Bootstrap

c++ 简单线程池

头文件

#pragma once
#include <iostream>
#include <vector>
#include <queue>
#include <thread>
#include <mutex>
#include <condition_variable>
#include <functional>
#include <future>

class ThreadPool {
public:
    ThreadPool(size_t numThreads);
    ~ThreadPool();

    template<class F, class... Args>
    auto enqueue(F&& f, Args&&... args)->std::future<typename std::result_of<F(Args...)>::type>;
private:
    void workerThread();
private:
    std::vector<std::thread> workers;
    std::queue<std::function<void()>> tasks;

    std::mutex queueMutex;
    std::condition_variable condition;
    bool stop;
};

cpp

 #include "threadPool.h"

ThreadPool::ThreadPool(size_t numThreads) : stop(false)
{
    for (size_t i = 0; i < numThreads; ++i)
    {
        workers.emplace_back(&ThreadPool::workerThread, this);
    }
}

ThreadPool::~ThreadPool() {
    {
        std::unique_lock<std::mutex> lock(queueMutex);
        stop = true;
    }
    condition.notify_all();
    for (std::thread& worker : workers) {
        worker.join();
    }
}

void ThreadPool::workerThread() {
    while (true) {
        std::function<void()> task;
        {
            std::unique_lock<std::mutex> lock(queueMutex);
            condition.wait(lock, [this] { return stop || !tasks.empty(); });
            if (stop && tasks.empty()) {
                return;
            }
            if (!tasks.empty()) {
                task = std::move(tasks.front());
                tasks.pop();
            }
        }
        if (task) {
            task();
        }
    }
}

template<class F, class... Args>
auto ThreadPool::enqueue(F&& f, Args&&... args) -> std::future<typename std::result_of<F(Args...)>::type> {
    using returnType = typename std::result_of<F(Args...)>::type;

    auto task = std::make_shared<std::packaged_task<returnType()>>(
        std::bind(std::forward<F>(f), std::forward<Args>(args)...)
        );

    std::future<returnType> res = task->get_future();
    {
        std::unique_lock<std::mutex> lock(queueMutex);

        if (stop) {
            throw std::runtime_error("enqueue on stopped ThreadPool");
        }

        tasks.emplace([task]() { (*task)(); });
    }
    condition.notify_one();
    return res;
}

int main() {
    ThreadPool pool(4);

    auto result1 = pool.enqueue([](int answer) { return answer; }, 42);
    auto result2 = pool.enqueue([](int a, int b) { return a + b; }, 5, 7);

    std::cout << "Result1: " << result1.get() << std::endl;
    std::cout << "Result2: " << result2.get() << std::endl;

    return 0;
}

解释

enqueue

Sure, let’s break down the enqueue function template step-by-step:

Template Definition

template<class F, class... Args>

This defines a function template that can take any callable object F (such as a function, function pointer, lambda, or functor) and a variadic list of arguments Args. The class... Args syntax allows the function to accept any number of additional arguments, making it very flexible.

Return Type

auto enqueue(F&& f, Args&&... args) -> std::future<typename std::result_of<F(Args...)>::type>;

The return type of the function is a std::future that will hold the result of invoking the callable object F with arguments Args.... The std::result_of<F(Args...)>::type part determines the type that will be returned by calling F with Args....

  • std::result_of<F(Args...)>::type: This uses the std::result_of type trait to deduce the return type of calling the function F with arguments Args....

Function Body

auto task = std::make_shared<std::packaged_task<returnType()>>(
    std::bind(std::forward<F>(f), std::forward<Args>(args)...)
);

Here, we are creating a std::shared_ptr to a std::packaged_task. A std::packaged_task wraps a callable object, allowing it to be executed asynchronously and providing a future to retrieve the result.

  • returnType: This is a type alias for the return type of the callable object F with arguments Args....
  • std::make_shared<std::packaged_task<returnType()>>: This creates a shared pointer to a std::packaged_task that will eventually execute the callable F with the provided arguments.
  • std::bind(std::forward<F>(f), std::forward<Args>(args)...): This binds the callable F with the provided arguments Args..., allowing them to be stored and called later.

Storing the Task

{
    std::unique_lock<std::mutex> lock(queueMutex);

    if (stop) {
        throw std::runtime_error("enqueue on stopped ThreadPool");
    }

    tasks.emplace([task](){ (*task)(); });
}

This block ensures that access to the task queue is synchronized using a mutex lock. It checks if the thread pool is stopped and throws an exception if it is. Otherwise, it adds the task to the queue.

  • std::unique_lock<std::mutex> lock(queueMutex): Locks the mutex to ensure thread-safe access to the task queue.
  • if (stop) { throw std::runtime_error("enqueue on stopped ThreadPool"); }: Checks if the thread pool has been stopped. If so, it throws an exception.
  • tasks.emplace([task](){ (*task)(); }): Adds a new task to the task queue. The task is a lambda function that calls the packaged_task to execute the callable F with its arguments.

Notifying Workers

condition.notify_one();

This line notifies one of the worker threads that a new task is available. The worker threads are waiting on this condition variable and will wake up to process the new task.

Returning the Future

return res;

Finally, the function returns the std::future associated with the packaged_task. This future can be used by the caller to get the result of the task once it has been executed by a worker thread.

Full Example for Clarity

Here’s the complete enqueue function with the detailed explanation:

template<class F, class... Args>
auto ThreadPool::enqueue(F&& f, Args&&... args) -> std::future<typename std::result_of<F(Args...)>::type> {
    using returnType = typename std::result_of<F(Args...)>::type;

    // Create a packaged_task to wrap the callable and its arguments
    auto task = std::make_shared<std::packaged_task<returnType()>>(
        std::bind(std::forward<F>(f), std::forward<Args>(args)...)
    );

    // Get the future associated with the task
    std::future<returnType> res = task->get_future();
    {
        std::unique_lock<std::mutex> lock(queueMutex);

        // Ensure the thread pool is not stopped
        if (stop) {
            throw std::runtime_error("enqueue on stopped ThreadPool");
        }

        // Add the task to the queue
        tasks.emplace([task](){ (*task)(); });
    }

    // Notify one of the waiting worker threads
    condition.notify_one();
    return res;
}

Summary

  1. Template Parameters: F is the callable, Args... are the arguments.
  2. Return Type: A std::future that holds the result of the callable.
  3. Creating Packaged Task: Wraps the callable and its arguments to be executed asynchronously.
  4. Mutex Lock: Ensures thread-safe access to the task queue.
  5. Task Queue: Adds the task to the queue.
  6. Notify Worker: Notifies a worker thread to process the new task.
  7. Return Future: Provides a future to get the result of the task.

Next Steps

a. Add unit tests to verify the ThreadPool implementation.
b. Implement error handling for tasks within the ThreadPool.

enqueue

好的，下面是对每一行代码的详细解释：

函数模板声明

template<class F, class... Args>
auto ThreadPool::enqueue(F&& f, Args&&... args) -> std::future<typename std::result_of<F(Args...)>::type>
  • template<class F, class… Args>: 定义一个模板函数，该函数接受一个可调用对象 F 和一组参数 Args...。
  • auto ThreadPool::enqueue(F&& f, Args&&… args): enqueue 是 ThreadPool 类的成员函数，它使用了模板参数 F 和 Args...，并通过右值引用接收这些参数（即使用 &&）。
  • -> std::future<typename std::result_of<F(Args…)>::type>: 函数返回类型是一个 std::future，其类型由 F 调用 Args... 参数后返回的结果类型决定。

使用别名 returnType

using returnType = typename std::result_of<F(Args...)>::type;
  • using returnType: 使用类型别名将 std::result_of<F(Args...)>::type 定义为 returnType。
  • std::result_of<F(Args…)>::type: 通过 std::result_of 获取可调用对象 F 使用参数 Args... 后的返回类型。

创建 packaged_task

auto task = std::make_shared<std::packaged_task<returnType()>>(
    std::bind(std::forward<F>(f), std::forward<Args>(args)...)
);
  • std::make_shared<std::packaged_task<returnType()>>: 创建一个 std::shared_ptr，指向一个 std::packaged_task 对象。
  • std::packaged_task<returnType()>: std::packaged_task 是一个模板类，用于包装一个可调用对象，以便异步调用并获取其结果。
  • std::bind(std::forward(f), std::forward(args)…): 使用 std::bind 将可调用对象 F 和参数 Args... 绑定在一起，生成一个新的可调用对象，并将其传递给 std::packaged_task。std::forward 确保参数的完美转发。

获取 future

std::future<returnType> res = task->get_future();
  • std::future res: 定义一个 std::future 对象 res，用于保存 packaged_task 的结果。
  • task->get_future(): 调用 packaged_task 的 get_future 方法，获取与任务关联的 std::future 对象。

加锁并检查线程池状态

{
    std::unique_lock<std::mutex> lock(queueMutex);
    if (stop) {
        throw std::runtime_error("enqueue on stopped ThreadPool");
    }
    tasks.emplace([task](){ (*task)(); });
}
  • std::unique_lockstd::mutex lock(queueMutex): 创建一个互斥锁对象 lock，并锁定 queueMutex，确保对任务队列的访问是线程安全的。
  • if (stop): 检查线程池是否已停止接受新任务。
  • throw std::runtime_error(“enqueue on stopped ThreadPool”): 如果线程池已停止，抛出一个运行时错误异常。
  • tasks.emplace(task{ (*task)(); }): 将一个新的任务添加到任务队列中。这个任务是一个 lambda 函数，调用 task 对象的 operator() 来执行实际的任务。

通知一个等待的线程

condition.notify_one();
  • condition.notify_one(): 通知一个正在等待 condition 的线程，让它从等待中醒来，以便处理新添加的任务。

返回 future

return res;
  • return res: 返回先前获取的 std::future 对象，让调用者可以在稍后获取任务的结果。

为什么使用 std::future 的详细解释：

  1. 异步任务的结果获取
    当我们将一个任务提交到线程池时，任务是异步执行的。std::future 提供了一种机制，让我们可以在任务执行完成后，获取其结果，而不需要阻塞主线程或者轮询状态。
auto result = pool.enqueue([](int x) { return x * x; }, 10);
std::cout << result.get() << std::endl; // 获取任务结果
  1. 同步等待任务完成
    std::future 提供了 get() 方法，这个方法会阻塞调用它的线程，直到任务完成并返回结果。这样，我们可以在需要的时候等待任务完成并获取结果。
auto result = pool.enqueue([](int x) { return x * x; }, 10);
result.get(); // 阻塞等待任务完成并获取结果
  1. 异常传播
    如果任务在执行过程中抛出异常，std::future 会捕获这个异常，并在调用 get() 时重新抛出。这使得我们可以在调用 get() 时处理任务中的异常。
auto result = pool.enqueue([]() { throw std::runtime_error("Error"); });
try {
    result.get();
} catch (const std::exception& e) {
    std::cout << "Caught exception: " << e.what() << std::endl;
}
  1. 使用简单方便
    std::future 和 std::promise 以及 std::packaged_task 的组合使用，使得在多线程环境中管理任务和获取任务结果变得非常简单和方便。

参考资料
chatgpt
现代 C++ 编程实战

悦读

道可道，非常道；名可名，非常名。 无名，天地之始，有名，万物之母。 故常无欲，以观其妙，常有欲，以观其徼。 此两者，同出而异名，同谓之玄，玄之又玄，众妙之门。

最新收录
js根据已有数组，从数组对象中筛选数据
DNSSec：网络安全的守护者
【教程】Ollama 部署 MindSpore 训练的大模型
Visual Studio 2019安装使用教程-详细
docker 详解
Android开发之页面跳转传递list集合
Docker的基本使用
IDEA快捷键（Mac）
点云检测算法之PointNet深度解读
深度学习之基于Matlab的鸢尾花分类
;