前言
CyclicBarrier
要做的事情是,让一组线程到达一个屏障(也可以叫同步点)时被阻塞,直到最后一个线程到达屏障时,屏障才会开门,所有被屏障拦截的线程才会继续运行. 简单地说就是人到齐了后才可以让每个人继续去做自己的事情.
CycliBarrier
是通过ReentrantLock
和Condition
实现的一个数据结构.
本文代码: 代码下载
例子1
先通过一个简单的例子了解一下
CyclicBarrier
.
package com.sourcecode.concurrencytools_CyclicBarrier;
import java.util.concurrent.BrokenBarrierException;
import java.util.concurrent.CyclicBarrier;
public class CyclicBarrierTest4 {
static CyclicBarrier c = new CyclicBarrier(5);
public static void main(String[] args) throws InterruptedException, BrokenBarrierException {
for (int i = 0; i < 5; i++) {
Thread thread = new MyThread();
thread.start();
}
}
static class MyThread extends Thread {
@Override
public void run() {
try {
System.out.println(Thread.currentThread().getName() + " tries to wait!");
c.await();
} catch (Exception e) {
System.out.println(e);
//System.out.println(Thread.currentThread().getName() + "------>" + c.isBroken() + ", interrupted status:" + Thread.currentThread().isInterrupted());
} finally {
System.out.println(Thread.currentThread().getName() + " finishes!");
}
}
}
}
运行结果如下: 初始化
CyclicBarrier
的时候参数是5
,表示需要等待5
个线程达到后才可以打开屏障,正如结果所示,thread-0
到thread-3
在等待,到最后一个线程thread-4
到达屏障时,此时屏障打开,每个线程执行各自接下来的模块.
如果初始化参数大于
5
,比如6
,此程序将一直阻塞,因为没有第6
个线程到达该屏障.
Thread-0 tries to wait!
Thread-1 tries to wait!
Thread-2 tries to wait!
Thread-3 tries to wait!
Thread-4 tries to wait!
Thread-4 finishes!
Thread-1 finishes!
Thread-0 finishes!
Thread-2 finishes!
Thread-3 finishes!
实现思路分析
private static class Generation {
boolean broken = false;
}
/** 重入锁 */
private final ReentrantLock lock = new ReentrantLock();
/** 一个lock对象的Condition实例 */
private final Condition trip = lock.newCondition();
/** 拦截线程的总个数 */
private final int parties;
/** The command to run when tripped */
private final Runnable barrierCommand;
/** The current generation */
private Generation generation = new Generation();
/** 拦截线程的剩余需要数量 */
private int count;
从该图可以看出
CyclicBarrier
有一个重入锁的变量lock
并且持有一个该锁的Condition
实例trip
,就可以大概知道该CyclicBarrier
会让线程尝试获取锁并且在拿到锁后将屏障个数减减操作,然后根据count
的数量来决定是否调用trip.await()
操作,比如count==0
表示最后一个到达屏障的线程,那么就不需要调用trip
的方法了.
构造方法
public CyclicBarrier(int parties) {
this(parties, null);
}
public CyclicBarrier(int parties, Runnable barrierAction) {
if (parties <= 0) throw new IllegalArgumentException();
this.parties = parties;
this.count = parties;
this.barrierCommand = barrierAction;
}
第二个参数
Runnable barrierAction
表示的是当最后一个到达屏障的线程先执行完该barrierAction
的run
方法后再执行唤醒其他线程的操作.简单地说当到达屏障时,先执行barrierAction
的业务再执行其他线程的业务.
await方法
await
方法有两个,分别为await()
和await(long timeout, TimeUnit unit)
方法,一个没有超时返回,另外一个有超时返回,但是两者都是调用dowait(boolean timed, long nanos)
,该方法是整个CyclicBarrier
的核心实现.
public int await() throws InterruptedException, BrokenBarrierException {
try {
return dowait(false, 0L);
} catch (TimeoutException toe) {
throw new Error(toe); // cannot happen
}
}
public int await(long timeout, TimeUnit unit)
throws InterruptedException,
BrokenBarrierException,
TimeoutException {
return dowait(true, unit.toNanos(timeout));
}
所以接下来的看看该方法
dowait
是如何实现的.
/**
* @param timed 是否需要超时
* @param nanos 时长
* @return 返回还需要等待多少个线程才可以到达屏障
* @throws InterruptedException 当前线程中断
* @throws BrokenBarrierException 有其他线程中断或者其他线程超时
* @throws TimeoutException 当前线程等待超时
*/
private int dowait(boolean timed, long nanos)
throws InterruptedException, BrokenBarrierException,
TimeoutException {
// 获取重入锁
final ReentrantLock lock = this.lock;
// 尝试获取锁
lock.lock();
try {
//System.out.println(Thread.currentThread().getName() + " get locks.");
// 获得当前代
final Generation g = generation;
// 如果有线程中断或者超时
if (g.broken)
throw new BrokenBarrierException();
// 如果当前线程被中断
if (Thread.interrupted()) {
breakBarrier();
throw new InterruptedException();
}
int index = --count;
//System.out.format("index=%d\n", index);
if (index == 0) { // 最后一个到达屏障的线程
boolean ranAction = false;
try {
final Runnable command = barrierCommand;
if (command != null)
command.run();
ranAction = true;
nextGeneration(); //更新下一代
return 0;
} finally {
// 如果执行command.run发生异常,则breakBarrier
if (!ranAction)
breakBarrier();
}
}
// loop until tripped, broken, interrupted, or timed out
for (;;) {
try {
if (!timed)
trip.await();
else if (nanos > 0L)
nanos = trip.awaitNanos(nanos);
} catch (InterruptedException ie) {
// 如果等待过程中有被线程中断
if (g == generation && ! g.broken) {
breakBarrier();
throw ie;
} else {
// We're about to finish waiting even if we had not
// been interrupted, so this interrupt is deemed to
// "belong" to subsequent execution.
Thread.currentThread().interrupt();
}
}
// 如果当代的broken为true,表明有线程被中断
if (g.broken)
throw new BrokenBarrierException();
// 如果换代了 表示可以返回了
if (g != generation)
return index;
// 如果超时则先break the current generation
// 再抛出超时异常
if (timed && nanos <= 0L) {
breakBarrier();
throw new TimeoutException();
}
}
} finally {
// 释放锁
//System.out.println(Thread.currentThread().getName() + " release locks.");
lock.unlock();
}
}
/**
* break the current generation
* 1. broken设置为true
* 2. count 重新设置为parties
* 3. 唤醒所有线程
*/
private void breakBarrier() {
generation.broken = true;
count = parties;
trip.signalAll();
}
/**
* start a new generation
* 1. 唤醒所有等待中的线程
* 2. count 重新设置为parties
* 3. generation 设置成一个新的Generation对象
*/
private void nextGeneration() {
// signal completion of last generation
trip.signalAll();
// set up next generation
count = parties;
generation = new Generation();
}
该方法的流程大概如下:
1. 尝试获取锁
2. 如果不是最后一个到达屏障的线程,则进入for
循环中一直等待(此时该线程会释放锁)直到被最后一个线程唤醒或者被某个线程中断后调用breakBarrier
方法唤醒. 唤醒后需要竞争再次获得锁后才可以继续执行.
3. 如果是最后一个到达屏障的线程,如果barrierCommand
不为空,则需要先执行barrierCommand.run()
方法,然后通过nextGeneration
唤醒等待的线程.
4. 在所有异常退出或者正常退出都需要释放锁.
流程图如下
例子2
设置线程屏障为3,启动两个线程2秒超时等待,让最后一个线程3秒后才到达屏障.
package com.sourcecode.concurrencytools_CyclicBarrier;
import java.util.concurrent.TimeUnit;
public class CyclicBarrierTest5 {
static CyclicBarrier c = new CyclicBarrier(3);
public static void main(String[] args) throws InterruptedException, BrokenBarrierException {
for (int i = 0; i < 2; i++) {
Thread thread = new MyThread();
thread.start();
}
TimeUnit.SECONDS.sleep(3);
System.out.println(Thread.currentThread().getName() + "------>" + "tries to wait!");
c.await();
System.out.println(Thread.currentThread().getName() + "------>" + "finishes!");
}
static class MyThread extends Thread {
@Override
public void run() {
try {
System.out.println(Thread.currentThread().getName() + " tries to wait!");
c.await(1, TimeUnit.SECONDS);
//c.await();
} catch (Exception e) {
System.out.println(Thread.currentThread().getName() + "---->" + e);
//System.out.println(Thread.currentThread().getName() + "------>" + c.isBroken() + ", interrupted status:" + Thread.currentThread().isInterrupted());
} finally {
System.out.println(Thread.currentThread().getName() + " finishes!");
}
}
}
}
结果如下: 可以看到第一个线程出现超时异常后,表示该线程已经调用了
breakBarrier
方法,所以可以看到后续的两个线程都是抛出BrokenBarrierException
异常.
Thread-0 tries to wait!
Thread-1 tries to wait!
Thread-1---->java.util.concurrent.TimeoutException
Thread-0---->com.sourcecode.concurrencytools_CyclicBarrier.BrokenBarrierException
Thread-0 finishes!
Thread-1 finishes!
main------>tries to wait!
Exception in thread "main" com.sourcecode.concurrencytools_CyclicBarrier.BrokenBarrierException
at com.sourcecode.concurrencytools_CyclicBarrier.CyclicBarrier.dowait(CyclicBarrier.java:69)
at com.sourcecode.concurrencytools_CyclicBarrier.CyclicBarrier.await(CyclicBarrier.java:39)
at com.sourcecode.concurrencytools_CyclicBarrier.CyclicBarrierTest5.main(CyclicBarrierTest5.java:14)
isBroken方法和reset方法
/**
* @return 当前代是否被破坏, 被破坏的两种情况, 某个线程中断或者等待超时
*/
public boolean isBroken() {
final ReentrantLock lock = this.lock;
lock.lock();
try {
return generation.broken;
} finally {
lock.unlock();
}
}
public void reset() {
final ReentrantLock lock = this.lock;
lock.lock();
try {
breakBarrier(); // break the current generation
nextGeneration(); // start a new generation
} finally {
lock.unlock();
}
}
reset留作遇到好的例子后再分析
参考
1. Java并发编程的艺术