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 // expected-no-diagnostics
7 // RUN: %clang_cc1 -DCK1 -verify -Wno-vla -fopenmp -x c++ -triple powerpc64le-unknown-unknown -fopenmp-targets=powerpc64le-ibm-linux-gnu -emit-llvm %s -o - | FileCheck %s --check-prefix=CHECK1
8 // RUN: %clang_cc1 -DCK1 -fopenmp -x c++ -std=c++11 -triple powerpc64le-unknown-unknown -fopenmp-targets=powerpc64le-ibm-linux-gnu -emit-pch -o %t %s
9 // RUN: %clang_cc1 -DCK1 -fopenmp -x c++ -triple powerpc64le-unknown-unknown -fopenmp-targets=powerpc64le-ibm-linux-gnu -std=c++11 -include-pch %t -verify -Wno-vla %s -emit-llvm -o - | FileCheck %s --check-prefix=CHECK1
10 // RUN: %clang_cc1 -DCK1 -verify -Wno-vla -fopenmp -x c++ -triple i386-unknown-unknown -fopenmp-targets=i386-pc-linux-gnu -emit-llvm %s -o - | FileCheck %s --check-prefix=CHECK3
11 // RUN: %clang_cc1 -DCK1 -fopenmp -x c++ -std=c++11 -triple i386-unknown-unknown -fopenmp-targets=i386-pc-linux-gnu -emit-pch -o %t %s
12 // RUN: %clang_cc1 -DCK1 -fopenmp -x c++ -triple i386-unknown-unknown -fopenmp-targets=i386-pc-linux-gnu -std=c++11 -include-pch %t -verify -Wno-vla %s -emit-llvm -o - | FileCheck %s --check-prefix=CHECK3
14 // RUN: %clang_cc1 -DCK1 -verify -Wno-vla -fopenmp -fopenmp-version=45 -x c++ -triple powerpc64le-unknown-unknown -fopenmp-targets=powerpc64le-ibm-linux-gnu -emit-llvm %s -o - | FileCheck %s --check-prefix=CHECK5
15 // RUN: %clang_cc1 -DCK1 -fopenmp -fopenmp-version=45 -x c++ -std=c++11 -triple powerpc64le-unknown-unknown -fopenmp-targets=powerpc64le-ibm-linux-gnu -emit-pch -o %t %s
16 // RUN: %clang_cc1 -DCK1 -fopenmp -fopenmp-version=45 -x c++ -triple powerpc64le-unknown-unknown -fopenmp-targets=powerpc64le-ibm-linux-gnu -std=c++11 -include-pch %t -verify -Wno-vla %s -emit-llvm -o - | FileCheck %s --check-prefix=CHECK5
17 // RUN: %clang_cc1 -DCK1 -verify -Wno-vla -fopenmp -fopenmp-version=45 -x c++ -triple i386-unknown-unknown -fopenmp-targets=i386-pc-linux-gnu -emit-llvm %s -o - | FileCheck %s --check-prefix=CHECK7
18 // RUN: %clang_cc1 -DCK1 -fopenmp -fopenmp-version=45 -x c++ -std=c++11 -triple i386-unknown-unknown -fopenmp-targets=i386-pc-linux-gnu -emit-pch -o %t %s
19 // RUN: %clang_cc1 -DCK1 -fopenmp -fopenmp-version=45 -x c++ -triple i386-unknown-unknown -fopenmp-targets=i386-pc-linux-gnu -std=c++11 -include-pch %t -verify -Wno-vla %s -emit-llvm -o - | FileCheck %s --check-prefix=CHECK7
21 // RUN: %clang_cc1 -DCK1 -verify -Wno-vla -fopenmp-simd -x c++ -triple powerpc64le-unknown-unknown -fopenmp-targets=powerpc64le-ibm-linux-gnu -emit-llvm %s -o - | FileCheck %s --implicit-check-not="{{__kmpc|__tgt}}"
22 // RUN: %clang_cc1 -DCK1 -fopenmp-simd -x c++ -std=c++11 -triple powerpc64le-unknown-unknown -fopenmp-targets=powerpc64le-ibm-linux-gnu -emit-pch -o %t %s
23 // RUN: %clang_cc1 -DCK1 -fopenmp-simd -x c++ -triple powerpc64le-unknown-unknown -fopenmp-targets=powerpc64le-ibm-linux-gnu -std=c++11 -include-pch %t -verify -Wno-vla %s -emit-llvm -o - | FileCheck %s --implicit-check-not="{{__kmpc|__tgt}}"
24 // RUN: %clang_cc1 -DCK1 -verify -Wno-vla -fopenmp-simd -x c++ -triple i386-unknown-unknown -fopenmp-targets=i386-pc-linux-gnu -emit-llvm %s -o - | FileCheck %s --implicit-check-not="{{__kmpc|__tgt}}"
25 // RUN: %clang_cc1 -DCK1 -fopenmp-simd -x c++ -std=c++11 -triple i386-unknown-unknown -fopenmp-targets=i386-pc-linux-gnu -emit-pch -o %t %s
26 // RUN: %clang_cc1 -DCK1 -fopenmp-simd -x c++ -triple i386-unknown-unknown -fopenmp-targets=i386-pc-linux-gnu -std=c++11 -include-pch %t -verify -Wno-vla %s -emit-llvm -o - | FileCheck %s --implicit-check-not="{{__kmpc|__tgt}}"
29 template <typename T
, int X
, long long Y
>
36 #pragma omp teams distribute parallel for
37 for(int i
= 0; i
< X
; i
++) {
41 #pragma omp teams distribute parallel for schedule(static)
42 for(int i
= 0; i
< X
; i
++) {
46 #pragma omp teams distribute parallel for schedule(static, X/2)
47 for(int i
= 0; i
< X
; i
++) {
52 #pragma omp teams distribute parallel for schedule(dynamic)
53 for(int i
= 0; i
< X
; i
++) {
58 #pragma omp teams distribute parallel for schedule(dynamic, X/2)
59 for(int i
= 0; i
< X
; i
++) {
82 int teams_template_struct(void) {
90 // RUN: %clang_cc1 -DCK2 -verify -Wno-vla -fopenmp -x c++ -triple powerpc64le-unknown-unknown -fopenmp-targets=powerpc64le-ibm-linux-gnu -emit-llvm %s -o - | FileCheck %s --check-prefix=CHECK13
91 // RUN: %clang_cc1 -DCK2 -fopenmp -x c++ -std=c++11 -triple powerpc64le-unknown-unknown -fopenmp-targets=powerpc64le-ibm-linux-gnu -emit-pch -o %t %s
92 // RUN: %clang_cc1 -DCK2 -fopenmp -x c++ -triple powerpc64le-unknown-unknown -fopenmp-targets=powerpc64le-ibm-linux-gnu -std=c++11 -include-pch %t -verify -Wno-vla %s -emit-llvm -o - | FileCheck %s --check-prefix=CHECK13
93 // RUN: %clang_cc1 -DCK2 -verify -Wno-vla -fopenmp -x c++ -triple i386-unknown-unknown -fopenmp-targets=i386-pc-linux-gnu -emit-llvm %s -o - | FileCheck %s --check-prefix=CHECK15
94 // RUN: %clang_cc1 -DCK2 -fopenmp -x c++ -std=c++11 -triple i386-unknown-unknown -fopenmp-targets=i386-pc-linux-gnu -emit-pch -o %t %s
95 // RUN: %clang_cc1 -DCK2 -fopenmp -x c++ -triple i386-unknown-unknown -fopenmp-targets=i386-pc-linux-gnu -std=c++11 -include-pch %t -verify -Wno-vla %s -emit-llvm -o - | FileCheck %s --check-prefix=CHECK15
97 // RUN: %clang_cc1 -DCK2 -verify -Wno-vla -fopenmp -fopenmp-version=45 -x c++ -triple powerpc64le-unknown-unknown -fopenmp-targets=powerpc64le-ibm-linux-gnu -emit-llvm %s -o - | FileCheck %s --check-prefix=CHECK17
98 // RUN: %clang_cc1 -DCK2 -fopenmp -fopenmp-version=45 -x c++ -std=c++11 -triple powerpc64le-unknown-unknown -fopenmp-targets=powerpc64le-ibm-linux-gnu -emit-pch -o %t %s
99 // RUN: %clang_cc1 -DCK2 -fopenmp -fopenmp-version=45 -x c++ -triple powerpc64le-unknown-unknown -fopenmp-targets=powerpc64le-ibm-linux-gnu -std=c++11 -include-pch %t -verify -Wno-vla %s -emit-llvm -o - | FileCheck %s --check-prefix=CHECK17
100 // RUN: %clang_cc1 -DCK2 -verify -Wno-vla -fopenmp -fopenmp-version=45 -x c++ -triple i386-unknown-unknown -fopenmp-targets=i386-pc-linux-gnu -emit-llvm %s -o - | FileCheck %s --check-prefix=CHECK19
101 // RUN: %clang_cc1 -DCK2 -fopenmp -fopenmp-version=45 -x c++ -std=c++11 -triple i386-unknown-unknown -fopenmp-targets=i386-pc-linux-gnu -emit-pch -o %t %s
102 // RUN: %clang_cc1 -DCK2 -fopenmp -fopenmp-version=45 -x c++ -triple i386-unknown-unknown -fopenmp-targets=i386-pc-linux-gnu -std=c++11 -include-pch %t -verify -Wno-vla %s -emit-llvm -o - | FileCheck %s --check-prefix=CHECK19
104 // RUN: %clang_cc1 -DCK2 -verify -Wno-vla -fopenmp-simd -x c++ -triple powerpc64le-unknown-unknown -fopenmp-targets=powerpc64le-ibm-linux-gnu -emit-llvm %s -o - | FileCheck %s --implicit-check-not="{{__kmpc|__tgt}}"
105 // RUN: %clang_cc1 -DCK2 -fopenmp-simd -x c++ -std=c++11 -triple powerpc64le-unknown-unknown -fopenmp-targets=powerpc64le-ibm-linux-gnu -emit-pch -o %t %s
106 // RUN: %clang_cc1 -DCK2 -fopenmp-simd -x c++ -triple powerpc64le-unknown-unknown -fopenmp-targets=powerpc64le-ibm-linux-gnu -std=c++11 -include-pch %t -verify -Wno-vla %s -emit-llvm -o - | FileCheck %s --implicit-check-not="{{__kmpc|__tgt}}"
107 // RUN: %clang_cc1 -DCK2 -verify -Wno-vla -fopenmp-simd -x c++ -triple i386-unknown-unknown -fopenmp-targets=i386-pc-linux-gnu -emit-llvm %s -o - | FileCheck %s --implicit-check-not="{{__kmpc|__tgt}}"
108 // RUN: %clang_cc1 -DCK2 -fopenmp-simd -x c++ -std=c++11 -triple i386-unknown-unknown -fopenmp-targets=i386-pc-linux-gnu -emit-pch -o %t %s
109 // RUN: %clang_cc1 -DCK2 -fopenmp-simd -x c++ -triple i386-unknown-unknown -fopenmp-targets=i386-pc-linux-gnu -std=c++11 -include-pch %t -verify -Wno-vla %s -emit-llvm -o - | FileCheck %s --implicit-check-not="{{__kmpc|__tgt}}"
112 template <typename T
, int n
>
117 #pragma omp teams distribute parallel for
118 for(int i
= 0; i
< n
; i
++) {
122 #pragma omp teams distribute parallel for schedule(static)
123 for(int i
= 0; i
< n
; i
++) {
127 #pragma omp teams distribute parallel for schedule(static, m)
128 for(int i
= 0; i
< n
; i
++) {
132 #pragma omp teams distribute parallel for schedule(dynamic)
133 for(int i
= 0; i
< n
; i
++) {
137 #pragma omp teams distribute parallel for schedule(dynamic, m)
138 for(int i
= 0; i
< n
; i
++) {
144 int main (int argc
, char **argv
) {
149 #pragma omp teams distribute parallel for
150 for(int i
= 0; i
< n
; i
++) {
154 #pragma omp teams distribute parallel for dist_schedule(static)
155 for(int i
= 0; i
< n
; i
++) {
159 #pragma omp teams distribute parallel for dist_schedule(static, m)
160 for(int i
= 0; i
< n
; i
++) {
164 #pragma omp teams distribute parallel for schedule(dynamic)
165 for(int i
= 0; i
< n
; i
++) {
169 #pragma omp teams distribute parallel for schedule(dynamic, m)
170 for(int i
= 0; i
< n
; i
++) {
173 return tmain
<int, 10>(argc
);
210 #endif // #ifndef HEADER
211 // CHECK1-LABEL: define {{[^@]+}}@_Z21teams_template_structv
212 // CHECK1-SAME: () #[[ATTR0:[0-9]+]] {
213 // CHECK1-NEXT: entry:
214 // CHECK1-NEXT: [[V:%.*]] = alloca [[STRUCT_SS:%.*]], align 4
215 // CHECK1-NEXT: [[CALL:%.*]] = call noundef signext i32 @_ZN2SSIiLi123ELx456EE3fooEv(ptr noundef nonnull align 4 dereferenceable(496) [[V]])
216 // CHECK1-NEXT: ret i32 [[CALL]]
219 // CHECK1-LABEL: define {{[^@]+}}@_ZN2SSIiLi123ELx456EE3fooEv
220 // CHECK1-SAME: (ptr noundef nonnull align 4 dereferenceable(496) [[THIS:%.*]]) #[[ATTR0]] comdat {
221 // CHECK1-NEXT: entry:
222 // CHECK1-NEXT: [[THIS_ADDR:%.*]] = alloca ptr, align 8
223 // CHECK1-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [1 x ptr], align 8
224 // CHECK1-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [1 x ptr], align 8
225 // CHECK1-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [1 x ptr], align 8
226 // CHECK1-NEXT: [[TMP:%.*]] = alloca i32, align 4
227 // CHECK1-NEXT: [[KERNEL_ARGS:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS:%.*]], align 8
228 // CHECK1-NEXT: [[DOTOFFLOAD_BASEPTRS3:%.*]] = alloca [1 x ptr], align 8
229 // CHECK1-NEXT: [[DOTOFFLOAD_PTRS4:%.*]] = alloca [1 x ptr], align 8
230 // CHECK1-NEXT: [[DOTOFFLOAD_MAPPERS5:%.*]] = alloca [1 x ptr], align 8
231 // CHECK1-NEXT: [[_TMP6:%.*]] = alloca i32, align 4
232 // CHECK1-NEXT: [[KERNEL_ARGS7:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS]], align 8
233 // CHECK1-NEXT: [[DOTOFFLOAD_BASEPTRS11:%.*]] = alloca [1 x ptr], align 8
234 // CHECK1-NEXT: [[DOTOFFLOAD_PTRS12:%.*]] = alloca [1 x ptr], align 8
235 // CHECK1-NEXT: [[DOTOFFLOAD_MAPPERS13:%.*]] = alloca [1 x ptr], align 8
236 // CHECK1-NEXT: [[_TMP14:%.*]] = alloca i32, align 4
237 // CHECK1-NEXT: [[KERNEL_ARGS15:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS]], align 8
238 // CHECK1-NEXT: [[DOTOFFLOAD_BASEPTRS19:%.*]] = alloca [1 x ptr], align 8
239 // CHECK1-NEXT: [[DOTOFFLOAD_PTRS20:%.*]] = alloca [1 x ptr], align 8
240 // CHECK1-NEXT: [[DOTOFFLOAD_MAPPERS21:%.*]] = alloca [1 x ptr], align 8
241 // CHECK1-NEXT: [[_TMP22:%.*]] = alloca i32, align 4
242 // CHECK1-NEXT: [[KERNEL_ARGS23:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS]], align 8
243 // CHECK1-NEXT: [[DOTOFFLOAD_BASEPTRS27:%.*]] = alloca [1 x ptr], align 8
244 // CHECK1-NEXT: [[DOTOFFLOAD_PTRS28:%.*]] = alloca [1 x ptr], align 8
245 // CHECK1-NEXT: [[DOTOFFLOAD_MAPPERS29:%.*]] = alloca [1 x ptr], align 8
246 // CHECK1-NEXT: [[_TMP30:%.*]] = alloca i32, align 4
247 // CHECK1-NEXT: [[KERNEL_ARGS31:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS]], align 8
248 // CHECK1-NEXT: store ptr [[THIS]], ptr [[THIS_ADDR]], align 8
249 // CHECK1-NEXT: [[THIS1:%.*]] = load ptr, ptr [[THIS_ADDR]], align 8
250 // CHECK1-NEXT: [[A:%.*]] = getelementptr inbounds nuw [[STRUCT_SS:%.*]], ptr [[THIS1]], i32 0, i32 0
251 // CHECK1-NEXT: [[TMP0:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
252 // CHECK1-NEXT: store ptr [[THIS1]], ptr [[TMP0]], align 8
253 // CHECK1-NEXT: [[TMP1:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
254 // CHECK1-NEXT: store ptr [[A]], ptr [[TMP1]], align 8
255 // CHECK1-NEXT: [[TMP2:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0
256 // CHECK1-NEXT: store ptr null, ptr [[TMP2]], align 8
257 // CHECK1-NEXT: [[TMP3:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
258 // CHECK1-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
259 // CHECK1-NEXT: [[TMP5:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 0
260 // CHECK1-NEXT: store i32 3, ptr [[TMP5]], align 4
261 // CHECK1-NEXT: [[TMP6:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 1
262 // CHECK1-NEXT: store i32 1, ptr [[TMP6]], align 4
263 // CHECK1-NEXT: [[TMP7:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 2
264 // CHECK1-NEXT: store ptr [[TMP3]], ptr [[TMP7]], align 8
265 // CHECK1-NEXT: [[TMP8:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 3
266 // CHECK1-NEXT: store ptr [[TMP4]], ptr [[TMP8]], align 8
267 // CHECK1-NEXT: [[TMP9:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 4
268 // CHECK1-NEXT: store ptr @.offload_sizes, ptr [[TMP9]], align 8
269 // CHECK1-NEXT: [[TMP10:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 5
270 // CHECK1-NEXT: store ptr @.offload_maptypes, ptr [[TMP10]], align 8
271 // CHECK1-NEXT: [[TMP11:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 6
272 // CHECK1-NEXT: store ptr null, ptr [[TMP11]], align 8
273 // CHECK1-NEXT: [[TMP12:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 7
274 // CHECK1-NEXT: store ptr null, ptr [[TMP12]], align 8
275 // CHECK1-NEXT: [[TMP13:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 8
276 // CHECK1-NEXT: store i64 123, ptr [[TMP13]], align 8
277 // CHECK1-NEXT: [[TMP14:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 9
278 // CHECK1-NEXT: store i64 0, ptr [[TMP14]], align 8
279 // CHECK1-NEXT: [[TMP15:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 10
280 // CHECK1-NEXT: store [3 x i32] zeroinitializer, ptr [[TMP15]], align 4
281 // CHECK1-NEXT: [[TMP16:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 11
282 // CHECK1-NEXT: store [3 x i32] zeroinitializer, ptr [[TMP16]], align 4
283 // CHECK1-NEXT: [[TMP17:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 12
284 // CHECK1-NEXT: store i32 0, ptr [[TMP17]], align 4
285 // CHECK1-NEXT: [[TMP18:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB3:[0-9]+]], i64 -1, i32 0, i32 0, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l35.region_id, ptr [[KERNEL_ARGS]])
286 // CHECK1-NEXT: [[TMP19:%.*]] = icmp ne i32 [[TMP18]], 0
287 // CHECK1-NEXT: br i1 [[TMP19]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
288 // CHECK1: omp_offload.failed:
289 // CHECK1-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l35(ptr [[THIS1]]) #[[ATTR2:[0-9]+]]
290 // CHECK1-NEXT: br label [[OMP_OFFLOAD_CONT]]
291 // CHECK1: omp_offload.cont:
292 // CHECK1-NEXT: [[A2:%.*]] = getelementptr inbounds nuw [[STRUCT_SS]], ptr [[THIS1]], i32 0, i32 0
293 // CHECK1-NEXT: [[TMP20:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS3]], i32 0, i32 0
294 // CHECK1-NEXT: store ptr [[THIS1]], ptr [[TMP20]], align 8
295 // CHECK1-NEXT: [[TMP21:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS4]], i32 0, i32 0
296 // CHECK1-NEXT: store ptr [[A2]], ptr [[TMP21]], align 8
297 // CHECK1-NEXT: [[TMP22:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_MAPPERS5]], i64 0, i64 0
298 // CHECK1-NEXT: store ptr null, ptr [[TMP22]], align 8
299 // CHECK1-NEXT: [[TMP23:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS3]], i32 0, i32 0
300 // CHECK1-NEXT: [[TMP24:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS4]], i32 0, i32 0
301 // CHECK1-NEXT: [[TMP25:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS7]], i32 0, i32 0
302 // CHECK1-NEXT: store i32 3, ptr [[TMP25]], align 4
303 // CHECK1-NEXT: [[TMP26:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS7]], i32 0, i32 1
304 // CHECK1-NEXT: store i32 1, ptr [[TMP26]], align 4
305 // CHECK1-NEXT: [[TMP27:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS7]], i32 0, i32 2
306 // CHECK1-NEXT: store ptr [[TMP23]], ptr [[TMP27]], align 8
307 // CHECK1-NEXT: [[TMP28:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS7]], i32 0, i32 3
308 // CHECK1-NEXT: store ptr [[TMP24]], ptr [[TMP28]], align 8
309 // CHECK1-NEXT: [[TMP29:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS7]], i32 0, i32 4
310 // CHECK1-NEXT: store ptr @.offload_sizes.1, ptr [[TMP29]], align 8
311 // CHECK1-NEXT: [[TMP30:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS7]], i32 0, i32 5
312 // CHECK1-NEXT: store ptr @.offload_maptypes.2, ptr [[TMP30]], align 8
313 // CHECK1-NEXT: [[TMP31:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS7]], i32 0, i32 6
314 // CHECK1-NEXT: store ptr null, ptr [[TMP31]], align 8
315 // CHECK1-NEXT: [[TMP32:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS7]], i32 0, i32 7
316 // CHECK1-NEXT: store ptr null, ptr [[TMP32]], align 8
317 // CHECK1-NEXT: [[TMP33:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS7]], i32 0, i32 8
318 // CHECK1-NEXT: store i64 123, ptr [[TMP33]], align 8
319 // CHECK1-NEXT: [[TMP34:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS7]], i32 0, i32 9
320 // CHECK1-NEXT: store i64 0, ptr [[TMP34]], align 8
321 // CHECK1-NEXT: [[TMP35:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS7]], i32 0, i32 10
322 // CHECK1-NEXT: store [3 x i32] zeroinitializer, ptr [[TMP35]], align 4
323 // CHECK1-NEXT: [[TMP36:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS7]], i32 0, i32 11
324 // CHECK1-NEXT: store [3 x i32] zeroinitializer, ptr [[TMP36]], align 4
325 // CHECK1-NEXT: [[TMP37:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS7]], i32 0, i32 12
326 // CHECK1-NEXT: store i32 0, ptr [[TMP37]], align 4
327 // CHECK1-NEXT: [[TMP38:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB3]], i64 -1, i32 0, i32 0, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l40.region_id, ptr [[KERNEL_ARGS7]])
328 // CHECK1-NEXT: [[TMP39:%.*]] = icmp ne i32 [[TMP38]], 0
329 // CHECK1-NEXT: br i1 [[TMP39]], label [[OMP_OFFLOAD_FAILED8:%.*]], label [[OMP_OFFLOAD_CONT9:%.*]]
330 // CHECK1: omp_offload.failed8:
331 // CHECK1-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l40(ptr [[THIS1]]) #[[ATTR2]]
332 // CHECK1-NEXT: br label [[OMP_OFFLOAD_CONT9]]
333 // CHECK1: omp_offload.cont9:
334 // CHECK1-NEXT: [[A10:%.*]] = getelementptr inbounds nuw [[STRUCT_SS]], ptr [[THIS1]], i32 0, i32 0
335 // CHECK1-NEXT: [[TMP40:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS11]], i32 0, i32 0
336 // CHECK1-NEXT: store ptr [[THIS1]], ptr [[TMP40]], align 8
337 // CHECK1-NEXT: [[TMP41:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS12]], i32 0, i32 0
338 // CHECK1-NEXT: store ptr [[A10]], ptr [[TMP41]], align 8
339 // CHECK1-NEXT: [[TMP42:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_MAPPERS13]], i64 0, i64 0
340 // CHECK1-NEXT: store ptr null, ptr [[TMP42]], align 8
341 // CHECK1-NEXT: [[TMP43:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS11]], i32 0, i32 0
342 // CHECK1-NEXT: [[TMP44:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS12]], i32 0, i32 0
343 // CHECK1-NEXT: [[TMP45:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS15]], i32 0, i32 0
344 // CHECK1-NEXT: store i32 3, ptr [[TMP45]], align 4
345 // CHECK1-NEXT: [[TMP46:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS15]], i32 0, i32 1
346 // CHECK1-NEXT: store i32 1, ptr [[TMP46]], align 4
347 // CHECK1-NEXT: [[TMP47:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS15]], i32 0, i32 2
348 // CHECK1-NEXT: store ptr [[TMP43]], ptr [[TMP47]], align 8
349 // CHECK1-NEXT: [[TMP48:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS15]], i32 0, i32 3
350 // CHECK1-NEXT: store ptr [[TMP44]], ptr [[TMP48]], align 8
351 // CHECK1-NEXT: [[TMP49:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS15]], i32 0, i32 4
352 // CHECK1-NEXT: store ptr @.offload_sizes.3, ptr [[TMP49]], align 8
353 // CHECK1-NEXT: [[TMP50:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS15]], i32 0, i32 5
354 // CHECK1-NEXT: store ptr @.offload_maptypes.4, ptr [[TMP50]], align 8
355 // CHECK1-NEXT: [[TMP51:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS15]], i32 0, i32 6
356 // CHECK1-NEXT: store ptr null, ptr [[TMP51]], align 8
357 // CHECK1-NEXT: [[TMP52:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS15]], i32 0, i32 7
358 // CHECK1-NEXT: store ptr null, ptr [[TMP52]], align 8
359 // CHECK1-NEXT: [[TMP53:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS15]], i32 0, i32 8
360 // CHECK1-NEXT: store i64 123, ptr [[TMP53]], align 8
361 // CHECK1-NEXT: [[TMP54:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS15]], i32 0, i32 9
362 // CHECK1-NEXT: store i64 0, ptr [[TMP54]], align 8
363 // CHECK1-NEXT: [[TMP55:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS15]], i32 0, i32 10
364 // CHECK1-NEXT: store [3 x i32] zeroinitializer, ptr [[TMP55]], align 4
365 // CHECK1-NEXT: [[TMP56:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS15]], i32 0, i32 11
366 // CHECK1-NEXT: store [3 x i32] zeroinitializer, ptr [[TMP56]], align 4
367 // CHECK1-NEXT: [[TMP57:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS15]], i32 0, i32 12
368 // CHECK1-NEXT: store i32 0, ptr [[TMP57]], align 4
369 // CHECK1-NEXT: [[TMP58:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB3]], i64 -1, i32 0, i32 0, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l45.region_id, ptr [[KERNEL_ARGS15]])
370 // CHECK1-NEXT: [[TMP59:%.*]] = icmp ne i32 [[TMP58]], 0
371 // CHECK1-NEXT: br i1 [[TMP59]], label [[OMP_OFFLOAD_FAILED16:%.*]], label [[OMP_OFFLOAD_CONT17:%.*]]
372 // CHECK1: omp_offload.failed16:
373 // CHECK1-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l45(ptr [[THIS1]]) #[[ATTR2]]
374 // CHECK1-NEXT: br label [[OMP_OFFLOAD_CONT17]]
375 // CHECK1: omp_offload.cont17:
376 // CHECK1-NEXT: [[A18:%.*]] = getelementptr inbounds nuw [[STRUCT_SS]], ptr [[THIS1]], i32 0, i32 0
377 // CHECK1-NEXT: [[TMP60:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS19]], i32 0, i32 0
378 // CHECK1-NEXT: store ptr [[THIS1]], ptr [[TMP60]], align 8
379 // CHECK1-NEXT: [[TMP61:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS20]], i32 0, i32 0
380 // CHECK1-NEXT: store ptr [[A18]], ptr [[TMP61]], align 8
381 // CHECK1-NEXT: [[TMP62:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_MAPPERS21]], i64 0, i64 0
382 // CHECK1-NEXT: store ptr null, ptr [[TMP62]], align 8
383 // CHECK1-NEXT: [[TMP63:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS19]], i32 0, i32 0
384 // CHECK1-NEXT: [[TMP64:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS20]], i32 0, i32 0
385 // CHECK1-NEXT: [[TMP65:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS23]], i32 0, i32 0
386 // CHECK1-NEXT: store i32 3, ptr [[TMP65]], align 4
387 // CHECK1-NEXT: [[TMP66:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS23]], i32 0, i32 1
388 // CHECK1-NEXT: store i32 1, ptr [[TMP66]], align 4
389 // CHECK1-NEXT: [[TMP67:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS23]], i32 0, i32 2
390 // CHECK1-NEXT: store ptr [[TMP63]], ptr [[TMP67]], align 8
391 // CHECK1-NEXT: [[TMP68:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS23]], i32 0, i32 3
392 // CHECK1-NEXT: store ptr [[TMP64]], ptr [[TMP68]], align 8
393 // CHECK1-NEXT: [[TMP69:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS23]], i32 0, i32 4
394 // CHECK1-NEXT: store ptr @.offload_sizes.5, ptr [[TMP69]], align 8
395 // CHECK1-NEXT: [[TMP70:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS23]], i32 0, i32 5
396 // CHECK1-NEXT: store ptr @.offload_maptypes.6, ptr [[TMP70]], align 8
397 // CHECK1-NEXT: [[TMP71:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS23]], i32 0, i32 6
398 // CHECK1-NEXT: store ptr null, ptr [[TMP71]], align 8
399 // CHECK1-NEXT: [[TMP72:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS23]], i32 0, i32 7
400 // CHECK1-NEXT: store ptr null, ptr [[TMP72]], align 8
401 // CHECK1-NEXT: [[TMP73:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS23]], i32 0, i32 8
402 // CHECK1-NEXT: store i64 123, ptr [[TMP73]], align 8
403 // CHECK1-NEXT: [[TMP74:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS23]], i32 0, i32 9
404 // CHECK1-NEXT: store i64 0, ptr [[TMP74]], align 8
405 // CHECK1-NEXT: [[TMP75:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS23]], i32 0, i32 10
406 // CHECK1-NEXT: store [3 x i32] zeroinitializer, ptr [[TMP75]], align 4
407 // CHECK1-NEXT: [[TMP76:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS23]], i32 0, i32 11
408 // CHECK1-NEXT: store [3 x i32] zeroinitializer, ptr [[TMP76]], align 4
409 // CHECK1-NEXT: [[TMP77:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS23]], i32 0, i32 12
410 // CHECK1-NEXT: store i32 0, ptr [[TMP77]], align 4
411 // CHECK1-NEXT: [[TMP78:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB3]], i64 -1, i32 0, i32 0, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l51.region_id, ptr [[KERNEL_ARGS23]])
412 // CHECK1-NEXT: [[TMP79:%.*]] = icmp ne i32 [[TMP78]], 0
413 // CHECK1-NEXT: br i1 [[TMP79]], label [[OMP_OFFLOAD_FAILED24:%.*]], label [[OMP_OFFLOAD_CONT25:%.*]]
414 // CHECK1: omp_offload.failed24:
415 // CHECK1-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l51(ptr [[THIS1]]) #[[ATTR2]]
416 // CHECK1-NEXT: br label [[OMP_OFFLOAD_CONT25]]
417 // CHECK1: omp_offload.cont25:
418 // CHECK1-NEXT: [[A26:%.*]] = getelementptr inbounds nuw [[STRUCT_SS]], ptr [[THIS1]], i32 0, i32 0
419 // CHECK1-NEXT: [[TMP80:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS27]], i32 0, i32 0
420 // CHECK1-NEXT: store ptr [[THIS1]], ptr [[TMP80]], align 8
421 // CHECK1-NEXT: [[TMP81:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS28]], i32 0, i32 0
422 // CHECK1-NEXT: store ptr [[A26]], ptr [[TMP81]], align 8
423 // CHECK1-NEXT: [[TMP82:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_MAPPERS29]], i64 0, i64 0
424 // CHECK1-NEXT: store ptr null, ptr [[TMP82]], align 8
425 // CHECK1-NEXT: [[TMP83:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS27]], i32 0, i32 0
426 // CHECK1-NEXT: [[TMP84:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS28]], i32 0, i32 0
427 // CHECK1-NEXT: [[TMP85:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS31]], i32 0, i32 0
428 // CHECK1-NEXT: store i32 3, ptr [[TMP85]], align 4
429 // CHECK1-NEXT: [[TMP86:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS31]], i32 0, i32 1
430 // CHECK1-NEXT: store i32 1, ptr [[TMP86]], align 4
431 // CHECK1-NEXT: [[TMP87:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS31]], i32 0, i32 2
432 // CHECK1-NEXT: store ptr [[TMP83]], ptr [[TMP87]], align 8
433 // CHECK1-NEXT: [[TMP88:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS31]], i32 0, i32 3
434 // CHECK1-NEXT: store ptr [[TMP84]], ptr [[TMP88]], align 8
435 // CHECK1-NEXT: [[TMP89:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS31]], i32 0, i32 4
436 // CHECK1-NEXT: store ptr @.offload_sizes.7, ptr [[TMP89]], align 8
437 // CHECK1-NEXT: [[TMP90:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS31]], i32 0, i32 5
438 // CHECK1-NEXT: store ptr @.offload_maptypes.8, ptr [[TMP90]], align 8
439 // CHECK1-NEXT: [[TMP91:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS31]], i32 0, i32 6
440 // CHECK1-NEXT: store ptr null, ptr [[TMP91]], align 8
441 // CHECK1-NEXT: [[TMP92:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS31]], i32 0, i32 7
442 // CHECK1-NEXT: store ptr null, ptr [[TMP92]], align 8
443 // CHECK1-NEXT: [[TMP93:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS31]], i32 0, i32 8
444 // CHECK1-NEXT: store i64 123, ptr [[TMP93]], align 8
445 // CHECK1-NEXT: [[TMP94:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS31]], i32 0, i32 9
446 // CHECK1-NEXT: store i64 0, ptr [[TMP94]], align 8
447 // CHECK1-NEXT: [[TMP95:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS31]], i32 0, i32 10
448 // CHECK1-NEXT: store [3 x i32] zeroinitializer, ptr [[TMP95]], align 4
449 // CHECK1-NEXT: [[TMP96:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS31]], i32 0, i32 11
450 // CHECK1-NEXT: store [3 x i32] zeroinitializer, ptr [[TMP96]], align 4
451 // CHECK1-NEXT: [[TMP97:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS31]], i32 0, i32 12
452 // CHECK1-NEXT: store i32 0, ptr [[TMP97]], align 4
453 // CHECK1-NEXT: [[TMP98:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB3]], i64 -1, i32 0, i32 0, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l57.region_id, ptr [[KERNEL_ARGS31]])
454 // CHECK1-NEXT: [[TMP99:%.*]] = icmp ne i32 [[TMP98]], 0
455 // CHECK1-NEXT: br i1 [[TMP99]], label [[OMP_OFFLOAD_FAILED32:%.*]], label [[OMP_OFFLOAD_CONT33:%.*]]
456 // CHECK1: omp_offload.failed32:
457 // CHECK1-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l57(ptr [[THIS1]]) #[[ATTR2]]
458 // CHECK1-NEXT: br label [[OMP_OFFLOAD_CONT33]]
459 // CHECK1: omp_offload.cont33:
460 // CHECK1-NEXT: [[A34:%.*]] = getelementptr inbounds nuw [[STRUCT_SS]], ptr [[THIS1]], i32 0, i32 0
461 // CHECK1-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds [123 x i32], ptr [[A34]], i64 0, i64 0
462 // CHECK1-NEXT: [[TMP100:%.*]] = load i32, ptr [[ARRAYIDX]], align 4
463 // CHECK1-NEXT: ret i32 [[TMP100]]
466 // CHECK1-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l35
467 // CHECK1-SAME: (ptr noundef [[THIS:%.*]]) #[[ATTR1:[0-9]+]] {
468 // CHECK1-NEXT: entry:
469 // CHECK1-NEXT: [[THIS_ADDR:%.*]] = alloca ptr, align 8
470 // CHECK1-NEXT: store ptr [[THIS]], ptr [[THIS_ADDR]], align 8
471 // CHECK1-NEXT: [[TMP0:%.*]] = load ptr, ptr [[THIS_ADDR]], align 8
472 // CHECK1-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_teams(ptr @[[GLOB3]], i32 1, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l35.omp_outlined, ptr [[TMP0]])
473 // CHECK1-NEXT: ret void
476 // CHECK1-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l35.omp_outlined
477 // CHECK1-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], ptr noundef [[THIS:%.*]]) #[[ATTR1]] {
478 // CHECK1-NEXT: entry:
479 // CHECK1-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 8
480 // CHECK1-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 8
481 // CHECK1-NEXT: [[THIS_ADDR:%.*]] = alloca ptr, align 8
482 // CHECK1-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
483 // CHECK1-NEXT: [[TMP:%.*]] = alloca i32, align 4
484 // CHECK1-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4
485 // CHECK1-NEXT: [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4
486 // CHECK1-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
487 // CHECK1-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
488 // CHECK1-NEXT: [[I:%.*]] = alloca i32, align 4
489 // CHECK1-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 8
490 // CHECK1-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 8
491 // CHECK1-NEXT: store ptr [[THIS]], ptr [[THIS_ADDR]], align 8
492 // CHECK1-NEXT: [[TMP0:%.*]] = load ptr, ptr [[THIS_ADDR]], align 8
493 // CHECK1-NEXT: store i32 0, ptr [[DOTOMP_COMB_LB]], align 4
494 // CHECK1-NEXT: store i32 122, ptr [[DOTOMP_COMB_UB]], align 4
495 // CHECK1-NEXT: store i32 1, ptr [[DOTOMP_STRIDE]], align 4
496 // CHECK1-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
497 // CHECK1-NEXT: [[TMP1:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
498 // CHECK1-NEXT: [[TMP2:%.*]] = load i32, ptr [[TMP1]], align 4
499 // CHECK1-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB1:[0-9]+]], i32 [[TMP2]], i32 92, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
500 // CHECK1-NEXT: [[TMP3:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
501 // CHECK1-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP3]], 122
502 // CHECK1-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
503 // CHECK1: cond.true:
504 // CHECK1-NEXT: br label [[COND_END:%.*]]
505 // CHECK1: cond.false:
506 // CHECK1-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
507 // CHECK1-NEXT: br label [[COND_END]]
509 // CHECK1-NEXT: [[COND:%.*]] = phi i32 [ 122, [[COND_TRUE]] ], [ [[TMP4]], [[COND_FALSE]] ]
510 // CHECK1-NEXT: store i32 [[COND]], ptr [[DOTOMP_COMB_UB]], align 4
511 // CHECK1-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
512 // CHECK1-NEXT: store i32 [[TMP5]], ptr [[DOTOMP_IV]], align 4
513 // CHECK1-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
514 // CHECK1: omp.inner.for.cond:
515 // CHECK1-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
516 // CHECK1-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
517 // CHECK1-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP6]], [[TMP7]]
518 // CHECK1-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
519 // CHECK1: omp.inner.for.body:
520 // CHECK1-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
521 // CHECK1-NEXT: [[TMP9:%.*]] = zext i32 [[TMP8]] to i64
522 // CHECK1-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
523 // CHECK1-NEXT: [[TMP11:%.*]] = zext i32 [[TMP10]] to i64
524 // CHECK1-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_call(ptr @[[GLOB3]], i32 3, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l35.omp_outlined.omp_outlined, i64 [[TMP9]], i64 [[TMP11]], ptr [[TMP0]])
525 // CHECK1-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
526 // CHECK1: omp.inner.for.inc:
527 // CHECK1-NEXT: [[TMP12:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
528 // CHECK1-NEXT: [[TMP13:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4
529 // CHECK1-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP12]], [[TMP13]]
530 // CHECK1-NEXT: store i32 [[ADD]], ptr [[DOTOMP_IV]], align 4
531 // CHECK1-NEXT: br label [[OMP_INNER_FOR_COND]]
532 // CHECK1: omp.inner.for.end:
533 // CHECK1-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
534 // CHECK1: omp.loop.exit:
535 // CHECK1-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB1]], i32 [[TMP2]])
536 // CHECK1-NEXT: ret void
539 // CHECK1-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l35.omp_outlined.omp_outlined
540 // CHECK1-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], i64 noundef [[DOTPREVIOUS_LB_:%.*]], i64 noundef [[DOTPREVIOUS_UB_:%.*]], ptr noundef [[THIS:%.*]]) #[[ATTR1]] {
541 // CHECK1-NEXT: entry:
542 // CHECK1-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 8
543 // CHECK1-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 8
544 // CHECK1-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i64, align 8
545 // CHECK1-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i64, align 8
546 // CHECK1-NEXT: [[THIS_ADDR:%.*]] = alloca ptr, align 8
547 // CHECK1-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
548 // CHECK1-NEXT: [[TMP:%.*]] = alloca i32, align 4
549 // CHECK1-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4
550 // CHECK1-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4
551 // CHECK1-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
552 // CHECK1-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
553 // CHECK1-NEXT: [[I:%.*]] = alloca i32, align 4
554 // CHECK1-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 8
555 // CHECK1-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 8
556 // CHECK1-NEXT: store i64 [[DOTPREVIOUS_LB_]], ptr [[DOTPREVIOUS_LB__ADDR]], align 8
557 // CHECK1-NEXT: store i64 [[DOTPREVIOUS_UB_]], ptr [[DOTPREVIOUS_UB__ADDR]], align 8
558 // CHECK1-NEXT: store ptr [[THIS]], ptr [[THIS_ADDR]], align 8
559 // CHECK1-NEXT: [[TMP0:%.*]] = load ptr, ptr [[THIS_ADDR]], align 8
560 // CHECK1-NEXT: store i32 0, ptr [[DOTOMP_LB]], align 4
561 // CHECK1-NEXT: store i32 122, ptr [[DOTOMP_UB]], align 4
562 // CHECK1-NEXT: [[TMP1:%.*]] = load i64, ptr [[DOTPREVIOUS_LB__ADDR]], align 8
563 // CHECK1-NEXT: [[CONV:%.*]] = trunc i64 [[TMP1]] to i32
564 // CHECK1-NEXT: [[TMP2:%.*]] = load i64, ptr [[DOTPREVIOUS_UB__ADDR]], align 8
565 // CHECK1-NEXT: [[CONV1:%.*]] = trunc i64 [[TMP2]] to i32
566 // CHECK1-NEXT: store i32 [[CONV]], ptr [[DOTOMP_LB]], align 4
567 // CHECK1-NEXT: store i32 [[CONV1]], ptr [[DOTOMP_UB]], align 4
568 // CHECK1-NEXT: store i32 1, ptr [[DOTOMP_STRIDE]], align 4
569 // CHECK1-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
570 // CHECK1-NEXT: [[TMP3:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
571 // CHECK1-NEXT: [[TMP4:%.*]] = load i32, ptr [[TMP3]], align 4
572 // CHECK1-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB2:[0-9]+]], i32 [[TMP4]], i32 34, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
573 // CHECK1-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
574 // CHECK1-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP5]], 122
575 // CHECK1-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
576 // CHECK1: cond.true:
577 // CHECK1-NEXT: br label [[COND_END:%.*]]
578 // CHECK1: cond.false:
579 // CHECK1-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
580 // CHECK1-NEXT: br label [[COND_END]]
582 // CHECK1-NEXT: [[COND:%.*]] = phi i32 [ 122, [[COND_TRUE]] ], [ [[TMP6]], [[COND_FALSE]] ]
583 // CHECK1-NEXT: store i32 [[COND]], ptr [[DOTOMP_UB]], align 4
584 // CHECK1-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4
585 // CHECK1-NEXT: store i32 [[TMP7]], ptr [[DOTOMP_IV]], align 4
586 // CHECK1-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
587 // CHECK1: omp.inner.for.cond:
588 // CHECK1-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
589 // CHECK1-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
590 // CHECK1-NEXT: [[CMP2:%.*]] = icmp sle i32 [[TMP8]], [[TMP9]]
591 // CHECK1-NEXT: br i1 [[CMP2]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
592 // CHECK1: omp.inner.for.body:
593 // CHECK1-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
594 // CHECK1-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP10]], 1
595 // CHECK1-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
596 // CHECK1-NEXT: store i32 [[ADD]], ptr [[I]], align 4
597 // CHECK1-NEXT: [[A:%.*]] = getelementptr inbounds nuw [[STRUCT_SS:%.*]], ptr [[TMP0]], i32 0, i32 0
598 // CHECK1-NEXT: [[TMP11:%.*]] = load i32, ptr [[I]], align 4
599 // CHECK1-NEXT: [[IDXPROM:%.*]] = sext i32 [[TMP11]] to i64
600 // CHECK1-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds [123 x i32], ptr [[A]], i64 0, i64 [[IDXPROM]]
601 // CHECK1-NEXT: store i32 0, ptr [[ARRAYIDX]], align 4
602 // CHECK1-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
603 // CHECK1: omp.body.continue:
604 // CHECK1-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
605 // CHECK1: omp.inner.for.inc:
606 // CHECK1-NEXT: [[TMP12:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
607 // CHECK1-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP12]], 1
608 // CHECK1-NEXT: store i32 [[ADD3]], ptr [[DOTOMP_IV]], align 4
609 // CHECK1-NEXT: br label [[OMP_INNER_FOR_COND]]
610 // CHECK1: omp.inner.for.end:
611 // CHECK1-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
612 // CHECK1: omp.loop.exit:
613 // CHECK1-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB2]], i32 [[TMP4]])
614 // CHECK1-NEXT: ret void
617 // CHECK1-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l40
618 // CHECK1-SAME: (ptr noundef [[THIS:%.*]]) #[[ATTR1]] {
619 // CHECK1-NEXT: entry:
620 // CHECK1-NEXT: [[THIS_ADDR:%.*]] = alloca ptr, align 8
621 // CHECK1-NEXT: store ptr [[THIS]], ptr [[THIS_ADDR]], align 8
622 // CHECK1-NEXT: [[TMP0:%.*]] = load ptr, ptr [[THIS_ADDR]], align 8
623 // CHECK1-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_teams(ptr @[[GLOB3]], i32 1, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l40.omp_outlined, ptr [[TMP0]])
624 // CHECK1-NEXT: ret void
627 // CHECK1-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l40.omp_outlined
628 // CHECK1-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], ptr noundef [[THIS:%.*]]) #[[ATTR1]] {
629 // CHECK1-NEXT: entry:
630 // CHECK1-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 8
631 // CHECK1-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 8
632 // CHECK1-NEXT: [[THIS_ADDR:%.*]] = alloca ptr, align 8
633 // CHECK1-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
634 // CHECK1-NEXT: [[TMP:%.*]] = alloca i32, align 4
635 // CHECK1-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4
636 // CHECK1-NEXT: [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4
637 // CHECK1-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
638 // CHECK1-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
639 // CHECK1-NEXT: [[I:%.*]] = alloca i32, align 4
640 // CHECK1-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 8
641 // CHECK1-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 8
642 // CHECK1-NEXT: store ptr [[THIS]], ptr [[THIS_ADDR]], align 8
643 // CHECK1-NEXT: [[TMP0:%.*]] = load ptr, ptr [[THIS_ADDR]], align 8
644 // CHECK1-NEXT: store i32 0, ptr [[DOTOMP_COMB_LB]], align 4
645 // CHECK1-NEXT: store i32 122, ptr [[DOTOMP_COMB_UB]], align 4
646 // CHECK1-NEXT: store i32 1, ptr [[DOTOMP_STRIDE]], align 4
647 // CHECK1-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
648 // CHECK1-NEXT: [[TMP1:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
649 // CHECK1-NEXT: [[TMP2:%.*]] = load i32, ptr [[TMP1]], align 4
650 // CHECK1-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB1]], i32 [[TMP2]], i32 92, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
651 // CHECK1-NEXT: [[TMP3:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
652 // CHECK1-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP3]], 122
653 // CHECK1-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
654 // CHECK1: cond.true:
655 // CHECK1-NEXT: br label [[COND_END:%.*]]
656 // CHECK1: cond.false:
657 // CHECK1-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
658 // CHECK1-NEXT: br label [[COND_END]]
660 // CHECK1-NEXT: [[COND:%.*]] = phi i32 [ 122, [[COND_TRUE]] ], [ [[TMP4]], [[COND_FALSE]] ]
661 // CHECK1-NEXT: store i32 [[COND]], ptr [[DOTOMP_COMB_UB]], align 4
662 // CHECK1-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
663 // CHECK1-NEXT: store i32 [[TMP5]], ptr [[DOTOMP_IV]], align 4
664 // CHECK1-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
665 // CHECK1: omp.inner.for.cond:
666 // CHECK1-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
667 // CHECK1-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
668 // CHECK1-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP6]], [[TMP7]]
669 // CHECK1-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
670 // CHECK1: omp.inner.for.body:
671 // CHECK1-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
672 // CHECK1-NEXT: [[TMP9:%.*]] = zext i32 [[TMP8]] to i64
673 // CHECK1-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
674 // CHECK1-NEXT: [[TMP11:%.*]] = zext i32 [[TMP10]] to i64
675 // CHECK1-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_call(ptr @[[GLOB3]], i32 3, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l40.omp_outlined.omp_outlined, i64 [[TMP9]], i64 [[TMP11]], ptr [[TMP0]])
676 // CHECK1-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
677 // CHECK1: omp.inner.for.inc:
678 // CHECK1-NEXT: [[TMP12:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
679 // CHECK1-NEXT: [[TMP13:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4
680 // CHECK1-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP12]], [[TMP13]]
681 // CHECK1-NEXT: store i32 [[ADD]], ptr [[DOTOMP_IV]], align 4
682 // CHECK1-NEXT: br label [[OMP_INNER_FOR_COND]]
683 // CHECK1: omp.inner.for.end:
684 // CHECK1-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
685 // CHECK1: omp.loop.exit:
686 // CHECK1-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB1]], i32 [[TMP2]])
687 // CHECK1-NEXT: ret void
690 // CHECK1-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l40.omp_outlined.omp_outlined
691 // CHECK1-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], i64 noundef [[DOTPREVIOUS_LB_:%.*]], i64 noundef [[DOTPREVIOUS_UB_:%.*]], ptr noundef [[THIS:%.*]]) #[[ATTR1]] {
692 // CHECK1-NEXT: entry:
693 // CHECK1-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 8
694 // CHECK1-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 8
695 // CHECK1-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i64, align 8
696 // CHECK1-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i64, align 8
697 // CHECK1-NEXT: [[THIS_ADDR:%.*]] = alloca ptr, align 8
698 // CHECK1-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
699 // CHECK1-NEXT: [[TMP:%.*]] = alloca i32, align 4
700 // CHECK1-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4
701 // CHECK1-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4
702 // CHECK1-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
703 // CHECK1-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
704 // CHECK1-NEXT: [[I:%.*]] = alloca i32, align 4
705 // CHECK1-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 8
706 // CHECK1-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 8
707 // CHECK1-NEXT: store i64 [[DOTPREVIOUS_LB_]], ptr [[DOTPREVIOUS_LB__ADDR]], align 8
708 // CHECK1-NEXT: store i64 [[DOTPREVIOUS_UB_]], ptr [[DOTPREVIOUS_UB__ADDR]], align 8
709 // CHECK1-NEXT: store ptr [[THIS]], ptr [[THIS_ADDR]], align 8
710 // CHECK1-NEXT: [[TMP0:%.*]] = load ptr, ptr [[THIS_ADDR]], align 8
711 // CHECK1-NEXT: store i32 0, ptr [[DOTOMP_LB]], align 4
712 // CHECK1-NEXT: store i32 122, ptr [[DOTOMP_UB]], align 4
713 // CHECK1-NEXT: [[TMP1:%.*]] = load i64, ptr [[DOTPREVIOUS_LB__ADDR]], align 8
714 // CHECK1-NEXT: [[CONV:%.*]] = trunc i64 [[TMP1]] to i32
715 // CHECK1-NEXT: [[TMP2:%.*]] = load i64, ptr [[DOTPREVIOUS_UB__ADDR]], align 8
716 // CHECK1-NEXT: [[CONV1:%.*]] = trunc i64 [[TMP2]] to i32
717 // CHECK1-NEXT: store i32 [[CONV]], ptr [[DOTOMP_LB]], align 4
718 // CHECK1-NEXT: store i32 [[CONV1]], ptr [[DOTOMP_UB]], align 4
719 // CHECK1-NEXT: store i32 1, ptr [[DOTOMP_STRIDE]], align 4
720 // CHECK1-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
721 // CHECK1-NEXT: [[TMP3:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
722 // CHECK1-NEXT: [[TMP4:%.*]] = load i32, ptr [[TMP3]], align 4
723 // CHECK1-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB2]], i32 [[TMP4]], i32 34, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
724 // CHECK1-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
725 // CHECK1-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP5]], 122
726 // CHECK1-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
727 // CHECK1: cond.true:
728 // CHECK1-NEXT: br label [[COND_END:%.*]]
729 // CHECK1: cond.false:
730 // CHECK1-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
731 // CHECK1-NEXT: br label [[COND_END]]
733 // CHECK1-NEXT: [[COND:%.*]] = phi i32 [ 122, [[COND_TRUE]] ], [ [[TMP6]], [[COND_FALSE]] ]
734 // CHECK1-NEXT: store i32 [[COND]], ptr [[DOTOMP_UB]], align 4
735 // CHECK1-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4
736 // CHECK1-NEXT: store i32 [[TMP7]], ptr [[DOTOMP_IV]], align 4
737 // CHECK1-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
738 // CHECK1: omp.inner.for.cond:
739 // CHECK1-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
740 // CHECK1-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
741 // CHECK1-NEXT: [[CMP2:%.*]] = icmp sle i32 [[TMP8]], [[TMP9]]
742 // CHECK1-NEXT: br i1 [[CMP2]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
743 // CHECK1: omp.inner.for.body:
744 // CHECK1-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
745 // CHECK1-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP10]], 1
746 // CHECK1-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
747 // CHECK1-NEXT: store i32 [[ADD]], ptr [[I]], align 4
748 // CHECK1-NEXT: [[A:%.*]] = getelementptr inbounds nuw [[STRUCT_SS:%.*]], ptr [[TMP0]], i32 0, i32 0
749 // CHECK1-NEXT: [[TMP11:%.*]] = load i32, ptr [[I]], align 4
750 // CHECK1-NEXT: [[IDXPROM:%.*]] = sext i32 [[TMP11]] to i64
751 // CHECK1-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds [123 x i32], ptr [[A]], i64 0, i64 [[IDXPROM]]
752 // CHECK1-NEXT: store i32 0, ptr [[ARRAYIDX]], align 4
753 // CHECK1-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
754 // CHECK1: omp.body.continue:
755 // CHECK1-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
756 // CHECK1: omp.inner.for.inc:
757 // CHECK1-NEXT: [[TMP12:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
758 // CHECK1-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP12]], 1
759 // CHECK1-NEXT: store i32 [[ADD3]], ptr [[DOTOMP_IV]], align 4
760 // CHECK1-NEXT: br label [[OMP_INNER_FOR_COND]]
761 // CHECK1: omp.inner.for.end:
762 // CHECK1-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
763 // CHECK1: omp.loop.exit:
764 // CHECK1-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB2]], i32 [[TMP4]])
765 // CHECK1-NEXT: ret void
768 // CHECK1-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l45
769 // CHECK1-SAME: (ptr noundef [[THIS:%.*]]) #[[ATTR1]] {
770 // CHECK1-NEXT: entry:
771 // CHECK1-NEXT: [[THIS_ADDR:%.*]] = alloca ptr, align 8
772 // CHECK1-NEXT: store ptr [[THIS]], ptr [[THIS_ADDR]], align 8
773 // CHECK1-NEXT: [[TMP0:%.*]] = load ptr, ptr [[THIS_ADDR]], align 8
774 // CHECK1-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_teams(ptr @[[GLOB3]], i32 1, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l45.omp_outlined, ptr [[TMP0]])
775 // CHECK1-NEXT: ret void
778 // CHECK1-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l45.omp_outlined
779 // CHECK1-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], ptr noundef [[THIS:%.*]]) #[[ATTR1]] {
780 // CHECK1-NEXT: entry:
781 // CHECK1-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 8
782 // CHECK1-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 8
783 // CHECK1-NEXT: [[THIS_ADDR:%.*]] = alloca ptr, align 8
784 // CHECK1-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
785 // CHECK1-NEXT: [[TMP:%.*]] = alloca i32, align 4
786 // CHECK1-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4
787 // CHECK1-NEXT: [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4
788 // CHECK1-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
789 // CHECK1-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
790 // CHECK1-NEXT: [[I:%.*]] = alloca i32, align 4
791 // CHECK1-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 8
792 // CHECK1-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 8
793 // CHECK1-NEXT: store ptr [[THIS]], ptr [[THIS_ADDR]], align 8
794 // CHECK1-NEXT: [[TMP0:%.*]] = load ptr, ptr [[THIS_ADDR]], align 8
795 // CHECK1-NEXT: store i32 0, ptr [[DOTOMP_COMB_LB]], align 4
796 // CHECK1-NEXT: store i32 122, ptr [[DOTOMP_COMB_UB]], align 4
797 // CHECK1-NEXT: store i32 1, ptr [[DOTOMP_STRIDE]], align 4
798 // CHECK1-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
799 // CHECK1-NEXT: [[TMP1:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
800 // CHECK1-NEXT: [[TMP2:%.*]] = load i32, ptr [[TMP1]], align 4
801 // CHECK1-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB1]], i32 [[TMP2]], i32 92, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
802 // CHECK1-NEXT: [[TMP3:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
803 // CHECK1-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP3]], 122
804 // CHECK1-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
805 // CHECK1: cond.true:
806 // CHECK1-NEXT: br label [[COND_END:%.*]]
807 // CHECK1: cond.false:
808 // CHECK1-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
809 // CHECK1-NEXT: br label [[COND_END]]
811 // CHECK1-NEXT: [[COND:%.*]] = phi i32 [ 122, [[COND_TRUE]] ], [ [[TMP4]], [[COND_FALSE]] ]
812 // CHECK1-NEXT: store i32 [[COND]], ptr [[DOTOMP_COMB_UB]], align 4
813 // CHECK1-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
814 // CHECK1-NEXT: store i32 [[TMP5]], ptr [[DOTOMP_IV]], align 4
815 // CHECK1-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
816 // CHECK1: omp.inner.for.cond:
817 // CHECK1-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
818 // CHECK1-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
819 // CHECK1-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP6]], [[TMP7]]
820 // CHECK1-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
821 // CHECK1: omp.inner.for.body:
822 // CHECK1-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
823 // CHECK1-NEXT: [[TMP9:%.*]] = zext i32 [[TMP8]] to i64
824 // CHECK1-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
825 // CHECK1-NEXT: [[TMP11:%.*]] = zext i32 [[TMP10]] to i64
826 // CHECK1-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_call(ptr @[[GLOB3]], i32 3, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l45.omp_outlined.omp_outlined, i64 [[TMP9]], i64 [[TMP11]], ptr [[TMP0]])
827 // CHECK1-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
828 // CHECK1: omp.inner.for.inc:
829 // CHECK1-NEXT: [[TMP12:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
830 // CHECK1-NEXT: [[TMP13:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4
831 // CHECK1-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP12]], [[TMP13]]
832 // CHECK1-NEXT: store i32 [[ADD]], ptr [[DOTOMP_IV]], align 4
833 // CHECK1-NEXT: br label [[OMP_INNER_FOR_COND]]
834 // CHECK1: omp.inner.for.end:
835 // CHECK1-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
836 // CHECK1: omp.loop.exit:
837 // CHECK1-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB1]], i32 [[TMP2]])
838 // CHECK1-NEXT: ret void
841 // CHECK1-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l45.omp_outlined.omp_outlined
842 // CHECK1-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], i64 noundef [[DOTPREVIOUS_LB_:%.*]], i64 noundef [[DOTPREVIOUS_UB_:%.*]], ptr noundef [[THIS:%.*]]) #[[ATTR1]] {
843 // CHECK1-NEXT: entry:
844 // CHECK1-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 8
845 // CHECK1-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 8
846 // CHECK1-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i64, align 8
847 // CHECK1-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i64, align 8
848 // CHECK1-NEXT: [[THIS_ADDR:%.*]] = alloca ptr, align 8
849 // CHECK1-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
850 // CHECK1-NEXT: [[TMP:%.*]] = alloca i32, align 4
851 // CHECK1-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4
852 // CHECK1-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4
853 // CHECK1-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
854 // CHECK1-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
855 // CHECK1-NEXT: [[I:%.*]] = alloca i32, align 4
856 // CHECK1-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 8
857 // CHECK1-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 8
858 // CHECK1-NEXT: store i64 [[DOTPREVIOUS_LB_]], ptr [[DOTPREVIOUS_LB__ADDR]], align 8
859 // CHECK1-NEXT: store i64 [[DOTPREVIOUS_UB_]], ptr [[DOTPREVIOUS_UB__ADDR]], align 8
860 // CHECK1-NEXT: store ptr [[THIS]], ptr [[THIS_ADDR]], align 8
861 // CHECK1-NEXT: [[TMP0:%.*]] = load ptr, ptr [[THIS_ADDR]], align 8
862 // CHECK1-NEXT: store i32 0, ptr [[DOTOMP_LB]], align 4
863 // CHECK1-NEXT: store i32 122, ptr [[DOTOMP_UB]], align 4
864 // CHECK1-NEXT: [[TMP1:%.*]] = load i64, ptr [[DOTPREVIOUS_LB__ADDR]], align 8
865 // CHECK1-NEXT: [[CONV:%.*]] = trunc i64 [[TMP1]] to i32
866 // CHECK1-NEXT: [[TMP2:%.*]] = load i64, ptr [[DOTPREVIOUS_UB__ADDR]], align 8
867 // CHECK1-NEXT: [[CONV1:%.*]] = trunc i64 [[TMP2]] to i32
868 // CHECK1-NEXT: store i32 [[CONV]], ptr [[DOTOMP_LB]], align 4
869 // CHECK1-NEXT: store i32 [[CONV1]], ptr [[DOTOMP_UB]], align 4
870 // CHECK1-NEXT: store i32 1, ptr [[DOTOMP_STRIDE]], align 4
871 // CHECK1-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
872 // CHECK1-NEXT: [[TMP3:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
873 // CHECK1-NEXT: [[TMP4:%.*]] = load i32, ptr [[TMP3]], align 4
874 // CHECK1-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB2]], i32 [[TMP4]], i32 33, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 61)
875 // CHECK1-NEXT: br label [[OMP_DISPATCH_COND:%.*]]
876 // CHECK1: omp.dispatch.cond:
877 // CHECK1-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
878 // CHECK1-NEXT: [[TMP6:%.*]] = load i64, ptr [[DOTPREVIOUS_UB__ADDR]], align 8
879 // CHECK1-NEXT: [[CONV2:%.*]] = trunc i64 [[TMP6]] to i32
880 // CHECK1-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP5]], [[CONV2]]
881 // CHECK1-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
882 // CHECK1: cond.true:
883 // CHECK1-NEXT: [[TMP7:%.*]] = load i64, ptr [[DOTPREVIOUS_UB__ADDR]], align 8
884 // CHECK1-NEXT: [[CONV3:%.*]] = trunc i64 [[TMP7]] to i32
885 // CHECK1-NEXT: br label [[COND_END:%.*]]
886 // CHECK1: cond.false:
887 // CHECK1-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
888 // CHECK1-NEXT: br label [[COND_END]]
890 // CHECK1-NEXT: [[COND:%.*]] = phi i32 [ [[CONV3]], [[COND_TRUE]] ], [ [[TMP8]], [[COND_FALSE]] ]
891 // CHECK1-NEXT: store i32 [[COND]], ptr [[DOTOMP_UB]], align 4
892 // CHECK1-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4
893 // CHECK1-NEXT: store i32 [[TMP9]], ptr [[DOTOMP_IV]], align 4
894 // CHECK1-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
895 // CHECK1-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
896 // CHECK1-NEXT: [[CMP4:%.*]] = icmp sle i32 [[TMP10]], [[TMP11]]
897 // CHECK1-NEXT: br i1 [[CMP4]], label [[OMP_DISPATCH_BODY:%.*]], label [[OMP_DISPATCH_END:%.*]]
898 // CHECK1: omp.dispatch.body:
899 // CHECK1-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
900 // CHECK1: omp.inner.for.cond:
901 // CHECK1-NEXT: [[TMP12:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
902 // CHECK1-NEXT: [[TMP13:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
903 // CHECK1-NEXT: [[CMP5:%.*]] = icmp sle i32 [[TMP12]], [[TMP13]]
904 // CHECK1-NEXT: br i1 [[CMP5]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
905 // CHECK1: omp.inner.for.body:
906 // CHECK1-NEXT: [[TMP14:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
907 // CHECK1-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP14]], 1
908 // CHECK1-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
909 // CHECK1-NEXT: store i32 [[ADD]], ptr [[I]], align 4
910 // CHECK1-NEXT: [[A:%.*]] = getelementptr inbounds nuw [[STRUCT_SS:%.*]], ptr [[TMP0]], i32 0, i32 0
911 // CHECK1-NEXT: [[TMP15:%.*]] = load i32, ptr [[I]], align 4
912 // CHECK1-NEXT: [[IDXPROM:%.*]] = sext i32 [[TMP15]] to i64
913 // CHECK1-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds [123 x i32], ptr [[A]], i64 0, i64 [[IDXPROM]]
914 // CHECK1-NEXT: store i32 0, ptr [[ARRAYIDX]], align 4
915 // CHECK1-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
916 // CHECK1: omp.body.continue:
917 // CHECK1-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
918 // CHECK1: omp.inner.for.inc:
919 // CHECK1-NEXT: [[TMP16:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
920 // CHECK1-NEXT: [[ADD6:%.*]] = add nsw i32 [[TMP16]], 1
921 // CHECK1-NEXT: store i32 [[ADD6]], ptr [[DOTOMP_IV]], align 4
922 // CHECK1-NEXT: br label [[OMP_INNER_FOR_COND]]
923 // CHECK1: omp.inner.for.end:
924 // CHECK1-NEXT: br label [[OMP_DISPATCH_INC:%.*]]
925 // CHECK1: omp.dispatch.inc:
926 // CHECK1-NEXT: [[TMP17:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4
927 // CHECK1-NEXT: [[TMP18:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4
928 // CHECK1-NEXT: [[ADD7:%.*]] = add nsw i32 [[TMP17]], [[TMP18]]
929 // CHECK1-NEXT: store i32 [[ADD7]], ptr [[DOTOMP_LB]], align 4
930 // CHECK1-NEXT: [[TMP19:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
931 // CHECK1-NEXT: [[TMP20:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4
932 // CHECK1-NEXT: [[ADD8:%.*]] = add nsw i32 [[TMP19]], [[TMP20]]
933 // CHECK1-NEXT: store i32 [[ADD8]], ptr [[DOTOMP_UB]], align 4
934 // CHECK1-NEXT: br label [[OMP_DISPATCH_COND]]
935 // CHECK1: omp.dispatch.end:
936 // CHECK1-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB2]], i32 [[TMP4]])
937 // CHECK1-NEXT: ret void
940 // CHECK1-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l51
941 // CHECK1-SAME: (ptr noundef [[THIS:%.*]]) #[[ATTR1]] {
942 // CHECK1-NEXT: entry:
943 // CHECK1-NEXT: [[THIS_ADDR:%.*]] = alloca ptr, align 8
944 // CHECK1-NEXT: store ptr [[THIS]], ptr [[THIS_ADDR]], align 8
945 // CHECK1-NEXT: [[TMP0:%.*]] = load ptr, ptr [[THIS_ADDR]], align 8
946 // CHECK1-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_teams(ptr @[[GLOB3]], i32 1, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l51.omp_outlined, ptr [[TMP0]])
947 // CHECK1-NEXT: ret void
950 // CHECK1-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l51.omp_outlined
951 // CHECK1-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], ptr noundef [[THIS:%.*]]) #[[ATTR1]] {
952 // CHECK1-NEXT: entry:
953 // CHECK1-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 8
954 // CHECK1-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 8
955 // CHECK1-NEXT: [[THIS_ADDR:%.*]] = alloca ptr, align 8
956 // CHECK1-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
957 // CHECK1-NEXT: [[TMP:%.*]] = alloca i32, align 4
958 // CHECK1-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4
959 // CHECK1-NEXT: [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4
960 // CHECK1-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
961 // CHECK1-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
962 // CHECK1-NEXT: [[I:%.*]] = alloca i32, align 4
963 // CHECK1-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 8
964 // CHECK1-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 8
965 // CHECK1-NEXT: store ptr [[THIS]], ptr [[THIS_ADDR]], align 8
966 // CHECK1-NEXT: [[TMP0:%.*]] = load ptr, ptr [[THIS_ADDR]], align 8
967 // CHECK1-NEXT: store i32 0, ptr [[DOTOMP_COMB_LB]], align 4
968 // CHECK1-NEXT: store i32 122, ptr [[DOTOMP_COMB_UB]], align 4
969 // CHECK1-NEXT: store i32 1, ptr [[DOTOMP_STRIDE]], align 4
970 // CHECK1-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
971 // CHECK1-NEXT: [[TMP1:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
972 // CHECK1-NEXT: [[TMP2:%.*]] = load i32, ptr [[TMP1]], align 4
973 // CHECK1-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB1]], i32 [[TMP2]], i32 92, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
974 // CHECK1-NEXT: [[TMP3:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
975 // CHECK1-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP3]], 122
976 // CHECK1-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
977 // CHECK1: cond.true:
978 // CHECK1-NEXT: br label [[COND_END:%.*]]
979 // CHECK1: cond.false:
980 // CHECK1-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
981 // CHECK1-NEXT: br label [[COND_END]]
983 // CHECK1-NEXT: [[COND:%.*]] = phi i32 [ 122, [[COND_TRUE]] ], [ [[TMP4]], [[COND_FALSE]] ]
984 // CHECK1-NEXT: store i32 [[COND]], ptr [[DOTOMP_COMB_UB]], align 4
985 // CHECK1-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
986 // CHECK1-NEXT: store i32 [[TMP5]], ptr [[DOTOMP_IV]], align 4
987 // CHECK1-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
988 // CHECK1: omp.inner.for.cond:
989 // CHECK1-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
990 // CHECK1-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
991 // CHECK1-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP6]], [[TMP7]]
992 // CHECK1-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
993 // CHECK1: omp.inner.for.body:
994 // CHECK1-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
995 // CHECK1-NEXT: [[TMP9:%.*]] = zext i32 [[TMP8]] to i64
996 // CHECK1-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
997 // CHECK1-NEXT: [[TMP11:%.*]] = zext i32 [[TMP10]] to i64
998 // CHECK1-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_call(ptr @[[GLOB3]], i32 3, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l51.omp_outlined.omp_outlined, i64 [[TMP9]], i64 [[TMP11]], ptr [[TMP0]])
999 // CHECK1-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
1000 // CHECK1: omp.inner.for.inc:
1001 // CHECK1-NEXT: [[TMP12:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
1002 // CHECK1-NEXT: [[TMP13:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4
1003 // CHECK1-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP12]], [[TMP13]]
1004 // CHECK1-NEXT: store i32 [[ADD]], ptr [[DOTOMP_IV]], align 4
1005 // CHECK1-NEXT: br label [[OMP_INNER_FOR_COND]]
1006 // CHECK1: omp.inner.for.end:
1007 // CHECK1-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
1008 // CHECK1: omp.loop.exit:
1009 // CHECK1-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB1]], i32 [[TMP2]])
1010 // CHECK1-NEXT: ret void
1013 // CHECK1-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l51.omp_outlined.omp_outlined
1014 // CHECK1-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], i64 noundef [[DOTPREVIOUS_LB_:%.*]], i64 noundef [[DOTPREVIOUS_UB_:%.*]], ptr noundef [[THIS:%.*]]) #[[ATTR1]] {
1015 // CHECK1-NEXT: entry:
1016 // CHECK1-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 8
1017 // CHECK1-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 8
1018 // CHECK1-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i64, align 8
1019 // CHECK1-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i64, align 8
1020 // CHECK1-NEXT: [[THIS_ADDR:%.*]] = alloca ptr, align 8
1021 // CHECK1-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
1022 // CHECK1-NEXT: [[TMP:%.*]] = alloca i32, align 4
1023 // CHECK1-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4
1024 // CHECK1-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4
1025 // CHECK1-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
1026 // CHECK1-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
1027 // CHECK1-NEXT: [[I:%.*]] = alloca i32, align 4
1028 // CHECK1-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 8
1029 // CHECK1-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 8
1030 // CHECK1-NEXT: store i64 [[DOTPREVIOUS_LB_]], ptr [[DOTPREVIOUS_LB__ADDR]], align 8
1031 // CHECK1-NEXT: store i64 [[DOTPREVIOUS_UB_]], ptr [[DOTPREVIOUS_UB__ADDR]], align 8
1032 // CHECK1-NEXT: store ptr [[THIS]], ptr [[THIS_ADDR]], align 8
1033 // CHECK1-NEXT: [[TMP0:%.*]] = load ptr, ptr [[THIS_ADDR]], align 8
1034 // CHECK1-NEXT: store i32 0, ptr [[DOTOMP_LB]], align 4
1035 // CHECK1-NEXT: store i32 122, ptr [[DOTOMP_UB]], align 4
1036 // CHECK1-NEXT: [[TMP1:%.*]] = load i64, ptr [[DOTPREVIOUS_LB__ADDR]], align 8
1037 // CHECK1-NEXT: [[CONV:%.*]] = trunc i64 [[TMP1]] to i32
1038 // CHECK1-NEXT: [[TMP2:%.*]] = load i64, ptr [[DOTPREVIOUS_UB__ADDR]], align 8
1039 // CHECK1-NEXT: [[CONV1:%.*]] = trunc i64 [[TMP2]] to i32
1040 // CHECK1-NEXT: store i32 [[CONV]], ptr [[DOTOMP_LB]], align 4
1041 // CHECK1-NEXT: store i32 [[CONV1]], ptr [[DOTOMP_UB]], align 4
1042 // CHECK1-NEXT: store i32 1, ptr [[DOTOMP_STRIDE]], align 4
1043 // CHECK1-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
1044 // CHECK1-NEXT: [[TMP3:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4
1045 // CHECK1-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
1046 // CHECK1-NEXT: [[TMP5:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
1047 // CHECK1-NEXT: [[TMP6:%.*]] = load i32, ptr [[TMP5]], align 4
1048 // CHECK1-NEXT: call void @__kmpc_dispatch_init_4(ptr @[[GLOB3]], i32 [[TMP6]], i32 1073741859, i32 [[TMP3]], i32 [[TMP4]], i32 1, i32 1)
1049 // CHECK1-NEXT: br label [[OMP_DISPATCH_COND:%.*]]
1050 // CHECK1: omp.dispatch.cond:
1051 // CHECK1-NEXT: [[TMP7:%.*]] = call i32 @__kmpc_dispatch_next_4(ptr @[[GLOB3]], i32 [[TMP6]], ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]])
1052 // CHECK1-NEXT: [[TOBOOL:%.*]] = icmp ne i32 [[TMP7]], 0
1053 // CHECK1-NEXT: br i1 [[TOBOOL]], label [[OMP_DISPATCH_BODY:%.*]], label [[OMP_DISPATCH_END:%.*]]
1054 // CHECK1: omp.dispatch.body:
1055 // CHECK1-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4
1056 // CHECK1-NEXT: store i32 [[TMP8]], ptr [[DOTOMP_IV]], align 4
1057 // CHECK1-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
1058 // CHECK1: omp.inner.for.cond:
1059 // CHECK1-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP10:![0-9]+]]
1060 // CHECK1-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP10]]
1061 // CHECK1-NEXT: [[CMP:%.*]] = icmp sle i32 [[TMP9]], [[TMP10]]
1062 // CHECK1-NEXT: br i1 [[CMP]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
1063 // CHECK1: omp.inner.for.body:
1064 // CHECK1-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP10]]
1065 // CHECK1-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP11]], 1
1066 // CHECK1-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
1067 // CHECK1-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP10]]
1068 // CHECK1-NEXT: [[A:%.*]] = getelementptr inbounds nuw [[STRUCT_SS:%.*]], ptr [[TMP0]], i32 0, i32 0
1069 // CHECK1-NEXT: [[TMP12:%.*]] = load i32, ptr [[I]], align 4, !llvm.access.group [[ACC_GRP10]]
1070 // CHECK1-NEXT: [[IDXPROM:%.*]] = sext i32 [[TMP12]] to i64
1071 // CHECK1-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds [123 x i32], ptr [[A]], i64 0, i64 [[IDXPROM]]
1072 // CHECK1-NEXT: store i32 0, ptr [[ARRAYIDX]], align 4, !llvm.access.group [[ACC_GRP10]]
1073 // CHECK1-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
1074 // CHECK1: omp.body.continue:
1075 // CHECK1-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
1076 // CHECK1: omp.inner.for.inc:
1077 // CHECK1-NEXT: [[TMP13:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP10]]
1078 // CHECK1-NEXT: [[ADD2:%.*]] = add nsw i32 [[TMP13]], 1
1079 // CHECK1-NEXT: store i32 [[ADD2]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP10]]
1080 // CHECK1-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP11:![0-9]+]]
1081 // CHECK1: omp.inner.for.end:
1082 // CHECK1-NEXT: br label [[OMP_DISPATCH_INC:%.*]]
1083 // CHECK1: omp.dispatch.inc:
1084 // CHECK1-NEXT: br label [[OMP_DISPATCH_COND]]
1085 // CHECK1: omp.dispatch.end:
1086 // CHECK1-NEXT: call void @__kmpc_dispatch_deinit(ptr @[[GLOB3]], i32 [[TMP6]])
1087 // CHECK1-NEXT: ret void
1090 // CHECK1-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l57
1091 // CHECK1-SAME: (ptr noundef [[THIS:%.*]]) #[[ATTR1]] {
1092 // CHECK1-NEXT: entry:
1093 // CHECK1-NEXT: [[THIS_ADDR:%.*]] = alloca ptr, align 8
1094 // CHECK1-NEXT: store ptr [[THIS]], ptr [[THIS_ADDR]], align 8
1095 // CHECK1-NEXT: [[TMP0:%.*]] = load ptr, ptr [[THIS_ADDR]], align 8
1096 // CHECK1-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_teams(ptr @[[GLOB3]], i32 1, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l57.omp_outlined, ptr [[TMP0]])
1097 // CHECK1-NEXT: ret void
1100 // CHECK1-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l57.omp_outlined
1101 // CHECK1-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], ptr noundef [[THIS:%.*]]) #[[ATTR1]] {
1102 // CHECK1-NEXT: entry:
1103 // CHECK1-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 8
1104 // CHECK1-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 8
1105 // CHECK1-NEXT: [[THIS_ADDR:%.*]] = alloca ptr, align 8
1106 // CHECK1-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
1107 // CHECK1-NEXT: [[TMP:%.*]] = alloca i32, align 4
1108 // CHECK1-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4
1109 // CHECK1-NEXT: [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4
1110 // CHECK1-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
1111 // CHECK1-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
1112 // CHECK1-NEXT: [[I:%.*]] = alloca i32, align 4
1113 // CHECK1-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 8
1114 // CHECK1-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 8
1115 // CHECK1-NEXT: store ptr [[THIS]], ptr [[THIS_ADDR]], align 8
1116 // CHECK1-NEXT: [[TMP0:%.*]] = load ptr, ptr [[THIS_ADDR]], align 8
1117 // CHECK1-NEXT: store i32 0, ptr [[DOTOMP_COMB_LB]], align 4
1118 // CHECK1-NEXT: store i32 122, ptr [[DOTOMP_COMB_UB]], align 4
1119 // CHECK1-NEXT: store i32 1, ptr [[DOTOMP_STRIDE]], align 4
1120 // CHECK1-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
1121 // CHECK1-NEXT: [[TMP1:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
1122 // CHECK1-NEXT: [[TMP2:%.*]] = load i32, ptr [[TMP1]], align 4
1123 // CHECK1-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB1]], i32 [[TMP2]], i32 92, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
1124 // CHECK1-NEXT: [[TMP3:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
1125 // CHECK1-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP3]], 122
1126 // CHECK1-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
1127 // CHECK1: cond.true:
1128 // CHECK1-NEXT: br label [[COND_END:%.*]]
1129 // CHECK1: cond.false:
1130 // CHECK1-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
1131 // CHECK1-NEXT: br label [[COND_END]]
1132 // CHECK1: cond.end:
1133 // CHECK1-NEXT: [[COND:%.*]] = phi i32 [ 122, [[COND_TRUE]] ], [ [[TMP4]], [[COND_FALSE]] ]
1134 // CHECK1-NEXT: store i32 [[COND]], ptr [[DOTOMP_COMB_UB]], align 4
1135 // CHECK1-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
1136 // CHECK1-NEXT: store i32 [[TMP5]], ptr [[DOTOMP_IV]], align 4
1137 // CHECK1-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
1138 // CHECK1: omp.inner.for.cond:
1139 // CHECK1-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
1140 // CHECK1-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
1141 // CHECK1-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP6]], [[TMP7]]
1142 // CHECK1-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
1143 // CHECK1: omp.inner.for.body:
1144 // CHECK1-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
1145 // CHECK1-NEXT: [[TMP9:%.*]] = zext i32 [[TMP8]] to i64
1146 // CHECK1-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
1147 // CHECK1-NEXT: [[TMP11:%.*]] = zext i32 [[TMP10]] to i64
1148 // CHECK1-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_call(ptr @[[GLOB3]], i32 3, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l57.omp_outlined.omp_outlined, i64 [[TMP9]], i64 [[TMP11]], ptr [[TMP0]])
1149 // CHECK1-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
1150 // CHECK1: omp.inner.for.inc:
1151 // CHECK1-NEXT: [[TMP12:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
1152 // CHECK1-NEXT: [[TMP13:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4
1153 // CHECK1-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP12]], [[TMP13]]
1154 // CHECK1-NEXT: store i32 [[ADD]], ptr [[DOTOMP_IV]], align 4
1155 // CHECK1-NEXT: br label [[OMP_INNER_FOR_COND]]
1156 // CHECK1: omp.inner.for.end:
1157 // CHECK1-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
1158 // CHECK1: omp.loop.exit:
1159 // CHECK1-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB1]], i32 [[TMP2]])
1160 // CHECK1-NEXT: ret void
1163 // CHECK1-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l57.omp_outlined.omp_outlined
1164 // CHECK1-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], i64 noundef [[DOTPREVIOUS_LB_:%.*]], i64 noundef [[DOTPREVIOUS_UB_:%.*]], ptr noundef [[THIS:%.*]]) #[[ATTR1]] {
1165 // CHECK1-NEXT: entry:
1166 // CHECK1-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 8
1167 // CHECK1-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 8
1168 // CHECK1-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i64, align 8
1169 // CHECK1-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i64, align 8
1170 // CHECK1-NEXT: [[THIS_ADDR:%.*]] = alloca ptr, align 8
1171 // CHECK1-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
1172 // CHECK1-NEXT: [[TMP:%.*]] = alloca i32, align 4
1173 // CHECK1-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4
1174 // CHECK1-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4
1175 // CHECK1-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
1176 // CHECK1-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
1177 // CHECK1-NEXT: [[I:%.*]] = alloca i32, align 4
1178 // CHECK1-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 8
1179 // CHECK1-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 8
1180 // CHECK1-NEXT: store i64 [[DOTPREVIOUS_LB_]], ptr [[DOTPREVIOUS_LB__ADDR]], align 8
1181 // CHECK1-NEXT: store i64 [[DOTPREVIOUS_UB_]], ptr [[DOTPREVIOUS_UB__ADDR]], align 8
1182 // CHECK1-NEXT: store ptr [[THIS]], ptr [[THIS_ADDR]], align 8
1183 // CHECK1-NEXT: [[TMP0:%.*]] = load ptr, ptr [[THIS_ADDR]], align 8
1184 // CHECK1-NEXT: store i32 0, ptr [[DOTOMP_LB]], align 4
1185 // CHECK1-NEXT: store i32 122, ptr [[DOTOMP_UB]], align 4
1186 // CHECK1-NEXT: [[TMP1:%.*]] = load i64, ptr [[DOTPREVIOUS_LB__ADDR]], align 8
1187 // CHECK1-NEXT: [[CONV:%.*]] = trunc i64 [[TMP1]] to i32
1188 // CHECK1-NEXT: [[TMP2:%.*]] = load i64, ptr [[DOTPREVIOUS_UB__ADDR]], align 8
1189 // CHECK1-NEXT: [[CONV1:%.*]] = trunc i64 [[TMP2]] to i32
1190 // CHECK1-NEXT: store i32 [[CONV]], ptr [[DOTOMP_LB]], align 4
1191 // CHECK1-NEXT: store i32 [[CONV1]], ptr [[DOTOMP_UB]], align 4
1192 // CHECK1-NEXT: store i32 1, ptr [[DOTOMP_STRIDE]], align 4
1193 // CHECK1-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
1194 // CHECK1-NEXT: [[TMP3:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4
1195 // CHECK1-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
1196 // CHECK1-NEXT: [[TMP5:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
1197 // CHECK1-NEXT: [[TMP6:%.*]] = load i32, ptr [[TMP5]], align 4
1198 // CHECK1-NEXT: call void @__kmpc_dispatch_init_4(ptr @[[GLOB3]], i32 [[TMP6]], i32 1073741859, i32 [[TMP3]], i32 [[TMP4]], i32 1, i32 61)
1199 // CHECK1-NEXT: br label [[OMP_DISPATCH_COND:%.*]]
1200 // CHECK1: omp.dispatch.cond:
1201 // CHECK1-NEXT: [[TMP7:%.*]] = call i32 @__kmpc_dispatch_next_4(ptr @[[GLOB3]], i32 [[TMP6]], ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]])
1202 // CHECK1-NEXT: [[TOBOOL:%.*]] = icmp ne i32 [[TMP7]], 0
1203 // CHECK1-NEXT: br i1 [[TOBOOL]], label [[OMP_DISPATCH_BODY:%.*]], label [[OMP_DISPATCH_END:%.*]]
1204 // CHECK1: omp.dispatch.body:
1205 // CHECK1-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4
1206 // CHECK1-NEXT: store i32 [[TMP8]], ptr [[DOTOMP_IV]], align 4
1207 // CHECK1-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
1208 // CHECK1: omp.inner.for.cond:
1209 // CHECK1-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP13:![0-9]+]]
1210 // CHECK1-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP13]]
1211 // CHECK1-NEXT: [[CMP:%.*]] = icmp sle i32 [[TMP9]], [[TMP10]]
1212 // CHECK1-NEXT: br i1 [[CMP]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
1213 // CHECK1: omp.inner.for.body:
1214 // CHECK1-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP13]]
1215 // CHECK1-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP11]], 1
1216 // CHECK1-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
1217 // CHECK1-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP13]]
1218 // CHECK1-NEXT: [[A:%.*]] = getelementptr inbounds nuw [[STRUCT_SS:%.*]], ptr [[TMP0]], i32 0, i32 0
1219 // CHECK1-NEXT: [[TMP12:%.*]] = load i32, ptr [[I]], align 4, !llvm.access.group [[ACC_GRP13]]
1220 // CHECK1-NEXT: [[IDXPROM:%.*]] = sext i32 [[TMP12]] to i64
1221 // CHECK1-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds [123 x i32], ptr [[A]], i64 0, i64 [[IDXPROM]]
1222 // CHECK1-NEXT: store i32 0, ptr [[ARRAYIDX]], align 4, !llvm.access.group [[ACC_GRP13]]
1223 // CHECK1-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
1224 // CHECK1: omp.body.continue:
1225 // CHECK1-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
1226 // CHECK1: omp.inner.for.inc:
1227 // CHECK1-NEXT: [[TMP13:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP13]]
1228 // CHECK1-NEXT: [[ADD2:%.*]] = add nsw i32 [[TMP13]], 1
1229 // CHECK1-NEXT: store i32 [[ADD2]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP13]]
1230 // CHECK1-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP14:![0-9]+]]
1231 // CHECK1: omp.inner.for.end:
1232 // CHECK1-NEXT: br label [[OMP_DISPATCH_INC:%.*]]
1233 // CHECK1: omp.dispatch.inc:
1234 // CHECK1-NEXT: br label [[OMP_DISPATCH_COND]]
1235 // CHECK1: omp.dispatch.end:
1236 // CHECK1-NEXT: call void @__kmpc_dispatch_deinit(ptr @[[GLOB3]], i32 [[TMP6]])
1237 // CHECK1-NEXT: ret void
1240 // CHECK3-LABEL: define {{[^@]+}}@_Z21teams_template_structv
1241 // CHECK3-SAME: () #[[ATTR0:[0-9]+]] {
1242 // CHECK3-NEXT: entry:
1243 // CHECK3-NEXT: [[V:%.*]] = alloca [[STRUCT_SS:%.*]], align 4
1244 // CHECK3-NEXT: [[CALL:%.*]] = call noundef i32 @_ZN2SSIiLi123ELx456EE3fooEv(ptr noundef nonnull align 4 dereferenceable(496) [[V]])
1245 // CHECK3-NEXT: ret i32 [[CALL]]
1248 // CHECK3-LABEL: define {{[^@]+}}@_ZN2SSIiLi123ELx456EE3fooEv
1249 // CHECK3-SAME: (ptr noundef nonnull align 4 dereferenceable(496) [[THIS:%.*]]) #[[ATTR0]] comdat align 2 {
1250 // CHECK3-NEXT: entry:
1251 // CHECK3-NEXT: [[THIS_ADDR:%.*]] = alloca ptr, align 4
1252 // CHECK3-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [1 x ptr], align 4
1253 // CHECK3-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [1 x ptr], align 4
1254 // CHECK3-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [1 x ptr], align 4
1255 // CHECK3-NEXT: [[TMP:%.*]] = alloca i32, align 4
1256 // CHECK3-NEXT: [[KERNEL_ARGS:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS:%.*]], align 8
1257 // CHECK3-NEXT: [[DOTOFFLOAD_BASEPTRS3:%.*]] = alloca [1 x ptr], align 4
1258 // CHECK3-NEXT: [[DOTOFFLOAD_PTRS4:%.*]] = alloca [1 x ptr], align 4
1259 // CHECK3-NEXT: [[DOTOFFLOAD_MAPPERS5:%.*]] = alloca [1 x ptr], align 4
1260 // CHECK3-NEXT: [[_TMP6:%.*]] = alloca i32, align 4
1261 // CHECK3-NEXT: [[KERNEL_ARGS7:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS]], align 8
1262 // CHECK3-NEXT: [[DOTOFFLOAD_BASEPTRS11:%.*]] = alloca [1 x ptr], align 4
1263 // CHECK3-NEXT: [[DOTOFFLOAD_PTRS12:%.*]] = alloca [1 x ptr], align 4
1264 // CHECK3-NEXT: [[DOTOFFLOAD_MAPPERS13:%.*]] = alloca [1 x ptr], align 4
1265 // CHECK3-NEXT: [[_TMP14:%.*]] = alloca i32, align 4
1266 // CHECK3-NEXT: [[KERNEL_ARGS15:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS]], align 8
1267 // CHECK3-NEXT: [[DOTOFFLOAD_BASEPTRS19:%.*]] = alloca [1 x ptr], align 4
1268 // CHECK3-NEXT: [[DOTOFFLOAD_PTRS20:%.*]] = alloca [1 x ptr], align 4
1269 // CHECK3-NEXT: [[DOTOFFLOAD_MAPPERS21:%.*]] = alloca [1 x ptr], align 4
1270 // CHECK3-NEXT: [[_TMP22:%.*]] = alloca i32, align 4
1271 // CHECK3-NEXT: [[KERNEL_ARGS23:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS]], align 8
1272 // CHECK3-NEXT: [[DOTOFFLOAD_BASEPTRS27:%.*]] = alloca [1 x ptr], align 4
1273 // CHECK3-NEXT: [[DOTOFFLOAD_PTRS28:%.*]] = alloca [1 x ptr], align 4
1274 // CHECK3-NEXT: [[DOTOFFLOAD_MAPPERS29:%.*]] = alloca [1 x ptr], align 4
1275 // CHECK3-NEXT: [[_TMP30:%.*]] = alloca i32, align 4
1276 // CHECK3-NEXT: [[KERNEL_ARGS31:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS]], align 8
1277 // CHECK3-NEXT: store ptr [[THIS]], ptr [[THIS_ADDR]], align 4
1278 // CHECK3-NEXT: [[THIS1:%.*]] = load ptr, ptr [[THIS_ADDR]], align 4
1279 // CHECK3-NEXT: [[A:%.*]] = getelementptr inbounds nuw [[STRUCT_SS:%.*]], ptr [[THIS1]], i32 0, i32 0
1280 // CHECK3-NEXT: [[TMP0:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
1281 // CHECK3-NEXT: store ptr [[THIS1]], ptr [[TMP0]], align 4
1282 // CHECK3-NEXT: [[TMP1:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
1283 // CHECK3-NEXT: store ptr [[A]], ptr [[TMP1]], align 4
1284 // CHECK3-NEXT: [[TMP2:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0
1285 // CHECK3-NEXT: store ptr null, ptr [[TMP2]], align 4
1286 // CHECK3-NEXT: [[TMP3:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
1287 // CHECK3-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
1288 // CHECK3-NEXT: [[TMP5:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 0
1289 // CHECK3-NEXT: store i32 3, ptr [[TMP5]], align 4
1290 // CHECK3-NEXT: [[TMP6:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 1
1291 // CHECK3-NEXT: store i32 1, ptr [[TMP6]], align 4
1292 // CHECK3-NEXT: [[TMP7:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 2
1293 // CHECK3-NEXT: store ptr [[TMP3]], ptr [[TMP7]], align 4
1294 // CHECK3-NEXT: [[TMP8:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 3
1295 // CHECK3-NEXT: store ptr [[TMP4]], ptr [[TMP8]], align 4
1296 // CHECK3-NEXT: [[TMP9:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 4
1297 // CHECK3-NEXT: store ptr @.offload_sizes, ptr [[TMP9]], align 4
1298 // CHECK3-NEXT: [[TMP10:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 5
1299 // CHECK3-NEXT: store ptr @.offload_maptypes, ptr [[TMP10]], align 4
1300 // CHECK3-NEXT: [[TMP11:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 6
1301 // CHECK3-NEXT: store ptr null, ptr [[TMP11]], align 4
1302 // CHECK3-NEXT: [[TMP12:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 7
1303 // CHECK3-NEXT: store ptr null, ptr [[TMP12]], align 4
1304 // CHECK3-NEXT: [[TMP13:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 8
1305 // CHECK3-NEXT: store i64 123, ptr [[TMP13]], align 8
1306 // CHECK3-NEXT: [[TMP14:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 9
1307 // CHECK3-NEXT: store i64 0, ptr [[TMP14]], align 8
1308 // CHECK3-NEXT: [[TMP15:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 10
1309 // CHECK3-NEXT: store [3 x i32] zeroinitializer, ptr [[TMP15]], align 4
1310 // CHECK3-NEXT: [[TMP16:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 11
1311 // CHECK3-NEXT: store [3 x i32] zeroinitializer, ptr [[TMP16]], align 4
1312 // CHECK3-NEXT: [[TMP17:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 12
1313 // CHECK3-NEXT: store i32 0, ptr [[TMP17]], align 4
1314 // CHECK3-NEXT: [[TMP18:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB3:[0-9]+]], i64 -1, i32 0, i32 0, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l35.region_id, ptr [[KERNEL_ARGS]])
1315 // CHECK3-NEXT: [[TMP19:%.*]] = icmp ne i32 [[TMP18]], 0
1316 // CHECK3-NEXT: br i1 [[TMP19]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
1317 // CHECK3: omp_offload.failed:
1318 // CHECK3-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l35(ptr [[THIS1]]) #[[ATTR2:[0-9]+]]
1319 // CHECK3-NEXT: br label [[OMP_OFFLOAD_CONT]]
1320 // CHECK3: omp_offload.cont:
1321 // CHECK3-NEXT: [[A2:%.*]] = getelementptr inbounds nuw [[STRUCT_SS]], ptr [[THIS1]], i32 0, i32 0
1322 // CHECK3-NEXT: [[TMP20:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS3]], i32 0, i32 0
1323 // CHECK3-NEXT: store ptr [[THIS1]], ptr [[TMP20]], align 4
1324 // CHECK3-NEXT: [[TMP21:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS4]], i32 0, i32 0
1325 // CHECK3-NEXT: store ptr [[A2]], ptr [[TMP21]], align 4
1326 // CHECK3-NEXT: [[TMP22:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_MAPPERS5]], i32 0, i32 0
1327 // CHECK3-NEXT: store ptr null, ptr [[TMP22]], align 4
1328 // CHECK3-NEXT: [[TMP23:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS3]], i32 0, i32 0
1329 // CHECK3-NEXT: [[TMP24:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS4]], i32 0, i32 0
1330 // CHECK3-NEXT: [[TMP25:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS7]], i32 0, i32 0
1331 // CHECK3-NEXT: store i32 3, ptr [[TMP25]], align 4
1332 // CHECK3-NEXT: [[TMP26:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS7]], i32 0, i32 1
1333 // CHECK3-NEXT: store i32 1, ptr [[TMP26]], align 4
1334 // CHECK3-NEXT: [[TMP27:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS7]], i32 0, i32 2
1335 // CHECK3-NEXT: store ptr [[TMP23]], ptr [[TMP27]], align 4
1336 // CHECK3-NEXT: [[TMP28:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS7]], i32 0, i32 3
1337 // CHECK3-NEXT: store ptr [[TMP24]], ptr [[TMP28]], align 4
1338 // CHECK3-NEXT: [[TMP29:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS7]], i32 0, i32 4
1339 // CHECK3-NEXT: store ptr @.offload_sizes.1, ptr [[TMP29]], align 4
1340 // CHECK3-NEXT: [[TMP30:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS7]], i32 0, i32 5
1341 // CHECK3-NEXT: store ptr @.offload_maptypes.2, ptr [[TMP30]], align 4
1342 // CHECK3-NEXT: [[TMP31:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS7]], i32 0, i32 6
1343 // CHECK3-NEXT: store ptr null, ptr [[TMP31]], align 4
1344 // CHECK3-NEXT: [[TMP32:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS7]], i32 0, i32 7
1345 // CHECK3-NEXT: store ptr null, ptr [[TMP32]], align 4
1346 // CHECK3-NEXT: [[TMP33:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS7]], i32 0, i32 8
1347 // CHECK3-NEXT: store i64 123, ptr [[TMP33]], align 8
1348 // CHECK3-NEXT: [[TMP34:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS7]], i32 0, i32 9
1349 // CHECK3-NEXT: store i64 0, ptr [[TMP34]], align 8
1350 // CHECK3-NEXT: [[TMP35:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS7]], i32 0, i32 10
1351 // CHECK3-NEXT: store [3 x i32] zeroinitializer, ptr [[TMP35]], align 4
1352 // CHECK3-NEXT: [[TMP36:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS7]], i32 0, i32 11
1353 // CHECK3-NEXT: store [3 x i32] zeroinitializer, ptr [[TMP36]], align 4
1354 // CHECK3-NEXT: [[TMP37:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS7]], i32 0, i32 12
1355 // CHECK3-NEXT: store i32 0, ptr [[TMP37]], align 4
1356 // CHECK3-NEXT: [[TMP38:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB3]], i64 -1, i32 0, i32 0, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l40.region_id, ptr [[KERNEL_ARGS7]])
1357 // CHECK3-NEXT: [[TMP39:%.*]] = icmp ne i32 [[TMP38]], 0
1358 // CHECK3-NEXT: br i1 [[TMP39]], label [[OMP_OFFLOAD_FAILED8:%.*]], label [[OMP_OFFLOAD_CONT9:%.*]]
1359 // CHECK3: omp_offload.failed8:
1360 // CHECK3-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l40(ptr [[THIS1]]) #[[ATTR2]]
1361 // CHECK3-NEXT: br label [[OMP_OFFLOAD_CONT9]]
1362 // CHECK3: omp_offload.cont9:
1363 // CHECK3-NEXT: [[A10:%.*]] = getelementptr inbounds nuw [[STRUCT_SS]], ptr [[THIS1]], i32 0, i32 0
1364 // CHECK3-NEXT: [[TMP40:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS11]], i32 0, i32 0
1365 // CHECK3-NEXT: store ptr [[THIS1]], ptr [[TMP40]], align 4
1366 // CHECK3-NEXT: [[TMP41:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS12]], i32 0, i32 0
1367 // CHECK3-NEXT: store ptr [[A10]], ptr [[TMP41]], align 4
1368 // CHECK3-NEXT: [[TMP42:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_MAPPERS13]], i32 0, i32 0
1369 // CHECK3-NEXT: store ptr null, ptr [[TMP42]], align 4
1370 // CHECK3-NEXT: [[TMP43:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS11]], i32 0, i32 0
1371 // CHECK3-NEXT: [[TMP44:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS12]], i32 0, i32 0
1372 // CHECK3-NEXT: [[TMP45:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS15]], i32 0, i32 0
1373 // CHECK3-NEXT: store i32 3, ptr [[TMP45]], align 4
1374 // CHECK3-NEXT: [[TMP46:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS15]], i32 0, i32 1
1375 // CHECK3-NEXT: store i32 1, ptr [[TMP46]], align 4
1376 // CHECK3-NEXT: [[TMP47:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS15]], i32 0, i32 2
1377 // CHECK3-NEXT: store ptr [[TMP43]], ptr [[TMP47]], align 4
1378 // CHECK3-NEXT: [[TMP48:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS15]], i32 0, i32 3
1379 // CHECK3-NEXT: store ptr [[TMP44]], ptr [[TMP48]], align 4
1380 // CHECK3-NEXT: [[TMP49:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS15]], i32 0, i32 4
1381 // CHECK3-NEXT: store ptr @.offload_sizes.3, ptr [[TMP49]], align 4
1382 // CHECK3-NEXT: [[TMP50:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS15]], i32 0, i32 5
1383 // CHECK3-NEXT: store ptr @.offload_maptypes.4, ptr [[TMP50]], align 4
1384 // CHECK3-NEXT: [[TMP51:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS15]], i32 0, i32 6
1385 // CHECK3-NEXT: store ptr null, ptr [[TMP51]], align 4
1386 // CHECK3-NEXT: [[TMP52:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS15]], i32 0, i32 7
1387 // CHECK3-NEXT: store ptr null, ptr [[TMP52]], align 4
1388 // CHECK3-NEXT: [[TMP53:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS15]], i32 0, i32 8
1389 // CHECK3-NEXT: store i64 123, ptr [[TMP53]], align 8
1390 // CHECK3-NEXT: [[TMP54:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS15]], i32 0, i32 9
1391 // CHECK3-NEXT: store i64 0, ptr [[TMP54]], align 8
1392 // CHECK3-NEXT: [[TMP55:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS15]], i32 0, i32 10
1393 // CHECK3-NEXT: store [3 x i32] zeroinitializer, ptr [[TMP55]], align 4
1394 // CHECK3-NEXT: [[TMP56:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS15]], i32 0, i32 11
1395 // CHECK3-NEXT: store [3 x i32] zeroinitializer, ptr [[TMP56]], align 4
1396 // CHECK3-NEXT: [[TMP57:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS15]], i32 0, i32 12
1397 // CHECK3-NEXT: store i32 0, ptr [[TMP57]], align 4
1398 // CHECK3-NEXT: [[TMP58:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB3]], i64 -1, i32 0, i32 0, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l45.region_id, ptr [[KERNEL_ARGS15]])
1399 // CHECK3-NEXT: [[TMP59:%.*]] = icmp ne i32 [[TMP58]], 0
1400 // CHECK3-NEXT: br i1 [[TMP59]], label [[OMP_OFFLOAD_FAILED16:%.*]], label [[OMP_OFFLOAD_CONT17:%.*]]
1401 // CHECK3: omp_offload.failed16:
1402 // CHECK3-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l45(ptr [[THIS1]]) #[[ATTR2]]
1403 // CHECK3-NEXT: br label [[OMP_OFFLOAD_CONT17]]
1404 // CHECK3: omp_offload.cont17:
1405 // CHECK3-NEXT: [[A18:%.*]] = getelementptr inbounds nuw [[STRUCT_SS]], ptr [[THIS1]], i32 0, i32 0
1406 // CHECK3-NEXT: [[TMP60:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS19]], i32 0, i32 0
1407 // CHECK3-NEXT: store ptr [[THIS1]], ptr [[TMP60]], align 4
1408 // CHECK3-NEXT: [[TMP61:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS20]], i32 0, i32 0
1409 // CHECK3-NEXT: store ptr [[A18]], ptr [[TMP61]], align 4
1410 // CHECK3-NEXT: [[TMP62:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_MAPPERS21]], i32 0, i32 0
1411 // CHECK3-NEXT: store ptr null, ptr [[TMP62]], align 4
1412 // CHECK3-NEXT: [[TMP63:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS19]], i32 0, i32 0
1413 // CHECK3-NEXT: [[TMP64:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS20]], i32 0, i32 0
1414 // CHECK3-NEXT: [[TMP65:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS23]], i32 0, i32 0
1415 // CHECK3-NEXT: store i32 3, ptr [[TMP65]], align 4
1416 // CHECK3-NEXT: [[TMP66:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS23]], i32 0, i32 1
1417 // CHECK3-NEXT: store i32 1, ptr [[TMP66]], align 4
1418 // CHECK3-NEXT: [[TMP67:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS23]], i32 0, i32 2
1419 // CHECK3-NEXT: store ptr [[TMP63]], ptr [[TMP67]], align 4
1420 // CHECK3-NEXT: [[TMP68:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS23]], i32 0, i32 3
1421 // CHECK3-NEXT: store ptr [[TMP64]], ptr [[TMP68]], align 4
1422 // CHECK3-NEXT: [[TMP69:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS23]], i32 0, i32 4
1423 // CHECK3-NEXT: store ptr @.offload_sizes.5, ptr [[TMP69]], align 4
1424 // CHECK3-NEXT: [[TMP70:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS23]], i32 0, i32 5
1425 // CHECK3-NEXT: store ptr @.offload_maptypes.6, ptr [[TMP70]], align 4
1426 // CHECK3-NEXT: [[TMP71:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS23]], i32 0, i32 6
1427 // CHECK3-NEXT: store ptr null, ptr [[TMP71]], align 4
1428 // CHECK3-NEXT: [[TMP72:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS23]], i32 0, i32 7
1429 // CHECK3-NEXT: store ptr null, ptr [[TMP72]], align 4
1430 // CHECK3-NEXT: [[TMP73:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS23]], i32 0, i32 8
1431 // CHECK3-NEXT: store i64 123, ptr [[TMP73]], align 8
1432 // CHECK3-NEXT: [[TMP74:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS23]], i32 0, i32 9
1433 // CHECK3-NEXT: store i64 0, ptr [[TMP74]], align 8
1434 // CHECK3-NEXT: [[TMP75:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS23]], i32 0, i32 10
1435 // CHECK3-NEXT: store [3 x i32] zeroinitializer, ptr [[TMP75]], align 4
1436 // CHECK3-NEXT: [[TMP76:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS23]], i32 0, i32 11
1437 // CHECK3-NEXT: store [3 x i32] zeroinitializer, ptr [[TMP76]], align 4
1438 // CHECK3-NEXT: [[TMP77:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS23]], i32 0, i32 12
1439 // CHECK3-NEXT: store i32 0, ptr [[TMP77]], align 4
1440 // CHECK3-NEXT: [[TMP78:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB3]], i64 -1, i32 0, i32 0, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l51.region_id, ptr [[KERNEL_ARGS23]])
1441 // CHECK3-NEXT: [[TMP79:%.*]] = icmp ne i32 [[TMP78]], 0
1442 // CHECK3-NEXT: br i1 [[TMP79]], label [[OMP_OFFLOAD_FAILED24:%.*]], label [[OMP_OFFLOAD_CONT25:%.*]]
1443 // CHECK3: omp_offload.failed24:
1444 // CHECK3-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l51(ptr [[THIS1]]) #[[ATTR2]]
1445 // CHECK3-NEXT: br label [[OMP_OFFLOAD_CONT25]]
1446 // CHECK3: omp_offload.cont25:
1447 // CHECK3-NEXT: [[A26:%.*]] = getelementptr inbounds nuw [[STRUCT_SS]], ptr [[THIS1]], i32 0, i32 0
1448 // CHECK3-NEXT: [[TMP80:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS27]], i32 0, i32 0
1449 // CHECK3-NEXT: store ptr [[THIS1]], ptr [[TMP80]], align 4
1450 // CHECK3-NEXT: [[TMP81:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS28]], i32 0, i32 0
1451 // CHECK3-NEXT: store ptr [[A26]], ptr [[TMP81]], align 4
1452 // CHECK3-NEXT: [[TMP82:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_MAPPERS29]], i32 0, i32 0
1453 // CHECK3-NEXT: store ptr null, ptr [[TMP82]], align 4
1454 // CHECK3-NEXT: [[TMP83:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS27]], i32 0, i32 0
1455 // CHECK3-NEXT: [[TMP84:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS28]], i32 0, i32 0
1456 // CHECK3-NEXT: [[TMP85:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS31]], i32 0, i32 0
1457 // CHECK3-NEXT: store i32 3, ptr [[TMP85]], align 4
1458 // CHECK3-NEXT: [[TMP86:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS31]], i32 0, i32 1
1459 // CHECK3-NEXT: store i32 1, ptr [[TMP86]], align 4
1460 // CHECK3-NEXT: [[TMP87:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS31]], i32 0, i32 2
1461 // CHECK3-NEXT: store ptr [[TMP83]], ptr [[TMP87]], align 4
1462 // CHECK3-NEXT: [[TMP88:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS31]], i32 0, i32 3
1463 // CHECK3-NEXT: store ptr [[TMP84]], ptr [[TMP88]], align 4
1464 // CHECK3-NEXT: [[TMP89:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS31]], i32 0, i32 4
1465 // CHECK3-NEXT: store ptr @.offload_sizes.7, ptr [[TMP89]], align 4
1466 // CHECK3-NEXT: [[TMP90:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS31]], i32 0, i32 5
1467 // CHECK3-NEXT: store ptr @.offload_maptypes.8, ptr [[TMP90]], align 4
1468 // CHECK3-NEXT: [[TMP91:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS31]], i32 0, i32 6
1469 // CHECK3-NEXT: store ptr null, ptr [[TMP91]], align 4
1470 // CHECK3-NEXT: [[TMP92:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS31]], i32 0, i32 7
1471 // CHECK3-NEXT: store ptr null, ptr [[TMP92]], align 4
1472 // CHECK3-NEXT: [[TMP93:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS31]], i32 0, i32 8
1473 // CHECK3-NEXT: store i64 123, ptr [[TMP93]], align 8
1474 // CHECK3-NEXT: [[TMP94:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS31]], i32 0, i32 9
1475 // CHECK3-NEXT: store i64 0, ptr [[TMP94]], align 8
1476 // CHECK3-NEXT: [[TMP95:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS31]], i32 0, i32 10
1477 // CHECK3-NEXT: store [3 x i32] zeroinitializer, ptr [[TMP95]], align 4
1478 // CHECK3-NEXT: [[TMP96:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS31]], i32 0, i32 11
1479 // CHECK3-NEXT: store [3 x i32] zeroinitializer, ptr [[TMP96]], align 4
1480 // CHECK3-NEXT: [[TMP97:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS31]], i32 0, i32 12
1481 // CHECK3-NEXT: store i32 0, ptr [[TMP97]], align 4
1482 // CHECK3-NEXT: [[TMP98:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB3]], i64 -1, i32 0, i32 0, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l57.region_id, ptr [[KERNEL_ARGS31]])
1483 // CHECK3-NEXT: [[TMP99:%.*]] = icmp ne i32 [[TMP98]], 0
1484 // CHECK3-NEXT: br i1 [[TMP99]], label [[OMP_OFFLOAD_FAILED32:%.*]], label [[OMP_OFFLOAD_CONT33:%.*]]
1485 // CHECK3: omp_offload.failed32:
1486 // CHECK3-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l57(ptr [[THIS1]]) #[[ATTR2]]
1487 // CHECK3-NEXT: br label [[OMP_OFFLOAD_CONT33]]
1488 // CHECK3: omp_offload.cont33:
1489 // CHECK3-NEXT: [[A34:%.*]] = getelementptr inbounds nuw [[STRUCT_SS]], ptr [[THIS1]], i32 0, i32 0
1490 // CHECK3-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds [123 x i32], ptr [[A34]], i32 0, i32 0
1491 // CHECK3-NEXT: [[TMP100:%.*]] = load i32, ptr [[ARRAYIDX]], align 4
1492 // CHECK3-NEXT: ret i32 [[TMP100]]
1495 // CHECK3-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l35
1496 // CHECK3-SAME: (ptr noundef [[THIS:%.*]]) #[[ATTR1:[0-9]+]] {
1497 // CHECK3-NEXT: entry:
1498 // CHECK3-NEXT: [[THIS_ADDR:%.*]] = alloca ptr, align 4
1499 // CHECK3-NEXT: store ptr [[THIS]], ptr [[THIS_ADDR]], align 4
1500 // CHECK3-NEXT: [[TMP0:%.*]] = load ptr, ptr [[THIS_ADDR]], align 4
1501 // CHECK3-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_teams(ptr @[[GLOB3]], i32 1, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l35.omp_outlined, ptr [[TMP0]])
1502 // CHECK3-NEXT: ret void
1505 // CHECK3-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l35.omp_outlined
1506 // CHECK3-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], ptr noundef [[THIS:%.*]]) #[[ATTR1]] {
1507 // CHECK3-NEXT: entry:
1508 // CHECK3-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 4
1509 // CHECK3-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 4
1510 // CHECK3-NEXT: [[THIS_ADDR:%.*]] = alloca ptr, align 4
1511 // CHECK3-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
1512 // CHECK3-NEXT: [[TMP:%.*]] = alloca i32, align 4
1513 // CHECK3-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4
1514 // CHECK3-NEXT: [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4
1515 // CHECK3-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
1516 // CHECK3-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
1517 // CHECK3-NEXT: [[I:%.*]] = alloca i32, align 4
1518 // CHECK3-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 4
1519 // CHECK3-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 4
1520 // CHECK3-NEXT: store ptr [[THIS]], ptr [[THIS_ADDR]], align 4
1521 // CHECK3-NEXT: [[TMP0:%.*]] = load ptr, ptr [[THIS_ADDR]], align 4
1522 // CHECK3-NEXT: store i32 0, ptr [[DOTOMP_COMB_LB]], align 4
1523 // CHECK3-NEXT: store i32 122, ptr [[DOTOMP_COMB_UB]], align 4
1524 // CHECK3-NEXT: store i32 1, ptr [[DOTOMP_STRIDE]], align 4
1525 // CHECK3-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
1526 // CHECK3-NEXT: [[TMP1:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
1527 // CHECK3-NEXT: [[TMP2:%.*]] = load i32, ptr [[TMP1]], align 4
1528 // CHECK3-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB1:[0-9]+]], i32 [[TMP2]], i32 92, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
1529 // CHECK3-NEXT: [[TMP3:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
1530 // CHECK3-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP3]], 122
1531 // CHECK3-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
1532 // CHECK3: cond.true:
1533 // CHECK3-NEXT: br label [[COND_END:%.*]]
1534 // CHECK3: cond.false:
1535 // CHECK3-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
1536 // CHECK3-NEXT: br label [[COND_END]]
1537 // CHECK3: cond.end:
1538 // CHECK3-NEXT: [[COND:%.*]] = phi i32 [ 122, [[COND_TRUE]] ], [ [[TMP4]], [[COND_FALSE]] ]
1539 // CHECK3-NEXT: store i32 [[COND]], ptr [[DOTOMP_COMB_UB]], align 4
1540 // CHECK3-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
1541 // CHECK3-NEXT: store i32 [[TMP5]], ptr [[DOTOMP_IV]], align 4
1542 // CHECK3-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
1543 // CHECK3: omp.inner.for.cond:
1544 // CHECK3-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
1545 // CHECK3-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
1546 // CHECK3-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP6]], [[TMP7]]
1547 // CHECK3-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
1548 // CHECK3: omp.inner.for.body:
1549 // CHECK3-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
1550 // CHECK3-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
1551 // CHECK3-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_call(ptr @[[GLOB3]], i32 3, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l35.omp_outlined.omp_outlined, i32 [[TMP8]], i32 [[TMP9]], ptr [[TMP0]])
1552 // CHECK3-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
1553 // CHECK3: omp.inner.for.inc:
1554 // CHECK3-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
1555 // CHECK3-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4
1556 // CHECK3-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP10]], [[TMP11]]
1557 // CHECK3-NEXT: store i32 [[ADD]], ptr [[DOTOMP_IV]], align 4
1558 // CHECK3-NEXT: br label [[OMP_INNER_FOR_COND]]
1559 // CHECK3: omp.inner.for.end:
1560 // CHECK3-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
1561 // CHECK3: omp.loop.exit:
1562 // CHECK3-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB1]], i32 [[TMP2]])
1563 // CHECK3-NEXT: ret void
1566 // CHECK3-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l35.omp_outlined.omp_outlined
1567 // CHECK3-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], i32 noundef [[DOTPREVIOUS_LB_:%.*]], i32 noundef [[DOTPREVIOUS_UB_:%.*]], ptr noundef [[THIS:%.*]]) #[[ATTR1]] {
1568 // CHECK3-NEXT: entry:
1569 // CHECK3-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 4
1570 // CHECK3-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 4
1571 // CHECK3-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i32, align 4
1572 // CHECK3-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i32, align 4
1573 // CHECK3-NEXT: [[THIS_ADDR:%.*]] = alloca ptr, align 4
1574 // CHECK3-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
1575 // CHECK3-NEXT: [[TMP:%.*]] = alloca i32, align 4
1576 // CHECK3-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4
1577 // CHECK3-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4
1578 // CHECK3-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
1579 // CHECK3-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
1580 // CHECK3-NEXT: [[I:%.*]] = alloca i32, align 4
1581 // CHECK3-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 4
1582 // CHECK3-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 4
1583 // CHECK3-NEXT: store i32 [[DOTPREVIOUS_LB_]], ptr [[DOTPREVIOUS_LB__ADDR]], align 4
1584 // CHECK3-NEXT: store i32 [[DOTPREVIOUS_UB_]], ptr [[DOTPREVIOUS_UB__ADDR]], align 4
1585 // CHECK3-NEXT: store ptr [[THIS]], ptr [[THIS_ADDR]], align 4
1586 // CHECK3-NEXT: [[TMP0:%.*]] = load ptr, ptr [[THIS_ADDR]], align 4
1587 // CHECK3-NEXT: store i32 0, ptr [[DOTOMP_LB]], align 4
1588 // CHECK3-NEXT: store i32 122, ptr [[DOTOMP_UB]], align 4
1589 // CHECK3-NEXT: [[TMP1:%.*]] = load i32, ptr [[DOTPREVIOUS_LB__ADDR]], align 4
1590 // CHECK3-NEXT: [[TMP2:%.*]] = load i32, ptr [[DOTPREVIOUS_UB__ADDR]], align 4
1591 // CHECK3-NEXT: store i32 [[TMP1]], ptr [[DOTOMP_LB]], align 4
1592 // CHECK3-NEXT: store i32 [[TMP2]], ptr [[DOTOMP_UB]], align 4
1593 // CHECK3-NEXT: store i32 1, ptr [[DOTOMP_STRIDE]], align 4
1594 // CHECK3-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
1595 // CHECK3-NEXT: [[TMP3:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
1596 // CHECK3-NEXT: [[TMP4:%.*]] = load i32, ptr [[TMP3]], align 4
1597 // CHECK3-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB2:[0-9]+]], i32 [[TMP4]], i32 34, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
1598 // CHECK3-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
1599 // CHECK3-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP5]], 122
1600 // CHECK3-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
1601 // CHECK3: cond.true:
1602 // CHECK3-NEXT: br label [[COND_END:%.*]]
1603 // CHECK3: cond.false:
1604 // CHECK3-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
1605 // CHECK3-NEXT: br label [[COND_END]]
1606 // CHECK3: cond.end:
1607 // CHECK3-NEXT: [[COND:%.*]] = phi i32 [ 122, [[COND_TRUE]] ], [ [[TMP6]], [[COND_FALSE]] ]
1608 // CHECK3-NEXT: store i32 [[COND]], ptr [[DOTOMP_UB]], align 4
1609 // CHECK3-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4
1610 // CHECK3-NEXT: store i32 [[TMP7]], ptr [[DOTOMP_IV]], align 4
1611 // CHECK3-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
1612 // CHECK3: omp.inner.for.cond:
1613 // CHECK3-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
1614 // CHECK3-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
1615 // CHECK3-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP8]], [[TMP9]]
1616 // CHECK3-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
1617 // CHECK3: omp.inner.for.body:
1618 // CHECK3-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
1619 // CHECK3-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP10]], 1
1620 // CHECK3-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
1621 // CHECK3-NEXT: store i32 [[ADD]], ptr [[I]], align 4
1622 // CHECK3-NEXT: [[A:%.*]] = getelementptr inbounds nuw [[STRUCT_SS:%.*]], ptr [[TMP0]], i32 0, i32 0
1623 // CHECK3-NEXT: [[TMP11:%.*]] = load i32, ptr [[I]], align 4
1624 // CHECK3-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds [123 x i32], ptr [[A]], i32 0, i32 [[TMP11]]
1625 // CHECK3-NEXT: store i32 0, ptr [[ARRAYIDX]], align 4
1626 // CHECK3-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
1627 // CHECK3: omp.body.continue:
1628 // CHECK3-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
1629 // CHECK3: omp.inner.for.inc:
1630 // CHECK3-NEXT: [[TMP12:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
1631 // CHECK3-NEXT: [[ADD2:%.*]] = add nsw i32 [[TMP12]], 1
1632 // CHECK3-NEXT: store i32 [[ADD2]], ptr [[DOTOMP_IV]], align 4
1633 // CHECK3-NEXT: br label [[OMP_INNER_FOR_COND]]
1634 // CHECK3: omp.inner.for.end:
1635 // CHECK3-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
1636 // CHECK3: omp.loop.exit:
1637 // CHECK3-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB2]], i32 [[TMP4]])
1638 // CHECK3-NEXT: ret void
1641 // CHECK3-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l40
1642 // CHECK3-SAME: (ptr noundef [[THIS:%.*]]) #[[ATTR1]] {
1643 // CHECK3-NEXT: entry:
1644 // CHECK3-NEXT: [[THIS_ADDR:%.*]] = alloca ptr, align 4
1645 // CHECK3-NEXT: store ptr [[THIS]], ptr [[THIS_ADDR]], align 4
1646 // CHECK3-NEXT: [[TMP0:%.*]] = load ptr, ptr [[THIS_ADDR]], align 4
1647 // CHECK3-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_teams(ptr @[[GLOB3]], i32 1, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l40.omp_outlined, ptr [[TMP0]])
1648 // CHECK3-NEXT: ret void
1651 // CHECK3-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l40.omp_outlined
1652 // CHECK3-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], ptr noundef [[THIS:%.*]]) #[[ATTR1]] {
1653 // CHECK3-NEXT: entry:
1654 // CHECK3-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 4
1655 // CHECK3-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 4
1656 // CHECK3-NEXT: [[THIS_ADDR:%.*]] = alloca ptr, align 4
1657 // CHECK3-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
1658 // CHECK3-NEXT: [[TMP:%.*]] = alloca i32, align 4
1659 // CHECK3-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4
1660 // CHECK3-NEXT: [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4
1661 // CHECK3-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
1662 // CHECK3-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
1663 // CHECK3-NEXT: [[I:%.*]] = alloca i32, align 4
1664 // CHECK3-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 4
1665 // CHECK3-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 4
1666 // CHECK3-NEXT: store ptr [[THIS]], ptr [[THIS_ADDR]], align 4
1667 // CHECK3-NEXT: [[TMP0:%.*]] = load ptr, ptr [[THIS_ADDR]], align 4
1668 // CHECK3-NEXT: store i32 0, ptr [[DOTOMP_COMB_LB]], align 4
1669 // CHECK3-NEXT: store i32 122, ptr [[DOTOMP_COMB_UB]], align 4
1670 // CHECK3-NEXT: store i32 1, ptr [[DOTOMP_STRIDE]], align 4
1671 // CHECK3-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
1672 // CHECK3-NEXT: [[TMP1:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
1673 // CHECK3-NEXT: [[TMP2:%.*]] = load i32, ptr [[TMP1]], align 4
1674 // CHECK3-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB1]], i32 [[TMP2]], i32 92, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
1675 // CHECK3-NEXT: [[TMP3:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
1676 // CHECK3-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP3]], 122
1677 // CHECK3-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
1678 // CHECK3: cond.true:
1679 // CHECK3-NEXT: br label [[COND_END:%.*]]
1680 // CHECK3: cond.false:
1681 // CHECK3-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
1682 // CHECK3-NEXT: br label [[COND_END]]
1683 // CHECK3: cond.end:
1684 // CHECK3-NEXT: [[COND:%.*]] = phi i32 [ 122, [[COND_TRUE]] ], [ [[TMP4]], [[COND_FALSE]] ]
1685 // CHECK3-NEXT: store i32 [[COND]], ptr [[DOTOMP_COMB_UB]], align 4
1686 // CHECK3-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
1687 // CHECK3-NEXT: store i32 [[TMP5]], ptr [[DOTOMP_IV]], align 4
1688 // CHECK3-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
1689 // CHECK3: omp.inner.for.cond:
1690 // CHECK3-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
1691 // CHECK3-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
1692 // CHECK3-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP6]], [[TMP7]]
1693 // CHECK3-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
1694 // CHECK3: omp.inner.for.body:
1695 // CHECK3-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
1696 // CHECK3-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
1697 // CHECK3-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_call(ptr @[[GLOB3]], i32 3, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l40.omp_outlined.omp_outlined, i32 [[TMP8]], i32 [[TMP9]], ptr [[TMP0]])
1698 // CHECK3-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
1699 // CHECK3: omp.inner.for.inc:
1700 // CHECK3-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
1701 // CHECK3-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4
1702 // CHECK3-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP10]], [[TMP11]]
1703 // CHECK3-NEXT: store i32 [[ADD]], ptr [[DOTOMP_IV]], align 4
1704 // CHECK3-NEXT: br label [[OMP_INNER_FOR_COND]]
1705 // CHECK3: omp.inner.for.end:
1706 // CHECK3-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
1707 // CHECK3: omp.loop.exit:
1708 // CHECK3-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB1]], i32 [[TMP2]])
1709 // CHECK3-NEXT: ret void
1712 // CHECK3-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l40.omp_outlined.omp_outlined
1713 // CHECK3-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], i32 noundef [[DOTPREVIOUS_LB_:%.*]], i32 noundef [[DOTPREVIOUS_UB_:%.*]], ptr noundef [[THIS:%.*]]) #[[ATTR1]] {
1714 // CHECK3-NEXT: entry:
1715 // CHECK3-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 4
1716 // CHECK3-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 4
1717 // CHECK3-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i32, align 4
1718 // CHECK3-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i32, align 4
1719 // CHECK3-NEXT: [[THIS_ADDR:%.*]] = alloca ptr, align 4
1720 // CHECK3-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
1721 // CHECK3-NEXT: [[TMP:%.*]] = alloca i32, align 4
1722 // CHECK3-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4
1723 // CHECK3-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4
1724 // CHECK3-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
1725 // CHECK3-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
1726 // CHECK3-NEXT: [[I:%.*]] = alloca i32, align 4
1727 // CHECK3-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 4
1728 // CHECK3-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 4
1729 // CHECK3-NEXT: store i32 [[DOTPREVIOUS_LB_]], ptr [[DOTPREVIOUS_LB__ADDR]], align 4
1730 // CHECK3-NEXT: store i32 [[DOTPREVIOUS_UB_]], ptr [[DOTPREVIOUS_UB__ADDR]], align 4
1731 // CHECK3-NEXT: store ptr [[THIS]], ptr [[THIS_ADDR]], align 4
1732 // CHECK3-NEXT: [[TMP0:%.*]] = load ptr, ptr [[THIS_ADDR]], align 4
1733 // CHECK3-NEXT: store i32 0, ptr [[DOTOMP_LB]], align 4
1734 // CHECK3-NEXT: store i32 122, ptr [[DOTOMP_UB]], align 4
1735 // CHECK3-NEXT: [[TMP1:%.*]] = load i32, ptr [[DOTPREVIOUS_LB__ADDR]], align 4
1736 // CHECK3-NEXT: [[TMP2:%.*]] = load i32, ptr [[DOTPREVIOUS_UB__ADDR]], align 4
1737 // CHECK3-NEXT: store i32 [[TMP1]], ptr [[DOTOMP_LB]], align 4
1738 // CHECK3-NEXT: store i32 [[TMP2]], ptr [[DOTOMP_UB]], align 4
1739 // CHECK3-NEXT: store i32 1, ptr [[DOTOMP_STRIDE]], align 4
1740 // CHECK3-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
1741 // CHECK3-NEXT: [[TMP3:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
1742 // CHECK3-NEXT: [[TMP4:%.*]] = load i32, ptr [[TMP3]], align 4
1743 // CHECK3-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB2]], i32 [[TMP4]], i32 34, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
1744 // CHECK3-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
1745 // CHECK3-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP5]], 122
1746 // CHECK3-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
1747 // CHECK3: cond.true:
1748 // CHECK3-NEXT: br label [[COND_END:%.*]]
1749 // CHECK3: cond.false:
1750 // CHECK3-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
1751 // CHECK3-NEXT: br label [[COND_END]]
1752 // CHECK3: cond.end:
1753 // CHECK3-NEXT: [[COND:%.*]] = phi i32 [ 122, [[COND_TRUE]] ], [ [[TMP6]], [[COND_FALSE]] ]
1754 // CHECK3-NEXT: store i32 [[COND]], ptr [[DOTOMP_UB]], align 4
1755 // CHECK3-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4
1756 // CHECK3-NEXT: store i32 [[TMP7]], ptr [[DOTOMP_IV]], align 4
1757 // CHECK3-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
1758 // CHECK3: omp.inner.for.cond:
1759 // CHECK3-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
1760 // CHECK3-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
1761 // CHECK3-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP8]], [[TMP9]]
1762 // CHECK3-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
1763 // CHECK3: omp.inner.for.body:
1764 // CHECK3-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
1765 // CHECK3-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP10]], 1
1766 // CHECK3-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
1767 // CHECK3-NEXT: store i32 [[ADD]], ptr [[I]], align 4
1768 // CHECK3-NEXT: [[A:%.*]] = getelementptr inbounds nuw [[STRUCT_SS:%.*]], ptr [[TMP0]], i32 0, i32 0
1769 // CHECK3-NEXT: [[TMP11:%.*]] = load i32, ptr [[I]], align 4
1770 // CHECK3-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds [123 x i32], ptr [[A]], i32 0, i32 [[TMP11]]
1771 // CHECK3-NEXT: store i32 0, ptr [[ARRAYIDX]], align 4
1772 // CHECK3-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
1773 // CHECK3: omp.body.continue:
1774 // CHECK3-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
1775 // CHECK3: omp.inner.for.inc:
1776 // CHECK3-NEXT: [[TMP12:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
1777 // CHECK3-NEXT: [[ADD2:%.*]] = add nsw i32 [[TMP12]], 1
1778 // CHECK3-NEXT: store i32 [[ADD2]], ptr [[DOTOMP_IV]], align 4
1779 // CHECK3-NEXT: br label [[OMP_INNER_FOR_COND]]
1780 // CHECK3: omp.inner.for.end:
1781 // CHECK3-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
1782 // CHECK3: omp.loop.exit:
1783 // CHECK3-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB2]], i32 [[TMP4]])
1784 // CHECK3-NEXT: ret void
1787 // CHECK3-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l45
1788 // CHECK3-SAME: (ptr noundef [[THIS:%.*]]) #[[ATTR1]] {
1789 // CHECK3-NEXT: entry:
1790 // CHECK3-NEXT: [[THIS_ADDR:%.*]] = alloca ptr, align 4
1791 // CHECK3-NEXT: store ptr [[THIS]], ptr [[THIS_ADDR]], align 4
1792 // CHECK3-NEXT: [[TMP0:%.*]] = load ptr, ptr [[THIS_ADDR]], align 4
1793 // CHECK3-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_teams(ptr @[[GLOB3]], i32 1, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l45.omp_outlined, ptr [[TMP0]])
1794 // CHECK3-NEXT: ret void
1797 // CHECK3-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l45.omp_outlined
1798 // CHECK3-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], ptr noundef [[THIS:%.*]]) #[[ATTR1]] {
1799 // CHECK3-NEXT: entry:
1800 // CHECK3-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 4
1801 // CHECK3-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 4
1802 // CHECK3-NEXT: [[THIS_ADDR:%.*]] = alloca ptr, align 4
1803 // CHECK3-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
1804 // CHECK3-NEXT: [[TMP:%.*]] = alloca i32, align 4
1805 // CHECK3-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4
1806 // CHECK3-NEXT: [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4
1807 // CHECK3-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
1808 // CHECK3-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
1809 // CHECK3-NEXT: [[I:%.*]] = alloca i32, align 4
1810 // CHECK3-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 4
1811 // CHECK3-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 4
1812 // CHECK3-NEXT: store ptr [[THIS]], ptr [[THIS_ADDR]], align 4
1813 // CHECK3-NEXT: [[TMP0:%.*]] = load ptr, ptr [[THIS_ADDR]], align 4
1814 // CHECK3-NEXT: store i32 0, ptr [[DOTOMP_COMB_LB]], align 4
1815 // CHECK3-NEXT: store i32 122, ptr [[DOTOMP_COMB_UB]], align 4
1816 // CHECK3-NEXT: store i32 1, ptr [[DOTOMP_STRIDE]], align 4
1817 // CHECK3-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
1818 // CHECK3-NEXT: [[TMP1:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
1819 // CHECK3-NEXT: [[TMP2:%.*]] = load i32, ptr [[TMP1]], align 4
1820 // CHECK3-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB1]], i32 [[TMP2]], i32 92, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
1821 // CHECK3-NEXT: [[TMP3:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
1822 // CHECK3-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP3]], 122
1823 // CHECK3-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
1824 // CHECK3: cond.true:
1825 // CHECK3-NEXT: br label [[COND_END:%.*]]
1826 // CHECK3: cond.false:
1827 // CHECK3-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
1828 // CHECK3-NEXT: br label [[COND_END]]
1829 // CHECK3: cond.end:
1830 // CHECK3-NEXT: [[COND:%.*]] = phi i32 [ 122, [[COND_TRUE]] ], [ [[TMP4]], [[COND_FALSE]] ]
1831 // CHECK3-NEXT: store i32 [[COND]], ptr [[DOTOMP_COMB_UB]], align 4
1832 // CHECK3-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
1833 // CHECK3-NEXT: store i32 [[TMP5]], ptr [[DOTOMP_IV]], align 4
1834 // CHECK3-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
1835 // CHECK3: omp.inner.for.cond:
1836 // CHECK3-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
1837 // CHECK3-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
1838 // CHECK3-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP6]], [[TMP7]]
1839 // CHECK3-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
1840 // CHECK3: omp.inner.for.body:
1841 // CHECK3-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
1842 // CHECK3-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
1843 // CHECK3-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_call(ptr @[[GLOB3]], i32 3, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l45.omp_outlined.omp_outlined, i32 [[TMP8]], i32 [[TMP9]], ptr [[TMP0]])
1844 // CHECK3-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
1845 // CHECK3: omp.inner.for.inc:
1846 // CHECK3-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
1847 // CHECK3-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4
1848 // CHECK3-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP10]], [[TMP11]]
1849 // CHECK3-NEXT: store i32 [[ADD]], ptr [[DOTOMP_IV]], align 4
1850 // CHECK3-NEXT: br label [[OMP_INNER_FOR_COND]]
1851 // CHECK3: omp.inner.for.end:
1852 // CHECK3-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
1853 // CHECK3: omp.loop.exit:
1854 // CHECK3-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB1]], i32 [[TMP2]])
1855 // CHECK3-NEXT: ret void
1858 // CHECK3-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l45.omp_outlined.omp_outlined
1859 // CHECK3-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], i32 noundef [[DOTPREVIOUS_LB_:%.*]], i32 noundef [[DOTPREVIOUS_UB_:%.*]], ptr noundef [[THIS:%.*]]) #[[ATTR1]] {
1860 // CHECK3-NEXT: entry:
1861 // CHECK3-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 4
1862 // CHECK3-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 4
1863 // CHECK3-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i32, align 4
1864 // CHECK3-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i32, align 4
1865 // CHECK3-NEXT: [[THIS_ADDR:%.*]] = alloca ptr, align 4
1866 // CHECK3-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
1867 // CHECK3-NEXT: [[TMP:%.*]] = alloca i32, align 4
1868 // CHECK3-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4
1869 // CHECK3-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4
1870 // CHECK3-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
1871 // CHECK3-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
1872 // CHECK3-NEXT: [[I:%.*]] = alloca i32, align 4
1873 // CHECK3-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 4
1874 // CHECK3-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 4
1875 // CHECK3-NEXT: store i32 [[DOTPREVIOUS_LB_]], ptr [[DOTPREVIOUS_LB__ADDR]], align 4
1876 // CHECK3-NEXT: store i32 [[DOTPREVIOUS_UB_]], ptr [[DOTPREVIOUS_UB__ADDR]], align 4
1877 // CHECK3-NEXT: store ptr [[THIS]], ptr [[THIS_ADDR]], align 4
1878 // CHECK3-NEXT: [[TMP0:%.*]] = load ptr, ptr [[THIS_ADDR]], align 4
1879 // CHECK3-NEXT: store i32 0, ptr [[DOTOMP_LB]], align 4
1880 // CHECK3-NEXT: store i32 122, ptr [[DOTOMP_UB]], align 4
1881 // CHECK3-NEXT: [[TMP1:%.*]] = load i32, ptr [[DOTPREVIOUS_LB__ADDR]], align 4
1882 // CHECK3-NEXT: [[TMP2:%.*]] = load i32, ptr [[DOTPREVIOUS_UB__ADDR]], align 4
1883 // CHECK3-NEXT: store i32 [[TMP1]], ptr [[DOTOMP_LB]], align 4
1884 // CHECK3-NEXT: store i32 [[TMP2]], ptr [[DOTOMP_UB]], align 4
1885 // CHECK3-NEXT: store i32 1, ptr [[DOTOMP_STRIDE]], align 4
1886 // CHECK3-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
1887 // CHECK3-NEXT: [[TMP3:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
1888 // CHECK3-NEXT: [[TMP4:%.*]] = load i32, ptr [[TMP3]], align 4
1889 // CHECK3-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB2]], i32 [[TMP4]], i32 33, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 61)
1890 // CHECK3-NEXT: br label [[OMP_DISPATCH_COND:%.*]]
1891 // CHECK3: omp.dispatch.cond:
1892 // CHECK3-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
1893 // CHECK3-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTPREVIOUS_UB__ADDR]], align 4
1894 // CHECK3-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP5]], [[TMP6]]
1895 // CHECK3-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
1896 // CHECK3: cond.true:
1897 // CHECK3-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTPREVIOUS_UB__ADDR]], align 4
1898 // CHECK3-NEXT: br label [[COND_END:%.*]]
1899 // CHECK3: cond.false:
1900 // CHECK3-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
1901 // CHECK3-NEXT: br label [[COND_END]]
1902 // CHECK3: cond.end:
1903 // CHECK3-NEXT: [[COND:%.*]] = phi i32 [ [[TMP7]], [[COND_TRUE]] ], [ [[TMP8]], [[COND_FALSE]] ]
1904 // CHECK3-NEXT: store i32 [[COND]], ptr [[DOTOMP_UB]], align 4
1905 // CHECK3-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4
1906 // CHECK3-NEXT: store i32 [[TMP9]], ptr [[DOTOMP_IV]], align 4
1907 // CHECK3-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
1908 // CHECK3-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
1909 // CHECK3-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP10]], [[TMP11]]
1910 // CHECK3-NEXT: br i1 [[CMP1]], label [[OMP_DISPATCH_BODY:%.*]], label [[OMP_DISPATCH_END:%.*]]
1911 // CHECK3: omp.dispatch.body:
1912 // CHECK3-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
1913 // CHECK3: omp.inner.for.cond:
1914 // CHECK3-NEXT: [[TMP12:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
1915 // CHECK3-NEXT: [[TMP13:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
1916 // CHECK3-NEXT: [[CMP2:%.*]] = icmp sle i32 [[TMP12]], [[TMP13]]
1917 // CHECK3-NEXT: br i1 [[CMP2]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
1918 // CHECK3: omp.inner.for.body:
1919 // CHECK3-NEXT: [[TMP14:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
1920 // CHECK3-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP14]], 1
1921 // CHECK3-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
1922 // CHECK3-NEXT: store i32 [[ADD]], ptr [[I]], align 4
1923 // CHECK3-NEXT: [[A:%.*]] = getelementptr inbounds nuw [[STRUCT_SS:%.*]], ptr [[TMP0]], i32 0, i32 0
1924 // CHECK3-NEXT: [[TMP15:%.*]] = load i32, ptr [[I]], align 4
1925 // CHECK3-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds [123 x i32], ptr [[A]], i32 0, i32 [[TMP15]]
1926 // CHECK3-NEXT: store i32 0, ptr [[ARRAYIDX]], align 4
1927 // CHECK3-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
1928 // CHECK3: omp.body.continue:
1929 // CHECK3-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
1930 // CHECK3: omp.inner.for.inc:
1931 // CHECK3-NEXT: [[TMP16:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
1932 // CHECK3-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP16]], 1
1933 // CHECK3-NEXT: store i32 [[ADD3]], ptr [[DOTOMP_IV]], align 4
1934 // CHECK3-NEXT: br label [[OMP_INNER_FOR_COND]]
1935 // CHECK3: omp.inner.for.end:
1936 // CHECK3-NEXT: br label [[OMP_DISPATCH_INC:%.*]]
1937 // CHECK3: omp.dispatch.inc:
1938 // CHECK3-NEXT: [[TMP17:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4
1939 // CHECK3-NEXT: [[TMP18:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4
1940 // CHECK3-NEXT: [[ADD4:%.*]] = add nsw i32 [[TMP17]], [[TMP18]]
1941 // CHECK3-NEXT: store i32 [[ADD4]], ptr [[DOTOMP_LB]], align 4
1942 // CHECK3-NEXT: [[TMP19:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
1943 // CHECK3-NEXT: [[TMP20:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4
1944 // CHECK3-NEXT: [[ADD5:%.*]] = add nsw i32 [[TMP19]], [[TMP20]]
1945 // CHECK3-NEXT: store i32 [[ADD5]], ptr [[DOTOMP_UB]], align 4
1946 // CHECK3-NEXT: br label [[OMP_DISPATCH_COND]]
1947 // CHECK3: omp.dispatch.end:
1948 // CHECK3-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB2]], i32 [[TMP4]])
1949 // CHECK3-NEXT: ret void
1952 // CHECK3-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l51
1953 // CHECK3-SAME: (ptr noundef [[THIS:%.*]]) #[[ATTR1]] {
1954 // CHECK3-NEXT: entry:
1955 // CHECK3-NEXT: [[THIS_ADDR:%.*]] = alloca ptr, align 4
1956 // CHECK3-NEXT: store ptr [[THIS]], ptr [[THIS_ADDR]], align 4
1957 // CHECK3-NEXT: [[TMP0:%.*]] = load ptr, ptr [[THIS_ADDR]], align 4
1958 // CHECK3-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_teams(ptr @[[GLOB3]], i32 1, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l51.omp_outlined, ptr [[TMP0]])
1959 // CHECK3-NEXT: ret void
1962 // CHECK3-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l51.omp_outlined
1963 // CHECK3-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], ptr noundef [[THIS:%.*]]) #[[ATTR1]] {
1964 // CHECK3-NEXT: entry:
1965 // CHECK3-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 4
1966 // CHECK3-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 4
1967 // CHECK3-NEXT: [[THIS_ADDR:%.*]] = alloca ptr, align 4
1968 // CHECK3-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
1969 // CHECK3-NEXT: [[TMP:%.*]] = alloca i32, align 4
1970 // CHECK3-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4
1971 // CHECK3-NEXT: [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4
1972 // CHECK3-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
1973 // CHECK3-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
1974 // CHECK3-NEXT: [[I:%.*]] = alloca i32, align 4
1975 // CHECK3-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 4
1976 // CHECK3-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 4
1977 // CHECK3-NEXT: store ptr [[THIS]], ptr [[THIS_ADDR]], align 4
1978 // CHECK3-NEXT: [[TMP0:%.*]] = load ptr, ptr [[THIS_ADDR]], align 4
1979 // CHECK3-NEXT: store i32 0, ptr [[DOTOMP_COMB_LB]], align 4
1980 // CHECK3-NEXT: store i32 122, ptr [[DOTOMP_COMB_UB]], align 4
1981 // CHECK3-NEXT: store i32 1, ptr [[DOTOMP_STRIDE]], align 4
1982 // CHECK3-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
1983 // CHECK3-NEXT: [[TMP1:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
1984 // CHECK3-NEXT: [[TMP2:%.*]] = load i32, ptr [[TMP1]], align 4
1985 // CHECK3-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB1]], i32 [[TMP2]], i32 92, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
1986 // CHECK3-NEXT: [[TMP3:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
1987 // CHECK3-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP3]], 122
1988 // CHECK3-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
1989 // CHECK3: cond.true:
1990 // CHECK3-NEXT: br label [[COND_END:%.*]]
1991 // CHECK3: cond.false:
1992 // CHECK3-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
1993 // CHECK3-NEXT: br label [[COND_END]]
1994 // CHECK3: cond.end:
1995 // CHECK3-NEXT: [[COND:%.*]] = phi i32 [ 122, [[COND_TRUE]] ], [ [[TMP4]], [[COND_FALSE]] ]
1996 // CHECK3-NEXT: store i32 [[COND]], ptr [[DOTOMP_COMB_UB]], align 4
1997 // CHECK3-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
1998 // CHECK3-NEXT: store i32 [[TMP5]], ptr [[DOTOMP_IV]], align 4
1999 // CHECK3-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
2000 // CHECK3: omp.inner.for.cond:
2001 // CHECK3-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
2002 // CHECK3-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
2003 // CHECK3-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP6]], [[TMP7]]
2004 // CHECK3-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
2005 // CHECK3: omp.inner.for.body:
2006 // CHECK3-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
2007 // CHECK3-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
2008 // CHECK3-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_call(ptr @[[GLOB3]], i32 3, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l51.omp_outlined.omp_outlined, i32 [[TMP8]], i32 [[TMP9]], ptr [[TMP0]])
2009 // CHECK3-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
2010 // CHECK3: omp.inner.for.inc:
2011 // CHECK3-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
2012 // CHECK3-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4
2013 // CHECK3-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP10]], [[TMP11]]
2014 // CHECK3-NEXT: store i32 [[ADD]], ptr [[DOTOMP_IV]], align 4
2015 // CHECK3-NEXT: br label [[OMP_INNER_FOR_COND]]
2016 // CHECK3: omp.inner.for.end:
2017 // CHECK3-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
2018 // CHECK3: omp.loop.exit:
2019 // CHECK3-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB1]], i32 [[TMP2]])
2020 // CHECK3-NEXT: ret void
2023 // CHECK3-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l51.omp_outlined.omp_outlined
2024 // CHECK3-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], i32 noundef [[DOTPREVIOUS_LB_:%.*]], i32 noundef [[DOTPREVIOUS_UB_:%.*]], ptr noundef [[THIS:%.*]]) #[[ATTR1]] {
2025 // CHECK3-NEXT: entry:
2026 // CHECK3-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 4
2027 // CHECK3-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 4
2028 // CHECK3-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i32, align 4
2029 // CHECK3-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i32, align 4
2030 // CHECK3-NEXT: [[THIS_ADDR:%.*]] = alloca ptr, align 4
2031 // CHECK3-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
2032 // CHECK3-NEXT: [[TMP:%.*]] = alloca i32, align 4
2033 // CHECK3-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4
2034 // CHECK3-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4
2035 // CHECK3-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
2036 // CHECK3-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
2037 // CHECK3-NEXT: [[I:%.*]] = alloca i32, align 4
2038 // CHECK3-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 4
2039 // CHECK3-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 4
2040 // CHECK3-NEXT: store i32 [[DOTPREVIOUS_LB_]], ptr [[DOTPREVIOUS_LB__ADDR]], align 4
2041 // CHECK3-NEXT: store i32 [[DOTPREVIOUS_UB_]], ptr [[DOTPREVIOUS_UB__ADDR]], align 4
2042 // CHECK3-NEXT: store ptr [[THIS]], ptr [[THIS_ADDR]], align 4
2043 // CHECK3-NEXT: [[TMP0:%.*]] = load ptr, ptr [[THIS_ADDR]], align 4
2044 // CHECK3-NEXT: store i32 0, ptr [[DOTOMP_LB]], align 4
2045 // CHECK3-NEXT: store i32 122, ptr [[DOTOMP_UB]], align 4
2046 // CHECK3-NEXT: [[TMP1:%.*]] = load i32, ptr [[DOTPREVIOUS_LB__ADDR]], align 4
2047 // CHECK3-NEXT: [[TMP2:%.*]] = load i32, ptr [[DOTPREVIOUS_UB__ADDR]], align 4
2048 // CHECK3-NEXT: store i32 [[TMP1]], ptr [[DOTOMP_LB]], align 4
2049 // CHECK3-NEXT: store i32 [[TMP2]], ptr [[DOTOMP_UB]], align 4
2050 // CHECK3-NEXT: store i32 1, ptr [[DOTOMP_STRIDE]], align 4
2051 // CHECK3-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
2052 // CHECK3-NEXT: [[TMP3:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4
2053 // CHECK3-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
2054 // CHECK3-NEXT: [[TMP5:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
2055 // CHECK3-NEXT: [[TMP6:%.*]] = load i32, ptr [[TMP5]], align 4
2056 // CHECK3-NEXT: call void @__kmpc_dispatch_init_4(ptr @[[GLOB3]], i32 [[TMP6]], i32 1073741859, i32 [[TMP3]], i32 [[TMP4]], i32 1, i32 1)
2057 // CHECK3-NEXT: br label [[OMP_DISPATCH_COND:%.*]]
2058 // CHECK3: omp.dispatch.cond:
2059 // CHECK3-NEXT: [[TMP7:%.*]] = call i32 @__kmpc_dispatch_next_4(ptr @[[GLOB3]], i32 [[TMP6]], ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]])
2060 // CHECK3-NEXT: [[TOBOOL:%.*]] = icmp ne i32 [[TMP7]], 0
2061 // CHECK3-NEXT: br i1 [[TOBOOL]], label [[OMP_DISPATCH_BODY:%.*]], label [[OMP_DISPATCH_END:%.*]]
2062 // CHECK3: omp.dispatch.body:
2063 // CHECK3-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4
2064 // CHECK3-NEXT: store i32 [[TMP8]], ptr [[DOTOMP_IV]], align 4
2065 // CHECK3-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
2066 // CHECK3: omp.inner.for.cond:
2067 // CHECK3-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP11:![0-9]+]]
2068 // CHECK3-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP11]]
2069 // CHECK3-NEXT: [[CMP:%.*]] = icmp sle i32 [[TMP9]], [[TMP10]]
2070 // CHECK3-NEXT: br i1 [[CMP]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
2071 // CHECK3: omp.inner.for.body:
2072 // CHECK3-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP11]]
2073 // CHECK3-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP11]], 1
2074 // CHECK3-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
2075 // CHECK3-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP11]]
2076 // CHECK3-NEXT: [[A:%.*]] = getelementptr inbounds nuw [[STRUCT_SS:%.*]], ptr [[TMP0]], i32 0, i32 0
2077 // CHECK3-NEXT: [[TMP12:%.*]] = load i32, ptr [[I]], align 4, !llvm.access.group [[ACC_GRP11]]
2078 // CHECK3-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds [123 x i32], ptr [[A]], i32 0, i32 [[TMP12]]
2079 // CHECK3-NEXT: store i32 0, ptr [[ARRAYIDX]], align 4, !llvm.access.group [[ACC_GRP11]]
2080 // CHECK3-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
2081 // CHECK3: omp.body.continue:
2082 // CHECK3-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
2083 // CHECK3: omp.inner.for.inc:
2084 // CHECK3-NEXT: [[TMP13:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP11]]
2085 // CHECK3-NEXT: [[ADD1:%.*]] = add nsw i32 [[TMP13]], 1
2086 // CHECK3-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP11]]
2087 // CHECK3-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP12:![0-9]+]]
2088 // CHECK3: omp.inner.for.end:
2089 // CHECK3-NEXT: br label [[OMP_DISPATCH_INC:%.*]]
2090 // CHECK3: omp.dispatch.inc:
2091 // CHECK3-NEXT: br label [[OMP_DISPATCH_COND]]
2092 // CHECK3: omp.dispatch.end:
2093 // CHECK3-NEXT: call void @__kmpc_dispatch_deinit(ptr @[[GLOB3]], i32 [[TMP6]])
2094 // CHECK3-NEXT: ret void
2097 // CHECK3-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l57
2098 // CHECK3-SAME: (ptr noundef [[THIS:%.*]]) #[[ATTR1]] {
2099 // CHECK3-NEXT: entry:
2100 // CHECK3-NEXT: [[THIS_ADDR:%.*]] = alloca ptr, align 4
2101 // CHECK3-NEXT: store ptr [[THIS]], ptr [[THIS_ADDR]], align 4
2102 // CHECK3-NEXT: [[TMP0:%.*]] = load ptr, ptr [[THIS_ADDR]], align 4
2103 // CHECK3-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_teams(ptr @[[GLOB3]], i32 1, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l57.omp_outlined, ptr [[TMP0]])
2104 // CHECK3-NEXT: ret void
2107 // CHECK3-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l57.omp_outlined
2108 // CHECK3-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], ptr noundef [[THIS:%.*]]) #[[ATTR1]] {
2109 // CHECK3-NEXT: entry:
2110 // CHECK3-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 4
2111 // CHECK3-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 4
2112 // CHECK3-NEXT: [[THIS_ADDR:%.*]] = alloca ptr, align 4
2113 // CHECK3-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
2114 // CHECK3-NEXT: [[TMP:%.*]] = alloca i32, align 4
2115 // CHECK3-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4
2116 // CHECK3-NEXT: [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4
2117 // CHECK3-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
2118 // CHECK3-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
2119 // CHECK3-NEXT: [[I:%.*]] = alloca i32, align 4
2120 // CHECK3-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 4
2121 // CHECK3-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 4
2122 // CHECK3-NEXT: store ptr [[THIS]], ptr [[THIS_ADDR]], align 4
2123 // CHECK3-NEXT: [[TMP0:%.*]] = load ptr, ptr [[THIS_ADDR]], align 4
2124 // CHECK3-NEXT: store i32 0, ptr [[DOTOMP_COMB_LB]], align 4
2125 // CHECK3-NEXT: store i32 122, ptr [[DOTOMP_COMB_UB]], align 4
2126 // CHECK3-NEXT: store i32 1, ptr [[DOTOMP_STRIDE]], align 4
2127 // CHECK3-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
2128 // CHECK3-NEXT: [[TMP1:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
2129 // CHECK3-NEXT: [[TMP2:%.*]] = load i32, ptr [[TMP1]], align 4
2130 // CHECK3-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB1]], i32 [[TMP2]], i32 92, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
2131 // CHECK3-NEXT: [[TMP3:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
2132 // CHECK3-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP3]], 122
2133 // CHECK3-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
2134 // CHECK3: cond.true:
2135 // CHECK3-NEXT: br label [[COND_END:%.*]]
2136 // CHECK3: cond.false:
2137 // CHECK3-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
2138 // CHECK3-NEXT: br label [[COND_END]]
2139 // CHECK3: cond.end:
2140 // CHECK3-NEXT: [[COND:%.*]] = phi i32 [ 122, [[COND_TRUE]] ], [ [[TMP4]], [[COND_FALSE]] ]
2141 // CHECK3-NEXT: store i32 [[COND]], ptr [[DOTOMP_COMB_UB]], align 4
2142 // CHECK3-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
2143 // CHECK3-NEXT: store i32 [[TMP5]], ptr [[DOTOMP_IV]], align 4
2144 // CHECK3-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
2145 // CHECK3: omp.inner.for.cond:
2146 // CHECK3-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
2147 // CHECK3-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
2148 // CHECK3-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP6]], [[TMP7]]
2149 // CHECK3-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
2150 // CHECK3: omp.inner.for.body:
2151 // CHECK3-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
2152 // CHECK3-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
2153 // CHECK3-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_call(ptr @[[GLOB3]], i32 3, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l57.omp_outlined.omp_outlined, i32 [[TMP8]], i32 [[TMP9]], ptr [[TMP0]])
2154 // CHECK3-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
2155 // CHECK3: omp.inner.for.inc:
2156 // CHECK3-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
2157 // CHECK3-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4
2158 // CHECK3-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP10]], [[TMP11]]
2159 // CHECK3-NEXT: store i32 [[ADD]], ptr [[DOTOMP_IV]], align 4
2160 // CHECK3-NEXT: br label [[OMP_INNER_FOR_COND]]
2161 // CHECK3: omp.inner.for.end:
2162 // CHECK3-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
2163 // CHECK3: omp.loop.exit:
2164 // CHECK3-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB1]], i32 [[TMP2]])
2165 // CHECK3-NEXT: ret void
2168 // CHECK3-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l57.omp_outlined.omp_outlined
2169 // CHECK3-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], i32 noundef [[DOTPREVIOUS_LB_:%.*]], i32 noundef [[DOTPREVIOUS_UB_:%.*]], ptr noundef [[THIS:%.*]]) #[[ATTR1]] {
2170 // CHECK3-NEXT: entry:
2171 // CHECK3-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 4
2172 // CHECK3-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 4
2173 // CHECK3-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i32, align 4
2174 // CHECK3-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i32, align 4
2175 // CHECK3-NEXT: [[THIS_ADDR:%.*]] = alloca ptr, align 4
2176 // CHECK3-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
2177 // CHECK3-NEXT: [[TMP:%.*]] = alloca i32, align 4
2178 // CHECK3-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4
2179 // CHECK3-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4
2180 // CHECK3-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
2181 // CHECK3-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
2182 // CHECK3-NEXT: [[I:%.*]] = alloca i32, align 4
2183 // CHECK3-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 4
2184 // CHECK3-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 4
2185 // CHECK3-NEXT: store i32 [[DOTPREVIOUS_LB_]], ptr [[DOTPREVIOUS_LB__ADDR]], align 4
2186 // CHECK3-NEXT: store i32 [[DOTPREVIOUS_UB_]], ptr [[DOTPREVIOUS_UB__ADDR]], align 4
2187 // CHECK3-NEXT: store ptr [[THIS]], ptr [[THIS_ADDR]], align 4
2188 // CHECK3-NEXT: [[TMP0:%.*]] = load ptr, ptr [[THIS_ADDR]], align 4
2189 // CHECK3-NEXT: store i32 0, ptr [[DOTOMP_LB]], align 4
2190 // CHECK3-NEXT: store i32 122, ptr [[DOTOMP_UB]], align 4
2191 // CHECK3-NEXT: [[TMP1:%.*]] = load i32, ptr [[DOTPREVIOUS_LB__ADDR]], align 4
2192 // CHECK3-NEXT: [[TMP2:%.*]] = load i32, ptr [[DOTPREVIOUS_UB__ADDR]], align 4
2193 // CHECK3-NEXT: store i32 [[TMP1]], ptr [[DOTOMP_LB]], align 4
2194 // CHECK3-NEXT: store i32 [[TMP2]], ptr [[DOTOMP_UB]], align 4
2195 // CHECK3-NEXT: store i32 1, ptr [[DOTOMP_STRIDE]], align 4
2196 // CHECK3-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
2197 // CHECK3-NEXT: [[TMP3:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4
2198 // CHECK3-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
2199 // CHECK3-NEXT: [[TMP5:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
2200 // CHECK3-NEXT: [[TMP6:%.*]] = load i32, ptr [[TMP5]], align 4
2201 // CHECK3-NEXT: call void @__kmpc_dispatch_init_4(ptr @[[GLOB3]], i32 [[TMP6]], i32 1073741859, i32 [[TMP3]], i32 [[TMP4]], i32 1, i32 61)
2202 // CHECK3-NEXT: br label [[OMP_DISPATCH_COND:%.*]]
2203 // CHECK3: omp.dispatch.cond:
2204 // CHECK3-NEXT: [[TMP7:%.*]] = call i32 @__kmpc_dispatch_next_4(ptr @[[GLOB3]], i32 [[TMP6]], ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]])
2205 // CHECK3-NEXT: [[TOBOOL:%.*]] = icmp ne i32 [[TMP7]], 0
2206 // CHECK3-NEXT: br i1 [[TOBOOL]], label [[OMP_DISPATCH_BODY:%.*]], label [[OMP_DISPATCH_END:%.*]]
2207 // CHECK3: omp.dispatch.body:
2208 // CHECK3-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4
2209 // CHECK3-NEXT: store i32 [[TMP8]], ptr [[DOTOMP_IV]], align 4
2210 // CHECK3-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
2211 // CHECK3: omp.inner.for.cond:
2212 // CHECK3-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP14:![0-9]+]]
2213 // CHECK3-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP14]]
2214 // CHECK3-NEXT: [[CMP:%.*]] = icmp sle i32 [[TMP9]], [[TMP10]]
2215 // CHECK3-NEXT: br i1 [[CMP]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
2216 // CHECK3: omp.inner.for.body:
2217 // CHECK3-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP14]]
2218 // CHECK3-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP11]], 1
2219 // CHECK3-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
2220 // CHECK3-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP14]]
2221 // CHECK3-NEXT: [[A:%.*]] = getelementptr inbounds nuw [[STRUCT_SS:%.*]], ptr [[TMP0]], i32 0, i32 0
2222 // CHECK3-NEXT: [[TMP12:%.*]] = load i32, ptr [[I]], align 4, !llvm.access.group [[ACC_GRP14]]
2223 // CHECK3-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds [123 x i32], ptr [[A]], i32 0, i32 [[TMP12]]
2224 // CHECK3-NEXT: store i32 0, ptr [[ARRAYIDX]], align 4, !llvm.access.group [[ACC_GRP14]]
2225 // CHECK3-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
2226 // CHECK3: omp.body.continue:
2227 // CHECK3-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
2228 // CHECK3: omp.inner.for.inc:
2229 // CHECK3-NEXT: [[TMP13:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP14]]
2230 // CHECK3-NEXT: [[ADD1:%.*]] = add nsw i32 [[TMP13]], 1
2231 // CHECK3-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP14]]
2232 // CHECK3-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP15:![0-9]+]]
2233 // CHECK3: omp.inner.for.end:
2234 // CHECK3-NEXT: br label [[OMP_DISPATCH_INC:%.*]]
2235 // CHECK3: omp.dispatch.inc:
2236 // CHECK3-NEXT: br label [[OMP_DISPATCH_COND]]
2237 // CHECK3: omp.dispatch.end:
2238 // CHECK3-NEXT: call void @__kmpc_dispatch_deinit(ptr @[[GLOB3]], i32 [[TMP6]])
2239 // CHECK3-NEXT: ret void
2242 // CHECK5-LABEL: define {{[^@]+}}@_Z21teams_template_structv
2243 // CHECK5-SAME: () #[[ATTR0:[0-9]+]] {
2244 // CHECK5-NEXT: entry:
2245 // CHECK5-NEXT: [[V:%.*]] = alloca [[STRUCT_SS:%.*]], align 4
2246 // CHECK5-NEXT: [[CALL:%.*]] = call noundef signext i32 @_ZN2SSIiLi123ELx456EE3fooEv(ptr noundef nonnull align 4 dereferenceable(496) [[V]])
2247 // CHECK5-NEXT: ret i32 [[CALL]]
2250 // CHECK5-LABEL: define {{[^@]+}}@_ZN2SSIiLi123ELx456EE3fooEv
2251 // CHECK5-SAME: (ptr noundef nonnull align 4 dereferenceable(496) [[THIS:%.*]]) #[[ATTR0]] comdat {
2252 // CHECK5-NEXT: entry:
2253 // CHECK5-NEXT: [[THIS_ADDR:%.*]] = alloca ptr, align 8
2254 // CHECK5-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [1 x ptr], align 8
2255 // CHECK5-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [1 x ptr], align 8
2256 // CHECK5-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [1 x ptr], align 8
2257 // CHECK5-NEXT: [[TMP:%.*]] = alloca i32, align 4
2258 // CHECK5-NEXT: [[KERNEL_ARGS:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS:%.*]], align 8
2259 // CHECK5-NEXT: [[DOTOFFLOAD_BASEPTRS3:%.*]] = alloca [1 x ptr], align 8
2260 // CHECK5-NEXT: [[DOTOFFLOAD_PTRS4:%.*]] = alloca [1 x ptr], align 8
2261 // CHECK5-NEXT: [[DOTOFFLOAD_MAPPERS5:%.*]] = alloca [1 x ptr], align 8
2262 // CHECK5-NEXT: [[_TMP6:%.*]] = alloca i32, align 4
2263 // CHECK5-NEXT: [[KERNEL_ARGS7:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS]], align 8
2264 // CHECK5-NEXT: [[DOTOFFLOAD_BASEPTRS11:%.*]] = alloca [1 x ptr], align 8
2265 // CHECK5-NEXT: [[DOTOFFLOAD_PTRS12:%.*]] = alloca [1 x ptr], align 8
2266 // CHECK5-NEXT: [[DOTOFFLOAD_MAPPERS13:%.*]] = alloca [1 x ptr], align 8
2267 // CHECK5-NEXT: [[_TMP14:%.*]] = alloca i32, align 4
2268 // CHECK5-NEXT: [[KERNEL_ARGS15:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS]], align 8
2269 // CHECK5-NEXT: [[DOTOFFLOAD_BASEPTRS19:%.*]] = alloca [1 x ptr], align 8
2270 // CHECK5-NEXT: [[DOTOFFLOAD_PTRS20:%.*]] = alloca [1 x ptr], align 8
2271 // CHECK5-NEXT: [[DOTOFFLOAD_MAPPERS21:%.*]] = alloca [1 x ptr], align 8
2272 // CHECK5-NEXT: [[_TMP22:%.*]] = alloca i32, align 4
2273 // CHECK5-NEXT: [[KERNEL_ARGS23:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS]], align 8
2274 // CHECK5-NEXT: [[DOTOFFLOAD_BASEPTRS27:%.*]] = alloca [1 x ptr], align 8
2275 // CHECK5-NEXT: [[DOTOFFLOAD_PTRS28:%.*]] = alloca [1 x ptr], align 8
2276 // CHECK5-NEXT: [[DOTOFFLOAD_MAPPERS29:%.*]] = alloca [1 x ptr], align 8
2277 // CHECK5-NEXT: [[_TMP30:%.*]] = alloca i32, align 4
2278 // CHECK5-NEXT: [[KERNEL_ARGS31:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS]], align 8
2279 // CHECK5-NEXT: store ptr [[THIS]], ptr [[THIS_ADDR]], align 8
2280 // CHECK5-NEXT: [[THIS1:%.*]] = load ptr, ptr [[THIS_ADDR]], align 8
2281 // CHECK5-NEXT: [[A:%.*]] = getelementptr inbounds nuw [[STRUCT_SS:%.*]], ptr [[THIS1]], i32 0, i32 0
2282 // CHECK5-NEXT: [[TMP0:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
2283 // CHECK5-NEXT: store ptr [[THIS1]], ptr [[TMP0]], align 8
2284 // CHECK5-NEXT: [[TMP1:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
2285 // CHECK5-NEXT: store ptr [[A]], ptr [[TMP1]], align 8
2286 // CHECK5-NEXT: [[TMP2:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0
2287 // CHECK5-NEXT: store ptr null, ptr [[TMP2]], align 8
2288 // CHECK5-NEXT: [[TMP3:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
2289 // CHECK5-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
2290 // CHECK5-NEXT: [[TMP5:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 0
2291 // CHECK5-NEXT: store i32 3, ptr [[TMP5]], align 4
2292 // CHECK5-NEXT: [[TMP6:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 1
2293 // CHECK5-NEXT: store i32 1, ptr [[TMP6]], align 4
2294 // CHECK5-NEXT: [[TMP7:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 2
2295 // CHECK5-NEXT: store ptr [[TMP3]], ptr [[TMP7]], align 8
2296 // CHECK5-NEXT: [[TMP8:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 3
2297 // CHECK5-NEXT: store ptr [[TMP4]], ptr [[TMP8]], align 8
2298 // CHECK5-NEXT: [[TMP9:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 4
2299 // CHECK5-NEXT: store ptr @.offload_sizes, ptr [[TMP9]], align 8
2300 // CHECK5-NEXT: [[TMP10:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 5
2301 // CHECK5-NEXT: store ptr @.offload_maptypes, ptr [[TMP10]], align 8
2302 // CHECK5-NEXT: [[TMP11:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 6
2303 // CHECK5-NEXT: store ptr null, ptr [[TMP11]], align 8
2304 // CHECK5-NEXT: [[TMP12:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 7
2305 // CHECK5-NEXT: store ptr null, ptr [[TMP12]], align 8
2306 // CHECK5-NEXT: [[TMP13:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 8
2307 // CHECK5-NEXT: store i64 123, ptr [[TMP13]], align 8
2308 // CHECK5-NEXT: [[TMP14:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 9
2309 // CHECK5-NEXT: store i64 0, ptr [[TMP14]], align 8
2310 // CHECK5-NEXT: [[TMP15:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 10
2311 // CHECK5-NEXT: store [3 x i32] zeroinitializer, ptr [[TMP15]], align 4
2312 // CHECK5-NEXT: [[TMP16:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 11
2313 // CHECK5-NEXT: store [3 x i32] zeroinitializer, ptr [[TMP16]], align 4
2314 // CHECK5-NEXT: [[TMP17:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 12
2315 // CHECK5-NEXT: store i32 0, ptr [[TMP17]], align 4
2316 // CHECK5-NEXT: [[TMP18:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB3:[0-9]+]], i64 -1, i32 0, i32 0, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l35.region_id, ptr [[KERNEL_ARGS]])
2317 // CHECK5-NEXT: [[TMP19:%.*]] = icmp ne i32 [[TMP18]], 0
2318 // CHECK5-NEXT: br i1 [[TMP19]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
2319 // CHECK5: omp_offload.failed:
2320 // CHECK5-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l35(ptr [[THIS1]]) #[[ATTR2:[0-9]+]]
2321 // CHECK5-NEXT: br label [[OMP_OFFLOAD_CONT]]
2322 // CHECK5: omp_offload.cont:
2323 // CHECK5-NEXT: [[A2:%.*]] = getelementptr inbounds nuw [[STRUCT_SS]], ptr [[THIS1]], i32 0, i32 0
2324 // CHECK5-NEXT: [[TMP20:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS3]], i32 0, i32 0
2325 // CHECK5-NEXT: store ptr [[THIS1]], ptr [[TMP20]], align 8
2326 // CHECK5-NEXT: [[TMP21:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS4]], i32 0, i32 0
2327 // CHECK5-NEXT: store ptr [[A2]], ptr [[TMP21]], align 8
2328 // CHECK5-NEXT: [[TMP22:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_MAPPERS5]], i64 0, i64 0
2329 // CHECK5-NEXT: store ptr null, ptr [[TMP22]], align 8
2330 // CHECK5-NEXT: [[TMP23:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS3]], i32 0, i32 0
2331 // CHECK5-NEXT: [[TMP24:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS4]], i32 0, i32 0
2332 // CHECK5-NEXT: [[TMP25:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS7]], i32 0, i32 0
2333 // CHECK5-NEXT: store i32 3, ptr [[TMP25]], align 4
2334 // CHECK5-NEXT: [[TMP26:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS7]], i32 0, i32 1
2335 // CHECK5-NEXT: store i32 1, ptr [[TMP26]], align 4
2336 // CHECK5-NEXT: [[TMP27:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS7]], i32 0, i32 2
2337 // CHECK5-NEXT: store ptr [[TMP23]], ptr [[TMP27]], align 8
2338 // CHECK5-NEXT: [[TMP28:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS7]], i32 0, i32 3
2339 // CHECK5-NEXT: store ptr [[TMP24]], ptr [[TMP28]], align 8
2340 // CHECK5-NEXT: [[TMP29:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS7]], i32 0, i32 4
2341 // CHECK5-NEXT: store ptr @.offload_sizes.1, ptr [[TMP29]], align 8
2342 // CHECK5-NEXT: [[TMP30:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS7]], i32 0, i32 5
2343 // CHECK5-NEXT: store ptr @.offload_maptypes.2, ptr [[TMP30]], align 8
2344 // CHECK5-NEXT: [[TMP31:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS7]], i32 0, i32 6
2345 // CHECK5-NEXT: store ptr null, ptr [[TMP31]], align 8
2346 // CHECK5-NEXT: [[TMP32:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS7]], i32 0, i32 7
2347 // CHECK5-NEXT: store ptr null, ptr [[TMP32]], align 8
2348 // CHECK5-NEXT: [[TMP33:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS7]], i32 0, i32 8
2349 // CHECK5-NEXT: store i64 123, ptr [[TMP33]], align 8
2350 // CHECK5-NEXT: [[TMP34:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS7]], i32 0, i32 9
2351 // CHECK5-NEXT: store i64 0, ptr [[TMP34]], align 8
2352 // CHECK5-NEXT: [[TMP35:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS7]], i32 0, i32 10
2353 // CHECK5-NEXT: store [3 x i32] zeroinitializer, ptr [[TMP35]], align 4
2354 // CHECK5-NEXT: [[TMP36:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS7]], i32 0, i32 11
2355 // CHECK5-NEXT: store [3 x i32] zeroinitializer, ptr [[TMP36]], align 4
2356 // CHECK5-NEXT: [[TMP37:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS7]], i32 0, i32 12
2357 // CHECK5-NEXT: store i32 0, ptr [[TMP37]], align 4
2358 // CHECK5-NEXT: [[TMP38:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB3]], i64 -1, i32 0, i32 0, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l40.region_id, ptr [[KERNEL_ARGS7]])
2359 // CHECK5-NEXT: [[TMP39:%.*]] = icmp ne i32 [[TMP38]], 0
2360 // CHECK5-NEXT: br i1 [[TMP39]], label [[OMP_OFFLOAD_FAILED8:%.*]], label [[OMP_OFFLOAD_CONT9:%.*]]
2361 // CHECK5: omp_offload.failed8:
2362 // CHECK5-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l40(ptr [[THIS1]]) #[[ATTR2]]
2363 // CHECK5-NEXT: br label [[OMP_OFFLOAD_CONT9]]
2364 // CHECK5: omp_offload.cont9:
2365 // CHECK5-NEXT: [[A10:%.*]] = getelementptr inbounds nuw [[STRUCT_SS]], ptr [[THIS1]], i32 0, i32 0
2366 // CHECK5-NEXT: [[TMP40:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS11]], i32 0, i32 0
2367 // CHECK5-NEXT: store ptr [[THIS1]], ptr [[TMP40]], align 8
2368 // CHECK5-NEXT: [[TMP41:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS12]], i32 0, i32 0
2369 // CHECK5-NEXT: store ptr [[A10]], ptr [[TMP41]], align 8
2370 // CHECK5-NEXT: [[TMP42:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_MAPPERS13]], i64 0, i64 0
2371 // CHECK5-NEXT: store ptr null, ptr [[TMP42]], align 8
2372 // CHECK5-NEXT: [[TMP43:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS11]], i32 0, i32 0
2373 // CHECK5-NEXT: [[TMP44:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS12]], i32 0, i32 0
2374 // CHECK5-NEXT: [[TMP45:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS15]], i32 0, i32 0
2375 // CHECK5-NEXT: store i32 3, ptr [[TMP45]], align 4
2376 // CHECK5-NEXT: [[TMP46:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS15]], i32 0, i32 1
2377 // CHECK5-NEXT: store i32 1, ptr [[TMP46]], align 4
2378 // CHECK5-NEXT: [[TMP47:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS15]], i32 0, i32 2
2379 // CHECK5-NEXT: store ptr [[TMP43]], ptr [[TMP47]], align 8
2380 // CHECK5-NEXT: [[TMP48:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS15]], i32 0, i32 3
2381 // CHECK5-NEXT: store ptr [[TMP44]], ptr [[TMP48]], align 8
2382 // CHECK5-NEXT: [[TMP49:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS15]], i32 0, i32 4
2383 // CHECK5-NEXT: store ptr @.offload_sizes.3, ptr [[TMP49]], align 8
2384 // CHECK5-NEXT: [[TMP50:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS15]], i32 0, i32 5
2385 // CHECK5-NEXT: store ptr @.offload_maptypes.4, ptr [[TMP50]], align 8
2386 // CHECK5-NEXT: [[TMP51:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS15]], i32 0, i32 6
2387 // CHECK5-NEXT: store ptr null, ptr [[TMP51]], align 8
2388 // CHECK5-NEXT: [[TMP52:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS15]], i32 0, i32 7
2389 // CHECK5-NEXT: store ptr null, ptr [[TMP52]], align 8
2390 // CHECK5-NEXT: [[TMP53:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS15]], i32 0, i32 8
2391 // CHECK5-NEXT: store i64 123, ptr [[TMP53]], align 8
2392 // CHECK5-NEXT: [[TMP54:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS15]], i32 0, i32 9
2393 // CHECK5-NEXT: store i64 0, ptr [[TMP54]], align 8
2394 // CHECK5-NEXT: [[TMP55:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS15]], i32 0, i32 10
2395 // CHECK5-NEXT: store [3 x i32] zeroinitializer, ptr [[TMP55]], align 4
2396 // CHECK5-NEXT: [[TMP56:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS15]], i32 0, i32 11
2397 // CHECK5-NEXT: store [3 x i32] zeroinitializer, ptr [[TMP56]], align 4
2398 // CHECK5-NEXT: [[TMP57:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS15]], i32 0, i32 12
2399 // CHECK5-NEXT: store i32 0, ptr [[TMP57]], align 4
2400 // CHECK5-NEXT: [[TMP58:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB3]], i64 -1, i32 0, i32 0, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l45.region_id, ptr [[KERNEL_ARGS15]])
2401 // CHECK5-NEXT: [[TMP59:%.*]] = icmp ne i32 [[TMP58]], 0
2402 // CHECK5-NEXT: br i1 [[TMP59]], label [[OMP_OFFLOAD_FAILED16:%.*]], label [[OMP_OFFLOAD_CONT17:%.*]]
2403 // CHECK5: omp_offload.failed16:
2404 // CHECK5-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l45(ptr [[THIS1]]) #[[ATTR2]]
2405 // CHECK5-NEXT: br label [[OMP_OFFLOAD_CONT17]]
2406 // CHECK5: omp_offload.cont17:
2407 // CHECK5-NEXT: [[A18:%.*]] = getelementptr inbounds nuw [[STRUCT_SS]], ptr [[THIS1]], i32 0, i32 0
2408 // CHECK5-NEXT: [[TMP60:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS19]], i32 0, i32 0
2409 // CHECK5-NEXT: store ptr [[THIS1]], ptr [[TMP60]], align 8
2410 // CHECK5-NEXT: [[TMP61:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS20]], i32 0, i32 0
2411 // CHECK5-NEXT: store ptr [[A18]], ptr [[TMP61]], align 8
2412 // CHECK5-NEXT: [[TMP62:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_MAPPERS21]], i64 0, i64 0
2413 // CHECK5-NEXT: store ptr null, ptr [[TMP62]], align 8
2414 // CHECK5-NEXT: [[TMP63:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS19]], i32 0, i32 0
2415 // CHECK5-NEXT: [[TMP64:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS20]], i32 0, i32 0
2416 // CHECK5-NEXT: [[TMP65:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS23]], i32 0, i32 0
2417 // CHECK5-NEXT: store i32 3, ptr [[TMP65]], align 4
2418 // CHECK5-NEXT: [[TMP66:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS23]], i32 0, i32 1
2419 // CHECK5-NEXT: store i32 1, ptr [[TMP66]], align 4
2420 // CHECK5-NEXT: [[TMP67:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS23]], i32 0, i32 2
2421 // CHECK5-NEXT: store ptr [[TMP63]], ptr [[TMP67]], align 8
2422 // CHECK5-NEXT: [[TMP68:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS23]], i32 0, i32 3
2423 // CHECK5-NEXT: store ptr [[TMP64]], ptr [[TMP68]], align 8
2424 // CHECK5-NEXT: [[TMP69:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS23]], i32 0, i32 4
2425 // CHECK5-NEXT: store ptr @.offload_sizes.5, ptr [[TMP69]], align 8
2426 // CHECK5-NEXT: [[TMP70:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS23]], i32 0, i32 5
2427 // CHECK5-NEXT: store ptr @.offload_maptypes.6, ptr [[TMP70]], align 8
2428 // CHECK5-NEXT: [[TMP71:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS23]], i32 0, i32 6
2429 // CHECK5-NEXT: store ptr null, ptr [[TMP71]], align 8
2430 // CHECK5-NEXT: [[TMP72:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS23]], i32 0, i32 7
2431 // CHECK5-NEXT: store ptr null, ptr [[TMP72]], align 8
2432 // CHECK5-NEXT: [[TMP73:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS23]], i32 0, i32 8
2433 // CHECK5-NEXT: store i64 123, ptr [[TMP73]], align 8
2434 // CHECK5-NEXT: [[TMP74:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS23]], i32 0, i32 9
2435 // CHECK5-NEXT: store i64 0, ptr [[TMP74]], align 8
2436 // CHECK5-NEXT: [[TMP75:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS23]], i32 0, i32 10
2437 // CHECK5-NEXT: store [3 x i32] zeroinitializer, ptr [[TMP75]], align 4
2438 // CHECK5-NEXT: [[TMP76:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS23]], i32 0, i32 11
2439 // CHECK5-NEXT: store [3 x i32] zeroinitializer, ptr [[TMP76]], align 4
2440 // CHECK5-NEXT: [[TMP77:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS23]], i32 0, i32 12
2441 // CHECK5-NEXT: store i32 0, ptr [[TMP77]], align 4
2442 // CHECK5-NEXT: [[TMP78:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB3]], i64 -1, i32 0, i32 0, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l51.region_id, ptr [[KERNEL_ARGS23]])
2443 // CHECK5-NEXT: [[TMP79:%.*]] = icmp ne i32 [[TMP78]], 0
2444 // CHECK5-NEXT: br i1 [[TMP79]], label [[OMP_OFFLOAD_FAILED24:%.*]], label [[OMP_OFFLOAD_CONT25:%.*]]
2445 // CHECK5: omp_offload.failed24:
2446 // CHECK5-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l51(ptr [[THIS1]]) #[[ATTR2]]
2447 // CHECK5-NEXT: br label [[OMP_OFFLOAD_CONT25]]
2448 // CHECK5: omp_offload.cont25:
2449 // CHECK5-NEXT: [[A26:%.*]] = getelementptr inbounds nuw [[STRUCT_SS]], ptr [[THIS1]], i32 0, i32 0
2450 // CHECK5-NEXT: [[TMP80:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS27]], i32 0, i32 0
2451 // CHECK5-NEXT: store ptr [[THIS1]], ptr [[TMP80]], align 8
2452 // CHECK5-NEXT: [[TMP81:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS28]], i32 0, i32 0
2453 // CHECK5-NEXT: store ptr [[A26]], ptr [[TMP81]], align 8
2454 // CHECK5-NEXT: [[TMP82:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_MAPPERS29]], i64 0, i64 0
2455 // CHECK5-NEXT: store ptr null, ptr [[TMP82]], align 8
2456 // CHECK5-NEXT: [[TMP83:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS27]], i32 0, i32 0
2457 // CHECK5-NEXT: [[TMP84:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS28]], i32 0, i32 0
2458 // CHECK5-NEXT: [[TMP85:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS31]], i32 0, i32 0
2459 // CHECK5-NEXT: store i32 3, ptr [[TMP85]], align 4
2460 // CHECK5-NEXT: [[TMP86:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS31]], i32 0, i32 1
2461 // CHECK5-NEXT: store i32 1, ptr [[TMP86]], align 4
2462 // CHECK5-NEXT: [[TMP87:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS31]], i32 0, i32 2
2463 // CHECK5-NEXT: store ptr [[TMP83]], ptr [[TMP87]], align 8
2464 // CHECK5-NEXT: [[TMP88:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS31]], i32 0, i32 3
2465 // CHECK5-NEXT: store ptr [[TMP84]], ptr [[TMP88]], align 8
2466 // CHECK5-NEXT: [[TMP89:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS31]], i32 0, i32 4
2467 // CHECK5-NEXT: store ptr @.offload_sizes.7, ptr [[TMP89]], align 8
2468 // CHECK5-NEXT: [[TMP90:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS31]], i32 0, i32 5
2469 // CHECK5-NEXT: store ptr @.offload_maptypes.8, ptr [[TMP90]], align 8
2470 // CHECK5-NEXT: [[TMP91:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS31]], i32 0, i32 6
2471 // CHECK5-NEXT: store ptr null, ptr [[TMP91]], align 8
2472 // CHECK5-NEXT: [[TMP92:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS31]], i32 0, i32 7
2473 // CHECK5-NEXT: store ptr null, ptr [[TMP92]], align 8
2474 // CHECK5-NEXT: [[TMP93:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS31]], i32 0, i32 8
2475 // CHECK5-NEXT: store i64 123, ptr [[TMP93]], align 8
2476 // CHECK5-NEXT: [[TMP94:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS31]], i32 0, i32 9
2477 // CHECK5-NEXT: store i64 0, ptr [[TMP94]], align 8
2478 // CHECK5-NEXT: [[TMP95:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS31]], i32 0, i32 10
2479 // CHECK5-NEXT: store [3 x i32] zeroinitializer, ptr [[TMP95]], align 4
2480 // CHECK5-NEXT: [[TMP96:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS31]], i32 0, i32 11
2481 // CHECK5-NEXT: store [3 x i32] zeroinitializer, ptr [[TMP96]], align 4
2482 // CHECK5-NEXT: [[TMP97:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS31]], i32 0, i32 12
2483 // CHECK5-NEXT: store i32 0, ptr [[TMP97]], align 4
2484 // CHECK5-NEXT: [[TMP98:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB3]], i64 -1, i32 0, i32 0, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l57.region_id, ptr [[KERNEL_ARGS31]])
2485 // CHECK5-NEXT: [[TMP99:%.*]] = icmp ne i32 [[TMP98]], 0
2486 // CHECK5-NEXT: br i1 [[TMP99]], label [[OMP_OFFLOAD_FAILED32:%.*]], label [[OMP_OFFLOAD_CONT33:%.*]]
2487 // CHECK5: omp_offload.failed32:
2488 // CHECK5-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l57(ptr [[THIS1]]) #[[ATTR2]]
2489 // CHECK5-NEXT: br label [[OMP_OFFLOAD_CONT33]]
2490 // CHECK5: omp_offload.cont33:
2491 // CHECK5-NEXT: [[A34:%.*]] = getelementptr inbounds nuw [[STRUCT_SS]], ptr [[THIS1]], i32 0, i32 0
2492 // CHECK5-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds [123 x i32], ptr [[A34]], i64 0, i64 0
2493 // CHECK5-NEXT: [[TMP100:%.*]] = load i32, ptr [[ARRAYIDX]], align 4
2494 // CHECK5-NEXT: ret i32 [[TMP100]]
2497 // CHECK5-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l35
2498 // CHECK5-SAME: (ptr noundef [[THIS:%.*]]) #[[ATTR1:[0-9]+]] {
2499 // CHECK5-NEXT: entry:
2500 // CHECK5-NEXT: [[THIS_ADDR:%.*]] = alloca ptr, align 8
2501 // CHECK5-NEXT: store ptr [[THIS]], ptr [[THIS_ADDR]], align 8
2502 // CHECK5-NEXT: [[TMP0:%.*]] = load ptr, ptr [[THIS_ADDR]], align 8
2503 // CHECK5-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_teams(ptr @[[GLOB3]], i32 1, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l35.omp_outlined, ptr [[TMP0]])
2504 // CHECK5-NEXT: ret void
2507 // CHECK5-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l35.omp_outlined
2508 // CHECK5-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], ptr noundef [[THIS:%.*]]) #[[ATTR1]] {
2509 // CHECK5-NEXT: entry:
2510 // CHECK5-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 8
2511 // CHECK5-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 8
2512 // CHECK5-NEXT: [[THIS_ADDR:%.*]] = alloca ptr, align 8
2513 // CHECK5-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
2514 // CHECK5-NEXT: [[TMP:%.*]] = alloca i32, align 4
2515 // CHECK5-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4
2516 // CHECK5-NEXT: [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4
2517 // CHECK5-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
2518 // CHECK5-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
2519 // CHECK5-NEXT: [[I:%.*]] = alloca i32, align 4
2520 // CHECK5-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 8
2521 // CHECK5-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 8
2522 // CHECK5-NEXT: store ptr [[THIS]], ptr [[THIS_ADDR]], align 8
2523 // CHECK5-NEXT: [[TMP0:%.*]] = load ptr, ptr [[THIS_ADDR]], align 8
2524 // CHECK5-NEXT: store i32 0, ptr [[DOTOMP_COMB_LB]], align 4
2525 // CHECK5-NEXT: store i32 122, ptr [[DOTOMP_COMB_UB]], align 4
2526 // CHECK5-NEXT: store i32 1, ptr [[DOTOMP_STRIDE]], align 4
2527 // CHECK5-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
2528 // CHECK5-NEXT: [[TMP1:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
2529 // CHECK5-NEXT: [[TMP2:%.*]] = load i32, ptr [[TMP1]], align 4
2530 // CHECK5-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB1:[0-9]+]], i32 [[TMP2]], i32 92, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
2531 // CHECK5-NEXT: [[TMP3:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
2532 // CHECK5-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP3]], 122
2533 // CHECK5-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
2534 // CHECK5: cond.true:
2535 // CHECK5-NEXT: br label [[COND_END:%.*]]
2536 // CHECK5: cond.false:
2537 // CHECK5-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
2538 // CHECK5-NEXT: br label [[COND_END]]
2539 // CHECK5: cond.end:
2540 // CHECK5-NEXT: [[COND:%.*]] = phi i32 [ 122, [[COND_TRUE]] ], [ [[TMP4]], [[COND_FALSE]] ]
2541 // CHECK5-NEXT: store i32 [[COND]], ptr [[DOTOMP_COMB_UB]], align 4
2542 // CHECK5-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
2543 // CHECK5-NEXT: store i32 [[TMP5]], ptr [[DOTOMP_IV]], align 4
2544 // CHECK5-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
2545 // CHECK5: omp.inner.for.cond:
2546 // CHECK5-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
2547 // CHECK5-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
2548 // CHECK5-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP6]], [[TMP7]]
2549 // CHECK5-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
2550 // CHECK5: omp.inner.for.body:
2551 // CHECK5-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
2552 // CHECK5-NEXT: [[TMP9:%.*]] = zext i32 [[TMP8]] to i64
2553 // CHECK5-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
2554 // CHECK5-NEXT: [[TMP11:%.*]] = zext i32 [[TMP10]] to i64
2555 // CHECK5-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_call(ptr @[[GLOB3]], i32 3, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l35.omp_outlined.omp_outlined, i64 [[TMP9]], i64 [[TMP11]], ptr [[TMP0]])
2556 // CHECK5-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
2557 // CHECK5: omp.inner.for.inc:
2558 // CHECK5-NEXT: [[TMP12:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
2559 // CHECK5-NEXT: [[TMP13:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4
2560 // CHECK5-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP12]], [[TMP13]]
2561 // CHECK5-NEXT: store i32 [[ADD]], ptr [[DOTOMP_IV]], align 4
2562 // CHECK5-NEXT: br label [[OMP_INNER_FOR_COND]]
2563 // CHECK5: omp.inner.for.end:
2564 // CHECK5-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
2565 // CHECK5: omp.loop.exit:
2566 // CHECK5-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB1]], i32 [[TMP2]])
2567 // CHECK5-NEXT: ret void
2570 // CHECK5-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l35.omp_outlined.omp_outlined
2571 // CHECK5-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], i64 noundef [[DOTPREVIOUS_LB_:%.*]], i64 noundef [[DOTPREVIOUS_UB_:%.*]], ptr noundef [[THIS:%.*]]) #[[ATTR1]] {
2572 // CHECK5-NEXT: entry:
2573 // CHECK5-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 8
2574 // CHECK5-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 8
2575 // CHECK5-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i64, align 8
2576 // CHECK5-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i64, align 8
2577 // CHECK5-NEXT: [[THIS_ADDR:%.*]] = alloca ptr, align 8
2578 // CHECK5-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
2579 // CHECK5-NEXT: [[TMP:%.*]] = alloca i32, align 4
2580 // CHECK5-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4
2581 // CHECK5-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4
2582 // CHECK5-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
2583 // CHECK5-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
2584 // CHECK5-NEXT: [[I:%.*]] = alloca i32, align 4
2585 // CHECK5-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 8
2586 // CHECK5-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 8
2587 // CHECK5-NEXT: store i64 [[DOTPREVIOUS_LB_]], ptr [[DOTPREVIOUS_LB__ADDR]], align 8
2588 // CHECK5-NEXT: store i64 [[DOTPREVIOUS_UB_]], ptr [[DOTPREVIOUS_UB__ADDR]], align 8
2589 // CHECK5-NEXT: store ptr [[THIS]], ptr [[THIS_ADDR]], align 8
2590 // CHECK5-NEXT: [[TMP0:%.*]] = load ptr, ptr [[THIS_ADDR]], align 8
2591 // CHECK5-NEXT: store i32 0, ptr [[DOTOMP_LB]], align 4
2592 // CHECK5-NEXT: store i32 122, ptr [[DOTOMP_UB]], align 4
2593 // CHECK5-NEXT: [[TMP1:%.*]] = load i64, ptr [[DOTPREVIOUS_LB__ADDR]], align 8
2594 // CHECK5-NEXT: [[CONV:%.*]] = trunc i64 [[TMP1]] to i32
2595 // CHECK5-NEXT: [[TMP2:%.*]] = load i64, ptr [[DOTPREVIOUS_UB__ADDR]], align 8
2596 // CHECK5-NEXT: [[CONV1:%.*]] = trunc i64 [[TMP2]] to i32
2597 // CHECK5-NEXT: store i32 [[CONV]], ptr [[DOTOMP_LB]], align 4
2598 // CHECK5-NEXT: store i32 [[CONV1]], ptr [[DOTOMP_UB]], align 4
2599 // CHECK5-NEXT: store i32 1, ptr [[DOTOMP_STRIDE]], align 4
2600 // CHECK5-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
2601 // CHECK5-NEXT: [[TMP3:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
2602 // CHECK5-NEXT: [[TMP4:%.*]] = load i32, ptr [[TMP3]], align 4
2603 // CHECK5-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB2:[0-9]+]], i32 [[TMP4]], i32 34, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
2604 // CHECK5-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
2605 // CHECK5-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP5]], 122
2606 // CHECK5-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
2607 // CHECK5: cond.true:
2608 // CHECK5-NEXT: br label [[COND_END:%.*]]
2609 // CHECK5: cond.false:
2610 // CHECK5-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
2611 // CHECK5-NEXT: br label [[COND_END]]
2612 // CHECK5: cond.end:
2613 // CHECK5-NEXT: [[COND:%.*]] = phi i32 [ 122, [[COND_TRUE]] ], [ [[TMP6]], [[COND_FALSE]] ]
2614 // CHECK5-NEXT: store i32 [[COND]], ptr [[DOTOMP_UB]], align 4
2615 // CHECK5-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4
2616 // CHECK5-NEXT: store i32 [[TMP7]], ptr [[DOTOMP_IV]], align 4
2617 // CHECK5-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
2618 // CHECK5: omp.inner.for.cond:
2619 // CHECK5-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
2620 // CHECK5-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
2621 // CHECK5-NEXT: [[CMP2:%.*]] = icmp sle i32 [[TMP8]], [[TMP9]]
2622 // CHECK5-NEXT: br i1 [[CMP2]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
2623 // CHECK5: omp.inner.for.body:
2624 // CHECK5-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
2625 // CHECK5-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP10]], 1
2626 // CHECK5-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
2627 // CHECK5-NEXT: store i32 [[ADD]], ptr [[I]], align 4
2628 // CHECK5-NEXT: [[A:%.*]] = getelementptr inbounds nuw [[STRUCT_SS:%.*]], ptr [[TMP0]], i32 0, i32 0
2629 // CHECK5-NEXT: [[TMP11:%.*]] = load i32, ptr [[I]], align 4
2630 // CHECK5-NEXT: [[IDXPROM:%.*]] = sext i32 [[TMP11]] to i64
2631 // CHECK5-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds [123 x i32], ptr [[A]], i64 0, i64 [[IDXPROM]]
2632 // CHECK5-NEXT: store i32 0, ptr [[ARRAYIDX]], align 4
2633 // CHECK5-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
2634 // CHECK5: omp.body.continue:
2635 // CHECK5-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
2636 // CHECK5: omp.inner.for.inc:
2637 // CHECK5-NEXT: [[TMP12:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
2638 // CHECK5-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP12]], 1
2639 // CHECK5-NEXT: store i32 [[ADD3]], ptr [[DOTOMP_IV]], align 4
2640 // CHECK5-NEXT: br label [[OMP_INNER_FOR_COND]]
2641 // CHECK5: omp.inner.for.end:
2642 // CHECK5-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
2643 // CHECK5: omp.loop.exit:
2644 // CHECK5-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB2]], i32 [[TMP4]])
2645 // CHECK5-NEXT: ret void
2648 // CHECK5-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l40
2649 // CHECK5-SAME: (ptr noundef [[THIS:%.*]]) #[[ATTR1]] {
2650 // CHECK5-NEXT: entry:
2651 // CHECK5-NEXT: [[THIS_ADDR:%.*]] = alloca ptr, align 8
2652 // CHECK5-NEXT: store ptr [[THIS]], ptr [[THIS_ADDR]], align 8
2653 // CHECK5-NEXT: [[TMP0:%.*]] = load ptr, ptr [[THIS_ADDR]], align 8
2654 // CHECK5-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_teams(ptr @[[GLOB3]], i32 1, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l40.omp_outlined, ptr [[TMP0]])
2655 // CHECK5-NEXT: ret void
2658 // CHECK5-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l40.omp_outlined
2659 // CHECK5-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], ptr noundef [[THIS:%.*]]) #[[ATTR1]] {
2660 // CHECK5-NEXT: entry:
2661 // CHECK5-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 8
2662 // CHECK5-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 8
2663 // CHECK5-NEXT: [[THIS_ADDR:%.*]] = alloca ptr, align 8
2664 // CHECK5-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
2665 // CHECK5-NEXT: [[TMP:%.*]] = alloca i32, align 4
2666 // CHECK5-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4
2667 // CHECK5-NEXT: [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4
2668 // CHECK5-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
2669 // CHECK5-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
2670 // CHECK5-NEXT: [[I:%.*]] = alloca i32, align 4
2671 // CHECK5-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 8
2672 // CHECK5-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 8
2673 // CHECK5-NEXT: store ptr [[THIS]], ptr [[THIS_ADDR]], align 8
2674 // CHECK5-NEXT: [[TMP0:%.*]] = load ptr, ptr [[THIS_ADDR]], align 8
2675 // CHECK5-NEXT: store i32 0, ptr [[DOTOMP_COMB_LB]], align 4
2676 // CHECK5-NEXT: store i32 122, ptr [[DOTOMP_COMB_UB]], align 4
2677 // CHECK5-NEXT: store i32 1, ptr [[DOTOMP_STRIDE]], align 4
2678 // CHECK5-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
2679 // CHECK5-NEXT: [[TMP1:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
2680 // CHECK5-NEXT: [[TMP2:%.*]] = load i32, ptr [[TMP1]], align 4
2681 // CHECK5-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB1]], i32 [[TMP2]], i32 92, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
2682 // CHECK5-NEXT: [[TMP3:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
2683 // CHECK5-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP3]], 122
2684 // CHECK5-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
2685 // CHECK5: cond.true:
2686 // CHECK5-NEXT: br label [[COND_END:%.*]]
2687 // CHECK5: cond.false:
2688 // CHECK5-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
2689 // CHECK5-NEXT: br label [[COND_END]]
2690 // CHECK5: cond.end:
2691 // CHECK5-NEXT: [[COND:%.*]] = phi i32 [ 122, [[COND_TRUE]] ], [ [[TMP4]], [[COND_FALSE]] ]
2692 // CHECK5-NEXT: store i32 [[COND]], ptr [[DOTOMP_COMB_UB]], align 4
2693 // CHECK5-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
2694 // CHECK5-NEXT: store i32 [[TMP5]], ptr [[DOTOMP_IV]], align 4
2695 // CHECK5-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
2696 // CHECK5: omp.inner.for.cond:
2697 // CHECK5-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
2698 // CHECK5-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
2699 // CHECK5-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP6]], [[TMP7]]
2700 // CHECK5-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
2701 // CHECK5: omp.inner.for.body:
2702 // CHECK5-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
2703 // CHECK5-NEXT: [[TMP9:%.*]] = zext i32 [[TMP8]] to i64
2704 // CHECK5-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
2705 // CHECK5-NEXT: [[TMP11:%.*]] = zext i32 [[TMP10]] to i64
2706 // CHECK5-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_call(ptr @[[GLOB3]], i32 3, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l40.omp_outlined.omp_outlined, i64 [[TMP9]], i64 [[TMP11]], ptr [[TMP0]])
2707 // CHECK5-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
2708 // CHECK5: omp.inner.for.inc:
2709 // CHECK5-NEXT: [[TMP12:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
2710 // CHECK5-NEXT: [[TMP13:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4
2711 // CHECK5-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP12]], [[TMP13]]
2712 // CHECK5-NEXT: store i32 [[ADD]], ptr [[DOTOMP_IV]], align 4
2713 // CHECK5-NEXT: br label [[OMP_INNER_FOR_COND]]
2714 // CHECK5: omp.inner.for.end:
2715 // CHECK5-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
2716 // CHECK5: omp.loop.exit:
2717 // CHECK5-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB1]], i32 [[TMP2]])
2718 // CHECK5-NEXT: ret void
2721 // CHECK5-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l40.omp_outlined.omp_outlined
2722 // CHECK5-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], i64 noundef [[DOTPREVIOUS_LB_:%.*]], i64 noundef [[DOTPREVIOUS_UB_:%.*]], ptr noundef [[THIS:%.*]]) #[[ATTR1]] {
2723 // CHECK5-NEXT: entry:
2724 // CHECK5-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 8
2725 // CHECK5-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 8
2726 // CHECK5-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i64, align 8
2727 // CHECK5-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i64, align 8
2728 // CHECK5-NEXT: [[THIS_ADDR:%.*]] = alloca ptr, align 8
2729 // CHECK5-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
2730 // CHECK5-NEXT: [[TMP:%.*]] = alloca i32, align 4
2731 // CHECK5-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4
2732 // CHECK5-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4
2733 // CHECK5-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
2734 // CHECK5-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
2735 // CHECK5-NEXT: [[I:%.*]] = alloca i32, align 4
2736 // CHECK5-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 8
2737 // CHECK5-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 8
2738 // CHECK5-NEXT: store i64 [[DOTPREVIOUS_LB_]], ptr [[DOTPREVIOUS_LB__ADDR]], align 8
2739 // CHECK5-NEXT: store i64 [[DOTPREVIOUS_UB_]], ptr [[DOTPREVIOUS_UB__ADDR]], align 8
2740 // CHECK5-NEXT: store ptr [[THIS]], ptr [[THIS_ADDR]], align 8
2741 // CHECK5-NEXT: [[TMP0:%.*]] = load ptr, ptr [[THIS_ADDR]], align 8
2742 // CHECK5-NEXT: store i32 0, ptr [[DOTOMP_LB]], align 4
2743 // CHECK5-NEXT: store i32 122, ptr [[DOTOMP_UB]], align 4
2744 // CHECK5-NEXT: [[TMP1:%.*]] = load i64, ptr [[DOTPREVIOUS_LB__ADDR]], align 8
2745 // CHECK5-NEXT: [[CONV:%.*]] = trunc i64 [[TMP1]] to i32
2746 // CHECK5-NEXT: [[TMP2:%.*]] = load i64, ptr [[DOTPREVIOUS_UB__ADDR]], align 8
2747 // CHECK5-NEXT: [[CONV1:%.*]] = trunc i64 [[TMP2]] to i32
2748 // CHECK5-NEXT: store i32 [[CONV]], ptr [[DOTOMP_LB]], align 4
2749 // CHECK5-NEXT: store i32 [[CONV1]], ptr [[DOTOMP_UB]], align 4
2750 // CHECK5-NEXT: store i32 1, ptr [[DOTOMP_STRIDE]], align 4
2751 // CHECK5-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
2752 // CHECK5-NEXT: [[TMP3:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
2753 // CHECK5-NEXT: [[TMP4:%.*]] = load i32, ptr [[TMP3]], align 4
2754 // CHECK5-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB2]], i32 [[TMP4]], i32 34, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
2755 // CHECK5-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
2756 // CHECK5-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP5]], 122
2757 // CHECK5-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
2758 // CHECK5: cond.true:
2759 // CHECK5-NEXT: br label [[COND_END:%.*]]
2760 // CHECK5: cond.false:
2761 // CHECK5-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
2762 // CHECK5-NEXT: br label [[COND_END]]
2763 // CHECK5: cond.end:
2764 // CHECK5-NEXT: [[COND:%.*]] = phi i32 [ 122, [[COND_TRUE]] ], [ [[TMP6]], [[COND_FALSE]] ]
2765 // CHECK5-NEXT: store i32 [[COND]], ptr [[DOTOMP_UB]], align 4
2766 // CHECK5-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4
2767 // CHECK5-NEXT: store i32 [[TMP7]], ptr [[DOTOMP_IV]], align 4
2768 // CHECK5-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
2769 // CHECK5: omp.inner.for.cond:
2770 // CHECK5-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
2771 // CHECK5-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
2772 // CHECK5-NEXT: [[CMP2:%.*]] = icmp sle i32 [[TMP8]], [[TMP9]]
2773 // CHECK5-NEXT: br i1 [[CMP2]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
2774 // CHECK5: omp.inner.for.body:
2775 // CHECK5-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
2776 // CHECK5-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP10]], 1
2777 // CHECK5-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
2778 // CHECK5-NEXT: store i32 [[ADD]], ptr [[I]], align 4
2779 // CHECK5-NEXT: [[A:%.*]] = getelementptr inbounds nuw [[STRUCT_SS:%.*]], ptr [[TMP0]], i32 0, i32 0
2780 // CHECK5-NEXT: [[TMP11:%.*]] = load i32, ptr [[I]], align 4
2781 // CHECK5-NEXT: [[IDXPROM:%.*]] = sext i32 [[TMP11]] to i64
2782 // CHECK5-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds [123 x i32], ptr [[A]], i64 0, i64 [[IDXPROM]]
2783 // CHECK5-NEXT: store i32 0, ptr [[ARRAYIDX]], align 4
2784 // CHECK5-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
2785 // CHECK5: omp.body.continue:
2786 // CHECK5-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
2787 // CHECK5: omp.inner.for.inc:
2788 // CHECK5-NEXT: [[TMP12:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
2789 // CHECK5-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP12]], 1
2790 // CHECK5-NEXT: store i32 [[ADD3]], ptr [[DOTOMP_IV]], align 4
2791 // CHECK5-NEXT: br label [[OMP_INNER_FOR_COND]]
2792 // CHECK5: omp.inner.for.end:
2793 // CHECK5-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
2794 // CHECK5: omp.loop.exit:
2795 // CHECK5-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB2]], i32 [[TMP4]])
2796 // CHECK5-NEXT: ret void
2799 // CHECK5-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l45
2800 // CHECK5-SAME: (ptr noundef [[THIS:%.*]]) #[[ATTR1]] {
2801 // CHECK5-NEXT: entry:
2802 // CHECK5-NEXT: [[THIS_ADDR:%.*]] = alloca ptr, align 8
2803 // CHECK5-NEXT: store ptr [[THIS]], ptr [[THIS_ADDR]], align 8
2804 // CHECK5-NEXT: [[TMP0:%.*]] = load ptr, ptr [[THIS_ADDR]], align 8
2805 // CHECK5-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_teams(ptr @[[GLOB3]], i32 1, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l45.omp_outlined, ptr [[TMP0]])
2806 // CHECK5-NEXT: ret void
2809 // CHECK5-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l45.omp_outlined
2810 // CHECK5-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], ptr noundef [[THIS:%.*]]) #[[ATTR1]] {
2811 // CHECK5-NEXT: entry:
2812 // CHECK5-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 8
2813 // CHECK5-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 8
2814 // CHECK5-NEXT: [[THIS_ADDR:%.*]] = alloca ptr, align 8
2815 // CHECK5-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
2816 // CHECK5-NEXT: [[TMP:%.*]] = alloca i32, align 4
2817 // CHECK5-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4
2818 // CHECK5-NEXT: [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4
2819 // CHECK5-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
2820 // CHECK5-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
2821 // CHECK5-NEXT: [[I:%.*]] = alloca i32, align 4
2822 // CHECK5-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 8
2823 // CHECK5-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 8
2824 // CHECK5-NEXT: store ptr [[THIS]], ptr [[THIS_ADDR]], align 8
2825 // CHECK5-NEXT: [[TMP0:%.*]] = load ptr, ptr [[THIS_ADDR]], align 8
2826 // CHECK5-NEXT: store i32 0, ptr [[DOTOMP_COMB_LB]], align 4
2827 // CHECK5-NEXT: store i32 122, ptr [[DOTOMP_COMB_UB]], align 4
2828 // CHECK5-NEXT: store i32 1, ptr [[DOTOMP_STRIDE]], align 4
2829 // CHECK5-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
2830 // CHECK5-NEXT: [[TMP1:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
2831 // CHECK5-NEXT: [[TMP2:%.*]] = load i32, ptr [[TMP1]], align 4
2832 // CHECK5-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB1]], i32 [[TMP2]], i32 92, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
2833 // CHECK5-NEXT: [[TMP3:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
2834 // CHECK5-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP3]], 122
2835 // CHECK5-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
2836 // CHECK5: cond.true:
2837 // CHECK5-NEXT: br label [[COND_END:%.*]]
2838 // CHECK5: cond.false:
2839 // CHECK5-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
2840 // CHECK5-NEXT: br label [[COND_END]]
2841 // CHECK5: cond.end:
2842 // CHECK5-NEXT: [[COND:%.*]] = phi i32 [ 122, [[COND_TRUE]] ], [ [[TMP4]], [[COND_FALSE]] ]
2843 // CHECK5-NEXT: store i32 [[COND]], ptr [[DOTOMP_COMB_UB]], align 4
2844 // CHECK5-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
2845 // CHECK5-NEXT: store i32 [[TMP5]], ptr [[DOTOMP_IV]], align 4
2846 // CHECK5-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
2847 // CHECK5: omp.inner.for.cond:
2848 // CHECK5-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
2849 // CHECK5-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
2850 // CHECK5-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP6]], [[TMP7]]
2851 // CHECK5-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
2852 // CHECK5: omp.inner.for.body:
2853 // CHECK5-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
2854 // CHECK5-NEXT: [[TMP9:%.*]] = zext i32 [[TMP8]] to i64
2855 // CHECK5-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
2856 // CHECK5-NEXT: [[TMP11:%.*]] = zext i32 [[TMP10]] to i64
2857 // CHECK5-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_call(ptr @[[GLOB3]], i32 3, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l45.omp_outlined.omp_outlined, i64 [[TMP9]], i64 [[TMP11]], ptr [[TMP0]])
2858 // CHECK5-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
2859 // CHECK5: omp.inner.for.inc:
2860 // CHECK5-NEXT: [[TMP12:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
2861 // CHECK5-NEXT: [[TMP13:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4
2862 // CHECK5-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP12]], [[TMP13]]
2863 // CHECK5-NEXT: store i32 [[ADD]], ptr [[DOTOMP_IV]], align 4
2864 // CHECK5-NEXT: br label [[OMP_INNER_FOR_COND]]
2865 // CHECK5: omp.inner.for.end:
2866 // CHECK5-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
2867 // CHECK5: omp.loop.exit:
2868 // CHECK5-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB1]], i32 [[TMP2]])
2869 // CHECK5-NEXT: ret void
2872 // CHECK5-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l45.omp_outlined.omp_outlined
2873 // CHECK5-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], i64 noundef [[DOTPREVIOUS_LB_:%.*]], i64 noundef [[DOTPREVIOUS_UB_:%.*]], ptr noundef [[THIS:%.*]]) #[[ATTR1]] {
2874 // CHECK5-NEXT: entry:
2875 // CHECK5-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 8
2876 // CHECK5-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 8
2877 // CHECK5-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i64, align 8
2878 // CHECK5-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i64, align 8
2879 // CHECK5-NEXT: [[THIS_ADDR:%.*]] = alloca ptr, align 8
2880 // CHECK5-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
2881 // CHECK5-NEXT: [[TMP:%.*]] = alloca i32, align 4
2882 // CHECK5-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4
2883 // CHECK5-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4
2884 // CHECK5-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
2885 // CHECK5-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
2886 // CHECK5-NEXT: [[I:%.*]] = alloca i32, align 4
2887 // CHECK5-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 8
2888 // CHECK5-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 8
2889 // CHECK5-NEXT: store i64 [[DOTPREVIOUS_LB_]], ptr [[DOTPREVIOUS_LB__ADDR]], align 8
2890 // CHECK5-NEXT: store i64 [[DOTPREVIOUS_UB_]], ptr [[DOTPREVIOUS_UB__ADDR]], align 8
2891 // CHECK5-NEXT: store ptr [[THIS]], ptr [[THIS_ADDR]], align 8
2892 // CHECK5-NEXT: [[TMP0:%.*]] = load ptr, ptr [[THIS_ADDR]], align 8
2893 // CHECK5-NEXT: store i32 0, ptr [[DOTOMP_LB]], align 4
2894 // CHECK5-NEXT: store i32 122, ptr [[DOTOMP_UB]], align 4
2895 // CHECK5-NEXT: [[TMP1:%.*]] = load i64, ptr [[DOTPREVIOUS_LB__ADDR]], align 8
2896 // CHECK5-NEXT: [[CONV:%.*]] = trunc i64 [[TMP1]] to i32
2897 // CHECK5-NEXT: [[TMP2:%.*]] = load i64, ptr [[DOTPREVIOUS_UB__ADDR]], align 8
2898 // CHECK5-NEXT: [[CONV1:%.*]] = trunc i64 [[TMP2]] to i32
2899 // CHECK5-NEXT: store i32 [[CONV]], ptr [[DOTOMP_LB]], align 4
2900 // CHECK5-NEXT: store i32 [[CONV1]], ptr [[DOTOMP_UB]], align 4
2901 // CHECK5-NEXT: store i32 1, ptr [[DOTOMP_STRIDE]], align 4
2902 // CHECK5-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
2903 // CHECK5-NEXT: [[TMP3:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
2904 // CHECK5-NEXT: [[TMP4:%.*]] = load i32, ptr [[TMP3]], align 4
2905 // CHECK5-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB2]], i32 [[TMP4]], i32 33, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 61)
2906 // CHECK5-NEXT: br label [[OMP_DISPATCH_COND:%.*]]
2907 // CHECK5: omp.dispatch.cond:
2908 // CHECK5-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
2909 // CHECK5-NEXT: [[TMP6:%.*]] = load i64, ptr [[DOTPREVIOUS_UB__ADDR]], align 8
2910 // CHECK5-NEXT: [[CONV2:%.*]] = trunc i64 [[TMP6]] to i32
2911 // CHECK5-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP5]], [[CONV2]]
2912 // CHECK5-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
2913 // CHECK5: cond.true:
2914 // CHECK5-NEXT: [[TMP7:%.*]] = load i64, ptr [[DOTPREVIOUS_UB__ADDR]], align 8
2915 // CHECK5-NEXT: [[CONV3:%.*]] = trunc i64 [[TMP7]] to i32
2916 // CHECK5-NEXT: br label [[COND_END:%.*]]
2917 // CHECK5: cond.false:
2918 // CHECK5-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
2919 // CHECK5-NEXT: br label [[COND_END]]
2920 // CHECK5: cond.end:
2921 // CHECK5-NEXT: [[COND:%.*]] = phi i32 [ [[CONV3]], [[COND_TRUE]] ], [ [[TMP8]], [[COND_FALSE]] ]
2922 // CHECK5-NEXT: store i32 [[COND]], ptr [[DOTOMP_UB]], align 4
2923 // CHECK5-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4
2924 // CHECK5-NEXT: store i32 [[TMP9]], ptr [[DOTOMP_IV]], align 4
2925 // CHECK5-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
2926 // CHECK5-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
2927 // CHECK5-NEXT: [[CMP4:%.*]] = icmp sle i32 [[TMP10]], [[TMP11]]
2928 // CHECK5-NEXT: br i1 [[CMP4]], label [[OMP_DISPATCH_BODY:%.*]], label [[OMP_DISPATCH_END:%.*]]
2929 // CHECK5: omp.dispatch.body:
2930 // CHECK5-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
2931 // CHECK5: omp.inner.for.cond:
2932 // CHECK5-NEXT: [[TMP12:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
2933 // CHECK5-NEXT: [[TMP13:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
2934 // CHECK5-NEXT: [[CMP5:%.*]] = icmp sle i32 [[TMP12]], [[TMP13]]
2935 // CHECK5-NEXT: br i1 [[CMP5]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
2936 // CHECK5: omp.inner.for.body:
2937 // CHECK5-NEXT: [[TMP14:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
2938 // CHECK5-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP14]], 1
2939 // CHECK5-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
2940 // CHECK5-NEXT: store i32 [[ADD]], ptr [[I]], align 4
2941 // CHECK5-NEXT: [[A:%.*]] = getelementptr inbounds nuw [[STRUCT_SS:%.*]], ptr [[TMP0]], i32 0, i32 0
2942 // CHECK5-NEXT: [[TMP15:%.*]] = load i32, ptr [[I]], align 4
2943 // CHECK5-NEXT: [[IDXPROM:%.*]] = sext i32 [[TMP15]] to i64
2944 // CHECK5-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds [123 x i32], ptr [[A]], i64 0, i64 [[IDXPROM]]
2945 // CHECK5-NEXT: store i32 0, ptr [[ARRAYIDX]], align 4
2946 // CHECK5-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
2947 // CHECK5: omp.body.continue:
2948 // CHECK5-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
2949 // CHECK5: omp.inner.for.inc:
2950 // CHECK5-NEXT: [[TMP16:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
2951 // CHECK5-NEXT: [[ADD6:%.*]] = add nsw i32 [[TMP16]], 1
2952 // CHECK5-NEXT: store i32 [[ADD6]], ptr [[DOTOMP_IV]], align 4
2953 // CHECK5-NEXT: br label [[OMP_INNER_FOR_COND]]
2954 // CHECK5: omp.inner.for.end:
2955 // CHECK5-NEXT: br label [[OMP_DISPATCH_INC:%.*]]
2956 // CHECK5: omp.dispatch.inc:
2957 // CHECK5-NEXT: [[TMP17:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4
2958 // CHECK5-NEXT: [[TMP18:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4
2959 // CHECK5-NEXT: [[ADD7:%.*]] = add nsw i32 [[TMP17]], [[TMP18]]
2960 // CHECK5-NEXT: store i32 [[ADD7]], ptr [[DOTOMP_LB]], align 4
2961 // CHECK5-NEXT: [[TMP19:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
2962 // CHECK5-NEXT: [[TMP20:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4
2963 // CHECK5-NEXT: [[ADD8:%.*]] = add nsw i32 [[TMP19]], [[TMP20]]
2964 // CHECK5-NEXT: store i32 [[ADD8]], ptr [[DOTOMP_UB]], align 4
2965 // CHECK5-NEXT: br label [[OMP_DISPATCH_COND]]
2966 // CHECK5: omp.dispatch.end:
2967 // CHECK5-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB2]], i32 [[TMP4]])
2968 // CHECK5-NEXT: ret void
2971 // CHECK5-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l51
2972 // CHECK5-SAME: (ptr noundef [[THIS:%.*]]) #[[ATTR1]] {
2973 // CHECK5-NEXT: entry:
2974 // CHECK5-NEXT: [[THIS_ADDR:%.*]] = alloca ptr, align 8
2975 // CHECK5-NEXT: store ptr [[THIS]], ptr [[THIS_ADDR]], align 8
2976 // CHECK5-NEXT: [[TMP0:%.*]] = load ptr, ptr [[THIS_ADDR]], align 8
2977 // CHECK5-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_teams(ptr @[[GLOB3]], i32 1, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l51.omp_outlined, ptr [[TMP0]])
2978 // CHECK5-NEXT: ret void
2981 // CHECK5-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l51.omp_outlined
2982 // CHECK5-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], ptr noundef [[THIS:%.*]]) #[[ATTR1]] {
2983 // CHECK5-NEXT: entry:
2984 // CHECK5-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 8
2985 // CHECK5-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 8
2986 // CHECK5-NEXT: [[THIS_ADDR:%.*]] = alloca ptr, align 8
2987 // CHECK5-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
2988 // CHECK5-NEXT: [[TMP:%.*]] = alloca i32, align 4
2989 // CHECK5-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4
2990 // CHECK5-NEXT: [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4
2991 // CHECK5-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
2992 // CHECK5-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
2993 // CHECK5-NEXT: [[I:%.*]] = alloca i32, align 4
2994 // CHECK5-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 8
2995 // CHECK5-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 8
2996 // CHECK5-NEXT: store ptr [[THIS]], ptr [[THIS_ADDR]], align 8
2997 // CHECK5-NEXT: [[TMP0:%.*]] = load ptr, ptr [[THIS_ADDR]], align 8
2998 // CHECK5-NEXT: store i32 0, ptr [[DOTOMP_COMB_LB]], align 4
2999 // CHECK5-NEXT: store i32 122, ptr [[DOTOMP_COMB_UB]], align 4
3000 // CHECK5-NEXT: store i32 1, ptr [[DOTOMP_STRIDE]], align 4
3001 // CHECK5-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
3002 // CHECK5-NEXT: [[TMP1:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
3003 // CHECK5-NEXT: [[TMP2:%.*]] = load i32, ptr [[TMP1]], align 4
3004 // CHECK5-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB1]], i32 [[TMP2]], i32 92, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
3005 // CHECK5-NEXT: [[TMP3:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
3006 // CHECK5-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP3]], 122
3007 // CHECK5-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
3008 // CHECK5: cond.true:
3009 // CHECK5-NEXT: br label [[COND_END:%.*]]
3010 // CHECK5: cond.false:
3011 // CHECK5-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
3012 // CHECK5-NEXT: br label [[COND_END]]
3013 // CHECK5: cond.end:
3014 // CHECK5-NEXT: [[COND:%.*]] = phi i32 [ 122, [[COND_TRUE]] ], [ [[TMP4]], [[COND_FALSE]] ]
3015 // CHECK5-NEXT: store i32 [[COND]], ptr [[DOTOMP_COMB_UB]], align 4
3016 // CHECK5-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
3017 // CHECK5-NEXT: store i32 [[TMP5]], ptr [[DOTOMP_IV]], align 4
3018 // CHECK5-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
3019 // CHECK5: omp.inner.for.cond:
3020 // CHECK5-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
3021 // CHECK5-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
3022 // CHECK5-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP6]], [[TMP7]]
3023 // CHECK5-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
3024 // CHECK5: omp.inner.for.body:
3025 // CHECK5-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
3026 // CHECK5-NEXT: [[TMP9:%.*]] = zext i32 [[TMP8]] to i64
3027 // CHECK5-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
3028 // CHECK5-NEXT: [[TMP11:%.*]] = zext i32 [[TMP10]] to i64
3029 // CHECK5-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_call(ptr @[[GLOB3]], i32 3, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l51.omp_outlined.omp_outlined, i64 [[TMP9]], i64 [[TMP11]], ptr [[TMP0]])
3030 // CHECK5-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
3031 // CHECK5: omp.inner.for.inc:
3032 // CHECK5-NEXT: [[TMP12:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
3033 // CHECK5-NEXT: [[TMP13:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4
3034 // CHECK5-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP12]], [[TMP13]]
3035 // CHECK5-NEXT: store i32 [[ADD]], ptr [[DOTOMP_IV]], align 4
3036 // CHECK5-NEXT: br label [[OMP_INNER_FOR_COND]]
3037 // CHECK5: omp.inner.for.end:
3038 // CHECK5-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
3039 // CHECK5: omp.loop.exit:
3040 // CHECK5-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB1]], i32 [[TMP2]])
3041 // CHECK5-NEXT: ret void
3044 // CHECK5-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l51.omp_outlined.omp_outlined
3045 // CHECK5-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], i64 noundef [[DOTPREVIOUS_LB_:%.*]], i64 noundef [[DOTPREVIOUS_UB_:%.*]], ptr noundef [[THIS:%.*]]) #[[ATTR1]] {
3046 // CHECK5-NEXT: entry:
3047 // CHECK5-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 8
3048 // CHECK5-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 8
3049 // CHECK5-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i64, align 8
3050 // CHECK5-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i64, align 8
3051 // CHECK5-NEXT: [[THIS_ADDR:%.*]] = alloca ptr, align 8
3052 // CHECK5-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
3053 // CHECK5-NEXT: [[TMP:%.*]] = alloca i32, align 4
3054 // CHECK5-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4
3055 // CHECK5-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4
3056 // CHECK5-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
3057 // CHECK5-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
3058 // CHECK5-NEXT: [[I:%.*]] = alloca i32, align 4
3059 // CHECK5-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 8
3060 // CHECK5-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 8
3061 // CHECK5-NEXT: store i64 [[DOTPREVIOUS_LB_]], ptr [[DOTPREVIOUS_LB__ADDR]], align 8
3062 // CHECK5-NEXT: store i64 [[DOTPREVIOUS_UB_]], ptr [[DOTPREVIOUS_UB__ADDR]], align 8
3063 // CHECK5-NEXT: store ptr [[THIS]], ptr [[THIS_ADDR]], align 8
3064 // CHECK5-NEXT: [[TMP0:%.*]] = load ptr, ptr [[THIS_ADDR]], align 8
3065 // CHECK5-NEXT: store i32 0, ptr [[DOTOMP_LB]], align 4
3066 // CHECK5-NEXT: store i32 122, ptr [[DOTOMP_UB]], align 4
3067 // CHECK5-NEXT: [[TMP1:%.*]] = load i64, ptr [[DOTPREVIOUS_LB__ADDR]], align 8
3068 // CHECK5-NEXT: [[CONV:%.*]] = trunc i64 [[TMP1]] to i32
3069 // CHECK5-NEXT: [[TMP2:%.*]] = load i64, ptr [[DOTPREVIOUS_UB__ADDR]], align 8
3070 // CHECK5-NEXT: [[CONV1:%.*]] = trunc i64 [[TMP2]] to i32
3071 // CHECK5-NEXT: store i32 [[CONV]], ptr [[DOTOMP_LB]], align 4
3072 // CHECK5-NEXT: store i32 [[CONV1]], ptr [[DOTOMP_UB]], align 4
3073 // CHECK5-NEXT: store i32 1, ptr [[DOTOMP_STRIDE]], align 4
3074 // CHECK5-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
3075 // CHECK5-NEXT: [[TMP3:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4
3076 // CHECK5-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
3077 // CHECK5-NEXT: [[TMP5:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
3078 // CHECK5-NEXT: [[TMP6:%.*]] = load i32, ptr [[TMP5]], align 4
3079 // CHECK5-NEXT: call void @__kmpc_dispatch_init_4(ptr @[[GLOB3]], i32 [[TMP6]], i32 35, i32 [[TMP3]], i32 [[TMP4]], i32 1, i32 1)
3080 // CHECK5-NEXT: br label [[OMP_DISPATCH_COND:%.*]]
3081 // CHECK5: omp.dispatch.cond:
3082 // CHECK5-NEXT: [[TMP7:%.*]] = call i32 @__kmpc_dispatch_next_4(ptr @[[GLOB3]], i32 [[TMP6]], ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]])
3083 // CHECK5-NEXT: [[TOBOOL:%.*]] = icmp ne i32 [[TMP7]], 0
3084 // CHECK5-NEXT: br i1 [[TOBOOL]], label [[OMP_DISPATCH_BODY:%.*]], label [[OMP_DISPATCH_END:%.*]]
3085 // CHECK5: omp.dispatch.body:
3086 // CHECK5-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4
3087 // CHECK5-NEXT: store i32 [[TMP8]], ptr [[DOTOMP_IV]], align 4
3088 // CHECK5-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
3089 // CHECK5: omp.inner.for.cond:
3090 // CHECK5-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP10:![0-9]+]]
3091 // CHECK5-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP10]]
3092 // CHECK5-NEXT: [[CMP:%.*]] = icmp sle i32 [[TMP9]], [[TMP10]]
3093 // CHECK5-NEXT: br i1 [[CMP]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
3094 // CHECK5: omp.inner.for.body:
3095 // CHECK5-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP10]]
3096 // CHECK5-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP11]], 1
3097 // CHECK5-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
3098 // CHECK5-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP10]]
3099 // CHECK5-NEXT: [[A:%.*]] = getelementptr inbounds nuw [[STRUCT_SS:%.*]], ptr [[TMP0]], i32 0, i32 0
3100 // CHECK5-NEXT: [[TMP12:%.*]] = load i32, ptr [[I]], align 4, !llvm.access.group [[ACC_GRP10]]
3101 // CHECK5-NEXT: [[IDXPROM:%.*]] = sext i32 [[TMP12]] to i64
3102 // CHECK5-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds [123 x i32], ptr [[A]], i64 0, i64 [[IDXPROM]]
3103 // CHECK5-NEXT: store i32 0, ptr [[ARRAYIDX]], align 4, !llvm.access.group [[ACC_GRP10]]
3104 // CHECK5-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
3105 // CHECK5: omp.body.continue:
3106 // CHECK5-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
3107 // CHECK5: omp.inner.for.inc:
3108 // CHECK5-NEXT: [[TMP13:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP10]]
3109 // CHECK5-NEXT: [[ADD2:%.*]] = add nsw i32 [[TMP13]], 1
3110 // CHECK5-NEXT: store i32 [[ADD2]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP10]]
3111 // CHECK5-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP11:![0-9]+]]
3112 // CHECK5: omp.inner.for.end:
3113 // CHECK5-NEXT: br label [[OMP_DISPATCH_INC:%.*]]
3114 // CHECK5: omp.dispatch.inc:
3115 // CHECK5-NEXT: br label [[OMP_DISPATCH_COND]]
3116 // CHECK5: omp.dispatch.end:
3117 // CHECK5-NEXT: call void @__kmpc_dispatch_deinit(ptr @[[GLOB3]], i32 [[TMP6]])
3118 // CHECK5-NEXT: ret void
3121 // CHECK5-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l57
3122 // CHECK5-SAME: (ptr noundef [[THIS:%.*]]) #[[ATTR1]] {
3123 // CHECK5-NEXT: entry:
3124 // CHECK5-NEXT: [[THIS_ADDR:%.*]] = alloca ptr, align 8
3125 // CHECK5-NEXT: store ptr [[THIS]], ptr [[THIS_ADDR]], align 8
3126 // CHECK5-NEXT: [[TMP0:%.*]] = load ptr, ptr [[THIS_ADDR]], align 8
3127 // CHECK5-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_teams(ptr @[[GLOB3]], i32 1, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l57.omp_outlined, ptr [[TMP0]])
3128 // CHECK5-NEXT: ret void
3131 // CHECK5-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l57.omp_outlined
3132 // CHECK5-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], ptr noundef [[THIS:%.*]]) #[[ATTR1]] {
3133 // CHECK5-NEXT: entry:
3134 // CHECK5-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 8
3135 // CHECK5-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 8
3136 // CHECK5-NEXT: [[THIS_ADDR:%.*]] = alloca ptr, align 8
3137 // CHECK5-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
3138 // CHECK5-NEXT: [[TMP:%.*]] = alloca i32, align 4
3139 // CHECK5-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4
3140 // CHECK5-NEXT: [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4
3141 // CHECK5-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
3142 // CHECK5-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
3143 // CHECK5-NEXT: [[I:%.*]] = alloca i32, align 4
3144 // CHECK5-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 8
3145 // CHECK5-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 8
3146 // CHECK5-NEXT: store ptr [[THIS]], ptr [[THIS_ADDR]], align 8
3147 // CHECK5-NEXT: [[TMP0:%.*]] = load ptr, ptr [[THIS_ADDR]], align 8
3148 // CHECK5-NEXT: store i32 0, ptr [[DOTOMP_COMB_LB]], align 4
3149 // CHECK5-NEXT: store i32 122, ptr [[DOTOMP_COMB_UB]], align 4
3150 // CHECK5-NEXT: store i32 1, ptr [[DOTOMP_STRIDE]], align 4
3151 // CHECK5-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
3152 // CHECK5-NEXT: [[TMP1:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
3153 // CHECK5-NEXT: [[TMP2:%.*]] = load i32, ptr [[TMP1]], align 4
3154 // CHECK5-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB1]], i32 [[TMP2]], i32 92, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
3155 // CHECK5-NEXT: [[TMP3:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
3156 // CHECK5-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP3]], 122
3157 // CHECK5-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
3158 // CHECK5: cond.true:
3159 // CHECK5-NEXT: br label [[COND_END:%.*]]
3160 // CHECK5: cond.false:
3161 // CHECK5-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
3162 // CHECK5-NEXT: br label [[COND_END]]
3163 // CHECK5: cond.end:
3164 // CHECK5-NEXT: [[COND:%.*]] = phi i32 [ 122, [[COND_TRUE]] ], [ [[TMP4]], [[COND_FALSE]] ]
3165 // CHECK5-NEXT: store i32 [[COND]], ptr [[DOTOMP_COMB_UB]], align 4
3166 // CHECK5-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
3167 // CHECK5-NEXT: store i32 [[TMP5]], ptr [[DOTOMP_IV]], align 4
3168 // CHECK5-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
3169 // CHECK5: omp.inner.for.cond:
3170 // CHECK5-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
3171 // CHECK5-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
3172 // CHECK5-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP6]], [[TMP7]]
3173 // CHECK5-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
3174 // CHECK5: omp.inner.for.body:
3175 // CHECK5-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
3176 // CHECK5-NEXT: [[TMP9:%.*]] = zext i32 [[TMP8]] to i64
3177 // CHECK5-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
3178 // CHECK5-NEXT: [[TMP11:%.*]] = zext i32 [[TMP10]] to i64
3179 // CHECK5-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_call(ptr @[[GLOB3]], i32 3, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l57.omp_outlined.omp_outlined, i64 [[TMP9]], i64 [[TMP11]], ptr [[TMP0]])
3180 // CHECK5-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
3181 // CHECK5: omp.inner.for.inc:
3182 // CHECK5-NEXT: [[TMP12:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
3183 // CHECK5-NEXT: [[TMP13:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4
3184 // CHECK5-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP12]], [[TMP13]]
3185 // CHECK5-NEXT: store i32 [[ADD]], ptr [[DOTOMP_IV]], align 4
3186 // CHECK5-NEXT: br label [[OMP_INNER_FOR_COND]]
3187 // CHECK5: omp.inner.for.end:
3188 // CHECK5-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
3189 // CHECK5: omp.loop.exit:
3190 // CHECK5-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB1]], i32 [[TMP2]])
3191 // CHECK5-NEXT: ret void
3194 // CHECK5-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l57.omp_outlined.omp_outlined
3195 // CHECK5-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], i64 noundef [[DOTPREVIOUS_LB_:%.*]], i64 noundef [[DOTPREVIOUS_UB_:%.*]], ptr noundef [[THIS:%.*]]) #[[ATTR1]] {
3196 // CHECK5-NEXT: entry:
3197 // CHECK5-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 8
3198 // CHECK5-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 8
3199 // CHECK5-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i64, align 8
3200 // CHECK5-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i64, align 8
3201 // CHECK5-NEXT: [[THIS_ADDR:%.*]] = alloca ptr, align 8
3202 // CHECK5-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
3203 // CHECK5-NEXT: [[TMP:%.*]] = alloca i32, align 4
3204 // CHECK5-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4
3205 // CHECK5-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4
3206 // CHECK5-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
3207 // CHECK5-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
3208 // CHECK5-NEXT: [[I:%.*]] = alloca i32, align 4
3209 // CHECK5-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 8
3210 // CHECK5-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 8
3211 // CHECK5-NEXT: store i64 [[DOTPREVIOUS_LB_]], ptr [[DOTPREVIOUS_LB__ADDR]], align 8
3212 // CHECK5-NEXT: store i64 [[DOTPREVIOUS_UB_]], ptr [[DOTPREVIOUS_UB__ADDR]], align 8
3213 // CHECK5-NEXT: store ptr [[THIS]], ptr [[THIS_ADDR]], align 8
3214 // CHECK5-NEXT: [[TMP0:%.*]] = load ptr, ptr [[THIS_ADDR]], align 8
3215 // CHECK5-NEXT: store i32 0, ptr [[DOTOMP_LB]], align 4
3216 // CHECK5-NEXT: store i32 122, ptr [[DOTOMP_UB]], align 4
3217 // CHECK5-NEXT: [[TMP1:%.*]] = load i64, ptr [[DOTPREVIOUS_LB__ADDR]], align 8
3218 // CHECK5-NEXT: [[CONV:%.*]] = trunc i64 [[TMP1]] to i32
3219 // CHECK5-NEXT: [[TMP2:%.*]] = load i64, ptr [[DOTPREVIOUS_UB__ADDR]], align 8
3220 // CHECK5-NEXT: [[CONV1:%.*]] = trunc i64 [[TMP2]] to i32
3221 // CHECK5-NEXT: store i32 [[CONV]], ptr [[DOTOMP_LB]], align 4
3222 // CHECK5-NEXT: store i32 [[CONV1]], ptr [[DOTOMP_UB]], align 4
3223 // CHECK5-NEXT: store i32 1, ptr [[DOTOMP_STRIDE]], align 4
3224 // CHECK5-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
3225 // CHECK5-NEXT: [[TMP3:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4
3226 // CHECK5-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
3227 // CHECK5-NEXT: [[TMP5:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
3228 // CHECK5-NEXT: [[TMP6:%.*]] = load i32, ptr [[TMP5]], align 4
3229 // CHECK5-NEXT: call void @__kmpc_dispatch_init_4(ptr @[[GLOB3]], i32 [[TMP6]], i32 35, i32 [[TMP3]], i32 [[TMP4]], i32 1, i32 61)
3230 // CHECK5-NEXT: br label [[OMP_DISPATCH_COND:%.*]]
3231 // CHECK5: omp.dispatch.cond:
3232 // CHECK5-NEXT: [[TMP7:%.*]] = call i32 @__kmpc_dispatch_next_4(ptr @[[GLOB3]], i32 [[TMP6]], ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]])
3233 // CHECK5-NEXT: [[TOBOOL:%.*]] = icmp ne i32 [[TMP7]], 0
3234 // CHECK5-NEXT: br i1 [[TOBOOL]], label [[OMP_DISPATCH_BODY:%.*]], label [[OMP_DISPATCH_END:%.*]]
3235 // CHECK5: omp.dispatch.body:
3236 // CHECK5-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4
3237 // CHECK5-NEXT: store i32 [[TMP8]], ptr [[DOTOMP_IV]], align 4
3238 // CHECK5-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
3239 // CHECK5: omp.inner.for.cond:
3240 // CHECK5-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP13:![0-9]+]]
3241 // CHECK5-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP13]]
3242 // CHECK5-NEXT: [[CMP:%.*]] = icmp sle i32 [[TMP9]], [[TMP10]]
3243 // CHECK5-NEXT: br i1 [[CMP]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
3244 // CHECK5: omp.inner.for.body:
3245 // CHECK5-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP13]]
3246 // CHECK5-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP11]], 1
3247 // CHECK5-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
3248 // CHECK5-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP13]]
3249 // CHECK5-NEXT: [[A:%.*]] = getelementptr inbounds nuw [[STRUCT_SS:%.*]], ptr [[TMP0]], i32 0, i32 0
3250 // CHECK5-NEXT: [[TMP12:%.*]] = load i32, ptr [[I]], align 4, !llvm.access.group [[ACC_GRP13]]
3251 // CHECK5-NEXT: [[IDXPROM:%.*]] = sext i32 [[TMP12]] to i64
3252 // CHECK5-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds [123 x i32], ptr [[A]], i64 0, i64 [[IDXPROM]]
3253 // CHECK5-NEXT: store i32 0, ptr [[ARRAYIDX]], align 4, !llvm.access.group [[ACC_GRP13]]
3254 // CHECK5-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
3255 // CHECK5: omp.body.continue:
3256 // CHECK5-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
3257 // CHECK5: omp.inner.for.inc:
3258 // CHECK5-NEXT: [[TMP13:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP13]]
3259 // CHECK5-NEXT: [[ADD2:%.*]] = add nsw i32 [[TMP13]], 1
3260 // CHECK5-NEXT: store i32 [[ADD2]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP13]]
3261 // CHECK5-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP14:![0-9]+]]
3262 // CHECK5: omp.inner.for.end:
3263 // CHECK5-NEXT: br label [[OMP_DISPATCH_INC:%.*]]
3264 // CHECK5: omp.dispatch.inc:
3265 // CHECK5-NEXT: br label [[OMP_DISPATCH_COND]]
3266 // CHECK5: omp.dispatch.end:
3267 // CHECK5-NEXT: call void @__kmpc_dispatch_deinit(ptr @[[GLOB3]], i32 [[TMP6]])
3268 // CHECK5-NEXT: ret void
3271 // CHECK7-LABEL: define {{[^@]+}}@_Z21teams_template_structv
3272 // CHECK7-SAME: () #[[ATTR0:[0-9]+]] {
3273 // CHECK7-NEXT: entry:
3274 // CHECK7-NEXT: [[V:%.*]] = alloca [[STRUCT_SS:%.*]], align 4
3275 // CHECK7-NEXT: [[CALL:%.*]] = call noundef i32 @_ZN2SSIiLi123ELx456EE3fooEv(ptr noundef nonnull align 4 dereferenceable(496) [[V]])
3276 // CHECK7-NEXT: ret i32 [[CALL]]
3279 // CHECK7-LABEL: define {{[^@]+}}@_ZN2SSIiLi123ELx456EE3fooEv
3280 // CHECK7-SAME: (ptr noundef nonnull align 4 dereferenceable(496) [[THIS:%.*]]) #[[ATTR0]] comdat align 2 {
3281 // CHECK7-NEXT: entry:
3282 // CHECK7-NEXT: [[THIS_ADDR:%.*]] = alloca ptr, align 4
3283 // CHECK7-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [1 x ptr], align 4
3284 // CHECK7-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [1 x ptr], align 4
3285 // CHECK7-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [1 x ptr], align 4
3286 // CHECK7-NEXT: [[TMP:%.*]] = alloca i32, align 4
3287 // CHECK7-NEXT: [[KERNEL_ARGS:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS:%.*]], align 8
3288 // CHECK7-NEXT: [[DOTOFFLOAD_BASEPTRS3:%.*]] = alloca [1 x ptr], align 4
3289 // CHECK7-NEXT: [[DOTOFFLOAD_PTRS4:%.*]] = alloca [1 x ptr], align 4
3290 // CHECK7-NEXT: [[DOTOFFLOAD_MAPPERS5:%.*]] = alloca [1 x ptr], align 4
3291 // CHECK7-NEXT: [[_TMP6:%.*]] = alloca i32, align 4
3292 // CHECK7-NEXT: [[KERNEL_ARGS7:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS]], align 8
3293 // CHECK7-NEXT: [[DOTOFFLOAD_BASEPTRS11:%.*]] = alloca [1 x ptr], align 4
3294 // CHECK7-NEXT: [[DOTOFFLOAD_PTRS12:%.*]] = alloca [1 x ptr], align 4
3295 // CHECK7-NEXT: [[DOTOFFLOAD_MAPPERS13:%.*]] = alloca [1 x ptr], align 4
3296 // CHECK7-NEXT: [[_TMP14:%.*]] = alloca i32, align 4
3297 // CHECK7-NEXT: [[KERNEL_ARGS15:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS]], align 8
3298 // CHECK7-NEXT: [[DOTOFFLOAD_BASEPTRS19:%.*]] = alloca [1 x ptr], align 4
3299 // CHECK7-NEXT: [[DOTOFFLOAD_PTRS20:%.*]] = alloca [1 x ptr], align 4
3300 // CHECK7-NEXT: [[DOTOFFLOAD_MAPPERS21:%.*]] = alloca [1 x ptr], align 4
3301 // CHECK7-NEXT: [[_TMP22:%.*]] = alloca i32, align 4
3302 // CHECK7-NEXT: [[KERNEL_ARGS23:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS]], align 8
3303 // CHECK7-NEXT: [[DOTOFFLOAD_BASEPTRS27:%.*]] = alloca [1 x ptr], align 4
3304 // CHECK7-NEXT: [[DOTOFFLOAD_PTRS28:%.*]] = alloca [1 x ptr], align 4
3305 // CHECK7-NEXT: [[DOTOFFLOAD_MAPPERS29:%.*]] = alloca [1 x ptr], align 4
3306 // CHECK7-NEXT: [[_TMP30:%.*]] = alloca i32, align 4
3307 // CHECK7-NEXT: [[KERNEL_ARGS31:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS]], align 8
3308 // CHECK7-NEXT: store ptr [[THIS]], ptr [[THIS_ADDR]], align 4
3309 // CHECK7-NEXT: [[THIS1:%.*]] = load ptr, ptr [[THIS_ADDR]], align 4
3310 // CHECK7-NEXT: [[A:%.*]] = getelementptr inbounds nuw [[STRUCT_SS:%.*]], ptr [[THIS1]], i32 0, i32 0
3311 // CHECK7-NEXT: [[TMP0:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
3312 // CHECK7-NEXT: store ptr [[THIS1]], ptr [[TMP0]], align 4
3313 // CHECK7-NEXT: [[TMP1:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
3314 // CHECK7-NEXT: store ptr [[A]], ptr [[TMP1]], align 4
3315 // CHECK7-NEXT: [[TMP2:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0
3316 // CHECK7-NEXT: store ptr null, ptr [[TMP2]], align 4
3317 // CHECK7-NEXT: [[TMP3:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
3318 // CHECK7-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
3319 // CHECK7-NEXT: [[TMP5:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 0
3320 // CHECK7-NEXT: store i32 3, ptr [[TMP5]], align 4
3321 // CHECK7-NEXT: [[TMP6:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 1
3322 // CHECK7-NEXT: store i32 1, ptr [[TMP6]], align 4
3323 // CHECK7-NEXT: [[TMP7:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 2
3324 // CHECK7-NEXT: store ptr [[TMP3]], ptr [[TMP7]], align 4
3325 // CHECK7-NEXT: [[TMP8:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 3
3326 // CHECK7-NEXT: store ptr [[TMP4]], ptr [[TMP8]], align 4
3327 // CHECK7-NEXT: [[TMP9:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 4
3328 // CHECK7-NEXT: store ptr @.offload_sizes, ptr [[TMP9]], align 4
3329 // CHECK7-NEXT: [[TMP10:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 5
3330 // CHECK7-NEXT: store ptr @.offload_maptypes, ptr [[TMP10]], align 4
3331 // CHECK7-NEXT: [[TMP11:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 6
3332 // CHECK7-NEXT: store ptr null, ptr [[TMP11]], align 4
3333 // CHECK7-NEXT: [[TMP12:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 7
3334 // CHECK7-NEXT: store ptr null, ptr [[TMP12]], align 4
3335 // CHECK7-NEXT: [[TMP13:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 8
3336 // CHECK7-NEXT: store i64 123, ptr [[TMP13]], align 8
3337 // CHECK7-NEXT: [[TMP14:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 9
3338 // CHECK7-NEXT: store i64 0, ptr [[TMP14]], align 8
3339 // CHECK7-NEXT: [[TMP15:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 10
3340 // CHECK7-NEXT: store [3 x i32] zeroinitializer, ptr [[TMP15]], align 4
3341 // CHECK7-NEXT: [[TMP16:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 11
3342 // CHECK7-NEXT: store [3 x i32] zeroinitializer, ptr [[TMP16]], align 4
3343 // CHECK7-NEXT: [[TMP17:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 12
3344 // CHECK7-NEXT: store i32 0, ptr [[TMP17]], align 4
3345 // CHECK7-NEXT: [[TMP18:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB3:[0-9]+]], i64 -1, i32 0, i32 0, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l35.region_id, ptr [[KERNEL_ARGS]])
3346 // CHECK7-NEXT: [[TMP19:%.*]] = icmp ne i32 [[TMP18]], 0
3347 // CHECK7-NEXT: br i1 [[TMP19]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
3348 // CHECK7: omp_offload.failed:
3349 // CHECK7-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l35(ptr [[THIS1]]) #[[ATTR2:[0-9]+]]
3350 // CHECK7-NEXT: br label [[OMP_OFFLOAD_CONT]]
3351 // CHECK7: omp_offload.cont:
3352 // CHECK7-NEXT: [[A2:%.*]] = getelementptr inbounds nuw [[STRUCT_SS]], ptr [[THIS1]], i32 0, i32 0
3353 // CHECK7-NEXT: [[TMP20:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS3]], i32 0, i32 0
3354 // CHECK7-NEXT: store ptr [[THIS1]], ptr [[TMP20]], align 4
3355 // CHECK7-NEXT: [[TMP21:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS4]], i32 0, i32 0
3356 // CHECK7-NEXT: store ptr [[A2]], ptr [[TMP21]], align 4
3357 // CHECK7-NEXT: [[TMP22:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_MAPPERS5]], i32 0, i32 0
3358 // CHECK7-NEXT: store ptr null, ptr [[TMP22]], align 4
3359 // CHECK7-NEXT: [[TMP23:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS3]], i32 0, i32 0
3360 // CHECK7-NEXT: [[TMP24:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS4]], i32 0, i32 0
3361 // CHECK7-NEXT: [[TMP25:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS7]], i32 0, i32 0
3362 // CHECK7-NEXT: store i32 3, ptr [[TMP25]], align 4
3363 // CHECK7-NEXT: [[TMP26:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS7]], i32 0, i32 1
3364 // CHECK7-NEXT: store i32 1, ptr [[TMP26]], align 4
3365 // CHECK7-NEXT: [[TMP27:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS7]], i32 0, i32 2
3366 // CHECK7-NEXT: store ptr [[TMP23]], ptr [[TMP27]], align 4
3367 // CHECK7-NEXT: [[TMP28:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS7]], i32 0, i32 3
3368 // CHECK7-NEXT: store ptr [[TMP24]], ptr [[TMP28]], align 4
3369 // CHECK7-NEXT: [[TMP29:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS7]], i32 0, i32 4
3370 // CHECK7-NEXT: store ptr @.offload_sizes.1, ptr [[TMP29]], align 4
3371 // CHECK7-NEXT: [[TMP30:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS7]], i32 0, i32 5
3372 // CHECK7-NEXT: store ptr @.offload_maptypes.2, ptr [[TMP30]], align 4
3373 // CHECK7-NEXT: [[TMP31:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS7]], i32 0, i32 6
3374 // CHECK7-NEXT: store ptr null, ptr [[TMP31]], align 4
3375 // CHECK7-NEXT: [[TMP32:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS7]], i32 0, i32 7
3376 // CHECK7-NEXT: store ptr null, ptr [[TMP32]], align 4
3377 // CHECK7-NEXT: [[TMP33:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS7]], i32 0, i32 8
3378 // CHECK7-NEXT: store i64 123, ptr [[TMP33]], align 8
3379 // CHECK7-NEXT: [[TMP34:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS7]], i32 0, i32 9
3380 // CHECK7-NEXT: store i64 0, ptr [[TMP34]], align 8
3381 // CHECK7-NEXT: [[TMP35:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS7]], i32 0, i32 10
3382 // CHECK7-NEXT: store [3 x i32] zeroinitializer, ptr [[TMP35]], align 4
3383 // CHECK7-NEXT: [[TMP36:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS7]], i32 0, i32 11
3384 // CHECK7-NEXT: store [3 x i32] zeroinitializer, ptr [[TMP36]], align 4
3385 // CHECK7-NEXT: [[TMP37:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS7]], i32 0, i32 12
3386 // CHECK7-NEXT: store i32 0, ptr [[TMP37]], align 4
3387 // CHECK7-NEXT: [[TMP38:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB3]], i64 -1, i32 0, i32 0, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l40.region_id, ptr [[KERNEL_ARGS7]])
3388 // CHECK7-NEXT: [[TMP39:%.*]] = icmp ne i32 [[TMP38]], 0
3389 // CHECK7-NEXT: br i1 [[TMP39]], label [[OMP_OFFLOAD_FAILED8:%.*]], label [[OMP_OFFLOAD_CONT9:%.*]]
3390 // CHECK7: omp_offload.failed8:
3391 // CHECK7-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l40(ptr [[THIS1]]) #[[ATTR2]]
3392 // CHECK7-NEXT: br label [[OMP_OFFLOAD_CONT9]]
3393 // CHECK7: omp_offload.cont9:
3394 // CHECK7-NEXT: [[A10:%.*]] = getelementptr inbounds nuw [[STRUCT_SS]], ptr [[THIS1]], i32 0, i32 0
3395 // CHECK7-NEXT: [[TMP40:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS11]], i32 0, i32 0
3396 // CHECK7-NEXT: store ptr [[THIS1]], ptr [[TMP40]], align 4
3397 // CHECK7-NEXT: [[TMP41:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS12]], i32 0, i32 0
3398 // CHECK7-NEXT: store ptr [[A10]], ptr [[TMP41]], align 4
3399 // CHECK7-NEXT: [[TMP42:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_MAPPERS13]], i32 0, i32 0
3400 // CHECK7-NEXT: store ptr null, ptr [[TMP42]], align 4
3401 // CHECK7-NEXT: [[TMP43:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS11]], i32 0, i32 0
3402 // CHECK7-NEXT: [[TMP44:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS12]], i32 0, i32 0
3403 // CHECK7-NEXT: [[TMP45:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS15]], i32 0, i32 0
3404 // CHECK7-NEXT: store i32 3, ptr [[TMP45]], align 4
3405 // CHECK7-NEXT: [[TMP46:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS15]], i32 0, i32 1
3406 // CHECK7-NEXT: store i32 1, ptr [[TMP46]], align 4
3407 // CHECK7-NEXT: [[TMP47:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS15]], i32 0, i32 2
3408 // CHECK7-NEXT: store ptr [[TMP43]], ptr [[TMP47]], align 4
3409 // CHECK7-NEXT: [[TMP48:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS15]], i32 0, i32 3
3410 // CHECK7-NEXT: store ptr [[TMP44]], ptr [[TMP48]], align 4
3411 // CHECK7-NEXT: [[TMP49:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS15]], i32 0, i32 4
3412 // CHECK7-NEXT: store ptr @.offload_sizes.3, ptr [[TMP49]], align 4
3413 // CHECK7-NEXT: [[TMP50:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS15]], i32 0, i32 5
3414 // CHECK7-NEXT: store ptr @.offload_maptypes.4, ptr [[TMP50]], align 4
3415 // CHECK7-NEXT: [[TMP51:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS15]], i32 0, i32 6
3416 // CHECK7-NEXT: store ptr null, ptr [[TMP51]], align 4
3417 // CHECK7-NEXT: [[TMP52:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS15]], i32 0, i32 7
3418 // CHECK7-NEXT: store ptr null, ptr [[TMP52]], align 4
3419 // CHECK7-NEXT: [[TMP53:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS15]], i32 0, i32 8
3420 // CHECK7-NEXT: store i64 123, ptr [[TMP53]], align 8
3421 // CHECK7-NEXT: [[TMP54:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS15]], i32 0, i32 9
3422 // CHECK7-NEXT: store i64 0, ptr [[TMP54]], align 8
3423 // CHECK7-NEXT: [[TMP55:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS15]], i32 0, i32 10
3424 // CHECK7-NEXT: store [3 x i32] zeroinitializer, ptr [[TMP55]], align 4
3425 // CHECK7-NEXT: [[TMP56:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS15]], i32 0, i32 11
3426 // CHECK7-NEXT: store [3 x i32] zeroinitializer, ptr [[TMP56]], align 4
3427 // CHECK7-NEXT: [[TMP57:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS15]], i32 0, i32 12
3428 // CHECK7-NEXT: store i32 0, ptr [[TMP57]], align 4
3429 // CHECK7-NEXT: [[TMP58:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB3]], i64 -1, i32 0, i32 0, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l45.region_id, ptr [[KERNEL_ARGS15]])
3430 // CHECK7-NEXT: [[TMP59:%.*]] = icmp ne i32 [[TMP58]], 0
3431 // CHECK7-NEXT: br i1 [[TMP59]], label [[OMP_OFFLOAD_FAILED16:%.*]], label [[OMP_OFFLOAD_CONT17:%.*]]
3432 // CHECK7: omp_offload.failed16:
3433 // CHECK7-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l45(ptr [[THIS1]]) #[[ATTR2]]
3434 // CHECK7-NEXT: br label [[OMP_OFFLOAD_CONT17]]
3435 // CHECK7: omp_offload.cont17:
3436 // CHECK7-NEXT: [[A18:%.*]] = getelementptr inbounds nuw [[STRUCT_SS]], ptr [[THIS1]], i32 0, i32 0
3437 // CHECK7-NEXT: [[TMP60:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS19]], i32 0, i32 0
3438 // CHECK7-NEXT: store ptr [[THIS1]], ptr [[TMP60]], align 4
3439 // CHECK7-NEXT: [[TMP61:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS20]], i32 0, i32 0
3440 // CHECK7-NEXT: store ptr [[A18]], ptr [[TMP61]], align 4
3441 // CHECK7-NEXT: [[TMP62:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_MAPPERS21]], i32 0, i32 0
3442 // CHECK7-NEXT: store ptr null, ptr [[TMP62]], align 4
3443 // CHECK7-NEXT: [[TMP63:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS19]], i32 0, i32 0
3444 // CHECK7-NEXT: [[TMP64:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS20]], i32 0, i32 0
3445 // CHECK7-NEXT: [[TMP65:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS23]], i32 0, i32 0
3446 // CHECK7-NEXT: store i32 3, ptr [[TMP65]], align 4
3447 // CHECK7-NEXT: [[TMP66:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS23]], i32 0, i32 1
3448 // CHECK7-NEXT: store i32 1, ptr [[TMP66]], align 4
3449 // CHECK7-NEXT: [[TMP67:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS23]], i32 0, i32 2
3450 // CHECK7-NEXT: store ptr [[TMP63]], ptr [[TMP67]], align 4
3451 // CHECK7-NEXT: [[TMP68:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS23]], i32 0, i32 3
3452 // CHECK7-NEXT: store ptr [[TMP64]], ptr [[TMP68]], align 4
3453 // CHECK7-NEXT: [[TMP69:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS23]], i32 0, i32 4
3454 // CHECK7-NEXT: store ptr @.offload_sizes.5, ptr [[TMP69]], align 4
3455 // CHECK7-NEXT: [[TMP70:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS23]], i32 0, i32 5
3456 // CHECK7-NEXT: store ptr @.offload_maptypes.6, ptr [[TMP70]], align 4
3457 // CHECK7-NEXT: [[TMP71:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS23]], i32 0, i32 6
3458 // CHECK7-NEXT: store ptr null, ptr [[TMP71]], align 4
3459 // CHECK7-NEXT: [[TMP72:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS23]], i32 0, i32 7
3460 // CHECK7-NEXT: store ptr null, ptr [[TMP72]], align 4
3461 // CHECK7-NEXT: [[TMP73:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS23]], i32 0, i32 8
3462 // CHECK7-NEXT: store i64 123, ptr [[TMP73]], align 8
3463 // CHECK7-NEXT: [[TMP74:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS23]], i32 0, i32 9
3464 // CHECK7-NEXT: store i64 0, ptr [[TMP74]], align 8
3465 // CHECK7-NEXT: [[TMP75:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS23]], i32 0, i32 10
3466 // CHECK7-NEXT: store [3 x i32] zeroinitializer, ptr [[TMP75]], align 4
3467 // CHECK7-NEXT: [[TMP76:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS23]], i32 0, i32 11
3468 // CHECK7-NEXT: store [3 x i32] zeroinitializer, ptr [[TMP76]], align 4
3469 // CHECK7-NEXT: [[TMP77:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS23]], i32 0, i32 12
3470 // CHECK7-NEXT: store i32 0, ptr [[TMP77]], align 4
3471 // CHECK7-NEXT: [[TMP78:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB3]], i64 -1, i32 0, i32 0, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l51.region_id, ptr [[KERNEL_ARGS23]])
3472 // CHECK7-NEXT: [[TMP79:%.*]] = icmp ne i32 [[TMP78]], 0
3473 // CHECK7-NEXT: br i1 [[TMP79]], label [[OMP_OFFLOAD_FAILED24:%.*]], label [[OMP_OFFLOAD_CONT25:%.*]]
3474 // CHECK7: omp_offload.failed24:
3475 // CHECK7-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l51(ptr [[THIS1]]) #[[ATTR2]]
3476 // CHECK7-NEXT: br label [[OMP_OFFLOAD_CONT25]]
3477 // CHECK7: omp_offload.cont25:
3478 // CHECK7-NEXT: [[A26:%.*]] = getelementptr inbounds nuw [[STRUCT_SS]], ptr [[THIS1]], i32 0, i32 0
3479 // CHECK7-NEXT: [[TMP80:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS27]], i32 0, i32 0
3480 // CHECK7-NEXT: store ptr [[THIS1]], ptr [[TMP80]], align 4
3481 // CHECK7-NEXT: [[TMP81:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS28]], i32 0, i32 0
3482 // CHECK7-NEXT: store ptr [[A26]], ptr [[TMP81]], align 4
3483 // CHECK7-NEXT: [[TMP82:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_MAPPERS29]], i32 0, i32 0
3484 // CHECK7-NEXT: store ptr null, ptr [[TMP82]], align 4
3485 // CHECK7-NEXT: [[TMP83:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS27]], i32 0, i32 0
3486 // CHECK7-NEXT: [[TMP84:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS28]], i32 0, i32 0
3487 // CHECK7-NEXT: [[TMP85:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS31]], i32 0, i32 0
3488 // CHECK7-NEXT: store i32 3, ptr [[TMP85]], align 4
3489 // CHECK7-NEXT: [[TMP86:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS31]], i32 0, i32 1
3490 // CHECK7-NEXT: store i32 1, ptr [[TMP86]], align 4
3491 // CHECK7-NEXT: [[TMP87:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS31]], i32 0, i32 2
3492 // CHECK7-NEXT: store ptr [[TMP83]], ptr [[TMP87]], align 4
3493 // CHECK7-NEXT: [[TMP88:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS31]], i32 0, i32 3
3494 // CHECK7-NEXT: store ptr [[TMP84]], ptr [[TMP88]], align 4
3495 // CHECK7-NEXT: [[TMP89:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS31]], i32 0, i32 4
3496 // CHECK7-NEXT: store ptr @.offload_sizes.7, ptr [[TMP89]], align 4
3497 // CHECK7-NEXT: [[TMP90:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS31]], i32 0, i32 5
3498 // CHECK7-NEXT: store ptr @.offload_maptypes.8, ptr [[TMP90]], align 4
3499 // CHECK7-NEXT: [[TMP91:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS31]], i32 0, i32 6
3500 // CHECK7-NEXT: store ptr null, ptr [[TMP91]], align 4
3501 // CHECK7-NEXT: [[TMP92:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS31]], i32 0, i32 7
3502 // CHECK7-NEXT: store ptr null, ptr [[TMP92]], align 4
3503 // CHECK7-NEXT: [[TMP93:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS31]], i32 0, i32 8
3504 // CHECK7-NEXT: store i64 123, ptr [[TMP93]], align 8
3505 // CHECK7-NEXT: [[TMP94:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS31]], i32 0, i32 9
3506 // CHECK7-NEXT: store i64 0, ptr [[TMP94]], align 8
3507 // CHECK7-NEXT: [[TMP95:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS31]], i32 0, i32 10
3508 // CHECK7-NEXT: store [3 x i32] zeroinitializer, ptr [[TMP95]], align 4
3509 // CHECK7-NEXT: [[TMP96:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS31]], i32 0, i32 11
3510 // CHECK7-NEXT: store [3 x i32] zeroinitializer, ptr [[TMP96]], align 4
3511 // CHECK7-NEXT: [[TMP97:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS31]], i32 0, i32 12
3512 // CHECK7-NEXT: store i32 0, ptr [[TMP97]], align 4
3513 // CHECK7-NEXT: [[TMP98:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB3]], i64 -1, i32 0, i32 0, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l57.region_id, ptr [[KERNEL_ARGS31]])
3514 // CHECK7-NEXT: [[TMP99:%.*]] = icmp ne i32 [[TMP98]], 0
3515 // CHECK7-NEXT: br i1 [[TMP99]], label [[OMP_OFFLOAD_FAILED32:%.*]], label [[OMP_OFFLOAD_CONT33:%.*]]
3516 // CHECK7: omp_offload.failed32:
3517 // CHECK7-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l57(ptr [[THIS1]]) #[[ATTR2]]
3518 // CHECK7-NEXT: br label [[OMP_OFFLOAD_CONT33]]
3519 // CHECK7: omp_offload.cont33:
3520 // CHECK7-NEXT: [[A34:%.*]] = getelementptr inbounds nuw [[STRUCT_SS]], ptr [[THIS1]], i32 0, i32 0
3521 // CHECK7-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds [123 x i32], ptr [[A34]], i32 0, i32 0
3522 // CHECK7-NEXT: [[TMP100:%.*]] = load i32, ptr [[ARRAYIDX]], align 4
3523 // CHECK7-NEXT: ret i32 [[TMP100]]
3526 // CHECK7-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l35
3527 // CHECK7-SAME: (ptr noundef [[THIS:%.*]]) #[[ATTR1:[0-9]+]] {
3528 // CHECK7-NEXT: entry:
3529 // CHECK7-NEXT: [[THIS_ADDR:%.*]] = alloca ptr, align 4
3530 // CHECK7-NEXT: store ptr [[THIS]], ptr [[THIS_ADDR]], align 4
3531 // CHECK7-NEXT: [[TMP0:%.*]] = load ptr, ptr [[THIS_ADDR]], align 4
3532 // CHECK7-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_teams(ptr @[[GLOB3]], i32 1, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l35.omp_outlined, ptr [[TMP0]])
3533 // CHECK7-NEXT: ret void
3536 // CHECK7-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l35.omp_outlined
3537 // CHECK7-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], ptr noundef [[THIS:%.*]]) #[[ATTR1]] {
3538 // CHECK7-NEXT: entry:
3539 // CHECK7-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 4
3540 // CHECK7-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 4
3541 // CHECK7-NEXT: [[THIS_ADDR:%.*]] = alloca ptr, align 4
3542 // CHECK7-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
3543 // CHECK7-NEXT: [[TMP:%.*]] = alloca i32, align 4
3544 // CHECK7-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4
3545 // CHECK7-NEXT: [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4
3546 // CHECK7-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
3547 // CHECK7-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
3548 // CHECK7-NEXT: [[I:%.*]] = alloca i32, align 4
3549 // CHECK7-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 4
3550 // CHECK7-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 4
3551 // CHECK7-NEXT: store ptr [[THIS]], ptr [[THIS_ADDR]], align 4
3552 // CHECK7-NEXT: [[TMP0:%.*]] = load ptr, ptr [[THIS_ADDR]], align 4
3553 // CHECK7-NEXT: store i32 0, ptr [[DOTOMP_COMB_LB]], align 4
3554 // CHECK7-NEXT: store i32 122, ptr [[DOTOMP_COMB_UB]], align 4
3555 // CHECK7-NEXT: store i32 1, ptr [[DOTOMP_STRIDE]], align 4
3556 // CHECK7-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
3557 // CHECK7-NEXT: [[TMP1:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
3558 // CHECK7-NEXT: [[TMP2:%.*]] = load i32, ptr [[TMP1]], align 4
3559 // CHECK7-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB1:[0-9]+]], i32 [[TMP2]], i32 92, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
3560 // CHECK7-NEXT: [[TMP3:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
3561 // CHECK7-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP3]], 122
3562 // CHECK7-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
3563 // CHECK7: cond.true:
3564 // CHECK7-NEXT: br label [[COND_END:%.*]]
3565 // CHECK7: cond.false:
3566 // CHECK7-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
3567 // CHECK7-NEXT: br label [[COND_END]]
3568 // CHECK7: cond.end:
3569 // CHECK7-NEXT: [[COND:%.*]] = phi i32 [ 122, [[COND_TRUE]] ], [ [[TMP4]], [[COND_FALSE]] ]
3570 // CHECK7-NEXT: store i32 [[COND]], ptr [[DOTOMP_COMB_UB]], align 4
3571 // CHECK7-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
3572 // CHECK7-NEXT: store i32 [[TMP5]], ptr [[DOTOMP_IV]], align 4
3573 // CHECK7-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
3574 // CHECK7: omp.inner.for.cond:
3575 // CHECK7-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
3576 // CHECK7-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
3577 // CHECK7-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP6]], [[TMP7]]
3578 // CHECK7-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
3579 // CHECK7: omp.inner.for.body:
3580 // CHECK7-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
3581 // CHECK7-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
3582 // CHECK7-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_call(ptr @[[GLOB3]], i32 3, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l35.omp_outlined.omp_outlined, i32 [[TMP8]], i32 [[TMP9]], ptr [[TMP0]])
3583 // CHECK7-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
3584 // CHECK7: omp.inner.for.inc:
3585 // CHECK7-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
3586 // CHECK7-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4
3587 // CHECK7-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP10]], [[TMP11]]
3588 // CHECK7-NEXT: store i32 [[ADD]], ptr [[DOTOMP_IV]], align 4
3589 // CHECK7-NEXT: br label [[OMP_INNER_FOR_COND]]
3590 // CHECK7: omp.inner.for.end:
3591 // CHECK7-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
3592 // CHECK7: omp.loop.exit:
3593 // CHECK7-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB1]], i32 [[TMP2]])
3594 // CHECK7-NEXT: ret void
3597 // CHECK7-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l35.omp_outlined.omp_outlined
3598 // CHECK7-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], i32 noundef [[DOTPREVIOUS_LB_:%.*]], i32 noundef [[DOTPREVIOUS_UB_:%.*]], ptr noundef [[THIS:%.*]]) #[[ATTR1]] {
3599 // CHECK7-NEXT: entry:
3600 // CHECK7-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 4
3601 // CHECK7-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 4
3602 // CHECK7-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i32, align 4
3603 // CHECK7-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i32, align 4
3604 // CHECK7-NEXT: [[THIS_ADDR:%.*]] = alloca ptr, align 4
3605 // CHECK7-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
3606 // CHECK7-NEXT: [[TMP:%.*]] = alloca i32, align 4
3607 // CHECK7-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4
3608 // CHECK7-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4
3609 // CHECK7-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
3610 // CHECK7-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
3611 // CHECK7-NEXT: [[I:%.*]] = alloca i32, align 4
3612 // CHECK7-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 4
3613 // CHECK7-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 4
3614 // CHECK7-NEXT: store i32 [[DOTPREVIOUS_LB_]], ptr [[DOTPREVIOUS_LB__ADDR]], align 4
3615 // CHECK7-NEXT: store i32 [[DOTPREVIOUS_UB_]], ptr [[DOTPREVIOUS_UB__ADDR]], align 4
3616 // CHECK7-NEXT: store ptr [[THIS]], ptr [[THIS_ADDR]], align 4
3617 // CHECK7-NEXT: [[TMP0:%.*]] = load ptr, ptr [[THIS_ADDR]], align 4
3618 // CHECK7-NEXT: store i32 0, ptr [[DOTOMP_LB]], align 4
3619 // CHECK7-NEXT: store i32 122, ptr [[DOTOMP_UB]], align 4
3620 // CHECK7-NEXT: [[TMP1:%.*]] = load i32, ptr [[DOTPREVIOUS_LB__ADDR]], align 4
3621 // CHECK7-NEXT: [[TMP2:%.*]] = load i32, ptr [[DOTPREVIOUS_UB__ADDR]], align 4
3622 // CHECK7-NEXT: store i32 [[TMP1]], ptr [[DOTOMP_LB]], align 4
3623 // CHECK7-NEXT: store i32 [[TMP2]], ptr [[DOTOMP_UB]], align 4
3624 // CHECK7-NEXT: store i32 1, ptr [[DOTOMP_STRIDE]], align 4
3625 // CHECK7-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
3626 // CHECK7-NEXT: [[TMP3:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
3627 // CHECK7-NEXT: [[TMP4:%.*]] = load i32, ptr [[TMP3]], align 4
3628 // CHECK7-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB2:[0-9]+]], i32 [[TMP4]], i32 34, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
3629 // CHECK7-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
3630 // CHECK7-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP5]], 122
3631 // CHECK7-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
3632 // CHECK7: cond.true:
3633 // CHECK7-NEXT: br label [[COND_END:%.*]]
3634 // CHECK7: cond.false:
3635 // CHECK7-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
3636 // CHECK7-NEXT: br label [[COND_END]]
3637 // CHECK7: cond.end:
3638 // CHECK7-NEXT: [[COND:%.*]] = phi i32 [ 122, [[COND_TRUE]] ], [ [[TMP6]], [[COND_FALSE]] ]
3639 // CHECK7-NEXT: store i32 [[COND]], ptr [[DOTOMP_UB]], align 4
3640 // CHECK7-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4
3641 // CHECK7-NEXT: store i32 [[TMP7]], ptr [[DOTOMP_IV]], align 4
3642 // CHECK7-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
3643 // CHECK7: omp.inner.for.cond:
3644 // CHECK7-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
3645 // CHECK7-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
3646 // CHECK7-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP8]], [[TMP9]]
3647 // CHECK7-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
3648 // CHECK7: omp.inner.for.body:
3649 // CHECK7-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
3650 // CHECK7-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP10]], 1
3651 // CHECK7-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
3652 // CHECK7-NEXT: store i32 [[ADD]], ptr [[I]], align 4
3653 // CHECK7-NEXT: [[A:%.*]] = getelementptr inbounds nuw [[STRUCT_SS:%.*]], ptr [[TMP0]], i32 0, i32 0
3654 // CHECK7-NEXT: [[TMP11:%.*]] = load i32, ptr [[I]], align 4
3655 // CHECK7-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds [123 x i32], ptr [[A]], i32 0, i32 [[TMP11]]
3656 // CHECK7-NEXT: store i32 0, ptr [[ARRAYIDX]], align 4
3657 // CHECK7-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
3658 // CHECK7: omp.body.continue:
3659 // CHECK7-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
3660 // CHECK7: omp.inner.for.inc:
3661 // CHECK7-NEXT: [[TMP12:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
3662 // CHECK7-NEXT: [[ADD2:%.*]] = add nsw i32 [[TMP12]], 1
3663 // CHECK7-NEXT: store i32 [[ADD2]], ptr [[DOTOMP_IV]], align 4
3664 // CHECK7-NEXT: br label [[OMP_INNER_FOR_COND]]
3665 // CHECK7: omp.inner.for.end:
3666 // CHECK7-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
3667 // CHECK7: omp.loop.exit:
3668 // CHECK7-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB2]], i32 [[TMP4]])
3669 // CHECK7-NEXT: ret void
3672 // CHECK7-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l40
3673 // CHECK7-SAME: (ptr noundef [[THIS:%.*]]) #[[ATTR1]] {
3674 // CHECK7-NEXT: entry:
3675 // CHECK7-NEXT: [[THIS_ADDR:%.*]] = alloca ptr, align 4
3676 // CHECK7-NEXT: store ptr [[THIS]], ptr [[THIS_ADDR]], align 4
3677 // CHECK7-NEXT: [[TMP0:%.*]] = load ptr, ptr [[THIS_ADDR]], align 4
3678 // CHECK7-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_teams(ptr @[[GLOB3]], i32 1, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l40.omp_outlined, ptr [[TMP0]])
3679 // CHECK7-NEXT: ret void
3682 // CHECK7-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l40.omp_outlined
3683 // CHECK7-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], ptr noundef [[THIS:%.*]]) #[[ATTR1]] {
3684 // CHECK7-NEXT: entry:
3685 // CHECK7-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 4
3686 // CHECK7-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 4
3687 // CHECK7-NEXT: [[THIS_ADDR:%.*]] = alloca ptr, align 4
3688 // CHECK7-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
3689 // CHECK7-NEXT: [[TMP:%.*]] = alloca i32, align 4
3690 // CHECK7-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4
3691 // CHECK7-NEXT: [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4
3692 // CHECK7-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
3693 // CHECK7-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
3694 // CHECK7-NEXT: [[I:%.*]] = alloca i32, align 4
3695 // CHECK7-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 4
3696 // CHECK7-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 4
3697 // CHECK7-NEXT: store ptr [[THIS]], ptr [[THIS_ADDR]], align 4
3698 // CHECK7-NEXT: [[TMP0:%.*]] = load ptr, ptr [[THIS_ADDR]], align 4
3699 // CHECK7-NEXT: store i32 0, ptr [[DOTOMP_COMB_LB]], align 4
3700 // CHECK7-NEXT: store i32 122, ptr [[DOTOMP_COMB_UB]], align 4
3701 // CHECK7-NEXT: store i32 1, ptr [[DOTOMP_STRIDE]], align 4
3702 // CHECK7-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
3703 // CHECK7-NEXT: [[TMP1:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
3704 // CHECK7-NEXT: [[TMP2:%.*]] = load i32, ptr [[TMP1]], align 4
3705 // CHECK7-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB1]], i32 [[TMP2]], i32 92, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
3706 // CHECK7-NEXT: [[TMP3:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
3707 // CHECK7-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP3]], 122
3708 // CHECK7-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
3709 // CHECK7: cond.true:
3710 // CHECK7-NEXT: br label [[COND_END:%.*]]
3711 // CHECK7: cond.false:
3712 // CHECK7-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
3713 // CHECK7-NEXT: br label [[COND_END]]
3714 // CHECK7: cond.end:
3715 // CHECK7-NEXT: [[COND:%.*]] = phi i32 [ 122, [[COND_TRUE]] ], [ [[TMP4]], [[COND_FALSE]] ]
3716 // CHECK7-NEXT: store i32 [[COND]], ptr [[DOTOMP_COMB_UB]], align 4
3717 // CHECK7-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
3718 // CHECK7-NEXT: store i32 [[TMP5]], ptr [[DOTOMP_IV]], align 4
3719 // CHECK7-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
3720 // CHECK7: omp.inner.for.cond:
3721 // CHECK7-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
3722 // CHECK7-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
3723 // CHECK7-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP6]], [[TMP7]]
3724 // CHECK7-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
3725 // CHECK7: omp.inner.for.body:
3726 // CHECK7-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
3727 // CHECK7-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
3728 // CHECK7-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_call(ptr @[[GLOB3]], i32 3, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l40.omp_outlined.omp_outlined, i32 [[TMP8]], i32 [[TMP9]], ptr [[TMP0]])
3729 // CHECK7-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
3730 // CHECK7: omp.inner.for.inc:
3731 // CHECK7-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
3732 // CHECK7-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4
3733 // CHECK7-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP10]], [[TMP11]]
3734 // CHECK7-NEXT: store i32 [[ADD]], ptr [[DOTOMP_IV]], align 4
3735 // CHECK7-NEXT: br label [[OMP_INNER_FOR_COND]]
3736 // CHECK7: omp.inner.for.end:
3737 // CHECK7-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
3738 // CHECK7: omp.loop.exit:
3739 // CHECK7-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB1]], i32 [[TMP2]])
3740 // CHECK7-NEXT: ret void
3743 // CHECK7-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l40.omp_outlined.omp_outlined
3744 // CHECK7-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], i32 noundef [[DOTPREVIOUS_LB_:%.*]], i32 noundef [[DOTPREVIOUS_UB_:%.*]], ptr noundef [[THIS:%.*]]) #[[ATTR1]] {
3745 // CHECK7-NEXT: entry:
3746 // CHECK7-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 4
3747 // CHECK7-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 4
3748 // CHECK7-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i32, align 4
3749 // CHECK7-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i32, align 4
3750 // CHECK7-NEXT: [[THIS_ADDR:%.*]] = alloca ptr, align 4
3751 // CHECK7-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
3752 // CHECK7-NEXT: [[TMP:%.*]] = alloca i32, align 4
3753 // CHECK7-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4
3754 // CHECK7-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4
3755 // CHECK7-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
3756 // CHECK7-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
3757 // CHECK7-NEXT: [[I:%.*]] = alloca i32, align 4
3758 // CHECK7-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 4
3759 // CHECK7-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 4
3760 // CHECK7-NEXT: store i32 [[DOTPREVIOUS_LB_]], ptr [[DOTPREVIOUS_LB__ADDR]], align 4
3761 // CHECK7-NEXT: store i32 [[DOTPREVIOUS_UB_]], ptr [[DOTPREVIOUS_UB__ADDR]], align 4
3762 // CHECK7-NEXT: store ptr [[THIS]], ptr [[THIS_ADDR]], align 4
3763 // CHECK7-NEXT: [[TMP0:%.*]] = load ptr, ptr [[THIS_ADDR]], align 4
3764 // CHECK7-NEXT: store i32 0, ptr [[DOTOMP_LB]], align 4
3765 // CHECK7-NEXT: store i32 122, ptr [[DOTOMP_UB]], align 4
3766 // CHECK7-NEXT: [[TMP1:%.*]] = load i32, ptr [[DOTPREVIOUS_LB__ADDR]], align 4
3767 // CHECK7-NEXT: [[TMP2:%.*]] = load i32, ptr [[DOTPREVIOUS_UB__ADDR]], align 4
3768 // CHECK7-NEXT: store i32 [[TMP1]], ptr [[DOTOMP_LB]], align 4
3769 // CHECK7-NEXT: store i32 [[TMP2]], ptr [[DOTOMP_UB]], align 4
3770 // CHECK7-NEXT: store i32 1, ptr [[DOTOMP_STRIDE]], align 4
3771 // CHECK7-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
3772 // CHECK7-NEXT: [[TMP3:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
3773 // CHECK7-NEXT: [[TMP4:%.*]] = load i32, ptr [[TMP3]], align 4
3774 // CHECK7-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB2]], i32 [[TMP4]], i32 34, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
3775 // CHECK7-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
3776 // CHECK7-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP5]], 122
3777 // CHECK7-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
3778 // CHECK7: cond.true:
3779 // CHECK7-NEXT: br label [[COND_END:%.*]]
3780 // CHECK7: cond.false:
3781 // CHECK7-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
3782 // CHECK7-NEXT: br label [[COND_END]]
3783 // CHECK7: cond.end:
3784 // CHECK7-NEXT: [[COND:%.*]] = phi i32 [ 122, [[COND_TRUE]] ], [ [[TMP6]], [[COND_FALSE]] ]
3785 // CHECK7-NEXT: store i32 [[COND]], ptr [[DOTOMP_UB]], align 4
3786 // CHECK7-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4
3787 // CHECK7-NEXT: store i32 [[TMP7]], ptr [[DOTOMP_IV]], align 4
3788 // CHECK7-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
3789 // CHECK7: omp.inner.for.cond:
3790 // CHECK7-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
3791 // CHECK7-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
3792 // CHECK7-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP8]], [[TMP9]]
3793 // CHECK7-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
3794 // CHECK7: omp.inner.for.body:
3795 // CHECK7-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
3796 // CHECK7-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP10]], 1
3797 // CHECK7-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
3798 // CHECK7-NEXT: store i32 [[ADD]], ptr [[I]], align 4
3799 // CHECK7-NEXT: [[A:%.*]] = getelementptr inbounds nuw [[STRUCT_SS:%.*]], ptr [[TMP0]], i32 0, i32 0
3800 // CHECK7-NEXT: [[TMP11:%.*]] = load i32, ptr [[I]], align 4
3801 // CHECK7-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds [123 x i32], ptr [[A]], i32 0, i32 [[TMP11]]
3802 // CHECK7-NEXT: store i32 0, ptr [[ARRAYIDX]], align 4
3803 // CHECK7-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
3804 // CHECK7: omp.body.continue:
3805 // CHECK7-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
3806 // CHECK7: omp.inner.for.inc:
3807 // CHECK7-NEXT: [[TMP12:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
3808 // CHECK7-NEXT: [[ADD2:%.*]] = add nsw i32 [[TMP12]], 1
3809 // CHECK7-NEXT: store i32 [[ADD2]], ptr [[DOTOMP_IV]], align 4
3810 // CHECK7-NEXT: br label [[OMP_INNER_FOR_COND]]
3811 // CHECK7: omp.inner.for.end:
3812 // CHECK7-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
3813 // CHECK7: omp.loop.exit:
3814 // CHECK7-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB2]], i32 [[TMP4]])
3815 // CHECK7-NEXT: ret void
3818 // CHECK7-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l45
3819 // CHECK7-SAME: (ptr noundef [[THIS:%.*]]) #[[ATTR1]] {
3820 // CHECK7-NEXT: entry:
3821 // CHECK7-NEXT: [[THIS_ADDR:%.*]] = alloca ptr, align 4
3822 // CHECK7-NEXT: store ptr [[THIS]], ptr [[THIS_ADDR]], align 4
3823 // CHECK7-NEXT: [[TMP0:%.*]] = load ptr, ptr [[THIS_ADDR]], align 4
3824 // CHECK7-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_teams(ptr @[[GLOB3]], i32 1, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l45.omp_outlined, ptr [[TMP0]])
3825 // CHECK7-NEXT: ret void
3828 // CHECK7-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l45.omp_outlined
3829 // CHECK7-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], ptr noundef [[THIS:%.*]]) #[[ATTR1]] {
3830 // CHECK7-NEXT: entry:
3831 // CHECK7-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 4
3832 // CHECK7-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 4
3833 // CHECK7-NEXT: [[THIS_ADDR:%.*]] = alloca ptr, align 4
3834 // CHECK7-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
3835 // CHECK7-NEXT: [[TMP:%.*]] = alloca i32, align 4
3836 // CHECK7-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4
3837 // CHECK7-NEXT: [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4
3838 // CHECK7-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
3839 // CHECK7-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
3840 // CHECK7-NEXT: [[I:%.*]] = alloca i32, align 4
3841 // CHECK7-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 4
3842 // CHECK7-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 4
3843 // CHECK7-NEXT: store ptr [[THIS]], ptr [[THIS_ADDR]], align 4
3844 // CHECK7-NEXT: [[TMP0:%.*]] = load ptr, ptr [[THIS_ADDR]], align 4
3845 // CHECK7-NEXT: store i32 0, ptr [[DOTOMP_COMB_LB]], align 4
3846 // CHECK7-NEXT: store i32 122, ptr [[DOTOMP_COMB_UB]], align 4
3847 // CHECK7-NEXT: store i32 1, ptr [[DOTOMP_STRIDE]], align 4
3848 // CHECK7-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
3849 // CHECK7-NEXT: [[TMP1:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
3850 // CHECK7-NEXT: [[TMP2:%.*]] = load i32, ptr [[TMP1]], align 4
3851 // CHECK7-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB1]], i32 [[TMP2]], i32 92, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
3852 // CHECK7-NEXT: [[TMP3:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
3853 // CHECK7-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP3]], 122
3854 // CHECK7-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
3855 // CHECK7: cond.true:
3856 // CHECK7-NEXT: br label [[COND_END:%.*]]
3857 // CHECK7: cond.false:
3858 // CHECK7-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
3859 // CHECK7-NEXT: br label [[COND_END]]
3860 // CHECK7: cond.end:
3861 // CHECK7-NEXT: [[COND:%.*]] = phi i32 [ 122, [[COND_TRUE]] ], [ [[TMP4]], [[COND_FALSE]] ]
3862 // CHECK7-NEXT: store i32 [[COND]], ptr [[DOTOMP_COMB_UB]], align 4
3863 // CHECK7-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
3864 // CHECK7-NEXT: store i32 [[TMP5]], ptr [[DOTOMP_IV]], align 4
3865 // CHECK7-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
3866 // CHECK7: omp.inner.for.cond:
3867 // CHECK7-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
3868 // CHECK7-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
3869 // CHECK7-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP6]], [[TMP7]]
3870 // CHECK7-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
3871 // CHECK7: omp.inner.for.body:
3872 // CHECK7-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
3873 // CHECK7-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
3874 // CHECK7-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_call(ptr @[[GLOB3]], i32 3, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l45.omp_outlined.omp_outlined, i32 [[TMP8]], i32 [[TMP9]], ptr [[TMP0]])
3875 // CHECK7-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
3876 // CHECK7: omp.inner.for.inc:
3877 // CHECK7-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
3878 // CHECK7-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4
3879 // CHECK7-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP10]], [[TMP11]]
3880 // CHECK7-NEXT: store i32 [[ADD]], ptr [[DOTOMP_IV]], align 4
3881 // CHECK7-NEXT: br label [[OMP_INNER_FOR_COND]]
3882 // CHECK7: omp.inner.for.end:
3883 // CHECK7-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
3884 // CHECK7: omp.loop.exit:
3885 // CHECK7-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB1]], i32 [[TMP2]])
3886 // CHECK7-NEXT: ret void
3889 // CHECK7-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l45.omp_outlined.omp_outlined
3890 // CHECK7-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], i32 noundef [[DOTPREVIOUS_LB_:%.*]], i32 noundef [[DOTPREVIOUS_UB_:%.*]], ptr noundef [[THIS:%.*]]) #[[ATTR1]] {
3891 // CHECK7-NEXT: entry:
3892 // CHECK7-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 4
3893 // CHECK7-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 4
3894 // CHECK7-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i32, align 4
3895 // CHECK7-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i32, align 4
3896 // CHECK7-NEXT: [[THIS_ADDR:%.*]] = alloca ptr, align 4
3897 // CHECK7-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
3898 // CHECK7-NEXT: [[TMP:%.*]] = alloca i32, align 4
3899 // CHECK7-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4
3900 // CHECK7-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4
3901 // CHECK7-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
3902 // CHECK7-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
3903 // CHECK7-NEXT: [[I:%.*]] = alloca i32, align 4
3904 // CHECK7-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 4
3905 // CHECK7-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 4
3906 // CHECK7-NEXT: store i32 [[DOTPREVIOUS_LB_]], ptr [[DOTPREVIOUS_LB__ADDR]], align 4
3907 // CHECK7-NEXT: store i32 [[DOTPREVIOUS_UB_]], ptr [[DOTPREVIOUS_UB__ADDR]], align 4
3908 // CHECK7-NEXT: store ptr [[THIS]], ptr [[THIS_ADDR]], align 4
3909 // CHECK7-NEXT: [[TMP0:%.*]] = load ptr, ptr [[THIS_ADDR]], align 4
3910 // CHECK7-NEXT: store i32 0, ptr [[DOTOMP_LB]], align 4
3911 // CHECK7-NEXT: store i32 122, ptr [[DOTOMP_UB]], align 4
3912 // CHECK7-NEXT: [[TMP1:%.*]] = load i32, ptr [[DOTPREVIOUS_LB__ADDR]], align 4
3913 // CHECK7-NEXT: [[TMP2:%.*]] = load i32, ptr [[DOTPREVIOUS_UB__ADDR]], align 4
3914 // CHECK7-NEXT: store i32 [[TMP1]], ptr [[DOTOMP_LB]], align 4
3915 // CHECK7-NEXT: store i32 [[TMP2]], ptr [[DOTOMP_UB]], align 4
3916 // CHECK7-NEXT: store i32 1, ptr [[DOTOMP_STRIDE]], align 4
3917 // CHECK7-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
3918 // CHECK7-NEXT: [[TMP3:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
3919 // CHECK7-NEXT: [[TMP4:%.*]] = load i32, ptr [[TMP3]], align 4
3920 // CHECK7-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB2]], i32 [[TMP4]], i32 33, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 61)
3921 // CHECK7-NEXT: br label [[OMP_DISPATCH_COND:%.*]]
3922 // CHECK7: omp.dispatch.cond:
3923 // CHECK7-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
3924 // CHECK7-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTPREVIOUS_UB__ADDR]], align 4
3925 // CHECK7-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP5]], [[TMP6]]
3926 // CHECK7-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
3927 // CHECK7: cond.true:
3928 // CHECK7-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTPREVIOUS_UB__ADDR]], align 4
3929 // CHECK7-NEXT: br label [[COND_END:%.*]]
3930 // CHECK7: cond.false:
3931 // CHECK7-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
3932 // CHECK7-NEXT: br label [[COND_END]]
3933 // CHECK7: cond.end:
3934 // CHECK7-NEXT: [[COND:%.*]] = phi i32 [ [[TMP7]], [[COND_TRUE]] ], [ [[TMP8]], [[COND_FALSE]] ]
3935 // CHECK7-NEXT: store i32 [[COND]], ptr [[DOTOMP_UB]], align 4
3936 // CHECK7-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4
3937 // CHECK7-NEXT: store i32 [[TMP9]], ptr [[DOTOMP_IV]], align 4
3938 // CHECK7-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
3939 // CHECK7-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
3940 // CHECK7-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP10]], [[TMP11]]
3941 // CHECK7-NEXT: br i1 [[CMP1]], label [[OMP_DISPATCH_BODY:%.*]], label [[OMP_DISPATCH_END:%.*]]
3942 // CHECK7: omp.dispatch.body:
3943 // CHECK7-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
3944 // CHECK7: omp.inner.for.cond:
3945 // CHECK7-NEXT: [[TMP12:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
3946 // CHECK7-NEXT: [[TMP13:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
3947 // CHECK7-NEXT: [[CMP2:%.*]] = icmp sle i32 [[TMP12]], [[TMP13]]
3948 // CHECK7-NEXT: br i1 [[CMP2]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
3949 // CHECK7: omp.inner.for.body:
3950 // CHECK7-NEXT: [[TMP14:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
3951 // CHECK7-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP14]], 1
3952 // CHECK7-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
3953 // CHECK7-NEXT: store i32 [[ADD]], ptr [[I]], align 4
3954 // CHECK7-NEXT: [[A:%.*]] = getelementptr inbounds nuw [[STRUCT_SS:%.*]], ptr [[TMP0]], i32 0, i32 0
3955 // CHECK7-NEXT: [[TMP15:%.*]] = load i32, ptr [[I]], align 4
3956 // CHECK7-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds [123 x i32], ptr [[A]], i32 0, i32 [[TMP15]]
3957 // CHECK7-NEXT: store i32 0, ptr [[ARRAYIDX]], align 4
3958 // CHECK7-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
3959 // CHECK7: omp.body.continue:
3960 // CHECK7-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
3961 // CHECK7: omp.inner.for.inc:
3962 // CHECK7-NEXT: [[TMP16:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
3963 // CHECK7-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP16]], 1
3964 // CHECK7-NEXT: store i32 [[ADD3]], ptr [[DOTOMP_IV]], align 4
3965 // CHECK7-NEXT: br label [[OMP_INNER_FOR_COND]]
3966 // CHECK7: omp.inner.for.end:
3967 // CHECK7-NEXT: br label [[OMP_DISPATCH_INC:%.*]]
3968 // CHECK7: omp.dispatch.inc:
3969 // CHECK7-NEXT: [[TMP17:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4
3970 // CHECK7-NEXT: [[TMP18:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4
3971 // CHECK7-NEXT: [[ADD4:%.*]] = add nsw i32 [[TMP17]], [[TMP18]]
3972 // CHECK7-NEXT: store i32 [[ADD4]], ptr [[DOTOMP_LB]], align 4
3973 // CHECK7-NEXT: [[TMP19:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
3974 // CHECK7-NEXT: [[TMP20:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4
3975 // CHECK7-NEXT: [[ADD5:%.*]] = add nsw i32 [[TMP19]], [[TMP20]]
3976 // CHECK7-NEXT: store i32 [[ADD5]], ptr [[DOTOMP_UB]], align 4
3977 // CHECK7-NEXT: br label [[OMP_DISPATCH_COND]]
3978 // CHECK7: omp.dispatch.end:
3979 // CHECK7-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB2]], i32 [[TMP4]])
3980 // CHECK7-NEXT: ret void
3983 // CHECK7-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l51
3984 // CHECK7-SAME: (ptr noundef [[THIS:%.*]]) #[[ATTR1]] {
3985 // CHECK7-NEXT: entry:
3986 // CHECK7-NEXT: [[THIS_ADDR:%.*]] = alloca ptr, align 4
3987 // CHECK7-NEXT: store ptr [[THIS]], ptr [[THIS_ADDR]], align 4
3988 // CHECK7-NEXT: [[TMP0:%.*]] = load ptr, ptr [[THIS_ADDR]], align 4
3989 // CHECK7-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_teams(ptr @[[GLOB3]], i32 1, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l51.omp_outlined, ptr [[TMP0]])
3990 // CHECK7-NEXT: ret void
3993 // CHECK7-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l51.omp_outlined
3994 // CHECK7-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], ptr noundef [[THIS:%.*]]) #[[ATTR1]] {
3995 // CHECK7-NEXT: entry:
3996 // CHECK7-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 4
3997 // CHECK7-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 4
3998 // CHECK7-NEXT: [[THIS_ADDR:%.*]] = alloca ptr, align 4
3999 // CHECK7-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
4000 // CHECK7-NEXT: [[TMP:%.*]] = alloca i32, align 4
4001 // CHECK7-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4
4002 // CHECK7-NEXT: [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4
4003 // CHECK7-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
4004 // CHECK7-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
4005 // CHECK7-NEXT: [[I:%.*]] = alloca i32, align 4
4006 // CHECK7-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 4
4007 // CHECK7-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 4
4008 // CHECK7-NEXT: store ptr [[THIS]], ptr [[THIS_ADDR]], align 4
4009 // CHECK7-NEXT: [[TMP0:%.*]] = load ptr, ptr [[THIS_ADDR]], align 4
4010 // CHECK7-NEXT: store i32 0, ptr [[DOTOMP_COMB_LB]], align 4
4011 // CHECK7-NEXT: store i32 122, ptr [[DOTOMP_COMB_UB]], align 4
4012 // CHECK7-NEXT: store i32 1, ptr [[DOTOMP_STRIDE]], align 4
4013 // CHECK7-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
4014 // CHECK7-NEXT: [[TMP1:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
4015 // CHECK7-NEXT: [[TMP2:%.*]] = load i32, ptr [[TMP1]], align 4
4016 // CHECK7-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB1]], i32 [[TMP2]], i32 92, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
4017 // CHECK7-NEXT: [[TMP3:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
4018 // CHECK7-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP3]], 122
4019 // CHECK7-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
4020 // CHECK7: cond.true:
4021 // CHECK7-NEXT: br label [[COND_END:%.*]]
4022 // CHECK7: cond.false:
4023 // CHECK7-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
4024 // CHECK7-NEXT: br label [[COND_END]]
4025 // CHECK7: cond.end:
4026 // CHECK7-NEXT: [[COND:%.*]] = phi i32 [ 122, [[COND_TRUE]] ], [ [[TMP4]], [[COND_FALSE]] ]
4027 // CHECK7-NEXT: store i32 [[COND]], ptr [[DOTOMP_COMB_UB]], align 4
4028 // CHECK7-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
4029 // CHECK7-NEXT: store i32 [[TMP5]], ptr [[DOTOMP_IV]], align 4
4030 // CHECK7-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
4031 // CHECK7: omp.inner.for.cond:
4032 // CHECK7-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
4033 // CHECK7-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
4034 // CHECK7-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP6]], [[TMP7]]
4035 // CHECK7-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
4036 // CHECK7: omp.inner.for.body:
4037 // CHECK7-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
4038 // CHECK7-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
4039 // CHECK7-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_call(ptr @[[GLOB3]], i32 3, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l51.omp_outlined.omp_outlined, i32 [[TMP8]], i32 [[TMP9]], ptr [[TMP0]])
4040 // CHECK7-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
4041 // CHECK7: omp.inner.for.inc:
4042 // CHECK7-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
4043 // CHECK7-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4
4044 // CHECK7-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP10]], [[TMP11]]
4045 // CHECK7-NEXT: store i32 [[ADD]], ptr [[DOTOMP_IV]], align 4
4046 // CHECK7-NEXT: br label [[OMP_INNER_FOR_COND]]
4047 // CHECK7: omp.inner.for.end:
4048 // CHECK7-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
4049 // CHECK7: omp.loop.exit:
4050 // CHECK7-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB1]], i32 [[TMP2]])
4051 // CHECK7-NEXT: ret void
4054 // CHECK7-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l51.omp_outlined.omp_outlined
4055 // CHECK7-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], i32 noundef [[DOTPREVIOUS_LB_:%.*]], i32 noundef [[DOTPREVIOUS_UB_:%.*]], ptr noundef [[THIS:%.*]]) #[[ATTR1]] {
4056 // CHECK7-NEXT: entry:
4057 // CHECK7-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 4
4058 // CHECK7-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 4
4059 // CHECK7-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i32, align 4
4060 // CHECK7-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i32, align 4
4061 // CHECK7-NEXT: [[THIS_ADDR:%.*]] = alloca ptr, align 4
4062 // CHECK7-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
4063 // CHECK7-NEXT: [[TMP:%.*]] = alloca i32, align 4
4064 // CHECK7-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4
4065 // CHECK7-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4
4066 // CHECK7-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
4067 // CHECK7-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
4068 // CHECK7-NEXT: [[I:%.*]] = alloca i32, align 4
4069 // CHECK7-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 4
4070 // CHECK7-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 4
4071 // CHECK7-NEXT: store i32 [[DOTPREVIOUS_LB_]], ptr [[DOTPREVIOUS_LB__ADDR]], align 4
4072 // CHECK7-NEXT: store i32 [[DOTPREVIOUS_UB_]], ptr [[DOTPREVIOUS_UB__ADDR]], align 4
4073 // CHECK7-NEXT: store ptr [[THIS]], ptr [[THIS_ADDR]], align 4
4074 // CHECK7-NEXT: [[TMP0:%.*]] = load ptr, ptr [[THIS_ADDR]], align 4
4075 // CHECK7-NEXT: store i32 0, ptr [[DOTOMP_LB]], align 4
4076 // CHECK7-NEXT: store i32 122, ptr [[DOTOMP_UB]], align 4
4077 // CHECK7-NEXT: [[TMP1:%.*]] = load i32, ptr [[DOTPREVIOUS_LB__ADDR]], align 4
4078 // CHECK7-NEXT: [[TMP2:%.*]] = load i32, ptr [[DOTPREVIOUS_UB__ADDR]], align 4
4079 // CHECK7-NEXT: store i32 [[TMP1]], ptr [[DOTOMP_LB]], align 4
4080 // CHECK7-NEXT: store i32 [[TMP2]], ptr [[DOTOMP_UB]], align 4
4081 // CHECK7-NEXT: store i32 1, ptr [[DOTOMP_STRIDE]], align 4
4082 // CHECK7-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
4083 // CHECK7-NEXT: [[TMP3:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4
4084 // CHECK7-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
4085 // CHECK7-NEXT: [[TMP5:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
4086 // CHECK7-NEXT: [[TMP6:%.*]] = load i32, ptr [[TMP5]], align 4
4087 // CHECK7-NEXT: call void @__kmpc_dispatch_init_4(ptr @[[GLOB3]], i32 [[TMP6]], i32 35, i32 [[TMP3]], i32 [[TMP4]], i32 1, i32 1)
4088 // CHECK7-NEXT: br label [[OMP_DISPATCH_COND:%.*]]
4089 // CHECK7: omp.dispatch.cond:
4090 // CHECK7-NEXT: [[TMP7:%.*]] = call i32 @__kmpc_dispatch_next_4(ptr @[[GLOB3]], i32 [[TMP6]], ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]])
4091 // CHECK7-NEXT: [[TOBOOL:%.*]] = icmp ne i32 [[TMP7]], 0
4092 // CHECK7-NEXT: br i1 [[TOBOOL]], label [[OMP_DISPATCH_BODY:%.*]], label [[OMP_DISPATCH_END:%.*]]
4093 // CHECK7: omp.dispatch.body:
4094 // CHECK7-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4
4095 // CHECK7-NEXT: store i32 [[TMP8]], ptr [[DOTOMP_IV]], align 4
4096 // CHECK7-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
4097 // CHECK7: omp.inner.for.cond:
4098 // CHECK7-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP11:![0-9]+]]
4099 // CHECK7-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP11]]
4100 // CHECK7-NEXT: [[CMP:%.*]] = icmp sle i32 [[TMP9]], [[TMP10]]
4101 // CHECK7-NEXT: br i1 [[CMP]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
4102 // CHECK7: omp.inner.for.body:
4103 // CHECK7-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP11]]
4104 // CHECK7-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP11]], 1
4105 // CHECK7-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
4106 // CHECK7-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP11]]
4107 // CHECK7-NEXT: [[A:%.*]] = getelementptr inbounds nuw [[STRUCT_SS:%.*]], ptr [[TMP0]], i32 0, i32 0
4108 // CHECK7-NEXT: [[TMP12:%.*]] = load i32, ptr [[I]], align 4, !llvm.access.group [[ACC_GRP11]]
4109 // CHECK7-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds [123 x i32], ptr [[A]], i32 0, i32 [[TMP12]]
4110 // CHECK7-NEXT: store i32 0, ptr [[ARRAYIDX]], align 4, !llvm.access.group [[ACC_GRP11]]
4111 // CHECK7-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
4112 // CHECK7: omp.body.continue:
4113 // CHECK7-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
4114 // CHECK7: omp.inner.for.inc:
4115 // CHECK7-NEXT: [[TMP13:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP11]]
4116 // CHECK7-NEXT: [[ADD1:%.*]] = add nsw i32 [[TMP13]], 1
4117 // CHECK7-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP11]]
4118 // CHECK7-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP12:![0-9]+]]
4119 // CHECK7: omp.inner.for.end:
4120 // CHECK7-NEXT: br label [[OMP_DISPATCH_INC:%.*]]
4121 // CHECK7: omp.dispatch.inc:
4122 // CHECK7-NEXT: br label [[OMP_DISPATCH_COND]]
4123 // CHECK7: omp.dispatch.end:
4124 // CHECK7-NEXT: call void @__kmpc_dispatch_deinit(ptr @[[GLOB3]], i32 [[TMP6]])
4125 // CHECK7-NEXT: ret void
4128 // CHECK7-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l57
4129 // CHECK7-SAME: (ptr noundef [[THIS:%.*]]) #[[ATTR1]] {
4130 // CHECK7-NEXT: entry:
4131 // CHECK7-NEXT: [[THIS_ADDR:%.*]] = alloca ptr, align 4
4132 // CHECK7-NEXT: store ptr [[THIS]], ptr [[THIS_ADDR]], align 4
4133 // CHECK7-NEXT: [[TMP0:%.*]] = load ptr, ptr [[THIS_ADDR]], align 4
4134 // CHECK7-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_teams(ptr @[[GLOB3]], i32 1, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l57.omp_outlined, ptr [[TMP0]])
4135 // CHECK7-NEXT: ret void
4138 // CHECK7-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l57.omp_outlined
4139 // CHECK7-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], ptr noundef [[THIS:%.*]]) #[[ATTR1]] {
4140 // CHECK7-NEXT: entry:
4141 // CHECK7-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 4
4142 // CHECK7-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 4
4143 // CHECK7-NEXT: [[THIS_ADDR:%.*]] = alloca ptr, align 4
4144 // CHECK7-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
4145 // CHECK7-NEXT: [[TMP:%.*]] = alloca i32, align 4
4146 // CHECK7-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4
4147 // CHECK7-NEXT: [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4
4148 // CHECK7-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
4149 // CHECK7-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
4150 // CHECK7-NEXT: [[I:%.*]] = alloca i32, align 4
4151 // CHECK7-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 4
4152 // CHECK7-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 4
4153 // CHECK7-NEXT: store ptr [[THIS]], ptr [[THIS_ADDR]], align 4
4154 // CHECK7-NEXT: [[TMP0:%.*]] = load ptr, ptr [[THIS_ADDR]], align 4
4155 // CHECK7-NEXT: store i32 0, ptr [[DOTOMP_COMB_LB]], align 4
4156 // CHECK7-NEXT: store i32 122, ptr [[DOTOMP_COMB_UB]], align 4
4157 // CHECK7-NEXT: store i32 1, ptr [[DOTOMP_STRIDE]], align 4
4158 // CHECK7-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
4159 // CHECK7-NEXT: [[TMP1:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
4160 // CHECK7-NEXT: [[TMP2:%.*]] = load i32, ptr [[TMP1]], align 4
4161 // CHECK7-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB1]], i32 [[TMP2]], i32 92, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
4162 // CHECK7-NEXT: [[TMP3:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
4163 // CHECK7-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP3]], 122
4164 // CHECK7-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
4165 // CHECK7: cond.true:
4166 // CHECK7-NEXT: br label [[COND_END:%.*]]
4167 // CHECK7: cond.false:
4168 // CHECK7-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
4169 // CHECK7-NEXT: br label [[COND_END]]
4170 // CHECK7: cond.end:
4171 // CHECK7-NEXT: [[COND:%.*]] = phi i32 [ 122, [[COND_TRUE]] ], [ [[TMP4]], [[COND_FALSE]] ]
4172 // CHECK7-NEXT: store i32 [[COND]], ptr [[DOTOMP_COMB_UB]], align 4
4173 // CHECK7-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
4174 // CHECK7-NEXT: store i32 [[TMP5]], ptr [[DOTOMP_IV]], align 4
4175 // CHECK7-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
4176 // CHECK7: omp.inner.for.cond:
4177 // CHECK7-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
4178 // CHECK7-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
4179 // CHECK7-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP6]], [[TMP7]]
4180 // CHECK7-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
4181 // CHECK7: omp.inner.for.body:
4182 // CHECK7-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
4183 // CHECK7-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
4184 // CHECK7-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_call(ptr @[[GLOB3]], i32 3, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l57.omp_outlined.omp_outlined, i32 [[TMP8]], i32 [[TMP9]], ptr [[TMP0]])
4185 // CHECK7-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
4186 // CHECK7: omp.inner.for.inc:
4187 // CHECK7-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
4188 // CHECK7-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4
4189 // CHECK7-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP10]], [[TMP11]]
4190 // CHECK7-NEXT: store i32 [[ADD]], ptr [[DOTOMP_IV]], align 4
4191 // CHECK7-NEXT: br label [[OMP_INNER_FOR_COND]]
4192 // CHECK7: omp.inner.for.end:
4193 // CHECK7-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
4194 // CHECK7: omp.loop.exit:
4195 // CHECK7-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB1]], i32 [[TMP2]])
4196 // CHECK7-NEXT: ret void
4199 // CHECK7-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l57.omp_outlined.omp_outlined
4200 // CHECK7-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], i32 noundef [[DOTPREVIOUS_LB_:%.*]], i32 noundef [[DOTPREVIOUS_UB_:%.*]], ptr noundef [[THIS:%.*]]) #[[ATTR1]] {
4201 // CHECK7-NEXT: entry:
4202 // CHECK7-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 4
4203 // CHECK7-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 4
4204 // CHECK7-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i32, align 4
4205 // CHECK7-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i32, align 4
4206 // CHECK7-NEXT: [[THIS_ADDR:%.*]] = alloca ptr, align 4
4207 // CHECK7-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
4208 // CHECK7-NEXT: [[TMP:%.*]] = alloca i32, align 4
4209 // CHECK7-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4
4210 // CHECK7-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4
4211 // CHECK7-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
4212 // CHECK7-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
4213 // CHECK7-NEXT: [[I:%.*]] = alloca i32, align 4
4214 // CHECK7-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 4
4215 // CHECK7-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 4
4216 // CHECK7-NEXT: store i32 [[DOTPREVIOUS_LB_]], ptr [[DOTPREVIOUS_LB__ADDR]], align 4
4217 // CHECK7-NEXT: store i32 [[DOTPREVIOUS_UB_]], ptr [[DOTPREVIOUS_UB__ADDR]], align 4
4218 // CHECK7-NEXT: store ptr [[THIS]], ptr [[THIS_ADDR]], align 4
4219 // CHECK7-NEXT: [[TMP0:%.*]] = load ptr, ptr [[THIS_ADDR]], align 4
4220 // CHECK7-NEXT: store i32 0, ptr [[DOTOMP_LB]], align 4
4221 // CHECK7-NEXT: store i32 122, ptr [[DOTOMP_UB]], align 4
4222 // CHECK7-NEXT: [[TMP1:%.*]] = load i32, ptr [[DOTPREVIOUS_LB__ADDR]], align 4
4223 // CHECK7-NEXT: [[TMP2:%.*]] = load i32, ptr [[DOTPREVIOUS_UB__ADDR]], align 4
4224 // CHECK7-NEXT: store i32 [[TMP1]], ptr [[DOTOMP_LB]], align 4
4225 // CHECK7-NEXT: store i32 [[TMP2]], ptr [[DOTOMP_UB]], align 4
4226 // CHECK7-NEXT: store i32 1, ptr [[DOTOMP_STRIDE]], align 4
4227 // CHECK7-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
4228 // CHECK7-NEXT: [[TMP3:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4
4229 // CHECK7-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
4230 // CHECK7-NEXT: [[TMP5:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
4231 // CHECK7-NEXT: [[TMP6:%.*]] = load i32, ptr [[TMP5]], align 4
4232 // CHECK7-NEXT: call void @__kmpc_dispatch_init_4(ptr @[[GLOB3]], i32 [[TMP6]], i32 35, i32 [[TMP3]], i32 [[TMP4]], i32 1, i32 61)
4233 // CHECK7-NEXT: br label [[OMP_DISPATCH_COND:%.*]]
4234 // CHECK7: omp.dispatch.cond:
4235 // CHECK7-NEXT: [[TMP7:%.*]] = call i32 @__kmpc_dispatch_next_4(ptr @[[GLOB3]], i32 [[TMP6]], ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]])
4236 // CHECK7-NEXT: [[TOBOOL:%.*]] = icmp ne i32 [[TMP7]], 0
4237 // CHECK7-NEXT: br i1 [[TOBOOL]], label [[OMP_DISPATCH_BODY:%.*]], label [[OMP_DISPATCH_END:%.*]]
4238 // CHECK7: omp.dispatch.body:
4239 // CHECK7-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4
4240 // CHECK7-NEXT: store i32 [[TMP8]], ptr [[DOTOMP_IV]], align 4
4241 // CHECK7-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
4242 // CHECK7: omp.inner.for.cond:
4243 // CHECK7-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP14:![0-9]+]]
4244 // CHECK7-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP14]]
4245 // CHECK7-NEXT: [[CMP:%.*]] = icmp sle i32 [[TMP9]], [[TMP10]]
4246 // CHECK7-NEXT: br i1 [[CMP]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
4247 // CHECK7: omp.inner.for.body:
4248 // CHECK7-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP14]]
4249 // CHECK7-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP11]], 1
4250 // CHECK7-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
4251 // CHECK7-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP14]]
4252 // CHECK7-NEXT: [[A:%.*]] = getelementptr inbounds nuw [[STRUCT_SS:%.*]], ptr [[TMP0]], i32 0, i32 0
4253 // CHECK7-NEXT: [[TMP12:%.*]] = load i32, ptr [[I]], align 4, !llvm.access.group [[ACC_GRP14]]
4254 // CHECK7-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds [123 x i32], ptr [[A]], i32 0, i32 [[TMP12]]
4255 // CHECK7-NEXT: store i32 0, ptr [[ARRAYIDX]], align 4, !llvm.access.group [[ACC_GRP14]]
4256 // CHECK7-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
4257 // CHECK7: omp.body.continue:
4258 // CHECK7-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
4259 // CHECK7: omp.inner.for.inc:
4260 // CHECK7-NEXT: [[TMP13:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP14]]
4261 // CHECK7-NEXT: [[ADD1:%.*]] = add nsw i32 [[TMP13]], 1
4262 // CHECK7-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP14]]
4263 // CHECK7-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP15:![0-9]+]]
4264 // CHECK7: omp.inner.for.end:
4265 // CHECK7-NEXT: br label [[OMP_DISPATCH_INC:%.*]]
4266 // CHECK7: omp.dispatch.inc:
4267 // CHECK7-NEXT: br label [[OMP_DISPATCH_COND]]
4268 // CHECK7: omp.dispatch.end:
4269 // CHECK7-NEXT: call void @__kmpc_dispatch_deinit(ptr @[[GLOB3]], i32 [[TMP6]])
4270 // CHECK7-NEXT: ret void
4273 // CHECK13-LABEL: define {{[^@]+}}@main
4274 // CHECK13-SAME: (i32 noundef signext [[ARGC:%.*]], ptr noundef [[ARGV:%.*]]) #[[ATTR0:[0-9]+]] {
4275 // CHECK13-NEXT: entry:
4276 // CHECK13-NEXT: [[RETVAL:%.*]] = alloca i32, align 4
4277 // CHECK13-NEXT: [[ARGC_ADDR:%.*]] = alloca i32, align 4
4278 // CHECK13-NEXT: [[ARGV_ADDR:%.*]] = alloca ptr, align 8
4279 // CHECK13-NEXT: [[N:%.*]] = alloca i32, align 4
4280 // CHECK13-NEXT: [[SAVED_STACK:%.*]] = alloca ptr, align 8
4281 // CHECK13-NEXT: [[__VLA_EXPR0:%.*]] = alloca i64, align 8
4282 // CHECK13-NEXT: [[M:%.*]] = alloca i32, align 4
4283 // CHECK13-NEXT: [[N_CASTED:%.*]] = alloca i64, align 8
4284 // CHECK13-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [3 x ptr], align 8
4285 // CHECK13-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [3 x ptr], align 8
4286 // CHECK13-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [3 x ptr], align 8
4287 // CHECK13-NEXT: [[DOTOFFLOAD_SIZES:%.*]] = alloca [3 x i64], align 8
4288 // CHECK13-NEXT: [[TMP:%.*]] = alloca i32, align 4
4289 // CHECK13-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4
4290 // CHECK13-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4
4291 // CHECK13-NEXT: [[KERNEL_ARGS:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS:%.*]], align 8
4292 // CHECK13-NEXT: [[N_CASTED3:%.*]] = alloca i64, align 8
4293 // CHECK13-NEXT: [[DOTOFFLOAD_BASEPTRS4:%.*]] = alloca [3 x ptr], align 8
4294 // CHECK13-NEXT: [[DOTOFFLOAD_PTRS5:%.*]] = alloca [3 x ptr], align 8
4295 // CHECK13-NEXT: [[DOTOFFLOAD_MAPPERS6:%.*]] = alloca [3 x ptr], align 8
4296 // CHECK13-NEXT: [[DOTOFFLOAD_SIZES7:%.*]] = alloca [3 x i64], align 8
4297 // CHECK13-NEXT: [[_TMP8:%.*]] = alloca i32, align 4
4298 // CHECK13-NEXT: [[DOTCAPTURE_EXPR_9:%.*]] = alloca i32, align 4
4299 // CHECK13-NEXT: [[DOTCAPTURE_EXPR_10:%.*]] = alloca i32, align 4
4300 // CHECK13-NEXT: [[KERNEL_ARGS15:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS]], align 8
4301 // CHECK13-NEXT: [[M_CASTED:%.*]] = alloca i64, align 8
4302 // CHECK13-NEXT: [[N_CASTED18:%.*]] = alloca i64, align 8
4303 // CHECK13-NEXT: [[DOTOFFLOAD_BASEPTRS19:%.*]] = alloca [4 x ptr], align 8
4304 // CHECK13-NEXT: [[DOTOFFLOAD_PTRS20:%.*]] = alloca [4 x ptr], align 8
4305 // CHECK13-NEXT: [[DOTOFFLOAD_MAPPERS21:%.*]] = alloca [4 x ptr], align 8
4306 // CHECK13-NEXT: [[DOTOFFLOAD_SIZES22:%.*]] = alloca [4 x i64], align 8
4307 // CHECK13-NEXT: [[_TMP23:%.*]] = alloca i32, align 4
4308 // CHECK13-NEXT: [[DOTCAPTURE_EXPR_24:%.*]] = alloca i32, align 4
4309 // CHECK13-NEXT: [[DOTCAPTURE_EXPR_25:%.*]] = alloca i32, align 4
4310 // CHECK13-NEXT: [[KERNEL_ARGS30:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS]], align 8
4311 // CHECK13-NEXT: [[N_CASTED33:%.*]] = alloca i64, align 8
4312 // CHECK13-NEXT: [[DOTOFFLOAD_BASEPTRS34:%.*]] = alloca [3 x ptr], align 8
4313 // CHECK13-NEXT: [[DOTOFFLOAD_PTRS35:%.*]] = alloca [3 x ptr], align 8
4314 // CHECK13-NEXT: [[DOTOFFLOAD_MAPPERS36:%.*]] = alloca [3 x ptr], align 8
4315 // CHECK13-NEXT: [[DOTOFFLOAD_SIZES37:%.*]] = alloca [3 x i64], align 8
4316 // CHECK13-NEXT: [[_TMP38:%.*]] = alloca i32, align 4
4317 // CHECK13-NEXT: [[DOTCAPTURE_EXPR_39:%.*]] = alloca i32, align 4
4318 // CHECK13-NEXT: [[DOTCAPTURE_EXPR_40:%.*]] = alloca i32, align 4
4319 // CHECK13-NEXT: [[KERNEL_ARGS45:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS]], align 8
4320 // CHECK13-NEXT: [[M_CASTED48:%.*]] = alloca i64, align 8
4321 // CHECK13-NEXT: [[N_CASTED49:%.*]] = alloca i64, align 8
4322 // CHECK13-NEXT: [[DOTOFFLOAD_BASEPTRS50:%.*]] = alloca [4 x ptr], align 8
4323 // CHECK13-NEXT: [[DOTOFFLOAD_PTRS51:%.*]] = alloca [4 x ptr], align 8
4324 // CHECK13-NEXT: [[DOTOFFLOAD_MAPPERS52:%.*]] = alloca [4 x ptr], align 8
4325 // CHECK13-NEXT: [[DOTOFFLOAD_SIZES53:%.*]] = alloca [4 x i64], align 8
4326 // CHECK13-NEXT: [[_TMP54:%.*]] = alloca i32, align 4
4327 // CHECK13-NEXT: [[DOTCAPTURE_EXPR_55:%.*]] = alloca i32, align 4
4328 // CHECK13-NEXT: [[DOTCAPTURE_EXPR_56:%.*]] = alloca i32, align 4
4329 // CHECK13-NEXT: [[KERNEL_ARGS61:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS]], align 8
4330 // CHECK13-NEXT: store i32 0, ptr [[RETVAL]], align 4
4331 // CHECK13-NEXT: store i32 [[ARGC]], ptr [[ARGC_ADDR]], align 4
4332 // CHECK13-NEXT: store ptr [[ARGV]], ptr [[ARGV_ADDR]], align 8
4333 // CHECK13-NEXT: store i32 100, ptr [[N]], align 4
4334 // CHECK13-NEXT: [[TMP0:%.*]] = load i32, ptr [[N]], align 4
4335 // CHECK13-NEXT: [[TMP1:%.*]] = zext i32 [[TMP0]] to i64
4336 // CHECK13-NEXT: [[TMP2:%.*]] = call ptr @llvm.stacksave.p0()
4337 // CHECK13-NEXT: store ptr [[TMP2]], ptr [[SAVED_STACK]], align 8
4338 // CHECK13-NEXT: [[VLA:%.*]] = alloca i32, i64 [[TMP1]], align 4
4339 // CHECK13-NEXT: store i64 [[TMP1]], ptr [[__VLA_EXPR0]], align 8
4340 // CHECK13-NEXT: store i32 10, ptr [[M]], align 4
4341 // CHECK13-NEXT: [[TMP3:%.*]] = load i32, ptr [[N]], align 4
4342 // CHECK13-NEXT: store i32 [[TMP3]], ptr [[N_CASTED]], align 4
4343 // CHECK13-NEXT: [[TMP4:%.*]] = load i64, ptr [[N_CASTED]], align 8
4344 // CHECK13-NEXT: [[TMP5:%.*]] = mul nuw i64 [[TMP1]], 4
4345 // CHECK13-NEXT: call void @llvm.memcpy.p0.p0.i64(ptr align 8 [[DOTOFFLOAD_SIZES]], ptr align 8 @.offload_sizes, i64 24, i1 false)
4346 // CHECK13-NEXT: [[TMP6:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
4347 // CHECK13-NEXT: store i64 [[TMP4]], ptr [[TMP6]], align 8
4348 // CHECK13-NEXT: [[TMP7:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
4349 // CHECK13-NEXT: store i64 [[TMP4]], ptr [[TMP7]], align 8
4350 // CHECK13-NEXT: [[TMP8:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0
4351 // CHECK13-NEXT: store ptr null, ptr [[TMP8]], align 8
4352 // CHECK13-NEXT: [[TMP9:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1
4353 // CHECK13-NEXT: store i64 [[TMP1]], ptr [[TMP9]], align 8
4354 // CHECK13-NEXT: [[TMP10:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 1
4355 // CHECK13-NEXT: store i64 [[TMP1]], ptr [[TMP10]], align 8
4356 // CHECK13-NEXT: [[TMP11:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i64 0, i64 1
4357 // CHECK13-NEXT: store ptr null, ptr [[TMP11]], align 8
4358 // CHECK13-NEXT: [[TMP12:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
4359 // CHECK13-NEXT: store ptr [[VLA]], ptr [[TMP12]], align 8
4360 // CHECK13-NEXT: [[TMP13:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 2
4361 // CHECK13-NEXT: store ptr [[VLA]], ptr [[TMP13]], align 8
4362 // CHECK13-NEXT: [[TMP14:%.*]] = getelementptr inbounds [3 x i64], ptr [[DOTOFFLOAD_SIZES]], i32 0, i32 2
4363 // CHECK13-NEXT: store i64 [[TMP5]], ptr [[TMP14]], align 8
4364 // CHECK13-NEXT: [[TMP15:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i64 0, i64 2
4365 // CHECK13-NEXT: store ptr null, ptr [[TMP15]], align 8
4366 // CHECK13-NEXT: [[TMP16:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
4367 // CHECK13-NEXT: [[TMP17:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
4368 // CHECK13-NEXT: [[TMP18:%.*]] = getelementptr inbounds [3 x i64], ptr [[DOTOFFLOAD_SIZES]], i32 0, i32 0
4369 // CHECK13-NEXT: [[TMP19:%.*]] = load i32, ptr [[N]], align 4
4370 // CHECK13-NEXT: store i32 [[TMP19]], ptr [[DOTCAPTURE_EXPR_]], align 4
4371 // CHECK13-NEXT: [[TMP20:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4
4372 // CHECK13-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP20]], 0
4373 // CHECK13-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1
4374 // CHECK13-NEXT: [[SUB2:%.*]] = sub nsw i32 [[DIV]], 1
4375 // CHECK13-NEXT: store i32 [[SUB2]], ptr [[DOTCAPTURE_EXPR_1]], align 4
4376 // CHECK13-NEXT: [[TMP21:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4
4377 // CHECK13-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP21]], 1
4378 // CHECK13-NEXT: [[TMP22:%.*]] = zext i32 [[ADD]] to i64
4379 // CHECK13-NEXT: [[TMP23:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 0
4380 // CHECK13-NEXT: store i32 3, ptr [[TMP23]], align 4
4381 // CHECK13-NEXT: [[TMP24:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 1
4382 // CHECK13-NEXT: store i32 3, ptr [[TMP24]], align 4
4383 // CHECK13-NEXT: [[TMP25:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 2
4384 // CHECK13-NEXT: store ptr [[TMP16]], ptr [[TMP25]], align 8
4385 // CHECK13-NEXT: [[TMP26:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 3
4386 // CHECK13-NEXT: store ptr [[TMP17]], ptr [[TMP26]], align 8
4387 // CHECK13-NEXT: [[TMP27:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 4
4388 // CHECK13-NEXT: store ptr [[TMP18]], ptr [[TMP27]], align 8
4389 // CHECK13-NEXT: [[TMP28:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 5
4390 // CHECK13-NEXT: store ptr @.offload_maptypes, ptr [[TMP28]], align 8
4391 // CHECK13-NEXT: [[TMP29:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 6
4392 // CHECK13-NEXT: store ptr null, ptr [[TMP29]], align 8
4393 // CHECK13-NEXT: [[TMP30:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 7
4394 // CHECK13-NEXT: store ptr null, ptr [[TMP30]], align 8
4395 // CHECK13-NEXT: [[TMP31:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 8
4396 // CHECK13-NEXT: store i64 [[TMP22]], ptr [[TMP31]], align 8
4397 // CHECK13-NEXT: [[TMP32:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 9
4398 // CHECK13-NEXT: store i64 0, ptr [[TMP32]], align 8
4399 // CHECK13-NEXT: [[TMP33:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 10
4400 // CHECK13-NEXT: store [3 x i32] zeroinitializer, ptr [[TMP33]], align 4
4401 // CHECK13-NEXT: [[TMP34:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 11
4402 // CHECK13-NEXT: store [3 x i32] zeroinitializer, ptr [[TMP34]], align 4
4403 // CHECK13-NEXT: [[TMP35:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 12
4404 // CHECK13-NEXT: store i32 0, ptr [[TMP35]], align 4
4405 // CHECK13-NEXT: [[TMP36:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB3:[0-9]+]], i64 -1, i32 0, i32 0, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l148.region_id, ptr [[KERNEL_ARGS]])
4406 // CHECK13-NEXT: [[TMP37:%.*]] = icmp ne i32 [[TMP36]], 0
4407 // CHECK13-NEXT: br i1 [[TMP37]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
4408 // CHECK13: omp_offload.failed:
4409 // CHECK13-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l148(i64 [[TMP4]], i64 [[TMP1]], ptr [[VLA]]) #[[ATTR3:[0-9]+]]
4410 // CHECK13-NEXT: br label [[OMP_OFFLOAD_CONT]]
4411 // CHECK13: omp_offload.cont:
4412 // CHECK13-NEXT: [[TMP38:%.*]] = load i32, ptr [[N]], align 4
4413 // CHECK13-NEXT: store i32 [[TMP38]], ptr [[N_CASTED3]], align 4
4414 // CHECK13-NEXT: [[TMP39:%.*]] = load i64, ptr [[N_CASTED3]], align 8
4415 // CHECK13-NEXT: [[TMP40:%.*]] = mul nuw i64 [[TMP1]], 4
4416 // CHECK13-NEXT: call void @llvm.memcpy.p0.p0.i64(ptr align 8 [[DOTOFFLOAD_SIZES7]], ptr align 8 @.offload_sizes.1, i64 24, i1 false)
4417 // CHECK13-NEXT: [[TMP41:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_BASEPTRS4]], i32 0, i32 0
4418 // CHECK13-NEXT: store i64 [[TMP39]], ptr [[TMP41]], align 8
4419 // CHECK13-NEXT: [[TMP42:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_PTRS5]], i32 0, i32 0
4420 // CHECK13-NEXT: store i64 [[TMP39]], ptr [[TMP42]], align 8
4421 // CHECK13-NEXT: [[TMP43:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_MAPPERS6]], i64 0, i64 0
4422 // CHECK13-NEXT: store ptr null, ptr [[TMP43]], align 8
4423 // CHECK13-NEXT: [[TMP44:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_BASEPTRS4]], i32 0, i32 1
4424 // CHECK13-NEXT: store i64 [[TMP1]], ptr [[TMP44]], align 8
4425 // CHECK13-NEXT: [[TMP45:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_PTRS5]], i32 0, i32 1
4426 // CHECK13-NEXT: store i64 [[TMP1]], ptr [[TMP45]], align 8
4427 // CHECK13-NEXT: [[TMP46:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_MAPPERS6]], i64 0, i64 1
4428 // CHECK13-NEXT: store ptr null, ptr [[TMP46]], align 8
4429 // CHECK13-NEXT: [[TMP47:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_BASEPTRS4]], i32 0, i32 2
4430 // CHECK13-NEXT: store ptr [[VLA]], ptr [[TMP47]], align 8
4431 // CHECK13-NEXT: [[TMP48:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_PTRS5]], i32 0, i32 2
4432 // CHECK13-NEXT: store ptr [[VLA]], ptr [[TMP48]], align 8
4433 // CHECK13-NEXT: [[TMP49:%.*]] = getelementptr inbounds [3 x i64], ptr [[DOTOFFLOAD_SIZES7]], i32 0, i32 2
4434 // CHECK13-NEXT: store i64 [[TMP40]], ptr [[TMP49]], align 8
4435 // CHECK13-NEXT: [[TMP50:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_MAPPERS6]], i64 0, i64 2
4436 // CHECK13-NEXT: store ptr null, ptr [[TMP50]], align 8
4437 // CHECK13-NEXT: [[TMP51:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_BASEPTRS4]], i32 0, i32 0
4438 // CHECK13-NEXT: [[TMP52:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_PTRS5]], i32 0, i32 0
4439 // CHECK13-NEXT: [[TMP53:%.*]] = getelementptr inbounds [3 x i64], ptr [[DOTOFFLOAD_SIZES7]], i32 0, i32 0
4440 // CHECK13-NEXT: [[TMP54:%.*]] = load i32, ptr [[N]], align 4
4441 // CHECK13-NEXT: store i32 [[TMP54]], ptr [[DOTCAPTURE_EXPR_9]], align 4
4442 // CHECK13-NEXT: [[TMP55:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_9]], align 4
4443 // CHECK13-NEXT: [[SUB11:%.*]] = sub nsw i32 [[TMP55]], 0
4444 // CHECK13-NEXT: [[DIV12:%.*]] = sdiv i32 [[SUB11]], 1
4445 // CHECK13-NEXT: [[SUB13:%.*]] = sub nsw i32 [[DIV12]], 1
4446 // CHECK13-NEXT: store i32 [[SUB13]], ptr [[DOTCAPTURE_EXPR_10]], align 4
4447 // CHECK13-NEXT: [[TMP56:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_10]], align 4
4448 // CHECK13-NEXT: [[ADD14:%.*]] = add nsw i32 [[TMP56]], 1
4449 // CHECK13-NEXT: [[TMP57:%.*]] = zext i32 [[ADD14]] to i64
4450 // CHECK13-NEXT: [[TMP58:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS15]], i32 0, i32 0
4451 // CHECK13-NEXT: store i32 3, ptr [[TMP58]], align 4
4452 // CHECK13-NEXT: [[TMP59:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS15]], i32 0, i32 1
4453 // CHECK13-NEXT: store i32 3, ptr [[TMP59]], align 4
4454 // CHECK13-NEXT: [[TMP60:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS15]], i32 0, i32 2
4455 // CHECK13-NEXT: store ptr [[TMP51]], ptr [[TMP60]], align 8
4456 // CHECK13-NEXT: [[TMP61:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS15]], i32 0, i32 3
4457 // CHECK13-NEXT: store ptr [[TMP52]], ptr [[TMP61]], align 8
4458 // CHECK13-NEXT: [[TMP62:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS15]], i32 0, i32 4
4459 // CHECK13-NEXT: store ptr [[TMP53]], ptr [[TMP62]], align 8
4460 // CHECK13-NEXT: [[TMP63:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS15]], i32 0, i32 5
4461 // CHECK13-NEXT: store ptr @.offload_maptypes.2, ptr [[TMP63]], align 8
4462 // CHECK13-NEXT: [[TMP64:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS15]], i32 0, i32 6
4463 // CHECK13-NEXT: store ptr null, ptr [[TMP64]], align 8
4464 // CHECK13-NEXT: [[TMP65:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS15]], i32 0, i32 7
4465 // CHECK13-NEXT: store ptr null, ptr [[TMP65]], align 8
4466 // CHECK13-NEXT: [[TMP66:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS15]], i32 0, i32 8
4467 // CHECK13-NEXT: store i64 [[TMP57]], ptr [[TMP66]], align 8
4468 // CHECK13-NEXT: [[TMP67:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS15]], i32 0, i32 9
4469 // CHECK13-NEXT: store i64 0, ptr [[TMP67]], align 8
4470 // CHECK13-NEXT: [[TMP68:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS15]], i32 0, i32 10
4471 // CHECK13-NEXT: store [3 x i32] zeroinitializer, ptr [[TMP68]], align 4
4472 // CHECK13-NEXT: [[TMP69:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS15]], i32 0, i32 11
4473 // CHECK13-NEXT: store [3 x i32] zeroinitializer, ptr [[TMP69]], align 4
4474 // CHECK13-NEXT: [[TMP70:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS15]], i32 0, i32 12
4475 // CHECK13-NEXT: store i32 0, ptr [[TMP70]], align 4
4476 // CHECK13-NEXT: [[TMP71:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB3]], i64 -1, i32 0, i32 0, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l153.region_id, ptr [[KERNEL_ARGS15]])
4477 // CHECK13-NEXT: [[TMP72:%.*]] = icmp ne i32 [[TMP71]], 0
4478 // CHECK13-NEXT: br i1 [[TMP72]], label [[OMP_OFFLOAD_FAILED16:%.*]], label [[OMP_OFFLOAD_CONT17:%.*]]
4479 // CHECK13: omp_offload.failed16:
4480 // CHECK13-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l153(i64 [[TMP39]], i64 [[TMP1]], ptr [[VLA]]) #[[ATTR3]]
4481 // CHECK13-NEXT: br label [[OMP_OFFLOAD_CONT17]]
4482 // CHECK13: omp_offload.cont17:
4483 // CHECK13-NEXT: [[TMP73:%.*]] = load i32, ptr [[M]], align 4
4484 // CHECK13-NEXT: store i32 [[TMP73]], ptr [[M_CASTED]], align 4
4485 // CHECK13-NEXT: [[TMP74:%.*]] = load i64, ptr [[M_CASTED]], align 8
4486 // CHECK13-NEXT: [[TMP75:%.*]] = load i32, ptr [[N]], align 4
4487 // CHECK13-NEXT: store i32 [[TMP75]], ptr [[N_CASTED18]], align 4
4488 // CHECK13-NEXT: [[TMP76:%.*]] = load i64, ptr [[N_CASTED18]], align 8
4489 // CHECK13-NEXT: [[TMP77:%.*]] = mul nuw i64 [[TMP1]], 4
4490 // CHECK13-NEXT: call void @llvm.memcpy.p0.p0.i64(ptr align 8 [[DOTOFFLOAD_SIZES22]], ptr align 8 @.offload_sizes.3, i64 32, i1 false)
4491 // CHECK13-NEXT: [[TMP78:%.*]] = getelementptr inbounds [4 x ptr], ptr [[DOTOFFLOAD_BASEPTRS19]], i32 0, i32 0
4492 // CHECK13-NEXT: store i64 [[TMP74]], ptr [[TMP78]], align 8
4493 // CHECK13-NEXT: [[TMP79:%.*]] = getelementptr inbounds [4 x ptr], ptr [[DOTOFFLOAD_PTRS20]], i32 0, i32 0
4494 // CHECK13-NEXT: store i64 [[TMP74]], ptr [[TMP79]], align 8
4495 // CHECK13-NEXT: [[TMP80:%.*]] = getelementptr inbounds [4 x ptr], ptr [[DOTOFFLOAD_MAPPERS21]], i64 0, i64 0
4496 // CHECK13-NEXT: store ptr null, ptr [[TMP80]], align 8
4497 // CHECK13-NEXT: [[TMP81:%.*]] = getelementptr inbounds [4 x ptr], ptr [[DOTOFFLOAD_BASEPTRS19]], i32 0, i32 1
4498 // CHECK13-NEXT: store i64 [[TMP76]], ptr [[TMP81]], align 8
4499 // CHECK13-NEXT: [[TMP82:%.*]] = getelementptr inbounds [4 x ptr], ptr [[DOTOFFLOAD_PTRS20]], i32 0, i32 1
4500 // CHECK13-NEXT: store i64 [[TMP76]], ptr [[TMP82]], align 8
4501 // CHECK13-NEXT: [[TMP83:%.*]] = getelementptr inbounds [4 x ptr], ptr [[DOTOFFLOAD_MAPPERS21]], i64 0, i64 1
4502 // CHECK13-NEXT: store ptr null, ptr [[TMP83]], align 8
4503 // CHECK13-NEXT: [[TMP84:%.*]] = getelementptr inbounds [4 x ptr], ptr [[DOTOFFLOAD_BASEPTRS19]], i32 0, i32 2
4504 // CHECK13-NEXT: store i64 [[TMP1]], ptr [[TMP84]], align 8
4505 // CHECK13-NEXT: [[TMP85:%.*]] = getelementptr inbounds [4 x ptr], ptr [[DOTOFFLOAD_PTRS20]], i32 0, i32 2
4506 // CHECK13-NEXT: store i64 [[TMP1]], ptr [[TMP85]], align 8
4507 // CHECK13-NEXT: [[TMP86:%.*]] = getelementptr inbounds [4 x ptr], ptr [[DOTOFFLOAD_MAPPERS21]], i64 0, i64 2
4508 // CHECK13-NEXT: store ptr null, ptr [[TMP86]], align 8
4509 // CHECK13-NEXT: [[TMP87:%.*]] = getelementptr inbounds [4 x ptr], ptr [[DOTOFFLOAD_BASEPTRS19]], i32 0, i32 3
4510 // CHECK13-NEXT: store ptr [[VLA]], ptr [[TMP87]], align 8
4511 // CHECK13-NEXT: [[TMP88:%.*]] = getelementptr inbounds [4 x ptr], ptr [[DOTOFFLOAD_PTRS20]], i32 0, i32 3
4512 // CHECK13-NEXT: store ptr [[VLA]], ptr [[TMP88]], align 8
4513 // CHECK13-NEXT: [[TMP89:%.*]] = getelementptr inbounds [4 x i64], ptr [[DOTOFFLOAD_SIZES22]], i32 0, i32 3
4514 // CHECK13-NEXT: store i64 [[TMP77]], ptr [[TMP89]], align 8
4515 // CHECK13-NEXT: [[TMP90:%.*]] = getelementptr inbounds [4 x ptr], ptr [[DOTOFFLOAD_MAPPERS21]], i64 0, i64 3
4516 // CHECK13-NEXT: store ptr null, ptr [[TMP90]], align 8
4517 // CHECK13-NEXT: [[TMP91:%.*]] = getelementptr inbounds [4 x ptr], ptr [[DOTOFFLOAD_BASEPTRS19]], i32 0, i32 0
4518 // CHECK13-NEXT: [[TMP92:%.*]] = getelementptr inbounds [4 x ptr], ptr [[DOTOFFLOAD_PTRS20]], i32 0, i32 0
4519 // CHECK13-NEXT: [[TMP93:%.*]] = getelementptr inbounds [4 x i64], ptr [[DOTOFFLOAD_SIZES22]], i32 0, i32 0
4520 // CHECK13-NEXT: [[TMP94:%.*]] = load i32, ptr [[N]], align 4
4521 // CHECK13-NEXT: store i32 [[TMP94]], ptr [[DOTCAPTURE_EXPR_24]], align 4
4522 // CHECK13-NEXT: [[TMP95:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_24]], align 4
4523 // CHECK13-NEXT: [[SUB26:%.*]] = sub nsw i32 [[TMP95]], 0
4524 // CHECK13-NEXT: [[DIV27:%.*]] = sdiv i32 [[SUB26]], 1
4525 // CHECK13-NEXT: [[SUB28:%.*]] = sub nsw i32 [[DIV27]], 1
4526 // CHECK13-NEXT: store i32 [[SUB28]], ptr [[DOTCAPTURE_EXPR_25]], align 4
4527 // CHECK13-NEXT: [[TMP96:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_25]], align 4
4528 // CHECK13-NEXT: [[ADD29:%.*]] = add nsw i32 [[TMP96]], 1
4529 // CHECK13-NEXT: [[TMP97:%.*]] = zext i32 [[ADD29]] to i64
4530 // CHECK13-NEXT: [[TMP98:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS30]], i32 0, i32 0
4531 // CHECK13-NEXT: store i32 3, ptr [[TMP98]], align 4
4532 // CHECK13-NEXT: [[TMP99:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS30]], i32 0, i32 1
4533 // CHECK13-NEXT: store i32 4, ptr [[TMP99]], align 4
4534 // CHECK13-NEXT: [[TMP100:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS30]], i32 0, i32 2
4535 // CHECK13-NEXT: store ptr [[TMP91]], ptr [[TMP100]], align 8
4536 // CHECK13-NEXT: [[TMP101:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS30]], i32 0, i32 3
4537 // CHECK13-NEXT: store ptr [[TMP92]], ptr [[TMP101]], align 8
4538 // CHECK13-NEXT: [[TMP102:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS30]], i32 0, i32 4
4539 // CHECK13-NEXT: store ptr [[TMP93]], ptr [[TMP102]], align 8
4540 // CHECK13-NEXT: [[TMP103:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS30]], i32 0, i32 5
4541 // CHECK13-NEXT: store ptr @.offload_maptypes.4, ptr [[TMP103]], align 8
4542 // CHECK13-NEXT: [[TMP104:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS30]], i32 0, i32 6
4543 // CHECK13-NEXT: store ptr null, ptr [[TMP104]], align 8
4544 // CHECK13-NEXT: [[TMP105:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS30]], i32 0, i32 7
4545 // CHECK13-NEXT: store ptr null, ptr [[TMP105]], align 8
4546 // CHECK13-NEXT: [[TMP106:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS30]], i32 0, i32 8
4547 // CHECK13-NEXT: store i64 [[TMP97]], ptr [[TMP106]], align 8
4548 // CHECK13-NEXT: [[TMP107:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS30]], i32 0, i32 9
4549 // CHECK13-NEXT: store i64 0, ptr [[TMP107]], align 8
4550 // CHECK13-NEXT: [[TMP108:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS30]], i32 0, i32 10
4551 // CHECK13-NEXT: store [3 x i32] zeroinitializer, ptr [[TMP108]], align 4
4552 // CHECK13-NEXT: [[TMP109:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS30]], i32 0, i32 11
4553 // CHECK13-NEXT: store [3 x i32] zeroinitializer, ptr [[TMP109]], align 4
4554 // CHECK13-NEXT: [[TMP110:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS30]], i32 0, i32 12
4555 // CHECK13-NEXT: store i32 0, ptr [[TMP110]], align 4
4556 // CHECK13-NEXT: [[TMP111:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB3]], i64 -1, i32 0, i32 0, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l158.region_id, ptr [[KERNEL_ARGS30]])
4557 // CHECK13-NEXT: [[TMP112:%.*]] = icmp ne i32 [[TMP111]], 0
4558 // CHECK13-NEXT: br i1 [[TMP112]], label [[OMP_OFFLOAD_FAILED31:%.*]], label [[OMP_OFFLOAD_CONT32:%.*]]
4559 // CHECK13: omp_offload.failed31:
4560 // CHECK13-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l158(i64 [[TMP74]], i64 [[TMP76]], i64 [[TMP1]], ptr [[VLA]]) #[[ATTR3]]
4561 // CHECK13-NEXT: br label [[OMP_OFFLOAD_CONT32]]
4562 // CHECK13: omp_offload.cont32:
4563 // CHECK13-NEXT: [[TMP113:%.*]] = load i32, ptr [[N]], align 4
4564 // CHECK13-NEXT: store i32 [[TMP113]], ptr [[N_CASTED33]], align 4
4565 // CHECK13-NEXT: [[TMP114:%.*]] = load i64, ptr [[N_CASTED33]], align 8
4566 // CHECK13-NEXT: [[TMP115:%.*]] = mul nuw i64 [[TMP1]], 4
4567 // CHECK13-NEXT: call void @llvm.memcpy.p0.p0.i64(ptr align 8 [[DOTOFFLOAD_SIZES37]], ptr align 8 @.offload_sizes.5, i64 24, i1 false)
4568 // CHECK13-NEXT: [[TMP116:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_BASEPTRS34]], i32 0, i32 0
4569 // CHECK13-NEXT: store i64 [[TMP114]], ptr [[TMP116]], align 8
4570 // CHECK13-NEXT: [[TMP117:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_PTRS35]], i32 0, i32 0
4571 // CHECK13-NEXT: store i64 [[TMP114]], ptr [[TMP117]], align 8
4572 // CHECK13-NEXT: [[TMP118:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_MAPPERS36]], i64 0, i64 0
4573 // CHECK13-NEXT: store ptr null, ptr [[TMP118]], align 8
4574 // CHECK13-NEXT: [[TMP119:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_BASEPTRS34]], i32 0, i32 1
4575 // CHECK13-NEXT: store i64 [[TMP1]], ptr [[TMP119]], align 8
4576 // CHECK13-NEXT: [[TMP120:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_PTRS35]], i32 0, i32 1
4577 // CHECK13-NEXT: store i64 [[TMP1]], ptr [[TMP120]], align 8
4578 // CHECK13-NEXT: [[TMP121:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_MAPPERS36]], i64 0, i64 1
4579 // CHECK13-NEXT: store ptr null, ptr [[TMP121]], align 8
4580 // CHECK13-NEXT: [[TMP122:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_BASEPTRS34]], i32 0, i32 2
4581 // CHECK13-NEXT: store ptr [[VLA]], ptr [[TMP122]], align 8
4582 // CHECK13-NEXT: [[TMP123:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_PTRS35]], i32 0, i32 2
4583 // CHECK13-NEXT: store ptr [[VLA]], ptr [[TMP123]], align 8
4584 // CHECK13-NEXT: [[TMP124:%.*]] = getelementptr inbounds [3 x i64], ptr [[DOTOFFLOAD_SIZES37]], i32 0, i32 2
4585 // CHECK13-NEXT: store i64 [[TMP115]], ptr [[TMP124]], align 8
4586 // CHECK13-NEXT: [[TMP125:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_MAPPERS36]], i64 0, i64 2
4587 // CHECK13-NEXT: store ptr null, ptr [[TMP125]], align 8
4588 // CHECK13-NEXT: [[TMP126:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_BASEPTRS34]], i32 0, i32 0
4589 // CHECK13-NEXT: [[TMP127:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_PTRS35]], i32 0, i32 0
4590 // CHECK13-NEXT: [[TMP128:%.*]] = getelementptr inbounds [3 x i64], ptr [[DOTOFFLOAD_SIZES37]], i32 0, i32 0
4591 // CHECK13-NEXT: [[TMP129:%.*]] = load i32, ptr [[N]], align 4
4592 // CHECK13-NEXT: store i32 [[TMP129]], ptr [[DOTCAPTURE_EXPR_39]], align 4
4593 // CHECK13-NEXT: [[TMP130:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_39]], align 4
4594 // CHECK13-NEXT: [[SUB41:%.*]] = sub nsw i32 [[TMP130]], 0
4595 // CHECK13-NEXT: [[DIV42:%.*]] = sdiv i32 [[SUB41]], 1
4596 // CHECK13-NEXT: [[SUB43:%.*]] = sub nsw i32 [[DIV42]], 1
4597 // CHECK13-NEXT: store i32 [[SUB43]], ptr [[DOTCAPTURE_EXPR_40]], align 4
4598 // CHECK13-NEXT: [[TMP131:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_40]], align 4
4599 // CHECK13-NEXT: [[ADD44:%.*]] = add nsw i32 [[TMP131]], 1
4600 // CHECK13-NEXT: [[TMP132:%.*]] = zext i32 [[ADD44]] to i64
4601 // CHECK13-NEXT: [[TMP133:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS45]], i32 0, i32 0
4602 // CHECK13-NEXT: store i32 3, ptr [[TMP133]], align 4
4603 // CHECK13-NEXT: [[TMP134:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS45]], i32 0, i32 1
4604 // CHECK13-NEXT: store i32 3, ptr [[TMP134]], align 4
4605 // CHECK13-NEXT: [[TMP135:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS45]], i32 0, i32 2
4606 // CHECK13-NEXT: store ptr [[TMP126]], ptr [[TMP135]], align 8
4607 // CHECK13-NEXT: [[TMP136:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS45]], i32 0, i32 3
4608 // CHECK13-NEXT: store ptr [[TMP127]], ptr [[TMP136]], align 8
4609 // CHECK13-NEXT: [[TMP137:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS45]], i32 0, i32 4
4610 // CHECK13-NEXT: store ptr [[TMP128]], ptr [[TMP137]], align 8
4611 // CHECK13-NEXT: [[TMP138:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS45]], i32 0, i32 5
4612 // CHECK13-NEXT: store ptr @.offload_maptypes.6, ptr [[TMP138]], align 8
4613 // CHECK13-NEXT: [[TMP139:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS45]], i32 0, i32 6
4614 // CHECK13-NEXT: store ptr null, ptr [[TMP139]], align 8
4615 // CHECK13-NEXT: [[TMP140:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS45]], i32 0, i32 7
4616 // CHECK13-NEXT: store ptr null, ptr [[TMP140]], align 8
4617 // CHECK13-NEXT: [[TMP141:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS45]], i32 0, i32 8
4618 // CHECK13-NEXT: store i64 [[TMP132]], ptr [[TMP141]], align 8
4619 // CHECK13-NEXT: [[TMP142:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS45]], i32 0, i32 9
4620 // CHECK13-NEXT: store i64 0, ptr [[TMP142]], align 8
4621 // CHECK13-NEXT: [[TMP143:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS45]], i32 0, i32 10
4622 // CHECK13-NEXT: store [3 x i32] zeroinitializer, ptr [[TMP143]], align 4
4623 // CHECK13-NEXT: [[TMP144:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS45]], i32 0, i32 11
4624 // CHECK13-NEXT: store [3 x i32] zeroinitializer, ptr [[TMP144]], align 4
4625 // CHECK13-NEXT: [[TMP145:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS45]], i32 0, i32 12
4626 // CHECK13-NEXT: store i32 0, ptr [[TMP145]], align 4
4627 // CHECK13-NEXT: [[TMP146:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB3]], i64 -1, i32 0, i32 0, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l163.region_id, ptr [[KERNEL_ARGS45]])
4628 // CHECK13-NEXT: [[TMP147:%.*]] = icmp ne i32 [[TMP146]], 0
4629 // CHECK13-NEXT: br i1 [[TMP147]], label [[OMP_OFFLOAD_FAILED46:%.*]], label [[OMP_OFFLOAD_CONT47:%.*]]
4630 // CHECK13: omp_offload.failed46:
4631 // CHECK13-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l163(i64 [[TMP114]], i64 [[TMP1]], ptr [[VLA]]) #[[ATTR3]]
4632 // CHECK13-NEXT: br label [[OMP_OFFLOAD_CONT47]]
4633 // CHECK13: omp_offload.cont47:
4634 // CHECK13-NEXT: [[TMP148:%.*]] = load i32, ptr [[M]], align 4
4635 // CHECK13-NEXT: store i32 [[TMP148]], ptr [[M_CASTED48]], align 4
4636 // CHECK13-NEXT: [[TMP149:%.*]] = load i64, ptr [[M_CASTED48]], align 8
4637 // CHECK13-NEXT: [[TMP150:%.*]] = load i32, ptr [[N]], align 4
4638 // CHECK13-NEXT: store i32 [[TMP150]], ptr [[N_CASTED49]], align 4
4639 // CHECK13-NEXT: [[TMP151:%.*]] = load i64, ptr [[N_CASTED49]], align 8
4640 // CHECK13-NEXT: [[TMP152:%.*]] = mul nuw i64 [[TMP1]], 4
4641 // CHECK13-NEXT: call void @llvm.memcpy.p0.p0.i64(ptr align 8 [[DOTOFFLOAD_SIZES53]], ptr align 8 @.offload_sizes.7, i64 32, i1 false)
4642 // CHECK13-NEXT: [[TMP153:%.*]] = getelementptr inbounds [4 x ptr], ptr [[DOTOFFLOAD_BASEPTRS50]], i32 0, i32 0
4643 // CHECK13-NEXT: store i64 [[TMP149]], ptr [[TMP153]], align 8
4644 // CHECK13-NEXT: [[TMP154:%.*]] = getelementptr inbounds [4 x ptr], ptr [[DOTOFFLOAD_PTRS51]], i32 0, i32 0
4645 // CHECK13-NEXT: store i64 [[TMP149]], ptr [[TMP154]], align 8
4646 // CHECK13-NEXT: [[TMP155:%.*]] = getelementptr inbounds [4 x ptr], ptr [[DOTOFFLOAD_MAPPERS52]], i64 0, i64 0
4647 // CHECK13-NEXT: store ptr null, ptr [[TMP155]], align 8
4648 // CHECK13-NEXT: [[TMP156:%.*]] = getelementptr inbounds [4 x ptr], ptr [[DOTOFFLOAD_BASEPTRS50]], i32 0, i32 1
4649 // CHECK13-NEXT: store i64 [[TMP151]], ptr [[TMP156]], align 8
4650 // CHECK13-NEXT: [[TMP157:%.*]] = getelementptr inbounds [4 x ptr], ptr [[DOTOFFLOAD_PTRS51]], i32 0, i32 1
4651 // CHECK13-NEXT: store i64 [[TMP151]], ptr [[TMP157]], align 8
4652 // CHECK13-NEXT: [[TMP158:%.*]] = getelementptr inbounds [4 x ptr], ptr [[DOTOFFLOAD_MAPPERS52]], i64 0, i64 1
4653 // CHECK13-NEXT: store ptr null, ptr [[TMP158]], align 8
4654 // CHECK13-NEXT: [[TMP159:%.*]] = getelementptr inbounds [4 x ptr], ptr [[DOTOFFLOAD_BASEPTRS50]], i32 0, i32 2
4655 // CHECK13-NEXT: store i64 [[TMP1]], ptr [[TMP159]], align 8
4656 // CHECK13-NEXT: [[TMP160:%.*]] = getelementptr inbounds [4 x ptr], ptr [[DOTOFFLOAD_PTRS51]], i32 0, i32 2
4657 // CHECK13-NEXT: store i64 [[TMP1]], ptr [[TMP160]], align 8
4658 // CHECK13-NEXT: [[TMP161:%.*]] = getelementptr inbounds [4 x ptr], ptr [[DOTOFFLOAD_MAPPERS52]], i64 0, i64 2
4659 // CHECK13-NEXT: store ptr null, ptr [[TMP161]], align 8
4660 // CHECK13-NEXT: [[TMP162:%.*]] = getelementptr inbounds [4 x ptr], ptr [[DOTOFFLOAD_BASEPTRS50]], i32 0, i32 3
4661 // CHECK13-NEXT: store ptr [[VLA]], ptr [[TMP162]], align 8
4662 // CHECK13-NEXT: [[TMP163:%.*]] = getelementptr inbounds [4 x ptr], ptr [[DOTOFFLOAD_PTRS51]], i32 0, i32 3
4663 // CHECK13-NEXT: store ptr [[VLA]], ptr [[TMP163]], align 8
4664 // CHECK13-NEXT: [[TMP164:%.*]] = getelementptr inbounds [4 x i64], ptr [[DOTOFFLOAD_SIZES53]], i32 0, i32 3
4665 // CHECK13-NEXT: store i64 [[TMP152]], ptr [[TMP164]], align 8
4666 // CHECK13-NEXT: [[TMP165:%.*]] = getelementptr inbounds [4 x ptr], ptr [[DOTOFFLOAD_MAPPERS52]], i64 0, i64 3
4667 // CHECK13-NEXT: store ptr null, ptr [[TMP165]], align 8
4668 // CHECK13-NEXT: [[TMP166:%.*]] = getelementptr inbounds [4 x ptr], ptr [[DOTOFFLOAD_BASEPTRS50]], i32 0, i32 0
4669 // CHECK13-NEXT: [[TMP167:%.*]] = getelementptr inbounds [4 x ptr], ptr [[DOTOFFLOAD_PTRS51]], i32 0, i32 0
4670 // CHECK13-NEXT: [[TMP168:%.*]] = getelementptr inbounds [4 x i64], ptr [[DOTOFFLOAD_SIZES53]], i32 0, i32 0
4671 // CHECK13-NEXT: [[TMP169:%.*]] = load i32, ptr [[N]], align 4
4672 // CHECK13-NEXT: store i32 [[TMP169]], ptr [[DOTCAPTURE_EXPR_55]], align 4
4673 // CHECK13-NEXT: [[TMP170:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_55]], align 4
4674 // CHECK13-NEXT: [[SUB57:%.*]] = sub nsw i32 [[TMP170]], 0
4675 // CHECK13-NEXT: [[DIV58:%.*]] = sdiv i32 [[SUB57]], 1
4676 // CHECK13-NEXT: [[SUB59:%.*]] = sub nsw i32 [[DIV58]], 1
4677 // CHECK13-NEXT: store i32 [[SUB59]], ptr [[DOTCAPTURE_EXPR_56]], align 4
4678 // CHECK13-NEXT: [[TMP171:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_56]], align 4
4679 // CHECK13-NEXT: [[ADD60:%.*]] = add nsw i32 [[TMP171]], 1
4680 // CHECK13-NEXT: [[TMP172:%.*]] = zext i32 [[ADD60]] to i64
4681 // CHECK13-NEXT: [[TMP173:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS61]], i32 0, i32 0
4682 // CHECK13-NEXT: store i32 3, ptr [[TMP173]], align 4
4683 // CHECK13-NEXT: [[TMP174:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS61]], i32 0, i32 1
4684 // CHECK13-NEXT: store i32 4, ptr [[TMP174]], align 4
4685 // CHECK13-NEXT: [[TMP175:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS61]], i32 0, i32 2
4686 // CHECK13-NEXT: store ptr [[TMP166]], ptr [[TMP175]], align 8
4687 // CHECK13-NEXT: [[TMP176:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS61]], i32 0, i32 3
4688 // CHECK13-NEXT: store ptr [[TMP167]], ptr [[TMP176]], align 8
4689 // CHECK13-NEXT: [[TMP177:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS61]], i32 0, i32 4
4690 // CHECK13-NEXT: store ptr [[TMP168]], ptr [[TMP177]], align 8
4691 // CHECK13-NEXT: [[TMP178:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS61]], i32 0, i32 5
4692 // CHECK13-NEXT: store ptr @.offload_maptypes.8, ptr [[TMP178]], align 8
4693 // CHECK13-NEXT: [[TMP179:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS61]], i32 0, i32 6
4694 // CHECK13-NEXT: store ptr null, ptr [[TMP179]], align 8
4695 // CHECK13-NEXT: [[TMP180:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS61]], i32 0, i32 7
4696 // CHECK13-NEXT: store ptr null, ptr [[TMP180]], align 8
4697 // CHECK13-NEXT: [[TMP181:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS61]], i32 0, i32 8
4698 // CHECK13-NEXT: store i64 [[TMP172]], ptr [[TMP181]], align 8
4699 // CHECK13-NEXT: [[TMP182:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS61]], i32 0, i32 9
4700 // CHECK13-NEXT: store i64 0, ptr [[TMP182]], align 8
4701 // CHECK13-NEXT: [[TMP183:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS61]], i32 0, i32 10
4702 // CHECK13-NEXT: store [3 x i32] zeroinitializer, ptr [[TMP183]], align 4
4703 // CHECK13-NEXT: [[TMP184:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS61]], i32 0, i32 11
4704 // CHECK13-NEXT: store [3 x i32] zeroinitializer, ptr [[TMP184]], align 4
4705 // CHECK13-NEXT: [[TMP185:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS61]], i32 0, i32 12
4706 // CHECK13-NEXT: store i32 0, ptr [[TMP185]], align 4
4707 // CHECK13-NEXT: [[TMP186:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB3]], i64 -1, i32 0, i32 0, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l168.region_id, ptr [[KERNEL_ARGS61]])
4708 // CHECK13-NEXT: [[TMP187:%.*]] = icmp ne i32 [[TMP186]], 0
4709 // CHECK13-NEXT: br i1 [[TMP187]], label [[OMP_OFFLOAD_FAILED62:%.*]], label [[OMP_OFFLOAD_CONT63:%.*]]
4710 // CHECK13: omp_offload.failed62:
4711 // CHECK13-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l168(i64 [[TMP149]], i64 [[TMP151]], i64 [[TMP1]], ptr [[VLA]]) #[[ATTR3]]
4712 // CHECK13-NEXT: br label [[OMP_OFFLOAD_CONT63]]
4713 // CHECK13: omp_offload.cont63:
4714 // CHECK13-NEXT: [[TMP188:%.*]] = load i32, ptr [[ARGC_ADDR]], align 4
4715 // CHECK13-NEXT: [[CALL:%.*]] = call noundef signext i32 @_Z5tmainIiLi10EEiT_(i32 noundef signext [[TMP188]])
4716 // CHECK13-NEXT: store i32 [[CALL]], ptr [[RETVAL]], align 4
4717 // CHECK13-NEXT: [[TMP189:%.*]] = load ptr, ptr [[SAVED_STACK]], align 8
4718 // CHECK13-NEXT: call void @llvm.stackrestore.p0(ptr [[TMP189]])
4719 // CHECK13-NEXT: [[TMP190:%.*]] = load i32, ptr [[RETVAL]], align 4
4720 // CHECK13-NEXT: ret i32 [[TMP190]]
4723 // CHECK13-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l148
4724 // CHECK13-SAME: (i64 noundef [[N:%.*]], i64 noundef [[VLA:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[A:%.*]]) #[[ATTR2:[0-9]+]] {
4725 // CHECK13-NEXT: entry:
4726 // CHECK13-NEXT: [[N_ADDR:%.*]] = alloca i64, align 8
4727 // CHECK13-NEXT: [[VLA_ADDR:%.*]] = alloca i64, align 8
4728 // CHECK13-NEXT: [[A_ADDR:%.*]] = alloca ptr, align 8
4729 // CHECK13-NEXT: store i64 [[N]], ptr [[N_ADDR]], align 8
4730 // CHECK13-NEXT: store i64 [[VLA]], ptr [[VLA_ADDR]], align 8
4731 // CHECK13-NEXT: store ptr [[A]], ptr [[A_ADDR]], align 8
4732 // CHECK13-NEXT: [[TMP0:%.*]] = load i64, ptr [[VLA_ADDR]], align 8
4733 // CHECK13-NEXT: [[TMP1:%.*]] = load ptr, ptr [[A_ADDR]], align 8
4734 // CHECK13-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_teams(ptr @[[GLOB3]], i32 3, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l148.omp_outlined, ptr [[N_ADDR]], i64 [[TMP0]], ptr [[TMP1]])
4735 // CHECK13-NEXT: ret void
4738 // CHECK13-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l148.omp_outlined
4739 // CHECK13-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[N:%.*]], i64 noundef [[VLA:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[A:%.*]]) #[[ATTR2]] {
4740 // CHECK13-NEXT: entry:
4741 // CHECK13-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 8
4742 // CHECK13-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 8
4743 // CHECK13-NEXT: [[N_ADDR:%.*]] = alloca ptr, align 8
4744 // CHECK13-NEXT: [[VLA_ADDR:%.*]] = alloca i64, align 8
4745 // CHECK13-NEXT: [[A_ADDR:%.*]] = alloca ptr, align 8
4746 // CHECK13-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
4747 // CHECK13-NEXT: [[TMP:%.*]] = alloca i32, align 4
4748 // CHECK13-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4
4749 // CHECK13-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4
4750 // CHECK13-NEXT: [[I:%.*]] = alloca i32, align 4
4751 // CHECK13-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4
4752 // CHECK13-NEXT: [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4
4753 // CHECK13-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
4754 // CHECK13-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
4755 // CHECK13-NEXT: [[I3:%.*]] = alloca i32, align 4
4756 // CHECK13-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 8
4757 // CHECK13-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 8
4758 // CHECK13-NEXT: store ptr [[N]], ptr [[N_ADDR]], align 8
4759 // CHECK13-NEXT: store i64 [[VLA]], ptr [[VLA_ADDR]], align 8
4760 // CHECK13-NEXT: store ptr [[A]], ptr [[A_ADDR]], align 8
4761 // CHECK13-NEXT: [[TMP0:%.*]] = load ptr, ptr [[N_ADDR]], align 8
4762 // CHECK13-NEXT: [[TMP1:%.*]] = load i64, ptr [[VLA_ADDR]], align 8
4763 // CHECK13-NEXT: [[TMP2:%.*]] = load ptr, ptr [[A_ADDR]], align 8
4764 // CHECK13-NEXT: [[TMP3:%.*]] = load i32, ptr [[TMP0]], align 4
4765 // CHECK13-NEXT: store i32 [[TMP3]], ptr [[DOTCAPTURE_EXPR_]], align 4
4766 // CHECK13-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4
4767 // CHECK13-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP4]], 0
4768 // CHECK13-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1
4769 // CHECK13-NEXT: [[SUB2:%.*]] = sub nsw i32 [[DIV]], 1
4770 // CHECK13-NEXT: store i32 [[SUB2]], ptr [[DOTCAPTURE_EXPR_1]], align 4
4771 // CHECK13-NEXT: store i32 0, ptr [[I]], align 4
4772 // CHECK13-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4
4773 // CHECK13-NEXT: [[CMP:%.*]] = icmp slt i32 0, [[TMP5]]
4774 // CHECK13-NEXT: br i1 [[CMP]], label [[OMP_PRECOND_THEN:%.*]], label [[OMP_PRECOND_END:%.*]]
4775 // CHECK13: omp.precond.then:
4776 // CHECK13-NEXT: store i32 0, ptr [[DOTOMP_COMB_LB]], align 4
4777 // CHECK13-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4
4778 // CHECK13-NEXT: store i32 [[TMP6]], ptr [[DOTOMP_COMB_UB]], align 4
4779 // CHECK13-NEXT: store i32 1, ptr [[DOTOMP_STRIDE]], align 4
4780 // CHECK13-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
4781 // CHECK13-NEXT: [[TMP7:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
4782 // CHECK13-NEXT: [[TMP8:%.*]] = load i32, ptr [[TMP7]], align 4
4783 // CHECK13-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB1:[0-9]+]], i32 [[TMP8]], i32 92, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
4784 // CHECK13-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
4785 // CHECK13-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4
4786 // CHECK13-NEXT: [[CMP4:%.*]] = icmp sgt i32 [[TMP9]], [[TMP10]]
4787 // CHECK13-NEXT: br i1 [[CMP4]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
4788 // CHECK13: cond.true:
4789 // CHECK13-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4
4790 // CHECK13-NEXT: br label [[COND_END:%.*]]
4791 // CHECK13: cond.false:
4792 // CHECK13-NEXT: [[TMP12:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
4793 // CHECK13-NEXT: br label [[COND_END]]
4794 // CHECK13: cond.end:
4795 // CHECK13-NEXT: [[COND:%.*]] = phi i32 [ [[TMP11]], [[COND_TRUE]] ], [ [[TMP12]], [[COND_FALSE]] ]
4796 // CHECK13-NEXT: store i32 [[COND]], ptr [[DOTOMP_COMB_UB]], align 4
4797 // CHECK13-NEXT: [[TMP13:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
4798 // CHECK13-NEXT: store i32 [[TMP13]], ptr [[DOTOMP_IV]], align 4
4799 // CHECK13-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
4800 // CHECK13: omp.inner.for.cond:
4801 // CHECK13-NEXT: [[TMP14:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
4802 // CHECK13-NEXT: [[TMP15:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
4803 // CHECK13-NEXT: [[CMP5:%.*]] = icmp sle i32 [[TMP14]], [[TMP15]]
4804 // CHECK13-NEXT: br i1 [[CMP5]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
4805 // CHECK13: omp.inner.for.body:
4806 // CHECK13-NEXT: [[TMP16:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
4807 // CHECK13-NEXT: [[TMP17:%.*]] = zext i32 [[TMP16]] to i64
4808 // CHECK13-NEXT: [[TMP18:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
4809 // CHECK13-NEXT: [[TMP19:%.*]] = zext i32 [[TMP18]] to i64
4810 // CHECK13-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_call(ptr @[[GLOB3]], i32 5, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l148.omp_outlined.omp_outlined, i64 [[TMP17]], i64 [[TMP19]], ptr [[TMP0]], i64 [[TMP1]], ptr [[TMP2]])
4811 // CHECK13-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
4812 // CHECK13: omp.inner.for.inc:
4813 // CHECK13-NEXT: [[TMP20:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
4814 // CHECK13-NEXT: [[TMP21:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4
4815 // CHECK13-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP20]], [[TMP21]]
4816 // CHECK13-NEXT: store i32 [[ADD]], ptr [[DOTOMP_IV]], align 4
4817 // CHECK13-NEXT: br label [[OMP_INNER_FOR_COND]]
4818 // CHECK13: omp.inner.for.end:
4819 // CHECK13-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
4820 // CHECK13: omp.loop.exit:
4821 // CHECK13-NEXT: [[TMP22:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
4822 // CHECK13-NEXT: [[TMP23:%.*]] = load i32, ptr [[TMP22]], align 4
4823 // CHECK13-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB1]], i32 [[TMP23]])
4824 // CHECK13-NEXT: br label [[OMP_PRECOND_END]]
4825 // CHECK13: omp.precond.end:
4826 // CHECK13-NEXT: ret void
4829 // CHECK13-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l148.omp_outlined.omp_outlined
4830 // CHECK13-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], i64 noundef [[DOTPREVIOUS_LB_:%.*]], i64 noundef [[DOTPREVIOUS_UB_:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[N:%.*]], i64 noundef [[VLA:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[A:%.*]]) #[[ATTR2]] {
4831 // CHECK13-NEXT: entry:
4832 // CHECK13-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 8
4833 // CHECK13-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 8
4834 // CHECK13-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i64, align 8
4835 // CHECK13-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i64, align 8
4836 // CHECK13-NEXT: [[N_ADDR:%.*]] = alloca ptr, align 8
4837 // CHECK13-NEXT: [[VLA_ADDR:%.*]] = alloca i64, align 8
4838 // CHECK13-NEXT: [[A_ADDR:%.*]] = alloca ptr, align 8
4839 // CHECK13-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
4840 // CHECK13-NEXT: [[TMP:%.*]] = alloca i32, align 4
4841 // CHECK13-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4
4842 // CHECK13-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4
4843 // CHECK13-NEXT: [[I:%.*]] = alloca i32, align 4
4844 // CHECK13-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4
4845 // CHECK13-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4
4846 // CHECK13-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
4847 // CHECK13-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
4848 // CHECK13-NEXT: [[I4:%.*]] = alloca i32, align 4
4849 // CHECK13-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 8
4850 // CHECK13-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 8
4851 // CHECK13-NEXT: store i64 [[DOTPREVIOUS_LB_]], ptr [[DOTPREVIOUS_LB__ADDR]], align 8
4852 // CHECK13-NEXT: store i64 [[DOTPREVIOUS_UB_]], ptr [[DOTPREVIOUS_UB__ADDR]], align 8
4853 // CHECK13-NEXT: store ptr [[N]], ptr [[N_ADDR]], align 8
4854 // CHECK13-NEXT: store i64 [[VLA]], ptr [[VLA_ADDR]], align 8
4855 // CHECK13-NEXT: store ptr [[A]], ptr [[A_ADDR]], align 8
4856 // CHECK13-NEXT: [[TMP0:%.*]] = load ptr, ptr [[N_ADDR]], align 8
4857 // CHECK13-NEXT: [[TMP1:%.*]] = load i64, ptr [[VLA_ADDR]], align 8
4858 // CHECK13-NEXT: [[TMP2:%.*]] = load ptr, ptr [[A_ADDR]], align 8
4859 // CHECK13-NEXT: [[TMP3:%.*]] = load i32, ptr [[TMP0]], align 4
4860 // CHECK13-NEXT: store i32 [[TMP3]], ptr [[DOTCAPTURE_EXPR_]], align 4
4861 // CHECK13-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4
4862 // CHECK13-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP4]], 0
4863 // CHECK13-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1
4864 // CHECK13-NEXT: [[SUB2:%.*]] = sub nsw i32 [[DIV]], 1
4865 // CHECK13-NEXT: store i32 [[SUB2]], ptr [[DOTCAPTURE_EXPR_1]], align 4
4866 // CHECK13-NEXT: store i32 0, ptr [[I]], align 4
4867 // CHECK13-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4
4868 // CHECK13-NEXT: [[CMP:%.*]] = icmp slt i32 0, [[TMP5]]
4869 // CHECK13-NEXT: br i1 [[CMP]], label [[OMP_PRECOND_THEN:%.*]], label [[OMP_PRECOND_END:%.*]]
4870 // CHECK13: omp.precond.then:
4871 // CHECK13-NEXT: store i32 0, ptr [[DOTOMP_LB]], align 4
4872 // CHECK13-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4
4873 // CHECK13-NEXT: store i32 [[TMP6]], ptr [[DOTOMP_UB]], align 4
4874 // CHECK13-NEXT: [[TMP7:%.*]] = load i64, ptr [[DOTPREVIOUS_LB__ADDR]], align 8
4875 // CHECK13-NEXT: [[CONV:%.*]] = trunc i64 [[TMP7]] to i32
4876 // CHECK13-NEXT: [[TMP8:%.*]] = load i64, ptr [[DOTPREVIOUS_UB__ADDR]], align 8
4877 // CHECK13-NEXT: [[CONV3:%.*]] = trunc i64 [[TMP8]] to i32
4878 // CHECK13-NEXT: store i32 [[CONV]], ptr [[DOTOMP_LB]], align 4
4879 // CHECK13-NEXT: store i32 [[CONV3]], ptr [[DOTOMP_UB]], align 4
4880 // CHECK13-NEXT: store i32 1, ptr [[DOTOMP_STRIDE]], align 4
4881 // CHECK13-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
4882 // CHECK13-NEXT: [[TMP9:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
4883 // CHECK13-NEXT: [[TMP10:%.*]] = load i32, ptr [[TMP9]], align 4
4884 // CHECK13-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB2:[0-9]+]], i32 [[TMP10]], i32 34, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
4885 // CHECK13-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
4886 // CHECK13-NEXT: [[TMP12:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4
4887 // CHECK13-NEXT: [[CMP5:%.*]] = icmp sgt i32 [[TMP11]], [[TMP12]]
4888 // CHECK13-NEXT: br i1 [[CMP5]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
4889 // CHECK13: cond.true:
4890 // CHECK13-NEXT: [[TMP13:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4
4891 // CHECK13-NEXT: br label [[COND_END:%.*]]
4892 // CHECK13: cond.false:
4893 // CHECK13-NEXT: [[TMP14:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
4894 // CHECK13-NEXT: br label [[COND_END]]
4895 // CHECK13: cond.end:
4896 // CHECK13-NEXT: [[COND:%.*]] = phi i32 [ [[TMP13]], [[COND_TRUE]] ], [ [[TMP14]], [[COND_FALSE]] ]
4897 // CHECK13-NEXT: store i32 [[COND]], ptr [[DOTOMP_UB]], align 4
4898 // CHECK13-NEXT: [[TMP15:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4
4899 // CHECK13-NEXT: store i32 [[TMP15]], ptr [[DOTOMP_IV]], align 4
4900 // CHECK13-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
4901 // CHECK13: omp.inner.for.cond:
4902 // CHECK13-NEXT: [[TMP16:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
4903 // CHECK13-NEXT: [[TMP17:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
4904 // CHECK13-NEXT: [[CMP6:%.*]] = icmp sle i32 [[TMP16]], [[TMP17]]
4905 // CHECK13-NEXT: br i1 [[CMP6]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
4906 // CHECK13: omp.inner.for.body:
4907 // CHECK13-NEXT: [[TMP18:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
4908 // CHECK13-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP18]], 1
4909 // CHECK13-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
4910 // CHECK13-NEXT: store i32 [[ADD]], ptr [[I4]], align 4
4911 // CHECK13-NEXT: [[TMP19:%.*]] = load i32, ptr [[I4]], align 4
4912 // CHECK13-NEXT: [[IDXPROM:%.*]] = sext i32 [[TMP19]] to i64
4913 // CHECK13-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i32, ptr [[TMP2]], i64 [[IDXPROM]]
4914 // CHECK13-NEXT: store i32 0, ptr [[ARRAYIDX]], align 4
4915 // CHECK13-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
4916 // CHECK13: omp.body.continue:
4917 // CHECK13-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
4918 // CHECK13: omp.inner.for.inc:
4919 // CHECK13-NEXT: [[TMP20:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
4920 // CHECK13-NEXT: [[ADD7:%.*]] = add nsw i32 [[TMP20]], 1
4921 // CHECK13-NEXT: store i32 [[ADD7]], ptr [[DOTOMP_IV]], align 4
4922 // CHECK13-NEXT: br label [[OMP_INNER_FOR_COND]]
4923 // CHECK13: omp.inner.for.end:
4924 // CHECK13-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
4925 // CHECK13: omp.loop.exit:
4926 // CHECK13-NEXT: [[TMP21:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
4927 // CHECK13-NEXT: [[TMP22:%.*]] = load i32, ptr [[TMP21]], align 4
4928 // CHECK13-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB2]], i32 [[TMP22]])
4929 // CHECK13-NEXT: br label [[OMP_PRECOND_END]]
4930 // CHECK13: omp.precond.end:
4931 // CHECK13-NEXT: ret void
4934 // CHECK13-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l153
4935 // CHECK13-SAME: (i64 noundef [[N:%.*]], i64 noundef [[VLA:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[A:%.*]]) #[[ATTR2]] {
4936 // CHECK13-NEXT: entry:
4937 // CHECK13-NEXT: [[N_ADDR:%.*]] = alloca i64, align 8
4938 // CHECK13-NEXT: [[VLA_ADDR:%.*]] = alloca i64, align 8
4939 // CHECK13-NEXT: [[A_ADDR:%.*]] = alloca ptr, align 8
4940 // CHECK13-NEXT: store i64 [[N]], ptr [[N_ADDR]], align 8
4941 // CHECK13-NEXT: store i64 [[VLA]], ptr [[VLA_ADDR]], align 8
4942 // CHECK13-NEXT: store ptr [[A]], ptr [[A_ADDR]], align 8
4943 // CHECK13-NEXT: [[TMP0:%.*]] = load i64, ptr [[VLA_ADDR]], align 8
4944 // CHECK13-NEXT: [[TMP1:%.*]] = load ptr, ptr [[A_ADDR]], align 8
4945 // CHECK13-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_teams(ptr @[[GLOB3]], i32 3, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l153.omp_outlined, ptr [[N_ADDR]], i64 [[TMP0]], ptr [[TMP1]])
4946 // CHECK13-NEXT: ret void
4949 // CHECK13-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l153.omp_outlined
4950 // CHECK13-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[N:%.*]], i64 noundef [[VLA:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[A:%.*]]) #[[ATTR2]] {
4951 // CHECK13-NEXT: entry:
4952 // CHECK13-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 8
4953 // CHECK13-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 8
4954 // CHECK13-NEXT: [[N_ADDR:%.*]] = alloca ptr, align 8
4955 // CHECK13-NEXT: [[VLA_ADDR:%.*]] = alloca i64, align 8
4956 // CHECK13-NEXT: [[A_ADDR:%.*]] = alloca ptr, align 8
4957 // CHECK13-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
4958 // CHECK13-NEXT: [[TMP:%.*]] = alloca i32, align 4
4959 // CHECK13-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4
4960 // CHECK13-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4
4961 // CHECK13-NEXT: [[I:%.*]] = alloca i32, align 4
4962 // CHECK13-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4
4963 // CHECK13-NEXT: [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4
4964 // CHECK13-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
4965 // CHECK13-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
4966 // CHECK13-NEXT: [[I3:%.*]] = alloca i32, align 4
4967 // CHECK13-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 8
4968 // CHECK13-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 8
4969 // CHECK13-NEXT: store ptr [[N]], ptr [[N_ADDR]], align 8
4970 // CHECK13-NEXT: store i64 [[VLA]], ptr [[VLA_ADDR]], align 8
4971 // CHECK13-NEXT: store ptr [[A]], ptr [[A_ADDR]], align 8
4972 // CHECK13-NEXT: [[TMP0:%.*]] = load ptr, ptr [[N_ADDR]], align 8
4973 // CHECK13-NEXT: [[TMP1:%.*]] = load i64, ptr [[VLA_ADDR]], align 8
4974 // CHECK13-NEXT: [[TMP2:%.*]] = load ptr, ptr [[A_ADDR]], align 8
4975 // CHECK13-NEXT: [[TMP3:%.*]] = load i32, ptr [[TMP0]], align 4
4976 // CHECK13-NEXT: store i32 [[TMP3]], ptr [[DOTCAPTURE_EXPR_]], align 4
4977 // CHECK13-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4
4978 // CHECK13-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP4]], 0
4979 // CHECK13-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1
4980 // CHECK13-NEXT: [[SUB2:%.*]] = sub nsw i32 [[DIV]], 1
4981 // CHECK13-NEXT: store i32 [[SUB2]], ptr [[DOTCAPTURE_EXPR_1]], align 4
4982 // CHECK13-NEXT: store i32 0, ptr [[I]], align 4
4983 // CHECK13-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4
4984 // CHECK13-NEXT: [[CMP:%.*]] = icmp slt i32 0, [[TMP5]]
4985 // CHECK13-NEXT: br i1 [[CMP]], label [[OMP_PRECOND_THEN:%.*]], label [[OMP_PRECOND_END:%.*]]
4986 // CHECK13: omp.precond.then:
4987 // CHECK13-NEXT: store i32 0, ptr [[DOTOMP_COMB_LB]], align 4
4988 // CHECK13-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4
4989 // CHECK13-NEXT: store i32 [[TMP6]], ptr [[DOTOMP_COMB_UB]], align 4
4990 // CHECK13-NEXT: store i32 1, ptr [[DOTOMP_STRIDE]], align 4
4991 // CHECK13-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
4992 // CHECK13-NEXT: [[TMP7:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
4993 // CHECK13-NEXT: [[TMP8:%.*]] = load i32, ptr [[TMP7]], align 4
4994 // CHECK13-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB1]], i32 [[TMP8]], i32 92, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
4995 // CHECK13-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
4996 // CHECK13-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4
4997 // CHECK13-NEXT: [[CMP4:%.*]] = icmp sgt i32 [[TMP9]], [[TMP10]]
4998 // CHECK13-NEXT: br i1 [[CMP4]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
4999 // CHECK13: cond.true:
5000 // CHECK13-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4
5001 // CHECK13-NEXT: br label [[COND_END:%.*]]
5002 // CHECK13: cond.false:
5003 // CHECK13-NEXT: [[TMP12:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
5004 // CHECK13-NEXT: br label [[COND_END]]
5005 // CHECK13: cond.end:
5006 // CHECK13-NEXT: [[COND:%.*]] = phi i32 [ [[TMP11]], [[COND_TRUE]] ], [ [[TMP12]], [[COND_FALSE]] ]
5007 // CHECK13-NEXT: store i32 [[COND]], ptr [[DOTOMP_COMB_UB]], align 4
5008 // CHECK13-NEXT: [[TMP13:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
5009 // CHECK13-NEXT: store i32 [[TMP13]], ptr [[DOTOMP_IV]], align 4
5010 // CHECK13-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
5011 // CHECK13: omp.inner.for.cond:
5012 // CHECK13-NEXT: [[TMP14:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
5013 // CHECK13-NEXT: [[TMP15:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
5014 // CHECK13-NEXT: [[CMP5:%.*]] = icmp sle i32 [[TMP14]], [[TMP15]]
5015 // CHECK13-NEXT: br i1 [[CMP5]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
5016 // CHECK13: omp.inner.for.body:
5017 // CHECK13-NEXT: [[TMP16:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
5018 // CHECK13-NEXT: [[TMP17:%.*]] = zext i32 [[TMP16]] to i64
5019 // CHECK13-NEXT: [[TMP18:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
5020 // CHECK13-NEXT: [[TMP19:%.*]] = zext i32 [[TMP18]] to i64
5021 // CHECK13-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_call(ptr @[[GLOB3]], i32 5, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l153.omp_outlined.omp_outlined, i64 [[TMP17]], i64 [[TMP19]], ptr [[TMP0]], i64 [[TMP1]], ptr [[TMP2]])
5022 // CHECK13-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
5023 // CHECK13: omp.inner.for.inc:
5024 // CHECK13-NEXT: [[TMP20:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
5025 // CHECK13-NEXT: [[TMP21:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4
5026 // CHECK13-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP20]], [[TMP21]]
5027 // CHECK13-NEXT: store i32 [[ADD]], ptr [[DOTOMP_IV]], align 4
5028 // CHECK13-NEXT: br label [[OMP_INNER_FOR_COND]]
5029 // CHECK13: omp.inner.for.end:
5030 // CHECK13-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
5031 // CHECK13: omp.loop.exit:
5032 // CHECK13-NEXT: [[TMP22:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
5033 // CHECK13-NEXT: [[TMP23:%.*]] = load i32, ptr [[TMP22]], align 4
5034 // CHECK13-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB1]], i32 [[TMP23]])
5035 // CHECK13-NEXT: br label [[OMP_PRECOND_END]]
5036 // CHECK13: omp.precond.end:
5037 // CHECK13-NEXT: ret void
5040 // CHECK13-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l153.omp_outlined.omp_outlined
5041 // CHECK13-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], i64 noundef [[DOTPREVIOUS_LB_:%.*]], i64 noundef [[DOTPREVIOUS_UB_:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[N:%.*]], i64 noundef [[VLA:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[A:%.*]]) #[[ATTR2]] {
5042 // CHECK13-NEXT: entry:
5043 // CHECK13-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 8
5044 // CHECK13-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 8
5045 // CHECK13-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i64, align 8
5046 // CHECK13-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i64, align 8
5047 // CHECK13-NEXT: [[N_ADDR:%.*]] = alloca ptr, align 8
5048 // CHECK13-NEXT: [[VLA_ADDR:%.*]] = alloca i64, align 8
5049 // CHECK13-NEXT: [[A_ADDR:%.*]] = alloca ptr, align 8
5050 // CHECK13-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
5051 // CHECK13-NEXT: [[TMP:%.*]] = alloca i32, align 4
5052 // CHECK13-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4
5053 // CHECK13-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4
5054 // CHECK13-NEXT: [[I:%.*]] = alloca i32, align 4
5055 // CHECK13-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4
5056 // CHECK13-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4
5057 // CHECK13-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
5058 // CHECK13-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
5059 // CHECK13-NEXT: [[I4:%.*]] = alloca i32, align 4
5060 // CHECK13-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 8
5061 // CHECK13-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 8
5062 // CHECK13-NEXT: store i64 [[DOTPREVIOUS_LB_]], ptr [[DOTPREVIOUS_LB__ADDR]], align 8
5063 // CHECK13-NEXT: store i64 [[DOTPREVIOUS_UB_]], ptr [[DOTPREVIOUS_UB__ADDR]], align 8
5064 // CHECK13-NEXT: store ptr [[N]], ptr [[N_ADDR]], align 8
5065 // CHECK13-NEXT: store i64 [[VLA]], ptr [[VLA_ADDR]], align 8
5066 // CHECK13-NEXT: store ptr [[A]], ptr [[A_ADDR]], align 8
5067 // CHECK13-NEXT: [[TMP0:%.*]] = load ptr, ptr [[N_ADDR]], align 8
5068 // CHECK13-NEXT: [[TMP1:%.*]] = load i64, ptr [[VLA_ADDR]], align 8
5069 // CHECK13-NEXT: [[TMP2:%.*]] = load ptr, ptr [[A_ADDR]], align 8
5070 // CHECK13-NEXT: [[TMP3:%.*]] = load i32, ptr [[TMP0]], align 4
5071 // CHECK13-NEXT: store i32 [[TMP3]], ptr [[DOTCAPTURE_EXPR_]], align 4
5072 // CHECK13-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4
5073 // CHECK13-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP4]], 0
5074 // CHECK13-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1
5075 // CHECK13-NEXT: [[SUB2:%.*]] = sub nsw i32 [[DIV]], 1
5076 // CHECK13-NEXT: store i32 [[SUB2]], ptr [[DOTCAPTURE_EXPR_1]], align 4
5077 // CHECK13-NEXT: store i32 0, ptr [[I]], align 4
5078 // CHECK13-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4
5079 // CHECK13-NEXT: [[CMP:%.*]] = icmp slt i32 0, [[TMP5]]
5080 // CHECK13-NEXT: br i1 [[CMP]], label [[OMP_PRECOND_THEN:%.*]], label [[OMP_PRECOND_END:%.*]]
5081 // CHECK13: omp.precond.then:
5082 // CHECK13-NEXT: store i32 0, ptr [[DOTOMP_LB]], align 4
5083 // CHECK13-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4
5084 // CHECK13-NEXT: store i32 [[TMP6]], ptr [[DOTOMP_UB]], align 4
5085 // CHECK13-NEXT: [[TMP7:%.*]] = load i64, ptr [[DOTPREVIOUS_LB__ADDR]], align 8
5086 // CHECK13-NEXT: [[CONV:%.*]] = trunc i64 [[TMP7]] to i32
5087 // CHECK13-NEXT: [[TMP8:%.*]] = load i64, ptr [[DOTPREVIOUS_UB__ADDR]], align 8
5088 // CHECK13-NEXT: [[CONV3:%.*]] = trunc i64 [[TMP8]] to i32
5089 // CHECK13-NEXT: store i32 [[CONV]], ptr [[DOTOMP_LB]], align 4
5090 // CHECK13-NEXT: store i32 [[CONV3]], ptr [[DOTOMP_UB]], align 4
5091 // CHECK13-NEXT: store i32 1, ptr [[DOTOMP_STRIDE]], align 4
5092 // CHECK13-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
5093 // CHECK13-NEXT: [[TMP9:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
5094 // CHECK13-NEXT: [[TMP10:%.*]] = load i32, ptr [[TMP9]], align 4
5095 // CHECK13-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB2]], i32 [[TMP10]], i32 34, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
5096 // CHECK13-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
5097 // CHECK13-NEXT: [[TMP12:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4
5098 // CHECK13-NEXT: [[CMP5:%.*]] = icmp sgt i32 [[TMP11]], [[TMP12]]
5099 // CHECK13-NEXT: br i1 [[CMP5]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
5100 // CHECK13: cond.true:
5101 // CHECK13-NEXT: [[TMP13:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4
5102 // CHECK13-NEXT: br label [[COND_END:%.*]]
5103 // CHECK13: cond.false:
5104 // CHECK13-NEXT: [[TMP14:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
5105 // CHECK13-NEXT: br label [[COND_END]]
5106 // CHECK13: cond.end:
5107 // CHECK13-NEXT: [[COND:%.*]] = phi i32 [ [[TMP13]], [[COND_TRUE]] ], [ [[TMP14]], [[COND_FALSE]] ]
5108 // CHECK13-NEXT: store i32 [[COND]], ptr [[DOTOMP_UB]], align 4
5109 // CHECK13-NEXT: [[TMP15:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4
5110 // CHECK13-NEXT: store i32 [[TMP15]], ptr [[DOTOMP_IV]], align 4
5111 // CHECK13-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
5112 // CHECK13: omp.inner.for.cond:
5113 // CHECK13-NEXT: [[TMP16:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
5114 // CHECK13-NEXT: [[TMP17:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
5115 // CHECK13-NEXT: [[CMP6:%.*]] = icmp sle i32 [[TMP16]], [[TMP17]]
5116 // CHECK13-NEXT: br i1 [[CMP6]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
5117 // CHECK13: omp.inner.for.body:
5118 // CHECK13-NEXT: [[TMP18:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
5119 // CHECK13-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP18]], 1
5120 // CHECK13-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
5121 // CHECK13-NEXT: store i32 [[ADD]], ptr [[I4]], align 4
5122 // CHECK13-NEXT: [[TMP19:%.*]] = load i32, ptr [[I4]], align 4
5123 // CHECK13-NEXT: [[IDXPROM:%.*]] = sext i32 [[TMP19]] to i64
5124 // CHECK13-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i32, ptr [[TMP2]], i64 [[IDXPROM]]
5125 // CHECK13-NEXT: store i32 0, ptr [[ARRAYIDX]], align 4
5126 // CHECK13-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
5127 // CHECK13: omp.body.continue:
5128 // CHECK13-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
5129 // CHECK13: omp.inner.for.inc:
5130 // CHECK13-NEXT: [[TMP20:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
5131 // CHECK13-NEXT: [[ADD7:%.*]] = add nsw i32 [[TMP20]], 1
5132 // CHECK13-NEXT: store i32 [[ADD7]], ptr [[DOTOMP_IV]], align 4
5133 // CHECK13-NEXT: br label [[OMP_INNER_FOR_COND]]
5134 // CHECK13: omp.inner.for.end:
5135 // CHECK13-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
5136 // CHECK13: omp.loop.exit:
5137 // CHECK13-NEXT: [[TMP21:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
5138 // CHECK13-NEXT: [[TMP22:%.*]] = load i32, ptr [[TMP21]], align 4
5139 // CHECK13-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB2]], i32 [[TMP22]])
5140 // CHECK13-NEXT: br label [[OMP_PRECOND_END]]
5141 // CHECK13: omp.precond.end:
5142 // CHECK13-NEXT: ret void
5145 // CHECK13-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l158
5146 // CHECK13-SAME: (i64 noundef [[M:%.*]], i64 noundef [[N:%.*]], i64 noundef [[VLA:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[A:%.*]]) #[[ATTR2]] {
5147 // CHECK13-NEXT: entry:
5148 // CHECK13-NEXT: [[M_ADDR:%.*]] = alloca i64, align 8
5149 // CHECK13-NEXT: [[N_ADDR:%.*]] = alloca i64, align 8
5150 // CHECK13-NEXT: [[VLA_ADDR:%.*]] = alloca i64, align 8
5151 // CHECK13-NEXT: [[A_ADDR:%.*]] = alloca ptr, align 8
5152 // CHECK13-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4
5153 // CHECK13-NEXT: [[DOTCAPTURE_EXPR__CASTED:%.*]] = alloca i64, align 8
5154 // CHECK13-NEXT: store i64 [[M]], ptr [[M_ADDR]], align 8
5155 // CHECK13-NEXT: store i64 [[N]], ptr [[N_ADDR]], align 8
5156 // CHECK13-NEXT: store i64 [[VLA]], ptr [[VLA_ADDR]], align 8
5157 // CHECK13-NEXT: store ptr [[A]], ptr [[A_ADDR]], align 8
5158 // CHECK13-NEXT: [[TMP0:%.*]] = load i64, ptr [[VLA_ADDR]], align 8
5159 // CHECK13-NEXT: [[TMP1:%.*]] = load ptr, ptr [[A_ADDR]], align 8
5160 // CHECK13-NEXT: [[TMP2:%.*]] = load i32, ptr [[M_ADDR]], align 4
5161 // CHECK13-NEXT: store i32 [[TMP2]], ptr [[DOTCAPTURE_EXPR_]], align 4
5162 // CHECK13-NEXT: [[TMP3:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4
5163 // CHECK13-NEXT: store i32 [[TMP3]], ptr [[DOTCAPTURE_EXPR__CASTED]], align 4
5164 // CHECK13-NEXT: [[TMP4:%.*]] = load i64, ptr [[DOTCAPTURE_EXPR__CASTED]], align 8
5165 // CHECK13-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_teams(ptr @[[GLOB3]], i32 4, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l158.omp_outlined, ptr [[N_ADDR]], i64 [[TMP0]], ptr [[TMP1]], i64 [[TMP4]])
5166 // CHECK13-NEXT: ret void
5169 // CHECK13-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l158.omp_outlined
5170 // CHECK13-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[N:%.*]], i64 noundef [[VLA:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[A:%.*]], i64 noundef [[DOTCAPTURE_EXPR_:%.*]]) #[[ATTR2]] {
5171 // CHECK13-NEXT: entry:
5172 // CHECK13-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 8
5173 // CHECK13-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 8
5174 // CHECK13-NEXT: [[N_ADDR:%.*]] = alloca ptr, align 8
5175 // CHECK13-NEXT: [[VLA_ADDR:%.*]] = alloca i64, align 8
5176 // CHECK13-NEXT: [[A_ADDR:%.*]] = alloca ptr, align 8
5177 // CHECK13-NEXT: [[DOTCAPTURE_EXPR__ADDR:%.*]] = alloca i64, align 8
5178 // CHECK13-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
5179 // CHECK13-NEXT: [[TMP:%.*]] = alloca i32, align 4
5180 // CHECK13-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4
5181 // CHECK13-NEXT: [[DOTCAPTURE_EXPR_2:%.*]] = alloca i32, align 4
5182 // CHECK13-NEXT: [[I:%.*]] = alloca i32, align 4
5183 // CHECK13-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4
5184 // CHECK13-NEXT: [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4
5185 // CHECK13-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
5186 // CHECK13-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
5187 // CHECK13-NEXT: [[I4:%.*]] = alloca i32, align 4
5188 // CHECK13-NEXT: [[DOTCAPTURE_EXPR__CASTED:%.*]] = alloca i64, align 8
5189 // CHECK13-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 8
5190 // CHECK13-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 8
5191 // CHECK13-NEXT: store ptr [[N]], ptr [[N_ADDR]], align 8
5192 // CHECK13-NEXT: store i64 [[VLA]], ptr [[VLA_ADDR]], align 8
5193 // CHECK13-NEXT: store ptr [[A]], ptr [[A_ADDR]], align 8
5194 // CHECK13-NEXT: store i64 [[DOTCAPTURE_EXPR_]], ptr [[DOTCAPTURE_EXPR__ADDR]], align 8
5195 // CHECK13-NEXT: [[TMP0:%.*]] = load ptr, ptr [[N_ADDR]], align 8
5196 // CHECK13-NEXT: [[TMP1:%.*]] = load i64, ptr [[VLA_ADDR]], align 8
5197 // CHECK13-NEXT: [[TMP2:%.*]] = load ptr, ptr [[A_ADDR]], align 8
5198 // CHECK13-NEXT: [[TMP3:%.*]] = load i32, ptr [[TMP0]], align 4
5199 // CHECK13-NEXT: store i32 [[TMP3]], ptr [[DOTCAPTURE_EXPR_1]], align 4
5200 // CHECK13-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4
5201 // CHECK13-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP4]], 0
5202 // CHECK13-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1
5203 // CHECK13-NEXT: [[SUB3:%.*]] = sub nsw i32 [[DIV]], 1
5204 // CHECK13-NEXT: store i32 [[SUB3]], ptr [[DOTCAPTURE_EXPR_2]], align 4
5205 // CHECK13-NEXT: store i32 0, ptr [[I]], align 4
5206 // CHECK13-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4
5207 // CHECK13-NEXT: [[CMP:%.*]] = icmp slt i32 0, [[TMP5]]
5208 // CHECK13-NEXT: br i1 [[CMP]], label [[OMP_PRECOND_THEN:%.*]], label [[OMP_PRECOND_END:%.*]]
5209 // CHECK13: omp.precond.then:
5210 // CHECK13-NEXT: store i32 0, ptr [[DOTOMP_COMB_LB]], align 4
5211 // CHECK13-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_2]], align 4
5212 // CHECK13-NEXT: store i32 [[TMP6]], ptr [[DOTOMP_COMB_UB]], align 4
5213 // CHECK13-NEXT: store i32 1, ptr [[DOTOMP_STRIDE]], align 4
5214 // CHECK13-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
5215 // CHECK13-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR__ADDR]], align 4
5216 // CHECK13-NEXT: [[TMP8:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
5217 // CHECK13-NEXT: [[TMP9:%.*]] = load i32, ptr [[TMP8]], align 4
5218 // CHECK13-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB1]], i32 [[TMP9]], i32 91, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 [[TMP7]])
5219 // CHECK13-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
5220 // CHECK13-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_2]], align 4
5221 // CHECK13-NEXT: [[CMP5:%.*]] = icmp sgt i32 [[TMP10]], [[TMP11]]
5222 // CHECK13-NEXT: br i1 [[CMP5]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
5223 // CHECK13: cond.true:
5224 // CHECK13-NEXT: [[TMP12:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_2]], align 4
5225 // CHECK13-NEXT: br label [[COND_END:%.*]]
5226 // CHECK13: cond.false:
5227 // CHECK13-NEXT: [[TMP13:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
5228 // CHECK13-NEXT: br label [[COND_END]]
5229 // CHECK13: cond.end:
5230 // CHECK13-NEXT: [[COND:%.*]] = phi i32 [ [[TMP12]], [[COND_TRUE]] ], [ [[TMP13]], [[COND_FALSE]] ]
5231 // CHECK13-NEXT: store i32 [[COND]], ptr [[DOTOMP_COMB_UB]], align 4
5232 // CHECK13-NEXT: [[TMP14:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
5233 // CHECK13-NEXT: store i32 [[TMP14]], ptr [[DOTOMP_IV]], align 4
5234 // CHECK13-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
5235 // CHECK13: omp.inner.for.cond:
5236 // CHECK13-NEXT: [[TMP15:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
5237 // CHECK13-NEXT: [[TMP16:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_2]], align 4
5238 // CHECK13-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP16]], 1
5239 // CHECK13-NEXT: [[CMP6:%.*]] = icmp slt i32 [[TMP15]], [[ADD]]
5240 // CHECK13-NEXT: br i1 [[CMP6]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
5241 // CHECK13: omp.inner.for.body:
5242 // CHECK13-NEXT: [[TMP17:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
5243 // CHECK13-NEXT: [[TMP18:%.*]] = zext i32 [[TMP17]] to i64
5244 // CHECK13-NEXT: [[TMP19:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
5245 // CHECK13-NEXT: [[TMP20:%.*]] = zext i32 [[TMP19]] to i64
5246 // CHECK13-NEXT: [[TMP21:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR__ADDR]], align 4
5247 // CHECK13-NEXT: store i32 [[TMP21]], ptr [[DOTCAPTURE_EXPR__CASTED]], align 4
5248 // CHECK13-NEXT: [[TMP22:%.*]] = load i64, ptr [[DOTCAPTURE_EXPR__CASTED]], align 8
5249 // CHECK13-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_call(ptr @[[GLOB3]], i32 6, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l158.omp_outlined.omp_outlined, i64 [[TMP18]], i64 [[TMP20]], ptr [[TMP0]], i64 [[TMP1]], ptr [[TMP2]], i64 [[TMP22]])
5250 // CHECK13-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
5251 // CHECK13: omp.inner.for.inc:
5252 // CHECK13-NEXT: [[TMP23:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
5253 // CHECK13-NEXT: [[TMP24:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4
5254 // CHECK13-NEXT: [[ADD7:%.*]] = add nsw i32 [[TMP23]], [[TMP24]]
5255 // CHECK13-NEXT: store i32 [[ADD7]], ptr [[DOTOMP_IV]], align 4
5256 // CHECK13-NEXT: [[TMP25:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
5257 // CHECK13-NEXT: [[TMP26:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4
5258 // CHECK13-NEXT: [[ADD8:%.*]] = add nsw i32 [[TMP25]], [[TMP26]]
5259 // CHECK13-NEXT: store i32 [[ADD8]], ptr [[DOTOMP_COMB_LB]], align 4
5260 // CHECK13-NEXT: [[TMP27:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
5261 // CHECK13-NEXT: [[TMP28:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4
5262 // CHECK13-NEXT: [[ADD9:%.*]] = add nsw i32 [[TMP27]], [[TMP28]]
5263 // CHECK13-NEXT: store i32 [[ADD9]], ptr [[DOTOMP_COMB_UB]], align 4
5264 // CHECK13-NEXT: [[TMP29:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
5265 // CHECK13-NEXT: [[TMP30:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_2]], align 4
5266 // CHECK13-NEXT: [[CMP10:%.*]] = icmp sgt i32 [[TMP29]], [[TMP30]]
5267 // CHECK13-NEXT: br i1 [[CMP10]], label [[COND_TRUE11:%.*]], label [[COND_FALSE12:%.*]]
5268 // CHECK13: cond.true11:
5269 // CHECK13-NEXT: [[TMP31:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_2]], align 4
5270 // CHECK13-NEXT: br label [[COND_END13:%.*]]
5271 // CHECK13: cond.false12:
5272 // CHECK13-NEXT: [[TMP32:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
5273 // CHECK13-NEXT: br label [[COND_END13]]
5274 // CHECK13: cond.end13:
5275 // CHECK13-NEXT: [[COND14:%.*]] = phi i32 [ [[TMP31]], [[COND_TRUE11]] ], [ [[TMP32]], [[COND_FALSE12]] ]
5276 // CHECK13-NEXT: store i32 [[COND14]], ptr [[DOTOMP_COMB_UB]], align 4
5277 // CHECK13-NEXT: [[TMP33:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
5278 // CHECK13-NEXT: store i32 [[TMP33]], ptr [[DOTOMP_IV]], align 4
5279 // CHECK13-NEXT: br label [[OMP_INNER_FOR_COND]]
5280 // CHECK13: omp.inner.for.end:
5281 // CHECK13-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
5282 // CHECK13: omp.loop.exit:
5283 // CHECK13-NEXT: [[TMP34:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
5284 // CHECK13-NEXT: [[TMP35:%.*]] = load i32, ptr [[TMP34]], align 4
5285 // CHECK13-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB1]], i32 [[TMP35]])
5286 // CHECK13-NEXT: br label [[OMP_PRECOND_END]]
5287 // CHECK13: omp.precond.end:
5288 // CHECK13-NEXT: ret void
5291 // CHECK13-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l158.omp_outlined.omp_outlined
5292 // CHECK13-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], i64 noundef [[DOTPREVIOUS_LB_:%.*]], i64 noundef [[DOTPREVIOUS_UB_:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[N:%.*]], i64 noundef [[VLA:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[A:%.*]], i64 noundef [[DOTCAPTURE_EXPR_:%.*]]) #[[ATTR2]] {
5293 // CHECK13-NEXT: entry:
5294 // CHECK13-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 8
5295 // CHECK13-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 8
5296 // CHECK13-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i64, align 8
5297 // CHECK13-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i64, align 8
5298 // CHECK13-NEXT: [[N_ADDR:%.*]] = alloca ptr, align 8
5299 // CHECK13-NEXT: [[VLA_ADDR:%.*]] = alloca i64, align 8
5300 // CHECK13-NEXT: [[A_ADDR:%.*]] = alloca ptr, align 8
5301 // CHECK13-NEXT: [[DOTCAPTURE_EXPR__ADDR:%.*]] = alloca i64, align 8
5302 // CHECK13-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
5303 // CHECK13-NEXT: [[TMP:%.*]] = alloca i32, align 4
5304 // CHECK13-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4
5305 // CHECK13-NEXT: [[DOTCAPTURE_EXPR_2:%.*]] = alloca i32, align 4
5306 // CHECK13-NEXT: [[I:%.*]] = alloca i32, align 4
5307 // CHECK13-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4
5308 // CHECK13-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4
5309 // CHECK13-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
5310 // CHECK13-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
5311 // CHECK13-NEXT: [[I5:%.*]] = alloca i32, align 4
5312 // CHECK13-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 8
5313 // CHECK13-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 8
5314 // CHECK13-NEXT: store i64 [[DOTPREVIOUS_LB_]], ptr [[DOTPREVIOUS_LB__ADDR]], align 8
5315 // CHECK13-NEXT: store i64 [[DOTPREVIOUS_UB_]], ptr [[DOTPREVIOUS_UB__ADDR]], align 8
5316 // CHECK13-NEXT: store ptr [[N]], ptr [[N_ADDR]], align 8
5317 // CHECK13-NEXT: store i64 [[VLA]], ptr [[VLA_ADDR]], align 8
5318 // CHECK13-NEXT: store ptr [[A]], ptr [[A_ADDR]], align 8
5319 // CHECK13-NEXT: store i64 [[DOTCAPTURE_EXPR_]], ptr [[DOTCAPTURE_EXPR__ADDR]], align 8
5320 // CHECK13-NEXT: [[TMP0:%.*]] = load ptr, ptr [[N_ADDR]], align 8
5321 // CHECK13-NEXT: [[TMP1:%.*]] = load i64, ptr [[VLA_ADDR]], align 8
5322 // CHECK13-NEXT: [[TMP2:%.*]] = load ptr, ptr [[A_ADDR]], align 8
5323 // CHECK13-NEXT: [[TMP3:%.*]] = load i32, ptr [[TMP0]], align 4
5324 // CHECK13-NEXT: store i32 [[TMP3]], ptr [[DOTCAPTURE_EXPR_1]], align 4
5325 // CHECK13-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4
5326 // CHECK13-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP4]], 0
5327 // CHECK13-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1
5328 // CHECK13-NEXT: [[SUB3:%.*]] = sub nsw i32 [[DIV]], 1
5329 // CHECK13-NEXT: store i32 [[SUB3]], ptr [[DOTCAPTURE_EXPR_2]], align 4
5330 // CHECK13-NEXT: store i32 0, ptr [[I]], align 4
5331 // CHECK13-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4
5332 // CHECK13-NEXT: [[CMP:%.*]] = icmp slt i32 0, [[TMP5]]
5333 // CHECK13-NEXT: br i1 [[CMP]], label [[OMP_PRECOND_THEN:%.*]], label [[OMP_PRECOND_END:%.*]]
5334 // CHECK13: omp.precond.then:
5335 // CHECK13-NEXT: store i32 0, ptr [[DOTOMP_LB]], align 4
5336 // CHECK13-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_2]], align 4
5337 // CHECK13-NEXT: store i32 [[TMP6]], ptr [[DOTOMP_UB]], align 4
5338 // CHECK13-NEXT: [[TMP7:%.*]] = load i64, ptr [[DOTPREVIOUS_LB__ADDR]], align 8
5339 // CHECK13-NEXT: [[CONV:%.*]] = trunc i64 [[TMP7]] to i32
5340 // CHECK13-NEXT: [[TMP8:%.*]] = load i64, ptr [[DOTPREVIOUS_UB__ADDR]], align 8
5341 // CHECK13-NEXT: [[CONV4:%.*]] = trunc i64 [[TMP8]] to i32
5342 // CHECK13-NEXT: store i32 [[CONV]], ptr [[DOTOMP_LB]], align 4
5343 // CHECK13-NEXT: store i32 [[CONV4]], ptr [[DOTOMP_UB]], align 4
5344 // CHECK13-NEXT: store i32 1, ptr [[DOTOMP_STRIDE]], align 4
5345 // CHECK13-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
5346 // CHECK13-NEXT: [[TMP9:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
5347 // CHECK13-NEXT: [[TMP10:%.*]] = load i32, ptr [[TMP9]], align 4
5348 // CHECK13-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB2]], i32 [[TMP10]], i32 34, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
5349 // CHECK13-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
5350 // CHECK13-NEXT: [[TMP12:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_2]], align 4
5351 // CHECK13-NEXT: [[CMP6:%.*]] = icmp sgt i32 [[TMP11]], [[TMP12]]
5352 // CHECK13-NEXT: br i1 [[CMP6]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
5353 // CHECK13: cond.true:
5354 // CHECK13-NEXT: [[TMP13:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_2]], align 4
5355 // CHECK13-NEXT: br label [[COND_END:%.*]]
5356 // CHECK13: cond.false:
5357 // CHECK13-NEXT: [[TMP14:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
5358 // CHECK13-NEXT: br label [[COND_END]]
5359 // CHECK13: cond.end:
5360 // CHECK13-NEXT: [[COND:%.*]] = phi i32 [ [[TMP13]], [[COND_TRUE]] ], [ [[TMP14]], [[COND_FALSE]] ]
5361 // CHECK13-NEXT: store i32 [[COND]], ptr [[DOTOMP_UB]], align 4
5362 // CHECK13-NEXT: [[TMP15:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4
5363 // CHECK13-NEXT: store i32 [[TMP15]], ptr [[DOTOMP_IV]], align 4
5364 // CHECK13-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
5365 // CHECK13: omp.inner.for.cond:
5366 // CHECK13-NEXT: [[TMP16:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
5367 // CHECK13-NEXT: [[TMP17:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
5368 // CHECK13-NEXT: [[CMP7:%.*]] = icmp sle i32 [[TMP16]], [[TMP17]]
5369 // CHECK13-NEXT: br i1 [[CMP7]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
5370 // CHECK13: omp.inner.for.body:
5371 // CHECK13-NEXT: [[TMP18:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
5372 // CHECK13-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP18]], 1
5373 // CHECK13-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
5374 // CHECK13-NEXT: store i32 [[ADD]], ptr [[I5]], align 4
5375 // CHECK13-NEXT: [[TMP19:%.*]] = load i32, ptr [[I5]], align 4
5376 // CHECK13-NEXT: [[IDXPROM:%.*]] = sext i32 [[TMP19]] to i64
5377 // CHECK13-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i32, ptr [[TMP2]], i64 [[IDXPROM]]
5378 // CHECK13-NEXT: store i32 0, ptr [[ARRAYIDX]], align 4
5379 // CHECK13-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
5380 // CHECK13: omp.body.continue:
5381 // CHECK13-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
5382 // CHECK13: omp.inner.for.inc:
5383 // CHECK13-NEXT: [[TMP20:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
5384 // CHECK13-NEXT: [[ADD8:%.*]] = add nsw i32 [[TMP20]], 1
5385 // CHECK13-NEXT: store i32 [[ADD8]], ptr [[DOTOMP_IV]], align 4
5386 // CHECK13-NEXT: br label [[OMP_INNER_FOR_COND]]
5387 // CHECK13: omp.inner.for.end:
5388 // CHECK13-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
5389 // CHECK13: omp.loop.exit:
5390 // CHECK13-NEXT: [[TMP21:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
5391 // CHECK13-NEXT: [[TMP22:%.*]] = load i32, ptr [[TMP21]], align 4
5392 // CHECK13-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB2]], i32 [[TMP22]])
5393 // CHECK13-NEXT: br label [[OMP_PRECOND_END]]
5394 // CHECK13: omp.precond.end:
5395 // CHECK13-NEXT: ret void
5398 // CHECK13-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l163
5399 // CHECK13-SAME: (i64 noundef [[N:%.*]], i64 noundef [[VLA:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[A:%.*]]) #[[ATTR2]] {
5400 // CHECK13-NEXT: entry:
5401 // CHECK13-NEXT: [[N_ADDR:%.*]] = alloca i64, align 8
5402 // CHECK13-NEXT: [[VLA_ADDR:%.*]] = alloca i64, align 8
5403 // CHECK13-NEXT: [[A_ADDR:%.*]] = alloca ptr, align 8
5404 // CHECK13-NEXT: store i64 [[N]], ptr [[N_ADDR]], align 8
5405 // CHECK13-NEXT: store i64 [[VLA]], ptr [[VLA_ADDR]], align 8
5406 // CHECK13-NEXT: store ptr [[A]], ptr [[A_ADDR]], align 8
5407 // CHECK13-NEXT: [[TMP0:%.*]] = load i64, ptr [[VLA_ADDR]], align 8
5408 // CHECK13-NEXT: [[TMP1:%.*]] = load ptr, ptr [[A_ADDR]], align 8
5409 // CHECK13-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_teams(ptr @[[GLOB3]], i32 3, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l163.omp_outlined, ptr [[N_ADDR]], i64 [[TMP0]], ptr [[TMP1]])
5410 // CHECK13-NEXT: ret void
5413 // CHECK13-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l163.omp_outlined
5414 // CHECK13-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[N:%.*]], i64 noundef [[VLA:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[A:%.*]]) #[[ATTR2]] {
5415 // CHECK13-NEXT: entry:
5416 // CHECK13-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 8
5417 // CHECK13-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 8
5418 // CHECK13-NEXT: [[N_ADDR:%.*]] = alloca ptr, align 8
5419 // CHECK13-NEXT: [[VLA_ADDR:%.*]] = alloca i64, align 8
5420 // CHECK13-NEXT: [[A_ADDR:%.*]] = alloca ptr, align 8
5421 // CHECK13-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
5422 // CHECK13-NEXT: [[TMP:%.*]] = alloca i32, align 4
5423 // CHECK13-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4
5424 // CHECK13-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4
5425 // CHECK13-NEXT: [[I:%.*]] = alloca i32, align 4
5426 // CHECK13-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4
5427 // CHECK13-NEXT: [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4
5428 // CHECK13-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
5429 // CHECK13-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
5430 // CHECK13-NEXT: [[I3:%.*]] = alloca i32, align 4
5431 // CHECK13-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 8
5432 // CHECK13-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 8
5433 // CHECK13-NEXT: store ptr [[N]], ptr [[N_ADDR]], align 8
5434 // CHECK13-NEXT: store i64 [[VLA]], ptr [[VLA_ADDR]], align 8
5435 // CHECK13-NEXT: store ptr [[A]], ptr [[A_ADDR]], align 8
5436 // CHECK13-NEXT: [[TMP0:%.*]] = load ptr, ptr [[N_ADDR]], align 8
5437 // CHECK13-NEXT: [[TMP1:%.*]] = load i64, ptr [[VLA_ADDR]], align 8
5438 // CHECK13-NEXT: [[TMP2:%.*]] = load ptr, ptr [[A_ADDR]], align 8
5439 // CHECK13-NEXT: [[TMP3:%.*]] = load i32, ptr [[TMP0]], align 4
5440 // CHECK13-NEXT: store i32 [[TMP3]], ptr [[DOTCAPTURE_EXPR_]], align 4
5441 // CHECK13-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4
5442 // CHECK13-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP4]], 0
5443 // CHECK13-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1
5444 // CHECK13-NEXT: [[SUB2:%.*]] = sub nsw i32 [[DIV]], 1
5445 // CHECK13-NEXT: store i32 [[SUB2]], ptr [[DOTCAPTURE_EXPR_1]], align 4
5446 // CHECK13-NEXT: store i32 0, ptr [[I]], align 4
5447 // CHECK13-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4
5448 // CHECK13-NEXT: [[CMP:%.*]] = icmp slt i32 0, [[TMP5]]
5449 // CHECK13-NEXT: br i1 [[CMP]], label [[OMP_PRECOND_THEN:%.*]], label [[OMP_PRECOND_END:%.*]]
5450 // CHECK13: omp.precond.then:
5451 // CHECK13-NEXT: store i32 0, ptr [[DOTOMP_COMB_LB]], align 4
5452 // CHECK13-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4
5453 // CHECK13-NEXT: store i32 [[TMP6]], ptr [[DOTOMP_COMB_UB]], align 4
5454 // CHECK13-NEXT: store i32 1, ptr [[DOTOMP_STRIDE]], align 4
5455 // CHECK13-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
5456 // CHECK13-NEXT: [[TMP7:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
5457 // CHECK13-NEXT: [[TMP8:%.*]] = load i32, ptr [[TMP7]], align 4
5458 // CHECK13-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB1]], i32 [[TMP8]], i32 92, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
5459 // CHECK13-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
5460 // CHECK13-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4
5461 // CHECK13-NEXT: [[CMP4:%.*]] = icmp sgt i32 [[TMP9]], [[TMP10]]
5462 // CHECK13-NEXT: br i1 [[CMP4]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
5463 // CHECK13: cond.true:
5464 // CHECK13-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4
5465 // CHECK13-NEXT: br label [[COND_END:%.*]]
5466 // CHECK13: cond.false:
5467 // CHECK13-NEXT: [[TMP12:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
5468 // CHECK13-NEXT: br label [[COND_END]]
5469 // CHECK13: cond.end:
5470 // CHECK13-NEXT: [[COND:%.*]] = phi i32 [ [[TMP11]], [[COND_TRUE]] ], [ [[TMP12]], [[COND_FALSE]] ]
5471 // CHECK13-NEXT: store i32 [[COND]], ptr [[DOTOMP_COMB_UB]], align 4
5472 // CHECK13-NEXT: [[TMP13:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
5473 // CHECK13-NEXT: store i32 [[TMP13]], ptr [[DOTOMP_IV]], align 4
5474 // CHECK13-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
5475 // CHECK13: omp.inner.for.cond:
5476 // CHECK13-NEXT: [[TMP14:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
5477 // CHECK13-NEXT: [[TMP15:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
5478 // CHECK13-NEXT: [[CMP5:%.*]] = icmp sle i32 [[TMP14]], [[TMP15]]
5479 // CHECK13-NEXT: br i1 [[CMP5]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
5480 // CHECK13: omp.inner.for.body:
5481 // CHECK13-NEXT: [[TMP16:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
5482 // CHECK13-NEXT: [[TMP17:%.*]] = zext i32 [[TMP16]] to i64
5483 // CHECK13-NEXT: [[TMP18:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
5484 // CHECK13-NEXT: [[TMP19:%.*]] = zext i32 [[TMP18]] to i64
5485 // CHECK13-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_call(ptr @[[GLOB3]], i32 5, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l163.omp_outlined.omp_outlined, i64 [[TMP17]], i64 [[TMP19]], ptr [[TMP0]], i64 [[TMP1]], ptr [[TMP2]])
5486 // CHECK13-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
5487 // CHECK13: omp.inner.for.inc:
5488 // CHECK13-NEXT: [[TMP20:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
5489 // CHECK13-NEXT: [[TMP21:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4
5490 // CHECK13-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP20]], [[TMP21]]
5491 // CHECK13-NEXT: store i32 [[ADD]], ptr [[DOTOMP_IV]], align 4
5492 // CHECK13-NEXT: br label [[OMP_INNER_FOR_COND]]
5493 // CHECK13: omp.inner.for.end:
5494 // CHECK13-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
5495 // CHECK13: omp.loop.exit:
5496 // CHECK13-NEXT: [[TMP22:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
5497 // CHECK13-NEXT: [[TMP23:%.*]] = load i32, ptr [[TMP22]], align 4
5498 // CHECK13-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB1]], i32 [[TMP23]])
5499 // CHECK13-NEXT: br label [[OMP_PRECOND_END]]
5500 // CHECK13: omp.precond.end:
5501 // CHECK13-NEXT: ret void
5504 // CHECK13-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l163.omp_outlined.omp_outlined
5505 // CHECK13-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], i64 noundef [[DOTPREVIOUS_LB_:%.*]], i64 noundef [[DOTPREVIOUS_UB_:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[N:%.*]], i64 noundef [[VLA:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[A:%.*]]) #[[ATTR2]] {
5506 // CHECK13-NEXT: entry:
5507 // CHECK13-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 8
5508 // CHECK13-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 8
5509 // CHECK13-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i64, align 8
5510 // CHECK13-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i64, align 8
5511 // CHECK13-NEXT: [[N_ADDR:%.*]] = alloca ptr, align 8
5512 // CHECK13-NEXT: [[VLA_ADDR:%.*]] = alloca i64, align 8
5513 // CHECK13-NEXT: [[A_ADDR:%.*]] = alloca ptr, align 8
5514 // CHECK13-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
5515 // CHECK13-NEXT: [[TMP:%.*]] = alloca i32, align 4
5516 // CHECK13-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4
5517 // CHECK13-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4
5518 // CHECK13-NEXT: [[I:%.*]] = alloca i32, align 4
5519 // CHECK13-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4
5520 // CHECK13-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4
5521 // CHECK13-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
5522 // CHECK13-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
5523 // CHECK13-NEXT: [[I4:%.*]] = alloca i32, align 4
5524 // CHECK13-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 8
5525 // CHECK13-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 8
5526 // CHECK13-NEXT: store i64 [[DOTPREVIOUS_LB_]], ptr [[DOTPREVIOUS_LB__ADDR]], align 8
5527 // CHECK13-NEXT: store i64 [[DOTPREVIOUS_UB_]], ptr [[DOTPREVIOUS_UB__ADDR]], align 8
5528 // CHECK13-NEXT: store ptr [[N]], ptr [[N_ADDR]], align 8
5529 // CHECK13-NEXT: store i64 [[VLA]], ptr [[VLA_ADDR]], align 8
5530 // CHECK13-NEXT: store ptr [[A]], ptr [[A_ADDR]], align 8
5531 // CHECK13-NEXT: [[TMP0:%.*]] = load ptr, ptr [[N_ADDR]], align 8
5532 // CHECK13-NEXT: [[TMP1:%.*]] = load i64, ptr [[VLA_ADDR]], align 8
5533 // CHECK13-NEXT: [[TMP2:%.*]] = load ptr, ptr [[A_ADDR]], align 8
5534 // CHECK13-NEXT: [[TMP3:%.*]] = load i32, ptr [[TMP0]], align 4
5535 // CHECK13-NEXT: store i32 [[TMP3]], ptr [[DOTCAPTURE_EXPR_]], align 4
5536 // CHECK13-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4
5537 // CHECK13-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP4]], 0
5538 // CHECK13-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1
5539 // CHECK13-NEXT: [[SUB2:%.*]] = sub nsw i32 [[DIV]], 1
5540 // CHECK13-NEXT: store i32 [[SUB2]], ptr [[DOTCAPTURE_EXPR_1]], align 4
5541 // CHECK13-NEXT: store i32 0, ptr [[I]], align 4
5542 // CHECK13-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4
5543 // CHECK13-NEXT: [[CMP:%.*]] = icmp slt i32 0, [[TMP5]]
5544 // CHECK13-NEXT: br i1 [[CMP]], label [[OMP_PRECOND_THEN:%.*]], label [[OMP_PRECOND_END:%.*]]
5545 // CHECK13: omp.precond.then:
5546 // CHECK13-NEXT: store i32 0, ptr [[DOTOMP_LB]], align 4
5547 // CHECK13-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4
5548 // CHECK13-NEXT: store i32 [[TMP6]], ptr [[DOTOMP_UB]], align 4
5549 // CHECK13-NEXT: [[TMP7:%.*]] = load i64, ptr [[DOTPREVIOUS_LB__ADDR]], align 8
5550 // CHECK13-NEXT: [[CONV:%.*]] = trunc i64 [[TMP7]] to i32
5551 // CHECK13-NEXT: [[TMP8:%.*]] = load i64, ptr [[DOTPREVIOUS_UB__ADDR]], align 8
5552 // CHECK13-NEXT: [[CONV3:%.*]] = trunc i64 [[TMP8]] to i32
5553 // CHECK13-NEXT: store i32 [[CONV]], ptr [[DOTOMP_LB]], align 4
5554 // CHECK13-NEXT: store i32 [[CONV3]], ptr [[DOTOMP_UB]], align 4
5555 // CHECK13-NEXT: store i32 1, ptr [[DOTOMP_STRIDE]], align 4
5556 // CHECK13-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
5557 // CHECK13-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4
5558 // CHECK13-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
5559 // CHECK13-NEXT: [[TMP11:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
5560 // CHECK13-NEXT: [[TMP12:%.*]] = load i32, ptr [[TMP11]], align 4
5561 // CHECK13-NEXT: call void @__kmpc_dispatch_init_4(ptr @[[GLOB3]], i32 [[TMP12]], i32 1073741859, i32 [[TMP9]], i32 [[TMP10]], i32 1, i32 1)
5562 // CHECK13-NEXT: br label [[OMP_DISPATCH_COND:%.*]]
5563 // CHECK13: omp.dispatch.cond:
5564 // CHECK13-NEXT: [[TMP13:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
5565 // CHECK13-NEXT: [[TMP14:%.*]] = load i32, ptr [[TMP13]], align 4
5566 // CHECK13-NEXT: [[TMP15:%.*]] = call i32 @__kmpc_dispatch_next_4(ptr @[[GLOB3]], i32 [[TMP14]], ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]])
5567 // CHECK13-NEXT: [[TOBOOL:%.*]] = icmp ne i32 [[TMP15]], 0
5568 // CHECK13-NEXT: br i1 [[TOBOOL]], label [[OMP_DISPATCH_BODY:%.*]], label [[OMP_DISPATCH_END:%.*]]
5569 // CHECK13: omp.dispatch.body:
5570 // CHECK13-NEXT: [[TMP16:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4
5571 // CHECK13-NEXT: store i32 [[TMP16]], ptr [[DOTOMP_IV]], align 4
5572 // CHECK13-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
5573 // CHECK13: omp.inner.for.cond:
5574 // CHECK13-NEXT: [[TMP17:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP15:![0-9]+]]
5575 // CHECK13-NEXT: [[TMP18:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP15]]
5576 // CHECK13-NEXT: [[CMP5:%.*]] = icmp sle i32 [[TMP17]], [[TMP18]]
5577 // CHECK13-NEXT: br i1 [[CMP5]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
5578 // CHECK13: omp.inner.for.body:
5579 // CHECK13-NEXT: [[TMP19:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP15]]
5580 // CHECK13-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP19]], 1
5581 // CHECK13-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
5582 // CHECK13-NEXT: store i32 [[ADD]], ptr [[I4]], align 4, !llvm.access.group [[ACC_GRP15]]
5583 // CHECK13-NEXT: [[TMP20:%.*]] = load i32, ptr [[I4]], align 4, !llvm.access.group [[ACC_GRP15]]
5584 // CHECK13-NEXT: [[IDXPROM:%.*]] = sext i32 [[TMP20]] to i64
5585 // CHECK13-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i32, ptr [[TMP2]], i64 [[IDXPROM]]
5586 // CHECK13-NEXT: store i32 0, ptr [[ARRAYIDX]], align 4, !llvm.access.group [[ACC_GRP15]]
5587 // CHECK13-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
5588 // CHECK13: omp.body.continue:
5589 // CHECK13-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
5590 // CHECK13: omp.inner.for.inc:
5591 // CHECK13-NEXT: [[TMP21:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP15]]
5592 // CHECK13-NEXT: [[ADD6:%.*]] = add nsw i32 [[TMP21]], 1
5593 // CHECK13-NEXT: store i32 [[ADD6]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP15]]
5594 // CHECK13-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP16:![0-9]+]]
5595 // CHECK13: omp.inner.for.end:
5596 // CHECK13-NEXT: br label [[OMP_DISPATCH_INC:%.*]]
5597 // CHECK13: omp.dispatch.inc:
5598 // CHECK13-NEXT: br label [[OMP_DISPATCH_COND]]
5599 // CHECK13: omp.dispatch.end:
5600 // CHECK13-NEXT: [[TMP22:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
5601 // CHECK13-NEXT: [[TMP23:%.*]] = load i32, ptr [[TMP22]], align 4
5602 // CHECK13-NEXT: call void @__kmpc_dispatch_deinit(ptr @[[GLOB3]], i32 [[TMP23]])
5603 // CHECK13-NEXT: br label [[OMP_PRECOND_END]]
5604 // CHECK13: omp.precond.end:
5605 // CHECK13-NEXT: ret void
5608 // CHECK13-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l168
5609 // CHECK13-SAME: (i64 noundef [[M:%.*]], i64 noundef [[N:%.*]], i64 noundef [[VLA:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[A:%.*]]) #[[ATTR2]] {
5610 // CHECK13-NEXT: entry:
5611 // CHECK13-NEXT: [[M_ADDR:%.*]] = alloca i64, align 8
5612 // CHECK13-NEXT: [[N_ADDR:%.*]] = alloca i64, align 8
5613 // CHECK13-NEXT: [[VLA_ADDR:%.*]] = alloca i64, align 8
5614 // CHECK13-NEXT: [[A_ADDR:%.*]] = alloca ptr, align 8
5615 // CHECK13-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4
5616 // CHECK13-NEXT: [[DOTCAPTURE_EXPR__CASTED:%.*]] = alloca i64, align 8
5617 // CHECK13-NEXT: store i64 [[M]], ptr [[M_ADDR]], align 8
5618 // CHECK13-NEXT: store i64 [[N]], ptr [[N_ADDR]], align 8
5619 // CHECK13-NEXT: store i64 [[VLA]], ptr [[VLA_ADDR]], align 8
5620 // CHECK13-NEXT: store ptr [[A]], ptr [[A_ADDR]], align 8
5621 // CHECK13-NEXT: [[TMP0:%.*]] = load i64, ptr [[VLA_ADDR]], align 8
5622 // CHECK13-NEXT: [[TMP1:%.*]] = load ptr, ptr [[A_ADDR]], align 8
5623 // CHECK13-NEXT: [[TMP2:%.*]] = load i32, ptr [[M_ADDR]], align 4
5624 // CHECK13-NEXT: store i32 [[TMP2]], ptr [[DOTCAPTURE_EXPR_]], align 4
5625 // CHECK13-NEXT: [[TMP3:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4
5626 // CHECK13-NEXT: store i32 [[TMP3]], ptr [[DOTCAPTURE_EXPR__CASTED]], align 4
5627 // CHECK13-NEXT: [[TMP4:%.*]] = load i64, ptr [[DOTCAPTURE_EXPR__CASTED]], align 8
5628 // CHECK13-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_teams(ptr @[[GLOB3]], i32 4, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l168.omp_outlined, ptr [[N_ADDR]], i64 [[TMP0]], ptr [[TMP1]], i64 [[TMP4]])
5629 // CHECK13-NEXT: ret void
5632 // CHECK13-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l168.omp_outlined
5633 // CHECK13-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[N:%.*]], i64 noundef [[VLA:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[A:%.*]], i64 noundef [[DOTCAPTURE_EXPR_:%.*]]) #[[ATTR2]] {
5634 // CHECK13-NEXT: entry:
5635 // CHECK13-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 8
5636 // CHECK13-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 8
5637 // CHECK13-NEXT: [[N_ADDR:%.*]] = alloca ptr, align 8
5638 // CHECK13-NEXT: [[VLA_ADDR:%.*]] = alloca i64, align 8
5639 // CHECK13-NEXT: [[A_ADDR:%.*]] = alloca ptr, align 8
5640 // CHECK13-NEXT: [[DOTCAPTURE_EXPR__ADDR:%.*]] = alloca i64, align 8
5641 // CHECK13-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
5642 // CHECK13-NEXT: [[TMP:%.*]] = alloca i32, align 4
5643 // CHECK13-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4
5644 // CHECK13-NEXT: [[DOTCAPTURE_EXPR_2:%.*]] = alloca i32, align 4
5645 // CHECK13-NEXT: [[I:%.*]] = alloca i32, align 4
5646 // CHECK13-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4
5647 // CHECK13-NEXT: [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4
5648 // CHECK13-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
5649 // CHECK13-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
5650 // CHECK13-NEXT: [[I4:%.*]] = alloca i32, align 4
5651 // CHECK13-NEXT: [[DOTCAPTURE_EXPR__CASTED:%.*]] = alloca i64, align 8
5652 // CHECK13-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 8
5653 // CHECK13-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 8
5654 // CHECK13-NEXT: store ptr [[N]], ptr [[N_ADDR]], align 8
5655 // CHECK13-NEXT: store i64 [[VLA]], ptr [[VLA_ADDR]], align 8
5656 // CHECK13-NEXT: store ptr [[A]], ptr [[A_ADDR]], align 8
5657 // CHECK13-NEXT: store i64 [[DOTCAPTURE_EXPR_]], ptr [[DOTCAPTURE_EXPR__ADDR]], align 8
5658 // CHECK13-NEXT: [[TMP0:%.*]] = load ptr, ptr [[N_ADDR]], align 8
5659 // CHECK13-NEXT: [[TMP1:%.*]] = load i64, ptr [[VLA_ADDR]], align 8
5660 // CHECK13-NEXT: [[TMP2:%.*]] = load ptr, ptr [[A_ADDR]], align 8
5661 // CHECK13-NEXT: [[TMP3:%.*]] = load i32, ptr [[TMP0]], align 4
5662 // CHECK13-NEXT: store i32 [[TMP3]], ptr [[DOTCAPTURE_EXPR_1]], align 4
5663 // CHECK13-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4
5664 // CHECK13-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP4]], 0
5665 // CHECK13-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1
5666 // CHECK13-NEXT: [[SUB3:%.*]] = sub nsw i32 [[DIV]], 1
5667 // CHECK13-NEXT: store i32 [[SUB3]], ptr [[DOTCAPTURE_EXPR_2]], align 4
5668 // CHECK13-NEXT: store i32 0, ptr [[I]], align 4
5669 // CHECK13-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4
5670 // CHECK13-NEXT: [[CMP:%.*]] = icmp slt i32 0, [[TMP5]]
5671 // CHECK13-NEXT: br i1 [[CMP]], label [[OMP_PRECOND_THEN:%.*]], label [[OMP_PRECOND_END:%.*]]
5672 // CHECK13: omp.precond.then:
5673 // CHECK13-NEXT: store i32 0, ptr [[DOTOMP_COMB_LB]], align 4
5674 // CHECK13-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_2]], align 4
5675 // CHECK13-NEXT: store i32 [[TMP6]], ptr [[DOTOMP_COMB_UB]], align 4
5676 // CHECK13-NEXT: store i32 1, ptr [[DOTOMP_STRIDE]], align 4
5677 // CHECK13-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
5678 // CHECK13-NEXT: [[TMP7:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
5679 // CHECK13-NEXT: [[TMP8:%.*]] = load i32, ptr [[TMP7]], align 4
5680 // CHECK13-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB1]], i32 [[TMP8]], i32 92, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
5681 // CHECK13-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
5682 // CHECK13-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_2]], align 4
5683 // CHECK13-NEXT: [[CMP5:%.*]] = icmp sgt i32 [[TMP9]], [[TMP10]]
5684 // CHECK13-NEXT: br i1 [[CMP5]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
5685 // CHECK13: cond.true:
5686 // CHECK13-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_2]], align 4
5687 // CHECK13-NEXT: br label [[COND_END:%.*]]
5688 // CHECK13: cond.false:
5689 // CHECK13-NEXT: [[TMP12:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
5690 // CHECK13-NEXT: br label [[COND_END]]
5691 // CHECK13: cond.end:
5692 // CHECK13-NEXT: [[COND:%.*]] = phi i32 [ [[TMP11]], [[COND_TRUE]] ], [ [[TMP12]], [[COND_FALSE]] ]
5693 // CHECK13-NEXT: store i32 [[COND]], ptr [[DOTOMP_COMB_UB]], align 4
5694 // CHECK13-NEXT: [[TMP13:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
5695 // CHECK13-NEXT: store i32 [[TMP13]], ptr [[DOTOMP_IV]], align 4
5696 // CHECK13-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
5697 // CHECK13: omp.inner.for.cond:
5698 // CHECK13-NEXT: [[TMP14:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
5699 // CHECK13-NEXT: [[TMP15:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
5700 // CHECK13-NEXT: [[CMP6:%.*]] = icmp sle i32 [[TMP14]], [[TMP15]]
5701 // CHECK13-NEXT: br i1 [[CMP6]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
5702 // CHECK13: omp.inner.for.body:
5703 // CHECK13-NEXT: [[TMP16:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
5704 // CHECK13-NEXT: [[TMP17:%.*]] = zext i32 [[TMP16]] to i64
5705 // CHECK13-NEXT: [[TMP18:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
5706 // CHECK13-NEXT: [[TMP19:%.*]] = zext i32 [[TMP18]] to i64
5707 // CHECK13-NEXT: [[TMP20:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR__ADDR]], align 4
5708 // CHECK13-NEXT: store i32 [[TMP20]], ptr [[DOTCAPTURE_EXPR__CASTED]], align 4
5709 // CHECK13-NEXT: [[TMP21:%.*]] = load i64, ptr [[DOTCAPTURE_EXPR__CASTED]], align 8
5710 // CHECK13-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_call(ptr @[[GLOB3]], i32 6, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l168.omp_outlined.omp_outlined, i64 [[TMP17]], i64 [[TMP19]], ptr [[TMP0]], i64 [[TMP1]], ptr [[TMP2]], i64 [[TMP21]])
5711 // CHECK13-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
5712 // CHECK13: omp.inner.for.inc:
5713 // CHECK13-NEXT: [[TMP22:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
5714 // CHECK13-NEXT: [[TMP23:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4
5715 // CHECK13-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP22]], [[TMP23]]
5716 // CHECK13-NEXT: store i32 [[ADD]], ptr [[DOTOMP_IV]], align 4
5717 // CHECK13-NEXT: br label [[OMP_INNER_FOR_COND]]
5718 // CHECK13: omp.inner.for.end:
5719 // CHECK13-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
5720 // CHECK13: omp.loop.exit:
5721 // CHECK13-NEXT: [[TMP24:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
5722 // CHECK13-NEXT: [[TMP25:%.*]] = load i32, ptr [[TMP24]], align 4
5723 // CHECK13-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB1]], i32 [[TMP25]])
5724 // CHECK13-NEXT: br label [[OMP_PRECOND_END]]
5725 // CHECK13: omp.precond.end:
5726 // CHECK13-NEXT: ret void
5729 // CHECK13-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l168.omp_outlined.omp_outlined
5730 // CHECK13-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], i64 noundef [[DOTPREVIOUS_LB_:%.*]], i64 noundef [[DOTPREVIOUS_UB_:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[N:%.*]], i64 noundef [[VLA:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[A:%.*]], i64 noundef [[DOTCAPTURE_EXPR_:%.*]]) #[[ATTR2]] {
5731 // CHECK13-NEXT: entry:
5732 // CHECK13-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 8
5733 // CHECK13-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 8
5734 // CHECK13-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i64, align 8
5735 // CHECK13-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i64, align 8
5736 // CHECK13-NEXT: [[N_ADDR:%.*]] = alloca ptr, align 8
5737 // CHECK13-NEXT: [[VLA_ADDR:%.*]] = alloca i64, align 8
5738 // CHECK13-NEXT: [[A_ADDR:%.*]] = alloca ptr, align 8
5739 // CHECK13-NEXT: [[DOTCAPTURE_EXPR__ADDR:%.*]] = alloca i64, align 8
5740 // CHECK13-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
5741 // CHECK13-NEXT: [[TMP:%.*]] = alloca i32, align 4
5742 // CHECK13-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4
5743 // CHECK13-NEXT: [[DOTCAPTURE_EXPR_2:%.*]] = alloca i32, align 4
5744 // CHECK13-NEXT: [[I:%.*]] = alloca i32, align 4
5745 // CHECK13-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4
5746 // CHECK13-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4
5747 // CHECK13-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
5748 // CHECK13-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
5749 // CHECK13-NEXT: [[I5:%.*]] = alloca i32, align 4
5750 // CHECK13-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 8
5751 // CHECK13-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 8
5752 // CHECK13-NEXT: store i64 [[DOTPREVIOUS_LB_]], ptr [[DOTPREVIOUS_LB__ADDR]], align 8
5753 // CHECK13-NEXT: store i64 [[DOTPREVIOUS_UB_]], ptr [[DOTPREVIOUS_UB__ADDR]], align 8
5754 // CHECK13-NEXT: store ptr [[N]], ptr [[N_ADDR]], align 8
5755 // CHECK13-NEXT: store i64 [[VLA]], ptr [[VLA_ADDR]], align 8
5756 // CHECK13-NEXT: store ptr [[A]], ptr [[A_ADDR]], align 8
5757 // CHECK13-NEXT: store i64 [[DOTCAPTURE_EXPR_]], ptr [[DOTCAPTURE_EXPR__ADDR]], align 8
5758 // CHECK13-NEXT: [[TMP0:%.*]] = load ptr, ptr [[N_ADDR]], align 8
5759 // CHECK13-NEXT: [[TMP1:%.*]] = load i64, ptr [[VLA_ADDR]], align 8
5760 // CHECK13-NEXT: [[TMP2:%.*]] = load ptr, ptr [[A_ADDR]], align 8
5761 // CHECK13-NEXT: [[TMP3:%.*]] = load i32, ptr [[TMP0]], align 4
5762 // CHECK13-NEXT: store i32 [[TMP3]], ptr [[DOTCAPTURE_EXPR_1]], align 4
5763 // CHECK13-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4
5764 // CHECK13-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP4]], 0
5765 // CHECK13-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1
5766 // CHECK13-NEXT: [[SUB3:%.*]] = sub nsw i32 [[DIV]], 1
5767 // CHECK13-NEXT: store i32 [[SUB3]], ptr [[DOTCAPTURE_EXPR_2]], align 4
5768 // CHECK13-NEXT: store i32 0, ptr [[I]], align 4
5769 // CHECK13-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4
5770 // CHECK13-NEXT: [[CMP:%.*]] = icmp slt i32 0, [[TMP5]]
5771 // CHECK13-NEXT: br i1 [[CMP]], label [[OMP_PRECOND_THEN:%.*]], label [[OMP_PRECOND_END:%.*]]
5772 // CHECK13: omp.precond.then:
5773 // CHECK13-NEXT: store i32 0, ptr [[DOTOMP_LB]], align 4
5774 // CHECK13-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_2]], align 4
5775 // CHECK13-NEXT: store i32 [[TMP6]], ptr [[DOTOMP_UB]], align 4
5776 // CHECK13-NEXT: [[TMP7:%.*]] = load i64, ptr [[DOTPREVIOUS_LB__ADDR]], align 8
5777 // CHECK13-NEXT: [[CONV:%.*]] = trunc i64 [[TMP7]] to i32
5778 // CHECK13-NEXT: [[TMP8:%.*]] = load i64, ptr [[DOTPREVIOUS_UB__ADDR]], align 8
5779 // CHECK13-NEXT: [[CONV4:%.*]] = trunc i64 [[TMP8]] to i32
5780 // CHECK13-NEXT: store i32 [[CONV]], ptr [[DOTOMP_LB]], align 4
5781 // CHECK13-NEXT: store i32 [[CONV4]], ptr [[DOTOMP_UB]], align 4
5782 // CHECK13-NEXT: store i32 1, ptr [[DOTOMP_STRIDE]], align 4
5783 // CHECK13-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
5784 // CHECK13-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR__ADDR]], align 4
5785 // CHECK13-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4
5786 // CHECK13-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
5787 // CHECK13-NEXT: [[TMP12:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
5788 // CHECK13-NEXT: [[TMP13:%.*]] = load i32, ptr [[TMP12]], align 4
5789 // CHECK13-NEXT: call void @__kmpc_dispatch_init_4(ptr @[[GLOB3]], i32 [[TMP13]], i32 1073741859, i32 [[TMP10]], i32 [[TMP11]], i32 1, i32 [[TMP9]])
5790 // CHECK13-NEXT: br label [[OMP_DISPATCH_COND:%.*]]
5791 // CHECK13: omp.dispatch.cond:
5792 // CHECK13-NEXT: [[TMP14:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
5793 // CHECK13-NEXT: [[TMP15:%.*]] = load i32, ptr [[TMP14]], align 4
5794 // CHECK13-NEXT: [[TMP16:%.*]] = call i32 @__kmpc_dispatch_next_4(ptr @[[GLOB3]], i32 [[TMP15]], ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]])
5795 // CHECK13-NEXT: [[TOBOOL:%.*]] = icmp ne i32 [[TMP16]], 0
5796 // CHECK13-NEXT: br i1 [[TOBOOL]], label [[OMP_DISPATCH_BODY:%.*]], label [[OMP_DISPATCH_END:%.*]]
5797 // CHECK13: omp.dispatch.body:
5798 // CHECK13-NEXT: [[TMP17:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4
5799 // CHECK13-NEXT: store i32 [[TMP17]], ptr [[DOTOMP_IV]], align 4
5800 // CHECK13-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
5801 // CHECK13: omp.inner.for.cond:
5802 // CHECK13-NEXT: [[TMP18:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP18:![0-9]+]]
5803 // CHECK13-NEXT: [[TMP19:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP18]]
5804 // CHECK13-NEXT: [[CMP6:%.*]] = icmp sle i32 [[TMP18]], [[TMP19]]
5805 // CHECK13-NEXT: br i1 [[CMP6]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
5806 // CHECK13: omp.inner.for.body:
5807 // CHECK13-NEXT: [[TMP20:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP18]]
5808 // CHECK13-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP20]], 1
5809 // CHECK13-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
5810 // CHECK13-NEXT: store i32 [[ADD]], ptr [[I5]], align 4, !llvm.access.group [[ACC_GRP18]]
5811 // CHECK13-NEXT: [[TMP21:%.*]] = load i32, ptr [[I5]], align 4, !llvm.access.group [[ACC_GRP18]]
5812 // CHECK13-NEXT: [[IDXPROM:%.*]] = sext i32 [[TMP21]] to i64
5813 // CHECK13-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i32, ptr [[TMP2]], i64 [[IDXPROM]]
5814 // CHECK13-NEXT: store i32 0, ptr [[ARRAYIDX]], align 4, !llvm.access.group [[ACC_GRP18]]
5815 // CHECK13-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
5816 // CHECK13: omp.body.continue:
5817 // CHECK13-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
5818 // CHECK13: omp.inner.for.inc:
5819 // CHECK13-NEXT: [[TMP22:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP18]]
5820 // CHECK13-NEXT: [[ADD7:%.*]] = add nsw i32 [[TMP22]], 1
5821 // CHECK13-NEXT: store i32 [[ADD7]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP18]]
5822 // CHECK13-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP19:![0-9]+]]
5823 // CHECK13: omp.inner.for.end:
5824 // CHECK13-NEXT: br label [[OMP_DISPATCH_INC:%.*]]
5825 // CHECK13: omp.dispatch.inc:
5826 // CHECK13-NEXT: br label [[OMP_DISPATCH_COND]]
5827 // CHECK13: omp.dispatch.end:
5828 // CHECK13-NEXT: [[TMP23:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
5829 // CHECK13-NEXT: [[TMP24:%.*]] = load i32, ptr [[TMP23]], align 4
5830 // CHECK13-NEXT: call void @__kmpc_dispatch_deinit(ptr @[[GLOB3]], i32 [[TMP24]])
5831 // CHECK13-NEXT: br label [[OMP_PRECOND_END]]
5832 // CHECK13: omp.precond.end:
5833 // CHECK13-NEXT: ret void
5836 // CHECK13-LABEL: define {{[^@]+}}@_Z5tmainIiLi10EEiT_
5837 // CHECK13-SAME: (i32 noundef signext [[ARGC:%.*]]) #[[ATTR5:[0-9]+]] comdat {
5838 // CHECK13-NEXT: entry:
5839 // CHECK13-NEXT: [[ARGC_ADDR:%.*]] = alloca i32, align 4
5840 // CHECK13-NEXT: [[A:%.*]] = alloca [10 x i32], align 4
5841 // CHECK13-NEXT: [[M:%.*]] = alloca i32, align 4
5842 // CHECK13-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [1 x ptr], align 8
5843 // CHECK13-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [1 x ptr], align 8
5844 // CHECK13-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [1 x ptr], align 8
5845 // CHECK13-NEXT: [[TMP:%.*]] = alloca i32, align 4
5846 // CHECK13-NEXT: [[KERNEL_ARGS:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS:%.*]], align 8
5847 // CHECK13-NEXT: [[DOTOFFLOAD_BASEPTRS1:%.*]] = alloca [1 x ptr], align 8
5848 // CHECK13-NEXT: [[DOTOFFLOAD_PTRS2:%.*]] = alloca [1 x ptr], align 8
5849 // CHECK13-NEXT: [[DOTOFFLOAD_MAPPERS3:%.*]] = alloca [1 x ptr], align 8
5850 // CHECK13-NEXT: [[_TMP4:%.*]] = alloca i32, align 4
5851 // CHECK13-NEXT: [[KERNEL_ARGS5:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS]], align 8
5852 // CHECK13-NEXT: [[M_CASTED:%.*]] = alloca i64, align 8
5853 // CHECK13-NEXT: [[DOTOFFLOAD_BASEPTRS8:%.*]] = alloca [2 x ptr], align 8
5854 // CHECK13-NEXT: [[DOTOFFLOAD_PTRS9:%.*]] = alloca [2 x ptr], align 8
5855 // CHECK13-NEXT: [[DOTOFFLOAD_MAPPERS10:%.*]] = alloca [2 x ptr], align 8
5856 // CHECK13-NEXT: [[_TMP11:%.*]] = alloca i32, align 4
5857 // CHECK13-NEXT: [[KERNEL_ARGS12:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS]], align 8
5858 // CHECK13-NEXT: [[DOTOFFLOAD_BASEPTRS15:%.*]] = alloca [1 x ptr], align 8
5859 // CHECK13-NEXT: [[DOTOFFLOAD_PTRS16:%.*]] = alloca [1 x ptr], align 8
5860 // CHECK13-NEXT: [[DOTOFFLOAD_MAPPERS17:%.*]] = alloca [1 x ptr], align 8
5861 // CHECK13-NEXT: [[_TMP18:%.*]] = alloca i32, align 4
5862 // CHECK13-NEXT: [[KERNEL_ARGS19:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS]], align 8
5863 // CHECK13-NEXT: [[M_CASTED22:%.*]] = alloca i64, align 8
5864 // CHECK13-NEXT: [[DOTOFFLOAD_BASEPTRS23:%.*]] = alloca [2 x ptr], align 8
5865 // CHECK13-NEXT: [[DOTOFFLOAD_PTRS24:%.*]] = alloca [2 x ptr], align 8
5866 // CHECK13-NEXT: [[DOTOFFLOAD_MAPPERS25:%.*]] = alloca [2 x ptr], align 8
5867 // CHECK13-NEXT: [[_TMP26:%.*]] = alloca i32, align 4
5868 // CHECK13-NEXT: [[KERNEL_ARGS27:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS]], align 8
5869 // CHECK13-NEXT: store i32 [[ARGC]], ptr [[ARGC_ADDR]], align 4
5870 // CHECK13-NEXT: store i32 10, ptr [[M]], align 4
5871 // CHECK13-NEXT: [[TMP0:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
5872 // CHECK13-NEXT: store ptr [[A]], ptr [[TMP0]], align 8
5873 // CHECK13-NEXT: [[TMP1:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
5874 // CHECK13-NEXT: store ptr [[A]], ptr [[TMP1]], align 8
5875 // CHECK13-NEXT: [[TMP2:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0
5876 // CHECK13-NEXT: store ptr null, ptr [[TMP2]], align 8
5877 // CHECK13-NEXT: [[TMP3:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
5878 // CHECK13-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
5879 // CHECK13-NEXT: [[TMP5:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 0
5880 // CHECK13-NEXT: store i32 3, ptr [[TMP5]], align 4
5881 // CHECK13-NEXT: [[TMP6:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 1
5882 // CHECK13-NEXT: store i32 1, ptr [[TMP6]], align 4
5883 // CHECK13-NEXT: [[TMP7:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 2
5884 // CHECK13-NEXT: store ptr [[TMP3]], ptr [[TMP7]], align 8
5885 // CHECK13-NEXT: [[TMP8:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 3
5886 // CHECK13-NEXT: store ptr [[TMP4]], ptr [[TMP8]], align 8
5887 // CHECK13-NEXT: [[TMP9:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 4
5888 // CHECK13-NEXT: store ptr @.offload_sizes.9, ptr [[TMP9]], align 8
5889 // CHECK13-NEXT: [[TMP10:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 5
5890 // CHECK13-NEXT: store ptr @.offload_maptypes.10, ptr [[TMP10]], align 8
5891 // CHECK13-NEXT: [[TMP11:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 6
5892 // CHECK13-NEXT: store ptr null, ptr [[TMP11]], align 8
5893 // CHECK13-NEXT: [[TMP12:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 7
5894 // CHECK13-NEXT: store ptr null, ptr [[TMP12]], align 8
5895 // CHECK13-NEXT: [[TMP13:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 8
5896 // CHECK13-NEXT: store i64 10, ptr [[TMP13]], align 8
5897 // CHECK13-NEXT: [[TMP14:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 9
5898 // CHECK13-NEXT: store i64 0, ptr [[TMP14]], align 8
5899 // CHECK13-NEXT: [[TMP15:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 10
5900 // CHECK13-NEXT: store [3 x i32] zeroinitializer, ptr [[TMP15]], align 4
5901 // CHECK13-NEXT: [[TMP16:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 11
5902 // CHECK13-NEXT: store [3 x i32] zeroinitializer, ptr [[TMP16]], align 4
5903 // CHECK13-NEXT: [[TMP17:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 12
5904 // CHECK13-NEXT: store i32 0, ptr [[TMP17]], align 4
5905 // CHECK13-NEXT: [[TMP18:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB3]], i64 -1, i32 0, i32 0, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l116.region_id, ptr [[KERNEL_ARGS]])
5906 // CHECK13-NEXT: [[TMP19:%.*]] = icmp ne i32 [[TMP18]], 0
5907 // CHECK13-NEXT: br i1 [[TMP19]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
5908 // CHECK13: omp_offload.failed:
5909 // CHECK13-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l116(ptr [[A]]) #[[ATTR3]]
5910 // CHECK13-NEXT: br label [[OMP_OFFLOAD_CONT]]
5911 // CHECK13: omp_offload.cont:
5912 // CHECK13-NEXT: [[TMP20:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS1]], i32 0, i32 0
5913 // CHECK13-NEXT: store ptr [[A]], ptr [[TMP20]], align 8
5914 // CHECK13-NEXT: [[TMP21:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS2]], i32 0, i32 0
5915 // CHECK13-NEXT: store ptr [[A]], ptr [[TMP21]], align 8
5916 // CHECK13-NEXT: [[TMP22:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_MAPPERS3]], i64 0, i64 0
5917 // CHECK13-NEXT: store ptr null, ptr [[TMP22]], align 8
5918 // CHECK13-NEXT: [[TMP23:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS1]], i32 0, i32 0
5919 // CHECK13-NEXT: [[TMP24:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS2]], i32 0, i32 0
5920 // CHECK13-NEXT: [[TMP25:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS5]], i32 0, i32 0
5921 // CHECK13-NEXT: store i32 3, ptr [[TMP25]], align 4
5922 // CHECK13-NEXT: [[TMP26:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS5]], i32 0, i32 1
5923 // CHECK13-NEXT: store i32 1, ptr [[TMP26]], align 4
5924 // CHECK13-NEXT: [[TMP27:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS5]], i32 0, i32 2
5925 // CHECK13-NEXT: store ptr [[TMP23]], ptr [[TMP27]], align 8
5926 // CHECK13-NEXT: [[TMP28:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS5]], i32 0, i32 3
5927 // CHECK13-NEXT: store ptr [[TMP24]], ptr [[TMP28]], align 8
5928 // CHECK13-NEXT: [[TMP29:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS5]], i32 0, i32 4
5929 // CHECK13-NEXT: store ptr @.offload_sizes.11, ptr [[TMP29]], align 8
5930 // CHECK13-NEXT: [[TMP30:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS5]], i32 0, i32 5
5931 // CHECK13-NEXT: store ptr @.offload_maptypes.12, ptr [[TMP30]], align 8
5932 // CHECK13-NEXT: [[TMP31:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS5]], i32 0, i32 6
5933 // CHECK13-NEXT: store ptr null, ptr [[TMP31]], align 8
5934 // CHECK13-NEXT: [[TMP32:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS5]], i32 0, i32 7
5935 // CHECK13-NEXT: store ptr null, ptr [[TMP32]], align 8
5936 // CHECK13-NEXT: [[TMP33:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS5]], i32 0, i32 8
5937 // CHECK13-NEXT: store i64 10, ptr [[TMP33]], align 8
5938 // CHECK13-NEXT: [[TMP34:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS5]], i32 0, i32 9
5939 // CHECK13-NEXT: store i64 0, ptr [[TMP34]], align 8
5940 // CHECK13-NEXT: [[TMP35:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS5]], i32 0, i32 10
5941 // CHECK13-NEXT: store [3 x i32] zeroinitializer, ptr [[TMP35]], align 4
5942 // CHECK13-NEXT: [[TMP36:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS5]], i32 0, i32 11
5943 // CHECK13-NEXT: store [3 x i32] zeroinitializer, ptr [[TMP36]], align 4
5944 // CHECK13-NEXT: [[TMP37:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS5]], i32 0, i32 12
5945 // CHECK13-NEXT: store i32 0, ptr [[TMP37]], align 4
5946 // CHECK13-NEXT: [[TMP38:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB3]], i64 -1, i32 0, i32 0, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l121.region_id, ptr [[KERNEL_ARGS5]])
5947 // CHECK13-NEXT: [[TMP39:%.*]] = icmp ne i32 [[TMP38]], 0
5948 // CHECK13-NEXT: br i1 [[TMP39]], label [[OMP_OFFLOAD_FAILED6:%.*]], label [[OMP_OFFLOAD_CONT7:%.*]]
5949 // CHECK13: omp_offload.failed6:
5950 // CHECK13-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l121(ptr [[A]]) #[[ATTR3]]
5951 // CHECK13-NEXT: br label [[OMP_OFFLOAD_CONT7]]
5952 // CHECK13: omp_offload.cont7:
5953 // CHECK13-NEXT: [[TMP40:%.*]] = load i32, ptr [[M]], align 4
5954 // CHECK13-NEXT: store i32 [[TMP40]], ptr [[M_CASTED]], align 4
5955 // CHECK13-NEXT: [[TMP41:%.*]] = load i64, ptr [[M_CASTED]], align 8
5956 // CHECK13-NEXT: [[TMP42:%.*]] = getelementptr inbounds [2 x ptr], ptr [[DOTOFFLOAD_BASEPTRS8]], i32 0, i32 0
5957 // CHECK13-NEXT: store i64 [[TMP41]], ptr [[TMP42]], align 8
5958 // CHECK13-NEXT: [[TMP43:%.*]] = getelementptr inbounds [2 x ptr], ptr [[DOTOFFLOAD_PTRS9]], i32 0, i32 0
5959 // CHECK13-NEXT: store i64 [[TMP41]], ptr [[TMP43]], align 8
5960 // CHECK13-NEXT: [[TMP44:%.*]] = getelementptr inbounds [2 x ptr], ptr [[DOTOFFLOAD_MAPPERS10]], i64 0, i64 0
5961 // CHECK13-NEXT: store ptr null, ptr [[TMP44]], align 8
5962 // CHECK13-NEXT: [[TMP45:%.*]] = getelementptr inbounds [2 x ptr], ptr [[DOTOFFLOAD_BASEPTRS8]], i32 0, i32 1
5963 // CHECK13-NEXT: store ptr [[A]], ptr [[TMP45]], align 8
5964 // CHECK13-NEXT: [[TMP46:%.*]] = getelementptr inbounds [2 x ptr], ptr [[DOTOFFLOAD_PTRS9]], i32 0, i32 1
5965 // CHECK13-NEXT: store ptr [[A]], ptr [[TMP46]], align 8
5966 // CHECK13-NEXT: [[TMP47:%.*]] = getelementptr inbounds [2 x ptr], ptr [[DOTOFFLOAD_MAPPERS10]], i64 0, i64 1
5967 // CHECK13-NEXT: store ptr null, ptr [[TMP47]], align 8
5968 // CHECK13-NEXT: [[TMP48:%.*]] = getelementptr inbounds [2 x ptr], ptr [[DOTOFFLOAD_BASEPTRS8]], i32 0, i32 0
5969 // CHECK13-NEXT: [[TMP49:%.*]] = getelementptr inbounds [2 x ptr], ptr [[DOTOFFLOAD_PTRS9]], i32 0, i32 0
5970 // CHECK13-NEXT: [[TMP50:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS12]], i32 0, i32 0
5971 // CHECK13-NEXT: store i32 3, ptr [[TMP50]], align 4
5972 // CHECK13-NEXT: [[TMP51:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS12]], i32 0, i32 1
5973 // CHECK13-NEXT: store i32 2, ptr [[TMP51]], align 4
5974 // CHECK13-NEXT: [[TMP52:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS12]], i32 0, i32 2
5975 // CHECK13-NEXT: store ptr [[TMP48]], ptr [[TMP52]], align 8
5976 // CHECK13-NEXT: [[TMP53:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS12]], i32 0, i32 3
5977 // CHECK13-NEXT: store ptr [[TMP49]], ptr [[TMP53]], align 8
5978 // CHECK13-NEXT: [[TMP54:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS12]], i32 0, i32 4
5979 // CHECK13-NEXT: store ptr @.offload_sizes.13, ptr [[TMP54]], align 8
5980 // CHECK13-NEXT: [[TMP55:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS12]], i32 0, i32 5
5981 // CHECK13-NEXT: store ptr @.offload_maptypes.14, ptr [[TMP55]], align 8
5982 // CHECK13-NEXT: [[TMP56:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS12]], i32 0, i32 6
5983 // CHECK13-NEXT: store ptr null, ptr [[TMP56]], align 8
5984 // CHECK13-NEXT: [[TMP57:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS12]], i32 0, i32 7
5985 // CHECK13-NEXT: store ptr null, ptr [[TMP57]], align 8
5986 // CHECK13-NEXT: [[TMP58:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS12]], i32 0, i32 8
5987 // CHECK13-NEXT: store i64 10, ptr [[TMP58]], align 8
5988 // CHECK13-NEXT: [[TMP59:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS12]], i32 0, i32 9
5989 // CHECK13-NEXT: store i64 0, ptr [[TMP59]], align 8
5990 // CHECK13-NEXT: [[TMP60:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS12]], i32 0, i32 10
5991 // CHECK13-NEXT: store [3 x i32] zeroinitializer, ptr [[TMP60]], align 4
5992 // CHECK13-NEXT: [[TMP61:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS12]], i32 0, i32 11
5993 // CHECK13-NEXT: store [3 x i32] zeroinitializer, ptr [[TMP61]], align 4
5994 // CHECK13-NEXT: [[TMP62:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS12]], i32 0, i32 12
5995 // CHECK13-NEXT: store i32 0, ptr [[TMP62]], align 4
5996 // CHECK13-NEXT: [[TMP63:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB3]], i64 -1, i32 0, i32 0, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l126.region_id, ptr [[KERNEL_ARGS12]])
5997 // CHECK13-NEXT: [[TMP64:%.*]] = icmp ne i32 [[TMP63]], 0
5998 // CHECK13-NEXT: br i1 [[TMP64]], label [[OMP_OFFLOAD_FAILED13:%.*]], label [[OMP_OFFLOAD_CONT14:%.*]]
5999 // CHECK13: omp_offload.failed13:
6000 // CHECK13-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l126(i64 [[TMP41]], ptr [[A]]) #[[ATTR3]]
6001 // CHECK13-NEXT: br label [[OMP_OFFLOAD_CONT14]]
6002 // CHECK13: omp_offload.cont14:
6003 // CHECK13-NEXT: [[TMP65:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS15]], i32 0, i32 0
6004 // CHECK13-NEXT: store ptr [[A]], ptr [[TMP65]], align 8
6005 // CHECK13-NEXT: [[TMP66:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS16]], i32 0, i32 0
6006 // CHECK13-NEXT: store ptr [[A]], ptr [[TMP66]], align 8
6007 // CHECK13-NEXT: [[TMP67:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_MAPPERS17]], i64 0, i64 0
6008 // CHECK13-NEXT: store ptr null, ptr [[TMP67]], align 8
6009 // CHECK13-NEXT: [[TMP68:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS15]], i32 0, i32 0
6010 // CHECK13-NEXT: [[TMP69:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS16]], i32 0, i32 0
6011 // CHECK13-NEXT: [[TMP70:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS19]], i32 0, i32 0
6012 // CHECK13-NEXT: store i32 3, ptr [[TMP70]], align 4
6013 // CHECK13-NEXT: [[TMP71:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS19]], i32 0, i32 1
6014 // CHECK13-NEXT: store i32 1, ptr [[TMP71]], align 4
6015 // CHECK13-NEXT: [[TMP72:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS19]], i32 0, i32 2
6016 // CHECK13-NEXT: store ptr [[TMP68]], ptr [[TMP72]], align 8
6017 // CHECK13-NEXT: [[TMP73:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS19]], i32 0, i32 3
6018 // CHECK13-NEXT: store ptr [[TMP69]], ptr [[TMP73]], align 8
6019 // CHECK13-NEXT: [[TMP74:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS19]], i32 0, i32 4
6020 // CHECK13-NEXT: store ptr @.offload_sizes.15, ptr [[TMP74]], align 8
6021 // CHECK13-NEXT: [[TMP75:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS19]], i32 0, i32 5
6022 // CHECK13-NEXT: store ptr @.offload_maptypes.16, ptr [[TMP75]], align 8
6023 // CHECK13-NEXT: [[TMP76:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS19]], i32 0, i32 6
6024 // CHECK13-NEXT: store ptr null, ptr [[TMP76]], align 8
6025 // CHECK13-NEXT: [[TMP77:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS19]], i32 0, i32 7
6026 // CHECK13-NEXT: store ptr null, ptr [[TMP77]], align 8
6027 // CHECK13-NEXT: [[TMP78:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS19]], i32 0, i32 8
6028 // CHECK13-NEXT: store i64 10, ptr [[TMP78]], align 8
6029 // CHECK13-NEXT: [[TMP79:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS19]], i32 0, i32 9
6030 // CHECK13-NEXT: store i64 0, ptr [[TMP79]], align 8
6031 // CHECK13-NEXT: [[TMP80:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS19]], i32 0, i32 10
6032 // CHECK13-NEXT: store [3 x i32] zeroinitializer, ptr [[TMP80]], align 4
6033 // CHECK13-NEXT: [[TMP81:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS19]], i32 0, i32 11
6034 // CHECK13-NEXT: store [3 x i32] zeroinitializer, ptr [[TMP81]], align 4
6035 // CHECK13-NEXT: [[TMP82:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS19]], i32 0, i32 12
6036 // CHECK13-NEXT: store i32 0, ptr [[TMP82]], align 4
6037 // CHECK13-NEXT: [[TMP83:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB3]], i64 -1, i32 0, i32 0, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l131.region_id, ptr [[KERNEL_ARGS19]])
6038 // CHECK13-NEXT: [[TMP84:%.*]] = icmp ne i32 [[TMP83]], 0
6039 // CHECK13-NEXT: br i1 [[TMP84]], label [[OMP_OFFLOAD_FAILED20:%.*]], label [[OMP_OFFLOAD_CONT21:%.*]]
6040 // CHECK13: omp_offload.failed20:
6041 // CHECK13-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l131(ptr [[A]]) #[[ATTR3]]
6042 // CHECK13-NEXT: br label [[OMP_OFFLOAD_CONT21]]
6043 // CHECK13: omp_offload.cont21:
6044 // CHECK13-NEXT: [[TMP85:%.*]] = load i32, ptr [[M]], align 4
6045 // CHECK13-NEXT: store i32 [[TMP85]], ptr [[M_CASTED22]], align 4
6046 // CHECK13-NEXT: [[TMP86:%.*]] = load i64, ptr [[M_CASTED22]], align 8
6047 // CHECK13-NEXT: [[TMP87:%.*]] = getelementptr inbounds [2 x ptr], ptr [[DOTOFFLOAD_BASEPTRS23]], i32 0, i32 0
6048 // CHECK13-NEXT: store i64 [[TMP86]], ptr [[TMP87]], align 8
6049 // CHECK13-NEXT: [[TMP88:%.*]] = getelementptr inbounds [2 x ptr], ptr [[DOTOFFLOAD_PTRS24]], i32 0, i32 0
6050 // CHECK13-NEXT: store i64 [[TMP86]], ptr [[TMP88]], align 8
6051 // CHECK13-NEXT: [[TMP89:%.*]] = getelementptr inbounds [2 x ptr], ptr [[DOTOFFLOAD_MAPPERS25]], i64 0, i64 0
6052 // CHECK13-NEXT: store ptr null, ptr [[TMP89]], align 8
6053 // CHECK13-NEXT: [[TMP90:%.*]] = getelementptr inbounds [2 x ptr], ptr [[DOTOFFLOAD_BASEPTRS23]], i32 0, i32 1
6054 // CHECK13-NEXT: store ptr [[A]], ptr [[TMP90]], align 8
6055 // CHECK13-NEXT: [[TMP91:%.*]] = getelementptr inbounds [2 x ptr], ptr [[DOTOFFLOAD_PTRS24]], i32 0, i32 1
6056 // CHECK13-NEXT: store ptr [[A]], ptr [[TMP91]], align 8
6057 // CHECK13-NEXT: [[TMP92:%.*]] = getelementptr inbounds [2 x ptr], ptr [[DOTOFFLOAD_MAPPERS25]], i64 0, i64 1
6058 // CHECK13-NEXT: store ptr null, ptr [[TMP92]], align 8
6059 // CHECK13-NEXT: [[TMP93:%.*]] = getelementptr inbounds [2 x ptr], ptr [[DOTOFFLOAD_BASEPTRS23]], i32 0, i32 0
6060 // CHECK13-NEXT: [[TMP94:%.*]] = getelementptr inbounds [2 x ptr], ptr [[DOTOFFLOAD_PTRS24]], i32 0, i32 0
6061 // CHECK13-NEXT: [[TMP95:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS27]], i32 0, i32 0
6062 // CHECK13-NEXT: store i32 3, ptr [[TMP95]], align 4
6063 // CHECK13-NEXT: [[TMP96:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS27]], i32 0, i32 1
6064 // CHECK13-NEXT: store i32 2, ptr [[TMP96]], align 4
6065 // CHECK13-NEXT: [[TMP97:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS27]], i32 0, i32 2
6066 // CHECK13-NEXT: store ptr [[TMP93]], ptr [[TMP97]], align 8
6067 // CHECK13-NEXT: [[TMP98:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS27]], i32 0, i32 3
6068 // CHECK13-NEXT: store ptr [[TMP94]], ptr [[TMP98]], align 8
6069 // CHECK13-NEXT: [[TMP99:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS27]], i32 0, i32 4
6070 // CHECK13-NEXT: store ptr @.offload_sizes.17, ptr [[TMP99]], align 8
6071 // CHECK13-NEXT: [[TMP100:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS27]], i32 0, i32 5
6072 // CHECK13-NEXT: store ptr @.offload_maptypes.18, ptr [[TMP100]], align 8
6073 // CHECK13-NEXT: [[TMP101:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS27]], i32 0, i32 6
6074 // CHECK13-NEXT: store ptr null, ptr [[TMP101]], align 8
6075 // CHECK13-NEXT: [[TMP102:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS27]], i32 0, i32 7
6076 // CHECK13-NEXT: store ptr null, ptr [[TMP102]], align 8
6077 // CHECK13-NEXT: [[TMP103:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS27]], i32 0, i32 8
6078 // CHECK13-NEXT: store i64 10, ptr [[TMP103]], align 8
6079 // CHECK13-NEXT: [[TMP104:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS27]], i32 0, i32 9
6080 // CHECK13-NEXT: store i64 0, ptr [[TMP104]], align 8
6081 // CHECK13-NEXT: [[TMP105:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS27]], i32 0, i32 10
6082 // CHECK13-NEXT: store [3 x i32] zeroinitializer, ptr [[TMP105]], align 4
6083 // CHECK13-NEXT: [[TMP106:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS27]], i32 0, i32 11
6084 // CHECK13-NEXT: store [3 x i32] zeroinitializer, ptr [[TMP106]], align 4
6085 // CHECK13-NEXT: [[TMP107:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS27]], i32 0, i32 12
6086 // CHECK13-NEXT: store i32 0, ptr [[TMP107]], align 4
6087 // CHECK13-NEXT: [[TMP108:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB3]], i64 -1, i32 0, i32 0, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l136.region_id, ptr [[KERNEL_ARGS27]])
6088 // CHECK13-NEXT: [[TMP109:%.*]] = icmp ne i32 [[TMP108]], 0
6089 // CHECK13-NEXT: br i1 [[TMP109]], label [[OMP_OFFLOAD_FAILED28:%.*]], label [[OMP_OFFLOAD_CONT29:%.*]]
6090 // CHECK13: omp_offload.failed28:
6091 // CHECK13-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l136(i64 [[TMP86]], ptr [[A]]) #[[ATTR3]]
6092 // CHECK13-NEXT: br label [[OMP_OFFLOAD_CONT29]]
6093 // CHECK13: omp_offload.cont29:
6094 // CHECK13-NEXT: ret i32 0
6097 // CHECK13-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l116
6098 // CHECK13-SAME: (ptr noundef nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] {
6099 // CHECK13-NEXT: entry:
6100 // CHECK13-NEXT: [[A_ADDR:%.*]] = alloca ptr, align 8
6101 // CHECK13-NEXT: store ptr [[A]], ptr [[A_ADDR]], align 8
6102 // CHECK13-NEXT: [[TMP0:%.*]] = load ptr, ptr [[A_ADDR]], align 8
6103 // CHECK13-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_teams(ptr @[[GLOB3]], i32 1, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l116.omp_outlined, ptr [[TMP0]])
6104 // CHECK13-NEXT: ret void
6107 // CHECK13-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l116.omp_outlined
6108 // CHECK13-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], ptr noundef nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] {
6109 // CHECK13-NEXT: entry:
6110 // CHECK13-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 8
6111 // CHECK13-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 8
6112 // CHECK13-NEXT: [[A_ADDR:%.*]] = alloca ptr, align 8
6113 // CHECK13-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
6114 // CHECK13-NEXT: [[TMP:%.*]] = alloca i32, align 4
6115 // CHECK13-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4
6116 // CHECK13-NEXT: [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4
6117 // CHECK13-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
6118 // CHECK13-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
6119 // CHECK13-NEXT: [[I:%.*]] = alloca i32, align 4
6120 // CHECK13-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 8
6121 // CHECK13-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 8
6122 // CHECK13-NEXT: store ptr [[A]], ptr [[A_ADDR]], align 8
6123 // CHECK13-NEXT: [[TMP0:%.*]] = load ptr, ptr [[A_ADDR]], align 8
6124 // CHECK13-NEXT: store i32 0, ptr [[DOTOMP_COMB_LB]], align 4
6125 // CHECK13-NEXT: store i32 9, ptr [[DOTOMP_COMB_UB]], align 4
6126 // CHECK13-NEXT: store i32 1, ptr [[DOTOMP_STRIDE]], align 4
6127 // CHECK13-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
6128 // CHECK13-NEXT: [[TMP1:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
6129 // CHECK13-NEXT: [[TMP2:%.*]] = load i32, ptr [[TMP1]], align 4
6130 // CHECK13-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB1]], i32 [[TMP2]], i32 92, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
6131 // CHECK13-NEXT: [[TMP3:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
6132 // CHECK13-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP3]], 9
6133 // CHECK13-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
6134 // CHECK13: cond.true:
6135 // CHECK13-NEXT: br label [[COND_END:%.*]]
6136 // CHECK13: cond.false:
6137 // CHECK13-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
6138 // CHECK13-NEXT: br label [[COND_END]]
6139 // CHECK13: cond.end:
6140 // CHECK13-NEXT: [[COND:%.*]] = phi i32 [ 9, [[COND_TRUE]] ], [ [[TMP4]], [[COND_FALSE]] ]
6141 // CHECK13-NEXT: store i32 [[COND]], ptr [[DOTOMP_COMB_UB]], align 4
6142 // CHECK13-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
6143 // CHECK13-NEXT: store i32 [[TMP5]], ptr [[DOTOMP_IV]], align 4
6144 // CHECK13-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
6145 // CHECK13: omp.inner.for.cond:
6146 // CHECK13-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
6147 // CHECK13-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
6148 // CHECK13-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP6]], [[TMP7]]
6149 // CHECK13-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
6150 // CHECK13: omp.inner.for.body:
6151 // CHECK13-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
6152 // CHECK13-NEXT: [[TMP9:%.*]] = zext i32 [[TMP8]] to i64
6153 // CHECK13-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
6154 // CHECK13-NEXT: [[TMP11:%.*]] = zext i32 [[TMP10]] to i64
6155 // CHECK13-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_call(ptr @[[GLOB3]], i32 3, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l116.omp_outlined.omp_outlined, i64 [[TMP9]], i64 [[TMP11]], ptr [[TMP0]])
6156 // CHECK13-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
6157 // CHECK13: omp.inner.for.inc:
6158 // CHECK13-NEXT: [[TMP12:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
6159 // CHECK13-NEXT: [[TMP13:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4
6160 // CHECK13-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP12]], [[TMP13]]
6161 // CHECK13-NEXT: store i32 [[ADD]], ptr [[DOTOMP_IV]], align 4
6162 // CHECK13-NEXT: br label [[OMP_INNER_FOR_COND]]
6163 // CHECK13: omp.inner.for.end:
6164 // CHECK13-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
6165 // CHECK13: omp.loop.exit:
6166 // CHECK13-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB1]], i32 [[TMP2]])
6167 // CHECK13-NEXT: ret void
6170 // CHECK13-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l116.omp_outlined.omp_outlined
6171 // CHECK13-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], i64 noundef [[DOTPREVIOUS_LB_:%.*]], i64 noundef [[DOTPREVIOUS_UB_:%.*]], ptr noundef nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] {
6172 // CHECK13-NEXT: entry:
6173 // CHECK13-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 8
6174 // CHECK13-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 8
6175 // CHECK13-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i64, align 8
6176 // CHECK13-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i64, align 8
6177 // CHECK13-NEXT: [[A_ADDR:%.*]] = alloca ptr, align 8
6178 // CHECK13-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
6179 // CHECK13-NEXT: [[TMP:%.*]] = alloca i32, align 4
6180 // CHECK13-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4
6181 // CHECK13-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4
6182 // CHECK13-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
6183 // CHECK13-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
6184 // CHECK13-NEXT: [[I:%.*]] = alloca i32, align 4
6185 // CHECK13-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 8
6186 // CHECK13-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 8
6187 // CHECK13-NEXT: store i64 [[DOTPREVIOUS_LB_]], ptr [[DOTPREVIOUS_LB__ADDR]], align 8
6188 // CHECK13-NEXT: store i64 [[DOTPREVIOUS_UB_]], ptr [[DOTPREVIOUS_UB__ADDR]], align 8
6189 // CHECK13-NEXT: store ptr [[A]], ptr [[A_ADDR]], align 8
6190 // CHECK13-NEXT: [[TMP0:%.*]] = load ptr, ptr [[A_ADDR]], align 8
6191 // CHECK13-NEXT: store i32 0, ptr [[DOTOMP_LB]], align 4
6192 // CHECK13-NEXT: store i32 9, ptr [[DOTOMP_UB]], align 4
6193 // CHECK13-NEXT: [[TMP1:%.*]] = load i64, ptr [[DOTPREVIOUS_LB__ADDR]], align 8
6194 // CHECK13-NEXT: [[CONV:%.*]] = trunc i64 [[TMP1]] to i32
6195 // CHECK13-NEXT: [[TMP2:%.*]] = load i64, ptr [[DOTPREVIOUS_UB__ADDR]], align 8
6196 // CHECK13-NEXT: [[CONV1:%.*]] = trunc i64 [[TMP2]] to i32
6197 // CHECK13-NEXT: store i32 [[CONV]], ptr [[DOTOMP_LB]], align 4
6198 // CHECK13-NEXT: store i32 [[CONV1]], ptr [[DOTOMP_UB]], align 4
6199 // CHECK13-NEXT: store i32 1, ptr [[DOTOMP_STRIDE]], align 4
6200 // CHECK13-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
6201 // CHECK13-NEXT: [[TMP3:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
6202 // CHECK13-NEXT: [[TMP4:%.*]] = load i32, ptr [[TMP3]], align 4
6203 // CHECK13-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB2]], i32 [[TMP4]], i32 34, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
6204 // CHECK13-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
6205 // CHECK13-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP5]], 9
6206 // CHECK13-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
6207 // CHECK13: cond.true:
6208 // CHECK13-NEXT: br label [[COND_END:%.*]]
6209 // CHECK13: cond.false:
6210 // CHECK13-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
6211 // CHECK13-NEXT: br label [[COND_END]]
6212 // CHECK13: cond.end:
6213 // CHECK13-NEXT: [[COND:%.*]] = phi i32 [ 9, [[COND_TRUE]] ], [ [[TMP6]], [[COND_FALSE]] ]
6214 // CHECK13-NEXT: store i32 [[COND]], ptr [[DOTOMP_UB]], align 4
6215 // CHECK13-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4
6216 // CHECK13-NEXT: store i32 [[TMP7]], ptr [[DOTOMP_IV]], align 4
6217 // CHECK13-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
6218 // CHECK13: omp.inner.for.cond:
6219 // CHECK13-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
6220 // CHECK13-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
6221 // CHECK13-NEXT: [[CMP2:%.*]] = icmp sle i32 [[TMP8]], [[TMP9]]
6222 // CHECK13-NEXT: br i1 [[CMP2]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
6223 // CHECK13: omp.inner.for.body:
6224 // CHECK13-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
6225 // CHECK13-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP10]], 1
6226 // CHECK13-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
6227 // CHECK13-NEXT: store i32 [[ADD]], ptr [[I]], align 4
6228 // CHECK13-NEXT: [[TMP11:%.*]] = load i32, ptr [[I]], align 4
6229 // CHECK13-NEXT: [[IDXPROM:%.*]] = sext i32 [[TMP11]] to i64
6230 // CHECK13-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds [10 x i32], ptr [[TMP0]], i64 0, i64 [[IDXPROM]]
6231 // CHECK13-NEXT: store i32 0, ptr [[ARRAYIDX]], align 4
6232 // CHECK13-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
6233 // CHECK13: omp.body.continue:
6234 // CHECK13-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
6235 // CHECK13: omp.inner.for.inc:
6236 // CHECK13-NEXT: [[TMP12:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
6237 // CHECK13-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP12]], 1
6238 // CHECK13-NEXT: store i32 [[ADD3]], ptr [[DOTOMP_IV]], align 4
6239 // CHECK13-NEXT: br label [[OMP_INNER_FOR_COND]]
6240 // CHECK13: omp.inner.for.end:
6241 // CHECK13-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
6242 // CHECK13: omp.loop.exit:
6243 // CHECK13-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB2]], i32 [[TMP4]])
6244 // CHECK13-NEXT: ret void
6247 // CHECK13-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l121
6248 // CHECK13-SAME: (ptr noundef nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] {
6249 // CHECK13-NEXT: entry:
6250 // CHECK13-NEXT: [[A_ADDR:%.*]] = alloca ptr, align 8
6251 // CHECK13-NEXT: store ptr [[A]], ptr [[A_ADDR]], align 8
6252 // CHECK13-NEXT: [[TMP0:%.*]] = load ptr, ptr [[A_ADDR]], align 8
6253 // CHECK13-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_teams(ptr @[[GLOB3]], i32 1, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l121.omp_outlined, ptr [[TMP0]])
6254 // CHECK13-NEXT: ret void
6257 // CHECK13-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l121.omp_outlined
6258 // CHECK13-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], ptr noundef nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] {
6259 // CHECK13-NEXT: entry:
6260 // CHECK13-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 8
6261 // CHECK13-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 8
6262 // CHECK13-NEXT: [[A_ADDR:%.*]] = alloca ptr, align 8
6263 // CHECK13-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
6264 // CHECK13-NEXT: [[TMP:%.*]] = alloca i32, align 4
6265 // CHECK13-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4
6266 // CHECK13-NEXT: [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4
6267 // CHECK13-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
6268 // CHECK13-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
6269 // CHECK13-NEXT: [[I:%.*]] = alloca i32, align 4
6270 // CHECK13-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 8
6271 // CHECK13-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 8
6272 // CHECK13-NEXT: store ptr [[A]], ptr [[A_ADDR]], align 8
6273 // CHECK13-NEXT: [[TMP0:%.*]] = load ptr, ptr [[A_ADDR]], align 8
6274 // CHECK13-NEXT: store i32 0, ptr [[DOTOMP_COMB_LB]], align 4
6275 // CHECK13-NEXT: store i32 9, ptr [[DOTOMP_COMB_UB]], align 4
6276 // CHECK13-NEXT: store i32 1, ptr [[DOTOMP_STRIDE]], align 4
6277 // CHECK13-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
6278 // CHECK13-NEXT: [[TMP1:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
6279 // CHECK13-NEXT: [[TMP2:%.*]] = load i32, ptr [[TMP1]], align 4
6280 // CHECK13-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB1]], i32 [[TMP2]], i32 92, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
6281 // CHECK13-NEXT: [[TMP3:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
6282 // CHECK13-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP3]], 9
6283 // CHECK13-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
6284 // CHECK13: cond.true:
6285 // CHECK13-NEXT: br label [[COND_END:%.*]]
6286 // CHECK13: cond.false:
6287 // CHECK13-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
6288 // CHECK13-NEXT: br label [[COND_END]]
6289 // CHECK13: cond.end:
6290 // CHECK13-NEXT: [[COND:%.*]] = phi i32 [ 9, [[COND_TRUE]] ], [ [[TMP4]], [[COND_FALSE]] ]
6291 // CHECK13-NEXT: store i32 [[COND]], ptr [[DOTOMP_COMB_UB]], align 4
6292 // CHECK13-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
6293 // CHECK13-NEXT: store i32 [[TMP5]], ptr [[DOTOMP_IV]], align 4
6294 // CHECK13-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
6295 // CHECK13: omp.inner.for.cond:
6296 // CHECK13-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
6297 // CHECK13-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
6298 // CHECK13-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP6]], [[TMP7]]
6299 // CHECK13-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
6300 // CHECK13: omp.inner.for.body:
6301 // CHECK13-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
6302 // CHECK13-NEXT: [[TMP9:%.*]] = zext i32 [[TMP8]] to i64
6303 // CHECK13-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
6304 // CHECK13-NEXT: [[TMP11:%.*]] = zext i32 [[TMP10]] to i64
6305 // CHECK13-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_call(ptr @[[GLOB3]], i32 3, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l121.omp_outlined.omp_outlined, i64 [[TMP9]], i64 [[TMP11]], ptr [[TMP0]])
6306 // CHECK13-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
6307 // CHECK13: omp.inner.for.inc:
6308 // CHECK13-NEXT: [[TMP12:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
6309 // CHECK13-NEXT: [[TMP13:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4
6310 // CHECK13-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP12]], [[TMP13]]
6311 // CHECK13-NEXT: store i32 [[ADD]], ptr [[DOTOMP_IV]], align 4
6312 // CHECK13-NEXT: br label [[OMP_INNER_FOR_COND]]
6313 // CHECK13: omp.inner.for.end:
6314 // CHECK13-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
6315 // CHECK13: omp.loop.exit:
6316 // CHECK13-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB1]], i32 [[TMP2]])
6317 // CHECK13-NEXT: ret void
6320 // CHECK13-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l121.omp_outlined.omp_outlined
6321 // CHECK13-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], i64 noundef [[DOTPREVIOUS_LB_:%.*]], i64 noundef [[DOTPREVIOUS_UB_:%.*]], ptr noundef nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] {
6322 // CHECK13-NEXT: entry:
6323 // CHECK13-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 8
6324 // CHECK13-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 8
6325 // CHECK13-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i64, align 8
6326 // CHECK13-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i64, align 8
6327 // CHECK13-NEXT: [[A_ADDR:%.*]] = alloca ptr, align 8
6328 // CHECK13-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
6329 // CHECK13-NEXT: [[TMP:%.*]] = alloca i32, align 4
6330 // CHECK13-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4
6331 // CHECK13-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4
6332 // CHECK13-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
6333 // CHECK13-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
6334 // CHECK13-NEXT: [[I:%.*]] = alloca i32, align 4
6335 // CHECK13-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 8
6336 // CHECK13-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 8
6337 // CHECK13-NEXT: store i64 [[DOTPREVIOUS_LB_]], ptr [[DOTPREVIOUS_LB__ADDR]], align 8
6338 // CHECK13-NEXT: store i64 [[DOTPREVIOUS_UB_]], ptr [[DOTPREVIOUS_UB__ADDR]], align 8
6339 // CHECK13-NEXT: store ptr [[A]], ptr [[A_ADDR]], align 8
6340 // CHECK13-NEXT: [[TMP0:%.*]] = load ptr, ptr [[A_ADDR]], align 8
6341 // CHECK13-NEXT: store i32 0, ptr [[DOTOMP_LB]], align 4
6342 // CHECK13-NEXT: store i32 9, ptr [[DOTOMP_UB]], align 4
6343 // CHECK13-NEXT: [[TMP1:%.*]] = load i64, ptr [[DOTPREVIOUS_LB__ADDR]], align 8
6344 // CHECK13-NEXT: [[CONV:%.*]] = trunc i64 [[TMP1]] to i32
6345 // CHECK13-NEXT: [[TMP2:%.*]] = load i64, ptr [[DOTPREVIOUS_UB__ADDR]], align 8
6346 // CHECK13-NEXT: [[CONV1:%.*]] = trunc i64 [[TMP2]] to i32
6347 // CHECK13-NEXT: store i32 [[CONV]], ptr [[DOTOMP_LB]], align 4
6348 // CHECK13-NEXT: store i32 [[CONV1]], ptr [[DOTOMP_UB]], align 4
6349 // CHECK13-NEXT: store i32 1, ptr [[DOTOMP_STRIDE]], align 4
6350 // CHECK13-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
6351 // CHECK13-NEXT: [[TMP3:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
6352 // CHECK13-NEXT: [[TMP4:%.*]] = load i32, ptr [[TMP3]], align 4
6353 // CHECK13-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB2]], i32 [[TMP4]], i32 34, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
6354 // CHECK13-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
6355 // CHECK13-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP5]], 9
6356 // CHECK13-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
6357 // CHECK13: cond.true:
6358 // CHECK13-NEXT: br label [[COND_END:%.*]]
6359 // CHECK13: cond.false:
6360 // CHECK13-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
6361 // CHECK13-NEXT: br label [[COND_END]]
6362 // CHECK13: cond.end:
6363 // CHECK13-NEXT: [[COND:%.*]] = phi i32 [ 9, [[COND_TRUE]] ], [ [[TMP6]], [[COND_FALSE]] ]
6364 // CHECK13-NEXT: store i32 [[COND]], ptr [[DOTOMP_UB]], align 4
6365 // CHECK13-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4
6366 // CHECK13-NEXT: store i32 [[TMP7]], ptr [[DOTOMP_IV]], align 4
6367 // CHECK13-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
6368 // CHECK13: omp.inner.for.cond:
6369 // CHECK13-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
6370 // CHECK13-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
6371 // CHECK13-NEXT: [[CMP2:%.*]] = icmp sle i32 [[TMP8]], [[TMP9]]
6372 // CHECK13-NEXT: br i1 [[CMP2]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
6373 // CHECK13: omp.inner.for.body:
6374 // CHECK13-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
6375 // CHECK13-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP10]], 1
6376 // CHECK13-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
6377 // CHECK13-NEXT: store i32 [[ADD]], ptr [[I]], align 4
6378 // CHECK13-NEXT: [[TMP11:%.*]] = load i32, ptr [[I]], align 4
6379 // CHECK13-NEXT: [[IDXPROM:%.*]] = sext i32 [[TMP11]] to i64
6380 // CHECK13-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds [10 x i32], ptr [[TMP0]], i64 0, i64 [[IDXPROM]]
6381 // CHECK13-NEXT: store i32 0, ptr [[ARRAYIDX]], align 4
6382 // CHECK13-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
6383 // CHECK13: omp.body.continue:
6384 // CHECK13-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
6385 // CHECK13: omp.inner.for.inc:
6386 // CHECK13-NEXT: [[TMP12:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
6387 // CHECK13-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP12]], 1
6388 // CHECK13-NEXT: store i32 [[ADD3]], ptr [[DOTOMP_IV]], align 4
6389 // CHECK13-NEXT: br label [[OMP_INNER_FOR_COND]]
6390 // CHECK13: omp.inner.for.end:
6391 // CHECK13-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
6392 // CHECK13: omp.loop.exit:
6393 // CHECK13-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB2]], i32 [[TMP4]])
6394 // CHECK13-NEXT: ret void
6397 // CHECK13-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l126
6398 // CHECK13-SAME: (i64 noundef [[M:%.*]], ptr noundef nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] {
6399 // CHECK13-NEXT: entry:
6400 // CHECK13-NEXT: [[M_ADDR:%.*]] = alloca i64, align 8
6401 // CHECK13-NEXT: [[A_ADDR:%.*]] = alloca ptr, align 8
6402 // CHECK13-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4
6403 // CHECK13-NEXT: [[DOTCAPTURE_EXPR__CASTED:%.*]] = alloca i64, align 8
6404 // CHECK13-NEXT: store i64 [[M]], ptr [[M_ADDR]], align 8
6405 // CHECK13-NEXT: store ptr [[A]], ptr [[A_ADDR]], align 8
6406 // CHECK13-NEXT: [[TMP0:%.*]] = load ptr, ptr [[A_ADDR]], align 8
6407 // CHECK13-NEXT: [[TMP1:%.*]] = load i32, ptr [[M_ADDR]], align 4
6408 // CHECK13-NEXT: store i32 [[TMP1]], ptr [[DOTCAPTURE_EXPR_]], align 4
6409 // CHECK13-NEXT: [[TMP2:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4
6410 // CHECK13-NEXT: store i32 [[TMP2]], ptr [[DOTCAPTURE_EXPR__CASTED]], align 4
6411 // CHECK13-NEXT: [[TMP3:%.*]] = load i64, ptr [[DOTCAPTURE_EXPR__CASTED]], align 8
6412 // CHECK13-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_teams(ptr @[[GLOB3]], i32 2, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l126.omp_outlined, ptr [[TMP0]], i64 [[TMP3]])
6413 // CHECK13-NEXT: ret void
6416 // CHECK13-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l126.omp_outlined
6417 // CHECK13-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], ptr noundef nonnull align 4 dereferenceable(40) [[A:%.*]], i64 noundef [[DOTCAPTURE_EXPR_:%.*]]) #[[ATTR2]] {
6418 // CHECK13-NEXT: entry:
6419 // CHECK13-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 8
6420 // CHECK13-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 8
6421 // CHECK13-NEXT: [[A_ADDR:%.*]] = alloca ptr, align 8
6422 // CHECK13-NEXT: [[DOTCAPTURE_EXPR__ADDR:%.*]] = alloca i64, align 8
6423 // CHECK13-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
6424 // CHECK13-NEXT: [[TMP:%.*]] = alloca i32, align 4
6425 // CHECK13-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4
6426 // CHECK13-NEXT: [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4
6427 // CHECK13-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
6428 // CHECK13-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
6429 // CHECK13-NEXT: [[I:%.*]] = alloca i32, align 4
6430 // CHECK13-NEXT: [[DOTCAPTURE_EXPR__CASTED:%.*]] = alloca i64, align 8
6431 // CHECK13-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 8
6432 // CHECK13-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 8
6433 // CHECK13-NEXT: store ptr [[A]], ptr [[A_ADDR]], align 8
6434 // CHECK13-NEXT: store i64 [[DOTCAPTURE_EXPR_]], ptr [[DOTCAPTURE_EXPR__ADDR]], align 8
6435 // CHECK13-NEXT: [[TMP0:%.*]] = load ptr, ptr [[A_ADDR]], align 8
6436 // CHECK13-NEXT: store i32 0, ptr [[DOTOMP_COMB_LB]], align 4
6437 // CHECK13-NEXT: store i32 9, ptr [[DOTOMP_COMB_UB]], align 4
6438 // CHECK13-NEXT: store i32 1, ptr [[DOTOMP_STRIDE]], align 4
6439 // CHECK13-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
6440 // CHECK13-NEXT: [[TMP1:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
6441 // CHECK13-NEXT: [[TMP2:%.*]] = load i32, ptr [[TMP1]], align 4
6442 // CHECK13-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB1]], i32 [[TMP2]], i32 92, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
6443 // CHECK13-NEXT: [[TMP3:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
6444 // CHECK13-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP3]], 9
6445 // CHECK13-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
6446 // CHECK13: cond.true:
6447 // CHECK13-NEXT: br label [[COND_END:%.*]]
6448 // CHECK13: cond.false:
6449 // CHECK13-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
6450 // CHECK13-NEXT: br label [[COND_END]]
6451 // CHECK13: cond.end:
6452 // CHECK13-NEXT: [[COND:%.*]] = phi i32 [ 9, [[COND_TRUE]] ], [ [[TMP4]], [[COND_FALSE]] ]
6453 // CHECK13-NEXT: store i32 [[COND]], ptr [[DOTOMP_COMB_UB]], align 4
6454 // CHECK13-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
6455 // CHECK13-NEXT: store i32 [[TMP5]], ptr [[DOTOMP_IV]], align 4
6456 // CHECK13-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
6457 // CHECK13: omp.inner.for.cond:
6458 // CHECK13-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
6459 // CHECK13-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
6460 // CHECK13-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP6]], [[TMP7]]
6461 // CHECK13-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
6462 // CHECK13: omp.inner.for.body:
6463 // CHECK13-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
6464 // CHECK13-NEXT: [[TMP9:%.*]] = zext i32 [[TMP8]] to i64
6465 // CHECK13-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
6466 // CHECK13-NEXT: [[TMP11:%.*]] = zext i32 [[TMP10]] to i64
6467 // CHECK13-NEXT: [[TMP12:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR__ADDR]], align 4
6468 // CHECK13-NEXT: store i32 [[TMP12]], ptr [[DOTCAPTURE_EXPR__CASTED]], align 4
6469 // CHECK13-NEXT: [[TMP13:%.*]] = load i64, ptr [[DOTCAPTURE_EXPR__CASTED]], align 8
6470 // CHECK13-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_call(ptr @[[GLOB3]], i32 4, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l126.omp_outlined.omp_outlined, i64 [[TMP9]], i64 [[TMP11]], ptr [[TMP0]], i64 [[TMP13]])
6471 // CHECK13-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
6472 // CHECK13: omp.inner.for.inc:
6473 // CHECK13-NEXT: [[TMP14:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
6474 // CHECK13-NEXT: [[TMP15:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4
6475 // CHECK13-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP14]], [[TMP15]]
6476 // CHECK13-NEXT: store i32 [[ADD]], ptr [[DOTOMP_IV]], align 4
6477 // CHECK13-NEXT: br label [[OMP_INNER_FOR_COND]]
6478 // CHECK13: omp.inner.for.end:
6479 // CHECK13-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
6480 // CHECK13: omp.loop.exit:
6481 // CHECK13-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB1]], i32 [[TMP2]])
6482 // CHECK13-NEXT: ret void
6485 // CHECK13-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l126.omp_outlined.omp_outlined
6486 // CHECK13-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], i64 noundef [[DOTPREVIOUS_LB_:%.*]], i64 noundef [[DOTPREVIOUS_UB_:%.*]], ptr noundef nonnull align 4 dereferenceable(40) [[A:%.*]], i64 noundef [[DOTCAPTURE_EXPR_:%.*]]) #[[ATTR2]] {
6487 // CHECK13-NEXT: entry:
6488 // CHECK13-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 8
6489 // CHECK13-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 8
6490 // CHECK13-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i64, align 8
6491 // CHECK13-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i64, align 8
6492 // CHECK13-NEXT: [[A_ADDR:%.*]] = alloca ptr, align 8
6493 // CHECK13-NEXT: [[DOTCAPTURE_EXPR__ADDR:%.*]] = alloca i64, align 8
6494 // CHECK13-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
6495 // CHECK13-NEXT: [[TMP:%.*]] = alloca i32, align 4
6496 // CHECK13-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4
6497 // CHECK13-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4
6498 // CHECK13-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
6499 // CHECK13-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
6500 // CHECK13-NEXT: [[I:%.*]] = alloca i32, align 4
6501 // CHECK13-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 8
6502 // CHECK13-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 8
6503 // CHECK13-NEXT: store i64 [[DOTPREVIOUS_LB_]], ptr [[DOTPREVIOUS_LB__ADDR]], align 8
6504 // CHECK13-NEXT: store i64 [[DOTPREVIOUS_UB_]], ptr [[DOTPREVIOUS_UB__ADDR]], align 8
6505 // CHECK13-NEXT: store ptr [[A]], ptr [[A_ADDR]], align 8
6506 // CHECK13-NEXT: store i64 [[DOTCAPTURE_EXPR_]], ptr [[DOTCAPTURE_EXPR__ADDR]], align 8
6507 // CHECK13-NEXT: [[TMP0:%.*]] = load ptr, ptr [[A_ADDR]], align 8
6508 // CHECK13-NEXT: store i32 0, ptr [[DOTOMP_LB]], align 4
6509 // CHECK13-NEXT: store i32 9, ptr [[DOTOMP_UB]], align 4
6510 // CHECK13-NEXT: [[TMP1:%.*]] = load i64, ptr [[DOTPREVIOUS_LB__ADDR]], align 8
6511 // CHECK13-NEXT: [[CONV:%.*]] = trunc i64 [[TMP1]] to i32
6512 // CHECK13-NEXT: [[TMP2:%.*]] = load i64, ptr [[DOTPREVIOUS_UB__ADDR]], align 8
6513 // CHECK13-NEXT: [[CONV1:%.*]] = trunc i64 [[TMP2]] to i32
6514 // CHECK13-NEXT: store i32 [[CONV]], ptr [[DOTOMP_LB]], align 4
6515 // CHECK13-NEXT: store i32 [[CONV1]], ptr [[DOTOMP_UB]], align 4
6516 // CHECK13-NEXT: store i32 1, ptr [[DOTOMP_STRIDE]], align 4
6517 // CHECK13-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
6518 // CHECK13-NEXT: [[TMP3:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR__ADDR]], align 4
6519 // CHECK13-NEXT: [[TMP4:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
6520 // CHECK13-NEXT: [[TMP5:%.*]] = load i32, ptr [[TMP4]], align 4
6521 // CHECK13-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB2]], i32 [[TMP5]], i32 33, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 [[TMP3]])
6522 // CHECK13-NEXT: br label [[OMP_DISPATCH_COND:%.*]]
6523 // CHECK13: omp.dispatch.cond:
6524 // CHECK13-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
6525 // CHECK13-NEXT: [[TMP7:%.*]] = load i64, ptr [[DOTPREVIOUS_UB__ADDR]], align 8
6526 // CHECK13-NEXT: [[CONV2:%.*]] = trunc i64 [[TMP7]] to i32
6527 // CHECK13-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP6]], [[CONV2]]
6528 // CHECK13-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
6529 // CHECK13: cond.true:
6530 // CHECK13-NEXT: [[TMP8:%.*]] = load i64, ptr [[DOTPREVIOUS_UB__ADDR]], align 8
6531 // CHECK13-NEXT: [[CONV3:%.*]] = trunc i64 [[TMP8]] to i32
6532 // CHECK13-NEXT: br label [[COND_END:%.*]]
6533 // CHECK13: cond.false:
6534 // CHECK13-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
6535 // CHECK13-NEXT: br label [[COND_END]]
6536 // CHECK13: cond.end:
6537 // CHECK13-NEXT: [[COND:%.*]] = phi i32 [ [[CONV3]], [[COND_TRUE]] ], [ [[TMP9]], [[COND_FALSE]] ]
6538 // CHECK13-NEXT: store i32 [[COND]], ptr [[DOTOMP_UB]], align 4
6539 // CHECK13-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4
6540 // CHECK13-NEXT: store i32 [[TMP10]], ptr [[DOTOMP_IV]], align 4
6541 // CHECK13-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
6542 // CHECK13-NEXT: [[TMP12:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
6543 // CHECK13-NEXT: [[CMP4:%.*]] = icmp sle i32 [[TMP11]], [[TMP12]]
6544 // CHECK13-NEXT: br i1 [[CMP4]], label [[OMP_DISPATCH_BODY:%.*]], label [[OMP_DISPATCH_END:%.*]]
6545 // CHECK13: omp.dispatch.body:
6546 // CHECK13-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
6547 // CHECK13: omp.inner.for.cond:
6548 // CHECK13-NEXT: [[TMP13:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
6549 // CHECK13-NEXT: [[TMP14:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
6550 // CHECK13-NEXT: [[CMP5:%.*]] = icmp sle i32 [[TMP13]], [[TMP14]]
6551 // CHECK13-NEXT: br i1 [[CMP5]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
6552 // CHECK13: omp.inner.for.body:
6553 // CHECK13-NEXT: [[TMP15:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
6554 // CHECK13-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP15]], 1
6555 // CHECK13-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
6556 // CHECK13-NEXT: store i32 [[ADD]], ptr [[I]], align 4
6557 // CHECK13-NEXT: [[TMP16:%.*]] = load i32, ptr [[I]], align 4
6558 // CHECK13-NEXT: [[IDXPROM:%.*]] = sext i32 [[TMP16]] to i64
6559 // CHECK13-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds [10 x i32], ptr [[TMP0]], i64 0, i64 [[IDXPROM]]
6560 // CHECK13-NEXT: store i32 0, ptr [[ARRAYIDX]], align 4
6561 // CHECK13-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
6562 // CHECK13: omp.body.continue:
6563 // CHECK13-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
6564 // CHECK13: omp.inner.for.inc:
6565 // CHECK13-NEXT: [[TMP17:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
6566 // CHECK13-NEXT: [[ADD6:%.*]] = add nsw i32 [[TMP17]], 1
6567 // CHECK13-NEXT: store i32 [[ADD6]], ptr [[DOTOMP_IV]], align 4
6568 // CHECK13-NEXT: br label [[OMP_INNER_FOR_COND]]
6569 // CHECK13: omp.inner.for.end:
6570 // CHECK13-NEXT: br label [[OMP_DISPATCH_INC:%.*]]
6571 // CHECK13: omp.dispatch.inc:
6572 // CHECK13-NEXT: [[TMP18:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4
6573 // CHECK13-NEXT: [[TMP19:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4
6574 // CHECK13-NEXT: [[ADD7:%.*]] = add nsw i32 [[TMP18]], [[TMP19]]
6575 // CHECK13-NEXT: store i32 [[ADD7]], ptr [[DOTOMP_LB]], align 4
6576 // CHECK13-NEXT: [[TMP20:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
6577 // CHECK13-NEXT: [[TMP21:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4
6578 // CHECK13-NEXT: [[ADD8:%.*]] = add nsw i32 [[TMP20]], [[TMP21]]
6579 // CHECK13-NEXT: store i32 [[ADD8]], ptr [[DOTOMP_UB]], align 4
6580 // CHECK13-NEXT: br label [[OMP_DISPATCH_COND]]
6581 // CHECK13: omp.dispatch.end:
6582 // CHECK13-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB2]], i32 [[TMP5]])
6583 // CHECK13-NEXT: ret void
6586 // CHECK13-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l131
6587 // CHECK13-SAME: (ptr noundef nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] {
6588 // CHECK13-NEXT: entry:
6589 // CHECK13-NEXT: [[A_ADDR:%.*]] = alloca ptr, align 8
6590 // CHECK13-NEXT: store ptr [[A]], ptr [[A_ADDR]], align 8
6591 // CHECK13-NEXT: [[TMP0:%.*]] = load ptr, ptr [[A_ADDR]], align 8
6592 // CHECK13-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_teams(ptr @[[GLOB3]], i32 1, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l131.omp_outlined, ptr [[TMP0]])
6593 // CHECK13-NEXT: ret void
6596 // CHECK13-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l131.omp_outlined
6597 // CHECK13-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], ptr noundef nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] {
6598 // CHECK13-NEXT: entry:
6599 // CHECK13-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 8
6600 // CHECK13-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 8
6601 // CHECK13-NEXT: [[A_ADDR:%.*]] = alloca ptr, align 8
6602 // CHECK13-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
6603 // CHECK13-NEXT: [[TMP:%.*]] = alloca i32, align 4
6604 // CHECK13-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4
6605 // CHECK13-NEXT: [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4
6606 // CHECK13-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
6607 // CHECK13-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
6608 // CHECK13-NEXT: [[I:%.*]] = alloca i32, align 4
6609 // CHECK13-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 8
6610 // CHECK13-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 8
6611 // CHECK13-NEXT: store ptr [[A]], ptr [[A_ADDR]], align 8
6612 // CHECK13-NEXT: [[TMP0:%.*]] = load ptr, ptr [[A_ADDR]], align 8
6613 // CHECK13-NEXT: store i32 0, ptr [[DOTOMP_COMB_LB]], align 4
6614 // CHECK13-NEXT: store i32 9, ptr [[DOTOMP_COMB_UB]], align 4
6615 // CHECK13-NEXT: store i32 1, ptr [[DOTOMP_STRIDE]], align 4
6616 // CHECK13-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
6617 // CHECK13-NEXT: [[TMP1:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
6618 // CHECK13-NEXT: [[TMP2:%.*]] = load i32, ptr [[TMP1]], align 4
6619 // CHECK13-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB1]], i32 [[TMP2]], i32 92, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
6620 // CHECK13-NEXT: [[TMP3:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
6621 // CHECK13-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP3]], 9
6622 // CHECK13-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
6623 // CHECK13: cond.true:
6624 // CHECK13-NEXT: br label [[COND_END:%.*]]
6625 // CHECK13: cond.false:
6626 // CHECK13-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
6627 // CHECK13-NEXT: br label [[COND_END]]
6628 // CHECK13: cond.end:
6629 // CHECK13-NEXT: [[COND:%.*]] = phi i32 [ 9, [[COND_TRUE]] ], [ [[TMP4]], [[COND_FALSE]] ]
6630 // CHECK13-NEXT: store i32 [[COND]], ptr [[DOTOMP_COMB_UB]], align 4
6631 // CHECK13-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
6632 // CHECK13-NEXT: store i32 [[TMP5]], ptr [[DOTOMP_IV]], align 4
6633 // CHECK13-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
6634 // CHECK13: omp.inner.for.cond:
6635 // CHECK13-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
6636 // CHECK13-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
6637 // CHECK13-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP6]], [[TMP7]]
6638 // CHECK13-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
6639 // CHECK13: omp.inner.for.body:
6640 // CHECK13-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
6641 // CHECK13-NEXT: [[TMP9:%.*]] = zext i32 [[TMP8]] to i64
6642 // CHECK13-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
6643 // CHECK13-NEXT: [[TMP11:%.*]] = zext i32 [[TMP10]] to i64
6644 // CHECK13-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_call(ptr @[[GLOB3]], i32 3, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l131.omp_outlined.omp_outlined, i64 [[TMP9]], i64 [[TMP11]], ptr [[TMP0]])
6645 // CHECK13-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
6646 // CHECK13: omp.inner.for.inc:
6647 // CHECK13-NEXT: [[TMP12:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
6648 // CHECK13-NEXT: [[TMP13:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4
6649 // CHECK13-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP12]], [[TMP13]]
6650 // CHECK13-NEXT: store i32 [[ADD]], ptr [[DOTOMP_IV]], align 4
6651 // CHECK13-NEXT: br label [[OMP_INNER_FOR_COND]]
6652 // CHECK13: omp.inner.for.end:
6653 // CHECK13-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
6654 // CHECK13: omp.loop.exit:
6655 // CHECK13-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB1]], i32 [[TMP2]])
6656 // CHECK13-NEXT: ret void
6659 // CHECK13-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l131.omp_outlined.omp_outlined
6660 // CHECK13-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], i64 noundef [[DOTPREVIOUS_LB_:%.*]], i64 noundef [[DOTPREVIOUS_UB_:%.*]], ptr noundef nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] {
6661 // CHECK13-NEXT: entry:
6662 // CHECK13-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 8
6663 // CHECK13-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 8
6664 // CHECK13-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i64, align 8
6665 // CHECK13-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i64, align 8
6666 // CHECK13-NEXT: [[A_ADDR:%.*]] = alloca ptr, align 8
6667 // CHECK13-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
6668 // CHECK13-NEXT: [[TMP:%.*]] = alloca i32, align 4
6669 // CHECK13-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4
6670 // CHECK13-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4
6671 // CHECK13-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
6672 // CHECK13-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
6673 // CHECK13-NEXT: [[I:%.*]] = alloca i32, align 4
6674 // CHECK13-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 8
6675 // CHECK13-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 8
6676 // CHECK13-NEXT: store i64 [[DOTPREVIOUS_LB_]], ptr [[DOTPREVIOUS_LB__ADDR]], align 8
6677 // CHECK13-NEXT: store i64 [[DOTPREVIOUS_UB_]], ptr [[DOTPREVIOUS_UB__ADDR]], align 8
6678 // CHECK13-NEXT: store ptr [[A]], ptr [[A_ADDR]], align 8
6679 // CHECK13-NEXT: [[TMP0:%.*]] = load ptr, ptr [[A_ADDR]], align 8
6680 // CHECK13-NEXT: store i32 0, ptr [[DOTOMP_LB]], align 4
6681 // CHECK13-NEXT: store i32 9, ptr [[DOTOMP_UB]], align 4
6682 // CHECK13-NEXT: [[TMP1:%.*]] = load i64, ptr [[DOTPREVIOUS_LB__ADDR]], align 8
6683 // CHECK13-NEXT: [[CONV:%.*]] = trunc i64 [[TMP1]] to i32
6684 // CHECK13-NEXT: [[TMP2:%.*]] = load i64, ptr [[DOTPREVIOUS_UB__ADDR]], align 8
6685 // CHECK13-NEXT: [[CONV1:%.*]] = trunc i64 [[TMP2]] to i32
6686 // CHECK13-NEXT: store i32 [[CONV]], ptr [[DOTOMP_LB]], align 4
6687 // CHECK13-NEXT: store i32 [[CONV1]], ptr [[DOTOMP_UB]], align 4
6688 // CHECK13-NEXT: store i32 1, ptr [[DOTOMP_STRIDE]], align 4
6689 // CHECK13-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
6690 // CHECK13-NEXT: [[TMP3:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4
6691 // CHECK13-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
6692 // CHECK13-NEXT: [[TMP5:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
6693 // CHECK13-NEXT: [[TMP6:%.*]] = load i32, ptr [[TMP5]], align 4
6694 // CHECK13-NEXT: call void @__kmpc_dispatch_init_4(ptr @[[GLOB3]], i32 [[TMP6]], i32 1073741859, i32 [[TMP3]], i32 [[TMP4]], i32 1, i32 1)
6695 // CHECK13-NEXT: br label [[OMP_DISPATCH_COND:%.*]]
6696 // CHECK13: omp.dispatch.cond:
6697 // CHECK13-NEXT: [[TMP7:%.*]] = call i32 @__kmpc_dispatch_next_4(ptr @[[GLOB3]], i32 [[TMP6]], ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]])
6698 // CHECK13-NEXT: [[TOBOOL:%.*]] = icmp ne i32 [[TMP7]], 0
6699 // CHECK13-NEXT: br i1 [[TOBOOL]], label [[OMP_DISPATCH_BODY:%.*]], label [[OMP_DISPATCH_END:%.*]]
6700 // CHECK13: omp.dispatch.body:
6701 // CHECK13-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4
6702 // CHECK13-NEXT: store i32 [[TMP8]], ptr [[DOTOMP_IV]], align 4
6703 // CHECK13-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
6704 // CHECK13: omp.inner.for.cond:
6705 // CHECK13-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP21:![0-9]+]]
6706 // CHECK13-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP21]]
6707 // CHECK13-NEXT: [[CMP:%.*]] = icmp sle i32 [[TMP9]], [[TMP10]]
6708 // CHECK13-NEXT: br i1 [[CMP]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
6709 // CHECK13: omp.inner.for.body:
6710 // CHECK13-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP21]]
6711 // CHECK13-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP11]], 1
6712 // CHECK13-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
6713 // CHECK13-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP21]]
6714 // CHECK13-NEXT: [[TMP12:%.*]] = load i32, ptr [[I]], align 4, !llvm.access.group [[ACC_GRP21]]
6715 // CHECK13-NEXT: [[IDXPROM:%.*]] = sext i32 [[TMP12]] to i64
6716 // CHECK13-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds [10 x i32], ptr [[TMP0]], i64 0, i64 [[IDXPROM]]
6717 // CHECK13-NEXT: store i32 0, ptr [[ARRAYIDX]], align 4, !llvm.access.group [[ACC_GRP21]]
6718 // CHECK13-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
6719 // CHECK13: omp.body.continue:
6720 // CHECK13-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
6721 // CHECK13: omp.inner.for.inc:
6722 // CHECK13-NEXT: [[TMP13:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP21]]
6723 // CHECK13-NEXT: [[ADD2:%.*]] = add nsw i32 [[TMP13]], 1
6724 // CHECK13-NEXT: store i32 [[ADD2]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP21]]
6725 // CHECK13-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP22:![0-9]+]]
6726 // CHECK13: omp.inner.for.end:
6727 // CHECK13-NEXT: br label [[OMP_DISPATCH_INC:%.*]]
6728 // CHECK13: omp.dispatch.inc:
6729 // CHECK13-NEXT: br label [[OMP_DISPATCH_COND]]
6730 // CHECK13: omp.dispatch.end:
6731 // CHECK13-NEXT: call void @__kmpc_dispatch_deinit(ptr @[[GLOB3]], i32 [[TMP6]])
6732 // CHECK13-NEXT: ret void
6735 // CHECK13-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l136
6736 // CHECK13-SAME: (i64 noundef [[M:%.*]], ptr noundef nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] {
6737 // CHECK13-NEXT: entry:
6738 // CHECK13-NEXT: [[M_ADDR:%.*]] = alloca i64, align 8
6739 // CHECK13-NEXT: [[A_ADDR:%.*]] = alloca ptr, align 8
6740 // CHECK13-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4
6741 // CHECK13-NEXT: [[DOTCAPTURE_EXPR__CASTED:%.*]] = alloca i64, align 8
6742 // CHECK13-NEXT: store i64 [[M]], ptr [[M_ADDR]], align 8
6743 // CHECK13-NEXT: store ptr [[A]], ptr [[A_ADDR]], align 8
6744 // CHECK13-NEXT: [[TMP0:%.*]] = load ptr, ptr [[A_ADDR]], align 8
6745 // CHECK13-NEXT: [[TMP1:%.*]] = load i32, ptr [[M_ADDR]], align 4
6746 // CHECK13-NEXT: store i32 [[TMP1]], ptr [[DOTCAPTURE_EXPR_]], align 4
6747 // CHECK13-NEXT: [[TMP2:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4
6748 // CHECK13-NEXT: store i32 [[TMP2]], ptr [[DOTCAPTURE_EXPR__CASTED]], align 4
6749 // CHECK13-NEXT: [[TMP3:%.*]] = load i64, ptr [[DOTCAPTURE_EXPR__CASTED]], align 8
6750 // CHECK13-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_teams(ptr @[[GLOB3]], i32 2, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l136.omp_outlined, ptr [[TMP0]], i64 [[TMP3]])
6751 // CHECK13-NEXT: ret void
6754 // CHECK13-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l136.omp_outlined
6755 // CHECK13-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], ptr noundef nonnull align 4 dereferenceable(40) [[A:%.*]], i64 noundef [[DOTCAPTURE_EXPR_:%.*]]) #[[ATTR2]] {
6756 // CHECK13-NEXT: entry:
6757 // CHECK13-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 8
6758 // CHECK13-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 8
6759 // CHECK13-NEXT: [[A_ADDR:%.*]] = alloca ptr, align 8
6760 // CHECK13-NEXT: [[DOTCAPTURE_EXPR__ADDR:%.*]] = alloca i64, align 8
6761 // CHECK13-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
6762 // CHECK13-NEXT: [[TMP:%.*]] = alloca i32, align 4
6763 // CHECK13-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4
6764 // CHECK13-NEXT: [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4
6765 // CHECK13-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
6766 // CHECK13-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
6767 // CHECK13-NEXT: [[I:%.*]] = alloca i32, align 4
6768 // CHECK13-NEXT: [[DOTCAPTURE_EXPR__CASTED:%.*]] = alloca i64, align 8
6769 // CHECK13-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 8
6770 // CHECK13-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 8
6771 // CHECK13-NEXT: store ptr [[A]], ptr [[A_ADDR]], align 8
6772 // CHECK13-NEXT: store i64 [[DOTCAPTURE_EXPR_]], ptr [[DOTCAPTURE_EXPR__ADDR]], align 8
6773 // CHECK13-NEXT: [[TMP0:%.*]] = load ptr, ptr [[A_ADDR]], align 8
6774 // CHECK13-NEXT: store i32 0, ptr [[DOTOMP_COMB_LB]], align 4
6775 // CHECK13-NEXT: store i32 9, ptr [[DOTOMP_COMB_UB]], align 4
6776 // CHECK13-NEXT: store i32 1, ptr [[DOTOMP_STRIDE]], align 4
6777 // CHECK13-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
6778 // CHECK13-NEXT: [[TMP1:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
6779 // CHECK13-NEXT: [[TMP2:%.*]] = load i32, ptr [[TMP1]], align 4
6780 // CHECK13-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB1]], i32 [[TMP2]], i32 92, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
6781 // CHECK13-NEXT: [[TMP3:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
6782 // CHECK13-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP3]], 9
6783 // CHECK13-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
6784 // CHECK13: cond.true:
6785 // CHECK13-NEXT: br label [[COND_END:%.*]]
6786 // CHECK13: cond.false:
6787 // CHECK13-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
6788 // CHECK13-NEXT: br label [[COND_END]]
6789 // CHECK13: cond.end:
6790 // CHECK13-NEXT: [[COND:%.*]] = phi i32 [ 9, [[COND_TRUE]] ], [ [[TMP4]], [[COND_FALSE]] ]
6791 // CHECK13-NEXT: store i32 [[COND]], ptr [[DOTOMP_COMB_UB]], align 4
6792 // CHECK13-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
6793 // CHECK13-NEXT: store i32 [[TMP5]], ptr [[DOTOMP_IV]], align 4
6794 // CHECK13-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
6795 // CHECK13: omp.inner.for.cond:
6796 // CHECK13-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
6797 // CHECK13-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
6798 // CHECK13-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP6]], [[TMP7]]
6799 // CHECK13-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
6800 // CHECK13: omp.inner.for.body:
6801 // CHECK13-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
6802 // CHECK13-NEXT: [[TMP9:%.*]] = zext i32 [[TMP8]] to i64
6803 // CHECK13-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
6804 // CHECK13-NEXT: [[TMP11:%.*]] = zext i32 [[TMP10]] to i64
6805 // CHECK13-NEXT: [[TMP12:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR__ADDR]], align 4
6806 // CHECK13-NEXT: store i32 [[TMP12]], ptr [[DOTCAPTURE_EXPR__CASTED]], align 4
6807 // CHECK13-NEXT: [[TMP13:%.*]] = load i64, ptr [[DOTCAPTURE_EXPR__CASTED]], align 8
6808 // CHECK13-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_call(ptr @[[GLOB3]], i32 4, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l136.omp_outlined.omp_outlined, i64 [[TMP9]], i64 [[TMP11]], ptr [[TMP0]], i64 [[TMP13]])
6809 // CHECK13-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
6810 // CHECK13: omp.inner.for.inc:
6811 // CHECK13-NEXT: [[TMP14:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
6812 // CHECK13-NEXT: [[TMP15:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4
6813 // CHECK13-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP14]], [[TMP15]]
6814 // CHECK13-NEXT: store i32 [[ADD]], ptr [[DOTOMP_IV]], align 4
6815 // CHECK13-NEXT: br label [[OMP_INNER_FOR_COND]]
6816 // CHECK13: omp.inner.for.end:
6817 // CHECK13-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
6818 // CHECK13: omp.loop.exit:
6819 // CHECK13-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB1]], i32 [[TMP2]])
6820 // CHECK13-NEXT: ret void
6823 // CHECK13-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l136.omp_outlined.omp_outlined
6824 // CHECK13-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], i64 noundef [[DOTPREVIOUS_LB_:%.*]], i64 noundef [[DOTPREVIOUS_UB_:%.*]], ptr noundef nonnull align 4 dereferenceable(40) [[A:%.*]], i64 noundef [[DOTCAPTURE_EXPR_:%.*]]) #[[ATTR2]] {
6825 // CHECK13-NEXT: entry:
6826 // CHECK13-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 8
6827 // CHECK13-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 8
6828 // CHECK13-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i64, align 8
6829 // CHECK13-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i64, align 8
6830 // CHECK13-NEXT: [[A_ADDR:%.*]] = alloca ptr, align 8
6831 // CHECK13-NEXT: [[DOTCAPTURE_EXPR__ADDR:%.*]] = alloca i64, align 8
6832 // CHECK13-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
6833 // CHECK13-NEXT: [[TMP:%.*]] = alloca i32, align 4
6834 // CHECK13-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4
6835 // CHECK13-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4
6836 // CHECK13-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
6837 // CHECK13-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
6838 // CHECK13-NEXT: [[I:%.*]] = alloca i32, align 4
6839 // CHECK13-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 8
6840 // CHECK13-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 8
6841 // CHECK13-NEXT: store i64 [[DOTPREVIOUS_LB_]], ptr [[DOTPREVIOUS_LB__ADDR]], align 8
6842 // CHECK13-NEXT: store i64 [[DOTPREVIOUS_UB_]], ptr [[DOTPREVIOUS_UB__ADDR]], align 8
6843 // CHECK13-NEXT: store ptr [[A]], ptr [[A_ADDR]], align 8
6844 // CHECK13-NEXT: store i64 [[DOTCAPTURE_EXPR_]], ptr [[DOTCAPTURE_EXPR__ADDR]], align 8
6845 // CHECK13-NEXT: [[TMP0:%.*]] = load ptr, ptr [[A_ADDR]], align 8
6846 // CHECK13-NEXT: store i32 0, ptr [[DOTOMP_LB]], align 4
6847 // CHECK13-NEXT: store i32 9, ptr [[DOTOMP_UB]], align 4
6848 // CHECK13-NEXT: [[TMP1:%.*]] = load i64, ptr [[DOTPREVIOUS_LB__ADDR]], align 8
6849 // CHECK13-NEXT: [[CONV:%.*]] = trunc i64 [[TMP1]] to i32
6850 // CHECK13-NEXT: [[TMP2:%.*]] = load i64, ptr [[DOTPREVIOUS_UB__ADDR]], align 8
6851 // CHECK13-NEXT: [[CONV1:%.*]] = trunc i64 [[TMP2]] to i32
6852 // CHECK13-NEXT: store i32 [[CONV]], ptr [[DOTOMP_LB]], align 4
6853 // CHECK13-NEXT: store i32 [[CONV1]], ptr [[DOTOMP_UB]], align 4
6854 // CHECK13-NEXT: store i32 1, ptr [[DOTOMP_STRIDE]], align 4
6855 // CHECK13-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
6856 // CHECK13-NEXT: [[TMP3:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR__ADDR]], align 4
6857 // CHECK13-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4
6858 // CHECK13-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
6859 // CHECK13-NEXT: [[TMP6:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
6860 // CHECK13-NEXT: [[TMP7:%.*]] = load i32, ptr [[TMP6]], align 4
6861 // CHECK13-NEXT: call void @__kmpc_dispatch_init_4(ptr @[[GLOB3]], i32 [[TMP7]], i32 1073741859, i32 [[TMP4]], i32 [[TMP5]], i32 1, i32 [[TMP3]])
6862 // CHECK13-NEXT: br label [[OMP_DISPATCH_COND:%.*]]
6863 // CHECK13: omp.dispatch.cond:
6864 // CHECK13-NEXT: [[TMP8:%.*]] = call i32 @__kmpc_dispatch_next_4(ptr @[[GLOB3]], i32 [[TMP7]], ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]])
6865 // CHECK13-NEXT: [[TOBOOL:%.*]] = icmp ne i32 [[TMP8]], 0
6866 // CHECK13-NEXT: br i1 [[TOBOOL]], label [[OMP_DISPATCH_BODY:%.*]], label [[OMP_DISPATCH_END:%.*]]
6867 // CHECK13: omp.dispatch.body:
6868 // CHECK13-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4
6869 // CHECK13-NEXT: store i32 [[TMP9]], ptr [[DOTOMP_IV]], align 4
6870 // CHECK13-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
6871 // CHECK13: omp.inner.for.cond:
6872 // CHECK13-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP24:![0-9]+]]
6873 // CHECK13-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP24]]
6874 // CHECK13-NEXT: [[CMP:%.*]] = icmp sle i32 [[TMP10]], [[TMP11]]
6875 // CHECK13-NEXT: br i1 [[CMP]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
6876 // CHECK13: omp.inner.for.body:
6877 // CHECK13-NEXT: [[TMP12:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP24]]
6878 // CHECK13-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP12]], 1
6879 // CHECK13-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
6880 // CHECK13-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP24]]
6881 // CHECK13-NEXT: [[TMP13:%.*]] = load i32, ptr [[I]], align 4, !llvm.access.group [[ACC_GRP24]]
6882 // CHECK13-NEXT: [[IDXPROM:%.*]] = sext i32 [[TMP13]] to i64
6883 // CHECK13-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds [10 x i32], ptr [[TMP0]], i64 0, i64 [[IDXPROM]]
6884 // CHECK13-NEXT: store i32 0, ptr [[ARRAYIDX]], align 4, !llvm.access.group [[ACC_GRP24]]
6885 // CHECK13-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
6886 // CHECK13: omp.body.continue:
6887 // CHECK13-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
6888 // CHECK13: omp.inner.for.inc:
6889 // CHECK13-NEXT: [[TMP14:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP24]]
6890 // CHECK13-NEXT: [[ADD2:%.*]] = add nsw i32 [[TMP14]], 1
6891 // CHECK13-NEXT: store i32 [[ADD2]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP24]]
6892 // CHECK13-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP25:![0-9]+]]
6893 // CHECK13: omp.inner.for.end:
6894 // CHECK13-NEXT: br label [[OMP_DISPATCH_INC:%.*]]
6895 // CHECK13: omp.dispatch.inc:
6896 // CHECK13-NEXT: br label [[OMP_DISPATCH_COND]]
6897 // CHECK13: omp.dispatch.end:
6898 // CHECK13-NEXT: call void @__kmpc_dispatch_deinit(ptr @[[GLOB3]], i32 [[TMP7]])
6899 // CHECK13-NEXT: ret void
6902 // CHECK15-LABEL: define {{[^@]+}}@main
6903 // CHECK15-SAME: (i32 noundef [[ARGC:%.*]], ptr noundef [[ARGV:%.*]]) #[[ATTR0:[0-9]+]] {
6904 // CHECK15-NEXT: entry:
6905 // CHECK15-NEXT: [[RETVAL:%.*]] = alloca i32, align 4
6906 // CHECK15-NEXT: [[ARGC_ADDR:%.*]] = alloca i32, align 4
6907 // CHECK15-NEXT: [[ARGV_ADDR:%.*]] = alloca ptr, align 4
6908 // CHECK15-NEXT: [[N:%.*]] = alloca i32, align 4
6909 // CHECK15-NEXT: [[SAVED_STACK:%.*]] = alloca ptr, align 4
6910 // CHECK15-NEXT: [[__VLA_EXPR0:%.*]] = alloca i32, align 4
6911 // CHECK15-NEXT: [[M:%.*]] = alloca i32, align 4
6912 // CHECK15-NEXT: [[N_CASTED:%.*]] = alloca i32, align 4
6913 // CHECK15-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [3 x ptr], align 4
6914 // CHECK15-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [3 x ptr], align 4
6915 // CHECK15-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [3 x ptr], align 4
6916 // CHECK15-NEXT: [[DOTOFFLOAD_SIZES:%.*]] = alloca [3 x i64], align 4
6917 // CHECK15-NEXT: [[TMP:%.*]] = alloca i32, align 4
6918 // CHECK15-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4
6919 // CHECK15-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4
6920 // CHECK15-NEXT: [[KERNEL_ARGS:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS:%.*]], align 8
6921 // CHECK15-NEXT: [[N_CASTED3:%.*]] = alloca i32, align 4
6922 // CHECK15-NEXT: [[DOTOFFLOAD_BASEPTRS4:%.*]] = alloca [3 x ptr], align 4
6923 // CHECK15-NEXT: [[DOTOFFLOAD_PTRS5:%.*]] = alloca [3 x ptr], align 4
6924 // CHECK15-NEXT: [[DOTOFFLOAD_MAPPERS6:%.*]] = alloca [3 x ptr], align 4
6925 // CHECK15-NEXT: [[DOTOFFLOAD_SIZES7:%.*]] = alloca [3 x i64], align 4
6926 // CHECK15-NEXT: [[_TMP8:%.*]] = alloca i32, align 4
6927 // CHECK15-NEXT: [[DOTCAPTURE_EXPR_9:%.*]] = alloca i32, align 4
6928 // CHECK15-NEXT: [[DOTCAPTURE_EXPR_10:%.*]] = alloca i32, align 4
6929 // CHECK15-NEXT: [[KERNEL_ARGS15:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS]], align 8
6930 // CHECK15-NEXT: [[M_CASTED:%.*]] = alloca i32, align 4
6931 // CHECK15-NEXT: [[N_CASTED18:%.*]] = alloca i32, align 4
6932 // CHECK15-NEXT: [[DOTOFFLOAD_BASEPTRS19:%.*]] = alloca [4 x ptr], align 4
6933 // CHECK15-NEXT: [[DOTOFFLOAD_PTRS20:%.*]] = alloca [4 x ptr], align 4
6934 // CHECK15-NEXT: [[DOTOFFLOAD_MAPPERS21:%.*]] = alloca [4 x ptr], align 4
6935 // CHECK15-NEXT: [[DOTOFFLOAD_SIZES22:%.*]] = alloca [4 x i64], align 4
6936 // CHECK15-NEXT: [[_TMP23:%.*]] = alloca i32, align 4
6937 // CHECK15-NEXT: [[DOTCAPTURE_EXPR_24:%.*]] = alloca i32, align 4
6938 // CHECK15-NEXT: [[DOTCAPTURE_EXPR_25:%.*]] = alloca i32, align 4
6939 // CHECK15-NEXT: [[KERNEL_ARGS30:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS]], align 8
6940 // CHECK15-NEXT: [[N_CASTED33:%.*]] = alloca i32, align 4
6941 // CHECK15-NEXT: [[DOTOFFLOAD_BASEPTRS34:%.*]] = alloca [3 x ptr], align 4
6942 // CHECK15-NEXT: [[DOTOFFLOAD_PTRS35:%.*]] = alloca [3 x ptr], align 4
6943 // CHECK15-NEXT: [[DOTOFFLOAD_MAPPERS36:%.*]] = alloca [3 x ptr], align 4
6944 // CHECK15-NEXT: [[DOTOFFLOAD_SIZES37:%.*]] = alloca [3 x i64], align 4
6945 // CHECK15-NEXT: [[_TMP38:%.*]] = alloca i32, align 4
6946 // CHECK15-NEXT: [[DOTCAPTURE_EXPR_39:%.*]] = alloca i32, align 4
6947 // CHECK15-NEXT: [[DOTCAPTURE_EXPR_40:%.*]] = alloca i32, align 4
6948 // CHECK15-NEXT: [[KERNEL_ARGS45:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS]], align 8
6949 // CHECK15-NEXT: [[M_CASTED48:%.*]] = alloca i32, align 4
6950 // CHECK15-NEXT: [[N_CASTED49:%.*]] = alloca i32, align 4
6951 // CHECK15-NEXT: [[DOTOFFLOAD_BASEPTRS50:%.*]] = alloca [4 x ptr], align 4
6952 // CHECK15-NEXT: [[DOTOFFLOAD_PTRS51:%.*]] = alloca [4 x ptr], align 4
6953 // CHECK15-NEXT: [[DOTOFFLOAD_MAPPERS52:%.*]] = alloca [4 x ptr], align 4
6954 // CHECK15-NEXT: [[DOTOFFLOAD_SIZES53:%.*]] = alloca [4 x i64], align 4
6955 // CHECK15-NEXT: [[_TMP54:%.*]] = alloca i32, align 4
6956 // CHECK15-NEXT: [[DOTCAPTURE_EXPR_55:%.*]] = alloca i32, align 4
6957 // CHECK15-NEXT: [[DOTCAPTURE_EXPR_56:%.*]] = alloca i32, align 4
6958 // CHECK15-NEXT: [[KERNEL_ARGS61:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS]], align 8
6959 // CHECK15-NEXT: store i32 0, ptr [[RETVAL]], align 4
6960 // CHECK15-NEXT: store i32 [[ARGC]], ptr [[ARGC_ADDR]], align 4
6961 // CHECK15-NEXT: store ptr [[ARGV]], ptr [[ARGV_ADDR]], align 4
6962 // CHECK15-NEXT: store i32 100, ptr [[N]], align 4
6963 // CHECK15-NEXT: [[TMP0:%.*]] = load i32, ptr [[N]], align 4
6964 // CHECK15-NEXT: [[TMP1:%.*]] = call ptr @llvm.stacksave.p0()
6965 // CHECK15-NEXT: store ptr [[TMP1]], ptr [[SAVED_STACK]], align 4
6966 // CHECK15-NEXT: [[VLA:%.*]] = alloca i32, i32 [[TMP0]], align 4
6967 // CHECK15-NEXT: store i32 [[TMP0]], ptr [[__VLA_EXPR0]], align 4
6968 // CHECK15-NEXT: store i32 10, ptr [[M]], align 4
6969 // CHECK15-NEXT: [[TMP2:%.*]] = load i32, ptr [[N]], align 4
6970 // CHECK15-NEXT: store i32 [[TMP2]], ptr [[N_CASTED]], align 4
6971 // CHECK15-NEXT: [[TMP3:%.*]] = load i32, ptr [[N_CASTED]], align 4
6972 // CHECK15-NEXT: [[TMP4:%.*]] = mul nuw i32 [[TMP0]], 4
6973 // CHECK15-NEXT: [[TMP5:%.*]] = sext i32 [[TMP4]] to i64
6974 // CHECK15-NEXT: call void @llvm.memcpy.p0.p0.i32(ptr align 4 [[DOTOFFLOAD_SIZES]], ptr align 4 @.offload_sizes, i32 24, i1 false)
6975 // CHECK15-NEXT: [[TMP6:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
6976 // CHECK15-NEXT: store i32 [[TMP3]], ptr [[TMP6]], align 4
6977 // CHECK15-NEXT: [[TMP7:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
6978 // CHECK15-NEXT: store i32 [[TMP3]], ptr [[TMP7]], align 4
6979 // CHECK15-NEXT: [[TMP8:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0
6980 // CHECK15-NEXT: store ptr null, ptr [[TMP8]], align 4
6981 // CHECK15-NEXT: [[TMP9:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1
6982 // CHECK15-NEXT: store i32 [[TMP0]], ptr [[TMP9]], align 4
6983 // CHECK15-NEXT: [[TMP10:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 1
6984 // CHECK15-NEXT: store i32 [[TMP0]], ptr [[TMP10]], align 4
6985 // CHECK15-NEXT: [[TMP11:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i32 0, i32 1
6986 // CHECK15-NEXT: store ptr null, ptr [[TMP11]], align 4
6987 // CHECK15-NEXT: [[TMP12:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
6988 // CHECK15-NEXT: store ptr [[VLA]], ptr [[TMP12]], align 4
6989 // CHECK15-NEXT: [[TMP13:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 2
6990 // CHECK15-NEXT: store ptr [[VLA]], ptr [[TMP13]], align 4
6991 // CHECK15-NEXT: [[TMP14:%.*]] = getelementptr inbounds [3 x i64], ptr [[DOTOFFLOAD_SIZES]], i32 0, i32 2
6992 // CHECK15-NEXT: store i64 [[TMP5]], ptr [[TMP14]], align 4
6993 // CHECK15-NEXT: [[TMP15:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i32 0, i32 2
6994 // CHECK15-NEXT: store ptr null, ptr [[TMP15]], align 4
6995 // CHECK15-NEXT: [[TMP16:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
6996 // CHECK15-NEXT: [[TMP17:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
6997 // CHECK15-NEXT: [[TMP18:%.*]] = getelementptr inbounds [3 x i64], ptr [[DOTOFFLOAD_SIZES]], i32 0, i32 0
6998 // CHECK15-NEXT: [[TMP19:%.*]] = load i32, ptr [[N]], align 4
6999 // CHECK15-NEXT: store i32 [[TMP19]], ptr [[DOTCAPTURE_EXPR_]], align 4
7000 // CHECK15-NEXT: [[TMP20:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4
7001 // CHECK15-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP20]], 0
7002 // CHECK15-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1
7003 // CHECK15-NEXT: [[SUB2:%.*]] = sub nsw i32 [[DIV]], 1
7004 // CHECK15-NEXT: store i32 [[SUB2]], ptr [[DOTCAPTURE_EXPR_1]], align 4
7005 // CHECK15-NEXT: [[TMP21:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4
7006 // CHECK15-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP21]], 1
7007 // CHECK15-NEXT: [[TMP22:%.*]] = zext i32 [[ADD]] to i64
7008 // CHECK15-NEXT: [[TMP23:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 0
7009 // CHECK15-NEXT: store i32 3, ptr [[TMP23]], align 4
7010 // CHECK15-NEXT: [[TMP24:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 1
7011 // CHECK15-NEXT: store i32 3, ptr [[TMP24]], align 4
7012 // CHECK15-NEXT: [[TMP25:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 2
7013 // CHECK15-NEXT: store ptr [[TMP16]], ptr [[TMP25]], align 4
7014 // CHECK15-NEXT: [[TMP26:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 3
7015 // CHECK15-NEXT: store ptr [[TMP17]], ptr [[TMP26]], align 4
7016 // CHECK15-NEXT: [[TMP27:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 4
7017 // CHECK15-NEXT: store ptr [[TMP18]], ptr [[TMP27]], align 4
7018 // CHECK15-NEXT: [[TMP28:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 5
7019 // CHECK15-NEXT: store ptr @.offload_maptypes, ptr [[TMP28]], align 4
7020 // CHECK15-NEXT: [[TMP29:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 6
7021 // CHECK15-NEXT: store ptr null, ptr [[TMP29]], align 4
7022 // CHECK15-NEXT: [[TMP30:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 7
7023 // CHECK15-NEXT: store ptr null, ptr [[TMP30]], align 4
7024 // CHECK15-NEXT: [[TMP31:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 8
7025 // CHECK15-NEXT: store i64 [[TMP22]], ptr [[TMP31]], align 8
7026 // CHECK15-NEXT: [[TMP32:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 9
7027 // CHECK15-NEXT: store i64 0, ptr [[TMP32]], align 8
7028 // CHECK15-NEXT: [[TMP33:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 10
7029 // CHECK15-NEXT: store [3 x i32] zeroinitializer, ptr [[TMP33]], align 4
7030 // CHECK15-NEXT: [[TMP34:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 11
7031 // CHECK15-NEXT: store [3 x i32] zeroinitializer, ptr [[TMP34]], align 4
7032 // CHECK15-NEXT: [[TMP35:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 12
7033 // CHECK15-NEXT: store i32 0, ptr [[TMP35]], align 4
7034 // CHECK15-NEXT: [[TMP36:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB3:[0-9]+]], i64 -1, i32 0, i32 0, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l148.region_id, ptr [[KERNEL_ARGS]])
7035 // CHECK15-NEXT: [[TMP37:%.*]] = icmp ne i32 [[TMP36]], 0
7036 // CHECK15-NEXT: br i1 [[TMP37]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
7037 // CHECK15: omp_offload.failed:
7038 // CHECK15-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l148(i32 [[TMP3]], i32 [[TMP0]], ptr [[VLA]]) #[[ATTR3:[0-9]+]]
7039 // CHECK15-NEXT: br label [[OMP_OFFLOAD_CONT]]
7040 // CHECK15: omp_offload.cont:
7041 // CHECK15-NEXT: [[TMP38:%.*]] = load i32, ptr [[N]], align 4
7042 // CHECK15-NEXT: store i32 [[TMP38]], ptr [[N_CASTED3]], align 4
7043 // CHECK15-NEXT: [[TMP39:%.*]] = load i32, ptr [[N_CASTED3]], align 4
7044 // CHECK15-NEXT: [[TMP40:%.*]] = mul nuw i32 [[TMP0]], 4
7045 // CHECK15-NEXT: [[TMP41:%.*]] = sext i32 [[TMP40]] to i64
7046 // CHECK15-NEXT: call void @llvm.memcpy.p0.p0.i32(ptr align 4 [[DOTOFFLOAD_SIZES7]], ptr align 4 @.offload_sizes.1, i32 24, i1 false)
7047 // CHECK15-NEXT: [[TMP42:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_BASEPTRS4]], i32 0, i32 0
7048 // CHECK15-NEXT: store i32 [[TMP39]], ptr [[TMP42]], align 4
7049 // CHECK15-NEXT: [[TMP43:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_PTRS5]], i32 0, i32 0
7050 // CHECK15-NEXT: store i32 [[TMP39]], ptr [[TMP43]], align 4
7051 // CHECK15-NEXT: [[TMP44:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_MAPPERS6]], i32 0, i32 0
7052 // CHECK15-NEXT: store ptr null, ptr [[TMP44]], align 4
7053 // CHECK15-NEXT: [[TMP45:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_BASEPTRS4]], i32 0, i32 1
7054 // CHECK15-NEXT: store i32 [[TMP0]], ptr [[TMP45]], align 4
7055 // CHECK15-NEXT: [[TMP46:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_PTRS5]], i32 0, i32 1
7056 // CHECK15-NEXT: store i32 [[TMP0]], ptr [[TMP46]], align 4
7057 // CHECK15-NEXT: [[TMP47:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_MAPPERS6]], i32 0, i32 1
7058 // CHECK15-NEXT: store ptr null, ptr [[TMP47]], align 4
7059 // CHECK15-NEXT: [[TMP48:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_BASEPTRS4]], i32 0, i32 2
7060 // CHECK15-NEXT: store ptr [[VLA]], ptr [[TMP48]], align 4
7061 // CHECK15-NEXT: [[TMP49:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_PTRS5]], i32 0, i32 2
7062 // CHECK15-NEXT: store ptr [[VLA]], ptr [[TMP49]], align 4
7063 // CHECK15-NEXT: [[TMP50:%.*]] = getelementptr inbounds [3 x i64], ptr [[DOTOFFLOAD_SIZES7]], i32 0, i32 2
7064 // CHECK15-NEXT: store i64 [[TMP41]], ptr [[TMP50]], align 4
7065 // CHECK15-NEXT: [[TMP51:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_MAPPERS6]], i32 0, i32 2
7066 // CHECK15-NEXT: store ptr null, ptr [[TMP51]], align 4
7067 // CHECK15-NEXT: [[TMP52:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_BASEPTRS4]], i32 0, i32 0
7068 // CHECK15-NEXT: [[TMP53:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_PTRS5]], i32 0, i32 0
7069 // CHECK15-NEXT: [[TMP54:%.*]] = getelementptr inbounds [3 x i64], ptr [[DOTOFFLOAD_SIZES7]], i32 0, i32 0
7070 // CHECK15-NEXT: [[TMP55:%.*]] = load i32, ptr [[N]], align 4
7071 // CHECK15-NEXT: store i32 [[TMP55]], ptr [[DOTCAPTURE_EXPR_9]], align 4
7072 // CHECK15-NEXT: [[TMP56:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_9]], align 4
7073 // CHECK15-NEXT: [[SUB11:%.*]] = sub nsw i32 [[TMP56]], 0
7074 // CHECK15-NEXT: [[DIV12:%.*]] = sdiv i32 [[SUB11]], 1
7075 // CHECK15-NEXT: [[SUB13:%.*]] = sub nsw i32 [[DIV12]], 1
7076 // CHECK15-NEXT: store i32 [[SUB13]], ptr [[DOTCAPTURE_EXPR_10]], align 4
7077 // CHECK15-NEXT: [[TMP57:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_10]], align 4
7078 // CHECK15-NEXT: [[ADD14:%.*]] = add nsw i32 [[TMP57]], 1
7079 // CHECK15-NEXT: [[TMP58:%.*]] = zext i32 [[ADD14]] to i64
7080 // CHECK15-NEXT: [[TMP59:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS15]], i32 0, i32 0
7081 // CHECK15-NEXT: store i32 3, ptr [[TMP59]], align 4
7082 // CHECK15-NEXT: [[TMP60:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS15]], i32 0, i32 1
7083 // CHECK15-NEXT: store i32 3, ptr [[TMP60]], align 4
7084 // CHECK15-NEXT: [[TMP61:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS15]], i32 0, i32 2
7085 // CHECK15-NEXT: store ptr [[TMP52]], ptr [[TMP61]], align 4
7086 // CHECK15-NEXT: [[TMP62:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS15]], i32 0, i32 3
7087 // CHECK15-NEXT: store ptr [[TMP53]], ptr [[TMP62]], align 4
7088 // CHECK15-NEXT: [[TMP63:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS15]], i32 0, i32 4
7089 // CHECK15-NEXT: store ptr [[TMP54]], ptr [[TMP63]], align 4
7090 // CHECK15-NEXT: [[TMP64:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS15]], i32 0, i32 5
7091 // CHECK15-NEXT: store ptr @.offload_maptypes.2, ptr [[TMP64]], align 4
7092 // CHECK15-NEXT: [[TMP65:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS15]], i32 0, i32 6
7093 // CHECK15-NEXT: store ptr null, ptr [[TMP65]], align 4
7094 // CHECK15-NEXT: [[TMP66:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS15]], i32 0, i32 7
7095 // CHECK15-NEXT: store ptr null, ptr [[TMP66]], align 4
7096 // CHECK15-NEXT: [[TMP67:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS15]], i32 0, i32 8
7097 // CHECK15-NEXT: store i64 [[TMP58]], ptr [[TMP67]], align 8
7098 // CHECK15-NEXT: [[TMP68:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS15]], i32 0, i32 9
7099 // CHECK15-NEXT: store i64 0, ptr [[TMP68]], align 8
7100 // CHECK15-NEXT: [[TMP69:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS15]], i32 0, i32 10
7101 // CHECK15-NEXT: store [3 x i32] zeroinitializer, ptr [[TMP69]], align 4
7102 // CHECK15-NEXT: [[TMP70:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS15]], i32 0, i32 11
7103 // CHECK15-NEXT: store [3 x i32] zeroinitializer, ptr [[TMP70]], align 4
7104 // CHECK15-NEXT: [[TMP71:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS15]], i32 0, i32 12
7105 // CHECK15-NEXT: store i32 0, ptr [[TMP71]], align 4
7106 // CHECK15-NEXT: [[TMP72:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB3]], i64 -1, i32 0, i32 0, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l153.region_id, ptr [[KERNEL_ARGS15]])
7107 // CHECK15-NEXT: [[TMP73:%.*]] = icmp ne i32 [[TMP72]], 0
7108 // CHECK15-NEXT: br i1 [[TMP73]], label [[OMP_OFFLOAD_FAILED16:%.*]], label [[OMP_OFFLOAD_CONT17:%.*]]
7109 // CHECK15: omp_offload.failed16:
7110 // CHECK15-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l153(i32 [[TMP39]], i32 [[TMP0]], ptr [[VLA]]) #[[ATTR3]]
7111 // CHECK15-NEXT: br label [[OMP_OFFLOAD_CONT17]]
7112 // CHECK15: omp_offload.cont17:
7113 // CHECK15-NEXT: [[TMP74:%.*]] = load i32, ptr [[M]], align 4
7114 // CHECK15-NEXT: store i32 [[TMP74]], ptr [[M_CASTED]], align 4
7115 // CHECK15-NEXT: [[TMP75:%.*]] = load i32, ptr [[M_CASTED]], align 4
7116 // CHECK15-NEXT: [[TMP76:%.*]] = load i32, ptr [[N]], align 4
7117 // CHECK15-NEXT: store i32 [[TMP76]], ptr [[N_CASTED18]], align 4
7118 // CHECK15-NEXT: [[TMP77:%.*]] = load i32, ptr [[N_CASTED18]], align 4
7119 // CHECK15-NEXT: [[TMP78:%.*]] = mul nuw i32 [[TMP0]], 4
7120 // CHECK15-NEXT: [[TMP79:%.*]] = sext i32 [[TMP78]] to i64
7121 // CHECK15-NEXT: call void @llvm.memcpy.p0.p0.i32(ptr align 4 [[DOTOFFLOAD_SIZES22]], ptr align 4 @.offload_sizes.3, i32 32, i1 false)
7122 // CHECK15-NEXT: [[TMP80:%.*]] = getelementptr inbounds [4 x ptr], ptr [[DOTOFFLOAD_BASEPTRS19]], i32 0, i32 0
7123 // CHECK15-NEXT: store i32 [[TMP75]], ptr [[TMP80]], align 4
7124 // CHECK15-NEXT: [[TMP81:%.*]] = getelementptr inbounds [4 x ptr], ptr [[DOTOFFLOAD_PTRS20]], i32 0, i32 0
7125 // CHECK15-NEXT: store i32 [[TMP75]], ptr [[TMP81]], align 4
7126 // CHECK15-NEXT: [[TMP82:%.*]] = getelementptr inbounds [4 x ptr], ptr [[DOTOFFLOAD_MAPPERS21]], i32 0, i32 0
7127 // CHECK15-NEXT: store ptr null, ptr [[TMP82]], align 4
7128 // CHECK15-NEXT: [[TMP83:%.*]] = getelementptr inbounds [4 x ptr], ptr [[DOTOFFLOAD_BASEPTRS19]], i32 0, i32 1
7129 // CHECK15-NEXT: store i32 [[TMP77]], ptr [[TMP83]], align 4
7130 // CHECK15-NEXT: [[TMP84:%.*]] = getelementptr inbounds [4 x ptr], ptr [[DOTOFFLOAD_PTRS20]], i32 0, i32 1
7131 // CHECK15-NEXT: store i32 [[TMP77]], ptr [[TMP84]], align 4
7132 // CHECK15-NEXT: [[TMP85:%.*]] = getelementptr inbounds [4 x ptr], ptr [[DOTOFFLOAD_MAPPERS21]], i32 0, i32 1
7133 // CHECK15-NEXT: store ptr null, ptr [[TMP85]], align 4
7134 // CHECK15-NEXT: [[TMP86:%.*]] = getelementptr inbounds [4 x ptr], ptr [[DOTOFFLOAD_BASEPTRS19]], i32 0, i32 2
7135 // CHECK15-NEXT: store i32 [[TMP0]], ptr [[TMP86]], align 4
7136 // CHECK15-NEXT: [[TMP87:%.*]] = getelementptr inbounds [4 x ptr], ptr [[DOTOFFLOAD_PTRS20]], i32 0, i32 2
7137 // CHECK15-NEXT: store i32 [[TMP0]], ptr [[TMP87]], align 4
7138 // CHECK15-NEXT: [[TMP88:%.*]] = getelementptr inbounds [4 x ptr], ptr [[DOTOFFLOAD_MAPPERS21]], i32 0, i32 2
7139 // CHECK15-NEXT: store ptr null, ptr [[TMP88]], align 4
7140 // CHECK15-NEXT: [[TMP89:%.*]] = getelementptr inbounds [4 x ptr], ptr [[DOTOFFLOAD_BASEPTRS19]], i32 0, i32 3
7141 // CHECK15-NEXT: store ptr [[VLA]], ptr [[TMP89]], align 4
7142 // CHECK15-NEXT: [[TMP90:%.*]] = getelementptr inbounds [4 x ptr], ptr [[DOTOFFLOAD_PTRS20]], i32 0, i32 3
7143 // CHECK15-NEXT: store ptr [[VLA]], ptr [[TMP90]], align 4
7144 // CHECK15-NEXT: [[TMP91:%.*]] = getelementptr inbounds [4 x i64], ptr [[DOTOFFLOAD_SIZES22]], i32 0, i32 3
7145 // CHECK15-NEXT: store i64 [[TMP79]], ptr [[TMP91]], align 4
7146 // CHECK15-NEXT: [[TMP92:%.*]] = getelementptr inbounds [4 x ptr], ptr [[DOTOFFLOAD_MAPPERS21]], i32 0, i32 3
7147 // CHECK15-NEXT: store ptr null, ptr [[TMP92]], align 4
7148 // CHECK15-NEXT: [[TMP93:%.*]] = getelementptr inbounds [4 x ptr], ptr [[DOTOFFLOAD_BASEPTRS19]], i32 0, i32 0
7149 // CHECK15-NEXT: [[TMP94:%.*]] = getelementptr inbounds [4 x ptr], ptr [[DOTOFFLOAD_PTRS20]], i32 0, i32 0
7150 // CHECK15-NEXT: [[TMP95:%.*]] = getelementptr inbounds [4 x i64], ptr [[DOTOFFLOAD_SIZES22]], i32 0, i32 0
7151 // CHECK15-NEXT: [[TMP96:%.*]] = load i32, ptr [[N]], align 4
7152 // CHECK15-NEXT: store i32 [[TMP96]], ptr [[DOTCAPTURE_EXPR_24]], align 4
7153 // CHECK15-NEXT: [[TMP97:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_24]], align 4
7154 // CHECK15-NEXT: [[SUB26:%.*]] = sub nsw i32 [[TMP97]], 0
7155 // CHECK15-NEXT: [[DIV27:%.*]] = sdiv i32 [[SUB26]], 1
7156 // CHECK15-NEXT: [[SUB28:%.*]] = sub nsw i32 [[DIV27]], 1
7157 // CHECK15-NEXT: store i32 [[SUB28]], ptr [[DOTCAPTURE_EXPR_25]], align 4
7158 // CHECK15-NEXT: [[TMP98:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_25]], align 4
7159 // CHECK15-NEXT: [[ADD29:%.*]] = add nsw i32 [[TMP98]], 1
7160 // CHECK15-NEXT: [[TMP99:%.*]] = zext i32 [[ADD29]] to i64
7161 // CHECK15-NEXT: [[TMP100:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS30]], i32 0, i32 0
7162 // CHECK15-NEXT: store i32 3, ptr [[TMP100]], align 4
7163 // CHECK15-NEXT: [[TMP101:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS30]], i32 0, i32 1
7164 // CHECK15-NEXT: store i32 4, ptr [[TMP101]], align 4
7165 // CHECK15-NEXT: [[TMP102:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS30]], i32 0, i32 2
7166 // CHECK15-NEXT: store ptr [[TMP93]], ptr [[TMP102]], align 4
7167 // CHECK15-NEXT: [[TMP103:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS30]], i32 0, i32 3
7168 // CHECK15-NEXT: store ptr [[TMP94]], ptr [[TMP103]], align 4
7169 // CHECK15-NEXT: [[TMP104:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS30]], i32 0, i32 4
7170 // CHECK15-NEXT: store ptr [[TMP95]], ptr [[TMP104]], align 4
7171 // CHECK15-NEXT: [[TMP105:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS30]], i32 0, i32 5
7172 // CHECK15-NEXT: store ptr @.offload_maptypes.4, ptr [[TMP105]], align 4
7173 // CHECK15-NEXT: [[TMP106:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS30]], i32 0, i32 6
7174 // CHECK15-NEXT: store ptr null, ptr [[TMP106]], align 4
7175 // CHECK15-NEXT: [[TMP107:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS30]], i32 0, i32 7
7176 // CHECK15-NEXT: store ptr null, ptr [[TMP107]], align 4
7177 // CHECK15-NEXT: [[TMP108:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS30]], i32 0, i32 8
7178 // CHECK15-NEXT: store i64 [[TMP99]], ptr [[TMP108]], align 8
7179 // CHECK15-NEXT: [[TMP109:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS30]], i32 0, i32 9
7180 // CHECK15-NEXT: store i64 0, ptr [[TMP109]], align 8
7181 // CHECK15-NEXT: [[TMP110:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS30]], i32 0, i32 10
7182 // CHECK15-NEXT: store [3 x i32] zeroinitializer, ptr [[TMP110]], align 4
7183 // CHECK15-NEXT: [[TMP111:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS30]], i32 0, i32 11
7184 // CHECK15-NEXT: store [3 x i32] zeroinitializer, ptr [[TMP111]], align 4
7185 // CHECK15-NEXT: [[TMP112:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS30]], i32 0, i32 12
7186 // CHECK15-NEXT: store i32 0, ptr [[TMP112]], align 4
7187 // CHECK15-NEXT: [[TMP113:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB3]], i64 -1, i32 0, i32 0, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l158.region_id, ptr [[KERNEL_ARGS30]])
7188 // CHECK15-NEXT: [[TMP114:%.*]] = icmp ne i32 [[TMP113]], 0
7189 // CHECK15-NEXT: br i1 [[TMP114]], label [[OMP_OFFLOAD_FAILED31:%.*]], label [[OMP_OFFLOAD_CONT32:%.*]]
7190 // CHECK15: omp_offload.failed31:
7191 // CHECK15-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l158(i32 [[TMP75]], i32 [[TMP77]], i32 [[TMP0]], ptr [[VLA]]) #[[ATTR3]]
7192 // CHECK15-NEXT: br label [[OMP_OFFLOAD_CONT32]]
7193 // CHECK15: omp_offload.cont32:
7194 // CHECK15-NEXT: [[TMP115:%.*]] = load i32, ptr [[N]], align 4
7195 // CHECK15-NEXT: store i32 [[TMP115]], ptr [[N_CASTED33]], align 4
7196 // CHECK15-NEXT: [[TMP116:%.*]] = load i32, ptr [[N_CASTED33]], align 4
7197 // CHECK15-NEXT: [[TMP117:%.*]] = mul nuw i32 [[TMP0]], 4
7198 // CHECK15-NEXT: [[TMP118:%.*]] = sext i32 [[TMP117]] to i64
7199 // CHECK15-NEXT: call void @llvm.memcpy.p0.p0.i32(ptr align 4 [[DOTOFFLOAD_SIZES37]], ptr align 4 @.offload_sizes.5, i32 24, i1 false)
7200 // CHECK15-NEXT: [[TMP119:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_BASEPTRS34]], i32 0, i32 0
7201 // CHECK15-NEXT: store i32 [[TMP116]], ptr [[TMP119]], align 4
7202 // CHECK15-NEXT: [[TMP120:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_PTRS35]], i32 0, i32 0
7203 // CHECK15-NEXT: store i32 [[TMP116]], ptr [[TMP120]], align 4
7204 // CHECK15-NEXT: [[TMP121:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_MAPPERS36]], i32 0, i32 0
7205 // CHECK15-NEXT: store ptr null, ptr [[TMP121]], align 4
7206 // CHECK15-NEXT: [[TMP122:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_BASEPTRS34]], i32 0, i32 1
7207 // CHECK15-NEXT: store i32 [[TMP0]], ptr [[TMP122]], align 4
7208 // CHECK15-NEXT: [[TMP123:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_PTRS35]], i32 0, i32 1
7209 // CHECK15-NEXT: store i32 [[TMP0]], ptr [[TMP123]], align 4
7210 // CHECK15-NEXT: [[TMP124:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_MAPPERS36]], i32 0, i32 1
7211 // CHECK15-NEXT: store ptr null, ptr [[TMP124]], align 4
7212 // CHECK15-NEXT: [[TMP125:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_BASEPTRS34]], i32 0, i32 2
7213 // CHECK15-NEXT: store ptr [[VLA]], ptr [[TMP125]], align 4
7214 // CHECK15-NEXT: [[TMP126:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_PTRS35]], i32 0, i32 2
7215 // CHECK15-NEXT: store ptr [[VLA]], ptr [[TMP126]], align 4
7216 // CHECK15-NEXT: [[TMP127:%.*]] = getelementptr inbounds [3 x i64], ptr [[DOTOFFLOAD_SIZES37]], i32 0, i32 2
7217 // CHECK15-NEXT: store i64 [[TMP118]], ptr [[TMP127]], align 4
7218 // CHECK15-NEXT: [[TMP128:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_MAPPERS36]], i32 0, i32 2
7219 // CHECK15-NEXT: store ptr null, ptr [[TMP128]], align 4
7220 // CHECK15-NEXT: [[TMP129:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_BASEPTRS34]], i32 0, i32 0
7221 // CHECK15-NEXT: [[TMP130:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_PTRS35]], i32 0, i32 0
7222 // CHECK15-NEXT: [[TMP131:%.*]] = getelementptr inbounds [3 x i64], ptr [[DOTOFFLOAD_SIZES37]], i32 0, i32 0
7223 // CHECK15-NEXT: [[TMP132:%.*]] = load i32, ptr [[N]], align 4
7224 // CHECK15-NEXT: store i32 [[TMP132]], ptr [[DOTCAPTURE_EXPR_39]], align 4
7225 // CHECK15-NEXT: [[TMP133:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_39]], align 4
7226 // CHECK15-NEXT: [[SUB41:%.*]] = sub nsw i32 [[TMP133]], 0
7227 // CHECK15-NEXT: [[DIV42:%.*]] = sdiv i32 [[SUB41]], 1
7228 // CHECK15-NEXT: [[SUB43:%.*]] = sub nsw i32 [[DIV42]], 1
7229 // CHECK15-NEXT: store i32 [[SUB43]], ptr [[DOTCAPTURE_EXPR_40]], align 4
7230 // CHECK15-NEXT: [[TMP134:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_40]], align 4
7231 // CHECK15-NEXT: [[ADD44:%.*]] = add nsw i32 [[TMP134]], 1
7232 // CHECK15-NEXT: [[TMP135:%.*]] = zext i32 [[ADD44]] to i64
7233 // CHECK15-NEXT: [[TMP136:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS45]], i32 0, i32 0
7234 // CHECK15-NEXT: store i32 3, ptr [[TMP136]], align 4
7235 // CHECK15-NEXT: [[TMP137:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS45]], i32 0, i32 1
7236 // CHECK15-NEXT: store i32 3, ptr [[TMP137]], align 4
7237 // CHECK15-NEXT: [[TMP138:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS45]], i32 0, i32 2
7238 // CHECK15-NEXT: store ptr [[TMP129]], ptr [[TMP138]], align 4
7239 // CHECK15-NEXT: [[TMP139:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS45]], i32 0, i32 3
7240 // CHECK15-NEXT: store ptr [[TMP130]], ptr [[TMP139]], align 4
7241 // CHECK15-NEXT: [[TMP140:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS45]], i32 0, i32 4
7242 // CHECK15-NEXT: store ptr [[TMP131]], ptr [[TMP140]], align 4
7243 // CHECK15-NEXT: [[TMP141:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS45]], i32 0, i32 5
7244 // CHECK15-NEXT: store ptr @.offload_maptypes.6, ptr [[TMP141]], align 4
7245 // CHECK15-NEXT: [[TMP142:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS45]], i32 0, i32 6
7246 // CHECK15-NEXT: store ptr null, ptr [[TMP142]], align 4
7247 // CHECK15-NEXT: [[TMP143:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS45]], i32 0, i32 7
7248 // CHECK15-NEXT: store ptr null, ptr [[TMP143]], align 4
7249 // CHECK15-NEXT: [[TMP144:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS45]], i32 0, i32 8
7250 // CHECK15-NEXT: store i64 [[TMP135]], ptr [[TMP144]], align 8
7251 // CHECK15-NEXT: [[TMP145:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS45]], i32 0, i32 9
7252 // CHECK15-NEXT: store i64 0, ptr [[TMP145]], align 8
7253 // CHECK15-NEXT: [[TMP146:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS45]], i32 0, i32 10
7254 // CHECK15-NEXT: store [3 x i32] zeroinitializer, ptr [[TMP146]], align 4
7255 // CHECK15-NEXT: [[TMP147:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS45]], i32 0, i32 11
7256 // CHECK15-NEXT: store [3 x i32] zeroinitializer, ptr [[TMP147]], align 4
7257 // CHECK15-NEXT: [[TMP148:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS45]], i32 0, i32 12
7258 // CHECK15-NEXT: store i32 0, ptr [[TMP148]], align 4
7259 // CHECK15-NEXT: [[TMP149:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB3]], i64 -1, i32 0, i32 0, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l163.region_id, ptr [[KERNEL_ARGS45]])
7260 // CHECK15-NEXT: [[TMP150:%.*]] = icmp ne i32 [[TMP149]], 0
7261 // CHECK15-NEXT: br i1 [[TMP150]], label [[OMP_OFFLOAD_FAILED46:%.*]], label [[OMP_OFFLOAD_CONT47:%.*]]
7262 // CHECK15: omp_offload.failed46:
7263 // CHECK15-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l163(i32 [[TMP116]], i32 [[TMP0]], ptr [[VLA]]) #[[ATTR3]]
7264 // CHECK15-NEXT: br label [[OMP_OFFLOAD_CONT47]]
7265 // CHECK15: omp_offload.cont47:
7266 // CHECK15-NEXT: [[TMP151:%.*]] = load i32, ptr [[M]], align 4
7267 // CHECK15-NEXT: store i32 [[TMP151]], ptr [[M_CASTED48]], align 4
7268 // CHECK15-NEXT: [[TMP152:%.*]] = load i32, ptr [[M_CASTED48]], align 4
7269 // CHECK15-NEXT: [[TMP153:%.*]] = load i32, ptr [[N]], align 4
7270 // CHECK15-NEXT: store i32 [[TMP153]], ptr [[N_CASTED49]], align 4
7271 // CHECK15-NEXT: [[TMP154:%.*]] = load i32, ptr [[N_CASTED49]], align 4
7272 // CHECK15-NEXT: [[TMP155:%.*]] = mul nuw i32 [[TMP0]], 4
7273 // CHECK15-NEXT: [[TMP156:%.*]] = sext i32 [[TMP155]] to i64
7274 // CHECK15-NEXT: call void @llvm.memcpy.p0.p0.i32(ptr align 4 [[DOTOFFLOAD_SIZES53]], ptr align 4 @.offload_sizes.7, i32 32, i1 false)
7275 // CHECK15-NEXT: [[TMP157:%.*]] = getelementptr inbounds [4 x ptr], ptr [[DOTOFFLOAD_BASEPTRS50]], i32 0, i32 0
7276 // CHECK15-NEXT: store i32 [[TMP152]], ptr [[TMP157]], align 4
7277 // CHECK15-NEXT: [[TMP158:%.*]] = getelementptr inbounds [4 x ptr], ptr [[DOTOFFLOAD_PTRS51]], i32 0, i32 0
7278 // CHECK15-NEXT: store i32 [[TMP152]], ptr [[TMP158]], align 4
7279 // CHECK15-NEXT: [[TMP159:%.*]] = getelementptr inbounds [4 x ptr], ptr [[DOTOFFLOAD_MAPPERS52]], i32 0, i32 0
7280 // CHECK15-NEXT: store ptr null, ptr [[TMP159]], align 4
7281 // CHECK15-NEXT: [[TMP160:%.*]] = getelementptr inbounds [4 x ptr], ptr [[DOTOFFLOAD_BASEPTRS50]], i32 0, i32 1
7282 // CHECK15-NEXT: store i32 [[TMP154]], ptr [[TMP160]], align 4
7283 // CHECK15-NEXT: [[TMP161:%.*]] = getelementptr inbounds [4 x ptr], ptr [[DOTOFFLOAD_PTRS51]], i32 0, i32 1
7284 // CHECK15-NEXT: store i32 [[TMP154]], ptr [[TMP161]], align 4
7285 // CHECK15-NEXT: [[TMP162:%.*]] = getelementptr inbounds [4 x ptr], ptr [[DOTOFFLOAD_MAPPERS52]], i32 0, i32 1
7286 // CHECK15-NEXT: store ptr null, ptr [[TMP162]], align 4
7287 // CHECK15-NEXT: [[TMP163:%.*]] = getelementptr inbounds [4 x ptr], ptr [[DOTOFFLOAD_BASEPTRS50]], i32 0, i32 2
7288 // CHECK15-NEXT: store i32 [[TMP0]], ptr [[TMP163]], align 4
7289 // CHECK15-NEXT: [[TMP164:%.*]] = getelementptr inbounds [4 x ptr], ptr [[DOTOFFLOAD_PTRS51]], i32 0, i32 2
7290 // CHECK15-NEXT: store i32 [[TMP0]], ptr [[TMP164]], align 4
7291 // CHECK15-NEXT: [[TMP165:%.*]] = getelementptr inbounds [4 x ptr], ptr [[DOTOFFLOAD_MAPPERS52]], i32 0, i32 2
7292 // CHECK15-NEXT: store ptr null, ptr [[TMP165]], align 4
7293 // CHECK15-NEXT: [[TMP166:%.*]] = getelementptr inbounds [4 x ptr], ptr [[DOTOFFLOAD_BASEPTRS50]], i32 0, i32 3
7294 // CHECK15-NEXT: store ptr [[VLA]], ptr [[TMP166]], align 4
7295 // CHECK15-NEXT: [[TMP167:%.*]] = getelementptr inbounds [4 x ptr], ptr [[DOTOFFLOAD_PTRS51]], i32 0, i32 3
7296 // CHECK15-NEXT: store ptr [[VLA]], ptr [[TMP167]], align 4
7297 // CHECK15-NEXT: [[TMP168:%.*]] = getelementptr inbounds [4 x i64], ptr [[DOTOFFLOAD_SIZES53]], i32 0, i32 3
7298 // CHECK15-NEXT: store i64 [[TMP156]], ptr [[TMP168]], align 4
7299 // CHECK15-NEXT: [[TMP169:%.*]] = getelementptr inbounds [4 x ptr], ptr [[DOTOFFLOAD_MAPPERS52]], i32 0, i32 3
7300 // CHECK15-NEXT: store ptr null, ptr [[TMP169]], align 4
7301 // CHECK15-NEXT: [[TMP170:%.*]] = getelementptr inbounds [4 x ptr], ptr [[DOTOFFLOAD_BASEPTRS50]], i32 0, i32 0
7302 // CHECK15-NEXT: [[TMP171:%.*]] = getelementptr inbounds [4 x ptr], ptr [[DOTOFFLOAD_PTRS51]], i32 0, i32 0
7303 // CHECK15-NEXT: [[TMP172:%.*]] = getelementptr inbounds [4 x i64], ptr [[DOTOFFLOAD_SIZES53]], i32 0, i32 0
7304 // CHECK15-NEXT: [[TMP173:%.*]] = load i32, ptr [[N]], align 4
7305 // CHECK15-NEXT: store i32 [[TMP173]], ptr [[DOTCAPTURE_EXPR_55]], align 4
7306 // CHECK15-NEXT: [[TMP174:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_55]], align 4
7307 // CHECK15-NEXT: [[SUB57:%.*]] = sub nsw i32 [[TMP174]], 0
7308 // CHECK15-NEXT: [[DIV58:%.*]] = sdiv i32 [[SUB57]], 1
7309 // CHECK15-NEXT: [[SUB59:%.*]] = sub nsw i32 [[DIV58]], 1
7310 // CHECK15-NEXT: store i32 [[SUB59]], ptr [[DOTCAPTURE_EXPR_56]], align 4
7311 // CHECK15-NEXT: [[TMP175:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_56]], align 4
7312 // CHECK15-NEXT: [[ADD60:%.*]] = add nsw i32 [[TMP175]], 1
7313 // CHECK15-NEXT: [[TMP176:%.*]] = zext i32 [[ADD60]] to i64
7314 // CHECK15-NEXT: [[TMP177:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS61]], i32 0, i32 0
7315 // CHECK15-NEXT: store i32 3, ptr [[TMP177]], align 4
7316 // CHECK15-NEXT: [[TMP178:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS61]], i32 0, i32 1
7317 // CHECK15-NEXT: store i32 4, ptr [[TMP178]], align 4
7318 // CHECK15-NEXT: [[TMP179:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS61]], i32 0, i32 2
7319 // CHECK15-NEXT: store ptr [[TMP170]], ptr [[TMP179]], align 4
7320 // CHECK15-NEXT: [[TMP180:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS61]], i32 0, i32 3
7321 // CHECK15-NEXT: store ptr [[TMP171]], ptr [[TMP180]], align 4
7322 // CHECK15-NEXT: [[TMP181:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS61]], i32 0, i32 4
7323 // CHECK15-NEXT: store ptr [[TMP172]], ptr [[TMP181]], align 4
7324 // CHECK15-NEXT: [[TMP182:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS61]], i32 0, i32 5
7325 // CHECK15-NEXT: store ptr @.offload_maptypes.8, ptr [[TMP182]], align 4
7326 // CHECK15-NEXT: [[TMP183:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS61]], i32 0, i32 6
7327 // CHECK15-NEXT: store ptr null, ptr [[TMP183]], align 4
7328 // CHECK15-NEXT: [[TMP184:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS61]], i32 0, i32 7
7329 // CHECK15-NEXT: store ptr null, ptr [[TMP184]], align 4
7330 // CHECK15-NEXT: [[TMP185:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS61]], i32 0, i32 8
7331 // CHECK15-NEXT: store i64 [[TMP176]], ptr [[TMP185]], align 8
7332 // CHECK15-NEXT: [[TMP186:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS61]], i32 0, i32 9
7333 // CHECK15-NEXT: store i64 0, ptr [[TMP186]], align 8
7334 // CHECK15-NEXT: [[TMP187:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS61]], i32 0, i32 10
7335 // CHECK15-NEXT: store [3 x i32] zeroinitializer, ptr [[TMP187]], align 4
7336 // CHECK15-NEXT: [[TMP188:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS61]], i32 0, i32 11
7337 // CHECK15-NEXT: store [3 x i32] zeroinitializer, ptr [[TMP188]], align 4
7338 // CHECK15-NEXT: [[TMP189:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS61]], i32 0, i32 12
7339 // CHECK15-NEXT: store i32 0, ptr [[TMP189]], align 4
7340 // CHECK15-NEXT: [[TMP190:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB3]], i64 -1, i32 0, i32 0, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l168.region_id, ptr [[KERNEL_ARGS61]])
7341 // CHECK15-NEXT: [[TMP191:%.*]] = icmp ne i32 [[TMP190]], 0
7342 // CHECK15-NEXT: br i1 [[TMP191]], label [[OMP_OFFLOAD_FAILED62:%.*]], label [[OMP_OFFLOAD_CONT63:%.*]]
7343 // CHECK15: omp_offload.failed62:
7344 // CHECK15-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l168(i32 [[TMP152]], i32 [[TMP154]], i32 [[TMP0]], ptr [[VLA]]) #[[ATTR3]]
7345 // CHECK15-NEXT: br label [[OMP_OFFLOAD_CONT63]]
7346 // CHECK15: omp_offload.cont63:
7347 // CHECK15-NEXT: [[TMP192:%.*]] = load i32, ptr [[ARGC_ADDR]], align 4
7348 // CHECK15-NEXT: [[CALL:%.*]] = call noundef i32 @_Z5tmainIiLi10EEiT_(i32 noundef [[TMP192]])
7349 // CHECK15-NEXT: store i32 [[CALL]], ptr [[RETVAL]], align 4
7350 // CHECK15-NEXT: [[TMP193:%.*]] = load ptr, ptr [[SAVED_STACK]], align 4
7351 // CHECK15-NEXT: call void @llvm.stackrestore.p0(ptr [[TMP193]])
7352 // CHECK15-NEXT: [[TMP194:%.*]] = load i32, ptr [[RETVAL]], align 4
7353 // CHECK15-NEXT: ret i32 [[TMP194]]
7356 // CHECK15-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l148
7357 // CHECK15-SAME: (i32 noundef [[N:%.*]], i32 noundef [[VLA:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[A:%.*]]) #[[ATTR2:[0-9]+]] {
7358 // CHECK15-NEXT: entry:
7359 // CHECK15-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4
7360 // CHECK15-NEXT: [[VLA_ADDR:%.*]] = alloca i32, align 4
7361 // CHECK15-NEXT: [[A_ADDR:%.*]] = alloca ptr, align 4
7362 // CHECK15-NEXT: store i32 [[N]], ptr [[N_ADDR]], align 4
7363 // CHECK15-NEXT: store i32 [[VLA]], ptr [[VLA_ADDR]], align 4
7364 // CHECK15-NEXT: store ptr [[A]], ptr [[A_ADDR]], align 4
7365 // CHECK15-NEXT: [[TMP0:%.*]] = load i32, ptr [[VLA_ADDR]], align 4
7366 // CHECK15-NEXT: [[TMP1:%.*]] = load ptr, ptr [[A_ADDR]], align 4
7367 // CHECK15-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_teams(ptr @[[GLOB3]], i32 3, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l148.omp_outlined, ptr [[N_ADDR]], i32 [[TMP0]], ptr [[TMP1]])
7368 // CHECK15-NEXT: ret void
7371 // CHECK15-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l148.omp_outlined
7372 // CHECK15-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[N:%.*]], i32 noundef [[VLA:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[A:%.*]]) #[[ATTR2]] {
7373 // CHECK15-NEXT: entry:
7374 // CHECK15-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 4
7375 // CHECK15-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 4
7376 // CHECK15-NEXT: [[N_ADDR:%.*]] = alloca ptr, align 4
7377 // CHECK15-NEXT: [[VLA_ADDR:%.*]] = alloca i32, align 4
7378 // CHECK15-NEXT: [[A_ADDR:%.*]] = alloca ptr, align 4
7379 // CHECK15-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
7380 // CHECK15-NEXT: [[TMP:%.*]] = alloca i32, align 4
7381 // CHECK15-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4
7382 // CHECK15-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4
7383 // CHECK15-NEXT: [[I:%.*]] = alloca i32, align 4
7384 // CHECK15-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4
7385 // CHECK15-NEXT: [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4
7386 // CHECK15-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
7387 // CHECK15-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
7388 // CHECK15-NEXT: [[I3:%.*]] = alloca i32, align 4
7389 // CHECK15-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 4
7390 // CHECK15-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 4
7391 // CHECK15-NEXT: store ptr [[N]], ptr [[N_ADDR]], align 4
7392 // CHECK15-NEXT: store i32 [[VLA]], ptr [[VLA_ADDR]], align 4
7393 // CHECK15-NEXT: store ptr [[A]], ptr [[A_ADDR]], align 4
7394 // CHECK15-NEXT: [[TMP0:%.*]] = load ptr, ptr [[N_ADDR]], align 4
7395 // CHECK15-NEXT: [[TMP1:%.*]] = load i32, ptr [[VLA_ADDR]], align 4
7396 // CHECK15-NEXT: [[TMP2:%.*]] = load ptr, ptr [[A_ADDR]], align 4
7397 // CHECK15-NEXT: [[TMP3:%.*]] = load i32, ptr [[TMP0]], align 4
7398 // CHECK15-NEXT: store i32 [[TMP3]], ptr [[DOTCAPTURE_EXPR_]], align 4
7399 // CHECK15-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4
7400 // CHECK15-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP4]], 0
7401 // CHECK15-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1
7402 // CHECK15-NEXT: [[SUB2:%.*]] = sub nsw i32 [[DIV]], 1
7403 // CHECK15-NEXT: store i32 [[SUB2]], ptr [[DOTCAPTURE_EXPR_1]], align 4
7404 // CHECK15-NEXT: store i32 0, ptr [[I]], align 4
7405 // CHECK15-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4
7406 // CHECK15-NEXT: [[CMP:%.*]] = icmp slt i32 0, [[TMP5]]
7407 // CHECK15-NEXT: br i1 [[CMP]], label [[OMP_PRECOND_THEN:%.*]], label [[OMP_PRECOND_END:%.*]]
7408 // CHECK15: omp.precond.then:
7409 // CHECK15-NEXT: store i32 0, ptr [[DOTOMP_COMB_LB]], align 4
7410 // CHECK15-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4
7411 // CHECK15-NEXT: store i32 [[TMP6]], ptr [[DOTOMP_COMB_UB]], align 4
7412 // CHECK15-NEXT: store i32 1, ptr [[DOTOMP_STRIDE]], align 4
7413 // CHECK15-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
7414 // CHECK15-NEXT: [[TMP7:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
7415 // CHECK15-NEXT: [[TMP8:%.*]] = load i32, ptr [[TMP7]], align 4
7416 // CHECK15-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB1:[0-9]+]], i32 [[TMP8]], i32 92, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
7417 // CHECK15-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
7418 // CHECK15-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4
7419 // CHECK15-NEXT: [[CMP4:%.*]] = icmp sgt i32 [[TMP9]], [[TMP10]]
7420 // CHECK15-NEXT: br i1 [[CMP4]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
7421 // CHECK15: cond.true:
7422 // CHECK15-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4
7423 // CHECK15-NEXT: br label [[COND_END:%.*]]
7424 // CHECK15: cond.false:
7425 // CHECK15-NEXT: [[TMP12:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
7426 // CHECK15-NEXT: br label [[COND_END]]
7427 // CHECK15: cond.end:
7428 // CHECK15-NEXT: [[COND:%.*]] = phi i32 [ [[TMP11]], [[COND_TRUE]] ], [ [[TMP12]], [[COND_FALSE]] ]
7429 // CHECK15-NEXT: store i32 [[COND]], ptr [[DOTOMP_COMB_UB]], align 4
7430 // CHECK15-NEXT: [[TMP13:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
7431 // CHECK15-NEXT: store i32 [[TMP13]], ptr [[DOTOMP_IV]], align 4
7432 // CHECK15-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
7433 // CHECK15: omp.inner.for.cond:
7434 // CHECK15-NEXT: [[TMP14:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
7435 // CHECK15-NEXT: [[TMP15:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
7436 // CHECK15-NEXT: [[CMP5:%.*]] = icmp sle i32 [[TMP14]], [[TMP15]]
7437 // CHECK15-NEXT: br i1 [[CMP5]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
7438 // CHECK15: omp.inner.for.body:
7439 // CHECK15-NEXT: [[TMP16:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
7440 // CHECK15-NEXT: [[TMP17:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
7441 // CHECK15-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_call(ptr @[[GLOB3]], i32 5, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l148.omp_outlined.omp_outlined, i32 [[TMP16]], i32 [[TMP17]], ptr [[TMP0]], i32 [[TMP1]], ptr [[TMP2]])
7442 // CHECK15-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
7443 // CHECK15: omp.inner.for.inc:
7444 // CHECK15-NEXT: [[TMP18:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
7445 // CHECK15-NEXT: [[TMP19:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4
7446 // CHECK15-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP18]], [[TMP19]]
7447 // CHECK15-NEXT: store i32 [[ADD]], ptr [[DOTOMP_IV]], align 4
7448 // CHECK15-NEXT: br label [[OMP_INNER_FOR_COND]]
7449 // CHECK15: omp.inner.for.end:
7450 // CHECK15-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
7451 // CHECK15: omp.loop.exit:
7452 // CHECK15-NEXT: [[TMP20:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
7453 // CHECK15-NEXT: [[TMP21:%.*]] = load i32, ptr [[TMP20]], align 4
7454 // CHECK15-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB1]], i32 [[TMP21]])
7455 // CHECK15-NEXT: br label [[OMP_PRECOND_END]]
7456 // CHECK15: omp.precond.end:
7457 // CHECK15-NEXT: ret void
7460 // CHECK15-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l148.omp_outlined.omp_outlined
7461 // CHECK15-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], i32 noundef [[DOTPREVIOUS_LB_:%.*]], i32 noundef [[DOTPREVIOUS_UB_:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[N:%.*]], i32 noundef [[VLA:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[A:%.*]]) #[[ATTR2]] {
7462 // CHECK15-NEXT: entry:
7463 // CHECK15-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 4
7464 // CHECK15-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 4
7465 // CHECK15-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i32, align 4
7466 // CHECK15-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i32, align 4
7467 // CHECK15-NEXT: [[N_ADDR:%.*]] = alloca ptr, align 4
7468 // CHECK15-NEXT: [[VLA_ADDR:%.*]] = alloca i32, align 4
7469 // CHECK15-NEXT: [[A_ADDR:%.*]] = alloca ptr, align 4
7470 // CHECK15-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
7471 // CHECK15-NEXT: [[TMP:%.*]] = alloca i32, align 4
7472 // CHECK15-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4
7473 // CHECK15-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4
7474 // CHECK15-NEXT: [[I:%.*]] = alloca i32, align 4
7475 // CHECK15-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4
7476 // CHECK15-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4
7477 // CHECK15-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
7478 // CHECK15-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
7479 // CHECK15-NEXT: [[I3:%.*]] = alloca i32, align 4
7480 // CHECK15-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 4
7481 // CHECK15-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 4
7482 // CHECK15-NEXT: store i32 [[DOTPREVIOUS_LB_]], ptr [[DOTPREVIOUS_LB__ADDR]], align 4
7483 // CHECK15-NEXT: store i32 [[DOTPREVIOUS_UB_]], ptr [[DOTPREVIOUS_UB__ADDR]], align 4
7484 // CHECK15-NEXT: store ptr [[N]], ptr [[N_ADDR]], align 4
7485 // CHECK15-NEXT: store i32 [[VLA]], ptr [[VLA_ADDR]], align 4
7486 // CHECK15-NEXT: store ptr [[A]], ptr [[A_ADDR]], align 4
7487 // CHECK15-NEXT: [[TMP0:%.*]] = load ptr, ptr [[N_ADDR]], align 4
7488 // CHECK15-NEXT: [[TMP1:%.*]] = load i32, ptr [[VLA_ADDR]], align 4
7489 // CHECK15-NEXT: [[TMP2:%.*]] = load ptr, ptr [[A_ADDR]], align 4
7490 // CHECK15-NEXT: [[TMP3:%.*]] = load i32, ptr [[TMP0]], align 4
7491 // CHECK15-NEXT: store i32 [[TMP3]], ptr [[DOTCAPTURE_EXPR_]], align 4
7492 // CHECK15-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4
7493 // CHECK15-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP4]], 0
7494 // CHECK15-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1
7495 // CHECK15-NEXT: [[SUB2:%.*]] = sub nsw i32 [[DIV]], 1
7496 // CHECK15-NEXT: store i32 [[SUB2]], ptr [[DOTCAPTURE_EXPR_1]], align 4
7497 // CHECK15-NEXT: store i32 0, ptr [[I]], align 4
7498 // CHECK15-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4
7499 // CHECK15-NEXT: [[CMP:%.*]] = icmp slt i32 0, [[TMP5]]
7500 // CHECK15-NEXT: br i1 [[CMP]], label [[OMP_PRECOND_THEN:%.*]], label [[OMP_PRECOND_END:%.*]]
7501 // CHECK15: omp.precond.then:
7502 // CHECK15-NEXT: store i32 0, ptr [[DOTOMP_LB]], align 4
7503 // CHECK15-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4
7504 // CHECK15-NEXT: store i32 [[TMP6]], ptr [[DOTOMP_UB]], align 4
7505 // CHECK15-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTPREVIOUS_LB__ADDR]], align 4
7506 // CHECK15-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTPREVIOUS_UB__ADDR]], align 4
7507 // CHECK15-NEXT: store i32 [[TMP7]], ptr [[DOTOMP_LB]], align 4
7508 // CHECK15-NEXT: store i32 [[TMP8]], ptr [[DOTOMP_UB]], align 4
7509 // CHECK15-NEXT: store i32 1, ptr [[DOTOMP_STRIDE]], align 4
7510 // CHECK15-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
7511 // CHECK15-NEXT: [[TMP9:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
7512 // CHECK15-NEXT: [[TMP10:%.*]] = load i32, ptr [[TMP9]], align 4
7513 // CHECK15-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB2:[0-9]+]], i32 [[TMP10]], i32 34, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
7514 // CHECK15-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
7515 // CHECK15-NEXT: [[TMP12:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4
7516 // CHECK15-NEXT: [[CMP4:%.*]] = icmp sgt i32 [[TMP11]], [[TMP12]]
7517 // CHECK15-NEXT: br i1 [[CMP4]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
7518 // CHECK15: cond.true:
7519 // CHECK15-NEXT: [[TMP13:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4
7520 // CHECK15-NEXT: br label [[COND_END:%.*]]
7521 // CHECK15: cond.false:
7522 // CHECK15-NEXT: [[TMP14:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
7523 // CHECK15-NEXT: br label [[COND_END]]
7524 // CHECK15: cond.end:
7525 // CHECK15-NEXT: [[COND:%.*]] = phi i32 [ [[TMP13]], [[COND_TRUE]] ], [ [[TMP14]], [[COND_FALSE]] ]
7526 // CHECK15-NEXT: store i32 [[COND]], ptr [[DOTOMP_UB]], align 4
7527 // CHECK15-NEXT: [[TMP15:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4
7528 // CHECK15-NEXT: store i32 [[TMP15]], ptr [[DOTOMP_IV]], align 4
7529 // CHECK15-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
7530 // CHECK15: omp.inner.for.cond:
7531 // CHECK15-NEXT: [[TMP16:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
7532 // CHECK15-NEXT: [[TMP17:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
7533 // CHECK15-NEXT: [[CMP5:%.*]] = icmp sle i32 [[TMP16]], [[TMP17]]
7534 // CHECK15-NEXT: br i1 [[CMP5]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
7535 // CHECK15: omp.inner.for.body:
7536 // CHECK15-NEXT: [[TMP18:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
7537 // CHECK15-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP18]], 1
7538 // CHECK15-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
7539 // CHECK15-NEXT: store i32 [[ADD]], ptr [[I3]], align 4
7540 // CHECK15-NEXT: [[TMP19:%.*]] = load i32, ptr [[I3]], align 4
7541 // CHECK15-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i32, ptr [[TMP2]], i32 [[TMP19]]
7542 // CHECK15-NEXT: store i32 0, ptr [[ARRAYIDX]], align 4
7543 // CHECK15-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
7544 // CHECK15: omp.body.continue:
7545 // CHECK15-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
7546 // CHECK15: omp.inner.for.inc:
7547 // CHECK15-NEXT: [[TMP20:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
7548 // CHECK15-NEXT: [[ADD6:%.*]] = add nsw i32 [[TMP20]], 1
7549 // CHECK15-NEXT: store i32 [[ADD6]], ptr [[DOTOMP_IV]], align 4
7550 // CHECK15-NEXT: br label [[OMP_INNER_FOR_COND]]
7551 // CHECK15: omp.inner.for.end:
7552 // CHECK15-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
7553 // CHECK15: omp.loop.exit:
7554 // CHECK15-NEXT: [[TMP21:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
7555 // CHECK15-NEXT: [[TMP22:%.*]] = load i32, ptr [[TMP21]], align 4
7556 // CHECK15-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB2]], i32 [[TMP22]])
7557 // CHECK15-NEXT: br label [[OMP_PRECOND_END]]
7558 // CHECK15: omp.precond.end:
7559 // CHECK15-NEXT: ret void
7562 // CHECK15-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l153
7563 // CHECK15-SAME: (i32 noundef [[N:%.*]], i32 noundef [[VLA:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[A:%.*]]) #[[ATTR2]] {
7564 // CHECK15-NEXT: entry:
7565 // CHECK15-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4
7566 // CHECK15-NEXT: [[VLA_ADDR:%.*]] = alloca i32, align 4
7567 // CHECK15-NEXT: [[A_ADDR:%.*]] = alloca ptr, align 4
7568 // CHECK15-NEXT: store i32 [[N]], ptr [[N_ADDR]], align 4
7569 // CHECK15-NEXT: store i32 [[VLA]], ptr [[VLA_ADDR]], align 4
7570 // CHECK15-NEXT: store ptr [[A]], ptr [[A_ADDR]], align 4
7571 // CHECK15-NEXT: [[TMP0:%.*]] = load i32, ptr [[VLA_ADDR]], align 4
7572 // CHECK15-NEXT: [[TMP1:%.*]] = load ptr, ptr [[A_ADDR]], align 4
7573 // CHECK15-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_teams(ptr @[[GLOB3]], i32 3, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l153.omp_outlined, ptr [[N_ADDR]], i32 [[TMP0]], ptr [[TMP1]])
7574 // CHECK15-NEXT: ret void
7577 // CHECK15-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l153.omp_outlined
7578 // CHECK15-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[N:%.*]], i32 noundef [[VLA:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[A:%.*]]) #[[ATTR2]] {
7579 // CHECK15-NEXT: entry:
7580 // CHECK15-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 4
7581 // CHECK15-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 4
7582 // CHECK15-NEXT: [[N_ADDR:%.*]] = alloca ptr, align 4
7583 // CHECK15-NEXT: [[VLA_ADDR:%.*]] = alloca i32, align 4
7584 // CHECK15-NEXT: [[A_ADDR:%.*]] = alloca ptr, align 4
7585 // CHECK15-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
7586 // CHECK15-NEXT: [[TMP:%.*]] = alloca i32, align 4
7587 // CHECK15-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4
7588 // CHECK15-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4
7589 // CHECK15-NEXT: [[I:%.*]] = alloca i32, align 4
7590 // CHECK15-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4
7591 // CHECK15-NEXT: [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4
7592 // CHECK15-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
7593 // CHECK15-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
7594 // CHECK15-NEXT: [[I3:%.*]] = alloca i32, align 4
7595 // CHECK15-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 4
7596 // CHECK15-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 4
7597 // CHECK15-NEXT: store ptr [[N]], ptr [[N_ADDR]], align 4
7598 // CHECK15-NEXT: store i32 [[VLA]], ptr [[VLA_ADDR]], align 4
7599 // CHECK15-NEXT: store ptr [[A]], ptr [[A_ADDR]], align 4
7600 // CHECK15-NEXT: [[TMP0:%.*]] = load ptr, ptr [[N_ADDR]], align 4
7601 // CHECK15-NEXT: [[TMP1:%.*]] = load i32, ptr [[VLA_ADDR]], align 4
7602 // CHECK15-NEXT: [[TMP2:%.*]] = load ptr, ptr [[A_ADDR]], align 4
7603 // CHECK15-NEXT: [[TMP3:%.*]] = load i32, ptr [[TMP0]], align 4
7604 // CHECK15-NEXT: store i32 [[TMP3]], ptr [[DOTCAPTURE_EXPR_]], align 4
7605 // CHECK15-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4
7606 // CHECK15-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP4]], 0
7607 // CHECK15-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1
7608 // CHECK15-NEXT: [[SUB2:%.*]] = sub nsw i32 [[DIV]], 1
7609 // CHECK15-NEXT: store i32 [[SUB2]], ptr [[DOTCAPTURE_EXPR_1]], align 4
7610 // CHECK15-NEXT: store i32 0, ptr [[I]], align 4
7611 // CHECK15-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4
7612 // CHECK15-NEXT: [[CMP:%.*]] = icmp slt i32 0, [[TMP5]]
7613 // CHECK15-NEXT: br i1 [[CMP]], label [[OMP_PRECOND_THEN:%.*]], label [[OMP_PRECOND_END:%.*]]
7614 // CHECK15: omp.precond.then:
7615 // CHECK15-NEXT: store i32 0, ptr [[DOTOMP_COMB_LB]], align 4
7616 // CHECK15-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4
7617 // CHECK15-NEXT: store i32 [[TMP6]], ptr [[DOTOMP_COMB_UB]], align 4
7618 // CHECK15-NEXT: store i32 1, ptr [[DOTOMP_STRIDE]], align 4
7619 // CHECK15-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
7620 // CHECK15-NEXT: [[TMP7:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
7621 // CHECK15-NEXT: [[TMP8:%.*]] = load i32, ptr [[TMP7]], align 4
7622 // CHECK15-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB1]], i32 [[TMP8]], i32 92, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
7623 // CHECK15-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
7624 // CHECK15-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4
7625 // CHECK15-NEXT: [[CMP4:%.*]] = icmp sgt i32 [[TMP9]], [[TMP10]]
7626 // CHECK15-NEXT: br i1 [[CMP4]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
7627 // CHECK15: cond.true:
7628 // CHECK15-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4
7629 // CHECK15-NEXT: br label [[COND_END:%.*]]
7630 // CHECK15: cond.false:
7631 // CHECK15-NEXT: [[TMP12:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
7632 // CHECK15-NEXT: br label [[COND_END]]
7633 // CHECK15: cond.end:
7634 // CHECK15-NEXT: [[COND:%.*]] = phi i32 [ [[TMP11]], [[COND_TRUE]] ], [ [[TMP12]], [[COND_FALSE]] ]
7635 // CHECK15-NEXT: store i32 [[COND]], ptr [[DOTOMP_COMB_UB]], align 4
7636 // CHECK15-NEXT: [[TMP13:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
7637 // CHECK15-NEXT: store i32 [[TMP13]], ptr [[DOTOMP_IV]], align 4
7638 // CHECK15-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
7639 // CHECK15: omp.inner.for.cond:
7640 // CHECK15-NEXT: [[TMP14:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
7641 // CHECK15-NEXT: [[TMP15:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
7642 // CHECK15-NEXT: [[CMP5:%.*]] = icmp sle i32 [[TMP14]], [[TMP15]]
7643 // CHECK15-NEXT: br i1 [[CMP5]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
7644 // CHECK15: omp.inner.for.body:
7645 // CHECK15-NEXT: [[TMP16:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
7646 // CHECK15-NEXT: [[TMP17:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
7647 // CHECK15-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_call(ptr @[[GLOB3]], i32 5, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l153.omp_outlined.omp_outlined, i32 [[TMP16]], i32 [[TMP17]], ptr [[TMP0]], i32 [[TMP1]], ptr [[TMP2]])
7648 // CHECK15-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
7649 // CHECK15: omp.inner.for.inc:
7650 // CHECK15-NEXT: [[TMP18:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
7651 // CHECK15-NEXT: [[TMP19:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4
7652 // CHECK15-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP18]], [[TMP19]]
7653 // CHECK15-NEXT: store i32 [[ADD]], ptr [[DOTOMP_IV]], align 4
7654 // CHECK15-NEXT: br label [[OMP_INNER_FOR_COND]]
7655 // CHECK15: omp.inner.for.end:
7656 // CHECK15-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
7657 // CHECK15: omp.loop.exit:
7658 // CHECK15-NEXT: [[TMP20:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
7659 // CHECK15-NEXT: [[TMP21:%.*]] = load i32, ptr [[TMP20]], align 4
7660 // CHECK15-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB1]], i32 [[TMP21]])
7661 // CHECK15-NEXT: br label [[OMP_PRECOND_END]]
7662 // CHECK15: omp.precond.end:
7663 // CHECK15-NEXT: ret void
7666 // CHECK15-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l153.omp_outlined.omp_outlined
7667 // CHECK15-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], i32 noundef [[DOTPREVIOUS_LB_:%.*]], i32 noundef [[DOTPREVIOUS_UB_:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[N:%.*]], i32 noundef [[VLA:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[A:%.*]]) #[[ATTR2]] {
7668 // CHECK15-NEXT: entry:
7669 // CHECK15-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 4
7670 // CHECK15-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 4
7671 // CHECK15-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i32, align 4
7672 // CHECK15-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i32, align 4
7673 // CHECK15-NEXT: [[N_ADDR:%.*]] = alloca ptr, align 4
7674 // CHECK15-NEXT: [[VLA_ADDR:%.*]] = alloca i32, align 4
7675 // CHECK15-NEXT: [[A_ADDR:%.*]] = alloca ptr, align 4
7676 // CHECK15-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
7677 // CHECK15-NEXT: [[TMP:%.*]] = alloca i32, align 4
7678 // CHECK15-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4
7679 // CHECK15-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4
7680 // CHECK15-NEXT: [[I:%.*]] = alloca i32, align 4
7681 // CHECK15-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4
7682 // CHECK15-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4
7683 // CHECK15-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
7684 // CHECK15-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
7685 // CHECK15-NEXT: [[I3:%.*]] = alloca i32, align 4
7686 // CHECK15-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 4
7687 // CHECK15-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 4
7688 // CHECK15-NEXT: store i32 [[DOTPREVIOUS_LB_]], ptr [[DOTPREVIOUS_LB__ADDR]], align 4
7689 // CHECK15-NEXT: store i32 [[DOTPREVIOUS_UB_]], ptr [[DOTPREVIOUS_UB__ADDR]], align 4
7690 // CHECK15-NEXT: store ptr [[N]], ptr [[N_ADDR]], align 4
7691 // CHECK15-NEXT: store i32 [[VLA]], ptr [[VLA_ADDR]], align 4
7692 // CHECK15-NEXT: store ptr [[A]], ptr [[A_ADDR]], align 4
7693 // CHECK15-NEXT: [[TMP0:%.*]] = load ptr, ptr [[N_ADDR]], align 4
7694 // CHECK15-NEXT: [[TMP1:%.*]] = load i32, ptr [[VLA_ADDR]], align 4
7695 // CHECK15-NEXT: [[TMP2:%.*]] = load ptr, ptr [[A_ADDR]], align 4
7696 // CHECK15-NEXT: [[TMP3:%.*]] = load i32, ptr [[TMP0]], align 4
7697 // CHECK15-NEXT: store i32 [[TMP3]], ptr [[DOTCAPTURE_EXPR_]], align 4
7698 // CHECK15-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4
7699 // CHECK15-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP4]], 0
7700 // CHECK15-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1
7701 // CHECK15-NEXT: [[SUB2:%.*]] = sub nsw i32 [[DIV]], 1
7702 // CHECK15-NEXT: store i32 [[SUB2]], ptr [[DOTCAPTURE_EXPR_1]], align 4
7703 // CHECK15-NEXT: store i32 0, ptr [[I]], align 4
7704 // CHECK15-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4
7705 // CHECK15-NEXT: [[CMP:%.*]] = icmp slt i32 0, [[TMP5]]
7706 // CHECK15-NEXT: br i1 [[CMP]], label [[OMP_PRECOND_THEN:%.*]], label [[OMP_PRECOND_END:%.*]]
7707 // CHECK15: omp.precond.then:
7708 // CHECK15-NEXT: store i32 0, ptr [[DOTOMP_LB]], align 4
7709 // CHECK15-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4
7710 // CHECK15-NEXT: store i32 [[TMP6]], ptr [[DOTOMP_UB]], align 4
7711 // CHECK15-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTPREVIOUS_LB__ADDR]], align 4
7712 // CHECK15-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTPREVIOUS_UB__ADDR]], align 4
7713 // CHECK15-NEXT: store i32 [[TMP7]], ptr [[DOTOMP_LB]], align 4
7714 // CHECK15-NEXT: store i32 [[TMP8]], ptr [[DOTOMP_UB]], align 4
7715 // CHECK15-NEXT: store i32 1, ptr [[DOTOMP_STRIDE]], align 4
7716 // CHECK15-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
7717 // CHECK15-NEXT: [[TMP9:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
7718 // CHECK15-NEXT: [[TMP10:%.*]] = load i32, ptr [[TMP9]], align 4
7719 // CHECK15-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB2]], i32 [[TMP10]], i32 34, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
7720 // CHECK15-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
7721 // CHECK15-NEXT: [[TMP12:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4
7722 // CHECK15-NEXT: [[CMP4:%.*]] = icmp sgt i32 [[TMP11]], [[TMP12]]
7723 // CHECK15-NEXT: br i1 [[CMP4]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
7724 // CHECK15: cond.true:
7725 // CHECK15-NEXT: [[TMP13:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4
7726 // CHECK15-NEXT: br label [[COND_END:%.*]]
7727 // CHECK15: cond.false:
7728 // CHECK15-NEXT: [[TMP14:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
7729 // CHECK15-NEXT: br label [[COND_END]]
7730 // CHECK15: cond.end:
7731 // CHECK15-NEXT: [[COND:%.*]] = phi i32 [ [[TMP13]], [[COND_TRUE]] ], [ [[TMP14]], [[COND_FALSE]] ]
7732 // CHECK15-NEXT: store i32 [[COND]], ptr [[DOTOMP_UB]], align 4
7733 // CHECK15-NEXT: [[TMP15:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4
7734 // CHECK15-NEXT: store i32 [[TMP15]], ptr [[DOTOMP_IV]], align 4
7735 // CHECK15-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
7736 // CHECK15: omp.inner.for.cond:
7737 // CHECK15-NEXT: [[TMP16:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
7738 // CHECK15-NEXT: [[TMP17:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
7739 // CHECK15-NEXT: [[CMP5:%.*]] = icmp sle i32 [[TMP16]], [[TMP17]]
7740 // CHECK15-NEXT: br i1 [[CMP5]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
7741 // CHECK15: omp.inner.for.body:
7742 // CHECK15-NEXT: [[TMP18:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
7743 // CHECK15-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP18]], 1
7744 // CHECK15-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
7745 // CHECK15-NEXT: store i32 [[ADD]], ptr [[I3]], align 4
7746 // CHECK15-NEXT: [[TMP19:%.*]] = load i32, ptr [[I3]], align 4
7747 // CHECK15-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i32, ptr [[TMP2]], i32 [[TMP19]]
7748 // CHECK15-NEXT: store i32 0, ptr [[ARRAYIDX]], align 4
7749 // CHECK15-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
7750 // CHECK15: omp.body.continue:
7751 // CHECK15-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
7752 // CHECK15: omp.inner.for.inc:
7753 // CHECK15-NEXT: [[TMP20:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
7754 // CHECK15-NEXT: [[ADD6:%.*]] = add nsw i32 [[TMP20]], 1
7755 // CHECK15-NEXT: store i32 [[ADD6]], ptr [[DOTOMP_IV]], align 4
7756 // CHECK15-NEXT: br label [[OMP_INNER_FOR_COND]]
7757 // CHECK15: omp.inner.for.end:
7758 // CHECK15-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
7759 // CHECK15: omp.loop.exit:
7760 // CHECK15-NEXT: [[TMP21:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
7761 // CHECK15-NEXT: [[TMP22:%.*]] = load i32, ptr [[TMP21]], align 4
7762 // CHECK15-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB2]], i32 [[TMP22]])
7763 // CHECK15-NEXT: br label [[OMP_PRECOND_END]]
7764 // CHECK15: omp.precond.end:
7765 // CHECK15-NEXT: ret void
7768 // CHECK15-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l158
7769 // CHECK15-SAME: (i32 noundef [[M:%.*]], i32 noundef [[N:%.*]], i32 noundef [[VLA:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[A:%.*]]) #[[ATTR2]] {
7770 // CHECK15-NEXT: entry:
7771 // CHECK15-NEXT: [[M_ADDR:%.*]] = alloca i32, align 4
7772 // CHECK15-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4
7773 // CHECK15-NEXT: [[VLA_ADDR:%.*]] = alloca i32, align 4
7774 // CHECK15-NEXT: [[A_ADDR:%.*]] = alloca ptr, align 4
7775 // CHECK15-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4
7776 // CHECK15-NEXT: [[DOTCAPTURE_EXPR__CASTED:%.*]] = alloca i32, align 4
7777 // CHECK15-NEXT: store i32 [[M]], ptr [[M_ADDR]], align 4
7778 // CHECK15-NEXT: store i32 [[N]], ptr [[N_ADDR]], align 4
7779 // CHECK15-NEXT: store i32 [[VLA]], ptr [[VLA_ADDR]], align 4
7780 // CHECK15-NEXT: store ptr [[A]], ptr [[A_ADDR]], align 4
7781 // CHECK15-NEXT: [[TMP0:%.*]] = load i32, ptr [[VLA_ADDR]], align 4
7782 // CHECK15-NEXT: [[TMP1:%.*]] = load ptr, ptr [[A_ADDR]], align 4
7783 // CHECK15-NEXT: [[TMP2:%.*]] = load i32, ptr [[M_ADDR]], align 4
7784 // CHECK15-NEXT: store i32 [[TMP2]], ptr [[DOTCAPTURE_EXPR_]], align 4
7785 // CHECK15-NEXT: [[TMP3:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4
7786 // CHECK15-NEXT: store i32 [[TMP3]], ptr [[DOTCAPTURE_EXPR__CASTED]], align 4
7787 // CHECK15-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR__CASTED]], align 4
7788 // CHECK15-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_teams(ptr @[[GLOB3]], i32 4, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l158.omp_outlined, ptr [[N_ADDR]], i32 [[TMP0]], ptr [[TMP1]], i32 [[TMP4]])
7789 // CHECK15-NEXT: ret void
7792 // CHECK15-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l158.omp_outlined
7793 // CHECK15-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[N:%.*]], i32 noundef [[VLA:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[A:%.*]], i32 noundef [[DOTCAPTURE_EXPR_:%.*]]) #[[ATTR2]] {
7794 // CHECK15-NEXT: entry:
7795 // CHECK15-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 4
7796 // CHECK15-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 4
7797 // CHECK15-NEXT: [[N_ADDR:%.*]] = alloca ptr, align 4
7798 // CHECK15-NEXT: [[VLA_ADDR:%.*]] = alloca i32, align 4
7799 // CHECK15-NEXT: [[A_ADDR:%.*]] = alloca ptr, align 4
7800 // CHECK15-NEXT: [[DOTCAPTURE_EXPR__ADDR:%.*]] = alloca i32, align 4
7801 // CHECK15-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
7802 // CHECK15-NEXT: [[TMP:%.*]] = alloca i32, align 4
7803 // CHECK15-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4
7804 // CHECK15-NEXT: [[DOTCAPTURE_EXPR_2:%.*]] = alloca i32, align 4
7805 // CHECK15-NEXT: [[I:%.*]] = alloca i32, align 4
7806 // CHECK15-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4
7807 // CHECK15-NEXT: [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4
7808 // CHECK15-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
7809 // CHECK15-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
7810 // CHECK15-NEXT: [[I4:%.*]] = alloca i32, align 4
7811 // CHECK15-NEXT: [[DOTCAPTURE_EXPR__CASTED:%.*]] = alloca i32, align 4
7812 // CHECK15-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 4
7813 // CHECK15-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 4
7814 // CHECK15-NEXT: store ptr [[N]], ptr [[N_ADDR]], align 4
7815 // CHECK15-NEXT: store i32 [[VLA]], ptr [[VLA_ADDR]], align 4
7816 // CHECK15-NEXT: store ptr [[A]], ptr [[A_ADDR]], align 4
7817 // CHECK15-NEXT: store i32 [[DOTCAPTURE_EXPR_]], ptr [[DOTCAPTURE_EXPR__ADDR]], align 4
7818 // CHECK15-NEXT: [[TMP0:%.*]] = load ptr, ptr [[N_ADDR]], align 4
7819 // CHECK15-NEXT: [[TMP1:%.*]] = load i32, ptr [[VLA_ADDR]], align 4
7820 // CHECK15-NEXT: [[TMP2:%.*]] = load ptr, ptr [[A_ADDR]], align 4
7821 // CHECK15-NEXT: [[TMP3:%.*]] = load i32, ptr [[TMP0]], align 4
7822 // CHECK15-NEXT: store i32 [[TMP3]], ptr [[DOTCAPTURE_EXPR_1]], align 4
7823 // CHECK15-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4
7824 // CHECK15-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP4]], 0
7825 // CHECK15-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1
7826 // CHECK15-NEXT: [[SUB3:%.*]] = sub nsw i32 [[DIV]], 1
7827 // CHECK15-NEXT: store i32 [[SUB3]], ptr [[DOTCAPTURE_EXPR_2]], align 4
7828 // CHECK15-NEXT: store i32 0, ptr [[I]], align 4
7829 // CHECK15-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4
7830 // CHECK15-NEXT: [[CMP:%.*]] = icmp slt i32 0, [[TMP5]]
7831 // CHECK15-NEXT: br i1 [[CMP]], label [[OMP_PRECOND_THEN:%.*]], label [[OMP_PRECOND_END:%.*]]
7832 // CHECK15: omp.precond.then:
7833 // CHECK15-NEXT: store i32 0, ptr [[DOTOMP_COMB_LB]], align 4
7834 // CHECK15-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_2]], align 4
7835 // CHECK15-NEXT: store i32 [[TMP6]], ptr [[DOTOMP_COMB_UB]], align 4
7836 // CHECK15-NEXT: store i32 1, ptr [[DOTOMP_STRIDE]], align 4
7837 // CHECK15-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
7838 // CHECK15-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR__ADDR]], align 4
7839 // CHECK15-NEXT: [[TMP8:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
7840 // CHECK15-NEXT: [[TMP9:%.*]] = load i32, ptr [[TMP8]], align 4
7841 // CHECK15-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB1]], i32 [[TMP9]], i32 91, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 [[TMP7]])
7842 // CHECK15-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
7843 // CHECK15-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_2]], align 4
7844 // CHECK15-NEXT: [[CMP5:%.*]] = icmp sgt i32 [[TMP10]], [[TMP11]]
7845 // CHECK15-NEXT: br i1 [[CMP5]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
7846 // CHECK15: cond.true:
7847 // CHECK15-NEXT: [[TMP12:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_2]], align 4
7848 // CHECK15-NEXT: br label [[COND_END:%.*]]
7849 // CHECK15: cond.false:
7850 // CHECK15-NEXT: [[TMP13:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
7851 // CHECK15-NEXT: br label [[COND_END]]
7852 // CHECK15: cond.end:
7853 // CHECK15-NEXT: [[COND:%.*]] = phi i32 [ [[TMP12]], [[COND_TRUE]] ], [ [[TMP13]], [[COND_FALSE]] ]
7854 // CHECK15-NEXT: store i32 [[COND]], ptr [[DOTOMP_COMB_UB]], align 4
7855 // CHECK15-NEXT: [[TMP14:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
7856 // CHECK15-NEXT: store i32 [[TMP14]], ptr [[DOTOMP_IV]], align 4
7857 // CHECK15-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
7858 // CHECK15: omp.inner.for.cond:
7859 // CHECK15-NEXT: [[TMP15:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
7860 // CHECK15-NEXT: [[TMP16:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_2]], align 4
7861 // CHECK15-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP16]], 1
7862 // CHECK15-NEXT: [[CMP6:%.*]] = icmp slt i32 [[TMP15]], [[ADD]]
7863 // CHECK15-NEXT: br i1 [[CMP6]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
7864 // CHECK15: omp.inner.for.body:
7865 // CHECK15-NEXT: [[TMP17:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
7866 // CHECK15-NEXT: [[TMP18:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
7867 // CHECK15-NEXT: [[TMP19:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR__ADDR]], align 4
7868 // CHECK15-NEXT: store i32 [[TMP19]], ptr [[DOTCAPTURE_EXPR__CASTED]], align 4
7869 // CHECK15-NEXT: [[TMP20:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR__CASTED]], align 4
7870 // CHECK15-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_call(ptr @[[GLOB3]], i32 6, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l158.omp_outlined.omp_outlined, i32 [[TMP17]], i32 [[TMP18]], ptr [[TMP0]], i32 [[TMP1]], ptr [[TMP2]], i32 [[TMP20]])
7871 // CHECK15-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
7872 // CHECK15: omp.inner.for.inc:
7873 // CHECK15-NEXT: [[TMP21:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
7874 // CHECK15-NEXT: [[TMP22:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4
7875 // CHECK15-NEXT: [[ADD7:%.*]] = add nsw i32 [[TMP21]], [[TMP22]]
7876 // CHECK15-NEXT: store i32 [[ADD7]], ptr [[DOTOMP_IV]], align 4
7877 // CHECK15-NEXT: [[TMP23:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
7878 // CHECK15-NEXT: [[TMP24:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4
7879 // CHECK15-NEXT: [[ADD8:%.*]] = add nsw i32 [[TMP23]], [[TMP24]]
7880 // CHECK15-NEXT: store i32 [[ADD8]], ptr [[DOTOMP_COMB_LB]], align 4
7881 // CHECK15-NEXT: [[TMP25:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
7882 // CHECK15-NEXT: [[TMP26:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4
7883 // CHECK15-NEXT: [[ADD9:%.*]] = add nsw i32 [[TMP25]], [[TMP26]]
7884 // CHECK15-NEXT: store i32 [[ADD9]], ptr [[DOTOMP_COMB_UB]], align 4
7885 // CHECK15-NEXT: [[TMP27:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
7886 // CHECK15-NEXT: [[TMP28:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_2]], align 4
7887 // CHECK15-NEXT: [[CMP10:%.*]] = icmp sgt i32 [[TMP27]], [[TMP28]]
7888 // CHECK15-NEXT: br i1 [[CMP10]], label [[COND_TRUE11:%.*]], label [[COND_FALSE12:%.*]]
7889 // CHECK15: cond.true11:
7890 // CHECK15-NEXT: [[TMP29:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_2]], align 4
7891 // CHECK15-NEXT: br label [[COND_END13:%.*]]
7892 // CHECK15: cond.false12:
7893 // CHECK15-NEXT: [[TMP30:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
7894 // CHECK15-NEXT: br label [[COND_END13]]
7895 // CHECK15: cond.end13:
7896 // CHECK15-NEXT: [[COND14:%.*]] = phi i32 [ [[TMP29]], [[COND_TRUE11]] ], [ [[TMP30]], [[COND_FALSE12]] ]
7897 // CHECK15-NEXT: store i32 [[COND14]], ptr [[DOTOMP_COMB_UB]], align 4
7898 // CHECK15-NEXT: [[TMP31:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
7899 // CHECK15-NEXT: store i32 [[TMP31]], ptr [[DOTOMP_IV]], align 4
7900 // CHECK15-NEXT: br label [[OMP_INNER_FOR_COND]]
7901 // CHECK15: omp.inner.for.end:
7902 // CHECK15-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
7903 // CHECK15: omp.loop.exit:
7904 // CHECK15-NEXT: [[TMP32:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
7905 // CHECK15-NEXT: [[TMP33:%.*]] = load i32, ptr [[TMP32]], align 4
7906 // CHECK15-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB1]], i32 [[TMP33]])
7907 // CHECK15-NEXT: br label [[OMP_PRECOND_END]]
7908 // CHECK15: omp.precond.end:
7909 // CHECK15-NEXT: ret void
7912 // CHECK15-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l158.omp_outlined.omp_outlined
7913 // CHECK15-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], i32 noundef [[DOTPREVIOUS_LB_:%.*]], i32 noundef [[DOTPREVIOUS_UB_:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[N:%.*]], i32 noundef [[VLA:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[A:%.*]], i32 noundef [[DOTCAPTURE_EXPR_:%.*]]) #[[ATTR2]] {
7914 // CHECK15-NEXT: entry:
7915 // CHECK15-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 4
7916 // CHECK15-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 4
7917 // CHECK15-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i32, align 4
7918 // CHECK15-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i32, align 4
7919 // CHECK15-NEXT: [[N_ADDR:%.*]] = alloca ptr, align 4
7920 // CHECK15-NEXT: [[VLA_ADDR:%.*]] = alloca i32, align 4
7921 // CHECK15-NEXT: [[A_ADDR:%.*]] = alloca ptr, align 4
7922 // CHECK15-NEXT: [[DOTCAPTURE_EXPR__ADDR:%.*]] = alloca i32, align 4
7923 // CHECK15-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
7924 // CHECK15-NEXT: [[TMP:%.*]] = alloca i32, align 4
7925 // CHECK15-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4
7926 // CHECK15-NEXT: [[DOTCAPTURE_EXPR_2:%.*]] = alloca i32, align 4
7927 // CHECK15-NEXT: [[I:%.*]] = alloca i32, align 4
7928 // CHECK15-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4
7929 // CHECK15-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4
7930 // CHECK15-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
7931 // CHECK15-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
7932 // CHECK15-NEXT: [[I4:%.*]] = alloca i32, align 4
7933 // CHECK15-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 4
7934 // CHECK15-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 4
7935 // CHECK15-NEXT: store i32 [[DOTPREVIOUS_LB_]], ptr [[DOTPREVIOUS_LB__ADDR]], align 4
7936 // CHECK15-NEXT: store i32 [[DOTPREVIOUS_UB_]], ptr [[DOTPREVIOUS_UB__ADDR]], align 4
7937 // CHECK15-NEXT: store ptr [[N]], ptr [[N_ADDR]], align 4
7938 // CHECK15-NEXT: store i32 [[VLA]], ptr [[VLA_ADDR]], align 4
7939 // CHECK15-NEXT: store ptr [[A]], ptr [[A_ADDR]], align 4
7940 // CHECK15-NEXT: store i32 [[DOTCAPTURE_EXPR_]], ptr [[DOTCAPTURE_EXPR__ADDR]], align 4
7941 // CHECK15-NEXT: [[TMP0:%.*]] = load ptr, ptr [[N_ADDR]], align 4
7942 // CHECK15-NEXT: [[TMP1:%.*]] = load i32, ptr [[VLA_ADDR]], align 4
7943 // CHECK15-NEXT: [[TMP2:%.*]] = load ptr, ptr [[A_ADDR]], align 4
7944 // CHECK15-NEXT: [[TMP3:%.*]] = load i32, ptr [[TMP0]], align 4
7945 // CHECK15-NEXT: store i32 [[TMP3]], ptr [[DOTCAPTURE_EXPR_1]], align 4
7946 // CHECK15-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4
7947 // CHECK15-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP4]], 0
7948 // CHECK15-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1
7949 // CHECK15-NEXT: [[SUB3:%.*]] = sub nsw i32 [[DIV]], 1
7950 // CHECK15-NEXT: store i32 [[SUB3]], ptr [[DOTCAPTURE_EXPR_2]], align 4
7951 // CHECK15-NEXT: store i32 0, ptr [[I]], align 4
7952 // CHECK15-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4
7953 // CHECK15-NEXT: [[CMP:%.*]] = icmp slt i32 0, [[TMP5]]
7954 // CHECK15-NEXT: br i1 [[CMP]], label [[OMP_PRECOND_THEN:%.*]], label [[OMP_PRECOND_END:%.*]]
7955 // CHECK15: omp.precond.then:
7956 // CHECK15-NEXT: store i32 0, ptr [[DOTOMP_LB]], align 4
7957 // CHECK15-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_2]], align 4
7958 // CHECK15-NEXT: store i32 [[TMP6]], ptr [[DOTOMP_UB]], align 4
7959 // CHECK15-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTPREVIOUS_LB__ADDR]], align 4
7960 // CHECK15-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTPREVIOUS_UB__ADDR]], align 4
7961 // CHECK15-NEXT: store i32 [[TMP7]], ptr [[DOTOMP_LB]], align 4
7962 // CHECK15-NEXT: store i32 [[TMP8]], ptr [[DOTOMP_UB]], align 4
7963 // CHECK15-NEXT: store i32 1, ptr [[DOTOMP_STRIDE]], align 4
7964 // CHECK15-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
7965 // CHECK15-NEXT: [[TMP9:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
7966 // CHECK15-NEXT: [[TMP10:%.*]] = load i32, ptr [[TMP9]], align 4
7967 // CHECK15-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB2]], i32 [[TMP10]], i32 34, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
7968 // CHECK15-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
7969 // CHECK15-NEXT: [[TMP12:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_2]], align 4
7970 // CHECK15-NEXT: [[CMP5:%.*]] = icmp sgt i32 [[TMP11]], [[TMP12]]
7971 // CHECK15-NEXT: br i1 [[CMP5]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
7972 // CHECK15: cond.true:
7973 // CHECK15-NEXT: [[TMP13:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_2]], align 4
7974 // CHECK15-NEXT: br label [[COND_END:%.*]]
7975 // CHECK15: cond.false:
7976 // CHECK15-NEXT: [[TMP14:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
7977 // CHECK15-NEXT: br label [[COND_END]]
7978 // CHECK15: cond.end:
7979 // CHECK15-NEXT: [[COND:%.*]] = phi i32 [ [[TMP13]], [[COND_TRUE]] ], [ [[TMP14]], [[COND_FALSE]] ]
7980 // CHECK15-NEXT: store i32 [[COND]], ptr [[DOTOMP_UB]], align 4
7981 // CHECK15-NEXT: [[TMP15:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4
7982 // CHECK15-NEXT: store i32 [[TMP15]], ptr [[DOTOMP_IV]], align 4
7983 // CHECK15-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
7984 // CHECK15: omp.inner.for.cond:
7985 // CHECK15-NEXT: [[TMP16:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
7986 // CHECK15-NEXT: [[TMP17:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
7987 // CHECK15-NEXT: [[CMP6:%.*]] = icmp sle i32 [[TMP16]], [[TMP17]]
7988 // CHECK15-NEXT: br i1 [[CMP6]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
7989 // CHECK15: omp.inner.for.body:
7990 // CHECK15-NEXT: [[TMP18:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
7991 // CHECK15-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP18]], 1
7992 // CHECK15-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
7993 // CHECK15-NEXT: store i32 [[ADD]], ptr [[I4]], align 4
7994 // CHECK15-NEXT: [[TMP19:%.*]] = load i32, ptr [[I4]], align 4
7995 // CHECK15-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i32, ptr [[TMP2]], i32 [[TMP19]]
7996 // CHECK15-NEXT: store i32 0, ptr [[ARRAYIDX]], align 4
7997 // CHECK15-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
7998 // CHECK15: omp.body.continue:
7999 // CHECK15-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
8000 // CHECK15: omp.inner.for.inc:
8001 // CHECK15-NEXT: [[TMP20:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
8002 // CHECK15-NEXT: [[ADD7:%.*]] = add nsw i32 [[TMP20]], 1
8003 // CHECK15-NEXT: store i32 [[ADD7]], ptr [[DOTOMP_IV]], align 4
8004 // CHECK15-NEXT: br label [[OMP_INNER_FOR_COND]]
8005 // CHECK15: omp.inner.for.end:
8006 // CHECK15-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
8007 // CHECK15: omp.loop.exit:
8008 // CHECK15-NEXT: [[TMP21:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
8009 // CHECK15-NEXT: [[TMP22:%.*]] = load i32, ptr [[TMP21]], align 4
8010 // CHECK15-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB2]], i32 [[TMP22]])
8011 // CHECK15-NEXT: br label [[OMP_PRECOND_END]]
8012 // CHECK15: omp.precond.end:
8013 // CHECK15-NEXT: ret void
8016 // CHECK15-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l163
8017 // CHECK15-SAME: (i32 noundef [[N:%.*]], i32 noundef [[VLA:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[A:%.*]]) #[[ATTR2]] {
8018 // CHECK15-NEXT: entry:
8019 // CHECK15-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4
8020 // CHECK15-NEXT: [[VLA_ADDR:%.*]] = alloca i32, align 4
8021 // CHECK15-NEXT: [[A_ADDR:%.*]] = alloca ptr, align 4
8022 // CHECK15-NEXT: store i32 [[N]], ptr [[N_ADDR]], align 4
8023 // CHECK15-NEXT: store i32 [[VLA]], ptr [[VLA_ADDR]], align 4
8024 // CHECK15-NEXT: store ptr [[A]], ptr [[A_ADDR]], align 4
8025 // CHECK15-NEXT: [[TMP0:%.*]] = load i32, ptr [[VLA_ADDR]], align 4
8026 // CHECK15-NEXT: [[TMP1:%.*]] = load ptr, ptr [[A_ADDR]], align 4
8027 // CHECK15-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_teams(ptr @[[GLOB3]], i32 3, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l163.omp_outlined, ptr [[N_ADDR]], i32 [[TMP0]], ptr [[TMP1]])
8028 // CHECK15-NEXT: ret void
8031 // CHECK15-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l163.omp_outlined
8032 // CHECK15-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[N:%.*]], i32 noundef [[VLA:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[A:%.*]]) #[[ATTR2]] {
8033 // CHECK15-NEXT: entry:
8034 // CHECK15-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 4
8035 // CHECK15-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 4
8036 // CHECK15-NEXT: [[N_ADDR:%.*]] = alloca ptr, align 4
8037 // CHECK15-NEXT: [[VLA_ADDR:%.*]] = alloca i32, align 4
8038 // CHECK15-NEXT: [[A_ADDR:%.*]] = alloca ptr, align 4
8039 // CHECK15-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
8040 // CHECK15-NEXT: [[TMP:%.*]] = alloca i32, align 4
8041 // CHECK15-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4
8042 // CHECK15-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4
8043 // CHECK15-NEXT: [[I:%.*]] = alloca i32, align 4
8044 // CHECK15-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4
8045 // CHECK15-NEXT: [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4
8046 // CHECK15-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
8047 // CHECK15-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
8048 // CHECK15-NEXT: [[I3:%.*]] = alloca i32, align 4
8049 // CHECK15-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 4
8050 // CHECK15-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 4
8051 // CHECK15-NEXT: store ptr [[N]], ptr [[N_ADDR]], align 4
8052 // CHECK15-NEXT: store i32 [[VLA]], ptr [[VLA_ADDR]], align 4
8053 // CHECK15-NEXT: store ptr [[A]], ptr [[A_ADDR]], align 4
8054 // CHECK15-NEXT: [[TMP0:%.*]] = load ptr, ptr [[N_ADDR]], align 4
8055 // CHECK15-NEXT: [[TMP1:%.*]] = load i32, ptr [[VLA_ADDR]], align 4
8056 // CHECK15-NEXT: [[TMP2:%.*]] = load ptr, ptr [[A_ADDR]], align 4
8057 // CHECK15-NEXT: [[TMP3:%.*]] = load i32, ptr [[TMP0]], align 4
8058 // CHECK15-NEXT: store i32 [[TMP3]], ptr [[DOTCAPTURE_EXPR_]], align 4
8059 // CHECK15-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4
8060 // CHECK15-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP4]], 0
8061 // CHECK15-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1
8062 // CHECK15-NEXT: [[SUB2:%.*]] = sub nsw i32 [[DIV]], 1
8063 // CHECK15-NEXT: store i32 [[SUB2]], ptr [[DOTCAPTURE_EXPR_1]], align 4
8064 // CHECK15-NEXT: store i32 0, ptr [[I]], align 4
8065 // CHECK15-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4
8066 // CHECK15-NEXT: [[CMP:%.*]] = icmp slt i32 0, [[TMP5]]
8067 // CHECK15-NEXT: br i1 [[CMP]], label [[OMP_PRECOND_THEN:%.*]], label [[OMP_PRECOND_END:%.*]]
8068 // CHECK15: omp.precond.then:
8069 // CHECK15-NEXT: store i32 0, ptr [[DOTOMP_COMB_LB]], align 4
8070 // CHECK15-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4
8071 // CHECK15-NEXT: store i32 [[TMP6]], ptr [[DOTOMP_COMB_UB]], align 4
8072 // CHECK15-NEXT: store i32 1, ptr [[DOTOMP_STRIDE]], align 4
8073 // CHECK15-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
8074 // CHECK15-NEXT: [[TMP7:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
8075 // CHECK15-NEXT: [[TMP8:%.*]] = load i32, ptr [[TMP7]], align 4
8076 // CHECK15-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB1]], i32 [[TMP8]], i32 92, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
8077 // CHECK15-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
8078 // CHECK15-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4
8079 // CHECK15-NEXT: [[CMP4:%.*]] = icmp sgt i32 [[TMP9]], [[TMP10]]
8080 // CHECK15-NEXT: br i1 [[CMP4]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
8081 // CHECK15: cond.true:
8082 // CHECK15-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4
8083 // CHECK15-NEXT: br label [[COND_END:%.*]]
8084 // CHECK15: cond.false:
8085 // CHECK15-NEXT: [[TMP12:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
8086 // CHECK15-NEXT: br label [[COND_END]]
8087 // CHECK15: cond.end:
8088 // CHECK15-NEXT: [[COND:%.*]] = phi i32 [ [[TMP11]], [[COND_TRUE]] ], [ [[TMP12]], [[COND_FALSE]] ]
8089 // CHECK15-NEXT: store i32 [[COND]], ptr [[DOTOMP_COMB_UB]], align 4
8090 // CHECK15-NEXT: [[TMP13:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
8091 // CHECK15-NEXT: store i32 [[TMP13]], ptr [[DOTOMP_IV]], align 4
8092 // CHECK15-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
8093 // CHECK15: omp.inner.for.cond:
8094 // CHECK15-NEXT: [[TMP14:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
8095 // CHECK15-NEXT: [[TMP15:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
8096 // CHECK15-NEXT: [[CMP5:%.*]] = icmp sle i32 [[TMP14]], [[TMP15]]
8097 // CHECK15-NEXT: br i1 [[CMP5]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
8098 // CHECK15: omp.inner.for.body:
8099 // CHECK15-NEXT: [[TMP16:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
8100 // CHECK15-NEXT: [[TMP17:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
8101 // CHECK15-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_call(ptr @[[GLOB3]], i32 5, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l163.omp_outlined.omp_outlined, i32 [[TMP16]], i32 [[TMP17]], ptr [[TMP0]], i32 [[TMP1]], ptr [[TMP2]])
8102 // CHECK15-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
8103 // CHECK15: omp.inner.for.inc:
8104 // CHECK15-NEXT: [[TMP18:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
8105 // CHECK15-NEXT: [[TMP19:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4
8106 // CHECK15-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP18]], [[TMP19]]
8107 // CHECK15-NEXT: store i32 [[ADD]], ptr [[DOTOMP_IV]], align 4
8108 // CHECK15-NEXT: br label [[OMP_INNER_FOR_COND]]
8109 // CHECK15: omp.inner.for.end:
8110 // CHECK15-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
8111 // CHECK15: omp.loop.exit:
8112 // CHECK15-NEXT: [[TMP20:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
8113 // CHECK15-NEXT: [[TMP21:%.*]] = load i32, ptr [[TMP20]], align 4
8114 // CHECK15-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB1]], i32 [[TMP21]])
8115 // CHECK15-NEXT: br label [[OMP_PRECOND_END]]
8116 // CHECK15: omp.precond.end:
8117 // CHECK15-NEXT: ret void
8120 // CHECK15-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l163.omp_outlined.omp_outlined
8121 // CHECK15-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], i32 noundef [[DOTPREVIOUS_LB_:%.*]], i32 noundef [[DOTPREVIOUS_UB_:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[N:%.*]], i32 noundef [[VLA:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[A:%.*]]) #[[ATTR2]] {
8122 // CHECK15-NEXT: entry:
8123 // CHECK15-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 4
8124 // CHECK15-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 4
8125 // CHECK15-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i32, align 4
8126 // CHECK15-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i32, align 4
8127 // CHECK15-NEXT: [[N_ADDR:%.*]] = alloca ptr, align 4
8128 // CHECK15-NEXT: [[VLA_ADDR:%.*]] = alloca i32, align 4
8129 // CHECK15-NEXT: [[A_ADDR:%.*]] = alloca ptr, align 4
8130 // CHECK15-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
8131 // CHECK15-NEXT: [[TMP:%.*]] = alloca i32, align 4
8132 // CHECK15-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4
8133 // CHECK15-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4
8134 // CHECK15-NEXT: [[I:%.*]] = alloca i32, align 4
8135 // CHECK15-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4
8136 // CHECK15-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4
8137 // CHECK15-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
8138 // CHECK15-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
8139 // CHECK15-NEXT: [[I3:%.*]] = alloca i32, align 4
8140 // CHECK15-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 4
8141 // CHECK15-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 4
8142 // CHECK15-NEXT: store i32 [[DOTPREVIOUS_LB_]], ptr [[DOTPREVIOUS_LB__ADDR]], align 4
8143 // CHECK15-NEXT: store i32 [[DOTPREVIOUS_UB_]], ptr [[DOTPREVIOUS_UB__ADDR]], align 4
8144 // CHECK15-NEXT: store ptr [[N]], ptr [[N_ADDR]], align 4
8145 // CHECK15-NEXT: store i32 [[VLA]], ptr [[VLA_ADDR]], align 4
8146 // CHECK15-NEXT: store ptr [[A]], ptr [[A_ADDR]], align 4
8147 // CHECK15-NEXT: [[TMP0:%.*]] = load ptr, ptr [[N_ADDR]], align 4
8148 // CHECK15-NEXT: [[TMP1:%.*]] = load i32, ptr [[VLA_ADDR]], align 4
8149 // CHECK15-NEXT: [[TMP2:%.*]] = load ptr, ptr [[A_ADDR]], align 4
8150 // CHECK15-NEXT: [[TMP3:%.*]] = load i32, ptr [[TMP0]], align 4
8151 // CHECK15-NEXT: store i32 [[TMP3]], ptr [[DOTCAPTURE_EXPR_]], align 4
8152 // CHECK15-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4
8153 // CHECK15-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP4]], 0
8154 // CHECK15-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1
8155 // CHECK15-NEXT: [[SUB2:%.*]] = sub nsw i32 [[DIV]], 1
8156 // CHECK15-NEXT: store i32 [[SUB2]], ptr [[DOTCAPTURE_EXPR_1]], align 4
8157 // CHECK15-NEXT: store i32 0, ptr [[I]], align 4
8158 // CHECK15-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4
8159 // CHECK15-NEXT: [[CMP:%.*]] = icmp slt i32 0, [[TMP5]]
8160 // CHECK15-NEXT: br i1 [[CMP]], label [[OMP_PRECOND_THEN:%.*]], label [[OMP_PRECOND_END:%.*]]
8161 // CHECK15: omp.precond.then:
8162 // CHECK15-NEXT: store i32 0, ptr [[DOTOMP_LB]], align 4
8163 // CHECK15-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4
8164 // CHECK15-NEXT: store i32 [[TMP6]], ptr [[DOTOMP_UB]], align 4
8165 // CHECK15-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTPREVIOUS_LB__ADDR]], align 4
8166 // CHECK15-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTPREVIOUS_UB__ADDR]], align 4
8167 // CHECK15-NEXT: store i32 [[TMP7]], ptr [[DOTOMP_LB]], align 4
8168 // CHECK15-NEXT: store i32 [[TMP8]], ptr [[DOTOMP_UB]], align 4
8169 // CHECK15-NEXT: store i32 1, ptr [[DOTOMP_STRIDE]], align 4
8170 // CHECK15-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
8171 // CHECK15-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4
8172 // CHECK15-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
8173 // CHECK15-NEXT: [[TMP11:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
8174 // CHECK15-NEXT: [[TMP12:%.*]] = load i32, ptr [[TMP11]], align 4
8175 // CHECK15-NEXT: call void @__kmpc_dispatch_init_4(ptr @[[GLOB3]], i32 [[TMP12]], i32 1073741859, i32 [[TMP9]], i32 [[TMP10]], i32 1, i32 1)
8176 // CHECK15-NEXT: br label [[OMP_DISPATCH_COND:%.*]]
8177 // CHECK15: omp.dispatch.cond:
8178 // CHECK15-NEXT: [[TMP13:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
8179 // CHECK15-NEXT: [[TMP14:%.*]] = load i32, ptr [[TMP13]], align 4
8180 // CHECK15-NEXT: [[TMP15:%.*]] = call i32 @__kmpc_dispatch_next_4(ptr @[[GLOB3]], i32 [[TMP14]], ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]])
8181 // CHECK15-NEXT: [[TOBOOL:%.*]] = icmp ne i32 [[TMP15]], 0
8182 // CHECK15-NEXT: br i1 [[TOBOOL]], label [[OMP_DISPATCH_BODY:%.*]], label [[OMP_DISPATCH_END:%.*]]
8183 // CHECK15: omp.dispatch.body:
8184 // CHECK15-NEXT: [[TMP16:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4
8185 // CHECK15-NEXT: store i32 [[TMP16]], ptr [[DOTOMP_IV]], align 4
8186 // CHECK15-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
8187 // CHECK15: omp.inner.for.cond:
8188 // CHECK15-NEXT: [[TMP17:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP16:![0-9]+]]
8189 // CHECK15-NEXT: [[TMP18:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP16]]
8190 // CHECK15-NEXT: [[CMP4:%.*]] = icmp sle i32 [[TMP17]], [[TMP18]]
8191 // CHECK15-NEXT: br i1 [[CMP4]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
8192 // CHECK15: omp.inner.for.body:
8193 // CHECK15-NEXT: [[TMP19:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP16]]
8194 // CHECK15-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP19]], 1
8195 // CHECK15-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
8196 // CHECK15-NEXT: store i32 [[ADD]], ptr [[I3]], align 4, !llvm.access.group [[ACC_GRP16]]
8197 // CHECK15-NEXT: [[TMP20:%.*]] = load i32, ptr [[I3]], align 4, !llvm.access.group [[ACC_GRP16]]
8198 // CHECK15-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i32, ptr [[TMP2]], i32 [[TMP20]]
8199 // CHECK15-NEXT: store i32 0, ptr [[ARRAYIDX]], align 4, !llvm.access.group [[ACC_GRP16]]
8200 // CHECK15-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
8201 // CHECK15: omp.body.continue:
8202 // CHECK15-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
8203 // CHECK15: omp.inner.for.inc:
8204 // CHECK15-NEXT: [[TMP21:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP16]]
8205 // CHECK15-NEXT: [[ADD5:%.*]] = add nsw i32 [[TMP21]], 1
8206 // CHECK15-NEXT: store i32 [[ADD5]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP16]]
8207 // CHECK15-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP17:![0-9]+]]
8208 // CHECK15: omp.inner.for.end:
8209 // CHECK15-NEXT: br label [[OMP_DISPATCH_INC:%.*]]
8210 // CHECK15: omp.dispatch.inc:
8211 // CHECK15-NEXT: br label [[OMP_DISPATCH_COND]]
8212 // CHECK15: omp.dispatch.end:
8213 // CHECK15-NEXT: [[TMP22:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
8214 // CHECK15-NEXT: [[TMP23:%.*]] = load i32, ptr [[TMP22]], align 4
8215 // CHECK15-NEXT: call void @__kmpc_dispatch_deinit(ptr @[[GLOB3]], i32 [[TMP23]])
8216 // CHECK15-NEXT: br label [[OMP_PRECOND_END]]
8217 // CHECK15: omp.precond.end:
8218 // CHECK15-NEXT: ret void
8221 // CHECK15-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l168
8222 // CHECK15-SAME: (i32 noundef [[M:%.*]], i32 noundef [[N:%.*]], i32 noundef [[VLA:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[A:%.*]]) #[[ATTR2]] {
8223 // CHECK15-NEXT: entry:
8224 // CHECK15-NEXT: [[M_ADDR:%.*]] = alloca i32, align 4
8225 // CHECK15-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4
8226 // CHECK15-NEXT: [[VLA_ADDR:%.*]] = alloca i32, align 4
8227 // CHECK15-NEXT: [[A_ADDR:%.*]] = alloca ptr, align 4
8228 // CHECK15-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4
8229 // CHECK15-NEXT: [[DOTCAPTURE_EXPR__CASTED:%.*]] = alloca i32, align 4
8230 // CHECK15-NEXT: store i32 [[M]], ptr [[M_ADDR]], align 4
8231 // CHECK15-NEXT: store i32 [[N]], ptr [[N_ADDR]], align 4
8232 // CHECK15-NEXT: store i32 [[VLA]], ptr [[VLA_ADDR]], align 4
8233 // CHECK15-NEXT: store ptr [[A]], ptr [[A_ADDR]], align 4
8234 // CHECK15-NEXT: [[TMP0:%.*]] = load i32, ptr [[VLA_ADDR]], align 4
8235 // CHECK15-NEXT: [[TMP1:%.*]] = load ptr, ptr [[A_ADDR]], align 4
8236 // CHECK15-NEXT: [[TMP2:%.*]] = load i32, ptr [[M_ADDR]], align 4
8237 // CHECK15-NEXT: store i32 [[TMP2]], ptr [[DOTCAPTURE_EXPR_]], align 4
8238 // CHECK15-NEXT: [[TMP3:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4
8239 // CHECK15-NEXT: store i32 [[TMP3]], ptr [[DOTCAPTURE_EXPR__CASTED]], align 4
8240 // CHECK15-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR__CASTED]], align 4
8241 // CHECK15-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_teams(ptr @[[GLOB3]], i32 4, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l168.omp_outlined, ptr [[N_ADDR]], i32 [[TMP0]], ptr [[TMP1]], i32 [[TMP4]])
8242 // CHECK15-NEXT: ret void
8245 // CHECK15-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l168.omp_outlined
8246 // CHECK15-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[N:%.*]], i32 noundef [[VLA:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[A:%.*]], i32 noundef [[DOTCAPTURE_EXPR_:%.*]]) #[[ATTR2]] {
8247 // CHECK15-NEXT: entry:
8248 // CHECK15-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 4
8249 // CHECK15-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 4
8250 // CHECK15-NEXT: [[N_ADDR:%.*]] = alloca ptr, align 4
8251 // CHECK15-NEXT: [[VLA_ADDR:%.*]] = alloca i32, align 4
8252 // CHECK15-NEXT: [[A_ADDR:%.*]] = alloca ptr, align 4
8253 // CHECK15-NEXT: [[DOTCAPTURE_EXPR__ADDR:%.*]] = alloca i32, align 4
8254 // CHECK15-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
8255 // CHECK15-NEXT: [[TMP:%.*]] = alloca i32, align 4
8256 // CHECK15-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4
8257 // CHECK15-NEXT: [[DOTCAPTURE_EXPR_2:%.*]] = alloca i32, align 4
8258 // CHECK15-NEXT: [[I:%.*]] = alloca i32, align 4
8259 // CHECK15-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4
8260 // CHECK15-NEXT: [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4
8261 // CHECK15-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
8262 // CHECK15-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
8263 // CHECK15-NEXT: [[I4:%.*]] = alloca i32, align 4
8264 // CHECK15-NEXT: [[DOTCAPTURE_EXPR__CASTED:%.*]] = alloca i32, align 4
8265 // CHECK15-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 4
8266 // CHECK15-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 4
8267 // CHECK15-NEXT: store ptr [[N]], ptr [[N_ADDR]], align 4
8268 // CHECK15-NEXT: store i32 [[VLA]], ptr [[VLA_ADDR]], align 4
8269 // CHECK15-NEXT: store ptr [[A]], ptr [[A_ADDR]], align 4
8270 // CHECK15-NEXT: store i32 [[DOTCAPTURE_EXPR_]], ptr [[DOTCAPTURE_EXPR__ADDR]], align 4
8271 // CHECK15-NEXT: [[TMP0:%.*]] = load ptr, ptr [[N_ADDR]], align 4
8272 // CHECK15-NEXT: [[TMP1:%.*]] = load i32, ptr [[VLA_ADDR]], align 4
8273 // CHECK15-NEXT: [[TMP2:%.*]] = load ptr, ptr [[A_ADDR]], align 4
8274 // CHECK15-NEXT: [[TMP3:%.*]] = load i32, ptr [[TMP0]], align 4
8275 // CHECK15-NEXT: store i32 [[TMP3]], ptr [[DOTCAPTURE_EXPR_1]], align 4
8276 // CHECK15-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4
8277 // CHECK15-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP4]], 0
8278 // CHECK15-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1
8279 // CHECK15-NEXT: [[SUB3:%.*]] = sub nsw i32 [[DIV]], 1
8280 // CHECK15-NEXT: store i32 [[SUB3]], ptr [[DOTCAPTURE_EXPR_2]], align 4
8281 // CHECK15-NEXT: store i32 0, ptr [[I]], align 4
8282 // CHECK15-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4
8283 // CHECK15-NEXT: [[CMP:%.*]] = icmp slt i32 0, [[TMP5]]
8284 // CHECK15-NEXT: br i1 [[CMP]], label [[OMP_PRECOND_THEN:%.*]], label [[OMP_PRECOND_END:%.*]]
8285 // CHECK15: omp.precond.then:
8286 // CHECK15-NEXT: store i32 0, ptr [[DOTOMP_COMB_LB]], align 4
8287 // CHECK15-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_2]], align 4
8288 // CHECK15-NEXT: store i32 [[TMP6]], ptr [[DOTOMP_COMB_UB]], align 4
8289 // CHECK15-NEXT: store i32 1, ptr [[DOTOMP_STRIDE]], align 4
8290 // CHECK15-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
8291 // CHECK15-NEXT: [[TMP7:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
8292 // CHECK15-NEXT: [[TMP8:%.*]] = load i32, ptr [[TMP7]], align 4
8293 // CHECK15-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB1]], i32 [[TMP8]], i32 92, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
8294 // CHECK15-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
8295 // CHECK15-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_2]], align 4
8296 // CHECK15-NEXT: [[CMP5:%.*]] = icmp sgt i32 [[TMP9]], [[TMP10]]
8297 // CHECK15-NEXT: br i1 [[CMP5]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
8298 // CHECK15: cond.true:
8299 // CHECK15-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_2]], align 4
8300 // CHECK15-NEXT: br label [[COND_END:%.*]]
8301 // CHECK15: cond.false:
8302 // CHECK15-NEXT: [[TMP12:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
8303 // CHECK15-NEXT: br label [[COND_END]]
8304 // CHECK15: cond.end:
8305 // CHECK15-NEXT: [[COND:%.*]] = phi i32 [ [[TMP11]], [[COND_TRUE]] ], [ [[TMP12]], [[COND_FALSE]] ]
8306 // CHECK15-NEXT: store i32 [[COND]], ptr [[DOTOMP_COMB_UB]], align 4
8307 // CHECK15-NEXT: [[TMP13:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
8308 // CHECK15-NEXT: store i32 [[TMP13]], ptr [[DOTOMP_IV]], align 4
8309 // CHECK15-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
8310 // CHECK15: omp.inner.for.cond:
8311 // CHECK15-NEXT: [[TMP14:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
8312 // CHECK15-NEXT: [[TMP15:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
8313 // CHECK15-NEXT: [[CMP6:%.*]] = icmp sle i32 [[TMP14]], [[TMP15]]
8314 // CHECK15-NEXT: br i1 [[CMP6]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
8315 // CHECK15: omp.inner.for.body:
8316 // CHECK15-NEXT: [[TMP16:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
8317 // CHECK15-NEXT: [[TMP17:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
8318 // CHECK15-NEXT: [[TMP18:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR__ADDR]], align 4
8319 // CHECK15-NEXT: store i32 [[TMP18]], ptr [[DOTCAPTURE_EXPR__CASTED]], align 4
8320 // CHECK15-NEXT: [[TMP19:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR__CASTED]], align 4
8321 // CHECK15-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_call(ptr @[[GLOB3]], i32 6, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l168.omp_outlined.omp_outlined, i32 [[TMP16]], i32 [[TMP17]], ptr [[TMP0]], i32 [[TMP1]], ptr [[TMP2]], i32 [[TMP19]])
8322 // CHECK15-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
8323 // CHECK15: omp.inner.for.inc:
8324 // CHECK15-NEXT: [[TMP20:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
8325 // CHECK15-NEXT: [[TMP21:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4
8326 // CHECK15-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP20]], [[TMP21]]
8327 // CHECK15-NEXT: store i32 [[ADD]], ptr [[DOTOMP_IV]], align 4
8328 // CHECK15-NEXT: br label [[OMP_INNER_FOR_COND]]
8329 // CHECK15: omp.inner.for.end:
8330 // CHECK15-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
8331 // CHECK15: omp.loop.exit:
8332 // CHECK15-NEXT: [[TMP22:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
8333 // CHECK15-NEXT: [[TMP23:%.*]] = load i32, ptr [[TMP22]], align 4
8334 // CHECK15-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB1]], i32 [[TMP23]])
8335 // CHECK15-NEXT: br label [[OMP_PRECOND_END]]
8336 // CHECK15: omp.precond.end:
8337 // CHECK15-NEXT: ret void
8340 // CHECK15-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l168.omp_outlined.omp_outlined
8341 // CHECK15-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], i32 noundef [[DOTPREVIOUS_LB_:%.*]], i32 noundef [[DOTPREVIOUS_UB_:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[N:%.*]], i32 noundef [[VLA:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[A:%.*]], i32 noundef [[DOTCAPTURE_EXPR_:%.*]]) #[[ATTR2]] {
8342 // CHECK15-NEXT: entry:
8343 // CHECK15-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 4
8344 // CHECK15-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 4
8345 // CHECK15-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i32, align 4
8346 // CHECK15-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i32, align 4
8347 // CHECK15-NEXT: [[N_ADDR:%.*]] = alloca ptr, align 4
8348 // CHECK15-NEXT: [[VLA_ADDR:%.*]] = alloca i32, align 4
8349 // CHECK15-NEXT: [[A_ADDR:%.*]] = alloca ptr, align 4
8350 // CHECK15-NEXT: [[DOTCAPTURE_EXPR__ADDR:%.*]] = alloca i32, align 4
8351 // CHECK15-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
8352 // CHECK15-NEXT: [[TMP:%.*]] = alloca i32, align 4
8353 // CHECK15-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4
8354 // CHECK15-NEXT: [[DOTCAPTURE_EXPR_2:%.*]] = alloca i32, align 4
8355 // CHECK15-NEXT: [[I:%.*]] = alloca i32, align 4
8356 // CHECK15-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4
8357 // CHECK15-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4
8358 // CHECK15-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
8359 // CHECK15-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
8360 // CHECK15-NEXT: [[I4:%.*]] = alloca i32, align 4
8361 // CHECK15-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 4
8362 // CHECK15-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 4
8363 // CHECK15-NEXT: store i32 [[DOTPREVIOUS_LB_]], ptr [[DOTPREVIOUS_LB__ADDR]], align 4
8364 // CHECK15-NEXT: store i32 [[DOTPREVIOUS_UB_]], ptr [[DOTPREVIOUS_UB__ADDR]], align 4
8365 // CHECK15-NEXT: store ptr [[N]], ptr [[N_ADDR]], align 4
8366 // CHECK15-NEXT: store i32 [[VLA]], ptr [[VLA_ADDR]], align 4
8367 // CHECK15-NEXT: store ptr [[A]], ptr [[A_ADDR]], align 4
8368 // CHECK15-NEXT: store i32 [[DOTCAPTURE_EXPR_]], ptr [[DOTCAPTURE_EXPR__ADDR]], align 4
8369 // CHECK15-NEXT: [[TMP0:%.*]] = load ptr, ptr [[N_ADDR]], align 4
8370 // CHECK15-NEXT: [[TMP1:%.*]] = load i32, ptr [[VLA_ADDR]], align 4
8371 // CHECK15-NEXT: [[TMP2:%.*]] = load ptr, ptr [[A_ADDR]], align 4
8372 // CHECK15-NEXT: [[TMP3:%.*]] = load i32, ptr [[TMP0]], align 4
8373 // CHECK15-NEXT: store i32 [[TMP3]], ptr [[DOTCAPTURE_EXPR_1]], align 4
8374 // CHECK15-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4
8375 // CHECK15-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP4]], 0
8376 // CHECK15-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1
8377 // CHECK15-NEXT: [[SUB3:%.*]] = sub nsw i32 [[DIV]], 1
8378 // CHECK15-NEXT: store i32 [[SUB3]], ptr [[DOTCAPTURE_EXPR_2]], align 4
8379 // CHECK15-NEXT: store i32 0, ptr [[I]], align 4
8380 // CHECK15-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4
8381 // CHECK15-NEXT: [[CMP:%.*]] = icmp slt i32 0, [[TMP5]]
8382 // CHECK15-NEXT: br i1 [[CMP]], label [[OMP_PRECOND_THEN:%.*]], label [[OMP_PRECOND_END:%.*]]
8383 // CHECK15: omp.precond.then:
8384 // CHECK15-NEXT: store i32 0, ptr [[DOTOMP_LB]], align 4
8385 // CHECK15-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_2]], align 4
8386 // CHECK15-NEXT: store i32 [[TMP6]], ptr [[DOTOMP_UB]], align 4
8387 // CHECK15-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTPREVIOUS_LB__ADDR]], align 4
8388 // CHECK15-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTPREVIOUS_UB__ADDR]], align 4
8389 // CHECK15-NEXT: store i32 [[TMP7]], ptr [[DOTOMP_LB]], align 4
8390 // CHECK15-NEXT: store i32 [[TMP8]], ptr [[DOTOMP_UB]], align 4
8391 // CHECK15-NEXT: store i32 1, ptr [[DOTOMP_STRIDE]], align 4
8392 // CHECK15-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
8393 // CHECK15-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR__ADDR]], align 4
8394 // CHECK15-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4
8395 // CHECK15-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
8396 // CHECK15-NEXT: [[TMP12:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
8397 // CHECK15-NEXT: [[TMP13:%.*]] = load i32, ptr [[TMP12]], align 4
8398 // CHECK15-NEXT: call void @__kmpc_dispatch_init_4(ptr @[[GLOB3]], i32 [[TMP13]], i32 1073741859, i32 [[TMP10]], i32 [[TMP11]], i32 1, i32 [[TMP9]])
8399 // CHECK15-NEXT: br label [[OMP_DISPATCH_COND:%.*]]
8400 // CHECK15: omp.dispatch.cond:
8401 // CHECK15-NEXT: [[TMP14:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
8402 // CHECK15-NEXT: [[TMP15:%.*]] = load i32, ptr [[TMP14]], align 4
8403 // CHECK15-NEXT: [[TMP16:%.*]] = call i32 @__kmpc_dispatch_next_4(ptr @[[GLOB3]], i32 [[TMP15]], ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]])
8404 // CHECK15-NEXT: [[TOBOOL:%.*]] = icmp ne i32 [[TMP16]], 0
8405 // CHECK15-NEXT: br i1 [[TOBOOL]], label [[OMP_DISPATCH_BODY:%.*]], label [[OMP_DISPATCH_END:%.*]]
8406 // CHECK15: omp.dispatch.body:
8407 // CHECK15-NEXT: [[TMP17:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4
8408 // CHECK15-NEXT: store i32 [[TMP17]], ptr [[DOTOMP_IV]], align 4
8409 // CHECK15-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
8410 // CHECK15: omp.inner.for.cond:
8411 // CHECK15-NEXT: [[TMP18:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP19:![0-9]+]]
8412 // CHECK15-NEXT: [[TMP19:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP19]]
8413 // CHECK15-NEXT: [[CMP5:%.*]] = icmp sle i32 [[TMP18]], [[TMP19]]
8414 // CHECK15-NEXT: br i1 [[CMP5]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
8415 // CHECK15: omp.inner.for.body:
8416 // CHECK15-NEXT: [[TMP20:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP19]]
8417 // CHECK15-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP20]], 1
8418 // CHECK15-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
8419 // CHECK15-NEXT: store i32 [[ADD]], ptr [[I4]], align 4, !llvm.access.group [[ACC_GRP19]]
8420 // CHECK15-NEXT: [[TMP21:%.*]] = load i32, ptr [[I4]], align 4, !llvm.access.group [[ACC_GRP19]]
8421 // CHECK15-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i32, ptr [[TMP2]], i32 [[TMP21]]
8422 // CHECK15-NEXT: store i32 0, ptr [[ARRAYIDX]], align 4, !llvm.access.group [[ACC_GRP19]]
8423 // CHECK15-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
8424 // CHECK15: omp.body.continue:
8425 // CHECK15-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
8426 // CHECK15: omp.inner.for.inc:
8427 // CHECK15-NEXT: [[TMP22:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP19]]
8428 // CHECK15-NEXT: [[ADD6:%.*]] = add nsw i32 [[TMP22]], 1
8429 // CHECK15-NEXT: store i32 [[ADD6]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP19]]
8430 // CHECK15-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP20:![0-9]+]]
8431 // CHECK15: omp.inner.for.end:
8432 // CHECK15-NEXT: br label [[OMP_DISPATCH_INC:%.*]]
8433 // CHECK15: omp.dispatch.inc:
8434 // CHECK15-NEXT: br label [[OMP_DISPATCH_COND]]
8435 // CHECK15: omp.dispatch.end:
8436 // CHECK15-NEXT: [[TMP23:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
8437 // CHECK15-NEXT: [[TMP24:%.*]] = load i32, ptr [[TMP23]], align 4
8438 // CHECK15-NEXT: call void @__kmpc_dispatch_deinit(ptr @[[GLOB3]], i32 [[TMP24]])
8439 // CHECK15-NEXT: br label [[OMP_PRECOND_END]]
8440 // CHECK15: omp.precond.end:
8441 // CHECK15-NEXT: ret void
8444 // CHECK15-LABEL: define {{[^@]+}}@_Z5tmainIiLi10EEiT_
8445 // CHECK15-SAME: (i32 noundef [[ARGC:%.*]]) #[[ATTR5:[0-9]+]] comdat {
8446 // CHECK15-NEXT: entry:
8447 // CHECK15-NEXT: [[ARGC_ADDR:%.*]] = alloca i32, align 4
8448 // CHECK15-NEXT: [[A:%.*]] = alloca [10 x i32], align 4
8449 // CHECK15-NEXT: [[M:%.*]] = alloca i32, align 4
8450 // CHECK15-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [1 x ptr], align 4
8451 // CHECK15-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [1 x ptr], align 4
8452 // CHECK15-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [1 x ptr], align 4
8453 // CHECK15-NEXT: [[TMP:%.*]] = alloca i32, align 4
8454 // CHECK15-NEXT: [[KERNEL_ARGS:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS:%.*]], align 8
8455 // CHECK15-NEXT: [[DOTOFFLOAD_BASEPTRS1:%.*]] = alloca [1 x ptr], align 4
8456 // CHECK15-NEXT: [[DOTOFFLOAD_PTRS2:%.*]] = alloca [1 x ptr], align 4
8457 // CHECK15-NEXT: [[DOTOFFLOAD_MAPPERS3:%.*]] = alloca [1 x ptr], align 4
8458 // CHECK15-NEXT: [[_TMP4:%.*]] = alloca i32, align 4
8459 // CHECK15-NEXT: [[KERNEL_ARGS5:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS]], align 8
8460 // CHECK15-NEXT: [[M_CASTED:%.*]] = alloca i32, align 4
8461 // CHECK15-NEXT: [[DOTOFFLOAD_BASEPTRS8:%.*]] = alloca [2 x ptr], align 4
8462 // CHECK15-NEXT: [[DOTOFFLOAD_PTRS9:%.*]] = alloca [2 x ptr], align 4
8463 // CHECK15-NEXT: [[DOTOFFLOAD_MAPPERS10:%.*]] = alloca [2 x ptr], align 4
8464 // CHECK15-NEXT: [[_TMP11:%.*]] = alloca i32, align 4
8465 // CHECK15-NEXT: [[KERNEL_ARGS12:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS]], align 8
8466 // CHECK15-NEXT: [[DOTOFFLOAD_BASEPTRS15:%.*]] = alloca [1 x ptr], align 4
8467 // CHECK15-NEXT: [[DOTOFFLOAD_PTRS16:%.*]] = alloca [1 x ptr], align 4
8468 // CHECK15-NEXT: [[DOTOFFLOAD_MAPPERS17:%.*]] = alloca [1 x ptr], align 4
8469 // CHECK15-NEXT: [[_TMP18:%.*]] = alloca i32, align 4
8470 // CHECK15-NEXT: [[KERNEL_ARGS19:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS]], align 8
8471 // CHECK15-NEXT: [[M_CASTED22:%.*]] = alloca i32, align 4
8472 // CHECK15-NEXT: [[DOTOFFLOAD_BASEPTRS23:%.*]] = alloca [2 x ptr], align 4
8473 // CHECK15-NEXT: [[DOTOFFLOAD_PTRS24:%.*]] = alloca [2 x ptr], align 4
8474 // CHECK15-NEXT: [[DOTOFFLOAD_MAPPERS25:%.*]] = alloca [2 x ptr], align 4
8475 // CHECK15-NEXT: [[_TMP26:%.*]] = alloca i32, align 4
8476 // CHECK15-NEXT: [[KERNEL_ARGS27:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS]], align 8
8477 // CHECK15-NEXT: store i32 [[ARGC]], ptr [[ARGC_ADDR]], align 4
8478 // CHECK15-NEXT: store i32 10, ptr [[M]], align 4
8479 // CHECK15-NEXT: [[TMP0:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
8480 // CHECK15-NEXT: store ptr [[A]], ptr [[TMP0]], align 4
8481 // CHECK15-NEXT: [[TMP1:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
8482 // CHECK15-NEXT: store ptr [[A]], ptr [[TMP1]], align 4
8483 // CHECK15-NEXT: [[TMP2:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0
8484 // CHECK15-NEXT: store ptr null, ptr [[TMP2]], align 4
8485 // CHECK15-NEXT: [[TMP3:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
8486 // CHECK15-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
8487 // CHECK15-NEXT: [[TMP5:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 0
8488 // CHECK15-NEXT: store i32 3, ptr [[TMP5]], align 4
8489 // CHECK15-NEXT: [[TMP6:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 1
8490 // CHECK15-NEXT: store i32 1, ptr [[TMP6]], align 4
8491 // CHECK15-NEXT: [[TMP7:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 2
8492 // CHECK15-NEXT: store ptr [[TMP3]], ptr [[TMP7]], align 4
8493 // CHECK15-NEXT: [[TMP8:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 3
8494 // CHECK15-NEXT: store ptr [[TMP4]], ptr [[TMP8]], align 4
8495 // CHECK15-NEXT: [[TMP9:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 4
8496 // CHECK15-NEXT: store ptr @.offload_sizes.9, ptr [[TMP9]], align 4
8497 // CHECK15-NEXT: [[TMP10:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 5
8498 // CHECK15-NEXT: store ptr @.offload_maptypes.10, ptr [[TMP10]], align 4
8499 // CHECK15-NEXT: [[TMP11:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 6
8500 // CHECK15-NEXT: store ptr null, ptr [[TMP11]], align 4
8501 // CHECK15-NEXT: [[TMP12:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 7
8502 // CHECK15-NEXT: store ptr null, ptr [[TMP12]], align 4
8503 // CHECK15-NEXT: [[TMP13:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 8
8504 // CHECK15-NEXT: store i64 10, ptr [[TMP13]], align 8
8505 // CHECK15-NEXT: [[TMP14:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 9
8506 // CHECK15-NEXT: store i64 0, ptr [[TMP14]], align 8
8507 // CHECK15-NEXT: [[TMP15:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 10
8508 // CHECK15-NEXT: store [3 x i32] zeroinitializer, ptr [[TMP15]], align 4
8509 // CHECK15-NEXT: [[TMP16:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 11
8510 // CHECK15-NEXT: store [3 x i32] zeroinitializer, ptr [[TMP16]], align 4
8511 // CHECK15-NEXT: [[TMP17:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 12
8512 // CHECK15-NEXT: store i32 0, ptr [[TMP17]], align 4
8513 // CHECK15-NEXT: [[TMP18:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB3]], i64 -1, i32 0, i32 0, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l116.region_id, ptr [[KERNEL_ARGS]])
8514 // CHECK15-NEXT: [[TMP19:%.*]] = icmp ne i32 [[TMP18]], 0
8515 // CHECK15-NEXT: br i1 [[TMP19]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
8516 // CHECK15: omp_offload.failed:
8517 // CHECK15-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l116(ptr [[A]]) #[[ATTR3]]
8518 // CHECK15-NEXT: br label [[OMP_OFFLOAD_CONT]]
8519 // CHECK15: omp_offload.cont:
8520 // CHECK15-NEXT: [[TMP20:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS1]], i32 0, i32 0
8521 // CHECK15-NEXT: store ptr [[A]], ptr [[TMP20]], align 4
8522 // CHECK15-NEXT: [[TMP21:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS2]], i32 0, i32 0
8523 // CHECK15-NEXT: store ptr [[A]], ptr [[TMP21]], align 4
8524 // CHECK15-NEXT: [[TMP22:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_MAPPERS3]], i32 0, i32 0
8525 // CHECK15-NEXT: store ptr null, ptr [[TMP22]], align 4
8526 // CHECK15-NEXT: [[TMP23:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS1]], i32 0, i32 0
8527 // CHECK15-NEXT: [[TMP24:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS2]], i32 0, i32 0
8528 // CHECK15-NEXT: [[TMP25:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS5]], i32 0, i32 0
8529 // CHECK15-NEXT: store i32 3, ptr [[TMP25]], align 4
8530 // CHECK15-NEXT: [[TMP26:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS5]], i32 0, i32 1
8531 // CHECK15-NEXT: store i32 1, ptr [[TMP26]], align 4
8532 // CHECK15-NEXT: [[TMP27:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS5]], i32 0, i32 2
8533 // CHECK15-NEXT: store ptr [[TMP23]], ptr [[TMP27]], align 4
8534 // CHECK15-NEXT: [[TMP28:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS5]], i32 0, i32 3
8535 // CHECK15-NEXT: store ptr [[TMP24]], ptr [[TMP28]], align 4
8536 // CHECK15-NEXT: [[TMP29:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS5]], i32 0, i32 4
8537 // CHECK15-NEXT: store ptr @.offload_sizes.11, ptr [[TMP29]], align 4
8538 // CHECK15-NEXT: [[TMP30:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS5]], i32 0, i32 5
8539 // CHECK15-NEXT: store ptr @.offload_maptypes.12, ptr [[TMP30]], align 4
8540 // CHECK15-NEXT: [[TMP31:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS5]], i32 0, i32 6
8541 // CHECK15-NEXT: store ptr null, ptr [[TMP31]], align 4
8542 // CHECK15-NEXT: [[TMP32:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS5]], i32 0, i32 7
8543 // CHECK15-NEXT: store ptr null, ptr [[TMP32]], align 4
8544 // CHECK15-NEXT: [[TMP33:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS5]], i32 0, i32 8
8545 // CHECK15-NEXT: store i64 10, ptr [[TMP33]], align 8
8546 // CHECK15-NEXT: [[TMP34:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS5]], i32 0, i32 9
8547 // CHECK15-NEXT: store i64 0, ptr [[TMP34]], align 8
8548 // CHECK15-NEXT: [[TMP35:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS5]], i32 0, i32 10
8549 // CHECK15-NEXT: store [3 x i32] zeroinitializer, ptr [[TMP35]], align 4
8550 // CHECK15-NEXT: [[TMP36:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS5]], i32 0, i32 11
8551 // CHECK15-NEXT: store [3 x i32] zeroinitializer, ptr [[TMP36]], align 4
8552 // CHECK15-NEXT: [[TMP37:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS5]], i32 0, i32 12
8553 // CHECK15-NEXT: store i32 0, ptr [[TMP37]], align 4
8554 // CHECK15-NEXT: [[TMP38:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB3]], i64 -1, i32 0, i32 0, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l121.region_id, ptr [[KERNEL_ARGS5]])
8555 // CHECK15-NEXT: [[TMP39:%.*]] = icmp ne i32 [[TMP38]], 0
8556 // CHECK15-NEXT: br i1 [[TMP39]], label [[OMP_OFFLOAD_FAILED6:%.*]], label [[OMP_OFFLOAD_CONT7:%.*]]
8557 // CHECK15: omp_offload.failed6:
8558 // CHECK15-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l121(ptr [[A]]) #[[ATTR3]]
8559 // CHECK15-NEXT: br label [[OMP_OFFLOAD_CONT7]]
8560 // CHECK15: omp_offload.cont7:
8561 // CHECK15-NEXT: [[TMP40:%.*]] = load i32, ptr [[M]], align 4
8562 // CHECK15-NEXT: store i32 [[TMP40]], ptr [[M_CASTED]], align 4
8563 // CHECK15-NEXT: [[TMP41:%.*]] = load i32, ptr [[M_CASTED]], align 4
8564 // CHECK15-NEXT: [[TMP42:%.*]] = getelementptr inbounds [2 x ptr], ptr [[DOTOFFLOAD_BASEPTRS8]], i32 0, i32 0
8565 // CHECK15-NEXT: store i32 [[TMP41]], ptr [[TMP42]], align 4
8566 // CHECK15-NEXT: [[TMP43:%.*]] = getelementptr inbounds [2 x ptr], ptr [[DOTOFFLOAD_PTRS9]], i32 0, i32 0
8567 // CHECK15-NEXT: store i32 [[TMP41]], ptr [[TMP43]], align 4
8568 // CHECK15-NEXT: [[TMP44:%.*]] = getelementptr inbounds [2 x ptr], ptr [[DOTOFFLOAD_MAPPERS10]], i32 0, i32 0
8569 // CHECK15-NEXT: store ptr null, ptr [[TMP44]], align 4
8570 // CHECK15-NEXT: [[TMP45:%.*]] = getelementptr inbounds [2 x ptr], ptr [[DOTOFFLOAD_BASEPTRS8]], i32 0, i32 1
8571 // CHECK15-NEXT: store ptr [[A]], ptr [[TMP45]], align 4
8572 // CHECK15-NEXT: [[TMP46:%.*]] = getelementptr inbounds [2 x ptr], ptr [[DOTOFFLOAD_PTRS9]], i32 0, i32 1
8573 // CHECK15-NEXT: store ptr [[A]], ptr [[TMP46]], align 4
8574 // CHECK15-NEXT: [[TMP47:%.*]] = getelementptr inbounds [2 x ptr], ptr [[DOTOFFLOAD_MAPPERS10]], i32 0, i32 1
8575 // CHECK15-NEXT: store ptr null, ptr [[TMP47]], align 4
8576 // CHECK15-NEXT: [[TMP48:%.*]] = getelementptr inbounds [2 x ptr], ptr [[DOTOFFLOAD_BASEPTRS8]], i32 0, i32 0
8577 // CHECK15-NEXT: [[TMP49:%.*]] = getelementptr inbounds [2 x ptr], ptr [[DOTOFFLOAD_PTRS9]], i32 0, i32 0
8578 // CHECK15-NEXT: [[TMP50:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS12]], i32 0, i32 0
8579 // CHECK15-NEXT: store i32 3, ptr [[TMP50]], align 4
8580 // CHECK15-NEXT: [[TMP51:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS12]], i32 0, i32 1
8581 // CHECK15-NEXT: store i32 2, ptr [[TMP51]], align 4
8582 // CHECK15-NEXT: [[TMP52:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS12]], i32 0, i32 2
8583 // CHECK15-NEXT: store ptr [[TMP48]], ptr [[TMP52]], align 4
8584 // CHECK15-NEXT: [[TMP53:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS12]], i32 0, i32 3
8585 // CHECK15-NEXT: store ptr [[TMP49]], ptr [[TMP53]], align 4
8586 // CHECK15-NEXT: [[TMP54:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS12]], i32 0, i32 4
8587 // CHECK15-NEXT: store ptr @.offload_sizes.13, ptr [[TMP54]], align 4
8588 // CHECK15-NEXT: [[TMP55:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS12]], i32 0, i32 5
8589 // CHECK15-NEXT: store ptr @.offload_maptypes.14, ptr [[TMP55]], align 4
8590 // CHECK15-NEXT: [[TMP56:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS12]], i32 0, i32 6
8591 // CHECK15-NEXT: store ptr null, ptr [[TMP56]], align 4
8592 // CHECK15-NEXT: [[TMP57:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS12]], i32 0, i32 7
8593 // CHECK15-NEXT: store ptr null, ptr [[TMP57]], align 4
8594 // CHECK15-NEXT: [[TMP58:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS12]], i32 0, i32 8
8595 // CHECK15-NEXT: store i64 10, ptr [[TMP58]], align 8
8596 // CHECK15-NEXT: [[TMP59:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS12]], i32 0, i32 9
8597 // CHECK15-NEXT: store i64 0, ptr [[TMP59]], align 8
8598 // CHECK15-NEXT: [[TMP60:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS12]], i32 0, i32 10
8599 // CHECK15-NEXT: store [3 x i32] zeroinitializer, ptr [[TMP60]], align 4
8600 // CHECK15-NEXT: [[TMP61:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS12]], i32 0, i32 11
8601 // CHECK15-NEXT: store [3 x i32] zeroinitializer, ptr [[TMP61]], align 4
8602 // CHECK15-NEXT: [[TMP62:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS12]], i32 0, i32 12
8603 // CHECK15-NEXT: store i32 0, ptr [[TMP62]], align 4
8604 // CHECK15-NEXT: [[TMP63:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB3]], i64 -1, i32 0, i32 0, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l126.region_id, ptr [[KERNEL_ARGS12]])
8605 // CHECK15-NEXT: [[TMP64:%.*]] = icmp ne i32 [[TMP63]], 0
8606 // CHECK15-NEXT: br i1 [[TMP64]], label [[OMP_OFFLOAD_FAILED13:%.*]], label [[OMP_OFFLOAD_CONT14:%.*]]
8607 // CHECK15: omp_offload.failed13:
8608 // CHECK15-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l126(i32 [[TMP41]], ptr [[A]]) #[[ATTR3]]
8609 // CHECK15-NEXT: br label [[OMP_OFFLOAD_CONT14]]
8610 // CHECK15: omp_offload.cont14:
8611 // CHECK15-NEXT: [[TMP65:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS15]], i32 0, i32 0
8612 // CHECK15-NEXT: store ptr [[A]], ptr [[TMP65]], align 4
8613 // CHECK15-NEXT: [[TMP66:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS16]], i32 0, i32 0
8614 // CHECK15-NEXT: store ptr [[A]], ptr [[TMP66]], align 4
8615 // CHECK15-NEXT: [[TMP67:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_MAPPERS17]], i32 0, i32 0
8616 // CHECK15-NEXT: store ptr null, ptr [[TMP67]], align 4
8617 // CHECK15-NEXT: [[TMP68:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS15]], i32 0, i32 0
8618 // CHECK15-NEXT: [[TMP69:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS16]], i32 0, i32 0
8619 // CHECK15-NEXT: [[TMP70:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS19]], i32 0, i32 0
8620 // CHECK15-NEXT: store i32 3, ptr [[TMP70]], align 4
8621 // CHECK15-NEXT: [[TMP71:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS19]], i32 0, i32 1
8622 // CHECK15-NEXT: store i32 1, ptr [[TMP71]], align 4
8623 // CHECK15-NEXT: [[TMP72:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS19]], i32 0, i32 2
8624 // CHECK15-NEXT: store ptr [[TMP68]], ptr [[TMP72]], align 4
8625 // CHECK15-NEXT: [[TMP73:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS19]], i32 0, i32 3
8626 // CHECK15-NEXT: store ptr [[TMP69]], ptr [[TMP73]], align 4
8627 // CHECK15-NEXT: [[TMP74:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS19]], i32 0, i32 4
8628 // CHECK15-NEXT: store ptr @.offload_sizes.15, ptr [[TMP74]], align 4
8629 // CHECK15-NEXT: [[TMP75:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS19]], i32 0, i32 5
8630 // CHECK15-NEXT: store ptr @.offload_maptypes.16, ptr [[TMP75]], align 4
8631 // CHECK15-NEXT: [[TMP76:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS19]], i32 0, i32 6
8632 // CHECK15-NEXT: store ptr null, ptr [[TMP76]], align 4
8633 // CHECK15-NEXT: [[TMP77:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS19]], i32 0, i32 7
8634 // CHECK15-NEXT: store ptr null, ptr [[TMP77]], align 4
8635 // CHECK15-NEXT: [[TMP78:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS19]], i32 0, i32 8
8636 // CHECK15-NEXT: store i64 10, ptr [[TMP78]], align 8
8637 // CHECK15-NEXT: [[TMP79:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS19]], i32 0, i32 9
8638 // CHECK15-NEXT: store i64 0, ptr [[TMP79]], align 8
8639 // CHECK15-NEXT: [[TMP80:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS19]], i32 0, i32 10
8640 // CHECK15-NEXT: store [3 x i32] zeroinitializer, ptr [[TMP80]], align 4
8641 // CHECK15-NEXT: [[TMP81:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS19]], i32 0, i32 11
8642 // CHECK15-NEXT: store [3 x i32] zeroinitializer, ptr [[TMP81]], align 4
8643 // CHECK15-NEXT: [[TMP82:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS19]], i32 0, i32 12
8644 // CHECK15-NEXT: store i32 0, ptr [[TMP82]], align 4
8645 // CHECK15-NEXT: [[TMP83:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB3]], i64 -1, i32 0, i32 0, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l131.region_id, ptr [[KERNEL_ARGS19]])
8646 // CHECK15-NEXT: [[TMP84:%.*]] = icmp ne i32 [[TMP83]], 0
8647 // CHECK15-NEXT: br i1 [[TMP84]], label [[OMP_OFFLOAD_FAILED20:%.*]], label [[OMP_OFFLOAD_CONT21:%.*]]
8648 // CHECK15: omp_offload.failed20:
8649 // CHECK15-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l131(ptr [[A]]) #[[ATTR3]]
8650 // CHECK15-NEXT: br label [[OMP_OFFLOAD_CONT21]]
8651 // CHECK15: omp_offload.cont21:
8652 // CHECK15-NEXT: [[TMP85:%.*]] = load i32, ptr [[M]], align 4
8653 // CHECK15-NEXT: store i32 [[TMP85]], ptr [[M_CASTED22]], align 4
8654 // CHECK15-NEXT: [[TMP86:%.*]] = load i32, ptr [[M_CASTED22]], align 4
8655 // CHECK15-NEXT: [[TMP87:%.*]] = getelementptr inbounds [2 x ptr], ptr [[DOTOFFLOAD_BASEPTRS23]], i32 0, i32 0
8656 // CHECK15-NEXT: store i32 [[TMP86]], ptr [[TMP87]], align 4
8657 // CHECK15-NEXT: [[TMP88:%.*]] = getelementptr inbounds [2 x ptr], ptr [[DOTOFFLOAD_PTRS24]], i32 0, i32 0
8658 // CHECK15-NEXT: store i32 [[TMP86]], ptr [[TMP88]], align 4
8659 // CHECK15-NEXT: [[TMP89:%.*]] = getelementptr inbounds [2 x ptr], ptr [[DOTOFFLOAD_MAPPERS25]], i32 0, i32 0
8660 // CHECK15-NEXT: store ptr null, ptr [[TMP89]], align 4
8661 // CHECK15-NEXT: [[TMP90:%.*]] = getelementptr inbounds [2 x ptr], ptr [[DOTOFFLOAD_BASEPTRS23]], i32 0, i32 1
8662 // CHECK15-NEXT: store ptr [[A]], ptr [[TMP90]], align 4
8663 // CHECK15-NEXT: [[TMP91:%.*]] = getelementptr inbounds [2 x ptr], ptr [[DOTOFFLOAD_PTRS24]], i32 0, i32 1
8664 // CHECK15-NEXT: store ptr [[A]], ptr [[TMP91]], align 4
8665 // CHECK15-NEXT: [[TMP92:%.*]] = getelementptr inbounds [2 x ptr], ptr [[DOTOFFLOAD_MAPPERS25]], i32 0, i32 1
8666 // CHECK15-NEXT: store ptr null, ptr [[TMP92]], align 4
8667 // CHECK15-NEXT: [[TMP93:%.*]] = getelementptr inbounds [2 x ptr], ptr [[DOTOFFLOAD_BASEPTRS23]], i32 0, i32 0
8668 // CHECK15-NEXT: [[TMP94:%.*]] = getelementptr inbounds [2 x ptr], ptr [[DOTOFFLOAD_PTRS24]], i32 0, i32 0
8669 // CHECK15-NEXT: [[TMP95:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS27]], i32 0, i32 0
8670 // CHECK15-NEXT: store i32 3, ptr [[TMP95]], align 4
8671 // CHECK15-NEXT: [[TMP96:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS27]], i32 0, i32 1
8672 // CHECK15-NEXT: store i32 2, ptr [[TMP96]], align 4
8673 // CHECK15-NEXT: [[TMP97:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS27]], i32 0, i32 2
8674 // CHECK15-NEXT: store ptr [[TMP93]], ptr [[TMP97]], align 4
8675 // CHECK15-NEXT: [[TMP98:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS27]], i32 0, i32 3
8676 // CHECK15-NEXT: store ptr [[TMP94]], ptr [[TMP98]], align 4
8677 // CHECK15-NEXT: [[TMP99:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS27]], i32 0, i32 4
8678 // CHECK15-NEXT: store ptr @.offload_sizes.17, ptr [[TMP99]], align 4
8679 // CHECK15-NEXT: [[TMP100:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS27]], i32 0, i32 5
8680 // CHECK15-NEXT: store ptr @.offload_maptypes.18, ptr [[TMP100]], align 4
8681 // CHECK15-NEXT: [[TMP101:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS27]], i32 0, i32 6
8682 // CHECK15-NEXT: store ptr null, ptr [[TMP101]], align 4
8683 // CHECK15-NEXT: [[TMP102:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS27]], i32 0, i32 7
8684 // CHECK15-NEXT: store ptr null, ptr [[TMP102]], align 4
8685 // CHECK15-NEXT: [[TMP103:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS27]], i32 0, i32 8
8686 // CHECK15-NEXT: store i64 10, ptr [[TMP103]], align 8
8687 // CHECK15-NEXT: [[TMP104:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS27]], i32 0, i32 9
8688 // CHECK15-NEXT: store i64 0, ptr [[TMP104]], align 8
8689 // CHECK15-NEXT: [[TMP105:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS27]], i32 0, i32 10
8690 // CHECK15-NEXT: store [3 x i32] zeroinitializer, ptr [[TMP105]], align 4
8691 // CHECK15-NEXT: [[TMP106:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS27]], i32 0, i32 11
8692 // CHECK15-NEXT: store [3 x i32] zeroinitializer, ptr [[TMP106]], align 4
8693 // CHECK15-NEXT: [[TMP107:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS27]], i32 0, i32 12
8694 // CHECK15-NEXT: store i32 0, ptr [[TMP107]], align 4
8695 // CHECK15-NEXT: [[TMP108:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB3]], i64 -1, i32 0, i32 0, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l136.region_id, ptr [[KERNEL_ARGS27]])
8696 // CHECK15-NEXT: [[TMP109:%.*]] = icmp ne i32 [[TMP108]], 0
8697 // CHECK15-NEXT: br i1 [[TMP109]], label [[OMP_OFFLOAD_FAILED28:%.*]], label [[OMP_OFFLOAD_CONT29:%.*]]
8698 // CHECK15: omp_offload.failed28:
8699 // CHECK15-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l136(i32 [[TMP86]], ptr [[A]]) #[[ATTR3]]
8700 // CHECK15-NEXT: br label [[OMP_OFFLOAD_CONT29]]
8701 // CHECK15: omp_offload.cont29:
8702 // CHECK15-NEXT: ret i32 0
8705 // CHECK15-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l116
8706 // CHECK15-SAME: (ptr noundef nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] {
8707 // CHECK15-NEXT: entry:
8708 // CHECK15-NEXT: [[A_ADDR:%.*]] = alloca ptr, align 4
8709 // CHECK15-NEXT: store ptr [[A]], ptr [[A_ADDR]], align 4
8710 // CHECK15-NEXT: [[TMP0:%.*]] = load ptr, ptr [[A_ADDR]], align 4
8711 // CHECK15-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_teams(ptr @[[GLOB3]], i32 1, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l116.omp_outlined, ptr [[TMP0]])
8712 // CHECK15-NEXT: ret void
8715 // CHECK15-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l116.omp_outlined
8716 // CHECK15-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], ptr noundef nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] {
8717 // CHECK15-NEXT: entry:
8718 // CHECK15-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 4
8719 // CHECK15-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 4
8720 // CHECK15-NEXT: [[A_ADDR:%.*]] = alloca ptr, align 4
8721 // CHECK15-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
8722 // CHECK15-NEXT: [[TMP:%.*]] = alloca i32, align 4
8723 // CHECK15-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4
8724 // CHECK15-NEXT: [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4
8725 // CHECK15-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
8726 // CHECK15-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
8727 // CHECK15-NEXT: [[I:%.*]] = alloca i32, align 4
8728 // CHECK15-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 4
8729 // CHECK15-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 4
8730 // CHECK15-NEXT: store ptr [[A]], ptr [[A_ADDR]], align 4
8731 // CHECK15-NEXT: [[TMP0:%.*]] = load ptr, ptr [[A_ADDR]], align 4
8732 // CHECK15-NEXT: store i32 0, ptr [[DOTOMP_COMB_LB]], align 4
8733 // CHECK15-NEXT: store i32 9, ptr [[DOTOMP_COMB_UB]], align 4
8734 // CHECK15-NEXT: store i32 1, ptr [[DOTOMP_STRIDE]], align 4
8735 // CHECK15-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
8736 // CHECK15-NEXT: [[TMP1:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
8737 // CHECK15-NEXT: [[TMP2:%.*]] = load i32, ptr [[TMP1]], align 4
8738 // CHECK15-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB1]], i32 [[TMP2]], i32 92, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
8739 // CHECK15-NEXT: [[TMP3:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
8740 // CHECK15-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP3]], 9
8741 // CHECK15-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
8742 // CHECK15: cond.true:
8743 // CHECK15-NEXT: br label [[COND_END:%.*]]
8744 // CHECK15: cond.false:
8745 // CHECK15-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
8746 // CHECK15-NEXT: br label [[COND_END]]
8747 // CHECK15: cond.end:
8748 // CHECK15-NEXT: [[COND:%.*]] = phi i32 [ 9, [[COND_TRUE]] ], [ [[TMP4]], [[COND_FALSE]] ]
8749 // CHECK15-NEXT: store i32 [[COND]], ptr [[DOTOMP_COMB_UB]], align 4
8750 // CHECK15-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
8751 // CHECK15-NEXT: store i32 [[TMP5]], ptr [[DOTOMP_IV]], align 4
8752 // CHECK15-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
8753 // CHECK15: omp.inner.for.cond:
8754 // CHECK15-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
8755 // CHECK15-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
8756 // CHECK15-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP6]], [[TMP7]]
8757 // CHECK15-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
8758 // CHECK15: omp.inner.for.body:
8759 // CHECK15-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
8760 // CHECK15-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
8761 // CHECK15-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_call(ptr @[[GLOB3]], i32 3, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l116.omp_outlined.omp_outlined, i32 [[TMP8]], i32 [[TMP9]], ptr [[TMP0]])
8762 // CHECK15-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
8763 // CHECK15: omp.inner.for.inc:
8764 // CHECK15-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
8765 // CHECK15-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4
8766 // CHECK15-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP10]], [[TMP11]]
8767 // CHECK15-NEXT: store i32 [[ADD]], ptr [[DOTOMP_IV]], align 4
8768 // CHECK15-NEXT: br label [[OMP_INNER_FOR_COND]]
8769 // CHECK15: omp.inner.for.end:
8770 // CHECK15-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
8771 // CHECK15: omp.loop.exit:
8772 // CHECK15-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB1]], i32 [[TMP2]])
8773 // CHECK15-NEXT: ret void
8776 // CHECK15-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l116.omp_outlined.omp_outlined
8777 // CHECK15-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], i32 noundef [[DOTPREVIOUS_LB_:%.*]], i32 noundef [[DOTPREVIOUS_UB_:%.*]], ptr noundef nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] {
8778 // CHECK15-NEXT: entry:
8779 // CHECK15-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 4
8780 // CHECK15-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 4
8781 // CHECK15-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i32, align 4
8782 // CHECK15-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i32, align 4
8783 // CHECK15-NEXT: [[A_ADDR:%.*]] = alloca ptr, align 4
8784 // CHECK15-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
8785 // CHECK15-NEXT: [[TMP:%.*]] = alloca i32, align 4
8786 // CHECK15-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4
8787 // CHECK15-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4
8788 // CHECK15-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
8789 // CHECK15-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
8790 // CHECK15-NEXT: [[I:%.*]] = alloca i32, align 4
8791 // CHECK15-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 4
8792 // CHECK15-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 4
8793 // CHECK15-NEXT: store i32 [[DOTPREVIOUS_LB_]], ptr [[DOTPREVIOUS_LB__ADDR]], align 4
8794 // CHECK15-NEXT: store i32 [[DOTPREVIOUS_UB_]], ptr [[DOTPREVIOUS_UB__ADDR]], align 4
8795 // CHECK15-NEXT: store ptr [[A]], ptr [[A_ADDR]], align 4
8796 // CHECK15-NEXT: [[TMP0:%.*]] = load ptr, ptr [[A_ADDR]], align 4
8797 // CHECK15-NEXT: store i32 0, ptr [[DOTOMP_LB]], align 4
8798 // CHECK15-NEXT: store i32 9, ptr [[DOTOMP_UB]], align 4
8799 // CHECK15-NEXT: [[TMP1:%.*]] = load i32, ptr [[DOTPREVIOUS_LB__ADDR]], align 4
8800 // CHECK15-NEXT: [[TMP2:%.*]] = load i32, ptr [[DOTPREVIOUS_UB__ADDR]], align 4
8801 // CHECK15-NEXT: store i32 [[TMP1]], ptr [[DOTOMP_LB]], align 4
8802 // CHECK15-NEXT: store i32 [[TMP2]], ptr [[DOTOMP_UB]], align 4
8803 // CHECK15-NEXT: store i32 1, ptr [[DOTOMP_STRIDE]], align 4
8804 // CHECK15-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
8805 // CHECK15-NEXT: [[TMP3:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
8806 // CHECK15-NEXT: [[TMP4:%.*]] = load i32, ptr [[TMP3]], align 4
8807 // CHECK15-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB2]], i32 [[TMP4]], i32 34, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
8808 // CHECK15-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
8809 // CHECK15-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP5]], 9
8810 // CHECK15-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
8811 // CHECK15: cond.true:
8812 // CHECK15-NEXT: br label [[COND_END:%.*]]
8813 // CHECK15: cond.false:
8814 // CHECK15-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
8815 // CHECK15-NEXT: br label [[COND_END]]
8816 // CHECK15: cond.end:
8817 // CHECK15-NEXT: [[COND:%.*]] = phi i32 [ 9, [[COND_TRUE]] ], [ [[TMP6]], [[COND_FALSE]] ]
8818 // CHECK15-NEXT: store i32 [[COND]], ptr [[DOTOMP_UB]], align 4
8819 // CHECK15-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4
8820 // CHECK15-NEXT: store i32 [[TMP7]], ptr [[DOTOMP_IV]], align 4
8821 // CHECK15-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
8822 // CHECK15: omp.inner.for.cond:
8823 // CHECK15-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
8824 // CHECK15-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
8825 // CHECK15-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP8]], [[TMP9]]
8826 // CHECK15-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
8827 // CHECK15: omp.inner.for.body:
8828 // CHECK15-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
8829 // CHECK15-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP10]], 1
8830 // CHECK15-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
8831 // CHECK15-NEXT: store i32 [[ADD]], ptr [[I]], align 4
8832 // CHECK15-NEXT: [[TMP11:%.*]] = load i32, ptr [[I]], align 4
8833 // CHECK15-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds [10 x i32], ptr [[TMP0]], i32 0, i32 [[TMP11]]
8834 // CHECK15-NEXT: store i32 0, ptr [[ARRAYIDX]], align 4
8835 // CHECK15-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
8836 // CHECK15: omp.body.continue:
8837 // CHECK15-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
8838 // CHECK15: omp.inner.for.inc:
8839 // CHECK15-NEXT: [[TMP12:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
8840 // CHECK15-NEXT: [[ADD2:%.*]] = add nsw i32 [[TMP12]], 1
8841 // CHECK15-NEXT: store i32 [[ADD2]], ptr [[DOTOMP_IV]], align 4
8842 // CHECK15-NEXT: br label [[OMP_INNER_FOR_COND]]
8843 // CHECK15: omp.inner.for.end:
8844 // CHECK15-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
8845 // CHECK15: omp.loop.exit:
8846 // CHECK15-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB2]], i32 [[TMP4]])
8847 // CHECK15-NEXT: ret void
8850 // CHECK15-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l121
8851 // CHECK15-SAME: (ptr noundef nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] {
8852 // CHECK15-NEXT: entry:
8853 // CHECK15-NEXT: [[A_ADDR:%.*]] = alloca ptr, align 4
8854 // CHECK15-NEXT: store ptr [[A]], ptr [[A_ADDR]], align 4
8855 // CHECK15-NEXT: [[TMP0:%.*]] = load ptr, ptr [[A_ADDR]], align 4
8856 // CHECK15-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_teams(ptr @[[GLOB3]], i32 1, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l121.omp_outlined, ptr [[TMP0]])
8857 // CHECK15-NEXT: ret void
8860 // CHECK15-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l121.omp_outlined
8861 // CHECK15-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], ptr noundef nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] {
8862 // CHECK15-NEXT: entry:
8863 // CHECK15-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 4
8864 // CHECK15-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 4
8865 // CHECK15-NEXT: [[A_ADDR:%.*]] = alloca ptr, align 4
8866 // CHECK15-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
8867 // CHECK15-NEXT: [[TMP:%.*]] = alloca i32, align 4
8868 // CHECK15-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4
8869 // CHECK15-NEXT: [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4
8870 // CHECK15-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
8871 // CHECK15-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
8872 // CHECK15-NEXT: [[I:%.*]] = alloca i32, align 4
8873 // CHECK15-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 4
8874 // CHECK15-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 4
8875 // CHECK15-NEXT: store ptr [[A]], ptr [[A_ADDR]], align 4
8876 // CHECK15-NEXT: [[TMP0:%.*]] = load ptr, ptr [[A_ADDR]], align 4
8877 // CHECK15-NEXT: store i32 0, ptr [[DOTOMP_COMB_LB]], align 4
8878 // CHECK15-NEXT: store i32 9, ptr [[DOTOMP_COMB_UB]], align 4
8879 // CHECK15-NEXT: store i32 1, ptr [[DOTOMP_STRIDE]], align 4
8880 // CHECK15-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
8881 // CHECK15-NEXT: [[TMP1:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
8882 // CHECK15-NEXT: [[TMP2:%.*]] = load i32, ptr [[TMP1]], align 4
8883 // CHECK15-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB1]], i32 [[TMP2]], i32 92, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
8884 // CHECK15-NEXT: [[TMP3:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
8885 // CHECK15-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP3]], 9
8886 // CHECK15-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
8887 // CHECK15: cond.true:
8888 // CHECK15-NEXT: br label [[COND_END:%.*]]
8889 // CHECK15: cond.false:
8890 // CHECK15-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
8891 // CHECK15-NEXT: br label [[COND_END]]
8892 // CHECK15: cond.end:
8893 // CHECK15-NEXT: [[COND:%.*]] = phi i32 [ 9, [[COND_TRUE]] ], [ [[TMP4]], [[COND_FALSE]] ]
8894 // CHECK15-NEXT: store i32 [[COND]], ptr [[DOTOMP_COMB_UB]], align 4
8895 // CHECK15-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
8896 // CHECK15-NEXT: store i32 [[TMP5]], ptr [[DOTOMP_IV]], align 4
8897 // CHECK15-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
8898 // CHECK15: omp.inner.for.cond:
8899 // CHECK15-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
8900 // CHECK15-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
8901 // CHECK15-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP6]], [[TMP7]]
8902 // CHECK15-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
8903 // CHECK15: omp.inner.for.body:
8904 // CHECK15-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
8905 // CHECK15-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
8906 // CHECK15-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_call(ptr @[[GLOB3]], i32 3, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l121.omp_outlined.omp_outlined, i32 [[TMP8]], i32 [[TMP9]], ptr [[TMP0]])
8907 // CHECK15-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
8908 // CHECK15: omp.inner.for.inc:
8909 // CHECK15-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
8910 // CHECK15-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4
8911 // CHECK15-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP10]], [[TMP11]]
8912 // CHECK15-NEXT: store i32 [[ADD]], ptr [[DOTOMP_IV]], align 4
8913 // CHECK15-NEXT: br label [[OMP_INNER_FOR_COND]]
8914 // CHECK15: omp.inner.for.end:
8915 // CHECK15-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
8916 // CHECK15: omp.loop.exit:
8917 // CHECK15-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB1]], i32 [[TMP2]])
8918 // CHECK15-NEXT: ret void
8921 // CHECK15-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l121.omp_outlined.omp_outlined
8922 // CHECK15-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], i32 noundef [[DOTPREVIOUS_LB_:%.*]], i32 noundef [[DOTPREVIOUS_UB_:%.*]], ptr noundef nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] {
8923 // CHECK15-NEXT: entry:
8924 // CHECK15-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 4
8925 // CHECK15-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 4
8926 // CHECK15-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i32, align 4
8927 // CHECK15-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i32, align 4
8928 // CHECK15-NEXT: [[A_ADDR:%.*]] = alloca ptr, align 4
8929 // CHECK15-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
8930 // CHECK15-NEXT: [[TMP:%.*]] = alloca i32, align 4
8931 // CHECK15-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4
8932 // CHECK15-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4
8933 // CHECK15-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
8934 // CHECK15-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
8935 // CHECK15-NEXT: [[I:%.*]] = alloca i32, align 4
8936 // CHECK15-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 4
8937 // CHECK15-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 4
8938 // CHECK15-NEXT: store i32 [[DOTPREVIOUS_LB_]], ptr [[DOTPREVIOUS_LB__ADDR]], align 4
8939 // CHECK15-NEXT: store i32 [[DOTPREVIOUS_UB_]], ptr [[DOTPREVIOUS_UB__ADDR]], align 4
8940 // CHECK15-NEXT: store ptr [[A]], ptr [[A_ADDR]], align 4
8941 // CHECK15-NEXT: [[TMP0:%.*]] = load ptr, ptr [[A_ADDR]], align 4
8942 // CHECK15-NEXT: store i32 0, ptr [[DOTOMP_LB]], align 4
8943 // CHECK15-NEXT: store i32 9, ptr [[DOTOMP_UB]], align 4
8944 // CHECK15-NEXT: [[TMP1:%.*]] = load i32, ptr [[DOTPREVIOUS_LB__ADDR]], align 4
8945 // CHECK15-NEXT: [[TMP2:%.*]] = load i32, ptr [[DOTPREVIOUS_UB__ADDR]], align 4
8946 // CHECK15-NEXT: store i32 [[TMP1]], ptr [[DOTOMP_LB]], align 4
8947 // CHECK15-NEXT: store i32 [[TMP2]], ptr [[DOTOMP_UB]], align 4
8948 // CHECK15-NEXT: store i32 1, ptr [[DOTOMP_STRIDE]], align 4
8949 // CHECK15-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
8950 // CHECK15-NEXT: [[TMP3:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
8951 // CHECK15-NEXT: [[TMP4:%.*]] = load i32, ptr [[TMP3]], align 4
8952 // CHECK15-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB2]], i32 [[TMP4]], i32 34, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
8953 // CHECK15-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
8954 // CHECK15-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP5]], 9
8955 // CHECK15-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
8956 // CHECK15: cond.true:
8957 // CHECK15-NEXT: br label [[COND_END:%.*]]
8958 // CHECK15: cond.false:
8959 // CHECK15-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
8960 // CHECK15-NEXT: br label [[COND_END]]
8961 // CHECK15: cond.end:
8962 // CHECK15-NEXT: [[COND:%.*]] = phi i32 [ 9, [[COND_TRUE]] ], [ [[TMP6]], [[COND_FALSE]] ]
8963 // CHECK15-NEXT: store i32 [[COND]], ptr [[DOTOMP_UB]], align 4
8964 // CHECK15-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4
8965 // CHECK15-NEXT: store i32 [[TMP7]], ptr [[DOTOMP_IV]], align 4
8966 // CHECK15-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
8967 // CHECK15: omp.inner.for.cond:
8968 // CHECK15-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
8969 // CHECK15-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
8970 // CHECK15-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP8]], [[TMP9]]
8971 // CHECK15-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
8972 // CHECK15: omp.inner.for.body:
8973 // CHECK15-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
8974 // CHECK15-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP10]], 1
8975 // CHECK15-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
8976 // CHECK15-NEXT: store i32 [[ADD]], ptr [[I]], align 4
8977 // CHECK15-NEXT: [[TMP11:%.*]] = load i32, ptr [[I]], align 4
8978 // CHECK15-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds [10 x i32], ptr [[TMP0]], i32 0, i32 [[TMP11]]
8979 // CHECK15-NEXT: store i32 0, ptr [[ARRAYIDX]], align 4
8980 // CHECK15-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
8981 // CHECK15: omp.body.continue:
8982 // CHECK15-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
8983 // CHECK15: omp.inner.for.inc:
8984 // CHECK15-NEXT: [[TMP12:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
8985 // CHECK15-NEXT: [[ADD2:%.*]] = add nsw i32 [[TMP12]], 1
8986 // CHECK15-NEXT: store i32 [[ADD2]], ptr [[DOTOMP_IV]], align 4
8987 // CHECK15-NEXT: br label [[OMP_INNER_FOR_COND]]
8988 // CHECK15: omp.inner.for.end:
8989 // CHECK15-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
8990 // CHECK15: omp.loop.exit:
8991 // CHECK15-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB2]], i32 [[TMP4]])
8992 // CHECK15-NEXT: ret void
8995 // CHECK15-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l126
8996 // CHECK15-SAME: (i32 noundef [[M:%.*]], ptr noundef nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] {
8997 // CHECK15-NEXT: entry:
8998 // CHECK15-NEXT: [[M_ADDR:%.*]] = alloca i32, align 4
8999 // CHECK15-NEXT: [[A_ADDR:%.*]] = alloca ptr, align 4
9000 // CHECK15-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4
9001 // CHECK15-NEXT: [[DOTCAPTURE_EXPR__CASTED:%.*]] = alloca i32, align 4
9002 // CHECK15-NEXT: store i32 [[M]], ptr [[M_ADDR]], align 4
9003 // CHECK15-NEXT: store ptr [[A]], ptr [[A_ADDR]], align 4
9004 // CHECK15-NEXT: [[TMP0:%.*]] = load ptr, ptr [[A_ADDR]], align 4
9005 // CHECK15-NEXT: [[TMP1:%.*]] = load i32, ptr [[M_ADDR]], align 4
9006 // CHECK15-NEXT: store i32 [[TMP1]], ptr [[DOTCAPTURE_EXPR_]], align 4
9007 // CHECK15-NEXT: [[TMP2:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4
9008 // CHECK15-NEXT: store i32 [[TMP2]], ptr [[DOTCAPTURE_EXPR__CASTED]], align 4
9009 // CHECK15-NEXT: [[TMP3:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR__CASTED]], align 4
9010 // CHECK15-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_teams(ptr @[[GLOB3]], i32 2, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l126.omp_outlined, ptr [[TMP0]], i32 [[TMP3]])
9011 // CHECK15-NEXT: ret void
9014 // CHECK15-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l126.omp_outlined
9015 // CHECK15-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], ptr noundef nonnull align 4 dereferenceable(40) [[A:%.*]], i32 noundef [[DOTCAPTURE_EXPR_:%.*]]) #[[ATTR2]] {
9016 // CHECK15-NEXT: entry:
9017 // CHECK15-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 4
9018 // CHECK15-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 4
9019 // CHECK15-NEXT: [[A_ADDR:%.*]] = alloca ptr, align 4
9020 // CHECK15-NEXT: [[DOTCAPTURE_EXPR__ADDR:%.*]] = alloca i32, align 4
9021 // CHECK15-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
9022 // CHECK15-NEXT: [[TMP:%.*]] = alloca i32, align 4
9023 // CHECK15-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4
9024 // CHECK15-NEXT: [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4
9025 // CHECK15-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
9026 // CHECK15-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
9027 // CHECK15-NEXT: [[I:%.*]] = alloca i32, align 4
9028 // CHECK15-NEXT: [[DOTCAPTURE_EXPR__CASTED:%.*]] = alloca i32, align 4
9029 // CHECK15-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 4
9030 // CHECK15-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 4
9031 // CHECK15-NEXT: store ptr [[A]], ptr [[A_ADDR]], align 4
9032 // CHECK15-NEXT: store i32 [[DOTCAPTURE_EXPR_]], ptr [[DOTCAPTURE_EXPR__ADDR]], align 4
9033 // CHECK15-NEXT: [[TMP0:%.*]] = load ptr, ptr [[A_ADDR]], align 4
9034 // CHECK15-NEXT: store i32 0, ptr [[DOTOMP_COMB_LB]], align 4
9035 // CHECK15-NEXT: store i32 9, ptr [[DOTOMP_COMB_UB]], align 4
9036 // CHECK15-NEXT: store i32 1, ptr [[DOTOMP_STRIDE]], align 4
9037 // CHECK15-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
9038 // CHECK15-NEXT: [[TMP1:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
9039 // CHECK15-NEXT: [[TMP2:%.*]] = load i32, ptr [[TMP1]], align 4
9040 // CHECK15-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB1]], i32 [[TMP2]], i32 92, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
9041 // CHECK15-NEXT: [[TMP3:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
9042 // CHECK15-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP3]], 9
9043 // CHECK15-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
9044 // CHECK15: cond.true:
9045 // CHECK15-NEXT: br label [[COND_END:%.*]]
9046 // CHECK15: cond.false:
9047 // CHECK15-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
9048 // CHECK15-NEXT: br label [[COND_END]]
9049 // CHECK15: cond.end:
9050 // CHECK15-NEXT: [[COND:%.*]] = phi i32 [ 9, [[COND_TRUE]] ], [ [[TMP4]], [[COND_FALSE]] ]
9051 // CHECK15-NEXT: store i32 [[COND]], ptr [[DOTOMP_COMB_UB]], align 4
9052 // CHECK15-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
9053 // CHECK15-NEXT: store i32 [[TMP5]], ptr [[DOTOMP_IV]], align 4
9054 // CHECK15-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
9055 // CHECK15: omp.inner.for.cond:
9056 // CHECK15-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
9057 // CHECK15-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
9058 // CHECK15-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP6]], [[TMP7]]
9059 // CHECK15-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
9060 // CHECK15: omp.inner.for.body:
9061 // CHECK15-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
9062 // CHECK15-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
9063 // CHECK15-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR__ADDR]], align 4
9064 // CHECK15-NEXT: store i32 [[TMP10]], ptr [[DOTCAPTURE_EXPR__CASTED]], align 4
9065 // CHECK15-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR__CASTED]], align 4
9066 // CHECK15-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_call(ptr @[[GLOB3]], i32 4, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l126.omp_outlined.omp_outlined, i32 [[TMP8]], i32 [[TMP9]], ptr [[TMP0]], i32 [[TMP11]])
9067 // CHECK15-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
9068 // CHECK15: omp.inner.for.inc:
9069 // CHECK15-NEXT: [[TMP12:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
9070 // CHECK15-NEXT: [[TMP13:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4
9071 // CHECK15-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP12]], [[TMP13]]
9072 // CHECK15-NEXT: store i32 [[ADD]], ptr [[DOTOMP_IV]], align 4
9073 // CHECK15-NEXT: br label [[OMP_INNER_FOR_COND]]
9074 // CHECK15: omp.inner.for.end:
9075 // CHECK15-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
9076 // CHECK15: omp.loop.exit:
9077 // CHECK15-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB1]], i32 [[TMP2]])
9078 // CHECK15-NEXT: ret void
9081 // CHECK15-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l126.omp_outlined.omp_outlined
9082 // CHECK15-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], i32 noundef [[DOTPREVIOUS_LB_:%.*]], i32 noundef [[DOTPREVIOUS_UB_:%.*]], ptr noundef nonnull align 4 dereferenceable(40) [[A:%.*]], i32 noundef [[DOTCAPTURE_EXPR_:%.*]]) #[[ATTR2]] {
9083 // CHECK15-NEXT: entry:
9084 // CHECK15-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 4
9085 // CHECK15-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 4
9086 // CHECK15-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i32, align 4
9087 // CHECK15-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i32, align 4
9088 // CHECK15-NEXT: [[A_ADDR:%.*]] = alloca ptr, align 4
9089 // CHECK15-NEXT: [[DOTCAPTURE_EXPR__ADDR:%.*]] = alloca i32, align 4
9090 // CHECK15-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
9091 // CHECK15-NEXT: [[TMP:%.*]] = alloca i32, align 4
9092 // CHECK15-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4
9093 // CHECK15-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4
9094 // CHECK15-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
9095 // CHECK15-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
9096 // CHECK15-NEXT: [[I:%.*]] = alloca i32, align 4
9097 // CHECK15-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 4
9098 // CHECK15-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 4
9099 // CHECK15-NEXT: store i32 [[DOTPREVIOUS_LB_]], ptr [[DOTPREVIOUS_LB__ADDR]], align 4
9100 // CHECK15-NEXT: store i32 [[DOTPREVIOUS_UB_]], ptr [[DOTPREVIOUS_UB__ADDR]], align 4
9101 // CHECK15-NEXT: store ptr [[A]], ptr [[A_ADDR]], align 4
9102 // CHECK15-NEXT: store i32 [[DOTCAPTURE_EXPR_]], ptr [[DOTCAPTURE_EXPR__ADDR]], align 4
9103 // CHECK15-NEXT: [[TMP0:%.*]] = load ptr, ptr [[A_ADDR]], align 4
9104 // CHECK15-NEXT: store i32 0, ptr [[DOTOMP_LB]], align 4
9105 // CHECK15-NEXT: store i32 9, ptr [[DOTOMP_UB]], align 4
9106 // CHECK15-NEXT: [[TMP1:%.*]] = load i32, ptr [[DOTPREVIOUS_LB__ADDR]], align 4
9107 // CHECK15-NEXT: [[TMP2:%.*]] = load i32, ptr [[DOTPREVIOUS_UB__ADDR]], align 4
9108 // CHECK15-NEXT: store i32 [[TMP1]], ptr [[DOTOMP_LB]], align 4
9109 // CHECK15-NEXT: store i32 [[TMP2]], ptr [[DOTOMP_UB]], align 4
9110 // CHECK15-NEXT: store i32 1, ptr [[DOTOMP_STRIDE]], align 4
9111 // CHECK15-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
9112 // CHECK15-NEXT: [[TMP3:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR__ADDR]], align 4
9113 // CHECK15-NEXT: [[TMP4:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
9114 // CHECK15-NEXT: [[TMP5:%.*]] = load i32, ptr [[TMP4]], align 4
9115 // CHECK15-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB2]], i32 [[TMP5]], i32 33, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 [[TMP3]])
9116 // CHECK15-NEXT: br label [[OMP_DISPATCH_COND:%.*]]
9117 // CHECK15: omp.dispatch.cond:
9118 // CHECK15-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
9119 // CHECK15-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTPREVIOUS_UB__ADDR]], align 4
9120 // CHECK15-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP6]], [[TMP7]]
9121 // CHECK15-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
9122 // CHECK15: cond.true:
9123 // CHECK15-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTPREVIOUS_UB__ADDR]], align 4
9124 // CHECK15-NEXT: br label [[COND_END:%.*]]
9125 // CHECK15: cond.false:
9126 // CHECK15-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
9127 // CHECK15-NEXT: br label [[COND_END]]
9128 // CHECK15: cond.end:
9129 // CHECK15-NEXT: [[COND:%.*]] = phi i32 [ [[TMP8]], [[COND_TRUE]] ], [ [[TMP9]], [[COND_FALSE]] ]
9130 // CHECK15-NEXT: store i32 [[COND]], ptr [[DOTOMP_UB]], align 4
9131 // CHECK15-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4
9132 // CHECK15-NEXT: store i32 [[TMP10]], ptr [[DOTOMP_IV]], align 4
9133 // CHECK15-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
9134 // CHECK15-NEXT: [[TMP12:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
9135 // CHECK15-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP11]], [[TMP12]]
9136 // CHECK15-NEXT: br i1 [[CMP1]], label [[OMP_DISPATCH_BODY:%.*]], label [[OMP_DISPATCH_END:%.*]]
9137 // CHECK15: omp.dispatch.body:
9138 // CHECK15-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
9139 // CHECK15: omp.inner.for.cond:
9140 // CHECK15-NEXT: [[TMP13:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
9141 // CHECK15-NEXT: [[TMP14:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
9142 // CHECK15-NEXT: [[CMP2:%.*]] = icmp sle i32 [[TMP13]], [[TMP14]]
9143 // CHECK15-NEXT: br i1 [[CMP2]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
9144 // CHECK15: omp.inner.for.body:
9145 // CHECK15-NEXT: [[TMP15:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
9146 // CHECK15-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP15]], 1
9147 // CHECK15-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
9148 // CHECK15-NEXT: store i32 [[ADD]], ptr [[I]], align 4
9149 // CHECK15-NEXT: [[TMP16:%.*]] = load i32, ptr [[I]], align 4
9150 // CHECK15-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds [10 x i32], ptr [[TMP0]], i32 0, i32 [[TMP16]]
9151 // CHECK15-NEXT: store i32 0, ptr [[ARRAYIDX]], align 4
9152 // CHECK15-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
9153 // CHECK15: omp.body.continue:
9154 // CHECK15-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
9155 // CHECK15: omp.inner.for.inc:
9156 // CHECK15-NEXT: [[TMP17:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
9157 // CHECK15-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP17]], 1
9158 // CHECK15-NEXT: store i32 [[ADD3]], ptr [[DOTOMP_IV]], align 4
9159 // CHECK15-NEXT: br label [[OMP_INNER_FOR_COND]]
9160 // CHECK15: omp.inner.for.end:
9161 // CHECK15-NEXT: br label [[OMP_DISPATCH_INC:%.*]]
9162 // CHECK15: omp.dispatch.inc:
9163 // CHECK15-NEXT: [[TMP18:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4
9164 // CHECK15-NEXT: [[TMP19:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4
9165 // CHECK15-NEXT: [[ADD4:%.*]] = add nsw i32 [[TMP18]], [[TMP19]]
9166 // CHECK15-NEXT: store i32 [[ADD4]], ptr [[DOTOMP_LB]], align 4
9167 // CHECK15-NEXT: [[TMP20:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
9168 // CHECK15-NEXT: [[TMP21:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4
9169 // CHECK15-NEXT: [[ADD5:%.*]] = add nsw i32 [[TMP20]], [[TMP21]]
9170 // CHECK15-NEXT: store i32 [[ADD5]], ptr [[DOTOMP_UB]], align 4
9171 // CHECK15-NEXT: br label [[OMP_DISPATCH_COND]]
9172 // CHECK15: omp.dispatch.end:
9173 // CHECK15-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB2]], i32 [[TMP5]])
9174 // CHECK15-NEXT: ret void
9177 // CHECK15-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l131
9178 // CHECK15-SAME: (ptr noundef nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] {
9179 // CHECK15-NEXT: entry:
9180 // CHECK15-NEXT: [[A_ADDR:%.*]] = alloca ptr, align 4
9181 // CHECK15-NEXT: store ptr [[A]], ptr [[A_ADDR]], align 4
9182 // CHECK15-NEXT: [[TMP0:%.*]] = load ptr, ptr [[A_ADDR]], align 4
9183 // CHECK15-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_teams(ptr @[[GLOB3]], i32 1, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l131.omp_outlined, ptr [[TMP0]])
9184 // CHECK15-NEXT: ret void
9187 // CHECK15-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l131.omp_outlined
9188 // CHECK15-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], ptr noundef nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] {
9189 // CHECK15-NEXT: entry:
9190 // CHECK15-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 4
9191 // CHECK15-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 4
9192 // CHECK15-NEXT: [[A_ADDR:%.*]] = alloca ptr, align 4
9193 // CHECK15-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
9194 // CHECK15-NEXT: [[TMP:%.*]] = alloca i32, align 4
9195 // CHECK15-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4
9196 // CHECK15-NEXT: [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4
9197 // CHECK15-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
9198 // CHECK15-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
9199 // CHECK15-NEXT: [[I:%.*]] = alloca i32, align 4
9200 // CHECK15-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 4
9201 // CHECK15-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 4
9202 // CHECK15-NEXT: store ptr [[A]], ptr [[A_ADDR]], align 4
9203 // CHECK15-NEXT: [[TMP0:%.*]] = load ptr, ptr [[A_ADDR]], align 4
9204 // CHECK15-NEXT: store i32 0, ptr [[DOTOMP_COMB_LB]], align 4
9205 // CHECK15-NEXT: store i32 9, ptr [[DOTOMP_COMB_UB]], align 4
9206 // CHECK15-NEXT: store i32 1, ptr [[DOTOMP_STRIDE]], align 4
9207 // CHECK15-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
9208 // CHECK15-NEXT: [[TMP1:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
9209 // CHECK15-NEXT: [[TMP2:%.*]] = load i32, ptr [[TMP1]], align 4
9210 // CHECK15-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB1]], i32 [[TMP2]], i32 92, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
9211 // CHECK15-NEXT: [[TMP3:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
9212 // CHECK15-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP3]], 9
9213 // CHECK15-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
9214 // CHECK15: cond.true:
9215 // CHECK15-NEXT: br label [[COND_END:%.*]]
9216 // CHECK15: cond.false:
9217 // CHECK15-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
9218 // CHECK15-NEXT: br label [[COND_END]]
9219 // CHECK15: cond.end:
9220 // CHECK15-NEXT: [[COND:%.*]] = phi i32 [ 9, [[COND_TRUE]] ], [ [[TMP4]], [[COND_FALSE]] ]
9221 // CHECK15-NEXT: store i32 [[COND]], ptr [[DOTOMP_COMB_UB]], align 4
9222 // CHECK15-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
9223 // CHECK15-NEXT: store i32 [[TMP5]], ptr [[DOTOMP_IV]], align 4
9224 // CHECK15-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
9225 // CHECK15: omp.inner.for.cond:
9226 // CHECK15-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
9227 // CHECK15-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
9228 // CHECK15-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP6]], [[TMP7]]
9229 // CHECK15-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
9230 // CHECK15: omp.inner.for.body:
9231 // CHECK15-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
9232 // CHECK15-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
9233 // CHECK15-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_call(ptr @[[GLOB3]], i32 3, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l131.omp_outlined.omp_outlined, i32 [[TMP8]], i32 [[TMP9]], ptr [[TMP0]])
9234 // CHECK15-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
9235 // CHECK15: omp.inner.for.inc:
9236 // CHECK15-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
9237 // CHECK15-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4
9238 // CHECK15-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP10]], [[TMP11]]
9239 // CHECK15-NEXT: store i32 [[ADD]], ptr [[DOTOMP_IV]], align 4
9240 // CHECK15-NEXT: br label [[OMP_INNER_FOR_COND]]
9241 // CHECK15: omp.inner.for.end:
9242 // CHECK15-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
9243 // CHECK15: omp.loop.exit:
9244 // CHECK15-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB1]], i32 [[TMP2]])
9245 // CHECK15-NEXT: ret void
9248 // CHECK15-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l131.omp_outlined.omp_outlined
9249 // CHECK15-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], i32 noundef [[DOTPREVIOUS_LB_:%.*]], i32 noundef [[DOTPREVIOUS_UB_:%.*]], ptr noundef nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] {
9250 // CHECK15-NEXT: entry:
9251 // CHECK15-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 4
9252 // CHECK15-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 4
9253 // CHECK15-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i32, align 4
9254 // CHECK15-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i32, align 4
9255 // CHECK15-NEXT: [[A_ADDR:%.*]] = alloca ptr, align 4
9256 // CHECK15-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
9257 // CHECK15-NEXT: [[TMP:%.*]] = alloca i32, align 4
9258 // CHECK15-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4
9259 // CHECK15-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4
9260 // CHECK15-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
9261 // CHECK15-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
9262 // CHECK15-NEXT: [[I:%.*]] = alloca i32, align 4
9263 // CHECK15-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 4
9264 // CHECK15-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 4
9265 // CHECK15-NEXT: store i32 [[DOTPREVIOUS_LB_]], ptr [[DOTPREVIOUS_LB__ADDR]], align 4
9266 // CHECK15-NEXT: store i32 [[DOTPREVIOUS_UB_]], ptr [[DOTPREVIOUS_UB__ADDR]], align 4
9267 // CHECK15-NEXT: store ptr [[A]], ptr [[A_ADDR]], align 4
9268 // CHECK15-NEXT: [[TMP0:%.*]] = load ptr, ptr [[A_ADDR]], align 4
9269 // CHECK15-NEXT: store i32 0, ptr [[DOTOMP_LB]], align 4
9270 // CHECK15-NEXT: store i32 9, ptr [[DOTOMP_UB]], align 4
9271 // CHECK15-NEXT: [[TMP1:%.*]] = load i32, ptr [[DOTPREVIOUS_LB__ADDR]], align 4
9272 // CHECK15-NEXT: [[TMP2:%.*]] = load i32, ptr [[DOTPREVIOUS_UB__ADDR]], align 4
9273 // CHECK15-NEXT: store i32 [[TMP1]], ptr [[DOTOMP_LB]], align 4
9274 // CHECK15-NEXT: store i32 [[TMP2]], ptr [[DOTOMP_UB]], align 4
9275 // CHECK15-NEXT: store i32 1, ptr [[DOTOMP_STRIDE]], align 4
9276 // CHECK15-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
9277 // CHECK15-NEXT: [[TMP3:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4
9278 // CHECK15-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
9279 // CHECK15-NEXT: [[TMP5:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
9280 // CHECK15-NEXT: [[TMP6:%.*]] = load i32, ptr [[TMP5]], align 4
9281 // CHECK15-NEXT: call void @__kmpc_dispatch_init_4(ptr @[[GLOB3]], i32 [[TMP6]], i32 1073741859, i32 [[TMP3]], i32 [[TMP4]], i32 1, i32 1)
9282 // CHECK15-NEXT: br label [[OMP_DISPATCH_COND:%.*]]
9283 // CHECK15: omp.dispatch.cond:
9284 // CHECK15-NEXT: [[TMP7:%.*]] = call i32 @__kmpc_dispatch_next_4(ptr @[[GLOB3]], i32 [[TMP6]], ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]])
9285 // CHECK15-NEXT: [[TOBOOL:%.*]] = icmp ne i32 [[TMP7]], 0
9286 // CHECK15-NEXT: br i1 [[TOBOOL]], label [[OMP_DISPATCH_BODY:%.*]], label [[OMP_DISPATCH_END:%.*]]
9287 // CHECK15: omp.dispatch.body:
9288 // CHECK15-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4
9289 // CHECK15-NEXT: store i32 [[TMP8]], ptr [[DOTOMP_IV]], align 4
9290 // CHECK15-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
9291 // CHECK15: omp.inner.for.cond:
9292 // CHECK15-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP22:![0-9]+]]
9293 // CHECK15-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP22]]
9294 // CHECK15-NEXT: [[CMP:%.*]] = icmp sle i32 [[TMP9]], [[TMP10]]
9295 // CHECK15-NEXT: br i1 [[CMP]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
9296 // CHECK15: omp.inner.for.body:
9297 // CHECK15-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP22]]
9298 // CHECK15-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP11]], 1
9299 // CHECK15-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
9300 // CHECK15-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP22]]
9301 // CHECK15-NEXT: [[TMP12:%.*]] = load i32, ptr [[I]], align 4, !llvm.access.group [[ACC_GRP22]]
9302 // CHECK15-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds [10 x i32], ptr [[TMP0]], i32 0, i32 [[TMP12]]
9303 // CHECK15-NEXT: store i32 0, ptr [[ARRAYIDX]], align 4, !llvm.access.group [[ACC_GRP22]]
9304 // CHECK15-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
9305 // CHECK15: omp.body.continue:
9306 // CHECK15-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
9307 // CHECK15: omp.inner.for.inc:
9308 // CHECK15-NEXT: [[TMP13:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP22]]
9309 // CHECK15-NEXT: [[ADD1:%.*]] = add nsw i32 [[TMP13]], 1
9310 // CHECK15-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP22]]
9311 // CHECK15-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP23:![0-9]+]]
9312 // CHECK15: omp.inner.for.end:
9313 // CHECK15-NEXT: br label [[OMP_DISPATCH_INC:%.*]]
9314 // CHECK15: omp.dispatch.inc:
9315 // CHECK15-NEXT: br label [[OMP_DISPATCH_COND]]
9316 // CHECK15: omp.dispatch.end:
9317 // CHECK15-NEXT: call void @__kmpc_dispatch_deinit(ptr @[[GLOB3]], i32 [[TMP6]])
9318 // CHECK15-NEXT: ret void
9321 // CHECK15-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l136
9322 // CHECK15-SAME: (i32 noundef [[M:%.*]], ptr noundef nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] {
9323 // CHECK15-NEXT: entry:
9324 // CHECK15-NEXT: [[M_ADDR:%.*]] = alloca i32, align 4
9325 // CHECK15-NEXT: [[A_ADDR:%.*]] = alloca ptr, align 4
9326 // CHECK15-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4
9327 // CHECK15-NEXT: [[DOTCAPTURE_EXPR__CASTED:%.*]] = alloca i32, align 4
9328 // CHECK15-NEXT: store i32 [[M]], ptr [[M_ADDR]], align 4
9329 // CHECK15-NEXT: store ptr [[A]], ptr [[A_ADDR]], align 4
9330 // CHECK15-NEXT: [[TMP0:%.*]] = load ptr, ptr [[A_ADDR]], align 4
9331 // CHECK15-NEXT: [[TMP1:%.*]] = load i32, ptr [[M_ADDR]], align 4
9332 // CHECK15-NEXT: store i32 [[TMP1]], ptr [[DOTCAPTURE_EXPR_]], align 4
9333 // CHECK15-NEXT: [[TMP2:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4
9334 // CHECK15-NEXT: store i32 [[TMP2]], ptr [[DOTCAPTURE_EXPR__CASTED]], align 4
9335 // CHECK15-NEXT: [[TMP3:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR__CASTED]], align 4
9336 // CHECK15-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_teams(ptr @[[GLOB3]], i32 2, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l136.omp_outlined, ptr [[TMP0]], i32 [[TMP3]])
9337 // CHECK15-NEXT: ret void
9340 // CHECK15-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l136.omp_outlined
9341 // CHECK15-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], ptr noundef nonnull align 4 dereferenceable(40) [[A:%.*]], i32 noundef [[DOTCAPTURE_EXPR_:%.*]]) #[[ATTR2]] {
9342 // CHECK15-NEXT: entry:
9343 // CHECK15-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 4
9344 // CHECK15-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 4
9345 // CHECK15-NEXT: [[A_ADDR:%.*]] = alloca ptr, align 4
9346 // CHECK15-NEXT: [[DOTCAPTURE_EXPR__ADDR:%.*]] = alloca i32, align 4
9347 // CHECK15-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
9348 // CHECK15-NEXT: [[TMP:%.*]] = alloca i32, align 4
9349 // CHECK15-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4
9350 // CHECK15-NEXT: [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4
9351 // CHECK15-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
9352 // CHECK15-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
9353 // CHECK15-NEXT: [[I:%.*]] = alloca i32, align 4
9354 // CHECK15-NEXT: [[DOTCAPTURE_EXPR__CASTED:%.*]] = alloca i32, align 4
9355 // CHECK15-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 4
9356 // CHECK15-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 4
9357 // CHECK15-NEXT: store ptr [[A]], ptr [[A_ADDR]], align 4
9358 // CHECK15-NEXT: store i32 [[DOTCAPTURE_EXPR_]], ptr [[DOTCAPTURE_EXPR__ADDR]], align 4
9359 // CHECK15-NEXT: [[TMP0:%.*]] = load ptr, ptr [[A_ADDR]], align 4
9360 // CHECK15-NEXT: store i32 0, ptr [[DOTOMP_COMB_LB]], align 4
9361 // CHECK15-NEXT: store i32 9, ptr [[DOTOMP_COMB_UB]], align 4
9362 // CHECK15-NEXT: store i32 1, ptr [[DOTOMP_STRIDE]], align 4
9363 // CHECK15-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
9364 // CHECK15-NEXT: [[TMP1:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
9365 // CHECK15-NEXT: [[TMP2:%.*]] = load i32, ptr [[TMP1]], align 4
9366 // CHECK15-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB1]], i32 [[TMP2]], i32 92, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
9367 // CHECK15-NEXT: [[TMP3:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
9368 // CHECK15-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP3]], 9
9369 // CHECK15-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
9370 // CHECK15: cond.true:
9371 // CHECK15-NEXT: br label [[COND_END:%.*]]
9372 // CHECK15: cond.false:
9373 // CHECK15-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
9374 // CHECK15-NEXT: br label [[COND_END]]
9375 // CHECK15: cond.end:
9376 // CHECK15-NEXT: [[COND:%.*]] = phi i32 [ 9, [[COND_TRUE]] ], [ [[TMP4]], [[COND_FALSE]] ]
9377 // CHECK15-NEXT: store i32 [[COND]], ptr [[DOTOMP_COMB_UB]], align 4
9378 // CHECK15-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
9379 // CHECK15-NEXT: store i32 [[TMP5]], ptr [[DOTOMP_IV]], align 4
9380 // CHECK15-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
9381 // CHECK15: omp.inner.for.cond:
9382 // CHECK15-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
9383 // CHECK15-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
9384 // CHECK15-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP6]], [[TMP7]]
9385 // CHECK15-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
9386 // CHECK15: omp.inner.for.body:
9387 // CHECK15-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
9388 // CHECK15-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
9389 // CHECK15-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR__ADDR]], align 4
9390 // CHECK15-NEXT: store i32 [[TMP10]], ptr [[DOTCAPTURE_EXPR__CASTED]], align 4
9391 // CHECK15-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR__CASTED]], align 4
9392 // CHECK15-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_call(ptr @[[GLOB3]], i32 4, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l136.omp_outlined.omp_outlined, i32 [[TMP8]], i32 [[TMP9]], ptr [[TMP0]], i32 [[TMP11]])
9393 // CHECK15-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
9394 // CHECK15: omp.inner.for.inc:
9395 // CHECK15-NEXT: [[TMP12:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
9396 // CHECK15-NEXT: [[TMP13:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4
9397 // CHECK15-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP12]], [[TMP13]]
9398 // CHECK15-NEXT: store i32 [[ADD]], ptr [[DOTOMP_IV]], align 4
9399 // CHECK15-NEXT: br label [[OMP_INNER_FOR_COND]]
9400 // CHECK15: omp.inner.for.end:
9401 // CHECK15-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
9402 // CHECK15: omp.loop.exit:
9403 // CHECK15-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB1]], i32 [[TMP2]])
9404 // CHECK15-NEXT: ret void
9407 // CHECK15-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l136.omp_outlined.omp_outlined
9408 // CHECK15-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], i32 noundef [[DOTPREVIOUS_LB_:%.*]], i32 noundef [[DOTPREVIOUS_UB_:%.*]], ptr noundef nonnull align 4 dereferenceable(40) [[A:%.*]], i32 noundef [[DOTCAPTURE_EXPR_:%.*]]) #[[ATTR2]] {
9409 // CHECK15-NEXT: entry:
9410 // CHECK15-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 4
9411 // CHECK15-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 4
9412 // CHECK15-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i32, align 4
9413 // CHECK15-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i32, align 4
9414 // CHECK15-NEXT: [[A_ADDR:%.*]] = alloca ptr, align 4
9415 // CHECK15-NEXT: [[DOTCAPTURE_EXPR__ADDR:%.*]] = alloca i32, align 4
9416 // CHECK15-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
9417 // CHECK15-NEXT: [[TMP:%.*]] = alloca i32, align 4
9418 // CHECK15-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4
9419 // CHECK15-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4
9420 // CHECK15-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
9421 // CHECK15-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
9422 // CHECK15-NEXT: [[I:%.*]] = alloca i32, align 4
9423 // CHECK15-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 4
9424 // CHECK15-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 4
9425 // CHECK15-NEXT: store i32 [[DOTPREVIOUS_LB_]], ptr [[DOTPREVIOUS_LB__ADDR]], align 4
9426 // CHECK15-NEXT: store i32 [[DOTPREVIOUS_UB_]], ptr [[DOTPREVIOUS_UB__ADDR]], align 4
9427 // CHECK15-NEXT: store ptr [[A]], ptr [[A_ADDR]], align 4
9428 // CHECK15-NEXT: store i32 [[DOTCAPTURE_EXPR_]], ptr [[DOTCAPTURE_EXPR__ADDR]], align 4
9429 // CHECK15-NEXT: [[TMP0:%.*]] = load ptr, ptr [[A_ADDR]], align 4
9430 // CHECK15-NEXT: store i32 0, ptr [[DOTOMP_LB]], align 4
9431 // CHECK15-NEXT: store i32 9, ptr [[DOTOMP_UB]], align 4
9432 // CHECK15-NEXT: [[TMP1:%.*]] = load i32, ptr [[DOTPREVIOUS_LB__ADDR]], align 4
9433 // CHECK15-NEXT: [[TMP2:%.*]] = load i32, ptr [[DOTPREVIOUS_UB__ADDR]], align 4
9434 // CHECK15-NEXT: store i32 [[TMP1]], ptr [[DOTOMP_LB]], align 4
9435 // CHECK15-NEXT: store i32 [[TMP2]], ptr [[DOTOMP_UB]], align 4
9436 // CHECK15-NEXT: store i32 1, ptr [[DOTOMP_STRIDE]], align 4
9437 // CHECK15-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
9438 // CHECK15-NEXT: [[TMP3:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR__ADDR]], align 4
9439 // CHECK15-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4
9440 // CHECK15-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
9441 // CHECK15-NEXT: [[TMP6:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
9442 // CHECK15-NEXT: [[TMP7:%.*]] = load i32, ptr [[TMP6]], align 4
9443 // CHECK15-NEXT: call void @__kmpc_dispatch_init_4(ptr @[[GLOB3]], i32 [[TMP7]], i32 1073741859, i32 [[TMP4]], i32 [[TMP5]], i32 1, i32 [[TMP3]])
9444 // CHECK15-NEXT: br label [[OMP_DISPATCH_COND:%.*]]
9445 // CHECK15: omp.dispatch.cond:
9446 // CHECK15-NEXT: [[TMP8:%.*]] = call i32 @__kmpc_dispatch_next_4(ptr @[[GLOB3]], i32 [[TMP7]], ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]])
9447 // CHECK15-NEXT: [[TOBOOL:%.*]] = icmp ne i32 [[TMP8]], 0
9448 // CHECK15-NEXT: br i1 [[TOBOOL]], label [[OMP_DISPATCH_BODY:%.*]], label [[OMP_DISPATCH_END:%.*]]
9449 // CHECK15: omp.dispatch.body:
9450 // CHECK15-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4
9451 // CHECK15-NEXT: store i32 [[TMP9]], ptr [[DOTOMP_IV]], align 4
9452 // CHECK15-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
9453 // CHECK15: omp.inner.for.cond:
9454 // CHECK15-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP25:![0-9]+]]
9455 // CHECK15-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP25]]
9456 // CHECK15-NEXT: [[CMP:%.*]] = icmp sle i32 [[TMP10]], [[TMP11]]
9457 // CHECK15-NEXT: br i1 [[CMP]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
9458 // CHECK15: omp.inner.for.body:
9459 // CHECK15-NEXT: [[TMP12:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP25]]
9460 // CHECK15-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP12]], 1
9461 // CHECK15-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
9462 // CHECK15-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP25]]
9463 // CHECK15-NEXT: [[TMP13:%.*]] = load i32, ptr [[I]], align 4, !llvm.access.group [[ACC_GRP25]]
9464 // CHECK15-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds [10 x i32], ptr [[TMP0]], i32 0, i32 [[TMP13]]
9465 // CHECK15-NEXT: store i32 0, ptr [[ARRAYIDX]], align 4, !llvm.access.group [[ACC_GRP25]]
9466 // CHECK15-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
9467 // CHECK15: omp.body.continue:
9468 // CHECK15-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
9469 // CHECK15: omp.inner.for.inc:
9470 // CHECK15-NEXT: [[TMP14:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP25]]
9471 // CHECK15-NEXT: [[ADD1:%.*]] = add nsw i32 [[TMP14]], 1
9472 // CHECK15-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP25]]
9473 // CHECK15-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP26:![0-9]+]]
9474 // CHECK15: omp.inner.for.end:
9475 // CHECK15-NEXT: br label [[OMP_DISPATCH_INC:%.*]]
9476 // CHECK15: omp.dispatch.inc:
9477 // CHECK15-NEXT: br label [[OMP_DISPATCH_COND]]
9478 // CHECK15: omp.dispatch.end:
9479 // CHECK15-NEXT: call void @__kmpc_dispatch_deinit(ptr @[[GLOB3]], i32 [[TMP7]])
9480 // CHECK15-NEXT: ret void
9483 // CHECK17-LABEL: define {{[^@]+}}@main
9484 // CHECK17-SAME: (i32 noundef signext [[ARGC:%.*]], ptr noundef [[ARGV:%.*]]) #[[ATTR0:[0-9]+]] {
9485 // CHECK17-NEXT: entry:
9486 // CHECK17-NEXT: [[RETVAL:%.*]] = alloca i32, align 4
9487 // CHECK17-NEXT: [[ARGC_ADDR:%.*]] = alloca i32, align 4
9488 // CHECK17-NEXT: [[ARGV_ADDR:%.*]] = alloca ptr, align 8
9489 // CHECK17-NEXT: [[N:%.*]] = alloca i32, align 4
9490 // CHECK17-NEXT: [[SAVED_STACK:%.*]] = alloca ptr, align 8
9491 // CHECK17-NEXT: [[__VLA_EXPR0:%.*]] = alloca i64, align 8
9492 // CHECK17-NEXT: [[M:%.*]] = alloca i32, align 4
9493 // CHECK17-NEXT: [[N_CASTED:%.*]] = alloca i64, align 8
9494 // CHECK17-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [3 x ptr], align 8
9495 // CHECK17-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [3 x ptr], align 8
9496 // CHECK17-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [3 x ptr], align 8
9497 // CHECK17-NEXT: [[DOTOFFLOAD_SIZES:%.*]] = alloca [3 x i64], align 8
9498 // CHECK17-NEXT: [[TMP:%.*]] = alloca i32, align 4
9499 // CHECK17-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4
9500 // CHECK17-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4
9501 // CHECK17-NEXT: [[KERNEL_ARGS:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS:%.*]], align 8
9502 // CHECK17-NEXT: [[N_CASTED3:%.*]] = alloca i64, align 8
9503 // CHECK17-NEXT: [[DOTOFFLOAD_BASEPTRS4:%.*]] = alloca [3 x ptr], align 8
9504 // CHECK17-NEXT: [[DOTOFFLOAD_PTRS5:%.*]] = alloca [3 x ptr], align 8
9505 // CHECK17-NEXT: [[DOTOFFLOAD_MAPPERS6:%.*]] = alloca [3 x ptr], align 8
9506 // CHECK17-NEXT: [[DOTOFFLOAD_SIZES7:%.*]] = alloca [3 x i64], align 8
9507 // CHECK17-NEXT: [[_TMP8:%.*]] = alloca i32, align 4
9508 // CHECK17-NEXT: [[DOTCAPTURE_EXPR_9:%.*]] = alloca i32, align 4
9509 // CHECK17-NEXT: [[DOTCAPTURE_EXPR_10:%.*]] = alloca i32, align 4
9510 // CHECK17-NEXT: [[KERNEL_ARGS15:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS]], align 8
9511 // CHECK17-NEXT: [[M_CASTED:%.*]] = alloca i64, align 8
9512 // CHECK17-NEXT: [[N_CASTED18:%.*]] = alloca i64, align 8
9513 // CHECK17-NEXT: [[DOTOFFLOAD_BASEPTRS19:%.*]] = alloca [4 x ptr], align 8
9514 // CHECK17-NEXT: [[DOTOFFLOAD_PTRS20:%.*]] = alloca [4 x ptr], align 8
9515 // CHECK17-NEXT: [[DOTOFFLOAD_MAPPERS21:%.*]] = alloca [4 x ptr], align 8
9516 // CHECK17-NEXT: [[DOTOFFLOAD_SIZES22:%.*]] = alloca [4 x i64], align 8
9517 // CHECK17-NEXT: [[_TMP23:%.*]] = alloca i32, align 4
9518 // CHECK17-NEXT: [[DOTCAPTURE_EXPR_24:%.*]] = alloca i32, align 4
9519 // CHECK17-NEXT: [[DOTCAPTURE_EXPR_25:%.*]] = alloca i32, align 4
9520 // CHECK17-NEXT: [[KERNEL_ARGS30:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS]], align 8
9521 // CHECK17-NEXT: [[N_CASTED33:%.*]] = alloca i64, align 8
9522 // CHECK17-NEXT: [[DOTOFFLOAD_BASEPTRS34:%.*]] = alloca [3 x ptr], align 8
9523 // CHECK17-NEXT: [[DOTOFFLOAD_PTRS35:%.*]] = alloca [3 x ptr], align 8
9524 // CHECK17-NEXT: [[DOTOFFLOAD_MAPPERS36:%.*]] = alloca [3 x ptr], align 8
9525 // CHECK17-NEXT: [[DOTOFFLOAD_SIZES37:%.*]] = alloca [3 x i64], align 8
9526 // CHECK17-NEXT: [[_TMP38:%.*]] = alloca i32, align 4
9527 // CHECK17-NEXT: [[DOTCAPTURE_EXPR_39:%.*]] = alloca i32, align 4
9528 // CHECK17-NEXT: [[DOTCAPTURE_EXPR_40:%.*]] = alloca i32, align 4
9529 // CHECK17-NEXT: [[KERNEL_ARGS45:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS]], align 8
9530 // CHECK17-NEXT: [[M_CASTED48:%.*]] = alloca i64, align 8
9531 // CHECK17-NEXT: [[N_CASTED49:%.*]] = alloca i64, align 8
9532 // CHECK17-NEXT: [[DOTOFFLOAD_BASEPTRS50:%.*]] = alloca [4 x ptr], align 8
9533 // CHECK17-NEXT: [[DOTOFFLOAD_PTRS51:%.*]] = alloca [4 x ptr], align 8
9534 // CHECK17-NEXT: [[DOTOFFLOAD_MAPPERS52:%.*]] = alloca [4 x ptr], align 8
9535 // CHECK17-NEXT: [[DOTOFFLOAD_SIZES53:%.*]] = alloca [4 x i64], align 8
9536 // CHECK17-NEXT: [[_TMP54:%.*]] = alloca i32, align 4
9537 // CHECK17-NEXT: [[DOTCAPTURE_EXPR_55:%.*]] = alloca i32, align 4
9538 // CHECK17-NEXT: [[DOTCAPTURE_EXPR_56:%.*]] = alloca i32, align 4
9539 // CHECK17-NEXT: [[KERNEL_ARGS61:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS]], align 8
9540 // CHECK17-NEXT: store i32 0, ptr [[RETVAL]], align 4
9541 // CHECK17-NEXT: store i32 [[ARGC]], ptr [[ARGC_ADDR]], align 4
9542 // CHECK17-NEXT: store ptr [[ARGV]], ptr [[ARGV_ADDR]], align 8
9543 // CHECK17-NEXT: store i32 100, ptr [[N]], align 4
9544 // CHECK17-NEXT: [[TMP0:%.*]] = load i32, ptr [[N]], align 4
9545 // CHECK17-NEXT: [[TMP1:%.*]] = zext i32 [[TMP0]] to i64
9546 // CHECK17-NEXT: [[TMP2:%.*]] = call ptr @llvm.stacksave.p0()
9547 // CHECK17-NEXT: store ptr [[TMP2]], ptr [[SAVED_STACK]], align 8
9548 // CHECK17-NEXT: [[VLA:%.*]] = alloca i32, i64 [[TMP1]], align 4
9549 // CHECK17-NEXT: store i64 [[TMP1]], ptr [[__VLA_EXPR0]], align 8
9550 // CHECK17-NEXT: store i32 10, ptr [[M]], align 4
9551 // CHECK17-NEXT: [[TMP3:%.*]] = load i32, ptr [[N]], align 4
9552 // CHECK17-NEXT: store i32 [[TMP3]], ptr [[N_CASTED]], align 4
9553 // CHECK17-NEXT: [[TMP4:%.*]] = load i64, ptr [[N_CASTED]], align 8
9554 // CHECK17-NEXT: [[TMP5:%.*]] = mul nuw i64 [[TMP1]], 4
9555 // CHECK17-NEXT: call void @llvm.memcpy.p0.p0.i64(ptr align 8 [[DOTOFFLOAD_SIZES]], ptr align 8 @.offload_sizes, i64 24, i1 false)
9556 // CHECK17-NEXT: [[TMP6:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
9557 // CHECK17-NEXT: store i64 [[TMP4]], ptr [[TMP6]], align 8
9558 // CHECK17-NEXT: [[TMP7:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
9559 // CHECK17-NEXT: store i64 [[TMP4]], ptr [[TMP7]], align 8
9560 // CHECK17-NEXT: [[TMP8:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0
9561 // CHECK17-NEXT: store ptr null, ptr [[TMP8]], align 8
9562 // CHECK17-NEXT: [[TMP9:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1
9563 // CHECK17-NEXT: store i64 [[TMP1]], ptr [[TMP9]], align 8
9564 // CHECK17-NEXT: [[TMP10:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 1
9565 // CHECK17-NEXT: store i64 [[TMP1]], ptr [[TMP10]], align 8
9566 // CHECK17-NEXT: [[TMP11:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i64 0, i64 1
9567 // CHECK17-NEXT: store ptr null, ptr [[TMP11]], align 8
9568 // CHECK17-NEXT: [[TMP12:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
9569 // CHECK17-NEXT: store ptr [[VLA]], ptr [[TMP12]], align 8
9570 // CHECK17-NEXT: [[TMP13:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 2
9571 // CHECK17-NEXT: store ptr [[VLA]], ptr [[TMP13]], align 8
9572 // CHECK17-NEXT: [[TMP14:%.*]] = getelementptr inbounds [3 x i64], ptr [[DOTOFFLOAD_SIZES]], i32 0, i32 2
9573 // CHECK17-NEXT: store i64 [[TMP5]], ptr [[TMP14]], align 8
9574 // CHECK17-NEXT: [[TMP15:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i64 0, i64 2
9575 // CHECK17-NEXT: store ptr null, ptr [[TMP15]], align 8
9576 // CHECK17-NEXT: [[TMP16:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
9577 // CHECK17-NEXT: [[TMP17:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
9578 // CHECK17-NEXT: [[TMP18:%.*]] = getelementptr inbounds [3 x i64], ptr [[DOTOFFLOAD_SIZES]], i32 0, i32 0
9579 // CHECK17-NEXT: [[TMP19:%.*]] = load i32, ptr [[N]], align 4
9580 // CHECK17-NEXT: store i32 [[TMP19]], ptr [[DOTCAPTURE_EXPR_]], align 4
9581 // CHECK17-NEXT: [[TMP20:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4
9582 // CHECK17-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP20]], 0
9583 // CHECK17-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1
9584 // CHECK17-NEXT: [[SUB2:%.*]] = sub nsw i32 [[DIV]], 1
9585 // CHECK17-NEXT: store i32 [[SUB2]], ptr [[DOTCAPTURE_EXPR_1]], align 4
9586 // CHECK17-NEXT: [[TMP21:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4
9587 // CHECK17-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP21]], 1
9588 // CHECK17-NEXT: [[TMP22:%.*]] = zext i32 [[ADD]] to i64
9589 // CHECK17-NEXT: [[TMP23:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 0
9590 // CHECK17-NEXT: store i32 3, ptr [[TMP23]], align 4
9591 // CHECK17-NEXT: [[TMP24:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 1
9592 // CHECK17-NEXT: store i32 3, ptr [[TMP24]], align 4
9593 // CHECK17-NEXT: [[TMP25:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 2
9594 // CHECK17-NEXT: store ptr [[TMP16]], ptr [[TMP25]], align 8
9595 // CHECK17-NEXT: [[TMP26:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 3
9596 // CHECK17-NEXT: store ptr [[TMP17]], ptr [[TMP26]], align 8
9597 // CHECK17-NEXT: [[TMP27:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 4
9598 // CHECK17-NEXT: store ptr [[TMP18]], ptr [[TMP27]], align 8
9599 // CHECK17-NEXT: [[TMP28:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 5
9600 // CHECK17-NEXT: store ptr @.offload_maptypes, ptr [[TMP28]], align 8
9601 // CHECK17-NEXT: [[TMP29:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 6
9602 // CHECK17-NEXT: store ptr null, ptr [[TMP29]], align 8
9603 // CHECK17-NEXT: [[TMP30:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 7
9604 // CHECK17-NEXT: store ptr null, ptr [[TMP30]], align 8
9605 // CHECK17-NEXT: [[TMP31:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 8
9606 // CHECK17-NEXT: store i64 [[TMP22]], ptr [[TMP31]], align 8
9607 // CHECK17-NEXT: [[TMP32:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 9
9608 // CHECK17-NEXT: store i64 0, ptr [[TMP32]], align 8
9609 // CHECK17-NEXT: [[TMP33:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 10
9610 // CHECK17-NEXT: store [3 x i32] zeroinitializer, ptr [[TMP33]], align 4
9611 // CHECK17-NEXT: [[TMP34:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 11
9612 // CHECK17-NEXT: store [3 x i32] zeroinitializer, ptr [[TMP34]], align 4
9613 // CHECK17-NEXT: [[TMP35:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 12
9614 // CHECK17-NEXT: store i32 0, ptr [[TMP35]], align 4
9615 // CHECK17-NEXT: [[TMP36:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB3:[0-9]+]], i64 -1, i32 0, i32 0, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l148.region_id, ptr [[KERNEL_ARGS]])
9616 // CHECK17-NEXT: [[TMP37:%.*]] = icmp ne i32 [[TMP36]], 0
9617 // CHECK17-NEXT: br i1 [[TMP37]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
9618 // CHECK17: omp_offload.failed:
9619 // CHECK17-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l148(i64 [[TMP4]], i64 [[TMP1]], ptr [[VLA]]) #[[ATTR3:[0-9]+]]
9620 // CHECK17-NEXT: br label [[OMP_OFFLOAD_CONT]]
9621 // CHECK17: omp_offload.cont:
9622 // CHECK17-NEXT: [[TMP38:%.*]] = load i32, ptr [[N]], align 4
9623 // CHECK17-NEXT: store i32 [[TMP38]], ptr [[N_CASTED3]], align 4
9624 // CHECK17-NEXT: [[TMP39:%.*]] = load i64, ptr [[N_CASTED3]], align 8
9625 // CHECK17-NEXT: [[TMP40:%.*]] = mul nuw i64 [[TMP1]], 4
9626 // CHECK17-NEXT: call void @llvm.memcpy.p0.p0.i64(ptr align 8 [[DOTOFFLOAD_SIZES7]], ptr align 8 @.offload_sizes.1, i64 24, i1 false)
9627 // CHECK17-NEXT: [[TMP41:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_BASEPTRS4]], i32 0, i32 0
9628 // CHECK17-NEXT: store i64 [[TMP39]], ptr [[TMP41]], align 8
9629 // CHECK17-NEXT: [[TMP42:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_PTRS5]], i32 0, i32 0
9630 // CHECK17-NEXT: store i64 [[TMP39]], ptr [[TMP42]], align 8
9631 // CHECK17-NEXT: [[TMP43:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_MAPPERS6]], i64 0, i64 0
9632 // CHECK17-NEXT: store ptr null, ptr [[TMP43]], align 8
9633 // CHECK17-NEXT: [[TMP44:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_BASEPTRS4]], i32 0, i32 1
9634 // CHECK17-NEXT: store i64 [[TMP1]], ptr [[TMP44]], align 8
9635 // CHECK17-NEXT: [[TMP45:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_PTRS5]], i32 0, i32 1
9636 // CHECK17-NEXT: store i64 [[TMP1]], ptr [[TMP45]], align 8
9637 // CHECK17-NEXT: [[TMP46:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_MAPPERS6]], i64 0, i64 1
9638 // CHECK17-NEXT: store ptr null, ptr [[TMP46]], align 8
9639 // CHECK17-NEXT: [[TMP47:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_BASEPTRS4]], i32 0, i32 2
9640 // CHECK17-NEXT: store ptr [[VLA]], ptr [[TMP47]], align 8
9641 // CHECK17-NEXT: [[TMP48:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_PTRS5]], i32 0, i32 2
9642 // CHECK17-NEXT: store ptr [[VLA]], ptr [[TMP48]], align 8
9643 // CHECK17-NEXT: [[TMP49:%.*]] = getelementptr inbounds [3 x i64], ptr [[DOTOFFLOAD_SIZES7]], i32 0, i32 2
9644 // CHECK17-NEXT: store i64 [[TMP40]], ptr [[TMP49]], align 8
9645 // CHECK17-NEXT: [[TMP50:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_MAPPERS6]], i64 0, i64 2
9646 // CHECK17-NEXT: store ptr null, ptr [[TMP50]], align 8
9647 // CHECK17-NEXT: [[TMP51:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_BASEPTRS4]], i32 0, i32 0
9648 // CHECK17-NEXT: [[TMP52:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_PTRS5]], i32 0, i32 0
9649 // CHECK17-NEXT: [[TMP53:%.*]] = getelementptr inbounds [3 x i64], ptr [[DOTOFFLOAD_SIZES7]], i32 0, i32 0
9650 // CHECK17-NEXT: [[TMP54:%.*]] = load i32, ptr [[N]], align 4
9651 // CHECK17-NEXT: store i32 [[TMP54]], ptr [[DOTCAPTURE_EXPR_9]], align 4
9652 // CHECK17-NEXT: [[TMP55:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_9]], align 4
9653 // CHECK17-NEXT: [[SUB11:%.*]] = sub nsw i32 [[TMP55]], 0
9654 // CHECK17-NEXT: [[DIV12:%.*]] = sdiv i32 [[SUB11]], 1
9655 // CHECK17-NEXT: [[SUB13:%.*]] = sub nsw i32 [[DIV12]], 1
9656 // CHECK17-NEXT: store i32 [[SUB13]], ptr [[DOTCAPTURE_EXPR_10]], align 4
9657 // CHECK17-NEXT: [[TMP56:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_10]], align 4
9658 // CHECK17-NEXT: [[ADD14:%.*]] = add nsw i32 [[TMP56]], 1
9659 // CHECK17-NEXT: [[TMP57:%.*]] = zext i32 [[ADD14]] to i64
9660 // CHECK17-NEXT: [[TMP58:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS15]], i32 0, i32 0
9661 // CHECK17-NEXT: store i32 3, ptr [[TMP58]], align 4
9662 // CHECK17-NEXT: [[TMP59:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS15]], i32 0, i32 1
9663 // CHECK17-NEXT: store i32 3, ptr [[TMP59]], align 4
9664 // CHECK17-NEXT: [[TMP60:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS15]], i32 0, i32 2
9665 // CHECK17-NEXT: store ptr [[TMP51]], ptr [[TMP60]], align 8
9666 // CHECK17-NEXT: [[TMP61:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS15]], i32 0, i32 3
9667 // CHECK17-NEXT: store ptr [[TMP52]], ptr [[TMP61]], align 8
9668 // CHECK17-NEXT: [[TMP62:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS15]], i32 0, i32 4
9669 // CHECK17-NEXT: store ptr [[TMP53]], ptr [[TMP62]], align 8
9670 // CHECK17-NEXT: [[TMP63:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS15]], i32 0, i32 5
9671 // CHECK17-NEXT: store ptr @.offload_maptypes.2, ptr [[TMP63]], align 8
9672 // CHECK17-NEXT: [[TMP64:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS15]], i32 0, i32 6
9673 // CHECK17-NEXT: store ptr null, ptr [[TMP64]], align 8
9674 // CHECK17-NEXT: [[TMP65:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS15]], i32 0, i32 7
9675 // CHECK17-NEXT: store ptr null, ptr [[TMP65]], align 8
9676 // CHECK17-NEXT: [[TMP66:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS15]], i32 0, i32 8
9677 // CHECK17-NEXT: store i64 [[TMP57]], ptr [[TMP66]], align 8
9678 // CHECK17-NEXT: [[TMP67:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS15]], i32 0, i32 9
9679 // CHECK17-NEXT: store i64 0, ptr [[TMP67]], align 8
9680 // CHECK17-NEXT: [[TMP68:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS15]], i32 0, i32 10
9681 // CHECK17-NEXT: store [3 x i32] zeroinitializer, ptr [[TMP68]], align 4
9682 // CHECK17-NEXT: [[TMP69:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS15]], i32 0, i32 11
9683 // CHECK17-NEXT: store [3 x i32] zeroinitializer, ptr [[TMP69]], align 4
9684 // CHECK17-NEXT: [[TMP70:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS15]], i32 0, i32 12
9685 // CHECK17-NEXT: store i32 0, ptr [[TMP70]], align 4
9686 // CHECK17-NEXT: [[TMP71:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB3]], i64 -1, i32 0, i32 0, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l153.region_id, ptr [[KERNEL_ARGS15]])
9687 // CHECK17-NEXT: [[TMP72:%.*]] = icmp ne i32 [[TMP71]], 0
9688 // CHECK17-NEXT: br i1 [[TMP72]], label [[OMP_OFFLOAD_FAILED16:%.*]], label [[OMP_OFFLOAD_CONT17:%.*]]
9689 // CHECK17: omp_offload.failed16:
9690 // CHECK17-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l153(i64 [[TMP39]], i64 [[TMP1]], ptr [[VLA]]) #[[ATTR3]]
9691 // CHECK17-NEXT: br label [[OMP_OFFLOAD_CONT17]]
9692 // CHECK17: omp_offload.cont17:
9693 // CHECK17-NEXT: [[TMP73:%.*]] = load i32, ptr [[M]], align 4
9694 // CHECK17-NEXT: store i32 [[TMP73]], ptr [[M_CASTED]], align 4
9695 // CHECK17-NEXT: [[TMP74:%.*]] = load i64, ptr [[M_CASTED]], align 8
9696 // CHECK17-NEXT: [[TMP75:%.*]] = load i32, ptr [[N]], align 4
9697 // CHECK17-NEXT: store i32 [[TMP75]], ptr [[N_CASTED18]], align 4
9698 // CHECK17-NEXT: [[TMP76:%.*]] = load i64, ptr [[N_CASTED18]], align 8
9699 // CHECK17-NEXT: [[TMP77:%.*]] = mul nuw i64 [[TMP1]], 4
9700 // CHECK17-NEXT: call void @llvm.memcpy.p0.p0.i64(ptr align 8 [[DOTOFFLOAD_SIZES22]], ptr align 8 @.offload_sizes.3, i64 32, i1 false)
9701 // CHECK17-NEXT: [[TMP78:%.*]] = getelementptr inbounds [4 x ptr], ptr [[DOTOFFLOAD_BASEPTRS19]], i32 0, i32 0
9702 // CHECK17-NEXT: store i64 [[TMP74]], ptr [[TMP78]], align 8
9703 // CHECK17-NEXT: [[TMP79:%.*]] = getelementptr inbounds [4 x ptr], ptr [[DOTOFFLOAD_PTRS20]], i32 0, i32 0
9704 // CHECK17-NEXT: store i64 [[TMP74]], ptr [[TMP79]], align 8
9705 // CHECK17-NEXT: [[TMP80:%.*]] = getelementptr inbounds [4 x ptr], ptr [[DOTOFFLOAD_MAPPERS21]], i64 0, i64 0
9706 // CHECK17-NEXT: store ptr null, ptr [[TMP80]], align 8
9707 // CHECK17-NEXT: [[TMP81:%.*]] = getelementptr inbounds [4 x ptr], ptr [[DOTOFFLOAD_BASEPTRS19]], i32 0, i32 1
9708 // CHECK17-NEXT: store i64 [[TMP76]], ptr [[TMP81]], align 8
9709 // CHECK17-NEXT: [[TMP82:%.*]] = getelementptr inbounds [4 x ptr], ptr [[DOTOFFLOAD_PTRS20]], i32 0, i32 1
9710 // CHECK17-NEXT: store i64 [[TMP76]], ptr [[TMP82]], align 8
9711 // CHECK17-NEXT: [[TMP83:%.*]] = getelementptr inbounds [4 x ptr], ptr [[DOTOFFLOAD_MAPPERS21]], i64 0, i64 1
9712 // CHECK17-NEXT: store ptr null, ptr [[TMP83]], align 8
9713 // CHECK17-NEXT: [[TMP84:%.*]] = getelementptr inbounds [4 x ptr], ptr [[DOTOFFLOAD_BASEPTRS19]], i32 0, i32 2
9714 // CHECK17-NEXT: store i64 [[TMP1]], ptr [[TMP84]], align 8
9715 // CHECK17-NEXT: [[TMP85:%.*]] = getelementptr inbounds [4 x ptr], ptr [[DOTOFFLOAD_PTRS20]], i32 0, i32 2
9716 // CHECK17-NEXT: store i64 [[TMP1]], ptr [[TMP85]], align 8
9717 // CHECK17-NEXT: [[TMP86:%.*]] = getelementptr inbounds [4 x ptr], ptr [[DOTOFFLOAD_MAPPERS21]], i64 0, i64 2
9718 // CHECK17-NEXT: store ptr null, ptr [[TMP86]], align 8
9719 // CHECK17-NEXT: [[TMP87:%.*]] = getelementptr inbounds [4 x ptr], ptr [[DOTOFFLOAD_BASEPTRS19]], i32 0, i32 3
9720 // CHECK17-NEXT: store ptr [[VLA]], ptr [[TMP87]], align 8
9721 // CHECK17-NEXT: [[TMP88:%.*]] = getelementptr inbounds [4 x ptr], ptr [[DOTOFFLOAD_PTRS20]], i32 0, i32 3
9722 // CHECK17-NEXT: store ptr [[VLA]], ptr [[TMP88]], align 8
9723 // CHECK17-NEXT: [[TMP89:%.*]] = getelementptr inbounds [4 x i64], ptr [[DOTOFFLOAD_SIZES22]], i32 0, i32 3
9724 // CHECK17-NEXT: store i64 [[TMP77]], ptr [[TMP89]], align 8
9725 // CHECK17-NEXT: [[TMP90:%.*]] = getelementptr inbounds [4 x ptr], ptr [[DOTOFFLOAD_MAPPERS21]], i64 0, i64 3
9726 // CHECK17-NEXT: store ptr null, ptr [[TMP90]], align 8
9727 // CHECK17-NEXT: [[TMP91:%.*]] = getelementptr inbounds [4 x ptr], ptr [[DOTOFFLOAD_BASEPTRS19]], i32 0, i32 0
9728 // CHECK17-NEXT: [[TMP92:%.*]] = getelementptr inbounds [4 x ptr], ptr [[DOTOFFLOAD_PTRS20]], i32 0, i32 0
9729 // CHECK17-NEXT: [[TMP93:%.*]] = getelementptr inbounds [4 x i64], ptr [[DOTOFFLOAD_SIZES22]], i32 0, i32 0
9730 // CHECK17-NEXT: [[TMP94:%.*]] = load i32, ptr [[N]], align 4
9731 // CHECK17-NEXT: store i32 [[TMP94]], ptr [[DOTCAPTURE_EXPR_24]], align 4
9732 // CHECK17-NEXT: [[TMP95:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_24]], align 4
9733 // CHECK17-NEXT: [[SUB26:%.*]] = sub nsw i32 [[TMP95]], 0
9734 // CHECK17-NEXT: [[DIV27:%.*]] = sdiv i32 [[SUB26]], 1
9735 // CHECK17-NEXT: [[SUB28:%.*]] = sub nsw i32 [[DIV27]], 1
9736 // CHECK17-NEXT: store i32 [[SUB28]], ptr [[DOTCAPTURE_EXPR_25]], align 4
9737 // CHECK17-NEXT: [[TMP96:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_25]], align 4
9738 // CHECK17-NEXT: [[ADD29:%.*]] = add nsw i32 [[TMP96]], 1
9739 // CHECK17-NEXT: [[TMP97:%.*]] = zext i32 [[ADD29]] to i64
9740 // CHECK17-NEXT: [[TMP98:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS30]], i32 0, i32 0
9741 // CHECK17-NEXT: store i32 3, ptr [[TMP98]], align 4
9742 // CHECK17-NEXT: [[TMP99:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS30]], i32 0, i32 1
9743 // CHECK17-NEXT: store i32 4, ptr [[TMP99]], align 4
9744 // CHECK17-NEXT: [[TMP100:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS30]], i32 0, i32 2
9745 // CHECK17-NEXT: store ptr [[TMP91]], ptr [[TMP100]], align 8
9746 // CHECK17-NEXT: [[TMP101:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS30]], i32 0, i32 3
9747 // CHECK17-NEXT: store ptr [[TMP92]], ptr [[TMP101]], align 8
9748 // CHECK17-NEXT: [[TMP102:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS30]], i32 0, i32 4
9749 // CHECK17-NEXT: store ptr [[TMP93]], ptr [[TMP102]], align 8
9750 // CHECK17-NEXT: [[TMP103:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS30]], i32 0, i32 5
9751 // CHECK17-NEXT: store ptr @.offload_maptypes.4, ptr [[TMP103]], align 8
9752 // CHECK17-NEXT: [[TMP104:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS30]], i32 0, i32 6
9753 // CHECK17-NEXT: store ptr null, ptr [[TMP104]], align 8
9754 // CHECK17-NEXT: [[TMP105:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS30]], i32 0, i32 7
9755 // CHECK17-NEXT: store ptr null, ptr [[TMP105]], align 8
9756 // CHECK17-NEXT: [[TMP106:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS30]], i32 0, i32 8
9757 // CHECK17-NEXT: store i64 [[TMP97]], ptr [[TMP106]], align 8
9758 // CHECK17-NEXT: [[TMP107:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS30]], i32 0, i32 9
9759 // CHECK17-NEXT: store i64 0, ptr [[TMP107]], align 8
9760 // CHECK17-NEXT: [[TMP108:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS30]], i32 0, i32 10
9761 // CHECK17-NEXT: store [3 x i32] zeroinitializer, ptr [[TMP108]], align 4
9762 // CHECK17-NEXT: [[TMP109:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS30]], i32 0, i32 11
9763 // CHECK17-NEXT: store [3 x i32] zeroinitializer, ptr [[TMP109]], align 4
9764 // CHECK17-NEXT: [[TMP110:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS30]], i32 0, i32 12
9765 // CHECK17-NEXT: store i32 0, ptr [[TMP110]], align 4
9766 // CHECK17-NEXT: [[TMP111:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB3]], i64 -1, i32 0, i32 0, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l158.region_id, ptr [[KERNEL_ARGS30]])
9767 // CHECK17-NEXT: [[TMP112:%.*]] = icmp ne i32 [[TMP111]], 0
9768 // CHECK17-NEXT: br i1 [[TMP112]], label [[OMP_OFFLOAD_FAILED31:%.*]], label [[OMP_OFFLOAD_CONT32:%.*]]
9769 // CHECK17: omp_offload.failed31:
9770 // CHECK17-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l158(i64 [[TMP74]], i64 [[TMP76]], i64 [[TMP1]], ptr [[VLA]]) #[[ATTR3]]
9771 // CHECK17-NEXT: br label [[OMP_OFFLOAD_CONT32]]
9772 // CHECK17: omp_offload.cont32:
9773 // CHECK17-NEXT: [[TMP113:%.*]] = load i32, ptr [[N]], align 4
9774 // CHECK17-NEXT: store i32 [[TMP113]], ptr [[N_CASTED33]], align 4
9775 // CHECK17-NEXT: [[TMP114:%.*]] = load i64, ptr [[N_CASTED33]], align 8
9776 // CHECK17-NEXT: [[TMP115:%.*]] = mul nuw i64 [[TMP1]], 4
9777 // CHECK17-NEXT: call void @llvm.memcpy.p0.p0.i64(ptr align 8 [[DOTOFFLOAD_SIZES37]], ptr align 8 @.offload_sizes.5, i64 24, i1 false)
9778 // CHECK17-NEXT: [[TMP116:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_BASEPTRS34]], i32 0, i32 0
9779 // CHECK17-NEXT: store i64 [[TMP114]], ptr [[TMP116]], align 8
9780 // CHECK17-NEXT: [[TMP117:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_PTRS35]], i32 0, i32 0
9781 // CHECK17-NEXT: store i64 [[TMP114]], ptr [[TMP117]], align 8
9782 // CHECK17-NEXT: [[TMP118:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_MAPPERS36]], i64 0, i64 0
9783 // CHECK17-NEXT: store ptr null, ptr [[TMP118]], align 8
9784 // CHECK17-NEXT: [[TMP119:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_BASEPTRS34]], i32 0, i32 1
9785 // CHECK17-NEXT: store i64 [[TMP1]], ptr [[TMP119]], align 8
9786 // CHECK17-NEXT: [[TMP120:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_PTRS35]], i32 0, i32 1
9787 // CHECK17-NEXT: store i64 [[TMP1]], ptr [[TMP120]], align 8
9788 // CHECK17-NEXT: [[TMP121:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_MAPPERS36]], i64 0, i64 1
9789 // CHECK17-NEXT: store ptr null, ptr [[TMP121]], align 8
9790 // CHECK17-NEXT: [[TMP122:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_BASEPTRS34]], i32 0, i32 2
9791 // CHECK17-NEXT: store ptr [[VLA]], ptr [[TMP122]], align 8
9792 // CHECK17-NEXT: [[TMP123:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_PTRS35]], i32 0, i32 2
9793 // CHECK17-NEXT: store ptr [[VLA]], ptr [[TMP123]], align 8
9794 // CHECK17-NEXT: [[TMP124:%.*]] = getelementptr inbounds [3 x i64], ptr [[DOTOFFLOAD_SIZES37]], i32 0, i32 2
9795 // CHECK17-NEXT: store i64 [[TMP115]], ptr [[TMP124]], align 8
9796 // CHECK17-NEXT: [[TMP125:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_MAPPERS36]], i64 0, i64 2
9797 // CHECK17-NEXT: store ptr null, ptr [[TMP125]], align 8
9798 // CHECK17-NEXT: [[TMP126:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_BASEPTRS34]], i32 0, i32 0
9799 // CHECK17-NEXT: [[TMP127:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_PTRS35]], i32 0, i32 0
9800 // CHECK17-NEXT: [[TMP128:%.*]] = getelementptr inbounds [3 x i64], ptr [[DOTOFFLOAD_SIZES37]], i32 0, i32 0
9801 // CHECK17-NEXT: [[TMP129:%.*]] = load i32, ptr [[N]], align 4
9802 // CHECK17-NEXT: store i32 [[TMP129]], ptr [[DOTCAPTURE_EXPR_39]], align 4
9803 // CHECK17-NEXT: [[TMP130:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_39]], align 4
9804 // CHECK17-NEXT: [[SUB41:%.*]] = sub nsw i32 [[TMP130]], 0
9805 // CHECK17-NEXT: [[DIV42:%.*]] = sdiv i32 [[SUB41]], 1
9806 // CHECK17-NEXT: [[SUB43:%.*]] = sub nsw i32 [[DIV42]], 1
9807 // CHECK17-NEXT: store i32 [[SUB43]], ptr [[DOTCAPTURE_EXPR_40]], align 4
9808 // CHECK17-NEXT: [[TMP131:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_40]], align 4
9809 // CHECK17-NEXT: [[ADD44:%.*]] = add nsw i32 [[TMP131]], 1
9810 // CHECK17-NEXT: [[TMP132:%.*]] = zext i32 [[ADD44]] to i64
9811 // CHECK17-NEXT: [[TMP133:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS45]], i32 0, i32 0
9812 // CHECK17-NEXT: store i32 3, ptr [[TMP133]], align 4
9813 // CHECK17-NEXT: [[TMP134:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS45]], i32 0, i32 1
9814 // CHECK17-NEXT: store i32 3, ptr [[TMP134]], align 4
9815 // CHECK17-NEXT: [[TMP135:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS45]], i32 0, i32 2
9816 // CHECK17-NEXT: store ptr [[TMP126]], ptr [[TMP135]], align 8
9817 // CHECK17-NEXT: [[TMP136:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS45]], i32 0, i32 3
9818 // CHECK17-NEXT: store ptr [[TMP127]], ptr [[TMP136]], align 8
9819 // CHECK17-NEXT: [[TMP137:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS45]], i32 0, i32 4
9820 // CHECK17-NEXT: store ptr [[TMP128]], ptr [[TMP137]], align 8
9821 // CHECK17-NEXT: [[TMP138:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS45]], i32 0, i32 5
9822 // CHECK17-NEXT: store ptr @.offload_maptypes.6, ptr [[TMP138]], align 8
9823 // CHECK17-NEXT: [[TMP139:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS45]], i32 0, i32 6
9824 // CHECK17-NEXT: store ptr null, ptr [[TMP139]], align 8
9825 // CHECK17-NEXT: [[TMP140:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS45]], i32 0, i32 7
9826 // CHECK17-NEXT: store ptr null, ptr [[TMP140]], align 8
9827 // CHECK17-NEXT: [[TMP141:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS45]], i32 0, i32 8
9828 // CHECK17-NEXT: store i64 [[TMP132]], ptr [[TMP141]], align 8
9829 // CHECK17-NEXT: [[TMP142:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS45]], i32 0, i32 9
9830 // CHECK17-NEXT: store i64 0, ptr [[TMP142]], align 8
9831 // CHECK17-NEXT: [[TMP143:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS45]], i32 0, i32 10
9832 // CHECK17-NEXT: store [3 x i32] zeroinitializer, ptr [[TMP143]], align 4
9833 // CHECK17-NEXT: [[TMP144:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS45]], i32 0, i32 11
9834 // CHECK17-NEXT: store [3 x i32] zeroinitializer, ptr [[TMP144]], align 4
9835 // CHECK17-NEXT: [[TMP145:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS45]], i32 0, i32 12
9836 // CHECK17-NEXT: store i32 0, ptr [[TMP145]], align 4
9837 // CHECK17-NEXT: [[TMP146:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB3]], i64 -1, i32 0, i32 0, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l163.region_id, ptr [[KERNEL_ARGS45]])
9838 // CHECK17-NEXT: [[TMP147:%.*]] = icmp ne i32 [[TMP146]], 0
9839 // CHECK17-NEXT: br i1 [[TMP147]], label [[OMP_OFFLOAD_FAILED46:%.*]], label [[OMP_OFFLOAD_CONT47:%.*]]
9840 // CHECK17: omp_offload.failed46:
9841 // CHECK17-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l163(i64 [[TMP114]], i64 [[TMP1]], ptr [[VLA]]) #[[ATTR3]]
9842 // CHECK17-NEXT: br label [[OMP_OFFLOAD_CONT47]]
9843 // CHECK17: omp_offload.cont47:
9844 // CHECK17-NEXT: [[TMP148:%.*]] = load i32, ptr [[M]], align 4
9845 // CHECK17-NEXT: store i32 [[TMP148]], ptr [[M_CASTED48]], align 4
9846 // CHECK17-NEXT: [[TMP149:%.*]] = load i64, ptr [[M_CASTED48]], align 8
9847 // CHECK17-NEXT: [[TMP150:%.*]] = load i32, ptr [[N]], align 4
9848 // CHECK17-NEXT: store i32 [[TMP150]], ptr [[N_CASTED49]], align 4
9849 // CHECK17-NEXT: [[TMP151:%.*]] = load i64, ptr [[N_CASTED49]], align 8
9850 // CHECK17-NEXT: [[TMP152:%.*]] = mul nuw i64 [[TMP1]], 4
9851 // CHECK17-NEXT: call void @llvm.memcpy.p0.p0.i64(ptr align 8 [[DOTOFFLOAD_SIZES53]], ptr align 8 @.offload_sizes.7, i64 32, i1 false)
9852 // CHECK17-NEXT: [[TMP153:%.*]] = getelementptr inbounds [4 x ptr], ptr [[DOTOFFLOAD_BASEPTRS50]], i32 0, i32 0
9853 // CHECK17-NEXT: store i64 [[TMP149]], ptr [[TMP153]], align 8
9854 // CHECK17-NEXT: [[TMP154:%.*]] = getelementptr inbounds [4 x ptr], ptr [[DOTOFFLOAD_PTRS51]], i32 0, i32 0
9855 // CHECK17-NEXT: store i64 [[TMP149]], ptr [[TMP154]], align 8
9856 // CHECK17-NEXT: [[TMP155:%.*]] = getelementptr inbounds [4 x ptr], ptr [[DOTOFFLOAD_MAPPERS52]], i64 0, i64 0
9857 // CHECK17-NEXT: store ptr null, ptr [[TMP155]], align 8
9858 // CHECK17-NEXT: [[TMP156:%.*]] = getelementptr inbounds [4 x ptr], ptr [[DOTOFFLOAD_BASEPTRS50]], i32 0, i32 1
9859 // CHECK17-NEXT: store i64 [[TMP151]], ptr [[TMP156]], align 8
9860 // CHECK17-NEXT: [[TMP157:%.*]] = getelementptr inbounds [4 x ptr], ptr [[DOTOFFLOAD_PTRS51]], i32 0, i32 1
9861 // CHECK17-NEXT: store i64 [[TMP151]], ptr [[TMP157]], align 8
9862 // CHECK17-NEXT: [[TMP158:%.*]] = getelementptr inbounds [4 x ptr], ptr [[DOTOFFLOAD_MAPPERS52]], i64 0, i64 1
9863 // CHECK17-NEXT: store ptr null, ptr [[TMP158]], align 8
9864 // CHECK17-NEXT: [[TMP159:%.*]] = getelementptr inbounds [4 x ptr], ptr [[DOTOFFLOAD_BASEPTRS50]], i32 0, i32 2
9865 // CHECK17-NEXT: store i64 [[TMP1]], ptr [[TMP159]], align 8
9866 // CHECK17-NEXT: [[TMP160:%.*]] = getelementptr inbounds [4 x ptr], ptr [[DOTOFFLOAD_PTRS51]], i32 0, i32 2
9867 // CHECK17-NEXT: store i64 [[TMP1]], ptr [[TMP160]], align 8
9868 // CHECK17-NEXT: [[TMP161:%.*]] = getelementptr inbounds [4 x ptr], ptr [[DOTOFFLOAD_MAPPERS52]], i64 0, i64 2
9869 // CHECK17-NEXT: store ptr null, ptr [[TMP161]], align 8
9870 // CHECK17-NEXT: [[TMP162:%.*]] = getelementptr inbounds [4 x ptr], ptr [[DOTOFFLOAD_BASEPTRS50]], i32 0, i32 3
9871 // CHECK17-NEXT: store ptr [[VLA]], ptr [[TMP162]], align 8
9872 // CHECK17-NEXT: [[TMP163:%.*]] = getelementptr inbounds [4 x ptr], ptr [[DOTOFFLOAD_PTRS51]], i32 0, i32 3
9873 // CHECK17-NEXT: store ptr [[VLA]], ptr [[TMP163]], align 8
9874 // CHECK17-NEXT: [[TMP164:%.*]] = getelementptr inbounds [4 x i64], ptr [[DOTOFFLOAD_SIZES53]], i32 0, i32 3
9875 // CHECK17-NEXT: store i64 [[TMP152]], ptr [[TMP164]], align 8
9876 // CHECK17-NEXT: [[TMP165:%.*]] = getelementptr inbounds [4 x ptr], ptr [[DOTOFFLOAD_MAPPERS52]], i64 0, i64 3
9877 // CHECK17-NEXT: store ptr null, ptr [[TMP165]], align 8
9878 // CHECK17-NEXT: [[TMP166:%.*]] = getelementptr inbounds [4 x ptr], ptr [[DOTOFFLOAD_BASEPTRS50]], i32 0, i32 0
9879 // CHECK17-NEXT: [[TMP167:%.*]] = getelementptr inbounds [4 x ptr], ptr [[DOTOFFLOAD_PTRS51]], i32 0, i32 0
9880 // CHECK17-NEXT: [[TMP168:%.*]] = getelementptr inbounds [4 x i64], ptr [[DOTOFFLOAD_SIZES53]], i32 0, i32 0
9881 // CHECK17-NEXT: [[TMP169:%.*]] = load i32, ptr [[N]], align 4
9882 // CHECK17-NEXT: store i32 [[TMP169]], ptr [[DOTCAPTURE_EXPR_55]], align 4
9883 // CHECK17-NEXT: [[TMP170:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_55]], align 4
9884 // CHECK17-NEXT: [[SUB57:%.*]] = sub nsw i32 [[TMP170]], 0
9885 // CHECK17-NEXT: [[DIV58:%.*]] = sdiv i32 [[SUB57]], 1
9886 // CHECK17-NEXT: [[SUB59:%.*]] = sub nsw i32 [[DIV58]], 1
9887 // CHECK17-NEXT: store i32 [[SUB59]], ptr [[DOTCAPTURE_EXPR_56]], align 4
9888 // CHECK17-NEXT: [[TMP171:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_56]], align 4
9889 // CHECK17-NEXT: [[ADD60:%.*]] = add nsw i32 [[TMP171]], 1
9890 // CHECK17-NEXT: [[TMP172:%.*]] = zext i32 [[ADD60]] to i64
9891 // CHECK17-NEXT: [[TMP173:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS61]], i32 0, i32 0
9892 // CHECK17-NEXT: store i32 3, ptr [[TMP173]], align 4
9893 // CHECK17-NEXT: [[TMP174:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS61]], i32 0, i32 1
9894 // CHECK17-NEXT: store i32 4, ptr [[TMP174]], align 4
9895 // CHECK17-NEXT: [[TMP175:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS61]], i32 0, i32 2
9896 // CHECK17-NEXT: store ptr [[TMP166]], ptr [[TMP175]], align 8
9897 // CHECK17-NEXT: [[TMP176:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS61]], i32 0, i32 3
9898 // CHECK17-NEXT: store ptr [[TMP167]], ptr [[TMP176]], align 8
9899 // CHECK17-NEXT: [[TMP177:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS61]], i32 0, i32 4
9900 // CHECK17-NEXT: store ptr [[TMP168]], ptr [[TMP177]], align 8
9901 // CHECK17-NEXT: [[TMP178:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS61]], i32 0, i32 5
9902 // CHECK17-NEXT: store ptr @.offload_maptypes.8, ptr [[TMP178]], align 8
9903 // CHECK17-NEXT: [[TMP179:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS61]], i32 0, i32 6
9904 // CHECK17-NEXT: store ptr null, ptr [[TMP179]], align 8
9905 // CHECK17-NEXT: [[TMP180:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS61]], i32 0, i32 7
9906 // CHECK17-NEXT: store ptr null, ptr [[TMP180]], align 8
9907 // CHECK17-NEXT: [[TMP181:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS61]], i32 0, i32 8
9908 // CHECK17-NEXT: store i64 [[TMP172]], ptr [[TMP181]], align 8
9909 // CHECK17-NEXT: [[TMP182:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS61]], i32 0, i32 9
9910 // CHECK17-NEXT: store i64 0, ptr [[TMP182]], align 8
9911 // CHECK17-NEXT: [[TMP183:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS61]], i32 0, i32 10
9912 // CHECK17-NEXT: store [3 x i32] zeroinitializer, ptr [[TMP183]], align 4
9913 // CHECK17-NEXT: [[TMP184:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS61]], i32 0, i32 11
9914 // CHECK17-NEXT: store [3 x i32] zeroinitializer, ptr [[TMP184]], align 4
9915 // CHECK17-NEXT: [[TMP185:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS61]], i32 0, i32 12
9916 // CHECK17-NEXT: store i32 0, ptr [[TMP185]], align 4
9917 // CHECK17-NEXT: [[TMP186:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB3]], i64 -1, i32 0, i32 0, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l168.region_id, ptr [[KERNEL_ARGS61]])
9918 // CHECK17-NEXT: [[TMP187:%.*]] = icmp ne i32 [[TMP186]], 0
9919 // CHECK17-NEXT: br i1 [[TMP187]], label [[OMP_OFFLOAD_FAILED62:%.*]], label [[OMP_OFFLOAD_CONT63:%.*]]
9920 // CHECK17: omp_offload.failed62:
9921 // CHECK17-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l168(i64 [[TMP149]], i64 [[TMP151]], i64 [[TMP1]], ptr [[VLA]]) #[[ATTR3]]
9922 // CHECK17-NEXT: br label [[OMP_OFFLOAD_CONT63]]
9923 // CHECK17: omp_offload.cont63:
9924 // CHECK17-NEXT: [[TMP188:%.*]] = load i32, ptr [[ARGC_ADDR]], align 4
9925 // CHECK17-NEXT: [[CALL:%.*]] = call noundef signext i32 @_Z5tmainIiLi10EEiT_(i32 noundef signext [[TMP188]])
9926 // CHECK17-NEXT: store i32 [[CALL]], ptr [[RETVAL]], align 4
9927 // CHECK17-NEXT: [[TMP189:%.*]] = load ptr, ptr [[SAVED_STACK]], align 8
9928 // CHECK17-NEXT: call void @llvm.stackrestore.p0(ptr [[TMP189]])
9929 // CHECK17-NEXT: [[TMP190:%.*]] = load i32, ptr [[RETVAL]], align 4
9930 // CHECK17-NEXT: ret i32 [[TMP190]]
9933 // CHECK17-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l148
9934 // CHECK17-SAME: (i64 noundef [[N:%.*]], i64 noundef [[VLA:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[A:%.*]]) #[[ATTR2:[0-9]+]] {
9935 // CHECK17-NEXT: entry:
9936 // CHECK17-NEXT: [[N_ADDR:%.*]] = alloca i64, align 8
9937 // CHECK17-NEXT: [[VLA_ADDR:%.*]] = alloca i64, align 8
9938 // CHECK17-NEXT: [[A_ADDR:%.*]] = alloca ptr, align 8
9939 // CHECK17-NEXT: store i64 [[N]], ptr [[N_ADDR]], align 8
9940 // CHECK17-NEXT: store i64 [[VLA]], ptr [[VLA_ADDR]], align 8
9941 // CHECK17-NEXT: store ptr [[A]], ptr [[A_ADDR]], align 8
9942 // CHECK17-NEXT: [[TMP0:%.*]] = load i64, ptr [[VLA_ADDR]], align 8
9943 // CHECK17-NEXT: [[TMP1:%.*]] = load ptr, ptr [[A_ADDR]], align 8
9944 // CHECK17-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_teams(ptr @[[GLOB3]], i32 3, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l148.omp_outlined, ptr [[N_ADDR]], i64 [[TMP0]], ptr [[TMP1]])
9945 // CHECK17-NEXT: ret void
9948 // CHECK17-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l148.omp_outlined
9949 // CHECK17-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[N:%.*]], i64 noundef [[VLA:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[A:%.*]]) #[[ATTR2]] {
9950 // CHECK17-NEXT: entry:
9951 // CHECK17-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 8
9952 // CHECK17-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 8
9953 // CHECK17-NEXT: [[N_ADDR:%.*]] = alloca ptr, align 8
9954 // CHECK17-NEXT: [[VLA_ADDR:%.*]] = alloca i64, align 8
9955 // CHECK17-NEXT: [[A_ADDR:%.*]] = alloca ptr, align 8
9956 // CHECK17-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
9957 // CHECK17-NEXT: [[TMP:%.*]] = alloca i32, align 4
9958 // CHECK17-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4
9959 // CHECK17-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4
9960 // CHECK17-NEXT: [[I:%.*]] = alloca i32, align 4
9961 // CHECK17-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4
9962 // CHECK17-NEXT: [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4
9963 // CHECK17-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
9964 // CHECK17-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
9965 // CHECK17-NEXT: [[I3:%.*]] = alloca i32, align 4
9966 // CHECK17-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 8
9967 // CHECK17-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 8
9968 // CHECK17-NEXT: store ptr [[N]], ptr [[N_ADDR]], align 8
9969 // CHECK17-NEXT: store i64 [[VLA]], ptr [[VLA_ADDR]], align 8
9970 // CHECK17-NEXT: store ptr [[A]], ptr [[A_ADDR]], align 8
9971 // CHECK17-NEXT: [[TMP0:%.*]] = load ptr, ptr [[N_ADDR]], align 8
9972 // CHECK17-NEXT: [[TMP1:%.*]] = load i64, ptr [[VLA_ADDR]], align 8
9973 // CHECK17-NEXT: [[TMP2:%.*]] = load ptr, ptr [[A_ADDR]], align 8
9974 // CHECK17-NEXT: [[TMP3:%.*]] = load i32, ptr [[TMP0]], align 4
9975 // CHECK17-NEXT: store i32 [[TMP3]], ptr [[DOTCAPTURE_EXPR_]], align 4
9976 // CHECK17-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4
9977 // CHECK17-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP4]], 0
9978 // CHECK17-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1
9979 // CHECK17-NEXT: [[SUB2:%.*]] = sub nsw i32 [[DIV]], 1
9980 // CHECK17-NEXT: store i32 [[SUB2]], ptr [[DOTCAPTURE_EXPR_1]], align 4
9981 // CHECK17-NEXT: store i32 0, ptr [[I]], align 4
9982 // CHECK17-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4
9983 // CHECK17-NEXT: [[CMP:%.*]] = icmp slt i32 0, [[TMP5]]
9984 // CHECK17-NEXT: br i1 [[CMP]], label [[OMP_PRECOND_THEN:%.*]], label [[OMP_PRECOND_END:%.*]]
9985 // CHECK17: omp.precond.then:
9986 // CHECK17-NEXT: store i32 0, ptr [[DOTOMP_COMB_LB]], align 4
9987 // CHECK17-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4
9988 // CHECK17-NEXT: store i32 [[TMP6]], ptr [[DOTOMP_COMB_UB]], align 4
9989 // CHECK17-NEXT: store i32 1, ptr [[DOTOMP_STRIDE]], align 4
9990 // CHECK17-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
9991 // CHECK17-NEXT: [[TMP7:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
9992 // CHECK17-NEXT: [[TMP8:%.*]] = load i32, ptr [[TMP7]], align 4
9993 // CHECK17-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB1:[0-9]+]], i32 [[TMP8]], i32 92, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
9994 // CHECK17-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
9995 // CHECK17-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4
9996 // CHECK17-NEXT: [[CMP4:%.*]] = icmp sgt i32 [[TMP9]], [[TMP10]]
9997 // CHECK17-NEXT: br i1 [[CMP4]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
9998 // CHECK17: cond.true:
9999 // CHECK17-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4
10000 // CHECK17-NEXT: br label [[COND_END:%.*]]
10001 // CHECK17: cond.false:
10002 // CHECK17-NEXT: [[TMP12:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
10003 // CHECK17-NEXT: br label [[COND_END]]
10004 // CHECK17: cond.end:
10005 // CHECK17-NEXT: [[COND:%.*]] = phi i32 [ [[TMP11]], [[COND_TRUE]] ], [ [[TMP12]], [[COND_FALSE]] ]
10006 // CHECK17-NEXT: store i32 [[COND]], ptr [[DOTOMP_COMB_UB]], align 4
10007 // CHECK17-NEXT: [[TMP13:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
10008 // CHECK17-NEXT: store i32 [[TMP13]], ptr [[DOTOMP_IV]], align 4
10009 // CHECK17-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
10010 // CHECK17: omp.inner.for.cond:
10011 // CHECK17-NEXT: [[TMP14:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
10012 // CHECK17-NEXT: [[TMP15:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
10013 // CHECK17-NEXT: [[CMP5:%.*]] = icmp sle i32 [[TMP14]], [[TMP15]]
10014 // CHECK17-NEXT: br i1 [[CMP5]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
10015 // CHECK17: omp.inner.for.body:
10016 // CHECK17-NEXT: [[TMP16:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
10017 // CHECK17-NEXT: [[TMP17:%.*]] = zext i32 [[TMP16]] to i64
10018 // CHECK17-NEXT: [[TMP18:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
10019 // CHECK17-NEXT: [[TMP19:%.*]] = zext i32 [[TMP18]] to i64
10020 // CHECK17-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_call(ptr @[[GLOB3]], i32 5, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l148.omp_outlined.omp_outlined, i64 [[TMP17]], i64 [[TMP19]], ptr [[TMP0]], i64 [[TMP1]], ptr [[TMP2]])
10021 // CHECK17-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
10022 // CHECK17: omp.inner.for.inc:
10023 // CHECK17-NEXT: [[TMP20:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
10024 // CHECK17-NEXT: [[TMP21:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4
10025 // CHECK17-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP20]], [[TMP21]]
10026 // CHECK17-NEXT: store i32 [[ADD]], ptr [[DOTOMP_IV]], align 4
10027 // CHECK17-NEXT: br label [[OMP_INNER_FOR_COND]]
10028 // CHECK17: omp.inner.for.end:
10029 // CHECK17-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
10030 // CHECK17: omp.loop.exit:
10031 // CHECK17-NEXT: [[TMP22:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
10032 // CHECK17-NEXT: [[TMP23:%.*]] = load i32, ptr [[TMP22]], align 4
10033 // CHECK17-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB1]], i32 [[TMP23]])
10034 // CHECK17-NEXT: br label [[OMP_PRECOND_END]]
10035 // CHECK17: omp.precond.end:
10036 // CHECK17-NEXT: ret void
10039 // CHECK17-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l148.omp_outlined.omp_outlined
10040 // CHECK17-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], i64 noundef [[DOTPREVIOUS_LB_:%.*]], i64 noundef [[DOTPREVIOUS_UB_:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[N:%.*]], i64 noundef [[VLA:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[A:%.*]]) #[[ATTR2]] {
10041 // CHECK17-NEXT: entry:
10042 // CHECK17-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 8
10043 // CHECK17-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 8
10044 // CHECK17-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i64, align 8
10045 // CHECK17-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i64, align 8
10046 // CHECK17-NEXT: [[N_ADDR:%.*]] = alloca ptr, align 8
10047 // CHECK17-NEXT: [[VLA_ADDR:%.*]] = alloca i64, align 8
10048 // CHECK17-NEXT: [[A_ADDR:%.*]] = alloca ptr, align 8
10049 // CHECK17-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
10050 // CHECK17-NEXT: [[TMP:%.*]] = alloca i32, align 4
10051 // CHECK17-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4
10052 // CHECK17-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4
10053 // CHECK17-NEXT: [[I:%.*]] = alloca i32, align 4
10054 // CHECK17-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4
10055 // CHECK17-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4
10056 // CHECK17-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
10057 // CHECK17-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
10058 // CHECK17-NEXT: [[I4:%.*]] = alloca i32, align 4
10059 // CHECK17-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 8
10060 // CHECK17-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 8
10061 // CHECK17-NEXT: store i64 [[DOTPREVIOUS_LB_]], ptr [[DOTPREVIOUS_LB__ADDR]], align 8
10062 // CHECK17-NEXT: store i64 [[DOTPREVIOUS_UB_]], ptr [[DOTPREVIOUS_UB__ADDR]], align 8
10063 // CHECK17-NEXT: store ptr [[N]], ptr [[N_ADDR]], align 8
10064 // CHECK17-NEXT: store i64 [[VLA]], ptr [[VLA_ADDR]], align 8
10065 // CHECK17-NEXT: store ptr [[A]], ptr [[A_ADDR]], align 8
10066 // CHECK17-NEXT: [[TMP0:%.*]] = load ptr, ptr [[N_ADDR]], align 8
10067 // CHECK17-NEXT: [[TMP1:%.*]] = load i64, ptr [[VLA_ADDR]], align 8
10068 // CHECK17-NEXT: [[TMP2:%.*]] = load ptr, ptr [[A_ADDR]], align 8
10069 // CHECK17-NEXT: [[TMP3:%.*]] = load i32, ptr [[TMP0]], align 4
10070 // CHECK17-NEXT: store i32 [[TMP3]], ptr [[DOTCAPTURE_EXPR_]], align 4
10071 // CHECK17-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4
10072 // CHECK17-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP4]], 0
10073 // CHECK17-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1
10074 // CHECK17-NEXT: [[SUB2:%.*]] = sub nsw i32 [[DIV]], 1
10075 // CHECK17-NEXT: store i32 [[SUB2]], ptr [[DOTCAPTURE_EXPR_1]], align 4
10076 // CHECK17-NEXT: store i32 0, ptr [[I]], align 4
10077 // CHECK17-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4
10078 // CHECK17-NEXT: [[CMP:%.*]] = icmp slt i32 0, [[TMP5]]
10079 // CHECK17-NEXT: br i1 [[CMP]], label [[OMP_PRECOND_THEN:%.*]], label [[OMP_PRECOND_END:%.*]]
10080 // CHECK17: omp.precond.then:
10081 // CHECK17-NEXT: store i32 0, ptr [[DOTOMP_LB]], align 4
10082 // CHECK17-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4
10083 // CHECK17-NEXT: store i32 [[TMP6]], ptr [[DOTOMP_UB]], align 4
10084 // CHECK17-NEXT: [[TMP7:%.*]] = load i64, ptr [[DOTPREVIOUS_LB__ADDR]], align 8
10085 // CHECK17-NEXT: [[CONV:%.*]] = trunc i64 [[TMP7]] to i32
10086 // CHECK17-NEXT: [[TMP8:%.*]] = load i64, ptr [[DOTPREVIOUS_UB__ADDR]], align 8
10087 // CHECK17-NEXT: [[CONV3:%.*]] = trunc i64 [[TMP8]] to i32
10088 // CHECK17-NEXT: store i32 [[CONV]], ptr [[DOTOMP_LB]], align 4
10089 // CHECK17-NEXT: store i32 [[CONV3]], ptr [[DOTOMP_UB]], align 4
10090 // CHECK17-NEXT: store i32 1, ptr [[DOTOMP_STRIDE]], align 4
10091 // CHECK17-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
10092 // CHECK17-NEXT: [[TMP9:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
10093 // CHECK17-NEXT: [[TMP10:%.*]] = load i32, ptr [[TMP9]], align 4
10094 // CHECK17-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB2:[0-9]+]], i32 [[TMP10]], i32 34, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
10095 // CHECK17-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
10096 // CHECK17-NEXT: [[TMP12:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4
10097 // CHECK17-NEXT: [[CMP5:%.*]] = icmp sgt i32 [[TMP11]], [[TMP12]]
10098 // CHECK17-NEXT: br i1 [[CMP5]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
10099 // CHECK17: cond.true:
10100 // CHECK17-NEXT: [[TMP13:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4
10101 // CHECK17-NEXT: br label [[COND_END:%.*]]
10102 // CHECK17: cond.false:
10103 // CHECK17-NEXT: [[TMP14:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
10104 // CHECK17-NEXT: br label [[COND_END]]
10105 // CHECK17: cond.end:
10106 // CHECK17-NEXT: [[COND:%.*]] = phi i32 [ [[TMP13]], [[COND_TRUE]] ], [ [[TMP14]], [[COND_FALSE]] ]
10107 // CHECK17-NEXT: store i32 [[COND]], ptr [[DOTOMP_UB]], align 4
10108 // CHECK17-NEXT: [[TMP15:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4
10109 // CHECK17-NEXT: store i32 [[TMP15]], ptr [[DOTOMP_IV]], align 4
10110 // CHECK17-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
10111 // CHECK17: omp.inner.for.cond:
10112 // CHECK17-NEXT: [[TMP16:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
10113 // CHECK17-NEXT: [[TMP17:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
10114 // CHECK17-NEXT: [[CMP6:%.*]] = icmp sle i32 [[TMP16]], [[TMP17]]
10115 // CHECK17-NEXT: br i1 [[CMP6]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
10116 // CHECK17: omp.inner.for.body:
10117 // CHECK17-NEXT: [[TMP18:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
10118 // CHECK17-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP18]], 1
10119 // CHECK17-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
10120 // CHECK17-NEXT: store i32 [[ADD]], ptr [[I4]], align 4
10121 // CHECK17-NEXT: [[TMP19:%.*]] = load i32, ptr [[I4]], align 4
10122 // CHECK17-NEXT: [[IDXPROM:%.*]] = sext i32 [[TMP19]] to i64
10123 // CHECK17-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i32, ptr [[TMP2]], i64 [[IDXPROM]]
10124 // CHECK17-NEXT: store i32 0, ptr [[ARRAYIDX]], align 4
10125 // CHECK17-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
10126 // CHECK17: omp.body.continue:
10127 // CHECK17-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
10128 // CHECK17: omp.inner.for.inc:
10129 // CHECK17-NEXT: [[TMP20:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
10130 // CHECK17-NEXT: [[ADD7:%.*]] = add nsw i32 [[TMP20]], 1
10131 // CHECK17-NEXT: store i32 [[ADD7]], ptr [[DOTOMP_IV]], align 4
10132 // CHECK17-NEXT: br label [[OMP_INNER_FOR_COND]]
10133 // CHECK17: omp.inner.for.end:
10134 // CHECK17-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
10135 // CHECK17: omp.loop.exit:
10136 // CHECK17-NEXT: [[TMP21:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
10137 // CHECK17-NEXT: [[TMP22:%.*]] = load i32, ptr [[TMP21]], align 4
10138 // CHECK17-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB2]], i32 [[TMP22]])
10139 // CHECK17-NEXT: br label [[OMP_PRECOND_END]]
10140 // CHECK17: omp.precond.end:
10141 // CHECK17-NEXT: ret void
10144 // CHECK17-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l153
10145 // CHECK17-SAME: (i64 noundef [[N:%.*]], i64 noundef [[VLA:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[A:%.*]]) #[[ATTR2]] {
10146 // CHECK17-NEXT: entry:
10147 // CHECK17-NEXT: [[N_ADDR:%.*]] = alloca i64, align 8
10148 // CHECK17-NEXT: [[VLA_ADDR:%.*]] = alloca i64, align 8
10149 // CHECK17-NEXT: [[A_ADDR:%.*]] = alloca ptr, align 8
10150 // CHECK17-NEXT: store i64 [[N]], ptr [[N_ADDR]], align 8
10151 // CHECK17-NEXT: store i64 [[VLA]], ptr [[VLA_ADDR]], align 8
10152 // CHECK17-NEXT: store ptr [[A]], ptr [[A_ADDR]], align 8
10153 // CHECK17-NEXT: [[TMP0:%.*]] = load i64, ptr [[VLA_ADDR]], align 8
10154 // CHECK17-NEXT: [[TMP1:%.*]] = load ptr, ptr [[A_ADDR]], align 8
10155 // CHECK17-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_teams(ptr @[[GLOB3]], i32 3, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l153.omp_outlined, ptr [[N_ADDR]], i64 [[TMP0]], ptr [[TMP1]])
10156 // CHECK17-NEXT: ret void
10159 // CHECK17-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l153.omp_outlined
10160 // CHECK17-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[N:%.*]], i64 noundef [[VLA:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[A:%.*]]) #[[ATTR2]] {
10161 // CHECK17-NEXT: entry:
10162 // CHECK17-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 8
10163 // CHECK17-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 8
10164 // CHECK17-NEXT: [[N_ADDR:%.*]] = alloca ptr, align 8
10165 // CHECK17-NEXT: [[VLA_ADDR:%.*]] = alloca i64, align 8
10166 // CHECK17-NEXT: [[A_ADDR:%.*]] = alloca ptr, align 8
10167 // CHECK17-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
10168 // CHECK17-NEXT: [[TMP:%.*]] = alloca i32, align 4
10169 // CHECK17-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4
10170 // CHECK17-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4
10171 // CHECK17-NEXT: [[I:%.*]] = alloca i32, align 4
10172 // CHECK17-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4
10173 // CHECK17-NEXT: [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4
10174 // CHECK17-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
10175 // CHECK17-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
10176 // CHECK17-NEXT: [[I3:%.*]] = alloca i32, align 4
10177 // CHECK17-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 8
10178 // CHECK17-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 8
10179 // CHECK17-NEXT: store ptr [[N]], ptr [[N_ADDR]], align 8
10180 // CHECK17-NEXT: store i64 [[VLA]], ptr [[VLA_ADDR]], align 8
10181 // CHECK17-NEXT: store ptr [[A]], ptr [[A_ADDR]], align 8
10182 // CHECK17-NEXT: [[TMP0:%.*]] = load ptr, ptr [[N_ADDR]], align 8
10183 // CHECK17-NEXT: [[TMP1:%.*]] = load i64, ptr [[VLA_ADDR]], align 8
10184 // CHECK17-NEXT: [[TMP2:%.*]] = load ptr, ptr [[A_ADDR]], align 8
10185 // CHECK17-NEXT: [[TMP3:%.*]] = load i32, ptr [[TMP0]], align 4
10186 // CHECK17-NEXT: store i32 [[TMP3]], ptr [[DOTCAPTURE_EXPR_]], align 4
10187 // CHECK17-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4
10188 // CHECK17-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP4]], 0
10189 // CHECK17-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1
10190 // CHECK17-NEXT: [[SUB2:%.*]] = sub nsw i32 [[DIV]], 1
10191 // CHECK17-NEXT: store i32 [[SUB2]], ptr [[DOTCAPTURE_EXPR_1]], align 4
10192 // CHECK17-NEXT: store i32 0, ptr [[I]], align 4
10193 // CHECK17-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4
10194 // CHECK17-NEXT: [[CMP:%.*]] = icmp slt i32 0, [[TMP5]]
10195 // CHECK17-NEXT: br i1 [[CMP]], label [[OMP_PRECOND_THEN:%.*]], label [[OMP_PRECOND_END:%.*]]
10196 // CHECK17: omp.precond.then:
10197 // CHECK17-NEXT: store i32 0, ptr [[DOTOMP_COMB_LB]], align 4
10198 // CHECK17-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4
10199 // CHECK17-NEXT: store i32 [[TMP6]], ptr [[DOTOMP_COMB_UB]], align 4
10200 // CHECK17-NEXT: store i32 1, ptr [[DOTOMP_STRIDE]], align 4
10201 // CHECK17-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
10202 // CHECK17-NEXT: [[TMP7:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
10203 // CHECK17-NEXT: [[TMP8:%.*]] = load i32, ptr [[TMP7]], align 4
10204 // CHECK17-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB1]], i32 [[TMP8]], i32 92, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
10205 // CHECK17-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
10206 // CHECK17-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4
10207 // CHECK17-NEXT: [[CMP4:%.*]] = icmp sgt i32 [[TMP9]], [[TMP10]]
10208 // CHECK17-NEXT: br i1 [[CMP4]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
10209 // CHECK17: cond.true:
10210 // CHECK17-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4
10211 // CHECK17-NEXT: br label [[COND_END:%.*]]
10212 // CHECK17: cond.false:
10213 // CHECK17-NEXT: [[TMP12:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
10214 // CHECK17-NEXT: br label [[COND_END]]
10215 // CHECK17: cond.end:
10216 // CHECK17-NEXT: [[COND:%.*]] = phi i32 [ [[TMP11]], [[COND_TRUE]] ], [ [[TMP12]], [[COND_FALSE]] ]
10217 // CHECK17-NEXT: store i32 [[COND]], ptr [[DOTOMP_COMB_UB]], align 4
10218 // CHECK17-NEXT: [[TMP13:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
10219 // CHECK17-NEXT: store i32 [[TMP13]], ptr [[DOTOMP_IV]], align 4
10220 // CHECK17-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
10221 // CHECK17: omp.inner.for.cond:
10222 // CHECK17-NEXT: [[TMP14:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
10223 // CHECK17-NEXT: [[TMP15:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
10224 // CHECK17-NEXT: [[CMP5:%.*]] = icmp sle i32 [[TMP14]], [[TMP15]]
10225 // CHECK17-NEXT: br i1 [[CMP5]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
10226 // CHECK17: omp.inner.for.body:
10227 // CHECK17-NEXT: [[TMP16:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
10228 // CHECK17-NEXT: [[TMP17:%.*]] = zext i32 [[TMP16]] to i64
10229 // CHECK17-NEXT: [[TMP18:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
10230 // CHECK17-NEXT: [[TMP19:%.*]] = zext i32 [[TMP18]] to i64
10231 // CHECK17-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_call(ptr @[[GLOB3]], i32 5, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l153.omp_outlined.omp_outlined, i64 [[TMP17]], i64 [[TMP19]], ptr [[TMP0]], i64 [[TMP1]], ptr [[TMP2]])
10232 // CHECK17-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
10233 // CHECK17: omp.inner.for.inc:
10234 // CHECK17-NEXT: [[TMP20:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
10235 // CHECK17-NEXT: [[TMP21:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4
10236 // CHECK17-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP20]], [[TMP21]]
10237 // CHECK17-NEXT: store i32 [[ADD]], ptr [[DOTOMP_IV]], align 4
10238 // CHECK17-NEXT: br label [[OMP_INNER_FOR_COND]]
10239 // CHECK17: omp.inner.for.end:
10240 // CHECK17-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
10241 // CHECK17: omp.loop.exit:
10242 // CHECK17-NEXT: [[TMP22:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
10243 // CHECK17-NEXT: [[TMP23:%.*]] = load i32, ptr [[TMP22]], align 4
10244 // CHECK17-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB1]], i32 [[TMP23]])
10245 // CHECK17-NEXT: br label [[OMP_PRECOND_END]]
10246 // CHECK17: omp.precond.end:
10247 // CHECK17-NEXT: ret void
10250 // CHECK17-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l153.omp_outlined.omp_outlined
10251 // CHECK17-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], i64 noundef [[DOTPREVIOUS_LB_:%.*]], i64 noundef [[DOTPREVIOUS_UB_:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[N:%.*]], i64 noundef [[VLA:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[A:%.*]]) #[[ATTR2]] {
10252 // CHECK17-NEXT: entry:
10253 // CHECK17-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 8
10254 // CHECK17-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 8
10255 // CHECK17-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i64, align 8
10256 // CHECK17-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i64, align 8
10257 // CHECK17-NEXT: [[N_ADDR:%.*]] = alloca ptr, align 8
10258 // CHECK17-NEXT: [[VLA_ADDR:%.*]] = alloca i64, align 8
10259 // CHECK17-NEXT: [[A_ADDR:%.*]] = alloca ptr, align 8
10260 // CHECK17-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
10261 // CHECK17-NEXT: [[TMP:%.*]] = alloca i32, align 4
10262 // CHECK17-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4
10263 // CHECK17-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4
10264 // CHECK17-NEXT: [[I:%.*]] = alloca i32, align 4
10265 // CHECK17-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4
10266 // CHECK17-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4
10267 // CHECK17-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
10268 // CHECK17-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
10269 // CHECK17-NEXT: [[I4:%.*]] = alloca i32, align 4
10270 // CHECK17-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 8
10271 // CHECK17-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 8
10272 // CHECK17-NEXT: store i64 [[DOTPREVIOUS_LB_]], ptr [[DOTPREVIOUS_LB__ADDR]], align 8
10273 // CHECK17-NEXT: store i64 [[DOTPREVIOUS_UB_]], ptr [[DOTPREVIOUS_UB__ADDR]], align 8
10274 // CHECK17-NEXT: store ptr [[N]], ptr [[N_ADDR]], align 8
10275 // CHECK17-NEXT: store i64 [[VLA]], ptr [[VLA_ADDR]], align 8
10276 // CHECK17-NEXT: store ptr [[A]], ptr [[A_ADDR]], align 8
10277 // CHECK17-NEXT: [[TMP0:%.*]] = load ptr, ptr [[N_ADDR]], align 8
10278 // CHECK17-NEXT: [[TMP1:%.*]] = load i64, ptr [[VLA_ADDR]], align 8
10279 // CHECK17-NEXT: [[TMP2:%.*]] = load ptr, ptr [[A_ADDR]], align 8
10280 // CHECK17-NEXT: [[TMP3:%.*]] = load i32, ptr [[TMP0]], align 4
10281 // CHECK17-NEXT: store i32 [[TMP3]], ptr [[DOTCAPTURE_EXPR_]], align 4
10282 // CHECK17-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4
10283 // CHECK17-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP4]], 0
10284 // CHECK17-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1
10285 // CHECK17-NEXT: [[SUB2:%.*]] = sub nsw i32 [[DIV]], 1
10286 // CHECK17-NEXT: store i32 [[SUB2]], ptr [[DOTCAPTURE_EXPR_1]], align 4
10287 // CHECK17-NEXT: store i32 0, ptr [[I]], align 4
10288 // CHECK17-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4
10289 // CHECK17-NEXT: [[CMP:%.*]] = icmp slt i32 0, [[TMP5]]
10290 // CHECK17-NEXT: br i1 [[CMP]], label [[OMP_PRECOND_THEN:%.*]], label [[OMP_PRECOND_END:%.*]]
10291 // CHECK17: omp.precond.then:
10292 // CHECK17-NEXT: store i32 0, ptr [[DOTOMP_LB]], align 4
10293 // CHECK17-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4
10294 // CHECK17-NEXT: store i32 [[TMP6]], ptr [[DOTOMP_UB]], align 4
10295 // CHECK17-NEXT: [[TMP7:%.*]] = load i64, ptr [[DOTPREVIOUS_LB__ADDR]], align 8
10296 // CHECK17-NEXT: [[CONV:%.*]] = trunc i64 [[TMP7]] to i32
10297 // CHECK17-NEXT: [[TMP8:%.*]] = load i64, ptr [[DOTPREVIOUS_UB__ADDR]], align 8
10298 // CHECK17-NEXT: [[CONV3:%.*]] = trunc i64 [[TMP8]] to i32
10299 // CHECK17-NEXT: store i32 [[CONV]], ptr [[DOTOMP_LB]], align 4
10300 // CHECK17-NEXT: store i32 [[CONV3]], ptr [[DOTOMP_UB]], align 4
10301 // CHECK17-NEXT: store i32 1, ptr [[DOTOMP_STRIDE]], align 4
10302 // CHECK17-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
10303 // CHECK17-NEXT: [[TMP9:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
10304 // CHECK17-NEXT: [[TMP10:%.*]] = load i32, ptr [[TMP9]], align 4
10305 // CHECK17-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB2]], i32 [[TMP10]], i32 34, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
10306 // CHECK17-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
10307 // CHECK17-NEXT: [[TMP12:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4
10308 // CHECK17-NEXT: [[CMP5:%.*]] = icmp sgt i32 [[TMP11]], [[TMP12]]
10309 // CHECK17-NEXT: br i1 [[CMP5]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
10310 // CHECK17: cond.true:
10311 // CHECK17-NEXT: [[TMP13:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4
10312 // CHECK17-NEXT: br label [[COND_END:%.*]]
10313 // CHECK17: cond.false:
10314 // CHECK17-NEXT: [[TMP14:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
10315 // CHECK17-NEXT: br label [[COND_END]]
10316 // CHECK17: cond.end:
10317 // CHECK17-NEXT: [[COND:%.*]] = phi i32 [ [[TMP13]], [[COND_TRUE]] ], [ [[TMP14]], [[COND_FALSE]] ]
10318 // CHECK17-NEXT: store i32 [[COND]], ptr [[DOTOMP_UB]], align 4
10319 // CHECK17-NEXT: [[TMP15:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4
10320 // CHECK17-NEXT: store i32 [[TMP15]], ptr [[DOTOMP_IV]], align 4
10321 // CHECK17-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
10322 // CHECK17: omp.inner.for.cond:
10323 // CHECK17-NEXT: [[TMP16:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
10324 // CHECK17-NEXT: [[TMP17:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
10325 // CHECK17-NEXT: [[CMP6:%.*]] = icmp sle i32 [[TMP16]], [[TMP17]]
10326 // CHECK17-NEXT: br i1 [[CMP6]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
10327 // CHECK17: omp.inner.for.body:
10328 // CHECK17-NEXT: [[TMP18:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
10329 // CHECK17-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP18]], 1
10330 // CHECK17-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
10331 // CHECK17-NEXT: store i32 [[ADD]], ptr [[I4]], align 4
10332 // CHECK17-NEXT: [[TMP19:%.*]] = load i32, ptr [[I4]], align 4
10333 // CHECK17-NEXT: [[IDXPROM:%.*]] = sext i32 [[TMP19]] to i64
10334 // CHECK17-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i32, ptr [[TMP2]], i64 [[IDXPROM]]
10335 // CHECK17-NEXT: store i32 0, ptr [[ARRAYIDX]], align 4
10336 // CHECK17-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
10337 // CHECK17: omp.body.continue:
10338 // CHECK17-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
10339 // CHECK17: omp.inner.for.inc:
10340 // CHECK17-NEXT: [[TMP20:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
10341 // CHECK17-NEXT: [[ADD7:%.*]] = add nsw i32 [[TMP20]], 1
10342 // CHECK17-NEXT: store i32 [[ADD7]], ptr [[DOTOMP_IV]], align 4
10343 // CHECK17-NEXT: br label [[OMP_INNER_FOR_COND]]
10344 // CHECK17: omp.inner.for.end:
10345 // CHECK17-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
10346 // CHECK17: omp.loop.exit:
10347 // CHECK17-NEXT: [[TMP21:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
10348 // CHECK17-NEXT: [[TMP22:%.*]] = load i32, ptr [[TMP21]], align 4
10349 // CHECK17-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB2]], i32 [[TMP22]])
10350 // CHECK17-NEXT: br label [[OMP_PRECOND_END]]
10351 // CHECK17: omp.precond.end:
10352 // CHECK17-NEXT: ret void
10355 // CHECK17-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l158
10356 // CHECK17-SAME: (i64 noundef [[M:%.*]], i64 noundef [[N:%.*]], i64 noundef [[VLA:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[A:%.*]]) #[[ATTR2]] {
10357 // CHECK17-NEXT: entry:
10358 // CHECK17-NEXT: [[M_ADDR:%.*]] = alloca i64, align 8
10359 // CHECK17-NEXT: [[N_ADDR:%.*]] = alloca i64, align 8
10360 // CHECK17-NEXT: [[VLA_ADDR:%.*]] = alloca i64, align 8
10361 // CHECK17-NEXT: [[A_ADDR:%.*]] = alloca ptr, align 8
10362 // CHECK17-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4
10363 // CHECK17-NEXT: [[DOTCAPTURE_EXPR__CASTED:%.*]] = alloca i64, align 8
10364 // CHECK17-NEXT: store i64 [[M]], ptr [[M_ADDR]], align 8
10365 // CHECK17-NEXT: store i64 [[N]], ptr [[N_ADDR]], align 8
10366 // CHECK17-NEXT: store i64 [[VLA]], ptr [[VLA_ADDR]], align 8
10367 // CHECK17-NEXT: store ptr [[A]], ptr [[A_ADDR]], align 8
10368 // CHECK17-NEXT: [[TMP0:%.*]] = load i64, ptr [[VLA_ADDR]], align 8
10369 // CHECK17-NEXT: [[TMP1:%.*]] = load ptr, ptr [[A_ADDR]], align 8
10370 // CHECK17-NEXT: [[TMP2:%.*]] = load i32, ptr [[M_ADDR]], align 4
10371 // CHECK17-NEXT: store i32 [[TMP2]], ptr [[DOTCAPTURE_EXPR_]], align 4
10372 // CHECK17-NEXT: [[TMP3:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4
10373 // CHECK17-NEXT: store i32 [[TMP3]], ptr [[DOTCAPTURE_EXPR__CASTED]], align 4
10374 // CHECK17-NEXT: [[TMP4:%.*]] = load i64, ptr [[DOTCAPTURE_EXPR__CASTED]], align 8
10375 // CHECK17-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_teams(ptr @[[GLOB3]], i32 4, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l158.omp_outlined, ptr [[N_ADDR]], i64 [[TMP0]], ptr [[TMP1]], i64 [[TMP4]])
10376 // CHECK17-NEXT: ret void
10379 // CHECK17-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l158.omp_outlined
10380 // CHECK17-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[N:%.*]], i64 noundef [[VLA:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[A:%.*]], i64 noundef [[DOTCAPTURE_EXPR_:%.*]]) #[[ATTR2]] {
10381 // CHECK17-NEXT: entry:
10382 // CHECK17-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 8
10383 // CHECK17-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 8
10384 // CHECK17-NEXT: [[N_ADDR:%.*]] = alloca ptr, align 8
10385 // CHECK17-NEXT: [[VLA_ADDR:%.*]] = alloca i64, align 8
10386 // CHECK17-NEXT: [[A_ADDR:%.*]] = alloca ptr, align 8
10387 // CHECK17-NEXT: [[DOTCAPTURE_EXPR__ADDR:%.*]] = alloca i64, align 8
10388 // CHECK17-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
10389 // CHECK17-NEXT: [[TMP:%.*]] = alloca i32, align 4
10390 // CHECK17-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4
10391 // CHECK17-NEXT: [[DOTCAPTURE_EXPR_2:%.*]] = alloca i32, align 4
10392 // CHECK17-NEXT: [[I:%.*]] = alloca i32, align 4
10393 // CHECK17-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4
10394 // CHECK17-NEXT: [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4
10395 // CHECK17-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
10396 // CHECK17-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
10397 // CHECK17-NEXT: [[I4:%.*]] = alloca i32, align 4
10398 // CHECK17-NEXT: [[DOTCAPTURE_EXPR__CASTED:%.*]] = alloca i64, align 8
10399 // CHECK17-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 8
10400 // CHECK17-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 8
10401 // CHECK17-NEXT: store ptr [[N]], ptr [[N_ADDR]], align 8
10402 // CHECK17-NEXT: store i64 [[VLA]], ptr [[VLA_ADDR]], align 8
10403 // CHECK17-NEXT: store ptr [[A]], ptr [[A_ADDR]], align 8
10404 // CHECK17-NEXT: store i64 [[DOTCAPTURE_EXPR_]], ptr [[DOTCAPTURE_EXPR__ADDR]], align 8
10405 // CHECK17-NEXT: [[TMP0:%.*]] = load ptr, ptr [[N_ADDR]], align 8
10406 // CHECK17-NEXT: [[TMP1:%.*]] = load i64, ptr [[VLA_ADDR]], align 8
10407 // CHECK17-NEXT: [[TMP2:%.*]] = load ptr, ptr [[A_ADDR]], align 8
10408 // CHECK17-NEXT: [[TMP3:%.*]] = load i32, ptr [[TMP0]], align 4
10409 // CHECK17-NEXT: store i32 [[TMP3]], ptr [[DOTCAPTURE_EXPR_1]], align 4
10410 // CHECK17-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4
10411 // CHECK17-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP4]], 0
10412 // CHECK17-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1
10413 // CHECK17-NEXT: [[SUB3:%.*]] = sub nsw i32 [[DIV]], 1
10414 // CHECK17-NEXT: store i32 [[SUB3]], ptr [[DOTCAPTURE_EXPR_2]], align 4
10415 // CHECK17-NEXT: store i32 0, ptr [[I]], align 4
10416 // CHECK17-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4
10417 // CHECK17-NEXT: [[CMP:%.*]] = icmp slt i32 0, [[TMP5]]
10418 // CHECK17-NEXT: br i1 [[CMP]], label [[OMP_PRECOND_THEN:%.*]], label [[OMP_PRECOND_END:%.*]]
10419 // CHECK17: omp.precond.then:
10420 // CHECK17-NEXT: store i32 0, ptr [[DOTOMP_COMB_LB]], align 4
10421 // CHECK17-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_2]], align 4
10422 // CHECK17-NEXT: store i32 [[TMP6]], ptr [[DOTOMP_COMB_UB]], align 4
10423 // CHECK17-NEXT: store i32 1, ptr [[DOTOMP_STRIDE]], align 4
10424 // CHECK17-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
10425 // CHECK17-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR__ADDR]], align 4
10426 // CHECK17-NEXT: [[TMP8:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
10427 // CHECK17-NEXT: [[TMP9:%.*]] = load i32, ptr [[TMP8]], align 4
10428 // CHECK17-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB1]], i32 [[TMP9]], i32 91, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 [[TMP7]])
10429 // CHECK17-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
10430 // CHECK17-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_2]], align 4
10431 // CHECK17-NEXT: [[CMP5:%.*]] = icmp sgt i32 [[TMP10]], [[TMP11]]
10432 // CHECK17-NEXT: br i1 [[CMP5]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
10433 // CHECK17: cond.true:
10434 // CHECK17-NEXT: [[TMP12:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_2]], align 4
10435 // CHECK17-NEXT: br label [[COND_END:%.*]]
10436 // CHECK17: cond.false:
10437 // CHECK17-NEXT: [[TMP13:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
10438 // CHECK17-NEXT: br label [[COND_END]]
10439 // CHECK17: cond.end:
10440 // CHECK17-NEXT: [[COND:%.*]] = phi i32 [ [[TMP12]], [[COND_TRUE]] ], [ [[TMP13]], [[COND_FALSE]] ]
10441 // CHECK17-NEXT: store i32 [[COND]], ptr [[DOTOMP_COMB_UB]], align 4
10442 // CHECK17-NEXT: [[TMP14:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
10443 // CHECK17-NEXT: store i32 [[TMP14]], ptr [[DOTOMP_IV]], align 4
10444 // CHECK17-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
10445 // CHECK17: omp.inner.for.cond:
10446 // CHECK17-NEXT: [[TMP15:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
10447 // CHECK17-NEXT: [[TMP16:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_2]], align 4
10448 // CHECK17-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP16]], 1
10449 // CHECK17-NEXT: [[CMP6:%.*]] = icmp slt i32 [[TMP15]], [[ADD]]
10450 // CHECK17-NEXT: br i1 [[CMP6]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
10451 // CHECK17: omp.inner.for.body:
10452 // CHECK17-NEXT: [[TMP17:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
10453 // CHECK17-NEXT: [[TMP18:%.*]] = zext i32 [[TMP17]] to i64
10454 // CHECK17-NEXT: [[TMP19:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
10455 // CHECK17-NEXT: [[TMP20:%.*]] = zext i32 [[TMP19]] to i64
10456 // CHECK17-NEXT: [[TMP21:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR__ADDR]], align 4
10457 // CHECK17-NEXT: store i32 [[TMP21]], ptr [[DOTCAPTURE_EXPR__CASTED]], align 4
10458 // CHECK17-NEXT: [[TMP22:%.*]] = load i64, ptr [[DOTCAPTURE_EXPR__CASTED]], align 8
10459 // CHECK17-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_call(ptr @[[GLOB3]], i32 6, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l158.omp_outlined.omp_outlined, i64 [[TMP18]], i64 [[TMP20]], ptr [[TMP0]], i64 [[TMP1]], ptr [[TMP2]], i64 [[TMP22]])
10460 // CHECK17-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
10461 // CHECK17: omp.inner.for.inc:
10462 // CHECK17-NEXT: [[TMP23:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
10463 // CHECK17-NEXT: [[TMP24:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4
10464 // CHECK17-NEXT: [[ADD7:%.*]] = add nsw i32 [[TMP23]], [[TMP24]]
10465 // CHECK17-NEXT: store i32 [[ADD7]], ptr [[DOTOMP_IV]], align 4
10466 // CHECK17-NEXT: [[TMP25:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
10467 // CHECK17-NEXT: [[TMP26:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4
10468 // CHECK17-NEXT: [[ADD8:%.*]] = add nsw i32 [[TMP25]], [[TMP26]]
10469 // CHECK17-NEXT: store i32 [[ADD8]], ptr [[DOTOMP_COMB_LB]], align 4
10470 // CHECK17-NEXT: [[TMP27:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
10471 // CHECK17-NEXT: [[TMP28:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4
10472 // CHECK17-NEXT: [[ADD9:%.*]] = add nsw i32 [[TMP27]], [[TMP28]]
10473 // CHECK17-NEXT: store i32 [[ADD9]], ptr [[DOTOMP_COMB_UB]], align 4
10474 // CHECK17-NEXT: [[TMP29:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
10475 // CHECK17-NEXT: [[TMP30:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_2]], align 4
10476 // CHECK17-NEXT: [[CMP10:%.*]] = icmp sgt i32 [[TMP29]], [[TMP30]]
10477 // CHECK17-NEXT: br i1 [[CMP10]], label [[COND_TRUE11:%.*]], label [[COND_FALSE12:%.*]]
10478 // CHECK17: cond.true11:
10479 // CHECK17-NEXT: [[TMP31:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_2]], align 4
10480 // CHECK17-NEXT: br label [[COND_END13:%.*]]
10481 // CHECK17: cond.false12:
10482 // CHECK17-NEXT: [[TMP32:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
10483 // CHECK17-NEXT: br label [[COND_END13]]
10484 // CHECK17: cond.end13:
10485 // CHECK17-NEXT: [[COND14:%.*]] = phi i32 [ [[TMP31]], [[COND_TRUE11]] ], [ [[TMP32]], [[COND_FALSE12]] ]
10486 // CHECK17-NEXT: store i32 [[COND14]], ptr [[DOTOMP_COMB_UB]], align 4
10487 // CHECK17-NEXT: [[TMP33:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
10488 // CHECK17-NEXT: store i32 [[TMP33]], ptr [[DOTOMP_IV]], align 4
10489 // CHECK17-NEXT: br label [[OMP_INNER_FOR_COND]]
10490 // CHECK17: omp.inner.for.end:
10491 // CHECK17-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
10492 // CHECK17: omp.loop.exit:
10493 // CHECK17-NEXT: [[TMP34:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
10494 // CHECK17-NEXT: [[TMP35:%.*]] = load i32, ptr [[TMP34]], align 4
10495 // CHECK17-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB1]], i32 [[TMP35]])
10496 // CHECK17-NEXT: br label [[OMP_PRECOND_END]]
10497 // CHECK17: omp.precond.end:
10498 // CHECK17-NEXT: ret void
10501 // CHECK17-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l158.omp_outlined.omp_outlined
10502 // CHECK17-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], i64 noundef [[DOTPREVIOUS_LB_:%.*]], i64 noundef [[DOTPREVIOUS_UB_:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[N:%.*]], i64 noundef [[VLA:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[A:%.*]], i64 noundef [[DOTCAPTURE_EXPR_:%.*]]) #[[ATTR2]] {
10503 // CHECK17-NEXT: entry:
10504 // CHECK17-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 8
10505 // CHECK17-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 8
10506 // CHECK17-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i64, align 8
10507 // CHECK17-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i64, align 8
10508 // CHECK17-NEXT: [[N_ADDR:%.*]] = alloca ptr, align 8
10509 // CHECK17-NEXT: [[VLA_ADDR:%.*]] = alloca i64, align 8
10510 // CHECK17-NEXT: [[A_ADDR:%.*]] = alloca ptr, align 8
10511 // CHECK17-NEXT: [[DOTCAPTURE_EXPR__ADDR:%.*]] = alloca i64, align 8
10512 // CHECK17-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
10513 // CHECK17-NEXT: [[TMP:%.*]] = alloca i32, align 4
10514 // CHECK17-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4
10515 // CHECK17-NEXT: [[DOTCAPTURE_EXPR_2:%.*]] = alloca i32, align 4
10516 // CHECK17-NEXT: [[I:%.*]] = alloca i32, align 4
10517 // CHECK17-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4
10518 // CHECK17-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4
10519 // CHECK17-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
10520 // CHECK17-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
10521 // CHECK17-NEXT: [[I5:%.*]] = alloca i32, align 4
10522 // CHECK17-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 8
10523 // CHECK17-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 8
10524 // CHECK17-NEXT: store i64 [[DOTPREVIOUS_LB_]], ptr [[DOTPREVIOUS_LB__ADDR]], align 8
10525 // CHECK17-NEXT: store i64 [[DOTPREVIOUS_UB_]], ptr [[DOTPREVIOUS_UB__ADDR]], align 8
10526 // CHECK17-NEXT: store ptr [[N]], ptr [[N_ADDR]], align 8
10527 // CHECK17-NEXT: store i64 [[VLA]], ptr [[VLA_ADDR]], align 8
10528 // CHECK17-NEXT: store ptr [[A]], ptr [[A_ADDR]], align 8
10529 // CHECK17-NEXT: store i64 [[DOTCAPTURE_EXPR_]], ptr [[DOTCAPTURE_EXPR__ADDR]], align 8
10530 // CHECK17-NEXT: [[TMP0:%.*]] = load ptr, ptr [[N_ADDR]], align 8
10531 // CHECK17-NEXT: [[TMP1:%.*]] = load i64, ptr [[VLA_ADDR]], align 8
10532 // CHECK17-NEXT: [[TMP2:%.*]] = load ptr, ptr [[A_ADDR]], align 8
10533 // CHECK17-NEXT: [[TMP3:%.*]] = load i32, ptr [[TMP0]], align 4
10534 // CHECK17-NEXT: store i32 [[TMP3]], ptr [[DOTCAPTURE_EXPR_1]], align 4
10535 // CHECK17-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4
10536 // CHECK17-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP4]], 0
10537 // CHECK17-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1
10538 // CHECK17-NEXT: [[SUB3:%.*]] = sub nsw i32 [[DIV]], 1
10539 // CHECK17-NEXT: store i32 [[SUB3]], ptr [[DOTCAPTURE_EXPR_2]], align 4
10540 // CHECK17-NEXT: store i32 0, ptr [[I]], align 4
10541 // CHECK17-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4
10542 // CHECK17-NEXT: [[CMP:%.*]] = icmp slt i32 0, [[TMP5]]
10543 // CHECK17-NEXT: br i1 [[CMP]], label [[OMP_PRECOND_THEN:%.*]], label [[OMP_PRECOND_END:%.*]]
10544 // CHECK17: omp.precond.then:
10545 // CHECK17-NEXT: store i32 0, ptr [[DOTOMP_LB]], align 4
10546 // CHECK17-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_2]], align 4
10547 // CHECK17-NEXT: store i32 [[TMP6]], ptr [[DOTOMP_UB]], align 4
10548 // CHECK17-NEXT: [[TMP7:%.*]] = load i64, ptr [[DOTPREVIOUS_LB__ADDR]], align 8
10549 // CHECK17-NEXT: [[CONV:%.*]] = trunc i64 [[TMP7]] to i32
10550 // CHECK17-NEXT: [[TMP8:%.*]] = load i64, ptr [[DOTPREVIOUS_UB__ADDR]], align 8
10551 // CHECK17-NEXT: [[CONV4:%.*]] = trunc i64 [[TMP8]] to i32
10552 // CHECK17-NEXT: store i32 [[CONV]], ptr [[DOTOMP_LB]], align 4
10553 // CHECK17-NEXT: store i32 [[CONV4]], ptr [[DOTOMP_UB]], align 4
10554 // CHECK17-NEXT: store i32 1, ptr [[DOTOMP_STRIDE]], align 4
10555 // CHECK17-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
10556 // CHECK17-NEXT: [[TMP9:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
10557 // CHECK17-NEXT: [[TMP10:%.*]] = load i32, ptr [[TMP9]], align 4
10558 // CHECK17-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB2]], i32 [[TMP10]], i32 34, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
10559 // CHECK17-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
10560 // CHECK17-NEXT: [[TMP12:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_2]], align 4
10561 // CHECK17-NEXT: [[CMP6:%.*]] = icmp sgt i32 [[TMP11]], [[TMP12]]
10562 // CHECK17-NEXT: br i1 [[CMP6]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
10563 // CHECK17: cond.true:
10564 // CHECK17-NEXT: [[TMP13:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_2]], align 4
10565 // CHECK17-NEXT: br label [[COND_END:%.*]]
10566 // CHECK17: cond.false:
10567 // CHECK17-NEXT: [[TMP14:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
10568 // CHECK17-NEXT: br label [[COND_END]]
10569 // CHECK17: cond.end:
10570 // CHECK17-NEXT: [[COND:%.*]] = phi i32 [ [[TMP13]], [[COND_TRUE]] ], [ [[TMP14]], [[COND_FALSE]] ]
10571 // CHECK17-NEXT: store i32 [[COND]], ptr [[DOTOMP_UB]], align 4
10572 // CHECK17-NEXT: [[TMP15:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4
10573 // CHECK17-NEXT: store i32 [[TMP15]], ptr [[DOTOMP_IV]], align 4
10574 // CHECK17-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
10575 // CHECK17: omp.inner.for.cond:
10576 // CHECK17-NEXT: [[TMP16:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
10577 // CHECK17-NEXT: [[TMP17:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
10578 // CHECK17-NEXT: [[CMP7:%.*]] = icmp sle i32 [[TMP16]], [[TMP17]]
10579 // CHECK17-NEXT: br i1 [[CMP7]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
10580 // CHECK17: omp.inner.for.body:
10581 // CHECK17-NEXT: [[TMP18:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
10582 // CHECK17-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP18]], 1
10583 // CHECK17-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
10584 // CHECK17-NEXT: store i32 [[ADD]], ptr [[I5]], align 4
10585 // CHECK17-NEXT: [[TMP19:%.*]] = load i32, ptr [[I5]], align 4
10586 // CHECK17-NEXT: [[IDXPROM:%.*]] = sext i32 [[TMP19]] to i64
10587 // CHECK17-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i32, ptr [[TMP2]], i64 [[IDXPROM]]
10588 // CHECK17-NEXT: store i32 0, ptr [[ARRAYIDX]], align 4
10589 // CHECK17-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
10590 // CHECK17: omp.body.continue:
10591 // CHECK17-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
10592 // CHECK17: omp.inner.for.inc:
10593 // CHECK17-NEXT: [[TMP20:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
10594 // CHECK17-NEXT: [[ADD8:%.*]] = add nsw i32 [[TMP20]], 1
10595 // CHECK17-NEXT: store i32 [[ADD8]], ptr [[DOTOMP_IV]], align 4
10596 // CHECK17-NEXT: br label [[OMP_INNER_FOR_COND]]
10597 // CHECK17: omp.inner.for.end:
10598 // CHECK17-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
10599 // CHECK17: omp.loop.exit:
10600 // CHECK17-NEXT: [[TMP21:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
10601 // CHECK17-NEXT: [[TMP22:%.*]] = load i32, ptr [[TMP21]], align 4
10602 // CHECK17-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB2]], i32 [[TMP22]])
10603 // CHECK17-NEXT: br label [[OMP_PRECOND_END]]
10604 // CHECK17: omp.precond.end:
10605 // CHECK17-NEXT: ret void
10608 // CHECK17-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l163
10609 // CHECK17-SAME: (i64 noundef [[N:%.*]], i64 noundef [[VLA:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[A:%.*]]) #[[ATTR2]] {
10610 // CHECK17-NEXT: entry:
10611 // CHECK17-NEXT: [[N_ADDR:%.*]] = alloca i64, align 8
10612 // CHECK17-NEXT: [[VLA_ADDR:%.*]] = alloca i64, align 8
10613 // CHECK17-NEXT: [[A_ADDR:%.*]] = alloca ptr, align 8
10614 // CHECK17-NEXT: store i64 [[N]], ptr [[N_ADDR]], align 8
10615 // CHECK17-NEXT: store i64 [[VLA]], ptr [[VLA_ADDR]], align 8
10616 // CHECK17-NEXT: store ptr [[A]], ptr [[A_ADDR]], align 8
10617 // CHECK17-NEXT: [[TMP0:%.*]] = load i64, ptr [[VLA_ADDR]], align 8
10618 // CHECK17-NEXT: [[TMP1:%.*]] = load ptr, ptr [[A_ADDR]], align 8
10619 // CHECK17-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_teams(ptr @[[GLOB3]], i32 3, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l163.omp_outlined, ptr [[N_ADDR]], i64 [[TMP0]], ptr [[TMP1]])
10620 // CHECK17-NEXT: ret void
10623 // CHECK17-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l163.omp_outlined
10624 // CHECK17-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[N:%.*]], i64 noundef [[VLA:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[A:%.*]]) #[[ATTR2]] {
10625 // CHECK17-NEXT: entry:
10626 // CHECK17-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 8
10627 // CHECK17-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 8
10628 // CHECK17-NEXT: [[N_ADDR:%.*]] = alloca ptr, align 8
10629 // CHECK17-NEXT: [[VLA_ADDR:%.*]] = alloca i64, align 8
10630 // CHECK17-NEXT: [[A_ADDR:%.*]] = alloca ptr, align 8
10631 // CHECK17-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
10632 // CHECK17-NEXT: [[TMP:%.*]] = alloca i32, align 4
10633 // CHECK17-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4
10634 // CHECK17-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4
10635 // CHECK17-NEXT: [[I:%.*]] = alloca i32, align 4
10636 // CHECK17-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4
10637 // CHECK17-NEXT: [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4
10638 // CHECK17-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
10639 // CHECK17-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
10640 // CHECK17-NEXT: [[I3:%.*]] = alloca i32, align 4
10641 // CHECK17-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 8
10642 // CHECK17-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 8
10643 // CHECK17-NEXT: store ptr [[N]], ptr [[N_ADDR]], align 8
10644 // CHECK17-NEXT: store i64 [[VLA]], ptr [[VLA_ADDR]], align 8
10645 // CHECK17-NEXT: store ptr [[A]], ptr [[A_ADDR]], align 8
10646 // CHECK17-NEXT: [[TMP0:%.*]] = load ptr, ptr [[N_ADDR]], align 8
10647 // CHECK17-NEXT: [[TMP1:%.*]] = load i64, ptr [[VLA_ADDR]], align 8
10648 // CHECK17-NEXT: [[TMP2:%.*]] = load ptr, ptr [[A_ADDR]], align 8
10649 // CHECK17-NEXT: [[TMP3:%.*]] = load i32, ptr [[TMP0]], align 4
10650 // CHECK17-NEXT: store i32 [[TMP3]], ptr [[DOTCAPTURE_EXPR_]], align 4
10651 // CHECK17-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4
10652 // CHECK17-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP4]], 0
10653 // CHECK17-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1
10654 // CHECK17-NEXT: [[SUB2:%.*]] = sub nsw i32 [[DIV]], 1
10655 // CHECK17-NEXT: store i32 [[SUB2]], ptr [[DOTCAPTURE_EXPR_1]], align 4
10656 // CHECK17-NEXT: store i32 0, ptr [[I]], align 4
10657 // CHECK17-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4
10658 // CHECK17-NEXT: [[CMP:%.*]] = icmp slt i32 0, [[TMP5]]
10659 // CHECK17-NEXT: br i1 [[CMP]], label [[OMP_PRECOND_THEN:%.*]], label [[OMP_PRECOND_END:%.*]]
10660 // CHECK17: omp.precond.then:
10661 // CHECK17-NEXT: store i32 0, ptr [[DOTOMP_COMB_LB]], align 4
10662 // CHECK17-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4
10663 // CHECK17-NEXT: store i32 [[TMP6]], ptr [[DOTOMP_COMB_UB]], align 4
10664 // CHECK17-NEXT: store i32 1, ptr [[DOTOMP_STRIDE]], align 4
10665 // CHECK17-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
10666 // CHECK17-NEXT: [[TMP7:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
10667 // CHECK17-NEXT: [[TMP8:%.*]] = load i32, ptr [[TMP7]], align 4
10668 // CHECK17-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB1]], i32 [[TMP8]], i32 92, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
10669 // CHECK17-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
10670 // CHECK17-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4
10671 // CHECK17-NEXT: [[CMP4:%.*]] = icmp sgt i32 [[TMP9]], [[TMP10]]
10672 // CHECK17-NEXT: br i1 [[CMP4]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
10673 // CHECK17: cond.true:
10674 // CHECK17-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4
10675 // CHECK17-NEXT: br label [[COND_END:%.*]]
10676 // CHECK17: cond.false:
10677 // CHECK17-NEXT: [[TMP12:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
10678 // CHECK17-NEXT: br label [[COND_END]]
10679 // CHECK17: cond.end:
10680 // CHECK17-NEXT: [[COND:%.*]] = phi i32 [ [[TMP11]], [[COND_TRUE]] ], [ [[TMP12]], [[COND_FALSE]] ]
10681 // CHECK17-NEXT: store i32 [[COND]], ptr [[DOTOMP_COMB_UB]], align 4
10682 // CHECK17-NEXT: [[TMP13:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
10683 // CHECK17-NEXT: store i32 [[TMP13]], ptr [[DOTOMP_IV]], align 4
10684 // CHECK17-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
10685 // CHECK17: omp.inner.for.cond:
10686 // CHECK17-NEXT: [[TMP14:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
10687 // CHECK17-NEXT: [[TMP15:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
10688 // CHECK17-NEXT: [[CMP5:%.*]] = icmp sle i32 [[TMP14]], [[TMP15]]
10689 // CHECK17-NEXT: br i1 [[CMP5]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
10690 // CHECK17: omp.inner.for.body:
10691 // CHECK17-NEXT: [[TMP16:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
10692 // CHECK17-NEXT: [[TMP17:%.*]] = zext i32 [[TMP16]] to i64
10693 // CHECK17-NEXT: [[TMP18:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
10694 // CHECK17-NEXT: [[TMP19:%.*]] = zext i32 [[TMP18]] to i64
10695 // CHECK17-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_call(ptr @[[GLOB3]], i32 5, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l163.omp_outlined.omp_outlined, i64 [[TMP17]], i64 [[TMP19]], ptr [[TMP0]], i64 [[TMP1]], ptr [[TMP2]])
10696 // CHECK17-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
10697 // CHECK17: omp.inner.for.inc:
10698 // CHECK17-NEXT: [[TMP20:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
10699 // CHECK17-NEXT: [[TMP21:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4
10700 // CHECK17-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP20]], [[TMP21]]
10701 // CHECK17-NEXT: store i32 [[ADD]], ptr [[DOTOMP_IV]], align 4
10702 // CHECK17-NEXT: br label [[OMP_INNER_FOR_COND]]
10703 // CHECK17: omp.inner.for.end:
10704 // CHECK17-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
10705 // CHECK17: omp.loop.exit:
10706 // CHECK17-NEXT: [[TMP22:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
10707 // CHECK17-NEXT: [[TMP23:%.*]] = load i32, ptr [[TMP22]], align 4
10708 // CHECK17-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB1]], i32 [[TMP23]])
10709 // CHECK17-NEXT: br label [[OMP_PRECOND_END]]
10710 // CHECK17: omp.precond.end:
10711 // CHECK17-NEXT: ret void
10714 // CHECK17-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l163.omp_outlined.omp_outlined
10715 // CHECK17-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], i64 noundef [[DOTPREVIOUS_LB_:%.*]], i64 noundef [[DOTPREVIOUS_UB_:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[N:%.*]], i64 noundef [[VLA:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[A:%.*]]) #[[ATTR2]] {
10716 // CHECK17-NEXT: entry:
10717 // CHECK17-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 8
10718 // CHECK17-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 8
10719 // CHECK17-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i64, align 8
10720 // CHECK17-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i64, align 8
10721 // CHECK17-NEXT: [[N_ADDR:%.*]] = alloca ptr, align 8
10722 // CHECK17-NEXT: [[VLA_ADDR:%.*]] = alloca i64, align 8
10723 // CHECK17-NEXT: [[A_ADDR:%.*]] = alloca ptr, align 8
10724 // CHECK17-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
10725 // CHECK17-NEXT: [[TMP:%.*]] = alloca i32, align 4
10726 // CHECK17-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4
10727 // CHECK17-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4
10728 // CHECK17-NEXT: [[I:%.*]] = alloca i32, align 4
10729 // CHECK17-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4
10730 // CHECK17-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4
10731 // CHECK17-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
10732 // CHECK17-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
10733 // CHECK17-NEXT: [[I4:%.*]] = alloca i32, align 4
10734 // CHECK17-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 8
10735 // CHECK17-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 8
10736 // CHECK17-NEXT: store i64 [[DOTPREVIOUS_LB_]], ptr [[DOTPREVIOUS_LB__ADDR]], align 8
10737 // CHECK17-NEXT: store i64 [[DOTPREVIOUS_UB_]], ptr [[DOTPREVIOUS_UB__ADDR]], align 8
10738 // CHECK17-NEXT: store ptr [[N]], ptr [[N_ADDR]], align 8
10739 // CHECK17-NEXT: store i64 [[VLA]], ptr [[VLA_ADDR]], align 8
10740 // CHECK17-NEXT: store ptr [[A]], ptr [[A_ADDR]], align 8
10741 // CHECK17-NEXT: [[TMP0:%.*]] = load ptr, ptr [[N_ADDR]], align 8
10742 // CHECK17-NEXT: [[TMP1:%.*]] = load i64, ptr [[VLA_ADDR]], align 8
10743 // CHECK17-NEXT: [[TMP2:%.*]] = load ptr, ptr [[A_ADDR]], align 8
10744 // CHECK17-NEXT: [[TMP3:%.*]] = load i32, ptr [[TMP0]], align 4
10745 // CHECK17-NEXT: store i32 [[TMP3]], ptr [[DOTCAPTURE_EXPR_]], align 4
10746 // CHECK17-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4
10747 // CHECK17-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP4]], 0
10748 // CHECK17-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1
10749 // CHECK17-NEXT: [[SUB2:%.*]] = sub nsw i32 [[DIV]], 1
10750 // CHECK17-NEXT: store i32 [[SUB2]], ptr [[DOTCAPTURE_EXPR_1]], align 4
10751 // CHECK17-NEXT: store i32 0, ptr [[I]], align 4
10752 // CHECK17-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4
10753 // CHECK17-NEXT: [[CMP:%.*]] = icmp slt i32 0, [[TMP5]]
10754 // CHECK17-NEXT: br i1 [[CMP]], label [[OMP_PRECOND_THEN:%.*]], label [[OMP_PRECOND_END:%.*]]
10755 // CHECK17: omp.precond.then:
10756 // CHECK17-NEXT: store i32 0, ptr [[DOTOMP_LB]], align 4
10757 // CHECK17-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4
10758 // CHECK17-NEXT: store i32 [[TMP6]], ptr [[DOTOMP_UB]], align 4
10759 // CHECK17-NEXT: [[TMP7:%.*]] = load i64, ptr [[DOTPREVIOUS_LB__ADDR]], align 8
10760 // CHECK17-NEXT: [[CONV:%.*]] = trunc i64 [[TMP7]] to i32
10761 // CHECK17-NEXT: [[TMP8:%.*]] = load i64, ptr [[DOTPREVIOUS_UB__ADDR]], align 8
10762 // CHECK17-NEXT: [[CONV3:%.*]] = trunc i64 [[TMP8]] to i32
10763 // CHECK17-NEXT: store i32 [[CONV]], ptr [[DOTOMP_LB]], align 4
10764 // CHECK17-NEXT: store i32 [[CONV3]], ptr [[DOTOMP_UB]], align 4
10765 // CHECK17-NEXT: store i32 1, ptr [[DOTOMP_STRIDE]], align 4
10766 // CHECK17-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
10767 // CHECK17-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4
10768 // CHECK17-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
10769 // CHECK17-NEXT: [[TMP11:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
10770 // CHECK17-NEXT: [[TMP12:%.*]] = load i32, ptr [[TMP11]], align 4
10771 // CHECK17-NEXT: call void @__kmpc_dispatch_init_4(ptr @[[GLOB3]], i32 [[TMP12]], i32 35, i32 [[TMP9]], i32 [[TMP10]], i32 1, i32 1)
10772 // CHECK17-NEXT: br label [[OMP_DISPATCH_COND:%.*]]
10773 // CHECK17: omp.dispatch.cond:
10774 // CHECK17-NEXT: [[TMP13:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
10775 // CHECK17-NEXT: [[TMP14:%.*]] = load i32, ptr [[TMP13]], align 4
10776 // CHECK17-NEXT: [[TMP15:%.*]] = call i32 @__kmpc_dispatch_next_4(ptr @[[GLOB3]], i32 [[TMP14]], ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]])
10777 // CHECK17-NEXT: [[TOBOOL:%.*]] = icmp ne i32 [[TMP15]], 0
10778 // CHECK17-NEXT: br i1 [[TOBOOL]], label [[OMP_DISPATCH_BODY:%.*]], label [[OMP_DISPATCH_END:%.*]]
10779 // CHECK17: omp.dispatch.body:
10780 // CHECK17-NEXT: [[TMP16:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4
10781 // CHECK17-NEXT: store i32 [[TMP16]], ptr [[DOTOMP_IV]], align 4
10782 // CHECK17-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
10783 // CHECK17: omp.inner.for.cond:
10784 // CHECK17-NEXT: [[TMP17:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP15:![0-9]+]]
10785 // CHECK17-NEXT: [[TMP18:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP15]]
10786 // CHECK17-NEXT: [[CMP5:%.*]] = icmp sle i32 [[TMP17]], [[TMP18]]
10787 // CHECK17-NEXT: br i1 [[CMP5]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
10788 // CHECK17: omp.inner.for.body:
10789 // CHECK17-NEXT: [[TMP19:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP15]]
10790 // CHECK17-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP19]], 1
10791 // CHECK17-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
10792 // CHECK17-NEXT: store i32 [[ADD]], ptr [[I4]], align 4, !llvm.access.group [[ACC_GRP15]]
10793 // CHECK17-NEXT: [[TMP20:%.*]] = load i32, ptr [[I4]], align 4, !llvm.access.group [[ACC_GRP15]]
10794 // CHECK17-NEXT: [[IDXPROM:%.*]] = sext i32 [[TMP20]] to i64
10795 // CHECK17-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i32, ptr [[TMP2]], i64 [[IDXPROM]]
10796 // CHECK17-NEXT: store i32 0, ptr [[ARRAYIDX]], align 4, !llvm.access.group [[ACC_GRP15]]
10797 // CHECK17-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
10798 // CHECK17: omp.body.continue:
10799 // CHECK17-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
10800 // CHECK17: omp.inner.for.inc:
10801 // CHECK17-NEXT: [[TMP21:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP15]]
10802 // CHECK17-NEXT: [[ADD6:%.*]] = add nsw i32 [[TMP21]], 1
10803 // CHECK17-NEXT: store i32 [[ADD6]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP15]]
10804 // CHECK17-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP16:![0-9]+]]
10805 // CHECK17: omp.inner.for.end:
10806 // CHECK17-NEXT: br label [[OMP_DISPATCH_INC:%.*]]
10807 // CHECK17: omp.dispatch.inc:
10808 // CHECK17-NEXT: br label [[OMP_DISPATCH_COND]]
10809 // CHECK17: omp.dispatch.end:
10810 // CHECK17-NEXT: [[TMP22:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
10811 // CHECK17-NEXT: [[TMP23:%.*]] = load i32, ptr [[TMP22]], align 4
10812 // CHECK17-NEXT: call void @__kmpc_dispatch_deinit(ptr @[[GLOB3]], i32 [[TMP23]])
10813 // CHECK17-NEXT: br label [[OMP_PRECOND_END]]
10814 // CHECK17: omp.precond.end:
10815 // CHECK17-NEXT: ret void
10818 // CHECK17-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l168
10819 // CHECK17-SAME: (i64 noundef [[M:%.*]], i64 noundef [[N:%.*]], i64 noundef [[VLA:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[A:%.*]]) #[[ATTR2]] {
10820 // CHECK17-NEXT: entry:
10821 // CHECK17-NEXT: [[M_ADDR:%.*]] = alloca i64, align 8
10822 // CHECK17-NEXT: [[N_ADDR:%.*]] = alloca i64, align 8
10823 // CHECK17-NEXT: [[VLA_ADDR:%.*]] = alloca i64, align 8
10824 // CHECK17-NEXT: [[A_ADDR:%.*]] = alloca ptr, align 8
10825 // CHECK17-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4
10826 // CHECK17-NEXT: [[DOTCAPTURE_EXPR__CASTED:%.*]] = alloca i64, align 8
10827 // CHECK17-NEXT: store i64 [[M]], ptr [[M_ADDR]], align 8
10828 // CHECK17-NEXT: store i64 [[N]], ptr [[N_ADDR]], align 8
10829 // CHECK17-NEXT: store i64 [[VLA]], ptr [[VLA_ADDR]], align 8
10830 // CHECK17-NEXT: store ptr [[A]], ptr [[A_ADDR]], align 8
10831 // CHECK17-NEXT: [[TMP0:%.*]] = load i64, ptr [[VLA_ADDR]], align 8
10832 // CHECK17-NEXT: [[TMP1:%.*]] = load ptr, ptr [[A_ADDR]], align 8
10833 // CHECK17-NEXT: [[TMP2:%.*]] = load i32, ptr [[M_ADDR]], align 4
10834 // CHECK17-NEXT: store i32 [[TMP2]], ptr [[DOTCAPTURE_EXPR_]], align 4
10835 // CHECK17-NEXT: [[TMP3:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4
10836 // CHECK17-NEXT: store i32 [[TMP3]], ptr [[DOTCAPTURE_EXPR__CASTED]], align 4
10837 // CHECK17-NEXT: [[TMP4:%.*]] = load i64, ptr [[DOTCAPTURE_EXPR__CASTED]], align 8
10838 // CHECK17-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_teams(ptr @[[GLOB3]], i32 4, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l168.omp_outlined, ptr [[N_ADDR]], i64 [[TMP0]], ptr [[TMP1]], i64 [[TMP4]])
10839 // CHECK17-NEXT: ret void
10842 // CHECK17-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l168.omp_outlined
10843 // CHECK17-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[N:%.*]], i64 noundef [[VLA:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[A:%.*]], i64 noundef [[DOTCAPTURE_EXPR_:%.*]]) #[[ATTR2]] {
10844 // CHECK17-NEXT: entry:
10845 // CHECK17-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 8
10846 // CHECK17-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 8
10847 // CHECK17-NEXT: [[N_ADDR:%.*]] = alloca ptr, align 8
10848 // CHECK17-NEXT: [[VLA_ADDR:%.*]] = alloca i64, align 8
10849 // CHECK17-NEXT: [[A_ADDR:%.*]] = alloca ptr, align 8
10850 // CHECK17-NEXT: [[DOTCAPTURE_EXPR__ADDR:%.*]] = alloca i64, align 8
10851 // CHECK17-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
10852 // CHECK17-NEXT: [[TMP:%.*]] = alloca i32, align 4
10853 // CHECK17-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4
10854 // CHECK17-NEXT: [[DOTCAPTURE_EXPR_2:%.*]] = alloca i32, align 4
10855 // CHECK17-NEXT: [[I:%.*]] = alloca i32, align 4
10856 // CHECK17-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4
10857 // CHECK17-NEXT: [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4
10858 // CHECK17-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
10859 // CHECK17-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
10860 // CHECK17-NEXT: [[I4:%.*]] = alloca i32, align 4
10861 // CHECK17-NEXT: [[DOTCAPTURE_EXPR__CASTED:%.*]] = alloca i64, align 8
10862 // CHECK17-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 8
10863 // CHECK17-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 8
10864 // CHECK17-NEXT: store ptr [[N]], ptr [[N_ADDR]], align 8
10865 // CHECK17-NEXT: store i64 [[VLA]], ptr [[VLA_ADDR]], align 8
10866 // CHECK17-NEXT: store ptr [[A]], ptr [[A_ADDR]], align 8
10867 // CHECK17-NEXT: store i64 [[DOTCAPTURE_EXPR_]], ptr [[DOTCAPTURE_EXPR__ADDR]], align 8
10868 // CHECK17-NEXT: [[TMP0:%.*]] = load ptr, ptr [[N_ADDR]], align 8
10869 // CHECK17-NEXT: [[TMP1:%.*]] = load i64, ptr [[VLA_ADDR]], align 8
10870 // CHECK17-NEXT: [[TMP2:%.*]] = load ptr, ptr [[A_ADDR]], align 8
10871 // CHECK17-NEXT: [[TMP3:%.*]] = load i32, ptr [[TMP0]], align 4
10872 // CHECK17-NEXT: store i32 [[TMP3]], ptr [[DOTCAPTURE_EXPR_1]], align 4
10873 // CHECK17-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4
10874 // CHECK17-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP4]], 0
10875 // CHECK17-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1
10876 // CHECK17-NEXT: [[SUB3:%.*]] = sub nsw i32 [[DIV]], 1
10877 // CHECK17-NEXT: store i32 [[SUB3]], ptr [[DOTCAPTURE_EXPR_2]], align 4
10878 // CHECK17-NEXT: store i32 0, ptr [[I]], align 4
10879 // CHECK17-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4
10880 // CHECK17-NEXT: [[CMP:%.*]] = icmp slt i32 0, [[TMP5]]
10881 // CHECK17-NEXT: br i1 [[CMP]], label [[OMP_PRECOND_THEN:%.*]], label [[OMP_PRECOND_END:%.*]]
10882 // CHECK17: omp.precond.then:
10883 // CHECK17-NEXT: store i32 0, ptr [[DOTOMP_COMB_LB]], align 4
10884 // CHECK17-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_2]], align 4
10885 // CHECK17-NEXT: store i32 [[TMP6]], ptr [[DOTOMP_COMB_UB]], align 4
10886 // CHECK17-NEXT: store i32 1, ptr [[DOTOMP_STRIDE]], align 4
10887 // CHECK17-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
10888 // CHECK17-NEXT: [[TMP7:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
10889 // CHECK17-NEXT: [[TMP8:%.*]] = load i32, ptr [[TMP7]], align 4
10890 // CHECK17-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB1]], i32 [[TMP8]], i32 92, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
10891 // CHECK17-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
10892 // CHECK17-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_2]], align 4
10893 // CHECK17-NEXT: [[CMP5:%.*]] = icmp sgt i32 [[TMP9]], [[TMP10]]
10894 // CHECK17-NEXT: br i1 [[CMP5]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
10895 // CHECK17: cond.true:
10896 // CHECK17-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_2]], align 4
10897 // CHECK17-NEXT: br label [[COND_END:%.*]]
10898 // CHECK17: cond.false:
10899 // CHECK17-NEXT: [[TMP12:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
10900 // CHECK17-NEXT: br label [[COND_END]]
10901 // CHECK17: cond.end:
10902 // CHECK17-NEXT: [[COND:%.*]] = phi i32 [ [[TMP11]], [[COND_TRUE]] ], [ [[TMP12]], [[COND_FALSE]] ]
10903 // CHECK17-NEXT: store i32 [[COND]], ptr [[DOTOMP_COMB_UB]], align 4
10904 // CHECK17-NEXT: [[TMP13:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
10905 // CHECK17-NEXT: store i32 [[TMP13]], ptr [[DOTOMP_IV]], align 4
10906 // CHECK17-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
10907 // CHECK17: omp.inner.for.cond:
10908 // CHECK17-NEXT: [[TMP14:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
10909 // CHECK17-NEXT: [[TMP15:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
10910 // CHECK17-NEXT: [[CMP6:%.*]] = icmp sle i32 [[TMP14]], [[TMP15]]
10911 // CHECK17-NEXT: br i1 [[CMP6]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
10912 // CHECK17: omp.inner.for.body:
10913 // CHECK17-NEXT: [[TMP16:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
10914 // CHECK17-NEXT: [[TMP17:%.*]] = zext i32 [[TMP16]] to i64
10915 // CHECK17-NEXT: [[TMP18:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
10916 // CHECK17-NEXT: [[TMP19:%.*]] = zext i32 [[TMP18]] to i64
10917 // CHECK17-NEXT: [[TMP20:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR__ADDR]], align 4
10918 // CHECK17-NEXT: store i32 [[TMP20]], ptr [[DOTCAPTURE_EXPR__CASTED]], align 4
10919 // CHECK17-NEXT: [[TMP21:%.*]] = load i64, ptr [[DOTCAPTURE_EXPR__CASTED]], align 8
10920 // CHECK17-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_call(ptr @[[GLOB3]], i32 6, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l168.omp_outlined.omp_outlined, i64 [[TMP17]], i64 [[TMP19]], ptr [[TMP0]], i64 [[TMP1]], ptr [[TMP2]], i64 [[TMP21]])
10921 // CHECK17-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
10922 // CHECK17: omp.inner.for.inc:
10923 // CHECK17-NEXT: [[TMP22:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
10924 // CHECK17-NEXT: [[TMP23:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4
10925 // CHECK17-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP22]], [[TMP23]]
10926 // CHECK17-NEXT: store i32 [[ADD]], ptr [[DOTOMP_IV]], align 4
10927 // CHECK17-NEXT: br label [[OMP_INNER_FOR_COND]]
10928 // CHECK17: omp.inner.for.end:
10929 // CHECK17-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
10930 // CHECK17: omp.loop.exit:
10931 // CHECK17-NEXT: [[TMP24:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
10932 // CHECK17-NEXT: [[TMP25:%.*]] = load i32, ptr [[TMP24]], align 4
10933 // CHECK17-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB1]], i32 [[TMP25]])
10934 // CHECK17-NEXT: br label [[OMP_PRECOND_END]]
10935 // CHECK17: omp.precond.end:
10936 // CHECK17-NEXT: ret void
10939 // CHECK17-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l168.omp_outlined.omp_outlined
10940 // CHECK17-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], i64 noundef [[DOTPREVIOUS_LB_:%.*]], i64 noundef [[DOTPREVIOUS_UB_:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[N:%.*]], i64 noundef [[VLA:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[A:%.*]], i64 noundef [[DOTCAPTURE_EXPR_:%.*]]) #[[ATTR2]] {
10941 // CHECK17-NEXT: entry:
10942 // CHECK17-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 8
10943 // CHECK17-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 8
10944 // CHECK17-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i64, align 8
10945 // CHECK17-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i64, align 8
10946 // CHECK17-NEXT: [[N_ADDR:%.*]] = alloca ptr, align 8
10947 // CHECK17-NEXT: [[VLA_ADDR:%.*]] = alloca i64, align 8
10948 // CHECK17-NEXT: [[A_ADDR:%.*]] = alloca ptr, align 8
10949 // CHECK17-NEXT: [[DOTCAPTURE_EXPR__ADDR:%.*]] = alloca i64, align 8
10950 // CHECK17-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
10951 // CHECK17-NEXT: [[TMP:%.*]] = alloca i32, align 4
10952 // CHECK17-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4
10953 // CHECK17-NEXT: [[DOTCAPTURE_EXPR_2:%.*]] = alloca i32, align 4
10954 // CHECK17-NEXT: [[I:%.*]] = alloca i32, align 4
10955 // CHECK17-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4
10956 // CHECK17-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4
10957 // CHECK17-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
10958 // CHECK17-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
10959 // CHECK17-NEXT: [[I5:%.*]] = alloca i32, align 4
10960 // CHECK17-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 8
10961 // CHECK17-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 8
10962 // CHECK17-NEXT: store i64 [[DOTPREVIOUS_LB_]], ptr [[DOTPREVIOUS_LB__ADDR]], align 8
10963 // CHECK17-NEXT: store i64 [[DOTPREVIOUS_UB_]], ptr [[DOTPREVIOUS_UB__ADDR]], align 8
10964 // CHECK17-NEXT: store ptr [[N]], ptr [[N_ADDR]], align 8
10965 // CHECK17-NEXT: store i64 [[VLA]], ptr [[VLA_ADDR]], align 8
10966 // CHECK17-NEXT: store ptr [[A]], ptr [[A_ADDR]], align 8
10967 // CHECK17-NEXT: store i64 [[DOTCAPTURE_EXPR_]], ptr [[DOTCAPTURE_EXPR__ADDR]], align 8
10968 // CHECK17-NEXT: [[TMP0:%.*]] = load ptr, ptr [[N_ADDR]], align 8
10969 // CHECK17-NEXT: [[TMP1:%.*]] = load i64, ptr [[VLA_ADDR]], align 8
10970 // CHECK17-NEXT: [[TMP2:%.*]] = load ptr, ptr [[A_ADDR]], align 8
10971 // CHECK17-NEXT: [[TMP3:%.*]] = load i32, ptr [[TMP0]], align 4
10972 // CHECK17-NEXT: store i32 [[TMP3]], ptr [[DOTCAPTURE_EXPR_1]], align 4
10973 // CHECK17-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4
10974 // CHECK17-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP4]], 0
10975 // CHECK17-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1
10976 // CHECK17-NEXT: [[SUB3:%.*]] = sub nsw i32 [[DIV]], 1
10977 // CHECK17-NEXT: store i32 [[SUB3]], ptr [[DOTCAPTURE_EXPR_2]], align 4
10978 // CHECK17-NEXT: store i32 0, ptr [[I]], align 4
10979 // CHECK17-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4
10980 // CHECK17-NEXT: [[CMP:%.*]] = icmp slt i32 0, [[TMP5]]
10981 // CHECK17-NEXT: br i1 [[CMP]], label [[OMP_PRECOND_THEN:%.*]], label [[OMP_PRECOND_END:%.*]]
10982 // CHECK17: omp.precond.then:
10983 // CHECK17-NEXT: store i32 0, ptr [[DOTOMP_LB]], align 4
10984 // CHECK17-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_2]], align 4
10985 // CHECK17-NEXT: store i32 [[TMP6]], ptr [[DOTOMP_UB]], align 4
10986 // CHECK17-NEXT: [[TMP7:%.*]] = load i64, ptr [[DOTPREVIOUS_LB__ADDR]], align 8
10987 // CHECK17-NEXT: [[CONV:%.*]] = trunc i64 [[TMP7]] to i32
10988 // CHECK17-NEXT: [[TMP8:%.*]] = load i64, ptr [[DOTPREVIOUS_UB__ADDR]], align 8
10989 // CHECK17-NEXT: [[CONV4:%.*]] = trunc i64 [[TMP8]] to i32
10990 // CHECK17-NEXT: store i32 [[CONV]], ptr [[DOTOMP_LB]], align 4
10991 // CHECK17-NEXT: store i32 [[CONV4]], ptr [[DOTOMP_UB]], align 4
10992 // CHECK17-NEXT: store i32 1, ptr [[DOTOMP_STRIDE]], align 4
10993 // CHECK17-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
10994 // CHECK17-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR__ADDR]], align 4
10995 // CHECK17-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4
10996 // CHECK17-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
10997 // CHECK17-NEXT: [[TMP12:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
10998 // CHECK17-NEXT: [[TMP13:%.*]] = load i32, ptr [[TMP12]], align 4
10999 // CHECK17-NEXT: call void @__kmpc_dispatch_init_4(ptr @[[GLOB3]], i32 [[TMP13]], i32 35, i32 [[TMP10]], i32 [[TMP11]], i32 1, i32 [[TMP9]])
11000 // CHECK17-NEXT: br label [[OMP_DISPATCH_COND:%.*]]
11001 // CHECK17: omp.dispatch.cond:
11002 // CHECK17-NEXT: [[TMP14:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
11003 // CHECK17-NEXT: [[TMP15:%.*]] = load i32, ptr [[TMP14]], align 4
11004 // CHECK17-NEXT: [[TMP16:%.*]] = call i32 @__kmpc_dispatch_next_4(ptr @[[GLOB3]], i32 [[TMP15]], ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]])
11005 // CHECK17-NEXT: [[TOBOOL:%.*]] = icmp ne i32 [[TMP16]], 0
11006 // CHECK17-NEXT: br i1 [[TOBOOL]], label [[OMP_DISPATCH_BODY:%.*]], label [[OMP_DISPATCH_END:%.*]]
11007 // CHECK17: omp.dispatch.body:
11008 // CHECK17-NEXT: [[TMP17:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4
11009 // CHECK17-NEXT: store i32 [[TMP17]], ptr [[DOTOMP_IV]], align 4
11010 // CHECK17-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
11011 // CHECK17: omp.inner.for.cond:
11012 // CHECK17-NEXT: [[TMP18:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP18:![0-9]+]]
11013 // CHECK17-NEXT: [[TMP19:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP18]]
11014 // CHECK17-NEXT: [[CMP6:%.*]] = icmp sle i32 [[TMP18]], [[TMP19]]
11015 // CHECK17-NEXT: br i1 [[CMP6]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
11016 // CHECK17: omp.inner.for.body:
11017 // CHECK17-NEXT: [[TMP20:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP18]]
11018 // CHECK17-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP20]], 1
11019 // CHECK17-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
11020 // CHECK17-NEXT: store i32 [[ADD]], ptr [[I5]], align 4, !llvm.access.group [[ACC_GRP18]]
11021 // CHECK17-NEXT: [[TMP21:%.*]] = load i32, ptr [[I5]], align 4, !llvm.access.group [[ACC_GRP18]]
11022 // CHECK17-NEXT: [[IDXPROM:%.*]] = sext i32 [[TMP21]] to i64
11023 // CHECK17-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i32, ptr [[TMP2]], i64 [[IDXPROM]]
11024 // CHECK17-NEXT: store i32 0, ptr [[ARRAYIDX]], align 4, !llvm.access.group [[ACC_GRP18]]
11025 // CHECK17-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
11026 // CHECK17: omp.body.continue:
11027 // CHECK17-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
11028 // CHECK17: omp.inner.for.inc:
11029 // CHECK17-NEXT: [[TMP22:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP18]]
11030 // CHECK17-NEXT: [[ADD7:%.*]] = add nsw i32 [[TMP22]], 1
11031 // CHECK17-NEXT: store i32 [[ADD7]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP18]]
11032 // CHECK17-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP19:![0-9]+]]
11033 // CHECK17: omp.inner.for.end:
11034 // CHECK17-NEXT: br label [[OMP_DISPATCH_INC:%.*]]
11035 // CHECK17: omp.dispatch.inc:
11036 // CHECK17-NEXT: br label [[OMP_DISPATCH_COND]]
11037 // CHECK17: omp.dispatch.end:
11038 // CHECK17-NEXT: [[TMP23:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
11039 // CHECK17-NEXT: [[TMP24:%.*]] = load i32, ptr [[TMP23]], align 4
11040 // CHECK17-NEXT: call void @__kmpc_dispatch_deinit(ptr @[[GLOB3]], i32 [[TMP24]])
11041 // CHECK17-NEXT: br label [[OMP_PRECOND_END]]
11042 // CHECK17: omp.precond.end:
11043 // CHECK17-NEXT: ret void
11046 // CHECK17-LABEL: define {{[^@]+}}@_Z5tmainIiLi10EEiT_
11047 // CHECK17-SAME: (i32 noundef signext [[ARGC:%.*]]) #[[ATTR5:[0-9]+]] comdat {
11048 // CHECK17-NEXT: entry:
11049 // CHECK17-NEXT: [[ARGC_ADDR:%.*]] = alloca i32, align 4
11050 // CHECK17-NEXT: [[A:%.*]] = alloca [10 x i32], align 4
11051 // CHECK17-NEXT: [[M:%.*]] = alloca i32, align 4
11052 // CHECK17-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [1 x ptr], align 8
11053 // CHECK17-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [1 x ptr], align 8
11054 // CHECK17-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [1 x ptr], align 8
11055 // CHECK17-NEXT: [[TMP:%.*]] = alloca i32, align 4
11056 // CHECK17-NEXT: [[KERNEL_ARGS:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS:%.*]], align 8
11057 // CHECK17-NEXT: [[DOTOFFLOAD_BASEPTRS1:%.*]] = alloca [1 x ptr], align 8
11058 // CHECK17-NEXT: [[DOTOFFLOAD_PTRS2:%.*]] = alloca [1 x ptr], align 8
11059 // CHECK17-NEXT: [[DOTOFFLOAD_MAPPERS3:%.*]] = alloca [1 x ptr], align 8
11060 // CHECK17-NEXT: [[_TMP4:%.*]] = alloca i32, align 4
11061 // CHECK17-NEXT: [[KERNEL_ARGS5:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS]], align 8
11062 // CHECK17-NEXT: [[M_CASTED:%.*]] = alloca i64, align 8
11063 // CHECK17-NEXT: [[DOTOFFLOAD_BASEPTRS8:%.*]] = alloca [2 x ptr], align 8
11064 // CHECK17-NEXT: [[DOTOFFLOAD_PTRS9:%.*]] = alloca [2 x ptr], align 8
11065 // CHECK17-NEXT: [[DOTOFFLOAD_MAPPERS10:%.*]] = alloca [2 x ptr], align 8
11066 // CHECK17-NEXT: [[_TMP11:%.*]] = alloca i32, align 4
11067 // CHECK17-NEXT: [[KERNEL_ARGS12:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS]], align 8
11068 // CHECK17-NEXT: [[DOTOFFLOAD_BASEPTRS15:%.*]] = alloca [1 x ptr], align 8
11069 // CHECK17-NEXT: [[DOTOFFLOAD_PTRS16:%.*]] = alloca [1 x ptr], align 8
11070 // CHECK17-NEXT: [[DOTOFFLOAD_MAPPERS17:%.*]] = alloca [1 x ptr], align 8
11071 // CHECK17-NEXT: [[_TMP18:%.*]] = alloca i32, align 4
11072 // CHECK17-NEXT: [[KERNEL_ARGS19:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS]], align 8
11073 // CHECK17-NEXT: [[M_CASTED22:%.*]] = alloca i64, align 8
11074 // CHECK17-NEXT: [[DOTOFFLOAD_BASEPTRS23:%.*]] = alloca [2 x ptr], align 8
11075 // CHECK17-NEXT: [[DOTOFFLOAD_PTRS24:%.*]] = alloca [2 x ptr], align 8
11076 // CHECK17-NEXT: [[DOTOFFLOAD_MAPPERS25:%.*]] = alloca [2 x ptr], align 8
11077 // CHECK17-NEXT: [[_TMP26:%.*]] = alloca i32, align 4
11078 // CHECK17-NEXT: [[KERNEL_ARGS27:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS]], align 8
11079 // CHECK17-NEXT: store i32 [[ARGC]], ptr [[ARGC_ADDR]], align 4
11080 // CHECK17-NEXT: store i32 10, ptr [[M]], align 4
11081 // CHECK17-NEXT: [[TMP0:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
11082 // CHECK17-NEXT: store ptr [[A]], ptr [[TMP0]], align 8
11083 // CHECK17-NEXT: [[TMP1:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
11084 // CHECK17-NEXT: store ptr [[A]], ptr [[TMP1]], align 8
11085 // CHECK17-NEXT: [[TMP2:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0
11086 // CHECK17-NEXT: store ptr null, ptr [[TMP2]], align 8
11087 // CHECK17-NEXT: [[TMP3:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
11088 // CHECK17-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
11089 // CHECK17-NEXT: [[TMP5:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 0
11090 // CHECK17-NEXT: store i32 3, ptr [[TMP5]], align 4
11091 // CHECK17-NEXT: [[TMP6:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 1
11092 // CHECK17-NEXT: store i32 1, ptr [[TMP6]], align 4
11093 // CHECK17-NEXT: [[TMP7:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 2
11094 // CHECK17-NEXT: store ptr [[TMP3]], ptr [[TMP7]], align 8
11095 // CHECK17-NEXT: [[TMP8:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 3
11096 // CHECK17-NEXT: store ptr [[TMP4]], ptr [[TMP8]], align 8
11097 // CHECK17-NEXT: [[TMP9:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 4
11098 // CHECK17-NEXT: store ptr @.offload_sizes.9, ptr [[TMP9]], align 8
11099 // CHECK17-NEXT: [[TMP10:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 5
11100 // CHECK17-NEXT: store ptr @.offload_maptypes.10, ptr [[TMP10]], align 8
11101 // CHECK17-NEXT: [[TMP11:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 6
11102 // CHECK17-NEXT: store ptr null, ptr [[TMP11]], align 8
11103 // CHECK17-NEXT: [[TMP12:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 7
11104 // CHECK17-NEXT: store ptr null, ptr [[TMP12]], align 8
11105 // CHECK17-NEXT: [[TMP13:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 8
11106 // CHECK17-NEXT: store i64 10, ptr [[TMP13]], align 8
11107 // CHECK17-NEXT: [[TMP14:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 9
11108 // CHECK17-NEXT: store i64 0, ptr [[TMP14]], align 8
11109 // CHECK17-NEXT: [[TMP15:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 10
11110 // CHECK17-NEXT: store [3 x i32] zeroinitializer, ptr [[TMP15]], align 4
11111 // CHECK17-NEXT: [[TMP16:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 11
11112 // CHECK17-NEXT: store [3 x i32] zeroinitializer, ptr [[TMP16]], align 4
11113 // CHECK17-NEXT: [[TMP17:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 12
11114 // CHECK17-NEXT: store i32 0, ptr [[TMP17]], align 4
11115 // CHECK17-NEXT: [[TMP18:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB3]], i64 -1, i32 0, i32 0, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l116.region_id, ptr [[KERNEL_ARGS]])
11116 // CHECK17-NEXT: [[TMP19:%.*]] = icmp ne i32 [[TMP18]], 0
11117 // CHECK17-NEXT: br i1 [[TMP19]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
11118 // CHECK17: omp_offload.failed:
11119 // CHECK17-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l116(ptr [[A]]) #[[ATTR3]]
11120 // CHECK17-NEXT: br label [[OMP_OFFLOAD_CONT]]
11121 // CHECK17: omp_offload.cont:
11122 // CHECK17-NEXT: [[TMP20:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS1]], i32 0, i32 0
11123 // CHECK17-NEXT: store ptr [[A]], ptr [[TMP20]], align 8
11124 // CHECK17-NEXT: [[TMP21:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS2]], i32 0, i32 0
11125 // CHECK17-NEXT: store ptr [[A]], ptr [[TMP21]], align 8
11126 // CHECK17-NEXT: [[TMP22:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_MAPPERS3]], i64 0, i64 0
11127 // CHECK17-NEXT: store ptr null, ptr [[TMP22]], align 8
11128 // CHECK17-NEXT: [[TMP23:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS1]], i32 0, i32 0
11129 // CHECK17-NEXT: [[TMP24:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS2]], i32 0, i32 0
11130 // CHECK17-NEXT: [[TMP25:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS5]], i32 0, i32 0
11131 // CHECK17-NEXT: store i32 3, ptr [[TMP25]], align 4
11132 // CHECK17-NEXT: [[TMP26:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS5]], i32 0, i32 1
11133 // CHECK17-NEXT: store i32 1, ptr [[TMP26]], align 4
11134 // CHECK17-NEXT: [[TMP27:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS5]], i32 0, i32 2
11135 // CHECK17-NEXT: store ptr [[TMP23]], ptr [[TMP27]], align 8
11136 // CHECK17-NEXT: [[TMP28:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS5]], i32 0, i32 3
11137 // CHECK17-NEXT: store ptr [[TMP24]], ptr [[TMP28]], align 8
11138 // CHECK17-NEXT: [[TMP29:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS5]], i32 0, i32 4
11139 // CHECK17-NEXT: store ptr @.offload_sizes.11, ptr [[TMP29]], align 8
11140 // CHECK17-NEXT: [[TMP30:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS5]], i32 0, i32 5
11141 // CHECK17-NEXT: store ptr @.offload_maptypes.12, ptr [[TMP30]], align 8
11142 // CHECK17-NEXT: [[TMP31:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS5]], i32 0, i32 6
11143 // CHECK17-NEXT: store ptr null, ptr [[TMP31]], align 8
11144 // CHECK17-NEXT: [[TMP32:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS5]], i32 0, i32 7
11145 // CHECK17-NEXT: store ptr null, ptr [[TMP32]], align 8
11146 // CHECK17-NEXT: [[TMP33:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS5]], i32 0, i32 8
11147 // CHECK17-NEXT: store i64 10, ptr [[TMP33]], align 8
11148 // CHECK17-NEXT: [[TMP34:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS5]], i32 0, i32 9
11149 // CHECK17-NEXT: store i64 0, ptr [[TMP34]], align 8
11150 // CHECK17-NEXT: [[TMP35:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS5]], i32 0, i32 10
11151 // CHECK17-NEXT: store [3 x i32] zeroinitializer, ptr [[TMP35]], align 4
11152 // CHECK17-NEXT: [[TMP36:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS5]], i32 0, i32 11
11153 // CHECK17-NEXT: store [3 x i32] zeroinitializer, ptr [[TMP36]], align 4
11154 // CHECK17-NEXT: [[TMP37:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS5]], i32 0, i32 12
11155 // CHECK17-NEXT: store i32 0, ptr [[TMP37]], align 4
11156 // CHECK17-NEXT: [[TMP38:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB3]], i64 -1, i32 0, i32 0, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l121.region_id, ptr [[KERNEL_ARGS5]])
11157 // CHECK17-NEXT: [[TMP39:%.*]] = icmp ne i32 [[TMP38]], 0
11158 // CHECK17-NEXT: br i1 [[TMP39]], label [[OMP_OFFLOAD_FAILED6:%.*]], label [[OMP_OFFLOAD_CONT7:%.*]]
11159 // CHECK17: omp_offload.failed6:
11160 // CHECK17-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l121(ptr [[A]]) #[[ATTR3]]
11161 // CHECK17-NEXT: br label [[OMP_OFFLOAD_CONT7]]
11162 // CHECK17: omp_offload.cont7:
11163 // CHECK17-NEXT: [[TMP40:%.*]] = load i32, ptr [[M]], align 4
11164 // CHECK17-NEXT: store i32 [[TMP40]], ptr [[M_CASTED]], align 4
11165 // CHECK17-NEXT: [[TMP41:%.*]] = load i64, ptr [[M_CASTED]], align 8
11166 // CHECK17-NEXT: [[TMP42:%.*]] = getelementptr inbounds [2 x ptr], ptr [[DOTOFFLOAD_BASEPTRS8]], i32 0, i32 0
11167 // CHECK17-NEXT: store i64 [[TMP41]], ptr [[TMP42]], align 8
11168 // CHECK17-NEXT: [[TMP43:%.*]] = getelementptr inbounds [2 x ptr], ptr [[DOTOFFLOAD_PTRS9]], i32 0, i32 0
11169 // CHECK17-NEXT: store i64 [[TMP41]], ptr [[TMP43]], align 8
11170 // CHECK17-NEXT: [[TMP44:%.*]] = getelementptr inbounds [2 x ptr], ptr [[DOTOFFLOAD_MAPPERS10]], i64 0, i64 0
11171 // CHECK17-NEXT: store ptr null, ptr [[TMP44]], align 8
11172 // CHECK17-NEXT: [[TMP45:%.*]] = getelementptr inbounds [2 x ptr], ptr [[DOTOFFLOAD_BASEPTRS8]], i32 0, i32 1
11173 // CHECK17-NEXT: store ptr [[A]], ptr [[TMP45]], align 8
11174 // CHECK17-NEXT: [[TMP46:%.*]] = getelementptr inbounds [2 x ptr], ptr [[DOTOFFLOAD_PTRS9]], i32 0, i32 1
11175 // CHECK17-NEXT: store ptr [[A]], ptr [[TMP46]], align 8
11176 // CHECK17-NEXT: [[TMP47:%.*]] = getelementptr inbounds [2 x ptr], ptr [[DOTOFFLOAD_MAPPERS10]], i64 0, i64 1
11177 // CHECK17-NEXT: store ptr null, ptr [[TMP47]], align 8
11178 // CHECK17-NEXT: [[TMP48:%.*]] = getelementptr inbounds [2 x ptr], ptr [[DOTOFFLOAD_BASEPTRS8]], i32 0, i32 0
11179 // CHECK17-NEXT: [[TMP49:%.*]] = getelementptr inbounds [2 x ptr], ptr [[DOTOFFLOAD_PTRS9]], i32 0, i32 0
11180 // CHECK17-NEXT: [[TMP50:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS12]], i32 0, i32 0
11181 // CHECK17-NEXT: store i32 3, ptr [[TMP50]], align 4
11182 // CHECK17-NEXT: [[TMP51:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS12]], i32 0, i32 1
11183 // CHECK17-NEXT: store i32 2, ptr [[TMP51]], align 4
11184 // CHECK17-NEXT: [[TMP52:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS12]], i32 0, i32 2
11185 // CHECK17-NEXT: store ptr [[TMP48]], ptr [[TMP52]], align 8
11186 // CHECK17-NEXT: [[TMP53:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS12]], i32 0, i32 3
11187 // CHECK17-NEXT: store ptr [[TMP49]], ptr [[TMP53]], align 8
11188 // CHECK17-NEXT: [[TMP54:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS12]], i32 0, i32 4
11189 // CHECK17-NEXT: store ptr @.offload_sizes.13, ptr [[TMP54]], align 8
11190 // CHECK17-NEXT: [[TMP55:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS12]], i32 0, i32 5
11191 // CHECK17-NEXT: store ptr @.offload_maptypes.14, ptr [[TMP55]], align 8
11192 // CHECK17-NEXT: [[TMP56:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS12]], i32 0, i32 6
11193 // CHECK17-NEXT: store ptr null, ptr [[TMP56]], align 8
11194 // CHECK17-NEXT: [[TMP57:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS12]], i32 0, i32 7
11195 // CHECK17-NEXT: store ptr null, ptr [[TMP57]], align 8
11196 // CHECK17-NEXT: [[TMP58:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS12]], i32 0, i32 8
11197 // CHECK17-NEXT: store i64 10, ptr [[TMP58]], align 8
11198 // CHECK17-NEXT: [[TMP59:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS12]], i32 0, i32 9
11199 // CHECK17-NEXT: store i64 0, ptr [[TMP59]], align 8
11200 // CHECK17-NEXT: [[TMP60:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS12]], i32 0, i32 10
11201 // CHECK17-NEXT: store [3 x i32] zeroinitializer, ptr [[TMP60]], align 4
11202 // CHECK17-NEXT: [[TMP61:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS12]], i32 0, i32 11
11203 // CHECK17-NEXT: store [3 x i32] zeroinitializer, ptr [[TMP61]], align 4
11204 // CHECK17-NEXT: [[TMP62:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS12]], i32 0, i32 12
11205 // CHECK17-NEXT: store i32 0, ptr [[TMP62]], align 4
11206 // CHECK17-NEXT: [[TMP63:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB3]], i64 -1, i32 0, i32 0, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l126.region_id, ptr [[KERNEL_ARGS12]])
11207 // CHECK17-NEXT: [[TMP64:%.*]] = icmp ne i32 [[TMP63]], 0
11208 // CHECK17-NEXT: br i1 [[TMP64]], label [[OMP_OFFLOAD_FAILED13:%.*]], label [[OMP_OFFLOAD_CONT14:%.*]]
11209 // CHECK17: omp_offload.failed13:
11210 // CHECK17-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l126(i64 [[TMP41]], ptr [[A]]) #[[ATTR3]]
11211 // CHECK17-NEXT: br label [[OMP_OFFLOAD_CONT14]]
11212 // CHECK17: omp_offload.cont14:
11213 // CHECK17-NEXT: [[TMP65:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS15]], i32 0, i32 0
11214 // CHECK17-NEXT: store ptr [[A]], ptr [[TMP65]], align 8
11215 // CHECK17-NEXT: [[TMP66:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS16]], i32 0, i32 0
11216 // CHECK17-NEXT: store ptr [[A]], ptr [[TMP66]], align 8
11217 // CHECK17-NEXT: [[TMP67:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_MAPPERS17]], i64 0, i64 0
11218 // CHECK17-NEXT: store ptr null, ptr [[TMP67]], align 8
11219 // CHECK17-NEXT: [[TMP68:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS15]], i32 0, i32 0
11220 // CHECK17-NEXT: [[TMP69:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS16]], i32 0, i32 0
11221 // CHECK17-NEXT: [[TMP70:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS19]], i32 0, i32 0
11222 // CHECK17-NEXT: store i32 3, ptr [[TMP70]], align 4
11223 // CHECK17-NEXT: [[TMP71:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS19]], i32 0, i32 1
11224 // CHECK17-NEXT: store i32 1, ptr [[TMP71]], align 4
11225 // CHECK17-NEXT: [[TMP72:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS19]], i32 0, i32 2
11226 // CHECK17-NEXT: store ptr [[TMP68]], ptr [[TMP72]], align 8
11227 // CHECK17-NEXT: [[TMP73:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS19]], i32 0, i32 3
11228 // CHECK17-NEXT: store ptr [[TMP69]], ptr [[TMP73]], align 8
11229 // CHECK17-NEXT: [[TMP74:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS19]], i32 0, i32 4
11230 // CHECK17-NEXT: store ptr @.offload_sizes.15, ptr [[TMP74]], align 8
11231 // CHECK17-NEXT: [[TMP75:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS19]], i32 0, i32 5
11232 // CHECK17-NEXT: store ptr @.offload_maptypes.16, ptr [[TMP75]], align 8
11233 // CHECK17-NEXT: [[TMP76:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS19]], i32 0, i32 6
11234 // CHECK17-NEXT: store ptr null, ptr [[TMP76]], align 8
11235 // CHECK17-NEXT: [[TMP77:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS19]], i32 0, i32 7
11236 // CHECK17-NEXT: store ptr null, ptr [[TMP77]], align 8
11237 // CHECK17-NEXT: [[TMP78:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS19]], i32 0, i32 8
11238 // CHECK17-NEXT: store i64 10, ptr [[TMP78]], align 8
11239 // CHECK17-NEXT: [[TMP79:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS19]], i32 0, i32 9
11240 // CHECK17-NEXT: store i64 0, ptr [[TMP79]], align 8
11241 // CHECK17-NEXT: [[TMP80:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS19]], i32 0, i32 10
11242 // CHECK17-NEXT: store [3 x i32] zeroinitializer, ptr [[TMP80]], align 4
11243 // CHECK17-NEXT: [[TMP81:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS19]], i32 0, i32 11
11244 // CHECK17-NEXT: store [3 x i32] zeroinitializer, ptr [[TMP81]], align 4
11245 // CHECK17-NEXT: [[TMP82:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS19]], i32 0, i32 12
11246 // CHECK17-NEXT: store i32 0, ptr [[TMP82]], align 4
11247 // CHECK17-NEXT: [[TMP83:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB3]], i64 -1, i32 0, i32 0, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l131.region_id, ptr [[KERNEL_ARGS19]])
11248 // CHECK17-NEXT: [[TMP84:%.*]] = icmp ne i32 [[TMP83]], 0
11249 // CHECK17-NEXT: br i1 [[TMP84]], label [[OMP_OFFLOAD_FAILED20:%.*]], label [[OMP_OFFLOAD_CONT21:%.*]]
11250 // CHECK17: omp_offload.failed20:
11251 // CHECK17-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l131(ptr [[A]]) #[[ATTR3]]
11252 // CHECK17-NEXT: br label [[OMP_OFFLOAD_CONT21]]
11253 // CHECK17: omp_offload.cont21:
11254 // CHECK17-NEXT: [[TMP85:%.*]] = load i32, ptr [[M]], align 4
11255 // CHECK17-NEXT: store i32 [[TMP85]], ptr [[M_CASTED22]], align 4
11256 // CHECK17-NEXT: [[TMP86:%.*]] = load i64, ptr [[M_CASTED22]], align 8
11257 // CHECK17-NEXT: [[TMP87:%.*]] = getelementptr inbounds [2 x ptr], ptr [[DOTOFFLOAD_BASEPTRS23]], i32 0, i32 0
11258 // CHECK17-NEXT: store i64 [[TMP86]], ptr [[TMP87]], align 8
11259 // CHECK17-NEXT: [[TMP88:%.*]] = getelementptr inbounds [2 x ptr], ptr [[DOTOFFLOAD_PTRS24]], i32 0, i32 0
11260 // CHECK17-NEXT: store i64 [[TMP86]], ptr [[TMP88]], align 8
11261 // CHECK17-NEXT: [[TMP89:%.*]] = getelementptr inbounds [2 x ptr], ptr [[DOTOFFLOAD_MAPPERS25]], i64 0, i64 0
11262 // CHECK17-NEXT: store ptr null, ptr [[TMP89]], align 8
11263 // CHECK17-NEXT: [[TMP90:%.*]] = getelementptr inbounds [2 x ptr], ptr [[DOTOFFLOAD_BASEPTRS23]], i32 0, i32 1
11264 // CHECK17-NEXT: store ptr [[A]], ptr [[TMP90]], align 8
11265 // CHECK17-NEXT: [[TMP91:%.*]] = getelementptr inbounds [2 x ptr], ptr [[DOTOFFLOAD_PTRS24]], i32 0, i32 1
11266 // CHECK17-NEXT: store ptr [[A]], ptr [[TMP91]], align 8
11267 // CHECK17-NEXT: [[TMP92:%.*]] = getelementptr inbounds [2 x ptr], ptr [[DOTOFFLOAD_MAPPERS25]], i64 0, i64 1
11268 // CHECK17-NEXT: store ptr null, ptr [[TMP92]], align 8
11269 // CHECK17-NEXT: [[TMP93:%.*]] = getelementptr inbounds [2 x ptr], ptr [[DOTOFFLOAD_BASEPTRS23]], i32 0, i32 0
11270 // CHECK17-NEXT: [[TMP94:%.*]] = getelementptr inbounds [2 x ptr], ptr [[DOTOFFLOAD_PTRS24]], i32 0, i32 0
11271 // CHECK17-NEXT: [[TMP95:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS27]], i32 0, i32 0
11272 // CHECK17-NEXT: store i32 3, ptr [[TMP95]], align 4
11273 // CHECK17-NEXT: [[TMP96:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS27]], i32 0, i32 1
11274 // CHECK17-NEXT: store i32 2, ptr [[TMP96]], align 4
11275 // CHECK17-NEXT: [[TMP97:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS27]], i32 0, i32 2
11276 // CHECK17-NEXT: store ptr [[TMP93]], ptr [[TMP97]], align 8
11277 // CHECK17-NEXT: [[TMP98:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS27]], i32 0, i32 3
11278 // CHECK17-NEXT: store ptr [[TMP94]], ptr [[TMP98]], align 8
11279 // CHECK17-NEXT: [[TMP99:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS27]], i32 0, i32 4
11280 // CHECK17-NEXT: store ptr @.offload_sizes.17, ptr [[TMP99]], align 8
11281 // CHECK17-NEXT: [[TMP100:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS27]], i32 0, i32 5
11282 // CHECK17-NEXT: store ptr @.offload_maptypes.18, ptr [[TMP100]], align 8
11283 // CHECK17-NEXT: [[TMP101:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS27]], i32 0, i32 6
11284 // CHECK17-NEXT: store ptr null, ptr [[TMP101]], align 8
11285 // CHECK17-NEXT: [[TMP102:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS27]], i32 0, i32 7
11286 // CHECK17-NEXT: store ptr null, ptr [[TMP102]], align 8
11287 // CHECK17-NEXT: [[TMP103:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS27]], i32 0, i32 8
11288 // CHECK17-NEXT: store i64 10, ptr [[TMP103]], align 8
11289 // CHECK17-NEXT: [[TMP104:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS27]], i32 0, i32 9
11290 // CHECK17-NEXT: store i64 0, ptr [[TMP104]], align 8
11291 // CHECK17-NEXT: [[TMP105:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS27]], i32 0, i32 10
11292 // CHECK17-NEXT: store [3 x i32] zeroinitializer, ptr [[TMP105]], align 4
11293 // CHECK17-NEXT: [[TMP106:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS27]], i32 0, i32 11
11294 // CHECK17-NEXT: store [3 x i32] zeroinitializer, ptr [[TMP106]], align 4
11295 // CHECK17-NEXT: [[TMP107:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS27]], i32 0, i32 12
11296 // CHECK17-NEXT: store i32 0, ptr [[TMP107]], align 4
11297 // CHECK17-NEXT: [[TMP108:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB3]], i64 -1, i32 0, i32 0, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l136.region_id, ptr [[KERNEL_ARGS27]])
11298 // CHECK17-NEXT: [[TMP109:%.*]] = icmp ne i32 [[TMP108]], 0
11299 // CHECK17-NEXT: br i1 [[TMP109]], label [[OMP_OFFLOAD_FAILED28:%.*]], label [[OMP_OFFLOAD_CONT29:%.*]]
11300 // CHECK17: omp_offload.failed28:
11301 // CHECK17-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l136(i64 [[TMP86]], ptr [[A]]) #[[ATTR3]]
11302 // CHECK17-NEXT: br label [[OMP_OFFLOAD_CONT29]]
11303 // CHECK17: omp_offload.cont29:
11304 // CHECK17-NEXT: ret i32 0
11307 // CHECK17-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l116
11308 // CHECK17-SAME: (ptr noundef nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] {
11309 // CHECK17-NEXT: entry:
11310 // CHECK17-NEXT: [[A_ADDR:%.*]] = alloca ptr, align 8
11311 // CHECK17-NEXT: store ptr [[A]], ptr [[A_ADDR]], align 8
11312 // CHECK17-NEXT: [[TMP0:%.*]] = load ptr, ptr [[A_ADDR]], align 8
11313 // CHECK17-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_teams(ptr @[[GLOB3]], i32 1, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l116.omp_outlined, ptr [[TMP0]])
11314 // CHECK17-NEXT: ret void
11317 // CHECK17-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l116.omp_outlined
11318 // CHECK17-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], ptr noundef nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] {
11319 // CHECK17-NEXT: entry:
11320 // CHECK17-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 8
11321 // CHECK17-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 8
11322 // CHECK17-NEXT: [[A_ADDR:%.*]] = alloca ptr, align 8
11323 // CHECK17-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
11324 // CHECK17-NEXT: [[TMP:%.*]] = alloca i32, align 4
11325 // CHECK17-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4
11326 // CHECK17-NEXT: [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4
11327 // CHECK17-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
11328 // CHECK17-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
11329 // CHECK17-NEXT: [[I:%.*]] = alloca i32, align 4
11330 // CHECK17-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 8
11331 // CHECK17-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 8
11332 // CHECK17-NEXT: store ptr [[A]], ptr [[A_ADDR]], align 8
11333 // CHECK17-NEXT: [[TMP0:%.*]] = load ptr, ptr [[A_ADDR]], align 8
11334 // CHECK17-NEXT: store i32 0, ptr [[DOTOMP_COMB_LB]], align 4
11335 // CHECK17-NEXT: store i32 9, ptr [[DOTOMP_COMB_UB]], align 4
11336 // CHECK17-NEXT: store i32 1, ptr [[DOTOMP_STRIDE]], align 4
11337 // CHECK17-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
11338 // CHECK17-NEXT: [[TMP1:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
11339 // CHECK17-NEXT: [[TMP2:%.*]] = load i32, ptr [[TMP1]], align 4
11340 // CHECK17-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB1]], i32 [[TMP2]], i32 92, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
11341 // CHECK17-NEXT: [[TMP3:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
11342 // CHECK17-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP3]], 9
11343 // CHECK17-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
11344 // CHECK17: cond.true:
11345 // CHECK17-NEXT: br label [[COND_END:%.*]]
11346 // CHECK17: cond.false:
11347 // CHECK17-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
11348 // CHECK17-NEXT: br label [[COND_END]]
11349 // CHECK17: cond.end:
11350 // CHECK17-NEXT: [[COND:%.*]] = phi i32 [ 9, [[COND_TRUE]] ], [ [[TMP4]], [[COND_FALSE]] ]
11351 // CHECK17-NEXT: store i32 [[COND]], ptr [[DOTOMP_COMB_UB]], align 4
11352 // CHECK17-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
11353 // CHECK17-NEXT: store i32 [[TMP5]], ptr [[DOTOMP_IV]], align 4
11354 // CHECK17-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
11355 // CHECK17: omp.inner.for.cond:
11356 // CHECK17-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
11357 // CHECK17-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
11358 // CHECK17-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP6]], [[TMP7]]
11359 // CHECK17-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
11360 // CHECK17: omp.inner.for.body:
11361 // CHECK17-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
11362 // CHECK17-NEXT: [[TMP9:%.*]] = zext i32 [[TMP8]] to i64
11363 // CHECK17-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
11364 // CHECK17-NEXT: [[TMP11:%.*]] = zext i32 [[TMP10]] to i64
11365 // CHECK17-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_call(ptr @[[GLOB3]], i32 3, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l116.omp_outlined.omp_outlined, i64 [[TMP9]], i64 [[TMP11]], ptr [[TMP0]])
11366 // CHECK17-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
11367 // CHECK17: omp.inner.for.inc:
11368 // CHECK17-NEXT: [[TMP12:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
11369 // CHECK17-NEXT: [[TMP13:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4
11370 // CHECK17-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP12]], [[TMP13]]
11371 // CHECK17-NEXT: store i32 [[ADD]], ptr [[DOTOMP_IV]], align 4
11372 // CHECK17-NEXT: br label [[OMP_INNER_FOR_COND]]
11373 // CHECK17: omp.inner.for.end:
11374 // CHECK17-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
11375 // CHECK17: omp.loop.exit:
11376 // CHECK17-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB1]], i32 [[TMP2]])
11377 // CHECK17-NEXT: ret void
11380 // CHECK17-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l116.omp_outlined.omp_outlined
11381 // CHECK17-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], i64 noundef [[DOTPREVIOUS_LB_:%.*]], i64 noundef [[DOTPREVIOUS_UB_:%.*]], ptr noundef nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] {
11382 // CHECK17-NEXT: entry:
11383 // CHECK17-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 8
11384 // CHECK17-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 8
11385 // CHECK17-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i64, align 8
11386 // CHECK17-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i64, align 8
11387 // CHECK17-NEXT: [[A_ADDR:%.*]] = alloca ptr, align 8
11388 // CHECK17-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
11389 // CHECK17-NEXT: [[TMP:%.*]] = alloca i32, align 4
11390 // CHECK17-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4
11391 // CHECK17-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4
11392 // CHECK17-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
11393 // CHECK17-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
11394 // CHECK17-NEXT: [[I:%.*]] = alloca i32, align 4
11395 // CHECK17-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 8
11396 // CHECK17-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 8
11397 // CHECK17-NEXT: store i64 [[DOTPREVIOUS_LB_]], ptr [[DOTPREVIOUS_LB__ADDR]], align 8
11398 // CHECK17-NEXT: store i64 [[DOTPREVIOUS_UB_]], ptr [[DOTPREVIOUS_UB__ADDR]], align 8
11399 // CHECK17-NEXT: store ptr [[A]], ptr [[A_ADDR]], align 8
11400 // CHECK17-NEXT: [[TMP0:%.*]] = load ptr, ptr [[A_ADDR]], align 8
11401 // CHECK17-NEXT: store i32 0, ptr [[DOTOMP_LB]], align 4
11402 // CHECK17-NEXT: store i32 9, ptr [[DOTOMP_UB]], align 4
11403 // CHECK17-NEXT: [[TMP1:%.*]] = load i64, ptr [[DOTPREVIOUS_LB__ADDR]], align 8
11404 // CHECK17-NEXT: [[CONV:%.*]] = trunc i64 [[TMP1]] to i32
11405 // CHECK17-NEXT: [[TMP2:%.*]] = load i64, ptr [[DOTPREVIOUS_UB__ADDR]], align 8
11406 // CHECK17-NEXT: [[CONV1:%.*]] = trunc i64 [[TMP2]] to i32
11407 // CHECK17-NEXT: store i32 [[CONV]], ptr [[DOTOMP_LB]], align 4
11408 // CHECK17-NEXT: store i32 [[CONV1]], ptr [[DOTOMP_UB]], align 4
11409 // CHECK17-NEXT: store i32 1, ptr [[DOTOMP_STRIDE]], align 4
11410 // CHECK17-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
11411 // CHECK17-NEXT: [[TMP3:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
11412 // CHECK17-NEXT: [[TMP4:%.*]] = load i32, ptr [[TMP3]], align 4
11413 // CHECK17-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB2]], i32 [[TMP4]], i32 34, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
11414 // CHECK17-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
11415 // CHECK17-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP5]], 9
11416 // CHECK17-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
11417 // CHECK17: cond.true:
11418 // CHECK17-NEXT: br label [[COND_END:%.*]]
11419 // CHECK17: cond.false:
11420 // CHECK17-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
11421 // CHECK17-NEXT: br label [[COND_END]]
11422 // CHECK17: cond.end:
11423 // CHECK17-NEXT: [[COND:%.*]] = phi i32 [ 9, [[COND_TRUE]] ], [ [[TMP6]], [[COND_FALSE]] ]
11424 // CHECK17-NEXT: store i32 [[COND]], ptr [[DOTOMP_UB]], align 4
11425 // CHECK17-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4
11426 // CHECK17-NEXT: store i32 [[TMP7]], ptr [[DOTOMP_IV]], align 4
11427 // CHECK17-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
11428 // CHECK17: omp.inner.for.cond:
11429 // CHECK17-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
11430 // CHECK17-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
11431 // CHECK17-NEXT: [[CMP2:%.*]] = icmp sle i32 [[TMP8]], [[TMP9]]
11432 // CHECK17-NEXT: br i1 [[CMP2]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
11433 // CHECK17: omp.inner.for.body:
11434 // CHECK17-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
11435 // CHECK17-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP10]], 1
11436 // CHECK17-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
11437 // CHECK17-NEXT: store i32 [[ADD]], ptr [[I]], align 4
11438 // CHECK17-NEXT: [[TMP11:%.*]] = load i32, ptr [[I]], align 4
11439 // CHECK17-NEXT: [[IDXPROM:%.*]] = sext i32 [[TMP11]] to i64
11440 // CHECK17-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds [10 x i32], ptr [[TMP0]], i64 0, i64 [[IDXPROM]]
11441 // CHECK17-NEXT: store i32 0, ptr [[ARRAYIDX]], align 4
11442 // CHECK17-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
11443 // CHECK17: omp.body.continue:
11444 // CHECK17-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
11445 // CHECK17: omp.inner.for.inc:
11446 // CHECK17-NEXT: [[TMP12:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
11447 // CHECK17-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP12]], 1
11448 // CHECK17-NEXT: store i32 [[ADD3]], ptr [[DOTOMP_IV]], align 4
11449 // CHECK17-NEXT: br label [[OMP_INNER_FOR_COND]]
11450 // CHECK17: omp.inner.for.end:
11451 // CHECK17-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
11452 // CHECK17: omp.loop.exit:
11453 // CHECK17-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB2]], i32 [[TMP4]])
11454 // CHECK17-NEXT: ret void
11457 // CHECK17-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l121
11458 // CHECK17-SAME: (ptr noundef nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] {
11459 // CHECK17-NEXT: entry:
11460 // CHECK17-NEXT: [[A_ADDR:%.*]] = alloca ptr, align 8
11461 // CHECK17-NEXT: store ptr [[A]], ptr [[A_ADDR]], align 8
11462 // CHECK17-NEXT: [[TMP0:%.*]] = load ptr, ptr [[A_ADDR]], align 8
11463 // CHECK17-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_teams(ptr @[[GLOB3]], i32 1, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l121.omp_outlined, ptr [[TMP0]])
11464 // CHECK17-NEXT: ret void
11467 // CHECK17-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l121.omp_outlined
11468 // CHECK17-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], ptr noundef nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] {
11469 // CHECK17-NEXT: entry:
11470 // CHECK17-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 8
11471 // CHECK17-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 8
11472 // CHECK17-NEXT: [[A_ADDR:%.*]] = alloca ptr, align 8
11473 // CHECK17-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
11474 // CHECK17-NEXT: [[TMP:%.*]] = alloca i32, align 4
11475 // CHECK17-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4
11476 // CHECK17-NEXT: [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4
11477 // CHECK17-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
11478 // CHECK17-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
11479 // CHECK17-NEXT: [[I:%.*]] = alloca i32, align 4
11480 // CHECK17-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 8
11481 // CHECK17-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 8
11482 // CHECK17-NEXT: store ptr [[A]], ptr [[A_ADDR]], align 8
11483 // CHECK17-NEXT: [[TMP0:%.*]] = load ptr, ptr [[A_ADDR]], align 8
11484 // CHECK17-NEXT: store i32 0, ptr [[DOTOMP_COMB_LB]], align 4
11485 // CHECK17-NEXT: store i32 9, ptr [[DOTOMP_COMB_UB]], align 4
11486 // CHECK17-NEXT: store i32 1, ptr [[DOTOMP_STRIDE]], align 4
11487 // CHECK17-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
11488 // CHECK17-NEXT: [[TMP1:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
11489 // CHECK17-NEXT: [[TMP2:%.*]] = load i32, ptr [[TMP1]], align 4
11490 // CHECK17-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB1]], i32 [[TMP2]], i32 92, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
11491 // CHECK17-NEXT: [[TMP3:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
11492 // CHECK17-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP3]], 9
11493 // CHECK17-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
11494 // CHECK17: cond.true:
11495 // CHECK17-NEXT: br label [[COND_END:%.*]]
11496 // CHECK17: cond.false:
11497 // CHECK17-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
11498 // CHECK17-NEXT: br label [[COND_END]]
11499 // CHECK17: cond.end:
11500 // CHECK17-NEXT: [[COND:%.*]] = phi i32 [ 9, [[COND_TRUE]] ], [ [[TMP4]], [[COND_FALSE]] ]
11501 // CHECK17-NEXT: store i32 [[COND]], ptr [[DOTOMP_COMB_UB]], align 4
11502 // CHECK17-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
11503 // CHECK17-NEXT: store i32 [[TMP5]], ptr [[DOTOMP_IV]], align 4
11504 // CHECK17-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
11505 // CHECK17: omp.inner.for.cond:
11506 // CHECK17-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
11507 // CHECK17-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
11508 // CHECK17-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP6]], [[TMP7]]
11509 // CHECK17-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
11510 // CHECK17: omp.inner.for.body:
11511 // CHECK17-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
11512 // CHECK17-NEXT: [[TMP9:%.*]] = zext i32 [[TMP8]] to i64
11513 // CHECK17-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
11514 // CHECK17-NEXT: [[TMP11:%.*]] = zext i32 [[TMP10]] to i64
11515 // CHECK17-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_call(ptr @[[GLOB3]], i32 3, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l121.omp_outlined.omp_outlined, i64 [[TMP9]], i64 [[TMP11]], ptr [[TMP0]])
11516 // CHECK17-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
11517 // CHECK17: omp.inner.for.inc:
11518 // CHECK17-NEXT: [[TMP12:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
11519 // CHECK17-NEXT: [[TMP13:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4
11520 // CHECK17-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP12]], [[TMP13]]
11521 // CHECK17-NEXT: store i32 [[ADD]], ptr [[DOTOMP_IV]], align 4
11522 // CHECK17-NEXT: br label [[OMP_INNER_FOR_COND]]
11523 // CHECK17: omp.inner.for.end:
11524 // CHECK17-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
11525 // CHECK17: omp.loop.exit:
11526 // CHECK17-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB1]], i32 [[TMP2]])
11527 // CHECK17-NEXT: ret void
11530 // CHECK17-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l121.omp_outlined.omp_outlined
11531 // CHECK17-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], i64 noundef [[DOTPREVIOUS_LB_:%.*]], i64 noundef [[DOTPREVIOUS_UB_:%.*]], ptr noundef nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] {
11532 // CHECK17-NEXT: entry:
11533 // CHECK17-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 8
11534 // CHECK17-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 8
11535 // CHECK17-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i64, align 8
11536 // CHECK17-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i64, align 8
11537 // CHECK17-NEXT: [[A_ADDR:%.*]] = alloca ptr, align 8
11538 // CHECK17-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
11539 // CHECK17-NEXT: [[TMP:%.*]] = alloca i32, align 4
11540 // CHECK17-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4
11541 // CHECK17-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4
11542 // CHECK17-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
11543 // CHECK17-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
11544 // CHECK17-NEXT: [[I:%.*]] = alloca i32, align 4
11545 // CHECK17-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 8
11546 // CHECK17-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 8
11547 // CHECK17-NEXT: store i64 [[DOTPREVIOUS_LB_]], ptr [[DOTPREVIOUS_LB__ADDR]], align 8
11548 // CHECK17-NEXT: store i64 [[DOTPREVIOUS_UB_]], ptr [[DOTPREVIOUS_UB__ADDR]], align 8
11549 // CHECK17-NEXT: store ptr [[A]], ptr [[A_ADDR]], align 8
11550 // CHECK17-NEXT: [[TMP0:%.*]] = load ptr, ptr [[A_ADDR]], align 8
11551 // CHECK17-NEXT: store i32 0, ptr [[DOTOMP_LB]], align 4
11552 // CHECK17-NEXT: store i32 9, ptr [[DOTOMP_UB]], align 4
11553 // CHECK17-NEXT: [[TMP1:%.*]] = load i64, ptr [[DOTPREVIOUS_LB__ADDR]], align 8
11554 // CHECK17-NEXT: [[CONV:%.*]] = trunc i64 [[TMP1]] to i32
11555 // CHECK17-NEXT: [[TMP2:%.*]] = load i64, ptr [[DOTPREVIOUS_UB__ADDR]], align 8
11556 // CHECK17-NEXT: [[CONV1:%.*]] = trunc i64 [[TMP2]] to i32
11557 // CHECK17-NEXT: store i32 [[CONV]], ptr [[DOTOMP_LB]], align 4
11558 // CHECK17-NEXT: store i32 [[CONV1]], ptr [[DOTOMP_UB]], align 4
11559 // CHECK17-NEXT: store i32 1, ptr [[DOTOMP_STRIDE]], align 4
11560 // CHECK17-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
11561 // CHECK17-NEXT: [[TMP3:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
11562 // CHECK17-NEXT: [[TMP4:%.*]] = load i32, ptr [[TMP3]], align 4
11563 // CHECK17-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB2]], i32 [[TMP4]], i32 34, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
11564 // CHECK17-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
11565 // CHECK17-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP5]], 9
11566 // CHECK17-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
11567 // CHECK17: cond.true:
11568 // CHECK17-NEXT: br label [[COND_END:%.*]]
11569 // CHECK17: cond.false:
11570 // CHECK17-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
11571 // CHECK17-NEXT: br label [[COND_END]]
11572 // CHECK17: cond.end:
11573 // CHECK17-NEXT: [[COND:%.*]] = phi i32 [ 9, [[COND_TRUE]] ], [ [[TMP6]], [[COND_FALSE]] ]
11574 // CHECK17-NEXT: store i32 [[COND]], ptr [[DOTOMP_UB]], align 4
11575 // CHECK17-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4
11576 // CHECK17-NEXT: store i32 [[TMP7]], ptr [[DOTOMP_IV]], align 4
11577 // CHECK17-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
11578 // CHECK17: omp.inner.for.cond:
11579 // CHECK17-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
11580 // CHECK17-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
11581 // CHECK17-NEXT: [[CMP2:%.*]] = icmp sle i32 [[TMP8]], [[TMP9]]
11582 // CHECK17-NEXT: br i1 [[CMP2]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
11583 // CHECK17: omp.inner.for.body:
11584 // CHECK17-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
11585 // CHECK17-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP10]], 1
11586 // CHECK17-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
11587 // CHECK17-NEXT: store i32 [[ADD]], ptr [[I]], align 4
11588 // CHECK17-NEXT: [[TMP11:%.*]] = load i32, ptr [[I]], align 4
11589 // CHECK17-NEXT: [[IDXPROM:%.*]] = sext i32 [[TMP11]] to i64
11590 // CHECK17-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds [10 x i32], ptr [[TMP0]], i64 0, i64 [[IDXPROM]]
11591 // CHECK17-NEXT: store i32 0, ptr [[ARRAYIDX]], align 4
11592 // CHECK17-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
11593 // CHECK17: omp.body.continue:
11594 // CHECK17-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
11595 // CHECK17: omp.inner.for.inc:
11596 // CHECK17-NEXT: [[TMP12:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
11597 // CHECK17-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP12]], 1
11598 // CHECK17-NEXT: store i32 [[ADD3]], ptr [[DOTOMP_IV]], align 4
11599 // CHECK17-NEXT: br label [[OMP_INNER_FOR_COND]]
11600 // CHECK17: omp.inner.for.end:
11601 // CHECK17-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
11602 // CHECK17: omp.loop.exit:
11603 // CHECK17-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB2]], i32 [[TMP4]])
11604 // CHECK17-NEXT: ret void
11607 // CHECK17-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l126
11608 // CHECK17-SAME: (i64 noundef [[M:%.*]], ptr noundef nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] {
11609 // CHECK17-NEXT: entry:
11610 // CHECK17-NEXT: [[M_ADDR:%.*]] = alloca i64, align 8
11611 // CHECK17-NEXT: [[A_ADDR:%.*]] = alloca ptr, align 8
11612 // CHECK17-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4
11613 // CHECK17-NEXT: [[DOTCAPTURE_EXPR__CASTED:%.*]] = alloca i64, align 8
11614 // CHECK17-NEXT: store i64 [[M]], ptr [[M_ADDR]], align 8
11615 // CHECK17-NEXT: store ptr [[A]], ptr [[A_ADDR]], align 8
11616 // CHECK17-NEXT: [[TMP0:%.*]] = load ptr, ptr [[A_ADDR]], align 8
11617 // CHECK17-NEXT: [[TMP1:%.*]] = load i32, ptr [[M_ADDR]], align 4
11618 // CHECK17-NEXT: store i32 [[TMP1]], ptr [[DOTCAPTURE_EXPR_]], align 4
11619 // CHECK17-NEXT: [[TMP2:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4
11620 // CHECK17-NEXT: store i32 [[TMP2]], ptr [[DOTCAPTURE_EXPR__CASTED]], align 4
11621 // CHECK17-NEXT: [[TMP3:%.*]] = load i64, ptr [[DOTCAPTURE_EXPR__CASTED]], align 8
11622 // CHECK17-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_teams(ptr @[[GLOB3]], i32 2, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l126.omp_outlined, ptr [[TMP0]], i64 [[TMP3]])
11623 // CHECK17-NEXT: ret void
11626 // CHECK17-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l126.omp_outlined
11627 // CHECK17-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], ptr noundef nonnull align 4 dereferenceable(40) [[A:%.*]], i64 noundef [[DOTCAPTURE_EXPR_:%.*]]) #[[ATTR2]] {
11628 // CHECK17-NEXT: entry:
11629 // CHECK17-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 8
11630 // CHECK17-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 8
11631 // CHECK17-NEXT: [[A_ADDR:%.*]] = alloca ptr, align 8
11632 // CHECK17-NEXT: [[DOTCAPTURE_EXPR__ADDR:%.*]] = alloca i64, align 8
11633 // CHECK17-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
11634 // CHECK17-NEXT: [[TMP:%.*]] = alloca i32, align 4
11635 // CHECK17-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4
11636 // CHECK17-NEXT: [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4
11637 // CHECK17-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
11638 // CHECK17-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
11639 // CHECK17-NEXT: [[I:%.*]] = alloca i32, align 4
11640 // CHECK17-NEXT: [[DOTCAPTURE_EXPR__CASTED:%.*]] = alloca i64, align 8
11641 // CHECK17-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 8
11642 // CHECK17-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 8
11643 // CHECK17-NEXT: store ptr [[A]], ptr [[A_ADDR]], align 8
11644 // CHECK17-NEXT: store i64 [[DOTCAPTURE_EXPR_]], ptr [[DOTCAPTURE_EXPR__ADDR]], align 8
11645 // CHECK17-NEXT: [[TMP0:%.*]] = load ptr, ptr [[A_ADDR]], align 8
11646 // CHECK17-NEXT: store i32 0, ptr [[DOTOMP_COMB_LB]], align 4
11647 // CHECK17-NEXT: store i32 9, ptr [[DOTOMP_COMB_UB]], align 4
11648 // CHECK17-NEXT: store i32 1, ptr [[DOTOMP_STRIDE]], align 4
11649 // CHECK17-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
11650 // CHECK17-NEXT: [[TMP1:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
11651 // CHECK17-NEXT: [[TMP2:%.*]] = load i32, ptr [[TMP1]], align 4
11652 // CHECK17-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB1]], i32 [[TMP2]], i32 92, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
11653 // CHECK17-NEXT: [[TMP3:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
11654 // CHECK17-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP3]], 9
11655 // CHECK17-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
11656 // CHECK17: cond.true:
11657 // CHECK17-NEXT: br label [[COND_END:%.*]]
11658 // CHECK17: cond.false:
11659 // CHECK17-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
11660 // CHECK17-NEXT: br label [[COND_END]]
11661 // CHECK17: cond.end:
11662 // CHECK17-NEXT: [[COND:%.*]] = phi i32 [ 9, [[COND_TRUE]] ], [ [[TMP4]], [[COND_FALSE]] ]
11663 // CHECK17-NEXT: store i32 [[COND]], ptr [[DOTOMP_COMB_UB]], align 4
11664 // CHECK17-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
11665 // CHECK17-NEXT: store i32 [[TMP5]], ptr [[DOTOMP_IV]], align 4
11666 // CHECK17-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
11667 // CHECK17: omp.inner.for.cond:
11668 // CHECK17-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
11669 // CHECK17-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
11670 // CHECK17-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP6]], [[TMP7]]
11671 // CHECK17-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
11672 // CHECK17: omp.inner.for.body:
11673 // CHECK17-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
11674 // CHECK17-NEXT: [[TMP9:%.*]] = zext i32 [[TMP8]] to i64
11675 // CHECK17-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
11676 // CHECK17-NEXT: [[TMP11:%.*]] = zext i32 [[TMP10]] to i64
11677 // CHECK17-NEXT: [[TMP12:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR__ADDR]], align 4
11678 // CHECK17-NEXT: store i32 [[TMP12]], ptr [[DOTCAPTURE_EXPR__CASTED]], align 4
11679 // CHECK17-NEXT: [[TMP13:%.*]] = load i64, ptr [[DOTCAPTURE_EXPR__CASTED]], align 8
11680 // CHECK17-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_call(ptr @[[GLOB3]], i32 4, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l126.omp_outlined.omp_outlined, i64 [[TMP9]], i64 [[TMP11]], ptr [[TMP0]], i64 [[TMP13]])
11681 // CHECK17-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
11682 // CHECK17: omp.inner.for.inc:
11683 // CHECK17-NEXT: [[TMP14:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
11684 // CHECK17-NEXT: [[TMP15:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4
11685 // CHECK17-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP14]], [[TMP15]]
11686 // CHECK17-NEXT: store i32 [[ADD]], ptr [[DOTOMP_IV]], align 4
11687 // CHECK17-NEXT: br label [[OMP_INNER_FOR_COND]]
11688 // CHECK17: omp.inner.for.end:
11689 // CHECK17-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
11690 // CHECK17: omp.loop.exit:
11691 // CHECK17-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB1]], i32 [[TMP2]])
11692 // CHECK17-NEXT: ret void
11695 // CHECK17-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l126.omp_outlined.omp_outlined
11696 // CHECK17-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], i64 noundef [[DOTPREVIOUS_LB_:%.*]], i64 noundef [[DOTPREVIOUS_UB_:%.*]], ptr noundef nonnull align 4 dereferenceable(40) [[A:%.*]], i64 noundef [[DOTCAPTURE_EXPR_:%.*]]) #[[ATTR2]] {
11697 // CHECK17-NEXT: entry:
11698 // CHECK17-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 8
11699 // CHECK17-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 8
11700 // CHECK17-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i64, align 8
11701 // CHECK17-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i64, align 8
11702 // CHECK17-NEXT: [[A_ADDR:%.*]] = alloca ptr, align 8
11703 // CHECK17-NEXT: [[DOTCAPTURE_EXPR__ADDR:%.*]] = alloca i64, align 8
11704 // CHECK17-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
11705 // CHECK17-NEXT: [[TMP:%.*]] = alloca i32, align 4
11706 // CHECK17-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4
11707 // CHECK17-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4
11708 // CHECK17-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
11709 // CHECK17-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
11710 // CHECK17-NEXT: [[I:%.*]] = alloca i32, align 4
11711 // CHECK17-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 8
11712 // CHECK17-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 8
11713 // CHECK17-NEXT: store i64 [[DOTPREVIOUS_LB_]], ptr [[DOTPREVIOUS_LB__ADDR]], align 8
11714 // CHECK17-NEXT: store i64 [[DOTPREVIOUS_UB_]], ptr [[DOTPREVIOUS_UB__ADDR]], align 8
11715 // CHECK17-NEXT: store ptr [[A]], ptr [[A_ADDR]], align 8
11716 // CHECK17-NEXT: store i64 [[DOTCAPTURE_EXPR_]], ptr [[DOTCAPTURE_EXPR__ADDR]], align 8
11717 // CHECK17-NEXT: [[TMP0:%.*]] = load ptr, ptr [[A_ADDR]], align 8
11718 // CHECK17-NEXT: store i32 0, ptr [[DOTOMP_LB]], align 4
11719 // CHECK17-NEXT: store i32 9, ptr [[DOTOMP_UB]], align 4
11720 // CHECK17-NEXT: [[TMP1:%.*]] = load i64, ptr [[DOTPREVIOUS_LB__ADDR]], align 8
11721 // CHECK17-NEXT: [[CONV:%.*]] = trunc i64 [[TMP1]] to i32
11722 // CHECK17-NEXT: [[TMP2:%.*]] = load i64, ptr [[DOTPREVIOUS_UB__ADDR]], align 8
11723 // CHECK17-NEXT: [[CONV1:%.*]] = trunc i64 [[TMP2]] to i32
11724 // CHECK17-NEXT: store i32 [[CONV]], ptr [[DOTOMP_LB]], align 4
11725 // CHECK17-NEXT: store i32 [[CONV1]], ptr [[DOTOMP_UB]], align 4
11726 // CHECK17-NEXT: store i32 1, ptr [[DOTOMP_STRIDE]], align 4
11727 // CHECK17-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
11728 // CHECK17-NEXT: [[TMP3:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR__ADDR]], align 4
11729 // CHECK17-NEXT: [[TMP4:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
11730 // CHECK17-NEXT: [[TMP5:%.*]] = load i32, ptr [[TMP4]], align 4
11731 // CHECK17-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB2]], i32 [[TMP5]], i32 33, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 [[TMP3]])
11732 // CHECK17-NEXT: br label [[OMP_DISPATCH_COND:%.*]]
11733 // CHECK17: omp.dispatch.cond:
11734 // CHECK17-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
11735 // CHECK17-NEXT: [[TMP7:%.*]] = load i64, ptr [[DOTPREVIOUS_UB__ADDR]], align 8
11736 // CHECK17-NEXT: [[CONV2:%.*]] = trunc i64 [[TMP7]] to i32
11737 // CHECK17-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP6]], [[CONV2]]
11738 // CHECK17-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
11739 // CHECK17: cond.true:
11740 // CHECK17-NEXT: [[TMP8:%.*]] = load i64, ptr [[DOTPREVIOUS_UB__ADDR]], align 8
11741 // CHECK17-NEXT: [[CONV3:%.*]] = trunc i64 [[TMP8]] to i32
11742 // CHECK17-NEXT: br label [[COND_END:%.*]]
11743 // CHECK17: cond.false:
11744 // CHECK17-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
11745 // CHECK17-NEXT: br label [[COND_END]]
11746 // CHECK17: cond.end:
11747 // CHECK17-NEXT: [[COND:%.*]] = phi i32 [ [[CONV3]], [[COND_TRUE]] ], [ [[TMP9]], [[COND_FALSE]] ]
11748 // CHECK17-NEXT: store i32 [[COND]], ptr [[DOTOMP_UB]], align 4
11749 // CHECK17-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4
11750 // CHECK17-NEXT: store i32 [[TMP10]], ptr [[DOTOMP_IV]], align 4
11751 // CHECK17-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
11752 // CHECK17-NEXT: [[TMP12:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
11753 // CHECK17-NEXT: [[CMP4:%.*]] = icmp sle i32 [[TMP11]], [[TMP12]]
11754 // CHECK17-NEXT: br i1 [[CMP4]], label [[OMP_DISPATCH_BODY:%.*]], label [[OMP_DISPATCH_END:%.*]]
11755 // CHECK17: omp.dispatch.body:
11756 // CHECK17-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
11757 // CHECK17: omp.inner.for.cond:
11758 // CHECK17-NEXT: [[TMP13:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
11759 // CHECK17-NEXT: [[TMP14:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
11760 // CHECK17-NEXT: [[CMP5:%.*]] = icmp sle i32 [[TMP13]], [[TMP14]]
11761 // CHECK17-NEXT: br i1 [[CMP5]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
11762 // CHECK17: omp.inner.for.body:
11763 // CHECK17-NEXT: [[TMP15:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
11764 // CHECK17-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP15]], 1
11765 // CHECK17-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
11766 // CHECK17-NEXT: store i32 [[ADD]], ptr [[I]], align 4
11767 // CHECK17-NEXT: [[TMP16:%.*]] = load i32, ptr [[I]], align 4
11768 // CHECK17-NEXT: [[IDXPROM:%.*]] = sext i32 [[TMP16]] to i64
11769 // CHECK17-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds [10 x i32], ptr [[TMP0]], i64 0, i64 [[IDXPROM]]
11770 // CHECK17-NEXT: store i32 0, ptr [[ARRAYIDX]], align 4
11771 // CHECK17-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
11772 // CHECK17: omp.body.continue:
11773 // CHECK17-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
11774 // CHECK17: omp.inner.for.inc:
11775 // CHECK17-NEXT: [[TMP17:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
11776 // CHECK17-NEXT: [[ADD6:%.*]] = add nsw i32 [[TMP17]], 1
11777 // CHECK17-NEXT: store i32 [[ADD6]], ptr [[DOTOMP_IV]], align 4
11778 // CHECK17-NEXT: br label [[OMP_INNER_FOR_COND]]
11779 // CHECK17: omp.inner.for.end:
11780 // CHECK17-NEXT: br label [[OMP_DISPATCH_INC:%.*]]
11781 // CHECK17: omp.dispatch.inc:
11782 // CHECK17-NEXT: [[TMP18:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4
11783 // CHECK17-NEXT: [[TMP19:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4
11784 // CHECK17-NEXT: [[ADD7:%.*]] = add nsw i32 [[TMP18]], [[TMP19]]
11785 // CHECK17-NEXT: store i32 [[ADD7]], ptr [[DOTOMP_LB]], align 4
11786 // CHECK17-NEXT: [[TMP20:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
11787 // CHECK17-NEXT: [[TMP21:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4
11788 // CHECK17-NEXT: [[ADD8:%.*]] = add nsw i32 [[TMP20]], [[TMP21]]
11789 // CHECK17-NEXT: store i32 [[ADD8]], ptr [[DOTOMP_UB]], align 4
11790 // CHECK17-NEXT: br label [[OMP_DISPATCH_COND]]
11791 // CHECK17: omp.dispatch.end:
11792 // CHECK17-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB2]], i32 [[TMP5]])
11793 // CHECK17-NEXT: ret void
11796 // CHECK17-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l131
11797 // CHECK17-SAME: (ptr noundef nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] {
11798 // CHECK17-NEXT: entry:
11799 // CHECK17-NEXT: [[A_ADDR:%.*]] = alloca ptr, align 8
11800 // CHECK17-NEXT: store ptr [[A]], ptr [[A_ADDR]], align 8
11801 // CHECK17-NEXT: [[TMP0:%.*]] = load ptr, ptr [[A_ADDR]], align 8
11802 // CHECK17-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_teams(ptr @[[GLOB3]], i32 1, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l131.omp_outlined, ptr [[TMP0]])
11803 // CHECK17-NEXT: ret void
11806 // CHECK17-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l131.omp_outlined
11807 // CHECK17-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], ptr noundef nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] {
11808 // CHECK17-NEXT: entry:
11809 // CHECK17-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 8
11810 // CHECK17-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 8
11811 // CHECK17-NEXT: [[A_ADDR:%.*]] = alloca ptr, align 8
11812 // CHECK17-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
11813 // CHECK17-NEXT: [[TMP:%.*]] = alloca i32, align 4
11814 // CHECK17-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4
11815 // CHECK17-NEXT: [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4
11816 // CHECK17-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
11817 // CHECK17-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
11818 // CHECK17-NEXT: [[I:%.*]] = alloca i32, align 4
11819 // CHECK17-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 8
11820 // CHECK17-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 8
11821 // CHECK17-NEXT: store ptr [[A]], ptr [[A_ADDR]], align 8
11822 // CHECK17-NEXT: [[TMP0:%.*]] = load ptr, ptr [[A_ADDR]], align 8
11823 // CHECK17-NEXT: store i32 0, ptr [[DOTOMP_COMB_LB]], align 4
11824 // CHECK17-NEXT: store i32 9, ptr [[DOTOMP_COMB_UB]], align 4
11825 // CHECK17-NEXT: store i32 1, ptr [[DOTOMP_STRIDE]], align 4
11826 // CHECK17-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
11827 // CHECK17-NEXT: [[TMP1:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
11828 // CHECK17-NEXT: [[TMP2:%.*]] = load i32, ptr [[TMP1]], align 4
11829 // CHECK17-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB1]], i32 [[TMP2]], i32 92, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
11830 // CHECK17-NEXT: [[TMP3:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
11831 // CHECK17-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP3]], 9
11832 // CHECK17-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
11833 // CHECK17: cond.true:
11834 // CHECK17-NEXT: br label [[COND_END:%.*]]
11835 // CHECK17: cond.false:
11836 // CHECK17-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
11837 // CHECK17-NEXT: br label [[COND_END]]
11838 // CHECK17: cond.end:
11839 // CHECK17-NEXT: [[COND:%.*]] = phi i32 [ 9, [[COND_TRUE]] ], [ [[TMP4]], [[COND_FALSE]] ]
11840 // CHECK17-NEXT: store i32 [[COND]], ptr [[DOTOMP_COMB_UB]], align 4
11841 // CHECK17-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
11842 // CHECK17-NEXT: store i32 [[TMP5]], ptr [[DOTOMP_IV]], align 4
11843 // CHECK17-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
11844 // CHECK17: omp.inner.for.cond:
11845 // CHECK17-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
11846 // CHECK17-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
11847 // CHECK17-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP6]], [[TMP7]]
11848 // CHECK17-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
11849 // CHECK17: omp.inner.for.body:
11850 // CHECK17-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
11851 // CHECK17-NEXT: [[TMP9:%.*]] = zext i32 [[TMP8]] to i64
11852 // CHECK17-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
11853 // CHECK17-NEXT: [[TMP11:%.*]] = zext i32 [[TMP10]] to i64
11854 // CHECK17-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_call(ptr @[[GLOB3]], i32 3, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l131.omp_outlined.omp_outlined, i64 [[TMP9]], i64 [[TMP11]], ptr [[TMP0]])
11855 // CHECK17-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
11856 // CHECK17: omp.inner.for.inc:
11857 // CHECK17-NEXT: [[TMP12:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
11858 // CHECK17-NEXT: [[TMP13:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4
11859 // CHECK17-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP12]], [[TMP13]]
11860 // CHECK17-NEXT: store i32 [[ADD]], ptr [[DOTOMP_IV]], align 4
11861 // CHECK17-NEXT: br label [[OMP_INNER_FOR_COND]]
11862 // CHECK17: omp.inner.for.end:
11863 // CHECK17-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
11864 // CHECK17: omp.loop.exit:
11865 // CHECK17-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB1]], i32 [[TMP2]])
11866 // CHECK17-NEXT: ret void
11869 // CHECK17-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l131.omp_outlined.omp_outlined
11870 // CHECK17-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], i64 noundef [[DOTPREVIOUS_LB_:%.*]], i64 noundef [[DOTPREVIOUS_UB_:%.*]], ptr noundef nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] {
11871 // CHECK17-NEXT: entry:
11872 // CHECK17-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 8
11873 // CHECK17-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 8
11874 // CHECK17-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i64, align 8
11875 // CHECK17-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i64, align 8
11876 // CHECK17-NEXT: [[A_ADDR:%.*]] = alloca ptr, align 8
11877 // CHECK17-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
11878 // CHECK17-NEXT: [[TMP:%.*]] = alloca i32, align 4
11879 // CHECK17-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4
11880 // CHECK17-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4
11881 // CHECK17-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
11882 // CHECK17-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
11883 // CHECK17-NEXT: [[I:%.*]] = alloca i32, align 4
11884 // CHECK17-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 8
11885 // CHECK17-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 8
11886 // CHECK17-NEXT: store i64 [[DOTPREVIOUS_LB_]], ptr [[DOTPREVIOUS_LB__ADDR]], align 8
11887 // CHECK17-NEXT: store i64 [[DOTPREVIOUS_UB_]], ptr [[DOTPREVIOUS_UB__ADDR]], align 8
11888 // CHECK17-NEXT: store ptr [[A]], ptr [[A_ADDR]], align 8
11889 // CHECK17-NEXT: [[TMP0:%.*]] = load ptr, ptr [[A_ADDR]], align 8
11890 // CHECK17-NEXT: store i32 0, ptr [[DOTOMP_LB]], align 4
11891 // CHECK17-NEXT: store i32 9, ptr [[DOTOMP_UB]], align 4
11892 // CHECK17-NEXT: [[TMP1:%.*]] = load i64, ptr [[DOTPREVIOUS_LB__ADDR]], align 8
11893 // CHECK17-NEXT: [[CONV:%.*]] = trunc i64 [[TMP1]] to i32
11894 // CHECK17-NEXT: [[TMP2:%.*]] = load i64, ptr [[DOTPREVIOUS_UB__ADDR]], align 8
11895 // CHECK17-NEXT: [[CONV1:%.*]] = trunc i64 [[TMP2]] to i32
11896 // CHECK17-NEXT: store i32 [[CONV]], ptr [[DOTOMP_LB]], align 4
11897 // CHECK17-NEXT: store i32 [[CONV1]], ptr [[DOTOMP_UB]], align 4
11898 // CHECK17-NEXT: store i32 1, ptr [[DOTOMP_STRIDE]], align 4
11899 // CHECK17-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
11900 // CHECK17-NEXT: [[TMP3:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4
11901 // CHECK17-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
11902 // CHECK17-NEXT: [[TMP5:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
11903 // CHECK17-NEXT: [[TMP6:%.*]] = load i32, ptr [[TMP5]], align 4
11904 // CHECK17-NEXT: call void @__kmpc_dispatch_init_4(ptr @[[GLOB3]], i32 [[TMP6]], i32 35, i32 [[TMP3]], i32 [[TMP4]], i32 1, i32 1)
11905 // CHECK17-NEXT: br label [[OMP_DISPATCH_COND:%.*]]
11906 // CHECK17: omp.dispatch.cond:
11907 // CHECK17-NEXT: [[TMP7:%.*]] = call i32 @__kmpc_dispatch_next_4(ptr @[[GLOB3]], i32 [[TMP6]], ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]])
11908 // CHECK17-NEXT: [[TOBOOL:%.*]] = icmp ne i32 [[TMP7]], 0
11909 // CHECK17-NEXT: br i1 [[TOBOOL]], label [[OMP_DISPATCH_BODY:%.*]], label [[OMP_DISPATCH_END:%.*]]
11910 // CHECK17: omp.dispatch.body:
11911 // CHECK17-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4
11912 // CHECK17-NEXT: store i32 [[TMP8]], ptr [[DOTOMP_IV]], align 4
11913 // CHECK17-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
11914 // CHECK17: omp.inner.for.cond:
11915 // CHECK17-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP21:![0-9]+]]
11916 // CHECK17-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP21]]
11917 // CHECK17-NEXT: [[CMP:%.*]] = icmp sle i32 [[TMP9]], [[TMP10]]
11918 // CHECK17-NEXT: br i1 [[CMP]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
11919 // CHECK17: omp.inner.for.body:
11920 // CHECK17-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP21]]
11921 // CHECK17-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP11]], 1
11922 // CHECK17-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
11923 // CHECK17-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP21]]
11924 // CHECK17-NEXT: [[TMP12:%.*]] = load i32, ptr [[I]], align 4, !llvm.access.group [[ACC_GRP21]]
11925 // CHECK17-NEXT: [[IDXPROM:%.*]] = sext i32 [[TMP12]] to i64
11926 // CHECK17-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds [10 x i32], ptr [[TMP0]], i64 0, i64 [[IDXPROM]]
11927 // CHECK17-NEXT: store i32 0, ptr [[ARRAYIDX]], align 4, !llvm.access.group [[ACC_GRP21]]
11928 // CHECK17-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
11929 // CHECK17: omp.body.continue:
11930 // CHECK17-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
11931 // CHECK17: omp.inner.for.inc:
11932 // CHECK17-NEXT: [[TMP13:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP21]]
11933 // CHECK17-NEXT: [[ADD2:%.*]] = add nsw i32 [[TMP13]], 1
11934 // CHECK17-NEXT: store i32 [[ADD2]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP21]]
11935 // CHECK17-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP22:![0-9]+]]
11936 // CHECK17: omp.inner.for.end:
11937 // CHECK17-NEXT: br label [[OMP_DISPATCH_INC:%.*]]
11938 // CHECK17: omp.dispatch.inc:
11939 // CHECK17-NEXT: br label [[OMP_DISPATCH_COND]]
11940 // CHECK17: omp.dispatch.end:
11941 // CHECK17-NEXT: call void @__kmpc_dispatch_deinit(ptr @[[GLOB3]], i32 [[TMP6]])
11942 // CHECK17-NEXT: ret void
11945 // CHECK17-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l136
11946 // CHECK17-SAME: (i64 noundef [[M:%.*]], ptr noundef nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] {
11947 // CHECK17-NEXT: entry:
11948 // CHECK17-NEXT: [[M_ADDR:%.*]] = alloca i64, align 8
11949 // CHECK17-NEXT: [[A_ADDR:%.*]] = alloca ptr, align 8
11950 // CHECK17-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4
11951 // CHECK17-NEXT: [[DOTCAPTURE_EXPR__CASTED:%.*]] = alloca i64, align 8
11952 // CHECK17-NEXT: store i64 [[M]], ptr [[M_ADDR]], align 8
11953 // CHECK17-NEXT: store ptr [[A]], ptr [[A_ADDR]], align 8
11954 // CHECK17-NEXT: [[TMP0:%.*]] = load ptr, ptr [[A_ADDR]], align 8
11955 // CHECK17-NEXT: [[TMP1:%.*]] = load i32, ptr [[M_ADDR]], align 4
11956 // CHECK17-NEXT: store i32 [[TMP1]], ptr [[DOTCAPTURE_EXPR_]], align 4
11957 // CHECK17-NEXT: [[TMP2:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4
11958 // CHECK17-NEXT: store i32 [[TMP2]], ptr [[DOTCAPTURE_EXPR__CASTED]], align 4
11959 // CHECK17-NEXT: [[TMP3:%.*]] = load i64, ptr [[DOTCAPTURE_EXPR__CASTED]], align 8
11960 // CHECK17-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_teams(ptr @[[GLOB3]], i32 2, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l136.omp_outlined, ptr [[TMP0]], i64 [[TMP3]])
11961 // CHECK17-NEXT: ret void
11964 // CHECK17-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l136.omp_outlined
11965 // CHECK17-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], ptr noundef nonnull align 4 dereferenceable(40) [[A:%.*]], i64 noundef [[DOTCAPTURE_EXPR_:%.*]]) #[[ATTR2]] {
11966 // CHECK17-NEXT: entry:
11967 // CHECK17-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 8
11968 // CHECK17-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 8
11969 // CHECK17-NEXT: [[A_ADDR:%.*]] = alloca ptr, align 8
11970 // CHECK17-NEXT: [[DOTCAPTURE_EXPR__ADDR:%.*]] = alloca i64, align 8
11971 // CHECK17-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
11972 // CHECK17-NEXT: [[TMP:%.*]] = alloca i32, align 4
11973 // CHECK17-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4
11974 // CHECK17-NEXT: [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4
11975 // CHECK17-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
11976 // CHECK17-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
11977 // CHECK17-NEXT: [[I:%.*]] = alloca i32, align 4
11978 // CHECK17-NEXT: [[DOTCAPTURE_EXPR__CASTED:%.*]] = alloca i64, align 8
11979 // CHECK17-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 8
11980 // CHECK17-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 8
11981 // CHECK17-NEXT: store ptr [[A]], ptr [[A_ADDR]], align 8
11982 // CHECK17-NEXT: store i64 [[DOTCAPTURE_EXPR_]], ptr [[DOTCAPTURE_EXPR__ADDR]], align 8
11983 // CHECK17-NEXT: [[TMP0:%.*]] = load ptr, ptr [[A_ADDR]], align 8
11984 // CHECK17-NEXT: store i32 0, ptr [[DOTOMP_COMB_LB]], align 4
11985 // CHECK17-NEXT: store i32 9, ptr [[DOTOMP_COMB_UB]], align 4
11986 // CHECK17-NEXT: store i32 1, ptr [[DOTOMP_STRIDE]], align 4
11987 // CHECK17-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
11988 // CHECK17-NEXT: [[TMP1:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
11989 // CHECK17-NEXT: [[TMP2:%.*]] = load i32, ptr [[TMP1]], align 4
11990 // CHECK17-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB1]], i32 [[TMP2]], i32 92, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
11991 // CHECK17-NEXT: [[TMP3:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
11992 // CHECK17-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP3]], 9
11993 // CHECK17-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
11994 // CHECK17: cond.true:
11995 // CHECK17-NEXT: br label [[COND_END:%.*]]
11996 // CHECK17: cond.false:
11997 // CHECK17-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
11998 // CHECK17-NEXT: br label [[COND_END]]
11999 // CHECK17: cond.end:
12000 // CHECK17-NEXT: [[COND:%.*]] = phi i32 [ 9, [[COND_TRUE]] ], [ [[TMP4]], [[COND_FALSE]] ]
12001 // CHECK17-NEXT: store i32 [[COND]], ptr [[DOTOMP_COMB_UB]], align 4
12002 // CHECK17-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
12003 // CHECK17-NEXT: store i32 [[TMP5]], ptr [[DOTOMP_IV]], align 4
12004 // CHECK17-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
12005 // CHECK17: omp.inner.for.cond:
12006 // CHECK17-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
12007 // CHECK17-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
12008 // CHECK17-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP6]], [[TMP7]]
12009 // CHECK17-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
12010 // CHECK17: omp.inner.for.body:
12011 // CHECK17-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
12012 // CHECK17-NEXT: [[TMP9:%.*]] = zext i32 [[TMP8]] to i64
12013 // CHECK17-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
12014 // CHECK17-NEXT: [[TMP11:%.*]] = zext i32 [[TMP10]] to i64
12015 // CHECK17-NEXT: [[TMP12:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR__ADDR]], align 4
12016 // CHECK17-NEXT: store i32 [[TMP12]], ptr [[DOTCAPTURE_EXPR__CASTED]], align 4
12017 // CHECK17-NEXT: [[TMP13:%.*]] = load i64, ptr [[DOTCAPTURE_EXPR__CASTED]], align 8
12018 // CHECK17-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_call(ptr @[[GLOB3]], i32 4, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l136.omp_outlined.omp_outlined, i64 [[TMP9]], i64 [[TMP11]], ptr [[TMP0]], i64 [[TMP13]])
12019 // CHECK17-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
12020 // CHECK17: omp.inner.for.inc:
12021 // CHECK17-NEXT: [[TMP14:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
12022 // CHECK17-NEXT: [[TMP15:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4
12023 // CHECK17-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP14]], [[TMP15]]
12024 // CHECK17-NEXT: store i32 [[ADD]], ptr [[DOTOMP_IV]], align 4
12025 // CHECK17-NEXT: br label [[OMP_INNER_FOR_COND]]
12026 // CHECK17: omp.inner.for.end:
12027 // CHECK17-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
12028 // CHECK17: omp.loop.exit:
12029 // CHECK17-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB1]], i32 [[TMP2]])
12030 // CHECK17-NEXT: ret void
12033 // CHECK17-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l136.omp_outlined.omp_outlined
12034 // CHECK17-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], i64 noundef [[DOTPREVIOUS_LB_:%.*]], i64 noundef [[DOTPREVIOUS_UB_:%.*]], ptr noundef nonnull align 4 dereferenceable(40) [[A:%.*]], i64 noundef [[DOTCAPTURE_EXPR_:%.*]]) #[[ATTR2]] {
12035 // CHECK17-NEXT: entry:
12036 // CHECK17-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 8
12037 // CHECK17-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 8
12038 // CHECK17-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i64, align 8
12039 // CHECK17-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i64, align 8
12040 // CHECK17-NEXT: [[A_ADDR:%.*]] = alloca ptr, align 8
12041 // CHECK17-NEXT: [[DOTCAPTURE_EXPR__ADDR:%.*]] = alloca i64, align 8
12042 // CHECK17-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
12043 // CHECK17-NEXT: [[TMP:%.*]] = alloca i32, align 4
12044 // CHECK17-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4
12045 // CHECK17-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4
12046 // CHECK17-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
12047 // CHECK17-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
12048 // CHECK17-NEXT: [[I:%.*]] = alloca i32, align 4
12049 // CHECK17-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 8
12050 // CHECK17-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 8
12051 // CHECK17-NEXT: store i64 [[DOTPREVIOUS_LB_]], ptr [[DOTPREVIOUS_LB__ADDR]], align 8
12052 // CHECK17-NEXT: store i64 [[DOTPREVIOUS_UB_]], ptr [[DOTPREVIOUS_UB__ADDR]], align 8
12053 // CHECK17-NEXT: store ptr [[A]], ptr [[A_ADDR]], align 8
12054 // CHECK17-NEXT: store i64 [[DOTCAPTURE_EXPR_]], ptr [[DOTCAPTURE_EXPR__ADDR]], align 8
12055 // CHECK17-NEXT: [[TMP0:%.*]] = load ptr, ptr [[A_ADDR]], align 8
12056 // CHECK17-NEXT: store i32 0, ptr [[DOTOMP_LB]], align 4
12057 // CHECK17-NEXT: store i32 9, ptr [[DOTOMP_UB]], align 4
12058 // CHECK17-NEXT: [[TMP1:%.*]] = load i64, ptr [[DOTPREVIOUS_LB__ADDR]], align 8
12059 // CHECK17-NEXT: [[CONV:%.*]] = trunc i64 [[TMP1]] to i32
12060 // CHECK17-NEXT: [[TMP2:%.*]] = load i64, ptr [[DOTPREVIOUS_UB__ADDR]], align 8
12061 // CHECK17-NEXT: [[CONV1:%.*]] = trunc i64 [[TMP2]] to i32
12062 // CHECK17-NEXT: store i32 [[CONV]], ptr [[DOTOMP_LB]], align 4
12063 // CHECK17-NEXT: store i32 [[CONV1]], ptr [[DOTOMP_UB]], align 4
12064 // CHECK17-NEXT: store i32 1, ptr [[DOTOMP_STRIDE]], align 4
12065 // CHECK17-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
12066 // CHECK17-NEXT: [[TMP3:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR__ADDR]], align 4
12067 // CHECK17-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4
12068 // CHECK17-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
12069 // CHECK17-NEXT: [[TMP6:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
12070 // CHECK17-NEXT: [[TMP7:%.*]] = load i32, ptr [[TMP6]], align 4
12071 // CHECK17-NEXT: call void @__kmpc_dispatch_init_4(ptr @[[GLOB3]], i32 [[TMP7]], i32 35, i32 [[TMP4]], i32 [[TMP5]], i32 1, i32 [[TMP3]])
12072 // CHECK17-NEXT: br label [[OMP_DISPATCH_COND:%.*]]
12073 // CHECK17: omp.dispatch.cond:
12074 // CHECK17-NEXT: [[TMP8:%.*]] = call i32 @__kmpc_dispatch_next_4(ptr @[[GLOB3]], i32 [[TMP7]], ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]])
12075 // CHECK17-NEXT: [[TOBOOL:%.*]] = icmp ne i32 [[TMP8]], 0
12076 // CHECK17-NEXT: br i1 [[TOBOOL]], label [[OMP_DISPATCH_BODY:%.*]], label [[OMP_DISPATCH_END:%.*]]
12077 // CHECK17: omp.dispatch.body:
12078 // CHECK17-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4
12079 // CHECK17-NEXT: store i32 [[TMP9]], ptr [[DOTOMP_IV]], align 4
12080 // CHECK17-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
12081 // CHECK17: omp.inner.for.cond:
12082 // CHECK17-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP24:![0-9]+]]
12083 // CHECK17-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP24]]
12084 // CHECK17-NEXT: [[CMP:%.*]] = icmp sle i32 [[TMP10]], [[TMP11]]
12085 // CHECK17-NEXT: br i1 [[CMP]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
12086 // CHECK17: omp.inner.for.body:
12087 // CHECK17-NEXT: [[TMP12:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP24]]
12088 // CHECK17-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP12]], 1
12089 // CHECK17-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
12090 // CHECK17-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP24]]
12091 // CHECK17-NEXT: [[TMP13:%.*]] = load i32, ptr [[I]], align 4, !llvm.access.group [[ACC_GRP24]]
12092 // CHECK17-NEXT: [[IDXPROM:%.*]] = sext i32 [[TMP13]] to i64
12093 // CHECK17-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds [10 x i32], ptr [[TMP0]], i64 0, i64 [[IDXPROM]]
12094 // CHECK17-NEXT: store i32 0, ptr [[ARRAYIDX]], align 4, !llvm.access.group [[ACC_GRP24]]
12095 // CHECK17-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
12096 // CHECK17: omp.body.continue:
12097 // CHECK17-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
12098 // CHECK17: omp.inner.for.inc:
12099 // CHECK17-NEXT: [[TMP14:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP24]]
12100 // CHECK17-NEXT: [[ADD2:%.*]] = add nsw i32 [[TMP14]], 1
12101 // CHECK17-NEXT: store i32 [[ADD2]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP24]]
12102 // CHECK17-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP25:![0-9]+]]
12103 // CHECK17: omp.inner.for.end:
12104 // CHECK17-NEXT: br label [[OMP_DISPATCH_INC:%.*]]
12105 // CHECK17: omp.dispatch.inc:
12106 // CHECK17-NEXT: br label [[OMP_DISPATCH_COND]]
12107 // CHECK17: omp.dispatch.end:
12108 // CHECK17-NEXT: call void @__kmpc_dispatch_deinit(ptr @[[GLOB3]], i32 [[TMP7]])
12109 // CHECK17-NEXT: ret void
12112 // CHECK19-LABEL: define {{[^@]+}}@main
12113 // CHECK19-SAME: (i32 noundef [[ARGC:%.*]], ptr noundef [[ARGV:%.*]]) #[[ATTR0:[0-9]+]] {
12114 // CHECK19-NEXT: entry:
12115 // CHECK19-NEXT: [[RETVAL:%.*]] = alloca i32, align 4
12116 // CHECK19-NEXT: [[ARGC_ADDR:%.*]] = alloca i32, align 4
12117 // CHECK19-NEXT: [[ARGV_ADDR:%.*]] = alloca ptr, align 4
12118 // CHECK19-NEXT: [[N:%.*]] = alloca i32, align 4
12119 // CHECK19-NEXT: [[SAVED_STACK:%.*]] = alloca ptr, align 4
12120 // CHECK19-NEXT: [[__VLA_EXPR0:%.*]] = alloca i32, align 4
12121 // CHECK19-NEXT: [[M:%.*]] = alloca i32, align 4
12122 // CHECK19-NEXT: [[N_CASTED:%.*]] = alloca i32, align 4
12123 // CHECK19-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [3 x ptr], align 4
12124 // CHECK19-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [3 x ptr], align 4
12125 // CHECK19-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [3 x ptr], align 4
12126 // CHECK19-NEXT: [[DOTOFFLOAD_SIZES:%.*]] = alloca [3 x i64], align 4
12127 // CHECK19-NEXT: [[TMP:%.*]] = alloca i32, align 4
12128 // CHECK19-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4
12129 // CHECK19-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4
12130 // CHECK19-NEXT: [[KERNEL_ARGS:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS:%.*]], align 8
12131 // CHECK19-NEXT: [[N_CASTED3:%.*]] = alloca i32, align 4
12132 // CHECK19-NEXT: [[DOTOFFLOAD_BASEPTRS4:%.*]] = alloca [3 x ptr], align 4
12133 // CHECK19-NEXT: [[DOTOFFLOAD_PTRS5:%.*]] = alloca [3 x ptr], align 4
12134 // CHECK19-NEXT: [[DOTOFFLOAD_MAPPERS6:%.*]] = alloca [3 x ptr], align 4
12135 // CHECK19-NEXT: [[DOTOFFLOAD_SIZES7:%.*]] = alloca [3 x i64], align 4
12136 // CHECK19-NEXT: [[_TMP8:%.*]] = alloca i32, align 4
12137 // CHECK19-NEXT: [[DOTCAPTURE_EXPR_9:%.*]] = alloca i32, align 4
12138 // CHECK19-NEXT: [[DOTCAPTURE_EXPR_10:%.*]] = alloca i32, align 4
12139 // CHECK19-NEXT: [[KERNEL_ARGS15:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS]], align 8
12140 // CHECK19-NEXT: [[M_CASTED:%.*]] = alloca i32, align 4
12141 // CHECK19-NEXT: [[N_CASTED18:%.*]] = alloca i32, align 4
12142 // CHECK19-NEXT: [[DOTOFFLOAD_BASEPTRS19:%.*]] = alloca [4 x ptr], align 4
12143 // CHECK19-NEXT: [[DOTOFFLOAD_PTRS20:%.*]] = alloca [4 x ptr], align 4
12144 // CHECK19-NEXT: [[DOTOFFLOAD_MAPPERS21:%.*]] = alloca [4 x ptr], align 4
12145 // CHECK19-NEXT: [[DOTOFFLOAD_SIZES22:%.*]] = alloca [4 x i64], align 4
12146 // CHECK19-NEXT: [[_TMP23:%.*]] = alloca i32, align 4
12147 // CHECK19-NEXT: [[DOTCAPTURE_EXPR_24:%.*]] = alloca i32, align 4
12148 // CHECK19-NEXT: [[DOTCAPTURE_EXPR_25:%.*]] = alloca i32, align 4
12149 // CHECK19-NEXT: [[KERNEL_ARGS30:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS]], align 8
12150 // CHECK19-NEXT: [[N_CASTED33:%.*]] = alloca i32, align 4
12151 // CHECK19-NEXT: [[DOTOFFLOAD_BASEPTRS34:%.*]] = alloca [3 x ptr], align 4
12152 // CHECK19-NEXT: [[DOTOFFLOAD_PTRS35:%.*]] = alloca [3 x ptr], align 4
12153 // CHECK19-NEXT: [[DOTOFFLOAD_MAPPERS36:%.*]] = alloca [3 x ptr], align 4
12154 // CHECK19-NEXT: [[DOTOFFLOAD_SIZES37:%.*]] = alloca [3 x i64], align 4
12155 // CHECK19-NEXT: [[_TMP38:%.*]] = alloca i32, align 4
12156 // CHECK19-NEXT: [[DOTCAPTURE_EXPR_39:%.*]] = alloca i32, align 4
12157 // CHECK19-NEXT: [[DOTCAPTURE_EXPR_40:%.*]] = alloca i32, align 4
12158 // CHECK19-NEXT: [[KERNEL_ARGS45:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS]], align 8
12159 // CHECK19-NEXT: [[M_CASTED48:%.*]] = alloca i32, align 4
12160 // CHECK19-NEXT: [[N_CASTED49:%.*]] = alloca i32, align 4
12161 // CHECK19-NEXT: [[DOTOFFLOAD_BASEPTRS50:%.*]] = alloca [4 x ptr], align 4
12162 // CHECK19-NEXT: [[DOTOFFLOAD_PTRS51:%.*]] = alloca [4 x ptr], align 4
12163 // CHECK19-NEXT: [[DOTOFFLOAD_MAPPERS52:%.*]] = alloca [4 x ptr], align 4
12164 // CHECK19-NEXT: [[DOTOFFLOAD_SIZES53:%.*]] = alloca [4 x i64], align 4
12165 // CHECK19-NEXT: [[_TMP54:%.*]] = alloca i32, align 4
12166 // CHECK19-NEXT: [[DOTCAPTURE_EXPR_55:%.*]] = alloca i32, align 4
12167 // CHECK19-NEXT: [[DOTCAPTURE_EXPR_56:%.*]] = alloca i32, align 4
12168 // CHECK19-NEXT: [[KERNEL_ARGS61:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS]], align 8
12169 // CHECK19-NEXT: store i32 0, ptr [[RETVAL]], align 4
12170 // CHECK19-NEXT: store i32 [[ARGC]], ptr [[ARGC_ADDR]], align 4
12171 // CHECK19-NEXT: store ptr [[ARGV]], ptr [[ARGV_ADDR]], align 4
12172 // CHECK19-NEXT: store i32 100, ptr [[N]], align 4
12173 // CHECK19-NEXT: [[TMP0:%.*]] = load i32, ptr [[N]], align 4
12174 // CHECK19-NEXT: [[TMP1:%.*]] = call ptr @llvm.stacksave.p0()
12175 // CHECK19-NEXT: store ptr [[TMP1]], ptr [[SAVED_STACK]], align 4
12176 // CHECK19-NEXT: [[VLA:%.*]] = alloca i32, i32 [[TMP0]], align 4
12177 // CHECK19-NEXT: store i32 [[TMP0]], ptr [[__VLA_EXPR0]], align 4
12178 // CHECK19-NEXT: store i32 10, ptr [[M]], align 4
12179 // CHECK19-NEXT: [[TMP2:%.*]] = load i32, ptr [[N]], align 4
12180 // CHECK19-NEXT: store i32 [[TMP2]], ptr [[N_CASTED]], align 4
12181 // CHECK19-NEXT: [[TMP3:%.*]] = load i32, ptr [[N_CASTED]], align 4
12182 // CHECK19-NEXT: [[TMP4:%.*]] = mul nuw i32 [[TMP0]], 4
12183 // CHECK19-NEXT: [[TMP5:%.*]] = sext i32 [[TMP4]] to i64
12184 // CHECK19-NEXT: call void @llvm.memcpy.p0.p0.i32(ptr align 4 [[DOTOFFLOAD_SIZES]], ptr align 4 @.offload_sizes, i32 24, i1 false)
12185 // CHECK19-NEXT: [[TMP6:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
12186 // CHECK19-NEXT: store i32 [[TMP3]], ptr [[TMP6]], align 4
12187 // CHECK19-NEXT: [[TMP7:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
12188 // CHECK19-NEXT: store i32 [[TMP3]], ptr [[TMP7]], align 4
12189 // CHECK19-NEXT: [[TMP8:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0
12190 // CHECK19-NEXT: store ptr null, ptr [[TMP8]], align 4
12191 // CHECK19-NEXT: [[TMP9:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1
12192 // CHECK19-NEXT: store i32 [[TMP0]], ptr [[TMP9]], align 4
12193 // CHECK19-NEXT: [[TMP10:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 1
12194 // CHECK19-NEXT: store i32 [[TMP0]], ptr [[TMP10]], align 4
12195 // CHECK19-NEXT: [[TMP11:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i32 0, i32 1
12196 // CHECK19-NEXT: store ptr null, ptr [[TMP11]], align 4
12197 // CHECK19-NEXT: [[TMP12:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
12198 // CHECK19-NEXT: store ptr [[VLA]], ptr [[TMP12]], align 4
12199 // CHECK19-NEXT: [[TMP13:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 2
12200 // CHECK19-NEXT: store ptr [[VLA]], ptr [[TMP13]], align 4
12201 // CHECK19-NEXT: [[TMP14:%.*]] = getelementptr inbounds [3 x i64], ptr [[DOTOFFLOAD_SIZES]], i32 0, i32 2
12202 // CHECK19-NEXT: store i64 [[TMP5]], ptr [[TMP14]], align 4
12203 // CHECK19-NEXT: [[TMP15:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i32 0, i32 2
12204 // CHECK19-NEXT: store ptr null, ptr [[TMP15]], align 4
12205 // CHECK19-NEXT: [[TMP16:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
12206 // CHECK19-NEXT: [[TMP17:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
12207 // CHECK19-NEXT: [[TMP18:%.*]] = getelementptr inbounds [3 x i64], ptr [[DOTOFFLOAD_SIZES]], i32 0, i32 0
12208 // CHECK19-NEXT: [[TMP19:%.*]] = load i32, ptr [[N]], align 4
12209 // CHECK19-NEXT: store i32 [[TMP19]], ptr [[DOTCAPTURE_EXPR_]], align 4
12210 // CHECK19-NEXT: [[TMP20:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4
12211 // CHECK19-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP20]], 0
12212 // CHECK19-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1
12213 // CHECK19-NEXT: [[SUB2:%.*]] = sub nsw i32 [[DIV]], 1
12214 // CHECK19-NEXT: store i32 [[SUB2]], ptr [[DOTCAPTURE_EXPR_1]], align 4
12215 // CHECK19-NEXT: [[TMP21:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4
12216 // CHECK19-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP21]], 1
12217 // CHECK19-NEXT: [[TMP22:%.*]] = zext i32 [[ADD]] to i64
12218 // CHECK19-NEXT: [[TMP23:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 0
12219 // CHECK19-NEXT: store i32 3, ptr [[TMP23]], align 4
12220 // CHECK19-NEXT: [[TMP24:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 1
12221 // CHECK19-NEXT: store i32 3, ptr [[TMP24]], align 4
12222 // CHECK19-NEXT: [[TMP25:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 2
12223 // CHECK19-NEXT: store ptr [[TMP16]], ptr [[TMP25]], align 4
12224 // CHECK19-NEXT: [[TMP26:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 3
12225 // CHECK19-NEXT: store ptr [[TMP17]], ptr [[TMP26]], align 4
12226 // CHECK19-NEXT: [[TMP27:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 4
12227 // CHECK19-NEXT: store ptr [[TMP18]], ptr [[TMP27]], align 4
12228 // CHECK19-NEXT: [[TMP28:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 5
12229 // CHECK19-NEXT: store ptr @.offload_maptypes, ptr [[TMP28]], align 4
12230 // CHECK19-NEXT: [[TMP29:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 6
12231 // CHECK19-NEXT: store ptr null, ptr [[TMP29]], align 4
12232 // CHECK19-NEXT: [[TMP30:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 7
12233 // CHECK19-NEXT: store ptr null, ptr [[TMP30]], align 4
12234 // CHECK19-NEXT: [[TMP31:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 8
12235 // CHECK19-NEXT: store i64 [[TMP22]], ptr [[TMP31]], align 8
12236 // CHECK19-NEXT: [[TMP32:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 9
12237 // CHECK19-NEXT: store i64 0, ptr [[TMP32]], align 8
12238 // CHECK19-NEXT: [[TMP33:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 10
12239 // CHECK19-NEXT: store [3 x i32] zeroinitializer, ptr [[TMP33]], align 4
12240 // CHECK19-NEXT: [[TMP34:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 11
12241 // CHECK19-NEXT: store [3 x i32] zeroinitializer, ptr [[TMP34]], align 4
12242 // CHECK19-NEXT: [[TMP35:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 12
12243 // CHECK19-NEXT: store i32 0, ptr [[TMP35]], align 4
12244 // CHECK19-NEXT: [[TMP36:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB3:[0-9]+]], i64 -1, i32 0, i32 0, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l148.region_id, ptr [[KERNEL_ARGS]])
12245 // CHECK19-NEXT: [[TMP37:%.*]] = icmp ne i32 [[TMP36]], 0
12246 // CHECK19-NEXT: br i1 [[TMP37]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
12247 // CHECK19: omp_offload.failed:
12248 // CHECK19-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l148(i32 [[TMP3]], i32 [[TMP0]], ptr [[VLA]]) #[[ATTR3:[0-9]+]]
12249 // CHECK19-NEXT: br label [[OMP_OFFLOAD_CONT]]
12250 // CHECK19: omp_offload.cont:
12251 // CHECK19-NEXT: [[TMP38:%.*]] = load i32, ptr [[N]], align 4
12252 // CHECK19-NEXT: store i32 [[TMP38]], ptr [[N_CASTED3]], align 4
12253 // CHECK19-NEXT: [[TMP39:%.*]] = load i32, ptr [[N_CASTED3]], align 4
12254 // CHECK19-NEXT: [[TMP40:%.*]] = mul nuw i32 [[TMP0]], 4
12255 // CHECK19-NEXT: [[TMP41:%.*]] = sext i32 [[TMP40]] to i64
12256 // CHECK19-NEXT: call void @llvm.memcpy.p0.p0.i32(ptr align 4 [[DOTOFFLOAD_SIZES7]], ptr align 4 @.offload_sizes.1, i32 24, i1 false)
12257 // CHECK19-NEXT: [[TMP42:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_BASEPTRS4]], i32 0, i32 0
12258 // CHECK19-NEXT: store i32 [[TMP39]], ptr [[TMP42]], align 4
12259 // CHECK19-NEXT: [[TMP43:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_PTRS5]], i32 0, i32 0
12260 // CHECK19-NEXT: store i32 [[TMP39]], ptr [[TMP43]], align 4
12261 // CHECK19-NEXT: [[TMP44:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_MAPPERS6]], i32 0, i32 0
12262 // CHECK19-NEXT: store ptr null, ptr [[TMP44]], align 4
12263 // CHECK19-NEXT: [[TMP45:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_BASEPTRS4]], i32 0, i32 1
12264 // CHECK19-NEXT: store i32 [[TMP0]], ptr [[TMP45]], align 4
12265 // CHECK19-NEXT: [[TMP46:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_PTRS5]], i32 0, i32 1
12266 // CHECK19-NEXT: store i32 [[TMP0]], ptr [[TMP46]], align 4
12267 // CHECK19-NEXT: [[TMP47:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_MAPPERS6]], i32 0, i32 1
12268 // CHECK19-NEXT: store ptr null, ptr [[TMP47]], align 4
12269 // CHECK19-NEXT: [[TMP48:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_BASEPTRS4]], i32 0, i32 2
12270 // CHECK19-NEXT: store ptr [[VLA]], ptr [[TMP48]], align 4
12271 // CHECK19-NEXT: [[TMP49:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_PTRS5]], i32 0, i32 2
12272 // CHECK19-NEXT: store ptr [[VLA]], ptr [[TMP49]], align 4
12273 // CHECK19-NEXT: [[TMP50:%.*]] = getelementptr inbounds [3 x i64], ptr [[DOTOFFLOAD_SIZES7]], i32 0, i32 2
12274 // CHECK19-NEXT: store i64 [[TMP41]], ptr [[TMP50]], align 4
12275 // CHECK19-NEXT: [[TMP51:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_MAPPERS6]], i32 0, i32 2
12276 // CHECK19-NEXT: store ptr null, ptr [[TMP51]], align 4
12277 // CHECK19-NEXT: [[TMP52:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_BASEPTRS4]], i32 0, i32 0
12278 // CHECK19-NEXT: [[TMP53:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_PTRS5]], i32 0, i32 0
12279 // CHECK19-NEXT: [[TMP54:%.*]] = getelementptr inbounds [3 x i64], ptr [[DOTOFFLOAD_SIZES7]], i32 0, i32 0
12280 // CHECK19-NEXT: [[TMP55:%.*]] = load i32, ptr [[N]], align 4
12281 // CHECK19-NEXT: store i32 [[TMP55]], ptr [[DOTCAPTURE_EXPR_9]], align 4
12282 // CHECK19-NEXT: [[TMP56:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_9]], align 4
12283 // CHECK19-NEXT: [[SUB11:%.*]] = sub nsw i32 [[TMP56]], 0
12284 // CHECK19-NEXT: [[DIV12:%.*]] = sdiv i32 [[SUB11]], 1
12285 // CHECK19-NEXT: [[SUB13:%.*]] = sub nsw i32 [[DIV12]], 1
12286 // CHECK19-NEXT: store i32 [[SUB13]], ptr [[DOTCAPTURE_EXPR_10]], align 4
12287 // CHECK19-NEXT: [[TMP57:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_10]], align 4
12288 // CHECK19-NEXT: [[ADD14:%.*]] = add nsw i32 [[TMP57]], 1
12289 // CHECK19-NEXT: [[TMP58:%.*]] = zext i32 [[ADD14]] to i64
12290 // CHECK19-NEXT: [[TMP59:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS15]], i32 0, i32 0
12291 // CHECK19-NEXT: store i32 3, ptr [[TMP59]], align 4
12292 // CHECK19-NEXT: [[TMP60:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS15]], i32 0, i32 1
12293 // CHECK19-NEXT: store i32 3, ptr [[TMP60]], align 4
12294 // CHECK19-NEXT: [[TMP61:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS15]], i32 0, i32 2
12295 // CHECK19-NEXT: store ptr [[TMP52]], ptr [[TMP61]], align 4
12296 // CHECK19-NEXT: [[TMP62:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS15]], i32 0, i32 3
12297 // CHECK19-NEXT: store ptr [[TMP53]], ptr [[TMP62]], align 4
12298 // CHECK19-NEXT: [[TMP63:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS15]], i32 0, i32 4
12299 // CHECK19-NEXT: store ptr [[TMP54]], ptr [[TMP63]], align 4
12300 // CHECK19-NEXT: [[TMP64:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS15]], i32 0, i32 5
12301 // CHECK19-NEXT: store ptr @.offload_maptypes.2, ptr [[TMP64]], align 4
12302 // CHECK19-NEXT: [[TMP65:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS15]], i32 0, i32 6
12303 // CHECK19-NEXT: store ptr null, ptr [[TMP65]], align 4
12304 // CHECK19-NEXT: [[TMP66:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS15]], i32 0, i32 7
12305 // CHECK19-NEXT: store ptr null, ptr [[TMP66]], align 4
12306 // CHECK19-NEXT: [[TMP67:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS15]], i32 0, i32 8
12307 // CHECK19-NEXT: store i64 [[TMP58]], ptr [[TMP67]], align 8
12308 // CHECK19-NEXT: [[TMP68:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS15]], i32 0, i32 9
12309 // CHECK19-NEXT: store i64 0, ptr [[TMP68]], align 8
12310 // CHECK19-NEXT: [[TMP69:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS15]], i32 0, i32 10
12311 // CHECK19-NEXT: store [3 x i32] zeroinitializer, ptr [[TMP69]], align 4
12312 // CHECK19-NEXT: [[TMP70:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS15]], i32 0, i32 11
12313 // CHECK19-NEXT: store [3 x i32] zeroinitializer, ptr [[TMP70]], align 4
12314 // CHECK19-NEXT: [[TMP71:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS15]], i32 0, i32 12
12315 // CHECK19-NEXT: store i32 0, ptr [[TMP71]], align 4
12316 // CHECK19-NEXT: [[TMP72:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB3]], i64 -1, i32 0, i32 0, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l153.region_id, ptr [[KERNEL_ARGS15]])
12317 // CHECK19-NEXT: [[TMP73:%.*]] = icmp ne i32 [[TMP72]], 0
12318 // CHECK19-NEXT: br i1 [[TMP73]], label [[OMP_OFFLOAD_FAILED16:%.*]], label [[OMP_OFFLOAD_CONT17:%.*]]
12319 // CHECK19: omp_offload.failed16:
12320 // CHECK19-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l153(i32 [[TMP39]], i32 [[TMP0]], ptr [[VLA]]) #[[ATTR3]]
12321 // CHECK19-NEXT: br label [[OMP_OFFLOAD_CONT17]]
12322 // CHECK19: omp_offload.cont17:
12323 // CHECK19-NEXT: [[TMP74:%.*]] = load i32, ptr [[M]], align 4
12324 // CHECK19-NEXT: store i32 [[TMP74]], ptr [[M_CASTED]], align 4
12325 // CHECK19-NEXT: [[TMP75:%.*]] = load i32, ptr [[M_CASTED]], align 4
12326 // CHECK19-NEXT: [[TMP76:%.*]] = load i32, ptr [[N]], align 4
12327 // CHECK19-NEXT: store i32 [[TMP76]], ptr [[N_CASTED18]], align 4
12328 // CHECK19-NEXT: [[TMP77:%.*]] = load i32, ptr [[N_CASTED18]], align 4
12329 // CHECK19-NEXT: [[TMP78:%.*]] = mul nuw i32 [[TMP0]], 4
12330 // CHECK19-NEXT: [[TMP79:%.*]] = sext i32 [[TMP78]] to i64
12331 // CHECK19-NEXT: call void @llvm.memcpy.p0.p0.i32(ptr align 4 [[DOTOFFLOAD_SIZES22]], ptr align 4 @.offload_sizes.3, i32 32, i1 false)
12332 // CHECK19-NEXT: [[TMP80:%.*]] = getelementptr inbounds [4 x ptr], ptr [[DOTOFFLOAD_BASEPTRS19]], i32 0, i32 0
12333 // CHECK19-NEXT: store i32 [[TMP75]], ptr [[TMP80]], align 4
12334 // CHECK19-NEXT: [[TMP81:%.*]] = getelementptr inbounds [4 x ptr], ptr [[DOTOFFLOAD_PTRS20]], i32 0, i32 0
12335 // CHECK19-NEXT: store i32 [[TMP75]], ptr [[TMP81]], align 4
12336 // CHECK19-NEXT: [[TMP82:%.*]] = getelementptr inbounds [4 x ptr], ptr [[DOTOFFLOAD_MAPPERS21]], i32 0, i32 0
12337 // CHECK19-NEXT: store ptr null, ptr [[TMP82]], align 4
12338 // CHECK19-NEXT: [[TMP83:%.*]] = getelementptr inbounds [4 x ptr], ptr [[DOTOFFLOAD_BASEPTRS19]], i32 0, i32 1
12339 // CHECK19-NEXT: store i32 [[TMP77]], ptr [[TMP83]], align 4
12340 // CHECK19-NEXT: [[TMP84:%.*]] = getelementptr inbounds [4 x ptr], ptr [[DOTOFFLOAD_PTRS20]], i32 0, i32 1
12341 // CHECK19-NEXT: store i32 [[TMP77]], ptr [[TMP84]], align 4
12342 // CHECK19-NEXT: [[TMP85:%.*]] = getelementptr inbounds [4 x ptr], ptr [[DOTOFFLOAD_MAPPERS21]], i32 0, i32 1
12343 // CHECK19-NEXT: store ptr null, ptr [[TMP85]], align 4
12344 // CHECK19-NEXT: [[TMP86:%.*]] = getelementptr inbounds [4 x ptr], ptr [[DOTOFFLOAD_BASEPTRS19]], i32 0, i32 2
12345 // CHECK19-NEXT: store i32 [[TMP0]], ptr [[TMP86]], align 4
12346 // CHECK19-NEXT: [[TMP87:%.*]] = getelementptr inbounds [4 x ptr], ptr [[DOTOFFLOAD_PTRS20]], i32 0, i32 2
12347 // CHECK19-NEXT: store i32 [[TMP0]], ptr [[TMP87]], align 4
12348 // CHECK19-NEXT: [[TMP88:%.*]] = getelementptr inbounds [4 x ptr], ptr [[DOTOFFLOAD_MAPPERS21]], i32 0, i32 2
12349 // CHECK19-NEXT: store ptr null, ptr [[TMP88]], align 4
12350 // CHECK19-NEXT: [[TMP89:%.*]] = getelementptr inbounds [4 x ptr], ptr [[DOTOFFLOAD_BASEPTRS19]], i32 0, i32 3
12351 // CHECK19-NEXT: store ptr [[VLA]], ptr [[TMP89]], align 4
12352 // CHECK19-NEXT: [[TMP90:%.*]] = getelementptr inbounds [4 x ptr], ptr [[DOTOFFLOAD_PTRS20]], i32 0, i32 3
12353 // CHECK19-NEXT: store ptr [[VLA]], ptr [[TMP90]], align 4
12354 // CHECK19-NEXT: [[TMP91:%.*]] = getelementptr inbounds [4 x i64], ptr [[DOTOFFLOAD_SIZES22]], i32 0, i32 3
12355 // CHECK19-NEXT: store i64 [[TMP79]], ptr [[TMP91]], align 4
12356 // CHECK19-NEXT: [[TMP92:%.*]] = getelementptr inbounds [4 x ptr], ptr [[DOTOFFLOAD_MAPPERS21]], i32 0, i32 3
12357 // CHECK19-NEXT: store ptr null, ptr [[TMP92]], align 4
12358 // CHECK19-NEXT: [[TMP93:%.*]] = getelementptr inbounds [4 x ptr], ptr [[DOTOFFLOAD_BASEPTRS19]], i32 0, i32 0
12359 // CHECK19-NEXT: [[TMP94:%.*]] = getelementptr inbounds [4 x ptr], ptr [[DOTOFFLOAD_PTRS20]], i32 0, i32 0
12360 // CHECK19-NEXT: [[TMP95:%.*]] = getelementptr inbounds [4 x i64], ptr [[DOTOFFLOAD_SIZES22]], i32 0, i32 0
12361 // CHECK19-NEXT: [[TMP96:%.*]] = load i32, ptr [[N]], align 4
12362 // CHECK19-NEXT: store i32 [[TMP96]], ptr [[DOTCAPTURE_EXPR_24]], align 4
12363 // CHECK19-NEXT: [[TMP97:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_24]], align 4
12364 // CHECK19-NEXT: [[SUB26:%.*]] = sub nsw i32 [[TMP97]], 0
12365 // CHECK19-NEXT: [[DIV27:%.*]] = sdiv i32 [[SUB26]], 1
12366 // CHECK19-NEXT: [[SUB28:%.*]] = sub nsw i32 [[DIV27]], 1
12367 // CHECK19-NEXT: store i32 [[SUB28]], ptr [[DOTCAPTURE_EXPR_25]], align 4
12368 // CHECK19-NEXT: [[TMP98:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_25]], align 4
12369 // CHECK19-NEXT: [[ADD29:%.*]] = add nsw i32 [[TMP98]], 1
12370 // CHECK19-NEXT: [[TMP99:%.*]] = zext i32 [[ADD29]] to i64
12371 // CHECK19-NEXT: [[TMP100:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS30]], i32 0, i32 0
12372 // CHECK19-NEXT: store i32 3, ptr [[TMP100]], align 4
12373 // CHECK19-NEXT: [[TMP101:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS30]], i32 0, i32 1
12374 // CHECK19-NEXT: store i32 4, ptr [[TMP101]], align 4
12375 // CHECK19-NEXT: [[TMP102:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS30]], i32 0, i32 2
12376 // CHECK19-NEXT: store ptr [[TMP93]], ptr [[TMP102]], align 4
12377 // CHECK19-NEXT: [[TMP103:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS30]], i32 0, i32 3
12378 // CHECK19-NEXT: store ptr [[TMP94]], ptr [[TMP103]], align 4
12379 // CHECK19-NEXT: [[TMP104:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS30]], i32 0, i32 4
12380 // CHECK19-NEXT: store ptr [[TMP95]], ptr [[TMP104]], align 4
12381 // CHECK19-NEXT: [[TMP105:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS30]], i32 0, i32 5
12382 // CHECK19-NEXT: store ptr @.offload_maptypes.4, ptr [[TMP105]], align 4
12383 // CHECK19-NEXT: [[TMP106:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS30]], i32 0, i32 6
12384 // CHECK19-NEXT: store ptr null, ptr [[TMP106]], align 4
12385 // CHECK19-NEXT: [[TMP107:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS30]], i32 0, i32 7
12386 // CHECK19-NEXT: store ptr null, ptr [[TMP107]], align 4
12387 // CHECK19-NEXT: [[TMP108:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS30]], i32 0, i32 8
12388 // CHECK19-NEXT: store i64 [[TMP99]], ptr [[TMP108]], align 8
12389 // CHECK19-NEXT: [[TMP109:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS30]], i32 0, i32 9
12390 // CHECK19-NEXT: store i64 0, ptr [[TMP109]], align 8
12391 // CHECK19-NEXT: [[TMP110:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS30]], i32 0, i32 10
12392 // CHECK19-NEXT: store [3 x i32] zeroinitializer, ptr [[TMP110]], align 4
12393 // CHECK19-NEXT: [[TMP111:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS30]], i32 0, i32 11
12394 // CHECK19-NEXT: store [3 x i32] zeroinitializer, ptr [[TMP111]], align 4
12395 // CHECK19-NEXT: [[TMP112:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS30]], i32 0, i32 12
12396 // CHECK19-NEXT: store i32 0, ptr [[TMP112]], align 4
12397 // CHECK19-NEXT: [[TMP113:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB3]], i64 -1, i32 0, i32 0, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l158.region_id, ptr [[KERNEL_ARGS30]])
12398 // CHECK19-NEXT: [[TMP114:%.*]] = icmp ne i32 [[TMP113]], 0
12399 // CHECK19-NEXT: br i1 [[TMP114]], label [[OMP_OFFLOAD_FAILED31:%.*]], label [[OMP_OFFLOAD_CONT32:%.*]]
12400 // CHECK19: omp_offload.failed31:
12401 // CHECK19-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l158(i32 [[TMP75]], i32 [[TMP77]], i32 [[TMP0]], ptr [[VLA]]) #[[ATTR3]]
12402 // CHECK19-NEXT: br label [[OMP_OFFLOAD_CONT32]]
12403 // CHECK19: omp_offload.cont32:
12404 // CHECK19-NEXT: [[TMP115:%.*]] = load i32, ptr [[N]], align 4
12405 // CHECK19-NEXT: store i32 [[TMP115]], ptr [[N_CASTED33]], align 4
12406 // CHECK19-NEXT: [[TMP116:%.*]] = load i32, ptr [[N_CASTED33]], align 4
12407 // CHECK19-NEXT: [[TMP117:%.*]] = mul nuw i32 [[TMP0]], 4
12408 // CHECK19-NEXT: [[TMP118:%.*]] = sext i32 [[TMP117]] to i64
12409 // CHECK19-NEXT: call void @llvm.memcpy.p0.p0.i32(ptr align 4 [[DOTOFFLOAD_SIZES37]], ptr align 4 @.offload_sizes.5, i32 24, i1 false)
12410 // CHECK19-NEXT: [[TMP119:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_BASEPTRS34]], i32 0, i32 0
12411 // CHECK19-NEXT: store i32 [[TMP116]], ptr [[TMP119]], align 4
12412 // CHECK19-NEXT: [[TMP120:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_PTRS35]], i32 0, i32 0
12413 // CHECK19-NEXT: store i32 [[TMP116]], ptr [[TMP120]], align 4
12414 // CHECK19-NEXT: [[TMP121:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_MAPPERS36]], i32 0, i32 0
12415 // CHECK19-NEXT: store ptr null, ptr [[TMP121]], align 4
12416 // CHECK19-NEXT: [[TMP122:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_BASEPTRS34]], i32 0, i32 1
12417 // CHECK19-NEXT: store i32 [[TMP0]], ptr [[TMP122]], align 4
12418 // CHECK19-NEXT: [[TMP123:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_PTRS35]], i32 0, i32 1
12419 // CHECK19-NEXT: store i32 [[TMP0]], ptr [[TMP123]], align 4
12420 // CHECK19-NEXT: [[TMP124:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_MAPPERS36]], i32 0, i32 1
12421 // CHECK19-NEXT: store ptr null, ptr [[TMP124]], align 4
12422 // CHECK19-NEXT: [[TMP125:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_BASEPTRS34]], i32 0, i32 2
12423 // CHECK19-NEXT: store ptr [[VLA]], ptr [[TMP125]], align 4
12424 // CHECK19-NEXT: [[TMP126:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_PTRS35]], i32 0, i32 2
12425 // CHECK19-NEXT: store ptr [[VLA]], ptr [[TMP126]], align 4
12426 // CHECK19-NEXT: [[TMP127:%.*]] = getelementptr inbounds [3 x i64], ptr [[DOTOFFLOAD_SIZES37]], i32 0, i32 2
12427 // CHECK19-NEXT: store i64 [[TMP118]], ptr [[TMP127]], align 4
12428 // CHECK19-NEXT: [[TMP128:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_MAPPERS36]], i32 0, i32 2
12429 // CHECK19-NEXT: store ptr null, ptr [[TMP128]], align 4
12430 // CHECK19-NEXT: [[TMP129:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_BASEPTRS34]], i32 0, i32 0
12431 // CHECK19-NEXT: [[TMP130:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_PTRS35]], i32 0, i32 0
12432 // CHECK19-NEXT: [[TMP131:%.*]] = getelementptr inbounds [3 x i64], ptr [[DOTOFFLOAD_SIZES37]], i32 0, i32 0
12433 // CHECK19-NEXT: [[TMP132:%.*]] = load i32, ptr [[N]], align 4
12434 // CHECK19-NEXT: store i32 [[TMP132]], ptr [[DOTCAPTURE_EXPR_39]], align 4
12435 // CHECK19-NEXT: [[TMP133:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_39]], align 4
12436 // CHECK19-NEXT: [[SUB41:%.*]] = sub nsw i32 [[TMP133]], 0
12437 // CHECK19-NEXT: [[DIV42:%.*]] = sdiv i32 [[SUB41]], 1
12438 // CHECK19-NEXT: [[SUB43:%.*]] = sub nsw i32 [[DIV42]], 1
12439 // CHECK19-NEXT: store i32 [[SUB43]], ptr [[DOTCAPTURE_EXPR_40]], align 4
12440 // CHECK19-NEXT: [[TMP134:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_40]], align 4
12441 // CHECK19-NEXT: [[ADD44:%.*]] = add nsw i32 [[TMP134]], 1
12442 // CHECK19-NEXT: [[TMP135:%.*]] = zext i32 [[ADD44]] to i64
12443 // CHECK19-NEXT: [[TMP136:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS45]], i32 0, i32 0
12444 // CHECK19-NEXT: store i32 3, ptr [[TMP136]], align 4
12445 // CHECK19-NEXT: [[TMP137:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS45]], i32 0, i32 1
12446 // CHECK19-NEXT: store i32 3, ptr [[TMP137]], align 4
12447 // CHECK19-NEXT: [[TMP138:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS45]], i32 0, i32 2
12448 // CHECK19-NEXT: store ptr [[TMP129]], ptr [[TMP138]], align 4
12449 // CHECK19-NEXT: [[TMP139:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS45]], i32 0, i32 3
12450 // CHECK19-NEXT: store ptr [[TMP130]], ptr [[TMP139]], align 4
12451 // CHECK19-NEXT: [[TMP140:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS45]], i32 0, i32 4
12452 // CHECK19-NEXT: store ptr [[TMP131]], ptr [[TMP140]], align 4
12453 // CHECK19-NEXT: [[TMP141:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS45]], i32 0, i32 5
12454 // CHECK19-NEXT: store ptr @.offload_maptypes.6, ptr [[TMP141]], align 4
12455 // CHECK19-NEXT: [[TMP142:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS45]], i32 0, i32 6
12456 // CHECK19-NEXT: store ptr null, ptr [[TMP142]], align 4
12457 // CHECK19-NEXT: [[TMP143:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS45]], i32 0, i32 7
12458 // CHECK19-NEXT: store ptr null, ptr [[TMP143]], align 4
12459 // CHECK19-NEXT: [[TMP144:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS45]], i32 0, i32 8
12460 // CHECK19-NEXT: store i64 [[TMP135]], ptr [[TMP144]], align 8
12461 // CHECK19-NEXT: [[TMP145:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS45]], i32 0, i32 9
12462 // CHECK19-NEXT: store i64 0, ptr [[TMP145]], align 8
12463 // CHECK19-NEXT: [[TMP146:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS45]], i32 0, i32 10
12464 // CHECK19-NEXT: store [3 x i32] zeroinitializer, ptr [[TMP146]], align 4
12465 // CHECK19-NEXT: [[TMP147:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS45]], i32 0, i32 11
12466 // CHECK19-NEXT: store [3 x i32] zeroinitializer, ptr [[TMP147]], align 4
12467 // CHECK19-NEXT: [[TMP148:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS45]], i32 0, i32 12
12468 // CHECK19-NEXT: store i32 0, ptr [[TMP148]], align 4
12469 // CHECK19-NEXT: [[TMP149:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB3]], i64 -1, i32 0, i32 0, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l163.region_id, ptr [[KERNEL_ARGS45]])
12470 // CHECK19-NEXT: [[TMP150:%.*]] = icmp ne i32 [[TMP149]], 0
12471 // CHECK19-NEXT: br i1 [[TMP150]], label [[OMP_OFFLOAD_FAILED46:%.*]], label [[OMP_OFFLOAD_CONT47:%.*]]
12472 // CHECK19: omp_offload.failed46:
12473 // CHECK19-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l163(i32 [[TMP116]], i32 [[TMP0]], ptr [[VLA]]) #[[ATTR3]]
12474 // CHECK19-NEXT: br label [[OMP_OFFLOAD_CONT47]]
12475 // CHECK19: omp_offload.cont47:
12476 // CHECK19-NEXT: [[TMP151:%.*]] = load i32, ptr [[M]], align 4
12477 // CHECK19-NEXT: store i32 [[TMP151]], ptr [[M_CASTED48]], align 4
12478 // CHECK19-NEXT: [[TMP152:%.*]] = load i32, ptr [[M_CASTED48]], align 4
12479 // CHECK19-NEXT: [[TMP153:%.*]] = load i32, ptr [[N]], align 4
12480 // CHECK19-NEXT: store i32 [[TMP153]], ptr [[N_CASTED49]], align 4
12481 // CHECK19-NEXT: [[TMP154:%.*]] = load i32, ptr [[N_CASTED49]], align 4
12482 // CHECK19-NEXT: [[TMP155:%.*]] = mul nuw i32 [[TMP0]], 4
12483 // CHECK19-NEXT: [[TMP156:%.*]] = sext i32 [[TMP155]] to i64
12484 // CHECK19-NEXT: call void @llvm.memcpy.p0.p0.i32(ptr align 4 [[DOTOFFLOAD_SIZES53]], ptr align 4 @.offload_sizes.7, i32 32, i1 false)
12485 // CHECK19-NEXT: [[TMP157:%.*]] = getelementptr inbounds [4 x ptr], ptr [[DOTOFFLOAD_BASEPTRS50]], i32 0, i32 0
12486 // CHECK19-NEXT: store i32 [[TMP152]], ptr [[TMP157]], align 4
12487 // CHECK19-NEXT: [[TMP158:%.*]] = getelementptr inbounds [4 x ptr], ptr [[DOTOFFLOAD_PTRS51]], i32 0, i32 0
12488 // CHECK19-NEXT: store i32 [[TMP152]], ptr [[TMP158]], align 4
12489 // CHECK19-NEXT: [[TMP159:%.*]] = getelementptr inbounds [4 x ptr], ptr [[DOTOFFLOAD_MAPPERS52]], i32 0, i32 0
12490 // CHECK19-NEXT: store ptr null, ptr [[TMP159]], align 4
12491 // CHECK19-NEXT: [[TMP160:%.*]] = getelementptr inbounds [4 x ptr], ptr [[DOTOFFLOAD_BASEPTRS50]], i32 0, i32 1
12492 // CHECK19-NEXT: store i32 [[TMP154]], ptr [[TMP160]], align 4
12493 // CHECK19-NEXT: [[TMP161:%.*]] = getelementptr inbounds [4 x ptr], ptr [[DOTOFFLOAD_PTRS51]], i32 0, i32 1
12494 // CHECK19-NEXT: store i32 [[TMP154]], ptr [[TMP161]], align 4
12495 // CHECK19-NEXT: [[TMP162:%.*]] = getelementptr inbounds [4 x ptr], ptr [[DOTOFFLOAD_MAPPERS52]], i32 0, i32 1
12496 // CHECK19-NEXT: store ptr null, ptr [[TMP162]], align 4
12497 // CHECK19-NEXT: [[TMP163:%.*]] = getelementptr inbounds [4 x ptr], ptr [[DOTOFFLOAD_BASEPTRS50]], i32 0, i32 2
12498 // CHECK19-NEXT: store i32 [[TMP0]], ptr [[TMP163]], align 4
12499 // CHECK19-NEXT: [[TMP164:%.*]] = getelementptr inbounds [4 x ptr], ptr [[DOTOFFLOAD_PTRS51]], i32 0, i32 2
12500 // CHECK19-NEXT: store i32 [[TMP0]], ptr [[TMP164]], align 4
12501 // CHECK19-NEXT: [[TMP165:%.*]] = getelementptr inbounds [4 x ptr], ptr [[DOTOFFLOAD_MAPPERS52]], i32 0, i32 2
12502 // CHECK19-NEXT: store ptr null, ptr [[TMP165]], align 4
12503 // CHECK19-NEXT: [[TMP166:%.*]] = getelementptr inbounds [4 x ptr], ptr [[DOTOFFLOAD_BASEPTRS50]], i32 0, i32 3
12504 // CHECK19-NEXT: store ptr [[VLA]], ptr [[TMP166]], align 4
12505 // CHECK19-NEXT: [[TMP167:%.*]] = getelementptr inbounds [4 x ptr], ptr [[DOTOFFLOAD_PTRS51]], i32 0, i32 3
12506 // CHECK19-NEXT: store ptr [[VLA]], ptr [[TMP167]], align 4
12507 // CHECK19-NEXT: [[TMP168:%.*]] = getelementptr inbounds [4 x i64], ptr [[DOTOFFLOAD_SIZES53]], i32 0, i32 3
12508 // CHECK19-NEXT: store i64 [[TMP156]], ptr [[TMP168]], align 4
12509 // CHECK19-NEXT: [[TMP169:%.*]] = getelementptr inbounds [4 x ptr], ptr [[DOTOFFLOAD_MAPPERS52]], i32 0, i32 3
12510 // CHECK19-NEXT: store ptr null, ptr [[TMP169]], align 4
12511 // CHECK19-NEXT: [[TMP170:%.*]] = getelementptr inbounds [4 x ptr], ptr [[DOTOFFLOAD_BASEPTRS50]], i32 0, i32 0
12512 // CHECK19-NEXT: [[TMP171:%.*]] = getelementptr inbounds [4 x ptr], ptr [[DOTOFFLOAD_PTRS51]], i32 0, i32 0
12513 // CHECK19-NEXT: [[TMP172:%.*]] = getelementptr inbounds [4 x i64], ptr [[DOTOFFLOAD_SIZES53]], i32 0, i32 0
12514 // CHECK19-NEXT: [[TMP173:%.*]] = load i32, ptr [[N]], align 4
12515 // CHECK19-NEXT: store i32 [[TMP173]], ptr [[DOTCAPTURE_EXPR_55]], align 4
12516 // CHECK19-NEXT: [[TMP174:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_55]], align 4
12517 // CHECK19-NEXT: [[SUB57:%.*]] = sub nsw i32 [[TMP174]], 0
12518 // CHECK19-NEXT: [[DIV58:%.*]] = sdiv i32 [[SUB57]], 1
12519 // CHECK19-NEXT: [[SUB59:%.*]] = sub nsw i32 [[DIV58]], 1
12520 // CHECK19-NEXT: store i32 [[SUB59]], ptr [[DOTCAPTURE_EXPR_56]], align 4
12521 // CHECK19-NEXT: [[TMP175:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_56]], align 4
12522 // CHECK19-NEXT: [[ADD60:%.*]] = add nsw i32 [[TMP175]], 1
12523 // CHECK19-NEXT: [[TMP176:%.*]] = zext i32 [[ADD60]] to i64
12524 // CHECK19-NEXT: [[TMP177:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS61]], i32 0, i32 0
12525 // CHECK19-NEXT: store i32 3, ptr [[TMP177]], align 4
12526 // CHECK19-NEXT: [[TMP178:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS61]], i32 0, i32 1
12527 // CHECK19-NEXT: store i32 4, ptr [[TMP178]], align 4
12528 // CHECK19-NEXT: [[TMP179:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS61]], i32 0, i32 2
12529 // CHECK19-NEXT: store ptr [[TMP170]], ptr [[TMP179]], align 4
12530 // CHECK19-NEXT: [[TMP180:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS61]], i32 0, i32 3
12531 // CHECK19-NEXT: store ptr [[TMP171]], ptr [[TMP180]], align 4
12532 // CHECK19-NEXT: [[TMP181:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS61]], i32 0, i32 4
12533 // CHECK19-NEXT: store ptr [[TMP172]], ptr [[TMP181]], align 4
12534 // CHECK19-NEXT: [[TMP182:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS61]], i32 0, i32 5
12535 // CHECK19-NEXT: store ptr @.offload_maptypes.8, ptr [[TMP182]], align 4
12536 // CHECK19-NEXT: [[TMP183:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS61]], i32 0, i32 6
12537 // CHECK19-NEXT: store ptr null, ptr [[TMP183]], align 4
12538 // CHECK19-NEXT: [[TMP184:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS61]], i32 0, i32 7
12539 // CHECK19-NEXT: store ptr null, ptr [[TMP184]], align 4
12540 // CHECK19-NEXT: [[TMP185:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS61]], i32 0, i32 8
12541 // CHECK19-NEXT: store i64 [[TMP176]], ptr [[TMP185]], align 8
12542 // CHECK19-NEXT: [[TMP186:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS61]], i32 0, i32 9
12543 // CHECK19-NEXT: store i64 0, ptr [[TMP186]], align 8
12544 // CHECK19-NEXT: [[TMP187:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS61]], i32 0, i32 10
12545 // CHECK19-NEXT: store [3 x i32] zeroinitializer, ptr [[TMP187]], align 4
12546 // CHECK19-NEXT: [[TMP188:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS61]], i32 0, i32 11
12547 // CHECK19-NEXT: store [3 x i32] zeroinitializer, ptr [[TMP188]], align 4
12548 // CHECK19-NEXT: [[TMP189:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS61]], i32 0, i32 12
12549 // CHECK19-NEXT: store i32 0, ptr [[TMP189]], align 4
12550 // CHECK19-NEXT: [[TMP190:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB3]], i64 -1, i32 0, i32 0, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l168.region_id, ptr [[KERNEL_ARGS61]])
12551 // CHECK19-NEXT: [[TMP191:%.*]] = icmp ne i32 [[TMP190]], 0
12552 // CHECK19-NEXT: br i1 [[TMP191]], label [[OMP_OFFLOAD_FAILED62:%.*]], label [[OMP_OFFLOAD_CONT63:%.*]]
12553 // CHECK19: omp_offload.failed62:
12554 // CHECK19-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l168(i32 [[TMP152]], i32 [[TMP154]], i32 [[TMP0]], ptr [[VLA]]) #[[ATTR3]]
12555 // CHECK19-NEXT: br label [[OMP_OFFLOAD_CONT63]]
12556 // CHECK19: omp_offload.cont63:
12557 // CHECK19-NEXT: [[TMP192:%.*]] = load i32, ptr [[ARGC_ADDR]], align 4
12558 // CHECK19-NEXT: [[CALL:%.*]] = call noundef i32 @_Z5tmainIiLi10EEiT_(i32 noundef [[TMP192]])
12559 // CHECK19-NEXT: store i32 [[CALL]], ptr [[RETVAL]], align 4
12560 // CHECK19-NEXT: [[TMP193:%.*]] = load ptr, ptr [[SAVED_STACK]], align 4
12561 // CHECK19-NEXT: call void @llvm.stackrestore.p0(ptr [[TMP193]])
12562 // CHECK19-NEXT: [[TMP194:%.*]] = load i32, ptr [[RETVAL]], align 4
12563 // CHECK19-NEXT: ret i32 [[TMP194]]
12566 // CHECK19-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l148
12567 // CHECK19-SAME: (i32 noundef [[N:%.*]], i32 noundef [[VLA:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[A:%.*]]) #[[ATTR2:[0-9]+]] {
12568 // CHECK19-NEXT: entry:
12569 // CHECK19-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4
12570 // CHECK19-NEXT: [[VLA_ADDR:%.*]] = alloca i32, align 4
12571 // CHECK19-NEXT: [[A_ADDR:%.*]] = alloca ptr, align 4
12572 // CHECK19-NEXT: store i32 [[N]], ptr [[N_ADDR]], align 4
12573 // CHECK19-NEXT: store i32 [[VLA]], ptr [[VLA_ADDR]], align 4
12574 // CHECK19-NEXT: store ptr [[A]], ptr [[A_ADDR]], align 4
12575 // CHECK19-NEXT: [[TMP0:%.*]] = load i32, ptr [[VLA_ADDR]], align 4
12576 // CHECK19-NEXT: [[TMP1:%.*]] = load ptr, ptr [[A_ADDR]], align 4
12577 // CHECK19-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_teams(ptr @[[GLOB3]], i32 3, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l148.omp_outlined, ptr [[N_ADDR]], i32 [[TMP0]], ptr [[TMP1]])
12578 // CHECK19-NEXT: ret void
12581 // CHECK19-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l148.omp_outlined
12582 // CHECK19-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[N:%.*]], i32 noundef [[VLA:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[A:%.*]]) #[[ATTR2]] {
12583 // CHECK19-NEXT: entry:
12584 // CHECK19-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 4
12585 // CHECK19-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 4
12586 // CHECK19-NEXT: [[N_ADDR:%.*]] = alloca ptr, align 4
12587 // CHECK19-NEXT: [[VLA_ADDR:%.*]] = alloca i32, align 4
12588 // CHECK19-NEXT: [[A_ADDR:%.*]] = alloca ptr, align 4
12589 // CHECK19-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
12590 // CHECK19-NEXT: [[TMP:%.*]] = alloca i32, align 4
12591 // CHECK19-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4
12592 // CHECK19-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4
12593 // CHECK19-NEXT: [[I:%.*]] = alloca i32, align 4
12594 // CHECK19-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4
12595 // CHECK19-NEXT: [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4
12596 // CHECK19-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
12597 // CHECK19-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
12598 // CHECK19-NEXT: [[I3:%.*]] = alloca i32, align 4
12599 // CHECK19-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 4
12600 // CHECK19-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 4
12601 // CHECK19-NEXT: store ptr [[N]], ptr [[N_ADDR]], align 4
12602 // CHECK19-NEXT: store i32 [[VLA]], ptr [[VLA_ADDR]], align 4
12603 // CHECK19-NEXT: store ptr [[A]], ptr [[A_ADDR]], align 4
12604 // CHECK19-NEXT: [[TMP0:%.*]] = load ptr, ptr [[N_ADDR]], align 4
12605 // CHECK19-NEXT: [[TMP1:%.*]] = load i32, ptr [[VLA_ADDR]], align 4
12606 // CHECK19-NEXT: [[TMP2:%.*]] = load ptr, ptr [[A_ADDR]], align 4
12607 // CHECK19-NEXT: [[TMP3:%.*]] = load i32, ptr [[TMP0]], align 4
12608 // CHECK19-NEXT: store i32 [[TMP3]], ptr [[DOTCAPTURE_EXPR_]], align 4
12609 // CHECK19-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4
12610 // CHECK19-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP4]], 0
12611 // CHECK19-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1
12612 // CHECK19-NEXT: [[SUB2:%.*]] = sub nsw i32 [[DIV]], 1
12613 // CHECK19-NEXT: store i32 [[SUB2]], ptr [[DOTCAPTURE_EXPR_1]], align 4
12614 // CHECK19-NEXT: store i32 0, ptr [[I]], align 4
12615 // CHECK19-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4
12616 // CHECK19-NEXT: [[CMP:%.*]] = icmp slt i32 0, [[TMP5]]
12617 // CHECK19-NEXT: br i1 [[CMP]], label [[OMP_PRECOND_THEN:%.*]], label [[OMP_PRECOND_END:%.*]]
12618 // CHECK19: omp.precond.then:
12619 // CHECK19-NEXT: store i32 0, ptr [[DOTOMP_COMB_LB]], align 4
12620 // CHECK19-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4
12621 // CHECK19-NEXT: store i32 [[TMP6]], ptr [[DOTOMP_COMB_UB]], align 4
12622 // CHECK19-NEXT: store i32 1, ptr [[DOTOMP_STRIDE]], align 4
12623 // CHECK19-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
12624 // CHECK19-NEXT: [[TMP7:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
12625 // CHECK19-NEXT: [[TMP8:%.*]] = load i32, ptr [[TMP7]], align 4
12626 // CHECK19-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB1:[0-9]+]], i32 [[TMP8]], i32 92, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
12627 // CHECK19-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
12628 // CHECK19-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4
12629 // CHECK19-NEXT: [[CMP4:%.*]] = icmp sgt i32 [[TMP9]], [[TMP10]]
12630 // CHECK19-NEXT: br i1 [[CMP4]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
12631 // CHECK19: cond.true:
12632 // CHECK19-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4
12633 // CHECK19-NEXT: br label [[COND_END:%.*]]
12634 // CHECK19: cond.false:
12635 // CHECK19-NEXT: [[TMP12:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
12636 // CHECK19-NEXT: br label [[COND_END]]
12637 // CHECK19: cond.end:
12638 // CHECK19-NEXT: [[COND:%.*]] = phi i32 [ [[TMP11]], [[COND_TRUE]] ], [ [[TMP12]], [[COND_FALSE]] ]
12639 // CHECK19-NEXT: store i32 [[COND]], ptr [[DOTOMP_COMB_UB]], align 4
12640 // CHECK19-NEXT: [[TMP13:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
12641 // CHECK19-NEXT: store i32 [[TMP13]], ptr [[DOTOMP_IV]], align 4
12642 // CHECK19-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
12643 // CHECK19: omp.inner.for.cond:
12644 // CHECK19-NEXT: [[TMP14:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
12645 // CHECK19-NEXT: [[TMP15:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
12646 // CHECK19-NEXT: [[CMP5:%.*]] = icmp sle i32 [[TMP14]], [[TMP15]]
12647 // CHECK19-NEXT: br i1 [[CMP5]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
12648 // CHECK19: omp.inner.for.body:
12649 // CHECK19-NEXT: [[TMP16:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
12650 // CHECK19-NEXT: [[TMP17:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
12651 // CHECK19-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_call(ptr @[[GLOB3]], i32 5, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l148.omp_outlined.omp_outlined, i32 [[TMP16]], i32 [[TMP17]], ptr [[TMP0]], i32 [[TMP1]], ptr [[TMP2]])
12652 // CHECK19-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
12653 // CHECK19: omp.inner.for.inc:
12654 // CHECK19-NEXT: [[TMP18:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
12655 // CHECK19-NEXT: [[TMP19:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4
12656 // CHECK19-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP18]], [[TMP19]]
12657 // CHECK19-NEXT: store i32 [[ADD]], ptr [[DOTOMP_IV]], align 4
12658 // CHECK19-NEXT: br label [[OMP_INNER_FOR_COND]]
12659 // CHECK19: omp.inner.for.end:
12660 // CHECK19-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
12661 // CHECK19: omp.loop.exit:
12662 // CHECK19-NEXT: [[TMP20:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
12663 // CHECK19-NEXT: [[TMP21:%.*]] = load i32, ptr [[TMP20]], align 4
12664 // CHECK19-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB1]], i32 [[TMP21]])
12665 // CHECK19-NEXT: br label [[OMP_PRECOND_END]]
12666 // CHECK19: omp.precond.end:
12667 // CHECK19-NEXT: ret void
12670 // CHECK19-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l148.omp_outlined.omp_outlined
12671 // CHECK19-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], i32 noundef [[DOTPREVIOUS_LB_:%.*]], i32 noundef [[DOTPREVIOUS_UB_:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[N:%.*]], i32 noundef [[VLA:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[A:%.*]]) #[[ATTR2]] {
12672 // CHECK19-NEXT: entry:
12673 // CHECK19-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 4
12674 // CHECK19-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 4
12675 // CHECK19-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i32, align 4
12676 // CHECK19-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i32, align 4
12677 // CHECK19-NEXT: [[N_ADDR:%.*]] = alloca ptr, align 4
12678 // CHECK19-NEXT: [[VLA_ADDR:%.*]] = alloca i32, align 4
12679 // CHECK19-NEXT: [[A_ADDR:%.*]] = alloca ptr, align 4
12680 // CHECK19-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
12681 // CHECK19-NEXT: [[TMP:%.*]] = alloca i32, align 4
12682 // CHECK19-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4
12683 // CHECK19-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4
12684 // CHECK19-NEXT: [[I:%.*]] = alloca i32, align 4
12685 // CHECK19-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4
12686 // CHECK19-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4
12687 // CHECK19-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
12688 // CHECK19-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
12689 // CHECK19-NEXT: [[I3:%.*]] = alloca i32, align 4
12690 // CHECK19-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 4
12691 // CHECK19-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 4
12692 // CHECK19-NEXT: store i32 [[DOTPREVIOUS_LB_]], ptr [[DOTPREVIOUS_LB__ADDR]], align 4
12693 // CHECK19-NEXT: store i32 [[DOTPREVIOUS_UB_]], ptr [[DOTPREVIOUS_UB__ADDR]], align 4
12694 // CHECK19-NEXT: store ptr [[N]], ptr [[N_ADDR]], align 4
12695 // CHECK19-NEXT: store i32 [[VLA]], ptr [[VLA_ADDR]], align 4
12696 // CHECK19-NEXT: store ptr [[A]], ptr [[A_ADDR]], align 4
12697 // CHECK19-NEXT: [[TMP0:%.*]] = load ptr, ptr [[N_ADDR]], align 4
12698 // CHECK19-NEXT: [[TMP1:%.*]] = load i32, ptr [[VLA_ADDR]], align 4
12699 // CHECK19-NEXT: [[TMP2:%.*]] = load ptr, ptr [[A_ADDR]], align 4
12700 // CHECK19-NEXT: [[TMP3:%.*]] = load i32, ptr [[TMP0]], align 4
12701 // CHECK19-NEXT: store i32 [[TMP3]], ptr [[DOTCAPTURE_EXPR_]], align 4
12702 // CHECK19-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4
12703 // CHECK19-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP4]], 0
12704 // CHECK19-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1
12705 // CHECK19-NEXT: [[SUB2:%.*]] = sub nsw i32 [[DIV]], 1
12706 // CHECK19-NEXT: store i32 [[SUB2]], ptr [[DOTCAPTURE_EXPR_1]], align 4
12707 // CHECK19-NEXT: store i32 0, ptr [[I]], align 4
12708 // CHECK19-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4
12709 // CHECK19-NEXT: [[CMP:%.*]] = icmp slt i32 0, [[TMP5]]
12710 // CHECK19-NEXT: br i1 [[CMP]], label [[OMP_PRECOND_THEN:%.*]], label [[OMP_PRECOND_END:%.*]]
12711 // CHECK19: omp.precond.then:
12712 // CHECK19-NEXT: store i32 0, ptr [[DOTOMP_LB]], align 4
12713 // CHECK19-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4
12714 // CHECK19-NEXT: store i32 [[TMP6]], ptr [[DOTOMP_UB]], align 4
12715 // CHECK19-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTPREVIOUS_LB__ADDR]], align 4
12716 // CHECK19-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTPREVIOUS_UB__ADDR]], align 4
12717 // CHECK19-NEXT: store i32 [[TMP7]], ptr [[DOTOMP_LB]], align 4
12718 // CHECK19-NEXT: store i32 [[TMP8]], ptr [[DOTOMP_UB]], align 4
12719 // CHECK19-NEXT: store i32 1, ptr [[DOTOMP_STRIDE]], align 4
12720 // CHECK19-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
12721 // CHECK19-NEXT: [[TMP9:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
12722 // CHECK19-NEXT: [[TMP10:%.*]] = load i32, ptr [[TMP9]], align 4
12723 // CHECK19-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB2:[0-9]+]], i32 [[TMP10]], i32 34, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
12724 // CHECK19-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
12725 // CHECK19-NEXT: [[TMP12:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4
12726 // CHECK19-NEXT: [[CMP4:%.*]] = icmp sgt i32 [[TMP11]], [[TMP12]]
12727 // CHECK19-NEXT: br i1 [[CMP4]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
12728 // CHECK19: cond.true:
12729 // CHECK19-NEXT: [[TMP13:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4
12730 // CHECK19-NEXT: br label [[COND_END:%.*]]
12731 // CHECK19: cond.false:
12732 // CHECK19-NEXT: [[TMP14:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
12733 // CHECK19-NEXT: br label [[COND_END]]
12734 // CHECK19: cond.end:
12735 // CHECK19-NEXT: [[COND:%.*]] = phi i32 [ [[TMP13]], [[COND_TRUE]] ], [ [[TMP14]], [[COND_FALSE]] ]
12736 // CHECK19-NEXT: store i32 [[COND]], ptr [[DOTOMP_UB]], align 4
12737 // CHECK19-NEXT: [[TMP15:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4
12738 // CHECK19-NEXT: store i32 [[TMP15]], ptr [[DOTOMP_IV]], align 4
12739 // CHECK19-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
12740 // CHECK19: omp.inner.for.cond:
12741 // CHECK19-NEXT: [[TMP16:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
12742 // CHECK19-NEXT: [[TMP17:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
12743 // CHECK19-NEXT: [[CMP5:%.*]] = icmp sle i32 [[TMP16]], [[TMP17]]
12744 // CHECK19-NEXT: br i1 [[CMP5]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
12745 // CHECK19: omp.inner.for.body:
12746 // CHECK19-NEXT: [[TMP18:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
12747 // CHECK19-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP18]], 1
12748 // CHECK19-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
12749 // CHECK19-NEXT: store i32 [[ADD]], ptr [[I3]], align 4
12750 // CHECK19-NEXT: [[TMP19:%.*]] = load i32, ptr [[I3]], align 4
12751 // CHECK19-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i32, ptr [[TMP2]], i32 [[TMP19]]
12752 // CHECK19-NEXT: store i32 0, ptr [[ARRAYIDX]], align 4
12753 // CHECK19-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
12754 // CHECK19: omp.body.continue:
12755 // CHECK19-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
12756 // CHECK19: omp.inner.for.inc:
12757 // CHECK19-NEXT: [[TMP20:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
12758 // CHECK19-NEXT: [[ADD6:%.*]] = add nsw i32 [[TMP20]], 1
12759 // CHECK19-NEXT: store i32 [[ADD6]], ptr [[DOTOMP_IV]], align 4
12760 // CHECK19-NEXT: br label [[OMP_INNER_FOR_COND]]
12761 // CHECK19: omp.inner.for.end:
12762 // CHECK19-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
12763 // CHECK19: omp.loop.exit:
12764 // CHECK19-NEXT: [[TMP21:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
12765 // CHECK19-NEXT: [[TMP22:%.*]] = load i32, ptr [[TMP21]], align 4
12766 // CHECK19-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB2]], i32 [[TMP22]])
12767 // CHECK19-NEXT: br label [[OMP_PRECOND_END]]
12768 // CHECK19: omp.precond.end:
12769 // CHECK19-NEXT: ret void
12772 // CHECK19-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l153
12773 // CHECK19-SAME: (i32 noundef [[N:%.*]], i32 noundef [[VLA:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[A:%.*]]) #[[ATTR2]] {
12774 // CHECK19-NEXT: entry:
12775 // CHECK19-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4
12776 // CHECK19-NEXT: [[VLA_ADDR:%.*]] = alloca i32, align 4
12777 // CHECK19-NEXT: [[A_ADDR:%.*]] = alloca ptr, align 4
12778 // CHECK19-NEXT: store i32 [[N]], ptr [[N_ADDR]], align 4
12779 // CHECK19-NEXT: store i32 [[VLA]], ptr [[VLA_ADDR]], align 4
12780 // CHECK19-NEXT: store ptr [[A]], ptr [[A_ADDR]], align 4
12781 // CHECK19-NEXT: [[TMP0:%.*]] = load i32, ptr [[VLA_ADDR]], align 4
12782 // CHECK19-NEXT: [[TMP1:%.*]] = load ptr, ptr [[A_ADDR]], align 4
12783 // CHECK19-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_teams(ptr @[[GLOB3]], i32 3, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l153.omp_outlined, ptr [[N_ADDR]], i32 [[TMP0]], ptr [[TMP1]])
12784 // CHECK19-NEXT: ret void
12787 // CHECK19-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l153.omp_outlined
12788 // CHECK19-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[N:%.*]], i32 noundef [[VLA:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[A:%.*]]) #[[ATTR2]] {
12789 // CHECK19-NEXT: entry:
12790 // CHECK19-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 4
12791 // CHECK19-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 4
12792 // CHECK19-NEXT: [[N_ADDR:%.*]] = alloca ptr, align 4
12793 // CHECK19-NEXT: [[VLA_ADDR:%.*]] = alloca i32, align 4
12794 // CHECK19-NEXT: [[A_ADDR:%.*]] = alloca ptr, align 4
12795 // CHECK19-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
12796 // CHECK19-NEXT: [[TMP:%.*]] = alloca i32, align 4
12797 // CHECK19-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4
12798 // CHECK19-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4
12799 // CHECK19-NEXT: [[I:%.*]] = alloca i32, align 4
12800 // CHECK19-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4
12801 // CHECK19-NEXT: [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4
12802 // CHECK19-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
12803 // CHECK19-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
12804 // CHECK19-NEXT: [[I3:%.*]] = alloca i32, align 4
12805 // CHECK19-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 4
12806 // CHECK19-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 4
12807 // CHECK19-NEXT: store ptr [[N]], ptr [[N_ADDR]], align 4
12808 // CHECK19-NEXT: store i32 [[VLA]], ptr [[VLA_ADDR]], align 4
12809 // CHECK19-NEXT: store ptr [[A]], ptr [[A_ADDR]], align 4
12810 // CHECK19-NEXT: [[TMP0:%.*]] = load ptr, ptr [[N_ADDR]], align 4
12811 // CHECK19-NEXT: [[TMP1:%.*]] = load i32, ptr [[VLA_ADDR]], align 4
12812 // CHECK19-NEXT: [[TMP2:%.*]] = load ptr, ptr [[A_ADDR]], align 4
12813 // CHECK19-NEXT: [[TMP3:%.*]] = load i32, ptr [[TMP0]], align 4
12814 // CHECK19-NEXT: store i32 [[TMP3]], ptr [[DOTCAPTURE_EXPR_]], align 4
12815 // CHECK19-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4
12816 // CHECK19-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP4]], 0
12817 // CHECK19-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1
12818 // CHECK19-NEXT: [[SUB2:%.*]] = sub nsw i32 [[DIV]], 1
12819 // CHECK19-NEXT: store i32 [[SUB2]], ptr [[DOTCAPTURE_EXPR_1]], align 4
12820 // CHECK19-NEXT: store i32 0, ptr [[I]], align 4
12821 // CHECK19-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4
12822 // CHECK19-NEXT: [[CMP:%.*]] = icmp slt i32 0, [[TMP5]]
12823 // CHECK19-NEXT: br i1 [[CMP]], label [[OMP_PRECOND_THEN:%.*]], label [[OMP_PRECOND_END:%.*]]
12824 // CHECK19: omp.precond.then:
12825 // CHECK19-NEXT: store i32 0, ptr [[DOTOMP_COMB_LB]], align 4
12826 // CHECK19-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4
12827 // CHECK19-NEXT: store i32 [[TMP6]], ptr [[DOTOMP_COMB_UB]], align 4
12828 // CHECK19-NEXT: store i32 1, ptr [[DOTOMP_STRIDE]], align 4
12829 // CHECK19-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
12830 // CHECK19-NEXT: [[TMP7:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
12831 // CHECK19-NEXT: [[TMP8:%.*]] = load i32, ptr [[TMP7]], align 4
12832 // CHECK19-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB1]], i32 [[TMP8]], i32 92, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
12833 // CHECK19-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
12834 // CHECK19-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4
12835 // CHECK19-NEXT: [[CMP4:%.*]] = icmp sgt i32 [[TMP9]], [[TMP10]]
12836 // CHECK19-NEXT: br i1 [[CMP4]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
12837 // CHECK19: cond.true:
12838 // CHECK19-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4
12839 // CHECK19-NEXT: br label [[COND_END:%.*]]
12840 // CHECK19: cond.false:
12841 // CHECK19-NEXT: [[TMP12:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
12842 // CHECK19-NEXT: br label [[COND_END]]
12843 // CHECK19: cond.end:
12844 // CHECK19-NEXT: [[COND:%.*]] = phi i32 [ [[TMP11]], [[COND_TRUE]] ], [ [[TMP12]], [[COND_FALSE]] ]
12845 // CHECK19-NEXT: store i32 [[COND]], ptr [[DOTOMP_COMB_UB]], align 4
12846 // CHECK19-NEXT: [[TMP13:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
12847 // CHECK19-NEXT: store i32 [[TMP13]], ptr [[DOTOMP_IV]], align 4
12848 // CHECK19-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
12849 // CHECK19: omp.inner.for.cond:
12850 // CHECK19-NEXT: [[TMP14:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
12851 // CHECK19-NEXT: [[TMP15:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
12852 // CHECK19-NEXT: [[CMP5:%.*]] = icmp sle i32 [[TMP14]], [[TMP15]]
12853 // CHECK19-NEXT: br i1 [[CMP5]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
12854 // CHECK19: omp.inner.for.body:
12855 // CHECK19-NEXT: [[TMP16:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
12856 // CHECK19-NEXT: [[TMP17:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
12857 // CHECK19-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_call(ptr @[[GLOB3]], i32 5, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l153.omp_outlined.omp_outlined, i32 [[TMP16]], i32 [[TMP17]], ptr [[TMP0]], i32 [[TMP1]], ptr [[TMP2]])
12858 // CHECK19-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
12859 // CHECK19: omp.inner.for.inc:
12860 // CHECK19-NEXT: [[TMP18:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
12861 // CHECK19-NEXT: [[TMP19:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4
12862 // CHECK19-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP18]], [[TMP19]]
12863 // CHECK19-NEXT: store i32 [[ADD]], ptr [[DOTOMP_IV]], align 4
12864 // CHECK19-NEXT: br label [[OMP_INNER_FOR_COND]]
12865 // CHECK19: omp.inner.for.end:
12866 // CHECK19-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
12867 // CHECK19: omp.loop.exit:
12868 // CHECK19-NEXT: [[TMP20:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
12869 // CHECK19-NEXT: [[TMP21:%.*]] = load i32, ptr [[TMP20]], align 4
12870 // CHECK19-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB1]], i32 [[TMP21]])
12871 // CHECK19-NEXT: br label [[OMP_PRECOND_END]]
12872 // CHECK19: omp.precond.end:
12873 // CHECK19-NEXT: ret void
12876 // CHECK19-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l153.omp_outlined.omp_outlined
12877 // CHECK19-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], i32 noundef [[DOTPREVIOUS_LB_:%.*]], i32 noundef [[DOTPREVIOUS_UB_:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[N:%.*]], i32 noundef [[VLA:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[A:%.*]]) #[[ATTR2]] {
12878 // CHECK19-NEXT: entry:
12879 // CHECK19-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 4
12880 // CHECK19-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 4
12881 // CHECK19-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i32, align 4
12882 // CHECK19-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i32, align 4
12883 // CHECK19-NEXT: [[N_ADDR:%.*]] = alloca ptr, align 4
12884 // CHECK19-NEXT: [[VLA_ADDR:%.*]] = alloca i32, align 4
12885 // CHECK19-NEXT: [[A_ADDR:%.*]] = alloca ptr, align 4
12886 // CHECK19-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
12887 // CHECK19-NEXT: [[TMP:%.*]] = alloca i32, align 4
12888 // CHECK19-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4
12889 // CHECK19-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4
12890 // CHECK19-NEXT: [[I:%.*]] = alloca i32, align 4
12891 // CHECK19-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4
12892 // CHECK19-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4
12893 // CHECK19-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
12894 // CHECK19-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
12895 // CHECK19-NEXT: [[I3:%.*]] = alloca i32, align 4
12896 // CHECK19-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 4
12897 // CHECK19-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 4
12898 // CHECK19-NEXT: store i32 [[DOTPREVIOUS_LB_]], ptr [[DOTPREVIOUS_LB__ADDR]], align 4
12899 // CHECK19-NEXT: store i32 [[DOTPREVIOUS_UB_]], ptr [[DOTPREVIOUS_UB__ADDR]], align 4
12900 // CHECK19-NEXT: store ptr [[N]], ptr [[N_ADDR]], align 4
12901 // CHECK19-NEXT: store i32 [[VLA]], ptr [[VLA_ADDR]], align 4
12902 // CHECK19-NEXT: store ptr [[A]], ptr [[A_ADDR]], align 4
12903 // CHECK19-NEXT: [[TMP0:%.*]] = load ptr, ptr [[N_ADDR]], align 4
12904 // CHECK19-NEXT: [[TMP1:%.*]] = load i32, ptr [[VLA_ADDR]], align 4
12905 // CHECK19-NEXT: [[TMP2:%.*]] = load ptr, ptr [[A_ADDR]], align 4
12906 // CHECK19-NEXT: [[TMP3:%.*]] = load i32, ptr [[TMP0]], align 4
12907 // CHECK19-NEXT: store i32 [[TMP3]], ptr [[DOTCAPTURE_EXPR_]], align 4
12908 // CHECK19-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4
12909 // CHECK19-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP4]], 0
12910 // CHECK19-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1
12911 // CHECK19-NEXT: [[SUB2:%.*]] = sub nsw i32 [[DIV]], 1
12912 // CHECK19-NEXT: store i32 [[SUB2]], ptr [[DOTCAPTURE_EXPR_1]], align 4
12913 // CHECK19-NEXT: store i32 0, ptr [[I]], align 4
12914 // CHECK19-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4
12915 // CHECK19-NEXT: [[CMP:%.*]] = icmp slt i32 0, [[TMP5]]
12916 // CHECK19-NEXT: br i1 [[CMP]], label [[OMP_PRECOND_THEN:%.*]], label [[OMP_PRECOND_END:%.*]]
12917 // CHECK19: omp.precond.then:
12918 // CHECK19-NEXT: store i32 0, ptr [[DOTOMP_LB]], align 4
12919 // CHECK19-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4
12920 // CHECK19-NEXT: store i32 [[TMP6]], ptr [[DOTOMP_UB]], align 4
12921 // CHECK19-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTPREVIOUS_LB__ADDR]], align 4
12922 // CHECK19-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTPREVIOUS_UB__ADDR]], align 4
12923 // CHECK19-NEXT: store i32 [[TMP7]], ptr [[DOTOMP_LB]], align 4
12924 // CHECK19-NEXT: store i32 [[TMP8]], ptr [[DOTOMP_UB]], align 4
12925 // CHECK19-NEXT: store i32 1, ptr [[DOTOMP_STRIDE]], align 4
12926 // CHECK19-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
12927 // CHECK19-NEXT: [[TMP9:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
12928 // CHECK19-NEXT: [[TMP10:%.*]] = load i32, ptr [[TMP9]], align 4
12929 // CHECK19-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB2]], i32 [[TMP10]], i32 34, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
12930 // CHECK19-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
12931 // CHECK19-NEXT: [[TMP12:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4
12932 // CHECK19-NEXT: [[CMP4:%.*]] = icmp sgt i32 [[TMP11]], [[TMP12]]
12933 // CHECK19-NEXT: br i1 [[CMP4]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
12934 // CHECK19: cond.true:
12935 // CHECK19-NEXT: [[TMP13:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4
12936 // CHECK19-NEXT: br label [[COND_END:%.*]]
12937 // CHECK19: cond.false:
12938 // CHECK19-NEXT: [[TMP14:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
12939 // CHECK19-NEXT: br label [[COND_END]]
12940 // CHECK19: cond.end:
12941 // CHECK19-NEXT: [[COND:%.*]] = phi i32 [ [[TMP13]], [[COND_TRUE]] ], [ [[TMP14]], [[COND_FALSE]] ]
12942 // CHECK19-NEXT: store i32 [[COND]], ptr [[DOTOMP_UB]], align 4
12943 // CHECK19-NEXT: [[TMP15:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4
12944 // CHECK19-NEXT: store i32 [[TMP15]], ptr [[DOTOMP_IV]], align 4
12945 // CHECK19-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
12946 // CHECK19: omp.inner.for.cond:
12947 // CHECK19-NEXT: [[TMP16:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
12948 // CHECK19-NEXT: [[TMP17:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
12949 // CHECK19-NEXT: [[CMP5:%.*]] = icmp sle i32 [[TMP16]], [[TMP17]]
12950 // CHECK19-NEXT: br i1 [[CMP5]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
12951 // CHECK19: omp.inner.for.body:
12952 // CHECK19-NEXT: [[TMP18:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
12953 // CHECK19-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP18]], 1
12954 // CHECK19-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
12955 // CHECK19-NEXT: store i32 [[ADD]], ptr [[I3]], align 4
12956 // CHECK19-NEXT: [[TMP19:%.*]] = load i32, ptr [[I3]], align 4
12957 // CHECK19-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i32, ptr [[TMP2]], i32 [[TMP19]]
12958 // CHECK19-NEXT: store i32 0, ptr [[ARRAYIDX]], align 4
12959 // CHECK19-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
12960 // CHECK19: omp.body.continue:
12961 // CHECK19-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
12962 // CHECK19: omp.inner.for.inc:
12963 // CHECK19-NEXT: [[TMP20:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
12964 // CHECK19-NEXT: [[ADD6:%.*]] = add nsw i32 [[TMP20]], 1
12965 // CHECK19-NEXT: store i32 [[ADD6]], ptr [[DOTOMP_IV]], align 4
12966 // CHECK19-NEXT: br label [[OMP_INNER_FOR_COND]]
12967 // CHECK19: omp.inner.for.end:
12968 // CHECK19-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
12969 // CHECK19: omp.loop.exit:
12970 // CHECK19-NEXT: [[TMP21:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
12971 // CHECK19-NEXT: [[TMP22:%.*]] = load i32, ptr [[TMP21]], align 4
12972 // CHECK19-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB2]], i32 [[TMP22]])
12973 // CHECK19-NEXT: br label [[OMP_PRECOND_END]]
12974 // CHECK19: omp.precond.end:
12975 // CHECK19-NEXT: ret void
12978 // CHECK19-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l158
12979 // CHECK19-SAME: (i32 noundef [[M:%.*]], i32 noundef [[N:%.*]], i32 noundef [[VLA:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[A:%.*]]) #[[ATTR2]] {
12980 // CHECK19-NEXT: entry:
12981 // CHECK19-NEXT: [[M_ADDR:%.*]] = alloca i32, align 4
12982 // CHECK19-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4
12983 // CHECK19-NEXT: [[VLA_ADDR:%.*]] = alloca i32, align 4
12984 // CHECK19-NEXT: [[A_ADDR:%.*]] = alloca ptr, align 4
12985 // CHECK19-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4
12986 // CHECK19-NEXT: [[DOTCAPTURE_EXPR__CASTED:%.*]] = alloca i32, align 4
12987 // CHECK19-NEXT: store i32 [[M]], ptr [[M_ADDR]], align 4
12988 // CHECK19-NEXT: store i32 [[N]], ptr [[N_ADDR]], align 4
12989 // CHECK19-NEXT: store i32 [[VLA]], ptr [[VLA_ADDR]], align 4
12990 // CHECK19-NEXT: store ptr [[A]], ptr [[A_ADDR]], align 4
12991 // CHECK19-NEXT: [[TMP0:%.*]] = load i32, ptr [[VLA_ADDR]], align 4
12992 // CHECK19-NEXT: [[TMP1:%.*]] = load ptr, ptr [[A_ADDR]], align 4
12993 // CHECK19-NEXT: [[TMP2:%.*]] = load i32, ptr [[M_ADDR]], align 4
12994 // CHECK19-NEXT: store i32 [[TMP2]], ptr [[DOTCAPTURE_EXPR_]], align 4
12995 // CHECK19-NEXT: [[TMP3:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4
12996 // CHECK19-NEXT: store i32 [[TMP3]], ptr [[DOTCAPTURE_EXPR__CASTED]], align 4
12997 // CHECK19-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR__CASTED]], align 4
12998 // CHECK19-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_teams(ptr @[[GLOB3]], i32 4, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l158.omp_outlined, ptr [[N_ADDR]], i32 [[TMP0]], ptr [[TMP1]], i32 [[TMP4]])
12999 // CHECK19-NEXT: ret void
13002 // CHECK19-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l158.omp_outlined
13003 // CHECK19-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[N:%.*]], i32 noundef [[VLA:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[A:%.*]], i32 noundef [[DOTCAPTURE_EXPR_:%.*]]) #[[ATTR2]] {
13004 // CHECK19-NEXT: entry:
13005 // CHECK19-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 4
13006 // CHECK19-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 4
13007 // CHECK19-NEXT: [[N_ADDR:%.*]] = alloca ptr, align 4
13008 // CHECK19-NEXT: [[VLA_ADDR:%.*]] = alloca i32, align 4
13009 // CHECK19-NEXT: [[A_ADDR:%.*]] = alloca ptr, align 4
13010 // CHECK19-NEXT: [[DOTCAPTURE_EXPR__ADDR:%.*]] = alloca i32, align 4
13011 // CHECK19-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
13012 // CHECK19-NEXT: [[TMP:%.*]] = alloca i32, align 4
13013 // CHECK19-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4
13014 // CHECK19-NEXT: [[DOTCAPTURE_EXPR_2:%.*]] = alloca i32, align 4
13015 // CHECK19-NEXT: [[I:%.*]] = alloca i32, align 4
13016 // CHECK19-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4
13017 // CHECK19-NEXT: [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4
13018 // CHECK19-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
13019 // CHECK19-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
13020 // CHECK19-NEXT: [[I4:%.*]] = alloca i32, align 4
13021 // CHECK19-NEXT: [[DOTCAPTURE_EXPR__CASTED:%.*]] = alloca i32, align 4
13022 // CHECK19-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 4
13023 // CHECK19-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 4
13024 // CHECK19-NEXT: store ptr [[N]], ptr [[N_ADDR]], align 4
13025 // CHECK19-NEXT: store i32 [[VLA]], ptr [[VLA_ADDR]], align 4
13026 // CHECK19-NEXT: store ptr [[A]], ptr [[A_ADDR]], align 4
13027 // CHECK19-NEXT: store i32 [[DOTCAPTURE_EXPR_]], ptr [[DOTCAPTURE_EXPR__ADDR]], align 4
13028 // CHECK19-NEXT: [[TMP0:%.*]] = load ptr, ptr [[N_ADDR]], align 4
13029 // CHECK19-NEXT: [[TMP1:%.*]] = load i32, ptr [[VLA_ADDR]], align 4
13030 // CHECK19-NEXT: [[TMP2:%.*]] = load ptr, ptr [[A_ADDR]], align 4
13031 // CHECK19-NEXT: [[TMP3:%.*]] = load i32, ptr [[TMP0]], align 4
13032 // CHECK19-NEXT: store i32 [[TMP3]], ptr [[DOTCAPTURE_EXPR_1]], align 4
13033 // CHECK19-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4
13034 // CHECK19-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP4]], 0
13035 // CHECK19-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1
13036 // CHECK19-NEXT: [[SUB3:%.*]] = sub nsw i32 [[DIV]], 1
13037 // CHECK19-NEXT: store i32 [[SUB3]], ptr [[DOTCAPTURE_EXPR_2]], align 4
13038 // CHECK19-NEXT: store i32 0, ptr [[I]], align 4
13039 // CHECK19-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4
13040 // CHECK19-NEXT: [[CMP:%.*]] = icmp slt i32 0, [[TMP5]]
13041 // CHECK19-NEXT: br i1 [[CMP]], label [[OMP_PRECOND_THEN:%.*]], label [[OMP_PRECOND_END:%.*]]
13042 // CHECK19: omp.precond.then:
13043 // CHECK19-NEXT: store i32 0, ptr [[DOTOMP_COMB_LB]], align 4
13044 // CHECK19-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_2]], align 4
13045 // CHECK19-NEXT: store i32 [[TMP6]], ptr [[DOTOMP_COMB_UB]], align 4
13046 // CHECK19-NEXT: store i32 1, ptr [[DOTOMP_STRIDE]], align 4
13047 // CHECK19-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
13048 // CHECK19-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR__ADDR]], align 4
13049 // CHECK19-NEXT: [[TMP8:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
13050 // CHECK19-NEXT: [[TMP9:%.*]] = load i32, ptr [[TMP8]], align 4
13051 // CHECK19-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB1]], i32 [[TMP9]], i32 91, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 [[TMP7]])
13052 // CHECK19-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
13053 // CHECK19-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_2]], align 4
13054 // CHECK19-NEXT: [[CMP5:%.*]] = icmp sgt i32 [[TMP10]], [[TMP11]]
13055 // CHECK19-NEXT: br i1 [[CMP5]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
13056 // CHECK19: cond.true:
13057 // CHECK19-NEXT: [[TMP12:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_2]], align 4
13058 // CHECK19-NEXT: br label [[COND_END:%.*]]
13059 // CHECK19: cond.false:
13060 // CHECK19-NEXT: [[TMP13:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
13061 // CHECK19-NEXT: br label [[COND_END]]
13062 // CHECK19: cond.end:
13063 // CHECK19-NEXT: [[COND:%.*]] = phi i32 [ [[TMP12]], [[COND_TRUE]] ], [ [[TMP13]], [[COND_FALSE]] ]
13064 // CHECK19-NEXT: store i32 [[COND]], ptr [[DOTOMP_COMB_UB]], align 4
13065 // CHECK19-NEXT: [[TMP14:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
13066 // CHECK19-NEXT: store i32 [[TMP14]], ptr [[DOTOMP_IV]], align 4
13067 // CHECK19-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
13068 // CHECK19: omp.inner.for.cond:
13069 // CHECK19-NEXT: [[TMP15:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
13070 // CHECK19-NEXT: [[TMP16:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_2]], align 4
13071 // CHECK19-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP16]], 1
13072 // CHECK19-NEXT: [[CMP6:%.*]] = icmp slt i32 [[TMP15]], [[ADD]]
13073 // CHECK19-NEXT: br i1 [[CMP6]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
13074 // CHECK19: omp.inner.for.body:
13075 // CHECK19-NEXT: [[TMP17:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
13076 // CHECK19-NEXT: [[TMP18:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
13077 // CHECK19-NEXT: [[TMP19:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR__ADDR]], align 4
13078 // CHECK19-NEXT: store i32 [[TMP19]], ptr [[DOTCAPTURE_EXPR__CASTED]], align 4
13079 // CHECK19-NEXT: [[TMP20:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR__CASTED]], align 4
13080 // CHECK19-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_call(ptr @[[GLOB3]], i32 6, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l158.omp_outlined.omp_outlined, i32 [[TMP17]], i32 [[TMP18]], ptr [[TMP0]], i32 [[TMP1]], ptr [[TMP2]], i32 [[TMP20]])
13081 // CHECK19-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
13082 // CHECK19: omp.inner.for.inc:
13083 // CHECK19-NEXT: [[TMP21:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
13084 // CHECK19-NEXT: [[TMP22:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4
13085 // CHECK19-NEXT: [[ADD7:%.*]] = add nsw i32 [[TMP21]], [[TMP22]]
13086 // CHECK19-NEXT: store i32 [[ADD7]], ptr [[DOTOMP_IV]], align 4
13087 // CHECK19-NEXT: [[TMP23:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
13088 // CHECK19-NEXT: [[TMP24:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4
13089 // CHECK19-NEXT: [[ADD8:%.*]] = add nsw i32 [[TMP23]], [[TMP24]]
13090 // CHECK19-NEXT: store i32 [[ADD8]], ptr [[DOTOMP_COMB_LB]], align 4
13091 // CHECK19-NEXT: [[TMP25:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
13092 // CHECK19-NEXT: [[TMP26:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4
13093 // CHECK19-NEXT: [[ADD9:%.*]] = add nsw i32 [[TMP25]], [[TMP26]]
13094 // CHECK19-NEXT: store i32 [[ADD9]], ptr [[DOTOMP_COMB_UB]], align 4
13095 // CHECK19-NEXT: [[TMP27:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
13096 // CHECK19-NEXT: [[TMP28:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_2]], align 4
13097 // CHECK19-NEXT: [[CMP10:%.*]] = icmp sgt i32 [[TMP27]], [[TMP28]]
13098 // CHECK19-NEXT: br i1 [[CMP10]], label [[COND_TRUE11:%.*]], label [[COND_FALSE12:%.*]]
13099 // CHECK19: cond.true11:
13100 // CHECK19-NEXT: [[TMP29:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_2]], align 4
13101 // CHECK19-NEXT: br label [[COND_END13:%.*]]
13102 // CHECK19: cond.false12:
13103 // CHECK19-NEXT: [[TMP30:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
13104 // CHECK19-NEXT: br label [[COND_END13]]
13105 // CHECK19: cond.end13:
13106 // CHECK19-NEXT: [[COND14:%.*]] = phi i32 [ [[TMP29]], [[COND_TRUE11]] ], [ [[TMP30]], [[COND_FALSE12]] ]
13107 // CHECK19-NEXT: store i32 [[COND14]], ptr [[DOTOMP_COMB_UB]], align 4
13108 // CHECK19-NEXT: [[TMP31:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
13109 // CHECK19-NEXT: store i32 [[TMP31]], ptr [[DOTOMP_IV]], align 4
13110 // CHECK19-NEXT: br label [[OMP_INNER_FOR_COND]]
13111 // CHECK19: omp.inner.for.end:
13112 // CHECK19-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
13113 // CHECK19: omp.loop.exit:
13114 // CHECK19-NEXT: [[TMP32:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
13115 // CHECK19-NEXT: [[TMP33:%.*]] = load i32, ptr [[TMP32]], align 4
13116 // CHECK19-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB1]], i32 [[TMP33]])
13117 // CHECK19-NEXT: br label [[OMP_PRECOND_END]]
13118 // CHECK19: omp.precond.end:
13119 // CHECK19-NEXT: ret void
13122 // CHECK19-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l158.omp_outlined.omp_outlined
13123 // CHECK19-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], i32 noundef [[DOTPREVIOUS_LB_:%.*]], i32 noundef [[DOTPREVIOUS_UB_:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[N:%.*]], i32 noundef [[VLA:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[A:%.*]], i32 noundef [[DOTCAPTURE_EXPR_:%.*]]) #[[ATTR2]] {
13124 // CHECK19-NEXT: entry:
13125 // CHECK19-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 4
13126 // CHECK19-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 4
13127 // CHECK19-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i32, align 4
13128 // CHECK19-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i32, align 4
13129 // CHECK19-NEXT: [[N_ADDR:%.*]] = alloca ptr, align 4
13130 // CHECK19-NEXT: [[VLA_ADDR:%.*]] = alloca i32, align 4
13131 // CHECK19-NEXT: [[A_ADDR:%.*]] = alloca ptr, align 4
13132 // CHECK19-NEXT: [[DOTCAPTURE_EXPR__ADDR:%.*]] = alloca i32, align 4
13133 // CHECK19-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
13134 // CHECK19-NEXT: [[TMP:%.*]] = alloca i32, align 4
13135 // CHECK19-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4
13136 // CHECK19-NEXT: [[DOTCAPTURE_EXPR_2:%.*]] = alloca i32, align 4
13137 // CHECK19-NEXT: [[I:%.*]] = alloca i32, align 4
13138 // CHECK19-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4
13139 // CHECK19-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4
13140 // CHECK19-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
13141 // CHECK19-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
13142 // CHECK19-NEXT: [[I4:%.*]] = alloca i32, align 4
13143 // CHECK19-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 4
13144 // CHECK19-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 4
13145 // CHECK19-NEXT: store i32 [[DOTPREVIOUS_LB_]], ptr [[DOTPREVIOUS_LB__ADDR]], align 4
13146 // CHECK19-NEXT: store i32 [[DOTPREVIOUS_UB_]], ptr [[DOTPREVIOUS_UB__ADDR]], align 4
13147 // CHECK19-NEXT: store ptr [[N]], ptr [[N_ADDR]], align 4
13148 // CHECK19-NEXT: store i32 [[VLA]], ptr [[VLA_ADDR]], align 4
13149 // CHECK19-NEXT: store ptr [[A]], ptr [[A_ADDR]], align 4
13150 // CHECK19-NEXT: store i32 [[DOTCAPTURE_EXPR_]], ptr [[DOTCAPTURE_EXPR__ADDR]], align 4
13151 // CHECK19-NEXT: [[TMP0:%.*]] = load ptr, ptr [[N_ADDR]], align 4
13152 // CHECK19-NEXT: [[TMP1:%.*]] = load i32, ptr [[VLA_ADDR]], align 4
13153 // CHECK19-NEXT: [[TMP2:%.*]] = load ptr, ptr [[A_ADDR]], align 4
13154 // CHECK19-NEXT: [[TMP3:%.*]] = load i32, ptr [[TMP0]], align 4
13155 // CHECK19-NEXT: store i32 [[TMP3]], ptr [[DOTCAPTURE_EXPR_1]], align 4
13156 // CHECK19-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4
13157 // CHECK19-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP4]], 0
13158 // CHECK19-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1
13159 // CHECK19-NEXT: [[SUB3:%.*]] = sub nsw i32 [[DIV]], 1
13160 // CHECK19-NEXT: store i32 [[SUB3]], ptr [[DOTCAPTURE_EXPR_2]], align 4
13161 // CHECK19-NEXT: store i32 0, ptr [[I]], align 4
13162 // CHECK19-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4
13163 // CHECK19-NEXT: [[CMP:%.*]] = icmp slt i32 0, [[TMP5]]
13164 // CHECK19-NEXT: br i1 [[CMP]], label [[OMP_PRECOND_THEN:%.*]], label [[OMP_PRECOND_END:%.*]]
13165 // CHECK19: omp.precond.then:
13166 // CHECK19-NEXT: store i32 0, ptr [[DOTOMP_LB]], align 4
13167 // CHECK19-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_2]], align 4
13168 // CHECK19-NEXT: store i32 [[TMP6]], ptr [[DOTOMP_UB]], align 4
13169 // CHECK19-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTPREVIOUS_LB__ADDR]], align 4
13170 // CHECK19-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTPREVIOUS_UB__ADDR]], align 4
13171 // CHECK19-NEXT: store i32 [[TMP7]], ptr [[DOTOMP_LB]], align 4
13172 // CHECK19-NEXT: store i32 [[TMP8]], ptr [[DOTOMP_UB]], align 4
13173 // CHECK19-NEXT: store i32 1, ptr [[DOTOMP_STRIDE]], align 4
13174 // CHECK19-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
13175 // CHECK19-NEXT: [[TMP9:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
13176 // CHECK19-NEXT: [[TMP10:%.*]] = load i32, ptr [[TMP9]], align 4
13177 // CHECK19-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB2]], i32 [[TMP10]], i32 34, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
13178 // CHECK19-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
13179 // CHECK19-NEXT: [[TMP12:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_2]], align 4
13180 // CHECK19-NEXT: [[CMP5:%.*]] = icmp sgt i32 [[TMP11]], [[TMP12]]
13181 // CHECK19-NEXT: br i1 [[CMP5]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
13182 // CHECK19: cond.true:
13183 // CHECK19-NEXT: [[TMP13:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_2]], align 4
13184 // CHECK19-NEXT: br label [[COND_END:%.*]]
13185 // CHECK19: cond.false:
13186 // CHECK19-NEXT: [[TMP14:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
13187 // CHECK19-NEXT: br label [[COND_END]]
13188 // CHECK19: cond.end:
13189 // CHECK19-NEXT: [[COND:%.*]] = phi i32 [ [[TMP13]], [[COND_TRUE]] ], [ [[TMP14]], [[COND_FALSE]] ]
13190 // CHECK19-NEXT: store i32 [[COND]], ptr [[DOTOMP_UB]], align 4
13191 // CHECK19-NEXT: [[TMP15:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4
13192 // CHECK19-NEXT: store i32 [[TMP15]], ptr [[DOTOMP_IV]], align 4
13193 // CHECK19-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
13194 // CHECK19: omp.inner.for.cond:
13195 // CHECK19-NEXT: [[TMP16:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
13196 // CHECK19-NEXT: [[TMP17:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
13197 // CHECK19-NEXT: [[CMP6:%.*]] = icmp sle i32 [[TMP16]], [[TMP17]]
13198 // CHECK19-NEXT: br i1 [[CMP6]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
13199 // CHECK19: omp.inner.for.body:
13200 // CHECK19-NEXT: [[TMP18:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
13201 // CHECK19-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP18]], 1
13202 // CHECK19-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
13203 // CHECK19-NEXT: store i32 [[ADD]], ptr [[I4]], align 4
13204 // CHECK19-NEXT: [[TMP19:%.*]] = load i32, ptr [[I4]], align 4
13205 // CHECK19-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i32, ptr [[TMP2]], i32 [[TMP19]]
13206 // CHECK19-NEXT: store i32 0, ptr [[ARRAYIDX]], align 4
13207 // CHECK19-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
13208 // CHECK19: omp.body.continue:
13209 // CHECK19-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
13210 // CHECK19: omp.inner.for.inc:
13211 // CHECK19-NEXT: [[TMP20:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
13212 // CHECK19-NEXT: [[ADD7:%.*]] = add nsw i32 [[TMP20]], 1
13213 // CHECK19-NEXT: store i32 [[ADD7]], ptr [[DOTOMP_IV]], align 4
13214 // CHECK19-NEXT: br label [[OMP_INNER_FOR_COND]]
13215 // CHECK19: omp.inner.for.end:
13216 // CHECK19-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
13217 // CHECK19: omp.loop.exit:
13218 // CHECK19-NEXT: [[TMP21:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
13219 // CHECK19-NEXT: [[TMP22:%.*]] = load i32, ptr [[TMP21]], align 4
13220 // CHECK19-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB2]], i32 [[TMP22]])
13221 // CHECK19-NEXT: br label [[OMP_PRECOND_END]]
13222 // CHECK19: omp.precond.end:
13223 // CHECK19-NEXT: ret void
13226 // CHECK19-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l163
13227 // CHECK19-SAME: (i32 noundef [[N:%.*]], i32 noundef [[VLA:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[A:%.*]]) #[[ATTR2]] {
13228 // CHECK19-NEXT: entry:
13229 // CHECK19-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4
13230 // CHECK19-NEXT: [[VLA_ADDR:%.*]] = alloca i32, align 4
13231 // CHECK19-NEXT: [[A_ADDR:%.*]] = alloca ptr, align 4
13232 // CHECK19-NEXT: store i32 [[N]], ptr [[N_ADDR]], align 4
13233 // CHECK19-NEXT: store i32 [[VLA]], ptr [[VLA_ADDR]], align 4
13234 // CHECK19-NEXT: store ptr [[A]], ptr [[A_ADDR]], align 4
13235 // CHECK19-NEXT: [[TMP0:%.*]] = load i32, ptr [[VLA_ADDR]], align 4
13236 // CHECK19-NEXT: [[TMP1:%.*]] = load ptr, ptr [[A_ADDR]], align 4
13237 // CHECK19-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_teams(ptr @[[GLOB3]], i32 3, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l163.omp_outlined, ptr [[N_ADDR]], i32 [[TMP0]], ptr [[TMP1]])
13238 // CHECK19-NEXT: ret void
13241 // CHECK19-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l163.omp_outlined
13242 // CHECK19-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[N:%.*]], i32 noundef [[VLA:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[A:%.*]]) #[[ATTR2]] {
13243 // CHECK19-NEXT: entry:
13244 // CHECK19-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 4
13245 // CHECK19-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 4
13246 // CHECK19-NEXT: [[N_ADDR:%.*]] = alloca ptr, align 4
13247 // CHECK19-NEXT: [[VLA_ADDR:%.*]] = alloca i32, align 4
13248 // CHECK19-NEXT: [[A_ADDR:%.*]] = alloca ptr, align 4
13249 // CHECK19-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
13250 // CHECK19-NEXT: [[TMP:%.*]] = alloca i32, align 4
13251 // CHECK19-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4
13252 // CHECK19-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4
13253 // CHECK19-NEXT: [[I:%.*]] = alloca i32, align 4
13254 // CHECK19-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4
13255 // CHECK19-NEXT: [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4
13256 // CHECK19-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
13257 // CHECK19-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
13258 // CHECK19-NEXT: [[I3:%.*]] = alloca i32, align 4
13259 // CHECK19-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 4
13260 // CHECK19-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 4
13261 // CHECK19-NEXT: store ptr [[N]], ptr [[N_ADDR]], align 4
13262 // CHECK19-NEXT: store i32 [[VLA]], ptr [[VLA_ADDR]], align 4
13263 // CHECK19-NEXT: store ptr [[A]], ptr [[A_ADDR]], align 4
13264 // CHECK19-NEXT: [[TMP0:%.*]] = load ptr, ptr [[N_ADDR]], align 4
13265 // CHECK19-NEXT: [[TMP1:%.*]] = load i32, ptr [[VLA_ADDR]], align 4
13266 // CHECK19-NEXT: [[TMP2:%.*]] = load ptr, ptr [[A_ADDR]], align 4
13267 // CHECK19-NEXT: [[TMP3:%.*]] = load i32, ptr [[TMP0]], align 4
13268 // CHECK19-NEXT: store i32 [[TMP3]], ptr [[DOTCAPTURE_EXPR_]], align 4
13269 // CHECK19-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4
13270 // CHECK19-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP4]], 0
13271 // CHECK19-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1
13272 // CHECK19-NEXT: [[SUB2:%.*]] = sub nsw i32 [[DIV]], 1
13273 // CHECK19-NEXT: store i32 [[SUB2]], ptr [[DOTCAPTURE_EXPR_1]], align 4
13274 // CHECK19-NEXT: store i32 0, ptr [[I]], align 4
13275 // CHECK19-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4
13276 // CHECK19-NEXT: [[CMP:%.*]] = icmp slt i32 0, [[TMP5]]
13277 // CHECK19-NEXT: br i1 [[CMP]], label [[OMP_PRECOND_THEN:%.*]], label [[OMP_PRECOND_END:%.*]]
13278 // CHECK19: omp.precond.then:
13279 // CHECK19-NEXT: store i32 0, ptr [[DOTOMP_COMB_LB]], align 4
13280 // CHECK19-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4
13281 // CHECK19-NEXT: store i32 [[TMP6]], ptr [[DOTOMP_COMB_UB]], align 4
13282 // CHECK19-NEXT: store i32 1, ptr [[DOTOMP_STRIDE]], align 4
13283 // CHECK19-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
13284 // CHECK19-NEXT: [[TMP7:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
13285 // CHECK19-NEXT: [[TMP8:%.*]] = load i32, ptr [[TMP7]], align 4
13286 // CHECK19-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB1]], i32 [[TMP8]], i32 92, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
13287 // CHECK19-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
13288 // CHECK19-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4
13289 // CHECK19-NEXT: [[CMP4:%.*]] = icmp sgt i32 [[TMP9]], [[TMP10]]
13290 // CHECK19-NEXT: br i1 [[CMP4]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
13291 // CHECK19: cond.true:
13292 // CHECK19-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4
13293 // CHECK19-NEXT: br label [[COND_END:%.*]]
13294 // CHECK19: cond.false:
13295 // CHECK19-NEXT: [[TMP12:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
13296 // CHECK19-NEXT: br label [[COND_END]]
13297 // CHECK19: cond.end:
13298 // CHECK19-NEXT: [[COND:%.*]] = phi i32 [ [[TMP11]], [[COND_TRUE]] ], [ [[TMP12]], [[COND_FALSE]] ]
13299 // CHECK19-NEXT: store i32 [[COND]], ptr [[DOTOMP_COMB_UB]], align 4
13300 // CHECK19-NEXT: [[TMP13:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
13301 // CHECK19-NEXT: store i32 [[TMP13]], ptr [[DOTOMP_IV]], align 4
13302 // CHECK19-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
13303 // CHECK19: omp.inner.for.cond:
13304 // CHECK19-NEXT: [[TMP14:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
13305 // CHECK19-NEXT: [[TMP15:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
13306 // CHECK19-NEXT: [[CMP5:%.*]] = icmp sle i32 [[TMP14]], [[TMP15]]
13307 // CHECK19-NEXT: br i1 [[CMP5]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
13308 // CHECK19: omp.inner.for.body:
13309 // CHECK19-NEXT: [[TMP16:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
13310 // CHECK19-NEXT: [[TMP17:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
13311 // CHECK19-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_call(ptr @[[GLOB3]], i32 5, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l163.omp_outlined.omp_outlined, i32 [[TMP16]], i32 [[TMP17]], ptr [[TMP0]], i32 [[TMP1]], ptr [[TMP2]])
13312 // CHECK19-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
13313 // CHECK19: omp.inner.for.inc:
13314 // CHECK19-NEXT: [[TMP18:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
13315 // CHECK19-NEXT: [[TMP19:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4
13316 // CHECK19-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP18]], [[TMP19]]
13317 // CHECK19-NEXT: store i32 [[ADD]], ptr [[DOTOMP_IV]], align 4
13318 // CHECK19-NEXT: br label [[OMP_INNER_FOR_COND]]
13319 // CHECK19: omp.inner.for.end:
13320 // CHECK19-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
13321 // CHECK19: omp.loop.exit:
13322 // CHECK19-NEXT: [[TMP20:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
13323 // CHECK19-NEXT: [[TMP21:%.*]] = load i32, ptr [[TMP20]], align 4
13324 // CHECK19-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB1]], i32 [[TMP21]])
13325 // CHECK19-NEXT: br label [[OMP_PRECOND_END]]
13326 // CHECK19: omp.precond.end:
13327 // CHECK19-NEXT: ret void
13330 // CHECK19-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l163.omp_outlined.omp_outlined
13331 // CHECK19-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], i32 noundef [[DOTPREVIOUS_LB_:%.*]], i32 noundef [[DOTPREVIOUS_UB_:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[N:%.*]], i32 noundef [[VLA:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[A:%.*]]) #[[ATTR2]] {
13332 // CHECK19-NEXT: entry:
13333 // CHECK19-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 4
13334 // CHECK19-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 4
13335 // CHECK19-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i32, align 4
13336 // CHECK19-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i32, align 4
13337 // CHECK19-NEXT: [[N_ADDR:%.*]] = alloca ptr, align 4
13338 // CHECK19-NEXT: [[VLA_ADDR:%.*]] = alloca i32, align 4
13339 // CHECK19-NEXT: [[A_ADDR:%.*]] = alloca ptr, align 4
13340 // CHECK19-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
13341 // CHECK19-NEXT: [[TMP:%.*]] = alloca i32, align 4
13342 // CHECK19-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4
13343 // CHECK19-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4
13344 // CHECK19-NEXT: [[I:%.*]] = alloca i32, align 4
13345 // CHECK19-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4
13346 // CHECK19-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4
13347 // CHECK19-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
13348 // CHECK19-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
13349 // CHECK19-NEXT: [[I3:%.*]] = alloca i32, align 4
13350 // CHECK19-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 4
13351 // CHECK19-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 4
13352 // CHECK19-NEXT: store i32 [[DOTPREVIOUS_LB_]], ptr [[DOTPREVIOUS_LB__ADDR]], align 4
13353 // CHECK19-NEXT: store i32 [[DOTPREVIOUS_UB_]], ptr [[DOTPREVIOUS_UB__ADDR]], align 4
13354 // CHECK19-NEXT: store ptr [[N]], ptr [[N_ADDR]], align 4
13355 // CHECK19-NEXT: store i32 [[VLA]], ptr [[VLA_ADDR]], align 4
13356 // CHECK19-NEXT: store ptr [[A]], ptr [[A_ADDR]], align 4
13357 // CHECK19-NEXT: [[TMP0:%.*]] = load ptr, ptr [[N_ADDR]], align 4
13358 // CHECK19-NEXT: [[TMP1:%.*]] = load i32, ptr [[VLA_ADDR]], align 4
13359 // CHECK19-NEXT: [[TMP2:%.*]] = load ptr, ptr [[A_ADDR]], align 4
13360 // CHECK19-NEXT: [[TMP3:%.*]] = load i32, ptr [[TMP0]], align 4
13361 // CHECK19-NEXT: store i32 [[TMP3]], ptr [[DOTCAPTURE_EXPR_]], align 4
13362 // CHECK19-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4
13363 // CHECK19-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP4]], 0
13364 // CHECK19-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1
13365 // CHECK19-NEXT: [[SUB2:%.*]] = sub nsw i32 [[DIV]], 1
13366 // CHECK19-NEXT: store i32 [[SUB2]], ptr [[DOTCAPTURE_EXPR_1]], align 4
13367 // CHECK19-NEXT: store i32 0, ptr [[I]], align 4
13368 // CHECK19-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4
13369 // CHECK19-NEXT: [[CMP:%.*]] = icmp slt i32 0, [[TMP5]]
13370 // CHECK19-NEXT: br i1 [[CMP]], label [[OMP_PRECOND_THEN:%.*]], label [[OMP_PRECOND_END:%.*]]
13371 // CHECK19: omp.precond.then:
13372 // CHECK19-NEXT: store i32 0, ptr [[DOTOMP_LB]], align 4
13373 // CHECK19-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4
13374 // CHECK19-NEXT: store i32 [[TMP6]], ptr [[DOTOMP_UB]], align 4
13375 // CHECK19-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTPREVIOUS_LB__ADDR]], align 4
13376 // CHECK19-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTPREVIOUS_UB__ADDR]], align 4
13377 // CHECK19-NEXT: store i32 [[TMP7]], ptr [[DOTOMP_LB]], align 4
13378 // CHECK19-NEXT: store i32 [[TMP8]], ptr [[DOTOMP_UB]], align 4
13379 // CHECK19-NEXT: store i32 1, ptr [[DOTOMP_STRIDE]], align 4
13380 // CHECK19-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
13381 // CHECK19-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4
13382 // CHECK19-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
13383 // CHECK19-NEXT: [[TMP11:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
13384 // CHECK19-NEXT: [[TMP12:%.*]] = load i32, ptr [[TMP11]], align 4
13385 // CHECK19-NEXT: call void @__kmpc_dispatch_init_4(ptr @[[GLOB3]], i32 [[TMP12]], i32 35, i32 [[TMP9]], i32 [[TMP10]], i32 1, i32 1)
13386 // CHECK19-NEXT: br label [[OMP_DISPATCH_COND:%.*]]
13387 // CHECK19: omp.dispatch.cond:
13388 // CHECK19-NEXT: [[TMP13:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
13389 // CHECK19-NEXT: [[TMP14:%.*]] = load i32, ptr [[TMP13]], align 4
13390 // CHECK19-NEXT: [[TMP15:%.*]] = call i32 @__kmpc_dispatch_next_4(ptr @[[GLOB3]], i32 [[TMP14]], ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]])
13391 // CHECK19-NEXT: [[TOBOOL:%.*]] = icmp ne i32 [[TMP15]], 0
13392 // CHECK19-NEXT: br i1 [[TOBOOL]], label [[OMP_DISPATCH_BODY:%.*]], label [[OMP_DISPATCH_END:%.*]]
13393 // CHECK19: omp.dispatch.body:
13394 // CHECK19-NEXT: [[TMP16:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4
13395 // CHECK19-NEXT: store i32 [[TMP16]], ptr [[DOTOMP_IV]], align 4
13396 // CHECK19-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
13397 // CHECK19: omp.inner.for.cond:
13398 // CHECK19-NEXT: [[TMP17:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP16:![0-9]+]]
13399 // CHECK19-NEXT: [[TMP18:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP16]]
13400 // CHECK19-NEXT: [[CMP4:%.*]] = icmp sle i32 [[TMP17]], [[TMP18]]
13401 // CHECK19-NEXT: br i1 [[CMP4]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
13402 // CHECK19: omp.inner.for.body:
13403 // CHECK19-NEXT: [[TMP19:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP16]]
13404 // CHECK19-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP19]], 1
13405 // CHECK19-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
13406 // CHECK19-NEXT: store i32 [[ADD]], ptr [[I3]], align 4, !llvm.access.group [[ACC_GRP16]]
13407 // CHECK19-NEXT: [[TMP20:%.*]] = load i32, ptr [[I3]], align 4, !llvm.access.group [[ACC_GRP16]]
13408 // CHECK19-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i32, ptr [[TMP2]], i32 [[TMP20]]
13409 // CHECK19-NEXT: store i32 0, ptr [[ARRAYIDX]], align 4, !llvm.access.group [[ACC_GRP16]]
13410 // CHECK19-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
13411 // CHECK19: omp.body.continue:
13412 // CHECK19-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
13413 // CHECK19: omp.inner.for.inc:
13414 // CHECK19-NEXT: [[TMP21:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP16]]
13415 // CHECK19-NEXT: [[ADD5:%.*]] = add nsw i32 [[TMP21]], 1
13416 // CHECK19-NEXT: store i32 [[ADD5]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP16]]
13417 // CHECK19-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP17:![0-9]+]]
13418 // CHECK19: omp.inner.for.end:
13419 // CHECK19-NEXT: br label [[OMP_DISPATCH_INC:%.*]]
13420 // CHECK19: omp.dispatch.inc:
13421 // CHECK19-NEXT: br label [[OMP_DISPATCH_COND]]
13422 // CHECK19: omp.dispatch.end:
13423 // CHECK19-NEXT: [[TMP22:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
13424 // CHECK19-NEXT: [[TMP23:%.*]] = load i32, ptr [[TMP22]], align 4
13425 // CHECK19-NEXT: call void @__kmpc_dispatch_deinit(ptr @[[GLOB3]], i32 [[TMP23]])
13426 // CHECK19-NEXT: br label [[OMP_PRECOND_END]]
13427 // CHECK19: omp.precond.end:
13428 // CHECK19-NEXT: ret void
13431 // CHECK19-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l168
13432 // CHECK19-SAME: (i32 noundef [[M:%.*]], i32 noundef [[N:%.*]], i32 noundef [[VLA:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[A:%.*]]) #[[ATTR2]] {
13433 // CHECK19-NEXT: entry:
13434 // CHECK19-NEXT: [[M_ADDR:%.*]] = alloca i32, align 4
13435 // CHECK19-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4
13436 // CHECK19-NEXT: [[VLA_ADDR:%.*]] = alloca i32, align 4
13437 // CHECK19-NEXT: [[A_ADDR:%.*]] = alloca ptr, align 4
13438 // CHECK19-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4
13439 // CHECK19-NEXT: [[DOTCAPTURE_EXPR__CASTED:%.*]] = alloca i32, align 4
13440 // CHECK19-NEXT: store i32 [[M]], ptr [[M_ADDR]], align 4
13441 // CHECK19-NEXT: store i32 [[N]], ptr [[N_ADDR]], align 4
13442 // CHECK19-NEXT: store i32 [[VLA]], ptr [[VLA_ADDR]], align 4
13443 // CHECK19-NEXT: store ptr [[A]], ptr [[A_ADDR]], align 4
13444 // CHECK19-NEXT: [[TMP0:%.*]] = load i32, ptr [[VLA_ADDR]], align 4
13445 // CHECK19-NEXT: [[TMP1:%.*]] = load ptr, ptr [[A_ADDR]], align 4
13446 // CHECK19-NEXT: [[TMP2:%.*]] = load i32, ptr [[M_ADDR]], align 4
13447 // CHECK19-NEXT: store i32 [[TMP2]], ptr [[DOTCAPTURE_EXPR_]], align 4
13448 // CHECK19-NEXT: [[TMP3:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4
13449 // CHECK19-NEXT: store i32 [[TMP3]], ptr [[DOTCAPTURE_EXPR__CASTED]], align 4
13450 // CHECK19-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR__CASTED]], align 4
13451 // CHECK19-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_teams(ptr @[[GLOB3]], i32 4, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l168.omp_outlined, ptr [[N_ADDR]], i32 [[TMP0]], ptr [[TMP1]], i32 [[TMP4]])
13452 // CHECK19-NEXT: ret void
13455 // CHECK19-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l168.omp_outlined
13456 // CHECK19-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[N:%.*]], i32 noundef [[VLA:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[A:%.*]], i32 noundef [[DOTCAPTURE_EXPR_:%.*]]) #[[ATTR2]] {
13457 // CHECK19-NEXT: entry:
13458 // CHECK19-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 4
13459 // CHECK19-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 4
13460 // CHECK19-NEXT: [[N_ADDR:%.*]] = alloca ptr, align 4
13461 // CHECK19-NEXT: [[VLA_ADDR:%.*]] = alloca i32, align 4
13462 // CHECK19-NEXT: [[A_ADDR:%.*]] = alloca ptr, align 4
13463 // CHECK19-NEXT: [[DOTCAPTURE_EXPR__ADDR:%.*]] = alloca i32, align 4
13464 // CHECK19-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
13465 // CHECK19-NEXT: [[TMP:%.*]] = alloca i32, align 4
13466 // CHECK19-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4
13467 // CHECK19-NEXT: [[DOTCAPTURE_EXPR_2:%.*]] = alloca i32, align 4
13468 // CHECK19-NEXT: [[I:%.*]] = alloca i32, align 4
13469 // CHECK19-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4
13470 // CHECK19-NEXT: [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4
13471 // CHECK19-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
13472 // CHECK19-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
13473 // CHECK19-NEXT: [[I4:%.*]] = alloca i32, align 4
13474 // CHECK19-NEXT: [[DOTCAPTURE_EXPR__CASTED:%.*]] = alloca i32, align 4
13475 // CHECK19-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 4
13476 // CHECK19-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 4
13477 // CHECK19-NEXT: store ptr [[N]], ptr [[N_ADDR]], align 4
13478 // CHECK19-NEXT: store i32 [[VLA]], ptr [[VLA_ADDR]], align 4
13479 // CHECK19-NEXT: store ptr [[A]], ptr [[A_ADDR]], align 4
13480 // CHECK19-NEXT: store i32 [[DOTCAPTURE_EXPR_]], ptr [[DOTCAPTURE_EXPR__ADDR]], align 4
13481 // CHECK19-NEXT: [[TMP0:%.*]] = load ptr, ptr [[N_ADDR]], align 4
13482 // CHECK19-NEXT: [[TMP1:%.*]] = load i32, ptr [[VLA_ADDR]], align 4
13483 // CHECK19-NEXT: [[TMP2:%.*]] = load ptr, ptr [[A_ADDR]], align 4
13484 // CHECK19-NEXT: [[TMP3:%.*]] = load i32, ptr [[TMP0]], align 4
13485 // CHECK19-NEXT: store i32 [[TMP3]], ptr [[DOTCAPTURE_EXPR_1]], align 4
13486 // CHECK19-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4
13487 // CHECK19-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP4]], 0
13488 // CHECK19-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1
13489 // CHECK19-NEXT: [[SUB3:%.*]] = sub nsw i32 [[DIV]], 1
13490 // CHECK19-NEXT: store i32 [[SUB3]], ptr [[DOTCAPTURE_EXPR_2]], align 4
13491 // CHECK19-NEXT: store i32 0, ptr [[I]], align 4
13492 // CHECK19-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4
13493 // CHECK19-NEXT: [[CMP:%.*]] = icmp slt i32 0, [[TMP5]]
13494 // CHECK19-NEXT: br i1 [[CMP]], label [[OMP_PRECOND_THEN:%.*]], label [[OMP_PRECOND_END:%.*]]
13495 // CHECK19: omp.precond.then:
13496 // CHECK19-NEXT: store i32 0, ptr [[DOTOMP_COMB_LB]], align 4
13497 // CHECK19-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_2]], align 4
13498 // CHECK19-NEXT: store i32 [[TMP6]], ptr [[DOTOMP_COMB_UB]], align 4
13499 // CHECK19-NEXT: store i32 1, ptr [[DOTOMP_STRIDE]], align 4
13500 // CHECK19-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
13501 // CHECK19-NEXT: [[TMP7:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
13502 // CHECK19-NEXT: [[TMP8:%.*]] = load i32, ptr [[TMP7]], align 4
13503 // CHECK19-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB1]], i32 [[TMP8]], i32 92, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
13504 // CHECK19-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
13505 // CHECK19-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_2]], align 4
13506 // CHECK19-NEXT: [[CMP5:%.*]] = icmp sgt i32 [[TMP9]], [[TMP10]]
13507 // CHECK19-NEXT: br i1 [[CMP5]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
13508 // CHECK19: cond.true:
13509 // CHECK19-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_2]], align 4
13510 // CHECK19-NEXT: br label [[COND_END:%.*]]
13511 // CHECK19: cond.false:
13512 // CHECK19-NEXT: [[TMP12:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
13513 // CHECK19-NEXT: br label [[COND_END]]
13514 // CHECK19: cond.end:
13515 // CHECK19-NEXT: [[COND:%.*]] = phi i32 [ [[TMP11]], [[COND_TRUE]] ], [ [[TMP12]], [[COND_FALSE]] ]
13516 // CHECK19-NEXT: store i32 [[COND]], ptr [[DOTOMP_COMB_UB]], align 4
13517 // CHECK19-NEXT: [[TMP13:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
13518 // CHECK19-NEXT: store i32 [[TMP13]], ptr [[DOTOMP_IV]], align 4
13519 // CHECK19-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
13520 // CHECK19: omp.inner.for.cond:
13521 // CHECK19-NEXT: [[TMP14:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
13522 // CHECK19-NEXT: [[TMP15:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
13523 // CHECK19-NEXT: [[CMP6:%.*]] = icmp sle i32 [[TMP14]], [[TMP15]]
13524 // CHECK19-NEXT: br i1 [[CMP6]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
13525 // CHECK19: omp.inner.for.body:
13526 // CHECK19-NEXT: [[TMP16:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
13527 // CHECK19-NEXT: [[TMP17:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
13528 // CHECK19-NEXT: [[TMP18:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR__ADDR]], align 4
13529 // CHECK19-NEXT: store i32 [[TMP18]], ptr [[DOTCAPTURE_EXPR__CASTED]], align 4
13530 // CHECK19-NEXT: [[TMP19:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR__CASTED]], align 4
13531 // CHECK19-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_call(ptr @[[GLOB3]], i32 6, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l168.omp_outlined.omp_outlined, i32 [[TMP16]], i32 [[TMP17]], ptr [[TMP0]], i32 [[TMP1]], ptr [[TMP2]], i32 [[TMP19]])
13532 // CHECK19-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
13533 // CHECK19: omp.inner.for.inc:
13534 // CHECK19-NEXT: [[TMP20:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
13535 // CHECK19-NEXT: [[TMP21:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4
13536 // CHECK19-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP20]], [[TMP21]]
13537 // CHECK19-NEXT: store i32 [[ADD]], ptr [[DOTOMP_IV]], align 4
13538 // CHECK19-NEXT: br label [[OMP_INNER_FOR_COND]]
13539 // CHECK19: omp.inner.for.end:
13540 // CHECK19-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
13541 // CHECK19: omp.loop.exit:
13542 // CHECK19-NEXT: [[TMP22:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
13543 // CHECK19-NEXT: [[TMP23:%.*]] = load i32, ptr [[TMP22]], align 4
13544 // CHECK19-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB1]], i32 [[TMP23]])
13545 // CHECK19-NEXT: br label [[OMP_PRECOND_END]]
13546 // CHECK19: omp.precond.end:
13547 // CHECK19-NEXT: ret void
13550 // CHECK19-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l168.omp_outlined.omp_outlined
13551 // CHECK19-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], i32 noundef [[DOTPREVIOUS_LB_:%.*]], i32 noundef [[DOTPREVIOUS_UB_:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[N:%.*]], i32 noundef [[VLA:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[A:%.*]], i32 noundef [[DOTCAPTURE_EXPR_:%.*]]) #[[ATTR2]] {
13552 // CHECK19-NEXT: entry:
13553 // CHECK19-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 4
13554 // CHECK19-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 4
13555 // CHECK19-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i32, align 4
13556 // CHECK19-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i32, align 4
13557 // CHECK19-NEXT: [[N_ADDR:%.*]] = alloca ptr, align 4
13558 // CHECK19-NEXT: [[VLA_ADDR:%.*]] = alloca i32, align 4
13559 // CHECK19-NEXT: [[A_ADDR:%.*]] = alloca ptr, align 4
13560 // CHECK19-NEXT: [[DOTCAPTURE_EXPR__ADDR:%.*]] = alloca i32, align 4
13561 // CHECK19-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
13562 // CHECK19-NEXT: [[TMP:%.*]] = alloca i32, align 4
13563 // CHECK19-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4
13564 // CHECK19-NEXT: [[DOTCAPTURE_EXPR_2:%.*]] = alloca i32, align 4
13565 // CHECK19-NEXT: [[I:%.*]] = alloca i32, align 4
13566 // CHECK19-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4
13567 // CHECK19-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4
13568 // CHECK19-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
13569 // CHECK19-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
13570 // CHECK19-NEXT: [[I4:%.*]] = alloca i32, align 4
13571 // CHECK19-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 4
13572 // CHECK19-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 4
13573 // CHECK19-NEXT: store i32 [[DOTPREVIOUS_LB_]], ptr [[DOTPREVIOUS_LB__ADDR]], align 4
13574 // CHECK19-NEXT: store i32 [[DOTPREVIOUS_UB_]], ptr [[DOTPREVIOUS_UB__ADDR]], align 4
13575 // CHECK19-NEXT: store ptr [[N]], ptr [[N_ADDR]], align 4
13576 // CHECK19-NEXT: store i32 [[VLA]], ptr [[VLA_ADDR]], align 4
13577 // CHECK19-NEXT: store ptr [[A]], ptr [[A_ADDR]], align 4
13578 // CHECK19-NEXT: store i32 [[DOTCAPTURE_EXPR_]], ptr [[DOTCAPTURE_EXPR__ADDR]], align 4
13579 // CHECK19-NEXT: [[TMP0:%.*]] = load ptr, ptr [[N_ADDR]], align 4
13580 // CHECK19-NEXT: [[TMP1:%.*]] = load i32, ptr [[VLA_ADDR]], align 4
13581 // CHECK19-NEXT: [[TMP2:%.*]] = load ptr, ptr [[A_ADDR]], align 4
13582 // CHECK19-NEXT: [[TMP3:%.*]] = load i32, ptr [[TMP0]], align 4
13583 // CHECK19-NEXT: store i32 [[TMP3]], ptr [[DOTCAPTURE_EXPR_1]], align 4
13584 // CHECK19-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4
13585 // CHECK19-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP4]], 0
13586 // CHECK19-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1
13587 // CHECK19-NEXT: [[SUB3:%.*]] = sub nsw i32 [[DIV]], 1
13588 // CHECK19-NEXT: store i32 [[SUB3]], ptr [[DOTCAPTURE_EXPR_2]], align 4
13589 // CHECK19-NEXT: store i32 0, ptr [[I]], align 4
13590 // CHECK19-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4
13591 // CHECK19-NEXT: [[CMP:%.*]] = icmp slt i32 0, [[TMP5]]
13592 // CHECK19-NEXT: br i1 [[CMP]], label [[OMP_PRECOND_THEN:%.*]], label [[OMP_PRECOND_END:%.*]]
13593 // CHECK19: omp.precond.then:
13594 // CHECK19-NEXT: store i32 0, ptr [[DOTOMP_LB]], align 4
13595 // CHECK19-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_2]], align 4
13596 // CHECK19-NEXT: store i32 [[TMP6]], ptr [[DOTOMP_UB]], align 4
13597 // CHECK19-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTPREVIOUS_LB__ADDR]], align 4
13598 // CHECK19-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTPREVIOUS_UB__ADDR]], align 4
13599 // CHECK19-NEXT: store i32 [[TMP7]], ptr [[DOTOMP_LB]], align 4
13600 // CHECK19-NEXT: store i32 [[TMP8]], ptr [[DOTOMP_UB]], align 4
13601 // CHECK19-NEXT: store i32 1, ptr [[DOTOMP_STRIDE]], align 4
13602 // CHECK19-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
13603 // CHECK19-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR__ADDR]], align 4
13604 // CHECK19-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4
13605 // CHECK19-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
13606 // CHECK19-NEXT: [[TMP12:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
13607 // CHECK19-NEXT: [[TMP13:%.*]] = load i32, ptr [[TMP12]], align 4
13608 // CHECK19-NEXT: call void @__kmpc_dispatch_init_4(ptr @[[GLOB3]], i32 [[TMP13]], i32 35, i32 [[TMP10]], i32 [[TMP11]], i32 1, i32 [[TMP9]])
13609 // CHECK19-NEXT: br label [[OMP_DISPATCH_COND:%.*]]
13610 // CHECK19: omp.dispatch.cond:
13611 // CHECK19-NEXT: [[TMP14:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
13612 // CHECK19-NEXT: [[TMP15:%.*]] = load i32, ptr [[TMP14]], align 4
13613 // CHECK19-NEXT: [[TMP16:%.*]] = call i32 @__kmpc_dispatch_next_4(ptr @[[GLOB3]], i32 [[TMP15]], ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]])
13614 // CHECK19-NEXT: [[TOBOOL:%.*]] = icmp ne i32 [[TMP16]], 0
13615 // CHECK19-NEXT: br i1 [[TOBOOL]], label [[OMP_DISPATCH_BODY:%.*]], label [[OMP_DISPATCH_END:%.*]]
13616 // CHECK19: omp.dispatch.body:
13617 // CHECK19-NEXT: [[TMP17:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4
13618 // CHECK19-NEXT: store i32 [[TMP17]], ptr [[DOTOMP_IV]], align 4
13619 // CHECK19-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
13620 // CHECK19: omp.inner.for.cond:
13621 // CHECK19-NEXT: [[TMP18:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP19:![0-9]+]]
13622 // CHECK19-NEXT: [[TMP19:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP19]]
13623 // CHECK19-NEXT: [[CMP5:%.*]] = icmp sle i32 [[TMP18]], [[TMP19]]
13624 // CHECK19-NEXT: br i1 [[CMP5]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
13625 // CHECK19: omp.inner.for.body:
13626 // CHECK19-NEXT: [[TMP20:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP19]]
13627 // CHECK19-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP20]], 1
13628 // CHECK19-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
13629 // CHECK19-NEXT: store i32 [[ADD]], ptr [[I4]], align 4, !llvm.access.group [[ACC_GRP19]]
13630 // CHECK19-NEXT: [[TMP21:%.*]] = load i32, ptr [[I4]], align 4, !llvm.access.group [[ACC_GRP19]]
13631 // CHECK19-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i32, ptr [[TMP2]], i32 [[TMP21]]
13632 // CHECK19-NEXT: store i32 0, ptr [[ARRAYIDX]], align 4, !llvm.access.group [[ACC_GRP19]]
13633 // CHECK19-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
13634 // CHECK19: omp.body.continue:
13635 // CHECK19-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
13636 // CHECK19: omp.inner.for.inc:
13637 // CHECK19-NEXT: [[TMP22:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP19]]
13638 // CHECK19-NEXT: [[ADD6:%.*]] = add nsw i32 [[TMP22]], 1
13639 // CHECK19-NEXT: store i32 [[ADD6]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP19]]
13640 // CHECK19-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP20:![0-9]+]]
13641 // CHECK19: omp.inner.for.end:
13642 // CHECK19-NEXT: br label [[OMP_DISPATCH_INC:%.*]]
13643 // CHECK19: omp.dispatch.inc:
13644 // CHECK19-NEXT: br label [[OMP_DISPATCH_COND]]
13645 // CHECK19: omp.dispatch.end:
13646 // CHECK19-NEXT: [[TMP23:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
13647 // CHECK19-NEXT: [[TMP24:%.*]] = load i32, ptr [[TMP23]], align 4
13648 // CHECK19-NEXT: call void @__kmpc_dispatch_deinit(ptr @[[GLOB3]], i32 [[TMP24]])
13649 // CHECK19-NEXT: br label [[OMP_PRECOND_END]]
13650 // CHECK19: omp.precond.end:
13651 // CHECK19-NEXT: ret void
13654 // CHECK19-LABEL: define {{[^@]+}}@_Z5tmainIiLi10EEiT_
13655 // CHECK19-SAME: (i32 noundef [[ARGC:%.*]]) #[[ATTR5:[0-9]+]] comdat {
13656 // CHECK19-NEXT: entry:
13657 // CHECK19-NEXT: [[ARGC_ADDR:%.*]] = alloca i32, align 4
13658 // CHECK19-NEXT: [[A:%.*]] = alloca [10 x i32], align 4
13659 // CHECK19-NEXT: [[M:%.*]] = alloca i32, align 4
13660 // CHECK19-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [1 x ptr], align 4
13661 // CHECK19-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [1 x ptr], align 4
13662 // CHECK19-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [1 x ptr], align 4
13663 // CHECK19-NEXT: [[TMP:%.*]] = alloca i32, align 4
13664 // CHECK19-NEXT: [[KERNEL_ARGS:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS:%.*]], align 8
13665 // CHECK19-NEXT: [[DOTOFFLOAD_BASEPTRS1:%.*]] = alloca [1 x ptr], align 4
13666 // CHECK19-NEXT: [[DOTOFFLOAD_PTRS2:%.*]] = alloca [1 x ptr], align 4
13667 // CHECK19-NEXT: [[DOTOFFLOAD_MAPPERS3:%.*]] = alloca [1 x ptr], align 4
13668 // CHECK19-NEXT: [[_TMP4:%.*]] = alloca i32, align 4
13669 // CHECK19-NEXT: [[KERNEL_ARGS5:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS]], align 8
13670 // CHECK19-NEXT: [[M_CASTED:%.*]] = alloca i32, align 4
13671 // CHECK19-NEXT: [[DOTOFFLOAD_BASEPTRS8:%.*]] = alloca [2 x ptr], align 4
13672 // CHECK19-NEXT: [[DOTOFFLOAD_PTRS9:%.*]] = alloca [2 x ptr], align 4
13673 // CHECK19-NEXT: [[DOTOFFLOAD_MAPPERS10:%.*]] = alloca [2 x ptr], align 4
13674 // CHECK19-NEXT: [[_TMP11:%.*]] = alloca i32, align 4
13675 // CHECK19-NEXT: [[KERNEL_ARGS12:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS]], align 8
13676 // CHECK19-NEXT: [[DOTOFFLOAD_BASEPTRS15:%.*]] = alloca [1 x ptr], align 4
13677 // CHECK19-NEXT: [[DOTOFFLOAD_PTRS16:%.*]] = alloca [1 x ptr], align 4
13678 // CHECK19-NEXT: [[DOTOFFLOAD_MAPPERS17:%.*]] = alloca [1 x ptr], align 4
13679 // CHECK19-NEXT: [[_TMP18:%.*]] = alloca i32, align 4
13680 // CHECK19-NEXT: [[KERNEL_ARGS19:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS]], align 8
13681 // CHECK19-NEXT: [[M_CASTED22:%.*]] = alloca i32, align 4
13682 // CHECK19-NEXT: [[DOTOFFLOAD_BASEPTRS23:%.*]] = alloca [2 x ptr], align 4
13683 // CHECK19-NEXT: [[DOTOFFLOAD_PTRS24:%.*]] = alloca [2 x ptr], align 4
13684 // CHECK19-NEXT: [[DOTOFFLOAD_MAPPERS25:%.*]] = alloca [2 x ptr], align 4
13685 // CHECK19-NEXT: [[_TMP26:%.*]] = alloca i32, align 4
13686 // CHECK19-NEXT: [[KERNEL_ARGS27:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS]], align 8
13687 // CHECK19-NEXT: store i32 [[ARGC]], ptr [[ARGC_ADDR]], align 4
13688 // CHECK19-NEXT: store i32 10, ptr [[M]], align 4
13689 // CHECK19-NEXT: [[TMP0:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
13690 // CHECK19-NEXT: store ptr [[A]], ptr [[TMP0]], align 4
13691 // CHECK19-NEXT: [[TMP1:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
13692 // CHECK19-NEXT: store ptr [[A]], ptr [[TMP1]], align 4
13693 // CHECK19-NEXT: [[TMP2:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0
13694 // CHECK19-NEXT: store ptr null, ptr [[TMP2]], align 4
13695 // CHECK19-NEXT: [[TMP3:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
13696 // CHECK19-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
13697 // CHECK19-NEXT: [[TMP5:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 0
13698 // CHECK19-NEXT: store i32 3, ptr [[TMP5]], align 4
13699 // CHECK19-NEXT: [[TMP6:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 1
13700 // CHECK19-NEXT: store i32 1, ptr [[TMP6]], align 4
13701 // CHECK19-NEXT: [[TMP7:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 2
13702 // CHECK19-NEXT: store ptr [[TMP3]], ptr [[TMP7]], align 4
13703 // CHECK19-NEXT: [[TMP8:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 3
13704 // CHECK19-NEXT: store ptr [[TMP4]], ptr [[TMP8]], align 4
13705 // CHECK19-NEXT: [[TMP9:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 4
13706 // CHECK19-NEXT: store ptr @.offload_sizes.9, ptr [[TMP9]], align 4
13707 // CHECK19-NEXT: [[TMP10:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 5
13708 // CHECK19-NEXT: store ptr @.offload_maptypes.10, ptr [[TMP10]], align 4
13709 // CHECK19-NEXT: [[TMP11:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 6
13710 // CHECK19-NEXT: store ptr null, ptr [[TMP11]], align 4
13711 // CHECK19-NEXT: [[TMP12:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 7
13712 // CHECK19-NEXT: store ptr null, ptr [[TMP12]], align 4
13713 // CHECK19-NEXT: [[TMP13:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 8
13714 // CHECK19-NEXT: store i64 10, ptr [[TMP13]], align 8
13715 // CHECK19-NEXT: [[TMP14:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 9
13716 // CHECK19-NEXT: store i64 0, ptr [[TMP14]], align 8
13717 // CHECK19-NEXT: [[TMP15:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 10
13718 // CHECK19-NEXT: store [3 x i32] zeroinitializer, ptr [[TMP15]], align 4
13719 // CHECK19-NEXT: [[TMP16:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 11
13720 // CHECK19-NEXT: store [3 x i32] zeroinitializer, ptr [[TMP16]], align 4
13721 // CHECK19-NEXT: [[TMP17:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 12
13722 // CHECK19-NEXT: store i32 0, ptr [[TMP17]], align 4
13723 // CHECK19-NEXT: [[TMP18:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB3]], i64 -1, i32 0, i32 0, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l116.region_id, ptr [[KERNEL_ARGS]])
13724 // CHECK19-NEXT: [[TMP19:%.*]] = icmp ne i32 [[TMP18]], 0
13725 // CHECK19-NEXT: br i1 [[TMP19]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
13726 // CHECK19: omp_offload.failed:
13727 // CHECK19-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l116(ptr [[A]]) #[[ATTR3]]
13728 // CHECK19-NEXT: br label [[OMP_OFFLOAD_CONT]]
13729 // CHECK19: omp_offload.cont:
13730 // CHECK19-NEXT: [[TMP20:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS1]], i32 0, i32 0
13731 // CHECK19-NEXT: store ptr [[A]], ptr [[TMP20]], align 4
13732 // CHECK19-NEXT: [[TMP21:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS2]], i32 0, i32 0
13733 // CHECK19-NEXT: store ptr [[A]], ptr [[TMP21]], align 4
13734 // CHECK19-NEXT: [[TMP22:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_MAPPERS3]], i32 0, i32 0
13735 // CHECK19-NEXT: store ptr null, ptr [[TMP22]], align 4
13736 // CHECK19-NEXT: [[TMP23:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS1]], i32 0, i32 0
13737 // CHECK19-NEXT: [[TMP24:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS2]], i32 0, i32 0
13738 // CHECK19-NEXT: [[TMP25:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS5]], i32 0, i32 0
13739 // CHECK19-NEXT: store i32 3, ptr [[TMP25]], align 4
13740 // CHECK19-NEXT: [[TMP26:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS5]], i32 0, i32 1
13741 // CHECK19-NEXT: store i32 1, ptr [[TMP26]], align 4
13742 // CHECK19-NEXT: [[TMP27:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS5]], i32 0, i32 2
13743 // CHECK19-NEXT: store ptr [[TMP23]], ptr [[TMP27]], align 4
13744 // CHECK19-NEXT: [[TMP28:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS5]], i32 0, i32 3
13745 // CHECK19-NEXT: store ptr [[TMP24]], ptr [[TMP28]], align 4
13746 // CHECK19-NEXT: [[TMP29:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS5]], i32 0, i32 4
13747 // CHECK19-NEXT: store ptr @.offload_sizes.11, ptr [[TMP29]], align 4
13748 // CHECK19-NEXT: [[TMP30:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS5]], i32 0, i32 5
13749 // CHECK19-NEXT: store ptr @.offload_maptypes.12, ptr [[TMP30]], align 4
13750 // CHECK19-NEXT: [[TMP31:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS5]], i32 0, i32 6
13751 // CHECK19-NEXT: store ptr null, ptr [[TMP31]], align 4
13752 // CHECK19-NEXT: [[TMP32:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS5]], i32 0, i32 7
13753 // CHECK19-NEXT: store ptr null, ptr [[TMP32]], align 4
13754 // CHECK19-NEXT: [[TMP33:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS5]], i32 0, i32 8
13755 // CHECK19-NEXT: store i64 10, ptr [[TMP33]], align 8
13756 // CHECK19-NEXT: [[TMP34:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS5]], i32 0, i32 9
13757 // CHECK19-NEXT: store i64 0, ptr [[TMP34]], align 8
13758 // CHECK19-NEXT: [[TMP35:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS5]], i32 0, i32 10
13759 // CHECK19-NEXT: store [3 x i32] zeroinitializer, ptr [[TMP35]], align 4
13760 // CHECK19-NEXT: [[TMP36:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS5]], i32 0, i32 11
13761 // CHECK19-NEXT: store [3 x i32] zeroinitializer, ptr [[TMP36]], align 4
13762 // CHECK19-NEXT: [[TMP37:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS5]], i32 0, i32 12
13763 // CHECK19-NEXT: store i32 0, ptr [[TMP37]], align 4
13764 // CHECK19-NEXT: [[TMP38:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB3]], i64 -1, i32 0, i32 0, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l121.region_id, ptr [[KERNEL_ARGS5]])
13765 // CHECK19-NEXT: [[TMP39:%.*]] = icmp ne i32 [[TMP38]], 0
13766 // CHECK19-NEXT: br i1 [[TMP39]], label [[OMP_OFFLOAD_FAILED6:%.*]], label [[OMP_OFFLOAD_CONT7:%.*]]
13767 // CHECK19: omp_offload.failed6:
13768 // CHECK19-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l121(ptr [[A]]) #[[ATTR3]]
13769 // CHECK19-NEXT: br label [[OMP_OFFLOAD_CONT7]]
13770 // CHECK19: omp_offload.cont7:
13771 // CHECK19-NEXT: [[TMP40:%.*]] = load i32, ptr [[M]], align 4
13772 // CHECK19-NEXT: store i32 [[TMP40]], ptr [[M_CASTED]], align 4
13773 // CHECK19-NEXT: [[TMP41:%.*]] = load i32, ptr [[M_CASTED]], align 4
13774 // CHECK19-NEXT: [[TMP42:%.*]] = getelementptr inbounds [2 x ptr], ptr [[DOTOFFLOAD_BASEPTRS8]], i32 0, i32 0
13775 // CHECK19-NEXT: store i32 [[TMP41]], ptr [[TMP42]], align 4
13776 // CHECK19-NEXT: [[TMP43:%.*]] = getelementptr inbounds [2 x ptr], ptr [[DOTOFFLOAD_PTRS9]], i32 0, i32 0
13777 // CHECK19-NEXT: store i32 [[TMP41]], ptr [[TMP43]], align 4
13778 // CHECK19-NEXT: [[TMP44:%.*]] = getelementptr inbounds [2 x ptr], ptr [[DOTOFFLOAD_MAPPERS10]], i32 0, i32 0
13779 // CHECK19-NEXT: store ptr null, ptr [[TMP44]], align 4
13780 // CHECK19-NEXT: [[TMP45:%.*]] = getelementptr inbounds [2 x ptr], ptr [[DOTOFFLOAD_BASEPTRS8]], i32 0, i32 1
13781 // CHECK19-NEXT: store ptr [[A]], ptr [[TMP45]], align 4
13782 // CHECK19-NEXT: [[TMP46:%.*]] = getelementptr inbounds [2 x ptr], ptr [[DOTOFFLOAD_PTRS9]], i32 0, i32 1
13783 // CHECK19-NEXT: store ptr [[A]], ptr [[TMP46]], align 4
13784 // CHECK19-NEXT: [[TMP47:%.*]] = getelementptr inbounds [2 x ptr], ptr [[DOTOFFLOAD_MAPPERS10]], i32 0, i32 1
13785 // CHECK19-NEXT: store ptr null, ptr [[TMP47]], align 4
13786 // CHECK19-NEXT: [[TMP48:%.*]] = getelementptr inbounds [2 x ptr], ptr [[DOTOFFLOAD_BASEPTRS8]], i32 0, i32 0
13787 // CHECK19-NEXT: [[TMP49:%.*]] = getelementptr inbounds [2 x ptr], ptr [[DOTOFFLOAD_PTRS9]], i32 0, i32 0
13788 // CHECK19-NEXT: [[TMP50:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS12]], i32 0, i32 0
13789 // CHECK19-NEXT: store i32 3, ptr [[TMP50]], align 4
13790 // CHECK19-NEXT: [[TMP51:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS12]], i32 0, i32 1
13791 // CHECK19-NEXT: store i32 2, ptr [[TMP51]], align 4
13792 // CHECK19-NEXT: [[TMP52:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS12]], i32 0, i32 2
13793 // CHECK19-NEXT: store ptr [[TMP48]], ptr [[TMP52]], align 4
13794 // CHECK19-NEXT: [[TMP53:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS12]], i32 0, i32 3
13795 // CHECK19-NEXT: store ptr [[TMP49]], ptr [[TMP53]], align 4
13796 // CHECK19-NEXT: [[TMP54:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS12]], i32 0, i32 4
13797 // CHECK19-NEXT: store ptr @.offload_sizes.13, ptr [[TMP54]], align 4
13798 // CHECK19-NEXT: [[TMP55:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS12]], i32 0, i32 5
13799 // CHECK19-NEXT: store ptr @.offload_maptypes.14, ptr [[TMP55]], align 4
13800 // CHECK19-NEXT: [[TMP56:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS12]], i32 0, i32 6
13801 // CHECK19-NEXT: store ptr null, ptr [[TMP56]], align 4
13802 // CHECK19-NEXT: [[TMP57:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS12]], i32 0, i32 7
13803 // CHECK19-NEXT: store ptr null, ptr [[TMP57]], align 4
13804 // CHECK19-NEXT: [[TMP58:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS12]], i32 0, i32 8
13805 // CHECK19-NEXT: store i64 10, ptr [[TMP58]], align 8
13806 // CHECK19-NEXT: [[TMP59:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS12]], i32 0, i32 9
13807 // CHECK19-NEXT: store i64 0, ptr [[TMP59]], align 8
13808 // CHECK19-NEXT: [[TMP60:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS12]], i32 0, i32 10
13809 // CHECK19-NEXT: store [3 x i32] zeroinitializer, ptr [[TMP60]], align 4
13810 // CHECK19-NEXT: [[TMP61:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS12]], i32 0, i32 11
13811 // CHECK19-NEXT: store [3 x i32] zeroinitializer, ptr [[TMP61]], align 4
13812 // CHECK19-NEXT: [[TMP62:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS12]], i32 0, i32 12
13813 // CHECK19-NEXT: store i32 0, ptr [[TMP62]], align 4
13814 // CHECK19-NEXT: [[TMP63:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB3]], i64 -1, i32 0, i32 0, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l126.region_id, ptr [[KERNEL_ARGS12]])
13815 // CHECK19-NEXT: [[TMP64:%.*]] = icmp ne i32 [[TMP63]], 0
13816 // CHECK19-NEXT: br i1 [[TMP64]], label [[OMP_OFFLOAD_FAILED13:%.*]], label [[OMP_OFFLOAD_CONT14:%.*]]
13817 // CHECK19: omp_offload.failed13:
13818 // CHECK19-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l126(i32 [[TMP41]], ptr [[A]]) #[[ATTR3]]
13819 // CHECK19-NEXT: br label [[OMP_OFFLOAD_CONT14]]
13820 // CHECK19: omp_offload.cont14:
13821 // CHECK19-NEXT: [[TMP65:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS15]], i32 0, i32 0
13822 // CHECK19-NEXT: store ptr [[A]], ptr [[TMP65]], align 4
13823 // CHECK19-NEXT: [[TMP66:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS16]], i32 0, i32 0
13824 // CHECK19-NEXT: store ptr [[A]], ptr [[TMP66]], align 4
13825 // CHECK19-NEXT: [[TMP67:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_MAPPERS17]], i32 0, i32 0
13826 // CHECK19-NEXT: store ptr null, ptr [[TMP67]], align 4
13827 // CHECK19-NEXT: [[TMP68:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS15]], i32 0, i32 0
13828 // CHECK19-NEXT: [[TMP69:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS16]], i32 0, i32 0
13829 // CHECK19-NEXT: [[TMP70:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS19]], i32 0, i32 0
13830 // CHECK19-NEXT: store i32 3, ptr [[TMP70]], align 4
13831 // CHECK19-NEXT: [[TMP71:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS19]], i32 0, i32 1
13832 // CHECK19-NEXT: store i32 1, ptr [[TMP71]], align 4
13833 // CHECK19-NEXT: [[TMP72:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS19]], i32 0, i32 2
13834 // CHECK19-NEXT: store ptr [[TMP68]], ptr [[TMP72]], align 4
13835 // CHECK19-NEXT: [[TMP73:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS19]], i32 0, i32 3
13836 // CHECK19-NEXT: store ptr [[TMP69]], ptr [[TMP73]], align 4
13837 // CHECK19-NEXT: [[TMP74:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS19]], i32 0, i32 4
13838 // CHECK19-NEXT: store ptr @.offload_sizes.15, ptr [[TMP74]], align 4
13839 // CHECK19-NEXT: [[TMP75:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS19]], i32 0, i32 5
13840 // CHECK19-NEXT: store ptr @.offload_maptypes.16, ptr [[TMP75]], align 4
13841 // CHECK19-NEXT: [[TMP76:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS19]], i32 0, i32 6
13842 // CHECK19-NEXT: store ptr null, ptr [[TMP76]], align 4
13843 // CHECK19-NEXT: [[TMP77:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS19]], i32 0, i32 7
13844 // CHECK19-NEXT: store ptr null, ptr [[TMP77]], align 4
13845 // CHECK19-NEXT: [[TMP78:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS19]], i32 0, i32 8
13846 // CHECK19-NEXT: store i64 10, ptr [[TMP78]], align 8
13847 // CHECK19-NEXT: [[TMP79:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS19]], i32 0, i32 9
13848 // CHECK19-NEXT: store i64 0, ptr [[TMP79]], align 8
13849 // CHECK19-NEXT: [[TMP80:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS19]], i32 0, i32 10
13850 // CHECK19-NEXT: store [3 x i32] zeroinitializer, ptr [[TMP80]], align 4
13851 // CHECK19-NEXT: [[TMP81:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS19]], i32 0, i32 11
13852 // CHECK19-NEXT: store [3 x i32] zeroinitializer, ptr [[TMP81]], align 4
13853 // CHECK19-NEXT: [[TMP82:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS19]], i32 0, i32 12
13854 // CHECK19-NEXT: store i32 0, ptr [[TMP82]], align 4
13855 // CHECK19-NEXT: [[TMP83:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB3]], i64 -1, i32 0, i32 0, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l131.region_id, ptr [[KERNEL_ARGS19]])
13856 // CHECK19-NEXT: [[TMP84:%.*]] = icmp ne i32 [[TMP83]], 0
13857 // CHECK19-NEXT: br i1 [[TMP84]], label [[OMP_OFFLOAD_FAILED20:%.*]], label [[OMP_OFFLOAD_CONT21:%.*]]
13858 // CHECK19: omp_offload.failed20:
13859 // CHECK19-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l131(ptr [[A]]) #[[ATTR3]]
13860 // CHECK19-NEXT: br label [[OMP_OFFLOAD_CONT21]]
13861 // CHECK19: omp_offload.cont21:
13862 // CHECK19-NEXT: [[TMP85:%.*]] = load i32, ptr [[M]], align 4
13863 // CHECK19-NEXT: store i32 [[TMP85]], ptr [[M_CASTED22]], align 4
13864 // CHECK19-NEXT: [[TMP86:%.*]] = load i32, ptr [[M_CASTED22]], align 4
13865 // CHECK19-NEXT: [[TMP87:%.*]] = getelementptr inbounds [2 x ptr], ptr [[DOTOFFLOAD_BASEPTRS23]], i32 0, i32 0
13866 // CHECK19-NEXT: store i32 [[TMP86]], ptr [[TMP87]], align 4
13867 // CHECK19-NEXT: [[TMP88:%.*]] = getelementptr inbounds [2 x ptr], ptr [[DOTOFFLOAD_PTRS24]], i32 0, i32 0
13868 // CHECK19-NEXT: store i32 [[TMP86]], ptr [[TMP88]], align 4
13869 // CHECK19-NEXT: [[TMP89:%.*]] = getelementptr inbounds [2 x ptr], ptr [[DOTOFFLOAD_MAPPERS25]], i32 0, i32 0
13870 // CHECK19-NEXT: store ptr null, ptr [[TMP89]], align 4
13871 // CHECK19-NEXT: [[TMP90:%.*]] = getelementptr inbounds [2 x ptr], ptr [[DOTOFFLOAD_BASEPTRS23]], i32 0, i32 1
13872 // CHECK19-NEXT: store ptr [[A]], ptr [[TMP90]], align 4
13873 // CHECK19-NEXT: [[TMP91:%.*]] = getelementptr inbounds [2 x ptr], ptr [[DOTOFFLOAD_PTRS24]], i32 0, i32 1
13874 // CHECK19-NEXT: store ptr [[A]], ptr [[TMP91]], align 4
13875 // CHECK19-NEXT: [[TMP92:%.*]] = getelementptr inbounds [2 x ptr], ptr [[DOTOFFLOAD_MAPPERS25]], i32 0, i32 1
13876 // CHECK19-NEXT: store ptr null, ptr [[TMP92]], align 4
13877 // CHECK19-NEXT: [[TMP93:%.*]] = getelementptr inbounds [2 x ptr], ptr [[DOTOFFLOAD_BASEPTRS23]], i32 0, i32 0
13878 // CHECK19-NEXT: [[TMP94:%.*]] = getelementptr inbounds [2 x ptr], ptr [[DOTOFFLOAD_PTRS24]], i32 0, i32 0
13879 // CHECK19-NEXT: [[TMP95:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS27]], i32 0, i32 0
13880 // CHECK19-NEXT: store i32 3, ptr [[TMP95]], align 4
13881 // CHECK19-NEXT: [[TMP96:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS27]], i32 0, i32 1
13882 // CHECK19-NEXT: store i32 2, ptr [[TMP96]], align 4
13883 // CHECK19-NEXT: [[TMP97:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS27]], i32 0, i32 2
13884 // CHECK19-NEXT: store ptr [[TMP93]], ptr [[TMP97]], align 4
13885 // CHECK19-NEXT: [[TMP98:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS27]], i32 0, i32 3
13886 // CHECK19-NEXT: store ptr [[TMP94]], ptr [[TMP98]], align 4
13887 // CHECK19-NEXT: [[TMP99:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS27]], i32 0, i32 4
13888 // CHECK19-NEXT: store ptr @.offload_sizes.17, ptr [[TMP99]], align 4
13889 // CHECK19-NEXT: [[TMP100:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS27]], i32 0, i32 5
13890 // CHECK19-NEXT: store ptr @.offload_maptypes.18, ptr [[TMP100]], align 4
13891 // CHECK19-NEXT: [[TMP101:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS27]], i32 0, i32 6
13892 // CHECK19-NEXT: store ptr null, ptr [[TMP101]], align 4
13893 // CHECK19-NEXT: [[TMP102:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS27]], i32 0, i32 7
13894 // CHECK19-NEXT: store ptr null, ptr [[TMP102]], align 4
13895 // CHECK19-NEXT: [[TMP103:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS27]], i32 0, i32 8
13896 // CHECK19-NEXT: store i64 10, ptr [[TMP103]], align 8
13897 // CHECK19-NEXT: [[TMP104:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS27]], i32 0, i32 9
13898 // CHECK19-NEXT: store i64 0, ptr [[TMP104]], align 8
13899 // CHECK19-NEXT: [[TMP105:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS27]], i32 0, i32 10
13900 // CHECK19-NEXT: store [3 x i32] zeroinitializer, ptr [[TMP105]], align 4
13901 // CHECK19-NEXT: [[TMP106:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS27]], i32 0, i32 11
13902 // CHECK19-NEXT: store [3 x i32] zeroinitializer, ptr [[TMP106]], align 4
13903 // CHECK19-NEXT: [[TMP107:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS27]], i32 0, i32 12
13904 // CHECK19-NEXT: store i32 0, ptr [[TMP107]], align 4
13905 // CHECK19-NEXT: [[TMP108:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB3]], i64 -1, i32 0, i32 0, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l136.region_id, ptr [[KERNEL_ARGS27]])
13906 // CHECK19-NEXT: [[TMP109:%.*]] = icmp ne i32 [[TMP108]], 0
13907 // CHECK19-NEXT: br i1 [[TMP109]], label [[OMP_OFFLOAD_FAILED28:%.*]], label [[OMP_OFFLOAD_CONT29:%.*]]
13908 // CHECK19: omp_offload.failed28:
13909 // CHECK19-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l136(i32 [[TMP86]], ptr [[A]]) #[[ATTR3]]
13910 // CHECK19-NEXT: br label [[OMP_OFFLOAD_CONT29]]
13911 // CHECK19: omp_offload.cont29:
13912 // CHECK19-NEXT: ret i32 0
13915 // CHECK19-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l116
13916 // CHECK19-SAME: (ptr noundef nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] {
13917 // CHECK19-NEXT: entry:
13918 // CHECK19-NEXT: [[A_ADDR:%.*]] = alloca ptr, align 4
13919 // CHECK19-NEXT: store ptr [[A]], ptr [[A_ADDR]], align 4
13920 // CHECK19-NEXT: [[TMP0:%.*]] = load ptr, ptr [[A_ADDR]], align 4
13921 // CHECK19-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_teams(ptr @[[GLOB3]], i32 1, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l116.omp_outlined, ptr [[TMP0]])
13922 // CHECK19-NEXT: ret void
13925 // CHECK19-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l116.omp_outlined
13926 // CHECK19-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], ptr noundef nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] {
13927 // CHECK19-NEXT: entry:
13928 // CHECK19-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 4
13929 // CHECK19-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 4
13930 // CHECK19-NEXT: [[A_ADDR:%.*]] = alloca ptr, align 4
13931 // CHECK19-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
13932 // CHECK19-NEXT: [[TMP:%.*]] = alloca i32, align 4
13933 // CHECK19-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4
13934 // CHECK19-NEXT: [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4
13935 // CHECK19-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
13936 // CHECK19-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
13937 // CHECK19-NEXT: [[I:%.*]] = alloca i32, align 4
13938 // CHECK19-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 4
13939 // CHECK19-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 4
13940 // CHECK19-NEXT: store ptr [[A]], ptr [[A_ADDR]], align 4
13941 // CHECK19-NEXT: [[TMP0:%.*]] = load ptr, ptr [[A_ADDR]], align 4
13942 // CHECK19-NEXT: store i32 0, ptr [[DOTOMP_COMB_LB]], align 4
13943 // CHECK19-NEXT: store i32 9, ptr [[DOTOMP_COMB_UB]], align 4
13944 // CHECK19-NEXT: store i32 1, ptr [[DOTOMP_STRIDE]], align 4
13945 // CHECK19-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
13946 // CHECK19-NEXT: [[TMP1:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
13947 // CHECK19-NEXT: [[TMP2:%.*]] = load i32, ptr [[TMP1]], align 4
13948 // CHECK19-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB1]], i32 [[TMP2]], i32 92, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
13949 // CHECK19-NEXT: [[TMP3:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
13950 // CHECK19-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP3]], 9
13951 // CHECK19-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
13952 // CHECK19: cond.true:
13953 // CHECK19-NEXT: br label [[COND_END:%.*]]
13954 // CHECK19: cond.false:
13955 // CHECK19-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
13956 // CHECK19-NEXT: br label [[COND_END]]
13957 // CHECK19: cond.end:
13958 // CHECK19-NEXT: [[COND:%.*]] = phi i32 [ 9, [[COND_TRUE]] ], [ [[TMP4]], [[COND_FALSE]] ]
13959 // CHECK19-NEXT: store i32 [[COND]], ptr [[DOTOMP_COMB_UB]], align 4
13960 // CHECK19-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
13961 // CHECK19-NEXT: store i32 [[TMP5]], ptr [[DOTOMP_IV]], align 4
13962 // CHECK19-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
13963 // CHECK19: omp.inner.for.cond:
13964 // CHECK19-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
13965 // CHECK19-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
13966 // CHECK19-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP6]], [[TMP7]]
13967 // CHECK19-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
13968 // CHECK19: omp.inner.for.body:
13969 // CHECK19-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
13970 // CHECK19-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
13971 // CHECK19-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_call(ptr @[[GLOB3]], i32 3, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l116.omp_outlined.omp_outlined, i32 [[TMP8]], i32 [[TMP9]], ptr [[TMP0]])
13972 // CHECK19-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
13973 // CHECK19: omp.inner.for.inc:
13974 // CHECK19-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
13975 // CHECK19-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4
13976 // CHECK19-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP10]], [[TMP11]]
13977 // CHECK19-NEXT: store i32 [[ADD]], ptr [[DOTOMP_IV]], align 4
13978 // CHECK19-NEXT: br label [[OMP_INNER_FOR_COND]]
13979 // CHECK19: omp.inner.for.end:
13980 // CHECK19-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
13981 // CHECK19: omp.loop.exit:
13982 // CHECK19-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB1]], i32 [[TMP2]])
13983 // CHECK19-NEXT: ret void
13986 // CHECK19-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l116.omp_outlined.omp_outlined
13987 // CHECK19-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], i32 noundef [[DOTPREVIOUS_LB_:%.*]], i32 noundef [[DOTPREVIOUS_UB_:%.*]], ptr noundef nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] {
13988 // CHECK19-NEXT: entry:
13989 // CHECK19-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 4
13990 // CHECK19-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 4
13991 // CHECK19-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i32, align 4
13992 // CHECK19-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i32, align 4
13993 // CHECK19-NEXT: [[A_ADDR:%.*]] = alloca ptr, align 4
13994 // CHECK19-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
13995 // CHECK19-NEXT: [[TMP:%.*]] = alloca i32, align 4
13996 // CHECK19-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4
13997 // CHECK19-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4
13998 // CHECK19-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
13999 // CHECK19-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
14000 // CHECK19-NEXT: [[I:%.*]] = alloca i32, align 4
14001 // CHECK19-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 4
14002 // CHECK19-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 4
14003 // CHECK19-NEXT: store i32 [[DOTPREVIOUS_LB_]], ptr [[DOTPREVIOUS_LB__ADDR]], align 4
14004 // CHECK19-NEXT: store i32 [[DOTPREVIOUS_UB_]], ptr [[DOTPREVIOUS_UB__ADDR]], align 4
14005 // CHECK19-NEXT: store ptr [[A]], ptr [[A_ADDR]], align 4
14006 // CHECK19-NEXT: [[TMP0:%.*]] = load ptr, ptr [[A_ADDR]], align 4
14007 // CHECK19-NEXT: store i32 0, ptr [[DOTOMP_LB]], align 4
14008 // CHECK19-NEXT: store i32 9, ptr [[DOTOMP_UB]], align 4
14009 // CHECK19-NEXT: [[TMP1:%.*]] = load i32, ptr [[DOTPREVIOUS_LB__ADDR]], align 4
14010 // CHECK19-NEXT: [[TMP2:%.*]] = load i32, ptr [[DOTPREVIOUS_UB__ADDR]], align 4
14011 // CHECK19-NEXT: store i32 [[TMP1]], ptr [[DOTOMP_LB]], align 4
14012 // CHECK19-NEXT: store i32 [[TMP2]], ptr [[DOTOMP_UB]], align 4
14013 // CHECK19-NEXT: store i32 1, ptr [[DOTOMP_STRIDE]], align 4
14014 // CHECK19-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
14015 // CHECK19-NEXT: [[TMP3:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
14016 // CHECK19-NEXT: [[TMP4:%.*]] = load i32, ptr [[TMP3]], align 4
14017 // CHECK19-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB2]], i32 [[TMP4]], i32 34, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
14018 // CHECK19-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
14019 // CHECK19-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP5]], 9
14020 // CHECK19-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
14021 // CHECK19: cond.true:
14022 // CHECK19-NEXT: br label [[COND_END:%.*]]
14023 // CHECK19: cond.false:
14024 // CHECK19-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
14025 // CHECK19-NEXT: br label [[COND_END]]
14026 // CHECK19: cond.end:
14027 // CHECK19-NEXT: [[COND:%.*]] = phi i32 [ 9, [[COND_TRUE]] ], [ [[TMP6]], [[COND_FALSE]] ]
14028 // CHECK19-NEXT: store i32 [[COND]], ptr [[DOTOMP_UB]], align 4
14029 // CHECK19-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4
14030 // CHECK19-NEXT: store i32 [[TMP7]], ptr [[DOTOMP_IV]], align 4
14031 // CHECK19-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
14032 // CHECK19: omp.inner.for.cond:
14033 // CHECK19-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
14034 // CHECK19-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
14035 // CHECK19-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP8]], [[TMP9]]
14036 // CHECK19-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
14037 // CHECK19: omp.inner.for.body:
14038 // CHECK19-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
14039 // CHECK19-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP10]], 1
14040 // CHECK19-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
14041 // CHECK19-NEXT: store i32 [[ADD]], ptr [[I]], align 4
14042 // CHECK19-NEXT: [[TMP11:%.*]] = load i32, ptr [[I]], align 4
14043 // CHECK19-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds [10 x i32], ptr [[TMP0]], i32 0, i32 [[TMP11]]
14044 // CHECK19-NEXT: store i32 0, ptr [[ARRAYIDX]], align 4
14045 // CHECK19-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
14046 // CHECK19: omp.body.continue:
14047 // CHECK19-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
14048 // CHECK19: omp.inner.for.inc:
14049 // CHECK19-NEXT: [[TMP12:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
14050 // CHECK19-NEXT: [[ADD2:%.*]] = add nsw i32 [[TMP12]], 1
14051 // CHECK19-NEXT: store i32 [[ADD2]], ptr [[DOTOMP_IV]], align 4
14052 // CHECK19-NEXT: br label [[OMP_INNER_FOR_COND]]
14053 // CHECK19: omp.inner.for.end:
14054 // CHECK19-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
14055 // CHECK19: omp.loop.exit:
14056 // CHECK19-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB2]], i32 [[TMP4]])
14057 // CHECK19-NEXT: ret void
14060 // CHECK19-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l121
14061 // CHECK19-SAME: (ptr noundef nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] {
14062 // CHECK19-NEXT: entry:
14063 // CHECK19-NEXT: [[A_ADDR:%.*]] = alloca ptr, align 4
14064 // CHECK19-NEXT: store ptr [[A]], ptr [[A_ADDR]], align 4
14065 // CHECK19-NEXT: [[TMP0:%.*]] = load ptr, ptr [[A_ADDR]], align 4
14066 // CHECK19-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_teams(ptr @[[GLOB3]], i32 1, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l121.omp_outlined, ptr [[TMP0]])
14067 // CHECK19-NEXT: ret void
14070 // CHECK19-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l121.omp_outlined
14071 // CHECK19-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], ptr noundef nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] {
14072 // CHECK19-NEXT: entry:
14073 // CHECK19-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 4
14074 // CHECK19-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 4
14075 // CHECK19-NEXT: [[A_ADDR:%.*]] = alloca ptr, align 4
14076 // CHECK19-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
14077 // CHECK19-NEXT: [[TMP:%.*]] = alloca i32, align 4
14078 // CHECK19-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4
14079 // CHECK19-NEXT: [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4
14080 // CHECK19-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
14081 // CHECK19-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
14082 // CHECK19-NEXT: [[I:%.*]] = alloca i32, align 4
14083 // CHECK19-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 4
14084 // CHECK19-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 4
14085 // CHECK19-NEXT: store ptr [[A]], ptr [[A_ADDR]], align 4
14086 // CHECK19-NEXT: [[TMP0:%.*]] = load ptr, ptr [[A_ADDR]], align 4
14087 // CHECK19-NEXT: store i32 0, ptr [[DOTOMP_COMB_LB]], align 4
14088 // CHECK19-NEXT: store i32 9, ptr [[DOTOMP_COMB_UB]], align 4
14089 // CHECK19-NEXT: store i32 1, ptr [[DOTOMP_STRIDE]], align 4
14090 // CHECK19-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
14091 // CHECK19-NEXT: [[TMP1:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
14092 // CHECK19-NEXT: [[TMP2:%.*]] = load i32, ptr [[TMP1]], align 4
14093 // CHECK19-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB1]], i32 [[TMP2]], i32 92, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
14094 // CHECK19-NEXT: [[TMP3:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
14095 // CHECK19-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP3]], 9
14096 // CHECK19-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
14097 // CHECK19: cond.true:
14098 // CHECK19-NEXT: br label [[COND_END:%.*]]
14099 // CHECK19: cond.false:
14100 // CHECK19-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
14101 // CHECK19-NEXT: br label [[COND_END]]
14102 // CHECK19: cond.end:
14103 // CHECK19-NEXT: [[COND:%.*]] = phi i32 [ 9, [[COND_TRUE]] ], [ [[TMP4]], [[COND_FALSE]] ]
14104 // CHECK19-NEXT: store i32 [[COND]], ptr [[DOTOMP_COMB_UB]], align 4
14105 // CHECK19-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
14106 // CHECK19-NEXT: store i32 [[TMP5]], ptr [[DOTOMP_IV]], align 4
14107 // CHECK19-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
14108 // CHECK19: omp.inner.for.cond:
14109 // CHECK19-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
14110 // CHECK19-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
14111 // CHECK19-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP6]], [[TMP7]]
14112 // CHECK19-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
14113 // CHECK19: omp.inner.for.body:
14114 // CHECK19-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
14115 // CHECK19-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
14116 // CHECK19-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_call(ptr @[[GLOB3]], i32 3, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l121.omp_outlined.omp_outlined, i32 [[TMP8]], i32 [[TMP9]], ptr [[TMP0]])
14117 // CHECK19-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
14118 // CHECK19: omp.inner.for.inc:
14119 // CHECK19-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
14120 // CHECK19-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4
14121 // CHECK19-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP10]], [[TMP11]]
14122 // CHECK19-NEXT: store i32 [[ADD]], ptr [[DOTOMP_IV]], align 4
14123 // CHECK19-NEXT: br label [[OMP_INNER_FOR_COND]]
14124 // CHECK19: omp.inner.for.end:
14125 // CHECK19-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
14126 // CHECK19: omp.loop.exit:
14127 // CHECK19-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB1]], i32 [[TMP2]])
14128 // CHECK19-NEXT: ret void
14131 // CHECK19-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l121.omp_outlined.omp_outlined
14132 // CHECK19-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], i32 noundef [[DOTPREVIOUS_LB_:%.*]], i32 noundef [[DOTPREVIOUS_UB_:%.*]], ptr noundef nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] {
14133 // CHECK19-NEXT: entry:
14134 // CHECK19-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 4
14135 // CHECK19-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 4
14136 // CHECK19-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i32, align 4
14137 // CHECK19-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i32, align 4
14138 // CHECK19-NEXT: [[A_ADDR:%.*]] = alloca ptr, align 4
14139 // CHECK19-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
14140 // CHECK19-NEXT: [[TMP:%.*]] = alloca i32, align 4
14141 // CHECK19-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4
14142 // CHECK19-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4
14143 // CHECK19-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
14144 // CHECK19-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
14145 // CHECK19-NEXT: [[I:%.*]] = alloca i32, align 4
14146 // CHECK19-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 4
14147 // CHECK19-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 4
14148 // CHECK19-NEXT: store i32 [[DOTPREVIOUS_LB_]], ptr [[DOTPREVIOUS_LB__ADDR]], align 4
14149 // CHECK19-NEXT: store i32 [[DOTPREVIOUS_UB_]], ptr [[DOTPREVIOUS_UB__ADDR]], align 4
14150 // CHECK19-NEXT: store ptr [[A]], ptr [[A_ADDR]], align 4
14151 // CHECK19-NEXT: [[TMP0:%.*]] = load ptr, ptr [[A_ADDR]], align 4
14152 // CHECK19-NEXT: store i32 0, ptr [[DOTOMP_LB]], align 4
14153 // CHECK19-NEXT: store i32 9, ptr [[DOTOMP_UB]], align 4
14154 // CHECK19-NEXT: [[TMP1:%.*]] = load i32, ptr [[DOTPREVIOUS_LB__ADDR]], align 4
14155 // CHECK19-NEXT: [[TMP2:%.*]] = load i32, ptr [[DOTPREVIOUS_UB__ADDR]], align 4
14156 // CHECK19-NEXT: store i32 [[TMP1]], ptr [[DOTOMP_LB]], align 4
14157 // CHECK19-NEXT: store i32 [[TMP2]], ptr [[DOTOMP_UB]], align 4
14158 // CHECK19-NEXT: store i32 1, ptr [[DOTOMP_STRIDE]], align 4
14159 // CHECK19-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
14160 // CHECK19-NEXT: [[TMP3:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
14161 // CHECK19-NEXT: [[TMP4:%.*]] = load i32, ptr [[TMP3]], align 4
14162 // CHECK19-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB2]], i32 [[TMP4]], i32 34, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
14163 // CHECK19-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
14164 // CHECK19-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP5]], 9
14165 // CHECK19-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
14166 // CHECK19: cond.true:
14167 // CHECK19-NEXT: br label [[COND_END:%.*]]
14168 // CHECK19: cond.false:
14169 // CHECK19-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
14170 // CHECK19-NEXT: br label [[COND_END]]
14171 // CHECK19: cond.end:
14172 // CHECK19-NEXT: [[COND:%.*]] = phi i32 [ 9, [[COND_TRUE]] ], [ [[TMP6]], [[COND_FALSE]] ]
14173 // CHECK19-NEXT: store i32 [[COND]], ptr [[DOTOMP_UB]], align 4
14174 // CHECK19-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4
14175 // CHECK19-NEXT: store i32 [[TMP7]], ptr [[DOTOMP_IV]], align 4
14176 // CHECK19-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
14177 // CHECK19: omp.inner.for.cond:
14178 // CHECK19-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
14179 // CHECK19-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
14180 // CHECK19-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP8]], [[TMP9]]
14181 // CHECK19-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
14182 // CHECK19: omp.inner.for.body:
14183 // CHECK19-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
14184 // CHECK19-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP10]], 1
14185 // CHECK19-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
14186 // CHECK19-NEXT: store i32 [[ADD]], ptr [[I]], align 4
14187 // CHECK19-NEXT: [[TMP11:%.*]] = load i32, ptr [[I]], align 4
14188 // CHECK19-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds [10 x i32], ptr [[TMP0]], i32 0, i32 [[TMP11]]
14189 // CHECK19-NEXT: store i32 0, ptr [[ARRAYIDX]], align 4
14190 // CHECK19-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
14191 // CHECK19: omp.body.continue:
14192 // CHECK19-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
14193 // CHECK19: omp.inner.for.inc:
14194 // CHECK19-NEXT: [[TMP12:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
14195 // CHECK19-NEXT: [[ADD2:%.*]] = add nsw i32 [[TMP12]], 1
14196 // CHECK19-NEXT: store i32 [[ADD2]], ptr [[DOTOMP_IV]], align 4
14197 // CHECK19-NEXT: br label [[OMP_INNER_FOR_COND]]
14198 // CHECK19: omp.inner.for.end:
14199 // CHECK19-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
14200 // CHECK19: omp.loop.exit:
14201 // CHECK19-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB2]], i32 [[TMP4]])
14202 // CHECK19-NEXT: ret void
14205 // CHECK19-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l126
14206 // CHECK19-SAME: (i32 noundef [[M:%.*]], ptr noundef nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] {
14207 // CHECK19-NEXT: entry:
14208 // CHECK19-NEXT: [[M_ADDR:%.*]] = alloca i32, align 4
14209 // CHECK19-NEXT: [[A_ADDR:%.*]] = alloca ptr, align 4
14210 // CHECK19-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4
14211 // CHECK19-NEXT: [[DOTCAPTURE_EXPR__CASTED:%.*]] = alloca i32, align 4
14212 // CHECK19-NEXT: store i32 [[M]], ptr [[M_ADDR]], align 4
14213 // CHECK19-NEXT: store ptr [[A]], ptr [[A_ADDR]], align 4
14214 // CHECK19-NEXT: [[TMP0:%.*]] = load ptr, ptr [[A_ADDR]], align 4
14215 // CHECK19-NEXT: [[TMP1:%.*]] = load i32, ptr [[M_ADDR]], align 4
14216 // CHECK19-NEXT: store i32 [[TMP1]], ptr [[DOTCAPTURE_EXPR_]], align 4
14217 // CHECK19-NEXT: [[TMP2:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4
14218 // CHECK19-NEXT: store i32 [[TMP2]], ptr [[DOTCAPTURE_EXPR__CASTED]], align 4
14219 // CHECK19-NEXT: [[TMP3:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR__CASTED]], align 4
14220 // CHECK19-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_teams(ptr @[[GLOB3]], i32 2, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l126.omp_outlined, ptr [[TMP0]], i32 [[TMP3]])
14221 // CHECK19-NEXT: ret void
14224 // CHECK19-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l126.omp_outlined
14225 // CHECK19-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], ptr noundef nonnull align 4 dereferenceable(40) [[A:%.*]], i32 noundef [[DOTCAPTURE_EXPR_:%.*]]) #[[ATTR2]] {
14226 // CHECK19-NEXT: entry:
14227 // CHECK19-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 4
14228 // CHECK19-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 4
14229 // CHECK19-NEXT: [[A_ADDR:%.*]] = alloca ptr, align 4
14230 // CHECK19-NEXT: [[DOTCAPTURE_EXPR__ADDR:%.*]] = alloca i32, align 4
14231 // CHECK19-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
14232 // CHECK19-NEXT: [[TMP:%.*]] = alloca i32, align 4
14233 // CHECK19-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4
14234 // CHECK19-NEXT: [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4
14235 // CHECK19-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
14236 // CHECK19-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
14237 // CHECK19-NEXT: [[I:%.*]] = alloca i32, align 4
14238 // CHECK19-NEXT: [[DOTCAPTURE_EXPR__CASTED:%.*]] = alloca i32, align 4
14239 // CHECK19-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 4
14240 // CHECK19-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 4
14241 // CHECK19-NEXT: store ptr [[A]], ptr [[A_ADDR]], align 4
14242 // CHECK19-NEXT: store i32 [[DOTCAPTURE_EXPR_]], ptr [[DOTCAPTURE_EXPR__ADDR]], align 4
14243 // CHECK19-NEXT: [[TMP0:%.*]] = load ptr, ptr [[A_ADDR]], align 4
14244 // CHECK19-NEXT: store i32 0, ptr [[DOTOMP_COMB_LB]], align 4
14245 // CHECK19-NEXT: store i32 9, ptr [[DOTOMP_COMB_UB]], align 4
14246 // CHECK19-NEXT: store i32 1, ptr [[DOTOMP_STRIDE]], align 4
14247 // CHECK19-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
14248 // CHECK19-NEXT: [[TMP1:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
14249 // CHECK19-NEXT: [[TMP2:%.*]] = load i32, ptr [[TMP1]], align 4
14250 // CHECK19-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB1]], i32 [[TMP2]], i32 92, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
14251 // CHECK19-NEXT: [[TMP3:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
14252 // CHECK19-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP3]], 9
14253 // CHECK19-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
14254 // CHECK19: cond.true:
14255 // CHECK19-NEXT: br label [[COND_END:%.*]]
14256 // CHECK19: cond.false:
14257 // CHECK19-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
14258 // CHECK19-NEXT: br label [[COND_END]]
14259 // CHECK19: cond.end:
14260 // CHECK19-NEXT: [[COND:%.*]] = phi i32 [ 9, [[COND_TRUE]] ], [ [[TMP4]], [[COND_FALSE]] ]
14261 // CHECK19-NEXT: store i32 [[COND]], ptr [[DOTOMP_COMB_UB]], align 4
14262 // CHECK19-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
14263 // CHECK19-NEXT: store i32 [[TMP5]], ptr [[DOTOMP_IV]], align 4
14264 // CHECK19-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
14265 // CHECK19: omp.inner.for.cond:
14266 // CHECK19-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
14267 // CHECK19-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
14268 // CHECK19-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP6]], [[TMP7]]
14269 // CHECK19-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
14270 // CHECK19: omp.inner.for.body:
14271 // CHECK19-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
14272 // CHECK19-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
14273 // CHECK19-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR__ADDR]], align 4
14274 // CHECK19-NEXT: store i32 [[TMP10]], ptr [[DOTCAPTURE_EXPR__CASTED]], align 4
14275 // CHECK19-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR__CASTED]], align 4
14276 // CHECK19-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_call(ptr @[[GLOB3]], i32 4, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l126.omp_outlined.omp_outlined, i32 [[TMP8]], i32 [[TMP9]], ptr [[TMP0]], i32 [[TMP11]])
14277 // CHECK19-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
14278 // CHECK19: omp.inner.for.inc:
14279 // CHECK19-NEXT: [[TMP12:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
14280 // CHECK19-NEXT: [[TMP13:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4
14281 // CHECK19-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP12]], [[TMP13]]
14282 // CHECK19-NEXT: store i32 [[ADD]], ptr [[DOTOMP_IV]], align 4
14283 // CHECK19-NEXT: br label [[OMP_INNER_FOR_COND]]
14284 // CHECK19: omp.inner.for.end:
14285 // CHECK19-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
14286 // CHECK19: omp.loop.exit:
14287 // CHECK19-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB1]], i32 [[TMP2]])
14288 // CHECK19-NEXT: ret void
14291 // CHECK19-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l126.omp_outlined.omp_outlined
14292 // CHECK19-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], i32 noundef [[DOTPREVIOUS_LB_:%.*]], i32 noundef [[DOTPREVIOUS_UB_:%.*]], ptr noundef nonnull align 4 dereferenceable(40) [[A:%.*]], i32 noundef [[DOTCAPTURE_EXPR_:%.*]]) #[[ATTR2]] {
14293 // CHECK19-NEXT: entry:
14294 // CHECK19-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 4
14295 // CHECK19-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 4
14296 // CHECK19-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i32, align 4
14297 // CHECK19-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i32, align 4
14298 // CHECK19-NEXT: [[A_ADDR:%.*]] = alloca ptr, align 4
14299 // CHECK19-NEXT: [[DOTCAPTURE_EXPR__ADDR:%.*]] = alloca i32, align 4
14300 // CHECK19-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
14301 // CHECK19-NEXT: [[TMP:%.*]] = alloca i32, align 4
14302 // CHECK19-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4
14303 // CHECK19-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4
14304 // CHECK19-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
14305 // CHECK19-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
14306 // CHECK19-NEXT: [[I:%.*]] = alloca i32, align 4
14307 // CHECK19-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 4
14308 // CHECK19-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 4
14309 // CHECK19-NEXT: store i32 [[DOTPREVIOUS_LB_]], ptr [[DOTPREVIOUS_LB__ADDR]], align 4
14310 // CHECK19-NEXT: store i32 [[DOTPREVIOUS_UB_]], ptr [[DOTPREVIOUS_UB__ADDR]], align 4
14311 // CHECK19-NEXT: store ptr [[A]], ptr [[A_ADDR]], align 4
14312 // CHECK19-NEXT: store i32 [[DOTCAPTURE_EXPR_]], ptr [[DOTCAPTURE_EXPR__ADDR]], align 4
14313 // CHECK19-NEXT: [[TMP0:%.*]] = load ptr, ptr [[A_ADDR]], align 4
14314 // CHECK19-NEXT: store i32 0, ptr [[DOTOMP_LB]], align 4
14315 // CHECK19-NEXT: store i32 9, ptr [[DOTOMP_UB]], align 4
14316 // CHECK19-NEXT: [[TMP1:%.*]] = load i32, ptr [[DOTPREVIOUS_LB__ADDR]], align 4
14317 // CHECK19-NEXT: [[TMP2:%.*]] = load i32, ptr [[DOTPREVIOUS_UB__ADDR]], align 4
14318 // CHECK19-NEXT: store i32 [[TMP1]], ptr [[DOTOMP_LB]], align 4
14319 // CHECK19-NEXT: store i32 [[TMP2]], ptr [[DOTOMP_UB]], align 4
14320 // CHECK19-NEXT: store i32 1, ptr [[DOTOMP_STRIDE]], align 4
14321 // CHECK19-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
14322 // CHECK19-NEXT: [[TMP3:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR__ADDR]], align 4
14323 // CHECK19-NEXT: [[TMP4:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
14324 // CHECK19-NEXT: [[TMP5:%.*]] = load i32, ptr [[TMP4]], align 4
14325 // CHECK19-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB2]], i32 [[TMP5]], i32 33, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 [[TMP3]])
14326 // CHECK19-NEXT: br label [[OMP_DISPATCH_COND:%.*]]
14327 // CHECK19: omp.dispatch.cond:
14328 // CHECK19-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
14329 // CHECK19-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTPREVIOUS_UB__ADDR]], align 4
14330 // CHECK19-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP6]], [[TMP7]]
14331 // CHECK19-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
14332 // CHECK19: cond.true:
14333 // CHECK19-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTPREVIOUS_UB__ADDR]], align 4
14334 // CHECK19-NEXT: br label [[COND_END:%.*]]
14335 // CHECK19: cond.false:
14336 // CHECK19-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
14337 // CHECK19-NEXT: br label [[COND_END]]
14338 // CHECK19: cond.end:
14339 // CHECK19-NEXT: [[COND:%.*]] = phi i32 [ [[TMP8]], [[COND_TRUE]] ], [ [[TMP9]], [[COND_FALSE]] ]
14340 // CHECK19-NEXT: store i32 [[COND]], ptr [[DOTOMP_UB]], align 4
14341 // CHECK19-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4
14342 // CHECK19-NEXT: store i32 [[TMP10]], ptr [[DOTOMP_IV]], align 4
14343 // CHECK19-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
14344 // CHECK19-NEXT: [[TMP12:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
14345 // CHECK19-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP11]], [[TMP12]]
14346 // CHECK19-NEXT: br i1 [[CMP1]], label [[OMP_DISPATCH_BODY:%.*]], label [[OMP_DISPATCH_END:%.*]]
14347 // CHECK19: omp.dispatch.body:
14348 // CHECK19-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
14349 // CHECK19: omp.inner.for.cond:
14350 // CHECK19-NEXT: [[TMP13:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
14351 // CHECK19-NEXT: [[TMP14:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
14352 // CHECK19-NEXT: [[CMP2:%.*]] = icmp sle i32 [[TMP13]], [[TMP14]]
14353 // CHECK19-NEXT: br i1 [[CMP2]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
14354 // CHECK19: omp.inner.for.body:
14355 // CHECK19-NEXT: [[TMP15:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
14356 // CHECK19-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP15]], 1
14357 // CHECK19-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
14358 // CHECK19-NEXT: store i32 [[ADD]], ptr [[I]], align 4
14359 // CHECK19-NEXT: [[TMP16:%.*]] = load i32, ptr [[I]], align 4
14360 // CHECK19-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds [10 x i32], ptr [[TMP0]], i32 0, i32 [[TMP16]]
14361 // CHECK19-NEXT: store i32 0, ptr [[ARRAYIDX]], align 4
14362 // CHECK19-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
14363 // CHECK19: omp.body.continue:
14364 // CHECK19-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
14365 // CHECK19: omp.inner.for.inc:
14366 // CHECK19-NEXT: [[TMP17:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
14367 // CHECK19-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP17]], 1
14368 // CHECK19-NEXT: store i32 [[ADD3]], ptr [[DOTOMP_IV]], align 4
14369 // CHECK19-NEXT: br label [[OMP_INNER_FOR_COND]]
14370 // CHECK19: omp.inner.for.end:
14371 // CHECK19-NEXT: br label [[OMP_DISPATCH_INC:%.*]]
14372 // CHECK19: omp.dispatch.inc:
14373 // CHECK19-NEXT: [[TMP18:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4
14374 // CHECK19-NEXT: [[TMP19:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4
14375 // CHECK19-NEXT: [[ADD4:%.*]] = add nsw i32 [[TMP18]], [[TMP19]]
14376 // CHECK19-NEXT: store i32 [[ADD4]], ptr [[DOTOMP_LB]], align 4
14377 // CHECK19-NEXT: [[TMP20:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
14378 // CHECK19-NEXT: [[TMP21:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4
14379 // CHECK19-NEXT: [[ADD5:%.*]] = add nsw i32 [[TMP20]], [[TMP21]]
14380 // CHECK19-NEXT: store i32 [[ADD5]], ptr [[DOTOMP_UB]], align 4
14381 // CHECK19-NEXT: br label [[OMP_DISPATCH_COND]]
14382 // CHECK19: omp.dispatch.end:
14383 // CHECK19-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB2]], i32 [[TMP5]])
14384 // CHECK19-NEXT: ret void
14387 // CHECK19-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l131
14388 // CHECK19-SAME: (ptr noundef nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] {
14389 // CHECK19-NEXT: entry:
14390 // CHECK19-NEXT: [[A_ADDR:%.*]] = alloca ptr, align 4
14391 // CHECK19-NEXT: store ptr [[A]], ptr [[A_ADDR]], align 4
14392 // CHECK19-NEXT: [[TMP0:%.*]] = load ptr, ptr [[A_ADDR]], align 4
14393 // CHECK19-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_teams(ptr @[[GLOB3]], i32 1, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l131.omp_outlined, ptr [[TMP0]])
14394 // CHECK19-NEXT: ret void
14397 // CHECK19-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l131.omp_outlined
14398 // CHECK19-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], ptr noundef nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] {
14399 // CHECK19-NEXT: entry:
14400 // CHECK19-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 4
14401 // CHECK19-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 4
14402 // CHECK19-NEXT: [[A_ADDR:%.*]] = alloca ptr, align 4
14403 // CHECK19-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
14404 // CHECK19-NEXT: [[TMP:%.*]] = alloca i32, align 4
14405 // CHECK19-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4
14406 // CHECK19-NEXT: [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4
14407 // CHECK19-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
14408 // CHECK19-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
14409 // CHECK19-NEXT: [[I:%.*]] = alloca i32, align 4
14410 // CHECK19-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 4
14411 // CHECK19-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 4
14412 // CHECK19-NEXT: store ptr [[A]], ptr [[A_ADDR]], align 4
14413 // CHECK19-NEXT: [[TMP0:%.*]] = load ptr, ptr [[A_ADDR]], align 4
14414 // CHECK19-NEXT: store i32 0, ptr [[DOTOMP_COMB_LB]], align 4
14415 // CHECK19-NEXT: store i32 9, ptr [[DOTOMP_COMB_UB]], align 4
14416 // CHECK19-NEXT: store i32 1, ptr [[DOTOMP_STRIDE]], align 4
14417 // CHECK19-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
14418 // CHECK19-NEXT: [[TMP1:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
14419 // CHECK19-NEXT: [[TMP2:%.*]] = load i32, ptr [[TMP1]], align 4
14420 // CHECK19-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB1]], i32 [[TMP2]], i32 92, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
14421 // CHECK19-NEXT: [[TMP3:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
14422 // CHECK19-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP3]], 9
14423 // CHECK19-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
14424 // CHECK19: cond.true:
14425 // CHECK19-NEXT: br label [[COND_END:%.*]]
14426 // CHECK19: cond.false:
14427 // CHECK19-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
14428 // CHECK19-NEXT: br label [[COND_END]]
14429 // CHECK19: cond.end:
14430 // CHECK19-NEXT: [[COND:%.*]] = phi i32 [ 9, [[COND_TRUE]] ], [ [[TMP4]], [[COND_FALSE]] ]
14431 // CHECK19-NEXT: store i32 [[COND]], ptr [[DOTOMP_COMB_UB]], align 4
14432 // CHECK19-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
14433 // CHECK19-NEXT: store i32 [[TMP5]], ptr [[DOTOMP_IV]], align 4
14434 // CHECK19-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
14435 // CHECK19: omp.inner.for.cond:
14436 // CHECK19-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
14437 // CHECK19-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
14438 // CHECK19-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP6]], [[TMP7]]
14439 // CHECK19-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
14440 // CHECK19: omp.inner.for.body:
14441 // CHECK19-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
14442 // CHECK19-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
14443 // CHECK19-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_call(ptr @[[GLOB3]], i32 3, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l131.omp_outlined.omp_outlined, i32 [[TMP8]], i32 [[TMP9]], ptr [[TMP0]])
14444 // CHECK19-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
14445 // CHECK19: omp.inner.for.inc:
14446 // CHECK19-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
14447 // CHECK19-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4
14448 // CHECK19-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP10]], [[TMP11]]
14449 // CHECK19-NEXT: store i32 [[ADD]], ptr [[DOTOMP_IV]], align 4
14450 // CHECK19-NEXT: br label [[OMP_INNER_FOR_COND]]
14451 // CHECK19: omp.inner.for.end:
14452 // CHECK19-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
14453 // CHECK19: omp.loop.exit:
14454 // CHECK19-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB1]], i32 [[TMP2]])
14455 // CHECK19-NEXT: ret void
14458 // CHECK19-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l131.omp_outlined.omp_outlined
14459 // CHECK19-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], i32 noundef [[DOTPREVIOUS_LB_:%.*]], i32 noundef [[DOTPREVIOUS_UB_:%.*]], ptr noundef nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] {
14460 // CHECK19-NEXT: entry:
14461 // CHECK19-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 4
14462 // CHECK19-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 4
14463 // CHECK19-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i32, align 4
14464 // CHECK19-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i32, align 4
14465 // CHECK19-NEXT: [[A_ADDR:%.*]] = alloca ptr, align 4
14466 // CHECK19-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
14467 // CHECK19-NEXT: [[TMP:%.*]] = alloca i32, align 4
14468 // CHECK19-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4
14469 // CHECK19-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4
14470 // CHECK19-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
14471 // CHECK19-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
14472 // CHECK19-NEXT: [[I:%.*]] = alloca i32, align 4
14473 // CHECK19-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 4
14474 // CHECK19-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 4
14475 // CHECK19-NEXT: store i32 [[DOTPREVIOUS_LB_]], ptr [[DOTPREVIOUS_LB__ADDR]], align 4
14476 // CHECK19-NEXT: store i32 [[DOTPREVIOUS_UB_]], ptr [[DOTPREVIOUS_UB__ADDR]], align 4
14477 // CHECK19-NEXT: store ptr [[A]], ptr [[A_ADDR]], align 4
14478 // CHECK19-NEXT: [[TMP0:%.*]] = load ptr, ptr [[A_ADDR]], align 4
14479 // CHECK19-NEXT: store i32 0, ptr [[DOTOMP_LB]], align 4
14480 // CHECK19-NEXT: store i32 9, ptr [[DOTOMP_UB]], align 4
14481 // CHECK19-NEXT: [[TMP1:%.*]] = load i32, ptr [[DOTPREVIOUS_LB__ADDR]], align 4
14482 // CHECK19-NEXT: [[TMP2:%.*]] = load i32, ptr [[DOTPREVIOUS_UB__ADDR]], align 4
14483 // CHECK19-NEXT: store i32 [[TMP1]], ptr [[DOTOMP_LB]], align 4
14484 // CHECK19-NEXT: store i32 [[TMP2]], ptr [[DOTOMP_UB]], align 4
14485 // CHECK19-NEXT: store i32 1, ptr [[DOTOMP_STRIDE]], align 4
14486 // CHECK19-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
14487 // CHECK19-NEXT: [[TMP3:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4
14488 // CHECK19-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
14489 // CHECK19-NEXT: [[TMP5:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
14490 // CHECK19-NEXT: [[TMP6:%.*]] = load i32, ptr [[TMP5]], align 4
14491 // CHECK19-NEXT: call void @__kmpc_dispatch_init_4(ptr @[[GLOB3]], i32 [[TMP6]], i32 35, i32 [[TMP3]], i32 [[TMP4]], i32 1, i32 1)
14492 // CHECK19-NEXT: br label [[OMP_DISPATCH_COND:%.*]]
14493 // CHECK19: omp.dispatch.cond:
14494 // CHECK19-NEXT: [[TMP7:%.*]] = call i32 @__kmpc_dispatch_next_4(ptr @[[GLOB3]], i32 [[TMP6]], ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]])
14495 // CHECK19-NEXT: [[TOBOOL:%.*]] = icmp ne i32 [[TMP7]], 0
14496 // CHECK19-NEXT: br i1 [[TOBOOL]], label [[OMP_DISPATCH_BODY:%.*]], label [[OMP_DISPATCH_END:%.*]]
14497 // CHECK19: omp.dispatch.body:
14498 // CHECK19-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4
14499 // CHECK19-NEXT: store i32 [[TMP8]], ptr [[DOTOMP_IV]], align 4
14500 // CHECK19-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
14501 // CHECK19: omp.inner.for.cond:
14502 // CHECK19-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP22:![0-9]+]]
14503 // CHECK19-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP22]]
14504 // CHECK19-NEXT: [[CMP:%.*]] = icmp sle i32 [[TMP9]], [[TMP10]]
14505 // CHECK19-NEXT: br i1 [[CMP]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
14506 // CHECK19: omp.inner.for.body:
14507 // CHECK19-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP22]]
14508 // CHECK19-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP11]], 1
14509 // CHECK19-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
14510 // CHECK19-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP22]]
14511 // CHECK19-NEXT: [[TMP12:%.*]] = load i32, ptr [[I]], align 4, !llvm.access.group [[ACC_GRP22]]
14512 // CHECK19-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds [10 x i32], ptr [[TMP0]], i32 0, i32 [[TMP12]]
14513 // CHECK19-NEXT: store i32 0, ptr [[ARRAYIDX]], align 4, !llvm.access.group [[ACC_GRP22]]
14514 // CHECK19-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
14515 // CHECK19: omp.body.continue:
14516 // CHECK19-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
14517 // CHECK19: omp.inner.for.inc:
14518 // CHECK19-NEXT: [[TMP13:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP22]]
14519 // CHECK19-NEXT: [[ADD1:%.*]] = add nsw i32 [[TMP13]], 1
14520 // CHECK19-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP22]]
14521 // CHECK19-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP23:![0-9]+]]
14522 // CHECK19: omp.inner.for.end:
14523 // CHECK19-NEXT: br label [[OMP_DISPATCH_INC:%.*]]
14524 // CHECK19: omp.dispatch.inc:
14525 // CHECK19-NEXT: br label [[OMP_DISPATCH_COND]]
14526 // CHECK19: omp.dispatch.end:
14527 // CHECK19-NEXT: call void @__kmpc_dispatch_deinit(ptr @[[GLOB3]], i32 [[TMP6]])
14528 // CHECK19-NEXT: ret void
14531 // CHECK19-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l136
14532 // CHECK19-SAME: (i32 noundef [[M:%.*]], ptr noundef nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] {
14533 // CHECK19-NEXT: entry:
14534 // CHECK19-NEXT: [[M_ADDR:%.*]] = alloca i32, align 4
14535 // CHECK19-NEXT: [[A_ADDR:%.*]] = alloca ptr, align 4
14536 // CHECK19-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4
14537 // CHECK19-NEXT: [[DOTCAPTURE_EXPR__CASTED:%.*]] = alloca i32, align 4
14538 // CHECK19-NEXT: store i32 [[M]], ptr [[M_ADDR]], align 4
14539 // CHECK19-NEXT: store ptr [[A]], ptr [[A_ADDR]], align 4
14540 // CHECK19-NEXT: [[TMP0:%.*]] = load ptr, ptr [[A_ADDR]], align 4
14541 // CHECK19-NEXT: [[TMP1:%.*]] = load i32, ptr [[M_ADDR]], align 4
14542 // CHECK19-NEXT: store i32 [[TMP1]], ptr [[DOTCAPTURE_EXPR_]], align 4
14543 // CHECK19-NEXT: [[TMP2:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4
14544 // CHECK19-NEXT: store i32 [[TMP2]], ptr [[DOTCAPTURE_EXPR__CASTED]], align 4
14545 // CHECK19-NEXT: [[TMP3:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR__CASTED]], align 4
14546 // CHECK19-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_teams(ptr @[[GLOB3]], i32 2, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l136.omp_outlined, ptr [[TMP0]], i32 [[TMP3]])
14547 // CHECK19-NEXT: ret void
14550 // CHECK19-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l136.omp_outlined
14551 // CHECK19-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], ptr noundef nonnull align 4 dereferenceable(40) [[A:%.*]], i32 noundef [[DOTCAPTURE_EXPR_:%.*]]) #[[ATTR2]] {
14552 // CHECK19-NEXT: entry:
14553 // CHECK19-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 4
14554 // CHECK19-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 4
14555 // CHECK19-NEXT: [[A_ADDR:%.*]] = alloca ptr, align 4
14556 // CHECK19-NEXT: [[DOTCAPTURE_EXPR__ADDR:%.*]] = alloca i32, align 4
14557 // CHECK19-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
14558 // CHECK19-NEXT: [[TMP:%.*]] = alloca i32, align 4
14559 // CHECK19-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4
14560 // CHECK19-NEXT: [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4
14561 // CHECK19-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
14562 // CHECK19-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
14563 // CHECK19-NEXT: [[I:%.*]] = alloca i32, align 4
14564 // CHECK19-NEXT: [[DOTCAPTURE_EXPR__CASTED:%.*]] = alloca i32, align 4
14565 // CHECK19-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 4
14566 // CHECK19-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 4
14567 // CHECK19-NEXT: store ptr [[A]], ptr [[A_ADDR]], align 4
14568 // CHECK19-NEXT: store i32 [[DOTCAPTURE_EXPR_]], ptr [[DOTCAPTURE_EXPR__ADDR]], align 4
14569 // CHECK19-NEXT: [[TMP0:%.*]] = load ptr, ptr [[A_ADDR]], align 4
14570 // CHECK19-NEXT: store i32 0, ptr [[DOTOMP_COMB_LB]], align 4
14571 // CHECK19-NEXT: store i32 9, ptr [[DOTOMP_COMB_UB]], align 4
14572 // CHECK19-NEXT: store i32 1, ptr [[DOTOMP_STRIDE]], align 4
14573 // CHECK19-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
14574 // CHECK19-NEXT: [[TMP1:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
14575 // CHECK19-NEXT: [[TMP2:%.*]] = load i32, ptr [[TMP1]], align 4
14576 // CHECK19-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB1]], i32 [[TMP2]], i32 92, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
14577 // CHECK19-NEXT: [[TMP3:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
14578 // CHECK19-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP3]], 9
14579 // CHECK19-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
14580 // CHECK19: cond.true:
14581 // CHECK19-NEXT: br label [[COND_END:%.*]]
14582 // CHECK19: cond.false:
14583 // CHECK19-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
14584 // CHECK19-NEXT: br label [[COND_END]]
14585 // CHECK19: cond.end:
14586 // CHECK19-NEXT: [[COND:%.*]] = phi i32 [ 9, [[COND_TRUE]] ], [ [[TMP4]], [[COND_FALSE]] ]
14587 // CHECK19-NEXT: store i32 [[COND]], ptr [[DOTOMP_COMB_UB]], align 4
14588 // CHECK19-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
14589 // CHECK19-NEXT: store i32 [[TMP5]], ptr [[DOTOMP_IV]], align 4
14590 // CHECK19-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
14591 // CHECK19: omp.inner.for.cond:
14592 // CHECK19-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
14593 // CHECK19-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
14594 // CHECK19-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP6]], [[TMP7]]
14595 // CHECK19-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
14596 // CHECK19: omp.inner.for.body:
14597 // CHECK19-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
14598 // CHECK19-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
14599 // CHECK19-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR__ADDR]], align 4
14600 // CHECK19-NEXT: store i32 [[TMP10]], ptr [[DOTCAPTURE_EXPR__CASTED]], align 4
14601 // CHECK19-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR__CASTED]], align 4
14602 // CHECK19-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_call(ptr @[[GLOB3]], i32 4, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l136.omp_outlined.omp_outlined, i32 [[TMP8]], i32 [[TMP9]], ptr [[TMP0]], i32 [[TMP11]])
14603 // CHECK19-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
14604 // CHECK19: omp.inner.for.inc:
14605 // CHECK19-NEXT: [[TMP12:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
14606 // CHECK19-NEXT: [[TMP13:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4
14607 // CHECK19-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP12]], [[TMP13]]
14608 // CHECK19-NEXT: store i32 [[ADD]], ptr [[DOTOMP_IV]], align 4
14609 // CHECK19-NEXT: br label [[OMP_INNER_FOR_COND]]
14610 // CHECK19: omp.inner.for.end:
14611 // CHECK19-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
14612 // CHECK19: omp.loop.exit:
14613 // CHECK19-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB1]], i32 [[TMP2]])
14614 // CHECK19-NEXT: ret void
14617 // CHECK19-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l136.omp_outlined.omp_outlined
14618 // CHECK19-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], i32 noundef [[DOTPREVIOUS_LB_:%.*]], i32 noundef [[DOTPREVIOUS_UB_:%.*]], ptr noundef nonnull align 4 dereferenceable(40) [[A:%.*]], i32 noundef [[DOTCAPTURE_EXPR_:%.*]]) #[[ATTR2]] {
14619 // CHECK19-NEXT: entry:
14620 // CHECK19-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 4
14621 // CHECK19-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 4
14622 // CHECK19-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i32, align 4
14623 // CHECK19-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i32, align 4
14624 // CHECK19-NEXT: [[A_ADDR:%.*]] = alloca ptr, align 4
14625 // CHECK19-NEXT: [[DOTCAPTURE_EXPR__ADDR:%.*]] = alloca i32, align 4
14626 // CHECK19-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
14627 // CHECK19-NEXT: [[TMP:%.*]] = alloca i32, align 4
14628 // CHECK19-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4
14629 // CHECK19-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4
14630 // CHECK19-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
14631 // CHECK19-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
14632 // CHECK19-NEXT: [[I:%.*]] = alloca i32, align 4
14633 // CHECK19-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 4
14634 // CHECK19-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 4
14635 // CHECK19-NEXT: store i32 [[DOTPREVIOUS_LB_]], ptr [[DOTPREVIOUS_LB__ADDR]], align 4
14636 // CHECK19-NEXT: store i32 [[DOTPREVIOUS_UB_]], ptr [[DOTPREVIOUS_UB__ADDR]], align 4
14637 // CHECK19-NEXT: store ptr [[A]], ptr [[A_ADDR]], align 4
14638 // CHECK19-NEXT: store i32 [[DOTCAPTURE_EXPR_]], ptr [[DOTCAPTURE_EXPR__ADDR]], align 4
14639 // CHECK19-NEXT: [[TMP0:%.*]] = load ptr, ptr [[A_ADDR]], align 4
14640 // CHECK19-NEXT: store i32 0, ptr [[DOTOMP_LB]], align 4
14641 // CHECK19-NEXT: store i32 9, ptr [[DOTOMP_UB]], align 4
14642 // CHECK19-NEXT: [[TMP1:%.*]] = load i32, ptr [[DOTPREVIOUS_LB__ADDR]], align 4
14643 // CHECK19-NEXT: [[TMP2:%.*]] = load i32, ptr [[DOTPREVIOUS_UB__ADDR]], align 4
14644 // CHECK19-NEXT: store i32 [[TMP1]], ptr [[DOTOMP_LB]], align 4
14645 // CHECK19-NEXT: store i32 [[TMP2]], ptr [[DOTOMP_UB]], align 4
14646 // CHECK19-NEXT: store i32 1, ptr [[DOTOMP_STRIDE]], align 4
14647 // CHECK19-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
14648 // CHECK19-NEXT: [[TMP3:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR__ADDR]], align 4
14649 // CHECK19-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4
14650 // CHECK19-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
14651 // CHECK19-NEXT: [[TMP6:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
14652 // CHECK19-NEXT: [[TMP7:%.*]] = load i32, ptr [[TMP6]], align 4
14653 // CHECK19-NEXT: call void @__kmpc_dispatch_init_4(ptr @[[GLOB3]], i32 [[TMP7]], i32 35, i32 [[TMP4]], i32 [[TMP5]], i32 1, i32 [[TMP3]])
14654 // CHECK19-NEXT: br label [[OMP_DISPATCH_COND:%.*]]
14655 // CHECK19: omp.dispatch.cond:
14656 // CHECK19-NEXT: [[TMP8:%.*]] = call i32 @__kmpc_dispatch_next_4(ptr @[[GLOB3]], i32 [[TMP7]], ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]])
14657 // CHECK19-NEXT: [[TOBOOL:%.*]] = icmp ne i32 [[TMP8]], 0
14658 // CHECK19-NEXT: br i1 [[TOBOOL]], label [[OMP_DISPATCH_BODY:%.*]], label [[OMP_DISPATCH_END:%.*]]
14659 // CHECK19: omp.dispatch.body:
14660 // CHECK19-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4
14661 // CHECK19-NEXT: store i32 [[TMP9]], ptr [[DOTOMP_IV]], align 4
14662 // CHECK19-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
14663 // CHECK19: omp.inner.for.cond:
14664 // CHECK19-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP25:![0-9]+]]
14665 // CHECK19-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP25]]
14666 // CHECK19-NEXT: [[CMP:%.*]] = icmp sle i32 [[TMP10]], [[TMP11]]
14667 // CHECK19-NEXT: br i1 [[CMP]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
14668 // CHECK19: omp.inner.for.body:
14669 // CHECK19-NEXT: [[TMP12:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP25]]
14670 // CHECK19-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP12]], 1
14671 // CHECK19-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
14672 // CHECK19-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP25]]
14673 // CHECK19-NEXT: [[TMP13:%.*]] = load i32, ptr [[I]], align 4, !llvm.access.group [[ACC_GRP25]]
14674 // CHECK19-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds [10 x i32], ptr [[TMP0]], i32 0, i32 [[TMP13]]
14675 // CHECK19-NEXT: store i32 0, ptr [[ARRAYIDX]], align 4, !llvm.access.group [[ACC_GRP25]]
14676 // CHECK19-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
14677 // CHECK19: omp.body.continue:
14678 // CHECK19-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
14679 // CHECK19: omp.inner.for.inc:
14680 // CHECK19-NEXT: [[TMP14:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP25]]
14681 // CHECK19-NEXT: [[ADD1:%.*]] = add nsw i32 [[TMP14]], 1
14682 // CHECK19-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP25]]
14683 // CHECK19-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP26:![0-9]+]]
14684 // CHECK19: omp.inner.for.end:
14685 // CHECK19-NEXT: br label [[OMP_DISPATCH_INC:%.*]]
14686 // CHECK19: omp.dispatch.inc:
14687 // CHECK19-NEXT: br label [[OMP_DISPATCH_COND]]
14688 // CHECK19: omp.dispatch.end:
14689 // CHECK19-NEXT: call void @__kmpc_dispatch_deinit(ptr @[[GLOB3]], i32 [[TMP7]])
14690 // CHECK19-NEXT: ret void