这样终止线程,竟然会导致服务宕机?

2,314 阅读7分钟

在开始之前,我们先来看以下代码会有什么问题?

public class ThreadStopExample {
    public static void main(String[] args) throws InterruptedException {
        Thread t1 = new Thread(() -> {
            try {
                System.out.println("子线程开始执行");
                // 模拟业务处理
                Thread.sleep(1000);
            } catch (Exception e) { }
            // 伪代码:重要的业务方法
            System.out.println("子线程的重要业务方法");
        });
        t1.start();
        // 让子线程先运行一点业务
        Thread.sleep(100);
        // 终止子线程
        t1.stop();
        // 等待一段时间,确保子线程“执行完”
        Thread.sleep(3000);
        System.out.println("主线程执行完成");
    }
}

或许你已经发现了,上面这段代码使用了 Thread.stop() 来终止线程,在 Java 程序中是不允许这样终止线程的。什么?你问为什么不能这样?

首先来说 IDE 都会鄙视你了,它会阻止你使用 Thread.stop() !

什么?你不信。那么来看这张图:

image.png

好吧,那为什么不能这样用呢?总得给我一个敷衍的理由吧?

问题一:破坏了程序的完整性

其实是这样的,以文章刚开头的那段代码来说,它的执行结果是:

子线程开始执行

主线程执行完成

我们发现了一个惊天的大问题,最重要的那段伪代码竟然没执行,如下图所示:

image.png

可以看出使用 stop() 终止线程之后,线程剩余的部分代码会放弃执行,这样会造成严重的且不易被发现的惊天大 Bug,假如没有执行的那段代码是释放系统资源的代码,或者是此程序的主要逻辑处理代码。这就破坏了程序基本逻辑的完整性,导致意想不到的问题发生,而且它还很隐秘,不易被发现和修复。

有人说,这还不简单,我加个 finally 不就完了吗?

这???杠精哪都有,今年特别多。

行,既然这个说服不了你,咱接着往下看。

问题二:破坏了原子逻辑

我们知道在 Java 中 synchronized 属于独占式可重入悲观锁,如果我们使用它修饰代码,妥妥的多线程没问题,但如果碰到 stop() 方法就不一定了,直接来看代码吧。

public class ThreadStopExample {
    public static void main(String[] args) throws InterruptedException {
        MyThread myThread = new MyThread();
        Thread t2 = new Thread(myThread);
        // 开启线程
        t2.start();
        for (int i = 0; i < 10; i++) {
            Thread t = new Thread(myThread);
            t.start();
        }
        // 结束线程
        t2.stop();
    }

    /**
     * 自定义原子测试线程
     */
    static class MyThread implements Runnable {
        // 计数器
        int num = 0;

        @Override
        public void run() {
            // 同步代码块,保证原子操作
            synchronized (MyThread.class) {
                // 自增
                num++;
                try {
                    // 线程休眠 0.1 秒
                    Thread.sleep(100);
                } catch (InterruptedException e) {
                    e.printStackTrace();
                }
                // 自减
                num--;
                System.out.println(Thread.currentThread().getName() + " | num=" + num);
            }
        }
    }
}

以上程序的执行结果为:

Thread-5 | num=1

Thread-4 | num=1

Thread-2 | num=1

Thread-1 | num=1

Thread-8 | num=1

Thread-6 | num=1

Thread-9 | num=1

Thread-3 | num=1

Thread-7 | num=1

Thread-10 | num=1

从结果可以看出,以上代码经过 synchronized 修饰的 ++ 和 -- 操作,到最后打印的结果 num 竟然不是 0,而是 1。

这是因为 stop() 方法会释放此线程中的所有锁,导致程序执行紊乱,破坏了程序的原子操作逻辑

以上的这些问题,导致了 JDK 废弃了 stop() 的方法,它的废弃源码如下:

/**
 * Forces the thread to stop executing.
 * <p>
 * If there is a security manager installed, its <code>checkAccess</code>
 * method is called with <code>this</code>
 * as its argument. This may result in a
 * <code>SecurityException</code> being raised (in the current thread).
 * <p>
 * If this thread is different from the current thread (that is, the current
 * thread is trying to stop a thread other than itself), the
 * security manager's <code>checkPermission</code> method (with a
 * <code>RuntimePermission("stopThread")</code> argument) is called in
 * addition.
 * Again, this may result in throwing a
 * <code>SecurityException</code> (in the current thread).
 * <p>
 * The thread represented by this thread is forced to stop whatever
 * it is doing abnormally and to throw a newly created
 * <code>ThreadDeath</code> object as an exception.
 * <p>
 * It is permitted to stop a thread that has not yet been started.
 * If the thread is eventually started, it immediately terminates.
 * <p>
 * An application should not normally try to catch
 * <code>ThreadDeath</code> unless it must do some extraordinary
 * cleanup operation (note that the throwing of
 * <code>ThreadDeath</code> causes <code>finally</code> clauses of
 * <code>try</code> statements to be executed before the thread
 * officially dies).  If a <code>catch</code> clause catches a
 * <code>ThreadDeath</code> object, it is important to rethrow the
 * object so that the thread actually dies.
 * <p>
 * The top-level error handler that reacts to otherwise uncaught
 * exceptions does not print out a message or otherwise notify the
 * application if the uncaught exception is an instance of
 * <code>ThreadDeath</code>.
 *
 * @exception  SecurityException  if the current thread cannot
 *               modify this thread.
 * @see        #interrupt()
 * @see        #checkAccess()
 * @see        #run()
 * @see        #start()
 * @see        ThreadDeath
 * @see        ThreadGroup#uncaughtException(Thread,Throwable)
 * @see        SecurityManager#checkAccess(Thread)
 * @see        SecurityManager#checkPermission
 * @deprecated This method is inherently unsafe.  Stopping a thread with
 *       Thread.stop causes it to unlock all of the monitors that it
 *       has locked (as a natural consequence of the unchecked
 *       <code>ThreadDeath</code> exception propagating up the stack).  If
 *       any of the objects previously protected by these monitors were in
 *       an inconsistent state, the damaged objects become visible to
 *       other threads, potentially resulting in arbitrary behavior.  Many
 *       uses of <code>stop</code> should be replaced by code that simply
 *       modifies some variable to indicate that the target thread should
 *       stop running.  The target thread should check this variable
 *       regularly, and return from its run method in an orderly fashion
 *       if the variable indicates that it is to stop running.  If the
 *       target thread waits for long periods (on a condition variable,
 *       for example), the <code>interrupt</code> method should be used to
 *       interrupt the wait.
 *       For more information, see
 *       <a href="{@docRoot}/../technotes/guides/concurrency/threadPrimitiveDeprecation.html">Why
 *       are Thread.stop, Thread.suspend and Thread.resume Deprecated?</a>.
 */
@Deprecated
public final void stop() {
    SecurityManager security = System.getSecurityManager();
    if (security != null) {
        checkAccess();
        if (this != Thread.currentThread()) {
            security.checkPermission(SecurityConstants.STOP_THREAD_PERMISSION);
        }
    }
    // A zero status value corresponds to "NEW", it can't change to
    // not-NEW because we hold the lock.
    if (threadStatus != 0) {
        resume(); // Wake up thread if it was suspended; no-op otherwise
    }

    // The VM can handle all thread states
    stop0(new ThreadDeath());
}

可以看出 stop() 方法被 @Deprecated 注释修饰了,而被此注解修饰的代码表示为过时方法,不建议被使用。从 stop() 的备注信息可以看出,官方也不建议使用 stop() ,说它是一个非安全的方法。

正确终止线程

那如何终止线程呢?这里提供 2 个正确的方法:

  1. 设置退出标识退出线程;
  2. 使用 interrupt() 方法终止线程。

1.自定义退出标识

我们可以自定义一个布尔变量来标识是否需要退出线程,实现代码如下:

// 自定义退出标识退出线程
static class FlagThread extends Thread {
    public volatile boolean exit = false;

    public void run() {
        while (!exit) {
            // 执行正常的业务逻辑
        }
    }
}

可以看出我们使用了关键字 volatile 对线程进行了修饰,这样就可以保证多线程的执行安全了,在我们需要让线程退出时,只需要把变量 exit 赋值为 true 就可以了。

2.interrupt 终止线程

当我们使用 interrupt() 方法时,以上两个示例的执行结果就正常了,执行代码如下:

public class ThreadStopExample {
    public static void main(String[] args) throws InterruptedException {
        // 问题一:破坏了程序的完整性
        Thread t1 = new Thread(() -> {
            try {
                System.out.println("子线程开始执行");
                // 模拟业务处理
                Thread.sleep(1000);
            } catch (Exception e) { }
            // 伪代码:重要业务方法
            System.out.println("子线程的重要业务方法");
        });
        t1.start();
        // 让子线程先运行一点业务
        Thread.sleep(100);
        // 终止子线程
        t1.interrupt();
        // 等待一段时间,确保子线程“执行完”
        Thread.sleep(3000);
        System.out.println("主线程执行完成");

        // 问题二:破坏了原子逻辑
        MyThread myThread = new MyThread();
        Thread t2 = new Thread(myThread);
        // 开启线程
        t2.start();
        for (int i = 0; i < 10; i++) {
            Thread t = new Thread(myThread);
            t.start();
        }
        // 结束线程
        t2.interrupt();
    }

    /**
     * 自定义原子测试线程
     */
    static class MyThread implements Runnable {
        // 计数器
        int num = 0;

        @Override
        public void run() {
            // 同步代码块,保证原子操作
            synchronized (MyThread.class) {
                // 自增
                num++;
                try {
                    // 线程休眠 0.1 秒
                    Thread.sleep(100);
                } catch (InterruptedException e) {
                    System.out.println(e.getMessage());
                }
                // 自减
                num--;
                System.out.println(Thread.currentThread().getName() + " | num=" + num);
            }
        }
    }
}

以上程序的执行结果为:

子线程开始执行

子线程的重要业务方法

主线程执行完成

sleep interrupted

Thread-1 | num=0

Thread-9 | num=0

Thread-10 | num=0

Thread-7 | num=0

Thread-6 | num=0

Thread-5 | num=0

Thread-4 | num=0

Thread-2 | num=0

Thread-3 | num=0

Thread-11 | num=0

Thread-8 | num=0

可以看出以上的执行都符合我们的预期,这才是正确的终止线程的方式。

总结

本文我们讲了线程的三种终止方式,自定义退出标识的方式、使用 stop() 的方式或 interrupt() 的方式。其中 stop() 的方式会导致程序的完整性和原子性被破坏的问题,并且此方法被 JDK 标识为过期方法,不建议使用,而 interrupt() 方法无疑是最适合我们的终止线程的方式。

更多精彩内容,请关注微信公众号「Java中文社群」

Java中文社群公众号二维码