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 1.117 by jsr166, Sat Jan 28 04:32:25 2012 UTC vs.
Revision 1.137 by dl, Tue Oct 30 14:23:11 2012 UTC

#	Line 21 | Line 21 | import java.util.concurrent.RunnableFutu
21		import java.util.concurrent.TimeUnit;
22		import java.util.concurrent.atomic.AtomicInteger;
23		import java.util.concurrent.atomic.AtomicLong;
24	<	import java.util.concurrent.locks.ReentrantLock;
24	>	import java.util.concurrent.locks.AbstractQueuedSynchronizer;
25		import java.util.concurrent.locks.Condition;
26
27		/**
#	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 203 | Line 201 | public class ForkJoinPool extends Abstra
201		* take tasks, and they are multiplexed on to a finite number of
202		* shared work queues. However, classes are set up so that future
203		* extensions could allow submitters to optionally help perform
204	<	* tasks as well. Pool submissions from internal workers are also
205	<	* allowed, but use randomized rather than thread-hashed queue
206	<	* indices to avoid imbalance.  Insertion of tasks in shared mode
207	<	* requires a lock (mainly to protect in the case of resizing) but
208	<	* we use only a simple spinlock (using bits in field runState),
209	<	* because submitters encountering a busy queue try or create
212	<	* others so never block.
204	>	* tasks as well. Insertion of tasks in shared mode requires a
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 -- they block only when creating and registering
209	>	* new queues.
210		*
211		* Management
212		* ==========
#	Line 238 | Line 235 | public class ForkJoinPool extends Abstra
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
247	<	* scanning. To avoid flailing during start-up, the array is
248	<	* presized to hold twice #parallelism workers (which is unlikely
249	<	* to need further resizing during execution). But to avoid
253	<	* dealing with so many null slots, variable runState includes a
254	<	* mask for the nearest power of two that contains all current
255	<	* workers.  All worker thread creation is on-demand, triggered by
256	<	* task submissions, replacement of terminated workers, and/or
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
250		* compensation for blocked workers. However, all other support
251		* code is set up to work with other policies.  To ensure that we
252		* do not hold on to worker references that would prevent GC, ALL
#	Line 266 | Line 259 | public class ForkJoinPool extends Abstra
259		* both index-check and null-check the IDs. All such accesses
260		* ignore bad IDs by returning out early from what they are doing,
261		* since this can only be associated with termination, in which
262	<	* case it is OK to give up.
263	<	*
264	<	* All uses of the workQueues array check that it is non-null
265	<	* (even if previously non-null). This allows nulling during
266	<	* termination, which is currently not necessary, but remains an
267	<	* option for resource-revocation-based shutdown schemes. It also
275	<	* helps reduce JIT issuance of uncommon-trap code, which tends to
262	>	* case it is OK to give up.  All uses of the workQueues array
263	>	* also check that it is non-null (even if previously
264	>	* non-null). This allows nulling during termination, which is
265	>	* currently not necessary, but remains an option for
266	>	* resource-revocation-based shutdown schemes. It also helps
267	>	* reduce JIT issuance of uncommon-trap code, which tends to
268		* unnecessarily complicate control flow in some methods.
269		*
270		* Event Queuing. Unlike HPC work-stealing frameworks, we cannot
#	Line 323 | 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 383 | 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 workers 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 395 | 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 418 | Line 433 | public class ForkJoinPool extends Abstra
433		* managed by ForkJoinPool, so are directly accessed.  There is
434		* little point trying to reduce this, since any associated future
435		* changes in representations will need to be accompanied by
436	<	* algorithmic changes anyway. All together, these low-level
437	<	* implementation choices produce as much as a factor of 4
438	<	* performance improvement compared to naive implementations, and
439	<	* enable the processing of billions of tasks per second, at the
440	<	* expense of some ugliness.
441	<	*
442	<	* Methods signalWork() and scan() are the main bottlenecks, so are
443	<	* especially heavily micro-optimized/mangled.  There are lots of
444	<	* inline assignments (of form "while ((local = field) != 0)")
445	<	* which are usually the simplest way to ensure the required read
446	<	* orderings (which are sometimes critical). This leads to a
447	<	* "C"-like style of listing declarations of these locals at the
448	<	* heads of methods or blocks.  There are several occurrences of
449	<	* the unusual "do {} while (!cas...)"  which is the simplest way
435	<	* to force an update of a CAS'ed variable. There are also other
436	<	* coding oddities that help some methods perform reasonably even
437	<	* when interpreted (not compiled).
436	>	* algorithmic changes anyway. Several methods intrinsically
437	>	* sprawl because they must accumulate sets of consistent reads of
438	>	* volatiles held in local variables.  Methods signalWork() and
439	>	* scan() are the main bottlenecks, so are especially heavily
440	>	* micro-optimized/mangled.  There are lots of inline assignments
441	>	* (of form "while ((local = field) != 0)") which are usually the
442	>	* simplest way to ensure the required read orderings (which are
443	>	* sometimes critical). This leads to a "C"-like style of listing
444	>	* declarations of these locals at the heads of methods or blocks.
445	>	* There are several occurrences of the unusual "do {} while
446	>	* (!cas...)"  which is the simplest way to force an update of a
447	>	* CAS'ed variable. There are also other coding oddities that help
448	>	* some methods perform reasonably even when interpreted (not
449	>	* compiled).
450		*
451		* The order of declarations in this file is:
452	<	* (1) statics
453	<	* (2) fields (along with constants used when unpacking some of
454	<	*     them), listed in an order that tends to reduce contention
455	<	*     among them a bit under most JVMs;
456	<	* (3) nested classes
457	<	* (4) internal control methods
458	<	* (5) callbacks and other support for ForkJoinTask methods
459	<	* (6) exported methods (plus a few little helpers)
448	<	* (7) static block initializing all statics in a minimally
449	<	*     dependent order.
452	>	* (1) Static utility functions
453	>	* (2) Nested (static) classes
454	>	* (3) Static fields
455	>	* (4) Fields, along with constants used when unpacking some of them
456	>	* (5) Internal control methods
457	>	* (6) Callbacks and other support for ForkJoinTask methods
458	>	* (7) Exported methods
459	>	* (8) Static block initializing statics in minimally dependent order
460		*/
461
462	+	// Static utilities
463	+
464	+	/**
465	+	* If there is a security manager, makes sure caller has
466	+	* permission to modify threads.
467	+	*/
468	+	private static void checkPermission() {
469	+	SecurityManager security = System.getSecurityManager();
470	+	if (security != null)
471	+	security.checkPermission(modifyThreadPermission);
472	+	}
473	+
474	+	// Nested classes
475	+
476		/**
477		* Factory for creating new {@link ForkJoinWorkerThread}s.
478		* A {@code ForkJoinWorkerThreadFactory} must be defined and used
#	Line 477 | Line 501 | public class ForkJoinPool extends Abstra
501		}
502
503		/**
504	<	* Creates a new ForkJoinWorkerThread. This factory is used unless
505	<	* overridden in ForkJoinPool constructors.
506	<	*/
507	<	public static final ForkJoinWorkerThreadFactory
484	<	defaultForkJoinWorkerThreadFactory;
485	<
486	<	/**
487	<	* Permission required for callers of methods that may start or
488	<	* kill threads.
489	<	*/
490	<	private static final RuntimePermission modifyThreadPermission;
491	<
492	<	/**
493	<	* If there is a security manager, makes sure caller has
494	<	* permission to modify threads.
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
507	>	* actually do anything beyond having a unique identity.
508		*/
509	<	private static void checkPermission() {
510	<	SecurityManager security = System.getSecurityManager();
511	<	if (security != null)
512	<	security.checkPermission(modifyThreadPermission);
509	>	static final class EmptyTask extends ForkJoinTask<Void> {
510	>	EmptyTask() { status = ForkJoinTask.NORMAL; } // force done
511	>	public final Void getRawResult() { return null; }
512	>	public final void setRawResult(Void x) {}
513	>	public final boolean exec() { return true; }
514		}
515
516		/**
503	–	* Generator for assigning sequence numbers as pool names.
504	–	*/
505	–	private static final AtomicInteger poolNumberGenerator;
506	–
507	–	/**
508	–	* Bits and masks for control variables
509	–	*
510	–	* Field ctl is a long packed with:
511	–	* AC: Number of active running workers minus target parallelism (16 bits)
512	–	* TC: Number of total workers minus target parallelism (16 bits)
513	–	* ST: true if pool is terminating (1 bit)
514	–	* EC: the wait count of top waiting thread (15 bits)
515	–	* ID: ~(poolIndex >>> 1) of top of Treiber stack of waiters (16 bits)
516	–	*
517	–	* When convenient, we can extract the upper 32 bits of counts and
518	–	* the lower 32 bits of queue state, u = (int)(ctl >>> 32) and e =
519	–	* (int)ctl.  The ec field is never accessed alone, but always
520	–	* together with id and st. The offsets of counts by the target
521	–	* parallelism and the positionings of fields makes it possible to
522	–	* perform the most common checks via sign tests of fields: When
523	–	* ac is negative, there are not enough active workers, when tc is
524	–	* negative, there are not enough total workers, when id is
525	–	* negative, there is at least one waiting worker, and when e is
526	–	* negative, the pool is terminating.  To deal with these possibly
527	–	* negative fields, we use casts in and out of "short" and/or
528	–	* signed shifts to maintain signedness.
529	–	*
530	–	* When a thread is queued (inactivated), its eventCount field is
531	–	* negative, which is the only way to tell if a worker is
532	–	* prevented from executing tasks, even though it must continue to
533	–	* scan for them to avoid queuing races.
534	–	*
535	–	* Field runState is an int packed with:
536	–	* SHUTDOWN: true if shutdown is enabled (1 bit)
537	–	* SEQ:  a sequence number updated upon (de)registering workers (15 bits)
538	–	* MASK: mask (power of 2 - 1) covering all registered poolIndexes (16 bits)
539	–	*
540	–	* The combination of mask and sequence number enables simple
541	–	* consistency checks: Staleness of read-only operations on the
542	–	* workers and queues arrays can be checked by comparing runState
543	–	* before vs after the reads. The low 16 bits (i.e, anding with
544	–	* SMASK) hold the smallest power of two covering all worker
545	–	* indices, minus one.  The mask for queues (vs workers) is twice
546	–	* this value plus 1.
547	–	*/
548	–
549	–	// bit positions/shifts for fields
550	–	private static final int  AC_SHIFT   = 48;
551	–	private static final int  TC_SHIFT   = 32;
552	–	private static final int  ST_SHIFT   = 31;
553	–	private static final int  EC_SHIFT   = 16;
554	–
555	–	// bounds
556	–	private static final int  MAX_ID     = 0x7fff;  // max poolIndex
557	–	private static final int  SMASK      = 0xffff;  // mask short bits
558	–	private static final int  SHORT_SIGN = 1 << 15;
559	–	private static final int  INT_SIGN   = 1 << 31;
560	–
561	–	// masks
562	–	private static final long STOP_BIT   = 0x0001L << ST_SHIFT;
563	–	private static final long AC_MASK    = ((long)SMASK) << AC_SHIFT;
564	–	private static final long TC_MASK    = ((long)SMASK) << TC_SHIFT;
565	–
566	–	// units for incrementing and decrementing
567	–	private static final long TC_UNIT    = 1L << TC_SHIFT;
568	–	private static final long AC_UNIT    = 1L << AC_SHIFT;
569	–
570	–	// masks and units for dealing with u = (int)(ctl >>> 32)
571	–	private static final int  UAC_SHIFT  = AC_SHIFT - 32;
572	–	private static final int  UTC_SHIFT  = TC_SHIFT - 32;
573	–	private static final int  UAC_MASK   = SMASK << UAC_SHIFT;
574	–	private static final int  UTC_MASK   = SMASK << UTC_SHIFT;
575	–	private static final int  UAC_UNIT   = 1 << UAC_SHIFT;
576	–	private static final int  UTC_UNIT   = 1 << UTC_SHIFT;
577	–
578	–	// masks and units for dealing with e = (int)ctl
579	–	private static final int E_MASK      = 0x7fffffff; // no STOP_BIT
580	–	private static final int E_SEQ       = 1 << EC_SHIFT;
581	–
582	–	// runState bits
583	–	private static final int SHUTDOWN    = 1 << 31;
584	–	private static final int RS_SEQ      = 1 << 16;
585	–	private static final int RS_SEQ_MASK = 0x7fff0000;
586	–
587	–	// access mode for WorkQueue
588	–	static final int LIFO_QUEUE          =  0;
589	–	static final int FIFO_QUEUE          =  1;
590	–	static final int SHARED_QUEUE        = -1;
591	–
592	–	/**
593	–	* The wakeup interval (in nanoseconds) for a worker waiting for a
594	–	* task when the pool is quiescent to instead try to shrink the
595	–	* number of workers.  The exact value does not matter too
596	–	* much. It must be short enough to release resources during
597	–	* sustained periods of idleness, but not so short that threads
598	–	* are continually re-created.
599	–	*/
600	–	private static final long SHRINK_RATE =
601	–	4L * 1000L * 1000L * 1000L; // 4 seconds
602	–
603	–	/**
604	–	* The timeout value for attempted shrinkage, includes
605	–	* some slop to cope with system timer imprecision.
606	–	*/
607	–	private static final long SHRINK_TIMEOUT = SHRINK_RATE - (SHRINK_RATE / 10);
608	–
609	–	/**
610	–	* The maximum stolen->joining link depth allowed in tryHelpStealer.
611	–	* Depths for legitimate chains are unbounded, but we use a fixed
612	–	* constant to avoid (otherwise unchecked) cycles and to bound
613	–	* staleness of traversal parameters at the expense of sometimes
614	–	* blocking when we could be helping.
615	–	*/
616	–	private static final int MAX_HELP_DEPTH = 16;
617	–
618	–	/*
619	–	* Field layout order in this class tends to matter more than one
620	–	* would like. Runtime layout order is only loosely related to
621	–	* declaration order and may differ across JVMs, but the following
622	–	* empirically works OK on current JVMs.
623	–	*/
624	–
625	–	volatile long ctl;                       // main pool control
626	–	final int parallelism;                   // parallelism level
627	–	final int localMode;                     // per-worker scheduling mode
628	–	int nextPoolIndex;                       // hint used in registerWorker
629	–	volatile int runState;                   // shutdown status, seq, and mask
630	–	WorkQueue[] workQueues;                  // main registry
631	–	final ReentrantLock lock;                // for registration
632	–	final Condition termination;             // for awaitTermination
633	–	final ForkJoinWorkerThreadFactory factory; // factory for new workers
634	–	final Thread.UncaughtExceptionHandler ueh; // per-worker UEH
635	–	final AtomicLong stealCount;             // collect counts when terminated
636	–	final AtomicInteger nextWorkerNumber;    // to create worker name string
637	–	final String workerNamePrefix;           // Prefix for assigning worker names
638	–
639	–	/**
517		* Queues supporting work-stealing as well as external task
518		* submission. See above for main rationale and algorithms.
519		* Implementation relies heavily on "Unsafe" intrinsics
#	Line 685 | Line 562 | public class ForkJoinPool extends Abstra
562		* avoiding really bad worst-case access. (Until better JVM
563		* support is in place, this padding is dependent on transient
564		* properties of JVM field layout rules.)  We also take care in
565	<	* allocating and sizing and resizing the array. Non-shared queue
565	>	* allocating, sizing and resizing the array. Non-shared queue
566		* arrays are initialized (via method growArray) by workers before
567		* use. Others are allocated on first use.
568		*/
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 719 | 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) {
730	<	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.
748	–	* @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 774 | Line 673 | public class ForkJoinPool extends Abstra
673		boolean submitted = false;
674		if (runState == 0 && U.compareAndSwapInt(this, RUNSTATE, 0, 1)) {
675		ForkJoinTask<?>[] a = array;
676	<	int s = top, n = s - base;
676	>	int s = top;
677		try {
678	<	if ((a != null && n < a.length - 1) ||
678	>	if ((a != null && a.length > s + 1 - base) ||
679		(a = growArray(false)) != null) { // must presize
680		int j = (((a.length - 1) & s) << ASHIFT) + ABASE;
681		U.putObject(a, (long)j, task);    // don't need "ordered"
#	Line 791 | 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.
695		*/
696	<	final ForkJoinTask<?> poll() {
697	<	ForkJoinTask<?>[] a; int b, i;
698	<	while ((b = base) - top < 0 && (a = array) != null &&
699	<	(i = (a.length - 1) & b) >= 0) {
700	<	int j = (i << ASHIFT) + ABASE;
701	<	ForkJoinTask<?> t = (ForkJoinTask<?>)U.getObjectVolatile(a, j);
702	<	if (t != null && base == b &&
703	<	U.compareAndSwapObject(a, j, t, null)) {
704	<	base = b + 1;
705	<	return t;
696	>	final ForkJoinTask<?> pop() {
697	>	ForkJoinTask<?>[] a; ForkJoinTask<?> t; int m;
698	>	if ((a = array) != null && (m = a.length - 1) >= 0) {
699	>	for (int s; (s = top - 1) - base >= 0;) {
700	>	long j = ((m & s) << ASHIFT) + ABASE;
701	>	if ((t = (ForkJoinTask<?>)U.getObject(a, j)) == null)
702	>	break;
703	>	if (U.compareAndSwapObject(a, j, t, null)) {
704	>	top = s;
705	>	return t;
706	>	}
707		}
708		}
709		return null;
710		}
711
712	+	final ForkJoinTask<?> sharedPop() {
713	+	ForkJoinTask<?> task = null;
714	+	if (runState == 0 && U.compareAndSwapInt(this, RUNSTATE, 0, 1)) {
715	+	try {
716	+	ForkJoinTask<?>[] a; int m;
717	+	if ((a = array) != null && (m = a.length - 1) >= 0) {
718	+	for (int s; (s = top - 1) - base >= 0;) {
719	+	long j = ((m & s) << ASHIFT) + ABASE;
720	+	ForkJoinTask<?> t =
721	+	(ForkJoinTask<?>)U.getObject(a, j);
722	+	if (t == null)
723	+	break;
724	+	if (U.compareAndSwapObject(a, j, t, null)) {
725	+	top = s;
726	+	task = t;
727	+	break;
728	+	}
729	+	}
730	+	}
731	+	} finally {
732	+	runState = 0;
733	+	}
734	+	}
735	+	return task;
736	+	}
737	+
738		/**
739	<	* Takes next task, if one exists, in LIFO order.
740	<	* Call only by owner in unshared queues.
739	>	* Version of pop that takes top element only if it
740	>	* its root is the given CountedCompleter.
741		*/
742	<	final ForkJoinTask<?> pop() {
743	<	ForkJoinTask<?> t; int m;
744	<	ForkJoinTask<?>[] a = array;
818	<	if (a != null && (m = a.length - 1) >= 0) {
742	>	final ForkJoinTask<?> popCC(CountedCompleter<?> root) {
743	>	ForkJoinTask<?>[] a; int m;
744	>	if (root != null && (a = array) != null && (m = a.length - 1) >= 0) {
745		for (int s; (s = top - 1) - base >= 0;) {
746	<	int j = ((m & s) << ASHIFT) + ABASE;
747	<	if ((t = (ForkJoinTask<?>)U.getObjectVolatile(a, j)) == null)
746	>	long j = ((m & s) << ASHIFT) + ABASE;
747	>	ForkJoinTask<?> t =
748	>	(ForkJoinTask<?>)U.getObject(a, j);
749	>	if (t == null || !(t instanceof CountedCompleter) ||
750	>	((CountedCompleter<?>)t).getRoot() != root)
751		break;
752		if (U.compareAndSwapObject(a, j, t, null)) {
753		top = s;
754		return t;
755		}
756	+	if (root.status < 0)
757	+	break;
758	+	}
759	+	}
760	+	return null;
761	+	}
762	+
763	+	/**
764	+	* Shared version of popCC
765	+	*/
766	+	final ForkJoinTask<?> sharedPopCC(CountedCompleter<?> root) {
767	+	ForkJoinTask<?> task = null;
768	+	if (root != null &&
769	+	runState == 0 && U.compareAndSwapInt(this, RUNSTATE, 0, 1)) {
770	+	try {
771	+	ForkJoinTask<?>[] a; int m;
772	+	if ((a = array) != null && (m = a.length - 1) >= 0) {
773	+	for (int s; (s = top - 1) - base >= 0;) {
774	+	long j = ((m & s) << ASHIFT) + ABASE;
775	+	ForkJoinTask<?> t =
776	+	(ForkJoinTask<?>)U.getObject(a, j);
777	+	if (t == null || !(t instanceof CountedCompleter) ||
778	+	((CountedCompleter<?>)t).getRoot() != root)
779	+	break;
780	+	if (U.compareAndSwapObject(a, j, t, null)) {
781	+	top = s;
782	+	task = t;
783	+	break;
784	+	}
785	+	if (root.status < 0)
786	+	break;
787	+	}
788	+	}
789	+	} finally {
790	+	runState = 0;
791	+	}
792	+	}
793	+	return task;
794	+	}
795	+
796	+	/**
797	+	* Takes a task in FIFO order if b is base of queue and a task
798	+	* can be claimed without contention. Specialized versions
799	+	* appear in ForkJoinPool methods scan and tryHelpStealer.
800	+	*/
801	+	final ForkJoinTask<?> pollAt(int b) {
802	+	ForkJoinTask<?> t; ForkJoinTask<?>[] a;
803	+	if ((a = array) != null) {
804	+	int j = (((a.length - 1) & b) << ASHIFT) + ABASE;
805	+	if ((t = (ForkJoinTask<?>)U.getObjectVolatile(a, j)) != null &&
806	+	base == b &&
807	+	U.compareAndSwapObject(a, j, t, null)) {
808	+	base = b + 1;
809	+	return t;
810	+	}
811	+	}
812	+	return null;
813	+	}
814	+
815	+	/**
816	+	* Takes next task, if one exists, in FIFO order.
817	+	*/
818	+	final ForkJoinTask<?> poll() {
819	+	ForkJoinTask<?>[] a; int b; ForkJoinTask<?> t;
820	+	while ((b = base) - top < 0 && (a = array) != null) {
821	+	int j = (((a.length - 1) & b) << ASHIFT) + ABASE;
822	+	t = (ForkJoinTask<?>)U.getObjectVolatile(a, j);
823	+	if (t != null) {
824	+	if (base == b &&
825	+	U.compareAndSwapObject(a, j, t, null)) {
826	+	base = b + 1;
827	+	return t;
828	+	}
829	+	}
830	+	else if (base == b) {
831	+	if (b + 1 == top)
832	+	break;
833	+	Thread.yield(); // wait for lagging update
834		}
835		}
836		return null;
#	Line 849 | Line 856 | public class ForkJoinPool extends Abstra
856		}
857
858		/**
852	–	* Returns task at index b if b is current base of queue.
853	–	*/
854	–	final ForkJoinTask<?> pollAt(int b) {
855	–	ForkJoinTask<?>[] a; int i;
856	–	ForkJoinTask<?> task = null;
857	–	if ((a = array) != null && (i = ((a.length - 1) & b)) >= 0) {
858	–	int j = (i << ASHIFT) + ABASE;
859	–	ForkJoinTask<?> t = (ForkJoinTask<?>)U.getObjectVolatile(a, j);
860	–	if (t != null && base == b &&
861	–	U.compareAndSwapObject(a, j, t, null)) {
862	–	base = b + 1;
863	–	task = t;
864	–	}
865	–	}
866	–	return task;
867	–	}
868	–
869	–	/**
859		* Pops the given task only if it is at the current top.
860		*/
861		final boolean tryUnpush(ForkJoinTask<?> t) {
#	Line 881 | Line 870 | public class ForkJoinPool extends Abstra
870		}
871
872		/**
873	+	* Version of tryUnpush for shared queues; called by non-FJ
874	+	* submitters after prechecking that task probably exists.
875	+	*/
876	+	final boolean trySharedUnpush(ForkJoinTask<?> t) {
877	+	boolean success = false;
878	+	if (runState == 0 && U.compareAndSwapInt(this, RUNSTATE, 0, 1)) {
879	+	try {
880	+	ForkJoinTask<?>[] a; int s;
881	+	if ((a = array) != null && (s = top) != base &&
882	+	U.compareAndSwapObject
883	+	(a, (((a.length - 1) & --s) << ASHIFT) + ABASE, t, null)) {
884	+	top = s;
885	+	success = true;
886	+	}
887	+	} finally {
888	+	runState = 0;                         // unlock
889	+	}
890	+	}
891	+	return success;
892	+	}
893	+
894	+	/**
895		* Polls the given task only if it is at the current base.
896		*/
897		final boolean pollFor(ForkJoinTask<?> task) {
898	<	ForkJoinTask<?>[] a; int b, i;
899	<	if ((b = base) - top < 0 && (a = array) != null &&
900	<	(i = (a.length - 1) & b) >= 0) {
890	<	int j = (i << ASHIFT) + ABASE;
898	>	ForkJoinTask<?>[] a; int b;
899	>	if ((b = base) - top < 0 && (a = array) != null) {
900	>	int j = (((a.length - 1) & b) << ASHIFT) + ABASE;
901		if (U.getObjectVolatile(a, j) == task && base == b &&
902		U.compareAndSwapObject(a, j, task, null)) {
903		base = b + 1;
#	Line 898 | Line 908 | public class ForkJoinPool extends Abstra
908		}
909
910		/**
901	–	* If present, removes from queue and executes the given task, or
902	–	* any other cancelled task. Returns (true) immediately on any CAS
903	–	* or consistency check failure so caller can retry.
904	–	*
905	–	* @return false if no progress can be made
906	–	*/
907	–	final boolean tryRemoveAndExec(ForkJoinTask<?> task) {
908	–	boolean removed = false, empty = true, progress = true;
909	–	ForkJoinTask<?>[] a; int m, s, b, n;
910	–	if ((a = array) != null && (m = a.length - 1) >= 0 &&
911	–	(n = (s = top) - (b = base)) > 0) {
912	–	for (ForkJoinTask<?> t;;) {           // traverse from s to b
913	–	int j = ((--s & m) << ASHIFT) + ABASE;
914	–	t = (ForkJoinTask<?>)U.getObjectVolatile(a, j);
915	–	if (t == null)                    // inconsistent length
916	–	break;
917	–	else if (t == task) {
918	–	if (s + 1 == top) {           // pop
919	–	if (!U.compareAndSwapObject(a, j, task, null))
920	–	break;
921	–	top = s;
922	–	removed = true;
923	–	}
924	–	else if (base == b)           // replace with proxy
925	–	removed = U.compareAndSwapObject(a, j, task,
926	–	new EmptyTask());
927	–	break;
928	–	}
929	–	else if (t.status >= 0)
930	–	empty = false;
931	–	else if (s + 1 == top) {          // pop and throw away
932	–	if (U.compareAndSwapObject(a, j, t, null))
933	–	top = s;
934	–	break;
935	–	}
936	–	if (--n == 0) {
937	–	if (!empty && base == b)
938	–	progress = false;
939	–	break;
940	–	}
941	–	}
942	–	}
943	–	if (removed)
944	–	task.doExec();
945	–	return progress;
946	–	}
947	–
948	–	/**
911		* Initializes or doubles the capacity of array. Call either
912		* by owner or with lock held -- it is OK for base, but not
913		* top, to move while resizings are in progress.
#	Line 990 | Line 952 | public class ForkJoinPool extends Abstra
952		ForkJoinTask.cancelIgnoringExceptions(t);
953		}
954
955	+	/**
956	+	* Computes next value for random probes.  Scans don't require
957	+	* a very high quality generator, but also not a crummy one.
958	+	* Marsaglia xor-shift is cheap and works well enough.  Note:
959	+	* This is manually inlined in its usages in ForkJoinPool to
960	+	* avoid writes inside busy scan loops.
961	+	*/
962	+	final int nextSeed() {
963	+	int r = seed;
964	+	r ^= r << 13;
965	+	r ^= r >>> 17;
966	+	return seed = r ^= r << 5;
967	+	}
968	+
969		// Execution methods
970
971		/**
972	<	* Removes and runs tasks until empty, using local mode
997	<	* ordering.
972	>	* Pops and runs tasks until empty.
973		*/
974	<	final void runLocalTasks() {
975	<	if (base - top < 0) {
976	<	for (ForkJoinTask<?> t; (t = nextLocalTask()) != null; )
974	>	private void popAndExecAll() {
975	>	// A bit faster than repeated pop calls
976	>	ForkJoinTask<?>[] a; int m, s; long j; ForkJoinTask<?> t;
977	>	while ((a = array) != null && (m = a.length - 1) >= 0 &&
978	>	(s = top - 1) - base >= 0 &&
979	>	(t = ((ForkJoinTask<?>)
980	>	U.getObject(a, j = ((m & s) << ASHIFT) + ABASE)))
981	>	!= null) {
982	>	if (U.compareAndSwapObject(a, j, t, null)) {
983	>	top = s;
984		t.doExec();
985	+	}
986		}
987		}
988
989		/**
990	+	* Polls and runs tasks until empty.
991	+	*/
992	+	private void pollAndExecAll() {
993	+	for (ForkJoinTask<?> t; (t = poll()) != null;)
994	+	t.doExec();
995	+	}
996	+
997	+	/**
998	+	* If present, removes from queue and executes the given task, or
999	+	* any other cancelled task. Returns (true) immediately on any CAS
1000	+	* or consistency check failure so caller can retry.
1001	+	*
1002	+	* @return 0 if no progress can be made, else positive
1003	+	* (this unusual convention simplifies use with tryHelpStealer.)
1004	+	*/
1005	+	final int tryRemoveAndExec(ForkJoinTask<?> task) {
1006	+	int stat = 1;
1007	+	boolean removed = false, empty = true;
1008	+	ForkJoinTask<?>[] a; int m, s, b, n;
1009	+	if ((a = array) != null && (m = a.length - 1) >= 0 &&
1010	+	(n = (s = top) - (b = base)) > 0) {
1011	+	for (ForkJoinTask<?> t;;) {           // traverse from s to b
1012	+	int j = ((--s & m) << ASHIFT) + ABASE;
1013	+	t = (ForkJoinTask<?>)U.getObjectVolatile(a, j);
1014	+	if (t == null)                    // inconsistent length
1015	+	break;
1016	+	else if (t == task) {
1017	+	if (s + 1 == top) {           // pop
1018	+	if (!U.compareAndSwapObject(a, j, task, null))
1019	+	break;
1020	+	top = s;
1021	+	removed = true;
1022	+	}
1023	+	else if (base == b)           // replace with proxy
1024	+	removed = U.compareAndSwapObject(a, j, task,
1025	+	new EmptyTask());
1026	+	break;
1027	+	}
1028	+	else if (t.status >= 0)
1029	+	empty = false;
1030	+	else if (s + 1 == top) {          // pop and throw away
1031	+	if (U.compareAndSwapObject(a, j, t, null))
1032	+	top = s;
1033	+	break;
1034	+	}
1035	+	if (--n == 0) {
1036	+	if (!empty && base == b)
1037	+	stat = 0;
1038	+	break;
1039	+	}
1040	+	}
1041	+	}
1042	+	if (removed)
1043	+	task.doExec();
1044	+	return stat;
1045	+	}
1046	+
1047	+	/**
1048		* Executes a top-level task and any local tasks remaining
1049		* after execution.
1009	–	*
1010	–	* @return true unless terminating
1050		*/
1051	<	final boolean runTask(ForkJoinTask<?> t) {
1013	<	boolean alive = true;
1051	>	final void runTask(ForkJoinTask<?> t) {
1052		if (t != null) {
1053		currentSteal = t;
1054		t.doExec();
1055	<	runLocalTasks();
1055	>	if (top != base) {       // process remaining local tasks
1056	>	if (mode == 0)
1057	>	popAndExecAll();
1058	>	else
1059	>	pollAndExecAll();
1060	>	}
1061		++nsteals;
1062		currentSteal = null;
1063		}
1021	–	else if (runState < 0)            // terminating
1022	–	alive = false;
1023	–	return alive;
1064		}
1065
1066		/**
#	Line 1036 | Line 1076 | public class ForkJoinPool extends Abstra
1076		}
1077
1078		/**
1079	<	* Computes next value for random probes.  Scans don't require
1040	<	* a very high quality generator, but also not a crummy one.
1041	<	* Marsaglia xor-shift is cheap and works well enough.  Note:
1042	<	* This is manually inlined in several usages in ForkJoinPool
1043	<	* to avoid writes inside busy scan loops.
1079	>	* Returns true if owned and not known to be blocked.
1080		*/
1081	<	final int nextSeed() {
1082	<	int r = seed;
1083	<	r ^= r << 13;
1084	<	r ^= r >>> 17;
1085	<	r ^= r << 5;
1086	<	return seed = r;
1081	>	final boolean isApparentlyUnblocked() {
1082	>	Thread wt; Thread.State s;
1083	>	return (eventCount >= 0 &&
1084	>	(wt = owner) != null &&
1085	>	(s = wt.getState()) != Thread.State.BLOCKED &&
1086	>	s != Thread.State.WAITING &&
1087	>	s != Thread.State.TIMED_WAITING);
1088	>	}
1089	>
1090	>	/**
1091	>	* If this owned and is not already interrupted, try to
1092	>	* interrupt and/or unpark, ignoring exceptions.
1093	>	*/
1094	>	final void interruptOwner() {
1095	>	Thread wt, p;
1096	>	if ((wt = owner) != null && !wt.isInterrupted()) {
1097	>	try {
1098	>	wt.interrupt();
1099	>	} catch (SecurityException ignore) {
1100	>	}
1101	>	}
1102	>	if ((p = parker) != null)
1103	>	U.unpark(p);
1104		}
1105
1106		// Unsafe mechanics
#	Line 1075 | Line 1128 | public class ForkJoinPool extends Abstra
1128		}
1129
1130		/**
1131	<	* Class for artificial tasks that are used to replace the target
1132	<	* of local joins if they are removed from an interior queue slot
1133	<	* in WorkQueue.tryRemoveAndExec. We don't need the proxy to
1134	<	* actually do anything beyond having a unique identity.
1135	<	*/
1136	<	static final class EmptyTask extends ForkJoinTask<Void> {
1137	<	EmptyTask() { status = ForkJoinTask.NORMAL; } // force done
1138	<	public Void getRawResult() { return null; }
1139	<	public void setRawResult(Void x) {}
1140	<	public boolean exec() { return true; }
1141	<	}
1142	<
1143	<	/**
1144	<	* Per-thread records for (typically non-FJ) threads that submit
1092	<	* to pools. Cureently holds only psuedo-random seed / index that
1093	<	* is used to choose submission queues in method doSubmit. In the
1094	<	* future, this may incorporate a means to implement different
1095	<	* task rejection and resubmission policies.
1131	>	* Per-thread records for threads that submit to pools. Currently
1132	>	* holds only pseudo-random seed / index that is used to choose
1133	>	* submission queues in method doSubmit. In the future, this may
1134	>	* also incorporate a means to implement different task rejection
1135	>	* and resubmission policies.
1136	>	*
1137	>	* Seeds for submitters and workers/workQueues work in basically
1138	>	* the same way but are initialized and updated using slightly
1139	>	* different mechanics. Both are initialized using the same
1140	>	* approach as in class ThreadLocal, where successive values are
1141	>	* unlikely to collide with previous values. This is done during
1142	>	* registration for workers, but requires a separate AtomicInteger
1143	>	* for submitters. Seeds are then randomly modified upon
1144	>	* collisions using xorshifts, which requires a non-zero seed.
1145		*/
1146		static final class Submitter {
1147	<	int seed; // seed for random submission queue selection
1099	<
1100	<	// Heuristic padding to ameliorate unfortunate memory placements
1101	<	int p0, p1, p2, p3, p4, p5, p6, p7, p8, p9, pa, pb, pc, pd, pe;
1102	<
1147	>	int seed;
1148		Submitter() {
1149	<	// Use identityHashCode, forced negative, for seed
1150	<	seed = System.identityHashCode(Thread.currentThread()) | (1 << 31);
1106	<	}
1107	<
1108	<	/**
1109	<	* Computes next value for random probes.  Like method
1110	<	* WorkQueue.nextSeed, this is manually inlined in several
1111	<	* usages to avoid writes inside busy loops.
1112	<	*/
1113	<	final int nextSeed() {
1114	<	int r = seed;
1115	<	r ^= r << 13;
1116	<	r ^= r >>> 17;
1117	<	return seed = r ^= r << 5;
1149	>	int s = nextSubmitterSeed.getAndAdd(SEED_INCREMENT);
1150	>	seed = (s == 0) ? 1 : s; // ensure non-zero
1151		}
1152		}
1153
#	Line 1123 | Line 1156 | public class ForkJoinPool extends Abstra
1156		public Submitter initialValue() { return new Submitter(); }
1157		}
1158
1159	+	// static fields (initialized in static initializer below)
1160	+
1161	+	/**
1162	+	* Creates a new ForkJoinWorkerThread. This factory is used unless
1163	+	* overridden in ForkJoinPool constructors.
1164	+	*/
1165	+	public static final ForkJoinWorkerThreadFactory
1166	+	defaultForkJoinWorkerThreadFactory;
1167	+
1168	+
1169	+	/** Property prefix for constructing common pool */
1170	+	private static final String propPrefix =
1171	+	"java.util.concurrent.ForkJoinPool.common.";
1172	+
1173	+	/**
1174	+	* Common (static) pool. Non-null for public use unless a static
1175	+	* construction exception, but internal usages must null-check on
1176	+	* use.
1177	+	*/
1178	+	static final ForkJoinPool commonPool;
1179	+
1180	+	/**
1181	+	* Common pool parallelism. Must equal commonPool.parallelism.
1182	+	*/
1183	+	static final int commonPoolParallelism;
1184	+
1185		/**
1186	<	* Per-thread submission bookeeping. Shared across all pools
1186	>	* Generator for assigning sequence numbers as pool names.
1187	>	*/
1188	>	private static final AtomicInteger poolNumberGenerator;
1189	>
1190	>	/**
1191	>	* Generator for initial hashes/seeds for submitters. Accessed by
1192	>	* Submitter class constructor.
1193	>	*/
1194	>	static final AtomicInteger nextSubmitterSeed;
1195	>
1196	>	/**
1197	>	* Permission required for callers of methods that may start or
1198	>	* kill threads.
1199	>	*/
1200	>	private static final RuntimePermission modifyThreadPermission;
1201	>
1202	>	/**
1203	>	* Per-thread submission bookkeeping. Shared across all pools
1204		* to reduce ThreadLocal pollution and because random motion
1205		* to avoid contention in one pool is likely to hold for others.
1206		*/
1207	<	static final ThreadSubmitter submitters = new ThreadSubmitter();
1207	>	private static final ThreadSubmitter submitters;
1208	>
1209	>	// static constants
1210	>
1211	>	/**
1212	>	* Initial timeout value (in nanoseconds) for the thread triggering
1213	>	* quiescence to park waiting for new work. On timeout, the thread
1214	>	* will instead try to shrink the number of workers.
1215	>	*/
1216	>	private static final long IDLE_TIMEOUT      = 1000L * 1000L * 1000L; // 1sec
1217	>
1218	>	/**
1219	>	* Timeout value when there are more threads than parallelism level
1220	>	*/
1221	>	private static final long FAST_IDLE_TIMEOUT =  100L * 1000L * 1000L;
1222	>
1223	>	/**
1224	>	* The maximum stolen->joining link depth allowed in method
1225	>	* tryHelpStealer.  Must be a power of two. This value also
1226	>	* controls the maximum number of times to try to help join a task
1227	>	* without any apparent progress or change in pool state before
1228	>	* giving up and blocking (see awaitJoin).  Depths for legitimate
1229	>	* chains are unbounded, but we use a fixed constant to avoid
1230	>	* (otherwise unchecked) cycles and to bound staleness of
1231	>	* traversal parameters at the expense of sometimes blocking when
1232	>	* we could be helping.
1233	>	*/
1234	>	private static final int MAX_HELP = 64;
1235
1236		/**
1237	<	* Top-level runloop for workers
1237	>	* Secondary time-based bound (in nanosecs) for helping attempts
1238	>	* before trying compensated blocking in awaitJoin. Used in
1239	>	* conjunction with MAX_HELP to reduce variance due to different
1240	>	* polling rates associated with different helping options. The
1241	>	* value should roughly approximate the time required to create
1242	>	* and/or activate a worker thread.
1243	>	*/
1244	>	private static final long COMPENSATION_DELAY = 1L << 18; // ~0.25 millisec
1245	>
1246	>	/**
1247	>	* Increment for seed generators. See class ThreadLocal for
1248	>	* explanation.
1249	>	*/
1250	>	private static final int SEED_INCREMENT = 0x61c88647;
1251	>
1252	>	/**
1253	>	* Bits and masks for control variables
1254	>	*
1255	>	* Field ctl is a long packed with:
1256	>	* AC: Number of active running workers minus target parallelism (16 bits)
1257	>	* TC: Number of total workers minus target parallelism (16 bits)
1258	>	* ST: true if pool is terminating (1 bit)
1259	>	* EC: the wait count of top waiting thread (15 bits)
1260	>	* ID: poolIndex of top of Treiber stack of waiters (16 bits)
1261	>	*
1262	>	* When convenient, we can extract the upper 32 bits of counts and
1263	>	* the lower 32 bits of queue state, u = (int)(ctl >>> 32) and e =
1264	>	* (int)ctl.  The ec field is never accessed alone, but always
1265	>	* together with id and st. The offsets of counts by the target
1266	>	* parallelism and the positionings of fields makes it possible to
1267	>	* perform the most common checks via sign tests of fields: When
1268	>	* ac is negative, there are not enough active workers, when tc is
1269	>	* negative, there are not enough total workers, and when e is
1270	>	* negative, the pool is terminating.  To deal with these possibly
1271	>	* negative fields, we use casts in and out of "short" and/or
1272	>	* signed shifts to maintain signedness.
1273	>	*
1274	>	* When a thread is queued (inactivated), its eventCount field is
1275	>	* set negative, which is the only way to tell if a worker is
1276	>	* prevented from executing tasks, even though it must continue to
1277	>	* scan for them to avoid queuing races. Note however that
1278	>	* eventCount updates lag releases so usage requires care.
1279	>	*
1280	>	* Field runState is an int packed with:
1281	>	* SHUTDOWN: true if shutdown is enabled (1 bit)
1282	>	* SEQ:  a sequence number updated upon (de)registering workers (30 bits)
1283	>	* INIT: set true after workQueues array construction (1 bit)
1284	>	*
1285	>	* The sequence number enables simple consistency checks:
1286	>	* Staleness of read-only operations on the workQueues array can
1287	>	* be checked by comparing runState before vs after the reads.
1288		*/
1136	–	final void runWorker(ForkJoinWorkerThread wt) {
1137	–	// Initialize queue array and seed in this thread
1138	–	WorkQueue w = wt.workQueue;
1139	–	w.growArray(false);
1140	–	// Same initial hash as Submitters
1141	–	w.seed = System.identityHashCode(Thread.currentThread()) | (1 << 31);
1289
1290	<	do {} while (w.runTask(scan(w)));
1290	>	// bit positions/shifts for fields
1291	>	private static final int  AC_SHIFT   = 48;
1292	>	private static final int  TC_SHIFT   = 32;
1293	>	private static final int  ST_SHIFT   = 31;
1294	>	private static final int  EC_SHIFT   = 16;
1295	>
1296	>	// bounds
1297	>	private static final int  SMASK      = 0xffff;  // short bits
1298	>	private static final int  MAX_CAP    = 0x7fff;  // max #workers - 1
1299	>	private static final int  SQMASK     = 0xfffe;  // even short bits
1300	>	private static final int  SHORT_SIGN = 1 << 15;
1301	>	private static final int  INT_SIGN   = 1 << 31;
1302	>
1303	>	// masks
1304	>	private static final long STOP_BIT   = 0x0001L << ST_SHIFT;
1305	>	private static final long AC_MASK    = ((long)SMASK) << AC_SHIFT;
1306	>	private static final long TC_MASK    = ((long)SMASK) << TC_SHIFT;
1307	>
1308	>	// units for incrementing and decrementing
1309	>	private static final long TC_UNIT    = 1L << TC_SHIFT;
1310	>	private static final long AC_UNIT    = 1L << AC_SHIFT;
1311	>
1312	>	// masks and units for dealing with u = (int)(ctl >>> 32)
1313	>	private static final int  UAC_SHIFT  = AC_SHIFT - 32;
1314	>	private static final int  UTC_SHIFT  = TC_SHIFT - 32;
1315	>	private static final int  UAC_MASK   = SMASK << UAC_SHIFT;
1316	>	private static final int  UTC_MASK   = SMASK << UTC_SHIFT;
1317	>	private static final int  UAC_UNIT   = 1 << UAC_SHIFT;
1318	>	private static final int  UTC_UNIT   = 1 << UTC_SHIFT;
1319	>
1320	>	// masks and units for dealing with e = (int)ctl
1321	>	private static final int E_MASK      = 0x7fffffff; // no STOP_BIT
1322	>	private static final int E_SEQ       = 1 << EC_SHIFT;
1323	>
1324	>	// runState bits
1325	>	private static final int SHUTDOWN    = 1 << 31;
1326	>
1327	>	// access mode for WorkQueue
1328	>	static final int LIFO_QUEUE          =  0;
1329	>	static final int FIFO_QUEUE          =  1;
1330	>	static final int SHARED_QUEUE        = -1;
1331	>
1332	>	// Instance fields
1333	>
1334	>	/*
1335	>	* Field layout order in this class tends to matter more than one
1336	>	* would like. Runtime layout order is only loosely related to
1337	>	* declaration order and may differ across JVMs, but the following
1338	>	* empirically works OK on current JVMs.
1339	>	*/
1340	>
1341	>	volatile long stealCount;                  // collects worker counts
1342	>	volatile long ctl;                         // main pool control
1343	>	final int parallelism;                     // parallelism level
1344	>	final int localMode;                       // per-worker scheduling mode
1345	>	volatile int nextWorkerNumber;             // to create worker name string
1346	>	final int submitMask;                      // submit queue index bound
1347	>	int nextSeed;                              // for initializing worker seeds
1348	>	volatile int mainLock;                     // spinlock for array updates
1349	>	volatile int runState;                     // shutdown status and seq
1350	>	WorkQueue[] workQueues;                    // main registry
1351	>	final ForkJoinWorkerThreadFactory factory; // factory for new workers
1352	>	final Thread.UncaughtExceptionHandler ueh; // per-worker UEH
1353	>	final String workerNamePrefix;             // to create worker name string
1354	>
1355	>	/*
1356	>	* Mechanics for main lock protecting worker array updates.  Uses
1357	>	* the same strategy as ConcurrentHashMap bins -- a spinLock for
1358	>	* normal cases, but falling back to builtin lock when (rarely)
1359	>	* needed.  See internal ConcurrentHashMap documentation for
1360	>	* explanation.
1361	>	*/
1362	>
1363	>	static final int LOCK_WAITING = 2; // bit to indicate need for signal
1364	>	static final int MAX_LOCK_SPINS = 1 << 8;
1365	>
1366	>	private void tryAwaitMainLock() {
1367	>	int spins = MAX_LOCK_SPINS, r = 0, h;
1368	>	while (((h = mainLock) & 1) != 0) {
1369	>	if (r == 0)
1370	>	r = ThreadLocalRandom.current().nextInt(); // randomize spins
1371	>	else if (spins >= 0) {
1372	>	r ^= r << 1; r ^= r >>> 3; r ^= r << 10; // xorshift
1373	>	if (r >= 0)
1374	>	--spins;
1375	>	}
1376	>	else if (U.compareAndSwapInt(this, MAINLOCK, h, h | LOCK_WAITING)) {
1377	>	synchronized (this) {
1378	>	if ((mainLock & LOCK_WAITING) != 0) {
1379	>	try {
1380	>	wait();
1381	>	} catch (InterruptedException ie) {
1382	>	Thread.currentThread().interrupt();
1383	>	}
1384	>	}
1385	>	else
1386	>	notifyAll(); // possibly won race vs signaller
1387	>	}
1388	>	break;
1389	>	}
1390	>	}
1391		}
1392
1393	<	// Creating, registering and deregistering workers
1393	>	//  Creating, registering, and deregistering workers
1394
1395		/**
1396		* Tries to create and start a worker
1397		*/
1398		private void addWorker() {
1399		Throwable ex = null;
1400	<	ForkJoinWorkerThread w = null;
1400	>	ForkJoinWorkerThread wt = null;
1401		try {
1402	<	if ((w = factory.newThread(this)) != null) {
1403	<	w.start();
1402	>	if ((wt = factory.newThread(this)) != null) {
1403	>	wt.start();
1404		return;
1405		}
1406		} catch (Throwable e) {
1407		ex = e;
1408		}
1409	<	deregisterWorker(w, ex);
1409	>	deregisterWorker(wt, ex); // adjust counts etc on failure
1410		}
1411
1412		/**
#	Line 1169 | Line 1416 | public class ForkJoinPool extends Abstra
1416		* ForkJoinWorkerThread.
1417		*/
1418		final String nextWorkerName() {
1419	<	return workerNamePrefix.concat
1420	<	(Integer.toString(nextWorkerNumber.addAndGet(1)));
1419	>	int n;
1420	>	do {} while(!U.compareAndSwapInt(this, NEXTWORKERNUMBER,
1421	>	n = nextWorkerNumber, ++n));
1422	>	return workerNamePrefix.concat(Integer.toString(n));
1423		}
1424
1425		/**
1426	<	* Callback from ForkJoinWorkerThread constructor to establish and
1427	<	* record its WorkQueue.
1426	>	* Callback from ForkJoinWorkerThread constructor to establish its
1427	>	* poolIndex and record its WorkQueue. To avoid scanning bias due
1428	>	* to packing entries in front of the workQueues array, we treat
1429	>	* the array as a simple power-of-two hash table using per-thread
1430	>	* seed as hash, expanding as needed.
1431		*
1432	<	* @param wt the worker thread
1432	>	* @param w the worker's queue
1433		*/
1434	<	final void registerWorker(ForkJoinWorkerThread wt) {
1435	<	WorkQueue w = wt.workQueue;
1436	<	ReentrantLock lock = this.lock;
1185	<	lock.lock();
1434	>	final void registerWorker(WorkQueue w) {
1435	>	while (!U.compareAndSwapInt(this, MAINLOCK, 0, 1))
1436	>	tryAwaitMainLock();
1437		try {
1438	<	int k = nextPoolIndex;
1439	<	WorkQueue[] ws = workQueues;
1440	<	if (ws != null) {                       // ignore on shutdown
1441	<	int n = ws.length;
1442	<	if (k < 0 || (k & 1) == 0 || k >= n || ws[k] != null) {
1443	<	for (k = 1; k < n && ws[k] != null; k += 2)
1444	<	;                           // workers are at odd indices
1445	<	if (k >= n)                     // resize
1446	<	workQueues = ws = Arrays.copyOf(ws, n << 1);
1447	<	}
1448	<	w.poolIndex = k;
1449	<	w.eventCount = ~(k >>> 1) & SMASK;  // Set up wait count
1450	<	ws[k] = w;                          // record worker
1451	<	nextPoolIndex = k + 2;
1452	<	int rs = runState;
1453	<	int m = rs & SMASK;                 // recalculate runState mask
1454	<	if (k > m)
1455	<	m = (m << 1) + 1;
1456	<	runState = (rs & SHUTDOWN) | ((rs + RS_SEQ) & RS_SEQ_MASK) | m;
1438	>	WorkQueue[] ws;
1439	>	if ((ws = workQueues) == null)
1440	>	ws = workQueues = new WorkQueue[submitMask + 1];
1441	>	if (w != null) {
1442	>	int rs, n =  ws.length, m = n - 1;
1443	>	int s = nextSeed += SEED_INCREMENT; // rarely-colliding sequence
1444	>	w.seed = (s == 0) ? 1 : s;          // ensure non-zero seed
1445	>	int r = (s << 1) | 1;               // use odd-numbered indices
1446	>	if (ws[r &= m] != null) {           // collision
1447	>	int probes = 0;                 // step by approx half size
1448	>	int step = (n <= 4) ? 2 : ((n >>> 1) & SQMASK) + 2;
1449	>	while (ws[r = (r + step) & m] != null) {
1450	>	if (++probes >= n) {
1451	>	workQueues = ws = Arrays.copyOf(ws, n <<= 1);
1452	>	m = n - 1;
1453	>	probes = 0;
1454	>	}
1455	>	}
1456	>	}
1457	>	w.eventCount = w.poolIndex = r;     // establish before recording
1458	>	ws[r] = w;                          // also update seq
1459	>	runState = ((rs = runState) & SHUTDOWN) | ((rs + 2) & ~SHUTDOWN);
1460		}
1461		} finally {
1462	<	lock.unlock();
1462	>	if (!U.compareAndSwapInt(this, MAINLOCK, 1, 0)) {
1463	>	mainLock = 0;
1464	>	synchronized (this) { notifyAll(); };
1465	>	}
1466		}
1467	+
1468		}
1469
1470		/**
1471	<	* Final callback from terminating worker, as well as failure to
1472	<	* construct or start a worker in addWorker.  Removes record of
1471	>	* Final callback from terminating worker, as well as upon failure
1472	>	* to construct or start a worker in addWorker.  Removes record of
1473		* worker from array, and adjusts counts. If pool is shutting
1474		* down, tries to complete termination.
1475		*
#	Line 1222 | Line 1480 | public class ForkJoinPool extends Abstra
1480		WorkQueue w = null;
1481		if (wt != null && (w = wt.workQueue) != null) {
1482		w.runState = -1;                // ensure runState is set
1483	<	stealCount.getAndAdd(w.totalSteals + w.nsteals);
1483	>	long steals = w.totalSteals + w.nsteals, sc;
1484	>	do {} while(!U.compareAndSwapLong(this, STEALCOUNT,
1485	>	sc = stealCount, sc + steals));
1486		int idx = w.poolIndex;
1487	<	ReentrantLock lock = this.lock;
1488	<	lock.lock();
1489	<	try {                           // remove record from array
1487	>	while (!U.compareAndSwapInt(this, MAINLOCK, 0, 1))
1488	>	tryAwaitMainLock();
1489	>	try {
1490		WorkQueue[] ws = workQueues;
1491		if (ws != null && idx >= 0 && idx < ws.length && ws[idx] == w)
1492	<	ws[nextPoolIndex = idx] = null;
1492	>	ws[idx] = null;
1493		} finally {
1494	<	lock.unlock();
1494	>	if (!U.compareAndSwapInt(this, MAINLOCK, 1, 0)) {
1495	>	mainLock = 0;
1496	>	synchronized (this) { notifyAll(); };
1497	>	}
1498		}
1499		}
1500
#	Line 1241 | Line 1504 | public class ForkJoinPool extends Abstra
1504		((c - TC_UNIT) & TC_MASK) |
1505		(c & ~(AC_MASK|TC_MASK)))));
1506
1507	<	if (!tryTerminate(false) && w != null) {
1507	>	if (!tryTerminate(false, false) && w != null) {
1508		w.cancelAll();                  // cancel remaining tasks
1509		if (w.array != null)            // suppress signal if never ran
1510		signalWork();               // wake up or create replacement
1511	+	if (ex == null)                 // help clean refs on way out
1512	+	ForkJoinTask.helpExpungeStaleExceptions();
1513		}
1514
1515		if (ex != null)                     // rethrow
1516		U.throwException(ex);
1517		}
1518
1519	+	// Submissions
1520	+
1521		/**
1522	<	* Tries to add and register a new queue at the given index.
1522	>	* Unless shutting down, adds the given task to a submission queue
1523	>	* at submitter's current queue index (modulo submission
1524	>	* range). If no queue exists at the index, one is created.  If
1525	>	* the queue is busy, another index is randomly chosen. The
1526	>	* submitMask bounds the effective number of queues to the
1527	>	* (nearest power of two for) parallelism level.
1528		*
1529	<	* @param idx the workQueues array index to register the queue
1530	<	* @return the queue, or null if could not add because could
1531	<	* not acquire lock or idx is unusable
1532	<	*/
1533	<	private WorkQueue tryAddSharedQueue(int idx) {
1534	<	WorkQueue q = null;
1535	<	ReentrantLock lock = this.lock;
1536	<	if (idx >= 0 && (idx & 1) == 0 && !lock.isLocked()) {
1537	<	// create queue outside of lock but only if apparently free
1538	<	WorkQueue nq = new WorkQueue(null, SHARED_QUEUE);
1539	<	if (lock.tryLock()) {
1529	>	* @param task the task. Caller must ensure non-null.
1530	>	*/
1531	>	private void doSubmit(ForkJoinTask<?> task) {
1532	>	Submitter s = submitters.get();
1533	>	for (int r = s.seed, m = submitMask;;) {
1534	>	WorkQueue[] ws; WorkQueue q;
1535	>	int k = r & m & SQMASK;          // use only even indices
1536	>	if (runState < 0)
1537	>	throw new RejectedExecutionException(); // shutting down
1538	>	else if ((ws = workQueues) == null || ws.length <= k) {
1539	>	while (!U.compareAndSwapInt(this, MAINLOCK, 0, 1))
1540	>	tryAwaitMainLock();
1541		try {
1542	<	WorkQueue[] ws = workQueues;
1543	<	if (ws != null && idx < ws.length) {
1544	<	if ((q = ws[idx]) == null) {
1545	<	int rs;         // update runState seq
1546	<	ws[idx] = q = nq;
1547	<	runState = (((rs = runState) & SHUTDOWN) |
1548	<	((rs + RS_SEQ) & ~SHUTDOWN));
1549	<	}
1542	>	if (workQueues == null)
1543	>	workQueues = new WorkQueue[submitMask + 1];
1544	>	} finally {
1545	>	if (!U.compareAndSwapInt(this, MAINLOCK, 1, 0)) {
1546	>	mainLock = 0;
1547	>	synchronized (this) { notifyAll(); };
1548	>	}
1549	>	}
1550	>	}
1551	>	else if ((q = ws[k]) == null) {  // create new queue
1552	>	WorkQueue nq = new WorkQueue(this, null, SHARED_QUEUE);
1553	>	while (!U.compareAndSwapInt(this, MAINLOCK, 0, 1))
1554	>	tryAwaitMainLock();
1555	>	try {
1556	>	int rs = runState;       // to update seq
1557	>	if (ws == workQueues && ws[k] == null) {
1558	>	ws[k] = nq;
1559	>	runState = ((rs & SHUTDOWN) | ((rs + 2) & ~SHUTDOWN));
1560		}
1561		} finally {
1562	<	lock.unlock();
1562	>	if (!U.compareAndSwapInt(this, MAINLOCK, 1, 0)) {
1563	>	mainLock = 0;
1564	>	synchronized (this) { notifyAll(); };
1565	>	}
1566		}
1567		}
1568	+	else if (q.trySharedPush(task)) {
1569	+	signalWork();
1570	+	return;
1571	+	}
1572	+	else if (m > 1) {                // move to a different index
1573	+	r ^= r << 13;                // same xorshift as WorkQueues
1574	+	r ^= r >>> 17;
1575	+	s.seed = r ^= r << 5;
1576	+	}
1577	+	else
1578	+	Thread.yield();              // yield if no alternatives
1579		}
1283	–	return q;
1580		}
1581
1582	+	/**
1583	+	* Submits the given (non-null) task to the common pool, if possible.
1584	+	*/
1585	+	static void submitToCommonPool(ForkJoinTask<?> task) {
1586	+	ForkJoinPool p;
1587	+	if ((p = commonPool) == null)
1588	+	throw new RejectedExecutionException("Common Pool Unavailable");
1589	+	p.doSubmit(task);
1590	+	}
1591	+
1592	+	/**
1593	+	* Returns true if caller is (or may be) submitter to the common
1594	+	* pool, and not all workers are active, and there appear to be
1595	+	* tasks in the associated submission queue.
1596	+	*/
1597	+	static boolean canHelpCommonPool() {
1598	+	ForkJoinPool p; WorkQueue[] ws; WorkQueue q;
1599	+	int k = submitters.get().seed & SQMASK;
1600	+	return ((p = commonPool) != null &&
1601	+	(int)(p.ctl >> AC_SHIFT) < 0 &&
1602	+	(ws = p.workQueues) != null &&
1603	+	ws.length > (k &= p.submitMask) &&
1604	+	(q = ws[k]) != null &&
1605	+	q.top - q.base > 0);
1606	+	}
1607	+
1608	+	/**
1609	+	* Returns true if the given task was submitted to common pool
1610	+	* and has not yet commenced execution, and is available for
1611	+	* removal according to execution policies; if so removing the
1612	+	* submission from the pool.
1613	+	*
1614	+	* @param task the task
1615	+	* @return true if successful
1616	+	*/
1617	+	static boolean tryUnsubmitFromCommonPool(ForkJoinTask<?> task) {
1618	+	// Peek, looking for task and eligibility before
1619	+	// using trySharedUnpush to actually take it under lock
1620	+	ForkJoinPool p; WorkQueue[] ws; WorkQueue q;
1621	+	ForkJoinTask<?>[] a; int s;
1622	+	int k = submitters.get().seed & SQMASK;
1623	+	return ((p = commonPool) != null &&
1624	+	(int)(p.ctl >> AC_SHIFT) < 0 &&
1625	+	(ws = p.workQueues) != null &&
1626	+	ws.length > (k &= p.submitMask) &&
1627	+	(q = ws[k]) != null &&
1628	+	(a = q.array) != null &&
1629	+	(s = q.top - 1) - q.base >= 0 &&
1630	+	s >= 0 && s < a.length &&
1631	+	a[s] == task &&
1632	+	q.trySharedUnpush(task));
1633	+	}
1634	+
1635	+	/**
1636	+	* Tries to pop a task from common pool with given root
1637	+	*/
1638	+	static ForkJoinTask<?> popCCFromCommonPool(CountedCompleter<?> root) {
1639	+	ForkJoinPool p; WorkQueue[] ws; WorkQueue q;
1640	+	ForkJoinTask<?> t;
1641	+	int k = submitters.get().seed & SQMASK;
1642	+	if (root != null &&
1643	+	(p = commonPool) != null &&
1644	+	(int)(p.ctl >> AC_SHIFT) < 0 &&
1645	+	(ws = p.workQueues) != null &&
1646	+	ws.length > (k &= p.submitMask) &&
1647	+	(q = ws[k]) != null && q.top - q.base > 0 &&
1648	+	root.status < 0 &&
1649	+	(t = q.sharedPopCC(root)) != null)
1650	+	return t;
1651	+	return null;
1652	+	}
1653	+
1654	+
1655		// Maintaining ctl counts
1656
1657		/**
#	Line 1294 | Line 1663 | public class ForkJoinPool extends Abstra
1663		}
1664
1665		/**
1666	<	* Activates or creates a worker.
1666	>	* Tries to create one or activate one or more workers if too few are active.
1667		*/
1668		final void signalWork() {
1669	<	/*
1670	<	* The while condition is true if: (there is are too few total
1671	<	* workers OR there is at least one waiter) AND (there are too
1672	<	* few active workers OR the pool is terminating).  The value
1673	<	* of e distinguishes the remaining cases: zero (no waiters)
1674	<	* for create, negative if terminating (in which case do
1675	<	* nothing), else release a waiter. The secondary checks for
1676	<	* release (non-null array etc) can fail if the pool begins
1677	<	* terminating after the test, and don't impose any added cost
1678	<	* because JVMs must perform null and bounds checks anyway.
1679	<	*/
1680	<	long c; int e, u;
1681	<	while ((((e = (int)(c = ctl)) | (u = (int)(c >>> 32))) &
1682	<	(INT_SIGN|SHORT_SIGN)) == (INT_SIGN|SHORT_SIGN)) {
1314	<	WorkQueue[] ws = workQueues; int i; WorkQueue w; Thread p;
1315	<	if (e == 0) {                    // add a new worker
1316	<	if (U.compareAndSwapLong
1317	<	(this, CTL, c, (long)(((u + UTC_UNIT) & UTC_MASK) |
1318	<	((u + UAC_UNIT) & UAC_MASK)) << 32)) {
1319	<	addWorker();
1320	<	break;
1669	>	long c; int u;
1670	>	while ((u = (int)((c = ctl) >>> 32)) < 0) {     // too few active
1671	>	WorkQueue[] ws = workQueues; int e, i; WorkQueue w; Thread p;
1672	>	if ((e = (int)c) > 0) {                     // at least one waiting
1673	>	if (ws != null && (i = e & SMASK) < ws.length &&
1674	>	(w = ws[i]) != null && w.eventCount == (e | INT_SIGN)) {
1675	>	long nc = (((long)(w.nextWait & E_MASK)) |
1676	>	((long)(u + UAC_UNIT) << 32));
1677	>	if (U.compareAndSwapLong(this, CTL, c, nc)) {
1678	>	w.eventCount = (e + E_SEQ) & E_MASK;
1679	>	if ((p = w.parker) != null)
1680	>	U.unpark(p);                // activate and release
1681	>	break;
1682	>	}
1683		}
1684	<	}
1323	<	else if (e > 0 && ws != null &&
1324	<	(i = ((~e << 1) | 1) & SMASK) < ws.length &&
1325	<	(w = ws[i]) != null &&
1326	<	w.eventCount == (e | INT_SIGN)) {
1327	<	if (U.compareAndSwapLong
1328	<	(this, CTL, c, (((long)(w.nextWait & E_MASK)) |
1329	<	((long)(u + UAC_UNIT) << 32)))) {
1330	<	w.eventCount = (e + E_SEQ) & E_MASK;
1331	<	if ((p = w.parker) != null)
1332	<	U.unpark(p);         // release a waiting worker
1684	>	else
1685		break;
1334	–	}
1335	–	}
1336	–	else
1337	–	break;
1338	–	}
1339	–	}
1340	–
1341	–	/**
1342	–	* Tries to decrement active count (sometimes implicitly) and
1343	–	* possibly release or create a compensating worker in preparation
1344	–	* for blocking. Fails on contention or termination.
1345	–	*
1346	–	* @return true if the caller can block, else should recheck and retry
1347	–	*/
1348	–	final boolean tryCompensate() {
1349	–	WorkQueue[] ws; WorkQueue w; Thread p;
1350	–	int pc = parallelism, e, u, ac, tc, i;
1351	–	long c = ctl;
1352	–
1353	–	if ((e = (int)c) >= 0) {
1354	–	if ((ac = ((u = (int)(c >>> 32)) >> UAC_SHIFT)) <= 0 &&
1355	–	e != 0 && (ws = workQueues) != null &&
1356	–	(i = ((~e << 1) | 1) & SMASK) < ws.length &&
1357	–	(w = ws[i]) != null) {
1358	–	if (w.eventCount == (e | INT_SIGN) &&
1359	–	U.compareAndSwapLong
1360	–	(this, CTL, c, ((long)(w.nextWait & E_MASK) |
1361	–	(c & (AC_MASK|TC_MASK))))) {
1362	–	w.eventCount = (e + E_SEQ) & E_MASK;
1363	–	if ((p = w.parker) != null)
1364	–	U.unpark(p);
1365	–	return true;             // release an idle worker
1366	–	}
1367	–	}
1368	–	else if ((tc = (short)(u >>> UTC_SHIFT)) >= 0 && ac + pc > 1) {
1369	–	long nc = ((c - AC_UNIT) & AC_MASK) | (c & ~AC_MASK);
1370	–	if (U.compareAndSwapLong(this, CTL, c, nc))
1371	–	return true;             // no compensation needed
1686		}
1687	<	else if (tc + pc < MAX_ID) {
1688	<	long nc = ((c + TC_UNIT) & TC_MASK) | (c & ~TC_MASK);
1687	>	else if (e == 0 && (u & SHORT_SIGN) != 0) { // too few total
1688	>	long nc = (long)(((u + UTC_UNIT) & UTC_MASK) |
1689	>	((u + UAC_UNIT) & UAC_MASK)) << 32;
1690		if (U.compareAndSwapLong(this, CTL, c, nc)) {
1691		addWorker();
1692	<	return true;             // create replacement
1692	>	break;
1693		}
1694		}
1695	+	else
1696	+	break;
1697		}
1381	–	return false;
1698		}
1699
1700	<	// Submissions
1700	>	// Scanning for tasks
1701
1702		/**
1703	<	* Unless shutting down, adds the given task to a submission queue
1388	<	* at submitter's current queue index. If no queue exists at the
1389	<	* index, one is created unless pool lock is busy.  If the queue
1390	<	* and/or lock are busy, another index is randomly chosen.
1703	>	* Top-level runloop for workers, called by ForkJoinWorkerThread.run.
1704		*/
1705	<	private void doSubmit(ForkJoinTask<?> task) {
1706	<	if (task == null)
1707	<	throw new NullPointerException();
1395	<	Submitter s = submitters.get();
1396	<	for (int r = s.seed;;) {
1397	<	WorkQueue q; int k;
1398	<	int rs = runState, m = rs & SMASK;
1399	<	WorkQueue[] ws = workQueues;
1400	<	if (rs < 0 || ws == null)   // shutting down
1401	<	throw new RejectedExecutionException();
1402	<	if (ws.length > m &&        // k must be at index
1403	<	((q = ws[k = (r << 1) & m]) != null ||
1404	<	(q = tryAddSharedQueue(k)) != null) &&
1405	<	q.trySharedPush(task)) {
1406	<	signalWork();
1407	<	return;
1408	<	}
1409	<	r ^= r << 13;               // xorshift seed to new position
1410	<	r ^= r >>> 17;
1411	<	if (((s.seed = r ^= r << 5) & m) == 0)
1412	<	Thread.yield();         // occasionally yield if busy
1413	<	}
1705	>	final void runWorker(WorkQueue w) {
1706	>	w.growArray(false);         // initialize queue array in this thread
1707	>	do { w.runTask(scan(w)); } while (w.runState >= 0);
1708		}
1709
1416	–
1417	–	// Scanning for tasks
1418	–
1710		/**
1711		* Scans for and, if found, returns one task, else possibly
1712		* inactivates the worker. This method operates on single reads of
1713	<	* volatile state and is designed to be re-invoked continuously in
1714	<	* part because it returns upon detecting inconsistencies,
1713	>	* volatile state and is designed to be re-invoked continuously,
1714	>	* in part because it returns upon detecting inconsistencies,
1715		* contention, or state changes that indicate possible success on
1716		* re-invocation.
1717		*
1718	<	* The scan searches for tasks across queues, randomly selecting
1719	<	* the first #queues probes, favoring steals 2:1 over submissions
1720	<	* (by exploiting even/odd indexing), and then performing a
1721	<	* circular sweep of all queues.  The scan terminates upon either
1722	<	* finding a non-empty queue, or completing a full sweep. If the
1723	<	* worker is not inactivated, it takes and returns a task from
1724	<	* this queue.  On failure to find a task, we take one of the
1725	<	* following actions, after which the caller will retry calling
1726	<	* this method unless terminated.
1718	>	* The scan searches for tasks across a random permutation of
1719	>	* queues (starting at a random index and stepping by a random
1720	>	* relative prime, checking each at least once).  The scan
1721	>	* terminates upon either finding a non-empty queue, or completing
1722	>	* the sweep. If the worker is not inactivated, it takes and
1723	>	* returns a task from this queue.  On failure to find a task, we
1724	>	* take one of the following actions, after which the caller will
1725	>	* retry calling this method unless terminated.
1726	>	*
1727	>	* * If pool is terminating, terminate the worker.
1728		*
1729		* * If not a complete sweep, try to release a waiting worker.  If
1730		* the scan terminated because the worker is inactivated, then the
#	Line 1441 | Line 1733 | public class ForkJoinPool extends Abstra
1733		* another worker, but with same net effect. Releasing in other
1734		* cases as well ensures that we have enough workers running.
1735		*
1444	–	* * If the caller has run a task since the the last empty scan,
1445	–	* return (to allow rescan) if other workers are not also yet
1446	–	* enqueued.  Field WorkQueue.rescans counts down on each scan to
1447	–	* ensure eventual inactivation, and occasional calls to
1448	–	* Thread.yield to help avoid interference with more useful
1449	–	* activities on the system.
1450	–	*
1451	–	* * If pool is terminating, terminate the worker.
1452	–	*
1736		* * If not already enqueued, try to inactivate and enqueue the
1737	<	* worker on wait queue.
1737	>	* worker on wait queue. Or, if inactivating has caused the pool
1738	>	* to be quiescent, relay to idleAwaitWork to check for
1739	>	* termination and possibly shrink pool.
1740	>	*
1741	>	* * If already inactive, and the caller has run a task since the
1742	>	* last empty scan, return (to allow rescan) unless others are
1743	>	* also inactivated.  Field WorkQueue.rescans counts down on each
1744	>	* scan to ensure eventual inactivation and blocking.
1745		*
1746	<	* * If already enqueued and none of the above apply, either park
1747	<	* awaiting signal, or if this is the most recent waiter and pool
1458	<	* is quiescent, relay to idleAwaitWork to check for termination
1459	<	* and possibly shrink pool.
1746	>	* * If already enqueued and none of the above apply, park
1747	>	* awaiting signal,
1748		*
1749		* @param w the worker (via its WorkQueue)
1750	<	* @return a task or null of none found
1750	>	* @return a task or null if none found
1751		*/
1752		private final ForkJoinTask<?> scan(WorkQueue w) {
1753	<	boolean swept = false;                 // true after full empty scan
1754	<	WorkQueue[] ws;                        // volatile read order matters
1755	<	int r = w.seed, ec = w.eventCount;     // ec is negative if inactive
1756	<	int rs = runState, m = rs & SMASK;
1757	<	if ((ws = workQueues) != null && ws.length > m) {
1758	<	ForkJoinTask<?> task = null;
1759	<	for (int k = 0, j = -2 - m; ; ++j) {
1760	<	WorkQueue q; int b;
1761	<	if (j < 0) {                   // random probes while j negative
1762	<	r ^= r << 13; r ^= r >>> 17; k = (r ^= r << 5) | (j & 1);
1763	<	}                              // worker (not submit) for odd j
1764	<	else                           // cyclic scan when j >= 0
1765	<	k += (m >>> 1) | 1;        // step by half to reduce bias
1766	<
1767	<	if ((q = ws[k & m]) != null && (b = q.base) - q.top < 0) {
1768	<	if (ec >= 0)
1769	<	task = q.pollAt(b);    // steal
1770	<	break;
1753	>	WorkQueue[] ws;                       // first update random seed
1754	>	int r = w.seed; r ^= r << 13; r ^= r >>> 17; w.seed = r ^= r << 5;
1755	>	int rs = runState, m;                 // volatile read order matters
1756	>	if ((ws = workQueues) != null && (m = ws.length - 1) > 0) {
1757	>	int ec = w.eventCount;            // ec is negative if inactive
1758	>	int step = (r >>> 16) | 1;        // relative prime
1759	>	for (int j = (m + 1) << 2; ; r += step) {
1760	>	WorkQueue q; ForkJoinTask<?> t; ForkJoinTask<?>[] a; int b;
1761	>	if ((q = ws[r & m]) != null && (b = q.base) - q.top < 0 &&
1762	>	(a = q.array) != null) {  // probably nonempty
1763	>	int i = (((a.length - 1) & b) << ASHIFT) + ABASE;
1764	>	t = (ForkJoinTask<?>)U.getObjectVolatile(a, i);
1765	>	if (q.base == b && ec >= 0 && t != null &&
1766	>	U.compareAndSwapObject(a, i, t, null)) {
1767	>	if (q.top - (q.base = b + 1) > 0)
1768	>	signalWork();    // help pushes signal
1769	>	return t;
1770	>	}
1771	>	else if (ec < 0 || j <= m) {
1772	>	rs = 0;               // mark scan as imcomplete
1773	>	break;                // caller can retry after release
1774	>	}
1775		}
1776	<	else if (j > m) {
1485	<	if (rs == runState)        // staleness check
1486	<	swept = true;
1776	>	if (--j < 0)
1777		break;
1778	+	}
1779	+
1780	+	long c = ctl; int e = (int)c, a = (int)(c >> AC_SHIFT), nr, ns;
1781	+	if (e < 0)                        // decode ctl on empty scan
1782	+	w.runState = -1;              // pool is terminating
1783	+	else if (rs == 0 || rs != runState) { // incomplete scan
1784	+	WorkQueue v; Thread p;        // try to release a waiter
1785	+	if (e > 0 && a < 0 && w.eventCount == ec &&
1786	+	(v = ws[e & m]) != null && v.eventCount == (e | INT_SIGN)) {
1787	+	long nc = ((long)(v.nextWait & E_MASK) |
1788	+	((c + AC_UNIT) & (AC_MASK|TC_MASK)));
1789	+	if (ctl == c && U.compareAndSwapLong(this, CTL, c, nc)) {
1790	+	v.eventCount = (e + E_SEQ) & E_MASK;
1791	+	if ((p = v.parker) != null)
1792	+	U.unpark(p);
1793	+	}
1794		}
1795		}
1796	<	w.seed = r;                        // save seed for next scan
1797	<	if (task != null)
1798	<	return task;
1799	<	}
1800	<
1801	<	// Decode ctl on empty scan
1802	<	long c = ctl; int e = (int)c, a = (int)(c >> AC_SHIFT), nr, ns;
1803	<	if (!swept) {                          // try to release a waiter
1804	<	WorkQueue v; Thread p;
1805	<	if (e > 0 && a < 0 && ws != null &&
1806	<	(v = ws[((~e << 1) | 1) & m]) != null &&
1807	<	v.eventCount == (e | INT_SIGN) && U.compareAndSwapLong
1808	<	(this, CTL, c, ((long)(v.nextWait & E_MASK) |
1809	<	((c + AC_UNIT) & (AC_MASK|TC_MASK))))) {
1810	<	v.eventCount = (e + E_SEQ) & E_MASK;
1811	<	if ((p = v.parker) != null)
1812	<	U.unpark(p);
1813	<	}
1814	<	}
1815	<	else if ((nr = w.rescans) > 0) {       // continue rescanning
1816	<	int ac = a + parallelism;
1817	<	if ((w.rescans = (ac < nr) ? ac : nr - 1) > 0 && w.seed < 0 &&
1818	<	w.eventCount == ec)
1819	<	Thread.yield();                // 1 bit randomness for yield call
1820	<	}
1821	<	else if (e < 0)                        // pool is terminating
1822	<	w.runState = -1;
1823	<	else if (ec >= 0) {                    // try to enqueue
1824	<	long nc = (long)ec | ((c - AC_UNIT) & (AC_MASK|TC_MASK));
1825	<	w.nextWait = e;
1520	<	w.eventCount = ec | INT_SIGN;      // mark as inactive
1521	<	if (!U.compareAndSwapLong(this, CTL, c, nc))
1522	<	w.eventCount = ec;             // back out on CAS failure
1523	<	else if ((ns = w.nsteals) != 0) {  // set rescans if ran task
1524	<	if (a <= 0)                    // ... unless too many active
1525	<	w.rescans = a + parallelism;
1526	<	w.nsteals = 0;
1527	<	w.totalSteals += ns;
1528	<	}
1529	<	}
1530	<	else{                                  // already queued
1531	<	if (parallelism == -a)
1532	<	idleAwaitWork(w);              // quiescent
1533	<	if (w.eventCount == ec) {
1534	<	Thread.interrupted();          // clear status
1535	<	ForkJoinWorkerThread wt = w.owner;
1536	<	U.putObject(wt, PARKBLOCKER, this);
1537	<	w.parker = wt;                 // emulate LockSupport.park
1538	<	if (w.eventCount == ec)        // recheck
1539	<	U.park(false, 0L);         // block
1540	<	w.parker = null;
1541	<	U.putObject(wt, PARKBLOCKER, null);
1796	>	else if (ec >= 0) {               // try to enqueue/inactivate
1797	>	long nc = (long)ec | ((c - AC_UNIT) & (AC_MASK|TC_MASK));
1798	>	w.nextWait = e;
1799	>	w.eventCount = ec | INT_SIGN; // mark as inactive
1800	>	if (ctl != c || !U.compareAndSwapLong(this, CTL, c, nc))
1801	>	w.eventCount = ec;        // unmark on CAS failure
1802	>	else {
1803	>	if ((ns = w.nsteals) != 0) {
1804	>	w.nsteals = 0;        // set rescans if ran task
1805	>	w.rescans = (a > 0) ? 0 : a + parallelism;
1806	>	w.totalSteals += ns;
1807	>	}
1808	>	if (a == 1 - parallelism) // quiescent
1809	>	idleAwaitWork(w, nc, c);
1810	>	}
1811	>	}
1812	>	else if (w.eventCount < 0) {      // already queued
1813	>	int ac = a + parallelism;
1814	>	if ((nr = w.rescans) > 0)     // continue rescanning
1815	>	w.rescans = (ac < nr) ? ac : nr - 1;
1816	>	else if (((w.seed >>> 16) & ac) == 0) { // randomize park
1817	>	Thread.interrupted();     // clear status
1818	>	Thread wt = Thread.currentThread();
1819	>	U.putObject(wt, PARKBLOCKER, this);
1820	>	w.parker = wt;            // emulate LockSupport.park
1821	>	if (w.eventCount < 0)     // recheck
1822	>	U.park(false, 0L);
1823	>	w.parker = null;
1824	>	U.putObject(wt, PARKBLOCKER, null);
1825	>	}
1826		}
1827		}
1828		return null;
1829		}
1830
1831		/**
1832	<	* If inactivating worker w has caused pool to become quiescent,
1833	<	* checks for pool termination, and, so long as this is not the
1834	<	* only worker, waits for event for up to SHRINK_RATE nanosecs.
1835	<	* On timeout, if ctl has not changed, terminates the worker,
1836	<	* which will in turn wake up another worker to possibly repeat
1837	<	* this process.
1832	>	* If inactivating worker w has caused the pool to become
1833	>	* quiescent, checks for pool termination, and, so long as this is
1834	>	* not the only worker, waits for event for up to a given
1835	>	* duration.  On timeout, if ctl has not changed, terminates the
1836	>	* worker, which will in turn wake up another worker to possibly
1837	>	* repeat this process.
1838		*
1839		* @param w the calling worker
1840	+	* @param currentCtl the ctl value triggering possible quiescence
1841	+	* @param prevCtl the ctl value to restore if thread is terminated
1842		*/
1843	<	private void idleAwaitWork(WorkQueue w) {
1844	<	long c; int nw, ec;
1845	<	if (!tryTerminate(false) &&
1846	<	(int)((c = ctl) >> AC_SHIFT) + parallelism == 0 &&
1847	<	(ec = w.eventCount) == ((int)c | INT_SIGN) &&
1848	<	(nw = w.nextWait) != 0) {
1849	<	long nc = ((long)(nw & E_MASK) | // ctl to restore on timeout
1850	<	((c + AC_UNIT) & AC_MASK) | (c & TC_MASK));
1565	<	ForkJoinTask.helpExpungeStaleExceptions(); // help clean
1566	<	ForkJoinWorkerThread wt = w.owner;
1567	<	while (ctl == c) {
1568	<	long startTime = System.nanoTime();
1843	>	private void idleAwaitWork(WorkQueue w, long currentCtl, long prevCtl) {
1844	>	if (w.eventCount < 0 && !tryTerminate(false, false) &&
1845	>	(int)prevCtl != 0 && !hasQueuedSubmissions() && ctl == currentCtl) {
1846	>	int dc = -(short)(currentCtl >>> TC_SHIFT);
1847	>	long parkTime = dc < 0 ? FAST_IDLE_TIMEOUT: (dc + 1) * IDLE_TIMEOUT;
1848	>	long deadline = System.nanoTime() + parkTime - 100000L; // 1ms slop
1849	>	Thread wt = Thread.currentThread();
1850	>	while (ctl == currentCtl) {
1851		Thread.interrupted();  // timed variant of version in scan()
1852		U.putObject(wt, PARKBLOCKER, this);
1853		w.parker = wt;
1854	<	if (ctl == c)
1855	<	U.park(false, SHRINK_RATE);
1854	>	if (ctl == currentCtl)
1855	>	U.park(false, parkTime);
1856		w.parker = null;
1857		U.putObject(wt, PARKBLOCKER, null);
1858	<	if (ctl != c)
1858	>	if (ctl != currentCtl)
1859		break;
1860	<	if (System.nanoTime() - startTime >= SHRINK_TIMEOUT &&
1861	<	U.compareAndSwapLong(this, CTL, c, nc)) {
1862	<	w.runState = -1;          // shrink
1863	<	w.eventCount = (ec + E_SEQ) | E_MASK;
1860	>	if (deadline - System.nanoTime() <= 0L &&
1861	>	U.compareAndSwapLong(this, CTL, currentCtl, prevCtl)) {
1862	>	w.eventCount = (w.eventCount + E_SEQ) | E_MASK;
1863	>	w.runState = -1;   // shrink
1864		break;
1865		}
1866		}
#	Line 1596 | Line 1878 | public class ForkJoinPool extends Abstra
1878		* leaves hints in workers to speed up subsequent calls. The
1879		* implementation is very branchy to cope with potential
1880		* inconsistencies or loops encountering chains that are stale,
1881	<	* unknown, or of length greater than MAX_HELP_DEPTH links.  All
1600	<	* of these cases are dealt with by just retrying by caller.
1881	>	* unknown, or so long that they are likely cyclic.
1882		*
1883		* @param joiner the joining worker
1884		* @param task the task to join
1885	<	* @return true if found or ran a task (and so is immediately retryable)
1885	>	* @return 0 if no progress can be made, negative if task
1886	>	* known complete, else positive
1887		*/
1888	<	final boolean tryHelpStealer(WorkQueue joiner, ForkJoinTask<?> task) {
1889	<	ForkJoinTask<?> subtask;    // current target
1890	<	boolean progress = false;
1891	<	int depth = 0;              // current chain depth
1892	<	int m = runState & SMASK;
1893	<	WorkQueue[] ws = workQueues;
1894	<
1895	<	if (ws != null && ws.length > m && (subtask = task).status >= 0) {
1896	<	outer:for (WorkQueue j = joiner;;) {
1897	<	// Try to find the stealer of subtask, by first using hint
1616	<	WorkQueue stealer = null;
1617	<	WorkQueue v = ws[j.stealHint & m];
1618	<	if (v != null && v.currentSteal == subtask)
1619	<	stealer = v;
1620	<	else {
1621	<	for (int i = 1; i <= m; i += 2) {
1622	<	if ((v = ws[i]) != null && v.currentSteal == subtask) {
1623	<	stealer = v;
1624	<	j.stealHint = i; // save hint
1625	<	break;
1626	<	}
1888	>	private int tryHelpStealer(WorkQueue joiner, ForkJoinTask<?> task) {
1889	>	int stat = 0, steps = 0;                    // bound to avoid cycles
1890	>	if (joiner != null && task != null) {       // hoist null checks
1891	>	restart: for (;;) {
1892	>	ForkJoinTask<?> subtask = task;     // current target
1893	>	for (WorkQueue j = joiner, v;;) {   // v is stealer of subtask
1894	>	WorkQueue[] ws; int m, s, h;
1895	>	if ((s = task.status) < 0) {
1896	>	stat = s;
1897	>	break restart;
1898		}
1899	<	if (stealer == null)
1900	<	break;
1901	<	}
1902	<
1903	<	for (WorkQueue q = stealer;;) { // Try to help stealer
1904	<	ForkJoinTask<?> t; int b;
1905	<	if (task.status < 0)
1906	<	break outer;
1907	<	if ((b = q.base) - q.top < 0) {
1908	<	progress = true;
1909	<	if (subtask.status < 0)
1910	<	break outer;               // stale
1911	<	if ((t = q.pollAt(b)) != null) {
1912	<	stealer.stealHint = joiner.poolIndex;
1913	<	joiner.runSubtask(t);
1899	>	if ((ws = workQueues) == null || (m = ws.length - 1) <= 0)
1900	>	break restart;              // shutting down
1901	>	if ((v = ws[h = (j.stealHint | 1) & m]) == null ||
1902	>	v.currentSteal != subtask) {
1903	>	for (int origin = h;;) {    // find stealer
1904	>	if (((h = (h + 2) & m) & 15) == 1 &&
1905	>	(subtask.status < 0 || j.currentJoin != subtask))
1906	>	continue restart;   // occasional staleness check
1907	>	if ((v = ws[h]) != null &&
1908	>	v.currentSteal == subtask) {
1909	>	j.stealHint = h;    // save hint
1910	>	break;
1911	>	}
1912	>	if (h == origin)
1913	>	break restart;      // cannot find stealer
1914		}
1915		}
1916	<	else { // empty - try to descend to find stealer's stealer
1917	<	ForkJoinTask<?> next = stealer.currentJoin;
1918	<	if (++depth == MAX_HELP_DEPTH || subtask.status < 0 ||
1919	<	next == null || next == subtask)
1920	<	break outer;  // max depth, stale, dead-end, cyclic
1921	<	subtask = next;
1922	<	j = stealer;
1923	<	break;
1916	>	for (;;) { // help stealer or descend to its stealer
1917	>	ForkJoinTask[] a;  int b;
1918	>	if (subtask.status < 0)     // surround probes with
1919	>	continue restart;       //   consistency checks
1920	>	if ((b = v.base) - v.top < 0 && (a = v.array) != null) {
1921	>	int i = (((a.length - 1) & b) << ASHIFT) + ABASE;
1922	>	ForkJoinTask<?> t =
1923	>	(ForkJoinTask<?>)U.getObjectVolatile(a, i);
1924	>	if (subtask.status < 0 || j.currentJoin != subtask ||
1925	>	v.currentSteal != subtask)
1926	>	continue restart;   // stale
1927	>	stat = 1;               // apparent progress
1928	>	if (t != null && v.base == b &&
1929	>	U.compareAndSwapObject(a, i, t, null)) {
1930	>	v.base = b + 1;     // help stealer
1931	>	joiner.runSubtask(t);
1932	>	}
1933	>	else if (v.base == b && ++steps == MAX_HELP)
1934	>	break restart;      // v apparently stalled
1935	>	}
1936	>	else {                      // empty -- try to descend
1937	>	ForkJoinTask<?> next = v.currentJoin;
1938	>	if (subtask.status < 0 || j.currentJoin != subtask ||
1939	>	v.currentSteal != subtask)
1940	>	continue restart;   // stale
1941	>	else if (next == null || ++steps == MAX_HELP)
1942	>	break restart;      // dead-end or maybe cyclic
1943	>	else {
1944	>	subtask = next;
1945	>	j = v;
1946	>	break;
1947	>	}
1948	>	}
1949		}
1950		}
1951		}
1952		}
1953	<	return progress;
1953	>	return stat;
1954		}
1955
1956		/**
#	Line 1663 | Line 1959 | public class ForkJoinPool extends Abstra
1959		* @param joiner the joining worker
1960		* @param task the task
1961		*/
1962	<	final void tryPollForAndExec(WorkQueue joiner, ForkJoinTask<?> task) {
1962	>	private void tryPollForAndExec(WorkQueue joiner, ForkJoinTask<?> task) {
1963		WorkQueue[] ws;
1964	<	int m = runState & SMASK;
1965	<	if ((ws = workQueues) != null && ws.length > m) {
1670	<	for (int j = 1; j <= m && task.status >= 0; j += 2) {
1964	>	if ((ws = workQueues) != null) {
1965	>	for (int j = 1; j < ws.length && task.status >= 0; j += 2) {
1966		WorkQueue q = ws[j];
1967		if (q != null && q.pollFor(task)) {
1968		joiner.runSubtask(task);
#	Line 1678 | Line 1973 | public class ForkJoinPool extends Abstra
1973		}
1974
1975		/**
1976	<	* Returns a non-empty steal queue, if one is found during a random,
1977	<	* then cyclic scan, else null.  This method must be retried by
1978	<	* caller if, by the time it tries to use the queue, it is empty.
1976	>	* Tries to decrement active count (sometimes implicitly) and
1977	>	* possibly release or create a compensating worker in preparation
1978	>	* for blocking. Fails on contention or termination. Otherwise,
1979	>	* adds a new thread if no idle workers are available and either
1980	>	* pool would become completely starved or: (at least half
1981	>	* starved, and fewer than 50% spares exist, and there is at least
1982	>	* one task apparently available). Even though the availability
1983	>	* check requires a full scan, it is worthwhile in reducing false
1984	>	* alarms.
1985	>	*
1986	>	* @param task if non-null, a task being waited for
1987	>	* @param blocker if non-null, a blocker being waited for
1988	>	* @return true if the caller can block, else should recheck and retry
1989	>	*/
1990	>	final boolean tryCompensate(ForkJoinTask<?> task, ManagedBlocker blocker) {
1991	>	int pc = parallelism, e;
1992	>	long c = ctl;
1993	>	WorkQueue[] ws = workQueues;
1994	>	if ((e = (int)c) >= 0 && ws != null) {
1995	>	int u, a, ac, hc;
1996	>	int tc = (short)((u = (int)(c >>> 32)) >>> UTC_SHIFT) + pc;
1997	>	boolean replace = false;
1998	>	if ((a = u >> UAC_SHIFT) <= 0) {
1999	>	if ((ac = a + pc) <= 1)
2000	>	replace = true;
2001	>	else if ((e > 0 || (task != null &&
2002	>	ac <= (hc = pc >>> 1) && tc < pc + hc))) {
2003	>	WorkQueue w;
2004	>	for (int j = 0; j < ws.length; ++j) {
2005	>	if ((w = ws[j]) != null && !w.isEmpty()) {
2006	>	replace = true;
2007	>	break;   // in compensation range and tasks available
2008	>	}
2009	>	}
2010	>	}
2011	>	}
2012	>	if ((task == null || task.status >= 0) && // recheck need to block
2013	>	(blocker == null || !blocker.isReleasable()) && ctl == c) {
2014	>	if (!replace) {          // no compensation
2015	>	long nc = ((c - AC_UNIT) & AC_MASK) | (c & ~AC_MASK);
2016	>	if (U.compareAndSwapLong(this, CTL, c, nc))
2017	>	return true;
2018	>	}
2019	>	else if (e != 0) {       // release an idle worker
2020	>	WorkQueue w; Thread p; int i;
2021	>	if ((i = e & SMASK) < ws.length && (w = ws[i]) != null) {
2022	>	long nc = ((long)(w.nextWait & E_MASK) |
2023	>	(c & (AC_MASK|TC_MASK)));
2024	>	if (w.eventCount == (e | INT_SIGN) &&
2025	>	U.compareAndSwapLong(this, CTL, c, nc)) {
2026	>	w.eventCount = (e + E_SEQ) & E_MASK;
2027	>	if ((p = w.parker) != null)
2028	>	U.unpark(p);
2029	>	return true;
2030	>	}
2031	>	}
2032	>	}
2033	>	else if (tc < MAX_CAP) { // create replacement
2034	>	long nc = ((c + TC_UNIT) & TC_MASK) | (c & ~TC_MASK);
2035	>	if (U.compareAndSwapLong(this, CTL, c, nc)) {
2036	>	addWorker();
2037	>	return true;
2038	>	}
2039	>	}
2040	>	}
2041	>	}
2042	>	return false;
2043	>	}
2044	>
2045	>	/**
2046	>	* Helps and/or blocks until the given task is done.
2047	>	*
2048	>	* @param joiner the joining worker
2049	>	* @param task the task
2050	>	* @return task status on exit
2051	>	*/
2052	>	final int awaitJoin(WorkQueue joiner, ForkJoinTask<?> task) {
2053	>	int s;
2054	>	if ((s = task.status) >= 0) {
2055	>	ForkJoinTask<?> prevJoin = joiner.currentJoin;
2056	>	joiner.currentJoin = task;
2057	>	long startTime = 0L;
2058	>	for (int k = 0;;) {
2059	>	if ((s = (joiner.isEmpty() ?           // try to help
2060	>	tryHelpStealer(joiner, task) :
2061	>	joiner.tryRemoveAndExec(task))) == 0 &&
2062	>	(s = task.status) >= 0) {
2063	>	if (k == 0) {
2064	>	startTime = System.nanoTime();
2065	>	tryPollForAndExec(joiner, task); // check uncommon case
2066	>	}
2067	>	else if ((k & (MAX_HELP - 1)) == 0 &&
2068	>	System.nanoTime() - startTime >=
2069	>	COMPENSATION_DELAY &&
2070	>	tryCompensate(task, null)) {
2071	>	if (task.trySetSignal()) {
2072	>	synchronized (task) {
2073	>	if (task.status >= 0) {
2074	>	try {                // see ForkJoinTask
2075	>	task.wait();     //  for explanation
2076	>	} catch (InterruptedException ie) {
2077	>	}
2078	>	}
2079	>	else
2080	>	task.notifyAll();
2081	>	}
2082	>	}
2083	>	long c;                          // re-activate
2084	>	do {} while (!U.compareAndSwapLong
2085	>	(this, CTL, c = ctl, c + AC_UNIT));
2086	>	}
2087	>	}
2088	>	if (s < 0 || (s = task.status) < 0) {
2089	>	joiner.currentJoin = prevJoin;
2090	>	break;
2091	>	}
2092	>	else if ((k++ & (MAX_HELP - 1)) == MAX_HELP >>> 1)
2093	>	Thread.yield();                     // for politeness
2094	>	}
2095	>	}
2096	>	return s;
2097	>	}
2098	>
2099	>	/**
2100	>	* Stripped-down variant of awaitJoin used by timed joins. Tries
2101	>	* to help join only while there is continuous progress. (Caller
2102	>	* will then enter a timed wait.)
2103	>	*
2104	>	* @param joiner the joining worker
2105	>	* @param task the task
2106	>	* @return task status on exit
2107	>	*/
2108	>	final int helpJoinOnce(WorkQueue joiner, ForkJoinTask<?> task) {
2109	>	int s;
2110	>	while ((s = task.status) >= 0 &&
2111	>	(joiner.isEmpty() ?
2112	>	tryHelpStealer(joiner, task) :
2113	>	joiner.tryRemoveAndExec(task)) != 0)
2114	>	;
2115	>	return s;
2116	>	}
2117	>
2118	>	/**
2119	>	* Returns a (probably) non-empty steal queue, if one is found
2120	>	* during a random, then cyclic scan, else null.  This method must
2121	>	* be retried by caller if, by the time it tries to use the queue,
2122	>	* it is empty.
2123		*/
2124		private WorkQueue findNonEmptyStealQueue(WorkQueue w) {
2125	<	int r = w.seed;    // Same idea as scan(), but ignoring submissions
2125	>	// Similar to loop in scan(), but ignoring submissions
2126	>	int r = w.seed; r ^= r << 13; r ^= r >>> 17; w.seed = r ^= r << 5;
2127	>	int step = (r >>> 16) | 1;
2128		for (WorkQueue[] ws;;) {
2129	<	int m = runState & SMASK;
2130	<	if ((ws = workQueues) == null)
2129	>	int rs = runState, m;
2130	>	if ((ws = workQueues) == null || (m = ws.length - 1) < 1)
2131		return null;
2132	<	if (ws.length > m) {
2133	<	WorkQueue q;
2134	<	for (int n = m << 2, k = r, j = -n;;) {
2135	<	r ^= r << 13; r ^= r >>> 17; r ^= r << 5;
2136	<	if ((q = ws[(k | 1) & m]) != null && q.base - q.top < 0) {
2137	<	w.seed = r;
1697	<	return q;
1698	<	}
1699	<	else if (j > n)
2132	>	for (int j = (m + 1) << 2; ; r += step) {
2133	>	WorkQueue q = ws[((r << 1) | 1) & m];
2134	>	if (q != null && !q.isEmpty())
2135	>	return q;
2136	>	else if (--j < 0) {
2137	>	if (runState == rs)
2138		return null;
2139	<	else
1702	<	k = (j++ < 0) ? r : k + ((m >>> 1) | 1);
1703	<
2139	>	break;
2140		}
2141		}
2142		}
#	Line 1714 | Line 2150 | public class ForkJoinPool extends Abstra
2150		*/
2151		final void helpQuiescePool(WorkQueue w) {
2152		for (boolean active = true;;) {
2153	<	w.runLocalTasks();      // exhaust local queue
2153	>	ForkJoinTask<?> localTask; // exhaust local queue
2154	>	while ((localTask = w.nextLocalTask()) != null)
2155	>	localTask.doExec();
2156		WorkQueue q = findNonEmptyStealQueue(w);
2157		if (q != null) {
2158	<	ForkJoinTask<?> t;
2158	>	ForkJoinTask<?> t; int b;
2159		if (!active) {      // re-establish active count
2160		long c;
2161		active = true;
2162		do {} while (!U.compareAndSwapLong
2163		(this, CTL, c = ctl, c + AC_UNIT));
2164		}
2165	<	if ((t = q.poll()) != null)
2165	>	if ((b = q.base) - q.top < 0 && (t = q.pollAt(b)) != null)
2166		w.runSubtask(t);
2167		}
2168		else {
#	Line 1746 | Line 2184 | public class ForkJoinPool extends Abstra
2184		}
2185
2186		/**
2187	+	* Restricted version of helpQuiescePool for non-FJ callers
2188	+	*/
2189	+	static void externalHelpQuiescePool() {
2190	+	ForkJoinPool p; WorkQueue[] ws; WorkQueue q, sq;
2191	+	ForkJoinTask<?>[] a; int b;
2192	+	ForkJoinTask<?> t = null;
2193	+	int k = submitters.get().seed & SQMASK;
2194	+	if ((p = commonPool) != null &&
2195	+	(int)(p.ctl >> AC_SHIFT) < 0 &&
2196	+	(ws = p.workQueues) != null &&
2197	+	ws.length > (k &= p.submitMask) &&
2198	+	(q = ws[k]) != null) {
2199	+	while (q.top - q.base > 0) {
2200	+	if ((t = q.sharedPop()) != null)
2201	+	break;
2202	+	}
2203	+	if (t == null && (sq = p.findNonEmptyStealQueue(q)) != null &&
2204	+	(b = sq.base) - sq.top < 0)
2205	+	t = sq.pollAt(b);
2206	+	if (t != null)
2207	+	t.doExec();
2208	+	}
2209	+	}
2210	+
2211	+	/**
2212		* Gets and removes a local or stolen task for the given worker.
2213		*
2214		* @return a task, if available
2215		*/
2216		final ForkJoinTask<?> nextTaskFor(WorkQueue w) {
2217		for (ForkJoinTask<?> t;;) {
2218	<	WorkQueue q;
2218	>	WorkQueue q; int b;
2219		if ((t = w.nextLocalTask()) != null)
2220		return t;
2221		if ((q = findNonEmptyStealQueue(w)) == null)
2222		return null;
2223	<	if ((t = q.poll()) != null)
2223	>	if ((b = q.base) - q.top < 0 && (t = q.pollAt(b)) != null)
2224		return t;
2225		}
2226		}
#	Line 1778 | Line 2241 | public class ForkJoinPool extends Abstra
2241		8);
2242		}
2243
1781	–	// Termination
1782	–
2244		/**
2245	<	* Sets SHUTDOWN bit of runState under lock
2245	>	* Returns approximate submission queue length for the given caller
2246		*/
2247	<	private void enableShutdown() {
2248	<	ReentrantLock lock = this.lock;
2249	<	if (runState >= 0) {
2250	<	lock.lock();                       // don't need try/finally
2251	<	runState |= SHUTDOWN;
2252	<	lock.unlock();
2253	<	}
2247	>	static int getEstimatedSubmitterQueueLength() {
2248	>	ForkJoinPool p; WorkQueue[] ws; WorkQueue q;
2249	>	int k = submitters.get().seed & SQMASK;
2250	>	return ((p = commonPool) != null &&
2251	>	p.runState >= 0 &&
2252	>	(ws = p.workQueues) != null &&
2253	>	ws.length > (k &= p.submitMask) &&
2254	>	(q = ws[k]) != null) ?
2255	>	q.queueSize() : 0;
2256		}
2257
2258	+	//  Termination
2259	+
2260		/**
2261	<	* Possibly initiates and/or completes termination.  Upon
2262	<	* termination, cancels all queued tasks and then
2261	>	* Possibly initiates and/or completes termination.  The caller
2262	>	* triggering termination runs three passes through workQueues:
2263	>	* (0) Setting termination status, followed by wakeups of queued
2264	>	* workers; (1) cancelling all tasks; (2) interrupting lagging
2265	>	* threads (likely in external tasks, but possibly also blocked in
2266	>	* joins).  Each pass repeats previous steps because of potential
2267	>	* lagging thread creation.
2268		*
2269		* @param now if true, unconditionally terminate, else only
2270		* if no work and no active workers
2271	+	* @param enable if true, enable shutdown when next possible
2272		* @return true if now terminating or terminated
2273		*/
2274	<	private boolean tryTerminate(boolean now) {
2274	>	private boolean tryTerminate(boolean now, boolean enable) {
2275		for (long c;;) {
2276		if (((c = ctl) & STOP_BIT) != 0) {      // already terminating
2277		if ((short)(c >>> TC_SHIFT) == -parallelism) {
2278	<	ReentrantLock lock = this.lock; // signal when no workers
2279	<	lock.lock();                    // don't need try/finally
2280	<	termination.signalAll();        // signal when 0 workers
1810	<	lock.unlock();
2278	>	synchronized(this) {
2279	>	notifyAll();                // signal when 0 workers
2280	>	}
2281		}
2282		return true;
2283		}
2284	<	if (!now) {
2285	<	if ((int)(c >> AC_SHIFT) != -parallelism || runState >= 0 ||
2284	>	if (runState >= 0) {                    // not yet enabled
2285	>	if (!enable)
2286	>	return false;
2287	>	while (!U.compareAndSwapInt(this, MAINLOCK, 0, 1))
2288	>	tryAwaitMainLock();
2289	>	try {
2290	>	runState |= SHUTDOWN;
2291	>	} finally {
2292	>	if (!U.compareAndSwapInt(this, MAINLOCK, 1, 0)) {
2293	>	mainLock = 0;
2294	>	synchronized (this) { notifyAll(); };
2295	>	}
2296	>	}
2297	>	}
2298	>	if (!now) {                             // check if idle & no tasks
2299	>	if ((int)(c >> AC_SHIFT) != -parallelism ||
2300		hasQueuedSubmissions())
2301		return false;
2302		// Check for unqueued inactive workers. One pass suffices.
2303		WorkQueue[] ws = workQueues; WorkQueue w;
2304		if (ws != null) {
2305	<	int n = ws.length;
1822	<	for (int i = 1; i < n; i += 2) {
2305	>	for (int i = 1; i < ws.length; i += 2) {
2306		if ((w = ws[i]) != null && w.eventCount >= 0)
2307		return false;
2308		}
2309		}
2310		}
2311	<	if (U.compareAndSwapLong(this, CTL, c, c | STOP_BIT))
2312	<	startTerminating();
2313	<	}
2314	<	}
2315	<
2316	<	/**
2317	<	* Initiates termination: Runs three passes through workQueues:
2318	<	* (0) Setting termination status, followed by wakeups of queued
2319	<	* workers; (1) cancelling all tasks; (2) interrupting lagging
2320	<	* threads (likely in external tasks, but possibly also blocked in
2321	<	* joins).  Each pass repeats previous steps because of potential
2322	<	* lagging thread creation.
2323	<	*/
1841	<	private void startTerminating() {
1842	<	for (int pass = 0; pass < 3; ++pass) {
1843	<	WorkQueue[] ws = workQueues;
1844	<	if (ws != null) {
1845	<	WorkQueue w; Thread wt;
1846	<	int n = ws.length;
1847	<	for (int j = 0; j < n; ++j) {
1848	<	if ((w = ws[j]) != null) {
1849	<	w.runState = -1;
1850	<	if (pass > 0) {
1851	<	w.cancelAll();
1852	<	if (pass > 1 && (wt = w.owner) != null &&
1853	<	!wt.isInterrupted()) {
1854	<	try {
1855	<	wt.interrupt();
1856	<	} catch (SecurityException ignore) {
2311	>	if (U.compareAndSwapLong(this, CTL, c, c | STOP_BIT)) {
2312	>	for (int pass = 0; pass < 3; ++pass) {
2313	>	WorkQueue[] ws = workQueues;
2314	>	if (ws != null) {
2315	>	WorkQueue w;
2316	>	int n = ws.length;
2317	>	for (int i = 0; i < n; ++i) {
2318	>	if ((w = ws[i]) != null) {
2319	>	w.runState = -1;
2320	>	if (pass > 0) {
2321	>	w.cancelAll();
2322	>	if (pass > 1)
2323	>	w.interruptOwner();
2324		}
2325		}
2326		}
2327	<	}
2328	<	}
2329	<	// Wake up workers parked on event queue
2330	<	int i, e; long c; Thread p;
2331	<	while ((i = ((~(e = (int)(c = ctl)) << 1) | 1) & SMASK) < n &&
2332	<	(w = ws[i]) != null &&
2333	<	w.eventCount == (e | INT_SIGN)) {
2334	<	long nc = ((long)(w.nextWait & E_MASK) |
2335	<	((c + AC_UNIT) & AC_MASK) |
2336	<	(c & (TC_MASK|STOP_BIT)));
2337	<	if (U.compareAndSwapLong(this, CTL, c, nc)) {
2338	<	w.eventCount = (e + E_SEQ) & E_MASK;
2339	<	if ((p = w.parker) != null)
2340	<	U.unpark(p);
2327	>	// Wake up workers parked on event queue
2328	>	int i, e; long cc; Thread p;
2329	>	while ((e = (int)(cc = ctl) & E_MASK) != 0 &&
2330	>	(i = e & SMASK) < n &&
2331	>	(w = ws[i]) != null) {
2332	>	long nc = ((long)(w.nextWait & E_MASK) |
2333	>	((cc + AC_UNIT) & AC_MASK) |
2334	>	(cc & (TC_MASK|STOP_BIT)));
2335	>	if (w.eventCount == (e | INT_SIGN) &&
2336	>	U.compareAndSwapLong(this, CTL, cc, nc)) {
2337	>	w.eventCount = (e + E_SEQ) & E_MASK;
2338	>	w.runState = -1;
2339	>	if ((p = w.parker) != null)
2340	>	U.unpark(p);
2341	>	}
2342	>	}
2343		}
2344		}
2345		}
#	Line 1946 | Line 2415 | public class ForkJoinPool extends Abstra
2415		checkPermission();
2416		if (factory == null)
2417		throw new NullPointerException();
2418	<	if (parallelism <= 0 || parallelism > MAX_ID)
2418	>	if (parallelism <= 0 || parallelism > MAX_CAP)
2419		throw new IllegalArgumentException();
2420		this.parallelism = parallelism;
2421		this.factory = factory;
2422		this.ueh = handler;
2423		this.localMode = asyncMode ? FIFO_QUEUE : LIFO_QUEUE;
1955	–	this.nextPoolIndex = 1;
2424		long np = (long)(-parallelism); // offset ctl counts
2425		this.ctl = ((np << AC_SHIFT) & AC_MASK) | ((np << TC_SHIFT) & TC_MASK);
2426	<	// initialize workQueues array with room for 2*parallelism if possible
2427	<	int n = parallelism << 1;
2428	<	if (n >= MAX_ID)
2429	<	n = MAX_ID;
2430	<	else { // See Hackers Delight, sec 3.2, where n < (1 << 16)
1963	<	n |= n >>> 1; n |= n >>> 2; n |= n >>> 4; n |= n >>> 8;
1964	<	}
1965	<	this.workQueues = new WorkQueue[(n + 1) << 1];
1966	<	ReentrantLock lck = this.lock = new ReentrantLock();
1967	<	this.termination = lck.newCondition();
1968	<	this.stealCount = new AtomicLong();
1969	<	this.nextWorkerNumber = new AtomicInteger();
2426	>	// Use nearest power 2 for workQueues size. See Hackers Delight sec 3.2.
2427	>	int n = parallelism - 1;
2428	>	n |= n >>> 1; n |= n >>> 2; n |= n >>> 4; n |= n >>> 8; n |= n >>> 16;
2429	>	this.submitMask = ((n + 1) << 1) - 1;
2430	>	int pn = poolNumberGenerator.incrementAndGet();
2431		StringBuilder sb = new StringBuilder("ForkJoinPool-");
2432	<	sb.append(poolNumberGenerator.incrementAndGet());
2432	>	sb.append(Integer.toString(pn));
2433		sb.append("-worker-");
2434		this.workerNamePrefix = sb.toString();
2435	<	// Create initial submission queue
2436	<	WorkQueue sq = tryAddSharedQueue(0);
2437	<	if (sq != null)
2438	<	sq.growArray(false);
2435	>	this.runState = 1;              // set init flag
2436	>	}
2437	>
2438	>	/**
2439	>	* Constructor for common pool, suitable only for static initialization.
2440	>	* Basically the same as above, but uses smallest possible initial footprint.
2441	>	*/
2442	>	ForkJoinPool(int parallelism, int submitMask,
2443	>	ForkJoinWorkerThreadFactory factory,
2444	>	Thread.UncaughtExceptionHandler handler) {
2445	>	this.factory = factory;
2446	>	this.ueh = handler;
2447	>	this.submitMask = submitMask;
2448	>	this.parallelism = parallelism;
2449	>	long np = (long)(-parallelism);
2450	>	this.ctl = ((np << AC_SHIFT) & AC_MASK) | ((np << TC_SHIFT) & TC_MASK);
2451	>	this.localMode = LIFO_QUEUE;
2452	>	this.workerNamePrefix = "ForkJoinPool.commonPool-worker-";
2453	>	this.runState = 1;
2454	>	}
2455	>
2456	>	/**
2457	>	* Returns the common pool instance.
2458	>	*
2459	>	* @return the common pool instance
2460	>	*/
2461	>	public static ForkJoinPool commonPool() {
2462	>	ForkJoinPool p;
2463	>	if ((p = commonPool) == null)
2464	>	throw new Error("Common Pool Unavailable");
2465	>	return p;
2466		}
2467
2468		// Execution methods
#	Line 1996 | Line 2484 | public class ForkJoinPool extends Abstra
2484		*         scheduled for execution
2485		*/
2486		public <T> T invoke(ForkJoinTask<T> task) {
2487	+	if (task == null)
2488	+	throw new NullPointerException();
2489		doSubmit(task);
2490		return task.join();
2491		}
#	Line 2009 | Line 2499 | public class ForkJoinPool extends Abstra
2499		*         scheduled for execution
2500		*/
2501		public void execute(ForkJoinTask<?> task) {
2502	+	if (task == null)
2503	+	throw new NullPointerException();
2504		doSubmit(task);
2505		}
2506
#	Line 2026 | Line 2518 | public class ForkJoinPool extends Abstra
2518		if (task instanceof ForkJoinTask<?>) // avoid re-wrap
2519		job = (ForkJoinTask<?>) task;
2520		else
2521	<	job = ForkJoinTask.adapt(task, null);
2521	>	job = new ForkJoinTask.AdaptedRunnableAction(task);
2522		doSubmit(job);
2523		}
2524
#	Line 2040 | Line 2532 | public class ForkJoinPool extends Abstra
2532		*         scheduled for execution
2533		*/
2534		public <T> ForkJoinTask<T> submit(ForkJoinTask<T> task) {
2535	+	if (task == null)
2536	+	throw new NullPointerException();
2537		doSubmit(task);
2538		return task;
2539		}
#	Line 2050 | Line 2544 | public class ForkJoinPool extends Abstra
2544		*         scheduled for execution
2545		*/
2546		public <T> ForkJoinTask<T> submit(Callable<T> task) {
2547	<	if (task == null)
2054	<	throw new NullPointerException();
2055	<	ForkJoinTask<T> job = ForkJoinTask.adapt(task);
2547	>	ForkJoinTask<T> job = new ForkJoinTask.AdaptedCallable<T>(task);
2548		doSubmit(job);
2549		return job;
2550		}
#	Line 2063 | Line 2555 | public class ForkJoinPool extends Abstra
2555		*         scheduled for execution
2556		*/
2557		public <T> ForkJoinTask<T> submit(Runnable task, T result) {
2558	<	if (task == null)
2067	<	throw new NullPointerException();
2068	<	ForkJoinTask<T> job = ForkJoinTask.adapt(task, result);
2558	>	ForkJoinTask<T> job = new ForkJoinTask.AdaptedRunnable<T>(task, result);
2559		doSubmit(job);
2560		return job;
2561		}
#	Line 2082 | Line 2572 | public class ForkJoinPool extends Abstra
2572		if (task instanceof ForkJoinTask<?>) // avoid re-wrap
2573		job = (ForkJoinTask<?>) task;
2574		else
2575	<	job = ForkJoinTask.adapt(task, null);
2575	>	job = new ForkJoinTask.AdaptedRunnableAction(task);
2576		doSubmit(job);
2577		return job;
2578		}
#	Line 2092 | Line 2582 | public class ForkJoinPool extends Abstra
2582		* @throws RejectedExecutionException {@inheritDoc}
2583		*/
2584		public <T> List<Future<T>> invokeAll(Collection<? extends Callable<T>> tasks) {
2585	<	ArrayList<ForkJoinTask<T>> forkJoinTasks =
2586	<	new ArrayList<ForkJoinTask<T>>(tasks.size());
2587	<	for (Callable<T> task : tasks)
2588	<	forkJoinTasks.add(ForkJoinTask.adapt(task));
2589	<	invoke(new InvokeAll<T>(forkJoinTasks));
2590	<
2585	>	// In previous versions of this class, this method constructed
2586	>	// a task to run ForkJoinTask.invokeAll, but now external
2587	>	// invocation of multiple tasks is at least as efficient.
2588	>	List<ForkJoinTask<T>> fs = new ArrayList<ForkJoinTask<T>>(tasks.size());
2589	>	// Workaround needed because method wasn't declared with
2590	>	// wildcards in return type but should have been.
2591		@SuppressWarnings({"unchecked", "rawtypes"})
2592	<	List<Future<T>> futures = (List<Future<T>>) (List) forkJoinTasks;
2103	<	return futures;
2104	<	}
2592	>	List<Future<T>> futures = (List<Future<T>>) (List) fs;
2593
2594	<	static final class InvokeAll<T> extends RecursiveAction {
2595	<	final ArrayList<ForkJoinTask<T>> tasks;
2596	<	InvokeAll(ArrayList<ForkJoinTask<T>> tasks) { this.tasks = tasks; }
2597	<	public void compute() {
2598	<	try { invokeAll(tasks); }
2599	<	catch (Exception ignore) {}
2594	>	boolean done = false;
2595	>	try {
2596	>	for (Callable<T> t : tasks) {
2597	>	ForkJoinTask<T> f = new ForkJoinTask.AdaptedCallable<T>(t);
2598	>	doSubmit(f);
2599	>	fs.add(f);
2600	>	}
2601	>	for (ForkJoinTask<T> f : fs)
2602	>	f.quietlyJoin();
2603	>	done = true;
2604	>	return futures;
2605	>	} finally {
2606	>	if (!done)
2607	>	for (ForkJoinTask<T> f : fs)
2608	>	f.cancel(false);
2609		}
2113	–	private static final long serialVersionUID = -7914297376763021607L;
2610		}
2611
2612		/**
#	Line 2142 | Line 2638 | public class ForkJoinPool extends Abstra
2638		}
2639
2640		/**
2641	+	* Returns the targeted parallelism level of the common pool.
2642	+	*
2643	+	* @return the targeted parallelism level of the common pool
2644	+	*/
2645	+	public static int getCommonPoolParallelism() {
2646	+	return commonPoolParallelism;
2647	+	}
2648	+
2649	+	/**
2650		* Returns the number of worker threads that have started but not
2651		* yet terminated.  The result returned by this method may differ
2652		* from {@link #getParallelism} when threads are created to
#	Line 2175 | Line 2680 | public class ForkJoinPool extends Abstra
2680		int rc = 0;
2681		WorkQueue[] ws; WorkQueue w;
2682		if ((ws = workQueues) != null) {
2683	<	int n = ws.length;
2684	<	for (int i = 1; i < n; i += 2) {
2180	<	Thread.State s; ForkJoinWorkerThread wt;
2181	<	if ((w = ws[i]) != null && (wt = w.owner) != null &&
2182	<	w.eventCount >= 0 &&
2183	<	(s = wt.getState()) != Thread.State.BLOCKED &&
2184	<	s != Thread.State.WAITING &&
2185	<	s != Thread.State.TIMED_WAITING)
2683	>	for (int i = 1; i < ws.length; i += 2) {
2684	>	if ((w = ws[i]) != null && w.isApparentlyUnblocked())
2685		++rc;
2686		}
2687		}
#	Line 2228 | Line 2727 | public class ForkJoinPool extends Abstra
2727		* @return the number of steals
2728		*/
2729		public long getStealCount() {
2730	<	long count = stealCount.get();
2730	>	long count = stealCount;
2731		WorkQueue[] ws; WorkQueue w;
2732		if ((ws = workQueues) != null) {
2733	<	int n = ws.length;
2235	<	for (int i = 1; i < n; i += 2) {
2733	>	for (int i = 1; i < ws.length; i += 2) {
2734		if ((w = ws[i]) != null)
2735		count += w.totalSteals;
2736		}
#	Line 2254 | Line 2752 | public class ForkJoinPool extends Abstra
2752		long count = 0;
2753		WorkQueue[] ws; WorkQueue w;
2754		if ((ws = workQueues) != null) {
2755	<	int n = ws.length;
2258	<	for (int i = 1; i < n; i += 2) {
2755	>	for (int i = 1; i < ws.length; i += 2) {
2756		if ((w = ws[i]) != null)
2757		count += w.queueSize();
2758		}
#	Line 2274 | Line 2771 | public class ForkJoinPool extends Abstra
2771		int count = 0;
2772		WorkQueue[] ws; WorkQueue w;
2773		if ((ws = workQueues) != null) {
2774	<	int n = ws.length;
2278	<	for (int i = 0; i < n; i += 2) {
2774	>	for (int i = 0; i < ws.length; i += 2) {
2775		if ((w = ws[i]) != null)
2776		count += w.queueSize();
2777		}
#	Line 2292 | Line 2788 | public class ForkJoinPool extends Abstra
2788		public boolean hasQueuedSubmissions() {
2789		WorkQueue[] ws; WorkQueue w;
2790		if ((ws = workQueues) != null) {
2791	<	int n = ws.length;
2792	<	for (int i = 0; i < n; i += 2) {
2297	<	if ((w = ws[i]) != null && w.queueSize() != 0)
2791	>	for (int i = 0; i < ws.length; i += 2) {
2792	>	if ((w = ws[i]) != null && !w.isEmpty())
2793		return true;
2794		}
2795		}
#	Line 2311 | Line 2806 | public class ForkJoinPool extends Abstra
2806		protected ForkJoinTask<?> pollSubmission() {
2807		WorkQueue[] ws; WorkQueue w; ForkJoinTask<?> t;
2808		if ((ws = workQueues) != null) {
2809	<	int n = ws.length;
2315	<	for (int i = 0; i < n; i += 2) {
2809	>	for (int i = 0; i < ws.length; i += 2) {
2810		if ((w = ws[i]) != null && (t = w.poll()) != null)
2811		return t;
2812		}
#	Line 2341 | Line 2835 | public class ForkJoinPool extends Abstra
2835		int count = 0;
2836		WorkQueue[] ws; WorkQueue w; ForkJoinTask<?> t;
2837		if ((ws = workQueues) != null) {
2838	<	int n = ws.length;
2345	<	for (int i = 0; i < n; ++i) {
2838	>	for (int i = 0; i < ws.length; ++i) {
2839		if ((w = ws[i]) != null) {
2840		while ((t = w.poll()) != null) {
2841		c.add(t);
#	Line 2362 | Line 2855 | public class ForkJoinPool extends Abstra
2855		* @return a string identifying this pool, as well as its state
2856		*/
2857		public String toString() {
2858	<	long st = getStealCount();
2859	<	long qt = getQueuedTaskCount();
2860	<	long qs = getQueuedSubmissionCount();
2368	<	int rc = getRunningThreadCount();
2369	<	int pc = parallelism;
2858	>	// Use a single pass through workQueues to collect counts
2859	>	long qt = 0L, qs = 0L; int rc = 0;
2860	>	long st = stealCount;
2861		long c = ctl;
2862	+	WorkQueue[] ws; WorkQueue w;
2863	+	if ((ws = workQueues) != null) {
2864	+	for (int i = 0; i < ws.length; ++i) {
2865	+	if ((w = ws[i]) != null) {
2866	+	int size = w.queueSize();
2867	+	if ((i & 1) == 0)
2868	+	qs += size;
2869	+	else {
2870	+	qt += size;
2871	+	st += w.totalSteals;
2872	+	if (w.isApparentlyUnblocked())
2873	+	++rc;
2874	+	}
2875	+	}
2876	+	}
2877	+	}
2878	+	int pc = parallelism;
2879		int tc = pc + (short)(c >>> TC_SHIFT);
2880		int ac = pc + (int)(c >> AC_SHIFT);
2881		if (ac < 0) // ignore transient negative
#	Line 2390 | Line 2898 | public class ForkJoinPool extends Abstra
2898		}
2899
2900		/**
2901	<	* Initiates an orderly shutdown in which previously submitted
2902	<	* tasks are executed, but no new tasks will be accepted.
2903	<	* Invocation has no additional effect if already shut down.
2904	<	* Tasks that are in the process of being submitted concurrently
2905	<	* during the course of this method may or may not be rejected.
2901	>	* Possibly initiates an orderly shutdown in which previously
2902	>	* submitted tasks are executed, but no new tasks will be
2903	>	* accepted. Invocation has no effect on execution state if this
2904	>	* is the {@link #commonPool}, and no additional effect if
2905	>	* already shut down.  Tasks that are in the process of being
2906	>	* submitted concurrently during the course of this method may or
2907	>	* may not be rejected.
2908		*
2909		* @throws SecurityException if a security manager exists and
2910		*         the caller is not permitted to modify threads
#	Line 2403 | Line 2913 | public class ForkJoinPool extends Abstra
2913		*/
2914		public void shutdown() {
2915		checkPermission();
2916	<	enableShutdown();
2917	<	tryTerminate(false);
2916	>	if (this != commonPool)
2917	>	tryTerminate(false, true);
2918		}
2919
2920		/**
2921	<	* Attempts to cancel and/or stop all tasks, and reject all
2922	<	* subsequently submitted tasks.  Tasks that are in the process of
2923	<	* being submitted or executed concurrently during the course of
2924	<	* this method may or may not be rejected. This method cancels
2925	<	* both existing and unexecuted tasks, in order to permit
2926	<	* termination in the presence of task dependencies. So the method
2927	<	* always returns an empty list (unlike the case for some other
2928	<	* Executors).
2921	>	* Possibly attempts to cancel and/or stop all tasks, and reject
2922	>	* all subsequently submitted tasks.  Invocation has no effect on
2923	>	* execution state if this is the {@link #commonPool}, and no
2924	>	* additional effect if already shut down. Otherwise, tasks that
2925	>	* are in the process of being submitted or executed concurrently
2926	>	* during the course of this method may or may not be
2927	>	* rejected. This method cancels both existing and unexecuted
2928	>	* tasks, in order to permit termination in the presence of task
2929	>	* dependencies. So the method always returns an empty list
2930	>	* (unlike the case for some other Executors).
2931		*
2932		* @return an empty list
2933		* @throws SecurityException if a security manager exists and
#	Line 2425 | Line 2937 | public class ForkJoinPool extends Abstra
2937		*/
2938		public List<Runnable> shutdownNow() {
2939		checkPermission();
2940	<	enableShutdown();
2941	<	tryTerminate(true);
2940	>	if (this != commonPool)
2941	>	tryTerminate(true, true);
2942		return Collections.emptyList();
2943		}
2944
#	Line 2483 | Line 2995 | public class ForkJoinPool extends Abstra
2995		public boolean awaitTermination(long timeout, TimeUnit unit)
2996		throws InterruptedException {
2997		long nanos = unit.toNanos(timeout);
2998	<	final ReentrantLock lock = this.lock;
2999	<	lock.lock();
3000	<	try {
3001	<	for (;;) {
3002	<	if (isTerminated())
3003	<	return true;
3004	<	if (nanos <= 0)
3005	<	return false;
3006	<	nanos = termination.awaitNanos(nanos);
2998	>	if (isTerminated())
2999	>	return true;
3000	>	long startTime = System.nanoTime();
3001	>	boolean terminated = false;
3002	>	synchronized(this) {
3003	>	for (long waitTime = nanos, millis = 0L;;) {
3004	>	if (terminated = isTerminated() ||
3005	>	waitTime <= 0L ||
3006	>	(millis = unit.toMillis(waitTime)) <= 0L)
3007	>	break;
3008	>	wait(millis);
3009	>	waitTime = nanos - (System.nanoTime() - startTime);
3010		}
2496	–	} finally {
2497	–	lock.unlock();
3011		}
3012	+	return terminated;
3013		}
3014
3015		/**
#	Line 2597 | Line 3111 | public class ForkJoinPool extends Abstra
3111		ForkJoinPool p = ((t instanceof ForkJoinWorkerThread) ?
3112		((ForkJoinWorkerThread)t).pool : null);
3113		while (!blocker.isReleasable()) {
3114	<	if (p == null || p.tryCompensate()) {
3114	>	if (p == null || p.tryCompensate(null, blocker)) {
3115		try {
3116		do {} while (!blocker.isReleasable() && !blocker.block());
3117		} finally {
#	Line 2614 | Line 3128 | public class ForkJoinPool extends Abstra
3128		// implement RunnableFuture.
3129
3130		protected <T> RunnableFuture<T> newTaskFor(Runnable runnable, T value) {
3131	<	return (RunnableFuture<T>) ForkJoinTask.adapt(runnable, value);
3131	>	return new ForkJoinTask.AdaptedRunnable<T>(runnable, value);
3132		}
3133
3134		protected <T> RunnableFuture<T> newTaskFor(Callable<T> callable) {
3135	<	return (RunnableFuture<T>) ForkJoinTask.adapt(callable);
3135	>	return new ForkJoinTask.AdaptedCallable<T>(callable);
3136		}
3137
3138		// Unsafe mechanics
3139		private static final sun.misc.Unsafe U;
3140		private static final long CTL;
2627	–	private static final long RUNSTATE;
3141		private static final long PARKBLOCKER;
3142	+	private static final int ABASE;
3143	+	private static final int ASHIFT;
3144	+	private static final long NEXTWORKERNUMBER;
3145	+	private static final long STEALCOUNT;
3146	+	private static final long MAINLOCK;
3147
3148		static {
3149		poolNumberGenerator = new AtomicInteger();
3150	+	nextSubmitterSeed = new AtomicInteger(0x55555555);
3151		modifyThreadPermission = new RuntimePermission("modifyThread");
3152		defaultForkJoinWorkerThreadFactory =
3153		new DefaultForkJoinWorkerThreadFactory();
3154	+	submitters = new ThreadSubmitter();
3155		int s;
3156		try {
3157		U = getUnsafe();
3158		Class<?> k = ForkJoinPool.class;
3159	<	Class<?> tk = Thread.class;
3159	>	Class<?> ak = ForkJoinTask[].class;
3160		CTL = U.objectFieldOffset
3161		(k.getDeclaredField("ctl"));
3162	<	RUNSTATE = U.objectFieldOffset
3163	<	(k.getDeclaredField("runState"));
3162	>	NEXTWORKERNUMBER = U.objectFieldOffset
3163	>	(k.getDeclaredField("nextWorkerNumber"));
3164	>	STEALCOUNT = U.objectFieldOffset
3165	>	(k.getDeclaredField("stealCount"));
3166	>	MAINLOCK = U.objectFieldOffset
3167	>	(k.getDeclaredField("mainLock"));
3168	>	Class<?> tk = Thread.class;
3169		PARKBLOCKER = U.objectFieldOffset
3170		(tk.getDeclaredField("parkBlocker"));
3171	+	ABASE = U.arrayBaseOffset(ak);
3172	+	s = U.arrayIndexScale(ak);
3173	+	ASHIFT = 31 - Integer.numberOfLeadingZeros(s);
3174	+	} catch (Exception e) {
3175	+	throw new Error(e);
3176	+	}
3177	+	if ((s & (s-1)) != 0)
3178	+	throw new Error("data type scale not a power of two");
3179	+	try { // Establish common pool
3180	+	String pp = System.getProperty(propPrefix + "parallelism");
3181	+	String fp = System.getProperty(propPrefix + "threadFactory");
3182	+	String up = System.getProperty(propPrefix + "exceptionHandler");
3183	+	ForkJoinWorkerThreadFactory fac = (fp == null) ?
3184	+	defaultForkJoinWorkerThreadFactory :
3185	+	((ForkJoinWorkerThreadFactory)ClassLoader.
3186	+	getSystemClassLoader().loadClass(fp).newInstance());
3187	+	Thread.UncaughtExceptionHandler ueh = (up == null)? null :
3188	+	((Thread.UncaughtExceptionHandler)ClassLoader.
3189	+	getSystemClassLoader().loadClass(up).newInstance());
3190	+	int par;
3191	+	if ((pp == null || (par = Integer.parseInt(pp)) <= 0))
3192	+	par = Runtime.getRuntime().availableProcessors();
3193	+	if (par > MAX_CAP)
3194	+	par = MAX_CAP;
3195	+	commonPoolParallelism = par;
3196	+	int n = par - 1; // precompute submit mask
3197	+	n |= n >>> 1; n |= n >>> 2; n |= n >>> 4;
3198	+	n |= n >>> 8; n |= n >>> 16;
3199	+	int mask = ((n + 1) << 1) - 1;
3200	+	commonPool = new ForkJoinPool(par, mask, fac, ueh);
3201		} catch (Exception e) {
3202		throw new Error(e);
3203		}

Diff Legend

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