Future类的使用和原理

十月 29, 2020

java异步接口Future和Callable

Future是Concurrent包提供的一种异步得到结果的接口。
Future接口：

public interface Future<V> {
    // 取消当前的计算
    boolean cancel(boolean mayInterruptIfRunning);
    // 计算是否被取消
    boolean isCancelled();
    // 计算是否已经结束
    boolean isDone();
    // 得到计算的结果
    V get() throws InterruptedException, ExecutionException;
    // 带有超时时间的get方法
    V get(long timeout, TimeUnit unit)
        throws InterruptedException, ExecutionException, TimeoutException;
}

Future里面的任务，必须实现Callable接口。
Callable接口：

public interface Callable<V> {
    // 要实现具体的计算逻辑
    V call() throws Exception;
}

state状态
当前计算的情况有下面这些状态

private volatile int state;
private static final int NEW          = 0;//初始
private static final int COMPLETING   = 1;//任务已经执行完或者出错，准备赋值
private static final int NORMAL       = 2;//任务正常执行完，并且已经赋值完
private static final int EXCEPTIONAL  = 3;//任务失败，把异常赋值回去
private static final int CANCELLED    = 4;//取消
private static final int INTERRUPTING = 5;//准备中断计算过程
private static final int INTERRUPTED  = 6;//对计算进行中断

下面以FutureTask类为例，叙述一遍Future的使用方法和原理。

一、使用FutureTask

FutureTask的实现接口：
在这里插入图片描述

测试代码：

public class TestMain {
    public static void main(String[] args) {
        long start = System.currentTimeMillis();
        Callable<Clothes> callable = new Callable<Clothes>() {
            @Override
            public Clothes call() throws Exception {
                Clothes clothes = new Clothes();
                clothes.washClothes();
                return clothes;
            }
        };
        FutureTask<Clothes> futureTask = new FutureTask<Clothes>(callable);
        new Thread(futureTask).start();
        Dishes dishes = new Dishes();
        dishes.washDishes();
        Clothes clothes = null;
        try {
            clothes = futureTask.get();
        } catch (InterruptedException e) {
            e.printStackTrace();
        } catch (ExecutionException e) {
            e.printStackTrace();
        }
        if (clothes != null) {
            System.out.println("===========over===========");
        }
        long end = System.currentTimeMillis();
        System.out.println("all tasks cost "+(end - start)+" millis");
        /*运行结果：
        start to wash dishes
        machine starts to wash cloths
        all dishes are cleaned!!!
        all clothes are cleaned!!!
        ===========over===========
        all tasks cost 3003 millis
        */
    }
    static class Clothes {
        public void washClothes() {
            System.out.println("machine starts to wash cloths");
            try {
                Thread.sleep(3000);
            } catch (InterruptedException e) {
                e.printStackTrace();
            }
            System.out.println("all clothes are cleaned!!!");
        }
    }
    static class Dishes {
        public void washDishes() {
            System.out.println("start to wash dishes");
            try {
                Thread.sleep(2000);
            } catch (InterruptedException e) {
                e.printStackTrace();
            }
            System.out.println("all dishes are cleaned!!!");
        }
    }
}

由结果可看出Future的方法是异步的，所以总耗时是3s，如果同步的话应该是5s。
在代码的最开头，我们new了一个Callable对象，Callable对象会被当做任务丢到FutureTask里面执行。
启动线程就可以开始执行任务了。
下面具体说一下Future帮我们干了什么，为什么他是异步的。

二、 FutureTask的run()方法

public void run() {
        if (state != NEW ||
            !UNSAFE.compareAndSwapObject(this, runnerOffset,
                                         null, Thread.currentThread()))
            return;
        try {
            Callable<V> c = callable;// 全局变量，构造函数赋值
            if (c != null && state == NEW) {
                V result;
                boolean ran;
                try {
                    result = c.call();// callable的返回值
                    ran = true;// Callable计算已经结束了
                } catch (Throwable ex) {
                    result = null;
                    ran = false;
                    setException(ex);
                }
                if (ran)
                    set(result);
            }
        } finally {
            // runner must be non-null until state is settled to
            // prevent concurrent calls to run()
            runner = null;
            // state must be re-read after nulling runner to prevent
            // leaked interrupts
            int s = state;
            if (s >= INTERRUPTING)
                handlePossibleCancellationInterrupt(s);
        }
    }

run方法很直接，直接调用Callable的call方法获取值，获取到值了，就令ran为true，然后调用set方法。

三、 set()方法

set方法会调用finishCompletion，

protected void set(V v) {
        if (UNSAFE.compareAndSwapInt(this, stateOffset, NEW, COMPLETING)) {
            outcome = v;
            // outcome是计算出来的结果，或者抛出的异常
            UNSAFE.putOrderedInt(this, stateOffset, NORMAL); // final state
            finishCompletion();
        }
    }

3.1、 finishCompletion()方法

WaitNode节点是Treiber stack的节点，就是一个排队等待的线程队列。

private void finishCompletion() {
        // assert state > COMPLETING;
        for (WaitNode q; (q = waiters) != null;) {
            if (UNSAFE.compareAndSwapObject(this, waitersOffset, q, null)) {
                for (;;) {
                    //唤醒队列中等待的线程
                    Thread t = q.thread;
                    if (t != null) {
                        q.thread = null;
                        LockSupport.unpark(t);//唤醒线程，加锁在get里面，如果没有加锁，正常向下执行
                    }
                    WaitNode next = q.next;
                    if (next == null)
                        break;
                    q.next = null; // unlink to help gc
                    q = next;
                }
                break;
            }
        }

        done();// protected方法，钩子函数，留给程序员自己实现

        callable = null;        // to reduce footprint
    }

四、 get()方法

public V get() throws InterruptedException, ExecutionException {
        int s = state;
        if (s <= COMPLETING)
        //计算没完成，进入awaitDone
            s = awaitDone(false, 0L);
        return report(s);
    }

4.1 awaitDone()

awaitDone用死循环等待结果，也就是说会阻塞在这里。

private int awaitDone(boolean timed, long nanos)
        throws InterruptedException {
        final long deadline = timed ? System.nanoTime() + nanos : 0L;
        WaitNode q = null;
        boolean queued = false;
        for (;;) {
            //死循环等待
            if (Thread.interrupted()) {
                removeWaiter(q);
                throw new InterruptedException();
            }

            int s = state;
            if (s > COMPLETING) {
                // 计算完成了就返回
                if (q != null)
                    q.thread = null;
                return s;
            }
            // COMPLETING是一个很短暂的状态，调用Thread.yield期望让出时间片，之后重试循环
            else if (s == COMPLETING) // cannot time out yet
                Thread.yield();
            else if (q == null)
                q = new WaitNode();
            else if (!queued)
            /*  当前节点未入栈
             *  这是Treiber Stack算法入栈的逻辑。
             *  Treiber Stack是一个基于CAS的无锁并发栈实现,
             *  更多可以参考https://en.wikipedia.org/wiki/Treiber_Stack
             */
                queued = UNSAFE.compareAndSwapObject(this, waitersOffset,
                                                     q.next = waiters, q);
            else if (timed) {
                nanos = deadline - System.nanoTime();
                if (nanos <= 0L) {
                    // 超时，移除节点
                    removeWaiter(q);
                    return state;
                }
                LockSupport.parkNanos(this, nanos);
            }
            else
                LockSupport.park(this);
        }
    }

4.2 report()

report把结果赋值回去

private V report(int s) throws ExecutionException {
        Object x = outcome;
        if (s == NORMAL)//正常得到了结果
            return (V)x;
        if (s >= CANCELLED)
            throw new CancellationException();
        throw new ExecutionException((Throwable)x);
    }

总结

FutureTask在底层开了一个死循环用于等待结果，当线程得到结果时，跳出循环，借此实现的异步操作。注意，我们一般在使用Future的时候，都不会用本文中new线程的方式，而是采用连接池中的线程。

查看评论