Fork me on GitHub
如果对您有帮助,麻烦您点赞留言吧~~~

AbstractQueuedSynchronizer(AQS源码解读)

AbstractQueuedSynchronizer继承自AbstractOwnableSynchronizer。

双向链表

head->A->B->C->D

tail->D->C->B->A

参数及代码块


	// 获取Unsafe类的实例,用于对内存进行操作(CAS操作)
	private static final Unsafe unsafe = Unsafe.getUnsafe();

	// 在内存中的偏移量
    private static final long stateOffset;
    private static final long headOffset;
    private static final long tailOffset;
    private static final long waitStatusOffset;
    private static final long nextOffset;

    static {
        try {
        	// 获取偏移量
            stateOffset = unsafe.objectFieldOffset
                (AbstractQueuedSynchronizer.class.getDeclaredField("state"));
            headOffset = unsafe.objectFieldOffset
                (AbstractQueuedSynchronizer.class.getDeclaredField("head"));
            tailOffset = unsafe.objectFieldOffset
                (AbstractQueuedSynchronizer.class.getDeclaredField("tail"));
            waitStatusOffset = unsafe.objectFieldOffset
                (Node.class.getDeclaredField("waitStatus"));
            nextOffset = unsafe.objectFieldOffset
                (Node.class.getDeclaredField("next"));

        } catch (Exception ex) { throw new Error(ex); }
    }

AbstractQueuedSynchronizer.Node

Node为内部类,数据结构为双向链表。

compareAndSetState

如果期望值和更新值不一样,则返回false。

	/**
     * 比较并且设置状态
     * @param expect 期望值
     * @param update 更新值
     */
	protected final boolean compareAndSetState(int expect, int update) {

		// 通过unsafe中的原子方法来设置
        return unsafe.compareAndSwapInt(this, stateOffset, expect, update);
    }

unsafe.compareAndSwap**方法是基于JNI的原子操作

acquire

    public final void acquire(int arg) {
        // 如果当前线程未获取到锁(即被其他线程占有),把当前线程加到队列
        if (!tryAcquire(arg) &&
            acquireQueued(addWaiter(Node.EXCLUSIVE), arg))
            selfInterrupt();
    }

tryAcquire

子类必须重写tryAcquire方法,不然会抛出异常。

    protected boolean tryAcquire(int arg) {
        throw new UnsupportedOperationException();
    }

addWaiter

addWaiter流程

放入尾节点

    private Node addWaiter(Node mode) {

        // 创建node节点
        Node node = new Node(Thread.currentThread(), mode);
        
        // tail(尾节点)
        Node pred = tail;
        if (pred != null) {
            node.prev = pred;
            // 如果尾节点设置成功,直接返回创建的node节点
            if (compareAndSetTail(pred, node)) {
                pred.next = node;
                return node;
            }
        }
        enq(node);
        return node;
    }

enq

    private Node enq(final Node node) {

        // 一直循环去获取,直到尾节点设置成功才返回
        for (;;) {
            // 双向链表的知识
            Node t = tail;
            // 尾节点为空时,默认初始化头节点=尾节点=空节点
            if (t == null) {
                if (compareAndSetHead(new Node()))
                    tail = head;
            } else {
                // node的前驱节点指向为当前的尾节点
                node.prev = t;
                // 将尾节点设置成node节点(新的尾节点)
                if (compareAndSetTail(t, node)) {
                    // 当前尾节点的后继节点指向为node节点
                    t.next = node;
                    // 返回旧的尾节点
                    return t;
                }
            }
        }
    }

acquireQueued

acquireQueued流程

这里有三种状态

  1. pred.waitStatus => 0, 返回 interrupted => false
  2. pred.waitStatus => -1, 返回 interrupted => false
  3. pred.waitStatus => -1, 线程一直被挂起, 直到锁被释放(release), 返回 interrupted => true
    final boolean acquireQueued(final Node node, int arg) {

        boolean failed = true;
        try {

            // 是否中断
            boolean interrupted = false;

            for (;;) {
                // node的前驱节点
                final Node p = node.predecessor();

                // 当p为头节点,并且获取到锁  【FIFO】先进先出
                if (p == head && tryAcquire(arg)) {

                    // 设置头节点为node
                    setHead(node);

                    // 删除p后继节点的引用
                    p.next = null;
                    failed = false;
                    return interrupted;
                }

                // 线程一直被挂起,直到上面的if成立
                // 下面分析
                if (shouldParkAfterFailedAcquire(p, node) &&
                    parkAndCheckInterrupt())
                    // 已中断
                    interrupted = true;
            }
        } finally {

            // 一旦发生异常,则会进入
            if (failed)
                cancelAcquire(node);
        }
    }

shouldParkAfterFailedAcquire

第一次会进入else,将waitStatus设置为Node.SIGNAL-1,返回false,不会执行parkAndCheckInterrupt
第二次及之后进入直接返回true,就会执行parkAndCheckInterrupt

    private static boolean shouldParkAfterFailedAcquire(Node pred, Node node) {

        // 默认为0
        int ws = pred.waitStatus;
        if (ws == Node.SIGNAL)
            /*
             * This node has already set status asking a release
             * to signal it, so it can safely park.
             */
            return true;
        if (ws > 0) {
            /*
             * Predecessor was cancelled. Skip over predecessors and
             * indicate retry.
             */
            do {
                node.prev = pred = pred.prev;
            } while (pred.waitStatus > 0);
            pred.next = node;
        } else {
            /*
             * waitStatus must be 0 or PROPAGATE.  Indicate that we
             * need a signal, but don't park yet.  Caller will need to
             * retry to make sure it cannot acquire before parking.
             */
            compareAndSetWaitStatus(pred, ws, Node.SIGNAL);
        }
        return false;
    }

parkAndCheckInterrupt

为什么要返回当前线程的中断标识呢?因为LockSupport.park()会响应线程中断。
即,当线程中断时,无论是LockSupport.unpark()还是Thread.interrupt(),都会马上执行下面的return Thread.interrupted()

    private final boolean parkAndCheckInterrupt() {
        // 阻塞线程
        LockSupport.park(this);
        // 返回当前线程中断标识
        return Thread.interrupted();
    }

LockSupport

park

    public static void park(Object blocker) {
        Thread t = Thread.currentThread();
        setBlocker(t, blocker);
        // 设置线程无限阻塞
        UNSAFE.park(false, 0L);

        // 阻塞时不会执行
        setBlocker(t, null);
    }

setBlocker

设置t线程的parkBlocker属性,记录线程是被谁阻塞的。

    private static void setBlocker(Thread t, Object arg) {
        // Even though volatile, hotspot doesn't need a write barrier here.
        UNSAFE.putObject(t, parkBlockerOffset, arg);
    }

一个LockSupport的demo

import java.util.concurrent.locks.LockSupport;

public class LockSupportDemo {

    public static Object o = new Object();

    public static void main(String[] args) throws InterruptedException {
        Thread t = new MyThread1();

        t.start();

        Thread.sleep(4000);
        LockSupport.unpark(t);
    }

    static class MyThread1 extends Thread {

        @Override
        public void run() {

            System.out.println("等待挂起");

            try {
                Thread.sleep(2000);
            } catch (InterruptedException e) {
                e.printStackTrace();
            }

            System.out.println("已挂起");
            System.out.println("等待唤醒");
            LockSupport.park(o);

            System.out.println("已唤醒");
        }
    }
}

输出为

等待挂起
已挂起
等待唤醒
已唤醒

思考?

LockSupport.park()和Object.wait()的区别?

....下次补充

Thread.interrupted()和Thread.isInterrupted()的区别?

....下次补充

cancelAcquire

    private void cancelAcquire(Node node) {

        if (node == null)
            return;

        // 删除挂载的线程
        node.thread = null;

        Node pred = node.prev;

        // 把node的前驱节点指向挂载到没有被CANCELLED的节点上
        // 为什么不判断pred不为null呢?因为在enq()方法里将节点插入到队列的时候就已经初始化过了
        /**
         * private Node enq(final Node node) {
         *     for (;;) {
         *         Node t = tail;
         *         if (t == null) { // 如果为null就初始化
         *             if (compareAndSetHead(new Node()))
         *                 tail = head;
         *         } else {
         *             node.prev = t;
         *             if (compareAndSetTail(t, node)) {
         *                 t.next = node;
         *                 return t;
         *             }
         *         }
         *     }
         * }
         */
        while (pred.waitStatus > 0)
            node.prev = pred = pred.prev;

        Node predNext = pred.next;

        // 将node节点的状态设置为CANCELLED
        node.waitStatus = Node.CANCELLED;

        // 如果node是尾节点,将尾节点设置为node的前驱节点
        if (node == tail && compareAndSetTail(node, pred)) {

            // 将node前驱节点的后继节点指向设置为null,目的是为了切断与node节点的联系
            // pred.next设置为null
            compareAndSetNext(pred, predNext, null);
        } else {
            
            int ws;
            // node的前驱节点不是头节点
            // 将node的前驱节点的状态设置为Node.SIGNAL,如果已经是Node.SIGNAL则不需要设置
            // pred.thread != null 这个是干啥的,疑问??
            if (pred != head &&
                ((ws = pred.waitStatus) == Node.SIGNAL ||
                 (ws <= 0 && compareAndSetWaitStatus(pred, ws, Node.SIGNAL))) &&
                pred.thread != null) {

                Node next = node.next;
                // node的后继节点不为null,状态不为CANCELLED
                if (next != null && next.waitStatus <= 0)

                    // 将node的前驱节点的后继节点指向设置为node的后继节点
                    // 断开node节点的前后联系
                    // APrev - ANext - NodePrev - NodeNext - CPrev - CNext
                    //                 ||
                    //      APrev - ANext - CPrev - CNext
                    compareAndSetNext(pred, predNext, next);
            } else {
                // node的前驱节点是头节点,唤醒该节点线程
                unparkSuccessor(node);
            }

            // node的后继节点指向设置为node
            node.next = node; // help GC
        }
    }

selfInterrupt

线程在等待的过程中被中断,不响应,需要补上中断。

    static void selfInterrupt() {

        // 中断当前线程
        Thread.currentThread().interrupt();
    }

release

  1. waitStatus ===>>> 0 默认值
  2. waitStatus ===>>> 1 Node.CANCELLED
  3. waitStatus ===>>> -1 Node.SIGNAL
  4. waitStatus ===>>> -2 Node.CONDITION
  5. waitStatus ===>>> -3 Node.PROPAGATE
    public final boolean release(int arg) {

        // 尝试释放锁,返回true则表示已经释放
        if (tryRelease(arg)) {
            // 头节点
            Node h = head;
            // 头节点不为null 并且waitStatus不为0
            if (h != null && h.waitStatus != 0)
                unparkSuccessor(h);
            return true;
        }
        return false;
    }

tryRelease

子类必须重写tryRelease方法,不然会抛出异常。

    protected boolean tryRelease(int arg) {
        throw new UnsupportedOperationException();
    }

unparkSuccessor

唤醒队列中的头节点线程

    /**
     * node的前驱节点是头节点,唤醒该节点线程
     */
    private void unparkSuccessor(Node node) {
        
        int ws = node.waitStatus;
        // 状态不为CANCELLED
        if (ws < 0)
            // 将waitStatus设置为0
            compareAndSetWaitStatus(node, ws, 0);

        // node的后继节点
        Node s = node.next;
        // 后继节点为空 或者 状态是CANCELLED
        if (s == null || s.waitStatus > 0) {

            s = null;

            // 找到一个有效节点
            // 尾节点!=头节点,从尾节点找到头节点的下一个未被CANCELLED的节点
            // 疑问?为什么从尾节点往前遍历,而不从头节点往后遍历??
            // HeadPrev - HeadNext - APrev - ANext - (TailPrev - TailNext) => s=TailNode => t=t.prev=ANode
            //                           ||                    t
            //             HeadPrev - HeadNext - (APrev - ANext) => s=ANode => t=t.prev=HeadNode
            //                           ||             t
            // ------------------------退出循环---------------------
            for (Node t = tail; t != null && t != node; t = t.prev)
                if (t.waitStatus <= 0)
                    s = t;
        }
        // s就是找到的有效节点(头节点之后的第一个有效节点,因为头节点是空节点)
        if (s != null)
            LockSupport.unpark(s.thread);
    }

思考?

为什么从尾节点往前遍历,而不从头节点往后遍历?

因为head.next可能被设置为null??在哪里设置为null的呢

LockSupport

unpark

解除线程阻塞

    public static void unpark(Thread thread) {
        if (thread != null)
            UNSAFE.unpark(thread);
    }
posted @ 2019-06-12 18:05  jarjune  阅读(333)  评论(0编辑  收藏  举报