std:call_once是C++11引入的新特性，如需使用，只需要#include <mutex>即可，简单来说std:call_once的作用，确保函数或代码片段在多线程环境下，只需要执行一次，常用的场景如Init()操作或一些系统参数的获取等。

相对来说，std::call_once用法比较简单，配合std::once_flag即可实现，废话不多说，上代码：

#include "stdafx.h"
#include <iostream>
#include <thread>
#include <mutex>std::once_flag flag;void Initialize()
{std::cout << "Run into Initialize.." << std::endl;
}void Init()
{std::call_once(flag, Initialize);
}int main()
{std::thread t1(Init);std::thread t2(Init);std::thread t3(Init);std::thread t4(Init);t1.join();t2.join();t3.join();t4.join();
}

执行结果如下：

 

可以看到，Initialize()只调用了一次。

需要了解更多的话，可以参考​​https://en.cppreference.com/w/cpp/thread/call_once​​关于std::call_once的描述：

Executes the ​​Callable​​​ object ​​f​​ exactly once, even if called concurrently, from several threads.

In detail:

• If, by the time​​call_once​​​ is called,​​flag​​​ indicates that​​f​​​ was already called,​​call_once​​​ returns right away (such a call to​​call_once​​ is known aspassive).
• Otherwise, call_once invokes ​​std::forward​​<Callable>(f)with the arguments ​​std::forward​​<Args>(args)... (as if by​​std::invoke​​​). Unlike the​​std::thread​​​ constructor or​​std::async​​​, the arguments are not moved or copied because they don't need to be transferred to another thread of execution. (such a call to​​call_once​​ is known asactive).
• If that invocation throws an exception, it is propagated to the caller of​​call_once​​​, and the flag is not flipped so that another call will be attempted (such a call to​​call_once​​ is known asexceptional).
• If that invocation returns normally (such a call to​​call_once​​ is known asreturning), the flag is flipped, and all other calls to​​call_once​​ with the same flag are guaranteed to bepassive.

All active calls on the same ​​flag​​ form a single total order consisting of zero or more exceptional calls, followed by one returning call. The end of each active call synchronizes-with the next active call in that order.

The return from the returning call synchronizes-with the returns from all passive calls on the same ​​flag​​​: this means that all concurrent calls to ​​call_once​​ are guaranteed to observe any side-effects made by the active call, with no additional synchronization.