clang/test/OpenMP/target_teams_generic_loop_uses_allocators_codegen.cpp

   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 // Test host codegen.
   3 // RUN: %clang_cc1 -verify -fopenmp -fopenmp-version=50 -x c++ -triple powerpc64le-unknown-unknown -fopenmp-targets=powerpc64le-ibm-linux-gnu -emit-llvm %s -o - | FileCheck %s --check-prefix CHECK
   4 // RUN: %clang_cc1 -fopenmp -fopenmp-version=50 -x c++ -triple powerpc64le-unknown-unknown -fopenmp-targets=powerpc64le-ibm-linux-gnu -emit-pch -o %t %s
   5 // RUN: %clang_cc1 -fopenmp -fopenmp-version=50 -x c++ -triple powerpc64le-unknown-unknown -fopenmp-targets=powerpc64le-ibm-linux-gnu -include-pch %t -verify %s -emit-llvm -o - | FileCheck %s --check-prefix CHECK
   6
   7 // expected-no-diagnostics
   8 #ifndef HEADER
   9 #define HEADER
  10
  11 enum omp_allocator_handle_t {
  12   omp_null_allocator = 0,
  13   omp_default_mem_alloc = 1,
  14   omp_large_cap_mem_alloc = 2,
  15   omp_const_mem_alloc = 3,
  16   omp_high_bw_mem_alloc = 4,
  17   omp_low_lat_mem_alloc = 5,
  18   omp_cgroup_mem_alloc = 6,
  19   omp_pteam_mem_alloc = 7,
  20   omp_thread_mem_alloc = 8,
  21   KMP_ALLOCATOR_MAX_HANDLE = __UINTPTR_MAX__
  22 };
  23
  24 typedef enum omp_alloctrait_key_t { omp_atk_sync_hint = 1,
  25                                     omp_atk_alignment = 2,
  26                                     omp_atk_access = 3,
  27                                     omp_atk_pool_size = 4,
  28                                     omp_atk_fallback = 5,
  29                                     omp_atk_fb_data = 6,
  30                                     omp_atk_pinned = 7,
  31                                     omp_atk_partition = 8
  32 } omp_alloctrait_key_t;
  33 typedef enum omp_alloctrait_value_t {
  34   omp_atv_false = 0,
  35   omp_atv_true = 1,
  36   omp_atv_default = 2,
  37   omp_atv_contended = 3,
  38   omp_atv_uncontended = 4,
  39   omp_atv_sequential = 5,
  40   omp_atv_private = 6,
  41   omp_atv_all = 7,
  42   omp_atv_thread = 8,
  43   omp_atv_pteam = 9,
  44   omp_atv_cgroup = 10,
  45   omp_atv_default_mem_fb = 11,
  46   omp_atv_null_fb = 12,
  47   omp_atv_abort_fb = 13,
  48   omp_atv_allocator_fb = 14,
  49   omp_atv_environment = 15,
  50   omp_atv_nearest = 16,
  51   omp_atv_blocked = 17,
  52   omp_atv_interleaved = 18
  53 } omp_alloctrait_value_t;
  54
  55 typedef struct omp_alloctrait_t {
  56   omp_alloctrait_key_t key;
  57   __UINTPTR_TYPE__ value;
  58 } omp_alloctrait_t;
  59
  60 // Just map the traits variable as a firstprivate variable.
  61
  62 void foo() {
  63   omp_alloctrait_t traits[10];
  64   omp_allocator_handle_t my_allocator;
  65
  66 #pragma omp target teams loop uses_allocators(omp_null_allocator, omp_thread_mem_alloc, my_allocator(traits))
  67   for (int i = 0; i < 10; ++i)
  68     ;
  69 }
  70
  71
  72 // Destroy allocator upon exit from the region.
  73
  74 #endif
  75 // CHECK-64-LABEL: define {{[^@]+}}@_Z3foov
  76 // CHECK-64-SAME: () #[[ATTR0:[0-9]+]] {
  77 // CHECK-64-NEXT:  entry:
  78 // CHECK-64-NEXT:    [[TRAITS:%.*]] = alloca [10 x %struct.omp_alloctrait_t], align 8
  79 // CHECK-64-NEXT:    [[MY_ALLOCATOR:%.*]] = alloca i64, align 8
  80 // CHECK-64-NEXT:    [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [1 x ptr], align 8
  81 // CHECK-64-NEXT:    [[DOTOFFLOAD_PTRS:%.*]] = alloca [1 x ptr], align 8
  82 // CHECK-64-NEXT:    [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [1 x ptr], align 8
  83 // CHECK-64-NEXT:    [[TMP:%.*]] = alloca i32, align 4
  84 // CHECK-64-NEXT:    [[TMP0:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
  85 // CHECK-64-NEXT:    store ptr [[TRAITS]], ptr [[TMP0]], align 8
  86 // CHECK-64-NEXT:    [[TMP1:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
  87 // CHECK-64-NEXT:    store ptr [[TRAITS]], ptr [[TMP1]], align 8
  88 // CHECK-64-NEXT:    [[TMP2:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0
  89 // CHECK-64-NEXT:    store ptr null, ptr [[TMP2]], align 8
  90 // CHECK-64-NEXT:    [[TMP3:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
  91 // CHECK-64-NEXT:    [[TMP4:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
  92 // CHECK-64-NEXT:    [[KERNEL_ARGS:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS:%.*]], align 8
  93 // CHECK-64-NEXT:    [[TMP5:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 0
  94 // CHECK-64-NEXT:    store i32 2, ptr [[TMP5]], align 4
  95 // CHECK-64-NEXT:    [[TMP6:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 1
  96 // CHECK-64-NEXT:    store i32 1, ptr [[TMP6]], align 4
  97 // CHECK-64-NEXT:    [[TMP7:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 2
  98 // CHECK-64-NEXT:    store ptr [[TMP3]], ptr [[TMP7]], align 8
  99 // CHECK-64-NEXT:    [[TMP8:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 3
 100 // CHECK-64-NEXT:    store ptr [[TMP4]], ptr [[TMP8]], align 8
 101 // CHECK-64-NEXT:    [[TMP9:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 4
 102 // CHECK-64-NEXT:    store ptr @.offload_sizes, ptr [[TMP9]], align 8
 103 // CHECK-64-NEXT:    [[TMP10:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 5
 104 // CHECK-64-NEXT:    store ptr @.offload_maptypes, ptr [[TMP10]], align 8
 105 // CHECK-64-NEXT:    [[TMP11:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 6
 106 // CHECK-64-NEXT:    store ptr null, ptr [[TMP11]], align 8
 107 // CHECK-64-NEXT:    [[TMP12:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 7
 108 // CHECK-64-NEXT:    store ptr null, ptr [[TMP12]], align 8
 109 // CHECK-64-NEXT:    [[TMP13:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 8
 110 // CHECK-64-NEXT:    store i64 10, ptr [[TMP13]], align 8
 111 // CHECK-64-NEXT:    [[TMP14:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 9
 112 // CHECK-64-NEXT:    store i64 0, ptr [[TMP14]], align 8
 113 // CHECK-64-NEXT:    [[TMP15:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 10
 114 // CHECK-64-NEXT:    store [3 x i32] zeroinitializer, ptr [[TMP15]], align 4
 115 // CHECK-64-NEXT:    [[TMP16:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 11
 116 // CHECK-64-NEXT:    store [3 x i32] zeroinitializer, ptr [[TMP16]], align 4
 117 // CHECK-64-NEXT:    [[TMP17:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 12
 118 // CHECK-64-NEXT:    store i32 0, ptr [[TMP17]], align 4
 119 // CHECK-64-NEXT:    [[TMP18:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB1:[0-9]+]], i64 -1, i32 0, i32 0, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3foov_l73.region_id, ptr [[KERNEL_ARGS]])
 120 // CHECK-64-NEXT:    [[TMP19:%.*]] = icmp ne i32 [[TMP18]], 0
 121 // CHECK-64-NEXT:    br i1 [[TMP19]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
 122 // CHECK-64:       omp_offload.failed:
 123 // CHECK-64-NEXT:    call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3foov_l73(ptr [[TRAITS]]) #[[ATTR2:[0-9]+]]
 124 // CHECK-64-NEXT:    br label [[OMP_OFFLOAD_CONT]]
 125 // CHECK-64:       omp_offload.cont:
 126 // CHECK-64-NEXT:    ret void
 127 // CHECK-64-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3foov_l73
 128 // CHECK-64-SAME: (ptr noundef nonnull align 8 dereferenceable(160) [[TRAITS:%.*]]) #[[ATTR1:[0-9]+]] {
 129 // CHECK-64-NEXT:  entry:
 130 // CHECK-64-NEXT:    [[TRAITS_ADDR:%.*]] = alloca ptr, align 8
 131 // CHECK-64-NEXT:    [[MY_ALLOCATOR:%.*]] = alloca i64, align 8
 132 // CHECK-64-NEXT:    [[TMP0:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB1]])
 133 // CHECK-64-NEXT:    store ptr [[TRAITS]], ptr [[TRAITS_ADDR]], align 8
 134 // CHECK-64-NEXT:    [[TMP1:%.*]] = load ptr, ptr [[TRAITS_ADDR]], align 8
 135 // CHECK-64-NEXT:    [[TMP2:%.*]] = load ptr, ptr [[TMP1]], align 8
 136 // CHECK-64-NEXT:    [[TMP3:%.*]] = call ptr @__kmpc_init_allocator(i32 [[TMP0]], ptr null, i32 10, ptr [[TMP2]])
 137 // CHECK-64-NEXT:    [[CONV:%.*]] = ptrtoint ptr [[TMP3]] to i64
 138 // CHECK-64-NEXT:    store i64 [[CONV]], ptr [[MY_ALLOCATOR]], align 8
 139 // CHECK-64-NEXT:    call void (ptr, i32, ptr, ...) @__kmpc_fork_teams(ptr @[[GLOB1]], i32 0, ptr @.omp_outlined.)
 140 // CHECK-64-NEXT:    [[TMP4:%.*]] = load i64, ptr [[MY_ALLOCATOR]], align 8
 141 // CHECK-64-NEXT:    [[CONV1:%.*]] = inttoptr i64 [[TMP4]] to ptr
 142 // CHECK-64-NEXT:    call void @__kmpc_destroy_allocator(i32 [[TMP0]], ptr [[CONV1]])
 143 // CHECK-64-NEXT:    ret void
 144 // CHECK-64-LABEL: define {{[^@]+}}@.omp_outlined.
 145 // CHECK-64-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]]) #[[ATTR1]] {
 146 // CHECK-64-NEXT:  entry:
 147 // CHECK-64-NEXT:    [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 8
 148 // CHECK-64-NEXT:    [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 8
 149 // CHECK-64-NEXT:    [[DOTOMP_IV:%.*]] = alloca i32, align 4
 150 // CHECK-64-NEXT:    [[TMP:%.*]] = alloca i32, align 4
 151 // CHECK-64-NEXT:    [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4
 152 // CHECK-64-NEXT:    [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4
 153 // CHECK-64-NEXT:    [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
 154 // CHECK-64-NEXT:    [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
 155 // CHECK-64-NEXT:    [[I:%.*]] = alloca i32, align 4
 156 // CHECK-64-NEXT:    store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 8
 157 // CHECK-64-NEXT:    store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 8
 158 // CHECK-64-NEXT:    store i32 0, ptr [[DOTOMP_COMB_LB]], align 4
 159 // CHECK-64-NEXT:    store i32 9, ptr [[DOTOMP_COMB_UB]], align 4
 160 // CHECK-64-NEXT:    store i32 1, ptr [[DOTOMP_STRIDE]], align 4
 161 // CHECK-64-NEXT:    store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
 162 // CHECK-64-NEXT:    [[TMP0:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
 163 // CHECK-64-NEXT:    [[TMP1:%.*]] = load i32, ptr [[TMP0]], align 4
 164 // CHECK-64-NEXT:    call void @__kmpc_for_static_init_4(ptr @[[GLOB2:[0-9]+]], i32 [[TMP1]], i32 92, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
 165 // CHECK-64-NEXT:    [[TMP2:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
 166 // CHECK-64-NEXT:    [[CMP:%.*]] = icmp sgt i32 [[TMP2]], 9
 167 // CHECK-64-NEXT:    br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
 168 // CHECK-64:       cond.true:
 169 // CHECK-64-NEXT:    br label [[COND_END:%.*]]
 170 // CHECK-64:       cond.false:
 171 // CHECK-64-NEXT:    [[TMP3:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
 172 // CHECK-64-NEXT:    br label [[COND_END]]
 173 // CHECK-64:       cond.end:
 174 // CHECK-64-NEXT:    [[COND:%.*]] = phi i32 [ 9, [[COND_TRUE]] ], [ [[TMP3]], [[COND_FALSE]] ]
 175 // CHECK-64-NEXT:    store i32 [[COND]], ptr [[DOTOMP_COMB_UB]], align 4
 176 // CHECK-64-NEXT:    [[TMP4:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
 177 // CHECK-64-NEXT:    store i32 [[TMP4]], ptr [[DOTOMP_IV]], align 4
 178 // CHECK-64-NEXT:    br label [[OMP_INNER_FOR_COND:%.*]]
 179 // CHECK-64:       omp.inner.for.cond:
 180 // CHECK-64-NEXT:    [[TMP5:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
 181 // CHECK-64-NEXT:    [[TMP6:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
 182 // CHECK-64-NEXT:    [[CMP1:%.*]] = icmp sle i32 [[TMP5]], [[TMP6]]
 183 // CHECK-64-NEXT:    br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
 184 // CHECK-64:       omp.inner.for.body:
 185 // CHECK-64-NEXT:    [[TMP7:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
 186 // CHECK-64-NEXT:    [[TMP8:%.*]] = zext i32 [[TMP7]] to i64
 187 // CHECK-64-NEXT:    [[TMP9:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
 188 // CHECK-64-NEXT:    [[TMP10:%.*]] = zext i32 [[TMP9]] to i64
 189 // CHECK-64-NEXT:    call void (ptr, i32, ptr, ...) @__kmpc_fork_call(ptr @[[GLOB1]], i32 2, ptr @.omp_outlined..1, i64 [[TMP8]], i64 [[TMP10]])
 190 // CHECK-64-NEXT:    br label [[OMP_INNER_FOR_INC:%.*]]
 191 // CHECK-64:       omp.inner.for.inc:
 192 // CHECK-64-NEXT:    [[TMP11:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
 193 // CHECK-64-NEXT:    [[TMP12:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4
 194 // CHECK-64-NEXT:    [[ADD:%.*]] = add nsw i32 [[TMP11]], [[TMP12]]
 195 // CHECK-64-NEXT:    store i32 [[ADD]], ptr [[DOTOMP_IV]], align 4
 196 // CHECK-64-NEXT:    br label [[OMP_INNER_FOR_COND]]
 197 // CHECK-64:       omp.inner.for.end:
 198 // CHECK-64-NEXT:    br label [[OMP_LOOP_EXIT:%.*]]
 199 // CHECK-64:       omp.loop.exit:
 200 // CHECK-64-NEXT:    call void @__kmpc_for_static_fini(ptr @[[GLOB3:[0-9]+]], i32 [[TMP1]])
 201 // CHECK-64-NEXT:    ret void
 202 // CHECK-64-LABEL: define {{[^@]+}}@.omp_outlined..1
 203 // CHECK-64-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], i64 noundef [[DOTPREVIOUS_LB_:%.*]], i64 noundef [[DOTPREVIOUS_UB_:%.*]]) #[[ATTR1]] {
 204 // CHECK-64-NEXT:  entry:
 205 // CHECK-64-NEXT:    [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 8
 206 // CHECK-64-NEXT:    [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 8
 207 // CHECK-64-NEXT:    [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i64, align 8
 208 // CHECK-64-NEXT:    [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i64, align 8
 209 // CHECK-64-NEXT:    [[DOTOMP_IV:%.*]] = alloca i32, align 4
 210 // CHECK-64-NEXT:    [[TMP:%.*]] = alloca i32, align 4
 211 // CHECK-64-NEXT:    [[DOTOMP_LB:%.*]] = alloca i32, align 4
 212 // CHECK-64-NEXT:    [[DOTOMP_UB:%.*]] = alloca i32, align 4
 213 // CHECK-64-NEXT:    [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
 214 // CHECK-64-NEXT:    [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
 215 // CHECK-64-NEXT:    [[I:%.*]] = alloca i32, align 4
 216 // CHECK-64-NEXT:    store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 8
 217 // CHECK-64-NEXT:    store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 8
 218 // CHECK-64-NEXT:    store i64 [[DOTPREVIOUS_LB_]], ptr [[DOTPREVIOUS_LB__ADDR]], align 8
 219 // CHECK-64-NEXT:    store i64 [[DOTPREVIOUS_UB_]], ptr [[DOTPREVIOUS_UB__ADDR]], align 8
 220 // CHECK-64-NEXT:    store i32 0, ptr [[DOTOMP_LB]], align 4
 221 // CHECK-64-NEXT:    store i32 9, ptr [[DOTOMP_UB]], align 4
 222 // CHECK-64-NEXT:    [[TMP0:%.*]] = load i64, ptr [[DOTPREVIOUS_LB__ADDR]], align 8
 223 // CHECK-64-NEXT:    [[CONV:%.*]] = trunc i64 [[TMP0]] to i32
 224 // CHECK-64-NEXT:    [[TMP1:%.*]] = load i64, ptr [[DOTPREVIOUS_UB__ADDR]], align 8
 225 // CHECK-64-NEXT:    [[CONV1:%.*]] = trunc i64 [[TMP1]] to i32
 226 // CHECK-64-NEXT:    store i32 [[CONV]], ptr [[DOTOMP_LB]], align 4
 227 // CHECK-64-NEXT:    store i32 [[CONV1]], ptr [[DOTOMP_UB]], align 4
 228 // CHECK-64-NEXT:    store i32 1, ptr [[DOTOMP_STRIDE]], align 4
 229 // CHECK-64-NEXT:    store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
 230 // CHECK-64-NEXT:    [[TMP2:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
 231 // CHECK-64-NEXT:    [[TMP3:%.*]] = load i32, ptr [[TMP2]], align 4
 232 // CHECK-64-NEXT:    call void @__kmpc_for_static_init_4(ptr @[[GLOB3]], i32 [[TMP3]], i32 34, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
 233 // CHECK-64-NEXT:    [[TMP4:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
 234 // CHECK-64-NEXT:    [[CMP:%.*]] = icmp sgt i32 [[TMP4]], 9
 235 // CHECK-64-NEXT:    br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
 236 // CHECK-64:       cond.true:
 237 // CHECK-64-NEXT:    br label [[COND_END:%.*]]
 238 // CHECK-64:       cond.false:
 239 // CHECK-64-NEXT:    [[TMP5:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
 240 // CHECK-64-NEXT:    br label [[COND_END]]
 241 // CHECK-64:       cond.end:
 242 // CHECK-64-NEXT:    [[COND:%.*]] = phi i32 [ 9, [[COND_TRUE]] ], [ [[TMP5]], [[COND_FALSE]] ]
 243 // CHECK-64-NEXT:    store i32 [[COND]], ptr [[DOTOMP_UB]], align 4
 244 // CHECK-64-NEXT:    [[TMP6:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4
 245 // CHECK-64-NEXT:    store i32 [[TMP6]], ptr [[DOTOMP_IV]], align 4
 246 // CHECK-64-NEXT:    br label [[OMP_INNER_FOR_COND:%.*]]
 247 // CHECK-64:       omp.inner.for.cond:
 248 // CHECK-64-NEXT:    [[TMP7:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
 249 // CHECK-64-NEXT:    [[TMP8:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
 250 // CHECK-64-NEXT:    [[CMP2:%.*]] = icmp sle i32 [[TMP7]], [[TMP8]]
 251 // CHECK-64-NEXT:    br i1 [[CMP2]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
 252 // CHECK-64:       omp.inner.for.body:
 253 // CHECK-64-NEXT:    [[TMP9:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
 254 // CHECK-64-NEXT:    [[MUL:%.*]] = mul nsw i32 [[TMP9]], 1
 255 // CHECK-64-NEXT:    [[ADD:%.*]] = add nsw i32 0, [[MUL]]
 256 // CHECK-64-NEXT:    store i32 [[ADD]], ptr [[I]], align 4
 257 // CHECK-64-NEXT:    br label [[OMP_BODY_CONTINUE:%.*]]
 258 // CHECK-64:       omp.body.continue:
 259 // CHECK-64-NEXT:    br label [[OMP_INNER_FOR_INC:%.*]]
 260 // CHECK-64:       omp.inner.for.inc:
 261 // CHECK-64-NEXT:    [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
 262 // CHECK-64-NEXT:    [[ADD3:%.*]] = add nsw i32 [[TMP10]], 1
 263 // CHECK-64-NEXT:    store i32 [[ADD3]], ptr [[DOTOMP_IV]], align 4
 264 // CHECK-64-NEXT:    br label [[OMP_INNER_FOR_COND]]
 265 // CHECK-64:       omp.inner.for.end:
 266 // CHECK-64-NEXT:    br label [[OMP_LOOP_EXIT:%.*]]
 267 // CHECK-64:       omp.loop.exit:
 268 // CHECK-64-NEXT:    call void @__kmpc_for_static_fini(ptr @[[GLOB3]], i32 [[TMP3]])
 269 // CHECK-64-NEXT:    ret void
 270 // CHECK-64-LABEL: define {{[^@]+}}@.omp_offloading.requires_reg
 271 // CHECK-64-SAME: () #[[ATTR3:[0-9]+]] {
 272 // CHECK-64-NEXT:  entry:
 273 // CHECK-64-NEXT:    call void @__tgt_register_requires(i64 1)
 274 // CHECK-64-NEXT:    ret void
 275 // CHECK-LABEL: define {{[^@]+}}@_Z3foov
 276 // CHECK-SAME: () #[[ATTR0:[0-9]+]] {
 277 // CHECK-NEXT:  entry:
 278 // CHECK-NEXT:    [[TRAITS:%.*]] = alloca [10 x %struct.omp_alloctrait_t], align 8
 279 // CHECK-NEXT:    [[MY_ALLOCATOR:%.*]] = alloca i64, align 8
 280 // CHECK-NEXT:    [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [1 x ptr], align 8
 281 // CHECK-NEXT:    [[DOTOFFLOAD_PTRS:%.*]] = alloca [1 x ptr], align 8
 282 // CHECK-NEXT:    [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [1 x ptr], align 8
 283 // CHECK-NEXT:    [[TMP:%.*]] = alloca i32, align 4
 284 // CHECK-NEXT:    [[KERNEL_ARGS:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS:%.*]], align 8
 285 // CHECK-NEXT:    [[TMP0:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
 286 // CHECK-NEXT:    store ptr [[TRAITS]], ptr [[TMP0]], align 8
 287 // CHECK-NEXT:    [[TMP1:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
 288 // CHECK-NEXT:    store ptr [[TRAITS]], ptr [[TMP1]], align 8
 289 // CHECK-NEXT:    [[TMP2:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0
 290 // CHECK-NEXT:    store ptr null, ptr [[TMP2]], align 8
 291 // CHECK-NEXT:    [[TMP3:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
 292 // CHECK-NEXT:    [[TMP4:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
 293 // CHECK-NEXT:    [[TMP5:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 0
 294 // CHECK-NEXT:    store i32 2, ptr [[TMP5]], align 4
 295 // CHECK-NEXT:    [[TMP6:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 1
 296 // CHECK-NEXT:    store i32 1, ptr [[TMP6]], align 4
 297 // CHECK-NEXT:    [[TMP7:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 2
 298 // CHECK-NEXT:    store ptr [[TMP3]], ptr [[TMP7]], align 8
 299 // CHECK-NEXT:    [[TMP8:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 3
 300 // CHECK-NEXT:    store ptr [[TMP4]], ptr [[TMP8]], align 8
 301 // CHECK-NEXT:    [[TMP9:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 4
 302 // CHECK-NEXT:    store ptr @.offload_sizes, ptr [[TMP9]], align 8
 303 // CHECK-NEXT:    [[TMP10:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 5
 304 // CHECK-NEXT:    store ptr @.offload_maptypes, ptr [[TMP10]], align 8
 305 // CHECK-NEXT:    [[TMP11:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 6
 306 // CHECK-NEXT:    store ptr null, ptr [[TMP11]], align 8
 307 // CHECK-NEXT:    [[TMP12:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 7
 308 // CHECK-NEXT:    store ptr null, ptr [[TMP12]], align 8
 309 // CHECK-NEXT:    [[TMP13:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 8
 310 // CHECK-NEXT:    store i64 10, ptr [[TMP13]], align 8
 311 // CHECK-NEXT:    [[TMP14:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 9
 312 // CHECK-NEXT:    store i64 0, ptr [[TMP14]], align 8
 313 // CHECK-NEXT:    [[TMP15:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 10
 314 // CHECK-NEXT:    store [3 x i32] zeroinitializer, ptr [[TMP15]], align 4
 315 // CHECK-NEXT:    [[TMP16:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 11
 316 // CHECK-NEXT:    store [3 x i32] zeroinitializer, ptr [[TMP16]], align 4
 317 // CHECK-NEXT:    [[TMP17:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 12
 318 // CHECK-NEXT:    store i32 0, ptr [[TMP17]], align 4
 319 // CHECK-NEXT:    [[TMP18:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB1:[0-9]+]], i64 -1, i32 0, i32 0, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3foov_l66.region_id, ptr [[KERNEL_ARGS]])
 320 // CHECK-NEXT:    [[TMP19:%.*]] = icmp ne i32 [[TMP18]], 0
 321 // CHECK-NEXT:    br i1 [[TMP19]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
 322 // CHECK:       omp_offload.failed:
 323 // CHECK-NEXT:    call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3foov_l66(ptr [[TRAITS]]) #[[ATTR2:[0-9]+]]
 324 // CHECK-NEXT:    br label [[OMP_OFFLOAD_CONT]]
 325 // CHECK:       omp_offload.cont:
 326 // CHECK-NEXT:    ret void
 327 //
 328 //
 329 // CHECK-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3foov_l66
 330 // CHECK-SAME: (ptr noundef nonnull align 8 dereferenceable(160) [[TRAITS:%.*]]) #[[ATTR1:[0-9]+]] {
 331 // CHECK-NEXT:  entry:
 332 // CHECK-NEXT:    [[TRAITS_ADDR:%.*]] = alloca ptr, align 8
 333 // CHECK-NEXT:    [[MY_ALLOCATOR:%.*]] = alloca i64, align 8
 334 // CHECK-NEXT:    [[TMP0:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB1]])
 335 // CHECK-NEXT:    store ptr [[TRAITS]], ptr [[TRAITS_ADDR]], align 8
 336 // CHECK-NEXT:    [[TMP1:%.*]] = load ptr, ptr [[TRAITS_ADDR]], align 8
 337 // CHECK-NEXT:    [[TMP2:%.*]] = call ptr @__kmpc_init_allocator(i32 [[TMP0]], ptr null, i32 10, ptr [[TMP1]])
 338 // CHECK-NEXT:    [[CONV:%.*]] = ptrtoint ptr [[TMP2]] to i64
 339 // CHECK-NEXT:    store i64 [[CONV]], ptr [[MY_ALLOCATOR]], align 8
 340 // CHECK-NEXT:    call void (ptr, i32, ptr, ...) @__kmpc_fork_teams(ptr @[[GLOB1]], i32 0, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3foov_l66.omp_outlined)
 341 // CHECK-NEXT:    [[TMP3:%.*]] = load i64, ptr [[MY_ALLOCATOR]], align 8
 342 // CHECK-NEXT:    [[CONV1:%.*]] = inttoptr i64 [[TMP3]] to ptr
 343 // CHECK-NEXT:    call void @__kmpc_destroy_allocator(i32 [[TMP0]], ptr [[CONV1]])
 344 // CHECK-NEXT:    ret void
 345 //
 346 //
 347 // CHECK-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3foov_l66.omp_outlined
 348 // CHECK-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]]) #[[ATTR1]] {
 349 // CHECK-NEXT:  entry:
 350 // CHECK-NEXT:    [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 8
 351 // CHECK-NEXT:    [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 8
 352 // CHECK-NEXT:    [[DOTOMP_IV:%.*]] = alloca i32, align 4
 353 // CHECK-NEXT:    [[TMP:%.*]] = alloca i32, align 4
 354 // CHECK-NEXT:    [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4
 355 // CHECK-NEXT:    [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4
 356 // CHECK-NEXT:    [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
 357 // CHECK-NEXT:    [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
 358 // CHECK-NEXT:    [[I:%.*]] = alloca i32, align 4
 359 // CHECK-NEXT:    store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 8
 360 // CHECK-NEXT:    store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 8
 361 // CHECK-NEXT:    store i32 0, ptr [[DOTOMP_COMB_LB]], align 4
 362 // CHECK-NEXT:    store i32 9, ptr [[DOTOMP_COMB_UB]], align 4
 363 // CHECK-NEXT:    store i32 1, ptr [[DOTOMP_STRIDE]], align 4
 364 // CHECK-NEXT:    store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
 365 // CHECK-NEXT:    [[TMP0:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
 366 // CHECK-NEXT:    [[TMP1:%.*]] = load i32, ptr [[TMP0]], align 4
 367 // CHECK-NEXT:    call void @__kmpc_for_static_init_4(ptr @[[GLOB2:[0-9]+]], i32 [[TMP1]], i32 92, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
 368 // CHECK-NEXT:    [[TMP2:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
 369 // CHECK-NEXT:    [[CMP:%.*]] = icmp sgt i32 [[TMP2]], 9
 370 // CHECK-NEXT:    br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
 371 // CHECK:       cond.true:
 372 // CHECK-NEXT:    br label [[COND_END:%.*]]
 373 // CHECK:       cond.false:
 374 // CHECK-NEXT:    [[TMP3:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
 375 // CHECK-NEXT:    br label [[COND_END]]
 376 // CHECK:       cond.end:
 377 // CHECK-NEXT:    [[COND:%.*]] = phi i32 [ 9, [[COND_TRUE]] ], [ [[TMP3]], [[COND_FALSE]] ]
 378 // CHECK-NEXT:    store i32 [[COND]], ptr [[DOTOMP_COMB_UB]], align 4
 379 // CHECK-NEXT:    [[TMP4:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
 380 // CHECK-NEXT:    store i32 [[TMP4]], ptr [[DOTOMP_IV]], align 4
 381 // CHECK-NEXT:    br label [[OMP_INNER_FOR_COND:%.*]]
 382 // CHECK:       omp.inner.for.cond:
 383 // CHECK-NEXT:    [[TMP5:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
 384 // CHECK-NEXT:    [[TMP6:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
 385 // CHECK-NEXT:    [[CMP1:%.*]] = icmp sle i32 [[TMP5]], [[TMP6]]
 386 // CHECK-NEXT:    br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
 387 // CHECK:       omp.inner.for.body:
 388 // CHECK-NEXT:    [[TMP7:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
 389 // CHECK-NEXT:    [[TMP8:%.*]] = zext i32 [[TMP7]] to i64
 390 // CHECK-NEXT:    [[TMP9:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
 391 // CHECK-NEXT:    [[TMP10:%.*]] = zext i32 [[TMP9]] to i64
 392 // CHECK-NEXT:    call void (ptr, i32, ptr, ...) @__kmpc_fork_call(ptr @[[GLOB1]], i32 2, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3foov_l66.omp_outlined.omp_outlined, i64 [[TMP8]], i64 [[TMP10]])
 393 // CHECK-NEXT:    br label [[OMP_INNER_FOR_INC:%.*]]
 394 // CHECK:       omp.inner.for.inc:
 395 // CHECK-NEXT:    [[TMP11:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
 396 // CHECK-NEXT:    [[TMP12:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4
 397 // CHECK-NEXT:    [[ADD:%.*]] = add nsw i32 [[TMP11]], [[TMP12]]
 398 // CHECK-NEXT:    store i32 [[ADD]], ptr [[DOTOMP_IV]], align 4
 399 // CHECK-NEXT:    br label [[OMP_INNER_FOR_COND]]
 400 // CHECK:       omp.inner.for.end:
 401 // CHECK-NEXT:    br label [[OMP_LOOP_EXIT:%.*]]
 402 // CHECK:       omp.loop.exit:
 403 // CHECK-NEXT:    call void @__kmpc_for_static_fini(ptr @[[GLOB3:[0-9]+]], i32 [[TMP1]])
 404 // CHECK-NEXT:    ret void
 405 //
 406 //
 407 // CHECK-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3foov_l66.omp_outlined.omp_outlined
 408 // CHECK-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], i64 noundef [[DOTPREVIOUS_LB_:%.*]], i64 noundef [[DOTPREVIOUS_UB_:%.*]]) #[[ATTR1]] {
 409 // CHECK-NEXT:  entry:
 410 // CHECK-NEXT:    [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 8
 411 // CHECK-NEXT:    [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 8
 412 // CHECK-NEXT:    [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i64, align 8
 413 // CHECK-NEXT:    [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i64, align 8
 414 // CHECK-NEXT:    [[DOTOMP_IV:%.*]] = alloca i32, align 4
 415 // CHECK-NEXT:    [[TMP:%.*]] = alloca i32, align 4
 416 // CHECK-NEXT:    [[DOTOMP_LB:%.*]] = alloca i32, align 4
 417 // CHECK-NEXT:    [[DOTOMP_UB:%.*]] = alloca i32, align 4
 418 // CHECK-NEXT:    [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
 419 // CHECK-NEXT:    [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
 420 // CHECK-NEXT:    [[I:%.*]] = alloca i32, align 4
 421 // CHECK-NEXT:    store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 8
 422 // CHECK-NEXT:    store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 8
 423 // CHECK-NEXT:    store i64 [[DOTPREVIOUS_LB_]], ptr [[DOTPREVIOUS_LB__ADDR]], align 8
 424 // CHECK-NEXT:    store i64 [[DOTPREVIOUS_UB_]], ptr [[DOTPREVIOUS_UB__ADDR]], align 8
 425 // CHECK-NEXT:    store i32 0, ptr [[DOTOMP_LB]], align 4
 426 // CHECK-NEXT:    store i32 9, ptr [[DOTOMP_UB]], align 4
 427 // CHECK-NEXT:    [[TMP0:%.*]] = load i64, ptr [[DOTPREVIOUS_LB__ADDR]], align 8
 428 // CHECK-NEXT:    [[CONV:%.*]] = trunc i64 [[TMP0]] to i32
 429 // CHECK-NEXT:    [[TMP1:%.*]] = load i64, ptr [[DOTPREVIOUS_UB__ADDR]], align 8
 430 // CHECK-NEXT:    [[CONV1:%.*]] = trunc i64 [[TMP1]] to i32
 431 // CHECK-NEXT:    store i32 [[CONV]], ptr [[DOTOMP_LB]], align 4
 432 // CHECK-NEXT:    store i32 [[CONV1]], ptr [[DOTOMP_UB]], align 4
 433 // CHECK-NEXT:    store i32 1, ptr [[DOTOMP_STRIDE]], align 4
 434 // CHECK-NEXT:    store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
 435 // CHECK-NEXT:    [[TMP2:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
 436 // CHECK-NEXT:    [[TMP3:%.*]] = load i32, ptr [[TMP2]], align 4
 437 // CHECK-NEXT:    call void @__kmpc_for_static_init_4(ptr @[[GLOB3]], i32 [[TMP3]], i32 34, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
 438 // CHECK-NEXT:    [[TMP4:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
 439 // CHECK-NEXT:    [[CMP:%.*]] = icmp sgt i32 [[TMP4]], 9
 440 // CHECK-NEXT:    br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
 441 // CHECK:       cond.true:
 442 // CHECK-NEXT:    br label [[COND_END:%.*]]
 443 // CHECK:       cond.false:
 444 // CHECK-NEXT:    [[TMP5:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
 445 // CHECK-NEXT:    br label [[COND_END]]
 446 // CHECK:       cond.end:
 447 // CHECK-NEXT:    [[COND:%.*]] = phi i32 [ 9, [[COND_TRUE]] ], [ [[TMP5]], [[COND_FALSE]] ]
 448 // CHECK-NEXT:    store i32 [[COND]], ptr [[DOTOMP_UB]], align 4
 449 // CHECK-NEXT:    [[TMP6:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4
 450 // CHECK-NEXT:    store i32 [[TMP6]], ptr [[DOTOMP_IV]], align 4
 451 // CHECK-NEXT:    br label [[OMP_INNER_FOR_COND:%.*]]
 452 // CHECK:       omp.inner.for.cond:
 453 // CHECK-NEXT:    [[TMP7:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
 454 // CHECK-NEXT:    [[TMP8:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
 455 // CHECK-NEXT:    [[CMP2:%.*]] = icmp sle i32 [[TMP7]], [[TMP8]]
 456 // CHECK-NEXT:    br i1 [[CMP2]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
 457 // CHECK:       omp.inner.for.body:
 458 // CHECK-NEXT:    [[TMP9:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
 459 // CHECK-NEXT:    [[MUL:%.*]] = mul nsw i32 [[TMP9]], 1
 460 // CHECK-NEXT:    [[ADD:%.*]] = add nsw i32 0, [[MUL]]
 461 // CHECK-NEXT:    store i32 [[ADD]], ptr [[I]], align 4
 462 // CHECK-NEXT:    br label [[OMP_BODY_CONTINUE:%.*]]
 463 // CHECK:       omp.body.continue:
 464 // CHECK-NEXT:    br label [[OMP_INNER_FOR_INC:%.*]]
 465 // CHECK:       omp.inner.for.inc:
 466 // CHECK-NEXT:    [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
 467 // CHECK-NEXT:    [[ADD3:%.*]] = add nsw i32 [[TMP10]], 1
 468 // CHECK-NEXT:    store i32 [[ADD3]], ptr [[DOTOMP_IV]], align 4
 469 // CHECK-NEXT:    br label [[OMP_INNER_FOR_COND]]
 470 // CHECK:       omp.inner.for.end:
 471 // CHECK-NEXT:    br label [[OMP_LOOP_EXIT:%.*]]
 472 // CHECK:       omp.loop.exit:
 473 // CHECK-NEXT:    call void @__kmpc_for_static_fini(ptr @[[GLOB3]], i32 [[TMP3]])
 474 // CHECK-NEXT:    ret void
 475 //
 476 //
 477 // CHECK-LABEL: define {{[^@]+}}@.omp_offloading.requires_reg
 478 // CHECK-SAME: () #[[ATTR3:[0-9]+]] {
 479 // CHECK-NEXT:  entry:
 480 // CHECK-NEXT:    call void @__tgt_register_requires(i64 1)
 481 // CHECK-NEXT:    ret void
 482 //