1 // NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --function-signature --include-generated-funcs --replace-value-regex "__omp_offloading_[0-9a-z]+_[0-9a-z]+" "reduction_size[.].+[.]" "pl_cond[.].+[.|,]" --prefix-filecheck-ir-name _
2 // RUN: %clang_cc1 -verify -fopenmp -x c++ -triple x86_64-apple-darwin10 -emit-llvm %s -o - | FileCheck %s --check-prefix=CHECK1
3 // RUN: %clang_cc1 -fopenmp -x c++ -std=c++11 -triple x86_64-apple-darwin10 -emit-pch -o %t %s
4 // RUN: %clang_cc1 -fopenmp -x c++ -triple x86_64-apple-darwin10 -std=c++11 -include-pch %t -verify %s -emit-llvm -o - | FileCheck %s --check-prefix=CHECK2
5 // RUN: %clang_cc1 -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 -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 -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 -fopenmp-simd -x c++ -std=c++11 -triple x86_64-apple-darwin10 -emit-pch -o %t %s
10 // RUN: %clang_cc1 -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 -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 -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 enum omp_allocator_handle_t {
18   omp_null_allocator = 0,
19   omp_default_mem_alloc = 1,
20   omp_large_cap_mem_alloc = 2,
21   omp_const_mem_alloc = 3,
22   omp_high_bw_mem_alloc = 4,
23   omp_low_lat_mem_alloc = 5,
24   omp_cgroup_mem_alloc = 6,
25   omp_pteam_mem_alloc = 7,
26   omp_thread_mem_alloc = 8,
27   KMP_ALLOCATOR_MAX_HANDLE = __UINTPTR_MAX__
28 };
29 
30 template <class T>
31 struct S {
32   T f;
33   S(T a) : f(a) {}
34   S() : f() {}
35   S<T> &operator=(const S<T> &);
36   operator T() { return T(); }
37   ~S() {}
38 };
39 
40 volatile int g = 1212;
41 volatile int &g1 = g;
42 float f;
43 char cnt;
44 
45 struct SS {
46   int a;
47   int b : 4;
48   int &c;
49   SS(int &d) : a(0), b(0), c(d) {
50 #pragma omp parallel
51 #pragma omp for linear(a, b, c)
52     for (int i = 0; i < 2; ++i)
53 #ifdef LAMBDA
54       [&]() {
55         ++this->a, --b, (this)->c /= 1;
56 #pragma omp parallel
57 #pragma omp for linear(a, b) linear(ref(c))
58         for (int i = 0; i < 2; ++i)
59           ++(this)->a, --b, this->c /= 1;
60       }();
61 #elif defined(BLOCKS)
62       ^{
63         ++a;
64         --this->b;
65         (this)->c /= 1;
66 #pragma omp parallel
67 #pragma omp for linear(a, b) linear(uval(c))
68         for (int i = 0; i < 2; ++i)
69           ++(this)->a, --b, this->c /= 1;
70       }();
71 #else
72       ++this->a, --b, c /= 1;
73 #endif
74   }
75 };
76 
77 template <typename T>
78 struct SST {
79   T a;
80   SST() : a(T()) {
81 #pragma omp parallel
82 #pragma omp for linear(a)
83     for (int i = 0; i < 2; ++i)
84 #ifdef LAMBDA
85       [&]() {
86         [&]() {
87           ++this->a;
88 #pragma omp parallel
89 #pragma omp for linear(a)
90           for (int i = 0; i < 2; ++i)
91             ++(this)->a;
92         }();
93       }();
94 #elif defined(BLOCKS)
95       ^{
96         ^{
97           ++a;
98 #pragma omp parallel
99 #pragma omp for linear(a)
100           for (int i = 0; i < 2; ++i)
101             ++(this)->a;
102         }();
103       }();
104 #else
105       ++(this)->a;
106 #endif
107   }
108 };
109 
110 template <typename T>
111 T tmain() {
112   S<T> test;
113   SST<T> sst;
114   T *pvar = &test.f;
115   T &lvar = test.f;
116 #pragma omp parallel
117 #pragma omp for linear(pvar, lvar)
118   for (int i = 0; i < 2; ++i) {
119     ++pvar, ++lvar;
120   }
121   return T();
122 }
123 
124 int main() {
125   static int sivar;
126   SS ss(sivar);
127 #ifdef LAMBDA
128   [&]() {
129 #pragma omp parallel
130 #pragma omp for linear(g, g1:5)
131   for (int i = 0; i < 2; ++i) {
132 
133 
134 
135     g += 5;
136     g1 += 5;
137     [&]() {
138       g = 2;
139       g1 = 2;
140     }();
141   }
142   }();
143   return 0;
144 #elif defined(BLOCKS)
145   ^{
146 #pragma omp parallel
147 #pragma omp for linear(g, g1:5)
148   for (int i = 0; i < 2; ++i) {
149     g += 5;
150     g1 += 5;
151     g = 1;
152     g1 = 5;
153     ^{
154       g = 2;
155       g1 = 2;
156     }();
157   }
158   }();
159   return 0;
160 
161 
162 #else
163   S<float> test;
164   float *pvar = &test.f;
165   long long lvar = 0;
166 #pragma omp parallel
167 #pragma omp for linear(pvar, lvar : 3) allocate(omp_low_lat_mem_alloc: lvar)
168   for (int i = 0; i < 2; ++i) {
169     pvar += 3, lvar += 3;
170   }
171   return tmain<int>();
172 #endif
173 }
174 
175 
176 
177 // Check for default initialization.
178 
179 
180 
181 
182 
183 // Check for default initialization.
184 
185 #endif
186 
187 // CHECK1-LABEL: define {{[^@]+}}@main
188 // CHECK1-SAME: () #[[ATTR0:[0-9]+]] {
189 // CHECK1-NEXT:  entry:
190 // CHECK1-NEXT:    [[RETVAL:%.*]] = alloca i32, align 4
191 // CHECK1-NEXT:    [[SS:%.*]] = alloca [[STRUCT_SS:%.*]], align 8
192 // CHECK1-NEXT:    [[TEST:%.*]] = alloca [[STRUCT_S:%.*]], align 4
193 // CHECK1-NEXT:    [[PVAR:%.*]] = alloca float*, align 8
194 // CHECK1-NEXT:    [[LVAR:%.*]] = alloca i64, align 8
195 // CHECK1-NEXT:    store i32 0, i32* [[RETVAL]], align 4
196 // CHECK1-NEXT:    call void @_ZN2SSC1ERi(%struct.SS* nonnull align 8 dereferenceable(16) [[SS]], i32* nonnull align 4 dereferenceable(4) @_ZZ4mainE5sivar)
197 // CHECK1-NEXT:    call void @_ZN1SIfEC1Ev(%struct.S* nonnull align 4 dereferenceable(4) [[TEST]])
198 // CHECK1-NEXT:    [[F:%.*]] = getelementptr inbounds [[STRUCT_S]], %struct.S* [[TEST]], i32 0, i32 0
199 // CHECK1-NEXT:    store float* [[F]], float** [[PVAR]], align 8
200 // CHECK1-NEXT:    store i64 0, i64* [[LVAR]], align 8
201 // CHECK1-NEXT:    call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB3:[0-9]+]], i32 2, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, float**, i64*)* @.omp_outlined. to void (i32*, i32*, ...)*), float** [[PVAR]], i64* [[LVAR]])
202 // CHECK1-NEXT:    [[CALL:%.*]] = call i32 @_Z5tmainIiET_v()
203 // CHECK1-NEXT:    store i32 [[CALL]], i32* [[RETVAL]], align 4
204 // CHECK1-NEXT:    call void @_ZN1SIfED1Ev(%struct.S* nonnull align 4 dereferenceable(4) [[TEST]]) #[[ATTR4:[0-9]+]]
205 // CHECK1-NEXT:    [[TMP0:%.*]] = load i32, i32* [[RETVAL]], align 4
206 // CHECK1-NEXT:    ret i32 [[TMP0]]
207 //
208 //
209 // CHECK1-LABEL: define {{[^@]+}}@_ZN2SSC1ERi
210 // CHECK1-SAME: (%struct.SS* nonnull align 8 dereferenceable(16) [[THIS:%.*]], i32* nonnull align 4 dereferenceable(4) [[D:%.*]]) unnamed_addr #[[ATTR1:[0-9]+]] align 2 {
211 // CHECK1-NEXT:  entry:
212 // CHECK1-NEXT:    [[THIS_ADDR:%.*]] = alloca %struct.SS*, align 8
213 // CHECK1-NEXT:    [[D_ADDR:%.*]] = alloca i32*, align 8
214 // CHECK1-NEXT:    store %struct.SS* [[THIS]], %struct.SS** [[THIS_ADDR]], align 8
215 // CHECK1-NEXT:    store i32* [[D]], i32** [[D_ADDR]], align 8
216 // CHECK1-NEXT:    [[THIS1:%.*]] = load %struct.SS*, %struct.SS** [[THIS_ADDR]], align 8
217 // CHECK1-NEXT:    [[TMP0:%.*]] = load i32*, i32** [[D_ADDR]], align 8
218 // CHECK1-NEXT:    call void @_ZN2SSC2ERi(%struct.SS* nonnull align 8 dereferenceable(16) [[THIS1]], i32* nonnull align 4 dereferenceable(4) [[TMP0]])
219 // CHECK1-NEXT:    ret void
220 //
221 //
222 // CHECK1-LABEL: define {{[^@]+}}@_ZN1SIfEC1Ev
223 // CHECK1-SAME: (%struct.S* nonnull align 4 dereferenceable(4) [[THIS:%.*]]) unnamed_addr #[[ATTR1]] align 2 {
224 // CHECK1-NEXT:  entry:
225 // CHECK1-NEXT:    [[THIS_ADDR:%.*]] = alloca %struct.S*, align 8
226 // CHECK1-NEXT:    store %struct.S* [[THIS]], %struct.S** [[THIS_ADDR]], align 8
227 // CHECK1-NEXT:    [[THIS1:%.*]] = load %struct.S*, %struct.S** [[THIS_ADDR]], align 8
228 // CHECK1-NEXT:    call void @_ZN1SIfEC2Ev(%struct.S* nonnull align 4 dereferenceable(4) [[THIS1]])
229 // CHECK1-NEXT:    ret void
230 //
231 //
232 // CHECK1-LABEL: define {{[^@]+}}@.omp_outlined.
233 // CHECK1-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], float** nonnull align 8 dereferenceable(8) [[PVAR:%.*]], i64* nonnull align 8 dereferenceable(8) [[LVAR:%.*]]) #[[ATTR2:[0-9]+]] {
234 // CHECK1-NEXT:  entry:
235 // CHECK1-NEXT:    [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
236 // CHECK1-NEXT:    [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
237 // CHECK1-NEXT:    [[PVAR_ADDR:%.*]] = alloca float**, align 8
238 // CHECK1-NEXT:    [[LVAR_ADDR:%.*]] = alloca i64*, align 8
239 // CHECK1-NEXT:    [[DOTOMP_IV:%.*]] = alloca i32, align 4
240 // CHECK1-NEXT:    [[TMP:%.*]] = alloca i32, align 4
241 // CHECK1-NEXT:    [[DOTLINEAR_START:%.*]] = alloca float*, align 8
242 // CHECK1-NEXT:    [[DOTLINEAR_START1:%.*]] = alloca i64, align 8
243 // CHECK1-NEXT:    [[DOTOMP_LB:%.*]] = alloca i32, align 4
244 // CHECK1-NEXT:    [[DOTOMP_UB:%.*]] = alloca i32, align 4
245 // CHECK1-NEXT:    [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
246 // CHECK1-NEXT:    [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
247 // CHECK1-NEXT:    [[I:%.*]] = alloca i32, align 4
248 // CHECK1-NEXT:    [[PVAR2:%.*]] = alloca float*, align 8
249 // CHECK1-NEXT:    store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
250 // CHECK1-NEXT:    store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
251 // CHECK1-NEXT:    store float** [[PVAR]], float*** [[PVAR_ADDR]], align 8
252 // CHECK1-NEXT:    store i64* [[LVAR]], i64** [[LVAR_ADDR]], align 8
253 // CHECK1-NEXT:    [[TMP0:%.*]] = load float**, float*** [[PVAR_ADDR]], align 8
254 // CHECK1-NEXT:    [[TMP1:%.*]] = load i64*, i64** [[LVAR_ADDR]], align 8
255 // CHECK1-NEXT:    [[TMP2:%.*]] = load float*, float** [[TMP0]], align 8
256 // CHECK1-NEXT:    store float* [[TMP2]], float** [[DOTLINEAR_START]], align 8
257 // CHECK1-NEXT:    [[TMP3:%.*]] = load i64, i64* [[TMP1]], align 8
258 // CHECK1-NEXT:    store i64 [[TMP3]], i64* [[DOTLINEAR_START1]], align 8
259 // CHECK1-NEXT:    store i32 0, i32* [[DOTOMP_LB]], align 4
260 // CHECK1-NEXT:    store i32 1, i32* [[DOTOMP_UB]], align 4
261 // CHECK1-NEXT:    store i32 1, i32* [[DOTOMP_STRIDE]], align 4
262 // CHECK1-NEXT:    store i32 0, i32* [[DOTOMP_IS_LAST]], align 4
263 // CHECK1-NEXT:    [[TMP4:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
264 // CHECK1-NEXT:    [[TMP5:%.*]] = load i32, i32* [[TMP4]], align 4
265 // CHECK1-NEXT:    call void @__kmpc_barrier(%struct.ident_t* @[[GLOB1:[0-9]+]], i32 [[TMP5]])
266 // CHECK1-NEXT:    [[DOTLVAR__VOID_ADDR:%.*]] = call i8* @__kmpc_alloc(i32 [[TMP5]], i64 8, i8* inttoptr (i64 5 to i8*))
267 // CHECK1-NEXT:    [[DOTLVAR__ADDR:%.*]] = bitcast i8* [[DOTLVAR__VOID_ADDR]] to i64*
268 // CHECK1-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)
269 // CHECK1-NEXT:    [[TMP6:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
270 // CHECK1-NEXT:    [[CMP:%.*]] = icmp sgt i32 [[TMP6]], 1
271 // CHECK1-NEXT:    br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
272 // CHECK1:       cond.true:
273 // CHECK1-NEXT:    br label [[COND_END:%.*]]
274 // CHECK1:       cond.false:
275 // CHECK1-NEXT:    [[TMP7:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
276 // CHECK1-NEXT:    br label [[COND_END]]
277 // CHECK1:       cond.end:
278 // CHECK1-NEXT:    [[COND:%.*]] = phi i32 [ 1, [[COND_TRUE]] ], [ [[TMP7]], [[COND_FALSE]] ]
279 // CHECK1-NEXT:    store i32 [[COND]], i32* [[DOTOMP_UB]], align 4
280 // CHECK1-NEXT:    [[TMP8:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4
281 // CHECK1-NEXT:    store i32 [[TMP8]], i32* [[DOTOMP_IV]], align 4
282 // CHECK1-NEXT:    br label [[OMP_INNER_FOR_COND:%.*]]
283 // CHECK1:       omp.inner.for.cond:
284 // CHECK1-NEXT:    [[TMP9:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
285 // CHECK1-NEXT:    [[TMP10:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
286 // CHECK1-NEXT:    [[CMP3:%.*]] = icmp sle i32 [[TMP9]], [[TMP10]]
287 // CHECK1-NEXT:    br i1 [[CMP3]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_COND_CLEANUP:%.*]]
288 // CHECK1:       omp.inner.for.cond.cleanup:
289 // CHECK1-NEXT:    br label [[OMP_INNER_FOR_END:%.*]]
290 // CHECK1:       omp.inner.for.body:
291 // CHECK1-NEXT:    [[TMP11:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
292 // CHECK1-NEXT:    [[MUL:%.*]] = mul nsw i32 [[TMP11]], 1
293 // CHECK1-NEXT:    [[ADD:%.*]] = add nsw i32 0, [[MUL]]
294 // CHECK1-NEXT:    store i32 [[ADD]], i32* [[I]], align 4
295 // CHECK1-NEXT:    [[TMP12:%.*]] = load float*, float** [[DOTLINEAR_START]], align 8
296 // CHECK1-NEXT:    [[TMP13:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
297 // CHECK1-NEXT:    [[MUL4:%.*]] = mul nsw i32 [[TMP13]], 3
298 // CHECK1-NEXT:    [[IDX_EXT:%.*]] = sext i32 [[MUL4]] to i64
299 // CHECK1-NEXT:    [[ADD_PTR:%.*]] = getelementptr inbounds float, float* [[TMP12]], i64 [[IDX_EXT]]
300 // CHECK1-NEXT:    store float* [[ADD_PTR]], float** [[PVAR2]], align 8
301 // CHECK1-NEXT:    [[TMP14:%.*]] = load i64, i64* [[DOTLINEAR_START1]], align 8
302 // CHECK1-NEXT:    [[TMP15:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
303 // CHECK1-NEXT:    [[MUL5:%.*]] = mul nsw i32 [[TMP15]], 3
304 // CHECK1-NEXT:    [[CONV:%.*]] = sext i32 [[MUL5]] to i64
305 // CHECK1-NEXT:    [[ADD6:%.*]] = add nsw i64 [[TMP14]], [[CONV]]
306 // CHECK1-NEXT:    store i64 [[ADD6]], i64* [[DOTLVAR__ADDR]], align 8
307 // CHECK1-NEXT:    [[TMP16:%.*]] = load float*, float** [[PVAR2]], align 8
308 // CHECK1-NEXT:    [[ADD_PTR7:%.*]] = getelementptr inbounds float, float* [[TMP16]], i64 3
309 // CHECK1-NEXT:    store float* [[ADD_PTR7]], float** [[PVAR2]], align 8
310 // CHECK1-NEXT:    [[TMP17:%.*]] = load i64, i64* [[DOTLVAR__ADDR]], align 8
311 // CHECK1-NEXT:    [[ADD8:%.*]] = add nsw i64 [[TMP17]], 3
312 // CHECK1-NEXT:    store i64 [[ADD8]], i64* [[DOTLVAR__ADDR]], align 8
313 // CHECK1-NEXT:    br label [[OMP_BODY_CONTINUE:%.*]]
314 // CHECK1:       omp.body.continue:
315 // CHECK1-NEXT:    br label [[OMP_INNER_FOR_INC:%.*]]
316 // CHECK1:       omp.inner.for.inc:
317 // CHECK1-NEXT:    [[TMP18:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
318 // CHECK1-NEXT:    [[ADD9:%.*]] = add nsw i32 [[TMP18]], 1
319 // CHECK1-NEXT:    store i32 [[ADD9]], i32* [[DOTOMP_IV]], align 4
320 // CHECK1-NEXT:    br label [[OMP_INNER_FOR_COND]]
321 // CHECK1:       omp.inner.for.end:
322 // CHECK1-NEXT:    br label [[OMP_LOOP_EXIT:%.*]]
323 // CHECK1:       omp.loop.exit:
324 // CHECK1-NEXT:    call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB2]], i32 [[TMP5]])
325 // CHECK1-NEXT:    [[TMP19:%.*]] = bitcast i64* [[DOTLVAR__ADDR]] to i8*
326 // CHECK1-NEXT:    call void @__kmpc_free(i32 [[TMP5]], i8* [[TMP19]], i8* inttoptr (i64 5 to i8*))
327 // CHECK1-NEXT:    [[TMP20:%.*]] = load i32, i32* [[DOTOMP_IS_LAST]], align 4
328 // CHECK1-NEXT:    [[TMP21:%.*]] = icmp ne i32 [[TMP20]], 0
329 // CHECK1-NEXT:    br i1 [[TMP21]], label [[DOTOMP_LINEAR_PU:%.*]], label [[DOTOMP_LINEAR_PU_DONE:%.*]]
330 // CHECK1:       .omp.linear.pu:
331 // CHECK1-NEXT:    [[TMP22:%.*]] = load float*, float** [[PVAR2]], align 8
332 // CHECK1-NEXT:    store float* [[TMP22]], float** [[TMP0]], align 8
333 // CHECK1-NEXT:    [[TMP23:%.*]] = load i64, i64* [[DOTLVAR__ADDR]], align 8
334 // CHECK1-NEXT:    store i64 [[TMP23]], i64* [[TMP1]], align 8
335 // CHECK1-NEXT:    br label [[DOTOMP_LINEAR_PU_DONE]]
336 // CHECK1:       .omp.linear.pu.done:
337 // CHECK1-NEXT:    call void @__kmpc_barrier(%struct.ident_t* @[[GLOB1]], i32 [[TMP5]])
338 // CHECK1-NEXT:    ret void
339 //
340 //
341 // CHECK1-LABEL: define {{[^@]+}}@_Z5tmainIiET_v
342 // CHECK1-SAME: () #[[ATTR5:[0-9]+]] {
343 // CHECK1-NEXT:  entry:
344 // CHECK1-NEXT:    [[TEST:%.*]] = alloca [[STRUCT_S_0:%.*]], align 4
345 // CHECK1-NEXT:    [[SST:%.*]] = alloca [[STRUCT_SST:%.*]], align 4
346 // CHECK1-NEXT:    [[PVAR:%.*]] = alloca i32*, align 8
347 // CHECK1-NEXT:    [[LVAR:%.*]] = alloca i32*, align 8
348 // CHECK1-NEXT:    call void @_ZN1SIiEC1Ev(%struct.S.0* nonnull align 4 dereferenceable(4) [[TEST]])
349 // CHECK1-NEXT:    call void @_ZN3SSTIiEC1Ev(%struct.SST* nonnull align 4 dereferenceable(4) [[SST]])
350 // CHECK1-NEXT:    [[F:%.*]] = getelementptr inbounds [[STRUCT_S_0]], %struct.S.0* [[TEST]], i32 0, i32 0
351 // CHECK1-NEXT:    store i32* [[F]], i32** [[PVAR]], align 8
352 // CHECK1-NEXT:    [[F1:%.*]] = getelementptr inbounds [[STRUCT_S_0]], %struct.S.0* [[TEST]], i32 0, i32 0
353 // CHECK1-NEXT:    store i32* [[F1]], i32** [[LVAR]], align 8
354 // CHECK1-NEXT:    [[TMP0:%.*]] = load i32*, i32** [[LVAR]], align 8
355 // CHECK1-NEXT:    call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB3]], i32 2, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32**, i32*)* @.omp_outlined..2 to void (i32*, i32*, ...)*), i32** [[PVAR]], i32* [[TMP0]])
356 // CHECK1-NEXT:    call void @_ZN1SIiED1Ev(%struct.S.0* nonnull align 4 dereferenceable(4) [[TEST]]) #[[ATTR4]]
357 // CHECK1-NEXT:    ret i32 0
358 //
359 //
360 // CHECK1-LABEL: define {{[^@]+}}@_ZN1SIfED1Ev
361 // CHECK1-SAME: (%struct.S* nonnull align 4 dereferenceable(4) [[THIS:%.*]]) unnamed_addr #[[ATTR1]] align 2 {
362 // CHECK1-NEXT:  entry:
363 // CHECK1-NEXT:    [[THIS_ADDR:%.*]] = alloca %struct.S*, align 8
364 // CHECK1-NEXT:    store %struct.S* [[THIS]], %struct.S** [[THIS_ADDR]], align 8
365 // CHECK1-NEXT:    [[THIS1:%.*]] = load %struct.S*, %struct.S** [[THIS_ADDR]], align 8
366 // CHECK1-NEXT:    call void @_ZN1SIfED2Ev(%struct.S* nonnull align 4 dereferenceable(4) [[THIS1]]) #[[ATTR4]]
367 // CHECK1-NEXT:    ret void
368 //
369 //
370 // CHECK1-LABEL: define {{[^@]+}}@_ZN2SSC2ERi
371 // CHECK1-SAME: (%struct.SS* nonnull align 8 dereferenceable(16) [[THIS:%.*]], i32* nonnull align 4 dereferenceable(4) [[D:%.*]]) unnamed_addr #[[ATTR1]] align 2 {
372 // CHECK1-NEXT:  entry:
373 // CHECK1-NEXT:    [[THIS_ADDR:%.*]] = alloca %struct.SS*, align 8
374 // CHECK1-NEXT:    [[D_ADDR:%.*]] = alloca i32*, align 8
375 // CHECK1-NEXT:    store %struct.SS* [[THIS]], %struct.SS** [[THIS_ADDR]], align 8
376 // CHECK1-NEXT:    store i32* [[D]], i32** [[D_ADDR]], align 8
377 // CHECK1-NEXT:    [[THIS1:%.*]] = load %struct.SS*, %struct.SS** [[THIS_ADDR]], align 8
378 // CHECK1-NEXT:    [[A:%.*]] = getelementptr inbounds [[STRUCT_SS:%.*]], %struct.SS* [[THIS1]], i32 0, i32 0
379 // CHECK1-NEXT:    store i32 0, i32* [[A]], align 8
380 // CHECK1-NEXT:    [[B:%.*]] = getelementptr inbounds [[STRUCT_SS]], %struct.SS* [[THIS1]], i32 0, i32 1
381 // CHECK1-NEXT:    [[BF_LOAD:%.*]] = load i8, i8* [[B]], align 4
382 // CHECK1-NEXT:    [[BF_CLEAR:%.*]] = and i8 [[BF_LOAD]], -16
383 // CHECK1-NEXT:    store i8 [[BF_CLEAR]], i8* [[B]], align 4
384 // CHECK1-NEXT:    [[C:%.*]] = getelementptr inbounds [[STRUCT_SS]], %struct.SS* [[THIS1]], i32 0, i32 2
385 // CHECK1-NEXT:    [[TMP0:%.*]] = load i32*, i32** [[D_ADDR]], align 8
386 // CHECK1-NEXT:    store i32* [[TMP0]], i32** [[C]], align 8
387 // 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.SS*)* @.omp_outlined..1 to void (i32*, i32*, ...)*), %struct.SS* [[THIS1]])
388 // CHECK1-NEXT:    ret void
389 //
390 //
391 // CHECK1-LABEL: define {{[^@]+}}@.omp_outlined..1
392 // CHECK1-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], %struct.SS* [[THIS:%.*]]) #[[ATTR2]] {
393 // CHECK1-NEXT:  entry:
394 // CHECK1-NEXT:    [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
395 // CHECK1-NEXT:    [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
396 // CHECK1-NEXT:    [[THIS_ADDR:%.*]] = alloca %struct.SS*, align 8
397 // CHECK1-NEXT:    [[A:%.*]] = alloca i32*, align 8
398 // CHECK1-NEXT:    [[B:%.*]] = alloca i32, align 4
399 // CHECK1-NEXT:    [[C:%.*]] = alloca i32*, align 8
400 // CHECK1-NEXT:    [[TMP:%.*]] = alloca i32*, align 8
401 // CHECK1-NEXT:    [[_TMP3:%.*]] = alloca i32*, align 8
402 // CHECK1-NEXT:    [[DOTOMP_IV:%.*]] = alloca i32, align 4
403 // CHECK1-NEXT:    [[_TMP4:%.*]] = alloca i32, align 4
404 // CHECK1-NEXT:    [[DOTLINEAR_START:%.*]] = alloca i32, align 4
405 // CHECK1-NEXT:    [[DOTLINEAR_START5:%.*]] = alloca i32, align 4
406 // CHECK1-NEXT:    [[DOTLINEAR_START6:%.*]] = alloca i32, align 4
407 // CHECK1-NEXT:    [[DOTOMP_LB:%.*]] = alloca i32, align 4
408 // CHECK1-NEXT:    [[DOTOMP_UB:%.*]] = alloca i32, align 4
409 // CHECK1-NEXT:    [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
410 // CHECK1-NEXT:    [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
411 // CHECK1-NEXT:    [[I:%.*]] = alloca i32, align 4
412 // CHECK1-NEXT:    [[A7:%.*]] = alloca i32, align 4
413 // CHECK1-NEXT:    [[_TMP8:%.*]] = alloca i32*, align 8
414 // CHECK1-NEXT:    [[B9:%.*]] = alloca i32, align 4
415 // CHECK1-NEXT:    [[C10:%.*]] = alloca i32, align 4
416 // CHECK1-NEXT:    [[_TMP11:%.*]] = alloca i32*, align 8
417 // CHECK1-NEXT:    [[_TMP20:%.*]] = alloca i32*, align 8
418 // CHECK1-NEXT:    [[_TMP21:%.*]] = alloca i32*, align 8
419 // CHECK1-NEXT:    store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
420 // CHECK1-NEXT:    store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
421 // CHECK1-NEXT:    store %struct.SS* [[THIS]], %struct.SS** [[THIS_ADDR]], align 8
422 // CHECK1-NEXT:    [[TMP0:%.*]] = load %struct.SS*, %struct.SS** [[THIS_ADDR]], align 8
423 // CHECK1-NEXT:    [[A1:%.*]] = getelementptr inbounds [[STRUCT_SS:%.*]], %struct.SS* [[TMP0]], i32 0, i32 0
424 // CHECK1-NEXT:    store i32* [[A1]], i32** [[A]], align 8
425 // CHECK1-NEXT:    [[C2:%.*]] = getelementptr inbounds [[STRUCT_SS]], %struct.SS* [[TMP0]], i32 0, i32 2
426 // CHECK1-NEXT:    [[TMP1:%.*]] = load i32*, i32** [[C2]], align 8
427 // CHECK1-NEXT:    store i32* [[TMP1]], i32** [[C]], align 8
428 // CHECK1-NEXT:    [[TMP2:%.*]] = load i32*, i32** [[A]], align 8
429 // CHECK1-NEXT:    store i32* [[TMP2]], i32** [[TMP]], align 8
430 // CHECK1-NEXT:    [[TMP3:%.*]] = load i32*, i32** [[C]], align 8
431 // CHECK1-NEXT:    store i32* [[TMP3]], i32** [[_TMP3]], align 8
432 // CHECK1-NEXT:    [[TMP4:%.*]] = load i32*, i32** [[TMP]], align 8
433 // CHECK1-NEXT:    [[TMP5:%.*]] = load i32, i32* [[TMP4]], align 4
434 // CHECK1-NEXT:    store i32 [[TMP5]], i32* [[DOTLINEAR_START]], align 4
435 // CHECK1-NEXT:    [[TMP6:%.*]] = load i32, i32* [[B]], align 4
436 // CHECK1-NEXT:    store i32 [[TMP6]], i32* [[DOTLINEAR_START5]], align 4
437 // CHECK1-NEXT:    [[TMP7:%.*]] = load i32*, i32** [[_TMP3]], align 8
438 // CHECK1-NEXT:    [[TMP8:%.*]] = load i32, i32* [[TMP7]], align 4
439 // CHECK1-NEXT:    store i32 [[TMP8]], i32* [[DOTLINEAR_START6]], align 4
440 // CHECK1-NEXT:    store i32 0, i32* [[DOTOMP_LB]], align 4
441 // CHECK1-NEXT:    store i32 1, i32* [[DOTOMP_UB]], align 4
442 // CHECK1-NEXT:    store i32 1, i32* [[DOTOMP_STRIDE]], align 4
443 // CHECK1-NEXT:    store i32 0, i32* [[DOTOMP_IS_LAST]], align 4
444 // CHECK1-NEXT:    [[TMP9:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
445 // CHECK1-NEXT:    [[TMP10:%.*]] = load i32, i32* [[TMP9]], align 4
446 // CHECK1-NEXT:    call void @__kmpc_barrier(%struct.ident_t* @[[GLOB1]], i32 [[TMP10]])
447 // CHECK1-NEXT:    store i32* [[A7]], i32** [[_TMP8]], align 8
448 // CHECK1-NEXT:    store i32* [[C10]], i32** [[_TMP11]], align 8
449 // CHECK1-NEXT:    call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB2]], i32 [[TMP10]], i32 34, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_LB]], i32* [[DOTOMP_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1)
450 // CHECK1-NEXT:    [[TMP11:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
451 // CHECK1-NEXT:    [[CMP:%.*]] = icmp sgt i32 [[TMP11]], 1
452 // CHECK1-NEXT:    br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
453 // CHECK1:       cond.true:
454 // CHECK1-NEXT:    br label [[COND_END:%.*]]
455 // CHECK1:       cond.false:
456 // CHECK1-NEXT:    [[TMP12:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
457 // CHECK1-NEXT:    br label [[COND_END]]
458 // CHECK1:       cond.end:
459 // CHECK1-NEXT:    [[COND:%.*]] = phi i32 [ 1, [[COND_TRUE]] ], [ [[TMP12]], [[COND_FALSE]] ]
460 // CHECK1-NEXT:    store i32 [[COND]], i32* [[DOTOMP_UB]], align 4
461 // CHECK1-NEXT:    [[TMP13:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4
462 // CHECK1-NEXT:    store i32 [[TMP13]], i32* [[DOTOMP_IV]], align 4
463 // CHECK1-NEXT:    br label [[OMP_INNER_FOR_COND:%.*]]
464 // CHECK1:       omp.inner.for.cond:
465 // CHECK1-NEXT:    [[TMP14:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
466 // CHECK1-NEXT:    [[TMP15:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
467 // CHECK1-NEXT:    [[CMP12:%.*]] = icmp sle i32 [[TMP14]], [[TMP15]]
468 // CHECK1-NEXT:    br i1 [[CMP12]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
469 // CHECK1:       omp.inner.for.body:
470 // CHECK1-NEXT:    [[TMP16:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
471 // CHECK1-NEXT:    [[MUL:%.*]] = mul nsw i32 [[TMP16]], 1
472 // CHECK1-NEXT:    [[ADD:%.*]] = add nsw i32 0, [[MUL]]
473 // CHECK1-NEXT:    store i32 [[ADD]], i32* [[I]], align 4
474 // CHECK1-NEXT:    [[TMP17:%.*]] = load i32, i32* [[DOTLINEAR_START]], align 4
475 // CHECK1-NEXT:    [[TMP18:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
476 // CHECK1-NEXT:    [[MUL13:%.*]] = mul nsw i32 [[TMP18]], 1
477 // CHECK1-NEXT:    [[ADD14:%.*]] = add nsw i32 [[TMP17]], [[MUL13]]
478 // CHECK1-NEXT:    store i32 [[ADD14]], i32* [[A7]], align 4
479 // CHECK1-NEXT:    [[TMP19:%.*]] = load i32, i32* [[DOTLINEAR_START5]], align 4
480 // CHECK1-NEXT:    [[TMP20:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
481 // CHECK1-NEXT:    [[MUL15:%.*]] = mul nsw i32 [[TMP20]], 1
482 // CHECK1-NEXT:    [[ADD16:%.*]] = add nsw i32 [[TMP19]], [[MUL15]]
483 // CHECK1-NEXT:    store i32 [[ADD16]], i32* [[B9]], align 4
484 // CHECK1-NEXT:    [[TMP21:%.*]] = load i32, i32* [[DOTLINEAR_START6]], align 4
485 // CHECK1-NEXT:    [[TMP22:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
486 // CHECK1-NEXT:    [[MUL17:%.*]] = mul nsw i32 [[TMP22]], 1
487 // CHECK1-NEXT:    [[ADD18:%.*]] = add nsw i32 [[TMP21]], [[MUL17]]
488 // CHECK1-NEXT:    store i32 [[ADD18]], i32* [[C10]], align 4
489 // CHECK1-NEXT:    [[TMP23:%.*]] = load i32*, i32** [[_TMP8]], align 8
490 // CHECK1-NEXT:    [[TMP24:%.*]] = load i32, i32* [[TMP23]], align 4
491 // CHECK1-NEXT:    [[INC:%.*]] = add nsw i32 [[TMP24]], 1
492 // CHECK1-NEXT:    store i32 [[INC]], i32* [[TMP23]], align 4
493 // CHECK1-NEXT:    [[TMP25:%.*]] = load i32, i32* [[B9]], align 4
494 // CHECK1-NEXT:    [[DEC:%.*]] = add nsw i32 [[TMP25]], -1
495 // CHECK1-NEXT:    store i32 [[DEC]], i32* [[B9]], align 4
496 // CHECK1-NEXT:    [[TMP26:%.*]] = load i32*, i32** [[_TMP11]], align 8
497 // CHECK1-NEXT:    [[TMP27:%.*]] = load i32, i32* [[TMP26]], align 4
498 // CHECK1-NEXT:    [[DIV:%.*]] = sdiv i32 [[TMP27]], 1
499 // CHECK1-NEXT:    store i32 [[DIV]], i32* [[TMP26]], align 4
500 // CHECK1-NEXT:    br label [[OMP_BODY_CONTINUE:%.*]]
501 // CHECK1:       omp.body.continue:
502 // CHECK1-NEXT:    br label [[OMP_INNER_FOR_INC:%.*]]
503 // CHECK1:       omp.inner.for.inc:
504 // CHECK1-NEXT:    [[TMP28:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
505 // CHECK1-NEXT:    [[ADD19:%.*]] = add nsw i32 [[TMP28]], 1
506 // CHECK1-NEXT:    store i32 [[ADD19]], i32* [[DOTOMP_IV]], align 4
507 // CHECK1-NEXT:    br label [[OMP_INNER_FOR_COND]]
508 // CHECK1:       omp.inner.for.end:
509 // CHECK1-NEXT:    br label [[OMP_LOOP_EXIT:%.*]]
510 // CHECK1:       omp.loop.exit:
511 // CHECK1-NEXT:    call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB2]], i32 [[TMP10]])
512 // CHECK1-NEXT:    [[TMP29:%.*]] = load i32, i32* [[DOTOMP_IS_LAST]], align 4
513 // CHECK1-NEXT:    [[TMP30:%.*]] = icmp ne i32 [[TMP29]], 0
514 // CHECK1-NEXT:    br i1 [[TMP30]], label [[DOTOMP_LINEAR_PU:%.*]], label [[DOTOMP_LINEAR_PU_DONE:%.*]]
515 // CHECK1:       .omp.linear.pu:
516 // CHECK1-NEXT:    [[TMP31:%.*]] = load i32*, i32** [[TMP]], align 8
517 // CHECK1-NEXT:    store i32* [[TMP31]], i32** [[_TMP20]], align 8
518 // CHECK1-NEXT:    [[TMP32:%.*]] = load i32, i32* [[A7]], align 4
519 // CHECK1-NEXT:    [[TMP33:%.*]] = load i32*, i32** [[_TMP20]], align 8
520 // CHECK1-NEXT:    store i32 [[TMP32]], i32* [[TMP33]], align 4
521 // CHECK1-NEXT:    [[TMP34:%.*]] = load i32, i32* [[B9]], align 4
522 // CHECK1-NEXT:    store i32 [[TMP34]], i32* [[B]], align 4
523 // CHECK1-NEXT:    [[TMP35:%.*]] = load i32*, i32** [[_TMP3]], align 8
524 // CHECK1-NEXT:    store i32* [[TMP35]], i32** [[_TMP21]], align 8
525 // CHECK1-NEXT:    [[TMP36:%.*]] = load i32, i32* [[C10]], align 4
526 // CHECK1-NEXT:    [[TMP37:%.*]] = load i32*, i32** [[_TMP21]], align 8
527 // CHECK1-NEXT:    store i32 [[TMP36]], i32* [[TMP37]], align 4
528 // CHECK1-NEXT:    [[TMP38:%.*]] = load i32, i32* [[B]], align 4
529 // CHECK1-NEXT:    [[B22:%.*]] = getelementptr inbounds [[STRUCT_SS]], %struct.SS* [[TMP0]], i32 0, i32 1
530 // CHECK1-NEXT:    [[TMP39:%.*]] = trunc i32 [[TMP38]] to i8
531 // CHECK1-NEXT:    [[BF_LOAD:%.*]] = load i8, i8* [[B22]], align 4
532 // CHECK1-NEXT:    [[BF_VALUE:%.*]] = and i8 [[TMP39]], 15
533 // CHECK1-NEXT:    [[BF_CLEAR:%.*]] = and i8 [[BF_LOAD]], -16
534 // CHECK1-NEXT:    [[BF_SET:%.*]] = or i8 [[BF_CLEAR]], [[BF_VALUE]]
535 // CHECK1-NEXT:    store i8 [[BF_SET]], i8* [[B22]], align 4
536 // CHECK1-NEXT:    br label [[DOTOMP_LINEAR_PU_DONE]]
537 // CHECK1:       .omp.linear.pu.done:
538 // CHECK1-NEXT:    call void @__kmpc_barrier(%struct.ident_t* @[[GLOB1]], i32 [[TMP10]])
539 // CHECK1-NEXT:    ret void
540 //
541 //
542 // CHECK1-LABEL: define {{[^@]+}}@_ZN1SIfEC2Ev
543 // CHECK1-SAME: (%struct.S* nonnull align 4 dereferenceable(4) [[THIS:%.*]]) unnamed_addr #[[ATTR1]] align 2 {
544 // CHECK1-NEXT:  entry:
545 // CHECK1-NEXT:    [[THIS_ADDR:%.*]] = alloca %struct.S*, align 8
546 // CHECK1-NEXT:    store %struct.S* [[THIS]], %struct.S** [[THIS_ADDR]], align 8
547 // CHECK1-NEXT:    [[THIS1:%.*]] = load %struct.S*, %struct.S** [[THIS_ADDR]], align 8
548 // CHECK1-NEXT:    [[F:%.*]] = getelementptr inbounds [[STRUCT_S:%.*]], %struct.S* [[THIS1]], i32 0, i32 0
549 // CHECK1-NEXT:    store float 0.000000e+00, float* [[F]], align 4
550 // CHECK1-NEXT:    ret void
551 //
552 //
553 // CHECK1-LABEL: define {{[^@]+}}@_ZN1SIiEC1Ev
554 // CHECK1-SAME: (%struct.S.0* nonnull align 4 dereferenceable(4) [[THIS:%.*]]) unnamed_addr #[[ATTR1]] align 2 {
555 // CHECK1-NEXT:  entry:
556 // CHECK1-NEXT:    [[THIS_ADDR:%.*]] = alloca %struct.S.0*, align 8
557 // CHECK1-NEXT:    store %struct.S.0* [[THIS]], %struct.S.0** [[THIS_ADDR]], align 8
558 // CHECK1-NEXT:    [[THIS1:%.*]] = load %struct.S.0*, %struct.S.0** [[THIS_ADDR]], align 8
559 // CHECK1-NEXT:    call void @_ZN1SIiEC2Ev(%struct.S.0* nonnull align 4 dereferenceable(4) [[THIS1]])
560 // CHECK1-NEXT:    ret void
561 //
562 //
563 // CHECK1-LABEL: define {{[^@]+}}@_ZN3SSTIiEC1Ev
564 // CHECK1-SAME: (%struct.SST* nonnull align 4 dereferenceable(4) [[THIS:%.*]]) unnamed_addr #[[ATTR1]] align 2 {
565 // CHECK1-NEXT:  entry:
566 // CHECK1-NEXT:    [[THIS_ADDR:%.*]] = alloca %struct.SST*, align 8
567 // CHECK1-NEXT:    store %struct.SST* [[THIS]], %struct.SST** [[THIS_ADDR]], align 8
568 // CHECK1-NEXT:    [[THIS1:%.*]] = load %struct.SST*, %struct.SST** [[THIS_ADDR]], align 8
569 // CHECK1-NEXT:    call void @_ZN3SSTIiEC2Ev(%struct.SST* nonnull align 4 dereferenceable(4) [[THIS1]])
570 // CHECK1-NEXT:    ret void
571 //
572 //
573 // CHECK1-LABEL: define {{[^@]+}}@.omp_outlined..2
574 // CHECK1-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i32** nonnull align 8 dereferenceable(8) [[PVAR:%.*]], i32* nonnull align 4 dereferenceable(4) [[LVAR:%.*]]) #[[ATTR2]] {
575 // CHECK1-NEXT:  entry:
576 // CHECK1-NEXT:    [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
577 // CHECK1-NEXT:    [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
578 // CHECK1-NEXT:    [[PVAR_ADDR:%.*]] = alloca i32**, align 8
579 // CHECK1-NEXT:    [[LVAR_ADDR:%.*]] = alloca i32*, align 8
580 // CHECK1-NEXT:    [[TMP:%.*]] = alloca i32*, align 8
581 // CHECK1-NEXT:    [[_TMP1:%.*]] = alloca i32*, align 8
582 // CHECK1-NEXT:    [[DOTOMP_IV:%.*]] = alloca i32, align 4
583 // CHECK1-NEXT:    [[_TMP2:%.*]] = alloca i32, align 4
584 // CHECK1-NEXT:    [[DOTLINEAR_START:%.*]] = alloca i32*, align 8
585 // CHECK1-NEXT:    [[DOTLINEAR_START3:%.*]] = alloca i32, align 4
586 // CHECK1-NEXT:    [[DOTOMP_LB:%.*]] = alloca i32, align 4
587 // CHECK1-NEXT:    [[DOTOMP_UB:%.*]] = alloca i32, align 4
588 // CHECK1-NEXT:    [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
589 // CHECK1-NEXT:    [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
590 // CHECK1-NEXT:    [[I:%.*]] = alloca i32, align 4
591 // CHECK1-NEXT:    [[PVAR4:%.*]] = alloca i32*, align 8
592 // CHECK1-NEXT:    [[LVAR5:%.*]] = alloca i32, align 4
593 // CHECK1-NEXT:    [[_TMP6:%.*]] = alloca i32*, align 8
594 // CHECK1-NEXT:    [[_TMP12:%.*]] = alloca i32*, align 8
595 // CHECK1-NEXT:    store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
596 // CHECK1-NEXT:    store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
597 // CHECK1-NEXT:    store i32** [[PVAR]], i32*** [[PVAR_ADDR]], align 8
598 // CHECK1-NEXT:    store i32* [[LVAR]], i32** [[LVAR_ADDR]], align 8
599 // CHECK1-NEXT:    [[TMP0:%.*]] = load i32**, i32*** [[PVAR_ADDR]], align 8
600 // CHECK1-NEXT:    [[TMP1:%.*]] = load i32*, i32** [[LVAR_ADDR]], align 8
601 // CHECK1-NEXT:    store i32* [[TMP1]], i32** [[TMP]], align 8
602 // CHECK1-NEXT:    [[TMP2:%.*]] = load i32*, i32** [[TMP]], align 8
603 // CHECK1-NEXT:    store i32* [[TMP2]], i32** [[_TMP1]], align 8
604 // CHECK1-NEXT:    [[TMP3:%.*]] = load i32*, i32** [[TMP0]], align 8
605 // CHECK1-NEXT:    store i32* [[TMP3]], i32** [[DOTLINEAR_START]], align 8
606 // CHECK1-NEXT:    [[TMP4:%.*]] = load i32*, i32** [[_TMP1]], align 8
607 // CHECK1-NEXT:    [[TMP5:%.*]] = load i32, i32* [[TMP4]], align 4
608 // CHECK1-NEXT:    store i32 [[TMP5]], i32* [[DOTLINEAR_START3]], align 4
609 // CHECK1-NEXT:    store i32 0, i32* [[DOTOMP_LB]], align 4
610 // CHECK1-NEXT:    store i32 1, i32* [[DOTOMP_UB]], align 4
611 // CHECK1-NEXT:    store i32 1, i32* [[DOTOMP_STRIDE]], align 4
612 // CHECK1-NEXT:    store i32 0, i32* [[DOTOMP_IS_LAST]], align 4
613 // CHECK1-NEXT:    [[TMP6:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
614 // CHECK1-NEXT:    [[TMP7:%.*]] = load i32, i32* [[TMP6]], align 4
615 // CHECK1-NEXT:    call void @__kmpc_barrier(%struct.ident_t* @[[GLOB1]], i32 [[TMP7]])
616 // CHECK1-NEXT:    store i32* [[LVAR5]], i32** [[_TMP6]], align 8
617 // CHECK1-NEXT:    call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB2]], i32 [[TMP7]], i32 34, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_LB]], i32* [[DOTOMP_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1)
618 // CHECK1-NEXT:    [[TMP8:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
619 // CHECK1-NEXT:    [[CMP:%.*]] = icmp sgt i32 [[TMP8]], 1
620 // CHECK1-NEXT:    br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
621 // CHECK1:       cond.true:
622 // CHECK1-NEXT:    br label [[COND_END:%.*]]
623 // CHECK1:       cond.false:
624 // CHECK1-NEXT:    [[TMP9:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
625 // CHECK1-NEXT:    br label [[COND_END]]
626 // CHECK1:       cond.end:
627 // CHECK1-NEXT:    [[COND:%.*]] = phi i32 [ 1, [[COND_TRUE]] ], [ [[TMP9]], [[COND_FALSE]] ]
628 // CHECK1-NEXT:    store i32 [[COND]], i32* [[DOTOMP_UB]], align 4
629 // CHECK1-NEXT:    [[TMP10:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4
630 // CHECK1-NEXT:    store i32 [[TMP10]], i32* [[DOTOMP_IV]], align 4
631 // CHECK1-NEXT:    br label [[OMP_INNER_FOR_COND:%.*]]
632 // CHECK1:       omp.inner.for.cond:
633 // CHECK1-NEXT:    [[TMP11:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
634 // CHECK1-NEXT:    [[TMP12:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
635 // CHECK1-NEXT:    [[CMP7:%.*]] = icmp sle i32 [[TMP11]], [[TMP12]]
636 // CHECK1-NEXT:    br i1 [[CMP7]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
637 // CHECK1:       omp.inner.for.body:
638 // CHECK1-NEXT:    [[TMP13:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
639 // CHECK1-NEXT:    [[MUL:%.*]] = mul nsw i32 [[TMP13]], 1
640 // CHECK1-NEXT:    [[ADD:%.*]] = add nsw i32 0, [[MUL]]
641 // CHECK1-NEXT:    store i32 [[ADD]], i32* [[I]], align 4
642 // CHECK1-NEXT:    [[TMP14:%.*]] = load i32*, i32** [[DOTLINEAR_START]], align 8
643 // CHECK1-NEXT:    [[TMP15:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
644 // CHECK1-NEXT:    [[MUL8:%.*]] = mul nsw i32 [[TMP15]], 1
645 // CHECK1-NEXT:    [[IDX_EXT:%.*]] = sext i32 [[MUL8]] to i64
646 // CHECK1-NEXT:    [[ADD_PTR:%.*]] = getelementptr inbounds i32, i32* [[TMP14]], i64 [[IDX_EXT]]
647 // CHECK1-NEXT:    store i32* [[ADD_PTR]], i32** [[PVAR4]], align 8
648 // CHECK1-NEXT:    [[TMP16:%.*]] = load i32, i32* [[DOTLINEAR_START3]], align 4
649 // CHECK1-NEXT:    [[TMP17:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
650 // CHECK1-NEXT:    [[MUL9:%.*]] = mul nsw i32 [[TMP17]], 1
651 // CHECK1-NEXT:    [[ADD10:%.*]] = add nsw i32 [[TMP16]], [[MUL9]]
652 // CHECK1-NEXT:    store i32 [[ADD10]], i32* [[LVAR5]], align 4
653 // CHECK1-NEXT:    [[TMP18:%.*]] = load i32*, i32** [[PVAR4]], align 8
654 // CHECK1-NEXT:    [[INCDEC_PTR:%.*]] = getelementptr inbounds i32, i32* [[TMP18]], i32 1
655 // CHECK1-NEXT:    store i32* [[INCDEC_PTR]], i32** [[PVAR4]], align 8
656 // CHECK1-NEXT:    [[TMP19:%.*]] = load i32*, i32** [[_TMP6]], align 8
657 // CHECK1-NEXT:    [[TMP20:%.*]] = load i32, i32* [[TMP19]], align 4
658 // CHECK1-NEXT:    [[INC:%.*]] = add nsw i32 [[TMP20]], 1
659 // CHECK1-NEXT:    store i32 [[INC]], i32* [[TMP19]], align 4
660 // CHECK1-NEXT:    br label [[OMP_BODY_CONTINUE:%.*]]
661 // CHECK1:       omp.body.continue:
662 // CHECK1-NEXT:    br label [[OMP_INNER_FOR_INC:%.*]]
663 // CHECK1:       omp.inner.for.inc:
664 // CHECK1-NEXT:    [[TMP21:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
665 // CHECK1-NEXT:    [[ADD11:%.*]] = add nsw i32 [[TMP21]], 1
666 // CHECK1-NEXT:    store i32 [[ADD11]], i32* [[DOTOMP_IV]], align 4
667 // CHECK1-NEXT:    br label [[OMP_INNER_FOR_COND]]
668 // CHECK1:       omp.inner.for.end:
669 // CHECK1-NEXT:    br label [[OMP_LOOP_EXIT:%.*]]
670 // CHECK1:       omp.loop.exit:
671 // CHECK1-NEXT:    call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB2]], i32 [[TMP7]])
672 // CHECK1-NEXT:    [[TMP22:%.*]] = load i32, i32* [[DOTOMP_IS_LAST]], align 4
673 // CHECK1-NEXT:    [[TMP23:%.*]] = icmp ne i32 [[TMP22]], 0
674 // CHECK1-NEXT:    br i1 [[TMP23]], label [[DOTOMP_LINEAR_PU:%.*]], label [[DOTOMP_LINEAR_PU_DONE:%.*]]
675 // CHECK1:       .omp.linear.pu:
676 // CHECK1-NEXT:    [[TMP24:%.*]] = load i32*, i32** [[PVAR4]], align 8
677 // CHECK1-NEXT:    store i32* [[TMP24]], i32** [[TMP0]], align 8
678 // CHECK1-NEXT:    [[TMP25:%.*]] = load i32*, i32** [[_TMP1]], align 8
679 // CHECK1-NEXT:    store i32* [[TMP25]], i32** [[_TMP12]], align 8
680 // CHECK1-NEXT:    [[TMP26:%.*]] = load i32, i32* [[LVAR5]], align 4
681 // CHECK1-NEXT:    [[TMP27:%.*]] = load i32*, i32** [[_TMP12]], align 8
682 // CHECK1-NEXT:    store i32 [[TMP26]], i32* [[TMP27]], align 4
683 // CHECK1-NEXT:    br label [[DOTOMP_LINEAR_PU_DONE]]
684 // CHECK1:       .omp.linear.pu.done:
685 // CHECK1-NEXT:    call void @__kmpc_barrier(%struct.ident_t* @[[GLOB1]], i32 [[TMP7]])
686 // CHECK1-NEXT:    ret void
687 //
688 //
689 // CHECK1-LABEL: define {{[^@]+}}@_ZN1SIiED1Ev
690 // CHECK1-SAME: (%struct.S.0* nonnull align 4 dereferenceable(4) [[THIS:%.*]]) unnamed_addr #[[ATTR1]] align 2 {
691 // CHECK1-NEXT:  entry:
692 // CHECK1-NEXT:    [[THIS_ADDR:%.*]] = alloca %struct.S.0*, align 8
693 // CHECK1-NEXT:    store %struct.S.0* [[THIS]], %struct.S.0** [[THIS_ADDR]], align 8
694 // CHECK1-NEXT:    [[THIS1:%.*]] = load %struct.S.0*, %struct.S.0** [[THIS_ADDR]], align 8
695 // CHECK1-NEXT:    call void @_ZN1SIiED2Ev(%struct.S.0* nonnull align 4 dereferenceable(4) [[THIS1]]) #[[ATTR4]]
696 // CHECK1-NEXT:    ret void
697 //
698 //
699 // CHECK1-LABEL: define {{[^@]+}}@_ZN1SIiEC2Ev
700 // CHECK1-SAME: (%struct.S.0* nonnull align 4 dereferenceable(4) [[THIS:%.*]]) unnamed_addr #[[ATTR1]] align 2 {
701 // CHECK1-NEXT:  entry:
702 // CHECK1-NEXT:    [[THIS_ADDR:%.*]] = alloca %struct.S.0*, align 8
703 // CHECK1-NEXT:    store %struct.S.0* [[THIS]], %struct.S.0** [[THIS_ADDR]], align 8
704 // CHECK1-NEXT:    [[THIS1:%.*]] = load %struct.S.0*, %struct.S.0** [[THIS_ADDR]], align 8
705 // CHECK1-NEXT:    [[F:%.*]] = getelementptr inbounds [[STRUCT_S_0:%.*]], %struct.S.0* [[THIS1]], i32 0, i32 0
706 // CHECK1-NEXT:    store i32 0, i32* [[F]], align 4
707 // CHECK1-NEXT:    ret void
708 //
709 //
710 // CHECK1-LABEL: define {{[^@]+}}@_ZN3SSTIiEC2Ev
711 // CHECK1-SAME: (%struct.SST* nonnull align 4 dereferenceable(4) [[THIS:%.*]]) unnamed_addr #[[ATTR1]] align 2 {
712 // CHECK1-NEXT:  entry:
713 // CHECK1-NEXT:    [[THIS_ADDR:%.*]] = alloca %struct.SST*, align 8
714 // CHECK1-NEXT:    store %struct.SST* [[THIS]], %struct.SST** [[THIS_ADDR]], align 8
715 // CHECK1-NEXT:    [[THIS1:%.*]] = load %struct.SST*, %struct.SST** [[THIS_ADDR]], align 8
716 // CHECK1-NEXT:    [[A:%.*]] = getelementptr inbounds [[STRUCT_SST:%.*]], %struct.SST* [[THIS1]], i32 0, i32 0
717 // CHECK1-NEXT:    store i32 0, i32* [[A]], align 4
718 // 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.SST*)* @.omp_outlined..3 to void (i32*, i32*, ...)*), %struct.SST* [[THIS1]])
719 // CHECK1-NEXT:    ret void
720 //
721 //
722 // CHECK1-LABEL: define {{[^@]+}}@.omp_outlined..3
723 // CHECK1-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], %struct.SST* [[THIS:%.*]]) #[[ATTR2]] {
724 // CHECK1-NEXT:  entry:
725 // CHECK1-NEXT:    [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
726 // CHECK1-NEXT:    [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
727 // CHECK1-NEXT:    [[THIS_ADDR:%.*]] = alloca %struct.SST*, align 8
728 // CHECK1-NEXT:    [[A:%.*]] = alloca i32*, align 8
729 // CHECK1-NEXT:    [[TMP:%.*]] = alloca i32*, align 8
730 // CHECK1-NEXT:    [[DOTOMP_IV:%.*]] = alloca i32, align 4
731 // CHECK1-NEXT:    [[_TMP2:%.*]] = alloca i32, align 4
732 // CHECK1-NEXT:    [[DOTLINEAR_START:%.*]] = alloca i32, align 4
733 // CHECK1-NEXT:    [[DOTOMP_LB:%.*]] = alloca i32, align 4
734 // CHECK1-NEXT:    [[DOTOMP_UB:%.*]] = alloca i32, align 4
735 // CHECK1-NEXT:    [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
736 // CHECK1-NEXT:    [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
737 // CHECK1-NEXT:    [[I:%.*]] = alloca i32, align 4
738 // CHECK1-NEXT:    [[A3:%.*]] = alloca i32, align 4
739 // CHECK1-NEXT:    [[_TMP4:%.*]] = alloca i32*, align 8
740 // CHECK1-NEXT:    [[_TMP9:%.*]] = alloca i32*, align 8
741 // CHECK1-NEXT:    store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
742 // CHECK1-NEXT:    store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
743 // CHECK1-NEXT:    store %struct.SST* [[THIS]], %struct.SST** [[THIS_ADDR]], align 8
744 // CHECK1-NEXT:    [[TMP0:%.*]] = load %struct.SST*, %struct.SST** [[THIS_ADDR]], align 8
745 // CHECK1-NEXT:    [[A1:%.*]] = getelementptr inbounds [[STRUCT_SST:%.*]], %struct.SST* [[TMP0]], i32 0, i32 0
746 // CHECK1-NEXT:    store i32* [[A1]], i32** [[A]], align 8
747 // CHECK1-NEXT:    [[TMP1:%.*]] = load i32*, i32** [[A]], align 8
748 // CHECK1-NEXT:    store i32* [[TMP1]], i32** [[TMP]], align 8
749 // CHECK1-NEXT:    [[TMP2:%.*]] = load i32*, i32** [[TMP]], align 8
750 // CHECK1-NEXT:    [[TMP3:%.*]] = load i32, i32* [[TMP2]], align 4
751 // CHECK1-NEXT:    store i32 [[TMP3]], i32* [[DOTLINEAR_START]], align 4
752 // CHECK1-NEXT:    store i32 0, i32* [[DOTOMP_LB]], align 4
753 // CHECK1-NEXT:    store i32 1, i32* [[DOTOMP_UB]], align 4
754 // CHECK1-NEXT:    store i32 1, i32* [[DOTOMP_STRIDE]], align 4
755 // CHECK1-NEXT:    store i32 0, i32* [[DOTOMP_IS_LAST]], align 4
756 // CHECK1-NEXT:    [[TMP4:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
757 // CHECK1-NEXT:    [[TMP5:%.*]] = load i32, i32* [[TMP4]], align 4
758 // CHECK1-NEXT:    call void @__kmpc_barrier(%struct.ident_t* @[[GLOB1]], i32 [[TMP5]])
759 // CHECK1-NEXT:    store i32* [[A3]], i32** [[_TMP4]], align 8
760 // CHECK1-NEXT:    call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB2]], i32 [[TMP5]], i32 34, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_LB]], i32* [[DOTOMP_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1)
761 // CHECK1-NEXT:    [[TMP6:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
762 // CHECK1-NEXT:    [[CMP:%.*]] = icmp sgt i32 [[TMP6]], 1
763 // CHECK1-NEXT:    br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
764 // CHECK1:       cond.true:
765 // CHECK1-NEXT:    br label [[COND_END:%.*]]
766 // CHECK1:       cond.false:
767 // CHECK1-NEXT:    [[TMP7:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
768 // CHECK1-NEXT:    br label [[COND_END]]
769 // CHECK1:       cond.end:
770 // CHECK1-NEXT:    [[COND:%.*]] = phi i32 [ 1, [[COND_TRUE]] ], [ [[TMP7]], [[COND_FALSE]] ]
771 // CHECK1-NEXT:    store i32 [[COND]], i32* [[DOTOMP_UB]], align 4
772 // CHECK1-NEXT:    [[TMP8:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4
773 // CHECK1-NEXT:    store i32 [[TMP8]], i32* [[DOTOMP_IV]], align 4
774 // CHECK1-NEXT:    br label [[OMP_INNER_FOR_COND:%.*]]
775 // CHECK1:       omp.inner.for.cond:
776 // CHECK1-NEXT:    [[TMP9:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
777 // CHECK1-NEXT:    [[TMP10:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
778 // CHECK1-NEXT:    [[CMP5:%.*]] = icmp sle i32 [[TMP9]], [[TMP10]]
779 // CHECK1-NEXT:    br i1 [[CMP5]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
780 // CHECK1:       omp.inner.for.body:
781 // CHECK1-NEXT:    [[TMP11:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
782 // CHECK1-NEXT:    [[MUL:%.*]] = mul nsw i32 [[TMP11]], 1
783 // CHECK1-NEXT:    [[ADD:%.*]] = add nsw i32 0, [[MUL]]
784 // CHECK1-NEXT:    store i32 [[ADD]], i32* [[I]], align 4
785 // CHECK1-NEXT:    [[TMP12:%.*]] = load i32, i32* [[DOTLINEAR_START]], align 4
786 // CHECK1-NEXT:    [[TMP13:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
787 // CHECK1-NEXT:    [[MUL6:%.*]] = mul nsw i32 [[TMP13]], 1
788 // CHECK1-NEXT:    [[ADD7:%.*]] = add nsw i32 [[TMP12]], [[MUL6]]
789 // CHECK1-NEXT:    store i32 [[ADD7]], i32* [[A3]], align 4
790 // CHECK1-NEXT:    [[TMP14:%.*]] = load i32*, i32** [[_TMP4]], align 8
791 // CHECK1-NEXT:    [[TMP15:%.*]] = load i32, i32* [[TMP14]], align 4
792 // CHECK1-NEXT:    [[INC:%.*]] = add nsw i32 [[TMP15]], 1
793 // CHECK1-NEXT:    store i32 [[INC]], i32* [[TMP14]], align 4
794 // CHECK1-NEXT:    br label [[OMP_BODY_CONTINUE:%.*]]
795 // CHECK1:       omp.body.continue:
796 // CHECK1-NEXT:    br label [[OMP_INNER_FOR_INC:%.*]]
797 // CHECK1:       omp.inner.for.inc:
798 // CHECK1-NEXT:    [[TMP16:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
799 // CHECK1-NEXT:    [[ADD8:%.*]] = add nsw i32 [[TMP16]], 1
800 // CHECK1-NEXT:    store i32 [[ADD8]], i32* [[DOTOMP_IV]], align 4
801 // CHECK1-NEXT:    br label [[OMP_INNER_FOR_COND]]
802 // CHECK1:       omp.inner.for.end:
803 // CHECK1-NEXT:    br label [[OMP_LOOP_EXIT:%.*]]
804 // CHECK1:       omp.loop.exit:
805 // CHECK1-NEXT:    call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB2]], i32 [[TMP5]])
806 // CHECK1-NEXT:    [[TMP17:%.*]] = load i32, i32* [[DOTOMP_IS_LAST]], align 4
807 // CHECK1-NEXT:    [[TMP18:%.*]] = icmp ne i32 [[TMP17]], 0
808 // CHECK1-NEXT:    br i1 [[TMP18]], label [[DOTOMP_LINEAR_PU:%.*]], label [[DOTOMP_LINEAR_PU_DONE:%.*]]
809 // CHECK1:       .omp.linear.pu:
810 // CHECK1-NEXT:    [[TMP19:%.*]] = load i32*, i32** [[TMP]], align 8
811 // CHECK1-NEXT:    store i32* [[TMP19]], i32** [[_TMP9]], align 8
812 // CHECK1-NEXT:    [[TMP20:%.*]] = load i32, i32* [[A3]], align 4
813 // CHECK1-NEXT:    [[TMP21:%.*]] = load i32*, i32** [[_TMP9]], align 8
814 // CHECK1-NEXT:    store i32 [[TMP20]], i32* [[TMP21]], align 4
815 // CHECK1-NEXT:    br label [[DOTOMP_LINEAR_PU_DONE]]
816 // CHECK1:       .omp.linear.pu.done:
817 // CHECK1-NEXT:    call void @__kmpc_barrier(%struct.ident_t* @[[GLOB1]], i32 [[TMP5]])
818 // CHECK1-NEXT:    ret void
819 //
820 //
821 // CHECK1-LABEL: define {{[^@]+}}@_ZN1SIiED2Ev
822 // CHECK1-SAME: (%struct.S.0* nonnull align 4 dereferenceable(4) [[THIS:%.*]]) unnamed_addr #[[ATTR1]] align 2 {
823 // CHECK1-NEXT:  entry:
824 // CHECK1-NEXT:    [[THIS_ADDR:%.*]] = alloca %struct.S.0*, align 8
825 // CHECK1-NEXT:    store %struct.S.0* [[THIS]], %struct.S.0** [[THIS_ADDR]], align 8
826 // CHECK1-NEXT:    [[THIS1:%.*]] = load %struct.S.0*, %struct.S.0** [[THIS_ADDR]], align 8
827 // CHECK1-NEXT:    ret void
828 //
829 //
830 // CHECK1-LABEL: define {{[^@]+}}@_ZN1SIfED2Ev
831 // CHECK1-SAME: (%struct.S* nonnull align 4 dereferenceable(4) [[THIS:%.*]]) unnamed_addr #[[ATTR1]] align 2 {
832 // CHECK1-NEXT:  entry:
833 // CHECK1-NEXT:    [[THIS_ADDR:%.*]] = alloca %struct.S*, align 8
834 // CHECK1-NEXT:    store %struct.S* [[THIS]], %struct.S** [[THIS_ADDR]], align 8
835 // CHECK1-NEXT:    [[THIS1:%.*]] = load %struct.S*, %struct.S** [[THIS_ADDR]], align 8
836 // CHECK1-NEXT:    ret void
837 //
838 //
839 // CHECK2-LABEL: define {{[^@]+}}@main
840 // CHECK2-SAME: () #[[ATTR0:[0-9]+]] {
841 // CHECK2-NEXT:  entry:
842 // CHECK2-NEXT:    [[RETVAL:%.*]] = alloca i32, align 4
843 // CHECK2-NEXT:    [[SS:%.*]] = alloca [[STRUCT_SS:%.*]], align 8
844 // CHECK2-NEXT:    [[TEST:%.*]] = alloca [[STRUCT_S:%.*]], align 4
845 // CHECK2-NEXT:    [[PVAR:%.*]] = alloca float*, align 8
846 // CHECK2-NEXT:    [[LVAR:%.*]] = alloca i64, align 8
847 // CHECK2-NEXT:    store i32 0, i32* [[RETVAL]], align 4
848 // CHECK2-NEXT:    call void @_ZN2SSC1ERi(%struct.SS* nonnull align 8 dereferenceable(16) [[SS]], i32* nonnull align 4 dereferenceable(4) @_ZZ4mainE5sivar)
849 // CHECK2-NEXT:    call void @_ZN1SIfEC1Ev(%struct.S* nonnull align 4 dereferenceable(4) [[TEST]])
850 // CHECK2-NEXT:    [[F:%.*]] = getelementptr inbounds [[STRUCT_S]], %struct.S* [[TEST]], i32 0, i32 0
851 // CHECK2-NEXT:    store float* [[F]], float** [[PVAR]], align 8
852 // CHECK2-NEXT:    store i64 0, i64* [[LVAR]], align 8
853 // CHECK2-NEXT:    call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB3:[0-9]+]], i32 2, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, float**, i64*)* @.omp_outlined. to void (i32*, i32*, ...)*), float** [[PVAR]], i64* [[LVAR]])
854 // CHECK2-NEXT:    [[CALL:%.*]] = call i32 @_Z5tmainIiET_v()
855 // CHECK2-NEXT:    store i32 [[CALL]], i32* [[RETVAL]], align 4
856 // CHECK2-NEXT:    call void @_ZN1SIfED1Ev(%struct.S* nonnull align 4 dereferenceable(4) [[TEST]]) #[[ATTR4:[0-9]+]]
857 // CHECK2-NEXT:    [[TMP0:%.*]] = load i32, i32* [[RETVAL]], align 4
858 // CHECK2-NEXT:    ret i32 [[TMP0]]
859 //
860 //
861 // CHECK2-LABEL: define {{[^@]+}}@_ZN2SSC1ERi
862 // CHECK2-SAME: (%struct.SS* nonnull align 8 dereferenceable(16) [[THIS:%.*]], i32* nonnull align 4 dereferenceable(4) [[D:%.*]]) unnamed_addr #[[ATTR1:[0-9]+]] align 2 {
863 // CHECK2-NEXT:  entry:
864 // CHECK2-NEXT:    [[THIS_ADDR:%.*]] = alloca %struct.SS*, align 8
865 // CHECK2-NEXT:    [[D_ADDR:%.*]] = alloca i32*, align 8
866 // CHECK2-NEXT:    store %struct.SS* [[THIS]], %struct.SS** [[THIS_ADDR]], align 8
867 // CHECK2-NEXT:    store i32* [[D]], i32** [[D_ADDR]], align 8
868 // CHECK2-NEXT:    [[THIS1:%.*]] = load %struct.SS*, %struct.SS** [[THIS_ADDR]], align 8
869 // CHECK2-NEXT:    [[TMP0:%.*]] = load i32*, i32** [[D_ADDR]], align 8
870 // CHECK2-NEXT:    call void @_ZN2SSC2ERi(%struct.SS* nonnull align 8 dereferenceable(16) [[THIS1]], i32* nonnull align 4 dereferenceable(4) [[TMP0]])
871 // CHECK2-NEXT:    ret void
872 //
873 //
874 // CHECK2-LABEL: define {{[^@]+}}@_ZN1SIfEC1Ev
875 // CHECK2-SAME: (%struct.S* nonnull align 4 dereferenceable(4) [[THIS:%.*]]) unnamed_addr #[[ATTR1]] align 2 {
876 // CHECK2-NEXT:  entry:
877 // CHECK2-NEXT:    [[THIS_ADDR:%.*]] = alloca %struct.S*, align 8
878 // CHECK2-NEXT:    store %struct.S* [[THIS]], %struct.S** [[THIS_ADDR]], align 8
879 // CHECK2-NEXT:    [[THIS1:%.*]] = load %struct.S*, %struct.S** [[THIS_ADDR]], align 8
880 // CHECK2-NEXT:    call void @_ZN1SIfEC2Ev(%struct.S* nonnull align 4 dereferenceable(4) [[THIS1]])
881 // CHECK2-NEXT:    ret void
882 //
883 //
884 // CHECK2-LABEL: define {{[^@]+}}@.omp_outlined.
885 // CHECK2-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], float** nonnull align 8 dereferenceable(8) [[PVAR:%.*]], i64* nonnull align 8 dereferenceable(8) [[LVAR:%.*]]) #[[ATTR2:[0-9]+]] {
886 // CHECK2-NEXT:  entry:
887 // CHECK2-NEXT:    [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
888 // CHECK2-NEXT:    [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
889 // CHECK2-NEXT:    [[PVAR_ADDR:%.*]] = alloca float**, align 8
890 // CHECK2-NEXT:    [[LVAR_ADDR:%.*]] = alloca i64*, align 8
891 // CHECK2-NEXT:    [[DOTOMP_IV:%.*]] = alloca i32, align 4
892 // CHECK2-NEXT:    [[TMP:%.*]] = alloca i32, align 4
893 // CHECK2-NEXT:    [[DOTLINEAR_START:%.*]] = alloca float*, align 8
894 // CHECK2-NEXT:    [[DOTLINEAR_START1:%.*]] = alloca i64, align 8
895 // CHECK2-NEXT:    [[DOTOMP_LB:%.*]] = alloca i32, align 4
896 // CHECK2-NEXT:    [[DOTOMP_UB:%.*]] = alloca i32, align 4
897 // CHECK2-NEXT:    [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
898 // CHECK2-NEXT:    [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
899 // CHECK2-NEXT:    [[I:%.*]] = alloca i32, align 4
900 // CHECK2-NEXT:    [[PVAR2:%.*]] = alloca float*, align 8
901 // CHECK2-NEXT:    store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
902 // CHECK2-NEXT:    store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
903 // CHECK2-NEXT:    store float** [[PVAR]], float*** [[PVAR_ADDR]], align 8
904 // CHECK2-NEXT:    store i64* [[LVAR]], i64** [[LVAR_ADDR]], align 8
905 // CHECK2-NEXT:    [[TMP0:%.*]] = load float**, float*** [[PVAR_ADDR]], align 8
906 // CHECK2-NEXT:    [[TMP1:%.*]] = load i64*, i64** [[LVAR_ADDR]], align 8
907 // CHECK2-NEXT:    [[TMP2:%.*]] = load float*, float** [[TMP0]], align 8
908 // CHECK2-NEXT:    store float* [[TMP2]], float** [[DOTLINEAR_START]], align 8
909 // CHECK2-NEXT:    [[TMP3:%.*]] = load i64, i64* [[TMP1]], align 8
910 // CHECK2-NEXT:    store i64 [[TMP3]], i64* [[DOTLINEAR_START1]], align 8
911 // CHECK2-NEXT:    store i32 0, i32* [[DOTOMP_LB]], align 4
912 // CHECK2-NEXT:    store i32 1, i32* [[DOTOMP_UB]], align 4
913 // CHECK2-NEXT:    store i32 1, i32* [[DOTOMP_STRIDE]], align 4
914 // CHECK2-NEXT:    store i32 0, i32* [[DOTOMP_IS_LAST]], align 4
915 // CHECK2-NEXT:    [[TMP4:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
916 // CHECK2-NEXT:    [[TMP5:%.*]] = load i32, i32* [[TMP4]], align 4
917 // CHECK2-NEXT:    call void @__kmpc_barrier(%struct.ident_t* @[[GLOB1:[0-9]+]], i32 [[TMP5]])
918 // CHECK2-NEXT:    [[DOTLVAR__VOID_ADDR:%.*]] = call i8* @__kmpc_alloc(i32 [[TMP5]], i64 8, i8* inttoptr (i64 5 to i8*))
919 // CHECK2-NEXT:    [[DOTLVAR__ADDR:%.*]] = bitcast i8* [[DOTLVAR__VOID_ADDR]] to i64*
920 // CHECK2-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)
921 // CHECK2-NEXT:    [[TMP6:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
922 // CHECK2-NEXT:    [[CMP:%.*]] = icmp sgt i32 [[TMP6]], 1
923 // CHECK2-NEXT:    br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
924 // CHECK2:       cond.true:
925 // CHECK2-NEXT:    br label [[COND_END:%.*]]
926 // CHECK2:       cond.false:
927 // CHECK2-NEXT:    [[TMP7:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
928 // CHECK2-NEXT:    br label [[COND_END]]
929 // CHECK2:       cond.end:
930 // CHECK2-NEXT:    [[COND:%.*]] = phi i32 [ 1, [[COND_TRUE]] ], [ [[TMP7]], [[COND_FALSE]] ]
931 // CHECK2-NEXT:    store i32 [[COND]], i32* [[DOTOMP_UB]], align 4
932 // CHECK2-NEXT:    [[TMP8:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4
933 // CHECK2-NEXT:    store i32 [[TMP8]], i32* [[DOTOMP_IV]], align 4
934 // CHECK2-NEXT:    br label [[OMP_INNER_FOR_COND:%.*]]
935 // CHECK2:       omp.inner.for.cond:
936 // CHECK2-NEXT:    [[TMP9:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
937 // CHECK2-NEXT:    [[TMP10:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
938 // CHECK2-NEXT:    [[CMP3:%.*]] = icmp sle i32 [[TMP9]], [[TMP10]]
939 // CHECK2-NEXT:    br i1 [[CMP3]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_COND_CLEANUP:%.*]]
940 // CHECK2:       omp.inner.for.cond.cleanup:
941 // CHECK2-NEXT:    br label [[OMP_INNER_FOR_END:%.*]]
942 // CHECK2:       omp.inner.for.body:
943 // CHECK2-NEXT:    [[TMP11:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
944 // CHECK2-NEXT:    [[MUL:%.*]] = mul nsw i32 [[TMP11]], 1
945 // CHECK2-NEXT:    [[ADD:%.*]] = add nsw i32 0, [[MUL]]
946 // CHECK2-NEXT:    store i32 [[ADD]], i32* [[I]], align 4
947 // CHECK2-NEXT:    [[TMP12:%.*]] = load float*, float** [[DOTLINEAR_START]], align 8
948 // CHECK2-NEXT:    [[TMP13:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
949 // CHECK2-NEXT:    [[MUL4:%.*]] = mul nsw i32 [[TMP13]], 3
950 // CHECK2-NEXT:    [[IDX_EXT:%.*]] = sext i32 [[MUL4]] to i64
951 // CHECK2-NEXT:    [[ADD_PTR:%.*]] = getelementptr inbounds float, float* [[TMP12]], i64 [[IDX_EXT]]
952 // CHECK2-NEXT:    store float* [[ADD_PTR]], float** [[PVAR2]], align 8
953 // CHECK2-NEXT:    [[TMP14:%.*]] = load i64, i64* [[DOTLINEAR_START1]], align 8
954 // CHECK2-NEXT:    [[TMP15:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
955 // CHECK2-NEXT:    [[MUL5:%.*]] = mul nsw i32 [[TMP15]], 3
956 // CHECK2-NEXT:    [[CONV:%.*]] = sext i32 [[MUL5]] to i64
957 // CHECK2-NEXT:    [[ADD6:%.*]] = add nsw i64 [[TMP14]], [[CONV]]
958 // CHECK2-NEXT:    store i64 [[ADD6]], i64* [[DOTLVAR__ADDR]], align 8
959 // CHECK2-NEXT:    [[TMP16:%.*]] = load float*, float** [[PVAR2]], align 8
960 // CHECK2-NEXT:    [[ADD_PTR7:%.*]] = getelementptr inbounds float, float* [[TMP16]], i64 3
961 // CHECK2-NEXT:    store float* [[ADD_PTR7]], float** [[PVAR2]], align 8
962 // CHECK2-NEXT:    [[TMP17:%.*]] = load i64, i64* [[DOTLVAR__ADDR]], align 8
963 // CHECK2-NEXT:    [[ADD8:%.*]] = add nsw i64 [[TMP17]], 3
964 // CHECK2-NEXT:    store i64 [[ADD8]], i64* [[DOTLVAR__ADDR]], align 8
965 // CHECK2-NEXT:    br label [[OMP_BODY_CONTINUE:%.*]]
966 // CHECK2:       omp.body.continue:
967 // CHECK2-NEXT:    br label [[OMP_INNER_FOR_INC:%.*]]
968 // CHECK2:       omp.inner.for.inc:
969 // CHECK2-NEXT:    [[TMP18:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
970 // CHECK2-NEXT:    [[ADD9:%.*]] = add nsw i32 [[TMP18]], 1
971 // CHECK2-NEXT:    store i32 [[ADD9]], i32* [[DOTOMP_IV]], align 4
972 // CHECK2-NEXT:    br label [[OMP_INNER_FOR_COND]]
973 // CHECK2:       omp.inner.for.end:
974 // CHECK2-NEXT:    br label [[OMP_LOOP_EXIT:%.*]]
975 // CHECK2:       omp.loop.exit:
976 // CHECK2-NEXT:    call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB2]], i32 [[TMP5]])
977 // CHECK2-NEXT:    [[TMP19:%.*]] = bitcast i64* [[DOTLVAR__ADDR]] to i8*
978 // CHECK2-NEXT:    call void @__kmpc_free(i32 [[TMP5]], i8* [[TMP19]], i8* inttoptr (i64 5 to i8*))
979 // CHECK2-NEXT:    [[TMP20:%.*]] = load i32, i32* [[DOTOMP_IS_LAST]], align 4
980 // CHECK2-NEXT:    [[TMP21:%.*]] = icmp ne i32 [[TMP20]], 0
981 // CHECK2-NEXT:    br i1 [[TMP21]], label [[DOTOMP_LINEAR_PU:%.*]], label [[DOTOMP_LINEAR_PU_DONE:%.*]]
982 // CHECK2:       .omp.linear.pu:
983 // CHECK2-NEXT:    [[TMP22:%.*]] = load float*, float** [[PVAR2]], align 8
984 // CHECK2-NEXT:    store float* [[TMP22]], float** [[TMP0]], align 8
985 // CHECK2-NEXT:    [[TMP23:%.*]] = load i64, i64* [[DOTLVAR__ADDR]], align 8
986 // CHECK2-NEXT:    store i64 [[TMP23]], i64* [[TMP1]], align 8
987 // CHECK2-NEXT:    br label [[DOTOMP_LINEAR_PU_DONE]]
988 // CHECK2:       .omp.linear.pu.done:
989 // CHECK2-NEXT:    call void @__kmpc_barrier(%struct.ident_t* @[[GLOB1]], i32 [[TMP5]])
990 // CHECK2-NEXT:    ret void
991 //
992 //
993 // CHECK2-LABEL: define {{[^@]+}}@_Z5tmainIiET_v
994 // CHECK2-SAME: () #[[ATTR5:[0-9]+]] {
995 // CHECK2-NEXT:  entry:
996 // CHECK2-NEXT:    [[TEST:%.*]] = alloca [[STRUCT_S_0:%.*]], align 4
997 // CHECK2-NEXT:    [[SST:%.*]] = alloca [[STRUCT_SST:%.*]], align 4
998 // CHECK2-NEXT:    [[PVAR:%.*]] = alloca i32*, align 8
999 // CHECK2-NEXT:    [[LVAR:%.*]] = alloca i32*, align 8
1000 // CHECK2-NEXT:    call void @_ZN1SIiEC1Ev(%struct.S.0* nonnull align 4 dereferenceable(4) [[TEST]])
1001 // CHECK2-NEXT:    call void @_ZN3SSTIiEC1Ev(%struct.SST* nonnull align 4 dereferenceable(4) [[SST]])
1002 // CHECK2-NEXT:    [[F:%.*]] = getelementptr inbounds [[STRUCT_S_0]], %struct.S.0* [[TEST]], i32 0, i32 0
1003 // CHECK2-NEXT:    store i32* [[F]], i32** [[PVAR]], align 8
1004 // CHECK2-NEXT:    [[F1:%.*]] = getelementptr inbounds [[STRUCT_S_0]], %struct.S.0* [[TEST]], i32 0, i32 0
1005 // CHECK2-NEXT:    store i32* [[F1]], i32** [[LVAR]], align 8
1006 // CHECK2-NEXT:    [[TMP0:%.*]] = load i32*, i32** [[LVAR]], align 8
1007 // CHECK2-NEXT:    call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB3]], i32 2, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32**, i32*)* @.omp_outlined..2 to void (i32*, i32*, ...)*), i32** [[PVAR]], i32* [[TMP0]])
1008 // CHECK2-NEXT:    call void @_ZN1SIiED1Ev(%struct.S.0* nonnull align 4 dereferenceable(4) [[TEST]]) #[[ATTR4]]
1009 // CHECK2-NEXT:    ret i32 0
1010 //
1011 //
1012 // CHECK2-LABEL: define {{[^@]+}}@_ZN1SIfED1Ev
1013 // CHECK2-SAME: (%struct.S* nonnull align 4 dereferenceable(4) [[THIS:%.*]]) unnamed_addr #[[ATTR1]] align 2 {
1014 // CHECK2-NEXT:  entry:
1015 // CHECK2-NEXT:    [[THIS_ADDR:%.*]] = alloca %struct.S*, align 8
1016 // CHECK2-NEXT:    store %struct.S* [[THIS]], %struct.S** [[THIS_ADDR]], align 8
1017 // CHECK2-NEXT:    [[THIS1:%.*]] = load %struct.S*, %struct.S** [[THIS_ADDR]], align 8
1018 // CHECK2-NEXT:    call void @_ZN1SIfED2Ev(%struct.S* nonnull align 4 dereferenceable(4) [[THIS1]]) #[[ATTR4]]
1019 // CHECK2-NEXT:    ret void
1020 //
1021 //
1022 // CHECK2-LABEL: define {{[^@]+}}@_ZN2SSC2ERi
1023 // CHECK2-SAME: (%struct.SS* nonnull align 8 dereferenceable(16) [[THIS:%.*]], i32* nonnull align 4 dereferenceable(4) [[D:%.*]]) unnamed_addr #[[ATTR1]] align 2 {
1024 // CHECK2-NEXT:  entry:
1025 // CHECK2-NEXT:    [[THIS_ADDR:%.*]] = alloca %struct.SS*, align 8
1026 // CHECK2-NEXT:    [[D_ADDR:%.*]] = alloca i32*, align 8
1027 // CHECK2-NEXT:    store %struct.SS* [[THIS]], %struct.SS** [[THIS_ADDR]], align 8
1028 // CHECK2-NEXT:    store i32* [[D]], i32** [[D_ADDR]], align 8
1029 // CHECK2-NEXT:    [[THIS1:%.*]] = load %struct.SS*, %struct.SS** [[THIS_ADDR]], align 8
1030 // CHECK2-NEXT:    [[A:%.*]] = getelementptr inbounds [[STRUCT_SS:%.*]], %struct.SS* [[THIS1]], i32 0, i32 0
1031 // CHECK2-NEXT:    store i32 0, i32* [[A]], align 8
1032 // CHECK2-NEXT:    [[B:%.*]] = getelementptr inbounds [[STRUCT_SS]], %struct.SS* [[THIS1]], i32 0, i32 1
1033 // CHECK2-NEXT:    [[BF_LOAD:%.*]] = load i8, i8* [[B]], align 4
1034 // CHECK2-NEXT:    [[BF_CLEAR:%.*]] = and i8 [[BF_LOAD]], -16
1035 // CHECK2-NEXT:    store i8 [[BF_CLEAR]], i8* [[B]], align 4
1036 // CHECK2-NEXT:    [[C:%.*]] = getelementptr inbounds [[STRUCT_SS]], %struct.SS* [[THIS1]], i32 0, i32 2
1037 // CHECK2-NEXT:    [[TMP0:%.*]] = load i32*, i32** [[D_ADDR]], align 8
1038 // CHECK2-NEXT:    store i32* [[TMP0]], i32** [[C]], align 8
1039 // CHECK2-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.SS*)* @.omp_outlined..1 to void (i32*, i32*, ...)*), %struct.SS* [[THIS1]])
1040 // CHECK2-NEXT:    ret void
1041 //
1042 //
1043 // CHECK2-LABEL: define {{[^@]+}}@.omp_outlined..1
1044 // CHECK2-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], %struct.SS* [[THIS:%.*]]) #[[ATTR2]] {
1045 // CHECK2-NEXT:  entry:
1046 // CHECK2-NEXT:    [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
1047 // CHECK2-NEXT:    [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
1048 // CHECK2-NEXT:    [[THIS_ADDR:%.*]] = alloca %struct.SS*, align 8
1049 // CHECK2-NEXT:    [[A:%.*]] = alloca i32*, align 8
1050 // CHECK2-NEXT:    [[B:%.*]] = alloca i32, align 4
1051 // CHECK2-NEXT:    [[C:%.*]] = alloca i32*, align 8
1052 // CHECK2-NEXT:    [[TMP:%.*]] = alloca i32*, align 8
1053 // CHECK2-NEXT:    [[_TMP3:%.*]] = alloca i32*, align 8
1054 // CHECK2-NEXT:    [[DOTOMP_IV:%.*]] = alloca i32, align 4
1055 // CHECK2-NEXT:    [[_TMP4:%.*]] = alloca i32, align 4
1056 // CHECK2-NEXT:    [[DOTLINEAR_START:%.*]] = alloca i32, align 4
1057 // CHECK2-NEXT:    [[DOTLINEAR_START5:%.*]] = alloca i32, align 4
1058 // CHECK2-NEXT:    [[DOTLINEAR_START6:%.*]] = alloca i32, align 4
1059 // CHECK2-NEXT:    [[DOTOMP_LB:%.*]] = alloca i32, align 4
1060 // CHECK2-NEXT:    [[DOTOMP_UB:%.*]] = alloca i32, align 4
1061 // CHECK2-NEXT:    [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
1062 // CHECK2-NEXT:    [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
1063 // CHECK2-NEXT:    [[I:%.*]] = alloca i32, align 4
1064 // CHECK2-NEXT:    [[A7:%.*]] = alloca i32, align 4
1065 // CHECK2-NEXT:    [[_TMP8:%.*]] = alloca i32*, align 8
1066 // CHECK2-NEXT:    [[B9:%.*]] = alloca i32, align 4
1067 // CHECK2-NEXT:    [[C10:%.*]] = alloca i32, align 4
1068 // CHECK2-NEXT:    [[_TMP11:%.*]] = alloca i32*, align 8
1069 // CHECK2-NEXT:    [[_TMP20:%.*]] = alloca i32*, align 8
1070 // CHECK2-NEXT:    [[_TMP21:%.*]] = alloca i32*, align 8
1071 // CHECK2-NEXT:    store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
1072 // CHECK2-NEXT:    store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
1073 // CHECK2-NEXT:    store %struct.SS* [[THIS]], %struct.SS** [[THIS_ADDR]], align 8
1074 // CHECK2-NEXT:    [[TMP0:%.*]] = load %struct.SS*, %struct.SS** [[THIS_ADDR]], align 8
1075 // CHECK2-NEXT:    [[A1:%.*]] = getelementptr inbounds [[STRUCT_SS:%.*]], %struct.SS* [[TMP0]], i32 0, i32 0
1076 // CHECK2-NEXT:    store i32* [[A1]], i32** [[A]], align 8
1077 // CHECK2-NEXT:    [[C2:%.*]] = getelementptr inbounds [[STRUCT_SS]], %struct.SS* [[TMP0]], i32 0, i32 2
1078 // CHECK2-NEXT:    [[TMP1:%.*]] = load i32*, i32** [[C2]], align 8
1079 // CHECK2-NEXT:    store i32* [[TMP1]], i32** [[C]], align 8
1080 // CHECK2-NEXT:    [[TMP2:%.*]] = load i32*, i32** [[A]], align 8
1081 // CHECK2-NEXT:    store i32* [[TMP2]], i32** [[TMP]], align 8
1082 // CHECK2-NEXT:    [[TMP3:%.*]] = load i32*, i32** [[C]], align 8
1083 // CHECK2-NEXT:    store i32* [[TMP3]], i32** [[_TMP3]], align 8
1084 // CHECK2-NEXT:    [[TMP4:%.*]] = load i32*, i32** [[TMP]], align 8
1085 // CHECK2-NEXT:    [[TMP5:%.*]] = load i32, i32* [[TMP4]], align 4
1086 // CHECK2-NEXT:    store i32 [[TMP5]], i32* [[DOTLINEAR_START]], align 4
1087 // CHECK2-NEXT:    [[TMP6:%.*]] = load i32, i32* [[B]], align 4
1088 // CHECK2-NEXT:    store i32 [[TMP6]], i32* [[DOTLINEAR_START5]], align 4
1089 // CHECK2-NEXT:    [[TMP7:%.*]] = load i32*, i32** [[_TMP3]], align 8
1090 // CHECK2-NEXT:    [[TMP8:%.*]] = load i32, i32* [[TMP7]], align 4
1091 // CHECK2-NEXT:    store i32 [[TMP8]], i32* [[DOTLINEAR_START6]], align 4
1092 // CHECK2-NEXT:    store i32 0, i32* [[DOTOMP_LB]], align 4
1093 // CHECK2-NEXT:    store i32 1, i32* [[DOTOMP_UB]], align 4
1094 // CHECK2-NEXT:    store i32 1, i32* [[DOTOMP_STRIDE]], align 4
1095 // CHECK2-NEXT:    store i32 0, i32* [[DOTOMP_IS_LAST]], align 4
1096 // CHECK2-NEXT:    [[TMP9:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
1097 // CHECK2-NEXT:    [[TMP10:%.*]] = load i32, i32* [[TMP9]], align 4
1098 // CHECK2-NEXT:    call void @__kmpc_barrier(%struct.ident_t* @[[GLOB1]], i32 [[TMP10]])
1099 // CHECK2-NEXT:    store i32* [[A7]], i32** [[_TMP8]], align 8
1100 // CHECK2-NEXT:    store i32* [[C10]], i32** [[_TMP11]], align 8
1101 // CHECK2-NEXT:    call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB2]], i32 [[TMP10]], i32 34, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_LB]], i32* [[DOTOMP_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1)
1102 // CHECK2-NEXT:    [[TMP11:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
1103 // CHECK2-NEXT:    [[CMP:%.*]] = icmp sgt i32 [[TMP11]], 1
1104 // CHECK2-NEXT:    br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
1105 // CHECK2:       cond.true:
1106 // CHECK2-NEXT:    br label [[COND_END:%.*]]
1107 // CHECK2:       cond.false:
1108 // CHECK2-NEXT:    [[TMP12:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
1109 // CHECK2-NEXT:    br label [[COND_END]]
1110 // CHECK2:       cond.end:
1111 // CHECK2-NEXT:    [[COND:%.*]] = phi i32 [ 1, [[COND_TRUE]] ], [ [[TMP12]], [[COND_FALSE]] ]
1112 // CHECK2-NEXT:    store i32 [[COND]], i32* [[DOTOMP_UB]], align 4
1113 // CHECK2-NEXT:    [[TMP13:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4
1114 // CHECK2-NEXT:    store i32 [[TMP13]], i32* [[DOTOMP_IV]], align 4
1115 // CHECK2-NEXT:    br label [[OMP_INNER_FOR_COND:%.*]]
1116 // CHECK2:       omp.inner.for.cond:
1117 // CHECK2-NEXT:    [[TMP14:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
1118 // CHECK2-NEXT:    [[TMP15:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
1119 // CHECK2-NEXT:    [[CMP12:%.*]] = icmp sle i32 [[TMP14]], [[TMP15]]
1120 // CHECK2-NEXT:    br i1 [[CMP12]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
1121 // CHECK2:       omp.inner.for.body:
1122 // CHECK2-NEXT:    [[TMP16:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
1123 // CHECK2-NEXT:    [[MUL:%.*]] = mul nsw i32 [[TMP16]], 1
1124 // CHECK2-NEXT:    [[ADD:%.*]] = add nsw i32 0, [[MUL]]
1125 // CHECK2-NEXT:    store i32 [[ADD]], i32* [[I]], align 4
1126 // CHECK2-NEXT:    [[TMP17:%.*]] = load i32, i32* [[DOTLINEAR_START]], align 4
1127 // CHECK2-NEXT:    [[TMP18:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
1128 // CHECK2-NEXT:    [[MUL13:%.*]] = mul nsw i32 [[TMP18]], 1
1129 // CHECK2-NEXT:    [[ADD14:%.*]] = add nsw i32 [[TMP17]], [[MUL13]]
1130 // CHECK2-NEXT:    store i32 [[ADD14]], i32* [[A7]], align 4
1131 // CHECK2-NEXT:    [[TMP19:%.*]] = load i32, i32* [[DOTLINEAR_START5]], align 4
1132 // CHECK2-NEXT:    [[TMP20:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
1133 // CHECK2-NEXT:    [[MUL15:%.*]] = mul nsw i32 [[TMP20]], 1
1134 // CHECK2-NEXT:    [[ADD16:%.*]] = add nsw i32 [[TMP19]], [[MUL15]]
1135 // CHECK2-NEXT:    store i32 [[ADD16]], i32* [[B9]], align 4
1136 // CHECK2-NEXT:    [[TMP21:%.*]] = load i32, i32* [[DOTLINEAR_START6]], align 4
1137 // CHECK2-NEXT:    [[TMP22:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
1138 // CHECK2-NEXT:    [[MUL17:%.*]] = mul nsw i32 [[TMP22]], 1
1139 // CHECK2-NEXT:    [[ADD18:%.*]] = add nsw i32 [[TMP21]], [[MUL17]]
1140 // CHECK2-NEXT:    store i32 [[ADD18]], i32* [[C10]], align 4
1141 // CHECK2-NEXT:    [[TMP23:%.*]] = load i32*, i32** [[_TMP8]], align 8
1142 // CHECK2-NEXT:    [[TMP24:%.*]] = load i32, i32* [[TMP23]], align 4
1143 // CHECK2-NEXT:    [[INC:%.*]] = add nsw i32 [[TMP24]], 1
1144 // CHECK2-NEXT:    store i32 [[INC]], i32* [[TMP23]], align 4
1145 // CHECK2-NEXT:    [[TMP25:%.*]] = load i32, i32* [[B9]], align 4
1146 // CHECK2-NEXT:    [[DEC:%.*]] = add nsw i32 [[TMP25]], -1
1147 // CHECK2-NEXT:    store i32 [[DEC]], i32* [[B9]], align 4
1148 // CHECK2-NEXT:    [[TMP26:%.*]] = load i32*, i32** [[_TMP11]], align 8
1149 // CHECK2-NEXT:    [[TMP27:%.*]] = load i32, i32* [[TMP26]], align 4
1150 // CHECK2-NEXT:    [[DIV:%.*]] = sdiv i32 [[TMP27]], 1
1151 // CHECK2-NEXT:    store i32 [[DIV]], i32* [[TMP26]], align 4
1152 // CHECK2-NEXT:    br label [[OMP_BODY_CONTINUE:%.*]]
1153 // CHECK2:       omp.body.continue:
1154 // CHECK2-NEXT:    br label [[OMP_INNER_FOR_INC:%.*]]
1155 // CHECK2:       omp.inner.for.inc:
1156 // CHECK2-NEXT:    [[TMP28:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
1157 // CHECK2-NEXT:    [[ADD19:%.*]] = add nsw i32 [[TMP28]], 1
1158 // CHECK2-NEXT:    store i32 [[ADD19]], i32* [[DOTOMP_IV]], align 4
1159 // CHECK2-NEXT:    br label [[OMP_INNER_FOR_COND]]
1160 // CHECK2:       omp.inner.for.end:
1161 // CHECK2-NEXT:    br label [[OMP_LOOP_EXIT:%.*]]
1162 // CHECK2:       omp.loop.exit:
1163 // CHECK2-NEXT:    call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB2]], i32 [[TMP10]])
1164 // CHECK2-NEXT:    [[TMP29:%.*]] = load i32, i32* [[DOTOMP_IS_LAST]], align 4
1165 // CHECK2-NEXT:    [[TMP30:%.*]] = icmp ne i32 [[TMP29]], 0
1166 // CHECK2-NEXT:    br i1 [[TMP30]], label [[DOTOMP_LINEAR_PU:%.*]], label [[DOTOMP_LINEAR_PU_DONE:%.*]]
1167 // CHECK2:       .omp.linear.pu:
1168 // CHECK2-NEXT:    [[TMP31:%.*]] = load i32*, i32** [[TMP]], align 8
1169 // CHECK2-NEXT:    store i32* [[TMP31]], i32** [[_TMP20]], align 8
1170 // CHECK2-NEXT:    [[TMP32:%.*]] = load i32, i32* [[A7]], align 4
1171 // CHECK2-NEXT:    [[TMP33:%.*]] = load i32*, i32** [[_TMP20]], align 8
1172 // CHECK2-NEXT:    store i32 [[TMP32]], i32* [[TMP33]], align 4
1173 // CHECK2-NEXT:    [[TMP34:%.*]] = load i32, i32* [[B9]], align 4
1174 // CHECK2-NEXT:    store i32 [[TMP34]], i32* [[B]], align 4
1175 // CHECK2-NEXT:    [[TMP35:%.*]] = load i32*, i32** [[_TMP3]], align 8
1176 // CHECK2-NEXT:    store i32* [[TMP35]], i32** [[_TMP21]], align 8
1177 // CHECK2-NEXT:    [[TMP36:%.*]] = load i32, i32* [[C10]], align 4
1178 // CHECK2-NEXT:    [[TMP37:%.*]] = load i32*, i32** [[_TMP21]], align 8
1179 // CHECK2-NEXT:    store i32 [[TMP36]], i32* [[TMP37]], align 4
1180 // CHECK2-NEXT:    [[TMP38:%.*]] = load i32, i32* [[B]], align 4
1181 // CHECK2-NEXT:    [[B22:%.*]] = getelementptr inbounds [[STRUCT_SS]], %struct.SS* [[TMP0]], i32 0, i32 1
1182 // CHECK2-NEXT:    [[TMP39:%.*]] = trunc i32 [[TMP38]] to i8
1183 // CHECK2-NEXT:    [[BF_LOAD:%.*]] = load i8, i8* [[B22]], align 4
1184 // CHECK2-NEXT:    [[BF_VALUE:%.*]] = and i8 [[TMP39]], 15
1185 // CHECK2-NEXT:    [[BF_CLEAR:%.*]] = and i8 [[BF_LOAD]], -16
1186 // CHECK2-NEXT:    [[BF_SET:%.*]] = or i8 [[BF_CLEAR]], [[BF_VALUE]]
1187 // CHECK2-NEXT:    store i8 [[BF_SET]], i8* [[B22]], align 4
1188 // CHECK2-NEXT:    br label [[DOTOMP_LINEAR_PU_DONE]]
1189 // CHECK2:       .omp.linear.pu.done:
1190 // CHECK2-NEXT:    call void @__kmpc_barrier(%struct.ident_t* @[[GLOB1]], i32 [[TMP10]])
1191 // CHECK2-NEXT:    ret void
1192 //
1193 //
1194 // CHECK2-LABEL: define {{[^@]+}}@_ZN1SIfEC2Ev
1195 // CHECK2-SAME: (%struct.S* nonnull align 4 dereferenceable(4) [[THIS:%.*]]) unnamed_addr #[[ATTR1]] align 2 {
1196 // CHECK2-NEXT:  entry:
1197 // CHECK2-NEXT:    [[THIS_ADDR:%.*]] = alloca %struct.S*, align 8
1198 // CHECK2-NEXT:    store %struct.S* [[THIS]], %struct.S** [[THIS_ADDR]], align 8
1199 // CHECK2-NEXT:    [[THIS1:%.*]] = load %struct.S*, %struct.S** [[THIS_ADDR]], align 8
1200 // CHECK2-NEXT:    [[F:%.*]] = getelementptr inbounds [[STRUCT_S:%.*]], %struct.S* [[THIS1]], i32 0, i32 0
1201 // CHECK2-NEXT:    store float 0.000000e+00, float* [[F]], align 4
1202 // CHECK2-NEXT:    ret void
1203 //
1204 //
1205 // CHECK2-LABEL: define {{[^@]+}}@_ZN1SIiEC1Ev
1206 // CHECK2-SAME: (%struct.S.0* nonnull align 4 dereferenceable(4) [[THIS:%.*]]) unnamed_addr #[[ATTR1]] align 2 {
1207 // CHECK2-NEXT:  entry:
1208 // CHECK2-NEXT:    [[THIS_ADDR:%.*]] = alloca %struct.S.0*, align 8
1209 // CHECK2-NEXT:    store %struct.S.0* [[THIS]], %struct.S.0** [[THIS_ADDR]], align 8
1210 // CHECK2-NEXT:    [[THIS1:%.*]] = load %struct.S.0*, %struct.S.0** [[THIS_ADDR]], align 8
1211 // CHECK2-NEXT:    call void @_ZN1SIiEC2Ev(%struct.S.0* nonnull align 4 dereferenceable(4) [[THIS1]])
1212 // CHECK2-NEXT:    ret void
1213 //
1214 //
1215 // CHECK2-LABEL: define {{[^@]+}}@_ZN3SSTIiEC1Ev
1216 // CHECK2-SAME: (%struct.SST* nonnull align 4 dereferenceable(4) [[THIS:%.*]]) unnamed_addr #[[ATTR1]] align 2 {
1217 // CHECK2-NEXT:  entry:
1218 // CHECK2-NEXT:    [[THIS_ADDR:%.*]] = alloca %struct.SST*, align 8
1219 // CHECK2-NEXT:    store %struct.SST* [[THIS]], %struct.SST** [[THIS_ADDR]], align 8
1220 // CHECK2-NEXT:    [[THIS1:%.*]] = load %struct.SST*, %struct.SST** [[THIS_ADDR]], align 8
1221 // CHECK2-NEXT:    call void @_ZN3SSTIiEC2Ev(%struct.SST* nonnull align 4 dereferenceable(4) [[THIS1]])
1222 // CHECK2-NEXT:    ret void
1223 //
1224 //
1225 // CHECK2-LABEL: define {{[^@]+}}@.omp_outlined..2
1226 // CHECK2-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i32** nonnull align 8 dereferenceable(8) [[PVAR:%.*]], i32* nonnull align 4 dereferenceable(4) [[LVAR:%.*]]) #[[ATTR2]] {
1227 // CHECK2-NEXT:  entry:
1228 // CHECK2-NEXT:    [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
1229 // CHECK2-NEXT:    [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
1230 // CHECK2-NEXT:    [[PVAR_ADDR:%.*]] = alloca i32**, align 8
1231 // CHECK2-NEXT:    [[LVAR_ADDR:%.*]] = alloca i32*, align 8
1232 // CHECK2-NEXT:    [[TMP:%.*]] = alloca i32*, align 8
1233 // CHECK2-NEXT:    [[_TMP1:%.*]] = alloca i32*, align 8
1234 // CHECK2-NEXT:    [[DOTOMP_IV:%.*]] = alloca i32, align 4
1235 // CHECK2-NEXT:    [[_TMP2:%.*]] = alloca i32, align 4
1236 // CHECK2-NEXT:    [[DOTLINEAR_START:%.*]] = alloca i32*, align 8
1237 // CHECK2-NEXT:    [[DOTLINEAR_START3:%.*]] = alloca i32, align 4
1238 // CHECK2-NEXT:    [[DOTOMP_LB:%.*]] = alloca i32, align 4
1239 // CHECK2-NEXT:    [[DOTOMP_UB:%.*]] = alloca i32, align 4
1240 // CHECK2-NEXT:    [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
1241 // CHECK2-NEXT:    [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
1242 // CHECK2-NEXT:    [[I:%.*]] = alloca i32, align 4
1243 // CHECK2-NEXT:    [[PVAR4:%.*]] = alloca i32*, align 8
1244 // CHECK2-NEXT:    [[LVAR5:%.*]] = alloca i32, align 4
1245 // CHECK2-NEXT:    [[_TMP6:%.*]] = alloca i32*, align 8
1246 // CHECK2-NEXT:    [[_TMP12:%.*]] = alloca i32*, align 8
1247 // CHECK2-NEXT:    store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
1248 // CHECK2-NEXT:    store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
1249 // CHECK2-NEXT:    store i32** [[PVAR]], i32*** [[PVAR_ADDR]], align 8
1250 // CHECK2-NEXT:    store i32* [[LVAR]], i32** [[LVAR_ADDR]], align 8
1251 // CHECK2-NEXT:    [[TMP0:%.*]] = load i32**, i32*** [[PVAR_ADDR]], align 8
1252 // CHECK2-NEXT:    [[TMP1:%.*]] = load i32*, i32** [[LVAR_ADDR]], align 8
1253 // CHECK2-NEXT:    store i32* [[TMP1]], i32** [[TMP]], align 8
1254 // CHECK2-NEXT:    [[TMP2:%.*]] = load i32*, i32** [[TMP]], align 8
1255 // CHECK2-NEXT:    store i32* [[TMP2]], i32** [[_TMP1]], align 8
1256 // CHECK2-NEXT:    [[TMP3:%.*]] = load i32*, i32** [[TMP0]], align 8
1257 // CHECK2-NEXT:    store i32* [[TMP3]], i32** [[DOTLINEAR_START]], align 8
1258 // CHECK2-NEXT:    [[TMP4:%.*]] = load i32*, i32** [[_TMP1]], align 8
1259 // CHECK2-NEXT:    [[TMP5:%.*]] = load i32, i32* [[TMP4]], align 4
1260 // CHECK2-NEXT:    store i32 [[TMP5]], i32* [[DOTLINEAR_START3]], align 4
1261 // CHECK2-NEXT:    store i32 0, i32* [[DOTOMP_LB]], align 4
1262 // CHECK2-NEXT:    store i32 1, i32* [[DOTOMP_UB]], align 4
1263 // CHECK2-NEXT:    store i32 1, i32* [[DOTOMP_STRIDE]], align 4
1264 // CHECK2-NEXT:    store i32 0, i32* [[DOTOMP_IS_LAST]], align 4
1265 // CHECK2-NEXT:    [[TMP6:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
1266 // CHECK2-NEXT:    [[TMP7:%.*]] = load i32, i32* [[TMP6]], align 4
1267 // CHECK2-NEXT:    call void @__kmpc_barrier(%struct.ident_t* @[[GLOB1]], i32 [[TMP7]])
1268 // CHECK2-NEXT:    store i32* [[LVAR5]], i32** [[_TMP6]], align 8
1269 // CHECK2-NEXT:    call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB2]], i32 [[TMP7]], i32 34, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_LB]], i32* [[DOTOMP_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1)
1270 // CHECK2-NEXT:    [[TMP8:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
1271 // CHECK2-NEXT:    [[CMP:%.*]] = icmp sgt i32 [[TMP8]], 1
1272 // CHECK2-NEXT:    br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
1273 // CHECK2:       cond.true:
1274 // CHECK2-NEXT:    br label [[COND_END:%.*]]
1275 // CHECK2:       cond.false:
1276 // CHECK2-NEXT:    [[TMP9:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
1277 // CHECK2-NEXT:    br label [[COND_END]]
1278 // CHECK2:       cond.end:
1279 // CHECK2-NEXT:    [[COND:%.*]] = phi i32 [ 1, [[COND_TRUE]] ], [ [[TMP9]], [[COND_FALSE]] ]
1280 // CHECK2-NEXT:    store i32 [[COND]], i32* [[DOTOMP_UB]], align 4
1281 // CHECK2-NEXT:    [[TMP10:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4
1282 // CHECK2-NEXT:    store i32 [[TMP10]], i32* [[DOTOMP_IV]], align 4
1283 // CHECK2-NEXT:    br label [[OMP_INNER_FOR_COND:%.*]]
1284 // CHECK2:       omp.inner.for.cond:
1285 // CHECK2-NEXT:    [[TMP11:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
1286 // CHECK2-NEXT:    [[TMP12:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
1287 // CHECK2-NEXT:    [[CMP7:%.*]] = icmp sle i32 [[TMP11]], [[TMP12]]
1288 // CHECK2-NEXT:    br i1 [[CMP7]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
1289 // CHECK2:       omp.inner.for.body:
1290 // CHECK2-NEXT:    [[TMP13:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
1291 // CHECK2-NEXT:    [[MUL:%.*]] = mul nsw i32 [[TMP13]], 1
1292 // CHECK2-NEXT:    [[ADD:%.*]] = add nsw i32 0, [[MUL]]
1293 // CHECK2-NEXT:    store i32 [[ADD]], i32* [[I]], align 4
1294 // CHECK2-NEXT:    [[TMP14:%.*]] = load i32*, i32** [[DOTLINEAR_START]], align 8
1295 // CHECK2-NEXT:    [[TMP15:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
1296 // CHECK2-NEXT:    [[MUL8:%.*]] = mul nsw i32 [[TMP15]], 1
1297 // CHECK2-NEXT:    [[IDX_EXT:%.*]] = sext i32 [[MUL8]] to i64
1298 // CHECK2-NEXT:    [[ADD_PTR:%.*]] = getelementptr inbounds i32, i32* [[TMP14]], i64 [[IDX_EXT]]
1299 // CHECK2-NEXT:    store i32* [[ADD_PTR]], i32** [[PVAR4]], align 8
1300 // CHECK2-NEXT:    [[TMP16:%.*]] = load i32, i32* [[DOTLINEAR_START3]], align 4
1301 // CHECK2-NEXT:    [[TMP17:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
1302 // CHECK2-NEXT:    [[MUL9:%.*]] = mul nsw i32 [[TMP17]], 1
1303 // CHECK2-NEXT:    [[ADD10:%.*]] = add nsw i32 [[TMP16]], [[MUL9]]
1304 // CHECK2-NEXT:    store i32 [[ADD10]], i32* [[LVAR5]], align 4
1305 // CHECK2-NEXT:    [[TMP18:%.*]] = load i32*, i32** [[PVAR4]], align 8
1306 // CHECK2-NEXT:    [[INCDEC_PTR:%.*]] = getelementptr inbounds i32, i32* [[TMP18]], i32 1
1307 // CHECK2-NEXT:    store i32* [[INCDEC_PTR]], i32** [[PVAR4]], align 8
1308 // CHECK2-NEXT:    [[TMP19:%.*]] = load i32*, i32** [[_TMP6]], align 8
1309 // CHECK2-NEXT:    [[TMP20:%.*]] = load i32, i32* [[TMP19]], align 4
1310 // CHECK2-NEXT:    [[INC:%.*]] = add nsw i32 [[TMP20]], 1
1311 // CHECK2-NEXT:    store i32 [[INC]], i32* [[TMP19]], align 4
1312 // CHECK2-NEXT:    br label [[OMP_BODY_CONTINUE:%.*]]
1313 // CHECK2:       omp.body.continue:
1314 // CHECK2-NEXT:    br label [[OMP_INNER_FOR_INC:%.*]]
1315 // CHECK2:       omp.inner.for.inc:
1316 // CHECK2-NEXT:    [[TMP21:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
1317 // CHECK2-NEXT:    [[ADD11:%.*]] = add nsw i32 [[TMP21]], 1
1318 // CHECK2-NEXT:    store i32 [[ADD11]], i32* [[DOTOMP_IV]], align 4
1319 // CHECK2-NEXT:    br label [[OMP_INNER_FOR_COND]]
1320 // CHECK2:       omp.inner.for.end:
1321 // CHECK2-NEXT:    br label [[OMP_LOOP_EXIT:%.*]]
1322 // CHECK2:       omp.loop.exit:
1323 // CHECK2-NEXT:    call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB2]], i32 [[TMP7]])
1324 // CHECK2-NEXT:    [[TMP22:%.*]] = load i32, i32* [[DOTOMP_IS_LAST]], align 4
1325 // CHECK2-NEXT:    [[TMP23:%.*]] = icmp ne i32 [[TMP22]], 0
1326 // CHECK2-NEXT:    br i1 [[TMP23]], label [[DOTOMP_LINEAR_PU:%.*]], label [[DOTOMP_LINEAR_PU_DONE:%.*]]
1327 // CHECK2:       .omp.linear.pu:
1328 // CHECK2-NEXT:    [[TMP24:%.*]] = load i32*, i32** [[PVAR4]], align 8
1329 // CHECK2-NEXT:    store i32* [[TMP24]], i32** [[TMP0]], align 8
1330 // CHECK2-NEXT:    [[TMP25:%.*]] = load i32*, i32** [[_TMP1]], align 8
1331 // CHECK2-NEXT:    store i32* [[TMP25]], i32** [[_TMP12]], align 8
1332 // CHECK2-NEXT:    [[TMP26:%.*]] = load i32, i32* [[LVAR5]], align 4
1333 // CHECK2-NEXT:    [[TMP27:%.*]] = load i32*, i32** [[_TMP12]], align 8
1334 // CHECK2-NEXT:    store i32 [[TMP26]], i32* [[TMP27]], align 4
1335 // CHECK2-NEXT:    br label [[DOTOMP_LINEAR_PU_DONE]]
1336 // CHECK2:       .omp.linear.pu.done:
1337 // CHECK2-NEXT:    call void @__kmpc_barrier(%struct.ident_t* @[[GLOB1]], i32 [[TMP7]])
1338 // CHECK2-NEXT:    ret void
1339 //
1340 //
1341 // CHECK2-LABEL: define {{[^@]+}}@_ZN1SIiED1Ev
1342 // CHECK2-SAME: (%struct.S.0* nonnull align 4 dereferenceable(4) [[THIS:%.*]]) unnamed_addr #[[ATTR1]] align 2 {
1343 // CHECK2-NEXT:  entry:
1344 // CHECK2-NEXT:    [[THIS_ADDR:%.*]] = alloca %struct.S.0*, align 8
1345 // CHECK2-NEXT:    store %struct.S.0* [[THIS]], %struct.S.0** [[THIS_ADDR]], align 8
1346 // CHECK2-NEXT:    [[THIS1:%.*]] = load %struct.S.0*, %struct.S.0** [[THIS_ADDR]], align 8
1347 // CHECK2-NEXT:    call void @_ZN1SIiED2Ev(%struct.S.0* nonnull align 4 dereferenceable(4) [[THIS1]]) #[[ATTR4]]
1348 // CHECK2-NEXT:    ret void
1349 //
1350 //
1351 // CHECK2-LABEL: define {{[^@]+}}@_ZN1SIiEC2Ev
1352 // CHECK2-SAME: (%struct.S.0* nonnull align 4 dereferenceable(4) [[THIS:%.*]]) unnamed_addr #[[ATTR1]] align 2 {
1353 // CHECK2-NEXT:  entry:
1354 // CHECK2-NEXT:    [[THIS_ADDR:%.*]] = alloca %struct.S.0*, align 8
1355 // CHECK2-NEXT:    store %struct.S.0* [[THIS]], %struct.S.0** [[THIS_ADDR]], align 8
1356 // CHECK2-NEXT:    [[THIS1:%.*]] = load %struct.S.0*, %struct.S.0** [[THIS_ADDR]], align 8
1357 // CHECK2-NEXT:    [[F:%.*]] = getelementptr inbounds [[STRUCT_S_0:%.*]], %struct.S.0* [[THIS1]], i32 0, i32 0
1358 // CHECK2-NEXT:    store i32 0, i32* [[F]], align 4
1359 // CHECK2-NEXT:    ret void
1360 //
1361 //
1362 // CHECK2-LABEL: define {{[^@]+}}@_ZN3SSTIiEC2Ev
1363 // CHECK2-SAME: (%struct.SST* nonnull align 4 dereferenceable(4) [[THIS:%.*]]) unnamed_addr #[[ATTR1]] align 2 {
1364 // CHECK2-NEXT:  entry:
1365 // CHECK2-NEXT:    [[THIS_ADDR:%.*]] = alloca %struct.SST*, align 8
1366 // CHECK2-NEXT:    store %struct.SST* [[THIS]], %struct.SST** [[THIS_ADDR]], align 8
1367 // CHECK2-NEXT:    [[THIS1:%.*]] = load %struct.SST*, %struct.SST** [[THIS_ADDR]], align 8
1368 // CHECK2-NEXT:    [[A:%.*]] = getelementptr inbounds [[STRUCT_SST:%.*]], %struct.SST* [[THIS1]], i32 0, i32 0
1369 // CHECK2-NEXT:    store i32 0, i32* [[A]], align 4
1370 // CHECK2-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.SST*)* @.omp_outlined..3 to void (i32*, i32*, ...)*), %struct.SST* [[THIS1]])
1371 // CHECK2-NEXT:    ret void
1372 //
1373 //
1374 // CHECK2-LABEL: define {{[^@]+}}@.omp_outlined..3
1375 // CHECK2-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], %struct.SST* [[THIS:%.*]]) #[[ATTR2]] {
1376 // CHECK2-NEXT:  entry:
1377 // CHECK2-NEXT:    [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
1378 // CHECK2-NEXT:    [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
1379 // CHECK2-NEXT:    [[THIS_ADDR:%.*]] = alloca %struct.SST*, align 8
1380 // CHECK2-NEXT:    [[A:%.*]] = alloca i32*, align 8
1381 // CHECK2-NEXT:    [[TMP:%.*]] = alloca i32*, align 8
1382 // CHECK2-NEXT:    [[DOTOMP_IV:%.*]] = alloca i32, align 4
1383 // CHECK2-NEXT:    [[_TMP2:%.*]] = alloca i32, align 4
1384 // CHECK2-NEXT:    [[DOTLINEAR_START:%.*]] = alloca i32, align 4
1385 // CHECK2-NEXT:    [[DOTOMP_LB:%.*]] = alloca i32, align 4
1386 // CHECK2-NEXT:    [[DOTOMP_UB:%.*]] = alloca i32, align 4
1387 // CHECK2-NEXT:    [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
1388 // CHECK2-NEXT:    [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
1389 // CHECK2-NEXT:    [[I:%.*]] = alloca i32, align 4
1390 // CHECK2-NEXT:    [[A3:%.*]] = alloca i32, align 4
1391 // CHECK2-NEXT:    [[_TMP4:%.*]] = alloca i32*, align 8
1392 // CHECK2-NEXT:    [[_TMP9:%.*]] = alloca i32*, align 8
1393 // CHECK2-NEXT:    store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
1394 // CHECK2-NEXT:    store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
1395 // CHECK2-NEXT:    store %struct.SST* [[THIS]], %struct.SST** [[THIS_ADDR]], align 8
1396 // CHECK2-NEXT:    [[TMP0:%.*]] = load %struct.SST*, %struct.SST** [[THIS_ADDR]], align 8
1397 // CHECK2-NEXT:    [[A1:%.*]] = getelementptr inbounds [[STRUCT_SST:%.*]], %struct.SST* [[TMP0]], i32 0, i32 0
1398 // CHECK2-NEXT:    store i32* [[A1]], i32** [[A]], align 8
1399 // CHECK2-NEXT:    [[TMP1:%.*]] = load i32*, i32** [[A]], align 8
1400 // CHECK2-NEXT:    store i32* [[TMP1]], i32** [[TMP]], align 8
1401 // CHECK2-NEXT:    [[TMP2:%.*]] = load i32*, i32** [[TMP]], align 8
1402 // CHECK2-NEXT:    [[TMP3:%.*]] = load i32, i32* [[TMP2]], align 4
1403 // CHECK2-NEXT:    store i32 [[TMP3]], i32* [[DOTLINEAR_START]], align 4
1404 // CHECK2-NEXT:    store i32 0, i32* [[DOTOMP_LB]], align 4
1405 // CHECK2-NEXT:    store i32 1, i32* [[DOTOMP_UB]], align 4
1406 // CHECK2-NEXT:    store i32 1, i32* [[DOTOMP_STRIDE]], align 4
1407 // CHECK2-NEXT:    store i32 0, i32* [[DOTOMP_IS_LAST]], align 4
1408 // CHECK2-NEXT:    [[TMP4:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
1409 // CHECK2-NEXT:    [[TMP5:%.*]] = load i32, i32* [[TMP4]], align 4
1410 // CHECK2-NEXT:    call void @__kmpc_barrier(%struct.ident_t* @[[GLOB1]], i32 [[TMP5]])
1411 // CHECK2-NEXT:    store i32* [[A3]], i32** [[_TMP4]], align 8
1412 // CHECK2-NEXT:    call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB2]], i32 [[TMP5]], i32 34, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_LB]], i32* [[DOTOMP_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1)
1413 // CHECK2-NEXT:    [[TMP6:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
1414 // CHECK2-NEXT:    [[CMP:%.*]] = icmp sgt i32 [[TMP6]], 1
1415 // CHECK2-NEXT:    br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
1416 // CHECK2:       cond.true:
1417 // CHECK2-NEXT:    br label [[COND_END:%.*]]
1418 // CHECK2:       cond.false:
1419 // CHECK2-NEXT:    [[TMP7:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
1420 // CHECK2-NEXT:    br label [[COND_END]]
1421 // CHECK2:       cond.end:
1422 // CHECK2-NEXT:    [[COND:%.*]] = phi i32 [ 1, [[COND_TRUE]] ], [ [[TMP7]], [[COND_FALSE]] ]
1423 // CHECK2-NEXT:    store i32 [[COND]], i32* [[DOTOMP_UB]], align 4
1424 // CHECK2-NEXT:    [[TMP8:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4
1425 // CHECK2-NEXT:    store i32 [[TMP8]], i32* [[DOTOMP_IV]], align 4
1426 // CHECK2-NEXT:    br label [[OMP_INNER_FOR_COND:%.*]]
1427 // CHECK2:       omp.inner.for.cond:
1428 // CHECK2-NEXT:    [[TMP9:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
1429 // CHECK2-NEXT:    [[TMP10:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
1430 // CHECK2-NEXT:    [[CMP5:%.*]] = icmp sle i32 [[TMP9]], [[TMP10]]
1431 // CHECK2-NEXT:    br i1 [[CMP5]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
1432 // CHECK2:       omp.inner.for.body:
1433 // CHECK2-NEXT:    [[TMP11:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
1434 // CHECK2-NEXT:    [[MUL:%.*]] = mul nsw i32 [[TMP11]], 1
1435 // CHECK2-NEXT:    [[ADD:%.*]] = add nsw i32 0, [[MUL]]
1436 // CHECK2-NEXT:    store i32 [[ADD]], i32* [[I]], align 4
1437 // CHECK2-NEXT:    [[TMP12:%.*]] = load i32, i32* [[DOTLINEAR_START]], align 4
1438 // CHECK2-NEXT:    [[TMP13:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
1439 // CHECK2-NEXT:    [[MUL6:%.*]] = mul nsw i32 [[TMP13]], 1
1440 // CHECK2-NEXT:    [[ADD7:%.*]] = add nsw i32 [[TMP12]], [[MUL6]]
1441 // CHECK2-NEXT:    store i32 [[ADD7]], i32* [[A3]], align 4
1442 // CHECK2-NEXT:    [[TMP14:%.*]] = load i32*, i32** [[_TMP4]], align 8
1443 // CHECK2-NEXT:    [[TMP15:%.*]] = load i32, i32* [[TMP14]], align 4
1444 // CHECK2-NEXT:    [[INC:%.*]] = add nsw i32 [[TMP15]], 1
1445 // CHECK2-NEXT:    store i32 [[INC]], i32* [[TMP14]], align 4
1446 // CHECK2-NEXT:    br label [[OMP_BODY_CONTINUE:%.*]]
1447 // CHECK2:       omp.body.continue:
1448 // CHECK2-NEXT:    br label [[OMP_INNER_FOR_INC:%.*]]
1449 // CHECK2:       omp.inner.for.inc:
1450 // CHECK2-NEXT:    [[TMP16:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
1451 // CHECK2-NEXT:    [[ADD8:%.*]] = add nsw i32 [[TMP16]], 1
1452 // CHECK2-NEXT:    store i32 [[ADD8]], i32* [[DOTOMP_IV]], align 4
1453 // CHECK2-NEXT:    br label [[OMP_INNER_FOR_COND]]
1454 // CHECK2:       omp.inner.for.end:
1455 // CHECK2-NEXT:    br label [[OMP_LOOP_EXIT:%.*]]
1456 // CHECK2:       omp.loop.exit:
1457 // CHECK2-NEXT:    call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB2]], i32 [[TMP5]])
1458 // CHECK2-NEXT:    [[TMP17:%.*]] = load i32, i32* [[DOTOMP_IS_LAST]], align 4
1459 // CHECK2-NEXT:    [[TMP18:%.*]] = icmp ne i32 [[TMP17]], 0
1460 // CHECK2-NEXT:    br i1 [[TMP18]], label [[DOTOMP_LINEAR_PU:%.*]], label [[DOTOMP_LINEAR_PU_DONE:%.*]]
1461 // CHECK2:       .omp.linear.pu:
1462 // CHECK2-NEXT:    [[TMP19:%.*]] = load i32*, i32** [[TMP]], align 8
1463 // CHECK2-NEXT:    store i32* [[TMP19]], i32** [[_TMP9]], align 8
1464 // CHECK2-NEXT:    [[TMP20:%.*]] = load i32, i32* [[A3]], align 4
1465 // CHECK2-NEXT:    [[TMP21:%.*]] = load i32*, i32** [[_TMP9]], align 8
1466 // CHECK2-NEXT:    store i32 [[TMP20]], i32* [[TMP21]], align 4
1467 // CHECK2-NEXT:    br label [[DOTOMP_LINEAR_PU_DONE]]
1468 // CHECK2:       .omp.linear.pu.done:
1469 // CHECK2-NEXT:    call void @__kmpc_barrier(%struct.ident_t* @[[GLOB1]], i32 [[TMP5]])
1470 // CHECK2-NEXT:    ret void
1471 //
1472 //
1473 // CHECK2-LABEL: define {{[^@]+}}@_ZN1SIiED2Ev
1474 // CHECK2-SAME: (%struct.S.0* nonnull align 4 dereferenceable(4) [[THIS:%.*]]) unnamed_addr #[[ATTR1]] align 2 {
1475 // CHECK2-NEXT:  entry:
1476 // CHECK2-NEXT:    [[THIS_ADDR:%.*]] = alloca %struct.S.0*, align 8
1477 // CHECK2-NEXT:    store %struct.S.0* [[THIS]], %struct.S.0** [[THIS_ADDR]], align 8
1478 // CHECK2-NEXT:    [[THIS1:%.*]] = load %struct.S.0*, %struct.S.0** [[THIS_ADDR]], align 8
1479 // CHECK2-NEXT:    ret void
1480 //
1481 //
1482 // CHECK2-LABEL: define {{[^@]+}}@_ZN1SIfED2Ev
1483 // CHECK2-SAME: (%struct.S* nonnull align 4 dereferenceable(4) [[THIS:%.*]]) unnamed_addr #[[ATTR1]] align 2 {
1484 // CHECK2-NEXT:  entry:
1485 // CHECK2-NEXT:    [[THIS_ADDR:%.*]] = alloca %struct.S*, align 8
1486 // CHECK2-NEXT:    store %struct.S* [[THIS]], %struct.S** [[THIS_ADDR]], align 8
1487 // CHECK2-NEXT:    [[THIS1:%.*]] = load %struct.S*, %struct.S** [[THIS_ADDR]], align 8
1488 // CHECK2-NEXT:    ret void
1489 //
1490 //
1491 // CHECK3-LABEL: define {{[^@]+}}@main
1492 // CHECK3-SAME: () #[[ATTR0:[0-9]+]] {
1493 // CHECK3-NEXT:  entry:
1494 // CHECK3-NEXT:    [[RETVAL:%.*]] = alloca i32, align 4
1495 // CHECK3-NEXT:    [[SS:%.*]] = alloca [[STRUCT_SS:%.*]], align 8
1496 // CHECK3-NEXT:    [[REF_TMP:%.*]] = alloca [[CLASS_ANON:%.*]], align 1
1497 // CHECK3-NEXT:    store i32 0, i32* [[RETVAL]], align 4
1498 // CHECK3-NEXT:    call void @_ZN2SSC1ERi(%struct.SS* nonnull align 8 dereferenceable(16) [[SS]], i32* nonnull align 4 dereferenceable(4) @_ZZ4mainE5sivar)
1499 // CHECK3-NEXT:    call void @"_ZZ4mainENK3$_0clEv"(%class.anon* nonnull align 1 dereferenceable(1) [[REF_TMP]])
1500 // CHECK3-NEXT:    ret i32 0
1501 //
1502 //
1503 // CHECK3-LABEL: define {{[^@]+}}@_ZN2SSC1ERi
1504 // CHECK3-SAME: (%struct.SS* nonnull align 8 dereferenceable(16) [[THIS:%.*]], i32* nonnull align 4 dereferenceable(4) [[D:%.*]]) unnamed_addr #[[ATTR1:[0-9]+]] align 2 {
1505 // CHECK3-NEXT:  entry:
1506 // CHECK3-NEXT:    [[THIS_ADDR:%.*]] = alloca %struct.SS*, align 8
1507 // CHECK3-NEXT:    [[D_ADDR:%.*]] = alloca i32*, align 8
1508 // CHECK3-NEXT:    store %struct.SS* [[THIS]], %struct.SS** [[THIS_ADDR]], align 8
1509 // CHECK3-NEXT:    store i32* [[D]], i32** [[D_ADDR]], align 8
1510 // CHECK3-NEXT:    [[THIS1:%.*]] = load %struct.SS*, %struct.SS** [[THIS_ADDR]], align 8
1511 // CHECK3-NEXT:    [[TMP0:%.*]] = load i32*, i32** [[D_ADDR]], align 8
1512 // CHECK3-NEXT:    call void @_ZN2SSC2ERi(%struct.SS* nonnull align 8 dereferenceable(16) [[THIS1]], i32* nonnull align 4 dereferenceable(4) [[TMP0]])
1513 // CHECK3-NEXT:    ret void
1514 //
1515 //
1516 // CHECK3-LABEL: define {{[^@]+}}@_ZN2SSC2ERi
1517 // CHECK3-SAME: (%struct.SS* nonnull align 8 dereferenceable(16) [[THIS:%.*]], i32* nonnull align 4 dereferenceable(4) [[D:%.*]]) unnamed_addr #[[ATTR1]] align 2 {
1518 // CHECK3-NEXT:  entry:
1519 // CHECK3-NEXT:    [[THIS_ADDR:%.*]] = alloca %struct.SS*, align 8
1520 // CHECK3-NEXT:    [[D_ADDR:%.*]] = alloca i32*, align 8
1521 // CHECK3-NEXT:    store %struct.SS* [[THIS]], %struct.SS** [[THIS_ADDR]], align 8
1522 // CHECK3-NEXT:    store i32* [[D]], i32** [[D_ADDR]], align 8
1523 // CHECK3-NEXT:    [[THIS1:%.*]] = load %struct.SS*, %struct.SS** [[THIS_ADDR]], align 8
1524 // CHECK3-NEXT:    [[A:%.*]] = getelementptr inbounds [[STRUCT_SS:%.*]], %struct.SS* [[THIS1]], i32 0, i32 0
1525 // CHECK3-NEXT:    store i32 0, i32* [[A]], align 8
1526 // CHECK3-NEXT:    [[B:%.*]] = getelementptr inbounds [[STRUCT_SS]], %struct.SS* [[THIS1]], i32 0, i32 1
1527 // CHECK3-NEXT:    [[BF_LOAD:%.*]] = load i8, i8* [[B]], align 4
1528 // CHECK3-NEXT:    [[BF_CLEAR:%.*]] = and i8 [[BF_LOAD]], -16
1529 // CHECK3-NEXT:    store i8 [[BF_CLEAR]], i8* [[B]], align 4
1530 // CHECK3-NEXT:    [[C:%.*]] = getelementptr inbounds [[STRUCT_SS]], %struct.SS* [[THIS1]], i32 0, i32 2
1531 // CHECK3-NEXT:    [[TMP0:%.*]] = load i32*, i32** [[D_ADDR]], align 8
1532 // CHECK3-NEXT:    store i32* [[TMP0]], i32** [[C]], align 8
1533 // CHECK3-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.SS*)* @.omp_outlined. to void (i32*, i32*, ...)*), %struct.SS* [[THIS1]])
1534 // CHECK3-NEXT:    ret void
1535 //
1536 //
1537 // CHECK3-LABEL: define {{[^@]+}}@.omp_outlined.
1538 // CHECK3-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], %struct.SS* [[THIS:%.*]]) #[[ATTR3:[0-9]+]] {
1539 // CHECK3-NEXT:  entry:
1540 // CHECK3-NEXT:    [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
1541 // CHECK3-NEXT:    [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
1542 // CHECK3-NEXT:    [[THIS_ADDR:%.*]] = alloca %struct.SS*, align 8
1543 // CHECK3-NEXT:    [[A:%.*]] = alloca i32*, align 8
1544 // CHECK3-NEXT:    [[B:%.*]] = alloca i32, align 4
1545 // CHECK3-NEXT:    [[C:%.*]] = alloca i32*, align 8
1546 // CHECK3-NEXT:    [[TMP:%.*]] = alloca i32*, align 8
1547 // CHECK3-NEXT:    [[_TMP3:%.*]] = alloca i32*, align 8
1548 // CHECK3-NEXT:    [[DOTOMP_IV:%.*]] = alloca i32, align 4
1549 // CHECK3-NEXT:    [[_TMP4:%.*]] = alloca i32, align 4
1550 // CHECK3-NEXT:    [[DOTLINEAR_START:%.*]] = alloca i32, align 4
1551 // CHECK3-NEXT:    [[DOTLINEAR_START5:%.*]] = alloca i32, align 4
1552 // CHECK3-NEXT:    [[DOTLINEAR_START6:%.*]] = alloca i32, align 4
1553 // CHECK3-NEXT:    [[DOTOMP_LB:%.*]] = alloca i32, align 4
1554 // CHECK3-NEXT:    [[DOTOMP_UB:%.*]] = alloca i32, align 4
1555 // CHECK3-NEXT:    [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
1556 // CHECK3-NEXT:    [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
1557 // CHECK3-NEXT:    [[I:%.*]] = alloca i32, align 4
1558 // CHECK3-NEXT:    [[A7:%.*]] = alloca i32, align 4
1559 // CHECK3-NEXT:    [[_TMP8:%.*]] = alloca i32*, align 8
1560 // CHECK3-NEXT:    [[B9:%.*]] = alloca i32, align 4
1561 // CHECK3-NEXT:    [[C10:%.*]] = alloca i32, align 4
1562 // CHECK3-NEXT:    [[_TMP11:%.*]] = alloca i32*, align 8
1563 // CHECK3-NEXT:    [[REF_TMP:%.*]] = alloca [[CLASS_ANON_0:%.*]], align 8
1564 // CHECK3-NEXT:    [[_TMP20:%.*]] = alloca i32*, align 8
1565 // CHECK3-NEXT:    [[_TMP21:%.*]] = alloca i32*, align 8
1566 // CHECK3-NEXT:    store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
1567 // CHECK3-NEXT:    store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
1568 // CHECK3-NEXT:    store %struct.SS* [[THIS]], %struct.SS** [[THIS_ADDR]], align 8
1569 // CHECK3-NEXT:    [[TMP0:%.*]] = load %struct.SS*, %struct.SS** [[THIS_ADDR]], align 8
1570 // CHECK3-NEXT:    [[A1:%.*]] = getelementptr inbounds [[STRUCT_SS:%.*]], %struct.SS* [[TMP0]], i32 0, i32 0
1571 // CHECK3-NEXT:    store i32* [[A1]], i32** [[A]], align 8
1572 // CHECK3-NEXT:    [[C2:%.*]] = getelementptr inbounds [[STRUCT_SS]], %struct.SS* [[TMP0]], i32 0, i32 2
1573 // CHECK3-NEXT:    [[TMP1:%.*]] = load i32*, i32** [[C2]], align 8
1574 // CHECK3-NEXT:    store i32* [[TMP1]], i32** [[C]], align 8
1575 // CHECK3-NEXT:    [[TMP2:%.*]] = load i32*, i32** [[A]], align 8
1576 // CHECK3-NEXT:    store i32* [[TMP2]], i32** [[TMP]], align 8
1577 // CHECK3-NEXT:    [[TMP3:%.*]] = load i32*, i32** [[C]], align 8
1578 // CHECK3-NEXT:    store i32* [[TMP3]], i32** [[_TMP3]], align 8
1579 // CHECK3-NEXT:    [[TMP4:%.*]] = load i32*, i32** [[TMP]], align 8
1580 // CHECK3-NEXT:    [[TMP5:%.*]] = load i32, i32* [[TMP4]], align 4
1581 // CHECK3-NEXT:    store i32 [[TMP5]], i32* [[DOTLINEAR_START]], align 4
1582 // CHECK3-NEXT:    [[TMP6:%.*]] = load i32, i32* [[B]], align 4
1583 // CHECK3-NEXT:    store i32 [[TMP6]], i32* [[DOTLINEAR_START5]], align 4
1584 // CHECK3-NEXT:    [[TMP7:%.*]] = load i32*, i32** [[_TMP3]], align 8
1585 // CHECK3-NEXT:    [[TMP8:%.*]] = load i32, i32* [[TMP7]], align 4
1586 // CHECK3-NEXT:    store i32 [[TMP8]], i32* [[DOTLINEAR_START6]], align 4
1587 // CHECK3-NEXT:    store i32 0, i32* [[DOTOMP_LB]], align 4
1588 // CHECK3-NEXT:    store i32 1, i32* [[DOTOMP_UB]], align 4
1589 // CHECK3-NEXT:    store i32 1, i32* [[DOTOMP_STRIDE]], align 4
1590 // CHECK3-NEXT:    store i32 0, i32* [[DOTOMP_IS_LAST]], align 4
1591 // CHECK3-NEXT:    [[TMP9:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
1592 // CHECK3-NEXT:    [[TMP10:%.*]] = load i32, i32* [[TMP9]], align 4
1593 // CHECK3-NEXT:    call void @__kmpc_barrier(%struct.ident_t* @[[GLOB1:[0-9]+]], i32 [[TMP10]])
1594 // CHECK3-NEXT:    store i32* [[A7]], i32** [[_TMP8]], align 8
1595 // CHECK3-NEXT:    store i32* [[C10]], i32** [[_TMP11]], align 8
1596 // CHECK3-NEXT:    call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB2:[0-9]+]], i32 [[TMP10]], i32 34, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_LB]], i32* [[DOTOMP_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1)
1597 // CHECK3-NEXT:    [[TMP11:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
1598 // CHECK3-NEXT:    [[CMP:%.*]] = icmp sgt i32 [[TMP11]], 1
1599 // CHECK3-NEXT:    br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
1600 // CHECK3:       cond.true:
1601 // CHECK3-NEXT:    br label [[COND_END:%.*]]
1602 // CHECK3:       cond.false:
1603 // CHECK3-NEXT:    [[TMP12:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
1604 // CHECK3-NEXT:    br label [[COND_END]]
1605 // CHECK3:       cond.end:
1606 // CHECK3-NEXT:    [[COND:%.*]] = phi i32 [ 1, [[COND_TRUE]] ], [ [[TMP12]], [[COND_FALSE]] ]
1607 // CHECK3-NEXT:    store i32 [[COND]], i32* [[DOTOMP_UB]], align 4
1608 // CHECK3-NEXT:    [[TMP13:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4
1609 // CHECK3-NEXT:    store i32 [[TMP13]], i32* [[DOTOMP_IV]], align 4
1610 // CHECK3-NEXT:    br label [[OMP_INNER_FOR_COND:%.*]]
1611 // CHECK3:       omp.inner.for.cond:
1612 // CHECK3-NEXT:    [[TMP14:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
1613 // CHECK3-NEXT:    [[TMP15:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
1614 // CHECK3-NEXT:    [[CMP12:%.*]] = icmp sle i32 [[TMP14]], [[TMP15]]
1615 // CHECK3-NEXT:    br i1 [[CMP12]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
1616 // CHECK3:       omp.inner.for.body:
1617 // CHECK3-NEXT:    [[TMP16:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
1618 // CHECK3-NEXT:    [[MUL:%.*]] = mul nsw i32 [[TMP16]], 1
1619 // CHECK3-NEXT:    [[ADD:%.*]] = add nsw i32 0, [[MUL]]
1620 // CHECK3-NEXT:    store i32 [[ADD]], i32* [[I]], align 4
1621 // CHECK3-NEXT:    [[TMP17:%.*]] = load i32, i32* [[DOTLINEAR_START]], align 4
1622 // CHECK3-NEXT:    [[TMP18:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
1623 // CHECK3-NEXT:    [[MUL13:%.*]] = mul nsw i32 [[TMP18]], 1
1624 // CHECK3-NEXT:    [[ADD14:%.*]] = add nsw i32 [[TMP17]], [[MUL13]]
1625 // CHECK3-NEXT:    store i32 [[ADD14]], i32* [[A7]], align 4
1626 // CHECK3-NEXT:    [[TMP19:%.*]] = load i32, i32* [[DOTLINEAR_START5]], align 4
1627 // CHECK3-NEXT:    [[TMP20:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
1628 // CHECK3-NEXT:    [[MUL15:%.*]] = mul nsw i32 [[TMP20]], 1
1629 // CHECK3-NEXT:    [[ADD16:%.*]] = add nsw i32 [[TMP19]], [[MUL15]]
1630 // CHECK3-NEXT:    store i32 [[ADD16]], i32* [[B9]], align 4
1631 // CHECK3-NEXT:    [[TMP21:%.*]] = load i32, i32* [[DOTLINEAR_START6]], align 4
1632 // CHECK3-NEXT:    [[TMP22:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
1633 // CHECK3-NEXT:    [[MUL17:%.*]] = mul nsw i32 [[TMP22]], 1
1634 // CHECK3-NEXT:    [[ADD18:%.*]] = add nsw i32 [[TMP21]], [[MUL17]]
1635 // CHECK3-NEXT:    store i32 [[ADD18]], i32* [[C10]], align 4
1636 // CHECK3-NEXT:    [[TMP23:%.*]] = getelementptr inbounds [[CLASS_ANON_0]], %class.anon.0* [[REF_TMP]], i32 0, i32 0
1637 // CHECK3-NEXT:    store %struct.SS* [[TMP0]], %struct.SS** [[TMP23]], align 8
1638 // CHECK3-NEXT:    [[TMP24:%.*]] = getelementptr inbounds [[CLASS_ANON_0]], %class.anon.0* [[REF_TMP]], i32 0, i32 1
1639 // CHECK3-NEXT:    [[TMP25:%.*]] = load i32*, i32** [[_TMP8]], align 8
1640 // CHECK3-NEXT:    store i32* [[TMP25]], i32** [[TMP24]], align 8
1641 // CHECK3-NEXT:    [[TMP26:%.*]] = getelementptr inbounds [[CLASS_ANON_0]], %class.anon.0* [[REF_TMP]], i32 0, i32 2
1642 // CHECK3-NEXT:    store i32* [[B9]], i32** [[TMP26]], align 8
1643 // CHECK3-NEXT:    [[TMP27:%.*]] = getelementptr inbounds [[CLASS_ANON_0]], %class.anon.0* [[REF_TMP]], i32 0, i32 3
1644 // CHECK3-NEXT:    [[TMP28:%.*]] = load i32*, i32** [[_TMP11]], align 8
1645 // CHECK3-NEXT:    store i32* [[TMP28]], i32** [[TMP27]], align 8
1646 // CHECK3-NEXT:    call void @_ZZN2SSC1ERiENKUlvE_clEv(%class.anon.0* nonnull align 8 dereferenceable(32) [[REF_TMP]])
1647 // CHECK3-NEXT:    br label [[OMP_BODY_CONTINUE:%.*]]
1648 // CHECK3:       omp.body.continue:
1649 // CHECK3-NEXT:    br label [[OMP_INNER_FOR_INC:%.*]]
1650 // CHECK3:       omp.inner.for.inc:
1651 // CHECK3-NEXT:    [[TMP29:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
1652 // CHECK3-NEXT:    [[ADD19:%.*]] = add nsw i32 [[TMP29]], 1
1653 // CHECK3-NEXT:    store i32 [[ADD19]], i32* [[DOTOMP_IV]], align 4
1654 // CHECK3-NEXT:    br label [[OMP_INNER_FOR_COND]]
1655 // CHECK3:       omp.inner.for.end:
1656 // CHECK3-NEXT:    br label [[OMP_LOOP_EXIT:%.*]]
1657 // CHECK3:       omp.loop.exit:
1658 // CHECK3-NEXT:    call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB2]], i32 [[TMP10]])
1659 // CHECK3-NEXT:    [[TMP30:%.*]] = load i32, i32* [[DOTOMP_IS_LAST]], align 4
1660 // CHECK3-NEXT:    [[TMP31:%.*]] = icmp ne i32 [[TMP30]], 0
1661 // CHECK3-NEXT:    br i1 [[TMP31]], label [[DOTOMP_LINEAR_PU:%.*]], label [[DOTOMP_LINEAR_PU_DONE:%.*]]
1662 // CHECK3:       .omp.linear.pu:
1663 // CHECK3-NEXT:    [[TMP32:%.*]] = load i32*, i32** [[TMP]], align 8
1664 // CHECK3-NEXT:    store i32* [[TMP32]], i32** [[_TMP20]], align 8
1665 // CHECK3-NEXT:    [[TMP33:%.*]] = load i32, i32* [[A7]], align 4
1666 // CHECK3-NEXT:    [[TMP34:%.*]] = load i32*, i32** [[_TMP20]], align 8
1667 // CHECK3-NEXT:    store i32 [[TMP33]], i32* [[TMP34]], align 4
1668 // CHECK3-NEXT:    [[TMP35:%.*]] = load i32, i32* [[B9]], align 4
1669 // CHECK3-NEXT:    store i32 [[TMP35]], i32* [[B]], align 4
1670 // CHECK3-NEXT:    [[TMP36:%.*]] = load i32*, i32** [[_TMP3]], align 8
1671 // CHECK3-NEXT:    store i32* [[TMP36]], i32** [[_TMP21]], align 8
1672 // CHECK3-NEXT:    [[TMP37:%.*]] = load i32, i32* [[C10]], align 4
1673 // CHECK3-NEXT:    [[TMP38:%.*]] = load i32*, i32** [[_TMP21]], align 8
1674 // CHECK3-NEXT:    store i32 [[TMP37]], i32* [[TMP38]], align 4
1675 // CHECK3-NEXT:    [[TMP39:%.*]] = load i32, i32* [[B]], align 4
1676 // CHECK3-NEXT:    [[B22:%.*]] = getelementptr inbounds [[STRUCT_SS]], %struct.SS* [[TMP0]], i32 0, i32 1
1677 // CHECK3-NEXT:    [[TMP40:%.*]] = trunc i32 [[TMP39]] to i8
1678 // CHECK3-NEXT:    [[BF_LOAD:%.*]] = load i8, i8* [[B22]], align 4
1679 // CHECK3-NEXT:    [[BF_VALUE:%.*]] = and i8 [[TMP40]], 15
1680 // CHECK3-NEXT:    [[BF_CLEAR:%.*]] = and i8 [[BF_LOAD]], -16
1681 // CHECK3-NEXT:    [[BF_SET:%.*]] = or i8 [[BF_CLEAR]], [[BF_VALUE]]
1682 // CHECK3-NEXT:    store i8 [[BF_SET]], i8* [[B22]], align 4
1683 // CHECK3-NEXT:    br label [[DOTOMP_LINEAR_PU_DONE]]
1684 // CHECK3:       .omp.linear.pu.done:
1685 // CHECK3-NEXT:    call void @__kmpc_barrier(%struct.ident_t* @[[GLOB1]], i32 [[TMP10]])
1686 // CHECK3-NEXT:    ret void
1687 //
1688 //
1689 // CHECK3-LABEL: define {{[^@]+}}@_ZZN2SSC1ERiENKUlvE_clEv
1690 // CHECK3-SAME: (%class.anon.0* nonnull align 8 dereferenceable(32) [[THIS:%.*]]) #[[ATTR2:[0-9]+]] align 2 {
1691 // CHECK3-NEXT:  entry:
1692 // CHECK3-NEXT:    [[THIS_ADDR:%.*]] = alloca %class.anon.0*, align 8
1693 // CHECK3-NEXT:    store %class.anon.0* [[THIS]], %class.anon.0** [[THIS_ADDR]], align 8
1694 // CHECK3-NEXT:    [[THIS1:%.*]] = load %class.anon.0*, %class.anon.0** [[THIS_ADDR]], align 8
1695 // CHECK3-NEXT:    [[TMP0:%.*]] = getelementptr inbounds [[CLASS_ANON_0:%.*]], %class.anon.0* [[THIS1]], i32 0, i32 0
1696 // CHECK3-NEXT:    [[TMP1:%.*]] = load %struct.SS*, %struct.SS** [[TMP0]], align 8
1697 // CHECK3-NEXT:    [[TMP2:%.*]] = getelementptr inbounds [[CLASS_ANON_0]], %class.anon.0* [[THIS1]], i32 0, i32 1
1698 // CHECK3-NEXT:    [[TMP3:%.*]] = load i32*, i32** [[TMP2]], align 8
1699 // CHECK3-NEXT:    [[TMP4:%.*]] = load i32, i32* [[TMP3]], align 4
1700 // CHECK3-NEXT:    [[INC:%.*]] = add nsw i32 [[TMP4]], 1
1701 // CHECK3-NEXT:    store i32 [[INC]], i32* [[TMP3]], align 4
1702 // CHECK3-NEXT:    [[TMP5:%.*]] = getelementptr inbounds [[CLASS_ANON_0]], %class.anon.0* [[THIS1]], i32 0, i32 2
1703 // CHECK3-NEXT:    [[TMP6:%.*]] = load i32*, i32** [[TMP5]], align 8
1704 // CHECK3-NEXT:    [[TMP7:%.*]] = load i32, i32* [[TMP6]], align 4
1705 // CHECK3-NEXT:    [[DEC:%.*]] = add nsw i32 [[TMP7]], -1
1706 // CHECK3-NEXT:    store i32 [[DEC]], i32* [[TMP6]], align 4
1707 // CHECK3-NEXT:    [[TMP8:%.*]] = getelementptr inbounds [[CLASS_ANON_0]], %class.anon.0* [[THIS1]], i32 0, i32 3
1708 // CHECK3-NEXT:    [[TMP9:%.*]] = load i32*, i32** [[TMP8]], align 8
1709 // CHECK3-NEXT:    [[TMP10:%.*]] = load i32, i32* [[TMP9]], align 4
1710 // CHECK3-NEXT:    [[DIV:%.*]] = sdiv i32 [[TMP10]], 1
1711 // CHECK3-NEXT:    store i32 [[DIV]], i32* [[TMP9]], align 4
1712 // CHECK3-NEXT:    [[TMP11:%.*]] = getelementptr inbounds [[CLASS_ANON_0]], %class.anon.0* [[THIS1]], i32 0, i32 1
1713 // CHECK3-NEXT:    [[TMP12:%.*]] = load i32*, i32** [[TMP11]], align 8
1714 // CHECK3-NEXT:    [[TMP13:%.*]] = getelementptr inbounds [[CLASS_ANON_0]], %class.anon.0* [[THIS1]], i32 0, i32 2
1715 // CHECK3-NEXT:    [[TMP14:%.*]] = load i32*, i32** [[TMP13]], align 8
1716 // CHECK3-NEXT:    [[TMP15:%.*]] = getelementptr inbounds [[CLASS_ANON_0]], %class.anon.0* [[THIS1]], i32 0, i32 3
1717 // CHECK3-NEXT:    [[TMP16:%.*]] = load i32*, i32** [[TMP15]], align 8
1718 // CHECK3-NEXT:    call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB3]], i32 4, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, %struct.SS*, i32*, i32*, i32*)* @.omp_outlined..1 to void (i32*, i32*, ...)*), %struct.SS* [[TMP1]], i32* [[TMP12]], i32* [[TMP14]], i32* [[TMP16]])
1719 // CHECK3-NEXT:    ret void
1720 //
1721 //
1722 // CHECK3-LABEL: define {{[^@]+}}@.omp_outlined..1
1723 // CHECK3-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], %struct.SS* [[THIS:%.*]], i32* nonnull align 4 dereferenceable(4) [[A:%.*]], i32* nonnull align 4 dereferenceable(4) [[B:%.*]], i32* nonnull align 4 dereferenceable(4) [[C:%.*]]) #[[ATTR3]] {
1724 // CHECK3-NEXT:  entry:
1725 // CHECK3-NEXT:    [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
1726 // CHECK3-NEXT:    [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
1727 // CHECK3-NEXT:    [[THIS_ADDR:%.*]] = alloca %struct.SS*, align 8
1728 // CHECK3-NEXT:    [[A_ADDR:%.*]] = alloca i32*, align 8
1729 // CHECK3-NEXT:    [[B_ADDR:%.*]] = alloca i32*, align 8
1730 // CHECK3-NEXT:    [[C_ADDR:%.*]] = alloca i32*, align 8
1731 // CHECK3-NEXT:    [[TMP:%.*]] = alloca i32*, align 8
1732 // CHECK3-NEXT:    [[_TMP1:%.*]] = alloca i32*, align 8
1733 // CHECK3-NEXT:    [[_TMP2:%.*]] = alloca i32*, align 8
1734 // CHECK3-NEXT:    [[_TMP3:%.*]] = alloca i32*, align 8
1735 // CHECK3-NEXT:    [[DOTOMP_IV:%.*]] = alloca i32, align 4
1736 // CHECK3-NEXT:    [[_TMP4:%.*]] = alloca i32, align 4
1737 // CHECK3-NEXT:    [[DOTLINEAR_START:%.*]] = alloca i32, align 4
1738 // CHECK3-NEXT:    [[DOTLINEAR_START5:%.*]] = alloca i32, align 4
1739 // CHECK3-NEXT:    [[DOTLINEAR_START6:%.*]] = alloca i32, align 4
1740 // CHECK3-NEXT:    [[DOTOMP_LB:%.*]] = alloca i32, align 4
1741 // CHECK3-NEXT:    [[DOTOMP_UB:%.*]] = alloca i32, align 4
1742 // CHECK3-NEXT:    [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
1743 // CHECK3-NEXT:    [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
1744 // CHECK3-NEXT:    [[I:%.*]] = alloca i32, align 4
1745 // CHECK3-NEXT:    [[A7:%.*]] = alloca i32, align 4
1746 // CHECK3-NEXT:    [[_TMP8:%.*]] = alloca i32*, align 8
1747 // CHECK3-NEXT:    [[B9:%.*]] = alloca i32, align 4
1748 // CHECK3-NEXT:    [[C10:%.*]] = alloca i32, align 4
1749 // CHECK3-NEXT:    [[_TMP11:%.*]] = alloca i32*, align 8
1750 // CHECK3-NEXT:    [[_TMP20:%.*]] = alloca i32*, align 8
1751 // CHECK3-NEXT:    [[_TMP21:%.*]] = alloca i32*, align 8
1752 // CHECK3-NEXT:    store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
1753 // CHECK3-NEXT:    store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
1754 // CHECK3-NEXT:    store %struct.SS* [[THIS]], %struct.SS** [[THIS_ADDR]], align 8
1755 // CHECK3-NEXT:    store i32* [[A]], i32** [[A_ADDR]], align 8
1756 // CHECK3-NEXT:    store i32* [[B]], i32** [[B_ADDR]], align 8
1757 // CHECK3-NEXT:    store i32* [[C]], i32** [[C_ADDR]], align 8
1758 // CHECK3-NEXT:    [[TMP0:%.*]] = load %struct.SS*, %struct.SS** [[THIS_ADDR]], align 8
1759 // CHECK3-NEXT:    [[TMP1:%.*]] = load i32*, i32** [[A_ADDR]], align 8
1760 // CHECK3-NEXT:    [[TMP2:%.*]] = load i32*, i32** [[B_ADDR]], align 8
1761 // CHECK3-NEXT:    [[TMP3:%.*]] = load i32*, i32** [[C_ADDR]], align 8
1762 // CHECK3-NEXT:    store i32* [[TMP1]], i32** [[TMP]], align 8
1763 // CHECK3-NEXT:    store i32* [[TMP3]], i32** [[_TMP1]], align 8
1764 // CHECK3-NEXT:    [[TMP4:%.*]] = load i32*, i32** [[TMP]], align 8
1765 // CHECK3-NEXT:    store i32* [[TMP4]], i32** [[_TMP2]], align 8
1766 // CHECK3-NEXT:    [[TMP5:%.*]] = load i32*, i32** [[_TMP1]], align 8
1767 // CHECK3-NEXT:    store i32* [[TMP5]], i32** [[_TMP3]], align 8
1768 // CHECK3-NEXT:    [[TMP6:%.*]] = load i32*, i32** [[_TMP2]], align 8
1769 // CHECK3-NEXT:    [[TMP7:%.*]] = load i32, i32* [[TMP6]], align 4
1770 // CHECK3-NEXT:    store i32 [[TMP7]], i32* [[DOTLINEAR_START]], align 4
1771 // CHECK3-NEXT:    [[TMP8:%.*]] = load i32, i32* [[TMP2]], align 4
1772 // CHECK3-NEXT:    store i32 [[TMP8]], i32* [[DOTLINEAR_START5]], align 4
1773 // CHECK3-NEXT:    [[TMP9:%.*]] = load i32*, i32** [[_TMP3]], align 8
1774 // CHECK3-NEXT:    [[TMP10:%.*]] = load i32, i32* [[TMP9]], align 4
1775 // CHECK3-NEXT:    store i32 [[TMP10]], i32* [[DOTLINEAR_START6]], align 4
1776 // CHECK3-NEXT:    store i32 0, i32* [[DOTOMP_LB]], align 4
1777 // CHECK3-NEXT:    store i32 1, i32* [[DOTOMP_UB]], align 4
1778 // CHECK3-NEXT:    store i32 1, i32* [[DOTOMP_STRIDE]], align 4
1779 // CHECK3-NEXT:    store i32 0, i32* [[DOTOMP_IS_LAST]], align 4
1780 // CHECK3-NEXT:    [[TMP11:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
1781 // CHECK3-NEXT:    [[TMP12:%.*]] = load i32, i32* [[TMP11]], align 4
1782 // CHECK3-NEXT:    call void @__kmpc_barrier(%struct.ident_t* @[[GLOB1]], i32 [[TMP12]])
1783 // CHECK3-NEXT:    store i32* [[A7]], i32** [[_TMP8]], align 8
1784 // CHECK3-NEXT:    store i32* [[C10]], i32** [[_TMP11]], align 8
1785 // CHECK3-NEXT:    call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB2]], i32 [[TMP12]], i32 34, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_LB]], i32* [[DOTOMP_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1)
1786 // CHECK3-NEXT:    [[TMP13:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
1787 // CHECK3-NEXT:    [[CMP:%.*]] = icmp sgt i32 [[TMP13]], 1
1788 // CHECK3-NEXT:    br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
1789 // CHECK3:       cond.true:
1790 // CHECK3-NEXT:    br label [[COND_END:%.*]]
1791 // CHECK3:       cond.false:
1792 // CHECK3-NEXT:    [[TMP14:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
1793 // CHECK3-NEXT:    br label [[COND_END]]
1794 // CHECK3:       cond.end:
1795 // CHECK3-NEXT:    [[COND:%.*]] = phi i32 [ 1, [[COND_TRUE]] ], [ [[TMP14]], [[COND_FALSE]] ]
1796 // CHECK3-NEXT:    store i32 [[COND]], i32* [[DOTOMP_UB]], align 4
1797 // CHECK3-NEXT:    [[TMP15:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4
1798 // CHECK3-NEXT:    store i32 [[TMP15]], i32* [[DOTOMP_IV]], align 4
1799 // CHECK3-NEXT:    br label [[OMP_INNER_FOR_COND:%.*]]
1800 // CHECK3:       omp.inner.for.cond:
1801 // CHECK3-NEXT:    [[TMP16:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
1802 // CHECK3-NEXT:    [[TMP17:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
1803 // CHECK3-NEXT:    [[CMP12:%.*]] = icmp sle i32 [[TMP16]], [[TMP17]]
1804 // CHECK3-NEXT:    br i1 [[CMP12]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
1805 // CHECK3:       omp.inner.for.body:
1806 // CHECK3-NEXT:    [[TMP18:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
1807 // CHECK3-NEXT:    [[MUL:%.*]] = mul nsw i32 [[TMP18]], 1
1808 // CHECK3-NEXT:    [[ADD:%.*]] = add nsw i32 0, [[MUL]]
1809 // CHECK3-NEXT:    store i32 [[ADD]], i32* [[I]], align 4
1810 // CHECK3-NEXT:    [[TMP19:%.*]] = load i32, i32* [[DOTLINEAR_START]], align 4
1811 // CHECK3-NEXT:    [[TMP20:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
1812 // CHECK3-NEXT:    [[MUL13:%.*]] = mul nsw i32 [[TMP20]], 1
1813 // CHECK3-NEXT:    [[ADD14:%.*]] = add nsw i32 [[TMP19]], [[MUL13]]
1814 // CHECK3-NEXT:    store i32 [[ADD14]], i32* [[A7]], align 4
1815 // CHECK3-NEXT:    [[TMP21:%.*]] = load i32, i32* [[DOTLINEAR_START5]], align 4
1816 // CHECK3-NEXT:    [[TMP22:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
1817 // CHECK3-NEXT:    [[MUL15:%.*]] = mul nsw i32 [[TMP22]], 1
1818 // CHECK3-NEXT:    [[ADD16:%.*]] = add nsw i32 [[TMP21]], [[MUL15]]
1819 // CHECK3-NEXT:    store i32 [[ADD16]], i32* [[B9]], align 4
1820 // CHECK3-NEXT:    [[TMP23:%.*]] = load i32, i32* [[DOTLINEAR_START6]], align 4
1821 // CHECK3-NEXT:    [[TMP24:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
1822 // CHECK3-NEXT:    [[MUL17:%.*]] = mul nsw i32 [[TMP24]], 1
1823 // CHECK3-NEXT:    [[ADD18:%.*]] = add nsw i32 [[TMP23]], [[MUL17]]
1824 // CHECK3-NEXT:    store i32 [[ADD18]], i32* [[C10]], align 4
1825 // CHECK3-NEXT:    [[TMP25:%.*]] = load i32*, i32** [[_TMP8]], align 8
1826 // CHECK3-NEXT:    [[TMP26:%.*]] = load i32, i32* [[TMP25]], align 4
1827 // CHECK3-NEXT:    [[INC:%.*]] = add nsw i32 [[TMP26]], 1
1828 // CHECK3-NEXT:    store i32 [[INC]], i32* [[TMP25]], align 4
1829 // CHECK3-NEXT:    [[TMP27:%.*]] = load i32, i32* [[B9]], align 4
1830 // CHECK3-NEXT:    [[DEC:%.*]] = add nsw i32 [[TMP27]], -1
1831 // CHECK3-NEXT:    store i32 [[DEC]], i32* [[B9]], align 4
1832 // CHECK3-NEXT:    [[TMP28:%.*]] = load i32*, i32** [[_TMP11]], align 8
1833 // CHECK3-NEXT:    [[TMP29:%.*]] = load i32, i32* [[TMP28]], align 4
1834 // CHECK3-NEXT:    [[DIV:%.*]] = sdiv i32 [[TMP29]], 1
1835 // CHECK3-NEXT:    store i32 [[DIV]], i32* [[TMP28]], align 4
1836 // CHECK3-NEXT:    br label [[OMP_BODY_CONTINUE:%.*]]
1837 // CHECK3:       omp.body.continue:
1838 // CHECK3-NEXT:    br label [[OMP_INNER_FOR_INC:%.*]]
1839 // CHECK3:       omp.inner.for.inc:
1840 // CHECK3-NEXT:    [[TMP30:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
1841 // CHECK3-NEXT:    [[ADD19:%.*]] = add nsw i32 [[TMP30]], 1
1842 // CHECK3-NEXT:    store i32 [[ADD19]], i32* [[DOTOMP_IV]], align 4
1843 // CHECK3-NEXT:    br label [[OMP_INNER_FOR_COND]]
1844 // CHECK3:       omp.inner.for.end:
1845 // CHECK3-NEXT:    br label [[OMP_LOOP_EXIT:%.*]]
1846 // CHECK3:       omp.loop.exit:
1847 // CHECK3-NEXT:    call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB2]], i32 [[TMP12]])
1848 // CHECK3-NEXT:    [[TMP31:%.*]] = load i32, i32* [[DOTOMP_IS_LAST]], align 4
1849 // CHECK3-NEXT:    [[TMP32:%.*]] = icmp ne i32 [[TMP31]], 0
1850 // CHECK3-NEXT:    br i1 [[TMP32]], label [[DOTOMP_LINEAR_PU:%.*]], label [[DOTOMP_LINEAR_PU_DONE:%.*]]
1851 // CHECK3:       .omp.linear.pu:
1852 // CHECK3-NEXT:    [[TMP33:%.*]] = load i32*, i32** [[_TMP2]], align 8
1853 // CHECK3-NEXT:    store i32* [[TMP33]], i32** [[_TMP20]], align 8
1854 // CHECK3-NEXT:    [[TMP34:%.*]] = load i32, i32* [[A7]], align 4
1855 // CHECK3-NEXT:    [[TMP35:%.*]] = load i32*, i32** [[_TMP20]], align 8
1856 // CHECK3-NEXT:    store i32 [[TMP34]], i32* [[TMP35]], align 4
1857 // CHECK3-NEXT:    [[TMP36:%.*]] = load i32, i32* [[B9]], align 4
1858 // CHECK3-NEXT:    store i32 [[TMP36]], i32* [[TMP2]], align 4
1859 // CHECK3-NEXT:    [[TMP37:%.*]] = load i32*, i32** [[_TMP3]], align 8
1860 // CHECK3-NEXT:    store i32* [[TMP37]], i32** [[_TMP21]], align 8
1861 // CHECK3-NEXT:    [[TMP38:%.*]] = load i32, i32* [[C10]], align 4
1862 // CHECK3-NEXT:    [[TMP39:%.*]] = load i32*, i32** [[_TMP21]], align 8
1863 // CHECK3-NEXT:    store i32 [[TMP38]], i32* [[TMP39]], align 4
1864 // CHECK3-NEXT:    br label [[DOTOMP_LINEAR_PU_DONE]]
1865 // CHECK3:       .omp.linear.pu.done:
1866 // CHECK3-NEXT:    call void @__kmpc_barrier(%struct.ident_t* @[[GLOB1]], i32 [[TMP12]])
1867 // CHECK3-NEXT:    ret void
1868 //
1869 //
1870 // CHECK3-LABEL: define {{[^@]+}}@.omp_outlined..2
1871 // CHECK3-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]]) #[[ATTR3]] {
1872 // CHECK3-NEXT:  entry:
1873 // CHECK3-NEXT:    [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
1874 // CHECK3-NEXT:    [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
1875 // CHECK3-NEXT:    [[TMP:%.*]] = alloca i32*, align 8
1876 // CHECK3-NEXT:    [[DOTOMP_IV:%.*]] = alloca i32, align 4
1877 // CHECK3-NEXT:    [[_TMP1:%.*]] = alloca i32, align 4
1878 // CHECK3-NEXT:    [[DOTLINEAR_START:%.*]] = alloca i32, align 4
1879 // CHECK3-NEXT:    [[DOTLINEAR_START2:%.*]] = alloca i32, align 4
1880 // CHECK3-NEXT:    [[DOTOMP_LB:%.*]] = alloca i32, align 4
1881 // CHECK3-NEXT:    [[DOTOMP_UB:%.*]] = alloca i32, align 4
1882 // CHECK3-NEXT:    [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
1883 // CHECK3-NEXT:    [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
1884 // CHECK3-NEXT:    [[I:%.*]] = alloca i32, align 4
1885 // CHECK3-NEXT:    [[G:%.*]] = alloca i32, align 4
1886 // CHECK3-NEXT:    [[G1:%.*]] = alloca i32, align 4
1887 // CHECK3-NEXT:    [[_TMP3:%.*]] = alloca i32*, align 8
1888 // CHECK3-NEXT:    [[REF_TMP:%.*]] = alloca [[CLASS_ANON_1:%.*]], align 8
1889 // CHECK3-NEXT:    [[_TMP12:%.*]] = alloca i32*, align 8
1890 // CHECK3-NEXT:    store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
1891 // CHECK3-NEXT:    store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
1892 // CHECK3-NEXT:    [[TMP0:%.*]] = load i32*, i32** @g1, align 8
1893 // CHECK3-NEXT:    store i32* [[TMP0]], i32** [[TMP]], align 8
1894 // CHECK3-NEXT:    [[TMP1:%.*]] = load i32, i32* @g, align 4
1895 // CHECK3-NEXT:    store i32 [[TMP1]], i32* [[DOTLINEAR_START]], align 4
1896 // CHECK3-NEXT:    [[TMP2:%.*]] = load i32, i32* @g, align 4
1897 // CHECK3-NEXT:    store i32 [[TMP2]], i32* [[DOTLINEAR_START2]], align 4
1898 // CHECK3-NEXT:    store i32 0, i32* [[DOTOMP_LB]], align 4
1899 // CHECK3-NEXT:    store i32 1, i32* [[DOTOMP_UB]], align 4
1900 // CHECK3-NEXT:    store i32 1, i32* [[DOTOMP_STRIDE]], align 4
1901 // CHECK3-NEXT:    store i32 0, i32* [[DOTOMP_IS_LAST]], align 4
1902 // CHECK3-NEXT:    [[TMP3:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
1903 // CHECK3-NEXT:    [[TMP4:%.*]] = load i32, i32* [[TMP3]], align 4
1904 // CHECK3-NEXT:    call void @__kmpc_barrier(%struct.ident_t* @[[GLOB1]], i32 [[TMP4]])
1905 // CHECK3-NEXT:    store i32* [[G1]], i32** [[_TMP3]], align 8
1906 // CHECK3-NEXT:    call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB2]], i32 [[TMP4]], i32 34, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_LB]], i32* [[DOTOMP_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1)
1907 // CHECK3-NEXT:    [[TMP5:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
1908 // CHECK3-NEXT:    [[CMP:%.*]] = icmp sgt i32 [[TMP5]], 1
1909 // CHECK3-NEXT:    br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
1910 // CHECK3:       cond.true:
1911 // CHECK3-NEXT:    br label [[COND_END:%.*]]
1912 // CHECK3:       cond.false:
1913 // CHECK3-NEXT:    [[TMP6:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
1914 // CHECK3-NEXT:    br label [[COND_END]]
1915 // CHECK3:       cond.end:
1916 // CHECK3-NEXT:    [[COND:%.*]] = phi i32 [ 1, [[COND_TRUE]] ], [ [[TMP6]], [[COND_FALSE]] ]
1917 // CHECK3-NEXT:    store i32 [[COND]], i32* [[DOTOMP_UB]], align 4
1918 // CHECK3-NEXT:    [[TMP7:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4
1919 // CHECK3-NEXT:    store i32 [[TMP7]], i32* [[DOTOMP_IV]], align 4
1920 // CHECK3-NEXT:    br label [[OMP_INNER_FOR_COND:%.*]]
1921 // CHECK3:       omp.inner.for.cond:
1922 // CHECK3-NEXT:    [[TMP8:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
1923 // CHECK3-NEXT:    [[TMP9:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
1924 // CHECK3-NEXT:    [[CMP4:%.*]] = icmp sle i32 [[TMP8]], [[TMP9]]
1925 // CHECK3-NEXT:    br i1 [[CMP4]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
1926 // CHECK3:       omp.inner.for.body:
1927 // CHECK3-NEXT:    [[TMP10:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
1928 // CHECK3-NEXT:    [[MUL:%.*]] = mul nsw i32 [[TMP10]], 1
1929 // CHECK3-NEXT:    [[ADD:%.*]] = add nsw i32 0, [[MUL]]
1930 // CHECK3-NEXT:    store i32 [[ADD]], i32* [[I]], align 4
1931 // CHECK3-NEXT:    [[TMP11:%.*]] = load i32, i32* [[DOTLINEAR_START]], align 4
1932 // CHECK3-NEXT:    [[TMP12:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
1933 // CHECK3-NEXT:    [[MUL5:%.*]] = mul nsw i32 [[TMP12]], 5
1934 // CHECK3-NEXT:    [[ADD6:%.*]] = add nsw i32 [[TMP11]], [[MUL5]]
1935 // CHECK3-NEXT:    store i32 [[ADD6]], i32* [[G]], align 4
1936 // CHECK3-NEXT:    [[TMP13:%.*]] = load i32, i32* [[DOTLINEAR_START2]], align 4
1937 // CHECK3-NEXT:    [[TMP14:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
1938 // CHECK3-NEXT:    [[MUL7:%.*]] = mul nsw i32 [[TMP14]], 5
1939 // CHECK3-NEXT:    [[ADD8:%.*]] = add nsw i32 [[TMP13]], [[MUL7]]
1940 // CHECK3-NEXT:    store i32 [[ADD8]], i32* [[G1]], align 4
1941 // CHECK3-NEXT:    [[TMP15:%.*]] = load i32, i32* [[G]], align 4
1942 // CHECK3-NEXT:    [[ADD9:%.*]] = add nsw i32 [[TMP15]], 5
1943 // CHECK3-NEXT:    store i32 [[ADD9]], i32* [[G]], align 4
1944 // CHECK3-NEXT:    [[TMP16:%.*]] = load i32*, i32** [[_TMP3]], align 8
1945 // CHECK3-NEXT:    [[TMP17:%.*]] = load volatile i32, i32* [[TMP16]], align 4
1946 // CHECK3-NEXT:    [[ADD10:%.*]] = add nsw i32 [[TMP17]], 5
1947 // CHECK3-NEXT:    store volatile i32 [[ADD10]], i32* [[TMP16]], align 4
1948 // CHECK3-NEXT:    [[TMP18:%.*]] = getelementptr inbounds [[CLASS_ANON_1]], %class.anon.1* [[REF_TMP]], i32 0, i32 0
1949 // CHECK3-NEXT:    store i32* [[G]], i32** [[TMP18]], align 8
1950 // CHECK3-NEXT:    [[TMP19:%.*]] = getelementptr inbounds [[CLASS_ANON_1]], %class.anon.1* [[REF_TMP]], i32 0, i32 1
1951 // CHECK3-NEXT:    [[TMP20:%.*]] = load i32*, i32** [[_TMP3]], align 8
1952 // CHECK3-NEXT:    store i32* [[TMP20]], i32** [[TMP19]], align 8
1953 // CHECK3-NEXT:    call void @"_ZZZ4mainENK3$_0clEvENKUlvE_clEv"(%class.anon.1* nonnull align 8 dereferenceable(16) [[REF_TMP]])
1954 // CHECK3-NEXT:    br label [[OMP_BODY_CONTINUE:%.*]]
1955 // CHECK3:       omp.body.continue:
1956 // CHECK3-NEXT:    br label [[OMP_INNER_FOR_INC:%.*]]
1957 // CHECK3:       omp.inner.for.inc:
1958 // CHECK3-NEXT:    [[TMP21:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
1959 // CHECK3-NEXT:    [[ADD11:%.*]] = add nsw i32 [[TMP21]], 1
1960 // CHECK3-NEXT:    store i32 [[ADD11]], i32* [[DOTOMP_IV]], align 4
1961 // CHECK3-NEXT:    br label [[OMP_INNER_FOR_COND]]
1962 // CHECK3:       omp.inner.for.end:
1963 // CHECK3-NEXT:    br label [[OMP_LOOP_EXIT:%.*]]
1964 // CHECK3:       omp.loop.exit:
1965 // CHECK3-NEXT:    call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB2]], i32 [[TMP4]])
1966 // CHECK3-NEXT:    [[TMP22:%.*]] = load i32, i32* [[DOTOMP_IS_LAST]], align 4
1967 // CHECK3-NEXT:    [[TMP23:%.*]] = icmp ne i32 [[TMP22]], 0
1968 // CHECK3-NEXT:    br i1 [[TMP23]], label [[DOTOMP_LINEAR_PU:%.*]], label [[DOTOMP_LINEAR_PU_DONE:%.*]]
1969 // CHECK3:       .omp.linear.pu:
1970 // CHECK3-NEXT:    [[TMP24:%.*]] = load i32, i32* [[G]], align 4
1971 // CHECK3-NEXT:    store i32 [[TMP24]], i32* @g, align 4
1972 // CHECK3-NEXT:    [[TMP25:%.*]] = load i32*, i32** @g1, align 8
1973 // CHECK3-NEXT:    store i32* [[TMP25]], i32** [[_TMP12]], align 8
1974 // CHECK3-NEXT:    [[TMP26:%.*]] = load i32, i32* [[G1]], align 4
1975 // CHECK3-NEXT:    [[TMP27:%.*]] = load i32*, i32** [[_TMP12]], align 8
1976 // CHECK3-NEXT:    store volatile i32 [[TMP26]], i32* [[TMP27]], align 4
1977 // CHECK3-NEXT:    br label [[DOTOMP_LINEAR_PU_DONE]]
1978 // CHECK3:       .omp.linear.pu.done:
1979 // CHECK3-NEXT:    call void @__kmpc_barrier(%struct.ident_t* @[[GLOB1]], i32 [[TMP4]])
1980 // CHECK3-NEXT:    ret void
1981 //
1982 //
1983 // CHECK4-LABEL: define {{[^@]+}}@main
1984 // CHECK4-SAME: () #[[ATTR1:[0-9]+]] {
1985 // CHECK4-NEXT:  entry:
1986 // CHECK4-NEXT:    [[RETVAL:%.*]] = alloca i32, align 4
1987 // CHECK4-NEXT:    [[SS:%.*]] = alloca [[STRUCT_SS:%.*]], align 8
1988 // CHECK4-NEXT:    store i32 0, i32* [[RETVAL]], align 4
1989 // CHECK4-NEXT:    call void @_ZN2SSC1ERi(%struct.SS* nonnull align 8 dereferenceable(16) [[SS]], i32* nonnull align 4 dereferenceable(4) @_ZZ4mainE5sivar)
1990 // 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
1991 // CHECK4-NEXT:    [[TMP1:%.*]] = bitcast i8* [[TMP0]] to void (i8*)*
1992 // CHECK4-NEXT:    call void [[TMP1]](i8* bitcast ({ i8**, i32, i32, i8*, %struct.__block_descriptor* }* @__block_literal_global to i8*))
1993 // CHECK4-NEXT:    ret i32 0
1994 //
1995 //
1996 // CHECK4-LABEL: define {{[^@]+}}@_ZN2SSC1ERi
1997 // CHECK4-SAME: (%struct.SS* nonnull align 8 dereferenceable(16) [[THIS:%.*]], i32* nonnull align 4 dereferenceable(4) [[D:%.*]]) unnamed_addr #[[ATTR2:[0-9]+]] align 2 {
1998 // CHECK4-NEXT:  entry:
1999 // CHECK4-NEXT:    [[THIS_ADDR:%.*]] = alloca %struct.SS*, align 8
2000 // CHECK4-NEXT:    [[D_ADDR:%.*]] = alloca i32*, align 8
2001 // CHECK4-NEXT:    store %struct.SS* [[THIS]], %struct.SS** [[THIS_ADDR]], align 8
2002 // CHECK4-NEXT:    store i32* [[D]], i32** [[D_ADDR]], align 8
2003 // CHECK4-NEXT:    [[THIS1:%.*]] = load %struct.SS*, %struct.SS** [[THIS_ADDR]], align 8
2004 // CHECK4-NEXT:    [[TMP0:%.*]] = load i32*, i32** [[D_ADDR]], align 8
2005 // CHECK4-NEXT:    call void @_ZN2SSC2ERi(%struct.SS* nonnull align 8 dereferenceable(16) [[THIS1]], i32* nonnull align 4 dereferenceable(4) [[TMP0]])
2006 // CHECK4-NEXT:    ret void
2007 //
2008 //
2009 // CHECK4-LABEL: define {{[^@]+}}@__main_block_invoke
2010 // CHECK4-SAME: (i8* [[DOTBLOCK_DESCRIPTOR:%.*]]) #[[ATTR2]] {
2011 // CHECK4-NEXT:  entry:
2012 // CHECK4-NEXT:    [[DOTBLOCK_DESCRIPTOR_ADDR:%.*]] = alloca i8*, align 8
2013 // CHECK4-NEXT:    [[BLOCK_ADDR:%.*]] = alloca <{ i8*, i32, i32, i8*, %struct.__block_descriptor* }>*, align 8
2014 // CHECK4-NEXT:    store i8* [[DOTBLOCK_DESCRIPTOR]], i8** [[DOTBLOCK_DESCRIPTOR_ADDR]], align 8
2015 // CHECK4-NEXT:    [[BLOCK:%.*]] = bitcast i8* [[DOTBLOCK_DESCRIPTOR]] to <{ i8*, i32, i32, i8*, %struct.__block_descriptor* }>*
2016 // CHECK4-NEXT:    store <{ i8*, i32, i32, i8*, %struct.__block_descriptor* }>* [[BLOCK]], <{ i8*, i32, i32, i8*, %struct.__block_descriptor* }>** [[BLOCK_ADDR]], align 8
2017 // CHECK4-NEXT:    call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB3:[0-9]+]], i32 0, void (i32*, i32*, ...)* bitcast (void (i32*, i32*)* @.omp_outlined. to void (i32*, i32*, ...)*))
2018 // CHECK4-NEXT:    ret void
2019 //
2020 //
2021 // CHECK4-LABEL: define {{[^@]+}}@.omp_outlined.
2022 // CHECK4-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]]) #[[ATTR3:[0-9]+]] {
2023 // CHECK4-NEXT:  entry:
2024 // CHECK4-NEXT:    [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
2025 // CHECK4-NEXT:    [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
2026 // CHECK4-NEXT:    [[TMP:%.*]] = alloca i32*, align 8
2027 // CHECK4-NEXT:    [[DOTOMP_IV:%.*]] = alloca i32, align 4
2028 // CHECK4-NEXT:    [[_TMP1:%.*]] = alloca i32, align 4
2029 // CHECK4-NEXT:    [[DOTLINEAR_START:%.*]] = alloca i32, align 4
2030 // CHECK4-NEXT:    [[DOTLINEAR_START2:%.*]] = alloca i32, align 4
2031 // CHECK4-NEXT:    [[DOTOMP_LB:%.*]] = alloca i32, align 4
2032 // CHECK4-NEXT:    [[DOTOMP_UB:%.*]] = alloca i32, align 4
2033 // CHECK4-NEXT:    [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
2034 // CHECK4-NEXT:    [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
2035 // CHECK4-NEXT:    [[I:%.*]] = alloca i32, align 4
2036 // CHECK4-NEXT:    [[G:%.*]] = alloca i32, align 4
2037 // CHECK4-NEXT:    [[G1:%.*]] = alloca i32, align 4
2038 // CHECK4-NEXT:    [[_TMP3:%.*]] = alloca i32*, align 8
2039 // CHECK4-NEXT:    [[BLOCK:%.*]] = alloca <{ i8*, i32, i32, i8*, %struct.__block_descriptor*, i32*, i32 }>, align 8
2040 // CHECK4-NEXT:    [[_TMP13:%.*]] = alloca i32*, align 8
2041 // CHECK4-NEXT:    store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
2042 // CHECK4-NEXT:    store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
2043 // CHECK4-NEXT:    [[TMP0:%.*]] = load i32*, i32** @g1, align 8
2044 // CHECK4-NEXT:    store i32* [[TMP0]], i32** [[TMP]], align 8
2045 // CHECK4-NEXT:    [[TMP1:%.*]] = load i32, i32* @g, align 4
2046 // CHECK4-NEXT:    store i32 [[TMP1]], i32* [[DOTLINEAR_START]], align 4
2047 // CHECK4-NEXT:    [[TMP2:%.*]] = load i32, i32* @g, align 4
2048 // CHECK4-NEXT:    store i32 [[TMP2]], i32* [[DOTLINEAR_START2]], align 4
2049 // CHECK4-NEXT:    store i32 0, i32* [[DOTOMP_LB]], align 4
2050 // CHECK4-NEXT:    store i32 1, i32* [[DOTOMP_UB]], align 4
2051 // CHECK4-NEXT:    store i32 1, i32* [[DOTOMP_STRIDE]], align 4
2052 // CHECK4-NEXT:    store i32 0, i32* [[DOTOMP_IS_LAST]], align 4
2053 // CHECK4-NEXT:    [[TMP3:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
2054 // CHECK4-NEXT:    [[TMP4:%.*]] = load i32, i32* [[TMP3]], align 4
2055 // CHECK4-NEXT:    call void @__kmpc_barrier(%struct.ident_t* @[[GLOB1:[0-9]+]], i32 [[TMP4]])
2056 // CHECK4-NEXT:    store i32* [[G1]], i32** [[_TMP3]], align 8
2057 // CHECK4-NEXT:    call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB2:[0-9]+]], i32 [[TMP4]], i32 34, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_LB]], i32* [[DOTOMP_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1)
2058 // CHECK4-NEXT:    [[TMP5:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
2059 // CHECK4-NEXT:    [[CMP:%.*]] = icmp sgt i32 [[TMP5]], 1
2060 // CHECK4-NEXT:    br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
2061 // CHECK4:       cond.true:
2062 // CHECK4-NEXT:    br label [[COND_END:%.*]]
2063 // CHECK4:       cond.false:
2064 // CHECK4-NEXT:    [[TMP6:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
2065 // CHECK4-NEXT:    br label [[COND_END]]
2066 // CHECK4:       cond.end:
2067 // CHECK4-NEXT:    [[COND:%.*]] = phi i32 [ 1, [[COND_TRUE]] ], [ [[TMP6]], [[COND_FALSE]] ]
2068 // CHECK4-NEXT:    store i32 [[COND]], i32* [[DOTOMP_UB]], align 4
2069 // CHECK4-NEXT:    [[TMP7:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4
2070 // CHECK4-NEXT:    store i32 [[TMP7]], i32* [[DOTOMP_IV]], align 4
2071 // CHECK4-NEXT:    br label [[OMP_INNER_FOR_COND:%.*]]
2072 // CHECK4:       omp.inner.for.cond:
2073 // CHECK4-NEXT:    [[TMP8:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
2074 // CHECK4-NEXT:    [[TMP9:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
2075 // CHECK4-NEXT:    [[CMP4:%.*]] = icmp sle i32 [[TMP8]], [[TMP9]]
2076 // CHECK4-NEXT:    br i1 [[CMP4]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
2077 // CHECK4:       omp.inner.for.body:
2078 // CHECK4-NEXT:    [[TMP10:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
2079 // CHECK4-NEXT:    [[MUL:%.*]] = mul nsw i32 [[TMP10]], 1
2080 // CHECK4-NEXT:    [[ADD:%.*]] = add nsw i32 0, [[MUL]]
2081 // CHECK4-NEXT:    store i32 [[ADD]], i32* [[I]], align 4
2082 // CHECK4-NEXT:    [[TMP11:%.*]] = load i32, i32* [[DOTLINEAR_START]], align 4
2083 // CHECK4-NEXT:    [[TMP12:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
2084 // CHECK4-NEXT:    [[MUL5:%.*]] = mul nsw i32 [[TMP12]], 5
2085 // CHECK4-NEXT:    [[ADD6:%.*]] = add nsw i32 [[TMP11]], [[MUL5]]
2086 // CHECK4-NEXT:    store i32 [[ADD6]], i32* [[G]], align 4
2087 // CHECK4-NEXT:    [[TMP13:%.*]] = load i32, i32* [[DOTLINEAR_START2]], align 4
2088 // CHECK4-NEXT:    [[TMP14:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
2089 // CHECK4-NEXT:    [[MUL7:%.*]] = mul nsw i32 [[TMP14]], 5
2090 // CHECK4-NEXT:    [[ADD8:%.*]] = add nsw i32 [[TMP13]], [[MUL7]]
2091 // CHECK4-NEXT:    store i32 [[ADD8]], i32* [[G1]], align 4
2092 // CHECK4-NEXT:    [[TMP15:%.*]] = load i32, i32* [[G]], align 4
2093 // CHECK4-NEXT:    [[ADD9:%.*]] = add nsw i32 [[TMP15]], 5
2094 // CHECK4-NEXT:    store i32 [[ADD9]], i32* [[G]], align 4
2095 // CHECK4-NEXT:    [[TMP16:%.*]] = load i32*, i32** [[_TMP3]], align 8
2096 // CHECK4-NEXT:    [[TMP17:%.*]] = load volatile i32, i32* [[TMP16]], align 4
2097 // CHECK4-NEXT:    [[ADD10:%.*]] = add nsw i32 [[TMP17]], 5
2098 // CHECK4-NEXT:    store volatile i32 [[ADD10]], i32* [[TMP16]], align 4
2099 // CHECK4-NEXT:    store i32 1, i32* [[G]], align 4
2100 // CHECK4-NEXT:    [[TMP18:%.*]] = load i32*, i32** [[_TMP3]], align 8
2101 // CHECK4-NEXT:    store volatile i32 5, i32* [[TMP18]], align 4
2102 // CHECK4-NEXT:    [[BLOCK_ISA:%.*]] = getelementptr inbounds <{ i8*, i32, i32, i8*, %struct.__block_descriptor*, i32*, i32 }>, <{ i8*, i32, i32, i8*, %struct.__block_descriptor*, i32*, i32 }>* [[BLOCK]], i32 0, i32 0
2103 // CHECK4-NEXT:    store i8* bitcast (i8** @_NSConcreteStackBlock to i8*), i8** [[BLOCK_ISA]], align 8
2104 // CHECK4-NEXT:    [[BLOCK_FLAGS:%.*]] = getelementptr inbounds <{ i8*, i32, i32, i8*, %struct.__block_descriptor*, i32*, i32 }>, <{ i8*, i32, i32, i8*, %struct.__block_descriptor*, i32*, i32 }>* [[BLOCK]], i32 0, i32 1
2105 // CHECK4-NEXT:    store i32 1073741824, i32* [[BLOCK_FLAGS]], align 8
2106 // CHECK4-NEXT:    [[BLOCK_RESERVED:%.*]] = getelementptr inbounds <{ i8*, i32, i32, i8*, %struct.__block_descriptor*, i32*, i32 }>, <{ i8*, i32, i32, i8*, %struct.__block_descriptor*, i32*, i32 }>* [[BLOCK]], i32 0, i32 2
2107 // CHECK4-NEXT:    store i32 0, i32* [[BLOCK_RESERVED]], align 4
2108 // CHECK4-NEXT:    [[BLOCK_INVOKE:%.*]] = getelementptr inbounds <{ i8*, i32, i32, i8*, %struct.__block_descriptor*, i32*, i32 }>, <{ i8*, i32, i32, i8*, %struct.__block_descriptor*, i32*, i32 }>* [[BLOCK]], i32 0, i32 3
2109 // CHECK4-NEXT:    store i8* bitcast (void (i8*)* @g1_block_invoke to i8*), i8** [[BLOCK_INVOKE]], align 8
2110 // CHECK4-NEXT:    [[BLOCK_DESCRIPTOR:%.*]] = getelementptr inbounds <{ i8*, i32, i32, i8*, %struct.__block_descriptor*, i32*, i32 }>, <{ i8*, i32, i32, i8*, %struct.__block_descriptor*, i32*, i32 }>* [[BLOCK]], i32 0, i32 4
2111 // 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
2112 // CHECK4-NEXT:    [[BLOCK_CAPTURED:%.*]] = getelementptr inbounds <{ i8*, i32, i32, i8*, %struct.__block_descriptor*, i32*, i32 }>, <{ i8*, i32, i32, i8*, %struct.__block_descriptor*, i32*, i32 }>* [[BLOCK]], i32 0, i32 6
2113 // CHECK4-NEXT:    [[TMP19:%.*]] = load volatile i32, i32* [[G]], align 4
2114 // CHECK4-NEXT:    store volatile i32 [[TMP19]], i32* [[BLOCK_CAPTURED]], align 8
2115 // CHECK4-NEXT:    [[BLOCK_CAPTURED11:%.*]] = getelementptr inbounds <{ i8*, i32, i32, i8*, %struct.__block_descriptor*, i32*, i32 }>, <{ i8*, i32, i32, i8*, %struct.__block_descriptor*, i32*, i32 }>* [[BLOCK]], i32 0, i32 5
2116 // CHECK4-NEXT:    [[TMP20:%.*]] = load i32*, i32** [[_TMP3]], align 8
2117 // CHECK4-NEXT:    store i32* [[TMP20]], i32** [[BLOCK_CAPTURED11]], align 8
2118 // CHECK4-NEXT:    [[TMP21:%.*]] = bitcast <{ i8*, i32, i32, i8*, %struct.__block_descriptor*, i32*, i32 }>* [[BLOCK]] to void ()*
2119 // CHECK4-NEXT:    [[BLOCK_LITERAL:%.*]] = bitcast void ()* [[TMP21]] to %struct.__block_literal_generic*
2120 // CHECK4-NEXT:    [[TMP22:%.*]] = getelementptr inbounds [[STRUCT___BLOCK_LITERAL_GENERIC:%.*]], %struct.__block_literal_generic* [[BLOCK_LITERAL]], i32 0, i32 3
2121 // CHECK4-NEXT:    [[TMP23:%.*]] = bitcast %struct.__block_literal_generic* [[BLOCK_LITERAL]] to i8*
2122 // CHECK4-NEXT:    [[TMP24:%.*]] = load i8*, i8** [[TMP22]], align 8
2123 // CHECK4-NEXT:    [[TMP25:%.*]] = bitcast i8* [[TMP24]] to void (i8*)*
2124 // CHECK4-NEXT:    call void [[TMP25]](i8* [[TMP23]])
2125 // CHECK4-NEXT:    br label [[OMP_BODY_CONTINUE:%.*]]
2126 // CHECK4:       omp.body.continue:
2127 // CHECK4-NEXT:    br label [[OMP_INNER_FOR_INC:%.*]]
2128 // CHECK4:       omp.inner.for.inc:
2129 // CHECK4-NEXT:    [[TMP26:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
2130 // CHECK4-NEXT:    [[ADD12:%.*]] = add nsw i32 [[TMP26]], 1
2131 // CHECK4-NEXT:    store i32 [[ADD12]], i32* [[DOTOMP_IV]], align 4
2132 // CHECK4-NEXT:    br label [[OMP_INNER_FOR_COND]]
2133 // CHECK4:       omp.inner.for.end:
2134 // CHECK4-NEXT:    br label [[OMP_LOOP_EXIT:%.*]]
2135 // CHECK4:       omp.loop.exit:
2136 // CHECK4-NEXT:    call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB2]], i32 [[TMP4]])
2137 // CHECK4-NEXT:    [[TMP27:%.*]] = load i32, i32* [[DOTOMP_IS_LAST]], align 4
2138 // CHECK4-NEXT:    [[TMP28:%.*]] = icmp ne i32 [[TMP27]], 0
2139 // CHECK4-NEXT:    br i1 [[TMP28]], label [[DOTOMP_LINEAR_PU:%.*]], label [[DOTOMP_LINEAR_PU_DONE:%.*]]
2140 // CHECK4:       .omp.linear.pu:
2141 // CHECK4-NEXT:    [[TMP29:%.*]] = load i32, i32* [[G]], align 4
2142 // CHECK4-NEXT:    store i32 [[TMP29]], i32* @g, align 4
2143 // CHECK4-NEXT:    [[TMP30:%.*]] = load i32*, i32** @g1, align 8
2144 // CHECK4-NEXT:    store i32* [[TMP30]], i32** [[_TMP13]], align 8
2145 // CHECK4-NEXT:    [[TMP31:%.*]] = load i32, i32* [[G1]], align 4
2146 // CHECK4-NEXT:    [[TMP32:%.*]] = load i32*, i32** [[_TMP13]], align 8
2147 // CHECK4-NEXT:    store volatile i32 [[TMP31]], i32* [[TMP32]], align 4
2148 // CHECK4-NEXT:    br label [[DOTOMP_LINEAR_PU_DONE]]
2149 // CHECK4:       .omp.linear.pu.done:
2150 // CHECK4-NEXT:    call void @__kmpc_barrier(%struct.ident_t* @[[GLOB1]], i32 [[TMP4]])
2151 // CHECK4-NEXT:    ret void
2152 //
2153 //
2154 // CHECK4-LABEL: define {{[^@]+}}@g1_block_invoke
2155 // CHECK4-SAME: (i8* [[DOTBLOCK_DESCRIPTOR:%.*]]) #[[ATTR2]] {
2156 // CHECK4-NEXT:  entry:
2157 // CHECK4-NEXT:    [[DOTBLOCK_DESCRIPTOR_ADDR:%.*]] = alloca i8*, align 8
2158 // CHECK4-NEXT:    [[BLOCK_ADDR:%.*]] = alloca <{ i8*, i32, i32, i8*, %struct.__block_descriptor*, i32*, i32 }>*, align 8
2159 // CHECK4-NEXT:    store i8* [[DOTBLOCK_DESCRIPTOR]], i8** [[DOTBLOCK_DESCRIPTOR_ADDR]], align 8
2160 // CHECK4-NEXT:    [[BLOCK:%.*]] = bitcast i8* [[DOTBLOCK_DESCRIPTOR]] to <{ i8*, i32, i32, i8*, %struct.__block_descriptor*, i32*, i32 }>*
2161 // CHECK4-NEXT:    store <{ i8*, i32, i32, i8*, %struct.__block_descriptor*, i32*, i32 }>* [[BLOCK]], <{ i8*, i32, i32, i8*, %struct.__block_descriptor*, i32*, i32 }>** [[BLOCK_ADDR]], align 8
2162 // CHECK4-NEXT:    [[BLOCK_CAPTURE_ADDR:%.*]] = getelementptr inbounds <{ i8*, i32, i32, i8*, %struct.__block_descriptor*, i32*, i32 }>, <{ i8*, i32, i32, i8*, %struct.__block_descriptor*, i32*, i32 }>* [[BLOCK]], i32 0, i32 6
2163 // CHECK4-NEXT:    store i32 2, i32* [[BLOCK_CAPTURE_ADDR]], align 8
2164 // CHECK4-NEXT:    [[BLOCK_CAPTURE_ADDR1:%.*]] = getelementptr inbounds <{ i8*, i32, i32, i8*, %struct.__block_descriptor*, i32*, i32 }>, <{ i8*, i32, i32, i8*, %struct.__block_descriptor*, i32*, i32 }>* [[BLOCK]], i32 0, i32 5
2165 // CHECK4-NEXT:    [[TMP0:%.*]] = load i32*, i32** [[BLOCK_CAPTURE_ADDR1]], align 8
2166 // CHECK4-NEXT:    store i32 2, i32* [[TMP0]], align 4
2167 // CHECK4-NEXT:    ret void
2168 //
2169 //
2170 // CHECK4-LABEL: define {{[^@]+}}@_ZN2SSC2ERi
2171 // CHECK4-SAME: (%struct.SS* nonnull align 8 dereferenceable(16) [[THIS:%.*]], i32* nonnull align 4 dereferenceable(4) [[D:%.*]]) unnamed_addr #[[ATTR2]] align 2 {
2172 // CHECK4-NEXT:  entry:
2173 // CHECK4-NEXT:    [[THIS_ADDR:%.*]] = alloca %struct.SS*, align 8
2174 // CHECK4-NEXT:    [[D_ADDR:%.*]] = alloca i32*, align 8
2175 // CHECK4-NEXT:    store %struct.SS* [[THIS]], %struct.SS** [[THIS_ADDR]], align 8
2176 // CHECK4-NEXT:    store i32* [[D]], i32** [[D_ADDR]], align 8
2177 // CHECK4-NEXT:    [[THIS1:%.*]] = load %struct.SS*, %struct.SS** [[THIS_ADDR]], align 8
2178 // CHECK4-NEXT:    [[A:%.*]] = getelementptr inbounds [[STRUCT_SS:%.*]], %struct.SS* [[THIS1]], i32 0, i32 0
2179 // CHECK4-NEXT:    store i32 0, i32* [[A]], align 8
2180 // CHECK4-NEXT:    [[B:%.*]] = getelementptr inbounds [[STRUCT_SS]], %struct.SS* [[THIS1]], i32 0, i32 1
2181 // CHECK4-NEXT:    [[BF_LOAD:%.*]] = load i8, i8* [[B]], align 4
2182 // CHECK4-NEXT:    [[BF_CLEAR:%.*]] = and i8 [[BF_LOAD]], -16
2183 // CHECK4-NEXT:    store i8 [[BF_CLEAR]], i8* [[B]], align 4
2184 // CHECK4-NEXT:    [[C:%.*]] = getelementptr inbounds [[STRUCT_SS]], %struct.SS* [[THIS1]], i32 0, i32 2
2185 // CHECK4-NEXT:    [[TMP0:%.*]] = load i32*, i32** [[D_ADDR]], align 8
2186 // CHECK4-NEXT:    store i32* [[TMP0]], i32** [[C]], align 8
2187 // CHECK4-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.SS*)* @.omp_outlined..2 to void (i32*, i32*, ...)*), %struct.SS* [[THIS1]])
2188 // CHECK4-NEXT:    ret void
2189 //
2190 //
2191 // CHECK4-LABEL: define {{[^@]+}}@.omp_outlined..2
2192 // CHECK4-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], %struct.SS* [[THIS:%.*]]) #[[ATTR3]] {
2193 // CHECK4-NEXT:  entry:
2194 // CHECK4-NEXT:    [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
2195 // CHECK4-NEXT:    [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
2196 // CHECK4-NEXT:    [[THIS_ADDR:%.*]] = alloca %struct.SS*, align 8
2197 // CHECK4-NEXT:    [[A:%.*]] = alloca i32*, align 8
2198 // CHECK4-NEXT:    [[B:%.*]] = alloca i32, align 4
2199 // CHECK4-NEXT:    [[C:%.*]] = alloca i32*, align 8
2200 // CHECK4-NEXT:    [[TMP:%.*]] = alloca i32*, align 8
2201 // CHECK4-NEXT:    [[_TMP3:%.*]] = alloca i32*, align 8
2202 // CHECK4-NEXT:    [[DOTOMP_IV:%.*]] = alloca i32, align 4
2203 // CHECK4-NEXT:    [[_TMP4:%.*]] = alloca i32, align 4
2204 // CHECK4-NEXT:    [[DOTLINEAR_START:%.*]] = alloca i32, align 4
2205 // CHECK4-NEXT:    [[DOTLINEAR_START5:%.*]] = alloca i32, align 4
2206 // CHECK4-NEXT:    [[DOTLINEAR_START6:%.*]] = alloca i32, align 4
2207 // CHECK4-NEXT:    [[DOTOMP_LB:%.*]] = alloca i32, align 4
2208 // CHECK4-NEXT:    [[DOTOMP_UB:%.*]] = alloca i32, align 4
2209 // CHECK4-NEXT:    [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
2210 // CHECK4-NEXT:    [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
2211 // CHECK4-NEXT:    [[I:%.*]] = alloca i32, align 4
2212 // CHECK4-NEXT:    [[A7:%.*]] = alloca i32, align 4
2213 // CHECK4-NEXT:    [[_TMP8:%.*]] = alloca i32*, align 8
2214 // CHECK4-NEXT:    [[B9:%.*]] = alloca i32, align 4
2215 // CHECK4-NEXT:    [[C10:%.*]] = alloca i32, align 4
2216 // CHECK4-NEXT:    [[_TMP11:%.*]] = alloca i32*, align 8
2217 // CHECK4-NEXT:    [[BLOCK:%.*]] = alloca <{ i8*, i32, i32, i8*, %struct.__block_descriptor*, %struct.SS*, i32*, i32*, i32 }>, align 8
2218 // CHECK4-NEXT:    [[_TMP22:%.*]] = alloca i32*, align 8
2219 // CHECK4-NEXT:    [[_TMP23:%.*]] = alloca i32*, align 8
2220 // CHECK4-NEXT:    store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
2221 // CHECK4-NEXT:    store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
2222 // CHECK4-NEXT:    store %struct.SS* [[THIS]], %struct.SS** [[THIS_ADDR]], align 8
2223 // CHECK4-NEXT:    [[TMP0:%.*]] = load %struct.SS*, %struct.SS** [[THIS_ADDR]], align 8
2224 // CHECK4-NEXT:    [[A1:%.*]] = getelementptr inbounds [[STRUCT_SS:%.*]], %struct.SS* [[TMP0]], i32 0, i32 0
2225 // CHECK4-NEXT:    store i32* [[A1]], i32** [[A]], align 8
2226 // CHECK4-NEXT:    [[C2:%.*]] = getelementptr inbounds [[STRUCT_SS]], %struct.SS* [[TMP0]], i32 0, i32 2
2227 // CHECK4-NEXT:    [[TMP1:%.*]] = load i32*, i32** [[C2]], align 8
2228 // CHECK4-NEXT:    store i32* [[TMP1]], i32** [[C]], align 8
2229 // CHECK4-NEXT:    [[TMP2:%.*]] = load i32*, i32** [[A]], align 8
2230 // CHECK4-NEXT:    store i32* [[TMP2]], i32** [[TMP]], align 8
2231 // CHECK4-NEXT:    [[TMP3:%.*]] = load i32*, i32** [[C]], align 8
2232 // CHECK4-NEXT:    store i32* [[TMP3]], i32** [[_TMP3]], align 8
2233 // CHECK4-NEXT:    [[TMP4:%.*]] = load i32*, i32** [[TMP]], align 8
2234 // CHECK4-NEXT:    [[TMP5:%.*]] = load i32, i32* [[TMP4]], align 4
2235 // CHECK4-NEXT:    store i32 [[TMP5]], i32* [[DOTLINEAR_START]], align 4
2236 // CHECK4-NEXT:    [[TMP6:%.*]] = load i32, i32* [[B]], align 4
2237 // CHECK4-NEXT:    store i32 [[TMP6]], i32* [[DOTLINEAR_START5]], align 4
2238 // CHECK4-NEXT:    [[TMP7:%.*]] = load i32*, i32** [[_TMP3]], align 8
2239 // CHECK4-NEXT:    [[TMP8:%.*]] = load i32, i32* [[TMP7]], align 4
2240 // CHECK4-NEXT:    store i32 [[TMP8]], i32* [[DOTLINEAR_START6]], align 4
2241 // CHECK4-NEXT:    store i32 0, i32* [[DOTOMP_LB]], align 4
2242 // CHECK4-NEXT:    store i32 1, i32* [[DOTOMP_UB]], align 4
2243 // CHECK4-NEXT:    store i32 1, i32* [[DOTOMP_STRIDE]], align 4
2244 // CHECK4-NEXT:    store i32 0, i32* [[DOTOMP_IS_LAST]], align 4
2245 // CHECK4-NEXT:    [[TMP9:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
2246 // CHECK4-NEXT:    [[TMP10:%.*]] = load i32, i32* [[TMP9]], align 4
2247 // CHECK4-NEXT:    call void @__kmpc_barrier(%struct.ident_t* @[[GLOB1]], i32 [[TMP10]])
2248 // CHECK4-NEXT:    store i32* [[A7]], i32** [[_TMP8]], align 8
2249 // CHECK4-NEXT:    store i32* [[C10]], i32** [[_TMP11]], align 8
2250 // CHECK4-NEXT:    call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB2]], i32 [[TMP10]], i32 34, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_LB]], i32* [[DOTOMP_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1)
2251 // CHECK4-NEXT:    [[TMP11:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
2252 // CHECK4-NEXT:    [[CMP:%.*]] = icmp sgt i32 [[TMP11]], 1
2253 // CHECK4-NEXT:    br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
2254 // CHECK4:       cond.true:
2255 // CHECK4-NEXT:    br label [[COND_END:%.*]]
2256 // CHECK4:       cond.false:
2257 // CHECK4-NEXT:    [[TMP12:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
2258 // CHECK4-NEXT:    br label [[COND_END]]
2259 // CHECK4:       cond.end:
2260 // CHECK4-NEXT:    [[COND:%.*]] = phi i32 [ 1, [[COND_TRUE]] ], [ [[TMP12]], [[COND_FALSE]] ]
2261 // CHECK4-NEXT:    store i32 [[COND]], i32* [[DOTOMP_UB]], align 4
2262 // CHECK4-NEXT:    [[TMP13:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4
2263 // CHECK4-NEXT:    store i32 [[TMP13]], i32* [[DOTOMP_IV]], align 4
2264 // CHECK4-NEXT:    br label [[OMP_INNER_FOR_COND:%.*]]
2265 // CHECK4:       omp.inner.for.cond:
2266 // CHECK4-NEXT:    [[TMP14:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
2267 // CHECK4-NEXT:    [[TMP15:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
2268 // CHECK4-NEXT:    [[CMP12:%.*]] = icmp sle i32 [[TMP14]], [[TMP15]]
2269 // CHECK4-NEXT:    br i1 [[CMP12]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
2270 // CHECK4:       omp.inner.for.body:
2271 // CHECK4-NEXT:    [[TMP16:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
2272 // CHECK4-NEXT:    [[MUL:%.*]] = mul nsw i32 [[TMP16]], 1
2273 // CHECK4-NEXT:    [[ADD:%.*]] = add nsw i32 0, [[MUL]]
2274 // CHECK4-NEXT:    store i32 [[ADD]], i32* [[I]], align 4
2275 // CHECK4-NEXT:    [[TMP17:%.*]] = load i32, i32* [[DOTLINEAR_START]], align 4
2276 // CHECK4-NEXT:    [[TMP18:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
2277 // CHECK4-NEXT:    [[MUL13:%.*]] = mul nsw i32 [[TMP18]], 1
2278 // CHECK4-NEXT:    [[ADD14:%.*]] = add nsw i32 [[TMP17]], [[MUL13]]
2279 // CHECK4-NEXT:    store i32 [[ADD14]], i32* [[A7]], align 4
2280 // CHECK4-NEXT:    [[TMP19:%.*]] = load i32, i32* [[DOTLINEAR_START5]], align 4
2281 // CHECK4-NEXT:    [[TMP20:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
2282 // CHECK4-NEXT:    [[MUL15:%.*]] = mul nsw i32 [[TMP20]], 1
2283 // CHECK4-NEXT:    [[ADD16:%.*]] = add nsw i32 [[TMP19]], [[MUL15]]
2284 // CHECK4-NEXT:    store i32 [[ADD16]], i32* [[B9]], align 4
2285 // CHECK4-NEXT:    [[TMP21:%.*]] = load i32, i32* [[DOTLINEAR_START6]], align 4
2286 // CHECK4-NEXT:    [[TMP22:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
2287 // CHECK4-NEXT:    [[MUL17:%.*]] = mul nsw i32 [[TMP22]], 1
2288 // CHECK4-NEXT:    [[ADD18:%.*]] = add nsw i32 [[TMP21]], [[MUL17]]
2289 // CHECK4-NEXT:    store i32 [[ADD18]], i32* [[C10]], align 4
2290 // CHECK4-NEXT:    [[BLOCK_ISA:%.*]] = getelementptr inbounds <{ i8*, i32, i32, i8*, %struct.__block_descriptor*, %struct.SS*, i32*, i32*, i32 }>, <{ i8*, i32, i32, i8*, %struct.__block_descriptor*, %struct.SS*, i32*, i32*, i32 }>* [[BLOCK]], i32 0, i32 0
2291 // CHECK4-NEXT:    store i8* bitcast (i8** @_NSConcreteStackBlock to i8*), i8** [[BLOCK_ISA]], align 8
2292 // CHECK4-NEXT:    [[BLOCK_FLAGS:%.*]] = getelementptr inbounds <{ i8*, i32, i32, i8*, %struct.__block_descriptor*, %struct.SS*, i32*, i32*, i32 }>, <{ i8*, i32, i32, i8*, %struct.__block_descriptor*, %struct.SS*, i32*, i32*, i32 }>* [[BLOCK]], i32 0, i32 1
2293 // CHECK4-NEXT:    store i32 1073741824, i32* [[BLOCK_FLAGS]], align 8
2294 // CHECK4-NEXT:    [[BLOCK_RESERVED:%.*]] = getelementptr inbounds <{ i8*, i32, i32, i8*, %struct.__block_descriptor*, %struct.SS*, i32*, i32*, i32 }>, <{ i8*, i32, i32, i8*, %struct.__block_descriptor*, %struct.SS*, i32*, i32*, i32 }>* [[BLOCK]], i32 0, i32 2
2295 // CHECK4-NEXT:    store i32 0, i32* [[BLOCK_RESERVED]], align 4
2296 // CHECK4-NEXT:    [[BLOCK_INVOKE:%.*]] = getelementptr inbounds <{ i8*, i32, i32, i8*, %struct.__block_descriptor*, %struct.SS*, i32*, i32*, i32 }>, <{ i8*, i32, i32, i8*, %struct.__block_descriptor*, %struct.SS*, i32*, i32*, i32 }>* [[BLOCK]], i32 0, i32 3
2297 // CHECK4-NEXT:    store i8* bitcast (void (i8*)* @g1_block_invoke_2 to i8*), i8** [[BLOCK_INVOKE]], align 8
2298 // CHECK4-NEXT:    [[BLOCK_DESCRIPTOR:%.*]] = getelementptr inbounds <{ i8*, i32, i32, i8*, %struct.__block_descriptor*, %struct.SS*, i32*, i32*, i32 }>, <{ i8*, i32, i32, i8*, %struct.__block_descriptor*, %struct.SS*, i32*, i32*, i32 }>* [[BLOCK]], i32 0, i32 4
2299 // CHECK4-NEXT:    store %struct.__block_descriptor* bitcast ({ i64, i64, i8*, i8* }* @__block_descriptor_tmp.4 to %struct.__block_descriptor*), %struct.__block_descriptor** [[BLOCK_DESCRIPTOR]], align 8
2300 // CHECK4-NEXT:    [[BLOCK_CAPTURED_THIS_ADDR:%.*]] = getelementptr inbounds <{ i8*, i32, i32, i8*, %struct.__block_descriptor*, %struct.SS*, i32*, i32*, i32 }>, <{ i8*, i32, i32, i8*, %struct.__block_descriptor*, %struct.SS*, i32*, i32*, i32 }>* [[BLOCK]], i32 0, i32 5
2301 // CHECK4-NEXT:    store %struct.SS* [[TMP0]], %struct.SS** [[BLOCK_CAPTURED_THIS_ADDR]], align 8
2302 // CHECK4-NEXT:    [[BLOCK_CAPTURED:%.*]] = getelementptr inbounds <{ i8*, i32, i32, i8*, %struct.__block_descriptor*, %struct.SS*, i32*, i32*, i32 }>, <{ i8*, i32, i32, i8*, %struct.__block_descriptor*, %struct.SS*, i32*, i32*, i32 }>* [[BLOCK]], i32 0, i32 6
2303 // CHECK4-NEXT:    [[TMP23:%.*]] = load i32*, i32** [[_TMP8]], align 8
2304 // CHECK4-NEXT:    store i32* [[TMP23]], i32** [[BLOCK_CAPTURED]], align 8
2305 // CHECK4-NEXT:    [[BLOCK_CAPTURED19:%.*]] = getelementptr inbounds <{ i8*, i32, i32, i8*, %struct.__block_descriptor*, %struct.SS*, i32*, i32*, i32 }>, <{ i8*, i32, i32, i8*, %struct.__block_descriptor*, %struct.SS*, i32*, i32*, i32 }>* [[BLOCK]], i32 0, i32 8
2306 // CHECK4-NEXT:    [[TMP24:%.*]] = load i32, i32* [[B9]], align 4
2307 // CHECK4-NEXT:    store i32 [[TMP24]], i32* [[BLOCK_CAPTURED19]], align 8
2308 // CHECK4-NEXT:    [[BLOCK_CAPTURED20:%.*]] = getelementptr inbounds <{ i8*, i32, i32, i8*, %struct.__block_descriptor*, %struct.SS*, i32*, i32*, i32 }>, <{ i8*, i32, i32, i8*, %struct.__block_descriptor*, %struct.SS*, i32*, i32*, i32 }>* [[BLOCK]], i32 0, i32 7
2309 // CHECK4-NEXT:    [[TMP25:%.*]] = load i32*, i32** [[_TMP11]], align 8
2310 // CHECK4-NEXT:    store i32* [[TMP25]], i32** [[BLOCK_CAPTURED20]], align 8
2311 // CHECK4-NEXT:    [[TMP26:%.*]] = bitcast <{ i8*, i32, i32, i8*, %struct.__block_descriptor*, %struct.SS*, i32*, i32*, i32 }>* [[BLOCK]] to void ()*
2312 // CHECK4-NEXT:    [[BLOCK_LITERAL:%.*]] = bitcast void ()* [[TMP26]] to %struct.__block_literal_generic*
2313 // CHECK4-NEXT:    [[TMP27:%.*]] = getelementptr inbounds [[STRUCT___BLOCK_LITERAL_GENERIC:%.*]], %struct.__block_literal_generic* [[BLOCK_LITERAL]], i32 0, i32 3
2314 // CHECK4-NEXT:    [[TMP28:%.*]] = bitcast %struct.__block_literal_generic* [[BLOCK_LITERAL]] to i8*
2315 // CHECK4-NEXT:    [[TMP29:%.*]] = load i8*, i8** [[TMP27]], align 8
2316 // CHECK4-NEXT:    [[TMP30:%.*]] = bitcast i8* [[TMP29]] to void (i8*)*
2317 // CHECK4-NEXT:    call void [[TMP30]](i8* [[TMP28]])
2318 // CHECK4-NEXT:    br label [[OMP_BODY_CONTINUE:%.*]]
2319 // CHECK4:       omp.body.continue:
2320 // CHECK4-NEXT:    br label [[OMP_INNER_FOR_INC:%.*]]
2321 // CHECK4:       omp.inner.for.inc:
2322 // CHECK4-NEXT:    [[TMP31:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
2323 // CHECK4-NEXT:    [[ADD21:%.*]] = add nsw i32 [[TMP31]], 1
2324 // CHECK4-NEXT:    store i32 [[ADD21]], i32* [[DOTOMP_IV]], align 4
2325 // CHECK4-NEXT:    br label [[OMP_INNER_FOR_COND]]
2326 // CHECK4:       omp.inner.for.end:
2327 // CHECK4-NEXT:    br label [[OMP_LOOP_EXIT:%.*]]
2328 // CHECK4:       omp.loop.exit:
2329 // CHECK4-NEXT:    call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB2]], i32 [[TMP10]])
2330 // CHECK4-NEXT:    [[TMP32:%.*]] = load i32, i32* [[DOTOMP_IS_LAST]], align 4
2331 // CHECK4-NEXT:    [[TMP33:%.*]] = icmp ne i32 [[TMP32]], 0
2332 // CHECK4-NEXT:    br i1 [[TMP33]], label [[DOTOMP_LINEAR_PU:%.*]], label [[DOTOMP_LINEAR_PU_DONE:%.*]]
2333 // CHECK4:       .omp.linear.pu:
2334 // CHECK4-NEXT:    [[TMP34:%.*]] = load i32*, i32** [[TMP]], align 8
2335 // CHECK4-NEXT:    store i32* [[TMP34]], i32** [[_TMP22]], align 8
2336 // CHECK4-NEXT:    [[TMP35:%.*]] = load i32, i32* [[A7]], align 4
2337 // CHECK4-NEXT:    [[TMP36:%.*]] = load i32*, i32** [[_TMP22]], align 8
2338 // CHECK4-NEXT:    store i32 [[TMP35]], i32* [[TMP36]], align 4
2339 // CHECK4-NEXT:    [[TMP37:%.*]] = load i32, i32* [[B9]], align 4
2340 // CHECK4-NEXT:    store i32 [[TMP37]], i32* [[B]], align 4
2341 // CHECK4-NEXT:    [[TMP38:%.*]] = load i32*, i32** [[_TMP3]], align 8
2342 // CHECK4-NEXT:    store i32* [[TMP38]], i32** [[_TMP23]], align 8
2343 // CHECK4-NEXT:    [[TMP39:%.*]] = load i32, i32* [[C10]], align 4
2344 // CHECK4-NEXT:    [[TMP40:%.*]] = load i32*, i32** [[_TMP23]], align 8
2345 // CHECK4-NEXT:    store i32 [[TMP39]], i32* [[TMP40]], align 4
2346 // CHECK4-NEXT:    [[TMP41:%.*]] = load i32, i32* [[B]], align 4
2347 // CHECK4-NEXT:    [[B24:%.*]] = getelementptr inbounds [[STRUCT_SS]], %struct.SS* [[TMP0]], i32 0, i32 1
2348 // CHECK4-NEXT:    [[TMP42:%.*]] = trunc i32 [[TMP41]] to i8
2349 // CHECK4-NEXT:    [[BF_LOAD:%.*]] = load i8, i8* [[B24]], align 4
2350 // CHECK4-NEXT:    [[BF_VALUE:%.*]] = and i8 [[TMP42]], 15
2351 // CHECK4-NEXT:    [[BF_CLEAR:%.*]] = and i8 [[BF_LOAD]], -16
2352 // CHECK4-NEXT:    [[BF_SET:%.*]] = or i8 [[BF_CLEAR]], [[BF_VALUE]]
2353 // CHECK4-NEXT:    store i8 [[BF_SET]], i8* [[B24]], align 4
2354 // CHECK4-NEXT:    br label [[DOTOMP_LINEAR_PU_DONE]]
2355 // CHECK4:       .omp.linear.pu.done:
2356 // CHECK4-NEXT:    call void @__kmpc_barrier(%struct.ident_t* @[[GLOB1]], i32 [[TMP10]])
2357 // CHECK4-NEXT:    ret void
2358 //
2359 //
2360 // CHECK4-LABEL: define {{[^@]+}}@g1_block_invoke_2
2361 // CHECK4-SAME: (i8* [[DOTBLOCK_DESCRIPTOR:%.*]]) #[[ATTR2]] {
2362 // CHECK4-NEXT:  entry:
2363 // CHECK4-NEXT:    [[DOTBLOCK_DESCRIPTOR_ADDR:%.*]] = alloca i8*, align 8
2364 // CHECK4-NEXT:    [[BLOCK_ADDR:%.*]] = alloca <{ i8*, i32, i32, i8*, %struct.__block_descriptor*, %struct.SS*, i32*, i32*, i32 }>*, align 8
2365 // CHECK4-NEXT:    store i8* [[DOTBLOCK_DESCRIPTOR]], i8** [[DOTBLOCK_DESCRIPTOR_ADDR]], align 8
2366 // CHECK4-NEXT:    [[BLOCK:%.*]] = bitcast i8* [[DOTBLOCK_DESCRIPTOR]] to <{ i8*, i32, i32, i8*, %struct.__block_descriptor*, %struct.SS*, i32*, i32*, i32 }>*
2367 // CHECK4-NEXT:    store <{ i8*, i32, i32, i8*, %struct.__block_descriptor*, %struct.SS*, i32*, i32*, i32 }>* [[BLOCK]], <{ i8*, i32, i32, i8*, %struct.__block_descriptor*, %struct.SS*, i32*, i32*, i32 }>** [[BLOCK_ADDR]], align 8
2368 // CHECK4-NEXT:    [[BLOCK_CAPTURED_THIS:%.*]] = getelementptr inbounds <{ i8*, i32, i32, i8*, %struct.__block_descriptor*, %struct.SS*, i32*, i32*, i32 }>, <{ i8*, i32, i32, i8*, %struct.__block_descriptor*, %struct.SS*, i32*, i32*, i32 }>* [[BLOCK]], i32 0, i32 5
2369 // CHECK4-NEXT:    [[THIS:%.*]] = load %struct.SS*, %struct.SS** [[BLOCK_CAPTURED_THIS]], align 8
2370 // CHECK4-NEXT:    [[BLOCK_CAPTURE_ADDR:%.*]] = getelementptr inbounds <{ i8*, i32, i32, i8*, %struct.__block_descriptor*, %struct.SS*, i32*, i32*, i32 }>, <{ i8*, i32, i32, i8*, %struct.__block_descriptor*, %struct.SS*, i32*, i32*, i32 }>* [[BLOCK]], i32 0, i32 6
2371 // CHECK4-NEXT:    [[TMP0:%.*]] = load i32*, i32** [[BLOCK_CAPTURE_ADDR]], align 8
2372 // CHECK4-NEXT:    [[TMP1:%.*]] = load i32, i32* [[TMP0]], align 4
2373 // CHECK4-NEXT:    [[INC:%.*]] = add nsw i32 [[TMP1]], 1
2374 // CHECK4-NEXT:    store i32 [[INC]], i32* [[TMP0]], align 4
2375 // CHECK4-NEXT:    [[BLOCK_CAPTURE_ADDR1:%.*]] = getelementptr inbounds <{ i8*, i32, i32, i8*, %struct.__block_descriptor*, %struct.SS*, i32*, i32*, i32 }>, <{ i8*, i32, i32, i8*, %struct.__block_descriptor*, %struct.SS*, i32*, i32*, i32 }>* [[BLOCK]], i32 0, i32 8
2376 // CHECK4-NEXT:    [[TMP2:%.*]] = load i32, i32* [[BLOCK_CAPTURE_ADDR1]], align 8
2377 // CHECK4-NEXT:    [[DEC:%.*]] = add nsw i32 [[TMP2]], -1
2378 // CHECK4-NEXT:    store i32 [[DEC]], i32* [[BLOCK_CAPTURE_ADDR1]], align 8
2379 // CHECK4-NEXT:    [[BLOCK_CAPTURE_ADDR2:%.*]] = getelementptr inbounds <{ i8*, i32, i32, i8*, %struct.__block_descriptor*, %struct.SS*, i32*, i32*, i32 }>, <{ i8*, i32, i32, i8*, %struct.__block_descriptor*, %struct.SS*, i32*, i32*, i32 }>* [[BLOCK]], i32 0, i32 7
2380 // CHECK4-NEXT:    [[TMP3:%.*]] = load i32*, i32** [[BLOCK_CAPTURE_ADDR2]], align 8
2381 // CHECK4-NEXT:    [[TMP4:%.*]] = load i32, i32* [[TMP3]], align 4
2382 // CHECK4-NEXT:    [[DIV:%.*]] = sdiv i32 [[TMP4]], 1
2383 // CHECK4-NEXT:    store i32 [[DIV]], i32* [[TMP3]], align 4
2384 // CHECK4-NEXT:    [[BLOCK_CAPTURE_ADDR3:%.*]] = getelementptr inbounds <{ i8*, i32, i32, i8*, %struct.__block_descriptor*, %struct.SS*, i32*, i32*, i32 }>, <{ i8*, i32, i32, i8*, %struct.__block_descriptor*, %struct.SS*, i32*, i32*, i32 }>* [[BLOCK]], i32 0, i32 6
2385 // CHECK4-NEXT:    [[TMP5:%.*]] = load i32*, i32** [[BLOCK_CAPTURE_ADDR3]], align 8
2386 // CHECK4-NEXT:    [[BLOCK_CAPTURE_ADDR4:%.*]] = getelementptr inbounds <{ i8*, i32, i32, i8*, %struct.__block_descriptor*, %struct.SS*, i32*, i32*, i32 }>, <{ i8*, i32, i32, i8*, %struct.__block_descriptor*, %struct.SS*, i32*, i32*, i32 }>* [[BLOCK]], i32 0, i32 8
2387 // CHECK4-NEXT:    [[BLOCK_CAPTURE_ADDR5:%.*]] = getelementptr inbounds <{ i8*, i32, i32, i8*, %struct.__block_descriptor*, %struct.SS*, i32*, i32*, i32 }>, <{ i8*, i32, i32, i8*, %struct.__block_descriptor*, %struct.SS*, i32*, i32*, i32 }>* [[BLOCK]], i32 0, i32 7
2388 // CHECK4-NEXT:    [[TMP6:%.*]] = load i32*, i32** [[BLOCK_CAPTURE_ADDR5]], align 8
2389 // CHECK4-NEXT:    call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB3]], i32 4, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, %struct.SS*, i32*, i32*, i32*)* @.omp_outlined..3 to void (i32*, i32*, ...)*), %struct.SS* [[THIS]], i32* [[TMP5]], i32* [[BLOCK_CAPTURE_ADDR4]], i32* [[TMP6]])
2390 // CHECK4-NEXT:    ret void
2391 //
2392 //
2393 // CHECK4-LABEL: define {{[^@]+}}@.omp_outlined..3
2394 // CHECK4-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], %struct.SS* [[THIS:%.*]], i32* nonnull align 4 dereferenceable(4) [[A:%.*]], i32* nonnull align 4 dereferenceable(4) [[B:%.*]], i32* nonnull align 4 dereferenceable(4) [[C:%.*]]) #[[ATTR3]] {
2395 // CHECK4-NEXT:  entry:
2396 // CHECK4-NEXT:    [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
2397 // CHECK4-NEXT:    [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
2398 // CHECK4-NEXT:    [[THIS_ADDR:%.*]] = alloca %struct.SS*, align 8
2399 // CHECK4-NEXT:    [[A_ADDR:%.*]] = alloca i32*, align 8
2400 // CHECK4-NEXT:    [[B_ADDR:%.*]] = alloca i32*, align 8
2401 // CHECK4-NEXT:    [[C_ADDR:%.*]] = alloca i32*, align 8
2402 // CHECK4-NEXT:    [[TMP:%.*]] = alloca i32*, align 8
2403 // CHECK4-NEXT:    [[_TMP1:%.*]] = alloca i32*, align 8
2404 // CHECK4-NEXT:    [[_TMP2:%.*]] = alloca i32*, align 8
2405 // CHECK4-NEXT:    [[_TMP3:%.*]] = alloca i32*, align 8
2406 // CHECK4-NEXT:    [[DOTOMP_IV:%.*]] = alloca i32, align 4
2407 // CHECK4-NEXT:    [[_TMP4:%.*]] = alloca i32, align 4
2408 // CHECK4-NEXT:    [[DOTLINEAR_START:%.*]] = alloca i32, align 4
2409 // CHECK4-NEXT:    [[DOTLINEAR_START5:%.*]] = alloca i32, align 4
2410 // CHECK4-NEXT:    [[DOTLINEAR_START6:%.*]] = alloca i32, align 4
2411 // CHECK4-NEXT:    [[DOTOMP_LB:%.*]] = alloca i32, align 4
2412 // CHECK4-NEXT:    [[DOTOMP_UB:%.*]] = alloca i32, align 4
2413 // CHECK4-NEXT:    [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
2414 // CHECK4-NEXT:    [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
2415 // CHECK4-NEXT:    [[I:%.*]] = alloca i32, align 4
2416 // CHECK4-NEXT:    [[A7:%.*]] = alloca i32, align 4
2417 // CHECK4-NEXT:    [[_TMP8:%.*]] = alloca i32*, align 8
2418 // CHECK4-NEXT:    [[B9:%.*]] = alloca i32, align 4
2419 // CHECK4-NEXT:    [[C10:%.*]] = alloca i32, align 4
2420 // CHECK4-NEXT:    [[_TMP11:%.*]] = alloca i32*, align 8
2421 // CHECK4-NEXT:    [[_TMP20:%.*]] = alloca i32*, align 8
2422 // CHECK4-NEXT:    [[_TMP21:%.*]] = alloca i32*, align 8
2423 // CHECK4-NEXT:    store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
2424 // CHECK4-NEXT:    store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
2425 // CHECK4-NEXT:    store %struct.SS* [[THIS]], %struct.SS** [[THIS_ADDR]], align 8
2426 // CHECK4-NEXT:    store i32* [[A]], i32** [[A_ADDR]], align 8
2427 // CHECK4-NEXT:    store i32* [[B]], i32** [[B_ADDR]], align 8
2428 // CHECK4-NEXT:    store i32* [[C]], i32** [[C_ADDR]], align 8
2429 // CHECK4-NEXT:    [[TMP0:%.*]] = load %struct.SS*, %struct.SS** [[THIS_ADDR]], align 8
2430 // CHECK4-NEXT:    [[TMP1:%.*]] = load i32*, i32** [[A_ADDR]], align 8
2431 // CHECK4-NEXT:    [[TMP2:%.*]] = load i32*, i32** [[B_ADDR]], align 8
2432 // CHECK4-NEXT:    [[TMP3:%.*]] = load i32*, i32** [[C_ADDR]], align 8
2433 // CHECK4-NEXT:    store i32* [[TMP1]], i32** [[TMP]], align 8
2434 // CHECK4-NEXT:    store i32* [[TMP3]], i32** [[_TMP1]], align 8
2435 // CHECK4-NEXT:    [[TMP4:%.*]] = load i32*, i32** [[TMP]], align 8
2436 // CHECK4-NEXT:    store i32* [[TMP4]], i32** [[_TMP2]], align 8
2437 // CHECK4-NEXT:    [[TMP5:%.*]] = load i32*, i32** [[_TMP1]], align 8
2438 // CHECK4-NEXT:    store i32* [[TMP5]], i32** [[_TMP3]], align 8
2439 // CHECK4-NEXT:    [[TMP6:%.*]] = load i32*, i32** [[_TMP2]], align 8
2440 // CHECK4-NEXT:    [[TMP7:%.*]] = load i32, i32* [[TMP6]], align 4
2441 // CHECK4-NEXT:    store i32 [[TMP7]], i32* [[DOTLINEAR_START]], align 4
2442 // CHECK4-NEXT:    [[TMP8:%.*]] = load i32, i32* [[TMP2]], align 4
2443 // CHECK4-NEXT:    store i32 [[TMP8]], i32* [[DOTLINEAR_START5]], align 4
2444 // CHECK4-NEXT:    [[TMP9:%.*]] = load i32*, i32** [[_TMP3]], align 8
2445 // CHECK4-NEXT:    [[TMP10:%.*]] = load i32, i32* [[TMP9]], align 4
2446 // CHECK4-NEXT:    store i32 [[TMP10]], i32* [[DOTLINEAR_START6]], align 4
2447 // CHECK4-NEXT:    store i32 0, i32* [[DOTOMP_LB]], align 4
2448 // CHECK4-NEXT:    store i32 1, i32* [[DOTOMP_UB]], align 4
2449 // CHECK4-NEXT:    store i32 1, i32* [[DOTOMP_STRIDE]], align 4
2450 // CHECK4-NEXT:    store i32 0, i32* [[DOTOMP_IS_LAST]], align 4
2451 // CHECK4-NEXT:    [[TMP11:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
2452 // CHECK4-NEXT:    [[TMP12:%.*]] = load i32, i32* [[TMP11]], align 4
2453 // CHECK4-NEXT:    call void @__kmpc_barrier(%struct.ident_t* @[[GLOB1]], i32 [[TMP12]])
2454 // CHECK4-NEXT:    store i32* [[A7]], i32** [[_TMP8]], align 8
2455 // CHECK4-NEXT:    store i32* [[C10]], i32** [[_TMP11]], align 8
2456 // CHECK4-NEXT:    call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB2]], i32 [[TMP12]], i32 34, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_LB]], i32* [[DOTOMP_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1)
2457 // CHECK4-NEXT:    [[TMP13:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
2458 // CHECK4-NEXT:    [[CMP:%.*]] = icmp sgt i32 [[TMP13]], 1
2459 // CHECK4-NEXT:    br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
2460 // CHECK4:       cond.true:
2461 // CHECK4-NEXT:    br label [[COND_END:%.*]]
2462 // CHECK4:       cond.false:
2463 // CHECK4-NEXT:    [[TMP14:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
2464 // CHECK4-NEXT:    br label [[COND_END]]
2465 // CHECK4:       cond.end:
2466 // CHECK4-NEXT:    [[COND:%.*]] = phi i32 [ 1, [[COND_TRUE]] ], [ [[TMP14]], [[COND_FALSE]] ]
2467 // CHECK4-NEXT:    store i32 [[COND]], i32* [[DOTOMP_UB]], align 4
2468 // CHECK4-NEXT:    [[TMP15:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4
2469 // CHECK4-NEXT:    store i32 [[TMP15]], i32* [[DOTOMP_IV]], align 4
2470 // CHECK4-NEXT:    br label [[OMP_INNER_FOR_COND:%.*]]
2471 // CHECK4:       omp.inner.for.cond:
2472 // CHECK4-NEXT:    [[TMP16:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
2473 // CHECK4-NEXT:    [[TMP17:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
2474 // CHECK4-NEXT:    [[CMP12:%.*]] = icmp sle i32 [[TMP16]], [[TMP17]]
2475 // CHECK4-NEXT:    br i1 [[CMP12]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
2476 // CHECK4:       omp.inner.for.body:
2477 // CHECK4-NEXT:    [[TMP18:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
2478 // CHECK4-NEXT:    [[MUL:%.*]] = mul nsw i32 [[TMP18]], 1
2479 // CHECK4-NEXT:    [[ADD:%.*]] = add nsw i32 0, [[MUL]]
2480 // CHECK4-NEXT:    store i32 [[ADD]], i32* [[I]], align 4
2481 // CHECK4-NEXT:    [[TMP19:%.*]] = load i32, i32* [[DOTLINEAR_START]], align 4
2482 // CHECK4-NEXT:    [[TMP20:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
2483 // CHECK4-NEXT:    [[MUL13:%.*]] = mul nsw i32 [[TMP20]], 1
2484 // CHECK4-NEXT:    [[ADD14:%.*]] = add nsw i32 [[TMP19]], [[MUL13]]
2485 // CHECK4-NEXT:    store i32 [[ADD14]], i32* [[A7]], align 4
2486 // CHECK4-NEXT:    [[TMP21:%.*]] = load i32, i32* [[DOTLINEAR_START5]], align 4
2487 // CHECK4-NEXT:    [[TMP22:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
2488 // CHECK4-NEXT:    [[MUL15:%.*]] = mul nsw i32 [[TMP22]], 1
2489 // CHECK4-NEXT:    [[ADD16:%.*]] = add nsw i32 [[TMP21]], [[MUL15]]
2490 // CHECK4-NEXT:    store i32 [[ADD16]], i32* [[B9]], align 4
2491 // CHECK4-NEXT:    [[TMP23:%.*]] = load i32, i32* [[DOTLINEAR_START6]], align 4
2492 // CHECK4-NEXT:    [[TMP24:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
2493 // CHECK4-NEXT:    [[MUL17:%.*]] = mul nsw i32 [[TMP24]], 1
2494 // CHECK4-NEXT:    [[ADD18:%.*]] = add nsw i32 [[TMP23]], [[MUL17]]
2495 // CHECK4-NEXT:    store i32 [[ADD18]], i32* [[C10]], align 4
2496 // CHECK4-NEXT:    [[TMP25:%.*]] = load i32*, i32** [[_TMP8]], align 8
2497 // CHECK4-NEXT:    [[TMP26:%.*]] = load i32, i32* [[TMP25]], align 4
2498 // CHECK4-NEXT:    [[INC:%.*]] = add nsw i32 [[TMP26]], 1
2499 // CHECK4-NEXT:    store i32 [[INC]], i32* [[TMP25]], align 4
2500 // CHECK4-NEXT:    [[TMP27:%.*]] = load i32, i32* [[B9]], align 4
2501 // CHECK4-NEXT:    [[DEC:%.*]] = add nsw i32 [[TMP27]], -1
2502 // CHECK4-NEXT:    store i32 [[DEC]], i32* [[B9]], align 4
2503 // CHECK4-NEXT:    [[TMP28:%.*]] = load i32*, i32** [[_TMP11]], align 8
2504 // CHECK4-NEXT:    [[TMP29:%.*]] = load i32, i32* [[TMP28]], align 4
2505 // CHECK4-NEXT:    [[DIV:%.*]] = sdiv i32 [[TMP29]], 1
2506 // CHECK4-NEXT:    store i32 [[DIV]], i32* [[TMP28]], align 4
2507 // CHECK4-NEXT:    br label [[OMP_BODY_CONTINUE:%.*]]
2508 // CHECK4:       omp.body.continue:
2509 // CHECK4-NEXT:    br label [[OMP_INNER_FOR_INC:%.*]]
2510 // CHECK4:       omp.inner.for.inc:
2511 // CHECK4-NEXT:    [[TMP30:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
2512 // CHECK4-NEXT:    [[ADD19:%.*]] = add nsw i32 [[TMP30]], 1
2513 // CHECK4-NEXT:    store i32 [[ADD19]], i32* [[DOTOMP_IV]], align 4
2514 // CHECK4-NEXT:    br label [[OMP_INNER_FOR_COND]]
2515 // CHECK4:       omp.inner.for.end:
2516 // CHECK4-NEXT:    br label [[OMP_LOOP_EXIT:%.*]]
2517 // CHECK4:       omp.loop.exit:
2518 // CHECK4-NEXT:    call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB2]], i32 [[TMP12]])
2519 // CHECK4-NEXT:    [[TMP31:%.*]] = load i32, i32* [[DOTOMP_IS_LAST]], align 4
2520 // CHECK4-NEXT:    [[TMP32:%.*]] = icmp ne i32 [[TMP31]], 0
2521 // CHECK4-NEXT:    br i1 [[TMP32]], label [[DOTOMP_LINEAR_PU:%.*]], label [[DOTOMP_LINEAR_PU_DONE:%.*]]
2522 // CHECK4:       .omp.linear.pu:
2523 // CHECK4-NEXT:    [[TMP33:%.*]] = load i32*, i32** [[_TMP2]], align 8
2524 // CHECK4-NEXT:    store i32* [[TMP33]], i32** [[_TMP20]], align 8
2525 // CHECK4-NEXT:    [[TMP34:%.*]] = load i32, i32* [[A7]], align 4
2526 // CHECK4-NEXT:    [[TMP35:%.*]] = load i32*, i32** [[_TMP20]], align 8
2527 // CHECK4-NEXT:    store i32 [[TMP34]], i32* [[TMP35]], align 4
2528 // CHECK4-NEXT:    [[TMP36:%.*]] = load i32, i32* [[B9]], align 4
2529 // CHECK4-NEXT:    store i32 [[TMP36]], i32* [[TMP2]], align 4
2530 // CHECK4-NEXT:    [[TMP37:%.*]] = load i32*, i32** [[_TMP3]], align 8
2531 // CHECK4-NEXT:    store i32* [[TMP37]], i32** [[_TMP21]], align 8
2532 // CHECK4-NEXT:    [[TMP38:%.*]] = load i32, i32* [[C10]], align 4
2533 // CHECK4-NEXT:    [[C22:%.*]] = getelementptr inbounds [[STRUCT_SS:%.*]], %struct.SS* [[TMP0]], i32 0, i32 2
2534 // CHECK4-NEXT:    [[TMP39:%.*]] = load i32*, i32** [[C22]], align 8
2535 // CHECK4-NEXT:    store i32 [[TMP38]], i32* [[TMP39]], align 4
2536 // CHECK4-NEXT:    br label [[DOTOMP_LINEAR_PU_DONE]]
2537 // CHECK4:       .omp.linear.pu.done:
2538 // CHECK4-NEXT:    call void @__kmpc_barrier(%struct.ident_t* @[[GLOB1]], i32 [[TMP12]])
2539 // CHECK4-NEXT:    ret void
2540 //
2541