聊聊高并发（二十八）解析java.util.concurrent各个组件（十） 理解ReentrantReadWriteLock可重入读-写锁

这篇讲讲ReentrantReadWriteLock可重入读写锁，它不仅是读写锁的实现，而且支持可重入性。 聊聊高并发（十五）实现一个简单的读-写锁(共享-排他锁) 这篇讲了怎样模拟一个读写锁。

可重入的读写锁的特点是

1. 当有线程获取读锁时，不同意再有线程获得写锁

2. 当有线程获得写锁时。不同意其它线程获得读锁和写锁

这里隐含着几层含义：

 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; }

1. 能够同一时候有多个线程同一时候获得读锁，进入临界区。这时候的读锁的行为和Semaphore信号量是类似的

2. 因为是可重入的。所以1个线程假设获得了读锁，那么它能够重入这个读锁

3. 假设1个线程获得了读锁。那么它不能同一时候再获得写锁，这个就是所谓的“锁升级”，读锁升级到写锁可能会造成死锁，所以是不同意的

4. 假设1个线程获得了写锁，那么不同意其它线程再获得读锁和写锁，可是它自己能够获得读锁，就是所谓的“锁降级”，锁降级是同意的

关于读写锁的实现还要考虑的几个要点：

1. 释放锁时的优先级问题。是让写锁先获得还是先让读锁先获得

2. 是否同意读线程插队

3. 是否同意写线程插队。由于读写锁一般用在大量读，少量写的情况，假设写线程没有优先级，那么可能造成写线程的饥饿

4. 锁的升降级问题，通常是同意1个线程的写锁降级为读锁，不同意读锁升级成写锁

带着问题看看ReentrantReadWriteLock的源代码。 它相同提供了Sync来继承AQS并提供扩展，可是它的Sync相比較Semaphore和CountDownLatch要更加复杂。

1. 把State状态作为一个读写锁的计数器，包含了重入的次数。

state是32位的int值，所以把高位16位作为读锁的计数器，低位的16位作为写锁的计数器，并提供了响应的读写这两个计数器的位操作方法。

计算sharedCount时，採用无符号的移位操作，右移16位就是读锁计数器的值

写锁直接用EXCLUSIVE_MASK和state做与运算。EXCLUSIVE_MASK的值是00000000000000001111111111111111，相当于计算了低位16位的值

须要注意计算出来的值包括了重入的次数。

所以MAX_COUNT限定了最大值是2^17 - 1

 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; }

HoldCount类用来计算1个线程的重入次数，并使用了1个ThreadLocal类型的HoldCounter，能够记录每一个线程的锁的重入次数。 cachedHoldCounter记录了最后1个获取读锁的线程的重入次数。 firstReader指向了第一个获取读锁的线程，firstReaderHoldCounter记录了第一个获取读锁的线程的重入次数

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

        /**
         * ThreadLocal subclass. Easiest to explicitly define for sake
         * of deserialization mechanics.
         */
        static final class ThreadLocalHoldCounter
            extends ThreadLocal<HoldCounter> {
            public HoldCounter initialValue() {
                return new HoldCounter();
            }
        }

/**
         * The hold count of the last thread to successfully acquire
         * readLock. This saves ThreadLocal lookup in the common case
         * where the next thread to release is the last one to
         * acquire. This is non-volatile since it is just used
         * as a heuristic, and would be great for threads to cache.
         *
         * <p>Can outlive the Thread for which it is caching the read
         * hold count, but avoids garbage retention by not retaining a
         * reference to the Thread.
         *
         * <p>Accessed via a benign data race; relies on the memory
         * model's final field and out-of-thin-air guarantees.
         */
        private transient HoldCounter cachedHoldCounter;

Sync提供了两个抽象方法给子类扩展。用来表示读锁和写锁是否应该堵塞等待

/**
         * Returns true if the current thread, when trying to acquire
         * the read lock, and otherwise eligible to do so, should block
         * because of policy for overtaking other waiting threads.
         */
        abstract boolean readerShouldBlock();

        /**
         * Returns true if the current thread, when trying to acquire
         * the write lock, and otherwise eligible to do so, should block
         * because of policy for overtaking other waiting threads.
         */
        abstract boolean writerShouldBlock();

写锁的tryXXX获取和释放

1. 写锁释放时，因为没有其它线程获得临界区。它的tryRelease()方法仅仅须要设置状态的值。通过exclusiveCount计算写锁的计数器，假设为0表示释放了写锁，就把exclusiveOwnerThread设置为null.

2. 写锁的tryAcquire获取时。

先推断状态是否为0，为0表示没有线程获得锁，就能够直接设置状态。然后把exclusiveOwnerThread设置为当前线程

假设状态不为0，那表示有几种可能：写锁为0。读锁不为0。写锁不为0。读锁为0。写锁不为0，读锁也不为0。

所以它先推断写锁是否为0。写锁为0，那么表示读锁肯定不会为0，就失败，

或者写锁不为0，可是exclusiveOwnerThread不是自己。那么表示已经有其它线程获得了写锁，就失败

写锁不为0，而且exclusiveOwnerThread是自己。那么肯定表示是写锁的重入的情况，所以设置state状态。返回成功。

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;
        }

protected final boolean tryAcquire(int acquires) {
            /*
             * Walkthrough:
             * 1. If read count nonzero or write count nonzero
             *    and owner is a different thread, fail.
             * 2. If count would saturate, fail. (This can only
             *    happen if count is already nonzero.)
             * 3. Otherwise, this thread is eligible for lock if
             *    it is either a reentrant acquire or
             *    queue policy allows it. If so, update state
             *    and set owner.
             */
            Thread current = Thread.currentThread();
            int c = getState();
            int w = exclusiveCount(c);
            if (c != 0) {
                // (Note: if c != 0 and w == 0 then shared count != 0)
                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;
            }
            if (writerShouldBlock() ||
                !compareAndSetState(c, c + acquires))
                return false;
            setExclusiveOwnerThread(current);
            return true;
        }

读锁的tryXXX获取和释放

1. 读锁释放时基于共享的方式，改动线程各自的HoldCounter的值。最后採用位操作改动位于state的整体的读锁计数器。tryReleaseShared()之后详细的释放兴许线程的操作由AQS依据队列状态来决定。

2. 读所获取时先看写锁的计数器，假设写锁已经被获取。而且不是当前线程所获取的。就直接失败返回

这里会进行一次高速路径获取，尝试获取一次，假设readShouldBlock()返回false，而且CAS操作成功了，意思是能够获得锁，就更新相关读锁计数器

否则就进行轮询方式的获取fullTryAcquireShared()

也就是说假设当前没有线程获取写锁，或者是自己获取写锁。就能够获取读锁

一个线程获取了写锁之后，它还能够获取读锁，也就是所谓的“锁降级”，但这时候其它线程无法获取读锁。在检查到有其它写锁存在时就退出了

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 != current.getId())
                    rh = readHolds.get();
                int count = rh.count;
                if (count <= 1) {
                    readHolds.remove();
                    if (count <= 0)
                        throw unmatchedUnlockException();
                }
                --rh.count;
            }
            for (;;) {
                int c = getState();
                int nextc = c - SHARED_UNIT;
                if (compareAndSetState(c, nextc))
                    // Releasing the read lock has no effect on readers,
                    // but it may allow waiting writers to proceed if
                    // both read and write locks are now free.
                    return nextc == 0;
            }
        }

protected final int tryAcquireShared(int unused) {
            /*
             * Walkthrough:
             * 1. If write lock held by another thread, fail.
             * 2. Otherwise, this thread is eligible for
             *    lock wrt state, so ask if it should block
             *    because of queue policy. If not, try
             *    to grant by CASing state and updating count.
             *    Note that step does not check for reentrant
             *    acquires, which is postponed to full version
             *    to avoid having to check hold count in
             *    the more typical non-reentrant case.
             * 3. If step 2 fails either because thread
             *    apparently not eligible or CAS fails or count
             *    saturated, chain to version with full retry loop.
             */
            Thread current = Thread.currentThread();
            int c = getState();
            if (exclusiveCount(c) != 0 &&
                getExclusiveOwnerThread() != current)
                return -1;
            int r = sharedCount(c);
            if (!readerShouldBlock() &&
                r < MAX_COUNT &&
                compareAndSetState(c, c + SHARED_UNIT)) {
                if (r == 0) {
                    firstReader = current;
                    firstReaderHoldCount = 1;
                } else if (firstReader == current) {
                    firstReaderHoldCount++;
                } else {
                    HoldCounter rh = cachedHoldCounter;
                    if (rh == null || rh.tid != current.getId())
                        cachedHoldCounter = rh = readHolds.get();
                    else if (rh.count == 0)
                        readHolds.set(rh);
                    rh.count++;
                }
                return 1;
            }
            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) {
            /*
             * This code is in part redundant with that in
             * tryAcquireShared but is simpler overall by not
             * complicating tryAcquireShared with interactions between
             * retries and lazily reading hold counts.
             */
            HoldCounter rh = cachedHoldCounter;
            if (rh == null || rh.tid != current.getId())
                rh = readHolds.get();
            for (;;) {
                int c = getState();
                int w = exclusiveCount(c);
                if ((w != 0 && getExclusiveOwnerThread() != current) ||
                    ((rh.count | w) == 0 && readerShouldBlock(current)))
                    return -1;
                if (sharedCount(c) == MAX_COUNT)
                    throw new Error("Maximum lock count exceeded");
                if (compareAndSetState(c, c + SHARED_UNIT)) {
                    cachedHoldCounter = rh; // cache for release
                    rh.count++;
                    return 1;
                }
            }
        } 

tryWriteLock和tryReadLock操作和上面的操作类似，它们是读写锁的tryLock()的实际实现，表示尝试获取一次锁

1. tryWriteLock方法尝试获得写锁，先推断状态是否为0，为0而且CAS操作成功就表示获得锁。假设状态不为0，就推断写锁计数器的值。假设写锁计数器为0就表示存在读锁，就返回失败。获取写锁不为0，可是不是当前线程所获取的，也返回失败。仅仅有写锁不为0而且是当前线程自己获取的写锁，就是所谓的写锁重入操作。

CAS成功后就表示获得写锁

final boolean tryWriteLock() {
            Thread current = Thread.currentThread();
            int c = getState();
            if (c != 0) {
                int w = exclusiveCount(c);
                if (w == 0 ||current != getExclusiveOwnerThread())
                    return false;
                if (w == MAX_COUNT)
                    throw new Error("Maximum lock count exceeded");
            }
            if (!compareAndSetState(c, c + 1))
                return false;
            setExclusiveOwnerThread(current);
            return true;
        }

final boolean tryReadLock() {
            Thread current = Thread.currentThread();
            for (;;) {
                int c = getState();
                if (exclusiveCount(c) != 0 &&
                    getExclusiveOwnerThread() != current)
                    return false;
                if (sharedCount(c) == MAX_COUNT)
                    throw new Error("Maximum lock count exceeded");
                if (compareAndSetState(c, c + SHARED_UNIT)) {
                    HoldCounter rh = cachedHoldCounter;
                    if (rh == null || rh.tid != current.getId())
                        cachedHoldCounter = rh = readHolds.get();
                    rh.count++;
                    return true;
                }
            }
        }

ReentrantReadWriteLock也提供了非公平和公平的两个Sync版本号

非公平的版本号中

1. 写锁总是优先获取。不考虑AQS队列中先来的线程

2. 读锁也不按FIFO队列排队，而是看当前获得锁是否是写锁，假设是写锁，就等待。否则就尝试获得锁

而公平版本号中

1. 假设有其它锁存在，获取写锁操作就失败。应该(should)进AQS队列等待

2. 假设有其它锁存在。获取读锁操作就失败。应该(should)进AQS队列等待

final static class NonfairSync extends Sync {
        private static final long serialVersionUID = -8159625535654395037L;
        final boolean writerShouldBlock(Thread current) {
            return false; // writers can always barge
        }
        final boolean readerShouldBlock(Thread current) {
            /* 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 probablistic 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();
        }
    }

    /**
     * Fair version of Sync
     */
    final static class FairSync extends Sync {
        private static final long serialVersionUID = -2274990926593161451L;
        final boolean writerShouldBlock(Thread current) {
            // only proceed if queue is empty or current thread at head
            return !isFirst(current);
        }
        final boolean readerShouldBlock(Thread current) {
            // only proceed if queue is empty or current thread at head
            return !isFirst(current);
        }
    }

详细ReadLock和WriteLock的实现就是依赖Sync来实现的，默认是非公平版本号的Sync。

读锁採用共享默认的AQS，它提供了中断/不可中断的lock操作，tryLock操作，限时的tryLock操作。

值得注意的时读锁不支持newCondition操作。

 public static class ReadLock implements Lock, java.io.Serializable  {
        private static final long serialVersionUID = -5992448646407690164L;
        private final Sync sync;

        protected ReadLock(ReentrantReadWriteLock lock) {
            sync = lock.sync;
        }

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

        public void lockInterruptibly() throws InterruptedException {
            sync.acquireSharedInterruptibly(1);
        }

        public  boolean tryLock() {
            return sync.tryReadLock();
        }

        public boolean tryLock(long timeout, TimeUnit unit) throws InterruptedException {
            return sync.tryAcquireSharedNanos(1, unit.toNanos(timeout));
        }

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

        public Condition newCondition() {
            throw new UnsupportedOperationException();
        }

WriteLock基于独占模式的AQS，它提供了中断/不可中断的lock操作。tryLock操作，限时的tryLock操作

 public static class WriteLock implements Lock, java.io.Serializable  {
        private static final long serialVersionUID = -4992448646407690164L;
	private final Sync sync;

        protected WriteLock(ReentrantReadWriteLock lock) {
            sync = lock.sync;
        }

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

        public void lockInterruptibly() throws InterruptedException {
            sync.acquireInterruptibly(1);
        }

        public boolean tryLock( ) {
            return sync.tryWriteLock();
        }

        public boolean tryLock(long timeout, TimeUnit unit) throws InterruptedException {
            return sync.tryAcquireNanos(1, unit.toNanos(timeout));
        }

        public void unlock() {
            sync.release(1);
        }

        public Condition newCondition() {
            return sync.newCondition();
        }

最后再说一下AQS和各种同步器实现的关系，AQS提供了同步队列和条件队列的管理。包含各种情况下的入队出队操作。

而同步器子类实现了tryAcquire和tryRelease方法来操作状态。来表示什么情况下能够直接获得锁而不须要进入AQS。什么情况下获取锁失败则须要进入AQS队列等待