xref: /llvm-project/clang/test/OpenMP/parallel_for_linear_codegen.cpp (revision 7539e9cf811e590d9f12ae39673ca789e26386b4)
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 -x c++ -triple x86_64-apple-darwin10 -emit-llvm %s -o - | FileCheck %s --check-prefix=CHECK1
3 // RUN: %clang_cc1 -no-opaque-pointers -fopenmp -x c++ -std=c++11 -triple x86_64-apple-darwin10 -emit-pch -o %t %s
4 // RUN: %clang_cc1 -no-opaque-pointers -fopenmp -x c++ -triple x86_64-apple-darwin10 -std=c++11 -include-pch %t -verify %s -emit-llvm -o - | FileCheck %s --check-prefix=CHECK1
5 // RUN: %clang_cc1 -no-opaque-pointers -verify -fopenmp -x c++ -std=c++11 -DLAMBDA -triple x86_64-apple-darwin10 -emit-llvm %s -o - | FileCheck %s --check-prefix=CHECK3
6 // RUN: %clang_cc1 -no-opaque-pointers -verify -fopenmp -x c++ -fblocks -DBLOCKS -triple x86_64-apple-darwin10 -emit-llvm %s -o - | FileCheck %s --check-prefix=CHECK4
7 
8 // RUN: %clang_cc1 -no-opaque-pointers -verify -fopenmp-simd -x c++ -triple x86_64-apple-darwin10 -emit-llvm %s -o - | FileCheck %s --implicit-check-not="{{__kmpc|__tgt}}"
9 // RUN: %clang_cc1 -no-opaque-pointers -fopenmp-simd -x c++ -std=c++11 -triple x86_64-apple-darwin10 -emit-pch -o %t %s
10 // RUN: %clang_cc1 -no-opaque-pointers -fopenmp-simd -x c++ -triple x86_64-apple-darwin10 -std=c++11 -include-pch %t -verify %s -emit-llvm -o - | FileCheck %s --implicit-check-not="{{__kmpc|__tgt}}"
11 // RUN: %clang_cc1 -no-opaque-pointers -verify -fopenmp-simd -x c++ -std=c++11 -DLAMBDA -triple x86_64-apple-darwin10 -emit-llvm %s -o - | FileCheck %s --implicit-check-not="{{__kmpc|__tgt}}"
12 // RUN: %clang_cc1 -no-opaque-pointers -verify -fopenmp-simd -x c++ -fblocks -DBLOCKS -triple x86_64-apple-darwin10 -emit-llvm %s -o - | FileCheck %s --implicit-check-not="{{__kmpc|__tgt}}"
13 // expected-no-diagnostics
14 #ifndef HEADER
15 #define HEADER
16 
17 template <class T>
18 struct S {
19   T f;
20   S(T a) : f(a) {}
21   S() : f() {}
22   S<T> &operator=(const S<T> &);
23   operator T() { return T(); }
24   ~S() {}
25 };
26 
27 volatile int g = 1212;
28 float f;
29 char cnt;
30 
31 template <typename T>
32 T tmain() {
33   S<T> test;
34   T *pvar = &test.f;
35   T lvar = T();
36 #pragma omp parallel for linear(pvar, lvar)
37   for (int i = 0; i < 2; ++i) {
38     ++pvar, ++lvar;
39   }
40   return T();
41 }
42 
43 int main() {
44 #ifdef LAMBDA
45   [&]() {
46 #pragma omp parallel for linear(g:5)
47   for (int i = 0; i < 2; ++i) {
48     g += 5;
49     [&]() {
50       g = 2;
51     }();
52   }
53   }();
54   return 0;
55 #elif defined(BLOCKS)
56   ^{
57 #pragma omp parallel for linear(g:5)
58   for (int i = 0; i < 2; ++i) {
59     g += 5;
60     g = 1;
61     ^{
62       g = 2;
63     }();
64   }
65   }();
66   return 0;
67 #else
68   S<float> test;
69   float *pvar = &test.f;
70   long long lvar = 0;
71 #pragma omp parallel for linear(pvar, lvar : 3)
72   for (int i = 0; i < 2; ++i) {
73     pvar += 3, lvar += 3;
74   }
75   return tmain<int>();
76 #endif
77 }
78 
79 
80 
81 
82 // Check for default initialization.
83 
84 
85 
86 // Check for default initialization.
87 #endif
88 
89 // CHECK1-LABEL: define {{[^@]+}}@main
90 // CHECK1-SAME: () #[[ATTR0:[0-9]+]] {
91 // CHECK1-NEXT:  entry:
92 // CHECK1-NEXT:    [[RETVAL:%.*]] = alloca i32, align 4
93 // CHECK1-NEXT:    [[TEST:%.*]] = alloca [[STRUCT_S:%.*]], align 4
94 // CHECK1-NEXT:    [[PVAR:%.*]] = alloca float*, align 8
95 // CHECK1-NEXT:    [[LVAR:%.*]] = alloca i64, align 8
96 // CHECK1-NEXT:    [[OMP_OUTLINED_ARG_AGG_:%.*]] = alloca [[STRUCT_ANON:%.*]], align 8
97 // CHECK1-NEXT:    store i32 0, i32* [[RETVAL]], align 4
98 // CHECK1-NEXT:    call void @_ZN1SIfEC1Ev(%struct.S* noundef nonnull align 4 dereferenceable(4) [[TEST]])
99 // CHECK1-NEXT:    [[F:%.*]] = getelementptr inbounds [[STRUCT_S]], %struct.S* [[TEST]], i32 0, i32 0
100 // CHECK1-NEXT:    store float* [[F]], float** [[PVAR]], align 8
101 // CHECK1-NEXT:    store i64 0, i64* [[LVAR]], align 8
102 // CHECK1-NEXT:    [[TMP0:%.*]] = getelementptr inbounds [[STRUCT_ANON]], %struct.anon* [[OMP_OUTLINED_ARG_AGG_]], i32 0, i32 0
103 // CHECK1-NEXT:    store float** [[PVAR]], float*** [[TMP0]], align 8
104 // CHECK1-NEXT:    [[TMP1:%.*]] = getelementptr inbounds [[STRUCT_ANON]], %struct.anon* [[OMP_OUTLINED_ARG_AGG_]], i32 0, i32 1
105 // CHECK1-NEXT:    store i64* [[LVAR]], i64** [[TMP1]], align 8
106 // CHECK1-NEXT:    call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB3:[0-9]+]], i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, %struct.anon*)* @.omp_outlined. to void (i32*, i32*, ...)*), %struct.anon* [[OMP_OUTLINED_ARG_AGG_]])
107 // CHECK1-NEXT:    [[CALL:%.*]] = call noundef i32 @_Z5tmainIiET_v()
108 // CHECK1-NEXT:    store i32 [[CALL]], i32* [[RETVAL]], align 4
109 // CHECK1-NEXT:    call void @_ZN1SIfED1Ev(%struct.S* noundef nonnull align 4 dereferenceable(4) [[TEST]]) #[[ATTR4:[0-9]+]]
110 // CHECK1-NEXT:    [[TMP2:%.*]] = load i32, i32* [[RETVAL]], align 4
111 // CHECK1-NEXT:    ret i32 [[TMP2]]
112 //
113 //
114 // CHECK1-LABEL: define {{[^@]+}}@_ZN1SIfEC1Ev
115 // CHECK1-SAME: (%struct.S* noundef nonnull align 4 dereferenceable(4) [[THIS:%.*]]) unnamed_addr #[[ATTR1:[0-9]+]] align 2 {
116 // CHECK1-NEXT:  entry:
117 // CHECK1-NEXT:    [[THIS_ADDR:%.*]] = alloca %struct.S*, align 8
118 // CHECK1-NEXT:    store %struct.S* [[THIS]], %struct.S** [[THIS_ADDR]], align 8
119 // CHECK1-NEXT:    [[THIS1:%.*]] = load %struct.S*, %struct.S** [[THIS_ADDR]], align 8
120 // CHECK1-NEXT:    call void @_ZN1SIfEC2Ev(%struct.S* noundef nonnull align 4 dereferenceable(4) [[THIS1]])
121 // CHECK1-NEXT:    ret void
122 //
123 //
124 // CHECK1-LABEL: define {{[^@]+}}@.omp_outlined.
125 // CHECK1-SAME: (i32* noalias noundef [[DOTGLOBAL_TID_:%.*]], i32* noalias noundef [[DOTBOUND_TID_:%.*]], %struct.anon* noalias noundef [[__CONTEXT:%.*]]) #[[ATTR2:[0-9]+]] {
126 // CHECK1-NEXT:  entry:
127 // CHECK1-NEXT:    [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
128 // CHECK1-NEXT:    [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
129 // CHECK1-NEXT:    [[__CONTEXT_ADDR:%.*]] = alloca %struct.anon*, align 8
130 // CHECK1-NEXT:    [[DOTOMP_IV:%.*]] = alloca i32, align 4
131 // CHECK1-NEXT:    [[TMP:%.*]] = alloca i32, align 4
132 // CHECK1-NEXT:    [[DOTLINEAR_START:%.*]] = alloca float*, align 8
133 // CHECK1-NEXT:    [[DOTLINEAR_START1:%.*]] = alloca i64, align 8
134 // CHECK1-NEXT:    [[DOTOMP_LB:%.*]] = alloca i32, align 4
135 // CHECK1-NEXT:    [[DOTOMP_UB:%.*]] = alloca i32, align 4
136 // CHECK1-NEXT:    [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
137 // CHECK1-NEXT:    [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
138 // CHECK1-NEXT:    [[I:%.*]] = alloca i32, align 4
139 // CHECK1-NEXT:    [[PVAR:%.*]] = alloca float*, align 8
140 // CHECK1-NEXT:    [[LVAR:%.*]] = alloca i64, align 8
141 // CHECK1-NEXT:    store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
142 // CHECK1-NEXT:    store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
143 // CHECK1-NEXT:    store %struct.anon* [[__CONTEXT]], %struct.anon** [[__CONTEXT_ADDR]], align 8
144 // CHECK1-NEXT:    [[TMP0:%.*]] = load %struct.anon*, %struct.anon** [[__CONTEXT_ADDR]], align 8
145 // CHECK1-NEXT:    [[TMP1:%.*]] = getelementptr inbounds [[STRUCT_ANON:%.*]], %struct.anon* [[TMP0]], i32 0, i32 0
146 // CHECK1-NEXT:    [[TMP2:%.*]] = load float**, float*** [[TMP1]], align 8
147 // CHECK1-NEXT:    [[TMP3:%.*]] = getelementptr inbounds [[STRUCT_ANON]], %struct.anon* [[TMP0]], i32 0, i32 1
148 // CHECK1-NEXT:    [[TMP4:%.*]] = load i64*, i64** [[TMP3]], align 8
149 // CHECK1-NEXT:    [[TMP5:%.*]] = load float*, float** [[TMP2]], align 8
150 // CHECK1-NEXT:    store float* [[TMP5]], float** [[DOTLINEAR_START]], align 8
151 // CHECK1-NEXT:    [[TMP6:%.*]] = load i64, i64* [[TMP4]], align 8
152 // CHECK1-NEXT:    store i64 [[TMP6]], i64* [[DOTLINEAR_START1]], align 8
153 // CHECK1-NEXT:    store i32 0, i32* [[DOTOMP_LB]], align 4
154 // CHECK1-NEXT:    store i32 1, i32* [[DOTOMP_UB]], align 4
155 // CHECK1-NEXT:    store i32 1, i32* [[DOTOMP_STRIDE]], align 4
156 // CHECK1-NEXT:    store i32 0, i32* [[DOTOMP_IS_LAST]], align 4
157 // CHECK1-NEXT:    [[TMP7:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
158 // CHECK1-NEXT:    [[TMP8:%.*]] = load i32, i32* [[TMP7]], align 4
159 // CHECK1-NEXT:    call void @__kmpc_barrier(%struct.ident_t* @[[GLOB1:[0-9]+]], i32 [[TMP8]])
160 // CHECK1-NEXT:    call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB2:[0-9]+]], i32 [[TMP8]], i32 34, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_LB]], i32* [[DOTOMP_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1)
161 // CHECK1-NEXT:    [[TMP9:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
162 // CHECK1-NEXT:    [[CMP:%.*]] = icmp sgt i32 [[TMP9]], 1
163 // CHECK1-NEXT:    br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
164 // CHECK1:       cond.true:
165 // CHECK1-NEXT:    br label [[COND_END:%.*]]
166 // CHECK1:       cond.false:
167 // CHECK1-NEXT:    [[TMP10:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
168 // CHECK1-NEXT:    br label [[COND_END]]
169 // CHECK1:       cond.end:
170 // CHECK1-NEXT:    [[COND:%.*]] = phi i32 [ 1, [[COND_TRUE]] ], [ [[TMP10]], [[COND_FALSE]] ]
171 // CHECK1-NEXT:    store i32 [[COND]], i32* [[DOTOMP_UB]], align 4
172 // CHECK1-NEXT:    [[TMP11:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4
173 // CHECK1-NEXT:    store i32 [[TMP11]], i32* [[DOTOMP_IV]], align 4
174 // CHECK1-NEXT:    br label [[OMP_INNER_FOR_COND:%.*]]
175 // CHECK1:       omp.inner.for.cond:
176 // CHECK1-NEXT:    [[TMP12:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
177 // CHECK1-NEXT:    [[TMP13:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
178 // CHECK1-NEXT:    [[CMP2:%.*]] = icmp sle i32 [[TMP12]], [[TMP13]]
179 // CHECK1-NEXT:    br i1 [[CMP2]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
180 // CHECK1:       omp.inner.for.body:
181 // CHECK1-NEXT:    [[TMP14:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
182 // CHECK1-NEXT:    [[MUL:%.*]] = mul nsw i32 [[TMP14]], 1
183 // CHECK1-NEXT:    [[ADD:%.*]] = add nsw i32 0, [[MUL]]
184 // CHECK1-NEXT:    store i32 [[ADD]], i32* [[I]], align 4
185 // CHECK1-NEXT:    [[TMP15:%.*]] = load float*, float** [[DOTLINEAR_START]], align 8
186 // CHECK1-NEXT:    [[TMP16:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
187 // CHECK1-NEXT:    [[MUL3:%.*]] = mul nsw i32 [[TMP16]], 3
188 // CHECK1-NEXT:    [[IDX_EXT:%.*]] = sext i32 [[MUL3]] to i64
189 // CHECK1-NEXT:    [[ADD_PTR:%.*]] = getelementptr inbounds float, float* [[TMP15]], i64 [[IDX_EXT]]
190 // CHECK1-NEXT:    store float* [[ADD_PTR]], float** [[PVAR]], align 8
191 // CHECK1-NEXT:    [[TMP17:%.*]] = load i64, i64* [[DOTLINEAR_START1]], align 8
192 // CHECK1-NEXT:    [[TMP18:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
193 // CHECK1-NEXT:    [[MUL4:%.*]] = mul nsw i32 [[TMP18]], 3
194 // CHECK1-NEXT:    [[CONV:%.*]] = sext i32 [[MUL4]] to i64
195 // CHECK1-NEXT:    [[ADD5:%.*]] = add nsw i64 [[TMP17]], [[CONV]]
196 // CHECK1-NEXT:    store i64 [[ADD5]], i64* [[LVAR]], align 8
197 // CHECK1-NEXT:    [[TMP19:%.*]] = load float*, float** [[PVAR]], align 8
198 // CHECK1-NEXT:    [[ADD_PTR6:%.*]] = getelementptr inbounds float, float* [[TMP19]], i64 3
199 // CHECK1-NEXT:    store float* [[ADD_PTR6]], float** [[PVAR]], align 8
200 // CHECK1-NEXT:    [[TMP20:%.*]] = load i64, i64* [[LVAR]], align 8
201 // CHECK1-NEXT:    [[ADD7:%.*]] = add nsw i64 [[TMP20]], 3
202 // CHECK1-NEXT:    store i64 [[ADD7]], i64* [[LVAR]], align 8
203 // CHECK1-NEXT:    br label [[OMP_BODY_CONTINUE:%.*]]
204 // CHECK1:       omp.body.continue:
205 // CHECK1-NEXT:    br label [[OMP_INNER_FOR_INC:%.*]]
206 // CHECK1:       omp.inner.for.inc:
207 // CHECK1-NEXT:    [[TMP21:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
208 // CHECK1-NEXT:    [[ADD8:%.*]] = add nsw i32 [[TMP21]], 1
209 // CHECK1-NEXT:    store i32 [[ADD8]], i32* [[DOTOMP_IV]], align 4
210 // CHECK1-NEXT:    br label [[OMP_INNER_FOR_COND]]
211 // CHECK1:       omp.inner.for.end:
212 // CHECK1-NEXT:    br label [[OMP_LOOP_EXIT:%.*]]
213 // CHECK1:       omp.loop.exit:
214 // CHECK1-NEXT:    call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB2]], i32 [[TMP8]])
215 // CHECK1-NEXT:    [[TMP22:%.*]] = load i32, i32* [[DOTOMP_IS_LAST]], align 4
216 // CHECK1-NEXT:    [[TMP23:%.*]] = icmp ne i32 [[TMP22]], 0
217 // CHECK1-NEXT:    br i1 [[TMP23]], label [[DOTOMP_LINEAR_PU:%.*]], label [[DOTOMP_LINEAR_PU_DONE:%.*]]
218 // CHECK1:       .omp.linear.pu:
219 // CHECK1-NEXT:    [[TMP24:%.*]] = load float*, float** [[PVAR]], align 8
220 // CHECK1-NEXT:    store float* [[TMP24]], float** [[TMP2]], align 8
221 // CHECK1-NEXT:    [[TMP25:%.*]] = load i64, i64* [[LVAR]], align 8
222 // CHECK1-NEXT:    store i64 [[TMP25]], i64* [[TMP4]], align 8
223 // CHECK1-NEXT:    br label [[DOTOMP_LINEAR_PU_DONE]]
224 // CHECK1:       .omp.linear.pu.done:
225 // CHECK1-NEXT:    ret void
226 //
227 //
228 // CHECK1-LABEL: define {{[^@]+}}@_Z5tmainIiET_v
229 // CHECK1-SAME: () #[[ATTR5:[0-9]+]] {
230 // CHECK1-NEXT:  entry:
231 // CHECK1-NEXT:    [[TEST:%.*]] = alloca [[STRUCT_S_0:%.*]], align 4
232 // CHECK1-NEXT:    [[PVAR:%.*]] = alloca i32*, align 8
233 // CHECK1-NEXT:    [[LVAR:%.*]] = alloca i32, align 4
234 // CHECK1-NEXT:    [[OMP_OUTLINED_ARG_AGG_:%.*]] = alloca [[STRUCT_ANON_1:%.*]], align 8
235 // CHECK1-NEXT:    call void @_ZN1SIiEC1Ev(%struct.S.0* noundef nonnull align 4 dereferenceable(4) [[TEST]])
236 // CHECK1-NEXT:    [[F:%.*]] = getelementptr inbounds [[STRUCT_S_0]], %struct.S.0* [[TEST]], i32 0, i32 0
237 // CHECK1-NEXT:    store i32* [[F]], i32** [[PVAR]], align 8
238 // CHECK1-NEXT:    store i32 0, i32* [[LVAR]], align 4
239 // CHECK1-NEXT:    [[TMP0:%.*]] = getelementptr inbounds [[STRUCT_ANON_1]], %struct.anon.1* [[OMP_OUTLINED_ARG_AGG_]], i32 0, i32 0
240 // CHECK1-NEXT:    store i32** [[PVAR]], i32*** [[TMP0]], align 8
241 // CHECK1-NEXT:    [[TMP1:%.*]] = getelementptr inbounds [[STRUCT_ANON_1]], %struct.anon.1* [[OMP_OUTLINED_ARG_AGG_]], i32 0, i32 1
242 // CHECK1-NEXT:    store i32* [[LVAR]], i32** [[TMP1]], align 8
243 // CHECK1-NEXT:    call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB3]], i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, %struct.anon.1*)* @.omp_outlined..1 to void (i32*, i32*, ...)*), %struct.anon.1* [[OMP_OUTLINED_ARG_AGG_]])
244 // CHECK1-NEXT:    call void @_ZN1SIiED1Ev(%struct.S.0* noundef nonnull align 4 dereferenceable(4) [[TEST]]) #[[ATTR4]]
245 // CHECK1-NEXT:    ret i32 0
246 //
247 //
248 // CHECK1-LABEL: define {{[^@]+}}@_ZN1SIfED1Ev
249 // CHECK1-SAME: (%struct.S* noundef nonnull align 4 dereferenceable(4) [[THIS:%.*]]) unnamed_addr #[[ATTR1]] align 2 {
250 // CHECK1-NEXT:  entry:
251 // CHECK1-NEXT:    [[THIS_ADDR:%.*]] = alloca %struct.S*, align 8
252 // CHECK1-NEXT:    store %struct.S* [[THIS]], %struct.S** [[THIS_ADDR]], align 8
253 // CHECK1-NEXT:    [[THIS1:%.*]] = load %struct.S*, %struct.S** [[THIS_ADDR]], align 8
254 // CHECK1-NEXT:    call void @_ZN1SIfED2Ev(%struct.S* noundef nonnull align 4 dereferenceable(4) [[THIS1]]) #[[ATTR4]]
255 // CHECK1-NEXT:    ret void
256 //
257 //
258 // CHECK1-LABEL: define {{[^@]+}}@_ZN1SIfEC2Ev
259 // CHECK1-SAME: (%struct.S* noundef nonnull align 4 dereferenceable(4) [[THIS:%.*]]) unnamed_addr #[[ATTR1]] align 2 {
260 // CHECK1-NEXT:  entry:
261 // CHECK1-NEXT:    [[THIS_ADDR:%.*]] = alloca %struct.S*, align 8
262 // CHECK1-NEXT:    store %struct.S* [[THIS]], %struct.S** [[THIS_ADDR]], align 8
263 // CHECK1-NEXT:    [[THIS1:%.*]] = load %struct.S*, %struct.S** [[THIS_ADDR]], align 8
264 // CHECK1-NEXT:    [[F:%.*]] = getelementptr inbounds [[STRUCT_S:%.*]], %struct.S* [[THIS1]], i32 0, i32 0
265 // CHECK1-NEXT:    store float 0.000000e+00, float* [[F]], align 4
266 // CHECK1-NEXT:    ret void
267 //
268 //
269 // CHECK1-LABEL: define {{[^@]+}}@_ZN1SIiEC1Ev
270 // CHECK1-SAME: (%struct.S.0* noundef nonnull align 4 dereferenceable(4) [[THIS:%.*]]) unnamed_addr #[[ATTR1]] align 2 {
271 // CHECK1-NEXT:  entry:
272 // CHECK1-NEXT:    [[THIS_ADDR:%.*]] = alloca %struct.S.0*, align 8
273 // CHECK1-NEXT:    store %struct.S.0* [[THIS]], %struct.S.0** [[THIS_ADDR]], align 8
274 // CHECK1-NEXT:    [[THIS1:%.*]] = load %struct.S.0*, %struct.S.0** [[THIS_ADDR]], align 8
275 // CHECK1-NEXT:    call void @_ZN1SIiEC2Ev(%struct.S.0* noundef nonnull align 4 dereferenceable(4) [[THIS1]])
276 // CHECK1-NEXT:    ret void
277 //
278 //
279 // CHECK1-LABEL: define {{[^@]+}}@.omp_outlined..1
280 // CHECK1-SAME: (i32* noalias noundef [[DOTGLOBAL_TID_:%.*]], i32* noalias noundef [[DOTBOUND_TID_:%.*]], %struct.anon.1* noalias noundef [[__CONTEXT:%.*]]) #[[ATTR2]] {
281 // CHECK1-NEXT:  entry:
282 // CHECK1-NEXT:    [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
283 // CHECK1-NEXT:    [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
284 // CHECK1-NEXT:    [[__CONTEXT_ADDR:%.*]] = alloca %struct.anon.1*, align 8
285 // CHECK1-NEXT:    [[DOTOMP_IV:%.*]] = alloca i32, align 4
286 // CHECK1-NEXT:    [[TMP:%.*]] = alloca i32, align 4
287 // CHECK1-NEXT:    [[DOTLINEAR_START:%.*]] = alloca i32*, align 8
288 // CHECK1-NEXT:    [[DOTLINEAR_START1:%.*]] = alloca i32, align 4
289 // CHECK1-NEXT:    [[DOTOMP_LB:%.*]] = alloca i32, align 4
290 // CHECK1-NEXT:    [[DOTOMP_UB:%.*]] = alloca i32, align 4
291 // CHECK1-NEXT:    [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
292 // CHECK1-NEXT:    [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
293 // CHECK1-NEXT:    [[I:%.*]] = alloca i32, align 4
294 // CHECK1-NEXT:    [[PVAR:%.*]] = alloca i32*, align 8
295 // CHECK1-NEXT:    [[LVAR:%.*]] = alloca i32, align 4
296 // CHECK1-NEXT:    store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
297 // CHECK1-NEXT:    store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
298 // CHECK1-NEXT:    store %struct.anon.1* [[__CONTEXT]], %struct.anon.1** [[__CONTEXT_ADDR]], align 8
299 // CHECK1-NEXT:    [[TMP0:%.*]] = load %struct.anon.1*, %struct.anon.1** [[__CONTEXT_ADDR]], align 8
300 // CHECK1-NEXT:    [[TMP1:%.*]] = getelementptr inbounds [[STRUCT_ANON_1:%.*]], %struct.anon.1* [[TMP0]], i32 0, i32 0
301 // CHECK1-NEXT:    [[TMP2:%.*]] = load i32**, i32*** [[TMP1]], align 8
302 // CHECK1-NEXT:    [[TMP3:%.*]] = getelementptr inbounds [[STRUCT_ANON_1]], %struct.anon.1* [[TMP0]], i32 0, i32 1
303 // CHECK1-NEXT:    [[TMP4:%.*]] = load i32*, i32** [[TMP3]], align 8
304 // CHECK1-NEXT:    [[TMP5:%.*]] = load i32*, i32** [[TMP2]], align 8
305 // CHECK1-NEXT:    store i32* [[TMP5]], i32** [[DOTLINEAR_START]], align 8
306 // CHECK1-NEXT:    [[TMP6:%.*]] = load i32, i32* [[TMP4]], align 4
307 // CHECK1-NEXT:    store i32 [[TMP6]], i32* [[DOTLINEAR_START1]], align 4
308 // CHECK1-NEXT:    store i32 0, i32* [[DOTOMP_LB]], align 4
309 // CHECK1-NEXT:    store i32 1, i32* [[DOTOMP_UB]], align 4
310 // CHECK1-NEXT:    store i32 1, i32* [[DOTOMP_STRIDE]], align 4
311 // CHECK1-NEXT:    store i32 0, i32* [[DOTOMP_IS_LAST]], align 4
312 // CHECK1-NEXT:    [[TMP7:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
313 // CHECK1-NEXT:    [[TMP8:%.*]] = load i32, i32* [[TMP7]], align 4
314 // CHECK1-NEXT:    call void @__kmpc_barrier(%struct.ident_t* @[[GLOB1]], i32 [[TMP8]])
315 // CHECK1-NEXT:    call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB2]], i32 [[TMP8]], i32 34, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_LB]], i32* [[DOTOMP_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1)
316 // CHECK1-NEXT:    [[TMP9:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
317 // CHECK1-NEXT:    [[CMP:%.*]] = icmp sgt i32 [[TMP9]], 1
318 // CHECK1-NEXT:    br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
319 // CHECK1:       cond.true:
320 // CHECK1-NEXT:    br label [[COND_END:%.*]]
321 // CHECK1:       cond.false:
322 // CHECK1-NEXT:    [[TMP10:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
323 // CHECK1-NEXT:    br label [[COND_END]]
324 // CHECK1:       cond.end:
325 // CHECK1-NEXT:    [[COND:%.*]] = phi i32 [ 1, [[COND_TRUE]] ], [ [[TMP10]], [[COND_FALSE]] ]
326 // CHECK1-NEXT:    store i32 [[COND]], i32* [[DOTOMP_UB]], align 4
327 // CHECK1-NEXT:    [[TMP11:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4
328 // CHECK1-NEXT:    store i32 [[TMP11]], i32* [[DOTOMP_IV]], align 4
329 // CHECK1-NEXT:    br label [[OMP_INNER_FOR_COND:%.*]]
330 // CHECK1:       omp.inner.for.cond:
331 // CHECK1-NEXT:    [[TMP12:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
332 // CHECK1-NEXT:    [[TMP13:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
333 // CHECK1-NEXT:    [[CMP2:%.*]] = icmp sle i32 [[TMP12]], [[TMP13]]
334 // CHECK1-NEXT:    br i1 [[CMP2]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
335 // CHECK1:       omp.inner.for.body:
336 // CHECK1-NEXT:    [[TMP14:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
337 // CHECK1-NEXT:    [[MUL:%.*]] = mul nsw i32 [[TMP14]], 1
338 // CHECK1-NEXT:    [[ADD:%.*]] = add nsw i32 0, [[MUL]]
339 // CHECK1-NEXT:    store i32 [[ADD]], i32* [[I]], align 4
340 // CHECK1-NEXT:    [[TMP15:%.*]] = load i32*, i32** [[DOTLINEAR_START]], align 8
341 // CHECK1-NEXT:    [[TMP16:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
342 // CHECK1-NEXT:    [[MUL3:%.*]] = mul nsw i32 [[TMP16]], 1
343 // CHECK1-NEXT:    [[IDX_EXT:%.*]] = sext i32 [[MUL3]] to i64
344 // CHECK1-NEXT:    [[ADD_PTR:%.*]] = getelementptr inbounds i32, i32* [[TMP15]], i64 [[IDX_EXT]]
345 // CHECK1-NEXT:    store i32* [[ADD_PTR]], i32** [[PVAR]], align 8
346 // CHECK1-NEXT:    [[TMP17:%.*]] = load i32, i32* [[DOTLINEAR_START1]], align 4
347 // CHECK1-NEXT:    [[TMP18:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
348 // CHECK1-NEXT:    [[MUL4:%.*]] = mul nsw i32 [[TMP18]], 1
349 // CHECK1-NEXT:    [[ADD5:%.*]] = add nsw i32 [[TMP17]], [[MUL4]]
350 // CHECK1-NEXT:    store i32 [[ADD5]], i32* [[LVAR]], align 4
351 // CHECK1-NEXT:    [[TMP19:%.*]] = load i32*, i32** [[PVAR]], align 8
352 // CHECK1-NEXT:    [[INCDEC_PTR:%.*]] = getelementptr inbounds i32, i32* [[TMP19]], i32 1
353 // CHECK1-NEXT:    store i32* [[INCDEC_PTR]], i32** [[PVAR]], align 8
354 // CHECK1-NEXT:    [[TMP20:%.*]] = load i32, i32* [[LVAR]], align 4
355 // CHECK1-NEXT:    [[INC:%.*]] = add nsw i32 [[TMP20]], 1
356 // CHECK1-NEXT:    store i32 [[INC]], i32* [[LVAR]], align 4
357 // CHECK1-NEXT:    br label [[OMP_BODY_CONTINUE:%.*]]
358 // CHECK1:       omp.body.continue:
359 // CHECK1-NEXT:    br label [[OMP_INNER_FOR_INC:%.*]]
360 // CHECK1:       omp.inner.for.inc:
361 // CHECK1-NEXT:    [[TMP21:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
362 // CHECK1-NEXT:    [[ADD6:%.*]] = add nsw i32 [[TMP21]], 1
363 // CHECK1-NEXT:    store i32 [[ADD6]], i32* [[DOTOMP_IV]], align 4
364 // CHECK1-NEXT:    br label [[OMP_INNER_FOR_COND]]
365 // CHECK1:       omp.inner.for.end:
366 // CHECK1-NEXT:    br label [[OMP_LOOP_EXIT:%.*]]
367 // CHECK1:       omp.loop.exit:
368 // CHECK1-NEXT:    call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB2]], i32 [[TMP8]])
369 // CHECK1-NEXT:    [[TMP22:%.*]] = load i32, i32* [[DOTOMP_IS_LAST]], align 4
370 // CHECK1-NEXT:    [[TMP23:%.*]] = icmp ne i32 [[TMP22]], 0
371 // CHECK1-NEXT:    br i1 [[TMP23]], label [[DOTOMP_LINEAR_PU:%.*]], label [[DOTOMP_LINEAR_PU_DONE:%.*]]
372 // CHECK1:       .omp.linear.pu:
373 // CHECK1-NEXT:    [[TMP24:%.*]] = load i32*, i32** [[PVAR]], align 8
374 // CHECK1-NEXT:    store i32* [[TMP24]], i32** [[TMP2]], align 8
375 // CHECK1-NEXT:    [[TMP25:%.*]] = load i32, i32* [[LVAR]], align 4
376 // CHECK1-NEXT:    store i32 [[TMP25]], i32* [[TMP4]], align 4
377 // CHECK1-NEXT:    br label [[DOTOMP_LINEAR_PU_DONE]]
378 // CHECK1:       .omp.linear.pu.done:
379 // CHECK1-NEXT:    ret void
380 //
381 //
382 // CHECK1-LABEL: define {{[^@]+}}@_ZN1SIiED1Ev
383 // CHECK1-SAME: (%struct.S.0* noundef nonnull align 4 dereferenceable(4) [[THIS:%.*]]) unnamed_addr #[[ATTR1]] align 2 {
384 // CHECK1-NEXT:  entry:
385 // CHECK1-NEXT:    [[THIS_ADDR:%.*]] = alloca %struct.S.0*, align 8
386 // CHECK1-NEXT:    store %struct.S.0* [[THIS]], %struct.S.0** [[THIS_ADDR]], align 8
387 // CHECK1-NEXT:    [[THIS1:%.*]] = load %struct.S.0*, %struct.S.0** [[THIS_ADDR]], align 8
388 // CHECK1-NEXT:    call void @_ZN1SIiED2Ev(%struct.S.0* noundef nonnull align 4 dereferenceable(4) [[THIS1]]) #[[ATTR4]]
389 // CHECK1-NEXT:    ret void
390 //
391 //
392 // CHECK1-LABEL: define {{[^@]+}}@_ZN1SIiEC2Ev
393 // CHECK1-SAME: (%struct.S.0* noundef nonnull align 4 dereferenceable(4) [[THIS:%.*]]) unnamed_addr #[[ATTR1]] align 2 {
394 // CHECK1-NEXT:  entry:
395 // CHECK1-NEXT:    [[THIS_ADDR:%.*]] = alloca %struct.S.0*, align 8
396 // CHECK1-NEXT:    store %struct.S.0* [[THIS]], %struct.S.0** [[THIS_ADDR]], align 8
397 // CHECK1-NEXT:    [[THIS1:%.*]] = load %struct.S.0*, %struct.S.0** [[THIS_ADDR]], align 8
398 // CHECK1-NEXT:    [[F:%.*]] = getelementptr inbounds [[STRUCT_S_0:%.*]], %struct.S.0* [[THIS1]], i32 0, i32 0
399 // CHECK1-NEXT:    store i32 0, i32* [[F]], align 4
400 // CHECK1-NEXT:    ret void
401 //
402 //
403 // CHECK1-LABEL: define {{[^@]+}}@_ZN1SIiED2Ev
404 // CHECK1-SAME: (%struct.S.0* noundef nonnull align 4 dereferenceable(4) [[THIS:%.*]]) unnamed_addr #[[ATTR1]] align 2 {
405 // CHECK1-NEXT:  entry:
406 // CHECK1-NEXT:    [[THIS_ADDR:%.*]] = alloca %struct.S.0*, align 8
407 // CHECK1-NEXT:    store %struct.S.0* [[THIS]], %struct.S.0** [[THIS_ADDR]], align 8
408 // CHECK1-NEXT:    [[THIS1:%.*]] = load %struct.S.0*, %struct.S.0** [[THIS_ADDR]], align 8
409 // CHECK1-NEXT:    ret void
410 //
411 //
412 // CHECK1-LABEL: define {{[^@]+}}@_ZN1SIfED2Ev
413 // CHECK1-SAME: (%struct.S* noundef nonnull align 4 dereferenceable(4) [[THIS:%.*]]) unnamed_addr #[[ATTR1]] align 2 {
414 // CHECK1-NEXT:  entry:
415 // CHECK1-NEXT:    [[THIS_ADDR:%.*]] = alloca %struct.S*, align 8
416 // CHECK1-NEXT:    store %struct.S* [[THIS]], %struct.S** [[THIS_ADDR]], align 8
417 // CHECK1-NEXT:    [[THIS1:%.*]] = load %struct.S*, %struct.S** [[THIS_ADDR]], align 8
418 // CHECK1-NEXT:    ret void
419 //
420 //
421 // CHECK3-LABEL: define {{[^@]+}}@main
422 // CHECK3-SAME: () #[[ATTR0:[0-9]+]] {
423 // CHECK3-NEXT:  entry:
424 // CHECK3-NEXT:    [[RETVAL:%.*]] = alloca i32, align 4
425 // CHECK3-NEXT:    [[REF_TMP:%.*]] = alloca [[CLASS_ANON:%.*]], align 1
426 // CHECK3-NEXT:    store i32 0, i32* [[RETVAL]], align 4
427 // CHECK3-NEXT:    call void @"_ZZ4mainENK3$_0clEv"(%class.anon* noundef nonnull align 1 dereferenceable(1) [[REF_TMP]])
428 // CHECK3-NEXT:    ret i32 0
429 //
430 //
431 // CHECK3-LABEL: define {{[^@]+}}@.omp_outlined.
432 // CHECK3-SAME: (i32* noalias noundef [[DOTGLOBAL_TID_:%.*]], i32* noalias noundef [[DOTBOUND_TID_:%.*]], %struct.anon* noalias noundef [[__CONTEXT:%.*]]) #[[ATTR2:[0-9]+]] {
433 // CHECK3-NEXT:  entry:
434 // CHECK3-NEXT:    [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
435 // CHECK3-NEXT:    [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
436 // CHECK3-NEXT:    [[__CONTEXT_ADDR:%.*]] = alloca %struct.anon*, align 8
437 // CHECK3-NEXT:    [[DOTOMP_IV:%.*]] = alloca i32, align 4
438 // CHECK3-NEXT:    [[TMP:%.*]] = alloca i32, align 4
439 // CHECK3-NEXT:    [[DOTLINEAR_START:%.*]] = alloca i32, align 4
440 // CHECK3-NEXT:    [[DOTOMP_LB:%.*]] = alloca i32, align 4
441 // CHECK3-NEXT:    [[DOTOMP_UB:%.*]] = alloca i32, align 4
442 // CHECK3-NEXT:    [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
443 // CHECK3-NEXT:    [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
444 // CHECK3-NEXT:    [[I:%.*]] = alloca i32, align 4
445 // CHECK3-NEXT:    [[G:%.*]] = alloca i32, align 4
446 // CHECK3-NEXT:    [[REF_TMP:%.*]] = alloca [[CLASS_ANON_0:%.*]], align 8
447 // CHECK3-NEXT:    store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
448 // CHECK3-NEXT:    store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
449 // CHECK3-NEXT:    store %struct.anon* [[__CONTEXT]], %struct.anon** [[__CONTEXT_ADDR]], align 8
450 // CHECK3-NEXT:    [[TMP0:%.*]] = load %struct.anon*, %struct.anon** [[__CONTEXT_ADDR]], align 8
451 // CHECK3-NEXT:    [[TMP1:%.*]] = getelementptr inbounds [[STRUCT_ANON:%.*]], %struct.anon* [[TMP0]], i32 0, i32 0
452 // CHECK3-NEXT:    [[TMP2:%.*]] = load i32*, i32** [[TMP1]], align 8
453 // CHECK3-NEXT:    [[TMP3:%.*]] = load i32, i32* [[TMP2]], align 4
454 // CHECK3-NEXT:    store i32 [[TMP3]], i32* [[DOTLINEAR_START]], align 4
455 // CHECK3-NEXT:    store i32 0, i32* [[DOTOMP_LB]], align 4
456 // CHECK3-NEXT:    store i32 1, i32* [[DOTOMP_UB]], align 4
457 // CHECK3-NEXT:    store i32 1, i32* [[DOTOMP_STRIDE]], align 4
458 // CHECK3-NEXT:    store i32 0, i32* [[DOTOMP_IS_LAST]], align 4
459 // CHECK3-NEXT:    [[TMP4:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
460 // CHECK3-NEXT:    [[TMP5:%.*]] = load i32, i32* [[TMP4]], align 4
461 // CHECK3-NEXT:    call void @__kmpc_barrier(%struct.ident_t* @[[GLOB1:[0-9]+]], i32 [[TMP5]])
462 // CHECK3-NEXT:    call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB2:[0-9]+]], i32 [[TMP5]], i32 34, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_LB]], i32* [[DOTOMP_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1)
463 // CHECK3-NEXT:    [[TMP6:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
464 // CHECK3-NEXT:    [[CMP:%.*]] = icmp sgt i32 [[TMP6]], 1
465 // CHECK3-NEXT:    br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
466 // CHECK3:       cond.true:
467 // CHECK3-NEXT:    br label [[COND_END:%.*]]
468 // CHECK3:       cond.false:
469 // CHECK3-NEXT:    [[TMP7:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
470 // CHECK3-NEXT:    br label [[COND_END]]
471 // CHECK3:       cond.end:
472 // CHECK3-NEXT:    [[COND:%.*]] = phi i32 [ 1, [[COND_TRUE]] ], [ [[TMP7]], [[COND_FALSE]] ]
473 // CHECK3-NEXT:    store i32 [[COND]], i32* [[DOTOMP_UB]], align 4
474 // CHECK3-NEXT:    [[TMP8:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4
475 // CHECK3-NEXT:    store i32 [[TMP8]], i32* [[DOTOMP_IV]], align 4
476 // CHECK3-NEXT:    br label [[OMP_INNER_FOR_COND:%.*]]
477 // CHECK3:       omp.inner.for.cond:
478 // CHECK3-NEXT:    [[TMP9:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
479 // CHECK3-NEXT:    [[TMP10:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
480 // CHECK3-NEXT:    [[CMP1:%.*]] = icmp sle i32 [[TMP9]], [[TMP10]]
481 // CHECK3-NEXT:    br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
482 // CHECK3:       omp.inner.for.body:
483 // CHECK3-NEXT:    [[TMP11:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
484 // CHECK3-NEXT:    [[MUL:%.*]] = mul nsw i32 [[TMP11]], 1
485 // CHECK3-NEXT:    [[ADD:%.*]] = add nsw i32 0, [[MUL]]
486 // CHECK3-NEXT:    store i32 [[ADD]], i32* [[I]], align 4
487 // CHECK3-NEXT:    [[TMP12:%.*]] = load i32, i32* [[DOTLINEAR_START]], align 4
488 // CHECK3-NEXT:    [[TMP13:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
489 // CHECK3-NEXT:    [[MUL2:%.*]] = mul nsw i32 [[TMP13]], 5
490 // CHECK3-NEXT:    [[ADD3:%.*]] = add nsw i32 [[TMP12]], [[MUL2]]
491 // CHECK3-NEXT:    store i32 [[ADD3]], i32* [[G]], align 4
492 // CHECK3-NEXT:    [[TMP14:%.*]] = load i32, i32* [[G]], align 4
493 // CHECK3-NEXT:    [[ADD4:%.*]] = add nsw i32 [[TMP14]], 5
494 // CHECK3-NEXT:    store i32 [[ADD4]], i32* [[G]], align 4
495 // CHECK3-NEXT:    [[TMP15:%.*]] = getelementptr inbounds [[CLASS_ANON_0]], %class.anon.0* [[REF_TMP]], i32 0, i32 0
496 // CHECK3-NEXT:    store i32* [[G]], i32** [[TMP15]], align 8
497 // CHECK3-NEXT:    call void @"_ZZZ4mainENK3$_0clEvENKUlvE_clEv"(%class.anon.0* noundef nonnull align 8 dereferenceable(8) [[REF_TMP]])
498 // CHECK3-NEXT:    br label [[OMP_BODY_CONTINUE:%.*]]
499 // CHECK3:       omp.body.continue:
500 // CHECK3-NEXT:    br label [[OMP_INNER_FOR_INC:%.*]]
501 // CHECK3:       omp.inner.for.inc:
502 // CHECK3-NEXT:    [[TMP16:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
503 // CHECK3-NEXT:    [[ADD5:%.*]] = add nsw i32 [[TMP16]], 1
504 // CHECK3-NEXT:    store i32 [[ADD5]], i32* [[DOTOMP_IV]], align 4
505 // CHECK3-NEXT:    br label [[OMP_INNER_FOR_COND]]
506 // CHECK3:       omp.inner.for.end:
507 // CHECK3-NEXT:    br label [[OMP_LOOP_EXIT:%.*]]
508 // CHECK3:       omp.loop.exit:
509 // CHECK3-NEXT:    call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB2]], i32 [[TMP5]])
510 // CHECK3-NEXT:    [[TMP17:%.*]] = load i32, i32* [[DOTOMP_IS_LAST]], align 4
511 // CHECK3-NEXT:    [[TMP18:%.*]] = icmp ne i32 [[TMP17]], 0
512 // CHECK3-NEXT:    br i1 [[TMP18]], label [[DOTOMP_LINEAR_PU:%.*]], label [[DOTOMP_LINEAR_PU_DONE:%.*]]
513 // CHECK3:       .omp.linear.pu:
514 // CHECK3-NEXT:    [[TMP19:%.*]] = load i32, i32* [[G]], align 4
515 // CHECK3-NEXT:    store i32 [[TMP19]], i32* [[TMP2]], align 4
516 // CHECK3-NEXT:    br label [[DOTOMP_LINEAR_PU_DONE]]
517 // CHECK3:       .omp.linear.pu.done:
518 // CHECK3-NEXT:    ret void
519 //
520 //
521 // CHECK4-LABEL: define {{[^@]+}}@main
522 // CHECK4-SAME: () #[[ATTR1:[0-9]+]] {
523 // CHECK4-NEXT:  entry:
524 // CHECK4-NEXT:    [[RETVAL:%.*]] = alloca i32, align 4
525 // CHECK4-NEXT:    store i32 0, i32* [[RETVAL]], align 4
526 // CHECK4-NEXT:    [[TMP0:%.*]] = load i8*, i8** getelementptr inbounds ([[STRUCT___BLOCK_LITERAL_GENERIC:%.*]], %struct.__block_literal_generic* bitcast ({ i8**, i32, i32, i8*, %struct.__block_descriptor* }* @__block_literal_global to %struct.__block_literal_generic*), i32 0, i32 3), align 8
527 // CHECK4-NEXT:    [[TMP1:%.*]] = bitcast i8* [[TMP0]] to void (i8*)*
528 // CHECK4-NEXT:    call void [[TMP1]](i8* noundef bitcast ({ i8**, i32, i32, i8*, %struct.__block_descriptor* }* @__block_literal_global to i8*))
529 // CHECK4-NEXT:    ret i32 0
530 //
531 //
532 // CHECK4-LABEL: define {{[^@]+}}@__main_block_invoke
533 // CHECK4-SAME: (i8* noundef [[DOTBLOCK_DESCRIPTOR:%.*]]) #[[ATTR2:[0-9]+]] {
534 // CHECK4-NEXT:  entry:
535 // CHECK4-NEXT:    [[DOTBLOCK_DESCRIPTOR_ADDR:%.*]] = alloca i8*, align 8
536 // CHECK4-NEXT:    [[BLOCK_ADDR:%.*]] = alloca <{ i8*, i32, i32, i8*, %struct.__block_descriptor* }>*, align 8
537 // CHECK4-NEXT:    [[OMP_OUTLINED_ARG_AGG_:%.*]] = alloca [[STRUCT_ANON:%.*]], align 8
538 // CHECK4-NEXT:    store i8* [[DOTBLOCK_DESCRIPTOR]], i8** [[DOTBLOCK_DESCRIPTOR_ADDR]], align 8
539 // CHECK4-NEXT:    [[BLOCK:%.*]] = bitcast i8* [[DOTBLOCK_DESCRIPTOR]] to <{ i8*, i32, i32, i8*, %struct.__block_descriptor* }>*
540 // CHECK4-NEXT:    store <{ i8*, i32, i32, i8*, %struct.__block_descriptor* }>* [[BLOCK]], <{ i8*, i32, i32, i8*, %struct.__block_descriptor* }>** [[BLOCK_ADDR]], align 8
541 // CHECK4-NEXT:    [[TMP0:%.*]] = getelementptr inbounds [[STRUCT_ANON]], %struct.anon* [[OMP_OUTLINED_ARG_AGG_]], i32 0, i32 0
542 // CHECK4-NEXT:    store i32* @g, i32** [[TMP0]], align 8
543 // CHECK4-NEXT:    call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB3:[0-9]+]], i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, %struct.anon*)* @.omp_outlined. to void (i32*, i32*, ...)*), %struct.anon* [[OMP_OUTLINED_ARG_AGG_]])
544 // CHECK4-NEXT:    ret void
545 //
546 //
547 // CHECK4-LABEL: define {{[^@]+}}@.omp_outlined.
548 // CHECK4-SAME: (i32* noalias noundef [[DOTGLOBAL_TID_:%.*]], i32* noalias noundef [[DOTBOUND_TID_:%.*]], %struct.anon* noalias noundef [[__CONTEXT:%.*]]) #[[ATTR3:[0-9]+]] {
549 // CHECK4-NEXT:  entry:
550 // CHECK4-NEXT:    [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
551 // CHECK4-NEXT:    [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
552 // CHECK4-NEXT:    [[__CONTEXT_ADDR:%.*]] = alloca %struct.anon*, align 8
553 // CHECK4-NEXT:    [[DOTOMP_IV:%.*]] = alloca i32, align 4
554 // CHECK4-NEXT:    [[TMP:%.*]] = alloca i32, align 4
555 // CHECK4-NEXT:    [[DOTLINEAR_START:%.*]] = alloca i32, align 4
556 // CHECK4-NEXT:    [[DOTOMP_LB:%.*]] = alloca i32, align 4
557 // CHECK4-NEXT:    [[DOTOMP_UB:%.*]] = alloca i32, align 4
558 // CHECK4-NEXT:    [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
559 // CHECK4-NEXT:    [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
560 // CHECK4-NEXT:    [[I:%.*]] = alloca i32, align 4
561 // CHECK4-NEXT:    [[G:%.*]] = alloca i32, align 4
562 // CHECK4-NEXT:    [[BLOCK:%.*]] = alloca <{ i8*, i32, i32, i8*, %struct.__block_descriptor*, i32 }>, align 8
563 // CHECK4-NEXT:    store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
564 // CHECK4-NEXT:    store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
565 // CHECK4-NEXT:    store %struct.anon* [[__CONTEXT]], %struct.anon** [[__CONTEXT_ADDR]], align 8
566 // CHECK4-NEXT:    [[TMP0:%.*]] = load %struct.anon*, %struct.anon** [[__CONTEXT_ADDR]], align 8
567 // CHECK4-NEXT:    [[TMP1:%.*]] = getelementptr inbounds [[STRUCT_ANON:%.*]], %struct.anon* [[TMP0]], i32 0, i32 0
568 // CHECK4-NEXT:    [[TMP2:%.*]] = load i32*, i32** [[TMP1]], align 8
569 // CHECK4-NEXT:    [[TMP3:%.*]] = load i32, i32* [[TMP2]], align 4
570 // CHECK4-NEXT:    store i32 [[TMP3]], i32* [[DOTLINEAR_START]], align 4
571 // CHECK4-NEXT:    store i32 0, i32* [[DOTOMP_LB]], align 4
572 // CHECK4-NEXT:    store i32 1, i32* [[DOTOMP_UB]], align 4
573 // CHECK4-NEXT:    store i32 1, i32* [[DOTOMP_STRIDE]], align 4
574 // CHECK4-NEXT:    store i32 0, i32* [[DOTOMP_IS_LAST]], align 4
575 // CHECK4-NEXT:    [[TMP4:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
576 // CHECK4-NEXT:    [[TMP5:%.*]] = load i32, i32* [[TMP4]], align 4
577 // CHECK4-NEXT:    call void @__kmpc_barrier(%struct.ident_t* @[[GLOB1:[0-9]+]], i32 [[TMP5]])
578 // CHECK4-NEXT:    call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB2:[0-9]+]], i32 [[TMP5]], i32 34, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_LB]], i32* [[DOTOMP_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1)
579 // CHECK4-NEXT:    [[TMP6:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
580 // CHECK4-NEXT:    [[CMP:%.*]] = icmp sgt i32 [[TMP6]], 1
581 // CHECK4-NEXT:    br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
582 // CHECK4:       cond.true:
583 // CHECK4-NEXT:    br label [[COND_END:%.*]]
584 // CHECK4:       cond.false:
585 // CHECK4-NEXT:    [[TMP7:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
586 // CHECK4-NEXT:    br label [[COND_END]]
587 // CHECK4:       cond.end:
588 // CHECK4-NEXT:    [[COND:%.*]] = phi i32 [ 1, [[COND_TRUE]] ], [ [[TMP7]], [[COND_FALSE]] ]
589 // CHECK4-NEXT:    store i32 [[COND]], i32* [[DOTOMP_UB]], align 4
590 // CHECK4-NEXT:    [[TMP8:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4
591 // CHECK4-NEXT:    store i32 [[TMP8]], i32* [[DOTOMP_IV]], align 4
592 // CHECK4-NEXT:    br label [[OMP_INNER_FOR_COND:%.*]]
593 // CHECK4:       omp.inner.for.cond:
594 // CHECK4-NEXT:    [[TMP9:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
595 // CHECK4-NEXT:    [[TMP10:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
596 // CHECK4-NEXT:    [[CMP1:%.*]] = icmp sle i32 [[TMP9]], [[TMP10]]
597 // CHECK4-NEXT:    br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
598 // CHECK4:       omp.inner.for.body:
599 // CHECK4-NEXT:    [[TMP11:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
600 // CHECK4-NEXT:    [[MUL:%.*]] = mul nsw i32 [[TMP11]], 1
601 // CHECK4-NEXT:    [[ADD:%.*]] = add nsw i32 0, [[MUL]]
602 // CHECK4-NEXT:    store i32 [[ADD]], i32* [[I]], align 4
603 // CHECK4-NEXT:    [[TMP12:%.*]] = load i32, i32* [[DOTLINEAR_START]], align 4
604 // CHECK4-NEXT:    [[TMP13:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
605 // CHECK4-NEXT:    [[MUL2:%.*]] = mul nsw i32 [[TMP13]], 5
606 // CHECK4-NEXT:    [[ADD3:%.*]] = add nsw i32 [[TMP12]], [[MUL2]]
607 // CHECK4-NEXT:    store i32 [[ADD3]], i32* [[G]], align 4
608 // CHECK4-NEXT:    [[TMP14:%.*]] = load i32, i32* [[G]], align 4
609 // CHECK4-NEXT:    [[ADD4:%.*]] = add nsw i32 [[TMP14]], 5
610 // CHECK4-NEXT:    store i32 [[ADD4]], i32* [[G]], align 4
611 // CHECK4-NEXT:    store i32 1, i32* [[G]], align 4
612 // CHECK4-NEXT:    [[BLOCK_ISA:%.*]] = getelementptr inbounds <{ i8*, i32, i32, i8*, %struct.__block_descriptor*, i32 }>, <{ i8*, i32, i32, i8*, %struct.__block_descriptor*, i32 }>* [[BLOCK]], i32 0, i32 0
613 // CHECK4-NEXT:    store i8* bitcast (i8** @_NSConcreteStackBlock to i8*), i8** [[BLOCK_ISA]], align 8
614 // CHECK4-NEXT:    [[BLOCK_FLAGS:%.*]] = getelementptr inbounds <{ i8*, i32, i32, i8*, %struct.__block_descriptor*, i32 }>, <{ i8*, i32, i32, i8*, %struct.__block_descriptor*, i32 }>* [[BLOCK]], i32 0, i32 1
615 // CHECK4-NEXT:    store i32 1073741824, i32* [[BLOCK_FLAGS]], align 8
616 // CHECK4-NEXT:    [[BLOCK_RESERVED:%.*]] = getelementptr inbounds <{ i8*, i32, i32, i8*, %struct.__block_descriptor*, i32 }>, <{ i8*, i32, i32, i8*, %struct.__block_descriptor*, i32 }>* [[BLOCK]], i32 0, i32 2
617 // CHECK4-NEXT:    store i32 0, i32* [[BLOCK_RESERVED]], align 4
618 // CHECK4-NEXT:    [[BLOCK_INVOKE:%.*]] = getelementptr inbounds <{ i8*, i32, i32, i8*, %struct.__block_descriptor*, i32 }>, <{ i8*, i32, i32, i8*, %struct.__block_descriptor*, i32 }>* [[BLOCK]], i32 0, i32 3
619 // CHECK4-NEXT:    store i8* bitcast (void (i8*)* @g_block_invoke to i8*), i8** [[BLOCK_INVOKE]], align 8
620 // CHECK4-NEXT:    [[BLOCK_DESCRIPTOR:%.*]] = getelementptr inbounds <{ i8*, i32, i32, i8*, %struct.__block_descriptor*, i32 }>, <{ i8*, i32, i32, i8*, %struct.__block_descriptor*, i32 }>* [[BLOCK]], i32 0, i32 4
621 // CHECK4-NEXT:    store %struct.__block_descriptor* bitcast ({ i64, i64, i8*, i8* }* @__block_descriptor_tmp.1 to %struct.__block_descriptor*), %struct.__block_descriptor** [[BLOCK_DESCRIPTOR]], align 8
622 // CHECK4-NEXT:    [[BLOCK_CAPTURED:%.*]] = getelementptr inbounds <{ i8*, i32, i32, i8*, %struct.__block_descriptor*, i32 }>, <{ i8*, i32, i32, i8*, %struct.__block_descriptor*, i32 }>* [[BLOCK]], i32 0, i32 5
623 // CHECK4-NEXT:    [[TMP15:%.*]] = load volatile i32, i32* [[G]], align 4
624 // CHECK4-NEXT:    store volatile i32 [[TMP15]], i32* [[BLOCK_CAPTURED]], align 8
625 // CHECK4-NEXT:    [[TMP16:%.*]] = bitcast <{ i8*, i32, i32, i8*, %struct.__block_descriptor*, i32 }>* [[BLOCK]] to void ()*
626 // CHECK4-NEXT:    [[BLOCK_LITERAL:%.*]] = bitcast void ()* [[TMP16]] to %struct.__block_literal_generic*
627 // CHECK4-NEXT:    [[TMP17:%.*]] = getelementptr inbounds [[STRUCT___BLOCK_LITERAL_GENERIC:%.*]], %struct.__block_literal_generic* [[BLOCK_LITERAL]], i32 0, i32 3
628 // CHECK4-NEXT:    [[TMP18:%.*]] = bitcast %struct.__block_literal_generic* [[BLOCK_LITERAL]] to i8*
629 // CHECK4-NEXT:    [[TMP19:%.*]] = load i8*, i8** [[TMP17]], align 8
630 // CHECK4-NEXT:    [[TMP20:%.*]] = bitcast i8* [[TMP19]] to void (i8*)*
631 // CHECK4-NEXT:    call void [[TMP20]](i8* noundef [[TMP18]])
632 // CHECK4-NEXT:    br label [[OMP_BODY_CONTINUE:%.*]]
633 // CHECK4:       omp.body.continue:
634 // CHECK4-NEXT:    br label [[OMP_INNER_FOR_INC:%.*]]
635 // CHECK4:       omp.inner.for.inc:
636 // CHECK4-NEXT:    [[TMP21:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
637 // CHECK4-NEXT:    [[ADD5:%.*]] = add nsw i32 [[TMP21]], 1
638 // CHECK4-NEXT:    store i32 [[ADD5]], i32* [[DOTOMP_IV]], align 4
639 // CHECK4-NEXT:    br label [[OMP_INNER_FOR_COND]]
640 // CHECK4:       omp.inner.for.end:
641 // CHECK4-NEXT:    br label [[OMP_LOOP_EXIT:%.*]]
642 // CHECK4:       omp.loop.exit:
643 // CHECK4-NEXT:    call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB2]], i32 [[TMP5]])
644 // CHECK4-NEXT:    [[TMP22:%.*]] = load i32, i32* [[DOTOMP_IS_LAST]], align 4
645 // CHECK4-NEXT:    [[TMP23:%.*]] = icmp ne i32 [[TMP22]], 0
646 // CHECK4-NEXT:    br i1 [[TMP23]], label [[DOTOMP_LINEAR_PU:%.*]], label [[DOTOMP_LINEAR_PU_DONE:%.*]]
647 // CHECK4:       .omp.linear.pu:
648 // CHECK4-NEXT:    [[TMP24:%.*]] = load i32, i32* [[G]], align 4
649 // CHECK4-NEXT:    store i32 [[TMP24]], i32* [[TMP2]], align 4
650 // CHECK4-NEXT:    br label [[DOTOMP_LINEAR_PU_DONE]]
651 // CHECK4:       .omp.linear.pu.done:
652 // CHECK4-NEXT:    ret void
653 //
654 //
655 // CHECK4-LABEL: define {{[^@]+}}@g_block_invoke
656 // CHECK4-SAME: (i8* noundef [[DOTBLOCK_DESCRIPTOR:%.*]]) #[[ATTR2]] {
657 // CHECK4-NEXT:  entry:
658 // CHECK4-NEXT:    [[DOTBLOCK_DESCRIPTOR_ADDR:%.*]] = alloca i8*, align 8
659 // CHECK4-NEXT:    [[BLOCK_ADDR:%.*]] = alloca <{ i8*, i32, i32, i8*, %struct.__block_descriptor*, i32 }>*, align 8
660 // CHECK4-NEXT:    store i8* [[DOTBLOCK_DESCRIPTOR]], i8** [[DOTBLOCK_DESCRIPTOR_ADDR]], align 8
661 // CHECK4-NEXT:    [[BLOCK:%.*]] = bitcast i8* [[DOTBLOCK_DESCRIPTOR]] to <{ i8*, i32, i32, i8*, %struct.__block_descriptor*, i32 }>*
662 // CHECK4-NEXT:    store <{ i8*, i32, i32, i8*, %struct.__block_descriptor*, i32 }>* [[BLOCK]], <{ i8*, i32, i32, i8*, %struct.__block_descriptor*, i32 }>** [[BLOCK_ADDR]], align 8
663 // CHECK4-NEXT:    [[BLOCK_CAPTURE_ADDR:%.*]] = getelementptr inbounds <{ i8*, i32, i32, i8*, %struct.__block_descriptor*, i32 }>, <{ i8*, i32, i32, i8*, %struct.__block_descriptor*, i32 }>* [[BLOCK]], i32 0, i32 5
664 // CHECK4-NEXT:    store i32 2, i32* [[BLOCK_CAPTURE_ADDR]], align 8
665 // CHECK4-NEXT:    ret void
666 //
667