关于使用CyclicBarrier使主线程等待子线程执行完之后再向下执行的问题

CyclicBarrier
线程障碍，其实就是为线程制作一个集合点，相关知识网上一堆。

这次主要记录一下使用CyclicBarrier遇到的一点小问题。

需求：在主线程中创建两个子线程，希望能在子线程执行完成之后再执行主线程中的剩余代码。

代码一：

package com.iteye.wwwcomy.thread;

import java.util.concurrent.BrokenBarrierException;
import java.util.concurrent.CyclicBarrier;
import java.util.concurrent.ExecutorService;
import java.util.concurrent.Executors;

public class RaceCondition implements Runnable {
	Counter counter;
	final static CyclicBarrier cb = new CyclicBarrier(2, new Runnable() {
		@Override
		public void run() {
			System.out.println(Thread.currentThread().getName() + "执行");
		}
	});

	RaceCondition() {
	}

	RaceCondition(Counter counter) {
		this.counter = counter;
	}

	/**
	 * @param args
	 * @throws Exception
	 */
	public static void main(String[] args) throws Exception {
		Counter counter = new Counter();
		ExecutorService service = Executors.newCachedThreadPool();
		Thread t1 = new Thread(new RaceCondition(counter));
		Thread t2 = new Thread(new RaceCondition(counter));
		service.execute(t1);
		service.execute(t2);
		service.shutdown();
		System.out.println(counter.count);
	}

	@Override
	public void run() {
		for (int i = 0; i < 5000000; i++)
			counter.add(1);
		try {
			System.out.println("before");
			cb.await();
			System.out.println("complete");
		} catch (InterruptedException e) {
			e.printStackTrace();
		} catch (BrokenBarrierException e) {
			e.printStackTrace();
		} finally {
		}
	}
}

class Counter {

	/**
	 * 此处加上volatile并没有起到预期作用，依旧要把方法同步
	 * NOTICE 这里count的值的范围是多少？
	 * 循环十次的极限情况：
	 * 1.线程一拿到了0,等待线程二
	 * 2.线程二拿到了0,并且计算了9次,得到结果9并且写了回去
	 * 3.线程一第一次计算完,并且把1写入
	 * 4.线程二第十次取到了1
	 * 5.线程一计算完成,把结果写入
	 * 6.线程二把第十次的计算结果2写入
	 */
	protected volatile long count = 0;

	public void add(long value) {
		this.count++;
	}
}

结果发现：[b]主线程依旧在子线程之前执行到了[/b]。

再看代码二：

package com.iteye.wwwcomy.thread;

import java.util.ArrayList;
import java.util.concurrent.Callable;
import java.util.concurrent.CyclicBarrier;
import java.util.concurrent.ExecutorService;
import java.util.concurrent.Executors;

public class RaceConditionWithCyclicBarrier implements Callable<Boolean> {
	Counter2 counter;
	static CyclicBarrier cb = null;

	RaceConditionWithCyclicBarrier() {
	}

	RaceConditionWithCyclicBarrier(Counter2 counter) {
		this.counter = counter;
	}

	/**
	 * @param args
	 * @throws Throwable
	 */
	public static void main(String[] args) throws Throwable {
		Counter2 counter = new Counter2();

		cb = new CyclicBarrier(2, new Runnable() { // 等到线程到达后执行一个后续task
					@Override
					public void run() {
						System.out.println(Thread.currentThread().getName() + "a");
					}
				});

		ArrayList<Callable<Boolean>> listCall = new ArrayList<Callable<Boolean>>();
		listCall.add(new RaceConditionWithCyclicBarrier(counter));
		listCall.add(new RaceConditionWithCyclicBarrier(counter));

		ExecutorService executor = Executors.newFixedThreadPool(2); // 必须是allThread的个数
		try {
			executor.invokeAll(listCall);
		} catch (Throwable e) {
			executor.shutdown();
			throw e;
		} finally {
			System.out.println("over");
		}
		executor.shutdown();
		// 这句是在所有线程都跑完之后才会执行
		System.out.println(counter.count);
	}

	@Override
	public Boolean call() throws Exception {
		for (int i = 0; i < 500000; i++)
			counter.add(1);
		cb.await();
		return true;
	}
}

class Counter2 {

	/**
	 * 此处加上volatile并没有起到预期作用，依旧要把方法同步
	 * NOTICE 这里count的值的范围是多少？
	 * 循环十次的极限情况：
	 * 1.线程一拿到了0,等待线程二
	 * 2.线程二拿到了0,并且计算了9次,得到结果9并且写了回去
	 * 3.线程一第一次计算完,并且把1写入
	 * 4.线程二第十次取到了1
	 * 5.线程一计算完成,把结果写入
	 * 6.线程二把第十次的计算结果2写入
	 */
	protected volatile long count = 0;

	public void add(long value) {
		this.count++;
	}
}

竟然发现这次主线程在executor.invokeAll(listCall);后被挂起了，直到子线程执行完成之后才执行。

其实就是使用Runnable和Callable之后executor的API使用不同而已，怎么结果差距这么大。

断个点，在visualVM里面dump了一下线程发现Callable的executor执行代码中将主线程挂起了，见线程堆栈,具体源码按照堆栈进去跟就是了：

[quote]"pool-1-thread-2" prio=6 tid=0x00000000067f7800 nid=0x1534 at breakpoint[0x000000000876f000]
java.lang.Thread.State: RUNNABLE
at com.iteye.wwwcomy.thread.RaceConditionWithCyclicBarrier.call(RaceConditionWithCyclicBarrier.java:60)
at com.iteye.wwwcomy.thread.RaceConditionWithCyclicBarrier.call(RaceConditionWithCyclicBarrier.java:1)
at java.util.concurrent.FutureTask$Sync.innerRun(FutureTask.java:303)
at java.util.concurrent.FutureTask.run(FutureTask.java:138)
at java.util.concurrent.ThreadPoolExecutor$Worker.runTask(ThreadPoolExecutor.java:886)
at java.util.concurrent.ThreadPoolExecutor$Worker.run(ThreadPoolExecutor.java:908)
at java.lang.Thread.run(Thread.java:662)

Locked ownable synchronizers:
- <0x00000007d5f2ca10> (a java.util.concurrent.locks.ReentrantLock$NonfairSync)

"pool-1-thread-1" prio=6 tid=0x00000000067f6800 nid=0x1da4 at breakpoint[0x000000000866f000]
java.lang.Thread.State: RUNNABLE
at com.iteye.wwwcomy.thread.RaceConditionWithCyclicBarrier.call(RaceConditionWithCyclicBarrier.java:60)
at com.iteye.wwwcomy.thread.RaceConditionWithCyclicBarrier.call(RaceConditionWithCyclicBarrier.java:1)
at java.util.concurrent.FutureTask$Sync.innerRun(FutureTask.java:303)
at java.util.concurrent.FutureTask.run(FutureTask.java:138)
at java.util.concurrent.ThreadPoolExecutor$Worker.runTask(ThreadPoolExecutor.java:886)
at java.util.concurrent.ThreadPoolExecutor$Worker.run(ThreadPoolExecutor.java:908)
at java.lang.Thread.run(Thread.java:662)

Locked ownable synchronizers:
- <0x00000007d5f2c7d8> (a java.util.concurrent.locks.ReentrantLock$NonfairSync)

"main" prio=6 tid=0x00000000002ae800 nid=0x168c waiting on condition [0x000000000251f000]
java.lang.Thread.State: WAITING (parking)
at sun.misc.Unsafe.park(Native Method)
- parking to wait for <0x00000007d5f2c770> (a java.util.concurrent.FutureTask$Sync)
at java.util.concurrent.locks.LockSupport.park(LockSupport.java:156)
at java.util.concurrent.locks.AbstractQueuedSynchronizer.parkAndCheckInterrupt(AbstractQueuedSynchronizer.java:811)
at java.util.concurrent.locks.AbstractQueuedSynchronizer.doAcquireSharedInterruptibly(AbstractQueuedSynchronizer.java:969)
at java.util.concurrent.locks.AbstractQueuedSynchronizer.acquireSharedInterruptibly(AbstractQueuedSynchronizer.java:1281)
at java.util.concurrent.FutureTask$Sync.innerGet(FutureTask.java:218)
at java.util.concurrent.FutureTask.get(FutureTask.java:83)
at java.util.concurrent.AbstractExecutorService.invokeAll(AbstractExecutorService.java:205)
at com.iteye.wwwcomy.thread.RaceConditionWithCyclicBarrier.main(RaceConditionWithCyclicBarrier.java:46)

Locked ownable synchronizers:
- None[/quote]

可见只有代码二的做法才能使主线程在子线程完成之后再执行，CyclicBarrier自身并不保证主线程在子线程完成之后执行。


[b][color=red]CyclicBarrier仅仅是为调用await方法的线程设置一个集合点！[/[/color]b]

[b]代码一的写法其实稍微做一点改变，也可以实现效果：
1.修改CyclicBarrier的初始线程数+1,
2.在主线程里面也加入 cb.await(); 就可以了[/b]

当时先写的代码二，所以有点先入为主了。。本来正在研究volatile,遇到这个问题先记录一下。


另外，CyclicBarrier的第二个参数，是使用最后一个执行完的线程执行的。

----2017.05.10补充-------------
最近在面试，又回看了一下当时写的这个blog，实际上代码二中主线程在子线程之后执行的原因是调用的executor方法不同，这个在blog里面没提出来，
代码一里面使用的是（submit方法同样，区别见http://stackoverflow.com/questions/3929342/choose-between-executorservices-submit-and-executorservices-execute）

executor.execute()

代码二里面使用的是

executor.invokeAll()

代码二中会在子线程执行完之后才执行主线程的后续内容。代码一会直接执行主线程后续内容。