随笔- 261 文章- 0 评论- 12 阅读- 45万

Thread 之 join() 方法

先看一段代码,下面 count 的值是多少?

@Slf4j
public class ThreadDemo {
    private static int count = 0;
 
    public static void main(String[] args) throws InterruptedException {
        FutureTask<String> futureTask = new FutureTask(() -> {
            log.info(Thread.currentThread().getName() + " 开始运行");
            count = 10;
            TimeUnit.SECONDS.sleep(3);
            log.info(Thread.currentThread().getName() + " 运行结束");
            return "success";
        });
        log.info(Thread.currentThread().getName() + " 开始运行");
        Thread t1 = new Thread(futureTask, "t1");
        t1.start();
        count++;
        log.info("count: {}", count);
        log.info(Thread.currentThread().getName() + " 运行结束");
    }
}

分析一下结果

主线程开始运行,与此同时 t1 线程也开始运行,t1 运行过程中调用 sleep() 方法,t1 线程阻塞,main 线程获取到 CPU 时间片执行后续代码,count++,最终结果为 1

但是如果我想要等 t1 线程执行完毕之后才执行主线程的代码,那么该怎么做呢?

这个时候就需要使用 join(...) 方法,join 方法的作用就是等待其它线程运行结束

join 有两个重载方法,分别是 join() 和 join(long millis),但是实际上底层都是使用同一个方法

看一下 join() 的具体实现

public final void join() throws InterruptedException {
    // 调用的是 join(long millis) 这个方法,只是时间为 0
    join(0);
}

接着再看一下 join(long millis) 这个方法

public final synchronized void join(long millis)throws InterruptedException {
    long base = System.currentTimeMillis();
    long now = 0;
 
    if (millis < 0) {
        throw new IllegalArgumentException("timeout value is negative");
    }
    // 如果 millis 的值为 0,底层调用 wait() 方法一直等待,如果没有调用 notify() 或 notifyAll() 方法,线程将一直等待
    if (millis == 0) {
        while (isAlive()) {
            wait(0);
        }
        // millis 的值不为 0 的时候,会判断当前线程是否存活,如果存活的话会计算剩余等待时间,过时不候,底层也是调用 wait(delay),delay 时间到了之后会被唤醒 
    } else {
        while (isAlive()) {
            long delay = millis - now;
            if (delay <= 0) {
                break;
            }
            wait(delay);
            now = System.currentTimeMillis() - base;
        }
    }
}

看完了代码之后,继续来看一下如何使用 join() 方法

@Slf4j
public class ThreadDemo {
    private static int count = 0;
 
    public static void main(String[] args) throws InterruptedException {
        FutureTask<String> futureTask = new FutureTask(() -> {
            log.info(Thread.currentThread().getName() + " 开始运行");
            count = 10;
            TimeUnit.SECONDS.sleep(3);
            log.info(Thread.currentThread().getName() + " 运行结束");
            return "success";
        });
        log.info(Thread.currentThread().getName() + " 开始运行");
        Thread t1 = new Thread(futureTask, "t1");
        t1.start();
        // 在 t1 线程启动之后,在 main 线程中调用 t1 线程的 join 方法,那么 main 线程会等待 t1 线程运行结束,然后 main 线程再执行剩余的代码
        t1.join();
        count++;
        log.info("count: {}", count);
        log.info(Thread.currentThread().getName() + " 运行结束");
    }
}

从上面的结果可以看出,在 main 线程中,使用 t1.join() 方法,那么 main 线程会等待 t1 线程执行结束后,main 线程再执行剩余的代码

当有多个线程 join 时,又是怎样的结果呢?

@Slf4j
public class ThreadDemo {
    private static int count = 0;
 
    public static void main(String[] args) throws InterruptedException {
        FutureTask<String> futureTask = new FutureTask(() -> {
            log.info(Thread.currentThread().getName() + " 开始运行");
            TimeUnit.SECONDS.sleep(3);
            count = 10;
            log.info(Thread.currentThread().getName() + " 运行结束");
            return "success";
        });
        FutureTask<String> futureTask2 = new FutureTask(() -> {
            log.info(Thread.currentThread().getName() + " 开始运行");
            TimeUnit.SECONDS.sleep(5);
            count = 20;
            log.info(Thread.currentThread().getName() + " 运行结束");
            return "success";
        });
 
        log.info(Thread.currentThread().getName() + " 开始运行");
         
        Thread t1 = new Thread(futureTask, "t1");
        Thread t2 = new Thread(futureTask2, "t2");
        t1.start();
        t2.start();
 
        // 同时等待 t1、t2 执行完毕
        t1.join();
        t2.join();
 
        count++;
        log.info("count: {}", count);
        log.info(Thread.currentThread().getName() + " 运行结束");
    }
}