ViewVC Help

View File | Revision Log | Show Annotations | Download File | Root Listing

root/jsr166/jsr166/src/jsr166y/ForkJoinPool.java

Comparing jsr166/src/jsr166y/ForkJoinPool.java (file contents):
Revision 1.3 by dl, Wed Jan 7 19:12:36 2009 UTC vs.
Revision 1.48 by jsr166, Thu Aug 6 06:41:34 2009 UTC

#	Line 5 | Line 5
5		*/
6
7		package jsr166y;
8	<	import java.util.*;
8	>
9		import java.util.concurrent.*;
10	<	import java.util.concurrent.locks.*;
11	<	import java.util.concurrent.atomic.*;
12	<	import sun.misc.Unsafe;
13	<	import java.lang.reflect.*;
10	>
11	>	import java.util.ArrayList;
12	>	import java.util.Arrays;
13	>	import java.util.Collection;
14	>	import java.util.Collections;
15	>	import java.util.List;
16	>	import java.util.concurrent.locks.Condition;
17	>	import java.util.concurrent.locks.LockSupport;
18	>	import java.util.concurrent.locks.ReentrantLock;
19	>	import java.util.concurrent.atomic.AtomicInteger;
20	>	import java.util.concurrent.atomic.AtomicLong;
21
22		/**
23	<	* An {@link ExecutorService} for running {@link ForkJoinTask}s.  A
24	<	* 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
20	<	* submitted tasks. Otherwise, use would not usually outweigh the
21	<	* construction and bookkeeping overhead of creating a large set of
22	<	* threads.
23	>	* An {@link ExecutorService} for running {@link ForkJoinTask}s.
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. Otherwise, other
35	<	* ExecutorService implementations are typically more appropriate
36	<	* 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 <code>ManagedBlocker</code> interface enables extension of
48	<	* the kinds of synchronization accommodated.  The target parallelism
49	<	* level may also be changed dynamically (<code>setParallelism</code>)
50	<	* and dynamically thread construction can be limited using methods
51	<	* <code>setMaximumPoolSize</code> and/or
52	<	* <code>setMaintainsParallelism</code>.
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
60	<	* <code>getStealCount</code>) that are intended to aid in developing,
60	>	* {@link #getStealCount}) that are intended to aid in developing,
61		* tuning, and monitoring fork/join applications. Also, method
62	<	* <code>toString</code> returns indications of pool state in a
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
93		*/
94		public class ForkJoinPool extends AbstractExecutorService {
95
#	Line 69 | Line 105 | public class ForkJoinPool extends Abstra
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		}
122
123		/**
124	<	* Default ForkJoinWorkerThreadFactory implementation, creates a
124	>	* Default ForkJoinWorkerThreadFactory implementation; creates a
125		* new ForkJoinWorkerThread.
126		*/
127		static class  DefaultForkJoinWorkerThreadFactory
#	Line 131 | Line 167 | public class ForkJoinPool extends Abstra
167		new AtomicInteger();
168
169		/**
170	<	* Array holding all worker threads in the pool. Array size must
171	<	* be a power of two.  Updates and replacements are protected by
172	<	* workerLock, but it is always kept in a consistent enough state
173	<	* to be randomly accessed without locking by workers performing
174	<	* work-stealing.
170	>	* Array holding all worker threads in the pool. Initialized upon
171	>	* first use. Array size must be a power of two.  Updates and
172	>	* replacements are protected by workerLock, but it is always kept
173	>	* in a consistent enough state to be randomly accessed without
174	>	* locking by workers performing work-stealing.
175		*/
176		volatile ForkJoinWorkerThread[] workers;
177
#	Line 151 | Line 187 | public class ForkJoinPool extends Abstra
187
188		/**
189		* The uncaught exception handler used when any worker
190	<	* abrupty terminates
190	>	* abruptly terminates
191		*/
192		private Thread.UncaughtExceptionHandler ueh;
193
#	Line 179 | Line 215 | public class ForkJoinPool extends Abstra
215		private final LinkedTransferQueue<ForkJoinTask<?>> submissionQueue;
216
217		/**
218	<	* Head of Treiber stack for barrier sync. See below for explanation
218	>	* Head of Treiber stack for barrier sync. See below for explanation.
219		*/
220	<	private volatile WaitQueueNode barrierStack;
220	>	private volatile WaitQueueNode syncStack;
221
222		/**
223		* The count for event barrier
#	Line 204 | Line 240 | public class ForkJoinPool extends Abstra
240		private volatile int parallelism;
241
242		/**
243	+	* True if use local fifo, not default lifo, for local polling
244	+	*/
245	+	private volatile boolean locallyFifo;
246	+
247	+	/**
248		* Holds number of total (i.e., created and not yet terminated)
249		* and running (i.e., not blocked on joins or other managed sync)
250		* threads, packed into one int to ensure consistent snapshot when
251		* making decisions about creating and suspending spare
252		* threads. Updated only by CAS.  Note: CASes in
253	<	* updateRunningCount and preJoin running active count is in low
254	<	* word, so need to be modified if this changes
253	>	* updateRunningCount and preJoin assume that running active count
254	>	* is in low word, so need to be modified if this changes.
255		*/
256		private volatile int workerCounts;
257
#	Line 219 | Line 260 | public class ForkJoinPool extends Abstra
260		private static int workerCountsFor(int t, int r) { return (t << 16) + r; }
261
262		/**
263	<	* Add delta (which may be negative) to running count.  This must
263	>	* Adds delta (which may be negative) to running count.  This must
264		* be called before (with negative arg) and after (with positive)
265	<	* any managed synchronization (i.e., mainly, joins)
265	>	* any managed synchronization (i.e., mainly, joins).
266	>	*
267		* @param delta the number to add
268		*/
269		final void updateRunningCount(int delta) {
270		int s;
271	<	do;while (!casWorkerCounts(s = workerCounts, s + delta));
271	>	do {} while (!casWorkerCounts(s = workerCounts, s + delta));
272		}
273
274		/**
275	<	* Add delta (which may be negative) to both total and running
275	>	* Adds delta (which may be negative) to both total and running
276		* count.  This must be called upon creation and termination of
277		* worker threads.
278	+	*
279		* @param delta the number to add
280		*/
281		private void updateWorkerCount(int delta) {
282		int d = delta + (delta << 16); // add to both lo and hi parts
283		int s;
284	<	do;while (!casWorkerCounts(s = workerCounts, s + d));
284	>	do {} while (!casWorkerCounts(s = workerCounts, s + d));
285		}
286
287		/**
#	Line 264 | Line 307 | public class ForkJoinPool extends Abstra
307		private static int runControlFor(int r, int a)   { return (r << 16) + a; }
308
309		/**
310	<	* Increment active count. Called by workers before/during
311	<	* executing tasks.
310	>	* Tries incrementing active count; fails on contention.
311	>	* Called by workers before/during executing tasks.
312	>	*
313	>	* @return true on success
314		*/
315	<	final void incrementActiveCount() {
316	<	int c;
317	<	do;while (!casRunControl(c = runControl, c+1));
315	>	final boolean tryIncrementActiveCount() {
316	>	int c = runControl;
317	>	return casRunControl(c, c+1);
318		}
319
320		/**
321	<	* Decrement active count; possibly trigger termination.
321	>	* Tries decrementing active count; fails on contention.
322	>	* Possibly triggers termination on success.
323		* Called by workers when they can't find tasks.
324	+	*
325	+	* @return true on success
326		*/
327	<	final void decrementActiveCount() {
328	<	int c, nextc;
329	<	do;while (!casRunControl(c = runControl, nextc = c-1));
327	>	final boolean tryDecrementActiveCount() {
328	>	int c = runControl;
329	>	int nextc = c - 1;
330	>	if (!casRunControl(c, nextc))
331	>	return false;
332		if (canTerminateOnShutdown(nextc))
333		terminateOnShutdown();
334	+	return true;
335		}
336
337		/**
338	<	* Return true if argument represents zero active count and
339	<	* nonzero runstate, which is the triggering condition for
338	>	* Returns {@code true} if argument represents zero active count
339	>	* and nonzero runstate, which is the triggering condition for
340		* terminating on shutdown.
341		*/
342		private static boolean canTerminateOnShutdown(int c) {
343	<	return ((c & -c) >>> 16) != 0; // i.e. least bit is nonzero runState bit
343	>	// i.e. least bit is nonzero runState bit
344	>	return ((c & -c) >>> 16) != 0;
345		}
346
347		/**
#	Line 314 | Line 366 | public class ForkJoinPool extends Abstra
366		// Constructors
367
368		/**
369	<	* Creates a ForkJoinPool with a pool size equal to the number of
370	<	* processors available on the system and 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
375		*         because it does not hold {@link
376	<	*         java.lang.RuntimePermission}<code>("modifyThread")</code>,
376	>	*         java.lang.RuntimePermission}{@code ("modifyThread")}
377		*/
378		public ForkJoinPool() {
379		this(Runtime.getRuntime().availableProcessors(),
#	Line 328 | Line 381 | public class ForkJoinPool extends Abstra
381		}
382
383		/**
384	<	* Creates a ForkJoinPool with the indicated parellelism level
385	<	* threads, and using the default ForkJoinWorkerThreadFactory,
386	<	* @param parallelism the number of worker threads
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 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
394	<	*         java.lang.RuntimePermission}<code>("modifyThread")</code>,
394	>	*         java.lang.RuntimePermission}{@code ("modifyThread")}
395		*/
396		public ForkJoinPool(int parallelism) {
397		this(parallelism, defaultForkJoinWorkerThreadFactory);
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
410	<	*         java.lang.RuntimePermission}<code>("modifyThread")</code>,
410	>	*         java.lang.RuntimePermission}{@code ("modifyThread")}
411		*/
412		public ForkJoinPool(ForkJoinWorkerThreadFactory factory) {
413		this(Runtime.getRuntime().availableProcessors(), factory);
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
428	<	*         java.lang.RuntimePermission}<code>("modifyThread")</code>,
428	>	*         java.lang.RuntimePermission}{@code ("modifyThread")}
429		*/
430		public ForkJoinPool(int parallelism, ForkJoinWorkerThreadFactory factory) {
431		if (parallelism <= 0 || parallelism > MAX_THREADS)
#	Line 385 | Line 442 | public class ForkJoinPool extends Abstra
442		this.termination = workerLock.newCondition();
443		this.stealCount = new AtomicLong();
444		this.submissionQueue = new LinkedTransferQueue<ForkJoinTask<?>>();
445	<	createAndStartInitialWorkers(parallelism);
445	>	// worker array and workers are lazily constructed
446		}
447
448		/**
449	<	* Create new worker using factory.
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;
#	Line 399 | Line 457 | public class ForkJoinPool extends Abstra
457		if (w != null) {
458		w.poolIndex = index;
459		w.setDaemon(true);
460	+	w.setAsyncMode(locallyFifo);
461		w.setName("ForkJoinPool-" + poolNumber + "-worker-" + index);
462		if (h != null)
463		w.setUncaughtExceptionHandler(h);
#	Line 407 | Line 466 | public class ForkJoinPool extends Abstra
466		}
467
468		/**
469	<	* Return a good size for worker array given pool size.
469	>	* Returns a good size for worker array given pool size.
470		* Currently requires size to be a power of two.
471		*/
472	<	private static int arraySizeFor(int ps) {
473	<	return ps <= 1? 1 : (1 << (32 - Integer.numberOfLeadingZeros(ps-1)));
472	>	private static int arraySizeFor(int poolSize) {
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		/**
486	<	* Create or resize array if necessary to hold newLength
486	>	* Creates or resizes array if necessary to hold newLength.
487	>	* Call only under exclusion.
488	>	*
489		* @return the array
490		*/
491		private ForkJoinWorkerThread[] ensureWorkerArrayCapacity(int newLength) {
#	Line 429 | Line 499 | public class ForkJoinPool extends Abstra
499		}
500
501		/**
502	<	* Try to shrink workers into smaller array after one or more terminate
502	>	* Tries to shrink workers into smaller array after one or more terminate.
503		*/
504		private void tryShrinkWorkerArray() {
505		ForkJoinWorkerThread[] ws = workers;
506	<	int len = ws.length;
507	<	int last = len - 1;
508	<	while (last >= 0 && ws[last] == null)
509	<	--last;
510	<	int newLength = arraySizeFor(last+1);
511	<	if (newLength < len)
512	<	workers = Arrays.copyOf(ws, newLength);
506	>	if (ws != null) {
507	>	int len = ws.length;
508	>	int last = len - 1;
509	>	while (last >= 0 && ws[last] == null)
510	>	--last;
511	>	int newLength = arraySizeFor(last+1);
512	>	if (newLength < len)
513	>	workers = Arrays.copyOf(ws, newLength);
514	>	}
515		}
516
517		/**
518	<	* Initial worker array and worker creation and startup. (This
447	<	* must be done under lock to avoid interference by some of the
448	<	* newly started threads while creating others.)
518	>	* Initializes workers if necessary.
519		*/
520	<	private void createAndStartInitialWorkers(int ps) {
521	<	final ReentrantLock lock = this.workerLock;
522	<	lock.lock();
523	<	try {
524	<	ForkJoinWorkerThread[] ws = ensureWorkerArrayCapacity(ps);
525	<	for (int i = 0; i < ps; ++i) {
526	<	ForkJoinWorkerThread w = createWorker(i);
527	<	if (w != null) {
528	<	ws[i] = w;
529	<	w.start();
530	<	updateWorkerCount(1);
520	>	final void ensureWorkerInitialization() {
521	>	ForkJoinWorkerThread[] ws = workers;
522	>	if (ws == null) {
523	>	final ReentrantLock lock = this.workerLock;
524	>	lock.lock();
525	>	try {
526	>	ws = workers;
527	>	if (ws == null) {
528	>	int ps = parallelism;
529	>	ws = ensureWorkerArrayCapacity(ps);
530	>	for (int i = 0; i < ps; ++i) {
531	>	ForkJoinWorkerThread w = createWorker(i);
532	>	if (w != null) {
533	>	ws[i] = w;
534	>	w.start();
535	>	updateWorkerCount(1);
536	>	}
537	>	}
538		}
539	+	} finally {
540	+	lock.unlock();
541		}
463	–	} finally {
464	–	lock.unlock();
542		}
543		}
544
#	Line 505 | Line 582 | public class ForkJoinPool extends Abstra
582		* Common code for execute, invoke and submit
583		*/
584		private <T> void doSubmit(ForkJoinTask<T> task) {
585	+	if (task == null)
586	+	throw new NullPointerException();
587		if (isShutdown())
588		throw new RejectedExecutionException();
589	+	if (workers == null)
590	+	ensureWorkerInitialization();
591		submissionQueue.offer(task);
592	<	signalIdleWorkers(true);
592	>	signalIdleWorkers();
593		}
594
595		/**
596	<	* Performs the given task; returning its result upon completion
596	>	* Performs the given task, returning its result upon completion.
597	>	*
598		* @param task the task
599		* @return the task's result
600	<	* @throws NullPointerException if task is null
601	<	* @throws RejectedExecutionException if pool is shut down
600	>	* @throws NullPointerException if the task is null
601	>	* @throws RejectedExecutionException if the task cannot be
602	>	*         scheduled for execution
603		*/
604		public <T> T invoke(ForkJoinTask<T> task) {
605		doSubmit(task);
#	Line 525 | Line 608 | public class ForkJoinPool extends Abstra
608
609		/**
610		* Arranges for (asynchronous) execution of the given task.
611	+	*
612		* @param task the task
613	<	* @throws NullPointerException if task is null
614	<	* @throws RejectedExecutionException if pool is shut down
613	>	* @throws NullPointerException if the task is null
614	>	* @throws RejectedExecutionException if the task cannot be
615	>	*         scheduled for execution
616		*/
617	<	public <T> void execute(ForkJoinTask<T> task) {
617	>	public void execute(ForkJoinTask<?> task) {
618		doSubmit(task);
619		}
620
621		// AbstractExecutorService methods
622
623	+	/**
624	+	* @throws NullPointerException if the task is null
625	+	* @throws RejectedExecutionException if the task cannot be
626	+	*         scheduled for execution
627	+	*/
628		public void execute(Runnable task) {
629	<	doSubmit(new AdaptedRunnable<Void>(task, null));
629	>	ForkJoinTask<?> job;
630	>	if (task instanceof ForkJoinTask<?>) // avoid re-wrap
631	>	job = (ForkJoinTask<?>) task;
632	>	else
633	>	job = ForkJoinTask.adapt(task, null);
634	>	doSubmit(job);
635		}
636
637	+	/**
638	+	* @throws NullPointerException if the task is null
639	+	* @throws RejectedExecutionException if the task cannot be
640	+	*         scheduled for execution
641	+	*/
642		public <T> ForkJoinTask<T> submit(Callable<T> task) {
643	<	ForkJoinTask<T> job = new AdaptedCallable<T>(task);
643	>	ForkJoinTask<T> job = ForkJoinTask.adapt(task);
644		doSubmit(job);
645		return job;
646		}
647
648	+	/**
649	+	* @throws NullPointerException if the task is null
650	+	* @throws RejectedExecutionException if the task cannot be
651	+	*         scheduled for execution
652	+	*/
653		public <T> ForkJoinTask<T> submit(Runnable task, T result) {
654	<	ForkJoinTask<T> job = new AdaptedRunnable<T>(task, result);
654	>	ForkJoinTask<T> job = ForkJoinTask.adapt(task, result);
655		doSubmit(job);
656		return job;
657		}
658
659	+	/**
660	+	* @throws NullPointerException if the task is null
661	+	* @throws RejectedExecutionException if the task cannot be
662	+	*         scheduled for execution
663	+	*/
664		public ForkJoinTask<?> submit(Runnable task) {
665	<	ForkJoinTask<Void> job = new AdaptedRunnable<Void>(task, null);
665	>	ForkJoinTask<?> job;
666	>	if (task instanceof ForkJoinTask<?>) // avoid re-wrap
667	>	job = (ForkJoinTask<?>) task;
668	>	else
669	>	job = ForkJoinTask.adapt(task, null);
670		doSubmit(job);
671		return job;
672		}
673
674		/**
675	<	* Adaptor for Runnables. This implements RunnableFuture
676	<	* to be compliant with AbstractExecutorService constraints
675	>	* Submits a ForkJoinTask for execution.
676	>	*
677	>	* @param task the task to submit
678	>	* @return the task
679	>	* @throws NullPointerException if the task is null
680	>	* @throws RejectedExecutionException if the task cannot be
681	>	*         scheduled for execution
682		*/
683	<	static final class AdaptedRunnable<T> extends ForkJoinTask<T>
684	<	implements RunnableFuture<T> {
685	<	final Runnable runnable;
567	<	final T resultOnCompletion;
568	<	T result;
569	<	AdaptedRunnable(Runnable runnable, T result) {
570	<	if (runnable == null) throw new NullPointerException();
571	<	this.runnable = runnable;
572	<	this.resultOnCompletion = result;
573	<	}
574	<	public T getRawResult() { return result; }
575	<	public void setRawResult(T v) { result = v; }
576	<	public boolean exec() {
577	<	runnable.run();
578	<	result = resultOnCompletion;
579	<	return true;
580	<	}
581	<	public void run() { invoke(); }
683	>	public <T> ForkJoinTask<T> submit(ForkJoinTask<T> task) {
684	>	doSubmit(task);
685	>	return task;
686		}
687
688	+
689		/**
690	<	* Adaptor for Callables
690	>	* @throws NullPointerException       {@inheritDoc}
691	>	* @throws RejectedExecutionException {@inheritDoc}
692		*/
587	–	static final class AdaptedCallable<T> extends ForkJoinTask<T>
588	–	implements RunnableFuture<T> {
589	–	final Callable<T> callable;
590	–	T result;
591	–	AdaptedCallable(Callable<T> callable) {
592	–	if (callable == null) throw new NullPointerException();
593	–	this.callable = callable;
594	–	}
595	–	public T getRawResult() { return result; }
596	–	public void setRawResult(T v) { result = v; }
597	–	public boolean exec() {
598	–	try {
599	–	result = callable.call();
600	–	return true;
601	–	} catch (Error err) {
602	–	throw err;
603	–	} catch (RuntimeException rex) {
604	–	throw rex;
605	–	} catch (Exception ex) {
606	–	throw new RuntimeException(ex);
607	–	}
608	–	}
609	–	public void run() { invoke(); }
610	–	}
611	–
693		public <T> List<Future<T>> invokeAll(Collection<? extends Callable<T>> tasks) {
694	<	ArrayList<ForkJoinTask<T>> ts =
694	>	ArrayList<ForkJoinTask<T>> forkJoinTasks =
695		new ArrayList<ForkJoinTask<T>>(tasks.size());
696	<	for (Callable<T> c : tasks)
697	<	ts.add(new AdaptedCallable<T>(c));
698	<	invoke(new InvokeAll<T>(ts));
699	<	return (List<Future<T>>)(List)ts;
696	>	for (Callable<T> task : tasks)
697	>	forkJoinTasks.add(ForkJoinTask.adapt(task));
698	>	invoke(new InvokeAll<T>(forkJoinTasks));
699	>
700	>	@SuppressWarnings({"unchecked", "rawtypes"})
701	>	List<Future<T>> futures = (List<Future<T>>) (List) forkJoinTasks;
702	>	return futures;
703		}
704
705		static final class InvokeAll<T> extends RecursiveAction {
706		final ArrayList<ForkJoinTask<T>> tasks;
707		InvokeAll(ArrayList<ForkJoinTask<T>> tasks) { this.tasks = tasks; }
708		public void compute() {
709	<	try { invokeAll(tasks); } catch(Exception ignore) {}
709	>	try { invokeAll(tasks); }
710	>	catch (Exception ignore) {}
711		}
712	+	private static final long serialVersionUID = -7914297376763021607L;
713		}
714
715		// Configuration and status settings and queries
716
717		/**
718	<	* Returns the factory used for constructing new workers
718	>	* Returns the factory used for constructing new workers.
719		*
720		* @return the factory used for constructing new workers
721		*/
#	Line 640 | Line 726 | public class ForkJoinPool extends Abstra
726		/**
727		* Returns the handler for internal worker threads that terminate
728		* due to unrecoverable errors encountered while executing tasks.
729	<	* @return the handler, or null if none
729	>	*
730	>	* @return the handler, or {@code null} if none
731		*/
732		public Thread.UncaughtExceptionHandler getUncaughtExceptionHandler() {
733		Thread.UncaughtExceptionHandler h;
#	Line 661 | Line 748 | public class ForkJoinPool extends Abstra
748		* as handler.
749		*
750		* @param h the new handler
751	<	* @return the old handler, or null if none
751	>	* @return the old handler, or {@code null} if none
752		* @throws SecurityException if a security manager exists and
753		*         the caller is not permitted to modify threads
754		*         because it does not hold {@link
755	<	*         java.lang.RuntimePermission}<code>("modifyThread")</code>,
755	>	*         java.lang.RuntimePermission}{@code ("modifyThread")}
756		*/
757		public Thread.UncaughtExceptionHandler
758		setUncaughtExceptionHandler(Thread.UncaughtExceptionHandler h) {
#	Line 677 | Line 764 | public class ForkJoinPool extends Abstra
764		old = ueh;
765		ueh = h;
766		ForkJoinWorkerThread[] ws = workers;
767	<	for (int i = 0; i < ws.length; ++i) {
768	<	ForkJoinWorkerThread w = ws[i];
769	<	if (w != null)
770	<	w.setUncaughtExceptionHandler(h);
767	>	if (ws != null) {
768	>	for (int i = 0; i < ws.length; ++i) {
769	>	ForkJoinWorkerThread w = ws[i];
770	>	if (w != null)
771	>	w.setUncaughtExceptionHandler(h);
772	>	}
773		}
774		} finally {
775		lock.unlock();
#	Line 690 | Line 779 | public class ForkJoinPool extends Abstra
779
780
781		/**
782	<	* Sets the target paralleism level of this pool.
782	>	* Sets the target parallelism level of this pool.
783	>	*
784		* @param parallelism the target parallelism
785		* @throws IllegalArgumentException if parallelism less than or
786	<	* equal to zero or greater than maximum size bounds.
786	>	* equal to zero or greater than maximum size bounds
787		* @throws SecurityException if a security manager exists and
788		*         the caller is not permitted to modify threads
789		*         because it does not hold {@link
790	<	*         java.lang.RuntimePermission}<code>("modifyThread")</code>,
790	>	*         java.lang.RuntimePermission}{@code ("modifyThread")}
791		*/
792		public void setParallelism(int parallelism) {
793		checkPermission();
#	Line 706 | Line 796 | public class ForkJoinPool extends Abstra
796		final ReentrantLock lock = this.workerLock;
797		lock.lock();
798		try {
799	<	if (!isTerminating()) {
799	>	if (isProcessingTasks()) {
800		int p = this.parallelism;
801		this.parallelism = parallelism;
802		if (parallelism > p)
#	Line 717 | Line 807 | public class ForkJoinPool extends Abstra
807		} finally {
808		lock.unlock();
809		}
810	<	signalIdleWorkers(false);
810	>	signalIdleWorkers();
811		}
812
813		/**
814	<	* Returns the targeted number of worker threads in this pool.
814	>	* Returns the targeted parallelism level of this pool.
815		*
816	<	* @return the targeted number of worker threads in this pool
816	>	* @return the targeted parallelism level of this pool
817		*/
818		public int getParallelism() {
819		return parallelism;
#	Line 732 | Line 822 | public class ForkJoinPool extends Abstra
822		/**
823		* Returns the number of worker threads that have started but not
824		* yet terminated.  This result returned by this method may differ
825	<	* from <code>getParallelism</code> when threads are created to
825	>	* from {@link #getParallelism} when threads are created to
826		* maintain parallelism when others are cooperatively blocked.
827		*
828		* @return the number of worker threads
#	Line 743 | Line 833 | public class ForkJoinPool extends Abstra
833
834		/**
835		* Returns the maximum number of threads allowed to exist in the
836	<	* pool, even if there are insufficient unblocked running threads.
836	>	* pool. Unless set using {@link #setMaximumPoolSize}, the
837	>	* maximum is an implementation-defined value designed only to
838	>	* prevent runaway growth.
839	>	*
840		* @return the maximum
841		*/
842		public int getMaximumPoolSize() {
#	Line 752 | Line 845 | public class ForkJoinPool extends Abstra
845
846		/**
847		* Sets the maximum number of threads allowed to exist in the
848	<	* pool, even if there are insufficient unblocked running threads.
849	<	* Setting this value has no effect on current pool size. It
850	<	* controls construction of new threads.
851	<	* @throws IllegalArgumentException if negative or greater then
852	<	* internal implementation limit.
848	>	* pool. The given value should normally be greater than or equal
849	>	* to the {@link #getParallelism parallelism} level. Setting this
850	>	* value has no effect on current pool size. It controls
851	>	* construction of new threads.
852	>	*
853	>	* @throws IllegalArgumentException if negative or greater than
854	>	* internal implementation limit
855		*/
856		public void setMaximumPoolSize(int newMax) {
857		if (newMax < 0 || newMax > MAX_THREADS)
#	Line 766 | Line 861 | public class ForkJoinPool extends Abstra
861
862
863		/**
864	<	* Returns true if this pool dynamically maintains its target
865	<	* parallelism level. If false, new threads are added only to
866	<	* avoid possible starvation.
867	<	* This setting is by default true;
868	<	* @return true if maintains parallelism
864	>	* Returns {@code true} if this pool dynamically maintains its
865	>	* target parallelism level. If false, new threads are added only
866	>	* to avoid possible starvation.  This setting is by default true.
867	>	*
868	>	* @return {@code true} if maintains parallelism
869		*/
870		public boolean getMaintainsParallelism() {
871		return maintainsParallelism;
#	Line 780 | Line 875 | public class ForkJoinPool extends Abstra
875		* Sets whether this pool dynamically maintains its target
876		* parallelism level. If false, new threads are added only to
877		* avoid possible starvation.
878	<	* @param enable true to maintains parallelism
878	>	*
879	>	* @param enable {@code true} to maintain parallelism
880		*/
881		public void setMaintainsParallelism(boolean enable) {
882		maintainsParallelism = enable;
883		}
884
885		/**
886	+	* Establishes local first-in-first-out scheduling mode for forked
887	+	* tasks that are never joined. This mode may be more appropriate
888	+	* than default locally stack-based mode in applications in which
889	+	* worker threads only process asynchronous tasks.  This method is
890	+	* designed to be invoked only when the pool is quiescent, and
891	+	* typically only before any tasks are submitted. The effects of
892	+	* invocations at other times may be unpredictable.
893	+	*
894	+	* @param async if {@code true}, use locally FIFO scheduling
895	+	* @return the previous mode
896	+	* @see #getAsyncMode
897	+	*/
898	+	public boolean setAsyncMode(boolean async) {
899	+	boolean oldMode = locallyFifo;
900	+	locallyFifo = async;
901	+	ForkJoinWorkerThread[] ws = workers;
902	+	if (ws != null) {
903	+	for (int i = 0; i < ws.length; ++i) {
904	+	ForkJoinWorkerThread t = ws[i];
905	+	if (t != null)
906	+	t.setAsyncMode(async);
907	+	}
908	+	}
909	+	return oldMode;
910	+	}
911	+
912	+	/**
913	+	* Returns {@code true} if this pool uses local first-in-first-out
914	+	* scheduling mode for forked tasks that are never joined.
915	+	*
916	+	* @return {@code true} if this pool uses async mode
917	+	* @see #setAsyncMode
918	+	*/
919	+	public boolean getAsyncMode() {
920	+	return locallyFifo;
921	+	}
922	+
923	+	/**
924		* Returns an estimate of the number of worker threads that are
925		* not blocked waiting to join tasks or for other managed
926		* synchronization.
#	Line 801 | Line 935 | public class ForkJoinPool extends Abstra
935		* Returns an estimate of the number of threads that are currently
936		* stealing or executing tasks. This method may overestimate the
937		* number of active threads.
938	<	* @return the number of active threads.
938	>	*
939	>	* @return the number of active threads
940		*/
941		public int getActiveThreadCount() {
942		return activeCountOf(runControl);
#	Line 811 | Line 946 | public class ForkJoinPool extends Abstra
946		* Returns an estimate of the number of threads that are currently
947		* idle waiting for tasks. This method may underestimate the
948		* number of idle threads.
949	<	* @return the number of idle threads.
949	>	*
950	>	* @return the number of idle threads
951		*/
952		final int getIdleThreadCount() {
953		int c = runningCountOf(workerCounts) - activeCountOf(runControl);
954	<	return (c <= 0)? 0 : c;
954	>	return (c <= 0) ? 0 : c;
955		}
956
957		/**
958	<	* Returns true if all worker threads are currently idle. An idle
959	<	* worker is one that cannot obtain a task to execute because none
960	<	* are available to steal from other threads, and there are no
961	<	* pending submissions to the pool. This method is conservative:
962	<	* It might not return true immediately upon idleness of all
963	<	* threads, but will eventually become true if threads remain
964	<	* inactive.
965	<	* @return true if all threads are currently idle
958	>	* Returns {@code true} if all worker threads are currently idle.
959	>	* An idle worker is one that cannot obtain a task to execute
960	>	* because none are available to steal from other threads, and
961	>	* there are no pending submissions to the pool. This method is
962	>	* conservative; it might not return {@code true} immediately upon
963	>	* idleness of all threads, but will eventually become true if
964	>	* threads remain inactive.
965	>	*
966	>	* @return {@code true} if all threads are currently idle
967		*/
968		public boolean isQuiescent() {
969		return activeCountOf(runControl) == 0;
#	Line 837 | Line 974 | public class ForkJoinPool extends Abstra
974		* one thread's work queue by another. The reported value
975		* underestimates the actual total number of steals when the pool
976		* is not quiescent. This value may be useful for monitoring and
977	<	* tuning fork/join programs: In general, steal counts should be
977	>	* tuning fork/join programs: in general, steal counts should be
978		* high enough to keep threads busy, but low enough to avoid
979		* overhead and contention across threads.
980	<	* @return the number of steals.
980	>	*
981	>	* @return the number of steals
982		*/
983		public long getStealCount() {
984		return stealCount.get();
985		}
986
987		/**
988	<	* Accumulate steal count from a worker. Call only
989	<	* when worker known to be idle.
988	>	* Accumulates steal count from a worker.
989	>	* Call only when worker known to be idle.
990		*/
991		private void updateStealCount(ForkJoinWorkerThread w) {
992		int sc = w.getAndClearStealCount();
#	Line 863 | Line 1001 | public class ForkJoinPool extends Abstra
1001		* an approximation, obtained by iterating across all threads in
1002		* the pool. This method may be useful for tuning task
1003		* granularities.
1004	<	* @return the number of queued tasks.
1004	>	*
1005	>	* @return the number of queued tasks
1006		*/
1007		public long getQueuedTaskCount() {
1008		long count = 0;
1009		ForkJoinWorkerThread[] ws = workers;
1010	<	for (int i = 0; i < ws.length; ++i) {
1011	<	ForkJoinWorkerThread t = ws[i];
1012	<	if (t != null)
1013	<	count += t.getQueueSize();
1010	>	if (ws != null) {
1011	>	for (int i = 0; i < ws.length; ++i) {
1012	>	ForkJoinWorkerThread t = ws[i];
1013	>	if (t != null)
1014	>	count += t.getQueueSize();
1015	>	}
1016		}
1017		return count;
1018		}
1019
1020		/**
1021	<	* Returns an estimate of the number tasks submitted to this pool
1022	<	* that have not yet begun executing. This method takes time
1021	>	* Returns an estimate of the number of tasks submitted to this
1022	>	* pool that have not yet begun executing.  This method takes time
1023		* proportional to the number of submissions.
1024	<	* @return the number of queued submissions.
1024	>	*
1025	>	* @return the number of queued submissions
1026		*/
1027		public int getQueuedSubmissionCount() {
1028		return submissionQueue.size();
1029		}
1030
1031		/**
1032	<	* Returns true if there are any tasks submitted to this pool
1033	<	* that have not yet begun executing.
1034	<	* @return <code>true</code> if there are any queued submissions.
1032	>	* Returns {@code true} if there are any tasks submitted to this
1033	>	* pool that have not yet begun executing.
1034	>	*
1035	>	* @return {@code true} if there are any queued submissions
1036		*/
1037		public boolean hasQueuedSubmissions() {
1038		return !submissionQueue.isEmpty();
#	Line 899 | Line 1042 | public class ForkJoinPool extends Abstra
1042		* Removes and returns the next unexecuted submission if one is
1043		* available.  This method may be useful in extensions to this
1044		* class that re-assign work in systems with multiple pools.
1045	<	* @return the next submission, or null if none
1045	>	*
1046	>	* @return the next submission, or {@code null} if none
1047		*/
1048		protected ForkJoinTask<?> pollSubmission() {
1049		return submissionQueue.poll();
1050		}
1051
1052		/**
1053	+	* Removes all available unexecuted submitted and forked tasks
1054	+	* from scheduling queues and adds them to the given collection,
1055	+	* without altering their execution status. These may include
1056	+	* artificially generated or wrapped tasks. This method is
1057	+	* designed to be invoked only when the pool is known to be
1058	+	* quiescent. Invocations at other times may not remove all
1059	+	* tasks. A failure encountered while attempting to add elements
1060	+	* to collection {@code c} may result in elements being in
1061	+	* neither, either or both collections when the associated
1062	+	* exception is thrown.  The behavior of this operation is
1063	+	* undefined if the specified collection is modified while the
1064	+	* operation is in progress.
1065	+	*
1066	+	* @param c the collection to transfer elements into
1067	+	* @return the number of elements transferred
1068	+	*/
1069	+	protected int drainTasksTo(Collection<? super ForkJoinTask<?>> c) {
1070	+	int n = submissionQueue.drainTo(c);
1071	+	ForkJoinWorkerThread[] ws = workers;
1072	+	if (ws != null) {
1073	+	for (int i = 0; i < ws.length; ++i) {
1074	+	ForkJoinWorkerThread w = ws[i];
1075	+	if (w != null)
1076	+	n += w.drainTasksTo(c);
1077	+	}
1078	+	}
1079	+	return n;
1080	+	}
1081	+
1082	+	/**
1083		* Returns a string identifying this pool, as well as its state,
1084		* including indications of run state, parallelism level, and
1085		* worker and task counts.
#	Line 949 | Line 1123 | public class ForkJoinPool extends Abstra
1123		* Invocation has no additional effect if already shut down.
1124		* Tasks that are in the process of being submitted concurrently
1125		* during the course of this method may or may not be rejected.
1126	+	*
1127		* @throws SecurityException if a security manager exists and
1128		*         the caller is not permitted to modify threads
1129		*         because it does not hold {@link
1130	<	*         java.lang.RuntimePermission}<code>("modifyThread")</code>,
1130	>	*         java.lang.RuntimePermission}{@code ("modifyThread")}
1131		*/
1132		public void shutdown() {
1133		checkPermission();
1134		transitionRunStateTo(SHUTDOWN);
1135	<	if (canTerminateOnShutdown(runControl))
1135	>	if (canTerminateOnShutdown(runControl)) {
1136	>	if (workers == null) { // shutting down before workers created
1137	>	final ReentrantLock lock = this.workerLock;
1138	>	lock.lock();
1139	>	try {
1140	>	if (workers == null) {
1141	>	terminate();
1142	>	transitionRunStateTo(TERMINATED);
1143	>	termination.signalAll();
1144	>	}
1145	>	} finally {
1146	>	lock.unlock();
1147	>	}
1148	>	}
1149		terminateOnShutdown();
1150	+	}
1151		}
1152
1153		/**
1154	<	* Attempts to stop all actively executing tasks, and cancels all
1155	<	* waiting tasks.  Tasks that are in the process of being
1156	<	* submitted or executed concurrently during the course of this
1157	<	* method may or may not be rejected. Unlike some other executors,
1158	<	* this method cancels rather than collects non-executed tasks,
1159	<	* so always returns an empty list.
1154	>	* Attempts to cancel and/or stop all tasks, and reject all
1155	>	* subsequently submitted tasks.  Tasks that are in the process of
1156	>	* being submitted or executed concurrently during the course of
1157	>	* this method may or may not be rejected. This method cancels
1158	>	* both existing and unexecuted tasks, in order to permit
1159	>	* termination in the presence of task dependencies. So the method
1160	>	* always returns an empty list (unlike the case for some other
1161	>	* Executors).
1162	>	*
1163		* @return an empty list
1164		* @throws SecurityException if a security manager exists and
1165		*         the caller is not permitted to modify threads
1166		*         because it does not hold {@link
1167	<	*         java.lang.RuntimePermission}<code>("modifyThread")</code>,
1167	>	*         java.lang.RuntimePermission}{@code ("modifyThread")}
1168		*/
1169		public List<Runnable> shutdownNow() {
1170		checkPermission();
#	Line 981 | Line 1173 | public class ForkJoinPool extends Abstra
1173		}
1174
1175		/**
1176	<	* Returns <code>true</code> if all tasks have completed following shut down.
1176	>	* Returns {@code true} if all tasks have completed following shut down.
1177		*
1178	<	* @return <code>true</code> if all tasks have completed following shut down
1178	>	* @return {@code true} if all tasks have completed following shut down
1179		*/
1180		public boolean isTerminated() {
1181		return runStateOf(runControl) == TERMINATED;
1182		}
1183
1184		/**
1185	<	* Returns <code>true</code> if the process of termination has
1186	<	* commenced but possibly not yet completed.
1185	>	* Returns {@code true} if the process of termination has
1186	>	* commenced but not yet completed.  This method may be useful for
1187	>	* debugging. A return of {@code true} reported a sufficient
1188	>	* period after shutdown may indicate that submitted tasks have
1189	>	* ignored or suppressed interruption, causing this executor not
1190	>	* to properly terminate.
1191		*
1192	<	* @return <code>true</code> if terminating
1192	>	* @return {@code true} if terminating but not yet terminated
1193		*/
1194		public boolean isTerminating() {
1195	<	return runStateOf(runControl) >= TERMINATING;
1195	>	return runStateOf(runControl) == TERMINATING;
1196		}
1197
1198		/**
1199	<	* Returns <code>true</code> if this pool has been shut down.
1199	>	* Returns {@code true} if this pool has been shut down.
1200		*
1201	<	* @return <code>true</code> if this pool has been shut down
1201	>	* @return {@code true} if this pool has been shut down
1202		*/
1203		public boolean isShutdown() {
1204		return runStateOf(runControl) >= SHUTDOWN;
1205		}
1206
1207		/**
1208	+	* Returns true if pool is not terminating or terminated.
1209	+	* Used internally to suppress execution when terminating.
1210	+	*/
1211	+	final boolean isProcessingTasks() {
1212	+	return runStateOf(runControl) < TERMINATING;
1213	+	}
1214	+
1215	+	/**
1216		* Blocks until all tasks have completed execution after a shutdown
1217		* request, or the timeout occurs, or the current thread is
1218		* interrupted, whichever happens first.
1219		*
1220		* @param timeout the maximum time to wait
1221		* @param unit the time unit of the timeout argument
1222	<	* @return <code>true</code> if this executor terminated and
1223	<	*         <code>false</code> if the timeout elapsed before termination
1222	>	* @return {@code true} if this executor terminated and
1223	>	*         {@code false} if the timeout elapsed before termination
1224		* @throws InterruptedException if interrupted while waiting
1225		*/
1226		public boolean awaitTermination(long timeout, TimeUnit unit)
#	Line 1040 | Line 1244 | public class ForkJoinPool extends Abstra
1244		// Shutdown and termination support
1245
1246		/**
1247	<	* Callback from terminating worker. Null out the corresponding
1248	<	* workers slot, and if terminating, try to terminate, else try to
1249	<	* shrink workers array.
1247	>	* Callback from terminating worker. Nulls out the corresponding
1248	>	* workers slot, and if terminating, tries to terminate; else
1249	>	* tries to shrink workers array.
1250	>	*
1251		* @param w the worker
1252		*/
1253		final void workerTerminated(ForkJoinWorkerThread w) {
#	Line 1052 | Line 1257 | public class ForkJoinPool extends Abstra
1257		lock.lock();
1258		try {
1259		ForkJoinWorkerThread[] ws = workers;
1260	<	int idx = w.poolIndex;
1261	<	if (idx >= 0 && idx < ws.length && ws[idx] == w)
1262	<	ws[idx] = null;
1263	<	if (totalCountOf(workerCounts) == 0) {
1264	<	terminate(); // no-op if already terminating
1265	<	transitionRunStateTo(TERMINATED);
1266	<	termination.signalAll();
1267	<	}
1268	<	else if (!isTerminating()) {
1269	<	tryShrinkWorkerArray();
1270	<	tryResumeSpare(true); // allow replacement
1260	>	if (ws != null) {
1261	>	int idx = w.poolIndex;
1262	>	if (idx >= 0 && idx < ws.length && ws[idx] == w)
1263	>	ws[idx] = null;
1264	>	if (totalCountOf(workerCounts) == 0) {
1265	>	terminate(); // no-op if already terminating
1266	>	transitionRunStateTo(TERMINATED);
1267	>	termination.signalAll();
1268	>	}
1269	>	else if (isProcessingTasks()) {
1270	>	tryShrinkWorkerArray();
1271	>	tryResumeSpare(true); // allow replacement
1272	>	}
1273		}
1274		} finally {
1275		lock.unlock();
1276		}
1277	<	signalIdleWorkers(false);
1277	>	signalIdleWorkers();
1278		}
1279
1280		/**
1281	<	* Initiate termination.
1281	>	* Initiates termination.
1282		*/
1283		private void terminate() {
1284		if (transitionRunStateTo(TERMINATING)) {
1285		stopAllWorkers();
1286		resumeAllSpares();
1287	<	signalIdleWorkers(true);
1287	>	signalIdleWorkers();
1288		cancelQueuedSubmissions();
1289		cancelQueuedWorkerTasks();
1290		interruptUnterminatedWorkers();
1291	<	signalIdleWorkers(true); // resignal after interrupt
1291	>	signalIdleWorkers(); // resignal after interrupt
1292		}
1293		}
1294
1295		/**
1296	<	* Possibly terminate when on shutdown state
1296	>	* Possibly terminates when on shutdown state.
1297		*/
1298		private void terminateOnShutdown() {
1299		if (!hasQueuedSubmissions() && canTerminateOnShutdown(runControl))
#	Line 1094 | Line 1301 | public class ForkJoinPool extends Abstra
1301		}
1302
1303		/**
1304	<	* Clear out and cancel submissions
1304	>	* Clears out and cancels submissions.
1305		*/
1306		private void cancelQueuedSubmissions() {
1307		ForkJoinTask<?> task;
#	Line 1103 | Line 1310 | public class ForkJoinPool extends Abstra
1310		}
1311
1312		/**
1313	<	* Clean out worker queues.
1313	>	* Cleans out worker queues.
1314		*/
1315		private void cancelQueuedWorkerTasks() {
1316		final ReentrantLock lock = this.workerLock;
1317		lock.lock();
1318		try {
1319		ForkJoinWorkerThread[] ws = workers;
1320	<	for (int i = 0; i < ws.length; ++i) {
1321	<	ForkJoinWorkerThread t = ws[i];
1322	<	if (t != null)
1323	<	t.cancelTasks();
1320	>	if (ws != null) {
1321	>	for (int i = 0; i < ws.length; ++i) {
1322	>	ForkJoinWorkerThread t = ws[i];
1323	>	if (t != null)
1324	>	t.cancelTasks();
1325	>	}
1326		}
1327		} finally {
1328		lock.unlock();
#	Line 1121 | Line 1330 | public class ForkJoinPool extends Abstra
1330		}
1331
1332		/**
1333	<	* Set each worker's status to terminating. Requires lock to avoid
1334	<	* conflicts with add/remove
1333	>	* Sets each worker's status to terminating. Requires lock to avoid
1334	>	* conflicts with add/remove.
1335		*/
1336		private void stopAllWorkers() {
1337		final ReentrantLock lock = this.workerLock;
1338		lock.lock();
1339		try {
1340		ForkJoinWorkerThread[] ws = workers;
1341	<	for (int i = 0; i < ws.length; ++i) {
1342	<	ForkJoinWorkerThread t = ws[i];
1343	<	if (t != null)
1344	<	t.shutdownNow();
1341	>	if (ws != null) {
1342	>	for (int i = 0; i < ws.length; ++i) {
1343	>	ForkJoinWorkerThread t = ws[i];
1344	>	if (t != null)
1345	>	t.shutdownNow();
1346	>	}
1347		}
1348		} finally {
1349		lock.unlock();
#	Line 1140 | Line 1351 | public class ForkJoinPool extends Abstra
1351		}
1352
1353		/**
1354	<	* Interrupt all unterminated workers.  This is not required for
1354	>	* Interrupts all unterminated workers.  This is not required for
1355		* sake of internal control, but may help unstick user code during
1356		* shutdown.
1357		*/
#	Line 1149 | Line 1360 | public class ForkJoinPool extends Abstra
1360		lock.lock();
1361		try {
1362		ForkJoinWorkerThread[] ws = workers;
1363	<	for (int i = 0; i < ws.length; ++i) {
1364	<	ForkJoinWorkerThread t = ws[i];
1365	<	if (t != null && !t.isTerminated()) {
1366	<	try {
1367	<	t.interrupt();
1368	<	} catch (SecurityException ignore) {
1363	>	if (ws != null) {
1364	>	for (int i = 0; i < ws.length; ++i) {
1365	>	ForkJoinWorkerThread t = ws[i];
1366	>	if (t != null && !t.isTerminated()) {
1367	>	try {
1368	>	t.interrupt();
1369	>	} catch (SecurityException ignore) {
1370	>	}
1371		}
1372		}
1373		}
#	Line 1165 | Line 1378 | public class ForkJoinPool extends Abstra
1378
1379
1380		/*
1381	<	* Nodes for event barrier to manage idle threads.
1381	>	* Nodes for event barrier to manage idle threads.  Queue nodes
1382	>	* are basic Treiber stack nodes, also used for spare stack.
1383		*
1384		* The event barrier has an event count and a wait queue (actually
1385		* a Treiber stack).  Workers are enabled to look for work when
1386	<	* the eventCount is incremented. If they fail to find some,
1387	<	* they may wait for next count. Synchronization events occur only
1388	<	* in enough contexts to maintain overall liveness:
1386	>	* the eventCount is incremented. If they fail to find work, they
1387	>	* may wait for next count. Upon release, threads help others wake
1388	>	* up.
1389	>	*
1390	>	* Synchronization events occur only in enough contexts to
1391	>	* maintain overall liveness:
1392		*
1393		*   - Submission of a new task to the pool
1394	<	*   - Creation or termination of a worker
1394	>	*   - Resizes or other changes to the workers array
1395		*   - pool termination
1396		*   - A worker pushing a task on an empty queue
1397		*
1398	<	* The last case (pushing a task) occurs often enough, and is
1399	<	* heavy enough compared to simple stack pushes to require some
1400	<	* special handling: Method signalNonEmptyWorkerQueue returns
1401	<	* without advancing count if the queue appears to be empty.  This
1402	<	* would ordinarily result in races causing some queued waiters
1403	<	* not to be woken up. To avoid this, a worker in sync
1404	<	* rescans for tasks after being enqueued if it was the first to
1405	<	* enqueue, and aborts the wait if finding one, also helping to
1406	<	* signal others. This works well because the worker has nothing
1407	<	* better to do anyway, and so might as well help alleviate the
1408	<	* overhead and contention on the threads actually doing work.
1409	<	*
1410	<	* Queue nodes are basic Treiber stack nodes, also used for spare
1411	<	* stack.
1398	>	* The case of pushing a task occurs often enough, and is heavy
1399	>	* enough compared to simple stack pushes, to require special
1400	>	* handling: Method signalWork returns without advancing count if
1401	>	* the queue appears to be empty.  This would ordinarily result in
1402	>	* races causing some queued waiters not to be woken up. To avoid
1403	>	* this, the first worker enqueued in method sync (see
1404	>	* syncIsReleasable) rescans for tasks after being enqueued, and
1405	>	* helps signal if any are found. This works well because the
1406	>	* worker has nothing better to do, and so might as well help
1407	>	* alleviate the overhead and contention on the threads actually
1408	>	* doing work.  Also, since event counts increments on task
1409	>	* availability exist to maintain liveness (rather than to force
1410	>	* refreshes etc), it is OK for callers to exit early if
1411	>	* contending with another signaller.
1412		*/
1413		static final class WaitQueueNode {
1414		WaitQueueNode next; // only written before enqueued
1415		volatile ForkJoinWorkerThread thread; // nulled to cancel wait
1416		final long count; // unused for spare stack
1417	<	WaitQueueNode(ForkJoinWorkerThread w, long c) {
1417	>
1418	>	WaitQueueNode(long c, ForkJoinWorkerThread w) {
1419		count = c;
1420		thread = w;
1421		}
1422	<	final boolean signal() {
1422	>
1423	>	/**
1424	>	* Wakes up waiter, returning false if known to already
1425	>	*/
1426	>	boolean signal() {
1427		ForkJoinWorkerThread t = thread;
1428	+	if (t == null)
1429	+	return false;
1430		thread = null;
1431	<	if (t != null) {
1432	<	LockSupport.unpark(t);
1433	<	return true;
1431	>	LockSupport.unpark(t);
1432	>	return true;
1433	>	}
1434	>
1435	>	/**
1436	>	* Awaits release on sync.
1437	>	*/
1438	>	void awaitSyncRelease(ForkJoinPool p) {
1439	>	while (thread != null && !p.syncIsReleasable(this))
1440	>	LockSupport.park(this);
1441	>	}
1442	>
1443	>	/**
1444	>	* Awaits resumption as spare.
1445	>	*/
1446	>	void awaitSpareRelease() {
1447	>	while (thread != null) {
1448	>	if (!Thread.interrupted())
1449	>	LockSupport.park(this);
1450		}
1211	–	return false;
1451		}
1452		}
1453
1454		/**
1455	<	* Release at least one thread waiting for event count to advance,
1456	<	* if one exists. If initial attempt fails, release all threads.
1457	<	* @param all if false, at first try to only release one thread
1458	<	* @return current event
1455	>	* Ensures that no thread is waiting for count to advance from the
1456	>	* current value of eventCount read on entry to this method, by
1457	>	* releasing waiting threads if necessary.
1458	>	*
1459	>	* @return the count
1460		*/
1461	<	private long releaseIdleWorkers(boolean all) {
1462	<	long c;
1463	<	for (;;) {
1464	<	WaitQueueNode q = barrierStack;
1465	<	c = eventCount;
1226	<	long qc;
1227	<	if (q == null || (qc = q.count) >= c)
1228	<	break;
1229	<	if (!all) {
1230	<	if (casBarrierStack(q, q.next) && q.signal())
1231	<	break;
1232	<	all = true;
1233	<	}
1234	<	else if (casBarrierStack(q, null)) {
1461	>	final long ensureSync() {
1462	>	long c = eventCount;
1463	>	WaitQueueNode q;
1464	>	while ((q = syncStack) != null && q.count < c) {
1465	>	if (casBarrierStack(q, null)) {
1466		do {
1467	<	q.signal();
1467	>	q.signal();
1468		} while ((q = q.next) != null);
1469		break;
1470		}
#	Line 1242 | Line 1473 | public class ForkJoinPool extends Abstra
1473		}
1474
1475		/**
1476	<	* Returns current barrier event count
1246	<	* @return current barrier event count
1476	>	* Increments event count and releases waiting threads.
1477		*/
1478	<	final long getEventCount() {
1249	<	long ec = eventCount;
1250	<	releaseIdleWorkers(true); // release to ensure accurate result
1251	<	return ec;
1252	<	}
1253	<
1254	<	/**
1255	<	* Increment event count and release at least one waiting thread,
1256	<	* if one exists (released threads will in turn wake up others).
1257	<	* @param all if true, try to wake up all
1258	<	*/
1259	<	final void signalIdleWorkers(boolean all) {
1478	>	private void signalIdleWorkers() {
1479		long c;
1480	<	do;while (!casEventCount(c = eventCount, c+1));
1481	<	releaseIdleWorkers(all);
1480	>	do {} while (!casEventCount(c = eventCount, c+1));
1481	>	ensureSync();
1482		}
1483
1484		/**
1485	<	* Wake up threads waiting to steal a task. Because method
1486	<	* sync rechecks availability, it is OK to only proceed if
1487	<	* queue appears to be non-empty.
1485	>	* Signals threads waiting to poll a task. Because method sync
1486	>	* rechecks availability, it is OK to only proceed if queue
1487	>	* appears to be non-empty, and OK to skip under contention to
1488	>	* increment count (since some other thread succeeded).
1489		*/
1490	<	final void signalNonEmptyWorkerQueue() {
1271	<	// If CAS fails another signaller must have succeeded
1490	>	final void signalWork() {
1491		long c;
1492	<	if (barrierStack != null && casEventCount(c = eventCount, c+1))
1493	<	releaseIdleWorkers(false);
1492	>	WaitQueueNode q;
1493	>	if (syncStack != null &&
1494	>	casEventCount(c = eventCount, c+1) &&
1495	>	(((q = syncStack) != null && q.count <= c) &&
1496	>	(!casBarrierStack(q, q.next) || !q.signal())))
1497	>	ensureSync();
1498		}
1499
1500		/**
1501	<	* Waits until event count advances from count, or some thread is
1502	<	* waiting on a previous count, or there is stealable work
1503	<	* available. Help wake up others on release.
1501	>	* Waits until event count advances from last value held by
1502	>	* caller, or if excess threads, caller is resumed as spare, or
1503	>	* caller or pool is terminating. Updates caller's event on exit.
1504	>	*
1505		* @param w the calling worker thread
1282	–	* @param prev previous value returned by sync (or 0)
1283	–	* @return current event count
1506		*/
1507	<	final long sync(ForkJoinWorkerThread w, long prev) {
1508	<	updateStealCount(w);
1507	>	final void sync(ForkJoinWorkerThread w) {
1508	>	updateStealCount(w); // Transfer w's count while it is idle
1509
1510	<	while (!w.isShutdown() && !isTerminating() &&
1511	<	(parallelism >= runningCountOf(workerCounts) ||
1290	<	!suspendIfSpare(w))) { // prefer suspend to waiting here
1510	>	while (!w.isShutdown() && isProcessingTasks() && !suspendIfSpare(w)) {
1511	>	long prev = w.lastEventCount;
1512		WaitQueueNode node = null;
1513	<	boolean queued = false;
1514	<	for (;;) {
1515	<	if (!queued) {
1516	<	if (eventCount != prev)
1517	<	break;
1518	<	WaitQueueNode h = barrierStack;
1519	<	if (h != null && h.count != prev)
1299	<	break; // release below and maybe retry
1300	<	if (node == null)
1301	<	node = new WaitQueueNode(w, prev);
1302	<	queued = casBarrierStack(node.next = h, node);
1303	<	}
1304	<	else if (Thread.interrupted() ||
1305	<	node.thread == null ||
1306	<	(node.next == null && w.prescan()) ||
1307	<	eventCount != prev) {
1308	<	node.thread = null;
1309	<	if (eventCount == prev) // help trigger
1310	<	casEventCount(prev, prev+1);
1513	>	WaitQueueNode h;
1514	>	while (eventCount == prev &&
1515	>	((h = syncStack) == null || h.count == prev)) {
1516	>	if (node == null)
1517	>	node = new WaitQueueNode(prev, w);
1518	>	if (casBarrierStack(node.next = h, node)) {
1519	>	node.awaitSyncRelease(this);
1520		break;
1521		}
1313	–	else
1314	–	LockSupport.park(this);
1522		}
1523	+	long ec = ensureSync();
1524	+	if (ec != prev) {
1525	+	w.lastEventCount = ec;
1526	+	break;
1527	+	}
1528	+	}
1529	+	}
1530	+
1531	+	/**
1532	+	* Returns {@code true} if worker waiting on sync can proceed:
1533	+	*  - on signal (thread == null)
1534	+	*  - on event count advance (winning race to notify vs signaller)
1535	+	*  - on interrupt
1536	+	*  - if the first queued node, we find work available
1537	+	* If node was not signalled and event count not advanced on exit,
1538	+	* then we also help advance event count.
1539	+	*
1540	+	* @return {@code true} if node can be released
1541	+	*/
1542	+	final boolean syncIsReleasable(WaitQueueNode node) {
1543	+	long prev = node.count;
1544	+	if (!Thread.interrupted() && node.thread != null &&
1545	+	(node.next != null ||
1546	+	!ForkJoinWorkerThread.hasQueuedTasks(workers)) &&
1547	+	eventCount == prev)
1548	+	return false;
1549	+	if (node.thread != null) {
1550	+	node.thread = null;
1551		long ec = eventCount;
1552	<	if (releaseIdleWorkers(false) != prev)
1553	<	return ec;
1552	>	if (prev <= ec) // help signal
1553	>	casEventCount(ec, ec+1);
1554		}
1555	<	return prev; // return old count if aborted
1555	>	return true;
1556	>	}
1557	>
1558	>	/**
1559	>	* Returns {@code true} if a new sync event occurred since last
1560	>	* call to sync or this method, if so, updating caller's count.
1561	>	*/
1562	>	final boolean hasNewSyncEvent(ForkJoinWorkerThread w) {
1563	>	long lc = w.lastEventCount;
1564	>	long ec = ensureSync();
1565	>	if (ec == lc)
1566	>	return false;
1567	>	w.lastEventCount = ec;
1568	>	return true;
1569		}
1570
1571		//  Parallelism maintenance
1572
1573		/**
1574	<	* Decrement running count; if too low, add spare.
1574	>	* Decrements running count; if too low, adds spare.
1575		*
1576		* Conceptually, all we need to do here is add or resume a
1577		* spare thread when one is about to block (and remove or
1578		* suspend it later when unblocked -- see suspendIfSpare).
1579		* However, implementing this idea requires coping with
1580	<	* several problems: We have imperfect information about the
1580	>	* several problems: we have imperfect information about the
1581		* states of threads. Some count updates can and usually do
1582		* lag run state changes, despite arrangements to keep them
1583		* accurate (for example, when possible, updating counts
#	Line 1343 | Line 1591 | public class ForkJoinPool extends Abstra
1591		* only be suspended or removed when they are idle, not
1592		* immediately when they aren't needed. So adding threads will
1593		* raise parallelism level for longer than necessary.  Also,
1594	<	* FJ applications often enounter highly transient peaks when
1594	>	* FJ applications often encounter highly transient peaks when
1595		* many threads are blocked joining, but for less time than it
1596		* takes to create or resume spares.
1597		*
#	Line 1352 | Line 1600 | public class ForkJoinPool extends Abstra
1600		* target counts, else create only to avoid starvation
1601		* @return true if joinMe known to be done
1602		*/
1603	<	final boolean preJoin(ForkJoinTask<?> joinMe, boolean maintainParallelism) {
1603	>	final boolean preJoin(ForkJoinTask<?> joinMe,
1604	>	boolean maintainParallelism) {
1605		maintainParallelism &= maintainsParallelism; // overrride
1606		boolean dec = false;  // true when running count decremented
1607		while (spareStack == null || !tryResumeSpare(dec)) {
1608		int counts = workerCounts;
1609	<	if (dec || (dec = casWorkerCounts(counts, --counts))) { // CAS cheat
1609	>	if (dec || (dec = casWorkerCounts(counts, --counts))) {
1610		if (!needSpare(counts, maintainParallelism))
1611		break;
1612		if (joinMe.status < 0)
#	Line 1372 | Line 1621 | public class ForkJoinPool extends Abstra
1621		/**
1622		* Same idea as preJoin
1623		*/
1624	<	final boolean preBlock(ManagedBlocker blocker, boolean maintainParallelism){
1624	>	final boolean preBlock(ManagedBlocker blocker,
1625	>	boolean maintainParallelism) {
1626		maintainParallelism &= maintainsParallelism;
1627		boolean dec = false;
1628		while (spareStack == null || !tryResumeSpare(dec)) {
#	Line 1390 | Line 1640 | public class ForkJoinPool extends Abstra
1640		}
1641
1642		/**
1643	<	* Returns true if a spare thread appears to be needed.  If
1644	<	* maintaining parallelism, returns true when the deficit in
1643	>	* Returns {@code true} if a spare thread appears to be needed.
1644	>	* If maintaining parallelism, returns true when the deficit in
1645		* running threads is more than the surplus of total threads, and
1646		* there is apparently some work to do.  This self-limiting rule
1647		* means that the more threads that have already been added, the
1648		* less parallelism we will tolerate before adding another.
1649	+	*
1650		* @param counts current worker counts
1651		* @param maintainParallelism try to maintain parallelism
1652		*/
#	Line 1408 | Line 1659 | public class ForkJoinPool extends Abstra
1659		return (tc < maxPoolSize &&
1660		(rc == 0 || totalSurplus < 0 ||
1661		(maintainParallelism &&
1662	<	runningDeficit > totalSurplus && mayHaveQueuedWork())));
1662	>	runningDeficit > totalSurplus &&
1663	>	ForkJoinWorkerThread.hasQueuedTasks(workers))));
1664		}
1665
1666		/**
1667	<	* Returns true if at least one worker queue appears to be
1668	<	* nonempty. This is expensive but not often called. It is not
1669	<	* critical that this be accurate, but if not, more or fewer
1418	<	* running threads than desired might be maintained.
1419	<	*/
1420	<	private boolean mayHaveQueuedWork() {
1421	<	ForkJoinWorkerThread[] ws = workers;
1422	<	int len = ws.length;
1423	<	ForkJoinWorkerThread v;
1424	<	for (int i = 0; i < len; ++i) {
1425	<	if ((v = ws[i]) != null && v.getRawQueueSize() > 0) {
1426	<	releaseIdleWorkers(false); // help wake up stragglers
1427	<	return true;
1428	<	}
1429	<	}
1430	<	return false;
1431	<	}
1432	<
1433	<	/**
1434	<	* Add a spare worker if lock available and no more than the
1435	<	* expected numbers of threads exist
1667	>	* Adds a spare worker if lock available and no more than the
1668	>	* expected numbers of threads exist.
1669	>	*
1670		* @return true if successful
1671		*/
1672		private boolean tryAddSpare(int expectedCounts) {
#	Line 1465 | Line 1699 | public class ForkJoinPool extends Abstra
1699		}
1700
1701		/**
1702	<	* Add the kth spare worker. On entry, pool coounts are already
1702	>	* Adds the kth spare worker. On entry, pool counts are already
1703		* adjusted to reflect addition.
1704		*/
1705		private void createAndStartSpare(int k) {
#	Line 1477 | Line 1711 | public class ForkJoinPool extends Abstra
1711		for (k = 0; k < len && ws[k] != null; ++k)
1712		;
1713		}
1714	<	if (k < len && !isTerminating() && (w = createWorker(k)) != null) {
1714	>	if (k < len && isProcessingTasks() && (w = createWorker(k)) != null) {
1715		ws[k] = w;
1716		w.start();
1717		}
1718		else
1719		updateWorkerCount(-1); // adjust on failure
1720	<	signalIdleWorkers(false);
1720	>	signalIdleWorkers();
1721		}
1722
1723		/**
1724	<	* Suspend calling thread w if there are excess threads.  Called
1725	<	* only from sync.  Spares are enqueued in a Treiber stack
1726	<	* using the same WaitQueueNodes as barriers.  They are resumed
1727	<	* mainly in preJoin, but are also woken on pool events that
1728	<	* require all threads to check run state.
1724	>	* Suspends calling thread w if there are excess threads.  Called
1725	>	* only from sync.  Spares are enqueued in a Treiber stack using
1726	>	* the same WaitQueueNodes as barriers.  They are resumed mainly
1727	>	* in preJoin, but are also woken on pool events that require all
1728	>	* threads to check run state.
1729	>	*
1730		* @param w the caller
1731		*/
1732		private boolean suspendIfSpare(ForkJoinWorkerThread w) {
#	Line 1499 | Line 1734 | public class ForkJoinPool extends Abstra
1734		int s;
1735		while (parallelism < runningCountOf(s = workerCounts)) {
1736		if (node == null)
1737	<	node = new WaitQueueNode(w, 0);
1737	>	node = new WaitQueueNode(0, w);
1738		if (casWorkerCounts(s, s-1)) { // representation-dependent
1739		// push onto stack
1740	<	do;while (!casSpareStack(node.next = spareStack, node));
1506	<
1740	>	do {} while (!casSpareStack(node.next = spareStack, node));
1741		// block until released by resumeSpare
1742	<	while (node.thread != null) {
1509	<	if (!Thread.interrupted())
1510	<	LockSupport.park(this);
1511	<	}
1512	<	w.activate(); // help warm up
1742	>	node.awaitSpareRelease();
1743		return true;
1744		}
1745		}
#	Line 1517 | Line 1747 | public class ForkJoinPool extends Abstra
1747		}
1748
1749		/**
1750	<	* Try to pop and resume a spare thread.
1750	>	* Tries to pop and resume a spare thread.
1751	>	*
1752		* @param updateCount if true, increment running count on success
1753		* @return true if successful
1754		*/
#	Line 1535 | Line 1766 | public class ForkJoinPool extends Abstra
1766		}
1767
1768		/**
1769	<	* Pop and resume all spare threads. Same idea as
1770	<	* releaseIdleWorkers.
1769	>	* Pops and resumes all spare threads. Same idea as ensureSync.
1770	>	*
1771		* @return true if any spares released
1772		*/
1773		private boolean resumeAllSpares() {
#	Line 1554 | Line 1785 | public class ForkJoinPool extends Abstra
1785		}
1786
1787		/**
1788	<	* Pop and shutdown excessive spare threads. Call only while
1788	>	* Pops and shuts down excessive spare threads. Call only while
1789		* holding lock. This is not guaranteed to eliminate all excess
1790		* threads, only those suspended as spares, which are the ones
1791		* unlikely to be needed in the future.
#	Line 1577 | Line 1808 | public class ForkJoinPool extends Abstra
1808		}
1809
1810		/**
1580	–	* Returns approximate number of spares, just for diagnostics.
1581	–	*/
1582	–	private int countSpares() {
1583	–	int sum = 0;
1584	–	for (WaitQueueNode q = spareStack; q != null; q = q.next)
1585	–	++sum;
1586	–	return sum;
1587	–	}
1588	–
1589	–	/**
1811		* Interface for extending managed parallelism for tasks running
1812	<	* in ForkJoinPools. A ManagedBlocker provides two methods.
1813	<	* Method <code>isReleasable</code> must return true if blocking is not
1814	<	* necessary. Method <code>block</code> blocks the current thread
1815	<	* if necessary (perhaps internally invoking isReleasable before
1816	<	* actually blocking.).
1812	>	* in {@link ForkJoinPool}s.
1813	>	*
1814	>	* <p>A {@code ManagedBlocker} provides two methods.
1815	>	* Method {@code isReleasable} must return {@code true} if
1816	>	* blocking is not necessary. Method {@code block} blocks the
1817	>	* current thread if necessary (perhaps internally invoking
1818	>	* {@code isReleasable} before actually blocking).
1819	>	*
1820		* <p>For example, here is a ManagedBlocker based on a
1821		* ReentrantLock:
1822	<	* <pre>
1823	<	*   class ManagedLocker implements ManagedBlocker {
1824	<	*     final ReentrantLock lock;
1825	<	*     boolean hasLock = false;
1826	<	*     ManagedLocker(ReentrantLock lock) { this.lock = lock; }
1827	<	*     public boolean block() {
1828	<	*        if (!hasLock)
1829	<	*           lock.lock();
1830	<	*        return true;
1831	<	*     }
1832	<	*     public boolean isReleasable() {
1833	<	*        return hasLock || (hasLock = lock.tryLock());
1610	<	*     }
1822	>	*  <pre> {@code
1823	>	* class ManagedLocker implements ManagedBlocker {
1824	>	*   final ReentrantLock lock;
1825	>	*   boolean hasLock = false;
1826	>	*   ManagedLocker(ReentrantLock lock) { this.lock = lock; }
1827	>	*   public boolean block() {
1828	>	*     if (!hasLock)
1829	>	*       lock.lock();
1830	>	*     return true;
1831	>	*   }
1832	>	*   public boolean isReleasable() {
1833	>	*     return hasLock || (hasLock = lock.tryLock());
1834		*   }
1835	<	* </pre>
1835	>	* }}</pre>
1836		*/
1837		public static interface ManagedBlocker {
1838		/**
1839		* Possibly blocks the current thread, for example waiting for
1840		* a lock or condition.
1841	<	* @return true if no additional blocking is necessary (i.e.,
1842	<	* if isReleasable would return true).
1841	>	*
1842	>	* @return {@code true} if no additional blocking is necessary
1843	>	* (i.e., if isReleasable would return true)
1844		* @throws InterruptedException if interrupted while waiting
1845	<	* (the method is not required to do so, but is allowe to).
1845	>	* (the method is not required to do so, but is allowed to)
1846		*/
1847		boolean block() throws InterruptedException;
1848
1849		/**
1850	<	* Returns true if blocking is unnecessary.
1850	>	* Returns {@code true} if blocking is unnecessary.
1851		*/
1852		boolean isReleasable();
1853		}
1854
1855		/**
1856		* Blocks in accord with the given blocker.  If the current thread
1857	<	* is a ForkJoinWorkerThread, this method possibly arranges for a
1858	<	* spare thread to be activated if necessary to ensure parallelism
1859	<	* while the current thread is blocked.  If
1860	<	* <code>maintainParallelism</code> is true and the pool supports
1861	<	* it ({@link #getMaintainsParallelism}), this method attempts to
1862	<	* maintain the pool's nominal parallelism. Otherwise if activates
1863	<	* a thread only if necessary to avoid complete starvation. This
1864	<	* option may be preferable when blockages use timeouts, or are
1865	<	* almost always brief.
1866	<	*
1867	<	* <p> If the caller is not a ForkJoinTask, this method is behaviorally
1868	<	* equivalent to
1869	<	* <pre>
1870	<	*   while (!blocker.isReleasable())
1871	<	*      if (blocker.block())
1872	<	*         return;
1873	<	* </pre>
1874	<	* If the caller is a ForkJoinTask, then the pool may first
1875	<	* be expanded to ensure parallelism, and later adjusted.
1857	>	* is a {@link ForkJoinWorkerThread}, this method possibly
1858	>	* arranges for a spare thread to be activated if necessary to
1859	>	* ensure parallelism while the current thread is blocked.
1860	>	*
1861	>	* <p>If {@code maintainParallelism} is {@code true} and the pool
1862	>	* supports it ({@link #getMaintainsParallelism}), this method
1863	>	* attempts to maintain the pool's nominal parallelism. Otherwise
1864	>	* it activates a thread only if necessary to avoid complete
1865	>	* starvation. This option may be preferable when blockages use
1866	>	* timeouts, or are almost always brief.
1867	>	*
1868	>	* <p>If the caller is not a {@link ForkJoinTask}, this method is
1869	>	* behaviorally equivalent to
1870	>	*  <pre> {@code
1871	>	* while (!blocker.isReleasable())
1872	>	*   if (blocker.block())
1873	>	*     return;
1874	>	* }</pre>
1875	>	*
1876	>	* If the caller is a {@code ForkJoinTask}, then the pool may
1877	>	* first be expanded to ensure parallelism, and later adjusted.
1878		*
1879		* @param blocker the blocker
1880	<	* @param maintainParallelism if true and supported by this pool,
1881	<	* attempt to maintain the pool's nominal parallelism; otherwise
1882	<	* activate a thread only if necessary to avoid complete
1883	<	* starvation.
1884	<	* @throws InterruptedException if blocker.block did so.
1880	>	* @param maintainParallelism if {@code true} and supported by
1881	>	* this pool, attempt to maintain the pool's nominal parallelism;
1882	>	* otherwise activate a thread only if necessary to avoid
1883	>	* complete starvation.
1884	>	* @throws InterruptedException if blocker.block did so
1885		*/
1886		public static void managedBlock(ManagedBlocker blocker,
1887		boolean maintainParallelism)
1888		throws InterruptedException {
1889		Thread t = Thread.currentThread();
1890	<	ForkJoinPool pool = (t instanceof ForkJoinWorkerThread?
1891	<	((ForkJoinWorkerThread)t).pool : null);
1890	>	ForkJoinPool pool = ((t instanceof ForkJoinWorkerThread) ?
1891	>	((ForkJoinWorkerThread) t).pool : null);
1892		if (!blocker.isReleasable()) {
1893		try {
1894		if (pool == null ||
#	Line 1677 | Line 1903 | public class ForkJoinPool extends Abstra
1903
1904		private static void awaitBlocker(ManagedBlocker blocker)
1905		throws InterruptedException {
1906	<	do;while (!blocker.isReleasable() && !blocker.block());
1906	>	do {} while (!blocker.isReleasable() && !blocker.block());
1907		}
1908
1909	<	// AbstractExecutorService overrides
1909	>	// AbstractExecutorService overrides.  These rely on undocumented
1910	>	// fact that ForkJoinTask.adapt returns ForkJoinTasks that also
1911	>	// implement RunnableFuture.
1912
1913		protected <T> RunnableFuture<T> newTaskFor(Runnable runnable, T value) {
1914	<	return new AdaptedRunnable(runnable, value);
1914	>	return (RunnableFuture<T>) ForkJoinTask.adapt(runnable, value);
1915		}
1916
1917		protected <T> RunnableFuture<T> newTaskFor(Callable<T> callable) {
1918	<	return new AdaptedCallable(callable);
1918	>	return (RunnableFuture<T>) ForkJoinTask.adapt(callable);
1919		}
1920
1921	+	// Unsafe mechanics
1922
1923	<	// Temporary Unsafe mechanics for preliminary release
1924	<
1925	<	static final Unsafe _unsafe;
1926	<	static final long eventCountOffset;
1927	<	static final long workerCountsOffset;
1928	<	static final long runControlOffset;
1929	<	static final long barrierStackOffset;
1930	<	static final long spareStackOffset;
1931	<
1932	<	static {
1933	<	try {
1705	<	if (ForkJoinPool.class.getClassLoader() != null) {
1706	<	Field f = Unsafe.class.getDeclaredField("theUnsafe");
1707	<	f.setAccessible(true);
1708	<	_unsafe = (Unsafe)f.get(null);
1709	<	}
1710	<	else
1711	<	_unsafe = Unsafe.getUnsafe();
1712	<	eventCountOffset = _unsafe.objectFieldOffset
1713	<	(ForkJoinPool.class.getDeclaredField("eventCount"));
1714	<	workerCountsOffset = _unsafe.objectFieldOffset
1715	<	(ForkJoinPool.class.getDeclaredField("workerCounts"));
1716	<	runControlOffset = _unsafe.objectFieldOffset
1717	<	(ForkJoinPool.class.getDeclaredField("runControl"));
1718	<	barrierStackOffset = _unsafe.objectFieldOffset
1719	<	(ForkJoinPool.class.getDeclaredField("barrierStack"));
1720	<	spareStackOffset = _unsafe.objectFieldOffset
1721	<	(ForkJoinPool.class.getDeclaredField("spareStack"));
1722	<	} catch (Exception e) {
1723	<	throw new RuntimeException("Could not initialize intrinsics", e);
1724	<	}
1725	<	}
1923	>	private static final sun.misc.Unsafe UNSAFE = getUnsafe();
1924	>	private static final long eventCountOffset =
1925	>	objectFieldOffset("eventCount", ForkJoinPool.class);
1926	>	private static final long workerCountsOffset =
1927	>	objectFieldOffset("workerCounts", ForkJoinPool.class);
1928	>	private static final long runControlOffset =
1929	>	objectFieldOffset("runControl", ForkJoinPool.class);
1930	>	private static final long syncStackOffset =
1931	>	objectFieldOffset("syncStack",ForkJoinPool.class);
1932	>	private static final long spareStackOffset =
1933	>	objectFieldOffset("spareStack", ForkJoinPool.class);
1934
1935		private boolean casEventCount(long cmp, long val) {
1936	<	return _unsafe.compareAndSwapLong(this, eventCountOffset, cmp, val);
1936	>	return UNSAFE.compareAndSwapLong(this, eventCountOffset, cmp, val);
1937		}
1938		private boolean casWorkerCounts(int cmp, int val) {
1939	<	return _unsafe.compareAndSwapInt(this, workerCountsOffset, cmp, val);
1939	>	return UNSAFE.compareAndSwapInt(this, workerCountsOffset, cmp, val);
1940		}
1941		private boolean casRunControl(int cmp, int val) {
1942	<	return _unsafe.compareAndSwapInt(this, runControlOffset, cmp, val);
1942	>	return UNSAFE.compareAndSwapInt(this, runControlOffset, cmp, val);
1943		}
1944		private boolean casSpareStack(WaitQueueNode cmp, WaitQueueNode val) {
1945	<	return _unsafe.compareAndSwapObject(this, spareStackOffset, cmp, val);
1945	>	return UNSAFE.compareAndSwapObject(this, spareStackOffset, cmp, val);
1946		}
1947		private boolean casBarrierStack(WaitQueueNode cmp, WaitQueueNode val) {
1948	<	return _unsafe.compareAndSwapObject(this, barrierStackOffset, cmp, val);
1948	>	return UNSAFE.compareAndSwapObject(this, syncStackOffset, cmp, val);
1949	>	}
1950	>
1951	>	private static long objectFieldOffset(String field, Class<?> klazz) {
1952	>	try {
1953	>	return UNSAFE.objectFieldOffset(klazz.getDeclaredField(field));
1954	>	} catch (NoSuchFieldException e) {
1955	>	// Convert Exception to corresponding Error
1956	>	NoSuchFieldError error = new NoSuchFieldError(field);
1957	>	error.initCause(e);
1958	>	throw error;
1959	>	}
1960	>	}
1961	>
1962	>	/**
1963	>	* Returns a sun.misc.Unsafe.  Suitable for use in a 3rd party package.
1964	>	* Replace with a simple call to Unsafe.getUnsafe when integrating
1965	>	* into a jdk.
1966	>	*
1967	>	* @return a sun.misc.Unsafe
1968	>	*/
1969	>	private static sun.misc.Unsafe getUnsafe() {
1970	>	try {
1971	>	return sun.misc.Unsafe.getUnsafe();
1972	>	} catch (SecurityException se) {
1973	>	try {
1974	>	return java.security.AccessController.doPrivileged
1975	>	(new java.security
1976	>	.PrivilegedExceptionAction<sun.misc.Unsafe>() {
1977	>	public sun.misc.Unsafe run() throws Exception {
1978	>	java.lang.reflect.Field f = sun.misc
1979	>	.Unsafe.class.getDeclaredField("theUnsafe");
1980	>	f.setAccessible(true);
1981	>	return (sun.misc.Unsafe) f.get(null);
1982	>	}});
1983	>	} catch (java.security.PrivilegedActionException e) {
1984	>	throw new RuntimeException("Could not initialize intrinsics",
1985	>	e.getCause());
1986	>	}
1987	>	}
1988		}
1989		}

Diff Legend

–	Removed lines
+	Added lines
<	Changed lines
>	Changed lines