本文分析一下CountDownLatch是如何运用AQS的
CountDownLatch是什么
CountDownLatch顾名思义它是一个Latch（门闩），它是用一个计数器实现的，初始状态计数器的数值等于线程数，每当有线程完成任务后，计数器就会减一。当state为0时，锁就会被释放，凡是之前因抢占锁而等待的线程这时候就会被唤醒继续抢占锁。
CountDownLatch小栗子
public static void main(String[] args) throws InterruptedException{
int threadSize = 3;
CountDownLatch doneSignal = new CountDownLatch(threadSize);
for (int i = 1; i <= threadSize; i++) {
final int threadNum = i;
new Thread(() -> {
System.out.println("thread" + threadNum + ":start");
try {
Thread.sleep(1000 * threadNum);
} catch (InterruptedException e) {
System.out.println("thread" + threadNum + ":exception");
}
doneSignal.countDown();
System.out.println("thread" + threadNum + ":complete");
}).start();
}
System.out.println("main thread:await");
doneSignal.await();
System.out.println("main thread:go on");
}
例子中主线程启动了三条子线程，睡眠一段时间，此时主线程在等待所有子线程结束后才会继续执行下去；
看一下输出结果：
main thread:await
thread1:start
thread2:start
thread3:start
thread1:complete
thread2:complete
thread3:complete
main thread:go on
Process finished with exit code 0
CountDownLatch原理分析
既然CountDownLatch也是AQS的一种使用方式，我们看一下它的内部类Syc是怎么实现AQS的：
private static final class Sync extends AbstractQueuedSynchronizer {
private static final long serialVersionUID = 4982264981922014374L;
//构造函数，初始化同步状态state的值，即线程个数
Sync(int count) {
setState(count);
}
int getCount() {
return getState();
}
//这里重写了方法，在共享模式下，告诉调用者是否可以抢占state锁了，正数代表可以，负数代表否定；当state为0时返回正数
protected int tryAcquireShared(int acquires) {
return (getState() == 0) ? 1 : -1;
}
//共享模式下释放锁
protected boolean tryReleaseShared(int releases) {
// Decrement count; signal when transition to zero
for (;;) {
int c = getState();
//state为0时说明没有什么可释放
if (c == 0)
return false;
int nextc = c-1;
if (compareAndSetState(c, nextc))
//CAS对state操作成功后返回state值是否为0，为0则释放成功
return nextc == 0;
}
}
}
看完了重写的AQS同步器后，我们了解了CountDownLatch对state锁的描述。接下来先看主线程调用的await方法，在await方法里调用了AQS的acquireSharedInterruptibly：
//在共享模式下尝试抢占锁
public final void acquireSharedInterruptibly(int arg)
throws InterruptedException {
//线程中断抛出异常
if (Thread.interrupted())
throw new InterruptedException();
//尝试抢占前先查询一下是否可以抢占，如果返回值大于0程序往下执行，小于0则等待
if (tryAcquireShared(arg) < 0)
doAcquireSharedInterruptibly(arg);
}
private void doAcquireSharedInterruptibly(int arg)
throws InterruptedException {
//在Reentrant解析中我们看过，往队列中新增node(共享模式)
final Node node = addWaiter(Node.SHARED);
boolean failed = true;
try {
for (;;) {
final Node p = node.predecessor();
if (p == head) {
//如果当前node的前继时head，马上尝试抢占锁
int r = tryAcquireShared(arg);
if (r >= 0) {
//如果state==0即允许往下执行，重新设置head并往下传播信号
setHeadAndPropagate(node, r);
p.next = null; // help GC
failed = false;
//得到往下执行的允许
return;
}
}
//以下都跟Reentrant一样
if (shouldParkAfterFailedAcquire(p, node) &&
parkAndCheckInterrupt())
throw new InterruptedException();
}
} finally {
if (failed)
cancelAcquire(node);
}
}
private void setHeadAndPropagate(Node node, int propagate) {
Node h = head; // Record old head for check below
//将当前node设置为head，清空node的thread、prev
setHead(node);
/*
* Try to signal next queued node if:
* Propagation was indicated by caller,
* or was recorded (as h.waitStatus either before
* or after setHead) by a previous operation
* (note: this uses sign-check of waitStatus because
* PROPAGATE status may transition to SIGNAL.)
* and
* The next node is waiting in shared mode,
* or we don't know, because it appears null
*
* The conservatism in both of these checks may cause
* unnecessary wake-ups, but only when there are multiple
* racing acquires/releases, so most need signals now or soon
* anyway.
*/
//如果propagate大于0，或者原来head的等待状态小于0或者现在head的等待状态小于0
if (propagate > 0 || h == null || h.waitStatus < 0 ||
(h = head) == null || h.waitStatus < 0) {
Node s = node.next;
//准备唤醒下一个节点
if (s == null || s.isShared())
doReleaseShared();
}
}
private void doReleaseShared() {
/*
* Ensure that a release propagates, even if there are other
* in-progress acquires/releases. This proceeds in the usual
* way of trying to unparkSuccessor of head if it needs
* signal. But if it does not, status is set to PROPAGATE to
* ensure that upon release, propagation continues.
* Additionally, we must loop in case a new node is added
* while we are doing this. Also, unlike other uses of
* unparkSuccessor, we need to know if CAS to reset status
* fails, if so rechecking.
*/
for (;;) {
Node h = head;
if (h != null && h != tail) {
int ws = h.waitStatus;
if (ws == Node.SIGNAL) {
//如果head的状态为SIGNAL，更改状态为0
if (!compareAndSetWaitStatus(h, Node.SIGNAL, 0))
continue; // loop to recheck cases
//唤醒后继节点
unparkSuccessor(h);
}
//如果head状态为0，更改状态为PROPAGATE
else if (ws == 0 &&
!compareAndSetWaitStatus(h, 0, Node.PROPAGATE))
continue; // loop on failed CAS
}
//如果head没有改变，结束当前loop，如果遇到head被别的线程改变，继续loop
if (h == head) // loop if head changed
break;
}
}
释放锁的信号一直向后传播，直到所有node被唤醒并继续执行，那第一个信号时何时发起的呢？我们来看一下CountDownLatch的countDown方法，该方法调用了sync的releaseShared方法：
public final boolean releaseShared(int arg) {
if (tryReleaseShared(arg)) {
//如果同步状态state为0时，调用doReleaseShared，在这里就发出了第一个唤醒所有等待node的信号，然后信号自动往后传播
doReleaseShared();
return true;
}
return false;
}
总结
CountDownLatch在调用await的时候判断state释放为0，如果大于0则阻塞当前线程，将当前线程的node添加到队列中等待；在调用countDown时当遇到state减到0时，发出释放共享锁的信号，从头节点的后记节点开始往后传递信号，将队列等待的线程逐个唤醒并继续往下执行；
在这里state跟Reentrant的state独占锁含义不同，state的含义是由AQS的子类去描述的。
