作用

用来创建锁和其他同步类的基本线程阻塞的原语.在查看AbstractQueuedSynchronizer(AQS)中,发现其底层就是使用LockSupport.park()和LockSupport.unpark()实现线程的阻塞和唤醒的.它的作用与Object上的wait和notify很像,但是还是存在一些不同.

不需要在同步代码块中调用

之前在java多线程之wait和notify中有讲过,使用wait()和notify()必须在同步代码快中才是调用,如果不是则会抛出IllegalMonitorStateException异常.但是LockSupport.park()则没有该限制.下面示例代码,如果去掉synchronized同步代码块将抛出异常.

public class App1 {

public static void main(String[] args) throws InterruptedException {

Object lock = new Object();

Thread t1 = new Thread(() -> {

synchronized (lock){

try {

lock.wait();

System.out.println(Thread.currentThread().getName()+":执行完成");

} catch (InterruptedException e) {

e.printStackTrace();

}

}

}, "t1");

t1.start();

Thread t2 = new Thread(() -> {

LockSupport.park();

System.out.println(Thread.currentThread().getName()+":执行完成");

}, "t2");

t2.start();

Thread.sleep(1000L);

//唤醒t1,t2

synchronized (lock){

lock.notify();

}

LockSupport.unpark(t2);

}

}

调用先后不会导致死锁

在使用wait和notify时,如果线程A先调用了notify,然后线程B再调用的wait没有指定超时时间,那么线程B将一直处于WAITING状态,直到有其他线程调用notify或者notifyAll.使用park和unpark则不会导致该情况发生.代码如下:

public class App2 {

public static void main(String[] args) throws InterruptedException {

Object lock = new Object();

Thread t1 = new Thread(() -> {

synchronized (lock){

System.out.println(Thread.currentThread().getName()+":开始执行");

lock.notify();

}

synchronized (lock){

try {

lock.wait();

System.out.println(Thread.currentThread().getName()+":执行完成");

} catch (InterruptedException e) {

e.printStackTrace();

}

}

}, "t1");

Thread t2 = new Thread(() -> {

System.out.println(Thread.currentThread().getName()+":开始执行");

LockSupport.unpark(Thread.currentThread());

LockSupport.park();

System.out.println(Thread.currentThread().getName()+":执行完成");

}, "t2");

t1.start();

t2.start();

t1.join();

t2.join();

}

}

响应中断的不同

使用wait使线程进入WAITING或者TIMED_WAITING,如果中断线程,线程将抛出中断异常InterruptedException.如果使用park使线程进入等待状态,并不会抛出中断异常.

public class App3 {

public static void main(String[] args) throws InterruptedException {

Object lock = new Object();

Thread t1 = new Thread(() -> {

synchronized (lock){

try {

lock.wait();

} catch (InterruptedException e) {

Thread thread = Thread.currentThread();

System.out.println(thread.getName()+":响应中断退出");

}

}

}, "t1");

Thread t2 = new Thread(() -> {

LockSupport.park();

Thread thread = Thread.currentThread();

System.out.println(thread.getName()+":响应中断退出");

}, "t2");

t1.start();

t2.start();

Thread.sleep(1000L);

System.out.println("t1.getState() = " + t1.getState());

System.out.println("t2.getState() = " + t2.getState());

t1.interrupt();

t2.interrupt();

}

}

不可重入

LockSupport很类似于二元信号量(只有1个许可证可供使用),如果这个许可还没有被占用,当前线程获取许可并继续执行;如果许可已经被占用,当前线程阻塞,等待获取许可.如果调用unpark多次释放许可,再多次调用park获取许可只能获取到一次许可.如下面代码,第二次调用park将导致线程永远的等待下去.

public class App4 {

public static void main(String[] args) {

Thread thread = Thread.currentThread();

System.out.println("第一次调用unpark");

LockSupport.unpark(thread);

System.out.println("第二次调用unpark");

LockSupport.unpark(thread);

System.out.println("第一次调用park");

LockSupport.park();

System.out.println("第二次调用park");

LockSupport.park();

System.out.println("执行完成");

}

}

上面示例代码永远也不会打印执行完成.