// 获取共享资源,如果未能获得,则让线程入队 public final void acquireShared(int arg) { if (tryAcquireShared(arg) < 0) doAcquireShared(arg); }
作用:阻塞机制
/** * Acquires in shared uninterruptible mode. * 共享、不可中断模式,可类比于 acquireQueued 方法 */ private void doAcquireShared(int arg) { final Node node = addWaiter(Node.SHARED); boolean failed = true; try { boolean interrupted = false; for (;;) { final Node p = node.predecessor(); if (p == head) { /* tryAcquireShared:以共享模式 acquire,它会去看 state 是否允许被 acquire,如果可以的话就 acquire 1) CountDownLatch:获取操作表示,等待并直到闭锁状态结束(state == 0) protected int tryAcquireShared(int acquires) { return (getState() == 0) ? 1 : -1; } 2) Semaphore:提供公平和非公平的实现方式,这里放公平方式 protected int tryAcquireShared(int acquires) { for (;;) { if (hasQueuedPredecessors()) return -1; int available = getState(); int remaining = available - acquires; // 返回剩余的许可 if (remaining < 0 || compareAndSetState(available, remaining)) return remaining; } } */ int r = tryAcquireShared(arg); if (r >= 0) { setHeadAndPropagate(node, r); p.next = null; // help GC if (interrupted) selfInterrupt(); failed = false; return; } } // 如果 p 不是 head, 或者是 head 但是未能成功 acquire,则需要阻塞该线程 // shouldParkAfterFailedAcquire 方法中,如果 p 为 SINGAL 状态,将 node 的线程阻塞;否则将状态为0或者是 PROPAGATE 状态的 p 设置为 SINGAL 状态,准备进行阻塞 if (shouldParkAfterFailedAcquire(p, node) && parkAndCheckInterrupt()) interrupted = true; } } finally { if (failed) cancelAcquire(node); } }
作用:
releaseShare
传播/** * Sets head of queue, and checks if successor may be waiting * in shared mode, if so propagating if either propagate > 0 or * PROPAGATE status was set. * 在调用该方法之前,都会检查 propagate,只有大于等于0,才会调用该方法 */ private void setHeadAndPropagate(Node node, int propagate) { Node h = head; // Record old head for check below setHead(node); /* * Try to signal next queued node if: * Propagation was indicated by caller, * or was recorded (as h.waitStatus either before * or after setHead) by a previous operation * (note: this uses sign-check of waitStatus because * PROPAGATE status may transition to SIGNAL.) * and * The next node is waiting in shared mode, * or we don't know, because it appears null * * The conservatism in both of these checks may cause * unnecessary wake-ups, but only when there are multiple * racing acquires/releases, so most need signals now or soon * anyway. */ /* 唤醒下一个线程的情况包括:(在独占模式中,设置好 head 后,不会去唤醒下一个线程) 1)propagate 大于0,唤醒下一个线程; 2)propagate 等于0,但是其他线程将 h(旧的 head)的状态改为了 PROPAGATE 状态 h == null 的判断是防止空指针异常。 h.waitStatus < 0,表示 h 处于 SINGAL 或 PROPAGATE 状态,一般情况下是 PROPAGATE 状态,因为在 doReleaseShared 方法中 h 状态变化是 SINGAL -> 0 -> PROPAGATE。那么为什么 SINGAL 状态也要唤醒呢?这是因为在 doAcquireShared 中,第一次没有获得足够的资源时,shouldParkAfterFailedAcquire 将 PROPAGATE 状态转换成 SINGAL,准备阻塞线程,但是第二次进入本方法时发现资源刚好够,而此时 h 的状态是 SINGAL 状态 (h = head) == null 是再次检查 */ if (propagate > 0 || h == null || h.waitStatus < 0 || (h = head) == null || h.waitStatus < 0) { Node s = node.next; /* 如果 next node 是 SHARED 模式,或者是 null,则执行传播操作 final boolean isShared() { return nextWaiter == SHARED; } */ if (s == null || s.isShared()) doReleaseShared(); } }
public final boolean releaseShared(int arg) { if (tryReleaseShared(arg)) { // 唤醒后继节点以及设置传播状态 doReleaseShared(); return true; } return false; }
作用:
SINGLA
状态设置为0,或者将0设置为 PROPAGATE
unparkSuccessor
去唤醒/** * Release action for shared mode -- signals successor and ensures * propagation. (Note: For exclusive mode, release just amounts * to calling unparkSuccessor of head if it needs signal.) */ private void doReleaseShared() { /* * Ensure that a release propagates, even if there are other * in-progress acquires/releases. This proceeds in the usual * way of trying to unparkSuccessor of head if it needs * signal. But if it does not, status is set to PROPAGATE to * ensure that upon release, propagation continues. * Additionally, we must loop in case a new node is added * while we are doing this. Also, unlike other uses of * unparkSuccessor, we need to know if CAS to reset status * fails, if so rechecking. * 共享模式存在多个线程,需要 CAS 来观察操作是否成功 */ for (;;) { Node h = head; if (h != null && h != tail) { int ws = h.waitStatus; if (ws == Node.SIGNAL) { // head 是 SIGNAL 状态,尝试将状态设为0,设置成功的话唤醒下一个 Node if (!compareAndSetWaitStatus(h, Node.SIGNAL, 0)) continue; // loop to recheck cases unparkSuccessor(h); } // 状态为0表示已经有一个线程将 head 的状态设置为0了,当前线程尝试设置 PROPAGATE 状态 else if (ws == 0 && !compareAndSetWaitStatus(h, 0, Node.PROPAGATE)) continue; // loop on failed CAS } // 共享模式,可能有多个线程在操作,如果 head 发生变化,应该去唤醒下一个 Node if (h == head) // loop if head changed break; } }
PROPAGETE
表示需要将 releaseShared
传播给其他 Node。为什么需要将 releaseShared
传播呢?考虑有10个线程,3个资源,假设当前有3个线程在运行,那么有7个线程在 CLH 队列中处于阻塞状态(node.prev
的状态都是 SINGAL
)。不考虑外部有新的线程过来,有以下情况:
3个线程串行执行释放资源->唤醒线程->设置 head
,那么每个线程串行着将 head 的状态位修改为0,不会在 setHeadAndPropagate
方法中调用 doReleaseShared
3个线程同时释放资源,设3个线程分别是线程1、线程2、线程3。线程4是 CLH 队列中的第一个线程
线程1成功将 head 的状态设置为0,线程4被唤醒。线程1看 head 是否发生变化,线程2、3重试准备将 head 的状态从0设置为 PROPAGETE
,最终三个线程都会查看 head 是否发生变化
setHead
。线程1、2、3结束 doReleaseShared
,多出3个资源,活跃线程数为1,且 head 的状态一定是 PROPAGETE
。线程4通过**旧的 head **的状态,知道应该调用 doReleaseShared
来唤醒线程setHead
。线程1、2、3继续尝试唤醒线程因此,PROPAGETE
状态可以让活跃的线程数量尽可能达到资源的数量,以利用资源
阻塞机制由 doAcquireShared
方法实现,唤醒机制由 doReleaseShared
方法实现
PROPAGATE
,那么就调用 doReleaseShared
来唤醒线程;如果 Node 未能获得资源,则 head 的 waitStatus
设置为 SINGAL
,如果下一次检查它是 SINGAL
,则阻塞 Node。此外它做的另一个工作是清理 head 和当前 Node 之间处于 CANCELLED
状态的 Node (shouldParkAfterFailedAcquire
)SINGAL
改为0,如果改成功则调用 unparkSuccessor
唤醒下一个线程,以及 CAS 操作将 head 的状态从0改为 PROPAGATE
, CAS 操作失败后继续循环。如果发现 head 的状态不是 SINGAL
或者0,就去看 head 是否改变,如果改变了继续循环