xref: /llvm-project/clang/test/OpenMP/allocate_modifiers_codegen.cpp (revision ad38e24eb74e97148faec97c4f843b87768b6e9b)

// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --include-generated-funcs --replace-value-regex "__omp_offloading_[0-9a-z]+_[0-9a-z]+" "reduction_size[.].+[.]" "pl_cond[.].+[.|,]" --prefix-filecheck-ir-name _ --version 5
// RUN: %clang_cc1 -verify -fopenmp -fopenmp-version=52 -triple x86_64-unknown-linux-gnu -emit-llvm -o - %s | FileCheck %s
// RUN: %clang_cc1 -fopenmp -fopenmp-version=52 -triple x86_64-unknown-linux-gnu -x c++ -std=c++11 -emit-pch -o %t %s
// RUN: %clang_cc1 -fopenmp -fopenmp-version=52 -triple x86_64-unknown-linux-gnu -std=c++11 -include-pch %t -verify %s -emit-llvm -o - | FileCheck %s
// RUN: %clang_cc1 -verify -fopenmp -fopenmp-version=52 -triple x86_64-unknown-linux-gnu -emit-llvm -o - %s | FileCheck %s
// RUN: %clang_cc1 -fopenmp -fopenmp-version=52 -fnoopenmp-use-tls -triple x86_64-unknown-linux-gnu -x c++ -std=c++11 -emit-pch -o %t %s
// RUN: %clang_cc1 -fopenmp -fopenmp-version=52 -fnoopenmp-use-tls -triple x86_64-unknown-linux-gnu -std=c++11 -include-pch %t -verify %s -emit-llvm -o - | FileCheck --check-prefix CHECK-TLS %s

// RUN: %clang_cc1 -verify -fopenmp-simd -fopenmp-version=52 -triple x86_64-unknown-linux-gnu -emit-llvm -o - %s | FileCheck --check-prefix SIMD-ONLY0 %s
// RUN: %clang_cc1 -fopenmp-simd -fopenmp-version=52 -triple x86_64-unknown-linux-gnu -x c++ -std=c++11 -emit-pch -o %t %s
// RUN: %clang_cc1 -fopenmp-simd -fopenmp-version=52 -triple x86_64-unknown-linux-gnu -std=c++11 -include-pch %t -verify %s -emit-llvm -o - | FileCheck --check-prefix SIMD-ONLY0 %s
// RUN: %clang_cc1 -verify -fopenmp-simd -fopenmp-version=52 -triple x86_64-unknown-linux-gnu -emit-llvm -o - %s | FileCheck --check-prefix SIMD-ONLY0 %s
// RUN: %clang_cc1 -fopenmp-simd -fopenmp-version=52 -fnoopenmp-use-tls -triple x86_64-unknown-linux-gnu -x c++ -std=c++11 -emit-pch -o %t %s
// RUN: %clang_cc1 -fopenmp-simd -fopenmp-version=52 -fnoopenmp-use-tls -triple x86_64-unknown-linux-gnu -std=c++11 -include-pch %t -verify %s -emit-llvm -o - | FileCheck --check-prefix SIMD-ONLY0 %s
// expected-no-diagnostics

#ifndef HEADER
#define HEADER

enum omp_allocator_handle_t {
  omp_null_allocator = 0,
  omp_default_mem_alloc = 1,
  omp_large_cap_mem_alloc = 2,
  omp_const_mem_alloc = 3,
  omp_high_bw_mem_alloc = 4,
  omp_low_lat_mem_alloc = 5,
  omp_cgroup_mem_alloc = 6,
  omp_pteam_mem_alloc = 7,
  omp_thread_mem_alloc = 8,
  KMP_ALLOCATOR_MAX_HANDLE = __UINTPTR_MAX__
};

template <class T>
struct ST {
  static T m;
};

template <class T, omp_allocator_handle_t TY, unsigned al> T foo() {
  T v;
 #pragma omp scope private(v) allocate(allocator(TY):v)
  v = ST<T>::m;
  #pragma omp scope private(v) allocate(align(al), allocator(TY):v)
  ++v;
  return v;
}

namespace ns {
int a;
}

omp_allocator_handle_t foo();

int main() {
  static int a;
  static int temp;
  int v;
  #pragma omp scope private(ns::a) allocate(allocator(omp_pteam_mem_alloc):ns::a)
  ns::a++;
  #pragma omp scope private(a) allocate(align(8),allocator(omp_thread_mem_alloc):a)
  a = 2;
  #pragma omp scope private(v) allocate(align(1) : v)
  ++v;
  #pragma omp scope private(v) allocate(allocator(omp_default_mem_alloc) : v)
  ++v;
  #pragma omp scope private(v) allocate(allocator(omp_large_cap_mem_alloc), align(8) : v)
  ++v;
  #pragma omp scope private(v) allocate(align(4) : v)
  ++v;
  #pragma omp scope private(v) allocate(align(2), allocator(omp_default_mem_alloc) : v)
  ++v;
  #pragma omp scope private(v) allocate(align(8), allocator(foo()) : v)
  ++v;

  double b = 3;
  #pragma omp scope private(temp) allocate(temp)
  temp += foo<int, omp_cgroup_mem_alloc, 8>();
  return temp+ns::a;
}

extern template int ST<int>::m;

const int b = 8;

void bar(int a, float &z) {
  #pragma omp scope private(a,z) allocate(align(b), allocator(omp_default_mem_alloc) : a,z)
  a += b + z;
}
#endif
// CHECK-LABEL: define dso_local noundef i32 @main(
// CHECK-SAME: ) #[[ATTR0:[0-9]+]] {
// CHECK-NEXT:  [[ENTRY:.*:]]
// CHECK-NEXT:    [[RETVAL:%.*]] = alloca i32, align 4
// CHECK-NEXT:    [[V:%.*]] = alloca i32, align 4
// CHECK-NEXT:    [[B:%.*]] = alloca double, align 8
// CHECK-NEXT:    [[TMP0:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB1:[0-9]+]])
// CHECK-NEXT:    store i32 0, ptr [[RETVAL]], align 4
// CHECK-NEXT:    [[DOTA__VOID_ADDR:%.*]] = call ptr @__kmpc_alloc(i32 [[TMP0]], i64 4, ptr inttoptr (i64 7 to ptr))
// CHECK-NEXT:    [[TMP1:%.*]] = load i32, ptr [[DOTA__VOID_ADDR]], align 4
// CHECK-NEXT:    [[INC:%.*]] = add nsw i32 [[TMP1]], 1
// CHECK-NEXT:    store i32 [[INC]], ptr [[DOTA__VOID_ADDR]], align 4
// CHECK-NEXT:    call void @__kmpc_free(i32 [[TMP0]], ptr [[DOTA__VOID_ADDR]], ptr inttoptr (i64 7 to ptr))
// CHECK-NEXT:    call void @__kmpc_barrier(ptr @[[GLOB2:[0-9]+]], i32 [[TMP0]])
// CHECK-NEXT:    [[DOTA__VOID_ADDR1:%.*]] = call ptr @__kmpc_aligned_alloc(i32 [[TMP0]], i64 8, i64 4, ptr inttoptr (i64 8 to ptr))
// CHECK-NEXT:    store i32 2, ptr [[DOTA__VOID_ADDR1]], align 4
// CHECK-NEXT:    call void @__kmpc_free(i32 [[TMP0]], ptr [[DOTA__VOID_ADDR1]], ptr inttoptr (i64 8 to ptr))
// CHECK-NEXT:    call void @__kmpc_barrier(ptr @[[GLOB2]], i32 [[TMP0]])
// CHECK-NEXT:    [[DOTV__VOID_ADDR:%.*]] = call ptr @__kmpc_aligned_alloc(i32 [[TMP0]], i64 4, i64 4, ptr null)
// CHECK-NEXT:    [[TMP2:%.*]] = load i32, ptr [[DOTV__VOID_ADDR]], align 4
// CHECK-NEXT:    [[INC2:%.*]] = add nsw i32 [[TMP2]], 1
// CHECK-NEXT:    store i32 [[INC2]], ptr [[DOTV__VOID_ADDR]], align 4
// CHECK-NEXT:    call void @__kmpc_free(i32 [[TMP0]], ptr [[DOTV__VOID_ADDR]], ptr null)
// CHECK-NEXT:    call void @__kmpc_barrier(ptr @[[GLOB2]], i32 [[TMP0]])
// CHECK-NEXT:    [[DOTV__VOID_ADDR3:%.*]] = call ptr @__kmpc_alloc(i32 [[TMP0]], i64 4, ptr inttoptr (i64 1 to ptr))
// CHECK-NEXT:    [[TMP3:%.*]] = load i32, ptr [[DOTV__VOID_ADDR3]], align 4
// CHECK-NEXT:    [[INC4:%.*]] = add nsw i32 [[TMP3]], 1
// CHECK-NEXT:    store i32 [[INC4]], ptr [[DOTV__VOID_ADDR3]], align 4
// CHECK-NEXT:    call void @__kmpc_free(i32 [[TMP0]], ptr [[DOTV__VOID_ADDR3]], ptr inttoptr (i64 1 to ptr))
// CHECK-NEXT:    call void @__kmpc_barrier(ptr @[[GLOB2]], i32 [[TMP0]])
// CHECK-NEXT:    [[DOTV__VOID_ADDR5:%.*]] = call ptr @__kmpc_aligned_alloc(i32 [[TMP0]], i64 8, i64 4, ptr inttoptr (i64 2 to ptr))
// CHECK-NEXT:    [[TMP4:%.*]] = load i32, ptr [[DOTV__VOID_ADDR5]], align 4
// CHECK-NEXT:    [[INC6:%.*]] = add nsw i32 [[TMP4]], 1
// CHECK-NEXT:    store i32 [[INC6]], ptr [[DOTV__VOID_ADDR5]], align 4
// CHECK-NEXT:    call void @__kmpc_free(i32 [[TMP0]], ptr [[DOTV__VOID_ADDR5]], ptr inttoptr (i64 2 to ptr))
// CHECK-NEXT:    call void @__kmpc_barrier(ptr @[[GLOB2]], i32 [[TMP0]])
// CHECK-NEXT:    [[DOTV__VOID_ADDR7:%.*]] = call ptr @__kmpc_aligned_alloc(i32 [[TMP0]], i64 4, i64 4, ptr null)
// CHECK-NEXT:    [[TMP5:%.*]] = load i32, ptr [[DOTV__VOID_ADDR7]], align 4
// CHECK-NEXT:    [[INC8:%.*]] = add nsw i32 [[TMP5]], 1
// CHECK-NEXT:    store i32 [[INC8]], ptr [[DOTV__VOID_ADDR7]], align 4
// CHECK-NEXT:    call void @__kmpc_free(i32 [[TMP0]], ptr [[DOTV__VOID_ADDR7]], ptr null)
// CHECK-NEXT:    call void @__kmpc_barrier(ptr @[[GLOB2]], i32 [[TMP0]])
// CHECK-NEXT:    [[DOTV__VOID_ADDR9:%.*]] = call ptr @__kmpc_aligned_alloc(i32 [[TMP0]], i64 4, i64 4, ptr inttoptr (i64 1 to ptr))
// CHECK-NEXT:    [[TMP6:%.*]] = load i32, ptr [[DOTV__VOID_ADDR9]], align 4
// CHECK-NEXT:    [[INC10:%.*]] = add nsw i32 [[TMP6]], 1
// CHECK-NEXT:    store i32 [[INC10]], ptr [[DOTV__VOID_ADDR9]], align 4
// CHECK-NEXT:    call void @__kmpc_free(i32 [[TMP0]], ptr [[DOTV__VOID_ADDR9]], ptr inttoptr (i64 1 to ptr))
// CHECK-NEXT:    call void @__kmpc_barrier(ptr @[[GLOB2]], i32 [[TMP0]])
// CHECK-NEXT:    [[CALL:%.*]] = call noundef i64 @_Z3foov()
// CHECK-NEXT:    [[CONV:%.*]] = inttoptr i64 [[CALL]] to ptr
// CHECK-NEXT:    [[DOTV__VOID_ADDR11:%.*]] = call ptr @__kmpc_aligned_alloc(i32 [[TMP0]], i64 8, i64 4, ptr [[CONV]])
// CHECK-NEXT:    [[TMP7:%.*]] = load i32, ptr [[DOTV__VOID_ADDR11]], align 4
// CHECK-NEXT:    [[INC12:%.*]] = add nsw i32 [[TMP7]], 1
// CHECK-NEXT:    store i32 [[INC12]], ptr [[DOTV__VOID_ADDR11]], align 4
// CHECK-NEXT:    [[CALL13:%.*]] = call noundef i64 @_Z3foov()
// CHECK-NEXT:    [[CONV14:%.*]] = inttoptr i64 [[CALL13]] to ptr
// CHECK-NEXT:    call void @__kmpc_free(i32 [[TMP0]], ptr [[DOTV__VOID_ADDR11]], ptr [[CONV14]])
// CHECK-NEXT:    call void @__kmpc_barrier(ptr @[[GLOB2]], i32 [[TMP0]])
// CHECK-NEXT:    store double 3.000000e+00, ptr [[B]], align 8
// CHECK-NEXT:    [[DOTTEMP__VOID_ADDR:%.*]] = call ptr @__kmpc_alloc(i32 [[TMP0]], i64 4, ptr null)
// CHECK-NEXT:    [[CALL15:%.*]] = call noundef i32 @_Z3fooIiL22omp_allocator_handle_t6ELj8EET_v()
// CHECK-NEXT:    [[TMP8:%.*]] = load i32, ptr [[DOTTEMP__VOID_ADDR]], align 4
// CHECK-NEXT:    [[ADD:%.*]] = add nsw i32 [[TMP8]], [[CALL15]]
// CHECK-NEXT:    store i32 [[ADD]], ptr [[DOTTEMP__VOID_ADDR]], align 4
// CHECK-NEXT:    call void @__kmpc_free(i32 [[TMP0]], ptr [[DOTTEMP__VOID_ADDR]], ptr null)
// CHECK-NEXT:    call void @__kmpc_barrier(ptr @[[GLOB2]], i32 [[TMP0]])
// CHECK-NEXT:    [[TMP9:%.*]] = load i32, ptr @_ZZ4mainE4temp, align 4
// CHECK-NEXT:    [[TMP10:%.*]] = load i32, ptr @_ZN2ns1aE, align 4
// CHECK-NEXT:    [[ADD16:%.*]] = add nsw i32 [[TMP9]], [[TMP10]]
// CHECK-NEXT:    ret i32 [[ADD16]]
//
//
// CHECK-LABEL: define linkonce_odr noundef i32 @_Z3fooIiL22omp_allocator_handle_t6ELj8EET_v(
// CHECK-SAME: ) #[[ATTR4:[0-9]+]] comdat {
// CHECK-NEXT:  [[ENTRY:.*:]]
// CHECK-NEXT:    [[V:%.*]] = alloca i32, align 4
// CHECK-NEXT:    [[TMP0:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB1]])
// CHECK-NEXT:    [[DOTV__VOID_ADDR:%.*]] = call ptr @__kmpc_alloc(i32 [[TMP0]], i64 4, ptr inttoptr (i64 6 to ptr))
// CHECK-NEXT:    [[TMP1:%.*]] = load i32, ptr @_ZN2STIiE1mE, align 4
// CHECK-NEXT:    store i32 [[TMP1]], ptr [[DOTV__VOID_ADDR]], align 4
// CHECK-NEXT:    call void @__kmpc_free(i32 [[TMP0]], ptr [[DOTV__VOID_ADDR]], ptr inttoptr (i64 6 to ptr))
// CHECK-NEXT:    call void @__kmpc_barrier(ptr @[[GLOB2]], i32 [[TMP0]])
// CHECK-NEXT:    [[DOTV__VOID_ADDR1:%.*]] = call ptr @__kmpc_aligned_alloc(i32 [[TMP0]], i64 8, i64 4, ptr inttoptr (i64 6 to ptr))
// CHECK-NEXT:    [[TMP2:%.*]] = load i32, ptr [[DOTV__VOID_ADDR1]], align 4
// CHECK-NEXT:    [[INC:%.*]] = add nsw i32 [[TMP2]], 1
// CHECK-NEXT:    store i32 [[INC]], ptr [[DOTV__VOID_ADDR1]], align 4
// CHECK-NEXT:    call void @__kmpc_free(i32 [[TMP0]], ptr [[DOTV__VOID_ADDR1]], ptr inttoptr (i64 6 to ptr))
// CHECK-NEXT:    call void @__kmpc_barrier(ptr @[[GLOB2]], i32 [[TMP0]])
// CHECK-NEXT:    [[TMP3:%.*]] = load i32, ptr [[V]], align 4
// CHECK-NEXT:    ret i32 [[TMP3]]
//
//
// CHECK-LABEL: define dso_local void @_Z3bariRf(
// CHECK-SAME: i32 noundef [[A:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[Z:%.*]]) #[[ATTR4]] {
// CHECK-NEXT:  [[ENTRY:.*:]]
// CHECK-NEXT:    [[A_ADDR:%.*]] = alloca i32, align 4
// CHECK-NEXT:    [[Z_ADDR:%.*]] = alloca ptr, align 8
// CHECK-NEXT:    [[TMP:%.*]] = alloca ptr, align 8
// CHECK-NEXT:    [[TMP0:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB1]])
// CHECK-NEXT:    store i32 [[A]], ptr [[A_ADDR]], align 4
// CHECK-NEXT:    store ptr [[Z]], ptr [[Z_ADDR]], align 8
// CHECK-NEXT:    [[DOTA__VOID_ADDR:%.*]] = call ptr @__kmpc_aligned_alloc(i32 [[TMP0]], i64 8, i64 4, ptr inttoptr (i64 1 to ptr))
// CHECK-NEXT:    [[DOTZ__VOID_ADDR:%.*]] = call ptr @__kmpc_aligned_alloc(i32 [[TMP0]], i64 8, i64 4, ptr inttoptr (i64 1 to ptr))
// CHECK-NEXT:    store ptr [[DOTZ__VOID_ADDR]], ptr [[TMP]], align 8
// CHECK-NEXT:    [[TMP1:%.*]] = load ptr, ptr [[TMP]], align 8
// CHECK-NEXT:    [[TMP2:%.*]] = load float, ptr [[TMP1]], align 4
// CHECK-NEXT:    [[ADD:%.*]] = fadd float 8.000000e+00, [[TMP2]]
// CHECK-NEXT:    [[TMP3:%.*]] = load i32, ptr [[DOTA__VOID_ADDR]], align 4
// CHECK-NEXT:    [[CONV:%.*]] = sitofp i32 [[TMP3]] to float
// CHECK-NEXT:    [[ADD1:%.*]] = fadd float [[CONV]], [[ADD]]
// CHECK-NEXT:    [[CONV2:%.*]] = fptosi float [[ADD1]] to i32
// CHECK-NEXT:    store i32 [[CONV2]], ptr [[DOTA__VOID_ADDR]], align 4
// CHECK-NEXT:    call void @__kmpc_free(i32 [[TMP0]], ptr [[DOTZ__VOID_ADDR]], ptr inttoptr (i64 1 to ptr))
// CHECK-NEXT:    call void @__kmpc_free(i32 [[TMP0]], ptr [[DOTA__VOID_ADDR]], ptr inttoptr (i64 1 to ptr))
// CHECK-NEXT:    call void @__kmpc_barrier(ptr @[[GLOB2]], i32 [[TMP0]])
// CHECK-NEXT:    ret void
//
//
// CHECK-TLS-LABEL: define dso_local noundef i32 @main(
// CHECK-TLS-SAME: ) #[[ATTR0:[0-9]+]] {
// CHECK-TLS-NEXT:  [[ENTRY:.*:]]
// CHECK-TLS-NEXT:    [[RETVAL:%.*]] = alloca i32, align 4
// CHECK-TLS-NEXT:    [[V:%.*]] = alloca i32, align 4
// CHECK-TLS-NEXT:    [[B:%.*]] = alloca double, align 8
// CHECK-TLS-NEXT:    [[TMP0:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB1:[0-9]+]])
// CHECK-TLS-NEXT:    store i32 0, ptr [[RETVAL]], align 4
// CHECK-TLS-NEXT:    [[DOTA__VOID_ADDR:%.*]] = call ptr @__kmpc_alloc(i32 [[TMP0]], i64 4, ptr inttoptr (i64 7 to ptr))
// CHECK-TLS-NEXT:    [[TMP1:%.*]] = load i32, ptr [[DOTA__VOID_ADDR]], align 4
// CHECK-TLS-NEXT:    [[INC:%.*]] = add nsw i32 [[TMP1]], 1
// CHECK-TLS-NEXT:    store i32 [[INC]], ptr [[DOTA__VOID_ADDR]], align 4
// CHECK-TLS-NEXT:    call void @__kmpc_free(i32 [[TMP0]], ptr [[DOTA__VOID_ADDR]], ptr inttoptr (i64 7 to ptr))
// CHECK-TLS-NEXT:    call void @__kmpc_barrier(ptr @[[GLOB2:[0-9]+]], i32 [[TMP0]])
// CHECK-TLS-NEXT:    [[DOTA__VOID_ADDR1:%.*]] = call ptr @__kmpc_aligned_alloc(i32 [[TMP0]], i64 8, i64 4, ptr inttoptr (i64 8 to ptr))
// CHECK-TLS-NEXT:    store i32 2, ptr [[DOTA__VOID_ADDR1]], align 4
// CHECK-TLS-NEXT:    call void @__kmpc_free(i32 [[TMP0]], ptr [[DOTA__VOID_ADDR1]], ptr inttoptr (i64 8 to ptr))
// CHECK-TLS-NEXT:    call void @__kmpc_barrier(ptr @[[GLOB2]], i32 [[TMP0]])
// CHECK-TLS-NEXT:    [[DOTV__VOID_ADDR:%.*]] = call ptr @__kmpc_aligned_alloc(i32 [[TMP0]], i64 4, i64 4, ptr null)
// CHECK-TLS-NEXT:    [[TMP2:%.*]] = load i32, ptr [[DOTV__VOID_ADDR]], align 4
// CHECK-TLS-NEXT:    [[INC2:%.*]] = add nsw i32 [[TMP2]], 1
// CHECK-TLS-NEXT:    store i32 [[INC2]], ptr [[DOTV__VOID_ADDR]], align 4
// CHECK-TLS-NEXT:    call void @__kmpc_free(i32 [[TMP0]], ptr [[DOTV__VOID_ADDR]], ptr null)
// CHECK-TLS-NEXT:    call void @__kmpc_barrier(ptr @[[GLOB2]], i32 [[TMP0]])
// CHECK-TLS-NEXT:    [[DOTV__VOID_ADDR3:%.*]] = call ptr @__kmpc_alloc(i32 [[TMP0]], i64 4, ptr inttoptr (i64 1 to ptr))
// CHECK-TLS-NEXT:    [[TMP3:%.*]] = load i32, ptr [[DOTV__VOID_ADDR3]], align 4
// CHECK-TLS-NEXT:    [[INC4:%.*]] = add nsw i32 [[TMP3]], 1
// CHECK-TLS-NEXT:    store i32 [[INC4]], ptr [[DOTV__VOID_ADDR3]], align 4
// CHECK-TLS-NEXT:    call void @__kmpc_free(i32 [[TMP0]], ptr [[DOTV__VOID_ADDR3]], ptr inttoptr (i64 1 to ptr))
// CHECK-TLS-NEXT:    call void @__kmpc_barrier(ptr @[[GLOB2]], i32 [[TMP0]])
// CHECK-TLS-NEXT:    [[DOTV__VOID_ADDR5:%.*]] = call ptr @__kmpc_aligned_alloc(i32 [[TMP0]], i64 8, i64 4, ptr inttoptr (i64 2 to ptr))
// CHECK-TLS-NEXT:    [[TMP4:%.*]] = load i32, ptr [[DOTV__VOID_ADDR5]], align 4
// CHECK-TLS-NEXT:    [[INC6:%.*]] = add nsw i32 [[TMP4]], 1
// CHECK-TLS-NEXT:    store i32 [[INC6]], ptr [[DOTV__VOID_ADDR5]], align 4
// CHECK-TLS-NEXT:    call void @__kmpc_free(i32 [[TMP0]], ptr [[DOTV__VOID_ADDR5]], ptr inttoptr (i64 2 to ptr))
// CHECK-TLS-NEXT:    call void @__kmpc_barrier(ptr @[[GLOB2]], i32 [[TMP0]])
// CHECK-TLS-NEXT:    [[DOTV__VOID_ADDR7:%.*]] = call ptr @__kmpc_aligned_alloc(i32 [[TMP0]], i64 4, i64 4, ptr null)
// CHECK-TLS-NEXT:    [[TMP5:%.*]] = load i32, ptr [[DOTV__VOID_ADDR7]], align 4
// CHECK-TLS-NEXT:    [[INC8:%.*]] = add nsw i32 [[TMP5]], 1
// CHECK-TLS-NEXT:    store i32 [[INC8]], ptr [[DOTV__VOID_ADDR7]], align 4
// CHECK-TLS-NEXT:    call void @__kmpc_free(i32 [[TMP0]], ptr [[DOTV__VOID_ADDR7]], ptr null)
// CHECK-TLS-NEXT:    call void @__kmpc_barrier(ptr @[[GLOB2]], i32 [[TMP0]])
// CHECK-TLS-NEXT:    [[DOTV__VOID_ADDR9:%.*]] = call ptr @__kmpc_aligned_alloc(i32 [[TMP0]], i64 4, i64 4, ptr inttoptr (i64 1 to ptr))
// CHECK-TLS-NEXT:    [[TMP6:%.*]] = load i32, ptr [[DOTV__VOID_ADDR9]], align 4
// CHECK-TLS-NEXT:    [[INC10:%.*]] = add nsw i32 [[TMP6]], 1
// CHECK-TLS-NEXT:    store i32 [[INC10]], ptr [[DOTV__VOID_ADDR9]], align 4
// CHECK-TLS-NEXT:    call void @__kmpc_free(i32 [[TMP0]], ptr [[DOTV__VOID_ADDR9]], ptr inttoptr (i64 1 to ptr))
// CHECK-TLS-NEXT:    call void @__kmpc_barrier(ptr @[[GLOB2]], i32 [[TMP0]])
// CHECK-TLS-NEXT:    [[CALL:%.*]] = call noundef i64 @_Z3foov()
// CHECK-TLS-NEXT:    [[CONV:%.*]] = inttoptr i64 [[CALL]] to ptr
// CHECK-TLS-NEXT:    [[DOTV__VOID_ADDR11:%.*]] = call ptr @__kmpc_aligned_alloc(i32 [[TMP0]], i64 8, i64 4, ptr [[CONV]])
// CHECK-TLS-NEXT:    [[TMP7:%.*]] = load i32, ptr [[DOTV__VOID_ADDR11]], align 4
// CHECK-TLS-NEXT:    [[INC12:%.*]] = add nsw i32 [[TMP7]], 1
// CHECK-TLS-NEXT:    store i32 [[INC12]], ptr [[DOTV__VOID_ADDR11]], align 4
// CHECK-TLS-NEXT:    [[CALL13:%.*]] = call noundef i64 @_Z3foov()
// CHECK-TLS-NEXT:    [[CONV14:%.*]] = inttoptr i64 [[CALL13]] to ptr
// CHECK-TLS-NEXT:    call void @__kmpc_free(i32 [[TMP0]], ptr [[DOTV__VOID_ADDR11]], ptr [[CONV14]])
// CHECK-TLS-NEXT:    call void @__kmpc_barrier(ptr @[[GLOB2]], i32 [[TMP0]])
// CHECK-TLS-NEXT:    store double 3.000000e+00, ptr [[B]], align 8
// CHECK-TLS-NEXT:    [[DOTTEMP__VOID_ADDR:%.*]] = call ptr @__kmpc_alloc(i32 [[TMP0]], i64 4, ptr null)
// CHECK-TLS-NEXT:    [[CALL15:%.*]] = call noundef i32 @_Z3fooIiL22omp_allocator_handle_t6ELj8EET_v()
// CHECK-TLS-NEXT:    [[TMP8:%.*]] = load i32, ptr [[DOTTEMP__VOID_ADDR]], align 4
// CHECK-TLS-NEXT:    [[ADD:%.*]] = add nsw i32 [[TMP8]], [[CALL15]]
// CHECK-TLS-NEXT:    store i32 [[ADD]], ptr [[DOTTEMP__VOID_ADDR]], align 4
// CHECK-TLS-NEXT:    call void @__kmpc_free(i32 [[TMP0]], ptr [[DOTTEMP__VOID_ADDR]], ptr null)
// CHECK-TLS-NEXT:    call void @__kmpc_barrier(ptr @[[GLOB2]], i32 [[TMP0]])
// CHECK-TLS-NEXT:    [[TMP9:%.*]] = load i32, ptr @_ZZ4mainE4temp, align 4
// CHECK-TLS-NEXT:    [[TMP10:%.*]] = load i32, ptr @_ZN2ns1aE, align 4
// CHECK-TLS-NEXT:    [[ADD16:%.*]] = add nsw i32 [[TMP9]], [[TMP10]]
// CHECK-TLS-NEXT:    ret i32 [[ADD16]]
//
//
// CHECK-TLS-LABEL: define linkonce_odr noundef i32 @_Z3fooIiL22omp_allocator_handle_t6ELj8EET_v(
// CHECK-TLS-SAME: ) #[[ATTR4:[0-9]+]] comdat {
// CHECK-TLS-NEXT:  [[ENTRY:.*:]]
// CHECK-TLS-NEXT:    [[V:%.*]] = alloca i32, align 4
// CHECK-TLS-NEXT:    [[TMP0:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB1]])
// CHECK-TLS-NEXT:    [[DOTV__VOID_ADDR:%.*]] = call ptr @__kmpc_alloc(i32 [[TMP0]], i64 4, ptr inttoptr (i64 6 to ptr))
// CHECK-TLS-NEXT:    [[TMP1:%.*]] = load i32, ptr @_ZN2STIiE1mE, align 4
// CHECK-TLS-NEXT:    store i32 [[TMP1]], ptr [[DOTV__VOID_ADDR]], align 4
// CHECK-TLS-NEXT:    call void @__kmpc_free(i32 [[TMP0]], ptr [[DOTV__VOID_ADDR]], ptr inttoptr (i64 6 to ptr))
// CHECK-TLS-NEXT:    call void @__kmpc_barrier(ptr @[[GLOB2]], i32 [[TMP0]])
// CHECK-TLS-NEXT:    [[DOTV__VOID_ADDR1:%.*]] = call ptr @__kmpc_aligned_alloc(i32 [[TMP0]], i64 8, i64 4, ptr inttoptr (i64 6 to ptr))
// CHECK-TLS-NEXT:    [[TMP2:%.*]] = load i32, ptr [[DOTV__VOID_ADDR1]], align 4
// CHECK-TLS-NEXT:    [[INC:%.*]] = add nsw i32 [[TMP2]], 1
// CHECK-TLS-NEXT:    store i32 [[INC]], ptr [[DOTV__VOID_ADDR1]], align 4
// CHECK-TLS-NEXT:    call void @__kmpc_free(i32 [[TMP0]], ptr [[DOTV__VOID_ADDR1]], ptr inttoptr (i64 6 to ptr))
// CHECK-TLS-NEXT:    call void @__kmpc_barrier(ptr @[[GLOB2]], i32 [[TMP0]])
// CHECK-TLS-NEXT:    [[TMP3:%.*]] = load i32, ptr [[V]], align 4
// CHECK-TLS-NEXT:    ret i32 [[TMP3]]
//
//
// CHECK-TLS-LABEL: define dso_local void @_Z3bariRf(
// CHECK-TLS-SAME: i32 noundef [[A:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[Z:%.*]]) #[[ATTR4]] {
// CHECK-TLS-NEXT:  [[ENTRY:.*:]]
// CHECK-TLS-NEXT:    [[A_ADDR:%.*]] = alloca i32, align 4
// CHECK-TLS-NEXT:    [[Z_ADDR:%.*]] = alloca ptr, align 8
// CHECK-TLS-NEXT:    [[TMP:%.*]] = alloca ptr, align 8
// CHECK-TLS-NEXT:    [[TMP0:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB1]])
// CHECK-TLS-NEXT:    store i32 [[A]], ptr [[A_ADDR]], align 4
// CHECK-TLS-NEXT:    store ptr [[Z]], ptr [[Z_ADDR]], align 8
// CHECK-TLS-NEXT:    [[DOTA__VOID_ADDR:%.*]] = call ptr @__kmpc_aligned_alloc(i32 [[TMP0]], i64 8, i64 4, ptr inttoptr (i64 1 to ptr))
// CHECK-TLS-NEXT:    [[DOTZ__VOID_ADDR:%.*]] = call ptr @__kmpc_aligned_alloc(i32 [[TMP0]], i64 8, i64 4, ptr inttoptr (i64 1 to ptr))
// CHECK-TLS-NEXT:    store ptr [[DOTZ__VOID_ADDR]], ptr [[TMP]], align 8
// CHECK-TLS-NEXT:    [[TMP1:%.*]] = load ptr, ptr [[TMP]], align 8
// CHECK-TLS-NEXT:    [[TMP2:%.*]] = load float, ptr [[TMP1]], align 4
// CHECK-TLS-NEXT:    [[ADD:%.*]] = fadd float 8.000000e+00, [[TMP2]]
// CHECK-TLS-NEXT:    [[TMP3:%.*]] = load i32, ptr [[DOTA__VOID_ADDR]], align 4
// CHECK-TLS-NEXT:    [[CONV:%.*]] = sitofp i32 [[TMP3]] to float
// CHECK-TLS-NEXT:    [[ADD1:%.*]] = fadd float [[CONV]], [[ADD]]
// CHECK-TLS-NEXT:    [[CONV2:%.*]] = fptosi float [[ADD1]] to i32
// CHECK-TLS-NEXT:    store i32 [[CONV2]], ptr [[DOTA__VOID_ADDR]], align 4
// CHECK-TLS-NEXT:    call void @__kmpc_free(i32 [[TMP0]], ptr [[DOTZ__VOID_ADDR]], ptr inttoptr (i64 1 to ptr))
// CHECK-TLS-NEXT:    call void @__kmpc_free(i32 [[TMP0]], ptr [[DOTA__VOID_ADDR]], ptr inttoptr (i64 1 to ptr))
// CHECK-TLS-NEXT:    call void @__kmpc_barrier(ptr @[[GLOB2]], i32 [[TMP0]])
// CHECK-TLS-NEXT:    ret void
//
//
// SIMD-ONLY0-LABEL: define dso_local noundef i32 @main(
// SIMD-ONLY0-SAME: ) #[[ATTR0:[0-9]+]] {
// SIMD-ONLY0-NEXT:  [[ENTRY:.*:]]
// SIMD-ONLY0-NEXT:    [[RETVAL:%.*]] = alloca i32, align 4
// SIMD-ONLY0-NEXT:    [[V:%.*]] = alloca i32, align 4
// SIMD-ONLY0-NEXT:    [[A:%.*]] = alloca i32, align 4
// SIMD-ONLY0-NEXT:    [[A1:%.*]] = alloca i32, align 4
// SIMD-ONLY0-NEXT:    [[V2:%.*]] = alloca i32, align 4
// SIMD-ONLY0-NEXT:    [[V4:%.*]] = alloca i32, align 4
// SIMD-ONLY0-NEXT:    [[V6:%.*]] = alloca i32, align 4
// SIMD-ONLY0-NEXT:    [[V8:%.*]] = alloca i32, align 4
// SIMD-ONLY0-NEXT:    [[V10:%.*]] = alloca i32, align 4
// SIMD-ONLY0-NEXT:    [[V12:%.*]] = alloca i32, align 4
// SIMD-ONLY0-NEXT:    [[B:%.*]] = alloca double, align 8
// SIMD-ONLY0-NEXT:    [[TEMP:%.*]] = alloca i32, align 4
// SIMD-ONLY0-NEXT:    store i32 0, ptr [[RETVAL]], align 4
// SIMD-ONLY0-NEXT:    [[TMP0:%.*]] = load i32, ptr [[A]], align 4
// SIMD-ONLY0-NEXT:    [[INC:%.*]] = add nsw i32 [[TMP0]], 1
// SIMD-ONLY0-NEXT:    store i32 [[INC]], ptr [[A]], align 4
// SIMD-ONLY0-NEXT:    store i32 2, ptr [[A1]], align 4
// SIMD-ONLY0-NEXT:    [[TMP1:%.*]] = load i32, ptr [[V2]], align 4
// SIMD-ONLY0-NEXT:    [[INC3:%.*]] = add nsw i32 [[TMP1]], 1
// SIMD-ONLY0-NEXT:    store i32 [[INC3]], ptr [[V2]], align 4
// SIMD-ONLY0-NEXT:    [[TMP2:%.*]] = load i32, ptr [[V4]], align 4
// SIMD-ONLY0-NEXT:    [[INC5:%.*]] = add nsw i32 [[TMP2]], 1
// SIMD-ONLY0-NEXT:    store i32 [[INC5]], ptr [[V4]], align 4
// SIMD-ONLY0-NEXT:    [[TMP3:%.*]] = load i32, ptr [[V6]], align 4
// SIMD-ONLY0-NEXT:    [[INC7:%.*]] = add nsw i32 [[TMP3]], 1
// SIMD-ONLY0-NEXT:    store i32 [[INC7]], ptr [[V6]], align 4
// SIMD-ONLY0-NEXT:    [[TMP4:%.*]] = load i32, ptr [[V8]], align 4
// SIMD-ONLY0-NEXT:    [[INC9:%.*]] = add nsw i32 [[TMP4]], 1
// SIMD-ONLY0-NEXT:    store i32 [[INC9]], ptr [[V8]], align 4
// SIMD-ONLY0-NEXT:    [[TMP5:%.*]] = load i32, ptr [[V10]], align 4
// SIMD-ONLY0-NEXT:    [[INC11:%.*]] = add nsw i32 [[TMP5]], 1
// SIMD-ONLY0-NEXT:    store i32 [[INC11]], ptr [[V10]], align 4
// SIMD-ONLY0-NEXT:    [[TMP6:%.*]] = load i32, ptr [[V12]], align 4
// SIMD-ONLY0-NEXT:    [[INC13:%.*]] = add nsw i32 [[TMP6]], 1
// SIMD-ONLY0-NEXT:    store i32 [[INC13]], ptr [[V12]], align 4
// SIMD-ONLY0-NEXT:    store double 3.000000e+00, ptr [[B]], align 8
// SIMD-ONLY0-NEXT:    [[CALL:%.*]] = call noundef i32 @_Z3fooIiL22omp_allocator_handle_t6ELj8EET_v()
// SIMD-ONLY0-NEXT:    [[TMP7:%.*]] = load i32, ptr [[TEMP]], align 4
// SIMD-ONLY0-NEXT:    [[ADD:%.*]] = add nsw i32 [[TMP7]], [[CALL]]
// SIMD-ONLY0-NEXT:    store i32 [[ADD]], ptr [[TEMP]], align 4
// SIMD-ONLY0-NEXT:    [[TMP8:%.*]] = load i32, ptr @_ZZ4mainE4temp, align 4
// SIMD-ONLY0-NEXT:    [[TMP9:%.*]] = load i32, ptr @_ZN2ns1aE, align 4
// SIMD-ONLY0-NEXT:    [[ADD14:%.*]] = add nsw i32 [[TMP8]], [[TMP9]]
// SIMD-ONLY0-NEXT:    ret i32 [[ADD14]]
//
//
// SIMD-ONLY0-LABEL: define linkonce_odr noundef i32 @_Z3fooIiL22omp_allocator_handle_t6ELj8EET_v(
// SIMD-ONLY0-SAME: ) #[[ATTR1:[0-9]+]] comdat {
// SIMD-ONLY0-NEXT:  [[ENTRY:.*:]]
// SIMD-ONLY0-NEXT:    [[V:%.*]] = alloca i32, align 4
// SIMD-ONLY0-NEXT:    [[V1:%.*]] = alloca i32, align 4
// SIMD-ONLY0-NEXT:    [[V2:%.*]] = alloca i32, align 4
// SIMD-ONLY0-NEXT:    [[TMP0:%.*]] = load i32, ptr @_ZN2STIiE1mE, align 4
// SIMD-ONLY0-NEXT:    store i32 [[TMP0]], ptr [[V1]], align 4
// SIMD-ONLY0-NEXT:    [[TMP1:%.*]] = load i32, ptr [[V2]], align 4
// SIMD-ONLY0-NEXT:    [[INC:%.*]] = add nsw i32 [[TMP1]], 1
// SIMD-ONLY0-NEXT:    store i32 [[INC]], ptr [[V2]], align 4
// SIMD-ONLY0-NEXT:    [[TMP2:%.*]] = load i32, ptr [[V]], align 4
// SIMD-ONLY0-NEXT:    ret i32 [[TMP2]]
//
//
// SIMD-ONLY0-LABEL: define dso_local void @_Z3bariRf(
// SIMD-ONLY0-SAME: i32 noundef [[A:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[Z:%.*]]) #[[ATTR1]] {
// SIMD-ONLY0-NEXT:  [[ENTRY:.*:]]
// SIMD-ONLY0-NEXT:    [[A_ADDR:%.*]] = alloca i32, align 4
// SIMD-ONLY0-NEXT:    [[Z_ADDR:%.*]] = alloca ptr, align 8
// SIMD-ONLY0-NEXT:    [[A1:%.*]] = alloca i32, align 4
// SIMD-ONLY0-NEXT:    [[Z2:%.*]] = alloca float, align 4
// SIMD-ONLY0-NEXT:    [[TMP:%.*]] = alloca ptr, align 8
// SIMD-ONLY0-NEXT:    store i32 [[A]], ptr [[A_ADDR]], align 4
// SIMD-ONLY0-NEXT:    store ptr [[Z]], ptr [[Z_ADDR]], align 8
// SIMD-ONLY0-NEXT:    store ptr [[Z2]], ptr [[TMP]], align 8
// SIMD-ONLY0-NEXT:    [[TMP0:%.*]] = load ptr, ptr [[TMP]], align 8
// SIMD-ONLY0-NEXT:    [[TMP1:%.*]] = load float, ptr [[TMP0]], align 4
// SIMD-ONLY0-NEXT:    [[ADD:%.*]] = fadd float 8.000000e+00, [[TMP1]]
// SIMD-ONLY0-NEXT:    [[TMP2:%.*]] = load i32, ptr [[A1]], align 4
// SIMD-ONLY0-NEXT:    [[CONV:%.*]] = sitofp i32 [[TMP2]] to float
// SIMD-ONLY0-NEXT:    [[ADD3:%.*]] = fadd float [[CONV]], [[ADD]]
// SIMD-ONLY0-NEXT:    [[CONV4:%.*]] = fptosi float [[ADD3]] to i32
// SIMD-ONLY0-NEXT:    store i32 [[CONV4]], ptr [[A1]], align 4
// SIMD-ONLY0-NEXT:    ret void
//