深入理解CountDownLatch原理
一、简介
CountDownLatch是一个同步工具类，它允许一个或多个线程一直等待，直到其他线程执行完后再执行。例如，应用程序的主线程希望在负责启动框架服务的线程已经启动所有框架服务之后执行。CountDownLatch是在java1.5被引入的，跟它一起被引入的并发工具类还有CyclicBarrier、Semaphore、ConcurrentHashMap和BlockingQueue等，它们都存在于java.util.concurrent包下。
二、源码
函数列表
//构造一个用给定计数初始化的 CountDownLatch。
CountDownLatch(int count)
// 使当前线程在锁存器倒计数至零之前一直等待，除非线程被中断。
void await()
// 使当前线程在锁存器倒计数至零之前一直等待，除非线程被中断或超出了指定的等待时间。
boolean await(long timeout, TimeUnit unit)
// 递减锁存器的计数，如果计数到达零，则释放所有等待的线程。
void countDown()
// 返回当前计数。
long getCount()
// 返回标识此锁存器及其状态的字符串。
String toString()数据结构
CountDownLatch的UML类图如下：
CountDownLatch的数据结构很简单，它是通过"共享锁"实现的。它包含了sync对象，sync是Sync类型。Sync是实例类，它继承于AQS。
package java.util.concurrent;
import java.util.concurrent.locks.AbstractQueuedSynchronizer;
public class CountDownLatch {
/**
* Synchronization control For CountDownLatch.
* Uses AQS state to represent count.
*/
private static final class Sync extends AbstractQueuedSynchronizer {
private static final long serialVersionUID = 4982264981922014374L;
Sync(int count) {
setState(count);
}
int getCount() {
return getState();
}
protected int tryAcquireShared(int acquires) {
return (getState() == 0) ? 1 : -1;
}
protected boolean tryReleaseShared(int releases) {
// Decrement count; signal when transition to zero
for (;;) {
int c = getState();
if (c == 0)
return false;
int nextc = c-1;
if (compareAndSetState(c, nextc))
return nextc == 0;
}
}
}
private final Sync sync;
/**
* Constructs a {@code CountDownLatch} initialized with the given count.
*
* @param count the number of times {@link #countDown} must be invoked
* before threads can pass through {@link #await}
* @throws IllegalArgumentException if {@code count} is negative
*/
public CountDownLatch(int count) {
if (count < 0) throw new IllegalArgumentException("count < 0");
this.sync = new Sync(count);
}
/**
* Causes the current thread to wait until the latch has counted down to
* zero, unless the thread is {@linkplain Thread#interrupt interrupted}.
*
* <p>If the current count is zero then this method returns immediately.
*
* <p>If the current count is greater than zero then the current
* thread becomes disabled for thread scheduling purposes and lies
* dormant until one of two things happen:
* <ul>
* <li>The count reaches zero due to invocations of the
* {@link #countDown} method; or
* <li>Some other thread {@linkplain Thread#interrupt interrupts}
* the current thread.
* </ul>
*
* <p>If the current thread:
* <ul>
* <li>has its interrupted status set on entry to this method; or
* <li>is {@linkplain Thread#interrupt interrupted} while waiting,
* </ul>
* then {@link InterruptedException} is thrown and the current thread's
* interrupted status is cleared.
*
* @throws InterruptedException if the current thread is interrupted
* while waiting
*/
public void await() throws InterruptedException {
sync.acquireSharedInterruptibly(1);
}
/**
* Causes the current thread to wait until the latch has counted down to
* zero, unless the thread is {@linkplain Thread#interrupt interrupted},
* or the specified waiting time elapses.
*
* <p>If the current count is zero then this method returns immediately
* with the value {@code true}.
*
* <p>If the current count is greater than zero then the current
* thread becomes disabled for thread scheduling purposes and lies
* dormant until one of three things happen:
* <ul>
* <li>The count reaches zero due to invocations of the
* {@link #countDown} method; or
* <li>Some other thread {@linkplain Thread#interrupt interrupts}
* the current thread; or
* <li>The specified waiting time elapses.
* </ul>
*
* <p>If the count reaches zero then the method returns with the
* value {@code true}.
*
* <p>If the current thread:
* <ul>
* <li>has its interrupted status set on entry to this method; or
* <li>is {@linkplain Thread#interrupt interrupted} while waiting,
* </ul>
* then {@link InterruptedException} is thrown and the current thread's
* interrupted status is cleared.
*
* <p>If the specified waiting time elapses then the value {@code false}
* is returned. If the time is less than or equal to zero, the method
* will not wait at all.
*
* @param timeout the maximum time to wait
* @param unit the time unit of the {@code timeout} argument
* @return {@code true} if the count reached zero and {@code false}
* if the waiting time elapsed before the count reached zero
* @throws InterruptedException if the current thread is interrupted
* while waiting
*/
public boolean await(long timeout, TimeUnit unit)
throws InterruptedException {
return sync.tryAcquireSharedNanos(1, unit.toNanos(timeout));
}
/**
* Decrements the count of the latch, releasing all waiting threads if
* the count reaches zero.
*
* <p>If the current count is greater than zero then it is decremented.
* If the new count is zero then all waiting threads are re-enabled for
* thread scheduling purposes.
*
* <p>If the current count equals zero then nothing happens.
*/
public void countDown() {
sync.releaseShared(1);
}
/**
* Returns the current count.
*
* <p>This method is typically used for debugging and testing purposes.
*
* @return the current count
*/
public long getCount() {
return sync.getCount();
}
/**
* Returns a string identifying this latch, as well as its state.
* The state, in brackets, includes the String {@code "Count ="}
* followed by the current count.
*
* @return a string identifying this latch, as well as its state
*/
public String toString() {
return super.toString() + "[Count = " + sync.getCount() + "]";
}
}CountDownLatch是通过“共享锁”实现的。下面，我们分析CountDownLatch中3个核心函数: CountDownLatch(int count), await(), countDown()
构造函数
public CountDownLatch(int count) {
if (count < 0) throw new IllegalArgumentException("count < 0");
this.sync = new Sync(count);
}说明：该函数是创建一个Sync对象，而Sync是继承于AQS类。Sync构造函数如下：
Sync(int count) {
setState(count);
}setState()在AQS中实现，源码如下：
protected final void setState(long newState) {
state = newState;
} 说明：在AQS中，state是一个private volatile long类型的对象。对于CountDownLatch而言，state表示的”锁计数器“。CountDownLatch中的getCount()最终是调用AQS中的getState()，返回的state对象，即”锁计数器“。
await()函数
public void await() throws InterruptedException {
sync.acquireSharedInterruptibly(1);
}
//说明：该函数实际上是调用的AQS的acquireSharedInterruptibly(1);
public final void acquireSharedInterruptibly(long arg)
throws InterruptedException {
if (Thread.interrupted())
throw new InterruptedException();
if (tryAcquireShared(arg) < 0)
doAcquireSharedInterruptibly(arg);
}说明：acquireSharedInterruptibly()的作用是获取共享锁。
如果当前线程是中断状态，则抛出异常InterruptedException。否则，调用tryAcquireShared(arg)尝试获取共享锁；尝试成功则返回，否则就调用doAcquireSharedInterruptibly()。doAcquireSharedInterruptibly()会使当前线程一直等待，直到当前线程获取到共享锁(或被中断)才返回。
tryAcquireShared()在CountDownLatch.java中被重写，它的源码如下：
protected int tryAcquireShared(int acquires) {
return (getState() == 0) ? 1 : -1;
}说明：tryAcquireShared()的作用是尝试获取共享锁。
如果"锁计数器=0"，即锁是可获取状态，则返回1；否则，锁是不可获取状态，则返回-1。
private void doAcquireSharedInterruptibly(long arg)
throws InterruptedException {
// 创建"当前线程"的Node节点，且Node中记录的锁是"共享锁"类型；并将该节点添加到CLH队列末尾。
final Node node = addWaiter(Node.SHARED);
boolean failed = true;
try {
for (;;) {
// 获取上一个节点。
// 如果上一节点是CLH队列的表头，则"尝试获取共享锁"。
final Node p = node.predecessor();
if (p == head) {
int r = tryAcquireShared(arg);
if (r >= 0) {
setHeadAndPropagate(node, r);
p.next = null; // help GC
failed = false;
return;
}
}
// (上一节点不是CLH队列的表头) 当前线程一直等待，直到获取到共享锁。
// 如果线程在等待过程中被中断过，则再次中断该线程(还原之前的中断状态)。
if (shouldParkAfterFailedAcquire(p, node) &&
parkAndCheckInterrupt())
throw new InterruptedException();
}
} finally {
if (failed)
cancelAcquire(node);
}
}说明：
a. addWaiter(Node.SHARED)的作用是，创建”当前线程“的Node节点，且Node中记录的锁的类型是”共享锁“(Node.SHARED)；并将该节点添加到CLH队列末尾。关于Node和CLH在深入理解AQS（AbstractQueuedSynchronizer）已经详细介绍过，这里就不再重复说明了。
b. node.predecessor()的作用是，获取上一个节点。如果上一节点是CLH队列的表头，则”尝试获取共享锁“。
c. shouldParkAfterFailedAcquire()的作用和它的名称一样，如果在尝试获取锁失败之后，线程应该等待，则返回true；否则，返回false。
d. 当shouldParkAfterFailedAcquire()返回ture时，则调用parkAndCheckInterrupt()，当前线程会进入等待状态，直到获取到共享锁才继续运行。
doAcquireSharedInterruptibly()中的shouldParkAfterFailedAcquire(), parkAndCheckInterrupt等函数在深入理解AQS（AbstractQueuedSynchronizer）中介绍过，这里也就不再详细说明了。
再简述：await()方法是怎么“阻塞”当前线程的，已经非常明白了。其实说白了，就是当你调用了countDownLatch.await()方法后，你当前线程就会进入了一个死循环当中，在这个死循环里面，会不断的进行判断，通过调用tryAcquireShared方法，不断判断我们上面说的那个计数器，看看它的值是否为0了（为0的时候，其实就是我们调用了足够多次数的countDownLatch.countDown（）方法的时候），如果是为0的话，tryAcquireShared就会返回1，然后跳出了循环，也就不再“阻塞”当前线程了。
countDown()函数
public void countDown() {
sync.releaseShared(1);
}
public final boolean releaseShared(int arg) {
// 只有当 state 减为 0 的时候，tryReleaseShared 才返回 true
// 否则只是简单的 state = state - 1 那么 countDown 方法就结束了
if (tryReleaseShared(arg)) {
// 唤醒 await 的线程
doReleaseShared();
return true;
}
return false;
}
// 这个方法很简单，用自旋的方法实现 state 减 1
protected boolean tryReleaseShared(int releases) {
for (;;) {
int c = getState();
if (c == 0)
return false;
int nextc = c-1;
//通过CAS将state的值减1，失败就不会进入return,继续for循环，直至CAS成功
if (compareAndSetState(c, nextc))
//state减到0就返回true,否则返回false
return nextc == 0;
}
}countDown 方法就是每次调用都将 state 值减 1，如果 state 减到 0 了，那么就调用下面的方法进行唤醒阻塞队列中的线程：
// 调用这个方法的时候，state == 0
private void doReleaseShared() {
for (;;) {
Node h = head;
if (h != null && h != tail) {
int ws = h.waitStatus;
if (ws == Node.SIGNAL) {
if (!compareAndSetWaitStatus(h, Node.SIGNAL, 0))
continue; // loop to recheck cases
// 就是这里，唤醒 head 的后继节点，也就是阻塞队列中的第一个节点
// 在这里，也就是唤醒线程 ，可以接着运行了
unparkSuccessor(h);
}
else if (ws == 0 &&
!compareAndSetWaitStatus(h, 0, Node.PROPAGATE))
continue; // loop on failed CAS
}
if (h == head) // loop if head changed
break;
}
}我们继续回到 await 的这段代码，在第24行代码 parkAndCheckInterrupt 返回继续接着运行，我们先不考虑中断的情况：
private void doAcquireSharedInterruptibly(int arg)
throws InterruptedException {
final Node node = addWaiter(Node.SHARED);
boolean failed = true;
try {
for (;;) {
//p表示当前节点的前驱节点
final Node p = node.predecessor();
//此时被唤醒的是之前head的后继节点，所以此线程的前驱节点是head
if (p == head) {
//此时state已经为0，r为1
int r = tryAcquireShared(arg);
if (r >= 0) {
// 2. 这里将唤醒的后续节点,以此类推，后续节点被唤醒后，会在它的await中唤醒t4的后续节点
setHeadAndPropagate(node, r);
// 将已经唤醒的t3节点从队列中去除
p.next = null; // help GC
failed = false;
return;
}
}
if (shouldParkAfterFailedAcquire(p, node) &&
// 1. 唤醒后这个方法返回
parkAndCheckInterrupt())
throw new InterruptedException();
}
} finally {
if (failed)
cancelAcquire(node);
}
}会循环一次进到 setHeadAndPropagate(node, r) 这个方法，先把 head 给占了，然后唤醒队列中其他的线程：
private void setHeadAndPropagate(Node node, int propagate) {
Node h = head; // Record old head for check below
setHead(node);
// 下面说的是，唤醒当前 node 之后的节点，即当前后续节点已经醒了，马上唤醒之后后续节点
// 类似的，如果后面还有节点，那么醒了以后，马上将后面的节点给唤醒了
if (propagate > 0 || h == null || h.waitStatus < 0 ||
(h = head) == null || h.waitStatus < 0) {
Node s = node.next;
if (s == null || s.isShared())
// 又是这个方法，只是现在的 head 已经不是原来的空节点了，是当前的节点了
doReleaseShared();
}
}又回到这个方法了，那么接下来，我们好好分析 doReleaseShared 这个方法，我们根据流程
// 调用这个方法的时候，state == 0
private void doReleaseShared() {
for (;;) {
Node h = head;
if (h != null && h != tail) {
int ws = h.waitStatus;
// 将头节点的 waitStatus 设置为 Node.SIGNAL（-1） 了
if (ws == Node.SIGNAL) {
if (!compareAndSetWaitStatus(h, Node.SIGNAL, 0))
continue; // loop to recheck cases
// 就是这里，唤醒 head 的后继节点，也就是阻塞队列中的第一个节点
// 在这里，也就是唤醒后续节点
unparkSuccessor(h);
}
else if (ws == 0 &&
// 这个 CAS 失败的场景是：执行到这里的时候，刚好有一个节点入队，入队会将这个 ws 设置为 -1
!compareAndSetWaitStatus(h, 0, Node.PROPAGATE))
continue; // loop on failed CAS
}
// 如果到这里的时候，前面唤醒的线程已经占领了 head，那么再循环
// 否则，就是 head 没变，那么退出循环，
// 退出循环是不是意味着阻塞队列中的其他节点就不唤醒了？当然不是，唤醒的线程之后还是会在await()方法中调用此方法接着唤醒后续节点
if (h == head) // loop if head changed
break;
}
}总结：CountDownLatch是通过“共享锁”实现的。在创建CountDownLatch中时，会传递一个int类型参数count，该参数是“锁计数器”的初始状态，表示该“共享锁”最多能被count给线程同时获取。当某线程调用该CountDownLatch对象的await()方法时，该线程会等待“共享锁”可用时，才能获取“共享锁”进而继续运行。而“共享锁”可用的条件，就是“锁计数器”的值为0！而“锁计数器”的初始值为count，每当一个线程调用该CountDownLatch对象的countDown()方法时，才将“锁计数器”-1；通过这种方式，必须有count个线程调用countDown()之后，“锁计数器”才为0，而前面提到的等待线程才能继续运行！以上，就是CountDownLatch的实现原理。
三、示例
import java.util.Random;
import java.util.concurrent.CountDownLatch;
public class Test {
private static final int THREAD_COUNT_NUM = 7;
private static CountDownLatch countDownLatch = new CountDownLatch(THREAD_COUNT_NUM);
public static void main(String[] args) throws InterruptedException {
for (int i = 1; i <= THREAD_COUNT_NUM; i++) {
int index = i;
new Thread(() -> {
try {
System.out.println("第" + index + "颗龙珠已收集到！");
//模拟收集第i个龙珠,随机模拟不同的寻找时间
Thread.sleep(new Random().nextInt(3000));
} catch (InterruptedException e) {
e.printStackTrace();
}
//每收集到一颗龙珠,需要等待的颗数减1
countDownLatch.countDown();
}).start();
}
//等待检查，即上述7个线程执行完毕之后，执行await后边的代码
countDownLatch.await();
System.out.println("集齐七颗龙珠！召唤神龙！");
}
}
CountDownLatch和CyclicBarrier的区别
(01) CountDownLatch的作用是允许1或N个线程等待其他线程完成执行；而CyclicBarrier则是允许N个线程相互等待。
(02) CountDownLatch的计数器无法被重置；CyclicBarrier的计数器可以被重置后使用，因此它被称为是循环的barrier。
