File size: 11,108 Bytes
0dc1b04 |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 |
//===----------------------------------------------------------------------===//
//
// Part of libcu++, the C++ Standard Library for your entire system,
// under the Apache License v2.0 with LLVM Exceptions.
// See https://llvm.org/LICENSE.txt for license information.
// SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
//
//===----------------------------------------------------------------------===//
#ifndef _CUDA_ATOMIC
#define _CUDA_ATOMIC
#ifndef __CUDACC_RTC__
#include <atomic>
static_assert(ATOMIC_BOOL_LOCK_FREE == 2, "");
static_assert(ATOMIC_CHAR_LOCK_FREE == 2, "");
static_assert(ATOMIC_CHAR16_T_LOCK_FREE == 2, "");
static_assert(ATOMIC_CHAR32_T_LOCK_FREE == 2, "");
static_assert(ATOMIC_WCHAR_T_LOCK_FREE == 2, "");
static_assert(ATOMIC_SHORT_LOCK_FREE == 2, "");
static_assert(ATOMIC_INT_LOCK_FREE == 2, "");
static_assert(ATOMIC_LONG_LOCK_FREE == 2, "");
static_assert(ATOMIC_LLONG_LOCK_FREE == 2, "");
static_assert(ATOMIC_POINTER_LOCK_FREE == 2, "");
#undef ATOMIC_BOOL_LOCK_FREE
#undef ATOMIC_BOOL_LOCK_FREE
#undef ATOMIC_CHAR_LOCK_FREE
#undef ATOMIC_CHAR16_T_LOCK_FREE
#undef ATOMIC_CHAR32_T_LOCK_FREE
#undef ATOMIC_WCHAR_T_LOCK_FREE
#undef ATOMIC_SHORT_LOCK_FREE
#undef ATOMIC_INT_LOCK_FREE
#undef ATOMIC_LONG_LOCK_FREE
#undef ATOMIC_LLONG_LOCK_FREE
#undef ATOMIC_POINTER_LOCK_FREE
#undef ATOMIC_FLAG_INIT
#undef ATOMIC_VAR_INIT
#endif //__CUDACC_RTC__
// pre-define lock free query for heterogeneous compatibility
#ifndef _LIBCUDACXX_ATOMIC_IS_LOCK_FREE
#define _LIBCUDACXX_ATOMIC_IS_LOCK_FREE(__x) (__x <= 8)
#endif
#include "cassert"
#include "cstddef"
#include "cstdint"
#include "type_traits"
#include "version"
#include "detail/__config"
#include "detail/__pragma_push"
#include "detail/__threading_support"
#include "detail/libcxx/include/atomic"
_LIBCUDACXX_BEGIN_NAMESPACE_CUDA
using std::__detail::thread_scope;
using std::__detail::thread_scope_system;
using std::__detail::thread_scope_device;
using std::__detail::thread_scope_block;
using std::__detail::thread_scope_thread;
namespace __detail {
using std::__detail::__thread_scope_block_tag;
using std::__detail::__thread_scope_device_tag;
using std::__detail::__thread_scope_system_tag;
}
using memory_order = std::memory_order;
constexpr memory_order memory_order_relaxed = std::memory_order_relaxed;
constexpr memory_order memory_order_consume = std::memory_order_consume;
constexpr memory_order memory_order_acquire = std::memory_order_acquire;
constexpr memory_order memory_order_release = std::memory_order_release;
constexpr memory_order memory_order_acq_rel = std::memory_order_acq_rel;
constexpr memory_order memory_order_seq_cst = std::memory_order_seq_cst;
// atomic<T>
template <class _Tp, thread_scope _Sco = thread_scope::thread_scope_system>
struct atomic
: public std::__atomic_base<_Tp, _Sco>
{
typedef std::__atomic_base<_Tp, _Sco> __base;
constexpr atomic() noexcept = default;
__host__ __device__
constexpr atomic(_Tp __d) noexcept : __base(__d) {}
__host__ __device__
_Tp operator=(_Tp __d) volatile noexcept
{__base::store(__d); return __d;}
__host__ __device__
_Tp operator=(_Tp __d) noexcept
{__base::store(__d); return __d;}
__host__ __device__
_Tp fetch_max(const _Tp & __op, memory_order __m = memory_order_seq_cst) volatile noexcept
{
return std::__detail::__cxx_atomic_fetch_max(&this->__a_, __op, __m);
}
__host__ __device__
_Tp fetch_min(const _Tp & __op, memory_order __m = memory_order_seq_cst) volatile noexcept
{
return std::__detail::__cxx_atomic_fetch_min(&this->__a_, __op, __m);
}
};
// atomic<T*>
template <class _Tp, thread_scope _Sco>
struct atomic<_Tp*, _Sco>
: public std::__atomic_base<_Tp*, _Sco>
{
typedef std::__atomic_base<_Tp*, _Sco> __base;
constexpr atomic() noexcept = default;
__host__ __device__
constexpr atomic(_Tp* __d) noexcept : __base(__d) {}
__host__ __device__
_Tp* operator=(_Tp* __d) volatile noexcept
{__base::store(__d); return __d;}
__host__ __device__
_Tp* operator=(_Tp* __d) noexcept
{__base::store(__d); return __d;}
__host__ __device__
_Tp* fetch_add(ptrdiff_t __op, memory_order __m = memory_order_seq_cst)
volatile noexcept
{return __cxx_atomic_fetch_add(&this->__a_, __op, __m);}
__host__ __device__
_Tp* fetch_add(ptrdiff_t __op, memory_order __m = memory_order_seq_cst) noexcept
{return __cxx_atomic_fetch_add(&this->__a_, __op, __m);}
__host__ __device__
_Tp* fetch_sub(ptrdiff_t __op, memory_order __m = memory_order_seq_cst)
volatile noexcept
{return __cxx_atomic_fetch_sub(&this->__a_, __op, __m);}
__host__ __device__
_Tp* fetch_sub(ptrdiff_t __op, memory_order __m = memory_order_seq_cst) noexcept
{return __cxx_atomic_fetch_sub(&this->__a_, __op, __m);}
__host__ __device__
_Tp* operator++(int) volatile noexcept {return fetch_add(1);}
__host__ __device__
_Tp* operator++(int) noexcept {return fetch_add(1);}
__host__ __device__
_Tp* operator--(int) volatile noexcept {return fetch_sub(1);}
__host__ __device__
_Tp* operator--(int) noexcept {return fetch_sub(1);}
__host__ __device__
_Tp* operator++() volatile noexcept {return fetch_add(1) + 1;}
__host__ __device__
_Tp* operator++() noexcept {return fetch_add(1) + 1;}
__host__ __device__
_Tp* operator--() volatile noexcept {return fetch_sub(1) - 1;}
__host__ __device__
_Tp* operator--() noexcept {return fetch_sub(1) - 1;}
__host__ __device__
_Tp* operator+=(ptrdiff_t __op) volatile noexcept {return fetch_add(__op) + __op;}
__host__ __device__
_Tp* operator+=(ptrdiff_t __op) noexcept {return fetch_add(__op) + __op;}
__host__ __device__
_Tp* operator-=(ptrdiff_t __op) volatile noexcept {return fetch_sub(__op) - __op;}
__host__ __device__
_Tp* operator-=(ptrdiff_t __op) noexcept {return fetch_sub(__op) - __op;}
};
// atomic_ref<T>
template <class _Tp, thread_scope _Sco = thread_scope::thread_scope_system>
struct atomic_ref
: public std::__atomic_base_ref<_Tp, _Sco>
{
typedef std::__atomic_base_ref<_Tp, _Sco> __base;
__host__ __device__
constexpr atomic_ref(_Tp& __d) noexcept : __base(__d) {}
__host__ __device__
_Tp operator=(_Tp __d) const volatile noexcept
{__base::store(__d); return __d;}
__host__ __device__
_Tp operator=(_Tp __d) const noexcept
{__base::store(__d); return __d;}
__host__ __device__
_Tp fetch_max(const _Tp & __op, memory_order __m = memory_order_seq_cst) const volatile noexcept
{
return std::__detail::__cxx_atomic_fetch_max(&this->__a_, __op, __m);
}
__host__ __device__
_Tp fetch_min(const _Tp & __op, memory_order __m = memory_order_seq_cst) const volatile noexcept
{
return std::__detail::__cxx_atomic_fetch_min(&this->__a_, __op, __m);
}
};
// atomic_ref<T*>
template <class _Tp, thread_scope _Sco>
struct atomic_ref<_Tp*, _Sco>
: public std::__atomic_base_ref<_Tp*, _Sco>
{
typedef std::__atomic_base_ref<_Tp*, _Sco> __base;
__host__ __device__
constexpr atomic_ref(_Tp*& __d) noexcept : __base(__d) {}
__host__ __device__
_Tp* operator=(_Tp* __d) const volatile noexcept
{__base::store(__d); return __d;}
__host__ __device__
_Tp* operator=(_Tp* __d) const noexcept
{__base::store(__d); return __d;}
__host__ __device__
_Tp* fetch_add(ptrdiff_t __op,
memory_order __m = memory_order_seq_cst) const volatile noexcept
{return __cxx_atomic_fetch_add(&this->__a_, __op, __m);}
__host__ __device__
_Tp* fetch_add(ptrdiff_t __op,
memory_order __m = memory_order_seq_cst) const noexcept
{return __cxx_atomic_fetch_add(&this->__a_, __op, __m);}
__host__ __device__
_Tp* fetch_sub(ptrdiff_t __op,
memory_order __m = memory_order_seq_cst) const volatile noexcept
{return __cxx_atomic_fetch_sub(&this->__a_, __op, __m);}
__host__ __device__
_Tp* fetch_sub(ptrdiff_t __op,
memory_order __m = memory_order_seq_cst) const noexcept
{return __cxx_atomic_fetch_sub(&this->__a_, __op, __m);}
__host__ __device__
_Tp* operator++(int) const volatile noexcept {return fetch_add(1);}
__host__ __device__
_Tp* operator++(int) const noexcept {return fetch_add(1);}
__host__ __device__
_Tp* operator--(int) const volatile noexcept {return fetch_sub(1);}
__host__ __device__
_Tp* operator--(int) const noexcept {return fetch_sub(1);}
__host__ __device__
_Tp* operator++() const volatile noexcept {return fetch_add(1) + 1;}
__host__ __device__
_Tp* operator++() const noexcept {return fetch_add(1) + 1;}
__host__ __device__
_Tp* operator--() const volatile noexcept {return fetch_sub(1) - 1;}
__host__ __device__
_Tp* operator--() const noexcept {return fetch_sub(1) - 1;}
__host__ __device__
_Tp* operator+=(ptrdiff_t __op) const volatile noexcept {return fetch_add(__op) + __op;}
__host__ __device__
_Tp* operator+=(ptrdiff_t __op) const noexcept {return fetch_add(__op) + __op;}
__host__ __device__
_Tp* operator-=(ptrdiff_t __op) const volatile noexcept {return fetch_sub(__op) - __op;}
__host__ __device__
_Tp* operator-=(ptrdiff_t __op) const noexcept {return fetch_sub(__op) - __op;}
};
inline __host__ __device__ void atomic_thread_fence(memory_order __m, thread_scope _Scope = thread_scope::thread_scope_system) {
NV_DISPATCH_TARGET(
NV_IS_DEVICE, (
switch(_Scope) {
case thread_scope::thread_scope_system:
std::__detail::__atomic_thread_fence_cuda((int)__m, __detail::__thread_scope_system_tag());
break;
case thread_scope::thread_scope_device:
std::__detail::__atomic_thread_fence_cuda((int)__m, __detail::__thread_scope_device_tag());
break;
case thread_scope::thread_scope_block:
std::__detail::__atomic_thread_fence_cuda((int)__m, __detail::__thread_scope_block_tag());
break;
}
),
NV_IS_HOST, (
(void) _Scope;
std::atomic_thread_fence(__m);
)
)
}
inline __host__ __device__ void atomic_signal_fence(memory_order __m) {
std::atomic_signal_fence(__m);
}
_LIBCUDACXX_END_NAMESPACE_CUDA
#include "detail/__pragma_pop"
#endif //_CUDA_ATOMIC
|