std::atomic_fetch_add, std::atomic_fetch_add_explicit

 

C++

編譯器支援

自由（freestanding）與宿主（hosted）

語言

標準庫

標準庫標頭檔案

具名要求

特性測試宏 (C++20)

語言支援庫

概念庫 (C++20)

診斷庫

記憶體管理庫

超程式設計庫 (C++11)

通用工具庫

容器庫

迭代器庫

範圍庫 (C++20)

演算法庫

字串庫

文字處理庫

數值庫

日期和時間庫

輸入/輸出庫

檔案系統庫 (C++17)

併發支援庫 (C++11)

執行控制庫 (C++26)

技術規範

符號索引

外部庫

 

併發支援庫

執行緒 thread (C++11) jthread (C++20) hardware_destructive_interference_sizehardware_constructive_interference_size (C++17)(C++17) this_thread 名稱空間 get_id (C++11) yield (C++11) sleep_for (C++11) sleep_until (C++11) 協同取消 stop_token (C++20) inplace_stop_token (C++26) never_stop_token (C++26) stop_source (C++20) inplace_stop_source (C++26)   stop_callback (C++20) inplace_stop_callback (C++26) stop_callback_for_t (C++26) stoppable_token (C++26) unstoppable_token (C++26) stoppable-source (C++26) stoppable-callback-for (C++26)   互斥 mutex (C++11) recursive_mutex (C++11) shared_mutex (C++17) timed_mutex (C++11) recursive_timed_mutex (C++11) shared_timed_mutex (C++14) 通用鎖管理 lock (C++11) lock_guard (C++11) scoped_lock (C++17) unique_lock (C++11) shared_lock (C++14) once_flag (C++11) call_once (C++11) try_lock (C++11) defer_locktry_to_lockadopt_lockdefer_lock_ttry_to_lock_tadopt_lock_t (C++11)(C++11)(C++11)(C++11)(C++11)(C++11) 條件變數 condition_variable (C++11) condition_variable_any (C++11) notify_all_at_thread_exit (C++11) cv_status (C++11) 訊號量 counting_semaphorebinary_semaphore (C++20)(C++20) 門閂和屏障 latch (C++20) barrier (C++20) 期值 promise (C++11) future (C++11) shared_future (C++11) packaged_task (C++11) async (C++11) launch (C++11) future_status (C++11) future_error (C++11) future_category (C++11) future_errc (C++11) 安全回收 rcu_obj_base (C++26) rcu_domain (C++26) rcu_default_domain (C++26) rcu_synchronize (C++26) rcu_barrier (C++26) rcu_retire (C++26) 危險指標 hazard_pointer_obj_base (C++26) hazard_pointer (C++26) make_hazard_pointer (C++26)

原子型別 atomic (C++11) atomic_ref (C++20) atomic_flag (C++11) 原子型別的初始化 atomic_init (C++11)(C++20 中已棄用) ATOMIC_VAR_INIT (C++11)(C++20 中已棄用) ATOMIC_FLAG_INIT (C++11) 記憶體排序 memory_order (C++11) kill_dependency (C++11)(C++26 中已棄用) atomic_thread_fence (C++11) atomic_signal_fence (C++11) 原子操作的自由函式 atomic_storeatomic_store_explicit (C++11)(C++11) atomic_loadatomic_load_explicit (C++11)(C++11) atomic_exchangeatomic_exchange_explicit (C++11)(C++11) atomic_compare_exchange_weakatomic_compare_exchange_weak_explicitatomic_compare_exchange_strongatomic_compare_exchange_strong_explicit (C++11)(C++11)(C++11)(C++11) atomic_fetch_addatomic_fetch_add_explicit (C++11)(C++11) atomic_fetch_subatomic_fetch_sub_explicit (C++11)(C++11) atomic_fetch_andatomic_fetch_and_explicit (C++11)(C++11) atomic_fetch_oratomic_fetch_or_explicit (C++11)(C++11) atomic_fetch_xoratomic_fetch_xor_explicit (C++11)(C++11) atomic_fetch_maxatomic_fetch_max_explicit (C++26)(C++26) atomic_fetch_minatomic_fetch_min_explicit (C++26)(C++26) atomic_is_lock_free (C++11) atomic_waitatomic_wait_explicit (C++20)(C++20) atomic_notify_one (C++20) atomic_notify_all (C++20) 原子標誌的自由函式 atomic_flag_test_and_setatomic_flag_test_and_set_explicit (C++11)(C++11) atomic_flag_clearatomic_flag_clear_explicit (C++11)(C++11) atomic_flag_testatomic_flag_test_explicit (C++20)(C++20) atomic_flag_waitatomic_flag_wait_explicit (C++20)(C++20) atomic_flag_notify_one (C++20) atomic_flag_notify_all (C++20)

 

定義於標頭檔案 <atomic>
template< class T > T atomic_fetch_add( std::atomic<T>* obj,                     typename std::atomic<T>::difference_type arg ) noexcept;	(1)	(C++11 起)
template< class T > T atomic_fetch_add( volatile std::atomic<T>* obj,                     typename std::atomic<T>::difference_type arg ) noexcept;	(2)	(C++11 起)
template< class T > T atomic_fetch_add_explicit( std::atomic<T>* obj,                              typename std::atomic<T>::difference_type arg,                              std::memory_order order ) noexcept;	(3)	(C++11 起)
template< class T > T atomic_fetch_add_explicit( volatile std::atomic<T>* obj,                              typename std::atomic<T>::difference_type arg,                              std::memory_order order ) noexcept;	(4)	(C++11 起)

執行原子加法操作。原子地將 arg 加到 obj 指向的值，並返回 obj 之前持有的值。操作的執行方式如同以下程式碼被執行：

1,2) obj->fetch_add(arg)

3,4) obj->fetch_add(arg, order)

如果 std::atomic<T> 沒有 fetch_add 成員（該成員僅為整型、浮點型(C++20 起)和指標型別提供，除了 bool），則程式格式不正確。

引數

obj	-	指向要修改的原子物件的指標
arg	-	要加到原子物件中儲存的值上的值
順序	-	記憶體同步順序

返回值

在 *obj 的修改順序中，此函式效果緊前的值。

示例

可以使用 std::atomic_fetch_add 實現單寫入器/多讀取器鎖。請注意，這種簡單的實現並非無飢餓（lockout-free）。

執行此程式碼

#include <atomic>
#include <chrono>
#include <iostream>
#include <string>
#include <thread>
#include <vector>
 
using namespace std::chrono_literals;
 
// meaning of cnt:
//  5: readers and writer are in race. There are no active readers or writers.
//  4...0: there are 1...5 active readers, The writer is blocked.
// -1: writer won the race and readers are blocked.
 
const int N = 5; // four concurrent readers are allowed
std::atomic<int> cnt(N);
 
std::vector<int> data;
 
void reader(int id)
{
    for (;;)
    {
        // lock
        while (std::atomic_fetch_sub(&cnt, 1) <= 0)
            std::atomic_fetch_add(&cnt, 1);
 
        // read
        if (!data.empty())
            std::cout << ("reader " + std::to_string(id) +
                          " sees " + std::to_string(*data.rbegin()) + '\n');
        if (data.size() == 25)
            break;
 
        // unlock
        std::atomic_fetch_add(&cnt, 1);
 
        // pause
        std::this_thread::sleep_for(1ms);
    }
}
 
void writer()
{
    for (int n = 0; n < 25; ++n)
    {
        // lock
        while (std::atomic_fetch_sub(&cnt, N + 1) != N)
            std::atomic_fetch_add(&cnt, N + 1);
 
        // write
        data.push_back(n);
        std::cout << "writer pushed back " << n << '\n';
 
        // unlock
        std::atomic_fetch_add(&cnt, N + 1);
 
        // pause
        std::this_thread::sleep_for(1ms);
    }
}
 
int main()
{
    std::vector<std::thread> v;
    for (int n = 0; n < N; ++n)
        v.emplace_back(reader, n);
    v.emplace_back(writer);
 
    for (auto& t : v)
        t.join();
}

輸出

writer pushed back 0
reader 2 sees 0
reader 3 sees 0
reader 1 sees 0
<...>
reader 2 sees 24
reader 4 sees 24
reader 1 sees 24

缺陷報告

下列更改行為的缺陷報告追溯地應用於以前出版的 C++ 標準。

缺陷報告	應用於	釋出時的行為	正確的行為
P0558R1	C++11	因為要求精確型別匹配 T 是從多個引數推匯出來的	T 僅從 obj 中推導

參閱

fetch_add	原子地將引數新增到原子物件中儲存的值，並獲取之前持有的值 (std::atomic<T> 的公開成員函式)
atomic_fetch_subatomic_fetch_sub_explicit (C++11)(C++11)	從原子物件中減去非原子值並獲取原子的先前值 (函式模板)
C 文件 用於 atomic_fetch_add, atomic_fetch_add_explicit