JAVA并发(3)-ReentrantReadWriteLock的探索

1. 介绍

本文我们继续探究使用AQS的子类ReentrantReadWriteLock(读写锁)。老规矩,先贴一下类图
ReentrantReadWriteLock结构图
ReentrantReadWriteLock这个类包含读锁和写锁,这两种锁都存在是否公平的概念,这个后面会细讲。

此类跟ReentrantLock类似,有以下几种性质:

  • 可选的公平性政策
  • 重入,读锁和写锁同一个线程可以重复获取写锁可以获取读锁,反之不能
  • 锁的降级,重入还可以通过获取写锁,然后获取到读锁,通过释放写锁的方式,从而写锁降级为读锁。 然而,从读锁升级到写锁是不可能的。
  • 获取读写锁期间,支持不可中断

2. 源码剖析

先讲几个必要的知识点,然后我们再对写锁的获取与释放读锁的获取与释放进行讲解,中间穿插着讲公平与非公平的实现。

知识点一:
内部类Sync中,将AQS中的state(private volatile int state;长度是32位)逻辑分成了两份,高16位代表读锁持有的count,低16位代表写锁持有的count

Sync

    ...
        /*
         * Read vs write count extraction constants and functions.
         * Lock state is logically divided into two unsigned shorts:
         * The lower one representing the exclusive (writer) lock hold count,
         * and the upper the shared (reader) hold count.
         */
        static final int SHARED_SHIFT   = 16;
        static final int SHARED_UNIT    = (1 << SHARED_SHIFT);
        static final int MAX_COUNT      = (1 << SHARED_SHIFT) - 1;
        static final int EXCLUSIVE_MASK = (1 << SHARED_SHIFT) - 1;

        /** Returns the number of shared holds represented in count  */
        static int sharedCount(int c)    { return c >>> SHARED_SHIFT; }
        /** Returns the number of exclusive holds represented in count  */
        static int exclusiveCount(int c) { return c & EXCLUSIVE_MASK; }
   ...

This lock supports a maximum of 65535 recursive write locks and 65535 read locks. Attempts to exceed these limits result in Error throws from locking methods.(读锁与写锁都最大支持65535个)

知识点二
HoldCounter的作用,一个计数器记录每个线程持有的读锁count。使用ThreadLocal维护。缓存在cachedHoldCounter

        static final class HoldCounter {
            int count = 0;
            // Use id, not reference, to avoid garbage retention
            final long tid = getThreadId(Thread.currentThread());
        }

使用ThreadLocal维护 👇

        static final class ThreadLocalHoldCounter
            extends ThreadLocal<HoldCounter> {
            public HoldCounter initialValue() {
                return new HoldCounter();
            }
        }

维护最后一个使用HoldCounter的线程。简言之就是,假如A线程持有读锁,A线程重入获取读锁,在它之后没有其他线程获取读锁,那么当获取HoldCounter时,可以直接将cachedHoldCounter赋值给该线程,就不用从ThreadLocal中去查询了(ThreadLocal内部维持一个 Map ,想获取当前线程的值就需要去遍历查询),这样做可以节约时间。

        private transient HoldCounter cachedHoldCounter;

当前线程持有的重入读锁count,当某个线程持有的count降至0,将被删除。

        private transient ThreadLocalHoldCounter readHolds;

初始化在构造函数或readObject

        Sync() {
            readHolds = new ThreadLocalHoldCounter();
            setState(getState()); // ensures visibility of readHolds
        }
        private void readObject(java.io.ObjectInputStream s)
            throws java.io.IOException, ClassNotFoundException {
            s.defaultReadObject();
            readHolds = new ThreadLocalHoldCounter();
            setState(0); // reset to unlocked state
        }

知识点三:
是否互斥 :

读操作 写操作
读操作
写操作

只有读读不互斥,其余都互斥
获取到了写锁,也有资格获取读锁,反之不行.

知识点四
ReentranReadWriteLock,实现了ReadWriteLock接口。

public interface ReadWriteLock {
    /**
     * Returns the lock used for reading.
     *
     * @return the lock used for reading
     */
    Lock readLock();

    /**
     * Returns the lock used for writing.
     *
     * @return the lock used for writing
     */
    Lock writeLock();
}

ReadWriteLock维护着写锁读锁写锁是排他的,而读锁可以同时由多个线程持有。与互斥锁相比,读写锁的粒度更细

有了上面的知识,等会理解下面的源码就更容易了,故事从下面几个变量开始~

    private final ReentrantReadWriteLock rwl = new ReentrantReadWriteLock();
    private final Lock readLock = rwl.readLock();
    private final Lock writeLock = rwl.writeLock();

ReentrantReadWriteLock的构造器,默认是非公平模式


    /**
     * Creates a new {@code ReentrantReadWriteLock} with
     * default (nonfair) ordering properties.
     */
    public ReentrantReadWriteLock() {
        this(false);
    }

    /**
     * Creates a new {@code ReentrantReadWriteLock} with
     * the given fairness policy.
     *
     * @param fair {@code true} if this lock should use a fair ordering policy
     */
    public ReentrantReadWriteLock(boolean fair) {
        sync = fair ? new FairSync() : new NonfairSync();
        readerLock = new ReadLock(this);
        writerLock = new WriteLock(this);
    }

2.1 写锁

写锁,排他锁;一个线程获取了写锁,其他线程只能等待

2.1.1 写锁的获取

    writeLock.lock();

java.util.concurrent.locks.ReentrantReadWriteLock.WriteLock#lock

public void lock() {
            sync.acquire(1);
        }

又来到了AQSacquire方法

    public final void acquire(int arg) {
        if (!tryAcquire(arg) &&
            acquireQueued(addWaiter(Node.EXCLUSIVE), arg))
            selfInterrupt();
    }

我们看ReentrantReadWriteLock是如何实现tryAcquire

 protected final boolean tryAcquire(int acquires) {

            /*
            * 对下面的条件做一个总述:
            * 1. 当读锁或者写锁的count不为零时,同时拥有者不是当前线程,返回false
            * 2. 当count达到饱和(超过最大的限制65535),返回false
            * 3. 如果上面的情况都不是,该线程有资格去获取锁,如果它是可重入获取锁或队列策略允许它。那么就更新state并且设置锁的拥有者
            */
            // 结合总述看下面的代码,很容易就看懂了

            Thread current = Thread.currentThread();
            int c = getState();
            int w = exclusiveCount(c);

            // 存在读锁或写锁
            if (c != 0) {
                // 情况分析: 1. w = 0, 表示已经获取了读锁(不管是自己还是其他线程),直接返回false
                // 2. 写锁不为零,且当前锁的拥有者不是当前线程,返回false
                if (w == 0 || current != getExclusiveOwnerThread())
                    return false;
                if (w + exclusiveCount(acquires) > MAX_COUNT)
                    throw new Error("Maximum lock count exceeded");
                // Reentrant acquire
                setState(c + acquires);
                return true;
            }

            // 不存在读锁或写锁被占用的情况
            if (writerShouldBlock() ||
                !compareAndSetState(c, c + acquires))
                return false;
            setExclusiveOwnerThread(current);
            return true;
        }

上面代码中的writerShouldBlock方法就是tryAcquire控制公平与否的关键,我们分别看看公平与非公平是如何实现的

2.1.1.1 写锁的获取(非公平)

默认情况下是非公平的

 static final class NonfairSync extends Sync {
        ...
        // 直接返回false
        final boolean writerShouldBlock() {
            return false;
        }
        ...

2.1.1.2 写锁的获取(公平)

    static final class FairSync extends Sync {
        ...
        final boolean writerShouldBlock() {
            return hasQueuedPredecessors();
        }
        ...

判断是否该线程前面还有其他线程的结点,上一节有讲到过。

这里还贴一下,整个acquire的流程图
acquire的流程图

2.1.2 写锁的释放

下面的这段代码,记得一定放在finally中

    writeLock.unlock();
        public void unlock() {
            sync.release(1);
        }
    public final boolean release(int arg) {
        if (tryRelease(arg)) {
            Node h = head;
            if (h != null && h.waitStatus != 0)
                unparkSuccessor(h);
            return true;
        }
        return false;
    }

又看到了熟悉的面孔,但我们主要看的还是tryRelease, 👇

        protected final boolean tryRelease(int releases) {
            if (!isHeldExclusively())
                throw new IllegalMonitorStateException();
            int nextc = getState() - releases;
            boolean free = exclusiveCount(nextc) == 0;
            if (free)
                setExclusiveOwnerThread(null);
            setState(nextc);
            return free;
        }

很简单,就贴一下release的流程图
release流程图

2.2 读锁

读锁与读锁并不互斥,可以存在多个持有读锁的线程📕

2.2.1 读锁的获取

        readLock.lock();
        public void lock() {
            sync.acquireShared(1);
        }
    public final void acquireShared(int arg) {
        if (tryAcquireShared(arg) < 0)
            doAcquireShared(arg);
    }

之前的文章还没有讲解过tryAcquireShared在子类如何实现的。看看如何实现的 👇

  protected final int tryAcquireShared(int unused) {

            /*
             * 对下面的条件做一个总述:
             * 1. 如果写锁被其他线程持有,失败
             * 2. 否则,如果此线程有资格去获取写锁,首先查询是否需要阻塞(readerShouldBlock).如果不需要,就通过CAS更新state。
             *
             * 3. 如果上面两步都失败了,可能是当前线程没有资格(需要被阻塞),或CAS失败,或者count数量饱和了,然后执行fullTryAcquireShared进行重试
             */
            Thread current = Thread.currentThread();
            int c = getState();
            // 这里可以看出,若该线程持有写锁,同样也可以去获取读锁
            if (exclusiveCount(c) != 0 &&
                getExclusiveOwnerThread() != current)
                return -1;
            int r = sharedCount(c);

            // readerShouldBlock表示此线程是否被阻塞(后面细谈)
            // 多个线程执行compareAndSetState(c, c + SHARED_UNIT), 只有一个线程可以执行成功,其余线程去执行fullTryAcquireShared
            if (!readerShouldBlock() &&
                r < MAX_COUNT &&
                compareAndSetState(c, c + SHARED_UNIT)) {
                if (r == 0) {
                    // firstReader是第一个获取到读锁的线程
                    // firstReaderHoldCount是firstReader持有的count
                    firstReader = current;
                    firstReaderHoldCount = 1;
                } else if (firstReader == current) {
                    // 重入
                    firstReaderHoldCount++;
                } else {
                    // sync queue还存在其他线程持有读锁,且该线程不是第一个持有读锁的线程
                    // 结合上面的知识点二,理解下面的逻辑
                    HoldCounter rh = cachedHoldCounter;
                    if (rh == null || rh.tid != getThreadId(current))
                        cachedHoldCounter = rh = readHolds.get();
                    else if (rh.count == 0)
                        readHolds.set(rh);
                    rh.count++;
                }
                return 1;
            }
            // 到达这里的线程,1. 没有资格获取(需要阻塞) 2. CAS失败 3. 读锁count饱和
            return fullTryAcquireShared(current);
        }

Full version of acquire for reads, that handles CAS misses and reentrant reads not dealt with in tryAcquireShared.


        final int fullTryAcquireShared(Thread current) {

             // 主要逻辑跟上面的tryAcquireShared类似。tryAcquireShared的逻辑只有一个线程会成功CAS,
             // 其余的线程都进入fullTryAcquireShared,进行重试(代码不那么复杂)
            HoldCounter rh = null;
            for (;;) {
                int c = getState();
                if (exclusiveCount(c) != 0) {
                    if (getExclusiveOwnerThread() != current)
                        return -1;
                    // else we hold the exclusive lock; blocking here
                    // would cause deadlock.
                } else if (readerShouldBlock()) { // 需要被阻塞
                    // Make sure we're not acquiring read lock reentrantly
                    if (firstReader == current) {
                        // assert firstReaderHoldCount > 0;
                    } else {
                        // 此时, 如果该线程前面已经有线程获取了读锁,且当前线程持有的读锁count为0,从readHolds除去。
                        if (rh == null) {
                            rh = cachedHoldCounter;
                            if (rh == null || rh.tid != getThreadId(current)) {
                                rh = readHolds.get();
                                if (rh.count == 0)
                                    readHolds.remove();
                            }
                        }
                        // 这里表示该线程还是一个new reader, 还没有持有读锁
                        if (rh.count == 0)
                            return -1;
                    }
                }
                if (sharedCount(c) == MAX_COUNT)
                    throw new Error("Maximum lock count exceeded");
                // 这里只有一个线程会成功,若CAS失败,一直重试直到成功,或者读锁count饱和,或者需要被阻塞为止
                if (compareAndSetState(c, c + SHARED_UNIT)) {
                    if (sharedCount(c) == 0) {
                        firstReader = current;
                        firstReaderHoldCount = 1;
                    } else if (firstReader == current) {
                        firstReaderHoldCount++;
                    } else {
                        if (rh == null)
                            rh = cachedHoldCounter;
                        if (rh == null || rh.tid != getThreadId(current))
                            rh = readHolds.get();
                        else if (rh.count == 0)
                            readHolds.set(rh);
                        rh.count++;
                        cachedHoldCounter = rh; // cache for release
                    }
                    return 1;
                }
            }
        }

我们上面看到了readerShouldBlock,这个方法是控制读锁获取公平与否,下面我们分别看看在非公平与公平模式下的实现 👇

2.2.1.1 读锁的获取(非公平)

    static final class NonfairSync extends Sync {
        ...
        final boolean readerShouldBlock() {
            /* As a heuristic to avoid indefinite writer starvation,
             * block if the thread that momentarily appears to be head
             * of queue, if one exists, is a waiting writer.  This is
             * only a probabilistic effect since a new reader will not
             * block if there is a waiting writer behind other enabled
             * readers that have not yet drained from the queue.
             */
            return apparentlyFirstQueuedIsExclusive();
        }
        ...
    }

判断sync queue的head的后继结点是否是写锁(独占模式)

    final boolean apparentlyFirstQueuedIsExclusive() {
        Node h, s;
        return (h = head) != null &&
            (s = h.next)  != null &&
            !s.isShared()         &&
            s.thread != null;
    }

上面的方法是,获取读锁时,避免导致写锁饥饿(indefinite writer starvation)的一个措施,下面我们对它进行详细的解释

写锁饥饿
结合上面的图片,我们设想有一个这样的情况,写锁没有被获取,线程A获取到了读锁,此时另一个线程X想要获取写锁,但是写锁与读锁互斥,所以此时将线程X代表的node添加到sync queue中,等待读锁被释放,才有资格去获取写锁

上面的情况 + 不存在判断sync queue的head的后继结点是否是写锁(apparentlyFirstQueuedIsExclusive)的方法时,我们看看会出什么问题

时刻一: 线程B线程C线程D是新建线程想要去获取读锁(new reader),此时因为不存在写锁被获取,所以线程B线程C线程D都会在fullTryAcquireShared中不断重试,最终都获得读锁

时刻二: 线程A释放,会执行unparkSuccessor,此时线程X被唤醒,但是执行到tryAcquire,又检测到读锁被持有(不管是自己还是是其他线程)线程X又被阻塞。线程B释放,还是会出现这种情况,只有等到最后一个读锁被释放线程X才能获取到写锁。但是想想如果后面一连串的读锁线程X不是早就被'饿死了'

apparentlyFirstQueuedIsExclusive,就可以防止这种'写锁饥饿'的情况发生。线程B线程C线程D只有被阻塞,等待线程X获取到写锁,才有机会获取读锁

2.2.1.2 读锁的获取(公平)


    /**
     * Fair version of Sync
     */
    static final class FairSync extends Sync {
       ...

        final boolean readerShouldBlock() {
            return hasQueuedPredecessors();
        }
       ...
    }

这里关于读锁的获取(公平与非公平)分析完了,贴一张整个acquireShared的流程图
acquireShared流程图

2.2.2 读锁的释放

        readLock.unlock();
        public void unlock() {
            sync.releaseShared(1);
        }

AQS

    public final boolean releaseShared(int arg) {
        if (tryReleaseShared(arg)) {
            // 唤醒head后驱结点
            doReleaseShared();
            return true;
        }
        return false;
    }
 // unused = 1
 protected final boolean tryReleaseShared(int unused) {
            Thread current = Thread.currentThread();
            if (firstReader == current) {
                // assert firstReaderHoldCount > 0;
                if (firstReaderHoldCount == 1)
                    firstReader = null;
                else
                    firstReaderHoldCount--;
            } else {
                HoldCounter rh = cachedHoldCounter;
                if (rh == null || rh.tid != getThreadId(current))
                    rh = readHolds.get();
                int count = rh.count;
                if (count <= 1) {
                    readHolds.remove();
                    if (count <= 0)
                        throw unmatchedUnlockException();
                }
                --rh.count;
            }
            for (;;) {
                int c = getState();
                // 获取读锁时,会执行 compareAndSetState(c, c + SHARED_UNIT),这里就是减SHARED_UNIT
                int nextc = c - SHARED_UNIT;
                if (compareAndSetState(c, nextc))
                    // 释放读锁,对需要读锁的线程没有影响(读读不互斥)
                    // 但是,这里如果返回true,可以唤醒被阻塞的想要持有写锁的线程
                    return nextc == 0;
            }
        }

3. 总结

  1. ReentrantReadWriteLock中的防止'写锁饥饿'的操作,值得一看
  2. AQS中的state(private volatile int state;),逻辑分为高16位(代表读锁的state),低16位(代表写锁的state),是一个值得学习的办法
  3. 使用ThreadLocal维护每一个现成的读锁的重入数,使用cachedHoldCounter维护最后一个使用HoldCounter的读锁线程,节省在ThreadLocal中的查询

使用读写锁的情况,应该取决于与修改相比,读取数据的频率,读取和写入操作持续的时间。
例如:

  • 某个集合不经常修改,但是对其元素搜索很频繁,使用读写锁就是最佳选择。(简言之,就是读多写少)
  • 现在也是读多写少,但是读操作时间很短,只有一小段代码,而读写锁比互斥锁更加复杂,开销可能大于互斥锁,这种情况使用读写锁可能不合适。此时就要通过性能分析,判断使用读写锁在系统中是否合适。

4. 参考

posted on 2021-05-30 00:29  ukyu  阅读(220)  评论(0编辑  收藏  举报

导航