15. Reentrant Lock

ReentrantLock,意思是“可重入锁”,关于可重入锁的概念在下一节讲述。ReentrantLock是唯一实现了Lock接口的类,并且ReentrantLock提供了更多的方法。下面通过一些实例看具体看一下如何使用ReentrantLock。

1. Lock代码实例:

你可以在[github][1]中看到。

1.1 lock();

首先,我们尝试写一个错误的代码:

public class ReentrantLockWrongSample {
    public static void main(String[] args) {

        ArrayList<Integer> arrayList = new ArrayList<Integer>();

        WrongSampleThread thread1 = new WrongSampleThread(arrayList);
        WrongSampleThread thread2 = new WrongSampleThread(arrayList);

        new Thread(thread1).start();
        new Thread(thread1).start();
        new Thread(thread2).start();
        new Thread(thread2).start();

    }
}

public class WrongSampleThread implements Runnable {
    private ArrayList<Integer> arrayList;
    public WrongSampleThread(ArrayList<Integer> arrayList) {
        this.arrayList = arrayList;
    }

    @Override
    public void run() {
        Lock lock = new ReentrantLock();    //注意这个地方
        lock.lock();
        try {
            System.out.println(Thread.currentThread() + "得到了锁");
            for (int i = 0; i < 5; i++) arrayList.add(i);
            Thread.sleep(100);
        } catch (Exception e) {
        } finally {
            System.out.println(Thread.currentThread() + "释放了锁");
            lock.unlock();
        }
    }
}

output:

Thread[Thread-0,5,main]得到了锁
Thread[Thread-3,5,main]得到了锁
Thread[Thread-2,5,main]得到了锁
Thread[Thread-1,5,main]得到了锁
Thread[Thread-1,5,main]释放了锁
Thread[Thread-0,5,main]释放了锁
Thread[Thread-3,5,main]释放了锁
Thread[Thread-2,5,main]释放了锁

第二个线程怎么会在第一个线程释放锁之前得到了锁?原因在于,在insert方法中的lock变量是局部变量,每个线程执行该方法时都会保存一个副本,那么理所当然每个线程执行到lock.lock()处获取的是不同的锁,所以就不会发生冲突。 所以我们只需要把Lock设置为类的属性:

public class ReentrantLockRightSample {
    public static void main(String[] args) {
        ArrayList<Integer> list = new ArrayList<Integer>();
        Lock lock = new ReentrantLock();    //注意这个地方

        RightSampleThread thread1 = new RightSampleThread(lock, list);
        RightSampleThread thread2 = new RightSampleThread(lock, list);

        new Thread(thread1).start();
        new Thread(thread2).start();
    }
}

public class RightSampleThread implements Runnable {
    private Lock lock;
    private List<Integer> list;

    public RightSampleThread(Lock lock, List<Integer> list) {
        this.lock = lock;
        this.list = list;
    }

    @Override
    public void run() {
        lock.lock();
        try {
            System.out.println(Thread.currentThread() + "得到了锁");
            for (int i = 0; i < 5; i++) list.add(i);
            Thread.sleep(200);
        } catch (Exception e) {
            // TODO: handle exception
        } finally {
            System.out.println(Thread.currentThread() + "释放了锁");
            lock.unlock();
        }
    }
}

output:

Thread[Thread-0,5,main]得到了锁
Thread[Thread-0,5,main]释放了锁
Thread[Thread-1,5,main]得到了锁
Thread[Thread-1,5,main]释放了锁

1.2 tryLock()

public class ReentrantLockTryLockLearn {
    public static void main(String[] args) {
        List<Integer> list = new ArrayList<Integer>();
        Lock lock = new ReentrantLock();    //注意这个地方

        TryLockThread thread1 = new TryLockThread(list, lock);
        TryLockThread thread2 = new TryLockThread(list, lock);
        TryLockWithTimeThread thread3 = new TryLockWithTimeThread(list, lock);
        new Thread(thread1).start();
        new Thread(thread2).start();
        new Thread(thread3).start();
    }
}

public class TryLockThread implements Runnable {
    private List<Integer> list;
    private Lock lock;

    public TryLockThread(List<Integer> list, Lock lock) {
        this.list = list;
        this.lock = lock;
    }
    @Override
    public void run() {
        if (lock.tryLock()) {
            try {
                System.out.println(Thread.currentThread() + "得到了锁");
                for (int i = 0; i < 10; i++) {
                    list.add(i);
                }
                Thread.sleep(500);
            } catch (Exception e) {
                // TODO: handle exception
            } finally {
                System.out.println(Thread.currentThread() + "释放了锁");
                lock.unlock();
            }
        } else {
            System.out.println(Thread.currentThread() + "获取锁失败");
        }
    }
}

public class TryLockWithTimeThread implements Runnable {
    private List<Integer> list;
    private Lock lock;

    public TryLockWithTimeThread(List<Integer> list, Lock lock) {
        this.list = list;
        this.lock = lock;
    }
    @Override
    public void run() {
        try {
           if (lock.tryLock(5, TimeUnit.SECONDS)) {
                try {
                    System.out.println(Thread.currentThread() + "得到了锁");
                    for (int i = 0; i < 10; i++) {
                        list.add(i);
                    }
                    Thread.sleep(100);
                } catch (Exception e) {
                    // TODO: handle exception
                } finally {
                    System.out.println(Thread.currentThread() + "释放了锁");
                    lock.unlock();
                }
            } else {
                System.out.println(Thread.currentThread() + "获取锁失败");
            }
        } catch (InterruptedException e) {
            e.printStackTrace();
        }
    }
}

output:

Thread[Thread-0,5,main]得到了锁
Thread[Thread-1,5,main]获取锁失败
Thread[Thread-0,5,main]释放了锁
Thread[Thread-2,5,main]得到了锁
Thread[Thread-2,5,main]释放了锁

1.3 lockInterruptibly()

public class LockInterruptiblyLockLearn {
    public static void main(String[] args) {
        Lock lock = new ReentrantLock();
        LockInterruptiblyThread interruptiblyThread1 = new LockInterruptiblyThread(lock);
        LockInterruptiblyThread interruptiblyThread2 = new LockInterruptiblyThread(lock);
        Thread thread1 = new Thread(interruptiblyThread1);
        Thread thread2 = new Thread(interruptiblyThread2);
        thread1.start();
        thread2.start();
        try {
            Thread.sleep(2000);
        } catch (InterruptedException e) {
            e.printStackTrace();
        }
        thread2.interrupt();
    }
}

public class LockInterruptiblyThread implements Runnable {
    private Lock lock;
    public LockInterruptiblyThread(Lock lock) {
        this.lock = lock;
    }
    @Override
    public void run() {
        try {
            //注意,如果需要正确中断等待锁的线程,必须将获取锁放在外面,然后将InterruptedException抛出
            lock.lockInterruptibly();
            try {
                System.out.println(Thread.currentThread() + "得到了锁");
                long startTime = System.currentTimeMillis();
                for (; ; ) {
                    if (System.currentTimeMillis() - startTime >= 5000) break;
                    //插入数据
                }
            } finally {
                System.out.println(Thread.currentThread().getName() + "执行finally");
                lock.unlock();
                System.out.println(Thread.currentThread() + "释放了锁");
            }
        } catch (InterruptedException e) {
            e.printStackTrace();
            System.out.println(Thread.currentThread().getName() + "被中断");
        }
    }
}

output:

Thread[Thread-0,5,main]得到了锁
java.lang.InterruptedException
	at java.util.concurrent.locks.AbstractQueuedSynchronizer.doAcquireInterruptibly(AbstractQueuedSynchronizer.java:898)
	at java.util.concurrent.locks.AbstractQueuedSynchronizer.acquireInterruptibly(AbstractQueuedSynchronizer.java:1222)
	at java.util.concurrent.locks.ReentrantLock.lockInterruptibly(ReentrantLock.java:335)
	at com.lock.reentrantLock.reentrantLockApi.thread.LockInterruptiblyThread.run(LockInterruptiblyThread.java:38)
	at java.lang.Thread.run(Thread.java:748)
Thread-1被中断
Thread-0执行finally
Thread[Thread-0,5,main]释放了锁

2. Condition 例子:

Condition 将 Object 监视器方法(wait、notify 和 notifyAll)分解成截然不同的对象,以便通过将这些对象与任意 Lock 实现组合使用,为每个对象提供多个等待 set (wait-set)。

其中,Lock 替代了 synchronized 方法和语句的使用,Condition 替代了 Object 监视器方法的使用。

在Condition中,用await()替换wait(),用signal()替换notify(),用signalAll()替换notifyAll(),传统线程的通信方式,Condition都可以实现,这里注意,Condition是被绑定到Lock上的,要创建一个Lock的Condition必须用newCondition()方法。

这样看来,Condition和传统的线程通信没什么区别,Condition的强大之处在于它可以为多个线程间建立不同的Condition。

你可以在[github][4]中发现源码。

首先我们建立一个ArrayBuffer类,这个类中,我们设置了两个方法,read()put(String number)write方法是向队列中写入数据,read方法是从队列中读取数据。

public class ArrayBuffer {

    private Lock lock = new ReentrantLock();
    private Condition write;
    private Condition read;
    public ArrayBuffer() {
        write = lock.newCondition();
        read = lock.newCondition();
    }
    private String[] array = new String[30];
    private int putPoint, takePoint;
    private int count = 0;

    public String read() {
        lock.lock();
        try {

            if (count == 0) {
                try {
                    System.out.println("buffer is empty");
                    write.signal();
                    read.await();
                } catch (InterruptedException e) {e.printStackTrace();}
            }
            System.out.println("takePoint: " + takePoint);
            String temp = array[takePoint()];
            count--;
            takePoint++;
            return temp;
        }finally {
            lock.unlock();
        }
    }

    public void put(String number) {
        lock.lock();
        try{
            if (count>=array.length){
                try {
                    System.out.println("buffer is full.");
                    read.signal();
                    write.await();
                    System.out.println("count: " + count);
                } catch (InterruptedException e) {e.printStackTrace();}
            }
            array[putPoint()] = number;
            count++;
            System.out.println("putPoint: " + putPoint);
            putPoint++;
        }finally {lock.unlock();}
    }
    private int putPoint() {
        return putPoint%(array.length);
    }

    private int takePoint() {
        return takePoint%(array.length);
    }

}

其次,我们建立两个不同的thread,分别读取和写入。

public class ReadThread implements Runnable {
    private ArrayBuffer buffer;
    public ReadThread(ArrayBuffer buffer) {
        this.buffer = buffer;
    }
    @Override
    public void run() {
        System.out.println("read running.");
        for (int i =0;i<500;i++) {
            System.out.println("read result: " + buffer.read());
        }
    }
}

public class WriteThread implements Runnable {
    private ArrayBuffer buffer;
    public WriteThread(ArrayBuffer buffer) {
        this.buffer = buffer;
    }
    @Override
    public void run() {
        System.out.println("write running");
        for (int i =0;i<500;i++) {
            buffer.put(i+"");
        }
    }
}

测试类

public class ConditionOnReadWriteLearn {
    public static void main(String[] args) {
        ArrayBuffer buffer = new ArrayBuffer();
        ReadThread read = new ReadThread(buffer);
        WriteThread write = new WriteThread(buffer);

        new Thread(write).start();
        new Thread(read).start();

    }
}

output:

read running.
write running
putPoint: 0
takePoint: 0
read result: 0
putPoint: 1
···
putPoint: 9
putPoint: 10
takePoint: 1
read result: 1
putPoint: 11
···
putPoint: 31
buffer is full.
takePoint: 2
read result: 2
···
takePoint: 31
read result: 31
buffer is empty
count: 0
putPoint: 32
putPoint: 33
putPoint: 34
···

ReentrantLock源码分析

类的继承关系

ReentrantLock实现了Lock接口,Lock接口中定义了lock与unlock相关操作,并且还存在newCondition方法,表示生成一个条件。

public class ReentrantLock implements Lock, java.io.Serializable

类的内部类

ReentrantLock总共有三个内部类,并且三个内部类是紧密相关的,下面先看三个类的关系。

image

说明: ReentrantLock类内部总共存在Sync、NonfairSync、FairSync三个类,NonfairSync与FairSync类继承自Sync类,Sync类继承自AbstractQueuedSynchronizer抽象类。下面逐个进行分析。

  • Sync类

Sync类的源码如下:

abstract static class Sync extends AbstractQueuedSynchronizer {
    // 序列号
    private static final long serialVersionUID = -5179523762034025860L;
    
    // 获取锁
    abstract void lock();
    
    // 非公平方式获取
    final boolean nonfairTryAcquire(int acquires) {
        // 当前线程
        final Thread current = Thread.currentThread();
        // 获取状态
        int c = getState();
        if (c == 0) { // 表示没有线程正在竞争该锁
            if (compareAndSetState(0, acquires)) { // 比较并设置状态成功,状态0表示锁没有被占用
                // 设置当前线程独占
                setExclusiveOwnerThread(current); 
                return true; // 成功
            }
        }
        else if (current == getExclusiveOwnerThread()) { // 当前线程拥有该锁
            int nextc = c + acquires; // 增加重入次数
            if (nextc < 0) // overflow
                throw new Error("Maximum lock count exceeded");
            // 设置状态
            setState(nextc); 
            // 成功
            return true; 
        }
        // 失败
        return false;
    }
    
    // 试图在共享模式下获取对象状态,此方法应该查询是否允许它在共享模式下获取对象状态,如果允许,则获取它
    protected final boolean tryRelease(int releases) {
        int c = getState() - releases;
        if (Thread.currentThread() != getExclusiveOwnerThread()) // 当前线程不为独占线程
            throw new IllegalMonitorStateException(); // 抛出异常
        // 释放标识
        boolean free = false; 
        if (c == 0) {
            free = true;
            // 已经释放,清空独占
            setExclusiveOwnerThread(null); 
        }
        // 设置标识
        setState(c); 
        return free; 
    }
    
    // 判断资源是否被当前线程占有
    protected final boolean isHeldExclusively() {
        // While we must in general read state before owner,
        // we don't need to do so to check if current thread is owner
        return getExclusiveOwnerThread() == Thread.currentThread();
    }

    // 新生一个条件
    final ConditionObject newCondition() {
        return new ConditionObject();
    }

    // Methods relayed from outer class
    // 返回资源的占用线程
    final Thread getOwner() {        
        return getState() == 0 ? null : getExclusiveOwnerThread();
    }
    // 返回状态
    final int getHoldCount() {            
        return isHeldExclusively() ? getState() : 0;
    }

    // 资源是否被占用
    final boolean isLocked() {        
        return getState() != 0;
    }

    /**
        * Reconstitutes the instance from a stream (that is, deserializes it).
        */
    // 自定义反序列化逻辑
    private void readObject(java.io.ObjectInputStream s)
        throws java.io.IOException, ClassNotFoundException {
        s.defaultReadObject();
        setState(0); // reset to unlocked state
    }
}  

Sync类存在如下方法和作用如下。

image

  • NonfairSync类

NonfairSync类继承了Sync类,表示采用非公平策略获取锁,其实现了Sync类中抽象的lock方法,源码如下:

// 非公平锁
static final class NonfairSync extends Sync {
    // 版本号
    private static final long serialVersionUID = 7316153563782823691L;

    // 获得锁
    final void lock() {
        if (compareAndSetState(0, 1)) // 比较并设置状态成功,状态0表示锁没有被占用
            // 把当前线程设置独占了锁
            setExclusiveOwnerThread(Thread.currentThread());
        else // 锁已经被占用,或者set失败
            // 以独占模式获取对象,忽略中断
            acquire(1); 
    }

    protected final boolean tryAcquire(int acquires) {
        return nonfairTryAcquire(acquires);
    }
}

说明: 从lock方法的源码可知,每一次都尝试获取锁,而并不会按照公平等待的原则进行等待,让等待时间最久的线程获得锁。

  • FairSyn类

FairSync类也继承了Sync类,表示采用公平策略获取锁,其实现了Sync类中的抽象lock方法,源码如下:

// 公平锁
static final class FairSync extends Sync {
    // 版本序列化
    private static final long serialVersionUID = -3000897897090466540L;

    final void lock() {
        // 以独占模式获取对象,忽略中断
        acquire(1);
    }

    /**
        * Fair version of tryAcquire.  Don't grant access unless
        * recursive call or no waiters or is first.
        */
    // 尝试公平获取锁
    protected final boolean tryAcquire(int acquires) {
        // 获取当前线程
        final Thread current = Thread.currentThread();
        // 获取状态
        int c = getState();
        if (c == 0) { // 状态为0
            if (!hasQueuedPredecessors() &&
                compareAndSetState(0, acquires)) { // 不存在已经等待更久的线程并且比较并且设置状态成功
                // 设置当前线程独占
                setExclusiveOwnerThread(current);
                return true;
            }
        }
        else if (current == getExclusiveOwnerThread()) { // 状态不为0,即资源已经被线程占据
            // 下一个状态
            int nextc = c + acquires;
            if (nextc < 0) // 超过了int的表示范围
                throw new Error("Maximum lock count exceeded");
            // 设置状态
            setState(nextc);
            return true;
        }
        return false;
    }
}

说明: 跟踪lock方法的源码可知,当资源空闲时,它总是会先判断sync队列(AbstractQueuedSynchronizer中的数据结构)是否有等待时间更长的线程,如果存在,则将该线程加入到等待队列的尾部,实现了公平获取原则。其中,FairSync类的lock的方法调用如下,只给出了主要的方法。

image

说明: 可以看出只要资源被其他线程占用,该线程就会添加到sync queue中的尾部,而不会先尝试获取资源。这也是和Nonfair最大的区别,Nonfair每一次都会尝试去获取资源,如果此时该资源恰好被释放,则会被当前线程获取,这就造成了不公平的现象,当获取不成功,再加入队列尾部。

类的属性

ReentrantLock类的sync非常重要,对ReentrantLock类的操作大部分都直接转化为对Sync和AbstractQueuedSynchronizer类的操作。

public class ReentrantLock implements Lock, java.io.Serializable {
    // 序列号
    private static final long serialVersionUID = 7373984872572414699L;    
    // 同步队列
    private final Sync sync;
}

类的构造函数

  • ReentrantLock()型构造函数

默认是采用的非公平策略获取锁

public ReentrantLock() {
    // 默认非公平策略
    sync = new NonfairSync();
}
  • ReentrantLock(boolean)型构造函数

可以传递参数确定采用公平策略或者是非公平策略,参数为true表示公平策略,否则,采用非公平策略:

public ReentrantLock(boolean fair) {
    sync = fair ? new FairSync() : new NonfairSync();
}

核心函数分析

通过分析ReentrantLock的源码,可知对其操作都转化为对Sync对象的操作,由于Sync继承了AQS,所以基本上都可以转化为对AQS的操作。如将ReentrantLock的lock函数转化为对Sync的lock函数的调用,而具体会根据采用的策略(如公平策略或者非公平策略)的不同而调用到Sync的不同子类。

所以可知,在ReentrantLock的背后,是AQS对其服务提供了支持,由于之前我们分析AQS的核心源码,遂不再累赘。下面还是通过例子来更进一步分析源码。

示例分析

公平锁

import java.util.concurrent.locks.Lock;
import java.util.concurrent.locks.ReentrantLock;

class MyThread extends Thread {
    private Lock lock;
    public MyThread(String name, Lock lock) {
        super(name);
        this.lock = lock;
    }
    
    public void run () {
        lock.lock();
        try {
            System.out.println(Thread.currentThread() + " running");
            try {
                Thread.sleep(500);
            } catch (InterruptedException e) {
                e.printStackTrace();
            }
        } finally {
            lock.unlock();
        }
    }
}

public class AbstractQueuedSynchronizerDemo {
    public static void main(String[] args) throws InterruptedException {
        Lock lock = new ReentrantLock(true);
        
        MyThread t1 = new MyThread("t1", lock);        
        MyThread t2 = new MyThread("t2", lock);
        MyThread t3 = new MyThread("t3", lock);
        t1.start();
        t2.start();    
        t3.start();
    }
}

运行结果(某一次):

Thread[t1,5,main] running
Thread[t2,5,main] running
Thread[t3,5,main] running

说明: 该示例使用的是公平策略,由结果可知,可能会存在如下一种时序。

image

说明: 首先,t1线程的lock操作 -> t2线程的lock操作 -> t3线程的lock操作 -> t1线程的unlock操作 -> t2线程的unlock操作 -> t3线程的unlock操作。根据这个时序图来进一步分析源码的工作流程。

  • t1线程执行lock.lock,下图给出了方法调用中的主要方法。

image

说明: 由调用流程可知,t1线程成功获取了资源,可以继续执行。

  • t2线程执行lock.lock,下图给出了方法调用中的主要方法。

image

说明: 由上图可知,最后的结果是t2线程会被禁止,因为调用了LockSupport.park。

  • t3线程执行lock.lock,下图给出了方法调用中的主要方法。

image

说明: 由上图可知,最后的结果是t3线程会被禁止,因为调用了LockSupport.park。

  • t1线程调用了lock.unlock,下图给出了方法调用中的主要方法。

image

说明: 如上图所示,最后,head的状态会变为0,t2线程会被unpark,即t2线程可以继续运行。此时t3线程还是被禁止。

  • t2获得cpu资源,继续运行,由于t2之前被park了,现在需要恢复之前的状态,下图给出了方法调用中的主要方法。

image

说明: 在setHead函数中会将head设置为之前head的下一个结点,并且将pre域与thread域都设置为null,在acquireQueued返回之前,sync queue就只有两个结点了。

  • t2执行lock.unlock,下图给出了方法调用中的主要方法。

image

说明: 由上图可知,最终unpark t3线程,让t3线程可以继续运行。

  • t3线程获取cpu资源,恢复之前的状态,继续运行。

image

说明: 最终达到的状态是sync queue中只剩下了一个结点,并且该节点除了状态为0外,其余均为null。

  • t3执行lock.unlock,下图给出了方法调用中的主要方法。

image

说明: 最后的状态和之前的状态是一样的,队列中有一个空节点,头节点为尾节点均指向它。