C++11 執行緒同步介面std::condition_variable和std::future的簡單使用

robot2017發表於2024-09-17

std::condition_variable

條件變數std::condition_variable有wait和notify介面用於執行緒間的同步。如下圖所示,Thread 2阻塞在wait介面,Thread 1透過notify介面通知Thread 2繼續執行。

con_variable_result

具體參見示例程式碼:

#include<iostream>
#include<mutex>
#include<thread>
#include<queue>
std::mutex mt;
std::queue<int> data;
std::condition_variable cv;
auto start=std::chrono::high_resolution_clock::now();

void logCurrentTime()
{
	auto end = std::chrono::high_resolution_clock::now();
	auto elapsed = std::chrono::duration_cast<std::chrono::milliseconds>(end - start).count();
	std::cout << elapsed << ":";
}
void prepare_data()
{	
	logCurrentTime();
	std::cout << "this is " << __FUNCTION__ << " thread:" << std::this_thread::get_id() << std::endl;
	for (int i = 0; i < 10; i++)
	{
		data.push(i);
		logCurrentTime();
		std::cout << "data OK:" << i << std::endl;
	}
	//start to notify consume_data thread data is OK!
	cv.notify_one();
}


void consume_data()
{
	logCurrentTime();
	std::cout << "this is: " << __FUNCTION__ << " thread:" << std::this_thread::get_id() << std::endl;
	std::unique_lock<std::mutex> lk(mt);
	//wait first for notification
	cv.wait(lk);  //it must accept a unique_lock parameter to wait

	while (!data.empty())
	{
		logCurrentTime();
		std::cout << "data consumed: " << data.front() << std::endl;
		data.pop();
	}
}


int main()
{
	std::thread t2(consume_data);
	//wait for a while to wait first then prepare data,otherwise stuck on wait
	std::this_thread::sleep_for(std::chrono::milliseconds(10));
	std::thread t1(prepare_data);
	t1.join();
	t2.join();
	return 0;
}

輸出結果

con_variable_result

分析

主執行緒中另啟兩個執行緒,分別執行consume_data和prepare_data,其中consume_data要先執行,以保證先等待再通知,否則若先通知再等待就死鎖了。首先consume_data執行緒在從wait 處阻塞等待。後prepare_data執行緒中依次向佇列寫入0-10,寫完之後透過notify_one 通知consume_data執行緒解除阻塞,依次讀取0-10。

std::future

std::future與std::async配合非同步執行程式碼,再透過wait或get介面阻塞當前執行緒等待結果。如下圖所示,Thread 2中future介面的get或wait介面會阻塞當前執行緒,std::async非同步開啟的新執行緒Thread1執行結束後,將結果存於std::future後通知Thread 1獲取結果後繼續執行.

con_variable_result

具體參見如下程式碼:

#include <iostream>
#include <future>
#include<thread>

int test()
{
	std::cout << "this is " << __FUNCTION__ << " thread:" << std::this_thread::get_id() << std::endl;;
	std::this_thread::sleep_for(std::chrono::microseconds(1000));
	return 10;
}
int main()
{
	std::cout << "this is " <<__FUNCTION__<<" thread:" << std::this_thread::get_id() << std::endl;;
	//this will lanuch on another thread
	std::future<int> result = std::async(test);

	std::cout << "After lanuch a thread: "<< std::this_thread::get_id() << std::endl;

	//block the thread and wait for the result
	std::cout << "result is: " <<result.get()<< std::endl;

	std::cout << "After get result "<< std::endl;

	return 0;
}

輸出結果

執行結果

分析

主程式中呼叫std::async非同步呼叫test函式,可以看到main函式的執行緒ID 27428與test函式執行的執行緒ID 9704並不一樣,說明std::async另起了一個新的執行緒。在test執行緒中,先sleep 1000ms,所以可以看到"After lanuch a thread:"先輸出,說明主執行緒非同步執行,不受子執行緒影響。而"After get result "最後輸出,說明get()方法會阻塞主執行緒,直到獲取結果。

相關文章