ReentrantLock是如何实现可重入性

爱被打了一巴掌 2024-03-16 12:05 42阅读 0赞

**1** **问题**

ReentrantLock是如何实现可重入性的？ 并且在进入锁内之前进行了哪些判断？

**2** **方法**

<table> 
 <tbody> 
 <tr> 
 <td>// Sync继承于AQS abstract static class Sync extends AbstractQueuedSynchronizer { &nbsp;... } // ReentrantLock默认是非公平锁 public ReentrantLock() { &nbsp; &nbsp; &nbsp; &nbsp;sync = new NonfairSync(); } // 可以通过向构造方法中传true来实现公平锁 public ReentrantLock(boolean fair) { &nbsp; &nbsp;sync = fair ? new FairSync() : new NonfairSync(); } protected final boolean tryAcquire(int acquires) { &nbsp; &nbsp; &nbsp; &nbsp;// 当前想要获取锁的线程 &nbsp; &nbsp; &nbsp; &nbsp;final Thread current = Thread.currentThread(); &nbsp; &nbsp; &nbsp; &nbsp;// 当前锁的状态 &nbsp; &nbsp; &nbsp; &nbsp;int c = getState(); &nbsp; &nbsp; &nbsp; &nbsp;// state == 0 此时此刻没有线程持有锁 &nbsp; &nbsp; &nbsp; &nbsp;if (c == 0) { &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp;// 虽然此时此刻锁是可以用的，但是这是公平锁，既然是公平，就得讲究先来后到， &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp;// 看看有没有别人在队列中等了半天了 &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp;if (!hasQueuedPredecessors() &amp;&amp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp;// 如果没有线程在等待，那就用CAS尝试一下，成功了就获取到锁了， &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp;// 不成功的话，只能说明一个问题，就在刚刚几乎同一时刻有个线程抢先了 =_= &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp;// 因为刚刚还没人的，我判断过了 &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp;compareAndSetState(0, acquires)) { &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp;// 到这里就是获取到锁了，标记一下，告诉大家，现在是我占用了锁 &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp;setExclusiveOwnerThread(current); &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp;return true; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp;} &nbsp; &nbsp; &nbsp; &nbsp;} &nbsp; &nbsp; &nbsp; &nbsp; &nbsp;// 会进入这个else if分支，说明是重入了，需要操作：state=state+1 &nbsp; &nbsp; &nbsp; &nbsp;// 这里不存在并发问题 &nbsp; &nbsp; &nbsp; &nbsp;else if (current == getExclusiveOwnerThread()) { &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp;int nextc = c + acquires; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp;if (nextc &lt; 0) &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp;throw new Error("Maximum lock count exceeded"); &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp;setState(nextc); &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp;return true; &nbsp; &nbsp; &nbsp; &nbsp;} &nbsp; &nbsp; &nbsp; &nbsp;// 如果到这里，说明前面的if和else if都没有返回true，说明没有获取到锁 &nbsp; &nbsp; &nbsp; &nbsp;return false; &nbsp; &nbsp;}</td> 
 </tr> 
 </tbody> 
</table>

**3** **结语**

ReentrantLock使用内部类Sync来管理锁，所以真正的获取锁是由Sync的实现类控制的。在进行前先判断锁内是否有数据，没有锁就选择Sync继承AQS来实现进入锁内；如果锁内有数据则先判断数据是否是自己的，是自己的数据就将state的值+1，表示重入。