最简单的生产者-消费者模式实现

2026-10-17T00:00:00+00:00

生产者-消费者模式</h2>
这是多线程开发中一种常见的设计模式，用于解决数据生产和消费速度不一致的问题。</p>
在生产者和消费者二者之间增加一层缓冲区，生产者只需要关心缓冲区是否还能存放数据，不需要等待消费者处理完；同样消费者只关注缓冲区是否还存有数据。</p>
也就是说，这样的模型可以带来最显著的优势是：生产者和消费者完全解耦合，可以独立开发和扩展。如你可以增加并发数、支持分布式开发等。</p>

如何实现？</h2>

分三部分考虑：</p>

对于缓冲区：先生产的数据要被先消费，自然想到缓冲区用队列结构实现。缓冲区有大小限制，你不能在满了之后继续生产数据。当然也不能在空的时候消费数据。</p> </li>
对于生产者：在往缓冲区存放数据前，判断是否已满。当缓冲区已满时，使用条件变量等待数据被拿取，直到能够继续存放数据。存完数据后通知消费者。</p> </li>

对于消费者：在拿去缓冲区数据时，判断是否为空。当缓冲区为空时，使用条件变量等到新数据存放，直到能够拿取数据。拿完后通知生产者。</p> </li> </ul>

代码实现</h2>

假定缓冲区存放 int 型数据，缓冲区大小为 5 。</p>

#include</span> <iostream></span></span>
#include</span> <mutex></span></span>
#include</span> <condition_variable></span></span>
#include</span> <thread></span></span>
</span>
std</span>::</span>mutex mtx</span>;</span></span>
std</span>::</span>condition_variable cv</span>;</span></span>
std</span>::</span>queue</span><</span>int</span>></span> que</span>;</span></span>
const int</span> MAX_SIZE </span>=</span> 5</span>;</span></span>
</span>
void</span> producer</span>() {</span></span>
  for</span> (</span>int</span> i </span>=</span> 0</span>;</span> i </span><</span> 10</span>;</span> ++</span>i</span>) {</span></span>
    std</span>::</span>unique_lock</span><</span>std</span>::</span>mutex</span>></span> lock</span>(</span>mtx</span>);</span></span>
    cv</span>.</span>wait</span>(</span>lock</span>, [] {</span> return</span> que</span>.</span>size</span>()</span> <</span> MAX_SIZE</span>; });</span></span>
    que</span>.</span>push</span>(</span>i</span>);</span></span>
    std</span>::</span>cout </span><<</span> "produce "</span> <<</span> i </span><<</span> "/"</span> <<</span> que</span>.</span>size</span>()</span> <<</span> std</span>::</span>endl</span>;</span></span>
    cv</span>.</span>notify_all</span>();</span></span>
    std</span>::</span>this_thread</span>::</span>sleep_for</span>(</span>std</span>::</span>chrono</span>::</span>milliseconds</span>(</span>100</span>));</span></span>
  }</span></span>
}</span></span>
</span>
void</span> consumer</span>() {</span></span>
  for</span> (</span>int</span> i </span>=</span> 0</span>;</span> i </span><</span> 10</span>;</span> ++</span>i</span>) {</span></span>
    std</span>::</span>unique_lock</span><</span>std</span>::</span>mutex</span>></span> lock</span>(</span>mtx</span>);</span></span>
    cv</span>.</span>wait</span>(</span>lock</span>, [] {</span> return</span> !</span>que</span>.</span>empty</span>(); });</span></span>
    int</span> val </span>=</span> que</span>.</span>front</span>();</span></span>
    que</span>.</span>pop</span>();</span></span>
    std</span>::</span>cout </span><<</span> "consume "</span> <<</span> i </span><<</span> "/"</span> <<</span> que</span>.</span>size</span>()</span> <<</span> std</span>::</span>endl</span>;</span></span>
    cv</span>.</span>notify_all</span>();</span></span>
    std</span>::</span>this_thread</span>::</span>sleep_for</span>(</span>std</span>::</span>chrono</span>::</span>milliseconds</span>(</span>100</span>));</span></span>
  }</span></span>
}</span></span>
</span>
int</span> main</span>() {</span></span>
  std</span>::</span>thread</span> p</span>(</span>producer</span>);</span></span>
  std</span>::</span>thread</span> c</span>(</span>consumer</span>);</span></span>
  p</span>.</span>join</span>();</span></span>
  c</span>.</span>join</span>();</span></span>
  return</span> 0</span>;</span></span>
}</span></span>
</span></code></pre>
下次面试官让我使用 C++ 实现生产者-消费者模型，可以直接照着这个写，而不是陷入尴尬的境地 :<</p>
参考</h2>
生产者-消费者模型：理论讲解及实现（C++）</a></p>

shaozk - thread

最简单的生产者-消费者模式实现

参考</h2> 生产者-消费者模型：理论讲解及实现（C++）</a></p>

参考</h2>
生产者-消费者模型：理论讲解及实现（C++）</a></p>