AQS的conditionObject实现类似object的wait/notify/notify的功能,功能大概是:
1.object维护一个监视器和一个等待队列,condition对于一个lock可以有多个condition,对于每个condition维护一个条件队列;
2.提供wait/signal/signalall功能。
来个入门demo:
public class ConditionTest { private static ReentrantLock lock = new ReentrantLock(); private static Condition condition = lock.newCondition(); public static void main(String[] args) { new Thread(new Runnable() { @Override public void run() { try { lock.lock(); System.out.println(Thread.currentThread()+ "等待条件完成"); condition.await(); } catch (InterruptedException e) { e.printStackTrace(); } finally { System.out.println(Thread.currentThread()+ "终于等到条件完成了,gogogo"); lock.unlock(); } } }).start(); Thread b = new Thread(new Runnable() { @Override public void run() { try { lock.lock(); condition.signalAll(); System.out.println(Thread.currentThread()+ "条件完成了,释放吧"); } finally { lock.unlock(); } } }); b.start(); } }
ConditionObject每次new都会维护一个条件队列,通过node的nextWaiter串起来
/** 条件队列的第一个节点 */ private transient Node firstWaiter; /** 条件队列的最后一个节点 */ private transient Node lastWaiter; /** * 空的构造,看下ReentrantLock.newCondition()每次都会new ConditionObject()可以维护多个条件队列 */ public ConditionObject() { }
/** 响应中断的await 能调用await的方法的线程肯定获得锁 */ public final void await() throws InterruptedException { //线程中断直接异常 if (Thread.interrupted()) throw new InterruptedException(); //将当前线程封装加入condition的条件队列 Node node = addConditionWaiter(); //释放AQS同步等待队列中的节点 int savedState = fullyRelease(node); int interruptMode = 0; //看节点是否还在AQS的同步等待队列,因为signal/signalall调用的话会把节点加入到AQS的等待队列,如果没在那就说明需要park while (!isOnSyncQueue(node)) { //不在的话那就应该在条件队列了,那么park吧 LockSupport.park(this); //被signal/signalall唤醒后,检查中断状态,如果被中断,break,没有的话while if ((interruptMode = checkInterruptWhileWaiting(node)) != 0) break; } //这里说明已经加入到AQS的队列,重新acquire,注意的是acquireQueued返回值为是否中断,返回true肯定是中断,返回false if (acquireQueued(node, savedState) && interruptMode != THROW_IE) interruptMode = REINTERRUPT; if (node.nextWaiter != null) // clean up if cancelled unlinkCancelledWaiters(); if (interruptMode != 0)//中断时,直接throw还是设置中断状态 reportInterruptAfterWait(interruptMode); } /** 先判断lastWaiter的状态,如果不是condition就过一遍条件队列,将所有状态不为condition的都去掉 然后将节点加入到lastWaiter(类似AQS中的tail)的nextWaiter,如果last为null,就将first和last的nextWaiter都指向新节点 最后将lastWaiter指向新加入节点 */ private Node addConditionWaiter() { Node t = lastWaiter; // If lastWaiter is cancelled, clean out. if (t != null && t.waitStatus != Node.CONDITION) { unlinkCancelledWaiters(); t = lastWaiter; } Node node = new Node(Thread.currentThread(), Node.CONDITION); if (t == null) firstWaiter = node; else t.nextWaiter = node; lastWaiter = node; return node; } /** 从firstWaiter开始,过滤掉所有状态不为condition的节点 基本上按trail-t-next逐个节点向后移动,t从firstWaiter开始 当时看的时候,拿纸画了一遍才清楚 */ private void unlinkCancelledWaiters() { Node t = firstWaiter; Node trail = null; while (t != null) { Node next = t.nextWaiter; if (t.waitStatus != Node.CONDITION) { t.nextWaiter = null; if (trail == null) firstWaiter = next; else trail.nextWaiter = next; if (next == null) lastWaiter = trail; } else trail = t; t = next; } } /** 这里是释放掉AQS同步等待队列中的节点 返回释放前的state值 有异常的话就将节点的状态改为cancelled */ final int fullyRelease(Node node) { boolean failed = true; try { int savedState = getState(); if (release(savedState)) { failed = false; return savedState; } else { throw new IllegalMonitorStateException(); } } finally { if (failed) node.waitStatus = Node.CANCELLED; } } /** 看节点是否还在AQS的队列中 */ final boolean isOnSyncQueue(Node node) { if (node.waitStatus == Node.CONDITION || node.prev == null) return false; if (node.next != null) // If has successor, it must be on queue return true; /* 之前分析过AQS加入节点的顺序enq(),pre-tail-next,pre加入了,但是并不能说明这个节点就真正在AQS的等待队列, 所以需要从tail往前过滤一遍看是否存在 */ return findNodeFromTail(node); } /** 从tail往前判断节点是否在队列中,找到返回true */ private boolean findNodeFromTail(Node node) { Node t = tail; for (;;) { if (t == node) return true; if (t == null) return false; t = t.prev; } } /** 2个状态,表示await被唤醒后,如果检查线程是中断的,就需要判断是在什么时候被中断,然后判断怎么返回这个中断,是直接异常还是设置中断状态 */ private static final int REINTERRUPT = 1; private static final int THROW_IE = -1; /** 检查中断状态,0:未中断, THROW_IE: REINTERRUPT: */ private int checkInterruptWhileWaiting(Node node) { return Thread.interrupted() ? (transferAfterCancelledWait(node) ? THROW_IE : REINTERRUPT) : 0; } /** 检查中断是在什么时候发生的,是在signal前还是signal后 */ final boolean transferAfterCancelledWait(Node node) { //如果调用signal的话,先把节点的状态设置成0,再把节点从条件队列转移(enq)到AQS的等待队列 //所以下面这个cas成功,那么这个中断肯定是发生在signal前 if (compareAndSetWaitStatus(node, Node.CONDITION, 0)) { //把节点放入AQS队列,保证后面acquireQueued执行 enq(node); return true; } /* * If we lost out to a signal(), then we can't proceed * until it finishes its enq(). Cancelling during an * incomplete transfer is both rare and transient, so just * spin. */ /** 到这里的话,肯定是已经发生了signal,但是signal的enq没有完成,所以自旋,让signal的enq完成,返回false */ while (!isOnSyncQueue(node)) Thread.yield(); return false; } /** 这是根据之前的标识判断怎么处理中断 signal前就抛出,signal后就设置中断状态 */ private void reportInterruptAfterWait(int interruptMode) throws InterruptedException { if (interruptMode == THROW_IE) throw new InterruptedException(); else if (interruptMode == REINTERRUPT) selfInterrupt(); }整个await的流程为:
1.判断线程中断,中断直接抛出异常
2.将节点加入condition条件队列
3.释放AQS队列中的锁
4.while判断是否在AQS等待队列
5.如果不在AQS队列中,就park
6.唤醒后检查是被signal唤醒还是中断唤醒
7.中断唤醒要判断signal前还是signal后,设置怎么处理中断,signal前的话还需要将节点enq到AQS的等待队列,转到4
8.如果在就acquireQueued,重新获取,这里判断acquire返回,为true则为中断,然后设置中断处理方式
9.如果节点的nextWaiter不为null,就清理下condition的条件队列,清除所有状态不为condition的节点
10.最后看是否需要处理中断,如有,signal前的中断直接抛出,signal后设置中断状态。
awaitNanos/awaitUntil/await(long time, TimeUnit unit)基本流程跟响应中断的await差不多,只不过多了超时时间处理,跟前面讲过的响应超时没什么区别,都是底层unsafe的那些。
看下signal/signalAll:
public final void signal() { if (!isHeldExclusively()) throw new IllegalMonitorStateException(); Node first = firstWaiter; if (first != null) doSignal(first); } /**子类实现判断是否是自己拥有*/ protected boolean isHeldExclusively() { throw new UnsupportedOperationException(); } /** 这里从first开始释放一个condition状态的节点 */ private void doSignal(Node first) { do { if ( (firstWaiter = first.nextWaiter) == null) lastWaiter = null; first.nextWaiter = null; } while (!transferForSignal(first) && (first = firstWaiter) != null); } final boolean transferForSignal(Node node) { /* * If cannot change waitStatus, the node has been cancelled. */ //设置节点状态为0 if (!compareAndSetWaitStatus(node, Node.CONDITION, 0)) return false; /* * Splice onto queue and try to set waitStatus of predecessor to * indicate that thread is (probably) waiting. If cancelled or * attempt to set waitStatus fails, wake up to resync (in which * case the waitStatus can be transiently and harmlessly wrong). */ //将节点加入AQS的等待队列,返回的是加入节点的pre Node p = enq(node); int ws = p.waitStatus; //设置节点状态为SIGNAL,如果失败直接unpark新加入的节点 if (ws > 0 || !compareAndSetWaitStatus(p, ws, Node.SIGNAL)) LockSupport.unpark(node.thread); return true; }signal只释放first的开始第一个状态为condition的节点,然后将节点加入到AQS的同步等待队列,设置新加入节点的pre的状态为SIGNAL。看下signalAll的释放:
private void doSignalAll(Node first) { lastWaiter = firstWaiter = null; do { Node next = first.nextWaiter; first.nextWaiter = null; transferForSignal(first); first = next; } while (first != null); }看到signalAll最终是处理所有condition节点。
其实不管是await还是signal/signalAll都是模拟object.wait跟notify/notifyAll,可以对比来看。
AQS大概就这么多了,还有个AbstractQueuedLongSynchronizer这个类,跟AQS差不多,只是state状态采用的是long类型:
private volatile long state;
private volatile int state;
注意:
await会有虚假唤醒的情况,即使没有signal,await的线程也可能被唤醒。参考:多线程编程中条件变量和虚假唤醒(spurious wakeup)的讨论 http://siwind.iteye.com/blog/1469216,最后的建议就是使用 while判断条件而不是使用if判断:
while(条件不满足){
condition_wait(cond, mutex);
}
而不是:
If( 条件不满足 ){
Condition_wait(cond,mutex);
}
说实话最后我也没看懂什么原因导致虚假唤醒,后来去stackoverflow查询了下,这是解释,自己研究吧
http://stackoverflow.com/questions/1050592/do-spurious-wakeups-actually-happen,还有这篇http://blog.sina.com.cn/s/blog_e59371cc0102v29b.html
参考:
http://blog.csdn.net/yuenkin/article/details/50867530
http://brokendreams.iteye.com/blog/2250372
http://ifeve.com/understand-condition/comment-page-1/#comment-26901