clang/test/OpenMP/target_parallel_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 parallel 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-LABEL: define {{[^@]+}}@_Z3foov
  76 // CHECK-SAME: () #[[ATTR0:[0-9]+]] {
  77 // CHECK-NEXT:  entry:
  78 // CHECK-NEXT:    [[TRAITS:%.*]] = alloca [10 x %struct.omp_alloctrait_t], align 8
  79 // CHECK-NEXT:    [[MY_ALLOCATOR:%.*]] = alloca i64, align 8
  80 // CHECK-NEXT:    [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [1 x ptr], align 8
  81 // CHECK-NEXT:    [[DOTOFFLOAD_PTRS:%.*]] = alloca [1 x ptr], align 8
  82 // CHECK-NEXT:    [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [1 x ptr], align 8
  83 // CHECK-NEXT:    [[KERNEL_ARGS:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS:%.*]], align 8
  84 // CHECK-NEXT:    [[TMP0:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
  85 // CHECK-NEXT:    store ptr [[TRAITS]], ptr [[TMP0]], align 8
  86 // CHECK-NEXT:    [[TMP1:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
  87 // CHECK-NEXT:    store ptr [[TRAITS]], ptr [[TMP1]], align 8
  88 // CHECK-NEXT:    [[TMP2:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0
  89 // CHECK-NEXT:    store ptr null, ptr [[TMP2]], align 8
  90 // CHECK-NEXT:    [[TMP3:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
  91 // CHECK-NEXT:    [[TMP4:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
  92 // CHECK-NEXT:    [[TMP5:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 0
  93 // CHECK-NEXT:    store i32 2, ptr [[TMP5]], align 4
  94 // CHECK-NEXT:    [[TMP6:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 1
  95 // CHECK-NEXT:    store i32 1, ptr [[TMP6]], align 4
  96 // CHECK-NEXT:    [[TMP7:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 2
  97 // CHECK-NEXT:    store ptr [[TMP3]], ptr [[TMP7]], align 8
  98 // CHECK-NEXT:    [[TMP8:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 3
  99 // CHECK-NEXT:    store ptr [[TMP4]], ptr [[TMP8]], align 8
 100 // CHECK-NEXT:    [[TMP9:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 4
 101 // CHECK-NEXT:    store ptr @.offload_sizes, ptr [[TMP9]], align 8
 102 // CHECK-NEXT:    [[TMP10:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 5
 103 // CHECK-NEXT:    store ptr @.offload_maptypes, ptr [[TMP10]], align 8
 104 // CHECK-NEXT:    [[TMP11:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 6
 105 // CHECK-NEXT:    store ptr null, ptr [[TMP11]], align 8
 106 // CHECK-NEXT:    [[TMP12:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 7
 107 // CHECK-NEXT:    store ptr null, ptr [[TMP12]], align 8
 108 // CHECK-NEXT:    [[TMP13:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 8
 109 // CHECK-NEXT:    store i64 0, ptr [[TMP13]], align 8
 110 // CHECK-NEXT:    [[TMP14:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 9
 111 // CHECK-NEXT:    store i64 0, ptr [[TMP14]], align 8
 112 // CHECK-NEXT:    [[TMP15:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 10
 113 // CHECK-NEXT:    store [3 x i32] [i32 1, i32 0, i32 0], ptr [[TMP15]], align 4
 114 // CHECK-NEXT:    [[TMP16:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 11
 115 // CHECK-NEXT:    store [3 x i32] zeroinitializer, ptr [[TMP16]], align 4
 116 // CHECK-NEXT:    [[TMP17:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 12
 117 // CHECK-NEXT:    store i32 0, ptr [[TMP17]], align 4
 118 // CHECK-NEXT:    [[TMP18:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB1:[0-9]+]], i64 -1, i32 1, i32 0, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3foov_l66.region_id, ptr [[KERNEL_ARGS]])
 119 // CHECK-NEXT:    [[TMP19:%.*]] = icmp ne i32 [[TMP18]], 0
 120 // CHECK-NEXT:    br i1 [[TMP19]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
 121 // CHECK:       omp_offload.failed:
 122 // CHECK-NEXT:    call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3foov_l66(ptr [[TRAITS]]) #[[ATTR2:[0-9]+]]
 123 // CHECK-NEXT:    br label [[OMP_OFFLOAD_CONT]]
 124 // CHECK:       omp_offload.cont:
 125 // CHECK-NEXT:    ret void
 126 //
 127 //
 128 // CHECK-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3foov_l66
 129 // CHECK-SAME: (ptr noundef nonnull align 8 dereferenceable(160) [[TRAITS:%.*]]) #[[ATTR1:[0-9]+]] {
 130 // CHECK-NEXT:  entry:
 131 // CHECK-NEXT:    [[TRAITS_ADDR:%.*]] = alloca ptr, align 8
 132 // CHECK-NEXT:    [[MY_ALLOCATOR:%.*]] = alloca i64, align 8
 133 // CHECK-NEXT:    [[TMP0:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB1]])
 134 // CHECK-NEXT:    store ptr [[TRAITS]], ptr [[TRAITS_ADDR]], align 8
 135 // CHECK-NEXT:    [[TMP1:%.*]] = load ptr, ptr [[TRAITS_ADDR]], align 8
 136 // CHECK-NEXT:    [[TMP2:%.*]] = call ptr @__kmpc_init_allocator(i32 [[TMP0]], ptr null, i32 10, ptr [[TMP1]])
 137 // CHECK-NEXT:    [[CONV:%.*]] = ptrtoint ptr [[TMP2]] to i64
 138 // CHECK-NEXT:    store i64 [[CONV]], ptr [[MY_ALLOCATOR]], align 8
 139 // CHECK-NEXT:    call void (ptr, i32, ptr, ...) @__kmpc_fork_call(ptr @[[GLOB1]], i32 0, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3foov_l66.omp_outlined)
 140 // CHECK-NEXT:    [[TMP3:%.*]] = load i64, ptr [[MY_ALLOCATOR]], align 8
 141 // CHECK-NEXT:    [[CONV1:%.*]] = inttoptr i64 [[TMP3]] to ptr
 142 // CHECK-NEXT:    call void @__kmpc_destroy_allocator(i32 [[TMP0]], ptr [[CONV1]])
 143 // CHECK-NEXT:    ret void
 144 //
 145 //
 146 // CHECK-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3foov_l66.omp_outlined
 147 // CHECK-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]]) #[[ATTR1]] {
 148 // CHECK-NEXT:  entry:
 149 // CHECK-NEXT:    [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 8
 150 // CHECK-NEXT:    [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 8
 151 // CHECK-NEXT:    [[DOTOMP_IV:%.*]] = alloca i32, align 4
 152 // CHECK-NEXT:    [[TMP:%.*]] = alloca i32, align 4
 153 // CHECK-NEXT:    [[DOTOMP_LB:%.*]] = alloca i32, align 4
 154 // CHECK-NEXT:    [[DOTOMP_UB:%.*]] = alloca i32, align 4
 155 // CHECK-NEXT:    [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
 156 // CHECK-NEXT:    [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
 157 // CHECK-NEXT:    [[I:%.*]] = alloca i32, align 4
 158 // CHECK-NEXT:    store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 8
 159 // CHECK-NEXT:    store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 8
 160 // CHECK-NEXT:    store i32 0, ptr [[DOTOMP_LB]], align 4
 161 // CHECK-NEXT:    store i32 9, ptr [[DOTOMP_UB]], align 4
 162 // CHECK-NEXT:    store i32 1, ptr [[DOTOMP_STRIDE]], align 4
 163 // CHECK-NEXT:    store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
 164 // CHECK-NEXT:    [[TMP0:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
 165 // CHECK-NEXT:    [[TMP1:%.*]] = load i32, ptr [[TMP0]], align 4
 166 // CHECK-NEXT:    call void @__kmpc_for_static_init_4(ptr @[[GLOB2:[0-9]+]], i32 [[TMP1]], i32 34, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
 167 // CHECK-NEXT:    [[TMP2:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
 168 // CHECK-NEXT:    [[CMP:%.*]] = icmp sgt i32 [[TMP2]], 9
 169 // CHECK-NEXT:    br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
 170 // CHECK:       cond.true:
 171 // CHECK-NEXT:    br label [[COND_END:%.*]]
 172 // CHECK:       cond.false:
 173 // CHECK-NEXT:    [[TMP3:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
 174 // CHECK-NEXT:    br label [[COND_END]]
 175 // CHECK:       cond.end:
 176 // CHECK-NEXT:    [[COND:%.*]] = phi i32 [ 9, [[COND_TRUE]] ], [ [[TMP3]], [[COND_FALSE]] ]
 177 // CHECK-NEXT:    store i32 [[COND]], ptr [[DOTOMP_UB]], align 4
 178 // CHECK-NEXT:    [[TMP4:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4
 179 // CHECK-NEXT:    store i32 [[TMP4]], ptr [[DOTOMP_IV]], align 4
 180 // CHECK-NEXT:    br label [[OMP_INNER_FOR_COND:%.*]]
 181 // CHECK:       omp.inner.for.cond:
 182 // CHECK-NEXT:    [[TMP5:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
 183 // CHECK-NEXT:    [[TMP6:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
 184 // CHECK-NEXT:    [[CMP1:%.*]] = icmp sle i32 [[TMP5]], [[TMP6]]
 185 // CHECK-NEXT:    br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
 186 // CHECK:       omp.inner.for.body:
 187 // CHECK-NEXT:    [[TMP7:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
 188 // CHECK-NEXT:    [[MUL:%.*]] = mul nsw i32 [[TMP7]], 1
 189 // CHECK-NEXT:    [[ADD:%.*]] = add nsw i32 0, [[MUL]]
 190 // CHECK-NEXT:    store i32 [[ADD]], ptr [[I]], align 4
 191 // CHECK-NEXT:    br label [[OMP_BODY_CONTINUE:%.*]]
 192 // CHECK:       omp.body.continue:
 193 // CHECK-NEXT:    br label [[OMP_INNER_FOR_INC:%.*]]
 194 // CHECK:       omp.inner.for.inc:
 195 // CHECK-NEXT:    [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
 196 // CHECK-NEXT:    [[ADD2:%.*]] = add nsw i32 [[TMP8]], 1
 197 // CHECK-NEXT:    store i32 [[ADD2]], ptr [[DOTOMP_IV]], align 4
 198 // CHECK-NEXT:    br label [[OMP_INNER_FOR_COND]]
 199 // CHECK:       omp.inner.for.end:
 200 // CHECK-NEXT:    br label [[OMP_LOOP_EXIT:%.*]]
 201 // CHECK:       omp.loop.exit:
 202 // CHECK-NEXT:    call void @__kmpc_for_static_fini(ptr @[[GLOB2]], i32 [[TMP1]])
 203 // CHECK-NEXT:    ret void
 204 //
 205 //
 206 // CHECK-LABEL: define {{[^@]+}}@.omp_offloading.requires_reg
 207 // CHECK-SAME: () #[[ATTR3:[0-9]+]] {
 208 // CHECK-NEXT:  entry:
 209 // CHECK-NEXT:    call void @__tgt_register_requires(i64 1)
 210 // CHECK-NEXT:    ret void
 211 //