[ViewVC] Diff of: jsr166/jsr166/src/jsr166y/ForkJoinPool.java

Comparing jsr166/src/jsr166y/ForkJoinPool.java (file contents):
Revision 1.122 by jsr166, Tue Jan 31 01:51:13 2012 UTC vs.
Revision 1.136 by dl, Mon Oct 29 17:23:34 2012 UTC

#	Line 42 \| Line 42 \| import java.util.concurrent.locks.Condit
42		* ForkJoinPool}s may also be appropriate for use with event-style
43		* tasks that are never joined.
44		*
45	<	* <p>A {@code ForkJoinPool} is constructed with a given target
46	<	* parallelism level; by default, equal to the number of available
47	<	* processors. The pool attempts to maintain enough active (or
48	<	* available) threads by dynamically adding, suspending, or resuming
49	<	* internal worker threads, even if some tasks are stalled waiting to
50	<	* join others. However, no such adjustments are guaranteed in the
51	<	* face of blocked IO or other unmanaged synchronization. The nested
52	<	* {@link ManagedBlocker} interface enables extension of the kinds of
45	>	* <p>A static {@link #commonPool} is available and appropriate for
46	>	* most applications. The common pool is used by any ForkJoinTask that
47	>	* is not explicitly submitted to a specified pool. Using the common
48	>	* pool normally reduces resource usage (its threads are slowly
49	>	* reclaimed during periods of non-use, and reinstated upon subsequent
50	>	* use). The common pool is by default constructed with default
51	>	* parameters, but these may be controlled by setting any or all of
52	>	* the three properties {@code
53	>	* java.util.concurrent.ForkJoinPool.common.{parallelism,
54	>	* threadFactory, exceptionHandler}}.
55	>	*
56	>	* <p>For applications that require separate or custom pools, a {@code
57	>	* ForkJoinPool} may be constructed with a given target parallelism
58	>	* level; by default, equal to the number of available processors. The
59	>	* pool attempts to maintain enough active (or available) threads by
60	>	* dynamically adding, suspending, or resuming internal worker
61	>	* threads, even if some tasks are stalled waiting to join
62	>	* others. However, no such adjustments are guaranteed in the face of
63	>	* blocked IO or other unmanaged synchronization. The nested {@link
64	>	* ManagedBlocker} interface enables extension of the kinds of
65		* synchronization accommodated.
66		*
67		* <p>In addition to execution and lifecycle control methods, this
#	Line 94 \| Line 106 \| import java.util.concurrent.locks.Condit
106		* </tr>
107		* </table>
108		*
97	–	* <p><b>Sample Usage.</b> Normally a single {@code ForkJoinPool} is
98	–	* used for all parallel task execution in a program or subsystem.
99	–	* Otherwise, use would not usually outweigh the construction and
100	–	* bookkeeping overhead of creating a large set of threads. For
101	–	* example, a common pool could be used for the {@code SortTasks}
102	–	* illustrated in {@link RecursiveAction}. Because {@code
103	–	* ForkJoinPool} uses threads in {@linkplain java.lang.Thread#isDaemon
104	–	* daemon} mode, there is typically no need to explicitly {@link
105	–	* #shutdown} such a pool upon program exit.
106	–	*
107	–	* <pre> {@code
108	–	* static final ForkJoinPool mainPool = new ForkJoinPool();
109	–	* ...
110	–	* public void sort(long[] array) {
111	–	* mainPool.invoke(new SortTask(array, 0, array.length));
112	–	* }}</pre>
113	–	*
109		* <p><b>Implementation notes</b>: This implementation restricts the
110		* maximum number of running threads to 32767. Attempts to create
111		* pools with greater than the maximum number result in
#	Line 177 \| Line 172 \| public class ForkJoinPool extends Abstra
172		* If an attempted steal fails, a thief always chooses a different
173		* random victim target to try next. So, in order for one thief to
174		* progress, it suffices for any in-progress poll or new push on
175	<	* any empty queue to complete.
175	>	* any empty queue to complete. (This is why we normally use
176	>	* method pollAt and its variants that try once at the apparent
177	>	* base index, else consider alternative actions, rather than
178	>	* method poll.)
179		*
180		* This approach also enables support of a user mode in which local
181		* task processing is in FIFO, not LIFO order, simply by using
#	Line 207 \| Line 205 \| public class ForkJoinPool extends Abstra
205		* lock (mainly to protect in the case of resizing) but we use
206		* only a simple spinlock (using bits in field runState), because
207		* submitters encountering a busy queue move on to try or create
208	<	* other queues, so never block.
208	>	* other queues -- they block only when creating and registering
209	>	* new queues.
210		*
211		* Management
212		* ==========
#	Line 233 \| Line 232 \| public class ForkJoinPool extends Abstra
232		* deregister WorkQueues, as well as to enable shutdown. It is
233		* only modified under a lock (normally briefly held, but
234		* occasionally protecting allocations and resizings) but even
235	<	* when locked remains available to check consistency. An
237	<	* auxiliary field "growHints", also only modified under lock,
238	<	* contains a candidate index for the next WorkQueue and
239	<	* a mask for submission queue indices.
235	>	* when locked remains available to check consistency.
236		*
237		* Recording WorkQueues. WorkQueues are recorded in the
238	<	* "workQueues" array that is created upon pool construction and
239	<	* expanded if necessary. Updates to the array while recording
240	<	* new workers and unrecording terminated ones are protected from
241	<	* each other by a lock but the array is otherwise concurrently
242	<	* readable, and accessed directly. To simplify index-based
243	<	* operations, the array size is always a power of two, and all
244	<	* readers must tolerate null slots. Shared (submission) queues
245	<	* are at even indices, worker queues at odd indices. Grouping
246	<	* them together in this way simplifies and speeds up task
251	<	* scanning. To avoid flailing during start-up, the array is
252	<	* presized to hold twice #parallelism workers (which is unlikely
253	<	* to need further resizing during execution). But to avoid
254	<	* dealing with so many null slots, variable runState includes a
255	<	* mask for the nearest power of two that contains all currently
256	<	* used indices.
238	>	* "workQueues" array that is created upon first use and expanded
239	>	* if necessary. Updates to the array while recording new workers
240	>	* and unrecording terminated ones are protected from each other
241	>	* by a lock but the array is otherwise concurrently readable, and
242	>	* accessed directly. To simplify index-based operations, the
243	>	* array size is always a power of two, and all readers must
244	>	* tolerate null slots. Shared (submission) queues are at even
245	>	* indices, worker queues at odd indices. Grouping them together
246	>	* in this way simplifies and speeds up task scanning.
247		*
248		* All worker thread creation is on-demand, triggered by task
249		* submissions, replacement of terminated workers, and/or
#	Line 325 \| Line 315 \| public class ForkJoinPool extends Abstra
315		*
316		* Trimming workers. To release resources after periods of lack of
317		* use, a worker starting to wait when the pool is quiescent will
318	<	* time out and terminate if the pool has remained quiescent for
319	<	* SHRINK_RATE nanosecs. This will slowly propagate, eventually
320	<	* terminating all workers after long periods of non-use.
318	>	* time out and terminate if the pool has remained quiescent for a
319	>	* given period -- a short period if there are more threads than
320	>	* parallelism, longer as the number of threads decreases. This
321	>	* will slowly propagate, eventually terminating all workers after
322	>	* periods of non-use.
323		*
324		* Shutdown and Termination. A call to shutdownNow atomically sets
325		* a runState bit and then (non-atomically) sets each worker's
#	Line 385 \| Line 377 \| public class ForkJoinPool extends Abstra
377		* (http://portal.acm.org/citation.cfm?id=155354). It differs in
378		* that: (1) We only maintain dependency links across workers upon
379		* steals, rather than use per-task bookkeeping. This sometimes
380	<	* requires a linear scan of workQueues array to locate stealers, but
381	<	* often doesn't because stealers leave hints (that may become
380	>	* requires a linear scan of workQueues array to locate stealers,
381	>	* but often doesn't because stealers leave hints (that may become
382		* stale/wrong) of where to locate them. A stealHint is only a
383		* hint because a worker might have had multiple steals and the
384		* hint records only one of them (usually the most current).
#	Line 397 \| Line 389 \| public class ForkJoinPool extends Abstra
389		* which means that we miss links in the chain during long-lived
390		* tasks, GC stalls etc (which is OK since blocking in such cases
391		* is usually a good idea). (4) We bound the number of attempts
392	<	* to find work (see MAX_HELP_DEPTH) and fall back to suspending
393	<	* the worker and if necessary replacing it with another.
392	>	* to find work (see MAX_HELP) and fall back to suspending the
393	>	* worker and if necessary replacing it with another.
394		*
395		* It is impossible to keep exactly the target parallelism number
396		* of threads running at any given time. Determining the
397		* existence of conservatively safe helping targets, the
398		* availability of already-created spares, and the apparent need
399		* to create new spares are all racy, so we rely on multiple
400	<	* retries of each. Currently, in keeping with on-demand
401	<	* signalling policy, we compensate only if blocking would leave
402	<	* less than one active (non-waiting, non-blocked) worker.
403	<	* Additionally, to avoid some false alarms due to GC, lagging
404	<	* counters, system activity, etc, compensated blocking for joins
405	<	* is only attempted after rechecks stabilize in
406	<	* ForkJoinTask.awaitJoin. (Retries are interspersed with
407	<	* Thread.yield, for good citizenship.)
400	>	* retries of each. Compensation in the apparent absence of
401	>	* helping opportunities is challenging to control on JVMs, where
402	>	* GC and other activities can stall progress of tasks that in
403	>	* turn stall out many other dependent tasks, without us being
404	>	* able to determine whether they will ever require compensation.
405	>	* Even though work-stealing otherwise encounters little
406	>	* degradation in the presence of more threads than cores,
407	>	* aggressively adding new threads in such cases entails risk of
408	>	* unwanted positive feedback control loops in which more threads
409	>	* cause more dependent stalls (as well as delayed progress of
410	>	* unblocked threads to the point that we know they are available)
411	>	* leading to more situations requiring more threads, and so
412	>	* on. This aspect of control can be seen as an (analytically
413	>	* intractable) game with an opponent that may choose the worst
414	>	* (for us) active thread to stall at any time. We take several
415	>	* precautions to bound losses (and thus bound gains), mainly in
416	>	* methods tryCompensate and awaitJoin: (1) We only try
417	>	* compensation after attempting enough helping steps (measured
418	>	* via counting and timing) that we have already consumed the
419	>	* estimated cost of creating and activating a new thread. (2) We
420	>	* allow up to 50% of threads to be blocked before initially
421	>	* adding any others, and unless completely saturated, check that
422	>	* some work is available for a new worker before adding. Also, we
423	>	* create up to only 50% more threads until entering a mode that
424	>	* only adds a thread if all others are possibly blocked. All
425	>	* together, this means that we might be half as fast to react,
426	>	* and create half as many threads as possible in the ideal case,
427	>	* but present vastly fewer anomalies in all other cases compared
428	>	* to both more aggressive and more conservative alternatives.
429		*
430		* Style notes: There is a lot of representation-level coupling
431		* among classes ForkJoinPool, ForkJoinWorkerThread, and
#	Line 449 \| Line 462 \| public class ForkJoinPool extends Abstra
462		// Static utilities
463
464		/**
452	–	* Computes an initial hash code (also serving as a non-zero
453	–	* random seed) for a thread id. This method is expected to
454	–	* provide higher-quality hash codes than using method hashCode().
455	–	*/
456	–	static final int hashId(long id) {
457	–	int h = (int)id ^ (int)(id >>> 32); // Use MurmurHash of thread id
458	–	h ^= h >>> 16; h *= 0x85ebca6b;
459	–	h ^= h >>> 13; h *= 0xc2b2ae35;
460	–	h ^= h >>> 16;
461	–	return (h == 0) ? 1 : h; // ensure nonzero
462	–	}
463	–
464	–	/**
465		* If there is a security manager, makes sure caller has
466		* permission to modify threads.
467		*/
#	Line 501 \| Line 501 \| public class ForkJoinPool extends Abstra
501		}
502
503		/**
504	–	* A simple non-reentrant lock used for exclusion when managing
505	–	* queues and workers. We use a custom lock so that we can readily
506	–	* probe lock state in constructions that check among alternative
507	–	* actions. The lock is normally only very briefly held, and
508	–	* sometimes treated as a spinlock, but other usages block to
509	–	* reduce overall contention in those cases where locked code
510	–	* bodies perform allocation/resizing.
511	–	*/
512	–	static final class Mutex extends AbstractQueuedSynchronizer {
513	–	public final boolean tryAcquire(int ignore) {
514	–	return compareAndSetState(0, 1);
515	–	}
516	–	public final boolean tryRelease(int ignore) {
517	–	setState(0);
518	–	return true;
519	–	}
520	–	public final void lock() { acquire(0); }
521	–	public final void unlock() { release(0); }
522	–	public final boolean isHeldExclusively() { return getState() == 1; }
523	–	public final Condition newCondition() { return new ConditionObject(); }
524	–	}
525	–
526	–	/**
504		* Class for artificial tasks that are used to replace the target
505		* of local joins if they are removed from an interior queue slot
506		* in WorkQueue.tryRemoveAndExec. We don't need the proxy to
#	Line 592 \| Line 569 \| public class ForkJoinPool extends Abstra
569		static final class WorkQueue {
570		/**
571		* Capacity of work-stealing queue array upon initialization.
572	<	* Must be a power of two; at least 4, but set larger to
573	<	* reduce cacheline sharing among queues.
572	>	* Must be a power of two; at least 4, but should be larger to
573	>	* reduce or eliminate cacheline sharing among queues.
574	>	* Currently, it is much larger, as a partial workaround for
575	>	* the fact that JVMs often place arrays in locations that
576	>	* share GC bookkeeping (especially cardmarks) such that
577	>	* per-write accesses encounter serious memory contention.
578		*/
579	<	static final int INITIAL_QUEUE_CAPACITY = 1 << 8;
579	>	static final int INITIAL_QUEUE_CAPACITY = 1 << 13;
580
581		/**
582		* Maximum size for queue arrays. Must be a power of two less
#	Line 619 \| Line 600 \| public class ForkJoinPool extends Abstra
600		volatile int base; // index of next slot for poll
601		int top; // index of next slot for push
602		ForkJoinTask<?>[] array; // the elements (initially unallocated)
603	+	final ForkJoinPool pool; // the containing pool (may be null)
604		final ForkJoinWorkerThread owner; // owning thread or null if shared
605		volatile Thread parker; // == owner during call to park; else null
606	<	ForkJoinTask<?> currentJoin; // task being joined in awaitJoin
606	>	volatile ForkJoinTask<?> currentJoin; // task being joined in awaitJoin
607		ForkJoinTask<?> currentSteal; // current non-local task being executed
608		// Heuristic padding to ameliorate unfortunate memory placements
609	<	Object p00, p01, p02, p03, p04, p05, p06, p07, p08, p09, p0a;
609	>	Object p00, p01, p02, p03, p04, p05, p06, p07;
610	>	Object p08, p09, p0a, p0b, p0c, p0d, p0e;
611
612	<	WorkQueue(ForkJoinWorkerThread owner, int mode) {
630	<	this.owner = owner;
612	>	WorkQueue(ForkJoinPool pool, ForkJoinWorkerThread owner, int mode) {
613		this.mode = mode;
614	+	this.pool = pool;
615	+	this.owner = owner;
616		// Place indices in the center of array (that is not yet allocated)
617		base = top = INITIAL_QUEUE_CAPACITY >>> 1;
618		}
619
620		/**
621	<	* Returns number of tasks in the queue.
621	>	* Returns the approximate number of tasks in the queue.
622		*/
623		final int queueSize() {
624	<	int n = base - top; // non-owner callers must read base first
625	<	return (n >= 0) ? 0 : -n;
624	>	int n = base - top; // non-owner callers must read base first
625	>	return (n >= 0) ? 0 : -n; // ignore transient negative
626	>	}
627	>
628	>	/**
629	>	* Provides a more accurate estimate of whether this queue has
630	>	* any tasks than does queueSize, by checking whether a
631	>	* near-empty queue has at least one unclaimed task.
632	>	*/
633	>	final boolean isEmpty() {
634	>	ForkJoinTask<?>[] a; int m, s;
635	>	int n = base - (s = top);
636	>	return (n >= 0 \|\|
637	>	(n == -1 &&
638	>	((a = array) == null \|\|
639	>	(m = a.length - 1) < 0 \|\|
640	>	U.getObjectVolatile
641	>	(a, ((m & (s - 1)) << ASHIFT) + ABASE) == null)));
642		}
643
644		/**
645		* Pushes a task. Call only by owner in unshared queues.
646		*
647		* @param task the task. Caller must ensure non-null.
648	–	* @param p if non-null, pool to signal if necessary
648		* @throw RejectedExecutionException if array cannot be resized
649		*/
650	<	final void push(ForkJoinTask<?> task, ForkJoinPool p) {
651	<	ForkJoinTask<?>[] a;
650	>	final void push(ForkJoinTask<?> task) {
651	>	ForkJoinTask<?>[] a; ForkJoinPool p;
652		int s = top, m, n;
653		if ((a = array) != null) { // ignore if queue removed
654		U.putOrderedObject
655		(a, (((m = a.length - 1) & s) << ASHIFT) + ABASE, task);
656		if ((n = (top = s + 1) - base) <= 2) {
657	<	if (p != null)
657	>	if ((p = pool) != null)
658		p.signalWork();
659		}
660		else if (n >= m)
#	Line 691 \| Line 690 \| public class ForkJoinPool extends Abstra
690		}
691
692		/**
693	<	* Takes next task, if one exists, in FIFO order.
693	>	* Takes next task, if one exists, in LIFO order. Call only
694	>	* by owner in unshared queues. (We do not have a shared
695	>	* version of this method because it is never needed.)
696		*/
697	<	final ForkJoinTask<?> poll() {
698	<	ForkJoinTask<?>[] a; int b; ForkJoinTask<?> t;
699	<	while ((b = base) - top < 0 && (a = array) != null) {
697	>	final ForkJoinTask<?> pop() {
698	>	ForkJoinTask<?>[] a; ForkJoinTask<?> t; int m;
699	>	if ((a = array) != null && (m = a.length - 1) >= 0) {
700	>	for (int s; (s = top - 1) - base >= 0;) {
701	>	long j = ((m & s) << ASHIFT) + ABASE;
702	>	if ((t = (ForkJoinTask<?>)U.getObject(a, j)) == null)
703	>	break;
704	>	if (U.compareAndSwapObject(a, j, t, null)) {
705	>	top = s;
706	>	return t;
707	>	}
708	>	}
709	>	}
710	>	return null;
711	>	}
712	>
713	>	/**
714	>	* Takes a task in FIFO order if b is base of queue and a task
715	>	* can be claimed without contention. Specialized versions
716	>	* appear in ForkJoinPool methods scan and tryHelpStealer.
717	>	*/
718	>	final ForkJoinTask<?> pollAt(int b) {
719	>	ForkJoinTask<?> t; ForkJoinTask<?>[] a;
720	>	if ((a = array) != null) {
721		int j = (((a.length - 1) & b) << ASHIFT) + ABASE;
722		if ((t = (ForkJoinTask<?>)U.getObjectVolatile(a, j)) != null &&
723		base == b &&
#	Line 708 \| Line 730 \| public class ForkJoinPool extends Abstra
730		}
731
732		/**
733	<	* Takes next task, if one exists, in LIFO order. Call only
712	<	* by owner in unshared queues. (We do not have a shared
713	<	* version of this method because it is never needed.)
733	>	* Takes next task, if one exists, in FIFO order.
734		*/
735	<	final ForkJoinTask<?> pop() {
736	<	ForkJoinTask<?> t; int m;
737	<	ForkJoinTask<?>[] a = array;
738	<	if (a != null && (m = a.length - 1) >= 0) {
739	<	for (int s; (s = top - 1) - base >= 0;) {
740	<	int j = ((m & s) << ASHIFT) + ABASE;
741	<	if ((t = (ForkJoinTask<?>)U.getObjectVolatile(a, j)) == null)
742	<	break;
743	<	if (U.compareAndSwapObject(a, j, t, null)) {
724	<	top = s;
735	>	final ForkJoinTask<?> poll() {
736	>	ForkJoinTask<?>[] a; int b; ForkJoinTask<?> t;
737	>	while ((b = base) - top < 0 && (a = array) != null) {
738	>	int j = (((a.length - 1) & b) << ASHIFT) + ABASE;
739	>	t = (ForkJoinTask<?>)U.getObjectVolatile(a, j);
740	>	if (t != null) {
741	>	if (base == b &&
742	>	U.compareAndSwapObject(a, j, t, null)) {
743	>	base = b + 1;
744		return t;
745		}
746		}
747	+	else if (base == b) {
748	+	if (b + 1 == top)
749	+	break;
750	+	Thread.yield(); // wait for lagging update
751	+	}
752		}
753		return null;
754		}
#	Line 749 \| Line 773 \| public class ForkJoinPool extends Abstra
773		}
774
775		/**
752	–	* Returns task at index b if b is current base of queue.
753	–	*/
754	–	final ForkJoinTask<?> pollAt(int b) {
755	–	ForkJoinTask<?> t; ForkJoinTask<?>[] a;
756	–	if ((a = array) != null) {
757	–	int j = (((a.length - 1) & b) << ASHIFT) + ABASE;
758	–	if ((t = (ForkJoinTask<?>)U.getObjectVolatile(a, j)) != null &&
759	–	base == b &&
760	–	U.compareAndSwapObject(a, j, t, null)) {
761	–	base = b + 1;
762	–	return t;
763	–	}
764	–	}
765	–	return null;
766	–	}
767	–
768	–	/**
776		* Pops the given task only if it is at the current top.
777		*/
778		final boolean tryUnpush(ForkJoinTask<?> t) {
#	Line 780 \| Line 787 \| public class ForkJoinPool extends Abstra
787		}
788
789		/**
790	+	* Version of tryUnpush for shared queues; called by non-FJ
791	+	* submitters after prechecking that task probably exists.
792	+	*/
793	+	final boolean trySharedUnpush(ForkJoinTask<?> t) {
794	+	boolean success = false;
795	+	if (runState == 0 && U.compareAndSwapInt(this, RUNSTATE, 0, 1)) {
796	+	try {
797	+	ForkJoinTask<?>[] a; int s;
798	+	if ((a = array) != null && (s = top) != base &&
799	+	U.compareAndSwapObject
800	+	(a, (((a.length - 1) & --s) << ASHIFT) + ABASE, t, null)) {
801	+	top = s;
802	+	success = true;
803	+	}
804	+	} finally {
805	+	runState = 0; // unlock
806	+	}
807	+	}
808	+	return success;
809	+	}
810	+
811	+	/**
812		* Polls the given task only if it is at the current base.
813		*/
814		final boolean pollFor(ForkJoinTask<?> task) {
#	Line 796 \| Line 825 \| public class ForkJoinPool extends Abstra
825		}
826
827		/**
799	–	* If present, removes from queue and executes the given task, or
800	–	* any other cancelled task. Returns (true) immediately on any CAS
801	–	* or consistency check failure so caller can retry.
802	–	*
803	–	* @return false if no progress can be made
804	–	*/
805	–	final boolean tryRemoveAndExec(ForkJoinTask<?> task) {
806	–	boolean removed = false, empty = true, progress = true;
807	–	ForkJoinTask<?>[] a; int m, s, b, n;
808	–	if ((a = array) != null && (m = a.length - 1) >= 0 &&
809	–	(n = (s = top) - (b = base)) > 0) {
810	–	for (ForkJoinTask<?> t;;) { // traverse from s to b
811	–	int j = ((--s & m) << ASHIFT) + ABASE;
812	–	t = (ForkJoinTask<?>)U.getObjectVolatile(a, j);
813	–	if (t == null) // inconsistent length
814	–	break;
815	–	else if (t == task) {
816	–	if (s + 1 == top) { // pop
817	–	if (!U.compareAndSwapObject(a, j, task, null))
818	–	break;
819	–	top = s;
820	–	removed = true;
821	–	}
822	–	else if (base == b) // replace with proxy
823	–	removed = U.compareAndSwapObject(a, j, task,
824	–	new EmptyTask());
825	–	break;
826	–	}
827	–	else if (t.status >= 0)
828	–	empty = false;
829	–	else if (s + 1 == top) { // pop and throw away
830	–	if (U.compareAndSwapObject(a, j, t, null))
831	–	top = s;
832	–	break;
833	–	}
834	–	if (--n == 0) {
835	–	if (!empty && base == b)
836	–	progress = false;
837	–	break;
838	–	}
839	–	}
840	–	}
841	–	if (removed)
842	–	task.doExec();
843	–	return progress;
844	–	}
845	–
846	–	/**
828		* Initializes or doubles the capacity of array. Call either
829		* by owner or with lock held -- it is OK for base, but not
830		* top, to move while resizings are in progress.
#	Line 892 \| Line 873 \| public class ForkJoinPool extends Abstra
873		* Computes next value for random probes. Scans don't require
874		* a very high quality generator, but also not a crummy one.
875		* Marsaglia xor-shift is cheap and works well enough. Note:
876	<	* This is manually inlined in several usages in ForkJoinPool
877	<	* to avoid writes inside busy scan loops.
876	>	* This is manually inlined in its usages in ForkJoinPool to
877	>	* avoid writes inside busy scan loops.
878		*/
879		final int nextSeed() {
880		int r = seed;
#	Line 905 \| Line 886 \| public class ForkJoinPool extends Abstra
886		// Execution methods
887
888		/**
889	<	* Removes and runs tasks until empty, using local mode
909	<	* ordering.
889	>	* Pops and runs tasks until empty.
890		*/
891	<	final void runLocalTasks() {
892	<	if (base - top < 0) {
893	<	for (ForkJoinTask<?> t; (t = nextLocalTask()) != null; )
891	>	private void popAndExecAll() {
892	>	// A bit faster than repeated pop calls
893	>	ForkJoinTask<?>[] a; int m, s; long j; ForkJoinTask<?> t;
894	>	while ((a = array) != null && (m = a.length - 1) >= 0 &&
895	>	(s = top - 1) - base >= 0 &&
896	>	(t = ((ForkJoinTask<?>)
897	>	U.getObject(a, j = ((m & s) << ASHIFT) + ABASE)))
898	>	!= null) {
899	>	if (U.compareAndSwapObject(a, j, t, null)) {
900	>	top = s;
901		t.doExec();
902	+	}
903		}
904		}
905
906		/**
907	+	* Polls and runs tasks until empty.
908	+	*/
909	+	private void pollAndExecAll() {
910	+	for (ForkJoinTask<?> t; (t = poll()) != null;)
911	+	t.doExec();
912	+	}
913	+
914	+	/**
915	+	* If present, removes from queue and executes the given task, or
916	+	* any other cancelled task. Returns (true) immediately on any CAS
917	+	* or consistency check failure so caller can retry.
918	+	*
919	+	* @return 0 if no progress can be made, else positive
920	+	* (this unusual convention simplifies use with tryHelpStealer.)
921	+	*/
922	+	final int tryRemoveAndExec(ForkJoinTask<?> task) {
923	+	int stat = 1;
924	+	boolean removed = false, empty = true;
925	+	ForkJoinTask<?>[] a; int m, s, b, n;
926	+	if ((a = array) != null && (m = a.length - 1) >= 0 &&
927	+	(n = (s = top) - (b = base)) > 0) {
928	+	for (ForkJoinTask<?> t;;) { // traverse from s to b
929	+	int j = ((--s & m) << ASHIFT) + ABASE;
930	+	t = (ForkJoinTask<?>)U.getObjectVolatile(a, j);
931	+	if (t == null) // inconsistent length
932	+	break;
933	+	else if (t == task) {
934	+	if (s + 1 == top) { // pop
935	+	if (!U.compareAndSwapObject(a, j, task, null))
936	+	break;
937	+	top = s;
938	+	removed = true;
939	+	}
940	+	else if (base == b) // replace with proxy
941	+	removed = U.compareAndSwapObject(a, j, task,
942	+	new EmptyTask());
943	+	break;
944	+	}
945	+	else if (t.status >= 0)
946	+	empty = false;
947	+	else if (s + 1 == top) { // pop and throw away
948	+	if (U.compareAndSwapObject(a, j, t, null))
949	+	top = s;
950	+	break;
951	+	}
952	+	if (--n == 0) {
953	+	if (!empty && base == b)
954	+	stat = 0;
955	+	break;
956	+	}
957	+	}
958	+	}
959	+	if (removed)
960	+	task.doExec();
961	+	return stat;
962	+	}
963	+
964	+	/**
965		* Executes a top-level task and any local tasks remaining
966		* after execution.
921	–	*
922	–	* @return true unless terminating
967		*/
968	<	final boolean runTask(ForkJoinTask<?> t) {
925	<	boolean alive = true;
968	>	final void runTask(ForkJoinTask<?> t) {
969		if (t != null) {
970		currentSteal = t;
971		t.doExec();
972	<	runLocalTasks();
972	>	if (top != base) { // process remaining local tasks
973	>	if (mode == 0)
974	>	popAndExecAll();
975	>	else
976	>	pollAndExecAll();
977	>	}
978		++nsteals;
979		currentSteal = null;
980		}
933	–	else if (runState < 0) // terminating
934	–	alive = false;
935	–	return alive;
981		}
982
983		/**
#	Line 1005 \| Line 1050 \| public class ForkJoinPool extends Abstra
1050		* submission queues in method doSubmit. In the future, this may
1051		* also incorporate a means to implement different task rejection
1052		* and resubmission policies.
1053	+	*
1054	+	* Seeds for submitters and workers/workQueues work in basically
1055	+	* the same way but are initialized and updated using slightly
1056	+	* different mechanics. Both are initialized using the same
1057	+	* approach as in class ThreadLocal, where successive values are
1058	+	* unlikely to collide with previous values. This is done during
1059	+	* registration for workers, but requires a separate AtomicInteger
1060	+	* for submitters. Seeds are then randomly modified upon
1061	+	* collisions using xorshifts, which requires a non-zero seed.
1062		*/
1063		static final class Submitter {
1064		int seed;
1065	<	Submitter() { seed = hashId(Thread.currentThread().getId()); }
1065	>	Submitter() {
1066	>	int s = nextSubmitterSeed.getAndAdd(SEED_INCREMENT);
1067	>	seed = (s == 0) ? 1 : s; // ensure non-zero
1068	>	}
1069		}
1070
1071		/** ThreadLocal class for Submitters */
#	Line 1025 \| Line 1082 \| public class ForkJoinPool extends Abstra
1082		public static final ForkJoinWorkerThreadFactory
1083		defaultForkJoinWorkerThreadFactory;
1084
1085	+
1086	+	/** Property prefix for constructing common pool */
1087	+	private static final String propPrefix =
1088	+	"java.util.concurrent.ForkJoinPool.common.";
1089	+
1090	+	/**
1091	+	* Common (static) pool. Non-null for public use unless a static
1092	+	* construction exception, but internal usages must null-check on
1093	+	* use.
1094	+	*/
1095	+	static final ForkJoinPool commonPool;
1096	+
1097	+	/**
1098	+	* Common pool parallelism. Must equal commonPool.parallelism.
1099	+	*/
1100	+	static final int commonPoolParallelism;
1101	+
1102		/**
1103		* Generator for assigning sequence numbers as pool names.
1104		*/
1105		private static final AtomicInteger poolNumberGenerator;
1106
1107		/**
1108	+	* Generator for initial hashes/seeds for submitters. Accessed by
1109	+	* Submitter class constructor.
1110	+	*/
1111	+	static final AtomicInteger nextSubmitterSeed;
1112	+
1113	+	/**
1114		* Permission required for callers of methods that may start or
1115		* kill threads.
1116		*/
1117		private static final RuntimePermission modifyThreadPermission;
1118
1119		/**
1120	<	* Per-thread submission bookeeping. Shared across all pools
1120	>	* Per-thread submission bookkeeping. Shared across all pools
1121		* to reduce ThreadLocal pollution and because random motion
1122		* to avoid contention in one pool is likely to hold for others.
1123		*/
#	Line 1046 \| Line 1126 \| public class ForkJoinPool extends Abstra
1126		// static constants
1127
1128		/**
1129	<	* The wakeup interval (in nanoseconds) for a worker waiting for a
1130	<	* task when the pool is quiescent to instead try to shrink the
1131	<	* number of workers. The exact value does not matter too
1132	<	* much. It must be short enough to release resources during
1133	<	* sustained periods of idleness, but not so short that threads
1134	<	* are continually re-created.
1129	>	* Initial timeout value (in nanoseconds) for the thread triggering
1130	>	* quiescence to park waiting for new work. On timeout, the thread
1131	>	* will instead try to shrink the number of workers.
1132	>	*/
1133	>	private static final long IDLE_TIMEOUT = 1000L * 1000L * 1000L; // 1sec
1134	>
1135	>	/**
1136	>	* Timeout value when there are more threads than parallelism level
1137	>	*/
1138	>	private static final long FAST_IDLE_TIMEOUT = 100L * 1000L * 1000L;
1139	>
1140	>	/**
1141	>	* The maximum stolen->joining link depth allowed in method
1142	>	* tryHelpStealer. Must be a power of two. This value also
1143	>	* controls the maximum number of times to try to help join a task
1144	>	* without any apparent progress or change in pool state before
1145	>	* giving up and blocking (see awaitJoin). Depths for legitimate
1146	>	* chains are unbounded, but we use a fixed constant to avoid
1147	>	* (otherwise unchecked) cycles and to bound staleness of
1148	>	* traversal parameters at the expense of sometimes blocking when
1149	>	* we could be helping.
1150		*/
1151	<	private static final long SHRINK_RATE =
1057	<	4L * 1000L * 1000L * 1000L; // 4 seconds
1151	>	private static final int MAX_HELP = 64;
1152
1153		/**
1154	<	* The timeout value for attempted shrinkage, includes
1155	<	* some slop to cope with system timer imprecision.
1154	>	* Secondary time-based bound (in nanosecs) for helping attempts
1155	>	* before trying compensated blocking in awaitJoin. Used in
1156	>	* conjunction with MAX_HELP to reduce variance due to different
1157	>	* polling rates associated with different helping options. The
1158	>	* value should roughly approximate the time required to create
1159	>	* and/or activate a worker thread.
1160		*/
1161	<	private static final long SHRINK_TIMEOUT = SHRINK_RATE - (SHRINK_RATE / 10);
1161	>	private static final long COMPENSATION_DELAY = 1L << 18; // ~0.25 millisec
1162
1163		/**
1164	<	* The maximum stolen->joining link depth allowed in tryHelpStealer.
1165	<	* Depths for legitimate chains are unbounded, but we use a fixed
1068	<	* constant to avoid (otherwise unchecked) cycles and to bound
1069	<	* staleness of traversal parameters at the expense of sometimes
1070	<	* blocking when we could be helping.
1164	>	* Increment for seed generators. See class ThreadLocal for
1165	>	* explanation.
1166		*/
1167	<	private static final int MAX_HELP_DEPTH = 16;
1167	>	private static final int SEED_INCREMENT = 0x61c88647;
1168
1169		/**
1170		* Bits and masks for control variables
#	Line 1101 \| Line 1196 \| public class ForkJoinPool extends Abstra
1196		*
1197		* Field runState is an int packed with:
1198		* SHUTDOWN: true if shutdown is enabled (1 bit)
1199	<	* SEQ: a sequence number updated upon (de)registering workers (15 bits)
1200	<	* MASK: mask (power of 2 - 1) covering all registered poolIndexes (16 bits)
1199	>	* SEQ: a sequence number updated upon (de)registering workers (30 bits)
1200	>	* INIT: set true after workQueues array construction (1 bit)
1201		*
1202	<	* The combination of mask and sequence number enables simple
1203	<	* consistency checks: Staleness of read-only operations on the
1204	<	* workQueues array can be checked by comparing runState before vs
1110	<	* after the reads. The low 16 bits (i.e, anding with SMASK) hold
1111	<	* the smallest power of two covering all indices, minus
1112	<	* one.
1202	>	* The sequence number enables simple consistency checks:
1203	>	* Staleness of read-only operations on the workQueues array can
1204	>	* be checked by comparing runState before vs after the reads.
1205		*/
1206
1207		// bit positions/shifts for fields
#	Line 1119 \| Line 1211 \| public class ForkJoinPool extends Abstra
1211		private static final int EC_SHIFT = 16;
1212
1213		// bounds
1122	–	private static final int POOL_MAX = 0x7fff; // max #workers - 1
1214		private static final int SMASK = 0xffff; // short bits
1215	+	private static final int MAX_CAP = 0x7fff; // max #workers - 1
1216		private static final int SQMASK = 0xfffe; // even short bits
1217		private static final int SHORT_SIGN = 1 << 15;
1218		private static final int INT_SIGN = 1 << 31;
#	Line 1148 \| Line 1240 \| public class ForkJoinPool extends Abstra
1240
1241		// runState bits
1242		private static final int SHUTDOWN = 1 << 31;
1151	–	private static final int RS_SEQ = 1 << 16;
1152	–	private static final int RS_SEQ_MASK = 0x7fff0000;
1243
1244		// access mode for WorkQueue
1245		static final int LIFO_QUEUE = 0;
#	Line 1165 \| Line 1255 \| public class ForkJoinPool extends Abstra
1255		* empirically works OK on current JVMs.
1256		*/
1257
1258	+	volatile long stealCount; // collects worker counts
1259		volatile long ctl; // main pool control
1260		final int parallelism; // parallelism level
1261		final int localMode; // per-worker scheduling mode
1262	<	int growHints; // for expanding indices/ranges
1263	<	volatile int runState; // shutdown status, seq, and mask
1262	>	volatile int nextWorkerNumber; // to create worker name string
1263	>	final int submitMask; // submit queue index bound
1264	>	int nextSeed; // for initializing worker seeds
1265	>	volatile int mainLock; // spinlock for array updates
1266	>	volatile int runState; // shutdown status and seq
1267		WorkQueue[] workQueues; // main registry
1174	–	final Mutex lock; // for registration
1175	–	final Condition termination; // for awaitTermination
1268		final ForkJoinWorkerThreadFactory factory; // factory for new workers
1269		final Thread.UncaughtExceptionHandler ueh; // per-worker UEH
1178	–	final AtomicLong stealCount; // collect counts when terminated
1179	–	final AtomicInteger nextWorkerNumber; // to create worker name string
1270		final String workerNamePrefix; // to create worker name string
1271
1272	<	// Creating, registering, deregistering and running workers
1272	>	/*
1273	>	* Mechanics for main lock protecting worker array updates. Uses
1274	>	* the same strategy as ConcurrentHashMap bins -- a spinLock for
1275	>	* normal cases, but falling back to builtin lock when (rarely)
1276	>	* needed. See internal ConcurrentHashMap documentation for
1277	>	* explanation.
1278	>	*/
1279	>
1280	>	static final int LOCK_WAITING = 2; // bit to indicate need for signal
1281	>	static final int MAX_LOCK_SPINS = 1 << 8;
1282	>
1283	>	private void tryAwaitMainLock() {
1284	>	int spins = MAX_LOCK_SPINS, r = 0, h;
1285	>	while (((h = mainLock) & 1) != 0) {
1286	>	if (r == 0)
1287	>	r = ThreadLocalRandom.current().nextInt(); // randomize spins
1288	>	else if (spins >= 0) {
1289	>	r ^= r << 1; r ^= r >>> 3; r ^= r << 10; // xorshift
1290	>	if (r >= 0)
1291	>	--spins;
1292	>	}
1293	>	else if (U.compareAndSwapInt(this, MAINLOCK, h, h \| LOCK_WAITING)) {
1294	>	synchronized (this) {
1295	>	if ((mainLock & LOCK_WAITING) != 0) {
1296	>	try {
1297	>	wait();
1298	>	} catch (InterruptedException ie) {
1299	>	Thread.currentThread().interrupt();
1300	>	}
1301	>	}
1302	>	else
1303	>	notifyAll(); // possibly won race vs signaller
1304	>	}
1305	>	break;
1306	>	}
1307	>	}
1308	>	}
1309	>
1310	>	// Creating, registering, and deregistering workers
1311
1312		/**
1313		* Tries to create and start a worker
#	Line 1205 \| Line 1333 \| public class ForkJoinPool extends Abstra
1333		* ForkJoinWorkerThread.
1334		*/
1335		final String nextWorkerName() {
1336	<	return workerNamePrefix.concat
1337	<	(Integer.toString(nextWorkerNumber.addAndGet(1)));
1336	>	int n;
1337	>	do {} while(!U.compareAndSwapInt(this, NEXTWORKERNUMBER,
1338	>	n = nextWorkerNumber, ++n));
1339	>	return workerNamePrefix.concat(Integer.toString(n));
1340		}
1341
1342		/**
1343	<	* Callback from ForkJoinWorkerThread constructor to establish and
1344	<	* record its WorkQueue.
1343	>	* Callback from ForkJoinWorkerThread constructor to establish its
1344	>	* poolIndex and record its WorkQueue. To avoid scanning bias due
1345	>	* to packing entries in front of the workQueues array, we treat
1346	>	* the array as a simple power-of-two hash table using per-thread
1347	>	* seed as hash, expanding as needed.
1348		*
1349	<	* @param wt the worker thread
1349	>	* @param w the worker's queue
1350		*/
1351	<	final void registerWorker(ForkJoinWorkerThread wt) {
1352	<	WorkQueue w = wt.workQueue;
1353	<	Mutex lock = this.lock;
1221	<	lock.lock();
1351	>	final void registerWorker(WorkQueue w) {
1352	>	while (!U.compareAndSwapInt(this, MAINLOCK, 0, 1))
1353	>	tryAwaitMainLock();
1354		try {
1355	<	int g = growHints, k = g & SMASK;
1356	<	WorkQueue[] ws = workQueues;
1357	<	if (ws != null) { // ignore on shutdown
1358	<	int n = ws.length;
1359	<	if ((k & 1) == 0 \|\| k >= n \|\| ws[k] != null) {
1360	<	for (k = 1; k < n && ws[k] != null; k += 2)
1361	<	; // workers are at odd indices
1362	<	if (k >= n) // resize
1363	<	workQueues = ws = Arrays.copyOf(ws, n << 1);
1364	<	}
1365	<	w.eventCount = w.poolIndex = k; // establish before recording
1366	<	ws[k] = w;
1367	<	growHints = (g & ~SMASK) \| ((k + 2) & SMASK);
1368	<	int rs = runState;
1369	<	int m = rs & SMASK; // recalculate runState mask
1370	<	if (k > m)
1371	<	m = (m << 1) + 1;
1372	<	runState = (rs & SHUTDOWN) \| ((rs + RS_SEQ) & RS_SEQ_MASK) \| m;
1355	>	WorkQueue[] ws;
1356	>	if ((ws = workQueues) == null)
1357	>	ws = workQueues = new WorkQueue[submitMask + 1];
1358	>	if (w != null) {
1359	>	int rs, n = ws.length, m = n - 1;
1360	>	int s = nextSeed += SEED_INCREMENT; // rarely-colliding sequence
1361	>	w.seed = (s == 0) ? 1 : s; // ensure non-zero seed
1362	>	int r = (s << 1) \| 1; // use odd-numbered indices
1363	>	if (ws[r &= m] != null) { // collision
1364	>	int probes = 0; // step by approx half size
1365	>	int step = (n <= 4) ? 2 : ((n >>> 1) & SQMASK) + 2;
1366	>	while (ws[r = (r + step) & m] != null) {
1367	>	if (++probes >= n) {
1368	>	workQueues = ws = Arrays.copyOf(ws, n <<= 1);
1369	>	m = n - 1;
1370	>	probes = 0;
1371	>	}
1372	>	}
1373	>	}
1374	>	w.eventCount = w.poolIndex = r; // establish before recording
1375	>	ws[r] = w; // also update seq
1376	>	runState = ((rs = runState) & SHUTDOWN) \| ((rs + 2) & ~SHUTDOWN);
1377		}
1378		} finally {
1379	<	lock.unlock();
1379	>	if (!U.compareAndSwapInt(this, MAINLOCK, 1, 0)) {
1380	>	mainLock = 0;
1381	>	synchronized (this) { notifyAll(); };
1382	>	}
1383		}
1384	+
1385		}
1386
1387		/**
#	Line 1257 \| Line 1397 \| public class ForkJoinPool extends Abstra
1397		WorkQueue w = null;
1398		if (wt != null && (w = wt.workQueue) != null) {
1399		w.runState = -1; // ensure runState is set
1400	<	stealCount.getAndAdd(w.totalSteals + w.nsteals);
1400	>	long steals = w.totalSteals + w.nsteals, sc;
1401	>	do {} while(!U.compareAndSwapLong(this, STEALCOUNT,
1402	>	sc = stealCount, sc + steals));
1403		int idx = w.poolIndex;
1404	<	Mutex lock = this.lock;
1405	<	lock.lock();
1406	<	try { // remove record from array
1404	>	while (!U.compareAndSwapInt(this, MAINLOCK, 0, 1))
1405	>	tryAwaitMainLock();
1406	>	try {
1407		WorkQueue[] ws = workQueues;
1408	<	if (ws != null && idx >= 0 && idx < ws.length && ws[idx] == w) {
1408	>	if (ws != null && idx >= 0 && idx < ws.length && ws[idx] == w)
1409		ws[idx] = null;
1268	–	growHints = (growHints & ~SMASK) \| idx;
1269	–	}
1410		} finally {
1411	<	lock.unlock();
1411	>	if (!U.compareAndSwapInt(this, MAINLOCK, 1, 0)) {
1412	>	mainLock = 0;
1413	>	synchronized (this) { notifyAll(); };
1414	>	}
1415		}
1416		}
1417
#	Line 1290 \| Line 1433 \| public class ForkJoinPool extends Abstra
1433		U.throwException(ex);
1434		}
1435
1293	–	/**
1294	–	* Top-level runloop for workers, called by ForkJoinWorkerThread.run.
1295	–	*/
1296	–	final void runWorker(ForkJoinWorkerThread wt) {
1297	–	// Initialize queue array and seed in this thread
1298	–	WorkQueue w = wt.workQueue;
1299	–	w.growArray(false);
1300	–	w.seed = hashId(Thread.currentThread().getId());
1301	–
1302	–	do {} while (w.runTask(scan(w)));
1303	–	}
1304	–
1436		// Submissions
1437
1438		/**
1439		* Unless shutting down, adds the given task to a submission queue
1440		* at submitter's current queue index (modulo submission
1441	<	* range). If no queue exists at the index, one is created unless
1442	<	* pool lock is busy. If the queue and/or lock are busy, another
1443	<	* index is randomly chosen. The mask in growHints controls the
1444	<	* effective index range of queues considered. The mask is
1445	<	* expanded, up to the current workerQueue mask, upon any detected
1446	<	* contention but otherwise remains small to avoid needlessly
1316	<	* creating queues when there is no contention.
1441	>	* range). If no queue exists at the index, one is created. If
1442	>	* the queue is busy, another index is randomly chosen. The
1443	>	* submitMask bounds the effective number of queues to the
1444	>	* (nearest power of two for) parallelism level.
1445	>	*
1446	>	* @param task the task. Caller must ensure non-null.
1447		*/
1448		private void doSubmit(ForkJoinTask<?> task) {
1319	–	if (task == null)
1320	–	throw new NullPointerException();
1449		Submitter s = submitters.get();
1450	<	for (int r = s.seed, m = growHints >>> 16;;) {
1451	<	WorkQueue[] ws; WorkQueue q; Mutex lk;
1450	>	for (int r = s.seed, m = submitMask;;) {
1451	>	WorkQueue[] ws; WorkQueue q;
1452		int k = r & m & SQMASK; // use only even indices
1453	<	if (runState < 0 \|\| (ws = workQueues) == null \|\| ws.length <= k)
1453	>	if (runState < 0)
1454		throw new RejectedExecutionException(); // shutting down
1455	<	if ((q = ws[k]) == null && (lk = lock).tryAcquire(0)) {
1456	<	try { // try to create new queue
1457	<	if (ws == workQueues && (q = ws[k]) == null) {
1458	<	int rs; // update runState seq
1459	<	ws[k] = q = new WorkQueue(null, SHARED_QUEUE);
1460	<	runState = (((rs = runState) & SHUTDOWN) \|
1461	<	((rs + RS_SEQ) & ~SHUTDOWN));
1455	>	else if ((ws = workQueues) == null \|\| ws.length <= k) {
1456	>	while (!U.compareAndSwapInt(this, MAINLOCK, 0, 1))
1457	>	tryAwaitMainLock();
1458	>	try {
1459	>	if (workQueues == null)
1460	>	workQueues = new WorkQueue[submitMask + 1];
1461	>	} finally {
1462	>	if (!U.compareAndSwapInt(this, MAINLOCK, 1, 0)) {
1463	>	mainLock = 0;
1464	>	synchronized (this) { notifyAll(); };
1465	>	}
1466	>	}
1467	>	}
1468	>	else if ((q = ws[k]) == null) { // create new queue
1469	>	WorkQueue nq = new WorkQueue(this, null, SHARED_QUEUE);
1470	>	while (!U.compareAndSwapInt(this, MAINLOCK, 0, 1))
1471	>	tryAwaitMainLock();
1472	>	try {
1473	>	int rs = runState; // to update seq
1474	>	if (ws == workQueues && ws[k] == null) {
1475	>	ws[k] = nq;
1476	>	runState = ((rs & SHUTDOWN) \| ((rs + 2) & ~SHUTDOWN));
1477		}
1478		} finally {
1479	<	lk.unlock();
1479	>	if (!U.compareAndSwapInt(this, MAINLOCK, 1, 0)) {
1480	>	mainLock = 0;
1481	>	synchronized (this) { notifyAll(); };
1482	>	}
1483		}
1484		}
1485	<	if (q != null) {
1486	<	if (q.trySharedPush(task)) {
1487	<	signalWork();
1488	<	return;
1489	<	}
1490	<	else if (m < parallelism - 1 && m < (runState & SMASK)) {
1491	<	Mutex lock = this.lock;
1346	<	lock.lock(); // block until lock free
1347	<	int g = growHints;
1348	<	if (g >>> 16 == m) // expand range
1349	<	growHints = (((m << 1) + 1) << 16) \| (g & SMASK);
1350	<	lock.unlock(); // no need for try/finally
1351	<	}
1352	<	else if ((r & m) == 0)
1353	<	Thread.yield(); // occasionally yield if busy
1354	<	}
1355	<	if (m == (m = growHints >>> 16)) {
1356	<	r ^= r << 13; // update seed unless new range
1357	<	r ^= r >>> 17; // same xorshift as WorkQueues
1485	>	else if (q.trySharedPush(task)) {
1486	>	signalWork();
1487	>	return;
1488	>	}
1489	>	else if (m > 1) { // move to a different index
1490	>	r ^= r << 13; // same xorshift as WorkQueues
1491	>	r ^= r >>> 17;
1492		s.seed = r ^= r << 5;
1493		}
1494	+	else
1495	+	Thread.yield(); // yield if no alternatives
1496		}
1497		}
1498
1499	+	/**
1500	+	* Submits the given (non-null) task to the common pool, if possible.
1501	+	*/
1502	+	static void submitToCommonPool(ForkJoinTask<?> task) {
1503	+	ForkJoinPool p;
1504	+	if ((p = commonPool) == null)
1505	+	throw new RejectedExecutionException("Common Pool Unavailable");
1506	+	p.doSubmit(task);
1507	+	}
1508	+
1509	+	/**
1510	+	* Returns true if the given task was submitted to common pool
1511	+	* and has not yet commenced execution, and is available for
1512	+	* removal according to execution policies; if so removing the
1513	+	* submission from the pool.
1514	+	*
1515	+	* @param task the task
1516	+	* @return true if successful
1517	+	*/
1518	+	static boolean tryUnsubmitFromCommonPool(ForkJoinTask<?> task) {
1519	+	// Peek, looking for task and eligibility before
1520	+	// using trySharedUnpush to actually take it under lock
1521	+	ForkJoinPool p; WorkQueue[] ws; WorkQueue q;
1522	+	ForkJoinTask<?>[] a; int t, s, n;
1523	+	int k = submitters.get().seed & SQMASK;
1524	+	return ((p = commonPool) != null &&
1525	+	(ws = p.workQueues) != null &&
1526	+	ws.length > (k &= p.submitMask) &&
1527	+	(q = ws[k]) != null &&
1528	+	(a = q.array) != null &&
1529	+	(n = (t = q.top) - q.base) > 0 &&
1530	+	(n > 1 \|\| (int)(p.ctl >> AC_SHIFT) < 0) &&
1531	+	(s = t - 1) >= 0 && s < a.length && a[s] == task &&
1532	+	q.trySharedUnpush(task));
1533	+	}
1534	+
1535		// Maintaining ctl counts
1536
1537		/**
#	Line 1371 \| Line 1543 \| public class ForkJoinPool extends Abstra
1543		}
1544
1545		/**
1546	<	* Tries to activate or create a worker if too few are active.
1546	>	* Tries to create one or activate one or more workers if too few are active.
1547		*/
1548		final void signalWork() {
1549		long c; int u;
#	Line 1405 \| Line 1577 \| public class ForkJoinPool extends Abstra
1577		}
1578		}
1579
1580	+	// Scanning for tasks
1581	+
1582		/**
1583	<	* Tries to decrement active count (sometimes implicitly) and
1410	<	* possibly release or create a compensating worker in preparation
1411	<	* for blocking. Fails on contention or termination.
1412	<	*
1413	<	* @return true if the caller can block, else should recheck and retry
1583	>	* Top-level runloop for workers, called by ForkJoinWorkerThread.run.
1584		*/
1585	<	final boolean tryCompensate() {
1586	<	WorkQueue w; Thread p;
1587	<	int pc = parallelism, e, u, ac, tc, i;
1418	<	long c = ctl;
1419	<	WorkQueue[] ws = workQueues;
1420	<	if ((e = (int)c) >= 0) {
1421	<	if ((ac = ((u = (int)(c >>> 32)) >> UAC_SHIFT)) <= 0 &&
1422	<	e != 0 && ws != null && (i = e & SMASK) < ws.length &&
1423	<	(w = ws[i]) != null) {
1424	<	long nc = (long)(w.nextWait & E_MASK) \| (c & (AC_MASK\|TC_MASK));
1425	<	if (w.eventCount == (e \| INT_SIGN) &&
1426	<	U.compareAndSwapLong(this, CTL, c, nc)) {
1427	<	w.eventCount = (e + E_SEQ) & E_MASK;
1428	<	if ((p = w.parker) != null)
1429	<	U.unpark(p);
1430	<	return true; // release an idle worker
1431	<	}
1432	<	}
1433	<	else if ((tc = (short)(u >>> UTC_SHIFT)) >= 0 && ac + pc > 1) {
1434	<	long nc = ((c - AC_UNIT) & AC_MASK) \| (c & ~AC_MASK);
1435	<	if (U.compareAndSwapLong(this, CTL, c, nc))
1436	<	return true; // no compensation needed
1437	<	}
1438	<	else if (tc + pc < POOL_MAX) {
1439	<	long nc = ((c + TC_UNIT) & TC_MASK) \| (c & ~TC_MASK);
1440	<	if (U.compareAndSwapLong(this, CTL, c, nc)) {
1441	<	addWorker();
1442	<	return true; // create replacement
1443	<	}
1444	<	}
1445	<	}
1446	<	return false;
1585	>	final void runWorker(WorkQueue w) {
1586	>	w.growArray(false); // initialize queue array in this thread
1587	>	do { w.runTask(scan(w)); } while (w.runState >= 0);
1588		}
1589
1449	–	// Scanning for tasks
1450	–
1590		/**
1591		* Scans for and, if found, returns one task, else possibly
1592		* inactivates the worker. This method operates on single reads of
1593	<	* volatile state and is designed to be re-invoked continuously in
1594	<	* part because it returns upon detecting inconsistencies,
1593	>	* volatile state and is designed to be re-invoked continuously,
1594	>	* in part because it returns upon detecting inconsistencies,
1595		* contention, or state changes that indicate possible success on
1596		* re-invocation.
1597		*
1598	<	* The scan searches for tasks across queues, randomly selecting
1599	<	* the first #queues probes, favoring steals over submissions
1600	<	* (by exploiting even/odd indexing), and then performing a
1601	<	* circular sweep of all queues. The scan terminates upon either
1602	<	* finding a non-empty queue, or completing a full sweep. If the
1603	<	* worker is not inactivated, it takes and returns a task from
1604	<	* this queue. On failure to find a task, we take one of the
1605	<	* following actions, after which the caller will retry calling
1467	<	* this method unless terminated.
1598	>	* The scan searches for tasks across a random permutation of
1599	>	* queues (starting at a random index and stepping by a random
1600	>	* relative prime, checking each at least once). The scan
1601	>	* terminates upon either finding a non-empty queue, or completing
1602	>	* the sweep. If the worker is not inactivated, it takes and
1603	>	* returns a task from this queue. On failure to find a task, we
1604	>	* take one of the following actions, after which the caller will
1605	>	* retry calling this method unless terminated.
1606		*
1607		* * If pool is terminating, terminate the worker.
1608		*
#	Line 1475 \| Line 1613 \| public class ForkJoinPool extends Abstra
1613		* another worker, but with same net effect. Releasing in other
1614		* cases as well ensures that we have enough workers running.
1615		*
1478	–	* * If the caller has run a task since the last empty scan,
1479	–	* return (to allow rescan) if other workers are not also yet
1480	–	* enqueued. Field WorkQueue.rescans counts down on each scan to
1481	–	* ensure eventual inactivation and blocking.
1482	–	*
1616		* * If not already enqueued, try to inactivate and enqueue the
1617	<	* worker on wait queue.
1617	>	* worker on wait queue. Or, if inactivating has caused the pool
1618	>	* to be quiescent, relay to idleAwaitWork to check for
1619	>	* termination and possibly shrink pool.
1620	>	*
1621	>	* * If already inactive, and the caller has run a task since the
1622	>	* last empty scan, return (to allow rescan) unless others are
1623	>	* also inactivated. Field WorkQueue.rescans counts down on each
1624	>	* scan to ensure eventual inactivation and blocking.
1625		*
1626	<	* * If already enqueued and none of the above apply, either park
1627	<	* awaiting signal, or if this is the most recent waiter and pool
1488	<	* is quiescent, relay to idleAwaitWork to check for termination
1489	<	* and possibly shrink pool.
1626	>	* * If already enqueued and none of the above apply, park
1627	>	* awaiting signal,
1628		*
1629		* @param w the worker (via its WorkQueue)
1630	<	* @return a task or null of none found
1630	>	* @return a task or null if none found
1631		*/
1632		private final ForkJoinTask<?> scan(WorkQueue w) {
1633	<	boolean swept = false; // true after full empty scan
1634	<	WorkQueue[] ws; // volatile read order matters
1635	<	int r = w.seed, ec = w.eventCount; // ec is negative if inactive
1636	<	int rs = runState, m = rs & SMASK;
1637	<	if ((ws = workQueues) != null && ws.length > m) { // consistency check
1638	<	for (int k = 0, j = -1 - m; ; ++j) {
1639	<	WorkQueue q; int b;
1640	<	if (j < 0) { // random probes while j negative
1641	<	r ^= r << 13; r ^= r >>> 17; k = (r ^= r << 5) \| (j & 1);
1642	<	} // worker (not submit) for odd j
1643	<	else // cyclic scan when j >= 0
1644	<	k += 7; // step 7 reduces array packing bias
1645	<	if ((q = ws[k & m]) != null && (b = q.base) - q.top < 0) {
1646	<	ForkJoinTask<?> t = (ec >= 0) ? q.pollAt(b) : null;
1647	<	w.seed = r; // save seed for next scan
1648	<	if (t != null)
1633	>	WorkQueue[] ws; // first update random seed
1634	>	int r = w.seed; r ^= r << 13; r ^= r >>> 17; w.seed = r ^= r << 5;
1635	>	int rs = runState, m; // volatile read order matters
1636	>	if ((ws = workQueues) != null && (m = ws.length - 1) > 0) {
1637	>	int ec = w.eventCount; // ec is negative if inactive
1638	>	int step = (r >>> 16) \| 1; // relative prime
1639	>	for (int j = (m + 1) << 2; ; r += step) {
1640	>	WorkQueue q; ForkJoinTask<?> t; ForkJoinTask<?>[] a; int b;
1641	>	if ((q = ws[r & m]) != null && (b = q.base) - q.top < 0 &&
1642	>	(a = q.array) != null) { // probably nonempty
1643	>	int i = (((a.length - 1) & b) << ASHIFT) + ABASE;
1644	>	t = (ForkJoinTask<?>)U.getObjectVolatile(a, i);
1645	>	if (q.base == b && ec >= 0 && t != null &&
1646	>	U.compareAndSwapObject(a, i, t, null)) {
1647	>	if (q.top - (q.base = b + 1) > 0)
1648	>	signalWork(); // help pushes signal
1649		return t;
1650	<	break;
1650	>	}
1651	>	else if (ec < 0 \|\| j <= m) {
1652	>	rs = 0; // mark scan as imcomplete
1653	>	break; // caller can retry after release
1654	>	}
1655		}
1656	<	else if (j - m > m) {
1515	<	if (rs == runState) // staleness check
1516	<	swept = true;
1656	>	if (--j < 0)
1657		break;
1518	–	}
1658		}
1659
1521	–	// Decode ctl on empty scan
1660		long c = ctl; int e = (int)c, a = (int)(c >> AC_SHIFT), nr, ns;
1661	<	if (e < 0) // pool is terminating
1662	<	w.runState = -1;
1663	<	else if (!swept) { // try to release a waiter
1664	<	WorkQueue v; Thread p;
1665	<	if (e > 0 && a < 0 && (v = ws[e & m]) != null &&
1666	<	v.eventCount == (e \| INT_SIGN)) {
1661	>	if (e < 0) // decode ctl on empty scan
1662	>	w.runState = -1; // pool is terminating
1663	>	else if (rs == 0 \|\| rs != runState) { // incomplete scan
1664	>	WorkQueue v; Thread p; // try to release a waiter
1665	>	if (e > 0 && a < 0 && w.eventCount == ec &&
1666	>	(v = ws[e & m]) != null && v.eventCount == (e \| INT_SIGN)) {
1667		long nc = ((long)(v.nextWait & E_MASK) \|
1668		((c + AC_UNIT) & (AC_MASK\|TC_MASK)));
1669	<	if (U.compareAndSwapLong(this, CTL, c, nc)) {
1669	>	if (ctl == c && U.compareAndSwapLong(this, CTL, c, nc)) {
1670		v.eventCount = (e + E_SEQ) & E_MASK;
1671		if ((p = v.parker) != null)
1672		U.unpark(p);
1673		}
1674		}
1675		}
1676	<	else if ((nr = w.rescans) > 0) { // continue rescanning
1539	<	int ac = a + parallelism;
1540	<	if (((w.rescans = (ac < nr) ? ac : nr - 1) & 3) == 0 &&
1541	<	w.eventCount == ec)
1542	<	Thread.yield(); // occasionally yield
1543	<	}
1544	<	else if (ec >= 0) { // try to enqueue
1676	>	else if (ec >= 0) { // try to enqueue/inactivate
1677		long nc = (long)ec \| ((c - AC_UNIT) & (AC_MASK\|TC_MASK));
1678		w.nextWait = e;
1679	<	w.eventCount = ec \| INT_SIGN;// mark as inactive
1680	<	if (!U.compareAndSwapLong(this, CTL, c, nc))
1681	<	w.eventCount = ec; // unmark on CAS failure
1682	<	else if ((ns = w.nsteals) != 0) {
1683	<	w.nsteals = 0; // set rescans if ran task
1684	<	w.rescans = a + parallelism;
1685	<	w.totalSteals += ns;
1679	>	w.eventCount = ec \| INT_SIGN; // mark as inactive
1680	>	if (ctl != c \|\| !U.compareAndSwapLong(this, CTL, c, nc))
1681	>	w.eventCount = ec; // unmark on CAS failure
1682	>	else {
1683	>	if ((ns = w.nsteals) != 0) {
1684	>	w.nsteals = 0; // set rescans if ran task
1685	>	w.rescans = (a > 0) ? 0 : a + parallelism;
1686	>	w.totalSteals += ns;
1687	>	}
1688	>	if (a == 1 - parallelism) // quiescent
1689	>	idleAwaitWork(w, nc, c);
1690		}
1691		}
1692	<	else { // already queued
1693	<	if (parallelism == -a)
1694	<	idleAwaitWork(w); // quiescent
1695	<	if (w.eventCount == ec) {
1696	<	Thread.interrupted(); // clear status
1697	<	ForkJoinWorkerThread wt = w.owner;
1692	>	else if (w.eventCount < 0) { // already queued
1693	>	int ac = a + parallelism;
1694	>	if ((nr = w.rescans) > 0) // continue rescanning
1695	>	w.rescans = (ac < nr) ? ac : nr - 1;
1696	>	else if (((w.seed >>> 16) & ac) == 0) { // randomize park
1697	>	Thread.interrupted(); // clear status
1698	>	Thread wt = Thread.currentThread();
1699		U.putObject(wt, PARKBLOCKER, this);
1700	<	w.parker = wt; // emulate LockSupport.park
1701	<	if (w.eventCount == ec) // recheck
1702	<	U.park(false, 0L); // block
1700	>	w.parker = wt; // emulate LockSupport.park
1701	>	if (w.eventCount < 0) // recheck
1702	>	U.park(false, 0L);
1703		w.parker = null;
1704		U.putObject(wt, PARKBLOCKER, null);
1705		}
#	Line 1572 \| Line 1709 \| public class ForkJoinPool extends Abstra
1709		}
1710
1711		/**
1712	<	* If inactivating worker w has caused pool to become quiescent,
1713	<	* checks for pool termination, and, so long as this is not the
1714	<	* only worker, waits for event for up to SHRINK_RATE nanosecs.
1715	<	* On timeout, if ctl has not changed, terminates the worker,
1716	<	* which will in turn wake up another worker to possibly repeat
1717	<	* this process.
1712	>	* If inactivating worker w has caused the pool to become
1713	>	* quiescent, checks for pool termination, and, so long as this is
1714	>	* not the only worker, waits for event for up to a given
1715	>	* duration. On timeout, if ctl has not changed, terminates the
1716	>	* worker, which will in turn wake up another worker to possibly
1717	>	* repeat this process.
1718		*
1719		* @param w the calling worker
1720	+	* @param currentCtl the ctl value triggering possible quiescence
1721	+	* @param prevCtl the ctl value to restore if thread is terminated
1722		*/
1723	<	private void idleAwaitWork(WorkQueue w) {
1724	<	long c; int nw, ec;
1725	<	if (!tryTerminate(false, false) &&
1726	<	(int)((c = ctl) >> AC_SHIFT) + parallelism == 0 &&
1727	<	(ec = w.eventCount) == ((int)c \| INT_SIGN) &&
1728	<	(nw = w.nextWait) != 0) {
1729	<	long nc = ((long)(nw & E_MASK) \| // ctl to restore on timeout
1730	<	((c + AC_UNIT) & AC_MASK) \| (c & TC_MASK));
1592	<	ForkJoinWorkerThread wt = w.owner;
1593	<	while (ctl == c) {
1594	<	long startTime = System.nanoTime();
1723	>	private void idleAwaitWork(WorkQueue w, long currentCtl, long prevCtl) {
1724	>	if (w.eventCount < 0 && !tryTerminate(false, false) &&
1725	>	(int)prevCtl != 0 && !hasQueuedSubmissions() && ctl == currentCtl) {
1726	>	int dc = -(short)(currentCtl >>> TC_SHIFT);
1727	>	long parkTime = dc < 0 ? FAST_IDLE_TIMEOUT: (dc + 1) * IDLE_TIMEOUT;
1728	>	long deadline = System.nanoTime() + parkTime - 100000L; // 1ms slop
1729	>	Thread wt = Thread.currentThread();
1730	>	while (ctl == currentCtl) {
1731		Thread.interrupted(); // timed variant of version in scan()
1732		U.putObject(wt, PARKBLOCKER, this);
1733		w.parker = wt;
1734	<	if (ctl == c)
1735	<	U.park(false, SHRINK_RATE);
1734	>	if (ctl == currentCtl)
1735	>	U.park(false, parkTime);
1736		w.parker = null;
1737		U.putObject(wt, PARKBLOCKER, null);
1738	<	if (ctl != c)
1738	>	if (ctl != currentCtl)
1739		break;
1740	<	if (System.nanoTime() - startTime >= SHRINK_TIMEOUT &&
1741	<	U.compareAndSwapLong(this, CTL, c, nc)) {
1742	<	w.eventCount = (ec + E_SEQ) \| E_MASK;
1743	<	w.runState = -1; // shrink
1740	>	if (deadline - System.nanoTime() <= 0L &&
1741	>	U.compareAndSwapLong(this, CTL, currentCtl, prevCtl)) {
1742	>	w.eventCount = (w.eventCount + E_SEQ) \| E_MASK;
1743	>	w.runState = -1; // shrink
1744		break;
1745		}
1746		}
#	Line 1622 \| Line 1758 \| public class ForkJoinPool extends Abstra
1758		* leaves hints in workers to speed up subsequent calls. The
1759		* implementation is very branchy to cope with potential
1760		* inconsistencies or loops encountering chains that are stale,
1761	<	* unknown, or of length greater than MAX_HELP_DEPTH links. All
1626	<	* of these cases are dealt with by just retrying by caller.
1761	>	* unknown, or so long that they are likely cyclic.
1762		*
1763		* @param joiner the joining worker
1764		* @param task the task to join
1765	<	* @return true if found or ran a task (and so is immediately retryable)
1765	>	* @return 0 if no progress can be made, negative if task
1766	>	* known complete, else positive
1767		*/
1768	<	final boolean tryHelpStealer(WorkQueue joiner, ForkJoinTask<?> task) {
1769	<	ForkJoinTask<?> subtask; // current target
1770	<	boolean progress = false;
1771	<	int depth = 0; // current chain depth
1772	<	int m = runState & SMASK;
1773	<	WorkQueue[] ws = workQueues;
1774	<
1775	<	if (ws != null && ws.length > m && (subtask = task).status >= 0) {
1776	<	outer:for (WorkQueue j = joiner;;) {
1777	<	// Try to find the stealer of subtask, by first using hint
1642	<	WorkQueue stealer = null;
1643	<	WorkQueue v = ws[j.stealHint & m];
1644	<	if (v != null && v.currentSteal == subtask)
1645	<	stealer = v;
1646	<	else {
1647	<	for (int i = 1; i <= m; i += 2) {
1648	<	if ((v = ws[i]) != null && v.currentSteal == subtask) {
1649	<	stealer = v;
1650	<	j.stealHint = i; // save hint
1651	<	break;
1652	<	}
1768	>	private int tryHelpStealer(WorkQueue joiner, ForkJoinTask<?> task) {
1769	>	int stat = 0, steps = 0; // bound to avoid cycles
1770	>	if (joiner != null && task != null) { // hoist null checks
1771	>	restart: for (;;) {
1772	>	ForkJoinTask<?> subtask = task; // current target
1773	>	for (WorkQueue j = joiner, v;;) { // v is stealer of subtask
1774	>	WorkQueue[] ws; int m, s, h;
1775	>	if ((s = task.status) < 0) {
1776	>	stat = s;
1777	>	break restart;
1778		}
1779	<	if (stealer == null)
1780	<	break;
1781	<	}
1782	<
1783	<	for (WorkQueue q = stealer;;) { // Try to help stealer
1784	<	ForkJoinTask<?> t; int b;
1785	<	if (task.status < 0)
1786	<	break outer;
1787	<	if ((b = q.base) - q.top < 0) {
1788	<	progress = true;
1789	<	if (subtask.status < 0)
1790	<	break outer; // stale
1791	<	if ((t = q.pollAt(b)) != null) {
1792	<	stealer.stealHint = joiner.poolIndex;
1793	<	joiner.runSubtask(t);
1779	>	if ((ws = workQueues) == null \|\| (m = ws.length - 1) <= 0)
1780	>	break restart; // shutting down
1781	>	if ((v = ws[h = (j.stealHint \| 1) & m]) == null \|\|
1782	>	v.currentSteal != subtask) {
1783	>	for (int origin = h;;) { // find stealer
1784	>	if (((h = (h + 2) & m) & 15) == 1 &&
1785	>	(subtask.status < 0 \|\| j.currentJoin != subtask))
1786	>	continue restart; // occasional staleness check
1787	>	if ((v = ws[h]) != null &&
1788	>	v.currentSteal == subtask) {
1789	>	j.stealHint = h; // save hint
1790	>	break;
1791	>	}
1792	>	if (h == origin)
1793	>	break restart; // cannot find stealer
1794		}
1795		}
1796	<	else { // empty - try to descend to find stealer's stealer
1797	<	ForkJoinTask<?> next = stealer.currentJoin;
1798	<	if (++depth == MAX_HELP_DEPTH \|\| subtask.status < 0 \|\|
1799	<	next == null \|\| next == subtask)
1800	<	break outer; // max depth, stale, dead-end, cyclic
1801	<	subtask = next;
1802	<	j = stealer;
1803	<	break;
1796	>	for (;;) { // help stealer or descend to its stealer
1797	>	ForkJoinTask[] a; int b;
1798	>	if (subtask.status < 0) // surround probes with
1799	>	continue restart; // consistency checks
1800	>	if ((b = v.base) - v.top < 0 && (a = v.array) != null) {
1801	>	int i = (((a.length - 1) & b) << ASHIFT) + ABASE;
1802	>	ForkJoinTask<?> t =
1803	>	(ForkJoinTask<?>)U.getObjectVolatile(a, i);
1804	>	if (subtask.status < 0 \|\| j.currentJoin != subtask \|\|
1805	>	v.currentSteal != subtask)
1806	>	continue restart; // stale
1807	>	stat = 1; // apparent progress
1808	>	if (t != null && v.base == b &&
1809	>	U.compareAndSwapObject(a, i, t, null)) {
1810	>	v.base = b + 1; // help stealer
1811	>	joiner.runSubtask(t);
1812	>	}
1813	>	else if (v.base == b && ++steps == MAX_HELP)
1814	>	break restart; // v apparently stalled
1815	>	}
1816	>	else { // empty -- try to descend
1817	>	ForkJoinTask<?> next = v.currentJoin;
1818	>	if (subtask.status < 0 \|\| j.currentJoin != subtask \|\|
1819	>	v.currentSteal != subtask)
1820	>	continue restart; // stale
1821	>	else if (next == null \|\| ++steps == MAX_HELP)
1822	>	break restart; // dead-end or maybe cyclic
1823	>	else {
1824	>	subtask = next;
1825	>	j = v;
1826	>	break;
1827	>	}
1828	>	}
1829		}
1830		}
1831		}
1832		}
1833	<	return progress;
1833	>	return stat;
1834		}
1835
1836		/**
#	Line 1689 \| Line 1839 \| public class ForkJoinPool extends Abstra
1839		* @param joiner the joining worker
1840		* @param task the task
1841		*/
1842	<	final void tryPollForAndExec(WorkQueue joiner, ForkJoinTask<?> task) {
1842	>	private void tryPollForAndExec(WorkQueue joiner, ForkJoinTask<?> task) {
1843		WorkQueue[] ws;
1844	<	int m = runState & SMASK;
1845	<	if ((ws = workQueues) != null && ws.length > m) {
1696	<	for (int j = 1; j <= m && task.status >= 0; j += 2) {
1844	>	if ((ws = workQueues) != null) {
1845	>	for (int j = 1; j < ws.length && task.status >= 0; j += 2) {
1846		WorkQueue q = ws[j];
1847		if (q != null && q.pollFor(task)) {
1848		joiner.runSubtask(task);
#	Line 1704 \| Line 1853 \| public class ForkJoinPool extends Abstra
1853		}
1854
1855		/**
1856	<	* Returns a non-empty steal queue, if one is found during a random,
1857	<	* then cyclic scan, else null. This method must be retried by
1858	<	* caller if, by the time it tries to use the queue, it is empty.
1856	>	* Tries to decrement active count (sometimes implicitly) and
1857	>	* possibly release or create a compensating worker in preparation
1858	>	* for blocking. Fails on contention or termination. Otherwise,
1859	>	* adds a new thread if no idle workers are available and either
1860	>	* pool would become completely starved or: (at least half
1861	>	* starved, and fewer than 50% spares exist, and there is at least
1862	>	* one task apparently available). Even though the availability
1863	>	* check requires a full scan, it is worthwhile in reducing false
1864	>	* alarms.
1865	>	*
1866	>	* @param task if non-null, a task being waited for
1867	>	* @param blocker if non-null, a blocker being waited for
1868	>	* @return true if the caller can block, else should recheck and retry
1869	>	*/
1870	>	final boolean tryCompensate(ForkJoinTask<?> task, ManagedBlocker blocker) {
1871	>	int pc = parallelism, e;
1872	>	long c = ctl;
1873	>	WorkQueue[] ws = workQueues;
1874	>	if ((e = (int)c) >= 0 && ws != null) {
1875	>	int u, a, ac, hc;
1876	>	int tc = (short)((u = (int)(c >>> 32)) >>> UTC_SHIFT) + pc;
1877	>	boolean replace = false;
1878	>	if ((a = u >> UAC_SHIFT) <= 0) {
1879	>	if ((ac = a + pc) <= 1)
1880	>	replace = true;
1881	>	else if ((e > 0 \|\| (task != null &&
1882	>	ac <= (hc = pc >>> 1) && tc < pc + hc))) {
1883	>	WorkQueue w;
1884	>	for (int j = 0; j < ws.length; ++j) {
1885	>	if ((w = ws[j]) != null && !w.isEmpty()) {
1886	>	replace = true;
1887	>	break; // in compensation range and tasks available
1888	>	}
1889	>	}
1890	>	}
1891	>	}
1892	>	if ((task == null \|\| task.status >= 0) && // recheck need to block
1893	>	(blocker == null \|\| !blocker.isReleasable()) && ctl == c) {
1894	>	if (!replace) { // no compensation
1895	>	long nc = ((c - AC_UNIT) & AC_MASK) \| (c & ~AC_MASK);
1896	>	if (U.compareAndSwapLong(this, CTL, c, nc))
1897	>	return true;
1898	>	}
1899	>	else if (e != 0) { // release an idle worker
1900	>	WorkQueue w; Thread p; int i;
1901	>	if ((i = e & SMASK) < ws.length && (w = ws[i]) != null) {
1902	>	long nc = ((long)(w.nextWait & E_MASK) \|
1903	>	(c & (AC_MASK\|TC_MASK)));
1904	>	if (w.eventCount == (e \| INT_SIGN) &&
1905	>	U.compareAndSwapLong(this, CTL, c, nc)) {
1906	>	w.eventCount = (e + E_SEQ) & E_MASK;
1907	>	if ((p = w.parker) != null)
1908	>	U.unpark(p);
1909	>	return true;
1910	>	}
1911	>	}
1912	>	}
1913	>	else if (tc < MAX_CAP) { // create replacement
1914	>	long nc = ((c + TC_UNIT) & TC_MASK) \| (c & ~TC_MASK);
1915	>	if (U.compareAndSwapLong(this, CTL, c, nc)) {
1916	>	addWorker();
1917	>	return true;
1918	>	}
1919	>	}
1920	>	}
1921	>	}
1922	>	return false;
1923	>	}
1924	>
1925	>	/**
1926	>	* Helps and/or blocks until the given task is done.
1927	>	*
1928	>	* @param joiner the joining worker
1929	>	* @param task the task
1930	>	* @return task status on exit
1931	>	*/
1932	>	final int awaitJoin(WorkQueue joiner, ForkJoinTask<?> task) {
1933	>	int s;
1934	>	if ((s = task.status) >= 0) {
1935	>	ForkJoinTask<?> prevJoin = joiner.currentJoin;
1936	>	joiner.currentJoin = task;
1937	>	long startTime = 0L;
1938	>	for (int k = 0;;) {
1939	>	if ((s = (joiner.isEmpty() ? // try to help
1940	>	tryHelpStealer(joiner, task) :
1941	>	joiner.tryRemoveAndExec(task))) == 0 &&
1942	>	(s = task.status) >= 0) {
1943	>	if (k == 0) {
1944	>	startTime = System.nanoTime();
1945	>	tryPollForAndExec(joiner, task); // check uncommon case
1946	>	}
1947	>	else if ((k & (MAX_HELP - 1)) == 0 &&
1948	>	System.nanoTime() - startTime >=
1949	>	COMPENSATION_DELAY &&
1950	>	tryCompensate(task, null)) {
1951	>	if (task.trySetSignal()) {
1952	>	synchronized (task) {
1953	>	if (task.status >= 0) {
1954	>	try { // see ForkJoinTask
1955	>	task.wait(); // for explanation
1956	>	} catch (InterruptedException ie) {
1957	>	}
1958	>	}
1959	>	else
1960	>	task.notifyAll();
1961	>	}
1962	>	}
1963	>	long c; // re-activate
1964	>	do {} while (!U.compareAndSwapLong
1965	>	(this, CTL, c = ctl, c + AC_UNIT));
1966	>	}
1967	>	}
1968	>	if (s < 0 \|\| (s = task.status) < 0) {
1969	>	joiner.currentJoin = prevJoin;
1970	>	break;
1971	>	}
1972	>	else if ((k++ & (MAX_HELP - 1)) == MAX_HELP >>> 1)
1973	>	Thread.yield(); // for politeness
1974	>	}
1975	>	}
1976	>	return s;
1977	>	}
1978	>
1979	>	/**
1980	>	* Stripped-down variant of awaitJoin used by timed joins. Tries
1981	>	* to help join only while there is continuous progress. (Caller
1982	>	* will then enter a timed wait.)
1983	>	*
1984	>	* @param joiner the joining worker
1985	>	* @param task the task
1986	>	* @return task status on exit
1987	>	*/
1988	>	final int helpJoinOnce(WorkQueue joiner, ForkJoinTask<?> task) {
1989	>	int s;
1990	>	while ((s = task.status) >= 0 &&
1991	>	(joiner.isEmpty() ?
1992	>	tryHelpStealer(joiner, task) :
1993	>	joiner.tryRemoveAndExec(task)) != 0)
1994	>	;
1995	>	return s;
1996	>	}
1997	>
1998	>	/**
1999	>	* Returns a (probably) non-empty steal queue, if one is found
2000	>	* during a random, then cyclic scan, else null. This method must
2001	>	* be retried by caller if, by the time it tries to use the queue,
2002	>	* it is empty.
2003		*/
2004		private WorkQueue findNonEmptyStealQueue(WorkQueue w) {
2005	<	int r = w.seed; // Same idea as scan(), but ignoring submissions
2005	>	// Similar to loop in scan(), but ignoring submissions
2006	>	int r = w.seed; r ^= r << 13; r ^= r >>> 17; w.seed = r ^= r << 5;
2007	>	int step = (r >>> 16) \| 1;
2008		for (WorkQueue[] ws;;) {
2009	<	int m = runState & SMASK;
2010	<	if ((ws = workQueues) == null)
2009	>	int rs = runState, m;
2010	>	if ((ws = workQueues) == null \|\| (m = ws.length - 1) < 1)
2011		return null;
2012	<	if (ws.length > m) {
2013	<	WorkQueue q;
2014	<	for (int k = 0, j = -1 - m;; ++j) {
2015	<	if (j < 0) {
2016	<	r ^= r << 13; r ^= r >>> 17; k = r ^= r << 5;
2017	<	}
1723	<	else
1724	<	k += 7;
1725	<	if ((q = ws[(k \| 1) & m]) != null && q.base - q.top < 0) {
1726	<	w.seed = r;
1727	<	return q;
1728	<	}
1729	<	else if (j - m > m)
2012	>	for (int j = (m + 1) << 2; ; r += step) {
2013	>	WorkQueue q = ws[((r << 1) \| 1) & m];
2014	>	if (q != null && !q.isEmpty())
2015	>	return q;
2016	>	else if (--j < 0) {
2017	>	if (runState == rs)
2018		return null;
2019	+	break;
2020		}
2021		}
2022		}
#	Line 1741 \| Line 2030 \| public class ForkJoinPool extends Abstra
2030		*/
2031		final void helpQuiescePool(WorkQueue w) {
2032		for (boolean active = true;;) {
2033	<	w.runLocalTasks(); // exhaust local queue
2033	>	ForkJoinTask<?> localTask; // exhaust local queue
2034	>	while ((localTask = w.nextLocalTask()) != null)
2035	>	localTask.doExec();
2036		WorkQueue q = findNonEmptyStealQueue(w);
2037		if (q != null) {
2038	<	ForkJoinTask<?> t;
2038	>	ForkJoinTask<?> t; int b;
2039		if (!active) { // re-establish active count
2040		long c;
2041		active = true;
2042		do {} while (!U.compareAndSwapLong
2043		(this, CTL, c = ctl, c + AC_UNIT));
2044		}
2045	<	if ((t = q.poll()) != null)
2045	>	if ((b = q.base) - q.top < 0 && (t = q.pollAt(b)) != null)
2046		w.runSubtask(t);
2047		}
2048		else {
#	Line 1773 \| Line 2064 \| public class ForkJoinPool extends Abstra
2064		}
2065
2066		/**
2067	+	* Restricted version of helpQuiescePool for non-FJ callers
2068	+	*/
2069	+	static void externalHelpQuiescePool() {
2070	+	ForkJoinPool p; WorkQueue[] ws; WorkQueue w, q;
2071	+	ForkJoinTask<?> t; int b;
2072	+	int k = submitters.get().seed & SQMASK;
2073	+	if ((p = commonPool) != null &&
2074	+	(ws = p.workQueues) != null &&
2075	+	ws.length > (k &= p.submitMask) &&
2076	+	(w = ws[k]) != null &&
2077	+	(q = p.findNonEmptyStealQueue(w)) != null &&
2078	+	(b = q.base) - q.top < 0 &&
2079	+	(t = q.pollAt(b)) != null)
2080	+	t.doExec();
2081	+	}
2082	+
2083	+	/**
2084		* Gets and removes a local or stolen task for the given worker.
2085		*
2086		* @return a task, if available
2087		*/
2088		final ForkJoinTask<?> nextTaskFor(WorkQueue w) {
2089		for (ForkJoinTask<?> t;;) {
2090	<	WorkQueue q;
2090	>	WorkQueue q; int b;
2091		if ((t = w.nextLocalTask()) != null)
2092		return t;
2093		if ((q = findNonEmptyStealQueue(w)) == null)
2094		return null;
2095	<	if ((t = q.poll()) != null)
2095	>	if ((b = q.base) - q.top < 0 && (t = q.pollAt(b)) != null)
2096		return t;
2097		}
2098		}
#	Line 1805 \| Line 2113 \| public class ForkJoinPool extends Abstra
2113		8);
2114		}
2115
2116	+	/**
2117	+	* Returns approximate submission queue length for the given caller
2118	+	*/
2119	+	static int getEstimatedSubmitterQueueLength() {
2120	+	ForkJoinPool p; WorkQueue[] ws; WorkQueue q;
2121	+	int k = submitters.get().seed & SQMASK;
2122	+	return ((p = commonPool) != null &&
2123	+	p.runState >= 0 &&
2124	+	(ws = p.workQueues) != null &&
2125	+	ws.length > (k &= p.submitMask) &&
2126	+	(q = ws[k]) != null) ?
2127	+	q.queueSize() : 0;
2128	+	}
2129	+
2130		// Termination
2131
2132		/**
#	Line 1822 \| Line 2144 \| public class ForkJoinPool extends Abstra
2144		* @return true if now terminating or terminated
2145		*/
2146		private boolean tryTerminate(boolean now, boolean enable) {
1825	–	Mutex lock = this.lock;
2147		for (long c;;) {
2148		if (((c = ctl) & STOP_BIT) != 0) { // already terminating
2149		if ((short)(c >>> TC_SHIFT) == -parallelism) {
2150	<	lock.lock(); // don't need try/finally
2151	<	termination.signalAll(); // signal when 0 workers
2152	<	lock.unlock();
2150	>	synchronized(this) {
2151	>	notifyAll(); // signal when 0 workers
2152	>	}
2153		}
2154		return true;
2155		}
2156		if (runState >= 0) { // not yet enabled
2157		if (!enable)
2158		return false;
2159	<	lock.lock();
2160	<	runState \|= SHUTDOWN;
2161	<	lock.unlock();
2159	>	while (!U.compareAndSwapInt(this, MAINLOCK, 0, 1))
2160	>	tryAwaitMainLock();
2161	>	try {
2162	>	runState \|= SHUTDOWN;
2163	>	} finally {
2164	>	if (!U.compareAndSwapInt(this, MAINLOCK, 1, 0)) {
2165	>	mainLock = 0;
2166	>	synchronized (this) { notifyAll(); };
2167	>	}
2168	>	}
2169		}
2170		if (!now) { // check if idle & no tasks
2171		if ((int)(c >> AC_SHIFT) != -parallelism \|\|
#	Line 1959 \| Line 2287 \| public class ForkJoinPool extends Abstra
2287		checkPermission();
2288		if (factory == null)
2289		throw new NullPointerException();
2290	<	if (parallelism <= 0 \|\| parallelism > POOL_MAX)
2290	>	if (parallelism <= 0 \|\| parallelism > MAX_CAP)
2291		throw new IllegalArgumentException();
2292		this.parallelism = parallelism;
2293		this.factory = factory;
2294		this.ueh = handler;
2295		this.localMode = asyncMode ? FIFO_QUEUE : LIFO_QUEUE;
1968	–	this.growHints = 1;
2296		long np = (long)(-parallelism); // offset ctl counts
2297		this.ctl = ((np << AC_SHIFT) & AC_MASK) \| ((np << TC_SHIFT) & TC_MASK);
2298	<	// initialize workQueues array with room for 2*parallelism if possible
2299	<	int n = parallelism << 1;
2300	<	if (n >= POOL_MAX)
2301	<	n = POOL_MAX;
2302	<	else { // See Hackers Delight, sec 3.2, where n < (1 << 16)
1976	<	n \|= n >>> 1; n \|= n >>> 2; n \|= n >>> 4; n \|= n >>> 8;
1977	<	}
1978	<	this.workQueues = new WorkQueue[(n + 1) << 1]; // #slots = 2 * #workers
1979	<	this.termination = (this.lock = new Mutex()).newCondition();
1980	<	this.stealCount = new AtomicLong();
1981	<	this.nextWorkerNumber = new AtomicInteger();
2298	>	// Use nearest power 2 for workQueues size. See Hackers Delight sec 3.2.
2299	>	int n = parallelism - 1;
2300	>	n \|= n >>> 1; n \|= n >>> 2; n \|= n >>> 4; n \|= n >>> 8; n \|= n >>> 16;
2301	>	this.submitMask = ((n + 1) << 1) - 1;
2302	>	int pn = poolNumberGenerator.incrementAndGet();
2303		StringBuilder sb = new StringBuilder("ForkJoinPool-");
2304	<	sb.append(poolNumberGenerator.incrementAndGet());
2304	>	sb.append(Integer.toString(pn));
2305		sb.append("-worker-");
2306		this.workerNamePrefix = sb.toString();
2307	+	this.runState = 1; // set init flag
2308	+	}
2309	+
2310	+	/**
2311	+	* Constructor for common pool, suitable only for static initialization.
2312	+	* Basically the same as above, but uses smallest possible initial footprint.
2313	+	*/
2314	+	ForkJoinPool(int parallelism, int submitMask,
2315	+	ForkJoinWorkerThreadFactory factory,
2316	+	Thread.UncaughtExceptionHandler handler) {
2317	+	this.factory = factory;
2318	+	this.ueh = handler;
2319	+	this.submitMask = submitMask;
2320	+	this.parallelism = parallelism;
2321	+	long np = (long)(-parallelism);
2322	+	this.ctl = ((np << AC_SHIFT) & AC_MASK) \| ((np << TC_SHIFT) & TC_MASK);
2323	+	this.localMode = LIFO_QUEUE;
2324	+	this.workerNamePrefix = "ForkJoinPool.commonPool-worker-";
2325	+	this.runState = 1;
2326	+	}
2327	+
2328	+	/**
2329	+	* Returns the common pool instance.
2330	+	*
2331	+	* @return the common pool instance
2332	+	*/
2333	+	public static ForkJoinPool commonPool() {
2334	+	ForkJoinPool p;
2335	+	if ((p = commonPool) == null)
2336	+	throw new Error("Common Pool Unavailable");
2337	+	return p;
2338		}
2339
2340		// Execution methods
#	Line 2004 \| Line 2356 \| public class ForkJoinPool extends Abstra
2356		* scheduled for execution
2357		*/
2358		public <T> T invoke(ForkJoinTask<T> task) {
2359	+	if (task == null)
2360	+	throw new NullPointerException();
2361		doSubmit(task);
2362		return task.join();
2363		}
#	Line 2017 \| Line 2371 \| public class ForkJoinPool extends Abstra
2371		* scheduled for execution
2372		*/
2373		public void execute(ForkJoinTask<?> task) {
2374	+	if (task == null)
2375	+	throw new NullPointerException();
2376		doSubmit(task);
2377		}
2378
#	Line 2034 \| Line 2390 \| public class ForkJoinPool extends Abstra
2390		if (task instanceof ForkJoinTask<?>) // avoid re-wrap
2391		job = (ForkJoinTask<?>) task;
2392		else
2393	<	job = ForkJoinTask.adapt(task, null);
2393	>	job = new ForkJoinTask.AdaptedRunnableAction(task);
2394		doSubmit(job);
2395		}
2396
#	Line 2048 \| Line 2404 \| public class ForkJoinPool extends Abstra
2404		* scheduled for execution
2405		*/
2406		public <T> ForkJoinTask<T> submit(ForkJoinTask<T> task) {
2407	+	if (task == null)
2408	+	throw new NullPointerException();
2409		doSubmit(task);
2410		return task;
2411		}
#	Line 2058 \| Line 2416 \| public class ForkJoinPool extends Abstra
2416		* scheduled for execution
2417		*/
2418		public <T> ForkJoinTask<T> submit(Callable<T> task) {
2419	<	if (task == null)
2062	<	throw new NullPointerException();
2063	<	ForkJoinTask<T> job = ForkJoinTask.adapt(task);
2419	>	ForkJoinTask<T> job = new ForkJoinTask.AdaptedCallable<T>(task);
2420		doSubmit(job);
2421		return job;
2422		}
#	Line 2071 \| Line 2427 \| public class ForkJoinPool extends Abstra
2427		* scheduled for execution
2428		*/
2429		public <T> ForkJoinTask<T> submit(Runnable task, T result) {
2430	<	if (task == null)
2075	<	throw new NullPointerException();
2076	<	ForkJoinTask<T> job = ForkJoinTask.adapt(task, result);
2430	>	ForkJoinTask<T> job = new ForkJoinTask.AdaptedRunnable<T>(task, result);
2431		doSubmit(job);
2432		return job;
2433		}
#	Line 2090 \| Line 2444 \| public class ForkJoinPool extends Abstra
2444		if (task instanceof ForkJoinTask<?>) // avoid re-wrap
2445		job = (ForkJoinTask<?>) task;
2446		else
2447	<	job = ForkJoinTask.adapt(task, null);
2447	>	job = new ForkJoinTask.AdaptedRunnableAction(task);
2448		doSubmit(job);
2449		return job;
2450		}
#	Line 2112 \| Line 2466 \| public class ForkJoinPool extends Abstra
2466		boolean done = false;
2467		try {
2468		for (Callable<T> t : tasks) {
2469	<	ForkJoinTask<T> f = ForkJoinTask.adapt(t);
2469	>	ForkJoinTask<T> f = new ForkJoinTask.AdaptedCallable<T>(t);
2470		doSubmit(f);
2471		fs.add(f);
2472		}
#	Line 2156 \| Line 2510 \| public class ForkJoinPool extends Abstra
2510		}
2511
2512		/**
2513	+	* Returns the targeted parallelism level of the common pool.
2514	+	*
2515	+	* @return the targeted parallelism level of the common pool
2516	+	*/
2517	+	public static int getCommonPoolParallelism() {
2518	+	return commonPoolParallelism;
2519	+	}
2520	+
2521	+	/**
2522		* Returns the number of worker threads that have started but not
2523		* yet terminated. The result returned by this method may differ
2524		* from {@link #getParallelism} when threads are created to
#	Line 2236 \| Line 2599 \| public class ForkJoinPool extends Abstra
2599		* @return the number of steals
2600		*/
2601		public long getStealCount() {
2602	<	long count = stealCount.get();
2602	>	long count = stealCount;
2603		WorkQueue[] ws; WorkQueue w;
2604		if ((ws = workQueues) != null) {
2605		for (int i = 1; i < ws.length; i += 2) {
#	Line 2298 \| Line 2661 \| public class ForkJoinPool extends Abstra
2661		WorkQueue[] ws; WorkQueue w;
2662		if ((ws = workQueues) != null) {
2663		for (int i = 0; i < ws.length; i += 2) {
2664	<	if ((w = ws[i]) != null && w.queueSize() != 0)
2664	>	if ((w = ws[i]) != null && !w.isEmpty())
2665		return true;
2666		}
2667		}
#	Line 2366 \| Line 2729 \| public class ForkJoinPool extends Abstra
2729		public String toString() {
2730		// Use a single pass through workQueues to collect counts
2731		long qt = 0L, qs = 0L; int rc = 0;
2732	<	long st = stealCount.get();
2732	>	long st = stealCount;
2733		long c = ctl;
2734		WorkQueue[] ws; WorkQueue w;
2735		if ((ws = workQueues) != null) {
#	Line 2407 \| Line 2770 \| public class ForkJoinPool extends Abstra
2770		}
2771
2772		/**
2773	<	* Initiates an orderly shutdown in which previously submitted
2774	<	* tasks are executed, but no new tasks will be accepted.
2775	<	* Invocation has no additional effect if already shut down.
2776	<	* Tasks that are in the process of being submitted concurrently
2777	<	* during the course of this method may or may not be rejected.
2773	>	* Possibly initiates an orderly shutdown in which previously
2774	>	* submitted tasks are executed, but no new tasks will be
2775	>	* accepted. Invocation has no effect on execution state if this
2776	>	* is the {@link #commonPool}, and no additional effect if
2777	>	* already shut down. Tasks that are in the process of being
2778	>	* submitted concurrently during the course of this method may or
2779	>	* may not be rejected.
2780		*
2781		* @throws SecurityException if a security manager exists and
2782		* the caller is not permitted to modify threads
#	Line 2420 \| Line 2785 \| public class ForkJoinPool extends Abstra
2785		*/
2786		public void shutdown() {
2787		checkPermission();
2788	<	tryTerminate(false, true);
2788	>	if (this != commonPool)
2789	>	tryTerminate(false, true);
2790		}
2791
2792		/**
2793	<	* Attempts to cancel and/or stop all tasks, and reject all
2794	<	* subsequently submitted tasks. Tasks that are in the process of
2795	<	* being submitted or executed concurrently during the course of
2796	<	* this method may or may not be rejected. This method cancels
2797	<	* both existing and unexecuted tasks, in order to permit
2798	<	* termination in the presence of task dependencies. So the method
2799	<	* always returns an empty list (unlike the case for some other
2800	<	* Executors).
2793	>	* Possibly attempts to cancel and/or stop all tasks, and reject
2794	>	* all subsequently submitted tasks. Invocation has no effect on
2795	>	* execution state if this is the {@link #commonPool}, and no
2796	>	* additional effect if already shut down. Otherwise, tasks that
2797	>	* are in the process of being submitted or executed concurrently
2798	>	* during the course of this method may or may not be
2799	>	* rejected. This method cancels both existing and unexecuted
2800	>	* tasks, in order to permit termination in the presence of task
2801	>	* dependencies. So the method always returns an empty list
2802	>	* (unlike the case for some other Executors).
2803		*
2804		* @return an empty list
2805		* @throws SecurityException if a security manager exists and
#	Line 2441 \| Line 2809 \| public class ForkJoinPool extends Abstra
2809		*/
2810		public List<Runnable> shutdownNow() {
2811		checkPermission();
2812	<	tryTerminate(true, true);
2812	>	if (this != commonPool)
2813	>	tryTerminate(true, true);
2814		return Collections.emptyList();
2815		}
2816
#	Line 2498 \| Line 2867 \| public class ForkJoinPool extends Abstra
2867		public boolean awaitTermination(long timeout, TimeUnit unit)
2868		throws InterruptedException {
2869		long nanos = unit.toNanos(timeout);
2870	<	final Mutex lock = this.lock;
2871	<	lock.lock();
2872	<	try {
2873	<	for (;;) {
2874	<	if (isTerminated())
2875	<	return true;
2876	<	if (nanos <= 0)
2877	<	return false;
2878	<	nanos = termination.awaitNanos(nanos);
2870	>	if (isTerminated())
2871	>	return true;
2872	>	long startTime = System.nanoTime();
2873	>	boolean terminated = false;
2874	>	synchronized(this) {
2875	>	for (long waitTime = nanos, millis = 0L;;) {
2876	>	if (terminated = isTerminated() \|\|
2877	>	waitTime <= 0L \|\|
2878	>	(millis = unit.toMillis(waitTime)) <= 0L)
2879	>	break;
2880	>	wait(millis);
2881	>	waitTime = nanos - (System.nanoTime() - startTime);
2882		}
2511	–	} finally {
2512	–	lock.unlock();
2883		}
2884	+	return terminated;
2885		}
2886
2887		/**
#	Line 2612 \| Line 2983 \| public class ForkJoinPool extends Abstra
2983		ForkJoinPool p = ((t instanceof ForkJoinWorkerThread) ?
2984		((ForkJoinWorkerThread)t).pool : null);
2985		while (!blocker.isReleasable()) {
2986	<	if (p == null \|\| p.tryCompensate()) {
2986	>	if (p == null \|\| p.tryCompensate(null, blocker)) {
2987		try {
2988		do {} while (!blocker.isReleasable() && !blocker.block());
2989		} finally {
#	Line 2629 \| Line 3000 \| public class ForkJoinPool extends Abstra
3000		// implement RunnableFuture.
3001
3002		protected <T> RunnableFuture<T> newTaskFor(Runnable runnable, T value) {
3003	<	return (RunnableFuture<T>) ForkJoinTask.adapt(runnable, value);
3003	>	return new ForkJoinTask.AdaptedRunnable<T>(runnable, value);
3004		}
3005
3006		protected <T> RunnableFuture<T> newTaskFor(Callable<T> callable) {
3007	<	return (RunnableFuture<T>) ForkJoinTask.adapt(callable);
3007	>	return new ForkJoinTask.AdaptedCallable<T>(callable);
3008		}
3009
3010		// Unsafe mechanics
3011		private static final sun.misc.Unsafe U;
3012		private static final long CTL;
3013		private static final long PARKBLOCKER;
3014	+	private static final int ABASE;
3015	+	private static final int ASHIFT;
3016	+	private static final long NEXTWORKERNUMBER;
3017	+	private static final long STEALCOUNT;
3018	+	private static final long MAINLOCK;
3019
3020		static {
3021		poolNumberGenerator = new AtomicInteger();
3022	+	nextSubmitterSeed = new AtomicInteger(0x55555555);
3023		modifyThreadPermission = new RuntimePermission("modifyThread");
3024		defaultForkJoinWorkerThreadFactory =
3025		new DefaultForkJoinWorkerThreadFactory();
3026		submitters = new ThreadSubmitter();
3027	+	int s;
3028		try {
3029		U = getUnsafe();
3030		Class<?> k = ForkJoinPool.class;
3031	+	Class<?> ak = ForkJoinTask[].class;
3032		CTL = U.objectFieldOffset
3033		(k.getDeclaredField("ctl"));
3034	+	NEXTWORKERNUMBER = U.objectFieldOffset
3035	+	(k.getDeclaredField("nextWorkerNumber"));
3036	+	STEALCOUNT = U.objectFieldOffset
3037	+	(k.getDeclaredField("stealCount"));
3038	+	MAINLOCK = U.objectFieldOffset
3039	+	(k.getDeclaredField("mainLock"));
3040		Class<?> tk = Thread.class;
3041		PARKBLOCKER = U.objectFieldOffset
3042		(tk.getDeclaredField("parkBlocker"));
3043	+	ABASE = U.arrayBaseOffset(ak);
3044	+	s = U.arrayIndexScale(ak);
3045	+	ASHIFT = 31 - Integer.numberOfLeadingZeros(s);
3046	+	} catch (Exception e) {
3047	+	throw new Error(e);
3048	+	}
3049	+	if ((s & (s-1)) != 0)
3050	+	throw new Error("data type scale not a power of two");
3051	+	try { // Establish common pool
3052	+	String pp = System.getProperty(propPrefix + "parallelism");
3053	+	String fp = System.getProperty(propPrefix + "threadFactory");
3054	+	String up = System.getProperty(propPrefix + "exceptionHandler");
3055	+	ForkJoinWorkerThreadFactory fac = (fp == null) ?
3056	+	defaultForkJoinWorkerThreadFactory :
3057	+	((ForkJoinWorkerThreadFactory)ClassLoader.
3058	+	getSystemClassLoader().loadClass(fp).newInstance());
3059	+	Thread.UncaughtExceptionHandler ueh = (up == null)? null :
3060	+	((Thread.UncaughtExceptionHandler)ClassLoader.
3061	+	getSystemClassLoader().loadClass(up).newInstance());
3062	+	int par;
3063	+	if ((pp == null \|\| (par = Integer.parseInt(pp)) <= 0))
3064	+	par = Runtime.getRuntime().availableProcessors();
3065	+	if (par > MAX_CAP)
3066	+	par = MAX_CAP;
3067	+	commonPoolParallelism = par;
3068	+	int n = par - 1; // precompute submit mask
3069	+	n \|= n >>> 1; n \|= n >>> 2; n \|= n >>> 4;
3070	+	n \|= n >>> 8; n \|= n >>> 16;
3071	+	int mask = ((n + 1) << 1) - 1;
3072	+	commonPool = new ForkJoinPool(par, mask, fac, ueh);
3073		} catch (Exception e) {
3074		throw new Error(e);
3075		}

Diff Legend

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

Comparing jsr166/src/jsr166y/ForkJoinPool.java (file contents): Revision 1.122 by jsr166, Tue Jan 31 01:51:13 2012 UTC vs. Revision 1.136 by dl, Mon Oct 29 17:23:34 2012 UTC

Diff Legend

Comparing jsr166/src/jsr166y/ForkJoinPool.java (file contents):
Revision 1.122 by jsr166, Tue Jan 31 01:51:13 2012 UTC vs.
Revision 1.136 by dl, Mon Oct 29 17:23:34 2012 UTC