ThreadPoolExecutor with corePoolSize 0 should not

问题:

I was going through Java Concurrency In Practice and got stuck at the 8.3.1 Thread creation and teardown topic. The following footnote warns about keeping corePoolSize to zero.

Developers are sometimes tempted to set the core size to zero so that the worker threads will eventually be torn down and therefore won’t prevent the JVM from exiting, but this can cause some strange-seeming behavior in thread pools that don’t use a SynchronousQueue for their work queue (as newCachedThreadPool does). If the pool is already at the core size, ThreadPoolExecutor creates a new thread only if the work queue is full. So tasks submitted to a thread pool with a work queue that has any capacity and a core size of zero will not execute until the queue fills up, which is usually not what is desired.

So to verify this I wrote this program which does not work as stated above.

    final int corePoolSize = 0;
    ThreadPoolExecutor tp = new ThreadPoolExecutor(corePoolSize, 1, 5, TimeUnit.SECONDS,
            new LinkedBlockingQueue<>());

    // If the pool is already at the core size
    if (tp.getPoolSize() == corePoolSize) {
        ExecutorService ex = tp;

        // So tasks submitted to a thread pool with a work queue that has any capacity
        // and a core size of zero will not execute until the queue fills up.
        // So, this should not execute until queue fills up.
        ex.execute(() -> System.out.println("Hello"));
    }

Output: Hello

So, does the behavior of the program suggest that ThreadPoolExecutor creates at least one thread if a task is submitted irrespective of corePoolSize=0. If yes, then what is the warning about in the text book.

EDIT: Tested the code in jdk1.5.0_22 upon @S.K.'s suggestion with following change:

ThreadPoolExecutor tp = new ThreadPoolExecutor(corePoolSize, 1, 5, TimeUnit.SECONDS,
                new LinkedBlockingQueue<Runnable>(1));//Queue size is set to 1.

But with this change, the program terminates without printing any output.

So am I misinterpreting these statements from the book?

EDIT (@sjlee): It's hard to add code in the comment, so I'll add it as an edit here... Can you try out this modification and run it against both the latest JDK and JDK 1.5?

final int corePoolSize = 0;
ThreadPoolExecutor tp = new ThreadPoolExecutor(corePoolSize, 1, 5, TimeUnit.SECONDS, new LinkedBlockingQueue<>());

// If the pool is already at the core size
if (tp.getPoolSize() == corePoolSize) {
    ExecutorService ex = tp;

    // So tasks submitted to a thread pool with a work queue that has any capacity
    // and a core size of zero will not execute until the queue fills up.
    // So, this should not execute until queue fills up.
    ex.execute(() -> System.out.println("Hello"));
}
tp.shutdown();
if (tp.awaitTermination(1, TimeUnit.SECONDS)) {
    System.out.println("thread pool shut down. exiting.");
} else {
    System.out.println("shutdown timed out. exiting.");
}

回答1:

While running this program in jdk 1.5,1.6,1.7 and 1.8, I found different implementations of ThreadPoolExecutor#execute(Runnable) in 1.5,1.6 and 1.7+. Here's what I found:

JDK 1.5 implementation

 //Here poolSize is the number of core threads running.

 public void execute(Runnable command) {
    if (command == null)
        throw new NullPointerException();
    for (;;) {
        if (runState != RUNNING) {
            reject(command);
            return;
        }
        if (poolSize < corePoolSize && addIfUnderCorePoolSize(command))
            return;
        if (workQueue.offer(command))
            return;
        Runnable r = addIfUnderMaximumPoolSize(command);
        if (r == command)
            return;
        if (r == null) {
            reject(command);
            return;
        }
        // else retry
    }
}

This implementation does not create a thread when corePoolSize is 0, therefore the supplied task does not execute.

JDK 1.6 implementation

//Here poolSize is the number of core threads running.

  public void execute(Runnable command) {
    if (command == null)
        throw new NullPointerException();
    if (poolSize >= corePoolSize || !addIfUnderCorePoolSize(command)) {
        if (runState == RUNNING && workQueue.offer(command)) {
            if (runState != RUNNING || poolSize == 0)
                ensureQueuedTaskHandled(command);
        }
        else if (!addIfUnderMaximumPoolSize(command))
            reject(command); // is shutdown or saturated
    }
}

JDK 1.6 creates a new thread even if the corePoolSize is 0.

JDK 1.7+ implementation(Similar to JDK 1.6 but with better locks and state checks)

    public void execute(Runnable command) {
    if (command == null)
        throw new NullPointerException();
    /*
     * Proceed in 3 steps:
     *
     * 1. If fewer than corePoolSize threads are running, try to
     * start a new thread with the given command as its first
     * task.  The call to addWorker atomically checks runState and
     * workerCount, and so prevents false alarms that would add
     * threads when it shouldn't, by returning false.
     *
     * 2. If a task can be successfully queued, then we still need
     * to double-check whether we should have added a thread
     * (because existing ones died since last checking) or that
     * the pool shut down since entry into this method. So we
     * recheck state and if necessary roll back the enqueuing if
     * stopped, or start a new thread if there are none.
     *
     * 3. If we cannot queue task, then we try to add a new
     * thread.  If it fails, we know we are shut down or saturated
     * and so reject the task.
     */
    int c = ctl.get();
    if (workerCountOf(c) < corePoolSize) {
        if (addWorker(command, true))
            return;
        c = ctl.get();
    }
    if (isRunning(c) && workQueue.offer(command)) {
        int recheck = ctl.get();
        if (! isRunning(recheck) && remove(command))
            reject(command);
        else if (workerCountOf(recheck) == 0)
            addWorker(null, false);
    }
    else if (!addWorker(command, false))
        reject(command);
}

JDK 1.7 too creates a new thread even if the corePoolSize is 0.

So, it seems that corePoolSize=0 is a special case in each versions of JDK 1.5 and JDK 1.6+.

But it is strange that the book's explanation doesn't match any of the program results.

回答2:

Seems like it was a bug with older java versions but it doesn't exist now in Java 1.8.

According to the Java 1.8 documentation from ThreadPoolExecutor.execute():

     /*
     * Proceed in 3 steps:
     *
     * 1. If fewer than corePoolSize threads are running, try to
     * start a new thread with the given command as its first
     * task.  The call to addWorker atomically checks runState and
     * workerCount, and so prevents false alarms that would add
     * threads when it shouldn't, by returning false.
     *
     * 2. If a task can be successfully queued, then we still need
     * to double-check whether we should have added a thread
     * (because existing ones died since last checking) or that
     * the pool shut down since entry into this method. So we
     * recheck state and if necessary roll back the enqueuing if
     * stopped, or start a new thread if there are none.
     * ....
     */

In the second point, there is a recheck after adding a worker to the queue that if instead of queuing the task, a new thread can be started, than rollback the enqueuing and start a new thread.

This is what is happening. During first check the task is queued but during recheck, a new thread is started which executes your task.