关于c++:C11的conditionvariable实现WaitForSingleObject功能

85次阅读

共计 1818 个字符，预计需要花费 5 分钟才能阅读完成。

在多线程程序开发中，咱们有时会心愿期待其余线程完结后在继续执行某些工作。比方常见的，主线程须要期待其余所有子线程完结后再完结程序。在 Windows 下咱们个别应用 WaitForSingleObject，或者 WaitForMultipleObjects 来实现，它们都会期待 Object 被设置为有信号时才返回的。这样咱们在主线程完结的中央 wait，在子线程完结的中央设置信号，就能实现上述性能。

然而这个 API 是 Windows 下的，无奈跨平台应用。还好 C ++11 为咱们带来了多线程的反对。在 C ++11 中咱们能够通过 mutex 和 condition_variable 来达到类似成果。上面是我写的小例子。

#include <iostream>
#include <thread>
#include <mutex>
#include <condition_variable>
#include <chrono>

std::mutex mx;

std::mutex g_mx;
std::condition_variable g_cv;
uint32_t g_thread_num = 0;

void function_1()
{for (int i = 0; i < 100; i++)
  {std::lock_guard<std::mutex> loc(mx);
      std::cout << "from function 1:" << i << std::endl;
  }

  std::unique_lock<std::mutex> guard(g_mx);
  --g_thread_num;
  g_cv.notify_one();
  guard.unlock();

  std::cout << "function 1 end" << std::endl;
}
void function_2()
{for (int i = 0; i < 100; i++)
  {std::lock_guard<std::mutex> loc(mx);
      std::cout << "from function 2:" << i << std::endl;
  }

  std::unique_lock<std::mutex> guard(g_mx);
  --g_thread_num;
  g_cv.notify_one();
  guard.unlock();

  std::cout << "function 2 end" << std::endl;
}
void function_3()
{for (int i = 0; i < 100; i++)
  {std::lock_guard<std::mutex> loc(mx);
      std::cout << "from function 3:" << i << std::endl;
  }
  std::this_thread::sleep_for(std::chrono::seconds(3));

  std::unique_lock<std::mutex> guard(g_mx);
  --g_thread_num;
  g_cv.notify_one();
  guard.unlock();

  std::cout << "function 3 end" << std::endl;
}

int main()
{std::thread t1(function_1);
  {std::unique_lock<std::mutex> guard(g_mx);
      ++g_thread_num;
  }
  t1.detach();
  std::thread t2(function_2);
  {std::unique_lock<std::mutex> guard(g_mx);
      ++g_thread_num;
  }
  t2.detach();
  std::thread t3(function_3);
  {std::unique_lock<std::mutex> guard(g_mx);
      ++g_thread_num;
  }
  t3.detach();

  for (int i = 0; i < 100; i++)
  {std::lock_guard<std::mutex> loc(mx);
      std::cout << "from function main:" << i << std::endl;
  }
  

  {std::unique_lock<std::mutex> guard(g_mx);
      g_cv.wait(guard, []() { return g_thread_num == 0;});
  }

  std::cout << "main end" << std::endl;

  return 0;
}