21 |
|
|
22 |
|
/** |
23 |
|
* An {@link ExecutorService} for running {@link ForkJoinTask}s. |
24 |
< |
* A ForkJoinPool provides the entry point for submissions from |
25 |
< |
* non-ForkJoinTasks, as well as management and monitoring operations. |
26 |
< |
* Normally a single ForkJoinPool is used for a large number of |
27 |
< |
* submitted tasks. Otherwise, use would not usually outweigh the |
28 |
< |
* construction and bookkeeping overhead of creating a large set of |
29 |
< |
* threads. |
24 |
> |
* A {@code ForkJoinPool} provides the entry point for submissions |
25 |
> |
* from non-{@code ForkJoinTask}s, as well as management and |
26 |
> |
* monitoring operations. |
27 |
|
* |
28 |
< |
* <p>ForkJoinPools differ from other kinds of Executors mainly in |
29 |
< |
* that they provide <em>work-stealing</em>: all threads in the pool |
30 |
< |
* attempt to find and execute subtasks created by other active tasks |
31 |
< |
* (eventually blocking if none exist). This makes them efficient when |
32 |
< |
* most tasks spawn other subtasks (as do most ForkJoinTasks), as well |
33 |
< |
* as the mixed execution of some plain Runnable- or Callable- based |
34 |
< |
* activities along with ForkJoinTasks. When setting {@linkplain |
35 |
< |
* #setAsyncMode async mode}, a ForkJoinPool may also be appropriate |
36 |
< |
* for use with fine-grained tasks that are never joined. Otherwise, |
37 |
< |
* other ExecutorService implementations are typically more |
38 |
< |
* appropriate choices. |
28 |
> |
* <p>A {@code ForkJoinPool} differs from other kinds of {@link |
29 |
> |
* ExecutorService} mainly by virtue of employing |
30 |
> |
* <em>work-stealing</em>: all threads in the pool attempt to find and |
31 |
> |
* execute subtasks created by other active tasks (eventually blocking |
32 |
> |
* waiting for work if none exist). This enables efficient processing |
33 |
> |
* when most tasks spawn other subtasks (as do most {@code |
34 |
> |
* ForkJoinTask}s). A {@code ForkJoinPool} may also be used for mixed |
35 |
> |
* execution of some plain {@code Runnable}- or {@code Callable}- |
36 |
> |
* based activities along with {@code ForkJoinTask}s. When setting |
37 |
> |
* {@linkplain #setAsyncMode async mode}, a {@code ForkJoinPool} may |
38 |
> |
* also be appropriate for use with fine-grained tasks of any form |
39 |
> |
* that are never joined. Otherwise, other {@code ExecutorService} |
40 |
> |
* implementations are typically more appropriate choices. |
41 |
|
* |
42 |
< |
* <p>A ForkJoinPool may be constructed with a given parallelism level |
43 |
< |
* (target pool size), which it attempts to maintain by dynamically |
44 |
< |
* adding, suspending, or resuming threads, even if some tasks are |
45 |
< |
* waiting to join others. However, no such adjustments are performed |
46 |
< |
* in the face of blocked IO or other unmanaged synchronization. The |
47 |
< |
* nested {@link ManagedBlocker} interface enables extension of |
48 |
< |
* the kinds of synchronization accommodated. The target parallelism |
49 |
< |
* level may also be changed dynamically ({@link #setParallelism}) |
50 |
< |
* and thread construction can be limited using methods |
51 |
< |
* {@link #setMaximumPoolSize} and/or |
52 |
< |
* {@link #setMaintainsParallelism}. |
42 |
> |
* <p>A {@code ForkJoinPool} is constructed with a given target |
43 |
> |
* parallelism level; by default, equal to the number of available |
44 |
> |
* processors. Unless configured otherwise via {@link |
45 |
> |
* #setMaintainsParallelism}, the pool attempts to maintain this |
46 |
> |
* number of active (or available) threads by dynamically adding, |
47 |
> |
* suspending, or resuming internal worker threads, even if some tasks |
48 |
> |
* are stalled waiting to join others. However, no such adjustments |
49 |
> |
* are performed in the face of blocked IO or other unmanaged |
50 |
> |
* synchronization. The nested {@link ManagedBlocker} interface |
51 |
> |
* enables extension of the kinds of synchronization accommodated. |
52 |
> |
* The target parallelism level may also be changed dynamically |
53 |
> |
* ({@link #setParallelism}). The total number of threads may be |
54 |
> |
* limited using method {@link #setMaximumPoolSize}, in which case it |
55 |
> |
* may become possible for the activities of a pool to stall due to |
56 |
> |
* the lack of available threads to process new tasks. |
57 |
|
* |
58 |
|
* <p>In addition to execution and lifecycle control methods, this |
59 |
|
* class provides status check methods (for example |
62 |
|
* {@link #toString} returns indications of pool state in a |
63 |
|
* convenient form for informal monitoring. |
64 |
|
* |
65 |
+ |
* <p><b>Sample Usage.</b> Normally a single {@code ForkJoinPool} is |
66 |
+ |
* used for all parallel task execution in a program or subsystem. |
67 |
+ |
* Otherwise, use would not usually outweigh the construction and |
68 |
+ |
* bookkeeping overhead of creating a large set of threads. For |
69 |
+ |
* example, a common pool could be used for the {@code SortTasks} |
70 |
+ |
* illustrated in {@link RecursiveAction}. Because {@code |
71 |
+ |
* ForkJoinPool} uses threads in {@linkplain java.lang.Thread#isDaemon |
72 |
+ |
* daemon} mode, there is typically no need to explicitly {@link |
73 |
+ |
* #shutdown} such a pool upon program exit. |
74 |
+ |
* |
75 |
+ |
* <pre> |
76 |
+ |
* static final ForkJoinPool mainPool = new ForkJoinPool(); |
77 |
+ |
* ... |
78 |
+ |
* public void sort(long[] array) { |
79 |
+ |
* mainPool.invoke(new SortTask(array, 0, array.length)); |
80 |
+ |
* } |
81 |
+ |
* </pre> |
82 |
+ |
* |
83 |
|
* <p><b>Implementation notes</b>: This implementation restricts the |
84 |
|
* maximum number of running threads to 32767. Attempts to create |
85 |
< |
* pools with greater than the maximum result in |
86 |
< |
* IllegalArgumentExceptions. |
85 |
> |
* pools with greater than the maximum number result in |
86 |
> |
* {@code IllegalArgumentException}. |
87 |
> |
* |
88 |
> |
* <p>This implementation rejects submitted tasks (that is, by throwing |
89 |
> |
* {@link RejectedExecutionException}) only when the pool is shut down. |
90 |
|
* |
91 |
|
* @since 1.7 |
92 |
|
* @author Doug Lea |
105 |
|
private static final int MAX_THREADS = 0x7FFF; |
106 |
|
|
107 |
|
/** |
108 |
< |
* Factory for creating new ForkJoinWorkerThreads. A |
109 |
< |
* ForkJoinWorkerThreadFactory must be defined and used for |
110 |
< |
* ForkJoinWorkerThread subclasses that extend base functionality |
111 |
< |
* or initialize threads with different contexts. |
108 |
> |
* Factory for creating new {@link ForkJoinWorkerThread}s. |
109 |
> |
* A {@code ForkJoinWorkerThreadFactory} must be defined and used |
110 |
> |
* for {@code ForkJoinWorkerThread} subclasses that extend base |
111 |
> |
* functionality or initialize threads with different contexts. |
112 |
|
*/ |
113 |
|
public static interface ForkJoinWorkerThreadFactory { |
114 |
|
/** |
115 |
|
* Returns a new worker thread operating in the given pool. |
116 |
|
* |
117 |
|
* @param pool the pool this thread works in |
118 |
< |
* @throws NullPointerException if pool is null |
118 |
> |
* @throws NullPointerException if the pool is null |
119 |
|
*/ |
120 |
|
public ForkJoinWorkerThread newThread(ForkJoinPool pool); |
121 |
|
} |
366 |
|
// Constructors |
367 |
|
|
368 |
|
/** |
369 |
< |
* Creates a ForkJoinPool with a pool size equal to the number of |
370 |
< |
* processors available on the system, using the default |
371 |
< |
* ForkJoinWorkerThreadFactory. |
369 |
> |
* Creates a {@code ForkJoinPool} with parallelism equal to {@link |
370 |
> |
* java.lang.Runtime#availableProcessors}, and using the {@linkplain |
371 |
> |
* #defaultForkJoinWorkerThreadFactory default thread factory}. |
372 |
|
* |
373 |
|
* @throws SecurityException if a security manager exists and |
374 |
|
* the caller is not permitted to modify threads |
381 |
|
} |
382 |
|
|
383 |
|
/** |
384 |
< |
* Creates a ForkJoinPool with the indicated parallelism level |
385 |
< |
* threads and using the default ForkJoinWorkerThreadFactory. |
384 |
> |
* Creates a {@code ForkJoinPool} with the indicated parallelism |
385 |
> |
* level and using the {@linkplain |
386 |
> |
* #defaultForkJoinWorkerThreadFactory default thread factory}. |
387 |
|
* |
388 |
< |
* @param parallelism the number of worker threads |
388 |
> |
* @param parallelism the parallelism level |
389 |
|
* @throws IllegalArgumentException if parallelism less than or |
390 |
< |
* equal to zero |
390 |
> |
* equal to zero, or greater than implementation limit |
391 |
|
* @throws SecurityException if a security manager exists and |
392 |
|
* the caller is not permitted to modify threads |
393 |
|
* because it does not hold {@link |
398 |
|
} |
399 |
|
|
400 |
|
/** |
401 |
< |
* Creates a ForkJoinPool with parallelism equal to the number of |
402 |
< |
* processors available on the system and using the given |
403 |
< |
* ForkJoinWorkerThreadFactory. |
401 |
> |
* Creates a {@code ForkJoinPool} with parallelism equal to {@link |
402 |
> |
* java.lang.Runtime#availableProcessors}, and using the given |
403 |
> |
* thread factory. |
404 |
|
* |
405 |
|
* @param factory the factory for creating new threads |
406 |
< |
* @throws NullPointerException if factory is null |
406 |
> |
* @throws NullPointerException if the factory is null |
407 |
|
* @throws SecurityException if a security manager exists and |
408 |
|
* the caller is not permitted to modify threads |
409 |
|
* because it does not hold {@link |
414 |
|
} |
415 |
|
|
416 |
|
/** |
417 |
< |
* Creates a ForkJoinPool with the given parallelism and factory. |
417 |
> |
* Creates a {@code ForkJoinPool} with the given parallelism and |
418 |
> |
* thread factory. |
419 |
|
* |
420 |
< |
* @param parallelism the targeted number of worker threads |
420 |
> |
* @param parallelism the parallelism level |
421 |
|
* @param factory the factory for creating new threads |
422 |
|
* @throws IllegalArgumentException if parallelism less than or |
423 |
< |
* equal to zero, or greater than implementation limit |
424 |
< |
* @throws NullPointerException if factory is null |
423 |
> |
* equal to zero, or greater than implementation limit |
424 |
> |
* @throws NullPointerException if the factory is null |
425 |
|
* @throws SecurityException if a security manager exists and |
426 |
|
* the caller is not permitted to modify threads |
427 |
|
* because it does not hold {@link |
449 |
|
* Creates a new worker thread using factory. |
450 |
|
* |
451 |
|
* @param index the index to assign worker |
452 |
< |
* @return new worker, or null of factory failed |
452 |
> |
* @return new worker, or null if factory failed |
453 |
|
*/ |
454 |
|
private ForkJoinWorkerThread createWorker(int index) { |
455 |
|
Thread.UncaughtExceptionHandler h = ueh; |
470 |
|
* Currently requires size to be a power of two. |
471 |
|
*/ |
472 |
|
private static int arraySizeFor(int poolSize) { |
473 |
< |
return (poolSize <= 1) ? 1 : |
474 |
< |
(1 << (32 - Integer.numberOfLeadingZeros(poolSize-1))); |
473 |
> |
if (poolSize <= 1) |
474 |
> |
return 1; |
475 |
> |
// See Hackers Delight, sec 3.2 |
476 |
> |
int c = poolSize >= MAX_THREADS ? MAX_THREADS : (poolSize - 1); |
477 |
> |
c |= c >>> 1; |
478 |
> |
c |= c >>> 2; |
479 |
> |
c |= c >>> 4; |
480 |
> |
c |= c >>> 8; |
481 |
> |
c |= c >>> 16; |
482 |
> |
return c + 1; |
483 |
|
} |
484 |
|
|
485 |
|
/** |
526 |
|
ws = workers; |
527 |
|
if (ws == null) { |
528 |
|
int ps = parallelism; |
529 |
+ |
updateWorkerCount(ps); |
530 |
|
ws = ensureWorkerArrayCapacity(ps); |
531 |
|
for (int i = 0; i < ps; ++i) { |
532 |
|
ForkJoinWorkerThread w = createWorker(i); |
533 |
|
if (w != null) { |
534 |
|
ws[i] = w; |
535 |
|
w.start(); |
501 |
– |
updateWorkerCount(1); |
536 |
|
} |
537 |
+ |
else |
538 |
+ |
updateWorkerCount(-1); |
539 |
|
} |
540 |
|
} |
541 |
|
} finally { |
599 |
|
* |
600 |
|
* @param task the task |
601 |
|
* @return the task's result |
602 |
< |
* @throws NullPointerException if task is null |
603 |
< |
* @throws RejectedExecutionException if pool is shut down |
602 |
> |
* @throws NullPointerException if the task is null |
603 |
> |
* @throws RejectedExecutionException if the task cannot be |
604 |
> |
* scheduled for execution |
605 |
|
*/ |
606 |
|
public <T> T invoke(ForkJoinTask<T> task) { |
607 |
|
doSubmit(task); |
612 |
|
* Arranges for (asynchronous) execution of the given task. |
613 |
|
* |
614 |
|
* @param task the task |
615 |
< |
* @throws NullPointerException if task is null |
616 |
< |
* @throws RejectedExecutionException if pool is shut down |
615 |
> |
* @throws NullPointerException if the task is null |
616 |
> |
* @throws RejectedExecutionException if the task cannot be |
617 |
> |
* scheduled for execution |
618 |
|
*/ |
619 |
< |
public <T> void execute(ForkJoinTask<T> task) { |
619 |
> |
public void execute(ForkJoinTask<?> task) { |
620 |
|
doSubmit(task); |
621 |
|
} |
622 |
|
|
623 |
|
// AbstractExecutorService methods |
624 |
|
|
625 |
+ |
/** |
626 |
+ |
* @throws NullPointerException if the task is null |
627 |
+ |
* @throws RejectedExecutionException if the task cannot be |
628 |
+ |
* scheduled for execution |
629 |
+ |
*/ |
630 |
|
public void execute(Runnable task) { |
631 |
|
ForkJoinTask<?> job; |
632 |
|
if (task instanceof ForkJoinTask<?>) // avoid re-wrap |
633 |
|
job = (ForkJoinTask<?>) task; |
634 |
|
else |
635 |
< |
job = new AdaptedRunnable<Void>(task, null); |
635 |
> |
job = ForkJoinTask.adapt(task, null); |
636 |
|
doSubmit(job); |
637 |
|
} |
638 |
|
|
639 |
+ |
/** |
640 |
+ |
* @throws NullPointerException if the task is null |
641 |
+ |
* @throws RejectedExecutionException if the task cannot be |
642 |
+ |
* scheduled for execution |
643 |
+ |
*/ |
644 |
|
public <T> ForkJoinTask<T> submit(Callable<T> task) { |
645 |
< |
ForkJoinTask<T> job = new AdaptedCallable<T>(task); |
645 |
> |
ForkJoinTask<T> job = ForkJoinTask.adapt(task); |
646 |
|
doSubmit(job); |
647 |
|
return job; |
648 |
|
} |
649 |
|
|
650 |
+ |
/** |
651 |
+ |
* @throws NullPointerException if the task is null |
652 |
+ |
* @throws RejectedExecutionException if the task cannot be |
653 |
+ |
* scheduled for execution |
654 |
+ |
*/ |
655 |
|
public <T> ForkJoinTask<T> submit(Runnable task, T result) { |
656 |
< |
ForkJoinTask<T> job = new AdaptedRunnable<T>(task, result); |
656 |
> |
ForkJoinTask<T> job = ForkJoinTask.adapt(task, result); |
657 |
|
doSubmit(job); |
658 |
|
return job; |
659 |
|
} |
660 |
|
|
661 |
+ |
/** |
662 |
+ |
* @throws NullPointerException if the task is null |
663 |
+ |
* @throws RejectedExecutionException if the task cannot be |
664 |
+ |
* scheduled for execution |
665 |
+ |
*/ |
666 |
|
public ForkJoinTask<?> submit(Runnable task) { |
667 |
|
ForkJoinTask<?> job; |
668 |
|
if (task instanceof ForkJoinTask<?>) // avoid re-wrap |
669 |
|
job = (ForkJoinTask<?>) task; |
670 |
|
else |
671 |
< |
job = new AdaptedRunnable<Void>(task, null); |
671 |
> |
job = ForkJoinTask.adapt(task, null); |
672 |
|
doSubmit(job); |
673 |
|
return job; |
674 |
|
} |
678 |
|
* |
679 |
|
* @param task the task to submit |
680 |
|
* @return the task |
681 |
+ |
* @throws NullPointerException if the task is null |
682 |
|
* @throws RejectedExecutionException if the task cannot be |
683 |
|
* scheduled for execution |
625 |
– |
* @throws NullPointerException if the task is null |
684 |
|
*/ |
685 |
|
public <T> ForkJoinTask<T> submit(ForkJoinTask<T> task) { |
686 |
|
doSubmit(task); |
687 |
|
return task; |
688 |
|
} |
689 |
|
|
632 |
– |
/** |
633 |
– |
* Adaptor for Runnables. This implements RunnableFuture |
634 |
– |
* to be compliant with AbstractExecutorService constraints. |
635 |
– |
*/ |
636 |
– |
static final class AdaptedRunnable<T> extends ForkJoinTask<T> |
637 |
– |
implements RunnableFuture<T> { |
638 |
– |
final Runnable runnable; |
639 |
– |
final T resultOnCompletion; |
640 |
– |
T result; |
641 |
– |
AdaptedRunnable(Runnable runnable, T result) { |
642 |
– |
if (runnable == null) throw new NullPointerException(); |
643 |
– |
this.runnable = runnable; |
644 |
– |
this.resultOnCompletion = result; |
645 |
– |
} |
646 |
– |
public T getRawResult() { return result; } |
647 |
– |
public void setRawResult(T v) { result = v; } |
648 |
– |
public boolean exec() { |
649 |
– |
runnable.run(); |
650 |
– |
result = resultOnCompletion; |
651 |
– |
return true; |
652 |
– |
} |
653 |
– |
public void run() { invoke(); } |
654 |
– |
private static final long serialVersionUID = 5232453952276885070L; |
655 |
– |
} |
690 |
|
|
691 |
|
/** |
692 |
< |
* Adaptor for Callables |
692 |
> |
* @throws NullPointerException {@inheritDoc} |
693 |
> |
* @throws RejectedExecutionException {@inheritDoc} |
694 |
|
*/ |
660 |
– |
static final class AdaptedCallable<T> extends ForkJoinTask<T> |
661 |
– |
implements RunnableFuture<T> { |
662 |
– |
final Callable<T> callable; |
663 |
– |
T result; |
664 |
– |
AdaptedCallable(Callable<T> callable) { |
665 |
– |
if (callable == null) throw new NullPointerException(); |
666 |
– |
this.callable = callable; |
667 |
– |
} |
668 |
– |
public T getRawResult() { return result; } |
669 |
– |
public void setRawResult(T v) { result = v; } |
670 |
– |
public boolean exec() { |
671 |
– |
try { |
672 |
– |
result = callable.call(); |
673 |
– |
return true; |
674 |
– |
} catch (Error err) { |
675 |
– |
throw err; |
676 |
– |
} catch (RuntimeException rex) { |
677 |
– |
throw rex; |
678 |
– |
} catch (Exception ex) { |
679 |
– |
throw new RuntimeException(ex); |
680 |
– |
} |
681 |
– |
} |
682 |
– |
public void run() { invoke(); } |
683 |
– |
private static final long serialVersionUID = 2838392045355241008L; |
684 |
– |
} |
685 |
– |
|
695 |
|
public <T> List<Future<T>> invokeAll(Collection<? extends Callable<T>> tasks) { |
696 |
|
ArrayList<ForkJoinTask<T>> forkJoinTasks = |
697 |
|
new ArrayList<ForkJoinTask<T>>(tasks.size()); |
698 |
|
for (Callable<T> task : tasks) |
699 |
< |
forkJoinTasks.add(new AdaptedCallable<T>(task)); |
699 |
> |
forkJoinTasks.add(ForkJoinTask.adapt(task)); |
700 |
|
invoke(new InvokeAll<T>(forkJoinTasks)); |
701 |
|
|
702 |
|
@SuppressWarnings({"unchecked", "rawtypes"}) |
798 |
|
final ReentrantLock lock = this.workerLock; |
799 |
|
lock.lock(); |
800 |
|
try { |
801 |
< |
if (!isTerminating()) { |
801 |
> |
if (isProcessingTasks()) { |
802 |
|
int p = this.parallelism; |
803 |
|
this.parallelism = parallelism; |
804 |
< |
if (parallelism > p) |
805 |
< |
createAndStartAddedWorkers(); |
806 |
< |
else |
807 |
< |
trimSpares(); |
804 |
> |
if (workers != null) { |
805 |
> |
if (parallelism > p) |
806 |
> |
createAndStartAddedWorkers(); |
807 |
> |
else |
808 |
> |
trimSpares(); |
809 |
> |
} |
810 |
|
} |
811 |
|
} finally { |
812 |
|
lock.unlock(); |
815 |
|
} |
816 |
|
|
817 |
|
/** |
818 |
< |
* Returns the targeted number of worker threads in this pool. |
818 |
> |
* Returns the targeted parallelism level of this pool. |
819 |
|
* |
820 |
< |
* @return the targeted number of worker threads in this pool |
820 |
> |
* @return the targeted parallelism level of this pool |
821 |
|
*/ |
822 |
|
public int getParallelism() { |
823 |
|
return parallelism; |
837 |
|
|
838 |
|
/** |
839 |
|
* Returns the maximum number of threads allowed to exist in the |
840 |
< |
* pool, even if there are insufficient unblocked running threads. |
840 |
> |
* pool. Unless set using {@link #setMaximumPoolSize}, the |
841 |
> |
* maximum is an implementation-defined value designed only to |
842 |
> |
* prevent runaway growth. |
843 |
|
* |
844 |
|
* @return the maximum |
845 |
|
*/ |
849 |
|
|
850 |
|
/** |
851 |
|
* Sets the maximum number of threads allowed to exist in the |
852 |
< |
* pool, even if there are insufficient unblocked running threads. |
853 |
< |
* Setting this value has no effect on current pool size. It |
854 |
< |
* controls construction of new threads. |
852 |
> |
* pool. The given value should normally be greater than or equal |
853 |
> |
* to the {@link #getParallelism parallelism} level. Setting this |
854 |
> |
* value has no effect on current pool size. It controls |
855 |
> |
* construction of new threads. |
856 |
|
* |
857 |
< |
* @throws IllegalArgumentException if negative or greater then |
857 |
> |
* @throws IllegalArgumentException if negative or greater than |
858 |
|
* internal implementation limit |
859 |
|
*/ |
860 |
|
public void setMaximumPoolSize(int newMax) { |
1022 |
|
} |
1023 |
|
|
1024 |
|
/** |
1025 |
< |
* Returns an estimate of the number tasks submitted to this pool |
1026 |
< |
* that have not yet begun executing. This method takes time |
1025 |
> |
* Returns an estimate of the number of tasks submitted to this |
1026 |
> |
* pool that have not yet begun executing. This method takes time |
1027 |
|
* proportional to the number of submissions. |
1028 |
|
* |
1029 |
|
* @return the number of queued submissions |
1057 |
|
* Removes all available unexecuted submitted and forked tasks |
1058 |
|
* from scheduling queues and adds them to the given collection, |
1059 |
|
* without altering their execution status. These may include |
1060 |
< |
* artificially generated or wrapped tasks. This method is designed |
1061 |
< |
* to be invoked only when the pool is known to be |
1060 |
> |
* artificially generated or wrapped tasks. This method is |
1061 |
> |
* designed to be invoked only when the pool is known to be |
1062 |
|
* quiescent. Invocations at other times may not remove all |
1063 |
|
* tasks. A failure encountered while attempting to add elements |
1064 |
|
* to collection {@code c} may result in elements being in |
1110 |
|
} |
1111 |
|
|
1112 |
|
private static String runStateToString(int rs) { |
1113 |
< |
switch(rs) { |
1113 |
> |
switch (rs) { |
1114 |
|
case RUNNING: return "Running"; |
1115 |
|
case SHUTDOWN: return "Shutting down"; |
1116 |
|
case TERMINATING: return "Terminating"; |
1136 |
|
public void shutdown() { |
1137 |
|
checkPermission(); |
1138 |
|
transitionRunStateTo(SHUTDOWN); |
1139 |
< |
if (canTerminateOnShutdown(runControl)) |
1139 |
> |
if (canTerminateOnShutdown(runControl)) { |
1140 |
> |
if (workers == null) { // shutting down before workers created |
1141 |
> |
final ReentrantLock lock = this.workerLock; |
1142 |
> |
lock.lock(); |
1143 |
> |
try { |
1144 |
> |
if (workers == null) { |
1145 |
> |
terminate(); |
1146 |
> |
transitionRunStateTo(TERMINATED); |
1147 |
> |
termination.signalAll(); |
1148 |
> |
} |
1149 |
> |
} finally { |
1150 |
> |
lock.unlock(); |
1151 |
> |
} |
1152 |
> |
} |
1153 |
|
terminateOnShutdown(); |
1154 |
+ |
} |
1155 |
|
} |
1156 |
|
|
1157 |
|
/** |
1158 |
< |
* Attempts to stop all actively executing tasks, and cancels all |
1159 |
< |
* waiting tasks. Tasks that are in the process of being |
1160 |
< |
* submitted or executed concurrently during the course of this |
1161 |
< |
* method may or may not be rejected. Unlike some other executors, |
1162 |
< |
* this method cancels rather than collects non-executed tasks |
1163 |
< |
* upon termination, so always returns an empty list. However, you |
1164 |
< |
* can use method {@link #drainTasksTo} before invoking this |
1165 |
< |
* method to transfer unexecuted tasks to another collection. |
1158 |
> |
* Attempts to cancel and/or stop all tasks, and reject all |
1159 |
> |
* subsequently submitted tasks. Tasks that are in the process of |
1160 |
> |
* being submitted or executed concurrently during the course of |
1161 |
> |
* this method may or may not be rejected. This method cancels |
1162 |
> |
* both existing and unexecuted tasks, in order to permit |
1163 |
> |
* termination in the presence of task dependencies. So the method |
1164 |
> |
* always returns an empty list (unlike the case for some other |
1165 |
> |
* Executors). |
1166 |
|
* |
1167 |
|
* @return an empty list |
1168 |
|
* @throws SecurityException if a security manager exists and |
1187 |
|
|
1188 |
|
/** |
1189 |
|
* Returns {@code true} if the process of termination has |
1190 |
< |
* commenced but possibly not yet completed. |
1190 |
> |
* commenced but not yet completed. This method may be useful for |
1191 |
> |
* debugging. A return of {@code true} reported a sufficient |
1192 |
> |
* period after shutdown may indicate that submitted tasks have |
1193 |
> |
* ignored or suppressed interruption, causing this executor not |
1194 |
> |
* to properly terminate. |
1195 |
|
* |
1196 |
< |
* @return {@code true} if terminating |
1196 |
> |
* @return {@code true} if terminating but not yet terminated |
1197 |
|
*/ |
1198 |
|
public boolean isTerminating() { |
1199 |
< |
return runStateOf(runControl) >= TERMINATING; |
1199 |
> |
return runStateOf(runControl) == TERMINATING; |
1200 |
|
} |
1201 |
|
|
1202 |
|
/** |
1209 |
|
} |
1210 |
|
|
1211 |
|
/** |
1212 |
+ |
* Returns true if pool is not terminating or terminated. |
1213 |
+ |
* Used internally to suppress execution when terminating. |
1214 |
+ |
*/ |
1215 |
+ |
final boolean isProcessingTasks() { |
1216 |
+ |
return runStateOf(runControl) < TERMINATING; |
1217 |
+ |
} |
1218 |
+ |
|
1219 |
+ |
/** |
1220 |
|
* Blocks until all tasks have completed execution after a shutdown |
1221 |
|
* request, or the timeout occurs, or the current thread is |
1222 |
|
* interrupted, whichever happens first. |
1270 |
|
transitionRunStateTo(TERMINATED); |
1271 |
|
termination.signalAll(); |
1272 |
|
} |
1273 |
< |
else if (!isTerminating()) { |
1273 |
> |
else if (isProcessingTasks()) { |
1274 |
|
tryShrinkWorkerArray(); |
1275 |
|
tryResumeSpare(true); // allow replacement |
1276 |
|
} |
1380 |
|
} |
1381 |
|
} |
1382 |
|
|
1343 |
– |
|
1383 |
|
/* |
1384 |
|
* Nodes for event barrier to manage idle threads. Queue nodes |
1385 |
|
* are basic Treiber stack nodes, also used for spare stack. |
1403 |
|
* handling: Method signalWork returns without advancing count if |
1404 |
|
* the queue appears to be empty. This would ordinarily result in |
1405 |
|
* races causing some queued waiters not to be woken up. To avoid |
1406 |
< |
* this, the first worker enqueued in method sync (see |
1407 |
< |
* syncIsReleasable) rescans for tasks after being enqueued, and |
1408 |
< |
* helps signal if any are found. This works well because the |
1409 |
< |
* worker has nothing better to do, and so might as well help |
1410 |
< |
* alleviate the overhead and contention on the threads actually |
1411 |
< |
* doing work. Also, since event counts increments on task |
1412 |
< |
* availability exist to maintain liveness (rather than to force |
1413 |
< |
* refreshes etc), it is OK for callers to exit early if |
1375 |
< |
* contending with another signaller. |
1406 |
> |
* this, the first worker enqueued in method sync rescans for |
1407 |
> |
* tasks after being enqueued, and helps signal if any are |
1408 |
> |
* found. This works well because the worker has nothing better to |
1409 |
> |
* do, and so might as well help alleviate the overhead and |
1410 |
> |
* contention on the threads actually doing work. Also, since |
1411 |
> |
* event counts increments on task availability exist to maintain |
1412 |
> |
* liveness (rather than to force refreshes etc), it is OK for |
1413 |
> |
* callers to exit early if contending with another signaller. |
1414 |
|
*/ |
1415 |
|
static final class WaitQueueNode { |
1416 |
|
WaitQueueNode next; // only written before enqueued |
1423 |
|
} |
1424 |
|
|
1425 |
|
/** |
1426 |
< |
* Wakes up waiter, returning false if known to already |
1426 |
> |
* Wakes up waiter, returning false if known to already be awake |
1427 |
|
*/ |
1428 |
|
boolean signal() { |
1429 |
|
ForkJoinWorkerThread t = thread; |
1433 |
|
LockSupport.unpark(t); |
1434 |
|
return true; |
1435 |
|
} |
1398 |
– |
|
1399 |
– |
/** |
1400 |
– |
* Awaits release on sync. |
1401 |
– |
*/ |
1402 |
– |
void awaitSyncRelease(ForkJoinPool p) { |
1403 |
– |
while (thread != null && !p.syncIsReleasable(this)) |
1404 |
– |
LockSupport.park(this); |
1405 |
– |
} |
1406 |
– |
|
1407 |
– |
/** |
1408 |
– |
* Awaits resumption as spare. |
1409 |
– |
*/ |
1410 |
– |
void awaitSpareRelease() { |
1411 |
– |
while (thread != null) { |
1412 |
– |
if (!Thread.interrupted()) |
1413 |
– |
LockSupport.park(this); |
1414 |
– |
} |
1415 |
– |
} |
1436 |
|
} |
1437 |
|
|
1438 |
|
/** |
1439 |
|
* Ensures that no thread is waiting for count to advance from the |
1440 |
|
* current value of eventCount read on entry to this method, by |
1441 |
|
* releasing waiting threads if necessary. |
1422 |
– |
* |
1423 |
– |
* @return the count |
1442 |
|
*/ |
1443 |
< |
final long ensureSync() { |
1443 |
> |
final void ensureSync() { |
1444 |
|
long c = eventCount; |
1445 |
|
WaitQueueNode q; |
1446 |
|
while ((q = syncStack) != null && q.count < c) { |
1451 |
|
break; |
1452 |
|
} |
1453 |
|
} |
1436 |
– |
return c; |
1454 |
|
} |
1455 |
|
|
1456 |
|
/** |
1465 |
|
/** |
1466 |
|
* Signals threads waiting to poll a task. Because method sync |
1467 |
|
* rechecks availability, it is OK to only proceed if queue |
1468 |
< |
* appears to be non-empty, and OK to skip under contention to |
1469 |
< |
* increment count (since some other thread succeeded). |
1468 |
> |
* appears to be non-empty, and OK if CAS to increment count |
1469 |
> |
* fails (since some other thread succeeded). |
1470 |
|
*/ |
1471 |
|
final void signalWork() { |
1472 |
< |
long c; |
1473 |
< |
WaitQueueNode q; |
1474 |
< |
if (syncStack != null && |
1475 |
< |
casEventCount(c = eventCount, c+1) && |
1476 |
< |
(((q = syncStack) != null && q.count <= c) && |
1477 |
< |
(!casBarrierStack(q, q.next) || !q.signal()))) |
1478 |
< |
ensureSync(); |
1472 |
> |
if (syncStack != null) { |
1473 |
> |
long c; |
1474 |
> |
casEventCount(c = eventCount, c+1); |
1475 |
> |
WaitQueueNode q = syncStack; |
1476 |
> |
if (q != null && q.count <= c && |
1477 |
> |
(!casBarrierStack(q, q.next) || !q.signal())) |
1478 |
> |
ensureSync(); |
1479 |
> |
} |
1480 |
|
} |
1481 |
|
|
1482 |
|
/** |
1489 |
|
final void sync(ForkJoinWorkerThread w) { |
1490 |
|
updateStealCount(w); // Transfer w's count while it is idle |
1491 |
|
|
1492 |
< |
while (!w.isShutdown() && !isTerminating() && !suspendIfSpare(w)) { |
1492 |
> |
if (!w.isShutdown() && isProcessingTasks() && !suspendIfSpare(w)) { |
1493 |
|
long prev = w.lastEventCount; |
1494 |
|
WaitQueueNode node = null; |
1495 |
|
WaitQueueNode h; |
1496 |
< |
while (eventCount == prev && |
1496 |
> |
long ec; |
1497 |
> |
while ((ec = eventCount) == prev && |
1498 |
|
((h = syncStack) == null || h.count == prev)) { |
1499 |
|
if (node == null) |
1500 |
|
node = new WaitQueueNode(prev, w); |
1501 |
|
if (casBarrierStack(node.next = h, node)) { |
1502 |
< |
node.awaitSyncRelease(this); |
1502 |
> |
if (!Thread.interrupted() && |
1503 |
> |
node.thread != null && |
1504 |
> |
eventCount == prev && |
1505 |
> |
(h != null || // cover signalWork race |
1506 |
> |
(!ForkJoinWorkerThread.hasQueuedTasks(workers) && |
1507 |
> |
eventCount == prev))) |
1508 |
> |
LockSupport.park(this); |
1509 |
> |
ec = eventCount; |
1510 |
> |
if (node.thread != null) { |
1511 |
> |
node.thread = null; |
1512 |
> |
if (ec == prev) |
1513 |
> |
casEventCount(prev, prev + 1); // help signal |
1514 |
> |
} |
1515 |
|
break; |
1516 |
|
} |
1517 |
|
} |
1518 |
< |
long ec = ensureSync(); |
1519 |
< |
if (ec != prev) { |
1489 |
< |
w.lastEventCount = ec; |
1490 |
< |
break; |
1491 |
< |
} |
1492 |
< |
} |
1493 |
< |
} |
1494 |
< |
|
1495 |
< |
/** |
1496 |
< |
* Returns {@code true} if worker waiting on sync can proceed: |
1497 |
< |
* - on signal (thread == null) |
1498 |
< |
* - on event count advance (winning race to notify vs signaller) |
1499 |
< |
* - on interrupt |
1500 |
< |
* - if the first queued node, we find work available |
1501 |
< |
* If node was not signalled and event count not advanced on exit, |
1502 |
< |
* then we also help advance event count. |
1503 |
< |
* |
1504 |
< |
* @return {@code true} if node can be released |
1505 |
< |
*/ |
1506 |
< |
final boolean syncIsReleasable(WaitQueueNode node) { |
1507 |
< |
long prev = node.count; |
1508 |
< |
if (!Thread.interrupted() && node.thread != null && |
1509 |
< |
(node.next != null || |
1510 |
< |
!ForkJoinWorkerThread.hasQueuedTasks(workers)) && |
1511 |
< |
eventCount == prev) |
1512 |
< |
return false; |
1513 |
< |
if (node.thread != null) { |
1514 |
< |
node.thread = null; |
1515 |
< |
long ec = eventCount; |
1516 |
< |
if (prev <= ec) // help signal |
1517 |
< |
casEventCount(ec, ec+1); |
1518 |
> |
w.lastEventCount = ec; |
1519 |
> |
ensureSync(); |
1520 |
|
} |
1519 |
– |
return true; |
1521 |
|
} |
1522 |
|
|
1523 |
|
/** |
1526 |
|
*/ |
1527 |
|
final boolean hasNewSyncEvent(ForkJoinWorkerThread w) { |
1528 |
|
long lc = w.lastEventCount; |
1529 |
< |
long ec = ensureSync(); |
1530 |
< |
if (ec == lc) |
1531 |
< |
return false; |
1532 |
< |
w.lastEventCount = ec; |
1533 |
< |
return true; |
1529 |
> |
long ec = eventCount; |
1530 |
> |
if (lc != ec) |
1531 |
> |
w.lastEventCount = ec; |
1532 |
> |
ensureSync(); |
1533 |
> |
return lc != ec || lc != eventCount; |
1534 |
|
} |
1535 |
|
|
1536 |
|
// Parallelism maintenance |
1572 |
|
while (spareStack == null || !tryResumeSpare(dec)) { |
1573 |
|
int counts = workerCounts; |
1574 |
|
if (dec || (dec = casWorkerCounts(counts, --counts))) { |
1574 |
– |
// CAS cheat |
1575 |
|
if (!needSpare(counts, maintainParallelism)) |
1576 |
|
break; |
1577 |
|
if (joinMe.status < 0) |
1676 |
|
for (k = 0; k < len && ws[k] != null; ++k) |
1677 |
|
; |
1678 |
|
} |
1679 |
< |
if (k < len && !isTerminating() && (w = createWorker(k)) != null) { |
1679 |
> |
if (k < len && isProcessingTasks() && (w = createWorker(k)) != null) { |
1680 |
|
ws[k] = w; |
1681 |
|
w.start(); |
1682 |
|
} |
1696 |
|
*/ |
1697 |
|
private boolean suspendIfSpare(ForkJoinWorkerThread w) { |
1698 |
|
WaitQueueNode node = null; |
1699 |
< |
int s; |
1700 |
< |
while (parallelism < runningCountOf(s = workerCounts)) { |
1699 |
> |
for (;;) { |
1700 |
> |
int p = parallelism; |
1701 |
> |
int s = workerCounts; |
1702 |
> |
int r = runningCountOf(s); |
1703 |
> |
int t = totalCountOf(s); |
1704 |
> |
// use t as bound if r transiently out of sync |
1705 |
> |
if (t <= p || r <= p) |
1706 |
> |
return false; // not a spare |
1707 |
|
if (node == null) |
1708 |
|
node = new WaitQueueNode(0, w); |
1709 |
< |
if (casWorkerCounts(s, s-1)) { // representation-dependent |
1710 |
< |
// push onto stack |
1705 |
< |
do {} while (!casSpareStack(node.next = spareStack, node)); |
1706 |
< |
// block until released by resumeSpare |
1707 |
< |
node.awaitSpareRelease(); |
1708 |
< |
return true; |
1709 |
< |
} |
1709 |
> |
if (casWorkerCounts(s, workerCountsFor(t, r - 1))) |
1710 |
> |
break; |
1711 |
|
} |
1712 |
< |
return false; |
1712 |
> |
// push onto stack |
1713 |
> |
do {} while (!casSpareStack(node.next = spareStack, node)); |
1714 |
> |
// block until released by resumeSpare |
1715 |
> |
while (!Thread.interrupted() && node.thread != null) |
1716 |
> |
LockSupport.park(this); |
1717 |
> |
return true; |
1718 |
|
} |
1719 |
|
|
1720 |
|
/** |
1780 |
|
|
1781 |
|
/** |
1782 |
|
* Interface for extending managed parallelism for tasks running |
1783 |
< |
* in ForkJoinPools. A ManagedBlocker provides two methods. |
1783 |
> |
* in {@link ForkJoinPool}s. |
1784 |
> |
* |
1785 |
> |
* <p>A {@code ManagedBlocker} provides two methods. |
1786 |
|
* Method {@code isReleasable} must return {@code true} if |
1787 |
|
* blocking is not necessary. Method {@code block} blocks the |
1788 |
|
* current thread if necessary (perhaps internally invoking |
1789 |
< |
* {@code isReleasable} before actually blocking.). |
1789 |
> |
* {@code isReleasable} before actually blocking). |
1790 |
|
* |
1791 |
|
* <p>For example, here is a ManagedBlocker based on a |
1792 |
|
* ReentrantLock: |
1825 |
|
|
1826 |
|
/** |
1827 |
|
* Blocks in accord with the given blocker. If the current thread |
1828 |
< |
* is a ForkJoinWorkerThread, this method possibly arranges for a |
1829 |
< |
* spare thread to be activated if necessary to ensure parallelism |
1830 |
< |
* while the current thread is blocked. If |
1831 |
< |
* {@code maintainParallelism} is {@code true} and the pool supports |
1832 |
< |
* it ({@link #getMaintainsParallelism}), this method attempts to |
1833 |
< |
* maintain the pool's nominal parallelism. Otherwise it activates |
1834 |
< |
* a thread only if necessary to avoid complete starvation. This |
1835 |
< |
* option may be preferable when blockages use timeouts, or are |
1836 |
< |
* almost always brief. |
1828 |
> |
* is a {@link ForkJoinWorkerThread}, this method possibly |
1829 |
> |
* arranges for a spare thread to be activated if necessary to |
1830 |
> |
* ensure parallelism while the current thread is blocked. |
1831 |
> |
* |
1832 |
> |
* <p>If {@code maintainParallelism} is {@code true} and the pool |
1833 |
> |
* supports it ({@link #getMaintainsParallelism}), this method |
1834 |
> |
* attempts to maintain the pool's nominal parallelism. Otherwise |
1835 |
> |
* it activates a thread only if necessary to avoid complete |
1836 |
> |
* starvation. This option may be preferable when blockages use |
1837 |
> |
* timeouts, or are almost always brief. |
1838 |
|
* |
1839 |
< |
* <p> If the caller is not a ForkJoinTask, this method is behaviorally |
1840 |
< |
* equivalent to |
1839 |
> |
* <p>If the caller is not a {@link ForkJoinTask}, this method is |
1840 |
> |
* behaviorally equivalent to |
1841 |
|
* <pre> {@code |
1842 |
|
* while (!blocker.isReleasable()) |
1843 |
|
* if (blocker.block()) |
1844 |
|
* return; |
1845 |
|
* }</pre> |
1846 |
< |
* If the caller is a ForkJoinTask, then the pool may first |
1847 |
< |
* be expanded to ensure parallelism, and later adjusted. |
1846 |
> |
* |
1847 |
> |
* If the caller is a {@code ForkJoinTask}, then the pool may |
1848 |
> |
* first be expanded to ensure parallelism, and later adjusted. |
1849 |
|
* |
1850 |
|
* @param blocker the blocker |
1851 |
|
* @param maintainParallelism if {@code true} and supported by |
1877 |
|
do {} while (!blocker.isReleasable() && !blocker.block()); |
1878 |
|
} |
1879 |
|
|
1880 |
< |
// AbstractExecutorService overrides |
1880 |
> |
// AbstractExecutorService overrides. These rely on undocumented |
1881 |
> |
// fact that ForkJoinTask.adapt returns ForkJoinTasks that also |
1882 |
> |
// implement RunnableFuture. |
1883 |
|
|
1884 |
|
protected <T> RunnableFuture<T> newTaskFor(Runnable runnable, T value) { |
1885 |
< |
return new AdaptedRunnable<T>(runnable, value); |
1885 |
> |
return (RunnableFuture<T>) ForkJoinTask.adapt(runnable, value); |
1886 |
|
} |
1887 |
|
|
1888 |
|
protected <T> RunnableFuture<T> newTaskFor(Callable<T> callable) { |
1889 |
< |
return new AdaptedCallable<T>(callable); |
1889 |
> |
return (RunnableFuture<T>) ForkJoinTask.adapt(callable); |
1890 |
|
} |
1891 |
|
|
1892 |
|
// Unsafe mechanics |