1 // NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --function-signature --include-generated-funcs --replace-value-regex "__omp_offloading_[0-9a-z]+_[0-9a-z]+" "reduction_size[.].+[.]" "pl_cond[.].+[.|,]" --prefix-filecheck-ir-name _
2 // RUN: %clang_cc1 -no-opaque-pointers -verify -fopenmp -fopenmp-version=50 -x c++ -triple x86_64-unknown-linux -emit-llvm %s -fexceptions -fcxx-exceptions -o - | FileCheck %s --check-prefix=CHECK1
3 // RUN: %clang_cc1 -no-opaque-pointers -fopenmp -fopenmp-version=50 -x c++ -std=c++11 -triple x86_64-unknown-linux -fexceptions -fcxx-exceptions -emit-pch -o %t %s
4 // RUN: %clang_cc1 -no-opaque-pointers -fopenmp -fopenmp-version=50 -x c++ -triple x86_64-unknown-linux -fexceptions -fcxx-exceptions -std=c++11 -include-pch %t -verify %s -emit-llvm -o - | FileCheck %s --check-prefix=CHECK1
5
6 // RUN: %clang_cc1 -no-opaque-pointers -triple x86_64-unknown-linux -verify -fopenmp-simd -fopenmp-version=50 -x c++ -emit-llvm %s -fexceptions -fcxx-exceptions -o - | FileCheck %s --implicit-check-not="{{__kmpc|__tgt}}"
7 // RUN: %clang_cc1 -no-opaque-pointers -fopenmp-simd -fopenmp-version=50 -x c++ -std=c++11 -triple x86_64-unknown-unknown -fexceptions -fcxx-exceptions -emit-pch -o %t %s
8 // RUN: %clang_cc1 -no-opaque-pointers -fopenmp-simd -fopenmp-version=50 -x c++ -triple x86_64-unknown-unknown -fexceptions -fcxx-exceptions -debug-info-kind=limited -std=c++11 -include-pch %t -verify %s -emit-llvm -o - | FileCheck %s --implicit-check-not="{{__kmpc|__tgt}}"
9 // expected-no-diagnostics
10 #ifndef HEADER
11 #define HEADER
12
main(int argc,char ** argv)13 int main(int argc, char **argv) {
14 #pragma omp target teams distribute parallel for reduction(task, +: argc, argv[0:10][0:argc])
15 for (long long i = 0; i < 10; ++i) {
16 #pragma omp task in_reduction(+: argc, argv[0:10][0:argc])
17 ;
18 }
19 }
20
21
22
23 // Init firstprivate copy of argc
24
25 // Init firstprivate copy of argv[0:10][0:argc]
26
27 // Register task reduction.
28
29
30
31
32
33
34
35
36
37 #endif
38 // CHECK1-LABEL: define {{[^@]+}}@main
39 // CHECK1-SAME: (i32 noundef [[ARGC:%.*]], i8** noundef [[ARGV:%.*]]) #[[ATTR0:[0-9]+]] {
40 // CHECK1-NEXT: entry:
41 // CHECK1-NEXT: [[ARGC_ADDR:%.*]] = alloca i32, align 4
42 // CHECK1-NEXT: [[ARGV_ADDR:%.*]] = alloca i8**, align 8
43 // CHECK1-NEXT: store i32 [[ARGC]], i32* [[ARGC_ADDR]], align 4
44 // CHECK1-NEXT: store i8** [[ARGV]], i8*** [[ARGV_ADDR]], align 8
45 // CHECK1-NEXT: [[TMP0:%.*]] = load i8**, i8*** [[ARGV_ADDR]], align 8
46 // CHECK1-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l14(i32* [[ARGC_ADDR]], i8** [[TMP0]]) #[[ATTR5:[0-9]+]]
47 // CHECK1-NEXT: ret i32 0
48 //
49 //
50 // CHECK1-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l14
51 // CHECK1-SAME: (i32* noundef nonnull align 4 dereferenceable(4) [[ARGC:%.*]], i8** noundef [[ARGV:%.*]]) #[[ATTR1:[0-9]+]] {
52 // CHECK1-NEXT: entry:
53 // CHECK1-NEXT: [[ARGC_ADDR:%.*]] = alloca i32*, align 8
54 // CHECK1-NEXT: [[ARGV_ADDR:%.*]] = alloca i8**, align 8
55 // CHECK1-NEXT: store i32* [[ARGC]], i32** [[ARGC_ADDR]], align 8
56 // CHECK1-NEXT: store i8** [[ARGV]], i8*** [[ARGV_ADDR]], align 8
57 // CHECK1-NEXT: [[TMP0:%.*]] = load i32*, i32** [[ARGC_ADDR]], align 8
58 // CHECK1-NEXT: [[TMP1:%.*]] = load i8**, i8*** [[ARGV_ADDR]], align 8
59 // CHECK1-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB1:[0-9]+]], i32 2, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32*, i8**)* @.omp_outlined. to void (i32*, i32*, ...)*), i32* [[TMP0]], i8** [[TMP1]])
60 // CHECK1-NEXT: ret void
61 //
62 //
63 // CHECK1-LABEL: define {{[^@]+}}@.omp_outlined.
64 // CHECK1-SAME: (i32* noalias noundef [[DOTGLOBAL_TID_:%.*]], i32* noalias noundef [[DOTBOUND_TID_:%.*]], i32* noundef nonnull align 4 dereferenceable(4) [[ARGC:%.*]], i8** noundef [[ARGV:%.*]]) #[[ATTR1]] {
65 // CHECK1-NEXT: entry:
66 // CHECK1-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
67 // CHECK1-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
68 // CHECK1-NEXT: [[ARGC_ADDR:%.*]] = alloca i32*, align 8
69 // CHECK1-NEXT: [[ARGV_ADDR:%.*]] = alloca i8**, align 8
70 // CHECK1-NEXT: [[ARGC1:%.*]] = alloca i32, align 4
71 // CHECK1-NEXT: [[SAVED_STACK:%.*]] = alloca i8*, align 8
72 // CHECK1-NEXT: [[__VLA_EXPR0:%.*]] = alloca i64, align 8
73 // CHECK1-NEXT: [[TMP:%.*]] = alloca i8**, align 8
74 // CHECK1-NEXT: [[_TMP5:%.*]] = alloca i8*, align 8
75 // CHECK1-NEXT: [[DOTRD_INPUT_:%.*]] = alloca [2 x %struct.kmp_taskred_input_t], align 8
76 // CHECK1-NEXT: [[DOTTASK_RED_:%.*]] = alloca i8*, align 8
77 // CHECK1-NEXT: [[DOTOMP_IV:%.*]] = alloca i64, align 8
78 // CHECK1-NEXT: [[_TMP12:%.*]] = alloca i64, align 8
79 // CHECK1-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i64, align 8
80 // CHECK1-NEXT: [[DOTOMP_COMB_UB:%.*]] = alloca i64, align 8
81 // CHECK1-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i64, align 8
82 // CHECK1-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
83 // CHECK1-NEXT: [[I:%.*]] = alloca i64, align 8
84 // CHECK1-NEXT: [[DOTOMP_REDUCTION_RED_LIST:%.*]] = alloca [3 x i8*], align 8
85 // CHECK1-NEXT: [[ATOMIC_TEMP:%.*]] = alloca i8, align 1
86 // CHECK1-NEXT: [[_TMP27:%.*]] = alloca i8, align 1
87 // CHECK1-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
88 // CHECK1-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
89 // CHECK1-NEXT: store i32* [[ARGC]], i32** [[ARGC_ADDR]], align 8
90 // CHECK1-NEXT: store i8** [[ARGV]], i8*** [[ARGV_ADDR]], align 8
91 // CHECK1-NEXT: [[TMP0:%.*]] = load i32*, i32** [[ARGC_ADDR]], align 8
92 // CHECK1-NEXT: store i32 0, i32* [[ARGC1]], align 4
93 // CHECK1-NEXT: [[TMP1:%.*]] = load i8**, i8*** [[ARGV_ADDR]], align 8
94 // CHECK1-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i8*, i8** [[TMP1]], i64 0
95 // CHECK1-NEXT: [[TMP2:%.*]] = load i8*, i8** [[ARRAYIDX]], align 8
96 // CHECK1-NEXT: [[ARRAYIDX2:%.*]] = getelementptr inbounds i8, i8* [[TMP2]], i64 0
97 // CHECK1-NEXT: [[TMP3:%.*]] = load i32, i32* [[TMP0]], align 4
98 // CHECK1-NEXT: [[TMP4:%.*]] = sext i32 [[TMP3]] to i64
99 // CHECK1-NEXT: [[LB_ADD_LEN:%.*]] = add nsw i64 -1, [[TMP4]]
100 // CHECK1-NEXT: [[TMP5:%.*]] = load i8**, i8*** [[ARGV_ADDR]], align 8
101 // CHECK1-NEXT: [[ARRAYIDX3:%.*]] = getelementptr inbounds i8*, i8** [[TMP5]], i64 9
102 // CHECK1-NEXT: [[TMP6:%.*]] = load i8*, i8** [[ARRAYIDX3]], align 8
103 // CHECK1-NEXT: [[ARRAYIDX4:%.*]] = getelementptr inbounds i8, i8* [[TMP6]], i64 [[LB_ADD_LEN]]
104 // CHECK1-NEXT: [[TMP7:%.*]] = ptrtoint i8* [[ARRAYIDX4]] to i64
105 // CHECK1-NEXT: [[TMP8:%.*]] = ptrtoint i8* [[ARRAYIDX2]] to i64
106 // CHECK1-NEXT: [[TMP9:%.*]] = sub i64 [[TMP7]], [[TMP8]]
107 // CHECK1-NEXT: [[TMP10:%.*]] = sdiv exact i64 [[TMP9]], ptrtoint (i8* getelementptr (i8, i8* null, i32 1) to i64)
108 // CHECK1-NEXT: [[TMP11:%.*]] = add nuw i64 [[TMP10]], 1
109 // CHECK1-NEXT: [[TMP12:%.*]] = mul nuw i64 [[TMP11]], ptrtoint (i8* getelementptr (i8, i8* null, i32 1) to i64)
110 // CHECK1-NEXT: [[TMP13:%.*]] = call i8* @llvm.stacksave()
111 // CHECK1-NEXT: store i8* [[TMP13]], i8** [[SAVED_STACK]], align 8
112 // CHECK1-NEXT: [[VLA:%.*]] = alloca i8, i64 [[TMP11]], align 16
113 // CHECK1-NEXT: store i64 [[TMP11]], i64* [[__VLA_EXPR0]], align 8
114 // CHECK1-NEXT: [[TMP14:%.*]] = getelementptr i8, i8* [[VLA]], i64 [[TMP11]]
115 // CHECK1-NEXT: [[OMP_ARRAYINIT_ISEMPTY:%.*]] = icmp eq i8* [[VLA]], [[TMP14]]
116 // CHECK1-NEXT: br i1 [[OMP_ARRAYINIT_ISEMPTY]], label [[OMP_ARRAYINIT_DONE:%.*]], label [[OMP_ARRAYINIT_BODY:%.*]]
117 // CHECK1: omp.arrayinit.body:
118 // CHECK1-NEXT: [[OMP_ARRAYCPY_DESTELEMENTPAST:%.*]] = phi i8* [ [[VLA]], [[ENTRY:%.*]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT:%.*]], [[OMP_ARRAYINIT_BODY]] ]
119 // CHECK1-NEXT: store i8 0, i8* [[OMP_ARRAYCPY_DESTELEMENTPAST]], align 1
120 // CHECK1-NEXT: [[OMP_ARRAYCPY_DEST_ELEMENT]] = getelementptr i8, i8* [[OMP_ARRAYCPY_DESTELEMENTPAST]], i32 1
121 // CHECK1-NEXT: [[OMP_ARRAYCPY_DONE:%.*]] = icmp eq i8* [[OMP_ARRAYCPY_DEST_ELEMENT]], [[TMP14]]
122 // CHECK1-NEXT: br i1 [[OMP_ARRAYCPY_DONE]], label [[OMP_ARRAYINIT_DONE]], label [[OMP_ARRAYINIT_BODY]]
123 // CHECK1: omp.arrayinit.done:
124 // CHECK1-NEXT: [[TMP15:%.*]] = load i8**, i8*** [[ARGV_ADDR]], align 8
125 // CHECK1-NEXT: [[TMP16:%.*]] = load i8*, i8** [[TMP15]], align 8
126 // CHECK1-NEXT: [[TMP17:%.*]] = ptrtoint i8* [[TMP16]] to i64
127 // CHECK1-NEXT: [[TMP18:%.*]] = ptrtoint i8* [[ARRAYIDX2]] to i64
128 // CHECK1-NEXT: [[TMP19:%.*]] = sub i64 [[TMP17]], [[TMP18]]
129 // CHECK1-NEXT: [[TMP20:%.*]] = sdiv exact i64 [[TMP19]], ptrtoint (i8* getelementptr (i8, i8* null, i32 1) to i64)
130 // CHECK1-NEXT: [[TMP21:%.*]] = getelementptr i8, i8* [[VLA]], i64 [[TMP20]]
131 // CHECK1-NEXT: store i8** [[_TMP5]], i8*** [[TMP]], align 8
132 // CHECK1-NEXT: store i8* [[TMP21]], i8** [[_TMP5]], align 8
133 // CHECK1-NEXT: [[DOTRD_INPUT_GEP_:%.*]] = getelementptr inbounds [2 x %struct.kmp_taskred_input_t], [2 x %struct.kmp_taskred_input_t]* [[DOTRD_INPUT_]], i64 0, i64 0
134 // CHECK1-NEXT: [[TMP22:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASKRED_INPUT_T:%.*]], %struct.kmp_taskred_input_t* [[DOTRD_INPUT_GEP_]], i32 0, i32 0
135 // CHECK1-NEXT: [[TMP23:%.*]] = bitcast i32* [[ARGC1]] to i8*
136 // CHECK1-NEXT: store i8* [[TMP23]], i8** [[TMP22]], align 8
137 // CHECK1-NEXT: [[TMP24:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASKRED_INPUT_T]], %struct.kmp_taskred_input_t* [[DOTRD_INPUT_GEP_]], i32 0, i32 1
138 // CHECK1-NEXT: [[TMP25:%.*]] = bitcast i32* [[TMP0]] to i8*
139 // CHECK1-NEXT: store i8* [[TMP25]], i8** [[TMP24]], align 8
140 // CHECK1-NEXT: [[TMP26:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASKRED_INPUT_T]], %struct.kmp_taskred_input_t* [[DOTRD_INPUT_GEP_]], i32 0, i32 2
141 // CHECK1-NEXT: store i64 4, i64* [[TMP26]], align 8
142 // CHECK1-NEXT: [[TMP27:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASKRED_INPUT_T]], %struct.kmp_taskred_input_t* [[DOTRD_INPUT_GEP_]], i32 0, i32 3
143 // CHECK1-NEXT: store i8* bitcast (void (i8*, i8*)* @.red_init. to i8*), i8** [[TMP27]], align 8
144 // CHECK1-NEXT: [[TMP28:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASKRED_INPUT_T]], %struct.kmp_taskred_input_t* [[DOTRD_INPUT_GEP_]], i32 0, i32 4
145 // CHECK1-NEXT: store i8* null, i8** [[TMP28]], align 8
146 // CHECK1-NEXT: [[TMP29:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASKRED_INPUT_T]], %struct.kmp_taskred_input_t* [[DOTRD_INPUT_GEP_]], i32 0, i32 5
147 // CHECK1-NEXT: store i8* bitcast (void (i8*, i8*)* @.red_comb. to i8*), i8** [[TMP29]], align 8
148 // CHECK1-NEXT: [[TMP30:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASKRED_INPUT_T]], %struct.kmp_taskred_input_t* [[DOTRD_INPUT_GEP_]], i32 0, i32 6
149 // CHECK1-NEXT: [[TMP31:%.*]] = bitcast i32* [[TMP30]] to i8*
150 // CHECK1-NEXT: call void @llvm.memset.p0i8.i64(i8* align 8 [[TMP31]], i8 0, i64 4, i1 false)
151 // CHECK1-NEXT: [[DOTRD_INPUT_GEP_6:%.*]] = getelementptr inbounds [2 x %struct.kmp_taskred_input_t], [2 x %struct.kmp_taskred_input_t]* [[DOTRD_INPUT_]], i64 0, i64 1
152 // CHECK1-NEXT: [[TMP32:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASKRED_INPUT_T]], %struct.kmp_taskred_input_t* [[DOTRD_INPUT_GEP_6]], i32 0, i32 0
153 // CHECK1-NEXT: [[TMP33:%.*]] = load i8**, i8*** [[ARGV_ADDR]], align 8
154 // CHECK1-NEXT: [[ARRAYIDX7:%.*]] = getelementptr inbounds i8*, i8** [[TMP33]], i64 0
155 // CHECK1-NEXT: [[TMP34:%.*]] = load i8*, i8** [[ARRAYIDX7]], align 8
156 // CHECK1-NEXT: [[ARRAYIDX8:%.*]] = getelementptr inbounds i8, i8* [[TMP34]], i64 0
157 // CHECK1-NEXT: [[TMP35:%.*]] = load i32, i32* [[TMP0]], align 4
158 // CHECK1-NEXT: [[TMP36:%.*]] = sext i32 [[TMP35]] to i64
159 // CHECK1-NEXT: [[LB_ADD_LEN9:%.*]] = add nsw i64 -1, [[TMP36]]
160 // CHECK1-NEXT: [[TMP37:%.*]] = load i8**, i8*** [[ARGV_ADDR]], align 8
161 // CHECK1-NEXT: [[ARRAYIDX10:%.*]] = getelementptr inbounds i8*, i8** [[TMP37]], i64 9
162 // CHECK1-NEXT: [[TMP38:%.*]] = load i8*, i8** [[ARRAYIDX10]], align 8
163 // CHECK1-NEXT: [[ARRAYIDX11:%.*]] = getelementptr inbounds i8, i8* [[TMP38]], i64 [[LB_ADD_LEN9]]
164 // CHECK1-NEXT: store i8* [[VLA]], i8** [[TMP32]], align 8
165 // CHECK1-NEXT: [[TMP39:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASKRED_INPUT_T]], %struct.kmp_taskred_input_t* [[DOTRD_INPUT_GEP_6]], i32 0, i32 1
166 // CHECK1-NEXT: store i8* [[ARRAYIDX8]], i8** [[TMP39]], align 8
167 // CHECK1-NEXT: [[TMP40:%.*]] = ptrtoint i8* [[ARRAYIDX11]] to i64
168 // CHECK1-NEXT: [[TMP41:%.*]] = ptrtoint i8* [[ARRAYIDX8]] to i64
169 // CHECK1-NEXT: [[TMP42:%.*]] = sub i64 [[TMP40]], [[TMP41]]
170 // CHECK1-NEXT: [[TMP43:%.*]] = sdiv exact i64 [[TMP42]], ptrtoint (i8* getelementptr (i8, i8* null, i32 1) to i64)
171 // CHECK1-NEXT: [[TMP44:%.*]] = add nuw i64 [[TMP43]], 1
172 // CHECK1-NEXT: [[TMP45:%.*]] = mul nuw i64 [[TMP44]], ptrtoint (i8* getelementptr (i8, i8* null, i32 1) to i64)
173 // CHECK1-NEXT: [[TMP46:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASKRED_INPUT_T]], %struct.kmp_taskred_input_t* [[DOTRD_INPUT_GEP_6]], i32 0, i32 2
174 // CHECK1-NEXT: store i64 [[TMP45]], i64* [[TMP46]], align 8
175 // CHECK1-NEXT: [[TMP47:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASKRED_INPUT_T]], %struct.kmp_taskred_input_t* [[DOTRD_INPUT_GEP_6]], i32 0, i32 3
176 // CHECK1-NEXT: store i8* bitcast (void (i8*, i8*)* @.red_init..1 to i8*), i8** [[TMP47]], align 8
177 // CHECK1-NEXT: [[TMP48:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASKRED_INPUT_T]], %struct.kmp_taskred_input_t* [[DOTRD_INPUT_GEP_6]], i32 0, i32 4
178 // CHECK1-NEXT: store i8* null, i8** [[TMP48]], align 8
179 // CHECK1-NEXT: [[TMP49:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASKRED_INPUT_T]], %struct.kmp_taskred_input_t* [[DOTRD_INPUT_GEP_6]], i32 0, i32 5
180 // CHECK1-NEXT: store i8* bitcast (void (i8*, i8*)* @.red_comb..2 to i8*), i8** [[TMP49]], align 8
181 // CHECK1-NEXT: [[TMP50:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASKRED_INPUT_T]], %struct.kmp_taskred_input_t* [[DOTRD_INPUT_GEP_6]], i32 0, i32 6
182 // CHECK1-NEXT: store i32 1, i32* [[TMP50]], align 8
183 // CHECK1-NEXT: [[TMP51:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
184 // CHECK1-NEXT: [[TMP52:%.*]] = load i32, i32* [[TMP51]], align 4
185 // CHECK1-NEXT: [[TMP53:%.*]] = bitcast [2 x %struct.kmp_taskred_input_t]* [[DOTRD_INPUT_]] to i8*
186 // CHECK1-NEXT: [[TMP54:%.*]] = call i8* @__kmpc_taskred_modifier_init(%struct.ident_t* @[[GLOB1]], i32 [[TMP52]], i32 1, i32 2, i8* [[TMP53]])
187 // CHECK1-NEXT: store i8* [[TMP54]], i8** [[DOTTASK_RED_]], align 8
188 // CHECK1-NEXT: store i64 0, i64* [[DOTOMP_COMB_LB]], align 8
189 // CHECK1-NEXT: store i64 9, i64* [[DOTOMP_COMB_UB]], align 8
190 // CHECK1-NEXT: store i64 1, i64* [[DOTOMP_STRIDE]], align 8
191 // CHECK1-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4
192 // CHECK1-NEXT: [[TMP55:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
193 // CHECK1-NEXT: [[TMP56:%.*]] = load i32, i32* [[TMP55]], align 4
194 // CHECK1-NEXT: call void @__kmpc_for_static_init_8(%struct.ident_t* @[[GLOB2:[0-9]+]], i32 [[TMP56]], i32 92, i32* [[DOTOMP_IS_LAST]], i64* [[DOTOMP_COMB_LB]], i64* [[DOTOMP_COMB_UB]], i64* [[DOTOMP_STRIDE]], i64 1, i64 1)
195 // CHECK1-NEXT: [[TMP57:%.*]] = load i64, i64* [[DOTOMP_COMB_UB]], align 8
196 // CHECK1-NEXT: [[CMP:%.*]] = icmp sgt i64 [[TMP57]], 9
197 // CHECK1-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
198 // CHECK1: cond.true:
199 // CHECK1-NEXT: br label [[COND_END:%.*]]
200 // CHECK1: cond.false:
201 // CHECK1-NEXT: [[TMP58:%.*]] = load i64, i64* [[DOTOMP_COMB_UB]], align 8
202 // CHECK1-NEXT: br label [[COND_END]]
203 // CHECK1: cond.end:
204 // CHECK1-NEXT: [[COND:%.*]] = phi i64 [ 9, [[COND_TRUE]] ], [ [[TMP58]], [[COND_FALSE]] ]
205 // CHECK1-NEXT: store i64 [[COND]], i64* [[DOTOMP_COMB_UB]], align 8
206 // CHECK1-NEXT: [[TMP59:%.*]] = load i64, i64* [[DOTOMP_COMB_LB]], align 8
207 // CHECK1-NEXT: store i64 [[TMP59]], i64* [[DOTOMP_IV]], align 8
208 // CHECK1-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
209 // CHECK1: omp.inner.for.cond:
210 // CHECK1-NEXT: [[TMP60:%.*]] = load i64, i64* [[DOTOMP_IV]], align 8
211 // CHECK1-NEXT: [[TMP61:%.*]] = load i64, i64* [[DOTOMP_COMB_UB]], align 8
212 // CHECK1-NEXT: [[CMP13:%.*]] = icmp sle i64 [[TMP60]], [[TMP61]]
213 // CHECK1-NEXT: br i1 [[CMP13]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
214 // CHECK1: omp.inner.for.body:
215 // CHECK1-NEXT: [[TMP62:%.*]] = load i64, i64* [[DOTOMP_COMB_LB]], align 8
216 // CHECK1-NEXT: [[TMP63:%.*]] = load i64, i64* [[DOTOMP_COMB_UB]], align 8
217 // CHECK1-NEXT: [[TMP64:%.*]] = load i8**, i8*** [[TMP]], align 8
218 // CHECK1-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB1]], i32 4, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64, i64, i32*, i8**)* @.omp_outlined..3 to void (i32*, i32*, ...)*), i64 [[TMP62]], i64 [[TMP63]], i32* [[ARGC1]], i8** [[TMP64]])
219 // CHECK1-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
220 // CHECK1: omp.inner.for.inc:
221 // CHECK1-NEXT: [[TMP65:%.*]] = load i64, i64* [[DOTOMP_IV]], align 8
222 // CHECK1-NEXT: [[TMP66:%.*]] = load i64, i64* [[DOTOMP_STRIDE]], align 8
223 // CHECK1-NEXT: [[ADD:%.*]] = add nsw i64 [[TMP65]], [[TMP66]]
224 // CHECK1-NEXT: store i64 [[ADD]], i64* [[DOTOMP_IV]], align 8
225 // CHECK1-NEXT: br label [[OMP_INNER_FOR_COND]]
226 // CHECK1: omp.inner.for.end:
227 // CHECK1-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
228 // CHECK1: omp.loop.exit:
229 // CHECK1-NEXT: [[TMP67:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
230 // CHECK1-NEXT: [[TMP68:%.*]] = load i32, i32* [[TMP67]], align 4
231 // CHECK1-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB2]], i32 [[TMP68]])
232 // CHECK1-NEXT: [[TMP69:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
233 // CHECK1-NEXT: [[TMP70:%.*]] = load i32, i32* [[TMP69]], align 4
234 // CHECK1-NEXT: call void @__kmpc_task_reduction_modifier_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP70]], i32 1)
235 // CHECK1-NEXT: [[TMP71:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOMP_REDUCTION_RED_LIST]], i64 0, i64 0
236 // CHECK1-NEXT: [[TMP72:%.*]] = bitcast i32* [[ARGC1]] to i8*
237 // CHECK1-NEXT: store i8* [[TMP72]], i8** [[TMP71]], align 8
238 // CHECK1-NEXT: [[TMP73:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOMP_REDUCTION_RED_LIST]], i64 0, i64 1
239 // CHECK1-NEXT: store i8* [[VLA]], i8** [[TMP73]], align 8
240 // CHECK1-NEXT: [[TMP74:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOMP_REDUCTION_RED_LIST]], i64 0, i64 2
241 // CHECK1-NEXT: [[TMP75:%.*]] = inttoptr i64 [[TMP11]] to i8*
242 // CHECK1-NEXT: store i8* [[TMP75]], i8** [[TMP74]], align 8
243 // CHECK1-NEXT: [[TMP76:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
244 // CHECK1-NEXT: [[TMP77:%.*]] = load i32, i32* [[TMP76]], align 4
245 // CHECK1-NEXT: [[TMP78:%.*]] = bitcast [3 x i8*]* [[DOTOMP_REDUCTION_RED_LIST]] to i8*
246 // CHECK1-NEXT: [[TMP79:%.*]] = call i32 @__kmpc_reduce_nowait(%struct.ident_t* @[[GLOB4:[0-9]+]], i32 [[TMP77]], i32 2, i64 24, i8* [[TMP78]], void (i8*, i8*)* @.omp.reduction.reduction_func.9, [8 x i32]* @.gomp_critical_user_.reduction.var)
247 // CHECK1-NEXT: switch i32 [[TMP79]], label [[DOTOMP_REDUCTION_DEFAULT:%.*]] [
248 // CHECK1-NEXT: i32 1, label [[DOTOMP_REDUCTION_CASE1:%.*]]
249 // CHECK1-NEXT: i32 2, label [[DOTOMP_REDUCTION_CASE2:%.*]]
250 // CHECK1-NEXT: ]
251 // CHECK1: .omp.reduction.case1:
252 // CHECK1-NEXT: [[TMP80:%.*]] = load i32, i32* [[TMP0]], align 4
253 // CHECK1-NEXT: [[TMP81:%.*]] = load i32, i32* [[ARGC1]], align 4
254 // CHECK1-NEXT: [[ADD14:%.*]] = add nsw i32 [[TMP80]], [[TMP81]]
255 // CHECK1-NEXT: store i32 [[ADD14]], i32* [[TMP0]], align 4
256 // CHECK1-NEXT: [[TMP82:%.*]] = getelementptr i8, i8* [[ARRAYIDX2]], i64 [[TMP11]]
257 // CHECK1-NEXT: [[OMP_ARRAYCPY_ISEMPTY:%.*]] = icmp eq i8* [[ARRAYIDX2]], [[TMP82]]
258 // CHECK1-NEXT: br i1 [[OMP_ARRAYCPY_ISEMPTY]], label [[OMP_ARRAYCPY_DONE21:%.*]], label [[OMP_ARRAYCPY_BODY:%.*]]
259 // CHECK1: omp.arraycpy.body:
260 // CHECK1-NEXT: [[OMP_ARRAYCPY_SRCELEMENTPAST:%.*]] = phi i8* [ [[VLA]], [[DOTOMP_REDUCTION_CASE1]] ], [ [[OMP_ARRAYCPY_SRC_ELEMENT:%.*]], [[OMP_ARRAYCPY_BODY]] ]
261 // CHECK1-NEXT: [[OMP_ARRAYCPY_DESTELEMENTPAST15:%.*]] = phi i8* [ [[ARRAYIDX2]], [[DOTOMP_REDUCTION_CASE1]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT19:%.*]], [[OMP_ARRAYCPY_BODY]] ]
262 // CHECK1-NEXT: [[TMP83:%.*]] = load i8, i8* [[OMP_ARRAYCPY_DESTELEMENTPAST15]], align 1
263 // CHECK1-NEXT: [[CONV:%.*]] = sext i8 [[TMP83]] to i32
264 // CHECK1-NEXT: [[TMP84:%.*]] = load i8, i8* [[OMP_ARRAYCPY_SRCELEMENTPAST]], align 1
265 // CHECK1-NEXT: [[CONV16:%.*]] = sext i8 [[TMP84]] to i32
266 // CHECK1-NEXT: [[ADD17:%.*]] = add nsw i32 [[CONV]], [[CONV16]]
267 // CHECK1-NEXT: [[CONV18:%.*]] = trunc i32 [[ADD17]] to i8
268 // CHECK1-NEXT: store i8 [[CONV18]], i8* [[OMP_ARRAYCPY_DESTELEMENTPAST15]], align 1
269 // CHECK1-NEXT: [[OMP_ARRAYCPY_DEST_ELEMENT19]] = getelementptr i8, i8* [[OMP_ARRAYCPY_DESTELEMENTPAST15]], i32 1
270 // CHECK1-NEXT: [[OMP_ARRAYCPY_SRC_ELEMENT]] = getelementptr i8, i8* [[OMP_ARRAYCPY_SRCELEMENTPAST]], i32 1
271 // CHECK1-NEXT: [[OMP_ARRAYCPY_DONE20:%.*]] = icmp eq i8* [[OMP_ARRAYCPY_DEST_ELEMENT19]], [[TMP82]]
272 // CHECK1-NEXT: br i1 [[OMP_ARRAYCPY_DONE20]], label [[OMP_ARRAYCPY_DONE21]], label [[OMP_ARRAYCPY_BODY]]
273 // CHECK1: omp.arraycpy.done21:
274 // CHECK1-NEXT: call void @__kmpc_end_reduce_nowait(%struct.ident_t* @[[GLOB4]], i32 [[TMP77]], [8 x i32]* @.gomp_critical_user_.reduction.var)
275 // CHECK1-NEXT: br label [[DOTOMP_REDUCTION_DEFAULT]]
276 // CHECK1: .omp.reduction.case2:
277 // CHECK1-NEXT: [[TMP85:%.*]] = load i32, i32* [[ARGC1]], align 4
278 // CHECK1-NEXT: [[TMP86:%.*]] = atomicrmw add i32* [[TMP0]], i32 [[TMP85]] monotonic, align 4
279 // CHECK1-NEXT: [[TMP87:%.*]] = getelementptr i8, i8* [[ARRAYIDX2]], i64 [[TMP11]]
280 // CHECK1-NEXT: [[OMP_ARRAYCPY_ISEMPTY22:%.*]] = icmp eq i8* [[ARRAYIDX2]], [[TMP87]]
281 // CHECK1-NEXT: br i1 [[OMP_ARRAYCPY_ISEMPTY22]], label [[OMP_ARRAYCPY_DONE35:%.*]], label [[OMP_ARRAYCPY_BODY23:%.*]]
282 // CHECK1: omp.arraycpy.body23:
283 // CHECK1-NEXT: [[OMP_ARRAYCPY_SRCELEMENTPAST24:%.*]] = phi i8* [ [[VLA]], [[DOTOMP_REDUCTION_CASE2]] ], [ [[OMP_ARRAYCPY_SRC_ELEMENT33:%.*]], [[ATOMIC_EXIT:%.*]] ]
284 // CHECK1-NEXT: [[OMP_ARRAYCPY_DESTELEMENTPAST25:%.*]] = phi i8* [ [[ARRAYIDX2]], [[DOTOMP_REDUCTION_CASE2]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT32:%.*]], [[ATOMIC_EXIT]] ]
285 // CHECK1-NEXT: [[TMP88:%.*]] = load i8, i8* [[OMP_ARRAYCPY_SRCELEMENTPAST24]], align 1
286 // CHECK1-NEXT: [[CONV26:%.*]] = sext i8 [[TMP88]] to i32
287 // CHECK1-NEXT: [[ATOMIC_LOAD:%.*]] = load atomic i8, i8* [[OMP_ARRAYCPY_DESTELEMENTPAST25]] monotonic, align 1
288 // CHECK1-NEXT: br label [[ATOMIC_CONT:%.*]]
289 // CHECK1: atomic_cont:
290 // CHECK1-NEXT: [[TMP89:%.*]] = phi i8 [ [[ATOMIC_LOAD]], [[OMP_ARRAYCPY_BODY23]] ], [ [[TMP94:%.*]], [[ATOMIC_CONT]] ]
291 // CHECK1-NEXT: store i8 [[TMP89]], i8* [[_TMP27]], align 1
292 // CHECK1-NEXT: [[TMP90:%.*]] = load i8, i8* [[_TMP27]], align 1
293 // CHECK1-NEXT: [[CONV28:%.*]] = sext i8 [[TMP90]] to i32
294 // CHECK1-NEXT: [[TMP91:%.*]] = load i8, i8* [[OMP_ARRAYCPY_SRCELEMENTPAST24]], align 1
295 // CHECK1-NEXT: [[CONV29:%.*]] = sext i8 [[TMP91]] to i32
296 // CHECK1-NEXT: [[ADD30:%.*]] = add nsw i32 [[CONV28]], [[CONV29]]
297 // CHECK1-NEXT: [[CONV31:%.*]] = trunc i32 [[ADD30]] to i8
298 // CHECK1-NEXT: store i8 [[CONV31]], i8* [[ATOMIC_TEMP]], align 1
299 // CHECK1-NEXT: [[TMP92:%.*]] = load i8, i8* [[ATOMIC_TEMP]], align 1
300 // CHECK1-NEXT: [[TMP93:%.*]] = cmpxchg i8* [[OMP_ARRAYCPY_DESTELEMENTPAST25]], i8 [[TMP89]], i8 [[TMP92]] monotonic monotonic, align 1
301 // CHECK1-NEXT: [[TMP94]] = extractvalue { i8, i1 } [[TMP93]], 0
302 // CHECK1-NEXT: [[TMP95:%.*]] = extractvalue { i8, i1 } [[TMP93]], 1
303 // CHECK1-NEXT: br i1 [[TMP95]], label [[ATOMIC_EXIT]], label [[ATOMIC_CONT]]
304 // CHECK1: atomic_exit:
305 // CHECK1-NEXT: [[OMP_ARRAYCPY_DEST_ELEMENT32]] = getelementptr i8, i8* [[OMP_ARRAYCPY_DESTELEMENTPAST25]], i32 1
306 // CHECK1-NEXT: [[OMP_ARRAYCPY_SRC_ELEMENT33]] = getelementptr i8, i8* [[OMP_ARRAYCPY_SRCELEMENTPAST24]], i32 1
307 // CHECK1-NEXT: [[OMP_ARRAYCPY_DONE34:%.*]] = icmp eq i8* [[OMP_ARRAYCPY_DEST_ELEMENT32]], [[TMP87]]
308 // CHECK1-NEXT: br i1 [[OMP_ARRAYCPY_DONE34]], label [[OMP_ARRAYCPY_DONE35]], label [[OMP_ARRAYCPY_BODY23]]
309 // CHECK1: omp.arraycpy.done35:
310 // CHECK1-NEXT: br label [[DOTOMP_REDUCTION_DEFAULT]]
311 // CHECK1: .omp.reduction.default:
312 // CHECK1-NEXT: [[TMP96:%.*]] = load i8*, i8** [[SAVED_STACK]], align 8
313 // CHECK1-NEXT: call void @llvm.stackrestore(i8* [[TMP96]])
314 // CHECK1-NEXT: ret void
315 //
316 //
317 // CHECK1-LABEL: define {{[^@]+}}@.red_init.
318 // CHECK1-SAME: (i8* noalias noundef [[TMP0:%.*]], i8* noalias noundef [[TMP1:%.*]]) #[[ATTR3:[0-9]+]] {
319 // CHECK1-NEXT: entry:
320 // CHECK1-NEXT: [[DOTADDR:%.*]] = alloca i8*, align 8
321 // CHECK1-NEXT: [[DOTADDR1:%.*]] = alloca i8*, align 8
322 // CHECK1-NEXT: store i8* [[TMP0]], i8** [[DOTADDR]], align 8
323 // CHECK1-NEXT: store i8* [[TMP1]], i8** [[DOTADDR1]], align 8
324 // CHECK1-NEXT: [[TMP2:%.*]] = bitcast i8** [[DOTADDR]] to i32**
325 // CHECK1-NEXT: [[TMP3:%.*]] = load i32*, i32** [[TMP2]], align 8
326 // CHECK1-NEXT: store i32 0, i32* [[TMP3]], align 4
327 // CHECK1-NEXT: ret void
328 //
329 //
330 // CHECK1-LABEL: define {{[^@]+}}@.red_comb.
331 // CHECK1-SAME: (i8* noundef [[TMP0:%.*]], i8* noundef [[TMP1:%.*]]) #[[ATTR3]] {
332 // CHECK1-NEXT: entry:
333 // CHECK1-NEXT: [[DOTADDR:%.*]] = alloca i8*, align 8
334 // CHECK1-NEXT: [[DOTADDR1:%.*]] = alloca i8*, align 8
335 // CHECK1-NEXT: store i8* [[TMP0]], i8** [[DOTADDR]], align 8
336 // CHECK1-NEXT: store i8* [[TMP1]], i8** [[DOTADDR1]], align 8
337 // CHECK1-NEXT: [[TMP2:%.*]] = bitcast i8** [[DOTADDR]] to i32**
338 // CHECK1-NEXT: [[TMP3:%.*]] = load i32*, i32** [[TMP2]], align 8
339 // CHECK1-NEXT: [[TMP4:%.*]] = bitcast i8** [[DOTADDR1]] to i32**
340 // CHECK1-NEXT: [[TMP5:%.*]] = load i32*, i32** [[TMP4]], align 8
341 // CHECK1-NEXT: [[TMP6:%.*]] = load i32, i32* [[TMP3]], align 4
342 // CHECK1-NEXT: [[TMP7:%.*]] = load i32, i32* [[TMP5]], align 4
343 // CHECK1-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP6]], [[TMP7]]
344 // CHECK1-NEXT: store i32 [[ADD]], i32* [[TMP3]], align 4
345 // CHECK1-NEXT: ret void
346 //
347 //
348 // CHECK1-LABEL: define {{[^@]+}}@.red_init..1
349 // CHECK1-SAME: (i8* noalias noundef [[TMP0:%.*]], i8* noalias noundef [[TMP1:%.*]]) #[[ATTR3]] {
350 // CHECK1-NEXT: entry:
351 // CHECK1-NEXT: [[DOTADDR:%.*]] = alloca i8*, align 8
352 // CHECK1-NEXT: [[DOTADDR1:%.*]] = alloca i8*, align 8
353 // CHECK1-NEXT: store i8* [[TMP0]], i8** [[DOTADDR]], align 8
354 // CHECK1-NEXT: store i8* [[TMP1]], i8** [[DOTADDR1]], align 8
355 // CHECK1-NEXT: [[TMP2:%.*]] = load i8*, i8** [[DOTADDR]], align 8
356 // CHECK1-NEXT: [[TMP3:%.*]] = load i64, i64* @{{reduction_size[.].+[.]}}, align 8
357 // CHECK1-NEXT: [[TMP4:%.*]] = getelementptr i8, i8* [[TMP2]], i64 [[TMP3]]
358 // CHECK1-NEXT: [[OMP_ARRAYINIT_ISEMPTY:%.*]] = icmp eq i8* [[TMP2]], [[TMP4]]
359 // CHECK1-NEXT: br i1 [[OMP_ARRAYINIT_ISEMPTY]], label [[OMP_ARRAYINIT_DONE:%.*]], label [[OMP_ARRAYINIT_BODY:%.*]]
360 // CHECK1: omp.arrayinit.body:
361 // CHECK1-NEXT: [[OMP_ARRAYCPY_DESTELEMENTPAST:%.*]] = phi i8* [ [[TMP2]], [[ENTRY:%.*]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT:%.*]], [[OMP_ARRAYINIT_BODY]] ]
362 // CHECK1-NEXT: store i8 0, i8* [[OMP_ARRAYCPY_DESTELEMENTPAST]], align 1
363 // CHECK1-NEXT: [[OMP_ARRAYCPY_DEST_ELEMENT]] = getelementptr i8, i8* [[OMP_ARRAYCPY_DESTELEMENTPAST]], i32 1
364 // CHECK1-NEXT: [[OMP_ARRAYCPY_DONE:%.*]] = icmp eq i8* [[OMP_ARRAYCPY_DEST_ELEMENT]], [[TMP4]]
365 // CHECK1-NEXT: br i1 [[OMP_ARRAYCPY_DONE]], label [[OMP_ARRAYINIT_DONE]], label [[OMP_ARRAYINIT_BODY]]
366 // CHECK1: omp.arrayinit.done:
367 // CHECK1-NEXT: ret void
368 //
369 //
370 // CHECK1-LABEL: define {{[^@]+}}@.red_comb..2
371 // CHECK1-SAME: (i8* noundef [[TMP0:%.*]], i8* noundef [[TMP1:%.*]]) #[[ATTR3]] {
372 // CHECK1-NEXT: entry:
373 // CHECK1-NEXT: [[DOTADDR:%.*]] = alloca i8*, align 8
374 // CHECK1-NEXT: [[DOTADDR1:%.*]] = alloca i8*, align 8
375 // CHECK1-NEXT: store i8* [[TMP0]], i8** [[DOTADDR]], align 8
376 // CHECK1-NEXT: store i8* [[TMP1]], i8** [[DOTADDR1]], align 8
377 // CHECK1-NEXT: [[TMP2:%.*]] = load i64, i64* @{{reduction_size[.].+[.]}}, align 8
378 // CHECK1-NEXT: [[TMP3:%.*]] = load i8*, i8** [[DOTADDR]], align 8
379 // CHECK1-NEXT: [[TMP4:%.*]] = load i8*, i8** [[DOTADDR1]], align 8
380 // CHECK1-NEXT: [[TMP5:%.*]] = getelementptr i8, i8* [[TMP3]], i64 [[TMP2]]
381 // CHECK1-NEXT: [[OMP_ARRAYCPY_ISEMPTY:%.*]] = icmp eq i8* [[TMP3]], [[TMP5]]
382 // CHECK1-NEXT: br i1 [[OMP_ARRAYCPY_ISEMPTY]], label [[OMP_ARRAYCPY_DONE4:%.*]], label [[OMP_ARRAYCPY_BODY:%.*]]
383 // CHECK1: omp.arraycpy.body:
384 // CHECK1-NEXT: [[OMP_ARRAYCPY_SRCELEMENTPAST:%.*]] = phi i8* [ [[TMP4]], [[ENTRY:%.*]] ], [ [[OMP_ARRAYCPY_SRC_ELEMENT:%.*]], [[OMP_ARRAYCPY_BODY]] ]
385 // CHECK1-NEXT: [[OMP_ARRAYCPY_DESTELEMENTPAST:%.*]] = phi i8* [ [[TMP3]], [[ENTRY]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT:%.*]], [[OMP_ARRAYCPY_BODY]] ]
386 // CHECK1-NEXT: [[TMP6:%.*]] = load i8, i8* [[OMP_ARRAYCPY_DESTELEMENTPAST]], align 1
387 // CHECK1-NEXT: [[CONV:%.*]] = sext i8 [[TMP6]] to i32
388 // CHECK1-NEXT: [[TMP7:%.*]] = load i8, i8* [[OMP_ARRAYCPY_SRCELEMENTPAST]], align 1
389 // CHECK1-NEXT: [[CONV2:%.*]] = sext i8 [[TMP7]] to i32
390 // CHECK1-NEXT: [[ADD:%.*]] = add nsw i32 [[CONV]], [[CONV2]]
391 // CHECK1-NEXT: [[CONV3:%.*]] = trunc i32 [[ADD]] to i8
392 // CHECK1-NEXT: store i8 [[CONV3]], i8* [[OMP_ARRAYCPY_DESTELEMENTPAST]], align 1
393 // CHECK1-NEXT: [[OMP_ARRAYCPY_DEST_ELEMENT]] = getelementptr i8, i8* [[OMP_ARRAYCPY_DESTELEMENTPAST]], i32 1
394 // CHECK1-NEXT: [[OMP_ARRAYCPY_SRC_ELEMENT]] = getelementptr i8, i8* [[OMP_ARRAYCPY_SRCELEMENTPAST]], i32 1
395 // CHECK1-NEXT: [[OMP_ARRAYCPY_DONE:%.*]] = icmp eq i8* [[OMP_ARRAYCPY_DEST_ELEMENT]], [[TMP5]]
396 // CHECK1-NEXT: br i1 [[OMP_ARRAYCPY_DONE]], label [[OMP_ARRAYCPY_DONE4]], label [[OMP_ARRAYCPY_BODY]]
397 // CHECK1: omp.arraycpy.done4:
398 // CHECK1-NEXT: ret void
399 //
400 //
401 // CHECK1-LABEL: define {{[^@]+}}@.omp_outlined..3
402 // CHECK1-SAME: (i32* noalias noundef [[DOTGLOBAL_TID_:%.*]], i32* noalias noundef [[DOTBOUND_TID_:%.*]], i64 noundef [[DOTPREVIOUS_LB_:%.*]], i64 noundef [[DOTPREVIOUS_UB_:%.*]], i32* noundef nonnull align 4 dereferenceable(4) [[ARGC:%.*]], i8** noundef [[ARGV:%.*]]) #[[ATTR1]] {
403 // CHECK1-NEXT: entry:
404 // CHECK1-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
405 // CHECK1-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
406 // CHECK1-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i64, align 8
407 // CHECK1-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i64, align 8
408 // CHECK1-NEXT: [[ARGC_ADDR:%.*]] = alloca i32*, align 8
409 // CHECK1-NEXT: [[ARGV_ADDR:%.*]] = alloca i8**, align 8
410 // CHECK1-NEXT: [[DOTOMP_IV:%.*]] = alloca i64, align 8
411 // CHECK1-NEXT: [[TMP:%.*]] = alloca i64, align 8
412 // CHECK1-NEXT: [[DOTOMP_LB:%.*]] = alloca i64, align 8
413 // CHECK1-NEXT: [[DOTOMP_UB:%.*]] = alloca i64, align 8
414 // CHECK1-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i64, align 8
415 // CHECK1-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
416 // CHECK1-NEXT: [[ARGC1:%.*]] = alloca i32, align 4
417 // CHECK1-NEXT: [[SAVED_STACK:%.*]] = alloca i8*, align 8
418 // CHECK1-NEXT: [[__VLA_EXPR0:%.*]] = alloca i64, align 8
419 // CHECK1-NEXT: [[_TMP5:%.*]] = alloca i8**, align 8
420 // CHECK1-NEXT: [[_TMP6:%.*]] = alloca i8*, align 8
421 // CHECK1-NEXT: [[DOTRD_INPUT_:%.*]] = alloca [2 x %struct.kmp_taskred_input_t.0], align 8
422 // CHECK1-NEXT: [[DOTTASK_RED_:%.*]] = alloca i8*, align 8
423 // CHECK1-NEXT: [[I:%.*]] = alloca i64, align 8
424 // CHECK1-NEXT: [[AGG_CAPTURED:%.*]] = alloca [[STRUCT_ANON:%.*]], align 8
425 // CHECK1-NEXT: [[DOTOMP_REDUCTION_RED_LIST:%.*]] = alloca [3 x i8*], align 8
426 // CHECK1-NEXT: [[ATOMIC_TEMP:%.*]] = alloca i8, align 1
427 // CHECK1-NEXT: [[_TMP28:%.*]] = alloca i8, align 1
428 // CHECK1-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
429 // CHECK1-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
430 // CHECK1-NEXT: store i64 [[DOTPREVIOUS_LB_]], i64* [[DOTPREVIOUS_LB__ADDR]], align 8
431 // CHECK1-NEXT: store i64 [[DOTPREVIOUS_UB_]], i64* [[DOTPREVIOUS_UB__ADDR]], align 8
432 // CHECK1-NEXT: store i32* [[ARGC]], i32** [[ARGC_ADDR]], align 8
433 // CHECK1-NEXT: store i8** [[ARGV]], i8*** [[ARGV_ADDR]], align 8
434 // CHECK1-NEXT: [[TMP0:%.*]] = load i32*, i32** [[ARGC_ADDR]], align 8
435 // CHECK1-NEXT: store i64 0, i64* [[DOTOMP_LB]], align 8
436 // CHECK1-NEXT: store i64 9, i64* [[DOTOMP_UB]], align 8
437 // CHECK1-NEXT: [[TMP1:%.*]] = load i64, i64* [[DOTPREVIOUS_LB__ADDR]], align 8
438 // CHECK1-NEXT: [[TMP2:%.*]] = load i64, i64* [[DOTPREVIOUS_UB__ADDR]], align 8
439 // CHECK1-NEXT: store i64 [[TMP1]], i64* [[DOTOMP_LB]], align 8
440 // CHECK1-NEXT: store i64 [[TMP2]], i64* [[DOTOMP_UB]], align 8
441 // CHECK1-NEXT: store i64 1, i64* [[DOTOMP_STRIDE]], align 8
442 // CHECK1-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4
443 // CHECK1-NEXT: store i32 0, i32* [[ARGC1]], align 4
444 // CHECK1-NEXT: [[TMP3:%.*]] = load i8**, i8*** [[ARGV_ADDR]], align 8
445 // CHECK1-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i8*, i8** [[TMP3]], i64 0
446 // CHECK1-NEXT: [[TMP4:%.*]] = load i8*, i8** [[ARRAYIDX]], align 8
447 // CHECK1-NEXT: [[ARRAYIDX2:%.*]] = getelementptr inbounds i8, i8* [[TMP4]], i64 0
448 // CHECK1-NEXT: [[TMP5:%.*]] = load i32, i32* [[TMP0]], align 4
449 // CHECK1-NEXT: [[TMP6:%.*]] = sext i32 [[TMP5]] to i64
450 // CHECK1-NEXT: [[LB_ADD_LEN:%.*]] = add nsw i64 -1, [[TMP6]]
451 // CHECK1-NEXT: [[TMP7:%.*]] = load i8**, i8*** [[ARGV_ADDR]], align 8
452 // CHECK1-NEXT: [[ARRAYIDX3:%.*]] = getelementptr inbounds i8*, i8** [[TMP7]], i64 9
453 // CHECK1-NEXT: [[TMP8:%.*]] = load i8*, i8** [[ARRAYIDX3]], align 8
454 // CHECK1-NEXT: [[ARRAYIDX4:%.*]] = getelementptr inbounds i8, i8* [[TMP8]], i64 [[LB_ADD_LEN]]
455 // CHECK1-NEXT: [[TMP9:%.*]] = ptrtoint i8* [[ARRAYIDX4]] to i64
456 // CHECK1-NEXT: [[TMP10:%.*]] = ptrtoint i8* [[ARRAYIDX2]] to i64
457 // CHECK1-NEXT: [[TMP11:%.*]] = sub i64 [[TMP9]], [[TMP10]]
458 // CHECK1-NEXT: [[TMP12:%.*]] = sdiv exact i64 [[TMP11]], ptrtoint (i8* getelementptr (i8, i8* null, i32 1) to i64)
459 // CHECK1-NEXT: [[TMP13:%.*]] = add nuw i64 [[TMP12]], 1
460 // CHECK1-NEXT: [[TMP14:%.*]] = mul nuw i64 [[TMP13]], ptrtoint (i8* getelementptr (i8, i8* null, i32 1) to i64)
461 // CHECK1-NEXT: [[TMP15:%.*]] = call i8* @llvm.stacksave()
462 // CHECK1-NEXT: store i8* [[TMP15]], i8** [[SAVED_STACK]], align 8
463 // CHECK1-NEXT: [[VLA:%.*]] = alloca i8, i64 [[TMP13]], align 16
464 // CHECK1-NEXT: store i64 [[TMP13]], i64* [[__VLA_EXPR0]], align 8
465 // CHECK1-NEXT: [[TMP16:%.*]] = getelementptr i8, i8* [[VLA]], i64 [[TMP13]]
466 // CHECK1-NEXT: [[OMP_ARRAYINIT_ISEMPTY:%.*]] = icmp eq i8* [[VLA]], [[TMP16]]
467 // CHECK1-NEXT: br i1 [[OMP_ARRAYINIT_ISEMPTY]], label [[OMP_ARRAYINIT_DONE:%.*]], label [[OMP_ARRAYINIT_BODY:%.*]]
468 // CHECK1: omp.arrayinit.body:
469 // CHECK1-NEXT: [[OMP_ARRAYCPY_DESTELEMENTPAST:%.*]] = phi i8* [ [[VLA]], [[ENTRY:%.*]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT:%.*]], [[OMP_ARRAYINIT_BODY]] ]
470 // CHECK1-NEXT: store i8 0, i8* [[OMP_ARRAYCPY_DESTELEMENTPAST]], align 1
471 // CHECK1-NEXT: [[OMP_ARRAYCPY_DEST_ELEMENT]] = getelementptr i8, i8* [[OMP_ARRAYCPY_DESTELEMENTPAST]], i32 1
472 // CHECK1-NEXT: [[OMP_ARRAYCPY_DONE:%.*]] = icmp eq i8* [[OMP_ARRAYCPY_DEST_ELEMENT]], [[TMP16]]
473 // CHECK1-NEXT: br i1 [[OMP_ARRAYCPY_DONE]], label [[OMP_ARRAYINIT_DONE]], label [[OMP_ARRAYINIT_BODY]]
474 // CHECK1: omp.arrayinit.done:
475 // CHECK1-NEXT: [[TMP17:%.*]] = load i8**, i8*** [[ARGV_ADDR]], align 8
476 // CHECK1-NEXT: [[TMP18:%.*]] = load i8*, i8** [[TMP17]], align 8
477 // CHECK1-NEXT: [[TMP19:%.*]] = ptrtoint i8* [[TMP18]] to i64
478 // CHECK1-NEXT: [[TMP20:%.*]] = ptrtoint i8* [[ARRAYIDX2]] to i64
479 // CHECK1-NEXT: [[TMP21:%.*]] = sub i64 [[TMP19]], [[TMP20]]
480 // CHECK1-NEXT: [[TMP22:%.*]] = sdiv exact i64 [[TMP21]], ptrtoint (i8* getelementptr (i8, i8* null, i32 1) to i64)
481 // CHECK1-NEXT: [[TMP23:%.*]] = getelementptr i8, i8* [[VLA]], i64 [[TMP22]]
482 // CHECK1-NEXT: store i8** [[_TMP6]], i8*** [[_TMP5]], align 8
483 // CHECK1-NEXT: store i8* [[TMP23]], i8** [[_TMP6]], align 8
484 // CHECK1-NEXT: [[DOTRD_INPUT_GEP_:%.*]] = getelementptr inbounds [2 x %struct.kmp_taskred_input_t.0], [2 x %struct.kmp_taskred_input_t.0]* [[DOTRD_INPUT_]], i64 0, i64 0
485 // CHECK1-NEXT: [[TMP24:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASKRED_INPUT_T_0:%.*]], %struct.kmp_taskred_input_t.0* [[DOTRD_INPUT_GEP_]], i32 0, i32 0
486 // CHECK1-NEXT: [[TMP25:%.*]] = bitcast i32* [[ARGC1]] to i8*
487 // CHECK1-NEXT: store i8* [[TMP25]], i8** [[TMP24]], align 8
488 // CHECK1-NEXT: [[TMP26:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASKRED_INPUT_T_0]], %struct.kmp_taskred_input_t.0* [[DOTRD_INPUT_GEP_]], i32 0, i32 1
489 // CHECK1-NEXT: [[TMP27:%.*]] = bitcast i32* [[TMP0]] to i8*
490 // CHECK1-NEXT: store i8* [[TMP27]], i8** [[TMP26]], align 8
491 // CHECK1-NEXT: [[TMP28:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASKRED_INPUT_T_0]], %struct.kmp_taskred_input_t.0* [[DOTRD_INPUT_GEP_]], i32 0, i32 2
492 // CHECK1-NEXT: store i64 4, i64* [[TMP28]], align 8
493 // CHECK1-NEXT: [[TMP29:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASKRED_INPUT_T_0]], %struct.kmp_taskred_input_t.0* [[DOTRD_INPUT_GEP_]], i32 0, i32 3
494 // CHECK1-NEXT: store i8* bitcast (void (i8*, i8*)* @.red_init..4 to i8*), i8** [[TMP29]], align 8
495 // CHECK1-NEXT: [[TMP30:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASKRED_INPUT_T_0]], %struct.kmp_taskred_input_t.0* [[DOTRD_INPUT_GEP_]], i32 0, i32 4
496 // CHECK1-NEXT: store i8* null, i8** [[TMP30]], align 8
497 // CHECK1-NEXT: [[TMP31:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASKRED_INPUT_T_0]], %struct.kmp_taskred_input_t.0* [[DOTRD_INPUT_GEP_]], i32 0, i32 5
498 // CHECK1-NEXT: store i8* bitcast (void (i8*, i8*)* @.red_comb..5 to i8*), i8** [[TMP31]], align 8
499 // CHECK1-NEXT: [[TMP32:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASKRED_INPUT_T_0]], %struct.kmp_taskred_input_t.0* [[DOTRD_INPUT_GEP_]], i32 0, i32 6
500 // CHECK1-NEXT: [[TMP33:%.*]] = bitcast i32* [[TMP32]] to i8*
501 // CHECK1-NEXT: call void @llvm.memset.p0i8.i64(i8* align 8 [[TMP33]], i8 0, i64 4, i1 false)
502 // CHECK1-NEXT: [[DOTRD_INPUT_GEP_7:%.*]] = getelementptr inbounds [2 x %struct.kmp_taskred_input_t.0], [2 x %struct.kmp_taskred_input_t.0]* [[DOTRD_INPUT_]], i64 0, i64 1
503 // CHECK1-NEXT: [[TMP34:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASKRED_INPUT_T_0]], %struct.kmp_taskred_input_t.0* [[DOTRD_INPUT_GEP_7]], i32 0, i32 0
504 // CHECK1-NEXT: [[TMP35:%.*]] = load i8**, i8*** [[ARGV_ADDR]], align 8
505 // CHECK1-NEXT: [[ARRAYIDX8:%.*]] = getelementptr inbounds i8*, i8** [[TMP35]], i64 0
506 // CHECK1-NEXT: [[TMP36:%.*]] = load i8*, i8** [[ARRAYIDX8]], align 8
507 // CHECK1-NEXT: [[ARRAYIDX9:%.*]] = getelementptr inbounds i8, i8* [[TMP36]], i64 0
508 // CHECK1-NEXT: [[TMP37:%.*]] = load i32, i32* [[TMP0]], align 4
509 // CHECK1-NEXT: [[TMP38:%.*]] = sext i32 [[TMP37]] to i64
510 // CHECK1-NEXT: [[LB_ADD_LEN10:%.*]] = add nsw i64 -1, [[TMP38]]
511 // CHECK1-NEXT: [[TMP39:%.*]] = load i8**, i8*** [[ARGV_ADDR]], align 8
512 // CHECK1-NEXT: [[ARRAYIDX11:%.*]] = getelementptr inbounds i8*, i8** [[TMP39]], i64 9
513 // CHECK1-NEXT: [[TMP40:%.*]] = load i8*, i8** [[ARRAYIDX11]], align 8
514 // CHECK1-NEXT: [[ARRAYIDX12:%.*]] = getelementptr inbounds i8, i8* [[TMP40]], i64 [[LB_ADD_LEN10]]
515 // CHECK1-NEXT: store i8* [[VLA]], i8** [[TMP34]], align 8
516 // CHECK1-NEXT: [[TMP41:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASKRED_INPUT_T_0]], %struct.kmp_taskred_input_t.0* [[DOTRD_INPUT_GEP_7]], i32 0, i32 1
517 // CHECK1-NEXT: store i8* [[ARRAYIDX9]], i8** [[TMP41]], align 8
518 // CHECK1-NEXT: [[TMP42:%.*]] = ptrtoint i8* [[ARRAYIDX12]] to i64
519 // CHECK1-NEXT: [[TMP43:%.*]] = ptrtoint i8* [[ARRAYIDX9]] to i64
520 // CHECK1-NEXT: [[TMP44:%.*]] = sub i64 [[TMP42]], [[TMP43]]
521 // CHECK1-NEXT: [[TMP45:%.*]] = sdiv exact i64 [[TMP44]], ptrtoint (i8* getelementptr (i8, i8* null, i32 1) to i64)
522 // CHECK1-NEXT: [[TMP46:%.*]] = add nuw i64 [[TMP45]], 1
523 // CHECK1-NEXT: [[TMP47:%.*]] = mul nuw i64 [[TMP46]], ptrtoint (i8* getelementptr (i8, i8* null, i32 1) to i64)
524 // CHECK1-NEXT: [[TMP48:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASKRED_INPUT_T_0]], %struct.kmp_taskred_input_t.0* [[DOTRD_INPUT_GEP_7]], i32 0, i32 2
525 // CHECK1-NEXT: store i64 [[TMP47]], i64* [[TMP48]], align 8
526 // CHECK1-NEXT: [[TMP49:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASKRED_INPUT_T_0]], %struct.kmp_taskred_input_t.0* [[DOTRD_INPUT_GEP_7]], i32 0, i32 3
527 // CHECK1-NEXT: store i8* bitcast (void (i8*, i8*)* @.red_init..6 to i8*), i8** [[TMP49]], align 8
528 // CHECK1-NEXT: [[TMP50:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASKRED_INPUT_T_0]], %struct.kmp_taskred_input_t.0* [[DOTRD_INPUT_GEP_7]], i32 0, i32 4
529 // CHECK1-NEXT: store i8* null, i8** [[TMP50]], align 8
530 // CHECK1-NEXT: [[TMP51:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASKRED_INPUT_T_0]], %struct.kmp_taskred_input_t.0* [[DOTRD_INPUT_GEP_7]], i32 0, i32 5
531 // CHECK1-NEXT: store i8* bitcast (void (i8*, i8*)* @.red_comb..7 to i8*), i8** [[TMP51]], align 8
532 // CHECK1-NEXT: [[TMP52:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASKRED_INPUT_T_0]], %struct.kmp_taskred_input_t.0* [[DOTRD_INPUT_GEP_7]], i32 0, i32 6
533 // CHECK1-NEXT: store i32 1, i32* [[TMP52]], align 8
534 // CHECK1-NEXT: [[TMP53:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
535 // CHECK1-NEXT: [[TMP54:%.*]] = load i32, i32* [[TMP53]], align 4
536 // CHECK1-NEXT: [[TMP55:%.*]] = bitcast [2 x %struct.kmp_taskred_input_t.0]* [[DOTRD_INPUT_]] to i8*
537 // CHECK1-NEXT: [[TMP56:%.*]] = call i8* @__kmpc_taskred_modifier_init(%struct.ident_t* @[[GLOB1]], i32 [[TMP54]], i32 1, i32 2, i8* [[TMP55]])
538 // CHECK1-NEXT: store i8* [[TMP56]], i8** [[DOTTASK_RED_]], align 8
539 // CHECK1-NEXT: [[TMP57:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
540 // CHECK1-NEXT: [[TMP58:%.*]] = load i32, i32* [[TMP57]], align 4
541 // CHECK1-NEXT: call void @__kmpc_for_static_init_8(%struct.ident_t* @[[GLOB3:[0-9]+]], i32 [[TMP58]], i32 34, i32* [[DOTOMP_IS_LAST]], i64* [[DOTOMP_LB]], i64* [[DOTOMP_UB]], i64* [[DOTOMP_STRIDE]], i64 1, i64 1)
542 // CHECK1-NEXT: [[TMP59:%.*]] = load i64, i64* [[DOTOMP_UB]], align 8
543 // CHECK1-NEXT: [[CMP:%.*]] = icmp sgt i64 [[TMP59]], 9
544 // CHECK1-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
545 // CHECK1: cond.true:
546 // CHECK1-NEXT: br label [[COND_END:%.*]]
547 // CHECK1: cond.false:
548 // CHECK1-NEXT: [[TMP60:%.*]] = load i64, i64* [[DOTOMP_UB]], align 8
549 // CHECK1-NEXT: br label [[COND_END]]
550 // CHECK1: cond.end:
551 // CHECK1-NEXT: [[COND:%.*]] = phi i64 [ 9, [[COND_TRUE]] ], [ [[TMP60]], [[COND_FALSE]] ]
552 // CHECK1-NEXT: store i64 [[COND]], i64* [[DOTOMP_UB]], align 8
553 // CHECK1-NEXT: [[TMP61:%.*]] = load i64, i64* [[DOTOMP_LB]], align 8
554 // CHECK1-NEXT: store i64 [[TMP61]], i64* [[DOTOMP_IV]], align 8
555 // CHECK1-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
556 // CHECK1: omp.inner.for.cond:
557 // CHECK1-NEXT: [[TMP62:%.*]] = load i64, i64* [[DOTOMP_IV]], align 8
558 // CHECK1-NEXT: [[TMP63:%.*]] = load i64, i64* [[DOTOMP_UB]], align 8
559 // CHECK1-NEXT: [[CMP13:%.*]] = icmp sle i64 [[TMP62]], [[TMP63]]
560 // CHECK1-NEXT: br i1 [[CMP13]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_COND_CLEANUP:%.*]]
561 // CHECK1: omp.inner.for.cond.cleanup:
562 // CHECK1-NEXT: br label [[OMP_INNER_FOR_END:%.*]]
563 // CHECK1: omp.inner.for.body:
564 // CHECK1-NEXT: [[TMP64:%.*]] = load i64, i64* [[DOTOMP_IV]], align 8
565 // CHECK1-NEXT: [[MUL:%.*]] = mul nsw i64 [[TMP64]], 1
566 // CHECK1-NEXT: [[ADD:%.*]] = add nsw i64 0, [[MUL]]
567 // CHECK1-NEXT: store i64 [[ADD]], i64* [[I]], align 8
568 // CHECK1-NEXT: [[TMP65:%.*]] = getelementptr inbounds [[STRUCT_ANON]], %struct.anon* [[AGG_CAPTURED]], i32 0, i32 0
569 // CHECK1-NEXT: store i8** [[DOTTASK_RED_]], i8*** [[TMP65]], align 8
570 // CHECK1-NEXT: [[TMP66:%.*]] = getelementptr inbounds [[STRUCT_ANON]], %struct.anon* [[AGG_CAPTURED]], i32 0, i32 1
571 // CHECK1-NEXT: store i32* [[ARGC1]], i32** [[TMP66]], align 8
572 // CHECK1-NEXT: [[TMP67:%.*]] = getelementptr inbounds [[STRUCT_ANON]], %struct.anon* [[AGG_CAPTURED]], i32 0, i32 2
573 // CHECK1-NEXT: [[TMP68:%.*]] = load i8**, i8*** [[_TMP5]], align 8
574 // CHECK1-NEXT: store i8** [[TMP68]], i8*** [[TMP67]], align 8
575 // CHECK1-NEXT: [[TMP69:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
576 // CHECK1-NEXT: [[TMP70:%.*]] = load i32, i32* [[TMP69]], align 4
577 // CHECK1-NEXT: [[TMP71:%.*]] = call i8* @__kmpc_omp_task_alloc(%struct.ident_t* @[[GLOB1]], i32 [[TMP70]], i32 1, i64 48, i64 24, i32 (i32, i8*)* bitcast (i32 (i32, %struct.kmp_task_t_with_privates*)* @.omp_task_entry. to i32 (i32, i8*)*))
578 // CHECK1-NEXT: [[TMP72:%.*]] = bitcast i8* [[TMP71]] to %struct.kmp_task_t_with_privates*
579 // CHECK1-NEXT: [[TMP73:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASK_T_WITH_PRIVATES:%.*]], %struct.kmp_task_t_with_privates* [[TMP72]], i32 0, i32 0
580 // CHECK1-NEXT: [[TMP74:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASK_T:%.*]], %struct.kmp_task_t* [[TMP73]], i32 0, i32 0
581 // CHECK1-NEXT: [[TMP75:%.*]] = load i8*, i8** [[TMP74]], align 8
582 // CHECK1-NEXT: [[TMP76:%.*]] = bitcast %struct.anon* [[AGG_CAPTURED]] to i8*
583 // CHECK1-NEXT: call void @llvm.memcpy.p0i8.p0i8.i64(i8* align 8 [[TMP75]], i8* align 8 [[TMP76]], i64 24, i1 false)
584 // CHECK1-NEXT: [[TMP77:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASK_T_WITH_PRIVATES]], %struct.kmp_task_t_with_privates* [[TMP72]], i32 0, i32 1
585 // CHECK1-NEXT: [[TMP78:%.*]] = getelementptr inbounds [[STRUCT__KMP_PRIVATES_T:%.*]], %struct..kmp_privates.t* [[TMP77]], i32 0, i32 0
586 // CHECK1-NEXT: [[TMP79:%.*]] = load i8*, i8** [[DOTTASK_RED_]], align 8
587 // CHECK1-NEXT: store i8* [[TMP79]], i8** [[TMP78]], align 8
588 // CHECK1-NEXT: [[TMP80:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
589 // CHECK1-NEXT: [[TMP81:%.*]] = load i32, i32* [[TMP80]], align 4
590 // CHECK1-NEXT: [[TMP82:%.*]] = call i32 @__kmpc_omp_task(%struct.ident_t* @[[GLOB1]], i32 [[TMP81]], i8* [[TMP71]])
591 // CHECK1-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
592 // CHECK1: omp.body.continue:
593 // CHECK1-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
594 // CHECK1: omp.inner.for.inc:
595 // CHECK1-NEXT: [[TMP83:%.*]] = load i64, i64* [[DOTOMP_IV]], align 8
596 // CHECK1-NEXT: [[ADD14:%.*]] = add nsw i64 [[TMP83]], 1
597 // CHECK1-NEXT: store i64 [[ADD14]], i64* [[DOTOMP_IV]], align 8
598 // CHECK1-NEXT: br label [[OMP_INNER_FOR_COND]]
599 // CHECK1: omp.inner.for.end:
600 // CHECK1-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
601 // CHECK1: omp.loop.exit:
602 // CHECK1-NEXT: [[TMP84:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
603 // CHECK1-NEXT: [[TMP85:%.*]] = load i32, i32* [[TMP84]], align 4
604 // CHECK1-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB2]], i32 [[TMP85]])
605 // CHECK1-NEXT: [[TMP86:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
606 // CHECK1-NEXT: [[TMP87:%.*]] = load i32, i32* [[TMP86]], align 4
607 // CHECK1-NEXT: call void @__kmpc_task_reduction_modifier_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP87]], i32 1)
608 // CHECK1-NEXT: [[TMP88:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOMP_REDUCTION_RED_LIST]], i64 0, i64 0
609 // CHECK1-NEXT: [[TMP89:%.*]] = bitcast i32* [[ARGC1]] to i8*
610 // CHECK1-NEXT: store i8* [[TMP89]], i8** [[TMP88]], align 8
611 // CHECK1-NEXT: [[TMP90:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOMP_REDUCTION_RED_LIST]], i64 0, i64 1
612 // CHECK1-NEXT: store i8* [[VLA]], i8** [[TMP90]], align 8
613 // CHECK1-NEXT: [[TMP91:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOMP_REDUCTION_RED_LIST]], i64 0, i64 2
614 // CHECK1-NEXT: [[TMP92:%.*]] = inttoptr i64 [[TMP13]] to i8*
615 // CHECK1-NEXT: store i8* [[TMP92]], i8** [[TMP91]], align 8
616 // CHECK1-NEXT: [[TMP93:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
617 // CHECK1-NEXT: [[TMP94:%.*]] = load i32, i32* [[TMP93]], align 4
618 // CHECK1-NEXT: [[TMP95:%.*]] = bitcast [3 x i8*]* [[DOTOMP_REDUCTION_RED_LIST]] to i8*
619 // CHECK1-NEXT: [[TMP96:%.*]] = call i32 @__kmpc_reduce_nowait(%struct.ident_t* @[[GLOB4]], i32 [[TMP94]], i32 2, i64 24, i8* [[TMP95]], void (i8*, i8*)* @.omp.reduction.reduction_func, [8 x i32]* @.gomp_critical_user_.reduction.var)
620 // CHECK1-NEXT: switch i32 [[TMP96]], label [[DOTOMP_REDUCTION_DEFAULT:%.*]] [
621 // CHECK1-NEXT: i32 1, label [[DOTOMP_REDUCTION_CASE1:%.*]]
622 // CHECK1-NEXT: i32 2, label [[DOTOMP_REDUCTION_CASE2:%.*]]
623 // CHECK1-NEXT: ]
624 // CHECK1: .omp.reduction.case1:
625 // CHECK1-NEXT: [[TMP97:%.*]] = load i32, i32* [[TMP0]], align 4
626 // CHECK1-NEXT: [[TMP98:%.*]] = load i32, i32* [[ARGC1]], align 4
627 // CHECK1-NEXT: [[ADD15:%.*]] = add nsw i32 [[TMP97]], [[TMP98]]
628 // CHECK1-NEXT: store i32 [[ADD15]], i32* [[TMP0]], align 4
629 // CHECK1-NEXT: [[TMP99:%.*]] = getelementptr i8, i8* [[ARRAYIDX2]], i64 [[TMP13]]
630 // CHECK1-NEXT: [[OMP_ARRAYCPY_ISEMPTY:%.*]] = icmp eq i8* [[ARRAYIDX2]], [[TMP99]]
631 // CHECK1-NEXT: br i1 [[OMP_ARRAYCPY_ISEMPTY]], label [[OMP_ARRAYCPY_DONE22:%.*]], label [[OMP_ARRAYCPY_BODY:%.*]]
632 // CHECK1: omp.arraycpy.body:
633 // CHECK1-NEXT: [[OMP_ARRAYCPY_SRCELEMENTPAST:%.*]] = phi i8* [ [[VLA]], [[DOTOMP_REDUCTION_CASE1]] ], [ [[OMP_ARRAYCPY_SRC_ELEMENT:%.*]], [[OMP_ARRAYCPY_BODY]] ]
634 // CHECK1-NEXT: [[OMP_ARRAYCPY_DESTELEMENTPAST16:%.*]] = phi i8* [ [[ARRAYIDX2]], [[DOTOMP_REDUCTION_CASE1]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT20:%.*]], [[OMP_ARRAYCPY_BODY]] ]
635 // CHECK1-NEXT: [[TMP100:%.*]] = load i8, i8* [[OMP_ARRAYCPY_DESTELEMENTPAST16]], align 1
636 // CHECK1-NEXT: [[CONV:%.*]] = sext i8 [[TMP100]] to i32
637 // CHECK1-NEXT: [[TMP101:%.*]] = load i8, i8* [[OMP_ARRAYCPY_SRCELEMENTPAST]], align 1
638 // CHECK1-NEXT: [[CONV17:%.*]] = sext i8 [[TMP101]] to i32
639 // CHECK1-NEXT: [[ADD18:%.*]] = add nsw i32 [[CONV]], [[CONV17]]
640 // CHECK1-NEXT: [[CONV19:%.*]] = trunc i32 [[ADD18]] to i8
641 // CHECK1-NEXT: store i8 [[CONV19]], i8* [[OMP_ARRAYCPY_DESTELEMENTPAST16]], align 1
642 // CHECK1-NEXT: [[OMP_ARRAYCPY_DEST_ELEMENT20]] = getelementptr i8, i8* [[OMP_ARRAYCPY_DESTELEMENTPAST16]], i32 1
643 // CHECK1-NEXT: [[OMP_ARRAYCPY_SRC_ELEMENT]] = getelementptr i8, i8* [[OMP_ARRAYCPY_SRCELEMENTPAST]], i32 1
644 // CHECK1-NEXT: [[OMP_ARRAYCPY_DONE21:%.*]] = icmp eq i8* [[OMP_ARRAYCPY_DEST_ELEMENT20]], [[TMP99]]
645 // CHECK1-NEXT: br i1 [[OMP_ARRAYCPY_DONE21]], label [[OMP_ARRAYCPY_DONE22]], label [[OMP_ARRAYCPY_BODY]]
646 // CHECK1: omp.arraycpy.done22:
647 // CHECK1-NEXT: call void @__kmpc_end_reduce_nowait(%struct.ident_t* @[[GLOB4]], i32 [[TMP94]], [8 x i32]* @.gomp_critical_user_.reduction.var)
648 // CHECK1-NEXT: br label [[DOTOMP_REDUCTION_DEFAULT]]
649 // CHECK1: .omp.reduction.case2:
650 // CHECK1-NEXT: [[TMP102:%.*]] = load i32, i32* [[ARGC1]], align 4
651 // CHECK1-NEXT: [[TMP103:%.*]] = atomicrmw add i32* [[TMP0]], i32 [[TMP102]] monotonic, align 4
652 // CHECK1-NEXT: [[TMP104:%.*]] = getelementptr i8, i8* [[ARRAYIDX2]], i64 [[TMP13]]
653 // CHECK1-NEXT: [[OMP_ARRAYCPY_ISEMPTY23:%.*]] = icmp eq i8* [[ARRAYIDX2]], [[TMP104]]
654 // CHECK1-NEXT: br i1 [[OMP_ARRAYCPY_ISEMPTY23]], label [[OMP_ARRAYCPY_DONE36:%.*]], label [[OMP_ARRAYCPY_BODY24:%.*]]
655 // CHECK1: omp.arraycpy.body24:
656 // CHECK1-NEXT: [[OMP_ARRAYCPY_SRCELEMENTPAST25:%.*]] = phi i8* [ [[VLA]], [[DOTOMP_REDUCTION_CASE2]] ], [ [[OMP_ARRAYCPY_SRC_ELEMENT34:%.*]], [[ATOMIC_EXIT:%.*]] ]
657 // CHECK1-NEXT: [[OMP_ARRAYCPY_DESTELEMENTPAST26:%.*]] = phi i8* [ [[ARRAYIDX2]], [[DOTOMP_REDUCTION_CASE2]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT33:%.*]], [[ATOMIC_EXIT]] ]
658 // CHECK1-NEXT: [[TMP105:%.*]] = load i8, i8* [[OMP_ARRAYCPY_SRCELEMENTPAST25]], align 1
659 // CHECK1-NEXT: [[CONV27:%.*]] = sext i8 [[TMP105]] to i32
660 // CHECK1-NEXT: [[ATOMIC_LOAD:%.*]] = load atomic i8, i8* [[OMP_ARRAYCPY_DESTELEMENTPAST26]] monotonic, align 1
661 // CHECK1-NEXT: br label [[ATOMIC_CONT:%.*]]
662 // CHECK1: atomic_cont:
663 // CHECK1-NEXT: [[TMP106:%.*]] = phi i8 [ [[ATOMIC_LOAD]], [[OMP_ARRAYCPY_BODY24]] ], [ [[TMP111:%.*]], [[ATOMIC_CONT]] ]
664 // CHECK1-NEXT: store i8 [[TMP106]], i8* [[_TMP28]], align 1
665 // CHECK1-NEXT: [[TMP107:%.*]] = load i8, i8* [[_TMP28]], align 1
666 // CHECK1-NEXT: [[CONV29:%.*]] = sext i8 [[TMP107]] to i32
667 // CHECK1-NEXT: [[TMP108:%.*]] = load i8, i8* [[OMP_ARRAYCPY_SRCELEMENTPAST25]], align 1
668 // CHECK1-NEXT: [[CONV30:%.*]] = sext i8 [[TMP108]] to i32
669 // CHECK1-NEXT: [[ADD31:%.*]] = add nsw i32 [[CONV29]], [[CONV30]]
670 // CHECK1-NEXT: [[CONV32:%.*]] = trunc i32 [[ADD31]] to i8
671 // CHECK1-NEXT: store i8 [[CONV32]], i8* [[ATOMIC_TEMP]], align 1
672 // CHECK1-NEXT: [[TMP109:%.*]] = load i8, i8* [[ATOMIC_TEMP]], align 1
673 // CHECK1-NEXT: [[TMP110:%.*]] = cmpxchg i8* [[OMP_ARRAYCPY_DESTELEMENTPAST26]], i8 [[TMP106]], i8 [[TMP109]] monotonic monotonic, align 1
674 // CHECK1-NEXT: [[TMP111]] = extractvalue { i8, i1 } [[TMP110]], 0
675 // CHECK1-NEXT: [[TMP112:%.*]] = extractvalue { i8, i1 } [[TMP110]], 1
676 // CHECK1-NEXT: br i1 [[TMP112]], label [[ATOMIC_EXIT]], label [[ATOMIC_CONT]]
677 // CHECK1: atomic_exit:
678 // CHECK1-NEXT: [[OMP_ARRAYCPY_DEST_ELEMENT33]] = getelementptr i8, i8* [[OMP_ARRAYCPY_DESTELEMENTPAST26]], i32 1
679 // CHECK1-NEXT: [[OMP_ARRAYCPY_SRC_ELEMENT34]] = getelementptr i8, i8* [[OMP_ARRAYCPY_SRCELEMENTPAST25]], i32 1
680 // CHECK1-NEXT: [[OMP_ARRAYCPY_DONE35:%.*]] = icmp eq i8* [[OMP_ARRAYCPY_DEST_ELEMENT33]], [[TMP104]]
681 // CHECK1-NEXT: br i1 [[OMP_ARRAYCPY_DONE35]], label [[OMP_ARRAYCPY_DONE36]], label [[OMP_ARRAYCPY_BODY24]]
682 // CHECK1: omp.arraycpy.done36:
683 // CHECK1-NEXT: br label [[DOTOMP_REDUCTION_DEFAULT]]
684 // CHECK1: .omp.reduction.default:
685 // CHECK1-NEXT: [[TMP113:%.*]] = load i8*, i8** [[SAVED_STACK]], align 8
686 // CHECK1-NEXT: call void @llvm.stackrestore(i8* [[TMP113]])
687 // CHECK1-NEXT: ret void
688 //
689 //
690 // CHECK1-LABEL: define {{[^@]+}}@.red_init..4
691 // CHECK1-SAME: (i8* noalias noundef [[TMP0:%.*]], i8* noalias noundef [[TMP1:%.*]]) #[[ATTR3]] {
692 // CHECK1-NEXT: entry:
693 // CHECK1-NEXT: [[DOTADDR:%.*]] = alloca i8*, align 8
694 // CHECK1-NEXT: [[DOTADDR1:%.*]] = alloca i8*, align 8
695 // CHECK1-NEXT: store i8* [[TMP0]], i8** [[DOTADDR]], align 8
696 // CHECK1-NEXT: store i8* [[TMP1]], i8** [[DOTADDR1]], align 8
697 // CHECK1-NEXT: [[TMP2:%.*]] = bitcast i8** [[DOTADDR]] to i32**
698 // CHECK1-NEXT: [[TMP3:%.*]] = load i32*, i32** [[TMP2]], align 8
699 // CHECK1-NEXT: store i32 0, i32* [[TMP3]], align 4
700 // CHECK1-NEXT: ret void
701 //
702 //
703 // CHECK1-LABEL: define {{[^@]+}}@.red_comb..5
704 // CHECK1-SAME: (i8* noundef [[TMP0:%.*]], i8* noundef [[TMP1:%.*]]) #[[ATTR3]] {
705 // CHECK1-NEXT: entry:
706 // CHECK1-NEXT: [[DOTADDR:%.*]] = alloca i8*, align 8
707 // CHECK1-NEXT: [[DOTADDR1:%.*]] = alloca i8*, align 8
708 // CHECK1-NEXT: store i8* [[TMP0]], i8** [[DOTADDR]], align 8
709 // CHECK1-NEXT: store i8* [[TMP1]], i8** [[DOTADDR1]], align 8
710 // CHECK1-NEXT: [[TMP2:%.*]] = bitcast i8** [[DOTADDR]] to i32**
711 // CHECK1-NEXT: [[TMP3:%.*]] = load i32*, i32** [[TMP2]], align 8
712 // CHECK1-NEXT: [[TMP4:%.*]] = bitcast i8** [[DOTADDR1]] to i32**
713 // CHECK1-NEXT: [[TMP5:%.*]] = load i32*, i32** [[TMP4]], align 8
714 // CHECK1-NEXT: [[TMP6:%.*]] = load i32, i32* [[TMP3]], align 4
715 // CHECK1-NEXT: [[TMP7:%.*]] = load i32, i32* [[TMP5]], align 4
716 // CHECK1-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP6]], [[TMP7]]
717 // CHECK1-NEXT: store i32 [[ADD]], i32* [[TMP3]], align 4
718 // CHECK1-NEXT: ret void
719 //
720 //
721 // CHECK1-LABEL: define {{[^@]+}}@.red_init..6
722 // CHECK1-SAME: (i8* noalias noundef [[TMP0:%.*]], i8* noalias noundef [[TMP1:%.*]]) #[[ATTR3]] {
723 // CHECK1-NEXT: entry:
724 // CHECK1-NEXT: [[DOTADDR:%.*]] = alloca i8*, align 8
725 // CHECK1-NEXT: [[DOTADDR1:%.*]] = alloca i8*, align 8
726 // CHECK1-NEXT: store i8* [[TMP0]], i8** [[DOTADDR]], align 8
727 // CHECK1-NEXT: store i8* [[TMP1]], i8** [[DOTADDR1]], align 8
728 // CHECK1-NEXT: [[TMP2:%.*]] = load i8*, i8** [[DOTADDR]], align 8
729 // CHECK1-NEXT: [[TMP3:%.*]] = load i64, i64* @{{reduction_size[.].+[.]}}, align 8
730 // CHECK1-NEXT: [[TMP4:%.*]] = getelementptr i8, i8* [[TMP2]], i64 [[TMP3]]
731 // CHECK1-NEXT: [[OMP_ARRAYINIT_ISEMPTY:%.*]] = icmp eq i8* [[TMP2]], [[TMP4]]
732 // CHECK1-NEXT: br i1 [[OMP_ARRAYINIT_ISEMPTY]], label [[OMP_ARRAYINIT_DONE:%.*]], label [[OMP_ARRAYINIT_BODY:%.*]]
733 // CHECK1: omp.arrayinit.body:
734 // CHECK1-NEXT: [[OMP_ARRAYCPY_DESTELEMENTPAST:%.*]] = phi i8* [ [[TMP2]], [[ENTRY:%.*]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT:%.*]], [[OMP_ARRAYINIT_BODY]] ]
735 // CHECK1-NEXT: store i8 0, i8* [[OMP_ARRAYCPY_DESTELEMENTPAST]], align 1
736 // CHECK1-NEXT: [[OMP_ARRAYCPY_DEST_ELEMENT]] = getelementptr i8, i8* [[OMP_ARRAYCPY_DESTELEMENTPAST]], i32 1
737 // CHECK1-NEXT: [[OMP_ARRAYCPY_DONE:%.*]] = icmp eq i8* [[OMP_ARRAYCPY_DEST_ELEMENT]], [[TMP4]]
738 // CHECK1-NEXT: br i1 [[OMP_ARRAYCPY_DONE]], label [[OMP_ARRAYINIT_DONE]], label [[OMP_ARRAYINIT_BODY]]
739 // CHECK1: omp.arrayinit.done:
740 // CHECK1-NEXT: ret void
741 //
742 //
743 // CHECK1-LABEL: define {{[^@]+}}@.red_comb..7
744 // CHECK1-SAME: (i8* noundef [[TMP0:%.*]], i8* noundef [[TMP1:%.*]]) #[[ATTR3]] {
745 // CHECK1-NEXT: entry:
746 // CHECK1-NEXT: [[DOTADDR:%.*]] = alloca i8*, align 8
747 // CHECK1-NEXT: [[DOTADDR1:%.*]] = alloca i8*, align 8
748 // CHECK1-NEXT: store i8* [[TMP0]], i8** [[DOTADDR]], align 8
749 // CHECK1-NEXT: store i8* [[TMP1]], i8** [[DOTADDR1]], align 8
750 // CHECK1-NEXT: [[TMP2:%.*]] = load i64, i64* @{{reduction_size[.].+[.]}}, align 8
751 // CHECK1-NEXT: [[TMP3:%.*]] = load i8*, i8** [[DOTADDR]], align 8
752 // CHECK1-NEXT: [[TMP4:%.*]] = load i8*, i8** [[DOTADDR1]], align 8
753 // CHECK1-NEXT: [[TMP5:%.*]] = getelementptr i8, i8* [[TMP3]], i64 [[TMP2]]
754 // CHECK1-NEXT: [[OMP_ARRAYCPY_ISEMPTY:%.*]] = icmp eq i8* [[TMP3]], [[TMP5]]
755 // CHECK1-NEXT: br i1 [[OMP_ARRAYCPY_ISEMPTY]], label [[OMP_ARRAYCPY_DONE4:%.*]], label [[OMP_ARRAYCPY_BODY:%.*]]
756 // CHECK1: omp.arraycpy.body:
757 // CHECK1-NEXT: [[OMP_ARRAYCPY_SRCELEMENTPAST:%.*]] = phi i8* [ [[TMP4]], [[ENTRY:%.*]] ], [ [[OMP_ARRAYCPY_SRC_ELEMENT:%.*]], [[OMP_ARRAYCPY_BODY]] ]
758 // CHECK1-NEXT: [[OMP_ARRAYCPY_DESTELEMENTPAST:%.*]] = phi i8* [ [[TMP3]], [[ENTRY]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT:%.*]], [[OMP_ARRAYCPY_BODY]] ]
759 // CHECK1-NEXT: [[TMP6:%.*]] = load i8, i8* [[OMP_ARRAYCPY_DESTELEMENTPAST]], align 1
760 // CHECK1-NEXT: [[CONV:%.*]] = sext i8 [[TMP6]] to i32
761 // CHECK1-NEXT: [[TMP7:%.*]] = load i8, i8* [[OMP_ARRAYCPY_SRCELEMENTPAST]], align 1
762 // CHECK1-NEXT: [[CONV2:%.*]] = sext i8 [[TMP7]] to i32
763 // CHECK1-NEXT: [[ADD:%.*]] = add nsw i32 [[CONV]], [[CONV2]]
764 // CHECK1-NEXT: [[CONV3:%.*]] = trunc i32 [[ADD]] to i8
765 // CHECK1-NEXT: store i8 [[CONV3]], i8* [[OMP_ARRAYCPY_DESTELEMENTPAST]], align 1
766 // CHECK1-NEXT: [[OMP_ARRAYCPY_DEST_ELEMENT]] = getelementptr i8, i8* [[OMP_ARRAYCPY_DESTELEMENTPAST]], i32 1
767 // CHECK1-NEXT: [[OMP_ARRAYCPY_SRC_ELEMENT]] = getelementptr i8, i8* [[OMP_ARRAYCPY_SRCELEMENTPAST]], i32 1
768 // CHECK1-NEXT: [[OMP_ARRAYCPY_DONE:%.*]] = icmp eq i8* [[OMP_ARRAYCPY_DEST_ELEMENT]], [[TMP5]]
769 // CHECK1-NEXT: br i1 [[OMP_ARRAYCPY_DONE]], label [[OMP_ARRAYCPY_DONE4]], label [[OMP_ARRAYCPY_BODY]]
770 // CHECK1: omp.arraycpy.done4:
771 // CHECK1-NEXT: ret void
772 //
773 //
774 // CHECK1-LABEL: define {{[^@]+}}@.omp_task_privates_map.
775 // CHECK1-SAME: (%struct..kmp_privates.t* noalias noundef [[TMP0:%.*]], i8*** noalias noundef [[TMP1:%.*]]) #[[ATTR6:[0-9]+]] {
776 // CHECK1-NEXT: entry:
777 // CHECK1-NEXT: [[DOTADDR:%.*]] = alloca %struct..kmp_privates.t*, align 8
778 // CHECK1-NEXT: [[DOTADDR1:%.*]] = alloca i8***, align 8
779 // CHECK1-NEXT: store %struct..kmp_privates.t* [[TMP0]], %struct..kmp_privates.t** [[DOTADDR]], align 8
780 // CHECK1-NEXT: store i8*** [[TMP1]], i8**** [[DOTADDR1]], align 8
781 // CHECK1-NEXT: [[TMP2:%.*]] = load %struct..kmp_privates.t*, %struct..kmp_privates.t** [[DOTADDR]], align 8
782 // CHECK1-NEXT: [[TMP3:%.*]] = getelementptr inbounds [[STRUCT__KMP_PRIVATES_T:%.*]], %struct..kmp_privates.t* [[TMP2]], i32 0, i32 0
783 // CHECK1-NEXT: [[TMP4:%.*]] = load i8***, i8**** [[DOTADDR1]], align 8
784 // CHECK1-NEXT: store i8** [[TMP3]], i8*** [[TMP4]], align 8
785 // CHECK1-NEXT: ret void
786 //
787 //
788 // CHECK1-LABEL: define {{[^@]+}}@.omp_task_entry.
789 // CHECK1-SAME: (i32 noundef [[TMP0:%.*]], %struct.kmp_task_t_with_privates* noalias noundef [[TMP1:%.*]]) #[[ATTR3]] {
790 // CHECK1-NEXT: entry:
791 // CHECK1-NEXT: [[DOTGLOBAL_TID__ADDR_I:%.*]] = alloca i32, align 4
792 // CHECK1-NEXT: [[DOTPART_ID__ADDR_I:%.*]] = alloca i32*, align 8
793 // CHECK1-NEXT: [[DOTPRIVATES__ADDR_I:%.*]] = alloca i8*, align 8
794 // CHECK1-NEXT: [[DOTCOPY_FN__ADDR_I:%.*]] = alloca void (i8*, ...)*, align 8
795 // CHECK1-NEXT: [[DOTTASK_T__ADDR_I:%.*]] = alloca i8*, align 8
796 // CHECK1-NEXT: [[__CONTEXT_ADDR_I:%.*]] = alloca %struct.anon*, align 8
797 // CHECK1-NEXT: [[DOTFIRSTPRIV_PTR_ADDR_I:%.*]] = alloca i8**, align 8
798 // CHECK1-NEXT: [[TMP_I:%.*]] = alloca i8**, align 8
799 // CHECK1-NEXT: [[TMP4_I:%.*]] = alloca i8*, align 8
800 // CHECK1-NEXT: [[DOTADDR:%.*]] = alloca i32, align 4
801 // CHECK1-NEXT: [[DOTADDR1:%.*]] = alloca %struct.kmp_task_t_with_privates*, align 8
802 // CHECK1-NEXT: store i32 [[TMP0]], i32* [[DOTADDR]], align 4
803 // CHECK1-NEXT: store %struct.kmp_task_t_with_privates* [[TMP1]], %struct.kmp_task_t_with_privates** [[DOTADDR1]], align 8
804 // CHECK1-NEXT: [[TMP2:%.*]] = load i32, i32* [[DOTADDR]], align 4
805 // CHECK1-NEXT: [[TMP3:%.*]] = load %struct.kmp_task_t_with_privates*, %struct.kmp_task_t_with_privates** [[DOTADDR1]], align 8
806 // CHECK1-NEXT: [[TMP4:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASK_T_WITH_PRIVATES:%.*]], %struct.kmp_task_t_with_privates* [[TMP3]], i32 0, i32 0
807 // CHECK1-NEXT: [[TMP5:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASK_T:%.*]], %struct.kmp_task_t* [[TMP4]], i32 0, i32 2
808 // CHECK1-NEXT: [[TMP6:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASK_T]], %struct.kmp_task_t* [[TMP4]], i32 0, i32 0
809 // CHECK1-NEXT: [[TMP7:%.*]] = load i8*, i8** [[TMP6]], align 8
810 // CHECK1-NEXT: [[TMP8:%.*]] = bitcast i8* [[TMP7]] to %struct.anon*
811 // CHECK1-NEXT: [[TMP9:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASK_T_WITH_PRIVATES]], %struct.kmp_task_t_with_privates* [[TMP3]], i32 0, i32 1
812 // CHECK1-NEXT: [[TMP10:%.*]] = bitcast %struct..kmp_privates.t* [[TMP9]] to i8*
813 // CHECK1-NEXT: [[TMP11:%.*]] = bitcast %struct.kmp_task_t_with_privates* [[TMP3]] to i8*
814 // CHECK1-NEXT: call void @llvm.experimental.noalias.scope.decl(metadata [[META3:![0-9]+]])
815 // CHECK1-NEXT: call void @llvm.experimental.noalias.scope.decl(metadata [[META6:![0-9]+]])
816 // CHECK1-NEXT: call void @llvm.experimental.noalias.scope.decl(metadata [[META8:![0-9]+]])
817 // CHECK1-NEXT: call void @llvm.experimental.noalias.scope.decl(metadata [[META10:![0-9]+]])
818 // CHECK1-NEXT: store i32 [[TMP2]], i32* [[DOTGLOBAL_TID__ADDR_I]], align 4, !noalias !12
819 // CHECK1-NEXT: store i32* [[TMP5]], i32** [[DOTPART_ID__ADDR_I]], align 8, !noalias !12
820 // CHECK1-NEXT: store i8* [[TMP10]], i8** [[DOTPRIVATES__ADDR_I]], align 8, !noalias !12
821 // CHECK1-NEXT: store void (i8*, ...)* bitcast (void (%struct..kmp_privates.t*, i8***)* @.omp_task_privates_map. to void (i8*, ...)*), void (i8*, ...)** [[DOTCOPY_FN__ADDR_I]], align 8, !noalias !12
822 // CHECK1-NEXT: store i8* [[TMP11]], i8** [[DOTTASK_T__ADDR_I]], align 8, !noalias !12
823 // CHECK1-NEXT: store %struct.anon* [[TMP8]], %struct.anon** [[__CONTEXT_ADDR_I]], align 8, !noalias !12
824 // CHECK1-NEXT: [[TMP12:%.*]] = load %struct.anon*, %struct.anon** [[__CONTEXT_ADDR_I]], align 8, !noalias !12
825 // CHECK1-NEXT: [[TMP13:%.*]] = load void (i8*, ...)*, void (i8*, ...)** [[DOTCOPY_FN__ADDR_I]], align 8, !noalias !12
826 // CHECK1-NEXT: [[TMP14:%.*]] = load i8*, i8** [[DOTPRIVATES__ADDR_I]], align 8, !noalias !12
827 // CHECK1-NEXT: [[TMP15:%.*]] = bitcast void (i8*, ...)* [[TMP13]] to void (i8*, i8***)*
828 // CHECK1-NEXT: call void [[TMP15]](i8* [[TMP14]], i8*** [[DOTFIRSTPRIV_PTR_ADDR_I]]) #[[ATTR5]]
829 // CHECK1-NEXT: [[TMP16:%.*]] = load i8**, i8*** [[DOTFIRSTPRIV_PTR_ADDR_I]], align 8, !noalias !12
830 // CHECK1-NEXT: [[TMP17:%.*]] = getelementptr inbounds [[STRUCT_ANON:%.*]], %struct.anon* [[TMP12]], i32 0, i32 1
831 // CHECK1-NEXT: [[TMP18:%.*]] = load i32*, i32** [[TMP17]], align 8
832 // CHECK1-NEXT: [[TMP19:%.*]] = load i8*, i8** [[TMP16]], align 8
833 // CHECK1-NEXT: [[TMP20:%.*]] = load i32, i32* [[DOTGLOBAL_TID__ADDR_I]], align 4, !noalias !12
834 // CHECK1-NEXT: [[TMP21:%.*]] = bitcast i32* [[TMP18]] to i8*
835 // CHECK1-NEXT: [[TMP22:%.*]] = call i8* @__kmpc_task_reduction_get_th_data(i32 [[TMP20]], i8* [[TMP19]], i8* [[TMP21]])
836 // CHECK1-NEXT: [[CONV_I:%.*]] = bitcast i8* [[TMP22]] to i32*
837 // CHECK1-NEXT: [[TMP23:%.*]] = getelementptr inbounds [[STRUCT_ANON]], %struct.anon* [[TMP12]], i32 0, i32 2
838 // CHECK1-NEXT: [[TMP24:%.*]] = load i8**, i8*** [[TMP23]], align 8
839 // CHECK1-NEXT: [[TMP25:%.*]] = load i8*, i8** [[TMP24]], align 8
840 // CHECK1-NEXT: [[TMP26:%.*]] = getelementptr inbounds [[STRUCT_ANON]], %struct.anon* [[TMP12]], i32 0, i32 1
841 // CHECK1-NEXT: [[TMP27:%.*]] = load i32*, i32** [[TMP26]], align 8
842 // CHECK1-NEXT: [[TMP28:%.*]] = load i32, i32* [[TMP27]], align 4
843 // CHECK1-NEXT: [[TMP29:%.*]] = sext i32 [[TMP28]] to i64
844 // CHECK1-NEXT: [[LB_ADD_LEN_I:%.*]] = add nsw i64 -1, [[TMP29]]
845 // CHECK1-NEXT: [[TMP30:%.*]] = getelementptr inbounds [[STRUCT_ANON]], %struct.anon* [[TMP12]], i32 0, i32 2
846 // CHECK1-NEXT: [[TMP31:%.*]] = load i8**, i8*** [[TMP30]], align 8
847 // CHECK1-NEXT: [[ARRAYIDX2_I:%.*]] = getelementptr inbounds i8*, i8** [[TMP31]], i64 9
848 // CHECK1-NEXT: [[TMP32:%.*]] = load i8*, i8** [[ARRAYIDX2_I]], align 8
849 // CHECK1-NEXT: [[ARRAYIDX3_I:%.*]] = getelementptr inbounds i8, i8* [[TMP32]], i64 [[LB_ADD_LEN_I]]
850 // CHECK1-NEXT: [[TMP33:%.*]] = ptrtoint i8* [[ARRAYIDX3_I]] to i64
851 // CHECK1-NEXT: [[TMP34:%.*]] = ptrtoint i8* [[TMP25]] to i64
852 // CHECK1-NEXT: [[TMP35:%.*]] = sub i64 [[TMP33]], [[TMP34]]
853 // CHECK1-NEXT: [[TMP36:%.*]] = sdiv exact i64 [[TMP35]], ptrtoint (i8* getelementptr (i8, i8* null, i32 1) to i64)
854 // CHECK1-NEXT: [[TMP37:%.*]] = add nuw i64 [[TMP36]], 1
855 // CHECK1-NEXT: [[TMP38:%.*]] = mul nuw i64 [[TMP37]], ptrtoint (i8* getelementptr (i8, i8* null, i32 1) to i64)
856 // CHECK1-NEXT: store i64 [[TMP37]], i64* @{{reduction_size[.].+[.]}}, align 8, !noalias !12
857 // CHECK1-NEXT: [[TMP39:%.*]] = load i8*, i8** [[TMP16]], align 8
858 // CHECK1-NEXT: [[TMP40:%.*]] = call i8* @__kmpc_task_reduction_get_th_data(i32 [[TMP20]], i8* [[TMP39]], i8* [[TMP25]])
859 // CHECK1-NEXT: [[TMP41:%.*]] = getelementptr inbounds [[STRUCT_ANON]], %struct.anon* [[TMP12]], i32 0, i32 2
860 // CHECK1-NEXT: [[TMP42:%.*]] = load i8**, i8*** [[TMP41]], align 8
861 // CHECK1-NEXT: [[TMP43:%.*]] = load i8*, i8** [[TMP42]], align 8
862 // CHECK1-NEXT: [[TMP44:%.*]] = ptrtoint i8* [[TMP43]] to i64
863 // CHECK1-NEXT: [[TMP45:%.*]] = ptrtoint i8* [[TMP25]] to i64
864 // CHECK1-NEXT: [[TMP46:%.*]] = sub i64 [[TMP44]], [[TMP45]]
865 // CHECK1-NEXT: [[TMP47:%.*]] = sdiv exact i64 [[TMP46]], ptrtoint (i8* getelementptr (i8, i8* null, i32 1) to i64)
866 // CHECK1-NEXT: [[TMP48:%.*]] = getelementptr i8, i8* [[TMP40]], i64 [[TMP47]]
867 // CHECK1-NEXT: store i8** [[TMP4_I]], i8*** [[TMP_I]], align 8, !noalias !12
868 // CHECK1-NEXT: store i8* [[TMP48]], i8** [[TMP4_I]], align 8, !noalias !12
869 // CHECK1-NEXT: ret i32 0
870 //
871 //
872 // CHECK1-LABEL: define {{[^@]+}}@.omp.reduction.reduction_func
873 // CHECK1-SAME: (i8* noundef [[TMP0:%.*]], i8* noundef [[TMP1:%.*]]) #[[ATTR3]] {
874 // CHECK1-NEXT: entry:
875 // CHECK1-NEXT: [[DOTADDR:%.*]] = alloca i8*, align 8
876 // CHECK1-NEXT: [[DOTADDR1:%.*]] = alloca i8*, align 8
877 // CHECK1-NEXT: store i8* [[TMP0]], i8** [[DOTADDR]], align 8
878 // CHECK1-NEXT: store i8* [[TMP1]], i8** [[DOTADDR1]], align 8
879 // CHECK1-NEXT: [[TMP2:%.*]] = load i8*, i8** [[DOTADDR]], align 8
880 // CHECK1-NEXT: [[TMP3:%.*]] = bitcast i8* [[TMP2]] to [3 x i8*]*
881 // CHECK1-NEXT: [[TMP4:%.*]] = load i8*, i8** [[DOTADDR1]], align 8
882 // CHECK1-NEXT: [[TMP5:%.*]] = bitcast i8* [[TMP4]] to [3 x i8*]*
883 // CHECK1-NEXT: [[TMP6:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[TMP5]], i64 0, i64 0
884 // CHECK1-NEXT: [[TMP7:%.*]] = load i8*, i8** [[TMP6]], align 8
885 // CHECK1-NEXT: [[TMP8:%.*]] = bitcast i8* [[TMP7]] to i32*
886 // CHECK1-NEXT: [[TMP9:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[TMP3]], i64 0, i64 0
887 // CHECK1-NEXT: [[TMP10:%.*]] = load i8*, i8** [[TMP9]], align 8
888 // CHECK1-NEXT: [[TMP11:%.*]] = bitcast i8* [[TMP10]] to i32*
889 // CHECK1-NEXT: [[TMP12:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[TMP5]], i64 0, i64 1
890 // CHECK1-NEXT: [[TMP13:%.*]] = load i8*, i8** [[TMP12]], align 8
891 // CHECK1-NEXT: [[TMP14:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[TMP3]], i64 0, i64 1
892 // CHECK1-NEXT: [[TMP15:%.*]] = load i8*, i8** [[TMP14]], align 8
893 // CHECK1-NEXT: [[TMP16:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[TMP3]], i64 0, i64 2
894 // CHECK1-NEXT: [[TMP17:%.*]] = load i8*, i8** [[TMP16]], align 8
895 // CHECK1-NEXT: [[TMP18:%.*]] = ptrtoint i8* [[TMP17]] to i64
896 // CHECK1-NEXT: [[TMP19:%.*]] = load i32, i32* [[TMP11]], align 4
897 // CHECK1-NEXT: [[TMP20:%.*]] = load i32, i32* [[TMP8]], align 4
898 // CHECK1-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP19]], [[TMP20]]
899 // CHECK1-NEXT: store i32 [[ADD]], i32* [[TMP11]], align 4
900 // CHECK1-NEXT: [[TMP21:%.*]] = getelementptr i8, i8* [[TMP15]], i64 [[TMP18]]
901 // CHECK1-NEXT: [[OMP_ARRAYCPY_ISEMPTY:%.*]] = icmp eq i8* [[TMP15]], [[TMP21]]
902 // CHECK1-NEXT: br i1 [[OMP_ARRAYCPY_ISEMPTY]], label [[OMP_ARRAYCPY_DONE5:%.*]], label [[OMP_ARRAYCPY_BODY:%.*]]
903 // CHECK1: omp.arraycpy.body:
904 // CHECK1-NEXT: [[OMP_ARRAYCPY_SRCELEMENTPAST:%.*]] = phi i8* [ [[TMP13]], [[ENTRY:%.*]] ], [ [[OMP_ARRAYCPY_SRC_ELEMENT:%.*]], [[OMP_ARRAYCPY_BODY]] ]
905 // CHECK1-NEXT: [[OMP_ARRAYCPY_DESTELEMENTPAST:%.*]] = phi i8* [ [[TMP15]], [[ENTRY]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT:%.*]], [[OMP_ARRAYCPY_BODY]] ]
906 // CHECK1-NEXT: [[TMP22:%.*]] = load i8, i8* [[OMP_ARRAYCPY_DESTELEMENTPAST]], align 1
907 // CHECK1-NEXT: [[CONV:%.*]] = sext i8 [[TMP22]] to i32
908 // CHECK1-NEXT: [[TMP23:%.*]] = load i8, i8* [[OMP_ARRAYCPY_SRCELEMENTPAST]], align 1
909 // CHECK1-NEXT: [[CONV2:%.*]] = sext i8 [[TMP23]] to i32
910 // CHECK1-NEXT: [[ADD3:%.*]] = add nsw i32 [[CONV]], [[CONV2]]
911 // CHECK1-NEXT: [[CONV4:%.*]] = trunc i32 [[ADD3]] to i8
912 // CHECK1-NEXT: store i8 [[CONV4]], i8* [[OMP_ARRAYCPY_DESTELEMENTPAST]], align 1
913 // CHECK1-NEXT: [[OMP_ARRAYCPY_DEST_ELEMENT]] = getelementptr i8, i8* [[OMP_ARRAYCPY_DESTELEMENTPAST]], i32 1
914 // CHECK1-NEXT: [[OMP_ARRAYCPY_SRC_ELEMENT]] = getelementptr i8, i8* [[OMP_ARRAYCPY_SRCELEMENTPAST]], i32 1
915 // CHECK1-NEXT: [[OMP_ARRAYCPY_DONE:%.*]] = icmp eq i8* [[OMP_ARRAYCPY_DEST_ELEMENT]], [[TMP21]]
916 // CHECK1-NEXT: br i1 [[OMP_ARRAYCPY_DONE]], label [[OMP_ARRAYCPY_DONE5]], label [[OMP_ARRAYCPY_BODY]]
917 // CHECK1: omp.arraycpy.done5:
918 // CHECK1-NEXT: ret void
919 //
920 //
921 // CHECK1-LABEL: define {{[^@]+}}@.omp.reduction.reduction_func.9
922 // CHECK1-SAME: (i8* noundef [[TMP0:%.*]], i8* noundef [[TMP1:%.*]]) #[[ATTR3]] {
923 // CHECK1-NEXT: entry:
924 // CHECK1-NEXT: [[DOTADDR:%.*]] = alloca i8*, align 8
925 // CHECK1-NEXT: [[DOTADDR1:%.*]] = alloca i8*, align 8
926 // CHECK1-NEXT: store i8* [[TMP0]], i8** [[DOTADDR]], align 8
927 // CHECK1-NEXT: store i8* [[TMP1]], i8** [[DOTADDR1]], align 8
928 // CHECK1-NEXT: [[TMP2:%.*]] = load i8*, i8** [[DOTADDR]], align 8
929 // CHECK1-NEXT: [[TMP3:%.*]] = bitcast i8* [[TMP2]] to [3 x i8*]*
930 // CHECK1-NEXT: [[TMP4:%.*]] = load i8*, i8** [[DOTADDR1]], align 8
931 // CHECK1-NEXT: [[TMP5:%.*]] = bitcast i8* [[TMP4]] to [3 x i8*]*
932 // CHECK1-NEXT: [[TMP6:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[TMP5]], i64 0, i64 0
933 // CHECK1-NEXT: [[TMP7:%.*]] = load i8*, i8** [[TMP6]], align 8
934 // CHECK1-NEXT: [[TMP8:%.*]] = bitcast i8* [[TMP7]] to i32*
935 // CHECK1-NEXT: [[TMP9:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[TMP3]], i64 0, i64 0
936 // CHECK1-NEXT: [[TMP10:%.*]] = load i8*, i8** [[TMP9]], align 8
937 // CHECK1-NEXT: [[TMP11:%.*]] = bitcast i8* [[TMP10]] to i32*
938 // CHECK1-NEXT: [[TMP12:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[TMP5]], i64 0, i64 1
939 // CHECK1-NEXT: [[TMP13:%.*]] = load i8*, i8** [[TMP12]], align 8
940 // CHECK1-NEXT: [[TMP14:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[TMP3]], i64 0, i64 1
941 // CHECK1-NEXT: [[TMP15:%.*]] = load i8*, i8** [[TMP14]], align 8
942 // CHECK1-NEXT: [[TMP16:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[TMP3]], i64 0, i64 2
943 // CHECK1-NEXT: [[TMP17:%.*]] = load i8*, i8** [[TMP16]], align 8
944 // CHECK1-NEXT: [[TMP18:%.*]] = ptrtoint i8* [[TMP17]] to i64
945 // CHECK1-NEXT: [[TMP19:%.*]] = load i32, i32* [[TMP11]], align 4
946 // CHECK1-NEXT: [[TMP20:%.*]] = load i32, i32* [[TMP8]], align 4
947 // CHECK1-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP19]], [[TMP20]]
948 // CHECK1-NEXT: store i32 [[ADD]], i32* [[TMP11]], align 4
949 // CHECK1-NEXT: [[TMP21:%.*]] = getelementptr i8, i8* [[TMP15]], i64 [[TMP18]]
950 // CHECK1-NEXT: [[OMP_ARRAYCPY_ISEMPTY:%.*]] = icmp eq i8* [[TMP15]], [[TMP21]]
951 // CHECK1-NEXT: br i1 [[OMP_ARRAYCPY_ISEMPTY]], label [[OMP_ARRAYCPY_DONE5:%.*]], label [[OMP_ARRAYCPY_BODY:%.*]]
952 // CHECK1: omp.arraycpy.body:
953 // CHECK1-NEXT: [[OMP_ARRAYCPY_SRCELEMENTPAST:%.*]] = phi i8* [ [[TMP13]], [[ENTRY:%.*]] ], [ [[OMP_ARRAYCPY_SRC_ELEMENT:%.*]], [[OMP_ARRAYCPY_BODY]] ]
954 // CHECK1-NEXT: [[OMP_ARRAYCPY_DESTELEMENTPAST:%.*]] = phi i8* [ [[TMP15]], [[ENTRY]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT:%.*]], [[OMP_ARRAYCPY_BODY]] ]
955 // CHECK1-NEXT: [[TMP22:%.*]] = load i8, i8* [[OMP_ARRAYCPY_DESTELEMENTPAST]], align 1
956 // CHECK1-NEXT: [[CONV:%.*]] = sext i8 [[TMP22]] to i32
957 // CHECK1-NEXT: [[TMP23:%.*]] = load i8, i8* [[OMP_ARRAYCPY_SRCELEMENTPAST]], align 1
958 // CHECK1-NEXT: [[CONV2:%.*]] = sext i8 [[TMP23]] to i32
959 // CHECK1-NEXT: [[ADD3:%.*]] = add nsw i32 [[CONV]], [[CONV2]]
960 // CHECK1-NEXT: [[CONV4:%.*]] = trunc i32 [[ADD3]] to i8
961 // CHECK1-NEXT: store i8 [[CONV4]], i8* [[OMP_ARRAYCPY_DESTELEMENTPAST]], align 1
962 // CHECK1-NEXT: [[OMP_ARRAYCPY_DEST_ELEMENT]] = getelementptr i8, i8* [[OMP_ARRAYCPY_DESTELEMENTPAST]], i32 1
963 // CHECK1-NEXT: [[OMP_ARRAYCPY_SRC_ELEMENT]] = getelementptr i8, i8* [[OMP_ARRAYCPY_SRCELEMENTPAST]], i32 1
964 // CHECK1-NEXT: [[OMP_ARRAYCPY_DONE:%.*]] = icmp eq i8* [[OMP_ARRAYCPY_DEST_ELEMENT]], [[TMP21]]
965 // CHECK1-NEXT: br i1 [[OMP_ARRAYCPY_DONE]], label [[OMP_ARRAYCPY_DONE5]], label [[OMP_ARRAYCPY_BODY]]
966 // CHECK1: omp.arraycpy.done5:
967 // CHECK1-NEXT: ret void
968 //
969