ViewVC Help

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

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

Comparing jsr166/src/jsr166y/ForkJoinPool.java (file contents):
Revision 1.119 by dl, Tue Jan 31 00:44:13 2012 UTC vs.
Revision 1.132 by jsr166, Fri Oct 12 16:46:37 2012 UTC

#	Line 5 | Line 5
5		*/
6
7		package jsr166y;
8	–
8		import java.util.ArrayList;
9		import java.util.Arrays;
10		import java.util.Collection;
#	Line 177 | Line 176 | public class ForkJoinPool extends Abstra
176		* If an attempted steal fails, a thief always chooses a different
177		* random victim target to try next. So, in order for one thief to
178		* progress, it suffices for any in-progress poll or new push on
179	<	* any empty queue to complete.
179	>	* any empty queue to complete. (This is why we normally use
180	>	* method pollAt and its variants that try once at the apparent
181	>	* base index, else consider alternative actions, rather than
182	>	* method poll.)
183		*
184		* This approach also enables support of a user mode in which local
185		* task processing is in FIFO, not LIFO order, simply by using
#	Line 207 | Line 209 | public class ForkJoinPool extends Abstra
209		* lock (mainly to protect in the case of resizing) but we use
210		* only a simple spinlock (using bits in field runState), because
211		* submitters encountering a busy queue move on to try or create
212	<	* other queues, so never block.
212	>	* other queues -- they block only when creating and registering
213	>	* new queues.
214		*
215		* Management
216		* ==========
#	Line 233 | Line 236 | public class ForkJoinPool extends Abstra
236		* deregister WorkQueues, as well as to enable shutdown. It is
237		* only modified under a lock (normally briefly held, but
238		* occasionally protecting allocations and resizings) but even
239	<	* 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.
239	>	* when locked remains available to check consistency.
240		*
241		* Recording WorkQueues.  WorkQueues are recorded in the
242		* "workQueues" array that is created upon pool construction and
#	Line 248 | Line 248 | public class ForkJoinPool extends Abstra
248		* readers must tolerate null slots. Shared (submission) queues
249		* are at even indices, worker queues at odd indices. Grouping
250		* 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.
251	>	* scanning.
252		*
253		* All worker thread creation is on-demand, triggered by task
254		* submissions, replacement of terminated workers, and/or
#	Line 385 | Line 380 | public class ForkJoinPool extends Abstra
380		* (http://portal.acm.org/citation.cfm?id=155354). It differs in
381		* that: (1) We only maintain dependency links across workers upon
382		* steals, rather than use per-task bookkeeping.  This sometimes
383	<	* requires a linear scan of workQueues array to locate stealers, but
384	<	* often doesn't because stealers leave hints (that may become
383	>	* requires a linear scan of workQueues array to locate stealers,
384	>	* but often doesn't because stealers leave hints (that may become
385		* stale/wrong) of where to locate them.  A stealHint is only a
386		* hint because a worker might have had multiple steals and the
387		* hint records only one of them (usually the most current).
#	Line 397 | Line 392 | public class ForkJoinPool extends Abstra
392		* which means that we miss links in the chain during long-lived
393		* tasks, GC stalls etc (which is OK since blocking in such cases
394		* is usually a good idea).  (4) We bound the number of attempts
395	<	* to find work (see MAX_HELP_DEPTH) and fall back to suspending
396	<	* the worker and if necessary replacing it with another.
395	>	* to find work (see MAX_HELP) and fall back to suspending the
396	>	* worker and if necessary replacing it with another.
397		*
398		* It is impossible to keep exactly the target parallelism number
399		* of threads running at any given time.  Determining the
400		* existence of conservatively safe helping targets, the
401		* availability of already-created spares, and the apparent need
402		* to create new spares are all racy, so we rely on multiple
403	<	* retries of each.  Currently, in keeping with on-demand
404	<	* signalling policy, we compensate only if blocking would leave
405	<	* less than one active (non-waiting, non-blocked) worker.
406	<	* Additionally, to avoid some false alarms due to GC, lagging
407	<	* counters, system activity, etc, compensated blocking for joins
408	<	* is only attempted after rechecks stabilize in
409	<	* ForkJoinTask.awaitJoin. (Retries are interspersed with
410	<	* Thread.yield, for good citizenship.)
403	>	* retries of each.  Compensation in the apparent absence of
404	>	* helping opportunities is challenging to control on JVMs, where
405	>	* GC and other activities can stall progress of tasks that in
406	>	* turn stall out many other dependent tasks, without us being
407	>	* able to determine whether they will ever require compensation.
408	>	* Even though work-stealing otherwise encounters little
409	>	* degradation in the presence of more threads than cores,
410	>	* aggressively adding new threads in such cases entails risk of
411	>	* unwanted positive feedback control loops in which more threads
412	>	* cause more dependent stalls (as well as delayed progress of
413	>	* unblocked threads to the point that we know they are available)
414	>	* leading to more situations requiring more threads, and so
415	>	* on. This aspect of control can be seen as an (analytically
416	>	* intractable) game with an opponent that may choose the worst
417	>	* (for us) active thread to stall at any time.  We take several
418	>	* precautions to bound losses (and thus bound gains), mainly in
419	>	* methods tryCompensate and awaitJoin: (1) We only try
420	>	* compensation after attempting enough helping steps (measured
421	>	* via counting and timing) that we have already consumed the
422	>	* estimated cost of creating and activating a new thread.  (2) We
423	>	* allow up to 50% of threads to be blocked before initially
424	>	* adding any others, and unless completely saturated, check that
425	>	* some work is available for a new worker before adding. Also, we
426	>	* create up to only 50% more threads until entering a mode that
427	>	* only adds a thread if all others are possibly blocked.  All
428	>	* together, this means that we might be half as fast to react,
429	>	* and create half as many threads as possible in the ideal case,
430	>	* but present vastly fewer anomalies in all other cases compared
431	>	* to both more aggressive and more conservative alternatives.
432		*
433		* Style notes: There is a lot of representation-level coupling
434		* among classes ForkJoinPool, ForkJoinWorkerThread, and
#	Line 449 | Line 465 | public class ForkJoinPool extends Abstra
465		// Static utilities
466
467		/**
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	–	/**
468		* If there is a security manager, makes sure caller has
469		* permission to modify threads.
470		*/
#	Line 592 | Line 595 | public class ForkJoinPool extends Abstra
595		static final class WorkQueue {
596		/**
597		* Capacity of work-stealing queue array upon initialization.
598	<	* Must be a power of two; at least 4, but set larger to
599	<	* reduce cacheline sharing among queues.
598	>	* Must be a power of two; at least 4, but should be larger to
599	>	* reduce or eliminate cacheline sharing among queues.
600	>	* Currently, it is much larger, as a partial workaround for
601	>	* the fact that JVMs often place arrays in locations that
602	>	* share GC bookkeeping (especially cardmarks) such that
603	>	* per-write accesses encounter serious memory contention.
604		*/
605	<	static final int INITIAL_QUEUE_CAPACITY = 1 << 8;
605	>	static final int INITIAL_QUEUE_CAPACITY = 1 << 13;
606
607		/**
608		* Maximum size for queue arrays. Must be a power of two less
#	Line 619 | Line 626 | public class ForkJoinPool extends Abstra
626		volatile int base;         // index of next slot for poll
627		int top;                   // index of next slot for push
628		ForkJoinTask<?>[] array;   // the elements (initially unallocated)
629	+	final ForkJoinPool pool;   // the containing pool (may be null)
630		final ForkJoinWorkerThread owner; // owning thread or null if shared
631		volatile Thread parker;    // == owner during call to park; else null
632	<	ForkJoinTask<?> currentJoin;  // task being joined in awaitJoin
632	>	volatile ForkJoinTask<?> currentJoin;  // task being joined in awaitJoin
633		ForkJoinTask<?> currentSteal; // current non-local task being executed
634		// Heuristic padding to ameliorate unfortunate memory placements
635	<	Object p00, p01, p02, p03, p04, p05, p06, p07, p08, p09, p0a;
635	>	Object p00, p01, p02, p03, p04, p05, p06, p07;
636	>	Object p08, p09, p0a, p0b, p0c, p0d, p0e;
637
638	<	WorkQueue(ForkJoinWorkerThread owner, int mode) {
630	<	this.owner = owner;
638	>	WorkQueue(ForkJoinPool pool, ForkJoinWorkerThread owner, int mode) {
639		this.mode = mode;
640	+	this.pool = pool;
641	+	this.owner = owner;
642		// Place indices in the center of array (that is not yet allocated)
643		base = top = INITIAL_QUEUE_CAPACITY >>> 1;
644		}
645
646		/**
647	<	* Returns number of tasks in the queue.
647	>	* Returns the approximate number of tasks in the queue.
648		*/
649		final int queueSize() {
650	<	int n = base - top; // non-owner callers must read base first
651	<	return (n >= 0) ? 0 : -n;
650	>	int n = base - top;       // non-owner callers must read base first
651	>	return (n >= 0) ? 0 : -n; // ignore transient negative
652	>	}
653	>
654	>	/**
655	>	* Provides a more accurate estimate of whether this queue has
656	>	* any tasks than does queueSize, by checking whether a
657	>	* near-empty queue has at least one unclaimed task.
658	>	*/
659	>	final boolean isEmpty() {
660	>	ForkJoinTask<?>[] a; int m, s;
661	>	int n = base - (s = top);
662	>	return (n >= 0 ||
663	>	(n == -1 &&
664	>	((a = array) == null ||
665	>	(m = a.length - 1) < 0 ||
666	>	U.getObjectVolatile
667	>	(a, ((m & (s - 1)) << ASHIFT) + ABASE) == null)));
668		}
669
670		/**
671		* Pushes a task. Call only by owner in unshared queues.
672		*
673		* @param task the task. Caller must ensure non-null.
648	–	* @param p if non-null, pool to signal if necessary
674		* @throw RejectedExecutionException if array cannot be resized
675		*/
676	<	final void push(ForkJoinTask<?> task, ForkJoinPool p) {
677	<	ForkJoinTask<?>[] a;
676	>	final void push(ForkJoinTask<?> task) {
677	>	ForkJoinTask<?>[] a; ForkJoinPool p;
678		int s = top, m, n;
679		if ((a = array) != null) {    // ignore if queue removed
680		U.putOrderedObject
681		(a, (((m = a.length - 1) & s) << ASHIFT) + ABASE, task);
682		if ((n = (top = s + 1) - base) <= 2) {
683	<	if (p != null)
683	>	if ((p = pool) != null)
684		p.signalWork();
685		}
686		else if (n >= m)
#	Line 691 | Line 716 | public class ForkJoinPool extends Abstra
716		}
717
718		/**
719	<	* Takes next task, if one exists, in FIFO order.
719	>	* Takes next task, if one exists, in LIFO order.  Call only
720	>	* by owner in unshared queues. (We do not have a shared
721	>	* version of this method because it is never needed.)
722		*/
723	<	final ForkJoinTask<?> poll() {
724	<	ForkJoinTask<?>[] a; int b; ForkJoinTask<?> t;
725	<	while ((b = base) - top < 0 && (a = array) != null) {
723	>	final ForkJoinTask<?> pop() {
724	>	ForkJoinTask<?>[] a; ForkJoinTask<?> t; int m;
725	>	if ((a = array) != null && (m = a.length - 1) >= 0) {
726	>	for (int s; (s = top - 1) - base >= 0;) {
727	>	long j = ((m & s) << ASHIFT) + ABASE;
728	>	if ((t = (ForkJoinTask<?>)U.getObject(a, j)) == null)
729	>	break;
730	>	if (U.compareAndSwapObject(a, j, t, null)) {
731	>	top = s;
732	>	return t;
733	>	}
734	>	}
735	>	}
736	>	return null;
737	>	}
738	>
739	>	/**
740	>	* Takes a task in FIFO order if b is base of queue and a task
741	>	* can be claimed without contention. Specialized versions
742	>	* appear in ForkJoinPool methods scan and tryHelpStealer.
743	>	*/
744	>	final ForkJoinTask<?> pollAt(int b) {
745	>	ForkJoinTask<?> t; ForkJoinTask<?>[] a;
746	>	if ((a = array) != null) {
747		int j = (((a.length - 1) & b) << ASHIFT) + ABASE;
748		if ((t = (ForkJoinTask<?>)U.getObjectVolatile(a, j)) != null &&
749		base == b &&
#	Line 708 | Line 756 | public class ForkJoinPool extends Abstra
756		}
757
758		/**
759	<	* 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.)
759	>	* Takes next task, if one exists, in FIFO order.
760		*/
761	<	final ForkJoinTask<?> pop() {
762	<	ForkJoinTask<?> t; int m;
763	<	ForkJoinTask<?>[] a = array;
764	<	if (a != null && (m = a.length - 1) >= 0) {
765	<	for (int s; (s = top - 1) - base >= 0;) {
766	<	int j = ((m & s) << ASHIFT) + ABASE;
767	<	if ((t = (ForkJoinTask<?>)U.getObjectVolatile(a, j)) == null)
768	<	break;
769	<	if (U.compareAndSwapObject(a, j, t, null)) {
724	<	top = s;
761	>	final ForkJoinTask<?> poll() {
762	>	ForkJoinTask<?>[] a; int b; ForkJoinTask<?> t;
763	>	while ((b = base) - top < 0 && (a = array) != null) {
764	>	int j = (((a.length - 1) & b) << ASHIFT) + ABASE;
765	>	t = (ForkJoinTask<?>)U.getObjectVolatile(a, j);
766	>	if (t != null) {
767	>	if (base == b &&
768	>	U.compareAndSwapObject(a, j, t, null)) {
769	>	base = b + 1;
770		return t;
771		}
772		}
773	+	else if (base == b) {
774	+	if (b + 1 == top)
775	+	break;
776	+	Thread.yield(); // wait for lagging update
777	+	}
778		}
779		return null;
780		}
#	Line 749 | Line 799 | public class ForkJoinPool extends Abstra
799		}
800
801		/**
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	–	/**
802		* Pops the given task only if it is at the current top.
803		*/
804		final boolean tryUnpush(ForkJoinTask<?> t) {
#	Line 796 | Line 829 | public class ForkJoinPool extends Abstra
829		}
830
831		/**
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	–	/**
832		* Initializes or doubles the capacity of array. Call either
833		* by owner or with lock held -- it is OK for base, but not
834		* top, to move while resizings are in progress.
#	Line 892 | Line 877 | public class ForkJoinPool extends Abstra
877		* Computes next value for random probes.  Scans don't require
878		* a very high quality generator, but also not a crummy one.
879		* Marsaglia xor-shift is cheap and works well enough.  Note:
880	<	* This is manually inlined in several usages in ForkJoinPool
881	<	* to avoid writes inside busy scan loops.
880	>	* This is manually inlined in its usages in ForkJoinPool to
881	>	* avoid writes inside busy scan loops.
882		*/
883		final int nextSeed() {
884		int r = seed;
#	Line 905 | Line 890 | public class ForkJoinPool extends Abstra
890		// Execution methods
891
892		/**
893	<	* Removes and runs tasks until empty, using local mode
909	<	* ordering.
893	>	* Pops and runs tasks until empty.
894		*/
895	<	final void runLocalTasks() {
896	<	if (base - top < 0) {
897	<	for (ForkJoinTask<?> t; (t = nextLocalTask()) != null; )
895	>	private void popAndExecAll() {
896	>	// A bit faster than repeated pop calls
897	>	ForkJoinTask<?>[] a; int m, s; long j; ForkJoinTask<?> t;
898	>	while ((a = array) != null && (m = a.length - 1) >= 0 &&
899	>	(s = top - 1) - base >= 0 &&
900	>	(t = ((ForkJoinTask<?>)
901	>	U.getObject(a, j = ((m & s) << ASHIFT) + ABASE)))
902	>	!= null) {
903	>	if (U.compareAndSwapObject(a, j, t, null)) {
904	>	top = s;
905		t.doExec();
906	+	}
907	+	}
908	+	}
909	+
910	+	/**
911	+	* Polls and runs tasks until empty.
912	+	*/
913	+	private void pollAndExecAll() {
914	+	for (ForkJoinTask<?> t; (t = poll()) != null;)
915	+	t.doExec();
916	+	}
917	+
918	+	/**
919	+	* If present, removes from queue and executes the given task, or
920	+	* any other cancelled task. Returns (true) immediately on any CAS
921	+	* or consistency check failure so caller can retry.
922	+	*
923	+	* @return 0 if no progress can be made, else positive
924	+	* (this unusual convention simplifies use with tryHelpStealer.)
925	+	*/
926	+	final int tryRemoveAndExec(ForkJoinTask<?> task) {
927	+	int stat = 1;
928	+	boolean removed = false, empty = true;
929	+	ForkJoinTask<?>[] a; int m, s, b, n;
930	+	if ((a = array) != null && (m = a.length - 1) >= 0 &&
931	+	(n = (s = top) - (b = base)) > 0) {
932	+	for (ForkJoinTask<?> t;;) {           // traverse from s to b
933	+	int j = ((--s & m) << ASHIFT) + ABASE;
934	+	t = (ForkJoinTask<?>)U.getObjectVolatile(a, j);
935	+	if (t == null)                    // inconsistent length
936	+	break;
937	+	else if (t == task) {
938	+	if (s + 1 == top) {           // pop
939	+	if (!U.compareAndSwapObject(a, j, task, null))
940	+	break;
941	+	top = s;
942	+	removed = true;
943	+	}
944	+	else if (base == b)           // replace with proxy
945	+	removed = U.compareAndSwapObject(a, j, task,
946	+	new EmptyTask());
947	+	break;
948	+	}
949	+	else if (t.status >= 0)
950	+	empty = false;
951	+	else if (s + 1 == top) {          // pop and throw away
952	+	if (U.compareAndSwapObject(a, j, t, null))
953	+	top = s;
954	+	break;
955	+	}
956	+	if (--n == 0) {
957	+	if (!empty && base == b)
958	+	stat = 0;
959	+	break;
960	+	}
961	+	}
962		}
963	+	if (removed)
964	+	task.doExec();
965	+	return stat;
966		}
967
968		/**
969		* Executes a top-level task and any local tasks remaining
970		* after execution.
921	–	*
922	–	* @return true unless terminating
971		*/
972	<	final boolean runTask(ForkJoinTask<?> t) {
925	<	boolean alive = true;
972	>	final void runTask(ForkJoinTask<?> t) {
973		if (t != null) {
974		currentSteal = t;
975		t.doExec();
976	<	runLocalTasks();
976	>	if (top != base) {       // process remaining local tasks
977	>	if (mode == 0)
978	>	popAndExecAll();
979	>	else
980	>	pollAndExecAll();
981	>	}
982		++nsteals;
983		currentSteal = null;
984		}
933	–	else if (runState < 0)            // terminating
934	–	alive = false;
935	–	return alive;
985		}
986
987		/**
#	Line 1001 | Line 1050 | public class ForkJoinPool extends Abstra
1050
1051		/**
1052		* Per-thread records for threads that submit to pools. Currently
1053	<	* holds only psuedo-random seed / index that is used to choose
1053	>	* holds only pseudo-random seed / index that is used to choose
1054		* submission queues in method doSubmit. In the future, this may
1055		* also incorporate a means to implement different task rejection
1056		* and resubmission policies.
1057	+	*
1058	+	* Seeds for submitters and workers/workQueues work in basically
1059	+	* the same way but are initialized and updated using slightly
1060	+	* different mechanics. Both are initialized using the same
1061	+	* approach as in class ThreadLocal, where successive values are
1062	+	* unlikely to collide with previous values. This is done during
1063	+	* registration for workers, but requires a separate AtomicInteger
1064	+	* for submitters. Seeds are then randomly modified upon
1065	+	* collisions using xorshifts, which requires a non-zero seed.
1066		*/
1067		static final class Submitter {
1068		int seed;
1069	<	Submitter() { seed = hashId(Thread.currentThread().getId()); }
1069	>	Submitter() {
1070	>	int s = nextSubmitterSeed.getAndAdd(SEED_INCREMENT);
1071	>	seed = (s == 0) ? 1 : s; // ensure non-zero
1072	>	}
1073		}
1074
1075		/** ThreadLocal class for Submitters */
#	Line 1031 | Line 1092 | public class ForkJoinPool extends Abstra
1092		private static final AtomicInteger poolNumberGenerator;
1093
1094		/**
1095	+	* Generator for initial hashes/seeds for submitters. Accessed by
1096	+	* Submitter class constructor.
1097	+	*/
1098	+	static final AtomicInteger nextSubmitterSeed;
1099	+
1100	+	/**
1101		* Permission required for callers of methods that may start or
1102		* kill threads.
1103		*/
1104		private static final RuntimePermission modifyThreadPermission;
1105
1106		/**
1107	<	* Per-thread submission bookeeping. Shared across all pools
1107	>	* Per-thread submission bookkeeping. Shared across all pools
1108		* to reduce ThreadLocal pollution and because random motion
1109		* to avoid contention in one pool is likely to hold for others.
1110		*/
#	Line 1063 | Line 1130 | public class ForkJoinPool extends Abstra
1130		private static final long SHRINK_TIMEOUT = SHRINK_RATE - (SHRINK_RATE / 10);
1131
1132		/**
1133	<	* The maximum stolen->joining link depth allowed in tryHelpStealer.
1134	<	* Depths for legitimate chains are unbounded, but we use a fixed
1135	<	* constant to avoid (otherwise unchecked) cycles and to bound
1136	<	* staleness of traversal parameters at the expense of sometimes
1137	<	* blocking when we could be helping.
1133	>	* The maximum stolen->joining link depth allowed in method
1134	>	* tryHelpStealer.  Must be a power of two. This value also
1135	>	* controls the maximum number of times to try to help join a task
1136	>	* without any apparent progress or change in pool state before
1137	>	* giving up and blocking (see awaitJoin).  Depths for legitimate
1138	>	* chains are unbounded, but we use a fixed constant to avoid
1139	>	* (otherwise unchecked) cycles and to bound staleness of
1140	>	* traversal parameters at the expense of sometimes blocking when
1141	>	* we could be helping.
1142	>	*/
1143	>	private static final int MAX_HELP = 64;
1144	>
1145	>	/**
1146	>	* Secondary time-based bound (in nanosecs) for helping attempts
1147	>	* before trying compensated blocking in awaitJoin. Used in
1148	>	* conjunction with MAX_HELP to reduce variance due to different
1149	>	* polling rates associated with different helping options. The
1150	>	* value should roughly approximate the time required to create
1151	>	* and/or activate a worker thread.
1152	>	*/
1153	>	private static final long COMPENSATION_DELAY = 1L << 18; // ~0.25 millisec
1154	>
1155	>	/**
1156	>	* Increment for seed generators. See class ThreadLocal for
1157	>	* explanation.
1158		*/
1159	<	private static final int MAX_HELP_DEPTH = 16;
1159	>	private static final int SEED_INCREMENT = 0x61c88647;
1160
1161		/**
1162		* Bits and masks for control variables
#	Line 1101 | Line 1188 | public class ForkJoinPool extends Abstra
1188		*
1189		* Field runState is an int packed with:
1190		* SHUTDOWN: true if shutdown is enabled (1 bit)
1191	<	* SEQ:  a sequence number updated upon (de)registering workers (15 bits)
1192	<	* MASK: mask (power of 2 - 1) covering all registered poolIndexes (16 bits)
1191	>	* SEQ:  a sequence number updated upon (de)registering workers (30 bits)
1192	>	* INIT: set true after workQueues array construction (1 bit)
1193		*
1194	<	* The combination of mask and sequence number enables simple
1195	<	* consistency checks: Staleness of read-only operations on the
1196	<	* 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.
1194	>	* The sequence number enables simple consistency checks:
1195	>	* Staleness of read-only operations on the workQueues array can
1196	>	* be checked by comparing runState before vs after the reads.
1197		*/
1198
1199		// bit positions/shifts for fields
#	Line 1119 | Line 1203 | public class ForkJoinPool extends Abstra
1203		private static final int  EC_SHIFT   = 16;
1204
1205		// bounds
1122	–	private static final int  POOL_MAX   = 0x7fff;  // max #workers - 1
1206		private static final int  SMASK      = 0xffff;  // short bits
1207	+	private static final int  MAX_CAP    = 0x7fff;  // max #workers - 1
1208		private static final int  SQMASK     = 0xfffe;  // even short bits
1209		private static final int  SHORT_SIGN = 1 << 15;
1210		private static final int  INT_SIGN   = 1 << 31;
#	Line 1148 | Line 1232 | public class ForkJoinPool extends Abstra
1232
1233		// runState bits
1234		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;
1235
1236		// access mode for WorkQueue
1237		static final int LIFO_QUEUE          =  0;
#	Line 1168 | Line 1250 | public class ForkJoinPool extends Abstra
1250		volatile long ctl;                         // main pool control
1251		final int parallelism;                     // parallelism level
1252		final int localMode;                       // per-worker scheduling mode
1253	<	int growHints;                             // for expanding indices/ranges
1254	<	volatile int runState;                     // shutdown status, seq, and mask
1253	>	final int submitMask;                      // submit queue index bound
1254	>	int nextSeed;                              // for initializing worker seeds
1255	>	volatile int runState;                     // shutdown status and seq
1256		WorkQueue[] workQueues;                    // main registry
1257		final Mutex lock;                          // for registration
1258		final Condition termination;               // for awaitTermination
#	Line 1179 | Line 1262 | public class ForkJoinPool extends Abstra
1262		final AtomicInteger nextWorkerNumber;      // to create worker name string
1263		final String workerNamePrefix;             // to create worker name string
1264
1265	<	//  Creating, registering, deregistering and running workers
1265	>	//  Creating, registering, and deregistering workers
1266
1267		/**
1268		* Tries to create and start a worker
#	Line 1210 | Line 1293 | public class ForkJoinPool extends Abstra
1293		}
1294
1295		/**
1296	<	* Callback from ForkJoinWorkerThread constructor to establish and
1297	<	* record its WorkQueue.
1296	>	* Callback from ForkJoinWorkerThread constructor to establish its
1297	>	* poolIndex and record its WorkQueue. To avoid scanning bias due
1298	>	* to packing entries in front of the workQueues array, we treat
1299	>	* the array as a simple power-of-two hash table using per-thread
1300	>	* seed as hash, expanding as needed.
1301		*
1302	<	* @param wt the worker thread
1302	>	* @param w the worker's queue
1303		*/
1304	<	final void registerWorker(ForkJoinWorkerThread wt) {
1305	<	WorkQueue w = wt.workQueue;
1304	>
1305	>	final void registerWorker(WorkQueue w) {
1306		Mutex lock = this.lock;
1307		lock.lock();
1308		try {
1223	–	int g = growHints, k = g & SMASK;
1309		WorkQueue[] ws = workQueues;
1310	<	if (ws != null) {                       // ignore on shutdown
1311	<	int n = ws.length;
1312	<	if ((k & 1) == 0 || k >= n || ws[k] != null) {
1313	<	for (k = 1; k < n && ws[k] != null; k += 2)
1314	<	;                           // workers are at odd indices
1315	<	if (k >= n)                     // resize
1316	<	workQueues = ws = Arrays.copyOf(ws, n << 1);
1317	<	}
1318	<	w.eventCount = w.poolIndex = k;     // establish before recording
1319	<	ws[k] = w;
1320	<	growHints = (g & ~SMASK) | ((k + 2) & SMASK);
1321	<	int rs = runState;
1322	<	int m = rs & SMASK;                 // recalculate runState mask
1323	<	if (k > m)
1324	<	m = (m << 1) + 1;
1325	<	runState = (rs & SHUTDOWN) | ((rs + RS_SEQ) & RS_SEQ_MASK) | m;
1310	>	if (w != null && ws != null) {          // skip on shutdown/failure
1311	>	int rs, n = ws.length, m = n - 1;
1312	>	int s = nextSeed += SEED_INCREMENT; // rarely-colliding sequence
1313	>	w.seed = (s == 0) ? 1 : s;          // ensure non-zero seed
1314	>	int r = (s << 1) | 1;               // use odd-numbered indices
1315	>	if (ws[r &= m] != null) {           // collision
1316	>	int probes = 0;                 // step by approx half size
1317	>	int step = (n <= 4) ? 2 : ((n >>> 1) & SQMASK) + 2;
1318	>	while (ws[r = (r + step) & m] != null) {
1319	>	if (++probes >= n) {
1320	>	workQueues = ws = Arrays.copyOf(ws, n <<= 1);
1321	>	m = n - 1;
1322	>	probes = 0;
1323	>	}
1324	>	}
1325	>	}
1326	>	w.eventCount = w.poolIndex = r;     // establish before recording
1327	>	ws[r] = w;                          // also update seq
1328	>	runState = ((rs = runState) & SHUTDOWN) | ((rs + 2) & ~SHUTDOWN);
1329		}
1330		} finally {
1331		lock.unlock();
#	Line 1254 | Line 1342 | public class ForkJoinPool extends Abstra
1342		* @param ex the exception causing failure, or null if none
1343		*/
1344		final void deregisterWorker(ForkJoinWorkerThread wt, Throwable ex) {
1345	+	Mutex lock = this.lock;
1346		WorkQueue w = null;
1347		if (wt != null && (w = wt.workQueue) != null) {
1348		w.runState = -1;                // ensure runState is set
1349		stealCount.getAndAdd(w.totalSteals + w.nsteals);
1350		int idx = w.poolIndex;
1262	–	Mutex lock = this.lock;
1351		lock.lock();
1352		try {                           // remove record from array
1353		WorkQueue[] ws = workQueues;
1354	<	if (ws != null && idx >= 0 && idx < ws.length && ws[idx] == w) {
1354	>	if (ws != null && idx >= 0 && idx < ws.length && ws[idx] == w)
1355		ws[idx] = null;
1268	–	growHints = (growHints & ~SMASK) | idx;
1269	–	}
1356		} finally {
1357		lock.unlock();
1358		}
#	Line 1290 | Line 1376 | public class ForkJoinPool extends Abstra
1376		U.throwException(ex);
1377		}
1378
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	–	}
1379
1380		// Submissions
1381
1382		/**
1383		* Unless shutting down, adds the given task to a submission queue
1384		* at submitter's current queue index (modulo submission
1385	<	* range). If no queue exists at the index, one is created unless
1386	<	* pool lock is busy.  If the queue and/or lock are busy, another
1387	<	* index is randomly chosen. The mask in growHints controls the
1388	<	* effective index range of queues considered. The mask is
1389	<	* expanded, up to the current workerQueue mask, upon any detected
1390	<	* contention but otherwise remains small to avoid needlessly
1316	<	* creating queues when there is no contention.
1385	>	* range). If no queue exists at the index, one is created.  If
1386	>	* the queue is busy, another index is randomly chosen. The
1387	>	* submitMask bounds the effective number of queues to the
1388	>	* (nearest power of two for) parallelism level.
1389	>	*
1390	>	* @param task the task. Caller must ensure non-null.
1391		*/
1392		private void doSubmit(ForkJoinTask<?> task) {
1319	–	if (task == null)
1320	–	throw new NullPointerException();
1393		Submitter s = submitters.get();
1394	<	for (int r = s.seed, m = growHints >>> 16;;) {
1395	<	WorkQueue[] ws; WorkQueue q; Mutex lk;
1394	>	for (int r = s.seed, m = submitMask;;) {
1395	>	WorkQueue[] ws; WorkQueue q;
1396		int k = r & m & SQMASK;          // use only even indices
1397		if (runState < 0 || (ws = workQueues) == null || ws.length <= k)
1398		throw new RejectedExecutionException(); // shutting down
1399	<	if ((q = ws[k]) == null && (lk = lock).tryAcquire(0)) {
1400	<	try {                        // try to create new queue
1401	<	if (ws == workQueues && (q = ws[k]) == null) {
1402	<	int rs;              // update runState seq
1403	<	ws[k] = q = new WorkQueue(null, SHARED_QUEUE);
1404	<	runState = (((rs = runState) & SHUTDOWN) |
1405	<	((rs + RS_SEQ) & ~SHUTDOWN));
1399	>	else if ((q = ws[k]) == null) {  // create new queue
1400	>	WorkQueue nq = new WorkQueue(this, null, SHARED_QUEUE);
1401	>	Mutex lock = this.lock;      // construct outside lock
1402	>	lock.lock();
1403	>	try {                        // recheck under lock
1404	>	int rs = runState;       // to update seq
1405	>	if (ws == workQueues && ws[k] == null) {
1406	>	ws[k] = nq;
1407	>	runState = ((rs & SHUTDOWN) | ((rs + 2) & ~SHUTDOWN));
1408		}
1409		} finally {
1410	<	lk.unlock();
1410	>	lock.unlock();
1411		}
1412		}
1413	<	if (q != null) {
1414	<	if (q.trySharedPush(task)) {
1415	<	signalWork();
1416	<	return;
1417	<	}
1418	<	else if (m < parallelism - 1 && m < (runState & SMASK)) {
1419	<	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
1413	>	else if (q.trySharedPush(task)) {
1414	>	signalWork();
1415	>	return;
1416	>	}
1417	>	else if (m > 1) {                // move to a different index
1418	>	r ^= r << 13;                // same xorshift as WorkQueues
1419	>	r ^= r >>> 17;
1420		s.seed = r ^= r << 5;
1421		}
1422	+	else
1423	+	Thread.yield();              // yield if no alternatives
1424		}
1425		}
1426
#	Line 1405 | Line 1469 | public class ForkJoinPool extends Abstra
1469		}
1470		}
1471
1472	+	// Scanning for tasks
1473	+
1474		/**
1475	<	* 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
1475	>	* Top-level runloop for workers, called by ForkJoinWorkerThread.run.
1476		*/
1477	<	final boolean tryCompensate() {
1478	<	WorkQueue w; Thread p;
1479	<	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;
1477	>	final void runWorker(WorkQueue w) {
1478	>	w.growArray(false);         // initialize queue array in this thread
1479	>	do { w.runTask(scan(w)); } while (w.runState >= 0);
1480		}
1481
1449	–	// Scanning for tasks
1450	–
1482		/**
1483		* Scans for and, if found, returns one task, else possibly
1484		* inactivates the worker. This method operates on single reads of
1485	<	* volatile state and is designed to be re-invoked continuously in
1486	<	* part because it returns upon detecting inconsistencies,
1485	>	* volatile state and is designed to be re-invoked continuously,
1486	>	* in part because it returns upon detecting inconsistencies,
1487		* contention, or state changes that indicate possible success on
1488		* re-invocation.
1489		*
1490	<	* The scan searches for tasks across queues, randomly selecting
1491	<	* the first #queues probes, favoring steals over submissions
1492	<	* (by exploiting even/odd indexing), and then performing a
1493	<	* circular sweep of all queues.  The scan terminates upon either
1494	<	* finding a non-empty queue, or completing a full sweep. If the
1495	<	* worker is not inactivated, it takes and returns a task from
1496	<	* this queue.  On failure to find a task, we take one of the
1497	<	* following actions, after which the caller will retry calling
1467	<	* this method unless terminated.
1490	>	* The scan searches for tasks across a random permutation of
1491	>	* queues (starting at a random index and stepping by a random
1492	>	* relative prime, checking each at least once).  The scan
1493	>	* terminates upon either finding a non-empty queue, or completing
1494	>	* the sweep. If the worker is not inactivated, it takes and
1495	>	* returns a task from this queue.  On failure to find a task, we
1496	>	* take one of the following actions, after which the caller will
1497	>	* retry calling this method unless terminated.
1498		*
1499		* * If pool is terminating, terminate the worker.
1500		*
#	Line 1475 | Line 1505 | public class ForkJoinPool extends Abstra
1505		* another worker, but with same net effect. Releasing in other
1506		* cases as well ensures that we have enough workers running.
1507		*
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	–	*
1508		* * If not already enqueued, try to inactivate and enqueue the
1509	<	* worker on wait queue.
1509	>	* worker on wait queue. Or, if inactivating has caused the pool
1510	>	* to be quiescent, relay to idleAwaitWork to check for
1511	>	* termination and possibly shrink pool.
1512	>	*
1513	>	* * If already inactive, and the caller has run a task since the
1514	>	* last empty scan, return (to allow rescan) unless others are
1515	>	* also inactivated.  Field WorkQueue.rescans counts down on each
1516	>	* scan to ensure eventual inactivation and blocking.
1517		*
1518	<	* * If already enqueued and none of the above apply, either park
1519	<	* 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.
1518	>	* * If already enqueued and none of the above apply, park
1519	>	* awaiting signal,
1520		*
1521		* @param w the worker (via its WorkQueue)
1522		* @return a task or null of none found
1523		*/
1524		private final ForkJoinTask<?> scan(WorkQueue w) {
1525	<	boolean swept = false;               // true after full empty scan
1526	<	WorkQueue[] ws;                      // volatile read order matters
1527	<	int r = w.seed, ec = w.eventCount;   // ec is negative if inactive
1528	<	int rs = runState, m = rs & SMASK;
1529	<	if ((ws = workQueues) != null && ws.length > m) { // consistency check
1530	<	for (int k = 0, j = -1 - m; ; ++j) {
1531	<	WorkQueue q; int b;
1532	<	if (j < 0) {                 // random probes while j negative
1533	<	r ^= r << 13; r ^= r >>> 17; k = (r ^= r << 5) | (j & 1);
1534	<	}                            // worker (not submit) for odd j
1535	<	else                         // cyclic scan when j >= 0
1536	<	k += 7;                  // step 7 reduces array packing bias
1537	<	if ((q = ws[k & m]) != null && (b = q.base) - q.top < 0) {
1538	<	ForkJoinTask<?> t = (ec >= 0) ? q.pollAt(b) : null;
1539	<	w.seed = r;              // save seed for next scan
1540	<	if (t != null)
1525	>	WorkQueue[] ws;                       // first update random seed
1526	>	int r = w.seed; r ^= r << 13; r ^= r >>> 17; w.seed = r ^= r << 5;
1527	>	int rs = runState, m;                 // volatile read order matters
1528	>	if ((ws = workQueues) != null && (m = ws.length - 1) > 0) {
1529	>	int ec = w.eventCount;            // ec is negative if inactive
1530	>	int step = (r >>> 16) | 1;        // relative prime
1531	>	for (int j = (m + 1) << 2; ; r += step) {
1532	>	WorkQueue q; ForkJoinTask<?> t; ForkJoinTask<?>[] a; int b;
1533	>	if ((q = ws[r & m]) != null && (b = q.base) - q.top < 0 &&
1534	>	(a = q.array) != null) {  // probably nonempty
1535	>	int i = (((a.length - 1) & b) << ASHIFT) + ABASE;
1536	>	t = (ForkJoinTask<?>)U.getObjectVolatile(a, i);
1537	>	if (q.base == b && ec >= 0 && t != null &&
1538	>	U.compareAndSwapObject(a, i, t, null)) {
1539	>	if (q.top - (q.base = b + 1) > 1)
1540	>	signalWork();    // help pushes signal
1541		return t;
1542	<	break;
1542	>	}
1543	>	else if (ec < 0 || j <= m) {
1544	>	rs = 0;               // mark scan as imcomplete
1545	>	break;                // caller can retry after release
1546	>	}
1547		}
1548	<	else if (j - m > m) {
1515	<	if (rs == runState)      // staleness check
1516	<	swept = true;
1548	>	if (--j < 0)
1549		break;
1518	–	}
1550		}
1551
1521	–	// Decode ctl on empty scan
1552		long c = ctl; int e = (int)c, a = (int)(c >> AC_SHIFT), nr, ns;
1553	<	if (e < 0)                       // pool is terminating
1554	<	w.runState = -1;
1555	<	else if (!swept) {               // try to release a waiter
1556	<	WorkQueue v; Thread p;
1557	<	if (e > 0 && a < 0 && (v = ws[e & m]) != null &&
1558	<	v.eventCount == (e | INT_SIGN)) {
1553	>	if (e < 0)                        // decode ctl on empty scan
1554	>	w.runState = -1;              // pool is terminating
1555	>	else if (rs == 0 || rs != runState) { // incomplete scan
1556	>	WorkQueue v; Thread p;        // try to release a waiter
1557	>	if (e > 0 && a < 0 && w.eventCount == ec &&
1558	>	(v = ws[e & m]) != null && v.eventCount == (e | INT_SIGN)) {
1559		long nc = ((long)(v.nextWait & E_MASK) |
1560		((c + AC_UNIT) & (AC_MASK|TC_MASK)));
1561	<	if (U.compareAndSwapLong(this, CTL, c, nc)) {
1561	>	if (ctl == c && U.compareAndSwapLong(this, CTL, c, nc)) {
1562		v.eventCount = (e + E_SEQ) & E_MASK;
1563		if ((p = v.parker) != null)
1564		U.unpark(p);
1565		}
1566		}
1567		}
1568	<	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
1568	>	else if (ec >= 0) {               // try to enqueue/inactivate
1569		long nc = (long)ec | ((c - AC_UNIT) & (AC_MASK|TC_MASK));
1570		w.nextWait = e;
1571	<	w.eventCount = ec | INT_SIGN;// mark as inactive
1572	<	if (!U.compareAndSwapLong(this, CTL, c, nc))
1573	<	w.eventCount = ec;       // unmark on CAS failure
1574	<	else if ((ns = w.nsteals) != 0) {
1575	<	w.nsteals = 0;           // set rescans if ran task
1576	<	w.rescans = a + parallelism;
1577	<	w.totalSteals += ns;
1571	>	w.eventCount = ec | INT_SIGN; // mark as inactive
1572	>	if (ctl != c || !U.compareAndSwapLong(this, CTL, c, nc))
1573	>	w.eventCount = ec;        // unmark on CAS failure
1574	>	else {
1575	>	if ((ns = w.nsteals) != 0) {
1576	>	w.nsteals = 0;        // set rescans if ran task
1577	>	w.rescans = (a > 0) ? 0 : a + parallelism;
1578	>	w.totalSteals += ns;
1579	>	}
1580	>	if (a == 1 - parallelism) // quiescent
1581	>	idleAwaitWork(w, nc, c);
1582		}
1583		}
1584	<	else{                            // already queued
1585	<	if (parallelism == -a)
1586	<	idleAwaitWork(w);        // quiescent
1587	<	if (w.eventCount == ec) {
1588	<	Thread.interrupted();    // clear status
1589	<	ForkJoinWorkerThread wt = w.owner;
1584	>	else if (w.eventCount < 0) {      // already queued
1585	>	if ((nr = w.rescans) > 0) {   // continue rescanning
1586	>	int ac = a + parallelism;
1587	>	if (((w.rescans = (ac < nr) ? ac : nr - 1) & 3) == 0)
1588	>	Thread.yield();       // yield before block
1589	>	}
1590	>	else {
1591	>	Thread.interrupted();     // clear status
1592	>	Thread wt = Thread.currentThread();
1593		U.putObject(wt, PARKBLOCKER, this);
1594	<	w.parker = wt;           // emulate LockSupport.park
1595	<	if (w.eventCount == ec)  // recheck
1596	<	U.park(false, 0L);   // block
1594	>	w.parker = wt;            // emulate LockSupport.park
1595	>	if (w.eventCount < 0)     // recheck
1596	>	U.park(false, 0L);
1597		w.parker = null;
1598		U.putObject(wt, PARKBLOCKER, null);
1599		}
#	Line 1572 | Line 1603 | public class ForkJoinPool extends Abstra
1603		}
1604
1605		/**
1606	<	* If inactivating worker w has caused pool to become quiescent,
1607	<	* checks for pool termination, and, so long as this is not the
1608	<	* only worker, waits for event for up to SHRINK_RATE nanosecs.
1609	<	* On timeout, if ctl has not changed, terminates the worker,
1610	<	* which will in turn wake up another worker to possibly repeat
1611	<	* this process.
1606	>	* If inactivating worker w has caused the pool to become
1607	>	* quiescent, checks for pool termination, and, so long as this is
1608	>	* not the only worker, waits for event for up to SHRINK_RATE
1609	>	* nanosecs.  On timeout, if ctl has not changed, terminates the
1610	>	* worker, which will in turn wake up another worker to possibly
1611	>	* repeat this process.
1612		*
1613		* @param w the calling worker
1614	+	* @param currentCtl the ctl value triggering possible quiescence
1615	+	* @param prevCtl the ctl value to restore if thread is terminated
1616		*/
1617	<	private void idleAwaitWork(WorkQueue w) {
1618	<	long c; int nw, ec;
1619	<	if (!tryTerminate(false, false) &&
1620	<	(int)((c = ctl) >> AC_SHIFT) + parallelism == 0 &&
1621	<	(ec = w.eventCount) == ((int)c | INT_SIGN) &&
1622	<	(nw = w.nextWait) != 0) {
1590	<	long nc = ((long)(nw & E_MASK) | // ctl to restore on timeout
1591	<	((c + AC_UNIT) & AC_MASK) | (c & TC_MASK));
1592	<	ForkJoinWorkerThread wt = w.owner;
1593	<	while (ctl == c) {
1617	>	private void idleAwaitWork(WorkQueue w, long currentCtl, long prevCtl) {
1618	>	if (w.eventCount < 0 && !tryTerminate(false, false) &&
1619	>	(int)prevCtl != 0 && !hasQueuedSubmissions() && ctl == currentCtl) {
1620	>	Thread wt = Thread.currentThread();
1621	>	Thread.yield();            // yield before block
1622	>	while (ctl == currentCtl) {
1623		long startTime = System.nanoTime();
1624		Thread.interrupted();  // timed variant of version in scan()
1625		U.putObject(wt, PARKBLOCKER, this);
1626		w.parker = wt;
1627	<	if (ctl == c)
1627	>	if (ctl == currentCtl)
1628		U.park(false, SHRINK_RATE);
1629		w.parker = null;
1630		U.putObject(wt, PARKBLOCKER, null);
1631	<	if (ctl != c)
1631	>	if (ctl != currentCtl)
1632		break;
1633		if (System.nanoTime() - startTime >= SHRINK_TIMEOUT &&
1634	<	U.compareAndSwapLong(this, CTL, c, nc)) {
1635	<	w.eventCount = (ec + E_SEQ) | E_MASK;
1636	<	w.runState = -1;          // shrink
1634	>	U.compareAndSwapLong(this, CTL, currentCtl, prevCtl)) {
1635	>	w.eventCount = (w.eventCount + E_SEQ) | E_MASK;
1636	>	w.runState = -1;   // shrink
1637		break;
1638		}
1639		}
#	Line 1622 | Line 1651 | public class ForkJoinPool extends Abstra
1651		* leaves hints in workers to speed up subsequent calls. The
1652		* implementation is very branchy to cope with potential
1653		* inconsistencies or loops encountering chains that are stale,
1654	<	* unknown, or of length greater than MAX_HELP_DEPTH links.  All
1626	<	* of these cases are dealt with by just retrying by caller.
1654	>	* unknown, or so long that they are likely cyclic.
1655		*
1656		* @param joiner the joining worker
1657		* @param task the task to join
1658	<	* @return true if found or ran a task (and so is immediately retryable)
1658	>	* @return 0 if no progress can be made, negative if task
1659	>	* known complete, else positive
1660		*/
1661	<	final boolean tryHelpStealer(WorkQueue joiner, ForkJoinTask<?> task) {
1662	<	ForkJoinTask<?> subtask;    // current target
1663	<	boolean progress = false;
1664	<	int depth = 0;              // current chain depth
1665	<	int m = runState & SMASK;
1666	<	WorkQueue[] ws = workQueues;
1667	<
1668	<	if (ws != null && ws.length > m && (subtask = task).status >= 0) {
1669	<	outer:for (WorkQueue j = joiner;;) {
1670	<	// 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	<	}
1661	>	private int tryHelpStealer(WorkQueue joiner, ForkJoinTask<?> task) {
1662	>	int stat = 0, steps = 0;                    // bound to avoid cycles
1663	>	if (joiner != null && task != null) {       // hoist null checks
1664	>	restart: for (;;) {
1665	>	ForkJoinTask<?> subtask = task;     // current target
1666	>	for (WorkQueue j = joiner, v;;) {   // v is stealer of subtask
1667	>	WorkQueue[] ws; int m, s, h;
1668	>	if ((s = task.status) < 0) {
1669	>	stat = s;
1670	>	break restart;
1671		}
1672	<	if (stealer == null)
1673	<	break;
1674	<	}
1675	<
1676	<	for (WorkQueue q = stealer;;) { // Try to help stealer
1677	<	ForkJoinTask<?> t; int b;
1678	<	if (task.status < 0)
1679	<	break outer;
1680	<	if ((b = q.base) - q.top < 0) {
1681	<	progress = true;
1682	<	if (subtask.status < 0)
1683	<	break outer;               // stale
1684	<	if ((t = q.pollAt(b)) != null) {
1685	<	stealer.stealHint = joiner.poolIndex;
1686	<	joiner.runSubtask(t);
1672	>	if ((ws = workQueues) == null || (m = ws.length - 1) <= 0)
1673	>	break restart;              // shutting down
1674	>	if ((v = ws[h = (j.stealHint | 1) & m]) == null ||
1675	>	v.currentSteal != subtask) {
1676	>	for (int origin = h;;) {    // find stealer
1677	>	if (((h = (h + 2) & m) & 15) == 1 &&
1678	>	(subtask.status < 0 || j.currentJoin != subtask))
1679	>	continue restart;   // occasional staleness check
1680	>	if ((v = ws[h]) != null &&
1681	>	v.currentSteal == subtask) {
1682	>	j.stealHint = h;    // save hint
1683	>	break;
1684	>	}
1685	>	if (h == origin)
1686	>	break restart;      // cannot find stealer
1687		}
1688		}
1689	<	else { // empty - try to descend to find stealer's stealer
1690	<	ForkJoinTask<?> next = stealer.currentJoin;
1691	<	if (++depth == MAX_HELP_DEPTH || subtask.status < 0 ||
1692	<	next == null || next == subtask)
1693	<	break outer;  // max depth, stale, dead-end, cyclic
1694	<	subtask = next;
1695	<	j = stealer;
1696	<	break;
1689	>	for (;;) { // help stealer or descend to its stealer
1690	>	ForkJoinTask[] a;  int b;
1691	>	if (subtask.status < 0)     // surround probes with
1692	>	continue restart;       //   consistency checks
1693	>	if ((b = v.base) - v.top < 0 && (a = v.array) != null) {
1694	>	int i = (((a.length - 1) & b) << ASHIFT) + ABASE;
1695	>	ForkJoinTask<?> t =
1696	>	(ForkJoinTask<?>)U.getObjectVolatile(a, i);
1697	>	if (subtask.status < 0 || j.currentJoin != subtask ||
1698	>	v.currentSteal != subtask)
1699	>	continue restart;   // stale
1700	>	stat = 1;               // apparent progress
1701	>	if (t != null && v.base == b &&
1702	>	U.compareAndSwapObject(a, i, t, null)) {
1703	>	v.base = b + 1;     // help stealer
1704	>	joiner.runSubtask(t);
1705	>	}
1706	>	else if (v.base == b && ++steps == MAX_HELP)
1707	>	break restart;      // v apparently stalled
1708	>	}
1709	>	else {                      // empty -- try to descend
1710	>	ForkJoinTask<?> next = v.currentJoin;
1711	>	if (subtask.status < 0 || j.currentJoin != subtask ||
1712	>	v.currentSteal != subtask)
1713	>	continue restart;   // stale
1714	>	else if (next == null || ++steps == MAX_HELP)
1715	>	break restart;      // dead-end or maybe cyclic
1716	>	else {
1717	>	subtask = next;
1718	>	j = v;
1719	>	break;
1720	>	}
1721	>	}
1722		}
1723		}
1724		}
1725		}
1726	<	return progress;
1726	>	return stat;
1727		}
1728
1729		/**
#	Line 1689 | Line 1732 | public class ForkJoinPool extends Abstra
1732		* @param joiner the joining worker
1733		* @param task the task
1734		*/
1735	<	final void tryPollForAndExec(WorkQueue joiner, ForkJoinTask<?> task) {
1735	>	private void tryPollForAndExec(WorkQueue joiner, ForkJoinTask<?> task) {
1736		WorkQueue[] ws;
1737	<	int m = runState & SMASK;
1738	<	if ((ws = workQueues) != null && ws.length > m) {
1696	<	for (int j = 1; j <= m && task.status >= 0; j += 2) {
1737	>	if ((ws = workQueues) != null) {
1738	>	for (int j = 1; j < ws.length && task.status >= 0; j += 2) {
1739		WorkQueue q = ws[j];
1740		if (q != null && q.pollFor(task)) {
1741		joiner.runSubtask(task);
#	Line 1704 | Line 1746 | public class ForkJoinPool extends Abstra
1746		}
1747
1748		/**
1749	<	* Returns a non-empty steal queue, if one is found during a random,
1750	<	* then cyclic scan, else null.  This method must be retried by
1751	<	* caller if, by the time it tries to use the queue, it is empty.
1749	>	* Tries to decrement active count (sometimes implicitly) and
1750	>	* possibly release or create a compensating worker in preparation
1751	>	* for blocking. Fails on contention or termination. Otherwise,
1752	>	* adds a new thread if no idle workers are available and either
1753	>	* pool would become completely starved or: (at least half
1754	>	* starved, and fewer than 50% spares exist, and there is at least
1755	>	* one task apparently available). Even though the availability
1756	>	* check requires a full scan, it is worthwhile in reducing false
1757	>	* alarms.
1758	>	*
1759	>	* @param task if non-null, a task being waited for
1760	>	* @param blocker if non-null, a blocker being waited for
1761	>	* @return true if the caller can block, else should recheck and retry
1762	>	*/
1763	>	final boolean tryCompensate(ForkJoinTask<?> task, ManagedBlocker blocker) {
1764	>	int pc = parallelism, e;
1765	>	long c = ctl;
1766	>	WorkQueue[] ws = workQueues;
1767	>	if ((e = (int)c) >= 0 && ws != null) {
1768	>	int u, a, ac, hc;
1769	>	int tc = (short)((u = (int)(c >>> 32)) >>> UTC_SHIFT) + pc;
1770	>	boolean replace = false;
1771	>	if ((a = u >> UAC_SHIFT) <= 0) {
1772	>	if ((ac = a + pc) <= 1)
1773	>	replace = true;
1774	>	else if ((e > 0 || (task != null &&
1775	>	ac <= (hc = pc >>> 1) && tc < pc + hc))) {
1776	>	WorkQueue w;
1777	>	for (int j = 0; j < ws.length; ++j) {
1778	>	if ((w = ws[j]) != null && !w.isEmpty()) {
1779	>	replace = true;
1780	>	break;   // in compensation range and tasks available
1781	>	}
1782	>	}
1783	>	}
1784	>	}
1785	>	if ((task == null || task.status >= 0) && // recheck need to block
1786	>	(blocker == null || !blocker.isReleasable()) && ctl == c) {
1787	>	if (!replace) {          // no compensation
1788	>	long nc = ((c - AC_UNIT) & AC_MASK) | (c & ~AC_MASK);
1789	>	if (U.compareAndSwapLong(this, CTL, c, nc))
1790	>	return true;
1791	>	}
1792	>	else if (e != 0) {       // release an idle worker
1793	>	WorkQueue w; Thread p; int i;
1794	>	if ((i = e & SMASK) < ws.length && (w = ws[i]) != null) {
1795	>	long nc = ((long)(w.nextWait & E_MASK) |
1796	>	(c & (AC_MASK|TC_MASK)));
1797	>	if (w.eventCount == (e | INT_SIGN) &&
1798	>	U.compareAndSwapLong(this, CTL, c, nc)) {
1799	>	w.eventCount = (e + E_SEQ) & E_MASK;
1800	>	if ((p = w.parker) != null)
1801	>	U.unpark(p);
1802	>	return true;
1803	>	}
1804	>	}
1805	>	}
1806	>	else if (tc < MAX_CAP) { // create replacement
1807	>	long nc = ((c + TC_UNIT) & TC_MASK) | (c & ~TC_MASK);
1808	>	if (U.compareAndSwapLong(this, CTL, c, nc)) {
1809	>	addWorker();
1810	>	return true;
1811	>	}
1812	>	}
1813	>	}
1814	>	}
1815	>	return false;
1816	>	}
1817	>
1818	>	/**
1819	>	* Helps and/or blocks until the given task is done.
1820	>	*
1821	>	* @param joiner the joining worker
1822	>	* @param task the task
1823	>	* @return task status on exit
1824	>	*/
1825	>	final int awaitJoin(WorkQueue joiner, ForkJoinTask<?> task) {
1826	>	int s;
1827	>	if ((s = task.status) >= 0) {
1828	>	ForkJoinTask<?> prevJoin = joiner.currentJoin;
1829	>	joiner.currentJoin = task;
1830	>	long startTime = 0L;
1831	>	for (int k = 0;;) {
1832	>	if ((s = (joiner.isEmpty() ?           // try to help
1833	>	tryHelpStealer(joiner, task) :
1834	>	joiner.tryRemoveAndExec(task))) == 0 &&
1835	>	(s = task.status) >= 0) {
1836	>	if (k == 0) {
1837	>	startTime = System.nanoTime();
1838	>	tryPollForAndExec(joiner, task); // check uncommon case
1839	>	}
1840	>	else if ((k & (MAX_HELP - 1)) == 0 &&
1841	>	System.nanoTime() - startTime >=
1842	>	COMPENSATION_DELAY &&
1843	>	tryCompensate(task, null)) {
1844	>	if (task.trySetSignal()) {
1845	>	synchronized (task) {
1846	>	if (task.status >= 0) {
1847	>	try {                // see ForkJoinTask
1848	>	task.wait();     //  for explanation
1849	>	} catch (InterruptedException ie) {
1850	>	}
1851	>	}
1852	>	else
1853	>	task.notifyAll();
1854	>	}
1855	>	}
1856	>	long c;                          // re-activate
1857	>	do {} while (!U.compareAndSwapLong
1858	>	(this, CTL, c = ctl, c + AC_UNIT));
1859	>	}
1860	>	}
1861	>	if (s < 0 || (s = task.status) < 0) {
1862	>	joiner.currentJoin = prevJoin;
1863	>	break;
1864	>	}
1865	>	else if ((k++ & (MAX_HELP - 1)) == MAX_HELP >>> 1)
1866	>	Thread.yield();                     // for politeness
1867	>	}
1868	>	}
1869	>	return s;
1870	>	}
1871	>
1872	>	/**
1873	>	* Stripped-down variant of awaitJoin used by timed joins. Tries
1874	>	* to help join only while there is continuous progress. (Caller
1875	>	* will then enter a timed wait.)
1876	>	*
1877	>	* @param joiner the joining worker
1878	>	* @param task the task
1879	>	* @return task status on exit
1880	>	*/
1881	>	final int helpJoinOnce(WorkQueue joiner, ForkJoinTask<?> task) {
1882	>	int s;
1883	>	while ((s = task.status) >= 0 &&
1884	>	(joiner.isEmpty() ?
1885	>	tryHelpStealer(joiner, task) :
1886	>	joiner.tryRemoveAndExec(task)) != 0)
1887	>	;
1888	>	return s;
1889	>	}
1890	>
1891	>	/**
1892	>	* Returns a (probably) non-empty steal queue, if one is found
1893	>	* during a random, then cyclic scan, else null.  This method must
1894	>	* be retried by caller if, by the time it tries to use the queue,
1895	>	* it is empty.
1896		*/
1897		private WorkQueue findNonEmptyStealQueue(WorkQueue w) {
1898	<	int r = w.seed;    // Same idea as scan(), but ignoring submissions
1898	>	// Similar to loop in scan(), but ignoring submissions
1899	>	int r = w.seed; r ^= r << 13; r ^= r >>> 17; w.seed = r ^= r << 5;
1900	>	int step = (r >>> 16) | 1;
1901		for (WorkQueue[] ws;;) {
1902	<	int m = runState & SMASK;
1903	<	if ((ws = workQueues) == null)
1902	>	int rs = runState, m;
1903	>	if ((ws = workQueues) == null || (m = ws.length - 1) < 1)
1904		return null;
1905	<	if (ws.length > m) {
1906	<	WorkQueue q;
1907	<	for (int k = 0, j = -1 - m;; ++j) {
1908	<	if (j < 0) {
1909	<	r ^= r << 13; r ^= r >>> 17; k = r ^= r << 5;
1910	<	}
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)
1905	>	for (int j = (m + 1) << 2; ; r += step) {
1906	>	WorkQueue q = ws[((r << 1) | 1) & m];
1907	>	if (q != null && !q.isEmpty())
1908	>	return q;
1909	>	else if (--j < 0) {
1910	>	if (runState == rs)
1911		return null;
1912	+	break;
1913		}
1914		}
1915		}
1916		}
1917
1918	+
1919		/**
1920		* Runs tasks until {@code isQuiescent()}. We piggyback on
1921		* active count ctl maintenance, but rather than blocking
#	Line 1741 | Line 1924 | public class ForkJoinPool extends Abstra
1924		*/
1925		final void helpQuiescePool(WorkQueue w) {
1926		for (boolean active = true;;) {
1927	<	w.runLocalTasks();      // exhaust local queue
1927	>	ForkJoinTask<?> localTask; // exhaust local queue
1928	>	while ((localTask = w.nextLocalTask()) != null)
1929	>	localTask.doExec();
1930		WorkQueue q = findNonEmptyStealQueue(w);
1931		if (q != null) {
1932	<	ForkJoinTask<?> t;
1932	>	ForkJoinTask<?> t; int b;
1933		if (!active) {      // re-establish active count
1934		long c;
1935		active = true;
1936		do {} while (!U.compareAndSwapLong
1937		(this, CTL, c = ctl, c + AC_UNIT));
1938		}
1939	<	if ((t = q.poll()) != null)
1939	>	if ((b = q.base) - q.top < 0 && (t = q.pollAt(b)) != null)
1940		w.runSubtask(t);
1941		}
1942		else {
#	Line 1779 | Line 1964 | public class ForkJoinPool extends Abstra
1964		*/
1965		final ForkJoinTask<?> nextTaskFor(WorkQueue w) {
1966		for (ForkJoinTask<?> t;;) {
1967	<	WorkQueue q;
1967	>	WorkQueue q; int b;
1968		if ((t = w.nextLocalTask()) != null)
1969		return t;
1970		if ((q = findNonEmptyStealQueue(w)) == null)
1971		return null;
1972	<	if ((t = q.poll()) != null)
1972	>	if ((b = q.base) - q.top < 0 && (t = q.pollAt(b)) != null)
1973		return t;
1974		}
1975		}
#	Line 1818 | Line 2003 | public class ForkJoinPool extends Abstra
2003		*
2004		* @param now if true, unconditionally terminate, else only
2005		* if no work and no active workers
2006	<	* @paran enable if true, enable shutdown when next possible
2006	>	* @param enable if true, enable shutdown when next possible
2007		* @return true if now terminating or terminated
2008		*/
2009		private boolean tryTerminate(boolean now, boolean enable) {
#	Line 1959 | Line 2144 | public class ForkJoinPool extends Abstra
2144		checkPermission();
2145		if (factory == null)
2146		throw new NullPointerException();
2147	<	if (parallelism <= 0 || parallelism > POOL_MAX)
2147	>	if (parallelism <= 0 || parallelism > MAX_CAP)
2148		throw new IllegalArgumentException();
2149		this.parallelism = parallelism;
2150		this.factory = factory;
2151		this.ueh = handler;
2152		this.localMode = asyncMode ? FIFO_QUEUE : LIFO_QUEUE;
1968	–	this.growHints = 1;
2153		long np = (long)(-parallelism); // offset ctl counts
2154		this.ctl = ((np << AC_SHIFT) & AC_MASK) | ((np << TC_SHIFT) & TC_MASK);
2155	<	// initialize workQueues array with room for 2*parallelism if possible
2156	<	int n = parallelism << 1;
2157	<	if (n >= POOL_MAX)
2158	<	n = POOL_MAX;
2159	<	else { // See Hackers Delight, sec 3.2, where n < (1 << 16)
2160	<	n |= n >>> 1; n |= n >>> 2; n |= n >>> 4; n |= n >>> 8;
1977	<	}
1978	<	this.workQueues = new WorkQueue[(n + 1) << 1]; // #slots = 2 * #workers
2155	>	// Use nearest power 2 for workQueues size. See Hackers Delight sec 3.2.
2156	>	int n = parallelism - 1;
2157	>	n |= n >>> 1; n |= n >>> 2; n |= n >>> 4; n |= n >>> 8; n |= n >>> 16;
2158	>	int size = (n + 1) << 1;        // #slots = 2*#workers
2159	>	this.submitMask = size - 1;     // room for max # of submit queues
2160	>	this.workQueues = new WorkQueue[size];
2161		this.termination = (this.lock = new Mutex()).newCondition();
2162		this.stealCount = new AtomicLong();
2163		this.nextWorkerNumber = new AtomicInteger();
2164	+	int pn = poolNumberGenerator.incrementAndGet();
2165		StringBuilder sb = new StringBuilder("ForkJoinPool-");
2166	<	sb.append(poolNumberGenerator.incrementAndGet());
2166	>	sb.append(Integer.toString(pn));
2167		sb.append("-worker-");
2168		this.workerNamePrefix = sb.toString();
2169	+	lock.lock();
2170	+	this.runState = 1;              // set init flag
2171	+	lock.unlock();
2172		}
2173
2174		// Execution methods
#	Line 2004 | Line 2190 | public class ForkJoinPool extends Abstra
2190		*         scheduled for execution
2191		*/
2192		public <T> T invoke(ForkJoinTask<T> task) {
2193	+	if (task == null)
2194	+	throw new NullPointerException();
2195		doSubmit(task);
2196		return task.join();
2197		}
#	Line 2017 | Line 2205 | public class ForkJoinPool extends Abstra
2205		*         scheduled for execution
2206		*/
2207		public void execute(ForkJoinTask<?> task) {
2208	+	if (task == null)
2209	+	throw new NullPointerException();
2210		doSubmit(task);
2211		}
2212
#	Line 2034 | Line 2224 | public class ForkJoinPool extends Abstra
2224		if (task instanceof ForkJoinTask<?>) // avoid re-wrap
2225		job = (ForkJoinTask<?>) task;
2226		else
2227	<	job = ForkJoinTask.adapt(task, null);
2227	>	job = new ForkJoinTask.AdaptedRunnableAction(task);
2228		doSubmit(job);
2229		}
2230
#	Line 2048 | Line 2238 | public class ForkJoinPool extends Abstra
2238		*         scheduled for execution
2239		*/
2240		public <T> ForkJoinTask<T> submit(ForkJoinTask<T> task) {
2241	+	if (task == null)
2242	+	throw new NullPointerException();
2243		doSubmit(task);
2244		return task;
2245		}
#	Line 2058 | Line 2250 | public class ForkJoinPool extends Abstra
2250		*         scheduled for execution
2251		*/
2252		public <T> ForkJoinTask<T> submit(Callable<T> task) {
2253	<	if (task == null)
2062	<	throw new NullPointerException();
2063	<	ForkJoinTask<T> job = ForkJoinTask.adapt(task);
2253	>	ForkJoinTask<T> job = new ForkJoinTask.AdaptedCallable<T>(task);
2254		doSubmit(job);
2255		return job;
2256		}
#	Line 2071 | Line 2261 | public class ForkJoinPool extends Abstra
2261		*         scheduled for execution
2262		*/
2263		public <T> ForkJoinTask<T> submit(Runnable task, T result) {
2264	<	if (task == null)
2075	<	throw new NullPointerException();
2076	<	ForkJoinTask<T> job = ForkJoinTask.adapt(task, result);
2264	>	ForkJoinTask<T> job = new ForkJoinTask.AdaptedRunnable<T>(task, result);
2265		doSubmit(job);
2266		return job;
2267		}
#	Line 2090 | Line 2278 | public class ForkJoinPool extends Abstra
2278		if (task instanceof ForkJoinTask<?>) // avoid re-wrap
2279		job = (ForkJoinTask<?>) task;
2280		else
2281	<	job = ForkJoinTask.adapt(task, null);
2281	>	job = new ForkJoinTask.AdaptedRunnableAction(task);
2282		doSubmit(job);
2283		return job;
2284		}
#	Line 2112 | Line 2300 | public class ForkJoinPool extends Abstra
2300		boolean done = false;
2301		try {
2302		for (Callable<T> t : tasks) {
2303	<	ForkJoinTask<T> f = ForkJoinTask.adapt(t);
2303	>	ForkJoinTask<T> f = new ForkJoinTask.AdaptedCallable<T>(t);
2304		doSubmit(f);
2305		fs.add(f);
2306		}
#	Line 2298 | Line 2486 | public class ForkJoinPool extends Abstra
2486		WorkQueue[] ws; WorkQueue w;
2487		if ((ws = workQueues) != null) {
2488		for (int i = 0; i < ws.length; i += 2) {
2489	<	if ((w = ws[i]) != null && w.queueSize() != 0)
2489	>	if ((w = ws[i]) != null && !w.isEmpty())
2490		return true;
2491		}
2492		}
#	Line 2612 | Line 2800 | public class ForkJoinPool extends Abstra
2800		ForkJoinPool p = ((t instanceof ForkJoinWorkerThread) ?
2801		((ForkJoinWorkerThread)t).pool : null);
2802		while (!blocker.isReleasable()) {
2803	<	if (p == null || p.tryCompensate()) {
2803	>	if (p == null || p.tryCompensate(null, blocker)) {
2804		try {
2805		do {} while (!blocker.isReleasable() && !blocker.block());
2806		} finally {
#	Line 2629 | Line 2817 | public class ForkJoinPool extends Abstra
2817		// implement RunnableFuture.
2818
2819		protected <T> RunnableFuture<T> newTaskFor(Runnable runnable, T value) {
2820	<	return (RunnableFuture<T>) ForkJoinTask.adapt(runnable, value);
2820	>	return new ForkJoinTask.AdaptedRunnable<T>(runnable, value);
2821		}
2822
2823		protected <T> RunnableFuture<T> newTaskFor(Callable<T> callable) {
2824	<	return (RunnableFuture<T>) ForkJoinTask.adapt(callable);
2824	>	return new ForkJoinTask.AdaptedCallable<T>(callable);
2825		}
2826
2827		// Unsafe mechanics
2828		private static final sun.misc.Unsafe U;
2829		private static final long CTL;
2830		private static final long PARKBLOCKER;
2831	+	private static final int ABASE;
2832	+	private static final int ASHIFT;
2833
2834		static {
2835		poolNumberGenerator = new AtomicInteger();
2836	+	nextSubmitterSeed = new AtomicInteger(0x55555555);
2837		modifyThreadPermission = new RuntimePermission("modifyThread");
2838		defaultForkJoinWorkerThreadFactory =
2839		new DefaultForkJoinWorkerThreadFactory();
2840		submitters = new ThreadSubmitter();
2841	+	int s;
2842		try {
2843		U = getUnsafe();
2844		Class<?> k = ForkJoinPool.class;
2845	+	Class<?> ak = ForkJoinTask[].class;
2846		CTL = U.objectFieldOffset
2847		(k.getDeclaredField("ctl"));
2848		Class<?> tk = Thread.class;
2849		PARKBLOCKER = U.objectFieldOffset
2850		(tk.getDeclaredField("parkBlocker"));
2851	+	ABASE = U.arrayBaseOffset(ak);
2852	+	s = U.arrayIndexScale(ak);
2853		} catch (Exception e) {
2854		throw new Error(e);
2855		}
2856	+	if ((s & (s-1)) != 0)
2857	+	throw new Error("data type scale not a power of two");
2858	+	ASHIFT = 31 - Integer.numberOfLeadingZeros(s);
2859		}
2860
2861		/**

Diff Legend

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