java并发中CountDownLatch的使用
在java并发中，控制共享变量的访问非常重要，有时候我们也想控制并发线程的执行顺序，比如：等待所有线程都执行完毕之后再执行另外的线程，或者等所有线程都准备好了才开始所有线程的执行等。
这个时候我们就可以使用到CountDownLatch。
简单点讲，CountDownLatch存有一个放在QueuedSynchronizer中的计数器。当调用countdown() 方法时，该计数器将会减一。然后再调用await()来等待计数器归零。
private static final class Sync extends AbstractQueuedSynchronizer {
...
}
private final Sync sync;
public void countDown() {
sync.releaseShared(1);
}
public void await() throws InterruptedException {
sync.acquireSharedInterruptibly(1);
}
public boolean await(long timeout, TimeUnit unit)
throws InterruptedException {
return sync.tryAcquireSharedNanos(1, unit.toNanos(timeout));
}
下面我们举两个使用的例子:
主线程等待子线程全都结束之后再开始运行
这里我们定义子线程类，在子线程类里面，我们传入一个CountDownLatch用来计数，然后在子线程结束之前，调用该CountDownLatch的countDown方法。最后在主线程中调用await（）方法来等待子线程结束执行。
@Slf4j
public class MainThreadWaitUsage implements Runnable {
private List<String> outputScraper;
private CountDownLatch countDownLatch;
public MainThreadWaitUsage(List<String> outputScraper, CountDownLatch countDownLatch) {
this.outputScraper = outputScraper;
this.countDownLatch = countDownLatch;
}
@Override
public void run() {
outputScraper.add("Counted down");
countDownLatch.countDown();
}
}
看下怎么调用：
@Test
public void testCountDownLatch()
throws InterruptedException {
List<String> outputScraper = Collections.synchronizedList(new ArrayList<>());
CountDownLatch countDownLatch = new CountDownLatch(5);
List<Thread> workers = Stream
.generate(() -> new Thread(new MainThreadWaitUsage(outputScraper, countDownLatch)))
.limit(5)
.collect(toList());
workers.forEach(Thread::start);
countDownLatch.await();
outputScraper.add("Latch released");
log.info(outputScraper.toString());
}
执行结果如下：
07:37:27.388 [main] INFO MainThreadWaitUsageTest - [Counted down, Counted down, Counted down, Counted down, Counted down, Latch released]
等待所有线程都准备好再一起执行
上面的例子中，我们是主线程等待子线程，那么在这个例子中，我们将会看看怎么子线程一起等待到准备好的状态，再一起执行。
思路也很简单，在子线程开始之后，将等待的子线程计数器减一，在主线程中await该计数器，等计数器归零之后，主线程再通知子线程运行。
public class ThreadWaitThreadUsage implements Runnable {
private List<String> outputScraper;
private CountDownLatch readyThreadCounter;
private CountDownLatch callingThreadBlocker;
private CountDownLatch completedThreadCounter;
public ThreadWaitThreadUsage(
List<String> outputScraper,
CountDownLatch readyThreadCounter,
CountDownLatch callingThreadBlocker,
CountDownLatch completedThreadCounter) {
this.outputScraper = outputScraper;
this.readyThreadCounter = readyThreadCounter;
this.callingThreadBlocker = callingThreadBlocker;
this.completedThreadCounter = completedThreadCounter;
}
@Override
public void run() {
readyThreadCounter.countDown();
try {
callingThreadBlocker.await();
outputScraper.add("Counted down");
} catch (InterruptedException e) {
e.printStackTrace();
} finally {
completedThreadCounter.countDown();
}
}
}
看下怎么调用：
@Test
public void testCountDownLatch()
throws InterruptedException {
List<String> outputScraper = Collections.synchronizedList(new ArrayList<>());
CountDownLatch readyThreadCounter = new CountDownLatch(5);
CountDownLatch callingThreadBlocker = new CountDownLatch(1);
CountDownLatch completedThreadCounter = new CountDownLatch(5);
List<Thread> workers = Stream
.generate(() -> new Thread(new ThreadWaitThreadUsage(
outputScraper, readyThreadCounter, callingThreadBlocker, completedThreadCounter)))
.limit(5)
.collect(toList());
workers.forEach(Thread::start);
readyThreadCounter.await();
outputScraper.add("Workers ready");
callingThreadBlocker.countDown();
completedThreadCounter.await();
outputScraper.add("Workers complete");
log.info(outputScraper.toString());
}
输出结果如下：
07:41:47.861 [main] INFO ThreadWaitThreadUsageTest - [Workers ready, Counted down, Counted down, Counted down, Counted down, Counted down, Workers complete]
停止CountdownLatch的await
如果我们调用await（）方法，该方法将会等待一直到count=0才结束。但是如果在线程执行过程中出现了异常，可能导致countdown方法执行不了。那么await（）方法可能会出现无限等待的情况。
这个时候我们可以使用：
public boolean await(long timeout, TimeUnit unit)
throws InterruptedException {
return sync.tryAcquireSharedNanos(1, unit.toNanos(timeout));
}
本文的例子可以参考https://github.com/ddean2009/learn-java-concurrency/tree/master/CountDownLatch
更多教程请参考 flydean的博客
