ExecutorService, Callable, Future realize multi-threading principle analysis with returned results

Original / Zhu Jiqian

In a concurrent multi-thread scenario, there is a need to obtain the asynchronous execution results of each thread. In this case, it can be achieved through the ExecutorService thread pool combined with Callable and Future.

Let's start with a simple example -

public class ExecutorTest {
    public static void main(String[] args) throws ExecutionException, InterruptedException {
        ExecutorService executor = Executors.newSingleThreadExecutor();
        Callable callable = new MyCallable();
        Future future = executor.submit(callable);
        System.out.println("打印线程池返回值：" + future.get());
    }
}

class MyCallable implements Callable<String>{
    @Override
    public String call() throws Exception {
        return "测试返回值";
    }
}
复制代码

After the execution is complete, the following results will be printed:

打印线程池返回值：测试返回值
复制代码

It can be seen that after the thread pool executes the asynchronous thread task, we can obtain the return value of the asynchronous thread.

So, how do ExecutorService, Callable, and Future implement multithreading that returns results?

First, we need to create a class that implements the functional interface Callable. The Callable interface only defines a call method modified by generics, which means that what type of value needs to be returned can be defined by the specific implementation class——

@FunctionalInterface
public interface Callable<V> {
    V call() throws Exception;
}
复制代码

Therefore, I customized a class that implements the Callable interface, which overrides the call method. What kind of result we want to return when executing multi-threading can be defined in the overridden call method.

class MyCallable implements Callable<String>{
    @Override
    public String call() throws Exception {
        return "测试返回值";
    }
}
复制代码

In the custom MyCallable class, I set a very simple String return value "test return value" in the call method, which means that I hope that when the thread pool finishes executing the asynchronous thread task, I can return "test return value" "Give me this string.

Next, we can create an object of the MyCallable class, and then throw it into the thread pool through executor.submit(callable). The thread pool will use idle threads to help us perform an asynchronous thread task.

ExecutorService executor = Executors.newSingleThreadExecutor();
Callable callable = new MyCallable();
Future future = executor.submit(callable);
复制代码

It is worth noting that if you need to achieve the effect of obtaining the return value of the thread, you can only execute it through executor.submit(callable), but not through executor.execute(Runnable command), because executor.execute(Runnable command) can only pass Enter objects that implement the Runnable interface, but such objects do not have the ability to return thread effects.

进入到executor.submit(callable)底层，具体实现在AbstractExecutorService类中。可以看到，执行到submit方法内部时，会将我们传进来的new MyCallable()对象作为参数传入到newTaskFor(task)方法里——

public <T> Future<T> submit(Callable<T> task) {
    if (task == null) throw new NullPointerException();
    RunnableFuture<T> ftask = newTaskFor(task);
    execute(ftask);
    return ftask;
}
复制代码

这个newTaskFor(task)方法内部具体实现，是将new MyCallable()对象传入构造器中，生成了一个FutureTask对象。

protected <T> RunnableFuture<T> newTaskFor(Callable<T> callable) {
    return new FutureTask<T>(callable);
}
复制代码

这个FutureTask对象实现RunableFuture接口，这个RunableFuture接口又继承了Runnable，说明FutureTask类内部会实现一个run方法，然后本身就可以当做一个Runnable线程任务，借助线程Thread(new FutureTask(.....)).start()方式开启一个新线程，去异步执行其内部实现的run方法逻辑。

public class FutureTask<V> implements RunnableFuture<V>{.....}


public interface RunnableFuture<V> extends Runnable, Future<V> {
    /**
     * Sets this Future to the result of its computation
     * unless it has been cancelled.
     */
    void run();
}
复制代码

分析到这里，可以知道FutureTask的核心方法一定是run方法，线程执行start方法后，最后会去调用FutureTask的run方法。在讲解这个run方法前，我们先去看一下创建FutureTask的初始化构造方法底层逻辑new FutureTask(callable) ——

public class FutureTask<V> implements RunnableFuture<V> {
    
private Callable<V> callable;

......//省略其余源码
    
public FutureTask(Callable<V> callable) {
    if (callable == null)
        throw new NullPointerException();
    //通过构造方法初始化Callable<V> callable赋值
    this.callable = callable;
    this.state = NEW;       // ensure visibility of callable
}

......//省略其余源码
}
复制代码

可以看到，FutureTask(Callable callable)构造器，主要是将我们先前创建的new MyCallable()对象传进来，赋值给FutureTask内部定义的Callable callable引用，实现子类对象指向父类引用。这一点很关键，这就意味着，在初始化创建FutureTask对象后，我们是可以通过callable.call()来调用我们自定义设置可以返回“测试返回值”的call方法，这不就是我们希望在异步线程执行完后能够返回的值吗？

我们不妨猜测一下整体返数主流程，在Thread(new FutureTask(.....)).start()开启一个线程后，当线程获得了CPU时间片，就会去执行FutureTask对象里的run方法，这时run方法里可以通过callable.call()调用到我们自定义的MyCallable#call()方法，进而得到方法返回值 “测试返回值”——到这一步，只需要将这个返回值赋值给FutureTask里某个定义的对象属性，那么，在主线程在通过获取FutureTask里被赋值的X对象属性值，不就可以拿到返回字符串值 “测试返回值”了吗？

实现上，主体流程确实是这样，只不过忽略了一些细节而已。

接下来，让我们看一下FutureTask的run方法——

public void run() {
    //如果状态不是NEW或者设置runner为当前线程时，说明FutureTask任务已经取消，无法继续执行
    if (state != NEW ||
        !UNSAFE.compareAndSwapObject(this, runnerOffset,
                                     null, Thread.currentThread()))
        return;
    try {
        //在该文中，callable被赋值为指向我们定义的new MyCallable()对象引用
        Callable<V> c = callable;
        if (c != null && state == NEW) {
            V result;
            boolean ran;
            try {
                //c.call最后会调用new MyCallable()的call()方法，得到字符串返回值“测试返回值”给result
                result = c.call();
                ran = true;
            } catch (Throwable ex) {
                result = null;
                ran = false;
                setException(ex);
            }
            //正常执行完c.call()方法时，ran值为true,说明会执行set(result)方法。
            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);
    }
}
复制代码

根据以上源码简单分析，可以看到run方法当中，最终确实会执行new MyCallable()的call()方法，得到字符串返回值“测试返回值”给result，然后执行set(result)方法，根据set方法名就不难猜出，这是一个会赋值给某个字段的方法。

这里分析会忽略一些状态值的讲解，这块会包括线程的取消、终止等内容，后面我会出一片专门针对FutureTask源码分析的文章再介绍，本文主要还是介绍异步线程返回结果的主要原理。

沿着以上分析，追踪至set(result)方法里——

protected void set(V v) {
    //通过CAS原子操作，将运行的线程设置为COMPLETING，说明线程已经执行完成中
    if (UNSAFE.compareAndSwapInt(this, stateOffset, NEW, COMPLETING)) {
        //若CAS原子比较赋值成功，说明线程可以被正常执行完成的话，然后将result结果值赋值给outcome
        outcome = v;
        //线程正常完成结束
        UNSAFE.putOrderedInt(this, stateOffset, NORMAL); // final state
        finishCompletion();
    }
}
复制代码

这个方法的主要是，若该线程执行能够正常完成话，就将得到的返回值赋值给outcome，这个outcome是FutureTask的一个Object变量——

private Object outcome;
复制代码

至此，就完成了流程的这一步——

最后，就是执行主线程的根据ftask.get()获取执行完成的值，这个get可以设置超时时间，例如 ftask.get(2,TimeUnit.SECONDS)表示超过2秒还没有获取到线程返回值的话，就直接结束该get方法，继续主线程往下执行。

System.out.println("打印线程池返回值：" + ftask.get(2,TimeUnit.SECONDS));
复制代码

进入到get方法，可以看到当状态在s <= COMPLETING时，表示任务还没有执行完，就会去执行awaitDone(false, 0L)方法，这个方法表示，将一直做死循环等待线程执行完成，才会跳出等待循环继续往下走。若设置了超时时间，例如ftask.get(2,TimeUnit.SECONDS))，就会在awaitDone方法循环至2秒，在2秒内发现线程状态被设置为正常完成时，就会跳出循环，若2秒后线程没有执行完成，也会强制跳出循环了，但这种情况将无法获取到线程结果值。

public V get() throws InterruptedException, ExecutionException {
    int s = state;
    if (s <= COMPLETING)
        //循环等待线程执行状态
        s = awaitDone(false, 0L);
    return report(s);
}
复制代码

最后就是report(s)方法，可以看到outcome值最终赋值给Object x，若s==NORMAL表示线程任务已经正常完成结束，就可以根据我们定义的类型进行泛型转换返回，我们定义的是String字符串类型，故而会返回字符串值，也就是 “测试返回值”。

private V report(int s) throws ExecutionException {
    Object x = outcome;
    if (s == NORMAL)
        //返回线程任务执行结果
        return (V)x;
    if (s >= CANCELLED)
        throw new CancellationException();
    throw new ExecutionException((Throwable)x);
}
复制代码

你看，最后就能获取到了异步线程执行的结果返回给main主线程——

以上就是执行线程任务run方法后，如何将线程任务结果返回给主线程，其实，还少一个地方补充，就是如何将FutureTask任务丢给线程执行，我们这里用到了线程池， 但是execute(ftask)底层同样是使用一个了线程通过执行start方法开启一个线程，这个新运行的线程最终会执行FutureTask的run方法。

public <T> Future<T> submit(Callable<T> task) {
    if (task == null) throw new NullPointerException();
    RunnableFuture<T> ftask = newTaskFor(task);
    execute(ftask);
    return ftask;
}
复制代码

可以简单优化下，直接用一个线程演示该案例，这样看着更好理解些，当时，生产上是不会有这样直接用一个线程来执行的，更多是通过原生线程池——

public static void main(String[] args) throws Exception{
    Callable callable = new MyCallable();
    RunnableFuture<String> ftask = new FutureTask<String>(callable);
    new Thread(ftask).start();
    System.out.println("打印线程池返回值：" + ftask.get());
}
复制代码

作者：朱季谦