?  search:
summary | log | graphiclog1 | graphiclog2 | commit | commitdiff | tree | refs | edit | fork
blame | history | raw | HEAD
[clang] Implement lifetime analysis for lifetime_capture_by(X) (#115921)
[llvm-project.git] / clang / test / OpenMP / target_teams_distribute_simd_collapse_codegen.cpp
bloba5ce3c7d5bf8cb37dbe9eab3208cc07a70bfaaac
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
3 #ifndef HEADER
4 #define HEADER
5
6 // Test host codegen.
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
13
14 // 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 --check-prefix=CHECK5
15 // 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
16 // 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 --check-prefix=CHECK5
17 // 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 --check-prefix=CHECK7
18 // 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
19 // 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 --check-prefix=CHECK7
20 #ifdef CK1
21
22 template <typename T, int X, long long Y>
23 struct SS{
24 T a[X][Y];
25
26 int foo(void) {
27
28 #pragma omp target teams distribute simd collapse(2)
29 for(int i = 0; i < X; i++) {
30 for(int j = 0; j < Y; j++) {
31 a[i][j] = (T)0;
32 }
33 }
34
35 // discard loop variables not needed here
36
37 return a[0][0];
38 }
39 };
40
41 int teams_template_struct(void) {
42 SS<int, 123, 456> V;
43 return V.foo();
44
45 }
46 #endif // CK1
47
48 // Test host codegen.
49 // 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=CHECK9
50 // 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
51 // 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=CHECK9
52 // 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=CHECK11
53 // 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
54 // 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=CHECK11
55
56 // 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 --check-prefix=CHECK13
57 // 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
58 // 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 --check-prefix=CHECK13
59 // 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 --check-prefix=CHECK15
60 // 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
61 // 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 --check-prefix=CHECK15
62 #ifdef CK2
63
64 template <typename T, int n, int m>
65 int tmain(T argc) {
66 T a[n][m];
67 #pragma omp target teams distribute simd collapse(2)
68 for(int i = 0; i < n; i++) {
69 for(int j = 0; j < m; j++) {
70 a[i][j] = (T)0;
71 }
72 }
73 return 0;
74 }
75
76 int main (int argc, char **argv) {
77 int n = 100;
78 int m = 2;
79 int a[n][m];
80 #pragma omp target teams distribute simd collapse(2)
81 for(int i = 0; i < n; i++) {
82 for(int j = 0; j < m; j++) {
83 a[i][j] = 0;
84 }
85 }
86 return tmain<int, 10, 2>(argc);
87 }
88
89
90
91
92
93 // discard loop variables not needed here
94
95 #endif // CK2
96 #endif // #ifndef HEADER
97 // CHECK1-LABEL: define {{[^@]+}}@_Z21teams_template_structv
98 // CHECK1-SAME: () #[[ATTR0:[0-9]+]] {
99 // CHECK1-NEXT: entry:
100 // CHECK1-NEXT: [[V:%.*]] = alloca [[STRUCT_SS:%.*]], align 4
101 // CHECK1-NEXT: [[CALL:%.*]] = call noundef signext i32 @_ZN2SSIiLi123ELx456EE3fooEv(ptr noundef nonnull align 4 dereferenceable(224352) [[V]])
102 // CHECK1-NEXT: ret i32 [[CALL]]
103 //
104 //
105 // CHECK1-LABEL: define {{[^@]+}}@_ZN2SSIiLi123ELx456EE3fooEv
106 // CHECK1-SAME: (ptr noundef nonnull align 4 dereferenceable(224352) [[THIS:%.*]]) #[[ATTR0]] comdat {
107 // CHECK1-NEXT: entry:
108 // CHECK1-NEXT: [[THIS_ADDR:%.*]] = alloca ptr, align 8
109 // CHECK1-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [1 x ptr], align 8
110 // CHECK1-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [1 x ptr], align 8
111 // CHECK1-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [1 x ptr], align 8
112 // CHECK1-NEXT: [[TMP:%.*]] = alloca i32, align 4
113 // CHECK1-NEXT: [[_TMP2:%.*]] = alloca i32, align 4
114 // CHECK1-NEXT: [[KERNEL_ARGS:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS:%.*]], align 8
115 // CHECK1-NEXT: store ptr [[THIS]], ptr [[THIS_ADDR]], align 8
116 // CHECK1-NEXT: [[THIS1:%.*]] = load ptr, ptr [[THIS_ADDR]], align 8
117 // CHECK1-NEXT: [[A:%.*]] = getelementptr inbounds nuw [[STRUCT_SS:%.*]], ptr [[THIS1]], i32 0, i32 0
118 // CHECK1-NEXT: [[TMP0:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
119 // CHECK1-NEXT: store ptr [[THIS1]], ptr [[TMP0]], align 8
120 // CHECK1-NEXT: [[TMP1:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
121 // CHECK1-NEXT: store ptr [[A]], ptr [[TMP1]], align 8
122 // CHECK1-NEXT: [[TMP2:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0
123 // CHECK1-NEXT: store ptr null, ptr [[TMP2]], align 8
124 // CHECK1-NEXT: [[TMP3:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
125 // CHECK1-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
126 // CHECK1-NEXT: [[TMP5:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 0
127 // CHECK1-NEXT: store i32 3, ptr [[TMP5]], align 4
128 // CHECK1-NEXT: [[TMP6:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 1
129 // CHECK1-NEXT: store i32 1, ptr [[TMP6]], align 4
130 // CHECK1-NEXT: [[TMP7:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 2
131 // CHECK1-NEXT: store ptr [[TMP3]], ptr [[TMP7]], align 8
132 // CHECK1-NEXT: [[TMP8:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 3
133 // CHECK1-NEXT: store ptr [[TMP4]], ptr [[TMP8]], align 8
134 // CHECK1-NEXT: [[TMP9:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 4
135 // CHECK1-NEXT: store ptr @.offload_sizes, ptr [[TMP9]], align 8
136 // CHECK1-NEXT: [[TMP10:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 5
137 // CHECK1-NEXT: store ptr @.offload_maptypes, ptr [[TMP10]], align 8
138 // CHECK1-NEXT: [[TMP11:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 6
139 // CHECK1-NEXT: store ptr null, ptr [[TMP11]], align 8
140 // CHECK1-NEXT: [[TMP12:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 7
141 // CHECK1-NEXT: store ptr null, ptr [[TMP12]], align 8
142 // CHECK1-NEXT: [[TMP13:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 8
143 // CHECK1-NEXT: store i64 56088, ptr [[TMP13]], align 8
144 // CHECK1-NEXT: [[TMP14:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 9
145 // CHECK1-NEXT: store i64 0, ptr [[TMP14]], align 8
146 // CHECK1-NEXT: [[TMP15:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 10
147 // CHECK1-NEXT: store [3 x i32] zeroinitializer, ptr [[TMP15]], align 4
148 // CHECK1-NEXT: [[TMP16:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 11
149 // CHECK1-NEXT: store [3 x i32] [i32 1, i32 0, i32 0], ptr [[TMP16]], align 4
150 // CHECK1-NEXT: [[TMP17:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 12
151 // CHECK1-NEXT: store i32 0, ptr [[TMP17]], align 4
152 // CHECK1-NEXT: [[TMP18:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB2:[0-9]+]], i64 -1, i32 0, i32 1, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l28.region_id, ptr [[KERNEL_ARGS]])
153 // CHECK1-NEXT: [[TMP19:%.*]] = icmp ne i32 [[TMP18]], 0
154 // CHECK1-NEXT: br i1 [[TMP19]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
155 // CHECK1: omp_offload.failed:
156 // CHECK1-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l28(ptr [[THIS1]]) #[[ATTR2:[0-9]+]]
157 // CHECK1-NEXT: br label [[OMP_OFFLOAD_CONT]]
158 // CHECK1: omp_offload.cont:
159 // CHECK1-NEXT: [[A3:%.*]] = getelementptr inbounds nuw [[STRUCT_SS]], ptr [[THIS1]], i32 0, i32 0
160 // CHECK1-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds [123 x [456 x i32]], ptr [[A3]], i64 0, i64 0
161 // CHECK1-NEXT: [[ARRAYIDX4:%.*]] = getelementptr inbounds [456 x i32], ptr [[ARRAYIDX]], i64 0, i64 0
162 // CHECK1-NEXT: [[TMP20:%.*]] = load i32, ptr [[ARRAYIDX4]], align 4
163 // CHECK1-NEXT: ret i32 [[TMP20]]
164 //
165 //
166 // CHECK1-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l28
167 // CHECK1-SAME: (ptr noundef [[THIS:%.*]]) #[[ATTR1:[0-9]+]] {
168 // CHECK1-NEXT: entry:
169 // CHECK1-NEXT: [[THIS_ADDR:%.*]] = alloca ptr, align 8
170 // CHECK1-NEXT: store ptr [[THIS]], ptr [[THIS_ADDR]], align 8
171 // CHECK1-NEXT: [[TMP0:%.*]] = load ptr, ptr [[THIS_ADDR]], align 8
172 // CHECK1-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_teams(ptr @[[GLOB2]], i32 1, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l28.omp_outlined, ptr [[TMP0]])
173 // CHECK1-NEXT: ret void
174 //
175 //
176 // CHECK1-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l28.omp_outlined
177 // CHECK1-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], ptr noundef [[THIS:%.*]]) #[[ATTR1]] {
178 // CHECK1-NEXT: entry:
179 // CHECK1-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 8
180 // CHECK1-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 8
181 // CHECK1-NEXT: [[THIS_ADDR:%.*]] = alloca ptr, align 8
182 // CHECK1-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
183 // CHECK1-NEXT: [[TMP:%.*]] = alloca i32, align 4
184 // CHECK1-NEXT: [[_TMP1:%.*]] = alloca i32, align 4
185 // CHECK1-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4
186 // CHECK1-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4
187 // CHECK1-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
188 // CHECK1-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
189 // CHECK1-NEXT: [[I:%.*]] = alloca i32, align 4
190 // CHECK1-NEXT: [[J:%.*]] = alloca i32, align 4
191 // CHECK1-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 8
192 // CHECK1-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 8
193 // CHECK1-NEXT: store ptr [[THIS]], ptr [[THIS_ADDR]], align 8
194 // CHECK1-NEXT: [[TMP0:%.*]] = load ptr, ptr [[THIS_ADDR]], align 8
195 // CHECK1-NEXT: store i32 0, ptr [[DOTOMP_LB]], align 4
196 // CHECK1-NEXT: store i32 56087, ptr [[DOTOMP_UB]], align 4
197 // CHECK1-NEXT: store i32 1, ptr [[DOTOMP_STRIDE]], align 4
198 // CHECK1-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
199 // CHECK1-NEXT: [[TMP1:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
200 // CHECK1-NEXT: [[TMP2:%.*]] = load i32, ptr [[TMP1]], align 4
201 // CHECK1-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB1:[0-9]+]], i32 [[TMP2]], i32 92, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
202 // CHECK1-NEXT: [[TMP3:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
203 // CHECK1-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP3]], 56087
204 // CHECK1-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
205 // CHECK1: cond.true:
206 // CHECK1-NEXT: br label [[COND_END:%.*]]
207 // CHECK1: cond.false:
208 // CHECK1-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
209 // CHECK1-NEXT: br label [[COND_END]]
210 // CHECK1: cond.end:
211 // CHECK1-NEXT: [[COND:%.*]] = phi i32 [ 56087, [[COND_TRUE]] ], [ [[TMP4]], [[COND_FALSE]] ]
212 // CHECK1-NEXT: store i32 [[COND]], ptr [[DOTOMP_UB]], align 4
213 // CHECK1-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4
214 // CHECK1-NEXT: store i32 [[TMP5]], ptr [[DOTOMP_IV]], align 4
215 // CHECK1-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
216 // CHECK1: omp.inner.for.cond:
217 // CHECK1-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP4:![0-9]+]]
218 // CHECK1-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP4]]
219 // CHECK1-NEXT: [[CMP2:%.*]] = icmp sle i32 [[TMP6]], [[TMP7]]
220 // CHECK1-NEXT: br i1 [[CMP2]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
221 // CHECK1: omp.inner.for.body:
222 // CHECK1-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP4]]
223 // CHECK1-NEXT: [[DIV:%.*]] = sdiv i32 [[TMP8]], 456
224 // CHECK1-NEXT: [[MUL:%.*]] = mul nsw i32 [[DIV]], 1
225 // CHECK1-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
226 // CHECK1-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP4]]
227 // CHECK1-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP4]]
228 // CHECK1-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP4]]
229 // CHECK1-NEXT: [[DIV3:%.*]] = sdiv i32 [[TMP10]], 456
230 // CHECK1-NEXT: [[MUL4:%.*]] = mul nsw i32 [[DIV3]], 456
231 // CHECK1-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP9]], [[MUL4]]
232 // CHECK1-NEXT: [[MUL5:%.*]] = mul nsw i32 [[SUB]], 1
233 // CHECK1-NEXT: [[ADD6:%.*]] = add nsw i32 0, [[MUL5]]
234 // CHECK1-NEXT: store i32 [[ADD6]], ptr [[J]], align 4, !llvm.access.group [[ACC_GRP4]]
235 // CHECK1-NEXT: [[A:%.*]] = getelementptr inbounds nuw [[STRUCT_SS:%.*]], ptr [[TMP0]], i32 0, i32 0
236 // CHECK1-NEXT: [[TMP11:%.*]] = load i32, ptr [[I]], align 4, !llvm.access.group [[ACC_GRP4]]
237 // CHECK1-NEXT: [[IDXPROM:%.*]] = sext i32 [[TMP11]] to i64
238 // CHECK1-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds [123 x [456 x i32]], ptr [[A]], i64 0, i64 [[IDXPROM]]
239 // CHECK1-NEXT: [[TMP12:%.*]] = load i32, ptr [[J]], align 4, !llvm.access.group [[ACC_GRP4]]
240 // CHECK1-NEXT: [[IDXPROM7:%.*]] = sext i32 [[TMP12]] to i64
241 // CHECK1-NEXT: [[ARRAYIDX8:%.*]] = getelementptr inbounds [456 x i32], ptr [[ARRAYIDX]], i64 0, i64 [[IDXPROM7]]
242 // CHECK1-NEXT: store i32 0, ptr [[ARRAYIDX8]], align 4, !llvm.access.group [[ACC_GRP4]]
243 // CHECK1-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
244 // CHECK1: omp.body.continue:
245 // CHECK1-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
246 // CHECK1: omp.inner.for.inc:
247 // CHECK1-NEXT: [[TMP13:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP4]]
248 // CHECK1-NEXT: [[ADD9:%.*]] = add nsw i32 [[TMP13]], 1
249 // CHECK1-NEXT: store i32 [[ADD9]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP4]]
250 // CHECK1-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP5:![0-9]+]]
251 // CHECK1: omp.inner.for.end:
252 // CHECK1-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
253 // CHECK1: omp.loop.exit:
254 // CHECK1-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB1]], i32 [[TMP2]])
255 // CHECK1-NEXT: [[TMP14:%.*]] = load i32, ptr [[DOTOMP_IS_LAST]], align 4
256 // CHECK1-NEXT: [[TMP15:%.*]] = icmp ne i32 [[TMP14]], 0
257 // CHECK1-NEXT: br i1 [[TMP15]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
258 // CHECK1: .omp.final.then:
259 // CHECK1-NEXT: store i32 123, ptr [[I]], align 4
260 // CHECK1-NEXT: store i32 456, ptr [[J]], align 4
261 // CHECK1-NEXT: br label [[DOTOMP_FINAL_DONE]]
262 // CHECK1: .omp.final.done:
263 // CHECK1-NEXT: ret void
264 //
265 //
266 // CHECK3-LABEL: define {{[^@]+}}@_Z21teams_template_structv
267 // CHECK3-SAME: () #[[ATTR0:[0-9]+]] {
268 // CHECK3-NEXT: entry:
269 // CHECK3-NEXT: [[V:%.*]] = alloca [[STRUCT_SS:%.*]], align 4
270 // CHECK3-NEXT: [[CALL:%.*]] = call noundef i32 @_ZN2SSIiLi123ELx456EE3fooEv(ptr noundef nonnull align 4 dereferenceable(224352) [[V]])
271 // CHECK3-NEXT: ret i32 [[CALL]]
272 //
273 //
274 // CHECK3-LABEL: define {{[^@]+}}@_ZN2SSIiLi123ELx456EE3fooEv
275 // CHECK3-SAME: (ptr noundef nonnull align 4 dereferenceable(224352) [[THIS:%.*]]) #[[ATTR0]] comdat align 2 {
276 // CHECK3-NEXT: entry:
277 // CHECK3-NEXT: [[THIS_ADDR:%.*]] = alloca ptr, align 4
278 // CHECK3-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [1 x ptr], align 4
279 // CHECK3-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [1 x ptr], align 4
280 // CHECK3-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [1 x ptr], align 4
281 // CHECK3-NEXT: [[TMP:%.*]] = alloca i32, align 4
282 // CHECK3-NEXT: [[_TMP2:%.*]] = alloca i32, align 4
283 // CHECK3-NEXT: [[KERNEL_ARGS:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS:%.*]], align 8
284 // CHECK3-NEXT: store ptr [[THIS]], ptr [[THIS_ADDR]], align 4
285 // CHECK3-NEXT: [[THIS1:%.*]] = load ptr, ptr [[THIS_ADDR]], align 4
286 // CHECK3-NEXT: [[A:%.*]] = getelementptr inbounds nuw [[STRUCT_SS:%.*]], ptr [[THIS1]], i32 0, i32 0
287 // CHECK3-NEXT: [[TMP0:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
288 // CHECK3-NEXT: store ptr [[THIS1]], ptr [[TMP0]], align 4
289 // CHECK3-NEXT: [[TMP1:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
290 // CHECK3-NEXT: store ptr [[A]], ptr [[TMP1]], align 4
291 // CHECK3-NEXT: [[TMP2:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0
292 // CHECK3-NEXT: store ptr null, ptr [[TMP2]], align 4
293 // CHECK3-NEXT: [[TMP3:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
294 // CHECK3-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
295 // CHECK3-NEXT: [[TMP5:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 0
296 // CHECK3-NEXT: store i32 3, ptr [[TMP5]], align 4
297 // CHECK3-NEXT: [[TMP6:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 1
298 // CHECK3-NEXT: store i32 1, ptr [[TMP6]], align 4
299 // CHECK3-NEXT: [[TMP7:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 2
300 // CHECK3-NEXT: store ptr [[TMP3]], ptr [[TMP7]], align 4
301 // CHECK3-NEXT: [[TMP8:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 3
302 // CHECK3-NEXT: store ptr [[TMP4]], ptr [[TMP8]], align 4
303 // CHECK3-NEXT: [[TMP9:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 4
304 // CHECK3-NEXT: store ptr @.offload_sizes, ptr [[TMP9]], align 4
305 // CHECK3-NEXT: [[TMP10:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 5
306 // CHECK3-NEXT: store ptr @.offload_maptypes, ptr [[TMP10]], align 4
307 // CHECK3-NEXT: [[TMP11:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 6
308 // CHECK3-NEXT: store ptr null, ptr [[TMP11]], align 4
309 // CHECK3-NEXT: [[TMP12:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 7
310 // CHECK3-NEXT: store ptr null, ptr [[TMP12]], align 4
311 // CHECK3-NEXT: [[TMP13:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 8
312 // CHECK3-NEXT: store i64 56088, ptr [[TMP13]], align 8
313 // CHECK3-NEXT: [[TMP14:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 9
314 // CHECK3-NEXT: store i64 0, ptr [[TMP14]], align 8
315 // CHECK3-NEXT: [[TMP15:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 10
316 // CHECK3-NEXT: store [3 x i32] zeroinitializer, ptr [[TMP15]], align 4
317 // CHECK3-NEXT: [[TMP16:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 11
318 // CHECK3-NEXT: store [3 x i32] [i32 1, i32 0, i32 0], ptr [[TMP16]], align 4
319 // CHECK3-NEXT: [[TMP17:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 12
320 // CHECK3-NEXT: store i32 0, ptr [[TMP17]], align 4
321 // CHECK3-NEXT: [[TMP18:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB2:[0-9]+]], i64 -1, i32 0, i32 1, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l28.region_id, ptr [[KERNEL_ARGS]])
322 // CHECK3-NEXT: [[TMP19:%.*]] = icmp ne i32 [[TMP18]], 0
323 // CHECK3-NEXT: br i1 [[TMP19]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
324 // CHECK3: omp_offload.failed:
325 // CHECK3-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l28(ptr [[THIS1]]) #[[ATTR2:[0-9]+]]
326 // CHECK3-NEXT: br label [[OMP_OFFLOAD_CONT]]
327 // CHECK3: omp_offload.cont:
328 // CHECK3-NEXT: [[A3:%.*]] = getelementptr inbounds nuw [[STRUCT_SS]], ptr [[THIS1]], i32 0, i32 0
329 // CHECK3-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds [123 x [456 x i32]], ptr [[A3]], i32 0, i32 0
330 // CHECK3-NEXT: [[ARRAYIDX4:%.*]] = getelementptr inbounds [456 x i32], ptr [[ARRAYIDX]], i32 0, i32 0
331 // CHECK3-NEXT: [[TMP20:%.*]] = load i32, ptr [[ARRAYIDX4]], align 4
332 // CHECK3-NEXT: ret i32 [[TMP20]]
333 //
334 //
335 // CHECK3-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l28
336 // CHECK3-SAME: (ptr noundef [[THIS:%.*]]) #[[ATTR1:[0-9]+]] {
337 // CHECK3-NEXT: entry:
338 // CHECK3-NEXT: [[THIS_ADDR:%.*]] = alloca ptr, align 4
339 // CHECK3-NEXT: store ptr [[THIS]], ptr [[THIS_ADDR]], align 4
340 // CHECK3-NEXT: [[TMP0:%.*]] = load ptr, ptr [[THIS_ADDR]], align 4
341 // CHECK3-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_teams(ptr @[[GLOB2]], i32 1, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l28.omp_outlined, ptr [[TMP0]])
342 // CHECK3-NEXT: ret void
343 //
344 //
345 // CHECK3-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l28.omp_outlined
346 // CHECK3-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], ptr noundef [[THIS:%.*]]) #[[ATTR1]] {
347 // CHECK3-NEXT: entry:
348 // CHECK3-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 4
349 // CHECK3-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 4
350 // CHECK3-NEXT: [[THIS_ADDR:%.*]] = alloca ptr, align 4
351 // CHECK3-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
352 // CHECK3-NEXT: [[TMP:%.*]] = alloca i32, align 4
353 // CHECK3-NEXT: [[_TMP1:%.*]] = alloca i32, align 4
354 // CHECK3-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4
355 // CHECK3-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4
356 // CHECK3-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
357 // CHECK3-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
358 // CHECK3-NEXT: [[I:%.*]] = alloca i32, align 4
359 // CHECK3-NEXT: [[J:%.*]] = alloca i32, align 4
360 // CHECK3-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 4
361 // CHECK3-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 4
362 // CHECK3-NEXT: store ptr [[THIS]], ptr [[THIS_ADDR]], align 4
363 // CHECK3-NEXT: [[TMP0:%.*]] = load ptr, ptr [[THIS_ADDR]], align 4
364 // CHECK3-NEXT: store i32 0, ptr [[DOTOMP_LB]], align 4
365 // CHECK3-NEXT: store i32 56087, ptr [[DOTOMP_UB]], align 4
366 // CHECK3-NEXT: store i32 1, ptr [[DOTOMP_STRIDE]], align 4
367 // CHECK3-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
368 // CHECK3-NEXT: [[TMP1:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
369 // CHECK3-NEXT: [[TMP2:%.*]] = load i32, ptr [[TMP1]], align 4
370 // CHECK3-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB1:[0-9]+]], i32 [[TMP2]], i32 92, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
371 // CHECK3-NEXT: [[TMP3:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
372 // CHECK3-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP3]], 56087
373 // CHECK3-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
374 // CHECK3: cond.true:
375 // CHECK3-NEXT: br label [[COND_END:%.*]]
376 // CHECK3: cond.false:
377 // CHECK3-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
378 // CHECK3-NEXT: br label [[COND_END]]
379 // CHECK3: cond.end:
380 // CHECK3-NEXT: [[COND:%.*]] = phi i32 [ 56087, [[COND_TRUE]] ], [ [[TMP4]], [[COND_FALSE]] ]
381 // CHECK3-NEXT: store i32 [[COND]], ptr [[DOTOMP_UB]], align 4
382 // CHECK3-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4
383 // CHECK3-NEXT: store i32 [[TMP5]], ptr [[DOTOMP_IV]], align 4
384 // CHECK3-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
385 // CHECK3: omp.inner.for.cond:
386 // CHECK3-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP5:![0-9]+]]
387 // CHECK3-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP5]]
388 // CHECK3-NEXT: [[CMP2:%.*]] = icmp sle i32 [[TMP6]], [[TMP7]]
389 // CHECK3-NEXT: br i1 [[CMP2]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
390 // CHECK3: omp.inner.for.body:
391 // CHECK3-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP5]]
392 // CHECK3-NEXT: [[DIV:%.*]] = sdiv i32 [[TMP8]], 456
393 // CHECK3-NEXT: [[MUL:%.*]] = mul nsw i32 [[DIV]], 1
394 // CHECK3-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
395 // CHECK3-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP5]]
396 // CHECK3-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP5]]
397 // CHECK3-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP5]]
398 // CHECK3-NEXT: [[DIV3:%.*]] = sdiv i32 [[TMP10]], 456
399 // CHECK3-NEXT: [[MUL4:%.*]] = mul nsw i32 [[DIV3]], 456
400 // CHECK3-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP9]], [[MUL4]]
401 // CHECK3-NEXT: [[MUL5:%.*]] = mul nsw i32 [[SUB]], 1
402 // CHECK3-NEXT: [[ADD6:%.*]] = add nsw i32 0, [[MUL5]]
403 // CHECK3-NEXT: store i32 [[ADD6]], ptr [[J]], align 4, !llvm.access.group [[ACC_GRP5]]
404 // CHECK3-NEXT: [[A:%.*]] = getelementptr inbounds nuw [[STRUCT_SS:%.*]], ptr [[TMP0]], i32 0, i32 0
405 // CHECK3-NEXT: [[TMP11:%.*]] = load i32, ptr [[I]], align 4, !llvm.access.group [[ACC_GRP5]]
406 // CHECK3-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds [123 x [456 x i32]], ptr [[A]], i32 0, i32 [[TMP11]]
407 // CHECK3-NEXT: [[TMP12:%.*]] = load i32, ptr [[J]], align 4, !llvm.access.group [[ACC_GRP5]]
408 // CHECK3-NEXT: [[ARRAYIDX7:%.*]] = getelementptr inbounds [456 x i32], ptr [[ARRAYIDX]], i32 0, i32 [[TMP12]]
409 // CHECK3-NEXT: store i32 0, ptr [[ARRAYIDX7]], align 4, !llvm.access.group [[ACC_GRP5]]
410 // CHECK3-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
411 // CHECK3: omp.body.continue:
412 // CHECK3-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
413 // CHECK3: omp.inner.for.inc:
414 // CHECK3-NEXT: [[TMP13:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP5]]
415 // CHECK3-NEXT: [[ADD8:%.*]] = add nsw i32 [[TMP13]], 1
416 // CHECK3-NEXT: store i32 [[ADD8]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP5]]
417 // CHECK3-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP6:![0-9]+]]
418 // CHECK3: omp.inner.for.end:
419 // CHECK3-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
420 // CHECK3: omp.loop.exit:
421 // CHECK3-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB1]], i32 [[TMP2]])
422 // CHECK3-NEXT: [[TMP14:%.*]] = load i32, ptr [[DOTOMP_IS_LAST]], align 4
423 // CHECK3-NEXT: [[TMP15:%.*]] = icmp ne i32 [[TMP14]], 0
424 // CHECK3-NEXT: br i1 [[TMP15]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
425 // CHECK3: .omp.final.then:
426 // CHECK3-NEXT: store i32 123, ptr [[I]], align 4
427 // CHECK3-NEXT: store i32 456, ptr [[J]], align 4
428 // CHECK3-NEXT: br label [[DOTOMP_FINAL_DONE]]
429 // CHECK3: .omp.final.done:
430 // CHECK3-NEXT: ret void
431 //
432 //
433 // CHECK5-LABEL: define {{[^@]+}}@_Z21teams_template_structv
434 // CHECK5-SAME: () #[[ATTR0:[0-9]+]] {
435 // CHECK5-NEXT: entry:
436 // CHECK5-NEXT: [[V:%.*]] = alloca [[STRUCT_SS:%.*]], align 4
437 // CHECK5-NEXT: [[CALL:%.*]] = call noundef signext i32 @_ZN2SSIiLi123ELx456EE3fooEv(ptr noundef nonnull align 4 dereferenceable(224352) [[V]])
438 // CHECK5-NEXT: ret i32 [[CALL]]
439 //
440 //
441 // CHECK5-LABEL: define {{[^@]+}}@_ZN2SSIiLi123ELx456EE3fooEv
442 // CHECK5-SAME: (ptr noundef nonnull align 4 dereferenceable(224352) [[THIS:%.*]]) #[[ATTR0]] comdat {
443 // CHECK5-NEXT: entry:
444 // CHECK5-NEXT: [[THIS_ADDR:%.*]] = alloca ptr, align 8
445 // CHECK5-NEXT: [[TMP:%.*]] = alloca i32, align 4
446 // CHECK5-NEXT: [[_TMP2:%.*]] = alloca i32, align 4
447 // CHECK5-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4
448 // CHECK5-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4
449 // CHECK5-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
450 // CHECK5-NEXT: [[I:%.*]] = alloca i32, align 4
451 // CHECK5-NEXT: [[J:%.*]] = alloca i32, align 4
452 // CHECK5-NEXT: store ptr [[THIS]], ptr [[THIS_ADDR]], align 8
453 // CHECK5-NEXT: [[THIS1:%.*]] = load ptr, ptr [[THIS_ADDR]], align 8
454 // CHECK5-NEXT: store i32 0, ptr [[DOTOMP_LB]], align 4
455 // CHECK5-NEXT: store i32 56087, ptr [[DOTOMP_UB]], align 4
456 // CHECK5-NEXT: [[TMP0:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4
457 // CHECK5-NEXT: store i32 [[TMP0]], ptr [[DOTOMP_IV]], align 4
458 // CHECK5-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
459 // CHECK5: omp.inner.for.cond:
460 // CHECK5-NEXT: [[TMP1:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP2:![0-9]+]]
461 // CHECK5-NEXT: [[TMP2:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP2]]
462 // CHECK5-NEXT: [[CMP:%.*]] = icmp sle i32 [[TMP1]], [[TMP2]]
463 // CHECK5-NEXT: br i1 [[CMP]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
464 // CHECK5: omp.inner.for.body:
465 // CHECK5-NEXT: [[TMP3:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP2]]
466 // CHECK5-NEXT: [[DIV:%.*]] = sdiv i32 [[TMP3]], 456
467 // CHECK5-NEXT: [[MUL:%.*]] = mul nsw i32 [[DIV]], 1
468 // CHECK5-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
469 // CHECK5-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP2]]
470 // CHECK5-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP2]]
471 // CHECK5-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP2]]
472 // CHECK5-NEXT: [[DIV3:%.*]] = sdiv i32 [[TMP5]], 456
473 // CHECK5-NEXT: [[MUL4:%.*]] = mul nsw i32 [[DIV3]], 456
474 // CHECK5-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP4]], [[MUL4]]
475 // CHECK5-NEXT: [[MUL5:%.*]] = mul nsw i32 [[SUB]], 1
476 // CHECK5-NEXT: [[ADD6:%.*]] = add nsw i32 0, [[MUL5]]
477 // CHECK5-NEXT: store i32 [[ADD6]], ptr [[J]], align 4, !llvm.access.group [[ACC_GRP2]]
478 // CHECK5-NEXT: [[A:%.*]] = getelementptr inbounds nuw [[STRUCT_SS:%.*]], ptr [[THIS1]], i32 0, i32 0
479 // CHECK5-NEXT: [[TMP6:%.*]] = load i32, ptr [[I]], align 4, !llvm.access.group [[ACC_GRP2]]
480 // CHECK5-NEXT: [[IDXPROM:%.*]] = sext i32 [[TMP6]] to i64
481 // CHECK5-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds [123 x [456 x i32]], ptr [[A]], i64 0, i64 [[IDXPROM]]
482 // CHECK5-NEXT: [[TMP7:%.*]] = load i32, ptr [[J]], align 4, !llvm.access.group [[ACC_GRP2]]
483 // CHECK5-NEXT: [[IDXPROM7:%.*]] = sext i32 [[TMP7]] to i64
484 // CHECK5-NEXT: [[ARRAYIDX8:%.*]] = getelementptr inbounds [456 x i32], ptr [[ARRAYIDX]], i64 0, i64 [[IDXPROM7]]
485 // CHECK5-NEXT: store i32 0, ptr [[ARRAYIDX8]], align 4, !llvm.access.group [[ACC_GRP2]]
486 // CHECK5-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
487 // CHECK5: omp.body.continue:
488 // CHECK5-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
489 // CHECK5: omp.inner.for.inc:
490 // CHECK5-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP2]]
491 // CHECK5-NEXT: [[ADD9:%.*]] = add nsw i32 [[TMP8]], 1
492 // CHECK5-NEXT: store i32 [[ADD9]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP2]]
493 // CHECK5-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP3:![0-9]+]]
494 // CHECK5: omp.inner.for.end:
495 // CHECK5-NEXT: store i32 123, ptr [[I]], align 4
496 // CHECK5-NEXT: store i32 456, ptr [[J]], align 4
497 // CHECK5-NEXT: [[A10:%.*]] = getelementptr inbounds nuw [[STRUCT_SS]], ptr [[THIS1]], i32 0, i32 0
498 // CHECK5-NEXT: [[ARRAYIDX11:%.*]] = getelementptr inbounds [123 x [456 x i32]], ptr [[A10]], i64 0, i64 0
499 // CHECK5-NEXT: [[ARRAYIDX12:%.*]] = getelementptr inbounds [456 x i32], ptr [[ARRAYIDX11]], i64 0, i64 0
500 // CHECK5-NEXT: [[TMP9:%.*]] = load i32, ptr [[ARRAYIDX12]], align 4
501 // CHECK5-NEXT: ret i32 [[TMP9]]
502 //
503 //
504 // CHECK7-LABEL: define {{[^@]+}}@_Z21teams_template_structv
505 // CHECK7-SAME: () #[[ATTR0:[0-9]+]] {
506 // CHECK7-NEXT: entry:
507 // CHECK7-NEXT: [[V:%.*]] = alloca [[STRUCT_SS:%.*]], align 4
508 // CHECK7-NEXT: [[CALL:%.*]] = call noundef i32 @_ZN2SSIiLi123ELx456EE3fooEv(ptr noundef nonnull align 4 dereferenceable(224352) [[V]])
509 // CHECK7-NEXT: ret i32 [[CALL]]
510 //
511 //
512 // CHECK7-LABEL: define {{[^@]+}}@_ZN2SSIiLi123ELx456EE3fooEv
513 // CHECK7-SAME: (ptr noundef nonnull align 4 dereferenceable(224352) [[THIS:%.*]]) #[[ATTR0]] comdat align 2 {
514 // CHECK7-NEXT: entry:
515 // CHECK7-NEXT: [[THIS_ADDR:%.*]] = alloca ptr, align 4
516 // CHECK7-NEXT: [[TMP:%.*]] = alloca i32, align 4
517 // CHECK7-NEXT: [[_TMP2:%.*]] = alloca i32, align 4
518 // CHECK7-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4
519 // CHECK7-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4
520 // CHECK7-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
521 // CHECK7-NEXT: [[I:%.*]] = alloca i32, align 4
522 // CHECK7-NEXT: [[J:%.*]] = alloca i32, align 4
523 // CHECK7-NEXT: store ptr [[THIS]], ptr [[THIS_ADDR]], align 4
524 // CHECK7-NEXT: [[THIS1:%.*]] = load ptr, ptr [[THIS_ADDR]], align 4
525 // CHECK7-NEXT: store i32 0, ptr [[DOTOMP_LB]], align 4
526 // CHECK7-NEXT: store i32 56087, ptr [[DOTOMP_UB]], align 4
527 // CHECK7-NEXT: [[TMP0:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4
528 // CHECK7-NEXT: store i32 [[TMP0]], ptr [[DOTOMP_IV]], align 4
529 // CHECK7-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
530 // CHECK7: omp.inner.for.cond:
531 // CHECK7-NEXT: [[TMP1:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP3:![0-9]+]]
532 // CHECK7-NEXT: [[TMP2:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP3]]
533 // CHECK7-NEXT: [[CMP:%.*]] = icmp sle i32 [[TMP1]], [[TMP2]]
534 // CHECK7-NEXT: br i1 [[CMP]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
535 // CHECK7: omp.inner.for.body:
536 // CHECK7-NEXT: [[TMP3:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP3]]
537 // CHECK7-NEXT: [[DIV:%.*]] = sdiv i32 [[TMP3]], 456
538 // CHECK7-NEXT: [[MUL:%.*]] = mul nsw i32 [[DIV]], 1
539 // CHECK7-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
540 // CHECK7-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP3]]
541 // CHECK7-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP3]]
542 // CHECK7-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP3]]
543 // CHECK7-NEXT: [[DIV3:%.*]] = sdiv i32 [[TMP5]], 456
544 // CHECK7-NEXT: [[MUL4:%.*]] = mul nsw i32 [[DIV3]], 456
545 // CHECK7-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP4]], [[MUL4]]
546 // CHECK7-NEXT: [[MUL5:%.*]] = mul nsw i32 [[SUB]], 1
547 // CHECK7-NEXT: [[ADD6:%.*]] = add nsw i32 0, [[MUL5]]
548 // CHECK7-NEXT: store i32 [[ADD6]], ptr [[J]], align 4, !llvm.access.group [[ACC_GRP3]]
549 // CHECK7-NEXT: [[A:%.*]] = getelementptr inbounds nuw [[STRUCT_SS:%.*]], ptr [[THIS1]], i32 0, i32 0
550 // CHECK7-NEXT: [[TMP6:%.*]] = load i32, ptr [[I]], align 4, !llvm.access.group [[ACC_GRP3]]
551 // CHECK7-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds [123 x [456 x i32]], ptr [[A]], i32 0, i32 [[TMP6]]
552 // CHECK7-NEXT: [[TMP7:%.*]] = load i32, ptr [[J]], align 4, !llvm.access.group [[ACC_GRP3]]
553 // CHECK7-NEXT: [[ARRAYIDX7:%.*]] = getelementptr inbounds [456 x i32], ptr [[ARRAYIDX]], i32 0, i32 [[TMP7]]
554 // CHECK7-NEXT: store i32 0, ptr [[ARRAYIDX7]], align 4, !llvm.access.group [[ACC_GRP3]]
555 // CHECK7-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
556 // CHECK7: omp.body.continue:
557 // CHECK7-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
558 // CHECK7: omp.inner.for.inc:
559 // CHECK7-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP3]]
560 // CHECK7-NEXT: [[ADD8:%.*]] = add nsw i32 [[TMP8]], 1
561 // CHECK7-NEXT: store i32 [[ADD8]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP3]]
562 // CHECK7-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP4:![0-9]+]]
563 // CHECK7: omp.inner.for.end:
564 // CHECK7-NEXT: store i32 123, ptr [[I]], align 4
565 // CHECK7-NEXT: store i32 456, ptr [[J]], align 4
566 // CHECK7-NEXT: [[A9:%.*]] = getelementptr inbounds nuw [[STRUCT_SS]], ptr [[THIS1]], i32 0, i32 0
567 // CHECK7-NEXT: [[ARRAYIDX10:%.*]] = getelementptr inbounds [123 x [456 x i32]], ptr [[A9]], i32 0, i32 0
568 // CHECK7-NEXT: [[ARRAYIDX11:%.*]] = getelementptr inbounds [456 x i32], ptr [[ARRAYIDX10]], i32 0, i32 0
569 // CHECK7-NEXT: [[TMP9:%.*]] = load i32, ptr [[ARRAYIDX11]], align 4
570 // CHECK7-NEXT: ret i32 [[TMP9]]
571 //
572 //
573 // CHECK9-LABEL: define {{[^@]+}}@main
574 // CHECK9-SAME: (i32 noundef signext [[ARGC:%.*]], ptr noundef [[ARGV:%.*]]) #[[ATTR0:[0-9]+]] {
575 // CHECK9-NEXT: entry:
576 // CHECK9-NEXT: [[RETVAL:%.*]] = alloca i32, align 4
577 // CHECK9-NEXT: [[ARGC_ADDR:%.*]] = alloca i32, align 4
578 // CHECK9-NEXT: [[ARGV_ADDR:%.*]] = alloca ptr, align 8
579 // CHECK9-NEXT: [[N:%.*]] = alloca i32, align 4
580 // CHECK9-NEXT: [[M:%.*]] = alloca i32, align 4
581 // CHECK9-NEXT: [[SAVED_STACK:%.*]] = alloca ptr, align 8
582 // CHECK9-NEXT: [[__VLA_EXPR0:%.*]] = alloca i64, align 8
583 // CHECK9-NEXT: [[__VLA_EXPR1:%.*]] = alloca i64, align 8
584 // CHECK9-NEXT: [[N_CASTED:%.*]] = alloca i64, align 8
585 // CHECK9-NEXT: [[M_CASTED:%.*]] = alloca i64, align 8
586 // CHECK9-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [5 x ptr], align 8
587 // CHECK9-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [5 x ptr], align 8
588 // CHECK9-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [5 x ptr], align 8
589 // CHECK9-NEXT: [[DOTOFFLOAD_SIZES:%.*]] = alloca [5 x i64], align 8
590 // CHECK9-NEXT: [[TMP:%.*]] = alloca i32, align 4
591 // CHECK9-NEXT: [[_TMP1:%.*]] = alloca i32, align 4
592 // CHECK9-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4
593 // CHECK9-NEXT: [[DOTCAPTURE_EXPR_2:%.*]] = alloca i32, align 4
594 // CHECK9-NEXT: [[DOTCAPTURE_EXPR_3:%.*]] = alloca i64, align 8
595 // CHECK9-NEXT: [[KERNEL_ARGS:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS:%.*]], align 8
596 // CHECK9-NEXT: store i32 0, ptr [[RETVAL]], align 4
597 // CHECK9-NEXT: store i32 [[ARGC]], ptr [[ARGC_ADDR]], align 4
598 // CHECK9-NEXT: store ptr [[ARGV]], ptr [[ARGV_ADDR]], align 8
599 // CHECK9-NEXT: store i32 100, ptr [[N]], align 4
600 // CHECK9-NEXT: store i32 2, ptr [[M]], align 4
601 // CHECK9-NEXT: [[TMP0:%.*]] = load i32, ptr [[N]], align 4
602 // CHECK9-NEXT: [[TMP1:%.*]] = zext i32 [[TMP0]] to i64
603 // CHECK9-NEXT: [[TMP2:%.*]] = load i32, ptr [[M]], align 4
604 // CHECK9-NEXT: [[TMP3:%.*]] = zext i32 [[TMP2]] to i64
605 // CHECK9-NEXT: [[TMP4:%.*]] = call ptr @llvm.stacksave.p0()
606 // CHECK9-NEXT: store ptr [[TMP4]], ptr [[SAVED_STACK]], align 8
607 // CHECK9-NEXT: [[TMP5:%.*]] = mul nuw i64 [[TMP1]], [[TMP3]]
608 // CHECK9-NEXT: [[VLA:%.*]] = alloca i32, i64 [[TMP5]], align 4
609 // CHECK9-NEXT: store i64 [[TMP1]], ptr [[__VLA_EXPR0]], align 8
610 // CHECK9-NEXT: store i64 [[TMP3]], ptr [[__VLA_EXPR1]], align 8
611 // CHECK9-NEXT: [[TMP6:%.*]] = load i32, ptr [[N]], align 4
612 // CHECK9-NEXT: store i32 [[TMP6]], ptr [[N_CASTED]], align 4
613 // CHECK9-NEXT: [[TMP7:%.*]] = load i64, ptr [[N_CASTED]], align 8
614 // CHECK9-NEXT: [[TMP8:%.*]] = load i32, ptr [[M]], align 4
615 // CHECK9-NEXT: store i32 [[TMP8]], ptr [[M_CASTED]], align 4
616 // CHECK9-NEXT: [[TMP9:%.*]] = load i64, ptr [[M_CASTED]], align 8
617 // CHECK9-NEXT: [[TMP10:%.*]] = mul nuw i64 [[TMP1]], [[TMP3]]
618 // CHECK9-NEXT: [[TMP11:%.*]] = mul nuw i64 [[TMP10]], 4
619 // CHECK9-NEXT: call void @llvm.memcpy.p0.p0.i64(ptr align 8 [[DOTOFFLOAD_SIZES]], ptr align 8 @.offload_sizes, i64 40, i1 false)
620 // CHECK9-NEXT: [[TMP12:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
621 // CHECK9-NEXT: store i64 [[TMP7]], ptr [[TMP12]], align 8
622 // CHECK9-NEXT: [[TMP13:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
623 // CHECK9-NEXT: store i64 [[TMP7]], ptr [[TMP13]], align 8
624 // CHECK9-NEXT: [[TMP14:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0
625 // CHECK9-NEXT: store ptr null, ptr [[TMP14]], align 8
626 // CHECK9-NEXT: [[TMP15:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1
627 // CHECK9-NEXT: store i64 [[TMP9]], ptr [[TMP15]], align 8
628 // CHECK9-NEXT: [[TMP16:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 1
629 // CHECK9-NEXT: store i64 [[TMP9]], ptr [[TMP16]], align 8
630 // CHECK9-NEXT: [[TMP17:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i64 0, i64 1
631 // CHECK9-NEXT: store ptr null, ptr [[TMP17]], align 8
632 // CHECK9-NEXT: [[TMP18:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
633 // CHECK9-NEXT: store i64 [[TMP1]], ptr [[TMP18]], align 8
634 // CHECK9-NEXT: [[TMP19:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 2
635 // CHECK9-NEXT: store i64 [[TMP1]], ptr [[TMP19]], align 8
636 // CHECK9-NEXT: [[TMP20:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i64 0, i64 2
637 // CHECK9-NEXT: store ptr null, ptr [[TMP20]], align 8
638 // CHECK9-NEXT: [[TMP21:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 3
639 // CHECK9-NEXT: store i64 [[TMP3]], ptr [[TMP21]], align 8
640 // CHECK9-NEXT: [[TMP22:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 3
641 // CHECK9-NEXT: store i64 [[TMP3]], ptr [[TMP22]], align 8
642 // CHECK9-NEXT: [[TMP23:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i64 0, i64 3
643 // CHECK9-NEXT: store ptr null, ptr [[TMP23]], align 8
644 // CHECK9-NEXT: [[TMP24:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 4
645 // CHECK9-NEXT: store ptr [[VLA]], ptr [[TMP24]], align 8
646 // CHECK9-NEXT: [[TMP25:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 4
647 // CHECK9-NEXT: store ptr [[VLA]], ptr [[TMP25]], align 8
648 // CHECK9-NEXT: [[TMP26:%.*]] = getelementptr inbounds [5 x i64], ptr [[DOTOFFLOAD_SIZES]], i32 0, i32 4
649 // CHECK9-NEXT: store i64 [[TMP11]], ptr [[TMP26]], align 8
650 // CHECK9-NEXT: [[TMP27:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i64 0, i64 4
651 // CHECK9-NEXT: store ptr null, ptr [[TMP27]], align 8
652 // CHECK9-NEXT: [[TMP28:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
653 // CHECK9-NEXT: [[TMP29:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
654 // CHECK9-NEXT: [[TMP30:%.*]] = getelementptr inbounds [5 x i64], ptr [[DOTOFFLOAD_SIZES]], i32 0, i32 0
655 // CHECK9-NEXT: [[TMP31:%.*]] = load i32, ptr [[N]], align 4
656 // CHECK9-NEXT: store i32 [[TMP31]], ptr [[DOTCAPTURE_EXPR_]], align 4
657 // CHECK9-NEXT: [[TMP32:%.*]] = load i32, ptr [[M]], align 4
658 // CHECK9-NEXT: store i32 [[TMP32]], ptr [[DOTCAPTURE_EXPR_2]], align 4
659 // CHECK9-NEXT: [[TMP33:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4
660 // CHECK9-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP33]], 0
661 // CHECK9-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1
662 // CHECK9-NEXT: [[CONV:%.*]] = sext i32 [[DIV]] to i64
663 // CHECK9-NEXT: [[TMP34:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_2]], align 4
664 // CHECK9-NEXT: [[SUB4:%.*]] = sub nsw i32 [[TMP34]], 0
665 // CHECK9-NEXT: [[DIV5:%.*]] = sdiv i32 [[SUB4]], 1
666 // CHECK9-NEXT: [[CONV6:%.*]] = sext i32 [[DIV5]] to i64
667 // CHECK9-NEXT: [[MUL:%.*]] = mul nsw i64 [[CONV]], [[CONV6]]
668 // CHECK9-NEXT: [[SUB7:%.*]] = sub nsw i64 [[MUL]], 1
669 // CHECK9-NEXT: store i64 [[SUB7]], ptr [[DOTCAPTURE_EXPR_3]], align 8
670 // CHECK9-NEXT: [[TMP35:%.*]] = load i64, ptr [[DOTCAPTURE_EXPR_3]], align 8
671 // CHECK9-NEXT: [[ADD:%.*]] = add nsw i64 [[TMP35]], 1
672 // CHECK9-NEXT: [[TMP36:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 0
673 // CHECK9-NEXT: store i32 3, ptr [[TMP36]], align 4
674 // CHECK9-NEXT: [[TMP37:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 1
675 // CHECK9-NEXT: store i32 5, ptr [[TMP37]], align 4
676 // CHECK9-NEXT: [[TMP38:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 2
677 // CHECK9-NEXT: store ptr [[TMP28]], ptr [[TMP38]], align 8
678 // CHECK9-NEXT: [[TMP39:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 3
679 // CHECK9-NEXT: store ptr [[TMP29]], ptr [[TMP39]], align 8
680 // CHECK9-NEXT: [[TMP40:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 4
681 // CHECK9-NEXT: store ptr [[TMP30]], ptr [[TMP40]], align 8
682 // CHECK9-NEXT: [[TMP41:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 5
683 // CHECK9-NEXT: store ptr @.offload_maptypes, ptr [[TMP41]], align 8
684 // CHECK9-NEXT: [[TMP42:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 6
685 // CHECK9-NEXT: store ptr null, ptr [[TMP42]], align 8
686 // CHECK9-NEXT: [[TMP43:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 7
687 // CHECK9-NEXT: store ptr null, ptr [[TMP43]], align 8
688 // CHECK9-NEXT: [[TMP44:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 8
689 // CHECK9-NEXT: store i64 [[ADD]], ptr [[TMP44]], align 8
690 // CHECK9-NEXT: [[TMP45:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 9
691 // CHECK9-NEXT: store i64 0, ptr [[TMP45]], align 8
692 // CHECK9-NEXT: [[TMP46:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 10
693 // CHECK9-NEXT: store [3 x i32] zeroinitializer, ptr [[TMP46]], align 4
694 // CHECK9-NEXT: [[TMP47:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 11
695 // CHECK9-NEXT: store [3 x i32] [i32 1, i32 0, i32 0], ptr [[TMP47]], align 4
696 // CHECK9-NEXT: [[TMP48:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 12
697 // CHECK9-NEXT: store i32 0, ptr [[TMP48]], align 4
698 // CHECK9-NEXT: [[TMP49:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB2:[0-9]+]], i64 -1, i32 0, i32 1, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l80.region_id, ptr [[KERNEL_ARGS]])
699 // CHECK9-NEXT: [[TMP50:%.*]] = icmp ne i32 [[TMP49]], 0
700 // CHECK9-NEXT: br i1 [[TMP50]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
701 // CHECK9: omp_offload.failed:
702 // CHECK9-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l80(i64 [[TMP7]], i64 [[TMP9]], i64 [[TMP1]], i64 [[TMP3]], ptr [[VLA]]) #[[ATTR3:[0-9]+]]
703 // CHECK9-NEXT: br label [[OMP_OFFLOAD_CONT]]
704 // CHECK9: omp_offload.cont:
705 // CHECK9-NEXT: [[TMP51:%.*]] = load i32, ptr [[ARGC_ADDR]], align 4
706 // CHECK9-NEXT: [[CALL:%.*]] = call noundef signext i32 @_Z5tmainIiLi10ELi2EEiT_(i32 noundef signext [[TMP51]])
707 // CHECK9-NEXT: store i32 [[CALL]], ptr [[RETVAL]], align 4
708 // CHECK9-NEXT: [[TMP52:%.*]] = load ptr, ptr [[SAVED_STACK]], align 8
709 // CHECK9-NEXT: call void @llvm.stackrestore.p0(ptr [[TMP52]])
710 // CHECK9-NEXT: [[TMP53:%.*]] = load i32, ptr [[RETVAL]], align 4
711 // CHECK9-NEXT: ret i32 [[TMP53]]
712 //
713 //
714 // CHECK9-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l80
715 // CHECK9-SAME: (i64 noundef [[N:%.*]], i64 noundef [[M:%.*]], i64 noundef [[VLA:%.*]], i64 noundef [[VLA1:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[A:%.*]]) #[[ATTR2:[0-9]+]] {
716 // CHECK9-NEXT: entry:
717 // CHECK9-NEXT: [[N_ADDR:%.*]] = alloca i64, align 8
718 // CHECK9-NEXT: [[M_ADDR:%.*]] = alloca i64, align 8
719 // CHECK9-NEXT: [[VLA_ADDR:%.*]] = alloca i64, align 8
720 // CHECK9-NEXT: [[VLA_ADDR2:%.*]] = alloca i64, align 8
721 // CHECK9-NEXT: [[A_ADDR:%.*]] = alloca ptr, align 8
722 // CHECK9-NEXT: [[N_CASTED:%.*]] = alloca i64, align 8
723 // CHECK9-NEXT: [[M_CASTED:%.*]] = alloca i64, align 8
724 // CHECK9-NEXT: store i64 [[N]], ptr [[N_ADDR]], align 8
725 // CHECK9-NEXT: store i64 [[M]], ptr [[M_ADDR]], align 8
726 // CHECK9-NEXT: store i64 [[VLA]], ptr [[VLA_ADDR]], align 8
727 // CHECK9-NEXT: store i64 [[VLA1]], ptr [[VLA_ADDR2]], align 8
728 // CHECK9-NEXT: store ptr [[A]], ptr [[A_ADDR]], align 8
729 // CHECK9-NEXT: [[TMP0:%.*]] = load i64, ptr [[VLA_ADDR]], align 8
730 // CHECK9-NEXT: [[TMP1:%.*]] = load i64, ptr [[VLA_ADDR2]], align 8
731 // CHECK9-NEXT: [[TMP2:%.*]] = load ptr, ptr [[A_ADDR]], align 8
732 // CHECK9-NEXT: [[TMP3:%.*]] = load i32, ptr [[N_ADDR]], align 4
733 // CHECK9-NEXT: store i32 [[TMP3]], ptr [[N_CASTED]], align 4
734 // CHECK9-NEXT: [[TMP4:%.*]] = load i64, ptr [[N_CASTED]], align 8
735 // CHECK9-NEXT: [[TMP5:%.*]] = load i32, ptr [[M_ADDR]], align 4
736 // CHECK9-NEXT: store i32 [[TMP5]], ptr [[M_CASTED]], align 4
737 // CHECK9-NEXT: [[TMP6:%.*]] = load i64, ptr [[M_CASTED]], align 8
738 // CHECK9-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_teams(ptr @[[GLOB2]], i32 5, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l80.omp_outlined, i64 [[TMP4]], i64 [[TMP6]], i64 [[TMP0]], i64 [[TMP1]], ptr [[TMP2]])
739 // CHECK9-NEXT: ret void
740 //
741 //
742 // CHECK9-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l80.omp_outlined
743 // CHECK9-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], i64 noundef [[N:%.*]], i64 noundef [[M:%.*]], i64 noundef [[VLA:%.*]], i64 noundef [[VLA1:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[A:%.*]]) #[[ATTR2]] {
744 // CHECK9-NEXT: entry:
745 // CHECK9-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 8
746 // CHECK9-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 8
747 // CHECK9-NEXT: [[N_ADDR:%.*]] = alloca i64, align 8
748 // CHECK9-NEXT: [[M_ADDR:%.*]] = alloca i64, align 8
749 // CHECK9-NEXT: [[VLA_ADDR:%.*]] = alloca i64, align 8
750 // CHECK9-NEXT: [[VLA_ADDR2:%.*]] = alloca i64, align 8
751 // CHECK9-NEXT: [[A_ADDR:%.*]] = alloca ptr, align 8
752 // CHECK9-NEXT: [[DOTOMP_IV:%.*]] = alloca i64, align 8
753 // CHECK9-NEXT: [[TMP:%.*]] = alloca i32, align 4
754 // CHECK9-NEXT: [[_TMP3:%.*]] = alloca i32, align 4
755 // CHECK9-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4
756 // CHECK9-NEXT: [[DOTCAPTURE_EXPR_4:%.*]] = alloca i32, align 4
757 // CHECK9-NEXT: [[DOTCAPTURE_EXPR_5:%.*]] = alloca i64, align 8
758 // CHECK9-NEXT: [[I:%.*]] = alloca i32, align 4
759 // CHECK9-NEXT: [[J:%.*]] = alloca i32, align 4
760 // CHECK9-NEXT: [[DOTOMP_LB:%.*]] = alloca i64, align 8
761 // CHECK9-NEXT: [[DOTOMP_UB:%.*]] = alloca i64, align 8
762 // CHECK9-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i64, align 8
763 // CHECK9-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
764 // CHECK9-NEXT: [[I11:%.*]] = alloca i32, align 4
765 // CHECK9-NEXT: [[J12:%.*]] = alloca i32, align 4
766 // CHECK9-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 8
767 // CHECK9-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 8
768 // CHECK9-NEXT: store i64 [[N]], ptr [[N_ADDR]], align 8
769 // CHECK9-NEXT: store i64 [[M]], ptr [[M_ADDR]], align 8
770 // CHECK9-NEXT: store i64 [[VLA]], ptr [[VLA_ADDR]], align 8
771 // CHECK9-NEXT: store i64 [[VLA1]], ptr [[VLA_ADDR2]], align 8
772 // CHECK9-NEXT: store ptr [[A]], ptr [[A_ADDR]], align 8
773 // CHECK9-NEXT: [[TMP0:%.*]] = load i64, ptr [[VLA_ADDR]], align 8
774 // CHECK9-NEXT: [[TMP1:%.*]] = load i64, ptr [[VLA_ADDR2]], align 8
775 // CHECK9-NEXT: [[TMP2:%.*]] = load ptr, ptr [[A_ADDR]], align 8
776 // CHECK9-NEXT: [[TMP3:%.*]] = load i32, ptr [[N_ADDR]], align 4
777 // CHECK9-NEXT: store i32 [[TMP3]], ptr [[DOTCAPTURE_EXPR_]], align 4
778 // CHECK9-NEXT: [[TMP4:%.*]] = load i32, ptr [[M_ADDR]], align 4
779 // CHECK9-NEXT: store i32 [[TMP4]], ptr [[DOTCAPTURE_EXPR_4]], align 4
780 // CHECK9-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4
781 // CHECK9-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP5]], 0
782 // CHECK9-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1
783 // CHECK9-NEXT: [[CONV:%.*]] = sext i32 [[DIV]] to i64
784 // CHECK9-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_4]], align 4
785 // CHECK9-NEXT: [[SUB6:%.*]] = sub nsw i32 [[TMP6]], 0
786 // CHECK9-NEXT: [[DIV7:%.*]] = sdiv i32 [[SUB6]], 1
787 // CHECK9-NEXT: [[CONV8:%.*]] = sext i32 [[DIV7]] to i64
788 // CHECK9-NEXT: [[MUL:%.*]] = mul nsw i64 [[CONV]], [[CONV8]]
789 // CHECK9-NEXT: [[SUB9:%.*]] = sub nsw i64 [[MUL]], 1
790 // CHECK9-NEXT: store i64 [[SUB9]], ptr [[DOTCAPTURE_EXPR_5]], align 8
791 // CHECK9-NEXT: store i32 0, ptr [[I]], align 4
792 // CHECK9-NEXT: store i32 0, ptr [[J]], align 4
793 // CHECK9-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4
794 // CHECK9-NEXT: [[CMP:%.*]] = icmp slt i32 0, [[TMP7]]
795 // CHECK9-NEXT: br i1 [[CMP]], label [[LAND_LHS_TRUE:%.*]], label [[OMP_PRECOND_END:%.*]]
796 // CHECK9: land.lhs.true:
797 // CHECK9-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_4]], align 4
798 // CHECK9-NEXT: [[CMP10:%.*]] = icmp slt i32 0, [[TMP8]]
799 // CHECK9-NEXT: br i1 [[CMP10]], label [[OMP_PRECOND_THEN:%.*]], label [[OMP_PRECOND_END]]
800 // CHECK9: omp.precond.then:
801 // CHECK9-NEXT: store i64 0, ptr [[DOTOMP_LB]], align 8
802 // CHECK9-NEXT: [[TMP9:%.*]] = load i64, ptr [[DOTCAPTURE_EXPR_5]], align 8
803 // CHECK9-NEXT: store i64 [[TMP9]], ptr [[DOTOMP_UB]], align 8
804 // CHECK9-NEXT: store i64 1, ptr [[DOTOMP_STRIDE]], align 8
805 // CHECK9-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
806 // CHECK9-NEXT: [[TMP10:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
807 // CHECK9-NEXT: [[TMP11:%.*]] = load i32, ptr [[TMP10]], align 4
808 // CHECK9-NEXT: call void @__kmpc_for_static_init_8(ptr @[[GLOB1:[0-9]+]], i32 [[TMP11]], i32 92, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i64 1, i64 1)
809 // CHECK9-NEXT: [[TMP12:%.*]] = load i64, ptr [[DOTOMP_UB]], align 8
810 // CHECK9-NEXT: [[TMP13:%.*]] = load i64, ptr [[DOTCAPTURE_EXPR_5]], align 8
811 // CHECK9-NEXT: [[CMP13:%.*]] = icmp sgt i64 [[TMP12]], [[TMP13]]
812 // CHECK9-NEXT: br i1 [[CMP13]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
813 // CHECK9: cond.true:
814 // CHECK9-NEXT: [[TMP14:%.*]] = load i64, ptr [[DOTCAPTURE_EXPR_5]], align 8
815 // CHECK9-NEXT: br label [[COND_END:%.*]]
816 // CHECK9: cond.false:
817 // CHECK9-NEXT: [[TMP15:%.*]] = load i64, ptr [[DOTOMP_UB]], align 8
818 // CHECK9-NEXT: br label [[COND_END]]
819 // CHECK9: cond.end:
820 // CHECK9-NEXT: [[COND:%.*]] = phi i64 [ [[TMP14]], [[COND_TRUE]] ], [ [[TMP15]], [[COND_FALSE]] ]
821 // CHECK9-NEXT: store i64 [[COND]], ptr [[DOTOMP_UB]], align 8
822 // CHECK9-NEXT: [[TMP16:%.*]] = load i64, ptr [[DOTOMP_LB]], align 8
823 // CHECK9-NEXT: store i64 [[TMP16]], ptr [[DOTOMP_IV]], align 8
824 // CHECK9-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
825 // CHECK9: omp.inner.for.cond:
826 // CHECK9-NEXT: [[TMP17:%.*]] = load i64, ptr [[DOTOMP_IV]], align 8, !llvm.access.group [[ACC_GRP5:![0-9]+]]
827 // CHECK9-NEXT: [[TMP18:%.*]] = load i64, ptr [[DOTOMP_UB]], align 8, !llvm.access.group [[ACC_GRP5]]
828 // CHECK9-NEXT: [[CMP14:%.*]] = icmp sle i64 [[TMP17]], [[TMP18]]
829 // CHECK9-NEXT: br i1 [[CMP14]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
830 // CHECK9: omp.inner.for.body:
831 // CHECK9-NEXT: [[TMP19:%.*]] = load i64, ptr [[DOTOMP_IV]], align 8, !llvm.access.group [[ACC_GRP5]]
832 // CHECK9-NEXT: [[TMP20:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_4]], align 4, !llvm.access.group [[ACC_GRP5]]
833 // CHECK9-NEXT: [[SUB15:%.*]] = sub nsw i32 [[TMP20]], 0
834 // CHECK9-NEXT: [[DIV16:%.*]] = sdiv i32 [[SUB15]], 1
835 // CHECK9-NEXT: [[MUL17:%.*]] = mul nsw i32 1, [[DIV16]]
836 // CHECK9-NEXT: [[CONV18:%.*]] = sext i32 [[MUL17]] to i64
837 // CHECK9-NEXT: [[DIV19:%.*]] = sdiv i64 [[TMP19]], [[CONV18]]
838 // CHECK9-NEXT: [[MUL20:%.*]] = mul nsw i64 [[DIV19]], 1
839 // CHECK9-NEXT: [[ADD:%.*]] = add nsw i64 0, [[MUL20]]
840 // CHECK9-NEXT: [[CONV21:%.*]] = trunc i64 [[ADD]] to i32
841 // CHECK9-NEXT: store i32 [[CONV21]], ptr [[I11]], align 4, !llvm.access.group [[ACC_GRP5]]
842 // CHECK9-NEXT: [[TMP21:%.*]] = load i64, ptr [[DOTOMP_IV]], align 8, !llvm.access.group [[ACC_GRP5]]
843 // CHECK9-NEXT: [[TMP22:%.*]] = load i64, ptr [[DOTOMP_IV]], align 8, !llvm.access.group [[ACC_GRP5]]
844 // CHECK9-NEXT: [[TMP23:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_4]], align 4, !llvm.access.group [[ACC_GRP5]]
845 // CHECK9-NEXT: [[SUB22:%.*]] = sub nsw i32 [[TMP23]], 0
846 // CHECK9-NEXT: [[DIV23:%.*]] = sdiv i32 [[SUB22]], 1
847 // CHECK9-NEXT: [[MUL24:%.*]] = mul nsw i32 1, [[DIV23]]
848 // CHECK9-NEXT: [[CONV25:%.*]] = sext i32 [[MUL24]] to i64
849 // CHECK9-NEXT: [[DIV26:%.*]] = sdiv i64 [[TMP22]], [[CONV25]]
850 // CHECK9-NEXT: [[TMP24:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_4]], align 4, !llvm.access.group [[ACC_GRP5]]
851 // CHECK9-NEXT: [[SUB27:%.*]] = sub nsw i32 [[TMP24]], 0
852 // CHECK9-NEXT: [[DIV28:%.*]] = sdiv i32 [[SUB27]], 1
853 // CHECK9-NEXT: [[MUL29:%.*]] = mul nsw i32 1, [[DIV28]]
854 // CHECK9-NEXT: [[CONV30:%.*]] = sext i32 [[MUL29]] to i64
855 // CHECK9-NEXT: [[MUL31:%.*]] = mul nsw i64 [[DIV26]], [[CONV30]]
856 // CHECK9-NEXT: [[SUB32:%.*]] = sub nsw i64 [[TMP21]], [[MUL31]]
857 // CHECK9-NEXT: [[MUL33:%.*]] = mul nsw i64 [[SUB32]], 1
858 // CHECK9-NEXT: [[ADD34:%.*]] = add nsw i64 0, [[MUL33]]
859 // CHECK9-NEXT: [[CONV35:%.*]] = trunc i64 [[ADD34]] to i32
860 // CHECK9-NEXT: store i32 [[CONV35]], ptr [[J12]], align 4, !llvm.access.group [[ACC_GRP5]]
861 // CHECK9-NEXT: [[TMP25:%.*]] = load i32, ptr [[I11]], align 4, !llvm.access.group [[ACC_GRP5]]
862 // CHECK9-NEXT: [[IDXPROM:%.*]] = sext i32 [[TMP25]] to i64
863 // CHECK9-NEXT: [[TMP26:%.*]] = mul nsw i64 [[IDXPROM]], [[TMP1]]
864 // CHECK9-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i32, ptr [[TMP2]], i64 [[TMP26]]
865 // CHECK9-NEXT: [[TMP27:%.*]] = load i32, ptr [[J12]], align 4, !llvm.access.group [[ACC_GRP5]]
866 // CHECK9-NEXT: [[IDXPROM36:%.*]] = sext i32 [[TMP27]] to i64
867 // CHECK9-NEXT: [[ARRAYIDX37:%.*]] = getelementptr inbounds i32, ptr [[ARRAYIDX]], i64 [[IDXPROM36]]
868 // CHECK9-NEXT: store i32 0, ptr [[ARRAYIDX37]], align 4, !llvm.access.group [[ACC_GRP5]]
869 // CHECK9-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
870 // CHECK9: omp.body.continue:
871 // CHECK9-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
872 // CHECK9: omp.inner.for.inc:
873 // CHECK9-NEXT: [[TMP28:%.*]] = load i64, ptr [[DOTOMP_IV]], align 8, !llvm.access.group [[ACC_GRP5]]
874 // CHECK9-NEXT: [[ADD38:%.*]] = add nsw i64 [[TMP28]], 1
875 // CHECK9-NEXT: store i64 [[ADD38]], ptr [[DOTOMP_IV]], align 8, !llvm.access.group [[ACC_GRP5]]
876 // CHECK9-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP6:![0-9]+]]
877 // CHECK9: omp.inner.for.end:
878 // CHECK9-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
879 // CHECK9: omp.loop.exit:
880 // CHECK9-NEXT: [[TMP29:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
881 // CHECK9-NEXT: [[TMP30:%.*]] = load i32, ptr [[TMP29]], align 4
882 // CHECK9-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB1]], i32 [[TMP30]])
883 // CHECK9-NEXT: [[TMP31:%.*]] = load i32, ptr [[DOTOMP_IS_LAST]], align 4
884 // CHECK9-NEXT: [[TMP32:%.*]] = icmp ne i32 [[TMP31]], 0
885 // CHECK9-NEXT: br i1 [[TMP32]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
886 // CHECK9: .omp.final.then:
887 // CHECK9-NEXT: [[TMP33:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4
888 // CHECK9-NEXT: [[SUB39:%.*]] = sub nsw i32 [[TMP33]], 0
889 // CHECK9-NEXT: [[DIV40:%.*]] = sdiv i32 [[SUB39]], 1
890 // CHECK9-NEXT: [[MUL41:%.*]] = mul nsw i32 [[DIV40]], 1
891 // CHECK9-NEXT: [[ADD42:%.*]] = add nsw i32 0, [[MUL41]]
892 // CHECK9-NEXT: store i32 [[ADD42]], ptr [[I11]], align 4
893 // CHECK9-NEXT: [[TMP34:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_4]], align 4
894 // CHECK9-NEXT: [[SUB43:%.*]] = sub nsw i32 [[TMP34]], 0
895 // CHECK9-NEXT: [[DIV44:%.*]] = sdiv i32 [[SUB43]], 1
896 // CHECK9-NEXT: [[MUL45:%.*]] = mul nsw i32 [[DIV44]], 1
897 // CHECK9-NEXT: [[ADD46:%.*]] = add nsw i32 0, [[MUL45]]
898 // CHECK9-NEXT: store i32 [[ADD46]], ptr [[J12]], align 4
899 // CHECK9-NEXT: br label [[DOTOMP_FINAL_DONE]]
900 // CHECK9: .omp.final.done:
901 // CHECK9-NEXT: br label [[OMP_PRECOND_END]]
902 // CHECK9: omp.precond.end:
903 // CHECK9-NEXT: ret void
904 //
905 //
906 // CHECK9-LABEL: define {{[^@]+}}@_Z5tmainIiLi10ELi2EEiT_
907 // CHECK9-SAME: (i32 noundef signext [[ARGC:%.*]]) #[[ATTR5:[0-9]+]] comdat {
908 // CHECK9-NEXT: entry:
909 // CHECK9-NEXT: [[ARGC_ADDR:%.*]] = alloca i32, align 4
910 // CHECK9-NEXT: [[A:%.*]] = alloca [10 x [2 x i32]], align 4
911 // CHECK9-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [1 x ptr], align 8
912 // CHECK9-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [1 x ptr], align 8
913 // CHECK9-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [1 x ptr], align 8
914 // CHECK9-NEXT: [[TMP:%.*]] = alloca i32, align 4
915 // CHECK9-NEXT: [[_TMP1:%.*]] = alloca i32, align 4
916 // CHECK9-NEXT: [[KERNEL_ARGS:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS:%.*]], align 8
917 // CHECK9-NEXT: store i32 [[ARGC]], ptr [[ARGC_ADDR]], align 4
918 // CHECK9-NEXT: [[TMP0:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
919 // CHECK9-NEXT: store ptr [[A]], ptr [[TMP0]], align 8
920 // CHECK9-NEXT: [[TMP1:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
921 // CHECK9-NEXT: store ptr [[A]], ptr [[TMP1]], align 8
922 // CHECK9-NEXT: [[TMP2:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0
923 // CHECK9-NEXT: store ptr null, ptr [[TMP2]], align 8
924 // CHECK9-NEXT: [[TMP3:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
925 // CHECK9-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
926 // CHECK9-NEXT: [[TMP5:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 0
927 // CHECK9-NEXT: store i32 3, ptr [[TMP5]], align 4
928 // CHECK9-NEXT: [[TMP6:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 1
929 // CHECK9-NEXT: store i32 1, ptr [[TMP6]], align 4
930 // CHECK9-NEXT: [[TMP7:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 2
931 // CHECK9-NEXT: store ptr [[TMP3]], ptr [[TMP7]], align 8
932 // CHECK9-NEXT: [[TMP8:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 3
933 // CHECK9-NEXT: store ptr [[TMP4]], ptr [[TMP8]], align 8
934 // CHECK9-NEXT: [[TMP9:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 4
935 // CHECK9-NEXT: store ptr @.offload_sizes.1, ptr [[TMP9]], align 8
936 // CHECK9-NEXT: [[TMP10:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 5
937 // CHECK9-NEXT: store ptr @.offload_maptypes.2, ptr [[TMP10]], align 8
938 // CHECK9-NEXT: [[TMP11:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 6
939 // CHECK9-NEXT: store ptr null, ptr [[TMP11]], align 8
940 // CHECK9-NEXT: [[TMP12:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 7
941 // CHECK9-NEXT: store ptr null, ptr [[TMP12]], align 8
942 // CHECK9-NEXT: [[TMP13:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 8
943 // CHECK9-NEXT: store i64 20, ptr [[TMP13]], align 8
944 // CHECK9-NEXT: [[TMP14:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 9
945 // CHECK9-NEXT: store i64 0, ptr [[TMP14]], align 8
946 // CHECK9-NEXT: [[TMP15:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 10
947 // CHECK9-NEXT: store [3 x i32] zeroinitializer, ptr [[TMP15]], align 4
948 // CHECK9-NEXT: [[TMP16:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 11
949 // CHECK9-NEXT: store [3 x i32] [i32 1, i32 0, i32 0], ptr [[TMP16]], align 4
950 // CHECK9-NEXT: [[TMP17:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 12
951 // CHECK9-NEXT: store i32 0, ptr [[TMP17]], align 4
952 // CHECK9-NEXT: [[TMP18:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB2]], i64 -1, i32 0, i32 1, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10ELi2EEiT__l67.region_id, ptr [[KERNEL_ARGS]])
953 // CHECK9-NEXT: [[TMP19:%.*]] = icmp ne i32 [[TMP18]], 0
954 // CHECK9-NEXT: br i1 [[TMP19]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
955 // CHECK9: omp_offload.failed:
956 // CHECK9-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10ELi2EEiT__l67(ptr [[A]]) #[[ATTR3]]
957 // CHECK9-NEXT: br label [[OMP_OFFLOAD_CONT]]
958 // CHECK9: omp_offload.cont:
959 // CHECK9-NEXT: ret i32 0
960 //
961 //
962 // CHECK9-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10ELi2EEiT__l67
963 // CHECK9-SAME: (ptr noundef nonnull align 4 dereferenceable(80) [[A:%.*]]) #[[ATTR2]] {
964 // CHECK9-NEXT: entry:
965 // CHECK9-NEXT: [[A_ADDR:%.*]] = alloca ptr, align 8
966 // CHECK9-NEXT: store ptr [[A]], ptr [[A_ADDR]], align 8
967 // CHECK9-NEXT: [[TMP0:%.*]] = load ptr, ptr [[A_ADDR]], align 8
968 // CHECK9-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_teams(ptr @[[GLOB2]], i32 1, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10ELi2EEiT__l67.omp_outlined, ptr [[TMP0]])
969 // CHECK9-NEXT: ret void
970 //
971 //
972 // CHECK9-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10ELi2EEiT__l67.omp_outlined
973 // CHECK9-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], ptr noundef nonnull align 4 dereferenceable(80) [[A:%.*]]) #[[ATTR2]] {
974 // CHECK9-NEXT: entry:
975 // CHECK9-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 8
976 // CHECK9-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 8
977 // CHECK9-NEXT: [[A_ADDR:%.*]] = alloca ptr, align 8
978 // CHECK9-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
979 // CHECK9-NEXT: [[TMP:%.*]] = alloca i32, align 4
980 // CHECK9-NEXT: [[_TMP1:%.*]] = alloca i32, align 4
981 // CHECK9-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4
982 // CHECK9-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4
983 // CHECK9-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
984 // CHECK9-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
985 // CHECK9-NEXT: [[I:%.*]] = alloca i32, align 4
986 // CHECK9-NEXT: [[J:%.*]] = alloca i32, align 4
987 // CHECK9-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 8
988 // CHECK9-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 8
989 // CHECK9-NEXT: store ptr [[A]], ptr [[A_ADDR]], align 8
990 // CHECK9-NEXT: [[TMP0:%.*]] = load ptr, ptr [[A_ADDR]], align 8
991 // CHECK9-NEXT: store i32 0, ptr [[DOTOMP_LB]], align 4
992 // CHECK9-NEXT: store i32 19, ptr [[DOTOMP_UB]], align 4
993 // CHECK9-NEXT: store i32 1, ptr [[DOTOMP_STRIDE]], align 4
994 // CHECK9-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
995 // CHECK9-NEXT: [[TMP1:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
996 // CHECK9-NEXT: [[TMP2:%.*]] = load i32, ptr [[TMP1]], align 4
997 // CHECK9-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB1]], i32 [[TMP2]], i32 92, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
998 // CHECK9-NEXT: [[TMP3:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
999 // CHECK9-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP3]], 19
1000 // CHECK9-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
1001 // CHECK9: cond.true:
1002 // CHECK9-NEXT: br label [[COND_END:%.*]]
1003 // CHECK9: cond.false:
1004 // CHECK9-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
1005 // CHECK9-NEXT: br label [[COND_END]]
1006 // CHECK9: cond.end:
1007 // CHECK9-NEXT: [[COND:%.*]] = phi i32 [ 19, [[COND_TRUE]] ], [ [[TMP4]], [[COND_FALSE]] ]
1008 // CHECK9-NEXT: store i32 [[COND]], ptr [[DOTOMP_UB]], align 4
1009 // CHECK9-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4
1010 // CHECK9-NEXT: store i32 [[TMP5]], ptr [[DOTOMP_IV]], align 4
1011 // CHECK9-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
1012 // CHECK9: omp.inner.for.cond:
1013 // CHECK9-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP11:![0-9]+]]
1014 // CHECK9-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP11]]
1015 // CHECK9-NEXT: [[CMP2:%.*]] = icmp sle i32 [[TMP6]], [[TMP7]]
1016 // CHECK9-NEXT: br i1 [[CMP2]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
1017 // CHECK9: omp.inner.for.body:
1018 // CHECK9-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP11]]
1019 // CHECK9-NEXT: [[DIV:%.*]] = sdiv i32 [[TMP8]], 2
1020 // CHECK9-NEXT: [[MUL:%.*]] = mul nsw i32 [[DIV]], 1
1021 // CHECK9-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
1022 // CHECK9-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP11]]
1023 // CHECK9-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP11]]
1024 // CHECK9-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP11]]
1025 // CHECK9-NEXT: [[DIV3:%.*]] = sdiv i32 [[TMP10]], 2
1026 // CHECK9-NEXT: [[MUL4:%.*]] = mul nsw i32 [[DIV3]], 2
1027 // CHECK9-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP9]], [[MUL4]]
1028 // CHECK9-NEXT: [[MUL5:%.*]] = mul nsw i32 [[SUB]], 1
1029 // CHECK9-NEXT: [[ADD6:%.*]] = add nsw i32 0, [[MUL5]]
1030 // CHECK9-NEXT: store i32 [[ADD6]], ptr [[J]], align 4, !llvm.access.group [[ACC_GRP11]]
1031 // CHECK9-NEXT: [[TMP11:%.*]] = load i32, ptr [[I]], align 4, !llvm.access.group [[ACC_GRP11]]
1032 // CHECK9-NEXT: [[IDXPROM:%.*]] = sext i32 [[TMP11]] to i64
1033 // CHECK9-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds [10 x [2 x i32]], ptr [[TMP0]], i64 0, i64 [[IDXPROM]]
1034 // CHECK9-NEXT: [[TMP12:%.*]] = load i32, ptr [[J]], align 4, !llvm.access.group [[ACC_GRP11]]
1035 // CHECK9-NEXT: [[IDXPROM7:%.*]] = sext i32 [[TMP12]] to i64
1036 // CHECK9-NEXT: [[ARRAYIDX8:%.*]] = getelementptr inbounds [2 x i32], ptr [[ARRAYIDX]], i64 0, i64 [[IDXPROM7]]
1037 // CHECK9-NEXT: store i32 0, ptr [[ARRAYIDX8]], align 4, !llvm.access.group [[ACC_GRP11]]
1038 // CHECK9-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
1039 // CHECK9: omp.body.continue:
1040 // CHECK9-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
1041 // CHECK9: omp.inner.for.inc:
1042 // CHECK9-NEXT: [[TMP13:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP11]]
1043 // CHECK9-NEXT: [[ADD9:%.*]] = add nsw i32 [[TMP13]], 1
1044 // CHECK9-NEXT: store i32 [[ADD9]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP11]]
1045 // CHECK9-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP12:![0-9]+]]
1046 // CHECK9: omp.inner.for.end:
1047 // CHECK9-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
1048 // CHECK9: omp.loop.exit:
1049 // CHECK9-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB1]], i32 [[TMP2]])
1050 // CHECK9-NEXT: [[TMP14:%.*]] = load i32, ptr [[DOTOMP_IS_LAST]], align 4
1051 // CHECK9-NEXT: [[TMP15:%.*]] = icmp ne i32 [[TMP14]], 0
1052 // CHECK9-NEXT: br i1 [[TMP15]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
1053 // CHECK9: .omp.final.then:
1054 // CHECK9-NEXT: store i32 10, ptr [[I]], align 4
1055 // CHECK9-NEXT: store i32 2, ptr [[J]], align 4
1056 // CHECK9-NEXT: br label [[DOTOMP_FINAL_DONE]]
1057 // CHECK9: .omp.final.done:
1058 // CHECK9-NEXT: ret void
1059 //
1060 //
1061 // CHECK11-LABEL: define {{[^@]+}}@main
1062 // CHECK11-SAME: (i32 noundef [[ARGC:%.*]], ptr noundef [[ARGV:%.*]]) #[[ATTR0:[0-9]+]] {
1063 // CHECK11-NEXT: entry:
1064 // CHECK11-NEXT: [[RETVAL:%.*]] = alloca i32, align 4
1065 // CHECK11-NEXT: [[ARGC_ADDR:%.*]] = alloca i32, align 4
1066 // CHECK11-NEXT: [[ARGV_ADDR:%.*]] = alloca ptr, align 4
1067 // CHECK11-NEXT: [[N:%.*]] = alloca i32, align 4
1068 // CHECK11-NEXT: [[M:%.*]] = alloca i32, align 4
1069 // CHECK11-NEXT: [[SAVED_STACK:%.*]] = alloca ptr, align 4
1070 // CHECK11-NEXT: [[__VLA_EXPR0:%.*]] = alloca i32, align 4
1071 // CHECK11-NEXT: [[__VLA_EXPR1:%.*]] = alloca i32, align 4
1072 // CHECK11-NEXT: [[N_CASTED:%.*]] = alloca i32, align 4
1073 // CHECK11-NEXT: [[M_CASTED:%.*]] = alloca i32, align 4
1074 // CHECK11-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [5 x ptr], align 4
1075 // CHECK11-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [5 x ptr], align 4
1076 // CHECK11-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [5 x ptr], align 4
1077 // CHECK11-NEXT: [[DOTOFFLOAD_SIZES:%.*]] = alloca [5 x i64], align 4
1078 // CHECK11-NEXT: [[TMP:%.*]] = alloca i32, align 4
1079 // CHECK11-NEXT: [[_TMP1:%.*]] = alloca i32, align 4
1080 // CHECK11-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4
1081 // CHECK11-NEXT: [[DOTCAPTURE_EXPR_2:%.*]] = alloca i32, align 4
1082 // CHECK11-NEXT: [[DOTCAPTURE_EXPR_3:%.*]] = alloca i64, align 8
1083 // CHECK11-NEXT: [[KERNEL_ARGS:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS:%.*]], align 8
1084 // CHECK11-NEXT: store i32 0, ptr [[RETVAL]], align 4
1085 // CHECK11-NEXT: store i32 [[ARGC]], ptr [[ARGC_ADDR]], align 4
1086 // CHECK11-NEXT: store ptr [[ARGV]], ptr [[ARGV_ADDR]], align 4
1087 // CHECK11-NEXT: store i32 100, ptr [[N]], align 4
1088 // CHECK11-NEXT: store i32 2, ptr [[M]], align 4
1089 // CHECK11-NEXT: [[TMP0:%.*]] = load i32, ptr [[N]], align 4
1090 // CHECK11-NEXT: [[TMP1:%.*]] = load i32, ptr [[M]], align 4
1091 // CHECK11-NEXT: [[TMP2:%.*]] = call ptr @llvm.stacksave.p0()
1092 // CHECK11-NEXT: store ptr [[TMP2]], ptr [[SAVED_STACK]], align 4
1093 // CHECK11-NEXT: [[TMP3:%.*]] = mul nuw i32 [[TMP0]], [[TMP1]]
1094 // CHECK11-NEXT: [[VLA:%.*]] = alloca i32, i32 [[TMP3]], align 4
1095 // CHECK11-NEXT: store i32 [[TMP0]], ptr [[__VLA_EXPR0]], align 4
1096 // CHECK11-NEXT: store i32 [[TMP1]], ptr [[__VLA_EXPR1]], align 4
1097 // CHECK11-NEXT: [[TMP4:%.*]] = load i32, ptr [[N]], align 4
1098 // CHECK11-NEXT: store i32 [[TMP4]], ptr [[N_CASTED]], align 4
1099 // CHECK11-NEXT: [[TMP5:%.*]] = load i32, ptr [[N_CASTED]], align 4
1100 // CHECK11-NEXT: [[TMP6:%.*]] = load i32, ptr [[M]], align 4
1101 // CHECK11-NEXT: store i32 [[TMP6]], ptr [[M_CASTED]], align 4
1102 // CHECK11-NEXT: [[TMP7:%.*]] = load i32, ptr [[M_CASTED]], align 4
1103 // CHECK11-NEXT: [[TMP8:%.*]] = mul nuw i32 [[TMP0]], [[TMP1]]
1104 // CHECK11-NEXT: [[TMP9:%.*]] = mul nuw i32 [[TMP8]], 4
1105 // CHECK11-NEXT: [[TMP10:%.*]] = sext i32 [[TMP9]] to i64
1106 // CHECK11-NEXT: call void @llvm.memcpy.p0.p0.i32(ptr align 4 [[DOTOFFLOAD_SIZES]], ptr align 4 @.offload_sizes, i32 40, i1 false)
1107 // CHECK11-NEXT: [[TMP11:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
1108 // CHECK11-NEXT: store i32 [[TMP5]], ptr [[TMP11]], align 4
1109 // CHECK11-NEXT: [[TMP12:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
1110 // CHECK11-NEXT: store i32 [[TMP5]], ptr [[TMP12]], align 4
1111 // CHECK11-NEXT: [[TMP13:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0
1112 // CHECK11-NEXT: store ptr null, ptr [[TMP13]], align 4
1113 // CHECK11-NEXT: [[TMP14:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1
1114 // CHECK11-NEXT: store i32 [[TMP7]], ptr [[TMP14]], align 4
1115 // CHECK11-NEXT: [[TMP15:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 1
1116 // CHECK11-NEXT: store i32 [[TMP7]], ptr [[TMP15]], align 4
1117 // CHECK11-NEXT: [[TMP16:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i32 0, i32 1
1118 // CHECK11-NEXT: store ptr null, ptr [[TMP16]], align 4
1119 // CHECK11-NEXT: [[TMP17:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
1120 // CHECK11-NEXT: store i32 [[TMP0]], ptr [[TMP17]], align 4
1121 // CHECK11-NEXT: [[TMP18:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 2
1122 // CHECK11-NEXT: store i32 [[TMP0]], ptr [[TMP18]], align 4
1123 // CHECK11-NEXT: [[TMP19:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i32 0, i32 2
1124 // CHECK11-NEXT: store ptr null, ptr [[TMP19]], align 4
1125 // CHECK11-NEXT: [[TMP20:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 3
1126 // CHECK11-NEXT: store i32 [[TMP1]], ptr [[TMP20]], align 4
1127 // CHECK11-NEXT: [[TMP21:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 3
1128 // CHECK11-NEXT: store i32 [[TMP1]], ptr [[TMP21]], align 4
1129 // CHECK11-NEXT: [[TMP22:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i32 0, i32 3
1130 // CHECK11-NEXT: store ptr null, ptr [[TMP22]], align 4
1131 // CHECK11-NEXT: [[TMP23:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 4
1132 // CHECK11-NEXT: store ptr [[VLA]], ptr [[TMP23]], align 4
1133 // CHECK11-NEXT: [[TMP24:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 4
1134 // CHECK11-NEXT: store ptr [[VLA]], ptr [[TMP24]], align 4
1135 // CHECK11-NEXT: [[TMP25:%.*]] = getelementptr inbounds [5 x i64], ptr [[DOTOFFLOAD_SIZES]], i32 0, i32 4
1136 // CHECK11-NEXT: store i64 [[TMP10]], ptr [[TMP25]], align 4
1137 // CHECK11-NEXT: [[TMP26:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i32 0, i32 4
1138 // CHECK11-NEXT: store ptr null, ptr [[TMP26]], align 4
1139 // CHECK11-NEXT: [[TMP27:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
1140 // CHECK11-NEXT: [[TMP28:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
1141 // CHECK11-NEXT: [[TMP29:%.*]] = getelementptr inbounds [5 x i64], ptr [[DOTOFFLOAD_SIZES]], i32 0, i32 0
1142 // CHECK11-NEXT: [[TMP30:%.*]] = load i32, ptr [[N]], align 4
1143 // CHECK11-NEXT: store i32 [[TMP30]], ptr [[DOTCAPTURE_EXPR_]], align 4
1144 // CHECK11-NEXT: [[TMP31:%.*]] = load i32, ptr [[M]], align 4
1145 // CHECK11-NEXT: store i32 [[TMP31]], ptr [[DOTCAPTURE_EXPR_2]], align 4
1146 // CHECK11-NEXT: [[TMP32:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4
1147 // CHECK11-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP32]], 0
1148 // CHECK11-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1
1149 // CHECK11-NEXT: [[CONV:%.*]] = sext i32 [[DIV]] to i64
1150 // CHECK11-NEXT: [[TMP33:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_2]], align 4
1151 // CHECK11-NEXT: [[SUB4:%.*]] = sub nsw i32 [[TMP33]], 0
1152 // CHECK11-NEXT: [[DIV5:%.*]] = sdiv i32 [[SUB4]], 1
1153 // CHECK11-NEXT: [[CONV6:%.*]] = sext i32 [[DIV5]] to i64
1154 // CHECK11-NEXT: [[MUL:%.*]] = mul nsw i64 [[CONV]], [[CONV6]]
1155 // CHECK11-NEXT: [[SUB7:%.*]] = sub nsw i64 [[MUL]], 1
1156 // CHECK11-NEXT: store i64 [[SUB7]], ptr [[DOTCAPTURE_EXPR_3]], align 8
1157 // CHECK11-NEXT: [[TMP34:%.*]] = load i64, ptr [[DOTCAPTURE_EXPR_3]], align 8
1158 // CHECK11-NEXT: [[ADD:%.*]] = add nsw i64 [[TMP34]], 1
1159 // CHECK11-NEXT: [[TMP35:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 0
1160 // CHECK11-NEXT: store i32 3, ptr [[TMP35]], align 4
1161 // CHECK11-NEXT: [[TMP36:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 1
1162 // CHECK11-NEXT: store i32 5, ptr [[TMP36]], align 4
1163 // CHECK11-NEXT: [[TMP37:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 2
1164 // CHECK11-NEXT: store ptr [[TMP27]], ptr [[TMP37]], align 4
1165 // CHECK11-NEXT: [[TMP38:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 3
1166 // CHECK11-NEXT: store ptr [[TMP28]], ptr [[TMP38]], align 4
1167 // CHECK11-NEXT: [[TMP39:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 4
1168 // CHECK11-NEXT: store ptr [[TMP29]], ptr [[TMP39]], align 4
1169 // CHECK11-NEXT: [[TMP40:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 5
1170 // CHECK11-NEXT: store ptr @.offload_maptypes, ptr [[TMP40]], align 4
1171 // CHECK11-NEXT: [[TMP41:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 6
1172 // CHECK11-NEXT: store ptr null, ptr [[TMP41]], align 4
1173 // CHECK11-NEXT: [[TMP42:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 7
1174 // CHECK11-NEXT: store ptr null, ptr [[TMP42]], align 4
1175 // CHECK11-NEXT: [[TMP43:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 8
1176 // CHECK11-NEXT: store i64 [[ADD]], ptr [[TMP43]], align 8
1177 // CHECK11-NEXT: [[TMP44:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 9
1178 // CHECK11-NEXT: store i64 0, ptr [[TMP44]], align 8
1179 // CHECK11-NEXT: [[TMP45:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 10
1180 // CHECK11-NEXT: store [3 x i32] zeroinitializer, ptr [[TMP45]], align 4
1181 // CHECK11-NEXT: [[TMP46:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 11
1182 // CHECK11-NEXT: store [3 x i32] [i32 1, i32 0, i32 0], ptr [[TMP46]], align 4
1183 // CHECK11-NEXT: [[TMP47:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 12
1184 // CHECK11-NEXT: store i32 0, ptr [[TMP47]], align 4
1185 // CHECK11-NEXT: [[TMP48:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB2:[0-9]+]], i64 -1, i32 0, i32 1, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l80.region_id, ptr [[KERNEL_ARGS]])
1186 // CHECK11-NEXT: [[TMP49:%.*]] = icmp ne i32 [[TMP48]], 0
1187 // CHECK11-NEXT: br i1 [[TMP49]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
1188 // CHECK11: omp_offload.failed:
1189 // CHECK11-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l80(i32 [[TMP5]], i32 [[TMP7]], i32 [[TMP0]], i32 [[TMP1]], ptr [[VLA]]) #[[ATTR3:[0-9]+]]
1190 // CHECK11-NEXT: br label [[OMP_OFFLOAD_CONT]]
1191 // CHECK11: omp_offload.cont:
1192 // CHECK11-NEXT: [[TMP50:%.*]] = load i32, ptr [[ARGC_ADDR]], align 4
1193 // CHECK11-NEXT: [[CALL:%.*]] = call noundef i32 @_Z5tmainIiLi10ELi2EEiT_(i32 noundef [[TMP50]])
1194 // CHECK11-NEXT: store i32 [[CALL]], ptr [[RETVAL]], align 4
1195 // CHECK11-NEXT: [[TMP51:%.*]] = load ptr, ptr [[SAVED_STACK]], align 4
1196 // CHECK11-NEXT: call void @llvm.stackrestore.p0(ptr [[TMP51]])
1197 // CHECK11-NEXT: [[TMP52:%.*]] = load i32, ptr [[RETVAL]], align 4
1198 // CHECK11-NEXT: ret i32 [[TMP52]]
1199 //
1200 //
1201 // CHECK11-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l80
1202 // CHECK11-SAME: (i32 noundef [[N:%.*]], i32 noundef [[M:%.*]], i32 noundef [[VLA:%.*]], i32 noundef [[VLA1:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[A:%.*]]) #[[ATTR2:[0-9]+]] {
1203 // CHECK11-NEXT: entry:
1204 // CHECK11-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4
1205 // CHECK11-NEXT: [[M_ADDR:%.*]] = alloca i32, align 4
1206 // CHECK11-NEXT: [[VLA_ADDR:%.*]] = alloca i32, align 4
1207 // CHECK11-NEXT: [[VLA_ADDR2:%.*]] = alloca i32, align 4
1208 // CHECK11-NEXT: [[A_ADDR:%.*]] = alloca ptr, align 4
1209 // CHECK11-NEXT: [[N_CASTED:%.*]] = alloca i32, align 4
1210 // CHECK11-NEXT: [[M_CASTED:%.*]] = alloca i32, align 4
1211 // CHECK11-NEXT: store i32 [[N]], ptr [[N_ADDR]], align 4
1212 // CHECK11-NEXT: store i32 [[M]], ptr [[M_ADDR]], align 4
1213 // CHECK11-NEXT: store i32 [[VLA]], ptr [[VLA_ADDR]], align 4
1214 // CHECK11-NEXT: store i32 [[VLA1]], ptr [[VLA_ADDR2]], align 4
1215 // CHECK11-NEXT: store ptr [[A]], ptr [[A_ADDR]], align 4
1216 // CHECK11-NEXT: [[TMP0:%.*]] = load i32, ptr [[VLA_ADDR]], align 4
1217 // CHECK11-NEXT: [[TMP1:%.*]] = load i32, ptr [[VLA_ADDR2]], align 4
1218 // CHECK11-NEXT: [[TMP2:%.*]] = load ptr, ptr [[A_ADDR]], align 4
1219 // CHECK11-NEXT: [[TMP3:%.*]] = load i32, ptr [[N_ADDR]], align 4
1220 // CHECK11-NEXT: store i32 [[TMP3]], ptr [[N_CASTED]], align 4
1221 // CHECK11-NEXT: [[TMP4:%.*]] = load i32, ptr [[N_CASTED]], align 4
1222 // CHECK11-NEXT: [[TMP5:%.*]] = load i32, ptr [[M_ADDR]], align 4
1223 // CHECK11-NEXT: store i32 [[TMP5]], ptr [[M_CASTED]], align 4
1224 // CHECK11-NEXT: [[TMP6:%.*]] = load i32, ptr [[M_CASTED]], align 4
1225 // CHECK11-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_teams(ptr @[[GLOB2]], i32 5, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l80.omp_outlined, i32 [[TMP4]], i32 [[TMP6]], i32 [[TMP0]], i32 [[TMP1]], ptr [[TMP2]])
1226 // CHECK11-NEXT: ret void
1227 //
1228 //
1229 // CHECK11-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l80.omp_outlined
1230 // CHECK11-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], i32 noundef [[N:%.*]], i32 noundef [[M:%.*]], i32 noundef [[VLA:%.*]], i32 noundef [[VLA1:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[A:%.*]]) #[[ATTR2]] {
1231 // CHECK11-NEXT: entry:
1232 // CHECK11-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 4
1233 // CHECK11-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 4
1234 // CHECK11-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4
1235 // CHECK11-NEXT: [[M_ADDR:%.*]] = alloca i32, align 4
1236 // CHECK11-NEXT: [[VLA_ADDR:%.*]] = alloca i32, align 4
1237 // CHECK11-NEXT: [[VLA_ADDR2:%.*]] = alloca i32, align 4
1238 // CHECK11-NEXT: [[A_ADDR:%.*]] = alloca ptr, align 4
1239 // CHECK11-NEXT: [[DOTOMP_IV:%.*]] = alloca i64, align 8
1240 // CHECK11-NEXT: [[TMP:%.*]] = alloca i32, align 4
1241 // CHECK11-NEXT: [[_TMP3:%.*]] = alloca i32, align 4
1242 // CHECK11-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4
1243 // CHECK11-NEXT: [[DOTCAPTURE_EXPR_4:%.*]] = alloca i32, align 4
1244 // CHECK11-NEXT: [[DOTCAPTURE_EXPR_5:%.*]] = alloca i64, align 8
1245 // CHECK11-NEXT: [[I:%.*]] = alloca i32, align 4
1246 // CHECK11-NEXT: [[J:%.*]] = alloca i32, align 4
1247 // CHECK11-NEXT: [[DOTOMP_LB:%.*]] = alloca i64, align 8
1248 // CHECK11-NEXT: [[DOTOMP_UB:%.*]] = alloca i64, align 8
1249 // CHECK11-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i64, align 8
1250 // CHECK11-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
1251 // CHECK11-NEXT: [[I11:%.*]] = alloca i32, align 4
1252 // CHECK11-NEXT: [[J12:%.*]] = alloca i32, align 4
1253 // CHECK11-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 4
1254 // CHECK11-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 4
1255 // CHECK11-NEXT: store i32 [[N]], ptr [[N_ADDR]], align 4
1256 // CHECK11-NEXT: store i32 [[M]], ptr [[M_ADDR]], align 4
1257 // CHECK11-NEXT: store i32 [[VLA]], ptr [[VLA_ADDR]], align 4
1258 // CHECK11-NEXT: store i32 [[VLA1]], ptr [[VLA_ADDR2]], align 4
1259 // CHECK11-NEXT: store ptr [[A]], ptr [[A_ADDR]], align 4
1260 // CHECK11-NEXT: [[TMP0:%.*]] = load i32, ptr [[VLA_ADDR]], align 4
1261 // CHECK11-NEXT: [[TMP1:%.*]] = load i32, ptr [[VLA_ADDR2]], align 4
1262 // CHECK11-NEXT: [[TMP2:%.*]] = load ptr, ptr [[A_ADDR]], align 4
1263 // CHECK11-NEXT: [[TMP3:%.*]] = load i32, ptr [[N_ADDR]], align 4
1264 // CHECK11-NEXT: store i32 [[TMP3]], ptr [[DOTCAPTURE_EXPR_]], align 4
1265 // CHECK11-NEXT: [[TMP4:%.*]] = load i32, ptr [[M_ADDR]], align 4
1266 // CHECK11-NEXT: store i32 [[TMP4]], ptr [[DOTCAPTURE_EXPR_4]], align 4
1267 // CHECK11-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4
1268 // CHECK11-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP5]], 0
1269 // CHECK11-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1
1270 // CHECK11-NEXT: [[CONV:%.*]] = sext i32 [[DIV]] to i64
1271 // CHECK11-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_4]], align 4
1272 // CHECK11-NEXT: [[SUB6:%.*]] = sub nsw i32 [[TMP6]], 0
1273 // CHECK11-NEXT: [[DIV7:%.*]] = sdiv i32 [[SUB6]], 1
1274 // CHECK11-NEXT: [[CONV8:%.*]] = sext i32 [[DIV7]] to i64
1275 // CHECK11-NEXT: [[MUL:%.*]] = mul nsw i64 [[CONV]], [[CONV8]]
1276 // CHECK11-NEXT: [[SUB9:%.*]] = sub nsw i64 [[MUL]], 1
1277 // CHECK11-NEXT: store i64 [[SUB9]], ptr [[DOTCAPTURE_EXPR_5]], align 8
1278 // CHECK11-NEXT: store i32 0, ptr [[I]], align 4
1279 // CHECK11-NEXT: store i32 0, ptr [[J]], align 4
1280 // CHECK11-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4
1281 // CHECK11-NEXT: [[CMP:%.*]] = icmp slt i32 0, [[TMP7]]
1282 // CHECK11-NEXT: br i1 [[CMP]], label [[LAND_LHS_TRUE:%.*]], label [[OMP_PRECOND_END:%.*]]
1283 // CHECK11: land.lhs.true:
1284 // CHECK11-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_4]], align 4
1285 // CHECK11-NEXT: [[CMP10:%.*]] = icmp slt i32 0, [[TMP8]]
1286 // CHECK11-NEXT: br i1 [[CMP10]], label [[OMP_PRECOND_THEN:%.*]], label [[OMP_PRECOND_END]]
1287 // CHECK11: omp.precond.then:
1288 // CHECK11-NEXT: store i64 0, ptr [[DOTOMP_LB]], align 8
1289 // CHECK11-NEXT: [[TMP9:%.*]] = load i64, ptr [[DOTCAPTURE_EXPR_5]], align 8
1290 // CHECK11-NEXT: store i64 [[TMP9]], ptr [[DOTOMP_UB]], align 8
1291 // CHECK11-NEXT: store i64 1, ptr [[DOTOMP_STRIDE]], align 8
1292 // CHECK11-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
1293 // CHECK11-NEXT: [[TMP10:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
1294 // CHECK11-NEXT: [[TMP11:%.*]] = load i32, ptr [[TMP10]], align 4
1295 // CHECK11-NEXT: call void @__kmpc_for_static_init_8(ptr @[[GLOB1:[0-9]+]], i32 [[TMP11]], i32 92, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i64 1, i64 1)
1296 // CHECK11-NEXT: [[TMP12:%.*]] = load i64, ptr [[DOTOMP_UB]], align 8
1297 // CHECK11-NEXT: [[TMP13:%.*]] = load i64, ptr [[DOTCAPTURE_EXPR_5]], align 8
1298 // CHECK11-NEXT: [[CMP13:%.*]] = icmp sgt i64 [[TMP12]], [[TMP13]]
1299 // CHECK11-NEXT: br i1 [[CMP13]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
1300 // CHECK11: cond.true:
1301 // CHECK11-NEXT: [[TMP14:%.*]] = load i64, ptr [[DOTCAPTURE_EXPR_5]], align 8
1302 // CHECK11-NEXT: br label [[COND_END:%.*]]
1303 // CHECK11: cond.false:
1304 // CHECK11-NEXT: [[TMP15:%.*]] = load i64, ptr [[DOTOMP_UB]], align 8
1305 // CHECK11-NEXT: br label [[COND_END]]
1306 // CHECK11: cond.end:
1307 // CHECK11-NEXT: [[COND:%.*]] = phi i64 [ [[TMP14]], [[COND_TRUE]] ], [ [[TMP15]], [[COND_FALSE]] ]
1308 // CHECK11-NEXT: store i64 [[COND]], ptr [[DOTOMP_UB]], align 8
1309 // CHECK11-NEXT: [[TMP16:%.*]] = load i64, ptr [[DOTOMP_LB]], align 8
1310 // CHECK11-NEXT: store i64 [[TMP16]], ptr [[DOTOMP_IV]], align 8
1311 // CHECK11-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
1312 // CHECK11: omp.inner.for.cond:
1313 // CHECK11-NEXT: [[TMP17:%.*]] = load i64, ptr [[DOTOMP_IV]], align 8, !llvm.access.group [[ACC_GRP6:![0-9]+]]
1314 // CHECK11-NEXT: [[TMP18:%.*]] = load i64, ptr [[DOTOMP_UB]], align 8, !llvm.access.group [[ACC_GRP6]]
1315 // CHECK11-NEXT: [[CMP14:%.*]] = icmp sle i64 [[TMP17]], [[TMP18]]
1316 // CHECK11-NEXT: br i1 [[CMP14]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
1317 // CHECK11: omp.inner.for.body:
1318 // CHECK11-NEXT: [[TMP19:%.*]] = load i64, ptr [[DOTOMP_IV]], align 8, !llvm.access.group [[ACC_GRP6]]
1319 // CHECK11-NEXT: [[TMP20:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_4]], align 4, !llvm.access.group [[ACC_GRP6]]
1320 // CHECK11-NEXT: [[SUB15:%.*]] = sub nsw i32 [[TMP20]], 0
1321 // CHECK11-NEXT: [[DIV16:%.*]] = sdiv i32 [[SUB15]], 1
1322 // CHECK11-NEXT: [[MUL17:%.*]] = mul nsw i32 1, [[DIV16]]
1323 // CHECK11-NEXT: [[CONV18:%.*]] = sext i32 [[MUL17]] to i64
1324 // CHECK11-NEXT: [[DIV19:%.*]] = sdiv i64 [[TMP19]], [[CONV18]]
1325 // CHECK11-NEXT: [[MUL20:%.*]] = mul nsw i64 [[DIV19]], 1
1326 // CHECK11-NEXT: [[ADD:%.*]] = add nsw i64 0, [[MUL20]]
1327 // CHECK11-NEXT: [[CONV21:%.*]] = trunc i64 [[ADD]] to i32
1328 // CHECK11-NEXT: store i32 [[CONV21]], ptr [[I11]], align 4, !llvm.access.group [[ACC_GRP6]]
1329 // CHECK11-NEXT: [[TMP21:%.*]] = load i64, ptr [[DOTOMP_IV]], align 8, !llvm.access.group [[ACC_GRP6]]
1330 // CHECK11-NEXT: [[TMP22:%.*]] = load i64, ptr [[DOTOMP_IV]], align 8, !llvm.access.group [[ACC_GRP6]]
1331 // CHECK11-NEXT: [[TMP23:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_4]], align 4, !llvm.access.group [[ACC_GRP6]]
1332 // CHECK11-NEXT: [[SUB22:%.*]] = sub nsw i32 [[TMP23]], 0
1333 // CHECK11-NEXT: [[DIV23:%.*]] = sdiv i32 [[SUB22]], 1
1334 // CHECK11-NEXT: [[MUL24:%.*]] = mul nsw i32 1, [[DIV23]]
1335 // CHECK11-NEXT: [[CONV25:%.*]] = sext i32 [[MUL24]] to i64
1336 // CHECK11-NEXT: [[DIV26:%.*]] = sdiv i64 [[TMP22]], [[CONV25]]
1337 // CHECK11-NEXT: [[TMP24:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_4]], align 4, !llvm.access.group [[ACC_GRP6]]
1338 // CHECK11-NEXT: [[SUB27:%.*]] = sub nsw i32 [[TMP24]], 0
1339 // CHECK11-NEXT: [[DIV28:%.*]] = sdiv i32 [[SUB27]], 1
1340 // CHECK11-NEXT: [[MUL29:%.*]] = mul nsw i32 1, [[DIV28]]
1341 // CHECK11-NEXT: [[CONV30:%.*]] = sext i32 [[MUL29]] to i64
1342 // CHECK11-NEXT: [[MUL31:%.*]] = mul nsw i64 [[DIV26]], [[CONV30]]
1343 // CHECK11-NEXT: [[SUB32:%.*]] = sub nsw i64 [[TMP21]], [[MUL31]]
1344 // CHECK11-NEXT: [[MUL33:%.*]] = mul nsw i64 [[SUB32]], 1
1345 // CHECK11-NEXT: [[ADD34:%.*]] = add nsw i64 0, [[MUL33]]
1346 // CHECK11-NEXT: [[CONV35:%.*]] = trunc i64 [[ADD34]] to i32
1347 // CHECK11-NEXT: store i32 [[CONV35]], ptr [[J12]], align 4, !llvm.access.group [[ACC_GRP6]]
1348 // CHECK11-NEXT: [[TMP25:%.*]] = load i32, ptr [[I11]], align 4, !llvm.access.group [[ACC_GRP6]]
1349 // CHECK11-NEXT: [[TMP26:%.*]] = mul nsw i32 [[TMP25]], [[TMP1]]
1350 // CHECK11-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i32, ptr [[TMP2]], i32 [[TMP26]]
1351 // CHECK11-NEXT: [[TMP27:%.*]] = load i32, ptr [[J12]], align 4, !llvm.access.group [[ACC_GRP6]]
1352 // CHECK11-NEXT: [[ARRAYIDX36:%.*]] = getelementptr inbounds i32, ptr [[ARRAYIDX]], i32 [[TMP27]]
1353 // CHECK11-NEXT: store i32 0, ptr [[ARRAYIDX36]], align 4, !llvm.access.group [[ACC_GRP6]]
1354 // CHECK11-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
1355 // CHECK11: omp.body.continue:
1356 // CHECK11-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
1357 // CHECK11: omp.inner.for.inc:
1358 // CHECK11-NEXT: [[TMP28:%.*]] = load i64, ptr [[DOTOMP_IV]], align 8, !llvm.access.group [[ACC_GRP6]]
1359 // CHECK11-NEXT: [[ADD37:%.*]] = add nsw i64 [[TMP28]], 1
1360 // CHECK11-NEXT: store i64 [[ADD37]], ptr [[DOTOMP_IV]], align 8, !llvm.access.group [[ACC_GRP6]]
1361 // CHECK11-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP7:![0-9]+]]
1362 // CHECK11: omp.inner.for.end:
1363 // CHECK11-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
1364 // CHECK11: omp.loop.exit:
1365 // CHECK11-NEXT: [[TMP29:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
1366 // CHECK11-NEXT: [[TMP30:%.*]] = load i32, ptr [[TMP29]], align 4
1367 // CHECK11-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB1]], i32 [[TMP30]])
1368 // CHECK11-NEXT: [[TMP31:%.*]] = load i32, ptr [[DOTOMP_IS_LAST]], align 4
1369 // CHECK11-NEXT: [[TMP32:%.*]] = icmp ne i32 [[TMP31]], 0
1370 // CHECK11-NEXT: br i1 [[TMP32]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
1371 // CHECK11: .omp.final.then:
1372 // CHECK11-NEXT: [[TMP33:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4
1373 // CHECK11-NEXT: [[SUB38:%.*]] = sub nsw i32 [[TMP33]], 0
1374 // CHECK11-NEXT: [[DIV39:%.*]] = sdiv i32 [[SUB38]], 1
1375 // CHECK11-NEXT: [[MUL40:%.*]] = mul nsw i32 [[DIV39]], 1
1376 // CHECK11-NEXT: [[ADD41:%.*]] = add nsw i32 0, [[MUL40]]
1377 // CHECK11-NEXT: store i32 [[ADD41]], ptr [[I11]], align 4
1378 // CHECK11-NEXT: [[TMP34:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_4]], align 4
1379 // CHECK11-NEXT: [[SUB42:%.*]] = sub nsw i32 [[TMP34]], 0
1380 // CHECK11-NEXT: [[DIV43:%.*]] = sdiv i32 [[SUB42]], 1
1381 // CHECK11-NEXT: [[MUL44:%.*]] = mul nsw i32 [[DIV43]], 1
1382 // CHECK11-NEXT: [[ADD45:%.*]] = add nsw i32 0, [[MUL44]]
1383 // CHECK11-NEXT: store i32 [[ADD45]], ptr [[J12]], align 4
1384 // CHECK11-NEXT: br label [[DOTOMP_FINAL_DONE]]
1385 // CHECK11: .omp.final.done:
1386 // CHECK11-NEXT: br label [[OMP_PRECOND_END]]
1387 // CHECK11: omp.precond.end:
1388 // CHECK11-NEXT: ret void
1389 //
1390 //
1391 // CHECK11-LABEL: define {{[^@]+}}@_Z5tmainIiLi10ELi2EEiT_
1392 // CHECK11-SAME: (i32 noundef [[ARGC:%.*]]) #[[ATTR5:[0-9]+]] comdat {
1393 // CHECK11-NEXT: entry:
1394 // CHECK11-NEXT: [[ARGC_ADDR:%.*]] = alloca i32, align 4
1395 // CHECK11-NEXT: [[A:%.*]] = alloca [10 x [2 x i32]], align 4
1396 // CHECK11-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [1 x ptr], align 4
1397 // CHECK11-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [1 x ptr], align 4
1398 // CHECK11-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [1 x ptr], align 4
1399 // CHECK11-NEXT: [[TMP:%.*]] = alloca i32, align 4
1400 // CHECK11-NEXT: [[_TMP1:%.*]] = alloca i32, align 4
1401 // CHECK11-NEXT: [[KERNEL_ARGS:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS:%.*]], align 8
1402 // CHECK11-NEXT: store i32 [[ARGC]], ptr [[ARGC_ADDR]], align 4
1403 // CHECK11-NEXT: [[TMP0:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
1404 // CHECK11-NEXT: store ptr [[A]], ptr [[TMP0]], align 4
1405 // CHECK11-NEXT: [[TMP1:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
1406 // CHECK11-NEXT: store ptr [[A]], ptr [[TMP1]], align 4
1407 // CHECK11-NEXT: [[TMP2:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0
1408 // CHECK11-NEXT: store ptr null, ptr [[TMP2]], align 4
1409 // CHECK11-NEXT: [[TMP3:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
1410 // CHECK11-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
1411 // CHECK11-NEXT: [[TMP5:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 0
1412 // CHECK11-NEXT: store i32 3, ptr [[TMP5]], align 4
1413 // CHECK11-NEXT: [[TMP6:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 1
1414 // CHECK11-NEXT: store i32 1, ptr [[TMP6]], align 4
1415 // CHECK11-NEXT: [[TMP7:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 2
1416 // CHECK11-NEXT: store ptr [[TMP3]], ptr [[TMP7]], align 4
1417 // CHECK11-NEXT: [[TMP8:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 3
1418 // CHECK11-NEXT: store ptr [[TMP4]], ptr [[TMP8]], align 4
1419 // CHECK11-NEXT: [[TMP9:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 4
1420 // CHECK11-NEXT: store ptr @.offload_sizes.1, ptr [[TMP9]], align 4
1421 // CHECK11-NEXT: [[TMP10:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 5
1422 // CHECK11-NEXT: store ptr @.offload_maptypes.2, ptr [[TMP10]], align 4
1423 // CHECK11-NEXT: [[TMP11:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 6
1424 // CHECK11-NEXT: store ptr null, ptr [[TMP11]], align 4
1425 // CHECK11-NEXT: [[TMP12:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 7
1426 // CHECK11-NEXT: store ptr null, ptr [[TMP12]], align 4
1427 // CHECK11-NEXT: [[TMP13:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 8
1428 // CHECK11-NEXT: store i64 20, ptr [[TMP13]], align 8
1429 // CHECK11-NEXT: [[TMP14:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 9
1430 // CHECK11-NEXT: store i64 0, ptr [[TMP14]], align 8
1431 // CHECK11-NEXT: [[TMP15:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 10
1432 // CHECK11-NEXT: store [3 x i32] zeroinitializer, ptr [[TMP15]], align 4
1433 // CHECK11-NEXT: [[TMP16:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 11
1434 // CHECK11-NEXT: store [3 x i32] [i32 1, i32 0, i32 0], ptr [[TMP16]], align 4
1435 // CHECK11-NEXT: [[TMP17:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 12
1436 // CHECK11-NEXT: store i32 0, ptr [[TMP17]], align 4
1437 // CHECK11-NEXT: [[TMP18:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB2]], i64 -1, i32 0, i32 1, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10ELi2EEiT__l67.region_id, ptr [[KERNEL_ARGS]])
1438 // CHECK11-NEXT: [[TMP19:%.*]] = icmp ne i32 [[TMP18]], 0
1439 // CHECK11-NEXT: br i1 [[TMP19]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
1440 // CHECK11: omp_offload.failed:
1441 // CHECK11-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10ELi2EEiT__l67(ptr [[A]]) #[[ATTR3]]
1442 // CHECK11-NEXT: br label [[OMP_OFFLOAD_CONT]]
1443 // CHECK11: omp_offload.cont:
1444 // CHECK11-NEXT: ret i32 0
1445 //
1446 //
1447 // CHECK11-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10ELi2EEiT__l67
1448 // CHECK11-SAME: (ptr noundef nonnull align 4 dereferenceable(80) [[A:%.*]]) #[[ATTR2]] {
1449 // CHECK11-NEXT: entry:
1450 // CHECK11-NEXT: [[A_ADDR:%.*]] = alloca ptr, align 4
1451 // CHECK11-NEXT: store ptr [[A]], ptr [[A_ADDR]], align 4
1452 // CHECK11-NEXT: [[TMP0:%.*]] = load ptr, ptr [[A_ADDR]], align 4
1453 // CHECK11-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_teams(ptr @[[GLOB2]], i32 1, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10ELi2EEiT__l67.omp_outlined, ptr [[TMP0]])
1454 // CHECK11-NEXT: ret void
1455 //
1456 //
1457 // CHECK11-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10ELi2EEiT__l67.omp_outlined
1458 // CHECK11-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], ptr noundef nonnull align 4 dereferenceable(80) [[A:%.*]]) #[[ATTR2]] {
1459 // CHECK11-NEXT: entry:
1460 // CHECK11-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 4
1461 // CHECK11-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 4
1462 // CHECK11-NEXT: [[A_ADDR:%.*]] = alloca ptr, align 4
1463 // CHECK11-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
1464 // CHECK11-NEXT: [[TMP:%.*]] = alloca i32, align 4
1465 // CHECK11-NEXT: [[_TMP1:%.*]] = alloca i32, align 4
1466 // CHECK11-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4
1467 // CHECK11-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4
1468 // CHECK11-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
1469 // CHECK11-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
1470 // CHECK11-NEXT: [[I:%.*]] = alloca i32, align 4
1471 // CHECK11-NEXT: [[J:%.*]] = alloca i32, align 4
1472 // CHECK11-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 4
1473 // CHECK11-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 4
1474 // CHECK11-NEXT: store ptr [[A]], ptr [[A_ADDR]], align 4
1475 // CHECK11-NEXT: [[TMP0:%.*]] = load ptr, ptr [[A_ADDR]], align 4
1476 // CHECK11-NEXT: store i32 0, ptr [[DOTOMP_LB]], align 4
1477 // CHECK11-NEXT: store i32 19, ptr [[DOTOMP_UB]], align 4
1478 // CHECK11-NEXT: store i32 1, ptr [[DOTOMP_STRIDE]], align 4
1479 // CHECK11-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
1480 // CHECK11-NEXT: [[TMP1:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
1481 // CHECK11-NEXT: [[TMP2:%.*]] = load i32, ptr [[TMP1]], align 4
1482 // CHECK11-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB1]], i32 [[TMP2]], i32 92, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
1483 // CHECK11-NEXT: [[TMP3:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
1484 // CHECK11-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP3]], 19
1485 // CHECK11-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
1486 // CHECK11: cond.true:
1487 // CHECK11-NEXT: br label [[COND_END:%.*]]
1488 // CHECK11: cond.false:
1489 // CHECK11-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
1490 // CHECK11-NEXT: br label [[COND_END]]
1491 // CHECK11: cond.end:
1492 // CHECK11-NEXT: [[COND:%.*]] = phi i32 [ 19, [[COND_TRUE]] ], [ [[TMP4]], [[COND_FALSE]] ]
1493 // CHECK11-NEXT: store i32 [[COND]], ptr [[DOTOMP_UB]], align 4
1494 // CHECK11-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4
1495 // CHECK11-NEXT: store i32 [[TMP5]], ptr [[DOTOMP_IV]], align 4
1496 // CHECK11-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
1497 // CHECK11: omp.inner.for.cond:
1498 // CHECK11-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP12:![0-9]+]]
1499 // CHECK11-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP12]]
1500 // CHECK11-NEXT: [[CMP2:%.*]] = icmp sle i32 [[TMP6]], [[TMP7]]
1501 // CHECK11-NEXT: br i1 [[CMP2]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
1502 // CHECK11: omp.inner.for.body:
1503 // CHECK11-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP12]]
1504 // CHECK11-NEXT: [[DIV:%.*]] = sdiv i32 [[TMP8]], 2
1505 // CHECK11-NEXT: [[MUL:%.*]] = mul nsw i32 [[DIV]], 1
1506 // CHECK11-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
1507 // CHECK11-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP12]]
1508 // CHECK11-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP12]]
1509 // CHECK11-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP12]]
1510 // CHECK11-NEXT: [[DIV3:%.*]] = sdiv i32 [[TMP10]], 2
1511 // CHECK11-NEXT: [[MUL4:%.*]] = mul nsw i32 [[DIV3]], 2
1512 // CHECK11-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP9]], [[MUL4]]
1513 // CHECK11-NEXT: [[MUL5:%.*]] = mul nsw i32 [[SUB]], 1
1514 // CHECK11-NEXT: [[ADD6:%.*]] = add nsw i32 0, [[MUL5]]
1515 // CHECK11-NEXT: store i32 [[ADD6]], ptr [[J]], align 4, !llvm.access.group [[ACC_GRP12]]
1516 // CHECK11-NEXT: [[TMP11:%.*]] = load i32, ptr [[I]], align 4, !llvm.access.group [[ACC_GRP12]]
1517 // CHECK11-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds [10 x [2 x i32]], ptr [[TMP0]], i32 0, i32 [[TMP11]]
1518 // CHECK11-NEXT: [[TMP12:%.*]] = load i32, ptr [[J]], align 4, !llvm.access.group [[ACC_GRP12]]
1519 // CHECK11-NEXT: [[ARRAYIDX7:%.*]] = getelementptr inbounds [2 x i32], ptr [[ARRAYIDX]], i32 0, i32 [[TMP12]]
1520 // CHECK11-NEXT: store i32 0, ptr [[ARRAYIDX7]], align 4, !llvm.access.group [[ACC_GRP12]]
1521 // CHECK11-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
1522 // CHECK11: omp.body.continue:
1523 // CHECK11-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
1524 // CHECK11: omp.inner.for.inc:
1525 // CHECK11-NEXT: [[TMP13:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP12]]
1526 // CHECK11-NEXT: [[ADD8:%.*]] = add nsw i32 [[TMP13]], 1
1527 // CHECK11-NEXT: store i32 [[ADD8]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP12]]
1528 // CHECK11-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP13:![0-9]+]]
1529 // CHECK11: omp.inner.for.end:
1530 // CHECK11-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
1531 // CHECK11: omp.loop.exit:
1532 // CHECK11-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB1]], i32 [[TMP2]])
1533 // CHECK11-NEXT: [[TMP14:%.*]] = load i32, ptr [[DOTOMP_IS_LAST]], align 4
1534 // CHECK11-NEXT: [[TMP15:%.*]] = icmp ne i32 [[TMP14]], 0
1535 // CHECK11-NEXT: br i1 [[TMP15]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
1536 // CHECK11: .omp.final.then:
1537 // CHECK11-NEXT: store i32 10, ptr [[I]], align 4
1538 // CHECK11-NEXT: store i32 2, ptr [[J]], align 4
1539 // CHECK11-NEXT: br label [[DOTOMP_FINAL_DONE]]
1540 // CHECK11: .omp.final.done:
1541 // CHECK11-NEXT: ret void
1542 //
1543 //
1544 // CHECK13-LABEL: define {{[^@]+}}@main
1545 // CHECK13-SAME: (i32 noundef signext [[ARGC:%.*]], ptr noundef [[ARGV:%.*]]) #[[ATTR0:[0-9]+]] {
1546 // CHECK13-NEXT: entry:
1547 // CHECK13-NEXT: [[RETVAL:%.*]] = alloca i32, align 4
1548 // CHECK13-NEXT: [[ARGC_ADDR:%.*]] = alloca i32, align 4
1549 // CHECK13-NEXT: [[ARGV_ADDR:%.*]] = alloca ptr, align 8
1550 // CHECK13-NEXT: [[N:%.*]] = alloca i32, align 4
1551 // CHECK13-NEXT: [[M:%.*]] = alloca i32, align 4
1552 // CHECK13-NEXT: [[SAVED_STACK:%.*]] = alloca ptr, align 8
1553 // CHECK13-NEXT: [[__VLA_EXPR0:%.*]] = alloca i64, align 8
1554 // CHECK13-NEXT: [[__VLA_EXPR1:%.*]] = alloca i64, align 8
1555 // CHECK13-NEXT: [[TMP:%.*]] = alloca i32, align 4
1556 // CHECK13-NEXT: [[_TMP1:%.*]] = alloca i32, align 4
1557 // CHECK13-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4
1558 // CHECK13-NEXT: [[DOTCAPTURE_EXPR_2:%.*]] = alloca i32, align 4
1559 // CHECK13-NEXT: [[DOTCAPTURE_EXPR_3:%.*]] = alloca i64, align 8
1560 // CHECK13-NEXT: [[DOTOMP_LB:%.*]] = alloca i64, align 8
1561 // CHECK13-NEXT: [[DOTOMP_UB:%.*]] = alloca i64, align 8
1562 // CHECK13-NEXT: [[I:%.*]] = alloca i32, align 4
1563 // CHECK13-NEXT: [[J:%.*]] = alloca i32, align 4
1564 // CHECK13-NEXT: [[DOTOMP_IV:%.*]] = alloca i64, align 8
1565 // CHECK13-NEXT: [[I9:%.*]] = alloca i32, align 4
1566 // CHECK13-NEXT: [[J10:%.*]] = alloca i32, align 4
1567 // CHECK13-NEXT: store i32 0, ptr [[RETVAL]], align 4
1568 // CHECK13-NEXT: store i32 [[ARGC]], ptr [[ARGC_ADDR]], align 4
1569 // CHECK13-NEXT: store ptr [[ARGV]], ptr [[ARGV_ADDR]], align 8
1570 // CHECK13-NEXT: store i32 100, ptr [[N]], align 4
1571 // CHECK13-NEXT: store i32 2, ptr [[M]], align 4
1572 // CHECK13-NEXT: [[TMP0:%.*]] = load i32, ptr [[N]], align 4
1573 // CHECK13-NEXT: [[TMP1:%.*]] = zext i32 [[TMP0]] to i64
1574 // CHECK13-NEXT: [[TMP2:%.*]] = load i32, ptr [[M]], align 4
1575 // CHECK13-NEXT: [[TMP3:%.*]] = zext i32 [[TMP2]] to i64
1576 // CHECK13-NEXT: [[TMP4:%.*]] = call ptr @llvm.stacksave.p0()
1577 // CHECK13-NEXT: store ptr [[TMP4]], ptr [[SAVED_STACK]], align 8
1578 // CHECK13-NEXT: [[TMP5:%.*]] = mul nuw i64 [[TMP1]], [[TMP3]]
1579 // CHECK13-NEXT: [[VLA:%.*]] = alloca i32, i64 [[TMP5]], align 4
1580 // CHECK13-NEXT: store i64 [[TMP1]], ptr [[__VLA_EXPR0]], align 8
1581 // CHECK13-NEXT: store i64 [[TMP3]], ptr [[__VLA_EXPR1]], align 8
1582 // CHECK13-NEXT: [[TMP6:%.*]] = load i32, ptr [[N]], align 4
1583 // CHECK13-NEXT: store i32 [[TMP6]], ptr [[DOTCAPTURE_EXPR_]], align 4
1584 // CHECK13-NEXT: [[TMP7:%.*]] = load i32, ptr [[M]], align 4
1585 // CHECK13-NEXT: store i32 [[TMP7]], ptr [[DOTCAPTURE_EXPR_2]], align 4
1586 // CHECK13-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4
1587 // CHECK13-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP8]], 0
1588 // CHECK13-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1
1589 // CHECK13-NEXT: [[CONV:%.*]] = sext i32 [[DIV]] to i64
1590 // CHECK13-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_2]], align 4
1591 // CHECK13-NEXT: [[SUB4:%.*]] = sub nsw i32 [[TMP9]], 0
1592 // CHECK13-NEXT: [[DIV5:%.*]] = sdiv i32 [[SUB4]], 1
1593 // CHECK13-NEXT: [[CONV6:%.*]] = sext i32 [[DIV5]] to i64
1594 // CHECK13-NEXT: [[MUL:%.*]] = mul nsw i64 [[CONV]], [[CONV6]]
1595 // CHECK13-NEXT: [[SUB7:%.*]] = sub nsw i64 [[MUL]], 1
1596 // CHECK13-NEXT: store i64 [[SUB7]], ptr [[DOTCAPTURE_EXPR_3]], align 8
1597 // CHECK13-NEXT: store i64 0, ptr [[DOTOMP_LB]], align 8
1598 // CHECK13-NEXT: [[TMP10:%.*]] = load i64, ptr [[DOTCAPTURE_EXPR_3]], align 8
1599 // CHECK13-NEXT: store i64 [[TMP10]], ptr [[DOTOMP_UB]], align 8
1600 // CHECK13-NEXT: store i32 0, ptr [[I]], align 4
1601 // CHECK13-NEXT: store i32 0, ptr [[J]], align 4
1602 // CHECK13-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4
1603 // CHECK13-NEXT: [[CMP:%.*]] = icmp slt i32 0, [[TMP11]]
1604 // CHECK13-NEXT: br i1 [[CMP]], label [[LAND_LHS_TRUE:%.*]], label [[SIMD_IF_END:%.*]]
1605 // CHECK13: land.lhs.true:
1606 // CHECK13-NEXT: [[TMP12:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_2]], align 4
1607 // CHECK13-NEXT: [[CMP8:%.*]] = icmp slt i32 0, [[TMP12]]
1608 // CHECK13-NEXT: br i1 [[CMP8]], label [[SIMD_IF_THEN:%.*]], label [[SIMD_IF_END]]
1609 // CHECK13: simd.if.then:
1610 // CHECK13-NEXT: [[TMP13:%.*]] = load i64, ptr [[DOTOMP_LB]], align 8
1611 // CHECK13-NEXT: store i64 [[TMP13]], ptr [[DOTOMP_IV]], align 8
1612 // CHECK13-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
1613 // CHECK13: omp.inner.for.cond:
1614 // CHECK13-NEXT: [[TMP14:%.*]] = load i64, ptr [[DOTOMP_IV]], align 8, !llvm.access.group [[ACC_GRP2:![0-9]+]]
1615 // CHECK13-NEXT: [[TMP15:%.*]] = load i64, ptr [[DOTOMP_UB]], align 8, !llvm.access.group [[ACC_GRP2]]
1616 // CHECK13-NEXT: [[CMP11:%.*]] = icmp sle i64 [[TMP14]], [[TMP15]]
1617 // CHECK13-NEXT: br i1 [[CMP11]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
1618 // CHECK13: omp.inner.for.body:
1619 // CHECK13-NEXT: [[TMP16:%.*]] = load i64, ptr [[DOTOMP_IV]], align 8, !llvm.access.group [[ACC_GRP2]]
1620 // CHECK13-NEXT: [[TMP17:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_2]], align 4, !llvm.access.group [[ACC_GRP2]]
1621 // CHECK13-NEXT: [[SUB12:%.*]] = sub nsw i32 [[TMP17]], 0
1622 // CHECK13-NEXT: [[DIV13:%.*]] = sdiv i32 [[SUB12]], 1
1623 // CHECK13-NEXT: [[MUL14:%.*]] = mul nsw i32 1, [[DIV13]]
1624 // CHECK13-NEXT: [[CONV15:%.*]] = sext i32 [[MUL14]] to i64
1625 // CHECK13-NEXT: [[DIV16:%.*]] = sdiv i64 [[TMP16]], [[CONV15]]
1626 // CHECK13-NEXT: [[MUL17:%.*]] = mul nsw i64 [[DIV16]], 1
1627 // CHECK13-NEXT: [[ADD:%.*]] = add nsw i64 0, [[MUL17]]
1628 // CHECK13-NEXT: [[CONV18:%.*]] = trunc i64 [[ADD]] to i32
1629 // CHECK13-NEXT: store i32 [[CONV18]], ptr [[I9]], align 4, !llvm.access.group [[ACC_GRP2]]
1630 // CHECK13-NEXT: [[TMP18:%.*]] = load i64, ptr [[DOTOMP_IV]], align 8, !llvm.access.group [[ACC_GRP2]]
1631 // CHECK13-NEXT: [[TMP19:%.*]] = load i64, ptr [[DOTOMP_IV]], align 8, !llvm.access.group [[ACC_GRP2]]
1632 // CHECK13-NEXT: [[TMP20:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_2]], align 4, !llvm.access.group [[ACC_GRP2]]
1633 // CHECK13-NEXT: [[SUB19:%.*]] = sub nsw i32 [[TMP20]], 0
1634 // CHECK13-NEXT: [[DIV20:%.*]] = sdiv i32 [[SUB19]], 1
1635 // CHECK13-NEXT: [[MUL21:%.*]] = mul nsw i32 1, [[DIV20]]
1636 // CHECK13-NEXT: [[CONV22:%.*]] = sext i32 [[MUL21]] to i64
1637 // CHECK13-NEXT: [[DIV23:%.*]] = sdiv i64 [[TMP19]], [[CONV22]]
1638 // CHECK13-NEXT: [[TMP21:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_2]], align 4, !llvm.access.group [[ACC_GRP2]]
1639 // CHECK13-NEXT: [[SUB24:%.*]] = sub nsw i32 [[TMP21]], 0
1640 // CHECK13-NEXT: [[DIV25:%.*]] = sdiv i32 [[SUB24]], 1
1641 // CHECK13-NEXT: [[MUL26:%.*]] = mul nsw i32 1, [[DIV25]]
1642 // CHECK13-NEXT: [[CONV27:%.*]] = sext i32 [[MUL26]] to i64
1643 // CHECK13-NEXT: [[MUL28:%.*]] = mul nsw i64 [[DIV23]], [[CONV27]]
1644 // CHECK13-NEXT: [[SUB29:%.*]] = sub nsw i64 [[TMP18]], [[MUL28]]
1645 // CHECK13-NEXT: [[MUL30:%.*]] = mul nsw i64 [[SUB29]], 1
1646 // CHECK13-NEXT: [[ADD31:%.*]] = add nsw i64 0, [[MUL30]]
1647 // CHECK13-NEXT: [[CONV32:%.*]] = trunc i64 [[ADD31]] to i32
1648 // CHECK13-NEXT: store i32 [[CONV32]], ptr [[J10]], align 4, !llvm.access.group [[ACC_GRP2]]
1649 // CHECK13-NEXT: [[TMP22:%.*]] = load i32, ptr [[I9]], align 4, !llvm.access.group [[ACC_GRP2]]
1650 // CHECK13-NEXT: [[IDXPROM:%.*]] = sext i32 [[TMP22]] to i64
1651 // CHECK13-NEXT: [[TMP23:%.*]] = mul nsw i64 [[IDXPROM]], [[TMP3]]
1652 // CHECK13-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i32, ptr [[VLA]], i64 [[TMP23]]
1653 // CHECK13-NEXT: [[TMP24:%.*]] = load i32, ptr [[J10]], align 4, !llvm.access.group [[ACC_GRP2]]
1654 // CHECK13-NEXT: [[IDXPROM33:%.*]] = sext i32 [[TMP24]] to i64
1655 // CHECK13-NEXT: [[ARRAYIDX34:%.*]] = getelementptr inbounds i32, ptr [[ARRAYIDX]], i64 [[IDXPROM33]]
1656 // CHECK13-NEXT: store i32 0, ptr [[ARRAYIDX34]], align 4, !llvm.access.group [[ACC_GRP2]]
1657 // CHECK13-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
1658 // CHECK13: omp.body.continue:
1659 // CHECK13-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
1660 // CHECK13: omp.inner.for.inc:
1661 // CHECK13-NEXT: [[TMP25:%.*]] = load i64, ptr [[DOTOMP_IV]], align 8, !llvm.access.group [[ACC_GRP2]]
1662 // CHECK13-NEXT: [[ADD35:%.*]] = add nsw i64 [[TMP25]], 1
1663 // CHECK13-NEXT: store i64 [[ADD35]], ptr [[DOTOMP_IV]], align 8, !llvm.access.group [[ACC_GRP2]]
1664 // CHECK13-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP3:![0-9]+]]
1665 // CHECK13: omp.inner.for.end:
1666 // CHECK13-NEXT: [[TMP26:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4
1667 // CHECK13-NEXT: [[SUB36:%.*]] = sub nsw i32 [[TMP26]], 0
1668 // CHECK13-NEXT: [[DIV37:%.*]] = sdiv i32 [[SUB36]], 1
1669 // CHECK13-NEXT: [[MUL38:%.*]] = mul nsw i32 [[DIV37]], 1
1670 // CHECK13-NEXT: [[ADD39:%.*]] = add nsw i32 0, [[MUL38]]
1671 // CHECK13-NEXT: store i32 [[ADD39]], ptr [[I9]], align 4
1672 // CHECK13-NEXT: [[TMP27:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_2]], align 4
1673 // CHECK13-NEXT: [[SUB40:%.*]] = sub nsw i32 [[TMP27]], 0
1674 // CHECK13-NEXT: [[DIV41:%.*]] = sdiv i32 [[SUB40]], 1
1675 // CHECK13-NEXT: [[MUL42:%.*]] = mul nsw i32 [[DIV41]], 1
1676 // CHECK13-NEXT: [[ADD43:%.*]] = add nsw i32 0, [[MUL42]]
1677 // CHECK13-NEXT: store i32 [[ADD43]], ptr [[J10]], align 4
1678 // CHECK13-NEXT: br label [[SIMD_IF_END]]
1679 // CHECK13: simd.if.end:
1680 // CHECK13-NEXT: [[TMP28:%.*]] = load i32, ptr [[ARGC_ADDR]], align 4
1681 // CHECK13-NEXT: [[CALL:%.*]] = call noundef signext i32 @_Z5tmainIiLi10ELi2EEiT_(i32 noundef signext [[TMP28]])
1682 // CHECK13-NEXT: store i32 [[CALL]], ptr [[RETVAL]], align 4
1683 // CHECK13-NEXT: [[TMP29:%.*]] = load ptr, ptr [[SAVED_STACK]], align 8
1684 // CHECK13-NEXT: call void @llvm.stackrestore.p0(ptr [[TMP29]])
1685 // CHECK13-NEXT: [[TMP30:%.*]] = load i32, ptr [[RETVAL]], align 4
1686 // CHECK13-NEXT: ret i32 [[TMP30]]
1687 //
1688 //
1689 // CHECK13-LABEL: define {{[^@]+}}@_Z5tmainIiLi10ELi2EEiT_
1690 // CHECK13-SAME: (i32 noundef signext [[ARGC:%.*]]) #[[ATTR2:[0-9]+]] comdat {
1691 // CHECK13-NEXT: entry:
1692 // CHECK13-NEXT: [[ARGC_ADDR:%.*]] = alloca i32, align 4
1693 // CHECK13-NEXT: [[A:%.*]] = alloca [10 x [2 x i32]], align 4
1694 // CHECK13-NEXT: [[TMP:%.*]] = alloca i32, align 4
1695 // CHECK13-NEXT: [[_TMP1:%.*]] = alloca i32, align 4
1696 // CHECK13-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4
1697 // CHECK13-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4
1698 // CHECK13-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
1699 // CHECK13-NEXT: [[I:%.*]] = alloca i32, align 4
1700 // CHECK13-NEXT: [[J:%.*]] = alloca i32, align 4
1701 // CHECK13-NEXT: store i32 [[ARGC]], ptr [[ARGC_ADDR]], align 4
1702 // CHECK13-NEXT: store i32 0, ptr [[DOTOMP_LB]], align 4
1703 // CHECK13-NEXT: store i32 19, ptr [[DOTOMP_UB]], align 4
1704 // CHECK13-NEXT: [[TMP0:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4
1705 // CHECK13-NEXT: store i32 [[TMP0]], ptr [[DOTOMP_IV]], align 4
1706 // CHECK13-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
1707 // CHECK13: omp.inner.for.cond:
1708 // CHECK13-NEXT: [[TMP1:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP6:![0-9]+]]
1709 // CHECK13-NEXT: [[TMP2:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP6]]
1710 // CHECK13-NEXT: [[CMP:%.*]] = icmp sle i32 [[TMP1]], [[TMP2]]
1711 // CHECK13-NEXT: br i1 [[CMP]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
1712 // CHECK13: omp.inner.for.body:
1713 // CHECK13-NEXT: [[TMP3:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP6]]
1714 // CHECK13-NEXT: [[DIV:%.*]] = sdiv i32 [[TMP3]], 2
1715 // CHECK13-NEXT: [[MUL:%.*]] = mul nsw i32 [[DIV]], 1
1716 // CHECK13-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
1717 // CHECK13-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP6]]
1718 // CHECK13-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP6]]
1719 // CHECK13-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP6]]
1720 // CHECK13-NEXT: [[DIV2:%.*]] = sdiv i32 [[TMP5]], 2
1721 // CHECK13-NEXT: [[MUL3:%.*]] = mul nsw i32 [[DIV2]], 2
1722 // CHECK13-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP4]], [[MUL3]]
1723 // CHECK13-NEXT: [[MUL4:%.*]] = mul nsw i32 [[SUB]], 1
1724 // CHECK13-NEXT: [[ADD5:%.*]] = add nsw i32 0, [[MUL4]]
1725 // CHECK13-NEXT: store i32 [[ADD5]], ptr [[J]], align 4, !llvm.access.group [[ACC_GRP6]]
1726 // CHECK13-NEXT: [[TMP6:%.*]] = load i32, ptr [[I]], align 4, !llvm.access.group [[ACC_GRP6]]
1727 // CHECK13-NEXT: [[IDXPROM:%.*]] = sext i32 [[TMP6]] to i64
1728 // CHECK13-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds [10 x [2 x i32]], ptr [[A]], i64 0, i64 [[IDXPROM]]
1729 // CHECK13-NEXT: [[TMP7:%.*]] = load i32, ptr [[J]], align 4, !llvm.access.group [[ACC_GRP6]]
1730 // CHECK13-NEXT: [[IDXPROM6:%.*]] = sext i32 [[TMP7]] to i64
1731 // CHECK13-NEXT: [[ARRAYIDX7:%.*]] = getelementptr inbounds [2 x i32], ptr [[ARRAYIDX]], i64 0, i64 [[IDXPROM6]]
1732 // CHECK13-NEXT: store i32 0, ptr [[ARRAYIDX7]], align 4, !llvm.access.group [[ACC_GRP6]]
1733 // CHECK13-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
1734 // CHECK13: omp.body.continue:
1735 // CHECK13-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
1736 // CHECK13: omp.inner.for.inc:
1737 // CHECK13-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP6]]
1738 // CHECK13-NEXT: [[ADD8:%.*]] = add nsw i32 [[TMP8]], 1
1739 // CHECK13-NEXT: store i32 [[ADD8]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP6]]
1740 // CHECK13-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP7:![0-9]+]]
1741 // CHECK13: omp.inner.for.end:
1742 // CHECK13-NEXT: store i32 10, ptr [[I]], align 4
1743 // CHECK13-NEXT: store i32 2, ptr [[J]], align 4
1744 // CHECK13-NEXT: ret i32 0
1745 //
1746 //
1747 // CHECK15-LABEL: define {{[^@]+}}@main
1748 // CHECK15-SAME: (i32 noundef [[ARGC:%.*]], ptr noundef [[ARGV:%.*]]) #[[ATTR0:[0-9]+]] {
1749 // CHECK15-NEXT: entry:
1750 // CHECK15-NEXT: [[RETVAL:%.*]] = alloca i32, align 4
1751 // CHECK15-NEXT: [[ARGC_ADDR:%.*]] = alloca i32, align 4
1752 // CHECK15-NEXT: [[ARGV_ADDR:%.*]] = alloca ptr, align 4
1753 // CHECK15-NEXT: [[N:%.*]] = alloca i32, align 4
1754 // CHECK15-NEXT: [[M:%.*]] = alloca i32, align 4
1755 // CHECK15-NEXT: [[SAVED_STACK:%.*]] = alloca ptr, align 4
1756 // CHECK15-NEXT: [[__VLA_EXPR0:%.*]] = alloca i32, align 4
1757 // CHECK15-NEXT: [[__VLA_EXPR1:%.*]] = alloca i32, align 4
1758 // CHECK15-NEXT: [[TMP:%.*]] = alloca i32, align 4
1759 // CHECK15-NEXT: [[_TMP1:%.*]] = alloca i32, align 4
1760 // CHECK15-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4
1761 // CHECK15-NEXT: [[DOTCAPTURE_EXPR_2:%.*]] = alloca i32, align 4
1762 // CHECK15-NEXT: [[DOTCAPTURE_EXPR_3:%.*]] = alloca i64, align 8
1763 // CHECK15-NEXT: [[DOTOMP_LB:%.*]] = alloca i64, align 8
1764 // CHECK15-NEXT: [[DOTOMP_UB:%.*]] = alloca i64, align 8
1765 // CHECK15-NEXT: [[I:%.*]] = alloca i32, align 4
1766 // CHECK15-NEXT: [[J:%.*]] = alloca i32, align 4
1767 // CHECK15-NEXT: [[DOTOMP_IV:%.*]] = alloca i64, align 8
1768 // CHECK15-NEXT: [[I9:%.*]] = alloca i32, align 4
1769 // CHECK15-NEXT: [[J10:%.*]] = alloca i32, align 4
1770 // CHECK15-NEXT: store i32 0, ptr [[RETVAL]], align 4
1771 // CHECK15-NEXT: store i32 [[ARGC]], ptr [[ARGC_ADDR]], align 4
1772 // CHECK15-NEXT: store ptr [[ARGV]], ptr [[ARGV_ADDR]], align 4
1773 // CHECK15-NEXT: store i32 100, ptr [[N]], align 4
1774 // CHECK15-NEXT: store i32 2, ptr [[M]], align 4
1775 // CHECK15-NEXT: [[TMP0:%.*]] = load i32, ptr [[N]], align 4
1776 // CHECK15-NEXT: [[TMP1:%.*]] = load i32, ptr [[M]], align 4
1777 // CHECK15-NEXT: [[TMP2:%.*]] = call ptr @llvm.stacksave.p0()
1778 // CHECK15-NEXT: store ptr [[TMP2]], ptr [[SAVED_STACK]], align 4
1779 // CHECK15-NEXT: [[TMP3:%.*]] = mul nuw i32 [[TMP0]], [[TMP1]]
1780 // CHECK15-NEXT: [[VLA:%.*]] = alloca i32, i32 [[TMP3]], align 4
1781 // CHECK15-NEXT: store i32 [[TMP0]], ptr [[__VLA_EXPR0]], align 4
1782 // CHECK15-NEXT: store i32 [[TMP1]], ptr [[__VLA_EXPR1]], align 4
1783 // CHECK15-NEXT: [[TMP4:%.*]] = load i32, ptr [[N]], align 4
1784 // CHECK15-NEXT: store i32 [[TMP4]], ptr [[DOTCAPTURE_EXPR_]], align 4
1785 // CHECK15-NEXT: [[TMP5:%.*]] = load i32, ptr [[M]], align 4
1786 // CHECK15-NEXT: store i32 [[TMP5]], ptr [[DOTCAPTURE_EXPR_2]], align 4
1787 // CHECK15-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4
1788 // CHECK15-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP6]], 0
1789 // CHECK15-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1
1790 // CHECK15-NEXT: [[CONV:%.*]] = sext i32 [[DIV]] to i64
1791 // CHECK15-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_2]], align 4
1792 // CHECK15-NEXT: [[SUB4:%.*]] = sub nsw i32 [[TMP7]], 0
1793 // CHECK15-NEXT: [[DIV5:%.*]] = sdiv i32 [[SUB4]], 1
1794 // CHECK15-NEXT: [[CONV6:%.*]] = sext i32 [[DIV5]] to i64
1795 // CHECK15-NEXT: [[MUL:%.*]] = mul nsw i64 [[CONV]], [[CONV6]]
1796 // CHECK15-NEXT: [[SUB7:%.*]] = sub nsw i64 [[MUL]], 1
1797 // CHECK15-NEXT: store i64 [[SUB7]], ptr [[DOTCAPTURE_EXPR_3]], align 8
1798 // CHECK15-NEXT: store i64 0, ptr [[DOTOMP_LB]], align 8
1799 // CHECK15-NEXT: [[TMP8:%.*]] = load i64, ptr [[DOTCAPTURE_EXPR_3]], align 8
1800 // CHECK15-NEXT: store i64 [[TMP8]], ptr [[DOTOMP_UB]], align 8
1801 // CHECK15-NEXT: store i32 0, ptr [[I]], align 4
1802 // CHECK15-NEXT: store i32 0, ptr [[J]], align 4
1803 // CHECK15-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4
1804 // CHECK15-NEXT: [[CMP:%.*]] = icmp slt i32 0, [[TMP9]]
1805 // CHECK15-NEXT: br i1 [[CMP]], label [[LAND_LHS_TRUE:%.*]], label [[SIMD_IF_END:%.*]]
1806 // CHECK15: land.lhs.true:
1807 // CHECK15-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_2]], align 4
1808 // CHECK15-NEXT: [[CMP8:%.*]] = icmp slt i32 0, [[TMP10]]
1809 // CHECK15-NEXT: br i1 [[CMP8]], label [[SIMD_IF_THEN:%.*]], label [[SIMD_IF_END]]
1810 // CHECK15: simd.if.then:
1811 // CHECK15-NEXT: [[TMP11:%.*]] = load i64, ptr [[DOTOMP_LB]], align 8
1812 // CHECK15-NEXT: store i64 [[TMP11]], ptr [[DOTOMP_IV]], align 8
1813 // CHECK15-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
1814 // CHECK15: omp.inner.for.cond:
1815 // CHECK15-NEXT: [[TMP12:%.*]] = load i64, ptr [[DOTOMP_IV]], align 8, !llvm.access.group [[ACC_GRP3:![0-9]+]]
1816 // CHECK15-NEXT: [[TMP13:%.*]] = load i64, ptr [[DOTOMP_UB]], align 8, !llvm.access.group [[ACC_GRP3]]
1817 // CHECK15-NEXT: [[CMP11:%.*]] = icmp sle i64 [[TMP12]], [[TMP13]]
1818 // CHECK15-NEXT: br i1 [[CMP11]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
1819 // CHECK15: omp.inner.for.body:
1820 // CHECK15-NEXT: [[TMP14:%.*]] = load i64, ptr [[DOTOMP_IV]], align 8, !llvm.access.group [[ACC_GRP3]]
1821 // CHECK15-NEXT: [[TMP15:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_2]], align 4, !llvm.access.group [[ACC_GRP3]]
1822 // CHECK15-NEXT: [[SUB12:%.*]] = sub nsw i32 [[TMP15]], 0
1823 // CHECK15-NEXT: [[DIV13:%.*]] = sdiv i32 [[SUB12]], 1
1824 // CHECK15-NEXT: [[MUL14:%.*]] = mul nsw i32 1, [[DIV13]]
1825 // CHECK15-NEXT: [[CONV15:%.*]] = sext i32 [[MUL14]] to i64
1826 // CHECK15-NEXT: [[DIV16:%.*]] = sdiv i64 [[TMP14]], [[CONV15]]
1827 // CHECK15-NEXT: [[MUL17:%.*]] = mul nsw i64 [[DIV16]], 1
1828 // CHECK15-NEXT: [[ADD:%.*]] = add nsw i64 0, [[MUL17]]
1829 // CHECK15-NEXT: [[CONV18:%.*]] = trunc i64 [[ADD]] to i32
1830 // CHECK15-NEXT: store i32 [[CONV18]], ptr [[I9]], align 4, !llvm.access.group [[ACC_GRP3]]
1831 // CHECK15-NEXT: [[TMP16:%.*]] = load i64, ptr [[DOTOMP_IV]], align 8, !llvm.access.group [[ACC_GRP3]]
1832 // CHECK15-NEXT: [[TMP17:%.*]] = load i64, ptr [[DOTOMP_IV]], align 8, !llvm.access.group [[ACC_GRP3]]
1833 // CHECK15-NEXT: [[TMP18:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_2]], align 4, !llvm.access.group [[ACC_GRP3]]
1834 // CHECK15-NEXT: [[SUB19:%.*]] = sub nsw i32 [[TMP18]], 0
1835 // CHECK15-NEXT: [[DIV20:%.*]] = sdiv i32 [[SUB19]], 1
1836 // CHECK15-NEXT: [[MUL21:%.*]] = mul nsw i32 1, [[DIV20]]
1837 // CHECK15-NEXT: [[CONV22:%.*]] = sext i32 [[MUL21]] to i64
1838 // CHECK15-NEXT: [[DIV23:%.*]] = sdiv i64 [[TMP17]], [[CONV22]]
1839 // CHECK15-NEXT: [[TMP19:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_2]], align 4, !llvm.access.group [[ACC_GRP3]]
1840 // CHECK15-NEXT: [[SUB24:%.*]] = sub nsw i32 [[TMP19]], 0
1841 // CHECK15-NEXT: [[DIV25:%.*]] = sdiv i32 [[SUB24]], 1
1842 // CHECK15-NEXT: [[MUL26:%.*]] = mul nsw i32 1, [[DIV25]]
1843 // CHECK15-NEXT: [[CONV27:%.*]] = sext i32 [[MUL26]] to i64
1844 // CHECK15-NEXT: [[MUL28:%.*]] = mul nsw i64 [[DIV23]], [[CONV27]]
1845 // CHECK15-NEXT: [[SUB29:%.*]] = sub nsw i64 [[TMP16]], [[MUL28]]
1846 // CHECK15-NEXT: [[MUL30:%.*]] = mul nsw i64 [[SUB29]], 1
1847 // CHECK15-NEXT: [[ADD31:%.*]] = add nsw i64 0, [[MUL30]]
1848 // CHECK15-NEXT: [[CONV32:%.*]] = trunc i64 [[ADD31]] to i32
1849 // CHECK15-NEXT: store i32 [[CONV32]], ptr [[J10]], align 4, !llvm.access.group [[ACC_GRP3]]
1850 // CHECK15-NEXT: [[TMP20:%.*]] = load i32, ptr [[I9]], align 4, !llvm.access.group [[ACC_GRP3]]
1851 // CHECK15-NEXT: [[TMP21:%.*]] = mul nsw i32 [[TMP20]], [[TMP1]]
1852 // CHECK15-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i32, ptr [[VLA]], i32 [[TMP21]]
1853 // CHECK15-NEXT: [[TMP22:%.*]] = load i32, ptr [[J10]], align 4, !llvm.access.group [[ACC_GRP3]]
1854 // CHECK15-NEXT: [[ARRAYIDX33:%.*]] = getelementptr inbounds i32, ptr [[ARRAYIDX]], i32 [[TMP22]]
1855 // CHECK15-NEXT: store i32 0, ptr [[ARRAYIDX33]], align 4, !llvm.access.group [[ACC_GRP3]]
1856 // CHECK15-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
1857 // CHECK15: omp.body.continue:
1858 // CHECK15-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
1859 // CHECK15: omp.inner.for.inc:
1860 // CHECK15-NEXT: [[TMP23:%.*]] = load i64, ptr [[DOTOMP_IV]], align 8, !llvm.access.group [[ACC_GRP3]]
1861 // CHECK15-NEXT: [[ADD34:%.*]] = add nsw i64 [[TMP23]], 1
1862 // CHECK15-NEXT: store i64 [[ADD34]], ptr [[DOTOMP_IV]], align 8, !llvm.access.group [[ACC_GRP3]]
1863 // CHECK15-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP4:![0-9]+]]
1864 // CHECK15: omp.inner.for.end:
1865 // CHECK15-NEXT: [[TMP24:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4
1866 // CHECK15-NEXT: [[SUB35:%.*]] = sub nsw i32 [[TMP24]], 0
1867 // CHECK15-NEXT: [[DIV36:%.*]] = sdiv i32 [[SUB35]], 1
1868 // CHECK15-NEXT: [[MUL37:%.*]] = mul nsw i32 [[DIV36]], 1
1869 // CHECK15-NEXT: [[ADD38:%.*]] = add nsw i32 0, [[MUL37]]
1870 // CHECK15-NEXT: store i32 [[ADD38]], ptr [[I9]], align 4
1871 // CHECK15-NEXT: [[TMP25:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_2]], align 4
1872 // CHECK15-NEXT: [[SUB39:%.*]] = sub nsw i32 [[TMP25]], 0
1873 // CHECK15-NEXT: [[DIV40:%.*]] = sdiv i32 [[SUB39]], 1
1874 // CHECK15-NEXT: [[MUL41:%.*]] = mul nsw i32 [[DIV40]], 1
1875 // CHECK15-NEXT: [[ADD42:%.*]] = add nsw i32 0, [[MUL41]]
1876 // CHECK15-NEXT: store i32 [[ADD42]], ptr [[J10]], align 4
1877 // CHECK15-NEXT: br label [[SIMD_IF_END]]
1878 // CHECK15: simd.if.end:
1879 // CHECK15-NEXT: [[TMP26:%.*]] = load i32, ptr [[ARGC_ADDR]], align 4
1880 // CHECK15-NEXT: [[CALL:%.*]] = call noundef i32 @_Z5tmainIiLi10ELi2EEiT_(i32 noundef [[TMP26]])
1881 // CHECK15-NEXT: store i32 [[CALL]], ptr [[RETVAL]], align 4
1882 // CHECK15-NEXT: [[TMP27:%.*]] = load ptr, ptr [[SAVED_STACK]], align 4
1883 // CHECK15-NEXT: call void @llvm.stackrestore.p0(ptr [[TMP27]])
1884 // CHECK15-NEXT: [[TMP28:%.*]] = load i32, ptr [[RETVAL]], align 4
1885 // CHECK15-NEXT: ret i32 [[TMP28]]
1886 //
1887 //
1888 // CHECK15-LABEL: define {{[^@]+}}@_Z5tmainIiLi10ELi2EEiT_
1889 // CHECK15-SAME: (i32 noundef [[ARGC:%.*]]) #[[ATTR2:[0-9]+]] comdat {
1890 // CHECK15-NEXT: entry:
1891 // CHECK15-NEXT: [[ARGC_ADDR:%.*]] = alloca i32, align 4
1892 // CHECK15-NEXT: [[A:%.*]] = alloca [10 x [2 x i32]], align 4
1893 // CHECK15-NEXT: [[TMP:%.*]] = alloca i32, align 4
1894 // CHECK15-NEXT: [[_TMP1:%.*]] = alloca i32, align 4
1895 // CHECK15-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4
1896 // CHECK15-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4
1897 // CHECK15-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
1898 // CHECK15-NEXT: [[I:%.*]] = alloca i32, align 4
1899 // CHECK15-NEXT: [[J:%.*]] = alloca i32, align 4
1900 // CHECK15-NEXT: store i32 [[ARGC]], ptr [[ARGC_ADDR]], align 4
1901 // CHECK15-NEXT: store i32 0, ptr [[DOTOMP_LB]], align 4
1902 // CHECK15-NEXT: store i32 19, ptr [[DOTOMP_UB]], align 4
1903 // CHECK15-NEXT: [[TMP0:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4
1904 // CHECK15-NEXT: store i32 [[TMP0]], ptr [[DOTOMP_IV]], align 4
1905 // CHECK15-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
1906 // CHECK15: omp.inner.for.cond:
1907 // CHECK15-NEXT: [[TMP1:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP7:![0-9]+]]
1908 // CHECK15-NEXT: [[TMP2:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP7]]
1909 // CHECK15-NEXT: [[CMP:%.*]] = icmp sle i32 [[TMP1]], [[TMP2]]
1910 // CHECK15-NEXT: br i1 [[CMP]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
1911 // CHECK15: omp.inner.for.body:
1912 // CHECK15-NEXT: [[TMP3:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP7]]
1913 // CHECK15-NEXT: [[DIV:%.*]] = sdiv i32 [[TMP3]], 2
1914 // CHECK15-NEXT: [[MUL:%.*]] = mul nsw i32 [[DIV]], 1
1915 // CHECK15-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
1916 // CHECK15-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP7]]
1917 // CHECK15-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP7]]
1918 // CHECK15-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP7]]
1919 // CHECK15-NEXT: [[DIV2:%.*]] = sdiv i32 [[TMP5]], 2
1920 // CHECK15-NEXT: [[MUL3:%.*]] = mul nsw i32 [[DIV2]], 2
1921 // CHECK15-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP4]], [[MUL3]]
1922 // CHECK15-NEXT: [[MUL4:%.*]] = mul nsw i32 [[SUB]], 1
1923 // CHECK15-NEXT: [[ADD5:%.*]] = add nsw i32 0, [[MUL4]]
1924 // CHECK15-NEXT: store i32 [[ADD5]], ptr [[J]], align 4, !llvm.access.group [[ACC_GRP7]]
1925 // CHECK15-NEXT: [[TMP6:%.*]] = load i32, ptr [[I]], align 4, !llvm.access.group [[ACC_GRP7]]
1926 // CHECK15-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds [10 x [2 x i32]], ptr [[A]], i32 0, i32 [[TMP6]]
1927 // CHECK15-NEXT: [[TMP7:%.*]] = load i32, ptr [[J]], align 4, !llvm.access.group [[ACC_GRP7]]
1928 // CHECK15-NEXT: [[ARRAYIDX6:%.*]] = getelementptr inbounds [2 x i32], ptr [[ARRAYIDX]], i32 0, i32 [[TMP7]]
1929 // CHECK15-NEXT: store i32 0, ptr [[ARRAYIDX6]], align 4, !llvm.access.group [[ACC_GRP7]]
1930 // CHECK15-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
1931 // CHECK15: omp.body.continue:
1932 // CHECK15-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
1933 // CHECK15: omp.inner.for.inc:
1934 // CHECK15-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP7]]
1935 // CHECK15-NEXT: [[ADD7:%.*]] = add nsw i32 [[TMP8]], 1
1936 // CHECK15-NEXT: store i32 [[ADD7]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP7]]
1937 // CHECK15-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP8:![0-9]+]]
1938 // CHECK15: omp.inner.for.end:
1939 // CHECK15-NEXT: store i32 10, ptr [[I]], align 4
1940 // CHECK15-NEXT: store i32 2, ptr [[J]], align 4
1941 // CHECK15-NEXT: ret i32 0
1942 //
LLVM monorepo