 0, i32 0)
11594 // CHECK11-NEXT: [[TMP1:%.*]] = icmp ne i32 [[TMP0]], 0
11595 // CHECK11-NEXT: br i1 [[TMP1]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
11596 // CHECK11: omp_offload.failed:
11597 // CHECK11-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiEiT__l57() #[[ATTR2]]
11598 // CHECK11-NEXT: br label [[OMP_OFFLOAD_CONT]]
11599 // CHECK11: omp_offload.cont:
11600 // CHECK11-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i64 100)
11601 // CHECK11-NEXT: [[TMP2:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiEiT__l62.region_id, i32 0, i8** null, i8** null, i64* null, i64* null, i8** null, i8** null, i32 0, i32 1)
11602 // CHECK11-NEXT: [[TMP3:%.*]] = icmp ne i32 [[TMP2]], 0
11603 // CHECK11-NEXT: br i1 [[TMP3]], label [[OMP_OFFLOAD_FAILED2:%.*]], label [[OMP_OFFLOAD_CONT3:%.*]]
11604 // CHECK11: omp_offload.failed2:
11605 // CHECK11-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiEiT__l62() #[[ATTR2]]
11606 // CHECK11-NEXT: br label [[OMP_OFFLOAD_CONT3]]
11607 // CHECK11: omp_offload.cont3:
11608 // CHECK11-NEXT: [[TMP4:%.*]] = load i32, i32* [[ARG_ADDR]], align 4
11609 // CHECK11-NEXT: [[CONV:%.*]] = bitcast i64* [[ARG_CASTED]] to i32*
11610 // CHECK11-NEXT: store i32 [[TMP4]], i32* [[CONV]], align 4
11611 // CHECK11-NEXT: [[TMP5:%.*]] = load i64, i64* [[ARG_CASTED]], align 8
11612 // CHECK11-NEXT: [[TMP6:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
11613 // CHECK11-NEXT: [[TMP7:%.*]] = bitcast i8** [[TMP6]] to i64*
11614 // CHECK11-NEXT: store i64 [[TMP5]], i64* [[TMP7]], align 8
11615 // CHECK11-NEXT: [[TMP8:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
11616 // CHECK11-NEXT: [[TMP9:%.*]] = bitcast i8** [[TMP8]] to i64*
11617 // CHECK11-NEXT: store i64 [[TMP5]], i64* [[TMP9]], align 8
11618 // CHECK11-NEXT: [[TMP10:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0
11619 // CHECK11-NEXT: store i8* null, i8** [[TMP10]], align 8
11620 // CHECK11-NEXT: [[TMP11:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
11621 // CHECK11-NEXT: [[TMP12:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
11622 // CHECK11-NEXT: [[TMP13:%.*]] = load i32, i32* [[ARG_ADDR]], align 4
11623 // CHECK11-NEXT: [[TOBOOL:%.*]] = icmp ne i32 [[TMP13]], 0
11624 // CHECK11-NEXT: [[FROMBOOL:%.*]] = zext i1 [[TOBOOL]] to i8
11625 // CHECK11-NEXT: store i8 [[FROMBOOL]], i8* [[DOTCAPTURE_EXPR_]], align 1
11626 // CHECK11-NEXT: [[TMP14:%.*]] = load i8, i8* [[DOTCAPTURE_EXPR_]], align 1
11627 // CHECK11-NEXT: [[TOBOOL4:%.*]] = trunc i8 [[TMP14]] to i1
11628 // CHECK11-NEXT: [[TMP15:%.*]] = select i1 [[TOBOOL4]], i32 0, i32 1
11629 // CHECK11-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i64 100)
11630 // CHECK11-NEXT: [[TMP16:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiEiT__l67.region_id, i32 1, i8** [[TMP11]], i8** [[TMP12]], i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_sizes.17, i32 0, i32 0), i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_maptypes.18, i32 0, i32 0), i8** null, i8** null, i32 0, i32 [[TMP15]])
11631 // CHECK11-NEXT: [[TMP17:%.*]] = icmp ne i32 [[TMP16]], 0
11632 // CHECK11-NEXT: br i1 [[TMP17]], label [[OMP_OFFLOAD_FAILED6:%.*]], label [[OMP_OFFLOAD_CONT7:%.*]]
11633 // CHECK11: omp_offload.failed6:
11634 // CHECK11-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiEiT__l67(i64 [[TMP5]]) #[[ATTR2]]
11635 // CHECK11-NEXT: br label [[OMP_OFFLOAD_CONT7]]
11636 // CHECK11: omp_offload.cont7:
11637 // CHECK11-NEXT: ret i32 0
11638 //
11639 //
11640 // CHECK11-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiEiT__l57
11641 // CHECK11-SAME: () #[[ATTR1]] {
11642 // CHECK11-NEXT: entry:
11643 // CHECK11-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 0, void (i32*, i32*, ...)* bitcast (void (i32*, i32*)* @.omp_outlined..11 to void (i32*, i32*, ...)*))
11644 // CHECK11-NEXT: ret void
11645 //
11646 //
11647 // CHECK11-LABEL: define {{[^@]+}}@.omp_outlined..11
11648 // CHECK11-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]]) #[[ATTR1]] {
11649 // CHECK11-NEXT: entry:
11650 // CHECK11-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
11651 // CHECK11-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
11652 // CHECK11-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
11653 // CHECK11-NEXT: [[TMP:%.*]] = alloca i32, align 4
11654 // CHECK11-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4
11655 // CHECK11-NEXT: [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4
11656 // CHECK11-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
11657 // CHECK11-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
11658 // CHECK11-NEXT: [[I:%.*]] = alloca i32, align 4
11659 // CHECK11-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
11660 // CHECK11-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
11661 // CHECK11-NEXT: store i32 0, i32* [[DOTOMP_COMB_LB]], align 4
11662 // CHECK11-NEXT: store i32 99, i32* [[DOTOMP_COMB_UB]], align 4
11663 // CHECK11-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4
11664 // CHECK11-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4
11665 // CHECK11-NEXT: [[TMP0:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
11666 // CHECK11-NEXT: [[TMP1:%.*]] = load i32, i32* [[TMP0]], align 4
11667 // CHECK11-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB1]], i32 [[TMP1]], i32 92, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_COMB_LB]], i32* [[DOTOMP_COMB_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1)
11668 // CHECK11-NEXT: [[TMP2:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
11669 // CHECK11-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP2]], 99
11670 // CHECK11-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
11671 // CHECK11: cond.true:
11672 // CHECK11-NEXT: br label [[COND_END:%.*]]
11673 // CHECK11: cond.false:
11674 // CHECK11-NEXT: [[TMP3:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
11675 // CHECK11-NEXT: br label [[COND_END]]
11676 // CHECK11: cond.end:
11677 // CHECK11-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP3]], [[COND_FALSE]] ]
11678 // CHECK11-NEXT: store i32 [[COND]], i32* [[DOTOMP_COMB_UB]], align 4
11679 // CHECK11-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4
11680 // CHECK11-NEXT: store i32 [[TMP4]], i32* [[DOTOMP_IV]], align 4
11681 // CHECK11-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
11682 // CHECK11: omp.inner.for.cond:
11683 // CHECK11-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
11684 // CHECK11-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
11685 // CHECK11-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP5]], [[TMP6]]
11686 // CHECK11-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
11687 // CHECK11: omp.inner.for.body:
11688 // CHECK11-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4
11689 // CHECK11-NEXT: [[TMP8:%.*]] = zext i32 [[TMP7]] to i64
11690 // CHECK11-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
11691 // CHECK11-NEXT: [[TMP10:%.*]] = zext i32 [[TMP9]] to i64
11692 // CHECK11-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB3]], i32 2, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64, i64)* @.omp_outlined..12 to void (i32*, i32*, ...)*), i64 [[TMP8]], i64 [[TMP10]])
11693 // CHECK11-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
11694 // CHECK11: omp.inner.for.inc:
11695 // CHECK11-NEXT: [[TMP11:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
11696 // CHECK11-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4
11697 // CHECK11-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP11]], [[TMP12]]
11698 // CHECK11-NEXT: store i32 [[ADD]], i32* [[DOTOMP_IV]], align 4
11699 // CHECK11-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP32:![0-9]+]]
11700 // CHECK11: omp.inner.for.end:
11701 // CHECK11-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
11702 // CHECK11: omp.loop.exit:
11703 // CHECK11-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP1]])
11704 // CHECK11-NEXT: [[TMP13:%.*]] = load i32, i32* [[DOTOMP_IS_LAST]], align 4
11705 // CHECK11-NEXT: [[TMP14:%.*]] = icmp ne i32 [[TMP13]], 0
11706 // CHECK11-NEXT: br i1 [[TMP14]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
11707 // CHECK11: .omp.final.then:
11708 // CHECK11-NEXT: store i32 100, i32* [[I]], align 4
11709 // CHECK11-NEXT: br label [[DOTOMP_FINAL_DONE]]
11710 // CHECK11: .omp.final.done:
11711 // CHECK11-NEXT: ret void
11712 //
11713 //
11714 // CHECK11-LABEL: define {{[^@]+}}@.omp_outlined..12
11715 // CHECK11-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[DOTPREVIOUS_LB_:%.*]], i64 [[DOTPREVIOUS_UB_:%.*]]) #[[ATTR1]] {
11716 // CHECK11-NEXT: entry:
11717 // CHECK11-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
11718 // CHECK11-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
11719 // CHECK11-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i64, align 8
11720 // CHECK11-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i64, align 8
11721 // CHECK11-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
11722 // CHECK11-NEXT: [[TMP:%.*]] = alloca i32, align 4
11723 // CHECK11-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4
11724 // CHECK11-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4
11725 // CHECK11-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
11726 // CHECK11-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
11727 // CHECK11-NEXT: [[I:%.*]] = alloca i32, align 4
11728 // CHECK11-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
11729 // CHECK11-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
11730 // CHECK11-NEXT: store i64 [[DOTPREVIOUS_LB_]], i64* [[DOTPREVIOUS_LB__ADDR]], align 8
11731 // CHECK11-NEXT: store i64 [[DOTPREVIOUS_UB_]], i64* [[DOTPREVIOUS_UB__ADDR]], align 8
11732 // CHECK11-NEXT: store i32 0, i32* [[DOTOMP_LB]], align 4
11733 // CHECK11-NEXT: store i32 99, i32* [[DOTOMP_UB]], align 4
11734 // CHECK11-NEXT: [[TMP0:%.*]] = load i64, i64* [[DOTPREVIOUS_LB__ADDR]], align 8
11735 // CHECK11-NEXT: [[CONV:%.*]] = trunc i64 [[TMP0]] to i32
11736 // CHECK11-NEXT: [[TMP1:%.*]] = load i64, i64* [[DOTPREVIOUS_UB__ADDR]], align 8
11737 // CHECK11-NEXT: [[CONV1:%.*]] = trunc i64 [[TMP1]] to i32
11738 // CHECK11-NEXT: store i32 [[CONV]], i32* [[DOTOMP_LB]], align 4
11739 // CHECK11-NEXT: store i32 [[CONV1]], i32* [[DOTOMP_UB]], align 4
11740 // CHECK11-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4
11741 // CHECK11-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4
11742 // CHECK11-NEXT: [[TMP2:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
11743 // CHECK11-NEXT: [[TMP3:%.*]] = load i32, i32* [[TMP2]], align 4
11744 // CHECK11-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB2]], i32 [[TMP3]], i32 34, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_LB]], i32* [[DOTOMP_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1)
11745 // CHECK11-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
11746 // CHECK11-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP4]], 99
11747 // CHECK11-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
11748 // CHECK11: cond.true:
11749 // CHECK11-NEXT: br label [[COND_END:%.*]]
11750 // CHECK11: cond.false:
11751 // CHECK11-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
11752 // CHECK11-NEXT: br label [[COND_END]]
11753 // CHECK11: cond.end:
11754 // CHECK11-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP5]], [[COND_FALSE]] ]
11755 // CHECK11-NEXT: store i32 [[COND]], i32* [[DOTOMP_UB]], align 4
11756 // CHECK11-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4
11757 // CHECK11-NEXT: store i32 [[TMP6]], i32* [[DOTOMP_IV]], align 4
11758 // CHECK11-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
11759 // CHECK11: omp.inner.for.cond:
11760 // CHECK11-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
11761 // CHECK11-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
11762 // CHECK11-NEXT: [[CMP2:%.*]] = icmp sle i32 [[TMP7]], [[TMP8]]
11763 // CHECK11-NEXT: br i1 [[CMP2]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
11764 // CHECK11: omp.inner.for.body:
11765 // CHECK11-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
11766 // CHECK11-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP9]], 1
11767 // CHECK11-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
11768 // CHECK11-NEXT: store i32 [[ADD]], i32* [[I]], align 4
11769 // CHECK11-NEXT: call void @_Z3fn1v()
11770 // CHECK11-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
11771 // CHECK11: omp.body.continue:
11772 // CHECK11-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
11773 // CHECK11: omp.inner.for.inc:
11774 // CHECK11-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
11775 // CHECK11-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP10]], 1
11776 // CHECK11-NEXT: store i32 [[ADD3]], i32* [[DOTOMP_IV]], align 4
11777 // CHECK11-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP33:![0-9]+]]
11778 // CHECK11: omp.inner.for.end:
11779 // CHECK11-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
11780 // CHECK11: omp.loop.exit:
11781 // CHECK11-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP3]])
11782 // CHECK11-NEXT: [[TMP11:%.*]] = load i32, i32* [[DOTOMP_IS_LAST]], align 4
11783 // CHECK11-NEXT: [[TMP12:%.*]] = icmp ne i32 [[TMP11]], 0
11784 // CHECK11-NEXT: br i1 [[TMP12]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
11785 // CHECK11: .omp.final.then:
11786 // CHECK11-NEXT: store i32 100, i32* [[I]], align 4
11787 // CHECK11-NEXT: br label [[DOTOMP_FINAL_DONE]]
11788 // CHECK11: .omp.final.done:
11789 // CHECK11-NEXT: ret void
11790 //
11791 //
11792 // CHECK11-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiEiT__l62
11793 // CHECK11-SAME: () #[[ATTR1]] {
11794 // CHECK11-NEXT: entry:
11795 // CHECK11-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 0, void (i32*, i32*, ...)* bitcast (void (i32*, i32*)* @.omp_outlined..13 to void (i32*, i32*, ...)*))
11796 // CHECK11-NEXT: ret void
11797 //
11798 //
11799 // CHECK11-LABEL: define {{[^@]+}}@.omp_outlined..13
11800 // CHECK11-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]]) #[[ATTR1]] {
11801 // CHECK11-NEXT: entry:
11802 // CHECK11-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
11803 // CHECK11-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
11804 // CHECK11-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
11805 // CHECK11-NEXT: [[TMP:%.*]] = alloca i32, align 4
11806 // CHECK11-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4
11807 // CHECK11-NEXT: [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4
11808 // CHECK11-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
11809 // CHECK11-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
11810 // CHECK11-NEXT: [[I:%.*]] = alloca i32, align 4
11811 // CHECK11-NEXT: [[DOTBOUND_ZERO_ADDR:%.*]] = alloca i32, align 4
11812 // CHECK11-NEXT: store i32 0, i32* [[DOTBOUND_ZERO_ADDR]], align 4
11813 // CHECK11-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
11814 // CHECK11-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
11815 // CHECK11-NEXT: store i32 0, i32* [[DOTOMP_COMB_LB]], align 4
11816 // CHECK11-NEXT: store i32 99, i32* [[DOTOMP_COMB_UB]], align 4
11817 // CHECK11-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4
11818 // CHECK11-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4
11819 // CHECK11-NEXT: [[TMP0:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
11820 // CHECK11-NEXT: [[TMP1:%.*]] = load i32, i32* [[TMP0]], align 4
11821 // CHECK11-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB1]], i32 [[TMP1]], i32 92, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_COMB_LB]], i32* [[DOTOMP_COMB_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1)
11822 // CHECK11-NEXT: [[TMP2:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
11823 // CHECK11-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP2]], 99
11824 // CHECK11-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
11825 // CHECK11: cond.true:
11826 // CHECK11-NEXT: br label [[COND_END:%.*]]
11827 // CHECK11: cond.false:
11828 // CHECK11-NEXT: [[TMP3:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
11829 // CHECK11-NEXT: br label [[COND_END]]
11830 // CHECK11: cond.end:
11831 // CHECK11-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP3]], [[COND_FALSE]] ]
11832 // CHECK11-NEXT: store i32 [[COND]], i32* [[DOTOMP_COMB_UB]], align 4
11833 // CHECK11-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4
11834 // CHECK11-NEXT: store i32 [[TMP4]], i32* [[DOTOMP_IV]], align 4
11835 // CHECK11-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
11836 // CHECK11: omp.inner.for.cond:
11837 // CHECK11-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
11838 // CHECK11-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
11839 // CHECK11-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP5]], [[TMP6]]
11840 // CHECK11-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
11841 // CHECK11: omp.inner.for.body:
11842 // CHECK11-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4
11843 // CHECK11-NEXT: [[TMP8:%.*]] = zext i32 [[TMP7]] to i64
11844 // CHECK11-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
11845 // CHECK11-NEXT: [[TMP10:%.*]] = zext i32 [[TMP9]] to i64
11846 // CHECK11-NEXT: call void @__kmpc_serialized_parallel(%struct.ident_t* @[[GLOB3]], i32 [[TMP1]])
11847 // CHECK11-NEXT: [[TMP11:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
11848 // CHECK11-NEXT: call void @.omp_outlined..14(i32* [[TMP11]], i32* [[DOTBOUND_ZERO_ADDR]], i64 [[TMP8]], i64 [[TMP10]]) #[[ATTR2]]
11849 // CHECK11-NEXT: call void @__kmpc_end_serialized_parallel(%struct.ident_t* @[[GLOB3]], i32 [[TMP1]])
11850 // CHECK11-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
11851 // CHECK11: omp.inner.for.inc:
11852 // CHECK11-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
11853 // CHECK11-NEXT: [[TMP13:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4
11854 // CHECK11-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP12]], [[TMP13]]
11855 // CHECK11-NEXT: store i32 [[ADD]], i32* [[DOTOMP_IV]], align 4
11856 // CHECK11-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP34:![0-9]+]]
11857 // CHECK11: omp.inner.for.end:
11858 // CHECK11-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
11859 // CHECK11: omp.loop.exit:
11860 // CHECK11-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP1]])
11861 // CHECK11-NEXT: [[TMP14:%.*]] = load i32, i32* [[DOTOMP_IS_LAST]], align 4
11862 // CHECK11-NEXT: [[TMP15:%.*]] = icmp ne i32 [[TMP14]], 0
11863 // CHECK11-NEXT: br i1 [[TMP15]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
11864 // CHECK11: .omp.final.then:
11865 // CHECK11-NEXT: store i32 100, i32* [[I]], align 4
11866 // CHECK11-NEXT: br label [[DOTOMP_FINAL_DONE]]
11867 // CHECK11: .omp.final.done:
11868 // CHECK11-NEXT: ret void
11869 //
11870 //
11871 // CHECK11-LABEL: define {{[^@]+}}@.omp_outlined..14
11872 // CHECK11-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[DOTPREVIOUS_LB_:%.*]], i64 [[DOTPREVIOUS_UB_:%.*]]) #[[ATTR1]] {
11873 // CHECK11-NEXT: entry:
11874 // CHECK11-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
11875 // CHECK11-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
11876 // CHECK11-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i64, align 8
11877 // CHECK11-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i64, align 8
11878 // CHECK11-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
11879 // CHECK11-NEXT: [[TMP:%.*]] = alloca i32, align 4
11880 // CHECK11-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4
11881 // CHECK11-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4
11882 // CHECK11-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
11883 // CHECK11-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
11884 // CHECK11-NEXT: [[I:%.*]] = alloca i32, align 4
11885 // CHECK11-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
11886 // CHECK11-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
11887 // CHECK11-NEXT: store i64 [[DOTPREVIOUS_LB_]], i64* [[DOTPREVIOUS_LB__ADDR]], align 8
11888 // CHECK11-NEXT: store i64 [[DOTPREVIOUS_UB_]], i64* [[DOTPREVIOUS_UB__ADDR]], align 8
11889 // CHECK11-NEXT: store i32 0, i32* [[DOTOMP_LB]], align 4
11890 // CHECK11-NEXT: store i32 99, i32* [[DOTOMP_UB]], align 4
11891 // CHECK11-NEXT: [[TMP0:%.*]] = load i64, i64* [[DOTPREVIOUS_LB__ADDR]], align 8
11892 // CHECK11-NEXT: [[CONV:%.*]] = trunc i64 [[TMP0]] to i32
11893 // CHECK11-NEXT: [[TMP1:%.*]] = load i64, i64* [[DOTPREVIOUS_UB__ADDR]], align 8
11894 // CHECK11-NEXT: [[CONV1:%.*]] = trunc i64 [[TMP1]] to i32
11895 // CHECK11-NEXT: store i32 [[CONV]], i32* [[DOTOMP_LB]], align 4
11896 // CHECK11-NEXT: store i32 [[CONV1]], i32* [[DOTOMP_UB]], align 4
11897 // CHECK11-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4
11898 // CHECK11-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4
11899 // CHECK11-NEXT: [[TMP2:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
11900 // CHECK11-NEXT: [[TMP3:%.*]] = load i32, i32* [[TMP2]], align 4
11901 // CHECK11-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB2]], i32 [[TMP3]], i32 34, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_LB]], i32* [[DOTOMP_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1)
11902 // CHECK11-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
11903 // CHECK11-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP4]], 99
11904 // CHECK11-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
11905 // CHECK11: cond.true:
11906 // CHECK11-NEXT: br label [[COND_END:%.*]]
11907 // CHECK11: cond.false:
11908 // CHECK11-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
11909 // CHECK11-NEXT: br label [[COND_END]]
11910 // CHECK11: cond.end:
11911 // CHECK11-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP5]], [[COND_FALSE]] ]
11912 // CHECK11-NEXT: store i32 [[COND]], i32* [[DOTOMP_UB]], align 4
11913 // CHECK11-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4
11914 // CHECK11-NEXT: store i32 [[TMP6]], i32* [[DOTOMP_IV]], align 4
11915 // CHECK11-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
11916 // CHECK11: omp.inner.for.cond:
11917 // CHECK11-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
11918 // CHECK11-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
11919 // CHECK11-NEXT: [[CMP2:%.*]] = icmp sle i32 [[TMP7]], [[TMP8]]
11920 // CHECK11-NEXT: br i1 [[CMP2]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
11921 // CHECK11: omp.inner.for.body:
11922 // CHECK11-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
11923 // CHECK11-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP9]], 1
11924 // CHECK11-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
11925 // CHECK11-NEXT: store i32 [[ADD]], i32* [[I]], align 4
11926 // CHECK11-NEXT: call void @_Z3fn2v()
11927 // CHECK11-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
11928 // CHECK11: omp.body.continue:
11929 // CHECK11-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
11930 // CHECK11: omp.inner.for.inc:
11931 // CHECK11-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
11932 // CHECK11-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP10]], 1
11933 // CHECK11-NEXT: store i32 [[ADD3]], i32* [[DOTOMP_IV]], align 4
11934 // CHECK11-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP35:![0-9]+]]
11935 // CHECK11: omp.inner.for.end:
11936 // CHECK11-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
11937 // CHECK11: omp.loop.exit:
11938 // CHECK11-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP3]])
11939 // CHECK11-NEXT: [[TMP11:%.*]] = load i32, i32* [[DOTOMP_IS_LAST]], align 4
11940 // CHECK11-NEXT: [[TMP12:%.*]] = icmp ne i32 [[TMP11]], 0
11941 // CHECK11-NEXT: br i1 [[TMP12]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
11942 // CHECK11: .omp.final.then:
11943 // CHECK11-NEXT: store i32 100, i32* [[I]], align 4
11944 // CHECK11-NEXT: br label [[DOTOMP_FINAL_DONE]]
11945 // CHECK11: .omp.final.done:
11946 // CHECK11-NEXT: ret void
11947 //
11948 //
11949 // CHECK11-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiEiT__l67
11950 // CHECK11-SAME: (i64 [[ARG:%.*]]) #[[ATTR1]] {
11951 // CHECK11-NEXT: entry:
11952 // CHECK11-NEXT: [[ARG_ADDR:%.*]] = alloca i64, align 8
11953 // CHECK11-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i8, align 1
11954 // CHECK11-NEXT: [[DOTCAPTURE_EXPR__CASTED:%.*]] = alloca i64, align 8
11955 // CHECK11-NEXT: store i64 [[ARG]], i64* [[ARG_ADDR]], align 8
11956 // CHECK11-NEXT: [[CONV:%.*]] = bitcast i64* [[ARG_ADDR]] to i32*
11957 // CHECK11-NEXT: [[TMP0:%.*]] = load i32, i32* [[CONV]], align 8
11958 // CHECK11-NEXT: [[TOBOOL:%.*]] = icmp ne i32 [[TMP0]], 0
11959 // CHECK11-NEXT: [[FROMBOOL:%.*]] = zext i1 [[TOBOOL]] to i8
11960 // CHECK11-NEXT: store i8 [[FROMBOOL]], i8* [[DOTCAPTURE_EXPR_]], align 1
11961 // CHECK11-NEXT: [[TMP1:%.*]] = load i8, i8* [[DOTCAPTURE_EXPR_]], align 1
11962 // CHECK11-NEXT: [[TOBOOL1:%.*]] = trunc i8 [[TMP1]] to i1
11963 // CHECK11-NEXT: [[CONV2:%.*]] = bitcast i64* [[DOTCAPTURE_EXPR__CASTED]] to i8*
11964 // CHECK11-NEXT: [[FROMBOOL3:%.*]] = zext i1 [[TOBOOL1]] to i8
11965 // CHECK11-NEXT: store i8 [[FROMBOOL3]], i8* [[CONV2]], align 1
11966 // CHECK11-NEXT: [[TMP2:%.*]] = load i64, i64* [[DOTCAPTURE_EXPR__CASTED]], align 8
11967 // CHECK11-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64)* @.omp_outlined..15 to void (i32*, i32*, ...)*), i64 [[TMP2]])
11968 // CHECK11-NEXT: ret void
11969 //
11970 //
11971 // CHECK11-LABEL: define {{[^@]+}}@.omp_outlined..15
11972 // CHECK11-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[DOTCAPTURE_EXPR_:%.*]]) #[[ATTR1]] {
11973 // CHECK11-NEXT: entry:
11974 // CHECK11-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
11975 // CHECK11-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
11976 // CHECK11-NEXT: [[DOTCAPTURE_EXPR__ADDR:%.*]] = alloca i64, align 8
11977 // CHECK11-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
11978 // CHECK11-NEXT: [[TMP:%.*]] = alloca i32, align 4
11979 // CHECK11-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4
11980 // CHECK11-NEXT: [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4
11981 // CHECK11-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
11982 // CHECK11-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
11983 // CHECK11-NEXT: [[I:%.*]] = alloca i32, align 4
11984 // CHECK11-NEXT: [[DOTBOUND_ZERO_ADDR:%.*]] = alloca i32, align 4
11985 // CHECK11-NEXT: store i32 0, i32* [[DOTBOUND_ZERO_ADDR]], align 4
11986 // CHECK11-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
11987 // CHECK11-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
11988 // CHECK11-NEXT: store i64 [[DOTCAPTURE_EXPR_]], i64* [[DOTCAPTURE_EXPR__ADDR]], align 8
11989 // CHECK11-NEXT: [[CONV:%.*]] = bitcast i64* [[DOTCAPTURE_EXPR__ADDR]] to i8*
11990 // CHECK11-NEXT: store i32 0, i32* [[DOTOMP_COMB_LB]], align 4
11991 // CHECK11-NEXT: store i32 99, i32* [[DOTOMP_COMB_UB]], align 4
11992 // CHECK11-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4
11993 // CHECK11-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4
11994 // CHECK11-NEXT: [[TMP0:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
11995 // CHECK11-NEXT: [[TMP1:%.*]] = load i32, i32* [[TMP0]], align 4
11996 // CHECK11-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB1]], i32 [[TMP1]], i32 92, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_COMB_LB]], i32* [[DOTOMP_COMB_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1)
11997 // CHECK11-NEXT: [[TMP2:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
11998 // CHECK11-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP2]], 99
11999 // CHECK11-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
12000 // CHECK11: cond.true:
12001 // CHECK11-NEXT: br label [[COND_END:%.*]]
12002 // CHECK11: cond.false:
12003 // CHECK11-NEXT: [[TMP3:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
12004 // CHECK11-NEXT: br label [[COND_END]]
12005 // CHECK11: cond.end:
12006 // CHECK11-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP3]], [[COND_FALSE]] ]
12007 // CHECK11-NEXT: store i32 [[COND]], i32* [[DOTOMP_COMB_UB]], align 4
12008 // CHECK11-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4
12009 // CHECK11-NEXT: store i32 [[TMP4]], i32* [[DOTOMP_IV]], align 4
12010 // CHECK11-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
12011 // CHECK11: omp.inner.for.cond:
12012 // CHECK11-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
12013 // CHECK11-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
12014 // CHECK11-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP5]], [[TMP6]]
12015 // CHECK11-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
12016 // CHECK11: omp.inner.for.body:
12017 // CHECK11-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4
12018 // CHECK11-NEXT: [[TMP8:%.*]] = zext i32 [[TMP7]] to i64
12019 // CHECK11-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
12020 // CHECK11-NEXT: [[TMP10:%.*]] = zext i32 [[TMP9]] to i64
12021 // CHECK11-NEXT: [[TMP11:%.*]] = load i8, i8* [[CONV]], align 8
12022 // CHECK11-NEXT: [[TOBOOL:%.*]] = trunc i8 [[TMP11]] to i1
12023 // CHECK11-NEXT: br i1 [[TOBOOL]], label [[OMP_IF_THEN:%.*]], label [[OMP_IF_ELSE:%.*]]
12024 // CHECK11: omp_if.then:
12025 // CHECK11-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB3]], i32 2, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64, i64)* @.omp_outlined..16 to void (i32*, i32*, ...)*), i64 [[TMP8]], i64 [[TMP10]])
12026 // CHECK11-NEXT: br label [[OMP_IF_END:%.*]]
12027 // CHECK11: omp_if.else:
12028 // CHECK11-NEXT: call void @__kmpc_serialized_parallel(%struct.ident_t* @[[GLOB3]], i32 [[TMP1]])
12029 // CHECK11-NEXT: [[TMP12:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
12030 // CHECK11-NEXT: call void @.omp_outlined..16(i32* [[TMP12]], i32* [[DOTBOUND_ZERO_ADDR]], i64 [[TMP8]], i64 [[TMP10]]) #[[ATTR2]]
12031 // CHECK11-NEXT: call void @__kmpc_end_serialized_parallel(%struct.ident_t* @[[GLOB3]], i32 [[TMP1]])
12032 // CHECK11-NEXT: br label [[OMP_IF_END]]
12033 // CHECK11: omp_if.end:
12034 // CHECK11-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
12035 // CHECK11: omp.inner.for.inc:
12036 // CHECK11-NEXT: [[TMP13:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
12037 // CHECK11-NEXT: [[TMP14:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4
12038 // CHECK11-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP13]], [[TMP14]]
12039 // CHECK11-NEXT: store i32 [[ADD]], i32* [[DOTOMP_IV]], align 4
12040 // CHECK11-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP36:![0-9]+]]
12041 // CHECK11: omp.inner.for.end:
12042 // CHECK11-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
12043 // CHECK11: omp.loop.exit:
12044 // CHECK11-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP1]])
12045 // CHECK11-NEXT: [[TMP15:%.*]] = load i32, i32* [[DOTOMP_IS_LAST]], align 4
12046 // CHECK11-NEXT: [[TMP16:%.*]] = icmp ne i32 [[TMP15]], 0
12047 // CHECK11-NEXT: br i1 [[TMP16]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
12048 // CHECK11: .omp.final.then:
12049 // CHECK11-NEXT: store i32 100, i32* [[I]], align 4
12050 // CHECK11-NEXT: br label [[DOTOMP_FINAL_DONE]]
12051 // CHECK11: .omp.final.done:
12052 // CHECK11-NEXT: ret void
12053 //
12054 //
12055 // CHECK11-LABEL: define {{[^@]+}}@.omp_outlined..16
12056 // CHECK11-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[DOTPREVIOUS_LB_:%.*]], i64 [[DOTPREVIOUS_UB_:%.*]]) #[[ATTR1]] {
12057 // CHECK11-NEXT: entry:
12058 // CHECK11-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
12059 // CHECK11-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
12060 // CHECK11-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i64, align 8
12061 // CHECK11-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i64, align 8
12062 // CHECK11-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
12063 // CHECK11-NEXT: [[TMP:%.*]] = alloca i32, align 4
12064 // CHECK11-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4
12065 // CHECK11-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4
12066 // CHECK11-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
12067 // CHECK11-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
12068 // CHECK11-NEXT: [[I:%.*]] = alloca i32, align 4
12069 // CHECK11-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
12070 // CHECK11-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
12071 // CHECK11-NEXT: store i64 [[DOTPREVIOUS_LB_]], i64* [[DOTPREVIOUS_LB__ADDR]], align 8
12072 // CHECK11-NEXT: store i64 [[DOTPREVIOUS_UB_]], i64* [[DOTPREVIOUS_UB__ADDR]], align 8
12073 // CHECK11-NEXT: store i32 0, i32* [[DOTOMP_LB]], align 4
12074 // CHECK11-NEXT: store i32 99, i32* [[DOTOMP_UB]], align 4
12075 // CHECK11-NEXT: [[TMP0:%.*]] = load i64, i64* [[DOTPREVIOUS_LB__ADDR]], align 8
12076 // CHECK11-NEXT: [[CONV:%.*]] = trunc i64 [[TMP0]] to i32
12077 // CHECK11-NEXT: [[TMP1:%.*]] = load i64, i64* [[DOTPREVIOUS_UB__ADDR]], align 8
12078 // CHECK11-NEXT: [[CONV1:%.*]] = trunc i64 [[TMP1]] to i32
12079 // CHECK11-NEXT: store i32 [[CONV]], i32* [[DOTOMP_LB]], align 4
12080 // CHECK11-NEXT: store i32 [[CONV1]], i32* [[DOTOMP_UB]], align 4
12081 // CHECK11-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4
12082 // CHECK11-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4
12083 // CHECK11-NEXT: [[TMP2:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
12084 // CHECK11-NEXT: [[TMP3:%.*]] = load i32, i32* [[TMP2]], align 4
12085 // CHECK11-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB2]], i32 [[TMP3]], i32 34, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_LB]], i32* [[DOTOMP_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1)
12086 // CHECK11-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
12087 // CHECK11-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP4]], 99
12088 // CHECK11-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
12089 // CHECK11: cond.true:
12090 // CHECK11-NEXT: br label [[COND_END:%.*]]
12091 // CHECK11: cond.false:
12092 // CHECK11-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
12093 // CHECK11-NEXT: br label [[COND_END]]
12094 // CHECK11: cond.end:
12095 // CHECK11-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP5]], [[COND_FALSE]] ]
12096 // CHECK11-NEXT: store i32 [[COND]], i32* [[DOTOMP_UB]], align 4
12097 // CHECK11-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4
12098 // CHECK11-NEXT: store i32 [[TMP6]], i32* [[DOTOMP_IV]], align 4
12099 // CHECK11-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
12100 // CHECK11: omp.inner.for.cond:
12101 // CHECK11-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
12102 // CHECK11-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
12103 // CHECK11-NEXT: [[CMP2:%.*]] = icmp sle i32 [[TMP7]], [[TMP8]]
12104 // CHECK11-NEXT: br i1 [[CMP2]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
12105 // CHECK11: omp.inner.for.body:
12106 // CHECK11-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
12107 // CHECK11-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP9]], 1
12108 // CHECK11-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
12109 // CHECK11-NEXT: store i32 [[ADD]], i32* [[I]], align 4
12110 // CHECK11-NEXT: call void @_Z3fn3v()
12111 // CHECK11-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
12112 // CHECK11: omp.body.continue:
12113 // CHECK11-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
12114 // CHECK11: omp.inner.for.inc:
12115 // CHECK11-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
12116 // CHECK11-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP10]], 1
12117 // CHECK11-NEXT: store i32 [[ADD3]], i32* [[DOTOMP_IV]], align 4
12118 // CHECK11-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP37:![0-9]+]]
12119 // CHECK11: omp.inner.for.end:
12120 // CHECK11-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
12121 // CHECK11: omp.loop.exit:
12122 // CHECK11-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP3]])
12123 // CHECK11-NEXT: [[TMP11:%.*]] = load i32, i32* [[DOTOMP_IS_LAST]], align 4
12124 // CHECK11-NEXT: [[TMP12:%.*]] = icmp ne i32 [[TMP11]], 0
12125 // CHECK11-NEXT: br i1 [[TMP12]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
12126 // CHECK11: .omp.final.then:
12127 // CHECK11-NEXT: store i32 100, i32* [[I]], align 4
12128 // CHECK11-NEXT: br label [[DOTOMP_FINAL_DONE]]
12129 // CHECK11: .omp.final.done:
12130 // CHECK11-NEXT: ret void
12131 //
12132 //
12133 // CHECK11-LABEL: define {{[^@]+}}@.omp_offloading.requires_reg
12134 // CHECK11-SAME: () #[[ATTR5:[0-9]+]] {
12135 // CHECK11-NEXT: entry:
12136 // CHECK11-NEXT: call void @__tgt_register_requires(i64 1)
12137 // CHECK11-NEXT: ret void
12138 //
12139 //
12140 // CHECK12-LABEL: define {{[^@]+}}@_Z9gtid_testv
12141 // CHECK12-SAME: () #[[ATTR0:[0-9]+]] {
12142 // CHECK12-NEXT: entry:
12143 // CHECK12-NEXT: [[TMP:%.*]] = alloca i32, align 4
12144 // CHECK12-NEXT: [[_TMP1:%.*]] = alloca i32, align 4
12145 // CHECK12-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3:[0-9]+]], i64 -1, i64 100)
12146 // CHECK12-NEXT: [[TMP0:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9gtid_testv_l43.region_id, i32 0, i8** null, i8** null, i64* null, i64* null, i8** null, i8** null, i32 0, i32 0)
12147 // CHECK12-NEXT: [[TMP1:%.*]] = icmp ne i32 [[TMP0]], 0
12148 // CHECK12-NEXT: br i1 [[TMP1]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
12149 // CHECK12: omp_offload.failed:
12150 // CHECK12-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9gtid_testv_l43() #[[ATTR2:[0-9]+]]
12151 // CHECK12-NEXT: br label [[OMP_OFFLOAD_CONT]]
12152 // CHECK12: omp_offload.cont:
12153 // CHECK12-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i64 100)
12154 // CHECK12-NEXT: [[TMP2:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9gtid_testv_l47.region_id, i32 0, i8** null, i8** null, i64* null, i64* null, i8** null, i8** null, i32 0, i32 1)
12155 // CHECK12-NEXT: [[TMP3:%.*]] = icmp ne i32 [[TMP2]], 0
12156 // CHECK12-NEXT: br i1 [[TMP3]], label [[OMP_OFFLOAD_FAILED2:%.*]], label [[OMP_OFFLOAD_CONT3:%.*]]
12157 // CHECK12: omp_offload.failed2:
12158 // CHECK12-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9gtid_testv_l47() #[[ATTR2]]
12159 // CHECK12-NEXT: br label [[OMP_OFFLOAD_CONT3]]
12160 // CHECK12: omp_offload.cont3:
12161 // CHECK12-NEXT: ret void
12162 //
12163 //
12164 // CHECK12-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9gtid_testv_l43
12165 // CHECK12-SAME: () #[[ATTR1:[0-9]+]] {
12166 // CHECK12-NEXT: entry:
12167 // CHECK12-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 0, void (i32*, i32*, ...)* bitcast (void (i32*, i32*)* @.omp_outlined. to void (i32*, i32*, ...)*))
12168 // CHECK12-NEXT: ret void
12169 //
12170 //
12171 // CHECK12-LABEL: define {{[^@]+}}@.omp_outlined.
12172 // CHECK12-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]]) #[[ATTR1]] {
12173 // CHECK12-NEXT: entry:
12174 // CHECK12-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
12175 // CHECK12-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
12176 // CHECK12-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
12177 // CHECK12-NEXT: [[TMP:%.*]] = alloca i32, align 4
12178 // CHECK12-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4
12179 // CHECK12-NEXT: [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4
12180 // CHECK12-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
12181 // CHECK12-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
12182 // CHECK12-NEXT: [[I:%.*]] = alloca i32, align 4
12183 // CHECK12-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
12184 // CHECK12-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
12185 // CHECK12-NEXT: store i32 0, i32* [[DOTOMP_COMB_LB]], align 4
12186 // CHECK12-NEXT: store i32 99, i32* [[DOTOMP_COMB_UB]], align 4
12187 // CHECK12-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4
12188 // CHECK12-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4
12189 // CHECK12-NEXT: [[TMP0:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
12190 // CHECK12-NEXT: [[TMP1:%.*]] = load i32, i32* [[TMP0]], align 4
12191 // CHECK12-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB1:[0-9]+]], i32 [[TMP1]], i32 92, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_COMB_LB]], i32* [[DOTOMP_COMB_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1)
12192 // CHECK12-NEXT: [[TMP2:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
12193 // CHECK12-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP2]], 99
12194 // CHECK12-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
12195 // CHECK12: cond.true:
12196 // CHECK12-NEXT: br label [[COND_END:%.*]]
12197 // CHECK12: cond.false:
12198 // CHECK12-NEXT: [[TMP3:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
12199 // CHECK12-NEXT: br label [[COND_END]]
12200 // CHECK12: cond.end:
12201 // CHECK12-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP3]], [[COND_FALSE]] ]
12202 // CHECK12-NEXT: store i32 [[COND]], i32* [[DOTOMP_COMB_UB]], align 4
12203 // CHECK12-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4
12204 // CHECK12-NEXT: store i32 [[TMP4]], i32* [[DOTOMP_IV]], align 4
12205 // CHECK12-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
12206 // CHECK12: omp.inner.for.cond:
12207 // CHECK12-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
12208 // CHECK12-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
12209 // CHECK12-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP5]], [[TMP6]]
12210 // CHECK12-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
12211 // CHECK12: omp.inner.for.body:
12212 // CHECK12-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4
12213 // CHECK12-NEXT: [[TMP8:%.*]] = zext i32 [[TMP7]] to i64
12214 // CHECK12-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
12215 // CHECK12-NEXT: [[TMP10:%.*]] = zext i32 [[TMP9]] to i64
12216 // CHECK12-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB3]], i32 2, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64, i64)* @.omp_outlined..1 to void (i32*, i32*, ...)*), i64 [[TMP8]], i64 [[TMP10]])
12217 // CHECK12-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
12218 // CHECK12: omp.inner.for.inc:
12219 // CHECK12-NEXT: [[TMP11:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
12220 // CHECK12-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4
12221 // CHECK12-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP11]], [[TMP12]]
12222 // CHECK12-NEXT: store i32 [[ADD]], i32* [[DOTOMP_IV]], align 4
12223 // CHECK12-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP14:![0-9]+]]
12224 // CHECK12: omp.inner.for.end:
12225 // CHECK12-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
12226 // CHECK12: omp.loop.exit:
12227 // CHECK12-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP1]])
12228 // CHECK12-NEXT: [[TMP13:%.*]] = load i32, i32* [[DOTOMP_IS_LAST]], align 4
12229 // CHECK12-NEXT: [[TMP14:%.*]] = icmp ne i32 [[TMP13]], 0
12230 // CHECK12-NEXT: br i1 [[TMP14]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
12231 // CHECK12: .omp.final.then:
12232 // CHECK12-NEXT: store i32 100, i32* [[I]], align 4
12233 // CHECK12-NEXT: br label [[DOTOMP_FINAL_DONE]]
12234 // CHECK12: .omp.final.done:
12235 // CHECK12-NEXT: ret void
12236 //
12237 //
12238 // CHECK12-LABEL: define {{[^@]+}}@.omp_outlined..1
12239 // CHECK12-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[DOTPREVIOUS_LB_:%.*]], i64 [[DOTPREVIOUS_UB_:%.*]]) #[[ATTR1]] {
12240 // CHECK12-NEXT: entry:
12241 // CHECK12-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
12242 // CHECK12-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
12243 // CHECK12-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i64, align 8
12244 // CHECK12-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i64, align 8
12245 // CHECK12-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
12246 // CHECK12-NEXT: [[TMP:%.*]] = alloca i32, align 4
12247 // CHECK12-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4
12248 // CHECK12-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4
12249 // CHECK12-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
12250 // CHECK12-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
12251 // CHECK12-NEXT: [[I:%.*]] = alloca i32, align 4
12252 // CHECK12-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
12253 // CHECK12-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
12254 // CHECK12-NEXT: store i64 [[DOTPREVIOUS_LB_]], i64* [[DOTPREVIOUS_LB__ADDR]], align 8
12255 // CHECK12-NEXT: store i64 [[DOTPREVIOUS_UB_]], i64* [[DOTPREVIOUS_UB__ADDR]], align 8
12256 // CHECK12-NEXT: store i32 0, i32* [[DOTOMP_LB]], align 4
12257 // CHECK12-NEXT: store i32 99, i32* [[DOTOMP_UB]], align 4
12258 // CHECK12-NEXT: [[TMP0:%.*]] = load i64, i64* [[DOTPREVIOUS_LB__ADDR]], align 8
12259 // CHECK12-NEXT: [[CONV:%.*]] = trunc i64 [[TMP0]] to i32
12260 // CHECK12-NEXT: [[TMP1:%.*]] = load i64, i64* [[DOTPREVIOUS_UB__ADDR]], align 8
12261 // CHECK12-NEXT: [[CONV1:%.*]] = trunc i64 [[TMP1]] to i32
12262 // CHECK12-NEXT: store i32 [[CONV]], i32* [[DOTOMP_LB]], align 4
12263 // CHECK12-NEXT: store i32 [[CONV1]], i32* [[DOTOMP_UB]], align 4
12264 // CHECK12-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4
12265 // CHECK12-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4
12266 // CHECK12-NEXT: [[TMP2:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
12267 // CHECK12-NEXT: [[TMP3:%.*]] = load i32, i32* [[TMP2]], align 4
12268 // CHECK12-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB2:[0-9]+]], i32 [[TMP3]], i32 34, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_LB]], i32* [[DOTOMP_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1)
12269 // CHECK12-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
12270 // CHECK12-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP4]], 99
12271 // CHECK12-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
12272 // CHECK12: cond.true:
12273 // CHECK12-NEXT: br label [[COND_END:%.*]]
12274 // CHECK12: cond.false:
12275 // CHECK12-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
12276 // CHECK12-NEXT: br label [[COND_END]]
12277 // CHECK12: cond.end:
12278 // CHECK12-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP5]], [[COND_FALSE]] ]
12279 // CHECK12-NEXT: store i32 [[COND]], i32* [[DOTOMP_UB]], align 4
12280 // CHECK12-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4
12281 // CHECK12-NEXT: store i32 [[TMP6]], i32* [[DOTOMP_IV]], align 4
12282 // CHECK12-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
12283 // CHECK12: omp.inner.for.cond:
12284 // CHECK12-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
12285 // CHECK12-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
12286 // CHECK12-NEXT: [[CMP2:%.*]] = icmp sle i32 [[TMP7]], [[TMP8]]
12287 // CHECK12-NEXT: br i1 [[CMP2]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
12288 // CHECK12: omp.inner.for.body:
12289 // CHECK12-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
12290 // CHECK12-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP9]], 1
12291 // CHECK12-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
12292 // CHECK12-NEXT: store i32 [[ADD]], i32* [[I]], align 4
12293 // CHECK12-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
12294 // CHECK12: omp.body.continue:
12295 // CHECK12-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
12296 // CHECK12: omp.inner.for.inc:
12297 // CHECK12-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
12298 // CHECK12-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP10]], 1
12299 // CHECK12-NEXT: store i32 [[ADD3]], i32* [[DOTOMP_IV]], align 4
12300 // CHECK12-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP16:![0-9]+]]
12301 // CHECK12: omp.inner.for.end:
12302 // CHECK12-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
12303 // CHECK12: omp.loop.exit:
12304 // CHECK12-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP3]])
12305 // CHECK12-NEXT: [[TMP11:%.*]] = load i32, i32* [[DOTOMP_IS_LAST]], align 4
12306 // CHECK12-NEXT: [[TMP12:%.*]] = icmp ne i32 [[TMP11]], 0
12307 // CHECK12-NEXT: br i1 [[TMP12]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
12308 // CHECK12: .omp.final.then:
12309 // CHECK12-NEXT: store i32 100, i32* [[I]], align 4
12310 // CHECK12-NEXT: br label [[DOTOMP_FINAL_DONE]]
12311 // CHECK12: .omp.final.done:
12312 // CHECK12-NEXT: ret void
12313 //
12314 //
12315 // CHECK12-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9gtid_testv_l47
12316 // CHECK12-SAME: () #[[ATTR1]] {
12317 // CHECK12-NEXT: entry:
12318 // CHECK12-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 0, void (i32*, i32*, ...)* bitcast (void (i32*, i32*)* @.omp_outlined..2 to void (i32*, i32*, ...)*))
12319 // CHECK12-NEXT: ret void
12320 //
12321 //
12322 // CHECK12-LABEL: define {{[^@]+}}@.omp_outlined..2
12323 // CHECK12-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]]) #[[ATTR1]] {
12324 // CHECK12-NEXT: entry:
12325 // CHECK12-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
12326 // CHECK12-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
12327 // CHECK12-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
12328 // CHECK12-NEXT: [[TMP:%.*]] = alloca i32, align 4
12329 // CHECK12-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4
12330 // CHECK12-NEXT: [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4
12331 // CHECK12-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
12332 // CHECK12-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
12333 // CHECK12-NEXT: [[I:%.*]] = alloca i32, align 4
12334 // CHECK12-NEXT: [[DOTBOUND_ZERO_ADDR:%.*]] = alloca i32, align 4
12335 // CHECK12-NEXT: store i32 0, i32* [[DOTBOUND_ZERO_ADDR]], align 4
12336 // CHECK12-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
12337 // CHECK12-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
12338 // CHECK12-NEXT: store i32 0, i32* [[DOTOMP_COMB_LB]], align 4
12339 // CHECK12-NEXT: store i32 99, i32* [[DOTOMP_COMB_UB]], align 4
12340 // CHECK12-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4
12341 // CHECK12-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4
12342 // CHECK12-NEXT: [[TMP0:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
12343 // CHECK12-NEXT: [[TMP1:%.*]] = load i32, i32* [[TMP0]], align 4
12344 // CHECK12-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB1]], i32 [[TMP1]], i32 92, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_COMB_LB]], i32* [[DOTOMP_COMB_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1)
12345 // CHECK12-NEXT: [[TMP2:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
12346 // CHECK12-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP2]], 99
12347 // CHECK12-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
12348 // CHECK12: cond.true:
12349 // CHECK12-NEXT: br label [[COND_END:%.*]]
12350 // CHECK12: cond.false:
12351 // CHECK12-NEXT: [[TMP3:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
12352 // CHECK12-NEXT: br label [[COND_END]]
12353 // CHECK12: cond.end:
12354 // CHECK12-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP3]], [[COND_FALSE]] ]
12355 // CHECK12-NEXT: store i32 [[COND]], i32* [[DOTOMP_COMB_UB]], align 4
12356 // CHECK12-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4
12357 // CHECK12-NEXT: store i32 [[TMP4]], i32* [[DOTOMP_IV]], align 4
12358 // CHECK12-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
12359 // CHECK12: omp.inner.for.cond:
12360 // CHECK12-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
12361 // CHECK12-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
12362 // CHECK12-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP5]], [[TMP6]]
12363 // CHECK12-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
12364 // CHECK12: omp.inner.for.body:
12365 // CHECK12-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4
12366 // CHECK12-NEXT: [[TMP8:%.*]] = zext i32 [[TMP7]] to i64
12367 // CHECK12-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
12368 // CHECK12-NEXT: [[TMP10:%.*]] = zext i32 [[TMP9]] to i64
12369 // CHECK12-NEXT: call void @__kmpc_serialized_parallel(%struct.ident_t* @[[GLOB3]], i32 [[TMP1]])
12370 // CHECK12-NEXT: [[TMP11:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
12371 // CHECK12-NEXT: call void @.omp_outlined..3(i32* [[TMP11]], i32* [[DOTBOUND_ZERO_ADDR]], i64 [[TMP8]], i64 [[TMP10]]) #[[ATTR2]]
12372 // CHECK12-NEXT: call void @__kmpc_end_serialized_parallel(%struct.ident_t* @[[GLOB3]], i32 [[TMP1]])
12373 // CHECK12-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
12374 // CHECK12: omp.inner.for.inc:
12375 // CHECK12-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
12376 // CHECK12-NEXT: [[TMP13:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4
12377 // CHECK12-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP12]], [[TMP13]]
12378 // CHECK12-NEXT: store i32 [[ADD]], i32* [[DOTOMP_IV]], align 4
12379 // CHECK12-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP19:![0-9]+]]
12380 // CHECK12: omp.inner.for.end:
12381 // CHECK12-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
12382 // CHECK12: omp.loop.exit:
12383 // CHECK12-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP1]])
12384 // CHECK12-NEXT: [[TMP14:%.*]] = load i32, i32* [[DOTOMP_IS_LAST]], align 4
12385 // CHECK12-NEXT: [[TMP15:%.*]] = icmp ne i32 [[TMP14]], 0
12386 // CHECK12-NEXT: br i1 [[TMP15]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
12387 // CHECK12: .omp.final.then:
12388 // CHECK12-NEXT: store i32 100, i32* [[I]], align 4
12389 // CHECK12-NEXT: br label [[DOTOMP_FINAL_DONE]]
12390 // CHECK12: .omp.final.done:
12391 // CHECK12-NEXT: ret void
12392 //
12393 //
12394 // CHECK12-LABEL: define {{[^@]+}}@.omp_outlined..3
12395 // CHECK12-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[DOTPREVIOUS_LB_:%.*]], i64 [[DOTPREVIOUS_UB_:%.*]]) #[[ATTR1]] {
12396 // CHECK12-NEXT: entry:
12397 // CHECK12-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
12398 // CHECK12-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
12399 // CHECK12-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i64, align 8
12400 // CHECK12-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i64, align 8
12401 // CHECK12-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
12402 // CHECK12-NEXT: [[TMP:%.*]] = alloca i32, align 4
12403 // CHECK12-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4
12404 // CHECK12-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4
12405 // CHECK12-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
12406 // CHECK12-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
12407 // CHECK12-NEXT: [[I:%.*]] = alloca i32, align 4
12408 // CHECK12-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
12409 // CHECK12-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
12410 // CHECK12-NEXT: store i64 [[DOTPREVIOUS_LB_]], i64* [[DOTPREVIOUS_LB__ADDR]], align 8
12411 // CHECK12-NEXT: store i64 [[DOTPREVIOUS_UB_]], i64* [[DOTPREVIOUS_UB__ADDR]], align 8
12412 // CHECK12-NEXT: store i32 0, i32* [[DOTOMP_LB]], align 4
12413 // CHECK12-NEXT: store i32 99, i32* [[DOTOMP_UB]], align 4
12414 // CHECK12-NEXT: [[TMP0:%.*]] = load i64, i64* [[DOTPREVIOUS_LB__ADDR]], align 8
12415 // CHECK12-NEXT: [[CONV:%.*]] = trunc i64 [[TMP0]] to i32
12416 // CHECK12-NEXT: [[TMP1:%.*]] = load i64, i64* [[DOTPREVIOUS_UB__ADDR]], align 8
12417 // CHECK12-NEXT: [[CONV1:%.*]] = trunc i64 [[TMP1]] to i32
12418 // CHECK12-NEXT: store i32 [[CONV]], i32* [[DOTOMP_LB]], align 4
12419 // CHECK12-NEXT: store i32 [[CONV1]], i32* [[DOTOMP_UB]], align 4
12420 // CHECK12-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4
12421 // CHECK12-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4
12422 // CHECK12-NEXT: [[TMP2:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
12423 // CHECK12-NEXT: [[TMP3:%.*]] = load i32, i32* [[TMP2]], align 4
12424 // CHECK12-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB2]], i32 [[TMP3]], i32 34, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_LB]], i32* [[DOTOMP_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1)
12425 // CHECK12-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
12426 // CHECK12-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP4]], 99
12427 // CHECK12-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
12428 // CHECK12: cond.true:
12429 // CHECK12-NEXT: br label [[COND_END:%.*]]
12430 // CHECK12: cond.false:
12431 // CHECK12-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
12432 // CHECK12-NEXT: br label [[COND_END]]
12433 // CHECK12: cond.end:
12434 // CHECK12-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP5]], [[COND_FALSE]] ]
12435 // CHECK12-NEXT: store i32 [[COND]], i32* [[DOTOMP_UB]], align 4
12436 // CHECK12-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4
12437 // CHECK12-NEXT: store i32 [[TMP6]], i32* [[DOTOMP_IV]], align 4
12438 // CHECK12-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
12439 // CHECK12: omp.inner.for.cond:
12440 // CHECK12-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
12441 // CHECK12-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
12442 // CHECK12-NEXT: [[CMP2:%.*]] = icmp sle i32 [[TMP7]], [[TMP8]]
12443 // CHECK12-NEXT: br i1 [[CMP2]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
12444 // CHECK12: omp.inner.for.body:
12445 // CHECK12-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
12446 // CHECK12-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP9]], 1
12447 // CHECK12-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
12448 // CHECK12-NEXT: store i32 [[ADD]], i32* [[I]], align 4
12449 // CHECK12-NEXT: call void @_Z9gtid_testv()
12450 // CHECK12-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
12451 // CHECK12: omp.body.continue:
12452 // CHECK12-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
12453 // CHECK12: omp.inner.for.inc:
12454 // CHECK12-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
12455 // CHECK12-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP10]], 1
12456 // CHECK12-NEXT: store i32 [[ADD3]], i32* [[DOTOMP_IV]], align 4
12457 // CHECK12-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP20:![0-9]+]]
12458 // CHECK12: omp.inner.for.end:
12459 // CHECK12-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
12460 // CHECK12: omp.loop.exit:
12461 // CHECK12-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP3]])
12462 // CHECK12-NEXT: [[TMP11:%.*]] = load i32, i32* [[DOTOMP_IS_LAST]], align 4
12463 // CHECK12-NEXT: [[TMP12:%.*]] = icmp ne i32 [[TMP11]], 0
12464 // CHECK12-NEXT: br i1 [[TMP12]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
12465 // CHECK12: .omp.final.then:
12466 // CHECK12-NEXT: store i32 100, i32* [[I]], align 4
12467 // CHECK12-NEXT: br label [[DOTOMP_FINAL_DONE]]
12468 // CHECK12: .omp.final.done:
12469 // CHECK12-NEXT: ret void
12470 //
12471 //
12472 // CHECK12-LABEL: define {{[^@]+}}@main
12473 // CHECK12-SAME: () #[[ATTR3:[0-9]+]] {
12474 // CHECK12-NEXT: entry:
12475 // CHECK12-NEXT: [[RETVAL:%.*]] = alloca i32, align 4
12476 // CHECK12-NEXT: [[TMP:%.*]] = alloca i32, align 4
12477 // CHECK12-NEXT: [[_TMP1:%.*]] = alloca i32, align 4
12478 // CHECK12-NEXT: [[ARG_CASTED:%.*]] = alloca i64, align 8
12479 // CHECK12-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [1 x i8*], align 8
12480 // CHECK12-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [1 x i8*], align 8
12481 // CHECK12-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [1 x i8*], align 8
12482 // CHECK12-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i8, align 1
12483 // CHECK12-NEXT: [[_TMP5:%.*]] = alloca i32, align 4
12484 // CHECK12-NEXT: store i32 0, i32* [[RETVAL]], align 4
12485 // CHECK12-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i64 100)
12486 // CHECK12-NEXT: [[TMP0:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l76.region_id, i32 0, i8** null, i8** null, i64* null, i64* null, i8** null, i8** null, i32 0, i32 0)
12487 // CHECK12-NEXT: [[TMP1:%.*]] = icmp ne i32 [[TMP0]], 0
12488 // CHECK12-NEXT: br i1 [[TMP1]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
12489 // CHECK12: omp_offload.failed:
12490 // CHECK12-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l76() #[[ATTR2]]
12491 // CHECK12-NEXT: br label [[OMP_OFFLOAD_CONT]]
12492 // CHECK12: omp_offload.cont:
12493 // CHECK12-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i64 100)
12494 // CHECK12-NEXT: [[TMP2:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l84.region_id, i32 0, i8** null, i8** null, i64* null, i64* null, i8** null, i8** null, i32 0, i32 1)
12495 // CHECK12-NEXT: [[TMP3:%.*]] = icmp ne i32 [[TMP2]], 0
12496 // CHECK12-NEXT: br i1 [[TMP3]], label [[OMP_OFFLOAD_FAILED2:%.*]], label [[OMP_OFFLOAD_CONT3:%.*]]
12497 // CHECK12: omp_offload.failed2:
12498 // CHECK12-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l84() #[[ATTR2]]
12499 // CHECK12-NEXT: br label [[OMP_OFFLOAD_CONT3]]
12500 // CHECK12: omp_offload.cont3:
12501 // CHECK12-NEXT: [[TMP4:%.*]] = load i32, i32* @Arg, align 4
12502 // CHECK12-NEXT: [[CONV:%.*]] = bitcast i64* [[ARG_CASTED]] to i32*
12503 // CHECK12-NEXT: store i32 [[TMP4]], i32* [[CONV]], align 4
12504 // CHECK12-NEXT: [[TMP5:%.*]] = load i64, i64* [[ARG_CASTED]], align 8
12505 // CHECK12-NEXT: [[TMP6:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
12506 // CHECK12-NEXT: [[TMP7:%.*]] = bitcast i8** [[TMP6]] to i64*
12507 // CHECK12-NEXT: store i64 [[TMP5]], i64* [[TMP7]], align 8
12508 // CHECK12-NEXT: [[TMP8:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
12509 // CHECK12-NEXT: [[TMP9:%.*]] = bitcast i8** [[TMP8]] to i64*
12510 // CHECK12-NEXT: store i64 [[TMP5]], i64* [[TMP9]], align 8
12511 // CHECK12-NEXT: [[TMP10:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0
12512 // CHECK12-NEXT: store i8* null, i8** [[TMP10]], align 8
12513 // CHECK12-NEXT: [[TMP11:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
12514 // CHECK12-NEXT: [[TMP12:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
12515 // CHECK12-NEXT: [[TMP13:%.*]] = load i32, i32* @Arg, align 4
12516 // CHECK12-NEXT: [[TOBOOL:%.*]] = icmp ne i32 [[TMP13]], 0
12517 // CHECK12-NEXT: [[FROMBOOL:%.*]] = zext i1 [[TOBOOL]] to i8
12518 // CHECK12-NEXT: store i8 [[FROMBOOL]], i8* [[DOTCAPTURE_EXPR_]], align 1
12519 // CHECK12-NEXT: [[TMP14:%.*]] = load i8, i8* [[DOTCAPTURE_EXPR_]], align 1
12520 // CHECK12-NEXT: [[TOBOOL4:%.*]] = trunc i8 [[TMP14]] to i1
12521 // CHECK12-NEXT: [[TMP15:%.*]] = select i1 [[TOBOOL4]], i32 0, i32 1
12522 // CHECK12-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i64 100)
12523 // CHECK12-NEXT: [[TMP16:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l92.region_id, i32 1, i8** [[TMP11]], i8** [[TMP12]], i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_sizes, i32 0, i32 0), i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_maptypes, i32 0, i32 0), i8** null, i8** null, i32 0, i32 [[TMP15]])
12524 // CHECK12-NEXT: [[TMP17:%.*]] = icmp ne i32 [[TMP16]], 0
12525 // CHECK12-NEXT: br i1 [[TMP17]], label [[OMP_OFFLOAD_FAILED6:%.*]], label [[OMP_OFFLOAD_CONT7:%.*]]
12526 // CHECK12: omp_offload.failed6:
12527 // CHECK12-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l92(i64 [[TMP5]]) #[[ATTR2]]
12528 // CHECK12-NEXT: br label [[OMP_OFFLOAD_CONT7]]
12529 // CHECK12: omp_offload.cont7:
12530 // CHECK12-NEXT: [[TMP18:%.*]] = load i32, i32* @Arg, align 4
12531 // CHECK12-NEXT: [[CALL:%.*]] = call i32 @_Z5tmainIiEiT_(i32 [[TMP18]])
12532 // CHECK12-NEXT: ret i32 [[CALL]]
12533 //
12534 //
12535 // CHECK12-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l76
12536 // CHECK12-SAME: () #[[ATTR1]] {
12537 // CHECK12-NEXT: entry:
12538 // CHECK12-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 0, void (i32*, i32*, ...)* bitcast (void (i32*, i32*)* @.omp_outlined..4 to void (i32*, i32*, ...)*))
12539 // CHECK12-NEXT: ret void
12540 //
12541 //
12542 // CHECK12-LABEL: define {{[^@]+}}@.omp_outlined..4
12543 // CHECK12-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]]) #[[ATTR1]] {
12544 // CHECK12-NEXT: entry:
12545 // CHECK12-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
12546 // CHECK12-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
12547 // CHECK12-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
12548 // CHECK12-NEXT: [[TMP:%.*]] = alloca i32, align 4
12549 // CHECK12-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4
12550 // CHECK12-NEXT: [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4
12551 // CHECK12-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
12552 // CHECK12-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
12553 // CHECK12-NEXT: [[I:%.*]] = alloca i32, align 4
12554 // CHECK12-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
12555 // CHECK12-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
12556 // CHECK12-NEXT: store i32 0, i32* [[DOTOMP_COMB_LB]], align 4
12557 // CHECK12-NEXT: store i32 99, i32* [[DOTOMP_COMB_UB]], align 4
12558 // CHECK12-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4
12559 // CHECK12-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4
12560 // CHECK12-NEXT: [[TMP0:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
12561 // CHECK12-NEXT: [[TMP1:%.*]] = load i32, i32* [[TMP0]], align 4
12562 // CHECK12-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB1]], i32 [[TMP1]], i32 92, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_COMB_LB]], i32* [[DOTOMP_COMB_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1)
12563 // CHECK12-NEXT: [[TMP2:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
12564 // CHECK12-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP2]], 99
12565 // CHECK12-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
12566 // CHECK12: cond.true:
12567 // CHECK12-NEXT: br label [[COND_END:%.*]]
12568 // CHECK12: cond.false:
12569 // CHECK12-NEXT: [[TMP3:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
12570 // CHECK12-NEXT: br label [[COND_END]]
12571 // CHECK12: cond.end:
12572 // CHECK12-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP3]], [[COND_FALSE]] ]
12573 // CHECK12-NEXT: store i32 [[COND]], i32* [[DOTOMP_COMB_UB]], align 4
12574 // CHECK12-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4
12575 // CHECK12-NEXT: store i32 [[TMP4]], i32* [[DOTOMP_IV]], align 4
12576 // CHECK12-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
12577 // CHECK12: omp.inner.for.cond:
12578 // CHECK12-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
12579 // CHECK12-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
12580 // CHECK12-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP5]], [[TMP6]]
12581 // CHECK12-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
12582 // CHECK12: omp.inner.for.body:
12583 // CHECK12-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4
12584 // CHECK12-NEXT: [[TMP8:%.*]] = zext i32 [[TMP7]] to i64
12585 // CHECK12-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
12586 // CHECK12-NEXT: [[TMP10:%.*]] = zext i32 [[TMP9]] to i64
12587 // CHECK12-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB3]], i32 2, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64, i64)* @.omp_outlined..5 to void (i32*, i32*, ...)*), i64 [[TMP8]], i64 [[TMP10]])
12588 // CHECK12-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
12589 // CHECK12: omp.inner.for.inc:
12590 // CHECK12-NEXT: [[TMP11:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
12591 // CHECK12-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4
12592 // CHECK12-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP11]], [[TMP12]]
12593 // CHECK12-NEXT: store i32 [[ADD]], i32* [[DOTOMP_IV]], align 4
12594 // CHECK12-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP21:![0-9]+]]
12595 // CHECK12: omp.inner.for.end:
12596 // CHECK12-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
12597 // CHECK12: omp.loop.exit:
12598 // CHECK12-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP1]])
12599 // CHECK12-NEXT: [[TMP13:%.*]] = load i32, i32* [[DOTOMP_IS_LAST]], align 4
12600 // CHECK12-NEXT: [[TMP14:%.*]] = icmp ne i32 [[TMP13]], 0
12601 // CHECK12-NEXT: br i1 [[TMP14]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
12602 // CHECK12: .omp.final.then:
12603 // CHECK12-NEXT: store i32 100, i32* [[I]], align 4
12604 // CHECK12-NEXT: br label [[DOTOMP_FINAL_DONE]]
12605 // CHECK12: .omp.final.done:
12606 // CHECK12-NEXT: ret void
12607 //
12608 //
12609 // CHECK12-LABEL: define {{[^@]+}}@.omp_outlined..5
12610 // CHECK12-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[DOTPREVIOUS_LB_:%.*]], i64 [[DOTPREVIOUS_UB_:%.*]]) #[[ATTR1]] {
12611 // CHECK12-NEXT: entry:
12612 // CHECK12-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
12613 // CHECK12-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
12614 // CHECK12-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i64, align 8
12615 // CHECK12-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i64, align 8
12616 // CHECK12-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
12617 // CHECK12-NEXT: [[TMP:%.*]] = alloca i32, align 4
12618 // CHECK12-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4
12619 // CHECK12-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4
12620 // CHECK12-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
12621 // CHECK12-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
12622 // CHECK12-NEXT: [[I:%.*]] = alloca i32, align 4
12623 // CHECK12-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
12624 // CHECK12-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
12625 // CHECK12-NEXT: store i64 [[DOTPREVIOUS_LB_]], i64* [[DOTPREVIOUS_LB__ADDR]], align 8
12626 // CHECK12-NEXT: store i64 [[DOTPREVIOUS_UB_]], i64* [[DOTPREVIOUS_UB__ADDR]], align 8
12627 // CHECK12-NEXT: store i32 0, i32* [[DOTOMP_LB]], align 4
12628 // CHECK12-NEXT: store i32 99, i32* [[DOTOMP_UB]], align 4
12629 // CHECK12-NEXT: [[TMP0:%.*]] = load i64, i64* [[DOTPREVIOUS_LB__ADDR]], align 8
12630 // CHECK12-NEXT: [[CONV:%.*]] = trunc i64 [[TMP0]] to i32
12631 // CHECK12-NEXT: [[TMP1:%.*]] = load i64, i64* [[DOTPREVIOUS_UB__ADDR]], align 8
12632 // CHECK12-NEXT: [[CONV1:%.*]] = trunc i64 [[TMP1]] to i32
12633 // CHECK12-NEXT: store i32 [[CONV]], i32* [[DOTOMP_LB]], align 4
12634 // CHECK12-NEXT: store i32 [[CONV1]], i32* [[DOTOMP_UB]], align 4
12635 // CHECK12-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4
12636 // CHECK12-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4
12637 // CHECK12-NEXT: [[TMP2:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
12638 // CHECK12-NEXT: [[TMP3:%.*]] = load i32, i32* [[TMP2]], align 4
12639 // CHECK12-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB2]], i32 [[TMP3]], i32 34, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_LB]], i32* [[DOTOMP_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1)
12640 // CHECK12-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
12641 // CHECK12-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP4]], 99
12642 // CHECK12-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
12643 // CHECK12: cond.true:
12644 // CHECK12-NEXT: br label [[COND_END:%.*]]
12645 // CHECK12: cond.false:
12646 // CHECK12-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
12647 // CHECK12-NEXT: br label [[COND_END]]
12648 // CHECK12: cond.end:
12649 // CHECK12-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP5]], [[COND_FALSE]] ]
12650 // CHECK12-NEXT: store i32 [[COND]], i32* [[DOTOMP_UB]], align 4
12651 // CHECK12-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4
12652 // CHECK12-NEXT: store i32 [[TMP6]], i32* [[DOTOMP_IV]], align 4
12653 // CHECK12-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
12654 // CHECK12: omp.inner.for.cond:
12655 // CHECK12-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
12656 // CHECK12-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
12657 // CHECK12-NEXT: [[CMP2:%.*]] = icmp sle i32 [[TMP7]], [[TMP8]]
12658 // CHECK12-NEXT: br i1 [[CMP2]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
12659 // CHECK12: omp.inner.for.body:
12660 // CHECK12-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
12661 // CHECK12-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP9]], 1
12662 // CHECK12-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
12663 // CHECK12-NEXT: store i32 [[ADD]], i32* [[I]], align 4
12664 // CHECK12-NEXT: call void @_Z3fn4v()
12665 // CHECK12-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
12666 // CHECK12: omp.body.continue:
12667 // CHECK12-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
12668 // CHECK12: omp.inner.for.inc:
12669 // CHECK12-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
12670 // CHECK12-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP10]], 1
12671 // CHECK12-NEXT: store i32 [[ADD3]], i32* [[DOTOMP_IV]], align 4
12672 // CHECK12-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP22:![0-9]+]]
12673 // CHECK12: omp.inner.for.end:
12674 // CHECK12-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
12675 // CHECK12: omp.loop.exit:
12676 // CHECK12-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP3]])
12677 // CHECK12-NEXT: [[TMP11:%.*]] = load i32, i32* [[DOTOMP_IS_LAST]], align 4
12678 // CHECK12-NEXT: [[TMP12:%.*]] = icmp ne i32 [[TMP11]], 0
12679 // CHECK12-NEXT: br i1 [[TMP12]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
12680 // CHECK12: .omp.final.then:
12681 // CHECK12-NEXT: store i32 100, i32* [[I]], align 4
12682 // CHECK12-NEXT: br label [[DOTOMP_FINAL_DONE]]
12683 // CHECK12: .omp.final.done:
12684 // CHECK12-NEXT: ret void
12685 //
12686 //
12687 // CHECK12-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l84
12688 // CHECK12-SAME: () #[[ATTR1]] {
12689 // CHECK12-NEXT: entry:
12690 // CHECK12-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 0, void (i32*, i32*, ...)* bitcast (void (i32*, i32*)* @.omp_outlined..6 to void (i32*, i32*, ...)*))
12691 // CHECK12-NEXT: ret void
12692 //
12693 //
12694 // CHECK12-LABEL: define {{[^@]+}}@.omp_outlined..6
12695 // CHECK12-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]]) #[[ATTR1]] {
12696 // CHECK12-NEXT: entry:
12697 // CHECK12-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
12698 // CHECK12-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
12699 // CHECK12-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
12700 // CHECK12-NEXT: [[TMP:%.*]] = alloca i32, align 4
12701 // CHECK12-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4
12702 // CHECK12-NEXT: [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4
12703 // CHECK12-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
12704 // CHECK12-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
12705 // CHECK12-NEXT: [[I:%.*]] = alloca i32, align 4
12706 // CHECK12-NEXT: [[DOTBOUND_ZERO_ADDR:%.*]] = alloca i32, align 4
12707 // CHECK12-NEXT: store i32 0, i32* [[DOTBOUND_ZERO_ADDR]], align 4
12708 // CHECK12-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
12709 // CHECK12-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
12710 // CHECK12-NEXT: store i32 0, i32* [[DOTOMP_COMB_LB]], align 4
12711 // CHECK12-NEXT: store i32 99, i32* [[DOTOMP_COMB_UB]], align 4
12712 // CHECK12-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4
12713 // CHECK12-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4
12714 // CHECK12-NEXT: [[TMP0:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
12715 // CHECK12-NEXT: [[TMP1:%.*]] = load i32, i32* [[TMP0]], align 4
12716 // CHECK12-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB1]], i32 [[TMP1]], i32 92, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_COMB_LB]], i32* [[DOTOMP_COMB_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1)
12717 // CHECK12-NEXT: [[TMP2:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
12718 // CHECK12-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP2]], 99
12719 // CHECK12-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
12720 // CHECK12: cond.true:
12721 // CHECK12-NEXT: br label [[COND_END:%.*]]
12722 // CHECK12: cond.false:
12723 // CHECK12-NEXT: [[TMP3:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
12724 // CHECK12-NEXT: br label [[COND_END]]
12725 // CHECK12: cond.end:
12726 // CHECK12-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP3]], [[COND_FALSE]] ]
12727 // CHECK12-NEXT: store i32 [[COND]], i32* [[DOTOMP_COMB_UB]], align 4
12728 // CHECK12-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4
12729 // CHECK12-NEXT: store i32 [[TMP4]], i32* [[DOTOMP_IV]], align 4
12730 // CHECK12-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
12731 // CHECK12: omp.inner.for.cond:
12732 // CHECK12-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
12733 // CHECK12-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
12734 // CHECK12-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP5]], [[TMP6]]
12735 // CHECK12-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
12736 // CHECK12: omp.inner.for.body:
12737 // CHECK12-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4
12738 // CHECK12-NEXT: [[TMP8:%.*]] = zext i32 [[TMP7]] to i64
12739 // CHECK12-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
12740 // CHECK12-NEXT: [[TMP10:%.*]] = zext i32 [[TMP9]] to i64
12741 // CHECK12-NEXT: call void @__kmpc_serialized_parallel(%struct.ident_t* @[[GLOB3]], i32 [[TMP1]])
12742 // CHECK12-NEXT: [[TMP11:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
12743 // CHECK12-NEXT: call void @.omp_outlined..7(i32* [[TMP11]], i32* [[DOTBOUND_ZERO_ADDR]], i64 [[TMP8]], i64 [[TMP10]]) #[[ATTR2]]
12744 // CHECK12-NEXT: call void @__kmpc_end_serialized_parallel(%struct.ident_t* @[[GLOB3]], i32 [[TMP1]])
12745 // CHECK12-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
12746 // CHECK12: omp.inner.for.inc:
12747 // CHECK12-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
12748 // CHECK12-NEXT: [[TMP13:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4
12749 // CHECK12-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP12]], [[TMP13]]
12750 // CHECK12-NEXT: store i32 [[ADD]], i32* [[DOTOMP_IV]], align 4
12751 // CHECK12-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP23:![0-9]+]]
12752 // CHECK12: omp.inner.for.end:
12753 // CHECK12-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
12754 // CHECK12: omp.loop.exit:
12755 // CHECK12-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP1]])
12756 // CHECK12-NEXT: [[TMP14:%.*]] = load i32, i32* [[DOTOMP_IS_LAST]], align 4
12757 // CHECK12-NEXT: [[TMP15:%.*]] = icmp ne i32 [[TMP14]], 0
12758 // CHECK12-NEXT: br i1 [[TMP15]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
12759 // CHECK12: .omp.final.then:
12760 // CHECK12-NEXT: store i32 100, i32* [[I]], align 4
12761 // CHECK12-NEXT: br label [[DOTOMP_FINAL_DONE]]
12762 // CHECK12: .omp.final.done:
12763 // CHECK12-NEXT: ret void
12764 //
12765 //
12766 // CHECK12-LABEL: define {{[^@]+}}@.omp_outlined..7
12767 // CHECK12-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[DOTPREVIOUS_LB_:%.*]], i64 [[DOTPREVIOUS_UB_:%.*]]) #[[ATTR1]] {
12768 // CHECK12-NEXT: entry:
12769 // CHECK12-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
12770 // CHECK12-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
12771 // CHECK12-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i64, align 8
12772 // CHECK12-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i64, align 8
12773 // CHECK12-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
12774 // CHECK12-NEXT: [[TMP:%.*]] = alloca i32, align 4
12775 // CHECK12-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4
12776 // CHECK12-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4
12777 // CHECK12-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
12778 // CHECK12-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
12779 // CHECK12-NEXT: [[I:%.*]] = alloca i32, align 4
12780 // CHECK12-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
12781 // CHECK12-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
12782 // CHECK12-NEXT: store i64 [[DOTPREVIOUS_LB_]], i64* [[DOTPREVIOUS_LB__ADDR]], align 8
12783 // CHECK12-NEXT: store i64 [[DOTPREVIOUS_UB_]], i64* [[DOTPREVIOUS_UB__ADDR]], align 8
12784 // CHECK12-NEXT: store i32 0, i32* [[DOTOMP_LB]], align 4
12785 // CHECK12-NEXT: store i32 99, i32* [[DOTOMP_UB]], align 4
12786 // CHECK12-NEXT: [[TMP0:%.*]] = load i64, i64* [[DOTPREVIOUS_LB__ADDR]], align 8
12787 // CHECK12-NEXT: [[CONV:%.*]] = trunc i64 [[TMP0]] to i32
12788 // CHECK12-NEXT: [[TMP1:%.*]] = load i64, i64* [[DOTPREVIOUS_UB__ADDR]], align 8
12789 // CHECK12-NEXT: [[CONV1:%.*]] = trunc i64 [[TMP1]] to i32
12790 // CHECK12-NEXT: store i32 [[CONV]], i32* [[DOTOMP_LB]], align 4
12791 // CHECK12-NEXT: store i32 [[CONV1]], i32* [[DOTOMP_UB]], align 4
12792 // CHECK12-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4
12793 // CHECK12-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4
12794 // CHECK12-NEXT: [[TMP2:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
12795 // CHECK12-NEXT: [[TMP3:%.*]] = load i32, i32* [[TMP2]], align 4
12796 // CHECK12-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB2]], i32 [[TMP3]], i32 34, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_LB]], i32* [[DOTOMP_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1)
12797 // CHECK12-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
12798 // CHECK12-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP4]], 99
12799 // CHECK12-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
12800 // CHECK12: cond.true:
12801 // CHECK12-NEXT: br label [[COND_END:%.*]]
12802 // CHECK12: cond.false:
12803 // CHECK12-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
12804 // CHECK12-NEXT: br label [[COND_END]]
12805 // CHECK12: cond.end:
12806 // CHECK12-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP5]], [[COND_FALSE]] ]
12807 // CHECK12-NEXT: store i32 [[COND]], i32* [[DOTOMP_UB]], align 4
12808 // CHECK12-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4
12809 // CHECK12-NEXT: store i32 [[TMP6]], i32* [[DOTOMP_IV]], align 4
12810 // CHECK12-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
12811 // CHECK12: omp.inner.for.cond:
12812 // CHECK12-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
12813 // CHECK12-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
12814 // CHECK12-NEXT: [[CMP2:%.*]] = icmp sle i32 [[TMP7]], [[TMP8]]
12815 // CHECK12-NEXT: br i1 [[CMP2]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
12816 // CHECK12: omp.inner.for.body:
12817 // CHECK12-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
12818 // CHECK12-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP9]], 1
12819 // CHECK12-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
12820 // CHECK12-NEXT: store i32 [[ADD]], i32* [[I]], align 4
12821 // CHECK12-NEXT: call void @_Z3fn5v()
12822 // CHECK12-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
12823 // CHECK12: omp.body.continue:
12824 // CHECK12-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
12825 // CHECK12: omp.inner.for.inc:
12826 // CHECK12-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
12827 // CHECK12-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP10]], 1
12828 // CHECK12-NEXT: store i32 [[ADD3]], i32* [[DOTOMP_IV]], align 4
12829 // CHECK12-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP25:![0-9]+]]
12830 // CHECK12: omp.inner.for.end:
12831 // CHECK12-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
12832 // CHECK12: omp.loop.exit:
12833 // CHECK12-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP3]])
12834 // CHECK12-NEXT: [[TMP11:%.*]] = load i32, i32* [[DOTOMP_IS_LAST]], align 4
12835 // CHECK12-NEXT: [[TMP12:%.*]] = icmp ne i32 [[TMP11]], 0
12836 // CHECK12-NEXT: br i1 [[TMP12]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
12837 // CHECK12: .omp.final.then:
12838 // CHECK12-NEXT: store i32 100, i32* [[I]], align 4
12839 // CHECK12-NEXT: br label [[DOTOMP_FINAL_DONE]]
12840 // CHECK12: .omp.final.done:
12841 // CHECK12-NEXT: ret void
12842 //
12843 //
12844 // CHECK12-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l92
12845 // CHECK12-SAME: (i64 [[ARG:%.*]]) #[[ATTR1]] {
12846 // CHECK12-NEXT: entry:
12847 // CHECK12-NEXT: [[ARG_ADDR:%.*]] = alloca i64, align 8
12848 // CHECK12-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i8, align 1
12849 // CHECK12-NEXT: [[DOTCAPTURE_EXPR__CASTED:%.*]] = alloca i64, align 8
12850 // CHECK12-NEXT: store i64 [[ARG]], i64* [[ARG_ADDR]], align 8
12851 // CHECK12-NEXT: [[CONV:%.*]] = bitcast i64* [[ARG_ADDR]] to i32*
12852 // CHECK12-NEXT: [[TMP0:%.*]] = load i32, i32* [[CONV]], align 8
12853 // CHECK12-NEXT: [[TOBOOL:%.*]] = icmp ne i32 [[TMP0]], 0
12854 // CHECK12-NEXT: [[FROMBOOL:%.*]] = zext i1 [[TOBOOL]] to i8
12855 // CHECK12-NEXT: store i8 [[FROMBOOL]], i8* [[DOTCAPTURE_EXPR_]], align 1
12856 // CHECK12-NEXT: [[TMP1:%.*]] = load i8, i8* [[DOTCAPTURE_EXPR_]], align 1
12857 // CHECK12-NEXT: [[TOBOOL1:%.*]] = trunc i8 [[TMP1]] to i1
12858 // CHECK12-NEXT: [[CONV2:%.*]] = bitcast i64* [[DOTCAPTURE_EXPR__CASTED]] to i8*
12859 // CHECK12-NEXT: [[FROMBOOL3:%.*]] = zext i1 [[TOBOOL1]] to i8
12860 // CHECK12-NEXT: store i8 [[FROMBOOL3]], i8* [[CONV2]], align 1
12861 // CHECK12-NEXT: [[TMP2:%.*]] = load i64, i64* [[DOTCAPTURE_EXPR__CASTED]], align 8
12862 // CHECK12-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64)* @.omp_outlined..8 to void (i32*, i32*, ...)*), i64 [[TMP2]])
12863 // CHECK12-NEXT: ret void
12864 //
12865 //
12866 // CHECK12-LABEL: define {{[^@]+}}@.omp_outlined..8
12867 // CHECK12-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[DOTCAPTURE_EXPR_:%.*]]) #[[ATTR1]] {
12868 // CHECK12-NEXT: entry:
12869 // CHECK12-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
12870 // CHECK12-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
12871 // CHECK12-NEXT: [[DOTCAPTURE_EXPR__ADDR:%.*]] = alloca i64, align 8
12872 // CHECK12-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
12873 // CHECK12-NEXT: [[TMP:%.*]] = alloca i32, align 4
12874 // CHECK12-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4
12875 // CHECK12-NEXT: [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4
12876 // CHECK12-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
12877 // CHECK12-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
12878 // CHECK12-NEXT: [[I:%.*]] = alloca i32, align 4
12879 // CHECK12-NEXT: [[DOTCAPTURE_EXPR__CASTED:%.*]] = alloca i64, align 8
12880 // CHECK12-NEXT: [[DOTBOUND_ZERO_ADDR:%.*]] = alloca i32, align 4
12881 // CHECK12-NEXT: [[DOTCAPTURE_EXPR__CASTED11:%.*]] = alloca i64, align 8
12882 // CHECK12-NEXT: [[DOTBOUND_ZERO_ADDR17:%.*]] = alloca i32, align 4
12883 // CHECK12-NEXT: store i32 0, i32* [[DOTBOUND_ZERO_ADDR17]], align 4
12884 // CHECK12-NEXT: store i32 0, i32* [[DOTBOUND_ZERO_ADDR]], align 4
12885 // CHECK12-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
12886 // CHECK12-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
12887 // CHECK12-NEXT: store i64 [[DOTCAPTURE_EXPR_]], i64* [[DOTCAPTURE_EXPR__ADDR]], align 8
12888 // CHECK12-NEXT: [[CONV:%.*]] = bitcast i64* [[DOTCAPTURE_EXPR__ADDR]] to i8*
12889 // CHECK12-NEXT: store i32 0, i32* [[DOTOMP_COMB_LB]], align 4
12890 // CHECK12-NEXT: store i32 99, i32* [[DOTOMP_COMB_UB]], align 4
12891 // CHECK12-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4
12892 // CHECK12-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4
12893 // CHECK12-NEXT: [[TMP0:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
12894 // CHECK12-NEXT: [[TMP1:%.*]] = load i32, i32* [[TMP0]], align 4
12895 // CHECK12-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB1]], i32 [[TMP1]], i32 92, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_COMB_LB]], i32* [[DOTOMP_COMB_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1)
12896 // CHECK12-NEXT: [[TMP2:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
12897 // CHECK12-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP2]], 99
12898 // CHECK12-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
12899 // CHECK12: cond.true:
12900 // CHECK12-NEXT: br label [[COND_END:%.*]]
12901 // CHECK12: cond.false:
12902 // CHECK12-NEXT: [[TMP3:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
12903 // CHECK12-NEXT: br label [[COND_END]]
12904 // CHECK12: cond.end:
12905 // CHECK12-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP3]], [[COND_FALSE]] ]
12906 // CHECK12-NEXT: store i32 [[COND]], i32* [[DOTOMP_COMB_UB]], align 4
12907 // CHECK12-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4
12908 // CHECK12-NEXT: store i32 [[TMP4]], i32* [[DOTOMP_IV]], align 4
12909 // CHECK12-NEXT: [[TMP5:%.*]] = load i8, i8* [[CONV]], align 8
12910 // CHECK12-NEXT: [[TOBOOL:%.*]] = trunc i8 [[TMP5]] to i1
12911 // CHECK12-NEXT: br i1 [[TOBOOL]], label [[OMP_IF_THEN:%.*]], label [[OMP_IF_ELSE6:%.*]]
12912 // CHECK12: omp_if.then:
12913 // CHECK12-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
12914 // CHECK12: omp.inner.for.cond:
12915 // CHECK12-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
12916 // CHECK12-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
12917 // CHECK12-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP6]], [[TMP7]]
12918 // CHECK12-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
12919 // CHECK12: omp.inner.for.body:
12920 // CHECK12-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4
12921 // CHECK12-NEXT: [[TMP9:%.*]] = zext i32 [[TMP8]] to i64
12922 // CHECK12-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
12923 // CHECK12-NEXT: [[TMP11:%.*]] = zext i32 [[TMP10]] to i64
12924 // CHECK12-NEXT: [[TMP12:%.*]] = load i8, i8* [[CONV]], align 8
12925 // CHECK12-NEXT: [[TOBOOL2:%.*]] = trunc i8 [[TMP12]] to i1
12926 // CHECK12-NEXT: [[CONV3:%.*]] = bitcast i64* [[DOTCAPTURE_EXPR__CASTED]] to i8*
12927 // CHECK12-NEXT: [[FROMBOOL:%.*]] = zext i1 [[TOBOOL2]] to i8
12928 // CHECK12-NEXT: store i8 [[FROMBOOL]], i8* [[CONV3]], align 1
12929 // CHECK12-NEXT: [[TMP13:%.*]] = load i64, i64* [[DOTCAPTURE_EXPR__CASTED]], align 8
12930 // CHECK12-NEXT: [[TMP14:%.*]] = load i8, i8* [[CONV]], align 8
12931 // CHECK12-NEXT: [[TOBOOL4:%.*]] = trunc i8 [[TMP14]] to i1
12932 // CHECK12-NEXT: br i1 [[TOBOOL4]], label [[OMP_IF_THEN5:%.*]], label [[OMP_IF_ELSE:%.*]]
12933 // CHECK12: omp_if.then5:
12934 // CHECK12-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB3]], i32 3, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64, i64, i64)* @.omp_outlined..9 to void (i32*, i32*, ...)*), i64 [[TMP9]], i64 [[TMP11]], i64 [[TMP13]])
12935 // CHECK12-NEXT: br label [[OMP_IF_END:%.*]]
12936 // CHECK12: omp_if.else:
12937 // CHECK12-NEXT: call void @__kmpc_serialized_parallel(%struct.ident_t* @[[GLOB3]], i32 [[TMP1]])
12938 // CHECK12-NEXT: [[TMP15:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
12939 // CHECK12-NEXT: call void @.omp_outlined..9(i32* [[TMP15]], i32* [[DOTBOUND_ZERO_ADDR]], i64 [[TMP9]], i64 [[TMP11]], i64 [[TMP13]]) #[[ATTR2]]
12940 // CHECK12-NEXT: call void @__kmpc_end_serialized_parallel(%struct.ident_t* @[[GLOB3]], i32 [[TMP1]])
12941 // CHECK12-NEXT: br label [[OMP_IF_END]]
12942 // CHECK12: omp_if.end:
12943 // CHECK12-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
12944 // CHECK12: omp.inner.for.inc:
12945 // CHECK12-NEXT: [[TMP16:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
12946 // CHECK12-NEXT: [[TMP17:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4
12947 // CHECK12-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP16]], [[TMP17]]
12948 // CHECK12-NEXT: store i32 [[ADD]], i32* [[DOTOMP_IV]], align 4
12949 // CHECK12-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP26:![0-9]+]]
12950 // CHECK12: omp.inner.for.end:
12951 // CHECK12-NEXT: br label [[OMP_IF_END22:%.*]]
12952 // CHECK12: omp_if.else6:
12953 // CHECK12-NEXT: br label [[OMP_INNER_FOR_COND7:%.*]]
12954 // CHECK12: omp.inner.for.cond7:
12955 // CHECK12-NEXT: [[TMP18:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
12956 // CHECK12-NEXT: [[TMP19:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
12957 // CHECK12-NEXT: [[CMP8:%.*]] = icmp sle i32 [[TMP18]], [[TMP19]]
12958 // CHECK12-NEXT: br i1 [[CMP8]], label [[OMP_INNER_FOR_BODY9:%.*]], label [[OMP_INNER_FOR_END21:%.*]]
12959 // CHECK12: omp.inner.for.body9:
12960 // CHECK12-NEXT: [[TMP20:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4
12961 // CHECK12-NEXT: [[TMP21:%.*]] = zext i32 [[TMP20]] to i64
12962 // CHECK12-NEXT: [[TMP22:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
12963 // CHECK12-NEXT: [[TMP23:%.*]] = zext i32 [[TMP22]] to i64
12964 // CHECK12-NEXT: [[TMP24:%.*]] = load i8, i8* [[CONV]], align 8
12965 // CHECK12-NEXT: [[TOBOOL10:%.*]] = trunc i8 [[TMP24]] to i1
12966 // CHECK12-NEXT: [[CONV12:%.*]] = bitcast i64* [[DOTCAPTURE_EXPR__CASTED11]] to i8*
12967 // CHECK12-NEXT: [[FROMBOOL13:%.*]] = zext i1 [[TOBOOL10]] to i8
12968 // CHECK12-NEXT: store i8 [[FROMBOOL13]], i8* [[CONV12]], align 1
12969 // CHECK12-NEXT: [[TMP25:%.*]] = load i64, i64* [[DOTCAPTURE_EXPR__CASTED11]], align 8
12970 // CHECK12-NEXT: [[TMP26:%.*]] = load i8, i8* [[CONV]], align 8
12971 // CHECK12-NEXT: [[TOBOOL14:%.*]] = trunc i8 [[TMP26]] to i1
12972 // CHECK12-NEXT: br i1 [[TOBOOL14]], label [[OMP_IF_THEN15:%.*]], label [[OMP_IF_ELSE16:%.*]]
12973 // CHECK12: omp_if.then15:
12974 // CHECK12-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB3]], i32 3, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64, i64, i64)* @.omp_outlined..10 to void (i32*, i32*, ...)*), i64 [[TMP21]], i64 [[TMP23]], i64 [[TMP25]])
12975 // CHECK12-NEXT: br label [[OMP_IF_END18:%.*]]
12976 // CHECK12: omp_if.else16:
12977 // CHECK12-NEXT: call void @__kmpc_serialized_parallel(%struct.ident_t* @[[GLOB3]], i32 [[TMP1]])
12978 // CHECK12-NEXT: [[TMP27:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
12979 // CHECK12-NEXT: call void @.omp_outlined..10(i32* [[TMP27]], i32* [[DOTBOUND_ZERO_ADDR17]], i64 [[TMP21]], i64 [[TMP23]], i64 [[TMP25]]) #[[ATTR2]]
12980 // CHECK12-NEXT: call void @__kmpc_end_serialized_parallel(%struct.ident_t* @[[GLOB3]], i32 [[TMP1]])
12981 // CHECK12-NEXT: br label [[OMP_IF_END18]]
12982 // CHECK12: omp_if.end18:
12983 // CHECK12-NEXT: br label [[OMP_INNER_FOR_INC19:%.*]]
12984 // CHECK12: omp.inner.for.inc19:
12985 // CHECK12-NEXT: [[TMP28:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
12986 // CHECK12-NEXT: [[TMP29:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4
12987 // CHECK12-NEXT: [[ADD20:%.*]] = add nsw i32 [[TMP28]], [[TMP29]]
12988 // CHECK12-NEXT: store i32 [[ADD20]], i32* [[DOTOMP_IV]], align 4
12989 // CHECK12-NEXT: br label [[OMP_INNER_FOR_COND7]], !llvm.loop [[LOOP27:![0-9]+]]
12990 // CHECK12: omp.inner.for.end21:
12991 // CHECK12-NEXT: br label [[OMP_IF_END22]]
12992 // CHECK12: omp_if.end22:
12993 // CHECK12-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
12994 // CHECK12: omp.loop.exit:
12995 // CHECK12-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP1]])
12996 // CHECK12-NEXT: [[TMP30:%.*]] = load i32, i32* [[DOTOMP_IS_LAST]], align 4
12997 // CHECK12-NEXT: [[TMP31:%.*]] = icmp ne i32 [[TMP30]], 0
12998 // CHECK12-NEXT: br i1 [[TMP31]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
12999 // CHECK12: .omp.final.then:
13000 // CHECK12-NEXT: store i32 100, i32* [[I]], align 4
13001 // CHECK12-NEXT: br label [[DOTOMP_FINAL_DONE]]
13002 // CHECK12: .omp.final.done:
13003 // CHECK12-NEXT: ret void
13004 //
13005 //
13006 // CHECK12-LABEL: define {{[^@]+}}@.omp_outlined..9
13007 // CHECK12-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[DOTPREVIOUS_LB_:%.*]], i64 [[DOTPREVIOUS_UB_:%.*]], i64 [[DOTCAPTURE_EXPR_:%.*]]) #[[ATTR1]] {
13008 // CHECK12-NEXT: entry:
13009 // CHECK12-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
13010 // CHECK12-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
13011 // CHECK12-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i64, align 8
13012 // CHECK12-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i64, align 8
13013 // CHECK12-NEXT: [[DOTCAPTURE_EXPR__ADDR:%.*]] = alloca i64, align 8
13014 // CHECK12-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
13015 // CHECK12-NEXT: [[TMP:%.*]] = alloca i32, align 4
13016 // CHECK12-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4
13017 // CHECK12-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4
13018 // CHECK12-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
13019 // CHECK12-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
13020 // CHECK12-NEXT: [[I:%.*]] = alloca i32, align 4
13021 // CHECK12-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
13022 // CHECK12-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
13023 // CHECK12-NEXT: store i64 [[DOTPREVIOUS_LB_]], i64* [[DOTPREVIOUS_LB__ADDR]], align 8
13024 // CHECK12-NEXT: store i64 [[DOTPREVIOUS_UB_]], i64* [[DOTPREVIOUS_UB__ADDR]], align 8
13025 // CHECK12-NEXT: store i64 [[DOTCAPTURE_EXPR_]], i64* [[DOTCAPTURE_EXPR__ADDR]], align 8
13026 // CHECK12-NEXT: [[CONV:%.*]] = bitcast i64* [[DOTCAPTURE_EXPR__ADDR]] to i8*
13027 // CHECK12-NEXT: store i32 0, i32* [[DOTOMP_LB]], align 4
13028 // CHECK12-NEXT: store i32 99, i32* [[DOTOMP_UB]], align 4
13029 // CHECK12-NEXT: [[TMP0:%.*]] = load i64, i64* [[DOTPREVIOUS_LB__ADDR]], align 8
13030 // CHECK12-NEXT: [[CONV1:%.*]] = trunc i64 [[TMP0]] to i32
13031 // CHECK12-NEXT: [[TMP1:%.*]] = load i64, i64* [[DOTPREVIOUS_UB__ADDR]], align 8
13032 // CHECK12-NEXT: [[CONV2:%.*]] = trunc i64 [[TMP1]] to i32
13033 // CHECK12-NEXT: store i32 [[CONV1]], i32* [[DOTOMP_LB]], align 4
13034 // CHECK12-NEXT: store i32 [[CONV2]], i32* [[DOTOMP_UB]], align 4
13035 // CHECK12-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4
13036 // CHECK12-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4
13037 // CHECK12-NEXT: [[TMP2:%.*]] = load i8, i8* [[CONV]], align 8
13038 // CHECK12-NEXT: [[TOBOOL:%.*]] = trunc i8 [[TMP2]] to i1
13039 // CHECK12-NEXT: br i1 [[TOBOOL]], label [[OMP_IF_THEN:%.*]], label [[OMP_IF_ELSE:%.*]]
13040 // CHECK12: omp_if.then:
13041 // CHECK12-NEXT: [[TMP3:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
13042 // CHECK12-NEXT: [[TMP4:%.*]] = load i32, i32* [[TMP3]], align 4
13043 // CHECK12-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB2]], i32 [[TMP4]], i32 34, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_LB]], i32* [[DOTOMP_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1)
13044 // CHECK12-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
13045 // CHECK12-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP5]], 99
13046 // CHECK12-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
13047 // CHECK12: cond.true:
13048 // CHECK12-NEXT: br label [[COND_END:%.*]]
13049 // CHECK12: cond.false:
13050 // CHECK12-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
13051 // CHECK12-NEXT: br label [[COND_END]]
13052 // CHECK12: cond.end:
13053 // CHECK12-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP6]], [[COND_FALSE]] ]
13054 // CHECK12-NEXT: store i32 [[COND]], i32* [[DOTOMP_UB]], align 4
13055 // CHECK12-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4
13056 // CHECK12-NEXT: store i32 [[TMP7]], i32* [[DOTOMP_IV]], align 4
13057 // CHECK12-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
13058 // CHECK12: omp.inner.for.cond:
13059 // CHECK12-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
13060 // CHECK12-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
13061 // CHECK12-NEXT: [[CMP3:%.*]] = icmp sle i32 [[TMP8]], [[TMP9]]
13062 // CHECK12-NEXT: br i1 [[CMP3]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
13063 // CHECK12: omp.inner.for.body:
13064 // CHECK12-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
13065 // CHECK12-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP10]], 1
13066 // CHECK12-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
13067 // CHECK12-NEXT: store i32 [[ADD]], i32* [[I]], align 4
13068 // CHECK12-NEXT: call void @_Z3fn6v()
13069 // CHECK12-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
13070 // CHECK12: omp.body.continue:
13071 // CHECK12-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
13072 // CHECK12: omp.inner.for.inc:
13073 // CHECK12-NEXT: [[TMP11:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
13074 // CHECK12-NEXT: [[ADD4:%.*]] = add nsw i32 [[TMP11]], 1
13075 // CHECK12-NEXT: store i32 [[ADD4]], i32* [[DOTOMP_IV]], align 4
13076 // CHECK12-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP28:![0-9]+]]
13077 // CHECK12: omp.inner.for.end:
13078 // CHECK12-NEXT: br label [[OMP_IF_END:%.*]]
13079 // CHECK12: omp_if.else:
13080 // CHECK12-NEXT: [[TMP12:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
13081 // CHECK12-NEXT: [[TMP13:%.*]] = load i32, i32* [[TMP12]], align 4
13082 // CHECK12-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB2]], i32 [[TMP13]], i32 34, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_LB]], i32* [[DOTOMP_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1)
13083 // CHECK12-NEXT: [[TMP14:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
13084 // CHECK12-NEXT: [[CMP5:%.*]] = icmp sgt i32 [[TMP14]], 99
13085 // CHECK12-NEXT: br i1 [[CMP5]], label [[COND_TRUE6:%.*]], label [[COND_FALSE7:%.*]]
13086 // CHECK12: cond.true6:
13087 // CHECK12-NEXT: br label [[COND_END8:%.*]]
13088 // CHECK12: cond.false7:
13089 // CHECK12-NEXT: [[TMP15:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
13090 // CHECK12-NEXT: br label [[COND_END8]]
13091 // CHECK12: cond.end8:
13092 // CHECK12-NEXT: [[COND9:%.*]] = phi i32 [ 99, [[COND_TRUE6]] ], [ [[TMP15]], [[COND_FALSE7]] ]
13093 // CHECK12-NEXT: store i32 [[COND9]], i32* [[DOTOMP_UB]], align 4
13094 // CHECK12-NEXT: [[TMP16:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4
13095 // CHECK12-NEXT: store i32 [[TMP16]], i32* [[DOTOMP_IV]], align 4
13096 // CHECK12-NEXT: br label [[OMP_INNER_FOR_COND10:%.*]]
13097 // CHECK12: omp.inner.for.cond10:
13098 // CHECK12-NEXT: [[TMP17:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
13099 // CHECK12-NEXT: [[TMP18:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
13100 // CHECK12-NEXT: [[CMP11:%.*]] = icmp sle i32 [[TMP17]], [[TMP18]]
13101 // CHECK12-NEXT: br i1 [[CMP11]], label [[OMP_INNER_FOR_BODY12:%.*]], label [[OMP_INNER_FOR_END18:%.*]]
13102 // CHECK12: omp.inner.for.body12:
13103 // CHECK12-NEXT: [[TMP19:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
13104 // CHECK12-NEXT: [[MUL13:%.*]] = mul nsw i32 [[TMP19]], 1
13105 // CHECK12-NEXT: [[ADD14:%.*]] = add nsw i32 0, [[MUL13]]
13106 // CHECK12-NEXT: store i32 [[ADD14]], i32* [[I]], align 4
13107 // CHECK12-NEXT: call void @_Z3fn6v()
13108 // CHECK12-NEXT: br label [[OMP_BODY_CONTINUE15:%.*]]
13109 // CHECK12: omp.body.continue15:
13110 // CHECK12-NEXT: br label [[OMP_INNER_FOR_INC16:%.*]]
13111 // CHECK12: omp.inner.for.inc16:
13112 // CHECK12-NEXT: [[TMP20:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
13113 // CHECK12-NEXT: [[ADD17:%.*]] = add nsw i32 [[TMP20]], 1
13114 // CHECK12-NEXT: store i32 [[ADD17]], i32* [[DOTOMP_IV]], align 4
13115 // CHECK12-NEXT: br label [[OMP_INNER_FOR_COND10]], !llvm.loop [[LOOP29:![0-9]+]]
13116 // CHECK12: omp.inner.for.end18:
13117 // CHECK12-NEXT: br label [[OMP_IF_END]]
13118 // CHECK12: omp_if.end:
13119 // CHECK12-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
13120 // CHECK12: omp.loop.exit:
13121 // CHECK12-NEXT: [[TMP21:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
13122 // CHECK12-NEXT: [[TMP22:%.*]] = load i32, i32* [[TMP21]], align 4
13123 // CHECK12-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP22]])
13124 // CHECK12-NEXT: [[TMP23:%.*]] = load i32, i32* [[DOTOMP_IS_LAST]], align 4
13125 // CHECK12-NEXT: [[TMP24:%.*]] = icmp ne i32 [[TMP23]], 0
13126 // CHECK12-NEXT: br i1 [[TMP24]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
13127 // CHECK12: .omp.final.then:
13128 // CHECK12-NEXT: store i32 100, i32* [[I]], align 4
13129 // CHECK12-NEXT: br label [[DOTOMP_FINAL_DONE]]
13130 // CHECK12: .omp.final.done:
13131 // CHECK12-NEXT: ret void
13132 //
13133 //
13134 // CHECK12-LABEL: define {{[^@]+}}@.omp_outlined..10
13135 // CHECK12-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[DOTPREVIOUS_LB_:%.*]], i64 [[DOTPREVIOUS_UB_:%.*]], i64 [[DOTCAPTURE_EXPR_:%.*]]) #[[ATTR1]] {
13136 // CHECK12-NEXT: entry:
13137 // CHECK12-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
13138 // CHECK12-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
13139 // CHECK12-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i64, align 8
13140 // CHECK12-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i64, align 8
13141 // CHECK12-NEXT: [[DOTCAPTURE_EXPR__ADDR:%.*]] = alloca i64, align 8
13142 // CHECK12-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
13143 // CHECK12-NEXT: [[TMP:%.*]] = alloca i32, align 4
13144 // CHECK12-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4
13145 // CHECK12-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4
13146 // CHECK12-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
13147 // CHECK12-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
13148 // CHECK12-NEXT: [[I:%.*]] = alloca i32, align 4
13149 // CHECK12-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
13150 // CHECK12-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
13151 // CHECK12-NEXT: store i64 [[DOTPREVIOUS_LB_]], i64* [[DOTPREVIOUS_LB__ADDR]], align 8
13152 // CHECK12-NEXT: store i64 [[DOTPREVIOUS_UB_]], i64* [[DOTPREVIOUS_UB__ADDR]], align 8
13153 // CHECK12-NEXT: store i64 [[DOTCAPTURE_EXPR_]], i64* [[DOTCAPTURE_EXPR__ADDR]], align 8
13154 // CHECK12-NEXT: [[CONV:%.*]] = bitcast i64* [[DOTCAPTURE_EXPR__ADDR]] to i8*
13155 // CHECK12-NEXT: store i32 0, i32* [[DOTOMP_LB]], align 4
13156 // CHECK12-NEXT: store i32 99, i32* [[DOTOMP_UB]], align 4
13157 // CHECK12-NEXT: [[TMP0:%.*]] = load i64, i64* [[DOTPREVIOUS_LB__ADDR]], align 8
13158 // CHECK12-NEXT: [[CONV1:%.*]] = trunc i64 [[TMP0]] to i32
13159 // CHECK12-NEXT: [[TMP1:%.*]] = load i64, i64* [[DOTPREVIOUS_UB__ADDR]], align 8
13160 // CHECK12-NEXT: [[CONV2:%.*]] = trunc i64 [[TMP1]] to i32
13161 // CHECK12-NEXT: store i32 [[CONV1]], i32* [[DOTOMP_LB]], align 4
13162 // CHECK12-NEXT: store i32 [[CONV2]], i32* [[DOTOMP_UB]], align 4
13163 // CHECK12-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4
13164 // CHECK12-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4
13165 // CHECK12-NEXT: [[TMP2:%.*]] = load i8, i8* [[CONV]], align 8
13166 // CHECK12-NEXT: [[TOBOOL:%.*]] = trunc i8 [[TMP2]] to i1
13167 // CHECK12-NEXT: br i1 [[TOBOOL]], label [[OMP_IF_THEN:%.*]], label [[OMP_IF_ELSE:%.*]]
13168 // CHECK12: omp_if.then:
13169 // CHECK12-NEXT: [[TMP3:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
13170 // CHECK12-NEXT: [[TMP4:%.*]] = load i32, i32* [[TMP3]], align 4
13171 // CHECK12-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB2]], i32 [[TMP4]], i32 34, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_LB]], i32* [[DOTOMP_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1)
13172 // CHECK12-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
13173 // CHECK12-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP5]], 99
13174 // CHECK12-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
13175 // CHECK12: cond.true:
13176 // CHECK12-NEXT: br label [[COND_END:%.*]]
13177 // CHECK12: cond.false:
13178 // CHECK12-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
13179 // CHECK12-NEXT: br label [[COND_END]]
13180 // CHECK12: cond.end:
13181 // CHECK12-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP6]], [[COND_FALSE]] ]
13182 // CHECK12-NEXT: store i32 [[COND]], i32* [[DOTOMP_UB]], align 4
13183 // CHECK12-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4
13184 // CHECK12-NEXT: store i32 [[TMP7]], i32* [[DOTOMP_IV]], align 4
13185 // CHECK12-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
13186 // CHECK12: omp.inner.for.cond:
13187 // CHECK12-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
13188 // CHECK12-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
13189 // CHECK12-NEXT: [[CMP3:%.*]] = icmp sle i32 [[TMP8]], [[TMP9]]
13190 // CHECK12-NEXT: br i1 [[CMP3]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
13191 // CHECK12: omp.inner.for.body:
13192 // CHECK12-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
13193 // CHECK12-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP10]], 1
13194 // CHECK12-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
13195 // CHECK12-NEXT: store i32 [[ADD]], i32* [[I]], align 4
13196 // CHECK12-NEXT: call void @_Z3fn6v()
13197 // CHECK12-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
13198 // CHECK12: omp.body.continue:
13199 // CHECK12-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
13200 // CHECK12: omp.inner.for.inc:
13201 // CHECK12-NEXT: [[TMP11:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
13202 // CHECK12-NEXT: [[ADD4:%.*]] = add nsw i32 [[TMP11]], 1
13203 // CHECK12-NEXT: store i32 [[ADD4]], i32* [[DOTOMP_IV]], align 4
13204 // CHECK12-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP30:![0-9]+]]
13205 // CHECK12: omp.inner.for.end:
13206 // CHECK12-NEXT: br label [[OMP_IF_END:%.*]]
13207 // CHECK12: omp_if.else:
13208 // CHECK12-NEXT: [[TMP12:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
13209 // CHECK12-NEXT: [[TMP13:%.*]] = load i32, i32* [[TMP12]], align 4
13210 // CHECK12-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB2]], i32 [[TMP13]], i32 34, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_LB]], i32* [[DOTOMP_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1)
13211 // CHECK12-NEXT: [[TMP14:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
13212 // CHECK12-NEXT: [[CMP5:%.*]] = icmp sgt i32 [[TMP14]], 99
13213 // CHECK12-NEXT: br i1 [[CMP5]], label [[COND_TRUE6:%.*]], label [[COND_FALSE7:%.*]]
13214 // CHECK12: cond.true6:
13215 // CHECK12-NEXT: br label [[COND_END8:%.*]]
13216 // CHECK12: cond.false7:
13217 // CHECK12-NEXT: [[TMP15:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
13218 // CHECK12-NEXT: br label [[COND_END8]]
13219 // CHECK12: cond.end8:
13220 // CHECK12-NEXT: [[COND9:%.*]] = phi i32 [ 99, [[COND_TRUE6]] ], [ [[TMP15]], [[COND_FALSE7]] ]
13221 // CHECK12-NEXT: store i32 [[COND9]], i32* [[DOTOMP_UB]], align 4
13222 // CHECK12-NEXT: [[TMP16:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4
13223 // CHECK12-NEXT: store i32 [[TMP16]], i32* [[DOTOMP_IV]], align 4
13224 // CHECK12-NEXT: br label [[OMP_INNER_FOR_COND10:%.*]]
13225 // CHECK12: omp.inner.for.cond10:
13226 // CHECK12-NEXT: [[TMP17:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
13227 // CHECK12-NEXT: [[TMP18:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
13228 // CHECK12-NEXT: [[CMP11:%.*]] = icmp sle i32 [[TMP17]], [[TMP18]]
13229 // CHECK12-NEXT: br i1 [[CMP11]], label [[OMP_INNER_FOR_BODY12:%.*]], label [[OMP_INNER_FOR_END18:%.*]]
13230 // CHECK12: omp.inner.for.body12:
13231 // CHECK12-NEXT: [[TMP19:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
13232 // CHECK12-NEXT: [[MUL13:%.*]] = mul nsw i32 [[TMP19]], 1
13233 // CHECK12-NEXT: [[ADD14:%.*]] = add nsw i32 0, [[MUL13]]
13234 // CHECK12-NEXT: store i32 [[ADD14]], i32* [[I]], align 4
13235 // CHECK12-NEXT: call void @_Z3fn6v()
13236 // CHECK12-NEXT: br label [[OMP_BODY_CONTINUE15:%.*]]
13237 // CHECK12: omp.body.continue15:
13238 // CHECK12-NEXT: br label [[OMP_INNER_FOR_INC16:%.*]]
13239 // CHECK12: omp.inner.for.inc16:
13240 // CHECK12-NEXT: [[TMP20:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
13241 // CHECK12-NEXT: [[ADD17:%.*]] = add nsw i32 [[TMP20]], 1
13242 // CHECK12-NEXT: store i32 [[ADD17]], i32* [[DOTOMP_IV]], align 4
13243 // CHECK12-NEXT: br label [[OMP_INNER_FOR_COND10]], !llvm.loop [[LOOP31:![0-9]+]]
13244 // CHECK12: omp.inner.for.end18:
13245 // CHECK12-NEXT: br label [[OMP_IF_END]]
13246 // CHECK12: omp_if.end:
13247 // CHECK12-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
13248 // CHECK12: omp.loop.exit:
13249 // CHECK12-NEXT: [[TMP21:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
13250 // CHECK12-NEXT: [[TMP22:%.*]] = load i32, i32* [[TMP21]], align 4
13251 // CHECK12-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP22]])
13252 // CHECK12-NEXT: [[TMP23:%.*]] = load i32, i32* [[DOTOMP_IS_LAST]], align 4
13253 // CHECK12-NEXT: [[TMP24:%.*]] = icmp ne i32 [[TMP23]], 0
13254 // CHECK12-NEXT: br i1 [[TMP24]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
13255 // CHECK12: .omp.final.then:
13256 // CHECK12-NEXT: store i32 100, i32* [[I]], align 4
13257 // CHECK12-NEXT: br label [[DOTOMP_FINAL_DONE]]
13258 // CHECK12: .omp.final.done:
13259 // CHECK12-NEXT: ret void
13260 //
13261 //
13262 // CHECK12-LABEL: define {{[^@]+}}@_Z5tmainIiEiT_
13263 // CHECK12-SAME: (i32 [[ARG:%.*]]) #[[ATTR0]] comdat {
13264 // CHECK12-NEXT: entry:
13265 // CHECK12-NEXT: [[ARG_ADDR:%.*]] = alloca i32, align 4
13266 // CHECK12-NEXT: [[TMP:%.*]] = alloca i32, align 4
13267 // CHECK12-NEXT: [[_TMP1:%.*]] = alloca i32, align 4
13268 // CHECK12-NEXT: [[ARG_CASTED:%.*]] = alloca i64, align 8
13269 // CHECK12-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [1 x i8*], align 8
13270 // CHECK12-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [1 x i8*], align 8
13271 // CHECK12-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [1 x i8*], align 8
13272 // CHECK12-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i8, align 1
13273 // CHECK12-NEXT: [[_TMP5:%.*]] = alloca i32, align 4
13274 // CHECK12-NEXT: store i32 [[ARG]], i32* [[ARG_ADDR]], align 4
13275 // CHECK12-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i64 100)
13276 // CHECK12-NEXT: [[TMP0:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiEiT__l57.region_id, i32 0, i8** null, i8** null, i64* null, i64* null, i8** null, i8** null, i32 0, i32 0)
13277 // CHECK12-NEXT: [[TMP1:%.*]] = icmp ne i32 [[TMP0]], 0
13278 // CHECK12-NEXT: br i1 [[TMP1]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
13279 // CHECK12: omp_offload.failed:
13280 // CHECK12-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiEiT__l57() #[[ATTR2]]
13281 // CHECK12-NEXT: br label [[OMP_OFFLOAD_CONT]]
13282 // CHECK12: omp_offload.cont:
13283 // CHECK12-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i64 100)
13284 // CHECK12-NEXT: [[TMP2:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiEiT__l62.region_id, i32 0, i8** null, i8** null, i64* null, i64* null, i8** null, i8** null, i32 0, i32 1)
13285 // CHECK12-NEXT: [[TMP3:%.*]] = icmp ne i32 [[TMP2]], 0
13286 // CHECK12-NEXT: br i1 [[TMP3]], label [[OMP_OFFLOAD_FAILED2:%.*]], label [[OMP_OFFLOAD_CONT3:%.*]]
13287 // CHECK12: omp_offload.failed2:
13288 // CHECK12-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiEiT__l62() #[[ATTR2]]
13289 // CHECK12-NEXT: br label [[OMP_OFFLOAD_CONT3]]
13290 // CHECK12: omp_offload.cont3:
13291 // CHECK12-NEXT: [[TMP4:%.*]] = load i32, i32* [[ARG_ADDR]], align 4
13292 // CHECK12-NEXT: [[CONV:%.*]] = bitcast i64* [[ARG_CASTED]] to i32*
13293 // CHECK12-NEXT: store i32 [[TMP4]], i32* [[CONV]], align 4
13294 // CHECK12-NEXT: [[TMP5:%.*]] = load i64, i64* [[ARG_CASTED]], align 8
13295 // CHECK12-NEXT: [[TMP6:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
13296 // CHECK12-NEXT: [[TMP7:%.*]] = bitcast i8** [[TMP6]] to i64*
13297 // CHECK12-NEXT: store i64 [[TMP5]], i64* [[TMP7]], align 8
13298 // CHECK12-NEXT: [[TMP8:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
13299 // CHECK12-NEXT: [[TMP9:%.*]] = bitcast i8** [[TMP8]] to i64*
13300 // CHECK12-NEXT: store i64 [[TMP5]], i64* [[TMP9]], align 8
13301 // CHECK12-NEXT: [[TMP10:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0
13302 // CHECK12-NEXT: store i8* null, i8** [[TMP10]], align 8
13303 // CHECK12-NEXT: [[TMP11:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
13304 // CHECK12-NEXT: [[TMP12:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
13305 // CHECK12-NEXT: [[TMP13:%.*]] = load i32, i32* [[ARG_ADDR]], align 4
13306 // CHECK12-NEXT: [[TOBOOL:%.*]] = icmp ne i32 [[TMP13]], 0
13307 // CHECK12-NEXT: [[FROMBOOL:%.*]] = zext i1 [[TOBOOL]] to i8
13308 // CHECK12-NEXT: store i8 [[FROMBOOL]], i8* [[DOTCAPTURE_EXPR_]], align 1
13309 // CHECK12-NEXT: [[TMP14:%.*]] = load i8, i8* [[DOTCAPTURE_EXPR_]], align 1
13310 // CHECK12-NEXT: [[TOBOOL4:%.*]] = trunc i8 [[TMP14]] to i1
13311 // CHECK12-NEXT: [[TMP15:%.*]] = select i1 [[TOBOOL4]], i32 0, i32 1
13312 // CHECK12-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i64 100)
13313 // CHECK12-NEXT: [[TMP16:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiEiT__l67.region_id, i32 1, i8** [[TMP11]], i8** [[TMP12]], i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_sizes.17, i32 0, i32 0), i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_maptypes.18, i32 0, i32 0), i8** null, i8** null, i32 0, i32 [[TMP15]])
13314 // CHECK12-NEXT: [[TMP17:%.*]] = icmp ne i32 [[TMP16]], 0
13315 // CHECK12-NEXT: br i1 [[TMP17]], label [[OMP_OFFLOAD_FAILED6:%.*]], label [[OMP_OFFLOAD_CONT7:%.*]]
13316 // CHECK12: omp_offload.failed6:
13317 // CHECK12-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiEiT__l67(i64 [[TMP5]]) #[[ATTR2]]
13318 // CHECK12-NEXT: br label [[OMP_OFFLOAD_CONT7]]
13319 // CHECK12: omp_offload.cont7:
13320 // CHECK12-NEXT: ret i32 0
13321 //
13322 //
13323 // CHECK12-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiEiT__l57
13324 // CHECK12-SAME: () #[[ATTR1]] {
13325 // CHECK12-NEXT: entry:
13326 // CHECK12-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 0, void (i32*, i32*, ...)* bitcast (void (i32*, i32*)* @.omp_outlined..11 to void (i32*, i32*, ...)*))
13327 // CHECK12-NEXT: ret void
13328 //
13329 //
13330 // CHECK12-LABEL: define {{[^@]+}}@.omp_outlined..11
13331 // CHECK12-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]]) #[[ATTR1]] {
13332 // CHECK12-NEXT: entry:
13333 // CHECK12-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
13334 // CHECK12-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
13335 // CHECK12-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
13336 // CHECK12-NEXT: [[TMP:%.*]] = alloca i32, align 4
13337 // CHECK12-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4
13338 // CHECK12-NEXT: [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4
13339 // CHECK12-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
13340 // CHECK12-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
13341 // CHECK12-NEXT: [[I:%.*]] = alloca i32, align 4
13342 // CHECK12-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
13343 // CHECK12-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
13344 // CHECK12-NEXT: store i32 0, i32* [[DOTOMP_COMB_LB]], align 4
13345 // CHECK12-NEXT: store i32 99, i32* [[DOTOMP_COMB_UB]], align 4
13346 // CHECK12-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4
13347 // CHECK12-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4
13348 // CHECK12-NEXT: [[TMP0:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
13349 // CHECK12-NEXT: [[TMP1:%.*]] = load i32, i32* [[TMP0]], align 4
13350 // CHECK12-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB1]], i32 [[TMP1]], i32 92, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_COMB_LB]], i32* [[DOTOMP_COMB_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1)
13351 // CHECK12-NEXT: [[TMP2:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
13352 // CHECK12-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP2]], 99
13353 // CHECK12-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
13354 // CHECK12: cond.true:
13355 // CHECK12-NEXT: br label [[COND_END:%.*]]
13356 // CHECK12: cond.false:
13357 // CHECK12-NEXT: [[TMP3:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
13358 // CHECK12-NEXT: br label [[COND_END]]
13359 // CHECK12: cond.end:
13360 // CHECK12-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP3]], [[COND_FALSE]] ]
13361 // CHECK12-NEXT: store i32 [[COND]], i32* [[DOTOMP_COMB_UB]], align 4
13362 // CHECK12-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4
13363 // CHECK12-NEXT: store i32 [[TMP4]], i32* [[DOTOMP_IV]], align 4
13364 // CHECK12-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
13365 // CHECK12: omp.inner.for.cond:
13366 // CHECK12-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
13367 // CHECK12-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
13368 // CHECK12-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP5]], [[TMP6]]
13369 // CHECK12-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
13370 // CHECK12: omp.inner.for.body:
13371 // CHECK12-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4
13372 // CHECK12-NEXT: [[TMP8:%.*]] = zext i32 [[TMP7]] to i64
13373 // CHECK12-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
13374 // CHECK12-NEXT: [[TMP10:%.*]] = zext i32 [[TMP9]] to i64
13375 // CHECK12-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB3]], i32 2, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64, i64)* @.omp_outlined..12 to void (i32*, i32*, ...)*), i64 [[TMP8]], i64 [[TMP10]])
13376 // CHECK12-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
13377 // CHECK12: omp.inner.for.inc:
13378 // CHECK12-NEXT: [[TMP11:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
13379 // CHECK12-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4
13380 // CHECK12-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP11]], [[TMP12]]
13381 // CHECK12-NEXT: store i32 [[ADD]], i32* [[DOTOMP_IV]], align 4
13382 // CHECK12-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP32:![0-9]+]]
13383 // CHECK12: omp.inner.for.end:
13384 // CHECK12-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
13385 // CHECK12: omp.loop.exit:
13386 // CHECK12-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP1]])
13387 // CHECK12-NEXT: [[TMP13:%.*]] = load i32, i32* [[DOTOMP_IS_LAST]], align 4
13388 // CHECK12-NEXT: [[TMP14:%.*]] = icmp ne i32 [[TMP13]], 0
13389 // CHECK12-NEXT: br i1 [[TMP14]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
13390 // CHECK12: .omp.final.then:
13391 // CHECK12-NEXT: store i32 100, i32* [[I]], align 4
13392 // CHECK12-NEXT: br label [[DOTOMP_FINAL_DONE]]
13393 // CHECK12: .omp.final.done:
13394 // CHECK12-NEXT: ret void
13395 //
13396 //
13397 // CHECK12-LABEL: define {{[^@]+}}@.omp_outlined..12
13398 // CHECK12-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[DOTPREVIOUS_LB_:%.*]], i64 [[DOTPREVIOUS_UB_:%.*]]) #[[ATTR1]] {
13399 // CHECK12-NEXT: entry:
13400 // CHECK12-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
13401 // CHECK12-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
13402 // CHECK12-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i64, align 8
13403 // CHECK12-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i64, align 8
13404 // CHECK12-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
13405 // CHECK12-NEXT: [[TMP:%.*]] = alloca i32, align 4
13406 // CHECK12-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4
13407 // CHECK12-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4
13408 // CHECK12-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
13409 // CHECK12-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
13410 // CHECK12-NEXT: [[I:%.*]] = alloca i32, align 4
13411 // CHECK12-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
13412 // CHECK12-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
13413 // CHECK12-NEXT: store i64 [[DOTPREVIOUS_LB_]], i64* [[DOTPREVIOUS_LB__ADDR]], align 8
13414 // CHECK12-NEXT: store i64 [[DOTPREVIOUS_UB_]], i64* [[DOTPREVIOUS_UB__ADDR]], align 8
13415 // CHECK12-NEXT: store i32 0, i32* [[DOTOMP_LB]], align 4
13416 // CHECK12-NEXT: store i32 99, i32* [[DOTOMP_UB]], align 4
13417 // CHECK12-NEXT: [[TMP0:%.*]] = load i64, i64* [[DOTPREVIOUS_LB__ADDR]], align 8
13418 // CHECK12-NEXT: [[CONV:%.*]] = trunc i64 [[TMP0]] to i32
13419 // CHECK12-NEXT: [[TMP1:%.*]] = load i64, i64* [[DOTPREVIOUS_UB__ADDR]], align 8
13420 // CHECK12-NEXT: [[CONV1:%.*]] = trunc i64 [[TMP1]] to i32
13421 // CHECK12-NEXT: store i32 [[CONV]], i32* [[DOTOMP_LB]], align 4
13422 // CHECK12-NEXT: store i32 [[CONV1]], i32* [[DOTOMP_UB]], align 4
13423 // CHECK12-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4
13424 // CHECK12-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4
13425 // CHECK12-NEXT: [[TMP2:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
13426 // CHECK12-NEXT: [[TMP3:%.*]] = load i32, i32* [[TMP2]], align 4
13427 // CHECK12-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB2]], i32 [[TMP3]], i32 34, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_LB]], i32* [[DOTOMP_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1)
13428 // CHECK12-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
13429 // CHECK12-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP4]], 99
13430 // CHECK12-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
13431 // CHECK12: cond.true:
13432 // CHECK12-NEXT: br label [[COND_END:%.*]]
13433 // CHECK12: cond.false:
13434 // CHECK12-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
13435 // CHECK12-NEXT: br label [[COND_END]]
13436 // CHECK12: cond.end:
13437 // CHECK12-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP5]], [[COND_FALSE]] ]
13438 // CHECK12-NEXT: store i32 [[COND]], i32* [[DOTOMP_UB]], align 4
13439 // CHECK12-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4
13440 // CHECK12-NEXT: store i32 [[TMP6]], i32* [[DOTOMP_IV]], align 4
13441 // CHECK12-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
13442 // CHECK12: omp.inner.for.cond:
13443 // CHECK12-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
13444 // CHECK12-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
13445 // CHECK12-NEXT: [[CMP2:%.*]] = icmp sle i32 [[TMP7]], [[TMP8]]
13446 // CHECK12-NEXT: br i1 [[CMP2]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
13447 // CHECK12: omp.inner.for.body:
13448 // CHECK12-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
13449 // CHECK12-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP9]], 1
13450 // CHECK12-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
13451 // CHECK12-NEXT: store i32 [[ADD]], i32* [[I]], align 4
13452 // CHECK12-NEXT: call void @_Z3fn1v()
13453 // CHECK12-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
13454 // CHECK12: omp.body.continue:
13455 // CHECK12-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
13456 // CHECK12: omp.inner.for.inc:
13457 // CHECK12-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
13458 // CHECK12-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP10]], 1
13459 // CHECK12-NEXT: store i32 [[ADD3]], i32* [[DOTOMP_IV]], align 4
13460 // CHECK12-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP33:![0-9]+]]
13461 // CHECK12: omp.inner.for.end:
13462 // CHECK12-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
13463 // CHECK12: omp.loop.exit:
13464 // CHECK12-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP3]])
13465 // CHECK12-NEXT: [[TMP11:%.*]] = load i32, i32* [[DOTOMP_IS_LAST]], align 4
13466 // CHECK12-NEXT: [[TMP12:%.*]] = icmp ne i32 [[TMP11]], 0
13467 // CHECK12-NEXT: br i1 [[TMP12]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
13468 // CHECK12: .omp.final.then:
13469 // CHECK12-NEXT: store i32 100, i32* [[I]], align 4
13470 // CHECK12-NEXT: br label [[DOTOMP_FINAL_DONE]]
13471 // CHECK12: .omp.final.done:
13472 // CHECK12-NEXT: ret void
13473 //
13474 //
13475 // CHECK12-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiEiT__l62
13476 // CHECK12-SAME: () #[[ATTR1]] {
13477 // CHECK12-NEXT: entry:
13478 // CHECK12-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 0, void (i32*, i32*, ...)* bitcast (void (i32*, i32*)* @.omp_outlined..13 to void (i32*, i32*, ...)*))
13479 // CHECK12-NEXT: ret void
13480 //
13481 //
13482 // CHECK12-LABEL: define {{[^@]+}}@.omp_outlined..13
13483 // CHECK12-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]]) #[[ATTR1]] {
13484 // CHECK12-NEXT: entry:
13485 // CHECK12-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
13486 // CHECK12-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
13487 // CHECK12-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
13488 // CHECK12-NEXT: [[TMP:%.*]] = alloca i32, align 4
13489 // CHECK12-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4
13490 // CHECK12-NEXT: [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4
13491 // CHECK12-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
13492 // CHECK12-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
13493 // CHECK12-NEXT: [[I:%.*]] = alloca i32, align 4
13494 // CHECK12-NEXT: [[DOTBOUND_ZERO_ADDR:%.*]] = alloca i32, align 4
13495 // CHECK12-NEXT: store i32 0, i32* [[DOTBOUND_ZERO_ADDR]], align 4
13496 // CHECK12-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
13497 // CHECK12-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
13498 // CHECK12-NEXT: store i32 0, i32* [[DOTOMP_COMB_LB]], align 4
13499 // CHECK12-NEXT: store i32 99, i32* [[DOTOMP_COMB_UB]], align 4
13500 // CHECK12-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4
13501 // CHECK12-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4
13502 // CHECK12-NEXT: [[TMP0:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
13503 // CHECK12-NEXT: [[TMP1:%.*]] = load i32, i32* [[TMP0]], align 4
13504 // CHECK12-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB1]], i32 [[TMP1]], i32 92, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_COMB_LB]], i32* [[DOTOMP_COMB_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1)
13505 // CHECK12-NEXT: [[TMP2:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
13506 // CHECK12-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP2]], 99
13507 // CHECK12-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
13508 // CHECK12: cond.true:
13509 // CHECK12-NEXT: br label [[COND_END:%.*]]
13510 // CHECK12: cond.false:
13511 // CHECK12-NEXT: [[TMP3:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
13512 // CHECK12-NEXT: br label [[COND_END]]
13513 // CHECK12: cond.end:
13514 // CHECK12-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP3]], [[COND_FALSE]] ]
13515 // CHECK12-NEXT: store i32 [[COND]], i32* [[DOTOMP_COMB_UB]], align 4
13516 // CHECK12-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4
13517 // CHECK12-NEXT: store i32 [[TMP4]], i32* [[DOTOMP_IV]], align 4
13518 // CHECK12-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
13519 // CHECK12: omp.inner.for.cond:
13520 // CHECK12-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
13521 // CHECK12-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
13522 // CHECK12-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP5]], [[TMP6]]
13523 // CHECK12-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
13524 // CHECK12: omp.inner.for.body:
13525 // CHECK12-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4
13526 // CHECK12-NEXT: [[TMP8:%.*]] = zext i32 [[TMP7]] to i64
13527 // CHECK12-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
13528 // CHECK12-NEXT: [[TMP10:%.*]] = zext i32 [[TMP9]] to i64
13529 // CHECK12-NEXT: call void @__kmpc_serialized_parallel(%struct.ident_t* @[[GLOB3]], i32 [[TMP1]])
13530 // CHECK12-NEXT: [[TMP11:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
13531 // CHECK12-NEXT: call void @.omp_outlined..14(i32* [[TMP11]], i32* [[DOTBOUND_ZERO_ADDR]], i64 [[TMP8]], i64 [[TMP10]]) #[[ATTR2]]
13532 // CHECK12-NEXT: call void @__kmpc_end_serialized_parallel(%struct.ident_t* @[[GLOB3]], i32 [[TMP1]])
13533 // CHECK12-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
13534 // CHECK12: omp.inner.for.inc:
13535 // CHECK12-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
13536 // CHECK12-NEXT: [[TMP13:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4
13537 // CHECK12-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP12]], [[TMP13]]
13538 // CHECK12-NEXT: store i32 [[ADD]], i32* [[DOTOMP_IV]], align 4
13539 // CHECK12-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP34:![0-9]+]]
13540 // CHECK12: omp.inner.for.end:
13541 // CHECK12-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
13542 // CHECK12: omp.loop.exit:
13543 // CHECK12-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP1]])
13544 // CHECK12-NEXT: [[TMP14:%.*]] = load i32, i32* [[DOTOMP_IS_LAST]], align 4
13545 // CHECK12-NEXT: [[TMP15:%.*]] = icmp ne i32 [[TMP14]], 0
13546 // CHECK12-NEXT: br i1 [[TMP15]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
13547 // CHECK12: .omp.final.then:
13548 // CHECK12-NEXT: store i32 100, i32* [[I]], align 4
13549 // CHECK12-NEXT: br label [[DOTOMP_FINAL_DONE]]
13550 // CHECK12: .omp.final.done:
13551 // CHECK12-NEXT: ret void
13552 //
13553 //
13554 // CHECK12-LABEL: define {{[^@]+}}@.omp_outlined..14
13555 // CHECK12-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[DOTPREVIOUS_LB_:%.*]], i64 [[DOTPREVIOUS_UB_:%.*]]) #[[ATTR1]] {
13556 // CHECK12-NEXT: entry:
13557 // CHECK12-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
13558 // CHECK12-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
13559 // CHECK12-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i64, align 8
13560 // CHECK12-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i64, align 8
13561 // CHECK12-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
13562 // CHECK12-NEXT: [[TMP:%.*]] = alloca i32, align 4
13563 // CHECK12-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4
13564 // CHECK12-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4
13565 // CHECK12-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
13566 // CHECK12-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
13567 // CHECK12-NEXT: [[I:%.*]] = alloca i32, align 4
13568 // CHECK12-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
13569 // CHECK12-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
13570 // CHECK12-NEXT: store i64 [[DOTPREVIOUS_LB_]], i64* [[DOTPREVIOUS_LB__ADDR]], align 8
13571 // CHECK12-NEXT: store i64 [[DOTPREVIOUS_UB_]], i64* [[DOTPREVIOUS_UB__ADDR]], align 8
13572 // CHECK12-NEXT: store i32 0, i32* [[DOTOMP_LB]], align 4
13573 // CHECK12-NEXT: store i32 99, i32* [[DOTOMP_UB]], align 4
13574 // CHECK12-NEXT: [[TMP0:%.*]] = load i64, i64* [[DOTPREVIOUS_LB__ADDR]], align 8
13575 // CHECK12-NEXT: [[CONV:%.*]] = trunc i64 [[TMP0]] to i32
13576 // CHECK12-NEXT: [[TMP1:%.*]] = load i64, i64* [[DOTPREVIOUS_UB__ADDR]], align 8
13577 // CHECK12-NEXT: [[CONV1:%.*]] = trunc i64 [[TMP1]] to i32
13578 // CHECK12-NEXT: store i32 [[CONV]], i32* [[DOTOMP_LB]], align 4
13579 // CHECK12-NEXT: store i32 [[CONV1]], i32* [[DOTOMP_UB]], align 4
13580 // CHECK12-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4
13581 // CHECK12-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4
13582 // CHECK12-NEXT: [[TMP2:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
13583 // CHECK12-NEXT: [[TMP3:%.*]] = load i32, i32* [[TMP2]], align 4
13584 // CHECK12-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB2]], i32 [[TMP3]], i32 34, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_LB]], i32* [[DOTOMP_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1)
13585 // CHECK12-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
13586 // CHECK12-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP4]], 99
13587 // CHECK12-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
13588 // CHECK12: cond.true:
13589 // CHECK12-NEXT: br label [[COND_END:%.*]]
13590 // CHECK12: cond.false:
13591 // CHECK12-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
13592 // CHECK12-NEXT: br label [[COND_END]]
13593 // CHECK12: cond.end:
13594 // CHECK12-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP5]], [[COND_FALSE]] ]
13595 // CHECK12-NEXT: store i32 [[COND]], i32* [[DOTOMP_UB]], align 4
13596 // CHECK12-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4
13597 // CHECK12-NEXT: store i32 [[TMP6]], i32* [[DOTOMP_IV]], align 4
13598 // CHECK12-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
13599 // CHECK12: omp.inner.for.cond:
13600 // CHECK12-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
13601 // CHECK12-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
13602 // CHECK12-NEXT: [[CMP2:%.*]] = icmp sle i32 [[TMP7]], [[TMP8]]
13603 // CHECK12-NEXT: br i1 [[CMP2]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
13604 // CHECK12: omp.inner.for.body:
13605 // CHECK12-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
13606 // CHECK12-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP9]], 1
13607 // CHECK12-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
13608 // CHECK12-NEXT: store i32 [[ADD]], i32* [[I]], align 4
13609 // CHECK12-NEXT: call void @_Z3fn2v()
13610 // CHECK12-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
13611 // CHECK12: omp.body.continue:
13612 // CHECK12-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
13613 // CHECK12: omp.inner.for.inc:
13614 // CHECK12-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
13615 // CHECK12-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP10]], 1
13616 // CHECK12-NEXT: store i32 [[ADD3]], i32* [[DOTOMP_IV]], align 4
13617 // CHECK12-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP35:![0-9]+]]
13618 // CHECK12: omp.inner.for.end:
13619 // CHECK12-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
13620 // CHECK12: omp.loop.exit:
13621 // CHECK12-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP3]])
13622 // CHECK12-NEXT: [[TMP11:%.*]] = load i32, i32* [[DOTOMP_IS_LAST]], align 4
13623 // CHECK12-NEXT: [[TMP12:%.*]] = icmp ne i32 [[TMP11]], 0
13624 // CHECK12-NEXT: br i1 [[TMP12]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
13625 // CHECK12: .omp.final.then:
13626 // CHECK12-NEXT: store i32 100, i32* [[I]], align 4
13627 // CHECK12-NEXT: br label [[DOTOMP_FINAL_DONE]]
13628 // CHECK12: .omp.final.done:
13629 // CHECK12-NEXT: ret void
13630 //
13631 //
13632 // CHECK12-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiEiT__l67
13633 // CHECK12-SAME: (i64 [[ARG:%.*]]) #[[ATTR1]] {
13634 // CHECK12-NEXT: entry:
13635 // CHECK12-NEXT: [[ARG_ADDR:%.*]] = alloca i64, align 8
13636 // CHECK12-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i8, align 1
13637 // CHECK12-NEXT: [[DOTCAPTURE_EXPR__CASTED:%.*]] = alloca i64, align 8
13638 // CHECK12-NEXT: store i64 [[ARG]], i64* [[ARG_ADDR]], align 8
13639 // CHECK12-NEXT: [[CONV:%.*]] = bitcast i64* [[ARG_ADDR]] to i32*
13640 // CHECK12-NEXT: [[TMP0:%.*]] = load i32, i32* [[CONV]], align 8
13641 // CHECK12-NEXT: [[TOBOOL:%.*]] = icmp ne i32 [[TMP0]], 0
13642 // CHECK12-NEXT: [[FROMBOOL:%.*]] = zext i1 [[TOBOOL]] to i8
13643 // CHECK12-NEXT: store i8 [[FROMBOOL]], i8* [[DOTCAPTURE_EXPR_]], align 1
13644 // CHECK12-NEXT: [[TMP1:%.*]] = load i8, i8* [[DOTCAPTURE_EXPR_]], align 1
13645 // CHECK12-NEXT: [[TOBOOL1:%.*]] = trunc i8 [[TMP1]] to i1
13646 // CHECK12-NEXT: [[CONV2:%.*]] = bitcast i64* [[DOTCAPTURE_EXPR__CASTED]] to i8*
13647 // CHECK12-NEXT: [[FROMBOOL3:%.*]] = zext i1 [[TOBOOL1]] to i8
13648 // CHECK12-NEXT: store i8 [[FROMBOOL3]], i8* [[CONV2]], align 1
13649 // CHECK12-NEXT: [[TMP2:%.*]] = load i64, i64* [[DOTCAPTURE_EXPR__CASTED]], align 8
13650 // CHECK12-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64)* @.omp_outlined..15 to void (i32*, i32*, ...)*), i64 [[TMP2]])
13651 // CHECK12-NEXT: ret void
13652 //
13653 //
13654 // CHECK12-LABEL: define {{[^@]+}}@.omp_outlined..15
13655 // CHECK12-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[DOTCAPTURE_EXPR_:%.*]]) #[[ATTR1]] {
13656 // CHECK12-NEXT: entry:
13657 // CHECK12-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
13658 // CHECK12-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
13659 // CHECK12-NEXT: [[DOTCAPTURE_EXPR__ADDR:%.*]] = alloca i64, align 8
13660 // CHECK12-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
13661 // CHECK12-NEXT: [[TMP:%.*]] = alloca i32, align 4
13662 // CHECK12-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4
13663 // CHECK12-NEXT: [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4
13664 // CHECK12-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
13665 // CHECK12-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
13666 // CHECK12-NEXT: [[I:%.*]] = alloca i32, align 4
13667 // CHECK12-NEXT: [[DOTBOUND_ZERO_ADDR:%.*]] = alloca i32, align 4
13668 // CHECK12-NEXT: store i32 0, i32* [[DOTBOUND_ZERO_ADDR]], align 4
13669 // CHECK12-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
13670 // CHECK12-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
13671 // CHECK12-NEXT: store i64 [[DOTCAPTURE_EXPR_]], i64* [[DOTCAPTURE_EXPR__ADDR]], align 8
13672 // CHECK12-NEXT: [[CONV:%.*]] = bitcast i64* [[DOTCAPTURE_EXPR__ADDR]] to i8*
13673 // CHECK12-NEXT: store i32 0, i32* [[DOTOMP_COMB_LB]], align 4
13674 // CHECK12-NEXT: store i32 99, i32* [[DOTOMP_COMB_UB]], align 4
13675 // CHECK12-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4
13676 // CHECK12-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4
13677 // CHECK12-NEXT: [[TMP0:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
13678 // CHECK12-NEXT: [[TMP1:%.*]] = load i32, i32* [[TMP0]], align 4
13679 // CHECK12-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB1]], i32 [[TMP1]], i32 92, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_COMB_LB]], i32* [[DOTOMP_COMB_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1)
13680 // CHECK12-NEXT: [[TMP2:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
13681 // CHECK12-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP2]], 99
13682 // CHECK12-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
13683 // CHECK12: cond.true:
13684 // CHECK12-NEXT: br label [[COND_END:%.*]]
13685 // CHECK12: cond.false:
13686 // CHECK12-NEXT: [[TMP3:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
13687 // CHECK12-NEXT: br label [[COND_END]]
13688 // CHECK12: cond.end:
13689 // CHECK12-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP3]], [[COND_FALSE]] ]
13690 // CHECK12-NEXT: store i32 [[COND]], i32* [[DOTOMP_COMB_UB]], align 4
13691 // CHECK12-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4
13692 // CHECK12-NEXT: store i32 [[TMP4]], i32* [[DOTOMP_IV]], align 4
13693 // CHECK12-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
13694 // CHECK12: omp.inner.for.cond:
13695 // CHECK12-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
13696 // CHECK12-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
13697 // CHECK12-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP5]], [[TMP6]]
13698 // CHECK12-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
13699 // CHECK12: omp.inner.for.body:
13700 // CHECK12-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4
13701 // CHECK12-NEXT: [[TMP8:%.*]] = zext i32 [[TMP7]] to i64
13702 // CHECK12-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
13703 // CHECK12-NEXT: [[TMP10:%.*]] = zext i32 [[TMP9]] to i64
13704 // CHECK12-NEXT: [[TMP11:%.*]] = load i8, i8* [[CONV]], align 8
13705 // CHECK12-NEXT: [[TOBOOL:%.*]] = trunc i8 [[TMP11]] to i1
13706 // CHECK12-NEXT: br i1 [[TOBOOL]], label [[OMP_IF_THEN:%.*]], label [[OMP_IF_ELSE:%.*]]
13707 // CHECK12: omp_if.then:
13708 // CHECK12-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB3]], i32 2, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64, i64)* @.omp_outlined..16 to void (i32*, i32*, ...)*), i64 [[TMP8]], i64 [[TMP10]])
13709 // CHECK12-NEXT: br label [[OMP_IF_END:%.*]]
13710 // CHECK12: omp_if.else:
13711 // CHECK12-NEXT: call void @__kmpc_serialized_parallel(%struct.ident_t* @[[GLOB3]], i32 [[TMP1]])
13712 // CHECK12-NEXT: [[TMP12:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
13713 // CHECK12-NEXT: call void @.omp_outlined..16(i32* [[TMP12]], i32* [[DOTBOUND_ZERO_ADDR]], i64 [[TMP8]], i64 [[TMP10]]) #[[ATTR2]]
13714 // CHECK12-NEXT: call void @__kmpc_end_serialized_parallel(%struct.ident_t* @[[GLOB3]], i32 [[TMP1]])
13715 // CHECK12-NEXT: br label [[OMP_IF_END]]
13716 // CHECK12: omp_if.end:
13717 // CHECK12-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
13718 // CHECK12: omp.inner.for.inc:
13719 // CHECK12-NEXT: [[TMP13:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
13720 // CHECK12-NEXT: [[TMP14:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4
13721 // CHECK12-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP13]], [[TMP14]]
13722 // CHECK12-NEXT: store i32 [[ADD]], i32* [[DOTOMP_IV]], align 4
13723 // CHECK12-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP36:![0-9]+]]
13724 // CHECK12: omp.inner.for.end:
13725 // CHECK12-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
13726 // CHECK12: omp.loop.exit:
13727 // CHECK12-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP1]])
13728 // CHECK12-NEXT: [[TMP15:%.*]] = load i32, i32* [[DOTOMP_IS_LAST]], align 4
13729 // CHECK12-NEXT: [[TMP16:%.*]] = icmp ne i32 [[TMP15]], 0
13730 // CHECK12-NEXT: br i1 [[TMP16]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
13731 // CHECK12: .omp.final.then:
13732 // CHECK12-NEXT: store i32 100, i32* [[I]], align 4
13733 // CHECK12-NEXT: br label [[DOTOMP_FINAL_DONE]]
13734 // CHECK12: .omp.final.done:
13735 // CHECK12-NEXT: ret void
13736 //
13737 //
13738 // CHECK12-LABEL: define {{[^@]+}}@.omp_outlined..16
13739 // CHECK12-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[DOTPREVIOUS_LB_:%.*]], i64 [[DOTPREVIOUS_UB_:%.*]]) #[[ATTR1]] {
13740 // CHECK12-NEXT: entry:
13741 // CHECK12-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
13742 // CHECK12-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
13743 // CHECK12-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i64, align 8
13744 // CHECK12-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i64, align 8
13745 // CHECK12-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
13746 // CHECK12-NEXT: [[TMP:%.*]] = alloca i32, align 4
13747 // CHECK12-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4
13748 // CHECK12-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4
13749 // CHECK12-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
13750 // CHECK12-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
13751 // CHECK12-NEXT: [[I:%.*]] = alloca i32, align 4
13752 // CHECK12-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
13753 // CHECK12-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
13754 // CHECK12-NEXT: store i64 [[DOTPREVIOUS_LB_]], i64* [[DOTPREVIOUS_LB__ADDR]], align 8
13755 // CHECK12-NEXT: store i64 [[DOTPREVIOUS_UB_]], i64* [[DOTPREVIOUS_UB__ADDR]], align 8
13756 // CHECK12-NEXT: store i32 0, i32* [[DOTOMP_LB]], align 4
13757 // CHECK12-NEXT: store i32 99, i32* [[DOTOMP_UB]], align 4
13758 // CHECK12-NEXT: [[TMP0:%.*]] = load i64, i64* [[DOTPREVIOUS_LB__ADDR]], align 8
13759 // CHECK12-NEXT: [[CONV:%.*]] = trunc i64 [[TMP0]] to i32
13760 // CHECK12-NEXT: [[TMP1:%.*]] = load i64, i64* [[DOTPREVIOUS_UB__ADDR]], align 8
13761 // CHECK12-NEXT: [[CONV1:%.*]] = trunc i64 [[TMP1]] to i32
13762 // CHECK12-NEXT: store i32 [[CONV]], i32* [[DOTOMP_LB]], align 4
13763 // CHECK12-NEXT: store i32 [[CONV1]], i32* [[DOTOMP_UB]], align 4
13764 // CHECK12-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4
13765 // CHECK12-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4
13766 // CHECK12-NEXT: [[TMP2:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
13767 // CHECK12-NEXT: [[TMP3:%.*]] = load i32, i32* [[TMP2]], align 4
13768 // CHECK12-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB2]], i32 [[TMP3]], i32 34, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_LB]], i32* [[DOTOMP_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1)
13769 // CHECK12-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
13770 // CHECK12-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP4]], 99
13771 // CHECK12-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
13772 // CHECK12: cond.true:
13773 // CHECK12-NEXT: br label [[COND_END:%.*]]
13774 // CHECK12: cond.false:
13775 // CHECK12-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
13776 // CHECK12-NEXT: br label [[COND_END]]
13777 // CHECK12: cond.end:
13778 // CHECK12-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP5]], [[COND_FALSE]] ]
13779 // CHECK12-NEXT: store i32 [[COND]], i32* [[DOTOMP_UB]], align 4
13780 // CHECK12-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4
13781 // CHECK12-NEXT: store i32 [[TMP6]], i32* [[DOTOMP_IV]], align 4
13782 // CHECK12-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
13783 // CHECK12: omp.inner.for.cond:
13784 // CHECK12-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
13785 // CHECK12-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
13786 // CHECK12-NEXT: [[CMP2:%.*]] = icmp sle i32 [[TMP7]], [[TMP8]]
13787 // CHECK12-NEXT: br i1 [[CMP2]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
13788 // CHECK12: omp.inner.for.body:
13789 // CHECK12-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
13790 // CHECK12-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP9]], 1
13791 // CHECK12-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
13792 // CHECK12-NEXT: store i32 [[ADD]], i32* [[I]], align 4
13793 // CHECK12-NEXT: call void @_Z3fn3v()
13794 // CHECK12-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
13795 // CHECK12: omp.body.continue:
13796 // CHECK12-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
13797 // CHECK12: omp.inner.for.inc:
13798 // CHECK12-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
13799 // CHECK12-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP10]], 1
13800 // CHECK12-NEXT: store i32 [[ADD3]], i32* [[DOTOMP_IV]], align 4
13801 // CHECK12-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP37:![0-9]+]]
13802 // CHECK12: omp.inner.for.end:
13803 // CHECK12-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
13804 // CHECK12: omp.loop.exit:
13805 // CHECK12-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP3]])
13806 // CHECK12-NEXT: [[TMP11:%.*]] = load i32, i32* [[DOTOMP_IS_LAST]], align 4
13807 // CHECK12-NEXT: [[TMP12:%.*]] = icmp ne i32 [[TMP11]], 0
13808 // CHECK12-NEXT: br i1 [[TMP12]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
13809 // CHECK12: .omp.final.then:
13810 // CHECK12-NEXT: store i32 100, i32* [[I]], align 4
13811 // CHECK12-NEXT: br label [[DOTOMP_FINAL_DONE]]
13812 // CHECK12: .omp.final.done:
13813 // CHECK12-NEXT: ret void
13814 //
13815 //
13816 // CHECK12-LABEL: define {{[^@]+}}@.omp_offloading.requires_reg
13817 // CHECK12-SAME: () #[[ATTR5:[0-9]+]] {
13818 // CHECK12-NEXT: entry:
13819 // CHECK12-NEXT: call void @__tgt_register_requires(i64 1)
13820 // CHECK12-NEXT: ret void
13821 //
13822 //
13823 // CHECK13-LABEL: define {{[^@]+}}@_Z9gtid_testv
13824 // CHECK13-SAME: () #[[ATTR0:[0-9]+]] {
13825 // CHECK13-NEXT: entry:
13826 // CHECK13-NEXT: [[TMP:%.*]] = alloca i32, align 4
13827 // CHECK13-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4
13828 // CHECK13-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4
13829 // CHECK13-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
13830 // CHECK13-NEXT: [[I:%.*]] = alloca i32, align 4
13831 // CHECK13-NEXT: [[_TMP2:%.*]] = alloca i32, align 4
13832 // CHECK13-NEXT: [[DOTOMP_LB3:%.*]] = alloca i32, align 4
13833 // CHECK13-NEXT: [[DOTOMP_UB4:%.*]] = alloca i32, align 4
13834 // CHECK13-NEXT: [[DOTOMP_IV5:%.*]] = alloca i32, align 4
13835 // CHECK13-NEXT: [[I6:%.*]] = alloca i32, align 4
13836 // CHECK13-NEXT: store i32 0, i32* [[DOTOMP_LB]], align 4
13837 // CHECK13-NEXT: store i32 99, i32* [[DOTOMP_UB]], align 4
13838 // CHECK13-NEXT: [[TMP0:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4
13839 // CHECK13-NEXT: store i32 [[TMP0]], i32* [[DOTOMP_IV]], align 4
13840 // CHECK13-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
13841 // CHECK13: omp.inner.for.cond:
13842 // CHECK13-NEXT: [[TMP1:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !6
13843 // CHECK13-NEXT: [[TMP2:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4, !llvm.access.group !6
13844 // CHECK13-NEXT: [[CMP:%.*]] = icmp sle i32 [[TMP1]], [[TMP2]]
13845 // CHECK13-NEXT: br i1 [[CMP]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
13846 // CHECK13: omp.inner.for.body:
13847 // CHECK13-NEXT: [[TMP3:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !6
13848 // CHECK13-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP3]], 1
13849 // CHECK13-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
13850 // CHECK13-NEXT: store i32 [[ADD]], i32* [[I]], align 4, !llvm.access.group !6
13851 // CHECK13-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
13852 // CHECK13: omp.body.continue:
13853 // CHECK13-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
13854 // CHECK13: omp.inner.for.inc:
13855 // CHECK13-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !6
13856 // CHECK13-NEXT: [[ADD1:%.*]] = add nsw i32 [[TMP4]], 1
13857 // CHECK13-NEXT: store i32 [[ADD1]], i32* [[DOTOMP_IV]], align 4, !llvm.access.group !6
13858 // CHECK13-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP7:![0-9]+]]
13859 // CHECK13: omp.inner.for.end:
13860 // CHECK13-NEXT: store i32 100, i32* [[I]], align 4
13861 // CHECK13-NEXT: store i32 0, i32* [[DOTOMP_LB3]], align 4
13862 // CHECK13-NEXT: store i32 99, i32* [[DOTOMP_UB4]], align 4
13863 // CHECK13-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_LB3]], align 4
13864 // CHECK13-NEXT: store i32 [[TMP5]], i32* [[DOTOMP_IV5]], align 4
13865 // CHECK13-NEXT: br label [[OMP_INNER_FOR_COND7:%.*]]
13866 // CHECK13: omp.inner.for.cond7:
13867 // CHECK13-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_IV5]], align 4, !llvm.access.group !10
13868 // CHECK13-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_UB4]], align 4, !llvm.access.group !10
13869 // CHECK13-NEXT: [[CMP8:%.*]] = icmp sle i32 [[TMP6]], [[TMP7]]
13870 // CHECK13-NEXT: br i1 [[CMP8]], label [[OMP_INNER_FOR_BODY9:%.*]], label [[OMP_INNER_FOR_END15:%.*]]
13871 // CHECK13: omp.inner.for.body9:
13872 // CHECK13-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_IV5]], align 4, !llvm.access.group !10
13873 // CHECK13-NEXT: [[MUL10:%.*]] = mul nsw i32 [[TMP8]], 1
13874 // CHECK13-NEXT: [[ADD11:%.*]] = add nsw i32 0, [[MUL10]]
13875 // CHECK13-NEXT: store i32 [[ADD11]], i32* [[I6]], align 4, !llvm.access.group !10
13876 // CHECK13-NEXT: call void @_Z9gtid_testv(), !llvm.access.group !10
13877 // CHECK13-NEXT: br label [[OMP_BODY_CONTINUE12:%.*]]
13878 // CHECK13: omp.body.continue12:
13879 // CHECK13-NEXT: br label [[OMP_INNER_FOR_INC13:%.*]]
13880 // CHECK13: omp.inner.for.inc13:
13881 // CHECK13-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_IV5]], align 4, !llvm.access.group !10
13882 // CHECK13-NEXT: [[ADD14:%.*]] = add nsw i32 [[TMP9]], 1
13883 // CHECK13-NEXT: store i32 [[ADD14]], i32* [[DOTOMP_IV5]], align 4, !llvm.access.group !10
13884 // CHECK13-NEXT: br label [[OMP_INNER_FOR_COND7]], !llvm.loop [[LOOP11:![0-9]+]]
13885 // CHECK13: omp.inner.for.end15:
13886 // CHECK13-NEXT: store i32 100, i32* [[I6]], align 4
13887 // CHECK13-NEXT: ret void
13888 //
13889 //
13890 // CHECK13-LABEL: define {{[^@]+}}@main
13891 // CHECK13-SAME: () #[[ATTR1:[0-9]+]] {
13892 // CHECK13-NEXT: entry:
13893 // CHECK13-NEXT: [[RETVAL:%.*]] = alloca i32, align 4
13894 // CHECK13-NEXT: [[TMP:%.*]] = alloca i32, align 4
13895 // CHECK13-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4
13896 // CHECK13-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4
13897 // CHECK13-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
13898 // CHECK13-NEXT: [[I:%.*]] = alloca i32, align 4
13899 // CHECK13-NEXT: [[_TMP2:%.*]] = alloca i32, align 4
13900 // CHECK13-NEXT: [[DOTOMP_LB3:%.*]] = alloca i32, align 4
13901 // CHECK13-NEXT: [[DOTOMP_UB4:%.*]] = alloca i32, align 4
13902 // CHECK13-NEXT: [[DOTOMP_IV5:%.*]] = alloca i32, align 4
13903 // CHECK13-NEXT: [[I6:%.*]] = alloca i32, align 4
13904 // CHECK13-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i8, align 1
13905 // CHECK13-NEXT: [[_TMP16:%.*]] = alloca i32, align 4
13906 // CHECK13-NEXT: [[DOTOMP_LB17:%.*]] = alloca i32, align 4
13907 // CHECK13-NEXT: [[DOTOMP_UB18:%.*]] = alloca i32, align 4
13908 // CHECK13-NEXT: [[DOTOMP_IV19:%.*]] = alloca i32, align 4
13909 // CHECK13-NEXT: [[I20:%.*]] = alloca i32, align 4
13910 // CHECK13-NEXT: store i32 0, i32* [[RETVAL]], align 4
13911 // CHECK13-NEXT: store i32 0, i32* [[DOTOMP_LB]], align 4
13912 // CHECK13-NEXT: store i32 99, i32* [[DOTOMP_UB]], align 4
13913 // CHECK13-NEXT: [[TMP0:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4
13914 // CHECK13-NEXT: store i32 [[TMP0]], i32* [[DOTOMP_IV]], align 4
13915 // CHECK13-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
13916 // CHECK13: omp.inner.for.cond:
13917 // CHECK13-NEXT: [[TMP1:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !13
13918 // CHECK13-NEXT: [[TMP2:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4, !llvm.access.group !13
13919 // CHECK13-NEXT: [[CMP:%.*]] = icmp sle i32 [[TMP1]], [[TMP2]]
13920 // CHECK13-NEXT: br i1 [[CMP]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
13921 // CHECK13: omp.inner.for.body:
13922 // CHECK13-NEXT: [[TMP3:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !13
13923 // CHECK13-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP3]], 1
13924 // CHECK13-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
13925 // CHECK13-NEXT: store i32 [[ADD]], i32* [[I]], align 4, !llvm.access.group !13
13926 // CHECK13-NEXT: call void @_Z3fn4v(), !llvm.access.group !13
13927 // CHECK13-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
13928 // CHECK13: omp.body.continue:
13929 // CHECK13-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
13930 // CHECK13: omp.inner.for.inc:
13931 // CHECK13-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !13
13932 // CHECK13-NEXT: [[ADD1:%.*]] = add nsw i32 [[TMP4]], 1
13933 // CHECK13-NEXT: store i32 [[ADD1]], i32* [[DOTOMP_IV]], align 4, !llvm.access.group !13
13934 // CHECK13-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP14:![0-9]+]]
13935 // CHECK13: omp.inner.for.end:
13936 // CHECK13-NEXT: store i32 100, i32* [[I]], align 4
13937 // CHECK13-NEXT: store i32 0, i32* [[DOTOMP_LB3]], align 4
13938 // CHECK13-NEXT: store i32 99, i32* [[DOTOMP_UB4]], align 4
13939 // CHECK13-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_LB3]], align 4
13940 // CHECK13-NEXT: store i32 [[TMP5]], i32* [[DOTOMP_IV5]], align 4
13941 // CHECK13-NEXT: br label [[OMP_INNER_FOR_COND7:%.*]]
13942 // CHECK13: omp.inner.for.cond7:
13943 // CHECK13-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_IV5]], align 4, !llvm.access.group !16
13944 // CHECK13-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_UB4]], align 4, !llvm.access.group !16
13945 // CHECK13-NEXT: [[CMP8:%.*]] = icmp sle i32 [[TMP6]], [[TMP7]]
13946 // CHECK13-NEXT: br i1 [[CMP8]], label [[OMP_INNER_FOR_BODY9:%.*]], label [[OMP_INNER_FOR_END15:%.*]]
13947 // CHECK13: omp.inner.for.body9:
13948 // CHECK13-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_IV5]], align 4, !llvm.access.group !16
13949 // CHECK13-NEXT: [[MUL10:%.*]] = mul nsw i32 [[TMP8]], 1
13950 // CHECK13-NEXT: [[ADD11:%.*]] = add nsw i32 0, [[MUL10]]
13951 // CHECK13-NEXT: store i32 [[ADD11]], i32* [[I6]], align 4, !llvm.access.group !16
13952 // CHECK13-NEXT: call void @_Z3fn5v(), !llvm.access.group !16
13953 // CHECK13-NEXT: br label [[OMP_BODY_CONTINUE12:%.*]]
13954 // CHECK13: omp.body.continue12:
13955 // CHECK13-NEXT: br label [[OMP_INNER_FOR_INC13:%.*]]
13956 // CHECK13: omp.inner.for.inc13:
13957 // CHECK13-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_IV5]], align 4, !llvm.access.group !16
13958 // CHECK13-NEXT: [[ADD14:%.*]] = add nsw i32 [[TMP9]], 1
13959 // CHECK13-NEXT: store i32 [[ADD14]], i32* [[DOTOMP_IV5]], align 4, !llvm.access.group !16
13960 // CHECK13-NEXT: br label [[OMP_INNER_FOR_COND7]], !llvm.loop [[LOOP17:![0-9]+]]
13961 // CHECK13: omp.inner.for.end15:
13962 // CHECK13-NEXT: store i32 100, i32* [[I6]], align 4
13963 // CHECK13-NEXT: [[TMP10:%.*]] = load i32, i32* @Arg, align 4
13964 // CHECK13-NEXT: [[TOBOOL:%.*]] = icmp ne i32 [[TMP10]], 0
13965 // CHECK13-NEXT: [[FROMBOOL:%.*]] = zext i1 [[TOBOOL]] to i8
13966 // CHECK13-NEXT: store i8 [[FROMBOOL]], i8* [[DOTCAPTURE_EXPR_]], align 1
13967 // CHECK13-NEXT: store i32 0, i32* [[DOTOMP_LB17]], align 4
13968 // CHECK13-NEXT: store i32 99, i32* [[DOTOMP_UB18]], align 4
13969 // CHECK13-NEXT: [[TMP11:%.*]] = load i32, i32* [[DOTOMP_LB17]], align 4
13970 // CHECK13-NEXT: store i32 [[TMP11]], i32* [[DOTOMP_IV19]], align 4
13971 // CHECK13-NEXT: br label [[OMP_INNER_FOR_COND21:%.*]]
13972 // CHECK13: omp.inner.for.cond21:
13973 // CHECK13-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTOMP_IV19]], align 4, !llvm.access.group !19
13974 // CHECK13-NEXT: [[TMP13:%.*]] = load i32, i32* [[DOTOMP_UB18]], align 4, !llvm.access.group !19
13975 // CHECK13-NEXT: [[CMP22:%.*]] = icmp sle i32 [[TMP12]], [[TMP13]]
13976 // CHECK13-NEXT: br i1 [[CMP22]], label [[OMP_INNER_FOR_BODY23:%.*]], label [[OMP_INNER_FOR_END29:%.*]]
13977 // CHECK13: omp.inner.for.body23:
13978 // CHECK13-NEXT: [[TMP14:%.*]] = load i32, i32* [[DOTOMP_IV19]], align 4, !llvm.access.group !19
13979 // CHECK13-NEXT: [[MUL24:%.*]] = mul nsw i32 [[TMP14]], 1
13980 // CHECK13-NEXT: [[ADD25:%.*]] = add nsw i32 0, [[MUL24]]
13981 // CHECK13-NEXT: store i32 [[ADD25]], i32* [[I20]], align 4, !llvm.access.group !19
13982 // CHECK13-NEXT: call void @_Z3fn6v(), !llvm.access.group !19
13983 // CHECK13-NEXT: br label [[OMP_BODY_CONTINUE26:%.*]]
13984 // CHECK13: omp.body.continue26:
13985 // CHECK13-NEXT: br label [[OMP_INNER_FOR_INC27:%.*]]
13986 // CHECK13: omp.inner.for.inc27:
13987 // CHECK13-NEXT: [[TMP15:%.*]] = load i32, i32* [[DOTOMP_IV19]], align 4, !llvm.access.group !19
13988 // CHECK13-NEXT: [[ADD28:%.*]] = add nsw i32 [[TMP15]], 1
13989 // CHECK13-NEXT: store i32 [[ADD28]], i32* [[DOTOMP_IV19]], align 4, !llvm.access.group !19
13990 // CHECK13-NEXT: br label [[OMP_INNER_FOR_COND21]], !llvm.loop [[LOOP20:![0-9]+]]
13991 // CHECK13: omp.inner.for.end29:
13992 // CHECK13-NEXT: store i32 100, i32* [[I20]], align 4
13993 // CHECK13-NEXT: [[TMP16:%.*]] = load i32, i32* @Arg, align 4
13994 // CHECK13-NEXT: [[CALL:%.*]] = call i32 @_Z5tmainIiEiT_(i32 [[TMP16]])
13995 // CHECK13-NEXT: ret i32 [[CALL]]
13996 //
13997 //
13998 // CHECK13-LABEL: define {{[^@]+}}@_Z5tmainIiEiT_
13999 // CHECK13-SAME: (i32 [[ARG:%.*]]) #[[ATTR0]] comdat {
14000 // CHECK13-NEXT: entry:
14001 // CHECK13-NEXT: [[ARG_ADDR:%.*]] = alloca i32, align 4
14002 // CHECK13-NEXT: [[TMP:%.*]] = alloca i32, align 4
14003 // CHECK13-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4
14004 // CHECK13-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4
14005 // CHECK13-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
14006 // CHECK13-NEXT: [[I:%.*]] = alloca i32, align 4
14007 // CHECK13-NEXT: [[_TMP2:%.*]] = alloca i32, align 4
14008 // CHECK13-NEXT: [[DOTOMP_LB3:%.*]] = alloca i32, align 4
14009 // CHECK13-NEXT: [[DOTOMP_UB4:%.*]] = alloca i32, align 4
14010 // CHECK13-NEXT: [[DOTOMP_IV5:%.*]] = alloca i32, align 4
14011 // CHECK13-NEXT: [[I6:%.*]] = alloca i32, align 4
14012 // CHECK13-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i8, align 1
14013 // CHECK13-NEXT: [[_TMP16:%.*]] = alloca i32, align 4
14014 // CHECK13-NEXT: [[DOTOMP_LB17:%.*]] = alloca i32, align 4
14015 // CHECK13-NEXT: [[DOTOMP_UB18:%.*]] = alloca i32, align 4
14016 // CHECK13-NEXT: [[DOTOMP_IV19:%.*]] = alloca i32, align 4
14017 // CHECK13-NEXT: [[I20:%.*]] = alloca i32, align 4
14018 // CHECK13-NEXT: store i32 [[ARG]], i32* [[ARG_ADDR]], align 4
14019 // CHECK13-NEXT: store i32 0, i32* [[DOTOMP_LB]], align 4
14020 // CHECK13-NEXT: store i32 99, i32* [[DOTOMP_UB]], align 4
14021 // CHECK13-NEXT: [[TMP0:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4
14022 // CHECK13-NEXT: store i32 [[TMP0]], i32* [[DOTOMP_IV]], align 4
14023 // CHECK13-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
14024 // CHECK13: omp.inner.for.cond:
14025 // CHECK13-NEXT: [[TMP1:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !22
14026 // CHECK13-NEXT: [[TMP2:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4, !llvm.access.group !22
14027 // CHECK13-NEXT: [[CMP:%.*]] = icmp sle i32 [[TMP1]], [[TMP2]]
14028 // CHECK13-NEXT: br i1 [[CMP]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
14029 // CHECK13: omp.inner.for.body:
14030 // CHECK13-NEXT: [[TMP3:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !22
14031 // CHECK13-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP3]], 1
14032 // CHECK13-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
14033 // CHECK13-NEXT: store i32 [[ADD]], i32* [[I]], align 4, !llvm.access.group !22
14034 // CHECK13-NEXT: call void @_Z3fn1v(), !llvm.access.group !22
14035 // CHECK13-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
14036 // CHECK13: omp.body.continue:
14037 // CHECK13-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
14038 // CHECK13: omp.inner.for.inc:
14039 // CHECK13-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !22
14040 // CHECK13-NEXT: [[ADD1:%.*]] = add nsw i32 [[TMP4]], 1
14041 // CHECK13-NEXT: store i32 [[ADD1]], i32* [[DOTOMP_IV]], align 4, !llvm.access.group !22
14042 // CHECK13-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP23:![0-9]+]]
14043 // CHECK13: omp.inner.for.end:
14044 // CHECK13-NEXT: store i32 100, i32* [[I]], align 4
14045 // CHECK13-NEXT: store i32 0, i32* [[DOTOMP_LB3]], align 4
14046 // CHECK13-NEXT: store i32 99, i32* [[DOTOMP_UB4]], align 4
14047 // CHECK13-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_LB3]], align 4
14048 // CHECK13-NEXT: store i32 [[TMP5]], i32* [[DOTOMP_IV5]], align 4
14049 // CHECK13-NEXT: br label [[OMP_INNER_FOR_COND7:%.*]]
14050 // CHECK13: omp.inner.for.cond7:
14051 // CHECK13-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_IV5]], align 4, !llvm.access.group !25
14052 // CHECK13-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_UB4]], align 4, !llvm.access.group !25
14053 // CHECK13-NEXT: [[CMP8:%.*]] = icmp sle i32 [[TMP6]], [[TMP7]]
14054 // CHECK13-NEXT: br i1 [[CMP8]], label [[OMP_INNER_FOR_BODY9:%.*]], label [[OMP_INNER_FOR_END15:%.*]]
14055 // CHECK13: omp.inner.for.body9:
14056 // CHECK13-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_IV5]], align 4, !llvm.access.group !25
14057 // CHECK13-NEXT: [[MUL10:%.*]] = mul nsw i32 [[TMP8]], 1
14058 // CHECK13-NEXT: [[ADD11:%.*]] = add nsw i32 0, [[MUL10]]
14059 // CHECK13-NEXT: store i32 [[ADD11]], i32* [[I6]], align 4, !llvm.access.group !25
14060 // CHECK13-NEXT: call void @_Z3fn2v(), !llvm.access.group !25
14061 // CHECK13-NEXT: br label [[OMP_BODY_CONTINUE12:%.*]]
14062 // CHECK13: omp.body.continue12:
14063 // CHECK13-NEXT: br label [[OMP_INNER_FOR_INC13:%.*]]
14064 // CHECK13: omp.inner.for.inc13:
14065 // CHECK13-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_IV5]], align 4, !llvm.access.group !25
14066 // CHECK13-NEXT: [[ADD14:%.*]] = add nsw i32 [[TMP9]], 1
14067 // CHECK13-NEXT: store i32 [[ADD14]], i32* [[DOTOMP_IV5]], align 4, !llvm.access.group !25
14068 // CHECK13-NEXT: br label [[OMP_INNER_FOR_COND7]], !llvm.loop [[LOOP26:![0-9]+]]
14069 // CHECK13: omp.inner.for.end15:
14070 // CHECK13-NEXT: store i32 100, i32* [[I6]], align 4
14071 // CHECK13-NEXT: [[TMP10:%.*]] = load i32, i32* [[ARG_ADDR]], align 4
14072 // CHECK13-NEXT: [[TOBOOL:%.*]] = icmp ne i32 [[TMP10]], 0
14073 // CHECK13-NEXT: [[FROMBOOL:%.*]] = zext i1 [[TOBOOL]] to i8
14074 // CHECK13-NEXT: store i8 [[FROMBOOL]], i8* [[DOTCAPTURE_EXPR_]], align 1
14075 // CHECK13-NEXT: store i32 0, i32* [[DOTOMP_LB17]], align 4
14076 // CHECK13-NEXT: store i32 99, i32* [[DOTOMP_UB18]], align 4
14077 // CHECK13-NEXT: [[TMP11:%.*]] = load i32, i32* [[DOTOMP_LB17]], align 4
14078 // CHECK13-NEXT: store i32 [[TMP11]], i32* [[DOTOMP_IV19]], align 4
14079 // CHECK13-NEXT: br label [[OMP_INNER_FOR_COND21:%.*]]
14080 // CHECK13: omp.inner.for.cond21:
14081 // CHECK13-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTOMP_IV19]], align 4, !llvm.access.group !28
14082 // CHECK13-NEXT: [[TMP13:%.*]] = load i32, i32* [[DOTOMP_UB18]], align 4, !llvm.access.group !28
14083 // CHECK13-NEXT: [[CMP22:%.*]] = icmp sle i32 [[TMP12]], [[TMP13]]
14084 // CHECK13-NEXT: br i1 [[CMP22]], label [[OMP_INNER_FOR_BODY23:%.*]], label [[OMP_INNER_FOR_END29:%.*]]
14085 // CHECK13: omp.inner.for.body23:
14086 // CHECK13-NEXT: [[TMP14:%.*]] = load i32, i32* [[DOTOMP_IV19]], align 4, !llvm.access.group !28
14087 // CHECK13-NEXT: [[MUL24:%.*]] = mul nsw i32 [[TMP14]], 1
14088 // CHECK13-NEXT: [[ADD25:%.*]] = add nsw i32 0, [[MUL24]]
14089 // CHECK13-NEXT: store i32 [[ADD25]], i32* [[I20]], align 4, !llvm.access.group !28
14090 // CHECK13-NEXT: call void @_Z3fn3v(), !llvm.access.group !28
14091 // CHECK13-NEXT: br label [[OMP_BODY_CONTINUE26:%.*]]
14092 // CHECK13: omp.body.continue26:
14093 // CHECK13-NEXT: br label [[OMP_INNER_FOR_INC27:%.*]]
14094 // CHECK13: omp.inner.for.inc27:
14095 // CHECK13-NEXT: [[TMP15:%.*]] = load i32, i32* [[DOTOMP_IV19]], align 4, !llvm.access.group !28
14096 // CHECK13-NEXT: [[ADD28:%.*]] = add nsw i32 [[TMP15]], 1
14097 // CHECK13-NEXT: store i32 [[ADD28]], i32* [[DOTOMP_IV19]], align 4, !llvm.access.group !28
14098 // CHECK13-NEXT: br label [[OMP_INNER_FOR_COND21]], !llvm.loop [[LOOP29:![0-9]+]]
14099 // CHECK13: omp.inner.for.end29:
14100 // CHECK13-NEXT: store i32 100, i32* [[I20]], align 4
14101 // CHECK13-NEXT: ret i32 0
14102 //
14103 //
14104 // CHECK14-LABEL: define {{[^@]+}}@_Z9gtid_testv
14105 // CHECK14-SAME: () #[[ATTR0:[0-9]+]] {
14106 // CHECK14-NEXT: entry:
14107 // CHECK14-NEXT: [[TMP:%.*]] = alloca i32, align 4
14108 // CHECK14-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4
14109 // CHECK14-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4
14110 // CHECK14-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
14111 // CHECK14-NEXT: [[I:%.*]] = alloca i32, align 4
14112 // CHECK14-NEXT: [[_TMP2:%.*]] = alloca i32, align 4
14113 // CHECK14-NEXT: [[DOTOMP_LB3:%.*]] = alloca i32, align 4
14114 // CHECK14-NEXT: [[DOTOMP_UB4:%.*]] = alloca i32, align 4
14115 // CHECK14-NEXT: [[DOTOMP_IV5:%.*]] = alloca i32, align 4
14116 // CHECK14-NEXT: [[I6:%.*]] = alloca i32, align 4
14117 // CHECK14-NEXT: store i32 0, i32* [[DOTOMP_LB]], align 4
14118 // CHECK14-NEXT: store i32 99, i32* [[DOTOMP_UB]], align 4
14119 // CHECK14-NEXT: [[TMP0:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4
14120 // CHECK14-NEXT: store i32 [[TMP0]], i32* [[DOTOMP_IV]], align 4
14121 // CHECK14-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
14122 // CHECK14: omp.inner.for.cond:
14123 // CHECK14-NEXT: [[TMP1:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !6
14124 // CHECK14-NEXT: [[TMP2:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4, !llvm.access.group !6
14125 // CHECK14-NEXT: [[CMP:%.*]] = icmp sle i32 [[TMP1]], [[TMP2]]
14126 // CHECK14-NEXT: br i1 [[CMP]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
14127 // CHECK14: omp.inner.for.body:
14128 // CHECK14-NEXT: [[TMP3:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !6
14129 // CHECK14-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP3]], 1
14130 // CHECK14-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
14131 // CHECK14-NEXT: store i32 [[ADD]], i32* [[I]], align 4, !llvm.access.group !6
14132 // CHECK14-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
14133 // CHECK14: omp.body.continue:
14134 // CHECK14-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
14135 // CHECK14: omp.inner.for.inc:
14136 // CHECK14-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !6
14137 // CHECK14-NEXT: [[ADD1:%.*]] = add nsw i32 [[TMP4]], 1
14138 // CHECK14-NEXT: store i32 [[ADD1]], i32* [[DOTOMP_IV]], align 4, !llvm.access.group !6
14139 // CHECK14-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP7:![0-9]+]]
14140 // CHECK14: omp.inner.for.end:
14141 // CHECK14-NEXT: store i32 100, i32* [[I]], align 4
14142 // CHECK14-NEXT: store i32 0, i32* [[DOTOMP_LB3]], align 4
14143 // CHECK14-NEXT: store i32 99, i32* [[DOTOMP_UB4]], align 4
14144 // CHECK14-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_LB3]], align 4
14145 // CHECK14-NEXT: store i32 [[TMP5]], i32* [[DOTOMP_IV5]], align 4
14146 // CHECK14-NEXT: br label [[OMP_INNER_FOR_COND7:%.*]]
14147 // CHECK14: omp.inner.for.cond7:
14148 // CHECK14-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_IV5]], align 4, !llvm.access.group !10
14149 // CHECK14-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_UB4]], align 4, !llvm.access.group !10
14150 // CHECK14-NEXT: [[CMP8:%.*]] = icmp sle i32 [[TMP6]], [[TMP7]]
14151 // CHECK14-NEXT: br i1 [[CMP8]], label [[OMP_INNER_FOR_BODY9:%.*]], label [[OMP_INNER_FOR_END15:%.*]]
14152 // CHECK14: omp.inner.for.body9:
14153 // CHECK14-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_IV5]], align 4, !llvm.access.group !10
14154 // CHECK14-NEXT: [[MUL10:%.*]] = mul nsw i32 [[TMP8]], 1
14155 // CHECK14-NEXT: [[ADD11:%.*]] = add nsw i32 0, [[MUL10]]
14156 // CHECK14-NEXT: store i32 [[ADD11]], i32* [[I6]], align 4, !llvm.access.group !10
14157 // CHECK14-NEXT: call void @_Z9gtid_testv(), !llvm.access.group !10
14158 // CHECK14-NEXT: br label [[OMP_BODY_CONTINUE12:%.*]]
14159 // CHECK14: omp.body.continue12:
14160 // CHECK14-NEXT: br label [[OMP_INNER_FOR_INC13:%.*]]
14161 // CHECK14: omp.inner.for.inc13:
14162 // CHECK14-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_IV5]], align 4, !llvm.access.group !10
14163 // CHECK14-NEXT: [[ADD14:%.*]] = add nsw i32 [[TMP9]], 1
14164 // CHECK14-NEXT: store i32 [[ADD14]], i32* [[DOTOMP_IV5]], align 4, !llvm.access.group !10
14165 // CHECK14-NEXT: br label [[OMP_INNER_FOR_COND7]], !llvm.loop [[LOOP11:![0-9]+]]
14166 // CHECK14: omp.inner.for.end15:
14167 // CHECK14-NEXT: store i32 100, i32* [[I6]], align 4
14168 // CHECK14-NEXT: ret void
14169 //
14170 //
14171 // CHECK14-LABEL: define {{[^@]+}}@main
14172 // CHECK14-SAME: () #[[ATTR1:[0-9]+]] {
14173 // CHECK14-NEXT: entry:
14174 // CHECK14-NEXT: [[RETVAL:%.*]] = alloca i32, align 4
14175 // CHECK14-NEXT: [[TMP:%.*]] = alloca i32, align 4
14176 // CHECK14-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4
14177 // CHECK14-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4
14178 // CHECK14-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
14179 // CHECK14-NEXT: [[I:%.*]] = alloca i32, align 4
14180 // CHECK14-NEXT: [[_TMP2:%.*]] = alloca i32, align 4
14181 // CHECK14-NEXT: [[DOTOMP_LB3:%.*]] = alloca i32, align 4
14182 // CHECK14-NEXT: [[DOTOMP_UB4:%.*]] = alloca i32, align 4
14183 // CHECK14-NEXT: [[DOTOMP_IV5:%.*]] = alloca i32, align 4
14184 // CHECK14-NEXT: [[I6:%.*]] = alloca i32, align 4
14185 // CHECK14-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i8, align 1
14186 // CHECK14-NEXT: [[_TMP16:%.*]] = alloca i32, align 4
14187 // CHECK14-NEXT: [[DOTOMP_LB17:%.*]] = alloca i32, align 4
14188 // CHECK14-NEXT: [[DOTOMP_UB18:%.*]] = alloca i32, align 4
14189 // CHECK14-NEXT: [[DOTOMP_IV19:%.*]] = alloca i32, align 4
14190 // CHECK14-NEXT: [[I20:%.*]] = alloca i32, align 4
14191 // CHECK14-NEXT: store i32 0, i32* [[RETVAL]], align 4
14192 // CHECK14-NEXT: store i32 0, i32* [[DOTOMP_LB]], align 4
14193 // CHECK14-NEXT: store i32 99, i32* [[DOTOMP_UB]], align 4
14194 // CHECK14-NEXT: [[TMP0:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4
14195 // CHECK14-NEXT: store i32 [[TMP0]], i32* [[DOTOMP_IV]], align 4
14196 // CHECK14-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
14197 // CHECK14: omp.inner.for.cond:
14198 // CHECK14-NEXT: [[TMP1:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !13
14199 // CHECK14-NEXT: [[TMP2:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4, !llvm.access.group !13
14200 // CHECK14-NEXT: [[CMP:%.*]] = icmp sle i32 [[TMP1]], [[TMP2]]
14201 // CHECK14-NEXT: br i1 [[CMP]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
14202 // CHECK14: omp.inner.for.body:
14203 // CHECK14-NEXT: [[TMP3:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !13
14204 // CHECK14-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP3]], 1
14205 // CHECK14-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
14206 // CHECK14-NEXT: store i32 [[ADD]], i32* [[I]], align 4, !llvm.access.group !13
14207 // CHECK14-NEXT: call void @_Z3fn4v(), !llvm.access.group !13
14208 // CHECK14-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
14209 // CHECK14: omp.body.continue:
14210 // CHECK14-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
14211 // CHECK14: omp.inner.for.inc:
14212 // CHECK14-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !13
14213 // CHECK14-NEXT: [[ADD1:%.*]] = add nsw i32 [[TMP4]], 1
14214 // CHECK14-NEXT: store i32 [[ADD1]], i32* [[DOTOMP_IV]], align 4, !llvm.access.group !13
14215 // CHECK14-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP14:![0-9]+]]
14216 // CHECK14: omp.inner.for.end:
14217 // CHECK14-NEXT: store i32 100, i32* [[I]], align 4
14218 // CHECK14-NEXT: store i32 0, i32* [[DOTOMP_LB3]], align 4
14219 // CHECK14-NEXT: store i32 99, i32* [[DOTOMP_UB4]], align 4
14220 // CHECK14-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_LB3]], align 4
14221 // CHECK14-NEXT: store i32 [[TMP5]], i32* [[DOTOMP_IV5]], align 4
14222 // CHECK14-NEXT: br label [[OMP_INNER_FOR_COND7:%.*]]
14223 // CHECK14: omp.inner.for.cond7:
14224 // CHECK14-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_IV5]], align 4, !llvm.access.group !16
14225 // CHECK14-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_UB4]], align 4, !llvm.access.group !16
14226 // CHECK14-NEXT: [[CMP8:%.*]] = icmp sle i32 [[TMP6]], [[TMP7]]
14227 // CHECK14-NEXT: br i1 [[CMP8]], label [[OMP_INNER_FOR_BODY9:%.*]], label [[OMP_INNER_FOR_END15:%.*]]
14228 // CHECK14: omp.inner.for.body9:
14229 // CHECK14-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_IV5]], align 4, !llvm.access.group !16
14230 // CHECK14-NEXT: [[MUL10:%.*]] = mul nsw i32 [[TMP8]], 1
14231 // CHECK14-NEXT: [[ADD11:%.*]] = add nsw i32 0, [[MUL10]]
14232 // CHECK14-NEXT: store i32 [[ADD11]], i32* [[I6]], align 4, !llvm.access.group !16
14233 // CHECK14-NEXT: call void @_Z3fn5v(), !llvm.access.group !16
14234 // CHECK14-NEXT: br label [[OMP_BODY_CONTINUE12:%.*]]
14235 // CHECK14: omp.body.continue12:
14236 // CHECK14-NEXT: br label [[OMP_INNER_FOR_INC13:%.*]]
14237 // CHECK14: omp.inner.for.inc13:
14238 // CHECK14-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_IV5]], align 4, !llvm.access.group !16
14239 // CHECK14-NEXT: [[ADD14:%.*]] = add nsw i32 [[TMP9]], 1
14240 // CHECK14-NEXT: store i32 [[ADD14]], i32* [[DOTOMP_IV5]], align 4, !llvm.access.group !16
14241 // CHECK14-NEXT: br label [[OMP_INNER_FOR_COND7]], !llvm.loop [[LOOP17:![0-9]+]]
14242 // CHECK14: omp.inner.for.end15:
14243 // CHECK14-NEXT: store i32 100, i32* [[I6]], align 4
14244 // CHECK14-NEXT: [[TMP10:%.*]] = load i32, i32* @Arg, align 4
14245 // CHECK14-NEXT: [[TOBOOL:%.*]] = icmp ne i32 [[TMP10]], 0
14246 // CHECK14-NEXT: [[FROMBOOL:%.*]] = zext i1 [[TOBOOL]] to i8
14247 // CHECK14-NEXT: store i8 [[FROMBOOL]], i8* [[DOTCAPTURE_EXPR_]], align 1
14248 // CHECK14-NEXT: store i32 0, i32* [[DOTOMP_LB17]], align 4
14249 // CHECK14-NEXT: store i32 99, i32* [[DOTOMP_UB18]], align 4
14250 // CHECK14-NEXT: [[TMP11:%.*]] = load i32, i32* [[DOTOMP_LB17]], align 4
14251 // CHECK14-NEXT: store i32 [[TMP11]], i32* [[DOTOMP_IV19]], align 4
14252 // CHECK14-NEXT: br label [[OMP_INNER_FOR_COND21:%.*]]
14253 // CHECK14: omp.inner.for.cond21:
14254 // CHECK14-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTOMP_IV19]], align 4, !llvm.access.group !19
14255 // CHECK14-NEXT: [[TMP13:%.*]] = load i32, i32* [[DOTOMP_UB18]], align 4, !llvm.access.group !19
14256 // CHECK14-NEXT: [[CMP22:%.*]] = icmp sle i32 [[TMP12]], [[TMP13]]
14257 // CHECK14-NEXT: br i1 [[CMP22]], label [[OMP_INNER_FOR_BODY23:%.*]], label [[OMP_INNER_FOR_END29:%.*]]
14258 // CHECK14: omp.inner.for.body23:
14259 // CHECK14-NEXT: [[TMP14:%.*]] = load i32, i32* [[DOTOMP_IV19]], align 4, !llvm.access.group !19
14260 // CHECK14-NEXT: [[MUL24:%.*]] = mul nsw i32 [[TMP14]], 1
14261 // CHECK14-NEXT: [[ADD25:%.*]] = add nsw i32 0, [[MUL24]]
14262 // CHECK14-NEXT: store i32 [[ADD25]], i32* [[I20]], align 4, !llvm.access.group !19
14263 // CHECK14-NEXT: call void @_Z3fn6v(), !llvm.access.group !19
14264 // CHECK14-NEXT: br label [[OMP_BODY_CONTINUE26:%.*]]
14265 // CHECK14: omp.body.continue26:
14266 // CHECK14-NEXT: br label [[OMP_INNER_FOR_INC27:%.*]]
14267 // CHECK14: omp.inner.for.inc27:
14268 // CHECK14-NEXT: [[TMP15:%.*]] = load i32, i32* [[DOTOMP_IV19]], align 4, !llvm.access.group !19
14269 // CHECK14-NEXT: [[ADD28:%.*]] = add nsw i32 [[TMP15]], 1
14270 // CHECK14-NEXT: store i32 [[ADD28]], i32* [[DOTOMP_IV19]], align 4, !llvm.access.group !19
14271 // CHECK14-NEXT: br label [[OMP_INNER_FOR_COND21]], !llvm.loop [[LOOP20:![0-9]+]]
14272 // CHECK14: omp.inner.for.end29:
14273 // CHECK14-NEXT: store i32 100, i32* [[I20]], align 4
14274 // CHECK14-NEXT: [[TMP16:%.*]] = load i32, i32* @Arg, align 4
14275 // CHECK14-NEXT: [[CALL:%.*]] = call i32 @_Z5tmainIiEiT_(i32 [[TMP16]])
14276 // CHECK14-NEXT: ret i32 [[CALL]]
14277 //
14278 //
14279 // CHECK14-LABEL: define {{[^@]+}}@_Z5tmainIiEiT_
14280 // CHECK14-SAME: (i32 [[ARG:%.*]]) #[[ATTR0]] comdat {
14281 // CHECK14-NEXT: entry:
14282 // CHECK14-NEXT: [[ARG_ADDR:%.*]] = alloca i32, align 4
14283 // CHECK14-NEXT: [[TMP:%.*]] = alloca i32, align 4
14284 // CHECK14-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4
14285 // CHECK14-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4
14286 // CHECK14-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
14287 // CHECK14-NEXT: [[I:%.*]] = alloca i32, align 4
14288 // CHECK14-NEXT: [[_TMP2:%.*]] = alloca i32, align 4
14289 // CHECK14-NEXT: [[DOTOMP_LB3:%.*]] = alloca i32, align 4
14290 // CHECK14-NEXT: [[DOTOMP_UB4:%.*]] = alloca i32, align 4
14291 // CHECK14-NEXT: [[DOTOMP_IV5:%.*]] = alloca i32, align 4
14292 // CHECK14-NEXT: [[I6:%.*]] = alloca i32, align 4
14293 // CHECK14-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i8, align 1
14294 // CHECK14-NEXT: [[_TMP16:%.*]] = alloca i32, align 4
14295 // CHECK14-NEXT: [[DOTOMP_LB17:%.*]] = alloca i32, align 4
14296 // CHECK14-NEXT: [[DOTOMP_UB18:%.*]] = alloca i32, align 4
14297 // CHECK14-NEXT: [[DOTOMP_IV19:%.*]] = alloca i32, align 4
14298 // CHECK14-NEXT: [[I20:%.*]] = alloca i32, align 4
14299 // CHECK14-NEXT: store i32 [[ARG]], i32* [[ARG_ADDR]], align 4
14300 // CHECK14-NEXT: store i32 0, i32* [[DOTOMP_LB]], align 4
14301 // CHECK14-NEXT: store i32 99, i32* [[DOTOMP_UB]], align 4
14302 // CHECK14-NEXT: [[TMP0:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4
14303 // CHECK14-NEXT: store i32 [[TMP0]], i32* [[DOTOMP_IV]], align 4
14304 // CHECK14-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
14305 // CHECK14: omp.inner.for.cond:
14306 // CHECK14-NEXT: [[TMP1:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !22
14307 // CHECK14-NEXT: [[TMP2:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4, !llvm.access.group !22
14308 // CHECK14-NEXT: [[CMP:%.*]] = icmp sle i32 [[TMP1]], [[TMP2]]
14309 // CHECK14-NEXT: br i1 [[CMP]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
14310 // CHECK14: omp.inner.for.body:
14311 // CHECK14-NEXT: [[TMP3:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !22
14312 // CHECK14-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP3]], 1
14313 // CHECK14-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
14314 // CHECK14-NEXT: store i32 [[ADD]], i32* [[I]], align 4, !llvm.access.group !22
14315 // CHECK14-NEXT: call void @_Z3fn1v(), !llvm.access.group !22
14316 // CHECK14-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
14317 // CHECK14: omp.body.continue:
14318 // CHECK14-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
14319 // CHECK14: omp.inner.for.inc:
14320 // CHECK14-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !22
14321 // CHECK14-NEXT: [[ADD1:%.*]] = add nsw i32 [[TMP4]], 1
14322 // CHECK14-NEXT: store i32 [[ADD1]], i32* [[DOTOMP_IV]], align 4, !llvm.access.group !22
14323 // CHECK14-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP23:![0-9]+]]
14324 // CHECK14: omp.inner.for.end:
14325 // CHECK14-NEXT: store i32 100, i32* [[I]], align 4
14326 // CHECK14-NEXT: store i32 0, i32* [[DOTOMP_LB3]], align 4
14327 // CHECK14-NEXT: store i32 99, i32* [[DOTOMP_UB4]], align 4
14328 // CHECK14-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_LB3]], align 4
14329 // CHECK14-NEXT: store i32 [[TMP5]], i32* [[DOTOMP_IV5]], align 4
14330 // CHECK14-NEXT: br label [[OMP_INNER_FOR_COND7:%.*]]
14331 // CHECK14: omp.inner.for.cond7:
14332 // CHECK14-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_IV5]], align 4, !llvm.access.group !25
14333 // CHECK14-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_UB4]], align 4, !llvm.access.group !25
14334 // CHECK14-NEXT: [[CMP8:%.*]] = icmp sle i32 [[TMP6]], [[TMP7]]
14335 // CHECK14-NEXT: br i1 [[CMP8]], label [[OMP_INNER_FOR_BODY9:%.*]], label [[OMP_INNER_FOR_END15:%.*]]
14336 // CHECK14: omp.inner.for.body9:
14337 // CHECK14-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_IV5]], align 4, !llvm.access.group !25
14338 // CHECK14-NEXT: [[MUL10:%.*]] = mul nsw i32 [[TMP8]], 1
14339 // CHECK14-NEXT: [[ADD11:%.*]] = add nsw i32 0, [[MUL10]]
14340 // CHECK14-NEXT: store i32 [[ADD11]], i32* [[I6]], align 4, !llvm.access.group !25
14341 // CHECK14-NEXT: call void @_Z3fn2v(), !llvm.access.group !25
14342 // CHECK14-NEXT: br label [[OMP_BODY_CONTINUE12:%.*]]
14343 // CHECK14: omp.body.continue12:
14344 // CHECK14-NEXT: br label [[OMP_INNER_FOR_INC13:%.*]]
14345 // CHECK14: omp.inner.for.inc13:
14346 // CHECK14-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_IV5]], align 4, !llvm.access.group !25
14347 // CHECK14-NEXT: [[ADD14:%.*]] = add nsw i32 [[TMP9]], 1
14348 // CHECK14-NEXT: store i32 [[ADD14]], i32* [[DOTOMP_IV5]], align 4, !llvm.access.group !25
14349 // CHECK14-NEXT: br label [[OMP_INNER_FOR_COND7]], !llvm.loop [[LOOP26:![0-9]+]]
14350 // CHECK14: omp.inner.for.end15:
14351 // CHECK14-NEXT: store i32 100, i32* [[I6]], align 4
14352 // CHECK14-NEXT: [[TMP10:%.*]] = load i32, i32* [[ARG_ADDR]], align 4
14353 // CHECK14-NEXT: [[TOBOOL:%.*]] = icmp ne i32 [[TMP10]], 0
14354 // CHECK14-NEXT: [[FROMBOOL:%.*]] = zext i1 [[TOBOOL]] to i8
14355 // CHECK14-NEXT: store i8 [[FROMBOOL]], i8* [[DOTCAPTURE_EXPR_]], align 1
14356 // CHECK14-NEXT: store i32 0, i32* [[DOTOMP_LB17]], align 4
14357 // CHECK14-NEXT: store i32 99, i32* [[DOTOMP_UB18]], align 4
14358 // CHECK14-NEXT: [[TMP11:%.*]] = load i32, i32* [[DOTOMP_LB17]], align 4
14359 // CHECK14-NEXT: store i32 [[TMP11]], i32* [[DOTOMP_IV19]], align 4
14360 // CHECK14-NEXT: br label [[OMP_INNER_FOR_COND21:%.*]]
14361 // CHECK14: omp.inner.for.cond21:
14362 // CHECK14-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTOMP_IV19]], align 4, !llvm.access.group !28
14363 // CHECK14-NEXT: [[TMP13:%.*]] = load i32, i32* [[DOTOMP_UB18]], align 4, !llvm.access.group !28
14364 // CHECK14-NEXT: [[CMP22:%.*]] = icmp sle i32 [[TMP12]], [[TMP13]]
14365 // CHECK14-NEXT: br i1 [[CMP22]], label [[OMP_INNER_FOR_BODY23:%.*]], label [[OMP_INNER_FOR_END29:%.*]]
14366 // CHECK14: omp.inner.for.body23:
14367 // CHECK14-NEXT: [[TMP14:%.*]] = load i32, i32* [[DOTOMP_IV19]], align 4, !llvm.access.group !28
14368 // CHECK14-NEXT: [[MUL24:%.*]] = mul nsw i32 [[TMP14]], 1
14369 // CHECK14-NEXT: [[ADD25:%.*]] = add nsw i32 0, [[MUL24]]
14370 // CHECK14-NEXT: store i32 [[ADD25]], i32* [[I20]], align 4, !llvm.access.group !28
14371 // CHECK14-NEXT: call void @_Z3fn3v(), !llvm.access.group !28
14372 // CHECK14-NEXT: br label [[OMP_BODY_CONTINUE26:%.*]]
14373 // CHECK14: omp.body.continue26:
14374 // CHECK14-NEXT: br label [[OMP_INNER_FOR_INC27:%.*]]
14375 // CHECK14: omp.inner.for.inc27:
14376 // CHECK14-NEXT: [[TMP15:%.*]] = load i32, i32* [[DOTOMP_IV19]], align 4, !llvm.access.group !28
14377 // CHECK14-NEXT: [[ADD28:%.*]] = add nsw i32 [[TMP15]], 1
14378 // CHECK14-NEXT: store i32 [[ADD28]], i32* [[DOTOMP_IV19]], align 4, !llvm.access.group !28
14379 // CHECK14-NEXT: br label [[OMP_INNER_FOR_COND21]], !llvm.loop [[LOOP29:![0-9]+]]
14380 // CHECK14: omp.inner.for.end29:
14381 // CHECK14-NEXT: store i32 100, i32* [[I20]], align 4
14382 // CHECK14-NEXT: ret i32 0
14383 //
14384 //
14385 // CHECK15-LABEL: define {{[^@]+}}@_Z9gtid_testv
14386 // CHECK15-SAME: () #[[ATTR0:[0-9]+]] {
14387 // CHECK15-NEXT: entry:
14388 // CHECK15-NEXT: [[TMP:%.*]] = alloca i32, align 4
14389 // CHECK15-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4
14390 // CHECK15-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4
14391 // CHECK15-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
14392 // CHECK15-NEXT: [[I:%.*]] = alloca i32, align 4
14393 // CHECK15-NEXT: [[_TMP2:%.*]] = alloca i32, align 4
14394 // CHECK15-NEXT: [[DOTOMP_LB3:%.*]] = alloca i32, align 4
14395 // CHECK15-NEXT: [[DOTOMP_UB4:%.*]] = alloca i32, align 4
14396 // CHECK15-NEXT: [[DOTOMP_IV5:%.*]] = alloca i32, align 4
14397 // CHECK15-NEXT: [[I6:%.*]] = alloca i32, align 4
14398 // CHECK15-NEXT: store i32 0, i32* [[DOTOMP_LB]], align 4
14399 // CHECK15-NEXT: store i32 99, i32* [[DOTOMP_UB]], align 4
14400 // CHECK15-NEXT: [[TMP0:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4
14401 // CHECK15-NEXT: store i32 [[TMP0]], i32* [[DOTOMP_IV]], align 4
14402 // CHECK15-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
14403 // CHECK15: omp.inner.for.cond:
14404 // CHECK15-NEXT: [[TMP1:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !6
14405 // CHECK15-NEXT: [[TMP2:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4, !llvm.access.group !6
14406 // CHECK15-NEXT: [[CMP:%.*]] = icmp sle i32 [[TMP1]], [[TMP2]]
14407 // CHECK15-NEXT: br i1 [[CMP]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
14408 // CHECK15: omp.inner.for.body:
14409 // CHECK15-NEXT: [[TMP3:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !6
14410 // CHECK15-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP3]], 1
14411 // CHECK15-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
14412 // CHECK15-NEXT: store i32 [[ADD]], i32* [[I]], align 4, !llvm.access.group !6
14413 // CHECK15-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
14414 // CHECK15: omp.body.continue:
14415 // CHECK15-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
14416 // CHECK15: omp.inner.for.inc:
14417 // CHECK15-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !6
14418 // CHECK15-NEXT: [[ADD1:%.*]] = add nsw i32 [[TMP4]], 1
14419 // CHECK15-NEXT: store i32 [[ADD1]], i32* [[DOTOMP_IV]], align 4, !llvm.access.group !6
14420 // CHECK15-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP7:![0-9]+]]
14421 // CHECK15: omp.inner.for.end:
14422 // CHECK15-NEXT: store i32 100, i32* [[I]], align 4
14423 // CHECK15-NEXT: store i32 0, i32* [[DOTOMP_LB3]], align 4
14424 // CHECK15-NEXT: store i32 99, i32* [[DOTOMP_UB4]], align 4
14425 // CHECK15-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_LB3]], align 4
14426 // CHECK15-NEXT: store i32 [[TMP5]], i32* [[DOTOMP_IV5]], align 4
14427 // CHECK15-NEXT: br label [[OMP_INNER_FOR_COND7:%.*]]
14428 // CHECK15: omp.inner.for.cond7:
14429 // CHECK15-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_IV5]], align 4, !llvm.access.group !10
14430 // CHECK15-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_UB4]], align 4, !llvm.access.group !10
14431 // CHECK15-NEXT: [[CMP8:%.*]] = icmp sle i32 [[TMP6]], [[TMP7]]
14432 // CHECK15-NEXT: br i1 [[CMP8]], label [[OMP_INNER_FOR_BODY9:%.*]], label [[OMP_INNER_FOR_END15:%.*]]
14433 // CHECK15: omp.inner.for.body9:
14434 // CHECK15-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_IV5]], align 4, !llvm.access.group !10
14435 // CHECK15-NEXT: [[MUL10:%.*]] = mul nsw i32 [[TMP8]], 1
14436 // CHECK15-NEXT: [[ADD11:%.*]] = add nsw i32 0, [[MUL10]]
14437 // CHECK15-NEXT: store i32 [[ADD11]], i32* [[I6]], align 4, !llvm.access.group !10
14438 // CHECK15-NEXT: call void @_Z9gtid_testv(), !llvm.access.group !10
14439 // CHECK15-NEXT: br label [[OMP_BODY_CONTINUE12:%.*]]
14440 // CHECK15: omp.body.continue12:
14441 // CHECK15-NEXT: br label [[OMP_INNER_FOR_INC13:%.*]]
14442 // CHECK15: omp.inner.for.inc13:
14443 // CHECK15-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_IV5]], align 4, !llvm.access.group !10
14444 // CHECK15-NEXT: [[ADD14:%.*]] = add nsw i32 [[TMP9]], 1
14445 // CHECK15-NEXT: store i32 [[ADD14]], i32* [[DOTOMP_IV5]], align 4, !llvm.access.group !10
14446 // CHECK15-NEXT: br label [[OMP_INNER_FOR_COND7]], !llvm.loop [[LOOP11:![0-9]+]]
14447 // CHECK15: omp.inner.for.end15:
14448 // CHECK15-NEXT: store i32 100, i32* [[I6]], align 4
14449 // CHECK15-NEXT: ret void
14450 //
14451 //
14452 // CHECK15-LABEL: define {{[^@]+}}@main
14453 // CHECK15-SAME: () #[[ATTR1:[0-9]+]] {
14454 // CHECK15-NEXT: entry:
14455 // CHECK15-NEXT: [[RETVAL:%.*]] = alloca i32, align 4
14456 // CHECK15-NEXT: [[TMP:%.*]] = alloca i32, align 4
14457 // CHECK15-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4
14458 // CHECK15-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4
14459 // CHECK15-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
14460 // CHECK15-NEXT: [[I:%.*]] = alloca i32, align 4
14461 // CHECK15-NEXT: [[_TMP2:%.*]] = alloca i32, align 4
14462 // CHECK15-NEXT: [[DOTOMP_LB3:%.*]] = alloca i32, align 4
14463 // CHECK15-NEXT: [[DOTOMP_UB4:%.*]] = alloca i32, align 4
14464 // CHECK15-NEXT: [[DOTOMP_IV5:%.*]] = alloca i32, align 4
14465 // CHECK15-NEXT: [[I6:%.*]] = alloca i32, align 4
14466 // CHECK15-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i8, align 1
14467 // CHECK15-NEXT: [[_TMP16:%.*]] = alloca i32, align 4
14468 // CHECK15-NEXT: [[DOTOMP_LB17:%.*]] = alloca i32, align 4
14469 // CHECK15-NEXT: [[DOTOMP_UB18:%.*]] = alloca i32, align 4
14470 // CHECK15-NEXT: [[DOTOMP_IV19:%.*]] = alloca i32, align 4
14471 // CHECK15-NEXT: [[I20:%.*]] = alloca i32, align 4
14472 // CHECK15-NEXT: store i32 0, i32* [[RETVAL]], align 4
14473 // CHECK15-NEXT: store i32 0, i32* [[DOTOMP_LB]], align 4
14474 // CHECK15-NEXT: store i32 99, i32* [[DOTOMP_UB]], align 4
14475 // CHECK15-NEXT: [[TMP0:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4
14476 // CHECK15-NEXT: store i32 [[TMP0]], i32* [[DOTOMP_IV]], align 4
14477 // CHECK15-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
14478 // CHECK15: omp.inner.for.cond:
14479 // CHECK15-NEXT: [[TMP1:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !13
14480 // CHECK15-NEXT: [[TMP2:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4, !llvm.access.group !13
14481 // CHECK15-NEXT: [[CMP:%.*]] = icmp sle i32 [[TMP1]], [[TMP2]]
14482 // CHECK15-NEXT: br i1 [[CMP]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
14483 // CHECK15: omp.inner.for.body:
14484 // CHECK15-NEXT: [[TMP3:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !13
14485 // CHECK15-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP3]], 1
14486 // CHECK15-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
14487 // CHECK15-NEXT: store i32 [[ADD]], i32* [[I]], align 4, !llvm.access.group !13
14488 // CHECK15-NEXT: call void @_Z3fn4v(), !llvm.access.group !13
14489 // CHECK15-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
14490 // CHECK15: omp.body.continue:
14491 // CHECK15-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
14492 // CHECK15: omp.inner.for.inc:
14493 // CHECK15-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !13
14494 // CHECK15-NEXT: [[ADD1:%.*]] = add nsw i32 [[TMP4]], 1
14495 // CHECK15-NEXT: store i32 [[ADD1]], i32* [[DOTOMP_IV]], align 4, !llvm.access.group !13
14496 // CHECK15-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP14:![0-9]+]]
14497 // CHECK15: omp.inner.for.end:
14498 // CHECK15-NEXT: store i32 100, i32* [[I]], align 4
14499 // CHECK15-NEXT: store i32 0, i32* [[DOTOMP_LB3]], align 4
14500 // CHECK15-NEXT: store i32 99, i32* [[DOTOMP_UB4]], align 4
14501 // CHECK15-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_LB3]], align 4
14502 // CHECK15-NEXT: store i32 [[TMP5]], i32* [[DOTOMP_IV5]], align 4
14503 // CHECK15-NEXT: br label [[OMP_INNER_FOR_COND7:%.*]]
14504 // CHECK15: omp.inner.for.cond7:
14505 // CHECK15-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_IV5]], align 4
14506 // CHECK15-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_UB4]], align 4
14507 // CHECK15-NEXT: [[CMP8:%.*]] = icmp sle i32 [[TMP6]], [[TMP7]]
14508 // CHECK15-NEXT: br i1 [[CMP8]], label [[OMP_INNER_FOR_BODY9:%.*]], label [[OMP_INNER_FOR_END15:%.*]]
14509 // CHECK15: omp.inner.for.body9:
14510 // CHECK15-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_IV5]], align 4
14511 // CHECK15-NEXT: [[MUL10:%.*]] = mul nsw i32 [[TMP8]], 1
14512 // CHECK15-NEXT: [[ADD11:%.*]] = add nsw i32 0, [[MUL10]]
14513 // CHECK15-NEXT: store i32 [[ADD11]], i32* [[I6]], align 4
14514 // CHECK15-NEXT: call void @_Z3fn5v()
14515 // CHECK15-NEXT: br label [[OMP_BODY_CONTINUE12:%.*]]
14516 // CHECK15: omp.body.continue12:
14517 // CHECK15-NEXT: br label [[OMP_INNER_FOR_INC13:%.*]]
14518 // CHECK15: omp.inner.for.inc13:
14519 // CHECK15-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_IV5]], align 4
14520 // CHECK15-NEXT: [[ADD14:%.*]] = add nsw i32 [[TMP9]], 1
14521 // CHECK15-NEXT: store i32 [[ADD14]], i32* [[DOTOMP_IV5]], align 4
14522 // CHECK15-NEXT: br label [[OMP_INNER_FOR_COND7]], !llvm.loop [[LOOP16:![0-9]+]]
14523 // CHECK15: omp.inner.for.end15:
14524 // CHECK15-NEXT: store i32 100, i32* [[I6]], align 4
14525 // CHECK15-NEXT: [[TMP10:%.*]] = load i32, i32* @Arg, align 4
14526 // CHECK15-NEXT: [[TOBOOL:%.*]] = icmp ne i32 [[TMP10]], 0
14527 // CHECK15-NEXT: [[FROMBOOL:%.*]] = zext i1 [[TOBOOL]] to i8
14528 // CHECK15-NEXT: store i8 [[FROMBOOL]], i8* [[DOTCAPTURE_EXPR_]], align 1
14529 // CHECK15-NEXT: store i32 0, i32* [[DOTOMP_LB17]], align 4
14530 // CHECK15-NEXT: store i32 99, i32* [[DOTOMP_UB18]], align 4
14531 // CHECK15-NEXT: [[TMP11:%.*]] = load i32, i32* [[DOTOMP_LB17]], align 4
14532 // CHECK15-NEXT: store i32 [[TMP11]], i32* [[DOTOMP_IV19]], align 4
14533 // CHECK15-NEXT: [[TMP12:%.*]] = load i8, i8* [[DOTCAPTURE_EXPR_]], align 1
14534 // CHECK15-NEXT: [[TOBOOL21:%.*]] = trunc i8 [[TMP12]] to i1
14535 // CHECK15-NEXT: br i1 [[TOBOOL21]], label [[OMP_IF_THEN:%.*]], label [[OMP_IF_ELSE:%.*]]
14536 // CHECK15: omp_if.then:
14537 // CHECK15-NEXT: br label [[OMP_INNER_FOR_COND22:%.*]]
14538 // CHECK15: omp.inner.for.cond22:
14539 // CHECK15-NEXT: [[TMP13:%.*]] = load i32, i32* [[DOTOMP_IV19]], align 4, !llvm.access.group !18
14540 // CHECK15-NEXT: [[TMP14:%.*]] = load i32, i32* [[DOTOMP_UB18]], align 4, !llvm.access.group !18
14541 // CHECK15-NEXT: [[CMP23:%.*]] = icmp sle i32 [[TMP13]], [[TMP14]]
14542 // CHECK15-NEXT: br i1 [[CMP23]], label [[OMP_INNER_FOR_BODY24:%.*]], label [[OMP_INNER_FOR_END30:%.*]]
14543 // CHECK15: omp.inner.for.body24:
14544 // CHECK15-NEXT: [[TMP15:%.*]] = load i32, i32* [[DOTOMP_IV19]], align 4, !llvm.access.group !18
14545 // CHECK15-NEXT: [[MUL25:%.*]] = mul nsw i32 [[TMP15]], 1
14546 // CHECK15-NEXT: [[ADD26:%.*]] = add nsw i32 0, [[MUL25]]
14547 // CHECK15-NEXT: store i32 [[ADD26]], i32* [[I20]], align 4, !llvm.access.group !18
14548 // CHECK15-NEXT: call void @_Z3fn6v(), !llvm.access.group !18
14549 // CHECK15-NEXT: br label [[OMP_BODY_CONTINUE27:%.*]]
14550 // CHECK15: omp.body.continue27:
14551 // CHECK15-NEXT: br label [[OMP_INNER_FOR_INC28:%.*]]
14552 // CHECK15: omp.inner.for.inc28:
14553 // CHECK15-NEXT: [[TMP16:%.*]] = load i32, i32* [[DOTOMP_IV19]], align 4, !llvm.access.group !18
14554 // CHECK15-NEXT: [[ADD29:%.*]] = add nsw i32 [[TMP16]], 1
14555 // CHECK15-NEXT: store i32 [[ADD29]], i32* [[DOTOMP_IV19]], align 4, !llvm.access.group !18
14556 // CHECK15-NEXT: br label [[OMP_INNER_FOR_COND22]], !llvm.loop [[LOOP19:![0-9]+]]
14557 // CHECK15: omp.inner.for.end30:
14558 // CHECK15-NEXT: br label [[OMP_IF_END:%.*]]
14559 // CHECK15: omp_if.else:
14560 // CHECK15-NEXT: br label [[OMP_INNER_FOR_COND31:%.*]]
14561 // CHECK15: omp.inner.for.cond31:
14562 // CHECK15-NEXT: [[TMP17:%.*]] = load i32, i32* [[DOTOMP_IV19]], align 4
14563 // CHECK15-NEXT: [[TMP18:%.*]] = load i32, i32* [[DOTOMP_UB18]], align 4
14564 // CHECK15-NEXT: [[CMP32:%.*]] = icmp sle i32 [[TMP17]], [[TMP18]]
14565 // CHECK15-NEXT: br i1 [[CMP32]], label [[OMP_INNER_FOR_BODY33:%.*]], label [[OMP_INNER_FOR_END39:%.*]]
14566 // CHECK15: omp.inner.for.body33:
14567 // CHECK15-NEXT: [[TMP19:%.*]] = load i32, i32* [[DOTOMP_IV19]], align 4
14568 // CHECK15-NEXT: [[MUL34:%.*]] = mul nsw i32 [[TMP19]], 1
14569 // CHECK15-NEXT: [[ADD35:%.*]] = add nsw i32 0, [[MUL34]]
14570 // CHECK15-NEXT: store i32 [[ADD35]], i32* [[I20]], align 4
14571 // CHECK15-NEXT: call void @_Z3fn6v()
14572 // CHECK15-NEXT: br label [[OMP_BODY_CONTINUE36:%.*]]
14573 // CHECK15: omp.body.continue36:
14574 // CHECK15-NEXT: br label [[OMP_INNER_FOR_INC37:%.*]]
14575 // CHECK15: omp.inner.for.inc37:
14576 // CHECK15-NEXT: [[TMP20:%.*]] = load i32, i32* [[DOTOMP_IV19]], align 4
14577 // CHECK15-NEXT: [[ADD38:%.*]] = add nsw i32 [[TMP20]], 1
14578 // CHECK15-NEXT: store i32 [[ADD38]], i32* [[DOTOMP_IV19]], align 4
14579 // CHECK15-NEXT: br label [[OMP_INNER_FOR_COND31]], !llvm.loop [[LOOP21:![0-9]+]]
14580 // CHECK15: omp.inner.for.end39:
14581 // CHECK15-NEXT: br label [[OMP_IF_END]]
14582 // CHECK15: omp_if.end:
14583 // CHECK15-NEXT: store i32 100, i32* [[I20]], align 4
14584 // CHECK15-NEXT: [[TMP21:%.*]] = load i32, i32* @Arg, align 4
14585 // CHECK15-NEXT: [[CALL:%.*]] = call i32 @_Z5tmainIiEiT_(i32 [[TMP21]])
14586 // CHECK15-NEXT: ret i32 [[CALL]]
14587 //
14588 //
14589 // CHECK15-LABEL: define {{[^@]+}}@_Z5tmainIiEiT_
14590 // CHECK15-SAME: (i32 [[ARG:%.*]]) #[[ATTR0]] comdat {
14591 // CHECK15-NEXT: entry:
14592 // CHECK15-NEXT: [[ARG_ADDR:%.*]] = alloca i32, align 4
14593 // CHECK15-NEXT: [[TMP:%.*]] = alloca i32, align 4
14594 // CHECK15-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4
14595 // CHECK15-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4
14596 // CHECK15-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
14597 // CHECK15-NEXT: [[I:%.*]] = alloca i32, align 4
14598 // CHECK15-NEXT: [[_TMP2:%.*]] = alloca i32, align 4
14599 // CHECK15-NEXT: [[DOTOMP_LB3:%.*]] = alloca i32, align 4
14600 // CHECK15-NEXT: [[DOTOMP_UB4:%.*]] = alloca i32, align 4
14601 // CHECK15-NEXT: [[DOTOMP_IV5:%.*]] = alloca i32, align 4
14602 // CHECK15-NEXT: [[I6:%.*]] = alloca i32, align 4
14603 // CHECK15-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i8, align 1
14604 // CHECK15-NEXT: [[_TMP16:%.*]] = alloca i32, align 4
14605 // CHECK15-NEXT: [[DOTOMP_LB17:%.*]] = alloca i32, align 4
14606 // CHECK15-NEXT: [[DOTOMP_UB18:%.*]] = alloca i32, align 4
14607 // CHECK15-NEXT: [[DOTOMP_IV19:%.*]] = alloca i32, align 4
14608 // CHECK15-NEXT: [[I20:%.*]] = alloca i32, align 4
14609 // CHECK15-NEXT: store i32 [[ARG]], i32* [[ARG_ADDR]], align 4
14610 // CHECK15-NEXT: store i32 0, i32* [[DOTOMP_LB]], align 4
14611 // CHECK15-NEXT: store i32 99, i32* [[DOTOMP_UB]], align 4
14612 // CHECK15-NEXT: [[TMP0:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4
14613 // CHECK15-NEXT: store i32 [[TMP0]], i32* [[DOTOMP_IV]], align 4
14614 // CHECK15-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
14615 // CHECK15: omp.inner.for.cond:
14616 // CHECK15-NEXT: [[TMP1:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !22
14617 // CHECK15-NEXT: [[TMP2:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4, !llvm.access.group !22
14618 // CHECK15-NEXT: [[CMP:%.*]] = icmp sle i32 [[TMP1]], [[TMP2]]
14619 // CHECK15-NEXT: br i1 [[CMP]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
14620 // CHECK15: omp.inner.for.body:
14621 // CHECK15-NEXT: [[TMP3:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !22
14622 // CHECK15-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP3]], 1
14623 // CHECK15-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
14624 // CHECK15-NEXT: store i32 [[ADD]], i32* [[I]], align 4, !llvm.access.group !22
14625 // CHECK15-NEXT: call void @_Z3fn1v(), !llvm.access.group !22
14626 // CHECK15-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
14627 // CHECK15: omp.body.continue:
14628 // CHECK15-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
14629 // CHECK15: omp.inner.for.inc:
14630 // CHECK15-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !22
14631 // CHECK15-NEXT: [[ADD1:%.*]] = add nsw i32 [[TMP4]], 1
14632 // CHECK15-NEXT: store i32 [[ADD1]], i32* [[DOTOMP_IV]], align 4, !llvm.access.group !22
14633 // CHECK15-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP23:![0-9]+]]
14634 // CHECK15: omp.inner.for.end:
14635 // CHECK15-NEXT: store i32 100, i32* [[I]], align 4
14636 // CHECK15-NEXT: store i32 0, i32* [[DOTOMP_LB3]], align 4
14637 // CHECK15-NEXT: store i32 99, i32* [[DOTOMP_UB4]], align 4
14638 // CHECK15-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_LB3]], align 4
14639 // CHECK15-NEXT: store i32 [[TMP5]], i32* [[DOTOMP_IV5]], align 4
14640 // CHECK15-NEXT: br label [[OMP_INNER_FOR_COND7:%.*]]
14641 // CHECK15: omp.inner.for.cond7:
14642 // CHECK15-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_IV5]], align 4
14643 // CHECK15-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_UB4]], align 4
14644 // CHECK15-NEXT: [[CMP8:%.*]] = icmp sle i32 [[TMP6]], [[TMP7]]
14645 // CHECK15-NEXT: br i1 [[CMP8]], label [[OMP_INNER_FOR_BODY9:%.*]], label [[OMP_INNER_FOR_END15:%.*]]
14646 // CHECK15: omp.inner.for.body9:
14647 // CHECK15-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_IV5]], align 4
14648 // CHECK15-NEXT: [[MUL10:%.*]] = mul nsw i32 [[TMP8]], 1
14649 // CHECK15-NEXT: [[ADD11:%.*]] = add nsw i32 0, [[MUL10]]
14650 // CHECK15-NEXT: store i32 [[ADD11]], i32* [[I6]], align 4
14651 // CHECK15-NEXT: call void @_Z3fn2v()
14652 // CHECK15-NEXT: br label [[OMP_BODY_CONTINUE12:%.*]]
14653 // CHECK15: omp.body.continue12:
14654 // CHECK15-NEXT: br label [[OMP_INNER_FOR_INC13:%.*]]
14655 // CHECK15: omp.inner.for.inc13:
14656 // CHECK15-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_IV5]], align 4
14657 // CHECK15-NEXT: [[ADD14:%.*]] = add nsw i32 [[TMP9]], 1
14658 // CHECK15-NEXT: store i32 [[ADD14]], i32* [[DOTOMP_IV5]], align 4
14659 // CHECK15-NEXT: br label [[OMP_INNER_FOR_COND7]], !llvm.loop [[LOOP25:![0-9]+]]
14660 // CHECK15: omp.inner.for.end15:
14661 // CHECK15-NEXT: store i32 100, i32* [[I6]], align 4
14662 // CHECK15-NEXT: [[TMP10:%.*]] = load i32, i32* [[ARG_ADDR]], align 4
14663 // CHECK15-NEXT: [[TOBOOL:%.*]] = icmp ne i32 [[TMP10]], 0
14664 // CHECK15-NEXT: [[FROMBOOL:%.*]] = zext i1 [[TOBOOL]] to i8
14665 // CHECK15-NEXT: store i8 [[FROMBOOL]], i8* [[DOTCAPTURE_EXPR_]], align 1
14666 // CHECK15-NEXT: store i32 0, i32* [[DOTOMP_LB17]], align 4
14667 // CHECK15-NEXT: store i32 99, i32* [[DOTOMP_UB18]], align 4
14668 // CHECK15-NEXT: [[TMP11:%.*]] = load i32, i32* [[DOTOMP_LB17]], align 4
14669 // CHECK15-NEXT: store i32 [[TMP11]], i32* [[DOTOMP_IV19]], align 4
14670 // CHECK15-NEXT: br label [[OMP_INNER_FOR_COND21:%.*]]
14671 // CHECK15: omp.inner.for.cond21:
14672 // CHECK15-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTOMP_IV19]], align 4, !llvm.access.group !26
14673 // CHECK15-NEXT: [[TMP13:%.*]] = load i32, i32* [[DOTOMP_UB18]], align 4, !llvm.access.group !26
14674 // CHECK15-NEXT: [[CMP22:%.*]] = icmp sle i32 [[TMP12]], [[TMP13]]
14675 // CHECK15-NEXT: br i1 [[CMP22]], label [[OMP_INNER_FOR_BODY23:%.*]], label [[OMP_INNER_FOR_END29:%.*]]
14676 // CHECK15: omp.inner.for.body23:
14677 // CHECK15-NEXT: [[TMP14:%.*]] = load i32, i32* [[DOTOMP_IV19]], align 4, !llvm.access.group !26
14678 // CHECK15-NEXT: [[MUL24:%.*]] = mul nsw i32 [[TMP14]], 1
14679 // CHECK15-NEXT: [[ADD25:%.*]] = add nsw i32 0, [[MUL24]]
14680 // CHECK15-NEXT: store i32 [[ADD25]], i32* [[I20]], align 4, !llvm.access.group !26
14681 // CHECK15-NEXT: call void @_Z3fn3v(), !llvm.access.group !26
14682 // CHECK15-NEXT: br label [[OMP_BODY_CONTINUE26:%.*]]
14683 // CHECK15: omp.body.continue26:
14684 // CHECK15-NEXT: br label [[OMP_INNER_FOR_INC27:%.*]]
14685 // CHECK15: omp.inner.for.inc27:
14686 // CHECK15-NEXT: [[TMP15:%.*]] = load i32, i32* [[DOTOMP_IV19]], align 4, !llvm.access.group !26
14687 // CHECK15-NEXT: [[ADD28:%.*]] = add nsw i32 [[TMP15]], 1
14688 // CHECK15-NEXT: store i32 [[ADD28]], i32* [[DOTOMP_IV19]], align 4, !llvm.access.group !26
14689 // CHECK15-NEXT: br label [[OMP_INNER_FOR_COND21]], !llvm.loop [[LOOP27:![0-9]+]]
14690 // CHECK15: omp.inner.for.end29:
14691 // CHECK15-NEXT: store i32 100, i32* [[I20]], align 4
14692 // CHECK15-NEXT: ret i32 0
14693 //
14694 //
14695 // CHECK16-LABEL: define {{[^@]+}}@_Z9gtid_testv
14696 // CHECK16-SAME: () #[[ATTR0:[0-9]+]] {
14697 // CHECK16-NEXT: entry:
14698 // CHECK16-NEXT: [[TMP:%.*]] = alloca i32, align 4
14699 // CHECK16-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4
14700 // CHECK16-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4
14701 // CHECK16-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
14702 // CHECK16-NEXT: [[I:%.*]] = alloca i32, align 4
14703 // CHECK16-NEXT: [[_TMP2:%.*]] = alloca i32, align 4
14704 // CHECK16-NEXT: [[DOTOMP_LB3:%.*]] = alloca i32, align 4
14705 // CHECK16-NEXT: [[DOTOMP_UB4:%.*]] = alloca i32, align 4
14706 // CHECK16-NEXT: [[DOTOMP_IV5:%.*]] = alloca i32, align 4
14707 // CHECK16-NEXT: [[I6:%.*]] = alloca i32, align 4
14708 // CHECK16-NEXT: store i32 0, i32* [[DOTOMP_LB]], align 4
14709 // CHECK16-NEXT: store i32 99, i32* [[DOTOMP_UB]], align 4
14710 // CHECK16-NEXT: [[TMP0:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4
14711 // CHECK16-NEXT: store i32 [[TMP0]], i32* [[DOTOMP_IV]], align 4
14712 // CHECK16-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
14713 // CHECK16: omp.inner.for.cond:
14714 // CHECK16-NEXT: [[TMP1:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !6
14715 // CHECK16-NEXT: [[TMP2:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4, !llvm.access.group !6
14716 // CHECK16-NEXT: [[CMP:%.*]] = icmp sle i32 [[TMP1]], [[TMP2]]
14717 // CHECK16-NEXT: br i1 [[CMP]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
14718 // CHECK16: omp.inner.for.body:
14719 // CHECK16-NEXT: [[TMP3:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !6
14720 // CHECK16-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP3]], 1
14721 // CHECK16-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
14722 // CHECK16-NEXT: store i32 [[ADD]], i32* [[I]], align 4, !llvm.access.group !6
14723 // CHECK16-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
14724 // CHECK16: omp.body.continue:
14725 // CHECK16-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
14726 // CHECK16: omp.inner.for.inc:
14727 // CHECK16-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !6
14728 // CHECK16-NEXT: [[ADD1:%.*]] = add nsw i32 [[TMP4]], 1
14729 // CHECK16-NEXT: store i32 [[ADD1]], i32* [[DOTOMP_IV]], align 4, !llvm.access.group !6
14730 // CHECK16-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP7:![0-9]+]]
14731 // CHECK16: omp.inner.for.end:
14732 // CHECK16-NEXT: store i32 100, i32* [[I]], align 4
14733 // CHECK16-NEXT: store i32 0, i32* [[DOTOMP_LB3]], align 4
14734 // CHECK16-NEXT: store i32 99, i32* [[DOTOMP_UB4]], align 4
14735 // CHECK16-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_LB3]], align 4
14736 // CHECK16-NEXT: store i32 [[TMP5]], i32* [[DOTOMP_IV5]], align 4
14737 // CHECK16-NEXT: br label [[OMP_INNER_FOR_COND7:%.*]]
14738 // CHECK16: omp.inner.for.cond7:
14739 // CHECK16-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_IV5]], align 4, !llvm.access.group !10
14740 // CHECK16-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_UB4]], align 4, !llvm.access.group !10
14741 // CHECK16-NEXT: [[CMP8:%.*]] = icmp sle i32 [[TMP6]], [[TMP7]]
14742 // CHECK16-NEXT: br i1 [[CMP8]], label [[OMP_INNER_FOR_BODY9:%.*]], label [[OMP_INNER_FOR_END15:%.*]]
14743 // CHECK16: omp.inner.for.body9:
14744 // CHECK16-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_IV5]], align 4, !llvm.access.group !10
14745 // CHECK16-NEXT: [[MUL10:%.*]] = mul nsw i32 [[TMP8]], 1
14746 // CHECK16-NEXT: [[ADD11:%.*]] = add nsw i32 0, [[MUL10]]
14747 // CHECK16-NEXT: store i32 [[ADD11]], i32* [[I6]], align 4, !llvm.access.group !10
14748 // CHECK16-NEXT: call void @_Z9gtid_testv(), !llvm.access.group !10
14749 // CHECK16-NEXT: br label [[OMP_BODY_CONTINUE12:%.*]]
14750 // CHECK16: omp.body.continue12:
14751 // CHECK16-NEXT: br label [[OMP_INNER_FOR_INC13:%.*]]
14752 // CHECK16: omp.inner.for.inc13:
14753 // CHECK16-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_IV5]], align 4, !llvm.access.group !10
14754 // CHECK16-NEXT: [[ADD14:%.*]] = add nsw i32 [[TMP9]], 1
14755 // CHECK16-NEXT: store i32 [[ADD14]], i32* [[DOTOMP_IV5]], align 4, !llvm.access.group !10
14756 // CHECK16-NEXT: br label [[OMP_INNER_FOR_COND7]], !llvm.loop [[LOOP11:![0-9]+]]
14757 // CHECK16: omp.inner.for.end15:
14758 // CHECK16-NEXT: store i32 100, i32* [[I6]], align 4
14759 // CHECK16-NEXT: ret void
14760 //
14761 //
14762 // CHECK16-LABEL: define {{[^@]+}}@main
14763 // CHECK16-SAME: () #[[ATTR1:[0-9]+]] {
14764 // CHECK16-NEXT: entry:
14765 // CHECK16-NEXT: [[RETVAL:%.*]] = alloca i32, align 4
14766 // CHECK16-NEXT: [[TMP:%.*]] = alloca i32, align 4
14767 // CHECK16-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4
14768 // CHECK16-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4
14769 // CHECK16-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
14770 // CHECK16-NEXT: [[I:%.*]] = alloca i32, align 4
14771 // CHECK16-NEXT: [[_TMP2:%.*]] = alloca i32, align 4
14772 // CHECK16-NEXT: [[DOTOMP_LB3:%.*]] = alloca i32, align 4
14773 // CHECK16-NEXT: [[DOTOMP_UB4:%.*]] = alloca i32, align 4
14774 // CHECK16-NEXT: [[DOTOMP_IV5:%.*]] = alloca i32, align 4
14775 // CHECK16-NEXT: [[I6:%.*]] = alloca i32, align 4
14776 // CHECK16-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i8, align 1
14777 // CHECK16-NEXT: [[_TMP16:%.*]] = alloca i32, align 4
14778 // CHECK16-NEXT: [[DOTOMP_LB17:%.*]] = alloca i32, align 4
14779 // CHECK16-NEXT: [[DOTOMP_UB18:%.*]] = alloca i32, align 4
14780 // CHECK16-NEXT: [[DOTOMP_IV19:%.*]] = alloca i32, align 4
14781 // CHECK16-NEXT: [[I20:%.*]] = alloca i32, align 4
14782 // CHECK16-NEXT: store i32 0, i32* [[RETVAL]], align 4
14783 // CHECK16-NEXT: store i32 0, i32* [[DOTOMP_LB]], align 4
14784 // CHECK16-NEXT: store i32 99, i32* [[DOTOMP_UB]], align 4
14785 // CHECK16-NEXT: [[TMP0:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4
14786 // CHECK16-NEXT: store i32 [[TMP0]], i32* [[DOTOMP_IV]], align 4
14787 // CHECK16-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
14788 // CHECK16: omp.inner.for.cond:
14789 // CHECK16-NEXT: [[TMP1:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !13
14790 // CHECK16-NEXT: [[TMP2:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4, !llvm.access.group !13
14791 // CHECK16-NEXT: [[CMP:%.*]] = icmp sle i32 [[TMP1]], [[TMP2]]
14792 // CHECK16-NEXT: br i1 [[CMP]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
14793 // CHECK16: omp.inner.for.body:
14794 // CHECK16-NEXT: [[TMP3:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !13
14795 // CHECK16-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP3]], 1
14796 // CHECK16-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
14797 // CHECK16-NEXT: store i32 [[ADD]], i32* [[I]], align 4, !llvm.access.group !13
14798 // CHECK16-NEXT: call void @_Z3fn4v(), !llvm.access.group !13
14799 // CHECK16-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
14800 // CHECK16: omp.body.continue:
14801 // CHECK16-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
14802 // CHECK16: omp.inner.for.inc:
14803 // CHECK16-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !13
14804 // CHECK16-NEXT: [[ADD1:%.*]] = add nsw i32 [[TMP4]], 1
14805 // CHECK16-NEXT: store i32 [[ADD1]], i32* [[DOTOMP_IV]], align 4, !llvm.access.group !13
14806 // CHECK16-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP14:![0-9]+]]
14807 // CHECK16: omp.inner.for.end:
14808 // CHECK16-NEXT: store i32 100, i32* [[I]], align 4
14809 // CHECK16-NEXT: store i32 0, i32* [[DOTOMP_LB3]], align 4
14810 // CHECK16-NEXT: store i32 99, i32* [[DOTOMP_UB4]], align 4
14811 // CHECK16-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_LB3]], align 4
14812 // CHECK16-NEXT: store i32 [[TMP5]], i32* [[DOTOMP_IV5]], align 4
14813 // CHECK16-NEXT: br label [[OMP_INNER_FOR_COND7:%.*]]
14814 // CHECK16: omp.inner.for.cond7:
14815 // CHECK16-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_IV5]], align 4
14816 // CHECK16-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_UB4]], align 4
14817 // CHECK16-NEXT: [[CMP8:%.*]] = icmp sle i32 [[TMP6]], [[TMP7]]
14818 // CHECK16-NEXT: br i1 [[CMP8]], label [[OMP_INNER_FOR_BODY9:%.*]], label [[OMP_INNER_FOR_END15:%.*]]
14819 // CHECK16: omp.inner.for.body9:
14820 // CHECK16-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_IV5]], align 4
14821 // CHECK16-NEXT: [[MUL10:%.*]] = mul nsw i32 [[TMP8]], 1
14822 // CHECK16-NEXT: [[ADD11:%.*]] = add nsw i32 0, [[MUL10]]
14823 // CHECK16-NEXT: store i32 [[ADD11]], i32* [[I6]], align 4
14824 // CHECK16-NEXT: call void @_Z3fn5v()
14825 // CHECK16-NEXT: br label [[OMP_BODY_CONTINUE12:%.*]]
14826 // CHECK16: omp.body.continue12:
14827 // CHECK16-NEXT: br label [[OMP_INNER_FOR_INC13:%.*]]
14828 // CHECK16: omp.inner.for.inc13:
14829 // CHECK16-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_IV5]], align 4
14830 // CHECK16-NEXT: [[ADD14:%.*]] = add nsw i32 [[TMP9]], 1
14831 // CHECK16-NEXT: store i32 [[ADD14]], i32* [[DOTOMP_IV5]], align 4
14832 // CHECK16-NEXT: br label [[OMP_INNER_FOR_COND7]], !llvm.loop [[LOOP16:![0-9]+]]
14833 // CHECK16: omp.inner.for.end15:
14834 // CHECK16-NEXT: store i32 100, i32* [[I6]], align 4
14835 // CHECK16-NEXT: [[TMP10:%.*]] = load i32, i32* @Arg, align 4
14836 // CHECK16-NEXT: [[TOBOOL:%.*]] = icmp ne i32 [[TMP10]], 0
14837 // CHECK16-NEXT: [[FROMBOOL:%.*]] = zext i1 [[TOBOOL]] to i8
14838 // CHECK16-NEXT: store i8 [[FROMBOOL]], i8* [[DOTCAPTURE_EXPR_]], align 1
14839 // CHECK16-NEXT: store i32 0, i32* [[DOTOMP_LB17]], align 4
14840 // CHECK16-NEXT: store i32 99, i32* [[DOTOMP_UB18]], align 4
14841 // CHECK16-NEXT: [[TMP11:%.*]] = load i32, i32* [[DOTOMP_LB17]], align 4
14842 // CHECK16-NEXT: store i32 [[TMP11]], i32* [[DOTOMP_IV19]], align 4
14843 // CHECK16-NEXT: [[TMP12:%.*]] = load i8, i8* [[DOTCAPTURE_EXPR_]], align 1
14844 // CHECK16-NEXT: [[TOBOOL21:%.*]] = trunc i8 [[TMP12]] to i1
14845 // CHECK16-NEXT: br i1 [[TOBOOL21]], label [[OMP_IF_THEN:%.*]], label [[OMP_IF_ELSE:%.*]]
14846 // CHECK16: omp_if.then:
14847 // CHECK16-NEXT: br label [[OMP_INNER_FOR_COND22:%.*]]
14848 // CHECK16: omp.inner.for.cond22:
14849 // CHECK16-NEXT: [[TMP13:%.*]] = load i32, i32* [[DOTOMP_IV19]], align 4, !llvm.access.group !18
14850 // CHECK16-NEXT: [[TMP14:%.*]] = load i32, i32* [[DOTOMP_UB18]], align 4, !llvm.access.group !18
14851 // CHECK16-NEXT: [[CMP23:%.*]] = icmp sle i32 [[TMP13]], [[TMP14]]
14852 // CHECK16-NEXT: br i1 [[CMP23]], label [[OMP_INNER_FOR_BODY24:%.*]], label [[OMP_INNER_FOR_END30:%.*]]
14853 // CHECK16: omp.inner.for.body24:
14854 // CHECK16-NEXT: [[TMP15:%.*]] = load i32, i32* [[DOTOMP_IV19]], align 4, !llvm.access.group !18
14855 // CHECK16-NEXT: [[MUL25:%.*]] = mul nsw i32 [[TMP15]], 1
14856 // CHECK16-NEXT: [[ADD26:%.*]] = add nsw i32 0, [[MUL25]]
14857 // CHECK16-NEXT: store i32 [[ADD26]], i32* [[I20]], align 4, !llvm.access.group !18
14858 // CHECK16-NEXT: call void @_Z3fn6v(), !llvm.access.group !18
14859 // CHECK16-NEXT: br label [[OMP_BODY_CONTINUE27:%.*]]
14860 // CHECK16: omp.body.continue27:
14861 // CHECK16-NEXT: br label [[OMP_INNER_FOR_INC28:%.*]]
14862 // CHECK16: omp.inner.for.inc28:
14863 // CHECK16-NEXT: [[TMP16:%.*]] = load i32, i32* [[DOTOMP_IV19]], align 4, !llvm.access.group !18
14864 // CHECK16-NEXT: [[ADD29:%.*]] = add nsw i32 [[TMP16]], 1
14865 // CHECK16-NEXT: store i32 [[ADD29]], i32* [[DOTOMP_IV19]], align 4, !llvm.access.group !18
14866 // CHECK16-NEXT: br label [[OMP_INNER_FOR_COND22]], !llvm.loop [[LOOP19:![0-9]+]]
14867 // CHECK16: omp.inner.for.end30:
14868 // CHECK16-NEXT: br label [[OMP_IF_END:%.*]]
14869 // CHECK16: omp_if.else:
14870 // CHECK16-NEXT: br label [[OMP_INNER_FOR_COND31:%.*]]
14871 // CHECK16: omp.inner.for.cond31:
14872 // CHECK16-NEXT: [[TMP17:%.*]] = load i32, i32* [[DOTOMP_IV19]], align 4
14873 // CHECK16-NEXT: [[TMP18:%.*]] = load i32, i32* [[DOTOMP_UB18]], align 4
14874 // CHECK16-NEXT: [[CMP32:%.*]] = icmp sle i32 [[TMP17]], [[TMP18]]
14875 // CHECK16-NEXT: br i1 [[CMP32]], label [[OMP_INNER_FOR_BODY33:%.*]], label [[OMP_INNER_FOR_END39:%.*]]
14876 // CHECK16: omp.inner.for.body33:
14877 // CHECK16-NEXT: [[TMP19:%.*]] = load i32, i32* [[DOTOMP_IV19]], align 4
14878 // CHECK16-NEXT: [[MUL34:%.*]] = mul nsw i32 [[TMP19]], 1
14879 // CHECK16-NEXT: [[ADD35:%.*]] = add nsw i32 0, [[MUL34]]
14880 // CHECK16-NEXT: store i32 [[ADD35]], i32* [[I20]], align 4
14881 // CHECK16-NEXT: call void @_Z3fn6v()
14882 // CHECK16-NEXT: br label [[OMP_BODY_CONTINUE36:%.*]]
14883 // CHECK16: omp.body.continue36:
14884 // CHECK16-NEXT: br label [[OMP_INNER_FOR_INC37:%.*]]
14885 // CHECK16: omp.inner.for.inc37:
14886 // CHECK16-NEXT: [[TMP20:%.*]] = load i32, i32* [[DOTOMP_IV19]], align 4
14887 // CHECK16-NEXT: [[ADD38:%.*]] = add nsw i32 [[TMP20]], 1
14888 // CHECK16-NEXT: store i32 [[ADD38]], i32* [[DOTOMP_IV19]], align 4
14889 // CHECK16-NEXT: br label [[OMP_INNER_FOR_COND31]], !llvm.loop [[LOOP21:![0-9]+]]
14890 // CHECK16: omp.inner.for.end39:
14891 // CHECK16-NEXT: br label [[OMP_IF_END]]
14892 // CHECK16: omp_if.end:
14893 // CHECK16-NEXT: store i32 100, i32* [[I20]], align 4
14894 // CHECK16-NEXT: [[TMP21:%.*]] = load i32, i32* @Arg, align 4
14895 // CHECK16-NEXT: [[CALL:%.*]] = call i32 @_Z5tmainIiEiT_(i32 [[TMP21]])
14896 // CHECK16-NEXT: ret i32 [[CALL]]
14897 //
14898 //
14899 // CHECK16-LABEL: define {{[^@]+}}@_Z5tmainIiEiT_
14900 // CHECK16-SAME: (i32 [[ARG:%.*]]) #[[ATTR0]] comdat {
14901 // CHECK16-NEXT: entry:
14902 // CHECK16-NEXT: [[ARG_ADDR:%.*]] = alloca i32, align 4
14903 // CHECK16-NEXT: [[TMP:%.*]] = alloca i32, align 4
14904 // CHECK16-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4
14905 // CHECK16-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4
14906 // CHECK16-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
14907 // CHECK16-NEXT: [[I:%.*]] = alloca i32, align 4
14908 // CHECK16-NEXT: [[_TMP2:%.*]] = alloca i32, align 4
14909 // CHECK16-NEXT: [[DOTOMP_LB3:%.*]] = alloca i32, align 4
14910 // CHECK16-NEXT: [[DOTOMP_UB4:%.*]] = alloca i32, align 4
14911 // CHECK16-NEXT: [[DOTOMP_IV5:%.*]] = alloca i32, align 4
14912 // CHECK16-NEXT: [[I6:%.*]] = alloca i32, align 4
14913 // CHECK16-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i8, align 1
14914 // CHECK16-NEXT: [[_TMP16:%.*]] = alloca i32, align 4
14915 // CHECK16-NEXT: [[DOTOMP_LB17:%.*]] = alloca i32, align 4
14916 // CHECK16-NEXT: [[DOTOMP_UB18:%.*]] = alloca i32, align 4
14917 // CHECK16-NEXT: [[DOTOMP_IV19:%.*]] = alloca i32, align 4
14918 // CHECK16-NEXT: [[I20:%.*]] = alloca i32, align 4
14919 // CHECK16-NEXT: store i32 [[ARG]], i32* [[ARG_ADDR]], align 4
14920 // CHECK16-NEXT: store i32 0, i32* [[DOTOMP_LB]], align 4
14921 // CHECK16-NEXT: store i32 99, i32* [[DOTOMP_UB]], align 4
14922 // CHECK16-NEXT: [[TMP0:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4
14923 // CHECK16-NEXT: store i32 [[TMP0]], i32* [[DOTOMP_IV]], align 4
14924 // CHECK16-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
14925 // CHECK16: omp.inner.for.cond:
14926 // CHECK16-NEXT: [[TMP1:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !22
14927 // CHECK16-NEXT: [[TMP2:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4, !llvm.access.group !22
14928 // CHECK16-NEXT: [[CMP:%.*]] = icmp sle i32 [[TMP1]], [[TMP2]]
14929 // CHECK16-NEXT: br i1 [[CMP]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
14930 // CHECK16: omp.inner.for.body:
14931 // CHECK16-NEXT: [[TMP3:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !22
14932 // CHECK16-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP3]], 1
14933 // CHECK16-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
14934 // CHECK16-NEXT: store i32 [[ADD]], i32* [[I]], align 4, !llvm.access.group !22
14935 // CHECK16-NEXT: call void @_Z3fn1v(), !llvm.access.group !22
14936 // CHECK16-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
14937 // CHECK16: omp.body.continue:
14938 // CHECK16-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
14939 // CHECK16: omp.inner.for.inc:
14940 // CHECK16-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !22
14941 // CHECK16-NEXT: [[ADD1:%.*]] = add nsw i32 [[TMP4]], 1
14942 // CHECK16-NEXT: store i32 [[ADD1]], i32* [[DOTOMP_IV]], align 4, !llvm.access.group !22
14943 // CHECK16-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP23:![0-9]+]]
14944 // CHECK16: omp.inner.for.end:
14945 // CHECK16-NEXT: store i32 100, i32* [[I]], align 4
14946 // CHECK16-NEXT: store i32 0, i32* [[DOTOMP_LB3]], align 4
14947 // CHECK16-NEXT: store i32 99, i32* [[DOTOMP_UB4]], align 4
14948 // CHECK16-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_LB3]], align 4
14949 // CHECK16-NEXT: store i32 [[TMP5]], i32* [[DOTOMP_IV5]], align 4
14950 // CHECK16-NEXT: br label [[OMP_INNER_FOR_COND7:%.*]]
14951 // CHECK16: omp.inner.for.cond7:
14952 // CHECK16-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_IV5]], align 4
14953 // CHECK16-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_UB4]], align 4
14954 // CHECK16-NEXT: [[CMP8:%.*]] = icmp sle i32 [[TMP6]], [[TMP7]]
14955 // CHECK16-NEXT: br i1 [[CMP8]], label [[OMP_INNER_FOR_BODY9:%.*]], label [[OMP_INNER_FOR_END15:%.*]]
14956 // CHECK16: omp.inner.for.body9:
14957 // CHECK16-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_IV5]], align 4
14958 // CHECK16-NEXT: [[MUL10:%.*]] = mul nsw i32 [[TMP8]], 1
14959 // CHECK16-NEXT: [[ADD11:%.*]] = add nsw i32 0, [[MUL10]]
14960 // CHECK16-NEXT: store i32 [[ADD11]], i32* [[I6]], align 4
14961 // CHECK16-NEXT: call void @_Z3fn2v()
14962 // CHECK16-NEXT: br label [[OMP_BODY_CONTINUE12:%.*]]
14963 // CHECK16: omp.body.continue12:
14964 // CHECK16-NEXT: br label [[OMP_INNER_FOR_INC13:%.*]]
14965 // CHECK16: omp.inner.for.inc13:
14966 // CHECK16-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_IV5]], align 4
14967 // CHECK16-NEXT: [[ADD14:%.*]] = add nsw i32 [[TMP9]], 1
14968 // CHECK16-NEXT: store i32 [[ADD14]], i32* [[DOTOMP_IV5]], align 4
14969 // CHECK16-NEXT: br label [[OMP_INNER_FOR_COND7]], !llvm.loop [[LOOP25:![0-9]+]]
14970 // CHECK16: omp.inner.for.end15:
14971 // CHECK16-NEXT: store i32 100, i32* [[I6]], align 4
14972 // CHECK16-NEXT: [[TMP10:%.*]] = load i32, i32* [[ARG_ADDR]], align 4
14973 // CHECK16-NEXT: [[TOBOOL:%.*]] = icmp ne i32 [[TMP10]], 0
14974 // CHECK16-NEXT: [[FROMBOOL:%.*]] = zext i1 [[TOBOOL]] to i8
14975 // CHECK16-NEXT: store i8 [[FROMBOOL]], i8* [[DOTCAPTURE_EXPR_]], align 1
14976 // CHECK16-NEXT: store i32 0, i32* [[DOTOMP_LB17]], align 4
14977 // CHECK16-NEXT: store i32 99, i32* [[DOTOMP_UB18]], align 4
14978 // CHECK16-NEXT: [[TMP11:%.*]] = load i32, i32* [[DOTOMP_LB17]], align 4
14979 // CHECK16-NEXT: store i32 [[TMP11]], i32* [[DOTOMP_IV19]], align 4
14980 // CHECK16-NEXT: br label [[OMP_INNER_FOR_COND21:%.*]]
14981 // CHECK16: omp.inner.for.cond21:
14982 // CHECK16-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTOMP_IV19]], align 4, !llvm.access.group !26
14983 // CHECK16-NEXT: [[TMP13:%.*]] = load i32, i32* [[DOTOMP_UB18]], align 4, !llvm.access.group !26
14984 // CHECK16-NEXT: [[CMP22:%.*]] = icmp sle i32 [[TMP12]], [[TMP13]]
14985 // CHECK16-NEXT: br i1 [[CMP22]], label [[OMP_INNER_FOR_BODY23:%.*]], label [[OMP_INNER_FOR_END29:%.*]]
14986 // CHECK16: omp.inner.for.body23:
14987 // CHECK16-NEXT: [[TMP14:%.*]] = load i32, i32* [[DOTOMP_IV19]], align 4, !llvm.access.group !26
14988 // CHECK16-NEXT: [[MUL24:%.*]] = mul nsw i32 [[TMP14]], 1
14989 // CHECK16-NEXT: [[ADD25:%.*]] = add nsw i32 0, [[MUL24]]
14990 // CHECK16-NEXT: store i32 [[ADD25]], i32* [[I20]], align 4, !llvm.access.group !26
14991 // CHECK16-NEXT: call void @_Z3fn3v(), !llvm.access.group !26
14992 // CHECK16-NEXT: br label [[OMP_BODY_CONTINUE26:%.*]]
14993 // CHECK16: omp.body.continue26:
14994 // CHECK16-NEXT: br label [[OMP_INNER_FOR_INC27:%.*]]
14995 // CHECK16: omp.inner.for.inc27:
14996 // CHECK16-NEXT: [[TMP15:%.*]] = load i32, i32* [[DOTOMP_IV19]], align 4, !llvm.access.group !26
14997 // CHECK16-NEXT: [[ADD28:%.*]] = add nsw i32 [[TMP15]], 1
14998 // CHECK16-NEXT: store i32 [[ADD28]], i32* [[DOTOMP_IV19]], align 4, !llvm.access.group !26
14999 // CHECK16-NEXT: br label [[OMP_INNER_FOR_COND21]], !llvm.loop [[LOOP27:![0-9]+]]
15000 // CHECK16: omp.inner.for.end29:
15001 // CHECK16-NEXT: store i32 100, i32* [[I20]], align 4
15002 // CHECK16-NEXT: ret i32 0
15003 //
15004